Haematologica, Volume 106, Issue 2 by Haematologica

haematologica Journal of the Ferrata Storti Foundation

Editor-in-Chief Jacob M. Rowe (Haifa)

Deputy Editors Carlo Balduini (Pavia), Jerry Radich (Seattle)

Managing Director Antonio Majocchi (Pavia)

Associate Editors Hélène Cavé (Paris), Monika Engelhardt (Freiburg), Steve Lane (Brisbane), PierMannuccio Mannucci (Milan), Simon Mendez-Ferrer (Cambridge), Pavan Reddy (Ann Arbor), David C. Rees (London), Francesco Rodeghiero (Vicenza), Davide Rossi (Bellinzona), Gilles Salles (New York), Aaron Schimmer (Toronto), Richard F Schlenk (Heidelberg), Sonali Smith (Chicago)

Assistant Editors Anne Freckleton (English Editor), Britta Dorst (English Editor), Cristiana Pascutto (Statistical Consultant), Rachel Stenner (English Editor),

Editorial Board Jeremy Abramson (Boston); Paolo Arosio (Brescia); Raphael Bejar (San Diego); Erik Berntorp (Malmö); Dominique Bonnet (London); Jean-Pierre Bourquin (Zurich); Suzanne Cannegieter (Leiden); Francisco Cervantes (Barcelona); Nicholas Chiorazzi (Manhasset); Oliver Cornely (Köln); Michel Delforge (Leuven); Ruud Delwel (Rotterdam); Meletios A. Dimopoulos (Athens); Inderjeet Dokal (London); Hervé Dombret (Paris); Peter Dreger (Hamburg); Martin Dreyling (München); Kieron Dunleavy (Bethesda); Dimitar Efremov (Rome); Sabine Eichinger (Vienna); Jean Feuillard (Limoges); Carlo Gambacorti-Passerini (Monza); Guillermo Garcia Manero (Houston); Christian Geisler (Copenhagen); Piero Giordano (Leiden); Christian Gisselbrecht (Paris); Andreas Greinacher (Greifswals); Hildegard Greinix (Vienna); Paolo Gresele (Perugia); Thomas M. Habermann (Rochester); Claudia Haferlach (München); Oliver Hantschel (Lausanne); Christine Harrison (Southampton); Brian Huntly (Cambridge); Ulrich Jaeger (Vienna); Elaine Jaffe (Bethesda); Arnon Kater (Amsterdam); Gregory Kato (Pittsburg); Christoph Klein (Munich); Steven Knapper (Cardiff); Seiji Kojima (Nagoya); John Koreth (Boston); Robert Kralovics (Vienna); Ralf Küppers (Essen); Ola Landgren (New York); Peter Lenting (Le Kremlin-Bicetre); Per Ljungman (Stockholm); Henk M. Lokhorst (Utrecht); John Mascarenhas (New York); Maria-Victoria Mateos (Salamanca); Giampaolo Merlini (Pavia); Anna Rita Migliaccio (New York); Mohamad Mohty (Nantes); Martina Muckenthaler (Heidelberg); Ann Mullally (Boston); Stephen Mulligan (Sydney); German Ott (Stuttgart); Jakob Passweg (Basel); Melanie Percy (Ireland); Rob Pieters (Utrecht); Stefano Pileri (Milan); Miguel Piris (Madrid); Andreas Reiter (Mannheim); Jose-Maria Ribera (Barcelona); Stefano Rivella (New York); Francesco Rodeghiero (Vicenza); Richard Rosenquist (Uppsala); Simon Rule (Plymouth); Claudia Scholl (Heidelberg); Martin Schrappe (Kiel); Radek C. Skoda (Basel); Gérard Socié (Paris); Kostas Stamatopoulos (Thessaloniki); David P. Steensma (Rochester); Martin H. Steinberg (Boston); Ali Taher (Beirut); Evangelos Terpos (Athens); Takanori Teshima (Sapporo); Pieter Van Vlierberghe (Gent); Alessandro M. Vannucchi (Firenze); George Vassiliou (Cambridge); Edo Vellenga (Groningen); Umberto Vitolo (Torino); Guenter Weiss (Innsbruck).

Editorial Office Simona Giri (Production & Marketing Manager), Lorella Ripari (Peer Review Manager), Paola Cariati (Senior Graphic Designer), Igor Ebuli Poletti (Senior Graphic Designer), Marta Fossati (Peer Review), Diana Serena Ravera (Peer Review)

Affiliated Scientific Societies SIE (Italian Society of Hematology, www.siematologia.it) SIES (Italian Society of Experimental Hematology, www.siesonline.it)

haematologica Journal of the Ferrata Storti Foundation

Information for readers, authors and subscribers Haematologica (print edition, pISSN 0390-6078, eISSN 1592-8721) publishes peer-reviewed papers on all areas of experimental and clinical hematology. The journal is owned by a non-profit organization, the Ferrata Storti Foundation, and serves the scientific community following the recommendations of the World Association of Medical Editors (www.wame.org) and the International Committee of Medical Journal Editors (www.icmje.org). Haematologica publishes editorials, research articles, review articles, guideline articles and letters. Manuscripts should be prepared according to our guidelines (www.haematologica.org/information-for-authors), and the Uniform Requirements for Manuscripts Submitted to Biomedical Journals, prepared by the International Committee of Medical Journal Editors (www.icmje.org). Manuscripts should be submitted online at http://www.haematologica.org/. Conflict of interests. According to the International Committee of Medical Journal Editors (http://www.icmje.org/#conflicts), “Public trust in the peer review process and the credibility of published articles depend in part on how well conflict of interest is handled during writing, peer review, and editorial decision making”. The ad hoc journal’s policy is reported in detail online (www.haematologica.org/content/policies). Transfer of Copyright and Permission to Reproduce Parts of Published Papers. Authors will grant copyright of their articles to the Ferrata Storti Foundation. No formal permission will be required to reproduce parts (tables or illustrations) of published papers, provided the source is quoted appropriately and reproduction has no commercial intent. Reproductions with commercial intent will require written permission and payment of royalties. Detailed information about subscriptions is available online at www.haematologica.org. Haematologica is an open access journal. Access to the online journal is free. Use of the Haematologica App (available on the App Store and on Google Play) is free. For subscriptions to the printed issue of the journal, please contact: Haematologica Office, via Giuseppe Belli 4, 27100 Pavia, Italy (phone +39.0382.27129, fax +39.0382.394705, E-mail: info@haematologica.org). Rates of the International edition for the year 2021 are as following: Print edition

Institutional Euro 700

Personal Euro 170

Advertisements. Contact the Advertising Manager, Haematologica Office, via Giuseppe Belli 4, 27100 Pavia, Italy (phone +39.0382.27129, fax +39.0382.394705, e-mail: marketing@haematologica.org). Disclaimer. Whilst every effort is made by the publishers and the editorial board to see that no inaccurate or misleading data, opinion or statement appears in this journal, they wish to make it clear that the data and opinions appearing in the articles or advertisements herein are the responsibility of the contributor or advisor concerned. Accordingly, the publisher, the editorial board and their respective employees, officers and agents accept no liability whatsoever for the consequences of any inaccurate or misleading data, opinion or statement. Whilst all due care is taken to ensure that drug doses and other quantities are presented accurately, readers are advised that new methods and techniques involving drug usage, and described within this journal, should only be followed in conjunction with the drug manufacturer’s own published literature. Direttore responsabile: Prof. Carlo Balduini; Autorizzazione del Tribunale di Pavia n. 63 del 5 marzo 1955. Printing: Press Up, zona Via Cassia Km 36, 300 Zona Ind.le Settevene - 01036 Nepi (VT)

Associated with USPI, Unione Stampa Periodica Italiana. Premiato per l’alto valore culturale dal Ministero dei Beni Culturali ed Ambientali

haematologica Journal of the Ferrata Storti Foundation

Table of Contents Volume 106, Issue 2: February 2021 About the Cover 323

Images from the Haematologica Atlas of Hematologic Cytology: bone marrow metastases from gastric adenocarcinoma Rosangela Invernizzi https://doi.org/10.3324/haematol.2020.275529

Editorials 324

Concurrent inhibition of IDH and methyltransferase maximizes therapeutic efficacy in IDH mutant acute myeloid leukemia Zhihong Zeng and Marina Konopleva https://doi.org/10.3324/haematol.2020.266809

326

Ironing out an approach to alleviate the hypoferremia of acute inflammation Karin E. Finberg https://doi.org/10.3324/haematol.2020.266627

329

Chromosome Y loss and drivers of clonal hematopoiesis in myelodysplastic syndrome Panagiotis Baliakas and Lars A. Forsberg https://doi.org/10.3324/haematol.2020.266601

332

Central nervous system prophylaxis in diffuse large B-cell lymphoma: a race to the bottom Mark Roschewski https://doi.org/10.3324/haematol.2020.266635

335

An early start of Coup-TFII promotes Îł-globin gene expression in adult erythroid cells Stefania Bottardi and Eric Milot

https://doi.org/10.3324/haematol.2020.266791

Review Articles 337

Inherited platelet diseases with normal platelet count: phenotypes, genotypes and diagnostic strategy Paquita Nurden et al. https://doi.org/10.3324/haematol.2020.248153

351

Biochemical, molecular and clinical aspects of coagulation factor VII and its role in hemostasis and thrombosis Francesco Bernardi and Guglielmo Mariani https://doi.org/10.3324/haematol.2020.248542

Articles Transplantation

363

Standardized monitoring of cytomegalovirus-specific immunity can improve risk stratification of recurrent cytomegalovirus reactivation after hematopoietic stem cell transplantation Eva Wagner-Drouet et al. https://doi.org/10.3324/haematol.2019.229252

375

Autologous stem cell transplantation for progressive systemic sclerosis: a prospective non-interventional study from the European Society for Blood and Marrow Transplantation Autoimmune Disease Working Party Joerg Henes et al. https://doi.org/10.3324/haematol.2019.230128

Iron Metabolism & its Disorders

384

Induction of erythroferrone in healthy humans by micro-dose recombinant erythropoietin or high-altitude exposure Paul Robach et al. https://doi.org/10.3324/haematol.2019.233874

Haematologica 2021; vol. 106 no. 2 - February 2021 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

391

C-FGF23 peptide alleviates hypoferremia during acute inflammation Rafiou Agoro et al. https://doi.org/10.3324/haematol.2019.237040

Hematopoiesis

404

A gain-of-function RAC2 mutation is associated with bone marrow hypoplasia and an autosomal dominant form of severe combined immunodeficiency Chantal Lagresle-Peyrou et al. https://doi.org/10.3324/haematol.2019.230250

412

MicroRNA-21 maintains hematopoietic stem cell homeostasis through sustaining the nuclear factor-ÎşB signaling pathway in mice Mengjia Hu et al. https://doi.org/10.3324/haematol.2019.236927

424

Signal-transducing adaptor protein-2 delays recovery of B-lineage lymphocytes during hematopoietic stress Michiko Ichii et al. https://doi.org/10.3324/haematol.2019.225573

Gaucher Disease

437

Accuracy of chitotriosidase activity and CCL18 concentration in assessing type I Gaucher disease severity. A systematic review with meta-analysis of individual participant data Tatiana Raskovalova et al. https://doi.org/10.3324/haematol.2019.236083

Complications in Hematology

446

Predicting sinusoidal obstruction syndrome after allogeneic stem cell transplantation with the EASIX biomarker panel Sihe Jiang et al. https://doi.org/10.3324/haematol.2019.238790

Leukocyte Biology

454

Alternative activation of human macrophages enhances tissue factor expression and production of extracellular vesicles Philipp J. Hohensinner et al. https://doi.org/10.3324/haematol.2019.220210

Red Cell Biology & its Disorders

464

Hemoglobin switching in mice carrying the Klf1Nan variant Anne Korporaal et al. https://doi.org/10.3324/haematol.2019.239830

474

The Coup-TFII orphan nuclear receptor is an activator of the Îł-globin gene Cristina Fugazza et al.

https://doi.org/10.3324/haematol.2019.241224

Chronic Myeloid Leukemia

483

Targeting of plasminogen activator inhibitor-1 activity promotes elimination of chronic myeloid leukemia stem cells Takashi Yahata et al. https://doi.org/10.3324/haematol.2019.230227

Non-Hodgkin Lymphoma

495

Ataxia-telangiectasia mutated interacts with Parkin and induces mitophagy independent of kinase activity. Evidence from mantle cell lymphoma Aloke Sarkar et al. https://doi.org/10.3324/haematol.2019.234385

513

Cell free circulating tumor DNA in cerebrospinal fluid detects and monitors central nervous system involvement of B-cell lymphomas Sabela Bobillo et al. https://doi.org/10.3324/haematol.2019.241208

Haematologica 2021; vol. 106 no. 2 - February 2021 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

Coagulation & its Disorders

522

uPA-mediated plasminogen activation is enhanced by polyphosphate Claire S. Whyte and Nicola J. Mutch https://doi.org/10.3324/haematol.2019.237966

Acute Lymphoblastic Leukemia

532

Clinical significance of soluble CADM1 as a novel marker for adult T-cell leukemia/lymphoma Shingo Nakahata et al. https://doi.org/10.3324/haematol.2019.234096

Chronic Lymphocytic Leukemia

543

Bendamustine followed by ofatumumab and ibrutinib in chronic lymphocytic leukemia: primary endpoint analysis of a multicenter, open-label, phase II trial (CLL2-BIO) Paula Cramer et al. https://doi.org/10.3324/haematol.2019.223693

Myelodysplastic Syndromes

555

Genomic alterations in patients with somatic loss of the Y chromosome as the sole cytogenetic finding in bone marrow cells Madhu M. Ouseph et al. https://doi.org/10.3324/haematol.2019.240689

Acute Myeloid Leukemia

565

Synergistic activity of IDH1 inhibitor BAY1436032 with azacitidine in IDH1 mutant acute myeloid leukemia Anuhar Chaturvedi et al. https://doi.org/10.3324/haematol.2019.236992

Letters to the Editor 574

Novel L-nucleoside analog, 5-fluorotroxacitabine, displays potent efficacy against acute myeloid leukemia Aniket Bankar et al. https://doi.org/10.3324/haematol.2019.226795

580

Wip1 regulates hematopoietic stem cell development in the mouse embryo Wenyan He et al. https://doi.org/10.3324/haematol.2019.235481

585

CD4+ T-cell alloreactivity after haploidentical hematopoietic stem cell transplantation Burak Kalin et al. https://doi.org/10.3324/haematol.2019.244152

589

Daratumumab inhibits acute myeloid leukemia metabolic capacity by blocking mitochondrial transfer from mesenchymal stromal cells Jayna J. Mistry et al. https://doi.org/10.3324/haematol.2019.242974

593

597

601

Heterogenous mutation spectrum and deregulated cellular pathways in aberrant plasma cells underline molecular pathology of light-chain amyloidosis Zuzana Chyra et al. https://doi.org/10.3324/haematol.2019.239756

Haematologica 2021; vol. 106 no. 2 - February 2021 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation 605

Selectively targeting FLT3-ITD mutants over FLT3-wt by a novel inhibitor for acute myeloid leukemia Aoli Wang, et al. https://doi.org/10.3324/haematol.2019.244186

610

miR-939 acts as tumor suppressor by modulating JUNB transcriptional activity in pediatric anaplastic large cell lymphoma Anna Garbin et al. https://doi.org/10.3324/haematol.2019.241307

614

Myeloid/lymphoid neoplasms with eosinophilia/basophilia and ETV6-ABL1 fusion: cell-of-origin and response to tyrosine kinase inhibition Jinjuan Yao et al. https://doi.org/10.3324/haematol.2020.249649

619

SMAD1 promoter hypermethylation and lack of SMAD1 expression in Hodgkin lymphoma: a potential target for hypomethylating drug therapy Magdalena M. Gerlach et al. https://doi.org/10.3324/haematol.2020.249276

622

Efficacy of recombinant erythropoietin in autoimmune hemolytic anemia: a multicenter international study Bruno Fattizzo et al. https://doi.org/10.3324/haematol.2020.250522

626

Heme induces human and mouse platelet activation through C-type-lectin-like receptor-2 Joshua H. Bourne et al. https://doi.org/10.3324/haematol.2019.246488

630

The clinical and biological characteristics of NUP98-KDM5A pediatric acute myeloid leukemia Sanne Noort et al. https://doi.org/10.3324/haematol.2019.236745

635

Down syndrome-related transient abnormal myelopoiesis is attributed to a specific erythro-megakaryocytic subpopulation with GATA1 mutation Yoko Nishinaka-Arai et al. https://doi.org/10.3324/haematol.2019.242693

Case Reports 641

Emapalumab treatment in an ADA-SCID patient with refractory hemophagocytic lymphohistiocytosis-related graft failure and disseminated bacillus Calmette-GuĂŠrin infection Francesca Tucci et al. https://doi.org/10.3324/haematol.2020.255620

647

Detectable mutations precede late myeloid neoplasia in aplastic anemia Bhavisha A. Patel et al. https://doi.org/10.3324/haematol.2020.263046

651

A deep molecular response of splenic marginal zone lymphoma to front-line checkpoint blockade Peter G. Miller et al. https://doi.org/10.3324/haematol.2020.258426

Haematologica 2021; vol. 106 no. 2 - February 2021 http://www.haematologica.org/

ABOUT THEâ&#x20AC;&#x2C6;COVER Images from the Haematologica Atlas of Hematologic Cytology: bone marrow metastases from gastric adenocarcinoma Rosangela Invernizzi University of Pavia, Pavia, Italy E-mail: ROSANGELA INVERNIZZI - rosangela.invernizzi@unipv.it doi:10.3324/haematol.2020.275529

n some patients with disseminated gastric adenocarcinoma, microangiopathic hemolytic anemia is the presenting feature of the tumor. In this case of gastric adenocarcinoma with bone marrow metastases, an erythroblast, polychromatic red cells, microspherocytes and red cell fragmentation are observed in the peripheral blood smear (panels A and B). Bone marrow smear reveals a group of large tumor cells with irregular nuclear shape, and polychromatic, inhomogeneous, foamy or vacuolated cytoplasm (panels C and D). Periodic acid-Schiff (PAS) stain shows intense granular or diffuse cytoplasmic positivity suggesting the accumulation of polysaccharides in the neoplastic cells (panel E). The strong immunocytochemical expression of cytokeratins (immunoperoxidase technique), in association with the absence of the common leukocyte antigen, confirms the non-hematopoietic origin of tumor cells (panel F).1

References 1. Invernizzi R. Metastases of solid tumors. Haematologica. 2020; 105(Suppl 1):261-269.

haematologica | 2021; 106(2)

323

EDITORIALS Concurrent inhibition of IDH and methyltransferase maximizes therapeutic efficacy in IDH mutant acute myeloid leukemia Zhihong Zeng1 and Marina Konopleva1,2 1

Department of Leukemia, The University of Texas MD Anderson Cancer Center and 2Department of Stem Cell Transplantation and Cellular Therapy, The University of Texas MD Anderson Cancer Center, Houston, TX, USA E-mail: MARINA KONOPLEVA - mkonople@mdanderson.org doi:10.3324/haematol.2020.266809

ith findings of the molecular mechanisms of cotargeting isocitrate dehydrogenase (IDH) and methyltransferase with IDH inhibitors and a hypomethylating agent (HMA), Chaturvedi et al.1 provide new insights into combination therapy for IDH mutant acute myeloid leukemia (AML). Their manuscript entitled “Synergistic activity of IDH1 inhibitor BAY1436032 with azacitidine in IDH1 mutant acute myeloid leukemia,” published in this issue of Haematologica, reports that the concurrent administration of an IDH inhibitor and an HMA maximizes antileukemia efficacy in vitro and in vivo. IDH is an enzyme that catalyzes conversion of isocitrate to α-ketoglutarate (α-KG); its metabolism plays an essential role in balancing cellular energy. IDH1 and IDH2 are 2 of the 3 isoforms identified in humans; mutations in both isoforms are found in several malignancies, including AML. Mutant IDH1/2 converts α-KG to 2-hydroxyglutarate (2-HG), a metabolite structurally similar to α-KG, competitively inhibits α-KG–dependent enzymes, ultimately alters DNA and histone methylation, and impairs cellular growth and differentiation.2,3 Recurrent IDH1/2 mutations occur in up to 20% of patients with AML.4,5 In IDH mutant AML, overproduction of 2-HG leads to DNA and histone hypermethylation and myeloid cell differentiation arrest.6 Several small-molecule inhibitors of mutant IDH1 and IDH2 have been developed and are progressing through preclinical and clinical development.7,8 Two of them, ivosidenib and enasidenib, are approved by the Food and Drug Administration (FDA) for the treatment of newly diagnosed and relapsed/refractory IDH1 or IDH2 mutant AML, respectively. BAY1436032 is an oral, pan-mutant IDH1 inhibitor characterized by researchers in this study.9 BAY1436032 has shown strong antileukemic activity in two separate IDH1 mutant AML xenograft mouse models. This compound is currently undergoing safety/efficacy testing in phase I dose escalation trials in patients with relapsed IDH1 mutant AML (clinicaltrials gov. Indentifier: 03127735) and advanced solid tumors (clinicaltrials gov. Indentifier: 02746081).10 Despite high efficacy and on-target activity, IDH-targeted monotherapy in AML offers response rates of less than 50%. Resistance mechanisms, such as co-occurring mutations in receptor tyrosine kinase signaling (FLT3, PTPN11, RAS, KIT), transcription factors (RUNX1, GATA2, CEBPA) and restoration of 2-HG through mutations in other IDH protein or IDH1 second site mutations, may account for the therapeutic failure; therefore, combination regimens are essential to improve clinical responses.11,12 The combination of an IDH1/2 inhibitor (ivosidenib or enasidenib) with the HMA azacitidine is being evaluated in an ongoing phase Ib/II study in patients with newly diagnosed IDH mutant AML who are ineligible for intensive chemotherapy (clinicaltrials gov. Indentifier: 02677922). This combination was well tolerated, with a safety profile consistent with that of ivosidenib or azacitidine monotherapy. 324

Clinical response rates exceeded those of azacitidine alone, and importantly, most responders achieved IDH1 mutation clearance.13 Based on these findings, a phase III study of ivosidenib and azacitidine is actively enrolling patients (clinicaltrials gov. Indentifier: 03173248). Despite these encouraging clinical responses, the molecular mechanisms of the interaction between IDH inhibitors and HMA are not well understood. In this issue of Haematologica, Chaturvedi et al.1 investigated the efficacy and elucidated the mechanism of action of the novel IDH1 inhibitor BAY1436032 combined with azacitidine in both in vitro and in vivo preclinical models of IDH1 mutant AML. The authors observed that ex vivo treatment with BAY1436032 and azacitidine is more effective than singleagent treatment in its ability to induce the cell cycle S phase block, resulting in synergistic inhibition of colony formation in primary IDH1 mutant AML. Using two human IDH1 mutant AML xenograft models, the team evaluated the efficacy of BAY1436032 and azacitidine as single agents and in combination, with sequential (azacitidine followed by BAY1436032) or concurrent applications in vivo. Combination therapy induced differentiation and significantly prolonged survival compared to single agent or control cohorts. Importantly, concurrent administration of these agents, similar to the design implemented in the ongoing clinical trials, produced the highest efficacy; this finding was further confirmed in a secondary transplantation assay indicating that the concurrent combination therapy elicits the greatest reduction in the number of leukemia stem cells (LSC). Using transcriptomic (RNA sequencing) and epigenomic (DNA methylation arrays) analyses, the authors explored the mechanisms underlying this additive/synergistic efficacy, studying AML cells collected in vivo. Combination therapy decreased the expression of LSC gene sets and suppressed transcriptional factors in MAP kinase (RAS/RAF) and retinoblastoma/E2F (RB/E2F) pathways. Quantitative RTPCR analysis confirmed that the cell survival/proliferation genes ELK1, ETS1, and CCND1 in the MAP kinase pathway and E2F1, CCNA2, and CCNE1 in RB/E2F signaling were additively suppressed, while the myeloid differentiation genes PU.1, CEBPA, and GABPA were upregulated in cells retrieved from the combination therapy arm. Using an IDH1mutant fibrosarcoma cell line, the authors evaluated these findings at the translational level: the concurrently administered combination dephosphorylated ERK1/2 and downregulated the downstream targets of ELK1, ETS1 and CYCLIN D1. The lack of CYCLIN D1 limited CYCLIN D-CDK4 complex formation and consequently inhibited RB phosphorylation on serine 795 and 807/811, thereby preventing RB from releasing E2F to regulate cell cycle G1 to S transition (Figure 1). Correspondingly, directly targeting the MAP kinase pathway with a MEK1/2 inhibitor, trametinib, or blocking the cell cycle with a CDK4/6 inhibitor, abemaciclib, more effectively haematologica | 2021; 106(2)

Editorials

Figure 1. Schema of the molecular mechanisms of cotargeting IDH and methyltransferase in IDH mutant acute myeloid leukemia. The concurrently administered combination of the isocitrate dehydrogenase (IDH) inhibitor BAY1436032 (BAY) and the hypomethylating agent azacitidine (AZA) synergistically inhibits RAS/RAF/ERK1/2 and its downstream targets ELK1, ETS1, and CCND1 and blocks complex formation of CYCLIN D1/CDK4 to prevent RB phosphorylation, consequently inhibiting E2F release from the RB/E2F complex to promote cell cycle transition from G1 to S, leading to suppression of cell proliferation and self-renewal. In parallel, the combination of BAY and AZA upregulates the myeloid differentiation transcription factors PU.1, CEBPA, and GABPA to promote cell differentiation.

inhibited proliferation in IDH mutant AML than in IDH wild-type AML. These data support the synergistic activity of BAY1436032 and azacitidine and identify inhibition of the MAPK/RB pathways and activation of differentiation-related transcriptional factors as the key mechanisms underlying this synergism in IDH mutant AML. Chaturvedi et al.â&#x20AC;&#x2122;s1 work is important for several reasons. It is highly relevant to the ongoing clinical trials exploring combinations of IDH inhibitors and HMA, which have reported encouraging initial findings in patients with IDH mutant AML.13 These trials are designed to concurrently administer these agents, which was also confirmed in this study as being most efficacious. Furthermore, this work not only advances our understanding of the molecular regulators affected by cotargeting IDH and methyltransferase, but also hints at the complexity of the inhibitory mechanisms that can be impacted by administration sequence and contribute to differential outcomes. The authors identified the MAPK/ERKCYCLIN D1/CDK4-RB/E2F axis as critical to the regulation of LSC proliferation and the response to concurrent BAY1436032 and azacitidine. Whether this crosstalk between MAPK/ERK and RB/E2F signaling is intrinsic to IDH mutant AML, how it is associated with the terminal differentiation of LSC, and whether it can be utilized as a biomarker to predict outcomes are questions worthy of future exploration. Notably, despite its striking antileukemic activity in IDH mutant AML, the authors haematologica | 2021; 106(2)

indicated that concurrent BAY1436032 and azacitidine failed to completely eliminate LSC. Similarly, in ongoing trials, a small fraction of patients was primary refractory or experienced AML relapse while being treated with the combination regimen of IDH inhibitor and HMA. Hypotheses to explain treatment insensitivity include incomplete mutation clearance, polyclonal resistance, and/or clonal expansion associated with activation of multiple kinases. Clearly, identification of the molecular determinants of primary and adaptive resistance is essential to refine the future therapeutic strategy. Given the genetic complexity and heterogeneity of AML, future large cohort studies and personalized molecular profiling at the singlecell level are needed to identify optimal therapeutic combinations, aiming to achieve a curative response in AML patients carrying IDH mutations. Disclosures No conflicts of interest to disclose. Contributions ZZ and MK wrote and edited the manuscript.

References 1. Chaturvedi A, Gupta C, Gabdoulline R, et al. Synergistic activity of IDH1 inhibitor BAY1436032 with azacitidine in IDH1 mutant acute myeloid leukemia. Haematologica. 2020;106(2):565-573. 2. Waitkus MS, Diplas BH, Yan H. Biological role and therapeutic

325

Editorials

potential of IDH mutations in cancer. Cancer Cell. 2018;34(2):186195. 3. Clark O, Yen K, Mellinghoff IK. Molecular pathways: isocitrate dehydrogenase mutations in cancer. Clin Cancer Res. 2016;22(8):1837-1842. 4. Cancer Genome Atlas Research N, Ley TJ, Miller C, et al. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med. 2013;368(22):2059-2074. 5. Metzeler KH, Herold T, Rothenberg-Thurley M, et al. Spectrum and prognostic relevance of driver gene mutations in acute myeloid leukemia. Blood. 2016;128(5):686-698. 6. Medeiros BC, Fathi AT, DiNardo CD, Pollyea DA, Chan SM, Swords R. Isocitrate dehydrogenase mutations in myeloid malignancies. Leukemia. 2017;31(2):272-281. 7. Ragon BK, DiNardo CD. Targeting IDH1 and IDH2 mutations in acute myeloid leukemia. Curr Hematol Malig Rep. 2017;12(6):537546. 8. Golub D, Iyengar N, Dogra S, et al. Mutant isocitrate dehydrogenase

inhibitors as targeted cancer therapeutics. Front Oncol. 2019;9:417. 9. Chaturvedi A, Herbst L, Pusch S, et al. Pan-mutant-IDH1 inhibitor BAY1436032 is highly effective against human IDH1 mutant acute myeloid leukemia in vivo. Leukemia. 2017;31(10):2020-2028. 10. Heuser M, Palmisiano N, Mantzaris I, et al. Safety and efficacy of BAY1436032 in IDH1-mutant AML: phase I study results. Leukemia. 2020;34(11):2903-2913. 11. Intlekofer AM, Shih AH, Wang B, et al. Acquired resistance to IDH inhibition through trans or cis dimer-interface mutations. Nature. 2018;559(7712):125-129. 12. Choe S, Wang H, DiNardo CD, et al. Molecular mechanisms mediating relapse following ivosidenib monotherapy in IDH1-mutant relapsed or refractory AML. Blood Adv. 2020;4(9):1894-1905. 13. Dinardo CD, Stein AS, Stein EM, et al. Mutant IDH (mIDH) inhibitors, ivosidenib or enasidenib, with azacitidine (AZA) in patients with acute myeloid leukemia (AML). J Clin Oncol. 2018;36(15 Suppl):S7042.

Ironing out an approach to alleviate the hypoferremia of acute inflammation Karin E. Finberg Yale School of Medicine, New Haven, CT, USA E-mail: KARIN E. FINBERG - karin.finberg@yale.edu doi:10.3324/haematol.2020.266627

n the steady state, iron levels in the plasma are regulated by the recycling of iron from senescent red blood cells by macrophages of the reticuloendothelial system. In systemic infections and inflammatory states, perturbation of this process can result in hypoferremia (a decrease in circulating iron levels), which may represent a host defense mechanism to limit iron availability to pathogens.1 Because hypoferremia restricts the availability of iron to erythroid precursors, if sustained, it contributes to the development of the anemia of inflammation. In this issue of Haematologica, Agoro et al.2 report that the acute hypoferremic response to lipopolysaccharide (LPS), a major component of the outer membrane of Gram-negative bacteria, is modulated in mice by pretreatment with a truncated, C-terminal fragment of the hormone fibroblast growth factor 23. During inflammatory states, iron is sequestered in cells due to a reduction in activity of ferroportin, the major cellular iron exporter that is expressed by multiple cell types, including macrophages.3 Activity of the ferroportin transporter on cell membranes is regulated by hepcidin, the key iron regulatory hormone synthesized primarily by the liver; hepcidin occludes the ferroportin transporter and triggers its endocytosis and degradation.4 Hepcidin expression is induced in response to several proinflammatory cytokines, including interleukin-6 (IL6) and IL-1Î˛, as well as LPS.1 In addition to its post-translational regulation by hepcidin, ferroportin expression is regulated at the mRNA level by inflammatory stimuli. In macrophages, stimulation of toll-like receptor 4 (TLR4), a member of the pattern recognition receptor family, with LPS suppresses ferroportin mRNA and protein levels and also induces hepcidin expression.5 Stimulation of TLR2 and TLR6 also promotes ferroportin downregulation in a hepcidin-independent manner.6 In humans, LPS injection induces an acute rise in plasma cytokines such 326

as IL-6 and tumor necrosis factor (TNF), which is followed by hepcidin elevation, and ultimately a reduction in serum iron levels.7 Recent studies have suggested intriguing crosstalk between inflammation, iron homeostasis, and fibroblast growth factor 23 (FGF23),8 a hormone that functions as a key regulator of phosphate and calcium homeostasis. FGF23 in the circulation appears to be mainly derived from bone, although FGF23 expression has also been detected in other tissues.9 Physiological actions of FGF23 are mediated through fibroblast growth factor receptors (FGFR) and by the co-receptor Klotho, which increases affinity of FGF23 for FGFR and is required for the hormone's ability to promote renal phosphate excretion.10 The phosphaturic activity of the mature, biologicallyactive FGF23 peptide can be abrogated by proteolytic cleavage at an RXXR motif located at the boundary between the FGF core homology domain and the 72amino-acid C-terminal portion of FGF23.9 The isolated Cterminal FGF23 fragment (referred to here as C-FGF23) competes with full-length FGF23 for binding to the FGFRKlotho complex, thereby impairing FGF23 signaling; accordingly, in healthy rodents, C-FGF23 administration inhibits renal phosphate excretion and induces hyperphosphatemia.11 Circulating FGF23 levels are markedly elevated in chronic kidney disease,9 a condition in which disruption of systemic iron homeostasis, mediated by factors such as inflammation, therapy-related iron losses, decreased glomerular filtration rate, low serum erythropoietin (EPO), and elevated serum hepcidin, contributes to the development of anemia.12 Both iron deficiency and inflammation stimulate the production of FGF23 as well as its proteolytic cleavage.8 Interestingly, in mice with established chronic kidney disease (induced by subtotal nephrectomy), a single dose of C-FGF23 induced acute haematologica | 2021; 106(2)

Editorials

Figure 1. Physiological responses to lipopolysaccharide (LPS), an inducer of acute inflammation, are modulated in mice by pretreatment with the C-terminal portion of the endocrine hormone FGF23 (C-FGF23). The attenuated responses involve parameters that are relevant to systemic iron balance and the development of the anemia of inflammation.

alterations in several physiological parameters relevant to iron homeostasis and erythropoiesis.13 In the present study, Agoro et al.2 considered the ability of C-FGF23 administration to modulate iron balance in a model of acute systemic inflammation. First, the authors characterized the detailed time course of early physiological responses in wild-type mice subjected to a single, sublethal dose of LPS. As expected, compared to vehicletreated control mice, mice treated with LPS showed transient elevations in hepatic levels of mRNAs encoding proinflammatory cytokines, which peaked one hour after injection. These increases in inflammatory markers were soon followed by anticipated changes in iron-related parameters, including an elevation of hepatic hepcidin mRNA, a reduction in serum iron, a reduction in ferroportin mRNA in liver and spleen, and a rise in splenic iron content. Additionally, compatible with prior studies indicating that FGF23 is regulated by inflammation, transient elevations of Fgf23 mRNA were observed in bone, bone marrow, liver, and spleen, and were accompanied by a transient rise in FGF23 protein in the circulation. Next, the authors examined the ability of C-FGF23 pretreatment to modulate these acute inflammatory and ironrelated physiological responses in LPS-treated mice. Mice received an intraperitoneal injection of C-FGF23 (or vehicle control) 8 hours prior to challenge with an intraperitoneal injection of LPS (or vehicle control); all mice were analyzed 4 hours after the challenge. LPS-mediated induction of hepatic Tnf and Hamp (hepcidin) mRNA, as well as induction of hyperhepcidinemia, was blunted by C-FGF23 pretreatment (Figure 1). Additionally, C-FGF23 pretreatment attenuated the ability of LPS to induce hypoferremia and to increase tissue iron levels in liver and spleen. haematologica | 2021; 106(2)

While changes in iron trafficking in response to systemic immune activation contribute significantly to the development of the anemia of inflammation, inflammatory suppression of erythropoietic activity also plays a major role.1 Notably, LPS-treatment has been shown to suppress renal Epo mRNA levels in rodents.14 Interestingly, Agoro and colleagues found that C-FGF23 pretreatment alleviated LPS-mediated suppression of mRNAs encoding EPO and the EPO receptor in the kidney and also in liver and spleen. C-FGF23 pretreatment also raised serum EPO levels in LPS-treated mice. Additionally, the ability of LPS to induce Fgf23 mRNA in bone and FGF23 protein in the circulation was attenuated by C-FGF23 pretreatment. By demonstrating a novel link between C-FGF23 and systemic iron modulation in the acute inflammatory setting, the study of Agoro et al.2 raises a number of questions to stimulate future investigation. First, what are the physiological mechanisms by which C-FGF23 pretreatment alters LPS-induced hypoferremia and attenuates iron storage in the liver and spleen? Although hepcidin induction by LPS was blunted in mice pretreated with C-FGF23, serum hepcidin levels nevertheless remained markedly elevated compared to control groups. Furthermore, the suppression of hepatic and splenic ferroportin mRNA in response to LPS was not alleviated by C-FGF23 pretreatment. Whether C-FGF23 has direct actions on the duodenum, the major site of dietary iron absorption in the intestine, remains to be explored. The underlying mechanism by which C-FGF23 modulates LPS-mediated hepcidin induction also requires further clarification. Interestingly, this effect was not explained by a reduction in hepatic expression of IL-6, the most established cytokine regulator of hepcidin in 327

Editorials

response to inflammation,15 nor did C-FGF23 pretreatment reduce hepatic levels of phosphorylated STAT3, the major intracellular mediator of hepcidin induction by inflammatory cytokines.1 However, interestingly, in livers of LPStreated mice, C-FGF23 pretreatment lowered levels of mRNA encoding BMP6, a key ligand for a major signaling pathway promoting hepcidin transcription.16 Additionally, in LPS-treated mice, C-FGF23 pretreatment raised splenic mRNA levels of erythroferrone, a negative regulator of hepcidin expression that is synthesized by erythroblasts.17 Another area to be clarified is whether the observed effects of C-FGF23 pretreatment on gene expression, as well as on iron storage, in liver and spleen are mediated through the ability of C-FGF23 to act as a competitive inhibitor of FGFR expressed by these organs. Indeed, although the co-receptor Klotho is expressed primarily in the kidney, parathyroid gland, and brain, recent work suggests that under conditions of FGF23 elevation the FGF23 hormone can participate in FGFR-dependent signaling mechanisms in tissues lacking Klotho.10 Another possibility is that C-FGF23 may mediate FGFR-independent biological actions that either directly or indirectly impact iron regulatory pathways. Because the gene expression responses found to be modulated by C-FGF23 were measured in organs of heterogenous cellular composition, such as liver and spleen, it will also be valuable to clarify the participating cell types. It is also worth considering the fact that the FGF23 Cterminal peptide introduced into mice in this study corresponds to the human peptide sequence; the murine and humans forms of FGF23 share 64% amino acid homology across this portion of the protein. To exclude the possibility that some observed responses to human C-FGF23 in mice reflect more general immunomodulatory effects that occur due to the introduction of a foreign peptide, it will be valuable to clarify if pretreatment with murine CFGF23 produces similar effects on iron homeostasis in LPS-treated mice. In terms of clinical relevance, the ability of C-FGF23 to attenuate the hypoferremia, hepatic and splenic iron sequestration, and renal suppression of Epo mRNA observed 4 hours after LPS injection raises the intriguing possibility that C-FGF23-based approaches may have relevance for treatment of anemia of inflammation. Future studies in which the ability of C-FGF23 to modulate these physiological parameters is assessed longitudinally after LPS injection will be informative. Additionally, it will be of great interest to determine if long-term administration of C-FGF23 produces beneficial effects on iron

328

metabolism and erythropoiesis in animal models in which anemia of inflammation has already been achieved through chronic exposure to infectious or noninfectious stimuli.18 It will also be of interest to characterize the effects of chronic C-FGF23 administration on phosphate and calcium homeostasis in animal models of the anemia of inflammation. Disclosures No conflicts of interest to disclose.

References 1. Weiss G, Ganz T, Goodnough LT. Anemia of inflammation. Blood. 2019;133(1):40-50. 2. Agoro R, Park MY, Le Henaff C, et al. C-FGF23 peptide alleviates hypoferremia during acute inflammation. Haematologica. 2020;106(2):391-403. 3. Drakesmith H, Nemeth E, Ganz T. Ironing out ferroportin. Cell Metab. 2015;22(5):777-787. 4. Camaschella C, Nai A, Silvestri L. Iron metabolism and iron disorders revisited in the hepcidin era. Haematologica. 2020;105(2):260-272. 5. Peyssonnaux C, Zinkernagel AS, Datta V, Lauth X, Johnson RS, Nizet V. TLR4-dependent hepcidin expression by myeloid cells in response to bacterial pathogens. Blood. 2006;107(9):3727-3732. 6. Guida C, Altamura S, Klein FA, et al. A novel inflammatory pathway mediating rapid hepcidin-independent hypoferremia. Blood. 2015;125(14):2265-2275. 7. Kemna E, Pickkers P, Nemeth E, van der Hoeven H, Swinkels D. Time-course analysis of hepcidin, serum iron, and plasma cytokine levels in humans injected with LPS. Blood. 2005;106(5):1864-1866. 8. Babitt JL, Sitara D. Crosstalk between fibroblast growth factor 23, iron, erythropoietin, and inflammation in kidney disease. Curr Opin Nephrol Hypertens. 2019;28(4):304-310. 9. Edmonston D, Wolf M. FGF23 at the crossroads of phosphate, iron economy and erythropoiesis. Nat Rev Nephrol. 2020;16(1):7-19. 10. Richter B, Faul C. FGF23 actions on target tissues-with and without Klotho. Front Endocrinol (Lausanne). 2018;9:189. 11. Goetz R, Nakada Y, Hu MC, et al. Isolated C-terminal tail of FGF23 alleviates hypophosphatemia by inhibiting FGF23-FGFR-Klotho complex formation. Proc Natl Acad Sci U S A. 2010;107(1):407-412. 12. van Swelm RPL, Wetzels JFM, Swinkels DW. The multifaceted role of iron in renal health and disease. Nat Rev Nephrol. 2020;16(2):77-98. 13. Agoro R, Montagna A, Goetz R, et al. Inhibition of fibroblast growth factor 23 (FGF23) signaling rescues renal anemia. FASEB J. 2018;32(7):3752-3764. 14. Frede S, Fandrey J, Pagel H, Hellwig T, Jelkmann W. Erythropoietin gene expression is suppressed after lipopolysaccharide or interleukin-1 beta injections in rats. Am J Physiol. 1997;273(3 Pt 2):R10671071. 15. Nemeth E, Rivera S, Gabayan V, et al. IL-6 mediates hypoferremia of inflammation by inducing the synthesis of the iron regulatory hormone hepcidin. J Clin Invest. 2004;113(9):1271-1276. 16. Wang CY, Babitt JL. Liver iron sensing and body iron homeostasis. Blood. 2019;133(1):18-29. 17. Kautz L, Jung G, Valore EV, Rivella S, Nemeth E, Ganz T. Identification of erythroferrone as an erythroid regulator of iron metabolism. Nat Genet. 2014;46(7):678-684. 18. Rivera S, Ganz T. Animal models of anemia of inflammation. Semin Hematol. 2009;46(4):351-357.

haematologica | 2021; 106(2)

Editorials

Chromosome Y loss and drivers of clonal hematopoiesis in myelodysplastic syndrome Panagiotis Baliakas1 and Lars A. Forsberg1, 2 1 Department of Immunology, Genetics and Pathology, Science for Life Laboratory, Uppsala University and 2The Beijer Laboratory, Uppsala University, Uppsala, Sweden

E-mail: LARS A. FORSBERG - lars.forsberg@igp.uu.se doi:10.3324/haematol.2020.266601

n the current issue of Haematologica, Ouseph et al.1 present new results on the relation between somatic Y chromosome loss and hematologic malignancies. Mosaic loss of chromosome Y (LOY) refers to chromosome Y aneuploidy acquired during life and it is the most common somatic mutation in human blood cells.2-4 Affected men carry cells without the entire Y chromosome in a mosaic fashion due to its loss from progenitor cells during the lifetime. Thus, in a single cell LOY is a binary event causing the absence of almost 2% of the male haploid nuclear genome and when measured in bulk blood samples, it is present as a continuous mosaicism affecting a fraction of cells. In normally aging populations, LOY is detectable in at least 10% of peripheral blood cells in about 5%, 20%, 40% and 60% of men around 50, 60, 70 and 90 years of age, respectively.4-12 Furthermore, longitudinal results from subjects without known hematologic malignancy suggest that the frequency of blood cells without the Y chromosome typically increases over time.11 However, profound variation in longitudinal patterns can be observed between individuals, with some subjects recovering from LOY over time.11 Interestingly, longitudinal data presented by Ouseph et al. suggest a general increase in the frequency of LOY within subjects developing myelodysplastic syndrome (MDS). Since a paper published in 1963,13 mosaic Y chromosome loss has been recognized as a frequent event in cells of the hematopoietic system and in blood cells of aging men and it was long viewed as a neutral event. In contrast to that old paradigm, recent epidemiological results suggest the opposite, showing that men with LOY in blood have an increased risk of all-cause mortality.5,14 For example, in a cohort of old men we described that men with LOY in at least 35% of peripheral blood cells at study entry survived on average only half as long as controls during the follow-up period.5 During the last years, LOY in blood has been linked with an increased risk of a continuously growing list of diseases, such as various forms of hematologic and non-hematologic cancer,1,5,7,15-18 Alzheimer disease,6 autoimmune conditions,19,20 cardiovascular diseases,14,21 schizophrenia,22 diabetes14 and age-related macular degeneration.10 Although LOY is more common in the elderly, it is detectable in younger subjects4,9 and associations with outcomes such as testicular germ cell tumors17 and age-related macular degeneration10 have been reported also in younger men. Of note, epidemiological and clinical data suggest that, on average, men with COVID-19 have more severe symptoms and higher rates of death than women with this disease. However, the impact of LOY in males vulnerable to the Sars-Cov-2 virus and other infectious diseases is largely unknown. As further discussed by Ouseph et al.,1 the role of LOY in hematologic haematologica | 2021; 106(2)

malignancies has been elusive. However, in their study they showed that LOY in at least 75% of metaphases in bone marrow samples from a cohort of clinically defined patients was associated with incident as well as prospective diagnoses of myelodysplastic syndrome (MDS). Furthermore, they also describe new results showing that patients with â&#x2030;Ľ75% LOY had a high prevalence of somatic mutations in genes linked with myeloid neoplasia (especially MDS) and clonal hematopoiesis of indeterminate potential (CHIP). The authors suggest that a high level of LOY (â&#x2030;Ľ75%) is associated with an MDS-type mutation signature, representing a disease-associated clonal proliferation, and conclude that high-proportion LOY is likely to be a true MDS-associated cytogenetic aberration rather than an incidental finding due to aging. Their study provides molecular proof supporting the current praxis of the majority of diagnostics laboratories when it comes to the interpretation of LOY in patients with a suspicion/diagnosis of MDS. In particular, when LOY is observed in low number of metaphases (<80%) it is considered more an aging-related phenomenon, not important for the clinical management, while when present in a high number of metaphases it is included in the prognostic algorithm. In addition to age itself, a number of intrinsic and external risk factors influence individual risk of being a carrier of blood cells without chromosome Y. Genetic susceptibility to LOY has been reported in several recent genomewide association studies.4,7,8,23 In the largest to date, variants in 156 genetic loci preferentially found near genes involved in cell-cycle regulation, cell proliferation and genome instability were identified.4 The nature of the risk loci suggests that LOY might primarily be an effect of chromosome mis-segregation during mitosis. Given the genetic component in individual propensity to be affected with LOY during lifetime, it is interesting that men of African ancestry display, on average, a lower frequency of LOY compared with men of European ancestry.14 Other reported risk factors for LOY in blood include smoking,7,8,10,14,24 obesity14 and exposure to environmental risk factors such as air pollution25 and polycyclic aromatic hydrocarbons.26 For example, we showed that current smokers have an up to 4-fold increased risk of being affected compared with never smokers, with the increase being dose-dependent, and that smoking cessation is associated with recovery from LOY.24 The transient effect from smoking has been replicated7 and causality established by Mendelian randomization.8 Still largely unresolved questions include how LOY in blood might be associated with risk of disease in other organs. As illustrated in Figure 1, LOY in the immune cells of blood might represent a new and understudied disease mechanism driven by the accumulation of different types 329

Editorials

Figure 1. During lifetime, acquired mutations, such as point mutations, structural variants and aneuploidies, in hematopoietic progenitors give rise to accumulation of immune the cells carrying somatic variants associated with clonal hematopoiesis. Harboring genetically altered leukocytes is associated with disease vulnerability and could be linked with dysregulation of normal immune cell functions. Known risk factors for common diseases and clonal hematopoiesis are largely overlapping and their relative contributions to disease risks need further evaluation.

of somatic variants (including LOY, CHIP and other types of somatic alterations) in immune cells. In addition to being associated with clonal hematopoiesis in aging individuals, it has been hypothesized that such somatic variants in leukocytes could have a negative impact on normal immune cell functions. For instance, LOY has been found to be associated with a substantial impact on genomewide transcriptional regulation27-30 while we reported that patients diagnosed with prostate cancer and Alzheimer disease were primarily affected with LOY in different types of leukocytes.29 Our studies of gene expression in immune cells from blood (i.e., different types of leukocytes and single cells with LOY collected in vivo) identified LOY-associated dysregulation of nearly 500 autosomal genes.29 The nature of the affected genes support the hypothesis that LOY could contribute to disease vulnerability by altering normal immune cell biology. However, it is still not known how the expression of hundreds of autosomal genes is dysregulated in cells after losing the Y chromosome. That said, the Y chromosome in not a ‘genetic wasteland’ and encodes proteins involved in processes such as transcription, translation and chromatin modifications,29 processes that would be affected in cells without chromosome Y. A direct physiological effect could help to elucidate how LOY in blood is associated with an increased risk of disease in tissues of the entire body. However, it should be emphasized that the risk factors for many of the dis330

eases associated with LOY show a substantial overlap with the known risk factors for LOY and clonal hematopoiesis, such as age, smoking and genetic susceptibility (Figure 1). Clearly, such covariance is challenging when trying to make causal inferences and deduction of links with potential disease mechanisms. For example, many of the identified LOY-associated risk loci are also linked with cancer susceptibility and other types of disease, such as type 2 diabetes, as well as age at menopause in women.4 Hence, associations between LOY and disease are in part explained by a ‘common soil’ of genetic predisposition to genomic instability,4 independently associated with an increased risk of LOY in the blood and disease development in various tissues. Remarkably, Ouseph et al.1 report a high prevalence of somatic variants in genes linked with myeloid neoplasia and CHIP in samples from MDS patients with high levels of LOY. Further studies are needed to understand the co-occurrence and implications of different types of somatic variants in immune cells associated with clonal hematopoiesis in normally aging populations. In conclusion, the increased risk of disease in men with LOY and other somatic variants in immune cells can likely be explained by several non-exclusive disease mechanisms. Hence, a reduced capability of altered immune cells to combat disease processes (driven by somatic variants and clonal hematopoiesis) would be operating in parallel with more traditional risk factors (Figure 1). haematologica | 2021; 106(2)

Editorials

Finally, it has been known for centuries that, throughout the world, men live a shorter time than women31 but until recently, the biological mechanisms behind this sex difference were poorly understood. Regardless of the underlying disease mechanism(s) discussed here, LOY is a malespecific somatic aneuploidy that can help to explain part of this bias in longevity. Interestingly, a recent study demonstrated that across 229 different species, the heterogametic sex has on average a shorter lifespan than the homogametic sex.32 For example, females tend to live longer in mammals and males tend to live longer in birds, as predicted by the “unguarded X hypothesis”. Accordingly, a reduced sex chromosome in the heterogametic sex (e.g., chromosome Y in mammals and chromosome W in birds) might explain part of the differences in lifespan due to exposure of recessive deleterious mutations in the heterogametic sex.32 Future studies could shed light on the relative contribution of effects associated with LOY and the unguarded X hypothesis to the observed differences in disease incidence and mortality between men and women in human populations. Disclosures LAF is cofounder and shareholder in Cray Innovation AB. Contributions PB and LAF wrote the manuscript. Acknowlegments This work was supported by the European Research Council ERC (#679744), the Swedish Research Council (#201703762) and Kjell och Märta Beijers Stiftelse to LAF.

References 1. Ouseph MM, Hasserjian RP, Dal Cin P, et al. Genomic alterations in patients with somatic loss of the Y chromosome as the sole cytogenetic finding in bone marrow cells. Haematologica. 2020;106(2):555564. 2. Forsberg LA, Gisselsson D, Dumanski JP. Mosaicism in health and disease - clones picking up speed. Nat Rev Genet. 2017;18(2):128-142. 3. Forsberg LA. Loss of chromosome Y (LOY) in blood cells is associated with increased risk for disease and mortality in aging men. Hum Genet. 2017;136(5):657-663. 4. Thompson DJ, Genovese G, Halvardson J, et al. Genetic predisposition to mosaic Y chromosome loss in blood. Nature. 2019;575(7784):652-657. 5. Forsberg LA, Rasi C, Malmqvist N, et al. Mosaic loss of chromosome Y in peripheral blood is associated with shorter survival and higher risk of cancer. Nat Genet. 2014;46(6):624-628. 6. Dumanski JP, Lambert JC, Rasi C, et al. Mosaic loss of chromosome Y in blood is associated with Alzheimer disease. Am J Hum Genet. 2016;98(6):1208-1219. 7. Zhou W, Machiela MJ, Freedman ND, et al. Mosaic loss of chromosome Y is associated with common variation near TCL1A. Nat Genet. 2016;48(5):563-568. 8. Wright DJ, Day FR, Kerrison ND, et al. Genetic variants associated with mosaic Y chromosome loss highlight cell cycle genes and overlap with cancer susceptibility. Nat Genet. 2017;49(5):674-679. 9. Zink F, Stacey SN, Norddahl GL, et al. Clonal hematopoiesis, with

haematologica | 2021; 106(2)

and without candidate driver mutations, is common in the elderly. Blood. 2017;130(6):742-752. 10. Grassmann F, Kiel C, den Hollander AI, et al. Y chromosome mosaicism is associated with age-related macular degeneration. Eur J Hum Genet. 2019;27(1):36-41. 11. Danielsson M, Halvardson J, Davies H, et al. Longitudinal changes in the frequency of mosaic chromosome Y loss in peripheral blood cells of aging men varies profoundly between individuals. Eur J Hum Genet. 2020;28(3):349-357. 12. Forsberg LA, Halvardson J, Rychlicka-Buniowska E, et al. Mosaic loss of chromosome Y in leukocytes matters. Nat Genet. 2019;51(1):4-7. 13. Jacobs PA, Brunton M, Court Brown WM, Doll R, Goldstein H. Change of human chromosome count distribution with age: evidence for a sex differences. Nature. 1963;197:1080-1081. 14. Loftfield E, Zhou W, Graubard BI, et al. Predictors of mosaic chromosome Y loss and associations with mortality in the UK Biobank. Sci Rep. 2018;8(1):12316. 15. Ganster C, Kampfe D, Jung K, et al. New data shed light on Y-lossrelated pathogenesis in myelodysplastic syndromes. Genes Chromosomes Cancer. 2015;54(12):717-724. 16. Noveski P, Madjunkova S, Sukarova Stefanovska E, et al. Loss of Y chromosome in peripheral blood of colorectal and prostate cancer patients. PLoS One. 2016;11(1):e0146264. 17. Machiela MJ, Dagnall CL, Pathak A, et al. Mosaic chromosome Y loss and testicular germ cell tumor risk. J Hum Genet. 2017;62(6):637-640. 18. Loftfield E, Zhou W, Yeager M, Chanock SJ, Freedman ND, Machiela MJ. Mosaic Y loss is moderately associated with solid tumor risk. Cancer Res. 2019;79(3):461-466. 19. Persani L, Bonomi M, Lleo A, et al. Increased loss of the Y chromosome in peripheral blood cells in male patients with autoimmune thyroiditis. J Autoimmun. 2012;38(2-3):J193-196. 20. Lleo A, Oertelt-Prigione S, Bianchi I, et al. Y chromosome loss in male patients with primary biliary cirrhosis. J Autoimmun. 2013;41:87-91. 21. Haitjema S, Kofink D, van Setten J, et al. Loss of Y chromosome in blood is associated with major cardiovascular events during followup in men after carotid endarterectomy. Circ Cardiovasc Genet. 2017;10(4):e001544. 22. Hirata T, Hishimoto A, Otsuka I, et al. Investigation of chromosome Y loss in men with schizophrenia. Neuropsychiatr Dis Treat. 2018;14:2115-2122. 23. Terao C, Momozawa Y, Ishigaki K, et al. GWAS of mosaic loss of chromosome Y highlights genetic effects on blood cell differentiation. Nat Commun. 2019;10(1):4719. 24. Dumanski JP, Rasi C, Lonn M, et al. Smoking is associated with mosaic loss of chromosome Y. Science. 2015;347(6217):81-83. 25. Wong JYY, Margolis HG, Machiela M, et al. Outdoor air pollution and mosaic loss of chromosome Y in older men from the Cardiovascular Health Study. Environ Int. 2018;116:239-247. 26. Liu Y, Bai Y, Wu X, et al. Polycyclic aromatic hydrocarbons exposure and their joint effects with age, smoking, and TCL1A variants on mosaic loss of chromosome Y among coke-oven workers. Environ Pollut. 2020;258:113655. 27. Caceres A, Jene A, Esko T, Perez-Jurado LA, Gonzalez JR. Extreme downregulation of chromosome Y and Alzheimer's disease in men. Neurobiol Aging. 2020;90:150.e1-150.e4. 28. Caceres A, Jene A, Esko T, Perez-Jurado LA, Gonzalez JR. Extreme downregulation of chromosome Y and cancer risk in men. J Natl Cancer Inst. 2020;112(9):913-920. 29. Dumanski JP, Halvardson J, Davies H, et al. Loss of Y in leukocytes, dysregulation of autosomal immune genes and disease risks. BioRxiv. 2019:673459. 30. Case LK, Wall EH, Dragon JA, et al. The Y chromosome as a regulatory element shaping immune cell transcriptomes and susceptibility to autoimmune disease. Genome Res. 2013;23(9):1474-1485. 31. Nathanson CA. Sex differences in mortality. Annu Rev Sociol. 1984;10:191-213. 32. Xirocostas ZA, Everingham SE, Moles AT. The sex with the reduced sex chromosome dies earlier: a comparison across the tree of life. Biol Lett. 2020;16(3):20190867.

331

Editorials

Central nervous system prophylaxis in diffuse large B-cell lymphoma: a race to the bottom Mark Roschewski Lymphoid Malignancies Branch, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA E-mail: MARK ROSCHEWSKI - mark.roschewski@nih.gov doi:10.3324/haematol.2020.266635

n this issue of Haematologica, Bobillo et al. address the controversial topic of chemotherapy as central nervous system (CNS) prophylaxis during front-line management of diffuse large B-cell lymphoma (DLBCL).1 Despite the fact that CNS spread is a feared and often terminal complication of DLBCL treatment, there is no broad consensus regarding which patients should receive CNS prophylaxis or which is the most effective delivery method.2 Overall, the incidence of CNS relapse across all DLBCL subsets is only approximately 5%, but clinical risk factors, including the involvement of specific anatomic sites, are associated with a significantly higher rate of CNS spread. Further, we are beginning to uncover the biological basis for DLBCL involving the CNS, as specific genetic subtypes demonstrate an inherently higher rate of CNS tropism.3-5 The CNS International Prognostic Index (CNS-IPI) is a commonly used risk model that stratifies patients into risk categories,6 and combining this model with the cell-of-origin phenotype may improve patient selection.7 However, even the most robust predictive models cannot overcome the fundamental problem that the chemotherapy agents most effective for the cure of systemic DLBCL do not reliably penetrate the bloodbrain barrier (Figure 1).8 Conversely, methotrexate (MTX), which reliably penetrates the CNS, is not highly effective for DLBCL. The most commonly used prophylactic strategy is repeated intrathecal (IT) injections of chemotherapy such as MTX during front-line therapy, but since brain parenchymal sites are the commonest site of CNS relapse, some advocate the use of deeply penetrant drugs such as high-dose methotrexate (HDMTX).9,10 No randomized prospective study has directly addressed this specific question, and, as a result, practice patterns rely on consensus guidelines and vary widely across institutions and individual providers.11 In essence, the debate about optimal delivery methods is essentially a 'race to the bottom' that compares two strategies that do not adequately address the clinical problem. Bobillo et al. utilize this lack of consensus to retrospectively compare clinical outcomes in 585 patients with DLBCL considered high-risk for CNS relapse (29% of all DLBCL cases screened) treated at their own institution with a variety of CNS prophylactic strategies including no prophylaxis at all.1 Most (86%) of the 295 patients who received CNS prophylaxis were treated with a median of four IT injections of MTX or cytarabine during front-line therapy while a significant minority (14%) of patients were treated with a median of two cycles of HDMTX at a median dose of 3.5g/m2 either during front-line therapy (45% of cases) or immediately following (55% of cases). Notably, 11 (26%) patients who received HDMTX also received concomitant IT chemotherapy. Importantly, 290 (50%) patients received no form of CNS 332

prophylaxis and these patients were significantly older (median age 72 years). This observation highlights the fact that patient-related factors such as age and perceived ability to tolerate treatment-related toxicity greatly influence treatment decisions beyond prognostic scores and/or involvement of extranodal sites. Since all forms of CNS prophylaxis have clinically meaningful toxicities, this underscores the fact that an important limitation of all available datasets is patient selection bias. Bobillo et al. first confirmed the expected finding that patients with relapse in the CNS had a worse median overall survival of only 4.9 months compared to 17.1 months for those with systemic-only relapse (P=0.03).1 Regarding efficacy, after a median follow-up of 6.8 years, the 5-year CNS relapse rate was still 5.6% in patients who received any form of CNS prophylaxis, confirming that currently employed chemotherapy strategies are not universally effective. An important observation in this study was that CNS relapses in patients treated with prophylaxis occurred a median of 19 months after front-line therapy compared to only 8 months in patients who received no prophylaxis. As a result, even though CNS prophylaxis appeared to reduce the risk of CNS relapse at 1 year, there was no difference in the 5-year CNS relapse rate between patients who received prophylaxis compared to those who received no CNS prophylaxis (5.6% vs. 7.5%) (Risk Ratio 0.76, 95% Confidence Interval: 0.35-1.50). Further, there was no difference in the 5-year CNS relapse risk in patients who received IT MTX compared to those who received HD-MTX, although the number of events was small. Taken together, these data suggest that actual risk reduction of any form of CNS prophylaxis with chemotherapy is likely to be modest at best and currently employed strategies may simply delay the timing of CNS recurrence. These findings also have implications for the reporting of CNS relapse. It is well-described that patients with primary DLBCL of the CNS (PCNSL) often have late relapses, and early reporting of clinical outcomes will miss a significant number of events and may overestimate the true cure rate in subsets of DLBCL with the highest CNS tropism.12 Many retrospective studies that have described the incidence of CNS relapse do not include long-term follow-up beyond 2 or 3 years and therefore may be underreporting the true incidence.2 However, the risk of CNS involvement is not equally distributed across all subsets of DLBCL, which may allow for precision medicine strategies. In fact, DLBCL is not a singular disease but comprises a spectrum of aggressive lymphomas with striking underlying genetic diversity. The current classification system recognizes both ABC DLBCL and GCB DLBCL as distinct molecular subtypes and introduced a new entity, high-grade B-cell lymphoma, defined by the presence of MYC and BCL2 haematologica | 2021; 106(2)

Editorials

Figure 1. A subset of patients with diffuse large B-cell lymphoma (DLBCL) are at high risk of disease spread to the central nervous system (CNS) and are often treated with chemotherapy prophylaxis. A critical barrier to effective CNS prophylaxis is the blood brain barrier (1) which limits the entry of the chemotherapy agents most effective for systemic DLBCL (2). Current therapeutic options for chemotherapy CNS prophylaxis are systemic chemotherapy (3) or intrathecal chemotherapy (4) which are both limited in efficacy and increase toxicity. Novel small molecule inhibitors are being tested in DLBCL involving the CNS that effectively penetrate the blood brain barrier and may improve treatment options.

and/or BCL6 rearrangements (HGBCL-DH/TH).13 Indeed, Bobillo et al. observed that patients with ABC (non-GCB) DLBCL subtype had an overall higher risk of CNS relapse in their series.1 Furthermore, recent multiplatform genomic profiling studies have identified genetic subtypes of DLBCL with shared genetic features.3,4 One genetic subtype MCD is characterized by frequent co-occurrence of MYD88L265P and CD79B mutations, prominent immuneediting features, and PIM1 mutations.3 These tumors occur almost exclusively within ABC DLBCL and frequently involve extranodal sites including the testes, breast and CNS.3 It is noteworthy that a separate multiplatform genomic profiling study described a very similar subtype termed Cluster 5 (C5) tumors which were characterized by MYD88L265P and CD79B mutations, gain of 18q, and PIM1 mutations, and also exhibited a propensity for extranodal sites, including the CNS and testis.4 Moreover, a recently reported series of 26 cases of secondary DLBCL of the CNS confirmed a higher prevalence of MCD subtype than observed in a reference cohort of relapsed DLBCL without CNS spread (38% vs. 8%; P=0.003).14 In this study, the majority of other DLBCL cases with CNS spread were either HGBCL-DH/TH or associated with TP53 mutations. Another recent study investigated the genomic predictors of CNS relapse in 82 cases of primary testicular DLBCL, which has a strong predilection for CNS spread.15 The authors identified BCL6 and/or PDL1 or PDL2 rearrangements as the most common genetic aberrations associated with CNS relapse after treatment for primary testicular DLBCL. Although haematologica | 2021; 106(2)

the precise mechanisms by which various genetic aberrations co-operate to promote CNS spread remains undetermined, these results suggest that a more nuanced understanding of the molecular biology of DLBCL involving the CNS may lead to novel therapeutic targets. In order to improve clinical outcomes, however, novel therapies with demonstrable efficacy within genetically defined subtypes will be necessary. Multiple clinical studies have reported impressive clinical activity of the Bruton tyrosine kinase (BTK) inhibitor ibrutinib and ibrutinibbased regimens in DLBCL involving the CNS, including patients refractory to chemotherapy.16,17 Even though a randomized phase III study did not show an overall benefit from adding ibrutinib to R-CHOP as part of front-line therapy for non-GCB DLBCL, certain subsets appeared to have improved outcomes.18 Further studies of BTK inhibitors with R-CHOP are currently ongoing that should provide additional data regarding rates of CNS relapse. In addition, the immunomodulatory agent lenalidomide has demonstrated good clinical activity and favorable safety in DLBCL involving the CNS.19 Lenalidomide has also been added to R-CHOP as part of front-line therapy for DLBCL that may benefit certain subsets of DLBCL.20 The currently available data do not support the use of either ibrutinib or lenalidomide as part of front-line therapy to prevent CNS spread of DLBCL, but all clinical trials testing novel agents should report CNS-specific outcomes within genetically defined subtypes. In summary, chemotherapy as CNS prophylaxis is not universally effective no matter what the delivery method, 333

Editorials

and the prevention and treatment of CNS relapse remains an unmet clinical need in the management of DLBCL. Penetrating the blood-brain barrier is an important consideration, but improved therapies will be required to overcome intrinsic chemotherapy resistance. A nuanced mechanistic understanding of targetable pathways underpinning DLBCL involving the CNS has led to novel targeted agents and immunotherapy approaches that demonstrate promising clinical activity and good CNS penetrance. Novel agents that target oncogenic drivers based on the underlying biology of DLBCL subtypes may ultimately prove to be the most effective way to prevent and/or treat CNS recurrence. Disclosures No conflicts of interest to disclose.

References 1. Bobillo S, Younes A. Central nervous system prophylaxis with highdose methotrexate or intrathecal chemotherapy in diffuse large Bcell lymphoma and high-risk of CNS relapse. Haematologica. 2020; 106(2):513-521. 2. Eyre TA, Kirkwood AA, Wolf J, et al. Stand-alone intrathecal central nervous system (CNS) prophylaxis provide unclear benefit in reducing CNS relapse risk in elderly DLBCL patients treated with RCHOP and is associated increased infection-related toxicity. Br J Haematol. 2019;187(2):185-194. 3. Schmitz R, Wright GW, Huang DW, et al. Genetics and pathogenesis of diffuse large B-cell lymphoma. N Engl J Med. 2018;378(15):13961407. 4. Chapuy B, Stewart C, Dunford AJ, et al. Molecular subtypes of diffuse large B cell lymphoma are associated with distinct pathogenic mechanisms and outcomes. Nat Med. 2018;24(5):679-690. 5. Wright GW, Huang DW, Phelan JD, et al. A probabilistic classification tool for genetic subtypes of diffuse large B cell lymphoma with therapeutic implications. Cancer Cell. 2020;37(4):551-568. 6. Schmitz N, Zeynalova S, Nickelsen M, et al. CNS international prognostic index: a risk model for CNS relapse in patients with diffuse large B-cell lymphoma treated with R-CHOP. J Clin Oncol. 2016;34(26):3150-3156. 7. Klanova M, Sehn LH, Bence-Bruckler I, et al. Integration of cell of origin into the clinical CNS International Prognostic Index improves CNS relapse prediction in DLBCL. Blood. 2019;133(9):919-926.

334

8. Arvanitis CD, Ferraro GB, Jain RK. The blood-brain barrier and blood-tumour barrier in brain tumours and metastases. Nat Rev Cancer. 2020;20(1):26-41. 9. Eyre TA, Djebbari F, Kirkwood AA, Collins GP. Efficacy of central nervous system prophylaxis with stand-alone intrathecal chemotherapy in diffuse large B-cell lymphoma patients treated with anthracycline-based chemotherapy in the rituximab era: a systematic review. Haematologica. 2020;105(7):1914-1924. 10. Abramson JS, Hellmann M, Barnes JA, et al. Intravenous methotrexate as central nervous system (CNS) prophylaxis is associated with a low risk of CNS recurrence in high-risk patients with diffuse large Bcell lymphoma. Cancer. 2010;116(18):4283-4290. 11. McKay P, Wilson MR, Chaganti S, et al. The prevention of central nervous system relapse in diffuse large B-cell lymphoma: a British Society for Haematology good practice paper. Br J Haematol. 2020;190(5):708-714. 12. Ambady P, Holdhoff M, Bonekamp D, Wong F, Grossman SA. Late relapses in primary CNS lymphoma after complete remissions with high-dose methotrexate monotherapy. CNS Oncol. 2015;4(6):393-398. 13. Swerdlow SH, Campo E, Pileri SA, et al. The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood. 2016;127(20):2375-2390. 14. Ollila TA, Kurt H, Waroich J, et al. Genomic subtypes may predict the risk of central nervous system recurrence in diffuse large B-cell lymphoma. Blood. 2020 Sep 2. [Epub ahead of print] 15. Twa DDW, Lee DG, Tan KL, et al. Genomic predictors of central nervous system relapse in primary testicular diffuse large B-cell lymphoma (DLBCL). Blood. 2020 Sep 23. [Epub ahead of print] 16. Lionakis MS, Dunleavy K, Roschewski M, et al. Inhibition of B cell receptor signaling by ibrutinib in primary CNS lymphoma. Cancer Cell. 2017;31(6):833-843.e5. 17. Grommes C, Pastore A, Palaskas N, et al. Ibrutinib unmasks critical role of bruton tyrosine kinase in primary CNS lymphoma. Cancer Discov. 2017;7(9):1018-1029. 18. Younes A, Sehn LH, Johnson P, et al. Randomized phase III trial of ibrutinib and rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone in non-germinal center B-cell diffuse large Bcell lymphoma. J Clin Oncol. 2019;37(15):1285-1295. 19. Ghesquieres H, Chevrier M, Laadhari M, et al. Lenalidomide in combination with intravenous rituximab (REVRI) in relapsed/refractory primary CNS lymphoma or primary intraocular lymphoma: a multicenter prospective 'proof of concept' phase II study of the French Oculo-Cerebral lymphoma (LOC) Network and the Lymphoma Study Association (LYSA). Ann Oncol. 2019;30(4):621-628. 20. Nowakowski GS, Hong F, Scott DW, et al. Addition of Lenalidomide to R-CHOP (R2CHOP) improves outcomes in newly diagnosed diffuse large B-Cell lymphoma (DLBCL): first report of ECOGACRIN1412 a randomized phase 2 US intergroup study of R2CHOP vs R-CHOP. Hematol Oncol. 2019;37(S2):37-8.

haematologica | 2021; 106(2)

Editorials

An early start of Coup-TFII promotes γ-globin gene expression in adult erythroid cells Stefania Bottardi1 and Eric Milot1,2 1

Maisonneuve Rosemont Hospital Research Center, CIUSSS Est de l’Île de Montréal, Montréal, and 2Department of Medicine, University of Montreal, Montréal, Québec, Canada E-mail: ERIC MILOT - e.milot.1@umontreal.ca doi:10.3324/haematol.2020.266791

utations of the human β-globin gene can affect the quantity or quality of the β-globin subunit and result in β-thalassemia or sickle cell anemia (respectively). These disorders can be regrouped under the designation of β-hemoglobinopathies and are the most frequent monogenic inherited diseases worldwide.1 The severity of β-hemoglobinopathies is strongly related to the nature of the mutation. Allogeneic hematopoietic stem cell transplantation is the most established cure for severe forms of β-hemoglobinopathies. However, like other clinical approaches, including gene therapy, which are used to correct this group of genetic diseases, allogeneic hematopoietic stem cell transplantation is not without risk and can be associated with complications.2 The developmental switching from γ- to β-globin gene expression occurs around birth and results in the formation of the adult form of hemoglobin, which is primarily composed of β and α-globin chains (HbA, α2β2). HbA gradually replaces the fetal hemoglobin (HbF, α2γ2) and hence, the developmental globin switching represents the genetic basis of HbF decline, which will drop to less than 1% within a few months of birth (by 6 months of age).3 However, a condition known as hereditary persistence of fetal hemoglobin (HPFH) is characterized by the maintenance of HbF at levels as high as 10% to 40% of total hemoglobin in adult blood.4 This condition is asymptomatic and can compensate for β-globin gene mutations that otherwise promote β-hemoglobinopathies. Indeed, increased HbF levels in adult blood cells is beneficial to patients with mutations affecting the quantity or quality of the β-globin chains. Thus, a substantial number of studies have been performed to determine how the reactivation of γ-globin genes could be efficiently reproduced in adult blood cells. However, the complexity of gene regulation at this multigenic locus and the absence of a known specific activator of the γ-globin genes that would not simultaneously increase the expression of the β-globin gene have delayed this achievement. To modulate the expression levels of human β-type globin genes, the influence of multiple control elements and different mechanisms of gene regulation have to be considered. Furthermore, tissue- and development-specific expression of β-type globin genes relies on a complex network of transcription factors, cofactors and regulatory complexes.5-7 For instance, in addition to binding gene regulatory regions, different transcription factors favor longrange interactions between promoters and distal regulatory elements of the locus in order to induce high levels of globin gene expression.8 Nonetheless, while many different transcriptional activators can influence these longrange interactions in the β-globin locus, it has been demonstrated that one engineered transcriptional regulator can be sufficient to induce these structures and reacti-

haematologica | 2021; 106(2)

vate γ-globin genes in adult erythroid cells.9 Another example of the complex regulation of globin genes is the number of transcription factors and complexes involved in γ-globin gene repression in adult erythroid cells. Indeed, although BCL11A is instrumental in the repression of γglobin genes at an adult stage of development,10,11 the mutation of many different transcription factors and cofactors results in the partial reactivation of γ-globin genes.6 Thus, even if discoveries such as the dominant effect of BCL11A on γ-globin gene repression are promising for the development of novel therapies to fight βhemoglobinopathies, the main objective of different studies remains the identification of a transcriptional activator capable of reactivating the γ-globin genes in adult erythroid cells. A specific activator of the human β-globin gene was identified in KLF1 (EKLF),12 but until now such a transcriptional factor or cofactor specifically involved in the γ-globin gene activation was lacking. Importantly, in this issue of Haematologica, Fugazza et al. report that CoupTFII, which was originally considered to be a γ-globin gene repressor, is a specific transcriptional activator of eand γ-globin genes.13 Fugazza et al. report that, when expressed in adult erythroid cells of healthy donors or thalassemic patients carrying the Sardinian β039 mutation (β039-Thal), Coup-TFII can reactivate γ-globin gene expression.13 Their identification of Coup-TFII as a specific activator of γ-globin genes that can enhance HbF production in adult red blood cells brings hope, and increases the therapeutic possibilities to fight β-hemoglobinopathies. The Coup-TFII (NR2F2/ARP-1) orphan nuclear receptor was initially identified as a component of the e- and γ-globin gene-regulating complex NF-E3. It was proposed to be a positive regulator of the e-globin gene but a negative regulator of γ-globin genes.14 Further investigation indicated that NF-E3/Coup-TFII can interfere with the recruitment of different transcriptional regulators to e- and γ-globin promoters, and unveiled conflicting results that could imply its involvement in the transcriptional activation of γ-globin genes.15 However, detailed molecular analysis of the precise effect of Coup-TFII on globin gene regulation during development has been complicated by the fact that it binds different regulatory regions of the β-globin locus and because Coup-TFII knockout is lethal at the embryonic stage of mouse development. To study the importance of the transcription factor Coup-TFII in the regulation of embryonic and fetal globin genes, Fugazza et al. used transgenic mice carrying the human β-globin locus, a β-K562 human cell line and human blood cells from healthy or β039-Thal donors.13 In mice, they found that Coup-TFII expression occurs at the same developmental stage as endogenous murine or transgenic human embryonic/fetal globin genes and, like the 335

Editorials

embryonic/fetal globin genes, Coup-TFII expression extinguishes at embryonic day 12.5. However, they showed that only a small percentage of erythroid cells are characterized by the simultaneous expression of Coup-TFII and mouse or human embryonic/fetal globin genes. Their results indicate that Coup-TFII expression is initiated in the erythroid-myeloid precursor cells and precedes γ-globin gene transcription during erythroid cell maturation, thus suggesting that the effect of Coup-TFII precedes the transcriptional activation of e- and γ-globin genes during erythroid cell differentiation/formation. Interestingly, in hematopoietic progenitor cells the human β-like globin genes are ‘primed’ for their subsequent transcriptional activation in erythroid cells.16-18 A few transcription factors have been reported to favor globin gene priming.19,20 Thus, Coup-TFII might also be involved in globin gene priming and, thereby, promote the specific activation of e- and γglobin genes. The effect of Coup-TFII expression was also tested in adult erythroid progenitor cells obtained from healthy donors and β039-Thal patients. As in the transgenic mouse model carrying the human β-globin locus, Coup-TFII expression was sufficient to reactivate the γ-globin genes in human adult erythroid cells. Furthermore, as indicated by the fold change in gene transcription, the effect of Coup-TFII was specific for γ- but not β-globin gene expression. Therefore, the results presented by Fugazza et al. indicate that the induction of Coup-TFII expression in adult erythroid cells could be beneficial for the treatment of patients with β-hemoglobinopathies. To further characterize the molecular aspects linked to Coup-TFII-related activation of γ-globin genes, Coup-TFII knockout or overexpression was engineered in a human model cell line, β-K562 cells. Genome-wide analysis of Coup-TFII chromatin binding performed by chromatin immunoprecipitation (ChIP)-sequencing suggested that Coup-TFII is frequently recruited to genes/loci associated with hematologic diseases. This analysis identified significant peaks of Coup-TFII binding within the β-globin locus control region hypersensitive sites 2 (HS2), HS3, HS4, and the γ−d intergenic region. Accordingly, it was found that Coup-TFII overexpression increases the longrange interaction between the locus control region and this γ−d intergenic region. Overall the analyses performed in β-K562 cells indicate that Coup-TFII binds to regulatory regions and influences the organization of the human βglobin locus. Furthermore, the ChIP-sequencing analysis and the finding that Coup-TFII peaks frequently overlap with GATA consensus binding sequences, led the authors to propose the participation of Coup-TFII in the erythropoietic program. Considered collectively, the results presented by Fugazza et al. elucidate the role of Coup-TFII in the regulation of e- and γ-globin genes during development.13 Unlike previously reported findings, these results reveal that Coup-TFII favors the transcriptional activation of γglobin genes. Furthermore, they indicate that, when expressed in adult erythroid progenitor cells, Coup-TFII preferentially reactivate the expression of γ- over β-globin genes, thus providing new possibilities for the induction of γ-globin gene expression in adult erythroid cells.

336

Disclosures No conflicts of interest to disclose. Contributions SB and EM wrote the manuscript. Funding This work was supported by a grant from the Canadian Institute of Health Research (CIHR: MOP133420) to EM.

References 1. Weatherall DJ. Phenotype-genotype relationships in monogenic disease: lessons from the thalassaemias. Nat Rev Genet. 2001;2(4):245255. 2. Payen E, Leboulch P. Advances in stem cell transplantation and gene therapy in the beta-hemoglobinopathies. Hematology Am Soc Hematol Educ Program. 2012;2012:276-283. 3. Stamatoyannopoulos G. Control of globin gene expression during development and erythroid differentiation. Exp Hematol. 2005;33(3):259-271. 4. Menzel S, Thein SL. Genetic architecture of hemoglobin F control. Curr Opin Hematol. 2009;16(3):179-186. 5. Johnson KD, Grass JA, Boyer ME, et al. Cooperative activities of hematopoietic regulators recruit RNA polymerase II to a tissue-specific chromatin domain. Proc Natl Acad Sci U S A. 2002;99(18): 11760-11765. 6. Andrieu-Soler C, Soler E. When basic science reaches into rational therapeutic design: from historical to novel leads for the treatment of beta-globinopathies. Curr Opin Hematol. 2020;27(3):141-148. 7. Rodriguez P, Bonte E, Krijgsveld J, et al. GATA-1 forms distinct activating and repressive complexes in erythroid cells. EMBO J. 2005;24(13):2354-2366. 8. Noordermeer D, de Laat W. Joining the loops: beta-globin gene regulation. IUBMB Life. 2008;60(12):824-833. 9. Deng W, Rupon JW, Krivega I, et al. Reactivation of developmentally silenced globin genes by forced chromatin looping. Cell. 2014;158(4):849-860. 10. Liu N, Hargreaves VV, Zhu Q, et al. Direct promoter repression by BCL11A controls the fetal to adult hemoglobin switch. Cell. 2018;173(2):430-442.e17. 11. Sankaran VG, Orkin SH. The switch from fetal to adult hemoglobin. Cold Spring Harb Perspect Med. 2013;3(1):a011643. 12. Siatecka M, Bieker JJ. The multifunctional role of EKLF/KLF1 during erythropoiesis. Blood. 2011;118(8):2044-2054. 13. Fugazza C, Barbarani G, Elangovan S, et al. The Coup-TFII orphan nuclear receptor is an activator of the gamma-globin gene. Haematologica. 2021;106(2):474-482. 14. Ronchi AE, Bottardi S, Mazzucchelli C, Ottolenghi S, Santoro C. Differential binding of the NFE3 and CP1/NFY transcription factors to the human gamma- and epsilon-globin CCAAT boxes. J Biol Chem. 1995;270(37):21934-21941. 15. Liberati C, Cera MR, Secco P, et al. Cooperation and competition between the binding of COUP-TFII and NF-Y on human epsilonand gamma-globin gene promoters. J Biol Chem. 2001;276(45): 41700-41709. 16. Papayannopoulou T, Brice M, Stamatoyannopoulos G. Hemoglobin F synthesis in vitro: evidence for control at the level of primitive erythroid stem cells. Proc Natl Acad Sci U S A. 1977;74(7):2923-2927. 17. Papayannopoulou TH, Brice M, Stamatoyannopoulos G. Stimulation of fetal hemoglobin synthesis in bone marrow cultures from adult individuals. Proc Natl Acad Sci U S A. 1976;73(6):2033-2037. 18. Bottardi S, Aumont A, Grosveld F, Milot E. Developmental stagespecific epigenetic control of human beta-globin gene expression is potentiated in hematopoietic progenitor cells prior to their transcriptional activation. Blood. 2003;102(12):3989-3997. 19. Bottardi S, Ross J, Pierre-Charles N, Blank V, Milot E. Lineage-specific activators affect beta-globin locus chromatin in multipotent hematopoietic progenitors. EMBO J. 2006;25(15):3586-3595. 20. Dulmovits BM, Appiah-Kubi AO, Papoin J, et al. Pomalidomide reverses gamma-globin silencing through the transcriptional reprogramming of adult hematopoietic progenitors. Blood. 2016;127(11): 1481-1492.

haematologica | 2021; 106(2)

REVIEW ARTICLE

Inherited platelet diseases with normal platelet count: phenotypes, genotypes and diagnostic strategy

Ferrata Storti Foundation

Paquita Nurden,1 Simon Stritt,2 Remi Favier3 and Alan T. Nurden1

Institut Hospitalo-Universitaire LIRYC, Pessac, France; 2Department of Immunology, Genetics and Pathology, Uppsala University, Uppsala, Sweden and 3French National Reference Center for Inherited Platelet Disorders, Armand Trousseau Hospital, Assistance Publique–Hôpitaux de Paris, Paris, France

ABSTRACT

Haematologica 2021 Volume 106(2):337-350

nherited platelet disorders resulting from platelet function defects and a normal platelet count cause a moderate or severe bleeding diathesis. Since the description of Glanzmann thrombasthenia resulting from defects of ITGA2B and ITGB3, new inherited platelet disorders have been discovered, facilitated by the use of high throughput sequencing and genomic analyses. Defects of RASGRP2 and FERMT3 responsible for severe bleeding syndromes and integrin activation have illustrated the critical role of signaling molecules. Important are mutations of P2RY12 encoding the major ADP receptor causal for an inherited platelet disorder with inheritance characteristics that depend on the variant identified. Interestingly, variants of GP6 encoding the major subunit of the collagen receptor GPVI/FcRγ associate only with mild bleeding. The numbers of genes involved in dense granule defects including Hermansky-Pudlak and Chediak Higashi syndromes continue to progress and are updated. The ANO6 gene encoding a Ca2+-activated ion channel required for phospholipid scrambling is responsible for the rare Scott syndrome and decreased procoagulant activity. A novel EPHB2 defect in a familial bleeding syndrome demonstrates a role for this tyrosine kinase receptor independent of the classical model of its interaction with ephrins. Such advances highlight the large diversity of variants affecting platelet function but not their production, despite the difficulties in establishing a clear phenotype when few families are affected. They have provided insights into essential pathways of platelet function and have been at the origin of new and improved therapies for ischemic disease. Nevertheless, many patients remain without a diagnosis and requiring new strategies that are now discussed.

Introduction In this review we compare clinical, biological and genetic characteristics of inherited platelet disorders (IPD) with abnormal platelet function but a normal platelet count (listed in Table 1) and highlight anti-ischemic drugs developed based on the discoveries made regarding these disorders.1-10 A low platelet count does not exclude altered platelet function but we refrain from repeating details for disorders such as Bernard-Soulier and gray platelet syndromes included in our companion paper in this journal and summarized in Online Supplementary Table S1.11 After briefly detailing therapy, we discuss the current lessons and perspectives for clinical practice and highlight how it is important for clinicians to constitute a pivot between genetic platforms and basic research.

Glanzmann thrombasthenia With a worldwide distribution, Glanzmann thrombasthenia (GT) is the most common platelet function disorder.12 Most importantly, studies on it led to the characterization of the αIIbβ3 integrin mediator of platelet aggregation (Figure 1); haematologica | 2021; 106(2)

Correspondence: PAQUITA NURDEN paquita.nurden@gmail.com Received: June 13, 2020. Accepted: August 12, 2020. Pre-published: November 5, 2020. https://doi.org/10.3324/haematol.2020.248153

©2021 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

337

P. Nurden et al. Table 1. Principal named inherited platelet function disorders with a normal platelet count

Disease - name - OMIM

Gene - transmission - protein

Platelets - count - volume - granules

Isolated or syndromic - bleeding severity

Principal platelet function defects

Glanzmann thrombasthenia 273800

ITGA2B/ ITGB3 AR (biallelic) αIIbβ3 integrin

N/N/N

Isolated Major bleeding

Absence of platelet aggregation with all agonists. Positive response to ristocetin Clot retraction often defective

N/N/N

Isolated Major bleeding syndrome

Reduced or absent aggregation Positive response to ristocetin Normal response with PMA

N/N/N Leukocytosis

Syndromic with severe infections Osteopetrosis (inconstant) Major bleeding Isolated Bleeding after surgery Isolated Spontaneous bleeding and after lesions Isolated Bleeding can be severe. For AD disease bleeding frequently mild Isolated Bleeding absent or mild.

Absence of platelet aggregation with all agonists Positive response to ristocetin. Defective white cell functions

RASGRP2RASGRP2 (CalDAG-GEFI)- AR (biallelic) RD CalDAG-GEFI 615888 Leukocyte FERMT3 adhesion AR (biallelic) deficiency-III Kindlin-3 (LAD-III syndrome) 612840 Scott syndrome ANO6 (TREM16F) 262890 AR (biallelic) Anoctamin-6 EPHB2-RD EPHB2 618462 AR (biallelic) EPHB2

N/N/N

N/some elongated platelets/N

P2Y12 receptor-RD P2RY12 609821 AR (biallelic) AD P2Y12

N/N/N

GPVI-RD 614201

N/N/N

GP6 AR (biallelic) GPVI

TBXA2 receptorRD 614009

TBXA2R AD TPα HermanskyAR (biallelic) Pudlak syndrome 10 Genes according (HPS): 10 subtypes subtypes OMIM: HPS1,3,4,5,6 HPS1:203300 encoded for HPS2:608233 HPS1,3,4,5,6 HPS3: 614072 HPS2 for AP3B1 HPS4: 614073 HPS7 for Dysbindin HPS5: 614074 HPS8 for BLOS3 HPS6: 614075 HPS9 for Pallidin HPS7: 614076 HPS10 for AP3D1 HPS8: 614077 HPS9: 614171 HPS10: 617050 Chediak-Higashi LYST syndrome (CHS) AR (biallelic) 214500 LYST

Cytosolic phospholipase A2- syndrome 600522

PLA2G4A AR (biallelic) cPLA2

N/N/N

N/N/defect of dense granules

N/N/N

Isolated Post-surgery bleeding Syndromic OCA HPS1-10 -Pulmonary fibrosis and granulomatous colitis, HPS1,4 -Immunodeficiency and neutropenia, HPS2,10 -Mild or severe bleeding syndrome HPS1-10

Mild bleeding Multi-syndromic OCA, recurrent infections, immune dysregulation with lymphoproliferative histiocytosis Isolated Moderate bleeding Recurrent GI ulceration

Biology

Ref

Defects of the αIIbβ3 integrin: Type 1: <5% Type 2: 5% -20% Variant: >20% but non functional No binding of Fg Normal presence but defective activation of αIIbβ3. Normal presence of αIIbβ3. Defect of β1, β2 and β3 activation

12-22

25-28

29-34

Defective platelet-derived thrombin formation Defect of microvesiculation Defective platelet response to collagen, ADP, AA and TXA analogs

Lack of Ca2+-dependent translocation 35-42 of PS to the surface of platelets and red cells Tyrosine phosphorylation of intermediates 43 involved in GPVI and G-protein-coupled receptor mediating signaling

ADP reversible aggregation even at high doses. Response reduced with collagen, AA; reversible. with TRAP. Defective VASP de-phosphorylation after ADP stimulation Absence of response with collagen, Cvx or CRP. Decreased binding of fibrin and other ligands Absence of response to AA and TXA analogs.

Defect of Giα-coupled P2Y12 receptor linked to adenylate cyclase and PI3-kinase activation

44-53

Lack or nonfunctioning of GPVI in complex with FcRγ. Impaired signaling through a pathway involving PLCγ2

54-59

Defect of aggregation to ADP, collagen, AA Defect of secretion whatever the agonist used Decreased or absence of dense granules Defect of CD63

Defective TPα-coupled signaling via 60-64 Gq-protein; signaling pathway involves PLCβ Platelet storage pool disease affecting 67-73 dense granules member of the lysosomal-related organelle family. Gene defects manifested via altered function of BLOC1-3 and AP-3 complexes during organelle biogenesis

Defect of aggregation to ADP, collagen, AA Defect of secretion whatever the agonist Decrease or absence of dense granules

Lysosomal storage disorder affecting platelet dense (d) granules, member of the lysosomal-related organelle family. LYST is a member of the BEACH protein family

Defect of platelet response to AA, collagen, and absence of second wave with ADP

Signaling pathway defect Impaired liberation of AA from membrane phospholipids

67,69,74

84-86

continued on the next page

338

haematologica | 2021; 106(2)

Inherited platelet function disorders continued from the previuos page

Disease - Name - OMIM

Gene - Transmission - Protein

Platelets - count - volume - granules

Isolated or syndromic -bleeding severity

Principle platelet function defects

Biology

Aspirin-like syndrome AR (biallelic and monoallelic) 176805 Ghosal syndrome 231095

PTGS1 AD or AR COX-1

N/N/N

Isolated Aspirin-like Mild bleeding

Defect of platelet response to AA, collagen, no second wave with ADP Normal response to TXA agonists

Signaling pathway defect Defect of conversion of AA into prostaglandin

TBXAS1 AR(biallelic) Thromboxane synthetase

N/N/N

Osteopetrosis Mild bleeding

Defect of platelet response to AA, collagen and no second wave with ADP Normal response to TXA agonists

Signaling pathway defect due to absent thromboxane synthesis

Ref

87-89

OMIM: Online Mendelian Inheritance in Man: Ref: references; RD: related disease; AR: autosomal recessive; AD: autosomal dominant; N: normal; Fg: fibrinogen; CalDAG-GEFI: calcium and diacylglycerol-regulated guanine nucleotide exchange factor I; PMA: phorbol 12-myristate 13-acetate; PS: phosphatidylserine; ADP: adenine disphosphate; AA: arachidonic acid; VASP: vasodilator stimulated phosphoprotein; PI3-kinase: phosphoinositide 3-kinase; Cvx: convulxin; CRP: collagen-related peptide; PLCγ2: phospholipase Cγ2; TXA2: thromboxane A2; TPα: thromboxane receptor α; PLCβ: phospholipase Cβ; OCA: oculocutaneous albinism; BLOC1-3: biogenesis of lysosome-related organelles complex 1-3; GI: gastrointestinal; PLA2: phospholipase-A2; Cox-1: cyclooxygenase-1

this was accompanied by the emergence of a new category of powerful anti-ischemic drugs.13,14

Definition Classically, GT is a severe bleeding diathesis with autosomal recessive transmission caused by an absence of platelet aggregation induced by all physiologic agonists.

Clinical phenotype The clinical phenotype is marked by life-long spontaneous easy bruising, epistaxis and gum bleeding, often requiring platelet transfusions. Gastro-intestinal hemorrhage, menorrhagia and severe bleeding at menarche feature prominently, as does trauma-related hemorrhage. The first signs usually appear at birth, are intense during childhood but tend to decrease in later life. GT is prevalent in ethnic communities.15

Biological phenotype Patients have a normal platelet count and morphology. Most have platelets lacking αIIbβ3 although some express residual amounts but insufficient to support platelet aggregation: these have been termed type I and type II thrombasthenia.12 A third category consists of phenotypic variants in which platelets have reduced to normal amounts of αIIbβ3 unable to function. While type I GT patients possess platelets that fail to promote clot retraction, this can be subnormal or normal for some with residual or variant αIIbβ3. Platelets respond to ristocetin and adhere to subendothelial matrix components under flow yet fail to spread; intriguingly they can attach to fibrin. If bleeding first occurs late in life, care must be taken to exclude acquired GT, which often has an immune origin.16 The severity of bleeding in GT is independent of the subtype.

Genotype Several hundred GT patients and their families have undergone gene sequencing.12,17,18 Mutations extend across both ITGA2B and ITGB3 and include missense mutations, stop codons, small deletions, inserts or duplications, splice defects often with frame-shifts. In classic GT, defects prevent or greatly limit the passage of the affected subunit (and the pro-αIIbβ3 precursor) across the endoplasmic reticulum and the Golgi apparatus. The absence or incorrect conformation of one subunit results in the destruction of both the haematologica | 2021; 106(2)

mutated and the unused normal subunit in maturing megakaryocytes. An exception is the ability of β3 to complex with αv, forming αvβ3, which is present in small amounts in platelets but is a major integrin of endothelial and other cells. For both ITGA2B and ITGB3 defects the bleeding phenotype predominates and the absence of αvβ3 does not lead to major additive defects. Large gene deletions are rare.17 Recent haplotype analysis suggests that the mutational landscape of GT is constantly renewing.19 Missense mutations in domains such as the terminal βpropeller of αIIb and in the βI (or βA) region of β3 have helped to identify interfaces where αIIb and β3 are in contact (Figure 1).19 Rare non-synonymous missense substitutions such as the classic p.D145N/Y and p.R240Q/W substitutions permit αIIbβ3 expression and give rise to variant GT in which the integrin fails to function (Figure 1). These have been vital to the identification of key residues within the metal ion-dependent adhesion site (MIDAS) and adjacent domains of β3 that are essential for Ca2+ and ligand binding.12,13,18 Variant GT can also involve intracellular domains abrogating “inside-out” signaling linked to integrin activation; these include the p.S778P and p.R750ter mutations in β3.18 Resting αIIbβ3 has a bent configuration; on platelet activation; conformational changes emanating from the cytoplasmic domains promote exposure of the ligand-binding epitopes and lead to its extension.20,21 Fibrinogen is the principal ligand for aggregation although other adhesive proteins can participate. Occupied αIIbβ3 undergoes further conformational changes, including clustering, providing “outside-in” signaling pathways essential for thrombus consolidation, platelet spreading and clot retraction.13 Variants within the cytoplasmic domains of αIIb or β3 interfere with the binding of talin and kindlin-3 during early αIIbβ3 activation.12 Rare gain-of-function monoallelic variants in cytoplasmic and membrane proximal domains promote spontaneous conformational changes that result in macrothrombocytopenia, as detailed in our companion paper.11 Significantly, recent data obtained by next-generation sequencing have begun to reveal combinations of variants in different genes that modify the GT phenotype.22

Target for anti-platelet therapy

Because of the importance of αIIbβ3 in the formation of arterial thrombi, a generation of antiplatelet drugs block339

P. Nurden et al. ing αIIbβ3 function was developed some 20 years ago.13 The main drugs are abciximab, a humanized Fab fragment of the murine monoclonal antibody 7E3 (c7E3), integrilin, a KGD-based cyclic heptapeptide and tirofiban a synthetic non-peptide inhibitor; all are given intravenously.13,23 These inhibitors have basically been used in acute coronary syndromes. While the associated bleeding risk has now restricted their use, they are still recommended for bailout in shunt thrombosis or as a bridge to supplement platelet inhibition by orally taken drugs such as antiP2Y12 inhibitors.24

RASGRP2-(CalDAG-GEFI)-related disease In 2014 a missense mutation (p.G248W) in RASGRP2 coding for calcium and diacylglycerol-regulated guanine nucleotide exchange factor I (CalDAG-GEFI) was reported for siblings with a severe bleeding disorder and much reduced platelet aggregation despite normal αIIbβ3 levels.25 Since this report, a characteristic GT-like defect of platelet function has been confirmed worldwide for 23 cases, as recently reviewed by Canault and Alessi.26

Definition Gene variants of RASGRP2 represent a new autosomal recessive inherited bleeding disorder linked to major defects of platelet signaling resulting in impaired αIIbβ3 activation (Figure 2). CalDAG-GEFI activates Rap1, which initiates “inside-out” signaling for αIIbβ3 promoting the binding of talin to shift the integrin into its high-affinity state.27 Both mouse and human platelets lacking functional CalDAG-GEFI undergo a slow but sustained activation that supports the formation of small thrombi at low shear

such as those found in the venous system but not the large thrombi characteristic of arterial blood flow.

Clinical phenotype Bleeding manifestations start early in life and can be severe. Epistaxis is common and most patients have other cutaneous bleeds. Gastrointestinal bleeding has been reported in four patients and menorrhagia in seven out of nine women.26 The bleeding diathesis is not truly syndromic; CalDAG-GEFI is found in other blood cell lineages, vascular cells and the brain striatum and has been implicated in Huntington disease.26 It often occurs in ethnic groups. One case, a male of Jamaican origin, successfully underwent surgery for a brain tumor under recombinant activated factor VII administration despite inefficacy of platelet transfusions.28

Biological phenotype and genetics A critical characteristic is that residual platelet aggregation and elevation of Ca2+ level, which is key for CalDAGGEFI activity, is retained for high doses of some physiological agonists.26 Furthermore, αIIbβ3 fails to activate normally with decreased binding of the PAC1 IgM monoclonal antibody. Phorbol 12-myristate 13-acetate (PMA) activates Rap1 independently of surface receptors and CalDAG-GEFI and PMA-induced platelet aggregation is a diagnostic test for this disease. Interestingly, platelet spreading on a collagen surface is much decreased and while platelets attach to fibrinogen, lamellipodia are not formed. In the profile of RASGRP2 variants updated by Canault & Alessi in 2020,26 12 of the 23 families had a premature stop codon or frameshift mutations while the others possessed missense mutations in functional domains important for CaLDAG-GEFI activity.

Figure 1. Integrin αIIbβ3 and Glanzmann thrombasthenia. The model illustrates the bent resting mature integrin and shows the many interactions between αIIb (green) and β3 (violet). Much information has been obtained from variant forms (noted in blue) that block functional sites in the extracellular and cytoplasmic domains while allowing αIIbβ3 expression. Other rare mutations (in red) give rise to activated integrin. Full information on mutations giving rise to Glanzmann thrombasthenia is found in Nurden & Pillois.18 Fg: fibrinogen; GT: Glanzmann thrombasthenia.

340

haematologica | 2021; 106(2)

Inherited platelet function disorders

Leukocyte adhesion deficiency III syndrome Around 2009, several groups reported that genetic variants of FERMT3 encoding kindlin-3, which is expressed in all hematopoietic cells, caused leukocyte adhesion deficiency III (LAD-III) syndrome.29 Kindlin-3 together with its paralogs kindlin-1 and kindlin-2 are key regulators of cellular functions and cell-matrix interactions but kindlin-3 predominates in platelets.

Definition LAD-III syndrome, with an autosomal recessive inheritance, combines a rare immunodeficiency and bleeding syndrome of extreme severity characterized by leukocytosis, platelet dysfunction and recurrent infections. The genetic defects result in loss of activation of β1, β2 and β3 integrins.

Clinical phenotype Patients are detected when children have GT-like bleeding with severe non-purulent infections and often osteopetrosis. The clinical phenotype is based on the description of a limited number of largely homozygous patients mostly from ethnic Turkish and Arab populations.29

Biological phenotype and genetics As in GT, platelets of patients with LAD-III syndrome fail to aggregate in response to all physiological agonists and PMA.30 This translates into markedly reduced thrombus formation when whole blood is perfused over collagen, von Willebrand factor or fibrinogen-containing microspots.31 αIIbβ3 integrin is present but fails to bind fibrinogen or the PAC-1 monoclonal antibody when platelets are activated (Figure 2). Most disease-causing variants in FERMT3 abrogate kindlin-3 expression.29 As first shown in mice, kindlin-3 is an essential cytoplasmic cofactor for the activation of β1, β2 and β3 integrins.32 Kindlin-3 binds in a phosphorylation-dependent manner to a subterminal NITY motif of the β3 cytoplasmic tail; it coordinates with talin, which binds independently to trigger “inside-out” integrin activation.27,32 Initially there was controversy about the role of a CaLDAG-GEFI mutation in this disorder but this was dispelled when a splice site mutation in RASGRP2 observed in seven subjects with LAD-III syndrome was shown to be in linkage disequilibrium with FERMT3; the two genes are located very closely on chromosome 11q.33,34 Most mutations are nonsense or involve frameshifts.

Figure 2. Loss of CalDAG-GEFI and kindlin-3 function abrogates αIIbβ3 activation. (A) A schematic representation showing examples of agonistic receptors and αIIbβ3 in the resting and activated forms. Intracytoplasmic signaling pathways induced by the binding of appropriate ligands lead to parallel roles of CalDAG-GEFI and kindlin-3 in promoting αIIbβ3 activation. CalDAG-GEFI is critical for RAP1 activation both in the circulation and at sites of vascular injury. It responds to changes in cytoplasmic Ca2+ providing sensitivity and speed to the activation response; its absence leads to a Glanzmann thrombasthenia-like phenotype. Kindlin-3 binds directly to β-integrin cytoplasmic tails in platelets but also in white blood cells, explaining the susceptibility to infections and immune disorders in its absence. (B) Key elements of the syndromes resulting from RASGRP2 and FERMT3 defects are summarized. Bleeding in both cases can be very severe but defects in FERMT3 are syndromic and life-threatening. TRAP: thrombin receptor agonist peptid; PAR: protease activated receptor; Cvx: convulxin; GPVI: glycoprotein VI; TXA2: thromboxane A2; PLCβ/γ: phospholipase Cβ/γ; DAG: diacylglyerol: CalDAG-GEFI: calcium and diacylglycerol-regulated guanine nucleotide exchange factor I; RAP1: RAS-related protein 1; RD: related disease; AR: autosomal recessive; LTA: light transmission aggregometry; ADP: adenosine triphosphate; Col: collagen; Cvx: convulxin; AA: arachidonic acid; PMA: phorbol 12-myristate 13-acetate.

haematologica | 2021; 106(2)

341

P. Nurden et al.

Defect of platelet procoagulant activity: Scott syndrome The original patient reported many years ago in New York had impaired platelet procoagulant activity and fibrin formation but other platelet functions were normal.35 The affected gene ANO6 (also called TMEM16F) was discovered in 2010 when a constitutively active mutant was identified in a cDNA library obtained from a Ba/F3 cell line overexposing phosphatidylserine on the cell surface; a homozygous deleterious variant of ANO6 in a patient with Scott syndrome validated the findings.36

Definition Scott syndrome is characterized by autosomal recessive transmission, a lack of phosphatidylserine exposure and phosphatidylserine-bearing microparticle production from platelets and red blood cells.37,38 Loss of Ca2+-dependent binding of the intrinsic tenase (FVIIIa, IXa, X) and prothrombinase (FVa, Xa, prothrombin) complexes results in reduced thrombin generation.

Definition Recurrent bleeding, normal platelet counts but abnormal function were linked to a variant of EPHB2, encoding a member of the EPH receptor family of transmembrane tyrosine kinases. The molecular basis of this IPD confirms a specific role for EPHB2 in platelet function via tyrosine phosphorylation of intermediates involved in GPVI and G-protein–coupled receptor-mediated signaling.

Clinical phenotype Both affected family members showed excessive spontaneous subcutaneous and heavy bleeding upon minor wounds.43 One sibling developed anemia after chronic gastrointestinal bleeding requiring iron supplementation. Due to its rarity the clinical phenotype remains poorly defined.

Biological phenotype

Only six cases of Scott syndrome have been detailed and all are women.39 Bleeding is predominant after surgery or trauma; menorrhagia and post-partum hemorrhage are marked. Although spontaneous mucocutaneous bleeding is rare, it was severe in one young woman and included epistaxis, gum and gastrointestinal bleeding.39

Ephrin receptors consist of a N-terminal glycosylated ligand-binding domain, a transmembrane region, and an intracellular kinase domain. Interactions largely occur between neighboring cells; studies using mice revealed a role in thrombus formation and in both contact-dependent and -independent signaling.43 Platelets from our family showed decreased aggregation and secretion in response to ADP, arachidonic acid, collagen, and analogs of thromboxane A . The mutation did not affect EPHB2ephrin interactions but phosphorylation of Lyn, Syk and FcRγ, the initial steps in GPVI signaling and of Src were drastically impaired as was inside-out αIIbβ3 activation.

Biological phenotype

Genotype

The major diagnostic clues are elevated residual levels of prothrombin in sera and a marked defect of platelet procoagulant activity. Full diagnosis requires flow cytometry measurements of platelet phosphatidylserine exposure (annexin V-binding) and microparticle release upon activation by thrombin in combination with collagen or by the ionophore A23187.40 Loss of mitochondrial membrane potential, a major factor in phosphatidylserine exposure, late phase procoagulant ballooning and capping of adhesive proteins all still occur but may be reduced. Caspasedependent apoptotic or necrotic pathways of phosphatidylserine expression are normal.40 Platelet numbers, morphology, adhesion, secretion and aggregation are largely unaffected.

A homozygous missense variant (p.R745C) in the EPHB2 gene was identified by whole exome sequencing in both siblings; their asymptomatic parents were heterozygous. The p.R745C substitution is located within the tyrosine kinase domain.

Clinical phenotype

Genotype Gene variants of ANO6 encoding a multi-pass transmembrane protein that is an essential component of the Ca2+-dependent “scramblase” activity cause Scott syndrome.39 For the original patient a homozygous variant at a splice acceptor site in intron 12 resulted in a premature stop codon.36 Two French siblings showed compound heterozygosity for a large deletion of exons 1-10 in one allele and a premature stop codon on the other.41 Compound heterozygosity with premature termination was also seen in a Welsh family.42 All mutations result in abrogated TMEM16F expression.

P2Y12 receptor-related disorder The use of ticlopidine as an anti-platelet agent in the 1970s showed the benefit of blocking the interaction between ADP and platelets in anti-ischemic treatment.44 The recognition in 1992 of an Italian family with a ticlopidine-like platelet function profile and in 1995 of a French family with a similar defect was crucial to identifying how ADP activates platelets.45,46 Some 20 years later only a few such families have been identified.

Definition P2Y12-related disease affects the full aggregation response of platelets to ADP and the stabilization of aggregates. Human platelets possess two ADP receptors that act in synergy: (i) P2Y1, a G -coupled receptor, which initiates platelet shape changes and ADP-induced aggregation through the mobilization of internal Ca2+ stores, and (ii) P2Y12, a G -coupled receptor linked to adenylyl cyclase and phosphoinositide 3-kinase activation (Figure 3). q

Clinical phenotype

Ephrin type-B receptor 2-related disorder Recently, we demonstrated that defects in Ephrin typeB receptor 2 (EPHB2) cause a bleeding syndrome and IPD in two siblings from a consanguineous French family.43 342

The disease is inherited in an autosomal recessive manner and is manifested clinically as moderate to severe mucocutaneous bleeding, including after surgery.45,46 Beyond the classical forms, a family with severe hemorrhage has been recently observed with autosomal domihaematologica | 2021; 106(2)

Inherited platelet function disorders

nant transmission; other cases with a monoallelic form have mild bleeding.47,50

Biological phenotype ADP, even at high doses, fails to induce full and irreversible platelet aggregation. With other agonists such as thrombin receptor agonist peptide (TRAP), the snake venom protein convulxin or ristocetin, platelet aggregation may remain partially reversible. A key feature is defective GiÎą-mediated inhibition of platelet adenylyl cyclase by ADP with loss of vasodilator-stimulated phosphoprotein (VASP) de-phosphorylation, readily evaluated in a diagnostic test.48 Variants can result in qualitative or quantitative defects of P2Y12, the latter can also be attributed to defective receptor recycling.

Genotype The patient that we first described in 1995 has a two nucleotide deletion in the coding region of P2RY12 moving the reading frame before the introduction of a premature stop (p.I240fs*29).10 Subsequently described mutations with autosomal recessive transmission include those of the initially reported Italian family, heterozygous for two missense mutations, p.R256Q and p.R265W (Figure

3).48,49 Here, two asymptomatic family members carried the heterozygous p.R265W variant. This is important because recently a single allele p.R265P variant, affecting the same amino acid but with a different substitution, was associated with a severe phenotype and autosomal dominant transmission in an unrelated family (Figure 3).47 Significantly, platelet mRNA for this allele was three-fold overexpressed, limiting wild-type homodimer formation (key to ADP-receptor signaling) and introducing a dominant-negative effect.47 Other mutations with a single affected P2RY12 allele and mild bleeding in families with autosomal dominant transmission are shown in Figure 3.48 Missense mutations allowing normal P2Y12 synthesis and affecting domains close to or part of the ADP-binding pockets cause qualitative defects (Figure 3).48,50 Special mention should be made of a novel cytoplasmic domain p.P341A variant affecting the PDZ-binding domain and causing abnormal endosomal recycling of platelet internal pools leading to a surface deficit of P2Y12.51 Interestingly, p.R122C and p.K174E P2RY12 variants have been associated with an intronic polymorphism within FR2 (encoding PAR-1) in a patient with chronic bleeding and in a patient with type 1 von Willebrand disease showing that hemorrhage can be part of a complex trait.52,53

Figure 3. G-protein-coupled ADP receptors and signaling pathways highlighting P2RY12 variants and their different subgroups. (A) A schematic representation of ADP-induced signaling pathways. Binding of extracellular ADP to the G -coupled P2Y1 receptor initiates aggregation whereas the G -coupled P2Y12 receptor enhances and sustains the platelet response. Specific signaling pathways are involved with PLCÎ˛ and PI3K enabling both RAP1 activation and RASA3 inactivation; both are required for the formation of a stable hemostatic thrombus. The release of ADP from dense granule storage pools after protein kinase C (PKC) activation contributes to the stabilization of thrombi. (B) The schematic representation shows the structure of P2Y12; variants reported as causal for a bleeding syndrome are highlighted (red spots). (C) Summarizes the genotype/phenotype relationship when the variants are regrouped according to bleeding severity and their mode of transmission. The first group consists of patients with autosomal recessive transmission and platelets showing defects of both the binding of ADP analogs and/or receptor cycling. The p.H187Q variant located in the fifth transmembrane domain has normal receptor expression but suppressed function.50 Interestingly p.R256 has a side chain that inserts into a hydrophobic pocket and in docking models it potentially makes contact with the phosphate groups of ADP;52 p.R265W impairs receptor activation and probably alters the conformational state of the receptor. Significantly, a severe phenotype-monoallelic form has a different amino acid substitution on the same residue, p.R265P associated with an increased expression of mutated mRNA and an impaired wild-type homodimer formation, possibly accounting for the bleeding severity. The third subgroup has a mild phenotype and monoallelic expression of mutations that affect receptor function: note that p.P341A located in the PDZ intracellular domain is associated with both abnormal ligand binding and a defect of re-sensitization of the receptor after ADP agonist-induced desensitization. (D) Electron microscopy of platelet aggregates stimulated with 10 mM ADP examined at the peak of aggregation. For the control the platelets are in close contact with some of the cells having lost their granule content. For our patient with the p.I240fs*29 mutation, platelets remain loosely bound and their granule contents are still present, as shown by the immunogold (black dots) localization of fibrinogen. Îąq

haematologica | 2021; 106(2)

343

P. Nurden et al.

Antiplatelet therapy P2Y12 identification was key to the development of nextgeneration therapies including the orally available thienopyridines (clopidogrel and prasugrel). These are prodrugs requiring cytochrome P450-based conversion to a short-lived active metabolite that binds covalently to P2Y12 although drug-resistance to clopidogrel can affect up to a third of patients.14 Ticagrelor and cangrelor are direct-acting and rapidly reversible antagonists; cangrelor given intravenously has the advantage of acting immediately.

antibody blocking GPVI, ACT017, have exhibited excellent safety and tolerability profiles in a phase 1 study in humans.65 An exciting challenge for anti-GPVI drugs will be to block thrombus build up through inhibition of the GPVI-fibrin interaction while preserving the initial interaction with collagen. Anti-GPVI agents may also have a role in cancer therapy by inhibiting platelet-tumor cell interactions; galactin-3 is a novel counter-receptor for GPVI on tumor cells.66 Attempts at treating ischemic disease based on inhibition of TXA2R have largely failed compared to the efficiency of aspirin and related drugs.14

Other platelet primary receptor defects Classic dense granule defects Glycoprotein VI-related disease Rare inherited autosomal recessive defects of the GP6 gene cause a mild bleeding syndrome. GPVI is a receptor for collagen as well as other extracellular matrix components and, as recently identified, fibrin.54,55 A member of the immunoglobulin superfamily, GPVI forms a non-covalent complex with FcRγ and acts in synergy with integrin α2β1. Early evidence of an inherited GPVI deficiency came from a Japanese woman with lifelong although mild mucocutaneous bleeding and platelets refractory to collagen despite responding to other agonists; immunological studies confirmed the specific absence of GPVI.56 The first genotyped patients with GP6 variants belonged to families from France and Belgium with autosomal recessive compound heterozygous mutations.57,58 However, most studies have centered on a cluster of patients from Chile with a homozygous 2 bp insertion followed by a stop at position 242.59 All evidence suggests a founder effect. The Chilean patients had mild bleeding and heterozygous family members were asymptomatic. A defective platelet interaction with fibrin and immobilized fibrinogen helps to explain the diminished thrombus build up under flow.58

Thromboxane prostanoid type α receptor-related disease

Hermansky-Pudlak syndrome Hermansky-Pudlak syndrome (HPS) is named after the Czechoslovakian physicians who in 1959 described two patients with oculocutaneous albinism, prolonged bleeding and pigmented macrophages in the bone marrow.67 Many of the advances regarding HPS have been linked to the numerous mouse and other animal models of storage pool diseases.68,69

Definition HPS is a heterogeneous autosomal recessive multisystem disorder resulting from defects of ten genes encoding proteins essential for the biogenesis of platelet dense granules, a member of the lysosomal-related organelle family.67-69 Association with albinism reflects defects in the biogenesis of melanosomes in the melanocytes of the skin, hair and choroid of the eye while pigment epithelial cells of the retina can also be affected.

Clinical phenotype

The TBXA2R gene encodes the thromboxane prostanoid type α receptor (TPα), the major thromboxane A2 (TXA ) receptor on platelets.60 Its activation leads to Ca2+ mobilization and platelet aggregation. It associates mainly with G and principally modulates platelet activation through the PLCβ pathway. Platelet aggregation is characterized by a decreased response to arachidonic acid or TXA analogs such as U46619 and I-BOP.61 From as early as 1994, Japanese families were described with a bleeding disorder due to a non-responsive TPα receptor.62,63 A p.R60L variant in the first intracytoplasmic loop of TPα with autosomal dominant transmission predominated, but it was homozygous in one member with mild mucocutaneous bleeding. Globally, three more families were later reported with autosomal dominant transmission; two had missense mutations with a dominant-negative effect linked to impaired receptor dimerization while a third had a frameshift variant, a stop codon and reduced TPα expression.60,64 2

Antiplatelet therapy The apparent preservation of hemostasis in the absence of GPVI has prompted research for anti-ischemic drugs and, in particular, treatment for stroke.14 Revacept is a recombinant soluble dimeric GPVI-Fc fusion protein blocking GPVI-reactive sites in collagen and is the object of ongoing clinical trials. Fab fragments of a monoclonal 344

Much information has been obtained from studies on disorders of platelet lysosome-related organelles.67,68

The bleeding diathesis includes easy bruising, epistaxis, gingival bleeding, menorrhagia, postpartum hemorrhage and prolonged bleeding after surgery.67 Platelet transfusions prevent the hemorrhagic risk during surgery, while tranexamic acid is often used preventively or to stop moderate bleeding. Oculocutaneous albinism involves hypopigmentation of the skin and hair associated with characteristic ocular findings. Granulomatous colitis, neutropenia, immunodeficiency and fatal pulmonary fibrosis can feature according to the subtype.67-69

Biological phenotype Platelet aggregation is characterized by an impaired second wave.68,69 The specific absence of dense granules means that the platelet secretory pool of ADP, ATP, serotonin, calcium, and polyphosphate is reduced or lacking. Dense granules can be evaluated by whole mount electron microscopy due to their opacity to electrons (Figure 4). Total platelet ADP and ATP and the secretion of ATP during platelet aggregation is now mostly measured by bioluminescence. Quantifying mepacrine uptake by flow cytometry is an indirect method for assessing dense granule content. Granule secretion can also be measured by flow cytometry based on the translocation of P-selectin and lysosomal-associated membrane protein (LAMP-3, CD63) to the platelet surface; exposure of LAMP-3 can be spontaneous. haematologica | 2021; 106(2)

Inherited platelet function disorders

Genotype Identification of pathogenic variants for HPS involves ten subtypes (HPS1-10) and ten genes: AP3B1, AP3D1, BLOC1S3 encoding for BLOS-3, BLOC1S6 encoding for pallidin, DTNBP1 encoding for dysbindin, HPS1, HPS36.69-71 All of the genes encode subunits organized in four multi-subunit protein complexes, named biogenesis of lysosome-related organelles complex (BLOC)-1, -2 and -3 and the adaptor protein-3 complex (AP-3).67-71 The HPS phenotype is related to the BLOC complex affected (Figure 4). BLOC-1 consists of eight subunits of which dysbindin (HPS7), BLOS-3 (HPS8) and pallidin (HPS9) are mutated in rare patients with a mild or absent bleeding diathesis and variable hypopigmentation. HPS3, HPS5, and HPS6 are subunits of BLOC-2; mutations in these genes are more frequent than those in the BLOC-1 and affected individuals have a mild clinical phenotype with a bleeding syndrome and hypopigmentation only. HPS1 and HPS4 are subunits of BLOC-3; mutations in these cases are predominantly associated with severe manifestations (severe oculocutaneous albinism, pulmonary fibrosis, granulomatous colitis). BLOC-3 defects are the most frequent form of HPS with p.His497Glnfs*90 in HPS1, the

first HPS mutation to be discovered; this form is common in Puerto Rico.67 The AP-3 complex has four subunits with AP3B1 encoding Î˛3A mutated in HPS-2 and AP3D1 encoding AP-3d mutated in HPS-10. AP-3 is important for dense granule biogenesis and protein delivery to the forming granule (e.g., the serotonin transporter, VMAT2).69-72 Variants of AP3B1 are said to cause HPS type II. Interestingly, Enders et al. detected a homozygous nonsense mutation in AP3B1 together with a heterozygous bystander RAB27A mutation in a child with bleeding, albinism, developmental delay and a susceptibility to infections; HPS type II was diagnosed rather than Griscelli syndrome as would indicate a RAB27A mutation.72 HPS-10 is a very rare form of HPS characterized by partial oculocutaneous albinism, bleeding, neutropenia and immunodeficiency. Defective protein disulfide isomerase release from platelets and endothelial cells in HPS contributes to the impaired thrombus formation and could represent a novel target for antithrombotic drugs.73

Chediak-Higashi syndrome The most important clinical problem of patients with Chediak-Higashi syndrome is immunodeficiency with life-

Figure 4. Hermansky-Pudlak syndrome. Syndromic platelet defect resulting from the abnormal biogenesis of dense granules with an autosomal recessive inheritance. (A) Summarizes the genotype/phenotype relationship of different types of Hermansky-Pudlak syndrome (HPS) regrouped according to the four multi-protein complexes involved in different steps of lysosome-related organelle (LRO) biogenesis including vesicle formation and/or trafficking: BLOC1, 2 and 3 and AP-3 with each containing several subunits. Variants in HPS 7, 8 and 9 are causal in BLOC1 but with mild clinical consequences. For BLOC2, variants in HPS 3, 5 or 6 give syndromes of moderate severity; they are rare but a variant of HPS 3 occurs in Puerto Rican communities. The most frequent causes of HPS are BLOC3 defects (variants in HPS 1 or 4). The syndromes are severe and are associated with lung fibrosis and gastro-intestinal defects; in Puerto Rico a community is affected by a common HPS1 variant. In the last group, defects of AP-3 are associated with neutropenia and infections. (B) A scheme illustrates the major steps of granule biogenesis. Dense granules may derive from the endolysosomal system, including either early or late endosomes. As secretory granules the membrane constituents including membrane transporters come from the Golgi complex. A network of interconnected and functionally distinct tubular subdomains transports their cargoes along microtubule tracks from the endosomes. Tubules ferry contents from the trans-Golgi-network and the plasma membrane to the LRO, a system that requires the coordination of numerous effectors. Components stored in dense granules or associated with their membranes are shown. (C) Platelets from a control and a patient with HPS3 examined by electron microscopy as whole mounts are shown. The dense granules observed in controls as black spots are absent from the patientâ&#x20AC;&#x2122;s platelet. AP-3: adaptor protein-3; BLOC: biogenesis of lysosome-related organelles complex; LRO: lysosome-related organelle; OCA: oculocutaneous albinism; MVB: multivesicular body.

haematologica | 2021; 106(2)

345

P. Nurden et al.

threatening infections. There is mild bleeding with a partial to severe reduction of dense granules.67,69,74 It is a rare but fatal autosomal recessive disorder with multiple clinical features, including hypopigmentation of skin and hair, reduction of leukocytes, giant organelles in circulating granulated cells, and neurological dysfunction. Recurrent infections are associated with defective function of natural killer cells. Immune dysregulation often leads to lymphoproliferative histiocytosis with an accelerated phase that determines the prognosis.73 Platelet function defects resemble those seen in HPS. The LYST gene (also called CHS1) alone is responsible for this disease and a wide variety of mutations have been described.67,69,74 Chediak-Higashi syndrome is particularly prevalent in Puerto Rico.67,69 LYST is a member of the BEACH protein family and is required for sorting resident endosomal proteins into late multivesicular endosomes (Figure 4B).71,74

Other defects involving platelet granules As well as the above disorders, qualitative defects can lead to decreased dense granule content, inability to secrete their contents or combined dense- and Îą-granule

deficiency.71,75,76 An absent second wave of aggregation in response to ADP and collagen is characteristic. This can be secondary to mutations of genes for membrane receptors: P2RY12, GP6, TBXA2R, EPHB2, G6b-B; or genetic defects often accompanied by thrombocytopenia (Online Supplementary Table S1).11 Applying next-generation sequencing procedures targeting 329 preselected genes, Leo et al. analyzed 18 patients including six with secretion defects, highlighting 49 genes not previously known to be causal of IPD.77 Gorski et al. performed a similar study using whole exome sequencing of 14 patients with bleeding caused by primary platelet secretion defects.78 Both studies exposed the complications of whole exome sequencing and next-generation sequencing when isolated gene variants are identified as potentially causal without statistical validation or comprehensive biological studies or usage of mouse models. A specific condition with defective granule secretion from platelets is familial hemophagocytic lymphohistiocytosis, a rare primary immunodeficiency syndrome in which autosomal recessively inherited variants of UNC13D, STX11 and STYXBP2 can result in defects in the secretory machinery of dense bodies and Îą-granules as well as lysosomes.79-82

Figure 5. Schema with management options for patients with inherited defects of platelet function. It is important to limit risks in daily life. It is also important to provide each patient with a card detailing the type of disorder and proposing the medication to be used according to the bleeding risk, in particular when platelet expert medical centers are unavailable to take charge of the patientâ&#x20AC;&#x2122;s needs. While the main therapeutic approaches for all disorders, including those patients with Glanzmann thrombasthenia (GT), RASGRP2 (CalDAG-GEFI)-related disease, and leukocyte adhesion deficiency III (LAD-III) syndrome for whom bleeding can be severe, are similar, some situations (e.g., isoantibody formation in GT) or infections in LAD-III syndrome need special care. Surgery requires a multidisciplinary consensus to evaluate the risk of bleeding, the benefit-risk ratio of prophylaxis, and to assess therapeutic efficiency. Pregnancy is always a challenge, from women with mild bleeding risks for whom minimal measures are needed to those with a severe bleeding history and for whom maximal prophylactic care is required. Here again, the management must be planned between obstetricians, anesthetists and hematologists. The contraindication of epidural or spinal anesthesia for those with mild risk remains a difficult problem, as is the choice of vaginal delivery or Cesarean section for the most severe forms. In the figure, it is noted that for mild bleeding risk, an antifibrinolytic medication or desmopressin is preferred as the first line of bleeding prevention and that platelet transfusions should be considered only when these approaches are insufficient. There is a risk of allo- or iso-immunization which, if it occurs, can be a lifelong handicap, and recombinant activated factor VII is increasingly recommended. Finally, stem cell transplantation is reserved for young patients with recurrent, severe bleeding that fails to respond to standard treatments and when the survival of the patient is at risk. Gene therapy is not yet an option for the disorders covered in this review but will probably become available in the future.

346

haematologica | 2021; 106(2)

Inherited platelet function disorders

Cytosolic signaling pathways Rao and his colleagues were pioneers in identifying candidate proteins in patients with mild bleeding diatheses deficient in key signaling pathways and second messengers.83 Gαq, PLC-β2, as well as defects in Ca2+-mobilization and protein kinase C-induced phosphorylation were highlighted. Abnormalities extended to granule secretion as well as αIIbβ3 activation. However, most of the early patients still await full genetic analysis. We have chosen to highlight three genes involved in arachidonic acid metabolism in platelets. The first is a defect of PLA2G4A encoding cytosolic PLA2 in three families. In the first family compound heterozygous missense mutations induced marked structural changes within cPLA2. In the second family heterozygous p.S111P and p.R485H missense variants respectively affected cPLA2 trafficking and abrogated the enzyme’s catalytic activity; while for the third family a homozygous p.D575H PLA2G4A variant was present.84-86 In all three cases, platelet aggregation and dense granule secretion with collagen and ADP were impaired; gastrointestinal bleeding was a feature. An inherited defect of cyclooxygenase-1 (COX-1) is more specifically linked to defective platelet aggregation with arachidonic acid and absence of the second wave response to epinephrine, ADP and collagen. While failing to respond to arachidonic acid, platelets aggregate normally with TXA agonists. Several such patients with mild bleeding have been reported either with autosomal dominant or, more recently, autosomal recessive transmission of variants in PTGS1.87-89 Patients may lack COX1 or have a functionally defective but normally expressed enzyme. Bastida et al., using a high-throughput sequencing gene panel, found a heterozygous variant with an aspirin-like platelet functional defect linked to a heterozygous p.N143S missense mutation in PTGS1.88 Very recently, whole genome sequencing helped to identify a homozygous missense variant (p.W322S) in a large consanguineous family of Iranian descent with mild bleeding.89 The third disease is Ghosal syndrome, a disorder with increased bone density that was linked to a deletion mapped to chromosome 7q33/34. The deletion includes TBXAS1, the gene for thromboxane synthase, a terminal enzyme in the arachidonic acid cascade converting prostaglandin H2 into TXA . A key fiding was homozygous missense mutations alone in TBXAS1 in consanguineous families with Ghosal syndrome in ethnic groups; the affected residues were all highly conserved.90 Platelets from the patients had reduced aggregation and secretion in response to arachidonic acid; in contrast the response to U46619 directly activating TPα was normal. A role for TXA in bone remodeling is also indicated. Somewhat surprisingly, none of the patients had a bleeding diathesis (distinguishing them from patients with congenital defects of TBXA2R). 2

mimetics is not applicable.11 As detailed in Figure 5, strategies involve the maintenance of a lifestyle that minimizes risk and being prepared in case of an emergency. Procedures range from local measures to the use of platelet transfusions or recombinant activated FVII to stop major bleeding as well as the establishment of and adherence to standardized protocols for prophylaxis prior to surgery or childbirth.15,91-93 A special case is type I GT in which the absence of αIIbβ3 can lead to the formation of isoantibodies, making patients refractory to normal platelets.

Conclusions Of the disorders covered by this review, GT is the most frequent and widely screened disease.12,15 Genotyping has now identified several hundred disease-causing mutations in a constantly changing spectrum, so the classification of patients into type I and type II subgroups based on residual platelet αIIbβ3 content should be updated to take into account mutation analysis.18,19 From 2001, the year in which P2RY12 was first characterized,10 newly identified genes include FERMT3, ANO6, RASGRP2 and EPHB2 thereby expanding the gene repertoire for IPD.25,26,33,34,36,43 The development of next-generation sequencing platforms targeting IPD genes,6,88 together with widening access to whole exome sequencing,4,7,8 and now whole genome sequencing,9 has both increased the number of patients with a genetic diagnosis and the speed at which this is performed. The multiplication of data has underlined the great diversity of variants responsible for gene dysfunction, the rarity of hot-spot mutations, the frequency of compound heterozygosity within the same gene (limiting the notion of consanguinity for recessive diseases); and raised the likelihood of additive effects of heterozygous variants on different genes modulating the phenotype. It is also worthy of mention that some inherited thrombocytopenias with associated functional platelet defects (Online Supplementary Table SI) may not have complete penetrance and some patients may therefore not be thrombocytopenic (e.g., those with mutations of RUNX1 and FLI1).11 The increasing understanding of the gene spectrum causing IPD is leading to better care of patients and it should be underlined how major anti-ischemic treatments have been initiated or improved as a result of this.

Therapy The management of IPD involves both prevention of bleeding and treatment of mild and severe bleeding episodes and differs little for disorders of platelet count with the exception that the recent use of thrombopoietin haematologica | 2021; 106(2)

Future strategies Despite the progress in gene identification a significant proportion of patients still fail to be genotyped. One successful strategy for the identification of causal genes in patients with inherited thrombocytopenia was by comparison with mouse models with similar phenotypes;11, 69 Another approach is ongoing screening of local whole exome or whole genome sequencing databases of patients for variants of each newly published IPD gene. A third strategy involves screening large families or those with consanguinity, as done for RASGRP2 and EPHB2, or statistical analysis of large cohorts of patients, aligning variant detection in unrelated families with a similar phenotype.26 The final proof that variants in new genes are causal will still require their expression in heterologous cells, the use of in silico modeling to assess changes in protein structure, 347

P. Nurden et al.

or their reproduction in mouse, zebrafish or other animal models.3,4,94 When no genetic diagnosis has been achieved with standard approaches, gene analysis should be expanded to evaluate complex copy number variations, deep variants in non-coding regions and secondary defects involving the action of microRNA or retrotransposons, as we have detailed elsewhere.11 Currently the diagnosis of IPD is based largely on the concept of a monogenic heritable platelet disease, but a proportion of these diseases may be di- or polygenic in origin. Families with unexplained heritability may have unrecognized variants or other changes in regulatory regions in combinations that only advanced technologies may help to resolve. Thus in a large cohort of patients with primary immunodeficiency, genome-wide association studies have revealed loci that modulate platelet function with the co-localization of and interplay between novel high-penetrance monogenic variants and common variants.95 It is reassuring that IPD without thrombocytopenia are not associated with a propensity to develop hematological malignancies or bone marrow aplasia or fibrosis, as in some inherited thrombocytopenias such as RUNX1-related disease and ETV6-related disease. Current advances suggest the need to redefine the circuit for patient screening from the first consultation. An information-gathering first consultation, blood cell counts and basic coagulation screening remain essential as is assigning a bleeding score and determining whether the platelet function abnormalities are isolated or syndromic.96 But flow charts with extensive biological testing in order to select the gene(s) to analyze need to be revisited and the choice of extensive biological testing before genotyping questioned. Platelet aggregation, flow cytometry and the examination of blood smears provide early information to guide the diagnosis.75,97 Nevertheless, when clinical and biological data are in favor of an IPD, in selected cases upfront DNA analysis may save time, money and improve patients’ care: however longitudinal studies are required to substantiate this expectation. For patients without access

References 1. Nurden P, Nurden AT. Congenital disorders associated with platelet dysfunctions. Thromb Haemost. 2008;99(2):253-263. 2. Favier R, Raslova H. Progress in understanding the diagnosis and molecular genetics of macrothrombocytopenias. Br J Haematol. 2015;170(5):626-639. 3. Heremans J, Freson K. High-throughput sequencing for diagnosing platelet disorders: lessons learned from exploring the causes of bleeding disorders. Int J Lab Hematol. 2018;40(Suppl):89-96. 4. Lentaigne C, Freson K, Laffan MA, et al. Inherited platelet disorders: towards DNAbased diagnosis. Blood. 2016;127(33):28142823. 5. Johnson B, Lowe GC, Futterer J, et al. Whole exome sequencing identifies genetic variants in inherited thrombocytopenia with secondary qualitative function defects. Haematologica. 2016;101(10):1170-1179. 6. Simeoni I, Stephens JC, Hu F, et al. A highthroughput sequencing test for diagnosing inherited bleeding, thrombotic and platelet disorders. Blood. 2016;127(23):2791-2803. 7. Downes K, Megy K, Duarte D, et al.

348

to a specialized platelet center, sending blood and DNA samples to expert centers is recommended. Validation of novel variant(s) on next-generation sequencing targeted gene platforms or after whole exome or whole genome sequencing requires confirmation by co-segregation or other studies as discussed above and it is here that selected biological phenotyping may be necessary. As part of this roadmap, national centers regrouping platelet experts have a key inter-connecting role between the patients, local hematology laboratories and genomic platforms. In France, our creation of a Reference Center for IPD, led to the organization, on a national basis, of platelet studies with European links to next-generation sequencing and the preparation of personalized followup both for the patient and the local general practitioner. In this context it is indispensable that the medical personnel directly responsible for the patient remain in control of the diffusion of information both to the patient and in the scientific literature and especially take into account ethical considerations.98 In everyday clinical practice a patient-centered approach governed by the overarching principle of patient utility versus in-depth, detailed laboratory and genetic characterization is acceptable and in generally sufficient for patients’ care.99 On the other hand, our understanding of these disorders cannot be expanded without advanced research involving modern technologies. Here, informed explicit collaboration of the patient should be obtained. Disclosures No conflicts of interests to diclose. Contributions All the authors participated in the manuscript preparation. PN and ATN supervised the writing of the text. Acknowledgment We thank Xavier Pillois for his contribution to the section on GT and for the Pymol model of αIIbβ3.

Diagnostic high-throughput sequencing of 2,396 patients with bleeding, thrombotic and platelet disorders. Blood. 2019;134(23): 2082-2091. 8. Megy K, Downes K, Simeoni I, et al. Curated disease-causing genes for bleeding, thrombotic, and platelet disorders: Communication from the SSC of the ISTH. J Thromb Haemost. 2019;17(8):1253-1260. 9. Gunay-Aygum M, Zivony-Elboum Y, Gumruk F, et al. Gray platelet syndrome: natural history of a large patient cohort and locus assignment to chromosome 3p. Blood. 2010;116(23):4990-5001. 10. Hollopeter G, Jantzen AM, Vincent D, et al. Identification of the platelet ADP receptor targeted by antithrombotic drugs. Nature. 2001;409(6817):202-207. 11. Nurden A, Nurden P. Inherited thrombocytopenias: history, advances and perspectives. Haematologica. 2020;105(8):1-16. 12. Nurden AT, Fiore M, Nurden P, Pillois X. Glanzmann thrombasthenia: a review of ITGA2B and ITGB3 defects with emphasis on variants, phenotypic variability, and mouse models. Blood. 2011;118(23):59966005. 13. Coller BS, Shattil SA. The GPIIb/IIIa (integrin αIIbβ3) odyssey: a technology-driven

saga of a receptor with twists, turns, and even a bend. Blood. 2008;112(8):3011-3025. 14. Mackman N, Bergmeier W, Stouffer GA, Weitz JI. Therapeutic strategies for thrombosis: new targets and approaches. Nat Rev Drug Disc. 2020;19(5):333-352. 15. Botero JP, Lee K, Branchford BR, et al. ClinGen Platelet Disorder Variant Curation Expert Panel. Haematologica. 2020;105(4): 888-894. 16. Nurden AT. Acquired Glanzmann thrombasthenia: from antibodies to anti-platelet drugs. Blood Rev. 2019;36:10-22. 17. Nurden AT, Pillois X, Fiore M, et al. Expanding the mutation spectrum of the αIIbβ3 integrin in Glanzmann thrombasthenia: screening of the ITGA2B and ITGB3 genes in a large international cohort. Hum Mutat. 2015;36(5):548-61. 18. Nurden AT, Pillois X. ITGA2B and ITGB3 gene mutations associated with Glanzmann thrombasthenia. Platelets. 2018;29(1):98101. 19. Pillois X, Nurden AT. Linkage disequilibrium amongst ITGA2B and ITGB3 gene variants in patients with Glanzmann thrombasthenia confirms that most disease-causing mutations are recent. Br J Haematol. 2016;175(4):686-695.

haematologica | 2021; 106(2)

Inherited platelet function disorders 20. Takagi J, Petre BM, Walz T, Springer TA. Global conformational rearrangements in integrin extracellular domains in outside-in and inside-out signaling. Cell. 2002;110(5): 599-611. 21. Zhu J, Zhu J, Springer TA. Complete integrin headpiece opening in eight steps. J Cell Biol. 2013;201(7):1053-1068. 22. Guillet B, Bayart S, Pillois X, Nurden P, Caen JP, Nurden AT. A Glanzmann thrombasthenia family associated with a TUBB1-related macrothrombasthenia. J Thromb Haemost. 2019;17(12):2211-2215. 23. Jamasbi J, Ayabe K, Goto S, Nieswandt B, Peter K, Siess W. Platelet receptors as therapeutic targets: past, present and future. Thromb Haemost. 2017;117(7):1249-1257. 24. Roule V, Agueznai M, Sabatier R, et al. Safety and efficacy of IIb/IIIa inhibitors in combination with highly active oral antiplatelet regimens in acute coronary syndromes: a meta-analysis of pivotal trials. Platelets. 2017;28(2):174-181. 25. Canault M, Ghalloussi D, Grosdidier C, et al. Human CalDAG-GEFI gene (RASGRP2) mutation affects platelet function and causes severe bleeding. J Exp Med. 2014;211(7): 1349-1362. 26. Canault M, Alessi MC. RASGRP2 structure, function and genetic variants in platelet pathophysiology. Int J Molec Sci. 2020; 21(3):1075. 27. Stefanini L, Bergmeier W. RAP GTPAses and platelet integrin signaling. Platelets. 2019;30(1):41-47. 28. Desai A, Bergmeier W, Canault M, et al. Phenotype analysis and clinical management in a large family with a novel truncating mutation in RASGRP2, the CalDAGGEFI encoding gene. Res Pract Thromb Haemost. 2017;1(1):128-133. 29. Rognoni E, Ruppert R, Fassler R. The kindlin family: functions, signaling properties and implications in disease. J Cell Sci. 2016;129 (1):17-27. 30. Jurk K, Schulz AS, Kehrel BE, et al. Novelintegrin dependent platelet malfunction in siblings with leukocyte adhesion deficiencyIII (LAD-III) caused by a point mutation in FERMT3. Thromb Haemost 2010;103(5): 1053-1054. 31. Nagy M, Mastenbroek TG, Mattheij NJA, et al. Variable impairment of platelet functions in patients with severe genetically linked immune deficiencies. Haematologica. 2018;103(3):540-549. 32. Moser M, Nieswandt B, Ussar S, Pozgajova M, Fassler R. Kindlin-3 is an essential cofactor for integrin activation and platelet aggregation. Nat Med. 2008;14(3):325-330. 33. Kuijpers T.W, van de Vijver E, Weterman MA et al. LAD-1/variant syndrome is caused by mutations in FERMT3. Blood. 2009;113(19):4740-4746. 34. Malinin NL, Plow EF, Byzova TV. Kindlins in FERM adhesion. Blood. 2010;115(20):40114017. 35. Weiss HJ. Scott syndrome: a disorder of platelet coagulant activity. Semin Hematol. 1994;31(4):312-319. 36. Suzuki J, Umeda M, Sims PJ, Nagata S. Calcium-dependent phospholipid scrambling by TMEM16F. Nature. 2010;458(7325): 834-838. 37. Toti F, Satta N, Fressinaud E, Meyer D, Freyssinet JM. Scott syndrome characterized by impaired transmembrane migration of procoagulant phosphatidylserine and hemorrhagic complications, is an inherited disorder. Blood. 1996;87(4):1409-1415. 38. Lhermusier T, Chap H, Payrastre B. Platelet

haematologica | 2021; 106(2)

membrane phospholipid asymmetry: from the characterization of a scramblase activity to the identification of an essential protein mutated in Scott syndrome. J Thromb Haemost. 2011;9(10):1883-1891. 39. Millington-Burgess SL, Harper MT. Gene of the issue: ANO6 and Scott syndrome. Platelets. 2020;31(7):964-967 40. Reddy EC, Rand ML. Procoagulant phosphatidylserine-exposing platelets in vitro and in vivo. Front Cardiovasc Med. 2020;7:15. 41. Boisseau P, Bene MC, Besnard T, et al. A new mutation of ANO6 in two familial cases of Scott syndrome. Br J Haematol. 2018;180(5):750-752. 42. Castoldi E, Collins PW, Williamson PL, Bevers EM. Compound heterozygosity for 2 novel TMEM16F mutations in a patient with Scott syndrome. Blood. 2011;117(16): 4399-4400. 43. Berrou E, Soukaseum C, Favier R, et al. A mutation of the human EPHB2 gene leads to a major platelet functional defect. Blood. 2018;132(19):2067-2077. 44. Cattaneo M. The platelet P2Y12 receptor for adenosine diphosphate: congenital and drug-induced defects. Blood. 2011;117(7): 2102-2012. 45. Cattaneo M, Lecchi A, Randi AM, McGregor JL, Mannucci PM. Identification of a new congenital detect of platelet function characterized by severe impairment of platelet responses to adenosine diphosphate. Blood. 1992;80(11):2787-2796. 46. Nurden P, Savi P, Heilmann E, et al. An inherited bleeding disorder linked to a defective interaction between ADP and its receptor on platelets. Its influence on glycoprotein IIb– IIIa complex function. J Clin Invest. 1995;95(4):1612-1622. 47. Mundell SJ, Rabbolini D, Gabrielli S, et al. Receptor homodimerization plays a critical role in a novel dominant negative P2RY12 variant identified in a family with severe bleeding. J Thromb Haemost. 2018;16(1):4453. 48. Lecchi A, Femia EA, Paoletta S, et al. Inherited dysfunctional platelet P2Y12 receptor mutations associated with bleeding disorders. Hamostaseologie. 2016;36(4):279283. 49. Cattaneo M, Zighetti ML, Lombardi R, et al. Molecular bases of defective signal transduction in the platelet P2Y12 receptor of a patient with congenital bleeding. Proc Natl Acad Sci U S A. 2003;100(4):1978-1983. 50. Zhang K, Zhang J, Gao ZG, et al. Structure of the human P2Y12 receptor in complex with an antithrombotic drug. Nature. 2014;509(7498):115-118. 51. Nisar S, Daly ME, Federici AB, et al. An intact PDZ motif is essential for correct P2Y12 purinoceptor traffic in human platelets. Blood. 2011;118(20):5641-5651. 52. Patel YM, Lordkipanidze M, Lowe GC, et al. A novel mutation in the P2Y12 receptor and a function-reducing polymorphism in protease-activated receptor 1 in a patient with chronic bleeding. J Thromb Haemost. 2014;12(5):716-725. 53. Daly ME, Dawood BB, Lester WA, et al. Identification and characterization of a novel P2Y12 variant in a patient diagnosed with type I von Willebrand disease in the European MCMDM-1VWD study Blood. 2009;113(17):4110-4113. 54. Mammadova-Bach E, Ollivier V, Loyau S, et al. Platelet glycoprotein VI binds to polymerized fibrin and promotes thrombin generation. Blood. 2015;126():683-691.

55. Reyes J, Watson SP, Nieswandt B. Functional significance of the platelet immune receptors CLEC-2 and GPVI. J Clin Invest. 2019;129(1):12-23. 56. Moroi M, Jung SM, Okuma M, Shinmyozu K. A patient with platelets deficient in glycoprotein VI that lack both collagen-induced aggregation and adhesion. J Clin Invest. 1989;84(5):1440-1445. 57. Nurden AT. Clinical significance of altered platelet collagen-receptor functioning in platelets with emphasis on glycoprotein VI. Blood Rev. 2019;38:100592. 58. Jandrot-Perrus M, Hermans C, Mezzano D. Platelet glycoprotein VI genetic quantitative and qualitative defects. Platelets. 2019;30(6):708-713. 59. Matus V, Valenzuela G, Saez CG, et al. An adenine insertion in exon 6 of human GP6 generates a truncated protein associated with a bleeding disorder in four Chilean families. J Thromb Haemost. 2013;11(9): 1751-1759. 60. Nisar SP, Jones ML, Cunningham MR, Mumford AD, Mundell SJ, on behalf of the UK GAPP study group. Rare platelet GPCR variants: what can we learn? Br J Pharmacol. 2015;172(13):3242-3253. 61. Habib, A, FitzGerald, GA, Maclouf, J. Phosphorylation of the thromboxane receptor alpha, the predominant isoform expressed in human platelets. J Biol Chem. 1999;274(5):2645–265. 62. Hirata T, Kakizuka A, Ushikubi F, Fuse I, Okuma M, Narumiya S. Arg60 to Leu mutation of the human thromboxane A2 receptor in a dominantly inherited bleeding disorder J Clin Invest. 1994;94(4):1662-1667. 63. Fuse I, Hattori A, Mito M, et al. Pathogenetic analysis of five cases with a platelet disorder characterized by the absence of thromboxane A2 (TXA2)-induced platelet aggregation in spite of normal TXA2 binding activity. Thromb Haemost. 1996;76(6):1080-1085. 64. Mundell SJ, Mumford A. TBXA2R gene variants associated with bleeding. Platelets. 2018;29(7):739-742. 65. Lebozec K, Jandrot-Perrus M, Avenard G, Favre-Bulle O, Billiald P. Design, development and characterization of ACT017, a humanized Fab that blocks platelet’s glycoprotein VI function without causing bleeding risks. MAbs. 2017;9(6):945-958. 66. Mammadova-Bach E, Gil-Pulido J, Sarukhanyan E, et al. Platelet glycoprotein VI promotes metastasis through interaction with cancer cell-derived galectin-3. Blood. 2020;135(14):1146-1160. 67. Huizing M, Helip-Wooley A, Westbrook W, Gunay-Aygun M, Gahl WA. Disorders of lysosome-related organelle biogenesis: clinical and molecular genetics. Ann Rev Genomics Hum Genet. 2008;9:359-386. 68. Bowman SL, Bi-Karchin J, Le L, Marks MS. The road to lysosome-related organelles: insights from Hermansky-Pudlak syndrome and other rare diseases. Traffic. 2019;20(6): 404-435. 69. Ambrosio AL, Di Pietro SM. Storage pool diseases illuminate platelet dense granule biogenesis. Platelets. 2017;28(2):138-146. 70. Huizing M, Malicdan MCV, Wang JA, et al. Hermansky-Pudlak syndrome: mutation update. Hum Mutat. 2020;42(3):543-580. 71. Karampini E, Bierings R, Voorberg J. Orchestration of primary hemostasis by platelet and endothelial lysosome-related organelles. Arterioscler Thromb Vasc Biol. 2020;40(6):1441-1453. 72. Enders A, Zieger B, Schwartz K, et al Lethal hemophagocytic lymphohistiocytosis in

349

P. Nurden et al. Hermansky-Pudlak syndrome type II. Blood. 2006;108(1):81-87. 73. Sharda A, Kim SH, Jasuja R, et al. Defective PDI release from platelets and endothelial cells impairs thrombus formation in Hermansky-Pudlak syndrome. Blood. 2015;125(10):1633-1642. 74. Kaplan J, De Domenico I, McVey Ward D. Chediak-Higashi syndrome. Curr Opin Hematol. 2008;15(1):22-29. 75. Dawood BB, Lowe GC, Lordkipanidze M, et al. Evaluation of participants with suspected heritable platelet function disorders including recommendation and validation of a streamlined agonist panel. Blood. 2012;120(5):5041-5049. 76. Rabbolini D, Connor D, Morel-Kopp MC, et al. An integrated approach to inherited platelet disorders: results from a research collaborative, the Sydney Platelet Group. Pathology. 2020;52(2):243-255. 77. Leo VC, Morgan NV, Bem D, et al. Use of next-generation sequencing and candidate gene analysis to identify underlying defects in patients with inherited platelet function disorders. J Thromb Haemost. 2015;13(4): 643-650. 78. Gorski MM, Lecchi A, Femia EA, et al. Complications of whole-exome sequencing for causal gene discovery in primary platelet secretion defects. Haematologica. 2019;104 (10):2084-2090. 79. Sandrock K, Nakamura L, Vraetz T, Beutel K, Ehl S, Zieger B. Platelet secretion defect in patients with familial hemophagocytic lymphohistiocytosis type 5 (FLH5). Blood. 2010;116(26):6148-6150. 80. Ye S, Karim ZA, Al Hawas R, Pessin JE, Filipovich AH, Whiteheart SW. Syntaxin-11 but not syntaxin-2 or syntaxin-4, is required for platelet secretion. Blood. 2012;120(12): 2484-2492. 81. Nakamura L, Bertling A, Brodde MF, et al.

350

First characterization of platelet secretion defect in patients with familial hemophagocytic lymphohistiocytosis type 3 (FHL-3). Blood. 2015;125(2):412-414. 82. Fager Ferrari MF, Leinoe E, Rossing M, et al. Germline heterozygous variants in genes associated with familial hemophagocytic lymphohistiocytosis as a cause of increased bleeding. Platelets. 2017;29(1):56-64. 83. Rao AK. Inherited defects in platelet signaling mechanisms. J Thromb Haemost. 2003;1(4):671-681. 84. Adler DH, Cogan JD, Phillips JA III, et al. Inherited human cPLA2 deficiency is associated with impaired eicosanoid biosynthesis, small intestinal ulceration, and platelet dysfunction. J Clin Invest. 2008;118(6):21212131. 85. Faioni EM, Razzari C, Zuleta A, et al. Bleeding diathesis and gastro-duodenal ulcers in inherited cytosolic phospholipaseA2 alpha deficiency. Thromb Haemost. 2014;112(6):1182-1189. 86. Reed K, Tucker DE, Aloulou A, et al. Functional characterization of mutations in inherited human cPLA2 deficiency. Biochemistry. 2011;50(10):1731-1738. 87. Rolf N, Knoefler R, Bugert P, et al. Clinical and laboratory phenotypes associated with the aspirin-like defect: a study in 17 unrelated families. Br J Haematol. 2009;144(3):416424. 88. Bastida JM, Lozano ML, Benito R, et al. Introducing high-throughput sequencing into mainstream genetic diagnosis practice in inherited platelet disorders. Haematologica. 2018;103(1):148-162. 89. Chan MV, Hayman MA, Sivapalaratnam S, et al. Identification of a homozygous recessive variant in PTGS1 resulting in a congenital aspirin-like defect in platelet function. Haematologica. 2020 April 16. [Epub ahead of print]

90. GeneviĂ¨ve D, Proulle V, Isidor B, et al. Thromboxane synthase mutations in an increased bone density disorder (Ghosal syndrome). Nat Genet. 2008;40(3):284-286. 91. Seligsohn U. Treatment of inherited platelet disorders. Haemophilia. 2012;18 (Suppl 4:)161-165. 92. Civaschi E, Klersy C, Melazzini F, et al. Analysis of 65 pregnacies in 24 women with five different forms of platelet function disorders. Br J Haematol. 2015;170(4):559-563. 93. Lambert MP. Inherited platelet disorders: A modern approach to evaluation and treatment. Hematol Oncol Clin North Am. 2019;33(3):471-487. 94. Freson K, Turro E. High-throughput sequencing approaches for diagnosing hereditary bleeding and platelet disorders. J Thromb Haemost. 2017;15(7):1262-1272. 95. Thaventhiran JED, Lango Allen H, Burren OS et al. Whole-genome sequencing of a sporadic primary immunodeficiency cohort. Nature. 2020;583(7814);90-95. 96. Westbury SK, Turro E, Greene D, et al. Human Phenotype Ontology annotation and cluster analysis to unravel genetic defects in 707 cases with unexplained bleeding and platelet disorders. Genome Med. 2015;7(1):36. 97. Greinacher A, Pecci A, Kunishima S, et al. Diagnosis of inherited platelet disorders on a blood smear: a tool to facilitate worldwide diagnosis of platelet disorders. J Thromb Haemost. 2017;15(7):1511-1521. 98. Greinacher A, Eekels JJM. Simplifying the diagnosis of inherited platelet disorders? The new tools do not make it easier. Blood. 2019;133(23):2478-2483. 99. Rodeghiero F, Pabinger I, Ragni M, et al. Fundamentals for a systematic approach to mild and inherited bleeding disorders: An EHA concensus report. Hemasphere. 2019;3 (5):e286.

haematologica | 2021; 106(2)

REVIEW ARTICLE

Biochemical, molecular and clinical aspects of coagulation factor VII and its role in hemostasis and thrombosis

Ferrata Storti Foundation

Francesco Bernardi1 and Guglielmo Mariani2 1 2

Department of Life Science and Biotechnology, University of Ferrara, Ferrara, Italy and Department of Science and Technology, University of Westminster, London, UK

ABSTRACT

Haematologica 2021 Volume 106(2):351-362

ctivated factor VII (FVIIa), the first protease of clotting, expresses its physiological procoagulant potential only after complexing with tissue factor (TF) exposed to blood. Deep knowledge of the FVIIa-TF complex and F7 gene helps to understand the Janus-faced clinical findings associated to low or elevated FVII activity (FVIIc). Congenital FVII deficiency, the most frequent among the recessively inherited bleeding disorders, is caused by heterogeneous mutations in the F7 gene. Complete FVII deficiency causes perinatal lethality. A wide range of bleeding symptoms, from life-threatening intracranial hemorrhage to mild mucosal bleeding, is observed in patients with apparently modest differences in FVIIc levels. Though clinically relevant FVIIc threshold levels are still uncertain, effective management, including prophylaxis, has been devised, substantially improving the quality of life of patients. The exposure of TF in diseased arteries fostered investigation on the role of FVII in cardiovascular disease. FVIIc levels were found to be predictors of cardiovascular death and to be markedly associated to F7 gene variation. These genotype-phenotype relationships are among the most extensively investigated in humans. Genome-wide analyses extended association to numerous loci that, together with F7, explain >50% of FVII level plasma variance. However, the ability of F7 variation to predict thrombosis was not consistently evidenced in the numerous population studies. Main aims of this review are to highlight i) the biological and clinical information that distinguishes FVII deficiency from the other clotting disorders and ii) the impact exerted by genetically predicted FVII level variation on bleeding as well as on the thrombotic states.

Introduction Blood coagulation is initiated by the formation of a complex between tissue factor (TF), a single-pass transmembrane glycoprotein, and activated factor VII (FVIIa), a serine protease highly dependent for its procoagulant activity on TF.1-4 The absence of either of these components is incompatible with life.5 However, small amounts of these proteins, interacting in the FVIIa-TF complex, are sufficient to initiate a number of reactions6 on membrane surfaces.4 The FVIIa-TF complex might also possess non-hemostatic, signaling properties.7 Detailed knowledge of the physiological and biochemical properties of FVIIa has enabled its pharmacological application as recombinant FVIIa (rFVIIa), a landmark in the management of bleeding disorders.8 Genetic9 and clinical studies have defined the heterogeneous molecular basis10-12 of FVII deficiency13 and could lay the foundations for gene therapy. Extensive plasma studies and genetic investigations14 have defined the importance of the F7 gene variation in determining the large FVII variance in plasma concentration,15 with implications in predisposition to thrombosis16-18 in both individuals and the population as a whole. A comprehensive review on FVII is complex because of the wealth of information available both in the field of hemostasis and thrombosis (Figure 1). The main aim of this review is to provide an integrated and balanced perspective of the biohaematologica | 2021; 106(2)

Correspondence: FRANCESCO BERNARDI ber@unife.it Received: July 3, 2020. Accepted: October 29, 2020. Pre-published: January 7, 2021. https://doi.org/10.3324/haematol.2020.248542

ÂŠ2021 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

351

F. Bernardi and G. Mariani et al.

logical and clinical information currently available. That is, the relationship between the reduced levels of FVII/bleeding tendency and high levels of FVII/ risk of cardiovascular disease. We will particularly focus on distinctive features of the FVII protein and F7 gene (Table 1) as well as on open issues (Table 2).

Genetics and biochemical aspects Expression of the F7 gene and FVII protein is shown in Figure 2. Key biochemical and genetic findings are summarized on a historical time line in Figure 1A, and some of the key unanswered questions are summarized in Table 2. The F7 gene9 (12.8 KB) is located on chromosome 13q34, 2 KB apart from the homologous F10 gene, which suggests evolution by duplication. In addition to F10, the F7 gene structure and coding sequence19 displays a noticeable homology also with F9 (factor IX, FIX) and PROC (protein C). Mutations that disrupt promoter activity in severe FVII deficiency20,21 highlight the importance of the transcription factors SP1 and HNF4Îą, but do not sustain the androgen-

dependent rescue observed for mutations in the F9 promoter in hemophilia B. The evolutionary history of the F7 promoter region, which contains frequent polymorphisms22,23 modulating FVII expression (see dedicated paragraph), may differ in human populations. The F7 gene gives rise to three mRNA transcripts12 through FVII mRNA alternative splicing. Whereas two transcripts (NM_019616.4 and NM_000131.4) encode an identical mature and circulating FVII protein, the third (NM_001267554.1) lacks the amino terminal domains and is of unknown physiological significance.

The FVII protein and FVIIa-TF complex A scheme of FVII activation, activity and inhibition is shown in Figure 3. Whereas the mature circulating FVII protein (50 KDa) sequence19 is composed of 406 residues, the mRNA (NM_019616.4) encodes 60 additional aminoacids, the pre-propeptide sequences which drive FVII biosynthesis/secretion and are removed by intracellular proteolysis (Figure 2). The first numbering (1-406 residues) is currently used in protein studies, and the second one (1-466 codons) in molecular genetics.

Figure 1. Publications over 70 years and some key achievements related to coagulation factor VII. (A) Coagulation factor VII (FVII) biochemistry and F7 genetics. (B) FVII deficiency, FVII level associated cardiovascular disease, FVII levels and F7 gene: N: number of publications reported in Pubmed. Only some of the discoveries are referenced in the text due to space constraints. A complete list is available on request. GWAS: Genome Wide Analysis Study; NGS: next-generation sequencing.

352

haematologica | 2021; 106(2)

Coagulation factor VII, hemostasis and thrombosis

FVII zymogen circulates in plasma at low concentration (500 ng/mL, 5nM) and the extracellular proteolytic cleavage1 of a minute portion (about 1%, 0.1 nM) of FVII giving rise to FVIIa, occurs between residues Arg152-Ile153, (Figure 2)2 producing the light and heavy chains. The origin of FVIIa in plasma is still debated. Substantially decreased levels of plasma FVIIa in individuals with congenital FIX deficiency suggest that the generation of FVIIa is dependent on an activation product of FIX. Recently, it has been proposed that different forms of activated FIX (FIXα and FIXβ) participate in a reciprocal activation mechanisms of FVII(a) and FIX(a) (white and red-curved arrows, Figure 3, left panel) that would not require a protein cofactor.24 The FVIIa heavy chain contains the domain characterized by the (chymotrypsin) serine protease family catalytic triad. The light chain contains the calcium binding, vitamin K-dependent domain (gamma-carboxyl-glutamic, GLA) and two epidermal growth factor-like domains, essential for the interaction with membranes and other proteins2 of the coagulation cascade. Considerable variation in FVIIa plasma concentrations between individuals has been reported (Figure 4),25 exceeding one order of magnitude. Differently from other serine protease of the coagulation cascade, FVIIa displays a plasma half‐life (2-3 hours) remarkably close to that of the FVII zymogen, which might be explained by the “zymogenlike” properties of FVIIa that have been weakened by mutagenesis, thus increasing its activity.26 FVIIa interacts with TF, a membrane receptor exposed following vascular lesion (extrinsic pathway). The FVIIaTF complex,1 essentially conserved in all jawed vertebrates,27 activates both FIX and factor X (FX)28 (Figure

3) on the platelet surface in the presence of Ca2+, leading to the generation of thrombin and the subsequent deposition of fibrin. The crystal structure of the complex between the activesite-inhibited FVIIa and the cleaved, soluble extracellular domain of TF (sTF) revealed the details of the contoured embrace of FVIIa and sTF domains and the extensive intermolecular contacts.29 Neutron and X-ray scattering experiments30 suggested that FVIIa in solution has an elongated domain structure with significant flexibility, which allows rapid interaction with sTF over a large surface area to form the high-affinity complex (dissociation constant, KD 2−5 nM). The FVIIa-TF complex is highly dependent on specific lipids for physiological activity.2 Activated phospholipid membranes host both TF, an integral membrane protein, and FVIIa, which binds membranes through its GLA domain. Externalization of phosphatidylserine to the outer membrane leaflet and allosteric TF disulphide bond exchanges (decryption on membranes)4 make TF the FVIIaactivating cofactor. The FVIIa-TF complex formation shapes the FVIIa catalytic site by allosteric interactions and increases its activity up to 106–fold. As a matter of fact, circulating FVIIa is the active portion of the total FVII mass only after high-affinity binding with TF. These FVIIa-TF complex-specific molecular events that provide FVIIa with its physiological activity are believed to represent the true initiation of the extrinsic pathway (Table 1). The physiological negative control of the FVIIa-TF complex (Figure 3) occurs through the reversible inhibition by tissue factor pathway inhibitor (TFPI),31 mediated by the TFPI Kunitz domain 1 and with protein S as cofactor.32 An

Figure 2. Schematic diagram of the F7 gene and factor VII protein expression. Upper part: Exons are numbered (1-9) and colored in accordance with the encoded protein domains (lower part). Exon 2 is in parenthesis because it is not included in the most abundant mature mRNA. FVII: factor VII; F7: factor VII gene; KB: kilobase. In addition to the complete nuclear transcript the most represented FVII mRNA is indicated. Lower part: Protein domains are indicated with different colors. GLA, triangle, γ-carboxyglutamic acid-containing domain. EGF, trapezium, epidermal growth factor-like domain. Green box, promoter. The complete intracellular protein, and the circulating forms are depicted. The numbers of residues in the pre-pro-leader sequence (n=60), in the circulating forms (n=406) and in the light (n=152) and heavy (n=254) chains are indicated. The interchain disulphide bridge is also depicted (yellow bracket).

haematologica | 2021; 106(2)

353

F. Bernardi and G. Mariani et al.

apparently redundant and irreversible inhibition of FVIIa is also exerted by the serpin antithrombin (AT),33 and a substantial portion of FVIIa may be cleared through this relatively stable complex (FVIIa-AT),34 which circulates in plasma at a concentration similar to that of FVIIa. FVII and FVIIa also bind the protein C receptor on endothelial cells (EPCR) with relevant affinity, but at present the pathophysiological significance of this interaction is unclear.

Congenital factor VII deficiency Definition, prevalence and epidemiology FVII deficiency was first described as a bleeding disorder

by Alexander and colleagues13 and represents a model disease to understand the pathophysiology of recessively inherited coagulation deficiencies. Relevant findings on FVII deficiency are summarized on a historical time scale in Figure 1B, and some of the open issues are summarized in Table 2. Congenital FVII deficiency (OMIM 227500, ORPHA 327) is defined as a bleeding disorder associated with FVII coagulant activity below 50% of normal. However, this threshold includes asymptomatic heterozygotes for causative mutations and may comprise individuals homozygous for frequent FVII lowering single nucleotide polymorphisms (SNP).35 These subgroups substantially contribute to the belief that FVII deficiency is a mild bleeding disorder.36

Figure 3. Schema showing factor VII activation, activity and inhibition (see text for specific information and references). TF: tissue factor; FIXaα: FIXa cleaved only after Arg226; FIXaβ: FIX cleaved after Arg191 and Arg226 (ref. 24); AT: antithrombin; PS: protein S.

Figure 4. F7 genotype driven differences in FVIIa, FVIIc and FVIIAg in the European population. Mean values of FVIIa (mU/mL), FVIIc (% of PNP), and FVIIAg (% of PNP) in genotype groups determined by the promoter (5′F7 ins del) and missense 353(413)Arg/Gln polymorphisms. Standard deviation is reported above each column. The number of subjects is reported in parentheses. Significant differences, groups 1–6 (P<0.001): FVIIc: 1 vs. 2, 4, 6; 4 vs. 6; FVIIa: 1 vs. 4, 6; 2 vs. 6; 4 vs. 6; and FVIIag 1 vs. 4, 6. Adapted from “Contribution of factor VII genotype to activated FVII levels. Differences in genotype frequencies between Northern and Southern European 107 FVII: factor VII; Populations”. FVIIc: FVII activity; FVIIAg: FVII antigen; PNP: pooled normal plasma; ins: insertion; del: deletion

354

haematologica | 2021; 106(2)

Coagulation factor VII, hemostasis and thrombosis

FVII deficiency is believed to be very rare (one case in 500,000 individuals).36,37 The World Federation of Hemophilia38 estimates around 5,000 FVII-deficient individuals in the countries covered by this organization. However, the Italian Registry of Congenital Rare Bleeding Disorders39 (RBD) reported 1,023 cases with FVII deficiency among the 2,178 patients registered. This figure, similar to that reported by the UK registry (n=1.373),38 suggests a prevalence only slightly lower than that of hemophilia B, at least in Italy (FVII deficiency, 1:59,000; haemophilia B, 1:67,000 individuals). Accordingly, FVII deficiency accounts for the most frequent RBD in most registries (3649%).37,39 These data support the concept (Table 1) that FVII deficiency, including mild, moderate and severe forms, might be much more prevalent than previously reported. Large differences in prevalence, with variation reaching one order of magnitude, have been reported in those geographical areas where consanguineous marriages are frequent.

Clinical picture The bleeding tendency in FVII deficiency varies from forms more severe than those seen in the hemophilias40 to very mild forms or asymptomatic cases. A simple classification of the clinical severity was proposed:10 i) severe forms (central nervous system [CNS] and/or gastrointestinal [GI] bleeding and/or hemarthrosis; ii) moderate forms

(three or more symptoms other than CNS and/or GI bleeding and/or hemarthrosis); iii) mild forms (one or two symptoms, mostly muco-cutaneous, and other than CNS and/or GI and/or hemarthrosis). In the International F7 Registry and the Seven Treatment Evaluation Registry (STER)41 that include 787 cases, 12% of individuals had a severe to very severe phenotype and three fourth of them bled within the first year of age (Table 4). About 20% of individuals with FVII deficiency suffer from a bleeding disorder that can be lifethreatening and/or be a major handicap. On the other hand, the most frequently observed patient cohort (>50%) is characterized by muco-cutaneous types of bleeding: epistaxis (two thirds), gum bleeding (one third), easy bruising (one third), and menorrhagia (70% of the females). In general, the bleeding tendency remains the same during the lifetime of the patient with the first bleeding event being a strong and independent predictor of the risk for the subsequent bleeding phenotype.41 Coinheritance of factor V Leiden enhances thrombin formation and is associated with a mild bleeding phenotype.42 Levels of FVIIc higher than 26%11,41,43 may have a nil or minimal clinical relevance in terms of spontaneous bleeding and subjects with FVIIc levels <10% who remain asymptomatic most of their lives are not rare in being diagnosed during family studies or hemostatic screenings.10,11,35

Table 1. Distinctive features of coagulation factor VII (FVII), activated FVII , FVII deficiency, FVII levels and cardiovascular disease.

FVII

FVII (FVIIa)

FVII deficiency

FVII levels

Very low plasma concentrations (0.5 mg/mL)

Highest prevalence among autosomally recessive rare bleeding disorders

High variability in human populations High number of environmental determinants

Short half-life 2-4 hours Stable in plasma

First protease in the coagulation cascade. Lowest catalytic activity without cofactor (TF). Half-life of FVIIa and FVII similar (2-3 hours) rFVIIa for replacement therapy

Conserved in all jawed vertebrates

F7 polymorphisms are strong predictors of FVIIa levels

High similarity with factor IX, factor X and Protein C

Very low amounts of FVIIa sufficient to initiate coagulation ÂŤStableÂť FVIIa-AT complex

Complete deficiency results in perinatal mortality High incidence of intracranial bleeding (severe form)

High proportion of variance explained by F7 gene variation The genotype/phenotype relationship is one of the most extensively investigated among human genes Wide variation of bleeding phenotype Frequent FVII level- lowering alleles in relation to modest FVII level variations with wide differences in human populations Homozygous FVII-lowering alleles High/low levels associated with mimic Mendelian heterozygous predisposition/ protection deficiency towards/from cardiovascular disease

FVII: factor VII; FVIIa: activated FVII; rFVIIa: recombinant FVIIa; TF: tissue factor; AT: serpin antithrombin; F7: factor VII gene.

Table 2. Some open issues in factor VII (FVII) protein and F7 genetics, FVII deficiency, FVII levels and cardiovascular disease.

FVII / F7

FVII deficiency

F7 gene variation cardiovascular disease

Mechanisms underlying FVIIa levels

Residual FVII concentration predicting the most severe symptoms is still puzzling. Residual FVII concentration preventing spontaneous bleeding is difficult to define. Short plasma half life of FVII/rFVIIa for therapeutic purposes. Discrepancy between rFVIIa pharmacokinetic and pharmacodynamic data Impact of FVII-lowering alleles in increasing the clinical penetrance of pathological F7 gene lesions

Population features causing large differences in the prediction of cardiovascular disease risk by F7 genotypes

Decryption of TF for FVIIa binding and catalytic activity

Physiological balance between FVIIa inhibitors

Impact of FVII-lowering alleles in the protection from cardiovascular disease

Impact of FVII-increasing alleles in the predisposition toward specific cardiovascular disease forms

FVII: factor VII; FVIIa: activated FVII; rFVIIa: recombinant FVIIa; TF: tissue factor; F7: factor VII gene.

haematologica | 2021; 106(2)

355

F. Bernardi and G. Mariani et al.

However, the therapeutic thresholds conferring protection from traumatic or post-operative bleeds are still a matter of debate.44,45 Surgical bleeding has been reported in 15-24% of cases42,45 with FVIIc <7%.42 Bleeds in the first post-operative day46 are particularly frequent if the bleeding disorder was not diagnosed before, and are mostly reported during orthopedic procedures,46 a major hemostatic challenge. Presenting symptoms as post-circumcision bleeds and hemorrhage from the umbilical stump frequently herald a severe disease form.

Diagnosis The spectrum of bleeding symptoms is wide, involves both males and females, can start after birth and may occur at any age. The most important predictors of bleeding risk, familial and personal histories, can be silent as observed in diseases with recessive inheritance. The relevance of the gynecological and obstetric histories are discussed in a dedicated paragraph. The laboratory workup47 performed after a bleeding episode or during a family screening includes routine screening assays (prothrombin time [PT], activated partial thromboplastin time [aPTT]), followed by FVIIc measurement, which is necessary to confirm the diagnosis. Evaluation of FVIIc is determined with a one-stage assay adding diluted plasma samples to FVII-deficient plasma and thromboplastin as the source of TF. The accuracy of FVIIc assays is related to the sensitivity of the thromboplastin reagent and to the quality of the FVII-deficient plasma and of the calibrators. The most suitable activators are the human placenta-derived and the recombinant preparations,47 that are also recommended to monitor the treatment with rFVIIa.25 New chromogenic or fluorogenic assays for FVIIa are available for research purposes and to monitor treatment with rFVIIa.48 The sensitivity of the FVIIc assay to levels below 2% may be poor, with negative implications for the accurate definition of the relationship between FVII levels and bleeding phenotypes (Table 1), as well as between genotype and phenotype. A low assay sensitivity may explain some inconsistencies such as the paradoxical observation of asymptomatic cases with low FVII levels. Furthermore, severe bleeding is sporadically observed in individuals with moderately reduced levels of FVII. The combination of the low assay sensitivity with the minimal amount of FVIIa-TF complex needed to trigger coagulation does not permit one to efficiently differentiate clinical phenotypes. As a consequence, in FVII deficiency a variety of bleeding phenotypes are observed in the presence of apparently modest differences in FVIIc levels (Table 1). In summary, the clotting tests do not always predict the degree of bleeding tendency in the single individual. The presence of the FVII protein (FVII antigen [FVIIag]) can be determined by enzyme-linked immunosorbent assays or immune-turbidimetric assays, using monoclonal epitope-specific antibodies.47 The FVIIag assay47 does not predict the bleeding tendency, but does allow one to distinguish between type I (quantitative defects, with decrease of both FVIIc and FVIIAg), and type II defects (qualitative defects, with low FVIIc and normal or reduced levels of FVIIAg), and may help to clarify the different mutational mechanisms.

Molecular genetics The key findings are summarized on a historical time 356

scale in Figure 1B, and the main findings in individuals with FVII deficiency are shown in Table 3. The functional characterization of protein variants in plasma was carried out in seminal studies49 well before the genetic characterization of FVII deficiencies. The first F7 gene mutations in FVII deficiency were reported in the UK50 and Italian populations, and more than 1,000 genetic diagnoses in several countries are reported in the International Registry on FVII deficiency (IRF7)10 and the Greifswald Registry.11 Mutations and residual FVII levels have been presented in databases.51 The recent EHAD database12 (http://www.umd.be/F7/W_F7/index.html) includes 221 unique variants identified in 728 individuals, of great help for the evaluation of F7 mutations found during genetic counseling and diagnosis, including prenatal diagnosis.5 F7 mutations have been detected by sequencing coding, splicing and in promoter regions in the vast majority of patients. High throughput genomic sequencing will help in the case no mutations are found.52 Cases with severe symptoms are virtually all homozygous or doubly heterozygous for F7 mutations. Symptomatic mutations impairing vitamin K metabolism, and thus posttranslational modification of GLA residues, have been found to cause combined vitamin K-dependent clotting factor deficiency that includes an abnormal FVII biosynthesis and low FVIIc levels. As reported in registries11 and databases,12 single nucleotide variants are the most common type of gene defect (around 90%), and include a large proportion (between two thirds and three quarters) of missense mutations. Numerous missense mutations have been found in the homozygous condition, related to geographical, ethnic and mutational pattern features. Clusters of homozygotes often indicate genetic isolates or ethnic customs favoring consanguinity (see below). Since they have been well presented in databases,12 we shall briefly comment on a few of them. The severe Gln100(160)Arg53,54 and the moderate/mild Ala294(354)Val55,56 are prevalent in Northern Europe. The frameshift mutation in the last carboxyl terminal codons (Pro404(464)Hfs) is very prevalent in cis with the Ala294(354)Val missense change and causes a 28 residueselongated FVII molecule. This moderate to severe and combined variant is frequent in Central Europe and especially in Slovakia.10 The mild Gly331(391)Ser mutation is prevalent in the gulf of Naples, Italy, as well as in other countries.57 Other mild/asymptomatic changes, the Cys310(370)Phe/Ser and Arg304(364)Gln50, Thr359(419)Met substitutions, have been found in several populations, suggesting recurrent nucleotide changes at a CpG site. The frequent Arg304(364)Gln is also characterized by discrepant FVII activity findings depending on the thromboplastin reagent.49 Splicing mutations are not rare,52 and the homozygous splicing mutation IVS7+5 g>a, relatively frequent in Italy, has been used for cellular58,59 and animal60 models of gene therapy. F7 deletions are also not rare and together with duplications, insertions and indel rearrangements represent about 10% of all genetic lesions.12 However, large and extended deletions have not been described in homozygous individuals. Differently from hemophilia A and B, individuals with complete FVII deficiency might die perinatally or haematologica | 2021; 106(2)

Coagulation factor VII, hemostasis and thrombosis Table 3. Overview of molecular genetics findings in factor VII deficiency.

Molecular genetics of FVII deficiency Genetic diagnosis

Mutation zygosity and disease severity

Mutation type frequency Recombinant expression

Sanger sequencing of exons and splicing junctions in >1,000 individuals with FVII deficiency Next-generation sequencing in a few individuals Phenotypic assays do not define specific genetic defects* No highly frequent gene lesions Large heterogeneity of genetic causes (n>300 different mutations) Geographical/ethnic clustering of identical-by-descent mutations (n>15) Several recurrent mutations in different populations Severe deficiencies caused by compound heterozygous/homozygous mutations Increased prevalence of homozygotes in genetic isolates/consanguineous marriages Otherwise asymptomatic heterozygous variants associated with frequent FVII-lowering single nucleotide polymorphism in mild deficiencies Missense>>Small deletions>Splicing>Nonsense>Large deletions (not reported in homozygosity) Numerous variants expressed after site directed mutagenesis A few purified protein variants Translational read through over premature termination codons Gain-of-function premature termination codon

*The asymptomatic Arg364Gln is detectable by using different thromboplastins59. Small deletions include also insertions and indels. FVII: factor VII.

Table 4. Management of factor VII deficiency using recombinant factor VII (synopsis of the STER study data).

Clinical Context

Treatments (n)

Treatment days Median, range

Total dose of rFVIIa Median, range (µg/kg of body weight)

Spontaneous bleeding

1 (1-14)

60 (10-3,600)

Major surgery^

3.5 (1-16)

31.9 (12-120)

29 31 18 (3/week) 13 (2/week)

1 (1-4)

20 (7.2-510)

3/week or 2/week

30 (3 per week) 24 (3 per week)

Minor surgery and invasive procedures Prophylaxis§

Outcome

3 rebleeding 1 inhibitor 3 bleeding (orthopaedic surgery) 1 inhibitor Partly effective in one case in each of the two arms§

The data express data analysis, not recommendations; ^ Mariani et al. 2011; §Napolitano et al. 2013 (no statistical difference between arms). FVII: factor VII; rFVII: recombinant FVII; STER: Seven Treatment Evaluation Registry.

shortly after birth due to severe hemorrhage.5 Complete FVII deficiency causes perinatal mortality in “knockout” mice, in which the presence of a trace amount of FVII (0.7%) seems sufficient for survival with symptoms.61 Several cases with heterozygous combined FVII and FX deficiency, due to large deletions within the terminal end of chromosome 13, have been reported.62 Recombinant expression after site-directed mutagenesis has been performed for a number of mutations detected in patients, which permits one to investigate the molecular bases of the deficiency. Recombinant studies of the rare and potentially null homozygous nonsense mutations support the notion that gain-of-function and translational readthrough over premature stop codons may prevent truly null conditions and may contribute to the unexpectedly variable bleeding phenotype in individuals homozygous for F7 nonsense changes.63 Whereas life-threatening symptoms have been associated with the p.Ser52(112)X, moderate bleeding was observed in p.Cys82 (132)X homozygotes. The Arg402(462)X change permits the secretion of a small amount of protein with gain-of-function features, causing a mild phenotype paradoxically associated with a “null” mutation.64 Chimeric fluorescent FVII molecules have also been constructed to investigate intracellular processing of variants, and chemical chaperones have been used to stimuhaematologica | 2021; 106(2)

late their secretion.54 The FVII protein has been purified and biochemically characterized for very few variants, among these the frequent Arg304(364)Gln50 and Ala294(354)Val55 changes, as well as the severe Gln160Arg33 and the activation sequence Val154(214)Gly65 changes. These laborious experiments provided strong support for the understanding of the mild or severe coagulation phenotypes in vivo and for FVII altered expression and properties. The homozygous conditions for polymorphisms predicting lower FVII levels might not be pathogenic but mimic Mendelian heterozygous FVII deficiency35 (Table 1), which frequently leads to the request for clotting defect diagnosis in pre-surgical screenings, and generates confusion in estimating the incidence of inherited FVII deficiency (see dedicated section). In light of their effects on lowering FVII plasma levels, these variants might contribute to pathogenicity of co-inherited variants.52 As a matter of fact the frequencies of the FVII-lowering alleles were significantly higher51 in the FVII-deficient individuals tested than in controls, suggesting that the presence of these alleles may have increased, through increased clinical significance (Table 2), the likelihood of identifying “causative” F7 gene lesions. Genotyping of these frequent variants would be especially useful for the interpretation of mild and moderate FVII deficiency. 357

F. Bernardi and G. Mariani et al.

Replacement therapy and prophylaxis Replacement therapy (RT) options are determined by: the rarity of the disorder; the availability and supply of products and the economic and geographical factors. These include: i) recombinant FVIIa (rFVIIa), ii) plasmaderived FVII (pdFVII), iii) fresh frozen plasma (FFP) and prothrombin complex concentrates (PCC). In the STER prospective trial (comprising 312 RT) most therapies were carried out with rFVIIa (78%), the remaining with FFP (10%), pdFVII concentrates (10%%) and PCC (2%). Based on the modest catalytic activity of FVIIa in the absence of TF, the therapeutic/prophylactic use of FVIIa and rFVIIa (initially proposed for bleeding diathesis other than FVII deficiency) represents a therapeutic milestone8 and distinguishes replacement therapy in FVII deficiency from that in the other coagulation defects (Table 1). Although in FVII deficiency rFVIIa is employed (Figure 1A) at doses (Table 3) much lower than those needed for patients with FVIII inhibitors, supraphysiological FVIIa concentrations are also produced in plasma of FVII deficient subjects. It is still a matter of debate whether rFVIIa acts through the binding with TF, provided by microparticles shed into the circulation following diverse stimuli, or binds at high concentration66 to anionic phospholipids exposed on activated platelets, thus directly activating FX to FXa. FXa would in turn generate thrombin, bypassing the tenase complex. It is tempting to speculate that the physiological FVIIa-TF function (Figure 3) prevails at the rFVIIa doses used in FVII deficiency and in the presence of normal FVIII levels. rFVIIa easily diffuses into the extravascular spaces where it could be retained for extended time periods. As supported by pharmacokinetic studies,67 rFVIIa prolongs its pharmacological effects at low concentration, in accordance with the hypothesis of physiological binding to TF. rFVIIa has a very good safety-to-efficacy ratio.46 Oneday therapy with ‘intermediate’ doses of rFVIIa can be effective and safe for the treatment of most of spontaneous bleeds as well as for minor surgery and invasive procedures (Table 3). Replacement with rFVIIa is also effective to prevent bleeding in major surgical procedures.46 The large volume of the infusions and the limited availability make pd-FVII concentrates less appealing. Currently, the average pd-FVII dosages used are 15–20 IU/kg for mucosal bleeding, and 30–40 IU/kg for severe or life-threatening hemorrhages. Although FFP is easily available in developing countries, its effectiveness is limited owing to the high risk of fluid overload and the consequent need for slow infusions. In the case of mild mucosal bleeds (epistaxis, mild menorrhagia) tranexamic acid and hormones are currently considered. Anti-fibrinolytic agents are contraindicated in hematuria, and may trigger thrombosis in association to PCC. Prophylaxis is warranted for patients with the most severe bleeding picture and should be prescribed from childhood or soon after the first bleeding event.45,68 Effective prophylaxis schedules have been reported for rFVIIa68,69 and pdFVII concentrates.70

Treatment complications Inhibitors to FVII are a rare (1-2%)71 complication that 358

occurs mainly in severely deficient patients, particularly in children younger than 1 year on prophylaxis. We observed only high responders (>5 BU Bethesda Unit)71, and in the presence of a high-titer inhibitor to FVII treatment becomes a problem.71 At variance with the homologous deficiency of FIX (F9, HB), complete homozygous gene deletions that predispose to FIX inhibitors have never been detected in FVII deficiency (Table 1),5,12 and FVII inhibitors have never been shown to be complicated by allergic reactions. Paradoxically, a dozen of cases with FVII deficiency and thrombosis have been reported. Surgical interventions and/or replacement therapies had a close temporal relationship with thrombotic episodes, but apparently spontaneous events were also reported.68,72 Different replacement therapies were associated with the thrombotic events: PCC (three cases), rFVIIa (three cases), pdFVII (two cases), FFP (one case), no replacement (three cases).72 This suggests that FVII deficiency does not seem to offer protection from strong thrombosis risk factors such as surgery/high dose radiotherapy.

Women with factor VII deficiency Autosomally transmitted RBD occur as frequently in women as in men, but women may experience more bleeding than men because gynecological and obstetric challenges to hemostasis add an important burden to the background bleeding related to the hemostatic defect. Menstruation and ovulation are associated with an increased risk of bleeding, as are pregnancy and delivery.73 While in normal and heterozygous women FVII plasma levels rise during pregnancy, no such increase is observed in women with a severe homozygous deficiency. As a consequence, symptomatic women with FVII deficiency may continue to bleed during pregnancy and postpartum. In a large study (234 women with FVII deficiency),74 menorrhagia during the reproductive age occurred in half of the cases, and in 12% represented the first bleed. Further, frequent gynecological problems, such as uterine fibroids, were diagnosed earlier because of the hemostatic defect. Although FVIIc was an important predictor of gynecological bleeding, other determinants including endocrine pathologies may also play a role. For severe menorrhagia, management has been moved from hemostatic agents like tranexamic acid, oral contraceptives and intra-uterine devices, to replacement therapy and prophylaxis with effective single- or multiple-dose schedules.68,73,74 This will hopefully change clinical practice that, until recently, included surgical approaches such as endometrial ablation or hysterectomy.

Therapeutics in development The first attempts to improve rFVIIa activity through mutagenesis, or its half-life in plasma by glycoPEGylation, have increased FVII antigenicity. Longer-acting FVIIa has been tested in different recombinant preparations, which are currently in clinical trials (reviewed in Menegatti et al.)75: i) addition of a c-terminal peptide76 potentially suitable for subcutaneous administration that led to prolonged pharmacodynamics effects, ii) an Fc receptor-fused rFVIIa that displayed a 5-fold longer plasma half-life and iii) an albumin-fused rFVIIa molecules that showed improved features compared to rFVIIa (2- to 3-fold longer half-life and 4- to 8-fold lower clearance) in patients with congenital FVII deficiency.77 Recently, an engineered albuhaematologica | 2021; 106(2)

Coagulation factor VII, hemostasis and thrombosis

min-fused rFVIIa molecule showed enhanced transcellular transport upon intranasal delivery and extended plasma half-life in transgenic mice.78 The development of non-substitutive therapy for the hemophilias, such as the use of anti-antithrombin RNA interference or of aptamer and monoclonal antibodies directed against TFPI, have raised expectations for improvement of the quality of life in individuals with FVII deficiency.75 Based on adeno-associated viral-mediated expression of FVII79, gene therapy has produced sustained correction of severe FVII deficiency in dogs. In addition, gene therapy using FVIIa, inserted into the same viral vector, could also be used to treat congenital FVII deficiency with lower vector doses, which increases the chance of efficient and long term expression of the transgene. Small engineered RNA have also been used as an approach of personalized gene therapy to correct a human FVII splicing mutation in mice.59

Factor VII levels, F7 genotypes and cardiovascular disease

40% of variance. Genotype effects, perhaps stronger on FVIIa than on FVIIag (Figure 4),15 might modulate the response to environmental stimuli and the sex-dependent regulation.89 Serum phospholipids were found to be strong and F7 genotype-associated FVIIa determinants.90 The number of F7 single SNP was substantially increased by highthroughput F7 gene sequencing, which enabled very informative F7 population studies.91 Genome-wide association studies (GWAS) confirmed14,92 the strong association between FVII levels and F7 gene variation.93 Importantly, GWAS have detected a number of genomic regions associated with FVII levels (GCKR, ADH4, MS4A6A, PROCR, APOA5, HNF4A, REEP3JMJD1C, JAZF1-AS1, MLXIPL and XXYLT1),14,92 that could explain an additional one fifth of the FVII variance in plasma levels and are also, in part, associated with lipids which in turn modulate FVII activity. These gene-based association scan initiatives have been established in several populations.94 Overall, this picture defines one of the most extensively investigated relationship between genotypes and multiple quantitative phenotypes (FVIIc, FVIIag and FVIIa) (Table 1).

Levels, genotypes and cardiovascular disease Several findings have created interest and still lead to an intense investigation on the role of FVII in cardiovascular disease (Figure 1B). Exposure of TF to blood in (coronary) artery disease, and especially plaque rupture, may favor FVIIa-TF complex formation. Further, lipids are important components of atherosclerosis and determinants of FVII activation and activity as well.79,80 Additional links between high FVII levels and risk for thrombosis have been suggested81 by finding that the TF-FVIIa-Xa complex activates FVIII before coagulation amplification and that heparanase increases generation of FXa by the FVIIa-TFcomplex.82 Furthermore, the FVIIa-TF complex is believed to mediate non-hemostatic functions in diverse biological processes, such as angiogenesis, inflammation, atherosclerosis and vascular and cardiac remodeling, by activating signaling pathways (reviewed in D'Alessandro et al.)84 through FVIIa-integrin binding and PAR2 cleavage.85 The F7 promoter may respond to a number of metabolic components, and FVIIc levels are associated with several environmental factors linked to atherosclerosis i.e., body mass index, dietary fat intake, plasma lipids and particularly triglyceride concentration. Age- and sex-related variations in FVII levels add complexity to the investigation of clinical correlates. FVII levels were found to increase with age and to be significantly lower in women than men at younger ages. Postmenopausal women displayed the highest levels, except when undergoing hormone replacement therapy.86 This conundrum of environmental factors interacts with the F7 gene variation.

Factor VII levels and F7/genome wide genotypes FVII levels show ample variation in the normal population and have a substantial heritable component, also in relation to polymorphisms15 (Figure 4). Missense,87 repeat number variation,58 insertion/deletion22 and SNP23 were found to be associated with FVIIc, FVIIa and FVIIag levels (Figure 4) through epigenetic, transcription, biosynthesisor stability-mediated mechanisms. Whereas twin studies88 estimated about 60% of genetic-associated level of variance in plasma, the F7 locus variation accounted for up to haematologica | 2021; 106(2)

The Northwick Park Heart Study investigators were the first to report that high FVII levels were predictors of death due to coronary disease,16 and a number of studies confirmed this observation in different populations by also evaluating FVIIa levels.95 Recent investigations on the level of the FVIIa-AT complex, which may reflect levels of FVIIa as well its interaction with TF, indicated that higher complex concentrations were associated with increased mortality in the Cardiovascular Health Study.96 In patients with stable coronary artery disease (Verona Heart Study) higher FVIIaAT complex levels were associated with increased cardiovascular mortality and increased thrombin and FXa generation,97 particularly in the coagulation initiation phase.98 However, in small groups of patients with unstable angina, acute myocardial infarction99 or post-infarction80 FVIIa levels were not higher than in controls. High plasma FVIIag levels were associated with failure of thrombolytic therapy in patients with myocardial infarction.100 Elevated levels of FVII have not been consistently associated with venous thromboembolism.101 The influence of F7 genotypes on the hemostatic balance and on the susceptibility to cardiovascular disease has been extensively investigated in several large cohorts of patients in relation to both myocardial infarction and stroke. F7 polymorphisms with an opposite effect on FVIIa levels may positively or negatively modulate the risk of MI in males with advanced coronary artery disease,18 and some FVII genotypes may protect against myocardial infarction17 by affecting transcription levels and reducing protein functional activity. The modulation of stroke risk in atrial fibrillation by F7 genotypes may follow a similar scheme102 and recently, in a meta-analysis of several GWAS14 variations in FVII-related genes and FVIIc levels were associated with a risk of the incidence of ischemic stroke in the general population. The physiological effects of FVII lowering alleles may represent a natural model for anticoagulation; in fact F7 polymorphisms were shown to play a role in determining the initial response to warfarin103 and influencing the risk of thrombosis in patients with essential thrombocythemia104. 359

F. Bernardi and G. Mariani et al.

Large haplotype studies confirmed that the F7 gene strongly influences FVII levels, but associations with coronary artery disease and FVII level were inconsistent.105 Variable genetic, environmental and atherogenic risk factors in different populations may explain this discrepancy (Table 2). The frequencies of F7 genotypes predicting lower FVII levels are higher in Caucasians and much lower in Chinese and Malays populations. On the other hand, the distribution of these F7 genotypes covaries across populations in Europe with the rate of myocardial infarction mortality, being higher in countries at low risk.106 The most recent meta-analysis of large studies14 appears to be consistent with positive and potentially clinically important causal effects of FVIIc levels, both on coronary artery disease and venous thromboembolism. However, several open issues remain (Table 2) and the usefulness of FVIIc and FVIIa evaluation in patients with cardiovascular risks is still unclear.

Conclusive remarks In summary, in the 70 years since FVII investigation started, research has been particularly intense (Figure 1), and provided excellent examples at the basic science and

References 1. Nemerson Y, Esnouf MP. Activation of a proteolytic system by a membrane lipoprotein: mechanism of action of tissue factor. Proc Natl Acad Sci U S A. 1973;70(2):310314. 2. Gajsiewicz JM, Morrissey JH. Structurefunction relationship of the interaction between tissue factor and factor VIIa. Semin Thromb Hemost. 2015;41(7):682690. 3. McVey JH. The role of the tissue factor pathway in haemostasis and beyond. Curr Opin Hematol. 2016;23(5):453-461. 4. Ansari SA, Pendurthi UR, Rao LVM. Role of cell surface lipids and thiol-disulphide exchange pathways in regulating the encryption and decryption of Tissue Factor. Thromb Haemost. 2019;119(6):860-870. 5. McVey JH, Boswell EJ, Takamiya O, et al. Exclusion of the first EGF domain of factor VII by a splice site mutation causes lethal factor VII deficiency. Blood 1998;92(3):920926. 6. Lawson JH, Kalafatis M, Stram S, Mann KG. A model for the tissue factor pathway to thrombin. I. An empirical study. Biol Chem. 1994;269(37):23357-23366. 7. Zelaya H, Rothmeier AS and Ruf W. Tissue factor at the crossroad of coagulation and cell signaling. J Thromb Haemost. 2018;16(10):1941-1952. 8. Hedner U, Glazer S, Pingel K, et al. Successful use of recombinant factor VIIa in a patient with severe hemophilia A during synovectomy. Lancet. 1988;2(8621):1193. 9. O'Hara PJ, Grant FJ, Haldeman BA, et al. Nucleotide sequence of the gene coding for human factor VII, a vitamin K-dependent protein participating in blood coagulation. Proc Natl Acad Sci U S A. 1987;84(15):51585162. 10. Mariani G, Herrmann FH, Dolce A, et al. International Factor VII Deficiency Study Group. Clinical phenotypes and factor VII genotype in congenital factor VII deficiency. Thromb Haemost. 2005;93(3):481-487.

360

translational levels. The multidisciplinary approaches have spanned structural biology, biochemistry, recombinant protein biotechnology, molecular and population genetics, pharmacology, clinics and epidemiology, and encompassed various areas of hematology, hemostasis and thrombosis. The large number of studies on FVII have not only served to substantially improve our knowledge on FVII biology and related phenotypes and to improve the quality of life of the individuals with FVII deficiency, but also to train three generations of scientists in different fields. Disclosures Pfizer Research Grant Contributions FB and GM wrote the Manuscript Acknowledgments The authors express their gratitude to Prof. Peter Lydyard (emeritus at UCL London, UK and associate at Georgia University, Tbilisi Georgia) for his very helpful text revision and Dr Barbara Lunghi for her help in the selection of references.

11. Herrmann FH, Wulff K, Auerswald G, et al. Greifswald Factor FVII Deficiency Study Group. Factor VII deficiency: clinical manifestation of 717 subjects from Europe and Latin America with mutations in the factor 7 gene. Haemophilia. 2009;15(1):267-280. 12. Giansily-Blaizot M, Rallapalli PM, Perkins SJ, et al. The EAHAD blood coagulation factor VII variant database. Hum Mutat. 2020;41(7):1209-1219. 13. Alexander B, Goldstein R, Landwehr G, Cook CD. Congenital SPCA deficiency: a hitherto unrecognized coagulation defect with hemorrhage rectified by serum and serum fractions. J Clin Invest. 1951;30(6): 596-608. 14. de Vries PS, Sabater-Lleal M, Huffman JE, et al. A genome-wide association study identifies new loci for factor VII and implicates factor VII in ischemic stroke etiology. Blood. 2019;133(9):967-977. 15. Bernardi F, Marchetti G, Pinotti M, et al. Factor VII gene polymorphisms contribute about one third of the factor VII level variation in plasma. Arterioscler Thromb Vasc Biol. 1996;16(1):72-76. 16. Meade TW, Mellows S, Brozovic M, et al. Haemostatic function and ischaemic heart disease: principal results of the Northwick Park Heart Study. Lancet. 1986;2(8506): 533-537. 17. Iacoviello L, Di Castelnuovo A, De Knijff P, et al. Polymorphisms in the coagulation factor VII gene and the risk of myocardial infarction. N Engl J Med. 1998;338(2):7985. 18. Girelli D, Russo C, Ferraresi P, et al. Polymorphisms in the factor VII gene and the risk of myocardial infarction in patients with coronary artery disease. N Engl J Med. 2000;343(11):774-780. 19. Hagen FS, Gray CL, O'Hara P, et al. Characterization of a cDNA coding for human factor VII. Proc Natl Acad Sci U S A. 1986;83(8):2412-2416. 20. Arbini AA, Pollak ES, Bayleran JK, High KA, Bauer KA. Severe factor VII deficiency due to a mutation disrupting a hepatocyte nuclear factor 4 binding site in the factor

VII promoter. Blood. 1997;89(1):176-82. 21. Barbon E, Pignani S, Branchini A, Bernardi F, Pinotti M, Bovolenta M. An engineered tale-transcription factor rescues transcription of factor VII impaired by promoter mutations and enhances its endogenous expression in hepatocytes. Sci Rep. 2016;6:28304. 22. Marchetti G, Patracchini P, Papacchini M, Ferrati M, Bernardi F. A polymorphism in the 5' region of coagulation factor VII gene (F7) caused by an inserted decanucleotide. Hum Genet. 1993;90(5):575-576. 23. van't Hooft FM, Silveira A, Tornvall P, et al. Two common functional polymorphisms in the promoter region of the coagulation factor VII gene determining plasma factor VII activity and mass concentration. Blood. 1999;93(10):3432-3441. 24. Misenheimer TM, Kumfer KT, Bates BE, Nettesheim ER, Schwartz BS. A candidate activation pathway for coagulation factor VII. Biochem J. 2019;476(19):2909-2926. 25. Morrisey JH, Macik BG, Neuenschwander PF, Comp PC. Qantitation of activated factor VII levels in plasma using a tissue factor mutant selectively deficient in promoting factor VII activation. Blood. 1993;81(3):734744. 26. Sorensen AB, Tuneew I, Anders Svensson L, et al. Beating tissue factor at its own game: Design and properties of a soluble tissue factor-independent coagulation factor VIIa. J Biol Chem. 2020;295(2):517-528. 27. Davidson CJ, Hirt RP, Lal K, et al. Molecular evolution of the vertebrate blood coagulation network. Thromb Haemost. 2003;89 (3):420-428. 28. Osterud B, Rapaport SI. Activation of factor IX by the reaction product of tissue factor and factor VII: additional pathway for initiating blood coagulation. Proc Natl Acad Sci U S A. 1977;74(12):5260-5264. 29. Banner DW, D'Arcy A, Chene C, et al. The crystal structure of the complex of blood coagulation factor VIIa with soluble tissue factor. Nature 1996;380(6569):41-46. 30. Rode-Mosbaek C, Nolan D, Persson E, Svergun DI, Bukrinsky JT, Vestergaard B.

haematologica | 2021; 106(2)

Coagulation factor VII, hemostasis and thrombosis Extensive small-angle X-ray scattering studies of blood coagulation factor VIIa reveal interdomain flexibility Biochemistry. 2010;49(45):9739-9745. 31. Girard TJ, Warren LA, Novotny WF, Bejcek BE, Miletich JP, Broze GJ Jr. Identification of the 1.4 kb and 4.0 kb messages for the lipoprotein associated coagulation inhibitor and expression of the encoded protein. Thromb Res. 1989;55(1):37-50. 32. Hackeng TM, Rosing J. Protein S as cofactor for TFPI. Arterioscler Thromb Vasc Biol. 2009;29(12):2015-2020. 33. Broze GJ Jr, Majerus PW. Purification and properties of human coagulation factor VII. J Biol Chem. 1980;255(4):1242-1247. 34. Agersø H, Brophy DF, Pelzer H, et al. Recombinant human factor VIIa (rFVIIa) cleared principally by antithrombin following intravenous administration in hemophilia patients. J Thromb Haemost. 2011;9(2):333-338. 35. Bernardi F, Castaman G, Pinotti M, et al. Mutation pattern in clinically asymptomatic coagulation factor VII deficiency. Hum Mutat. 1996;8(2):108-115. 36. Gallani D, Wheeler AP, Neff AT. Rare coagulation factor deficiencies. In Hoffman et al. Hematology, Basic Principles and Practice. Chapter 137:2034-50. Elsevier, Philadelphia (PA) USA, 2018. 37. Palla R, Peyvandi F, Shapiro A. Rare bleeding disorders: diagnosis and treatment. Blood. 2015;125(13):2052-2061. 38. World Federation of Hemophilia. Report on the Annual Global Survey 2017. 39. Abbonizio F, Hassan JH, Riccioni R, Arcieri R, Giampaolo A. National Registry of congenital bleeding disorders (AICE). Report 2017. Istituto Superiore di Sanità, Rapporti Istisan. 2019,III:54. 40. Bernardi F, Dolce A, Pinotti M, et al. International Factor VII Deficiency Study Group. Major differences in bleeding symptoms between factor VII deficiency and hemophilia B. J Thromb Haemost. 2009;7(5):774-779. 41. Di Minno MND, Dolce A, Mariani G on behalf of the STER Study Group. Bleeding symptoms at disease presentation and prediction of ensuing bleeding in inherited FVII deficiency. Thromb Haemost. 2013;109(6):1051-1059. 42. Castoldi E, Govers-Riemslag JW, Pinotti M, et al. Coinheritance of factor V (FV) Leiden enhances thrombin formation and is associated with a mild bleeding phenotype in patients homozygous for the FVII 9726þ5G>A (FVII Lazio) mutation. Blood. 2003;102(12):4014-4020. 43. Benlakhl F, Mura T, Schved JK, GiansilyBlaizot M on behalf of the French Study Group of FVII Deficiency. A retrospective analysis of 157 surgical procedures performed without replacement therapy in 83 unrelated factor VII-deficient patients. J Thromb Haemost. 2011;152(6):340-346. 44. Giansily-Blaizot M, Verdier R, BironAdréani C, et al. Analysis of biological phenotypes from 42 patients with inherited factor VII deficiency: can biological tests predict the bleeding risk? Haematologica. 2004;89(6):704-709. 45. Siboni SM, Biguzzi E, Mistretta C, Garagiola L, Peyvandi F. Long-term prophylaxis in severe FVII deficiency. Haemophilia. 2015;21(6):812-819. 46. Mariani G, Dolce A, Batorova A et al. Recombinant, activated factor VII for surgery in factor VII deficiency: a prospective evaluation – the surgical STER. Br J Haematol. 2011;152(3):340-346.

haematologica | 2021; 106(2)

47. Sevenet PO, Kaczor DA, Depasse F. Factor VII deficiency: from basics to clinical laboratory diagnosis and patient management. Clin Appl Thromb Hemost. 2017;23 (7):703-710. 48. Cid AR, Lorenzo JI, Haya S, Montoro JM, Casana P, Aznar JA. A comparison of FVII:c and FVIIa assays for the monitoring of recombinant factor VIIa treatment. Haemophilia. 2001,7(1):39-41. 49. Girolami A, Fabris F, Dal Bo Zanon R, Ghiotto G, Burul A. Factor VII Padua: a congenital coagulation disorder due to an abnormal factor VII with a peculiar activation pattern. J Lab Clin Med. 1978;91(3): 387-395. 50. O'Brien DP, Gale KM, Anderson JS, et al. Purification and characterization of factor VII 304-Gln: a variant molecule with reduced activity isolated from a clinically unaffected male. Blood. 1991;78(1):132140. 51. McVey JH, Boswell E, Mumford AD, Kemball-Cook G, Tuddenham EG. Factor VII deficiency and the FVII mutation database. Hum Mutat. 2001;17(1):3-17. 52. Ferraresi P, Balestra D, Guittard C, et al. Next-generation sequencing and recombinant expression characterized aberrant splicing mechanisms and provided correction strategies in factor VII deficiency. Haematologica. 2020;105(3):829-837. 53. Kemball-Cook G, Johnson DJ, Takamiya O, Banner DW, McVey JH, Tuddenham EG. Coagulation factor VII Gln100 --> Arg. Amino acid substitution at the epidermal growth factor 2-protease domain interface results in severely reduced tissue factor binding and procoagulant function. J Biol Chem. 1998;273(14):8516-8521. 54. Andersen E, Chollet ME, Baroni M, et al. The effect of the chemical chaperone 4phenylbutyrate on secretion and activity of the p.Q160R missense variant of coagulation factor FVII. Cell Biosci. 2019;9:69. 55. Toso R, Pinotti M, High KA, Pollak ES, Bernardi F. A frequent human coagulation Factor VII mutation (A294V, c152) in loop 140s affects the interaction with activators, tissue factor and substrates. Biochem J. 2002;363(Pt 2):411-416. 56. Herrmann FH, Wulff K, Strey R, Siegemund A, Astermark J, Schulman S, International Greifswald Registry of FVII deficiency. Variability of clinical manifestation of factor VII-deficiency in homozygous and heterozygous subjects of the european F7 gene mutation A294V. Haematologica. 2008;93 (8):1273-1275. 57. Etro D, Pinotti M, Wulff K, et al. The Gly331Ser mutation in factor VII in Europe and the Middle East. Haematologica. 2003;88(12):1434-1436. 58. Pinotti M, Toso R, Redaelli R, Berrettini M, Marchetti G, Bernardi F. Molecular mechanisms of FVII deficiency: expression of mutations clustered in the IVS7 donor splice site of factor VII gene. Blood. 1998;92(5):1646-1651. 59. Pinotti M, Rizzotto L, Balestra D, et al. U1snRNA-mediated rescue of mRNA processing in severe factor VII deficiency. Blood. 2008;111(5):2681-2684. 60. Balestra D, Faella A, Margaritis P, et al. An engineered U1 small nuclear RNA rescues splicing defective coagulation F7 gene expression in mice. J Thromb Haemost. 2014;12(2):177-185. 61. Rosen ED, Xu H, Liang Z, Martin JA, Suckow M, Castellino FJ. Generation of genetically-altered mice producing very low levels of coagulation factor VII.

Thromb Haemost. 2005;94(3):493-497. 62. Pavlova A, Preisler B, Driesen J, et al. Congenital combined deficiency of coagulation factors VII and X--different genetic mechanisms. Haemophilia. 2015;21(3):386391. 63. Branchini A, Ferrarese M, Lombardi S, Mari R, Bernardi F, Pinotti M. Differential functional readthrough over homozygous nonsense mutations contributes to the bleeding phenotype in coagulation factor VII deficiency. J Thromb Haemost. 2016;14(10): 1994-2000. 64. Branchini A, Rizzotto L, Mariani G, et al. Natural and engineered carboxy-terminal variants: decreased secretion and gain-offunction result in asymptomatic coagulation factor VII deficiency. Haematologica. 2012;97(5):705-709. 65. Toso R, Bernardi F, Tidd T, et al. Factor VII mutant V154G models a zymogen-like form of factor VIIa. Biochem J. 2003;369(Pt 3):563-571. 66. Augustsson C, Persson E. In vitro evidence of a tissue factor-independent mode of action of recombinant factor VIIa in hemophilia. Blood. 2104;124(20):3172-3174. 67. Morfini M, Batorova A, Mariani G, et al. Pharmacokinetic properties of recombinant FVIIa in inherited FVII deficiency account for a large distribution at steady state and a prolonged pharmakodinamic effect. Thromb Haemost. 2014;112(2):424-425. 68. Napolitano M, Giansily-Blaizot M, Dolce A, et al. Prophylaxis in congenital factor VII deficiency: indications, efficacy and safety. Results from the Seven Treatment Evaluation Registry (STER). Haematologica. 2013;98(4):538-544. 69. Kuperman AA, Barg AA, Fructhman Y, et al. Primary prophylaxis for children with severe congenital FVII deficiency. Clinical and laboratory assessment. Blood Cells Mol Dis. 2017;67:86-90. 70. Mariani G, Mannucci PM, Mazzucconi MG, Capitanio A. Treatment of congenital factor VII deficiency with a new concentrate. Thromb Haemost. 1978;39(3):675682. 71. Mariani G, Konkle AB, Kessler CM. Inhibitors in hemophilias. In Hoffman et al. Hematology, Basic Principles and Practice. 5th edition. Chapter 136:2023-33. Elsevier, Philadelphia (PA) USA, 2018. 72. Mariani G, Herrmann FH, Shulman S, et al. Thrombosis in inherited FVII deficiency. J Thromb Haemost. 2003;1(10):2153-2158. 73. Peyvandi F, Garagiola I, Menegatti M. Gynecological and obstetrical manifestations of inherited bleeding disorders in women. J Thromb Haemost. 2011;9(Suppl 1):S236-245. 74. Napolitano M, Di Minno MN, Batorova A, et al. Women with congenital factor VII deficiency: clinical phenotype and treatment options from two international studies. Haemophilia. 2016;22(5):752-759. 75. Menegatti M, Peyvandi F. Treatment of rare factor deficiencies other than hemophilia. Blood. 2019;133(5):415-424. 76. Bar-Ilan A, Livnat T, Hoffmann M, et al. In vitro characterization of MOD-5014, a novel longacting carboxy-terminal peptide (CTP)modified activated FVII. Haemophilia. 2018;24(3):477-486. 77. Laros-van Gorkom B, André Holme P, Joch C, et al. Pharmacokinetics and pharmacodynamics of a recombinant fusion protein linking activated coagulation factor VII with human albumin (rVIIa-FP) in patients with congenital FVII deficiency. Hematology. 2020;25(1):17-25.

361

F. Bernardi and G. Mariani et al. 78. Bern M, Nilsen J, Ferrarese M, et al. An engineered human albumin enhances halflife and transmucosal delivery when fused to protein-based biologics. Sci Transl Med. 2020;12(565):eabb0580. 79. Marcos-Contreras OA, Smith SM, Bellinger DA, et al. Sustained correction of FVII deficiency in dogs using AAV-mediated expression of zymogen FVII. Blood. 2016;127(5): 565-571. 80. Miller GJ. Dietary fatty acids and the haemostatic system. Atherosclerosis. 2005; 179(2):213-227. 81. Moor E, Silveira A, van't Hooft F, et al. Coagulation factor VII mass and activity in young men with myocardial infarction at a young age. Role of plasma lipoproteins and factor VII genotype. Arterioscler Thromb Vasc Biol. 1995;15(5):655-664. 82. Kamikubo Y, Mendolicchio GL, Zampolli A, et al. Selective factor VIII activation by the tissue factor–factor VIIa–factor Xa complex Blood. 2017; 130(14):1661-1670. 83. Nadir Y, Brenner B, Fux L, Shafat I, Attias J, Vlodavsky I. Heparanase enhances the generation of activated factor X in the presence of tissue factor and activated factor VII. Haematologica. 2010;95(11):1927-1934. 84. D'Alessandro E, Posma JJN, Spronk HMH, Ten Cate H. Tissue Factor (:Factor VIIa) in the heart and vasculature: more than an envelope. Thromb Res. 2018;168:130-137. 85. Rothmeier AS, Liu E, Chakrabarty S, et al. Identification of the integrin-binding site on coagulation factor VIIa required for proangiogenic PAR2 signaling. Blood. 2018;131(6):674-685. 86. Scarabin PY, Vissac AM, Kirzin JM, et al. Population correlates of coagulation factor VII. Importance of age, sex, and menopausal status as determinants of activated factor VII. Arterioscler Thromb Vasc Biol. 1996;16(9):1170-1176. 87. Green F, Kelleher C, Wilkes H, Temple A, Meade T, Humphries S. A common genetic polymorphism associated with lower coagulation factor VII levels in healthy individuals. Arterioscler Thromb. 1991;11(3):540546. 88. de Lange M, Snieder H, Ariens RA, Spector TD, Grant PJ. The genetics of haemostasis: a twin study. Lancet. 2001;357(9250):101105. 89. Eriksson-Berg M, Deguchi H, Hawe E, et al. Influence of factor VII gene polymorphisms

362

and environmental factors on plasma coagulation factor VII concentrations in middleaged women with and without manifest coronary heart disease. Thromb Haemost. 2005;93(2):351-358. 90. Mariani G, Bernardi F, Bertina R, et al. Serum phospholipids are the main environmental determinants of activated factor VII in the most common FVII genotype. European Union Concerted Action "Clotart". Haematologica. 1999;84(7):620626. 91. Sabater-Lleal M, Almasy L, MartínezMarchán E, et al. Genetic architecture of the F7 gene in a Spanish population: implication for mapping complex diseases and for functional assays. Clin Genet. 2006;69(5):420-428. 92. Smith NL, Chen MH, Dehghan A, et al. Novel associations of multiple genetic loci with plasma levels of factor VII, factor VIII, and von Willebrand Factor: The CHARGE (Cohorts for Heart and Aging Research in Genome Epidemiology) Consortium. Circulation. 2010;121(12):1382-1392. 93. Tang W, Schwienbacher C, Lopez LM, et al. Genetic associations for activated partial thromboplastin time and prothrombin time, their gene expression profiles, and risk of coronary artery disease meta-analysis. Am J Hum Genet. 2012;91(1):152-162. 94. Taylor KC, Lange LA, Zabaneh D, et al. A gene-centric association scan for coagulation factor VII levels in european and African Americans: the candidate gene association resource (CARe) consortium. Hum Mol Genet. 2011;20(17):3525-3534. 95. Kario K, Miyata T, Sakata T, Matsuo T, Kato H. Fluorogenic assay of activated factor VII. Plasma factor VIIa levels in relation to arterial cardiovascular diseases in Japanese. Arterioscler Thromb. 1994;14(2): 265-274. 96. Olson NC, Raffield LM, Lange LA, et al. Associations of activated coagulation factor VII and factor VIIa-antithrombin levels with genome-wide polymorphisms and cardiovascular disease risk. J Thromb Haemost. 2018;16(1):19-30. 97. Martinelli N, Girelli D, Baroni M, et al. Activated factor VII-antithrombin complex predicts mortality in patients with stable coronary artery disease: a cohort study. J Thromb Haemost. 2016;14(4):655-666. 98. Baroni M, Martinelli N, Lunghi B, et al.

Aptamer-modified FXa generation assays to investigate hypercoagulability in plasma from patients with ischemic heart disease. Thromb Res. 2020;189:140-146. 99. Merlini PA, Ardissino D, Oltrona L, Broccolino M, Coppola R, Mannucci PM. Heightened thrombin formation but normal plasma levels of activated factor VII in patients with acute coronary syndromes. Arterioscler Thromb Vasc Biol. 1995;15(10): 1675-1679. 100.Holm J, Tödt T, Berntorp E, Erhardt L. Failure of thrombolytic therapy in patients with myocardial infarction is associated with high plasma levels of factor VII antigen. Thromb Haemost. 1998;79(5):928931. 101.Folsom AR, Cushman M, Heckbert SR, Ohira T, Rasmussen-Torvik L, Tsai MY. Factor VII coagulant activity, factor VII 670A/C and -402G/A polymorphisms, and risk of venous thromboembolism. J Thromb Haemost. 2007;5(8):1674-1678. 102.Roldán V, Marín F, González-Conejero R, et al. Factor VII -323 decanucleotide D/I polymorphism in atrial fibrillation: implications for the prothrombotic state and stroke risk. Ann Med. 2008;40(7):553-559. 103.D'Ambrosio RL, D'Andrea G, Cappucci F, et al. Polymorphisms in factor II and factor VII genes modulate oral anticoagulation with warfarin. Haematologica. 2004;89 (12):1510-1516. 104.Buxhofer-Ausch V, Olcaydu D, Gisslinger B, et al. Decanucleotide insertion polymorphism of F7 significantly influences the risk of thrombosis in patients with essential thrombocythemia. Eur J Haematol. 2014;93 (2):103-111. 105.Ken-Dror G, Drenos F, Humphries SE, et al. Haplotype and genotype effects of the F7 gene on circulating factor VII, coagulation activation markers and incident coronary heart disease in UK men. J Thromb Haemost. 2010;8(11):2394-2403. 106.Donati MB, Zito F, Castelnuovo AD, Iacoviello L. Genes, coagulation and cardiovascular risk. J Hum Hypertens. 2000;14(6): 369-372. 107.Bernardi F, Arcieri P, Bertina RM, et al. Contribution of factor VII genotype to activated FVII levels. Differences in genotype frequencies between northern and southern European populations. Arterioscler Thromb Vasc Biol. 1997;17(11):2548-2553.

haematologica | 2021; 106(2)

ARTICLE

Transplantation

Standardized monitoring of cytomegalovirusspecific immunity can improve risk stratification of recurrent cytomegalovirus reactivation after hematopoietic stem cell transplantation

Eva Wagner-Drouet,1,* Daniel Teschner,1,* Christine Wolschke,2 Dietlinde Janson,2 Kerstin Schäfer-Eckart,3 Johannes Gärtner,3 Stephan Mielke,4,5 Martin Schreder,4,§ Guido Kobbe,6 Mustafa Kondakci,6 Inken Hilgendorf,7 Marie von Lilienfeld-Toal,7 Stefan Klein,8 Daniela Heidenreich,8 Sebastian Kreil,8 Mareike Verbeek,9 Sandra Graß,9 Markus Ditschkowski,10 Tanja Gromke,10 Martina Koch,11,‡ Monika Lindemann,12 Thomas Hünig,13 Traudel Schmidt,14 Anne Rascle,14 Harald Guldan,14 Sascha Barabas,14 Ludwig Deml,14 Ralf Wagner14,15 and Daniel Wolff16

Ferrata Storti Foundation

Haematologica 2021 Volume 106(2):363-374

Department of Hematology, Medical Oncology, and Pneumology, University Medical Center of the Johannes Gutenberg University, Mainz, Germany; 2Department of Stem Cell Transplantation, University Medical Center Hamburg-Eppendorf, Germany; 3 Oncology, Hematology and Bone Marrow Transplantation Unit, Klinikum Nord, Nürnberg, Germany; 4Department of Medicine II, University Medical Center, Würzburg, Germany; 5Department of Laboratory Medicine, CAST, Karolinska Institutet and University Hospital, Stockholm, Sweden; 6Department of Hematology, University Hospital Düsseldorf, Medical Faculty, Heinrich Heine University, Düsseldorf, Germany; 7Klinik für Innere Medizin II, Abteilung für Hämatologie und Internistische Onkologie, Universitätsklinikum, Jena, Germany; 8Department of Hematology and Oncology, UMM University Medical Center Mannheim, University of Heidelberg, Mannheim, Germany; 9III Medical Department, Hematology and Oncology, Klinikum Rechts der Isar, Technical University, Munich, Germany; 10Innere Klinik, Tumorforschung, University Hospital Essen, Germany; 11Department of Hepatobiliary Surgery and Transplantation, University Medical Center Hamburg-Eppendorf, Germany; 12Institute for Transfusion Medicine, University Hospital, Essen, Germany; 13Institute of Virology and Immunobiology, University Medical Center, Würzburg, Germany; 14Lophius Biosciences, Regensburg, Germany; 15Institute of Clinical Microbiology and Hygiene, University Medical Center Regensburg, Germany and 16Department of Internal Medicine III, Hematology and Oncology, University Medical Center Regensburg, Germany. 1

*EW-D and DT contributed equally as co-first authors.

Current address: First Department of Medicine, Center for Oncology and Hematology, Wilhelminenspital, Vienna, Austria

Current address: Department of General, Visceral and Transplantation Surgery, University Medical Center of the Johannes Gutenberg University, Mainz, Germany

‡

Correspondence: DANIEL WOLFF daniel.wolff@ukr.de RALF WAGNER ralf.wagner@ukr.de

ABSTRACT

ecurrence of cytomegalovirus reactivation remains a major cause of morbidity and mortality following allogeneic hematopoietic stem cell transplantation. Monitoring cytomegalovirus-specific cellular immunity using a standardized assay might improve the risk stratification of patients. A prospective multicenter study was conducted in 175 intermediate- and high-risk allogeneic hematopoietic stem cell transplant recipients under preemptive antiviral therapy. Cytomegalovirusspecific cellular immunity was measured using a standardized interferonγ enzyme-linked immunospot assay (T-Track® CMV). The primary aim was to evaluate the suitability of measuring cytomegalovirus-specific immunity after the end of treatment for a first cytomegalovirus reactivation to predict recurrent reactivation. Forty of 101 (39.6%) patients with a first cytomegalovirus reactivation experienced recurrent reactivations, mainly in the high-risk group (cytomegalovirus-seronegative donor/cytomegalovirus-seropositive recipient). The positive predictive value of T-Track® CMV (patients with a negative test after the first reactivation who experienced at least one recurrent reactivation) was 84.2% in high-risk patients. Kaplan-Meier analysis revealed a higher probability of recurrent cytomegalovirus reactivation in high-risk patients with a haematologica | 2021; 106(2)

Received: June 10, 2019. Accepted: December 18, 2019. Pre-published: December 26, 2019. https://doi.org/10.3324/haematol.2019.229252

363

E. Wagner-Drouet et al.

negative test after the first reactivation (hazard ratio 2.73; P=0.007). Interestingly, a post-hoc analysis considering T-Track® CMV measurements at day 100 after transplantation, a time point highly relevant for outpatient care, showed a positive predictive value of 90.0% in high-risk patients. Our results indicate that standardized cytomegalovirus-specific cellular immunity monitoring may allow improved risk stratification and management of recurrent cytomegalovirus reactivation after hematopoietic stem cell transplantation. This study was registered at www.clinicaltrials.gov as #NCT02156479.

Introduction Cytomegalovirus (CMV) infection and disease remain a serious cause of morbidity and mortality after allogeneic hematopoietic stem cell transplantation (HSCT).1–3 Thanks to efficient diagnosis and management of CMV, the incidence of death from CMV disease after HSCT has dropped to <10%.4–9 Adequate risk stratification is essential to identify and properly manage patients at risk of CMV reactivation. The main risk factors include donor (D) and recipient (R) CMV serostatus (D-/R+ defining high-risk patients), the use of mismatched or unrelated donors, graft-versus-host disease (GvHD), and intense immunosuppression.1,2 Close monitoring during the first 100 days after transplantation, in accordance with current guidelines, has greatly reduced the incidence of CMV-related complications. However, recurrent and late-occurring CMV reactivation remain major lifethreatening issues, and effective strategies for the prevention of late CMV disease prevail as an unmet medical need. This is particularly critical in outpatient care more than 100 days after HSCT when patients are less frequently monitored. Studies clearly identified a delay in global and CMVspecific immune reconstitution as a major risk factor for recurrent and late-onset CMV reactivation.10–13 Several CMV-specific immune monitoring assays have been described. They are based on the quantification of the number and/or functionality of immune cells targeted against CMV, using flow cytometry detection, an enzyme-linked immunosorbent assay (ELISA) or enzyme-linked immunospot (ELISpot).14,15 Multiple studies demonstrated the suitability of these methods for predicting recurrent and/or late CMV reactivation, resulting in the emergence of new risk stratification models based on the monitoring of CMV-specific cell-mediated immunity (CMV-CMI) together with CMV viral load.8,16–25 The lack of standardized assays does, however, render the comparison of most reported results difficult. Two standardized CMV-specific interferon (IFN)-γ ELISpot assays, based on the in vitro stimulation of peripheral blood mononuclear cells (PBMC) with IE-1 and pp65 peptides (T-SPOT®.CMV) or proteins (T-Track® CMV), have been described. T-Track® CMV is highly sensitive due to the use of urea-formulated T-activated® IE-1 and pp65 proteins, resulting in the activation of a broad spectrum of CMV-specific effector cells (including CD4+, CD8+ and NK cells).14,26–28 One study reported the utility of TSPOT®.CMV to predict the risk of a first treatment-requiring CMV reactivation after HSCT.29 Here we describe – to the best of our knowledge for the first time – the utility of the T-Track® CMV assay to predict recurrent and late-onset CMV reactivation after HSCT.

Methods Study design and participants A prospective multicenter study was conducted in 175 interme364

diate- and high-risk (D+/R+, D+/R-, D-/R+) allogeneic HSCT recipients. Recipients of a first-time bone marrow or peripheral blood transplantation from either a matched sibling, matched unrelated or mismatched unrelated donor of any gender and race, aged at least 18 years, and scheduled for preemptive antiviral therapy were eligible for participation in this study, which was approved by the relevant ethics committees (DIMDI’s registration number 00008544; University of Regensburg’s approval number 13-122-0282). Additional information is provided in the Online Supplementary Methods.

Viral load measurement CMV load was measured by quantitative polymerase chain reaction using validated, non-standardized protocols and equipment (Online Supplementary Methods). CMV reactivation was defined as a CMV viral load requiring antiviral treatment based on center-specific guidelines and/or physician’s decision. Accordingly, the term “CMV reactivation” will thereafter refer to antivirallytreated CMV reactivation.

Measurement and analysis of CMV-specific cell-mediated immunity Blood collection, peripheral blood mononuclear cell isolation and T-Track® CMV assays (Lophius Biosciences GmbH, Regensburg, Germany) were performed as described26,27,30 and as detailed in the Online Supplementary Methods. IFN-γ ELISpot test results were interpreted on the basis of square-root-transformed spot-forming cells (sqrt-SFC). Briefly, a test was considered positive if the mean of four replicate sqrt-SFC for 200,000 cells (SRM) resulting from IE-1 and/or pp65 stimulation was ≥sqrt(10) and if the difference of the mean of sqrt-SFC of the stimulated condition to that of the unstimulated condition (SRM[stimulated] SRM[unstimulated]) was ≥0.742.

Lymphocyte subpopulation count determination Lymphocyte subpopulations were characterized by flow cytometry from the same peripheral blood mononuclear cells and absolute cell counts were calculated using the peripheral blood absolute lymphocyte count determined at the same visit (Online Supplementary Methods).

Statistical analysis Calculations were performed with SAS 9.4 software, as detailed in the Online Supplementary Methods. Statistical analyses are presented for high-risk (D-/R+) and all patients. Differences in IE-1and pp65-specific SFC distributions between groups were tested using the Mann-Whitney U test. Qualitative (positive/negative) test results were compared using a χ2 test. The probability of recurrent CMV reactivation according to qualitative test results was estimated using Kaplan-Meier curves. Hazard ratio (HR) estimates were obtained by Cox regression analysis and differences in CMV reactivation probability between groups were tested using a log-rank test. Receiver operating characteristic (ROC) curves were generated and area under the curve (AUC) estimates were obtained by logistic regression. Two-sided P-values <0.05 were considered statistically significant. haematologica | 2021; 106(2)

Risk stratification of CMV recurrence after HSCT

Results Patients’ characteristics One hundred and seventy-five allogeneic HSCT recipients were enrolled. Twenty-one patients were excluded from the analysis, due to either protocol violation (n=8) or the absence of valid T-Track® CMV test results (n=13). The study flow diagram is shown in Online Supplementary Figure S2. Of the 154 HSCT recipients included in the final analysis, 101 (65.5%) experienced at least one treatmentrequiring CMV reactivation (hereafter designated as “CMV reactivation”) up to day 225 after transplantation (Table 1). Eight patients (4.6%) were diagnosed with CMV disease and 69 (44.8%) with GvHD (Table 1 and Online Supplementary Table S2). The majority of patients with GvHD (50/69 [72.5%] of all patients and 33/40 [82.5%] of D-/R+ patients) experienced at least one episode of CMV reactivation. This is in line with the strong immunosuppressive effect of steroids used for the treatment of GvHD.31,32 Of the 101 patients who had a CMV reactivation, 65 (64.4%) belonged to the D-/R+ high-risk group (Table 2). Sixty-one (60.4%) patients experienced only one CMV reactivation while 40 patients developed either two (n=24) or three (n=16) CMV reacti-

Table 1. Patients’ characteristics.

Study population, N (%) Gender, n (%) Male Female Age in years, median (range) Underlying disease, n (%) Acute myeloid leukemia Myelodysplastic syndrome Acute lymphoid leukemia Non-Hodgkin lymphoma Multiple myeloma Osteomyelofibrosis Chronic myeloid leukemia Chronic lymphoid leukemia Severe aplastic anemia Donor (D) / recipient (R) CMV serostatus, n (%) D+/R+ D+/RD-/R+ Stem cell source, n (%) Bone marrow Peripheral blood Donor source, n (%) Matched sibling Matched unrelated donor Mismatched unrelated donor Conditioning regimen, n (%) Non-myeloablative Myeloablative, standard Myeloablative, toxicity-reduced At least one treatment-requiring CMV reactivation, n (%) CMV disease, n (%) Graft-versus-host disease, n (%) Infections other than CMV (after day 45), n (%) Death, n (%) CMV: cytomegalovirus.

haematologica | 2021; 106(2)

154 (100%) 88 (57.1) 66 (42.9) 58 (20-75) 79 (51.3) 24 (15.6) 17 (11.0) 12 (7.8) 10 (6.5) 4 (2.6) 3 (1.95) 3 (1.95) 2 (1.3)

vations (hereafter referred to as “recurrent CMV reactivation”) (Table 2). Most of the HSCT recipients who had recurrent CMV reactivation were high-risk patients (37/40 [92.5%]) (Table 2). Therefore, we focused on highrisk patients as the clinically relevant population with regards to the risk of CMV recurrence after HSCT. The median (range) time to CMV reactivation in D-/R+ patients for the first, second and third CMV reactivation was 37 (19-58), 109 (63-188) and 174 (119-219) days after transplantation, respectively (Online Supplementary Figure S3B). The respective median (range) CMV viral load is presented in Figure 1.

Measurement of CMV-specific cell-mediated immunity over time after hematopoietic stem cell transplantation CMV-CMI was evaluated using a standardized IFN-γ ELISpot-based assay (T-Track® CMV).26,27,30 A total of 647 valid test results were included in the analysis. The distribution of spot-forming cells (SFC) was analyzed over time after transplantation in response to the CMV proteins IE1 and pp65 (Online Supplementary Figure S4A). Overall, the response to IE-1 antigen was lower than that to pp65 throughout the study, especially in D-/R+ patients. A significant increase in the response to pp65 was apparent in D-/R+ patients over time (Online Supplementary Figure S4A). Accordingly, while the proportion of IE-1-positive tests remained low (up to 33.3% around day 145), that of pp65-positive tests increased to 64.5% in D-/R+ patients (Online Supplementary Figure S4B). Interestingly, the percentage of T-Track® CMV-positive tests (considering both IE-1 and pp65 markers) was consistently higher than that of pp65 alone, reaching 77.4% of positive tests around day 145 in D-/R+ patients (Online Supplementary Figure S4B). Thus, although generating lower spot counts, IE-1 antigen contributes significantly to T-Track® CMV test positivity. The patterns of SFC distribution relative to the start of the first CMV reactivation were comparable (Online Supplementary Figure S4C).

CMV-specific cell-mediated immunity measured after the end of treatment of a first CMV reactivation can predict recurrence of CMV reactivation

9 (5.8) 145 (94.2)

The primary aim of the study was to evaluate the suitability of measuring CMV-CMI after the end of treatment of a first-occurring CMV reactivation to predict freedom from or occurrence of a subsequent CMV reactivation (Figure 2A). CMV-CMI was measured on the day of discontinuation of antiviral treatment (day 0), and on days 7

31 (20.1) 92 (59.8) 31 (20.1)

Table 2. Cytomegalovirus reactivation according to the patients’ cytomegalovirus serostatus.

53 (34.4) 19 (12.3) 82 (53.3)

All 41 (26.6) 75 (48.7) 38 (24.7) 101 (65.6) 8 (4.6) 69 (44.8) 52 (33.8) 21 (13.6)

Study population, n No documented CMV reactivation, n At least one CMV reactivation, n One CMV reactivation only, n Recurrent CMV reactivation, n Two CMV reactivations, n Three CMV reactivations, n

154 53 101 61 40 24 16

D-/R+ D+/R+ D+/R82 17 65 28 37 21 16

53 20 33 30 3 3 0

19 16 3 3 0 0 0

CMV: cytomegalovirus; D-: CMV-negative donor; R+: CMV-positive recipient; D+: CMVpositive donor; R-: CMV-negative recipient.

365

E. Wagner-Drouet et al.

Figure 1. Cytomegalovirus (CMV) viral load at start of the first, second and third CMV reactivations. Median (range) CMV viral load (VL) at the time of the first, second and third CMV reactivations is shown for all patients and high-risk CMV-negative donor/CMV-positive recipient (D-/R+) pairs. It should be emphasized that VL measurements were not standardized among centers. CMV load was measured by quantitative polymerase chain reaction from whole blood (9 centers) or plasma (1 center) using either a commercial assay (Abbott RealTime CMV; 2 centers) or validated in-house protocols and equipment (8 centers). Accordingly, treatment-requiring viral load thresholds were center-specific and no analysis correlating spot-forming-cell counts to VL was planned in this study.

and 14 thereafter. The first available measurement was considered for the analysis. In fact, measurements on days 0, 7 and 14 contributed 36/76 (47.4%), 29/76 (38.1%) and 11/76 (14.5%) test results, respectively, to this analysis. SFC levels after a first CMV reactivation were compared between patients who experienced no further CMV reactivation and those with one or two subsequent (i.e., recurrent) CMV reactivations. A significant difference in SFC distribution was observed between the two groups, for both IE-1- and pp65-specific SFC levels, when considering all patients (Mann-Whitney U test, P<0.001). The median SFC was, respectively, 10- and 40-times higher in patients free from recurrent CMV reactivation (based on squared mean of square-root-transformed [SRM^2] values) (Figure 2B). In the high-risk population, a significant difference was observed for pp65-induced response (Mann-Whitney U test, P=0.001), but not for IE-1-mediated response (Mann-Whitney U test, P=0.724) which was very low in both groups (Figure 2B). In accordance with these results, a ROC analysis in D-/R+ patients revealed AUC estimates significant for pp65-specific tests (AUC 0.780, 95% confidence interval [95% CI]: 0.642-0.917]; P<0.001) but not for IE-1-specific tests (Figure 2C). Interestingly, a post-hoc analysis of these data upon normalization of SFC values to absolute lymphocyte counts derived from peripheral blood counts (expressed as SFC/mL blood) showed a stronger discrimination of SFC distributions between patients without and with recurrent CMV reactivation, including in D-/R+ patients (Mann-Whitney U test, P=0.010 [IE-1] and P<0.001 [pp65]), and an improved predictive value in ROC analysis (AUC 0.760 [95% CI: 0.599-0.921], P=0.002 for IE-1 and AUC 0.863 [95% CI: 0.741-0.986], P<0.001 for pp65) (Online Supplementary Figure S5), compared to the normalization to 200,000 lymphocytes. SFC distributions according to conditioning regimen and GvHD occurrence 366

are presented in Online Supplementary Figures S6 and S7, respectively. The diagnostic accuracy of the ELISpot assay was determined in terms of sensitivity (patients with recurrent CMV reactivation had a negative test result after a first CMV reactivation) and specificity (patients free from recurrent CMV reactivation had a positive test result after a first CMV reactivation) (Table 3). Likely due to the low SFC levels and high proportion of negative test results induced by IE-1, the IE-1-specific test alone showed limited performance in the D-/R+ population (Table 3). By contrast, the performance of pp65-specific positive tests, alone or in combination with IE-1, to correctly identify patients free of future recurrent CMV reactivation was high, with a specificity of 77.8% (pp65) and 83.3% (pp65 and IE-1 combined) in D-/R+ patients (Table 3). The sensitivity of pp65-specific tests, alone or in combination with IE-1, in D-/R+ patients was 58.6% and 55.2%, respectively (Table 3). The positive predictive value (PPV; patients with a negative test after the first CMV reactivation had a subsequent recurrent CMV reactivation) of pp65-specific tests in D-/R+ patients reached 80.9% (pp65 test alone) and 84.2% (pp65 and IE-1 tests combined) while the respective negative predictive values (NPV; patients with a positive test after the first CMV reactivation did not have a subsequent CMV reactivation) were low (53.8% and 53.6%, respectively) (Table 3). The probability of recurrent CMV reactivation in patients with positive and negative ELISpot tests was estimated using Kaplan-Meier curves (Figure 2D). In line with the previous observations, the difference in probability of recurrent CMV reactivation between patients with a positive or negative pp65-specific test result after the first reactivation was highly significant, both in the total population (HR 4.91; log-rank test, P<0.001) and in the D-/R+ group (HR 2.52; log-rank test, P=0.013) (Figure haematologica | 2021; 106(2)

Risk stratification of CMV recurrence after HSCT

Figure 2. Legend on following page.

haematologica | 2021; 106(2)

367

E. Wagner-Drouet et al. Figure 2 (previous page). Performance of cytomegalovirus (CMV)-specific cell-mediated immunity measured after the end of a first CMV reactivation to predict freedom from and/or occurrence of recurrent CMV reactivation. (A) Interferon-γ enzyme-linked immunospot (ELISpot) was performed after the end of antiviral therapy for a first CMV reactivation, at up to three time points relative to the end of treatment, namely day 0 (d0), day 7 (d7) and day 14 (d14). The first available measurement was considered for the analysis. (B) Quantitative ELISpot results in response to CMV proteins IE-1 and pp65 were evaluated on the basis of the mean of squareroot-transformed (SRM) spot-forming cells (SFC), as described in the Methods section. Differences in SFC distribution between patients with only one CMV reactivation and those with recurrent CMV reactivation were evaluated using a Mann-Whitney U test. Respective P-values are shown under each graph. For the sake of simplicity, scatter plots are depicted as squared SRM values (SRM^2). The median and interquartile range of the SRM^2 SFC are shown above each graph. Additional information (minimum, maximum, 10th and 90th percentiles) are shown in Online Supplementary Table S3. Due to the log scale representation, values of zero SRM^2 were replaced by 0.01 (y-axis), meaning that baseline values shown at y=0.01 are actually equal to zero. Red triangles and blue dots represent negative and positive tests, respectively, defined according to the rules described in the Methods section. Of note, of the three CMV-negative donor/CMV-positive recipients (D-/R+) with a documented recurrent CMV reactivation and with high pp65-SFC after the first CMV reactivation (251 to 386 SFC/200,000 lymphocytes) one was treated for recurrent CMV although the viral load was below the center-specific threshold (0 or 100 copies/mL) in the 10 days preceding the start of treatment and at all time points thereafter; a second patient had a first treatment initiated for a viral load below the center-specific threshold, after which high pp65-specific SFC dropped dramatically over time before the start of treatment of a CMV reactivation with a viral load above the threshold; the third patient had a lengthy (>3 months) first CMV reactivation with a high sustained viral load (up to 110,000 copies/mL), likely reflecting refractory CMV.55 Importantly, the duration of antiviral therapy for a first CMV reactivation was comparable in patients without and with recurrent reactivation (median [range] duration of 25 [4-94] and 31 [3-77] days, respectively, in all patients [Mann-Whitney U test, P=0.336]; median [range] duration of 31 [4-94] and 34 [7-77] days, respectively, in D-/R+ ptients [Mann-Whitney U test, P=0.677]). (C) Prediction of CMV reactivation recurrence based on IE-1- and pp65-specific SFC counts measured at the end of treatment of a first CMV reactivation was evaluated by receiver operating characteristic curve analysis. Area under the curve estimates, 95% confidence intervals and respective P-values are indicated within each graph. (D) Cumulative probability of CMV reactivation recurrence based on IE-1- and pp65-specific qualitative test results after a first CMV reactivation, evaluated as described in the Methods section. In the case that both IE-1 and pp65 test results are considered (T-Track® CMV; right panels), a test is positive when at least one IE-1 and/or pp65 test is positive and a test is negative when both IE-1 and pp65 tests are negative. Kaplan-Meier analyses were performed and the respective hazard ratios and P-values are shown within each graph. / indicate censored observations. The median (range) follow-up time after the T-Track® CMV measurement was 137 (35-180) days in patients with no documented recurrent CMV reactivation (censored). The median (range) time to recurrent CMV reactivation after the T-Track® CMV measurement was 24 (2-77) days. Moreover, the last recurrent CMV event in the case of a pp65- (and T-Track® CMV)-positive test result occurred 56 days after the end of antiviral therapy, compared to 77 days in the case of a pp65- and IE-1-negative test result. In (B-D), statistically significant P-values are in bold. CMI: cellmediated immunity; HSCT: hematopoietic stem cell transplantation; IQR: interquartile range; MWU: Mann-Whitney U test; D-/R+: CMV-negative donor/CMV-positive recipient; AUC: area under curve; 95% CI: 95% confidence interval; HR: hazard ratio.

Table 3. Diagnostic accuracy in identifying patients with and without recurrent cytomegalovirus (CMV) reactivation based on CMV-specific negative and positive enzyme-linked immunospot test results after the first CMV reactivation.

Population

All patients

Marker

Sensitivity

Specificity

Chi-square

PPVb

NPVb

IE-1

73.3% (22/30) [95% CI: 54.1-87.7%] 58.1% (18/31) [95% CI: 39.1-75.5%] 53.3% (16/30) [95% CI: 34.3-71.7%] 75.9% (22/29) [95% CI: 56.5-89.7%] 58.6% (17/29) [95% CI: 38.9-76.5%] 55.2% (16/29) [95% CI: 35.7-73.6%]

56.8% (25/44) [95% CI: 41.0-71.7%] 88.4% (38/43) [95% CI: 74.9-96.1%] 93.2% (41/44) [95% CI: 81.3-98.6%] 17.7% (3/17) [95% CI: 3.8-43.4%] 77.8% (14/18) [95% CI: 52.4-93.6%] 83.3% (15/18) [95% CI: 58.6-96.4%]

P=0.010

P<0.001

P=0.606

61.1% (22/36) 80.9% (17/21) 84.2% (16/19)

30.0% (3/10) 53.8% (14/26) 53.6% (15/28)

pp65 IE-1, pp65a IE-1

D-/R+ patients

pp65 IE-1, pp65a

P=0.015 P=0.009

a T-Track® CMV assay: the test is positive when at least one of the IE-1- and/or pp65-specific response is positive, and the test is negative when both IE-1- and pp65-specific responses are negative; bPositive and negative predictive values were not calculated in the “all patients” population because of the imbalance toward CMV-negative donor/CMV-positive recipient patients in that group. 95% CI: 95% confidence interval; CMV: cytomegalovirus; D/R: donor/recipient CMV serostatus; NPV: negative predictive value; PPV: positive predictive value.

2D). Interestingly, the performance of the pp65 test was improved by the combination with IE-1 (T-Track® CMV test), in all patients (HR 5.68; log-rank test, P<0.001) and in D-/R+ patients (HR 2.73; log-rank test, P=0.007) (Figure 2D). To better understand the usability of the assay in terms of clinical cutoff, we evaluated the PPV (patients with SFC ≤ threshold after the first CMV reactivation had a recurrent CMV reactivation) and NPV (patients with SFC > threshold after the first CMV reactivation did not have a recurrent CMV reactivation) of pp65-specific response in D-/R+ patients at low, intermediate and high SFC counts (Table 4). A NPV of 100% (3/3) was observed for a threshold of 386 SFC/200,000 lymphocytes in association with a PPV of 65.9% (29/44). A NPV of 75.0% (9/12) and PPV of 74.3% (26/35) were found for a SFC of 40 SFC/200,000 lymphocytes. Higher PPV (87.5% [21/24] to 100% [8/8]) and lower NPV (65.2% [15/23] to 46.2% [18/39]) were observed at lower (9 and less) SFC counts (Table 4), in line with the results described above (Table 3). 368

Benefit of monitoring CMV-specific cell-mediated immunity over absolute T-cell counts We next compared the performance of CMV-CMI to that of absolute lymphocyte (T and NK cells) counts measured after end of treatment of a first CMV reactivation to predict subsequent CMV reactivation episodes. Multicolor flow cytometry was performed using remaining peripheral blood mononuclear cells. NK and T (total, naïve and memory) cell levels were expressed as absolute cell counts. The first visit with existing absolute cell counts following the end of antiviral therapy (of day 0, 7 and 14) was considered for the analysis. Absolute cell counts were significantly higher in patients with no recurrent CMV reactivation in all cases, except for total and memory CD4+ T cells in D-/R+ patients (Online Supplementary Figure S8A). In ROC analyses, total lymphocytes as well as total and naïve CD8+ cell populations showed a good predictive value for recurrent CMV reactivation, with AUC estimates between 0.813 and 0.833 in D-/R+ patients (Online Supplementary Figure S8B), thus in a haematologica | 2021; 106(2)

Risk stratification of CMV recurrence after HSCT Table 4. Positive and negative predictive values of low, intermediate and high spot-forming-cell counts of a pp65-specific enzyme-linked immunospot assay after the end of treatment for a first cytomegalovirus (CMV) reactivation to predict the occurrence of future recurrent CMV reactivation in high-risk hematopoietic stem cell transplant patients.

Patient population D-/R+

Marker

Thresholda

PPV

NPV

pp65

386 40 9 0

65.9% (29/44) 74.3% (26/35) 87.5% (21/24) 100% (8/8)

100% (3/3) 75.0% (9/12) 65.2% (15/23) 46.2% (18/39)

% patients above threshold 6.4% 25.5% 48.9% 83.0%

(3/47) (12/47) (23/47) (39/47)

SFC (SRM^2)/200,000 lymphocytes (stimulated minus unstimulated condition) rounded to closest spot count; thresholds were derived from Receiver operating characteristic curve data; the threshold of 9 SFC (SRM^2)/200,000 lymphocytes showed the highest sensitivity+specificity value. SFC: spot-forming cells; SRM^2: squared mean of square-roottransformed; CMV,: cytomegalovirus; D/R: donor/recipient CMV serostatus; NPV: negative predictive value; PPV: positive predictive value. a

comparable range to that of the AUC derived from pp65specific ELISpot test results (Figure 2C and Online Supplementary Figure S5C). NK cells as well as total, naïve and memory CD4+ T-cell subpopulations showed a lower predictive value, especially in D-/R+ patients, with AUC estimates between 0.615 and 0.729 (Online Supplementary Figure S8B). A ROC analysis of paired pp65-specific ELISpot results and absolute lymphocyte and T-cell counts obtained at the same visit after the end of therapy showed no statistically significant difference in predictability, although pp65-specific ELISpot results normalized to absolute lymphocyte counts tended to perform better than absolute cell counts (Online Supplementary Figure S8C).

CMV-specific cell-mediated immunity at day 100 post-transplantation can predict late recurrent CMV reactivation (post-hoc analysis) Accurate prediction of future recurrent CMV reactivation is particularly critical when patients are released from close monitoring in an outpatient setting around 3 months after HSCT.4–6 We conducted two post-hoc analyses to determine whether CMV-CMI monitoring at the fixed time of day 100 could identify patients at risk of future (i.e. late) CMV reactivation. The first analysis considered all patients, regardless of a possible existence of CMV reactivation prior to day 100, thus assessing occurrence of late CMV reactivation generally (Online Supplementary Figure S9). The second analysis focused on patients who experienced CMV reactivation prior to day 100, thus investigating the usefulness of the IFN-γ ELISpot measured around day 100 to predict late recurrent CMV reactivation (Figure 3). T-Track® CMV test results acquired between day 80 and day 100 after transplantation were considered in patients with no ongoing CMV reactivation (Online Supplementary Methods). Interestingly, patients (including those in the high-risk group) who did not experience CMV reactivation up to day 80-100 did not experience CMV reactivation thereafter (Online Supplementary Figure S9B, orange-labeled dots). A majority of these patients presented low IE-1 and pp65-specific test results throughout the study (Online Supplementary Figure S9B and data not shown). Such sustained low responsiveness in patients with no CMV reactivation was previously reported.33 Its cause remains to be investigated. Consequently, performance of the ELISpot test at day 100 in this global analysis of late CMV reactivation was low (Online Supplementary Figure S9B-D; Online Supplementary Table S5). On the other hand, the analysis of ELISpot test results at day 100 in patients with an earlier CMV reactivation (Figure 3A) revealed a significant difference in SFC counts between patients without and with haematologica | 2021; 106(2)

late recurrent CMV reactivation (Mann-Whitney U test, Pvalue between 0.013 and <0.001), with the exception of IE-1-specific response in D-/R+ patients (Mann-Whitney U test, P=0.277) (Figure 3B). All patients with late recurrent CMV reactivation belonged to the high-risk group. In ROC analyses, AUC estimates for pp65 test results were 0.811 (P=0.003) in the D-/R+ population and 0.911 (P<0.001) in the total population (Figure 3C). pp65-specific response showed the best diagnostic accuracy, with a sensitivity and specificity of 75.0% and 90.9%, and a PPV and NPV of 90.0% and 76.9% in the D-/R+ population, respectively (Table 5). The probability of late recurrent CMV reactivation was significantly higher in D-/R+ patients with a pp65-negative test at day 100 (HR 6.34; log-rank test, P=0.002). In this setting, the response to IE1 did not improve the performance of pp65 (Figure 3D). Remarkably, the probability of recurrent CMV in the case of pp65- and T-Track® CMV-positive tests was lower when measured at day 100 than when measured directly after a first CMV reactivation (compare blue curves in Figures 2D and 3D).

Discussion This study demonstrates for the first time the suitability of a standardized CMV-specific IFN-γ ELISpot assay to predict recurrence of CMV reactivation after HSCT, with a particular focus on clinically relevant high-risk D-/R+ patients. Overall, the response to pp65 antigen in the ELISpot assay was higher than that of IE-1, as previously reported.26,27,30 Although the IE-1-specific response had low or no predictive value alone, it improved the assay performance in combination with pp65-specific tests after the first CMV reactivation. Diagnostic accuracy and time-to-event analyses indicated that negative IFN-γ ELISpot test results could predict recurrence of CMV reactivation when measured after the end of anti-CMV therapy or around day 100 after HSCT. Positivity of the T-Track® CMV test is based on several rules, the main one being that CMV antigen-stimulated conditions must yield ≥10 SFC (or mean of sqrt SFC ≥3.16) (Online Supplementary Methods). This raises the possibility that the technical cutoff of the assay might be a relevant clinical cutoff for the prediction of recurrent CMV reactivation (PPV >80%). Whether the positivity cutoff of TTrack® CMV is indeed a valid clinical cutoff to predict recurrent CMV reactivation in high-risk D-/R+ patients could be addressed in a randomized interventional study. In fact, such an approach is currently being appraised in high-risk solid-organ transplant recipients in a study aim369

E. Wagner-Drouet et al. A

Figure 3. Legend on following page.

370

haematologica | 2021; 106(2)

Risk stratification of CMV recurrence after HSCT Figure 3 (previous page). Performance of cytomegalovirus (CMV) cell-mediated immunity measured around day 100 after hematopoietic stem cell transplantation to predict freedom from and occurrence of late recurrent CMV reactivation (post-hoc). (A) Interferon-γ enzyme-linked immunospot (ELISpot) tests performed between day 80 (d80) and day 100 (d100) after hematopoietic stem cell transplantation in patients with an earlier CMV reactivation were used for the analysis. In the case that several measurement were available, the one closest to day 100 was considered. (B) Quantitative ELISpot results in response to CMV IE-1 and pp65 proteins were evaluated as detailed in the legend to Figure 2. Differences in spot-forming cell (SFC) distribution between patients with a CMV reactivation before day 100 only and those with a late (after day 100) recurrent CMV reactivation were evaluated using a Mann-Whitney U test. The median and interquartile range of the squared mean of square-root-transformed (SRM^2) SFC are shown above each graph. Additional information (minimum, maximum, 10th and 90th percentiles) are shown in Online Supplementary Table S4. Due to the log scale representation, values of zero SRM^2 were replaced by 0.01 (y-axis), meaning that baseline values shown at y=0.01 are actually equal to zero. Red triangles and blue dots represent negative and positive tests, respectively, defined according to the rules described in the Online Supplementary Methods. (C) Prediction of late recurrent CMV reactivation based on IE-1- and pp65-specific SFC counts measured around day 100 was evaluated by receiver operating characteristic curve analysis, as detailed in the legend to Figure 2. (D) Cumulative incidence of recurrent CMV reactivation based on IE1- and pp65-specific qualitative test results around day 100, was evaluated as detailed in the legend to Figure 2. Median (range) follow up time after T-Track® CMV measurement was 132 (59-145) days in (censored) patients with no documented late recurrent CMV reactivation. The median (range) time to late recurrent CMV reactivation after T-Track® CMV measurement was 119 (105-165) days. Moreover, the last recurrent CMV event in the case of a pp65- (and T-Track® CMV)-positive test result occurred 31 days after the day-100 measurement, compared to 77 days in the case of a pp65- and IE-1-negative test result. CMI: cell-mediated immunity; HSCT: hematopoietic stem cell transplantation; IQR: interquartile range; MWU: Mann-Whitney U test; D-/R+: CMV-negative donor/CMV-positive recipient; AUC: area under curve; 95% CI: 95% confidence interval; HR: hazard ratio.

Table 5. Diagnostic accuracy in identifying patients with and without late (beyond day 100 after hematopoietic stem cell transplantation) recurrent cytomegalovirus (CMV) reactivation based on CMV-specific negative and positive enzyme-linked immunospot test results at day 80 to 100.

Population

Marker

Sensitivity

Specificity

Chi-square

PPVb

NPVb

IE-1

76.9% (10/13) [95% CI: 46.2-95.0%] 75.0% (9/12) [95% CI: 42.8-94.5%] 69.2% (9/13) [95% CI: 38.6-90.9%] 76.9% (10/13) [95% CI: 46.2-95.0%] 75.0% (9/12) [95% CI: 42.8-94.5%] 69.2% (9/13) [95% CI: 38.6-90.9%]

60.0% (18/30) [95% CI: 40.6-77.3%] 96.7% (29/30) [95% CI: 82.8-99.9%] 96.7% (29/30) [95% CI: 82.8-99.9%] 20.0% (2/10) [95% CI: 2.5-55.6%] 90.9% (10/11) [95% CI: 58.7-99.8%] 90.9% (10/11) [95% CI: 58.7-99.8%]

P=0.026

P<0.001

P<0.001 P=0.859

55.6% (10/18) 90.0% (9/10) 90.0% (9/10)

40.0% (2/5) 76.9% (10/13) 71.4% (10/14)

All patients

pp65 IE-1, pp65a IE-1

D-/R+ patients

pp65 IE-1, pp65a

P=0.001 P=0.003

T-Track® CMV assay: test is positive when at least one of the IE-1- and/or pp65-specific response is positive, and the test is negative when both IE-1- and pp65-specific responses are negative; bPositive and negative predictive values were not calculated in the “all patients” population because of the imbalance toward donor-negative/recipient-positive pairs in that group. 95% CI, 95% confidence interval; CMV, cytomegalovirus; D/R, donor/recipient CMV serostatus; NPV, negative predictive value; PPV, positive predictive value. a

ing to steer the duration of antiviral prophylaxis based on CMV CMV-CMI monitoring using T-Track® (ClinicalTrials.gov Identifier: NCT02538172). The finding that the IFN-γ ELISpot assay measured after a first CMV reactivation performs at least as well as the measurement of absolute lymphocyte subset counts to predict subsequent CMV reactivation offers the attractive possibility of introducing T-Track® CMV in complement to the currently implemented absolute T-cell count measurement to improve risk stratification of recurrent CMV after HSCT. The value of absolute lymphocyte counts13,34,35 and of absolute lymphocyte subsets15,17,32,36–40 to predict CMV infection and/or recurrence is well documented, in line with the described protective role of CD4+, CD8+, γd T and NK cells against CMV reactivation.15,40–48 In general, absolute T-cell counts recovery correlates well with CMVspecific T-cell reconstitution in vivo,17,32,47,49 and absolute lymphocyte, CD4+ and/or CD8+ T-cell counts have been proposed as surrogate markers for CMV-specific cellular immunity.34,35,47 The good predictive value of absolute lymphocyte counts probably explains the improved performance of the ELISpot assay observed upon normalization to absolute lymphocytes, which might be a relevant normalization method in HSCT patients. On the other hand, absolute T-cell counts do not always reflect CMV-specific immune reconstitution.49,50 Recovery of CMV-specific T cells is often delayed relative to that of bulk T cells, especially in patients with recurrent CMV reactivation.49 Moreover, dysfunctional CD8+ T cells spehaematologica | 2021; 106(2)

cific for CMV accumulate in patients with recurrent CMV reactivation.51 Thus, functionality (i.e., the capacity of cells to respond to antigen stimulation) rather than the number of T cells is more likely to accurately reflect protection against recurrent CMV reactivation. Accordingly, we observed several cases of discordant absolute CD8+ T cell counts and IFN-γ ELISpot measurements in favor of the ELISpot assay, in particular cases of high CD8+ T-cell count together with low T-Track® CMV results prior to or during CMV reactivation episodes (data not shown). Therefore, CMV-specific immune monitoring using a standardized assay, while complementing the current approach of absolute T-cell count monitoring, provides an additional value for the risk stratification of recurrent CMV reactivation in HSCT patients. Negative ELISpot test results after the first CMV reactivation and at day 100 predicted subsequent recurrent CMV reactivation with a PPV of 80% and 90%, respectively. Measuring pp65-specific response at day 100 rather than directly after the end of a first CMV reactivation also improved the NPV (77% vs. 54%, respectively). This suggests that measuring CMV-CMI at the fixed time point of day 100 might be of greater clinical relevance to predict recurrence of CMV. The apparent improvement in NPV (and PPV) of the day-100 measurement might be in part due to the overall later measurement time (median [range] time of 93 [80-100] days), compared to the time of measurement at the end of treatment for a first CMV reactivation, which ranged from day 45 to 186 after transplanta371

E. Wagner-Drouet et al.

tion (median 82 days). Moreover, most (31/44 [70.4%]) of the day-100 measurements took place 14 days or later relative to the end of a first antiviral treatment, while most (36/76 [47.4%]) of the measurements at the end of a first CMV reactivation were from the day of discontinuation of antiviral treatment (day 0). This raises the possibility that measuring CMV-CMI 14 days after the end of a treatment-requiring CMV reactivation – rather than on the day of antiviral treatment discontinuation – might improve the predictive value for protection against recurrent reactivation. This might be explained by the ongoing immune reconstitution that takes place after HSCT. Indeed, a paired comparison of existing day 0, day 7 and day 14 measurements revealed a statistically significant increase in SFC counts between day 7 (or day 0) and day 14 in patients with no future recurrent CMV reactivation (data not shown). A future study should determine whether measuring CMV-CMI 14 days after the end of antiviral treatment and/or whether monitoring the dynamics of response might improve prediction of recurrent CMV reactivation. While the current focus on high-risk (D-/R+) HSCT patients for the prediction of recurrent CMV reactivation is of major clinical relevance, it constitutes the main limitation of this study. A recent study in intermediate-risk (D+/R+) HSCT patients partly addressed this question by measuring the dynamic changes in CMV-CMI using TTrack® CMV,52 underscoring the usefulness of multiple measurements for risk stratification of HSCT recipients. It should also be emphasized that despite the use of nonstandardized conditioning regimens, viral load testing and treatment protocols in this multicenter study, the IFN-γ ELISpot assay performed very well at predicting recurrent CMV reactivation. This further highlights the applicability of using a sensitive and standardized CMV-CMI monitoring assay in a real-world setting, as well as its ease of implementation in clinical routine, for the risk stratification of HSCT patients. In conclusion, our results suggest that using a standardized CMV-CMI monitoring assay can support clinicians in the identification and management of patients with increased risk of recurrent CMV reactivation following HSCT. Beside its applicability in the preemptive setting, the door is also open to its implementation in novel clinical settings with unmet medical needs, such as steering the duration of antiviral prophylaxis (e.g., using letermovir)53 and monitoring immune reconstitution following adoptive T-cell transfer.54 Disclosures The participating clinical and measurement centers (D.W., EW, DT, CW, DJ, KS-E, JG, SM, MS, GK, MK, IH, ML-T, SK, DH, SK, MV, SG, MD, TG, MK, TH, ML) received research funding from Lophius Biosciences for this study. LD is

References 1. Ljungman P, Hakki M, Boeckh M. Cytomegalovirus in hematopoietic stem cell transplant recipients. Hematol Oncol Clin North Am. 2011;25(1):151-169. 2. Azevedo LS, Pierrotti LC, Abdala E, et al. Cytomegalovirus infection in transplant

372

an employee, co-founder, Chief Scientific Officer and shareholder of Lophius Biosciences. SB is an employee and shareholder of Lophius Biosciences. TS, H.G. and A.R. are employees of Lophius Biosciences. RW is Chairman of the Board and a shareholder of Lophius Biosciences. Issued patents relevant to the TTrack® CMV assay used in this study: WO/2010/115984 “METHOD FOR POLYPEPTIDE TRANSFER INTO CELLS”. WO/2003/080792 “USE OF UREA-ADJUVATED POLYPEPTIDES FOR DIAGNOSIS, PROPHYLAXIS AND TREATMENT”. WO/2003/046212 “METHOD FOR IDENTIFYING TARGET EPITOPES OF THE T CELL MEDIATED IMMUNE RESPONSE AND FOR ASSAYING EPITOPE-SPECIFIC T CELLS" Contributions LD, DW, SB and RW contributed to the study conception and design; TS, LD, SB and RW organized, coordinated and supervised the study and its logistics; DW, EW, DT, CW, DJ, KS-E, JG, SM, MS, GK, MK, IH, ML-T, SK, DH, SK, MV, SG, MD and TG acquired patient samples and contributed to data collection; EW, TH, M.K., M.L. and S.B. performed the ELISpot assays; SB and HG analyzed the ELISpot results; AR, TS, LD and RW conducted and supervised the interpretation of results; AR drafted the manuscript and figures; All authors contributed to, reviewed and approved the manuscript. Acknowledgments This study was supported in part by the German Federal Ministry of Education and Research (BMBF grant number 031A215) to Lophius Biosciences and by research funding from Lophius Biosciences to DW, EW, DT, CW, DJ, KS-E, JG, SM, MS, GK, MK, IH, ML-T, SK, DH, SK, MV, SG, MD, TG, MK, TH, and ML. DW was supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation, grant number 324392634 - TRR 221). We are grateful to all study participants for their contribution. The authors acknowledge the members of the respective participating centers, as well as Anja Svenson, Julia Batzilla, Hanna Bendfeldt, Astrid Starke and Alcedis GmbH (Giessen, Germany) for their expertise and assistance in data collection, data management, and study planning and coordination. We thank Thomas Keller and Stephan Weber (ACOMED Statistik, Leipzig, Germany) for their statistical evaluation. T-Track® CMV tests were provided by Lophius Biosciences GmbH, Regensburg, Germany. Funding This study was supported in part by the German Federal Ministry of Education and Research (BMBF grant number 031A215) to Lophius Biosciences and by research funding from Lophius Biosciences to DW, EW, DT, CW, DJ, KS-E, JG, SM, MS, GK, MK, IH, ML-T, SK, DH, SK, MV, SG, MD, TG, MK, TH, and ML. DW was supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation, grant number 324392634 - TRR 221).

recipients. Clinics. 2015;70(7):515-523. 3. de la Cámara R. CMV in Hematopoietic stem cell transplantation. Mediterr J Hematol Infect Dis. 2016;8(1):e2016031. 4. Ljungman P, de la Camara R, Cordonnier C, et al. Management of CMV, HHV-6, HHV-7 and Kaposi-sarcoma herpesvirus (HHV-8) infections in patients with hematological malignancies and after SCT. Bone Marrow Transplant. 2008;42(4):227-240.

5. Tomblyn M, Chiller T, Einsele H, et al. Guidelines for preventing infectious complications among hematopoietic cell transplant recipients: a global perspective. Preface. Bone Marrow Transplant. 2009;44(8):453455. 6. Emery V, Zuckerman M, Jackson G, et al. Management of cytomegalovirus infection in haemopoietic stem cell transplantation. Br J Haematol. 2013;162(1):25-39.

haematologica | 2021; 106(2)

Risk stratification of CMV recurrence after HSCT 7. Ljungman P, Boeckh M, Hirsch HH, et al. Definitions of cytomegalovirus infection and disease in transplant patients for use in clinical trials. Clin Infect Dis. 2017;64(1):8791. 8. Lilleri D, Gerna G. Strategies to control human cytomegalovirus infection in adult hematopoietic stem cell transplant recipients. Immunotherapy. 2016;8(9):1135-1149. 9. Ljungman P, de la Camara R, Robin C, et al. Guidelines for the management of cytomegalovirus infection in patients with haematological malignancies and after stem cell transplantation from the 2017 European Conference on Infections in Leukaemia (ECIL 7). Lancet Infect Dis. 2019;19(8):e260e272. 10. Krause H, Hebart H, Jahn G, Müller CA, Einsele H. Screening for CMV-specific T cell proliferation to identify patients at risk of developing late onset CMV disease. Bone Marrow Transplant. 1997;19(11):1111-1116. 11. Boeckh M, Leisenring W, Riddell SR, et al. Late cytomegalovirus disease and mortality in recipients of allogeneic hematopoietic stem cell transplants: importance of viral load and T-cell immunity. Blood. 2003;101(2):407-414. 12. Lamba R, Carrum G, Myers GD, et al. Cytomegalovirus (CMV) infections and CMV-specific cellular immune reconstitution following reduced intensity conditioning allogeneic stem cell transplantation with alemtuzumab. Bone Marrow Transplant. 2005;36(9):797-802. 13. Ozdemir E, Saliba RM, Champlin RE, et al. Risk factors associated with late cytomegalovirus reactivation after allogeneic stem cell transplantation for hematological malignancies. Bone Marrow Transplant. 2007;40(2):125-136. 14. Yong MK, Lewin SR, Manuel O. Immune monitoring for CMV in transplantation. Curr Infect Dis Rep. 2018;20(4):4. 15. Ciáurriz M, Zabalza A, Beloki L, et al. The immune response to cytomegalovirus in allogeneic hematopoietic stem cell transplant recipients. Cell Mol Life Sci. 2015;72(21):4049-4062. 16. Avetisyan G, Aschan J, Hägglund H, Ringdén O, Ljungman P. Evaluation of intervention strategy based on CMV-specific immune responses after allogeneic SCT. Bone Marrow Transplant. 2007;40(9):865869. 17. Moins-Teisserenc H, Busson M, Scieux C, et al. Patterns of cytomegalovirus reactivation are associated with distinct evolutive profiles of immune reconstitution after allogeneic hematopoietic stem cell transplantation. J Infect Dis. 2008;198(6):818-826. 18. Gratama JW, Brooimans RA, van der Holt B, et al. Monitoring cytomegalovirus IE-1 and pp65-specific CD4+ and CD8+ T-cell responses after allogeneic stem cell transplantation may identify patients at risk for recurrent CMV reactivations. Cytometry B Clin Cytom. 2008;74(4):211-220. 19. Gratama JW, Boeckh M, Nakamura R, et al. Immune monitoring with iTAg MHC tetramers for prediction of recurrent or persistent cytomegalovirus infection or disease in allogeneic hematopoietic stem cell transplant recipients: a prospective multicenter study. Blood. 2010;116(10):1655-1662. 20. Fleming T, Dunne J, Crowley B. Ex vivo monitoring of human cytomegalovirus-specific CD8(+) T-cell responses using the QuantiFERON-CMV assay in allogeneic hematopoietic stem cell transplant recipients attending an Irish hospital. J Med Virol. 2010;82(3):433-440.

haematologica | 2021; 106(2)

21. Borchers S, Luther S, Lips U, et al. Tetramer monitoring to assess risk factors for recurrent cytomegalovirus reactivation and reconstitution of antiviral immunity post allogeneic hematopoietic stem cell transplantation. Transpl Infect Dis. 2011;13(3): 222-236. 22. Borchers S, Bremm M, Lehrnbecher T, et al. Sequential anti-cytomegalovirus response monitoring may allow prediction of cytomegalovirus reactivation after allogeneic stem cell transplantation. PloS One. 2012;7(12):e50248. 23. Tey S-K, Kennedy GA, Cromer D, et al. Clinical assessment of anti-viral CD8+ T cell immune monitoring using QuantiFERONCMV® assay to identify high risk allogeneic hematopoietic stem cell transplant patients with CMV infection complications. PloS One. 2013;8(10):e74744. 24. Lee SM, Kim YJ, Yoo KH, et al. Clinical usefulness of monitoring cytomegalovirus-specific immunity by Quantiferon-CMV in pediatric allogeneic hematopoietic stem cell transplantation recipients. Ann Lab Med. 2017;37(3):277-281. 25. Yong MK, Cameron PU, Slavin M, et al. Identifying cytomegalovirus complications using the quantiferon-CMV assay after allogeneic hematopoietic stem cell transplantation. J Infect Dis. 2017;215(11): 1684-1694. 26. Barabas S, Spindler T, Kiener R, et al. An optimized IFN-γ ELISpot assay for the sensitive and standardized monitoring of CMV protein-reactive effector cells of cell-mediated immunity. BMC Immunol. 2017;18(1):14. 27. Banas B, Böger CA, Lückhoff G, et al. Validation of T-Track® CMV to assess the functionality of cytomegalovirus-reactive cell-mediated immunity in hemodialysis patients. BMC Immunol. 2017;18(1):15. 28. Barabas S, Gary R, Bauer T, et al. Urea-mediated cross-presentation of soluble EpsteinBarr virus BZLF1 protein. PLoS Pathog. 2008;4(11):e1000198. 29. Nesher L, Shah DP, Ariza-Heredia EJ, et al. Utility of the enzyme-linked immunospot interferon-γ-release assay to predict the risk of cytomegalovirus infection in hematopoietic cell transplant recipients. J Infect Dis. 2016;213(11):1701-1707. 30. Banas B, Steubl D, Renders L, et al. Clinical validation of a novel ELISpot-based in vitro diagnostic assay to monitor CMV-specific cell-mediated immunity in kidney transplant recipients: a multicenter, longitudinal, prospective, observational study. Transpl Int. 2018;31(4):436-450. 31. Ljungman P, Perez-Bercoff L, Jonsson J, et al. Risk factors for the development of cytomegalovirus disease after allogeneic stem cell transplantation. Haematologica. 2006;91(1):78-83. 32. Hakki M, Riddell SR, Storek J, et al. Immune reconstitution to cytomegalovirus after allogeneic hematopoietic stem cell transplantation: impact of host factors, drug therapy, and subclinical reactivation. Blood. 2003;102(8):3060-3067. 33. Bae S, Jung J, Kim S-M, et al. The detailed kinetics of cytomegalovirus-specific T cell responses after hematopoietic stem cell transplantation: 1 Year Follow-up Data. Immune Netw. 2018;18(2):e2. 34. Meesing A, Razonable RR. Absolute lymphocyte count thresholds: a simple, readily available tool to predict the risk of cytomegalovirus infection after transplantation. Open Forum Infect. Dis. 2018;5 (10):ofy230. 35. Watanabe M, Kanda J, Hishizawa M, et al. Lymphocyte area under the curve as a pre-

dictive factor for viral infection after allogenic hematopoietic stem cell transplantation. Biol Blood Marrow Transplant. 2019;25(3):587-593. 36. Gutiérrez A, Muñoz I, Solano C, et al. Reconstitution of lymphocyte populations and cytomegalovirus viremia or disease after allogeneic peripheral blood stem cell transplantation. J Med Virol. 2003;70(3):399-403. 37. Rénard C, Barlogis V, Mialou V, et al. Lymphocyte subset reconstitution after unrelated cord blood or bone marrow transplantation in children. Br J Haematol. 2011;152(3):322-330. 38. Chakrabarti S, Mackinnon S, Chopra R, et al. High incidence of cytomegalovirus infection after nonmyeloablative stem cell transplantation: potential role of Campath-1H in delaying immune reconstitution. Blood. 2002;99(12):4357-4363. 39. Hiwarkar P, Gajdosova E, Qasim W, et al. Frequent occurrence of cytomegalovirus retinitis during immune reconstitution warrants regular ophthalmic screening in highrisk pediatric allogeneic hematopoietic stem cell transplant recipients. Clin Infect Dis. 2014;58(12):1700-1706. 40. Ogonek J, Kralj Juric M, Ghimire S, et al. Immune reconstitution after allogeneic hematopoietic stem cell transplantation. Front Immunol. 2016;7:507. 41. Stemberger C, Graef P, Odendahl M, et al. Lowest numbers of primary CD8(+) T cells can reconstitute protective immunity upon adoptive immunotherapy. Blood. 2014;124 (4):628-637. 42. Gabanti E, Lilleri D, Ripamonti F, et al. Reconstitution of human cytomegalovirusspecific CD4+ T cells is critical for control of virus reactivation in hematopoietic stem cell transplant recipients but does not prevent organ infection. Biol Blood Marrow Transplant. 2015;21(12):2192-2202. 43. Shams El-Din AA, El-Desoukey NA, Amin Tawadrous DG, et al. The potential association of CMV-specific CD8+ T lymphocyte reconstitution with the risk of CMV reactivation and persistency in post allogeneic stem cell transplant patients. Hematol Amst Neth. 2018;23(8):463-469. 44. Barron MA, Gao D, Springer KL, et al. Relationship of reconstituted adaptive and innate cytomegalovirus (CMV)-specific immune responses with CMV viremia in hematopoietic stem cell transplant recipients. Clin Infect Dis. 2009;49(12):1777-1783. 45. Drylewicz J, Schellens IMM, Gaiser R, et al. Rapid reconstitution of CD4 T cells and NK cells protects against CMV-reactivation after allogeneic stem cell transplantation. J Transl Med. 2016;14(1):230. 46. Foster AE, Gottlieb DJ, Sartor M, Hertzberg MS, Bradstock KF. Cytomegalovirus-specific CD4+ and CD8+ T-cells follow a similar reconstitution pattern after allogeneic stem cell transplantation. Biol Blood Marrow Transplant. 2002;8(9):501-511. 47. Pourgheysari B, Piper KP, McLarnon A, et al. Early reconstitution of effector memory CD4+ CMV-specific T cells protects against CMV reactivation following allogeneic SCT. Bone Marrow Transplant. 2009;43(11):853861. 48. Lilleri D, Gerna G, Fornara C, et al. Prospective simultaneous quantification of human cytomegalovirus-specific CD4+ and CD8+ T-cell reconstitution in young recipients of allogeneic hematopoietic stem cell transplants. Blood. 2006;108(4):1406-1412. 49. Widmann T, Sester U, Schmidt T, et al. Rapid reconstitution of CMV-specific T-cells after stem-cell transplantation. Eur J

373

E. Wagner-Drouet et al. Haematol. 2018;101(1):38-47. 50. Lilleri D, Fornara C, Chiesa A, et al. Human cytomegalovirus-specific CD4+ and CD8+ T-cell reconstitution in adult allogeneic hematopoietic stem cell transplant recipients and immune control of viral infection. Haematologica. 2008;93(2):248-256. 51. Ozdemir E, St John LS, Gillespie G, et al. Cytomegalovirus reactivation following allogeneic stem cell transplantation is associated with the presence of dysfunctional anti-

374

gen-specific CD8+ T cells. Blood. 2002;100(10):3690-3697. 52. Jung J, Lee H-J, Kim S-M, et al. Diagnostic usefulness of dynamic changes of CMV-specific T-cell responses in predicting CMV infections in HCT recipients. J Clin Virol. 2017;87:5-11. 53. Marty FM, Ljungman P, Chemaly RF, et al. Letermovir prophylaxis for cytomegalovirus in hematopoietic-cell transplantation. N Engl J Med. 2017;377 (25):2433-2444.

54. Lindemann M, Eiz-Vesper B, Steckel NK, et al. Adoptive transfer of cellular immunity against cytomegalovirus by virus-specific lymphocytes from a third-party family donor. Bone Marrow Transplant. 2018;53 (10):1351-1355. 55. Chemaly RF, Chou S, Einsele H, et al. Definitions of resistant and refractory cytomegalovirus infection and disease in transplant recipients for use in clinical trials. Clin Infect Dis. 2019;68(8):1420-1426.

haematologica | 2021; 106(2)

ARTICLE

Transplantation

Joerg Henes,1 Maria Carolina Oliveira,2 Myriam Labopin,3 Manuela Badoglio,4 Hans Ulrich Scherer,5 Nicoletta Del Papa,6 Thomas Daikeler,7 Marc Schmalzing,8 Roland Schroers,9 Thierry Martin,10 Gregory Pugnet,11 Belinda Simoes,12 David Michonneau,13 Erik W.A. Marijt,14 Bruno Lioure,15 Jacques Olivier Bay,16 John A Snowden,17 Montserrat Rovira,18 Anne Huynh,19 Francesco Onida,20 Lothar Kanz,21 Zora Marjanovic22 and Dominique Farge23 for the NISSC1 members

Center for Interdisciplinary Clinical Immunology, Rheumatology and Auto-inflammatory Diseases and Department of Internal Medicine II (Oncology, Hematology, Immunology and Rheumatology), University Hospital, Tübingen, Germany; 2Ribeirão Preto Medical School, University of São Paulo, Ribeirão Preto, Brazil; 3EBMT Paris Study Office, Saint Antoine Hospital, Université Pierre et Marie Curie, INSERM UMR 938, Department of Haematology, Paris, France; 4EBMT Paris Study Office, Hôpital St Antoine, INSERM UMR 938, Paris, France; 5Leiden University Medical Center, Department of Rheumatology, Leiden, the Netherlands; 6Scleroderma Clinic, Ospedale Pini, Department of Rheumatology, Milan, Italy; 7University and University Hospital of Basel, Department of Rheumatology, Basel, Switzerland; 8University Hospital of Würzburg, Department of Rheumatology/Immunology, Würzburg, Germany 9Universitäts Klinik of Bochum, Med. Klinik, Bochum, Germany; 10Hôpitaux Universitaires de Strasbourg, Service de Médecine Interne et Immunologie Clinique, Centre de Référence des Maladies Auto-Immunes Systémiques Rares Est-Sud-Ouest (RESO), Strasbourg, France; 11CHU de Toulouse, Hôpital Purpan, Service de Médecine Interne, Toulouse, France; 12Ribeirão Preto Medical School, University of São Paulo, Division of Hematology, Ribeirão Preto, Brazil; 13 Hôpital Saint Louis & Université Paris 7, Denis Diderot, Department of Hematology, Paris, France; 14Leiden University Medical Center, Department of Hematology, Leiden, the Netherlands; 15Strasbourg University Hospital, Department of Hematology, Strasbourg, France; 16CHU de Clermont Ferrand, Department of Hematology, Clermont Ferrand, France; 17Department of Haematology, Sheffield Teaching Hospitals NHS Foundation Trust, Sheffield, UK; 18Hospital Clínic of Barcelona, Department of Hematology, Barcelona, Spain; 19UCT Oncopole, Department of Hematology, Toulouse, France; 20Fondazione IRCCS Ca Granda Ospedale Maggiore Policlinico, University of Milan, Milan, Italy; 21University Hospital Tübingen, Department of Internal Medicine II (Oncology, Haematology, Immunology and Rheumatology), Tübingen, Germany 22Saint Antoine Hospital, Department of Hematology, Paris, France; 23Hôpital Saint-Louis, Assistance Publique des Hôpitaux de Paris, Centre de Référence des Maladies AutoImmunes Systémiques Rares d’Ile-de-France, Filière ‘FAI2R’, Unité de Médecine Interne: Maladies Auto-immunes et Pathologie Vasculaire (UF 04), Université de Paris, EA 3518, Paris, France and 24Department of Medicine, McGill University, Montreal, Quebec, Canada

Ferrata Storti Foundation

Haematologica 2021 Volume 106(2):375-383

ABSTRACT

hree randomized controlled trials in early severe systemic sclerosis demonstrated that autologous hematopoietic stem cell transplantation was superior to standard cyclophosphamide therapy. This European Society for Blood and Marrow Transplantation multicenter, prospective, non-interventional study was designed to further decipher efficacy and safety of this procedure for severe systemic sclerosis patients in real-life practice and to search for prognostic factors. All consecutive adult patients with systemic sclerosis undergoing a first autologous hematopoietic stem cell transplant between December 2012 and February 2016 were prospectively included in the study. The primary endpoint was progression-free survival. Secondary endpoints were overall survival, non-relapse mortality, response and incidence of progression. Eighty patients with systemic sclerosis were included. The median durahaematologica | 2021; 106(2)

Correspondence: DOMINIQUE FARGE dominique.farge-bancel@aphp.fr; dominique.farge@mcgill.ca Received: June 22, 2019. Accepted: January 9, 2020. Pre-published: January 16, 2020. https://doi.org/10.3324/haematol.2019.230128

375

J, Henes et al.

tion of the follow-up was 24 (range, 6-57) months after stem cell transplantation using cyclophosphamide plus antithymocyte globulin conditioning for all, with CD34+ selection in 35 patients. At 2 years, the progression-free survival rate was 81.8%, the overall survival rate was 90%, the response rate was 88.7% and the incidence of progression was 11.9%. The 100-day non-relapse mortality rate was 6.25% (n=5) with four deaths from cardiac events, including three due to cyclophosphamide toxicity. Modified Rodnan skin score and forced vital capacity improved with time (P<0.001). By multivariate analysis, baseline skin score >24 and older age at transplantation were associated with lower progression-free survival (hazard ratios 3.32 and 1.77, respectively). CD34+-cell selection was associated with better response (hazard ratio 0.46). This study confirms the efficacy of autologous stem cell transplantation, using nonmyeloablative conditioning, in real-life practice for severe systemic sclerosis. Careful cardio-pulmonary assessment to identify organ involvement at the time of the patient’s referral, reduced cyclophosphamide doses and CD34+-cell selection may improve outcomes. The study was registered at ClinicalTrials.gov: NCT02516124.

Introduction Systemic sclerosis (SSc) is a rare systemic autoimmune connective tissue disease characterized by vasculopathy, immune dysregulation and progressive fibrosis, affecting primarily the skin, gastrointestinal tract, lungs, heart and kidneys.1 It is associated with high disease-related mortality2 and the main causes of death are related to cardiac, pulmonary and renal involvement.2-5 Mortality rates in patients with SSc have remained higher than those in the general population (standardized mortality ratio above 3.5) over the past decades, in contrast to reductions in mortality for other rheumatic diseases or cancer.6 After the diagnosis of non-Raynaud phenomenon, the following risk factors for higher mortality related to progressive disease have been identified:2 diffuse skin fibrosis measured by the modified Rodnan skin score (mRSS, range 0-51),7 elevated C-reactive protein levels, altered left or right ventricular ejection fraction, interstitial lung disease with reduced forced vital capacity (FVC) or diffusion capacity for carbon monoxide (DLCO) on lung function testing, proteinuria and male sex. Earlier studies had found that the presence of anti-topoisomerase II (Scl-70) auto-antibodies, reduced functional status - as measured by the Scleroderma Health Assessment Questionnaire (sHAQ, range from 0 to 3, with lower values indicating a better quality of life)8 - or pulmonary arterial hypertension adversely affect survival.3-5 Treatment with biologic or immunosuppressive drugs did not improve SSc survival.2,9,10 Following early phase I-II trials,11-13 three randomized controlled trials, namely ASSIST (American scleroderma stem cell versus immune suppression trial, phase II),14 ASTIS (Autologous stem cell transplantation international scleroderma trial, phase III)15 and SCOT (Scleroderma: cyclophosphamide or transplantation, phase II)16 demonstrated the superiority of autologous hematopoietic stem cell transplantation (HSCT) compared to the standard cyclophosphamide pulse treatment with regards to overall and event-free survival. In this context, autologous HSCT has become the best treatment option for patients with severe or rapidly progressive SSc, with a grade 1 level of evidence since 2012 according to the Autoimmune Disease Working Party (ADWP) of the European Society for Blood and Marrow Transplantation (EBMT) recommendations.17,18 After two decades of progress in HSCT for autoimmune diseases, the use of autologous HSCT for severe or rapidly progressive SSc has also been endorsed by the European League Against Rheumatism (EULAR)10 376

and the American Society for Blood and Marrow Transplantation10,17,19 and the demand for this therapeutic procedure is the fastest growing indication for HSCT in Europe.10,17,19,20 We therefore designed this ADWP-EBMT non-interventional study, NISSC1, to further evaluate the efficacy and safety of autologous HSCT for early, severe, progressive SSc, when performed in real-life practice, and to identify risk factors for early adverse events and transplant-related mortality, in order to make this approach safer.

Methods Study design This study is a prospective, open, multicenter, non-interventional study. Data regarding patients included in this study were prospectively collected using a specific case report form, designed according to EBMT guidelines for HSCT in SSc.17 All EBMT participating centers agreed to report their consecutive patients aged 18 to 65 years with severe progressive SSc undergoing their first autologous HSCT between December 2012 and February 2016. Inclusion and exclusion criteria are detailed in the Online Supplementary Appendix. Participating centers followed their own local protocols for the transplant procedure as wells as the patients’ evaluation at baseline and during the 2-year follow-up. The study was approved by the EBMT board in 2012 and by local ethics committees in participating countries and registered at ClinicalTrials.gov (NCT02516124). All patients gave written informed consent to their participation in the study.

Patients’ evaluation The SSc patients’ characteristics were assessed at baseline (within 3 months before autologous HSCT) and at 3, 6, 12, 18 and 24 months after transplantation (day 0 was defined as the time point of reinfusion of stem cells). Data collected included mRSS, erythrocyte sedimentation rate (in mm/h), creatine kinase levels, presence or absence of autoantibodies, left ventricular ejection fraction (LVEF in %) and systolic pulmonary arterial pressure (in mmHg) on echocardiography, resting and 24 h Holter-electrocardiograms, abnormal cardiac magnetic resonance imaging (MRI) or pulmonary arterial hypertension on right heart catheterization (RHC), FVC and DLCO (% of predicted values), high resolution computed tomography of the lungs and kidney function as well as immunosuppressive drugs given before transplantation. The sHAQ was used to evaluate patients’ quality of life. haematologica | 2021; 106(2)

ASCT for progressive SSc

Autologous hematopoietic stem cell transplantation procedure Data collected included details on the mobilization process with or without CD34+-cell selection, type of conditioning regimen; total dose of CD34+ cells infused; number of days of hospitalization for mobilization and conditioning, use and duration of granulocyte colony-stimulating factor (G-CSF), and time to engraftment.

Endpoints The primary endpoint was progression-free survival (PFS), defined as survival after autologous HSCT without death or evidence of progression of SSc (see Online Supplementary Appendix S1). The secondary endpoints were: (i) infectious or non-infectious adverse events during the first 100 days after autologous HSCT; (ii) engraftment, defined as 3 consecutive days of neutrophils >0.5x109/L and platelets >20x109/L; (iii) response to treatment, defined as a >25% improvement in mRSS and/or ≥10% improvement in FVC or DLCO as compared to baseline and without need of further immunosuppression; (iv) incidence of progression; (v) non-relapse mortality (NRM), defined as any death without progression; and (vi) overall survival.

Statistical analysis Cumulative incidence functions were used to estimate response

Table 1. Baseline characteristics of the patients (n=80).

Parameter

N (%) or median (range)

Patients Age at aHSCT, years Sex, female Disease duration from SSc diagnosis, months Body mass index (BMI) BMI <18·5 BMI 18·5-30 BMI >30 Skin involvement Modified Rodnan skin score (mRSS) Lung involvement Lung crepitations Pulmonary function tests, done FVC, % of predicted value FEV1, % of predicted value DLCO, % of predicted value Chest imaging, done Chest X-rays, abnormal Thoracic computed tomography (HRCT), abnormal* Cardiac involvement Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Heart rate Resting ECG, abnormal 24h Holter ECG, abnormal Echocardiography, done LVEF (in %) Pericardial effusion, present sPAP by cardiac echo (mmHg) Cardiac MRI, done Cardiac MRI, abnormal**

80 43 (20-62) 57 (71.3) 23.8 (5.3-103.7)

haematologica | 2021; 106(2)

8 (10.3) 65 (83.3) 5 (6.4) 24 (2 - 49) 41 (51.3) 80 (100) 72 (43-132) 78 (52-153.3) 59 (34.3-120) 80 (100) 31/71 (43.7) 66/77 (85.7) 110 (76-145) 70 (40-90) 82 (50-105) 4/80 (5) 12/60 (20) 80 (100) 65 (47-84) 5/79 (6.3) 29 (8-59) 59/80 (74) 11 (18.6)

to treatment, NRM, and progression to accommodate for competing risks. Probabilities of overall survival and PFS were calculated using the Kaplan-Meier method. Univariate analyses were performed using the Gray test for cumulative incidence and the log rank test for overall survival and PFS. Multivariate analysis was performed using a Cox regression model including variables associated with the outcome with a P-value less than 0.10 in univariate analyses. Age was included as a continuous variable. A generalized linear model was used to analyze evolution over time of mRSS, FVC, DLCO and sHAQ. Results are expressed as a hazard ratio (HR) with 95% confidence interval (95% CI).

Results Eighty patients (71.3% female), who had undergone a first autologous HSCT between December 2012 and February 2016 in 13 EBMT centers from six European countries and Brazil were included in this study. The centers are listed in Online Supplementary Appendix S2. The median (range) follow-up after transplantation was 24 (6-57) months. Table 1 summarizes the patients’ baseline characteristics. The median disease duration since the diagnosis of SSc was 23.8 (5.3103.7) months. At baseline, all 80 patients had skin involvement, and mRSS was above 15 in 64 (80%) of them. Abnormal chest

Right heart catheterization, done PAP (mmHg) PAP, systolic PAP, diastolic Fluid challenge, done Musculo-skeletal involvement Joints involvement Muscle weakness CK levels, U/L Patients with CK level ≥ 250 U/mL Kidney function Creatinine clearance, mL/min*** Proteinuria on labstix Haematuria on labstix SSc auto-antibodies Anti-nuclear antibody positive Anti-topoisomerase I (Scl70) positive Anti-centromere positive Previous immunosuppressive SSc medication, yes**** Cyclophosphamide (i.v./oral) Dose, g Methotrexate Mycophenolate mofetil Prednisone or equivalent Azathioprine ATG rabbit Cyclosporine

17/80 (21.3) 17 (8-23) 26 (16-33) 10 (6-16) 7 (41) 42 (52.5) 22 (27.5) 80 (16-1368) 9 (12.5) 111 (58-190) 9 (12) 7 (9.2) 74 (92.7) 57 (71.3) 3 (4.2) 76 (95.0) 48 (60) 6 (1-17) 46 (58.2) 16 (20.3) 64 (81.0) 8 (10.0) 2 (2.5) 3 (3.8)

If not otherwise stated all data are n (%) or median (range) *Abnormal computed tomography scan includes fibrosis, honeycombing or ground-glass pattern. **Abnormal cardiac magnetic resonance imaging includes reduced left or right ventricular function or late enhancement. ***The Cockcroft-Gault formula was recommended for estimating creatinine clearance, i.e., creatine clearance = (140 – age in years)/serum creatinine in mg/dL) x body weight in kg/72. ****Previous treatment ever before, multiple answers possible. aHSCT: autologous hematopoietic stem cell transplantation; SSc: systemic sclerosis; BMI: body mass index; FVC: forced vital capacity; FEV1: forced experiratory volume in 1 s; DLCO: diffusing capacity of lung for carbon monoxide; HRCT: high-resolution computed tomography; ECG: electrocardiography; LVEF: left ventricular ejection fraction; sPAP: systolic pulmonary artery pressure measured by echo; echo: echocardiography; MRI: magnetic resonance imaging; PAP: pulmonary artery pressure; CK: creatine kinase; SSc: systemic sclerosis; ATG: antithymocyte globulins;

377

J, Henes et al.

x-rays or abnormal high-resolution computed tomography was reported in 70 (87%) patients. Heart involvement with an abnormal 24 h electrocardiogram was reported in 12/60 (20%) or with abnormal cardiac MRI in 11/59 (19%) patients. Among the 11 patients with abnormal MRI at baseline, echocardiography showed normal LVEF in all cases and pericardial effusion in one patient, the resting ECG was abnormal in one patient and the 24 h ECG was abnormal in two patients; baseline RHC performed in six patients (with fluid challenge in 2 cases) was normal. Overall, 21 patients (26%) were transplanted without any prior cardiac MRI or RHC at baseline; among these, two had pericardial effusion and seven had a systolic pulmonary artery pressure above 25 mmHg on echocardiography, one had ventricular disturbances on resting ECG (n=20), and 2/8 had an abnormal 24 h ECG.

Mobilization and conditioning procedures All patients were mobilized using cyclophosphamide with a median dose of 2 g/m2 (range, 1-4 g/m2) plus G-CSF. The median duration of hospitalization was 4 (1-30) days for mobilization with large variation according to centers. CD34+-cell selection was performed in 35 (44%) patients (Table 2). The median time from mobilization to the start of the conditioning regimen for autologous HSCT was 52 (22-254) days. The conditioning regimen included cyclophosphamide in all 80 patients, alone in 76 (95%) patients, with a median total dose of 200 (50-240) mg/kg. Four patients (5%) received thiotepa 10 mg/kg with a total cyclophosphamide dose of 100 mg/kg as a single-center treatment for patients with cardiac abnormalities at the pre-transplant evaluation. Additional rabbit antithymocyte globulin (ATG), with different doses according to the brand, was also included in the conditioning regimen for all patients and administered over 4 to 5 days. Sixty-seven patients (82%) received methylprednisolone during the ATG treatment. No patient received total body irradiation. A median number of 5.2 x 106/kg (2.53-23.31) CD34+ cells were reinfused at transplantation. Following autologous HSCT, G-CSF was administered to 34 (43%) patients for a median duration of 5 (1-13) days. Following autologous HSCT, the median time to neutrophil engraftment was 11 (8-24) days and the median time to platelet engraftment was 9 (1-25) days. In 13 (16.9%) patients, the platelet counts were never below 20x109/L. There was an inverse correlation between the number of CD34+ cells infused and the time to engraftment (Spearman: r2= -0.31; P<0.01), while the number of infused CD34+ cells and addition of G-CSF until recovery of aplasia were not associated with a reduced number of infectious complications (P=0.32). The median duration of inpatient hospitalization for conditioning, autologous peripheral blood stem cell re-infusion and subsequent supportive care until engraftment and discharge was 24 (8-128) days (Table 2).

Efficacy At 1 and 2 years after autologous HSCT, the PFS rate was 87.5% (95% CI: 80.2-94.7) and 81.8% (95% CI: 73.1-90.5), respectively, and the rate of response to treatment was 75% (95% CI: 63.7-83.2) and 88.7% (95% CI: 79-94.1), respectively. At the 1- and 2-year time-points, the incidence of progression was 6.3% (95% CI: 2.3-13.1) and 11.9% (95% CI: 5.8-20.5), respectively, and the overall survival rate was 91.2% (95% CI: 85.1-97.4) and 90% (95% CI 83.4 - 96.6) at 1 and 2 years, respectively (Figure 1). Three of the four 378

Table 2. Treatment regimen used for mobilization and conditioning.

Parameter

N (%) or median (range)

Mobilization CYC 2 g/m² CYC 4 g/m² Other dose Missing = 2 G-CSF, yes Leukapheresis 1 2 3 CD34+ selection Yes None Conditioning regimen CYC 200 mg/kg CYC other dose mg/kg CYC 100 mg/kg + Thiotepa 10 mg/kg Rabbit ATG, yes Thymoglobulin (Sanofi/Genzyme), mg/kg Grafalon (Neovi/Fresenius), mg/kg Growth factors G-CSF administration after conditioning N. of days of administration Cells infused CD34+ cells reinfused, 106/kg Duration of hospitalization Mobilization, days Conditioning, days

45 (57.7%) 23 (29.5%) 10 (12.8%) 80 (100%) 61 (77.2%) 11 (13.9%) 7 (8.9%) 35 (43.8%) 45 (56.3%) 72 (90.0%) 4 (5.0%) 4 (5.0%) 80 (100%) 7.5 (2.5-7.5) 40 (30-41) 34 (43.0%) 5 (1-13) 5.2 (2.53-23.31) 4 (1-30) 24 (7-128)

CYC: cyclophosphamide; G-CSF: granulocyte – colony-stimulating factor; ATG: antithymocyte globulin;

patients transplanted with thiotepa and a reduced-dose cyclophosphamide regimen responded; one with worsening skin and lung involvement at day 350 was then treated with prednisone, mycophenolate mofetil and rituximab. By multivariate analysis, mRSS >24 at baseline and older age at transplantation were significantly associated with a lower PFS with hazard ratios of 3.33 (95% CI: 1.04-10.62) and 1.77 (95% CI: 1.07-2.94), respectively. CD34+-cell selection was associated with a better response to therapy (HR: 0.46; 95% CI: 0.27-0.76). Figure 2 illustrates the evolution of the parameters over time. The mean mRSS decreased from 23.9 (standard deviation [SD] 9.7) at baseline to 14.2 (SD 9.2) at 1 year and 12.6 (SD 8.3) at 2 years (n=55, P<0.001). Regarding lung function, the mean FVC increased from 73.6% (SD 16.9) of predicted value at baseline to 79.5% (SD 16.9) at 1 year and 80.6% (SD 19.1) at 2 years (n=37, P<0.001); the mean DLCO was 60.2% (SD 19.3) at baseline, 59.7% (SD 17.7) at 1 year and 60.4% (SD 19.1) at 2 years (n=35). The sHAQ score was available in 15 patients and decreased significantly from 0.96 (SD 0.8) at baseline to 0.73 (SD 0.6) at 1 year and 0.70 (SD 0.6) at 2 years (P<0.001). The mean erythrocyte sedimentation rate decreased from 23.2 mm/h (SD 19.8) at baseline to 16.1 mm/h (SD 11.3) at 1 year and 18.7 mm/h (SD 23.6) at 2 years (n=38, P<0.001). Anti-Scl-70 and anti-centromere autoantibodies did not change after autologous HSCT.

Safety The NRM rate at 100 days and 2 years was 6.25% (95% CI 2.3-13) (Figure 1). Four of the five deaths not due to haematologica | 2021; 106(2)

ASCT for progressive SSc

Figure 1. Kaplan-Meier curves of outcomes of patients undergoing autologous hematopoietic stem cell transplantation for severe systemic sclerosis. Overall survival (OS), progression-free survival (PFS), progression and non-relapse (treatment)related mortality (NRM) over time.

Figure 2. Boxplots of the four most important efficacy parameters in the 24 months following autologous hematopoietic stem cell transplantation for severe systemic sclerosis. (A) Improvement of the modified Rodnan skin score (mRSS) (n=55; P<0.001). (B) Improvement of forced vital capacity (FVC) (n=37; P=0.001. (C) Stabilization of diffusion capacity of carbon monoxide (DLCO) (n=35; P=0.01). (D) Improvement of Scleroderma Health Assessment Questionnaire (sHAQ) score (n=15, P=0.001).

haematologica | 2021; 106(2)

379

J, Henes et al. Table 3. Baseline characteristics and results from the pretransplant evaluation in patients who died during the follow-up.

Age at SAP/DBP Resting 24h-Holter Echo Echo Cardiac aHSCT/ (HR) ECG ECG LVEF PASP MRI gender (%) 28/F

90/60 (80)

46/F

Normal

RHC

Proteins Haematuria Serum Mobilization Conditioning Time to Cause of death creatinine CYC regimen death (mmol/L) (g/mÂ˛) (mg/kg) (days)

Normal Not done

Positive

CYC 200

aHSCT related: cardiac toxicity

110/75 (90) Normal Not done

Normal Not done

Negative

CYC 200

aHSCT related: cardiac toxicity and CYC toxicity

55/F

125/75 (65) Normal Abnormal

Normal Not done

Negative

CYC 200

aHSCT related: cardiac toxicity and CYC toxicity

35/F

100/70 (88) Normal Not done

Not done Not done

Negative

CYC 200

aHSCT related: cardiac toxicity and CYC toxicity

52/M

135/85 (72) Normal Abnormal

Abnormal

Mean PAP=17 No fluid challenge

Positive

Negative

CYC 100 + Thiotepa 10

aHSCT related: infection + MOF

58/M

120/80 (82) Normal

Normal Not done

Negative

115

CYC 200

128

Progression: Pulmonary toxicity

43/M

120/80 (100) Normal

Normal

Normal Not done

Positive

CYC 200

131

Progression

51/F

120/70 (70) Normal

Normal

Normal Not done

Negative

CYC 200

414

Progression

Normal

aHSCT: autologous hematopoietic stem cell transplantation; SAP/DAP: systolic arterial pressure/diastolic blood pressure; HR: heart rate; ECG: electrocardiography; Echo: echocardiographic; LVEF: left ventricular ejection fraction; PASP: pulmonary arterial systolic pressure; MRI: magnetic resonance imaging; RHC: right heart catheterization; CYC: cyclophosphamide; M: male; F: female; MOF: multiorgan failure; NA: not applicable.

relapse were related to a cardiac event, due to cyclophosphamide toxicity in three patients. A 35-year old female, who died from cardiac toxicity and multiorgan failure on day +61 after autologous HSCT, had normal resting ECG and LVEF (62%) on echocardiography, and abnormal LVEF on heart scintigraphy (47%); MRI and RHC were not performed at baseline. Three other patients died from cardiac toxicity at day +1 (n=2) and day +9 (n=1), with normal cardiac MRI and no RHC evaluation at baseline. A fifth patient (conditioning with thiotepa + reduced-dose cyclophosphamide) died from cytomegalovirus reactivation with viral copies detectable also in the lung, with bacterial (Pseudomonas) pneumonia and multiorgan failure at day +21. This patient had a normal resting ECG, with an abnormal 24 h ECG, abnormal MRI and a mean pulmonary artery pressure of 17 mmHg on RHC performed without fluid challenge at baseline. During the 24 months of follow-up, three additional patients died, all due to disease progression. Baseline characteristics of all patients who died during the follow-up and the respective causes of death are presented in Table 3. During the 100 days after transplantation, a total of 119 adverse events were reported among the 80 patients as detailed in Table 4. The percentage of infectious complications was not significantly different between the patients who did or did not receive CD34+-selected grafts (P=0.44), or according to the cyclophosphamide dose at the time of mobilization (2 g/m2 vs. 4 g/m2, P=0.16) or depending on whether they were or were not given G-CSF after their transplant (P=0.32) (Online Supplementary Table S3).

Discussion The NISCC1 study is the largest prospective cohort today analyzing the efficacy and safety of autologous 380

HSCT in 80 patients with severe SSc in a real-life setting. The PFS rate was 82% and significant improvements were observed in skin score, lung function and quality of life during the 2 years of follow-up, consistent with results of previous randomized controlled trials14-16 and earlier studies.11-13 Nevertheless, the NRM rate associated with autologous HSCT was 6.25%, with most (4 out of 5) of the deaths being related to early cardiac events; this is similar to the 6% NRM rate reported retrospectively among 89 SSc patients transplanted between 2002 to 2011 in one USA and one Brazilian expert center using a non-myeloablative conditioning regimen.21 We used PFS as the primary endpoint to assess autologous HSCT efficacy rather than event-free survival, as previously used14,15 and which was defined as survival without respiratory, renal, or cardiac failure. PFS, which encompasses disease progression and toxic deaths, better reflects safety and efficacy of HSCT. In NISSC1, the incidence of progression was 6.3% and 11.9% at 1 and 2 years, respectively, highlighting the need to identify patients who may benefit from early maintenance therapy after autologous HSCT. Several mechanisms contribute to the resetting of the immunological response and disease control after autologous HSCT, and vary according to individual patients.12 We showed that the immune reconstitution profile and Tcell repertoire diversity together with a decrease in mRSS >25% and in FVC >10% at 1-2 years after transplantation could identify long-term responders and nonresponders/relapsing patients at 4-5 years.22 The immune reconstitution process and the balance between reinduction of tolerance and sustained autoreactivity after transplantation will depend on the type of conditioning regimen used to ablate the original immune system, the effect of ATG or ex vivo CD34+-cell selection to haematologica | 2021; 106(2)

ASCT for progressive SSc

further deplete/eliminate residual autoreactive cells and the use of post-transplant maintenance immunosuppression.15,16,22-24 All patients in the NISCC1 study received a ‘non-myeloablative’ cyclophosphamide-based conditioning regimen, which included rabbit-derived ATG in all cases. The degree of T-cell depletion for each patient varied according to the type and total dose of ATG as well as timing of its administration in relation to the peripheral blood stem cell infusion, and whether the infused cells had undergone ex vivo CD34+ selection. In a previous retrospective EBMT analysis, we failed to demonstrate a benefit of ex vivo CD34+-cell selection23 in 138 SSc patients with similar baseline characteristics transplanted between 2000 and 2012. Surprisingly, the NISCC1 study showed a superior response to treatment in patients who received a CD34+selected graft, with no increased rate of infectious complications, consistent with the recent findings of Ayano et al. in a retrospective analysis of 19 SSc patients.24 The reason for the difference in outcomes between these studies is unclear, but the population included in the NISCC1 prospective cohort was more homogenous in terms of conditioning, including in vivo T-cell depletion with rabbitderived ATG. All patients in the NISCC1 study received rabbit ATG formulations (although of different brands) while, in the EBMT retrospective study,23 64% of patients treated with ATG received a rabbit-derived product and 35% the horse-derived ATG formulation, which were shown to be associated with significantly different clinical outcomes in aplastic anemia.25 This important finding raises the question of whether a randomized, controlled trial should be performed to clarify whether more profound depletion of autoreactive cells from the peripheral blood stem cell graft by CD34+-cell selection may translate into sustained clinical benefit in SSc patients after autologous HSCT, without adversely affecting safety and cost-effectiveness. During the immune reconstitution process, the re-emergence of naïve T and B cells, the renewal of the immune repertoire and reinstatement of synergistic immunoregulatory mechanisms are expected. Therefore maintenance or rapid reintroduction of immunosuppression after autologous HSCT may improve patients’ outcome. The ongoing open-label trial (ClinicalTrials.gov identifier: NCT01413100) analyzing the addition of mycophenolate mofetil for 2 years after autologous HSCT and the next NISCC2 study (ClinicalTrials.gov identifier: NCT 03444805) analyzing posttransplantation management may clarify the potential benefit of post-transplant maintenance immunosuppression. A critical point for autologous HSCT in autoimmune diseases is safety. The risks of early transplant-related complications and mortality vary according to the type of disease, pre-existing internal organ involvement, the experience of the transplant center and the intensity of the conditioning regimen.12,26,27 Major cardiac adverse events and NRM after autologous HSCT in SSc are predominantly related to primary cardiac and lung involvement, which are underevaluated in the absence of systematic MRI and RHC at baseline.28,29 Our study revealed important variations in clinical practice in the pre-transplant cardio-pulmonary evaluation process. Although all patients who died during the first 100 days had normal resting ECG and LVEF on echocardiography at baseline, they were not all assessed by 24 h ECG (3/5), MRI (4/5) or RHC with fluid haematologica | 2021; 106(2)

challenge (1/5). Overall, 24 h ECG was only used in 75% of the NISCC1 patients to detect subclinical rhythm or conduction disturbances at baseline, although it has been recommended since 2004.27 Similarly, cardiac MRI and RHC with fluid challenge were only performed in 74% and 21% of the patients, respectively, while their use has been advocated among expert centers since 2012 to mini-

Table 4. Complications reported in the 100 days following autologous hematopoietic stem cell transplantation among the 80 patients treated for severe systemic sclerosis.

Post-transplant non-infectious complications in 21 (26%) patients

Events (n=24)

CYC-related acute cardiomyopathy / 4 deaths Cardiomyopathy with myocardial infarction ARDS + acute heart failure Arrhythmias Pericardial effusion Atrial flutter/fibrillation Allergy to ATG with respiratory failure + pulmonary hemorrhagic syndrome ATG-related fever Acute pulmonary edema Renal failure Psychosis/depression Epistaxis Oral mucositis Anal fissure Hemorrhagic cystitis

Post-transplant infections reported in 60 (75%) patients Neutropenic fever with no documented infection Catheter infection Sepsis, including 1 septic shock Sinusitis Pneumonia Endocarditis Bacterial Pneumonia Pleuritis Sepsis Gram-positive bacillus Sepsis Gram-negative bacillus Catheter infection not documented Cellulitis Viral CMV reactivation EBV reactivation Pneumonia Herpes simplex infection BK virus cystitis Hemorrhagic cystitis, adenovirus-positive Fungal Fungal sinus infection Skin infection-right foot--Fusarium sppCandida stomatitis Parasitic Toxoplasma reactivation

4 1 1 2 1 2 1 1 1 3 2 1 1 1 2

Events (n=95) 25 2 2 2 8 1 4 1 4 1 3 3 13 13 3 4 1 1 1 1 1 1

CYC: cyclophosphamide; ARDS: acute respiratory distress syndrome; ATG: antithymocyte globulins; CMV: cytomegalovirus; EBV: Epstein-Barr virus.

381

J, Henes et al.

mize the risk of the transplant procedure.21,30,31 From 2013 to 2017, the EBMT and collaborative partners worked collectively to establish common standards for the use and interpretation of cardiac MRI results and RHC with fluid challenge for a rigorous cardio-pulmonary assessment of SSc patients, carefully defining eligibility and selection criteria for transplant referral.32 Optimized treatment regimens are warranted to improve the safety of autologous HSCT, even in patients without cardiac involvement. Interestingly, the NRM rate in the NISCC1 study was similar to the 6% reported among 32 SSc patients transplanted between 2005 and 2011, using a ‘myeloablative’ conditioning regimen comprising total body irradiation (8 Gy), cyclophosphamide (120 mg/kg) and horse-derived ATG in the prospective SCOT randomized controlled trial.16 However, in the SCOT trial, NRM was primarily related to late fatal myelodysplastic syndromes, while no cyclophosphamide toxicity was reported. Within the follow-up period of the NISSC1 study, there were no reports of myelodysplastic syndrome or other myeloid malignancies. In addition, there were no reports of serious late infectious complications or secondary autoimmune phenomena that may occasionally present as long-term complications following autologous HSCT in autoimmune diseases. Although further follow-up is required, it is possible that there are long-term advantages, in terms of late effects, of the ‘non-myeloablative’ regimen used in the NISCC1 study. Early toxicity remains the principal challenge, including a NRM rate of 6%. To reduce toxicity and thus treatmentrelated mortality of the conditioning, a more individualized regimen with respect to heart, lung or kidney involvement might further improve patients’ safety, without excluding patients with organ manifestations. Two studies addressing patients with heart involvement are ongoing, both using a conditioning regimen with a reduced dose of cyclophosphamide and the addition of either thiotepa, as in four patients in our NISCC1 cohort (NCT01895244), or rituximab and fludarabine followed by reinfusion of autologous hematopoietic stem cells (NCT03593902), although the hematopoietic recovery from this reduced-dose regimen is prompt and reinfusion is not necessary in this setting. Because mobilization with G-CSF alone may exacerbate autoimmune diseases, there is a clinical rationale for combining G-CSF with cyclophosphamide. The use of 2 g/m2 cyclophosphamide for mobilization was as effective as 4 g/m2 at achieving the EBMT recommended minimum CD34+-cell yield, supporting dose reduction in the mobilization regimen to limit early toxicity as well as long-term risks associated with cumulative doses of cyclophosphamide. During the autologous HSCT procedures, inevitable infectious and non-infectious toxicity occurred especially following the high-dose cytotoxic conditioning and the cytopenic phase. A correlation between the num-

References 1. Denton CP, Khanna D. Systemic sclerosis. Lancet. 2017;390(10103):1685-1699. 2. Elhai M, Meune C, Boubaya M, et al. Mapping and predicting mortality from systemic sclerosis. Ann Rheum Dis.

382

ber of infused CD34+ cells and a reduced duration of aplasia after autologous HSCT has been described previously,33,34 which was confirmed in the NISCC1 study, although CD34+-cell dose and addition of G-CSF until recovery from aplasia were not associated with a reduced number of infectious complications. The NISCC1 study has several limitations. Since the study was a non-interventional trial, follow-up physical and laboratory evaluations, although recommended by EBMT guidelines,17 were not mandatory, and were not obtained at all time-points. The non-interventional study design was also the cause of significant variation in data collection practices by the 13 participating sites. Data were prospectively collected and checked for consistency by the EBMT study office; however, when some examinations were not performed, as may occur in routine clinical practice, data were classified as “missing”. The NISCC1 results showed significant variation in local practices when evaluating patients at baseline. Similarly, the number of reported days for mobilization varied largely according to centers; indeed, some patients were hospitalized for the whole duration of this procedure, whereas in other cases it was carried in part on an outpatient basis. We aimed to investigate the factors influencing the safety and efficacy of autologous HSCT. We used validated parameters to assess post-transplant outcome, such as mRSS or lung function tests despite their large intra- and inter-individual variability. The results enable deeper analysis of the benefits and risks of the transplant procedure in relation to both SSc-related factors, including the extent of heart, lung and kidney involvement at the time of patient referral, and transplant-related factors such as the conditioning regimen and graft manipulation. No international consensus has yet validated a unique activity score in SSc and further studies will be necessary in the post-transplant setting to investigate disease activity, the concept of drug-free remission and longer term outcomes, with or without post-transplant interventions. In conclusion, the NISSC1 is a large prospective real‐world data study of importance for patients, clinicians, and healthcare payers, given the poor prognosis and burden of disability of patients with severe SSc. It allowed us to assess multicenter adherence and compliance to good clinical practice and adds significantly to the evidence base supporting the delivery of autologous HSCT for severe SSc in routine clinical practice. Disclosures No conflicts of interests to diclose. Contributions JH, DF, MB, JS and ML: conception, study design, analyses and interpretation of data. MO, HUS, NDP, TD, MS, RS, TM, GP, BS, DM, EM, BL, JOB, MS, AH, FO, LK and ZM: acquisition of data.

2017;76(11):1897-1905. 3. Herrick AL, Peytrignet S, Lunt M, et al. Patterns and predictors of skin score change in early diffuse systemic sclerosis from the European Scleroderma Observational Study. Ann Rheum Dis. 2018;77(4):563-570. 4. Fransen J, Popa-Diaconu D, Hesselstrand R, et al. Clinical prediction of 5-year survival in

systemic sclerosis: validation of a simple prognostic model in EUSTAR centres. Ann Rheum Dis. 2011;70(10):1788-1792. 5. Tyndall AJ, Bannert B, Vonk M, et al. Causes and risk factors for death in systemic sclerosis: a study from the EULAR Scleroderma Trials and Research (EUSTAR) database. Ann Rheum Dis. 2010;69(10):1809-1815.

haematologica | 2021; 106(2)

ASCT for progressive SSc

6. Hao Y, Hudson M, Baron M, et al. Early mortality in a multinational systemic sclerosis inception cohort. Arthritis Rheumatol. 2017;69(5):1067-1077. 7. Steen VD, Medsger TA Jr, Rodnan GP. Dpenicillamine therapy in progressive systemic sclerosis (scleroderma): a retrospective analysis. Ann Intern Med. 1982;97(5):652-659. 8. Steen VD, Medsger TA, Jr. The value of the Health Assessment Questionnaire and special patient-generated scales to demonstrate change in systemic sclerosis patients over time. Arthritis Rheum. 1997;40(11):19841991. 9. Herrick AL, Pan X, Peytrignet S, et al. Treatment outcome in early diffuse cutaneous systemic sclerosis: the European Scleroderma Observational Study (ESOS). Ann Rheum Dis. 2017;76(7):1207-1218. 10. Kowal-Bielecka O, Fransen J, Avouac J, et al. Update of EULAR recommendations for the treatment of systemic sclerosis. Ann Rheum Dis. 2017;76(8):1327-1339. 11. Farge D, Marolleau JP, Zohar S, et al. Autologous bone marrow transplantation in the treatment of refractory systemic sclerosis: early results from a French multicentre phase I-II study. Br J Haematol. 2002;119 (3):726-739. 12. Farge D, Labopin M, Tyndall A, et al. Autologous hematopoietic stem cell transplantation for autoimmune diseases: an observational study on 12 years' experience from the European Group for Blood and Marrow Transplantation Working Party on Autoimmune Diseases. Haematologica. 2010;95(2):284-292. 13. Henes JC, Schmalzing M, Vogel W, et al. Optimization of autologous stem cell transplantation for systemic sclerosis -- a singlecenter longterm experience in 26 patients with severe organ manifestations. J Rheumatol. 2012;39(2):269-275. 14. Burt RK, Shah SJ, Dill K, et al. Autologous non-myeloablative haemopoietic stem-cell transplantation compared with pulse cyclophosphamide once per month for systemic sclerosis (ASSIST): an open-label, randomised phase 2 trial. Lancet. 2011;378 (9790):498-506. 15. van Laar JM, Farge D, Sont JK, et al. Autologous hematopoietic stem cell transplantation vs intravenous pulse cyclophosphamide in diffuse cutaneous systemic sclerosis: a randomized clinical trial. JAMA. 2014;311(24):2490-2498.

haematologica | 2021; 106(2)

16. Sullivan KM, Goldmuntz EA, Keyes-Elstein L, et al. Myeloablative autologous stem-cell transplantation for severe scleroderma. N Engl J Med. 2018;378(1):35-47. 17. Snowden JA, Saccardi R, Allez M, et al. Haematopoietic SCT in severe autoimmune diseases: updated guidelines of the European Group for Blood and Marrow Transplantation. Bone Marrow Transplant. 2012;47(6):770-790. 18. Duarte RF, Labopin M, Bader P, et al. Indications for haematopoietic stem cell transplantation for haematological diseases, solid tumours and immune disorders: current practice in Europe, 2019. Bone Marrow Transplant. 2019;54(10):1525-1552. 19. Sullivan KM, Majhail NS, Bredeson C, et al. Systemic sclerosis as an indication for autologous hematopoietic cell transplantation: position statement from the american society for blood and marrow transplantation. Biol Blood Marrow Transplant. 2018;24(10): 1961-1964. 20. Passweg JR, Baldomero H, Bader P, et al. Is the use of unrelated donor transplantation leveling off in Europe? The 2016 European Society for Blood and Marrow Transplant activity survey report. Bone Marrow Transplant. 2018;53(9):1139-1148. 21. Burt RK, Oliveira MC, Shah SJ, et al. Cardiac involvement and treatment-related mortality after non-myeloablative haemopoietic stem-cell transplantation with unselected autologous peripheral blood for patients with systemic sclerosis: a retrospective analysis. Lancet. 2013;381 (9872):1116-1124. 22. Farge D, Arruda LC, Brigant F, et al. Longterm immune reconstitution and T cell repertoire analysis after autologous hematopoietic stem cell transplantation in systemic sclerosis patients. J Hematol Oncol. 2017;10(1):21. 23. Oliveira MC, Labopin M, Henes J, et al. Does ex vivo CD34+ positive selection influence outcome after autologous hematopoietic stem cell transplantation in systemic sclerosis patients? Bone Marrow Transplant. 2016;51(4):501-505. 24. Ayano M, Tsukamoto H, Mitoma H, et al. CD34-selected versus unmanipulated autologous haematopoietic stem cell transplantation in the treatment of severe systemic sclerosis: a post hoc analysis of a phase I/II clinical trial conducted in Japan. Arthritis Res Ther. 2019;21(1):30.

25. Young NS. Aplastic anemia. N Engl J Med. 2018;379(17):1643-1656. 26. Gratwohl A, Passweg J, Bocelli-Tyndall C, et al. Autologous hematopoietic stem cell transplantation for autoimmune diseases. Bone Marrow Transplant. 2005;35(9):869-879. 27. Saccardi R, Tyndall A, Coghlan G, et al. Consensus statement concerning cardiotoxicity occurring during haematopoietic stem cell transplantation in the treatment of autoimmune diseases, with special reference to systemic sclerosis and multiple sclerosis. Bone Marrow Transplant. 2004;34(10):877881. 28. Mousseaux E, Agoston-Coldea L, Marjanovic Z, et al. Left ventricle replacement fibrosis detected by CMR associated with cardiovascular events in systemic sclerosis patients. J Am Coll Cardiol. 2018;71(6): 703-705. 29. Mueller KA, Mueller, II, Eppler D, et al. Clinical and histopathological features of patients with systemic sclerosis undergoing endomyocardial biopsy. PLoS One. 2015;10(5):e0126707. 30. Burt RK, Shah SJ, Gheorghiade M, Ruderman E, Schroeder J. Hematopoietic stem cell transplantation for systemic sclerosis: if you are confused, remember: "it is a matter of the heart". J Rheumatol. 2012;39(2):206-209. 31. Krumm P, Mueller KA, Klingel K, et al. Cardiovascular magnetic resonance patterns of biopsy proven cardiac involvement in systemic sclerosis. J Cardiovasc Magn Reson. 2016;18(1):70. 32. Farge D, Burt RK, Oliveira MC, et al. Cardiopulmonary assessment of patients with systemic sclerosis for hematopoietic stem cell transplantation: recommendations from the European Society for Blood and Marrow Transplantation Autoimmune Diseases Working Party and collaborating partners. Bone Marrow Transplant. 2017;52(11):1495-1503. 33. Keever-Taylor CA, Heimfeld S, Steinmiller KC, et al. Manufacture of autologous CD34(+) selected grafts in the NIAID-sponsored HALT-MS and SCOT multicenter clinical trials for autoimmune diseases. Biol Blood Marrow Transplant. 2017;23(9):14631472. 34. McSweeney PA, Nash RA, Sullivan KM, et al. High-dose immunosuppressive therapy for severe systemic sclerosis: initial outcomes. Blood. 2002;100(5):1602-1610.

383

ARTICLE Ferrata Storti Foundation

Iron Metabolism & its Disorders

Induction of erythroferrone in healthy humans by micro-dose recombinant erythropoietin or high-altitude exposure Paul Robach,1* Elena Gammella,2* Stefania Recalcati,2 Domenico Girelli,3 Annalisa Castagna,3 Matthieu Roustit,4 Carsten Lundby,5 Anne-Kristine Lundby,5 Pierre Bouzat,4 Samuel Vergès,6 Guillaume Séchaud,4 Pierluigi Banco,4 Mario Uhr,7 Catherine Cornu,8 Pierre Sallet9 and Gaetano Cairo2

Haematologica 2021 Volume 106(2):384-390

1 National School for Mountain Sports, Chamonix, France; 2Department of Biomedical Sciences for Health, University of Milan, Italy; 3Department of Medicine, University of Verona, Italy; 4Grenoble Alpes University Hospital, Grenoble, France; 5Center for Physical Activity Research, University Hospital, Copenhagen, Denmark; 6HP2 Laboratory, U1042, Grenoble Alpes University, INSERM, Grenoble, France; 7Department of Hematology Synlab-Suisse, Lugano, Switzerland; 8Hospices Civils de Lyon INSERM CIC1407/UMR5558, Hôpital Louis Pradel, Bron, France and 9“Athletes for Transparency” Association, Lyon, France

*PR and EG contributed equally as co-first authors.

ABSTRACT

Received: July 31, 2019. Accepted: January 2, 2020. Pre-published: January 9, 2020.

he erythropoietin (Epo)-erythroferrone (ERFE)-hepcidin axis coordinates erythropoiesis and iron homeostasis. While mouse studies have established that Epo-induced ERFE production represses hepcidin synthesis by inhibiting hepatic BMP/SMAD signaling, evidence for the role of ERFE in humans is limited. To investigate the role of ERFE as a physiological erythroid regulator in humans, we conducted two studies. First, 24 males were given six injections of saline (placebo), recombinant Epo (rhEpo) at a dose of 20 IU/kg (micro-dose) or rhEpo at 50 IU/kg (low dose). Second, we quantified ERFE in 22 subjects exposed to high altitude (3800 m) for 15 h. In the first study, total hemoglobin mass (Hb ) increased after low- but not after micro-dose injections, when compared to the mass after placebo injections. Serum ERFE levels were enhanced by rhEpo, remaining higher than after placebo for 48 h (micro-dose) or 72 h (low-dose) after injections. Conversely, hepcidin levels decreased when Epo and ERFE rose, before any changes in serum iron parameters occurred. In the second study, serum Epo and ERFE increased at high altitude. The present results demonstrate that in healthy humans ERFE responds to slightly increased Epo levels not associated with Hb expansion and downregulates hepcidin in an apparently iron-independent way. Notably, ERFE flags micro-dose Epo, thus holding promise as a novel biomarker of doping.

https://doi.org/10.3324/haematol.2019.233874

Introduction

mass

Correspondence: GAETANO CAIRO gaetano.cairo@unimi.it PAUL ROBACH paul.robach@ensm.sports.gouv.fr

384

mass

Erythropoiesis and iron metabolism are tightly linked and an inadequate iron supply to developing erythrocytes results in anemia, a condition affecting a large segment of the world’s population. The coordination between erythropoietic activity and iron homeostasis is provided by hepcidin, which controls body iron balance by negatively regulating the activity of the iron exporter, ferroportin.1,2 Hepcidin expression is inhibited by iron deficiency and high erythropoietic activity,1,2 a response that increases iron availability to meet iron needs for hemoglobin synthesis. Accordingly, we and others have demonstrated that recombinant human erythropoietin (rhEpo) administration to healthy humans is followed by a prompt downregulation of hepcidin.3-5 The identification and characterization of erythroferrone (ERFE), a hepcidin-inhibiting factor produced by erythroblasts in response to Epo, provided an additional link between erythropoietic activity and iron homeostasis.6 Mouse studies established that ERFE synthesized in response to Epo impairs hepcidin transcription by inhibiting hepatic BMP/SMAD signaling.6,7 The haematologica | 2021; 106(2)

Low-dose erythropoietin induces erythroferrone

Figure 1. Effects of repeated administration of very low to low doses of recombinant human erythropoietin on total hemoglobin mass, erythropoietin, erythroferrone and hepcidin. (A) Total hemoglobin mass (Hb ), determined at baseline (Pre) and 72 h after the last of the six injections (Post) of placebo (n=7), recombinant human erythropoietin (rhEpo) 20 IU/kg (micro-dose) (n=7) or rhEpo 50 IU/kg (low dose) (n=8). ANOVA for a significant time × treatment interaction (P=0.049). *Indicates a significant difference between low-dose rhEpo and placebo. Data are means ± standard deviations. Individual Hb values are available in Online Supplementary Figure S1. (B-D) Serum concentrations of erythropoietin (B), erythroferrone (ln transformed, n=7 for the placebo condition) (C) and hepcidin (ln transformed) (D) before, during and after six injections of placebo (n=8), micro-dose rhEpo (n=8) or low-dose rhEpo (n=8). Vertical dotted lines indicate injections. Baselines are means from duplicate blood samples collected on separated days. Blood sampling was repeatedly performed at 24 h (2 occurrences), 48 h (4 occurrences) and 72 h (2 occurrences) after an injection. Additional samples were obtained at 7 and 14 days after the last injection. Numbers above the abscissa line (i.e., 24, 48, 72, 7d and 14d) indicate these sampling times, respectively. Data are means ± standard deviations. For each panel, the inserted table reports estimated marginal means ± standard errors and P-values for the effect of the group (placebo, micro-dose, low dose), the occurrence (i.e., the number of samples at 24 h, 48 h, 72 h, 7 days and 14 days, from 1 to 4) and the group × occurrence (G × O) interaction. Multiple comparisons between groups at 24 h, 48 h or 72 h are reported in each table, *Denotes a difference from placebo and ‡denotes a difference from micro-dose rhEPO (see the Online Supplementary Material, Statistical analysis section). Epo: erythropoietin; ERFE: erythroferrone; PLA: placebo. EPO20: recombinant human erythropoietin 20 IU/kg; EPO50: recombinant human erythropoietin 50 IU/kg. mass

mass

haematologica | 2021; 106(2)

385

P. Robach et al. A

Figure 2. Effects of repeated administration of very low to low doses of recombinant human erythropoietin on hematologic and iron parameters. (A-D) Blood hemoglobin concentration (A), hematocrit (B), serum concentration of ferritin (ln transformed) (C) and transferrin saturation (D) before, during and after six injections of placebo (n=8), recombinant human erythropoietin (rhEpo) 20 IU/kg (micro-dose) (n=8) or rhEpo 50 IU/kg (low dose) (n=8). Vertical dotted lines indicate injections. Baselines are means from duplicate blood samples collected on separated days. Blood sampling was repeatedly performed at 24 h (2 occurrences), 48 h (4 occurrences) and 72 h (2 occurrences) after an injection. Additional samples were obtained at 7 and 14 days after the last injection. Numbers above the abscissa line (i.e., 24, 48, 72, 7d and 14d) indicate these sampling times. Data are means ± standard deviations. For each panel, the inserted table reports estimated marginal means ± standard errors and P-values for the effect of the group (placebo, micro-dose, low dose), the occurrence (i.e., the number of samples at 24 h, 48 h, 72 h, 7 days and 14 days, from 1 to 4) and the group × occurrence (G × O) interaction. Multiple comparisons between groups at 24 h, 48 h or 72 h are reported in each table, *Denotes a difference from placebo and ‡denotes a difference from micro-dose (see the Online Supplementary Material, Statistical analysis section). Epo: erythropoietin; ERFE: erythroferrone; PLA: placebo. EPO20: recombinant human erythropoietin 20 IU/kg; EPO50: recombinant human erythropoietin 50 IU/kg.

386

haematologica | 2021; 106(2)

Low-dose erythropoietin induces erythroferrone

development of a validated assay for human ERFE led to the demonstration that ERFE is increased in patients with Î˛-thalassemia and in response to blood donation or administration of high doses of rhEpo.8 Moreover, serum ERFE levels were found to be elevated in patients with chronic kidney disease treated with rhEpo9 and in children affected by iron deficiency anemia.10 However, another study did not find increased ERFE levels in patients with chronic kidney disease.11 In addition, ERFE did not change in patients with reduced erythropoiesis caused by androgen deprivation therapy12 or in athletes exposed to altitude13 and, unexpectedly, decreased in parallel with hepcidin in subjects experiencing expansion of hemoglobin mass (Hb ) due to exposure to altitude.14 The effect of accelerated erythropoiesis on ERFE in humans does, therefore, remain uncertain. Here, we tested the hypothesis that, in healthy humans, ERFE is a physiological erythroid regulator, i.e. it responds to moderate erythropoietic stimulation, such as very low rhEpo doses or exposure to high altitude.

The pattern of variation of ERFE levels mirrored that of serum Epo (Figure 3A). Conversely, serum hepcidin decreased when Epo and ERFE rose, independently of the dose, and remained low as long as Epo and ERFE were above baseline values (Figure 1D). Both ERFE and hepcidin returned to placebo levels 1 week after the last injection. Interestingly, there was no cumulative effect of repeated rhEpo injections on ERFE levels (Figure 1C). Consistent with previous studies,3-5 the decrease in serum ferritin

mass

Methods Study design For the rhEpo study, Hb , indices of iron homeostasis and hematologic parameters were repeatedly measured in 24 healthy males given six injections (every second/third day) of saline (placebo), rhEpo 20 IU/kg (micro-dose) or rhEpo 50 IU/kg (low dose) Written informed consent to participation in the study was provided by all these subjects. The study (NCT03276910) was approved by the French ethics committee (CPP Est-III, EudraCT 2017-000375-82). For the high-altitude study, iron parameters were determined in 22 healthy subjects exposed to high altitude (3800 m) for 15 h. Written informed consent to participation in the study was provided by all these subjects. The study (NCT02778659) was approved by the French ethics committee (CPP Sud-Est-III, EudraCT 2015-004512-38). Details of the methods are available in the Online Supplementary Material. mass

Results Three days after the last rhEpo injection, Hb was higher in subjects treated with low-dose rhEpo than in those given the placebo (Figure 1A). In contrast, treatment with micro-doses of rhEpo did not induce significantly higher Hb levels in comparison to those induced by placebo treatment. Hb was not significantly altered following placebo treatment but nonetheless tended to decrease, possibly because of frequent blood sampling. Short treatment duration may otherwise explain the marginal change of Hb following micro-doses of rhEpo. Unlike Hb , hemoglobin concentration and hematocrit progressively increased with both rhEpo doses (Figure 2A, B). As expected, circulating Epo levels increased after each injection, but rapidly declined, in particular after micro-dose treatment (Figure 1B). ERFE levels showed a significant dose-related increase after each injection, remaining above placebo levels for up to 48 h (micro-dose) or 72 h (low dose) and decreasing thereafter (Figure 1C). Following low-dose rhEpo treatment, ERFE reached levels similar to those found in patients with anemia induced by bleeding (see the Online Supplementary Material, Analyses section). mass

mass

haematologica | 2021; 106(2)

Figure 3. Relationships between erythroferrone and erythropoietin with recombinant human erythropoietin and high altitude. (A, B) Relationship between individual serum concentrations (ln transformed) of erythroferrone (ERFE) and erythropoietin (Epo), before, during and after six injections of placebo, micro-dose or low-dose recombinant human erythropoietin (rhEpo) in healthy subjects (A), and at sea level and after 15 h of exposure to high altitude (3800 m) in healthy subjects (B). Regression equations, coefficients of determination and P values are reported in each panel. Relationships between hepcidin, Epo and ERFE during rhEpo injections are available in Online Supplementary Figure S2.

387

P. Robach et al.

with rhEpo was progressive (Figure 2C). In contrast, transferrin saturation (Tfsat) was not significantly altered by rhEpo injections (Figure 2D). We also found concomitant increases in Epo and ERFE in healthy subjects exposed to a high altitude condition associated with O saturation of 85 Âą 3% (Figure 4A, B). The correlation between ERFE and serum Epo found at high altitude (Fig 3B) and with rhEpo treatment (Figure 3A) suggests that hypoxia-related signaling is not directly 2

involved in ERFE induction, as previously shown in mice.6 ERFE tended to increase more with micro-dose rhEpo than with exposure to high altitude (P=0.22), whereas Epo increased less with the micro-dose injections than with exposure to high altitude (P=0.04) (data not shown). We speculate that the shorter time of exposure to elevated Epo levels at high altitude (15 h) versus micro-dose injections (24 h) may account for the observed trend. Tfsat, which was close to the level defining iron deficiency

Figure 4. Effects of acute exposure to high altitude. Individual serum concentrations and means Âą standard deviations of erythropoietin (n=21) (A), erythroferrone (ln transformed, n=21) (B), transferrin saturation (n=22) (C), hepcidin (ln transformed, n=7) (D) and ferritin (n=22) (E), at sea level and after 15 h of exposure to high altitude (3800 m). Of note, out of 22 subjects, 15 subjects had sea-level hepcidin concentrations which were below the detection limit; therefore, the hepcidin statistical analysis was performed on seven subjects. P values denote differences between sea level and high altitude. Epo: erythropoietin; ERFE: erythroferrone.

388

haematologica | 2021; 106(2)

Low-dose erythropoietin induces erythroferrone

(<20%), did not change (Figure 4C). Consistent with the low Tfsat, hepcidin concentration at sea level was below the detection limit in 15 subjects and was unchanged after exposure to high altitude in the subjects with detectable baseline values (Figure 4D). Lack of correlation between ERFE and hepcidin was previously found in patients with chronic kidney disease, in whom increased ERFE levels were not accompanied by lower hepcidin.9 In line with a previous report showing unaltered iron availability and no signs of inflammation at high altitude,15 the inflammatory marker interleukin-6 was not affected by the subjects’ exposure to high altitude (1.28 ± 1.04 pg/mL vs. 1.25 ± 0.9 pg/mL at sea level), whereas the concentration of ferritin increased slightly (from 125 ± 8 to 132 ± 7 ng/mL), although remaining within the normal range (Figure 4E).

Discussion Recently, mouse studies have shown that acute rhEpo treatment downregulates hepcidin in an ERFE-independent manner by decreasing serum iron and Tfsat.16,17 Moreover, the demonstration that rhEpo administration also downregulates hepcidin in mice lacking ERFE18 suggests that prolonged erythropoietic stimulation inhibits hepcidin expression in mice by depleting iron stores, whereas ERFE represents an acute regulator of stress erythropoiesis.18 Conversely, the present results show that in healthy humans ERFE responds even to low Epo levels which are not associated with an expansion of Hb , a functional marker of erythropoietic response.19 This conclusion is also supported by our findings in a physiological condition such as high-altitude hypoxia (Figure 4). Furthermore, our data showing no alterations in Tfsat and a progressive decrease in ferritin with repeated rhEpo injections are consistent with the view that ERFE may inhibit hepcidin transcription directly in the absence of changes in serum and liver iron.7 In fact, the ERFE-hepcidin axis was affected early, i.e. 24 h after the first rhEpo injection, while other serum iron parameters were unchanged at that time, as t test analysis showed no difference in ferritin between groups at 24 and 48 h after the first injection (see Online Supplementary Material, Statistical analysis section). However, it is well conceivable that under conditions of strong erythropoietic stimulation, such as in mice treated with high doses of rhEpo (8000 IU/kg),16-18 increased iron consumption for erythropoiesis leads to iron depletion and repression of hepcidin. The introduction of the Athlete’s Blood Passport20 has improved the detection of blood doping, although the Passport does have several limitations,21 in particular in detecting micro-dose rhEpo doping.22 A study in which

ERFE was measured in six subjects receiving relatively high doses of rhEpo or analogs, intravenously or subcutaneously, using an assay different from the one used in this study did not suggest that ERFE would be a reliable marker for rhEpo doping,23 although (while our manuscript was under revision) the same group reported that a different enzyme-linked immunosorbent assay was able to detect increased ERFE levels in the same samples.24 Conversely, increased erythropoiesis induced by training did not affect ERFE and hepcidin levels in runners.25 Our results demonstrate that ERFE is sensitive enough to flag even microdose rhEpo, correlates with Epo levels (Figure 3A, B) and has a detection window longer than that of Epo, thereby indicating that ERFE holds promise as a novel biomarker of doping for implementation in the ABP, although additional studies are required. In view of our results, ferritin or hepcidin could also be considered as potential biomarkers of the use of micro-dose rhEpo, although both factors may be confounded by iron supplementation, a legal practice commonly used by athletes. In summary, the present results demonstrate that in healthy humans ERFE is promptly enhanced in response to moderately increased Epo levels and represses hepcidin in an iron-independent way. Assaying ERFE levels may provide additional analytical support for the fight against doping. Disclosures No conflicts of interests to disclose.

mass

References 1. Ganz T. Erythropoietic regulators of iron metabolism. Free Radic Biol Med. 2019;133:69-74. 2. Camaschella C, Pagani A. Advances in understanding iron metabolism and its crosstalk with erythropoiesis. Br J Haematol. 2018;182(4):481-494. 3. Robach P, Recalcati S, Girelli D, et al. Alterations of systemic and muscle iron

haematologica | 2021; 106(2)

Contributions PR designed and coordinated the project, planned and performed experiments and co-wrote the manuscript. EG collected and interpreted data and co-wrote the manuscript. SR, DG and AC collected and analyzed data. MR analyzed data and performed the statistical analysis, A-K L, PBa, PBo, SV, GS and MU collected data, CL analyzed data and revised the manuscript, CC provided clinical support, PS designed and coordinated the project, discussed the results and co-wrote the manuscript. GC conceived and coordinated the study, interpreted data and cowrote the manuscript. All authors discussed the results and commented on the manuscript. Acknowledgments The authors thank Dr. Catherine Mercier (Hospices Civils de Lyon) for expert statistical advice, as well as Dr. Jean-Pierre Herry, Dr. Alice Gavet and Laura Oberholzer for their assistance during the protocol. The work involving rhEpo administration was supported by a grant from Partnership for Clean Competition to GC. The altitude study was supported by the Fédération Française des Clubs Alpins et de Montagne and Fondation Petzl through grants to PBo.

metabolism in human subjects treated with low-dose recombinant erythropoietin. Blood. 2009;113(26):6707-6715. 4. Ashby DR, Gale DP, Busbridge M, et al. Erythropoietin administration in humans causes a marked and prolonged reduction in circulating hepcidin. Haematologica. 2010;95(3):505-508. 5. Robach P, Recalcati S, Girelli D, et al. Serum hepcidin levels and muscle iron proteins in humans injected with low- or high-dose erythropoietin. Eur J Haematol. 2013;91

(1):74-84. 6. Kautz L, Jung G, Valore EV, Rivella S, Nemeth E, Ganz T. Identification of erythroferrone as an erythroid regulator of iron metabolism. Nat Genet. 2014;46(7): 678-684. 7. Arezes J, Foy N, McHugh K, et al. Erythroferrone inhibits the induction of hepcidin by BMP6. Blood. 2018;132(14): 1473-1477. 8. Ganz T, Jung G, Naeim A, et al. Immunoassay for human serum erythrofer-

389

P. Robach et al. rone. Blood. 2017;130(10):1243-1246. 9. Hanudel MR, Rappaport M, Chua K, et al. Levels of the erythropoietin-responsive hormone erythroferrone in mice and humans with chronic kidney disease. Haematologica. 2018;103(4):e141-e142. 10. El Gendy FM, El-Hawy MA, Shehata AMF, Osheba HE. Erythroferrone and iron status parameters levels in pediatric patients with iron deficiency anemia. Eur J Haematol. 2018;100(4):356-360. 11. Honda H, Kobayashi Y, Onuma S, et al. Associations among erythroferrone and biomarkers of erythropoiesis and iron metabolism, and treatment with long-term erythropoiesis-stimulating agents in patients on hemodialysis. PLoS One. 2016;11(3):e0151601. 12. Gagliano-Jucรก T, Pencina KM, Ganz T, et al. Mechanisms responsible for reduced erythropoiesis during androgen deprivation therapy in men with prostate cancer. Am J Physiol Endocrinol Metab. 2018;315(6): E1185-E1193. 13. Garvican-Lewis LA, Vuong VL, Govus AD, et al. Intravenous iron does not augment the hemoglobin mass response to simulated hypoxia. Med Sci Sports Exerc.

390

2018;50(8):1669-1678. 14. Hall R, Peeling P, Nemeth E, Bergland D, McCluskey WTP, Stellingwerff T. Single versus split dose of iron optimizes hemoglobin mass gains at 2106 m altitude. Med Sci Sports Exerc. 2019;51(4):751-759. 15. Talbot NP, Lakhal S, Smith TG, et al. Regulation of hepcidin expression at high altitude. Blood. 2012;119(3):857-860. 16. Artuso I, Pettinato M, Nai A, et al. Transient decrease of serum iron after acute erythropoietin treatment contributes to hepcidin inhibition by ERFE in mice. Haematologica. 2019;104(3):e87-e90. 17. Mirciov CSG, Wilkins SJ, Hung GCC, Helman SL, Anderson GJ, Frazer DM. Circulating iron levels influence the regulation of hepcidin following stimulated erythropoiesis. Haematologica. 2018;103(10): 1616-1626. 18. Coffey R, Sardo U, Kautz L, Gabayan V, Nemeth E, Ganz T. Erythroferrone is not required for the glucoregulatory and hematologic effects of chronic erythropoietin treatment in mice. Physiol Rep. 2018;6(19):e13890. 19. Siebenmann C, Keiser S, Robach P, Lundby C. CORP: the assessment of total hemoglo-

bin mass by carbon monoxide rebreathing. J Appl Physiol (1985). 2017;123(3):645-654. 20. Cazzola M. A global strategy for prevention and detection of blood doping with erythropoietin and related drugs. Haematologica. 2000;85(6):561-563. 21. Jelkmann W, Lundby C. Blood doping and its detection. Blood. 2011;118(9):23952404. 22. Ashenden M, Gough CE, Garnham A, Gore CJ, Sharpe K. Current markers of the Athlete Blood Passport do not flag microdose EPO doping. Eur J Appl Physiol. 2011;111(9):2307-2314. 23. Leuenberger N, Bulla E, Salamin O, et al. Hepcidin as a potential biomarker for blood doping. Drug Test Anal. 2017;9(7):10931097. 24. Cuevas KR, Schobinger C, Gottardo E, et al. Erythroferrone as a sensitive biomarker to detect stimulation of erythropoiesis. Drug Test Anal. 2020;12(2):261-267. 25. Moretti D, Mettler S, Zeder C, et al. An intensified training schedule in recreational male runners is associated with increases in erythropoiesis and inflammation and a net reduction in plasma hepcidin. Am J Clin Nutr. 2018;108(6):1324-1333.

haematologica | 2021; 106(2)

ARTICLE

Iron Metabolism & its Disorders

C-FGF23 peptide alleviates hypoferremia during acute inflammation

Ferrata Storti Foundation

Rafiou Agoro,1 Min Young Park,1 Carole Le Henaff,1 Stanislovas Jankauskas,1 Alina Gaias,1 Gaozhi Chen,2 Moosa Mohammadi3 and Despina Sitara1,4

Department of Basic Science and Craniofacial Biology, NYU College of Dentistry, New York, NY, USA; 2Chemical Biology Research Center, School of Pharmaceutical Sciences, Wenzhou Medical University, Wenzhou, China; 3Departments of Biochemistry and Molecular Pharmacology, NYU School of Medicine, New York, NY, USA and 4Department of Medicine, NYU School of Medicine, New York, NY, USA 1

ABSTRACT

Haematologica 2021 Volume 106(2):391-403

ypoferremia results as an acute phase response to infection and inflammation aiming to reduce iron availability to pathogens. Activation of toll-like receptors (TLR), the key sensors of the innate immune system, induces hypoferremia mainly through the rise of the iron hormone hepcidin. Conversely, stimulation of erythropoiesis suppresses hepcidin expression via induction of the erythropoietin-responsive hormone erythroferrone. Iron deficiency stimulates transcription of the osteocyte-secreted protein FGF23. Here we hypothesized that induction of FGF23 in response to TLR4 activation is a potent contributor to hypoferremia and, thus, impairment of its activity may alleviate hypoferremia induced by lipopolysaccharide (LPS), a TLR 4 agonist. We used the C-terminal tail of FGF23 to impair endogenous full-length FGF23 signaling in wildtype mice, and investigated its impact on hypoferremia. Our data show that FGF23 is induced as early as pro-inflammatory cytokines in response to LPS, followed by upregulation of hepcidin and downregulation of erythropoietin (Epo) expression in addition to decreased serum iron and transferrin saturation. Further, LPS-induced hepatic and circulating hepcidin were significantly reduced by FGF23 signaling disruption. Accordingly, iron sequestration in liver and spleen caused by TLR4 activation was completely abrogated by FGF23 signaling inhibition, resulting in alleviation of serum iron and transferrin saturation deficit. Taken together, our studies highlight for the first time that inhibition of FGF23 signaling alleviates LPS-induced acute hypoferremia.

Introduction Hypoferremia develops as a response to innate immune system activation following inflammation or infection. Iron is an essential element required for the survival, proliferation, and virulence of pathogens. The hypoferremic response is an important defense mechanism that aims to prevent iron bioavailability to pathogens by sequestration of iron within macrophages, resulting in a significant reduction in circulating iron levels.1-3 Toll-like receptors (TLR) are key components of the innate immune system that recognize pathogen-associated molecular patterns (PAMP), and are responsible for sensing invading pathogens. Activation of TLR signaling plays a pivotal role in the development of the hypoferremic host response.2,3 Lipopolysaccharide (LPS), a PAMP that is the major component of the outer membrane of Gram-negative bacteria, is known to activate TLR4. Injection of LPS to mice or humans causes a rapid release of pro-inflammatory cytokines and triggers a well characterized acute phase inflammatory response followed by an induction of the hepatic hormone hepcidin, the central regulator of iron homeostasis.4-6 Hepcidin binds to ferroportin, the only known iron exporter, and causes its internalization and degradation,7 resulting in iron sequestration in reticulo-endothelial macrophages, hepatocytes, and duodenal enterocytes, and subsequently decreased circulating iron levels.8 This situation of circulating iron restriction promotes ineffective erythropoiesis and results in anemia of inflammation, a form of anemia commonly associated with infectious diseases, cancer, and chronic kidney disease (CKD).9 haematologica | 2021; 106(2)

Correspondence: DESPINA SITARA ds199@nyu.edu Received: September 2, 2019. Accepted: March 13, 2020. Pre-published: March 19, 2020. https://doi.org/10.3324/haematol.2019.237040

391

R. Agoro et al.

CKD patients have increased iron losses due to chronic bleeding from impaired intestinal iron absorption, uremiaassociated platelet dysfunction, frequent phlebotomy, and blood trapping in the dialysis apparatus.10 In addition, circulating levels of fibroblast growth factor 23 (FGF23) dramatically increase in CKD, reaching up to 2,000-fold above the normal range in advanced renal failure.11 Recent studies have associated iron deficiency with the increase in FGF23 levels.12,13 FGF23 has been identified as the gene responsible for the phosphate wasting disorder autosomal dominant hypophosphatemic rickets (ADHR)14 and as a causative factor of tumor-induced osteomalacia.15 FGF23 is a bonederived endocrine regulator of phosphate and vitamin D homeostasis and bone mineralization.16-20 FGF23 inhibits phosphate reabsorption in the renal proximal tubule by suppressing the expression of the type II sodium-phosphate transporters NaPi2a and NaPi2c in the brush border membrane.17,21-23 Furthermore, FGF23 reduces circulating levels of 1,25-dihydroxyvitamin D3 by suppressing the renal expression of 1α-hydroxylase, the enzyme required for production of 1,25-dihydroxyvitamin D3, and increasing the activity of 24-hydroxylase, the enzyme that metabolizes 1,25-dihydroxyvitamin D3.17 FGF23 requires a co-receptor, klotho, for FGF receptor (FGFR) activation, and its effects are mediated through multiple FGFR, with FGFR1 being the principal receptor to mediate the effects of FGF23 on phosphate regulation, and FGFR3 and 4 mediating the FGF23 effects on vitamin D metabolism.24 In the kidney, FGF23 binds to the FGFR-klotho complex and induces signaling through Ras/MAPK, PI3K/Akt, and PLCγ/PKC downstream pathways which regulate mineral metabolism genes.25 FGF23 activity is regulated post-translationally by proteolytic cleavage between Arg179 and Ser180, and O-linked glycosylation. Cleavage of the intact full-length FGF23 peptide by a subtilisin-like pro-protein convertase abolishes its biologic activity. Intact FGF23 contains an arginine–X–X–arginine (RXXR) motif which is recognized by pro-protein convertases such as furin. This motif is located at the boundary between the FGF core homology domain, which interacts with the FGFR, and the 72-residue-long C-terminal tail of FGF23, which interacts with the co-receptor Klotho.26-28 We recently reported that inhibition of FGF23 signaling rescues renal anemia in a mouse model of CKD.29 Here, we investigated the effect of inhibition of FGF23 signaling in a mouse model of LPS-induced hypoferremia. Our results show that FGF23 is induced in response to LPS prior to hepcidin upregulation followed by downregulation of Epo expression. Moreover, we show that inhibition of FGF23 signaling alleviates hypoferremia and attenuates dysregulation of erythropoiesis in a mouse model of acute inflammation induced by LPS.

Methods Animals C57BL/6J wild-type (WT) male mice between 8 and 10 weeks of age were purchased from The Jackson Laboratory (Bar Harbor, ME, USA) and housed at the New York University (NYU) College of Dentistry Animal Facility, where they were kept on a light/dark (12 hours [h]/12h) cycle at 23ºC, and received food (standard lab chow) containing 185 parts per million iron (Teklad 2018S; Harlan) and water ad libitum. All animal protocols were approved by the 392

Institutional Animal Care and Use Committee (IACUC) at NYU. Mice were treated with a single intraperitoneal (i.p.) injection of 1 mg/kg of an FGF23 blocking peptide (provided by Dr. M. Mohammadi, NYU School of Medicine) or vehicle (HEPESbuffered saline; 25 mM HEPES-NaOH pH7.5, 150 mM NaCl). The FGF23 blocking peptide is the 72 aa long C-terminal tail of human FGF23 (residues Ser180-Ile251), which corresponds to the C-terminal fragment of FGF23 generated by proteolytic cleavage at the RXXR motif.27 Eight hours post treatment, acute inflammation was induced by a single i.p injection of 50 mg/kg of LPS (E. coli 055:B5, Sigma-Aldrich, St Louis, MO, USA). An equal volume of sterile saline solution (0.9% NaCl) was used as vehicle control. Mice were euthanized 4 h after LPS treatment. In longitudinal studies, saline- and LPS-treated mice were euthanized after 1, 2, 4, 6, 12, or 24 h of LPS treatment.

Blood, tissue, and serum collection Mice were immediately necropsied after euthanasia and blood was collected by cardiac puncture. Complete blood count was performed using the VetScan HM5 Hematology Analyzer (ABAXIS, Union City, CA, USA). Serum was obtained after blood centrifugation at 3,500 rpm for 10 minutes (min). Liver, spleen, kidney, bones, and bone marrow cells obtained from femora and tibiae, were collected, snap-frozen in liquid nitrogen, and stored at -80°C until further use.

Serum measurements Serum FGF23 levels were measured using the mouse FGF23 Intact and C-terminal ELISA assays (Quidel Corporation/Immutopics International, San Clemente, CA, USA). The intact assay detects exclusively biologically intact FGF23 (iFGF23), whereas the C-terminus (cFGF23) assay is capable of detecting both the intact molecule and its C-terminal fragments. Serum phosphorus was determined by colorimetric measurements using the Stanbio Phosphorus Liqui-UV Test reagent (Stanbio Laboratory, Boerne, TX, USA). Serum iron and transferrin saturation were measured using the Iron-TIBC kit from Pointe Scientific (Canton, MI, USA). Serum hepcidin was measured using the Hepcidin Murine-Compete ELISA kit (Intrinsic LifeSciences, La Jolla, CA, USA). Serum EPO levels were measured using the Rat/Mouse EPO Quantikine ELISA kit (R&D Systems, Minneapolis, MN, USA).

Tissue iron measurement Hepatic and splenic iron concentration was measured by the Ferrozine colorimetric method, as described elsewhere.30,31

RNA isolation, reverse transcription, and real-time quantitative polymerase chain reaction analysis Total RNA was extracted from kidneys, liver, whole bone, spleen, and bone marrow using Trizol (Ambion; Life Technologies, Carlsbad, CA, USA) according to the manufacturer’s protocol (Molecular Research Center, Cincinnati, OH, USA). Synthesis of cDNA was performed using the High Capacity cDNA Reverse Transcription Kit as described by the manufacturer (Applied Biosystems; Thermo Fisher Scientific, Waltham, MA, USA). cDNA was amplified by quantitative polymerase chain reaction (qPCR) using the PerfecCTa SYBR Green SuperMix (Quanta Biosciences, Gaithersburg, MD, USA). All primers used in this study are listed in the Online Supplementary Table S1. mRNA levels were normalized to the housekeeping gene (Gadph or Hprt) in the same cDNA sample. The relative transcript expression of a gene is given as DCt=Cttarget−Ctreference. The fold change in gene expression, as compared to control mice, was determined as 2−DDCt values (DDCt=DCttreated−DCtcontrol). haematologica | 2021; 106(2)

C-FGF23 peptide alleviates hypoferremia

Figure 1. Early induction of Fgr23 in response to lipopolysaccharide (LPS). C57BL/6J mice were injected intraperitoneally (i.p.) with a single dose of saline (0.9% NaCl, indicated as Vehicle) or LPS (50 mg/kg). Samples were collected at 0 (saline injection only), 1, 2, 4, 6, 12, and 24 hours (h) after treatment and total mRNA was isolated. (A-C) Quantitative real-time polymerase chain reaction (qRT-PCR) of (A) IL-6, (B) TNF-α, and (C) IL-1β expression in liver. (D and E) Serum concentration of (D) intact and (E) C-terminal FGF23 measured by ELISA. (F and G) qRT-PCR for Fgf23 expression in (F) liver and (G) spleen. Data are expressed as fold change (2-DDCt) relative to housekeeping genes Gapdh or Hprt. Samples were measured in duplicates (vehicle, n=3-4; LPS, n=5-8), and data are represented as mean+standard deviation. All data were analyzed for normality with Shapiro-Wilk test and equivalence of variance using Levene’s test. For samples with normal distribution, two-way ANOVA was performed in each vehicle- or LPS-treated group compared to 0 h with Bonferroni’s multiple comparison test (D). When the samples were not in normal distribution and equivalence of variance, data were analyzed with non-parametric Kruskal-Wallis test (A-C, E-G). ns: not significant, *P<0.05, **P<0.01, ***P<0.001 compared to 0 h.

Western blotting Snap frozen samples of spleen and liver were homogenized in an ice-cold RIPA buffer (Sigma-Aldrich, St. Louis, MO, USA) with addition of EDTA-free HaltTM Protease Inhibitor Single-use Cocktail and HaltTM Phosphatase Inhibitor Single-Use Cocktail (both from Thermo Scientific, Rockford, lL, USA) and were lysed for 2 h on ice with vortexing every 15 min. After lysis, samples were centrifuged for 20 min at 13,000 g and protein concentration was measured in supernatants by Quick Start Bradford Dye Reagent according to the manufacturer’s instructions. For each sample, 30 mg of protein was loaded into 4-20% Mini-Protean® TGX™ Precast Protein Gels. Electrophoresis was performed for 18 min on a constant voltage of 300 V in a Tris/Glycine/SDS buffer. Then proteins were transferred to 0.45 mm nitrocellulose membrane by wet method with Tris/Glycine Transfer buffer (Quality Biological Inc., Gaithersburg, MD, USA) with the addition of 20% methanol (Fisher Scientific, Hampton, NH, USA) for 1 h on a constant voltage of 110 V. Membranes were then washed with Trisbuffered saline Tween (TBST) (Fisher Scientific, Hampton, NH, USA) and blocked for 2 h at room temperature (RT) with either 5% non-fat dry milk (LabScientific Inc., Highlands, NJ, USA) in TBST (for anti-ferroportin and anti-STAT3 antibodies) or 5% BSA (Sigma-Aldrich, St. Louis, MO, USA) in TBST (for anti-phosphoSTAT3 antibodies) and washed four times with TBST. Membranes were stained with primary antibodies overnight at 4°C: anti-STAT3 mouse antibody (1:1,000 on 5% non-fat dry milk-TBST), anti-phospho-STAT3 (Tyr705) rabbit antibody haematologica | 2021; 106(2)

(1:2,000 on 5% BSA-TBST) (#9139 and #9145, respectively, Cell Signaling Technology, Danvers, MA, USA), anti-ferroportin rabbit antibody (1:1,000 on 5% non-fat dry milk TBST) (#MTP11-S, Alpha Diagnostic Intl Inc., San Antonio, TX, USA), and anti-Actin goat antibody (1:1,000 on TBST) (sc-1616, Santa Cruz Biotechnology Inc., Dallas, TX, USA). Membranes were stained with secondary antibodies diluted 1:10,000 on TBST for 30 min at RT: mouse-anti-rabbit IgG-HRP, goat-anti-mouse IgG-HRP and Rb-anti-Gt IgG-HRP (all from Santa Cruz Biotechnology Inc., Dallas, TX, USA). After incubation, membranes were washed with TBST four times. Visualization was performed with Clarity Western ECL Substrate and ChemiDocTM XRS+ imager, images were analyzed with Image LabTM, and protein signal intensity was calculated with ImageJ (NIH, Bethesda, MD, USA). All reagents, materials, devices and software were purchased from Bio-Rad Laboratories Inc., Hercules, CA, USA, unless otherwise specified.

Statistical analysis All data were analyzed for normal distribution by Shapiro-Wilk tests and “Normal Q-Q Plot” graphs. If normal distribution was not achieved, data were aligned in rank transformation and a normality test was performed. The equality of variance was determined with Levene’s test. If both normal distribution and equality of variance were demonstrated, data were analyzed by two-way analysis of variance (ANOVA) with Bonferroni’s multiple comparison test. When data were not normally distributed, non-parametric test was performed by Kruskal-Wallis test using SigmaStat soft393

R. Agoro et al. A

Figure 2. Regulation of phosphate homeostasis following lipopolysaccharide (LPS) administration. (A) Phosphate levels measured in serum and urine at 0, 1, 2, 4, 6, 12, and 24 hours (h) after i.p. injection of LPS (50 mg/kg). (B-D) Quantitative real-time polymerase chain reaction (qRT-PCR) for renal (B) Klotho, (C) NaPi2a, and (D) NaPi2c expression. Data are expressed as fold change (2-DDCt) relative to housekeeping gene Gapdh. Samples were measured in duplicates (vehicle, n=3-4; LPS, n=5-8). Data are represented as mean+standard deviation. All data were analyzed for normality with Shapiro-Wilk test and equivalence of variance using Levene’s test. For serum Pi, data were analyzed by non-parametric Kruskal-Wallis test; for the ratio of urinary Pi to creatinine, data were aligned in RANK transformation and analyzed with one-way ANOVA followed by Bonferroni’s multiple comparison test (A). With the samples showing normal distribution, two-way ANOVA was performed in each vehicle- or LPS-treated group compared to 0 h with Bonferroni’s multiple comparison test (C). The samples not in normal distribution were analyzed with nonparametric Kruskal-Wallis test (B and D). ns: not significant, *P<0.05, **P<0.01, ***P<0.001 compared to 0 h.

ware (SPSS Sciences, Chicago, IL, USA) and GraphPad Prism 8 (2019 GraphPad Software, Inc. La Jolla, CA, USA). All values are presented as mean+standard deviation (SD). P<0.05 was considered statistically significant.

Results Early induction of Fgf23 and regulation of phosphate homeostasis in response to lipopolysaccharide To identify the series of events that take place in response to inflammation, we carried out longitudinal studies during which we injected a sub-lethal dose (50 mg/kg) of LPS to wild-type mice and evaluated its effect on inflammatory markers, FGF23 levels, and phosphate homeostasis after 1, 2, 4, 6, 12, and 24 h. As expected, liver pro-inflammatory cytokines such as IL-6, TNF-α, and IL-1β were significantly induced during the first 6 h of LPS treatment, reaching a peak at 1 h, and returning progressively to baseline levels by 12 h upon resolution of inflammation (Figure 1A-C). Furthermore, we investigated the effect of LPS treatment on Fgf23 expression and circulating FGF23 levels. Both serum intact and Cterminal FGF23 levels were significantly elevated between 2-6 h or 2-12 h, respectively, in response to LPS (Figure 1D and E). Bone Fgf23 expression increased 4 h 394

after LPS treatment (Online Supplementary Figure S1A). However, in liver, Fgf23 expression was significantly upregulated 2 h after LPS treatment and remained elevated up to 12 h, before progressively returning to basal levels by 24 h (Figure 1F). In addition, Fgf23 expression was markedly increased in spleen and bone marrow during the first 12 h after LPS treatment and returned to basal levels by 24 h (Figure 1G and Online Supplementary Figure S1B). Serum phosphate levels were significantly increased the first 2 h of LPS treatment, and they were associated with a concurrent significant decrease in urinary phosphate excretion (Figure 2A), possibly causing the induction of FGF23. In a counter-regulatory manner, the increase in circulating FGF23 led to a significant increase in urinary phosphate levels 6 h after LPS treatment, resulting in normalization of serum phosphate levels (Figure 2A). In kidney, FGF23 binds to the FGFRKlotho complex to induce phosphate excretion by downregulating the sodium-phosphate transporters NaPi2a/c. Consistent with increased FGF23 levels and urinary phosphate excretion, renal expression of NaPi2a, NaPi2c, as well as Klotho, decreased significantly during the period of FGF23 elevation in response to LPS, before returning to basal levels by 24 h (Figure 2B-D). Taken together, our data demonstrate that, in response to LPS, FGF23 is induced as early as inflammatory markers, resulting in haematologica | 2021; 106(2)

C-FGF23 peptide alleviates hypoferremia Table 1. Effects of lipopolysaccharide (LPS) and fibroblast growth factor 23 (FGF23) blocking peptide on hematologic parameters in mice. 9

WBC, x10 /L Lymphocyte, x109/L Neutrophil, x109/L RBC, x1012/L Hb, g/dL Hematocrit, % RDW, % MCV, fL a

Ctl

LPS

FGF23 BL

FGF23 BL + LPS

2.47±0.41 1.73±0.66 0.19±0.22 9.80±0.35 14.18±0.41 45.24±1.55 18.14±1.18 46.00±0.00

2.79±0.78 1.22±0.30 1.49±0.53a*** 10.85±0.59a** 15.23±0.94a* 48.96±2.49a** 17.41±0.43 45.60±1.65

2.16±0.80 2.09±0.53 0.23±0.22 10.09±0.22 14.62±0.69 46.96±1.12 17.50±0.37 46.40±0.55

1.72±0.76b** 0.80±0.36b*, c*** 0.81±0.45b**, c* 10.28±0.68b* 14.45±0.67b* 46.32±2.23b* 16.73±0.50b**, c* 45.20±1.48

Values of LPS or FGF23 BL (FGF23 blocking peptide) differ significantly compared to control (vehicle) (Ctl). bValues of FGF23 BL+LPS differ significantly compared to LPS. Values of FGF23 BL+LPS differ significantly compared to FGF23 BL. *P<0.05, **P<0.01, ***P<0.001.

regulation of phosphate homeostasis by promoting phosphate excretion.

Acute inflammation induces hypoferremia and downregulates Epo expression Inflammation is a major stimulus for hepcidin synthesis in the liver. Specifically, treatment of hepatocytes with LPS leads to increase in hepcidin levels.32,33 In agreement with published studies, our data show that hepcidin expression started rising 1 h after LPS treatment with significant upregulation between 4-6 h after LPS (Figure 3A), consistent with the presence of inflammation (Figure 1A-C), and returned to basal levels after resolution of inflammation (Figure 3A). The increase in hepcidin resulted in progressive and significant reduction in circulating iron and transferrin saturation levels (Figure 3B and C), followed by recovery from hypoferremia once inflammation was resolved and hepcidin levels returned to normal. This drop in serum iron levels was accompanied by a significant decrease in ferroportin expression in liver and spleen (Figure 3D and E), leading to iron retention in these organs (Figure 3F and Online Supplementary Figure S2) and increased spleen weight (Figure 3G). In addition, expression of the acute phase protein lipocalin 2 (Lcn2), which plays a role as an iron sequester,34 was increased after 4 h of LPS and remained elevated up to 24 h post LPS treatment (Figure 3H). These data demonstrate that LPS induces hypoferremia associated with downregulation of ferroportin and iron retention in liver and spleen. Studies have shown that pro-inflammatory cytokines, such as IL-6, TNF-α, and IL-1β, increased by inflammation and/or infection, inhibit renal erythropoietin (Epo) production, the primary regulator of erythropoiesis, and therefore suppress erythropoiesis.35-37 This is known as anemia of inflammation. Specifically, it has been shown that administration of LPS in rats significantly suppressed renal Epo gene expression.36 In our studies, we found that renal Epo expression was significantly downregulated after 4-6 h (Figure 3I).

Inhibition of FGF23 signaling alleviates hypoferremia induced by acute inflammation It has been reported that the C-terminal tail of FGF23 mediates the binding of FGF23 to the FGFR-Klotho complex.27,38 Studies have shown that the 72-aa-long C-terminal tail of FGF23 (residues Ser180-Ile251), which corresponds to the C-terminal fragment of FGF23 generated by proteolytic cleavage at the RXXR motif, impairs FGF23 haematologica | 2021; 106(2)

signaling and activity by acting as endogenous inhibitor of the full-length FGF23 for binding to the FGFR-Klotho complex.27 We previously reported that inhibition of FGF23 signaling using this FGF23 blocking peptide (BL) ameliorates iron deficiency in a mouse model of CKD.29 Since FGF23 significantly increases in response to LPS, here we assessed the effect of inhibiting FGF23 signaling in a mouse model of hypoferremia induced by acute inflammation. Based on the results of our longitudinal studies, we opted to assess the effect of FGF23 signaling inhibition 4 h post LPS treatment. At this time point, we observed maximum induction of FGF23 and hepcidin, associated with a significant decrease in serum iron, transferrin saturation, and erythropoietin, during upregulation of pro-inflammatory cytokines. Our data show that the increase in FGF23 in response to LPS was significantly reduced after inhibition of FGF23 signaling (Figure 4A-C). Administration of LPS triggered an acute inflammatory response (Figure 5A-C) that led to induction of hepcidin (Figure 5D and E). However, blocking FGF23 signaling significantly decreased hepatic expression of TNF-α without affecting IL-6 and IL-1β mRNA expression in LPS-treated mice (Figure 5A-C). Importantly, the increase in hepatic and circulating hepcidin levels induced by LPS was significantly reduced when FGF23 signaling was impaired (Figure 5D and E). We also found that inhibition of FGF23 signaling did not impact phosphorylation of STAT3, the main intracellular regulator of hepcidin induction by IL6, as treatment with cFGF23 did not affect LPS-induced upregulation of hepatic STAT3 phosphorylation (Online Supplementary Figure S3). Therefore, suppression of hepcidin by cFGF23 appears to be independent of the IL6/STAT3 pathway. These data suggest that inhibiting FGF23 signaling reduces LPS-induced hepcidin in mice with normal kidney function. Increased hepcidin leads to iron sequestration in tissues and subsequent decrease in circulating iron levels.7,8,39 As expected, LPS elicited hypoferremia after 4 h with a significant decrease in serum iron and transferrin saturation (Figure 6A and B), associated with a robust reduction in liver and spleen ferroportin (Fpn) mRNA (Figure 6C and D). Downregulation of ferroportin resulted in increased hepatic and splenic iron content (Figure 6E and F), which correlated with increased hepatic expression of the iron storage protein ferritin H (Fth) (Figure 6G) and the iron sequester lipocalin 2 (Lcn2) (Figure 6H). The number of circulating neutrophils, which are the main source of lipocalin secretion, was 395

R. Agoro et al. A

Figure 3. Effect of lipopolysaccharide (LPS) on iron homeostasis and renal Epo mRNA expression. C57BL/6J mice were injected intraperitoneally (i.p.) with saline (0.9% NaCl, indicated as Vehicle) or LPS (50 mg/kg). Samples were collected at 0, 1, 2, 4, 6, 12, and 24 hours (h) after treatment. (A) Quantitative real-time polymerase chain reaction (qRT-PCR) for hepcidin expression in liver. (B) Serum iron and (C) serum transferrin saturation. (D and E) qRT-PCR for ferroportin (Fpn) expression in (D) liver and (E) spleen. (F) Iron content in spleen. (G) Spleen weight normalized to the body weight of the animals. (H) qRT-PCR for lipocalin (Lcn2) in liver. (I) qRT-PCR for renal Epo mRNA expression. Data are expressed as fold change (2-DDCt) relative to housekeeping genes Gapdh or Hprt. Samples were measured in duplicates (vehicle, n=3-4; LPS, n=5-8). Data are represented as mean±standard deviation. All data were analyzed for normality with Shapiro-Wilk test and equivalence of variance using Levene’s test. The samples not in normal distribution were analyzed with non-parametric Kruskal-Wallis test (A and B, D-I). When the samples showed normal distribution, two-way ANOVA was performed with Bonferroni’s multiple comparison test in each vehicle- or LPS-treated group compared to 0 h (C). ns: not significant, *P<0.05, **P<0.01, ***P<0.001 compared to 0 h.

also increased in response to LPS (Table 1). Importantly, inhibition of FGF23 signaling increased serum iron and transferrin saturation levels in both basal and LPS conditions (Figure 6A and B), and prevented LPS-induced liver and spleen iron sequestration (Figure 6E and F). Circulating neutrophils and expression of ferritin H and lipocalin 2 in the liver were also significantly reduced after administration of the FGF23 C-tail blocking peptide in LPS-treated mice (Table 1 and Figure 6G and H). However, blocking FGF23 signaling did not change the expression of liver and spleen ferroportin in response to LPS (Figure 6C and D), but in basal conditions it selectively induced splenic ferroportin mRNA and protein expression compared to controls (Figure 6D and Online Supplementary Figure S4). Taken together, these data suggest that inhibition of FGF23 signaling can alleviate 396

hypoferremia induced by acute inflammation in mice in the absence of renal disease.

Inhibition of FGF23 signaling increases renal and EPO and EpoR Consistent with previously published data,36 our studies show that LPS significantly downregulated renal Epo and EpoR expression after 4 h (Figure 7A and B). We also found that renal Hif2α mRNA was significantly decreased after LPS treatment (Figure 7C), in line with the role of HIF2 as a regulator of Epo synthesis and iron metabolism.40 Importantly, inhibition of FGF23 signaling prevented the downregulation of renal Epo, EpoR, and Hif2α expression caused by LPS (Figure 7A-C). We previously reported that treatment with the FGF23 C-tail blocking peptide significantly increases circulating Epo levels in mice.29 Here, we haematologica | 2021; 106(2)

C-FGF23 peptide alleviates hypoferremia

Figure 4. Inhibition of fibroblast growth factor 23 (FGF23) signaling decreases Fgf23 expression and circulating levels induced by lipopolysaccharide (LPS). C57BL/6J wild-type mice were injected intraperitoneally (i.p.) with C-tail FGF23 (1 mg/kg, indicated as FGF23 BL) or vehicle (HEPES buffer) for 8 hours (h). Mice were then challenged with LPS (i.p. 50 mg/kg) or vehicle (0.9% NaCl) for 4 h. (A) Quantitative real-time polymerase chain reaction (qRT-PCR) for Fgf23 expression in bone. Data are expressed as fold change (2-DDCt) relative to housekeeping gene Gapdh. (B and C) Serum concentration of (B) C-terminal FGF23 (cFGF23) and (C) intact FGF23 measured by ELISA. Samples were measured in duplicates (n=5-7 per group). Data are represented as mean+standard deviation. All data were analyzed for normality with Shapiro-Wilk test and equivalence of variance using Levene’s test. Because the samples did not show normal distribution, data were aligned in RANK transformation, and confirmed for normality. As the samples showed normal distribution, two-way ANOVA was performed with Bonferroni’s multiple comparison test. Ctl: control (vehicle), ns: not significant, *P<0.05, **P<0.01, ***P<0.001.

confirmed this effect in mice injected with LPS or saline (Figure 7D) using a 10 times lower concentration of the FGF23 C-tail blocking peptide than the one we used in our previous work.29 However, we found that 4 h of LPS treatment at the concentration of 50 mg/kg were not sufficient to alter circulating Epo levels (Figure 7D). In addition to its effect on renal Epo and EpoR, LPS significantly decreased the expression of splenic and hepatic Epo and EpoR (Figure 7E-H), which was rescued by treatment with the FGF23 C-tail blocking peptide (Figure 7E-H). These data clearly demonstrate the efficiency of the C-tail FGF23 in increasing renal and extra-renal Epo and EpoR expression in the presence of inflammation, leading to increased serum Epo levels. Our findings also indicate that in the absence of inflammation, the C-tail FGF23 increases circulating Epo levels by upregulating splenic Epo expression without affecting renal or hepatic Epo expression.

Discussion Iron is an essential micronutrient for host-pathogen interaction. During infection, pathogens develop multiple strategies to acquire iron for proliferation and virulence, such as secretion of ferric iron-binding molecules known as siderophores.41 A host defense mechanism to inhibit pathogen growth is to deprive invading pathogens of iron, a process known as nutritional immunity by sequestering iron in tissues, thus limiting availability of circulating iron and resulting in hypoferremia.42 Iron homeostasis is controlled by two proteins, the hepatic hormone hepcidin which regulates iron absorption, and the iron exporter ferroportin which exports iron into the circulation from duohaematologica | 2021; 106(2)

denal enterocytes, hepatocytes, and splenic reticuloendothelial macrophages that recycle the iron of senescent erythrocytes.43 Toll-like receptors (TLR) are key components of the innate immune system that recognize pathogen-associated molecular patterns (PAMP). Activation of TLR signaling plays a pivotal role in the development of the hypoferremic host response.2,3,44 Several toll like receptors such as TLR3, 4, 7, 7/8 and 9 have been described to induce hepcidin expression in vitro and in vivo.44 Several studies have demonstrated that iron deficiency induces FGF23 production.12,13,45 We recently reported that inhibition of FGF23 signaling using the C-tail FGF23 rescues anemia and iron deficiency in a CKD mouse model characterized by chronic inflammation, by increasing EPO, serum iron and ferritin levels, and attenuating inflammation.29 Our goals in the present study were: (i) to decipher the kinetics of Fgf23 in response to LPS in comparison to the kinetics of inflammatory, iron metabolism, and erythropoiesis parameters; and (ii) to investigate the impact of FGF23 signaling inhibition during LPS-induced acute inflammation. Our study showed that FGF23 is induced as early as proinflammatory cytokines in response to LPS (Figure 1), followed by upregulation of hepcidin and decrease in serum iron and transferrin saturation (Figure 3A-C), suggesting that the early increase in FGF23 by LPS may contribute to hypoferremia. Previous studies have shown that FGF23 is produced by macrophages and acts locally to stimulate TNF-α expression via activation of FGFR1/α-Klotho signaling.46 Moreover, acute inflammation induced by interleukin-1β (IL-1β) or heat-killed Brucella abortus in mice significantly increased Fgf23 production.12 Thus, since FGF23 397

R. Agoro et al. A

E Figure 5. Effect of fibroblast growth factor 23 (FGF23) signaling disruption on inflammation and hepcidin secretion. C57BL/6J wild-type mice were treated with C-tail FGF23 (1 mg/kg, indicated as FGF23 BL) or vehicle (HEPES buffer) for 8 hours (h). Mice were then challenged with LPS (intraperitoneal 50 mg/kg) or vehicle (0.9% NaCl) for 4 h. (A-D) Quantitative real-time RT-PCR for hepatic expression of (A) IL-6, (B) TNF-α, (C) IL-1β, and (D) hepcidin. Data are expressed as fold change (2-DDCt) relative to housekeeping gene Gapdh. (E) Serum concentration of hepcidin measured by ELISA. Samples were measured in duplicates (n=5-7 per group). Data are represented as mean+standard deviation. All data were analyzed for normality with Shapiro-Wilk test and equivalence of variance using Levene’s test. Because the samples did not show normal distribution, data were aligned in RANK transformation, and confirmed for normality. As the samples showed normal distribution, two-way ANOVA was performed followed by Bonferroni’s multiple comparison test (A-E). Ctl: control (vehicle), ns: not significant, *P<0.05, **P<0.01, ***P<0.001.

and inflammation positively regulate each other, it is possible that one condition may precede the other. FGF23 is a hormone primarily produced by osteocytes that affects many systems in the body. FGF23 is crucial for maintaining normal phosphate and vitamin D homeostasis. Its target organ for these actions is the kidney, where it inhibits phosphate reabsorption by downregulating the type II sodium-phosphate co-transporters (NaPi2a and NaPi2c), and reduces circulating calcitriol levels (1,25(OH)2 vitamin D3) by decreasing renal expression of 1-α-hydroxylase.17-20,47-50 FGF23 levels increase significantly in CKD and are associated with kidney disease progression, greater cardiovascular risk, left ventricular hypertrophy, and mortality.25,51 Phosphate, calcium, vitamin D, and parathyroid hormone (PTH) are known regulators of FGF23. However, in recent years, several other factors, such as inflammation, iron status, and EPO have been identified to both regulate FGF23 and be regulated by it.13,29,52-58 Several mechanisms have been proposed to explain the association between FGF23 and inflammation. It has been shown that inflammation stimulates FGF23 either indirectly through 398

increased hepcidin production, causing iron sequestration in macrophages and subsequent functional iron deficiency, or directly by stimulating Hif1α transcription leading to increased HIF1α which together with HIF1β bind to HIF response element on the Fgf23 promoter inducing its transcription.12 More recently, IL6 was reported to induce Fgf23 promoter activity through a STAT3 response element on the Fgf23 promoter,59 and a distal enhancer upstream of the Fgf23 transcription start site was shown to mediate inflammation-induced expression of Fgf23.60 FGF23 can directly target a number of different cell types and affect a variety of organs. Depending on the target, the mechanisms of FGF23 actions can differ in their requirement for klotho and specific FGFR isoforms, thereby activating different signaling pathways and leading to cell type- and tissue-specific effects of FGF23.61 For example, FGF23 can directly stimulate secretion of proinflammatory cytokines through activation of FGFR4 and PLC /calcineurin/NFAT signaling in the absence of klotho in the liver and bronchial epithelial cells in patients with CKD and chronic obstructive pulmonary disease (COPD), haematologica | 2021; 106(2)

C-FGF23 peptide alleviates hypoferremia

Figure 6. Inhibition of fibroblast growth factor 23 (FGF23) signaling alleviates lipopolysaccharide (LPS)-induced hypoferremia. C57BL/6J wild-type mice were treated with C-tail FGF23 (1 mg/kg, indicated as FGF23 BL) or vehicle (HEPES buffer) for 8 hours (h). Mice were then challenged with LPS (intraperitoneal 50 mg/kg) or vehicle (0.9% NaCl) for 4 h. (A) Serum iron and (B) serum transferrin saturation. (C and D) Quantitative real-time polymerase chain reaction (qRT-PCR) for ferroportin (Fpn) expression in (C) liver and (D) spleen. (E and F) Iron content in (E) liver and (F) spleen. (G and H) qRT-PCR for hepatic expression of (G) ferritin H and (H) lipocalin (Lcn2). Data are expressed as fold change (2-DDCt) relative to housekeeping gene Gapdh. Samples were measured in duplicates (n=5-7 per group). Data are represented as mean+standard deviation. All data were analyzed for normality with Shapiro-Wilk test and equivalence of variance using Levene’s test. For samples with normal distribution, two-way ANOVA with Bonferroni’s multiple comparison test was performed (B, E, F, and G). When the samples did not show normal distribution, they were aligned in RANK transformation, and confirmed for normality. As the samples showed normal distribution, two-way ANOVA was performed with Bonferroni’s multiple comparison test (A, C, D, and H). Ctl: control (vehicle), ns: not significant, *P<0.05, **P<0.01, ***P<0.001.

haematologica | 2021; 106(2)

399

R. Agoro et al. A

Figure 7. Inhibition of fibroblast growth factor 23 (FGF23) signaling increases renal and extra-renal EPO and EpoR mRNA expression under lipopolysaccharide (LPS)-induced hypoferremia. C57BL/6J wild-type mice were treated with C-tail FGF23 (1 mg/kg, indicated as FGF23 BL) or vehicle (HEPES buffer) for 8 hours (h). Mice were then challenged with LPS (intraperitoneal 50 mg/kg) or vehicle (0.9% NaCl) for 4 h. (A-C) Quantitative real-time polymerase chain reaction (qRT-PCR) for renal expression of (A) Epo, (B) EpoR, and (C) Hif2α. (D) Serum concentration of Epo measured by ELISA. (E-H) qRT-PCR for Epo expression in (E) spleen and (F) liver, and EpoR expression in (G) spleen and (H) liver. Data are expressed as fold change (2-DDCt) relative to housekeeping genes Gapdh or Hprt. Samples were measured in duplicates (n=5-7 per group). Data are represented as mean+standard deviation. All data were analyzed for normality by Shapiro-Wilk test and equivalence of variance using Levene’s test. When the samples did not show normal distribution, they were aligned in RANK transformation, and confirmed for normality. As the samples showed normal distribution, two-way ANOVA was performed with Bonferroni’s multiple comparison test (B, F, and H). The samples not in normal distribution were analyzed with non-parametric Kruskal-Wallis test (A, C-E, and G). Ctl: control (vehicle), ns: not significant, *P<0.05, **P<0.01, ***P<0.001.

400

haematologica | 2021; 106(2)

C-FGF23 peptide alleviates hypoferremia

respectively.58,62 The lack of hepatic FGFR4 signaling investigation is a limitation of our study. Therefore, further experiments are required to delineate the hepatic FGFR signaling that contributes to hypoferremia. Mice deficient in Fgf23 (Fgf23-/-) exhibit aberrant bone mineralization along with high serum phosphate, calcium, and vitamin D levels, tissue and vascular calcifications, early lethality within 12 weeks after birth,19 in addition to inflammation and dysregulation of iron homeostasis (data not shown), thus limiting the possibility of using these mice to investigate the impact of FGF23 in response to hypoferremia. Here we evaluated the effect of pharmacologic inhibition of FGF23 signaling using the C-tail FGF23 in a mouse model of LPS-induced hypoferremia. Binding and cell signaling studies have shown that the C-tail of FGF23 can inhibit FGF23 signaling and antagonize its phosphaturic activity in vivo by competing with the intact FGF23 for binding to the FGFR-Klotho complex.27 Moreover, we have previously successfully used the C-tail FGF23 in a CKD mouse model to rescue renal anemia and iron deficiency and attenuate chronic inflammation.29 CKD patients, as well as mice with 5/6 nephrectomy, exhibit elevation in FGF23 levels associated with downregulation of renal Klotho expression. Our longitudinal studies showed that serum phosphate levels significantly increased the first 2 h of LPS treatment (Figure 2A), consistent with previous studies.63 Although the mechanism of this increase is not fully understood, studies have shown that LPS and TNF-α stimulate osteoclastogenesis and bone resorption, suggesting that, in conditions of systemic inflammation, there is a transcellular shift of phosphate from bone cells into the extracellular compartment, leading to hyperphosphatemia.63,64 In response to high serum phosphate levels, circulating FGF23 and Fgf23 expression significantly increased (Figure 1D-G), followed by appropriate decrease in renal Klotho and NaPi2a/c expression (Figure 2B-D) to increase urinary phosphate excretion (Figure 2A) in order to achieve normophosphatemia. We also investigated any potential impact of PTH on phosphate homeostasis and we found that PTH levels are 2.5fold increased 4 h after LPS treatment and returned to basal levels by 6 h after treatment in comparison to vehicle-treated mice (0 h) (data not shown), suggesting the possibility that the rise in serum Pi 1-2 h after LPS may contribute to PTH induction, which in turn would induce phosphaturia. However, our data exclude the possibility that PTH contributes to the increase in FGF23 because the latter reaches peak levels in circulation at the same time point as PTH. Moreover, since vitamin D is a potent stimulator of FGF23, we measured the expression of Cyp27b1, the enzyme responsible for the synthesis of the bioactive form of vitamin D, in the kidney. We observed a 10-fold increase in renal Cyp27b1 expression 2 h after LPS treatment (data not shown), suggesting that circulating levels of vitamin D would be increased. However, FGF23 levels are already significantly elevated by 2 h after LPS, also excluding the possibility that vitamin D contributes to the increase in FGF23. The present study cannot determine whether the increase in FGF23 levels in response to LPS is due to the rise of inflammatory cytokines such as IL-6, TNF-α and IL-1β, the decrease in serum iron, and/or the increase in serum phosphate, all occurring after LPS treatment and prior to the increase in FGF23 levels. In this study we also show that, although the main source of circulating FGF23 is normally the bone, in LPShaematologica | 2021; 106(2)

induced inflammation, organs involved in the immune system, such as the liver and spleen, significantly contribute to the increased circulating FGF23 levels (Figure 1D-G). In addition, blocking FGF23 signaling in LPS conditions resulted in significant reduction in bone (Figure 4A) and bone marrow (data not shown) Fgf23 mRNA levels, resulting in a modest but significant decrease in serum FGF23 levels (Figure 4B and C). However, spleen and liver Fgf23 expression were not affected by the cFGF23 in LPStreated animals (data not shown), suggesting that these two tissues contribute to the residual induction of circulating FGF23. This is in agreement with published data showing that spleen significantly contributes to elevated circulating FGF23 levels that rise in response to acute or chronic exposure to LPS.65,66 Moreover, mice with genetic deletion of Fgf23 (Fgf23-/-) or its co-receptor Klotho (Klotho-/-) exhibit a reduced number of splenocytes and thymic atrophy,49,67 suggesting that Fgf23 may play a role in the regulation of the innate and/or acquired immune system. Bone marrow Fgf23 was also significantly increased between 1-12 h after LPS treatment, possibly contributing also to the increased circulating FGF23 levels (Online Supplementary Figure S1B). Fgf23 production in bone was only significantly increased after 4 h of LPS; however, this increase correlates with a similar increase in bone Fgf23 in the saline-injected group at the same time point, suggesting that it may not be LPSdependent (Online Supplementary Figure S1A). Moreover, in line with previously acquired knowledge that LPS-induced hypoferremia is caused through hepcidin-dependent and -independent mechanisms,68,69 our results show that treatment of mice with LPS significantly downregulates renal Epo expression (Figure 3I) following upregulation of pro-inflammatory cytokines and Fgf23 secretion. This is consistent with studies showing that inflammatory cytokines impair erythropoiesis by inhibiting production of Epo, and by directly inhibiting erythroid progenitor cell proliferation and differentiation.35-37,69 The present study is the first to highlight the effect of FGF23 signaling on iron metabolism and its contribution to hypoferremia. Here, we show that at steady-state conditions, disruption of FGF23 signaling results in increased circulating iron levels (Figure 6A) without affecting hepcidin (Figure 5D and E) or inflammatory status (Figure 5AC). This result is likely due to the rise of Epo induced by the C-tail FGF23 (Figure 7D) resulting in increased splenic Fpn mRNA and protein levels (Figure 7D and Online Supplementary Figure S4). Moreover, in response to LPS, mice treated with the C-tail FGF23 resist downregulation of Hif2α, a potent inducer of Epo, resulting in increased circulating EPO and Epo expression (Figure 7). Indeed, it has been reported that Epo directly stimulates intestinal iron absorption by increasing both iron uptake through upregulation of the divalent metal transporter 1 (Dmt1) and iron efflux through upregulation of the Fpn transporter.70 Studies have also shown that HIF2α maintains iron balance by regulating transcription of Dmt1 and Fpn.71,72 Fpn is rapidly induced under low iron conditions via a HIF2α-dependent mechanism and disruption of HIF2α in the intestine inhibits the adaptive increase of intestinal Fpn during iron deficiency in mice.72 Together, the results of these studies suggest that HIF2α activation may induce iron mobilization, whereas inhibition of HIF2α will favor decreased iron absorption.71,72 A limitation of our study is the lack of data on the effect of cFGF23 on intestinal HIF2a, DMT1, and FPN protein levels, which 401

R. Agoro et al.

could have provided evidence for a possible mechanism for the effect of cFGF23 on iron absorption during hypoferremia. Our study also demonstrates that disruption of FGF23 signaling prevents LPS-induced hypoferremia by decreasing hepcidin (Figure 5D and E) and abrogating iron sequestration in liver and spleen (Figure 6). Suppression of hepcidin by cFGF23 in LPS-treated mice appears to be independent of the IL6/STAT3 pathway (Online Supplementary Figure S3). However, cFGF23 significantly decreases induction of Bmp6 transcription by LPS (Online Supplementary Figure S5), suggesting that cFGF23 may be affecting the BMP6/SMAD pathway, an important transcriptional pathway for hepcidin regulation. In support of this, our data show that splenic expression of erythroferrone (Erfe), a negative regulator of hepcidin, is decreased in mice treated with LPS, whereas mice treated with the cFGF23 under LPS conditions have significantly higher Erfe expression (Online Supplementary Figure S6). Erfe has been shown to suppress BMP/SMAD signaling in vivo and in vitro, and to inhibit hepcidin induction by BMP6.73 Thus, our data suggest that cFGF23 decreases hepcidin induction by LPS by increasing Erfe, and possibly suppressing the BMP6/SMAD pathway. We recently described that inhibition of FGF23 signaling using 10 mg/kg of the C-tail FGF23 rescues anemia and iron deficiency in a CKD mouse model by stimulating erythropoiesis, increasing serum iron and ferritin levels, and attenuating chronic inflammation.29 In this experimental condition, where we observed highly elevated FGF23 levels and absolute iron deficiency characterized by a decrease in circulating iron29 and iron stores (Online Supplementary Figure S7), the C-tail FGF23 significantly reduced circulating FGF23 levels and improved the impaired iron distribution by elevating circulating iron and ferritin levels29 and increasing liver iron content (Online Supplementary Figure S7). In the present study, we show that even by decreasing the concentration

References 1. Gozzelino R, Arosio P. Iron homeostasis in health and disease. Int J Mol Sci. 2016;17(1). 2. Layoun A, Huang H, Calve A, Santos MM. Toll-like receptor signal adaptor protein MyD88 is required for sustained endotoxin-induced acute hypoferremic response in mice. Am J Pathol. 2012;180(6):2340-2350. 3. Schmidt PJ. Regulation of iron metabolism by hepcidin under conditions of inflammation. J Biol Chem. 2015;290(31):1897518983. 4. Ganz T. Hepcidin, a key regulator of iron metabolism and mediator of anemia of inflammation. Blood. 2003;102(3):783-788. 5. Kemna E, Pickkers P, Nemeth E, van der Hoeven H, Swinkels D. Time-course analysis of hepcidin, serum iron, and plasma cytokine levels in humans injected with LPS. Blood. 2005;106(5):1864-1866. 6. Nemeth E, Valore EV, Territo M, Schiller G, Lichtenstein A, Ganz T. Hepcidin, a putative mediator of anemia of inflammation, is a type II acute-phase protein. Blood. 2003;101(7):2461-2463. 7. Nemeth E, Tuttle MS, Powelson J, et al. Hepcidin regulates cellular iron efflux by binding to ferroportin and inducing its internalization. Science. 2004;306(5704):

402

of the C-tail FGF23 to 1 mg/kg, Epo is still induced at the steady state and after LPS (Figure 7D), and this concentration of the C-tail FGF23 is also sufficient to alleviate LPS-induced functional iron deficiency (Figure 6). However, this concentration is not sufficient to attenuate hepatic IL-6 and IL-1α induction (Figure 5A and C), but it did significantly decrease TNF-α expression (Figure 5B). This is consistent with other studies showing that recombinant FGF23 and increased production of endogenous FGF23, such as in Hyp mice, stimulate TNF-α expression in macrophages.46 Thus, inhibition of FGF23 signaling represents a therapeutic target against absolute and functional iron deficiency pathologies. Taken together, our results show for the first time that FGF23 signaling inhibition alleviates LPS-induced acute hypoferremia. Disclosures No conflicts of interests to disclose. Contributions RA developed and directed the project, performed research and data analysis, and wrote the manuscript; MYP and CLH performed research, data and statistical analyses; SJ and AG performed research; GC purified and crystallized the C-tail FGF23 and analyzed its crystal structure; MM designed, refined, analyzed, and interpreted the crystal structure of the C-tail FGF23; DS developed and directed the project, performed data analysis, supervised the study, and wrote the manuscript. Acknowledgments The authors would like to thank Anna Montagna and Elena Healy for technical assistance. Funding This work was supported by funds from the US Department of Defense (W81XWH-16-1-0598) to DS, and the NIH (NIDCR) to MM (DE 13686).

2090-2093. 8. Ganz T, Nemeth E. Iron homeostasis in host defence and inflammation. Nat Rev Immunol. 2015;15(8):500-510. 9. Weiss G. Iron metabolism in the anemia of chronic disease. Biochim Biophys Acta. 2009;1790(7):682-693. 10. Babitt JL, Lin HY. Mechanisms of anemia in CKD. J Am Soc Nephrol. 2012;23(10):16311634. 11. Wolf M. Update on fibroblast growth factor 23 in chronic kidney disease. Kidney Int. 2012;82(7):737-747. 12. David V, Martin A, Isakova T, Spaulding C, Qi L, Ramirez V, et al. Inflammation and functional iron deficiency regulate fibroblast growth factor 23 production. Kidney Int. 2016;89(1):135-146. 13. Farrow EG, Yu X, Summers LJ, et al. Iron deficiency drives an autosomal dominant hypophosphatemic rickets (ADHR) phenotype in fibroblast growth factor-23 (Fgf23) knock-in mice. Proc Natl Acad Sci U S A. 2011;108(46):E1146-55. 14. White KE, Carn G, Lorenz-Depiereux B, Benet-Pages A, Strom TM, Econs MJ. Autosomal-dominant hypophosphatemic rickets (ADHR) mutations stabilize FGF-23. Kidney Int. 2001;60(6):2079-2086. 15. Shimada T, Mizutani S, Muto T, et al. Cloning and characterization of FGF23 as a

causative factor of tumor-induced osteomalacia. Proc Natl Acad Sci U S A. 2001; 98(11):6500-6505. 16. Perwad F, Zhang MY, Tenenhouse HS, Portale AA. Fibroblast growth factor 23 impairs phosphorus and vitamin D metabolism in vivo and suppresses 25-hydroxyvitamin D-1alpha-hydroxylase expression in vitro. Am J Physiol Renal Physiol. 2007;293(5):F1577-1583. 17. Shimada T, Hasegawa H, Yamazaki Y, et al. FGF-23 is a potent regulator of vitamin D metabolism and phosphate homeostasis. J Bone Miner Res. 2004;19(3):429-435. 18. Sitara D, Kim S, Razzaque MS, et al. Genetic evidence of serum phosphateindependent functions of FGF-23 on bone. PLoS Genet. 2008;4(8):e1000154. 19. Sitara D, Razzaque MS, Hesse M, et al. Homozygous ablation of fibroblast growth factor-23 results in hyperphosphatemia and impaired skeletogenesis, and reverses hypophosphatemia in Phex-deficient mice. Matrix Biol. 2004;23(7):421-432. 20. Sitara D, Razzaque MS, St-Arnaud R, et al. Genetic ablation of vitamin D activation pathway reverses biochemical and skeletal anomalies in Fgf-23-null animals. Am J Pathol. 2006;169(6):2161-2170. 21. Eicher EM, Southard JL, Scriver CR, Glorieux FH. Hypophosphatemia: mouse

haematologica | 2021; 106(2)

C-FGF23 peptide alleviates hypoferremia

model for human familial hypophosphatemic (vitamin D-resistant) rickets. Proc Natl Acad Sci U S A. 1976;73(12):4667-4671. 22. Loffing J, Lotscher M, Kaissling B, et al. Renal Na/H exchanger NHE-3 and Na-PO4 cotransporter NaPi-2 protein expression in glucocorticoid excess and deficient states. J Am Soc Nephrol. 1998;9(9):1560-1567. 23. Strom TM, Francis F, Lorenz B, et al. Pex gene deletions in Gy and Hyp mice provide mouse models for X-linked hypophosphatemia. Hum Mol Genet. 1997;6(2):165171. 24. Martin A, David V, Quarles LD. Regulation and function of the FGF23/klotho endocrine pathways. Physiol Rev. 2012;92(1):131-155. 25. Faul C, Amaral AP, Oskouei B, et al. FGF23 induces left ventricular hypertrophy. J Clin Invest. 2011;121(11):4393-4408. 26. Bacic D, Capuano P, Baum M, et al. Activation of dopamine D1-like receptors induces acute internalization of the renal Na+/phosphate cotransporter NaPi-IIa in mouse kidney and OK cells. Am J Physiol Renal Physiol. 2005;288(4):F740-F747. 27. Goetz R, Nakada Y, Hu MC, et al. Isolated C-terminal tail of FGF23 alleviates hypophosphatemia by inhibiting FGF23FGFR-Klotho complex formation. Proc Natl Acad Sci U S A. 2010;107(1):407-412. 28. Moe OW, Tejedor A, Levi M, Seldin DW, Preisig PA, Alpern RJ. Dietary NaCl modulates Na(+)-H+ antiporter activity in renal cortical apical membrane vesicles. Am J Physiol. 1991;260(1 Pt 2):F130-137. 29. Agoro R, Montagna A, Goetz R, et al. Inhibition of fibroblast growth factor 23 (FGF23) signaling rescues renal anemia. FASEB J. 2018;32(7):3752-3764. 30. Agoro R, Taleb M, Quesniaux VFJ, Mura C. Cell iron status influences macrophage polarization. PLoS One. 2018; 13(5):e0196921. 31. Barry M, Sherlock S. Measurement of liveriron concentration in needle-biopsy specimens. Lancet. 1971;1(7690):100-103. 32. Pigeon C, Ilyin G, Courselaud B, et al. A new mouse liver-specific gene, encoding a protein homologous to human antimicrobial peptide hepcidin, is overexpressed during iron overload. J Biol Chem. 2001;276(11):7811-7819. 33. Rishi G, Wallace DF, Subramaniam VN. Hepcidin: regulation of the master iron regulator. Biosci Rep. 2015;35(3). 34. Flo TH, Smith KD, Sato S, et al. Lipocalin 2 mediates an innate immune response to bacterial infection by sequestrating iron. Nature. 2004;432(7019):917-21. 35. Faquin WC, Schneider TJ, Goldberg MA. Effect of inflammatory cytokines on hypoxia-induced erythropoietin production. Blood. 1992;79(8):1987-1994. 36. Frede S, Fandrey J, Pagel H, Hellwig T, Jelkmann W. Erythropoietin gene expression is suppressed after lipopolysaccharide or interleukin-1 beta injections in rats. Am J Physiol. 1997;273(3 Pt 2):R1067-1071. 37. Vannucchi AM, Grossi A, Rafanelli D, Statello M, Cinotti S, Rossi-Ferrini P. Inhibition of erythropoietin production in vitro by human interferon gamma. Br J Haematol. 1994;87(1):18-23. 38. Goetz R, Beenken A, Ibrahimi OA, et al. Molecular insights into the klotho-dependent, endocrine mode of action of fibroblast growth factor 19 subfamily members. Mol Cell Biol. 2007;27(9):3417-3428. 39. Rivera S, Nemeth E, Gabayan V, Lopez MA, Farshidi D, Ganz T. Synthetic hep-

haematologica | 2021; 106(2)

40. 41.

42.

43. 44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

cidin causes rapid dose-dependent hypoferremia and is concentrated in ferroportincontaining organs. Blood. 2005;106(6): 2196-2199. Haase VH. Regulation of erythropoiesis by hypoxia-inducible factors. Blood Rev. 2013; 27(1):41-53. Ward RJ, Crichton RR, Taylor DL, Della Corte L, Srai SK, Dexter DT. Iron and the immune system. J Neural Transm (Vienna). 2011;118(3):315-328. Hennigar SR, McClung JP. Nutritional immunity: starving pathogens of trace minerals. Am J Lifestyle Med. 2016;10(3):170173. Ganz T. Hepcidin and iron regulation, 10 years later. Blood. 2011;117(17):4425-4433. Agoro R, Mura C. Inflammation-induced up-regulation of hepcidin and down-regulation of ferroportin transcription are dependent on macrophage polarization. Blood Cells, Molecules and Diseases. 2016;61:16-25. Hanudel MR, Laster M, Salusky IB. Nonrenal-related mechanisms of FGF23 pathophysiology. Curr Osteoporos Rep. 2018;16(6):724-729. Han X, Li L, Yang J, King G, Xiao Z, Quarles LD. Counter-regulatory paracrine actions of FGF-23 and 1,25(OH)2 D in macrophages. FEBS Lett. 2016;590(1):53-67. Babitt JL, Sitara D. Crosstalk between fibroblast growth factor 23, iron, erythropoietin, and inflammation in kidney disease. Curr Opin Nephrol Hypertens. 2019;28(4):304-310. Gattineni J, Bates C, Twombley K, et al. FGF23 decreases renal NaPi-2a and NaPi-2c expression and induces hypophosphatemia in vivo predominantly via FGF receptor 1. Am J Physiol Renal Physiol. 2009; 297(2):F282-F291. Razzaque MS, Sitara D, Taguchi T, StArnaud R, Lanske B. Premature aging-like phenotype in fibroblast growth factor 23 null mice is a vitamin D-mediated process. FASEB J. 2006;20(6):720-722. Shimada T, Kakitani M, Yamazaki Y, et al. Targeted ablation of Fgf23 demonstrates an essential physiological role of FGF23 in phosphate and vitamin D metabolism. J Clin Invest. 2004;113(4):561-568. Leaf DE, Siew ED, Eisenga MF, et al. Fibroblast growth factor 23 associates with death in critically ill patients. Clin J Am Soc Nephrol. 2018;13(4):531-541. Clinkenbeard EL, Hanudel MR, Stayrook KR, et al. Erythropoietin stimulates murine and human fibroblast growth factor-23, revealing novel roles for bone and bone marrow. Haematologica. 2017; 102(11):e427-e430. Coe LM, Madathil SV, Casu C, Lanske B, Rivella S, Sitara D. FGF-23 is a negative regulator of prenatal and postnatal erythropoiesis. J Biol Chem. 2014;289(14):97959810. Daryadel A, Bettoni C, Haider T, et al. Erythropoietin stimulates fibroblast growth factor 23 (FGF23) in mice and men. Pflugers Arch. 2018;470(10):1569-1582. Hanudel MR, Eisenga MF, Rappaport M, et al. Effects of erythropoietin on fibroblast growth factor 23 in mice and humans. Nephrol Dial Transplant. 2019;34(12):20572065. Imel EA, Peacock M, Gray AK, Padgett LR, Hui SL, Econs MJ. Iron modifies plasma FGF23 differently in autosomal dominant hypophosphatemic rickets and healthy

humans. J Clin Endocrinol Metab. 2011;96(11):3541-3549. 57. Nam KH, Kim H, An SY, et al. Circulating fibroblast growth factor-23 levels are associated with an increased risk of anemia development in patients with nondialysis chronic kidney disease. Sci Rep. 2018;8 (1):7294. 58. Singh S, Grabner A, Yanucil C, et al. Fibroblast growth factor 23 directly targets hepatocytes to promote inflammation in chronic kidney disease. Kidney Int. 2016;90(5):985-996. 59. Durlacher-Betzer K, Hassan A, Levi R, Axelrod J, Silver J, Naveh-Many T. Interleukin-6 contributes to the increase in fibroblast growth factor 23 expression in acute and chronic kidney disease. Kidney Int. 2018;94(2):315-325. 60. Onal M, Carlson AH, Thostenson JD, et al. A novel distal enhancer mediates inflammation-, PTH-, and early onset murine kidney disease-induced expression of the mouse Fgf23 gene. JBMR Plus. 2018;2(1):32-47. 61. Richter B, Faul C. FGF23 actions on target tissues-with and without Klotho. Front Endocrinol (Lausanne). 2018;9:189. 62. Krick S, Grabner A, Baumlin N, et al. Fibroblast growth factor 23 and Klotho contribute to airway inflammation. Eur Respir J. 2018;52(1). 63. Ikeda S, Yamamoto H, Masuda M, et al. Downregulation of renal type IIa sodiumdependent phosphate cotransporter during lipopolysaccharide-induced acute inflammation. Am J Physiol Renal Physiol. 2014;306(7):F744-F750. 64. Abu-Amer Y, Ross FP, Edwards J, Teitelbaum SL. Lipopolysaccharide-stimulated osteoclastogenesis is mediated by tumor necrosis factor via its P55 receptor. J Clin Invest. 1997;100(6):1557-1565. 65. Bansal S, Friedrichs WE, Velagapudi C, et al. Spleen contributes significantly to increased circulating levels of fibroblast growth factor 23 in response to lipopolysaccharideinduced inflammation. Nephrol Dial Transplant. 2017;32(6):960-968. 66. Masuda Y, Ohta H, Morita Y, et al. Expression of Fgf23 in activated dendritic cells and macrophages in response to immunological stimuli in mice. Biol Pharm Bull. 2015;38(5):687-693. 67. Okada S, Yoshida T, Hong Z, et al. Impairment of B lymphopoiesis in precocious aging (klotho) mice. Int Immunol. 2000;12(6):861-871. 68. Deschemin JC, Vaulont S. Role of hepcidin in the setting of hypoferremia during acute inflammation. PLoS One. 2013;8(4):e61050. 69. Wang CY, Babitt JL. Hepcidin regulation in the anemia of inflammation. Curr Opin Hematol. 2016;23(3):189-197. 70. Srai SK, Chung B, Marks J, et al. Erythropoietin regulates intestinal iron absorption in a rat model of chronic renal failure. Kidney Int. 2010;78(7):660-667. 71. Mastrogiannaki M, Matak P, Keith B, Simon MC, Vaulont S, Peyssonnaux C. HIF-2alpha, but not HIF-1alpha, promotes iron absorption in mice. J Clin Invest. 2009;119(5):1159-1166. 72. Taylor M, Qu A, Anderson ER, et al. Hypoxia-inducible factor-2alpha mediates the adaptive increase of intestinal ferroportin during iron deficiency in mice. Gastroenterology. 2011;140(7):2044-2055. 73.Arezes J, Foy N, McHugh K, et al. Erythroferrone inhibits the induction of hepcidin by BMP6. Blood. 2018;132(14): 1473-1477.

403

ARTICLE Ferrata Storti Foundation

Haematologica 2021 Volume 106(2):404-411

Hematopoiesis

A gain-of-function RAC2 mutation is associated with bone marrow hypoplasia and an autosomal dominant form of severe combined immunodeficiency Chantal Lagresle-Peyrou,1,2,3 Aurélien Olichon,4 Hanem Sadek,3 Philippe Roche,5 Claudine Tardy,4 Cindy Da Silva,3 Alexandrine Garrigue,3 Alain Fischer,2,6,7 Despina Moshous,6 Yves Collette,5 Capucine Picard,2,6,8,9 Jean Laurent Casanova,2,6,10,11,12 Isabelle André1,2 and Marina Cavazzana1,2,3

Laboratory of Human Lymphohematopoiesis, INSERM UMR 1163, Imagine Institute, F-75015 Paris, France; 2Paris Descartes University – Sorbonne Paris Cité, Imagine Institute UMR1163, F-75015 Paris, France; 3Biotherapy Clinical Investigation Center, Groupe Hospitalier Universitaire Ouest, Assistance Publique-Hôpitaux de Paris, INSERM CIC 1416, F-75015 Paris, France; 4Cancer Research Center of Toulouse, CRCT, University of Toulouse, UPS, INSERM U1037, F-31037 Toulouse, France; 5Marseille Cancer Research Center, CRCM, Aix Marseille University, Institut Paoli-Calmettes, CNRS, INSERM, Team ISCB, F-13273 Marseille, France; 6Department of Pediatric Immunology, Hematology and Rheumatology, Necker-Enfants Malades University Hospital, APHP, F75015 Paris, France; 7College de France, F-75231 Paris, France; 8Study Center for Primary Immunodeficiencies, Assistance Publique–Hôpitaux de Paris (AP-HP), NeckerEnfants Malades University Hospital, F-75015 Paris, France; 9Laboratory of Lymphocyte Activation and Susceptibility to EBV, INSERM UMR 1163, Imagine Institute, F-75015 Paris, France; 10Laboratory of Human Genetics of Infectious Diseases, Necker Branch INSERM UMR 1163, Imagine Institute, F-75015 Paris, France; 11St. Giles Laboratory of Human Genetics of Infectious Diseases, Rockefeller Branch, Rockefeller University, New York, NY, USA and 12Howard Hughes Medical Institute, New York, NY, USA 1

ABSTRACT

Correspondence: MARINA CAVAZZANA m.cavazzana@aphp.fr CHANTAL LAGRESLE-PEYRO chantal.lagresle@inserm.fr Received: June 20, 2019. Accepted: January 8, 2020. Pre-published: January 9, 2020.

vere combined immunodeficiencies (SCID) constitute a heterogeneous group of life-threatening genetic disorders that typically present in the first year of life. They are defined by the absence of autologous T cells and the presence of an intrinsic or extrinsic defect in the B-cell compartment. In three newborns presenting with frequent infections and profound leukopenia, we identified a private, heterozygous mutation in the RAC2 gene (p.G12R). This mutation was de novo in the index case, who had been cured by hematopoietic stem cell transplantation but had transmitted the mutation to her sick daughter. Biochemical assays showed that the mutation was associated with a gain of function. The results of in vitro differentiation assays showed that RAC2 is essential for the survival and differentiation of hematopoietic stem/progenitor cells. Therefore, screening for RAC2 gain-offunction mutations should be considered in patients with a SCID phenotype and who lack a molecular diagnosis.

https://doi.org/10.3324/haematol.2019.230250

404

Introduction Severe combined immunodeficiencies (SCID) are inherited primary immunodeficiencies characterized by a profound impairment of T-cell development and an intrinsic or functional defect in the B-cell compartment. Life-threatening SCID are the most severe primary immunodeficiencies; in the absence of treatment, SCID lead to death within the first year of life. The only curative treatments are allogeneic hematopoietic stem cell transplantation (HSCT) or, for two types of SCID, autologous gene-modified HSCT.1 At present, all the SCID with a known molecular defect have an autosomal recessive pattern of inheritance or, in one case, Xlinked inheritance. Less than 5% of SCID patients do not have a molecular diagnosis.2,3 Reticular dysgenesis (RD) is the SCID form with the earlier clinical presentation (i.e., a few days after birth), due to the absence of both neutrophils and T cells. In haematologica | 2021; 106(2)

Autosomal dominant form of SCID due to gain-of-function RAC2 mutation

addition to the hematopoietic defect, patients with RD have bilateral sensorineural deafness.4 We and others have previously demonstrated that bi-allelic mutations in the gene coding for adenylate kinase 2 (AK2) are involved in the RD phenotype by disrupting the ATP production required to sustain the survival and differentiation of hematopoietic stem/progenitor cells (HSPC). The high AK2 expression levels in the inner ear may also account for the sensorineural hearing loss observed in patients with RD.5-7 We have identified an autosomal dominant (AD) missense mutation in the RAC2 gene (coding for Ras-related C3 botulinum toxin substrate 2 [RAC2]) in three SCID patients whose clinical presentation overlaps with the RD SCID but who lack AK2 mutations and deafness. RAC2 belongs to the Rac subfamily of RHO small GTPases. In the inactive GDP-bound state, RAC2 is located in the cytosol. Upon stimulation and activation by guanine nucleotide exchange factors, the active RAC2-GTP-bound form translocates to the plasma membrane. There, active RAC2 protein triggers various signaling pathways until the GTP is hydrolyzed following binding to GTPase-activating proteins (GAP).8-10 Unlike the other members of the Rac subfamily (RAC1 and RAC3), RAC2 is mostly expressed on hematopoietic cells.11,12 Using biochemical and in vitro differentiation assays, we demonstrated that the RAC2 mutation was closely related to an impairment in cell differentiation capacity and defects in cellular and mitochondrial networks. Taken as a whole, our data demonstrate that a dominant gain-of-function (GOF) mutation in the GDP/GTP binding site of the RAC2 protein inhibits HSPC differentiation and leads to a severe AD form of SCID with a clinical presentation of RD.

300 mM NaCl; 1% Nonidet P-40) supplemented with protease and phosphatase inhibitors. Cell extracts were separated by SDSPAGE, blotted, and stained with anti-RAC2 (ab154711, Abcam) or anti-GAPDH (SC-32233, Santa Cruz, CA, USA) antibodies. After staining with an HRP-conjugated secondary antibody, the immunoblot was developed and quantified in an Odyssey system (LiCor).

RAC2 activation assays and immunoblotting with the HEK293T cell line The HEK293T cells were cultured overnight prior to transduction with the appropriate lentiviral supernatant (Online Supplementary Methods). After two days of culture, the cell pellets were frozen in liquid nitrogen. Levels of activated RAC2 were determined using the G-LISA® RAC Activation Assay Biochem KitTM (#BK125, Cytoskeleton Inc.) according to the manufacturer’s instructions, except that RAC2 monoclonal specific antibody (AT2G11, sc-517424 Santa Cruz Biotechnology, Inc.) and HRPconjugated anti-mouse antibody (#1721011, Bio-Rad) were used to detect the amount of captured active RAC2 (details in the Online Supplementary Methods).

Homology modeling Three-dimensional homology models were built for the G12R mutant of human RAC2 (Uniprot P15153) using MODELLER software (version 20, Webb and Sali, 2016, 2017). The crystal structure of wild-type (WT) human RAC2 (PDB code: 1DS6) was used as a template.

Statistical analysis For all analyses, three or more independent experiments were performed. Data are reported as the mean±standard error of the mean (SEM). Two-tailed, unpaired t-tests were performed using Prism 4 software (GraphPad). P<0.05 was considered statistically significant.

Methods Patients and human cord blood samples The study was conducted in accordance with French legislation and the principles of the Declaration of Helsinki. Informed consent was obtained from the patients’ parents or legal guardians, and the study protocol was approved by the regional independent ethics committee and the French Ministry of Research (DC 20142272/2015/ DC-2008-329). Primary fibroblasts were obtained from skin biopsies. Human cord blood (CB) samples eligible for research purposes were obtained from the Cord Blood Bank at St Louis Hospital (Paris, France; authorization 2014/09/23). Mononuclear cells were isolated by density separation on Lymphoprep (Abcys). CD34+ HSPC were sorted magnetically using the autoMACSpro separator, and cell purity was checked with a MACSQuant analyzer (Miltenyi Biotec) before transduction and culture (Online Supplementary Methods).

Gene sequencing and gene expression in the patient's fibroblasts Genomic DNA was isolated by phenol/chloroform extraction from primary fibroblasts (from P1, P2 and P3) or peripheral blood mononuclear cells (PBMC, from P3’s father). Whole-exome sequencing was performed with an Illumina TruSeq exome enrichment kit (Illumina), using 100 bp paired-end reads. Eightyfive percent of the target regions were observed with a coverage >20×. Variant calling was not based on a particular genetic model. The mutation was confirmed by Sanger sequencing. For western blot analyses, cells were lysed in Tris buffer (20 mM Tris, pH 7.9; haematologica | 2021; 106(2)

Results A missense mutation (G12R) in the RAC2 GTPase protein leads to a severe combined immunodeficiencies phenotype A few members of our cohort of patients with a SCID phenotype do not yet have a molecular diagnosis. Three patients (Figure 1A) from two unrelated kindreds (P1 from one kindred, and mother and daughter P2 and P3 from another kindred) presented with all the immunological and clinical features of RD other than deafness. As observed in patients with RD, the complete absence of neutrophils was responsible for the occurrence of severe infections (Table 1) earlier than is usually observed in other forms of SCID.3 The analysis of other blood cell lineages highlighted the absence of T and B lymphocytes and circulating monocytes. The T-cell receptor excision circle (TREC) level was very low (<5 copies/mL in P3, the only patient tested, vs. >34 copies for a control), reflecting defective T-cell maturation. Conversely, the platelet count was normal and (in P3 only) the hemoglobin level was slightly below normal (Table 1). Overall, the peripheral cell count profile was in line with the severe hypoplastic content of the patients’ bone marrow (BM) (Table 1 and Online Supplementary Table S1). HSCT performed soon after birth was successful in P1 (who was previously described as P4 by André-Schmutz et al.13) and P2, with full donor immune reconstitution, demonstrating that the 405

C. Lagresle-Peyrou et al. A

Figure 1. Identification and effects of the p.G12R RAC2 mutation on the GTPase activity of RAC2. (A) Pedigrees of the three patients from two unrelated kindred. Black boxes and black circles respectively represent the affected male (P1) and affected females (P2, P3). White boxes and circles respectively represent unaffected males and females. Arrows represent the probands and a double horizontal bar represents consanguinity. (B) A representative electropherogram of RAC2 DNA sequencing for control cells and patient cells, showing the c.34G>A mutation. (C) A representative immunoblot of RAC2 protein expression in lysates from control fibroblasts (Ctrl) and fibroblasts derived from the affected individuals (P1, P2, and P3). The loading control corresponds to GAPDH expression. (D) 3D models of the RAC2 G12R mutant. The structures are shown with GDP (left panel) or GTP (right panel) in the G1 binding pocket. For each model, a close-up view of the GDP/GTP binding pocket (dotted brown circle) is shown below the overall view. The figure was generated with the pymol program (www.pymol.org/). (E) HEK293T cells were either not transduced (NT, control) or transduced with a lentiviral empty vector (WPI) containing the wild-type (WT) form of RAC2 cDNA, the mutated form described here (G12R) or (as a positive control) the constitutively activated GTP-bound RAC2 form (G12V). Two days after transduction, cells were recovered for analysis using the G-LISA assay (15 µg of total protein per well) for the quantification of the GTP-bound RAC2 form (RAC2 GTP). The results come from three independent experiments, and the table below the graph represents the mean of the percentage of GFP expressing cells (GFP+) in the three independent experiments. ***P<0.001; ****P<0.0001.

inherited defect was intrinsic and not extrinsic (i.e., not micro-environmental) (Table 1). No other clinical symptoms have been reported so far for these patients (12 and 21 years after HSCT for P1 and P2, respectively), both of whom have achieved a good quality of life. By performing whole-genome sequencing on the patients’ fibroblasts (the only available cell source, since the patients underwent HSCT early in life), we identified a heterozygous missense mutation (c.34G>A, p.G12R) in the RAC2 gene that was absent in P3’s father, the only relative from whom we could obtain a DNA sample (Figure 1A and B, and Online Supplementary Figure S1A). This variant (confirmed by Sanger sequencing) was not annotated in our in-house database (n=14,154 samples in the cohort) or in the human Genome Aggregation Database, and was predicted to be deleterious by four different in silico prediction software tools, including Combined Annotation Dependent Depletion (Online Supplementary Figure S1B). The mutation was, therefore, considered to be diseasecausing. The presence of the same clinical phenotype and the same RAC2 G12R missense mutation in P2 and her daughter P3 highlighted an AD inheritance pattern. 406

The p.G12R missense mutation is located in the G1 box, a highly conserved guanine nucleotide binding region.14 It is noteworthy that this mutation differed from the loss-offunction (LOF) or gain-of-function (GOF) mutations (Online Supplementary Figure S1A) previously reported as being responsible for mild neutrophil defects and/or lymphopenia.15-20 Interestingly, the G12R missense mutation did not substantially alter RAC2 protein expression level in patient fibroblasts (Figure 1C and Online Supplementary Figure S1C).

G12R mutation in the GDP/GTP-binding domain disrupts cell homeostasis In order to understand the mutation’s functional impact, we generated a 3D homology model by using in silico models of WT RAC221 as a template. The structural models correspond to the RAC2 G12R mutant form with GDP bound (Figure 1D, left panel) or GTP bound (Figure 1D, right panel) state. In the GTP-bound RAC2 state, the bulky flexible arginine is in tight contact with the terminal phosphate group and may induce steric clashes that prevent GAP proteins from accessing the G1 binding pocket haematologica | 2021; 106(2)

Autosomal dominant form of SCID due to gain-of-function RAC2 mutation

Table 1. Hematologic characteristics and outcomes for the three patients.

Patients (age at presentation)

P1 (3 days)

P2 (10 days)

P3 (9 days)

Infection at birth

Sepsis

Colored amniotic fluid, sepsis/pneumonia 0.3 0.1 0.004 0.07 NE NE 220 13 Hypoplasia HSCT (2 M) Busulfan 8 mg/kg Cyclophosphamide 2,000 mg/kg

Sepsis/meningitis, brain abscesses 0.5 0.5 0 0 0 0 429 10 Hypoplasia 1st T-depleted HSCT (2 M) Busulfan 3.6 mg/kg Fludarabine 160 mg/m2 ALS 5 mg/kg 2nd HSCT (3M) Fludarabine 120 mg/m2 ALS 5 mg/kg

MMFD/f

1st HSCT: MMFD/f 2nd HSCT: MMFD/f Graft failure; death (5.5 M post-HSCT)

White blood cells (x109/L) Lymphocytes (x109/L) B lymphocytes (x109/L) T lymphocytes (x109/L) Monocytes (x109/L) Neutrophils (x109L) Platelets (x109/L) Hemoglobin (g/dL) Bone marrow aspirate HSCT (age at transplantation) conditioning regimen

Donor cells Outcome

0.6 0.4 0.09 0.24 0.01 0,2 248 18 Hypoplasia 1st T-depleted HSCT (3M) Busulfan 8 mg/kg Cyclophosphamide 200 mg/kg ALS 25 mg/kg 2nd T-depleted HSCT (6M) Busulfan 16 mg/kg Cyclophosphamide 200 mg/kg ALS 25 mg/kg 1st HSCT: MMFD/f 2nd HSCT: MMFD/f A/W

GvHD resolved at D120; A/W

Age-matched control value

7-18 3.4-7.6 0.3-2 2.5-5.5 0.1-1.1 1.5-8.5 175-500 12.5-16.6

BM: bone marrow; HSCT: hematopoietic stem cell transplantation; M: months; ALS: antilymphocyte serum; MMFD/f: mismatched family donor/father; GvHD: graft-versus-host disease; A/W: alive and well (full donor immune reconstitution, and no other symptoms/diseases); NE: not evaluated.

(Figure 1D) and thus may impair the GTP hydrolysis rate, as previously demonstrated for other small GTPases.22 To test this model biochemically, we quantified the active GTP-bound RAC2 form (RAC2 GTP) in extracts of HEK293T cells not expressing RAC2 at the basal level. The cells were transduced with an empty lentiviral backbone (WPI) with green fluorescent protein (GFP) as a tracker or the vector containing either the WT form of RAC2 cDNA, the mutated form described here (G12R) or (as positive control) the constitutively activated GTPbound form (G12V).23 A high level of the active GTPbound RAC2 state was observed for both G12V and G12R (Figure 1E), demonstrating that substitution by arginine at position 12 increases the level of active RAC2 protein; hence, the G12R mutation is associated with GOF. To determine how the expression of a constitutively active form of RAC2 impacts cell division and survival, we used in vivo live cell imaging to measure the growth kinetics of primary fibroblasts. The proliferation rate (the change in the percentage of confluence) was significantly slower for P3’s cells than for control cells throughout the culture period (Figure 2A, upper panel). Conversely, the proportion of dead cells (measured using Cytotox green reagent) was significantly higher for the P3 experiment (Figure 2A, lower panel). These results demonstrated that, in vitro, G12R-mutated cells show impaired proliferative capacity and a high mortality rate. To further characterize the cellular defects, we used non-invasive holotomographic live cell imaging to observe the cells’ biological features and organelles. When compared with control fibroblasts, haematologica | 2021; 106(2)

P3’s cells displayed very slow cell dynamics over the 12hour culture period (Online Supplementary Movie V1); this was correlated with the low proliferative response. In P3’s fibroblasts, the nuclear membrane was highly segmented, and the mitochondrial network was disrupted (Figure 2B and Online Supplementary Figure S2A). The alteration of the mitochondrial network was confirmed by confocal microscopy (Online Supplementary Figure S2B); the proportion of fragmented mitochondrial networks was higher in P3’s fibroblasts (35%) than in control fibroblasts (10%).

G12R mutation inhibits hematopoietic stem/progenitor cell proliferation and differentiation To understand the impact of constitutive RAC2 activation on hematopoiesis and given the absence of patient BM samples, CD34+ human cord blood HSPC were transduced with the WPI, WT, G12R or G12V RAC2 cDNAs (with GFP as a tracker) and then cultured with cytokines for 7 days (Figure 3A). The expression of constitutively activated RAC2 forms (G12R and G12V) led to the disappearance of GFP+ transduced cells within 4 days. The GFP+ cell ability to generate reactive hydroxyl radical (a common reactive oxygen species [ROS]) was significantly impaired in the G12R and G12V conditions (quantified as the proportion of “ROS low” cells) (Figure 3B). It is noteworthy that the addition of an ROS inducer was not associated with an elevation in ROS production (data not shown). To analyze this phenomenon in more detail, we evaluated the mitochondrial membrane potential in GFP transduced cells. In the G12R and G12V experiments, 407

C. Lagresle-Peyrou et al. B A

Figure 2. Patient fibroblasts are affected by the G12R mutation. (A) Proliferation kinetics for control fibroblasts (Ctrl) and patient 3’s fibroblasts (P3), using the Incucyte assay. The percentage of confluence (%, upper graph) was measured every 6 hours (h) for 6 days (n=3 wells for both Ctrl and P3). The graph is representative of two independent experiments. Evaluation of cell death (represented as the green fluorescent cells, lower panel) for control (Ctrl) and patient 3’s fibroblasts (P3), with the Cytotox green reagent added 6 h after the start of the culture. Measurements were made every 6 h for 6 days (n=3 wells for Ctrl and P3), using the Incucyte assay. The graph is representative of two independent experiments. (B) Holotomographic live cell imaging was performed with a 3D Cell Explorer microscope. Cultures of control fibroblasts (left panels) and patient 3’s fibroblasts (right panels) are shown. White arrows indicate the nuclear envelope, and brown arrows indicate the mitochondrial network. Scale bar =10 mm, 1 pixel= 0.188 mm. See also Online Supplementary Video M1 and Online Supplementary Figure S2A.

mitochondrial membrane depolarization (quantified as the proportion of “DILC1(5) low” cells) and apoptosis (measured as the proportion of annexin-V-positive cells) were significantly higher than in controls (Figure 3B). It is worthy of note that these alterations were not observed in the GFP- subset (Online Supplementary Figure S1D). Taken as a whole, these findings emphasize the specific correlation between RAC2 mutations at position 12 and impaired cell survival. As RAC2 is highly expressed in human hematopoietic BM subsets and during human thymopoiesis (Online Supplementary Figure S1E), we transduced HSPC with WPI, WT, G12R or G12V RAC2 cDNAs and induced their differentiation along the granulocyte, monocyte and Tlymphoid lineages, all of which were absent in the three patients. After 7 days of culture with granulocyte colony stimulating factor, the percentage of GFP-expressing cells and the GFP+CD15+CD11b+ neutrophil counts were significantly lower in the G12R and G12V conditions than in the WPI or WT conditions (Figure 3C). The mitochondrial membrane potential and ROS production were also low, and were associated with a high level of apoptosis (Figure 3D). Differentiation towards the monocyte lineage and differentiation in colony-forming unit assays gave similar results (data not shown). Upon exposure to the Notch ligand -like-4 (DL4) culture system (which enables the differen408

tiation of HSPC into CD7+ T-cell progenitors in 7 days24), the GFP+ subset count in the G12R and G12V conditions was very low (relative to the WT and WPI conditions), and GFP+CD7+ T-cell progenitors were almost completely absent (Figure 3E). Taken as a whole, our results underline: (i) the key function of the GDP/GTP-bound RAC2 balance in the survival and differentiation of the lympho-myeloid compartment; and (ii) the involvement of the RAC2 signaling pathway in HSPC function. We next compared the impact of G12R mutation on HSPC proliferation with that of previously described RAC2 LOF and GOF missense mutations (p.D57N and p.E62K, respectively).15,16,19 During the 8-day culture, the number of GFP+-transduced HSPC was significantly lower in G12R conditions and slightly lower in E62K conditions (Online Supplementary Figure S3A and B). Unlike the G12R mutation, the D57N and E62K mutations had no impact on ROS production or mitochondrial membrane depolarization (Online Supplementary Figure S3C). Taken as a whole, our observations highlighted the correlation between the high active GTP-bound RAC2 state produced by the G12R mutation and the impairment of HSPC survival and differentiation. These findings fit with the clinical and immunological phenotype observed in our three patients and with the less severe phenotype displayed by patients carrying E62K or D57N mutations. haematologica | 2021; 106(2)

Autosomal dominant form of SCID due to gain-of-function RAC2 mutation

C Figure 3. The G12R mutation inhibits hematopoietic stem/progenitor cell (HSPC) proliferation and differentiation. (A) Proliferation of CD34+ cells in an 8-day culture. Flow cytometry was used to analyze the change over time in the percentage (%) of GFP-expressing cells (GFP+) from day 2 (when the transduction efficiency was measured) until day 8 (the end of the culture). (B) The proportions of “ROS low”, “DILC1(5) low” and annexin-V-positive GFP+ live cells were determined on day 4. All the analyses were performed in the live cell (7-AADnegative) gate. The results are quoted as the mean±standard error of mean (SEM) of four independent experiments. (C) Neutrophil differentiation in a 7-day culture. Flow cytometry was used to analyze the change over time in the percentage (%) of GFP+ cells from day 2 (when the transduction efficiency is measured) until day 8 (end of the culture) and the number of granulocytes (CD11b+CD15+) among the GFP+ live cells. (D) The proportions of “ROS low”, “DILC1(5) low” and annexin-V-positive cells among the GFP+ live cells were analyzed on day 4. Results are quoted as the mean±SEM of three independent experiments. (E) T-cell differentiation in a 7-day culture (n=3). Flow cytometry was used to analyze the change over time in the percentage (%) of GFP+ from day 2 (when the transduction efficiency is measured) until day 7 (end of the culture). The number of T-cell progenitors (CD7+) was evaluated by flow cytometry on day 7 in the GFP+ live cell gate (7-AAD-negative). The results are quoted as the mean±SEM of three independent experiments. *P<0.05; **P<0.01; ***P<0.001. WT: wild-type.

Discussion In three SCID patients with innate and adaptive immune system defects, we identified a heterozygous, dominant missense mutation (p.G12R) in the highly conserved GDP/GTP binding domain of the RAC2 protein. Given the severity of the clinical presentation, the patients had to undergo HSCT in the first few weeks of life. The observation of AD inheritance broadens the clinical spectrum of RD, as we can now distinguish between two forms: a recessive syndromic form associated with deafness and AK2 mutations, and a non-syndromic disease associated with a hypoplastic BM and a specific AD G12R mutation in RAC2. The absence of sensorineural hearing loss in the latter form might be correlated with the predominant expression of RAC2 in hematopoietic lineages. These data fit with the observation whereby in murine models, RAC1 and RAC3 (but not RAC2) are involved in the development of the inner ear;25 however, this has yet to be confirmed in human studies. The HSCT performed as soon as possible after birth in two patients with the G12R mutation restored hematopoiesis and highlights the non-redundant regulatory role of RAC2 in this process. In line with this observation, we demonstrated the drastic effect of the G12R RAC2 mutation on cell proliferation and survival, especially in cord blood HSPC. These results are in agreement with previous reports on the role of RAC2 in the regulation of HSPC10,26 and during T-cell difhaematologica | 2021; 106(2)

ferentiation via the activation of Wnt/β-catenin pathways.9 Furthermore, the constitutive active form of RAC2 G12V has already been described as preventing thymocyte differentiation and inducing apoptosis in the mouse.27 It should be noted that, despite the observation of fibroblast defects in vitro, skin alterations have not been noted for P1 and P2 more than 10 years after HSCT, suggesting that, in vivo, other RHO family members (or perhaps other molecules present in the microenvironment) provide compensatory survival signals to non-hematopoietic cells. The same dichotomy has been reported in RD: fibroblasts from RD patients displayed impairments in apoptosis and mitochondrial function in vitro,5 whereas the patients having undergone HSCT have not reported any skin or tissue lesions (24 years after HSCT, for the case with the longest follow-up). Taken as a whole, these data demonstrate that RAC2 is mandatory at different hematopoietic checkpoints; hence, the unregulated activity of GTP-bound RAC2 driven by the G12R mutation might explain the severe BM hypoplasia and the absence of circulating leukocytes in P1, P2, and P3. In HSPC, RHO GTPases also regulate cell trafficking and ROS production.9,26 In particular, the literature data show that RAC2 mutant cells are associated with unbalanced actin cycling and hyper-segmented nuclei,15,16,18,19,28 two features associated with defective cytokinesis, a process involving RHO GTPases.29-31 The highly segmented nuclear membranes found in P3’s fibroblasts are consistent 409

C. Lagresle-Peyrou et al.

with these data and suggest that the organization of the cytoskeleton is altered in these cells. However, the mutation described herein could not be classified as an actinopathy such as Wiskott-Aldrich syndrome,32 Dock8 deficiency33 or warts, hypogammaglobulinemia, infections, and myelokathexis (WHIM) syndrome.34 Although these diseases are characterized by impaired leukocyte migration, autoimmunity, and/or malignancies, none is associated with BM hypoplasia or a severe SCID phenotype. Therefore, our present observation is the first to have evidenced cytokinesis failure in SCID patients. In cell lines, actin dynamics have been described as a key regulator of the mitochondrial network,35 suggesting that the mitochondrial defects observed in the G12R mutated HSPC might be a consequence of actin dysregulation. In patients bearing D57N or E62K Rac2 mutations, the fact that actin cycling was disrupted in neutrophils15,19 but that mitochondrial membrane depolarization was not evident in the HSPC challenges this argument and suggests that RAC2 could regulate mitochondrial integrity and function through an actin-independent mechanism. This assumption fits with previous reports demonstrating that: (i) RAC2 interacts with SAM50, a mitochondrial transporter involved in the regulation of mitochondrial membrane potential and ROS production:36,37 (ii) unregulated RAC2 activation disrupts ROS production,28,38,39 which is required for the regulation of mitochondrial dynamics.40 Moreover, it is now well established that mitochondrial activity is a key regulator of HSPC homeostasis and functions,41,42 suggesting that the inappropriate RAC2 activation driven by G12 mutations directly affects mitochondria and ROS production. The RAC2 signaling pathways targeted by the G12R mutation are not reported here; however, given that G12R and G12V mutations have the same impact on HSPC, we can reasonably assume that they have the same downstream targets. This assumption is reinforced by the observations in KRAS mutants; both G12R and G12V mutations impaired GTP hydrolysis,22 suggesting that the mutant proteins are catalytically inactive. Interestingly, the G12V GOF RAC2 mutation constitutively activates the phospholipases C (PLC) isoforms PLCβ2 and PLCγ.23,43 These two PLC interact with phosphatidylinositol 3 kinase (PI3K) and protein kinase C (PKC), both of which are involved in various signaling pathways (including the ones that regulate HSPC homeostasis).44,45 The comparison of RAC2 mutants also highlighted the complexity of an unbalanced RAC2 network, as the D57N and E62K variants had a limited impact on HSPC. This finding might be related to the position of the amino-acid substitution, i.e., inside the GDP/GTP binding pocket (G12R) or in the G3 (D57N) or switch II (E62K) domains (Online Supplementary Figure S1A). In particular, the replacement of glutamate by a lysine at position 62 may impact the GTP bound RAC2 form stability and thus explain the absence of a GTP-bound RAC2 form and a lower level of RAC2 expression, relative to the G12R condition (Online Supplementary Figure S3D). These findings differ from the report of Hsu et al.,19 probably because we did not use the

410

same cells or the same techniques to measure RAC2 GTP activity. However, our conclusions fit with the patients’ clinical phenotype; we can reasonably assume that the highly active, constitutive GTP-bound form of RAC2 driven by the G12R mutation leads to BM failure and a SCID phenotype. In contrast, the transient overactivation of the RAC2-GTP form driven by the E62K RAC2 mutation is associated with lymphopenia19 but not with BM abnormalities.46 Taken as a whole, our findings emphasize that various RAC2 mutations differ in their effects on HSPC. These differences may account for the broad observed spectrum of clinical phenotypes, ranging from neutrophil defects to a severe form of SCID. This type of heterogeneity has already been reported for RAG1 mutations involved in common variable immunodeficiency or SCID phenotype.47 In summary, the p.G12R RAC2 mutation has a drastic impact on the homeostatic regulation of hematopoiesis, which explains the severity of the patients’ clinical and immunological phenotypes. To the best of our knowledge, the present study is the first to have described an AD form of a SCID. Physicians should consider RAC2 gene sequencing for patients with SCID and a clinical presentation of RD. This gene should also be included in newborn screening programs for SCID detection. Finally, future research should seek to characterize the RAC2 signaling pathway and novel downstream targets specifically involved in hematopoietic cell commitment. Disclosures No conflicts of interests to disclose. Contributions CLP and MC supervised and designed the project, CLP analyzed and interpreted data and drafted the manuscript; AO and CT performed the biochemical studies and drafted part of the manuscript; HS, AG and CDS performed some experiments and analyzed data; PR and YC performed the protein homology modelling and drafted part of the manuscript; AF, DM, CP, JLC and MC provided patient care; CP, JLC, IA and MC helped to draft the manuscript. Acknowledgments We especially thank Prof. Stephane Blanche and the nursing staff of the Department of Pediatric Immunology, Hematology and Rheumatology at Necker-Enfants Malades Hospital for patient care. We also thank Gisèle Froment, Didier Nègre, and Caroline Costa for lentiviral vector production (UMS3444/US8, BioSciences Gerland, Lyon), Olivier Pellé for cell sorting (at the flow cytometry core facility, SFR Necker, Paris), Thibault Courtheoux for the 3D cell imager analysis (Nanolive SA), Myriam Chouteau for technical assistance, and Patrick Revy (from the Genome Dynamics in the Immune System group, Imagine Institute, Paris) for fruitful discussions. Funding This study was funded by the French National Institute of Health and Medical Research (INSERM). The work at the CRCT was funded by an “Equipe Labellisée par la Fondation pour la Recherche Médicale” grant.

haematologica | 2021; 106(2)

Autosomal dominant form of SCID due to gain-of-function RAC2 mutation

References 1. Cavazzana M, Bushman FD, Miccio A, André-Schmutz I, Six E. Gene therapy targeting haematopoietic stem cells for inherited diseases: progress and challenges. Nat Rev Drug Discov. 2019;18(6):447-462. 2. Fischer A, Notarangelo LD, Neven B, Cavazzana M, Puck JM. Severe combined immunodeficiencies and related disorders. Nat Rev Dis Primers. 2015;1(1):15061. 3. Picard C, Bobby Gaspar H, Al-Herz W, et al. International Union of Immunological Societies: 2017 Primary Immunodeficiency Diseases Committee Report on Inborn Errors of Immunity. J Clin Immunol. 2018; 38(1):96-128. 4. de Vaal O, Seynhaeve V. Reticular dysgenesia. Lancet. 1959;2(7112):1123-1125. 5. Pannicke U, Hönig M, Hess I, et al. Reticular dysgenesis (aleukocytosis) is caused by mutations in the gene encoding mitochondrial adenylate kinase 2. Nat Genet. 2009;41(1):101-105. 6. Lagresle-Peyrou C, Six EM, Picard C, et al. Human adenylate kinase 2 deficiency causes a profound hematopoietic defect associated with sensorineural deafness. Nat Genet. 2009;41(1):106-111. 7. Six E, Lagresle-Peyrou C, Susini S, et al. AK2 deficiency compromises the mitochondrial energy metabolism required for differentiation of human neutrophil and lymphoid lineages. Cell Death Dis. 2015;6:e1856. 8. Troeger A, Williams DA. Hematopoieticspecific Rho GTPases Rac2 and RhoH and human blood disorders. Exp Cell Res. 2013; 319(15):2375-2383. 9. Nayak RC, Chang K-H, Vaitinadin N-S, Cancelas JA. Rho GTPases control specific cytoskeleton-dependent functions of hematopoietic stem cells. Immunol Rev. 2013;256(1):255-268. 10. Gu Y, Filippi M-D, Cancelas JA, et al. Hematopoietic cell regulation by Rac1 and Rac2 guanosine triphosphatases. Science. 2003;302(5644):445-449. 11. Shirsat NV, Pignolo RJ, Kreider BL, Rovera G. A member of the ras gene superfamily is expressed specifically in T, B and myeloid hemopoietic cells. Oncogene. 1990;5(5):769772. 12. Gu Y, Byrne MC, Paranavitana NC, et al. Rac2, a hematopoiesis-specific Rho GTPase, specifically regulates mast cell protease gene expression in bone marrowderived mast cells. Mol Cell Biol. 2002;22(21):7645-7657. 13. André-Schmutz I, Le Deist F, Hacein-BeyAbina S, et al. Immune reconstitution without graft-versus-host disease after haemopoietic stem-cell transplantation: a phase 1/2 study. Lancet. 2002; 360(9327): 130-137. 14. Olson MF. Rho GTPases, their post-translational modifications, disease-associated mutations and pharmacological inhibitors. Small GTPases. 2018;9(3):203-215. 15. Ambruso DR, Knall C, Abell AN, et al. Human neutrophil immunodeficiency syndrome is associated with an inhibitory Rac2 mutation. Proc Natl Acad Sci U S A. 2000; 97(9):4654-4659. 16. Accetta D, Syverson G, Bonacci B, et al.

haematologica | 2021; 106(2)

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30. 31.

32.

Human phagocyte defect caused by a Rac2 mutation detected by means of neonatal screening for T-cell lymphopenia. J Allergy Clin Immunol. 2011;127(2):535-538.e1-2. Alkhairy OK, Rezaei N, Graham RR, et al. RAC2 loss-of-function mutation in 2 siblings with characteristics of common variable immunodeficiency. J Allergy Clin Immunol. 2015;135(5):1380-1384.e1-5. Lougaris V, Chou J, Beano A, et al. A monoallelic activating mutation in RAC2 resulting in a combined immunodeficiency. J Allergy Clin Immunol. 2019;143(4):16491653.e3. Hsu AP, Donkó A, Arrington ME, et al. Dominant activating RAC2 mutation with lymphopenia, immunodeficiency, and cytoskeletal defects. Blood. 2019; 133(18):1977-1988. Sharapova SO, Haapaniemi E, Sakovich IS, et al. Heterozygous activating mutation in RAC2 causes infantile-onset combined immunodeficiency with susceptibility to viral infections. Clin Immunol. 2019;205:1-5. Scheffzek K, Stephan I, Jensen ON, Illenberger D, Gierschik P. The Rac-RhoGDI complex and the structural basis for the regulation of Rho proteins by RhoGDI. Nat Struct Biol. 2000;7(2):122-126. Hunter JC, Manandhar A, Carrasco MA, Gurbani D, Gondi S, Westover KD. Biochemical and structural analysis of common cancer-associated KRAS mutations. Mol Cancer Res. 2015;13(9):1325-1335. Illenberger D, Walliser C, Strobel J, et al. Rac2 regulation of phospholipase C-beta 2 activity and mode of membrane interactions in intact cells. J Biol Chem. 2003;278(10):8645-8652. Reimann C, Six E, Dal-Cortivo L, et al. Human T-lymphoid progenitors generated in a feeder-cell-Free delta-like-4 culture system promote T-cell reconstitution in NOD/SCID/γc−/− Mice. Stem Cells. 2012; 30(8):1771-1780. Grimsley-Myers CM, Sipe CW, Wu DK, Lu X. Redundant functions of Rac GTPases in inner ear morphogenesis. Dev Biol. 2012; 362(2):172-186. Mulloy JC, Cancelas JA, Filippi M-D, Kalfa TA, Guo F, Zheng Y. Rho GTPases in hematopoiesis and hemopathies. Blood. 2010;115(5):936-947. Lorès P, Morin L, Luna R, Gacon G. Enhanced apoptosis in the thymus of transgenic mice expressing constitutively activated forms of human Rac2GTPase. Oncogene. 1997;15(5):601-605. Tao W, Filippi M-D, Bailey JR, et al. The TRQQKRP motif located near the C-terminus of Rac2 is essential for Rac2 biologic functions and intracellular localization. Blood. 2002;100(5):1679-1688. Canman JC, Lewellyn L, Laband K, et al. Inhibition of Rac by the GAP activity of centralspindlin is essential for cytokinesis. Science. 2008;322(5907):1543-1546. Normand G, King RW. Understanding cytokinesis failure. Adv Exp Med Biol. 2010; 676:27-55. Verma V, Mogilner A, Maresca TJ. Classical and emerging regulatory mechanisms of cytokinesis in animal cells. Biology (Basel). 2019;8(3). Moulding DA, Moeendarbary E, Valon L,

33.

34. 35.

36.

37.

38.

39.

40.

41. 42.

43.

44.

45.

46.

47.

Record J, Charras GT, Thrasher AJ. Excess Factin mechanically impedes mitosis leading to cytokinesis failure in X-linked neutropenia by exceeding Aurora B kinase error correction capacity. Blood. 2012;120(18):38033811. Zhang Q, Dove CG, Hor JL, et al. DOCK8 regulates lymphocyte shape integrity for skin antiviral immunity. J Exp Med. 2014; 211(13):2549-2566. Al Ustwani O, Kurzrock R, Wetzler M. Genetics on a WHIM. Br J Haematol. 2014; 164(1):15-23. Moore AS, Wong YC, Simpson CL, Holzbaur ELF. Dynamic actin cycling through mitochondrial subpopulations locally regulates the fission-fusion balance within mitochondrial networks. Nat Commun. 2016;7:12886. Capala ME, Maat H, Bonardi F, et al. Mitochondrial dysfunction in human leukemic stem/progenitor cells upon loss of RAC2. PLoS One. 2015;10(5):e0128585. Capala ME, Pruis M, Vellenga E, Schuringa JJ. Depletion of SAM50 specifically targets BCR-ABL-expressing leukemic stem and progenitor cells by interfering with mitochondrial functions. Stem Cells Dev. 2016; 25(5):427-437. Pei H, Zhang J, Nie J, et al. RAC2-P38 MAPK-dependent NADPH oxidase activity is associated with the resistance of quiescent cells to ionizing radiation. Cell Cycle. 2017; 16(1):113-122. Nieborowska-Skorska M, Kopinski PK, Ray R, et al. Rac2-MRC-cIII-generated ROS cause genomic instability in chronic myeloid leukemia stem cells and primitive progenitors. Blood. 2012;119(18):4253-4263. Ježek J, Cooper KF, Strich R. Reactive oxygen species and mitochondrial dynamics: the yin and yang of mitochondrial dysfunction and cancer progression. Antioxidants (Basel). 2018;7(1). Bigarella CL, Liang R, Ghaffari S. Stem cells and the impact of ROS signaling. Development. 2014;141(22):4206-4218. Ansó E, Weinberg SE, Diebold LP, et al. The mitochondrial respiratory chain is essential for haematopoietic stem cell function. Nat Cell Biol. 2017;19(6):614-625. Bunney TD, Opaleye O, Roe SM, et al. Structural insights into formation of an active signaling complex between Rac and phospholipase C gamma 2. Mol Cell. 2009; 34(2):223-233. Edling CE, Pedersen M, Carlsson L, Rönnstrand L, Palmer RH, Hallberg B. Haematopoietic progenitor cells utilise conventional PKC to suppress PKB/Akt activity in response to c-Kit stimulation. Br J Haematol. 2007;136(2):260-268. Elich M, Sauer K. Regulation of hematopoietic cell development and function through phosphoinositides. Front Immunol. 2018; 9:931. Smits BM, Lelieveld PHC, Ververs FA, et al. A dominant activating RAC2 variant associated with immunodeficiency and pulmonary disease. Clin Immunol. 2019: 108248. Notarangelo LD, Kim M-S, Walter JE, Lee YN. Human RAG mutations: biochemistry and clinical implications. Nat Rev Immunol. 2016;16(4):234-246.

411

ARTICLE Ferrata Storti Foundation

Hematopoiesis

MicroRNA-21 maintains hematopoietic stem cell homeostasis through sustaining the nuclear factor-κB signaling pathway in mice

Mengjia Hu, Yukai Lu, Hao Zeng, Zihao Zhang, Shilei Chen, Yan Qi, Yang Xu, Fang Chen, Yong Tang, Mo Chen, Changhong Du, Mingqiang Shen, Fengchao Wang, Yongping Su, Song Wang# and Junping Wang#

State Key Laboratory of Trauma, Burns and Combined Injury, Institute of Combined Injury, Chongqing Engineering Research Center for Nanomedicine, College of Preventive Medicine, Third Military Medical University, Chongqing, China

Haematologica 2021 Volume 106(2):412-423

SW and JW contributed equally as co-senior authors

ABSTRACT

Correspondence: JUNPING WANG wangjunping@tmmu.edu.cn Received: August 30, 2019. Accepted: January 20, 2020. Pre-published: January 23, 2020. https://doi.org/10.3324/haematol.2019.236927

412

ong-term hematopoietic output is dependent on hematopoietic stem cell (HSC) homeostasis which is maintained by a complex molecular network in which microRNA play crucial roles, although the underlying molecular basis has not been fully elucidated. Here we show that microRNA-21 (miR-21) is enriched in murine HSC, and that mice with conditional knockout of miR-21 exhibit an obvious perturbation in hematopoiesis. Moreover, significant loss of HSC quiescence and long-term reconstituting ability are observed in the absence of miR-21. Further studies revealed that miR-21 deficiency markedly decreases the nuclear factor kappa B (NF-κB) pathway, accompanied by increased expression of PDCD4, a direct target of miR-21, in HSC. Interestingly, overexpression of PDCD4 in wild-type HSC generates similar phenotypes as those of miR-21-deficient HSC. More importantly, knockdown of PDCD4 can significantly rescue the attenuation of NF-κB activity, thereby improving the defects in miR-21-null HSC. On the other hand, we found that miR-21 is capable of preventing HSC from ionizing radiation-induced DNA damage via activation of the NF-κB pathway. Collectively, our data demonstrate that miR-21 is involved in maintaining HSC homeostasis and function, at least in part, by regulating the PDCD4-mediated NF-κB pathway and provide a new insight into radioprotection of HSC.

Introduction Hematopoiesis is a well-organized developmental process in which hematopoietic stem cells (HSC) can self-renew and differentiate into all kinds of blood cells.1,2 Under steady-state conditions, most adult HSC are retained in a relatively undifferentiated and quiescent state in the bone marrow (BM) microenvironment, which is necessary for sustaining long-term hematopoietic function.3,4 In contrast, the perturbation of HSC homeostasis may result in hematopoietic failure, immunodeficiencies or hematologic malignancies.5,6 It was known that HSC homeostasis is tightly modulated by a complicated molecular network.7 In the past decades, many factors, including cell cycle proteins, transcription factors, surface receptors, epigenetic regulatory factors, metabolic regulators, long non-coding RNA and cytokines, have been found to be involved in the control of HSC homeostasis.1,3,8,9 However, the underlying molecular mechanism is still not completely clear. MicroRNA (miRNA) are small non-coding RNA that participate in a wide range of biological processes by negatively controlling the expression of their target genes through post-transcriptional regulation.10,11 Previous studies have shown that miRNA have distinct expression patterns in the hematopoietic system, and specific miRNA can affect the development of different blood-cell lineages.12,13 In recent years, several miRNA, such as miR-22, miR-29a, miR-125a, miR-126, and the miR132/122 cluster,7,14-17 have been shown to play important roles in HSC biology. miR-21, a well-known short RNA, has multiple physiological functions in mamhaematologica | 2021; 106(2)

MicroRNA-21 maintains HSC homeostasis

mals. It was found that miR-21 is involved in Gfi1-mediated modulation of myelopoiesis and regulates macrophage polarization and anti-inflammatory effects.1820 Besides, mice lacking miR-21 have reduced eosinophil levels in the peripheral blood (PB) and impaired eosinophil colony-forming capacity in the BM.21 Furthermore, another study revealed that miR-21 mediates hematopoietic suppression in myelodysplastic syndromes by targeting smad7, which is a negative regulator of the TGF-β/smad pathway.22 In particular, recent studies have defined miR21 as an onco-miRNA, which is frequently upregulated in many kinds of tumors, including hematologic malignancies.17,23 Taken together, these findings demonstrate that miR-21 plays significant roles in the hematopoietic system. On the other hand, miR-21 was reported to be implicated in the biology of several types of stem cells, including mesenchymal stem cells, cancer stem cells and embryonic stem cells.17,24,25 However, the exact role of miR-21 in HSC populations is largely unclear. In the present study, we first found that mouse HSC express high levels of miR-21 and that targeted deletion of miR-21 leads to abnormal hematopoiesis. Furthermore, miR-21 deficiency significantly impairs the quiescence and long-term reconstituting function of HSC. We demonstrated that miR-21 is involved in the maintenance of HSC homeostasis and function through modulation of the nuclear factor kappa B (NF-κB) pathway by regulating programmed cell death 4 (PDCD4). Of note, miR-21 is also able to mitigate radiation-induced DNA damage in HSC. Collectively, these data indicate a key role for miR-21 in HSC biology and therefore broaden our knowledge of the physiological functions of miR-21.

Methods Animals Normal, wild-type (WT) C57BL/6J mice were purchased from the Institute of Zoology (Chinese Academy of Sciences, Beijing, China). miR-21flox/+ (miR-21fl/+) mice and Mx1-Cre mice were obtained from Shanghai Model Organisms Center (China). miR21fl/fl;Mx1-Cre mice were generated by crossing miR-21fl/fl mice with Mx1-Cre mice. Unless otherwise stated, miR-21 deletion was induced by intraperitoneally injecting 4- to 6-week old miR21fl/fl;Mx1-Cre+ mice with 250 mg of polyinosinic:polycytidylic acid (pIpC) (Sigma, St. Louis, MO, USA) every other day for a total of seven doses. Four weeks after pIpC treatment, these mice were used in subsequent experiments. Identically treated miR21fl/fl;Mx1-Cre- littermates served as controls. Congenic C57Bl/6 SJL CD45.1+ mice were kindly provided by Prof. Jinyong Wang (Guangzhou Institutes of Biomedicine and Health, Chinese Academy of Science, Guangzhou, China). All animal experiments were approved by the Animal Care Committee of The Third Military Medical University (Chongqing, China).

Flow cytometry Single-cell suspensions of BM, spleen and PB were prepared as we described previously.8,26 Detailed information about the antibodies used for flow cytometric analyses is provided in Online Supplementary Table S1. The cell cycle, apoptosis, in vivo 5-bromodeoxyuridine incorporation and intracellular protein staining were analyzed as we reported elsewhere.8 Samples were detected using a FACSverse or FACSCanto flow cytometer (BD Biosciences, San Jose, CA, USA) and data were analyzed using FlowJo10.0 software (TreeStar, San Carlos, CA, USA). Cell sorthaematologica | 2021; 106(2)

ing was performed using a FACSAriaII or FACSAriaIII sorter (BD Biosciences).

Lentiviral transduction The recombinant lentivirus carrying the PDCD4 gene or specific short hairpin RNA (shRNA) against PDCD4, as well as the corresponding negative controls, were obtained from Hanbio Co. Ltd. (Shanghai, China). The lentivirus was then transduced into HSC as we described previously.8 Subsequently, transduced LinSCA1+c-KIT+ (LSK) cells (6×103) were purified with GFP expression through flow cytometry, and transplanted into 10.0 Gy-irradiated CD45.1+ WT recipients along with 5×105 CD45.1+ competitor BM cells. The sequence of sh-PDCD4 is as follows: 5’-GAGCTTGTATATGAAGCCATTGTAA-3’.

Microarray analysis Total RNA was isolated from freshly sorted miR-21fl/fl or miR21D/D LSK cells. After that, samples were hybridized on Mouse Clariom D arrays (Affymetrix, Santa Clara, CA, USA) in triplicate using the procedures described in the User Manuals. Raw data were then normalized using the robust multi-array average. Genes with a fold change in expression >1.4 and a P value <0.05 were defined as differentially expressed and are listed in Online Supplementary Table S3. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment were used to analyze the microarray data. The data have been deposited in the Gene Expression Omnibus database (accession number GSE131603).

Statistical analysis All results were analyzed using GraphPad Prism 6.0 software (La Jolla, CA, USA). Differences in data between two groups and multiple groups were analyzed by a two-tailed Student t-test and one-way analysis of variance, respectively. The survival rates were compared by a log-rank nonparametric test and displayed as Kaplan-Meier survival curves. Unless otherwise stated, data were obtained from at least three independent experiments. P values <0.05 were defined as statistically significant.

Results miR-21 is enriched in hematopoietic stem cells and its conditional ablation skews hematopoietic differentiation Previous miRNA expression profiling studies have shown that miR-21 is highly expressed in mouse BM.12,27,28 To evaluate the role of miR-21 in hematopoiesis, we first measured the expression of miR-21 in murine hematopoietic stem and progenitor cells (HSPC). It was found that the expression of miR-21 was relatively enriched in HSC compared with that in committed hematopoietic progenitors (Figure 1A), which hints that miR-21 may play a potential role in HSC biology. We then generated a conditional knockout mouse model (miR-21fl/fl;Mx1-Cre) by crossing miR-21fl/fl mice with Mx1-Cre mice (Figure 1B). The deletion of miR-21 was induced in the hematopoietic compartment by seven injections of pIpC (hereafter, miR21fl/fl;Mx1-Cre- and miR-21fl/fl;Mx1-Cre+ mice are referred to as miR-21fl/fl and miR-21D/D mice, respectively), which was confirmed by genomic polymerase chain reaction (PCR) and quantitative real-time PCR (qRT-PCR) analysis (Figure 1C, D; Online Supplementary Figure S1D). However, we did not find significant differences in total cell number in the BM and spleen, or in the counts of 413

M. Hu et al.

white blood cells, red blood cells and platelets in the PB, after miR-21 knockout (Online Supplementary Figure S1E, F). Interestingly, miR-21D/D mice exhibited a significant increase in the percentage of myeloid cells but a marked decrease in the percentage of B cells in the BM and PB (Figure 1E; Online Supplementary Figure S1G). Consistently, we found that the percentage of granulocyte-monocyte progenitors (GMP) was obviously enhanced, while the frequencies of common myeloid progenitors (CMP),

megakaryocyte-erythroid progenitors (MEP) and common lymphoid progenitors (CLP) were reduced in the BM after miR-21 deletion (Figure 1F, G). These findings suggest that miR-21 is crucial for steady-state hematopoiesis.

Targeted deletion of miR-21 generates an aberrant hematopoietic stem cell pool To determine how miR-21 affects hematopoiesis, we next analyzed the phenotypes of HSC from miR-21D/D mice

Figure 1. miR-21 is enriched in hematopoietic stem cells and its conditional ablation skews hematopoietic differentiation. (A) Quantitative real-time polymerase chain reaction (PCR) analysis of miR-21 expression in long-term hematopoietic stem cells (LT-HSC), short-term hematopoietic stem cells (ST-HSC), multipotent progenitors (MPP), common myeloid progenitors (CMP), common lymphoid progenitors (CLP), granulocyte-monocyte progenitors (GMP) and megakaryocyte-erythroid progenitors (MEP) isolated from normal 8-week old wild-type mice (n=4 mice). miR-21 expression was compared with that in LT-HSC. Gating strategies are provided in Online Supplementary Figure S1A, B. (B) The strategy for the generation of the conditional miR-21 knockout mouse model. For detailed genotyping see Online Supplementary Figure S1C. (C) Schematic for pIpC-inducible deletion of miR-21 in the hematopoietic system. (D) PCR-based analysis of genomic DNA from the bone marrow (BM) and spleen (Sp) of miR-21fl/fl and miR-21∆/∆ mice. (E) Flow cytometric analysis of the percentages of T cells (CD3e+), B cells (B220+) and myeloid cells (Gr-1+ and Mac-1+) in the BM of miR-21fl/fl and miR-21∆/∆ mice (n=6 mice per group). (F, G) Flow cytometric analysis of the percentages of (F) CMP, MEP, GMP and (G) CLP in the BM of miR-21fl/fl and miR-21∆/∆ mice (n=6 mice per group). All data are shown as means ± standard deviation. *P<0.05, **P<0.01.

414

haematologica | 2021; 106(2)

MicroRNA-21 maintains HSC homeostasis

by flow cytometry. There were evident increases in the percentage and number of LSK, but not lineage-negative cells and myeloid progenitors, in the BM from miR-21D/D mice compared with those in the BM of miR-21fl/fl mice (Figure 2A). Further analysis revealed that the frequency of longterm HSC (LT-HSC) was increased, whereas the proportion of multipotent progenitors was decreased in the LSK compartments in the absence of miR-21 (Figure 2B). Indeed, the numbers of three LSK subpopulations, especially the LTHSC, were markedly increased in the BM after miR-21 knockout (Figure 2C). A similar result was obtained by staining LSK with another set of HSC surface markers, CD150 and CD48 (Figure 2D; Online Supplementary Figure S2A). We also observed a dramatic increase in the percentage of LSK in the spleen, but not in the PB, when miR-21

was deleted (Figure 2E; Online Supplementary Figure S2B). These results indicate that miR-21 is responsible for sustaining the normal HSC pool.

Loss of miR-21 impairs the quiescence and facilitates the proliferation of hematopoietic stem cells We then set out to explore the possible reasons for the accumulation of phenotypic HSC after miR-21 knockout. Actually, we found a minor but not significantly different decrease in the apoptosis of HSC upon miR-21 ablation (Online Supplementary Figure S3A). Importantly, cell cycle and in vivo 5-bromodeoxyuridine incorporation analysis revealed that miR-21 deficiency strikingly reduced the quiescence and increased the proliferation of HSC, but not of myeloid progenitors (Figure 3A, B; Online Supplementary

Figure 2. Targeted deletion of miR-21 generates an aberrant hematopoietic stem cell pool. (A) Flow cytometric analysis of the percentages and numbers of lineagenegative (Lin-) cells, myeloid progenitors (MP) and Lin-Sca1+c-Kit+ (LSK) cells in miR-21fl/fl and miR-21∆/∆ bone marrow (BM) (n=6 mice per group). (B) Flow cytometric analysis of the proportions of long-term hematopoietic stem cells (LT-HSC), short-term hematopoietic stem cells (ST-HSC) and multipotent progenitors (MPP) in LSK from miR-21fl/fl and miR-21∆/∆ mice (n=6 mice per group). (C) The numbers of LT-HSC, ST-HSC and MPP in miR-21fl/fl and miR-21∆/∆ BM (n=6 mice per group). (D) Flow cytometric analysis of the number of CD150+ CD48- LSK in miR-21fl/fl and miR-21∆/∆ BM (n=6 mice per group). (E) Flow cytometric analysis of the percentage of LSK in the spleen (Sp) of miR-21fl/fl and miR-21∆/∆ mice (n=6 mice per group). All data are shown as means ± standard deviation. *P<0.05, **P<0.01.

haematologica | 2021; 106(2)

415

M. Hu et al.

Figure S3B, C), which is in line with the expression pattern of miR-21. Moreover, qRT-PCR data showed that cyclindependent kinase inhibitors (p21 and p27) were downregulated, whereas cell cycle-associated genes (cyclin E1, cyclin E2, cyclin A2 and CDK6) were upregulated in LT-HSC in the absence of miR-21 (Figure 3C). Given that proliferating active cells are highly sensitive to genotoxic stress,29 we therefore administered miR-21fl/fl and miR-21D/D mice with a single dose of 5-fluorouracil. As expected, miR-21D/D mice displayed more severe myelosup-

pression compared with that of control mice (Figure 3D, E). Likewise, the survival rate of miR-21D/D mice was significantly reduced after weekly 5-fluorouracil injections (Figure 3F). Taken together, these data demonstrate that miR-21 is required to maintain HSC in a quiescent state.

miR-21 deficiency intrinsically compromises hematopoietic stem cell function in long-term reconstitution It has been well accepted that the maintenance of HSC

Figure 3. Loss of miR-21 impairs the quiescence and facilitates the proliferation of hematopoietic stem cells. (A) Cell cycle analysis of Lin-Sca1+c-Kit+ (LSK) cells and long-term hematopoietic stem cells (LT-HSC) in the bone marrow (BM) of miR-21fl/fl and miR-21∆/∆ mice (n=6 mice per group). (B) miR-21fl/fl and miR-21∆/∆ mice were intraperitoneally injected with 5-bromodeoxyuridine (BrdU: 1 mg/6 g mouse weight). Twelve hours later, BrdU incorporation into LSK and LT-HSC from these mice was analyzed by flow cytometry (n=6 mice per group). (C) Quantitative real-time polymerase chain reaction analysis of the relative expression of cyclin-dependent kinase inhibitors and cell cycle-associated genes in sorted miR-21fl/fl and miR-21∆/∆ LT-HSC (n=3 mice per group). (D, E) The number of BM cells (D) and LSK (E) from miR-21fl/fl and miR-21∆/∆ mice after a single injection of 5-fluorouracil (5-FU) (n=6 mice per group). (F) The survival rates of miR-21fl/fl and miR-21∆/∆ mice after sequential 5-FU injections (n=10 mice per group). All data are shown as means ± standard deviation. *P<0.05, **P<0.01.

416

haematologica | 2021; 106(2)

MicroRNA-21 maintains HSC homeostasis

quiescence is closely connected with these cells’ long-term reconstituting capability.6,29 To study this, we first conducted non-competitive bone marrow transplantation (BMT) assays (Figure 4A). Mice transplanted with miR-21D/D BM cells died earlier than those transplanted with miR-21fl/fl BM cells in secondary non-competitive BMT (Figure 4B, C). To further determine the function of miR-21D/D HSC, we next performed competitive BMT assays (Figure 4D). It was observed that miR-21 deficiency led to a reduced percentage of donor-derived cells in recipients’ PB and LSK compartments after primary and secondary transplants, accompanied by myeloid-biased hematopoietic differentiation (Figure 4E-H; Online Supplementary Figure S4A, B). To substantiate these findings, we performed another competitive BMT. BM cells from miR-21fl/fl;Mx1-Cre- or miR-21fl/fl;Mx1-Cre+ mice which were not administered pIpC were transplanted, and at 4 weeks after transplantation, miR-21 ablation was induced by injecting pIpC into the recipients (Figure 4I). Consistently, the compromised function and biased differentiation were also observed (Figure 4J, K). Furthermore, the reciprocal BMT experiment indicates that BM microenvironmental changes are not enough to mediate the functional defects in miR-21D/D HSC (Online Supplementary Figure S4C-E). Collectively, these data confirm that miR-21 intrinsically modulates the function of HSC.

Specific knockout of miR-21 dramatically decreases the NF-κB pathway in hematopoietic stem cells To gain insight into the underlying mechanisms by which miR-21 regulates HSC homeostasis and function, we conducted a microarray analysis of sorted LSK from miR-21fl/fl and miR-21D/D mice. As shown in Figure 5A, we identified 1,050 differentially expressed genes, of which 495 genes were upregulated and 555 genes were downregulated in HSC when miR-21 was deleted. GO enrichment analysis showed that the upregulated genes were markedly enriched in nucleosome assembly, mitotic nuclear division, cell division, DNA replication-dependent nucleosome assembly, cell cycle and chromosome segregation (Figure 5B), which was in accordance with the findings that miR-21 knockout reduced the quiescence and promoted the proliferation of HSC. Besides, consistent with the myeloid bias that manifested in miR-21D/D mice, we noticed a robust increase in the expression of myeloid differentiation genes, such as Mpo, Fcgr3, Csf1r, Igsf6, Ly6c1 and Ms4a3,30,31 in HSC after miR-21 knockout (Online Supplementary Figure S5). Importantly, KEGG pathway analysis revealed that miR-21 ablation dramatically decreased the NF-κB pathway in HSC (Figure 5C), which has been reported to play a critical role in maintaining hematopoietic homeostasis.30,32 In fact, a marked downregulation of NF-κB downstream genes was observed in miR-21D/D LT-HSC (Figure 5D), including p21 (Figure 2C). The attenuation of NF-κB activity was further verified by western blotting, immunofluorescence and flow cytometry (Figure 5E-G). Notably, the defects displayed in miR-21D/D mice, including myeloid bias, accumulation of HSC in the BM and spleen, and reduced quiescence and impaired reconstituting capacity, resemble such aspects of hematopoietic p65-null mice.30 Altogether, these results suggest that miR-21 deficiency inhibits the NF-κB pathway in HSC, which may contribute to the defects in HSC. haematologica | 2021; 106(2)

The upregulation of PDCD4 is responsible for the defects in miR-21-null hematopoietic stem cells Considering that miRNA negatively regulate gene expression at the post-transcriptional level, we then analyzed the expression of miR-21 target genes throughly. Of the miR-21 target genes that have a potential role in affecting NF-κB activity,25,33-35 PDCD4, but not PTEN, Spry1 or Spry2, was markedly upregulated in HSC with miR-21 deficiency (Figure 6A-C; Online Supplementary Figure S6A). The binding between miR-21 and the PDCD4 3’ untranslated region was then confirmed by luciferase reporter assay (Online Supplementary Figure S6B). Intriguingly, we observed that PDCD4 expression was abundant in HSC in mouse BM (Online Supplementary Figure S6C), which is consistent with the specific role of miR-21 in HSC. We next sought to determine whether the upregulation of PDCD4 in HSC contributes to the decreased NF-κB activity, and defective phenotype and function observed in miR-21D/D mice. For this purpose, we transduced normal HSC from WT mice with a lentivirus expressing PDCD4 and found that NF-κB activity was significantly inhibited after overexpression of PDCD4 (Figure 6D). As a consequence, ectopic expression of PDCD4 reduced the quiescence and long-term repopulating potential of HSC, and also biased differentiation (Figure 6E, F; Online Supplementary Figure S6D, E). To further verify this notion, miR-21D/D HSC were transduced with a lentivirus carrying shRNA against PDCD4 (Online Supplementary Figure S6F). As expected, knockdown of PDCD4 partly rescued the defects in miR-21-null HSC (Figure 6H, I; Online Supplementary Figure S6G-I). These data underscore a critical role of miR-21 in supporting the NF-κB pathway by targeting PDCD4, which is essential for the maintenance of HSC homeostasis.

miR-21 protects hematopoietic stem cells from irradiation-induced damage by activating the NF-κB pathway Evidence has indicated a vital role of NF-κB signaling in mitigating irradiation-induced hematopoietic injury,36-38 which led us to speculate that miR-21 may affect irradiation-induced biological processes in HSC. To confirm this, miR-21fl/fl and miR-21D/D mice were simultaneously subjected to ionizing radiation. We found that the loss of miR-21 led to distinctly decreased DNA damage repair, accompanied by increased apoptosis, in HSC exposed to irradiation (Figure 7A, B; Online Supplementary Figure S7A). Moreover, several NF-κB target genes (Ier3, Xrcc5 and Gadd45b) involved in DNA damage repair were significantly decreased in HSC with miR-21 deficiency (Figure 7C). Additionally, a more serious decrease in the number of HSC and white blood cells and an increased death rate were observed in miR-21D/D mice after irradiation (Figure 7D, E; Online Supplementary Figure S7B). Based on the finding that NF-κB signaling is implicated in thrombopoietin-promoted DNA damage repair in HSC, we administered thrombopoietin to the mice and observed that it did not work in miR-21D/D HSC (Online Supplementary Figure S7F, G). Finally, we determined whether the addition of exogenous miR-21 can protect HSC from irradiation by activating the NF-κB pathway. Treatment of mice with a miR-21 agomir, which is an engineered miRNA mimic, increased NF-κB activity in HSC by suppressing PDCD4 (Figure 417

M. Hu et al.

7H), thereby significantly alleviating irradiation-induced damage (Figure 7J; Online Supplementary Figure S7C). These effects were abrogated by a specific NF-κB inhibitor, QNZ38 (Figure 7J; Online Supplementary Figure S7C). In conclusion, our findings demonstrate that miR21 can also protect HSC from radiation damage.

Discussion The maintenance of HSC homeostasis contributes to the continuous supplementation of blood cells throughout the lifetime.1,6 In the past few decades, studies aimed at researching hematopoiesis have revealed many factors

Figure 4. miR-21 deficiency intrinsically compromises hematopoietic stem cell function in long-term reconstitution. (A-C) CD45.2+ miR-21fl/fl or miR-21∆/∆ bone marrow (BM) cells were transplanted into 10.0 Gy-irradiated CD45.1+ recipients. Sixteen weeks later, BM cells harvested from primary recipients were transplanted into 10.0 Gy-irradiated secondary CD45.1+ recipients. The survival rates of these recipient mice after primary (B) and secondary (C) non-competitive bone marrow transplantation (BMT) were monitored (n=10 mice per group). (D-H) CD45.2+ miR-21fl/fl or miR-21∆/∆ BM cells, mixed (1:1) with CD45.1+ competitor BM cells, were transplanted into 10.0 Gy-irradiated CD45.1+ recipients. Sixteen weeks later, BM cells harvested from primary recipients were transplanted into 10.0 Gy-irradiated secondary CD45.1+ recipients. At the indicated times, CD45.2+ donor chimerism and lineage distribution in recipients’ peripheral blood (PB) after primary (E, F) and secondary (G, H) competitive BMT were analyzed by flow cytometry (n=8 mice per group). The data shown on lineage distribution were obtained at 16 weeks after primary and secondary competitive BMT. (I-K) CD45.2+ BM cells from miR-21fl/fl;Mx1-Cre- or miR-21fl/fl;Mx1-Cre+ mice without pIpC treatment, together (1:1) with CD45.1+ competitor BM cells, were transplanted into 10 Gy-irradiated CD45.1+ WT recipients. Eight weeks after transplantation, miR-21 deletion was induced by injecting the recipients with pIpC. At the indicated times, the donor chimerism (J) and lineage distribution (K) in recipients’ PB after transplantation were analyzed by flow cytometry. The data shown on lineage distribution were obtained at 24 weeks after transplantation. A diagram is shown in the left panel. All data are shown as means ± standard deviation. *P<0.05, **P<0.01.

418

haematologica | 2021; 106(2)

MicroRNA-21 maintains HSC homeostasis

capable of controlling the homeostasis of HSC. Recently, growing evidence has suggested that miRNA play prominent roles in hematopoietic cells, including HSC.17,27 In this work, we showed for the first time that miR-21 is required to maintain HSC homeostasis and function by sustaining the NF-κB pathway. As a multifaceted, non-coding RNA, miR-21 is present in multiple tissues and implicated in various physiological and pathological processes.39 Here, we observed that miR21 is relatively enriched in HSC in adult mouse BM, implying that this microRNA may be involved in HSC biology. As a consequence, we found that specific knockout of miR-21 in the hematopoietic system causes an abnormal expansion of the HSC pool in the BM and spleen. Besides, mice with conditional deletion of miR-21 display a distinctly myeloid-skewed differentiation, along with a decrease of B cells. The reason might be that miR-

21 deficiency affects the differentiation of HSPC, a concept with is consistent with the findings of a previous study.18 In addition, a previous study showed that miR 21 can inhibit the apoptosis of diffuse large B cell lymphoma cells through upregulating BCL-2.40 Thus, miR-21 may also play a direct role in regulating B-cell survival. Taken together, our results reveal that miR-21 is essential to maintain normal hematopoiesis in mice. Recently, miR-21 has been increasingly regarded as an oncogene due to its function in promoting tumor cell proliferation and inhibiting apoptosis.10 In the present study, we found that miR-21 deficiency leads to a marked impairment in HSC quiescence, accompanied by increased proliferation, which is very different from the functions of miR-21 in other types of cells. Additionally, deletion of miR-21 in HSC had no effect on cell apoptosis at a steady state. These results suggest that miR-21 regu-

Figure 5. Specific knockout of miR-21 dramatically inhibits the NF-κB pathway in hematopoietic stem cells. (A-C) Lin-Sca1+c-Kit+ (LSK) cells sorted from miR-21∆/∆ and miR-21fl/fl mice were used for microarray analysis. (A) Heatmap of genes differentially expressed between miR-21fl/fl and miR-21∆/∆ LSK. (B) Gene Ontology (GO) analysis of genes upregulated in LSK after deletion of miR-21. The data shown are the top 15 enriched GO terms. (C) KEGG pathway analysis of downregulated genes in miR-21∆/∆ LSK relative to miR-21fl/fl LSK. The top five enriched pathways are shown. The microarray data were gathered from one experiment with three biological replicates. (D) Quantitative real-time polymerase chain reaction analysis of the relative expression of NF-κB target genes (including Egr1, Tnf, Birc3, Nr4a2, Tnfaip3 and Nfkbia) in miR-21fl/fl and miR-21∆/∆ long-term hematopoietic stem cells (n=3 mice per group). (E, F) The expression of p-p65 in LSK from miR-21fl/fl and miR-21∆/∆ bone marrow (n=5 mice per group), determined by western blotting (E) and immunofluorescence (F), respectively. DAPI staining indicates the nucleus of cells. Scale bar represents 5 mm. (G) Flow cytometric analysis of the expression of p-p65 in LSK and LT-HSC from miR-21fl/fl and miR-21∆/∆ bone marrow (n=5 mice per group). MFI: mean fluorescence intensity. All data are shown as means ± standard deviation. **P<0.01.

haematologica | 2021; 106(2)

419

M. Hu et al.

lates the cell cycle and apoptosis, most likely depending on the cellular contexts. It was well recognized that the loss of HSC quiescence can bring about a transient augmentation of phenotypic HSC but eventually compromises their function.6 These results could explain our observation that miR-21D/D HSC have a diminished long-term reconstituting capacity. Notably, the reciprocal transplantation assay validated that the defects manifested in miR21-deficient HSC are cell-intrinsic. In addition, we observed a myeloid bias in recipients transplanted with miR-21D/D BM cells, which is consistent with the changes in non-transplanted miR-21D/D mice. In an effort to characterize the molecular mechanisms by which miR-21 regulates HSC homeostasis, we performed a microarray analysis. Notably, we observed a marked downregulation of the NF-κB pathway when

miR-21 was deleted. It is well established that the NF-κB transcription factor family plays a key role in various physiological processes, including cell proliferation, apoptosis, inflammation and immune responses.41 Current studies using mouse genetic models have indicated that, although aberrant activation of NF-κB is not beneficial for hematopoiesis, basal NF-κB signaling is indispensable for HSC homeostasis.42 Interestingly, miR-21D/D HSC showed similar phenotypes to those of p65-null HSC.30 However, it is unknown whether miR-21 regulates HSC homeostasis and function via the NF-κB pathway. Further investigations revealed that a previously recognized target of miR21, the tumor suppressor PDCD4,33,43 was obviously upregulated in HSC with miR-21 deficiency. However, there is controversy about the function of PDCD4 in regulating NF-κB activity. Most studies have shown that

Figure 6. The upregulation of PDCD4 is responsible for the defects in miR-21-null hematopoietic stem cells. (A, B) The expression of PDCD4 in Lin-Sca1+c-Kit+ (LSK) cells from miR-21fl/fl and miR-21∆/∆ bone marrow (BM) (n=5 mice per group), determined by western blotting (A) and immunofluorescence (B), respectively. DAPI staining indicates the nucleus of cells. Scale bar represents 5 mm. (C) Flow cytometric analysis of the expression of PDCD4 in LSK and long-term hematopoietic stem cells (LT-HSC) from miR-21fl/fl and miR-21∆/∆ BM (n=5 mice per group). MFI: mean fluorescence intensity. (D) Western blotting analysis of the expression of PDCD4, p-p65 and p65 in LSK transduced with control or PDCD4 (n=5 mice per group). (E, F) Normal LSK from CD45.2+ wild-type mice were transduced with the lentivirus carrying control or PDCD4, then transduced cells (6×103), mixed with CD45.1+ competitor BM cells (5×105), were transplantated into 10.0 Gy-irradiated CD45.1+ recipients. At 12 weeks after transplantation, the cell cycle of CD45.2+ donor-derived LT-HSCs in recipients’ BM (E), and the CD45.2+ donor chimerism in recipients’ peripheral blood (PB) (F) were analyzed by flow cytometry (n=6 mice per group). (G-I) CD45.2+ miR-21fl/fl or miR-21∆/∆ LSK (6×103) transduced with the lentivirus carrying control or shRNA against PDCD4 (sh-PDCD4), mixed with CD45.1+ competitor BM cells (5×105), were transplanted into 10.0 Gy-irradiated CD45.1+ recipients. At 12 weeks after transplantation, the cell cycle of CD45.2+ donor-derived LT-HSC in recipients’ BM (H), and the CD45.2+ donor chimerism in recipients’ PB (I) were analyzed by flow cytometry (n=6 mice per group). All data are shown as means ± standard deviation. **P<0.01.

420

haematologica | 2021; 106(2)

MicroRNA-21 maintains HSC homeostasis

PDCD4 is involved in promoting NF-κB activation.44,45 However, in our study, we found that overexpression of PDCD4 in WT HSC by lentiviral transfection inhibited NF-κB activity, while knockdown of PDCD4 expression significantly improved the defects in miR-21-null HSC. These results elucidate that miR-21 sustains the NF-κB pathway, at least partly, by directly targeting PDCD4, which is consistent with the results of a previous study.33 However, whether miR-21 can regulate other target genes still needs further research. In addition, there is an opposing point of view that miR-21 inhibits NF-κB activity,45 which may involve miR-21 regulation of physiological functions in a cellular context-dependent manner.

Ionizing radiation-caused hematopoietic cell death is primarily attributed to the generation of double-strand breaks, which are the most serious form of DNA damage.2 At present, the acute myelosuppression induced by irradiation can be temporarily treated using various hematopoietic growth factors that are able to accelerate the recovery of hematopoietic function by stimulating HSPC proliferation and differentiation.46 However, an effective strategy to protect DNA from irradiation-induced damage in HSC is still lacking. Previous studies have revealed that miR-21 mediates resistance to radiation in many types of tumor cells, which is a great challenge for cancer radiotherapy.47 Furthermore, it has been shown that miR-21 can be stim-

Figure 7. miR-21 protects hematopoietic stem cells from irradiation-induced damage by activating the NF-κB pathway. (A) Flow cytometric analysis of the percentage of γ-H2AX+ cells at 0 (No irradiation) or 2, 6, 12, 24 h after 4.0 Gy total body irradiation (TBI) (n=6 mice per group). (B) Representative immunofluorescence plots showing the expression of γ-H2AX in Lin-Sca1+c-Kit+ (LSK) cells from miR-21fl/fl and miR-21∆/∆ mice 24 h after 4.0 Gy TBI. DAPI staining indicates the nucleus of cells. Scale bar represents 5 mm. (C) Quantitative real-time polymerase chain reaction (qRT-PCR) analysis of the relative expression of NF-κB target genes (Ier3, Xrcc5 and Gadd45b) in LSK from miR-21fl/fl and miR-21∆/∆ mice 5 h after 4.0 Gy TBI (n=3 mice per group). (D) The number of LSK in the bone marrow (BM) of miR-21fl/fl and miR21∆/∆ mice after 4.5 Gy TBI (n=6 mice per group). (E) The survival rates of miR-21fl/fl and miR-21∆/∆ mice after 6.5 Gy TBI (n=10 mice per group). (F, G) miR-21fl/fl and miR-21∆/∆ mice were treated with saline or thrombopoietin (TPO; 8 μg/kg) 30 min before 4.0 Gy TBI. Then, the expression of p-p65 in LSK was measured 40 min after irradiation (n=6 mice per group) (F), and the percentage of γ-H2AX+ LSK in these mice was measured by flow cytometry 24 h after irradiation (n=6 mice per group) (G). (H) Normal wild-type mice were treated with three doses of miR-21 agomir or negative control (NC) on consecutive days by tail intravenous injection. Twenty-four hours after the last injection, the relative miR-21 expression in LSK from these mice was detected by qRT-PCR (n=6 mice per group), and the expression of PDCD4 and p-p65 in LSK from these mice was analyzed by flow cytometry (n=6 mice per group). (I) Mice were treated with miR-21 agomir or NC as described before. Twentyone hours after the last injection, mice were intraperitoneally injected with one dose of QNZ or control. Three hours later, mice were subjected to 4.0 Gy TBI. The percentage of γ-H2AX+ cells in LSK from these mice was detected 24 h after irradiation by flow cytometry (n=6 mice per group). All data are shown as means ± standard deviation. *P<0.05, **P<0.01.

haematologica | 2021; 106(2)

421

M. Hu et al.

ulated by irradiation via the EGFR/STAT3 and ATM pathways,48,49 which is line with our data (Online Supplementary Figure S7D). However, the role of miR-21 in HSC in the context of irradiation remains obscure. In the present study, we discovered that miR-21 can prevent HSC from irradiation-induced DNA damage through activating the NF-κB pathway, which is consistent with a recent finding that miR-21 promotes hematopoietic cell survival after exposure to irradiation.50 Moreover, our study revealed that miR-21 may also be involved in thrombopoietinmediated non-homologous end junction repair in HSC. Nevertheless, we do not deny that there may be other NFκB-independent mechanisms that mediate the radioprotective effects of miR-21 in HSC.47 On the other hand, some studies have reported that miR-21 is also a downstream target gene of the NF-κB pathway.44 Indeed, overexpression of PDCD4 or treatment with a NF-κB inhibitor (QNZ) significantly inhibited the expression of miR-21 in HSC from normal WT mice (data not shown). These findings led us to hypothesize that there may be a positive regulatory feedback loop that regulates HSC function. In summary, our findings reveal a novel function of miR-21 as an important regulator of HSC homeostasis via modulation of the NF-κB pathway by targeting PDCD4,

References 1. Rathinam C, Matesic LE, Flavell RA. The E3 ligase Itch is a negative regulator of the homeostasis and function of hematopoietic stem cells. Nat Immunol. 2011;12(5):399407. 2. Yamashita M, Nitta E, Suda T. Aspp1 preserves hematopoietic stem cell pool integrity and prevents malignant transformation. Cell Stem Cell. 2015;17(1): 23-34. 3. Suda T, Takubo K, Semenza GL. Metabolic regulation of hematopoietic stem cells in the hypoxic niche. Cell Stem Cell. 2011;9(4): 298-310. 4. Miyamoto K, Araki KY, Naka K, et al. Foxo3a is essential for maintenance of the hematopoietic stem cell pool. Cell Stem Cell. 2007;1(1):101-112. 5. Anso E, Weinberg SE, Diebold LP, et al. The mitochondrial respiratory chain is essential for haematopoietic stem cell function. Nat Cell Biol. 2017;19(6):614-625. 6. Wang H, Diao D, Shi Z, et al. SIRT6 controls hematopoietic stem cell homeostasis through epigenetic regulation of Wnt signaling. Cell Stem Cell. 2016;18(4):495-507. 7. Mehta A, Zhao JL, Sinha N, et al. The microRNA-132 and microRNA-212 cluster regulates hematopoietic stem cell maintenance and survival with age by buffering FOXO3 expression. Immunity. 2015;42(6): 1021-1032. 8. Hu M, Zeng H, Chen S, et al. SRC-3 is involved in maintaining hematopoietic stem cell quiescence by regulation of mitochondrial metabolism in mice. Blood. 2018;132(9): 911-923. 9. Qian P, He XC, Paulson A, et al. The Dlk1Gtl2 locus preserves LT-HSC function by inhibiting the PI3K-mTOR pathway to restrict mitochondrial metabolism. Cell Stem Cell. 2016;18(2):214-228. 10. Medina PP, Nolde M, Slack FJ. OncomiR addiction in an in vivo model of microRNA21-induced pre-B-cell lymphoma. Nature.

422

thereby extending our understanding of the biological function of miR-21. At the same time, we demonstrate that targeting miR-21 may be a promising avenue for safeguarding HSC against irradiation damage. Disclosures No conflicts of interest to disclose. Contributions MH designed the study, performed experiments, analyzed data and wrote the paper; YL, ZZ and HZ performed experiments and analyzed data. SC, YQ, YX, FC and YT participated in the animal experiments and data analysis; MC, CD and MS contributed to the in vitro experiments; FW and YS participated in the initial experimental design and discussed the manuscript; JW and SW conceived and supervised the study, and revised the manuscript. Acknowledgments We thank Prof. Jinyong Wang for the gift of the CD45.1 mice, Yang Liu for technical support in flow cytometry, and Liting Wang for technical assistance in immunofluorescence microscopy. This work was supported by grants from the National Natural Science Fund of China (n. 81725019, 81930090, 81573084, 81500087) and the Scientific Research Project of PLA (AWS16J014).

2010;467(7311):86-90. 11. Thum T, Gross C, Fiedler J, et al. MicroRNA-21 contributes to myocardial disease by stimulating MAP kinase signalling in fibroblasts. Nature. 2008;456 (7224):980-984. 12. O'Connell RM, Chaudhuri AA, Rao DS, Gibson WS, Balazs AB, Baltimore D. MicroRNAs enriched in hematopoietic stem cells differentially regulate long-term hematopoietic output. Proc Natl Acad Sci U S A. 2010;107(32):14235-14240. 13. Kotaki R, Koyama-Nasu R, Yamakawa N, Kotani A. miRNAs in normal and malignant hematopoiesis. Int J Mol Sci. 2017;18(7). 14. Song SJ, Ito K, Ala U, et al. The oncogenic microRNA miR-22 targets the TET2 tumor suppressor to promote hematopoietic stem cell self-renewal and transformation. Cell Stem Cell. 2013;13(1):87-101. 15. Hu W, Dooley J, Chung SS, et al. miR-29a maintains mouse hematopoietic stem cell self-renewal by regulating Dnmt3a. Blood. 2015;125(14):2206-2216. 16. Lechman ER, Gentner B, Ng SWK, et al. miR126 regulates distinct self-renewal outcomes in normal and malignant hematopoietic stem cells. Cancer Cell. 2016;29(4):602-606. 17. Roden C, Lu J. MicroRNAs in control of stem cells in normal and malignant hematopoiesis. Curr Stem Cell Rep. 2016;2 (3):183-196. 18. Velu CS, Baktula AM, Grimes HL. Gfi1 regulates miR-21 and miR-196b to control myelopoiesis. Blood. 2009;113(19):47204728. 19. Wang Z, Brandt S, Medeiros A, et al. MicroRNA 21 is a homeostatic regulator of macrophage polarization and prevents prostaglandin E2-mediated M2 generation. PLoS One. 2015;10(2):e0115855. 20. Canfran-Duque A, Rotllan N, Zhang X, et al. Macrophage deficiency of miR-21 promotes apoptosis, plaque necrosis, and vascular inflammation during atherogenesis. EMBO Mol Med. 2017;9(9):1244-1262.

21. Lu TX, Lim EJ, Itskovich S, et al. Targeted ablation of miR-21 decreases murine eosinophil progenitor cell growth. PLoS One. 2013;8(3):e59397. 22. Bhagat TD, Zhou L, Sokol L, et al. miR-21 mediates hematopoietic suppression in MDS by activating TGF-beta signaling. Blood. 2013;121(15):2875-2881. 23. Panagal M, S RS, P S, et al. MicroRNA21 and the various types of myeloid leukemia. Cancer Gene Ther. 2018;25(7-8):161-166. 24. Yu Y, Nangia-Makker P, Farhana L, S GR, Levi E, Majumdar AP. miR-21 and miR-145 cooperation in regulation of colon cancer stem cells. Mol Cancer. 2015;14:98. 25. Wu T, Liu Y, Fan Z, et al. miR-21 Modulates the immunoregulatory function of bone marrow mesenchymal stem cells through the PTEN/Akt/TGF-beta1 pathway. Stem Cells. 2015;33(11):3281-3290. 26. Zeng H, Hu M, Lu Y, et al. MicroRNA 34a promotes ionizing radiation-induced DNA damage repair in murine hematopoietic stem cells. FASEB J. 2019;33(7):8138-8147. 27. Petriv OI, Kuchenbauer F, Delaney AD, et al. Comprehensive microRNA expression profiling of the hematopoietic hierarchy. Proc Natl Acad Sci U S A. 2010;107(35):1544315448. 28. Heiser D, Tan YS, Kaplan I, et al. Correlated miR-mRNA expression signatures of mouse hematopoietic stem and progenitor cell subsets predict "stemness" and "myeloid" interaction networks. PLoS One. 2014;9(4): e94852. 29. Venkatraman A, He XC, Thorvaldsen JL, et al. Maternal imprinting at the H19-Igf2 locus maintains adult haematopoietic stem cell quiescence. Nature. 2013;500(7462):345349. 30. Stein SJ, Baldwin AS. Deletion of the NFκB subunit p65/RelA in the hematopoietic compartment leads to defects in hematopoietic stem cell function. Blood. 2013;121(25):5015-5024. 31. Mullenders J, Aranda-Orgilles B, Lhoumaud

haematologica | 2021; 106(2)

MicroRNA-21 maintains HSC homeostasis

P, et al. Cohesin loss alters adult hematopoietic stem cell homeostasis, leading to myeloproliferative neoplasms. J Exp Med. 2015;212(11):1833-1850. 32. Fang J, Muto T, Kleppe M, et al. TRAF6 mediates basal activation of NF-κB necessary for hematopoietic stem cell homeostasis. Cell Rep. 2018;22(5):1250-1262. 33. Shi C, Yang Y, Xia Y, et al. Novel evidence for an oncogenic role of microRNA-21 in colitis-associated colorectal cancer. Gut. 2016;65(9):1470-1481. 34. Wang X, Zhang L, Wei Z, et al. The inhibitory action of PDCD4 in lipopolysaccharide/ D-galactosamine-induced acute liver injury. Lab Invest. 2013;93(3):291-302. 35. Mao XH, Chen M, Wang Y, Cui PG, Liu SB, Xu ZY. MicroRNA-21 regulates the ERK/NFκB signaling pathway to affect the proliferation, migration, and apoptosis of human melanoma A375 cells by targeting SPRY1, PDCD4, and PTEN. Mol Carcinog. 2017;56(3):886-894. 36. Kraft D, Rall M, Volcic M, et al. NF-κBdependent DNA damage-signaling differentially regulates DNA double-strand break repair mechanisms in immature and mature human hematopoietic cells. Leukemia. 2015;29(7):1543-1554. 37. Burdelya LG, Krivokrysenko VI, Tallant TC, et al. An agonist of toll-like receptor 5 has

haematologica | 2021; 106(2)

radioprotective activity in mouse and primate models. Science. 2008;320(5873):226-230. 38. de Laval B, Pawlikowska P, Barbieri D, et al. Thrombopoietin promotes NHEJ DNA repair in hematopoietic stem cells through specific activation of Erk and NF-κB pathways and their target, IEX-1. Blood. 2014;123(4):509-519. 39. Krichevsky AM, Gabriely G. miR-21: a small multi-faceted RNA. J Cell Mol Med. 2009;13(1):39-53. 40. Liu K, Du J, Ruan L. MicroRNA-21 regulates the viability and apoptosis of diffuse large Bcell lymphoma cells by upregulating B cell lymphoma-2. Exp Ther Med. 2017;14(5): 4489-4496. 41. Zhao C, Xiu Y, Ashton J, et al. Noncanonical NF-κB signaling regulates hematopoietic stem cell self-renewal and microenvironment interactions. Stem Cells. 2012;30(4): 709-718. 42. Nakagawa MM, Rathinam CV. Constitutive Activation of the canonical NF-κB pathway leads to bone marrow failure and induction of erythroid signature in hematopoietic stem cells. Cell Rep. 2018;25 (8):2094-2109 e2094. 43. Zhang Z, Zha Y, Hu W, et al. The autoregulatory feedback loop of microRNA-21/programmed cell death protein 4/activation protein-1 (MiR-21/PDCD4/AP-1) as a driv-

ing force for hepatic fibrosis development. J Biol Chem. 2013;288(52):37082-37093. 44. Sheedy FJ, Palsson-McDermott E, Hennessy EJ, et al. Negative regulation of TLR4 via targeting of the proinflammatory tumor suppressor PDCD4 by the microRNA miR-21. Nat Immunol. 2010;11(2):141-147. 45. Ma X, Becker Buscaglia LE, Barker JR, Li Y. MicroRNAs in NF-kappaB signaling. J Mol Cell Biol. 2011;3(3):159-166. 46. Xu G, Wu H, Zhang J, et al. Metformin ameliorates ionizing irradiation-induced longterm hematopoietic stem cell injury in mice. Free Radic Biol Med. 2015;87:15-25. 47. Hu B, Wang X, Hu S, et al. miR-21-mediated Radioresistance occurs via promoting repair of DNA double strand breaks. J Biol Chem. 2017;292(8):3531-3540. 48. Zhang X, Wan G, Berger FG, He X, Lu X. The ATM kinase induces microRNA biogenesis in the DNA damage response. Mol Cell. 2011;41(4):371-383. 49. Zhu Y, Yu X, Fu H, et al. MicroRNA-21 is involved in ionizing radiation-promoted liver carcinogenesis. Int J Clin Exp Med. 2010;3(3):211-222. 50. Puccetti MV, Adams CM, Dan TD, et al. MicroRNA-21 is required for hematopoietic cell viability after radiation exposure. Int J Radiat Oncol Biol Phys. 2019;104(5):11651174.

423

ARTICLE Ferrata Storti Foundation

Hematopoiesis

Signal-transducing adaptor protein-2 delays recovery of B-lineage lymphocytes during hematopoietic stress

Michiko Ichii,1 Kenji Oritani,2 Jun Toda,1 Hideaki Saito,1 Henyun Shi,1 Hirohiko Shibayama,1 Daisuke Motooka,3 Yuichi Kitai,4 Ryuta Muromoto,4 Jun-ichi Kashiwakura,4 Kodai Saitoh,4 Daisuke Okuzaki,3 Tadashi Matsuda,4 and Yuzuru Kanakura1,5

Haematologica 2021 Volume 106(2):424-436

Department of Hematology and Oncology, Osaka University Graduate School of Medicine, Suita; 2Department of Hematology, Graduate School of Medical Science, International University of Health and Welfare, Narita; 3Genome Information Research Center, Research Institute for Microbial Diseases, Osaka University, Suita; 4Department of Immunology, Graduate School of Pharmaceutical Sciences, Hokkaido University, Sapporo and 5 Sumitomo Hospital, Osaka, Japan 1

ABSTRACT

Received: April 26, 2019. Accepted: January 23, 2020. Pre-published: January 23, 2020.

ignal-transducing adaptor protein-2 (STAP-2) was discovered as a C-FMS/M-CSFR interacting protein and subsequently found to function as an adaptor of signaling or transcription factors. These include STAT5, MyD88 and IκB kinase in macrophages, mast cells, and T cells. There is additional information about roles for STAP-2 in several types of malignant diseases including chronic myeloid leukemia; however, none have been reported concerning B-lineage lymphocytes. We have now exploited gene targeted and transgenic mice to address this lack of knowledge, and demonstrated that STAP-2 is not required under normal, steadystate conditions. However, recovery of B cells following transplantation was augmented in the absence of STAP-2. This appeared to be restricted to cells of B-cell lineage with myeloid rebound noted as unremarkable. Furthermore, all hematologic parameters were observed to be normal once recovery from transplantation was complete. In addition, overexpression of STAP-2, specifically in lymphoid cells, resulted in reduced numbers of latestage B-cell progenitors within the bone marrow. While numbers of mature peripheral B and T cells were unaffected, recovery from sub-lethal irradiation or transplantation was dramatically reduced. Lipopolysaccharide (LPS) normally suppresses B precursor expansion in response to interleukin 7; however, STAP-2 deficiency made these cells more resistant. Preliminary RNA-sequencing analyses indicated multiple signaling pathways in B progenitors to be STAP-2-dependent. These findings suggest that STAP-2 modulates formation of B lymphocytes in demand conditions. Further study of this adapter protein could reveal ways to speed recovery of humoral immunity following chemotherapy or transplantation.

https://doi.org/10.3324/haematol.2019.225573

Introduction

Correspondence: MICHIKO ICHII michii@bldon.med.osaka-u.ac.jp

424

Production of blood cells in bone marrow (BM) is highly regulated. Billions of blood cells are derived from multipotent hematopoietic stem cells (HSC). Indeed, a wide spectrum of hematologic lineages is produced on a daily basis over an individual’s lifetime.1,2 Hematopoiesis is flexible enough to respond to various types of stress, including chemotherapy, acute or chronic infections, and injuries.3 In such situations, myeloid lineage cells often respond first to resolve inflammatory events, after which they need to be rapidly regenerated.4 Recent studies have shown that HSC play an important role in driving this emergency myelopoiesis. For example, hematopoietic progenitors (HPC) and HSC in BM can respond to stimulation of toll-like receptors (TLR) that detect microbial products. This results in HSC expansion, increased myeloid differentiation and depletion of lymphoid progenitors via direct and indirect ways.5-8 Besides this, proinflammatory cytokines such as interleukin (IL)-1, IL-6, tumor necrosis factor alpha (TNFα), interferons (IFN), and granulocyte-colony stimulating factor (G-CSF) impact the fate of multipotent haematologica | 2021; 106(2)

Effects of STAP-2 on B-cell recovery

hematopoietic stem/progenitor cells (HSPC).5,9,10 Many studies have focused on the pathophysiology of HSC, while few have investigated the role of lineage-committed progenitors, which have great capacity for proliferation. Treatments for hematologic malignancies such as leukemia and lymphoma have been dramatically improved by recent advances in chemotherapy, immunotherapy and HSC transplantation (HSCT). However, compromising the immune system remains a frequent complication of various types of therapy, and induces the risk of non-relapse mortality. Especially in allogeneic HSCT settings, which is the only curative therapy for patients with refractory malignancies and severe BM failure diseases, regeneration of cellular and humoral immunity occurs over one year, while the recovery of innate immune cells, megakaryocytes and erythrocytes is usually observed within one month of HSCT.11 Similar to clinical observations, murine HSCT experiments show relatively slow recovery of lymphocytes. Under regenerative conditions, HSC and myeloid-biased multipotent progenitors (MPP) enter cell-cycle, supporting early recovery of myeloid cells.12,13 However, the mechanisms of lymphoid reconstitution is less well understood. In 2003, we identified signal-transducing adaptor protein-2 (STAP-2) as a C-FMS/M-CSFR interacting protein.14 STAP-2 contains an N-terminal pleckstrin homology domain, a proline-rich region and an YXXQ motif. Its central region is distantly related to the Src homology 2-like (SH2) domain. As the adaptor protein structure predicts, we and others identified roles in inflammatory reactions, cell survival, migration and cell adhesion in macrophages, T cells or mast cells.15-18 Although interactions with inflammatory molecules such as STAT5, MyD88, and IκB kinase (IKK) have been shown in immune cells, the importance of STAP-2 to hematopoiesis in BM remains unknown. Therefore, we investigated STAP-2-mediated regulation of stress hematopoiesis using genetically modified mice.

Flow cytometry Flow cytometric analysis and sorting were performed using a FACS Aria IIu or FACSCanto (BD Biosciences). The antibodies used for flow cytometric sorting and analysis are listed in Online Supplementary Table S1. Antibodies to CD19, CD45R/B220, CD11b/Mac1, Gr1, Ter119, CD3, and CD8 were used as lineage markers. FITC-Annexin V apoptosis Detection Kit (BD Biosciences) for apoptosis detection, and BrdU Flow Kit (BD Biosciences) for cell proliferation were also used according to the manufacturer’s protocol. FlowJo software (Tree Star) was used for data analysis.

Cultures The details of culture methods are provided in the Online Supplementary Methods.

RNA-sequencing Sequencing was performed using an Illumina HiSeq 2500 platform in a 75-base single-end mode. The raw data were deposited in the National Center for Biotechnology Information Gene Expression Omnibus database (GSE127939). The normalized values compared to control were defined to show signal ratios with greater than 2-fold increases or decreases. Bioinformatic analyses were conducted with Ingenuity Pathway Analysis software (Ingenuity Systems).

Real-time quantitative polymerase chain reaction analyses of gene expression Semi-quantitative and real-time reverse transcriptase polymerase chain reaction (RT-PCR) was used to assess mRNA expression. Reactions were quantified using fluorescent TaqMan technology.

Statistical analysis Statistical analysis was carried out using unpaired two-tailed Mann-Whitney tests. The error bars used throughout indicate standard deviation (SD) of the mean.

Results Methods Mice STAP-2 knockout (KO) and transgenic (Tg) mice of the C57BL/6J strain were generated and maintained as described previously.14 For the generation of STAP-2 Tg mice, a cDNA fragment including the full coding region of the human STAP-2 gene was subcloned into a p1026X vector, which consisted of the murine Lck proximal promoter, the Ig intronic H chain enhancer E , and a human growth hormone (hGH) cassette, as previously described.19 Eight to 16-week old mice were used in all experiments except for those involving aged mice. Some mice were administered BrdU intraperitoneally (100 mg/kg of body weight) 12 hours prior to BM collection for cell cycle analyses. To examine age-related hematopoiesis, 12-22-month old mice were used. All experimental procedures were conducted under protocols approved by the Institutional Animal Care and Use Committee of Osaka University.

Bone marrow transplantation Six thousand lineage– Sca1+ cKithigh (LSK) cells sorted from C57BL/6-Ly5.2 (CD45.2) mice were mixed with 6x105 unfractionated adult BM cells obtained from wild-type (WT) C57BL/6-Ly5.1 (CD45.1) mice. The mixture of cells was transplanted into C57BL/6-Ly5.1 mice lethally irradiated at a dose of 8.5 Gy. haematologica | 2021; 106(2)

Loss of STAP-2 accelerates B-cell recovery in bone marrow transplantation It has been reported that STAP-2 affects immune cell responses by directly binding to a variety of proinflammatory molecules, including MyD88, IKKα/β, C-FMS, STAT3, and STAT5.14-18 In tumorigenesis, STAP-2 regulates cell migration, proliferation and therapy resistance for tyrosine kinase inhibitors.20-23 Hematopoiesis under steady state appears normal in STAP-2 deficient mice;24 we, therefore, hypothesized that STAP-2 may regulate stress hematopoiesis. Here, we first conducted immunophenotypic characterization of STAP-2 deficient HSPC with flow cytometry, which were found to be comparable between the WT and knockout (KO) BM under homeostatic conditions (Figure 1A). Next, LSK derived from STAP-2 KO or WT mice (CD45.2+) was intravenously injected into congenic recipients (CD45.1+). One month after HSCT, we found that CD45.2+ donor chimerism in peripheral B cells in KO HSPC transplanted mice was significantly higher than that in controls (WT donor, 51.8±1.0%; STAP-2 KO donor, 72.0±1.5%; P<0.001), while recovery of other lineages including Mac1+ or Gr-1+ myeloid and CD3+ T cells was comparable (Figure 1B). The calculation of blood lin425

M. Ichii et al.

Figure 1. B-cell reconstitution following transplantation is improved in the absence of STAP-2. (A) The proportion of the indicated subsets (LK; Lineage– Sca1– cKithigh, LSK; Lineage– Sca1+ cKithigh, early lymphoid progenitor (ELP); Flk2high IL7Rα– LSK, common lymphoid progenitor (CLP); Lin– Sca1+ cKitlow Flk2high IL7Rα+ , LT-HSC and multipotent progenitor (MPP)1; Flk2– CD48– CD150+ LSK, MPP2; Flk2– CD48+ CD150+ LSK, MPP3; Flk2– CD48+ CD150- LSK, MPP4; Flk2- CD48- CD150- LSK) in bone marrow (BM) were analyzed with flow cytometry using wild-type (WT) and STAP-2 knockout (KO) mice (WT, n=5; STAP-2 KO, n=4). (B-E) Competitive transplantation experiments were conducted using KO mice. LSK cells (CD45.2+) derived from KO or WT were injected intravenously with competitor BM cells (CD45.1+). (B) Flow cytometry was used to evaluate hematopoietic recovery of Mac1+ or Gr1+ myeloid (M), CD19+ B- and CD3+ T-lineage cells in CD45.2+ donor-derived peripheral blood at the indicated times after transplantation. (C) The lineage distribution in CD45.2+ cells one month after transplantation. (D) BM cells were collected one month after transplantation, and donor chimerism (CD45.2+) was evaluated in Flk2– or Flk2+ LSK cells, ELP and CLP (left panel). The proportion of CD45.2+ cells in the B lineage progenitor subsets (pre-pro-B; B220+ CD43+ CD19– IgM–, pro-B; B220+ CD43+ CD19+ IgM–, pre-B; B220+ CD43- CD19+ IgM-, immature B; B220+ CD43- CD19+ IgM+) was also analyzed (right panel). (E) The same experimental strategy was used to assess the effects of STAP-2 on aged hematopoiesis in 12-22-month old KO mice, using age-matched WT mice as a control. Similar results were obtained in three independent experiments. Six to ten mice were used for each experiment. The results are shown as means±standard deviation. Statistical significances relative to WT controls were determined by unpaired two-tailed Mann-Whitney tests: *P<0.05; **P<0.01.

426

haematologica | 2021; 106(2)

Effects of STAP-2 on B-cell recovery

eage distribution in donor-derived cells showed that the proportion of B cells was increased in KO donors, compared to that in WT donors (Figure 1C). Two months following transplantation, B-cell chimerism reached a plateau at which the engraftment and lineage distribution were indistinguishable between KO and WT donors

(Figure 1B). Interestingly, we noticed that the proportion of KO donor-derived B cells one month after transplantation was similar to that at two months, indicating that Bcell reconstitution was completed by one month in KO LSK transplanted mice, while it took two months in WT control.

Figure 2. STAP-2 overexpression impairs early B lymphopoiesis in bone marrow (BM). BM cells from transgenic mice (Tg) that overexpress human STAP-2 under the control of an Em enhancer and Lck proximal promoter were used to evaluate roles of STAP-2 in early B lymphopoiesis. (A) The expression level of the inserted human STAP-2 was analyzed with real-time polymerase chain reaction (RT-PCR). The transcript levels in the indicated subsets (LSK; Lineage- Sca1+ cKithigh, common lymphoid progenitor [CLP]; Lin– Sca1+ cKitlow Flk2high IL7Rα+ , pro-B; B220+ CD43+ CD19+ IgM–, pre-B; B220+ CD43- CD19+ IgM) were normalized to the median of lymph node cell (LN) samples in Tg mice. (B) The percentages of Ter119+ erythroid cells, Gr1+ or Mac1+ myeloid cells, CD19+ B cells, or CD3+ T cells, were determined by flow cytometry, and exact numbers were calculated. (C and E) Without initial separation, the samples were stained for lineage-associated surface markers for flow cytometric analysis. Representative gating for B-progenitor subsets is shown in (C) (top right panel). Proportion and number of cells in BM of B220+ CD19– CD11clow Ly6C+ plasmacytoid dendritic cells (pDC) and B220– CD19– CD11c+ conventional dendritic cells (cDC) were analyzed (E). (D) Ki67 and 7AAD staining was used for the analysis of the cell cycle status in B220+ CD19+ cells (left panel). Cell apoptosis was analyzed by Annexin V and 7AAD in B220+ cells using flow cytometry (right panel). Similar results were obtained in three independent experiments. Six to ten mice were used for each experiment. Results are presented as mean±standard deviation. Statistical significances relative to wild-type (WT) control were determined by unpaired two-tailed Mann-Whitney test: *P<0.05; **P<0.01. KO: knockout; CLP: common lymphoid progenitor; HSPC: hematopoietic stem/progenitor cells.

haematologica | 2021; 106(2)

427

M. Ichii et al. B

Figure 3. STAP-2 is essential for B-cell recovery during stress hematopoiesis. (A) Percentages of Mac1+ or Gr1+ myeloid (M), CD19+ B, CD3+ T-lineage cells and NK1.1+ natural killer (NK) cells in peripheral blood of STAP-2 transgenic (Tg) and wild-type (WT) mice were analyzed via flow cytometry, and the number of white blood cells (WBC) and different lineages are shown (n=6 in each). (B) Weight of spleen was measured and the ratio to body weight (BW) was calculated (n=6 in each group). (C and D) WT or Tg mice were sub-lethally irradiated (3.5 Gy). (C) Peripheral blood was examined twice per week, and WBC counts were compared at each time point (top panel). Percentages of B220+ CD19+ B cells (bottom panel) were calculated using flow cytometry. Data represent three independent trials with similar results. (D) BrdU was administered 12 hours before analysis of the cell cycle status of B220+ CD19+ cells in bone marrow (BM) via flow cytometry (top panel). Cell apoptosis was analyzed by Annexin V and 7AAD in B220+ cells in BM using flow cytometry (bottom panel). Data were pooled from two independent experiments (n=5 in each). (E) LSK cells (CD45.2+) derived from Tg or WT were injected intravenously with competitor BM cells (CD45.1+) (n=7 in each). Flow cytometry was used to evaluate hematopoietic recovery of myeloid, B- and T-lineage cells in CD45.2+ donor-derived peripheral blood at the indicated time after transplantation (top panels). BM cells were collected one month after transplantation, and donor chimerism (CD45.2+) was evaluated in B-lineage progenitor subsets (bottom panels). Similar results were obtained in two independent experiments. Results are shown as meanÂąstandard deviation. Statistical significances relative to WT control were determined using unpaired two-tailed Mann-Whitney tests: *P<0.05; **P<0.01.

428

haematologica | 2021; 106(2)

Effects of STAP-2 on B-cell recovery

Figure 4. STAP-2 impairs B-cell differentiation and pre-B-cell proliferation. (A) LSK cells derived from STAP-2 transgenic (Tg) marrow were cultured under stromalcell free, B-cell differentiation-supporting conditions. Flow cytometry was used to evaluate differentiation. (Left) Representative gating for Mac1+ or Gr1+ myeloid and B220+ CD19+ B cells. The number of myeloid and B cells recovered were used to calculate yields per input cell (fold expansion). (B and C) Lin– Sca1+ cKitlow Flk2high IL7Rα+ CLP derived from Tg were cultured without (B) or with (C) OP9 stromal cells under B-cell conditions, and the production of B220+ CD19+ B cells was analyzed. (D) In colony-forming unit (CFU pre-B) assays, the ability to generate pre-B colonies was evaluated. (E) The stromal cell-free culture system was used to evaluate the proliferation of pre-B cells and differentiation to IgM+ immature B cells. Similar results were obtained in three independent experiments. The results are shown as mean±standard deviation. Statistical significances relative to wild-type (WT) control were determined by unpaired two-tailed Mann-Whitney tests: *P<0.05; **P<0.01. KO: knockout.

We analyzed HSPC in BM one month after transplantation via flow cytometry and found that there were no differences in the number of each subset of multipotent HSPC and early lymphoid progenitors between WT and KO donors (Figure 1D). In contrast, STAP-2 deficiency significantly increased pre-B-cell subset. These results indicate that STAP-2 in B progenitors is involved with Bcell recovery under regenerative conditions. It is known that self-renewal capacity and lineage commitment in HSPC change according to the type of hematopoietic stress to which they are exposed;1,2 therefore, the effects of STAP-2 on aged hematopoiesis were also analyzed. In aged KO mice, no abnormalities were observed with respect to hematopoietic parameters in BM or peripheral blood (Online Supplementary Figure S1). In old KO LSK-transplanted mice, comparable long-term engraftment was observed. We found B-cell reconstitution from aged HSPC was also accelerated by STAP-2 deletion (Figure 1E). Collectively, our findings indicate that STAP-2 suppresses B-cell recovery during stress hematopoiesis at the pre-B-cell stage in BM, while STAP-2 does not affect the functions of multipotent HSPC.

STAP-2 is critical for B-cell differentiation during stress hematopoiesis To further investigate the role of STAP-2 in early B lymphopoiesis, we generated transgenic mice that overexpress STAP-2 under the control of an Em enhancer and Lck proximal promoter. This promoter could drive haematologica | 2021; 106(2)

expression of the inserted cDNA in lymphoid lineage cells in BM, T-cell precursors in thymus, and peripheral mature lymphoid cells. The consistent upregulation from the common lymphoid progenitor (CLP) stage to mature lymphoid cells in lymph nodes was confirmed with RT-PCR (Figure 2A). Tg mice showed no apparent defects, and had normal life-spans. We conducted a thorough analysis of marrow hematopoiesis in STAP-2 Tg mice that were 8-16 weeks old. No abnormalities were found with respect to marrow cellularity or proportions and numbers of HSPC and CLP (Figure 2B and C). In contrast, numbers of B220+ B progenitors were significantly reduced (Figure 2B). By immunophenotyping B-cell development, the reduction of all BM B-lineage-committed progenitor stages (pre-proB, pro-B and pre-B) was confirmed (Figure 2C). Analyses of the proliferation and apoptosis status showed no statistical differences among WT, KO, and Tg mice under steady states (Figure 2D). Dendritic cells that can be generated from CLP, were also affected by STAP-2 overexpression (Figure 2E). These findings suggest that STAP-2 has a critical function in maintaining the integrity of the B-progenitor compartment. Although overexpression of STAP-2 in B precursors seemed to increase the sensitivity to subtle stress under homeostasis, peripheral B-cell cellularity could be maintained through compensatory mechanisms, and the sizes of spleen were rather enlarged (Figure 3A and B). We therefore investigated whether B-cell recovery is compro429

M. Ichii et al.

Figure 5. STAP-2-dependent inflammatory signals at the pre-B stage regulate B lymphopoiesis during hematologic stress. (A-D) RNA-Seq experiments with pre-B cells derived from wild-type (WT) and STAP-2 transgenic (Tg) bone marrow (BM) under steady-state as well as WT and STAP-2 knockout (KO) BM one month after transplantation (Tx) was conducted, and the genes were grouped according to a hierarchical clustering technique with Ingenuity Pathway Analysis based on patterns of expression. (A) Number of genes extracted. (B) A colorcoded heatmap allows for the visualization of the differential gene expression data categorized by their functions in the Ingenuity Knowledge Base. Color bar indicates the z-score for each category: the strongest predicted increases (orange square) correspond to z-score > 2, the strongest predicted decreases (blue square) correspond to z-score < -2. Gray and white colors indicate categories with a -2 < z-score < 2 and without zscore, respectively. Size reflects the associated log of the calculated P-value with s significance threshold of P=0.05 (Fisher's exact test). Larger squares indicate more significant overlap among the affected genes contained in the dataset. (C and D) Datasets were compared, and a bioinformative approach identified shared pathways regulated under hematologic stress and by STAP-2 expression. (E) The proportion of IL7 receptor α+ in the B-lineage progenitor subsets in WT, KO, and Tg BM (pro-B; B220+ CD43+ CD19+ IgM–, pre-B; B220+ CD43– CD19+ IgM) was analyzed with flow cytometry (n=6 in each). (F) Indicated transcript levels in pro-B cells (top panel) and pre-B cells (bottom panel) were evaluated with real-time polymerase chain reaction and normalized to the median of WT samples (WT, n=9; KO, n=4; Tg, n=8).

430

haematologica | 2021; 106(2)

Effects of STAP-2 on B-cell recovery

Figure 6. STAP-2 deficiency attenuates the inhibitory effects of lipopolysaccharide (LPS) in pre-B cells. (A and B) Phosphorylation status of P38/MAPK (pT180/pY182), ERK1/2 (pT202/pY204), and STAT5 (pY649) in B220+ cells from wild-type (WT) or STAP-2 transgenic (Tg) mice, under steady state (A) and 12 hours after sub-lethal irradiation (B), were analyzed by flow cytometry. (C-E) Three days after 1.0 mg /kg of LPS had been administered intraperitoneally, flow cytometric analysis was conducted. Number of B-lineage progenitor subsets (pro-B; B220+ CD43+ CD19+ IgM–, pre-B; B220+ CD43– CD19+ IgM–, immature B; B220+ CD43– CD19+ IgM+) in bone marrow (BM) (C), as well as white blood cells (WBC) and B220+ CD19+ B cells in peripheral blood (D) are shown. BrdU was administered 12 hours before analysis of the cell cycle status in B220+ CD19+ cells (E, left). Cell apoptosis was analyzed by Annexin V and 7AAD in B220+ cells using flow cytometry (E, right). Data represent three independent experiments (n=6 in each). (F-I) Indicated concentration of LPS was added into CFU pre-B (F) or B-progenitor suspension cultures (G-I) using cells derived from STAP-2 knockout (KO), Tg, or WT mice. Recovered colony (F) or cell (G) numbers were counted, and the proportion normalized to cells without LPS stimulation (control) were calculated. (H) Cell apoptosis was analyzed by Annexin V and 7AAD using flow cytometry after 4-day cultures with LPS (1.0 mg/mL). (I) Differentiation to IgM+ immature cells was evaluated via flow cytometry. Results are shown as mean±standard deviation. Statistical significances relative to WT were determined by unpaired two-tailed Mann-Whitney tests: *P<0.05 and **P<0.01.

haematologica | 2021; 106(2)

431

M. Ichii et al. A

Figure 7. STAP-2 expression under hematologic stress. (A-C) The indicated subsets (pro-B; B220+ CD43+ CD19+ IgM–, pre-B; B220+ CD43– CD19+ IgM–, Lin(–); lineage marker negative, LSK; Lineage– Sca1+ cKithigh, common lymphoid progenitor [CLP]; Lin– Sca1+ cKitlow Flk2high IL7Rα+) of hematopoietic cells derived from wild-type (WT) mice three days after LPS administration (A) (n=6 in each), or 12 hours (B) (control, n=6; with irradiation, n=13) or one month (C) (control, n=3; with irradiation, n=5) after sub-lethal irradiation (3.5 Gy) were analyzed for Stap-2 expression by real-time polymerase chain reaction (RT-PCR). Stap-2 transcript levels were normalized to the median of control samples without treatment. (D) Indicated cytokines (IFN-γ, 0.1 ng/mL; IL-6, 50 ng/mL; IL-1β, 5.0 ng/mL; TNFα, 50 ng/mL; LPS, 5.0 mg/mL) were added to B-progenitor cultures. Cells were collected after 24-hour stimulation, and RT-PCR was performed. Results are expressed as fold change relative to the controls, and are representative of results obtained in three independent experiments. Results are shown as mean ± standard deviation. Statistical significances relative to controls without stimulation were determined by unpaired two-tailed Mann-Whitney tests: *P<0.05.

mised by STAP-2. Sub-lethal irradiation eliminates most hematopoietic cells, leading to rapid proliferation and differentiation of HSPC, and STAP-2 requirement was assessed following 3.5 Gy sub-lethal irradiation (Figure 3C). Four days post irradiation, loss of myeloid, T, and B cells was observed in both WT and Tg mice, whose total white blood cell counts were indistinguishable before and after irradiation. However, while control mice regained normal levels of B cells two weeks after irradiation, almost no recovery was observed in STAP-2 Tg mice at that stage. Analyses of the proliferation and apoptosis status after irradiation revealed that B progenitors in Tg mice showed more progressive cell division, and rapid induction of apoptosis (Figure 3D). To test the effect of STAP-2 on undamaged B progenitors following direct stress, B-cell recovery was monitored after congenic BM transplantation. One month after transplantation to lethally irradiated recipients, CD45.2+ donor chimerism in blood B cells of Tg LSK-transplanted recipients was significantly lower than in WT transplanted mice (WT donor, 63.3±7.5%; STAP-2 Tg donor, 23.0±2.8%; P<0.001). The percentage and cellularity of BM B progenitors one month after transplantation were significantly reduced in Tg LSK transplanted mice compared to control (Figure 3E). The suppression of B-cell recovery lasted for 4 months at the end of follow-up. Taken together, we concluded that STAP-2 governs B-cell reconstitution following hematologic stress. 432

STAP-2 impairs proliferation of B progenitor at the pre-B stage To gain further insight into STAP-2 functions, we evaluated B-cell differentiation of Stap-2 gene targeted mice in vitro. The capacity for B-cell lineage development was reduced, when HSPC derived from Tg were cultured under stromal-cell free, B-cell conditions in the presence of IL-7, FL and SCF (Figure 4A). The same was true in cultures of LSK cells as well as CLP with or without OP9 stromal cells (Figure 4B and C). After rearrangement and expression of immunoglobulin (Ig) heavy chain loci at the pro-B stage, large pre-B cells enter cell-cycle and undergo several rounds of proliferation. The Ig light chain gene is then rearranged, and cells begin to express a functional B-cell receptor, progressing to an IgM+ immature B-cell stage. Colony-forming unit (CFU) assays were used to study the effect on pre-B-cell proliferation, and we found that STAP-2 deficiency promoted the generation of pre-B colonies, while overexpression of STAP-2 inhibited colony recovery with statistical significance (recovered colony number; 90.0±9.5 in WT, 40.7±3.1 in Tg; P<0.001, and 200.0±5.3 in KO; P<0.001) (Figure 4D). Under stromal-cell free culture conditions of Mac1– B220+ CD19+ CD43+ IgM– pro-B cells, STAP-2 regulated the expansion in the same way as was found in the CFU assay, while the differentiation into immature B cells was unaffected (Figure 4E). These results indicate that STAP-2 is involved in the haematologica | 2021; 106(2)

Effects of STAP-2 on B-cell recovery

Figure 8. STAP-2 attenuates B-cell recovery under hematologic stress. A graphic depiction of hematologic stresses that variably regulate STAP-2 mRNA expression via the direct effects of several hematologic stress factors, such as TNFα, as well as indirect effects imposed by the affected microenvironment. STAP-2 impairs B-cell recovery under hematologic stress.

proliferation of pre-B cells leading to compromised B-cell recovery during hematologic stress.

Inflammatory signals upregulated following hematopoietic stem cell transplantation are exaggerated by STAP-2 To determine the molecular basis for the inefficient B lymphopoiesis in STAP-2 Tg BM, transcriptome analysis was used to identify the pathways regulated by STAP-2 under hematologic stress. We evaluated global gene expression profiles of pre-B cells derived from WT and Tg BM under steady-state as well as WT and KO BM one month post transplantation by RNA-sequencing (RNASeq) experiments (Figure 5). The number of significantly regulated transcripts is shown in Figure 5A. First we compared pre-B cells isolated one month after transplantation with cells under homeostasis. It was expected that pathways related to hematologic development and genes with lymphoid signatures would be significantly regulated (Figure 5B). Furthermore, a bioinformatic approach with Ingenuity Pathway Analysis revealed that groups of genes involved in inflammatory responses were strongly upregulated during stress hematopoiesis. Differential display between WT and KO after transplantation showed that the deletion of STAP-2 prevented inflammatory changes induced by stress. STAP-2 overexpression activated these pathways even under steadystate conditions. Next, we investigated which signals were commonly haematologica | 2021; 106(2)

regulated among the following three pairs: WT with or without transplantation, WT or KO with transplantation, and WT or Tg under steady state (Figure 5C and D). As a result, pathways related to cytokines (IFN, TNF, IL-1β and IL-6), innate immune system (TLR and complement system), NFκB and p38 MAPK were extracted. Among them, the pathway activated to the greatest degree by STAP-2 and regenerative stress triggered receptor expression on myeloid cells type-1 (TREM-1) signaling (Figure 5C). This molecule has recently been shown to play an important role in controlling the intensity of innate immune responses.25-27 Upregulation of TREM-1 was verified by quantitative PCR and flow cytometry (Online Supplementary Figure S2A-C). Similar to macrophages, the activation of TLR4 signaling in Baf-3 pre-B cells upregulated the expression of TREM-1 (Online Supplementary Figure S2D). There were no shared pathways involved in B-cell development, and protein expression of IL7R, and mRNA expression of E2a, Ebf1, Pax5, Rag1 and Rag2 were not regulated under these conditions (Figure 5E and F).

Inflammatory network regulates stress B lymphopoiesis in bone marrow Our findings from RNA-Seq experiments indicate STAP-2 modulation with inflammatory signals delays Bcell recovery after hematologic stress. This was also confirmed with the evaluation of phosphorylated forms of the target proteins of inflammatory signaling pathways, such as JAK/STAT and MAPK in B progenitors in vivo. 433

M. Ichii et al.

When the levels of p38 MAPK, ERK1/2 and STAT5 phosphorylation in B progenitors were evaluated under steady state, there were no differences between WT and Tg (Figure 6A). After mice were irradiated, the phosphorylation was activated. We found that STAP-2 overexpression exaggerated p38 phosphorylation significantly, compared to WT B progenitors (Figure 6B). While several inflammatory molecules including LPS are known to be activators of B cells at the maturation stage following export to the periphery, recent studies showed that the TLR4 signaling pathway inhibits proliferation of pro-B and pre-B cells, and promotes the differentiation to IgM+ immature B cells in BM.28,29 Consistent with previous studies, the suppression of B-cell differentiation in BM was observed 3 days following sub-lethal dosing of LPS treatment (1.0 mg/kg) (Figure 6C). When Tg mice were treated, the actual B-cell number in peripheral blood decreased and a significant reduction of B progenitors in BM was observed (Figure 6C and D). The proliferation and apoptosis analyses, including BrdU and Annexin V staining methods, revealed a similar pattern to those after sub-lethal irradiation in vivo, indicating that the same mechanism is employed under these hematologic stressors (Figure 6E). Treatment of LPS-induced septic Tg mice with the p38 MAPK inhibitor SB203580 increased the number of B progenitors in BM, but this increase was not statistically significant (Online Supplementary Figure S3). To study the specific interaction between each signal and STAP-2, pre-B-culture systems were used. This made it possible to eliminate the influence of HSPC and other lineage cells, which are also targets of the same signals such as IFN, TNFα, IL-1β, MAPK, and TLRs.5-10 When LPS was added to CFU pre-B cultures, we found that deletion of STAP-2 attenuated the inhibitory effect of LPS (Figure 6F). The cancelation, although incomplete, was statistically significant, and the same was true in stroma-free liquid cultures (Figure 6G). Similar to reports following LPS treatment in vivo, apoptosis was induced (Figure 6H); however, BrdU analysis revealed that differences were not observed with or without treatment (data not shown). Moreover, STAP-2 deficiency led to a statistically significant reduction in the effects (Figure 6H). Regarding effects on differentiation, the expression of IgM after one week of culture was identical between WT and KO pre-B-cell cultures (Figure 6I). The limited effects on TLR4 signaling indicates the existence of other pathways involved with STAP-2. The addition of IFNγ, TNFα, IL-1β and IL-6 into pre-B-cell cultures did not induce statistically significant differences between WT and KO cultures (data not shown). Our previous studies showed that Stap-2 mRNA is induced in murine hepatocytes in response to stimulation by LPS and IL-6.14 However, the mRNA level in pre-B or pro-B cells did not change after LPS treatment in vivo, indicating the sustained protein level is sufficient to serve as the suppressor of B-progenitor proliferation (Figure 7A). In contrast, we found the expression level varied after sublethal irradiation (Figure 7B). When 13 mice were irradiated, the expression in pre-B cells was upregulated in five mice (38.5%) while in the same number of mice that was completely lost. The same was observed in pro-B cells. Interestingly, Stap-2 expression in lineage negative cells was increased approximately four times compared to controls. The changes recovered to normal levels one month after irradiation (Figure 7C). These results indicated that Stap-2 mRNA expression may be regulated via the direct 434

effects of several hematologic stress factors, as well as indirectly by the affected microenvironment. Lastly, we sought to determine if the upregulated inflammation during hematologic stress could alter STAP2 expression in pre-B cells. Several cytokines such as IFNγ, IL-6, IL-1β, TNFα, and LPS were added to pre-B-cell cultures, and the expression of Stap-2 mRNA was analyzed after 24 hours of stimulation (Figure 7D). Interestingly, only TNF significantly increased Stap-2 expression. These results indicate that inflammatory signals during hematologic stress may use STAP-2 for negative regulation of B lymphopoiesis.

Discussion Here we demonstrate that STAP-2 governs early B lymphopoiesis in BM during hematopoietic stress. While STAP-2 did not influence stemness of HSC or homeostatic hematopoiesis, short-term recovery of B cells was impaired by STAP-2 following transplantation. This effect was due to the inhibited proliferation of pre-B cells. RNASeq analyses showed that activation of several inflammatory signals occurs one month after transplantation in Tg mice. Culture experiments indicated developmental stagespecific inhibitory regulation of the TLR4-STAP-2 pathway as one of the mechanisms. Our study unveiled the critical role of STAP-2 in B-cell lineage progenitors in short-term responses to hematologic stress (Figure 8). Previously, we and others reported that STAP-2 functions as an adaptor protein in a variety of signaling pathways, including immune responses and tumorigenesis. STAP-2 regulates cell growth, integrin-mediated adhesion, migration induced by CXCL12, and apoptosis in T cells.17,30,31 The production of cytokines in macrophages is also affected by STAP-2. In tumorigenesis, STAP-2 binds to BCR-ABL in chronic myeloid leukemia cells, and also interacts with Brk, which mediates STAT5 activation in breast cancers.21,22 Recent studies reported that STAP-2 has a significant impact on disease severity in mouse models of IgE allergy and inflammatory bowel disease.18,24 In this study, we show a novel function of STAP-2 during stress hematopoiesis. Consistent with previous reports, STAP-2 modulates inflammatory signals such as TLR4 in pre-B cells. Sustained hematopoiesis is regulated by various mechanisms, including cell-intrinsic and -extrinsic factors, and the role of HSC in hematologic stress has been intensively studied.1-3,5,9 Unlike HSC, lineage-committed progenitors lose multipotency, yet retain the ability to proliferate during differentiation. In murine erythropoiesis, it is reported that robust expansion of erythroid progenitors in BM and spleen occurs to counteract anemic stress using signals different from those in homeostatic conditions.32,33 In B lymphopoiesis, various factors such as infection, aging and GCSF administration, are known suppressors of BM B lymphopoiesis.34-38 However, some studies focused on the immune functions of B lymphocytes,39 and the effects of G-CSF on B progenitors in BM are indirect.36,37 How B progenitors in BM respond to stress is not well understood. In the current study, we show that STAP-2 impairs pre-Bstage proliferation and delays the recovery of B cells during hematologic stress, such as transplantation and irradiation. The proliferation of lineage-committed progenitors appears to be more important in hematologic regeneration than previously believed. haematologica | 2021; 106(2)

Effects of STAP-2 on B-cell recovery

Interestingly, pre-B cells were strongly affected by various inflammatory signals at the early phase after hematopoietic stem cell transplantation (Figure 5). Among these signals, we focused on the TLR4 pathway. B-lineage cells express multiple pathogen recognition receptors, such as TLR, from the earliest stage.7,40 As far as the impact on B progenitors in BM is concerned, two research groups previously showed that signaling through TLR4 inhibits pro-B and pre-B proliferation, and enhances differentiation to IgM+ immature B cells, using different methods.28,29 In our genetically modified mice, STAP-2 displayed synergistic effects on pre-B-cell proliferation with the TLR4 pathway. However, STAP-2 effects did not involve differentiation to IgM+ immature B cells. In macrophages, STAP-2 binds directly to MyD88 and IKKα/β but not to TNFR-associated factor 6 or IL-1R-associated kinase 1.16 As in macrophages, STAP-2 may modulate specific TLR4 functions. We also found that TREM-1 expression is upregulated by STAP-2 overexpression and LPS stimulation. Although it is known that TREM-1 is expressed on B-lineage cells, its potential roles in B lymphopoiesis remain unknown.25-27,41 When the antagonized antibody and neutralized Fc chimera was added into the pre-B-cell cultures, no effects were observed (data not shown). Thus, a gene-modified approach may be required to clarify the function. Considering the nature of STAP-2 as an adaptor protein, the total protein expression as well as its phosphorylation is essential for the acting process. Previously, we showed that phosphorylated STAP-2 binds directly to several types of inflammatory proteins, including MyD88.16,17,30,31 In this study, we evaluated the transcript under hematologic stress. As shown in Figure 7, the expression is sustained after LPS treatment. Alternatively, we observed expression changes according to the individual condition after sub-lethal irradiation, indicating that several direct and indirect factors affected by hematologic stress are associated with the expression of STAP-2. Among several candidates, TNFα upregulates the expression of STAP-2 in pre-B cells (Figure 7D). In clinical settings, prolongation of hematologic suppression after chemotherapy is sometimes observed when patients suffer from severe complications such as infection. Our findings indicate the existence of a complex network induced by hematologic stress under inflammatory milieu. Interestingly, we found Stap-2 expression in the lineage negative HSPC fraction was upregulated following irradiation (Figure 7B). Since the intensity of cKit decreases while that of Sca1 increases under hematologic stresses, such as irradiation and LPS administration, it is difficult to

References 1. Haas S, Trumpp A, Milsom MD. Causes and consequences of hematopoietic stem cell heterogeneity. Cell Stem Cell. 2018; 22(5):627-638. 2. Laurenti E, Gottgens B. From haematopoietic stem cells to complex differentiation landscapes. Nature. 2018;553(7689):418-426. 3. Pietras EM. Inflammation: a key regulator of hematopoietic stem cell fate in health and disease. Blood. 2017;130(15):1693-1698. 4. Baldridge MT, King KY, Goodell MA. Inflammatory signals regulate hematopoiet-

haematologica | 2021; 106(2)

sort LSK or CLP cells via flow cytometry, and, therefore we were unable to identify in which population HSPC expression increased. Based on our in vivo and in vitro results, STAP-2 in CLP may affect B lymphopoiesis under hematologic stress. In this study, we revealed a detailed mechanism underlying the impaired effects of STAP-2 on B-cell reconstitution under hematopoietic stress. The target of STAP-2 during stress hematopoiesis appears to be very specific to B progenitors in BM, and its expression is altered following exposure to hematologic stress, while the expression of STAP-2 is ubiquitous under steady state. Our findings provide insights that may aid in the development of new therapeutic approaches, with fewer adverse effects, or increased prognostic value, for patients given severe myeloablative therapy or allogeneic HSCT. Disclosures MI have received speakers bureau from Celgene, Kowa Pharmaceutical, Takeda Pharmaceutical and Novartis outside the submitted work. HS have received honoraria from Jansen, honoraria and research funding from Bristol-Meyer Squibb, Celgene, Fujimoto Pharmaceutical, Mundipharma, Novartis, Ono Pharmaceutical, and Takeda Pharmaceutical outside the submitted work. KO have received speakers bureau from Novartis outside the submitted work. YK have received consultancy, honoraria, and research funding from Alexion outside the submitted work. The remaining authors declare no competing financial interests. Contributions MI designed research, collected data, analyzed data, and wrote the paper. KO, HShib, and TM analyzed data, and wrote the paper. HShih, JT, HS, DM, and DO collected and analyzed data. YK, RM, JK and KS supplied materials. YK supervised the research and wrote the paper. All authors reviewed and approved the manuscript. Acknowledgments The authors would like to thank Dr. Kincade (Oklahoma Medical Research Foundation) for constructive suggestions. We are also grateful for Ms. Habuchi, and Mr. Takashima for technical support. Funding This work was supported in part by Japan Society for the Promotion of Science KAKENHI (grant no. 16K09872, 19H03364, 19K08815), Otsuka Pharmaceutical, and Kyowa Kirin Co., Ltd.

ic stem cells. Trends Immunol. 2011; 32(2):57-65. 5. Zhao JL, Ma C, O'Connell RM, et al. Conversion of danger signals into cytokine signals by hematopoietic stem and progenitor cells for regulation of stress-induced hematopoiesis. Cell Stem Cell. 2014; 14(4):445-459. 6. Liu A, Wang Y, Ding Y, Baez I, Payne KJ, Borghesi L. Cutting edge: hematopoietic stem cell expansion and common lymphoid progenitor depletion require hematopoieticderived, cell-autonomous TLR4 in a model of chronic endotoxin. J Immunol. 2015; 195(6):2524-2528.

7. Nagai Y, Garrett KP, Ohta S, et al. Toll-like receptors on hematopoietic progenitor cells stimulate innate immune system replenishment. Immunity. 2006;24(6):801-812. 8. Esplin BL, Shimazu T, Welner RS, et al. Chronic exposure to a TLR ligand injures hematopoietic stem cells. J Immunol. 2011;186(9):5367-5375. 9. Pietras EM, Mirantes-Barbeito C, Fong S, et al. Chronic interleukin-1 exposure drives haematopoietic stem cells towards precocious myeloid differentiation at the expense of self-renewal. Nat Cell Biol. 2016;18(6): 607-618. 10. Matatall KA, Jeong M, Chen S, et al. Chronic

435

M. Ichii et al. infection depletes hematopoietic stem cells through stress-induced terminal differentiation. Cell Rep. 2016;17(10):2584-2595. 11. Ogonek J, Kralj Juric M, Ghimire S, et al. Immune reconstitution after allogeneic hematopoietic stem cell transplantation. Front Immunol. 2016;7:507. 12. Pietras EM, Reynaud D, Kang YA, et al. Functionally distinct subsets of lineagebiased multipotent progenitors control blood production in normal and regenerative conditions. Cell Stem Cell. 2015;17 (1):35-46. 13. Karigane D, Kobayashi H, Morikawa T, et al. p38alpha activates purine metabolism to initiate hematopoietic stem/progenitor cell cycling in response to stress. Cell Stem Cell. 2016;19(2):192-204. 14. Minoguchi M, Minoguchi S, Aki D, et al. STAP-2/BKS, an adaptor/docking protein, modulates STAT3 activation in acute-phase response through its YXXQ motif. J Biol Chem. 2003;278(13):11182-11189. 15. Muraoka D, Seo N, Hayashi T, et al. Signaltransducing adaptor protein-2 promotes generation of functional long-term memory CD8+ T cells by preventing terminal effector differentiation. Oncotarget. 2017; 8(19):30766-30780. 16. Sekine Y, Yumioka T, Yamamoto T, et al. Modulation of TLR4 signaling by a novel adaptor protein signal-transducing adaptor protein-2 in macrophages. J Immunol. 2005;176(1):380-389. 17. Sekine Y, Ikeda O, Tsuji S, et al. Signal-transducing adaptor protein-2 regulates stromal cell-derived factor-1 alpha-induced chemotaxis in T cells. J Immunol. 2009; 183(12):7966-7974. 18. Sekine Y, Nishida K, Yamasaki S, et al. Signal-transducing adaptor protein-2 controls the IgE-mediated, mast cell-mediated anaphylactic responses. J Immunol. 2014; 192(8):3488-3495. 19. Takaki S, Tezuka Y, Sauer K, et al. Impaired lymphopoiesis and altered B cell subpopulations in mice overexpressing Lnk adaptor protein. J Immunol. 2003;170(2):703-710. 20. Fujisawa M, Sakata-Yanagimoto M,

436

Nishizawa S, et al. Activation of RHOAVAV1 signaling in angioimmunoblastic Tcell lymphoma. Leukemia. 2018;32(3):694702. 21. Sekine Y, Ikeda O, Mizushima A, et al. STAP-2 interacts with and modulates BCRABL-mediated tumorigenesis. Oncogene. 2012;31(40):4384-4396. 22. Ikeda O, Sekine Y, Mizushima A, et al. Interactions of STAP-2 with Brk and STAT3 participate in cell growth of human breast cancer cells. J Biol Chem. 2010; 285(49):38093-38103. 23. Sekine Y, Togi S, Muromoto R, et al. STAP-2 protein expression in B16F10 melanoma cells positively regulates protein levels of tyrosinase, which determines organs to infiltrate in the body. J Biol Chem. 2015; 290(28):17462-17473. 24. Fujita N, Oritani K, Ichii M, et al. Signal-transducing adaptor protein-2 regulates macrophage migration into inflammatory sites during dextran sodium sulfate induced colitis. Eur J Immunol. 2014;44(6):1791-1801. 25. Arts RJ, Joosten LA, van der Meer JW, Netea MG. TREM-1: intracellular signaling pathways and interaction with pattern recognition receptors. J Leukoc Biol. 2013;93(2):209215. 26. Ford JW, McVicar DW. TREM and TREMlike receptors in inflammation and disease. Curr Opin Immunol. 2009;21(1):38-46. 27. Sharif O, Knapp S. From expression to signaling: roles of TREM-1 and TREM-2 in innate immunity and bacterial infection. Immunobiology. 2008;213(9-10):701-713. 28. Li Q, Han D, Wang W, et al. Toll-like receptor 4-mediated signaling regulates IL-7-driven proliferation and differentiation of B-cell precursors. Cell Mol Immunol. 2014; 11(2):132-140. 29. Hayashi EA, Akira S, Nobrega A. Role of TLR in B cell development: signaling through TLR4 promotes B cell maturation and is inhibited by TLR2. J Immunol. 2005; 174(11):6639-6647. 30. Sekine Y, Yamamoto C, Kakisaka M, et al. Signal-transducing adaptor protein-2 modulates Fas-mediated T cell apoptosis by inter-

acting with caspase-8. J Immunol. 2012; 188(12):6194-6204. 31. Sekine Y, Tsuji S, Ikeda O, et al. Signal-transducing adaptor protein-2 regulates integrinmediated T cell adhesion through protein degradation of focal adhesion kinase. J Immunol. 2007;179(4):2397-2407. 32. Liao C, Hardison RC, Kennett MJ, Carlson BA, Paulson RF, Prabhu KS. Selenoproteins regulate stress erythroid progenitors and spleen microenvironment during stress erythropoiesis. Blood. 2018;131(23):2568-2580. 33. Peslak SA, Wenger J, Bemis JC, et al. EPOmediated expansion of late-stage erythroid progenitors in the bone marrow initiates recovery from sublethal radiation stress. Blood. 2012;120(12):2501-2511. 34. Ueda Y, Yang K, Foster SJ, Kondo M, Kelsoe G. Inflammation controls B lymphopoiesis by regulating chemokine CXCL12 expression. J Exp Med. 2004;199(1):47-58. 35. Melamed D, Scott DW. Aging and neoteny in the B lineage. Blood. 2012;120(20):41434149. 36. Winkler IG, Bendall LJ, Forristal CE, et al. Blymphopoiesis is stopped by mobilizing doses of G-CSF and is rescued by overexpression of the anti-apoptotic protein Bcl2. Haematologica. 2013;98(3):325-333. 37. Day RB, Bhattacharya D, Nagasawa T, Link DC. Granulocyte colony-stimulating factor reprograms bone marrow stromal cells to actively suppress B lymphopoiesis in mice. Blood. 2015;125(20):3114-3117. 38. Wang W, Org T, Montel-Hagen A, et al. MEF2C protects bone marrow B-lymphoid progenitors during stress haematopoiesis. Nat Commun. 2016;7:12376. 39. Moreau JM, Berger A, Nelles ME, et al. Inflammation rapidly reorganizes mouse bone marrow B cells and their environment in conjunction with early IgM responses. Blood. 2015;126(10):1184-1192. 40. Hua Z, Hou B. TLR signaling in B-cell development and activation. Cell Mol Immunol. 2013;10(2):103-106. 41. Matesanz-Isabel J, Sintes J, Llinas L, de Salort J, Lazaro A, Engel P. New B-cell CD molecules. Immunol Lett. 2011;134(2):104-112.

haematologica | 2021; 106(2)

ARTICLE

Gaucher Disease

Accuracy of chitotriosidase activity and CCL18 concentration in assessing type I Gaucher disease severity. A systematic review with meta-analysis of individual participant data

Tatiana Raskovalova,1 Patrick B. Deegan,2 Pramod K. Mistry,3 Elena Pavlova,2 Ruby Yang,3 Ari Zimran,4 Juliette Berger,5,6 Céline Bourgne,5,6 Bruno Pereira,7 José Labarère8,9 and Marc G. Berger5,6,10

Laboratoire d’immunologie, Grenoble University Hospital, Université Grenoble Alpes, Grenoble, France; 2Lysosomal Disorders Unit, Department of Medicine, University of Cambridge, Addenbrooke’s Hospital, Cambridge, UK; 3Pediatric Gastroenterology and Hepatology, Yale University School of Medicine, New Haven, CT, USA; 4Gaucher Clinic, Shaare Zedek Medical Center, Hebrew University-Hadassah Medical School, Jerusalem, Israel; 5CHU Clermont-Ferrand, Hôpital Estaing, Hématologie Biologique, ClermontFerrand, France; 6Université Clermont Auvergne, EA 7453 CHELTER, Clermont-Ferrand, France; 7DRCI, CHU Clermont-Ferrand, Clermont-Ferrand, France; 8Quality of Care Unit, INSERM CIC1406, Grenoble University Hospital, Université Grenoble Alpes, Grenoble, France; 9TIMC-IMAG, UMR 5525 CNRS, Université Grenoble Alpes, Grenoble, France and 10 CHU Clermont-Ferrand, Service d’Hématologie Clinique Adulte et Thérapie Cellulaire, Hôpital Estaing, Clermont-Ferrand, France 1

Ferrata Storti Foundation

Haematologica 2021 Volume 106(2):437-445

ABSTRACT

hitotriosidase activity and CCL18 concentration are interchangeably used for monitoring Gaucher disease (GD) activity, together with clinical assessment. However, comparative studies of these two biomarkers are scarce and of limited sample size. The aim of this systematic review with meta-analysis of individual participant data (IPD) was to compare the accuracy of chitotriosidase activity and CCL18 concentration for assessing type I GD severity. We identified cross-sectional and prospective cohort studies by searching Medline, EMBASE, and CENTRAL from January 1995 to June 2017, and by contacting research groups. The primary outcome was a composite of liver volume >1.25 multiple of normal (MN), spleen volume >5 MN, hemoglobin concentration <11 g/dL, and platelet count <100x109/L. Overall, IPD included 1,109 observations from 334 patients enrolled in nine primary studies, after excluding 111 patients with undocumented values and 18 patients with deficient chitotriosidase activity. IPD were unavailable for 14 eligible primary studies. The primary outcome was associated with a 5.3-fold (95% Confidence Interval [CI]: 4.2-6.6) and 3.0-fold (95% CI: 2.6-3.6) increase of the geometric mean for chitotriosidase activity and CCL18 concentration, respectively. The corresponding areas under the receiver operating characteristics curves were 0.82 and 0.84 (summary difference, 0.02, 95% CI: -0.02 to 0.05). The addition of chitotriosidase activity did not improve the accuracy of the CCL18 concentration. Estimates remained robust in the sensitivity analysis and consistent across subgroups. Neither the chitotriosidase activity nor the CCL18 concentration varied significantly according to a recent history of bone events among 97 patients. In conclusion, the CCL18 concentration is as accurate as chitotriosidase activity in assessing hematological and visceral parameters of GD severity and can be measured in all GD patients. This meta-analysis supports the use of CCL18 rather than chitotriosidase activity for monito-ring GD activity in routine practice. haematologica | 2021; 106(2)

Correspondence: JOSÉ LABARÈRE jlabarere@chu-grenoble.fr MARC G. BERGER mberger@chu-clermontferrand.fr Received: August 21, 2019. Accepted: January 20, 2020. Pre-published: January 30, 2020. https://doi.org/10.3324/haematol.2019.236083

437

T. Raskovalova et al.

Introduction Gaucher disease (GD; OMIM#230800) is a recessively inherited lysosomal sto-rage disorder caused by biallelic mutations in the GBA1 gene that encodes the lysosomal acid Î˛-glucosidase.1 The metabolic defect in GD results in bone marrow and visceral organ infiltration by Gaucher cells (i.e., glucosylceramide-laden macrophages), leading to anemia, thrombocytopenia, hepatosplenomegaly, and skeletal manifestations.2-4 Three main GD types (I, II and III) have been described, on the basis of the clinical features and age of onset. Type I (i.e., non-neuropathic) GD is predominant (85-94%).5 GD natural course is rather unpredictable, even within subgroups of patients with the same GBA1 mutation.6 Therefore, evaluating the disease severity and prognosis is challenging for clinicians. In this context, a surrogate blood biochemical marker is highly desirable for assessing GD severity and helping decision-making for specific therapy initiation or adjustment.7 Several biomarkers have been identified, including chitotriosidase, CCL18, and glucosylsphingosine.8-10 Chitotriosidase, the human analogue of the non-vertebrate chitinase, is directly secreted by Gaucher cells and is considered an indicator of the overall Gaucher cell burden.11 In patients with GD, chitotriosidase activity is about 1,000-fold higher than the normal values, and its level correlates with liver and spleen volume, hemoglobin concentration, platelet count, and some bone manifestations.12,13 Plasma chitotriosidase activity decreases dramatically after initiation of enzyme replacement therapy (ERT) and rises again when the treatment is stopped.14 However, plasma chitotriosidase activity has major limitations for monitoring GD activity. Indeed, measuring plasma chitotriosidase activity is technically complex and not standardized across laboratories.15 Moreover, 6% of the general population is homozygous for a chitotriosidase variant harboring a 24-base pair duplication in the CHIT1 gene and is deficient in chitotriosidase activity.16 In addition, 35% of the general population is heterozygous for this chitotriosidase variant, and displays about half of the activity observed in people with wild-type CHIT1. Finally, other CHIT1 gene polymorphisms have been reported that slightly impair the enzyme activity.11 CC chemokine ligand 18 (CCL18), originally named pulmonary and activation-regulated chemokine (PARC), is also directly secreted by Gaucher cells and is considered an indicator of the overall Gaucher cell burden.8 Plasma CCL18 concentration in GD patient is 10- to 50-fold higher than in healthy subjects.8 CCL18 concentration correlates with the liver and spleen volume, platelet count, history of osteonecrosis, and number of anatomical sites of osteonecrosis.12,17 Importantly, the CCL18 concentration is not subject to genetic variation and can be measured in all patients, including those with deficient chitotriosidase activity.8 Hitherto, both chitotriosidase activity and CCL18 concentration are used for monitoring GD activity and assessing response to treatment, based on the findings from single-center studies of relatively limited sample size. The lack of large scale, multicenter, head-to-head comparison studies have hindered full validation of these biomarkers and formulation of context of use in the clinical practice. We hypothesized that evidence on the comparative accuracy of these two biomarkers could be 438

strengthened by the secondary analysis of individual participant data (IPD) from primary studies on patients with GD. Therefore, our primary objective was to compare the accuracy of CCL18 concentration and chitotriosidase activity for assessing hematological and visceral parameters of type I GD severity. The secondary objective was to compare the accuracy of these two biomarkers for discriminating type I GD patients with symptomatic bone events.

Methods This systematic review with IPD meta-analysis was performed according to current guidelines18,19 and complied with the Preferred Reporting Items for Systematic review and MetaAnalysis (PRISMA)-IPD statement.20 The rationale and methods were pre-specified and reported in a protocol21 registered at PROSPERO (CRD42015027243). This meta-analysis was carried out on data from primary studies for which ethical approval had been obtained by the investigators. The ComitĂŠ de Protection des Personnes Sud Est 6, Clermont-Ferrand, France (IRB 00008526) reviewed the protocol and considered that it did not qualify for biomedical research requiring patient informed consent, provided that no supplementary data would be collected from the participants enrolled in primary studies.21

Eligibility criteria Eligible studies included cross-sectional and cohort studies that measured both chitotriosidase activity and CCL18 concentration at baseline and/or at follow-up in consecutive patients with type I GD (Online Supplementary Materials and Methods). Studies with fewer than 10 participants were excluded from this systematic review.

Information sources Studies were identified by searching Medline, EMBASE, and Cochrane Central Register of Controlled Trials (CENTRAL) from January 1995 to June 2017. Our electronic search was supplemented by scanning the reference lists of the retrieved articles and by contacting research groups.

Search strategy The search concepts included chitotriosidase activity, CCL18, biological markers, ERT, and GD (Online Supplementary Appendices S1-3). No restriction of document type and language was applied, and no methodology filter was used.

Study selection Two review authors (TR and JL) assessed potentially relevant full-text articles against pre-specified eligibility criteria.

Data collection Information on primary studies were collected using a standardized data extraction form. Where possible, IPD were extracted from published articles. Otherwise, the corresponding authors or principal investigators were invited to collaborate in this systematic review project by supplying de-identified IPD.21 The requested IPD included baseline characteristics and pre-specified outcomes.21 Organ volumes were expressed as multiples of normal (MN) adjusted for body weight.

Outcomes The primary outcome was a composite of hemoglobin concentration <11 g/dL (<10 g/dL for patients aged 12-59 months of age), haematologica | 2021; 106(2)

Biomarkers in Gaucher disease

platelet count <100x109/L, spleen volume >5 MN, and liver volume >1.25 MN. The secondary outcomes included symptomatic bone events, a composite of hemoglobin concentration <8 g/dL (<7 g/dL for patients 12-59 months of age), platelet count <50x109/L, spleen volume >15 MN, and liver volume >2.5 MN, and individual components of the primary and secondary composite outcomes. Bone events included skeletal fracture,

osteonecrosis, or avascular necrosis that occurred within the previous 12 months of biomarker analysis.21 All outcomes and cut-off values for continuous parameters were set according to published guidelines or previous studies,22,23 and were pre-specified.21

Statistical analysis We performed logarithm transformation for chitotriosidase

Figure 1. Identification and selection of studies for the meta-analysis of individual participant data. *Seventy three additional participants (222 observations) had splenectomy at baseline. IDP: individual participant data.

haematologica | 2021; 106(2)

439

T. Raskovalova et al.

activity and CCL18 concentration and derived geometric mean ratios along with 95% Confidence Intervals (CI), in order to account for skewed distributions.24,25 The comparative accuracy of CCL18 concentration relative to chitotriosidase activity in discriminating patients with the outcomes of interest was quantified by the difference in the areas under the receiver operating characteristic (AUC-ROC) curves. Data synthesis was performed with one- and two-stage approaches.26,27

Results Study selection and obtained IPD Overall, 2,636 records were identified by database searching (Figure 1). Two additional records were found by contacting field experts and searching clinical trial registries, respectively. After removing 13 duplicates, the titles and abstracts of 2,625 records were screened for eligibility. Of these, 318 records were identified as being potentially relevant and full-text articles were retrieved for a more thorough review. After excluding 295 records based on the full-text article, our systematic review included 23 primary studies from which IPD were sought. IPD were available only for nine of these studies totaling 463 patients with type I GD (Figure 1). The sponsor of four trials of ERT with velaglucerase28-31 and the principal investigators of three observational studies17,32,33 supplied computerized IPD upon request. The IPD of the other two randomized controlled trials on ERT with taliglucerase were extracted from the published articles34,35 (Online Supplementary Appendix S4). Our meta-analytical sample consisted of 1,109 observations from 334 patients with type I GD, after excluding 111 patients with undocumented values for chitotriosidase activity and/or serum CCL18 concentration, and 18 patients with deficient chitotriosidase activity (Figure 1). The median number of observations per patient was 2 (range: 1-14). IPD were not available for the other 14 eligible primary studies that included 11 academic observational studies8,9,14,36-43 and three industry-sponsored clinical trials of ERT or substrate reduction therapy44-46 (Online Supplementary Appendix S5). Overall, these studies enrolled 565 patients with GD, although overlap in study populations could not be excluded for five studies carried out in the Netherlands8,9,14,37,42 (201 patients) and four studies performed in Spain36,39,40,41 (203 patients).

Study characteristics Among the nine studies that supplied IPD, six were international industry-funded clinical trials of ERT28-31,34,35 and three were observational in design17,32,33 (Online Supplementary Appendix S4). The study enrollment periods extended from 2003 to 2015, and the length of the followup ranged from 12 to 132 months for longitudinal studies. The median number of primary study participants and observations contributing to the meta-analysis was 32 (range: 9 -98) and 117 (range: 20-224), respectively. All but one study recruited adult or mixed populations. The exception was a clinical trial performed in a pediatric setting.35 Children younger than 16 years of age accounted for 13% of all participants. Chitotriosidase activity was measured using the 4MUdeoxy-chitobiose28-31,33,35 or 4MU-chitotriose17,32 fluorogenic substrates, in compliance with the published methods. Serum CCL18 concentration was assayed using DELFIA or 440

ELISA in five and three studies, respectively (Online Supplementary Appendix S6). Information on chitotriosidase activity and serum CCL18 assays was not available for one study.34 The median values ranged from 1,340-14,809 nmol/mL/h for chitotriosidase activity, and from 237-1,113 ng/mL for serum CCL18 concentration. In five clinical trials,28-31,35 chitotriosidase activity and serum CCL18 concentration were both measured at a single central core laboratory (i.e., the Academic Medical Center in Amsterdam, the Netherlands). Liver and spleen volumes were quantified using magnetic resonance imaging in eight studies, assessed by independent blinded reviewers in five studies, and undocumented in one study. The median values in primary studies ranged from 0.8-1.7 MN for liver volume, 2.714.1 MN for spleen volume, 11.7-14.0 g/dL for hemoglobin concentration, and 82-260 x109/L for platelet count (Online Supplementary Appendix S6).

Study quality assessment Six of the eight studies that evaluated the primary composite outcome fulfilled five or more QUADAS-2 tool criteria (Online Supplementary Appendix S7). The other two studies met four QUADAS-2 tool criteria. As data were collected by retrospective chart review, the Yaleâ&#x20AC;&#x2122;s National Gaucher Disease Treatment Center study33 was considered at high risk of bias for index tests and pre-specified surrogate outcome assessment. This study was also considered at high risk of flow bias due to undocumented chitotriosidase activity or serum CCL18 concentration in 68% (113 of 167) of participants. In a randomized controlled trial that enrolled only treatment-naive patients, the applicability concern was high and it was not possible to formally determine whether biomaker assessment was blinded to the pre-specified outcomes and which analytical methods were used.34

Chitotriosidase activity and serum CCL18 concentration according to the outcomes In one-stage meta-analysis involving 492 observations nested within 177 participants from eight primary studies (Figure 1), the primary composite outcome was associated with increased geometric mean of chitotriosidase activity (7,623 vs. 1,478 nmol/mL/h; geometric mean ratio: 5.29, 95% CI: 4.24-6.61, P<0.001) and CCL18 concentration (679 vs. 198 ng/mL; geometric mean ratio: 3.04, 95% CI: 2.57-3.58, P<0.001) compared with the absence of outcome (Table 1). Despite substantial between-study heterogeneity, the two-stage meta-analysis yielded comparable results, using random-effect models (Online Supplementary Appendices S8-9). The effect size was quite homogeneous among the individual components of the primary composite outcome, with geometric mean ratio point estimates ranging from 2.96-5.43 for chitotriosidase activity and from 2.28-3.22 for serum CCL18 concentration (Table 1). Similar results were obtained for the secondary composite outcome and its individual components (Table 1), except for serum CCL18 concentration, which did not differ according to severe anemia (445 vs. 666 ng/mL, geometric mean ratio, 1.39, 95% CI: 0.82-2.37, P=0.22). The geometric means of chitotriosidase activity (3,556 vs. 1,618 nmol/mL/h; geometric mean ratio: 1.47, 95% CI: 0.78-2.79, P=0.24) and serum CCL18 concentration (786 vs. 449 ng/mL; geometric mean ratio: 1.22, 95% CI: 0.811.81, P=0.34) did not differ according to symptomatic haematologica | 2021; 106(2)

Biomarkers in Gaucher disease

bone events with imaging confirmation, among 218 observations nested within 97 participants from a single primary study (Table 1). The decrease in CCL18 concentration was paralleled by a similar trend in chitotriosidase activity over the 24 months of follow-up, among participants enrolled in four industry-sponsored clinical trials evaluating enzyme replacement therapy (ERT) (Online Supplementary Appendix S10).

Comparative accuracy of chitotriosidase activity and CCL18 concentration for assessing GD severity In one-stage meta-analysis, area under the curve receiv-

er operating characteristics (AUC-ROC) curve point estimates were 0.82 for chitotriosidase activity and 0.84 for serum CCL18 concentration for discriminating patients with GD according to the primary composite outcome (summary difference: 0.02, 95% CI: -0.02 to 0.05, P=0.32, Table 2). Adding chitotriosidase activity did not improve serum CCL18 concentration accuracy, as indicated by the AUC-ROC curve estimates (0.85 with and 0.84 without chitotriosidase activity, P=0.18) (Figure 2). The summary estimates of the difference in the AUCROC curves were consistent in the two-stage meta-analysis (Online Supplementary Appendix S11). No evidence of

Table 1. One-stage unpaired comparisons of chitotriosidase activity and CCL18 concentration according to pre-specified outcomes among type I Gaucher disease patients.

Outcomes

No.

Primary composite outcome† No outcome 212 ≥ 1 outcome 280 Secondary composite outcome‡ No outcome 391 ≥ 1 outcome 101 Hemoglobin concentration ≥ 11 g/dL 934 < 11 g/dL 102 Hemoglobin concentration ≥ 8 g/dL 1,029 < 8 g/dL 7 Platelet count ≥ 100x109/L 758 < 100x109/L 313 Platelet count ≥ 50x109/L 958 < 50x109/L 113 Liver volume < 1.25 MN 447 ≥ 1.25 MN 240 Liver volume < 2.5 MN 678 ≥ 2.5 MN 9 Spleen volume < 5 MN 240 ≥ 5 MN 265 Spleen volume < 15 MN 434 ≥ 15 MN 71 Symptomatic bone events# No 206 Yes 12

Chitotriosidase activity, nmol/mL/h Geometric mean Mean ratio (95%CI) (95% CI)*

CCL18, ng/mL Geometric mean Mean ratio (95% CI) (95% CI)*

<0.001 1,478 7,623

(1,235 - 1,768) (6,520 - 8,913)

1.00 (…) 5.29 (4.24 - 6.61)

<0.001 198 679

(177 - 221) (612 - 755)

1.00 3.04

(…) (2.57 - 3.58)

311 (283 - 342) 1,050 (879 - 1,254)

1.00 3.05

(…) (2.53 - 3.68)

406 (381 - 432) 1,057 (883 - 1,265)

1.00 2.28

(…) (1.97 - 2.64)

<0.001 2,701 (2,349 - 3,106) 1.00 (…) 13,516 (10,143 - 18,011) 4.35 (3.35 - 5.64)

<0.001

<.001 3,136 (2,836 - 3,468) 10,984 (7,841 - 15,386)

1.00 (…) 2.96 (2.44 - 3.59)

<0.001

.05 3,509 (3,178 - 3,876) 17,520 (8,076 - 38,005)

1.00 (…) 2.02 (1.00 - 4.10)

2,413 8,495

(2,156 - 2,700) (7,339 - 9,833)

1.00 (…) 4.03 (3.46 - 4.70)

3,121 (2,825 - 3,449) 8,890 (6,569 - 12,030)

1.00 (…) 2.94 (2.32 - 3.73)

2,242 (1,954 - 2,572) 10,181 (8,724 11,880)

1.00 (…) 3.96 (3.30 - 4.76)

0.22 445 666

(418 - 473) (331 - 1,340)

1.00 1.39

(…) (0.82 - 2.37)

343 856

(321 - 368) (782 - 937)

1.00 2.71

(…) (2.42 - 3.03)

412 928

(387 - 438) (786 - 1,094)

1.00 2.15

(…) (1.81 - 2.55)

317 886

(291 - 346) (800 - 980)

1.00 2.59

(…) (2.26 - 2.96)

447 (415 - 482) 1.00 1,479 (1,019 - 2,145) 2.25

(…) (1.35 - 3.74)

<.001

<0.001

<.001

0.001 3,700 (3,290 - 4,161) 1.00 (…) 30,353 (16,362 - 56,308) 3.16 (1.59 - 6.27)

0.002

<0.001 1,583 8,111

(1,338 - 1,873) (6,933 - 9,488)

1.00 (…) 5.43 (4.45 - 6.63)

<0.001 209 740

(188 - 232) (666 - 822)

1.00 3.22

(…) (2.77 - 3.75)

335 (306 - 367) 1.00 1,319 (1,076 - 1,616) 3.21

(…) (2.59 - 3.97)

449 786

(…) (0.81 - 1.81)

<0.001 2,871 (2,515 - 3,276) 1.00 (…) 18,539 (13,323 - 25,796) 4.14 (3.08 - 5.56)

<0.001

0.24 1,618 3,556

(1,335 - 1,961) (1,894 - 6,674)

1.00 (…) 1.47 (0.78 - 2.79)

0.34 (402 - 501) (407 - 1,518)

1.00 1.22

CI: confidence interval; MN: multiple of normal. *Summary geometric mean ratios and P-values for unpaired comparisons were derived from 3-level random intercept regression models for continuous dependent variables, with observations nested within patients and studies. †The primary outcome was a composite of hemoglobin concentration < 11 g/dL (< 10 g/dL for patients 12-59 months of age), platelet count < 100x109/L, spleen volume > 5 MN, and liver volume> 1.25 MN. Patients with splenectomy were excluded from this analysis. ‡The secondary outcome was a composite of hemoglobin concentration < 8 g/dL (< 7 g/dL for patients 12-59 months of age), platelet count < 50x109/L, spleen volume > 15 MN, and liver volume > 2.5 MN. Patients with splenectomy were excluded from this analysis. # Osteonecrosis or fracture with imaging confirmation within the previous 12 months.

haematologica | 2021; 106(2)

441

T. Raskovalova et al.

Subgroup analysis No evidence of heterogeneity in the biomarker levels associated with the primary composite outcome was observed across the age groups and provision of ERT (Online Supplementary Appendix S13). Conversely, lowerquality studies (<5 QUADAS-2 criteria) displayed higher geometric mean ratios for chitotriosidase activity, according to the primary outcome (Online Supplementary Appendix 13). Differences in the AUC-ROC curves for the primary composite outcome were homogeneous across the age groups, QUADAS-2 criteria, fluorogenic substrate, and CCL18 assay type (Online Supplementary Appendix S14). However, serum CCL18 concentration was more accurate than chitotriosidase activity in discriminating patients with the primary composite outcome among those receiving ERT (Online Supplementary Appendix S14). No evidence of heterogeneity in serum CCL18 concentration accuracy as a function of chitotriosidase activity deficiency was found (Online Supplementary Appendices S15-16).

Sensitivity analysis In leave-one-out sensitivity analysis that iteratively removed one study at a time, geometric mean ratios (Online Supplementary Appendix 17) and differences in the AUCROC curves (Online Supplementary Appendix 18) for the two biomarkers remained unchanged relative to the primary outcome, confirming that our results were not driven by any single study. In the additional sensitivity analysis, coding splenectomy as splenomegaly did not modify the associations of the primary and secondary composite outcomes with the two biomarkers (Online Supplementary Appendix S19), although the AUC-ROC curve for chitotriosidase activity decreased significantly (Online Supplementary Appendix S20).

Discussion Most published studies reporting on chitotriosidase activity and CCL18 concentration accuracy in assessing GD severity are of relatively limited sample size. In this context, our meta-analysis summarizes evidence from 1,109 observations nested within 334 participants enrolled in nine primary studies, with a broad range of patient age and disease severity. Our analysis indicates that CCL18 concentration is as accurate as chitotriosidase activity in discriminating patients on the basis of our primary composite outcome. This finding was robust in the sensitivity analysis, and we did not find evidence of between-study heterogeneity (IÂ˛=0%, P=0.68) in two-stage meta-analysis. Additionally, we did not detect any significant heterogeneity among 442

0.9 0.8 0.7 0.6

Sensitivity

selective reporting for the primary composite outcome was found graphically (Online Supplementary Appendix 12) or statistically (P=0.20). However, this result should be interpreted with caution, given the limited number of primary studies. Comparable effect sizes were observed for the secondary composite outcome and most of the individual components (Table 2). However, the AUC-ROC curve of serum CCL18 concentration for predicting severe anemia was lower than that of chitotriosidase activity (0.64 vs. 0.81, summary difference, -0.17, 95% CI: -0.33 to -0.08, P=0.008).

0.5 0.4 0.3 0.2 0.1 0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10 1 . Specificity (95% CI: 0.77-0.87)

(95% CI: 0.79-0.88)

(95% CI: 0.80-0.89)

Figure 2. The AUC-ROC curves receiver operating characteristics curves for chitotriosidase activity, CCL18 concentration, and the combination of these two biomarkers. CI: Confidence Interval. AUC: area under the curve; ROC: receiver operating characterists.

studies and patient subgroups, although CCL18 concentration may be more accurate in patients receiving ERT. Our primary result is also supported by consistency in the effect sizes for most of the individual outcome components and the secondary composite outcomes. Our meta-analysis corroborates and extends evidence from previous studies showing that chitotriosidase activity and CCL18 concentration relate to visceral and hematological parameter abnormalities in treated and untreated patients.8,12,13,15 A noteworthy exception was severe anemia (< 8 g/dL), for which CCL18 showed a lower AUC-ROC curve than chitotriosidase activity. This observation is consistent with our failure to show a significant difference in the CCL18 level for patients with and without severe anemia. Interestingly, weak or variable associations were previously reported between hemoglobin concentration and chitotriosidase activity or CCL18 concentration.12,13,15 A potential explanation might relate to the failure of physicians to correctly diagnose concurrent causes of severe anemia in GD patients. Yet, this cannot explain why the accuracy differed between chitotriosidase activity and CCL18 concentration. Due to the limited number of severe anemia cases in our meta-analytical sample (i.e., 7 among the 1,036 observations from 309 participants), our findings should be interpreted with caution and deserve confirmation in independent samples. Consistent with previous studies,47,48 we found that splenectomy altered the accuracy of chitotriosidase activity in discriminating patients with the primary outcome. Another striking result of our meta-analysis is that the addition of chitotriosidase activity did not improve the haematologica | 2021; 106(2)

Biomarkers in Gaucher disease Table 2. One-stage paired comparisons of AUC-ROC curves for chitotriosidase activity and CCL18 concentration for discriminating type I Gaucher disease patients with prespecified outcomes.

Outcome

n/N †

Primary composite outcome Secondary composite outcome‡ Hemoglobin concentration < 11 g/dL Hemoglobin concentration < 8 g/dL Platelet count < 100x109/L Platelet count < 50x109/L Liver volume > 1.25 MN Liver volume > 2.5 MN Spleen volume > 5 MN Spleen volume > 15 MN Symptomatic bone events#

280/492 101/492 102/1,036 7/1,036 313/1,071 113/1,071 240/687 9/687 265/505 71/505 12/218

AUC (95% CI)* Chitotriosidase activity CCL18 0.82 0.82 0.75 0.81 0.74 0.71 0.79 0.90 0.82 0.86 0.67

(0.77.- 0.87) (0.72 - 0.89) (0.65 - 0.82) (0.58 - 0.95) (0.68 - 0.79) (0.60 - 0.80) (0.74 - 0.84) (0.81 - 0.95) (0.77 - 0.87) (0.77 - 0.92) (0.40 - 0.83)

0.84 0.83 0.78 0.64 0.76 0.74 0.80 0.86 0.85 0.86 0.71

(0.79 - 0.88) (0.74 - 0.89) (0.69 - 0.84) (0.25 - 0.77) (0.71 - 0.81) (0.64 - 0.82) (0.76 - 0.84) (0.77 - 0.93) (0.80 - 0.89) (0.79 - 0.92) (0.40 - 0.88)

Difference in AUC (95% CI)* 0.02 0.01 0.02 -0.17 0.02 0.02 0.01 -0.04 0.03 0.00 0.04

(-0.02 to 0.05) (-0.04 to 0.05) (-0.04 to 0.11) (-0.33 to -0.08) (-0.02 to 0.06) (-0.03 to 0.10) (-0.02 to 0.05) (-0.11 to 0.05) (0.00 to 0.06) (-0.05 to 0.06) (-0.06 to 0.14)

P* 0.32 0.66 0.53 0.008 0.41 0.50 0.55 0.34 0.09 0.92 0.45

AUC-ROC: area under the curve receiver operating characteristics; CI: Confidence Interval; MN: multiple of normal. *Summary estimates for area under ROC curves and P-values for paired comparisons were derived from non-parametric ROC analysis with bootstrap resampling that accounted for observation clustering within patients and primary studies. †The primary outcome was a composite of hemoglobin concentration < 11 g/dL (< 10 g/dL for patients 12 to 59 months of age), platelet count < 100x109/L, spleen volume > 5 MN, and liver volume > 1.25 MN. Patients with splenectomy were excluded from this analysis. ‡The secondary outcome was a composite of hemoglobin concentration < 8 g/dL (< 7 g/dL for patients 12 to 59 months of age), platelet count < 50x109/L, spleen > 15 MN, and liver volume > 2.5 MN. Patients with splenectomy were excluded from this analysis. # Osteonecrosis or fracture with imaging confirmation within the previous 12 months.

accuracy of serum CCL18 concentration in identifying patients with the primary outcome, as shown by the unchanged AUC-ROC curve. This negative finding suggests that combining chitotriosidase activity and serum CCL18 concentration would have a limited incremental value, probably because they convey redundant information on Gaucher cell burden.14,49-51 The recent glucosylsphingosine biomarker9 may provide complementary information about the pathophysiological process in GD.14,33 Indeed, plasma glucosylsphingosine relates to sphingolipid turnover while chitotriosidase activity and CCL18 concentration are indicative of overall Gaucher cell mass. Glucosylsphingosine appeared to be highly sensitive and specific for the primary diagnosis and severity assessment of GD in previous studies.33,52 It would be interesting to compare the accuracy of glucosylsphingosine and CCL18 concentration using the same approach as in our meta-analysis although insufficient data exist to perform this analysis to date. Meanwhile a prospective comparative study of the three biomarkers is warranted to evaluate their contribution in monitoring GD activity. In contrast to CCL18, glucosylsphingosine can be measured in dried blood spot and has emerged as a second-tier biomarker for newborn screening for GD.53 Measuring biomarkers in dried blood spot samples might facilitate GD activity monitoring for clinics where the technology is not available. A limitation of our primary composite outcome is that it did not capture bone manifestations, despite their major impact on the quality of life.2,24 Previous studies reported that chitotriosidase activity level correlates with bone complications,13 history of osteonecrosis,17 and number of anatomical sites of osteonecrosis,17 although these associations were inconsistent.12,15 Similarly, CCL18 concentration has been inconsistently associated with history of osteonecrosis and number of anatomical sites of osteonecrosis.12,17 Unfortunately, information on symptomatic bone events was available in only one study included in our meta-analysis,17 and we could not reach a definitive conclusion for this outcome. haematologica | 2021; 106(2)

There are several potential explanations for the see-mingly inconsistent findings between our analysis of bone events and the study by Deegan et al.17 First, we used a stricter definition of bone manifestations, which included symptomatic major events (i.e., skeletal fracture, osteonecrosis, or avascular necrosis) that could be dated and excluded progressive alterations (i.e., osteoporosis, Erlenmeyer flask femur deformity). Second, we took into account the data temporality by including bone events that occurred within the previous 12 months of biomarker analysis while Deegan et al.17 analyzed the previous history of bone events that occurred during the lifetime. Third, the unit of analysis was different between our analysis (218 observations nested within 97 patients) and the study by Deegan et al.17 (100 patients). Two other clinical trials included in our meta-analysis recorded bone mineral density,30,34 but they did not report clinically relevant endpoints for bone disease (i.e., skeletal fracture, osteonecrosis, or avascular necrosis). Thus, the imprecise definition of diagnostic terms, the lack of standardized criteria for bone assessment, and the slow improvement in bone outcomes after specific treatment probably hamper the identification of a relationship between biomarkers and this important clinical outcome.5457 Another approach in assessing skeletal disease is to measure residual biomarker levels after remission of visceral disease (reversal of hepatomegaly in splenectomized patients or reversal of hepatosplenomegaly in intact spleen patients). Our findings have clinical implications for GD monitoring. Due to the comparable accuracy of the two biomarkers, the impact of common genetic polymorphism on chitotriosidase activity and the analytical robustness of CCL18 assays, it should be logical to recommend measuring CCL18 concentration rather than chitotriosidase activity as a biomarker of Gaucher cell burden and visceral and hematological damage. Measuring chitotriosidase activity in addition to CCL18 concentration does not appear useful in the routine practice or in clinical research. Indeed, the two biomarkers are strongly correlated and convey redundant information on 443

T. Raskovalova et al.

GD severity. Additionally, chitotriosidase activity has limited incremental value in improving CCL18 concentration accuracy, as reflected by the unchanged AUC-ROC curve estimate. Our meta-analysis demonstrates that chitotriosidase activity and CCL18 concentration relate to visceral and hematological parameters that can be easily monitored in the routine practice. Hence, additional studies are needed to further investigate the usefulness of these two biomar-kers. In the current practice, they are used to monitor an early response to treatment although their value for initiating specific treatment or adjusting treatment dosage remains questioned. Only prospective patient management studies or randomized controlled trials can establish the effectiveness of biomarker-guided therapy. The lack of robust data and measurement heterogeneity may also explain why some of the GD severity scores do not include biomarkers.58 Lastly, the relationships between GD biomarkers and patient-centered outcomes, including quality of life, fatigue, chronic pain, and restriction of daily activities should be evaluated.55,59 This meta-analysis analyzed IPD from nine primary studies that were completed since 2004, summarizing the most recent available evidence on the accuracy of chitotriosidase activity and CCL18 concentration in assessing type 1 GD severity. All studies included in this meta-analysis evaluated both biomarkers in the same patients, providing unconfounded comparative accuracy estimates.60 Moreover, our findings have strong generali-zability because we combined primary studies that enrolled various populations of patients worldwide.60 However, our meta-analysis also has a few caveats that must be considered. First, the finding interpretation is inevitably limited by the unavailability of IPD from the other 14 potentially eligible primary studies. Second, there was substantial between-study heterogeneity in chitotriosidase activity and, to a lesser extent, in CCL18 concentration. The lack of assay standardization may have contributed to this heterogeneity, although chitotriosidase activity and CCL18 concentration were measured at a single central core laboratory in five clinical trials. Therefore, we used mixed-effect regressions to account for this heterogeneity and performed sensitivity and subgroup analyses that supported the robustness of the summary estimates. In conclusion, CCL18 concentration is as accurate as chitotriosidase activity in assessing hematological and visceral parameters of GD activity and can be measured in patients

References 1. Zimran A, Elstein D. Chapter 72: Gaucher disease and related lysosomal storage diseases. In: Kaushansky K, Lichtman MA, Prchal JT, et al., eds. Williams Hematology. New York: McGraw-Hill Education. 2016:1121-1131. 2. Charrow J, Andersson HC, Kaplan P, et al. The Gaucher registry: demographics and disease characteristics of 1698 patients with Gaucher disease. Arch Intern Med. 2000; 160(18):2835-2843. 3. Grabowski GA, Zimran A, Ida H. Gaucher disease types 1 and 3: phenotypic characterization of large populations from the ICGG Gaucher Registry. Am J Hematol. 2015; 90(Suppl 1):S12-18.

444

who are homozygous for null chitotriosidase variants. However, the observed between-study heterogeneity and the limitations of this meta-analysis should encourage the international community to i) implement a strategy of homogenization and standardization of dosage techniques, more than 15 years after their implementation in clinical practice and ii) set up a prospective large-scale study to evaluate GD biomarkers, including glucosyl-sphingosine. This latter biomarker seems interesting from a pathophysiological perspective, but its superiority over chitotriosidase activity and CCL18 concentration remains to be documented. Disclosures PD received speaker and board membership fees from Takeda and consulting fees from Sanofi Corporation; PM received research grant, lecture fee honoraria, and travel support from SanofiGenzyme; AZ received consulting fees from Shire and Prevail Therapeutics, honoraria from Shire Corporation and Pfizer, he is an employee at the Gaucher Clinic, Shaare Zedek Medical Center, Israel, which received research grants from Shire Corporation, Centogene and Pfizer, and support from Shire Corporation, Pfizer and Sanofi-Genzyme for participation in their respective registries (GOS, TALIAS and ICGG). MGB received speaker fees and unrestricted research grants from SanofiGenzyme and Shire Corporation. The other authors have no conflict of interest relevant to this study. Contributions TR contributed to the study design, data acquisition, interpretation of the results, and manuscript preparation; PD, EP, PM, RY, AZ, JB, CB contributed to data acquisition and manuscript preparation; BP and JL contributed to the study design, data acquisition, data management, statistical analysis, interpretation of the results, and manuscript preparation; MB provided project leadership, contributed to the interpretation of the results, and manuscript preparation. All authors approved the final version of the manuscript. Acknowledgments The authors are indebted to the investigators of all primary studies for their contributions that allowed carrying out this secondary analysis. They are also grateful to Zoya Panahloo, Farid Merazi, Hak-Myung Lee, Bjorn Mellgard, Michael Craig, and Hartmann Wellhoefer from Shire, Lexington, MA, for sharing individual participant data from clinical trials and facilitating the pooled analysis. Statistical analysis was performed within the framework of the Grenoble Alpes Data Institute (ANR-15-IDEX-02).

4. Zimran A, Belmatoug N, Bembi B, et al. Demographics and patient characteristics of 1209 patients with Gaucher disease: descriptive analysis from the Gaucher Outcome Survey (GOS). Am J Hematol. 2018;93(2): 205-212. 5. Nalysnyk L, Rotella P, Simeone JC, et al. Gaucher disease epidemiology and natural history: a comprehensive review of the literature. Hematology. 2017;22(2):65-73. 6. Mistry PK, Belmatoug N, Vom Dahl S, et al. Understanding the natural history of Gaucher disease. Am J Hematol. 2015; 90(Suppl 1):S6S11. 7. Weinreb NJ, Aggio MC, Andersson HC, et al. Gaucher disease type 1: revised recommendations on evaluations and monitoring for adult patients. Semin Hematol. 2004; 41(4 Suppl 5):S15-22.

8. Boot RG, Verhoek M, de Fost M, et al. Marked elevation of the chemokine CCL18/PARC in Gaucher disease: a novel surrogate marker for assessing therapeutic intervention. Blood. 2004;103(1):33-39. 9. Dekker N, van Dussen L, Hollak CE, et al. Elevated plasma glucosylsphingosine in Gaucher disease: relation to phenotype, storage cell markers, and therapeutic response. Blood. 2011;118(16):e118-127. 10. Hollak CE, van Weely S, van Oers MH, et al. Marked elevation of plasma chitotriosidase activity. A novel hallmark of Gaucher disease. J Clin Invest. 1994;93(3):1288-1292. 11. Aerts JM, Kallemeijn WW, Wegdam W, et al. Biomarkers in the diagnosis of lysosomal storage disorders: proteins, lipids, and inhibodies. J Inherit Metab Dis. 2011;34(3):605619.

haematologica | 2021; 106(2)

Biomarkers in Gaucher disease

12. Deegan PB, Moran MT, McFarlane I, et al. Clinical evaluation of chemokine and enzymatic biomarkers of Gaucher disease. Blood Cells Mol Dis. 2005;35(2):259-267. 13. van Dussen L, Hendriks EJ, Groener JE, et al. Value of plasma chitotriosidase to assess nonneuronopathic Gaucher disease severity and progression in the era of enzyme replacement therapy. J Inherit Metab Dis. 2014;37(6):9911001. 14. Smid BE, Ferraz MJ, Verhoek M, et al. Biochemical response to substrate reduction therapy versus enzyme replacement therapy in Gaucher disease type 1 patients. Orphanet J Rare Dis. 2016;11:28. 15. Schoonhoven A, Rudensky B, Elstein D, et al. Monitoring of Gaucher patients with a novel chitotriosidase assay. Clin Chim Acta. 2007;381(2):136-139. 16. Boot RG, Renkema GH, Verhoek M, et al. The human chitotriosidase gene. Nature of inherited enzyme deficiency. J Biol Chem. 1998;273(40):25680-25685. 17. Deegan PB, Pavlova E, Tindall J, et al. Osseous manifestations of adult Gaucher disease in the era of enzyme replacement therapy. Medicine (Baltimore). 2011; 90(1):52-60. 18. Riley RD, Lambert PC, Abo-Zaid G. Metaanalysis of individual participant data: rationale, conduct, and reporting. BMJ. 2010; 340:c221. 19. Stewart LA, Clarke MJ. Practical methodology of meta-analyses (overviews) using updated individual patient data. Cochrane Working Group. Stat Med. 1995;14(19): 2057-2079. 20. Stewart LA, Clarke M, Rovers M, et al. Preferred reporting items for systematic review and meta-analyses of individual participant data: the PRISMA-IPD Statement. JAMA. 2015;313(16):1657-1665. 21. Raskovalova T, Deegan PB, Yang R, et al. Plasma chitotriosidase activity versus CCL18 level for assessing type I Gaucher disease severity: protocol for a systematic review with meta-analysis of individual participant data. Syst Rev. 2017;6(1):87. 22. Hollak CE, Belmatoug N, Cole JA, et al. Characteristics of type I Gaucher disease associated with persistent thrombocytopenia after treatment with imiglucerase for 4-5 years. Br J Haematol. 2012;158(4):528-538. 23. WHO. Haemoglobin concentrations for the diagnosis of anaemia and assessment of severity. Vitamin and Mineral Nutrition Information System. Geneva: World Health Organization. 2011. 24. Bland JM, Altman DG. Transformations, means, and confidence intervals. BMJ. 1996;312(7038):1079. 25. Bland JM, Altman DG. The use of transformation when comparing two means. BMJ. 1996;312(7039):1153. 26. Burke DL, Ensor J, Riley RD. Meta-analysis using individual participant data: one-stage and two-stage approaches, and why they may differ. Stat Med. 2017;36(5):855-875. 27. Stewart GB, Altman DG, Askie LM, et al. Statistical analysis of individual participant data meta-analyses: a comparison of methods and recommendations for practice. PLoS One. 2012;7(10):e46042. 28. Ben Turkia H, Gonzalez DE, Barton NW, et al. Velaglucerase alfa enzyme replacement therapy compared with imiglucerase in patients with Gaucher disease. Am J Hematol. 2013;88(3):179-184. 29. Elstein D, Mehta A, Hughes DA, et al. Safety and efficacy results of switch from

haematologica | 2021; 106(2)

imiglucerase to velaglucerase alfa treatment in patients with type 1 Gaucher disease. Am J Hematol. 2015;90(7):592-597. 30. Gonzalez DE, Turkia HB, Lukina EA, et al. Enzyme replacement therapy with velaglucerase alfa in Gaucher disease: results from a randomized, double-blind, multinational, Phase 3 study. Am J Hematol. 2013; 88(3):166-171. 31. Zimran A, Altarescu G, Philips M, et al. Phase 1/2 and extension study of velaglucerase alfa replacement therapy in adults with type 1 Gaucher disease: 48-month experience. Blood. 2010;115(23): 4651-4656. 32. Berger J, Vigan M, Pereira B, et al. Intra-monocyte pharmacokinetics of imiglucerase supports a possible personalized management of Gaucher disease type 1. Clin Pharmacokinet. 2019;58(4):469-482. 33. Murugesan V, Chuang WL, Liu J, et al. Glucosylsphingosine is a key biomarker of Gaucher disease. Am J Hematol. 2016; 91(11):1082-1089. 34. Zimran A, Brill-Almon E, Chertkoff R, et al. Pivotal trial with plant cell-expressed recombinant glucocerebrosidase, taliglucerase alfa, a novel enzyme replacement therapy for Gaucher disease. Blood. 2011;118(22):57675773. 35. Zimran A, Gonzalez-Rodriguez DE, Abrahamov A, et al. Safety and efficacy of two dose levels of taliglucerase alfa in pediatric patients with Gaucher disease. Blood Cells Mol Dis. 2015;54(1):9-16. 36. Andrade-Campos MM, Garcia-Sobreviela MP, Medrano-Engay B, et al. Assessing underlying chronic inflammation status in type 1 Gaucher disease patients (GD1): Involvement of lipocaline (LCN2) and other monocyte cytokines. Mol Genet Metab. 2017;120:S22. 37. Boot RG, Verhoek M, Langeveld M, et al. CCL18: a urinary marker of Gaucher cell burden in Gaucher patients. J Inherit Metab Dis. 2006;29(4):564-571. 38. Di Rocco M, Giona F, Carubbi F, et al. A new severity score index for phenotypic classification and evaluation of responses to treatment in type I Gaucher disease. Haematologica. 2008;93(8):1211-1218. 39. Giraldo P, Alfonso P, Atutxa K, et al. Realworld clinical experience with long-term miglustat maintenance therapy in type 1 Gaucher disease: the ZAGAL project. Haematologica. 2009;94(12):1771-1775. 40. Giraldo P, Irun P, Alfonso P, et al. Evaluation of Spanish Gaucher disease patients after a 6-month imiglucerase shortage. Blood Cells Mol Dis. 2011;46(1):115-118. 41. Giraldo P, Pérez-López J, Núñez R, et al. Patients with type 1 Gaucher disease in Spain: a cross-sectional evaluation of health status. Blood Cells Mol Dis. 2016;56(1):2330. 42. Groener JE, Poorthuis BJ, Kuiper S, et al. Plasma glucosylceramide and ceramide in type 1 Gaucher disease patients: correlations with disease severity and response to therapeutic intervention. Biochim Biophys Acta. 2008;1781(1-2):72-78. 43. Limgala RP, Ioanou C, Plassmeyer M, et al. Time of initiating enzyme replacement therapy affects immune abnormalities and disease severity in patients with Gaucher disease. PLoS One. 2016;11(12):e0168135. 44. Lukina E, Watman N, Dragosky M, et al. Eliglustat, an investigational oral therapy for Gaucher disease type 1: phase 2 trial results after 4 years of treatment. Blood Cells Mol

Dis. 2014;53(4):274-276. 45. Mistry PK, Lukina E, Ben Turkia H, et al. Effect of oral eliglustat on splenomegaly in patients with Gaucher disease type 1: the ENGAGE randomized clinical trial. JAMA. 2015;313(7):695-706. 46. Pastores GM, Petakov M, Giraldo P, et al. A phase 3, multicenter, open-label, switchover trial to assess the safety and efficacy of taliglucerase alfa, a plant cell-expressed recombinant human glucocerebrosidase, in adult and pediatric patients with Gaucher disease previously treated with imiglucerase. Blood Cells Mol Dis. 2014;53(4):253-260. 47. Drugan C, Drugan TC, Grigorescu-Sido P, et al. Modelling long-term evolution of chitotriosidase in non-neuronopathic Gaucher disease. Scand J Clin Lab Invest. 2017; 77(4):275-282. 48. Stirnemann J, Belmatoug N, Vincent C, et al. Bone events and evolution of biologic markers in Gaucher disease before and during treatment. Arthritis Res Ther. 2010;12(4): R156. 49. Boven LA, van Meurs M, Boot RG, et al. Gaucher cells demonstrate a distinct macrophage phenotype and resemble alternatively activated macrophages. Am J Clin Pathol. 2004;122(3):359-369. 50. Gordon S. Alternative activation of macrophages. Nat Rev Immunol 2003;3(1): 23-35. 51. Mantovani A, Sica A, Sozzani S, et al. The chemokine system in diverse forms of macrophage activation and polarization. Trends Immunol. 2004;25(12):677-686. 52. Rolfs A, Giese AK, Grittner U, et al. Glucosylsphingosine is a highly sensitive and specific biomarker for primary diagnostic and follow-up monitoring in Gaucher disease in a non-Jewish, Caucasian cohort of Gaucher disease patients. PLoS One. 2013;8(11): e79732. 53. Gelb MH. Newborn screening for lysosomal storage diseases: methodologies, screen positive rates, normalization of datasets, second-tier tests, and post-analysis tools. Int J Neonatal Screen. 2018;4(3). 54. Charrow J, Scott CR. Long-term treatment outcomes in Gaucher disease. Am J Hematol. 2015;90(Suppl 1):S19-24. 55. Hollak CE, Maas M, Aerts JM. Clinically relevant therapeutic endpoints in type I Gaucher disease. J Inherit Metab Dis. 2001;24(Suppl 2):S97-105. 56. Shemesh E, Deroma L, Bembi B, et al. Enzyme replacement and substrate reduction therapy for Gaucher disease. Cochrane Database Syst Rev. 2015;3:CD010324. 57. Sims KB, Pastores GM, Weinreb NJ, et al. Improvement of bone disease by imiglucerase (Cerezyme) therapy in patients with skeletal manifestations of type 1 Gaucher disease: results of a 48-month longitudinal cohort study. Clin Genet. 2008; 73(5):430-440. 58. Weinreb NJ, Cappellini MD, Cox TM, et al. A validated disease severity scoring system for adults with type 1 Gaucher disease. Genet Med. 2010;12(1):44-51. 59. Johnston BC, Miller PA, Agarwal A, et al. Limited responsiveness related to the minimal important difference of patient-reported outcomes in rare diseases. J Clin Epidemiol. 2016;79(11):10-21. 60. Leeflang MM, Deeks JJ, Gatsonis C, et al. Systematic reviews of diagnostic test accuracy. Ann Intern Med. 2008;149(12):889-897.

445

ARTICLE Ferrata Storti Foundation

Complications in Hematology

Predicting sinusoidal obstruction syndrome after allogeneic stem cell transplantation with the EASIX biomarker panel Sihe Jiang,1* Olaf Penack,1* Tobias Terzer,2 David Schult,3 Joshua Majer-Lauterbach,3 Aleksandar Radujkovic,3 Igor W. Blau,1 Lars Bullinger,1 Carsten Müller-Tidow,3 Peter Dreger3 and Thomas Luft3

Charité Universitätsmedizin Berlin, Campus Virchow Clinic, Hematology, Oncology and Tumorimmunology, Berlin; 2Biostatistics, German Cancer Research Centre, Heidelberg and 3 Medicine V, University Hospital Heidelberg, Heidelberg, Germany 1

Haematologica 2021 Volume 106(2):446-453

*SJ and OP contributed equally as co-first authors.

ABSTRACT

Correspondence: OLAF PENACK olaf.penack@charite.de Received: September 26, 2019. Accepted: January 22, 2020. Pre-published: January 23, 2020.

o biomarker panel has been established for prediction of sinusoidal obstruction syndrome/veno-occlusive disease (SOS/VOD), a major complication of allogeneic stem cell transplantation (alloSCT). We compared the potential of the Endothelial Activation and Stress Index (EASIX), based on lactate dehydrogenase, creatinine, and thrombocytes, with that of the SOS/VOD clinical risk score of the Center for International Blood and Marrow Transplant Research (CIBMTR) to predict SOS/VOD in two independent cohorts. In a third cohort, we studied the impact of endothelium-active prophylaxis with pravastatin and ursodeoxycholic acid (UDA) on SOS/VOD risk. The cumulative incidence of SOS/VOD within 28 days after alloSCT in the training cohort (Berlin, 2013-2015, n=446) and in the validation cohort (Heidelberg, 2002-2009, n=380) was 9.6% and 8.4%, respectively. In both cohorts, EASIX assessed at the day of alloSCT (EASIX-d0) was significantly associated with SOS/VOD incidence (P<0.0001), overall survival (OS), and non-relapse mortality (NRM). In contrast, the CIBMTR score showed no statistically significant association with SOS/VOD incidence, and did not predict OS and NRM. In patients receiving pravastatin/UDA, the cumulative incidence of SOS/VOD was significantly lower at 1.7% (Heidelberg, 2010-2015, n=359, P<0.0001) than in the two cohorts not receiving pravastatin/UDA. The protective effect was most pronounced in patients with high EASIX-d0. The cumulative SOS/VOD incidence in the highest EASIX-d0 quartiles were 18.1% and 16.8% in both cohorts without endothelial prophylaxis as compared to 2.2% in patients with pravastatin/UDA prophylaxis (P<0.0001). EASIX-d0 is the first validated biomarker for defining a subpopulation of alloSCT recipients at high risk for SOS/VOD. Statin/UDA endothelial prophylaxis could constitute a prophylactic measure for patients at increased SOS/VOD risk.

https://doi.org/10.3324/haematol.2019.238790

Introduction ©2021 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

446

Sinusoidal obstruction syndrome (SOS), also known as veno-occlusive disease (VOD), is a potentially fatal complication after allogeneic stem cell transplantation (alloSCT).1-3 Clinical management of SOS/VOD remains challenging, since there are no standardized predictive tools4 and diagnostic criteria are not uniform.2,4-6 The reported incidences of SOS/VOD after alloSCT range from 5.3% to 13.7% and vary depending on conditioning regimens, type of transplant, diagnostic criteria, patient characteristics, and other factors.7-9 The pathophysiology of SOS/VOD is characterized by endothelial injury caused by the conditioning regimen as well as pre-transplant damage.1,2,10,11 The resulting post-sinusoidal portal hypertension leads to the clinical syndrome of SOS/VOD.2,5,6,11 In severe SOS/VOD, which is strongly associated with multi-organ failure, mortality remains high.2,8,11 Early detection or prediction of SOS/VOD could allow identification of patients who could benefit from prophylactic meashaematologica | 2021; 106(2)

EASIX predicts SOS/VOD after alloSCT

ures,3 and preliminary data show that pre-emptive treatment with defibrotide might be effective,12 making tools to predict SOS/VOD highly desirable. Recently, a SOS/VOD clinical risk score (age, Karnofsky status, sirolimus use, hepatitis B/C status, conditioning regimen, disease type) has been published by the Center for International Blood and Marrow Transplant Research (CIBMTR).13 However, this score has not yet been validated outside of the CIBMTR database. In addition, it has been suggested that serum biomarkers, including microparticles and plasminogen activator inhibitor (PAI)1, may be useful to predict SOS/VOD,14-18 but validation and clinical implementation of these non-routine biomarkers will be difficult due to the lack of standardization. We had previously developed a standard biomarker panel to assess endothelial dysfunction and activation: the Endothelial Activation and Stress Index (EASIX). EASIX is based on the simple formula “lactate dehydrogenase (LDH) (U/L) * creatinine (mg/dL) / thrombocytes (109 cells per L)” and thus calculated using three of the diagnostic parameters of thrombotic microangiopathy (TMA),19 which is another endothelial complication after alloSCT.20 EASIX has also been shown to predict mortality of patients with acute graft-versus-host disease (GvHD).19 With this background, the aim of the current study was to test the SOS/VOD predictive capacity of EASIX compared with that of the CIBMTR score in two independent cohorts.

assess the risk of developing SOS/VOD after alloSCT.13 It incorporates age, hepatitis B/C serology, Karnofsky performance status, use of sirolimus prophylaxis, disease, disease status at the time of transplant, and conditioning regimen. It was developed using the CIBMTR database.13 We used the ‘VOD Risk Calculator’22 and recorded the probability of SOS/VOD development for each patient in the two independent cohorts when possible.

EASIX score The EASIX score was calculated using the formula: “LDH (U/L) *creatinine (mg/dL) / thrombocytes (109 cells per L)” as described previously, assessed at the day of alloSCT (EASIX-d0).19

Statistical analysis The primary objective was prediction of SOS/VOD occurrence. Primary analysis was performed for the binary endpoint “cumulative incidence of SOS/VOD within 28 days after alloSCT” and the time-to-event endpoint “time to VOD (TTV)” which is defined as time from alloSCT to diagnosis of SOS/VOD. Secondary objectives were the prediction of overall survival (OS) and time to nonrelapse mortality (NRM) measured from the day of alloSCT. TTV and NRM were analyzed using competing event models. The competing event for TTV is "non-SOS/VOD-mortality" defined as time from alloSCT to death without prior SOS/VOD, whereas the competing event for NRM is "time to relapse" defined as time from alloSCT to relapse of disease. Further details on statistical analyses are provided in the Online Supplementary Appendix.

Results Methods Patients' characteristics Study population For this retrospective cohort analysis, a training cohort and a validation cohort comprising consecutive adult patients who had undergone alloSCT at two independent institutions were investigated. Patients from the training cohort received alloSCT at the Charité - Campus Virchow Clinic, Berlin, between January 2013 and December 2015. The validation cohort consisted of patients who were allografted at the University of Heidelberg between September 2001 and December 2009. Patients undergoing alloSCT in Heidelberg after January 2010 received pravastatin and ursodeoxycholic acid (UDA) as prophylaxis of endothelial complications after alloSCT. To reduce confounding influences, the validation cohort was restricted to the time period before the introduction of pravastatin and UDA. The study was performed according to the Declaration of Helsinki, with written informed consent obtained by all eligible patients. The study has been approved by the institutional review board of both institutions.

Definitions SOS/VOD was defined according to the 2016 European Group for Blood and Marrow Transplantation (EBMT) criteria for SOS/VOD diagnosis in adults.4 The disease score we applied for the patients classifies the disease stage of the main diseases: acute leukemia, myelodysplastic syndrome, chronic myeloid leukemia, and non-Hodgkin lymphoma/multiple myeloma. The disease stages are assigned according to the remission status at transplant or the phases of chronic myeloid leukemia. The stages include ‘early disease stage' (0), ‘intermediate disease stage’ (1), and ‘late stage disease’ (2).21

CIBMTR sinusoidal obstruction syndrome/veno-occlusive disease risk score The SOS/VOD CIBMTR risk score has been established to haematologica | 2021; 106(2)

The training cohort and the validation cohort consisted of 446 and 380 patients, respectively. The baseline characteristics are presented in Table 1. The patient cohorts were similar with regards to age, sex, conditioning regimen, and SOS/VOD incidence. However, there were significant differences in the categories donor type (more matched unrelated donors in the training cohort), underlying disease (most frequent disease: acute myeloid leukemia, 51% and 27% in the training and validation cohort, respectively), stem cell source (bone marrow used more frequently in the validation cohort), and use of anti-thymocyte globulin (ATG) (more commonly used in the training cohort).

EASIX-d0 and sinusoidal obstruction syndrome/veno-occlusive disease risk SOS/VOD was diagnosed in 43 patients (9.6%, median onset day [d] +9) in the training cohort and in 32 patients (8.4%, median onset d +7) in the validation cohort. In the training cohort, median EASIX-d0 in patients who later went on to develop SOS/VOD was significantly higher than in patients who did not develop SOS/VOD (40.26; interquartile range [IQR] 14.72-80.38 vs. 16.06; IQR 6.0036.54, P<0.0001) (Figure 1A). These findings were confirmed in the validation cohort, where median EASIX-d0 in patients who subsequently developed SOS/VOD was also significantly higher than in patients who did not develop SOS/VOD (8.64; IQR 3.38-15.40 vs. 2.28; IQR 0.92-7.48, P<0.0001) (Figure 1B). Increasing EASIX-d0 was significantly associated with SOS/VOD incidence in the training cohort in both univariable (OR per log2 increase 1.39, 95%CI:1.18-1.66, P=0.0001) and multivariable analysis with the CIBMTR score as confounder (incorporating information on 6 clinical variables) 447

S. Jiang et al. Table 1. Baseline characteristics of the three patient cohorts: Berlin (training), Heidelberg no statins/ursodeoxycholic acid (UDA) (validation), and Heidelberg with statins/UDA. Berlin cohort Heidelberg Heidelberg P P P (training cohort) no statins/ with statins/ Berlin vs. Berlin vs. Heidelberg (n=446) UDA cohort UDA cohort Heidelberg Heidelberg no statins/UDA (validation cohort) (n=359) no statins/UDA with statins/UDA vs. Heidelberg (n=380) with statins/UDA Date of alloSCT 01/2013 – 12/2015 09/2001 – 01/2010 01/2010 – 12/2015 Age at transplant in years - median (range) 54 (18-75) 50 (17-70) 56 (19-75) Recipient sex (n, %) Female 172 (39) 153 (40) 140 (39) Male 274 (61) 227 (60) 219 (61) Donor sex (n, %) Female 124 (28) 121 (32) 122 (34) Male 274 (61) 259 (68) 237 (66) NA 48 (11) Donor (n, %) Matched related donor 88 (20) 141 (37) 94 (26) Matched unrelated donor 263 (59) 140 (37) 196 (55) Mismatched related donor 6 (1) 14 (4) 11 (3) Mismatched unrelated donor 89 (20) 85(22) 58 (16) Disease (n, %) AML 229 (51) 101 (27) 126 (35) MDS/MPN 82 (18) 53 (14) 69 (19) ALL 35 (8) 39 (10) 20 (6) Lymphoma 41 (9) 96 (25) 100 (28) MM 39 (9) 75 (20) 36 (10) Other 20 (4) 16 (4) 8 (2) Disease score (n, %) 0 142 (32) 131 (34) 122 (34) 1 47 (11) 65 (17) 122 (34) 2 248 (56) 184 (48) 115 (32) NA 9 (2) Stem-cell source (n, %) Peripheral blood 443 (99) 351 (92) 336 (94) Bone marrow 2 (1) 29 (8) 23 (6) NA 1 (0) Conditioning (n, %) RIC 341 (76) 291 (77) 294 (82) MAC 105 (24) 89 (23) 64 (18) Use of ATG (n, %) Yes 399 (89) 193 (51) 259 (72) No 47 (11) 187 (49) 100 (28) SOS/VOD development (EBMT criteria)(n, %) SOS/VOD 43 (10) 32 (8) 6 (2) No SOS/VOD 401 (90) 348 (92) 353 (98) Onset of SOS/VOD (median, range) d+9 (d+3 to d+30) d+8 (d0 to d+24) d+10 (d+1 to d+17) Median CIBMTR score (range, IQR) SOS/VOD n=32, rest n=29, rest n=5, rest of data NA: 1.51 of data NA: 1.9 of data NA: 1.7 (0.56-4.48; 0.82-2.37) (0.3-8.7; 1.4-3.1) (0.7-3.7; 1.0-2.7) No SOS/VOD n=338, rest n=333, rest n=347, rest of data NA: 1.01 of data NA: 1.2 of data NA: 1.0 (0.32-9.72; 0.76-1.80) (0.4-20.6; 0.9-2.1) (0.3-9.1; 0.8-1.7) Median EASIX-d0 (range, IQR) SOS/VOD n=41, rest n=32, rest n=6 rest of data NA: 40.26: of data NA: 8.6 of data NA: 7.4 (5.23-865.06; 14.72-80.38) (0.2-41.0; 3.4-15.4) (4.0-17.1; 5.8-14.8) No SOS/VOD n=376, rest n=348, rest n=353, rest of data NA: 16.06 of data NA: 2.3 of data NA: 7.5 (0.38-575; 6.00-36.54) (0.2-99.2; 0.9-7.5) (0.2-195.8; 2.1-14.9)

0.67

0.942

0.764

0.877

0.437

0.584

<0.001

0,034

<0.001

0.012

<0.001

0.566

0.999

0.055

0.069

<0.001

0.472

<0.001

SOS/VOD: sinusoidal obstruction syndrome/veno-occlusive disease; UDA: ursodeoxycholic acid; alloSCT: allogeneic stem cell transplantation; NA: not available; AML: acute myeloid leukemia; MDS: myelodysplastic syndrome; ALL: acute lymphocytic leukemia; CLL: chronic lymphocytic leukemia; MPN: myeloproliferative neoplasms; CML: chronic myeloid leukemia; MM: multiple myeloma; RIC: reduced intensity conditioning; MAC: myeloablative conditioning; ATG: anti-thymocyte globulin; KPS: Karnofsky Performance Score; IQR: interquartile range; CIBMTR: Center for International Blood and Marrow Transplant Research; EBMT: European Group for Blood and Marrow Transplantation; EASIX-d0: EASIX assessed at the day of alloSCT; n: number; d: day.

448

haematologica | 2021; 106(2)

EASIX predicts SOS/VOD after alloSCT

(OR per log2 increase 1.31, 95%CI:1.08-1.59, P=0.0067) (Figure 2A). Similarly, EASIX-d0 was strongly associated with the incidence of SOS/VOD in the validation cohort (univariable analysis: OR per log2 increase 1.50, 95%CI: 1.22-1.88, P=0.0002; multivariable analysis: OR per log2 increase 1.57, 95%CI: 1.26-2.01, P=0.0001) (Figure 2B). Based on these data, we have created the EASIX-d0 SOS/VOD calculator for free public use (http://biostatistics.dkfz.de/EASIX/).

Association of EASIX-d0 with overall survival and non-relapse mortality In the training cohort, EASIX-d0 was significantly associated with OS and NRM in univariable analysis (OS: HR per log2 increase 1.20, 95%CI: 1.12-1.29, P<0.0001; NRM: cause-specific hazard ratio [CSHR] per log2 increase 1.25, 95%CI: 1.14-1.38, P<0.0001) (Figure 3A and C). Likewise, EASIX-d0 was significantly associated with OS and NRM in the validation cohort in univariable analysis (OS: HR

per log2 increase 1.10, 95%CI: 1.02-1.18, P=0.0124; NRM: CSHR per log2 increase 1.18, 95%CI: 1.06-1.32, P=0.0024) (Figure 3B and D).

Association of the CIBMTR clinical risk score with sinusoidal obstruction syndrome/veno-occlusive disease incidence, overall survival and non-relapse mortality In the training cohort, a non-significant trend towards a higher median CIBMTR score was observed in patients who subsequently developed SOS/VOD as compared to patients who did not develop SOS/VOD (1.51; IQR 0.822.37 vs. 1.01; IQR 0.76-1.80, P=0.069). In the validation cohort, the median CIBMTR score in patients with SOS/VOD was significantly higher than that in patients without SOS/VOD (1.92; IQR 1.44-3.09 vs. 1.20; IQR 0.89-2.15, P=0.00053). On time-to-event analysis, however, the association of the CIBMTR score with SOS/VOD incidence was not statistically significant in the training

Figure 1. Endothelial Activation and Stress Index (EASIX) assessed at the day of alloSCT (EASIX-d0) in patients without sinusoidal obstruction syndrome/veno-occlusive disease (SOS/VOD) versus EASIX-d0 in patients with SOS/VOD. Box plot of EASIX-d0 in patients without SOS/VOD (No VOD) versus EASIX-d0 in patients with SOS/VOD (VOD)) in (A) the training and (B) the validation cohort.

Figure 2. Time to sinusoidal obstruction syndrome/veno-occlusive disease (SOS/VOD) depending on Endothelial Activation and Stress Index (EASIX) assessed at the day of alloSCT (EASIX-d0) quartiles. (A) Training and (B) validation cohort. SCT: stem cell transplantation.

haematologica | 2021; 106(2)

449

S. Jiang et al. A

Figure 3. Univariable effect of Endothelial Activation and Stress Index (EASIX) assessed at the day of alloSCT (EASIX-d0) on overall survival (OS) and non-relapse mortality (NRM). Univariable effect of EASIX-d0 on OS in (A) the training cohort and (B) the validation cohort. Univariable association of EASIX-d0 with NRM in (C) the training cohort and (D) the validation cohort.

cohort (univariable analysis: OR per log2 increase 1.19, 95%CI: 0.87-1.54, P=0.2288) (Figure 4A) nor in the validation cohort (univariable analysis: OR per log2 increase 1.07, 95%CI: 0.92-1.20, P=0.308) (Figure 4B). Brier score based on observed SOS/VOD incidence in the training cohort (null model) is 0.0774 in the validation cohort. Inclusion of EASIX-d0 leads to a reduction in the quadratic prediction error (0.0735) of approximately 5%. In contrast, Brier score of the CIBMTR score model (0.0771) is similar to the Brier score of the null model. In both cohorts, the CIBMTR score was not predictive of OS or NRM (univariable analysis, training cohort, OS: HR per log2 increase 1.08, 95%CI: 0.95-1.22, P=0.2264; NRM: CSHR per log2 increase 1.15, 95%CI: 1.00-1.33, P=0.0565; validation cohort, OS: HR per log2 increase 1.03, 95%CI: 0.97-1.09, P=0.4126; NRM: CSHR per log2 increase 0.95, 95%CI: 0.84-1.08, P=0.4648).

Effect of pravastatin/ursodeoxycholic acid on sinusoidal obstruction syndrome/veno-occlusive disease incidence Patients undergoing alloSCT in Heidelberg after January 2010 received pravastatin and UDA as routine prophylaxis of endothelial complications after alloSCT. In this cohort of 359 consecutive patients transplanted in Heidelberg between January 2010 and December 2015, the SOS/VOD incidence was significantly lower than in the training and 450

validation cohorts treated without endothelial prophylaxis (1.7%, vs. 9.6% and 8.4%, P<0.0001) (Figure 5A). Next, we focused on the effect of pravastatin/UDA prophylaxis on SOS/VOD incidence, NRM and OS in a population at increased risk for SOS/VOD, defined by the highest EASIX-d0 quartile in each of the three cohorts. The patient cohort at increased risk that received pravastatin/UDA showed a significantly lower SOS/VOD incidence (Figure 5B), lower NRM (Figure 5C), and higher OS (Figure 5D) compared to the high-risk patient populations in the training and validation cohorts.

Discussion In this retrospective cohort analysis, EASIX-d0 was found to be an independent predictor of SOS/VOD risk, OS and NRM in adult patients receiving alloSCT. EASIXd0 constitutes the first validated biomarker for defining a subpopulation of alloSCT recipients at high risk for SOS/VOD. It consists of routine laboratory parameters enabling easy implementation in any transplant center. EASIX-d0 seems to be a readily available tool for stratifying patients into high- and low-risk populations, which could be useful to improve clinical management of SOS/VOD, and for identifying patient subsets for clinical trials on SOS/VOD prophylaxis. Outside of clinical studhaematologica | 2021; 106(2)

EASIX predicts SOS/VOD after alloSCT

Figure 4. Time to sinusoidal obstruction syndrome/veno-occlusive disease (SOS/VOD) depending on Center for International Blood and Marrow Transplant Research (CIBMTR) score quartiles. (A) Training and (B) validation cohort. SCT: stem cell transplantation.

Figure 5. Effects of pravastatin/ursodeoxycholic acid prophylaxis (blue) compared to the training cohort (green) and the validation cohort (red). (A) Time to sinusoidal obstruction syndrome/veno-occlusive disease (SOS/VOD) in the three cohorts. (B) Time to SOS/VOD in patients with the highest Endothelial Activation and Stress Index (EASIX) assessed at the day of alloSCT (EASIX-d0) quartiles in the three cohorts. (C) Time to non-relapse mortality (NRM) in patients with the highest EASIX-d0 quartile in the three cohorts. (D) Overall survival (OS) in patients with the highest EASIX-d0 quartile in the three cohorts. SCT: stem cell transplantation.

haematologica | 2021; 106(2)

451

S. Jiang et al.

ies, patients with high EASIX-d0 scores might benefit from closer monitoring of emerging clinical SOS/VOD signs and early interventions. EASIX-d0 has to be put into perspective with the CIBMTR SOS/VOD clinical risk score, which has been recently described as a predictive tool to identify patients at high risk of developing SOS/VOD.13 The CIBMTR score has been established using a large sample from the CIBMTR database and consists of select baseline parameters which are partly fixed (age, hepatitis B/C serology, Karnofsky performance score, diagnosis, disease status at transplant) and partly subject to intervention (use of sirolimus, conditioning regimen). It has been shown to be predictive for SOS/VOD but not for NRM and OS.13 In the present study, the CIBMTR score exhibited only a weak association with SOS/VOD incidence and no association with NRM or OS. One explanation for this discrepancy might be that the two European cohorts investigated in the current study differed in some respects from the CIBMTR cohort. First, in vivo T-cell depletion with ATG was administered in most patients in the two European cohorts, whereas the majority of the CIBMTR patients did not receive ATG for GvHD prophylaxis.13 Second, most patients in our two cohorts were conditioned with reduced intensity conditioning (RIC), whereas only a minority of patients from the CIBMTR database received an RIC regimen.13 Third, sirolimus was administered in 8% of the patients from the CIBMTR database,13 whereas none of our patients received sirolimus prophylaxis. This is relevant because sirolimus was a risk factor for SOS/VOD development in the CIBMTR analysis. In addition, registry data might be prone to consistency deficits, while we had immediate access to the primary data ensuring the high quality of the data used in our study. Of note, the CIBMTR score has been primarily validated in a large population. We have validated the current EASIX-VOD score in smaller cohorts. Therefore, further validation is needed. The EBMT is currently conducting a prospective non-interventional study on the value of the EASIX score to predict alloSCT-related endothelial complications. Furthermore, we expect that data from several retrospective cohorts in different centers will soon be available. Based on previous publications on the efficacy of UDA and statins in the protection of the endothelium,23,24 the Heidelberg alloSCT group decided to administer pravastatin and UDA as prophylaxis for alloSCT recipients transplanted after January 2010. UDA is a synthetic bile acid that reduces the incidence of SOS/VOD, and is associated with less liver toxicity and better survival rates.3,25 Statins have not been extensively studied for the prevention of endothelial complications. However, its pleiotropic effects include, besides the inhibition of cholesterol synthesis, improving endothelial function, reducing oxidative stress and inflammation, and decreasing thrombogenic properties.26,27 Statins may therefore be beneficial in the prevention of endothelial complications. Accordingly, we

References 1. Mohty M, Malard F, Abecassis M, et al. Sinusoidal obstruction syndrome/venoocclusive disease: current situation and perspectives - a position statement from the

452

observed a reduction of endothelial post-transplant complications, as previously shown in TMA20 and refractory GvHD28 upon implementation of statin/UDA endothelial prophylaxis. In the present study, the SOS/VOD incidence was markedly reduced after the introduction of pravastatin/UDA prophylaxis. These protective effects, both in terms of SOS/VOD risk reduction and lower NRM and overall mortality, were particularly pronounced in patients with high EASIX-d0 scores, as compared to highrisk patients who did not receive pravastatin/UDA. This suggests that patients at high risk for SOS/VOD may benefit most from prophylactic SOS/VOD strategies. Limitations of our study are its retrospective design and the validation in only one independent cohort of patients with similar patients' characteristics, which are typical for adult European alloSCT transplant centers. Therefore, the results cannot be extrapolated to pediatric alloSCT populations. Since EASIX-d0 is very easy to assess, any transplant center now has the opportunity to evaluate its potential in their respective patient population and we hope that more data on different patient populations, including pediatric patients or haploidentical transplantation, will soon be available. To facilitate this process, we have created the EASIX-d0 SOS/VOD calculator for free public use (http://biostatistics.dkfz.de/EASIX/). In conclusion, EASIX-d0 seems to be an easy-to-use biomarker for identifying patient populations at high risk for SOS/VOD, and thus could be a promising tool both for clinical trials and tailored monitoring strategies. Statin/UDA endothelial prophylaxis could be an option to overcome the increased SOS/VOD risk indicated by high EASIX-d0 scores. Disclosures Olaf Penack has received research support and honorarium as a speaker from Jazz Pharmaceuticals. Thomas Luft has received research support from Jazz Pharmaceuticals and Neovii. The remaining authors declare no competing interests. Contributions SJ, OP and TL designed the study, performed research, analysed data and wrote the manuscript; DS, AR and JM-L performed research; TT analyzed data; IWB, LB, CM-T and PD designed the study and edited the manuscript. All authors have read the manuscript and have agreed to submit it. Acknowledgments The authors would like to thank Jazz Pharmaceuticals for financial support. Funding This study was partly financially supported by Jazz Pharmaceuticals. The company played no role in the acquisition or analyses of data. Jazz was not involved in the writing or editing of the manuscript.

European Society for Blood and Marrow Transplantation (EBMT). Bone Marrow Transplant. 2015;50(6):781-789. 2. Mcdonald GB, Sharma P, Matthews DE, Shulman HM, Thomas ED. Venocclusive disease of the liver after bone marrow transplantation: diagnosis, incidence, and predis-

posing factors. Hepatology. 1984;4(1):116122. 3. Carreras E. How I manage sinusoidal obstruction syndrome after haematopoietic cell transplantation. Br J Haematol. 2014; 168(4):481-491. 4. Mohty M, Malard F, Abecassis M, et al.

haematologica | 2021; 106(2)

EASIX predicts SOS/VOD after alloSCT

10.

11. 12.

Revised diagnosis and severity criteria for sinusoidal obstruction syndrome/venoocclusive disease in adult patients: a new classification from the European Society for Blood and Marrow Transplantation. Bone Marrow Transplant. 2016;51(7):906-912. Jones RJ, Lee KS, Beschorner WE, et al. Venoocclusive disease of the liver following bone marrow transplantation. Transplantation. 1987;44(6):778-783. McDonald GB, Hinds MS, Fisher LD, et al. Veno-occlusive disease of the liver and multiorgan failure after bone marrow transplantation: a cohort study of 355 patients. Ann Intern Med. 1993;118(4):255-267. Carreras E, Bertz H, Arcese W, et al. Incidence and outcome of hepatic venoocclusive disease after blood or marrow transplantation: a prospective cohort study of the European Group for Blood and Marrow Transplantation. Blood. 1998;92 (10):3599-3604. Coppell JA, Richardson PG, Soiffer R, et al. Hepatic veno-occlusive disease following stem cell transplantation: incidence, clinical course, and outcome. Biol Blood Marrow Transplant. 2010;16(2):157-168. Dalle J-H, Giralt SA. Hepatic veno-occlusive disease after hematopoietic stem cell transplantation: risk factors and stratification, prophylaxis, and treatment. Biol Blood Marrow Transplant. 2016;22(3):400-409. Carreras E, Diaz-Ricart M. The role of the endothelium in the short-term complications of hematopoietic SCT. Bone Marrow Transplant. 2011;46(12):1495-1502. Bearman S. The syndrome of hepatic venoocclusive disease after marrow transplantation. Blood. 1995;85(11):3005-3020. Richardson PG, Smith AR, Triplett BM, et al. Earlier defibrotide initiation post-diagnosis of veno-occlusive disease/sinusoidal obstruction syndrome improves day +100

haematologica | 2021; 106(2)

13.

14.

15.

16.

17.

18.

19.

20.

survival following haematopoietic stem cell transplantation. Br J Haematol. 2017; 178(1):112-118. Strouse C, Zhang Y, Zhang M-J, et al. Risk score for the development of veno-occlusive disease after allogeneic hematopoietic cell transplant. Biol Blood Marrow Transplant. 2018;24(10):2072-2080. Salat C, Holler E, Kolb H-J, et al. Plasminogen activator inhibitor-1 confirms the diagnosis of hepatic veno-occlusive disease in patients with hyperbilirubinemia after bone marrow transplantation. Blood. 1997;89(6):2184-2188. Tanikawa S, Mori S, Ohhashi K, et al. Predictive markers for hepatic veno-occlusive disease after hematopoietic stem cell transplantation in adults: a prospective single center study. Bone Marrow Transplant. 2000;26(8):881-886. Cutler C, Kim HT, Ayanian S, et al. Prediction of veno-occlusive disease using biomarkers of endothelial injury. Biol Blood Marrow Transplant. 2010;16(8):1180-1185. Akil A, Zhang Q, Mumaw CL, et al. Biomarkers for diagnosis and prognosis of sinusoidal obstruction syndrome after hematopoietic cell transplantation. Biol Blood Marrow Transplant. 2015; 21(10):1739-1745. Piccin A, Sartori MT, Bisogno G, et al. New insights into sinusoidal obstruction syndrome: microparticles and VOD. Intern Med J. 2017;47(10):1173-1183. Luft T, Benner A, Jodele S, et al. EASIX in patients with acute graft-versus-host disease: a retrospective cohort analysis. Lancet Haematol. 2017;4(9):e414-e423. Zeisbrich M, Becker N, Benner A, et al. Transplant-associated thrombotic microangiopathy is an endothelial complication associated with refractoriness of acute GvHD. Bone Marrow Transplant.

2017;52(10):1399-1405. 21. Gratwohl A, Stern M, Brand R, et al. Risk score for outcome after allogeneic hematopoietic stem cell transplantation: a retrospective analysis. Cancer. 2009; 115(20):4715-4726. 22. Center for International Blood & Marrow Transplant Research. VOD Risk Calculator. 2018. https:// www.cibmtr.org/ Reference Center/Statistical/Tools/Pages/VOD.aspx (Last accessed August 24, 2018). 23. Ruutu T, Eriksson B, Remes K, et al. Ursodeoxycholic acid for the prevention of hepatic complications in allogeneic stem cell transplantation. Blood. 2002;100(6):19771983. 24. Ruutu T, Juvonen E, Remberger M, et al. Improved survival with ursodeoxycholic acid prophylaxis in allogeneic stem cell transplantation: long-term follow-up of a randomized study. Biol Blood Marrow Transplant. 2014;20(1):135-138. 25. Cheuk DKL, Chiang AKS, Ha SY, Chan GCF. Interventions for prophylaxis of hepatic veno-occlusive disease in people undergoing haematopoietic stem cell transplantation. Cochrane Database Syst Rev. 2015;(5):CD009311. 26. Lahera V, Goicoechea M, de Vinuesa SG, et al. Endothelial dysfunction, oxidative stress and inflammation in atherosclerosis: beneficial effects of statins. Curr Med Chem. 2007;14(2):243-248. 27. Blum A, Shamburek R. The pleiotropic effects of statins on endothelial function, vascular inflammation, immunomodulation and thrombogenesis. Atherosclerosis. 2009; 203(2):325-330. 28. Dietrich S, Okun JG, Schmidt K, et al. High pre-transplant serum nitrate levels predict risk of acute steroid-refractory graft-versushost disease in the absence of statin therapy. Haematologica. 2014;99(3):541-547.

453

ARTICLE Ferrata Storti Foundation

Haematologica 2021 Volume 106(2):454-463

Leukocyte Biology

Alternative activation of human macrophages enhances tissue factor expression and production of extracellular vesicles

Philipp J. Hohensinner,1,2 Julia Mayer,1 Julia Kirchbacher,1 Julia Kral-Pointner,1,2 Barbara Thaler,1 Christoph Kaun,1 Lena Hell,3 Patrick Haider,1 Marion Mussbacher,4 Johannes A. Schmid,4 Stefan Stojkovic,1 Svitlana Demyanets,1,5 Michael B. Fischer,6,7 Kurt Huber,2,8,9 Katharina Wöran,10 Christian Hengstenberg,1 Walter S. Speidl,1 Rudolf Oehler,11 Ingrid Pabinger3 and Johann Wojta1,2,12

Department of Internal Medicine II/Cardiology, Medical University of Vienna, Vienna; Ludwig Boltzmann Institute for Cardiovascular Research, Vienna; 3Department of Internal Medicine I, Medical University of Vienna, Vienna; 4Institute of Vascular Biology and Thrombosis Research, Medical University of Vienna, Vienna; 5Department of Laboratory Medicine, Medical University of Vienna, Vienna; 6Clinic for Blood Group Serology and Transfusion Medicine, Medical University of Vienna, Vienna; 7Department for Health Science and Biomedicine, Danube University Krems, Krems; 83rd Medical Department, Wilhelminenhospital, Vienna; 9Medical Faculty, Sigmund Freud University, Vienna; 10Department of Pathology, Medical University of Vienna, Vienna; 11Department of Surgery, Medical University of Vienna, Vienna and 12Core Facilities, Medical University of Vienna, Vienna, Austria 1 2

ABSTRACT

M Correspondence: JOHANN WOJTA johann.wojta@meduniwien.ac.at Received: February 26, 2019. Accepted: January 23, 2020. Pre-published: January 23, 2020. https://doi.org/10.3324/haematol.2019.220210

454

acrophages are versatile cells that can be polarized by the tissue environment to fulfill required needs. Proinflammatory polarization is associated with increased tissue degradation and propagation of inflammation whereas alternative polarization within a Th2 cytokine environment is associated with wound healing and angiogenesis. To understand whether polarization of macrophages can lead to a procoagulant macrophage subset we polarized human monocyte-derived macrophages to proinflammatory and alternative activation states. Alternative polarization with interleukin-4 and interleukin-13 led to a macrophage phenotype characterized by increased tissue factor (TF) production and release and by an increase in extracellular vesicle production. In addition, TF activity was enhanced in extracellular vesicles of alternatively polarized macrophages. This TF induction was dependent on signal transducer and activator of transcription-6 signaling and poly ADP ribose polymerase activity. In contrast to monocytes, human macrophages did not show increased TF expression upon stimulation with lipopolysaccharide and interferon-γ. Previous polarization to either a proinflammatory or an alternative activation subset did not change the subsequent stimulation of TF. The inability of proinflammatory activated macrophages to respond to lipopolysaccharide and interferonγ with an increase in TF production seemed to be due to an increase in TF promoter methylation and was reversible when these macrophages were treated with a demethylating agent. In conclusion, we provide evidence that proinflammatory polarization of macrophages does not lead to enhanced procoagulatory function, whereas alternative polarization of macrophages leads to an increased expression of TF and increased production of TF-bearing extracellular vesicles by these cells suggesting a procoagulatory phenotype of alternatively polarized macrophages.

Introduction Macrophages are cells of the innate immune system which play numerous and vastly different functions within the body. Macrophages reside in all tissues of the body and each population of macrophages within a tissue can take on specialized functions that are tuned to the developmental and functional requirements of that haematologica | 2021; 106(2)

Alternative polarization induces tissue factor

tissue.1 Some tissue macrophages and precursors are already established embryonically in the yolk sac and fetal liver before the onset of definitive hematopoiesis.2 Upon inflammation, the pool of resident macrophages gets quickly replaced by macrophages derived from circulating monocytes.3 In order to be able to respond to multiple tasks, macrophages can adapt to the environment when exposed to specific cues. This macrophage polarization can be simulated in vitro using lipopolysaccharide (LPS) and interferon (IFN)-γ stimulation for a proinflammatory subset termed classical activation and stimulation with interleukin (IL)-4 and IL-13 for an alternative polarization phenotype.4 Upon polarization, macrophages react to the respective stimulus with the expression of a distinct phenotype. Classical polarization is usually associated with a proinflammatory response, including the increased production of tumor necrosis factor (TNF)-α, IL-1, and IL-6.5 Functionally, proinflammatory polarization leads to potent effector cells that kill intracellular micro-organisms and tumor cells.6 In addition, these cells are present during early wound healing and proinflammatory macrophages are characterized by a pronounced ability to degrade tissue.7 In contrast, alternatively activated macrophages are characterized by increased expression of IL-10 and of scavenger receptors. Besides scavenging debris, promoting angiogenesis, tissue remodeling and repair, alternatively activated macrophages are able to fine tune inflammatory responses.8 In vivo, distinct macrophage subtypes are more subtle. Nevertheless, in vitro polarization phenotypes are present in in vivo situations.4 Macrophages resembling classical macrophages can be found in environments with bacterial infection or in inflammatory pathologies, whereas alternatively polarized macrophages are prominent in cancer and in diseases with a Th2 cytokine signature.9 An inflammatory milieu is usually accompanied by an increased procoagulatory risk. This risk increase is due to the close link between inflammation and tissue factor (TF) induction. TF is the primary activator of the coagulation cascade.10 TF-bearing monocytes are essential for the initiation of coagulation11 and TF expression in monocytes is triggered strongly by inflammatory mediators including LPS.10 Nonetheless, whereas TF expression in human monocytes is well studied, data on TF production in human macrophages, especially under different polarization conditions, are scarce. TF is a transmembrane protein that functions as a high affinity receptor for factor VII and the activated form of this clotting factor.10 Especially in circulation, TF is present on circulating extracellular vesicles. In pathological conditions, high levels of procoagulant, TFexposing particles can be found in the circulation.12 Interestingly, even though TF and inflammation are closely connected, cancer and hence a Th2-promoting environment were also found to be associated with increased TFbearing extracellular vesicles.13 Extracellular vesicles are membrane-enclosed structures of different sizes.14 TF is mainly associated with vesicles within the microvesicle fraction of extracellular vesicles which range in size from 200-1000 nm.15 These vesicles form through an active budding mechanism from the parental cell and, therefore, include membrane-linked proteins.16 In addition to TF those vesicles can have phosphatidylserine exposed on the surface which, in turn, can both support coagulation17 and activate TF.10 haematologica | 2021; 106(2)

In order to understand a possible role of macrophages in fostering a procoagulatory environment in different disease states, we investigated the contribution of macrophages and their respective polarization to the production and release of TF and of extracellular vesicles.

Methods Full details of the experimental setups and conditions are available in the Online Supplement.

Generation of human monocytes and macrophages Human macrophages were derived from peripheral blood monocytes from leukapheresis chambers, as described previously, obtained from healthy platelet donors according to recommendations of the ethical board of the Medical University of Vienna including informed consent (approval number 1575/2014).7,18 Human macrophages were always prepared from fresh blood. Macrophages from 75 healthy volunteers were used in the course of the study.

Flow cytometry Extracellular vesicles were analyzed using flow cytometry on a Cytoflex (Beckman Coulter, CA, USA) or AttuneNXT (Thermo Fisher, MA, USA) flow cytometer. Values are given as events/mL. Microvesicles were defined as being between 200 nm and 900 nm according to size-specific fluorescence beads (Megamix Plus, Biocytex, France). Human and mouse monocytes were investigated for membrane-bound TF using a specific antibody (Thermo Fisher [MA, USA] for human TF, Biotechne [MN, USA] for mouse TF) as published previously.19

Enzyme-linked immunosorbent assay of extracellular vesicles We used a commercially available enzyme-linked immunosorbent assay (ELISA) kit (Hyphen Biomed, France) to determine the concentration of annexin V+ extracellular vesicles in the circulation according to the manufacturer’s instructions.

RNA isolation and quantitative polymerase chain reaction analysis Detailed information about RNA isolation and quantitative polymerase chain reactions are available in the Online Supplement.

Protein determination TF was determined by a commercially available ELISA (Biotechne, MN, USA) as suggested by the manufacturer on extracellular vesicles and in cell lysates. Extracellular vesicles were isolated by centrifugation at 18,000 g for 20 min at 4°C.

Tissue factor activity assay Extracellular vesicle-associated TF activity was measured as previously described.20 A detailed protocol is available in the Online Supplement.

Immunohistochemistry and fluorescence microscopy Immunohistochemical staining was performed on cryopreserved sections of TissueTek (Agilent, CA, USA)-embedded colon carcinoma tissue (4 patients) or on paraffin-embedded atherosclerotic tissue sections, as published previously.7,21 Specimens were collected according to the recommendations of the institutional ethics board including informed consent. Plaque tissue was derived from patients undergoing carotid endarterectomy (mean age 71±6.4 years, 72% male, 27% symptomatic, n=16). 455

P.J. Hohensinner et al.

Statistics Sample groups were compared using a paired Student t-test using SPSS 21 (IBM, CA, USA). P-values ≤0.05 were considered statistically significant. The number of individual donors per experiment is given in the figure legends. All graphs depict the mean values ± standard deviation.

Results We determined the expression of TF in macrophages under baseline condition as well as under polarized conditions to understand the impact of macrophages and their polarization on coagulation. When TF protein was measured in cell lysates of human macrophages, IL-4 and IL-13 significantly upregulated TF protein content whereas LPS and IFN-γ did not change protein levels of TF compared to the levels in unpolarized macrophages (M0) (Figure 1A). Polarization conditions did not influence viability as determined by lactate dehydrogenase assay (Online Supplementary Figure S1A). TF is associated with extracellular vesicles. We, therefore, determined the TF concentration in macrophage-derived extracellular vesicles at baseline and under polarization conditions. Again, TF levels were increased under alternative polarization conditions whereas proinflammatory stimulation did not alter TF protein levels compared to the levels in microvesicles obtained

from M0 (Figure 1B). The main role of TF is the initiation of coagulation. We, therefore, analyzed the functional capacity of macrophage-derived TF-bearing microvesicles to initiate coagulation by determining the change of factor X to its activated form via generation of activated factor VII. Again, polarization of macrophages with IL-4 and IL13 led to a marked induction of active TF microvesicles (Figure 1C). To rule out contamination and hence lowgrade induction with LPS during alternative polarization we determined the induction of TF under alternative polarization with IL-4 and IL-13 with or without the TLR-4 inhibitor polymyxin B. Similar results were obtained in the presence and absence of polymyxin B (Online Supplementary Figure S1B). To determine whether TF protein can be detected at an early time point, we analyzed microvesicles secreted after 6 h for TF content. We did not observe significant changes with any polarization condition, although we did observe a non-significant increase for TF in alternatively activated macrophages (Online Supplementary Figure S1C). Besides its role in coagulation, TF has been reported to influence the migratory behavior of cells via its interaction with integrins. This cross-talk was described, among other examples, for cell migration on laminin 5 resulting in reduced migration of TF+ cells.22 To analyze whether increased expression of TF reduces the formation of filopodia we analyzed filopodia formation of polarized macrophages when migrating into a laminin 5-

Figure 1. Tissue factor production after polarization of macrophages. (A) Tissue factor (TF) protein was determined on extracellular vesicles from supernatant (n=6) and (B) from lysed cells (n=7) using a specific enzyme-linked immunosorbent assay as indicated in the Methods section. (C) TF activity on extracellular vesicles from polarized macrophages was evaluated using an activity assay as indicated in the Methods section (n=13). Values are given in pg/mL and represent mean values ± standard deviation. (D) Capability of human polarized macrophages to form filopodia when migrating onto laminin-coated areas was evaluated by cytoskeletal staining (n=3). M0: unpolarized macrophages; M(LPS+IFN): classically activated polarized macrophages; M(IL-4+IL13): alternatively activated polarized macrophages; ns: not significant.

456

haematologica | 2021; 106(2)

Alternative polarization induces tissue factor

coated area. Our results indicate that filipodia of macrophages of classically polarized macrophages M(LPS+IFN) were not affected by the laminin 5 coating whereas in alternatively polarized macrophages, M(IL4+IL13), a significant reduction of filopodia was seen under these conditions (Figure 1D). Interestingly, the integrin described for the laminin 5 effect, namely integrin α3 was also downregulated in proinflammatory macrophages (Online Supplementary Figure S1D). A significant induction of TF mRNA after IL-4 and IL-13 polarization in human macrophages was seen after 2 h and 6 h of stimulation (Figure 2A). To further control for the influence of proinflammatory polarization on TF mRNA, we also performed a time course for TF mRNA levels in M(LPS+IFN). No significant regulation of TF mRNA was observed in proinflammatory macrophages at 2 h (2.9±4.2 fold the level in M0, P=0.5 [n=3]), 6 h (0.9±0.4 fold the level in M0, P=0.8 [n=3]) and 24 h (1.2±0.7 fold the level in M0, P=0.6 [n=3]). The main signaling pathway observed for IL4 and IL-13 is an activation cascade dependent on signal transducer and activator of transcription (STAT) 6 signaling, as indicated by the increase of phosphorylated STAT6 in macrophages after treatment with IL-4 and IL-13 (Figure 2B). When a STAT6 inhibitor was used, the increase in TFspecific mRNA induced by IL-4 and IL-13 was completely inhibited in these cells (Figure 2C). Blocking poly ADP ribose polymerase (PARP), a possible STAT6 downstream target, also abrogated the IL-4 and IL-13-induced TF expression in these cells (Figure 2D). Macrophages can be differentiated from monocytes via macrophage colony-stimulating factor (MCSF) or via granulocyte-macrophage colony-stimulating factor (GMCSF). Baseline extracellular vesicle TF levels were higher in GMCSF-derived macrophages than in MCSF-derived macrophages (Figure 3A). However, as can be seen from Figure 3B, when macrophages generated from monocytes with either MCSF or GMCSF were polarized with IL-4 and IL-13, these cells produced significantly more TF than the

respective M0 macrophages and the respective macrophages polarized with LPS and IFN-γ. Polarization of macrophages to M(LPS+IFN-γ) and M(IL-4+IL-13) is in part reversible but has the potential to change subsequent behavior of already polarized macrophages. TF expression was again significantly induced in macrophages by IL-4 and IL-13 in both previous M0 and M(LPS+IFN-γ). Interestingly, previous M(IL-4+IL-13) polarization prevented a significant increase in TF protein after restimulation (Figure 3C). In contrast to M(LPS+IFN-γ), monocytes treated with LPS and IFN-γ showed significant upregulation of TF (Figure 4A). We found greater methylation around the nuclear factor-kappa B (NF-κB) response element within the TF promoter region in macrophages compared to monocytes obtained from the same donors (Figure 4B). When a chemical demethylating agent23 was used during polarization, TF was induced by LPS and IFN-γ in macrophages, whereas the increase in TF was similar in macrophages polarized with IL-4 and IL-13 and the demethylating agent and in macrophages treated with only IL-4 and IL-13 (Figure 4C). TF is associated with extracellular vesicles. To understand the effect of polarization on extracellular vesicles more globally we determined the changes in extracellular vesicle production in unpolarized and polarized macrophages. As can be seen from Figure 5A, alternative polarization of macrophages using IL-4 and IL-13 resulted in a significant increase of extracellular vesicle production over time when compared to extracellular vesicle production by unpolarized macrophages. In contrast, proinflammatory polarization of macrophages with LPS and IFN-γ did not have a significant effect on the production of extracellular vesicles. Similar results were obtained using an ELISA, which specifically recognizes phosphatidylserine present on extracellular vesicles (Figure 5B). In addition, we used extracellular vesicle RNA as a surrogate marker for the amount of circulating vesicles. The total amount of

Figure 2. Tissue factor-specific mRNA induction by alternative polarization. (A) Tissue factor (TF) mRNA levels at the indicated time points in macrophages after polarization induced by interleukin (IL)-4 and IL-13 (n=3). (B) Phosphorylated STAT6 as determined in macrophages 30 min after IL-4+IL-13-induced polarization in comparison to that in unpolarized M0 macrophages using flow cytometry and specific antibodies as described in the Methods (n=3). A representative image is shown. (C) TF mRNA levels in macrophages 2 h after IL-4+IL-13-induced polarization in the presence and absence of a specific STAT6-inhibitor (S6Inh.) at 250 mM (n=6). (D) TF mRNA levels in macrophages 2 h after IL-4+IL-13-induced polarization in the presence and absence of the PARP-inhibitor PJ34 at 100 mM (n=6). TF mRNA in panels A, C and D was determined by quantitative polymerase chain reaction and GAPDH was used as a housekeeping gene as indicated in the Methods section. Values are given as fold changes compared to the respective unpolarized control (M0) and represent mean values ± standard deviation.

haematologica | 2021; 106(2)

457

P.J. Hohensinner et al.

RNA derived from the extracellular vesicle fraction was highest in vesicles derived from M(IL-4+IL-13) (Figure 5C). This was not, however, due to increased individual loading of extracellular vesicles with RNA, as the percentage of RNA+ extracellular vesicles was similar, at around 45%, for all polarization conditions: M0, M(IL-4+IL-13) and M(LPS+IFN-γ) (Online Supplementary Figure S2A). pSTAT6+ macrophages were found within tumor tissue (Online Supplementary Figure S2B). Staining for TF and the macrophage marker CD206 also revealed macrophages

positive for TF (Figure 6A). Not only did cells stain positive for TF but so too did the intracellular space (Figure 6B). These positive streaks of TF also stained positive for specks of CD206, which might indicate macrophagederived extracellular vesicles. To understand a possible link between TF+ areas and areas positive for CD206 we evaluated tumor tissue from different human donors. At least seven sections from each tumor were scored for the presence of CD206 or TF. We found that especially regions with high TF expression also showed strong

Figure 3. Tissue factor protein induction in different macrophage populations. (A) Tissue factor (TF) protein levels in supernatants from M0 macrophages generated via stimulation with macrophage colony-stimulation factor (MCSF) or granulocyte-macrophage colony-stimulating factor (GMCSF) (n=6). (B) TF protein in supernatants from macrophages that were generated either by stimulation with MCSF or GMCSF and polarized into M(LPS+IFN) and M(IL-4+IL-13) as indicated in the Methods. The respective M0 was used to determine the fold changes induced by the polarization conditions (n=6). (C) Macrophages were polarized for 48 h and afterwards repolarized for 24 h as indicated (n=5). TF protein levels were determined using a specific enzyme-linked immunosorbent assay as indicated in the Methods section. Values are given as fold changes compared to MCSF-differentiated macrophages in panel (A) or the respective unpolarized control (M0) in panels (B) and (C) and represent mean values ± standard deviation. M0: unpolarized macrophages; M(LPS+IFNγ): classically activated polarized macrophages; M(IL-4+IL13): alternatively activated polarized macrophages.

Figure 4. Epigenetic regulation of tissue factor. (A) Tissue factor (TF) expression on the surface of human monocytes cultured for 24 h in the presence of 100 ng/mL lipopolysaccharide (LPS) and 100 ng/mL interferon (IFN)-γ or 20 ng/mL interleukin (IL)-4 and 20 ng/mL IL-13 or without any addition (control) was analyzed by flow cytometry using a specific antibody as described in the Methods. Data are shown as mean fluorescence intensity (MFI) (n=4). (B) Methylation of the TF-promoter was analyzed by quantitative polymerase chain reaction as indicated in the Methods section. Macrophage values are given as fold increases compared to the monocyte value, which was set at 1. Monocytes and macrophages from the same individuals were compared (n=4). (C) TF protein was determined using a specific enzymelinked immunosorbent assay as indicated in the Methods section in the presence of the demethylating agent RG108 at a concentration of 5 mM during polarization. Values are given as fold changes compared to the respective unpolarized control (M0) (n=3) and represent mean values ± standard deviation. MFI: mean fluorescence intensity; M0: unpolarized macrophages; M(LPS+IFN): classically activated polarized macrophages; M(IL-4+IL13): alternatively activated polarized macrophages.

458

haematologica | 2021; 106(2)

Alternative polarization induces tissue factor

CD206 expression and regions with low TF expression were predominantly negative or low for CD206 (Figure 6C). Besides tumors, atherosclerotic plaques also contain areas of high TF expression.10 We, therefore, analyzed the presence of TF in CD206+ regions or CD80+ regions within human atherosclerotic lesions. We were able to detect TF in both CD206+ regions and in CD80+ regions. However, adjusted TF intensity was higher in regions positive for CD206, indicating an association of increased TF presence close to alternatively activated macrophages (Figure 6D). Furthermore, using flow cytometry, we analyzed macrophages isolated from atherosclerotic plaques from ApoE-/- mice fed a high fat diet for 20 weeks. The gating strategy is shown in Online Supplementary Figure S3. Overall, proinflammatory CD80high macrophages were less positive for TF surface expression as compared to CD206high macrophages (Figure 6E). Furthermore, CD206high macrophages had a two-fold higher TF surface intensity staining as determined by a comparison of mean fluorescence intensity of

CD206high macrophages and CD80high macrophages (Figure 6F).

Discussion Based on extensive experimental work and clinical data it has been well established that circulating monocytes are an important source of TF and, that by expressing this major component of coagulation, they play a key role in linking inflammation and thrombosis in various pathologies.10 Much less is known about the expression of TF and its regulation in macrophages. Here, for the first time we provide evidence that in human macrophages the expression of TF is not altered in a proinflammatory environment but is significantly enhanced when these cells are alternatively polarized. When human macrophages were polarized with IL-4 and IL-13 a significant increase in mRNA specific for TF was observed after 2 h and 6 h. Alternative polarization also caused a significant increase in TF protein

Figure 5. Influence of polarization conditions on extracellular vesicle production of human macrophages. (A) Numbers of extracellular vesicles in the supernatant of unpolarized macrophages and macrophages polarized in the presence of 100 ng/mL lipopolysaccharide (LPS) and 100 ng/mL interferon (IFN)-γ or 20 ng/mL interleukin (IL)-4 and 20 ng/mL IL-13 were determined by flow cytometry as indicated in the Methods section. Values are given as total vesicle count per mL: n=9 for 4 h and 12 h; n=12 for 24 h; and n=14 for 48 h. (B) Phosphatidylserine (PS) content of extracellular vesicles in the supernatant of M0 and macrophages polarized in the presence of 100 ng/mL LPS and 100 ng/mL IFN-γ or 20 ng/mL IL-4 and 20 ng/mL IL-13 for 48 h was determined using a specific enzyme-linked immunosorbent assay as indicated in the Methods. Values are given in nM PS (n=6). (C) Total extracellular vesicle-derived RNA was evaluated in the supernatant of M0 and macrophages polarized in the presence of 100 ng/mL LPS and 100 ng/mL IFN-γ or 20 ng/mL IL-4 and 20 ng/mL IL-13 for 48 h as indicated in the Methods section. Values are given in ng/mL RNA (n=5) and represent mean values ± standard deviation. M0: unpolarized macrophages; M(LPS+IFN): classically activated polarized macrophages; M(IL-4+IL13): alternatively activated polarized macrophages.

haematologica | 2021; 106(2)

459

P.J. Hohensinner et al.

in such cells. In contrast, TF production in macrophages was not affected by inflammatory polarization with LPS and IFN-γ. Differentiation of monocytes to macrophages can be induced by MCSF or GMCSF to simulate a chronic inflammatory state or an acute event, respectively.24,25 Both cytokines were reported to result in differentiation of monocytes into mature macrophages, although GMCSFderived macrophages display a higher proinflammatory potential.24 Here we show that human macrophages generated from monocytes by stimulation with GMCSF produce significantly more TF than macrophages generated from monocytes by stimulation with MCSF. We hypothesize that the increased proinflammatory state induced by GMCSF might be responsible for this higher basal TF protein level in GMCSF-derived macrophages as compared to their MCSF-derived counterparts. However, a significant increase in TF after alternative polarization was evident for macrophages differentiated from monocytes by either MCSF or GMCSF. Thus, this increase in TF was not dependent on the initial monocyte to macrophage differentiation by these colony-stimulating factors. Besides coagulation, TF might also modulate the migra-

tory capacity of macrophages in a way similar to that observed for other cell types.22 Several integrins have already been reported to interact with TF.26 Interestingly, integrin α4 was associated with TF+ microvesicles shed from macrophages.27 This proinflammatory shedding was dependent on activation of caspase-1.28 Integrin α3 was associated with TF-dependent migration in keratinocytes.22 Our results indicate that integrin α3 is downregulated with polarization. We speculate that low levels of both integrin α3 and TF in proinflammatory macrophages might be responsible for the ability to form filopodia in laminin 5positive areas. Macrophages within inflammatory tissues or within tumors are derived from differentiated monocytes.29 Signals encountered in the microenvironment have the potential to specifically shape the developing macrophages.30 Previous results demonstrated that stimulating macrophages with pro- or anti-inflammatory cytokines leads to opposing transcriptional functional programs and results in silencing or activation of genes causing a memory function of initial polarization conditions.31,32 Previously, we were able to demonstrate that an initial alternative activation of macrophages led to a subsequent

D E

Figure 6. Staining of tissue factor-positive macrophages in sections of colon carcinoma. (A) Tissue factor (TF) (red) and CD206 (green) were stained in colon cancer tissue using specific antibodies as described in the Methods. CD206+ macrophages positive for TF are indicated with white arrows. The boxed area is shown in more detail. (B) White arrows indicate acellular regions that showed double staining for CD206 and TF (orange), which could represent extracellular vesicles derived from macrophages positive for CD206 and TF. (C) Distribution of CD206+ macrophages was evaluated and scored in areas with low, medium, and high TF density. (D) Human atherosclerotic plaque tissue was stained for TF (green) and either CD206 (red) alternatively activated macrophages or CD80 (red) for proinflammatory macrophages. Adjusted TF intensity to macrophage intensity demonstrated an increase in TF in CD206+ regions. Values are given as adjusted tissue factor intensity (arbitrary units) mean values ± standard deviation (SD) (n=16 patients). (E, F) Mouse macrophages from atherosclerotic plaques were isolated as indicated in the Online Supplement. Proinflammatory macrophages were less positive for TF and showed reduced mean fluorescence intensity compared to alternatively activated CD206 macrophages (n=11). Values represent mean values ± SD. DAPI: 4′,6-diamidino-2-phenylindole; a.u.: arbitrary units.

460

haematologica | 2021; 106(2)

Alternative polarization induces tissue factor

activation of the plasminogen activator inhibitor-1 promoter for inflammatory induction.7 Here we show that TF inducibility with IL-4 and IL-13 was reduced after prior alternative polarization. However, prior inflammatory polarization did not alter the capability of IL-4 and IL-13 to induce TF. Previous reports suggested a dysfunctional mitochondrial phenotype after proinflammatory polarization resulting in an altered capability of IL-4 to induce its target genes and preventing macrophage repolarization.33 Our results suggest that TF is an alternative activationdependent target gene that is not affected by changes in mitochondrial function as TF production could still be induced after alternative polarization of previously proinflammatory polarized macrophages. Furthermore, TF release seems to be delayed and in parallel to microvesicle release as both TF amount and microvesicle number were significantly different in alternatively polarized macrophages 24 h after polarization. In monocytes TF was shown to be induced by NF-κB and AP-1-dependent signaling.34 In murine macrophages an induction of TF was demonstrated after LPS stimulation.35 Human cells were also reported to retain the capacity to react to LPS or IFN-γ for a certain time while differentiating from monocytes to macrophages.36 In addition to NF-κB and AP-1, the TF promoter contains Sp1 binding sites37 and is enhanced by Pin1.38 Furthermore, in mouse macrophages loss of PARP14 was demonstrated to increase the levels of TF expression.39 In addition, PARP14 is in part regulated by STAT6 as STAT6 activation induces a switch in PARP14 from a repressor to a promoter of STAT6 signaling via the ribosylation of histone deacetylases previously recruited to IL-4 response elements.40 Of note, STAT6 is the main downstream target activated by IL-4 and IL-13 signaling.41 In the present work we demonstrate that inhibition of either STAT6 or PARP activity abolished the induction of TF mRNA in macrophages by IL-4 and IL-13. Our findings suggest that macrophages upregulate TF in vivo in a STAT6inducing environment and that TF might be a marker protein for STAT6 signaling and, therefore, alternative polarization in macrophages. Nonetheless, it must be emphasized that macrophages in general express TF already at baseline. There is ample evidence to support the notion that monocytes are an important source of circulating TF. At resting conditions already around 1.5% of CD14+ monocytes are positive for TF.42 When monocytes are exposed to LPS TF expression is quickly and transiently upregulated in these cells.43,44 We have identified in our study a difference of TF induction between monocytes and macrophages as our results indicate that human macrophages do not react to LPS treatment with upregulation of TF but display increased TF production only after alternative polarization. This is somehow in contrast to earlier reports that suggested a similar behavior of monocytes and macrophages in response to defined stimuli.45,46 However, whereas we used MCSF or GMCSF for macrophage maturation those studies were performed on macrophages derived from monocytes in the presence of human serum without a defined macrophage maturation factor suggesting that the cells studied could be monocytes still in transition to macrophages. Epigenetic changes within a promoter region hold the potential to alter the response of cells to certain stimuli.47 We, therefore, determined the methylation state around the NF-κB response element, which is responsible for reacting to LPS signaling, in monocytes and macrophages haematologica | 2021; 106(2)

derived from the same donor and observed a higher grade of methylation of the TF promoter in macrophages as compared to the respective monocytes from the same donor. Interestingly, addition of a demethylating agent restored the capacity of macrophages to react to LPS stimulation with upregulation of TF. We, therefore, hypothesize that the difference between monocytes and macrophages in their respective ability to react towards LPS with an increase in TF could be due to epigenetic changes within the TF promoter. Extracellular vesicles were identified because of their ability to support coagulation.48 Among such vesicles, TFbearing ones were demonstrated to propagate thrombus formation.49 Whereas monocyte-derived, TF-positive extracellular vesicles are common in the circulation50 and monocyte-derived extracellular vesicles are associated with several pathological states including cardiovascular disease and inflammatory disorders,51 little is known about the production of TF-positive extracellular vesicles derived from macrophages. Our results indicate that macrophages show a basal production of extracellular vesicles, which can be increased by alternative activation. In addition, we found that these extracellular vesicles shed from macrophages are phosphatidylserine positive. It should be emphasized that TF it is not only regulated at the level of mRNA expression but also at the post-translational level with phosphatidylserine converting TF from a cryptic to a decrypted prothrombotic form.10,52,53 Our findings described above therefore support the notion that macrophagederived vesicles might have the capacity to activate TF. We also show here that alternative polarization increased both the total amount of extracellular vesicles as well as the amount of TF on these extracellular vesicles significantly as compared to vesicles from unpolarized macrophages or macrophages that had undergone proinflammatory polarization. Finally, our in vitro data showing that alternatively activated macrophages produce TF were supported by immunohistochemical analysis of colon carcinoma sections. As alternative polarization is most common in tumors, we analyzed histochemical sections of colon carcinoma, a tumor which has already been shown to have increased IL-4 expression54 and contain TF+ cells.55 To identify proinflammatory macrophages we used CD80, whereas alternatively activated macrophages were identified by CD206. Tumor-associated macrophages have already been associated with IL-4 activation.56 Here we were able to identify STAT6+ macrophages within these sections of colon carcinoma, indicating IL-4 and IL-13 signaling. Furthermore, macrophages within these sections stained positive for CD206, an established marker for alternatively polarized macrophages, and for TF. We also showed that extracellular TF in part co-localized with extracellular CD206 possibly suggesting the presence of macrophagederived extracellular vesicles in these sections. To understand a possible link between TF+ areas and areas positive for CD206 we evaluated tumor tissue from four different human donors. At least seven sections from each tumor were scored for the presence of CD206 or TF. We found that especially regions with high expression of TF also had high expression of CD206 and regions with low TF expression were predominantly negative or low for CD206. Besides tumor tissue, atherosclerotic plaques have also been demonstrated to contain TF with TF protein localization associated with plaque macrophages.57 Interestingly, 461

P.J. Hohensinner et al.

already back in 1989, Wilcox et al. demonstrated that not all TF was associated with cells,57 a notion our data from both atherosclerotic tissue and cancer tissue would support. Within the atherosclerotic environment macrophages are polarized into different subsets. Both proinflammatory and alternatively activated macrophages were reported to coexist within this environment.58 Our data demonstrate that both proinflammatory CD80+ macrophages as well as alternatively activated CD206+ macrophages are sources for TF. Nonetheless, CD206+ macrophages are associated with an increased production of TF within the atherosclerotic lesion, supporting our in vitro data showing induction of TF through alternative activation. In conclusion, we provide new evidence that alternative polarization of macrophages leads to a procoagulatory phenotype through the expression of TF and the production of TF-containing extracellular vesicles and might change overall cellular behavior including migratory preferences. Together with our previously published results showing that alternatively activated macrophages express

References 1. Glass CK. Genetic and genomic approaches to understanding macrophage identity and function. Arterioscler Thromb Vasc Biol. 2015;35(4):755-762. 2. Cybulsky MI, Cheong C, Robbins CS. Macrophages and dendritic cells: partners in atherogenesis. Circ Res. 2016;118(4):637-652. 3. Murray PJ, Wynn TA. Protective and pathogenic functions of macrophage subsets. Nat Rev Immunol. 2011;11(11):723-737. 4. Murray PJ. Macrophage polarization. Annu Rev Physiol. 2017;79:541-66. 5. Lech M, Anders HJ. Macrophages and fibrosis: how resident and infiltrating mononuclear phagocytes orchestrate all phases of tissue injury and repair. Biochim Biophys Acta. 2013;1832(7):989-997. 6. Porta C, Rimoldi M, Raes G, et al. Tolerance and M2 (alternative) macrophage polarization are related processes orchestrated by p50 nuclear factor kappaB. Proc Natl Acad Sci U S A. 2009;106(35):14978-14983. 7. Hohensinner PJ, Baumgartner J, KralPointner JB, et al. PAI-1 (plasminogen activator inhibitor-1) expression renders alternatively activated human macrophages proteolytically quiescent. Arterioscler Thromb Vasc Biol. 2017;37 (10):1913-1922. 8. Mantovani A, Sozzani S, Locati M, Allavena P, Sica A. Macrophage polarization: tumorassociated macrophages as a paradigm for polarized M2 mononuclear phagocytes. Trends Immunol. 2002;23(11):549-555. 9. Sica A, Mantovani A. Macrophage plasticity and polarization: in vivo veritas. J Clin Invest. 2012;122(3):787-795. 10. Grover SP, Mackman N. Tissue factor: an essential mdiator of hemostasis and trigger of thrombosis. Arterioscler Thromb Vasc Biol. 2018;38(4):709-725. 11. Pawlinski R, Wang JG, Owens AP 3rd, et al. Hematopoietic and nonhematopoietic cell tissue factor activates the coagulation cascade in endotoxemic mice. Blood. 2010;116(5):806-814. 12. Berckmans RJ, Sturk A, van Tienen LM, Schaap MC, Nieuwland R. Cell-derived vesicles exposing coagulant tissue factor in saliva. Blood. 2011;117(11):3172-3180.

462

increased levels of the antifibrinolytic serpin plasminogen activator inhibitor-1,7 our results suggest that alternatively polarized macrophages support thrombus formation and suppress thrombolysis59 and thus represent a novel cellular mediator linking macrophages and thrombosis in pathologies characterized by these events. Disclosures This work was supported by a grant from the FWF, the Austrian Science Fund to JW (SFB-54). Contributions PJH, JM, JK, JKP, BT, CK, LH, PH, MM, SS and KW performed the research; JKPWSS analyzed the data; PJH and JW wrote the manuscript; JAS, SD, MBF, KH, CH, RO, IP and JW supervised the study. Acknowledgments This work was supported by a grant from the FWF, the Austrian Science Fund, to IP, JS and JW (SFB-54).

13. Gardiner C, Harrison P, Belting M, et al. Extracellular vesicles, tissue factor, cancer and thrombosis - discussion themes of the ISEV 2014 Educational Day. J Extracell Vesicles. 2015;4:26901. 14. Butler JT, Abdelhamed S, Kurre P. Extracellular vesicles in the hematopoietic microenvironment. Haematologica. 2018; 103(3):382-394. 15. Ettelaie C, Collier ME, Maraveyas A, Ettelaie R. Characterization of physical properties of tissue factor-containing microvesicles and a comparison of ultracentrifuge-based recovery procedures. J Extracell Vesicles. 2014;3. 16. Yanez-Mo M, Siljander PR, Andreu Z, et al. Biological properties of extracellular vesicles and their physiological functions. J Extracell Vesicles. 2015;4:27066. 17. Tripisciano C, Weiss R, Eichhorn T, et al. Different potential of extracellular vesicles to support thrombin generation: contributions of phosphatidylserine, tissue factor, and cellular origin. Sci Rep. 2017;7(1):6522. 18. Hohensinner PJ, Baumgartner J, Ebenbauer B, et al. Statin treatment reduces matrix degradation capacity of proinflammatory polarized macrophages. Vascul Pharmacol. 2018;110:49-54. 19. Thaler B, Hohensinner PJ, Krychtiuk KA, et al. Differential in vivo activation of monocyte subsets during low-grade inflammation through experimental endotoxemia in humans. Sci Rep. 2016;6:30162. 20. Khorana AA, Francis CW, Menzies KE, et al. Plasma tissue factor may be predictive of venous thromboembolism in pancreatic cancer. J Thromb Haemost. 2008;6(11): 1983-1985. 21. Hohensinner PJ, Takacs N, Kaun C, et al. Urokinase plasminogen activator protects cardiac myocytes from oxidative damage and apoptosis via hOGG1 induction. Apoptosis. 2017;22(8):1048-1055. 22. Dorfleutner A, Hintermann E, Tarui T, Takada Y, Ruf W. Cross-talk of integrin alpha3beta1 and tissue factor in cell migration. Mol Biol Cell. 2004;15(10):4416-4425. 23. Brueckner B, Garcia Boy R, Siedlecki P, et al. Epigenetic reactivation of tumor suppressor genes by a novel small-molecule inhibitor of human DNA methyltransferases. Cancer

Res. 2005;65(14):6305-6311. 24. Hamilton JA. Colony-stimulating factors in inflammation and autoimmunity. Nat Rev Immunol. 2008;8(7):533-544. 25. Hamilton TA, Zhao C, Pavicic PG, Jr., Datta S. Myeloid colony-stimulating factors as regulators of macrophage polarization. Front Immunol. 2014;5:554. 26. Kocaturk B, Versteeg HH. Tissue factor-integrin interactions in cancer and thrombosis: every Jack has his Jill. J Thromb Haemost. 2013;11 Suppl 1:285-293. 27. Rothmeier AS, Marchese P, Langer F, et al. Tissue factor prothrombotic activity is regulated by integrin-arf6 trafficking. Arterioscler Thromb Vasc Biol. 2017;37(7): 1323-1331. 28. Rothmeier AS, Marchese P, Petrich BG, et al. Caspase-1-mediated pathway promotes generation of thromboinflammatory microparticles. J Clin Invest. 2015;125 (4):1471-1484. 29. Wynn TA, Chawla A, Pollard JW. Macrophage biology in development, homeostasis and disease. Nature. 2013;496 (7446):445-455. 30. Hoeksema MA, Glass CK. Nature and nurture of tissue-specific macrophage phenotypes. Atherosclerosis. 2019;281:159-167. 31. Czimmerer Z, Daniel B, Horvath A, et al. The transcription factor STAT6 mediates direct repression of inflammatory enhancers and lmits activation of alternatively polarized macrophages. Immunity. 2018;48(1): 75-90 e6. 32. Qiao Y, Kang K, Giannopoulou E, Fang C, Ivashkiv LB. IFN-Îł induces histone 3 lysine 27 trimethylation in a small subset of promoters to stably silence gene expression in human macrophages. Cell Rep. 2016;16(12): 3121-3129. 33. Van den Bossche J, Baardman J, Otto NA, et al. Mitochondrial dysfunction prevents repolarization of inflammatory macrophages. Cell Rep. 2016;17(3):684-696. 34. Mackman N, Brand K, Edgington TS. Lipopolysaccharide-mediated transcriptional activation of the human tissue factor gene in THP-1 monocytic cells requires both activator protein 1 and nuclear factor kappa B binding sites. J Exp Med. 1991;174(6):15171526.

haematologica | 2021; 106(2)

Alternative polarization induces tissue factor

35. Ahamed J, Niessen F, Kurokawa T, et al. Regulation of macrophage procoagulant responses by the tissue factor cytoplasmic domain in endotoxemia. Blood. 2007;109 (12):5251-5259. 36. Scheibenbogen C, Moser H, Krause S, Andreesen R. Interferon-gamma-induced expression of tissue factor activity during human monocyte to macrophage maturation. Haemostasis. 1992;22(4):173-178. 37. Moll T, Czyz M, Holzmuller H, et al. Regulation of the tissue factor promoter in endothelial cells. Binding of NFÎşB-, AP-1-, and Sp1-like transcription factors. J Biol Chem. 1995;270(8):3849-3857. 38. Kurakula K, Koenis DS, Herzik MA Jr, et al. Structural and cellular mechanisms of peptidyl-prolyl isomerase Pin1-mediated enhancement of tissue factor gene expression, protein half-life, and pro-coagulant activity. Haematologica. 2018;103(6):10731082. 39. Iqbal MB, Johns M, Cao J, et al. PARP-14 combines with tristetraprolin in the selective posttranscriptional control of macrophage tissue factor expression. Blood. 2014;124(24):3646-3655. 40. Mehrotra P, Riley JP, Patel R, Li F, Voss L, Goenka S. PARP-14 functions as a transcriptional switch for Stat6-dependent gene activation. J Biol Chem. 2011;286(3):1767-1776. 41. Goenka S, Kaplan MH. Transcriptional regulation by STAT6. Immunol Res. 2011;50(1):87-96. 42. Egorina EM, Sovershaev MA, Bjorkoy G, et al. Intracellular and surface distribution of monocyte tissue factor: application to inter-

haematologica | 2021; 106(2)

subject variability. Arterioscler Thromb Vasc Biol. 2005;25(7):1493-1498. 43. Franco RF, de Jonge E, Dekkers PE, et al. The in vivo kinetics of tissue factor messenger RNA expression during human endotoxemia: relationship with activation of coagulation. Blood. 2000;96(2):554-559. 44. Gregory SA, Morrissey JH, Edgington TS. Regulation of tissue factor gene expression in the monocyte procoagulant response to endotoxin. Mol Cell Biol. 1989;9(6):27522755. 45. Miserez R, Jungi TW. LPS-induced, but not interferon-gamma-induced procoagulant activity of suspended human macrophages is followed by a refractory state of low procoagulant expression. Thromb Res. 1992;65(6):733-744. 46. Schwager I, Jungi TW. Effect of human recombinant cytokines on the induction of macrophage procoagulant activity. Blood. 1994;83(1):152-160. 47. Jaenisch R, Bird A. Epigenetic regulation of gene expression: how the genome integrates intrinsic and environmental signals. Nat Genet. 2003;33 Suppl:245-254. 48. Wolf P. Nature and significance of platelet products in human plasma. Br J Haematol. 1967;13(3):269-288. 49. Key NS. Analysis of tissue factor positive microparticles. Thromb Res. 2010;125 Suppl 1:S42-S45. 50. Angelillo-Scherrer A. Leukocyte-derived microparticles in vascular homeostasis. Circ Res. 2012;110(2):356-369. 51. Suades R, Padro T, Badimon L. The role of blood-borne microparticles in inflammation

and hemostasis. Semin Thromb Hemost. 2015;41(6):590-606. 52. Bach RR, Moldow CF. Mechanism of tissue factor activation on HL-60 cells. Blood. 1997;89(9):3270-3276. 53. Zelaya H, Rothmeier AS, Ruf W. Tissue factor at the crossroad of coagulation and cell signaling. J Thromb Haemost. 2018;16(10): 1941-1952. 54. Baier PK, Wolff-Vorbeck G, Eggstein S, Baumgartner U, Hopt UT. Cytokine expression in colon carcinoma. Anticancer Res. 2005;25(3B):2135-2139. 55. Eriksson O, Asplund A, Hegde G, et al. A stromal cell population in the large intestine identified by tissue factor expression that is lost during colorectal cancer progression. Thromb Haemost. 2016;116(6):1050-1059. 56. Wang HW, Joyce JA. Alternative activation of tumor-associated macrophages by IL-4: priming for protumoral functions. Cell Cycle. 2010;9(24):4824-4835. 57. Wilcox JN, Smith KM, Schwartz SM, Gordon D. Localization of tissue factor in the normal vessel wall and in the atherosclerotic plaque. Proc Natl Acad Sci U S A. 1989;86(8):2839-2843. 58. Stoger JL, Gijbels MJ, van der Velden S, et al. Distribution of macrophage polarization markers in human atherosclerosis. Atherosclerosis. 2012;225(2):461-468. 59. Semeraro F, Ammollo CT, Semeraro N, Colucci M. Tissue factor-expressing monocytes inhibit fibrinolysis through a TAFImediated mechanism, and make clots resistant to heparins. Haematologica. 2009;94(6):819-826.

463

ARTICLE Ferrata Storti Foundation

Red Cell Biology & its Disorders

Hemoglobin switching in mice carrying the Klf1Nan variant

Anne Korporaal,1 Nynke Gillemans,1 Steven Heshusius,1,2 Ileana Cantú,1 Emile van den Akker,2 Thamar B. van Dijk,1 Marieke von Lindern2 and Sjaak Philipsen1

Erasmus MC Department of Cell Biology, Rotterdam and 2Sanquin Research Department of Hematopoiesis, Amsterdam and Landsteiner Laboratory, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands 1

Haematologica 2021 Volume 106(2):464-473

ABSTRACT

Correspondence: SJAAK PHILIPSEN j.philipsen@erasmusmc.nl Received: October 4, 2019. Accepted: January 23, 2020. Pre-published: May 28, 2020. https://doi.org/10.3324/haematol.2019.239830

464

aploinsufficiency for transcription factor KLF1 causes a variety of human erythroid phenotypes, such as the In(Lu) blood type, increased HbA2 levels, and hereditary persistence of fetal hemoglobin. Severe dominant congenital dyserythropoietic anemia IV (CDA-IV) (OMIM 613673) is associated with the KLF1 p.E325K variant. CDA-IV patients display ineffective erythropoiesis and hemolysis resulting in anemia, accompanied by persistently high levels of embryonic and fetal hemoglobin. The mouse Nan strain carries a variant in the orthologous residue, KLF1 p.E339D. Klf1Nan causes dominant hemolytic anemia with many similarities to CDA-IV. Here we investigated the impact of Klf1Nan on the developmental expression patterns of the endogenous α-like and β-like globins, and the human β-like globins carried on a HBB locus transgene. We observe that the switch from primitive, yolk sac-derived, erythropoiesis to definitive, fetal liver-derived, erythropoiesis is delayed in Klf1wt/Nan embryos. This is reflected in globin expression patterns measured between embryonic day 12.5 (E12.5) and E14.5. Cultured Klf1wt/Nan E12.5 fetal liver cells display growth- and differentiation defects. These defects likely contribute to the delayed appearance of definitive erythrocytes in the circulation of Klf1wt/Nan embryos. After E14.5, expression of the embryonic/fetal globin genes is silenced rapidly. In adult Klf1wt/Nan animals, silencing of the embryonic/fetal globin genes is impeded, but only minute amounts are expressed. Thus, in contrast to human KLF1 p.E325K, mouse KLF1 p.E339D does not lead to persistent high levels of embryonic/fetal globins. Our results support the notion that KLF1 affects gene expression in a variant-specific manner, highlighting the necessity to characterize KLF1 variant-specific phenotypes of patients in detail.

Introduction KLF1 is an erythroid-specific transcription factor with diverse and essential roles during terminal erythroid differentiation.1 Cloned from mouse erythroleukemia cells it was initially called erythroid Krüppel-like factor (EKLF) in honor of its erythroid-specific expression and DNA-binding domain.2 This domain is composed of three Cys -His zinc fingers similar to those found in the Drosophila Krüppel transcription factor. KLF1 is the founding member of the KLF branch of the 27-strong SP/KLF family.3 Despite the fact that other SP/KLF factors such as SP1, SP3, KLF2, KLF3 and KLF8 are abundantly expressed in erythroid cells,4,5 gene inactivation in mice demonstrated that KLF1 is essential for definitive erythropoiesis.6,7 Specifically, activation of β-globin expression was strongly impaired leading to a severe β-thalassemia phenotype. Restoring globin chain imbalance with a human γ-globin transgene failed to rescue the embryonic lethality of KLF1 deficiency,8 indicating that other KLF1 target genes contribute to the erythroid defects. This was confirmed by genome-wide expression analyses which established that KLF1 is involved in virtually every aspect of terminal erythroid differentiation.9-12 In humans, the first KLF1 variants reported were identified as the molecular basis of the rare blood type In(Lu).13 In all cases one normal KLF1 allele was present showing that, similar to Klf1wt/ko mice,6,7 this is sufficient to sustain erythropoiesis. These 2

haematologica | 2021; 106(2)

Hemoglobin switching in Klf1wt/Nan mice

observations were extended by the discovery that KLF1 haploinsufficiency caused hereditary persistence of fetal hemoglobin (HPFH) in a Maltese pedigree.14 To date, over 140 different KLF1 variants have been reported, and these have been linked to a broad range of hitherto unrelated human red cell disorders.1 In Sardinia, Thailand and southern China the frequency of KLF1 variants reaches endemic proportions, e.g., 1.25% in southern China.15 In these populations, cases with compound heterozygosity for KLF1 variants occur.16-18 In such cases, one allele invariably carries a missense variant which retains partial activity. KLF1 compound heterozygotes display more pronounced phenotypes, including HbF levels of >20%,1618 persistence of embryonic globins,18 microcytic hypochromic anemia,16,18 and pyruvate kinase deficiency.18 One case of a KLF1 null neonate was reported.19 This infant displayed severe non-spherocytic hemolytic anemia with elevated HbF levels (>70%). Thus, the vast majority of KLF1 variants displays classical autosomal recessive inheritance, and KLF1 haploinsufficiency is associated with mild erythroid phenotypes. The exception is a KLF1 variant in which an ultra-conserved residue in the second zinc finger is affected. This KLF1 p.E325K variant causes congenital dyserythropoietic anemia type IV (CDA-IV; OMIM 613673).20-22 CDA-IV patients suffer from severe hemolytic anemia, splenomegaly, elevated HbF, iron overload, and dyserythropoiesis. Notably, the mouse neonatal anemia (Nan) phenotype is caused by a variant of the orthologous residue in mouse KLF1 p.E339D.23,24 Klf1Nan displays semidominant inheritance. animals Klf1wt/Nan suffer from hemolytic anemia while Klf1Nan/Nan embryos die around embryonic day 10.5 (E10.5) due to failure of pri-mitive erythropoiesis.23,24 This phenotype is more severe than that of Klf1ko/ko embryos which die around E14.5 due to failure of definitive erythropoiesis.6,7 During definitive erythropoiesis KLF1Nan is thought to exert a dominant-ne-gative effect on the function of wild-type KLF1.23,24 In primitive erythropoiesis, KLF1 and KLF2 have compensatory roles25 and the early lethality of Klf1Nan/Nan embryos could therefore be due to interference of KLF1Nan with normal KLF2 function. In addition, KLF1Nan leads to aberrant gene expression which exerts negative effects on erythropoiesis.26,27 In human HBB locus transgenic mice, the fetal HBG1/2 genes are expressed highly in primitive erythrocytes and early definitive erythrocytes. Switching to expression of the adult HBB gene occurs in the fetal liver between E12.5 and E14.5.28 Given the profound deregulation of embryonic and fetal globin genes in CDA-IV patients, we investigated expression of the α-like and β-like globins in Klf1wt/Nan mice carrying a single-copy human HBB locus transgene29 at embryonic, fetal and adult stages of development.

Methods Animals All animal studies were approved by the Erasmus MC Animal Ethics Committee. Mouse strains used were Klf1wt/Nan (C3H101H-Klf1Nan/H30) crossed with PAC8.1 mice carrying a single-copy human HBB locus transgene (Tg(HBB)8.1Gvs29). For details see the Online Supplementary Materials and Methods.

haematologica | 2021; 106(2)

Culture of mouse erythroid progenitors E12.5 fetal livers were disrupted and single-cell suspensions were cultured as described.31 For details see the Online Supplementary Materials and Methods.

RNA isolation and RT-qPCR analyses RNA was extracted using TRI reagent (Sigma-Aldrich). For details see the Online Supplementary Materials and Methods.

Flow cytometry analysis Single-cell suspensions were washed twice with PBS and resuspended in FACS buffer (PBS containing 1% [w/v] bovine serum albumin, 2 mM EDTA). Approximately 106 cells were incubated for 30 minutes at room temperature with the appropriate antibodies. Data were acquired on a Fortessa instrument (BD Biosciences), and analyzed with FlowJo software v10 (Tree Star). For details see the Online Supplementary Materials and Methods.

Cell morphology Cell morphology was analyzed using cytospins stained with May Grünwald-Giemsa (Medion Diagnostics) and O-dianisidine (Sigma-Aldrich).32 Pictures were taken with a BX40 microscope (40x objective, NA 0.65) equipped with a DP50 CCD camera and Viewfinder Lite 1.0 acquisition software (all Olympus).

Statistical analysis Statistical analysis of gene expression data was performed by using Mann-Whitney tests (GraphPad Prism). Excel 2010 was used to draw the graphs. Standard deviations and P-values <0.05 are displayed in the relevant figures (*).

Results Expression patterns of the globin genes in Klf1wt/Nan mice In order to assess the impact of KLF1Nan on developmental regulation of globin gene expression, the Klf1wt/Nan strain30 was crossed with PAC8.1 mice carrying a singlecopy human HBB locus transgene.29 RT-qPCR analysis was performed to determine α-like and β-like globin expression in Klf1wt/wt::HBB (control) and Klf1wt/Nan::HBB (Klf1wt/Nan) embryos at E11.5, E12.5, E13.5, E14.5 and E16.5. Primer pairs were designed to amplify the mouse and human embryonic, fetal and adult globin mRNA specifically, aiming to minimize inter-globin and inter-species crossreactivity. Of note, the human HBG1 and HBG2 genes, encoding Aγ- and Gγ-globin respectively, arose by a recent duplication event and expression of these genes is assessed by a single primer pair. The same is true for the mouse Hba-a1/Hba-a2 genes, encoding α1- and α2-globin respectively, and Hbb-b1/Hbb-b2 genes, encoding majorand minor-globin respectively (see the Online Supplementary Information). Thus, we measured expression of mouse α-like globins (z and α1/2), mouse β-like globins (ey, βh1 and β major/minor) and human β-like globins (e, Aγ/Gγ, and β). For reasons of clarity we will refer to mouse α-like globins as mz and mα, mouse β-like globins as mey, mβh1 and mβ, and human β-like globins as he, hγ, and hβ in the remainder of this paper.

Globin gene expression patterns during development The first erythroid cells are derived extra-embryonically from the blood islands in the yolk sac. They enter the cir465

A. Korporaal et al.

Figure 1. Developmental expression patterns of mouse β-like globins in control and Klf1wt/Nan embryos. Expression of mey, mβh1 and mβ was determined by reverse transcriptase quantitatice PCR. Data are displayed as fraction of total mβ-like globin (mey+mβh1+mβ) expression. Embryonic day (E) and number of embryos (N) are indicated. *P<0.05; error bars indicate standard deviations.

culation in the embryo properly after E8.5 as large nucleated cells referred to as primitive erythrocytes.33 Expression of embryonic globins is a distinctive hallmark of primitive erythrocytes. The first intra-embryonically derived erythrocytes appear in the circulation around E12.5. These enucleated cells are generated in the fetal liver and referred to as definitive erythrocytes. They are characterized by predominant expression of adult globins; unlike the human HBB locus the mouse Hbb locus does not harbor fetal β-like globin genes. Nevertheless, mice carrying human HBB locus transgenes have been extensively used to study fetal-to-adult hemoglobin switching.34 In order to analyze the developmental dynamics of globin expression in Klf1wt/Nan embryos, we determined the globin expression profiles in RNA isolated from yolk sacs and fetal livers harvested between E11.5 and E16.5. Both the yolk sac and the fetal liver contain tissue cells plus circulating blood cells. First, we assessed expression of mβ-like globins. At E11.5, both the yolk sac and the fetal liver of control mice contained mainly mee globin, reflecting the presence of primitive erythrocytes in the circulation and the fact that the fetal liver only just starts to produce definitive erythroid cells (Figure 1 A and C). In E12.5 fetal liver the production of large numbers of definitive erythroid cells was reflected in the dominant expression of mβ, while meγ still constituted >90% of mβ-like globins in the yolk sac (Figure 1 A and C). At E13.5, expression of mβ-like globins detected in yolk sac was 70% mee and 30% mβ, whereas >95% of fetal liver mβlike globin was mβ. Finally, mainly mβ was detected in fetal liver and yolk sac from E14.5 onward (Figure 1 A and 466

C). In comparison, Klf1wt/Nan yolk sac and fetal liver expressed a larger fraction of mβ at both E12.5 and day E13.5 (Figure 1 B and D). The increase of mβ expression over time is delayed in Klf1wt/Nan yolk sac and fetal liver (Figure 1B-D), indicating a delayed shift in the expression of primitive to definitive mβ-like globins. Next, we investigated whether this delay is specific for the Hbb locus, or whether it also occurs in the Hba locus. Yolk sacs and fetal livers from E11.5 control embryos expressed mz as the major α-like globin, with mα contributing 25-30% to total α-like globin expression (Figure 2 A and C). At E12.5 mα became the dominant α-like globin in fetal liver, mz was gradually replaced in the yolk sac by mα at E12.5 and E13.5, with the major shift to mα expression completed by E14.5 (Figure 2A), corresponding with fetal liver output in circulation. A different pattern was observed in yolk sacs and fetal livers from Klf1wt/Nan embryos. Compared to the controls, at E11.5 the contribution of mz was increased at the expense of mα (Figure 2B and D). At E12.5 and E13.5 there was no increase in mα globin expression in the yolk sac (Figure 2B), and the increase in expression of mα in fetal livers was reduced compared to control fetal livers (Figure 2B and D). Thus, Klf1wt/Nan affects the developmental expression patterns of the mouse α-like and β-like globins in a very similar manner, displaying a delayed switch to expression of the adult genes. As CDA-IV patients display very high HbF levels,20-22 we extended the observations on the mouse globins to the human β-like globin genes thus adding analysis of the developmental expression patterns of fetal-stage globin haematologica | 2021; 106(2)

Hemoglobin switching in Klf1wt/Nan mice

Figure 2. Developmental expression patterns of mouse α-like globins in control and Klf1wt/Nan embryos. Expression of mz and mα was determined by reverse transcriptase quantitative PCR. Data are displayed as fraction of total mα-like globin (mz+mα) expression. Embryonic day (E) and number of embryos (N) are indicated. *P<0.05; error bars indicate standard deviations.

genes. The E11.5 yolk sac and fetal liver of control embryos expressed very similar ratios of he and hγ, while expression of hβ was very low (Figure 3A). Compared to the controls, E11.5 Klf1wt/Nan yolk sac and fetal liver displayed a small but significant shift to he expression. Relatively increased he expression was also observed in Klf1wt/Nan E12.5 yolk sac and fetal liver (Figure 3B and D). Next to the difference in he expression at E12.5, hβ made up ~75% of total hβ-like globins in control compared to ~43% in Klf1wt/Nan fetal liver. This apparent lag in switching to hβ expression was also observed in Klf1wt/Nan E13.5 fetal liver, and in E13.5 and E14.5 Klf1wt/Nan yolk sacs (Figure 3 B and D). In control yolk sacs, expression of hβ increased rapidly to ~35% at E13.5, with hβ remaining the most abundantly expressed hβ-like globin accounting for ~50% of the total output of the HBB locus (Figure 3A). Compared to control yolk sacs, expression of hγ in Klf1wt/Nan E13.5 yolk sacs was even higher at ~60%, with hβ expression also rapidly increasing but reaching a lower level of ~25% of hβ-like globins (Figure 3B). At E14.5, the hγ:hβ ratio shifted to 4:96 in control yolk sacs, while in Klf1wt/Nan yolk sacs this ratio remained higher at 28:72 (Figure 3A and B). At E16.5, hβ expression accounted for >97% of total hβ-like globin in all yolk sacs and fetal livers, showing that hemoglobin switching had quantitatively proceeded to the adult profile in both genotypes (Figure 3). We conclude that expression of he is maintained at higher levels in E11.5-E12.5 Klf1wt/Nan embryos. This is followed by a lag in switching to hβ expression, which favors expression of hγ, in E13.5-E14.5 Klf1wt/Nan embryos. Nevertheless, at E16.5 expression of he and hγ has receded to <3% of total hβ-like globins. haematologica | 2021; 106(2)

Delayed appearance of definitive erythrocytes in the circulation of Klf1wt/Nan embryos Having established that the shift from embryonic to fetal and adult globin expression is delayed in Klf1wt/Nan embryos, we investigated whether this could be due to a delay in embryonic development. The erythroid compartment changes dynamically during mouse development,33 and any alterations in this dynamic change would be reflected in globin expression patterns. Primitive erythrocytes originate in the yolk sac and are the sole erythrocytes in the circulation until E12.5, when the first definitive erythrocytes are released from the fetal liver. As fetal liver erythropoiesis gathers momentum, the majority of cells in the circulation are definitive erythrocytes by E14.5. In contrast to primitive erythrocytes, definitive erythrocytes enucleate before they are released in the circulation. We used this characteristic to determine the contribution of primitive cells to the circulation by making cytospins of peripheral blood collected from E14.5 and E16.5 control and Klf1wt/Nan embryos. Compared to the controls, nucleated erythrocytes were more abundant in cytospins of Klf1wt/Nan blood. They were still easily detected in E16.5 cytospins of Klf1wt/Nan blood, while such cells were virtually absent in control samples (Figure 4A). In order to assess the switch from primitive to definitive erythropoiesis, E14.5 cytospins were split into early, mid and late of litter harvest (Figure 4B). Consistent with the previous results, we observed that the fraction of nucleated cells declined very rapidly at this stage of development, in the controls from ~0.34 at early E14.5 to ~0.01 at late E14.5, and in the Klf1wt/Nan samples from ~0.52 at early E14.5 to ~0.13 at late E14.5. Importantly, compared to the controls the fraction 467

A. Korporaal et al.

Figure 3. Developmental expression patterns of human β-like globins in control and Klf1wt/Nan embryos. Expression of he, hγ and hβ was determined by reverse transcriptase quantitative PCR. Data are displayed as fraction of total hβ-like globin (he+hγ+hβ) expression. Embryonic day (E) and number of embryos (N) are indicated. *P<0.05; error bars indicate standard deviations.

of nucleated cells remained significantly higher in the Klf1wt/Nan samples in all E14.5 litters. Using a CASY cell counter, we determined the cell size distributions in E14.5 blood samples. In E14.5 control samples, the two peaks representing primitive (large) and definitive (small) cells are clearly separated (Figure 4C). In E14.5 Klf1wt/Nan samples, these two peaks are not clearly separated. The apparent continuum of cell sizes is in agreement with the rampant anisocytosis observed in the cytospins of E14.5 Klf1wt/Nan blood (Figure 4A). Finally, we sought to use flow cytometry as an orthogonal approach to determine the contribution of primitive erythrocytes to the circulation. In an attempt to better distinguish primitive from definitive erythrocytes, we performed flow cytometry using CD71 (transferrin receptor), Ter119, and CD9 (Tetraspanin) which is a marker for primitive erythrocytes.35 Compared to the controls, expression of CD71 was slightly increased on Klf1wt/Nan E10.5 primitive cells (Figure 4D). This might indicate a delay in maturation, similar to what has been proposed for Klf1wt/ko reticulocytes.36 Expression of Ter119 was virtually absent in E10.5 Klf1wt/Nan erythrocytes, while CD9 expression was strongly reduced (Figure 4D). At E14.5, CD9 was unable to distinguish primitive from definitive erythrocytes in Klf1wt/Nan blood, in contrast to E14.5 Klf1wt/Nan blood in which a distinct fraction of CD9+ primitive cells was observed (Figure 4D, arrow). Collectively, we conclude that the contribution of primitive erythrocytes to the circulation of Klf1wt/Nan embryos cannot be determined by flow cytome468

try using CD71, Ter119 and CD9 as markers. Despite these technical limitations, the analysis of blood samples is consistent with the notion that, compared to control embryos, the contribution of primitive erythrocytes to the pool of circulating erythrocytes in Klf1wt/Nan embryos is extended during development.

Klf1wt/Nan erythroblasts display impaired proliferation and differentiation The delays in hemoglobin switching and appearance of definitive erythrocytes in the circulation of Klf1wt/Nan embryos suggest that the production of definitive cells in the fetal liver might be affected by impaired proliferation or differentiation. Since KLF1 is critically involved in regulation of the erythroid cell cycle,11,37,38 central to both these processes, we cultured primary cells derived from E12.5 fetal livers to assess the proliferative capacity of Klf1wt/Nan erythroblasts. RT-qPCR analysis at day 6 of culture showed deregulated expression of the embryonic/fetal globin genes in the Klf1wt/Nan cells compared to the controls, demonstrating that this aspect of the phenotype is maintained in the culture system (Online Supplementary Table S1). We have previously shown that Klf1ko/ko E12.5 fetal liver cells expand well when grown under selfrenewal conditions.9 In contrast, Klf1wt/Nan erythroblasts from E12.5 fetal liver expanded very poorly under these growth conditions (Figure 5A). Up to day 3-4 of culture, Klf1wt/Nan fetal liver cells expanded similar to those derived from control embryos. After day 4, expansion of the Klf1wt/Nan cultures haematologica | 2021; 106(2)

Hemoglobin switching in Klf1wt/Nan mice

Figure 4. Analysis of blood from control and Klf1wt/Nan embryos. (A) Cytospins of E14.5 and E16.5 blood isolated from control (Klf1wt/wt) and Klf1wt/Nan embryos, stained with dianisidine and histological dyes.32 (B) Fraction of nucleated cells in blood isolated from control ( Klf1wt/wt) and Klf1wt/Nan littermates at early, mid and late E14.5. *P<0.05; error bars indicate standard deviations. N=3 for each genotype at each developmental time point. (C) Cell size analysis of E14.5 and E16.5 blood isolated from control (Klf1wt/wt) and Klf1wt/Nan embryos, obtained with a CASY cell counter. (D) Histograms of flow cytometry analysis of blood isolated at E10.5 and E14.5 from control (Klf1wt/wt) and Klf1wt/Nan embryos. Arrow indicates CD9+ primitive erythrocytes in Klf1wt/wt E14.5 blood.

slowed down and the percentage of smaller cells increased, suggesting spontaneous differentiation. Consistent with previously reported RT-qPCR data of Klf1wt/Nan fetal liver RNA,24 expression of cell cycle regulators E2F2, E2F4, and P18, all known KLF1 target genes,11 37,38 was downregulated in Klf1wt/Nan cells compared to the controls, while expression of P21 was unchanged (Figure 5B). Next, the cells were switched to differentiation medium at day 6. During differentiation the control cells, but not the Klf1wt/Nan cells, displayed the characteristic differenti-ation divisions, i.e., the cell number increased while cell size decreased (Figure 5 C and D). Cytospins taken at day 2 of differentiation revealed many mature, enucleated and hemoglobinised cells in the control cultures (Figure 5Ea). Rare macrophages still present in the culhaematologica | 2021; 106(2)

tures were surrounded by healthy maturing erythroblasts (Figure 5Eb). Macrophage inclusions resembled nuclei, presumably resulting from phagocytosis of pyrenocytes (expelled erythroid nuclei33). In contrast, the Klf1wt/Nan cultures showed few enucleated cells and the cells displayed much larger nuclei (Figure 5Ec). Macrophages were not surrounded by enucleating erythroblasts, but appeared to engulf the entire erythroblast (Figure 5Ed). Combined these observations suggest that Klf1wt/Nan erythroblasts are impaired in both proliferation and differentiation. It is known that KLF1 blocks progression to myeloid lineages.39 In the Klf1wt/Nan cultures we observed cells with morphological features of megakaryocytes (Figure 5Ee). Flow cytometry analysis of the cultured cells at day 9 revealed that control cultures were essentially free of non-erythroid cells, while Klf1wt/Nan cultures displayed pan469

A. Korporaal et al. A

Figure 5. Erythroid progenitor cultures of control and Klf1wt/Nan embryonic day 12.5 fetal liver cells. (A) Growth curves of primary erythroid progenitors cultured from control (Klf1wt/wt) and Klf1wt/Nan E12.5 fetal livers in StemPro medium containing SCF, dex and EPO. (B) Reverse transcriptase quantitative PCR analysis of cell cycle regulators in cultured erythroid progenitors (n=3 for each group). P-values<0.05; error bars indicate standard deviations. (C-D) Differentiation was induced on day 6, transferring the cells to StemPro medium supplemented with EPO and transferrin. Cell number (C) and cell volume (D) was recorded. (E) At day 2 of differentiation cells were centrifuged on glass slides, fixed and stained with dianisidine and histological dyes.32 (F) Flow cytometry analysis of cells cultured from control (Klf1wt/wt) and Klf1wt/Nan fetal livers at day 9 of culture. Cells were stained with antibodies indicated. The percentage of cells double-positive for the erythroid marker CD71 is shown on top, ± standard deviation. (nd): not detectable, due to the virtual absence of expression of myeloid markers on the control cells.

myeloid markers on 20-50% of the cells (Figure 5F). Of note, the majority of these cells were positive for the erythroid marker CD71. Collectively, these data suggest that compromised lineage fidelity, reduced proliferative capacity and impaired terminal differentiation all contribute to the delay in abundance of definitive erythrocytes in the circulation of Klf1wt/Nan embryos, thus bearing weight on the observed changes in globin expression during embryonic development.

Expression of globins in adult Klf1wt/Nan mice The analysis of developmental expression patterns of the globins demonstrated that by E16.5 Klf1wt/Nan embryos had quantitatively switched to expression of the adult genes (Figures 1-3). However, it remains possible that the maintenance of embryonic/fetal globin silencing is perturbed in adult Klf1wt/Nan mice. Indeed, derepression of embryonic globin genes in the spleen of Klf1wt/Nan mice has been reported.24 In order to investigate this further, we isolated RNA from spleen and bone marrow derived from control and Klf1wt/Nan mice. By RT-qPCR analysis we found that mz and mβh1, but not me, expression was increased between 35-800-fold in Klf1wt/Nan samples check comparison to control samples (Figure 6). For the human β-like globins, we observed 4-9-fold increased expression of he and hγ. In quantitative terms, even in the case of the most highly expressed embryonic globin mz, this amounted to less than 0.3% of total α-like globin. We conclude that maintenance of embryonic/fetal globin silencing is perturbed in the bone marrow and spleen of 470

adult Klf1wt/Nan mice. Since the amount of embryonic/fetal globins produced remains below 0.3% of the total amount of globins, this is a qua-litative rather than a quantitative trait.

Discussion In humans, reduced KLF1 activity has been associated with persistent expression of fetal hemoglobin in adults.1,14 A severe phenotype of hemolytic anemia characterizes patients suffering from CDA-IV, caused by the p.E325K variant in the DNA binding domain of KLF1. In the Klf1wt/Nan mouse, the orthologous glutamic acid residue (p.E339) is changed. Biochemically, these amino acid substitutions are very different. In CDA-IV, the glutamic acid (E) is replaced by a basic amino acid (lysine, K) while in KLF1Nan it is replaced by aspartic acid (D), an acidic amino acid. Despite these biochemically opposing properties, the erythroid phenotypes of CDA-IV patients and Klf1wt/Nan mice share many similarities. A hallmark of CDA-IV patients is that they maintain high expression levels of embryonic and fetal globins. In order to investigate whether this is also the case in Klf1wt/Nan mice, we surveyed expression of the α-like and β-like globins during development. Our main observations are summarized in the Online Supplementary Figure S1. We found that in Klf1wt/Nan embryos, switching from embryonic/fetal to adult globin genes is delayed in the endogenous Hbb and Hba loci, and in the single-copy human HBB locus transhaematologica | 2021; 106(2)

Hemoglobin switching in Klf1wt/Nan mice

Figure 6. Expression of human and mouse globins in bone marrow and spleen of adult control and Klf1wt/Nan animals. Expression of mouse and human globins was determined by reverse transcriptase quantitative PCR. Expression ratios of individual globins in Klf1wt/Nan samples were calculated relative to those observed for the control samples (Klf1wt/wt). Note logarithmic scale of the y-axis. N=3 for each group. *P<0.05; error bars indicate standard deviations.

gene. Two mechanisms may contribute to this phenomenon. Firstly, decreased proliferation and differentiation of fetal erythroblasts delays the replacement of primitive by definitive erythroid cells. Since we isolated RNA from populations of cells, the ratio of primitive/definitive cells will have an impact on the globin levels measured. Secondly, although by E16.5 adult globins are quantitatively the dominant globins in Klf1wt/Nan embryos, the embryonic and fetal globin genes retained expression in adult spleen and bone marrow. Qualitatively, this persistent expression is another phenotypic similarity with CDA-IV patients. However, quantitatively there is a major difference. While in CDA-IV patients HbF levels of up to 37% of total hemoglobin have been reported,20 even the most highly expressed embryonic globin in adult Klf1wt/Nan mice, mz, contributes only 0.3% to the total amount of Îą-like globins. KLF1 activates expression of BCL11A14,40 and LRF,41 two transcriptional repressors directly involved in hemoglobin switching.42,43 RT-qPCR analysis of BCL11A and LRF expression indicates that reduced expression of these two factors contributes to the sustained expression of the embryonic/fetal genes in Klf1wt/nan erythroid cells (Online Supplementary Figure S2). We note that expression of BCL11A is also significantly reduced in Klf1wt/ko cells with little effect on expression of the embryonic/fetal genes.36 Mechanistically, this suggests that the repressor proteins are expressed well above the critical threshold level in mice, and a reduction to 4060% of normal expression would still be sufficient for quantitative silencing of the embryonic/fetal genes. The impaired proliferation of E12.5 fetal liver-derived Klf1wt/Nan erythroblasts was initially surprising because a lack of KLF1 increased proliferation, likely due to impaired spontaneous differentiation.9 Whereas erythropoiesis in Klf1ko/ko mice is severely affected during terminal differentiation, erythropoiesis in Klf1wt/nan mice is much less affected with respect to terminal differentiation. The presence of KLF1Nan not only results in the reduced expression of KLF1 target genes, but also induces expression of genes not normally regulated by KLF1.26,27,44 The combined effects of deregulation of canonical KLF1 target genes and ectopic gene expression likely underlie the observed lineage commitment infidelity and impaired proliferation and differentiation of Klf1wt/nan erythroblast cultures. Defective haematologica | 2021; 106(2)

growth of erythroid progenitor cultures derived from a CDA-IV patient has been reported,45,46 indicating that impaired proliferation of erythroid progenitors is another hallmark that CDA-IV patients and Klf1wt/Nan mice have in common. In contrast, adult Klf1wt/Nan mice expressed mainly adult-type globin genes, as opposed to adult CDA-IV patients who maintain expression of embryonic and fetal globins at substantial levels.20-22,45 Importantly, the dominant effect of KLF1Nan is illustrated by comparison with Klf1wt/ko mice which do not display deregulated expression of mouse embryonic globins in E14.5 yolk sac and fetal liver and adult bone morrow36 (Online Supplementary Figure S3). We found that the effects of KLF1Nan on developmental regulation of globin expression are very consistent, but surprisingly subtle. KLF1Nan affects the dynamics of progression from primitive to definitive erythropoiesis during mouse development. Compared to control embryos, definitive erythrocytes emerge at a later stage as the dominant cell type in the circulation of Klf1wt/Nan embryos. We propose that this is at least in part due to the reduced expansion and differentiation capacity of the fetal liver progenitors, since cultured Klf1wt/Nan E12.5 fetal liver cells display growth and differentiation defects. Consistent with impaired erythroid diffe-rentiation, we observed aberrant expression of erythroid flow cytometry markers (CD71, Ter119, CD9). Furthermore, we found misexpression of myeloid markers, in particular the megakaryocyte marker CD41, indicating that Klf1wt/Nan erythroid progenitors display lineage infidelity. This is akin to the previously reported aberrantly activated megakaryocyte program in Klf1ko/ko erythroid cells.39,47 Collectively, our data support the notion that KLF1CDA and KLF1Nan present with similar but also variant-specific phenotypes. Thus, our study further highlights the need to investigate the effects of individual KLF1 variants in detail.1 The recently developed human adult erythroid progenitor cell lines HUDEP-248 and BEL-A49 could be combined with CRISPR-mediated homology-directed recombination50 to investigate the impact of individual human KLF1 variants on the molecular control of erythropoiesis, with a view to increase understanding of the broad spectrum of human red blood cell disorders caused by KLF1 variants.1 471

A. Korporaal et al.

Disclosures No conflicts of interests to disclose. Contributions AK, NG, IC, SH and EvdA performed experiments. AK, TBvD, SH, SP, EvdA and MvL analyzed data. MvL and SP conceived the study. AK, MvL and SP made the figures and wrote the paper. All authors reviewed the paper and agree with its contents.

References 1. Perkins A, Xu X, Higgs DR, et al. Kruppeling erythropoiesis: an unexpected broad spectrum of human red blood cell disorders due to KLF1 variants. Blood. 2016;127(15):1856-1862. 2. Miller IJ, Bieker JJ. A novel, erythroid cellspecific murine transcription factor that binds to the CACCC element and is related to the Kruppel family of nuclear proteins. Mol Cell Biol. 1993;13(5):2776-2786. 3. Suske G, Bruford E, Philipsen S. Mammalian SP/KLF transcription factors: bring in the family. Genomics. 2005;85(5):551-556. 4. Eaton SA, Funnell AP, Sue N, Nicholas H, Pearson RC, Crossley M. A network of Kruppel-like Factors (Klfs). Klf8 is repressed by Klf3 and activated by Klf1 in vivo. J Biol Chem. 2008;283(40):26937-26947. 5. Zhang P, Basu P, Redmond LC, et al. A functional screen for Kruppel-like factors that regulate the human gamma-globin gene through the CACCC promoter element. Blood Cells Mol Dis. 2005;35(2):227-235. 6. Nuez B, Michalovich D, Bygrave A, Ploemacher R, Grosveld F. Defective haematopoiesis in fetal liver resulting from inactivation of the EKLF gene. Nature. 1995;375(6529):316-318. 7. Perkins AC, Sharpe AH, Orkin SH. Lethal beta-thalassaemia in mice lacking the erythroid CACCC-transcription factor EKLF. Nature. 1995;375(6529):318-322. 8. Perkins AC, Peterson KR, Stamatoyannopoulos G, Witkowska HE, Orkin SH. Fetal expression of a human Agamma globin transgene rescues globin chain imbalance but not hemolysis in EKLF null mouse embryos. Blood. 2000; 95(5):1827-1833. 9. Drissen R, von Lindern M, Kolbus A, et al. The erythroid phenotype of EKLF-null mice: defects in hemoglobin metabolism and membrane stability. Mol Cell Biol. 2005;25(12):5205-5214. 10. Hodge D, Coghill E, Keys J, et al. A global role for EKLF in definitive and primitive erythropoiesis. Blood. 2006;107(8):33593370. 11. Pilon AM, Arcasoy MO, Dressman HK, et al. Failure of terminal erythroid differentiation in EKLF-deficient mice is associated with cell cycle perturbation and reduced expression of E2F2. Mol Cell Biol. 2008; 28(24):7394-7401. 12. Tallack MR, Magor GW, Dartigues B, et al. Novel roles for KLF1 in erythropoiesis revealed by mRNA-seq. Genome Res. 2012;22(12):2385-2398. 13. Singleton BK, Burton NM, Green C, Brady RL, Anstee DJ. Mutations in EKLF/KLF1 form the molecular basis of the rare blood group In(Lu) phenotype. Blood. 2008; 112(5):2081-2088.

472

Funding Research in our laboratories was funded by the Landsteiner Foundation for Blood Transfusion Research (LSBR 1040 and 1627), the Netherlands Organization for Scientific Research (ZonMw TOP 40-00812-98-12128), the Netherlands Genomics Initiative (NGI Zenith 93511036), and EU fp7 Specific Cooperation Research Project THALAMOSS (306201).

14. Borg J, Papadopoulos P, Georgitsi M, et al. Haploinsufficiency for the erythroid transcription factor KLF1 causes hereditary persistence of fetal hemoglobin. Nat Genet. 2010;42(9):801-805. 15. Liu D, Zhang X, Yu L, et al. KLF1 mutations are relatively more common in a thalassemia endemic region and ameliorate the severity of beta-thalassemia. Blood. 2014;124(5):803-811. 16. Huang J, Zhang X, Liu D, et al. Compound heterozygosity for KLF1 mutations is associated with microcytic hypochromic anemia and increased fetal hemoglobin. Eur J Hum Genet. 2015;23(10):1341-1348. 17. Satta S, Perseu L, Moi P, et al. Compound heterozygosity for KLF1 mutations associated with remarkable increase of fetal hemoglobin and red cell protoporphyrin. Haematologica. 2011;96(5):767-770. 18. Viprakasit V, Ekwattanakit S, Riolueang S, et al. Mutations in Kruppel-like factor 1 cause transfusion-dependent hemolytic anemia and persistence of embryonic globin gene expression. Blood. 2014; 123(10):1586-1595. 19. Magor GW, Tallack MR, Gillinder KR, et al. KLF1-null neonates display hydrops fetalis and a deranged erythroid transcriptome. Blood. 2015;125(15):2405-2417. 20. Arnaud L, Saison C, Helias V, et al. A dominant mutation in the gene encoding the erythroid transcription factor KLF1 causes a congenital dyserythropoietic anemia. Am J Hum Genet. 2010;87(5):721-727. 21. de la Iglesia-Inigo S, Moreno-Carralero MI, Lemes-Castellano A, Molero-Labarta T, Mendez M, Moran-Jimenez MJ. A case of congenital dyserythropoietic anemia type IV. Clin Case Rep. 2017;5(3):248-252. 22. Jaffray JA, Mitchell WB, Gnanapragasam MN, et al. Erythroid transcription factor EKLF/KLF1 mutation causing congenital dyserythropoietic anemia type IV in a patient of Taiwanese origin: review of all reported cases and development of a clinical diagnostic paradigm. Blood Cells Mol Dis. 2013;51(2):71-75. 23. Heruth DP, Hawkins T, Logsdon DP, et al. Mutation in erythroid specific transcription factor KLF1 causes hereditary spherocytosis in the nan hemolytic anemia mouse model. Genomics. 2010;96(5):303-307. 24. Siatecka M, Sahr KE, Andersen SG, Mezei M, Bieker JJ, Peters LL. Severe anemia in the Nan mutant mouse caused by sequenceselective disruption of erythroid Kruppellike factor. Proc Natl Acad Sci U S A. 2010; 107(34):15151-15156. 25. Basu P, Lung TK, Lemsaddek W, et al. EKLF and KLF2 have compensatory roles in embryonic beta-globin gene expression and primitive erythropoiesis. Blood. 2007; 110(9):3417-3425. 26. Nebor D, Graber JH, Ciciotte SL, et al. Mutant KLF1 in adult anemic Nan mice leads to profound transcriptome changes

and disordered erythropoiesis. Sci Rep. 2018;8(1):12793. 27. Planutis A, Xue L, Trainor CD, et al. Neomorphic effects of the neonatal anemia (Nan-Eklf) mutation contribute to deficits throughout development. Development. 2017;144(3):430-440. 28. Strouboulis J, Dillon N, Grosveld F. Developmental regulation of a complete 70kb human beta-globin locus in transgenic mice. Genes Dev. 1992;6(10):1857-1864. 29. de Krom M, van de Corput M, von Lindern M, Grosveld F, Strouboulis J. Stochastic patterns in globin gene expression are established prior to transcriptional activation and are clonally inherited. Mol Cell. 2002;9(6):1319-1326. 30. Lyon MF, Glenister PH, Loutit JF, Peters J. Dominant haemolytic anemia. Mouse News Letters. 1983;68:68. 31. Dolznig H, Kolbus A, Leberbauer C, et al. Expansion and differentiation of immature mouse and human hematopoietic progenitors. Methods Mol Med. 2005;105:323-344. 32. Beug H, Leutz A, Kahn P, Graf T. Ts mutants of E26 leukemia virus allow transformed myeloblasts, but not erythroblasts or fibroblasts, to differentiate at the nonpermissive temperature. Cell. 1984;39(3 Pt 2):579-588. 33. Palis J. Primitive and definitive erythropoiesis in mammals. Front Physiol. 2014; 5:3. 34. Peterson KR. Hemoglobin switching: new insights. Curr Opin Hematol. 2003; 10(2):123-129. 35. Isern J, Fraser ST, He Z, Zhang H, Baron MH. Dose-dependent regulation of primitive erythroid maturation and identity by the transcription factor Eklf. Blood. 2010; 116(19):3972-3980. 36. Esteghamat F, Gillemans N, Bilic I, et al. Erythropoiesis and globin switching in compound Klf1::Bcl11a mutant mice. Blood. 2013;121(13):2553-2562. 37. Gnanapragasam MN, McGrath KE, Catherman S, Xue L, Palis J, Bieker JJ. EKLF/KLF1-regulated cell cycle exit is essential for erythroblast enucleation. Blood. 2016;128(12):1631-1641. 38. Tallack MR, Keys JR, Humbert PO, Perkins AC. EKLF/KLF1 controls cell cycle entry via direct regulation of E2f2. J Biol Chem. 2009;284(31):20966-20974. 39. Tallack MR, Perkins AC. Megakaryocyteerythroid lineage promiscuity in EKLF null mouse blood. Haematologica. 2010; 95(1):144-147. 40. Zhou D, Liu K, Sun CW, Pawlik KM, Townes TM. KLF1 regulates BCL11A expression and gamma- to beta-globin gene switching. Nat Genet. 2010;42(9):742-744. 41. Norton LJ, Funnell APW, Burdach J, et al. KLF1 directly activates expression of the novel fetal globin repressor ZBTB7A/LRF in erythroid cells. Blood Adv. 2017;1(11):685692.

haematologica | 2021; 106(2)

Hemoglobin switching in Klf1wt/Nan mice

42. Masuda T, Wang X, Maeda M, et al. Transcription factors LRF and BCL11A independently repress expression of fetal hemoglobin. Science. 2016;351(6270):285289. 43. Sankaran VG, Xu J, Ragoczy T, et al. Developmental and species-divergent globin switching are driven by BCL11A. Nature. 2009;460(7259):1093-1097. 44. Gillinder KR, Ilsley MD, Nebor D, et al. Promiscuous DNA-binding of a mutant zinc finger protein corrupts the transcriptome and diminishes cell viability. Nucleic Acids Res. 2017;45(3):1130-1143.

haematologica | 2021; 106(2)

45. Ravindranath Y, Johnson RM, Goyette G, Buck S, Gadgeel M, Gallagher PG. KLF1 E325K-associated congenital dyserythropoietic anemia type IV: insights into the variable clinical severity. J Pediatr Hematol Oncol. 2018;40(6):e405-e409. 46. Varricchio L, Planutis A, Manwani D, et al. Genetic disarray follows mutant KLF1E325K expression in a congenital dyserythropoietic anemia patient. Haematologica. 2019;104(12):2372-2380. 47. Frontelo P, Manwani D, Galdass M, et al. Novel role for EKLF in megakaryocyte lineage commitment. Blood. 2007;

110(12):3871-3880. 48. Kurita R, Suda N, Sudo K, et al. Establishment of immortalized human erythroid progenitor cell lines able to produce enucleated red blood cells. PLoS One. 2013; 8(3):e59890. 49. Trakarnsanga K, Griffiths RE, Wilson MC, et al. An immortalized adult human erythroid line facilitates sustainable and scalable generation of functional red cells. Nat Commun. 2017;8:14750. 50. Mali P, Esvelt KM, Church GM. Cas9 as a versatile tool for engineering biology. Nat Methods. 2013;10(10):957-963.

473

ARTICLE Ferrata Storti Foundation

Red Cell Biology & its Disorders

The Coup-TFII orphan nuclear receptor is an activator of the γ-globin gene

Cristina Fugazza,1* Gloria Barbarani,1* Sudharshan Elangovan,1*° Maria Giuseppina Marini,2 Serena Giolitto,1 Isaura Font-Monclus,1 Maria Franca Marongiu,2 Laura Manunza,3 John Strouboulis,4 Claudio Cantù,5 Fabio Gasparri,6 Silvia M.L. Barabino,1 Yukio Nakamura,7 Sergio Ottolenghi,1 Paolo Moi3 and Antonella Ellena Ronchi1

Dipartimento di Biotecnologie e Bioscienze, Università degli Studi di Milano-Bicocca, Milano, Italy; 2Istituto di Ricerca Genetica e Biomedica del Consiglio Nazionale delle Ricerche, Cagliari, Italy; 3Dipartimento di Sanità Pubblica, Medicina Clinica e Molecolare, Università degli Studi di Cagliari, Cagliari, Italy; 4School of Cancer and Pharmaceutical Sciences, King’s College London, London, UK; 5Wallenberg Center for Molecular Medicine, Department of Clinical and Experimental Medicine (IKE), Linköping University, Linköping, Sweden; 6Department of Biology, Nerviano Medical Sciences S.r.l., Nerviano, Milano, Italy and 7University of Tsukuba, Tsukuba, Ibaraki, Japan 1

Haematologica 2021 Volume 106(2):474-482

*CF, GB and SE contributed equally as co-first authors.

°Present address: Genomics Division, Wipro Life Sciences Laboratory, WIPRO Limited, Bengaluru, Karnataka, India

ABSTRACT

he human fetal γ-globin gene is repressed in adulthood through complex regulatory mechanisms involving transcription factors and epigenetic modifiers. Reversing γ-globin repression, or maintaining its expression by manipulating regulatory mechanisms, has become a major clinical goal in the treatment of β-hemoglobinopathies. Here we identify the orphan nuclear receptor Coup-TFII (NR2F2/ARP1) as an embryonic/fetal stage activator of γ-globin expression. We show that Coup-TFII is expressed in early erythropoiesis of yolk sac origin, together with embryonic/fetal globins. When overexpressed in adult cells (including peripheral blood cells from human healthy donors and β039 thalassemic patients) Coup-TFII activates the embryonic/fetal globin genes, overcoming the repression imposed by the adult erythroid environment. Conversely, the knockout of Coup-TFII increases the β/γ+β globin ratio. Molecular analysis indicates that Coup-TFII binds in vivo to the β-locus and contributes to its three-dimensional conformation. Overall, our data identify Coup-TFII as a specific activator of the γglobin gene.

Correspondence: ANTONELLA ELLENA RONCHI1 antonella.ronchi@unimib.it Received: October 30, 2019. Accepted: February 20, 2020. Pre-published: February 27, 2020. https://doi.org/10.3324/haematol.2019.241224

474

Introduction In mammals, the developmental regulation of erythropoiesis ensures the production of the different types of hemoglobin, from embryonic to fetal and adult, required at the different developmental stages. In humans, the switch from γ- to βglobin expression, resulting in the change from fetal hemoglobin (HbF) to adult hemoglobin (HbA) production, is the subject of intense investigation. In fact, the maintenance of expression of the fetal γ-globin gene in the adult can compensate for defects in β-globin synthesis causing β-hemoglobinopathies, the most common monogenic inherited human diseases.1-5 This evidence has focused the research on discovering γ-globin repressors in adult cells, several of which have been identified to-date, the best example being the transcriptional repressor Bcl11a-XL.6 By contrast, less is known about possible specific activators of embryonic/fetal genes in early developmental stages. Early murine erythropoiesis is sustained by two waves of cells of yolk sac origin. The first wave generates “primitive” red blood cells (EryP), which synthesize embryonic eY and βH1-globins; the second wave produces erythro-myeloid progenitors (EMP), which colonize the fetal liver and differentiate into erythroid and haematologica | 2021; 106(2)

Coup-TFII activates γ-globin

myeloid cells. EMP-derived erythroid progenitors later on will generate the first “definitive” erythroid cells, which synthesize mainly adult βmaj and βmin globins and little βH1, prior to the emergence of the “adult” aorta-gonadmesonephros-derived hematopoietic stem cells.7,8 In mice transgenic for the human β-locus, EryP synthesize e- and γ-globin chains, whereas the EMP erythroid progeny mainly expresses γ- and some β-globin; hematopoietic stem cell-derived definitive cells express human β- and very little, if any, γ-globin.8 In this work, we focused on the chicken ovalbumin upstream promoter-transcription factor II (Coup-TFII or NR2F2/ARP1). Coup-TFII is an orphan nuclear receptor essential for many biological processes, as demonstrated by early embryonic death of Coup-TFII knockout mice.9 Conditional knockout models further demonstrated the requirement of COUP-TFII in different developmental programs, such as angiogenesis, organogenesis, cell fate determination and metabolic homeostasis.10 Within the erythroid lineage, Coup-TFII is expressed in the early embryo and its levels decline around embryonic day E12.5 11,12 Coup-TFII was originally identified as part of the NF-E3 complex, proposed to act as a repressor of e- and γ-globin genes on the basis of in vitro DNA binding studies.11,13,14 Subsequently, transfection experiments showed that the binding of Coup-TFII to both e- and γ-globin promoters interferes with that of NF-Y and possibly of other transcription factors and/or cofactors,15-17 resulting in either cooperation or competition.15 Here we show that Coup-TFII is co-expressed with embryonic globins in cells of yolk sac origin. Significantly, Coup-TFII overexpression induces γ-globin (and βH1) in different adult erythroid cellular systems. At the molecular level, overexpression or knockout of Coup-TFII in βK562 cells18 (to our knowledge the only erythroid model system expressing Coup-TFII), alters the γ/β expression ratio and the frequencies of the interactions within the βlocus in opposite directions, further demonstrating that Coup-TFII favors the expression of embryonic/fetal globins.

Methods Cell cultures

β-K562 cells were cultured in RPMI with 10% fetal bovine serum.18 Ex vivo cultures of mouse E13.5 fetal liver were established from human β-locus transgenic mice (a gift from Prof. Frank Grosveld). After expansion of erythroblasts, cells were infected, with a multiplicity of infection of 100, and analyzed 72 h after transduction.19 Human burst-forming unit erythroid cultures from peripheral blood were obtained essentially as described by Migliaccio et al.,20 by using a two-phase protocol (details in the Online Supplementary Methods). Cells were monitored by flow cytometry for the expression of CD71, Ter119/CD235a, CD117 and DNGFR. Experiments on mouse and human samples were approved by the Italian Ministry of Health.

Coup-TFII overexpression Coup-TFII murine cDNA (a gift from Dr. Michèle Studer) was cloned into a bicistronic IRES-eGFP or IRES-DNGFR CSI vector (a gift from Prof. Tariq Enver) (Online Supplementary Figure S1). Exogenous expression was monitored by quantitative real-time polymerase chain reaction (RTqPCR) analysis and/or western haematologica | 2021; 106(2)

blotting and flow cytometry (Figure 3 and Online Supplementary Figures S2, 3 and 5).

Coup-TFII CRISPR/Cas9 knockout The 39900 single guide RNA from the GeCKOv2 library (https://www.addgene.org/crispr/libraries/geckov2) was cloned into the LentiCrisprV2 plasmid (Addgene #52961), as described in the Online Supplementary Methods. After selection, single knockout (KO) cells were isolated to obtain single clones.

Lentiviral production and transduction experiments HEK-293T cells were transfected with the Coup-TFII or empty vector plus the packaging plasmids psPAX2 and pMD-VSVG (www.lentiweb.com). Seventy-two hours after transfection, recombinant particles were collected, concentrated and resuspended in phosphate-buffered saline.

RNA isolation and polymerase chain reaction The procedures used for RNA isolation and RTqPCR are detailed in the Online Supplementary Methods.

Western blotting Protein extracts were prepared according to standard protocols21 and subjected to sodium dodecylsulfate polyacrylamide gel electrophoresis and blotting.

Immunofluorescence Cells were fixed in 4% paraformaldehyde, washed, permeabilized and incubated overnight with the appropriate antibodies. DAPI was added 20 min before the final wash in phosphatebuffered saline. Cells were analyzed using an ArrayScan VTI highcontent screening reader (Thermo-Fisher), as described by Durlak et al.,18 or by flow cytometry.

Chromatin immunoprecipitation, sequencing and analysis

β-K562 overexpressing Coup-TFII were fixed with 1% formaldehyde for 10 min at room temperature. Chromatin was sonicated to a size of about 500 bp. DNA immunoprecipitation was obtained by incubation with the appropriate antibodies and subsequent isolation with protein A-agarose beads. DNA was sequenced on an Illumina platform. The sequencing data were uploaded to the Galaxy web platform.22 Reads were mapped to the human genome (GRCh37/hg19) using the Burrows-Wheeler aligner23 and peaks were detected using the HOMER tool package.24 Peaks in different experiments were called as the same if the minimal overlap of intervals was 1 bp. To identify peaks bound in one experiment but not in another, intervals should not overlap.25,26 Motif discovery was conducted using Multiple Expression motifs for Motif Elicitation (MEME)27 with default parameters.

Chromosome conformation capture Chromosome conformation capture (3C) experiments28,29 are described in detail in the Online Supplementary Methods.

Flow cytometry Transduced cells were washed in phosphate-buffered saline, fixed in 4% paraformaldehyde, stained with appropriate antibodies, acquired on FACSCalibur (Becton-Dickinson) or a CytoflexS (Beckman-Coulter) and analyzed with Summit V4.3 software or CytExpert, respectively.

Primers, antibodies and reagents The primers, antibodies and reagents used in this study are listed in Online Supplementary Tables S1 and S2. 475

C. Fugazza et al. A

Figure 1. Expression profile of Coup-TFII during the early fetal liver globin switching developmental window. (A) Representative western blot to assay the protein levels of Coup-TFII in nuclear extracts from E11.5, E12.5 and E13.5 mouse fetal livers. β-actin was used as a loading control. (B) Quantitative real-time polymerase reaction expression analysis relative to glyceraldehyde-3-phosphate dehydrogenase (Gapdh) (n≥4, error bars: standard error of mean; *P<0.05; **P<0.01; ***P<0.001). (C) Flow cytometry (FC) analysis defining Coup-TFII in erythro-myeloid progenitors (CD41+/CD117+/CD16/14+) and “primitive” red blood cells (EryP: ey+γ+) in fetal liver cells from human β-globin locus transgenic mouse embryos at E11.5. (D) FC analysis identifying Coup-TFII in EryP (ey+γ+) in blood at E11.5. FSC: forward scatter; EMP: erythro-myeloid progenitors.

Statistical analysis

Data are expressed as the mean ± standard error of mean of ≥3 replicates and were analyzed using a paired two-tailed Student ttest (GraphPad Prism).

Results Coup-TFII expression declines during the time of hemoglobin switching To gain insight into the role of Coup-TFII in the process of hemoglobin switching, we first analyzed its expression in mouse fetal liver at E11.5-E13.5, i.e., in the timeframe during which globin switching takes place. Coup-TFII is expressed at E11.5 and both its protein and RNA levels decline sharply (Figure 1A, B) at E12.5-E13.5; its expression profile follows that of the decline of embryonic mouse Hbb-bh1/Hbb-y globins and the parallel accumulation of adult Hbb-b1/2 globins (Figure 1B). This trend is consistent with a potential role of Coup-TFII as an activator of embryonic globins, which is the subject of this work. In order to investigate the role(s) of Coup-TFII in embryonic/fetal to adult hemoglobin switching in greater detail, we took advantage of a mouse model that is transgenic for the entire human HBB locus,30 in which the human γ to β switch occurs in parallel to the embryonic/adult switch of the murine genes. In E11.5 fetal liver, erythropoiesis is sustained by EryP cells derived from the yolk sac which express murine embryonic eY and βH1 globins and human transgenic eand γ-globins,8 and by EMP that will generate the first definitive red blood cells expressing adult β-globins.7,8 476

The overall percentage of Coup-TFII+ cells in the E11.5 fetal liver is about 2% (Figure 1C). Only approximately 0.63% of EryP cells (i.e., ey+γ+) cells express Coup-TFII, whereas approximately 12.2% of the rare (0.06%) EMP population, identified as CD41+/CD117+/CD16/32+ cells,7 express it (Figure 1C). In blood, about 97.7% of cells are ey+γ+, but only a small percentage of them (1.79%) also express Coup-TFII (Figure 1D). This suggests asynchrony between Coup-TFII expression and hemoglobin synthesis, Coup-TFII being predominantly expressed at early stages and then declining during maturation and hemoglobinization. Accordingly, the few ey+γ+ Coup-TFII+ cells found in the fetal liver could possibly represent EryP progenitors undergoing maturation.

Coup-TFII overexpression activates embryonic/fetal globin genes in adult cells To test whether Coup-TFII activates embryonic/fetal globin genes, we first reintroduced Coup-TFII into fetal liver cells isolated from mouse embryos transgenic for the human β-locus at E13.5, when Coup-TFII expression is already greatly reduced (Figure 1). We transduced the expanded proerythroblasts from E13.5 fetal liver with viral particles carrying either the Coup-TFII-DNGFR overexpressing vector, or the empty vector EV-DNGFR control (Online Supplementary Figure S1). As shown in Figure 2A, Coup-TFII significantly increased human γ-globin and mouse βH1 expression, without affecting the expression of mouse (Hbb-b1/2) and human (HBB) adult globins. This resulted in a significantly increased γ/(γ+β) globin ratio. No major changes were observed in α-like mouse globin genes (Online Supplementary Figure S2). haematologica | 2021; 106(2)

Coup-TFII activates γ-globin

Figure 2. Coup-TFII overexpression increases γ- (and βh1)-globin expression. (A) Quantitative real-time polymerase reaction (RTqPCR) analysis of globin gene expression using cDNA from erythroblasts derived from E13.5 ex-vivo fetal liver erythroid cultures transduced with either empty vector (EV) or with the Coup-TFII-overexpressing lentiviral particles (Coup-TFII). Left panel: the relative fold-change on the control is shown (error bars: standard error of mean; *P<0.05, **P<0.01). Expression levels on Gapdh are shown in Online Supplementary Figure S2. Middle panel: the same data are expressed as γ to β ratio (γ+β =1). Right panel: RTqPCR analysis for mouse globin genes. In all experiments >90% cells were transduced as assessed by flow cytometry (FC) analysis for DNGFR expression (Online Supplementary Figure S2). (B) Coup-TFII overexpression in both healthy and thalassemic β039 ex-vivo peripheral blood cultures. The analysis was carried out as in (A). In all experiments, cells were magnetically purified on the basis of DNGFR expression (>85% cells were isolated as assessed by FC analysis on DNGFR expression (Online Supplementary Figure S3). Expression levels on Gapdh are shown in Online Supplementary Figure S3. (C) FC analysis on CD71 and CD235ab (GpA).

This result prompted us to overexpress Coup-TFII in the context of human adult cells, in cultures from peripheral blood from healthy donors and from thalassemic patients carrying the Sardinian β039 mutation, which generates an in-frame stop codon resulting in β-globin mRNA degradation by nonsense mediated decay.31,32 As shown in Figure 2B, Coup-TFII overexpression in normal cells led to a 7-fold increase in γ-globin RNA expression, with no significant change in β-globin expression. Similarly, a 3fold activation of γ-globin was also seen in β039 cells following COUP-TFII overexpression. Of interest, in β039 thalassemic, but not in wild-type cultures, the increase in γ expression was accompanied by a decrease in the level of the residual β RNA produced from the β039 mutationcarrying alleles, which we tested as a control. Overexpressing Coup-TFII in these cells did not result in statistically significant maturational changes (CD71/GpA profiles in Figure 2C). haematologica | 2021; 106(2)

Generation of overexpressing and knockout β-K562 cells

Together, the above data indicate that Coup-TFII is an activator of γ-globin. To better characterize the role of Coup-TFII at the molecular level, we took advantage of βK562 cells, a K562 subclone that, to our knowledge, is the only erythroid cell model system that expresses CoupTFII together with e-, γ- and β-globins18 (Online Supplementary Figure S4). We used these cells to either knockout by CRISPR/Cas9 (KO cells) or overexpress by lentiviral transduction (OE cells) Coup-TFII (Figure 3A). In OE cells, we observed an increase in embryonic HBE and HBG and a decrease in adult HBA and HBB (Figure 3B). By contrast, in KO cells there was a general induction of all globin genes transcription, resulting in increased hemoglobinization (Figure 3B and Online Supplementary Figure S6A). In this context, β-globin expression increased by an average of 45-fold, whereas the increases in α-, e- and γ-globins was about 10 times lower (Figure 3B). This differen477

C. Fugazza et al.

Figure 3. Gain/loss of Coup-TFII expression alters the β to γ ratio in opposite directions. Coup-TFII overexpression (OE) was obtained as described in Figure 2; CoupTFII knockout (KO) was achieved using a CRISPR/Cas9 lentiviral vector and the corresponding empty vectors (EV) were used as a control. (A) Western blot confirming the Coup-TFII KO and OE. U2AF: protein loading control. The Coup-TFII KO was also confirmed by quantitative real-time polymerase reaction RTqPCR (not shown). (B) RTqPCR analysis of individual globin genes in a pool of six single clones at day 60 (KO) and on a pooled population of Coup-TFII OE cells. As a control, a pool of two clones infected with the EV were analyzed. Globin expression levels relative to GAPDH (n≥3, error bars: standard error of mean; *P<0.05) are shown. Right panel: γto β-globin ratio (γ+β=1).

Figure 4. Chromatin immunoprecipitation experiment. (A) Venn diagram illustrating the selection criteria used to identify bona fide Coup-TFII-bound regions. (B) Gene Ontology mouse phenotype association. Erythroid-related phenotypes are shaded in pink. (C) Results of the MEME motif discovery analysis. (D) Pie chart depicting the relative percentages of motifs found by MEME.

478

haematologica | 2021; 106(2)

Coup-TFII activates γ-globin tial increase of β versus γ levels resulted in an increase in the β/(γ+β) ratio from about 20% in cells transduced with the empty vector (EV cells) to about 30% in KO cells; this is in clear contrast to the decrease in the β/(γ+β) ratio from about 20% in EV cells to about 8% in OE cells (Figure 3B, right panel). This result was confirmed at the protein level (Online Supplementary Figure S6).

Identification of Coup-TFII in vivo binding sites by chromatin immunoprecipitation sequencing analysis

To elucidate whether Coup-TFII binds directly to the βlocus in vivo and to identify other high-confidence relevant Coup-TFII genomic target regions in β-K562 cells, we designed a stringent chromatin immunoprecipitation (ChIP)-sequencing experiment. We selected Coup-TFII peaks common to the OE and not transduced (NT) cells, from which we subtracted the background peaks present

also in KO cells (Figure 4A). Almost 100% of the remaining 1,783 immunoprecipitated regions mapped to an associated gene and showed a peak distribution profile within -500 and +500 kb relative to the transcription start site (Online Supplementary Figure S7). Interestingly, Gene Ontology analysis of mouse phenotypes of genes associated with Coup-TFII peaks clearly pointed to hematologic diseases and in particular to erythroid diseases (shaded in pink in Figure 4B). The MEME algorithm27 for enriched motifs discovery identified the canonical (GGTCA) nuclear receptor DNA binding consensus and the GATAfamily (GATAA) consensus as the most represented sequences centered under the Coup-TFII peaks (Figure 4C, D). About 39% of sequences bound by Coup-TFII, also contain the GATA consensus, suggesting that Coup-TFII does indeed participate in the regulation of the erythroidspecific genetic program.33

Figure 5. Coup-TFII binds in vivo to the human β-globin locus. (A) Schematic representation of the human β-globin locus. (B) Close-up of the HS2, HS3 and HS4 core regions within the locus control region (LCR). (C) Validation of the chromatin immunoprecipitation-sequencing sites within the LCR. (D) Chromosome conformation capture experiment on Coup-TFII-overexpressed (red), not transduced (black) and -knockout (blue) cells. The structure of the locus with the relative positions of globin genes are shown at the top of the Figure. The anchor region is shown in black; the positions and sizes of the analyzed interacting fragments are shaded in light gray. (n=3; error bars: standard error of mean; *P<0.05; **P<0.01; ***P<0.001). OE: overexpressed; NT: not transduced; KO: knockout.

haematologica | 2021; 106(2)

479

C. Fugazza et al.

Coup-TFII binds in vivo within the human β-globin locus and contributes to its three-dimensional conformation The ChIP-sequencing data show that Coup-TFII binds in vivo within the β-locus, with the strongest peaks (present in both NT and OE cells but absent in KO cells) mapping within the locus control region (LCR), at hypersensitive sites HS2, HS3 and HS4 (Figure 5A, B). Scanning of the core sequence of these elements using JASPAR software34 identified several potential Coup-TFII consensus sequences within these elements (Online Supplementary Figure S8), which were confirmed by ChIP assays (Figure 5B, C). To determine whether the level of Coup-TFII could influence the chromatin structure of the β-locus, we performed a 3C assay on the OE, NT and KO β-K562 cell lines described above. In this assay, the physical interaction within two nearby loci is measured on the basis of the relative crosslinking efficiency between a predefined anchor region and given downstream positions. In our experiment, in the absence of Coup-TFII the relative frequency of interactions of the γ-gene with the LCR is significantly lower (≈4%) than that of β (≈30%), pointing to an effect on the β-locus favoring β-globin expression (in agreement with data in Figure 3B). By contrast, in the presence of Coup-TFII (OE and NT cells), the interaction of β with the LCR is around 15%. Of note, in KO cells, the absolute frequencies of interactions within the β-locus with the LCR increases, possibly reflecting concomitant overlapping differentiation effects. Of interest, in OE cells, a significant interaction was observed in the Aγ-d region which has been implicated in hereditary persistence of fetal hemoglobin35 (Figure 5D). These results, together with ChIP-sequencing data identi-

fying Coup-TFII peaks in different regions within the βlocus, suggest a potential broad role of Coup-TFII in the differential regulation of globin genes and of the β-locus structure.

Discussion In recent years, several repressors of embryonic/fetal globin genes have been identified in murine and human adult erythroid cells.6,36 These studies have not directly addressed the regulation of these genes in earlier developmental stages. In particular, there is no evidence so far of stage-specific factors activating human and mouse embryonic/fetal globins prior to the switch to adult globin synthesis. Here we show that Coup-TFII is expressed in cells of yolk sac origin (Figure 1) and that its overexpression selectively activates embyonic/fetal globins in a variety of adult erythroid cells, including β039-thal cells (Figure 2). Coup-TFII overexpression increases γ-globin relative to βglobin expression and shows that Coup-TFII is not a repressor of γ-globin, as previously suggested by in vitro DNA binding studies.37 This evidence helps to reconcile the previously unexplained observation that a mutation specifically abolishing Coup-TFII in vitro binding to the γ promoter (-107-108)38 failed to generate the phenotype of hereditary persistence of fetal hemoglobin in a mouse line transgenic for the human β-locus. In the absence of Coup-TFII, the γ- to βglobin ratio is decreased, pointing to a possible alternative interaction between γ- and β-globin genes with the LCR.39 To answer the question of whether Coup-TFII binds directly to the β-locus in vivo, we carried out ChIP-

Figure 6. Hypothetical model for the differential γ-globin activation/silencing in embryonic vs. adult erythroid cells. In embryo/fetal cells, Coup-TFII is expressed and recognizes the GGTCA consensus sequences within the locus control region (LCR) and possibly at other positions within the β-locus. The resulting β-locus architecture favors the expression of γ-genes. In adult cells, the same positions are occupied by Bcl-11A. The switch between the two factors results in γ-silencing. The overexpression of Coup-TFII is capable of reverting γ-silencing in adult cells, suggesting that the erythroid cellular environment is similar in embryo/fetal and adult cell types, possibly because of the presence of a large set of common factors/cofactors. Whether Coup-TFII and Bcl-11A co-exist in some transient cellular stage during development or are distinctly expressed in different cells, remains to be elucidated. Continuous-line arrows point to the strong binding sites within the LCR; dashed-line arrows point to the other possible binding sites within the locus.

480

haematologica | 2021; 106(2)

Coup-TFII activates γ-globin sequencing and 3C experiments in β-K562 cells, which offer the advantage of co-expression of Coup-TFII together with e-, γ- and β-globins, making it possible to evaluate the effects of the modulation of Coup-TFII levels. We demonstrated that Coup-TFII does indeed bind the GGTCA binding sites present within the LCR and that its KO favors LCR-β-globin interactions (Figure 5). The presence of Coup-TFII binding at different positions suggests a possible widespread action of Coup-TFII in the transcriptional regulation of the β-locus. The functional relevance of these single binding sites will require their precise editing, with the caveat that the presence of overlapping/adjacent binding sites, bound by different transcription factors/cofactors (as in the case of the CCAAT box region of γ genes), could complicate the interpretation of the results. Of interest, the Coup-TFII consensus sequence retrieved by MEME (GGTCA) in ChIP-sequencing almost perfectly matches the one described for Bcl11a-XL40 and for TR2/TR4, two known repressors of γ-globin in adult cells.40,41 As for Bcl11a-XL (for which direct binding to the γ promoter could only be seen using the very sensitive CUT&RUN technique40), we could only detect very weak peaks on the CCAAT box region of the γ promoter, although previous in vitro binding and dimethyl sulfate interference experiments clearly mapped the Coup-TFII binding sites to the double CCAAT box γ-promoter region.14 The binding of Coup-TFII to the LCR (and possibly its interaction with the weaker sites on the double CCAAT box region), suggests that it might concur to establish/maintain the architecture of the β-locus in early developmental stages, when preferential γ activation is required, as confirmed by the 3C experiment (Figure 5). The same regions (γ-promoter and LCR) bound by CoupTFII within the LCR, are occupied in adult cells by the γ repressors Bcl11a-XL and/or TR2/TR4 that keep γ-globin expression switched off. Coup-TFII is expressed in an early developmental stage, whereas expression of Bcl11aXL and TR2/TR4 is confined to adult cells. This would suggest that repression of γ-globin in the adult stage may involve a switch between factors binding to the CCAAT box within the γ promoters and potentially within the LCR, i.e., from Coup-TFII in erythroid cells of yolk sac origin to Bcl11a-XL and TR2/TR4 in adult erythroid cells. The existence of specific developmental activators and repressors could be made necessary by the presence in both embryonic/fetal and adult cells of a common set of activators/coactivators (such as Gata1 and its coactivators) that create a “permissive environment” to which the γ promoter is responsive. This hypothesis would also help to explain why Coup-TFII, when overexpressed in

References 1. Forget BG. Molecular basis of hereditary persistence of fetal hemoglobin. Ann N Y Acad Sci. 1998;850:38-44. 2. Nienhuis AW. Gene transfer into hematopoietic stem cells. Blood Cells. 1994;20(1):141-147; discussion 147-148. 3. Thein SL, Craig JE. Genetics of Hb F/F cell variance in adults and heterocellular hereditary persistence of fetal hemoglobin. Hemoglobin. 1998;22(5-6):401-414. 4. Wahlberg K, Jiang J, Rooks H, et al. The

haematologica | 2021; 106(2)

adult cells, is indeed capable of activating γ-globin, overcoming γ-globin repression (Figure 2). Whether Coup-TFII and Bcl11a-XL and/or TR2/TR4 coexist in the same cell in a given transitory status in which they directly compete/substitute each other to allow the switch (from γ “ON” to γ “OFF”), or whether their expression is restricted to non-overlapping spatial and temporal windows of development, remains to be elucidated. The novel role of Coup-TFII as an activator of fetal globin genes provides new insight into the early activation of embryonic/fetal genes and, possibly, on the mechanisms of hemoglobin switching. In the light of these observations, Coup-TFII could be considered as a target for γ-globin reactivation. Although COUP-TFII has been extensively studied genetically, present knowledge of upstream regulatory signals and of possible ligands, remains poor. Retinoids,42-44 sonic hedgehog,45 ETS and their coactivators46 activate COUP-TFII expression in different cellular contexts. Moreover, whereas structural data suggest that the Coup-TFII protein is an “orphan” receptor capable of an auto-repressive conformation, there is evidence that Coup-TFII ligands may exist47 and, in addition, descriptions of synthetic Coup-TF agonists have been recently published.48 Collectively, the evidence that Coup-TFII can be transcriptionally induced and functionally activated suggests that it may be an attractive novel target for γ-globin reactivation in β-hemoglobinopathies. Disclosures No conflicts of interest to disclose. Contributions SE, CF, GB and MGM performed experiments, and analyzed and discussed data; MFM performed FC analysis; LM performed human cultures; CC performed ChIP-seq experiments; FG performed immunofluorescence analysis; IFM and SG performed experiments. YN provided HUDEP cells; JS contributed with support and discussion of the research; PM, SO and SMLB critically discussed the manuscript; AR led the project and wrote the paper. Funding The authors would like to thank Dr. Andrea Ditadi for help with experimental work, Fondazione Cariplo grant n. 2012.0517 for support to AR, PM and JS, the People Programme (Marie Curie Actions) of the European Union's Seventh Framework Programme FP7/2007-2013/ under REA grant agreement n. 289611 (HEM_ID Project) for support to AR and JS and the Knut and Alice Wallenberg foundation for support to CC.

HBS1L-MYB intergenic interval associated with elevated HbF levels shows characteristics of a distal regulatory region in erythroid cells. Blood. 2009;114(6):1254-1262. 5. Wilber A, Nienhuis AW, Persons DA. Transcriptional regulation of fetal to adult hemoglobin switching: new therapeutic opportunities. Blood. 2011;117(15):39453953. 6. Sankaran VG, Orkin SH. The switch from fetal to adult hemoglobin. Cold Spring Harb Perspect Med. 2013;3(1):a011643. 7. McGrath KE, Frame JM, Fegan KH, et al.

Distinct sources of hematopoietic progenitors emerge before HSCs and provide functional blood cells in the mammalian embryo. Cell Rep. 2015;11(12):1892-1904. 8. McGrath KE, Frame JM, Fromm GJ, et al. A transient definitive erythroid lineage with unique regulation of the β-globin locus in the mammalian embryo. Blood. 2011;117(17):4600-4608. 9. Pereira FA, Qiu Y, Zhou G, Tsai MJ, Tsai SY. The orphan nuclear receptor COUP-TFII is required for angiogenesis and heart development. Genes Dev. 1999;13(8):1037-1049.

481

C. Fugazza et al. 10. Lin FJ, Qin J, Tang K, Tsai SY, Tsai MJ. Coup d’etat: an orphan takes control. Endocr Rev. 2011;32(3):404-421. 11. Filipe A, Li Q, Deveaux S, et al. Regulation of embryonic/fetal globin genes by nuclear hormone receptors: a novel perspective on hemoglobin switching. EMBO J. 1999;18(3): 687-697. 12. Cui S, Tanabe O, Sierant M, et al. Compound loss of function of nuclear receptors Tr2 and Tr4 leads to induction of murine embryonic β-type globin genes. Blood. 2015;125(9):1477-1487. 13. Liberati C, Ronchi A, Lievens P, Ottolenghi S, Mantovani R. NF-Y organizes the gamma-globin CCAAT boxes region. J Biol Chem. 1998;273(27):16880-16889. 14. Ronchi AE, Bottardi S, Mazzucchelli C, Ottolenghi S, Santoro C. Differential binding of the NFE3 and CP1/NFY transcription factors to the human gamma- and epsilonglobin CCAAT boxes. J Biol Chem. 1995;270(37):21934-21941. 15. Liberati C, Cera MR, Secco P, et al. Cooperation and competition between the binding of COUP-TFII and NF-Y on human epsilon- and gamma-globin gene promoters. J Biol Chem. 2001;276(45):41700-41709. 16. Cui S, Kolodziej KE, Obara N, et al. Nuclear receptors TR2 and TR4 recruit multiple epigenetic transcriptional corepressors that associate specifically with the embryonic βtype globin promoters in differentiated adult erythroid cells. Mol Cell Biol. 2011;31(16): 3298-3311. 17. Zhu X, Wang Y, Pi W, Liu H, Wickrema A, Tuan D. NF-Y recruits both transcription activator and repressor to modulate tissueand developmental stage-specific expression of human γ-globin gene. PLoS One. 2012;7(10):e47175. 18. Durlak M, Fugazza C, Elangovan S, et al. A novel high-content immunofluorescence assay as a tool to identify at the single cell level γ-globin inducing compounds. PLoS One. 2015;10(10):e0141083. 19. von Lindern M, Deiner EM, Dolznig H, et al. Leukemic transformation of normal murine erythroid progenitors: v- and c-ErbB act through signaling pathways activated by the EpoR and c-Kit in stress erythropoiesis. Oncogene. 2001;20(28):3651-3664. 20. Migliaccio G, Di Pietro R, Di Giacomo V, et al. In vitro mass production of human erythroid cells from the blood of normal donors and thalassemic patients. Blood Cells Mol Dis. 2002;28(2):169-180. 21. Schreiber E, Matthias P, Muller MM, Schaffner W. Rapid detection of octamer binding proteins with 'mini-extracts', pre-

482

pared from a small number of cells. Nucleic Acids Res. 1989;17(15):6419. 22. Afgan E, Baker D, van den Beek M, et al. The Galaxy platform for accessible, reproducible and collaborative biomedical analyses: 2016 update. Nucleic Acids Res. 2016;44(W1):W3-W10. 23. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25(14):1754-1760. 24. Heinz S, Benner C, Spann N, et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol Cell. 2010;38(4):576-589. 25. Ramirez F, Dundar F, Diehl S, Gruning BA, Manke T. deepTools: a flexible platform for exploring deep-sequencing data. Nucleic Acids Res. 2014;42(Web Server issue):W187191. 26. Quinlan AR, Hall IM. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics. 2010;26(6):841-842. 27. Machanick P, Bailey TL. MEME-ChIP: motif analysis of large DNA datasets. Bioinformatics. 2011;27(12):1696-1697. 28. Palstra RJ, Tolhuis B, Splinter E, Nijmeijer R, Grosveld F, de Laat W. The beta-globin nuclear compartment in development and erythroid differentiation. Nat Genet. 2003;35(2):190-194. 29. Xu J, Sankaran VG, Ni M, et al. Transcriptional silencing of γ-globin by BCL11A involves long-range interactions and cooperation with SOX6. Genes Dev. 2010;24(8):783-798. 30. Patrinos GP, de Krom M, de Boer E, et al. Multiple interactions between regulatory regions are required to stabilize an active chromatin hub. Genes Dev. 2004;18(12): 1495-1509. 31. Cao A, Galanello R, Furbetta M, et al. Thalassaemia types and their incidence in Sardinia. J Med Genet. 1978;15(6):443-447. 32. Baserga SJ, Benz EJ, Jr. Beta-globin nonsense mutation: deficient accumulation of mRNA occurs despite normal cytoplasmic stability. Proc Natl Acad Sci U S A. 1992;89(7):29352939. 33. Erdos E, Balint BL. COUP-TFII is a modulator of cell-type-specific genetic programs based on genomic localization maps. J Biotechnol. 2019;301:11-17. 34. Mathelier A, Fornes O, Arenillas DJ, et al. JASPAR 2016: a major expansion and update of the open-access database of transcription factor binding profiles. Nucleic Acids Res. 2016;44(D1):D110-115. 35. Higgs DR, Engel JD, Stamatoyannopoulos G. Thalassaemia. Lancet. 2012;379(9813):

373-383. 36. Suzuki M, Yamamoto M, Engel JD. Fetal globin gene repressors as drug targets for molecular therapies to treat the β-globinopathies. Mol Cell Biol. 2014;34(19): 3560-3569. 37. Ottolenghi S, Nicolis S, Bertini C, Ronchi A, Crotta S, Giglioni B. Regulation of gammaglobin expression in hereditary persistence of fetal hemoglobin. Ann N Y Acad Sci. 1990;612:191-195. 38. Ronchi A, Berry M, Raguz S, et al. Role of the duplicated CCAAT box region in gamma-globin gene regulation and hereditary persistence of fetal haemoglobin. EMBO J. 1996;15(1):143-149. 39. Wijgerde M, Grosveld F, Fraser P. Transcription complex stability and chromatin dynamics in vivo. Nature. 1995;377 (6546):209-213. 40. Liu N, Hargreaves VV, Zhu Q, et al. Direct promoter repression by BCL11A controls the fetal to adult hemoglobin switch. Cell. 2018;173(2):430-442. 41. Tanabe O, McPhee D, Kobayashi S, et al. Embryonic and fetal beta-globin gene repression by the orphan nuclear receptors, TR2 and TR4. EMBO J. 2007;26(9):2295-2306. 42. Jonk LJ, de Jonge ME, Vervaart JM, Wissink S, Kruijer W. Isolation and developmental expression of retinoic-acid-induced genes. Dev Biol. 1994;161(2):604-614. 43. Qiu Y, Krishnan V, Pereira FA, Tsai SY, Tsai MJ. Chicken ovalbumin upstream promoter-transcription factors and their regulation. J Steroid Biochem Mol Biol. 1996;56(1-6 Spec No):81-85. 44. Soosaar A, Neuman K, Nornes HO, Neuman T. Cell type specific regulation of COUP-TF II promoter activity. FEBS Lett. 1996;391(1-2):95-100. 45. Krishnan V, Elberg G, Tsai MJ, Tsai SY. Identification of a novel sonic hedgehog response element in the chicken ovalbumin upstream promoter-transcription factor II promoter. Mol Endocrinol. 1997;11(10): 1458-1466. 46. Petit FG, Salas R, Tsai MJ, Tsai SY. The regulation of COUP-TFII gene expression by Ets-1 is enhanced by the steroid receptor coactivators. Mech Ageing Dev. 2004;125(1011):719-732. 47. Kruse SW, Suino-Powell K, Zhou XE, et al. Identification of COUP-TFII orphan nuclear receptor as a retinoic acid-activated receptor. PLoS Biol. 2008;6(9):e227. 48. Yoon K, Chen CC, Orr AA, Barreto PN, Tamamis P, Safe S. Activation of COUP-TFI by a novel diindolylmethane derivative. Cells. 2019;8(3):220.

haematologica | 2021; 106(2)

ARTICLE

Chronic Myeloid Leukemia

Targeting of plasminogen activator inhibitor-1 activity promotes elimination of chronic myeloid leukemia stem cells

Ferrata Storti Foundation

Takashi Yahata,1,2*Abd Aziz Ibrahim,1,3* Ken-ichi Hirano,1,4 Yukari Muguruma,5 Kazuhito Naka,6 Katsuto Hozumi,1,4 Douglas E. Vaughan,7 Toshio Miyata8 and Kiyoshi Ando1,3

Research Center for Regenerative Medicine, Tokai University School of Medicine, Kanagawa; Japan; 2Department of Innovative Medical Science, Tokai University School of Medicine, Kanagawa; Japan; 3Department of Hematology and Oncology, Tokai University School of Medicine, Kanagawa, Japan; 4Department of Immunology, Tokai University School of Medicine, Kanagawa; Japan; 5National Cancer Center Research Institute, Tokyo, Japan; 6Department of Stem Cell Biology, Research Institute for Radiation Biology and Medicine, Hiroshima University, Hiroshima, Japan; 7Department of Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL, USA and 8United Centers for Advanced Research and Translational Medicine, Tohoku University Graduate School of Medicine, Miyagi, Japan. 1

Haematologica 2021 Volume 106(2):483-494

*TY and AAI contributed equally as co-first authors.

ABSTRACT

herapeutic strategies that target leukemic stem cells (LSC) provide potential advantages in the treatment of chronic myeloid leukemia (CML). Here we showed that selective blockade of plasminogen activator inhibitor-1 (PAI-1) enhances the susceptibility of CML-LSC to tyrosine kinase inhibitor (TKI), which facilitates the eradication of CML-LSC and leads to sustained remission of the disease. We demonstrated for the first time that the TGF-β−PAI-1 axis was selectively augmented in CMLLSC in the bone marrow (BM), thereby protecting CML-LSC from TKI treatment. Furthermore, the combined administration of the TKI imatib plus a PAI-1 inhibitor, in a mouse model of CML, significantly enhanced the eradication of CML cells in the BM and prolonged the survival of CML mice. The combined therapy of imatinib and a PAI-1 inhibitor prevented the recurrence of CML-like disease in serially transplanted recipients, indicating the elimination of CML-LSC. Interestingly, PAI-1 inhibitor treatment augmented membrane-type matrix metalloprotease-1 (MT1-MMP)-dependent motility of CML-LSC, and the anti-CML effect of PAI-1 inhibitor was extinguished by the neutralizing antibody for MT1-MMP, underlining the mechanistic importance of MT1-MMP. Our findings provide evidence of, and a rationale for, a novel therapeutic tactic, based on the blockade of PAI1 activity, for CML patients.

Correspondence: TAKASHI YAHATA yahata@tokai-u.jp Received: June 23, 2019. Accepted: January 24, 2020. Pre-published: January 30, 2020.

Introduction Chronic myeloid leukemia (CML) was the first luekemia to be identified with a specific tumorigenic chromosomal abnormality - the Philadelphia chromosome.1 Subsequent studies identified that the translocation event, occurred between t(9;22) (q34;q11), fused the breakpoint cluster region gene (BCR) with the Abelson kinase gene (ABL1) and produced the BCR/ABL oncogene.2,3 This BCR/ABL fusion protein possesses constitutive tyrosine kinase activity resulting in the development of myeloid leukemia through aberrant differentiation of hematopoietic stem/progenitor cells (HSPC) toward the myeloid lineage. Although development of tyrosine kinase inhibitors (TKI) that target the abnormal activation of tyrosine kinase, such as imatinib, has dramatically improved the prognosis of CML,4,5 the disease often relapses, even after complete remission achieved under TKI therapy, thus remaining a central problem in the treatment of CML. It has been hypothesized that leukemic stem cells (LSC), also identified as leukemic initiating cells, are the cells that possess the unique ability to resist the cytotoxic agents and are responsible for the relapse of leukemia,6–8 highlighting the need for therapeutic strategies that specifically target this population of cells. haematologica | 2021; 106(2)

https://doi.org/10.3324/haematol.2019.230227

483

T. Yahata et al.

LSC are thought to possess properties similar to those characterizing normal HSPC, including the capacity for self-renewal, cell cycle quiescence, and resistance to traditional chemotherapy.6,9 Studies have shown that transforming growth factor-β (TGF-β) signaling plays supportive roles in normal hematopoiesis and leukemogenesis. Yamazaki et al.10,11 reported that TGF-β signaling is essential for the maintenance of HSPC within the bone marrow (BM) microenvironmental niche and promotes Smaddependent HSPC hibernation. In the same vein, Naka et al.12,13 showed that TGF-β signaling is also essential for the maintenance of CML-LSC. Thus, suppression of TGF-β signaling deserves investigation for the purpose of eliminating CML-LSC. Like their normal hematopoietic counterparts, CMLLSC are presumed to reside in specific niches in the BM microenvironment, which likely contribute to relapse after chemotherapy.14–16 We have recently uncovered a pathway by which TGF-β signaling regulates HSPC retention in the niche.17 TGF-β induces the expression of plasminogen activator inhibitor-1 (PAI-1), a major physiologic serine protease inhibitor (serpin) of the fibrinolytic system. In HSPC, PAI-1, known conventionally as an extracellular fibrinolysis-related serpin, functions intracellularly (subsequently referred to as iPAI-1) and inhibits the proteolytic activity of proprotein convertase, furin, which consequently diminishes membrane type-1 metalloprotease (MT1-MMP) activity.17 MT1-MMP promotes HSPC motility by breaking down pericellular matrix proteins and adhesion molecules necessary for anchoring HSPC to the niche.18,19 Genetic or pharmacological inhibition of TGF-β-iPAI-1 signaling increases MT1-MMP-dependent cellular motility, causing detachment of HSPC from the niche, making them receptive to extraneous stimuli.17,20 In line with our report, a number of studies have demonstrated that both TGF-β and PAI-1 play important roles in numerous physiological and pathological conditions, including wound healing, obesity, cardiovascular disease and cancer.21–23 Therefore, we hypothesize that blockade of iPAI-1 activity facilitates the dislocation of CML-LSC in the BM, which in turn renders CML-LSC susceptible to TKI therapy, resulting in the eradication of CML-LSC and sustained disease remission.

Methods

plantation-based CML model (BCR/ABL1-CML mice) as previously described.12,25 Briefly, normal LSK cells (4–5×104 cells per recipient mouse) isolated from mononuclear cells of fetal liver (embryonic day E14.5) or adult BM (8-12 weeks old) were transduced with the human BCR/ABL1-ires green fluorescent protein (GFP) retrovirus and transplanted into irradiated (9 Gy) recipient C57BL/6J mice purchased from CLEA Japan (Tokyo). Another set of experiments, gene-modified 32D cell lines were also transplanted into non-irradiated C3H/HeNCrl mice (CLEA Japan). CMLlike disease developed in these recipient mice by 12–20 days post transplantation. A tetracycline (tet)-inducible CML mouse model26,27 was also used in this study. Tal1-tTA mice (JAX, #006209) and TREBCR/ABL1 transgenic mice (JAX, #006202), both on an FVB/N genetic background, were purchased from the Jackson Laboratory. Tal1-tTA and TRE-BCR/ABL1 transgenic mice were interbred to generate Tal1-tTAxTRE-BCR/ABL1 double-transgenic mice. These animals were maintained in cages supplied with drinking water containing 20 mg/L doxycycline (Sigma-Aldrich). At 8-10 weeks after birth, BM cells of Scl/Tal1-tTAxTRE-BCR/ABL double-transgenic (Ly5.2; CD45.2) mice were mixed with competitor BM cells obtained from congenic (Ly5.1; CD45.1) mice and then were transplanted to lethally irradiated (9 Gy) wild-type (WT) hosts. The recipients bred in the presence of doxycycline, and BCR/ABL expression was induced by doxycycline withdrawal 2 months after transplantation. All induced recipient mice progressively developed CML-like disease associated with a severe myeloid cell expansion in the BM, spleen and peripheral blood (PB), and splenomegaly. The mice became moribund ~3 weeks after induction. In order to examine the in vivo effects of the combined administration of imatinib (IM) plus PAI-1 inhibitors, BCR/ABL1-CMLaffected mice received vehicle alone, or IM (Gleevec; 150-400 mg/kg/day; Novartis) and/or PAI-1 inhibitor (TM5257, 100 mg/kg/day; TM5009 or TM5614, 10 mg/kg/day) in vehicle, and, at the same time, mice were intraperitoneally injected with 1 mg/kg anti-MT1-MMP antibody (Merck Millipore) or nonimmune species- and isotype-matched control antibody for 5 consecutive days. Treatment was delivered by oral gavage on days 8–30 post transplantation. Three to 4 months after transplantation, PB, spleen, and BM samples were collected from individual recipients and subjected to analyses and secondary transfer experiments. Protocols concerning animal experiments and recombinant gene experiments were approved by the Animal Care Committee and the Gene Recombination Experiment Safety Committee of Tokai University.

Additional methods are presented in the Online Supplementary Methods, available on the Haematologica web site.

Results Pharmacological properties of inhibitors Three different PAI-1 inhibitory compounds (TM5275, TM5509 and TM5614) were used.24 All three selectively and effectively inhibited PAI-1 activity with a half-maximal inhibition (IC50) value <6.95 mM in a tPA-dependent hydrolysis assay. The PAI-1 inhibitors (up to 100 mM) did not interfere with other serpin/serine protease systems such as 1-antitrypsin/trypsin and α2-antiplasmin/plasmin. The TGF-β inhibitor, LY364947, is a potent and selective inhibitor of the TGF- β receptor I, inhibiting phosphorylation of Smad3 by TGF-β receptor I kinase (IC50=59 nM).

CML mouse model Several different mouse models of CML-like disease were utilized in this study. First, we used a BCR/ABL1 transduction/trans484

TGF-β-iPAI-1 axis is activated in CML-LSC LSC are analogous to HSPC in so many aspects; both cell types are characteristically slow cell-cycling and have a high tendency to localize in particular sites, so called niche dependency.14,15,28–30 In addition, the surface markers of CML-LSC and normal HSPC are shown to be almost identical both in murine models and in human samples. For example, transplantation studies of BCR/ABL-induced murine CML models demonstrated that BCR/ABLexpressing LSC activity is confined to the fraction of Lineage (Lin)–Sca1+c-Kit+ (LSK) cells that contains murine HSPC population.12,31 We have recently shown that TGF-β signaling regulates the motility of HSPC in the niche through selective upregulation of iPAI-1 expression in haematologica | 2021; 106(2)

PAI-1 blockade eliminates leukemia stem cells

HSPC.17,20 Therefore, we hypothesized that the TGF-β−iPAI-1 axis is similarly operational in CML-LSC. In order to investigate the activation of TGF-β−signaling in CML cells in vivo, we generated a CML-like myeloproliferative disease mouse model. Normal immature LSK cells obtained from fetal liver (FL) or adult BM were transduced with retrovirus carrying a human BCR/ABL-ires-GFP vector and transplanted into irradiated (9 Gy) recipient mice (Figure 1A). Consistent with previous reports,12,13 we found that BCR/ABL-transduced LSK cells efficiently induced CML-like disease in recipient mice by 12-20-day post transplantation (Figure 1B). In line with our previous reports, the highest levels of TGF-β signal activation was confined to LSK cells, the majority of which are considered to be CML-LSC, as demonstrated by flow cytometric analysis for the phosphorylation of Smad-3 protein, a downstream signaling molecule of TGF-β which forms a complex with Smad4 in the PAI-1 promoter and enhances transcriptional activation of PAI-1 gene (Figure 1C). A gradual increase in TGF β activity in the CML-LSC fraction correlated with an increase in iPAI-1 expression within the same fraction (Figure 1C). The higher expression of iPAI-1 in CML-LSC was abolished by administration of TGF-β inhibitor (LY364947, 10 mg/kg) (Figure 1D), indicating that the TGF-β−iPAI-1 axis is indeed activated in CML-LSC.

iPAI-1 protects CML cells from TKI treatment HSPC are known to be remarkably resilient (in part) because they are kept in cell cycle dormancy through activation of TGF-β signaling where iPAI-1 plays a pivotal role in retaining HSPC in their protective niche.17,20 Since we found that the TGF-β−iPAI-1 axis is similarly activated in CML-LSC (Figure 1), we hypothesized that functional activation of the TGF-β−iPAI-1 axis contributes to TKI resistance of CML-LSC by facilitating the supportive interaction within the niche. In order to test this hypothesis, we systemically investigated the influence of iPAI-1 on sensitivity of CML cells to TKI by using established CML-like cell lines (32D cells transduced with retrovirus carrying the human BCR/ABL-ires-GFP vector) that are genetically modified to enhance (PAI-1 overexpression [OE]) or nullify (PAI-1 knockout [KD]) the expression of iPAI-1 (Figure 2A). Twenty hours after being treated with IM in vitro, where extracellular fibrinolytic factors, such as tissue plasminogen activator, a PAI-1 target, did not exist, the apoptotic response of these cells was assessed by annexin V/PI staining. We found that a significantly higher percentage of PAI-1 OE cells survived IM treatment than parental 32D cells while a majority of PAI-1 KD cells underwent apoptosis (Figure 2B). This in vitro experiment clearly ruled out the involvement of extracellular anti-fibrinolytic function of PAI-1 in TKI resistance and suggested direct involvement of iPAI-1 in TKI resistance. Next, we examined the effect of iPAI-1 overexpression on the sensitivity to TKI in vivo (Figure 2C). Seven days after transplantation, the presence of BCR/ABLGFP+CML cells were confirmed in mice that received any of the genetically modified CML cells. IM was orally administered to these recipient mice for 7 consecutive days, and the percentage of GFP+CML cells in the BM was analyzed on the day after final IM administration. At day 7 of post transplantation, initial engraftment of PAI-1 KD CML cells in the BM was found to be slightly lower than that of WT CML cells. In addition, consistent with the in haematologica | 2021; 106(2)

vitro findings described above, downregulation of iPAI-1 expression resulted in a significant reduction of CML cells in the BM of TKI-treated mice (Figure 2D). On the contrary, although PAI-1 OE CML cells were significantly less successful in initial engraftment compared to WT CML cells and PAI-1 KD CML cells, PAI-1 OE CML cells were resistant to TKI treatment and overgrew in the BM. These results confirmed that the intensity of iPAI-1 expression governs the susceptibility of CML cells to TKI treatment.

Pharmacological inhibition of iPAI-1 activity increases the susceptibility of CML cells to TKI treatment The findings described above prompted us to investigate whether PAI-1 inhibitor administration can increase the susceptibility of CML cells to TKI treatment. CMLbearing mice created by transplantation of the parental BCR/ABL-ires-GFP-transduced 32D cells to non-irradiated mice were treated with either IM alone, a PAI-1 inhibitor alone or IM in combination with a PAI-1 inhibitor for 7 consecutive days and were observed for the following 40 days (Figure 3A). In this study, three different orallyactive, selective PAI-1 inhibitors, namely TM5275, TM5509 and TM5614 were used. The effect of the PAI-1 inhibitors was evaluated by assessing the percentage of BCR/ABL-GFP+CML cells in the BM and the overall survival of CML-bearing mice. Seven days after transplantation, all recipient mice developed a CML-like disease characterized by myeloid cell expansion in the BM, spleen and PB, accompanied by splenomegaly (Figure 3BC). Treatment with any one of the PAI-1 inhibitors alone did not markedly extend the survival of CML-bearing mice compared to the vehicle-treated CML mice (Figure 3D). Although treatment with IM by itself delayed disease onset, all IM-treated CML-bearing mice experienced relapse and died before the end of the observation period (Figure 3D). Interestingly, the combined treatment of IM plus any one of the PAI-1 inhibitors significantly inhibited the persistence of CML cells in the BM, reduced spleen size, and prolonged the survival of CML-bearing mice (Figure 3B-D). Furthermore, PAI-1 blockade combined with TKI was found to be effective in overcoming the TKI-resistance in CML-mice transplanted with PAI-1 OE CML cells (Figure 3E-F). In order to further evaluate the potential therapeutic benefit of PAI-1 inhibitor treatment, we examined the effects of the PAI-1 inhibitor using a CML-induction mouse model, a model that closely resembles CML-development in human. In this previously described tetracycline (tet)–inducible CML mouse model,26,27 BM cells of Scl/Tal1-tTAxTRE-BCR/ABL doubletransgenic (Ly5.2; CD45.2) mice were transplanted to lethally irradiated (9 Gy) WT hosts, and the recipients were maintained in the presence of doxycycline at least for 2 months (Figure 4A). Within 3 weeks upon the withdrawal of doxycycline and induction of BCR/ABL expression, mice progressively developed CML-like disease. At the time of BCR/ABL induction, mice were treated with either IM alone, a PAI-1 inhibitor alone, or the combination of IM and a PAI-1 inhibitor for 14 consecutive days. On day 15, the expression of BCR/ABL in CML cells in the BM was determined by quantitative real-time PCR. IM treatment alone reduced the level of BCR/ABL–expression in the BM LSK fraction of CML cells to approximately 1/4 of saline treated mice, whereas combined treatment of IM plus the PAI-1 inhibitor reduced the expression of 485

T. Yahata et al.

BCR/ABL in CML cells to barely detectable levels (Figure 4B). In addition, the overall survival of this CML-affected mice treated with IM plus the PAI-1 inhibitor was markedly extended (Figure 4C). These results are in accordance with our earlier experiments using genetically modified CML cell lines and BCR/ABL transduced LSK cells and indicate that inhibition of iPAI-1 activity increases the susceptibility of CML cells to TKI treatment.

iPAI-1 blockade induces the dislocation of CML-LSC in the BM We then sought to identify how iPAI-1 signaling influences the sensitivity of CML cells to TKI. Since the blockade of iPAI-1 causes detachment of HSPC from the niche,17,20 we hypothesized that iPAI-1 is similarly involved in the motility of CML-LSC in the BM. In order to test this hypothesis, we examined the expression of

Figure 1. TGF-β−iPAI-1 signaling is activated in chronic myeloid leukemia - leukemic stem cells. (A) Schema for experiments. (B) Representative flow cytometric profiles of contribution of BCR/ABL-GFP+ cells to the myeloid (Mac-1+/Gr-1+) peripheral blood (PB) output post transplantation. (C) Representative flow cytometric profiles and mean fluorescent intensity (MFI) (n=6) for p-Smad3 and intracellular plasminogen activator inhibitor-1 (iPAI-1) expressions in freshly isolated BCR/ABL-GFP+ immature chronic myeloid leukemia (CML) cells in the bone marrow (BM). MFI of LSK: Lin–c-kit+Sca-1+; LS–K: Lin–c-kit+Sca–1–; LS–K–: Lin–c-kit–Sca-1–. (D) Representative flow cytometric profile and MFI for iPAI-1 expression in freshly isolated BCR/ABL-GFP+ immature CML cells in the BM of vehicle- or LY364947-treated mice (n=5). Data represent means ± standard deviation. Statistical significance was determined by Mann-Whitney unpaired t-test. P<0.001, by a Kruskal-Wallis test. TGF-β: transforming growth factor-β; BCR: breakpoint cluster region; ABL: Abelson kinase; GFP: green florescent protein; SSC: side scatter.

486

haematologica | 2021; 106(2)

PAI-1 blockade eliminates leukemia stem cells

Figure 2. iPAI-1 protects chronic myeloid leukemia cells from tyrosine kinase inhibitor treatment. (A) A representative flow cytometric profile for intracellular plasminogen activator inhibitor-1 (iPAI-1) expressions in PAI-1-overexpression (PAI-1 OE), PAI-1-knockdown (PAI-1 KD), or wild-type parental 32D p210 cells (PAI-1 WT). (B) Representative flow cytometric profiles and percentage of Annexin V+/PI+ population in PAI-1 OE, PAI-1 KD, or PAI-1 WT cells before imatinib (Pre-IM) and after imatinib (Post-IM) treatment in vitro (n=6). (C) Schema for in vivo experiments. (D) Representative flow cytometric profiles and percentages of leukemic cells in the bone marrow (BM) (n=7). Data represent means Âą standard deviation. Statistical significance was determined by Mann-Whitney unpaired t-test. P<0.001, by a Kruskal-Wallis test. GFP: green fluorescent protein.

haematologica | 2021; 106(2)

487

T. Yahata et al.

Figure 3. Combined treatment with tyrosine kinase inhibitor plus PAI-1 inhibitor prolongs survival of chronic myeloid leukemia-bearing mice. (A) Schema for experiments. (B) Representative flow cytometric profiles for the percentage of chronic myeloid leukemia (CML) cells in the bone marrow (BM) at the indicated time points. (C) Analysis of spleen taken from imatinib (IM)- and IM plus plasminogen activator inhibitor-1 (PAI-1) inhibitor-treated mice (n=6). (D) Kaplan Meier survival curves of CML-bearing mice treated with saline (n=10), PAI-1 inhibitors (TM5275, TM5509, and TM5614) alone (n=10), IM alone (n=10), and IM plus PAI-1 inhibitor (n=15). (E) Percentages of leukemic cells in the BM (n=6). (F) Kaplan Meier survival curves of PAI-1 wild-type (PAI-1 WT) or PAI-1-overexpression (PAI-1 OE) CML cell-bearing mice treated with saline (n=0), IM alone (n=10), or IM plus PAI-1 inhibitor (TM5614) (n=10). Data represent means Âą standard deviation. Statistical significance was determined by a log-rank non-parametric test (D, F) or Mann-Whitney unpaired t-test (E). P<0.001, by a Kruskal-Wallis test. SSC: side scatter.

488

haematologica | 2021; 106(2)

PAI-1 blockade eliminates leukemia stem cells

MT1-MMP, which has been shown to control the motility of HSPC,17–19 in CML-LSC derived from PAI-1 inhibitor treated mice. Mice were transplanted with normal FL LSK cells that were transduced with retrovirus carrying the human BCR/ABL-ires-GFP and were treated with a PAI-1 inhibitor for 7 consecutive days starting at day 7 post transplantation (Figure 5A). As expected, treatment with a PAI-1 inhibitor reduced the expression of iPAI-1 in the LSK fraction of BCR/ABL-GFP+ CML cells, which represent CML-LSC, and at the same time, increased the expression of MT1-MMP in the same fraction of cells (Figure 5B). Furthermore, iPAI-1 blockade altered the expression levels of CD44 and VLA-4, key adhesion/de-adhesion molecules known to determine the localization of hematopoietic cells within the BM14,30,32–34 (Figure 5B). Consistent with this in vivo observation, expression of CD44 and VLA-4 were altered in iPAI-1 OE and KO CML cells, confirming the link between the expression of adhesion/de-adhesion molecules and the iPAI-1 activity (Online Supplementary Figure S1). In order to confirm that iPAI-1 regulates the motility of CML-LSC through the action of MT1-MMP, we examined the directional motility of CML-LSC toward a chemokine gradient using a trans-reconstituted basement membrane (MatrigelTM) migration assay system. Lin–c-Kit+ immature CML cells isolated from PAI-1 inhibitor treated CML mice exhibited significantly higher trans-Matrigel migration activity compared to those of saline-treated mice (Figure 5C). Importantly, addition of anti-MT1-MMP neutralizing antibody to the culture abo-lished the enhanced migration activity of immature CML cells (Figure 5C), confirming the critical role of MT1-MMP in regulating the motility of CML-LSC. We therefore examined the effect of PAI-1 blockage in the localization of CML-LCS in vivo. BM sections were stained with antibodies against hematopoi-

etic lineage markers (CD3, B220, Mac-1, Gr-1, Ter119, CD41, and CD48) and c-kit to identify CML-LSC and were simultaneously stained with antibody against TGF-β to identify niche cells (Figure 5D, the fluorescence images in lower magnification are shown in the Online Supplementary Figure S2), and the sections were evaluated to determine the positional relationships between CML-LSC and niche cells. In PAI-1 inhibitor-treated mice, BCR/ABL-GFP+Lin–ckit+ CML-LSC were frequently found further away from TGF-β-expressing niche cells (Figure 5D), in contrast to vehicle-treated mice in which CML-LSC were often in contact with, or within close proximity to, TGF- -expressing niche cells (Figure 5D). Importantly, the administration of an anti-MT1-MMP neutralizing antibody counteracted the observed detachment of CML-LSC from niche cells in PAI-1 inhibitor-treated mice (Figure 5D). These results suggest that iPAI-1 activity is an important determinant of CML-LSC’ sensitivity to TKI, whereby controlling the MT1-MMP dependent retention of CML-LCS in the BM protective environment.

Blockade of iPAI-1 activity in combination with TKI efficiently eliminates CML-LSC As shown earlier in this study that CML-LSC exhibited higher iPAI-1 expression (Figure 1), we focused on the effectiveness of iPAI-1 targeting therapy on primitive CML-LSC. Recipient mice that received BCR/ABL-GFP+ cells obtained from the BM of mice transplanted with human BCR/ABL-ires-GFP retrovirus transduced FL LSK cells were treated with IM alone or with IM plus a PAI-1 inhibitor for 14 consecutive days (Figure 6A). IM treatment alone reduced the percentage of BCR/ABL-GFP+ LSK cells in the BM approximately by half and mildly decreased the spleen size (Figure 6B-C). Notably, the com-

Figure 4. Blockade of iPAI-1 increases the sensitivity of chronic myeloid leukemia cells to tyrosine kinase inhibitor treatment. (A) Schema for experiments. (B) Quantitative real-time PCR analysis of the expression of BCR/ABL+ in bone marrow (BM) LSK cells of SCLtTAxBCR/ABL Tg mice that were treated with saline (n=10), imatinib (IM) alone (n=11), or IM plus plasminogen activator inhibitor-1 (PAI-1) inhibitor (TM5614) (n=12). (C) Kaplan Meier survival curves of SCLtTAxBCR/ABL Tg mice treated with saline (n=10), IM alone (n=10), or IM plus PAI-1 inhibitor (TM5614) (n=10). Data represent means ± standard deviation. Statistical significance was determined by Mann-Whitney unpaired t-test (B) or a log-rank non-parametric test (C). P<0.001, by a Kruskal-Wallis test. BCR: breakpoint cluster region; ABL: Abelson kinase: GFP: green fluorescent protein; LSK: Lineage (Lin)Sca1c-Kit.

haematologica | 2021; 106(2)

489

T. Yahata et al.

Figure 5. iPAI-1 blockade increases membrane type-1 metalloproteasedependent cellular motility. (A) Schema for experiments. (B) Representative flow cytometric profiles and mean fluorescent intensity (MFI) (n=6-9) for iPAI-1, MT1-MMP, CD44, and VLA-4 expressions in freshly isolated bone marrow (BM) BCR/ABL-GFP+ LSK cells of saline or PAI-1 inhibitor treated mice. (C) Percentages of migrated BCR/ABLGFP+Lin−c-kit+ immature chronic myeloid leukemia (CML) cells in trans-Matrigel migration assay (n=6 each). Data represent means ± standard deviation. Statistical significance was determined by Mann-Whitney unpaired t-test. P<0.001, by a Kruskal-Wallis test. (D) Representative pictures of the BM cavity of vehicle- or TM5614-treated mice. BM sections were stained with antitransforming growth factor-β (TGF-β) (purple), anti-c-kit (red) and anti-lineage markers (white) antibodies. Red arrowheads indicate TGF-β-expressing niches. Blue arrow heads indicate BCR/ABLGFP+Lin–c-kit+ CML cells. Bars represent 100 mm. Graph indicates percentages of TGF-β-expressing niches closely contact to immature CML cells. More than 50 in random fields on a slide were counted for two independent experiments (n=4 each). Each dot represents % of contact cells in the one field. Statistical significance was determined by Mann-Whitney unpaired t-test. P<0.001, by a Kruskal-Wallis test. BCR: breakpoint cluster region; ABL: Abelson kinase: GFP: green fluorescent protein; LSK: Lineage (Lin)Sca1c-Kit L; MT1MNP: membrane type-1 metalloprotease; Ab: antibody.

490

haematologica | 2021; 106(2)

PAI-1 blockade eliminates leukemia stem cells

bined treatment of IM plus a PAI-1 inhibitor diminished BCR/ABL-GFP+ LSK cell numbers in the BM and the size of spleen (Figure 6B-C). Next, we attempted to validate the rationale of PAI-1 inhibitor therapy for overcoming the IM resistance of CML-LSC using a serial BM transplantation assay system. Spleen and BM cells obtained from mice undergoing the

drug treatment described above were transplanted to recipient mice. As expected, mice that received cells of saline-treated donor mice developed CML and died within 20 days post transplantation, indicating the persistence of CML-LSC. Even though IM treatment reduced BCR/ABL-GFP+ LSK cell numbers in the BM of donor mice by half, it did not protect recipient mice from relapse and

Figure 6. Combined treatment with tyrosine kinase inhibitor plus PAI-1 inhibitor eradicates chronic myeloid leukemia - leukemic stem cells. (A) Schema for experiments. (B) Analysis of spleens taken from mice treated with saline, imatinib (IM) alone, or IM plus TM5614 in combination either with anti-MT1-MMP antibody or isotype-matched control immunoglobulin (n=6-7). (C) Representative flow cytometric profiles and percentage of chronic myeloid leukemia - leukemic stem cells (CMLLSC) in the bone marrow (BM) (n=9). (D) Kaplan Meier survival curves of serially transplanted mice (n=10-14). Data represent means Âą standard devaition. Statistical significance was determined by Mann-Whitney unpaired t-test (B) or a log-rank non-parametric test (D). P<0.001, by a Kruskal-Wallis test. PAI-1: plasminogen activator inhibitor-1; GFP: green fluorescent protein.

haematologica | 2021; 106(2)

491

T. Yahata et al.

all recipient mice died within 30 days. In contrast, more than 70% of mice that received cells from the donor mice that received the combined treatment survived until the end of the observation period (Figure 6D). In line with our in vitro findings, the effect of PAI-1 inhibitor on CML-LCS was completely abolished by the administration of a neutralizing antibody specific for MT1-MMP, indicating a critical role of this molecule in TKI sensitivity of CMLLSC (Figure 6B-D). These data confirmed that iPAI-1 blockade in combination with TKI effectively eliminates CML-LSC.

Discussion An effective strategy to eliminate cancer stem cells is critical in securing a complete cure of patients with cancers such as CML. The evidence that the TGF-β signaling pathway plays an essential role in the maintenance of immature CML-LSC12,13 led us to investigate the involvement of iPAI-1 in the function of CML-LSC and their susceptibility to TKI. Our data unambiguously demonstrate that the forced expression of iPAI-1 in a CML cell line endows tumor cells with resistance to TKI treatment both in vitro and in vivo, which likely explains why iPAI-1expressing immature CML-LSC are refractory to TKI treatment in the CML-bearing mice. This intimate association of TGF-β−iPAI-1 with CML has led to suggest that TGF-β−iPAI-1 axis itself contributes directly to malignant behavior, and that the inhibition of the TGF-β−iPAI-1 axis is beneficial in treating CML. A study has shown that iPAI-1 is able to directly bind to caspase 3,35 thus influencing activation of the apoptotic pathways. Indeed, our in vitro analysis demonstrated that the level of iPAI-1 expression is inversely correlated with the cellular susceptibility to TKI treatment, indicating that iPAI-1 contributes to TKI resistance by negatively regulating apoptosis in leukemic cells. Consistent with this, it has recently been reported that PAI-1 inhibitors induce apoptosis in vitro in a variety of tumor cell lines through the activation of the intrinsic apoptotic pathway,36,37 suggesting that PAI-1 inhibitors possess intrinsic anti-cancer activity. The combined administration of imatinib plus a PAI-1 inhibitor markedly improved the therapeutic outcome of TKI as evidenced by a reduction in the number of CML cells in the BM, reduced spleen size, and prolonged survival in a mouse model of CML. Furthermore, pharmacological blockade of iPAI-1 in combination with TKI effectively eliminated CML-LSC in the BM, which in turn prevented the recurrence of CML-like disease in recipients of serial transplantation experiments. HSPC are localized in the niche where local factors that keep them in cell-cycle dormancy, such as TGF-β, are abundantly present. LSC co-opt the HSPC BM niche to gain survival benefits by hijacking the molecular physiological mechanisms utilized by HSPC.15,28,29 Therefore, disruption of LSC localization has been proposed to be an effective means to eradicate LSC. Recently, we have shown that functional inhibition of iPAI-1 increased MT1MMP-dependent cellular motility and caused a detachment of HSPC from the niche. In this report, we demonstrated that the therapeutic effect of a PAI-1 inhibitor appears to depend on the activity of MT1-MMP. Although the involvement of MT1-MMP has been implicated in physiology and pathophysiology of many types of cells, 492

its exact role(s) has not been determined. One of the main functions of MT1-MMP is to break down pericellular matrix proteins. It also activates pro-MMP-2, which then activates other proteases such as MMP-9 and MMP-13. The interplay of these multiple protease controls motility of both normal and malignant cells. In addition, MT1MMP can promote cellular motility by breaking down membrane-bound adhesion molecules necessary for niche-anchoring, such as CD44, which then triggers the induction of molecules involved in hematopoietic cell trafficking, such as VLA4.17,18 CD44 and VLA-4 have been shown to play key roles in homing and engraftment of LSC.14,30,32–34 Consistent with this, the present study indicates that CML cells overexpressing iPAI-1 (where the expression of MT1-MMP is consequently suppressed17), manifest poor migration activity in the BM environment. Furthermore, analogous to our HSPC study, treatment of CML mice with a PAI-1 inhibitor increased MT1-MMP activity and altered the surface expression of CD44 and VLA-4 on CML-LSC, which resulted in enhanced motility of CML-LSC. This altered expression of adhesion/deadhesion molecules appear to disrupt the interaction between CML-LSC and their protective environment, which renders CML-LSC susceptible to TKI therapy. In fact, it has been reported that blockade of CD44 and VLA4 markedly reduced LSC engraftment and prolonged the survival of LSC-transplanted mice.14,30,33,34,38 Thus, MT1MMP activity appears to determine the therapeutic sensitivity of LSC. Taken together, these findings suggest that, besides its well-known role in cell cycle regulation, TGF-β signaling, in which iPAI-1-MT1-MMP axis plays an integral part of its coordinated circuitry, controls the motility of both normal and malignant stem cells, thereby protecting them from environmental stimuli. In line with our findings, several studies have proposed mobilization of CML-LSC from the niche as a measure to counteract TKI resistance, using granulocyte-colony stimulating factor (G-GSF)39,40 or inhibitors for C-X-C motif chemokine receptor (CXCR) 4,41,42 E-selectin,43 or TGF-β,13 further exploring the therapeutic potential of this approach. The present study also demonstrates that selective upregulation of the iPAI-1-MT1-MMP axis in CML-LSC provides a newly recognized mechanism of drug-resistance and that modulation of this signaling pathway overcomes TKI-resistance in CML mouse. The evidence we provide in this study has led us to suggest that iPAI-1 signaling contributes directly to pathology of cancers beyond the scope of this study, and that inhibition of the TGF-β−iPAI1 axis can become an effective approach to treat a wide variety of cancers. We have developed a low molecular weight synthetic inhibitor of PAI-1, TM5275 and a series of analogues with improved pharmacological and toxicological properties, including TM5509 and TM5614.44,45 A number of preclinical studies have demonstrated that this series of compounds is capable of affecting a number of metabolic, fibrotic, aging-related disorders, as well as hematopoietic regeneration.46–50 In this study, we provide the first evidence that inhibition of iPAI-1 can be of therapeutic benefit in the treatment of leukemia. iPAI-1 has almost identical roles in retaining normal HSC and LSC in the niche. In that sense, when CML patients are treated with a PAI-1 inhibitor, both HSC and LSC are supposed to be released from the niche. In a physiological setting, HSC are believed to be in a cycle of being released from and haematologica | 2021; 106(2)

PAI-1 blockade eliminates leukemia stem cells

returning to the niche.17,20 Even in the presence of a PAI-1 inhibitor plus TKI, HSC should be able to return to the niche because they are refractory to TKI. On the contrary, TKI specifically targets CML cells once they are released from the niche. In fact, the PAI-1 inhibitor used in this study, has successfully completed a phase I trial without any noticeable adverse effects, including myelosupplession, and has advanced to the phase II trial. In summary, this study describes a novel role of iPAI-1 in drug-resistance in CML and suggests that the selective therapeutic targeting of this serpin may be of value in the treatment of patients with this malignant hematopoietic disorder. Disclosures The PAI-1 inhibitor used in this study is protected as intellectual property. TY, TM, and KA are listed as inventors in the patent application. Contributions TY and AAI designed research, performed experiments, analyzed data, and co-wrote the paper; KH performed experiments;

References 1. Nowell PC, Hungerford DA. Chromosome studies on normal and leukemic human leukocytes. J Natl Cancer Inst. 1960;25:85109. 2. Rowley JD. Letter: A new consistent chromosomal abnormality in chronic myelogenous leukaemia identified by quinacrine fluorescence and Giemsa staining. Nature. 1973;243(5405):290-293. 3. Bartram CR, de Klein A, Hagemeijer A, et al. Translocation of c-ab1 oncogene correlates with the presence of a Philadelphia chromosome in chronic myelocytic leukaemia. Nature. 1983;306(5940):277-280. 4. Druker BJ, Guilhot F, O’Brien SG, et al. Fiveyear follow-up of patients receiving imatinib for chronic myeloid leukemia. N Engl J Med. 2006;355(23):2408-2417. 5. Hochhaus A, Larson RA, Guilhot F, et al. Long-term outcomes of imatinib treatment for chronic myeloid leukemia. N Engl J Med. 2017;376(10):917-927. 6. Huntly BJP, Gilliland DG. Leukaemia stem cells and the evolution of cancer-stem-cell research. Nat Rev Cancer. 2005;5(4):311321. 7. Bhatia R, Holtz M, Niu N, et al. Persistence of malignant hematopoietic progenitors in chronic myelogenous leukemia patients in complete cytogenetic remission following imatinib mesylate treatment. Blood. 2003; 101(12):4701-4707. 8. Corbin AS, Agarwal A, Loriaux M, et al. Human chronic meyloid leukemia stem cells insensitive to imatinib despite inhibition of BCR-ABL activity. J Clin Invest. 2011; 121(1):396-409. 9. Bonnet D, Dick JE. Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell. Nat Med. 1997;3(7):730-737. 10. Yamazaki S, Iwama A, Takayanagi S, Eto K, Ema H, Nakauchi H. TGF-β as a candidate bone marrow niche signal to induce hematopoietic stem cell hibernation. Blood. 2009; 113(6):1250-1256. 11. Yamazaki S, Ema H, Karlsson G, et al.

haematologica | 2021; 106(2)

KN provided study materials; MY and DEV wrote the paper; TM and KA conceived and designed the study. Funding TY, HK, and KA were supported by a grant-in-aid for Scientific Research and Strategic Research Foundation Grantaided Project for private universities from the Ministry of Education, Culture, Sports, Science and Technology, Japan (grant number: 15H04301 and 17H03072); TY was also supported by Project Research from the Tokai University School of Medicine, Takeda Science Foundation, and The Science Research Promotion Fund from The Promotion and Mutual Aid Corporation for private schools of Japan. AAI was supported by grant-in aid for young scientists from the Ministry of Education, Culture, Sports, Science and Technology, Japan (grant number: 18K14702). TY, TM, and KA were supported by the Project for Development of Innovative Research on Cancer Therapeutics (PDIRECT) and the Adaptable and Seamless Technology Transfer Program through Target-driven R&D (A-STEP), from the Japanese Agency for Medical Research and Development (grant number: 14533192, 12103262, and 14529205).

Nonmyelinating Schwann cells maintain hematopoietic stem cell hibernation in the bone marrow niche. Cell. 2011;147(5):11461158. 12. Naka K, Hoshii T, Muraguchi T, et al. TGFβ-FOXO signalling maintains leukaemia-initiating cells in chronic myeloid leukaemia. Nature. 2010;463(7281):676-680. 13. Naka K, Ishihara K, Jomen Y, et al. Novel oral transforming growth factor-β signaling inhibitor EW-7197 eradicates CML-initiating cells. Cancer Sci. 2016;107(2):140-148. 14. Krause DS, Lazarides K, von Andrian UH, Van Etten RA. Requirement for CD44 in homing and engraftment of BCR-ABLexpressing leukemic stem cells. Nat Med. 2006;12(10):1175-1180. 15. Krause DS, Fulzele K, Catic A, et al. Differential regulation of myeloid leukemias by the bone marrow microenvironment. Nat Med. 2013;19(11):1513-1517. 16. Guarnerio J, Mendez LM, Asada N, et al. A non-cell-autonomous role for Pml in the maintenance of leukemia from the niche. Nat Commun. 2018;9(1):66. 17. Yahata T, Ibrahim AA, Muguruma Y, et al. TGF-β–induced intracellular PAI-1 is responsible for retaining hematopoietic stem cells in the niche. Blood. 2017;130(21):2283-2294. 18. Vagima Y, Avigdor A, Goichberg P, et al. MT1-MMP and RECK are involved in human CD34+ progenitor cell retention, egress, and mobilization. J Clin Invest. 2009; 119(3):492-503. 19. Shirvaikar N, Marquez-Curtis LA, Shaw AR, Turner AR, Janowska-Wieczorek A. MT1MMP association with membrane lipid rafts facilitates G-CSF-induced hematopoietic stem/progenitor cell mobilization. Exp Hematol. 2010;38(9):823-835. 20. Ibrahim AA, Yahata T, Muguruma Y, Miyata T, Ando K. Blockade of plasminogen activator inhibitor-1 empties bone marrow niche sufficient for donor hematopoietic stem cell engraftment without myeloablative conditioning. Biochem Biophys Res Commun. 2019;516(2):500-505. 21. Ghosh AK, Vaughan DE. PAI-1 in tissue fibrosis. J Cell Physiol. 2012;227(2):493-507. 22. Placencio VR, DeClerck YA. Plasminogen

activator inhibitor-1 in cancer: rationale and insight for future therapeutic testing. Cancer Res. 2015;75(15):2969-2974. 23. Van De Craen B, Declerck PJ, Gils A. The biochemistry, physiology and pathological roles of PAI-1 and the requirements for PAI1 inhibition in vivo. Thromb Res. 2012; 130(4):576-585. 24. Yamaoka N, Murano K, Kodama H, et al. Identification of novel plasminogen activator inhibitor-1 inhibitors with improved oral bioavailability: structure optimization of Nacylanthranilic acid derivatives. Bioorg Med Chem Lett. 2018;28(4):809-813. 25. Daley GQ, Van Etten RA, Baltimore D. Induction of chronic myelogenous leukemia in mice by the P210bcr/abl gene of the Philadelphia chromosome. Science. 1990; 247(4944):824-830. 26. Huettner CS, Zhang P, Van Etten RA, Tenen DG. Reversibility of acute B-cell leukaemia induced by BCR–ABL1. Nat Genet. 2000;24(1):57-60. 27. Koschmieder S, Göttgens B, Zhang P, et al. Inducible chronic phase of myeloid leukemia with expansion of hematopoietic stem cells in a transgenic model of BCR-ABL leukemogenesis. Blood. 2005;105(1):324334. 28. Saito Y, Uchida N, Tanaka S, et al. Induction of cell cycle entry eliminates human leukemia stem cells in a mouse model of AML. Nat Biotechnol. 2010;28(3):275-280. 29. Medyouf H. The microenvironment in human myeloid malignancies: emerging concepts and therapeutic implications. Blood. 2017;129(12):1617-1626. 30. Jin L, Hope KJ, Zhai Q, Smadja-Joffe F, Dick JE. Targeting of CD44 eradicates human acute myeloid leukemic stem cells. Nat Med. 2006;12(10):1167-1174. 31. Neering SJ, Bushnell T, Sozer S, et al. Leukemia stem cells in a genetically defined murine model of blast-crisis CML. Blood. 2007;110(7):2578-2585. 32. Lapidot T, Petit I. Current understanding of stem cell mobilization. Exp Hematol. 2002; 30(9):973-981. 33. Lapid K, Glait-Santar C, Gur-Cohen S, Canaani J, Kollet O, Lapidot T. Egress and

493

T. Yahata et al. mobilization of hematopoietic stem and progenitor cells: a dynamic multi-facet process. Cambridge, MA: The Stem Cell Research Community, StemBook.; p1-37. 34. Krause DS, Lazarides K, Lewis JB, von Andrian UH, Van Etten RA. Selectins and their ligands are required for homing and engraftment of BCR-ABL1+ leukemic stem cells in the bone marrow niche. Blood. 2014; 123(9):1361-1371. 35. Chen Y, Kelm RJ, Budd RC, Sobel BE, Schneider DJ. Inhibition of apoptosis and caspase-3 in vascular smooth muscle cells by plasminogen activator inhibitor type-1. J Cell Biochem. 2004;92(1):178-188. 36. Placencio VR, Ichimura A, Miyata T, DeClerck YA. Small molecule inhibitors of plasminogen activator inhibitor-1 elicit antitumorigenic and anti-angiogenic activity. PLoS One. 2015;10(7):e0133786. 37. Mashiko S, Kitatani K, Toyoshima M, et al. Inhibition of plasminogen activator inhibitor-1 is a potential therapeutic strategy in ovarian cancer. Cancer Biol Ther. 2015; 16(2):253-260. 38. Lapidot T, Petit I. Current understanding of stem cell mobilization: the roles of chemokines, proteolytic enzymes, adhesion molecules, cytokines, and stromal cells. Exp Hematol. 2002;30(9):973-981. 39. Borthakur G, Kantarjian H, Wang X, et al.

494

Treatment of core-binding-factor in acute myelogenous leukemia with fludarabine, cytarabine, and granulocyte colony-stimulating factor results in improved event-free survival. Cancer. 2008;113(11):3181-3185. 40. Holtz M, Forman SJ, Bhatia R. Growth factor stimulation reduces residual quiescent chronic myelogenous leukemia progenitors remaining after imatinib treatment. Cancer Res. 2007;67(3):1113-1120. 41. Uy GL, Rettig MP, Motabi IH, et al. A phase 1/2 study of chemosensitization with the CXCR4 antagonist plerixafor in relapsed or refractory acute myeloid leukemia. Blood. 2012;119(17):3917-3924. 42. Beider K, Darash-Yahana M, Blaier O, et al. Combination of imatinib with CXCR4 antagonist BKT140 overcomes the protective effect of stroma and targets CML in vitro and in vivo. Mol Cancer Ther. 2014; 13(5):1155-1169. 43. Godavarthy PS, Kumar R, Herkt SC, et al. The vascular bone marrow niche influences outcome in chronic myeloid leukemia via the E-selectin - SCL/TAL1 - CD44 axis. Haematologica. 2020;105(1):136-147. 44. Izuhara Y, Takahashi S, Nangaku M, et al. Inhibition of plasminogen activator inhibitor-1: its mechanism and effectiveness on coagulation and fibrosis. Arterioscler Thromb Vasc Biol. 2008;28(4):672-677.

45. Izuhara Y, Yamaoka N, Kodama H, et al. A novel inhibitor of plasminogen activator inhibitor-1 provides antithrombotic benefits devoid of bleeding effect in nonhuman primates. J Cereb Blood Flow Metab. 2010; 30(5):904-912. 46. Eren M, Boe AE, Murphy SB, et al. PAI-1-regulated extracellular proteolysis governs senescence and survival in Klotho mice. Proc Natl Acad Sci U S A. 2014;111(19):7090-7095. 47. Boe AE, Eren M, Murphy SB, et al. Plasminogen activator inhibitor-1 antagonist TM5441 attenuates NĎ&#x2030;-nitro-L-arginine methyl ester-induced hypertension and vascular senescence. Circulation. 2013; 128(21): 2318-2324. 48. Pelisch N, Dan T, Ichimura A, et al. Plasminogen activator inhibitor-1 antagonist TM5484 attenuates demyelination and axonal degeneration in a mice model of multiple sclerosis. PLoS One. 2015;10(4): e0124510. 49. Ibrahim AA, Yahata T, Onizuka M, et al. Inhibition of plasminogen activator inhibitor type-1 activity enhances rapid and sustainable hematopoietic regeneration. Stem Cells. 2014;32(4):946-958. 50. Eren M, Place AT, Thomas PM, Flevaris P, Miyata T, Vaughan DE. PAI-1 is a critical regulator of FGF23 homeostasis. Sci Adv. 2017;3(9):e1603259.

haematologica | 2021; 106(2)

ARTICLE

Non-Hodgkin Lymphoma

Ataxia-telangiectasia mutated interacts with Parkin and induces mitophagy independent of kinase activity. Evidence from mantle cell lymphoma Aloke Sarkar,1 Christine M. Stellrecht,1,2 Hima V. Vangapandu,1 Mary Ayres,1 Benny A. Kaipparettu,3 Jun Hyoung Park,3 Kumudha Balakrishnan,1,4 Jared K. Burks,5 Tej K. Pandita,6 Walter N. Hittelman,1 Sattva S. Neelapu4 and Varsha Gandhi1,2,5

Department of Experimental Therapeutics, The University of Texas MD Anderson Cancer Center; 2The University of Texas Graduate School of Biomedical Sciences at Houston; 3Department of Molecular and Human Genetics, Dan L. Duncan Comprehensive Cancer Center, Baylor College of Medicine; 4Department of Lymphoma/Myeloma, The University of Texas MD Anderson Cancer Center; 5Department of Leukemia, The University of Texas MD Anderson Cancer Center and 6Department of Radiation Oncology, The Houston Methodist Research Institute, Houston, TX, USA 1

Ferrata Storti Foundation

Haematologica 2021 Volume 106(2):495-512

ABSTRACT

taxia telangiectasia mutated (ATM), a critical DNA damage sensor with protein kinase activity, is frequently altered in human cancers including mantle cell lymphoma. Loss of ATM protein is linked to accumulation of nonfunctional mitochondria and defective mitophagy in both murine thymocytes and in ataxia-telangiectasia cells. However, the mechanistic role of ATM kinase in cancer cell mitophagy is unknown. Here, we provide evidence that FCCP-induced mitophagy in mantle cell lymphoma and other cancer cell lines is dependent on ATM but independent of its kinase function. While Granta-519 mantle cell lymphoma cells possess single copy kinase-dead ATM and are resistant to FCCP-induced mitophagy, both Jeko-1 and Mino cells are ATMproficient and induce mitophagy. Stable knockdown of ATM in Jeko-1 and Mino cells conferred resistance to mitophagy and was associated with reduced ATP production, oxygen consumption, and increased mitochondrial reactive oxygen species. ATM interacts with the E3 ubiquitin ligase Parkin in a kinase-independent manner. Knockdown of ATM in HeLa cells resulted in proteasomal degradation of GFP-Parkin which was rescued by the proteasome inhibitor, MG132, suggesting that the ATMParkin interaction is important for Parkin stability. Neither loss of ATM kinase activity in primary B-cell lymphomas nor inhibition of ATM kinase in mantle cell lymphoma, ataxia-telangiectasia and HeLa cell lines mitigated FCCP- or CCCP-induced mitophagy suggesting that ATM kinase activity is dispensable for mitophagy. Malignant B-cell lymphomas without detectable ATM, Parkin, Pink1, and Parkin-UbSer65 phosphorylation were resistant to mitophagy, providing the first molecular evidence of the role of ATM in mitophagy in mantle cell lymphoma and other B-cell lymphomas.

Introduction Mitochondria are indispensable for generating the solitary cellular energy currency, namely ATP, via oxidative phosphorylation and yet they exert damaging functions in altered pathophysiological scenarios including cancer.1-5 Reactive oxygen species (ROS) emanating from mitochondria can cause inevitable damage to these organellesâ&#x20AC;&#x2122; own histone-devoid circular DNA with minimum proof-reading capacity encoding 37 genes - the prerequisite of the mitochondrial DNA (mtDNA) electron transport chain. Widespread metabolic reprogramming, excessive generation haematologica | 2021; 106(2)

Correspondence: VARSHA GANDHI vgandhi@mdanderson.org Received: August 9, 2019. Accepted: January 31, 2020. Pre-published: February 6, 2020. https://doi.org/10.3324/haematol.2019.234385

495

A. Sarkar et al.

of ROS due to leakage of electrons from complexes I and III of the electron transport chain,6 and dysregulation of fundamental cellular functions describe abnormal mitochondria that are structurally and functionally different from their normal counterparts.7-9 Maintenance of mitochondrial homeostasis in a cell is dependent on strict and highly dynamic mitochondrial fusion and fission cycles10,11 guided by two opposing events: reparation of damage by fusion and removal of damage by fission. In a unique scenario, a defective mitochondrion incapable of making ATP via F -F -ATPase, instead produces excessive amounts of ROS and is forced to consume ATP to generate membrane potential (DΨ ), impeding normal metabolic function.12 Although low levels of mitochondrial damage can be repaired by complementation through the fusion process, an excessively damaged pool of mitochondria may endanger functional mitochondria during their coexistence, affecting the quality control of mitochondria.12,13 Cells have an inherent capacity to sense damaged mitochondria and selectively degrade these defective organelles by a process called mitochondrial autophagy or mitophagy. During mitophagy, Pink1 accumulates on the outer membrane of depolarized mitochondria14 and recruits the cytosolic ubiquitin ligase Parkin and phosphorylates both Parkin and ubiquitin, resulting in Parkin activation. Activated Parkin in turn ubiquitylates scores of outer mitochondrial membrane proteins of depolarized mitochondria followed by recruitment of multiple autophagy cargo adaptors, such as OPTN and NDP52. Finally all these cargos bind directly to LC3 in the autophagosome leading to degradation of the entire mitochondrion within autophagolysosomes.15 Ataxia telangiectasia mutated (ATM) is obligatory to initiate cellular responses to DNA double-strand breaks and DNA repair to preserve genomic integrity, and loss of ATM results in genetic disorders characterized by neurodegeneration, immunodeficiency, and cancer.16-20 ATM also possesses non-nuclear functions associated with its cytoplasmic localization in various cell types21,22 and loss of ATM leads to increased accumulation of ROS and aberrant mitochondria leading to abnormal mitochondrial homeostasis and may trigger cancer progression.16,23 Further, loss of ATM leads to global dysregulation of ribonucleotide reductase activity and abrogation of mitochondrial biogenesis and mtDNA content.24 Cells from patients with ataxia telangiectasia (A-T) contain a greater mitochondrial mass and are defective in mitochondrial respiration compared to wild-type (WT) fibroblasts.25 However, none of the phenotypic abnormalities commonly observed in ATM-deficient cells are explained by defects in canonical DNA damage response pathways, particularly in neurodegeneration, cancer predisposition, and premature aging. Even though ATM is frequently mutated in cancer, a comprehensive study relating ATM dysfunction in mitophagy in cancer cells and in cancer patients is lacking. Mantle cell lymphoma (MCL) is a genetically unstable and fatal B-cell non-Hodgkin lymphoma, in which deletions or inactivating mutations in the ATM gene are frequently acquired (40-75% of cases).26,27 We demonstrate the role of ATM in mitophagy using MCL cell lines and primary cells obtained directly from patients, and show that ATM controls mitophagy through interaction with Parkin in a kinase-independent manner. 1

Methods Cell lines and primary lymphoma analysis Cell lines are described in the Online Supplementary Methods. Primary B-cell lymphomas were obtained following informed consent from every patient on an Institutional Review Boardapproved protocol and were in accordance to the declaration of Helsinki.

Reagents and antibodies The reagents and antibodies used in this study are presented in Online Supplementary Table S1.

496

Ionizing radiation, intracellular staining and flow cytometry analysis Ionizing radiation (IR) or neocarzinostatin was used to induce DNA damage. Two potent uncoupling agents, CCCP or FCCP, were used to induce mitophagy. Cells were stained with appropriate dyes to determine mitophagy, total and mitochrondrial ROS (mROS) and acquired by flow cytometry (FCS) assay.

Plasmids, lentiviral infection and transfection pcDNA3.1+ Flag-His-ATM wt (#31985) and pcDNA3.1+ FlagHis-ATM kd (#31986) were obtained from Addgene. GFPParkin, GFP-LC3 and GFP vector plasmids were gifts. Cells were transfected with Lipofectamine 2000 (Invitrogen) or Fugene6 (Promega). Lentiviral non-target short hairpin (sh)RNA and Mission shATM clones and lentivirus particles were prepared according to the manufacturer’s instructions (Sigma).

Immunoprecipitation and co-immunoprecipitation experiments Standard immunoprecipitation and co-immunoprecipitation experiments were performed by transfecting plasmids in either HEK293T, WT-HeLa or A-T cells.

Measurement of nucleoside triphosphates Intracellular nucleotides were determined by high performance liquid chromatography analysis after perchloric acid extraction.

Oxygen consumption analyses Cellular oxygen consumption rate was measured by a standard Seahorse assay (Seahorse Bioscience, Billerica, MA, USA).

Cell fractionation and immunoblot analysis Cell fractionation was performed using either a NE-PER kit (Thermo Fisher) or Cell fractionation kit (Abcam; AB109719) following the manufacturers’ guidelines. Protein lysates and immunoblots were prepared and protein bands were quantified using a LI-COR Odyssey CLx Infrared Imaging System.

Quantitative reverse-transcriptase polymerase chain reaction and mitochondrial DNA analysis Total RNA was isolated using an RNeasy Mini Kit and the reverse transcriptase reaction was performed using a RevertAidTM H Minus First Strand cDNA Synthesis kit. Reverse transcriptase polymerase chain reaction (RT-PCR) was performed using the SYBR® Green PCR Master Mix. mtDNA copy number was analyzed in total DNA by quantitative PCR (qPCR) using mtDNA- and nuclear DNA-specific primers (PMC3769921) and the copy number was calculated as described before.28

haematologica | 2021; 106(2)

ATM-induced mitophagy in MCL

Confocal analysis Cells were grown in chamber slides, stained with primary antibodies and processed for standard fixation for confocal analysis using an Olympus FV1000 laser confocal microscope with a 40x 1.3-oil immersion lens. All captured images were analyzed using 3I Slidebook 5.5 (SB645.0.0.30) software.

Statistical analysis All numerical results are presented as mean ± standard error of mean. The statistical significance of differences was analyzed using a Student t-test (paired) or analysis of variance. The statistical computations were conducted with Prism (GraphPad Software, San Diego, CA, USA). Full details of the methods are available in the Online Supplementary Methods file.

Results ATM-proficient and -deficient mantle cell lymphoma cell lines display differential sensitivity to FCCPinduced mitophagy. To determine the role of ATM in mitophagy we used three MCL cell lines (Granta-519, Jeko-1 and Mino). The

ATM-deficient Granta-519 line possesses a single copy of the ATM gene harboring a point mutation within the conserved residues of the kinase domain while both Jeko-1 and Mino cells are ATM-proficient.29,30 These cell lines contain the distinguishing t(11;14)(q13;q32) translocation resulting in cyclin D1 (CCND1) overexpression.31 Cells were treated with the mitochondrial uncoupler FCCP to induce mitophagy. Induction of mitophagy was associated with loss of mitochondrial membrane potential (DΨ ) in both Jeko-1 and Mino cells, while Granta-519 retained intermediate to high DΨ and greater mitochondrial mass (Figure 1A, B). Both basal mROS and global ROS were relatively higher in Granta-519 (Figure 1C-E) compared to other MCL cell lines. Typically mitophagy is accompanied by loss of both COXIV and the mitochondrial outer membrane protein Tom20. Immunoblot analysis revealed defective FCCP-induced mitophagy in Granta-519 as COXIV and Tom20 levels were not affected by FCCP treatment while they were reduced in FCCP-treated Jeko1 and Mino cells (Figure 1F) with <10% cell death (Online Supplementary Figure S1A). Phosphorylation of ATMSer1981 or ATM targets, Kap1Ser824 and Smc1Ser966 was detected in both Jeko-1 and Mino cells but not in Granta-519 (Figure 1F). FCS analysis also confirmed IR-induced ATMSer1981 and m

Figure 1. Mitophagy disparity in ATM-deficient and -proficient mantle cell lymphoma cell lines. (A) Representative flow cytometry (FCS) profile of mantle cell lymphoma (MCL) cell lines (5x106 live cells) treated with FCCP (75 mM for 3 h), stained with TMRE (PE) or Mitotracker deep red (APC) and acquired by a FACSCalibur and analyzed by FlowJo software. (B) Relative mitochondrial mass and membrane potential (DΨ ) in dimethylsulfoxide (DMSO)-treated (control) or FCCP-treated MCL cell lines (geometric mean; n=7; mean ± standard error of mean [SEM]). *P<0.05; **P<0.01; ***P<0.001 showing significant difference from DMSO controls. (C) Representative FCS profile of MitoSOX Red-stained, untreated MCL cell lines showing basal levels of mitochondrial reactive oxygen species (mROS). (D) Relative basal mROS in untreated MCL cell lines (geometric mean; n=3; *P<0.05 significant difference between Mino and Granta-519 cells). (E) Representative FCS profile of H DCFDA-stained global ROS levels in MCL cell lines. m

haematologica | 2021; 106(2)

497

A. Sarkar et al. F

Figure 1. (Continued from the previous page) (F) Immunoblot analysis of MCL cells lines (30 mg total protein) treated with DMSO or FCCP, as in (A) or irradiation (IR) (5x106 cells; 5 Gray) showing disparities in mitophagy. Separate blots were cut into pieces and probed with the indicated antibodies. Total and phosphorylated bands were merged and shown in color for specificity. Phospho-ATM, Smc-1, Kap-1 and p53 proteins represent DNA damage sensors. Decreases in Tom20 and COXIV levels reflect mitophagy and LC3 lipidation represents global autophagy. GAPDH was probed as a loading control each time. (G) FCS analysis showing IR (5 Gray)-induced ATMSer1981 and H2AXSer139 phosphorylation in MCL cell lines. One hour after IR treatment, cells were washed and stained with PE-ATMSer1981 and FITC-H2AXSer139, washed and acquired by a FACSCalibur and analyzed by FlowJo software, as in (1A). (H) Cell fractionation immunoblot analysis of MCL cell lines (30x106 cells per treatment) showing basal and FCCP (75 mM for 3 h)-induced mitophagy from whole cells (20 mg) and mitochondrial fractions (10 mg). GAPDH was probed to distinguish whole cell and mitochondrial proteins. Nuclear contamination in the mitochondrial fraction was detected by Lamin. Separate sets of blots were probed with the indicated antibodies. Parkin, phospho-Parkin and Pink1 were probed using an electrochemoluminescence method. Total and phosphorylated bands were merged and shown in color for specificity. GAPDH was probed as a loading control each time.

ÎłH2AXSer139 phosphorylations in both Jeko-1 and Mino cells but not in Granta-519 (Figure 1G). In contrast, global autophagy induced by FCCP, as detected by LC3 lipidation, was similar in all three cell lines indicating that ATM is not required for global autophagy. Furthermore, mitophagy was not affected by the loss of p53 in Jeko-1 cells demonstrating that p53 is not required for mitophagy. Similar experiments on human A-T isogenic cell lines revealed ATM dependency of mitophagy, as reported earlier.23 Further analyses confirmed that while autophagy was induced by CCCP in either cell line, mitophagy was restricted to only WT cells. Furthermore, we showed that A-T cells possess relatively lower levels of Pink1 and Parkin proteins with a significantly higher basal mROS level compared with the levels in WT cells (Online 498

Supplementary Figure S1B-F). Cell fractionation analysis revealed the presence of ATM and phospho-ATMSer1981 in the mitochondrial fraction of both Jeko-1 and Mino cells (Figure 1H). Although the majority of ATM protein was nuclear, some ATM was also detected in the cytoplasm of these cells (Online Supplementary Figure S2A). Mitophagy was detected in both Jeko-1 and Mino cells, but not in Granta-519, as evidenced by the loss of Tom20, and was associated with both Parkin and phospho-Parkin-UBSer65 activation (Figure 1H and Online Supplementary Figure S2B). Interestingly, despite selective induction of mitophagy in both Jeko-1 and Mino cells, neither p62 recruitment nor autophagy (LC3 lipidation) was affected in ATMdeficient Granta-519 cells. Together, these data suggest that a specific defect in mitophagy, but not in global haematologica | 2021; 106(2)

ATM-induced mitophagy in MCL

E Figure 2.. Legend on following page.

haematologica | 2021; 106(2)

499

A. Sarkar et al. I

Figure 2. Defective mitophagy in mantle cell lymphoma shATM cell lines. (A) Immunoblot analysis of stable lentiviral knockdown of ATM in Jeko-1 and Mino cells showing irradiation (IR)-induced defective phospho-ATMSer1981, -Kap1Ser824, -Smc1Ser966 and -p53Ser15 expression. Separate blots (20 mg total protein) were cut into pieces and probed with the indicated antibodies. Total and phosphorylated bands were merged and are shown in color for specificity. GAPDH was probed as a loading control each time. (B) Representative flow cytometry (FCS) profile showing IR (as in Figure 1G)-induced ATMSer1981 and H2AXSer139 phosphorylation in control and shATM mantle cell lymphoma (MCL) cell lines. (C) Representative FCS profile of MitoSOX Red-stained cells showing basal levels of mitochondrial reactive oxygen species (mROS) in control and shATM MCL cell lines. (D) Relative geometric mean of mROS levels in basal and untreated (as in Figure 1F) MCL shATM clones (n=3; mean ± standard error of mean [SEM]; *P<0.05) showing significant difference from respective control shRNA. (E) Representative FCS profile (as in Figure 1A) showing mitochondrial mass and mitochondrial membrane potential (DΨ ) in control or shATM clones treated with dimethylsulfoxide (DMSO) or FCCP in MCL cell lines. (F) Immunoblot analysis of whole cell extracts (30 µg total protein) from Jeko-1 and Mino shATM clones treated with FCCP or DMSO. Separate blots were cut into pieces and probed with the indicated antibodies. GAPDH was probed as a loading control each time. (G) Tom20 densitometry analysis from triplicate experiments (Figure 2F) with data presented as mean ± SEM; *P<0.05, **P<0.01: significant differences from respective DMSO controls. (H) Relative geometric mean of mitochondrial mass in DMSOor FCCP-treated (as in Figure 1B) MCL control and shATM clones (n=4; mean ± SEM). ***P<0.001, ****P<0.0001: significant differences from respective DMSO controls. (I) Quantitative polymerase chain reaction analysis of mitochondrial DNA copy number in untreated shRNA control and shATM MCL clones showing mean ± SEM (=3). *P<0.05, **P<0.001, ****P<0.0001: significant differences from respective control shRNA. mtDNA: mitochondrial DNA. m

autophagy resulted from the loss of functional ATM in Granta-519 cells.

Stable knockdown of ATM elicits high mitochondrial DNA copy number and defective mitophagy in mantle cell lymphoma cell lines To further delineate the role of ATM in mitophagy, we stably knocked down ATM in Jeko-1 and Mino cells via lentiviral shATM transduction. Immunoblot analysis confirmed loss of ATM protein in the shATM clones compared to control shRNA and were also defective in IRinduced ATMSer1981, Kap1Ser824, Smc1Ser966 and p53Ser15 phosphorylation as well as IR-induced ATMSer1981 and γH2AXSer139 phosphorylation, as shown by FCS analysis (Figure 2A, B). These shATM clones were also found to produce relatively higher mROS (Figure 2C, D), defective in FCCPinduced mitophagy and retained higher mitochondrial VDAC1, Tom20, COXIV and mtDNA copy number as compared to control shRNA. Neither, autophagy (LC3 lipidation) nor FCCP-induced loss of DΨ was affected by ATM ablation (Figure 2E-I and Online Supplementary Figure S3A-C). Therefore, loss of ATM provokes accumulation of mROS, preservation of mitochondrial mass, and inhibition of mitophagy in human cancer cells. m

Loss of ATM is associated with less mitochondrial ATP generation Endogenous levels of ATP and UTP were significantly higher in Jeko-1 and Mino cells (Online Supplementary Figure S4A) than in Granta-519, indicating higher respiratory capacity. Further, ATM ablation in MCL cell lines resulted in significantly less ATP production (Figure 3A). 500

This decline in ATP-linked respiration in the shATM clones resulted in depletion in the ATP pool, basal oxygen consumption rate (OCR) or in their respective stressinduced spare respiratory capacities (SRC) (Figure 3B-D). The maximal ATP output of mitochondria can be determined by addition of the mitochondrial uncoupler, FCCP, which collapses the proton gradient and generates DΨ and triggers maximal OCR and substrate oxidation by complex IV. Moreover, SRC allows us to determine the ability of the cells to respond to increased energy demands following FCCP treatment. Thus, the observed decreases in both basal OCR and SRC in shATM clones were consistent with significant reductions in the intracellular ATP pool in ATM-depleted cells. Consistent with these findings, we found higher OCR and SRC in WT cells than in A-T cells (Online Supplementary Figure S4B,C). m

ATM interacts with and confers Parkin stability in a kinase-independent manner Having observed that ATM ablation impedes mitophagy prompted us to determine the role of ATM kinase in Parkin-mediated mitophagy in HeLa cells. ATM in WT HeLa cells was stably knocked down (Kd-ATM HeLa) by lentiviral shATM infection (Figure 4A). The distributions of global, nuclear, cytoplasmic, and mitochondrial ATM foci were quantified by confocal analysis using a cellular masking method (Online Supplementary Figure S5A). As expected, significantly more abundant ATM foci were observed in all three compartments in WT cells than in Kd-ATM cells (Figure 4B-D). CCCP treatment resulted in greater mitochondrial ATM-Tom20 co-localization than in control cells treated with dimethylsulfoxide. While both global and cytoplasmic ATM foci remained haematologica | 2021; 106(2)

ATM-induced mitophagy in MCL

D B

Figure 3. Decreased mitochondrial respiration in shATM mantle cell lymphoma cell lines. (A) High performance liquid chromatography analysis of basal intracellular nucleotide (NTP) levels in untreated mantle cell lymphoma (MCL) control and shATM clones showing mean ± standard error of mean (SEM) (n=3). *P<0.05, **P<0.01: significant differences from respective control shRNA. (B) Graphical representation of the mitochondrial stress test assay in untreated Mino control and shATM clones (8×104 live cells). Treatment with a mitochondrial ATP synthesis uncoupler, FCCP, resulted in the maximal oxygen consumption rate (OCR) by the respiratory chain. Furthermore, addition of rotenone and antimycin A (complex I and III inhibitors, respectively) prevented the transfer of electrons and thereby diminished OCR, indicating an overall reduction in mitochondria-linked aerobic respiration and ATP production in shATM cells. (C) Relative OCR in untreated MCL shATM clones showing mean ± SEM; (n=3). **P<0.01, ***P<0.001: significant differences from control shRNA. (D) Relative mitochondrial spare respiratory capacity in untreated MCL shATM clones showing mean ± SEM; (n=3). **P<0.01, ***P<0.001, ****P<0.0001: significant differences from control shRNA.

unchanged following CCCP treatment, the loss of nuclear ATM and significant gain in mitochondrial ATM likely reflects translocation of ATM from the nucleus to the mitochondria. Confocal analysis also showed significantly higher mass of mitochondrial nucleoids in Kd-ATM HeLa cells than in WT controls (Figure 4E,F). Since HeLa cells lack detectable Parkin expression,32 both WT and Kd-ATM HeLa cells were transfected with GFP-Parkin plasmid to generate stable HeLa GFP-Parkin isogenic cell lines proficient or deficient in ATM. Surprisingly, while transient GFP-Parkin expression was fairly equal in both WT and Kd-ATM cell lines (Figure 4G), we failed to generate stable GFP-Parkinexpressing Kd-ATM HeLa cells. However, neither stable expression of GFP-vector nor GFP-LC3 was affected, suggesting a specific defect in GFP-Parkin stability in Kd-ATM HeLa cells (Figure 4H). Cell fractionation studies following exposure to CCCP (Figure 4I) revealed the presence of ATM protein in both nuclear and mitochondrial fractions in WT but not in Kd-ATM cells. GFP-Parkin expression was detected in both cytoplasm and mitochondrial fractions, and CCCP treatment enriched the abundance of mitochondrial GFP-Parkin accumulation in WT cells, resulting in a decrease in Tom20 expression via mitophagy. In contrast, mitochondrial GFP-Parkin translocation was undetectable in Kd-ATM cells following CCCP haematologica | 2021; 106(2)

treatment. However, the cellular distribution of the endogenous autophagy adaptor protein, p62/SQSTM1, was not affected in either of these cells, suggesting that autophagy is unrelated to the loss of ATM in HeLa cells. Confocal analysis reconfirmed that CCCP significantly induced mitochondrial GFP-Tom20 double-positive foci in WT cells compared to Kd-ATM cells (Figure 4J-L). We also showed that endogenous Parkin expression is significantly lower in the majority of the early passage MCL shATM cell lines than in control shRNA-transduced cells (Figure 4M, N). These data support the notion that defective mitophagy in MCL and HeLa shATM cells is likely due to lack of Parkin stability and mitochondrial Parkin translocation. Parkin is known to autoubiquitinate at the UBL domain and undergo proteasomal degradation in the cytosol.33 while endogenous Pink1 is known to be rapidly degraded by UBR1, UBR2 and UBR4 through the “N-end rule pathway”.34 Interestingly, the proteasome inhibitor, MG132 rescued loss of GFP-Parkin stability in Kd-ATM HeLa cells (Figure 5A upper panel; Online Supplementary Figure S5B). Surprisingly, endogenous Pink1 expression was lower in Kd-ATM cells than in WT ones and MG132 prevented Pink1 degradation in both cell lines (Figure 5A lower panel). The kinetics of GFP-Parkin degradation following cycloheximide treatment (Figure 5B, C) suggested a signif501

A. Sarkar et al. Figure 4. ATM is localized inside mitochondria: evidence from HeLa cells. (A) Immunoblot analysis (30 mg total protein) of five different stable lentiviral shATM in HeLa cells. A single blot was cut into pieces and probed with the indicated antibodies. Actin was probed as a loading control. (B) Representative confocal z-plane image analysis (scale: 10 mm) showing nuclear and extra-nuclear ATM localization in WT and Kd control (dimethylsulfoxide, DMSO) or CCCP-treated (50 mM for 3 h) HeLa cells. Nuclei are stained with DAPI. (C) Arrows in the inset of Figure 4B showing the localization of ATM in the nucleus (N), cytosol (C, green dots) or mitochondria (M, co-localized with Tom20, yellow dots). A Laplacian filter was used to identify both ATM and Tom20 co-localization in the merged image. (D) There were significantly fewer cellular ATM dots in whole cells and in all cellular compartments in Kd ATM compared to WT control cells. Significantly more mitochondrial ATM (co-localized ATM-Tom20 dots) was observed in WT HeLa cells treated with CCCP. WT control (162 cells), WT-CCCP (176 cells); Kd control (112 cells) and Kd-CCCP (121 cells) were scanned. Data represent mean Âą standard error of mean (n=3 from separate passages of cell lines following GFP-Parkin transfection). *P<0.05, ****P<0.0001 significant differences from respective controls, as indicated. DAPI represents nuclear staining. (E). Representative confocal z-plane image of WT and Kd HeLa cells showing mitochondrial nucleoids. Untreated cells were stained with anti-DNA (red) antibody (scale: 10 mm). DAPI represents nuclear staining. (F) Significantly higher (****P<0.0001) mitochondrial nucleoids in untreated Kd compared to WT HeLa cells. WT (228 cells), and Kd (385 cells) were scanned from three separate passages of cell lines and plotted. (G) WT and Kd HeLa cells were transiently transfected with plasmids (3 mg of each pcDNA control or GFP-Parkin). Representative FCS profile showing GFP expression 36 h after transfection (left panel) or merged GFP subsets (right panel).

502

haematologica | 2021; 106(2)

ATM-induced mitophagy in MCL

Figure 4. (Continued from the previous page). (H) Immunoblot analysis (30 mg total protein) showing ATM and GFP expression from transient (48 h) and stable (3 weeks) transfection with GFP vector, GFP-LC3 or GFP-Parkin plasmids in WT and Kd HeLa cells. A single blot was cut into pieces and probed with the indicated antibodies. GAPDH was probed as a loading control. (I) Cell fractionation immunoblot analysis of WT and Kd HeLa cells showing cellular GFP distribution following transient transfection with GFP-Parkin (48 h). Cells (10x106) treated with DMSO or CCCP (50 mM for 3 h) were fractionated and then 30 mg total or 10 mg each of nuclear, cytoplasmic, and mitochondrial proteins were loaded and probed with the indicated antibodies. Lamin and GAPDH were probed as nuclear and cytoplasmic loading controls. (J) Representative confocal z-plane image analysis (scale: 10 mm) 48 h after GFP-Parkin transfection, showing GFP-Parkin co-localization with the mitochondrial marker Tom20 in WT and Kd HeLa cells treated with DMSO (Ctrl) or CCCP (50 mM for 3 h). Nuclei were stained with DAPI. (K) Inset of Figure 4J, with arrows showing co-localization of GFP-Tom20 (yellow dots) inside mitochondria after WT HeLa cells had been treated with CCCP. A Laplacian filter was used to identify both GFP-Parkin and Tom20 foci revealing co-localization in the merged image. (L) CCCP treatment resulted in a greater abundance of mitochondrial GFP-Parkin-Tom20 co-localization in WT (****P<0.0001) compared to Kd (*P<0.05) HeLa cells. WT control (68 cells), WT-CCCP (200 cells); Kd control (59 cells) and Kd-CCCP (133 cells) were scanned from three separate passages of cell lines and plotted. (Continued on the next page)

icantly enhanced degradation of GFP in Kd-ATM cells than in WT HeLa cells. However, loss of ATM did not affect the stability or degradation of the long half-life (HSP90) or short half-life (MCl-1)35 proteins. Real-time RTPCR analysis from identical cells did not reveal any significant change in GFP expression (Figure 5D), arguing for a haematologica | 2021; 106(2)

specific defect in GFP-Parkin protein stability in Kd-ATM HeLa cells. This observation prompted us to investigate the role of ATM kinase activity in the ATM-Parkin interaction. GFPParkin-transfected WT Hela cells were immunoprecipitated with anti-ATM antibody (Figure 5E, F; Online 503

A. Sarkar et al. N

Figure 4. (Continued from the previous page). (M) Immunoblot analysis (30 µg total protein) from mantle cell lymphoma (MCL) shATM cell lines (between passage 2-5) showing loss of endogenous Parkin (electrochemoluminescence blot). Blots were cut into pieces and probed with the indicated antibodies. GAPDH was probed as a loading control. (N) Densitometry analysis showing basal Parkin expression from three separate replicates of MCL control and shATM clones. Data represent mean ± standard error of mean (n=3 from separate passages of cell lines until passage 5), *P<0.05, **P<0.01, ****P<0.0001: significant differences from respective controls, as indicated.

Supplementary Figure S6A). We showed that neither ATM kinase inhibitor (KU60019) nor neocarzinostatin treatment influenced their interaction, confirming that the ATM-Parkin interaction is kinase-independent. Similarly, the endogenous interaction of ATM-Parkin in the Mino cell line was also independent of ATM kinase inhibition and KU60019 did not inhibit the interaction when cells were treated with this in combination with FCCP (Figure 5G, H; Online Supplementary Figure S6B). The efficacy of KU60019 was demonstrated by loss of ATMSer1981 and Kap1Ser824 phosphorylation (Figure 6D). In other experiments, A-T cells were co-transfected with either WT or kinase-dead (KD)-Flag-ATM and GFP-Parkin plasmids and immunoprecipitated with anti-Flag antibody. Analysis of the immunoprecipitates showed that both WT and KdATM interacts with GFP-Parkin (Online Supplementary Figure S6C,D). Further immunoprecipitation analysis with anti-Parkin antibody from purified cytosolic and mitochondrial fractions from Mino cells suggests that the ATM-Parkin interaction was restricted in both these fractions (Figure 5I, J). In contrast, trypsin digestion of the mitochondrial fraction, which removes proteins from the outer membrane, revealed the presence of Parkin but not ATM in the inner membrane along with TIM23,36 a specific inner membrane translocase. These data confirm that the ATM-Parkin interaction is restricted to the outer mitochondrial membrane.

ATM kinase is dispensable in mitophagy: evidence from mantle cell lymphoma, HeLa cells and primary B-cell lymphomas The observed kinase-independent interaction of ATMParkin led us to test the role of ATM kinase function in mitophagy using pharmacological ATM inhibitors or activators. The ATM kinase inhibitor, KU60019, failed to inhibit mitophagy in either MCL or WT A-T cells (Figure 504

6A-C; Online Supplementary Figure S7A, B). As expected KU60019 inhibited FCCP-induced ATMSer1981, H2AXSer139 and Kap1Ser824 phosphorylation in both MCL cell lines, but failed to rescue FCCP-induced loss of Tom20 and mitophagy (Figure 6D-F). Neither FCCP-induced Pink1 and Parkin activation, nor Parkin-UBSer65 phosphorylation was inhibited by KU60019 (Online Supplementary Figure S7C, D). Furthermore, KU60019 could not inhibit CCCPinduced mitophagy in WT HeLa cells (Figure 6G, H). Thirty-eight primary lymphomas, including MCL (n=21), marginal zone lymphoma (MZL, n=5), follicular lymphoma (FL, n=6) and diffuse large B-cell lymphoma (DLBCL, n=6), were analyzed (Online Supplementary Table S2) for their response to FCCP-induced mitophagy. B cells isolated from healthy donors (n=3) or peripheral blood mononuclear cells (n=2) served as controls. All samples were treated with IR and their ATM kinase status was determined by FCS analysis of PE-phospho-ATMSer1981 and FITC-γH2AXSer139 co-staining, while their mitophagy status was determined by FCS and immunoblot analyses (Figure 6I-K, P; Online Supplementary Figures S8A-E, S9A, D and S10A-H). Among all B-cell lymphomas screened, 13 (34%) were negative for IR-induced phospho-ATMSer1981 and γH2AXSer139 activation (IR¯) (MCL=8; FL=2; DLBCL=3). Among MCL subtypes, 13 lymphomas were positive for IR-induced phospho-ATMSer1981 and γH2AXSer139 activation (IR+) and eight were IR¯; no statistical correlation was observed between IR+ or IR¯ subtypes and their mitophagy status (Table 1; Figure 6K). Similarly, among all non-MCL subjects, no statistical correlation was observed between mitophagy, basal mROS and their ATM kinase status (Figure 6M; Online Supplementary Figure S9A). A higher abundance of mtDNA copy number was seen in all IR¯ MCL and non-MCL lymphomas (Figure 6L; Online Supplementary Figure S9B). Consistent with cell line data, FCCP-induced loss of DΨ was prevalent in all primary m

haematologica | 2021; 106(2)

ATM-induced mitophagy in MCL

haematologica | 2021; 106(2)

Figure 5. ATM interacts with Parkin and confers Parkin stability independent of kinase activity. (A) Immunoblot analysis (30 mg total protein) from wildtype (WT) and knocked down (Kd) HeLa cells transiently transfected with 3 mg of GFP-Parkin and then, 96 h later, treated with the proteasome inhibitor MG132 (10 mM) for the indicated time points. Dimethylsulfoxide (DMSO) served as a control at 0 or 12 h. Blots were cut into pieces and probed with the indicated antibodies. ATM, GFP, Parkin and GAPDH were probed using the Licor method while Pink1 (separate blot from similar extract) was probed using an electrochemoluminescence (ECL) method with cells treated with DMSO at 0 h as the control. GAPDH was probed as a loading control in each case. (B) ATM alters the stability of Parkin protein. Immunoblot analysis of WT and Kd HeLa cells transfected with GFP-Parkin and then, 48 h later, treated with cycloheximide (10 mg/mL) for the indicated times. Immunoblot analysis (30 mg total protein) from WT and Kd HeLa cells were probed with the indicated antibodies. HSP90 and Mcl-1, representing long- and short halflife proteins, were probed. Merged image showing Mcl1 and GAPDH (as a loading control). (C) Quantification of immunoblots from triplicate experiments as in (B). Statistical analysis was performed using a two-tailed unpaired t-test. Points represent the mean ± standard error of mean (SEM). *P≤0.05; **P≤0.01. (D) Quantitative reverse transcriptase polymerase chain reaction analysis of WT and Kd HeLa cells 72 h after transfection with GFP-Parkin, as in (B), showing the lack of effect of ATM knockdown on GFP mRNA expression in either cell lines (n=3; mean ± SEM; ***P<0.001: significant difference from WT control). (E) WT and Kd HeLa cells were transiently transfected with 3 µg of GFP-Parkin and then 48 h later cell lysates (500 mg) were immunoprecipitated with mouse antiATM antibody and probed with both GFP and Parkin antibodies. WT-GFP transfected cells were treated with CCCP to induce mitophagy (50 mM for 3 h), KU60019 (10 mM for 2 h) or neocarzinostatin (NCS) (40 nM for 2 h) before immunoprecipitation. Kd HeLa cells were left untreated and served as a negative control. Arrows indicate GFP or Parkin bands superimposed in a Licor image showing their specificity. Untreated WT-GFP transfected input cell extract (10 mg) was loaded to show the specific GFP band. IgG mouse served as an isotype matched mouse IgG control. (F) Input controls (5%) of the immunoprecipitation (IP) analysis from (E). Immunoblot analysis showing total and ATMSer1981phosphorylation. Blots were cut into

505

A. Sarkar et al. pieces and probed with the indicated antibodies. GAPDH served as a loading control. (G) Endogenous co-immunoprecipitation of the Mino MCL cell line (500 mg for each treatment) either with rabbit anti-Parkin antibody and probed with mouse anti-ATM (upper lanes) or with rabbit anti-ATM antibody and probed with mouse antiParkin antibody (lower lanes, both ECL blots). Cells were pretreated with FCCP (50 mM for 3 h), KU60019 (10 mM for 1 h) or NCS (40 nM for 1 h) before IP. (H) Input controls (5%) of the IP analysis from Figure 5G. Immunoblot analysis showing FCCP-induced ATMSer1981 phosphorylation or inhibition by KU60019. NCS served as a positive control for both ATMSer1981 and Kap1Ser824 phosphorylation. Total and phosphorylated bands were merged and are shown in color for specificity. The Parkin blot was probed using ECL reagents. GAPDH was used as a loading control each time. (I) IP analysis of an endogenous ATM-Parkin interaction in total, cytoplasm, mitochondria or trypsin-digested mitochondria in untreated Mino cells (10x106 cells in each group) following cell fractionation. The Parkin-ATM interaction was detected in total, cytoplasm or in the untreated mitochondria but not in isolated mitochondria treated with trypsin. Isolated cellular fractions were immunoprecipitated with rabbit-Parkin antibody and probed with mouse ATM antibody. A 10 mL input control from the total cell extract was loaded to specify the ATM band. Rabbit IgG served as a negative control. (J) Input controls (5%) of the IP analysis from (I). Immunoblot analysis showing presence of ATM in total, cytoplasm, or in undigested mitochondria but not in isolated mitochondria treated with trypsin. GAPDH and HSP90 served as positive controls for their cytoplasmic abundance. Tom20 served as an outer mitochondrial membrane translocase while TIM23 represents an inner membrane translocase in trypsin-digested mitochondria.

MCL screened regardless of their IR status (Figure 6N; Online Supplementary Figure S9D). Cytogenetic analysis revealed that 18 subjects with MCL harbored the t(11;14) translocation and this was also unrelated to mitophagy (Figure 6O; Table 1; Online Supplementary Table S2). Immunoblot analysis established a strong correlation between IR-induced ATMSer1981, Kap1Ser824 and Smc1Ser966 phosphorylation, consistent with FCS analysis among all primary MCL and non-MCL lymphomas (Figure 6P; Online Supplementary Figure S10A-H). Total and phospho ATMSer1981 protein levels were either low or undetectable in the majority of subjects with MCL compared to the levels in MCL cell lines (Jeko-1 and Mino). Both Parkin and phospho-UBSer65 Parkin levels were either low or undetectable in a subset of MCL subjects (MCL 2, 8, 15, 17, 19) (Figure 6P; Online Supplementary Figure S10A-D); IR-induced ATMSer1981, Kap1Ser824 and Smc1Ser966 phosphorylation was not detected in these lymphomas which were resistant to mitophagy. Similarly despite a lack of IR-induced ATM kinase activation, a subset of primary MCL activated phospho-Parkin-UBSer65induced mitophagy (MCL 2, 4, 5) (Figure 6P). Conversely, analysis of basal Parkin protein expression in all primary lymphomas revealed a positive trend of higher Parkin expression among IR+ lymphomas than among IR¯ lymphomas (Figure 6Q). These data further suggest that ATM kinase activity is not required per se for phospho-UBSer65 activation-induced mitophagy in cells derived from patients. Pink1 expression was either low or undetectable in a subset of MCL (MCL 2, 4, 8, 17, 19) (Figure 6P; Online Supplementary Figures S8B and S10A-D) and a few of these lymphomas (MCL 8, 17, 19) could not activate phosphoUBSer65 Parkin and were resistant to mitophagy. Interestingly, mitophagy was not activated in a subset of IR+ MCL (MCL 8, 13, 14, 19, 21) (Online Supplementary Figures S8B and S10A-D) suggesting a heterogeneous response. Inducible mitophagy was also prevalent among non-MCL lymphomas (Online Supplementary Figures S9A and S10E-H). Despite detectable ATM expression, IR could not activate ATMSer1981, Kap1Ser824 and Smc1Ser966 phosphorylation in a subset of DLBCL and FL (IR¯) while all MZL lymphomas were IR+ (Online Supplementary Figure S8C-E). Consistent with MCL, a few IR+ lymphomas (DLBCL4, DLBCL5, LBCL, SMZL1, FL1 and FL4) were resistant to mitophagy (Online Supplementary Figure S9A; Supplementary Table S3) while a few of the IR¯ lymphomas (DLBCL 1, 2 and FL 6) were mitophagy-proficient. No statistical correlation was observed between IR status, mROS or loss of DΨ in these lymphomas, while mtDNA copy number (Online Supplementary Figure S9B-D) was higher in IR¯ FL and DLBCL. Although the majority of these lymphomas expressed Pink1 and Parkin, FCCP-induced phosm

506

pho-UBSer65 Parkin activation was not detected in a few lymphomas (DLBCL 4, 5 and FL 2) rendering them resistant to mitophagy.

Discussion Mitophagy is a selective process of macro-autophagy eliminating intracellular pathogens and dysfunctional mitochondria by engulfing cargo into autophagolysomes37 thereby maintaining mitochondrial homeostasis, genomic stability, and integrity of cells with other healthy organelles. Therefore, imposing an opportunistic and timely exclusion of dysfunctional mitochondria via mitophagy would be useful to protect cellular and genome integrity. ATMSer1981autophosphorylation is considered a hallmark of ATM activation38 and is important in maintaining genomic integrity.39 Recent observations suggest that ATM plays an important role in mitophagy in both murine thymocytes as well as in A-T cells23 but the mechanism is poorly understood. Parkin, a protein linked to Parkinson disease, is activated during mitophagy.40,41 These key observations raise important questions regarding a possible link between Parkin and ATM. Parkin is a tumor-suppressor protein and is known to regulate cell cycle proteins including Cyclin D1, Cyclin E, and CDK4 in cancers,42 while ATM is frequently lost and mutated in cancer thereby underscoring the need to evaluate their roles in mitophagy. To explore the mechanism of ATM-dependent mitophagy in cancer, we selected MCL as a model system since ATM is the second most common alteration in MCL (>50%), in which ATM is frequently lost either by 11q deletion or mutation in the kinase domain and is associated with a high number of chromosomal alterations.43 MCL is an aggressive form of non-Hodgkin lymphoma, and remains largely incurable; most affected patients eventually die of relapsed/refractory disease,44 thereby arguing for a need for an alternative method of treatment. We present evidence that cancer cell mitophagy is dependent on ATM but not its kinase activity and connects Parkin in this pathway. We show that ATM is required for mitophagy but not global autophagy. Furthermore we demonstrate that ATM, but not its kinase activity, is required for ionophore-induced dissipation of DΨ , a prerequisite signal required for Pink1-Parkin-mediated mitophagy.45 Our data are in agreement with the notion that Pink1 activation leading to Parkin activation via UBSer65 phosphorylation during mitophagy is in part ATM-dependent. We provide evidence that KU60019 failed to inhibit FCCP-induced Parkin-UBSer65 phosphorylation and mitophagy in multiple cell lines. Based on these m

haematologica | 2021; 106(2)

ATM-induced mitophagy in MCL Figure 6. Legend on page 509.

haematologica | 2021; 106(2)

507

A. Sarkar et al.

Figure 6. (continued) Legend on page 509.

508

haematologica | 2021; 106(2)

ATM-induced mitophagy in MCL

Figure 6. (continued) ATM kinase activity is dispensable in mitophagy in HeLa, mantle cell lymphoma cell lines and primary B-cell lymphomas. (A) Representative flow cytometry (FCS) data showing mitophagy in mantle cell lymphoma (MCL) cell lines (5x106 live cells) treated with FCCP (75 mM for 3 h) or the ATM kinase inhibitor, KU60019 (10 mM for 1 h), either alone or in combination. Following treatments, cells were washed and stained with TMRE (PE) and Mitotracker deep red (APC) and acquired by a FACSCalibur and analyzed with FlowJo software. (B) Relative geometric mean of mitochondrial mass, as in (A), in MCL cell lines treated with FCCP, KU60019 or their combination (mean Âą standard error of mean [SEM]; n=7; *P<0.05, **P<0.01: significant difference from respective dimethylsulfoxide [DMSO] or FCCP controls). (C) Representative FCS profile of mitochondrial retention in MCL cell lines treated with FCCP, KU60019 or both in combination, as in (A). (D) FCS analysis showing inhibition of ATM phosphorylation by KU60019 in MCL cell lines, as in (A). Cells treated with neocarzinostatin (NCS) (40 nM for 1 h) served as a positive control. Following treatments, cells were stained with PE-ATMSer1981 and FITC-ÎłH2AXSer139 (as in Figure 2B) and acquired by a FACSCalibur and analyzed with FlowJo software. (E) Immunoblot analysis (30 mg total protein) of MCL cell lines treated with FCCP, KU60019 or their combination, as in (A), and probed with the indicated antibodies. NCS treatment served as a positive control for both ATMSer1981 and Kap1Ser824 activation. Total and phosphorylated bands were merged and are shown in color for specificity. Separate blots were cut into pieces and probed with the indicated antibodies. The levels of Parkin, phospho-UBSer65 Parkin and Pink1 protein expression were detected by electrochemoluminescence (ECL). GAPDH was probed as a loading control each time. (F) Densitometry analysis of mitophagy (Tom20 expression) in MCL cell lines treated with the indicated compounds. Assays with FCCP, KU60019 or both in combination (n=7 in Jeko-1 or n=5 in Mino cells) and

haematologica | 2021; 106(2)

509

A. Sarkar et al. NCS (n=5 in Jeko-1 and n=3 in Mino) were done. Data represent the mean ± SEM; *P<0.05, **P<0.01, ***P<0.001: significant differences from respective controls. (G) Immunoblot analysis (30 µg total protein) of WT HeLa cells treated with FCCP, KU60019 or their combination, as in (E), and probed with the indicated antibodies. NCS served as a positive control for both ATMSer1981 and Kap1Ser824 activation. Total and phosphorylated bands were merged and are shown in color for specificity. Separate blots were cut into pieces and probed with the indicated antibodies. The levels of Parkin, phospho-UBSer65 Parkin and Pink1 protein expression were detected by ECL. GAPDH was probed as a loading control each time. (H) Densitometry analysis of mitophagy (Tom20 expression) in WT HeLa cells treated with the indicated compounds, as in (F). Assays with FCCP, KU60019 or both in combination (n=6) and NCS (n=4) were performed. Data represent the mean ± SEM; *P<0.05: significant difference from the respective control. (I) Representative FCS analysis of irradiation (IR)-induced (5 Gray as in Figure 1G) ATMSer1981 and +H2AXSer139 phosphorylation in B cells isolated from healthy donors (upper panel) or FCCP (75 mM for 3 h)-induced mitophagy (as in Figure 1A; lower panel). (J) Representative FCS analysis showing IR¯ and IR+ primary MCL lymphomas proficient or deficient in FCCP-induced mitophagy, respectively. (K) Line graph representation of the geometric mean of FCS analysis of mitochondrial retention in samples from four healthy donors (purified B cells in 3 cases and peripheral blood mononuclear cells in 1 case) or from 21 patients with primary MCL showing their individual IR status (IR+ n=13; IR– n=8) and their response to FCCP-induced mitophagy. (L) Quantitative polymerase chain reaction (qRT-PCR) analysis (as in Figure 2I) of mitochondrial DNA (mtDNA) copy number in 20 primary MCL (IR+ n=12 and IR– n=8) or control healthy donor B cells (n=3). WT DNA and Rho0-DNA served as positive and negative controls for mtDNA qRT-PCR analysis. Inset showing mean ± SEM; *P<0.05 significant difference from IR+ lymphomas. (M) Relative geometric mean of basal mitochondrial reactive oxygen species (mROS) levels (as in Figure 1D) in healthy donor B cells (n=3) or primary MCL (IR+ n=9 and IR– n=5). (N) Line graph representation of the geometric mean of FCS analysis of mitochondrial membrane potential (DΨ ) following FCCP treatment of cells from four healthy donors (3 purified B cells in 3 cases and peripheral blood mononuclear cells in 1 case) and from 21 patients with primary MCL showing their IR status (IR+ n=13; IR– n=8) and their response to FCCP-induced loss of DΨ . *P<0.05, **P<0.01: significant differences from respective DMSO controls. (O) FCS analysis showing geometric means of mitochondrial mass following FCCP-induced mitophagy in cells from 21 patients with primary MCL and their respective t(11;14) status (positive n=18; negative n=3). (P) Immunoblot analysis showing FCCP-induced mitophagy in control B cells, MCL cell lines or six primary MCL cases (30 μg total protein), probed with the indicated antibodies. Corresponding IR- or FCCP-induced FCS analysis data are shown in the tables underneath. The levels of Parkin, phospho-UBSer65 Parkin and Pink1 protein expression were determined by an ECL method. Both actin and GAPDH were probed as loading controls. D: DMSO; F: FCCP; I: IR. (Q) Densitometry analysis of Parkin expression in purified B cells from healthy donors (n=3) or IR+ (n=22) and IR– (n=8) lymphomas (including MCL, marginal zone lymphoma, diffuse large B-cell lymphoma and follicular lymphoma). Data represent mean ± SEM. m

observations, we propose that ATM kinase activity is dispensable for Parkin activation. Moreover, we observed that mitophagy inhibition following loss of ATM promotes high mROS and mtDNA copy number, low mitochondrial respiration, and ATP generation. However, none of these phenotypes clearly explains the role of ATM kinase in mitophagy. The loss of stable GFP-Parkin levels in ATM-deficient HeLa cells suggests a novel link between ATM and mitophagy via kinase-independent, ATM-Parkin physical interactions. In normal basal condition cytosolic Parkin exists in a coiled and auto-inhibited ‘closed’ conformation and in an inactive state.46-48 Conversely Parkin ubiquitinates itself and promotes its own degradation and we demonstrated that MG132 rescued GFP-Parkin degradation in shATM HeLa cells. Consistent with this observation, we also showed ablation of ATM in MCL cell lines triggers loss of endogenous Parkin expression. These observations support the notion that ATM may play a role in conferring Parkin stability and thereby contribute to mitophagy. However, this phenomenon is in contrast to the situation in A-T cells in which Parkin is expressed in the absence of ATM, suggesting a context-specific interaction. The stability of Parkin in tumor and non-tumor cells may enforce different mechanisms. This apparent discrepancy could be due to the possibility that non-tumor A-T fibroblasts utilize an ATM-independent mechanism to stabilize Parkin. Regardless, we show that ATM kinase activity is dispensable for the ATM-Parkin interaction. Both KU60019 and kinase-dead ATM (KD-ATM) complex with Parkin (or GFPParkin) in multiple cancer cell lines. This novel ATM-kinaseindependent affinity to complex with Parkin was further supported by the observation that ATM protein but not its kinase activity is required for mitophagy. Our data are in contrast with those of spermidine-induced mitophagy through ATM kinase-dependent activation of the PINK1/Parkin pathway in A-T fibroblasts.49 Moreover, recent studies provide evidence of the presence of mitochondrial Parkin23 in untreated A-T fibroblasts suggesting that ATM may not be required for mitochondrial Parkin translocation. However, we show that FCCP triggered both PINK1 and Parkin accumulation, consistent with spermidine-induced mitophagy.49 A recent study demonstrated that KU60019 treatment alleviates senescence via restora510

tion, functional recovery and re-acidification of the lysosome/autophagy system and metabolic reprogramming.50 Therefore, ATM kinase negatively regulates terminal mitochondrial destruction via strict control of lysosomal pH, underscoring our observation. Our data also contradict the role of ATM kinase in mitophagy in multiple cancer cell lines and in primary lymphomas. Consistent with data from shATM cell lines and given that upstream Pink1 kinase activates and recruits Parkin into depolarized mitochondria, we also present evidence that ATM kinase activity is not required for either Parkin-UBSer65 phosphorylation or upstream Pink1 activation since neither inhibition of ATM kinase by KU60019 nor activation of kinase activity by neocarzinostatin affected the ATM-Parkin interaction or ParkinUBSer65 phosphorylation. Similarly, despite heterogeneity among primary B-cell lymphomas, we did not see evidence of ATM kinase dependency in mitophagy. Consistent with the cell line data, we showed that ATM is not required for loss of DΨ , and mitophagy is not controlled by the MCL-specific t(11;14) translocation in primary lymphomas. Although we do not know the 11q status of these lymphomas, a few primary MCL do not express detectable ATM in immunoblots and also have low or undetectable Parkin expression. Based on our IRinduced kinase screening assay, we predict several lymphomas may have lost ATM protein and a subset of these lymphomas lack either Pink1 or Parkin proteins. Moreover, a few ATM kinase deficient (IR¯) lymphomas also activated FCCP-induced Parkin-UBSer65 phosphorylation suggesting that ATM kinase activity is dispensable for Parkin activation. It is likely that while Parkin (PARK2) is frequently deleted in human cancers and associated with CCND1 overexpression42 may also contribute to cell proliferation in MCL and confers resistance to mitophagy. In conclusion, loss of ATM protein may not only contribute to genotoxic stress through ROS production but can also provoke adverse effects including loss of Parkin, preservation of defective mitochondria, inhibition of mitophagy and low mitochondrial respiration, all of which may contribute to refractory diseases. We propose a pathway in which ATM but not its kinase activity is associated with induction of mitophagy through an ATM-Parkin interaction in cancer cells underscoring the m

haematologica | 2021; 106(2)

ATM-induced mitophagy in MCL Table 1. Flow cytometry analysis of the response of primary mantle cell lymphoma to FCCP-induced mitophagy is independent of ATM kinase activity.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

t(11;14) status

IR status

t(11;14) t(11;14) t(11;14) t(11;14)

+ + + + + + + + + + + + +

t(11;14) t(11;14) t(11;14) t(11;14) t(11;14) t(11;14) t(11;14) t(11;14) t(11;14) t(11;14) t(11;14) t(11;14) t(11;14) t(11;14)

Mitophagy Status DMSO FCCP 352 420 290 256 252 250 152 321 244 703 346 235 280 365 84 552 575 360 622 1120 1022

358 260 200 120 200 200 155 322 230 287 280 235 350 505 50 312 620 400 632 660 963

Dy Status DMSO FCCP m

90 120 40 32 177 88 55 58 260 54 175 170 195 336 29 265 188 50 198 70 305

45 36 20 18 82 57 51 32 180 16 50 77 40 99 21 40 58 28 63 53 162

mROS

mtDNA copy number

20 17 20 28 ND ND 153 28 20 23 27 ND ND ND 14 210 60 150 125 18 55

524 580 718 1621 405 489 5970 377 297 830 640 2711 236 573 468 512 1215 1141 1151 372 ND

Live cells (5x106) were irradiated (5 Gray, as in Figure 1G). Cells were stained with PE-ATMSer1981 and FITC-γH2AXSer139 to determine their ATM kinase status (Online Supplementary Figure S8) or treated with FCCP (75 mM for 3 h) and stained with PE-TMRE to determine mitochondrial membrane potential (DΨ ), or APC-Mitotracker deep red to determine mitochondria mass and deduce their mitophagy response. Untreated cells were stained with MitoSOX Red to determine the basal level of mitochondrial reactive oxygen species. Mitochondrial DNA copy number was analyzed from total cellular DNA isolated from untreated cells by quantitative polymerase chain reaction analysis (see Methods). ND: not determined. m

first described molecular role of ATM in mitochondrial autophagy. Pharmacological targeting of mitophagy through ATM and Parkin, in combination with other drugs or stresses, may offer opportunities to control tumor progression because of the acute sensitivity of tumor cells to mitochondrial dysfunction. Disclosures No conflicts of interests to disclose. Contributions A.S. conceptualized the study, designed experiments, generated and analyzed data, and wrote the manuscript. CMS contributed and provided critical reagents related to autophagy studies and was involved in designing and analyzing the protein half-life and mRNA qPCR studies. HVV contributed in Seahorse XF96 Analyzer and in OCR studies. M.A. performed NTP assays and HPLC analysis. B.A.K. and J.P contributed in qPCR analysis of mtDNA experiments. KB and A.S. contributed in FCS assays related to cytotoxicity. JKB and WNH contributed in confocal analyses. TKP contributed critical

References 1. Brandon M, Baldi P, Wallace DC. Mitochondrial mutations in cancer. Oncogene. 2006;25(34):4647-4662. 2. Fulda S, Galluzzi L, Kroemer G. Targeting

haematologica | 2021; 106(2)

reagents and technical help related to ATM kinase function. SSN identified patients and provided primary B-cell lymphoma cells and their cytogenetic characteristics. VG supervised the research project, analyzed data, obtained funding, and reviewed the manuscript. TKP, WNH, CMS and VG contributed in writing/editing the manuscript. Acknowledgments We thank Dr. Sankar Mitra, Methodist Research Institute, Houston, TX, USA for crtically reviewing the manuscript. We also thank Dr. Richard Youle, NIH for providing us with the YFP-Parkin plasmid, Dr. Noriyuki Matsuda, Tokyo Metropolitan Institute of Medical Science, Tokyo, Japan for providing us with the GFP-Parkin construct and Rakesh Sharma, Department of Lymphoma/Myeloma, MD Anderson Cancer Center, Houston, TX, USA for processing and distributing the primary B-cell lymphomas. This work was supported by a CLL Global Research Foundation Alliance grant. A part of this work was performed in the Flow Cytometry and Cellular Imaging Facility, supported in part by the NIH through MD Anderson’s Cancer Center Support grant CA016672.

mitochondria for cancer therapy. Nat Rev Drug Discov. 2010;9(6):447-464. 3. Galluzzi L, Joza N, Tasdemir E, et al. No death without life: vital functions of apoptotic effectors. Cell Death Differ. 200815(7): 1113-1123. 4. Sabharwal SS, Schumacker PT.

Mitochondrial ROS in cancer: initiators, amplifiers or an Achilles' heel? Nature Rev Cancer. 2014;14(11):709-721. 5. Vander Heiden MG, Cantley LC, et al. Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science. 2009;324(5930):1029-1033.

511

A. Sarkar et al. 6. Van Houten B, Woshner V, Santos JH. Role of mitochondrial DNA in toxic responses to oxidative stress. DNA Repair (Amst). 2006;5(2):145-152. 7. de Moura MB, dos Santos LS, Van Houten B. Mitochondrial dysfunction in neurodegenerative diseases and cancer. Environ Mol Mutagen. 2010;51(5):391-405. 8. Diehn M, Cho RW, Lobo NA, et al. Association of reactive oxygen species levels and radioresistance in cancer stem cells. Nature. 2009;458(7239):780-783. 9. Kroemer G, Pouyssegur J. Tumor cell metabolism: cancer's Achilles' heel. Cancer Cell. 2008;13(6):472-482. 10. Liesa M, Palacin M, Zorzano A. Mitochondrial dynamics in mammalian health and disease. Physiol Rev. 2009;89(3): 799-845. 11. Westermann B. Mitochondrial fusion and fission in cell life and death. Nat Rev Mol Cell Biol. 2010;11(12):872-884. 12. Youle RJ, van der Bliek AM. Mitochondrial fission, fusion, and stress. Science. 2012;337(6098):1062-1065. 13. Narendra DP, Jin SM, Tanaka A, et al. PINK1 is selectively stabilized on impaired mitochondria to activate Parkin. PLoS Biol. 2010;8(1):e1000298. 14. Youle RJ, Narendra DP. Mechanisms of mitophagy. Nat Rev Mol Cell Biol. 2011;12 (1):9-14. 15. Pickrell AM, Youle RJ. The roles of PINK1, parkin, and mitochondrial fidelity in Parkinson's disease. Neuron. 2015;85(2): 257-273. 16. Ditch S, Paull TT. The ATM protein kinase and cellular redox signaling: beyond the DNA damage response. Trends Biochem Sci. 2012;37(1):15-22. 17. Lavin MF, Shiloh Y. Ataxia-telangiectasia: a multifaceted genetic disorder associated with defective signal transduction. Curr Opin Immunol. 1996;8(4):459-464. 18. Udayakumar D, Horikoshi N, Mishra L, et al. Detecting ATM-dependent chromatin modification in DNA damage response. Methods Mol Biol. 2015;1288:317-336. 19. Shiloh Y. ATM and related protein kinases: safeguarding genome integrity. Nature Rev Cancer. 2003;3(3):155-168. 20. You Z, Shi LZ, Zhu Q, et al. CtIP links DNA double-strand break sensing to resection. Mol Cell. 2009;36(6):954-969. 21. Barlow C, Ribaut-Barassin C, Zwingman TA, et al. ATM is a cytoplasmic protein in mouse brain required to prevent lysosomal accumulation. Proc Natl Acad Sci U S A. 2000;97(2):871-876.

512

22. Lim DS, Kirsch DG, Canman CE, et al. ATM binds to beta-adaptin in cytoplasmic vesicles. Proc Natl Acad Sci U S A. 1998;95(17): 10146-10151. 23. Valentin-Vega YA, Maclean KH, Tait-Mulder J, et al. Mitochondrial dysfunction in ataxiatelangiectasia. Blood. 2012;119(6):14901500. 24. Eaton JS, Lin ZP, Sartorelli AC, et al. Ataxiatelangiectasia mutated kinase regulates ribonucleotide reductase and mitochondrial homeostasis. J Clin Invest. 2007;117(9): 2723-2734. 25. Ambrose M, Goldstine JV, Gatti RA. Intrinsic mitochondrial dysfunction in ATM-deficient lymphoblastoid cells. Hum Mol Genet. 2007;16(18):2154-2164. 26. Schaffner C, Idler I, Stilgenbauer S, et al. Mantle cell lymphoma is characterized by inactivation of the ATM gene. Proc Natl Acad Sci U S A. 2000;97(6):2773-2778. 27. Fang NY, Greiner TC, Weisenburger DD, et al. Oligonucleotide microarrays demonstrate the highest frequency of ATM mutations in the mantle cell subtype of lymphoma. Proc Natl Acad Sci U S A. 2003;100(9):5372-5377. 28. Venegas V, Halberg MC. Measurement of mitochondrial DNA copy number. Methods Mol Biol. 2012;837:327-335. 29. Amin HM, McDonnell TJ, Medeiros LJ, et al. Characterization of 4 mantle cell lymphoma cell lines. Arch Pathol Lab Med. 2003;127(4):424-431. 30. Vorechovsky I, Luo L, Dyer MJ, et al. Clustering of missense mutations in the ataxia-telangiectasia gene in a sporadic Tcell leukaemia. Nat Genet. 1997;17(1):96-99. 31. Salaverria I, Perez-Galan P, Colomer D, et al. Mantle cell lymphoma: from pathology and molecular pathogenesis to new therapeutic perspectives. Haematologica. 2006;91(1):1116. 32. Pawlyk AC, Giasson BI, Sampathu DM, et al. Novel monoclonal antibodies demonstrate biochemical variation of brain parkin with age. J Biol Chem. 2003;278(48):48120-48128. 33. Chaugule VK, Burchell L, Barber KR, et al. Autoregulation of Parkin activity through its ubiquitin-like domain. EMBO J. 2011;30 (14):2853-2867. 34. Yamano K, Youle RJ. PINK1 is degraded through the N-end rule pathway. Autophagy. 2013;9(11):1758-1769. 35. Schwanhausser B, Busse D, Li N, et al. Global quantification of mammalian gene expression control. Nature. 2011;473(7347): 337-342. 36. Donzeau M, Kaldi K, Adam A, et al. Tim23

links the inner and outer mitochondrial membranes. Cell. 2000;101(4):401-412. 37. Lazarou M, Sliter DA, Kane LA, et al. The ubiquitin kinase PINK1 recruits autophagy receptors to induce mitophagy. Nature. 2015;524(7565):309-314. 38. Bakkenist CJ, Kastan MB. DNA damage activates ATM through intermolecular autophosphorylation and dimer dissociation. Nature. 2003;421(6922):499-506. 39. Lee JH, Paull TT. Activation and regulation of ATM kinase activity in response to DNA double-strand breaks. Oncogene. 2007;26 (56):7741-7748. 40. Hang L, Thundyil J, Lim KL. Mitochondrial dysfunction and Parkinson disease: a ParkinAMPK alliance in neuroprotection. Ann N Y Acad Sci. 2015;1350: 37-47. 41. Kazlauskaite A, Muqit MM. PINK1 and Parkin - mitochondrial interplay between phosphorylation and ubiquitylation in Parkinson's disease. FEBS J. 2015;282(2):215223. 42. Gong Y, Zack TI, Morris LG, et al. Pan-cancer genetic analysis identifies PARK2 as a master regulator of G1/S cyclins. Nat Genet. 2014;46(6):588-594. 43. Bea S, Valdes-Mas R, Navarro A, et al. Landscape of somatic mutations and clonal evolution in mantle cell lymphoma. Proc Natl Acad Sci U S A. 2013 ;110(45):1825018255. 44. Campo E, Rule S. Mantle cell lymphoma: evolving management strategies. Blood. 2015;125(1):48-55. 45. Narendra D, Tanaka A, Suen DF, et al. Parkin is recruited selectively to impaired mitochondria and promotes their autophagy. J Cell Biol. 2008;183(5):795-803. 46. Riley BE, Lougheed JC, Callaway K, et al. Structure and function of Parkin E3 ubiquitin ligase reveals aspects of RING and HECT ligases. Nat Commun. 2013;4:1982-1997. 47. Wauer T, Komander D. Structure of the human Parkin ligase domain in an autoinhibited state. EMBO J. 2013;32(15):20992112. 48. Trempe JF, Sauve V, Grenier K, et al. Structure of parkin reveals mechanisms for ubiquitin ligase activation. Science. 2013;340(6139):1451-1455. 49. Qi Y, Qiu Q, Gu X, et al. ATM mediates spermidine-induced mitophagy via PINK1 and Parkin regulation in human fibroblasts. Sci Rep. 2016;6:24700. 50. Kang HT, Park JT, Choi K, et al. Chemical screening identifies ATM as a target for alleviating senescence. Nat Chem Biol. 2017;13(6):616-623.

haematologica | 2021; 106(2)

ARTICLE

Non-Hodgkin Lymphoma

Cell free circulating tumor DNA in cerebrospinal fluid detects and monitors central nervous system involvement of B-cell lymphomas Sabela Bobillo,1* Marta Crespo,1* Laura Escudero,2* Regina Mayor,2 Priyanka Raheja,1 Cecilia Carpio,1 Carlota Rubio-Perez,2 Bárbara Tazón-Vega,1Carlos Palacio,1 Júlia Carabia,1 Isabel Jiménez,1 Juan. C. Nieto,1 Julia Montoro,1 Francisco Martínez-Ricarte,3 Josep Castellví,4 Marc Simó,5 Lluis Puigdefàbregas,1 Pau Abrisqueta,1 Francesc Bosch1# and Joan Seoane2,6#

Laboratory of Experimental Hematology, Department of Hematology, Vall d'Hebron Institute of Oncology (VHIO), Vall d’Hebron University Hospital (HUVH), Universitat Autònoma de Barcelona, Department of Medicine; 2Translational Research Program, Vall d’Hebron Institute of Oncology (VHIO), Vall d’Hebron University Hospital (HUVH), Universitat Autònoma de Barcelona; 3Department of Neurosurgery, Universitat Autònoma de Barcelona, Hospital Vall d'Hebron; 4Department of Pathology, Vall d’Hebron University Hospital, Universitat Autònoma de Barcelona; 5Department of Nuclear Medicine, Vall d’Hebron University Hospital, Universitat Autònoma de Barcelona and 6Institució Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Spain and CIBERONC

Ferrata Storti Foundation

Haematologica 2021 Volume 106(2):513-521

*SB, MC and LE contributed equally as co-first authors. FB and JS contributed equally as co-senior authors.

ABSTRACT

he levels of cell free circulating tumor DNA (ctDNA) in plasma correlate with treatment response and outcome in systemic lymphomas. Notably, in brain tumors, the levels of ctDNA in the cerebrospinal fluid (CSF) are higher than in plasma. Nevertheless, their role in central nervous system (CNS) lymphomas remains elusive. We evaluated the CSF and plasma from 19 patients: 6 restricted CNS lymphomas, 1 systemic and CNS lymphoma, and 12 systemic lymphomas. We performed whole exome sequencing or targeted sequencing to identify somatic mutations of the primary tumor, then variant-specific droplet digital polymerase chain reaction was designed for each mutation. At time of enrollment, we found ctDNA in the CSF of all patients with restricted CNS lymphoma but not in patients with systemic lymphoma without CNS involvement. Conversely, plasma ctDNA was detected in only 2 out of 6 patients with restricted CNS lymphoma with lower variant allele frequencies than CSF ctDNA. Moreover, we detected CSF ctDNA in one patient with CNS lymphoma in complete remission and in one patient with systemic lymphoma, 3 and 8 months before CNS relapse was confirmed, indicating that CSF ctDNA might detect CNS relapse earlier than conventional methods. Finally, in two cases with CNS lymphoma, CSF ctDNA was still detected after treatment even though no tumoral cells were observed by flow cytometry (FC), indicating that CSF ctDNA detected residual disease better than FC. In conclusion, CSF ctDNA can detect CNS lesions better than plasma ctDNA and FC. In addition, CSF ctDNA predicted CNS relapse in CNS and systemic lymphomas.

Introduction Central nervous system (CNS) involvement in B-cell malignancies is associated with dismal prognosis, especially in the relapse setting.1 Primary central nervous system lymphoma (PCNSL) is defined by the presence of CNS lymphoma in the absence of systemic disease.2 Conversely, secondary involvement of the CNS (SCNSL) is characterized by CNS infiltration with concomitant or previous history of systemic lymphoma.3 The diagnosis of CNS lymphoma is commonly based on crahaematologica | 2021; 106(2)

Correspondence: JOAN SEOANE jseoane@vhio.net FRANCESC BOSCH fbosch@vhio.net Received: October 22, 2019. Accepted: February 18, 2020. Pre-published: February 20, 2020. https://doi.org/10.3324/haematol.2019.241208

513

S. Bobillo et al.

nial magnetic resonance imaging (MRI) along with brain excisional biopsy or analysis of cerebrospinal fluid (CSF) or vitreous fluid.4 In some cases, biopsies can be challenging owing to the difficulty in accessing the tumor and the inherent risks associated with cranial surgery. Patients with aggressive systemic lymphomas and risk factors for CNS relapse are routinely monitored by cytology and flow cytometry (FC) analysis of CSF. Cytology is highly specific but has a very limited sensitivity, showing up to 40% false negatives. In contrast, FC is more sensitive, detecting malignant cells in up to 5-15% of patients with negative cytology.5,6 However, FC still has some limitations, and in a relevant fraction of patients, recurrence in the CNS is found shortly after a negative result by FC, suggesting that tumor cells were undetected. Several studies have demonstrated that cell free circulating tumor DNA (ctDNA) can be detected in plasma from patients with B-cell lymphomas and correlates with metabolic tumor volume and outcome.7,8 Due to the difficulties in the diagnosis of brain tumors, there is growing interest in the analysis of ctDNA in plasma and CSF and their potential utility in the management of these patients. Thus, we and others have explored the analysis of CSF in solid brain tumors as a better source of ctDNA compared with plasma.9-12 As per B-cell lymphomas, only a minority of patients with restricted CNS lymphomas present detectable ctDNA in plasma13,14 and only a few studies of ctDNA in CSF have been performed so far.15-19 In this regard, the MYD88 L265P mutation has been identified in the CSF ctDNA from some patients with PCNSL and lymphoplasmacytic lymphoma with CNS involvement.17-19 More recently, a different study using a next-generation sequencing (NGS)-based assay, detected at least one tumor-derived genetic alteration in the CSF from eight patients with relapsed CNS lymphoma and demonstrated that CSF ctDNA levels correlated with treatment response.15 Taken together, these studies demonstrate that CSF ctDNA could be detected in patients with CNS lymphomas and could be useful to monitor the disease, but a thorough analysis of its relevance on the diagnosis and monitoring of the disease is still needed. Here, we conducted a study to explore whether the analysis of ctDNA in the CSF and plasma could be useful to complement the diagnosis and molecular profile of tumors, as well as to monitor treatment response in CNS lymphomas. In addition, we evaluated whether the presence of CSF ctDNA in patients with systemic lymphoma could be more sensitive to detect CNS malignancies than CSF standard analyses and plasma ctDNA.

Methods Patients and samples Nineteen patients diagnosed with the following conditions were included: restricted CNS lymphomas, n=6; PCNSL, n=1; SCNSL, n=5; systemic lymphoma with concomitant CNS involvement, n=1; and systemic lymphoma without CNS disease but risk factors for CNS relapse, n=12. Risk factors for CNS relapse were defined as: diffuse large B-cell lymphoma (DLBCL) with testis, kidney or adrenal involvement; DLBCL with involvement of two or more extranodal sites along with high lactate dehydrogenase; high-grade B-cell lymphoma (HGBCL) with MYC and/or BCL2 rearrangements or Burkitt lymphoma. All patients were diagnosed and treated at Vall dâ&#x20AC;&#x2122;Hebron University Hospital (Barcelona, 514

Spain). Diagnosis of CNS disease was established by MRI imaging and tumor biopsy in cases with parenchymal involvement, or by FC and cytology in cases with leptomeningeal disease. A written informed consent was obtained from all individuals in accordance with the Declaration of Helsinki and the study was approved by the local clinical research ethics committee. As part of standard of care practice, in patients with systemic lymphoma and risk factors for CNS relapse, prophylactic intrathecal (IT) methotrexate (MTX) was administered every 3 weeks together with systemic chemotherapy. In addition, in cases with leptomeningeal disease, IT chemotherapy was administrated every 3-4 days until malignant cells were not detected by FC. CSF (1-2 mL) and plasma were collected before treatment in all patients and sequentially in patients who received IT chemotherapy as part of routine practice. Cytology and FC were performed in all CSF samples. Two out of 7 patients with CNS lymphoma and 3 out of 12 with systemic lymphoma received steroids before the baseline CSF sample. Evaluation of response was assessed at the end of treatment and/or on suspicion of disease progression by using MRI imaging in cases with CNS parenchymal involvement and by CSF analysis (FC and cytology) in patients with leptomeningeal disease. Positron emission tomography/computed tomography (PET/CT) was used to assess response at the end of treatment in patients with systemic lymphoma.

Sample collection and DNA extraction The median volume of plasma and CSF obtained was 2.5 mL (range: 1-6) and 1 mL (range: 0.3-2), respectively. Peripheral blood was collected in tubes containing K2EDTA (Vacutainer) and centrifuged at 1600xg for 10 minutes (min). The plasma was then transferred to another tube that was further centrifuged at 3000xg for 5 min. CSF samples were centrifuged at 3000xg for 5 min. The supernatant was collected and CSF-derived and plasma-derived circulating cell free DNA (cfDNA) was extracted using the QIAamp Circulating Nucleid Acids kit and quantified using a fluorimeter. DNA was extracted from formalin-fixed paraffinembedded tumor samples using the QIAamp DNA FFPE tissue kit. Germline DNA was extracted from peripheral blood granulocytes using the Qiagen DNeasy Blood and Tissue kit.

DNA sequencing and mutation genomic analysis We performed DNA sequencing of the tumors of 16 out of 19 patients: 11 of 16 were sequenced using a 300 gene targeted NGS panel (Vall dâ&#x20AC;&#x2122;Hebron Institute of Oncology, Genomics Facility) and in the remaining five, whole-exome DNA sequencing (WES) was performed in both the tumor and germline DNA. SureSelect Human All Exon V5 (Agilent Technologies) was used to perform whole exome enrichment for the Illumina paired-read sequencing platform HiSeq2500 with a read length of 2x100bp (Online Supplementary Appendix). The amount of DNA used for NGS panel and WES was 1 mg and 0.4-4 mg, respectively. In 3 out of 19 patients (NHL1, NHL4 and NHL5), MYD88 L265P mutation was detected in the tumor by routine genomic analysis performed at diagnosis, therefore tumor DNA sequencing was not performed.

Droplet digital polymerase chain reaction and quantification of circulating tumor-specific DNA Given the limited amount of cfDNA available, ddPCR was performed to determine the presence of ctDNA mutations in the CSF and plasma, ensuring sensitivity and precision. A set of driver mutations along with additional mutations not predicted as drivers by Cancer Genome Interpreter (CGI), but relevant according to expert knowledge, were selected for further validation by ddPCR. Custom Taqman SNP genotyping assays for ddPCR were designed to specifically detect the selected point mutations and haematologica | 2021; 106(2)

ctDNA in cerebrospinal fluid of patients with CNS lymphoma

Figure 1. Genomic landscape of the cohort of study. (A) Pie chart representing the proportion of samples of each lymphoma type (see color legend). Two sectors in each sample depict whether the patient had systemic (gray) or central nervous system (CNS) restricted disease (black). (B) Circus plot representing all mutations identified across 16 lymphoma patients. A mutation was defined as driver according to Cancer Genome Interpreter (see Online Supplementary Appendix) colored in red (see Methods section). Genes with recurrent mutations have been highlighted in bold. (C) Bubble plot representing all driver mutations and droplet digital polymerase chain reaction (ddPCR) validated mutations identified in each patient. The bubble size represents the cancer allelic fraction (CAF) (see Online Supplementary Appendix) and the border has been highlighted in black if it has been found by ddPCR, in either the cerebrospinal fluid or plasma of each patient. NHL: non-Hodgkin lymphoma; DLBCL: diffuse large B-cell lymphoma; HGBCL: high-grade B-cell lymphoma; WM: Waldestrรถm macroglobulinemia; MCL: mantle cell lymphoma; BL: Burkitt lymphoma.

the corresponding wild-type alleles. Genomic DNA from tumor tissues (10 ng), germline DNA from peripheral blood granulocytes (10 ng), plasma DNA, and CSF DNA (1-5 ng) were used for ddPCR analysis using the QX200 Droplet Digital PCR system according to manufacturer's protocols and the literature.20

Results Circulating tumor DNA in the cerebrospinal fluid of patients with central nervous system restricted B-cell lymphoma is more abundant than in plasma Here, we sought to analyze the presence of ctDNA in the CSF and plasma from 19 patients with B-cell lymphomas with and without CNS disease at time of enrollment (Table 1). Six patients exhibited CNS restricted disease, one systemic and CNS disease, and 12 systemic disease with no haematologica | 2021; 106(2)

CNS involvement. CSF was obtained at the same time as plasma in all patients. WES or targeted sequencing of the tumors was performed to identify somatic mutations (Figure 1). Variant-specific ddPCR detecting 1-5 variants per patient was performed to detect ctDNA. To demonstrate that the variant-specific ddPCR was highly specific, we performed MYD88 L265P ddPCR in ten CSF samples obtained from patients with hydrocephalus without a brain tumor (n=6), and from patients diagnosed with glioma (n=3) and a cavernoma (n=1). No mutant allele was found in any of the cases. Analysis of the ctDNA in the CSF and plasma of the six CNS restricted lymphomas showed detectable ctDNA at high variant allele frequency (VAF) (ranging from 1% to 95%) in all cases. In contrast, ctDNA in plasma was only detected in 2 out of 6 cases at very low VAF (always <5%) (Table 1), highlighting that in CNS restricted lymphomas, 515

S. Bobillo et al. Table 1. Patients’ characteristics and tumor mutations at the time of enrolment and outcome.

Age

Histol

Disease status

CNS lymphoma NHL1 59 DLBCL PCNSL New dx NHL2 60 DLBCL SCNSL Relapse

Systemic CNS CNS Tumor inv inv (FC/C) seq No No

P P, LM

-/+/-

WES

Tumor mutations

VAF tumor

VAF CSF

VAF plasma

Outcome

MYD88 L265P CD79B Y197D MYD88 L265P FOXO1 D69A TMSB4A Q24P B2M M1T BCOR N145S CREBBP R1664C MYD88 L265P MYD88 L265P

17% 22% 13% 54% 41% 68% 21% 86%

48% 9% 25% 20% N 52% N 33% 29%

N N N N N 5%

CR CR

NHL3

DLBCL SCNSL Relapse

P, LM

+/+

Panel

NHL4 NHL5

75 58

WM SCNSL Relapse DLBCL SCNSL Relapse

No No

LM P, LM

+/+ +/+

NHL6

+/+

Panel

ATM H2872L MEF2B K23R

31% 90%

34% 95%

2% 4%

PD, died

+/+

Panel

BRAF N581S

44%

PD, died

LN, bone, intestine

-/-

WES

-/-

WES

-/-

Panel

-/-

Panel

-/-

Panel

28% 32% 33% 50% 29% 38% 98% 60% 34% 32% 25% 31% 26% 29% 27% 27% 22% 31% 63% 34%

N N N N N N N N N N N N N N N N N N N N

16% 20% 18% 31% 16% 12% 15% 19% 12% 12% 27% 0.9% N 5% 3% 5% N N N N

LN, intestine LN, kidney LN, lung, ileum LN, breast

MYC Y27C ARID1A E1104 MEF2B D83G MLL2 C533fs CCND3 P284fs ID3 Y48* DDX3X Q360fs MLL2 splicing MLL2 P3139A TP53 R273C CREBBP Y145C MCL1 L186F PTNPN6 T80fs MYD88 L265P PIM1 L2F MEF2B E77K HSD3B2 L108W MYD88 L265P BCOR S1517fs EZH2 P557L MYD88 V217F ASXL2 K927R CREBBP S1436I

28% 42% 50%

N N N

N CR, CNS relapse N 16% CR, CNS relapse

ID3 L54V KRAS G12V TCF3 D561E ABL1 G321R ARID1A del

37% 34% 42% 43% 9%

N N N N N

N 1% N 0.5% 1%

MCL

SCNSL Relapse

CNS and systemic lymphoma NHL7 72 MCL SCNSL Relapse Systemic lymphoma NHL8 50 HGBCL New dx

NHL9

HGBCL

New dx

NHL10

DLBCL

New dx

NHL11

DLBCL

New dx

NHL12

DLBCL

New dx

NHL13

DLBCL

New dx

LN, testes

-/-

WES

NHL14

DLBCL

New dx

Teste

-/-

Panel

NHL15

DLBCL

New dx

LN, testes, No liver, adrenal

-/-

Panel

NHL16†

DLBCL

New dx

-/-

Panel

NHL17

HGBCL

New dx

-/-

Panel

NHL18†

New dx

LN, intestine

-/-

WES

NHL19†

New dx

LN, bone

-/-

Panel

LN, sinus, bones LN

PD, died

CR CR, CNS relapse

CR CR CR CR CR

CR CR

BL: Burkitt lymphoma; C: cytology; CNS: central nervous system; CR: complete response; CSF: cerebrospinal fluid; DLBCL: diffuse large B-cell lymphoma; Dx: diagnosis; FC: flow cytometry; HGBCL: high-grade B-cell lymphoma; Histol: histology; ID: patient identification number; inv: involvement; LM: leptomeningeal; LN: lymph nodes; MCL: mantle cell lymphoma; N: not detectable; Neg: negative; NHL: non-Hodgkin lymphoma; P: parenchyma; PCNSL: primary central nervous system lymphoma; PD: progressive disease; Pos: positive; SCNSL: secondary central nervous system lymphoma; seq: sequencing; VAF: variant allele frequency; WES: whole exome sequencing; WM: Waldeström macroglobulinemia. † Plasma samples were collected after administration of systemic chemotherapy.

516

haematologica | 2021; 106(2)

ctDNA in cerebrospinal fluid of patients with CNS lymphoma

Figure 2. Levels of circulating tumor DNA (ctDNA) correlate with disease progression and response to treatment in Bcell lymphomas. (A-F) Longitudinal monitoring of two patients with central nervous system (CNS) lymphomas (NHL2 and NHL4) and (G-L) six patients with systemic lymphomas with no evidence of CNS disease (NHL8, NHL9, NHL10, NHL13, NHL18, and NHL19) through (A) cerebrospinal fluid (CSF) and (D, G and J) plasma ctDNA analysis (variant allele frequency [VAF] values for each mutation detected using droplet digital polymerase chain reaction [ddPCR]) and its correlation with (B) CSF flow cytometry (cells/ÂľL), (C) cytology (+/- presence/absence of cells), and (F, I and L) radiological imaging (magnetic resonance imaging [MRI] and positron emission tomography/computed tomography scans). Data points are vertically aligned (E, H and K). Clinical information for each patient includes type of treatment and radiological response assessment. MRI scans for patient NHL4 are not shown given that the patient only presented leptomeningeal involvement. Dx: time of diagnosis; IT MTX: intrathecal methotrexate; R-CHOP: combined therapy (rituximab, cyclophosphamide, doxorubicin hydrochloride, vincristine and prednisolone); Burk: Burkimab regimen; CR: complete response; EOT: end of treatment; m: month; w: week.

H I

K L

haematologica | 2021; 106(2)

517

S. Bobillo et al. A

C D

Figure 3. Cerebrospinal fluid (CSF) circulating tumor DNA (ctDNA) is more sensitive than flow cytometry (FC) to detect central nervous system (CNS) relapse and residual disease in CNS-restricted lymphomas. Longitudinal monitoring of three patients with CNS-restricted lymphomas (NHL3, NHL5 and NHL6) through (A) CSF and (D) plasma ctDNA analysis (variant allele frequency [VAF] values for each mutation obtained by droplet digital polymerase chain reaction [ddPCR]) and its correlation with (B) CSF flow cytometry (cells/mL), (C) cytology (+/- presence/absence of cells). (E) Type of treatment and response assessment, (F) radiological imaging (MRI). Data points are vertically aligned. MRI scans for patient NHL6 and NHL5 are not shown given that the patient only presented leptomeningeal involvement at diagnosis. Dx: time of diagnosis; IT MTX: intrathecal methotrexate; BAM-R: combined therapy (rituximab, carmustine, cytarabine and methotrexate); ASCT: autologous stem cell transplant; CR: complete response; LM: leptomeningeal; EOT: end of treatment; d: day; m: month; NA: not available. *CSF sample was not available for ctDNA analyses.

ctDNA is better identified in the CSF than in plasma. Interestingly, these two cases with restricted CNS lymphoma in which plasma ctDNA was detected had a previous history of systemic disease. This finding might reveal a systemic nature of some restricted SCNSL and deserves further investigation in larger studies. We then analyzed 12 patients with systemic lymphoma without CNS involvement but with high risk for CNS relapse. In these cases, lumbar puncture was performed to rule out CNS infiltration and to administer IT MTX. As opposed to cases with CNS involvement, no ctDNA was detected in the CSF while it was detected at high VAF in the plasma of the majority of cases (Table 1). Finally, the patient with systemic and concomitant CNS disease had ctDNA in plasma but not in the CSF, suggesting that some CNS lesions cannot be captured through CSF ctDNA.

Central nervous system circulating tumor DNA exhibits higher sensitivity than flow cytometry in detecting central nervous system lesions We next correlated the detection of ctDNA with the identification of malignant cells by conventional tests in the same CSF samples at time of enrollment. Cytology detected the presence of malignant cells in 4 out of 6 CNS restricted lymphomas. In these four cases the FC analysis also detect518

ed lymphoma cells. In one case (NHL2), cytology was normal but a small amount of tumor cells was observed by FC (0.06 cell/mL). Notably, all six cases exhibited CSF ctDNA. Importantly, in a patient with newly diagnosed PCNSL (NHL1) we could not detect lymphoma cells by FC or cytology; however, ctDNA was abundantly found in the CSF (Table 1). These results indicate that ctDNA analysis of the CSF can improve the detection of CNS lymphoma by conventional techniques.

Cerebrospinal fluid circulating tumor DNA can be used to monitor central nervous system tumor burden and response to treatment To address whether the amount of CSF ctDNA could be useful to monitor tumor burden in CNS lymphomas, we analyzed sequential concomitant samples of CSF and plasma. CSF was obtained as part of the standard procedures before each administration of IT MTX or on suspicion of relapse. Tumor burden was evaluated by MRI and/or FC depending on the lymphoma location. We observed that CSF ctDNA levels correlated with disease response and progression (Figure 2). In addition, tumor burden measured by MRI and FC was concordant with CSF ctDNA levels in two patients with CNS lymphoma (Figure 2). Importantly, plasma ctDNA levels remained constantly low (VAF <6%) haematologica | 2021; 106(2)

ctDNA in cerebrospinal fluid of patients with CNS lymphoma

C D

Figure 4. Cerebrospinal fluid (CSF) circulating tumor DNA (ctDNA) detects central nervous system (CNS) involvement and residual disease in patients with systemic lymphoma and is more sensitive than conventional tests. Longitudinal analyses of (A) CSF and (D) plasma ctDNA (VAF values for each mutation detected using ddPCR) and correlation with (B) CSF FC (cells/mL), (C) cytology (+/- presence/absence of cells), and (E) type of treatment and radiological response assessment. (F) Radiological imaging (magnetic resonance imaging [MRI] and positron emission tomography/computed tomography scans) in two patients with systemic lymphoma (NHL16 and NHL17) that present CNS relapse. Data points are vertically aligned. MRI scans for patient NHL17 are not shown given that the patient only presented leptomeningeal involvement. Dx: time of diagnosis; R-CHOP: combined therapy (rituximab, cyclophosphamide, doxorubicin, vincristine and prednisone); IT MTX: intrathecal methotrexate; CR: complete response; P: parenchymal; LM: leptomeningeal; EOT: end of treatment; m: month; d: days; NA: not available. *CSF sample was not available for ctDNA analyses

in sequential samples and did not correlate with the disease status, underlining the fact that plasma ctDNA is not useful to monitor the activity of CNS restricted lymphomas (Figure 2). Six patients with systemic lymphoma without CNS involvement had sequential plasma samples collected at the same time as CSF. We observed that plasma ctDNA became undetectable after the first cycle of treatment in all cases, in agreement with PET/CT results (Figure 2), suggesting that plasma ctDNA is a good tool to monitor response to treatment in these patients.

Cerebrospinal fluid circulating tumor DNA can identify central nervous system relapse and residual central nervous system disease better than flow cytometry We then addressed whether the presence of ctDNA in haematologica | 2021; 106(2)

the CSF could be more sensitive than conventional CSF analyses to monitor CNS lymphoma and detect CNS relapse. In two cases of CNS lymphoma (NHL3 and NHL5), we observed that CSF ctDNA levels increased earlier than the detection of cells by FC. In NHL3, after the first dose of IT MTX, despite the fact that the patient did not exhibit a complete neurological response, we observed a sharp decrease in CSF ctDNA levels concomitant with a decrease in the tumor cells by FC. However, after three doses of IT MTX, although FC was negative, a clear ctDNA increase was observed. The patient died 1 month later due to progression (Figure 3). In NHL5, CSF ctDNA levels also decreased with response to treatment and remained undetectable after complete remission. However, CSF ctDNA reappeared in two CSF samples collected 2 and 4 months 519

S. Bobillo et al.

after treatment due to the presence of minimal neurological symptoms, when brain-spinal MRI and CSF FC did not show any evidence of disease; 8 months later CNS relapse was confirmed (Figure 3). Taken together, our results indicate that ctDNA can be a more sensitive tool than FC to monitor relapse of CNS lymphomas. In one additional case of CNS lymphoma (NHL6) treatment with IT MTX decreased the number of cells in the CSF till they were no longer detectable by FC. However, CSF ctDNA was still detectable. Notably, this patient relapsed shortly after the analysis (Figure 3). In a similar way, we found persistence of ctDNA in the CSF from case NHL17 after treatment for CNS relapse, despite the fact that FC did not detect malignant cells. Four months later, the patient relapsed. (Figure 4). These results point out that the detection of CSF ctDNA may help to identify patients with residual disease who are likely to relapse. A subset of patients with systemic lymphoma presents high risk of CNS relapse, which frequently occurs during treatment or shortly thereafter. To address whether the presence of CSF ctDNA could be more sensitive than FC to detect CNS disease, we analyzed sequential CSF samples from eight patients with systemic lymphoma obtained while receiving prophylactic IT MTX. Two out of 8 patients relapsed in the CNS during the time of study (Figure 4). In one case (NHL17), we found CSF ctDNA at the same moment in which FC showed a positive result. However, we could not detect ctDNA in the CSF samples collected 9 and 7 months before CNS progression. Notably, in case NHL16, the detection of ctDNA in the CSF with a VAF of 7% antedated by 3 months the CNS relapse. Importantly, at the time of ctDNA detection, the patient was asymptomatic and CSF FC was normal (Figure 4). This case points out that CSF ctDNA could detect CNS involvement in systemic lymphomas earlier than standard procedures. Finally, in the remaining six patients with systemic lymphoma and sequential CSF samples, ctDNA was not detected and, with a median follow-up of 18 months (range: 12-28 months), these patients did not develop CNS disease.

Discussion The diagnosis and monitoring of CNS lymphoma are still challenging. MRI can be misleading and not sufficiently sensitive, and the difficulties of the intracranial tumor biopsies challenge histological diagnosis. Mutational analysis of the CSF could be useful in these cases, since the presence of MYD88 L265P mutation strongly suggests the diagnosis of PCNSL.13,14,16-18 Ours is a proof of concept study aiming to identify and characterize ctDNA in the CSF of patients with CNS lymphoma and to evaluate its role in the management of these patients. We were able to identify ctDNA in the CSF from all patients with restricted CNS disease and show that CSF ctDNA could complement the diagnosis, identify somatic mutations, monitor tumor burden and response to treatment, and facilitate early detection of relapse, as well as detect residual disease after treatment. In addition, our results support the notion that the CSF is a better source of ctDNA than plasma. Assessment of therapeutic response in CNS lymphomas is usually performed by brain MRI along with FC 520

and cytology. Regrettably, around 50-60% of patients with CNS lymphoma will eventually relapse.21 Notably, we observed that CSF ctDNA was able to detect CNS lesions better than FC and cytology. In addition, CSF ctDNA was able to predict CNS relapse months before MRI or FC, indicating that it can be used for early detection of tumor relapse. CNS relapse from systemic lymphoma is a fatal complication with an overall survival of less than 6 months, and early identification of patients with CNS relapse is of paramount value.3 High-risk patients are usually tested for CNS disease by CSF analysis including cytology and FC. Although FC is a more sensitive test than conventional cytology,5,6 it has some limitations22 and some patients with no evidence of CNS disease by FC still relapse, suggesting the presence of undetected malignant cells in the CSF. We identified ctDNA in the CSF in the absence of MRI or FC detection earlier than CNS relapse. This indicates that CSF ctDNA could facilitate early detection of CNS relapse in high-risk systemic lymphomas who undergo serial lumbar punctures to administrate prophylactic IT chemotherapy, therefore improving management of these patients. Moreover, we observed the persistence of high levels of CSF ctDNA in some patients even though FC did not detect tumor cells. This suggested that the detection of a residual disease by CSF ctDNA could be used to determine the type and duration of treatments. Taken together, and with the limitation of the reduced number of patients analyzed, our findings show that CSF ctDNA can be an important tool to complement standard procedures to evaluate the CNS lymphoma disease status. This technology can be exploited as a 'liquid biopsy' of CNS lymphoma opening a novel way forward for research in circulating biomarkers of CNS lymphoma with an important impact on the future characterization, diagnosis, prognosis, and management of this type of diseases. Disclosures JS is a co-founder of Mosaic Biomedicals and Northern Biologics. JS received grant/research support from Mosaic Biomedicals, Northern Biologics and Roche/Glycart. M. C. has received research funding from Karyopharm, Pharmacyclics, Roche, Arqule and AstraZeneca. Francesc Bosch has received research funding and honoraria from Roche, Celgene, Takeda, AstraZeneca, Novartis, Abbie and Janssen. All remaining authors have declared no conflicts of interest. Contributions SB, MC, LE, FB and JS contributed equally. Funding This work was supported by research funding from Fundación Asociación Española contra el Cáncer (AECC) (to JS, MC and PA); FERO (to JS), laCaixa (to JS), BBVA (CAIMI) (to JS), the Instituto de Salud Carlos III, Fondo de Investigaciones Sanitarias (PI16/01278 to JS; PI17/00950 to MC; PI17/00943 to FB) cofinanced by the European Regional Development Fund (ERDF) and Gilead Fellowships (GLD16/00144, GLD18/00047, to FB). MC holds a contract from Ministerio de Ciencia, Innovación y Universidades (RYC-2012-12018). SB received funding from Fundación Alfonso Martin Escudero. LE received funding from the Juan de la Cierva fellowship. We thank CERCA Programme / Generalitat de Catalunya for institutional support. haematologica | 2021; 106(2)

ctDNA in cerebrospinal fluid of patients with CNS lymphoma

References 1. Yamamoto W, Tomita N, Watanabe R, et al. Central nervous system involvement in diffuse large B-cell lymphoma. Eur J Haematol. 2010;85(1):6-10. 2. WHO classification of tumours of haematopoietic and lymphoid tissues. Revised 4th ed. Lyon, France: IARC; 2017. 3. El-Galaly TC, Cheah CY, Bendtsen MD, et al. Treatment strategies, outcomes and prognostic factors in 291 patients with secondary CNS involvement by diffuse large B-cell lymphoma. Eur J Cancer. 2018;93:57-68. 4. Grommes C, Rubenstein JL, DeAngelis LM, Ferreri AJM, Batchelor TT. Comprehensive approach to diagnosis and treatment of newly diagnosed primary CNS lymphoma. Neuro Oncol. 2019;21(3):296-305. 5. Quijano S, López A, Manuel Sancho J, et al. Identification of leptomeningeal disease in aggressive B-cell non-Hodgkin's lymphoma: improved sensitivity of flow cytometry. J Clin Oncol. 2009;27(9):1462-1469. 6. Hegde U, Filie A, Little RF, et al. High incidence of occult leptomeningeal disease detected by flow cytometry in newly diagnosed aggressive B-cell lymphomas at risk for central nervous system involvement: the role of flow cytometry versus cytology. Blood. 2005;105(2):496-502. 7. Roschewski M, Dunleavy K, Pittaluga S, et al. Circulating tumour DNA and CT monitoring in patients with untreated diffuse large B-cell lymphoma: a correlative biomarker study. Lancet Oncol. 2015;16(5):541-549. 8. Kurtz DM, Scherer F, Jin MC, et al. Circulating tumor DNA measurements as

haematologica | 2021; 106(2)

early outcome predictors in diffuse large Bcell lymphoma. J Clin Oncol. 2018; 36(28):2845-2853. 9. De Mattos-Arruda L, Mayor R, Ng CKY, et al. Cerebrospinal fluid-derived circulating tumour DNA better represents the genomic alterations of brain tumours than plasma. Nat Commun. 2015;6:8839. 10. Pentsova EI, Shah RH, Tang J, et al. Evaluating cancer of the central nervous system through next-generation sequencing of cerebrospinal fluid. J Clin Oncol. 2016; 34(20):2404-2415. 11. Momtaz P, Pentsova E, Abdel-Wahab O, et al. Quantification of tumor-derived cell free DNA(cfDNA) by digital PCR (DigPCR) in cerebrospinal fluid of patients with BRAFV600 mutated malignancies. Oncotarget. 2016;7(51):85430-85436. 12. Martínez-Ricarte F, Mayor R, Martínez-Sáez E, et al. Molecular diagnosis of diffuse gliomas through sequencing of cell-free circulating tumor DNA from cerebrospinal fluid. Clin Cancer Res. 2018;24(12):28122819. 13. Fontanilles M, Marguet F, Bohers É, et al. Non-invasive detection of somatic mutations using next-generation sequencing in primary central nervous system lymphoma. Oncotarget. 2017;8(29):48157-48168. 14. Hattori K, Sakata-Yanagimoto M, Suehara Y, et al. Clinical significance of disease-specific MYD88 mutations in circulating DNA in primary central nervous system lymphoma. Cancer Sci. 2018;109(1):225-230. 15. Grommes C, Tang SS, Wolfe J, et al. Phase 1b trial of an ibrutinib-based combination therapy in recurrent/refractory CNS lym-

phoma. Blood. 2019;133(5):436-445. 16. Hiemcke-Jiwa LS, Leguit RJ, Snijders TJ, et al. MYD88 p.(L265P) detection on cell-free DNA in liquid biopsies of patients with primary central nervous system lymphoma. Br J Haematol. 2019;185(5):974-977. 17. Hiemcke-Jiwa LS, Minnema MC, Radersma-van Loon JH, et al. The use of droplet digital PCR in liquid biopsies: a highly sensitive technique for MYD88 p.(L265P) detection in cerebrospinal fluid. Hematol Oncol. 2018;36(2):429-435. 18. Hickmann AK, Frick M, Hadaschik D, et al. Molecular tumor analysis and liquid biopsy: a feasibility investigation analyzing circulating tumor DNA in patients with central nervous system lymphomas. BMC Cancer. 2019;19(1):192. 19. Rimelen V, Ahle G, Pencreach E, et al. Tumor cell-free DNA detection in CSF for primary CNS lymphoma diagnosis. Acta Neuropathol Commun. 2019;7(1):43. 20. Forshew T, Murtaza M, Parkinson C, et al. Noninvasive identification and monitoring of cancer mutations by targeted deep sequencing of plasma DNA. Sci Transl Med. 2012;4(136):136ra168. 21. Langner-Lemercier S, Houillier C, Soussain C, et al. Primary CNS lymphoma at first relapse/progression: characteristics, management, and outcome of 256 patients from the French LOC network. Neuro Oncol. 2016;18(9):1297-1303. 22. Canovi S, Campioli D. Accuracy of flow cytometry and cytomorphology for the diagnosis of meningeal involvement in lymphoid neoplasms: a systematic review. Diagn Cytopathol. 2016;44(10):841-856.

521

ARTICLE Ferrata Storti Foundation

Coagulation & its Disorders

uPA-mediated plasminogen activation is enhanced by polyphosphate Claire S. Whyte and Nicola J. Mutch

Aberdeen Cardiovascular & Diabetes Center, School of Medicine, Medical Sciences & Nutrition, Institute of Medical Sciences, University of Aberdeen, Aberdeen, UK

Haematologica 2021 Volume 106(2):522-531

ABSTRACT

issue plasminogen activator (tPA) and urokinase plasminogen activator (uPA) differ in their modes of action. Efficient tPA-mediated plasminogen activation requires binding to fibrin. In contrast, uPA is fibrin independent and activates plasminogen in solution or when associated with its cellular receptor urokinase protease activated receptor (uPAR). We have previously shown that polyphosphate (polyP), alters fibrin structure and attenuates tPA and plasminogen binding to fibrin, thereby down-regulating fibrinolysis. Here we investigate the impact of polyP on uPA-mediated fibrinolysis. As previously reported polyP of an average chain length of 65 (polyP ) delays tPA-mediated fibrinolysis. The rate of plasmin generation was also delayed and reduced 1.6-fold in polyP -containing clots (0.74±0.06 vs. 1.17±0.14 pM/s in P<0.05). Analysis of tPA-mediated fibrinolysis in real-time by confocal microscopy was significantly slower in polyP -containing clots. In marked contrast, polyP augmented the rate of uPA-mediated plasmin generation 4.7-fold (3.96±0.34 vs. 0.84±0.08 pM/s; P<0.001) and acce-lerated fibrinolysis (t1/2 64.5±1.7 min vs. 108.2±3.8 min; P<0.001). Analysis of lysis in real-time confirmed that polyP enhanced uPA-mediated fibrino-lysis. Varying the plasminogen concentration (0.125-1 mM) in clots dose-dependently enhanced uPA-mediated fibrinolysis, while negligible changes were observed on tPA-mediated fibrinolysis. The accelerating effect of polyP on uPA-mediated fibrinolysis was overcome by additional plasminogen, while the down-regulation of tPA-mediated lysis and plasmin generation was largely unaffected. polyP exerts opposing effects on tPA- and uPA-mediated fibrinolysis, attenuating the fibrin cofactor function in tPA-media-ted plasminogen activation. In contrast, polyP may facilitate the interaction between fibrin-independent uPA and plasminogen thereby accelerating plasmin generation and downstream fibrinolysis. 65

Correspondence: NICOLA J. MUTCH n.j.mutch@abdn.ac.uk Received: September 12, 2019. Accepted: January 31, 2020. Pre-published: February 6, 2020. https://doi.org/10.3324/haematol.2019.237966

522

Introduction Polyphosphate (polyP) is a biomolecule composed of orthophosphate residues (Pi) linked by phosphoanhydride bonds.1 polyP of average chain length of 60-100-mers is a constituent of platelet dense granules and is released following stimulation of platelets with different agonists.1,2 polyP acts at numerous points in the coagulation cascade to augment clot formation, including stimulating factor XII (FXII) activation, thrombin-mediated factor XI (FXI) activation and interfering with the function of TFPI.3 Our work has shown that polyP interferes with fibrin polymerization by stunting protofibril growth, producing a heterogeneous network of dense ‘knotted’ regions interspersed by pores with altered mechanical properties.4 Plasmin is the serine protease responsible for degradation of fibrin. Two forms of the zymogen precursor plasminogen circulate in plasma, the native more abundant form, Glu-plasminogen, and the intermediate form Lys-plasminogen, formed by cleavage of the N-terminal peptide from Glu-plasminogen.5 Lys-plasminogen exists in flexible open conformation, with an approximately 10-fold higher binding affinity for plasminogen activators thereby facilitating its activation.6-9 Two-chain plasmin is formed by enzymatic cleavage of plasminogen at Arg561-Val562.5,10 The two haematologica | 2021; 106(2)

PolyP enhances uPA-mediated plasminogen activation

main physiological plasminogen activators are the serine proteases tissue plasminogen activator (tPA) and urokinase plasminogen activator (uPA). Efficient plasminogen activation by tPA requires fibrin as a cofactor, acting as a template for its own dissolution by binding tPA and plasminogen.11 uPA is fibrin-independent and efficiently activates plasminogen in solution, but is often found in association with its cellular receptor, urokinase protease activated receptor (uPAR).12 uPAR does not have a catalytic role but acts to localize plasminogen and uPA to the cell surface increasing local reactant concentration. We have previously shown that binding of tPA and plasminogen to fibrin is downregulated by polyP, in turn, delaying tPA-mediated fibrinolysis.4 Here, we characterize the impact of polyP on uPA-mediated plasminogen activation. Our data reveal that polyP has a strong accelerating effect on uPA-mediated plasminogen activation that is highly dependent on concentration of polyP and plasminogen. Enhanced uPA-mediated plasminogen activation in the presence of polyP translates as a robust profibrinolytic effect in clot lysis assays. These data exemplify the complexity of polyP’s action on hemostasis and illustrate that the presence of different activators and substrates in the local milieu can direct functional activity.

biotin-labelled polyP (71 mM) were performed using an adaptation of the protocol described by Choi et al.13 Bound tPA, uPA or plasmin was detected with chromogenic substrates (1.2 mM S2288, CS-61 44, or 0.6 mM S2251, respectively). FXII(a) was detected using a peroxidase conjugated goat anti-human FXII antibody.

Turbidimetric fibrinolysis assays

Purified human plasminogen-free fibrinogen (2.4 mM), Glu- or Lys-plasminogen (0–1 mM), tPA (20 pM) or uPA (180 pM) with or without polyP (0–1.3 mM) in TBST was added in triplicate to 96well polystyrene plates. Clotting was initiated by thrombin (0.25 U/mL) and CaCl (5 mM), and turbidity monitored every min at 340 nm for 5 h at 37°C in a FLX-800 plate reader (Biotek Instruments). 2

Confocal Microscopy Clots were prepared with human plasminogen-free fibrinogen (2.65 mM) of which 9% was DyLight 488-labeled, Glu-plasminogen (1.25 mM) of which 20% was DyLight 633-labeled, ±328 mM polyP in TBST in Ibidi -slides VI0.4. Thrombin (0.25 U/mL) and CaCl (5 mM) were added and fibrinolysis initiated by addition of tPA or uPA (75 nM). Cascade blue ethylenediamine (CB)-labeled polyP was prepared as described.13 Clots were formed ±328 mM CB-polyP by polymerizing fibrinogen (2.65 mM, with 9% labeled with DL550fibrinogen) as above. 65

Methods Statistical analysis Additional experimental details can be found in the Online Supplementary Appendix.

Ethical Consent

Statistical analysis was performed in GraphPad Prism® 5.04 using one-way analysis of variance or two-way analysis of variance with Bonferroni post hoc test or an unpaired Student’s t-test (2tailed). P<0.05 was considered to be significant.

Ethical approval was obtained from the University of Aberdeen College Ethics Review Board.

Results Materials PolyP was a kind gift from Dr Thomas Staffel BK Giulini GmbH (Ludwigshafen, Germany). PolyP 14, 60, 130 were a kind gift from Dr Toshikazu Shiba Regenetiss Inc. Medium chain (p100) polyP with and without biotin-labeling was from Kerafast Inc (Boston, MA, USA). PolyP concentrations are expressed as monomer concentrations throughout (monomer formula NaPO ). All reactions were carried out in TBST (50 mM Tris, 100 mM NaCl, 0.01% Tween-20, pH 7.4). 70

Plasmin generation and uPA activity

Purified human plasminogen-free fibrinogen (2.4 mM), Glu- or Lys-plasminogen (0–1 mM), tPA (20 pM) or uPA (180 pM) ± polyP (328 mM) in TBST was added in triplicate to 96-well polystyrene plates. Clotting was initiated by thrombin (0.25 U/mL) and CaCl (5 mM), and activity quantified using the fluorogenic substrate DVal-Leu-Lys 7-amido-4-methylcoumarin (D–VLK-AMC [0.35 mM]) by measuring fluorescence release (excitation 360/40 nm, emission 460/40 nm) every minute (min) for 5 hours (h) in a Biotek Flx800 fluorescence microplate reader at 37 °C. The rate of plasmin generation was calculated using; Longstaff C, 2016, Shiny App for calculating zymogen activation rates, version 0.6 (https://drclongstaff.shinyapps.io/zymogenactnCL/). UPA activity was determined by incubating the enzyme (180 pM) ±328 mM polyP with the chromogenic substrate CS-61 44 (1.25 mM). Change in absorbance was measured every 30 seconds (s) at 405 nm for 200 min.

Protein binding assays to biotin-labelled polyP Binding of tPA, uPA, plasmin, FXII and activated FXII (FXIIa) to haematologica | 2021; 106(2)

PolyP promotes uPA-mediated plasmin generation Our work has previously established that polyP interferes with the plasminogen activator function of tPA4 while augmenting plasminogen activation by FXIIa.14 Here we address the role of polyP on uPA-mediated plasminogen activation using the fluorogenic substrate D–VLKAMC. PolyP dramatically enhanced the rate of plasmin generation (3.96±0.34 vs. 0.84±0.08 pM/s; P<0.001) during uPA-mediated fibrinolysis (Figure 1A). In contrast, and consistent with our previous results4, a decrease in the rate (0.74±0.06 vs. 1.17±0.14 pM/s in P<0.05) and amount of plasmin generation was observed during tPA-mediated fibrinolysis (Figure 1B). The ability of polyP to enhance the rate of uPA-mediated lysis was dose-dependent up to 32.8 mM after which it decreased before increasing again at 164 mM (Figure 1C). Above this concentration the rate of plasmin generation was greatly enhanced with a 6-fold increase at 328 mM polyP. In marked contrast, downregulation of tPA-mediated plasmin generation required concentrations of greater than 70 mM and was not strongly does-dependent (Figure 1D).

Plasminogen concentration attenuates the cofactor function of polyP in uPA-mediated plasmin generation We next analysed the impact of plasminogen concentration on the cofactor function of polyP. Plasmin generation was monitored during clot lysis at various concentrations of plasminogen. The rate of plasmin generation by uPA and tPA in clots was directly proportional to the Glu-plas523

C.S. Whyte and N.J. Mutch

minogen concentration (Figure 2A-B). The presence of polyP significantly increased the rate of uPA-mediated plasmin generation by around 75–82% at all Glu-plasminogen concentrations (Figure 2A). In marked contrast, polyP attenuated tPA-mediated plasmin generation at all Glu-plasminogen concentrations, with a maximal reduction at 1 mM Glu-plasminogen (Figure 2B). Cleavage of the activation peptide of Glu-plasminogen by plasmin forms an intermediate form, Lys-plasminogen, which displays increased affinity for fibrin15 and is more readily cleaved by plasminogen activators.7,16,17 Overall plasmin generation by uPA occurs at a significantly faster rate in clots formed with Lys-plasminogen compared to Glu-plasminogen and is directly proportional to the concentration (Figure 2C-D). Interestingly, Lys-plasminogen overcame the cofactor function of polyP on uPA-mediated plasmin generation, significantly decreasing the magnitude of the effect from 79% at 0.125 mM Lys-plasminogen to 9% at 1 mM Lys-plasminogen (Figure 2C). With tPA there was a similar attenuation of plasmin generation (approximately 37%) at all Lys-plasminogen concentrations analysed (Figure 2B, D).

polyP binds with a higher affinity to uPA than tPA We have previously demonstrated that polyP binds to Glu-plasminogen.14 Here we use biotin-labelled polyP in a plate-based assay to determine the binding affinity to (A) uPA and (B) tPA. polyP bound with approximately 25-fold higher affinity to uPA (K =9.8 nM) than tPA (K =245.3 nM; Figure 3). Interestingly, binding of polyP did not alter the activity of uPA and no specific binding of polyP to plasmin was detected (data not shown). We have shown that polyP stimulates autoactivation of FXII to an active single chain d

FXII (scFXIIa)18 and enhances the plasminogen activator capacity of FXIIa.14 We found that the high affinity interaction of uPA and polyP was similar to the interaction of polyP with FXII (K =7.9 nM) and FXIIa (Kd=4.0 nM) (Online Supplementary Figure S1). d

polyP enhances uPA-mediated lysis in a concentration and polymer length dependent manner Incorporation of polyP during clot formation shortened the uPA-mediated 50% clot lysis time by 43.7±1.7 min (P<0.0001; Figure 4A). In contrast, tPA-mediated clot lysis was delayed on inclusion of polyP (Figure 4B), as previously reported.4 The cofactor function of polyP on uPA-mediated lysis was concentration dependent and required 66 mM or higher to significantly stimulate the rate of lysis (Figure 4C). The use of polyP of more refined chain length revealed that polymers of >14-mer were required to promote uPA-mediated lysis (Figure 4D).

Plasminogen modulates polyP-mediated effects on fibrinolysis The mode of action of tPA and uPA differ significantly, with tPA relying on the interaction with fibrin for efficient plasminogen activation, while uPA is fibrin independent. We investigated the relationship between plasminogen and polyP-mediated effects on fibrinolysis. A clear concentration dependence was observed with Glu-plasminogen on uPA-mediated lysis, with maximal lysis at 0.75 mM plasminogen (Figure 5A). The impact of Glu-plasminogen on tPA-mediated lysis was less pronounced, but a small effect was observed with maximal lysis at 0.5 mM plasminogen (P<0.01; Figure 5B). polyP significantly accelerated uPA-mediated lysis at all Glu-plasminogen concentra-

Figure 1. PolyP stimulates uPA-mediated plasmin generation in a concentration dependent manner. Fibrin clots were prepared containing 2.4 mM fibrinogen, 0.24 mM Glu-plasminogen, 20 pM tPA or 180 pM uPA ± 328 mM (polyP ). Clotting was initiated with thrombin (0.25 U/mL) and CaCl (5 mM) and plasmin generation was quantified by incorporating the fluorogenic substrate D–VLK-AMC and monitoring fluorescence release (FU; Ex 360 nm Em 460 nm). (A) uPA and (B) tPA mediated plasmin generation curves. Rate of plasmin generation in clots formed with varying polyP concentration (0-1.3 mM polyP ) for (C) uPA and (D) tPA. Data are expressed as mean ± standard error of the mean, n≥ 3. PolyP: polyphosphate; tPA: tissue plasminogen activator; uPA: urokinase plasminogen activator; FU: fluorescence units, Ex: excitation; EM: emission. 65

524

haematologica | 2021; 106(2)

PolyP enhances uPA-mediated plasminogen activation

tions tested, however, the cofactor function of polyP was tempered at high concentrations. Inclusion of polyP delayed tPA-mediated lysis at all Glu-plasminogen concentrations tested by an average of 40%; therefore, increasing the Glu-plasminogen concentration did not alter the efficacy of polyP on tPA-driven lysis. Clots containing Lys-plasminogen lyse significantly faster than those containing Glu-plasminogen (Figure 5C-

D). A dose-dependent relationship exists between Lys-plasminogen concentration and uPA-mediated lysis, with 50 % lysis times decreasing from 51.1±1.36 min at 0.125 mM to 17.2±0.32 min at 1 mM (Figure 5C). In contrast, there is no relationship between Lys-plasminogen concentration and tPA-mediated lysis, exemplifying the different modes of action of these plasminogen activators (Figure 5D). Similar to the results with plasmin generation,

Figure 2. High concentrations of plasminogen attenuate the cofactor function of polyP in uPA-mediated plasminogen activation. Fibrin clots were prepared with 2.4 mM fibrinogen, 0 -1 mM (A-B) Glu-plasminogen or (C-D) Lys-plasminogen, (A, C) 180 pM uPA or (B, D) 20 pM tPA ± 328 mM polyP . Clotting was initiated with thrombin (0.25 U/mL) and CaCl (5 mM). Plasmin generation in clots was quantified by incorporating the fluorogenic substrate D–VLK-AMC and monitoring fluorescence release (FU; Ex 360 nm Em 460 nm). *P<0.05; **P<0.01, ***P<0.001 and ****P<0.0001 compared with control clots. Data are expressed as mean ± standard error of the mean, n≥3. polyP: polyphosphate; tPA: tissue plasminogen activator; uPA: urokinase plasminogen activator; FU: fluorescence units; Ex: excitation; EM: emission. 65

Figure 3. polyP binds to uPA with a significantly higher affinity than tPA. Binding of uPA or tPA (0–400 nM) to biotin-labelled polyP (71 mM) bound to streptavidin coated stripwells. Bound uPA (orange) or tPA (blue) was detected with chromogenic substrates (S2288 or CS-61 44 respectively) by reading the change in absorbance at 405 nm every 30 seconds (s) for 200 minutes (min). No unspecific binding was detected in the absence of biotin-labelled polyP (black lines). Data are expressed as baseline corrected nonlinear fit as mean ± standard error of the mean (SEM), n=4. PolyP: polyphosphate; tPA: tissue plasminogen activator; uPA: urokinase plasminogen activator.

haematologica | 2021; 106(2)

525

C.S. Whyte and N.J. Mutch

the cofactor function of polyP on uPA-mediated lysis is lost at high Lys-plasminogen concentrations (Figure 5C). Whereas, polyP delayed tPA-mediated lysis by approximately 20 % at all Lys-plasminogen concentrations (Figure 5B, D).

poration of polyP significantly accelerates uPA-mediated lysis (Figure 7A) with full lysis observed on average at 6 min (Online Supplementary Video S1B) compared to greater than 9 min in the control (Online Supplementary Video S1A).

polyP localization during fibrinolysis

Discussion

polyP is found to co-localize with fibrinogen, as previously reported,19 and plasminogen in fibrin clots. During fibrinolysis polyP concentrates within the fibrin dense knots of the clot alongside plasminogen and fibrinogen (Figure 6A). Colocalization was observed with both uPA and tPA but only the data for uPA is shown. Lysis was initiated at the edge of the clot, by addition of plasminogen activator, and as it progressed through the fibrin network polyP and plasminogen were found to remain colocalized within the lysis front (Figure 6B).

Visualization of the cofactor function of polyP during uPA-mediated lysis in real-time Fluorescent confocal microscopy was used to visualize lysis in real-time following application of either uPA or tPA to the edge of fibrin clots. The rate of fibrinolysis in the control clot was similar in the presence of uPA or tPA (Figure 7A-B). As uPA-mediated lysis progresses plasmin(ogen) is dispersed throughout the clot, with a small concentrated zone at the leading edge (Figure 7A), in contrast plasminogen is restricted to a distinct zone at the leading edge in tPA-mediated lysis (Figure 7B). The incor-

Over the last decade a role for the biomolecule polyP in the regulation of hemostasis has been exposed. The ability of polyP to modulate fibrinolysis manifests on several levels. We and others4,20 have demonstrated altered fibrin structure in the presence of polyP, a consequence of which is tempered binding of tPA and plasminogen to fibrin and subsequent down-regulation of fibrinolysis.4 In contrast, in this study we show that that polyP significantly accelerates uPA-mediated plasminogen activation thereby augmenting fibrinolysis. The rate of plasmin formation was approximately 5-fold faster in the presence of polyP and was dependent on concentration and polymer length, with polymers of 60-mer or greater required to accelerate lysis. We determined that polyP binds with a 25-fold higher affinity to uPA than tPA and potentially accelerates conversion of Glu-plasminogen to the intermediate form Lysplasminogen via a template mediated effect. Importantly, for the first time we visualize polyP during real-time lysis of fibrin clots and find it to be localized within fibrin dense areas alongside plasminogen, consistent with its known binding to these proteins.4,14,19

Figure 4. polyP acts as a cofactor to accelerate uPA-mediated fibrinolysis in a concentration and polymer size dependent manner. Fibrin clots were prepared containing 2.4 mM fibrinogen, 0.24 mM Glu-plasminogen, 20 pM tPA or 180 pM uPA Âą328 mM polyP . Clotting was initiated with thrombin (0.25 U/mL) and CaCl (5 mM) and fibrinolysis monitored at 340 nm shown as percentage turbidity over time with (A) uPA or (B) tPA. (C) uPA-mediated fibrinolysis with 0-1.3 mM polyP or (D) polyP of various chain lengths at equivalent concentration of monomer (328 mm). *P<0.05, ***P<0.001 and ****P<0.0001 compared with control clots. Data are expressed as mean 50 % lysis time (t ) Âą standard error of the mean, nâ&#x2030;Ľ3. polyP: polyphosphate; tPA: tissue plasminogen activator; uPA: urokinase plasminogen activator. 65

1/2

526

haematologica | 2021; 106(2)

PolyP enhances uPA-mediated plasminogen activation

Figure 5. The cofactor function of polyP in uPA-mediated fibrinolysis is modulated by plasminogen concentration and form. Fibrin clots were prepared containing 2.4 mM fibrinogen, 0-1 mM (A-B) Glu-plasminogen or (C-D) Lys-plasminogen, (A, C) 180 pM uPA or (B, D) 20 pM tPA Âą328 mM polyP . Clotting was initiated with thrombin (0.25 U/mL) and CaCl (5 mM) and fibrinolysis monitored at 340 nm. *P<0.05, **P<0.01, ***P<0.001 and ****P<0.0001 compared with control clots. Data are expressed as mean Âą standard error of the mean, nâ&#x2030;Ľ3. polyP: polyphosphate; tPA: tissue plasminogen activator; uPA: urokinase plasminogen activator. 65

Despite their homology in structure, the mechanism of tPA and uPA-mediated plasminogen activation are very different. Efficient plasminogen activation by tPA requires its association with the fibrin surface, that is fibrin orchestrates its own destruction by plasmin. Binding of tPA to fibrin is largely attributed to its finger domain,21,22 while plasminogen associates with fibrin via kringle domains.23 The colocalization of enzyme and substrate on fibrin, as a surface, is crucial, as tPA is a poor plasminogen activator in solution. Circulating fibrinogen cannot accelerate plasminogen activation by tPA, as the sites are only exposed in fibrin24 thereby localizing plasmin formation to the fibrin clot. In marked contrast uPA does not bind fibrin and is reasonably efficient at activating plasminogen in solution. However, its cellular receptor, uPAR, localizes uPA via its amino terminal fragment to the cell surface, thereby augmenting plasminogen activation due to an increase in local reactant concentration. The high affinity interaction of uPA and plasminogen14 with polyP and dependence on substrate and template concentration are indicative of a template mechanism of activation. In line with this polyP polymers of around 60mer were required to stimulate plasminogen activation by uPA, suggesting this length is critical to accommodate binding of both enzyme and substrate to the same template. Indeed, a template effect of polyP in thrombinmediated activation of FV to FVa has previously been reported.25 Polymers of around 80-100-mer are secreted following platelet stimulation2 with insoluble divalent cation bound polyP nanoparticles remaining associated haematologica | 2021; 106(2)

with the activated platelet membrane.26 Platelet-derived polyP could actively participate in these template-mediated reactions within the milieu of a thrombus where local concentrations of the polymer will be high. Previously an inhibitory effect of polyP on uPA-mediated fibrinolysis was described in a plasma based system.25 The data presented here delineates the role of polyP specifically on plasminogen activation by uPA using a purified system. The roles of polyP in hemostasis are multifaceted and coagulation and fibrinolysis are inextricably linked. It is therefore difficult, at this current time, to exclude the fact that these differences cannot be attributed to different experimental setups, reagents and indeed an impact of polyP on a different part of the pathway. Intriguingly, the cofactor function of polyP described here in uPA-mediated plasminogen activation is markedly more pronounced than in uPA-mediated fibrinolysis. This can potentially be explained by an increase in the number of binding sites, that is plasminogen has the capacity to bind both polyP and lysine residues exposed on partially degraded fibrin generated following onset of lysis.23 Thus, the increase in surface binding sites on fibrin for plasminogen tempers the cofactor function of polyP. Therefore, there is a more evident effect at low plasminogen concentrations, where fewer lysine residues will be generated. Given the anionic nature of polyP it is feasible that it associates with positively charged lysine residues on fibrin, thereby obscuring the kringle-dependent binding of plasminogen to fibrin, however, high plasminogen concentrations eliminate the need for fibrin binding. 527

C.S. Whyte and N.J. Mutch A

Figure 6. polyP colocalizes with fibrinogen and plasminogen during fibrinolysis. Fibrin clots were formed containing fibrinogen (2.65 mM, 9% DyLight 550-labeled), Glu-plasminogen (1.25 mM, 20% DyLight 633-labeled) 328 mM Cascade-blue labeled polyP70 (CB-polyP), thrombin (0.25 U/mL), CaC (5 mM). (A) polyP and plasminogen largely accumulate in the knotted regions of fibrin. (B) Fibrinolysis was initiated by exogenous uPA (75 nM). Plasminogen and polyP strongly colocalize at the lysis front during clot lysis, as indicated by the arrows. Scale bars =10 mm. polyP: polyphosphate; uPA; urokinase plasminogen activator. l2

Our recent study19 investigating the molecular mechanisms underpinning the changes in fibrin structure in the presence of platelet-derived polyP revealed a significant change in fibrin polymerisation which stunts protofibril growth; thus providing an explanation for the characteristic ‘knotted’ appearance.4 Downstream this impacts on the mechanical properties of a clot, reducing overall stiffness and increasing its ability to deform plastically.19 Here, we find that polyP is localized within these ‘knotted’ regions of fibrin alongside plasminogen. Analysis of lysis in real-time reveals significant acceleration of uPA-mediated fibrinolysis presumably due to the direct colocalization of cofactor, enzyme and substrate. Previous work has shown that co-assembly of uPA and plasminogen on the same surface is not a prerequisite to stimulate plasmin formation.27 This crosstalk mechanism permits localization of plasminogen and uPA on different cellular surfaces or binding of plasminogen to fibrin while uPA is associated with cellular uPAR. A similar mechanism could explain our current observations; that is uPA is localized on polyP while plasminogen is bound to either polyP or partially degraded fibrin. Future work is necessary to ascertain the effect polyP may have on plasminogen activation on the surface of monocytes and neutrophils, which express both uPA and its cellular receptor uPAR. Plasminogen circulates in the native or Glu-plasminogen form but can be cleaved by plasmin at Lys77-Lys78 to generate Lys-plasminogen. This cleavage prompts changes in the properties of the zymogen thereby providing a positive feedback mechanism.28 Lys-plasminogen is a considerably better substrate for both tPA29 and uPA30 and displays enhanced affinity for fibrin.31,32 Surface-bound Glu-plasminogen is more readily cleaved to Lys- plasminogen than in solution.33 We have previously shown,14 and confirmed in this study, that plasminogen, but not plasmin, bindsdi528

rectly to polyP, as does the activator uPA. Co-assembly of Glu-plasminogen and uPA on the polyP surface will augment local concentrations of the reactants. An initial spark of plasmin formation will drive cleavage of Glu-plasminogen to Lys-plasminogen. Once formed Lys-plasminogen is more readily activated to plasmin than native Glu-plasminogen. Stimulation of uPA-mediated plasminogen activation by polyP is diminished when Lys-plasminogen is used, indicating that its cofactor function may lie in initial surface-mediated conversion of Glu- to Lys-plasminogen, ultimately accelerating plasmin formation. We have shown that plasmin does not bind to polyP4 or impact on its enzymatic activity indicating that the enhanced lysis arises at the level of plasminogen activation. The lysine analogue, tranexamic acid (TXA), downregulates lysis by blocking tPA-mediated plasmin generation. The effect of TXA on uPA-mediated plasmin generation is more complex, with high concentrations of TXA augmenting plasmin generation, despite this fibrinolysis is still inhibited. Enhanced activation of plasminogen by uPA in the presence of TXA is indicative of a conformational change in plasminogen, from a closed structure to an open and more readily activated form.34 Our observations with polyP are similar, in that binding of Glu-plasminogen appears to augment its susceptibility to uPA-mediated cleavage indicative of a conformational change in the protein. The fibrin specificity of tPA has led to the view that it is the dominant plasminogen activator in hemostasis, whereas, uPA has been implicated in plasmin-mediated cell migration, tissue remodelling and activation of latent growth factors and cytokines.35 TPA has been implicated in the degradation of deep vein thrombi in humans36 but genetic deficiency in mice has no impact. In marked contrast, uPA deficiency in mice markedly impairs venous thrombus resolution37 and conversely delivery of uPA to haematologica | 2021; 106(2)

PolyP enhances uPA-mediated plasminogen activation

Figure 7. Real-time lysis of fibrin clots reveals the cofactor function of polyP on uPA-mediated fibrinolysis. Fibrinolysis by exogenous PA (75 nM) was monitored by fluorescent confocal microscopy of clots formed from fibrinogen (2.65 mM, 9% DyLight 488-labeled), plasminogen (1.25 mM, 20% DyLight 633-labelled), thrombin (0.25 U/mL), CaCl (5 mM) Âą 328 mM polyP . Images were taken every 15 seconds (s). (A) uPA and (B) tPA mediated fibrinolysis over time showing fibrinogen (green), plasminogen (red) and merged images where co-localisation is visualised as yellow. Representative image from n=3, scale bar=10 mm. (C) Quantification of lysis time in s as determined by the time taken to lyse scan area (134.8 mm x 134.8 mm). Data expressed as mean Âąstandard error of the mean, n=3. polyA: polyphosphate; tPA: tissue plasminogen activator; uPA: urokinase plasminogen activator. 2

thrombi, either directly38 or via uPA expressing monocytes,39 improves thrombus dissolution. Cross-talk between neutrophils, monocytes and platelets contributes to deep vein thrombosis40 and it is feasible that plateletderived polyP could augment monocyte-derived uPAmediated plasmin formation and facilitate thrombus resolution. Platelets contain approximately 0.74 nmol/108 platelets1 therefore concentrations in whole blood are estimated to reach around â&#x2030;&#x2C6;3 mM following platelet activation.25 However, in platelet dense regions of thrombi concentrations could far exceed this; particularly as we and others have also shown that polyP remains bound to the activated platelet membrane.26,41 Elegant studies have revealed that solute transport within the core regions of thrombi is restricted42-44 and as such the diffusion rate of plateletderived polyP within this milieu will be restricted. PolyP nanoparticles formed on the activated platelet membrane trigger contact activation26 and our laboratory has described a role for polyP in augmenting FXIIa-mediated plasminogen activation, with the platelet surface acting as haematologica | 2021; 106(2)

a focal point for colocalization.14 By acting as a template for plasminogen activation by FXIIa or uPA, platelet-associated polyP nanoparticles could, under certain conditions, function to limit thrombus size. Here we show that polyP binds to uPA with a high affinity and enhances uPA-mediated plasmin generation thereby augmenting the rate of fibrinolysis. This is in sharp contrast to the inhibitory effect of this polyanion on tPA-mediated plasminogen activation.4 The strong binding of uPA and plasminogen is suggestive of a template mechanism that results in enhanced conversion of Glu-plasminogen to Lys-plasminogen and is supported by the colocalization of the reactants on fibrin. The interaction of Glu-plasminogen with polyP may additionally result in a conformational change thereby facilitating its activation by uPA. Our data define a novel cofactor function of polyP in regulation of plasminogen activation by uPA that drives fibrinolysis in real time and may have important implications for physiological processes such as thrombus resolution, cell migration and tissue remodeling. 529

C.S. Whyte and N.J. Mutch

Disclosures No conflicts of interests to disclose. Contributions CSW performed the research, analysed the data and wrote the manuscript; NJM designed the research, analysed the data and wrote the manuscript. Acknowledgments We would like to thank Ms Michela Donnarumma and Ms

References 1. Ruiz FA, Lea CR, Oldfield E, Docampo R. Human platelet dense granules contain polyphosphate and are similar to acidocalcisomes of bacteria and unicellular eukaryotes. J Biol Chem. 2004;279(43):44250-44257. 2. Muller F, Mutch NJ, Schenk WA, et al. Platelet polyphosphates are proinflammatory and procoagulant mediators in vivo. Cell. 2009;139(6):1143-1156. 3. Morrissey JH, Choi SH, Smith SA. Polyphosphate: an ancient molecule that links platelets, coagulation, and inflammation. Blood. 2012;119(25):5972-5979. 4. Mutch NJ, Engel R, Uitte de Willige S, Philippou H, Ariens RA. Polyphosphate modifies the fibrin network and down-regulates fibrinolysis by attenuating binding of tPA and plasminogen to fibrin. Blood. 2010;115(19):3980-3988. 5. Robbins KC, Bernabe P, Arzadon L, Summaria L. NH2-terminal sequences of mammalian plasminogens and plasmin Scarboxymethyl heavy (A) and light (B) chain derivatives. A re-evaluation of the mechanism of activation of plasminogen. J Biol Chem. 1973;248(20):7242-7246. 6. Mangel WF, Lin BH, Ramakrishnan V. Characterization of an extremely large, ligand-induced conformational change in plasminogen. Science. 1990;248(4951):69-73. 7. Sjoholm I. Studies on the conformational changes of plasminogen induced during activation to plasmin and by 6-aminohexanoic acid. Eur J Biochem. 1973;39(2):471-479. 8. Castellino FJ. Biochemistry of human plasminogen. Semin Thromb Hemost. 1984; 10(1):18-23. 9. Urano T, Chibber BA, Castellino FJ. The reciprocal effects of epsilon-aminohexanoic acid and chloride ion on the activation of human [Glu1]plasminogen by human urokinase. Proc Natl Acad Sci U S A. 1987; 84(12):4031-4034. 10. Robbins KC, Summaria L, Hsieh B, Shah RJ. The peptide chains of human plasmin. Mechanism of activation of human plasminogen to plasmin. J Biol Chem. 1967; 242(10):2333-2342. 11. Camiolo SM, Thorsen S, Astrup T. Fibrinogenolysis and fibrinolysis with tissue plasminogen activator, urokinase, streptokinase-activated human globulin, and plasmin. Proc Soc Exp Biol Med. 1971; 138(1):277-280. 12. Stoppelli MP, Corti A, Soffientini A, Cassani G, Blasi F, Assoian RK. Differentiationenhanced binding of the amino-terminal fragment of human urokinase plasminogen activator to a specific receptor on U937 monocytes. Proc Natl Acad Sci U S A. 1985;

530

Linda Robertson for their invaluable technical assistance. We thank Dr Colin Longstaff, NIBSC, for invaluable advice on the kinetic assays and data analysis. We acknowledge the Microscopy and Histology Core Facility at the University of Aberdeen for excellent advice and use of the facilities. Funding This research was supported by grants FS/11/2/28579 (NJM) and PG/11/1/28461 (NJM, CSW) from the British Heart Foundation.

82(15):4939-4943. 13. Choi SH, Collins JN, Smith SA, DavisHarrison RL, Rienstra CM, Morrissey JH. Phosphoramidate end labeling of inorganic polyphosphates: facile manipulation of polyphosphate for investigating and modulating its biological activities. Biochemistry. 2010;49(45):9935-9941. 14. Mitchell JL, Lionikiene AS, Georgiev G, et al. Polyphosphate colocalizes with factor XII on platelet-bound fibrin and augments its plasminogen activator activity. Blood. 2016; 128(24):2834-2845. 15. Cockell CS, Marshall JM, Dawson KM, Cederholm-Williams SA, Ponting CP. Evidence that the conformation of unliganded human plasminogen is maintained via an intramolecular interaction between the lysine-binding site of kringle 5 and the N-terminal peptide. Biochem J. 1998;333 ( Pt 1):99-105. 16. Claeys H, Vermylen J. Physico-chemical and proenzyme properties of NH2-terminal glutamic acid and NH2-terminal lysine human plasminogen. Influence of 6-aminohexanoic acid. Biochim Biophys Acta. 1974; 342(2):351-359. 17. Thorsen S, Mullertz S. Rate of activation and electrophoretic mobility of unmodified and partially degraded plasminogen. Effects of 6aminohexanoic acid and related compounds. Scand J Clin Lab Invest. 1974; 34(2):167-176. 18. Engel R, Brain CM, Paget J, Lionikiene AS, Mutch NJ. Single-chain factor XII exhibits activity when complexed to polyphosphate. J Thromb Haemost. 2014;12(9):1513-1522. 19. Whyte CS, Chernysh IN, Domingues MM, et al. Polyphosphate delays fibrin polymerisation and alters the mechanical properties of the fibrin network. Thromb Haemost. 2016;116(5):897-903. 20. Smith SA, Morrissey JH. Polyphosphate enhances fibrin clot structure. Blood. 2008; 112(7):2810-2816. 21. van Zonneveld AJ, Veerman H, Pannekoek H. On the interaction of the finger and the kringle-2 domain of tissue-type plasminogen activator with fibrin. Inhibition of kringle-2 binding to fibrin by epsilon-amino caproic acid. J Biol Chem. 1986; 261(30):14214-14218. 22. Longstaff C, Thelwell C, Williams SC, Silva MM, Szabo L, Kolev K. The interplay between tissue plasminogen activator domains and fibrin structures in the regulation of fibrinolysis: kinetic and microscopic studies. Blood. 2011;117(2):661-668. 23. de Vries C, Veerman H, Koornneef E, Pannekoek H. Tissue-type plasminogen activator and its substrate Glu-plasminogen share common binding sites in limited plasmin-digested fibrin. J Biol Chem. 1990; 265(23):13547-13552.

24. Medved L, Nieuwenhuizen W. Molecular mechanisms of initiation of fibrinolysis by fibrin. Thromb Haemost. 2003;89(3):409-419. 25. Smith SA, Mutch NJ, Baskar D, Rohloff P, Docampo R, Morrissey JH. Polyphosphate modulates blood coagulation and fibrinolysis. Proc Natl Acad Sci U S A. 2006; 103(4):903-908. 26. Verhoef JJ, Barendrecht AD, Nickel KF, et al. Polyphosphate nanoparticles on the platelet surface trigger contact system activation. Blood. 2017;129(12):1707-1717. 27 Dejouvencel T, Doeuvre L, Lacroix R, et al. Fibrinolytic cross-talk: a new mechanism for plasmin formation. Blood. 2010;115(10): 2048-2056. 28. Fredenburgh JC, Nesheim ME. Lys-plasminogen is a significant intermediate in the activation of Glu-plasminogen during fibrinolysis in vitro. J Biol Chem. 1992; 267(36):26150-26156. 29. Hoylaerts M, Rijken DC, Lijnen HR, Collen D. Kinetics of the activation of plasminogen by human tissue plasminogen activator. Role of fibrin. J Biol Chem. 1982;257(6): 2912-2919. 30. Wohl RC, Summaria L, Robbins KC. Kinetics of activation of human plasminogen by different activator species at pH 7.4 and 37 degrees C. J Biol Chem. 1980; 255(5):2005-2013. 31. Suenson E, Bjerrum P, Holm A, et al. The role of fragment X polymers in the fibrin enhancement of tissue plasminogen activator-catalyzed plasmin formation. J Biol Chem. 1990;265(36):22228-22237. 32. Nesheim M, Fredenburgh JC, Larsen GR. The dissociation constants and stoichiometries of the interactions of Lys-plasminogen and chloromethyl ketone derivatives of tissue plasminogen activator and the variant delta FEIX with intact fibrin. J Biol Chem. 1990;265(35):21541-21548. 33. Miles LA, Castellino FJ, Gong Y. Critical role for conversion of glu-plasminogen to Lysplasminogen for optimal stimulation of plasminogen activation on cell surfaces. Trends Cardiovasc Med. 2003;13(1):21-30. 34. Silva MM, Thelwell C, Williams SC, Longstaff C. Regulation of fibrinolysis by Cterminal lysines operates through plasminogen and plasmin but not tissue-type plasminogen activator. J Thromb Haemost. 2012;10(11):2354-2360. 35. Mondino A, Blasi F. uPA and uPAR in fibrinolysis, immunity and pathology. Trends Immunol. 2004;25(8):450-455. 36. Killewich LA, Macko RF, Cox K, et al. Regression of proximal deep venous thrombosis is associated with fibrinolytic enhancement. J Vasc Surg. 1997;26(5):861-868. 37. Singh I, Burnand KG, Collins M, et al. Failure of thrombus to resolve in urokinase-type plasminogen activator gene-knockout mice:

haematologica | 2021; 106(2)

PolyP enhances uPA-mediated plasminogen activation

rescue by normal bone marrow-derived cells. Circulation. 2003;107(6):869-875. 38. Gossage JA, Humphries J, Modarai B, Burnand KG, Smith A. Adenoviral urokinase-type plasminogen activator (uPA) gene transfer enhances venous thrombus resolution. J Vasc Surg. 2006;44(5):1085-1090. 39. Humphries J, Gossage JA, Modarai B, et al. Monocyte urokinase-type plasminogen activator up-regulation reduces thrombus size in a model of venous thrombosis. J Vasc Surg. 2009;50(5):1127-1134. 40. von Bruhl ML, Stark K, Steinhart A, et al.

haematologica | 2021; 106(2)

Monocytes, neutrophils, and platelets cooperate to initiate and propagate venous thrombosis in mice in vivo. J Exp Med. 2012; 209(4):819-835. 41. Mitchell JL, Lionikiene AS, Georgiev G, et al. Polyphosphate co-localizes with factor XII on platelet-bound fibrin and augments its plasminogen activator activity. Blood. 2016; 128(24):2834-2845. 42. Welsh JD, Stalker TJ, Voronov R, et al. A systems approach to hemostasis: 1. The interdependence of thrombus architecture and agonist movements in the gaps between

platelets. Blood. 2014;124(11):1808-1815. 43. Tomaiuolo M, Stalker TJ, Welsh JD, Diamond SL, Sinno T, Brass LF. A systems approach to hemostasis: 2. Computational analysis of molecular transport in the thrombus microenvironment. Blood. 2014; 124(11):1816-1823. 44. Stalker TJ, Welsh JD, Tomaiuolo M, et al. A systems approach to hemostasis: 3. Thrombus consolidation regulates intrathrombus solute transport and local thrombin activity. Blood. 2014; 124(11): 1824-1831.

531

ARTICLE Ferrata Storti Foundation

Haematologica 2021 Volume 106(2):532-542

Acute Lymphoblastic Leukemia

Clinical significance of soluble CADM1 as a novel marker for adult T-cell leukemia/lymphoma

Shingo Nakahata,1* Syahrul Chilmi,1* Ayako Nakatake,1 Kuniyo Sakamoto,1 Maki Yoshihama,1 Ichiro Nishikata,1 Yoshinori Ukai,2 Tadashi Matsuura,2 Takuro Kameda,3 Kotaro Shide,3 Yoko Kubuki,3 Tomonori Hidaka,3 Akira Kitanaka,4 Akihiko Ito,5 Shigeki Takemoto,6° Nobuaki Nakano,7 Masumichi Saito,8 Masako Iwanaga,9 Yasuko Sagara,10 Kosuke Mochida,11 Masahiro Amano,11 Kouichi Maeda,12 Eisaburo Sueoka,13 Akihiko Okayama,14 Atae Utsunomiya,7 Kazuya Shimoda,3 Toshiki Watanabe15 and Kazuhiro Morishita1

Division of Tumor and Cellular Biochemistry, Department of Medical Sciences, Faculty of Medicine, University of Miyazaki, Miyazaki; 2Perseus Proteomics Inc., Perseus Proteomics Inc., Tokyo; 3Division of Gastroenterology and Hematology, Department of Internal Medicine, Faculty of Medicine, University of Miyazaki, Miyazaki; 4Department of Laboratory Medicine, Kawasaki Medical School, Okayama; 5Department of Pathology, Kindai University School of Medicine, Osaka; 6National Hospital Organization Kumamoto Medical Center, Kumamoto; 7Department of Hematology, Imamura General Hospital, Kagoshima, 8Department of Safety Research on Blood and Biological Products, National Institute of Infectious Diseases, Tokyo; 9Department of Frontier Life Science, Nagasaki University Graduate School of Biomedical Sciences, Nagasaki; 10Japanese Red Cross Kyushu Block Blood Center, Fukuoka; 11Department of Dermatology, Faculty of Medicine, University of Miyazaki, Miyazaki; 12Internal Medicine, National Hospital Organization Miyakonojo Medical Center, Miyazaki; 13Department of Laboratory Medicine, Saga University Hospital and Department of Clinical Laboratory Medicine, Faculty of Medicine, Saga University, Saga; 14Department of Rheumatology, Infectious Diseases and Laboratory Medicine, University of Miyazaki, Miyazaki and 15Department of Computational Biology and Medical Sciences, Graduate School of Frontier Sciences, The University of Tokyo, Tokyo, Japan 1

*SN and CS contributed equally as co-first authors.

°Current address: JURAKU Internal Medicine Clinic, Kumamoto, Japan.

ABSTRACT

Correspondence: KAZUHIRO MORISHITA kmorishi@med.miyazaki-u.ac.jp Received: August 21, 2019. Accepted: February 7, 2020. Pre-published: February 13, 2020. https://doi.org/10.3324/haematol.2019.234096

532

dult T-cell leukemia/leukemia (ATLL) is an aggressive peripheral T-cell malignancy, caused by infection with the human T-cell leukemia virus type 1 (HTLV-1). We recently showed that the cell adhesion molecule 1 (CADM1), a member of the immunoglobulin superfamily, is specifically and consistently overexpressed in ATLL cells, and functions as a novel cell surface marker. In this study, we first show that a soluble form of CADM1 (sCADM1) is secreted from ATLL cells by mainly alternative splicing. After developing the Alpha linked immunosorbent assay (AlphaLISA) for sCADM1, we show that plasma sCADM1 concentrations gradually increased during disease progression from indolent to aggressive ATLL. Although other known biomarkers of tumor burden such as soluble interleukin-2 receptor α (sIL-2Rα) also increased with sCADM1 during ATLL progression, multivariate statistical analysis of biomarkers revealed that only plasma sCADM1 was selected as a specific biomarker for aggressive ATLL, suggesting that plasma sCADM1 may be a potential risk factor for aggressive ATLL. In addition, plasma sCADM1 is a useful marker for monitoring response to chemotherapy as well as for predicting relapse of ATLL. Furthermore, the change in sCADM1 concentration between indolent and aggressive type ATLL was more prominent than the change in the percentage of CD4+CADM1+ ATLL cells. As plasma sCADM1 values fell within normal ranges in HTLV-1-associated myelopathy/tropical spastic paraparesis (HAM/TSP) patients with higher levels of serum sIL-2Rα, the measurement of sCADM1 may become a useful tool to discriminate between ATLL and other inflammatory diseases, including HAM/TSP. haematologica | 2021; 106(2)

sCADM1 as a new diagnostic marker of ATLL

Introduction Adult T-cell leukemia/lymphoma (ATLL) is a refractory CD4+ T-cell malignancy associated with human T-cell leukemia virus type 1 (HTLV-1).1-3 It is estimated that 1520 million people are currently infected with HTLV-1 worldwide, and a high prevalence of HTLV-1 infection can be found in many parts of the world, including southwestern Japan, Melanesia, South America, sub-Saharan Africa, the Caribbean, Romania, central parts of Australia, and the Middle East. ATLL is classified into four subtypes – acute, lymphoma, chronic, and smoldering type. Patients with indolent ATLL (chronic or smoldering) have a better prognosis and watchful waiting or combined zidovudine (AZT) and interferon-α (IFN-α) therapy is standard treatment for indolent disease. Patients with aggressive forms (acute and lymphoma) have a very poor prognosis due to the intrinsic chemotherapy resistance of malignant cells. Allogeneic hematopoietic stem-cell transplantation (allo-HSCT), mogamulizumab, an anti-CC chemokine receptor 4 (CCR4) monoclonal antibody, or AZT/IFN therapy are important for the treatment of ATLL; however, the prognosis is unfavorable in many cases.4-7 Serum levels of interleukin-2 receptor α (sIL-2Rα, sCD25) are known to reflect tumor burden due to the high expression levels of IL-2R on ATLL cells.8 However, sIL2R levels are also increased during an inflammatory response related to HTLV-1-associated myelopathy/tropical spastic paraparesis (HAM/TSP),9 or graft-versus-host disease (GvHD)10,11 that is often seen after mogamulizumab and allo-HSCT treatment;12-14 consequently, ATLL reoccurrence and inflammatory responses may not be distinguishable. The elevation of sIL2R levels has also been reported in various hematologic and solid malignancies.15,16 Therefore, the development of a specific and reliable diagnostic marker for ATLL is of great clinical significance. CADM1 was originally isolated as a tumor suppressor gene in non-small cell lung cancer, and functions in the cell-cell adhesion of endothelial cells.17,18 We found that CADM1 was ectopically and highly expressed in HTLV-1infected T cells and ATLL cells, resulting in the enhanced adhesion of ATLL cells to promote the invasion of ATLL cells into various organs.19,20 As CADM1 is consistently and specifically expressed in HTLV-1-infected T cells and ATLL cells,21-23 but not in most of the non-ATLL lymphomas,21,24-26 CADM1 is now thought to be the best cell surface marker for ATLL. sCADM1 consists of the extracellular domain of CADM1, which is generated by alternative splicing of CADM1 pre-mRNA27,28 or shedding of CADM1 protein on the cell surface.29-31 We have previously shown that sCADM1 protein is detected in the serum of acute-type ATLL patients.21 In the present study, we developed a highly sensitive and efficient method for the measurement of sCADM1 using the Alpha linked immunosorbent assay (AlphaLISA) technology.32,33 sCADM1 levels were found to be increased in smoldering to acute type ATLL, which is highly correlated with various clinical parameters, including serum sIL-2α levels. Furthermore, sCADM1 levels correlated with the leukemic cell burden in ATLL patients during the course of chemotherapy treatments. These results suggest that sCADM1 can be a specific biomarker for ATLL, and that a measurement of sCADM1 may become a useful tool for accurately predicting leukemic haematologica | 2021; 106(2)

cell burden and the disease progression or status of ATLL patients.

Methods Patient samples Peripheral blood samples were collected from HTLV-1 carriers (n=94), patients with smoldering-type (n=80), chronic-type (n=70), acute-type (n=71), and lymphoma-type (n=37) ATLL, patients with HAM/TSP (n=12), and healthy volunteers as controls (n=35) (Online Supplementary Table S1). These samples were obtained from Miyazaki University Hospital (Miyazaki, Japan), National Hospital Organization Miyakonojo Medical Center (Miyazaki, Japan), Imamura General Hospital (Kagoshima, Japan), National Hospital Organization Kumamoto Medical Center (Kumamoto, Japan), and the Group of Joint Study on Predisposing Factors of ATL Development (JSPFAD, Japan). Informed consent was obtained from all patients. This study was approved by the Institutional Review Board at the Faculty of Medicine, University of Miyazaki, in accordance with the Declaration of Helsinki. Diagnosis of ATLL was based on clinical features, hematological characteristics, serum antibodies against HTLV-1, and monoclonal integration of the HTLV-1 proviral genome. Plasma and serum samples were prepared by centrifugation and stored at -80°C until use. Peripheral blood mononuclear cells (PBMC) were isolated by Histopaque density gradient centrifugation (Sigma-Aldrich, Tokyo, Japan) according to the manufacturer’s protocol. The procedures for purification of ATLL cells from patients by using antiCADM1-antibody-coated magnetic beads have been described elsewhere.21 CD4+ T cells from healthy volunteers were purified from PBMC using microbeads (Miltenyi Biotec, Auburn, CA, USA). Purity of isolated CD4+ CADM1+ cell populations from ATLL patients was confirmed by flow cytometry.21

Cell lines The HTLV-1-negative human T-cell acute lymphoblastic leukemia (T-ALL) cell line MOLT4, the cutaneous T-cell lymphoma (CTCL) cell line HUT78, the HTLV-1-infected T-cell line MT2, and ATLL-derived cell lines (S1T and ST1) were maintained in RPMI 1640 medium (Wako, Osaka, Japan) supplemented with 10% fetal bovine serum (FBS) and 50 mg/mL of penicillin/streptomycin in a 5% CO chamber at 37°C. The IL-2-dependent ATLLderived cell lines KK1 and KOB were maintained in complete RPMI 1640 medium supplemented with 0.75 g/mL of recombinant human IL-2 (Peprotech, Rocky Hill, NJ, USA). The human osteosarcoma cell line Saos-2 was cultured in Dulbecco’s modified Eagle’s medium (DMEM, Wako) supplemented with 10% FBS and 50 µg/mL of penicillin/streptomycin. MOLT4 was obtained from the Fujisaki Cell Center, Hayashibara Biochemical Laboratories (Okayama, Japan). MT2 was kindly provided by Dr. H. Iha (Oita University, Japan). ST1, KOB, and KK1 were kindly provided by Dr. Y. Yamada (Nagasaki University, Japan). S1T was a kind gift from Dr. N. Arima (Kagoshima University, Japan). Saos-2 was obtained from the RIKEN BioResource Center (Tsukuba, Japan). 2

Measurement of sCADM1 concentrations in the blood by AlphaLISA The anti-CADM1 antibody (103-109) was generated by phagedisplay technology.34 AlphaLISA was performed in 96-well microtiter plates containing 5 mL of plasma sample, 10 mL of biotinylated anti-CADM1 antibody (3E1, MBL, Nagoya, Japan) (0.1 nM), 10 L of 10 mg/mL anti-CADM1 antibody (103-109) -conjugated AlphaLISA acceptor beads (10 mg/mL), and 25 mL of streptavidin-coated AlphaLISA donor beads (40 mg/mL) in AlphaLISA 533

S. Nakahata et al.

ImmunoAssay Buffer (25 mM HEPES, pH 7.4, 0.1% Casein, 0.5% Triton X-100, 1 mg/mL Dextran-500, and 0.05% Proclin-300). The reaction mixture was incubated at 23ยบC for 1 hour (h) and the chemical emission was read on an EnSpire Alpha microplate reader (PerkinElmer Waltham, MA). The concentrations of sCADM1 were obtained from a standard curve generated by recombinant CADM1 protein (Sino Biological, Beijing, China) using four-parameter logistic curve fitting, and multiplied by the dilution factor. All measurements were done in duplicate. Additional methods are provided in the Online Supplementary Appendix.

Results The soluble form of CADM1 mRNA is efficiently expressed in ATLL As sCADM1 can be generated through alternative splicing by an intron retention event27,28 or ectodomain shedding of CADM1 protein,29-31 the mRNA expression of a soluble splice variant of CADM1 and the level of CADM1 shedding were initially determined in ATLL-related cell lines and leukemia cells from patients with acute-type ATLL by RT-PCR and western blotting, respectively. Because the sCADM1 transcript contains the 118-bp sequence of intron 7 (Figure 1A),27 we used PCR primers located in exon 6 and intron 7 of CADM1 to amplify the sCADM1 transcript. Semiquantitative RT-PCR showed that sCADM1 mRNA, along with the membrane-bound form of CADM1 (mbCADM1) mRNA, was highly expressed in all of the HTLV-1-positive ATLL-related cell lines (Online Supplementary Figure S1A). We then performed quantitative RT-PCR analysis with purified CD4+CADM1+ leukemic cell populations from various types of ATLL patients.21 A significantly higher abundance of sCADM1 mRNA was observed in all ATLL samples tested, including smoldering, chronic, and acute-type ATLL, compared with CD4+ T cells from healthy volunteers (Figure 1B). Similarly, the levels of mbCADM1 or total CADM1 (tCADM1) were significantly higher in all subtypes of ATLL than controls (Figure 1B). In order to quantify CADM1 shedding in ATLL, we analyzed the levels of two membrane-associated C-terminal fragments (ฮฑCTF, 18 kDa and CTF, 35 kDa), which are generated by the proteolytic cleavage of CADM1 by ADAM and unidentified proteases, respectively.29-31 Although the CTF fragment of CADM1 was detected in ATLL-related cell lines, very weak or no signal was found in leukemia cell samples from ATLL patients (Online Supplementary Figure S2A-B). Additionally, CADM1 undergoes alternative splicing between exons 7 and 11 (Figure 1A) and the inclusion of exon 9 confers shedding susceptibility to CADM1.31 RTPCR analysis revealed that the majority of the mbCADM1 mRNA skipped both exons 9 and 10 (Online Supplementary Figure S1A-B). Therefore, these results suggest that upregulated expression of sCADM1 mRNA, which is generated by retention of intron 7, might be the main cause of the increased level of sCADM1 protein in ATLL.

Plasma sCADM1 levels increase with disease progression in ATLL In order to investigate the clinical utility of sCADM1 in the diagnosis of ATLL patients, we developed a highly sensitive method for the measurement of plasma sCADM1 with the use of the AlphaLISA technology, which is based 534

on energy transfer from a streptavidin donor bead to an AlphaLISA acceptor bead in close proximity.32,33 The donor and the acceptor beads were conjugated with biotinylated anti-CADM1 (3E1) and anti-CADM1 (103-189) antibodies, respectively (see Methods). Using recombinant human CADM1 protein, the assay system was capable of detecting sCADM1 in the plasma at concentrations as low as ~0.2 ng/mL (Online Supplementary Figure S3). First, we determined the sCADM1 levels in the peripheral blood of healthy volunteers, HTLV-1 carriers, and patients with HAM/TSP and various types of ATLL who had not been previously treated. The median values for plasma and serum sCADM1 from healthy volunteers were 181.3 ng/mL and 173.5 ng/mL with 5th-95th percentile ranges of 142.7-234.0 ng/mL and 131.7-215.5 ng/mL, respectively (Online Supplementary Figure S4A). There were neither a significant differences between the sCADM1 levels of males and females, nor amongst the age groups (Online Supplementary Figure S4B-C). In addition, the majority of HTLV-1 carriers and HAM/TSP patients had sCADM1 levels within the standard reference range (Figure 2). The median plasma sCADM1 concentration was slightly higher in smoldering-type ATLL patients (210.2 ng/mL) than in healthy volunteers (181.3 ng/mL), and it was elevated in chronic-type ATLL patients, with a median peak value of 267.1 ng/mL (Figure 2). The median plasma sCADM1 level was 1055.0 ng/mL (range: 173.6-6,931.0 ng/mL) in acutetype ATLL patients, more than 5-fold higher than in the control group (Figure 2 and Online Supplementary Figure S5A). In addition, 2 of 10 patients with the unfavorable chronic-type ATLL35 had high levels of plasma sCADM1 (Online Supplementary Figure S6). Next, we determined the association between plasma sCADM1 concentrations and other clinical and/or biochemical parameters in ATLL patients (Online Supplementary Figure S5A-E). Among these subjects, serum concentrations of sIL2R were higher in patients with all subtypes of ATLL compared to healthy volunteers, and concentrations increased with disease progression to acute-type or lymphoma-type ATLL (Online Supplementary Figure S5B). Importantly, while the patients with HAM/TSP showed significantly higher sIL2R levels compared with healthy volunteers, as reported previously,9 plasma sCADM1 levels were in the healthy range (Online Supplementary Figure S5A-B), confirming that sCADM1 is a specific marker for ATLL. Most patients with acute-type or lymphoma-type ATLL had elevated serum lactate dehydrogenase (LDH) levels (Online Supplementary Figure S5C). HTLV-1 proviral load (PVL) in the peripheral blood increased in smoldering-type ATLL patients and was elevated in chronic-type and acute-type ATLL, and was similar to white blood cell (WBC) concentrations (Online Supplementary Figure S5D-E). Serum sIL2R levels strongly correlated with plasma sCADM1 levels (r=0.66, P<0.001) (Figure 3A), and moderate correlations were found between plasma sCADM1 and LDH levels (r=0.52, P<0.001) (Figure 3B) as well as plasma sCADM1 and PVL levels in the peripheral blood (r=0.47, P<0.001) (Figure 3C). The correlation between plasma sCADM1 and PVL levels was higher than that between plasma sCADM1 and the levels of oligoclonality of HTLV-1-infected clones (Online Supplementary Figure S7A-B). There was a weak correlation between WBC counts and plasma sCADM1 levels (r=0.37, P<0.001) (Figure 3D). Additionally, the increase in plasma concentrations of sCADM1 was much higher than the haematologica | 2021; 106(2)

sCADM1 as a new diagnostic marker of ATLL

Figure 1. High expression of sCADM1 transcript in adult T-cell leukemia/lymphoma. (A) Schematic representation of the structure and domain organization of membrane-bound cell adhesion molecule 1 CADM1 (mbCADM1) and its soluble form (sCADM1). mbCADM1 contains an extracellular domain (exons 2-10) containing three immunoglobulin-like C2-type domains, a transmembrane domain (exon 11), and a cytoplasmic domain (exon 12). sCADM1 is generated by alternative splicing in which intron 7 is retained and an in-frame stop codon is found immediately downstream of the immunoglobulin-like domain. Arrows indicate the position and direction of primers used in RT-PCR reactions. The mbCADM1, sCADM1, and total CADM1 (tCADM1) transcripts were amplified using primers for exon 7 and exon 8, exon 6 and intron 7, and exon 4 and exon 5, respectively. Numbers indicate exon numbers of the CADM1 gene. (B) Quantitative RT-PCR analysis of tCADM1, mbCADM1, and sCADM1 expression in CD4+ T lymphocytes from healthy volunteers (n=5) and sorted CADM1+ cells from peripheral blood of patients with smoldering-type (n=6), chronic-type (n=6), and acute-type (n=7) ATLL. The data were normalized to Î˛-actin gene expression and expressed relative to a healthy control sample. Mean Âą standard deviation is shown, **P<0.01 (Mann-Whitney U test).

haematologica | 2021; 106(2)

535

S. Nakahata et al. increases in the percentages of CD4+CADM1+ T cells, WBC concentrations, or serum sIL2R levels in acute-type ATLL patients compared to chronic-type ATLL patients (Online Supplementary Figure S8A-D). Notably, a strong positive correlation was found between the absolute number of circulating CD4+CADM1+ T cells and sCADM1 concentration in the peripheral blood of chronic-type and acutetype ATLL patients (Online Supplementary Figure S9A-B). Given that sCADM1 expression did not significantly differ between chronic-type and acute-type ATLL (Figure 1B), the drastic increase in sCADM1 levels in acute-type ATLL is possibly ascribed to an increase in numbers of circulating tumor cells. Additionally, univariate and multivariate logistic regression analyses were performed to assess factors associated with aggressive ATLL compared with indolent ATLL. In the univariate analysis, the risk factors associated with aggressive type ATLL were sCADM1, sIL2R , LDH,

and WBC (Table 1). On the other hand, in multivariate analysis, sCADM1 was the only risk factor associated with aggressive ATLL (Table 1). Furthermore, the Kaplan-Meier analysis in aggressive type ATLL patients showed a trend towards worse overall survival for patients with high sCADM1 (Online Supplementary Figure S10A-D, Online Supplementary Table S2). Taken together, these results suggest that sCADM1 has potential as a novel diagnostic biomarker in ATLL, and that it may be useful for monitoring disease progression.

The plasma sCADM1 level is a useful clinical indicator of the leukemic cell burden in ATLL patients In order to investigate the clinical value of plasma sCADM1 concentration, we evaluated the levels of plasma sCADM1 along with sIL2R , LDH, PVL, and WBC in the peripheral blood of patients with various types of ATLL

Table 1. Univariate and multivariate analysis of factors predicting aggressiveness of adult T-cell leukemia/lymphoma.

Variable

Univariate analysis 95% CI

Multivariate analysis 95% CI

sCADM1 sIL2Rα LDH WBC PVL* Age

5.46 1.85 6.93 1.72 1.14 1.01

3.13-9.54 1.49-2.28 3.48-13.8 1.22-2.44 0.94-1.38 0.98-1.04

<0.0001 <0.0001 <0.0001 0.002 0.158 0.385

4.33 1.22 2.22 0.617 0.855 1

2.06-9.10 0.895-1.67 0.944-5.21 0.349-1.09 0.611-1.2 0.957-1.05

<0.0001 0.205 0.068 0.096 0.359 0.963

For logistic regression analysis, numerical values of soluble for of cell adhesion molecule 1 (sCADM1), serum interleukin-2 receptor α (sIL2Rα ), lactate dehydrogenase (LDH), white blood cell (WBC), and proviral load (PVL) were converted to base 2 logarithm transformation. The data from 49 aggressive type (40 acute-type and nine lymphoma-type) and 83 indolent type adult T-cell leukemia/lymphoma (ATLL) cases (18 chronic-type and 65 smoldering-type) (Online Supplementary Figure S5) were used in the analysis. *Fortytwo cases with missing values on PVL (eight acute-type, eight chronic-type, and 17 smoldering-type) and nine lymphoma type ATLL were excluded.

Figure 2. Plasma sCADM1 levels in adult T-cell leukemia/lymphoma patients examined by AlphaLISA. The plasma soluble form of cell adhesion molecule 1 (sCADM1) levels in 34 healthy volunteers, 78 human T-cell leukemia virus type 1 (HTLV-1) carriers, 12 HTLV-1-associated myelopathy/tropical spastic paraparesis (HAM/TSP) patients, 77 smoldering-type, 23 chronic-type, 13 lymphoma-type, and 43 acute-type ATLL patients who were previously untreated were measured by AlphaLISA using anti-CADM1 antibodies. The box and whisker plots show the 5th, 25th, 50th (median), 75th, and 95th percentile values, with outliers marked by solid dots. *P<0.05, **P<0.01, ***P<0.001 versus healthy volunteers or HTLV-1 carriers (Kruskal-Wallis test/Dunn's multiple comparison test). th th Median and 5 and 95 percentile values are indicated at the top of each column. The dot line indicates the 95th percentile of plasma sCADM1 in healthy subjects.

536

haematologica | 2021; 106(2)

sCADM1 as a new diagnostic marker of ATLL

before and after treatment with chemotherapy (Figure 4). Although the pre- and post-treatment samples were not derived from the same patients, median plasma sCADM1 levels were significantly decreased in the peripheral blood

of patients with acute and lymphoma-type ATLL after chemotherapy compared to before treatment (Figure 4A). Additionally, the serum sIL2R levels in acute and lymphoma-type ATLL patients after chemothe-rapy were sig-

Table 2. Receiver operating characteristic (ROC) curve analysis of plasma sCADM1 concentrations. ATLL (acute/lymphoma type) versus healthy volunteers ATLL (chronic type) versus healthy volunteers ATLL (smoldering type) versus healthy volunteers

Marker

Cut-off Point

Sensitivity

Specificity

AUC

sCADM1 (ng/mL) WBC (×100/mL) sIL2Rα (U/mL) sCADM1 (ng/mL) WBC (×100/mL) sIL2Rα (U/mL) sCADM1 (ng/mL) WBC (×100/mL) sIL2Rα (U/mL)

232.4 92.5 494.5 233.4 76.5 470.0 214.6 62.5 436.5

0.873 0.630 1.0 0.609 1.0 1.0 0.481 0.627 0.955

0.971 0.969 1.0 0.971 0.875 1.0 0.912 0.719 1.0

0.937 0.786 1.0 0.739 0.981 1.0 0.657 0.692 0.986

P<0.0001 P=0.0002 P<0.0001 P=0.0024 P<0.0001 P<0.0001 P=0.0082 P=0.0020 P<0.0001

The Youden index was used to determine cut-off values. ATLL: adult T-cell leukemia/lymphoma; AUC: area under the ROC curve; SCADM1: soluble form of cell adhesion molecule 1; WBC: white blood cell; sIL2Rα: serum interleukin-2 receptor α.

Figure 3. Association between plasma sCADM1 levels and various prognostic indicators in adult T-cell leukemia/lymphoma. (A) Scatter plot of plasma soluable form of cell adhesion molecule 1 (sCADM1) concentrations versus serum interleukin-2 receptor α (sIL2Rα) concentrations in 34 healthy volunteers, 78 human T-cell leukemia virus type 1 (HTLV-1) carriers, 77 smoldering-type, 55 chronic-type, 34 lymphoma-type, and 69 acute-type adult T-cell leukemia/lymphoma (ATLL) patients, including those who were previously untreated and treated with chemotherapy. Spearman correlation coefficient values (r) and P-values are shown on each of the graphs. (B) Scatter plot of plasma sCADM1 concentrations versus serum lactate dehydrogenase (LDH) concentrations in 45 HTLV-1 carriers, 69 smolderingtype, 38 chronic-type, 18 lymphoma-type, and 60 acute-type ATLL patients, including those who were previously untreated and treated with chemotherapy. Spearman correlation coefficient values (r) and P-values are shown on each of the graphs. (C) Scatter plot of plasma sCADM1 concentrations versus proviral load (PVL) in 66 HTLV-1 carriers, 56 smoldering-type, 56 chronic-type, and 61 acute-type ATLL patients, including those who were previously untreated and treated with chemotherapy. Spearman correlation coefficient values (r) and P-values are shown on each of the graphs. (D) Scatter plot of plasma sCADM1 concentrations versus white blood cell (WBC) counts in 32 healthy volunteers, 71 HTLV-1 carriers, 74 smoldering-type, 57 chronic-type, 33 lymphoma-type, and 69 acute-type ATLL patients, including those who were previously untreated and treated with chemotherapy. Spearman correlation coefficient values (r) and P-values are shown on each of the graphs.

haematologica | 2021; 106(2)

537

S. Nakahata et al.

Figure 4. Plasma sCADM1 levels are related to treatment efficacy in adult T-cell leukemia/lymphoma patients. (A) Soluble form of cell adhesion molecule 1 (sCADM1) concentrations in the plasma of 23 chronic-type, 13 lymphoma-type, and 43 acute-type adult T-cell leukemia/lymphoma (ATLL) patients who were previously untreated and of 39 chronic-type, 18 lymphoma-type, and 7 acute-type ATLL patients who were previously treated with chemotherapy were measured by AlphaLISA. *P<0.05, ***P<0.001 (Mann-Whitney U test). The dot line indicates the 95th percentile of plasma sCADM1 in healthy subjects. Note that the pre-treatment and post-treatment samples were derived from different individuals. (B) Serum interleukin-2 receptor α (sIL2Rα) concentrations in the serum of 19 chronictype, 12 lymphoma-type, and 43 acute-type ATLL patients who were previously untreated and of 32 chronic-type, 16 lymphoma-type, and six acute-type ATLL patients who were previously treated with chemotherapy. *P<0.05, ***P<0.001 (Mann-Whitney U test). The dot line indicates the upper limit of normal serum sIL2Rα. The same samples as Figure 4A were used. Fifteen cases with missing values on sIL2Rα (one lymphoma-type and four chronic-type in the pre-treatment group and one acute-type, two lymphoma-type, and seven chronic-type in the post-treatment group) were excluded. (C) Lactate dehydrogenase (LDH) concentrations in the serum of 18 chronic-type, nine lymphoma-type, and 40 acute-type ATLL patients who were previously untreated and of 15 chronic-type, seven lymphoma-type, and five acute-type ATLL patients who were previously treated with chemotherapy. *P<0.05 (Mann-Whitney U test). The dot line indicates the upper limit of normal serum LDH. The same samples as Figure 4A were used. Forty-nine cases with missing values on LDH (three acute-type, four lymphoma-type, and five chronic-type in the pre-treatment group and two acute-type, 11 lymphoma-type, and 24 chronic-type in the post-treatment group) were excluded. (D) Human T-cell leukemia virus type 1 (HTLV-1) proviral load (PVL) in 15 chronic-type, 11 lymphoma-type, and 35 acute-type ATLL patients who were previously untreated and in 39 chronic-type, 18 lymphoma-type, and 7 acute-type ATLL patients who were previously treated with chemotherapy. **P<0.01 (Mann-Whitney U test). The same samples as Figure 4A were used. Eighteen cases with missing values on PVL (eight acute-type, two lymphoma-type, and eight chronic-type in the pre-treatment group) were excluded. (E) White blood cell counts (WBC) counts in 19 chronic-type, 11 lymphoma-type, and 43 acute-type ATLL patients who were previously untreated and in 33 chronic-type, 16 lymphoma-type, and 6 acute-type ATLL patients who were previously treated with chemotherapy. ***P<0.001 (Mann-Whitney U test). The dot line indicates the upper limit of normal for WBC counts. The same samples as Figure 4A were used. Thirteen cases with missing values on WBC counts (two lymphoma-type and four chronic-type in the pre-treatment group and one acute-type, two lymphoma-type, and four chronic-type in the post-treatment group) were excluded.

538

haematologica | 2021; 106(2)

sCADM1 as a new diagnostic marker of ATLL

nificantly lower than before treatment (Figure 4B). A significant reduction in serum LDH, PVL, and WBC after chemotherapy was also observed in acute-type ATLL patients (Figure 4C-E). Plasma sCADM1 levels were examined in three patients with acute-type ATLL who received

allo-HSCT (Online Supplementary Table S3). Although all three patients exhibited elevated levels of serum sIL2Rα, 2 of the 3 patients showed normal (case 3) or slightly high levels (case 1) of plasma sCADM1. Notably, in one patient (case 2) with elevated levels of both sIL2R and sCADM1,

Figure 5. Plasma sCADM1 is useful for monitoring disease progression in adult T-cell leukemia/lymphoma patients. Plasma soluble form of cell adhesion molecule 1 (sCADM1) concentrations along with various prognostic parameters including serum interleukin-2 receptor α (sIL2Rα ), lactate dehydrogenase (LDH), and white blood count (WBC) count were monitored in six patients with adult T-cell leukemia/lymphoma (ATLL) before and after acute crisis (day 0). The drug regimen is indicated at the top of the figure: VP-16, etoposide; LSG15, VCAP-AMP-VECP (vincristine, cyclophosphamide, doxorubicin, and prednisone [VCAP], doxorubicin, ranimustine, and prednisone [AMP], and vindesine, etoposide, carboplatin, and prednisone [VECP]); PSL: prednisolone; KW-0761: mogamulizumab; THP-COP: pirarubicin, cyclophosphamide, vincristine, and prednisolone.

haematologica | 2021; 106(2)

539

S. Nakahata et al.

recurrence of ATLL occurred shortly after allo-HSCT. These results suggest that mea-surement of plasma sCADM1 levels may be useful in monitoring response to chemotherapy as well as in predicting recurrence of ATLL after allo-HSCT. In order to further evaluate the relationship between plasma sCADM1 concentration and treatment response or the leukemic cell burden during the clinical course of ATLL patients, six ATLL patients were followed over a period of 45 to 1,219 days during treatment to assess changes in sCADM1 levels along with WBC, LDH, and sIL2R in the peripheral blood (Online Supplementary Table S4, Figure 5). In all of the patients except one (Pt #6), the levels of sCADM1 and sIL2R showed similar changes during the observation period, and importantly, increases in sCADM1 appears to correlate with poor outcome (Pt #1 and #3). While changes in LDH levels were related to changes in levels of sCADM1 in four patients (Pt #1, #2, #3, and #5), WBC counts did not appear to correlate with the levels of sCADM1 and sIL2R (Figure 5). Moreover, the sCADM1 level was also evaluated during the asymptomatic period and ATLL disease progression in each patient. In most cases, changes in the levels of sCADM1 from the asymptomatic state to ATLL stages showed a similar tendency to those observed in PVL (Online Supplementary Figure S11-12). In one case (MK1006), we observed a marked increase of sCADM1 from the smoldering to the lymphoma-type, which coincided with an increase in sIL2R (from 2,080 U/mL to 36,300 U/mL), whereas PVL was not increased in the peripheral blood (Online Supplementary Figure S12). As a few HTLV-1 carriers had an elevated plasma sCADM1 concentration compared to healthy subjects (Figure 2), we assessed whether sCADM1 levels are a predictive marker for the development of ATLL. Because PVL>4% is a known predictor for the development of ATLL,36 we compared the sCADM1 concentrations of asymptomatic HTLV-1 carriers with PVL>4% and PVL<4%, along with those who later progressed to ATLL (Online Supplementary Figure S13). There were no differences in plasma sCADM1 levels between the high and low PVL groups and no association was found between sCADM1 level and PVL in the univariate analysis (Online Supplementary Figure S13, Online Supplementary Table S5), although a slightly higher median sCADM1 concentration was observed in HTLV-1 carriers who later developed ATLL compared to the other groups (Online Supplementary Figure S13). Thus, these data suggest that while measurement of plasma sCADM1 alone may not be used to predict the progression of HTLV-1 carriers to indolent ATLL, it may be helpful for monitoring disease progression to an aggressive state, leukemic cell burden, and/or treatment response in ATLL patients. Finally, in order to assess the diagnostic efficiency of the measurement of plasma sCADM1, we examined receiver operating characteristics (ROC) in patients with different subtypes of ATLL and in healthy controls. The area under the ROC curve of sCADM1 was 0.94 for acute and lymphoma-type, 0.74 for chronic-type, and 0.66 for smoldering-type ATLL (Table 2), suggesting that sCADM1 is a promising functional biomarker for the diagnosis of aggressive ATLL.

Discussion In this study, we used AlphaLISA technology to measure sCADM1 concentrations in the peripheral blood. 540

Although ELISA is a well-known technique to measure the plasma concentration of target proteins, we found that detection sensitivity of plasma sCADM1 in our AlphaLISA system was enhanced ~10-fold compared to a conventional ELISA method (data not shown). By using this system, we successfully determined the range of plasma sCADM1 concentrations in healthy subjects and cut-off points for plasma sCADM1 between healthy and ATLL patients. The disease progression in ATLL patients may involve the accumulation of genetic alterations and clonal expansion of ATLL cells. Although recent studies have identified somatic mutations that act as drivers during ATLL progression,37,38 the application of this genetic information towards diagnosing patients is not easy, because the genomic abnormalities of ATLL cells are very complicated. In addition, clonality assessment is currently high-cost and time-consuming. Therefore, the development of effective methods for predicting disease progression is urgently needed. CADM1 is transcriptionally upregulated in HTLV-1infected T cells and ATLL cells in almost all cases,19,21,39,40 and in this study we found that the same CADM1 promoter appears to activate the expression of both sCADM1 and the membrane-bound form of CADM1 transcripts in ATLL. On the other hand, plasma sCADM1 levels were drastically increased in acute stage ATLL compared to chronic stage ATLL (median 1,066.7 ng/mL vs. 204.0 ng/mL, respectively), whereas the median percentages of CD4+CADM1+ T cells were 78.8% and 43.1% in acute and chronic-type ATLL, respectively (Online Supplementary Figure S8). Moreover, we observed increases in plasma sCADM1 levels during follow-ups with patients, which were accompanied by poor patient outcomes, whereas WBC counts were not changed (Figure 5). As multivariate analysis of several blood biomarkers for ATLL identified sCADM1 as the only independent serum marker for aggressive ATLL with significant differences (Table 1), plasma sCADM1 may be a potential risk factor of the aggressiveness of ATLL cells. In addition, while sIL2Rα is secreted not only by leukemia cells but also during various inflammatory responses,15,41 plasma sCADM1 was not increased in HAM/TSP patients or in a patient with GvHD after allo-HSCT for ATLL. Therefore, combined sCADM1 and sIL2Rα measurements may become a promising method for more accurately diagnosing ATLL development in HAM/TSP patients, or discriminating ATLL relapse and GvHD following allo-HSCT for ATLL. In the current clinical setting, assessment of treatment efficacy in ATLL is based on serum biochemical parameters such as sIL2Rα, LDH, and calcium levels, and clinical findings including computed tomography and positron emission tomography examination results, however, these tests are unable to determine the depth of response to treatment, which can be evaluated by detection of MRD. Given that sCADM1 seems to be a specific biomarker for ATLL, the measurement of plasma sCADM1 may be a useful in measuring the depth of response to therapy, and a change in the sCADM1 plasma level may become an important cli-nical criteria for not only determining the transplant adap-tability, but also determining the choice of consolidation or maintenance treatments and their periods. It has been reported that sCADM1 can bind to the extracellular domain of CADM1 through an interaction between their Ig domains.27 CADM1 contains a PSD95/Dlg/ZO-1 (PDZ) domain that interacts with T lymphoma invasion haematologica | 2021; 106(2)

sCADM1 as a new diagnostic marker of ATLL

and metastasis 1 (TIAM1), and induces signaling to actin filaments, leading to cell structure remodeling.42 sCADM1 may activate this TIAM1-actin signaling to modulate cellular functions. It has also been reported that CADM1 expression on tumor cells enhances the cytotoxic activities of cytotoxic T cells or natural killer cells via an interaction with Class I MHC-restricted T-cell-associated molecule (CRTAM).40,43 The CRTAM-CADM1 interaction acts as a costimulator of T-cell receptor (TCR) signaling that may help to eliminate tumor cells. Because sCADM1 has been reported to bind to the extracellular domain of CADM1 on the cell surface and inhibit CADM1 homophilic interactions,27 it is possible that sCADM1 may modulate immune responses under certain conditions. Future studies of the biological role of sCADM1 in anti-tumor immune responses will be important to obtain a better understanding of the molecular mechanisms of the immunological abnormalities in ATLL.44 To date, there has been no report identifying sCADM1 in cancers; however, it has been reported that CADM1 shedding was elevated in the lung epithelial cells of patients with idiopathic interstitial pneumonias45 or pulmonary emphysema,46 as well as in the pancreatic islets of patients with type 2 diabetes mellitus,47 which was accompanied by a decrease in full-length CADM1 levels and may be associated with the disruption of cell polarity and cell apoptosis.45,46 In ATLL, plasma sCADM1 levels drastically increased from chronic to acute-type ATLL with no apparent reduction in the expression of full-length CADM1, suggesting that sCADM1 may play a role in the malignant progression of ATLL. In summary, sCADM1 is a promising biomarker not

References 1. Uchiyama T, Yodoi J, Sagawa K, Takatsuki K, Uchino H. Adult T-cell leukemia: clinical and hematologic features of 16 cases. Blood. 1977;50(3):481-492. 2. Poiesz BJ, Ruscetti FW, Gazdar AF, Bunn PA, Minna JD, Gallo RC. Detection and isolation of type C retrovirus particles from fresh and cultured lymphocytes of a patient with cutaneous T-cell lymphoma. Proc Natl Acad Sci USA. 1980;77(12):7415-7419. 3. Bangham CRM, Matsuoka M. Human T-cell leukaemia virus type 1: parasitism and pathogenesis. Philos Trans R Soc Lond B Biol Sci. 2017;372:1732. 4. Watanabe T. Current status of HTLV-1 infection. Int J Hematol. 2011;94(5):430-434. 5. Utsunomiya A, Choi I, Chihara D, Seto M. Recent advances in the treatment of adult Tcell leukemia-lymphomas. Cancer Sci. 2015;106(4):344-351. 6. Mehta-Shah N, Ratner L, Horwitz SM. Adult T-cell leukemia/lymphoma. J Oncol Pract. 2017;13(8):487-492. 7. Phillips AA, Harewood JCK. Adult T cell leukemia-lymphoma (ATL): state of the art. Curr Hematol Malig Rep. 2018;13(4):300307. 8. Kamihira S, Atogami S, Sohda H, Momita S, Yamada Y, Tomonaga M. Significance of soluble interleukin-2 receptor levels for evaluation of the progression of adult T-cell leukemia. Cancer. 1994;73(11):2753-2758. 9. Toledo-Cornell C, Santos S, Orge G, Glesby MJ, Carvalho EM. Soluble IL-2 receptor and

haematologica | 2021; 106(2)

only for monitoring the leukemic burden of ATLL patients, but also for predicting disease status. Thus, sCADM1 measurement may be valuable for the diagnosis of ATLL. Disclosures No conflicts of interests to disclose. Contributions SN performed research, analyzed data and wrote the manuscript; CS, AN, KSa, MY, IN, and MS performed research and analyzed data; TK, KSh, YK, TH, AK, ST, NN, MI, YS, KMo, MA, KMa, ES, AO, KSh, and TW provided clinical information and samples. YU, TM, and AI provided critical reagents; AU provided clinical information and samples, and critical evaluation of the manuscript; and KM conceived and designed the study, directed and supervised the research, and wrote the manuscript. Acknowledgments The authors are grateful to Dr. Hidekatsu Iha (Oita University, Japan), Dr. Yasuaki Yamada (Nagasaki University, Japan), and Dr. Naomichi Arima (Kagoshima University, Japan) for providing cell lines. Funding This work was supported in part by Grant-in-Aid for Scientific Research (B) (25293081 and 17H03581) (KM) from the Japanese Society for the Promotion of Science (JSPS), by the Platform of Supporting Cohort Study and Biospecimen Analysis, Grant-in-Aid for Scientific Research on Innovative Areas (16H06277) (TW), and by the Takeda Science Foundation (KM).

beta-2 microglobulin as possible serologic markers of neurologic disease in HTLV-1 infection. J Med Virol. 2014;86(2):315-321. 10. Nakamura Y, Tanaka Y, Tanaka M, et al. Soluble interleukin-2 receptor index predicts the development of acute graft-versus-host disease after allogeneic hematopoietic stem cell transplantation from unrelated donors. Int J Hematol. 2016;103(4):436-443. 11. Budde H, Papert S, Maas JH, et al. Prediction of graft-versus-host disease: a biomarker panel based on lymphocytes and cytokines. Ann Hematol. 2017;96(7):1127-1133. 12. Fuji S, Inoue Y, Utsunomiya A, et al. Pretransplantation anti-CCR4 antibody Mogamulizumab against adult T-cell leukemia/lymphoma is associated with significantly increased risks of severe and corticosteroid-refractory graft-versus-host disease, nonrelapse mortality, and overall mortality. J Clin Oncol. 2016;34(28):3426-3433. 13. Kawano N, Kuriyama T, Yoshida S, et al. The Impact of a humanized CCR4 antibody (Mogamulizumab) on patients with aggressive-type adult T-cell leukemia-lymphoma treated with allogeneic hematopoietic stem cell transplantation. J Clin Exp Hematop. 2017;56(3):135-144. 14. Cook LB, Fuji S, Hermine O, et al. Revised adult T-cell leukemia-lymphoma International Consensus Meeting Report. J Clin Oncol. 2019;37(8):677-687. 15. Murakami S. Soluble interleukin-2 receptor in cancer. Front Biosci. 2004;9:3085-3090. 16. Bien E, Balcerska A. Serum soluble interleukin 2 receptor alpha in human cancer of adults and children: a review. Biomarkers.

2008;13(1):1-26 17. Kuramochi M, Fukuhara H, Nobukuni T, et al. TSLC1 is a tumor-suppressor gene in human non-small-cell lung cancer. Nat Genet. 2001;27(4):427-430. 18. Pletcher MT, Nobukuni T, Fukuhara H, et al. Identification of tumor suppressor candidate genes by physical and sequence mapping of the TSLC1 region of human chromosome 11q23. Gene. 2001;273(2):181-189. 19. Sasaki H, Nishikata I, Shiraga T, et al. Overexpression of a cell adhesion molecule, TSLC1, as a possible molecular marker for acute-type adult T-cell leukemia. Blood. 2005;105(3):1204-1213. 20. Dewan MZ, Takamatsu N, Hidaka T, et al. Critical role for TSLC1 expression in the growth and organ infiltration of adult T-cell leukemia cells in vivo. J Virol. 2008; 82(23):11958-11963. 21. Nakahata S, Saito Y, Marutsuka K, et al. Clinical significance of CADM1/TSLC1/IgSF4 expression in adult T-cell leukemia/lymphoma. Leukemia. 2012;26(6):1238-1246. 22. Kobayashi S, Watanabe E, Ishigaki T, et al. Advanced human T-cell leukemia virus type 1 carriers and early-stage indolent adult Tcell leukemia-lymphoma are indistinguishable based on CADM1 positivity in flow cytometry. Cancer Sci. 2015;106(5):598-603. 23. Kobayashi S, Nakano K, Watanabe E, et al. CADM1 expression and stepwise downregulation of CD7 are closely associated with clonal expansion of HTLV-I-infected cells in adult T-cell leukemia/lymphoma. Clin Cancer Res. 2014;20(11):2851-2861.

541

S. Nakahata et al. 24. Paulsson K, An Q, Moorman AV, et al. Methylation of tumour suppressor gene promoters in the presence and absence of transcriptional silencing in high hyperdiploid acute lymphoblastic leukaemia. Br JHaematol. 2009;144(6):838-847. 25. Murray PG, Fan Y, Davies G, et al. Epigenetic silencing of a proapoptotic cell adhesion molecule, the immunoglobulin superfamily member IGSF4, by promoter CpG methylation protects Hodgkin lymphoma cells from apoptosis. Am J Pathol. 2010;177(3):1480-1490. 26. Fu L, Gao Z, Zhang X, et al. Frequent concomitant epigenetic silencing of the stressresponsive tumor suppressor gene CADM1, and its interacting partner DAL-1 in nasal NK/T-cell lymphoma. Int J Cancer. 2009; 124(7):1572-1578. 27. Koma Y, Ito A, Wakayama T, et al. Cloning of a soluble isoform of the SgIGSF adhesion molecule that binds the extracellular domain of the membrane-bound isoform. Oncogene. 2004;23(33):5687-5692. 28. Hagiyama M, Ichiyanagi N, Kimura KB, Murakami Y, Ito A. Expression of a soluble isoform of cell adhesion molecule 1 in the brain and its involvement in directional neurite outgrowth. Am J Pathol. 2009; 174(6):2278-2289. 29. Nagara Y, Hagiyama M, Hatano N, et al. Tumor suppressor cell adhesion molecule 1 (CADM1) is cleaved by adisintegrin and metalloprotease 10 (ADAM10) and subsequently cleaved by Îł-secretase complex. Biochem Biophys Res Commun. 2012; 417(1):462-467. 30. Moiseeva EP, Leyland ML, Bradding P. CADM1 is expressed as multiple alternatively spliced functional and dysfunctional isoforms in human mast cells. Mol Immunol. 2013;53(4):345-354. 31. Shirakabe K, Omura T, Shibagaki Y, et al. Mechanistic insights into ectodomain shed-

542

ding: susceptibility of CADM1 adhesion molecule is determined by alternative splicing and O-glycosylation. Sci Rep. 2017; 7:46174. 32. Eglen RM, Reisine T, Roby P, et al. The use of AlphaScreen technology in HTS: current status. Curr Chem Genomics. 2008;1:2-10. 33. McGiven JA, Sawyer J, Perrett LL, et al. A new homogeneous assay for high throughput serological diagnosis of brucellosis in ruminants. J Immunol Methods. 2008; 337(1):7-15. 34. Kurosawa G, Akahori Y, Morita M, et al. Comprehensive screening for antigens overexpressed on carcinomas via isolation of human mAbs that may be therapeutic. Proc Natl Acad Sci U S A. 2008;105(20):72877292. 35. Tsukasaki K, Tobinai K. HTLV-1-associated T-cell diseases. In: Francine F, editor. T-cell lymphomas. New York: Humana; 2013. pp. 113-135. 36. Iwanaga M, Watanabe T, Utsunomiya A, et al. Human T-cell leukemia virus type I (HTLV-1) proviral load and disease progression in asymptomatic HTLV-1 carriers: a nationwide prospective study in Japan. Blood. 2010;116(8):1211-1219. 37. Kataoka K, Nagata Y, Kitanaka A, et al. Integrated molecular analysis of adult T cell leukemia/lymphoma. Nat Genet. 2015; 47(11):1304-1315. 38. Shah UA, Chung EY, Giricz O, et al. North American ATLL has a distinct mutational and transcriptional profile and responds to epigenetic therapies. Blood. 2018; 132(14):1507-1518. 39. Sarkar B, Nishikata I, Nakahata S, et al. Degradation of p47 by autophagy contributes to CADM1 overexpression in ATLL cells through the activation of NF-ÎşB. Sci Rep. 2019;9(1):3491. 40. Manivannan K, Rowan AG, Tanaka Y, Taylor GP, Bangham CR. CADM1/TSLC1

identifies HTLV-1-infected cells and determines their susceptibility to CTL-mediated lysis. PLoS Pathog. 2016;12(4):e1005560. 41. Seidler S, Zimmermann HW, Weiskirchen R, Trautwein C, Tacke F. Elevated circulating soluble interleukin-2 receptor in patients with chronic liver diseases is associated with non-classical monocytes. BMC Gastroenterol. 2012;12:38. 42. Masuda M, Maruyama T, Ohta T, et al. CADM1 interacts with Tiam1 and promotes invasive phenotype of human T-cell leukemia virus type I-transformed cells and adult T-cell leukemia cells. J Biol Chem. 2010;285(20):15511-15522. 143. Boles KS, Barchet W, Diacovo T, Cella M, Colonna M. The tumor suppressor TSLC1/NECL-2 triggers NK-cell and CD8+ T-cell responses through the cell-surface receptor CRTAM. Blood. 2005;106(3):779786. 44s. Kannagi M, Hasegawa A, Nagano Y, Kimpara S, Suehiro Y. Impact of host immunity on HTLV-1 pathogenesis: potential of Tax-targeted immunotherapy against ATL. Retrovirology. 2019;16(1):23. 45. Yoneshige A, Hagiyama M, Inoue T, et al. Increased ectodomain shedding of cell adhesion molecule 1 as a cause of type II alveolar epithelial cell apoptosis in patients with idiopathic interstitial pneumonia. Respir Res. 2015;16:90. 46. Mimae T, Hagiyama M, Inoue T, et al. Increased ectodomain shedding of lung epithelial cell adhesion molecule 1 as a cause of increased alveolar cell apoptosis in emphysema. Thorax. 2014;69(3):223231. 47. Inoue T, Hagiyama M, Yoneshige A, et al. Increased ectodomain shedding of cell adhesion molecule 1 from pancreatic islets in type 2 diabetic pancreata: correlation with hemoglobin A1c levels. PLoS One. 2014; 9(6):e100988.

haematologica | 2021; 106(2)

ARTICLE

Chronic Lymphocytic Leukemia

Bendamustine followed by ofatumumab and ibrutinib in chronic lymphocytic leukemia: primary endpoint analysis of a multicenter, open-label, phase II trial (CLL2-BIO)

Paula Cramer,1 Julia von Tresckow,1 Sandra Robrecht,1 Jasmin Bahlo,1 Moritz Fürstenau,1 Petra Langerbeins,1 Natali Pflug,1 Othman Al-Sawaf,1 Werner J. Heinz,2 Ursula Vehling-Kaiser,3 Jan Dürig,4 Eugen Tausch,5 Manfred Hensel,6 Stephanie Sasse,1 Anna-Maria Fink,1 Kirsten Fischer,1 Karl-Anton Kreuzer,1 Sebastian Böttcher,7, 8 Matthias Ritgen,8 Michael Kneba,8 Clemens-Martin Wendtner,1,9 Stephan Stilgenbauer,5 Barbara Eichhorst1 and Michael Hallek1

Department I of Internal Medicine, Center for Integrated Oncology Aachen Bonn Cologne Dusseldorf, Faculty of Medicine and University Hospital of Cologne, University of Cologne, Cologne; 2Medical Center, Medical Clinic II, University of Würzburg, Würzburg; 3Tagesklinik Landshut, Landshut; 4Department of Hematology, West German Cancer Center, University Hospital Essen, Essen; 5Department III of Internal Medicine, University Hospital Ulm, Ulm; 6Mannheimer Onkologie Praxis, Mannheim; 7Department III of Internal Medicine, Rostock University Medical Center, Rostock; 8Department of Internal Medicine II, Campus Kiel, University of Schleswig-Holstein, Kiel and 9 Department of Hematology, Oncology, Immunology, Palliative Care, Infectious Diseases and Tropical Medicine, Klinikum Schwabing, Munich, Germany

Ferrata Storti Foundation

Haematologica 2021 Volume 106(2):543-554

ABSTRACT

he introduction of targeted agents has revolutionized the treatment of chronic lymphocytic leukemia but only few patients achieve a complete remission and minimal residual disease negativity with ibrutinib monotherapy. This multicenter, investigator-initiated, phase II study is evaluating sequential treatment with two cycles of bendamustine debulking for patients with a higher tumor load, followed by ofatumumab and ibrutinib induction and maintenance treatment. An all-comer population, irrespective of prior treatment, physical fitness and genetic factors, was included. The primary endpoint was the investigator-assessed overall response rate at the end of induction treatment. Of 66 patients enrolled, one patient with early treatment discontinuation was excluded from the efficacy analysis as predefined by the protocol. Thirty-nine patients (60%) were treatment-naïve and 26 patients (40%) had relapsed/refractory chronic lymphocytic leukemia, 21 patients (32%) had a del(17p) and/or TP53 mutation and 45 patients (69%) had unmutated IGHV status. At the end of the induction, 60 of 65 patients (92%) responded and nine (14%) achieved minimal residual disease negativity (<10-4) in peripheral blood. No unexpected or cumulative toxicities occurred. The most common grade 3 or 4 adverse events, according to the Common Toxicity Criteria, were neutropenia, anemia, infusion-related reactions, and diarrhea. This sequential treatment of bendamustine debulking, followed by ofatumumab and ibrutinib was well tolerated without unexpected safety signals and showed a good efficacy with an overall response rate of 92%. Ongoing maintenance treatment aims at deeper responses with minimal residual disease negativity. However, ibrutinib should still be used as a single agent outside clinical trials. Clinicaltrials.gov number: NCT02689141.

Introduction The introduction of targeted agents has led to profound changes in the treatment of chronic lymphocytic leukemia (CLL)1 and the Bruton tyrosine kinase inhibitor ibrutinib has become a preferred treatment option not only in relapsed/refractory and high-risk patients, but also for first-line therapy.2-6 However, only very few patients treated with single-agent ibrutinib achieve a complete remission (CR) and haematologica | 2021; 106(2)

Correspondence: PAULA CRAMER paula.cramer@uk-koeln.de Received: April 9, 2019. Accepted: February 25, 2020. Pre-published: February 27, 2020. https://doi.org/10.3324/haematol.2019.223693

543

P. Cramer et al.

minimal residual disease (MRD) negativity and therefore a continuous treatment appears to be necessary to maintain responses.7-9 This prolonged treatment duration increases the risk of toxicities, non-compliance and selection of resistant clones, as well as inflating the costs of treatment. We hypothesized that the combination of ibrutinib with the second-generation, fully humanized antibody ofatumumab10 might have the potential to increase the depth of responses and achieve CR and MRD negativity, i.e., eradicate the disease below the detection limit and thus enable treatment discontinuations without risk of immediate relapse. Several phase II and III studies are testing the combination of ibrutinib with the anti-CD20 antibodies rituximab,5,11,12 ofatumumab13 and obinutuzumab;4,14 the CLL2-BIO trial is evaluating ibrutinib with ofatumumab after an optional debulking with bendamustine.

Methods The CLL2-BIO trial is a prospective, open-label, multicenter, investigator-initiated trial by the German CLL Study Group with the University of Cologne, Germany being the legal sponsor (participating sites are listed in the Online Supplementary Appendix, page 1). The design of this phase II trial is based on the previously published so-called “sequential triple-T concept” of a tailored and targeted treatment aiming at a total eradication of MRD.15,16 It is evaluating a sequential treatment consisting of two cycles of bendamustine debulking, followed by six induction cycles and up to 24 months of maintenance treatment with ofatumumab and ibrutinib (Online Supplementary Appendix, Figure S1, page 5). The debulking (bendamustine 70 mg/m2 on days 1 and 2 repeated after 28 days) was recommended for patients with an absolute lymphocyte count ≥25,000/mL and/or lymph nodes with a diameter ≥5 cm, unless the patients had contraindications (e.g., refractoriness or known hypersensitivity to the drug). In the subsequent induction treatment, 300 mg ofatumumab was administered on day 1 of the first cycle and 1,000 mg on days 8 and 15 of the first cycle and every 4 weeks in cycles 2 to 6. Daily oral intake of 420 mg ibrutinib started on day 1 of the second cycle. In the maintenance treatment, ibrutinib (420 mg) was continued daily and ofatumumab (1,000 mg) administered every 3 months until unacceptable toxicity, progression/new CLL treatment or for up to 24 months. Furthermore, treatment was stopped in the case of a CR and MRD negativity in peripheral blood in two assessments within an interval of 3 months. The trial was designed for adult patients with treatment-naïve or relapsed/refractory CLL requiring treatment according to International Workshop on CLL (IWCLL) criteria;17 eligibility criteria are described in the Online Supplementary Appendix (pages 13). All patients provided written informed consent. Responses were evaluated by the investigators according to IWCLL criteria17 and reviewed centrally. Computed or magnetic resonance tomography (CT/MRI), as well as a bone marrow aspirate were required for confirmation of a CR or CR with incomplete marrow recovery (CRi). The response of patients fulfilling all IWCLL criteria of a CR/CRi (no evidence of lymphadenopathy, hepatoor splenomegaly on clinical examination and ultrasound or other imaging investigations, no disease-related symptoms and normalization of the hematologic parameters) but lacking one or both of these diagnostic modalities were termed “clinical CR” or “clinical CRi”, respectively, and graded as a partial response (PR). Samples for detection of MRD, which were mostly peripheral blood and also bone marrow, were taken for the final restaging after the induction treatment onwards and were analyzed cen544

trally with four-color flow cytometry.18,19 Results were categorized into three different MRD levels: low (<10-4), intermediate (≥10-4 and <10-2) and high (≥10-2)20 and MRD "negativity" was defined as <10-4. The primary endpoint of the CLL2-BIO trial was the overall response rate (ORR) after induction treatment. Further details on statistical analyses are provided in the Online Supplementary Appendix (page 4). The study was approved by the health authorities and the institutional review board of each participating site, was registered at clinicaltrials.gov (NCT02689141) and was conducted in accordance with the Declaration of Helsinki and International Conference on Harmonization-Good Clinical Practice.

Results Between February 4 and October 4, 2016, 66 patients were enrolled. One patient treated first-line who received fewer than two induction cycles (treatment discontinuation due to a generalized seizure on day 4 of induction cycle 2) was excluded from the efficacy analysis as predefined by the protocol but remained in the safety population. The patients flow through the study is summarized in Figure 1. The patients’ baseline characteristics are shown in Table 1. Of note, 21 of 65 patients (32%) had a del(17p) and/or TP53 mutation and 45 patients (69%) had unmutated IGHV. Thirty-nine of the 65 patients (60%) were treatmentnaïve and 26 patients (40%) had relapsed/refractory CLL with a median of 1.5 prior therapies (range, 1-5; interquartile range, 1-2). The prior therapies are presented in the Online Supplementary Appendix (page 6); most common therapies were bendamustine plus rituximab (15 times in 14 patients) and fludarabine, cyclophosphamide plus rituximab (8 times in 8 patients); five patients had received novel agents (3 idelalisib with rituximab and 2 venetoclax). Fifty-one of 65 patients (78%) received bendamustine debulking and 44 patients completed the planned two cycles; 14 patients (22%) started the induction therapy immediately. Sixty-three of 65 patients (97%) received all six induction cycles; two patients discontinued treatment in the fourth cycle, one due to bronchial carcinoma and one due to atrial fibrillation. The majority of patients received all eight ofatumumab infusions in the induction phase (62 of 65 patients, 95%) and the mean dose intensity of ofatumumab was 99% of the planned dose. Regarding the daily oral intake of ibrutinib in the induction phase, more than half of the patients had at least one dose reduction (40 patients, 62%) and/or treatment interruption (37 patients, 57%). The median duration of the treatment interruptions was 1 day (range, 1-83 days); 19 patients (29%) had treatment interruptions lasting ≥7 days, which were mainly due to adverse events. The ibrutinib dose was modified in 89 of the total of 384 treatment cycles (23%); this was due to adverse events in 42 cycles given to 24 patients. Despite these dose modifications, the mean dose intensity of ibrutinib in the induction treatment was 94% of the planned dose. The administration of debulking treatment with bendamustine appeared not to increase the rate of dose reductions and treatment interruptions or to have an adverse impact on the dose intensity (Online Supplementary Appendix, Table S5, page 8). Adverse events leading to dose modifications of ibrutinib and/or ofatumumab were mainly infusion-related reachaematologica | 2021; 106(2)

The CLL2-BIO trial

tions (n=18), infections (n=13), hematologic and gastrointestinal adverse events (9 each), as well as atrial fibrillation, nervous system and respiratory/thoracic disorders (4 each). Maintenance treatment was started in 59 patients

(91%) and up to the data-cutoff all 59 patients have completed the second cycle. At the final restaging after induction treatment, all 39 patients being treated first-line (100%) and 21 of the 26

Table 1. Patients’ characteristics.

Age, years median (range) Time from initial diagnosis, months median (range) Sex, n (%) female male Binet stage, n (%) A B C Presence of B-symptoms, n (%) ECOG performance status, n (%) 0 I II CIRS score (comorbidity) median, range > 6, n (%) Creatinine-clearance, mL/min median (range) 30-69 mL/min, n (%) Bulky disease with lymph nodes >5 cm, n (%) Massive splenomegaly >20 cm, n (%) Cytogenetics*, n (%) del(17p) del(11q) trisomy 12 none del(13q) Molecular genetics, n (%) TP53 NOTCH1 SF3B1 IGHV status, n (%) unmutated Serum β -microglobulin,** mg/dL median (range) > 3.5 mg/dL, n (%) Serum thymidine kinase, U/L** median (range) > 10 U/L, n (%) ZAP-70 expression, n (%) >20% CD38 expression >30% CLL-IPI,** n (%) low intermediate high very high

Treatment-naïve (n=39)

Relapse/refractory (n=26)

All patients (n=65)

60 (32-80)

68 (41-81)

61 (32-81)

25 (4-134)

84 (2-335)

48 (2-335)

16 (41%) 23 (59%)

8 (31%) 18 (69%)

24 (37%) 41 (63%)

9 (23%) 16 (41%) 14 (36%) 23 (59%)

7 (27%) 10 (38%) 9 (35%) 10 (38%)

16 (25%) 26 (40%) 23 (35%) 33 (51%)

28 (72%) 11 (28%) -

16 (62%) 2 (8%)

44 (68%) 19 (29%) 2 (3%)

3 (0 - 16) 6 (15%)

3 (0 - 10) 6 (23%)

3 (0 - 16) 12 (18%)

81.9 (42.9-149.2) 10 (26%) 7 (18%) 2 (5%)

75.3 (31.3-131.5) 9 (35%) 6 (23%) 2 (8%)

80.5 (31.3-149.2) 19 (29%) 13 (20%) 4 (6%)

6 (15%) 8 (21%) 6 (15%) 13 (33%) 6 (15%)

6 (23%) 4 (15%) 4 (15%) 5 (19%) 7 (27%)

12 (18%) 12 (18%) 10 (15%) 18 (28%) 13 (20%)

7 (18%) 9 (23%) 3 (8%)

13 (50%) 3 (12%) 5 (19%)

20 (31%) 12 (18%) 8 (12%)

27 (69%)

18 (69%)

45 (69%)

3.7 (2.1 – 11.6) 21 (55%)

4.7 (2.4 – 15.5) 18 (72%)

4.2 (2.1 – 15.5) 39 (62%)

29.8 (11.5 – 258.0) 38 (100%)

38.9 (4.0 – 343.5) 23 (92%)

36.3 (4.0 – 343.5) 61 (97%)

15 (38%)

14 (54%)

29 (45%)

16 (41%)

13 (50%)

29 (45%)

5 (13%) 11 (29%) 17 (45%) 5 (13%)

1 (4%) 4 (16%) 8 (32%) 12 (48%)

6 (10%) 15 (24%) 25 (40%) 17 (27%)

CIRS: Cumulative Illness Rating Scale; ECOG: Eastern Cooperative Oncology Group; IGHV: immunoglobulin heavy-chain variable region; CLL: IPI: chronic lymphocytic leukemia International Prognostic Index; ZAP-70: zeta-chain associated protein kinase 70. *according to the hierarchical model by Döhner et al.; **data for five patients missing.

haematologica | 2021; 106(2)

545

P. Cramer et al.

relapsed/refractory patients (81%) had responded (Table 2). With a 92% ORR for the mixed population at the end of induction staging, the primary endpoint was met (95% confidence interval [95% CI]: 83â&#x20AC;&#x201C;98%; P=0.0013 when tested against the null hypothesis of an ORR of 75%). Fifty-nine patients achieved a PR according to IWCLL criteria and one had a PR with lymphocytosis; no CR/CRi were documented, but 18 patients with a PR lacked a bone marrow biopsy (18 patients) and/or a CT scan (14 patients) but fulfilled all other criteria for a CR. MRD was assessed in peripheral blood at the final restaging in 62 of

65 patients and nine patients (14%) had achieved MRD "negativity" (<10-4). The MRD "negativity" rate at the final restaging was 18% among patients treated first-line and 8% among the relapsed/refractory patients. Patients with adverse prognostic markers such as del(17p) and unmutated IGHV had lower overall response and MRD negativity rates: the ORR were 75% and 89%, respectively, for patients with these markers, while the MRD negativity rates were 0% and 9%, respectively (Online Supplementary Appendix, Table S2, page 6). After up to two cycles of the initial debulking treatment

Figure 1. CONSORT diagram. Patientsâ&#x20AC;&#x2122; flow through the study. pts: patients; AE: adverse event; MRD: minimal residual disease.

546

haematologica | 2021; 106(2)

The CLL2-BIO trial

with bendamustine, 30 of 51 patients (59%) achieved a response according to IWCLL criteria, including 23 of 35 (66%) patients treated first-line and seven of 16 (44%) relapsed/refractory patients. Nineteen patients (37%) had stable disease and two patients (4%) had progressive disease but proceeded with induction treatment on study. Eleven of the 21 patients with a del(17p)/TP53 mutation received debulking treatment: six achieved a response according to IWCLL criteria and the other five had stable disease. Nine of 15 patients who had previously been treated with bendamustine were re-exposed to bendamustine in the trial and four of them (44%) responded to the debulking treatment. Over time, the responses of the patients improved (Figure 2A). The best response achieved until 6 months after the final restaging was CR/CRi in four patients (6%), clinical CR/CRi without confirmatory CT scan and/or bone marrow biopsy in 28 patients (43%) and PR in 33 patients (51%). By the time of the data cut-off, 12 patients had stopped maintenance treatment: five due to a MRDnegative response, four with progressive disease, two due to adverse events and one patient died (Figure 2B). At the time of the data cut-off (median observation time of 20 months) six patients (all in the relapsed/refractory cohort) had progressed and two had died. One of the deaths was due to sepsis with multi-organ failure after the sixth induction cycle in a 73-year old female patient who had received five prior therapies. The other death was caused by ischemic cerebral infarction in the second maintenance cycle in an 73-year old male patient who had not received prior therapy and was initially hospitalized because of pneumonia, developed a tachyarrhythmia and in whom anticoagulant treatment was discontinued due to a fall and hemoptysis (see Online Supplementary Appendix, Table S3, page 7). The median progression-free and overall survival were not reached. The estimated 15month progression-free survival rate was 94% (95% CI: 88-100%) for the entire cohort, being 97% (93-100%) among those treated first-line and 89% (76-100%) among the relapsed/refractory patients (Figure 3A). Estimated 15month overall survival rates were 97% (95% CI: 93100%) for the entire cohort, 97% (93-100%) among patients treated first-line and 96% (89-100%) among the relapsed/refractory patients (Figure 3B). By the time of data cut-off, a total of 959 adverse events had been reported in all 66 patients: 602 (63%) occurred in the treatment-naïve patients and 357 (37%) in the relapsed/refractory cohort (Online Supplementary Appendix, Table S4, page 7). In total, 158 adverse events (16%) occurred during the debulking treatment, 547 (57%) during induction and 254 (26%) in the maintenance/followup phase. According to investigator assessment, 600 (63%) were deemed related to the study treatment. The majority of the adverse events (742, 77%) resolved and 796 adverse events (83%) did not require an adjustment of the study drug. Most adverse events were mild (Common Toxicity Criteria [CTC] grade 1: n=468, 49%) or moderate (CTC grade 2: n=355, 37%), although 120 adverse events (13%) were severe/medically significant (CTC grade 3), 14 (1%) were considered life-threatening (CTC grade 4) and two had a fatal outcome. Both deaths and 97 of the 134 grade 3 or 4 adverse events (72%) were considered related to the study treatment. Eighty-seven of the 959 adverse events (9%) were reported as serious adverse events and 69 of them (79%) were considered treatment-related. haematologica | 2021; 106(2)

Table 2. Response at the end of induction (efficacy population).

Treatment-naïve Rel./refractory All patients (n=39) (n=26) (n=65) Overall response rate, n (%) exact 95% CI Response, n (%) CR clinical CR/CRi* PR PR with lymphocytosis SD PD MRD in peripheral blood, n (%) negative (< 10-4) intermediate (≥ 10-4 and < 10-2) positive (≥ 10-2) missing

39 (100%) 91 - 100%

21 (81%) 61 - 94%

60 (92%) 83 - 98%

15 (38%) 24 (62%) -

3 (12%) 17 (65%) 1 (4%) 3 (12%) 2 (8%)

18 (28%) 41 (63%) 1 (2%) 3 (5%) 2 (3%)

7 (18%) 9 (23%) 21 (54%) 2 (5%)

2 (8%) 3 (12%) 20 (77%) 1 (4%)

9 (14%) 12 (18%) 41 (63%) 3 (5%)

95% CI: 95% confidence interval; CR: complete remission; CRi: CR with incomplete recovery of bone marrow; MRD: minimal residual disease; PD: progressive disease; PR: partial response; SD: stable disease. *Clinical CR/CRi is a remission fulfilling all criteria of a CR/CRi according to International Workshop on Chronic Lymphocytic Leukemia criteria except a bone marrow examination and/or computed tomography/magnetic resonance imaging scan: the lymphocyte count had to be normalized, patients had to be asymptomatic and have no evidence of lymphadenopathy and hepatosplenomegaly in clinical examination and ultrasound or other imaging.

During the two cycles of bendamustine debulking, 45 of 52 patients (87%) experienced adverse events of any grade (Table 3A), including 16 patients (31%) with grade 3 and three patients (6%) with grade 4 toxicities. Most common grade 1 or 2 toxicities were nausea (14 patients, 27%), rash/exanthema (8 patients, 15%) and fatigue (7 patients, 13%) and most common grade 3 or 4 toxicities were neutropenia (6 patients, 12%) and anemia (5 patients, 10%). Debulking did not lead to an increased rate of cytopenias or infections in the induction phase; 17% of patients with debulking had grade 3 or 4 hematologic toxicities and 8% experienced grade 3 or 4 infections in the induction phase compared to 50% and 21% among the 14 patients without prior debulking (Online Supplementary Appendix, Table S5, page 8). The rate of infusion-related reactions was 38%, including 4% grade 3 or 4 reactions among patients with prior debulking compared to 64% and 21%, respectively, among patients without debulking. During the induction treatment with ofatumumab and ibrutinib, all 66 patients experienced adverse events (Table 3B), including 25 (38%) who had grade 3 events, eight (12%) who had grade 4 adverse events, and one patient (2%) who died of septic shock. The most common grade 1 or 2 toxicities were infusion-related reactions (24 patients, 36%), upper respiratory tract infections (20 patients, 30%), diarrhea (18 patients, 27%), muscle disorders (17 patients, 26%), fatigue (14 patients, 21%) and rash/exanthema (13 patients, 20%). The most common grade 3 or 4 toxicities were neutropenia (14 patients, 21%), infusion-related reactions (5 patients, 8%) and diarrhea (4 patients, 6%). Cytopenias, including neutropenia and thrombocytopenia, as well as epistaxis and hematoma occurred more often among the relapsed/refractory patients, while CTC grade 1 or 2 infections, mild gastrointestinal toxicities and skin/subcutaneous disorders, as well as hypertension were reported more often among the treatment-naïve patients. Of note, four of 66 patients (6%) reported paresthesia or dysesthesia (all CTC grade 1 or 2) during the induction 547

P. Cramer et al.

treatment and three other patients had paresthesia or dysesthesia or sensory abnormalities during debulking treatment.

Discussion This paper reports on a sequential combination treatment strategy, consisting of bendamustine debulking, followed by ofatumumab and ibrutinib in induction and a maintenance phase in a mixed population of treatmentnaĂŻve and relapsed/refractory patients. The primary end-

point of the trial was met: at the end of the induction treatment, an ORR of 92% and a MRD negativity rate of 14% in peripheral blood were achieved. As expected, response and MRD negativity rates were somewhat lower in patients with relapsed/refractory disease or with the high-risk markers del(17p)/TP53 mutation or unmutated IGHV. The response and MRD negativity rates achieved are inferior to those reported with standard chemoimmunotherapies,21-24 venetoclax either as a single agent or in combination with rituximab or obinutuzumab,25-27 and with ibrutinib plus obinutuzumab4,14 (Online Supplementary Appendix, Table S6, page 9). The efficacy results after 8

548

Figure 2. Improvement of response, time to first minimal residual disease negativity and treatment discontinuation. (A) Change in response with continued treatment. (B) Time to first documentation of minimal residual disease negativity and treatment discontinuation. CR: complete remission; CRi: CR with incomplete recovery of bone marrow; PD: progressive disease; PR: partial response; SD: stable disease; *Clinical CR/CRi is a remission fulfilling all criteria of a CR/CRi according to International Workshop on Chronic Lymphocytic Leukemia criteria except a bone marrow examination and/or computed tomography/magnetic resonance imaging scan: the lymphocyte count had to be normalized, patients had to be asymptomatic and have no evidence of lymphadenopathy and hepatosplenomegaly in clinical examination and ultrasound or other imaging.

haematologica | 2021; 106(2)

The CLL2-BIOâ&#x20AC;&#x2C6;trial

Figure 3. Progression-free and overall survival. (A) Progression-free survival. (B) Overall survival. FL: patients treated first-line; RR: relapsed/refractory patients

haematologica | 2021; 106(2)

549

P. Cramer et al. Table 3A. Adverse events during debulking with bendamustine reported in â&#x2030;Ľ5% of all patients: data are number of patients (%)

Maximum grade Patients with AE Blood and lymphatic system disorders Anemia Neutropenias Thrombocytopenia Ear and labyrinth disorders Vertigo Gastrointestinal disorders Nausea Diarrhea Constipation Abdominal pain General disorders and administration site conditions Fatigue Pyrexia Edema Immune system disorders Infections and infestations Upper repiratory tract inf. Oral herpes Metabolism and nutrition disorders Hyperuricemia Nervous system disorders Renal and urinary disorders Acute kidney injury Respiratory, thoracic and mediastinal disorders Cough Skin and subcutaneous tissue disorders Rash, exanthema Pruritus

1/2

Treatment-naĂŻve (n=36) 3 4

1/2

All patients (n=52) 3 4

17 (47%) 13 (36%) 2 (6%) 4 (11%)

2 (6%) 1 (3%)

0 0

9 (56%) 1 (6%)

3 (19%) 1 (6%)

1 (6%) 1 (6%)

0 0

26 (50%) 3 (6%)

16 (31%) 5 (10%)

3 (6%) 2 (4%)

0 0

0 0 3 (8%) 1 (3%) 1 (3%) 15 (42%) 8 (22%) 4 (11%) 2 (6%) 1 (3%) 9 (25%)

3 (8%) 5 (14%) 1 (3%) 0 0 1 (3%) 0 0 0 0 1 (3%)

1 (3%) 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0

1 (6%) 0 0 2 (13%) 2 (13%) 8 (50%) 6 (38%) 0 1 (6%) 2 (13%) 5 (31%)

1 (6%) 0 0 0 0 1 (6%) 1 (6%) 0 0 0 0

0 1 (6%) 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0

1 (2%) 0 3 (6%) 3 (6%) 3 (6%) 23 (44%) 14 (27%) 4 (8%) 3 (6%) 3 (6%) 14 (27%)

4 (8%) 5 (10%) 1 (2%) 0 0 2 (4%) 1 (2%) 0 0 0 1 (2%)

1 (2%) 1 (2%) 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0

3 (8%) 4 (11%) 2 (6%) 1 (3%) 7 (19%) 2 (6%) 2 (6%) 4 (11%)

0 0 0 3 (8%) 3 (8%) 0 0 4 (11%)

0 0 0 0 0 0 0 0

4 (25%) 1 (6%) 1 (6%) 0 4 (25%) 2 (13%) 1 (6%) 0

0 0 0 0 0 0 0 1 (6%)

0 0 0 0 0 0 0 0

7 (13%) 5 (10%) 3 (6%) 1 (2%) 11 (21%) 4 (8%) 3 (6%) 4 (8%)

0 0 0 3 (6%) 3 (6%) 0 0 5 (10%)

0 0 0 0 0 0 0 0

1 (3%) 3 (8%) 4 (11%) 4 (11%) 3 (8%)

2 (6%) 0 0 0 1 (3%)

0 0 0 0 0

0 2 (13%) 0 0 2 (13%)

0 0 0 0 0

1 (2%) 5 (10%) 4 (8%) 4 (8%) 5 (10%)

2 (4%) 0 0 0 1 (2%)

0 0 0 0 0

2 (6%) 12 (33%)

0 1 (3%)

0 0

1 (6%) 2 (13%)

0 0

3 (6%) 14 (27%)

0 1 (2%)

0 0

7 (19%) 3 (8%)

1 (3%) 0

0 0

1 (6%) 0

0 0

8 (15%) 3 (6%)

1 (2%) 0

0 0

months of induction therapy are comparable to those reported for ibrutinib as a single agent3,5,12,28,29 (although MRD negativity is seen infrequently) or combined with the antibodies rituximab5,11,12 or ofatumumab.13 Two randomized trials showed faster and deeper remissions with the addition of rituximab to ibrutinib, but this did not translate into progression-free or overall survival differences.5,12 However, this might change with continued treatment and longer follow-up. Several trials showed that continued ibrutinib treatment is able to increase the rates of CR and MRD-negative remissions - responses also deepened in our study during the maintenance phase with continued ibrutinib and ofatumumab. No unexpected or cumulative toxicities were seen with the combination of bendamustine, followed by ofatumumab and ibrutinib. The majority of adverse events were mild/moderate (CTC grade 1 or 2). The most common CTC grade 3 or 4 adverse events were neutropenia and anemia during the debulking with bendamustine and neutropenia, infusion-related reactions and diarrhea during the induction with ofatumumab and ibrutinib. Except for two fatal adverse events, the toxicities were generally 550

Relapsed/refractory (n=16) 3 4

manageable: only six patients discontinued treatment prematurely due to toxicity and the mean dose intensity of both ibrutinib and ofatumumab in the induction phase was more than 90% of the planned dose. The rates of neutropenia, thrombocytopenia and infections in both cohorts compare favorably to those with chemoimmunotherapy and other targeted therapies (Online Supplementary Appendix, Table S6, page 9).3-5,12,14,21-31 The toxicities seen in this trial are in accordance with those reported in a phase-Ib/II trial evaluating different administration sequences of ibrutinib and ofatumumab in 71 patients with relapsed CLL, small lymphocytic lymphoma, prolymphocytic leukemia or Richter transformation.13 Importantly, the high incidence of peripheral sensory neuropathy reported in that trial (31 of 71 patients, mostly CTC grade 1 or 2) was not confirmed in our study despite a longer period of observation. The primary goal of the initial debulking with bendamustine was to reduce the tumor load before initiation of treatment with ofatumumab and ibrutinib to improve treatment tolerability. As intended, the rate of infusionrelated reactions was lower among patients who had haematologica | 2021; 106(2)

The CLL2-BIO trial Table 3B. Adverse events during induction with ofatumumab and ibrutinib reported in ≥5% of patients: data are number of patients (%).

Maximum grade Patients with AE Blood and lymphatic system disorders Neutropenia Thrombocytopenia Cardiac disorders Atrial fibrillation Ear and labyrinth disorders Eye disorders Gastrointestinal disorders Diarrhea Abdominal pain Oral soft tissue conditions Dyspepsia Nausea Gastroesophageal reflux General disorders and administration site conditions Fatigue Edema Immune system disorders Infections and infestations Upper repiratory tract inf. Lower repiratory tract inf. Infections NEC Dental/oral soft tissue inf. Urinary tract infections Abdominal GI infections Oral herpes Conjunctivitis Injury, poisoning and procedural complications Infusion-related reaction Investigations Metabolism and nutrition disorders Musculoskeletal/connective tissue disorders Muscle disorders Musculoskeletal/connective tissue disorders NEC Arthralgia Neoplasms benign/malignant* Nervous system disorders Dizziness Headaches Par-/dysesthesias Psychiatric disorders Renal and urinary disorders Urinary tract signs/symptoms Respiratory, thoracic and mediastinal disorders Cough Epistaxis Pneumonitis Oropharyngeal pain

1/2

Treatment-naïve (n=40) 3 4

Relapsed/refractory (n=26) 3 4

1/2

21 (53%) 14 (35%) 5 (13%) 0 4 (10%) 4 (10%)

0 0

11 (42%) 4 (15%)

11 (42%) 5 (19%)

0 0 5 (13%) 1 (3%) 2 (5%) 6 (15%) 21 (53%) 12 (30%) 4 (10%) 4 (10%) 3 (8%) 3 (8%) 2 (5%) 17 (43%)

3 (8%) 0 2 (5%) 2 (5%) 0 1 (3%) 3 (8%) 3 (8%) 0 0 0 0 0 0

4 (10%) 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 2 (8%) 2 (8%) 1 (4%) 2 (8%) 5 (19%) 9 (35%) 6 (23%) 4 (15%) 2 (8%) 2 (8%) 2 (8%) 3 (12%) 8 (31%)

9 (23%) 3 (8%) 1 (3%) 27 (68%) 16 (40%) 2 (5%) 3 (8%) 4 (10%) 4 (10%) 2 (5%) 4 (10%) 3 (8%) 20 (50%)

0 0 0 2 (5%) 0 0 0 0 0 1 (3%) 0 0 3 (8%)

0 0 0 0 0 0 0 0 0 0 0 0 0

16 (40%) 4 (10%) 5 (13%)

2 (5%) 0 0

0 0 0

22 (55%)

2 (5%)

10 (25%) 9 (23%)

All patients (n=66) 3 4

1/2

3 (12%) 3 (12%)

1 (4%) 0

32 (48%) 4 (6%)

25 (38%) 9 (14%)

8 (12%) 7 (11%)

1 (2%) 0

5 (19%) 1 (4%) 1 (4%) 1 (4%) 0 0 2 (8%) 1 (4%) 0 0 0 0 0 1 (4%)

2 (8%) 1 (4%) 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 2 (3%) 7 (11%) 2 (3%) 4 (6%) 11 (17%) 30 (45%) 18 (27%) 8 (12%) 6 (9%) 5 (8%) 5 (8%) 5 (8%) 25 (38%)

8 (12%) 1 (2%) 3 (5%) 3 (5%) 0 1 (2%) 5 (8%) 4 (6%) 0 0 0 0 0 1 (2%)

6 (9%) 1 (2%) 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0

5 (19%) 3 (12%) 3 (12%) 8 (31%) 4 (15%) 3 (12%) 2 (8%) 1 (4%) 1 (4%) 0 0 1 (4%) 9 (35%)

0 0 0 4 (15%) 0 2 (8%) 1 (4%) 0 0 1 (4%) 0 0 3 (12%)

0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 1 (4%) 0 0 0 0 0 0 0 0 0

14 (21%) 6 (9%) 4 (6%) 35 (53%) 20 (30%) 5 (8%) 5 (8%) 5 (8%) 5 (8%) 2 (3%) 4 (6%) 4 (6%) 29 (44%)

0 0 0 6 (9%) 0 2 (3%) 1 (2%) 0 0 2 (3%) 0 0 6 (9%)

0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 1 (2%) 0 0 0 0 0 0 0 0 0

0 0 0

8 (31%) 3 (12%) 4 (15%)

3 (12%) 0 0

0 0 0

24 (36%) 7 (11%) 9 (14%)

5 (8%) 0 0

0 0 0

10 (38%)

32 (48%)

2 (3%)

1 (3%) 1 (3%)

0 0

7 (27%) 2 (8%)

0 0

17 (26%) 11 (17%)

1 (2%) 1 (2%)

0 0

6 (15%) 0 10 (25%) 3 (8%) 3 (8%) 3 (8%) 2 (5%) 2 (5%) 2 (5%) 10 (25%)

0 1 (3%) 2 (5%) 0 0 0 0 0 0 1 (3%)

0 1 (3%) 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0

3 (12%) 1 (4%) 7 (27%) 2 (8%) 1 (4%) 1 (4%) 2 (8%) 3 (12%) 3 (12%) 9 (35%)

0 2 (8%) 2 (8%) 0 0 0 0 0 0 1 (4%)

0 0 0 0 0 0 0 0 0 0

9 (14%) 1 (2%) 17 (26%) 5 (8%) 4 (6%) 4 (6%) 4 (6%) 5 (8%) 5 (8%) 19 (29%)

0 3 (5%) 4 (6%) 0 0 0 0 0 0 2 (3%)

0 1 (2%) 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0

5 (13%) 3 (8%) 2 (5%) 3 (8%)

0 0 0 0

4 (15%) 4 (15%) 1 (4%) 1 (4%)

0 0 1 (4%) 0

0 0 0 0

9 (14%) 7 (11%) 3 (5%) 4 (6%)

0 0 1 (2%) 0

0 0 0 0

continued on the next page

haematologica | 2021; 106(2)

551

P. Cramer et al. continued from the previous page

Maximum grade Skin and subcutaneous tissue disorders Rash, exanthema Dermatitis, eczema Erythema Vascular disorders Hematoma Hypertension

Treatment-naĂŻve (n=40) 3 4

1/2

15 (38%)

2 (5%)

7 (27%)

2 (8%)

22 (33%)

4 (6%)

10 (25%) 2 (5%) 3 (8%) 9 (23%) 2 (5%) 4 (10%)

0 1 (3%) 0 1 (3%) 0 1 (3%)

0 0 0 0 0 0

3 (12%) 1 (4%) 1 (4%) 8 (31%) 6 (23%) 0

0 0 0 1 (4%) 0 0

0 0 0 0 0 0

13 (20%) 3 (5%) 4 (6%) 17 (26%) 8 (12%) 4 (6%)

0 1 (2%) 0 2 (3%) 0 1 (2%)

0 0 0 0 0 0

1/2

Relapsed/refractory (n=26) 3 4

All patients (n=66) 3 4

*including bronchial carcinoma (grade 4 adverse event occurring in cycle 4 in a first-line patient), squamous cell carcinoma of the skin (2 cases in 2 relapsed/refractory patients), basal cell carcinoma (1 case in a first-therapy naive patient), and seborrhoic keratosis (2 cases in 2 relapsed/refractory patients)

received bendamustine than in those who had not been given debulking treatment, and especially compared to the rate in patients in whom ibrutinib and ofatumumab were started concurrently in the above-mentioned phase Ib/II study.13 Although the majority of the patients experienced adverse events during the debulking treatment, including one third of patients with grade 3 or 4 toxicities, the administration of bendamustine did not seem to make patients more prone to cytopenias and infections in the induction phase and did not increase the risk of dose reductions and/or treatment interruptions. Due to low numbers of patients, it is difficult to draw a definite conclusion on whether debulking therapy with bendamustine is beneficial for all patients or only for certain subgroups. Additional analyses including two other phase II trials with a similar design also evaluating bendamustine debulking, followed by obinutuzumab with either ibrutinib or venetoclax (NCT02345863 and NCT02401503), are under way. First results suggest that the overall rate of adverse events and especially of cytopenias and infections are not increased, while infusion-related reactions seem to be less common in patients with prior debulking. The administration of single-agent ibrutinib over 2 to 3 months might serve as an alternative approach for debulking, as it effectively reduces the lymphadenopathy through a redistribution of the CLL cells to the blood.32 The inclusion of an all-comer population led to a quite heterogeneous group of patients. This might be seen as a limitation of the study but it can also be argued that it helps to transfer the results to routine clinical practice. The lack of CR confirmed by CT/MRI scan and bone marrow may also be criticized, but as the majority of patients continue treatment in a maintenance phase, it is planned that the missing examinations will be performed later. These assessments were not mandatory per study protocol at a defined time point in order to avoid unnecessary and repetitive CT scans or bone marrow biopsies. Instead it was recommended to perform these examinations as soon as the best clinical response was achieved, which is consistent with the updated IWCLL guidelines.17 Figure 2A suggests that the depth of response increases over time and consequently also the rates of CR and MRD "negativity". However, so far, only a few patients have achieved a deep, MRD-negative remission and have been able to stop maintenance treatment. The efficacy results in the current analysis are not superior to those achieved with ibrutinib monotherapy. The findings of six other clinical trials evaluating a combination of ibrutinib with an anti-CD20 antibody in patients with CLL have been published. Three 552

evaluated the combination of ibrutinib with rituximab (including 2 randomized studies with single-agent ibrutinib as a comparator arm),5,6,12 two with obinutuzumab4,14 and one with ofatumumab.13 Cumulative evidence from these trials indicates that the addition of an antibody leads to faster and deeper remissions and one study showed that the achievement of a CR is associated with longer progression-free survival.33 However, the two randomized trials with ibrutinib with or without rituximab did not show (or have not yet shown) a difference in survival times.5,12 When comparing the results of this trial with those of the CLL2-BIG trial, which evaluated the same treatment regimen but with obinutuzumab instead of ofatumumab, the responses appear to be deeper, as the MRD "negativity" rate at the end of induction treatment was 48% compared to 14% in CLL2-BIO.34 The addition of an antibody to ibrutinib cannot be recommended in routine practice, but individual patients in whom a faster response or clearance of the lymphocytes from the peripheral blood is needed might benefit from a combination treatment. However, in such situations obinutuzumab should be favored over rituximab and ofatumumab based on the current data and because ofatumumab is available only in the USA but not in Europe anymore. Further follow-up is necessary to determine whether bendamustine followed by ofatumumab and ibrutinib can produce deep, MRD negative remissions in a relevant proportion of patients and whether treatment can be stopped in this situation. For the regimen with bendamustine debulking followed by ibrutinib with obinutuzumab instead of ofatumumab it was already shown that patients, including those with high-risk cytogenetic abnormalities, remain in remission for a clinically meaningful time after termination of treatment.35 Termination of ibrutinib monotherapy would constitute a novel paradigm, as it overcomes the need for indefinite treatment, which is so far needed to maintain responses with the B-cell receptor signaling pathway inhibitors ibrutinib and idelalisib.7-9 Shortened treatment durations might reduce problems of patient compliance as well as treatment costs. It could also avoid resistance due to clonal evolution arising from a therapeutic pressure selecting small subclones with resistance mechanisms.9,36,37 Disclosures PC reports research funding from AbbVie, Acerta, F. Hoffmann-LaRoche, Gilead, GlaxoSmithKline, Janssen-Cilag and Novartis, honoraria for scientific talks from AbbVie, F. Hoffmann-LaRoche and Janssen-Cilag and for advisory boards haematologica | 2021; 106(2)

The CLL2-BIO trial

from AbbVie, Acerta, AstraZeneca, Janssen-Cilag and Novartis, as well as travel grants from AbbVie, F. Hoffmann LaRoche, Gilead, Janssen-Cilag and Mundipharma. JvT reports research funding from F. Hoffmann-LaRoche and Janssen-Cilag, honoraria for advisory boards from AbbVie, F. Hoffmann-LaRoche and Janssen-Cilag, as well as travel grants from Celgene, F. Hoffmann-LaRoche and Janssen-Cilag. JB reports honoraria and travel grants from F. Hoffmann-LaRoche. PL reports research funding from AbbVie, F. Hoffmann-LaRoche and Janssen-Cilag, honoraria from Janssen-Cilag and travel grants from F. Hoffmann-LaRoche, Janssen-Cilag and Mundipharma. OA-S reports personal fees from F. HoffmannLaRoche, AbbVie and Gilead. WH reports research funding from MSD (Merck Sharp & Dohme) and Pfizer, honoraria for scientific talks from Alexion, Basilea, Gilead, MSD (Merck Sharp & Dohme) and Pfizer, advisory boards from Gilead Science, MSD, Pfizer, travel grants from Alexion, Astellas, Basilea, Gilead Science, MSD and Pfizer (all outside the submitted work). UVK reports honoraria for advisory boards from AbbVie, Amgen, F. Hoffmann-LaRoche, Gilead, Lilly and MSD. JD reports honoraria for scientific talks from F. Hoffmann-LaRoche, Janssen-Cilag, Celgene; advisory boards from AbbVie, Amgen, Bristol-Myers Squibb, Janssen-Cilag, Novartis, Gilead; travel grants from Celgene, F. Hoffmann LaRoche, Gilead, Janssen-Cilag, and Mundipharma. ET reports travel grants from Abbvie, Celgene and Gilead. MHe reports honoraria for advisory boards from AbbVie and F. Hoffmann-LaRoche, and travel grants from AbbVie and F. Hoffmann-LaRoche. AMF reports research grants from Celgene and travel grants from AbbVie, F. Hoffmann-LaRoche and Mundipharma. KF reports travel grants from F. HoffmannLaRoche. KAK reports research support and honoraria for consultancy or advisory boards from AbbVie, F. Hoffmann-LaRoche and Mundipharma. SB reports grants and personal fees from AbbVie, F-Hoffmann-LaRoche and Janssen, grants from Celgene and Genentech, and personal fees from Becton Dickinson and Novartis. MR reports research funding from AbbVie, Amgen, F. Hoffmann-LaRoche and Gilead. MK reports honoraria for consultancy or advisory boards from AbbVie and F. Hoffmann-LaRoche. CMW reports research support, honoraria for consultancy or advisory boards and travel

References 1. Hallek M, Shanafelt TD, Eichhorst B. Chronic lymphocytic leukaemia. Lancet. 2018;391(10129):1524-1537. 2. Burger JA, Tedeschi A, Barr PM, et al. Ibrutinib as initial therapy for patients with chronic lymphocytic leukemia. N Engl J Med. 2015;373(25):2425-2437. 3. Byrd JC, Brown JR, O'Brien S, et al. Ibrutinib versus ofatumumab in previously treated chronic lymphoid leukemia. N Engl J Med. 2014;371(3):213-223. 4. Moreno C, Greil R, Demirkan F, et al. Ibrutinib plus obinutuzumab versus chlorambucil plus obinutuzumab in first-line treatment of chronic lymphocytic leukaemia (iLLUMINATE): a multicentre, randomised, open-label, phase 3 trial. Lancet Oncol. 2019;20(1):43-56. 5. Woyach JA, Ruppert AS, Heerema NA, et al. Ibrutinib regimens versus chemoimmunotherapy in older patients with untreated CLL. N Engl J Med. 2018;379(26): 2517-2528.

haematologica | 2021; 106(2)

support from AbbVie, F. Hoffmann-LaRoche, Genentech, Gilead, GlaxoSmithKline, Janssen-Cilag, Mundipharma and Pharmacyclics. SS reports research support, honoraria for consultancy or advisory boards and travel support from AbbVie, Amgen, Celgene, F. Hoffmann-LaRoche, Genentech, Gilead, GlaxoSmithKline, Janssen-Cilag, Mundipharma, Novartis and Pharmacyclics. BE reports research support and honoraria for consultancy or advisory boards from AbbVie and F. HoffmannLaRoche. MHa reports research support and honoraria from Roche, Gilead, Mundipharma, Janssen, Celgene, Pharmacyclics and Abbvie. NP, SR and MF report no conflicts of interest. Contributions PC, JvT, JB, SB, BE and MH were responsible for the conception and design of the study; PC, JvT, SR, JB, MF, PL, NP, OAS, AMF and KF were responsible for the trial management; PC, JvT, MF, PL, NP, OA-S, WH, UV-K, JD, ET, MHe, SS, SB, MR, MK, CMW, SS, BE and MH were responsible for the recruitment and treatment of patients; PC, JvT, SR, JB, MF, PL and NP had access to the raw data; PC, MF, PL and NP performed a central review of all clinical data; ET, KAK, SB, MR, MK and SS performed the laboratory analyses; SR and JB performed the statistical analysis. All authors interpreted the data; PC wrote the first draft of the manuscript and all authors approved the final manuscript. Acknowledgments The CLL2-BIO trial is an investigator-initiated trial with the University of Cologne being the legal sponsor. Financial support was provided by the pharmaceutical company Novartis and the study drugs were provided by Novartis and Janssen-Cilag. The authors wish to express their gratitude towards all patients participating in the trial, as well as the physicians and trial staff at the sites. We also thank the study management team with the project managers Tanja Annolleck, Johanna Wesselmann, Jana Kalz, and Miriam Schüler-Aparicio, the safety managers Sabine Frohs, and Tanja Annolleck, the data managers Viktoria Monar, Olga Korf, Florian Drey, Irene Stodden and Swetlana Dubnov, and Dr. Birgit Fath and the monitors from the “competence network, malignant lymphoma” (“Kompetenznetz Maligne Lymphome”).

6. Shanafelt TD, Wang XV, Kay NE, et al. Ibrutinib-rituximab or chemoimmunotherapy for chronic lymphocytic leukemia. N Engl J Med 2019; 381(5): 43243 7. Jain P, Keating M, Wierda W, et al. Outcomes of patients with chronic lymphocytic leukemia after discontinuing ibrutinib. Blood. 2015;125(13):2062-2067. 8. Maddocks KJ, Ruppert AS, Lozanski G, et al. Etiology of ibrutinib therapy discontinuation and outcomes in patients with chronic lymphocytic leukemia. JAMA Oncol. 2015; 1(1):80-87. 9. Lampson BL, Brown JR. Are BTK and PLCG2 mutations necessary and sufficient for ibrutinib resistance in chronic lymphocytic leukemia? Expert Rev Hematol. 2018;11(3):185-194. 10. Teeling JL, Mackus WJ, Wiegman LJ, et al. The biological activity of human CD20 monoclonal antibodies is linked to unique epitopes on CD20. J Immunol. 2006;177(1): 362-371. 11. Jain P, Keating MJ, Wierda WG, et al. Longterm follow-up of treatment with ibrutinib

and rituximab in patients with high-risk chronic lymphocytic leukemia. Clin Cancer Res. 2017;23(9):2154-2158. 12. Burger JA, Sivina M, Jain N, et al. Randomized trial of ibrutinib versus ibrutinib plus rituximab in patients with chronic lymphocytic leukemia. Blood. 2019;33(10): 1011-1019. 13. Jaglowski SM, Jones JA, Nagar V, et al. Safety and activity of BTK inhibitor ibrutinib combined with ofatumumab in chronic lymphocytic leukemia: a phase 1b/2 study. Blood. 2015;126(7):842-850. 14. von Tresckow J, Cramer P, Bahlo J, et al. CLL2-BIG: sequential treatment with bendamustine, ibrutinib and obinutuzumab (GA101) in chronic lymphocytic leukemia. Leukemia. 2019;33(5):1161-1172. 15. Cramer P, von Tresckow J, Bahlo J, et al. CLL2-BXX phase-II trials: sequential, targeted treatment for eradication of minimal residual disease in chronic lymphocytic leukemia. Future Oncol. 2018;14(6):499513. 16. Hallek M. Signaling the end of chronic lymphocytic leukemia: new frontline treatment

553

P. Cramer et al. strategies. Blood. 2013;122(23):3723-3734. 17. Hallek M, Cheson BD, Catovsky D, et al. iwCLL guidelines for diagnosis, indications for treatment, response assessment, and supportive management of CLL. Blood. 2018;131(25):2745-2760. 18. Bottcher S, Ritgen M, Kneba M. Flow cytometric MRD detection in selected mature B-cell malignancies. Methods Mol Biol. 2013;971:149-174. 19. Rawstron AC, Villamor N, Ritgen M, et al. International standardized approach for flow cytometric residual disease monitoring in chronic lymphocytic leukaemia. Leukemia. 2007;21(5):956-964. 20. Bottcher S, Ritgen M, Fischer K, et al. Minimal residual disease quantification is an independent predictor of progressionfree and overall survival in chronic lymphocytic leukemia: a multivariate analysis from the randomized GCLLSG CLL8 trial. J Clin Oncol. 2012;30(9):980-988. 21. Goede V, Fischer K, Busch R, et al. Obinutuzumab plus chlorambucil in patients with CLL and coexisting conditions. N Engl J Med. 2014;370(12):11011110. 22. Hallek M, Fischer K, Fingerle-Rowson G, et al. Addition of rituximab to fludarabine and cyclophosphamide in patients with chronic lymphocytic leukaemia: a randomised, open-label, phase 3 trial. Lancet. 2010;376 (9747):1164-1174. 23. Eichhorst B, Fink AM, Bahlo J, et al. Firstline chemoimmunotherapy with bendamustine and rituximab versus fludarabine, cyclophosphamide, and rituximab in patients with advanced chronic lymphocyt-

554

ic leukaemia (CLL10): an international, open-label, randomised, phase 3, non-inferiority trial. Lancet Oncol. 2016;17(7):928942. 24. Keating MJ, O'Brien S, Albitar M, et al. Early results of a chemoimmunotherapy regimen of fludarabine, cyclophosphamide, and rituximab as initial therapy for chronic lymphocytic leukemia. J Clin Oncol. 2005;23(18):4079-4088. 25. Stilgenbauer S, Eichhorst B, Schetelig J, et al. Venetoclax in relapsed or refractory chronic lymphocytic leukaemia with 17p deletion: a multicentre, open-label, phase 2 study. Lancet Oncol. 2016;17(6):768-778. 26. Seymour JF, Ma S, Brander DM, et al. Venetoclax plus rituximab in relapsed or refractory chronic lymphocytic leukaemia: a phase 1b study. Lancet Oncol. 2017;18(2):230-240. 27. Roberts AW, Davids MS, Pagel JM, et al. Targeting BCL2 with venetoclax in relapsed chronic lymphocytic leukemia. N Engl J Med. 2016;374(4):311-322. 28. Byrd JC, Furman RR, Coutre SE, et al. Three-year follow-up of treatment-naive and previously treated patients with CLL and SLL receiving single-agent ibrutinib. Blood. 2015;125(16):2497-2506. 29. Byrd JC, Furman RR, Coutre SE, et al. Targeting BTK with ibrutinib in relapsed chronic lymphocytic leukemia. N Engl J Med. 2013;369(1):32-42. 30. Furman RR, Sharman JP, Coutre SE, et al. Idelalisib and rituximab in relapsed chronic lymphocytic leukemia. N Engl J Med. 2014;370(11):997-1007. 31. Cramer P, von Tresckow J, Bahlo J, et al.

Bendamustine followed by obinutuzumab and venetoclax in chronic lymphocytic leukaemia (CLL2-BAG): primary endpoint analysis of a multicentre, open-label, phase 2 trial. Lancet Oncol. 2018;19(9):12151228. 32. Wierda WG, Byrd JC, O'Brien S, et al. Tumour debulking and reduction in predicted risk of tumour lysis syndrome with single-agent ibrutinib in patients with chronic lymphocytic leukaemia. Br J Haematol. 2019;186(1):184-188. 33. Strati P, Sivina M, Kim E, et al. Achievement of complete remission (CR) as an endpoint for patients with chronic lymphocytic leukemia (CLL) treated with ibrutinib. J Clin Oncol. 2018;36(15_suppl): 7522. 34. von Tresckow J, Cramer P, Bahlo J, et al. CLL2-BIG: sequential treatment with bendamustine, ibrutinib and obinutuzumab (GA101) in chronic lymphocytic leukemia. Leukemia. 2019;33(5):1161-1172. 35. Cramer P, von Tresckow J, Bahlo J, et al. Durable remissions after discontinuation of combined targeted treatment in patients with chronic lymphocytic leukemia (CLL) harbouring a high-risk genetic lesion (del(17p)/TP53 mutation) Blood. 2018;132 (Suppl 1):694. 36. Woyach JA, Furman RR, Liu TM, et al. Resistance mechanisms for the Bruton's tyrosine kinase inhibitor ibrutinib. N Engl J Med. 2014;370(24):2286-2294. 37. Woyach JA, Ruppert AS, Guinn D, et al. BTK(C481S)-mediated resistance to ibrutinib in chronic lymphocytic leukemia. J Clin Oncol. 2017;35(13):1437-1443.

haematologica | 2021; 106(2)

ARTICLE

Myelodysplastic Syndromes

Genomic alterations in patients with somatic loss of the Y chromosome as the sole cytogenetic finding in bone marrow cells

Ferrata Storti Foundation

Madhu M. Ouseph,1° Robert P. Hasserjian,2 Paola Dal Cin,1 Scott B. Lovitch,1 David P. Steensma,3 Valentina Nardi2 and Olga K. Weinberg1,4,∞

Department of Pathology, Brigham and Women's Hospital; 2Department of Pathology, Massachusetts General Hospital; 3Department of Medical Oncology, Dana-Farber Cancer Institute and 4Department of Pathology, Boston Children’s Hospital, Boston, MA, USA

°Current address: Department of Pathology and Laboratory Medicine, Weill Cornell Medical College, New York, NY, USA

Haematologica 2021 Volume 106(2):555-564

Current address: UT Southwestern Medical Center, BioCenter, Dallas, TX, USA

∞

ABSTRACT

oss of the Y chromosome (LOY) is one of the most common somatic genomic alterations in hematopoietic cells in men. However, due to the high prevalence of LOY as the sole cytogenetic finding in the healthy older population, differentiating isolated LOY associated with clonal hematologic processes from aging-associated mosaicism can be difficult in the absence of definitive morphological features of disease. In the past, various investigators have proposed that a high percentage of metaphases with LOY is more likely to represent expansion of a clonal myeloid disease-associated population. It is unknown whether the proportion of metaphases with LOY is associated with the incidence of myeloid neoplasia-associated genomic alterations. To address this question, we identified bone marrow samples with LOY as an isolated cytogenetic finding and used targeted next generation sequencing-based molecular analysis to identify common myeloid neoplasia-associated somatic mutations. Among 73 patients with a median age of 75 years (range, 29-90), the percentage of metaphases with LOY was <25% in 23 patients, 25-49% in 10, 50-74% in 8 and ≥75% in 32. A threshold of ≥75% LOY was significantly associated with a morphological diagnosis of myeloid neoplasm (P=0.004). Furthermore, ≥75% LOY was associated with a higher lifetime incidence of a diagnosis of myelodysplastic syndromes (MDS) (P<0.0001), and in multivariate analysis ≥75% LOY was a statistically significant independent predictor of myeloid neoplasia (odds ratio 6.17; 95% confidence interval: 2.15-17.68; P=0.0007]. Higher LOY percentage (≥75%) was associated with greater likelihood of having somatic mutations (P=0.0009) and a higher number of these mutations (P=0.0002). Our findings indicate that ≥75% LOY in bone marrow cells is associated with an increased likelihood of molecular aberrations in genes commonly seen to be altered in myeloid neoplasia and with morphological features of MDS. These observations suggest that ≥75% LOY in bone marrow should be considered an MDS-associated cytogenetic aberration.

Introduction Loss of the Y chromosome (LOY) was first described in 1963 in cultured peripheral blood leukocytes.1 LOY is one of the most common acquired somatic genomic alterations in men; present in approximately 6% of all male bone marrow karyotypes and in 16% of karyotypes from abnormal marrows.2,3 The incidence of LOY in bone marrow cells increases proportionally with age, with up to 20% of healthy men over 80 years of age showing LOY by conventional marrow karyotyping.4,5 LOY is observed in association with myeloproliferative neoplasms (MPN), myelodysplastic syndromes (MDS), acute myeloid leukemias (AML) and B-cell haematologica | 2021; 106(2)

Correspondence: OLGA K. WEINBERG Olga.weinberg@UTSouthwestern.edu Received: October 22, 2019. Accepted: March 19, 2020. Pre-published: March 19, 2020. https://doi.org/10.3324/haematol.2019.240689

555

M.M. Ouseph et al.

lymphomas/leukemias, often in the context of other chromosomal abnormalities.2,6-8 LOY is one of the most common recurrent cytogenetic abnormalities seen in MDS, with an incidence of up to 30% in males.3,5,8-13 The incidence of isolated LOY in MDS is lower, ranging from 4 to 10%.8,14 Although there is no definitive association between LOY and clinical or biological characteristics in MDS, some groups have reported a higher incidence of ring sideroblasts, lower bone marrow myeloid to erythroid ratio, blast proportion, and blood leukocyte count in MDS patients with LOY than in those without LOY.9 MDS patients with isolated LOY have longer overall and leukemia-free survival compared to MDS patients with normal cytogenetics; conversely, in AML, chronic myeloid leukemia and chronic lymphocytic leukemia, isolated LOY has been associated with a poorer prognosis.2,14-20 These observations suggest that MDS with LOY might be associated with a unique biology. In individual cases, it can be difficult to distinguish whether LOY in MDS is a disease-associated alteration or just incidental aging-associated somatic mosaicism. In current World Health Organization (WHO) diagnostic criteria, LOY by itself cannot be used to define MDS, unlike other more strongly MDS-associated cytogenetic aberrations such as del(5q) or del(7q).21 Although the incidence of LOY increases with age, the percentage of metaphases with LOY in healthy people typically does not increase with age, so that a high proportion of metaphases may be more likely to represent a marker of disease rather than progressive clonal expansion of an inconsequential LOY cell population during aging.5,22 Some studies have suggested that 75-100% of metaphases with LOY in bone marrow are likely to be disease-associated rather than being age-associated.2,23,24 Recently, large genome-wide association studies enrolling more than 200,000 men in the UK Biobank demonstrated an association of germline single nucleotide polymorphisms of 156 autosomal loci as well as differential methylation of various genes with LOY.4,25-27 Whether any of the LOY-associated single nucleotide polymorphisms or epigenetic changes contribute to a risk of hematologic neoplasia is unclear. While LOY involving a higher proportion of metaphases has been consistently associated with MDS, it is uncertain whether there is a corresponding increased incidence of pathogenic MDS-associated mutations, which typically affect genes involved in DNA methylation, DNA damage repair, chromatin modification, RNA processing, transcription and signal transduction.28 In addition, although karyotype evolution has been reported in patients with LOY over time, the rate at which patients with LOY accumulate additional mutations is unknown.23 The goal of this study was to evaluate the landscape of somatic mutations in patients with LOY.

apy within the preceding 3 months were excluded, as were those without a matching marrow specimen for morphological evaluation. Medical records were reviewed to identify age at presentation, clinical diagnosis, and clinical course including treatment. Marrow aspirate smears and biopsy cores were re-evaluated by two hematopathologists (OW and MO) and were classified according to the revised 2017 WHO classification of myeloid neoplasms.21 Marrow samples that did not fulfill WHO-defined diagnostic criteria were separated into cases with minimal dysplasia (affecting <10% cells in any lineage) and no dysplasia. Fresh marrow aspirates at the time of initial presentation of LOY or concurrent blood samples were subjected to next-generation sequencing (NGS)-based molecular analysis for genes associated with hematologic neoplasms. For marrow samples that had not undergone NGS for clinical indications, NGS was performed on DNA extracted from the corresponding archived frozen cytogenetic cell pellets. NGS panels including 95 (Rapid Heme Panel of Brigham and Women’s Hospital) or 103 (Heme SnapShot Panel of Massachusetts General Hospital) genes of importance in myeloid neoplasia (hotspots in oncogenes and most of the coding regions of tumor suppressors) (Online Supplementary Table S1) were used for mutation analysis. The former is an amplicon-based approach using an Illumina Truseq Custom Amplicon kit (San Diego, CA, USA), with an average depth of 1500x.29 The latter is an anchored multiplex polymerase chain reaction-based panel (Archer DX, Boulder, CO, USA) with an average demultiplexed coverage of 350x.30 After filtering for recurrent assay specific false positives, single nucleotide variants and small insertions/deletions at an allele fraction of ≥0.05 were filtered based on population databases to eliminate likely germline variants. Somatic variants were further classified based on information available in literature, in silico tools, ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/), Catalog of Somatic Mutations in Cancer (https://cancer.sanger.ac.uk/cosmic) and published guidelines for the interpretation and reporting of sequence variants in cancer.31 Only likely pathogenic and pathogenic variants identified in patients’ samples were analyzed for this study. The selection of patients and analysis of clinical/laboratory results for this study were approved by the Partners Healthcare Institutional Review Board (Protocol #: 2009P001369). Statistical analysis was performed using JMP®, version 14 (SAS Institute Inc., Cary, NC, USA). A P-value of <0.05 was considered statistically significant. Individual P-values and tests performed are provided in the Results section. Briefly, in addition to descriptive statistics, analysis of distribution of categorical variables were performed using the Fisher exact test or Pearson χ2 test. Continuous variables were compared by one-way analysis of variance (ANOVA) or the Wilcoxon/Kruskal-Wallis test with χ2 approximation. A log-rank test was used to compare progression among patients with long-term follow-up. Multivariate analysis with a row-wise method for estimation of correlation and logistic regression analysis were used to compare predictor variables; and generate correlation coefficients, confidence intervals and correlation probabilities.

Methods Results The cytogenetic databases of Brigham and Women’s Hospital and Massachusetts General Hospital (Boston, MA, USA) were searched for patients with isolated LOY cytogenetic abnormality on conventional karyotyping of bone marrow cells (defined as ≥3 metaphases with isolated LOY) reported between 01/2005 and 08/2018. Patients with lymphoid or plasma cell malignancies involving bone marrow and those who had received chemother556

Identification of cases We identified 106 LOY cases reported as an isolated cytogenetic abnormality from 88 patients (Online Supplementary Figure S1A). Of these 106 samples, 15 were subsequently excluded from analysis because of lack of molecular analysis data (i.e., they failed the quality control haematologica | 2021; 106(2)

High-proportion LOY is likely an MDS associated aberration

for NGS assay). The final study set comprised the remaining 91 samples from 73 patients.

Age at presentation and extent of loss of chromosome Y Most (86%) patients were older than 65 years; the median age was 75 years (range, 29-90; mean 73 years) (Online Supplementary Figure S1B). Twenty-three patients had <25% metaphases with LOY, 10 had 25-49%, 8 had 5074%, and 32 had ≥75%. Bivariate analysis did not demon-

strate any significant correlation between age and percentage of metaphases involved by LOY (Table 1, Online Supplementary Figure S1C). All identified patients had undergone bone marrow evaluation for either a history of cytopenias or cytosis; 56 patients (77%) had at least one cytopenia at the time of marrow evaluation. There was no association between percentage of metaphases with LOY and affected lineage or severity of peripheral blood cytopenias.

Figure 1. Loss of chromosome Y in ≥75% metaphases is associated with morphological features of myeloid neoplasia and progression to myelodysplastic syndrome. (A) Mosaic graph demonstrating pathological diagnoses in relation to percentage of metaphases involved by loss of chromosome Y (LOY) ; divided into four bins for ease of comparison. ≥75% LOY is significantly associated with myeloid neoplasia; while <25% LOY is associated with no/minimal dysplasia. (B) Failure plot demonstrating presence or absence of evidence of progression in patients with LOY with no dysplasia/minimal dysplasia at presentation. Each dot represents diagnosis of myelodysplastic syndrome during follow up. The red line represents patients with ≥75% LOY, while the blue line represents patients with <75% LOY.

haematologica | 2021; 106(2)

557

M.M. Ouseph et al.

Loss of chromosome Y in ≥75% metaphases is significantly associated with a morphological diagnosis of a myeloid neoplasm, especially myelodysplastic syndrome A WHO-defined diagnosis was made in 39/73 patients (53% of the total): 35 patients were diagnosed with different subtypes of MDS (MDS with single lineage dysplasia, MDS with multilineage dysplasia, MDS with ring sideroblasts and multilineage dysplasia, MDS with excess blasts1, MDS with excess blasts-2, and therapy-related MDS), as shown in Table 1. Four patients were diagnosed with myelodysplastic/myeloproliferative neoplasm with ring sideroblasts and thrombocytosis. No other MDS/MPN subtypes were identified and none of the patients met criteria for MPN or AML. Among the remaining 34 patients who did not meet diagnostic criteria for any WHO-defined myeloid neoplasm, 14 had minimal dysplasia not meeting the 10% diagnostic threshold for a diagnosis of MDS and one patient had a hypocellular bone marrow suggestive of aplastic anemia. The remaining 19 patients had a morphologically normal bone marrow evaluation. Of note, 73% of patients with minimal bone marrow dysplasia not diagnostic of MDS had blood cytopenias at presentation, while only 30% of patients with a morphologically normal bone marrow study had peripheral blood cytopenias; however, this difference did not reach statistical significance due to small numbers (P=0.32; Wilcoxon/KruskalWallis test, χ2 approximation). There was a significant association between higher percentages of metaphases with LOY and diagnosis of myeloid neoplasia (Figure 1A). Bone marrow samples with ≥75% LOY had a very high likelihood of a morphological diagnosis of a myeloid neoplasm (P=0.004; Fisher exact test, Pearson χ2 P-value), while samples with <25% LOY were associated with no/minimal dysplasia (P=0.0484; Fisher exact test, Pearson χ2 P-value).

Loss of chromosome Y in ≥75% metaphases is significantly associated with progression to myelodysplastic syndrome

Subsequent bone marrow evaluation had been performed on 34 patients who did not meet criteria for a WHO-defined diagnosis of myeloid neoplasia at the initial presentation with LOY, with an interval of 0.4 to 152.8 months (median 34.1 months) after the initial bone marrow evaluation. In the subsequent marrow evaluation, 4/7 patients (57%) with ≥75% LOY had a marrow morphological diagnosis of MDS, compared to only 1/27 patients with <75% LOY (P<0.0001; log-rank test, χ2 = 22.979) (Figure 1B).

Loss of chromosome Y in ≥75% metaphases is significantly associated with mutations in myeloid neoplasia-related genes

Among the 73 LOY patients with NGS performed at presentation, 25/32 (78%) patients with ≥75% LOY had pathogenic mutations in tested myeloid neoplasia-associated genes. The frequency of mutations was associated with LOY burden: 5/8 (63%) patients with 50-74% LOY, 5/10 (50%) with 25-49% LOY, and 8/23 (35%) with <25% LOY had mutations (Table 2, Online Supplementary Tables S2 and S3). The mean number of mutations per patient was 2.2 in samples with ≥75% LOY, 1.3 in samples with 50-74% LOY, 0.8 in samples with 25-49% LOY, 558

Table 1. Age and pathological diagnoses in the study population.

<25% Age (years) 25-49 50-74 75-100 Pathological diagnosis No dysplasia Minimal dysplasia, not diagnostic of MDS MDS/t-MDS MDS-SLD MDS-MLD MDS-RS-MLD MDS-EB-1 MDS-EB-2 t-MDS MDS/MPN-RS-T

[%] LOY 25-49% 50-74%

≥75%

1 11 11

0 4 6

0 4 4

2 11 19

8 6

4 3

4 1

4 4

9 6 2 0 0 1 0 0

3 1 2 0 0 0 0 0

3 1 1 0 0 0 1 0

20 4 10 3 1 1 1 4

[%] LOY: percentage of metaphases with loss of Y chromosome; MDS: myelodysplastic syndrome; MDS-SLD: MDS with single lineage dysplasia; MDS-MLD: MDS with multilineage dysplasia; MDS-RS-MLD: MDS with ring sideroblasts and multilineage dysplasia; MDS-EB-1: MDS with excess blasts-1; MDS-EB-2: MDS with excess blasts-2; t-MDS: therapy-related MDS; MDS/MPN-RS-T: myelodysplastic/myeloproliferative neoplasm with ring sideroblasts and thrombocytosis.

and 0.6 in samples with <25% LOY. Bivariate analyses demonstrated that ≥75% LOY was significantly associated with greater likelihood of having somatic mutations commonly associated with myeloid neoplasm (P=0.0009; one-way ANOVA) and a higher number of such mutations (P=0.0002; one-way ANOVA) (Figure 2). Multivariate analysis demonstrated a positive correlation between total number of mutations as well as number of mutated genes and % LOY (row-wise method) (Figure 3, Online Supplementary Tables S4 and S5). In multivariate analysis using a logistic regression model with effective likelihood ratio test, % LOY was found to be a statistically significant predictor of diagnosis of myeloid neoplasia (odds ratio [OR] 1.03, 95% confidence interval [95% CI]: 1.02-1.04; P=0.0005) per 20% increase in LOY; OR 6.17 (95% CI: 2.15-17.68; P=0.0007) for ≥75% LOY) (Table 3). The most commonly mutated genes were TET2, SF3B1, U2AF1, ZRSR2 and ASXL1 (Table 2, Online Supplementary Tables S2 and 3). Analysis of variance demonstrated that the variant allele frequencies (VAF) for pathogenic variants were significantly higher in patients with ≥75% LOY than in those with <25% LOY (P=0.000027, F-ratio=9.4; one-way ANOVA) (Table 2). Patients with ≥75% LOY had an overall higher incidence (50%) of mutations in spliceosome components (SFB31, SRSF2 and U2AF1), JAK2 and RUNX1 compared to the incidence of 9% in patients with LOY <25%, 10% in those with 25-49% LOY and 13% in patients with 5074% LOY (P=0.011; Pearson χ2 test) (Table 2).32 In addition, patients with ≥75% LOY had a higher incidence (50%) of two or more mutations compared to the incidence of 9% of in patients with <25% LOY, 10% in those with 25-49% LOY and 38% in patients with 50-74% LOY (P=0.01; Pearson χ2 test) (Table 2). Furthermore, a large subset of patients with ≥75% LOY had these mutations in combination with mutations in TET2, DNMT3A and/or ASXL1 (28%), compared to 4% of those with <25% LOY, 10% of 25-49% LOY patients and 25% of 50-74% LOY patients (P=0.01; Pearson χ2 test) (Table 2). haematologica | 2021; 106(2)

High-proportion LOY is likely an MDS associated aberration

Figure 2. Loss of chromosome Y in ≥75% metaphases is associated with somatic mutations in myeloid neoplasia-related genes. (A) Box plot demonstrating percentage of metaphases with loss of chromosome Y (LOY) compared to number of myeloid neoplasia related genes mutated. (B) Scatter plots demonstrating relation between total number of mutations and pathologic diagnosis in patients with ≥75% LOY and <75%. MDS/MPN-RS-T: myelodysplastic/myeloproliferative neoplasm with ring sideroblasts and thrombocytosis; MDS/t-MDS: myelodysplastic syndrome/therapy-related myelodysplastic syndrome.

Mutations with a high VAF as well as presence of two or more mutations, especially including mutations in TET2, DNMT3A and/or ASXL1, are known to correlate strongly with the presence of MDS or to have a clinical course indistinguishable from that of MDS with the same mutation profile.32

Incidence of somatic mutations in patients with ≥75% loss of chromosome Y is higher than that in the age-matched general population

We compared the proportion of patients with any mutation in myeloid neoplasia-related genes to the incidence of clonal hematopoiesis of indeterminate potential compiled from the published literature (Table 4).33 The overall incidence of mutations in myeloid neoplasia-related genes in our cohort of patients with any LOY was higher (58%) than that seen in healthy individuals >70 years of age (1015%). Moreover, the incidence was much higher when only patients with ≥75% LOY were considered (81%), close to the incidence of somatic mutations seen in haematologica | 2021; 106(2)

patients with MDS (85-90% in most series with broad testing panels).

Progression of loss of chromosome Y over time We identified 30 sequential samples with matching NGS data from 12 patients for evaluation of changes over time. These samples were obtained at varying time intervals after the first (median 22 months; range, 7-72 months). Of these 12 patients, seven did not demonstrate morphological features diagnostic of myeloid neoplasia at initial presentation with LOY (Figure 4). On follow-up, four of these seven patients progressed to having overt MDS, all of whom showed either an increased percentage of metaphases with LOY or persistent 100% LOY. The remaining three patients did not demonstrate evidence of progression to a myeloid neoplasm, despite an increase in the percentage of LOY metaphases in two. Of the five patients who had morphological diagnostic features of MDS at presentation, three maintained 100% LOY, while the remaining two showed an increase in percentage of LOY over the follow-up period. 559

M.M. Ouseph et al.

Figure 3. Percentage of loss of chromosome Y demonstrates positive correlation with total number of mutations as well as number of mutated genes by multivariate analysis. Scatterplot matrix demonstrating correlation between age, percentage of metaphases with loss of chromosome Y, total number of mutations and number of genes mutated. Correlation coefficient for individual correlations provided in each scatterplot. Horizontal bar graphs demonstrating distribution of each parameter is also shown. LOY: loss of Y chromosome.

The number of somatic mutations identified by NGS was variable for patients over the follow-up period. Two out of three patients who did not demonstrate progression to MDS showed a loss of detectable mutations (with an initial KDM5A mutation with a VAF of 0.05 and PPM1D with a VAF of 0.06) while the remaining patient showed no change in number of mutations. Of the four patients who showed progression to MDS, three had an increase in the number of mutations, while one patient did not demonstrate any change in number of mutations. Among five patients who had diagnostic morphological features of MDS at presentation, one patient showed a loss of detectable mutations (CUX1 with an initial VAF of 0.14) 560

with persistence of diagnostic MDS morphological features, while one patient did not demonstrate any change in number of mutations over time. The remaining three patients showed increases in the numbers of mutations over the follow-up period. The mutations that accumulated over time in patients who had a diagnosis of MDS occurred in TET2 (2 patients, VAF 0.18 and 0.21), ZRSR2 (1 patient, VAF 0.14), RUNX1 (1 patient, VAF 0.05), ASXL1 (2 patients, VAF 0.36 and 0.09), SETBP1 (2 patients; VAF 0.24 and 0.18), STAG2 (1 patient; VAF 0.23), IDH1 (1 patient, VAF 0.31), STAT3 (1 patient, VAF 0.37), CBL (1 patient, VAF 0.23), CBLB (1 patient, VAF 0.13) and PHF6 (1 patient; VAF 0.29). haematologica | 2021; 106(2)

High-proportion LOY is likely an MDS associated aberration

Table 2. Loss of chromosome Y is associated with somatic mutations that are frequently present in myeloid neoplasia.

<25% (23) Number of somatic mutations (number of patients with mutations; mean VAF) TET2 SF3B1 U2AF1 ZRSR2 ASXL1 JAK2 SETBP1 CBL SH2B3 STAG2 Total number of somatic mutations (number of patients with mutations) Number of patients with mutations in spliceosome components, JAK2 and RUNX1 (%) Number of patients with â&#x2030;Ľ2 mutations (%) Number of patients with combinations of mutations in TET2, DNMT3A and/or ASXL1 (%)

[%] LOY (total number of patients) 25-49% (10) 50-74% (8)

â&#x2030;Ľ75% (32)

2 (1; 0.33) 1 (1; 0.23) 0 0 1 (1; 0.43) 0 0 0 0 0 13 (7) 2 (8.7)

0 0 0 0 1 (1; 0.05) 1 (1; 0.46) 0 0 0 0 7 (5) 1 (10)

5 (4; 0.36) 1 (1; 0.41) 0 0 1 (1; 0.07) 0 0 1 (1; 0.49) 0 0 9 (5) 1 (12.5)

12 (9; 0.40) 10 (10; 0.36) 4 (4; 0.25) 4 (3; 0.48) 4 (3; 0.32) 3 (3; 0.37) 2 (2; 0.41) 2 (2; 0.27) 2 (2; 0.27) 2 (2; 0.31) 65 (25) 16 (50)

2 (8.7) 1 (4.4)

1 (10) 1 (10)

3 (37.5) 2 (25)

16 (50) 9 (28.1)

Number of mutations and variant allele fraction in commonly mutated myeloid neoplasia-related genes. For details please see Online Supplementary Tables S2 and S3. [%] LOY: percentage of metaphases with loss of Y chromosome; VAF: variant allele fraction.

Discussion LOY as a sole cytogenetic finding is one of the most common somatic genomic alterations in blood and marrow of elderly men.5,34 LOY as a post-zygote genomic alteration occurs at varying rates in different tissues, including non-hematopoietic cells.5,35,36 Interest in LOY is increasing because of the mounting evidence of a role of LOY in various disorders including atherosclerosis, latestage age-related macular degeneration, Alzheimer disease, and autoimmune disorders such as primary biliary cirrhosis and autoimmune thyroiditis.36-43 In addition, mosaic LOY in blood has been associated with a higher risk of all-cause mortality and overall shorter life expectancy.39,43-45 LOY in blood is associated with an increased incidence of concurrent diagnosis of nonhematopoietic malignancies such as head and neck, colorectal, prostatic and pancreatic carcinomas and may contribute to the increased incidence of some tumors in males.39,46-53 LOY in neoplastic tissues has been reported in various solid tumors, including carcinomas of urothelium, pancreas, esophagus, head and neck, and kidney.46,49,54-57 LOY is also one of the most common recurrent cytogenetic abnormalities seen in MDS. As MDS is predominantly observed in older people, it is difficult to separate age-associated LOY from disease-associated LOY. Although marrow sampling has an inherent age bias since the incidence of myeloid neoplasms increases with age, marrow samples taken from young patients evaluated for hematologic disorders do not demonstrate the same high incidence of LOY as that seen in the elderly.34 Consistent with published literature, in our cohort most patients were older than 65 years, with a median age of 75 years. In an analysis conducted in 1992, in which conventional karyotyping was used, the incidence of LOY was estimated at approximately 11% in patients with MDS and 4% in those with AML.8 Subsequently, in 1997, GarciaIsidoro et al. demonstrated the presence of LOY in 29% haematologica | 2021; 106(2)

Table 3. Multivariate analysis demonstrates that percentage loss of chromosome Y is a statistically significant independent predictor of a diagnosis of myeloid neoplasia.

P-value (effective likelihood ratio test)

Characteristic

Odds ratio (95% CI)

% LOY

1.026 (1.02-1.04) per 20% increase 8.83 (2.51-34.89) for entire range 6.17 (2.15-17.68) 0.99 (0.95-1.03) per 10 years 0.60 (0.04-7.87) for entire range 2 (0.78-5.15)

0.0005

1.21 (0.86-1.71) per gene mutated 3.12 (0.40-24.53) for entire range 1.22 (0.90-1.65) per mutation 4.02 (0.48-33.35) for entire range

0.27

â&#x2030;Ľ75% LOY Age Presence of mutations Number of genes mutated Total number of mutations

0.0007 0.69 0.15

0.18

Multivariate analysis using a logistic regression model for prediction of a diagnosis of myeloid neoplasia with odds of diagnosis of myeloid neoplasia, 95% confidence interval and P-value. 95% CI: 95% confidence interval; % LOY: percentage of metaphases with loss of Y chromosome).

of male patients diagnosed with MDS using fluorescence in situ hybridization studies on bone marrow.9 Some studies have suggested that the proportion of metaphases with LOY could be used to separate ageassociated LOY from disease-associated LOY. In previous analyses, the degree of LOY was not correlated with the severity of peripheral blood cytopenias.58,59 In an analysis published in 2000, Wiktor et al. identified that using 81% Y chromosome loss as a cut-off maximized the combined sensitivity (28%) and specificity (100%) for predicting the presence of disease states associated with LOY.2 In a subsequent study in 2011 by Wiktor and colleagues, restricted to men >50 years old,23 a myeloid neoplasm (MDS, MPN and AML) could be diagnosed in 64/161 patients with >75% metaphases with LOY.23 In another series, a cutoff of 21.5% LOY in CD34-positive blood cells was proposed to discriminate between age561

M.M. Ouseph et al.

and disease-associated LOY in MDS, but CD34 selection for chromosome analysis is not used in the clinical setting.14 In our series, marrow samples with ≥75% LOY had a high likelihood of morphological diagnosis of myeloid neoplasia, most commonly MDS. Our data also suggest that patients with ≥75% LOY who initially have a nondiagnostic marrow have a higher likelihood of progression to a morphological diagnosis of MDS on follow-up. The presence of <25% LOY metaphases, on the other hand, was associated with normal bone marrow morphology and a low likelihood of progression to MDS. These observations further support the notion that LOY involving a high proportion of metaphases might be a disease-associated genomic alteration. To our knowledge, no published study has evaluated the presence of myeloid-type mutations in patients with isolated LOY. In our cohort, patients with ≥75% LOY had a high prevalence (>80%) of somatic mutations associated with myeloid neoplasia, especially MDS. Although these mutations, as with LOY, can be seen in aging populations (so-called age-related clonal hematopoiesis or clonal hematopoiesis of indeterminate potential), the incidence of these frequent alterations was much higher when LOY was ≥75% compared to the incidence found in age-matched populations from the literature. These findings suggest that ≥75% LOY is associated with an MDS-type mutation signature and therefore likely represents disease-associated clonal proliferation.22 The molecular mechanisms behind the association of LOY, aging and neoplasia are not completely understood. There is no clear association between LOY and numerical abnormalities of other chromosomes.9,27,53 LOY does not seem to reflect an overall propensity to loss of small-sized chromosomes.60,61 Alternative proposed causes for a high

Table 4. Prevalence of mutations in myeloid neoplasia-related genes in the cohort with ≥75% loss of chromosome Y is comparable to the reported prevalence in patients with myelodysplastic syndrome.

Mutation frequency Myelodysplastic Current Current ≥75% Healthy syndrome LOY cohort LOY cohort individuals >70 years Proportion of patients 72-90% with any mutation Splicing factor genes SF3B1 15-32% SRSF2 10-17% U2AF1 7-12% ZRSR2 3-11% DNA methylation genes TET2 20-32% DNMT3A 5-13% Histone modification genes ASXL1 11-23% EZH2 5-12% Transcription factors SETBP1 2-5% BCOR/BCORL 4% Cohesin complex genes STAG1/STAG2 4-7% Signaling genes NRAS/KRAS 5-10% CBL 2-5% JAK2 2-5% Tumor suppressor gene TP53 5-10%

58%

81%

5-10%

12% 2% 4% 4%

32% 3% 13% 10%

0.1-0.2% <0.1% <0.1%

13% 4%

26% 3%

0.3-0.4% 1.5-2%

6% 2%

10% 7%

0.3-0.4%

2% 3%

7% 3%

2% 3% 4%

3% 7% 7%

<0.1% 0.2%

<0.1%

Comparison of mutation frequency in patients with loss of chromosome Y with a reported incidence of mutations in the healthy aging population and patients with myelodysplastic syndrome.The table is adapted from Table 45-3 of the paper by Hasserjian et al.33 % LOY: percentage of metaphases with loss of Y chromosome.

Figure 4. Progression of somatic mutations in 12 patients with loss of chromosome Y. Striped bars represent a pathological diagnosis of no dysplasia or minimal dysplasia, while solid bars represent pathological diagnosis of myelodysplastic syndrome (Each alphabet letter corresponds to one patient). The height of bars represents the percentage of metaphases with loss of chromosome Y. Black dots on bars highlight number of mutations identified for the corresponding samples. LOY: loss of Y chromosome.

562

<0.1%

haematologica | 2021; 106(2)

High-proportion LOY is likely an MDS associated aberration

incidence of LOY with aging include the presence of shortened telomeres as well as late S-phase replication of the Y chromosome.38,62 Smoking appears to be the most common epidemiological association with mosaic LOY in the bone marrow.4,25,63 It is unclear how LOY contributes to the phenotypes associated with LOY. LOY may not be limited to myeloid lineages, as more frequent LOY has been reported in CD3-positive cells in the blood of elderly patients with MDS compared to those without MDS.14 While some investigators have suggested an immune regulatory component, deregulation of proteins encoded by the male-specific region of the Y chromosome region has been seen in neoplasms with LOY, raising the possibility of a direct function of these proteins in carcinogenesis.48,64,65 In vitro models have demonstrated a tumor suppressor role of the Y chromosome in the context of prostatic carcinoma.66 Furthermore, identification of tumor suppressor roles of ZFY and UTY, which localize to the male-specific part of Y, suggests a more active role of LOY in oncogenesis.53 Both these genes have homologues on the X chromosome which escape X inactiva-

References 1. Jacobs PA, Brunton M, Court Brown WM, Doll R, Goldstein H. Change of human chromosome count distribution with age: evidence for a sex differences. Nature. 1963;197:1080-1081. 2. Wiktor A, Rybicki BA, Piao ZS, et al. Clinical significance of Y chromosome loss in hematologic disease. Genes Chromosomes Cancer. 2000;27(1):11-16. 3. Shahrabi S, Khodadi E, Saba F, Shahjahani M, Saki N. Sex chromosome changes in leukemia: cytogenetics and molecular aspects. Hematology. 2018;23(8):139-147. 4. Wright DJ, Day FR, Kerrison ND, et al. Genetic variants associated with mosaic Y chromosome loss highlight cell cycle genes and overlap with cancer susceptibility. Nat Genet. 2017;49(5):674-679. 5. Pierre RV, Hoagland HC. Age-associated aneuploidy: loss of Y chromosome from human bone marrow cells with aging. Cancer. 1972;30(4):889-894. 6. Zhang LJ, Shin ES, Yu ZX, Li SB. Molecular genetic evidence of Y chromosome loss in male patients with hematological disorders. Chin Med J (Engl). 2007;120(22):2002-2005. 7. Abe S, Golomb HM, Rowley JD, Mitelman F, Sandberg AA. Chromosomes and causation of human cancer and leukemia. XXXV. The missing Y in acute non-lymphocytic leukemia (ANLL). Cancer. 1980;45(1):84-90. 8. Loss of the Y chromosome from normal and neoplastic bone marrows. United Kingdom Cancer Cytogenetics Group (UKCCG). Genes Chromosomes Cancer. 1992;5(1):8388. 9. Garcia-Isidoro M, Tabernero MD, Najera ML, et al. Clinical and biological characteristics of myelodysplastic syndromes with nulisomy Y by FISH. Haematologica. 1997;82(5):537-541. 10. Haase D, Fonatsch C, Freund M, et al. Cytogenetic findings in 179 patients with myelodysplastic syndromes. Ann Hematol. 1995;70(4):171-87. 11. Levan G, Mitelman F. Clustering of aberra-

haematologica | 2021; 106(2)

tion.53,67 However, constitutional 45, XO (Turner syndrome) is not associated with an increased incidence of myeloid neoplasms.68 In conclusion, our findings demonstrate that a high proportion of metaphases with LOY (≥75% metaphases) in marrow is associated with a high frequency of molecular alterations in genes commonly mutated in myeloid neoplasia and strongly associated with a contemporaneous or subsequent diagnosis of MDS. These observations suggest that high-proportion LOY is likely to be a true MDSassociated cytogenetic aberration rather than an incidental finding due to aging. Disclosures No conflicts of interest to diclose. Contributions MO collected cases and compiled data. MO and OW reviewed cases. OW, RH and VN contributed cases. VN and SL analyzed NGS data. PC oversaw cytogenetic testing. DS provided clinical information related to cases. All authors contributed to the preparation of the manuscript and the revision.

tions to specific chromosomes in human neoplasms. Hereditas. 1975;79(1):156-160. 12. Suciu S, Kuse R, Weh HJ, Hossfeld DK. Results of chromosome studies and their relation to morphology, course, and prognosis in 120 patients with de novo myelodysplastic syndrome. Cancer Genet Cytogenet. 1990;44(1):15-26. 13. Nomdedeu M, Pereira A, Calvo X, et al. Clinical and biological significance of isolated Y chromosome loss in myelodysplastic syndromes and chronic myelomonocytic leukemia. A report from the Spanish MDS Group. Leuk Res. 2017;63:85-89. 14. Ganster C, Kampfe D, Jung K, et al. New data shed light on Y-loss-related pathogenesis in myelodysplastic syndromes. Genes Chromosomes Cancer. 2015;54(12):717724. 15. Lippert E, Etienne G, Mozziconacci MJ, et al. Loss of the Y chromosome in Philadelphiapositive cells predicts a poor response of chronic myeloid leukemia patients to imatinib mesylate therapy. Haematologica. 2010;95(9):1604-1607. 16. Holmes RI, Keating MJ, Cork A, Trujillo JM, McCredie KB, Freireich EJ. Loss of the Y chromosome in acute myelogenous leukemia: a report of 13 patients. Cancer Genet Cytogenet. 1985;17(3):269-278. 17. Chapiro E, Antony-Debre I, Marchay N, et al. Sex chromosome loss may represent a disease-associated clonal population in chronic lymphocytic leukemia. Genes Chromosomes Cancer. 2014;53(3):240-247. 18. Greenberg P, Cox C, LeBeau MM, et al. International scoring system for evaluating prognosis in myelodysplastic syndromes. Blood. 1997;89(6):2079-2088. 19. Slovak ML, Kopecky KJ, Cassileth PA, et al. Karyotypic analysis predicts outcome of preremission and postremission therapy in adult acute myeloid leukemia: a Southwest Oncology Group/Eastern Cooperative Oncology Group Study. Blood. 2000;96(13): 4075-4083. 20. Schanz J, Tüchler H, Solé F, et al. New comprehensive cytogenetic scoring system for primary myelodysplastic syndromes (MDS)

and oligoblastic acute myeloid leukemia after MDS derived from an international database merge. J Clin Oncol. 2012;30(8): 820-829. 21. Swerdlow SH. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. International Agency for Research on Cancer. 2017. 22. Danielsson M, Halvardson J, Davies H, et al. Longitudinal changes in the frequency of mosaic chromosome Y loss in peripheral blood cells of aging men varies profoundly between individuals. Eur J Hum Genet. 2020;28(3):349-357. 23. Wiktor AE, Van Dyke DL, Hodnefield JM, Eckel-Passow J, Hanson CA. The significance of isolated Y chromosome loss in bone marrow metaphase cells from males over age 50 years. Leuk Res. 2011;35(10):1297-300. 24. Wong AK, Fang B, Zhang L, Guo X, Lee S, Schreck R. Loss of the Y chromosome: an age-related or clonal phenomenon in acute myelogenous leukemia/myelodysplastic syndrome? Arch Pathol Lab Med. 2008;132(8):1329-1332. 25. Zhou W, Machiela MJ, Freedman ND, et al. Mosaic loss of chromosome Y is associated with common variation near TCL1A. Nat Genet. 2016;48(5):563-568. 26. Terao C, Momozawa Y, Ishigaki K, et al. GWAS of mosaic loss of chromosome Y highlights genetic effects on blood cell differentiation. Nat Commun. 2019;10(1):4719. 27. Thompson DJ, Genovese G, Halvardson J, et al. Genetic predisposition to mosaic Y chromosome loss in blood. Nature. 2019;575(7784):652-657. 28. Ogawa S. Genetics of MDS. Blood. 2019;133(10):1049-1059. 29. Kluk MJ, Lindsley RC, Aster JC, et al. Validation and implementation of a custom next-generation sequencing clinical assay for hematologic malignancies. J Mol Diagn. 2016;18(4):507-515. 30. Zheng Z, Liebers M, Zhelyazkova B, et al. Anchored multiplex PCR for targeted nextgeneration sequencing. Nat Med. 2014;20(12):1479-1484.

563

M.M. Ouseph et al. 31. Li MM, Datto M, Duncavage EJ, et al. Standards and guidelines for the interpretation and reporting of sequence variants in cancer: a joint consensus recommendation of the Association for Molecular Pathology, American Society of Clinical Oncology, and College of American Pathologists. J Mol Diagn. 2017;19(1):4-23. 32. Malcovati L, Galli A, Travaglino E, et al. Clinical significance of somatic mutation in unexplained blood cytopenia. Blood. 2017;129(25):3371-3378. 33. Hasserjian RP, Head DR. Myelodysplastic syndromes. In: Jaffe EA, Arber DA; Campo E; Harris, NL; Quintanilla-Martinez, L, editors Hematopathology. Second edition Elsevier, Inc. 2017:793-815. 34. Nowinski GP, Van Dyke DL, Tilley BC, et al. The frequency of aneuploidy in cultured lymphocytes is correlated with age and gender but not with reproductive history. Am J Hum Genet. 1990;46(6):1101-1111. 35. Forsberg LA, Halvardson J, RychlickaBuniowska E, et al. Mosaic loss of chromosome Y in leukocytes matters. Nat Genet. 2019;51(1):4-7. 36. Haitjema S, Kofink D, van Setten J, et al. Loss of Y chromosome in blood is associated with major cardiovascular events during follow-up in men after carotid endarterectomy. Circ Cardiovasc Genet. 2017;10(4): e001544. 37. Lleo A, Oertelt-Prigione S, Bianchi I, et al. Y chromosome loss in male patients with primary biliary cirrhosis. J Autoimmun. 2013;41:87-91. 38. Persani L, Bonomi M, Lleo A, et al. Increased loss of the Y chromosome in peripheral blood cells in male patients with autoimmune thyroiditis. J Autoimmun. 2012;38(23):J193-196. 39. Forsberg LA, Rasi C, Malmqvist N, et al. Mosaic loss of chromosome Y in peripheral blood is associated with shorter survival and higher risk of cancer. Nat Genet. 2014;46(6):624-628. 40. Charchar FJ, Bloomer LD, Barnes TA, et al. Inheritance of coronary artery disease in men: an analysis of the role of the Y chromosome. Lancet. 2012;379(9819):915-922. 41. Grassmann F, Kiel C, den Hollander AI, et al. Y chromosome mosaicism is associated with age-related macular degeneration. Eur J Hum Genet. 2019;27(1):36-41. 42. Grassmann F, International AMDGC, Weber BHF, Veitia RA. Insights into the loss of the Y chromosome with age in control individuals and in patients with age-related macular

564

degeneration using genotyping microarray data. Hum Genet. 2020;139(3):401-407. 43. Dumanski JP, Lambert JC, Rasi C, et al. Mosaic loss of chromosome Y in blood is associated with Alzheimer disease. Am J Hum Genet. 2016;98(6):1208-1219. 44. Forsberg LA. Loss of chromosome Y (LOY) in blood cells is associated with increased risk for disease and mortality in aging men. Hum Genet. 2017;136(5):657-663. 45. Loftfield E, Zhou W, Graubard BI, et al. Predictors of mosaic chromosome Y loss and associations with mortality in the UK Biobank. Sci Rep. 2018;8(1):12316. 46. Veiga LC, Bergamo NA, Reis PP, Kowalski LP, Rogatto SR. Loss of Y-chromosome does not correlate with age at onset of head and neck carcinoma: a case-control study. Braz J Med Biol Res. 2012;45(2):172-178. 47. Noveski P, Madjunkova S, Sukarova Stefanovska E, et al. Loss of Y chromosome in peripheral blood of colorectal and prostate cancer patients. PLoS One. 2016;11(1):e0146264. 48. Kido T, Lau YF. Roles of the Y chromosome genes in human cancers. Asian J Androl. 2015;17(3):373-380. 49. Hunter S, Gramlich T, Abbott K, Varma V. Y chromosome loss in esophageal carcinoma: an in situ hybridization study. Genes Chromosomes Cancer. 1993;8(3):172-177. 50. Park SJ, Jeong SY, Kim HJ. Y chromosome loss and other genomic alterations in hepatocellular carcinoma cell lines analyzed by CGH and CGH array. Cancer Genet Cytogenet. 2006;166(1):56-64. 51. Cook MB, McGlynn KA, Devesa SS, Freedman ND, Anderson WF. Sex disparities in cancer mortality and survival. Cancer Epidemiol Biomarkers Prev. 2011;20(8): 1629-1637. 52. Edgren G, Liang L, Adami HO, Chang ET. Enigmatic sex disparities in cancer incidence. Eur J Epidemiol. 2012;27(3):187-196. 53. Davoli T, Xu AW, Mengwasser KE, et al. Cumulative haploinsufficiency and triplosensitivity drive aneuploidy patterns and shape the cancer genome. Cell. 2013;155 (4):948-962. 54. Minner S, Kilgue A, Stahl P, et al. Y chromosome loss is a frequent early event in urothelial bladder cancer. Pathology. 2010;42(4): 356-359. 55. Wallrapp C, Hahnel S, Boeck W, et al. Loss of the Y chromosome is a frequent chromosomal imbalance in pancreatic cancer and allows differentiation to chronic pancreati-

tis. Int J Cancer. 2001;91(3):340-344. 56. Klatte T, Rao PN, de Martino M, et al. Cytogenetic profile predicts prognosis of patients with clear cell renal cell carcinoma. J Clin Oncol. 2009;27(5):746-753. 57. Bianchi NO. Y chromosome structural and functional changes in human malignant diseases. Mutat Res 2009;682(1):21-27. 58. Germing U, Strupp C, Giagounidis A, et al. Evaluation of dysplasia through detailed cytomorphology in 3156 patients from the Dusseldorf Registry on myelodysplastic syndromes. Leuk Res. 2012;36(6):727-734. 59. Maassen A, Strupp C, Giagounidis A, et al. Validation and proposals for a refinement of the WHO 2008 classification of myelodysplastic syndromes without excess of blasts. Leuk Res. 2013;37(1):64-70. 60. Duijf PH, Schultz N, Benezra R. Cancer cells preferentially lose small chromosomes. Int J Cancer. 2013;132(10):2316-2326. 61. Machiela MJ, Zhou W, Karlins E, et al. Female chromosome X mosaicism is agerelated and preferentially affects the inactivated X chromosome. Nat Commun. 2016;7:11843. 62. Stone JF, Sandberg AA. Sex chromosome aneuploidy and aging. Mutat Res. 1995;338 (1-6):107-113. 63. Dumanski JP, Rasi C, Lonn M, et al. Mutagenesis. Smoking is associated with mosaic loss of chromosome Y. Science. 2015;347(6217):81-83. 64. Kersemaekers AM, Honecker F, Stoop H, et al. Identification of germ cells at risk for neoplastic transformation in gonadoblastoma: an immunohistochemical study for OCT3/4 and TSPY. Hum Pathol. 2005;36(5):512-521. 65. Li Y, Tabatabai ZL, Lee TL, et al. The Yencoded TSPY protein: a significant marker potentially plays a role in the pathogenesis of testicular germ cell tumors. Hum Pathol. 2007;38(10):1470-1481. 66. Vijayakumar S, Garcia D, Hensel CH, et al. The human Y chromosome suppresses the tumorigenicity of PC-3, a human prostate cancer cell line, in athymic nude mice. Genes Chromosomes Cancer. 2005;44(4): 365-372. 67. Schneider-Gadicke A, Beer-Romero P, Brown LG, Nussbaum R, Page DC. ZFX has a gene structure similar to ZFY, the putative human sex determinant, and escapes X inactivation. Cell. 1989;57(7):1247-1258. 68. Hasle H. Turner syndrome and myelodysplastic syndrome. No reason to alert. J Pediatr Hematol Oncol. 1997;19(2):179.

haematologica | 2021; 106(2)

ARTICLE

Acute Myeloid Leukemia

Synergistic activity of IDH1 inhibitor BAY1436032 with azacitidine in IDH1 mutant acute myeloid leukemia Anuhar Chaturvedi,1 Charu Gupta,1 Razif Gabdoulline,1 Nora M. Borchert,1 Ramya Goparaju,1 Stefan Kaulfuss,2 Kerstin Görlich,1 Renate Schottmann,1 Basem Othman,1 Julia Welzenbach,3 Olaf Panknin,2 Markus Wagner,2 Robert Geffers,4 Arnold Ganser,1 Felicitas Thol,1 Michael Jeffers,5 Andrea Haegebarth2 and Michael Heuser1

Department of Hematology, Hemostasis, Oncology and Stem Cell Transplantation, Hannover Medical School, Hannover, Germany; 2Bayer AG, Berlin, Germany; 3Institute of Human Genetics, University of Bonn, School of Medicine & University Hospital Bonn, Bonn, Germany; 4Genome Analytics Research Group, Helmholtz Center for Infection Research, Braunschweig, Germany and 5Bayer AG, Whippany, NJ, USA 1

Ferrata Storti Foundation

Haematologica 2021 Volume 106(2):565-573

ABSTRACT

utant isocitrate dehydrogenase 1 (mIDH1) inhibitors have shown single-agent activity in relapsed/refractory acute myeloid leukemia (AML), even though most patients eventually relapse. We evaluated the efficacy and molecular mechanism of the combination treatment with azacitidine, which is currently the standard of care in older AML patients, and mIDH1 inhibitor BAY1436032. Both compounds were evaluated in vivo as single agents and in combination with sequential (azacitidine, followed by BAY1436032) or simultaneous application in two human IDH1 mutated AML xenograft models. Combination treatment significantly prolonged survival compared to single agent or control treatment (P<0.005). The sequential combination treatment depleted leukemia stem cells by 470-fold. Interestingly, the simultaneous combination treatment depleted leukemia stem cells by 33,150-fold compared to control mice. This strong synergy is mediated through inhibition of MAPK/ERK and Rb/E2F signaling. Our data strongly argues for the concurrent application of mIDH1 inhibitors and azacitidine and predicts improved outcome of this regimen in IDH1 mutated AML patients.

Introduction In acute myeloid leukemia (AML) patients, mutated isocitrate dehydrogenase 1 (IDH1) is found in about 6-10% of patients1,2 at amino acid arginine 132 in the active site where isocitrate and NADPH bind.3 Whereas the enzymatic function of mutant IDH to convert isocitrate to α-ketoglutarate is impaired, it gains a neomorphic function to convert α-ketoglutarate to R-2-hydroxyglutarate (R-2HG), which induces histone- and DNA hypermethylation through inhibition of demethylation, and leads to a block in cellular differentiation thus promoting tumorigenesis.3-6 IDH1 mutant-specific (mIDH1) inhibitors have emerged as a potent tool for the treatment of IDH1 mutant AML. A recent report on the first clinical IDH1 inhibitor ivosidenib as a single agent in IDH1-mutated relapsed or refractory AML showed an overall response rate of 41.6% and a complete remission rate of 21.6% with a median duration of response of 8.2 months.7 While these results are promising in this difficult to treat patient setting, they also suggest that mIDH1 inhibitors should be combined with other agents to improve efficacy. A possible mechanism of synergy in IDH1 mutant AML is the reversal of DNA methylation by hypomethylating agents. 5-azacitidine (AZA) is a hypomethylating agent and can activate key epigenetically silenced pathways in AML cells, leading to an arrest of AML cell proliferation.8 The AZA-AML-001 study showed that azacitidine has clear activity in AML as monotherapy in patients with unfavorable cytogenetics.9 Therefore, a successful reversal of IDH1 mutant induced DNA hypermethylation in AML cells represents a window of opportunity to enhance the efficacy of mIDH1 inhibitors. We have previously shown that BAY1436032 is a highly effective haematologica | 2021; 106(2)

Correspondence: MICHAEL HEUSER heuser.michael@mh-hannover.de ANUHAR CHATURVEDI chaturvedi.anuhar@mh-hannover.de Received: September 9, 2019. Accepted: March 26, 2020. Pre-published: April 2, 2020. https://doi.org/10.3324/haematol.2019.236992

565

A. Chaturvedi et al.

oral pan-mutant IDH1 inhibitor, which has strong antileukemic activity in patient-derived xenograft (PDX) models of IDH1 mutant leukemia in vivo by inducing differentiation of leukemic blasts as well as inhibition of leukemia stem cell proliferation and self-renewal.10 BAY1436032 is currently evaluated in phase I clinical trials in AML and glioma patients (NCT03127735 and NCT02746081). In this study, we evaluated the therapeutic efficacy and elucidated the mechanism of action of the combined treatment of azacitidine and BAY1436032 in IDH1 mutant AML in vitro as well as in vivo in two independent patient-derived xenograft models of IDH1 mutant AML. From these experiments, we conclude that the simultaneous combination of a mIDH inhibitor with a hypomethylating agent synergistically inhibits LSC through suppression of MAP kinase and RB/E2F signaling, which are involved in cell survival and proliferation.

groups were treated with BAY1436032 and azacitidine in the doses mentioned above either starting both drugs on day 1 (simultaneous group) or starting azacitidine on day 1 but BAY1436032 on day 6 (sequential group). In the second cycle in the sequential treatment group, BAY1436032 was stopped during azacitidine treatment to better model the biologic and therapeutic effects of the synergy. The treatment was stopped at day 84 or day 96. The proportion of leukemic cells in the peripheral blood of mice was measured with the human-specific CD45 antibody every 4 weeks by tail vein bleeds and fluorescence-activated cell sorting (FACS) analysis.13 Blood counts were performed using an ABC Vet Automated Blood counter (Scil animal care company GmbH, Viernheim, Germany). All animal experiments were started with 10 animals, however, the animals which died before the start of treatment due to engraftment failure, during injections or during bleeding were excluded from the study.

Statistical analysis Methods Combination index Drug synergy was evaluated using a combination index (CI) equation based on the multiple drug-effect equation of ChouTalalay.11,12 Colony-forming cell units were assayed in methylcellulose after treatment with varying doses of either azacitidine ( 62.5, 125, 250, 500, 1,000 or 2,000 nm), BAY 1436032 ( 6.25, 12.5, 25, 50, 100 or 200 nm) or a fixed ratio of the combination of azacitidine/BAY1436032 (62.5/6.25, 125/12.5, 250/25, 500/50, 1,000/100 or 2,000/200 nm). The analysis was performed with CompuSyn software (ComboSyn Inc., Paramus, USA).

Two-tailed unpaired comparisons were performed by MannWhitney test in Graph pad prism 8.2.1 (GraphPad Software, La Jolla, CA). Comparison of survival curves were performed using the log-rank test. Statistical analyses were performed with Microsoft Excel 2016 (Microsoft, Munich, Germany) or GraphPad Prism 8.2.1 (GraphPad Software, La Jolla, CA, USA). Graphical representation was prepared using Adobe Illustrator CS6 (Adobe systems GmbH, Munich, Germany). The size of the animal cohorts was based on our previous study.5,6,10 All in vitro experiments were performed at least three times and all attempts of replication were successful.

Results Patient samples Diagnostic bone marrow or peripheral blood collected from AML patients at Hannover Medical School were analyzed for mutations in IDH1 and IDH2 by Sanger sequencing. Details of the type of IDH mutation are described in the figure legends. Mononuclear cells were isolated by ficoll density centrifugation, washed with PBS, and red blood cells were lysed using a red blood cell (RBC) lysis buffer (BD Pharm Lyse, BD Biosciences, Heidelberg, Germany). The bone marrow samples for the development of the PDX models were collected prior to the start of AML treatment. Written informed consent was obtained according to the Declaration of Helsinki, and the study was approved by the Institutional Review Board of Hannover Medical School, Hannover, Germany.

Transplantation and treatment Six to eight-week old female NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mice were purchased from Hannover Medical School, Germany and kept in pathogen free conditions at the central animal laboratory of Hannover Medical School. Experimental procedures were approved by the governmental authorities of Lower Saxony, Germany, and supervised by local animal welfare officials. The IDH1 mutant AML PDX model was developed as described.10 One million patient-derived AML cells (hCD45+) were collected from the bone marrow and spleen of leukemic mice and were injected intravenously in the tail vein of sublethally (3 Gy) irradiated NSG mice. Neither randomization, nor blinding was used since all animal experiments were performed with a homogeneous strain, comparable age, and similar variance of the mice. Treatment was started 28 days after transplantation. The control groups were treated with either vehicle, BAY1436032 150 mg/kg once daily orally continuously, or azacitidine 1 mg/kg once daily subcutaneoulsy on days 1-5, repeated once after 28 days. The test 566

mIDH1 inhibitor BAY1436032 and azacitidine synergize to inhibit human IDH1 mutant acute myeloid leukemia cells ex vivo In order to test the effect of the combination of BAY1436032 and azacitidine on survival and proliferation of primary human AML cells, IDH1 wild-type (n=6) and IDH1 mutant (n=6) cells from AML patients were seeded in semi-solid medium supplemented without or with BAY1436032 in combination with varying concentrations of azacitidine. BAY1436032 reduced colony formation specifically in human IDH1 mutant AML cells with an IC of 100 nM, while IDH1 wild-type cells were not affected (Online Supplementary Figure S1A). Azacitidine alone induced a 20% reduction in colony formation in both IDH1 wild-type and IDH1 wild-type AML patient cells at 100 nM (Online Supplementary Figure S1B), while the combination of BAY1436032 with azacitidine reduced colony formation in a dose-dependent manner in IDH1 mutant cells with reduction of colonies by 80% in IDH1 mutant AML but only by 20% in IDH1 wild-type AML cells at 100 nM azacitidine, suggesting improved efficacy of an IDH1 inhibitor in combination with azacitidine (Figure 1A, normalized to cells treated with BAY1436032 at 100 nM). In order to test the effect on cell cycle progression and apoptosis, IDH1 mutant and IDH1 wild-type AML cells were cultured in suspension medium and treated with vehicle, 100 nM BAY1436032 or 100 nM azacitidine as single agents or in combination. The proportion of cells in S phase of the cell cycle was strongly decreased by the combination treatment compared with either monotherapy or vehicle in IDH1 mutant AML cells, however, remained largely unchanged in IDH1 wild-type cells (Figure 1B-C). No difference was 50

haematologica | 2021; 106(2)

Synergy between IDH1 inhibitor and hypomethylating agent

observed in the proportion of G0/G1 cells between the combination and monotherapy groups, while the subG1 population was increased in the combination treatment group indicating increased cell death. The percentage of Annexin V+ cells did not differ between any of the treatment groups in IDH1 wild-type or IDH1 mutant AML cells (Online Supplementary Figure S2). In order to further investigate the synergy between azacitidine and BAY1436032, we treated IDH1 mutant AML cells from six patients with serial dilutions of azacitidine and BAY1436032, either alone or in combination and performed colony forming unit assays. The combination of azacitidine with BAY14360932 significantly decreased the colony formation of cells treated with both agents (Figure 1D). In order to determine whether the combined effects of azacitidine and BAY1436032 are additive or synergistic, we performed an isobologram analysis from an average of six patients. The combination of azacitidine with BAY1436032 proved highly synergistic with a combination index (CI) of <0.68 at an effective dose (ED) of 95, a CI of <0.65 at ED 75, and a CI of <0.63 at ED 50, respectively (Figure 1E; a CI of 1 reflects an additive effect, a CI <1 a synergistic effect and a CI >1 indicates antagonism). This data suggest that BAY1436032 in combination with azacitidine more effectively inhibits proliferation of primary IDH1 mutant AML cells ex vivo than single agent treatment.

BAY1436032 synergizes with azacitidine to exert potent anti-leukemic activity in the patient-derived IDH1 mutant acute myeloid leukemia xenograft models in vivo We have previously developed and characterized an IDH1 mutant PDX mouse model (AML-PDX1) using primary AML cells from a patient harboring IDH1 R132C, FLT3-TKD (p.D835del), an atypical NPM1 (p.S254LfsTer4), and a NRAS (p.Q61R) mutation.10 After 28 days of transplantation, having confirmed engraftment between 2-5% in peripheral blood, IDH1 mutant PDX mice were treated with either vehicle, BAY1436032, two cycles of azacitidine as monotherapy or in sequential or simultaneous combination of both drugs for a total duration of 84 days (Figure 2A). We evaluated the simultaneous and sequential combination treatment to derive mechanistic insight into the combination treatment and to instruct the clinical use of this combination, as mutation analysis usually takes some time and leaves a time window in which treatment with azacitidine may be started before the mutation status is known, followed by sequential treatment with a mIDH1 inhibitor. While the engraftment of human leukemic cells increased in vehicle and azacitidine treated mice at week 8 after the start of treatment, the percentage of leukemic cells decreased in BAY1436032 treated mice as well as in the groups receiving the sequential and simultaneous combination treatment (Figure 2B). However, after the stop of treatment at week 12 the percentage of leukemic cells increased after week 16 in the treatment groups receiving BAY1436032 or the sequential combination of BAY1436032 and azacitidine (Figure 2B). Interestingly, the percentage of leukemic cells in mice treated with the simultaneous combination of BAY1436032 and azacitidine showed a delayed increase of blasts and slower kinetics (Figure 2B) and 2 of 6 mice from this cohort remained negative in peripheral blood until the end of the study at 36 weeks (Figure 2C). The mean IDH1 mutant allele fraction in peripheral blood haematologica | 2021; 106(2)

at 8 weeks after the stop of the treatment was 48.8% in vehicle, 40.56% in the azacitidine, 46.43% in the BAY1436032, and 43.2% in the sequential treatment cohorts. However, median IDH1 mutant allele burden was 10.89% in the simultaneous treatment cohort, and below the detection limit in 3 of 5 analysed mice (Online Supplementary Figure S3A). While the white blood cell counts constantly increased and platelet counts, as well as hemoglobin, decreased in mice receiving vehicle, azacitidine or BAY1436032 monotherapy or the sequential combination of BAY1436032 and azacitidine, all mice treated simultaneously had normal blood counts until week 36 (Figure 2D-E, Online Supplementary Figure S3B). In order to compare the effects of the combination therapy with single agents on the induction of myelomonocytic differentiation the expression of CD14 and CD15 in human CD45+ cells was monitored. At day 30 after initiation of treatment there was no significant difference in the expression of the myelomonocytic maturation markers CD14 and CD15 in the vehicle (CD14 0.05±0.05; CD15 0.86±0.3) and azacitidine treated groups (CD14 0; CD15 1.39±0.36). In contrast, mice treated with BAY1436032 had a significantly higher proportion of cells expressing CD14 (4.21±0.62) and CD15 (7.67±0.68). Mice treated in parallel with the combination of BAY1436032 and azacitidine had additive effects on the expression of these markers (CD14 7.44±1.27; CD15 11±0.79) (Online Supplementary Figure S4). While azacitidine treated mice survived longer with a median survival of 188 days compared to vehicle-treated mice with a median survival of 139 days, BAY1436032 treated mice had significantly longer latency with a median survival of 220 days. However, mice treated sequentially with the combination of BAY1436032 and azacitidine survived longer than with BAY1436032 monotherapy with a median survival of 299 days. Strikingly, 5 of 6 mice treated simultaneously with BAY1436032 and azacitidine survived until the end of the study at 300 days and the median survival was not reached (Figure 2F). In an independent, second PDX model (AML-PDX2), which harbored IDH1 p.R132H, DNMT3A p.R882H, PTPN11 p.A72T, and NPM1 p.T288CfsTer12 mutations (Online Supplementary Figure S5A), the percentage of human CD45+ cells increased in vehicle, azacitidine, BAY1436032 and sequential combination treatment groups albeit with different kinetics (Online Supplementary Figure S5B). However, even after the stop of treatment at week 12, the percentage of leukemic cells in mice treated with the simultaneous combination of BAY1436032 and azacitidine remained low and 2 of 6 mice from this cohort had less than 10% human CD45+ cells in the peripheral blood until the end of the study at 28 weeks (Online Supplementary Figure S5C). In contrast to the other groups, all mice treated simultaneously had normal blood counts until week 16 after the start of treatment (Online Supplementary Figure S5D-F). At day 90 after initiation of treatment mice treated simultaneously with the combination of BAY1436032 and azacitidine had significantly higher expression of CD15 than mice treated with BAY1436032 alone or with the sequential combination of BAY1436032 and azacitidine (Online Supplementary Figure S5G). The mice treated simultaneously with BAY1436032 and azacitidine survived significantly longer with a median survival of 250 days than sequentially treated mice with a median median survival of 186 days (Online Supplementary Figure S5G). In summary, the simultaneous combination of 567

A. Chaturvedi et al. A

Figure 1. Mutant IDH1 inhibitor BAY1436032 and azacitidine synergize to inhibit human IDH1 mutant acute myeloid leukemia cells ex vivo. (A) Inhibition of colony formation by combining BAY1436032 (100 nm, corresponding to the IC in previous experiments) with varying concentrations of azacitidine in colony forming assays using primary human acute myeloid leukemia (AML) cells with wild-type or mutant IDH1. The graph represents the proportion of colonies relative to cells treated with BAY1436032 at 100 nM (mean Âą standard error of the mean). From the six patients with IDH1 mutant AML, four harbored an IDH1 R132H mutation and one each an IDH1 R132C and IDH1 R132G mutation. (B) Proportion of viable cells in S phase of the cell cycle after treatment with BAY1436032 (100 nM) or azacitidine (100 nM) or the combination of both relative to DMSO-treated cells (mean Âą standard error of the mean). From the 5 patients with IDH1 mutant AML, 3 harbored an IDH1 R132H mutation and 1 each an IDH1 R132C and IDH1 R132G. (C) A representative fluorescence-activated cell sorting plot of IDH1 wild-type and IDH mutant primary AML cells treated ex vivo with either vehicle, BAY1436032, azacitidine or BAY1436032 and azacitidine in combination. (D) Inhibition of colony formation after treatment with serial dilutions of azacitidine and BAY1436032, alone or in combination using primary human IDH1 mutant AML cells. Five patients harbored an IDH1 R132H and one an IDH1 R132C mutation. (E) Isobologram analysis of the combination of azacitidine and BAY1436032 in IDH1 mutant AML patient cells. The individual doses of azacitidine and BAY1436032 to achieve 90% growth inhibition (effective dose [ED] or fraction affected [Fa]=0.9), 75% growth inhibition (ED 75 or Fa=0.75), and 50% growth inhibition (ED 50 or Fa=0.5) were plotted on the x- and y-axes. Combination index (CI) values calculated using CompuSyn software is depicted in the graph. A CI of 1 indicates an additive effect, a CI <1 a synergistic effect and a CI >1 antagonism. Wt: wild-type, mut: mutant. 50

568

haematologica | 2021; 106(2)

Synergy between IDH1 inhibitor and hypomethylating agent

Figure 2. BAY1436032 synergizes with azacitidine to exert potent anti-leukemic activity in a patient-derived IDH1 mutant acute myeloid leukemia xenograft model in vivo. (A) Schematic representation of the treatment regimens; sim: simultaneous treatment with BAY1436032 and azacitidine (Aza); seq: sequential treatment with BAY1436032 and azacitidine. (B) Percentage of hCD45+ leukemic cells in peripheral blood of IDH1 R132C PDX1 mice at different time points after treatment start with vehicle, azacitidine (1 mg/kg, subcutaneoulsy, days 1-5 and days 29-34), BAY1436032 (150 mg/kg orally, continuously), or the sequential or simultaneous combination of BAY1436032 and azacitidine according to the treatment regimen shown in Figure 2A (mean ± standard error of the mean [SEM]). (C) Percentage of hCD45+ leukemic cells in peripheral blood of individual mice transplanted with human IDH1 mutant AML cells and simultaneously treated with BAY1436032 and azacitidine. (D) White blood cell counts after different time points of treatment (mean ± SEM). (E) Platelet counts after different time points of treatment (mean ± SEM). (F) Kaplan–Meier survival curves of IDH1 mutant PDX1 mice treated with vehicle, azacitidine (1 mg/kg, subcutaneoulsy), BAY1436032 (150 mg/kg, orally), or the sequential or simultaneous combination of BAY1436032 and azacitidine according to the treatment regimen shown in Figure 2A. *P<0.5; **P<0.01; ***P<0.001; PDX: patient-derived xenograft; wt: wild-type; mut: mutant.

BAY1436032 and azacitidine is highly effective in inhibiting IDH1 mutated human leukemia in vivo, suggesting a specific dependency on the simultaneous presence of both drugs in vivo.

Combination treatment with BAY1436032 and azacitidine strongly depletes leukemia stem cells in vivo through inhibition of MAP-kinase signaling and activation of myeloid differentiation In order to assess the effect of simultaneous and sequential treatment with BAY1436032 and azacitidine on leukemia stem cell self-renewal we performed a limiting dilution transplantation experiment with IDH1 mutant AML cells. NSG mice transplanted with primary human IDH1R132C mutant AML cells were treated when leukemias were fully established (70-80% human AML cells in the peripheral blood) with either vehicle, azacitidine, BAY1436032, or the sequential or simultaneous combination of BAY1436032 and azacitidine for 4 weeks (Online Supplementary Figure S6A). Subsequently, we harvested bone marrow cells from these treatment groups and transplanted defined numbers of human AML cells in recipient mice at limiting dilution. Cell numbers of 2,000,000, 200,000, 20,000, 2,000, 200, and 20 human CD45+ AML cells were injected into three NSG recipient mice per cell dose for each treatment group. After 8 weeks the mice were bled and scored positive if more than 0.1% human CD45+ haematologica | 2021; 106(2)

cells were present in the peripheral blood (Figure 3A). The LSC frequency was 1 in 73 in vehicle-treated mice, 4.1-fold lower in azacitidine treated mice (1 in 304), and 117-fold lower in BAY1436032 treated mice (1 in 8,580). The sequential combination treatment of BAY1436032 and azacitidine decreased the stem cell frequency by 470-fold (1 in 34,300) compared to control treated mice and by 4-fold compared to BAY1436032 monotherapy. However, in mice treated simultaneously with BAY1436032 and azacitidine, the leukemia stem cell frequency was 33,150-fold lower compared to control treated mice, 282-fold lower compared to BAY1436032 treated mice, and 70-fold lower compared to mice treated sequentially with BAY1436032 and azacitidine (1 in 2,420,000, Figure 3A, Online Supplementary Figure 6B). Next, we performed gene expression profiling of hCD45+ cells from NSG mice treated with vehicle, azacitidine, BAY1436032 or simultaneous treatment with BAY1436032 and azacitidine for 4 weeks. In unsupervised hierarchical clustering, azacitidine and vehicle-treated groups clustered together and separated from BAY1436032 and BAY1436032 + azacitidine treated groups. However, many genes that were overexpressed in BAY1436032 treated cells were suppressed in the combination group (Figure 3B). Principal component analysis showed that BAY1436032 + azacitidine treated cells separated well from the control or singleagent treated cells (Figure 3C). Interestingly, 11 of 17 genes from the LSC17 gene signature were downregulated by the 569

A. Chaturvedi et al.

E F

570

haematologica | 2021; 106(2)

Synergy between IDH1 inhibitor and hypomethylating agent Figure 3 on previous page. Combination treatment with BAY1436032 and azacitidine strongly depletes leukemia stem cells in vivo through inhibition of MAPkinase signaling and activation of myeloid differentiation. (A) Limiting dilution transplantation of bone marrow cells from IDH1 mutant patient-derived xenograft (PDX) mice treated with vehicle, azacitidine (1 mg/kg, subcutaneously, days 1-5), BAY1436032 (150 mg/kg, orally [p.o.], daily [q.d.], 4 weeks) or the sequential or simultaneous combination of BAY1436032 and azacitidine with the same doses as in the single agent treated mice. 2,000,000, 200,000, 20,000, 2,000, 200 or 20 human acute myeloid leukemia (AML) cells per mouse were transplanted into three recipient mice per cell dose. Leukemia stem cells (LSC) frequencies are shown (mean ± standard error of the mean [SEM], n=3). Mice with hCD45+ cells in peripheral blood after 8 weeks (>0.1%) were scored positive. The log-fraction plot of the limiting dilution model has been generated using the ELDA software; the median LSC frequencies are shown by solid lines and the 95% confidence intervals (CI) by dotted lines. The fold change in LSC frequency normalized to vehicle treated control mice is shown as a bar graph. The fold decrease in LSC compared to vehicle treated mice is indicated on top of the bars. (B) Unsupervised hierarchical clustering using euclidean distance of cells from bone marrow of IDH1 mutant PDX mice treated with vehicle, azacitidine (1 mg/kg, subcutaneously, days 1-5), BAY1436032 (150 mg/kg, orally [p.o.], daily [q.d.], for 4 weeks) or the simultaneous combination of BAY1436032 and azacitidine. Cells were harvested from bone marrow at 4 weeks after treatment and sorted for hCD45+ cells. Gene expression profiling using RNA was performed on Affymetrix Human HG_U133 Plus 2.0 microarrays (n=3 per group). (C) Principal component analysis of all treatment groups using the top 4,000 differentially expressed genes. (D) Gene set enrichment analysis (MSigDB version 6.0) showing the most enriched transcription factor target gene sets from the indicated treatment comparisons. NES: normalized enrichment score; * gene sets involved in MAP Kinase signaling; ** gene sets involved in RB/E2F signaling (E) Heatmap from gene expression levels of MAP kinase signaling genes, RB/E2F signaling genes and myeloid differentiation genes from the bone marrow of IDH1 mutant PDX1 mice treated with vehicle, azacitidine (1 mg/kg, subcutaneous [s.c.] sq, days 1-5), BAY1436032 (150 mg/kg, p.o., q.d., 4 weeks) or the simultaneous combination of BAY1436032 and azacitidine. Gene expression was determined by quantitative RT-PCR relative to the housekeeping gene ABL and was normalized to gene expression in vehicle-treated cells (mean ± SEM, n=3 independent experiments). (F) Representative western blots of in vitro cultured HT1080, a fibrosarcoma cell line with an endogenous heterozygous IDH1 R132C mutation treated with vehicle, azacitidine, BAY1436032 or the simultaneous combination of BAY1436032 and azacitidine using antibodies against the indicated signaling proteins. (G) Inhibition of colony formation by the MEK1/2 inhibitor trametinib in colony-forming cell assays using primary human AML cells with wild-type IDH1 or mutant IDH1 (mean ± SEM). (H) Inhibition of colony formation by the CDK4/6 inhibitor abemaciclib in colony-forming cell assays using primary human AML cells IDH1 wild-type or IDH1 mutant (mean ± SEM). From the six patients with IDH1 mutant AML, three harbored a IDH1 R132H mutation and one each an IDH1 R132C, IDH1 R132L and IDH1 R132G mutation. *P<0.5; ***P<0.001, wt: wild-type, mut: mutant: AZA: azacitidine.

combination treatment, supporting the inhibitory effect of the combination therapy on LSC (Online Supplementary Figure S7).14 Gene set enrichment analysis revealed that the suppressed transcription factors in the combination treatment group were linked to MAP kinase and retinoblastoma/E2F (Rb/E2F) signaling (Figure 3D, Online Supplementary Table S1-3). We also performed epigenome wide asscociation studies by DNA methylation arrays and compared them with the transcriptome obtained by RNA sequencing (RNASeq) in IDH1 R132C cells isolated from AML-PDX1 mice and treated in vitro with either DMSO, 500 nM azacitidine (day 1-4), 50 nM of BAY1436032 (day 1-5), and sequential or simultaneous treatments with BAY1436032 and azacitidine. Principal component analysis showed that sequentially treated cells separated well from simultaneously treated cells (Online Supplementary Figure S8A-B). The simultaneous treatment induced strong synergistic gene expression (S-value >0), while the gene expression was mostly antagonistic in the sequential treatment group (S-value <0; Online Supplementary Figure S8C). The simultaneous treatment overall led to increased methylation compared to sequential treatment (Online Supplementary Figure S8D). Gene set enrichment analysis revealed that the suppressed transcription factors in the simultaneous treatment group were linked to MAP kinase signaling (RAS/RAF), while transcription factors enriched in single treatments were linked to Rb/E2F signaling (Online Supplementary Figure 8E). In concordance with gene expression, an epigenome-wide association showed increased methylation of transcription factors involved in MAP kinase (RAS/RAF) signaling upon simultaneous treatment, while it showed decreased methylation of gene sets involved in RB/E2F signaling upon single agent treatment (Online Supplementary Figure 8F). Gene expression analysis by quantitative RT-PCR validated an additive suppression of MAP kinase (ELK1, ETS1 and CCND1) and E2F signaling (E2F1, CCNA2 and CCNE1) by the combination treatment, which are involved in cell survival and proliferation. Myeloid differentiation genes (PU.1, CEBPA and GABPA) were upregulated by the combination of BAY1436032 and azacitidine, which might be due to the higher proportion of differentiated human CD45+ cells and may not be a direct effect of the drugs. (Figure 3E, Online Supplementary Figure haematologica | 2021; 106(2)

S9A-C). We further validated our findings at protein level in HT1080, a fibrosarcoma cell line with an endogenous heterozygous IDH1 R132C mutation. The phosphorylation of ERK1/2 and expression of its downstream target proteins ELK1 and CYCLIN D1 was decreased in cells simultaneously treated with BAY1436032 and azacitidine. Importantly, the phosphorylation of retinoblastoma (pRB), which is necessary for E2F mediated transcription of cell cycle transition genes, was strongly decreased in cells treated with the combination of BAY1436032 and azacitidine, while E2F1 protein levels were not affected (Figure 3F). We next evaluated if targeting the MEK pathway or cyclin dependent kinases alter the proliferation of IDH1mut AML cells using using the MEK1/2 inhibitor trametinib and the CDK4/6 inhibitor abemaciclib. Trametinib 4-fold more potently inhibited colony-forming potential in IDH1 mutant compared to IDH1 wild-type patients (Figure 3G). Similarly, the CDK4/6 inhibitor abemaciclib inhibited the colony-forming potential 3-fold more potently in IDH1 mutant compared to IDH1 wild-type patients (Figure 3F). This data suggests synergy between BAY1436032 and azacitidine in induction of differentiation through transcriptional activation of myeloid differentiation genes and inhibition of proliferation and self-renewal through inhibition of MAP-kinase and RB/E2F signaling.

Discussion Here we provide preclinical evidence that the simultaneous combination of a mIDH1 inhibitor with a hypomethylating agent significantly inhibits LSC in a synergistic manner. While the phase I dose-escalation study of mIDH1 inhibitor ivosidenib has shown encouraging results, 79% of IDH1 mutant AML patients who had a complete remission or complete remission with partial hematologic recovery had detectable allele burden of mutant IDH1, suggesting that the IDH1 mutant clone was not eradicated.7 It was shown previously that BAY1436032 has antileukemic activity, induces differentiation and depletes LSC 100-fold during a 4-week treatment.10 However, BAY1436032 neither has anti-leukemic activity nor induces myeloid differentiation in an IDH1 wild-type AML PDX model in vivo.10 571

A. Chaturvedi et al.

Figure 4. Model of the combined activity of BAY1436032 and azacitidine on leukemia stem cells. BAY1436032 and azacitidine (AZA) as single agents induce the expression of genes involved in myeloid differentiation (PU.1, CEBPA, and GABPA) and show additive effects in combination. MAP kinase signaling is synergistically inhibited by the combination treatment mediated by inhibition of ERK1/2 phosphorylation and suppression of its downstream targets (ELK1, ETS, CCND1). Suppression of the RB phosphorylation by the combined treatment of BAY1436032 and azacitidine suggests synergistic inhibition of the cyclinD1/CDK4 complex. Unphosphorylated retinoblastoma binds to E2F transcription factors and prevents the G1 to S transition of the cell cycle, thereby inhibiting cell proliferation.

Azacitidine also induces differentiation and reduces selfrenewal15,16 letting us hypothesize that the combination of a mIDH1 inhibitor and azacitidine may act synergistically. Our data suggest that the combination of BAY1436032 and azacitidine has additive effects on differentiation, while it synergistically reduces LSC by 33,150-fold. It was shown previously that mutant IDH1 activates MAPK/ERK signaling in leukemia and glioma cells.5,17 Therefore, we expected that a mIDH1 inhibitor would be able to reduce MAPK/ERK signaling. Data from pancreatic cancer cells suggested that treatment with azacitidine also inhibits MAPK/ERK signaling by promoter hypomethylation and induction of expression of DUSP6, which dephosphorylates members of the MAPK superfamily.18,19 However, the effect of azacitidine on MAPK signaling in leukemia patients is not known. We found that inhibition of the MAP kinase pathway is not a direct consequence of DNA hypomethylation of MAP kinase genes, but rather a transcriptional or posttranscriptional effect on genes that regulate the MAP kinase pathway. We evaluated the consequences of MAPK/ERK inhibition and found that reduced phosphorylation of ERK1/2 leads to suppression of cyclin D1, which is then unavailable to bind CDK4. CDK4 in turn cannot phosphorylate Rb and consequently phosphorylation of Rb is strongly inhibited by the combined treatment of BAY1436032 and azacitidine, providing a mechanistic basis for reduced proliferation and self-renewal of AML cells (Figure 4). 572

Phosphorylation of Serine 795 of Rb is required for the dissociation of E2F1 from Rb, so that it can bind to DNA and induce transcription of genes involved in the G1 to S transition.20 Similarly, the phosphorylation of Serine 807 and 811 is required to prime phosphorylation at other sites on Rb.21 E2F1 then remains bound to Rb and is unavailable for cell cycle progression, leading to the observed strong antileukemic effects. NRAS mutations activate MAPK/ERK signaling and have been associated with resistance to mIDH2 inhibitor treatment.22,23 Combination treatment with a mIDH1/2 inhibitor with azacitidine and inhibition of MAPK/ERK signaling may indeed target a specific dependency in IDH mutated leukemias and may reduce the risk of relapse. Although IDH1/2 mutations increase DNA methylation through the oncometabolite R-2HG,24 IDH1/2 mutated leukemias were not associated with better response to azacitidine monotherapy in myelodysplastic syndrome and AML patients.25,26 We conclude that the synergistic effect of BAY1436032 and azacitidine derives from specific inhibition of MAPK/ERK and RB/E2F signaling. This is supported by the striking difference between sequential and simultaneous treatment with BAY1436032 and azacitidine. Simultaneous treatment during 5 days was 70-fold more efficient in eliminating LSC than sequential treatment, suggesting improved efficacy of the combination when both drugs are present in the leukemic cell at the same time. This strongly argues for the concurrent application of both haematologica | 2021; 106(2)

Synergy between IDH1 inhibitor and hypomethylating agent

drugs in ongoing and future clinical trials. A phase I clinical study combining the IDH1 inhibitor ivosidenib with azacitidine is ongoing and initial data showed promising results.27 It may be worth investigating the effect of prolonged administration of azacitidine at a lower dose or with the oral formulation CC-486 concurrently with a mIDH1/2-inhibitor to further enhance its efficacy in IDH mutated AML patients. In conclusion, the combination of BAY1436032 and azacitidine shows a strong reduction of LSC by suppressing cell proliferation, self renewal and by inducing differentiation. Disclosures AC and MH received research support to their University from Bayer AG; SK and MJ are employees of Bayer AG. MW, OP, and AH are employees and equity owners of Bayer AG. All other authors have no conflict of interest to disclose. Contributions AC and MH conceived and designed the study; AC, CG, NB, RGo, KG, RS, BO, JW, and RGe collected the data; AC, CG,

References 1. Papaemmanuil E, Gerstung M, Bullinger L, et al. Genomic Classification and prognosis in acute Myeloid Leukemia. N Engl J Med. 2016;374(23):2209-2221. 2. Wagner K, Damm F, Gohring G, et al. Impact of IDH1 R132 mutations and an IDH1 single nucleotide polymorphism in cytogenetically normal acute myeloid leukemia: SNP rs11554137 is an adverse prognostic factor. J Clin Oncol. 2010; 28(14):2356-2364. 3. Dang L, White DW, Gross S, et al. Cancerassociated IDH1 mutations produce 2hydroxyglutarate. Nature. 2009; 462(7274):739-744. 4. Lu C, Ward PS, Kapoor GS, et al. IDH mutation impairs histone demethylation and results in a block to cell differentiation. Nature. 2012;483(7390):474-478. 5. Chaturvedi A, Araujo Cruz MM, Jyotsana N, et al. Mutant IDH1 promotes leukemogenesis in vivo and can be specifically targeted in human AML. Blood. 2013;122(16):28772887. 6. Chaturvedi A, Araujo Cruz MM, Jyotsana N, et al. Enantiomer-specific and paracrine leukemogenicity of mutant IDH metabolite 2-hydroxyglutarate. Leukemia. 2016;30(8): 1708-1715. 7. DiNardo CD, Stein EM, de Botton S, et al. Durable remissions with ivosidenib in IDH1mutated relapsed or refractory AML. N Engl J Med. 2018;378(25):2386-2398. 8. Hollenbach PW, Nguyen AN, Brady H, et al. A comparison of azacitidine and decitabine activities in acute myeloid leukemia cell lines. PLoS One. 2010;5(2):e9001. 9. Dombret H, Seymour JF, Butrym A, et al. International phase 3 study of azacitidine vs conventional care regimens in older patients with newly diagnosed AML with >30%

haematologica | 2021; 106(2)

RGa., and MH analyzed and assembled the data; SK, OP, MW, MJ, and AH provided critical reagents; FT, AG and MH collected patient samples and provided the patient data, AC and MH wrote the manuscript. All authors reviewed the data and edited and approved the final version of the manuscript. Acknowledgments We acknowledge the assistance of the Cell Sorting Core Facility of Hannover Medical School supported in part by the Braukmann-Wittenberg-Herz-Stiftung and the Deutsche Forschungsgemeinschaft. We would like to thank all participating patients and contributing doctors, the staff of the Central Animal Facility of Hannover Medical School, and Silke Glowotz, Martin Wichmann, Annett Reinsch and Nadine Kattre for their support. Funding This work was supported by funding from Bayer AG, an ERC grant under the European Unionâ&#x20AC;&#x2122;s Horizon 2020 research and innovation programme (No. 638035), by grant 70112697 from Deutsche Krebshilfe; the German Federal Ministry of Education and Research grant 01EO0802 (IFB-Tx); and DFG grants HE 5240/5-1, HE 5240/6-1 and HE5240/6-2.

blasts. Blood. 2015;126(3):291-299. 10. Chaturvedi A, Herbst L, Pusch S, et al. Panmutant-IDH1 inhibitor BAY1436032 is highly effective against human IDH1 mutant acute myeloid leukemia in vivo. Leukemia. 2017;31(10):2020-2028. 11. Chou TC. Drug combination studies and their synergy quantification using the ChouTalalay method. Cancer Res. 2010; 70(2):440446. 12. Chou TC. Theoretical basis, experimental design, and computerized simulation of synergism and antagonism in drug combination studies. Pharmacol Rev. 2006;58(3):621-681. 13. Heuser M, Argiropoulos B, Kuchenbauer F, et al. MN1 overexpression induces acute myeloid leukemia in mice and predicts ATRA resistance in patients with AML. Blood. 2007;110(5):1639-1647. 14. Ng SW, Mitchell A, Kennedy JA, et al. A 17gene stemness score for rapid determination of risk in acute leukaemia. Nature. 2016;540(7633):433-437. 15. Jones PA. Altering gene expression with 5azacytidine. Cell. 1985;40(3):485-486. 16. Jones PA. Effects of 5-azacytidine and its 2'deoxyderivative on cell differentiation and DNA methylation. Pharmacol Ther. 1985; 28(1):17-27. 17. Shibata T, Kokubu A, Miyamoto M, Sasajima Y, Yamazaki N. Mutant IDH1 confers an in vivo growth in a melanoma cell line with BRAF mutation. Am J Pathol. 2011; 178(3):1395-1402. 18. Xu S, Furukawa T, Kanai N, Sunamura M, Horii A. Abrogation of DUSP6 by hypermethylation in human pancreatic cancer. J Hum Genet. 2005;50(4):159-167. 19. Furukawa T, Tanji E, Xu S, Horii A. Feedback regulation of DUSP6 transcription responding to MAPK1 via ETS2 in human cells. Biochem Biophys Res Commun. 2008; 377(1):317-320. 20. Rubin SM, Gall AL, Zheng N, Pavletich NP.

Structure of the Rb C-terminal domain bound to E2F1-DP1: a mechanism for phosphorylation-induced E2F release. Cell. 2005;123(6): 1093-1106. 21. Driscoll B, T'Ang A, Hu YH, et al. Discovery of a regulatory motif that controls the exposure of specific upstream cyclin-dependent kinase sites that determine both conformation and growth suppressing activity of pRb. J Biol Chem. 1999;274(14):9463-9471. 22. Stein EM, DiNardo CD, Pollyea DA, et al. Enasidenib in mutant IDH2 relapsed or refractory acute myeloid leukemia. Blood. 2017;130(6):722-731. 23. Amatangelo MD, Quek L, Shih A, et al. Enasidenib induces acute myeloid leukemia cell differentiation to promote clinical response. Blood. 2017;130(6):732-741. 24. Figueroa ME, Abdel-Wahab O, Lu C, et al. Leukemic IDH1 and IDH2 mutations result in a hypermethylation phenotype, disrupt TET2 function, and impair hematopoietic differentiation. Cancer Cell. 2010;18(6):553567. 25. Sekeres MA, Othus M, List AF, et al. Randomized phase II study of azacitidine alone or in combination with lenalidomide or with vorinostat in higher-risk myelodysplastic syndromes and chronic myelomonocytic leukemia: North American Intergroup Study SWOG S1117. J Clin Oncol. 2017; 35(24):2745-2753. 26. Dohner H, Dolnik A, Tang L, et al. Cytogenetics and gene mutations influence survival in older patients with acute myeloid leukemia treated with azacitidine or conventional care. Leukemia. 2018;32(12):25462557. 27. Dinardo CD, Stein AS, Stein EM, et al. Mutant IDH (mIDH) inhibitors, ivosidenib or enasidenib, with azacitidine (AZA) in patients with acute myeloid leukemia (AML). J Clin Oncol. 2018;36(15):7042-7042.

573

LETTERS TOâ&#x20AC;&#x2C6;THE EDITOR Novel L-nucleoside analog, 5-fluorotroxacitabine, displays potent efficacy against acute myeloid leukemia Cytarabine (Ara-C) is a nucleotide analog and a cornerstone of standard chemotherapy in acute myeloid leukemia (AML).1 Ara-C is phosphorylated to its active metabolite Ara-CTP where it becomes incorporated into replicating DNA strands. Ara-CTP is deaminated to an inactive uridine metabolite by cytidine deaminase (CDA) and CDA overexpression is a clinically relevant mechanism of resistance to Ara-C.2-4 Here, we report the preclinical efficacy of a novel L-nucleoside analog 5-fluorotroxacitabine (5FTRX) in AML (Figure 1A). In a panel of 31 leukemia and lymphoma cell lines, 5FTRX reduced growth and viability with a median IC50 of 200 nM (range 12nM [EOL-1] to 6 mM [TF-1]) (Figure 1B and Online Supplementary Figure S1). 5FTRX also reduced the clonogenic growth of primary AML cells (Figure 1C; see Online Supplementary Tables S1 and S2 for patients' characteristics), thus demonstrating an ability to target leukemic progenitors. In contrast, normal hematopoietic cells were more resistant to 5FTRX (Figure 1D). 5FTRX is phosphorylated via monophosphate (MP) and diphosphate (DP) to the triphosphate (TP) before being incorporated into nascent DNA strands, leading to subsequent chain termination and cell death. AML cell lines MV4-11 and THP-1 or human peripheral blood mononuclear cells (PBMC) were incubated with 5FTRX and analyzed for the intracellular concentrations of 5FTRX and phosphorylated nucleotides (Online Supplementary Table S3). Detectible concentrations of MP, DP and TP were observed in PBMC and AML cells confirming bioactivation to the active TP in cells. To test whether 5FTRX damaged DNA in AML cells, we measured changes in phosphorylated H2AX (pH2AX) after 5FTRX treatment. 5FTRX increased pH2AX at concentrations associated with loss of viability (Online Supplementary Figure S2), indicating a likely chain termination during the proliferation of these cells. CDA overexpression is a mechanism of resistance to Ara-C.2-5 We investigated the effects of increased CDA on the anti-neoplastic action of 5FTRX. Wild-type HEK293 and HEK293 cells over-expressing CDA were treated with increasing concentrations of 5FTRX and Ara-C (Figure 2A and Online Supplementary Figure S3A). As expected, overexpression of CDA rendered cells approximately 7-fold more resistant to Ara-C (Figure 2B). Treatment of these cells with the CDA inhibitor tetrahydrouridine (THU) restored sensitivity to Ara-C (Online Supplementary Figure S3B). Interestingly, overexpression of CDA increased sensitivity to 5FTRX by approximately 10-fold (Figure 2C). Further, co-treatment with 5FTRX and THU returned sensitivity of the cells to 5FTRX towards baseline (Online Supplementary Figure S3C). Increased CDA expression did not change the growth rate of these cells (Online Supplementary Figure S4). Potentially, overexpression of CDA increased degradation of endogenous cytidine nucleotide pools enhancing the incorporation of 5FTRX-monophosphate into DNA. Of note, prior studies have reported that increased CDA expression increases the sensitivity of cells to troxacitabine.6,7 5FTRX overcomes drug resistance induced by CDA overexpression, but CDA expression did not fully explain 574

5FTRX sensitivity in AML cells (Online Supplementary Figure S5A and B), indicating additional mechanisms of resistance and sensitivity. Therefore, we sought to understand potential mechanisms of resistance to 5FTRX. We selected populations of THP-1 cells resistant to 5FTRX (THP1-5FR: 66-fold resistant to 5FTRX) and Ara-C (THP1-ACR; 35-fold resistant to Ara-C) by treating parental THP-1 cells with increasing concentrations of the drugs for up to 4 months (Figure 2D). THP1-5FR and THP1-ACR were cross-resistant to Ara-C and 5FTRX, respectively, suggesting that similar mechanisms were responsible for conferring resistance. Downregulation of dCK is known to confer resistance to Ara-C,6 so we examined dCK expression in the resistant cell lines. Indeed, dCK levels were reduced in THP1-5FR and THP1-ACR resistant cell populations (Figure 2E). To test whether the reduced dCK was functionally important for resistance to 5FTRX, we analyzed the metabolites of 5FTRX in THP1-5FR cells. THP1-5FR cells treated with 5FTRX showed decreased phosphorylated species, consistent with a reduced ability of dCK to generate the monophosphate species (Figure 2F). Moreover, the reported dCK inhibitor 15C7 rendered THP-1 cells resistant to 5FTRX and Ara-C with IC50 values >50 mM (Figure 2G). To assess the future potential for 5FTRX to be included in standard treatment regimens for AML treatment, the anti-proliferative effects of 5FTRX were examined in combination with Ara-C, azacytidine or doxorubicin in MV4-11 and THP-1 cells. In both cell types, 5FTRX exhibited strong synergy in combination with azacytidine or doxorubicin. The combination of 5FTRX and AraC was additive in MV4-11 cells, whereas in THP-1 cells, a moderately synergistic interaction was observed (Online Supplementary Table S5 and Online Supplementary Figure S6A-F). Finally, we examined the preclinical efficacy and toxicity of 5FTRX in mouse models of leukemia. In mice xenografted with MV4-11 cells, 5FTRX produced dramatic and sustained tumor regressions and showed a marked superiority over Ara-C at its maximum tolerated dose (Figure 3A) without evidence of toxicity or body weight loss (Online Supplementary Figure S7). Furthemore, we assessed the activity of 5FTRX at a reduced frequency of treatment (qD x5 instead of BIDx5) in vivo using MV411 cells. Sustained tumor regressions were observed at all doses (Figure 3C) without loss of body weight (Online Supplementary Figure S8). The OCI-AML2 model displayed even more sensitivity to 5FTRX, with tumor regression observed at both 100 mg/kg and 30 mg/kg doses. Mice remained tumor free for 45 days after termination of treatment (Figure 3B) and did not display any signs of toxicity (Online Supplementary Figures S9-S11). We determined the optimal treatment schedule by delivering a total dose of 150 mg/kg at different time points: a single dose of 150 mg/kg; three daily doses of 50 mg/kg; or five daily doses of 30 mg/kg. All regimens led to tumor regressions, although tumor regrowth varied between doses (Figure 3D). The maximal response was observed at five daily doses of 30 mg/kg. To test whether the tumors remained sensitive to the drug, tumors were allowed to regrow to approximately 1,000 mm3 from the 5x30 mg/kg dose group, and then challenged with the same schedule of 5FTRX treatment. Despite the larger size of these tumors at this re-treatment stage, they still showed pronounced regression (Figure 3D), confirming they remained sensitive to the agent. haematologica | 2021; 106(2)

Letters to the Editor

Figure 1. 5-fluorotroxacitabine (5FTRX) reduces cell growth and viability in hematologic cell lines and primary acute myeloid leukemia (AML) samples. (A) The structure of 5FTRX and its bio-activation by phosphorylation to the monophosphate (MP), diphosphate (DP), and triphosphate (TP). The enzymes known to catalyse these phosphorylation reactions are indicated. DCK: deoxycytidine kinase; CMPK: cytidine monophosphate (CMP) kinase; PGK: phosphoglycerate kinase. (B) Hematologic cell lines were treated with increasing concentrations of 5FTRX for 96 hours (h). After incubation, cell growth and viability were measured by Cell Titer-Glo assay. Black bars indicate the myeloid cell lines; gray bars represent lymphoid cell lines. Data represent IC50, mean±standard deviation (SD). (C) Primary AML cells were treated with increasing concentrations of 5FTRX for 24 h and then plated into clonogenic growth assays. Data represent the number of colonies: mean±SD. (D) Peripheral blood stem cell (PBSC) were treated with increasing concentrations of 5FTRX or control for 24 h and then plated into clonogenic growth assays. Data represent the number of colonies: mean±SD. For all experiments, P-value was calculated using Student’s t-test; P<0.05 was considered significant.

haematologica | 2021; 106(2)

575

Letters to the Editor

Figure 2. Resistance mechanisms to 5-fluorotroxacitabine (5FTRX ). (A) HEK293 cells were transduced with cDNA encoding cytidine deaminase (CDA) or vector control. Levels of CDA were measured by immunoblotting. (B and C) HEK293 cells over-expressing CDA or vector control were treated with increasing concentrations of cytarabine (Ara-C) (B) or 5FTRX (C) for 72 hours (h). After incubation, cell growth and viability were measured by MTS assay. Significance of difference in mean IC50 values for control and CDA+ cell line was calculated using Studentâ&#x20AC;&#x2122;s t-test. (D) THP-1 cells resistant to 5FTRX (THP1-5FR) and Ara-C (THP1-ACR) were selected. Data represent the IC50 in mM for Ara-C and 5FTRX in the parental and resistant cell populations. (E) Levels of deoxycytidine kinase (DCK) were measured by immunoblotting in the parental and resistant THP-1 cells. (F) THP-1 parental and resistant cells (THP1-5FR) were treated with 10 mM 5FTRX for 24 h. After incubation, levels of intracellular metabolites (5FTRX monophosphate [MP], 5FTRX diphosphate [DP], 5FTRX triphosphate [TP] were measured by liquid chromatography-mass spectrometry (LC-MS). (G) IC50 (mM) of gemcitabine, Ara-C and 5FTRX alone or in combination with 50 mM dCK inhibitor (dCKi 15c7) in THP-1 cells.

576

haematologica | 2021; 106(2)

Letters to the Editor

In order to understand the strong anti-tumor activity and mechanism of action of 5FTRX in vivo, pharmacokinetic and pharmacodynamic studies were undertaken in the MV4-11 model. Mice bearing subcutaneous tumor xenografts were treated with control or 5FTRX twice daily for 5 days and were sacrificed 2 hours after the last dose. Two hours prior to analysis the mice received BrdU and pimonidazole (i.p.) to label S-phase cells and hypoxic tumor regions. Proliferating BrdU positive cells were significantly decreased in a dose-dependent manner upon treatment with 5FTRX (Online Supplementary

Figure S12A). A concomitant increase in necrosis as assessed by the reduction in nuclear density was observed (Online Supplementary Figure S12B). Widespread elevation of DNA damage in tumor nuclei throughout the tumors was observed by measuring pH2AX levels at all doses (Online Supplementary Figure S12C), indicating double-strand breaks as a result of incorporation of chain terminating 5FTRX nucleotide. A reduction in %pH2AX-positive nuclei were observed in the 100 mg/kg group compared to the lower doses of 5FTRX, but this finding may reflect the extensive necro-

Figure 3. 5-fluorotroxacitabine (5FTRX) targets primary acute myeloid leukemia (AML) cells. (A) MV4-11 cells were xenografted into NOD/SCID mice. Once tumors were approximately 200 mm3, mice were treated with 5FTRX (100 mg/kg BID x 5 days [d]) or cytarabine (Ara-C) (60 mg/kg qD x 5/7 x 2 weeks). Tumor volume was measured over time. Tumor growth inhibition (TGI%) defined as (C1-T1)/(C0-T0) x 100 where C1 and T1 are the mean tumor volumes of control and 5FTRX or Ara-C treated groups at time of tumor extraction, while C0 and T0 are the mean tumor volumes at the start of treatment. Percent regression calculated using: 100 x DT/DC where DT = T1-T0 and DC = C1-C0. Tumor growth delay (TGD) represents anti-tumor efficacy defined as the time taken for tumors to quadruple in size (RTV4). (B) OCI-AML2 cells were xenografted into SCID mice. Once tumors were approximately 100 mm3, mice were treated with 5FTRX (30 or 100 mg/kg x 5 days). Tumor volume was measured over time. T/C represents the ratio of mean tumor volume for 5FTRX treated group versus the mean tumor volume for the control group. (C) MV4-11 cells were xenografted into NOD/SCID mice. Once tumors were approximately 200 mm3, mice were treated with increasing amounts of 5FTRX. Tumor volume was measured over time. (D) MV4-11 cells were xenografted into NOD/SCID mice. Once tumors were approximately 200 mm3, mice were treated with increasing amounts of 5FTRX on days 1-4. Mice treated with 30 mg/kg received a second treatment with 5FTRX 10 mg/kg on days 22 to 26. Tumor volume was measured over time.

haematologica | 2021; 106(2)

577

Letters to the Editor

Figure 4. 5-fluorotroxacitabine (5FTRX) targets acute myeloid leukemia (AML) stem cells in vivo. (A) Primary AML cells were injected intra-femorally into irradiated female NOD/SCID mice preconditioned with CD122. Mice were treated with 100 mg/kg per day of 5FTRX by intraperitoneal (i.p.) injection or control from days 12-14 for 5 days (n=10 per group). Six weeks after injection, cell engraftment was assessed by flow cytometry. Data represent mean±standard deviation (SD). (B) Secondary engraftment was assessed by injecting viable leukemia cells from the bone marrow of 5FTRX treated and control mice and injected into the secondary recipient mice, which remained untreated. Six weeks later, cell engraftment was measured by flow cytometry. Data represent mean±SD. For all experiments, P-value was calculated using Student’s t-test; P<0.05 was considered significant.

sis and profound anti-tumor effect induced by this dose. To assess changes in 5FTRX metabolites in vivo, mice bearing MV4-11 tumors were treated with 5FTRX and levels of 5FTRX and metabolites were measured in plasma and tumor samples over time. Approximately linear increases in plasma and tumor exposures were observed with increasing doses (Online Supplementary Figure S13A and B). 5FTRX MP, DP and TP exposures were observed in the tumor samples, indicating phosphorylation of the nucleoside within the tumor. High levels of TP were maintained within the tumor for a longer duration than plasma, consistent with a long half-life of this metabolite, and retention within the tumor, as expected from a charged metabolite (Online Supplementary Figure S13C). Finally, we assessed whether 5FTRX can target primary AML cells and stem cells in vivo. NOD/SCID mice were injected intra-femorally with primary AML cells (Online Supplementary Table S1). Eight days after injection of the primary cells, mice were treated with 5FTRX 100 mg/kg daily for 5 days. Four to six weeks after the initial injection of cells, mice were sacrificed, and the percentage of human myeloid cells defined as CD45+CD19–CD33+ was quantified by flow cytometry in the mouse marrow. 5FTRX reduced primary AML engraftment >95% without toxicity (P<0.0001, Student's t-test) (Figure 4A and Online Supplementary Table S4). Moreover, we assessed the effects of 5FTRX on the leukemic stem cells (LSC) by evaluating secondary engraftment. Primary AML cells harvested from bone marrow of 5FTRX treated mice were engrafted into secondary untreated mice and human leukemic cell engraftment was assessed after 5 weeks. 5FTRX significantly reduced leukemic engraftment (P<0.001, Student's t-test) in secondary transplants, demonstrating an effect of the drug on AML stem cells (Figure 4B). Thus, in summary, this study demonstrated that the Lnucleoside analog, 5FTRX, has a number of favorable properties for a potential new treatment for AML.1 5FTRX has a potent and broad-ranging anti-tumor activity against a range of AML and other leukemic cell lines, 578

which is maintained against primary patient AML cells.2 5FTRX targets patient-derived AML blasts and stem cells in physiologically relevant intra-femoral engraftment models in vivo.3 5FTRX overcomes CDA overexpression which is a known mechanism of Ara-C resistance.4 5FTRX has a strong anti-tumor activity in vivo that can lead to durable tumor regressions at well-tolerated doses.5 5FTRX is synergistic in combination with doxorubicin and azacytidine which could be good combination agents for clinical evaluation. Thus, 5FTRX warrants further preclinical study as a novel agent for patients with refractory AML and increased CDA expression. Aniket Bankar,1* Thirushi Piyumika Siriwardena,1* Biljana Rizoska,2 Christina Rydergård,2 Helen Kylefjord,2 Vilma Rraklli,2 Anders Eneroth,2 Pedro Pinho,2 Stefan Norin,2 Johan Bylund,2 Sara Moses,2 Richard Bethell,2 Simon Kavanagh,1 Neil Maclean,1 Marcela Gronda,1 Xiaoming Wang,1 Rose Hurren,1 Mark D. Minden,1,3,4 Paul Targett-Adams,2 Aaron D. Schimmer1,4 and Mark Albertella2 1 Princess Margaret Cancer Center, University Health Network, Ontario, Canada; 2Medivir AB, Huddinge, Sweden; 3Department of Medical Biophysics, Faculty of Medicine, University of Toronto, Ontario, Canada and 4Division of Hematology, Faculty of Medicine, University of Toronto, Ontario, Canada *AB and TPS contributed equally as co-first authors. Correspondence: AARON D. SCHIMMER - aaron.schimmer@uhn.ca doi:10.3324/haematol.2019.226795 Disclosures: BR was an employee of Medivir when the work was performed; MA and PTA were employees of Medivir at the time the research was performed and have an equity share in the company; ADS received research funding from Medivir and has received consulting fees from Novartis, Jazz, Takeda and Otsuka Pharmaceuticals; ADS holds shares in AbbVie. Contributions: AB, TPS, NM, MG, RH, XW, BR, SR, PTA and MA, performed research and analyzed data; BR, MA and PTA analyzed data, supervised research and provided critical reagents; ADS haematologica | 2021; 106(2)

Letters to the Editor

analyzed data and supervised research; AB, ADS, PTA and MA wrote the paper; all authors reviewed and edited the paper; 5FTRX was developed by Medivir AB, Sweden, which has interests in nucleotide/nucleoside science to develop pharmaceuticals that meet unmet medical needs. Acknowledgments: we thank Jill Flewelling (Princess Margaret Cancer Center) for administrative assistance. We would like to thank Andrew Minchinton and Alistair Kyle for the immunofluorescence analyses. Funding: this work was supported by Medivir AB, the Canadian Institutes of Health Research, the Princess Margaret Cancer Centre Foundation, and the Ministry of Long-Term Health and Planning in the Province of Ontario. ADS holds the Barbara Baker Chair in Leukemia and Related Diseases.

References 1. Dรถhner H, Estey EH, Amadori S, et al. Diagnosis and management of acute myeloid leukemia in adults: recommendations from an

haematologica | 2021; 106(2)

international expert panel, on behalf of the European LeukemiaNet. Blood. 2010;115(3):453-474. 2. Dumontet C, Fabianowska-Majewska K, Mantincic D, et al. Common resistance mechanisms to deoxynucleoside analogues in variants of the human erythroleukaemic line K562. Br J Haematol. 1999;106(1):78-85. 3. Galmarini CM, Thomas X, Calvo F, et al. Potential mechanisms of resistance to cytarabine in AML patients. Leuk Res. 2002;26(7):621629. 4. Hyo Kim L, Sub Cheong H, Koh Y, et al. Cytidine deaminase polymorphisms and worse treatment response in normal karyotype AML. J Hum Genet. 2015;60(12):749-754. 5. Schroder JK, Kirch C, Seeber S, Schutte J. Structural and functional analysis of the cytidine deaminase gene in patients with acute myeloid leukaemia. Br J Haematol. 1998;103(4):1096-1103. 6. Veuger MJ, Honders MW, Landegent JE, Willemze R, Barge RM. High incidence of alternatively spliced forms of deoxycytidine kinase in patients with resistant acute myeloid leukemia. Blood. 2000;96(4):1517-1524. 7. Murphy JM, Armijo AL, Nomme J, et al. Development of new deoxycytidine kinase inhibitors and noninvasive in vivo evaluation using positron emission tomography. J Med Chem. 2013; 56(17):6696-6708.

579

Letters to the Editor

Wip1 regulates hematopoietic development in the mouse embryo

stem

cell

The first adult-repopulating hematopoietic stem cells (HSC) are found in the aorta-gonad-mesonephros (AGM) region at embryonic day (E)10.5,1,2 as well as in other sites, such as the embryonic head,3,4 yolk sac (YS) and placenta.5-7 Accumulating evidence from ex vivo models suggests that HSC emerge from endothelial cells (EC) of the aorta by forming sprouting clusters,8-10 through the preHSC I (VE-Cadherin+CD41lowCD45–) and pre-HSC II (VECadherin+CD45+) stages of maturation, and are regulated by various transcription factors and signaling pathways. Less is known about the exact mechanism of maturation of pre-HSC in vivo. Wild-type p53-induced phosphatase (Wip1) is defined as a critical regulator involved in HSC function and lymphoid/neutrophil development.11,12 However, the role of Wip1 in regulating HSC function and pre-HSC maturation in the embryo is still unknown. Our data have shown that Wip1 is expressed in EC and pre-HSC in the AGM region and fetal liver (FL)/bone marrow HSC,13 suggesting that Wip1 may be involved in hematopoiesis (Online Supplementary Figure S1A). To test whether Wip1 plays a role in hematopoietic development of the embryo, Wip1 homozygous deficient embryos (Wip1-/-, KO) were used in this study. The total cell number was significantly decreased in E12.5 Wip1-/- FL compared with wild-type (Wip1+/+, WT) (2.8±0.4×106 vs. 4.1±0.5×106) and E13.5 (5.9±0.5×106 vs. 9.1±0.7×106) (Figure 1A, Online Supplementary Figure 1B). The percentages of CD45+ and CD34+c-Kit+ cells were indistinguishable, but the absolute cell numbers of both populations were obviously reduced, respectively (Figure 1B-E, Online Supplementary Figure S1C, D). To test the hematopoietic progenitor cell (HPC) function, colony-forming unit-culture (CFU-C) assays were performed, and revealed smaller size of burst-forming unit-erythroid (BFU-E), CFUgranulocyte macrophage (CFU-GM) and CFU-granulocyte erythrocyte monocyte megakaryocyte (CFU-Mix) colonies in the E12.5 Wip1-/- FL compared with E12.5 WT FL (Online Supplementary Figure S1E). The number of CFU-C per E12.5 Wip1-/- FL was reduced by more than 70% (3.4±0.3×103 E12.5 vs. 14.2±1.4×103), with dramatic reductions in BFU-E, CFU-GM and CFU-Mix numbers. The same trend in reduced CFU-C numbers was observed from E13.5 to E14.5 in Wip1-/- FL. Interestingly, the CFU-C numbers from the same input number of FL cells (2×104 cells) were decreased considerably (Figure 1F, G, Online Supplementary Figure S1F), demonstrating the influence of Wip1 deletion on the potential of HPC in the FL. Subsequently, flow analysis and transplantation were used to investigate Wip1 function in FL HSC development. Firstly, there was a significant reduction in the percentage of HSC of E12.5 Wip1-/- FL compared to WT FL (0.015±0.003% vs. 0.025±0.004%) as revealed by a cocktail of markers (Lin–CD48–Mac1lowSca-1+CD150+, SLAM HSC). Moreover, the absolute number of SLAM HSC was reduced by more than 30% (419±80 vs. 607±116/embryo equivalent), similar to the decrease in the percentage and absolute number of Lin–Mac1lowSca-1+ cells (including more immature HPC) (Figure 1H, J, Online Supplementary Figure S2A, B). Subsequently, assays following in vivo transplantation by injection of 0.05 embryo equivalent/recipient revealed that six out of 19 recipients given E12.5 Wip1-/- FL cells were repopulated, with 12.7±4.6% chimerism, whereas 15 out of 17 recipients of WT cells achieved chimerism of 65.0±7.8% (Figure 1K, 580

Online Supplementary Table S1). Multilineage assays showed the reduction of B-cell lineage output (B220+ %) in the peripheral blood from Wip1-/- FL derived-recipients. (Figure 1M, Online Supplementary Figure S2C-G). Secondary transplantation data showed that HSC selfrenewing ability was decreased in the Wip1-/- group (Online Supplementary Figure S2H). Although all the recipients were repopulated, the average chimerism was decreased in the E14.5 Wip1-/- FL, with reductions in Band T-cell lineage output and an increase of myeloid cells (Figure 1L, N). Taken together, our findings indicate the involvement of Wip1 in HPC development and HSC activity of embryonic FL with a decrease in lymphoid lineage differentiation. E11.5 embryos with Wip1 deficiency are smaller than their WT littermates (Online Supplementary Figure S3A). The total cell number was decreased from E10.5 to E12.5 AGM (Online Supplementary Figure S3B). Flow analysis data showed that the percentages of CD45 were comparable in the Wip1-/- AGM and YS and the corresponding WT tissues; however, the percentage of CD41lowCD45– cells was decreased significantly in the E11.5 AGM region, but not in the YS (Online Supplementary Figure S3C-F), indicating a reduction of HPC. CFU-C assays confirmed that Wip1 ablation leads to a dramatic decrease in HPC function (61±21 vs. 326±49 CFUC/AGM) at E10.5, including reduced numbers of CFUMix and CFU-GM. Similar decreases were found in the E11.5 and E12.5 Wip1-/- AGM (85±13 and 30±8 CFUC/AGM, respectively) compared with WT AGM (249±46 and 208±39 CFU-C/AGM, respectively) (Figure 2A). High proliferative potential colony-forming cells presenting immature HPC were clearly reduced in the Wip1-/- AGM (Online Supplementary Figure S3G). Meanwhile, similar trends in reduction were found in E10.5-E12.5 Wip1-/- YS (Figure 2B). These results suggest a positive regulatory role of Wip1 in HPC function in the AGM region and YS. Thereafter, in vivo transplantation assays were performed to test the influence of Wip1 on HSC functions. Four of 12 (33%) recipients of E11.5 WT AGM cells showed long-term, high-level multilineage repopulation, but none of nine recipients of Wip1-/- AGM cells were repopulated. Unexpectedly, two of seven recipients of E12.5 Wip1-/- AGM cells were repopulated, with a lower repopulation ratio compared with the WT group (6/6 recipients were repopulated). The B-cell lineage output was diminished in Wip1-/- AGM-derived recipients (Figure 2C, D, Online Supplementary Table S1), consistent with Wip1-/- FL and bone marrow HSC. In E11.5 YS, the ability to engraft was comparable (3/9 with chimerism of 9.1±4.1% vs. 2/6 with chimerism of 15.4±10.5%); however, the engraftment ability was decreased significantly in E12.5 Wip1-/- YS with an increase of myeloid output (8/11 vs. 9/9) (Figure 2E, F). These data, together with the HPC results, indicate that Wip1 affects hematopoietic stem and progenitor cell function mainly in the E11.5 AGM region. HSC are reported to emerge autonomously at E10.5 (>35 somite pairs) in the AGM region but are inefficient in direct transplantation because of their low activity.1 Explant culture is an efficient tool for HSC expansion/pre-HSC maturation. Explant culture assays revealed a decrease in HSC activity in the recipients receiving E10.5 Wip1-/- AGM (4/9) compared with WT AGM (11/12), with a greater myeloid output at the cost of T lymphoid output (Figure 2G, H), indicating that Wip1 ablation may affect HSC expansion and/or preHSC maturation. We also investigated whether Wip1 influences prehaematologica | 2021; 106(2)

Letters to the Editor

Figure 1. Wip1 knockout resulted in the reduction of hematopoietic progenitor cell number and hematopoietic stem cell activity in embryonic day 12.5-14.5 fetal liver. (A) The reduction of cell number on embryonic day (E)12.5 Wip1-/- fetal liver (FL) (n=9, **P=0.0028). (B-E) Percentages (B, D) and absolute numbers (C, E) of CD45+ and CD34+c-Kit+ cells on E12.5 (n=6, *P<0.05). (F, G) Colony-forming unit – culture assay showed the number of colonies per FL (F) and per 2x104 cells (G) from E12.5 to E14.5. Colony types are indicated by colored bars. E12.5, E13.5, and E14.5: n=3, 3, and 2, respectively; *P<0.05, **P<0.01, ***P<0.001. (H) Representative flow cytometric analysis of phenotypic hematopoietic stem cells (HSC) (Lin–Mac-1lowSca-1+CD150+CD48–, SLAM HSC). (I, J) Percentage (I) and number (J) of SLAM HSC in the E12.5 wild-type (WT) and Wip1-/- FL (n=6, *P=0.0189 and **P=0.0017). (K, L) Reconstituting potential of E12.5 and E14.5 WT and Wip1-/- FL. Chimerism (%CD45.2+) in peripheral blood of recipients at 16 weeks after transplantation. Each symbol represents one recipient. Green circle=WT, red triangle=Wip1-/-. The lines represent the average chimerism (n=3, ***P<0.001. (M, N) Multilineage output of peripheral blood of representative repopulated recipients. (n=3, *P=0.02 and ***P<0.001). ee: embryo equivalent; WT: wild-type; KO: knockout; CFU-C: colony-forming unit culture; CFU-Mix: colony-forming unit – granulocyte, erythrocyte, monocyte, megakaryocyte; CFU-GM: colony-forming unit – granulocyte-macrophage; BFU-E: burstforming unit – erythroid; GM: Gr-1+/Mac-1+ cells; T: CD3+ cells; B: B220+ cells.

HSC maturation. CD31+CD41lowCD45– (pre-HSC I) and CD31+CD45+ (pre-HSC II) cells were characterized by flow analysis. The percentages of pre-HSC I were comparable in the E11.5 Wip1-/- AGM and WT AGM (0.116±0.030% vs. 0.079±0.009%) as were the absolute number of cells (207±44 vs. 176±22, respectively) (Figure 3A-C). However, Wip1 deletion resulted in reduced percentages of HSC II (0.037±0.004% vs. 0.054±0.006%) and greater than 40% decreases of absolute numbers of pre-HSC II (68±9 vs. 121±15, respectively) (Figure 3A, D, E). The trends were similar when CD201 was included (Online Supplementary Figure S4A-D). To test the potential haematologica | 2021; 106(2)

of pre-HSC, pre-HSC I and pre-HSC II from E11.5 AGM were co-cultured with OP9-DL1 cells for 6 days prior to transplantation assays. The ratio of repopulated recipients was decreased slightly in the Wip1-/- pre-HSC I cultures (2/6; 33%) compared with WT cultures (3/4; 75%), with lower B lymphoid and higher myeloid lineage output. Meanwhile, 5/16 recipients were reconstituted by injecting Wip1-/- pre-HSC II cultures with lower chimerism (14.3±5.6%) and higher myeloid lineage in comparison to WT recipients (6/12, with 27.6±8.7% chimerism) (Figure 3F-G, J-K). To further confirm the roles of Wip1 in pre-HSC maturation, a Wip1 inhibitor 581

Letters to the Editor

(CCT007093, CCT) was added. None of the recipients that received pre-HSC I cultures with inhibitor was repopulated, whereas six of seven recipients given dimethylsulfoxide (DMSO) instead of the inhibitor achieved chimerism (33.3±9.0%). In the pre-HSC II cultures, the presence of the CCT inhibitor reduced the ability of engraftment of pre-HSC II-derived HSC cultures compared to the cultures with DMSO (5/10 [50%] vs. 7/9 [77.8%]; chimerism 13.5±5.2% vs. 46.2±8.8%, respec-

tively) (Figure 3H, I). Additionally, the lymphoid lineage output of pre-HSC II cultures was changed in the presence of the Wip1 inhibitor (Figure 3L, M). These results, together with the direct transplantation findings, indicate that Wip1 educates the maturation of pre-HSC in the AGM region. Hematopoietic stem and progenitor cells emerge from the EC of the aorta, by forming ‘hematopoietic clusters,8,10,14 including pre-HSC, HSC and HPC. Here, Wip1

582

Figure 2. Hematopoietic stem and progenitor cell function was reduced in the embryonic day 10.5-12.5 Wip1-/- aorta-gonadmesonephros region and yolk sac. (A, B) Colony-forming unit culture assay showed the number of colonies in the aorta-gonadmesonephros (AGM) (A) and yolk sac (YS) (B) on embryonic day (E)10.5, E11.5 and E12.5 (*P<0.05, **P<0.01, ***P<0.001 for wild-type [WT] -/vs. Wip1 ). (C) Direct transplantation assay showed the repopulating ability of E11.5 and E12.5 WT and Wip1-/- AGM. E11.5 (n=4) and E12.5 (n=3): *P=0.03 and ***P<0.001. (D) The lineage output of donor-derived granulocytemacrophage (GM), T and B cells in E12.5 WT and Wip1-/- AGM at 16 weeks after transplantation (n=3, **P=0.0075). (E) Repopulating ability of E11.5 and -/E12.5 WT and Wip1 YS by direct transplantation. (n=3, ***P=0.0004). (F) Donor-derived multilineage output in the peripheral blood of repopulated recipients of GM, T and B cells repopulated by E12.5 YS at 16 weeks after transplantation (n=3, *P=0.017). (G) Repopulating activity of E10.5 WT and Wip1-/AGM after 3 days of explant culture in vitro (AGMex). *P=0.0237. (H) Donor-derived lineage output of GM, T and B cells repopulated by E10.5 WT and Wip1-/- AGM explants at 16 weeks after transplantation. *P=0.0106, ***P<0.001. Each symbol representes one recipient. Green circles=WT, Red triangles= Wip1-/- . The lines represent average chimerism. CFU-C: colony-forming unit culture; ee: embryo equivalent; WT: wild-type; KO: knockout; CFU-Mix: colony-forming unit – granulocyte, erythrocyte, monocyte, megakaryocyte; CFU-GM: colony-forming unit – granulocytemacrophage; BFU-E: burst-forming unit – erythroid; GM: Gr1+/Mac-1+ cells; T: CD3+ cells; B: B220+ cells.

haematologica | 2021; 106(2)

Letters to the Editor

ablation decreased the percentage of CD31+c-Kithigh hematopoietic clusters in the E11.5 AGM (0.087±0.005% vs. 0.111±0.006%). Consistently, a greater than 35% reduction in the number of CD31+c-Kithigh cells was observed in the E11.5 Wip1-/- AGM (159±10 vs. 249±14

cells/AGM, respectively) (Figure 3N, O). Moreover, the hematopoiesis-related genes P2-Runx1 (a proximal P2 promoter of Runx1) and Gata2 were reduced significantly after the deletion of Wip1 in the E11.5 AGM clusters but increased in the E11.5 YS (Online Supplementary Figure

Figure 3. Pre-hematopoietic stem cell maturation was impeded by Wip1 ablation in the aorta-gonad-mesonephros region on embryonic day 11.5. (A) Flow cytometric analysis showing the percentages of pre-hematopoietic stem cell (HSC) I (CD31+ CD41lowCD45–) and pre-HSC II (CD31+CD45+) in wild-type (WT) and Wip1-/- aorta-gonad-mesonepros on embryonic day (E)11.5. (B-E) The percentages and absolute numbers of pre-HSC I (B and C, respectively) and pre-HSC II (D and E, respectively), respectively. n=6, **P<0.01. (F, G) The repopulating potential of pre-HSC I (CD31+CD45–) and pre-HSC II (CD31+CD45+) from E11.5 WT and Wip1-/- AGM regions after co-culture with OP9-DL1 cells for 6 days. Chimerism was determined in the peripheral blood at 16 weeks after transplantation. The lines represent average chimerism. (H, I) Transplantation assays showed the engraftment capacity of pre-HSC I (CD31+CD41lowCD45–) and pre-HSC II (CD31+CD45+) of E11.5 WT AGM regions co-cultured with OP9-DL1 with or without CCT007093 (CCT, 25 mM) for 4 days. **P<0.01. (J-M) Donor-derived multilineage output of myeloid (Gr-1+/Mac1+, GM), T (CD3+) and B (B220+) cells derived from pre-HSC I (J) and pre-HSC II (K) of E11.5 WT and Wip1-/- AGM region and of E11.5 WT AGM with or without the Wip1 inhibitor, CCT (L and M) at 16 weeks after transplantation. Ctr=dimethlysulfoxide control, *P<0.05, **P<0.01. (N) Representative flow cytometric analysis of CD31+c-Kithigh (CD31+c-Kithi) hematopoietic cluster cells in the E11.5 WT and Wip1-/- AGM region. (O) Percentages and absolute numbers of CD31+c-Kithigh cluster cells (n=9, ***P<0.001). (P) Cell cycle status of endothelial cells (EC, CD31+CD41–CD45–), pre-HSC I (I, CD31+CD41lowCD45–) and pre-HSC II (II, CD31+CD45+) from WT and Wip1-/- AGM regions was examined by Ki67 and 7-AAD staining (n=3, *P<0.05). WT: wildtype; KO: knockout; ee: embryo equivalent; GM: Gr-1+/Mac-1+ cells; T: CD3+ cells; B: B220+ cells.

haematologica | 2021; 106(2)

583

Letters to the Editor

S5A, B). Perhaps Wip1 is involved in the endothelial to hematopoietic transition and hematopoietic cell expansion by regulating Runx1 and Gata2. To test which step in hematopoiesis Wip1 deletion was affected, we sorted EC (CD31+CD41–CD45–Ter119–). The percentages of EC were indistinguishable between WT and Wip1-/- AGM (Online Supplementary Figure S5C). Three-day co-cultures of OP9 and EC showed similar numbers of hematopoietic clusters, related to the earlier stage of endothelial to hematopoietic transition in Wip1-/- AGM (Online Supplementary Figure S5D, E). However, after co-culture for 8 days, fewer CD45+ cells were found derived from Wip1-/- EC co-cultures (Online Supplementary Figure S5F), indicating that Wip1 mainly affects HPC maturation/expansion. As previously reported, Wip1 influences the cell proliferation/cell cycle status of bone marrow HSC.11,12,15 Flow cytometric assays showed no differences in cell cycle status of total cells between E11.5 Wip1-/- and WT AGM. Conversely, significantly higher percentages of EC in G1 phase were observed after Wip1 deletion (33.1±1.4% vs. 27.3±2.5%), with no differences in proliferation. Furthermore, a lower percentage of pre-HSC I was in the G0 phase (25.5±3.4% vs. 37.2±4.5%), however, the percentages of pre-HSC II in G1 phase were significantly higher in the Wip1-/- AGM region (40.0±3.1% vs. 31.26±2.6%) (Figure 3P). S/G2/M phases and proliferation status were not changed by Wip1 deficiency (Figure 3P, Online Supplementary Figure S6A-C). The apoptosis assay did not show any differences in the EC, pre-HSC I and II in Wip1-/- AGM (Online Supplementary Figure S6D6F), suggesting influences of Wip1 on the cell cycle. In summary, our study identifies a regulatory role for Wip1 in hematopoietic development during the mid-gestation stage. Transplantation assays after co-culture revealed that Wip1 regulates pre-HSC maturation in embryonic hematopoiesis, likely mediated by altered cell cycle status, laying a theoretical foundation for HSC regeneration in vitro. Wenyan He,1 Xiaobo Wang,2 Yanli Ni,2 Zongcheng Li,2 Wei Liu,2 Zhilin Chang,2 Haowen Li,1 Zhenyu Ju3 and Zhuan Li4 1 China National Clinical Research Center for Neurological Diseases, Beijing Tiantan Hospital, Capital Medical University, Beijing; 2 Laboratory of Oncology, Fifth Medical Center, General Hospital of PLA, Beijing; 3Key Laboratory of Regenerative Medicine of Ministry of Education, Guangzhou Regenerative Medicine and Health Guangdong Laboratory, Institute of Aging and Regenerative Medicine, Jinan University, Guangzhou and 4Department of Developmental Biology, School of Basic Medical Sciences, Southern Medical University, Guangzhou, China Correspondence: ZHUAN LI - zhuanli2018@smu.edu.cn ZHENYU JU - zhenyuju@163.com doi:10.3324/haematol.2019.235481 Disclosures: no conflicts of interests to disclose.

584

Contributions: ZL, ZJ and WH designed research; WH performed research; XW did some transplantation experiments; YN gave support on FACS; ZL supported bioinformatics analysis, WL, LC, HL prepared embryos; ZL and WH analyzed data; ZL, ZJ and WH wrote the paper. Acknowledgments: we are grateful to Dr. L.A. Donehower from Baylor College of Medicine for the Wip1 knockout mice. We thank Dr. B. Liu and Dr. Y. Lan for useful discussions. Funding: this work was supported by grants from the National Natural Science Foundation of China (81570095, 81870087, 91749203, 81525010), the National Key Research and Development Program (2019YFA0801802, SQ2019YFA0111100, 2016YFA0100602, 2017YFA0103302) and the Program for Guangdong Introducing Innovative and Entrepreneurial Teams (2017ZT07S347) and the Innovative Team Program of Guangzhou Regenerative Medicine and Health Guangdong Laboratory (2018GZR110103002).

References 1. Muller AM, Medvinsky A, Strouboulis J, Grosveld F, Dzierzak E. Development of hematopoietic stem cell activity in the mouse embryo. Immunity. 1994;1(4):291-301. 2. Medvinsky A, Dzierzak E. Definitive hematopoiesis is autonomously initiated by the AGM region. Cell. 1996;86(6):897-906. 3. Li Z, Lan Y, He W, Chen D, et al. Mouse embryonic head as a site for hematopoietic stem cell development. Cell Stem Cell. 2012; 11(5):663-675. 4. Li Z, Vink CS, Mariani SA, Dzierzak E. Subregional localization and characterization of Ly6aGFP-expressing hematopoietic cells in the mouse embryonic head. Dev Biol. 2016 ;416(1):34-41. 5. Ottersbach K, Dzierzak E. The murine placenta contains hematopoietic stem cells within the vascular labyrinth region. Dev Cell. 2005; 8(3):377-387. 6. Mikkola HK, Gekas C, Orkin SH, Dieterlen-Lievre F. Placenta as a site for hematopoietic stem cell development. Exp Hematol. 2005; 33(9):1048-1054. 7. Dzierzak E, Bigas A. Blood development: hematopoietic stem cell dependence and independence. Cell Stem Cell. 2018;22(5):639-651. 8. Chen MJ, Yokomizo T, Zeigler BM, Dzierzak E, Speck NA. Runx1 is required for the endothelial to haematopoietic cell transition but not thereafter. Nature. 2009;457(7231):887-891. 9. Bertrand JY, Chi NC, Santoso B, Teng S, Stainier DY, Traver D. Haematopoietic stem cells derive directly from aortic endothelium during development. Nature. 2010;464(7285):108-111. 10. Boisset JC, van Cappellen W, Andrieu-Soler C, Galjart N, Dzierzak E, Robin C. In vivo imaging of haematopoietic cells emerging from the mouse aortic endothelium. Nature. 2010;464(7285):116-120. 11. Chen Z, Yi W, Morita Y, et al. Wip1 deficiency impairs haematopoietic stem cell function via p53 and mTORC1 pathways. Nature Commun. 2015;6:6808. 12. Schito ML, Demidov ON, Saito S, Ashwell JD, Appella E. Wip1 phosphatase-deficient mice exhibit defective T cell maturation due to sustained p53 activation. J Immunol. 2006;176(8):4818-4825. 13. Zhou F, Li X, Wang W, et al. Tracing haematopoietic stem cell formation at single-cell resolution. Nature. 2016;533(7604):487-492. 14. Yokomizo T, Dzierzak E. Three-dimensional cartography of hematopoietic clusters in the vasculature of whole mouse embryos. Development. 2010;137(21):3651-3661. 15. Uyanik B, Grigorash BB, Goloudina AR, Demidov ON. DNA damage-induced phosphatase Wip1 in regulation of hematopoiesis, immune system and inflammation. Cell Death Discov. 2017; 3:17018.

haematologica | 2021; 106(2)

Letters to the Editor

CD4+ T-cell alloreactivity after haploidentical hematopoietic stem cell transplantation Allogeneic hematopoietic stem cell transplantation (HSCT) is an important, potentially curative therapeutic modality in many hematopoietic disorders.1 A fully human leukocyte antigens (HLA)-matched sibling or unrelated donor is the preferred source for stem cells.1 However, finding such a donor may pose a problem, especially for non-Caucasian patients. 1 Stem cells from a haploidentical donor or umbilical cord blood can serve as an alternative for those lacking a fully matched donor. In the past, haploidentical HSCT was associated with two problems; a higher rate of graft failure and, a higher incidence of graft-versus-host disease (GvHD), both necessitating intensive immunosuppressive measures.2 The development of conditioning regimens with intensified immunosuppressive therapy, including high-dose posttransplant cyclophosphamide (PT-Cy) therapy, improved results after haploindentical HSCT by ensuring engraftment, less GvHD, and a more favorable restoration of immune recovery.2 We earlier showed the strong presence of alloreactive CD4+ T cells in most patients early after double umbilical cord blood transplantation (dUCBT) and their potential role in enhancing the graftversus-leukemia effect.3,4 In this study, we addressed whether such HLA-class II specific CD4+ T cells are present early after haploidentical HSCT. The trial was approved by the ethics committees of the participating institutions and was conducted in accordance with the Declaration of Helsinki. All participants had given written informed consent. A total of 22 patients with high-risk hematologic diseases, eligible for haploidentical HSCT were retrospectively included in this study. The primary endpoint was the proportion of evaluable patients with activated class-II specific CD4+ T cells. Patients received myeloablative conditioning, consisting of thiotepa (10 mg/kg), busulfan 6.4 or 9.6 mg/kg and fludarabine 150 mg/m2, followed by transplantation of an unmanipulated bone marrow graft from a haploidentical family member, as previously described.2 GvHD prophylaxis consisted of PT-Cy, tacrolimus, and mycophenolate mofetil. Peripheral blood from patients was collected 1 month after transplantation for isolation of peripheral blood mononuclear cells (PBMC). Engraftment was achieved in all patients at the time of sampling. A library of stimulator cells expressing a single HLA-class II allele (HLA-DRA1/B1, HLA-DQA1/B1 or HLA-DPA1/B1) was generated by retroviral transduction of HLA-class II negative HeLa cells.4 Post-transplantation PBMC were assayed for the presence of alloreactive T cells. The very few alloreactive T cells were propagated in a co-culture with irradiated HeLa cells expressing a single mismatched HLA-class II allele derived from the recipient with addition of a cytokine mixture. Subsequently, alloreactivity was measured by flow cytometry and expressed as fold increase (%CD137+/CD4+) of reactivity toward HeLa cells transduced with mismatched HLA-class II alleles relative to reactivity towards the not-transduced ‘empty’ HeLa cells, as described previously.4 In addition, alloreactivity towards matched ‘control’ alleles were included (Online Supplementary Figure S1). The characteristics of the 22 patients and their grafts are presented in Table 1A and 1B, respectively. The majority of the patients were diagnosed with acute myeloid or lymphoblastic leukemia. Figure 1 shows immune recovery of natural killer (NK) cells, B cells, and haematologica | 2021; 106(2)

Table 1A. Patients’ characteristics.

Number of patients Diagnosis, n (%) Acute myeloid leukemia Acute lymphoblastic leukemia Myelofibrosis Non-Hodgkin lymphoma Plasma cell leukemia Male gender, n (%) Age, years, median (range) EBMT-risk score, n (%) 2-3 4-5 6-7

N=22

9 8 4 1 1 16

41 36 18 5 5 73 63(27-73)

7 13 2

32 59 9

EBMT: European Blood and Marrow Transplantation.

Table 1B. Graft characteristics.

Number of infused cells Nucleated cells (10 /kg) Viable CD34+ cells (103/kg) Viable CD3+ cells (106/kg) 6

Median (range) 398.8 (261.5-534.1) 3240.0 (1800.0-5800.0) 36.2 (25.2-60.1)

CD4+ and CD8+ T cells after haploidentical HSCT. Normal values were not reached during the follow-up of 3 months. High resolution typing was performed for all 22 patients. In vitro tests could be performed for 11 out of the 22 patients for whom HLA-class II transduced HeLa cells were available (Figure 2). T-cell numbers increased a median of 1.6-fold (range, 0.3-36.2) in the HLA-class II allele-specific propagation cultures of the post-haploidentical HSCT PBMC samples. In four out of 11 (36%) patients tested, CD4+ T-cell alloreactivity towards HLAclass II mismatched alleles of the recipient was detected, in three cases towards DR alleles and in one case towards a DQ allele (Figure 2). The median CD4+ T-cell alloreactivity towards DR, DQ and DP alleles was increased 19.7-fold (range, 1.3-200.0), 1.6-fold (range, 1.0-3.8) and 1.4-fold (range, 1.0-1.6), respectively, as compared to reactivity towards the not transduced ‘empty’ HeLa cells. In the present study, a positive CD4+ T-cell response towards ‘control’ matched alleles was not observed for any of the tested alleles. The magnitude of the CD4+ Tcell response was significantly higher towards mismatched alleles than towards matched alleles, with the median fold increases being 1.7 (range, 1.0-200.0) versus 1.0 (range, 0.9-1.8), respectively (P=0.03; Mann-Whitney test). It should be noted that alloreactivity towards controls could not be determined for three patients because of an insufficient number of expanded T cells (Figure 2). Both haploidentical HSCT and dUCBT are important alternative treatment modalities for patients in need of allogeneic HSCT but lacking a matched sibling or unrelated donor.1 The early challenges of transplant complications related to graft failure and GvHD have been overcome with new strategies such as using two cord blood units in dUCBT and intensified immunosuppressive therapy including high dose PT-Cy in haploidentical HSCT.2,5 Haploidentical HSCT has a number of advantages in comparison to dUCBT, including unlimited donor availability and accessibility of post-transplant donor-derived immune cells. We earlier showed the presence of donorderived CD4+ T cells specific for mismatched HLA class II 585

Letters to the Editor

Figure 1. Immune recovery after allogeneic hematopoietic stem cell transplantation. (A) CD4+ lymphocytes, (B) CD8+ lymphocytes, (C) natural killer cells, (D) B-lymphocytes. Dashed lines represent the upper and lower limits of reference values.

alleles in 92% of patients after dUCBT. These alloreactive CD4+ T cells showed activity towards patient-derived leukemic cells, especially when the mismatch was shared between recipient and the â&#x20AC;&#x2DC;loserâ&#x20AC;&#x2122; cord blood unit.3,4 In the present study, we set out to explore HLA-class II-specific CD4+ T-cell alloreactivity after haploidentical HSCT. Here we show the presence of such T cells for 36% of the tested alleles, which is much lower than the 92% detected after dUCBT, supporting our hypothesis that CD4+ T-cell alloreactivity may be less predominant after haploidentical HSCT.3 Current standard regimens of GvHD prophylaxis include the combination of a calcineurin inhibitor, cyclosporine, or tacrolimus, and either mycophenolate mofetil, or sirolimus, while prophylaxis in haploidentical HSCT consists of additional high-dose PT-Cy.6 Mycophenolate mofetil inhibits T-cell proliferation and adhesion, which hamper T-lymphocytic responses to allogeneic cells.7 Calcineurin inhibitors specifically inhibit intracellular signals responsible for T-cell activation after antigen binding, while PT-Cy induces apoptosis in alloreactive donor T cells after their activation and entry into the proliferative phase.8,9 Thus, it might further reduce the proliferation of early reconstituting alloreactive T cells with antileukemic potential. In addition, NK-cell recovery might also be compromised. Russo et al. recently studied NK-cell recovery and alloreactivity after haploidentical HSCT with PT-Cy.10 In that study, NK-cell proliferation was completely abrogated by day 8 after PT-Cy, suggesting selective elimination of cycling NK cells. 586

Importantly, donor NK cells harvested from patients at days 15 and 30 after haploidentical HSCT had a less mature phenotype. In addition these immature NK cells displayed less alloreactivity and an impaired antileukemic potential. Overall, haploidentical HSCT with PT-Cy is characterized by a delayed early immune reconstitution, especially of CD4+ T cells, in comparison to the reconstitution following transplantation with donor cells from other sources, reaching normal levels after roughly 1 year.11 Antigen-presenting cells play a central role in activating T cells and initiating immune responses. Several studies have shown that recovery of dendritic cells is also delayed after haploidentical HSCT and normal values are gradually achieved 1 year after transplantation.11,12 Delayed recovery of such antigen-presenting cells may lead to less presentation of recipient class II antigens and contribute to the relatively low numbers of alloreactive CD4+ T cells found in this study. A recent study explored dendritic cell recovery after dUCBT and matched sibling donor transplantation.13 Interestingly, umbilical cord blood recipients had better dendritic cell reconstitution in comparison to recipients of grafts from matched related donors, with normal values being reached at day 100 after transplantation. In addition, better dendritic cell recovery in the cord blood setting was associated with fewer relapses. Moreover, circulating class II-expressing cells after infusion of both cord blood units in a dUCBT might contribute to presenting mismatched HLA-class II to CD4+ T cells. Strong expression and/or a high number haematologica | 2021; 106(2)

Letters to the Editor

Figure 2. T-cell expansion, control reactivity and alloreactivity response. Eleven samples of peripheral blood mononuclear cells obtained from 22 patients 1 month after haploidentical hematopoietic stem cell transplantation were propagated towards mismatched HLA-class II alleles of the recipient in a T-cell and Hela-cell co-culture, using a panel of Hela cell lines, each transduced with a single HLA-class II allele. T-cell expansion of individual cultures at day 21 is presented as fold increase in cell number. Propagated T cells were subsequently tested for (allo)reactivity towards that same mismatched, and matched HLA-class II alleles. The response was quantified by CD137 upregulation on CD4+ T cells (%CD137+/CD4+) and presented for the alloreactivity as ‘fold increase (%CD137+/CD4+)’ of reactivity towards HeLa cells transduced with mismatched HLA-class II alleles relative to reactivity towards the not-transduced ‘empty’ HeLa cells. The level of T-cell expansion, control reactivity (towards matched alleles) and alloreactivity response are presented in a four-color grading scale. PBMC: peripheral blood mononuclear cells; HAPLO: haploidentical stem cell transplantation; HLA: human leukocyte antigen; FI: fold increase.

of antigen-presenting cells might evoke a stronger CD4+ T-cell response. In conclusion, this study showed the presence of alloreactive CD4+ T cells early after haploidentical HSCT, although less frequently than previously observed in the dUCBT-setting.3 Intensified immunosuppression, especially PT-Cy, to prevent GvHD might blunt the CD4+ T-cell response, with possible loss of part of the graft-versus-leukemia activity. Our observations might help the understanding of the kinetics of immune responses in both haploidentical and cord blood transplantation. However, a prospective comparison would be needed to quantify the different immune responses in more detail. Moreover bone marrow was the sole source of the grafts in our study, whereas the use of PBSC as a transplant source might be associated with increased alloreactivity after transplantation.14 Burak Kalin,1 Elisabetta Metafuni,2 Mariëtte ter Borg,1 Rebecca Wijers,3 Eric Braakman,1 Cor H. J. Lamers,3 Andrea Bacigalupo2 and Jan J. Cornelissen1 1 Erasmus University Medical Center, Department of Hematology, Rotterdam, the Netherlands; 2Istituto di Ematologia, Fondazione Policlinico Universitario A Gemelli IRCCS, Roma, Italy and 3Erasmus MC Cancer Institute, Department of Medical Oncology, Laboratory of Tumor Immunology, Rotterdam, the Netherlands Correspondence: JAN J. CORNELISSEN - j.cornelissen@erasmusmc.nl. doi:10.3324/haematol.2019.244152 haematologica | 2021; 106(2)

Disclosures: no conflicts of interests to disclose. Contributions: BK, AB, CHJL and JJC contributed to the study design; all authors provided study materials or patients; all authors were involved in collection and assembly of clinical data; BK, AB and JJC were involved in analyzing and interpreting the data and writing this report; and all authors reviewed and approved the final version of the manuscript.

References 1. Copelan EA. Hematopoietic stem-cell transplantation. N Engl J Med. 2006;354(17):1813-1826. 2. McCurdy SR, Luznik L. How we perform haploidentical stem cell transplantation with posttransplant cyclophosphamide. Blood. 2019;134(21):1802-1810. 3. Kalin B, ter Borg M, Wijers R, et al. Double umbilical cord blood transplantation in high-risk hematological patients: a phase II study focusing on the mechanism of graft predominance. HemaSphere. 2019;3(5):e285. 4. Lamers CH, Wijers R, van Bergen CA, et al. CD4+ T-cell alloreactivity toward mismatched HLA class II alleles early after double umbilical cord blood transplantation. Blood. 2016;128(17):2165-2174. 5. Barker JN, Weisdorf DJ, DeFor TE, et al. Transplantation of 2 partially HLA-matched umbilical cord blood units to enhance engraftment in adults with hematologic malignancy. Blood. 2005;105(3):1343-1347. 6. Luznik L, Fuchs EJ. High-dose, post-transplantation cyclophosphamide to promote graft-host tolerance after allogeneic hematopoietic stem cell transplantation. Immunol Res. 2010;47(1-3):65-77. 7. Allison AC, Eugui EM. Mycophenolate mofetil and its mechanisms of action. Immunopharmacology. 2000;47(2-3):85-118. 8. Liu J, Farmer JD, Jr., Lane WS, Friedman J, Weissman I, Schreiber SL. Calcineurin is a common target of cyclophilin-cyclosporin A and

587

Letters to the Editor

FKBP-FK506 complexes. Cell. 1991;66(4):807-815. 9. Eto M, Mayumi H, Tomita Y, Yoshikai Y, Nishimura Y, Nomoto K. Sequential mechanisms of cyclophosphamide-induced skin allograft tolerance including the intrathymic clonal deletion followed by late breakdown of the clonal deletion. J Immunol. 1990;145(5):13031310. 10. Russo A, Oliveira G, Berglund S, et al. NK cell recovery after haploidentical HSCT with posttransplant cyclophosphamide: dynamics and clinical implications. Blood. 2018;131(2):247-262. 11. McCurdy SR, Luznik L. Immune reconstitution after T-cell replete HLA-haploidentical transplantation. Semin Hematol. 2019;56(3):221226.

588

12. Pérez-Martínez A, González-Vicent M, Valentín J, et al. Early evaluation of immune reconstitution following allogeneic CD3/CD19depleted grafts from alternative donors in childhood acute leukemia. Bone Marrow Transplant. 2012;47(11):1419-1427. 13. Touma W, Brunstein CG, Cao Q, et al. Dendritic cell recovery impacts outcomes after umbilical cord blood and sibling donor transplantation for hematologic malignancies. Biol Blood Marrow Transplant. 2017;23(11):1925-1931. 14. Bashey A, Zhang MJ, McCurdy SR, et al. Mobilized peripheral blood stem cells versus unstimulated bone marrow as a graft source for Tcell-replete haploidentical donor transplantation using post-transplant cyclophosphamide. J Clin Oncol. 2017;35(26):3002-3009.

haematologica | 2021; 106(2)

Letters to the Editor

Daratumumab inhibits acute myeloid leukemia metabolic capacity by blocking mitochondrial transfer from mesenchymal stromal cells Acute myeloid leukemia (AML) proliferation is dependent on a complex multi-faceted interplay between the blasts and the bone marrow (BM) microenvironment.1 We and others have previously demonstrated that functional mitochondria are transferred from the mesenchymal stem cells (MSC) to the AML blasts facilitating progression of the disease,2,3 in a process which is hijacked from hematopoietic stem cell response to infection.4 Clinical observation trials using venetoclax, which targets BCL2 (which in turn is a key regulator of the mitochondrial apoptotic pathway), in combination with hypomethylating agents showed tolerable safety and favorable overall response rate in elderly patients with AML.5 Taken together, these data imply that mitochondria represent an attractive and biologically plausible drug target in the treatment of AML. CD38 is a transmembrane glycoprotein which is expressed on many cells of the hematopoietic system including malignant plasma cells, red blood cells, myeloid cells, lymphoid cells and subsets of leukemia-initiating AML blasts.6-8 Furthermore, we have recently shown that

CD38 mediates pro-tumoral mitochondrial transfer from MSC to malignant plasma cells in myeloma,9 which in part explains why daratumumab (an anti-CD38 monoclonal antibody) is clinically effective in treating patients with myeloma.10 More recently, daratumumab has been shown to have preclinical activity in AML.11 Therefore, taken together, we hypothesize that daratumumab treatment would impair AML metabolic capacity and consequently inhibit tumor proliferation, via a mechanism which blocks mitochondrial transfer from BMSC to the blasts. Initially, to determine if CD38 inhibition blocks mitochondrial transfer from MSC to AML blasts, we used an in vitro co-culture system. MitoTracker Green FM stain (MTG) was used to quantify mitochondria in AML after co-culture with MSC. We incubated both MSC and AML with MTG. The cells were washed twice in phosphate buffered saline (PBS) and incubated for 4 hours (h). The cells were then co-cultured for 24 h with and without daratumumab. Using flow cytometry, AML was shown to have less MTG fluorescence when treated with daratumumab (Figure 1A-C). Figure 1D shows the presence of mouse mtDNA in human AML after co-culture with mouse MSC, and the transfer of mouse mtDNA to human AML was inhibited by the addition of daratu-

Figure 1. CD38 inhibition blocks mitochondrial transfer from mesenchymal stem cells (MSC) to acute myeloid leukemia (AML) blasts. (A) Primary AML and MSC were pre-stained with MitoTracker green (MTG) fluorescent mitochondrial dye (FM) for 1 hour (h) and then cultured alone or together for 24 h. Flow cytometry was used to detect MTG FM in the AML blasts (n=8). (B) Primary AML and MSC were pre-stained with MTG FM for 1 h and then cultured together for 24 h in the presence of vehicle or daratumumab (Dara) (100 ng/mL). Flow cytometry was used to detect MTG FM in the AML blasts (n=6). (C) Primary AML were prestained with MTG FM for 1 h and then treated with vehicle or daratumumab for 24 h (100 ng/mL). Flow cytometry was used to detect MTG FM in the AML blasts (n=5). (D) Human AML was cultured with mouse MSC in the presence of vehicle or daratumumab for 24 h (100 ng/mL) for 24 h. AML were isolated and measured for mouse mitochondrial DNA content. Primary AML were transduced with CD38 targeted shRNA for 48 h. (E) RNA was then extracted and examined for CD38 mRNA expression. (F) AML and MSC were then pre-stained with MTG FM for 1 h and then cultured together for 24 h. Flow cytometry was used to detect MTG FM in the AML blasts (n=4). We used the Mann-Whitney U test and Wilcoxon matched pairs signal rank test to compare results between groups. MFI: mean fluorescence intensity. *P<0.05; **P<0.01.

haematologica | 2021; 106(2)

589

Letters to the Editor

mumab. We further showed that knockdown of CD38 inhibited mitochondrial content in AML when cultured with MSC (Figure 1E and F). We next investigated if daratumumab could inhibit AML disease progression in vivo. To do this we used an NSG mouse model of AML whereby on day 1 we transplanted 0.5x106 OCI-AML3 cells tagged with a luciferase construct into the tail vein of NSG mice.3 We then treated the NSG animals with either vehicle control (PBS) or daratumumab (5 mg/kg) on days 9 and 16 by intra-peritoneal injection (Figure 2A). The mice were then imaged using bioluminescence on day 21. Figure 2B shows bioluminescence from live in vivo imaging. The pre-treatment tumor burden was the same between treatment and control animals on day 9; however, following treatment with

daratumumab, the tumor burden in treated animals was significantly reduced when compared to mice in the vehicle control group. The densitometry measurement of these images are shown in Figure 2C to illustrate the differences between the vehicle control group and daratumumab-treated animals. These data show that there was less tumor-derived bioluminescence intensity in the daratumumab-treated animals when compared to control. Daratumumab-treated animals also had increased overall survival compared to control animals (Figure 2D). To determine how daratumumab altered the metabolic profile and function of the AML cells in vivo, we again transplanted 0.5x106 OCI-AML3-luc cells and treated the mice as described in Figure 2. The mice were then sacrificed on day 21. OCI-AML3-luc cells were isolated by cell

Figure 2. Daratumumab (Dara) inhibits acute myeloid leukemia (AML) disease progression in vivo. (A) Schematic of the in vivo model used. (B) 0.5x106 OCIAML3-luc cells were injected into the tail vein of NSG mice. Mice were imaged using bioluminescence at day 9 following injection to confirm tumor engraftment, and then split into two groups. Group 1 received vehicle (PBS) and group 2 received Dara (5 mg/kg) on days 9 and 16 by intra-peritoneal (IP) injection. Mice were then imaged using bioluminescence at day 21 and then sacrificed. (C) Densitometry of the bioluminescent images shown in (B) was performed to determine differences between vehicle and Dara-treated animals. (D) In a separate experiment, Kaplan-Meier survival curves were performed for C57BL/6 (n=5) mice injected with 0.5x106 OCI-AML3-luc and then treated with vehicle or Dara at days 9 and 16 post injection. We used the Mann-Whitney U test to compare results between groups. The Mantel-Cox test was used to analyze Kaplan-Meier survival curves.

590

haematologica | 2021; 106(2)

Letters to the Editor

To investigate if mitochondrial transfer from the BM microenvironment to the AML was altered with daratumumab treatment, the human OCI-AML3-luc cells were isolated from murine BM by cell sorting and analyzed for the presence of mouse mitochondrial DNA within the human OCI-AML3-luc cells by TaqMan real-time polymerase chain reaction. Figure 3E confirms that daratumumab treatment inhibited mouse mitochondrial DNA transfer to human OCI-AML3-luc cells. Together, these results suggest that, in vivo, daratumumab treatment causes OCI-AML3-luc cells to decrease in AML mitochondrial mass, and that AML transfer from MSC to AML blasts is reduced and may contribute to the inhibitory functional bio-energetic consequence in AML metabolism in the BM microenvironment. In conclusion, CD38 inhibition results in reduced mitochondrial transfer from MSC to AML blasts in the BM microenvironment, resulting in a reduction in AML-

sorting from mouse BM and the cells were analyzed ex vivo (Figure 3A). The mitochondrial mass within the OCI-AML3-luc cells was not significantly reduced in the animals treated with daratumumab compared to the control animals (Figure 3B). Mitochondrial potential measured by tetramethylrhodamine, methyl ester, perchlorate staining shows that the OCI-AML3-luc cells had a reduced mitochondrial potential in the daratumumabtreated animals compared to the NSG mice who received the control vehicle (Figure 3C). To determine whether daratumumab altered mitochondrial-based metabolism, we analyzed the OCI-AML3-luc cells isolated from the vehicle control and the daratumumab-treated animals and then measured oxygen consumption rate using the Seahorse Cell Mito Stress Test assay. Figure 3D shows decreased mitochondrial respiration was observed in the OCI-AML3-luc cells from daratumumab-treated animals compared to OCI-AML3-luc cells from the control mice.

Figure 3. Daratumumab (Dara) alters the metabolic function of acute myeloid leukemia (AML) cells in vivo. (A) Schematic of the in vivo model for these experiments. 0.5x106 OCI-AML3-luc cells were injected into the tail vein of NSG mice. Mice were imaged using bioluminescence at day 9 following injection to confirm tumor engraftment, and then split into two groups. Group 1 received vehicle (PBS) and group 2 received Dara (5 mg/kg) treatment on days 9 and 16 by intraperitoneal (IP) injection. Mice were then sacrificed at day 21 and OCI-AML3-luc cells were isolated by cell sorting from mouse bone marrow and analyzed for (B) mitochondrial content by MitoTracker green staining. (C) Mitochondrial potential by tetramethylrhodamine, methyl ester, perchlorate (TMRM) staining. (D) Oxygen consumption rate (OCR) by Seahorse Cell Mito Stress Test kit. (E) Presence of mouse mitochondrial DNA by real-time polymerase chain reaction. We used the Mann-Whitney U test to compare results between groups. *P<0.05.

haematologica | 2021; 106(2)

591

Letters to the Editor

derived oxidative phosphorylation and subsequent reduced tumor burden. While it is likely that daratumumab functions through a number of mechanisms of action, here we show in an NSG mouse model lacking functional B cells, T cells and natural killer cells, that inhibition of mitochondrial transfer can be added to the list of mechanisms of action for this drug in AML. These data support the further investigation of daratumumab as a therapeutic approach for the treatment of mitochondrial-dependent tumor growth. Jayna J. Mistry,1,2 Jamie A. Moore,1 Prakrit Kumar,1 Christopher R. Marlein,1 Charlotte Hellmich,1,3 Genevra Pillinger,1 Aisha Jibril,1 Federica Di Palma,1,2 Angela Collins,3 Kristian M. Bowles1,3 and Stuart A Rushworth1 1 Norwich Medical School, University of East Anglia, Norwich Research Park, Norwich; 2Earlham Institute, Norwich Research Park, Norwich and 3Department of Haematology, Norfolk and Norwich University Hospitals NHS Trust, Norwich, UK Correspondence: KRISTIAN BOWLES - k.bowles@uea.ac.uk STUART RUSHWORTH - s.rushworth@uea.ac.uk doi:10.3324/haematol.2019.242974 Disclosures: KMB and SAR received funding from Jannsen Pharmaceuticals for this study. Contributions: KMB and SAR designed the research; JJM, JAM, CRM, CH, GP, PK, AJ and SAR performed the research; AC, FDP, and KMB provided essential reagents and knowledge; JJM, KMB and SAR wrote the paper. Acknowledgments: the authors also thank Allyson Tyler, Dr. Ian Thirkettle and Karen Ashurst from the Laboratory Medicine Department at the Norfolk and Norwich University Hospital for technical assistance. Funding: the authors wish to thank the Norwich Medical School, the United Kingdom National Health Service, the Big C and the Rosetrees Trust for funding, and the Norwich Biorepository (UK)

592

for help with sample collection and storage. pCDH-Lucferase-T2AmCherry was kindly gifted by Prof. Irmela Jeremias, MD, from Helmholtz Zentrum MĂźnchen, Munich, Germany. KMB and SAR receive research funding from Janssen.

References 1. Shafat MS, Gnaneswaran B, Bowles KM, Rushworth SA. The bone marrow microenvironment - home of the leukemic blasts. Blood Rev. 2017;31(5):277-286. 2. Moschoi R, Imbert V, Nebout M, et al. Protective mitochondrial transfer from bone marrow stromal cells to acute myeloid leukemic cells during chemotherapy. Blood. 2016;128(2):253-264. 3. Marlein CR, Zaitseva L, Piddock RE, et al. NADPH oxidase-2 derived superoxide drives mitochondrial transfer from bone marrow stromal cells to leukemic blasts. Blood. 2017;130(14):1649-1660. 4. Mistry JJ, Marlein CR, Moore JA, et al. ROS-mediated PI3K activation drives mitochondrial transfer from stromal cells to hematopoietic stem cells in response to infection. Proc Natl Acad Sci U S A. 2019; 116(49):24610-24619. 5. DiNardo CD, Pratz K, Pullarkat V, et al. Venetoclax combined with decitabine or azacitidine in treatment-naive, elderly patients with acute myeloid leukemia. Blood. 2019;133(1):7-17. 6. Deaglio S, Mehta K, Malavasi F. Human CD38: a (r)evolutionary story of enzymes and receptors. Leuk Res. 2001;25(1):1-12. 7. Taussig DC, Vargaftig J, Miraki-Moud F, et al. Leukemia-initiating cells from some acute myeloid leukemia patients with mutated nucleophosmin reside in the CD34(-) fraction. Blood. 2010; 115(10):1976-1984. 8. Sarry J-E, Murphy K, Perry R, et al. Human acute myelogenous leukemia stem cells are rare and heterogeneous when assayed in NOD/SCID/IL2RÎłc-deficient mice. J Clin Invest. 2011;121(1):384395. 9. Marlein CR, Piddock RE, Mistry JJ, et al. CD38-driven mitochondrial trafficking promotes bioenergetic plasticity in multiple myeloma. Cancer Res. 2019;79(9):2285-2297. 10. van de Donk NWCJ, Usmani SZ. CD38 antibodies in multiple myeloma: mechanisms of action and modes of resistance. Front Immunol. 2018;9:2134. 11. Naik J, Themeli M, de Jong-Korlaar R, et al. CD38 as a therapeutic target for adult acute myeloid leukemia and T-cell acute lymphoblastic leukemia. Haematologica. 2019;104(3):e100-e103.

haematologica | 2021; 106(2)

Letters to the Editor

Clinicopathological and genetic features of limited-stage diffuse large B-cell lymphoma with late relapse: targeted sequencing analysis of gene alterations in the initial and late relapsed tumors According to the long-term analysis of several clinical studies, late relapse (LR) is more common in patients with limited-stage diffuse large B-cell lymphoma (DLBCL) than in those with advanced-stage DLBCL;1,2 therefore a unique but poorly understood tumor biology has been proposed for limited-stage DLBCL.3 Cell of origin (COO), defined by gene expression profiling4 or immunohistology,5 and patterns of gene alterations, identified by next-generation sequencing, have recently allowed the classification of DLBCL into biologically and clinically distinct subgroups.6,7 Several studies have evaluated the genetic alterations in DLBCL that occur between initial diagnosis and relapse;8,9 however, there is limited information concerning such alterations in LR. This study focused on patients with limited-stage DLBCL who developed LR and evaluated the clinicopathological and genetic features of paired specimens obtained at initial diagnosis and LR. We aimed to reveal the biological background and mechanisms of LR. The retrospective study protocol was approved by the Institutional Review Board of the National Cancer Center Hospital (NCCH). Detailed methods are available in the Online Supplementary Appendix. We defined LR as the first relapse of any B-cell lymphoma (including indolent B-cell lymphoma) that occurred at >5 years after the initial diagnosis. Between 1997 and 2012, 334 consecutive patients with limited-stage de novo DLBCL were treated using CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisolone) with or without rituximab therapy at the NCCH. Regular recurrences were observed (Figure 1), and of the 334 patients, 16 developed LR during a median follow-up period of 8.9 years (interquartile range [IQR]: 6.2–12.3 years). Additionally, three patients developed

LR after initial treatment at other institutions and were subsequently referred to the NCCH. Then, 19 patients (patients #1-19) were subjected to the clinicopathological and genetic evaluations (Tables 1, Online Supplementary Table S1–2). Immunohistochemical analysis of formalin-fixed, paraffin-embedded (FFPE) tissues was performed using antibodies against MYC, BCL2, programmed death-ligand 1 (PD-L1), and COO markers (CD10, BCL6, MUM1) (Online Supplementary Table S3). Fluorescence in situ hybridization (FISH) was performed using probe mapping 9p24.1 for specimens with PD-L1 tumor expression and commercial MYC break-apart probe. Genomic DNA (gDNA) was extracted from (i) the primary tumor, (ii) LR tumor, and (iii) control bone marrow without lymphoma infiltration in 12 patients (patient #112) who were eligible for further genetic evaluations. Immunoglobulin H (IgH) clonality asssays and target sequencing of 64 lymphoma-related genes (Online Supplementary Table S4) were performed using the paired gDNA samples from the initial diagnosis and relapsed tumor. Mutations in the paired tumor specimens, which were defined as described in the Online Supplementary Appendix and Online Supplementary Table S5, were compared between samples, and shared/non-shared mutations among the samples in the pair were identified. The median age at the initial diagnosis of the 19 patients was 60 years (range: 32–77 years), 13 (68%) patients had stage I disease, and all had low or low-intermediate International Prognostic Index (IPI) risk. The median duration from the initial diagnosis to LR was 8 years (range: 5–18 years). At LR, 13 (68%) patients had an advanced-stage disease, and 10 (53%) had high-intermediate or high IPI risk (Table 1). At initial diagnosis, 14 of the 19 patients (78%) had a non-germinal center B-cell like (non-GCB) DLBCL, although 4 patients experienced relapse that involved a different COO subtype (from the initial diagnosis). At LR, three patients relapsed with indolent lymphoma as

Figure 1. Progression-free survival of the 334 patients with limited-stage diffuse large B-cell lymphoma who received CHOP with/without rituximab. No plateau was observed in the progression-free survival (PFS) curve. CHOP: cyclophosphamide, doxorubicin, vincristine, and prednisolone.

haematologica | 2021; 106(2)

593

Letters to the Editor

Table 1. Characteristics of 19 patients who developed late relapsed diffuse large B-cell lymphoma Median age, years (range) Sex, male : female, n Ann Arbor clinical stage, I / II / III or IV, n Number of extranodal diseases, 0 / 1 / ≥2, n IPI risk, n Low/ low-int High-int/ high Site of extranodal disease, n Gingiva Nasal or paranasal cavity Testis Skin or subcutaneous Bone or bone marrow Intestine or colon Liver Other sites Immunohistochemistry Cell of origin GCB / non–GCB / NA / non–DLBCL, n PD–L1 on tumor cells, + / BCL2, + / - / NA, n MYC, + / - / NA, n

At initial diagnosis

At late relapse

60 (32–77) 11 : 8 13 / 6 / 0 8 / 11 / 0

71 (45–84)

19 0

9 10

3 3 3 0 0 0 0 2

0 2 3 6 4 3 2 6

4 / 14 / 1 / 0

5 / 12 / 0 / 2

0 / 19 9/7/3 6/7/6

2 / 17 11 / 5 / 3 5/8/6

5 / 1 / 13 2 / 10 / 7

DLBCL: diffuse large B-cell lymphoma; IPI: International Prognostic Index; GCB: germinal–center B-cell–like; NA: not assessed; PD–L1: programmed death–ligand 1; int: intermediate.

defined by histology. Two of the 19 patients (#8 and #16) exhibited tumor expression of PD-L1 at LR, despite having negative expression at initial diagnosis. The FISH analysis revealed a 9p24.1 gain in patient #8 (Online Supplementary Figure S1). The expression of MYC was evaluated in 13 pairs; it turned positive at LR in four patients, one of whom gained an MYC translocation detected by FISH analysis. BCL2 expression was evaluated in 15 pairs; seven pairs were positive, and three became positive at LR (Online Supplementary Table S6). The summary of the results and representative examples of the IgH clonality assays are shown in the Online Supplementary Table S7 and Online Supplementary Figure S2. Of the 12 pairs, 7 (58%) were judged to be clonally related, 2 (17%) clonally unrelated, while the clonal relationship could not be determined in 3 (25%). The summary of the target sequencing results is described in the Online Supplementary Table S8-9. Patient #9 was excluded from further analysis due to the poor specimen sequence quality at initial diagnosis. The key results of the targeted sequencing, plus the clinicopathological information and IgH clonality findings are shown in Figure 2. Shared mutations were detected in 9 of the 11 patients. The most frequent shared mutation (n=7) was a CD79B missense mutation that involved a single substitution in the immunoreceptor tyrosine-based activation motif domain (Y196 [n=5] or L199 [n=2]). Next to this was an MYD88 missense mutation (L265P) (n=5). Four patients had shared mutations in both of the CD79B and MYD88 genes. Of the eight patients with shared mutations in the CD79B and/or MYD88 genes, all had non-GCB DLBCL. From the IgH clonality analysis, 6 of the 8 patients had a clonally relat594

ed relapse. All except 1 (patient #11) with tumor mutations in the CD79B and/or MYD88 genes had initially presented with an extranodal disease in the testis (n=3), nasal/paranasal cavity (n=2), or gingiva (n=2). In our study, non-GCB types were more frequent in patients who developed LR, which is contrary to several previous studies.10,11 The difference in clinical characteristics of patients might explain this discrepancy. Our study focused only on the patients with limited-stage DLBCL and excluded those with transformed B-cell lymphoma at initial diagnosis, which might have lowered the number of patients with the GCB type. Two recent large-scale studies, which examined the genetic characteristics of untreated DLBCL reported a similar genetic subtype of DLBCL, which typically involved comutations in the CD79B and MYD88 genes. These subtypes were described as the MCD type7 and cluster-5 type,6 which was associated with the activated B-cell type of COO and extranodal involvement.6,7 Our results suggest that tumors with CD79B and/or MYD88 mutations in eight patients may belong to this genetic subgroup; at least, tumors with both MYD88 and MYD88 mutations would belong to this subgroup. In the current study, patients with MYD88 and/or MYD88 tumor mutations all had the non-GCB type and extranodal sites involvement (except for patient #11). The survival curves of patients with the MCD type and cluster-5 type showed an acute decline within 12 months; after that, a gradual but continuous deterioration.6,7 On the one hand, it can be speculated that most patients with these subtypes show aggressive clinical courses. However, limited-stage DLBCL, especially with low IPI risk of these subtypes, might respond to haematologica | 2021; 106(2)

Letters to the Editor

Figure 2. The results of shared and non-shared mutation analysis combined with the clinicopathological and IgH-clonality characteristics. The shared and nonshared mutations in the paired tumor samples were described separately. The numbers of mutations in each gene were expressed as shades of grey and black. All eight patients with tumor mutations in CD79B and/or MYD88 had non-germinal-center B-cell-like lymphoma (DLBCL). Clonal relationships between the paired tumors, which were determined using immunoglobulin heavy chain PCR analysis, were confirmed in 6 of the 8 patients. No clear trends were observed in the patterns of the non-shared mutations.

chemotherapy once, but be prone to continuous relapse, as observed in our study. The paired specimens of 2 patients (#8 and #10) were judged to be clonally unrelated based on the IgH clonality analysis. However, patient #8 had shared mutations in the paired specimens that involved CREBBP, KMT2D, MEF2B, and PIM1. Furthermore, FISH analysis using FFPE specimens revealed BCL2/IGH translocations in the paired tumors (data not shown). Abnormalities in the CREBBP, KMT2D, MEF2B, and PIM1 genes are reportedly associated with lymphomagenesis,8,12-14 and lymphoma-progenitor cells acquiring these mutations may develop DLBCL with different IgH rearrangements in the same patient. The background of the clonally unrelated relapse could not be determined in patient #10, and a rare second primary DLBCL might occur in this patient. Our findings suggested that the mechanisms of LR were heterogeneous. The acquired overexpression of MYC or BCL28 might be associated with LR in some patients, although it is not considered specific for LR. We cannot definitively comment on the possibility that recurrent mutations were associated with LR; the two most common non-shared mutations involved PIM1 (five samples) and KMT2D (two samples), although they were not consistently detected at initial diagnosis or LR. At LR, two patients expressed PD-L1, and three had acquired loss of function mutations involving B2M, CD58, or CIITA, which might be associated with an immune evasion mechanism related to avoidance of antigen presentation. The present study has several limitations. First, only a small number of patients developed LR. Second, many of the initial tumor specimens were old, which might have affected the results of the genetic analyses. However, the reliability of shared mutations is likely robust, as it is haematologica | 2021; 106(2)

extremely rare to identify the same mutation in paired samples by chance. Furthermore, we only considered 64 lymphoma-related genes, and other important genetic alterations might be missed. Again, we did not perform the genetic analyses or COO evaluations for all patients with limited-stage DLBCL; thus, it remains unclear whether our findings are truly specific for LR development. In conclusion, the present study revealed that patients with limited-stage DLBCL and LR tended to have CD79B and/or MYD88 tumor mutations, the non-GCB type of DLBCL, and extranodal disease at the initial presentation. However, the mechanisms of LR appear to be heterogeneous, and further studies are needed to confirm our results. Tomotaka Suzuki,1 Suguru Fukuhara,1 Junko Nomoto,1,4 Satoshi Yamashita,3 Akiko (Miyagi) Maeshima,2 Yuta Ito,1 Shunsuke Hatta,1 Sayako Yuda,1 Shinichi Makita,1 Wataru Munakata,1 Tatsuya Suzuki,1 Dai Maruyama,1 Hirokazu Taniguchi,2 Toshikazu Ushijima,3 Koji Izutsu,1 Kensei Tobinai1 and Yukio Kobayashi1,5 1 Department of Hematology, National Cancer Center Hospital; 2 Department of Pathology, National Cancer Center Hospital; 3 Department of Epigenomics, National Cancer Center Research Institute; 4Institute of Medical Genomics, IUHW, Chiba and 5 Department of Hematology, IUHW Mita Hospital, Tokyo, Japan Correspondence: SUGURU FUKUHARA - sufukuha@ncc.go.jp doi:10.3324/haematol.2019.235598 Disclosures: TS reports personal fees from Chugai Pharmaceutical, outside the submitted work; DM reports grants and personal fees from Chugai, grants and personal fees from Takeda, during the conduct of the study; grants from Sanofi, grants and personal fees from Janssen, grants and personal fees from Eisai, grants and personal fees from Celegene, grants and personal fees from Kyowa Hakko Kirin, 595

Letters to the Editor

grants and personal fees from Ono, grants and personal fees from Mundipharma, grants and personal fees from MSD, grants and personal fees from Zenyaku, personal fees from Sumitomo, personal fees from Asahi Kasei, grants and personal fees from Bristol-Myers Squibb, grants and personal fees from Daiichi Sankyo, grants and personal fees from AstraZeneca, personal fees from Linical, personal fees from Pharma International, personal fees from Fujimoto, grants from Abbvie, grants from Astellas, grants from Amgen Astellas Biopharma, grants from Otsuka, grants from Novartis, grants from Pfizer, grants from Solasia, grants from Bayer, grants from Symbio, grants from CMIC, grants from Quintiles, grants from IQvia, outside the submitted work; KI reports grants and personal fees from Eisai, grants and personal fees from MSD, grants and personal fees from Takeda, grants and personal fees from Janssen, personal fees from Bristol Myers Squib, personal fees from Dainihon Sumitomo, grants and personal fees from Mundipharma, personal fees from Nihon Mediphysics, grants and personal fees from Chugai, grants and personal fees from Astra Zeneca, grants and personal fees from Abbvie, grants and personal fees from Bayer, grants and personal fees from Ono, grants from Gilead, grants from Zenyaku, grants and personal fees from Celgene, grants from Solasia, grants from Symbio, grants from Astellas, grants from Astellas Amgen, grants from Bayer and Daiichi Sankyo, personal fees from Kyowa Hakko Kirin, outside the submitted work; KT reports personal fees from Zenyaku Kogyo, grants and personal fees from Eisai, grants and personal fees from Takeda, grants and personal fees from Mundipharma, personal fees from HUYA Bioscience International, grants and personal fees from Kyowa Hakko Kirin, grants and personal fees from Celgene, grants and personal fees from Chugai Pharma, grants and personal fees from Ono Pharmaceutical, personal fees from Yakult , personal fees from Daiichi Sankyo, personal fees from Bristol-Myers Squibb, personal fees from Meiji Seika Kaisha, personal fees from Solasia Pharma, personal fees from Verastem , grants from Janssen, grants from Abbvie, outside the submitted work; YK reports grants, personal fees and other from Pfizer, personal fees from Astellas, personal fees from Symbio, outside the submitted work. Contributions: ToS collected the data. ToS, SF, YI, SH, SY, SM, WM, TaS, DM, KI, KT, and YK provided the cases. AM and HT performed the immunohistological evaluations. ToS, JN, SY, and TU performed the genetic analyses and interpreted the data. ToS and SF wrote the preliminary draft of the manuscript. All authors revised the subsequent drafts and approved the final version for submission. Acknowledgments: the authors would like to thank laboratory technicians MS, ST, MM and IM for performing FISH analysis and extra immunostaining for this study.

596

References 1. Vannata B, Conconi A, Winkler J, et al. Late relapse in patients with diffuse large B-cell lymphoma: impact of rituximab on their incidence and outcome. Br J Haematol. 2019;187(4):478-487. 2. Larouche JF, Berger F, Chassagne-Clement C, et al. Lymphoma recurrence 5 years or later following diffuse large B-cell lymphoma: clinical characteristics and outcome. J Clin Oncol. 2010;28(12):2094-2100. 3. Stephens DM, Li H, LeBlanc ML, et al. Continued risk of relapse independent of treatment modality in limited-stage diffuse large BCell Lymphoma: Final and Long-Term Analysis of Southwest Oncology Group Study S8736. J Clin Oncol. 2016;34(25):2997-3004. 4. Lenz G, Wright G, Dave SS et al. Stromal gene signatures in large-Bcell lymphomas. N Engl J Med. 2008;359(22):2313-2323. 5. Hans CP, Weisenburger DD, Greiner TC et al. Confirmation of the molecular classification of diffuse large B-cell lymphoma by immunohistochemistry using a tissue microarray. Blood. 2004; 103(1):275-282. 6. Chapuy B, Stewart C, Dunford AJ et al. Molecular subtypes of diffuse large B cell lymphoma are associated with distinct pathogenic mechanisms and outcomes. Nat Med. 2018;24(5):679-690. 7. Schmitz R, Wright GW, Huang DW et al. Genetics and pathogenesis of diffuse large B-cell lymphoma. N Engl J Med. 2018;378(15):13961407. 8. Juskevicius D, Lorber T, Gsponer J et al. Distinct genetic evolution patterns of relapsing diffuse large B-cell lymphoma revealed by genome-wide copy number aberration and targeted sequencing analysis. Leukemia. 2016;30(12):2385-2395. 9. Broseus J, Chen G, Hergalant S et al. Relapsed diffuse large B-cell lymphoma present different genomic profiles between early and late relapses. Oncotarget. 2016;7(51):83987-84002. 10. Wang Y, Farooq U, Link BK et al. Late relapses in patients with diffuse large B-cell lymphoma treated with immunochemotherapy. J Clin Oncol. 2019;37(21):1819-1827 11. de Jong D, Glas AM, Boerrigter L et al. Very late relapse in diffuse large B-cell lymphoma represents clonally related disease and is marked by germinal center cell features. Blood. 2003;102(1):324-327. 12. Horton SJ, Giotopoulos G, Yun H et al. Early loss of Crebbp confers malignant stem cell properties on lymphoid progenitors. Nat Cell Biol. 2017;19(9):1093-1104. 13. Ying CY, Dominguez-Sola D, Fabi M et al. MEF2B mutations lead to deregulated expression of the oncogene BCL6 in diffuse large B cell lymphoma. Nat Immunol. 2013;14(10):1084-1092. 14. Ortega-Molina A, Boss IW, Canela A et al. The histone lysine methyltransferase KMT2D sustains a gene expression program that represses B cell lymphoma development. Nat Med. 2015; 21(10):1199-1208.

haematologica | 2021; 106(2)

Letters to the Editor

Primary therapy and survival in patients over 70 years old with primary central nervous system lymphoma: a contemporary, nationwide, population-based study in the Netherlands Primary central nervous system lymphoma (PCNSL) is an uncommon, but aggressive non-Hodgkin lymphoma confined to the brain, leptomeninges, spinal cord, and eyes. Its incidence has increased substantially over the past decades among over 60-year-olds.1 The median age at diagnosis is around 65 years, and approximately onethird of newly diagnosed patients are >70 years.1 Nevertheless, elderly PCNSL patients, especially those over the age of 70 years, are frequently excluded from or under-represented in clinical trials due to concomitant comorbidities, poor performance status, or concerns regarding treatment-related sequelae.2,3 Prospective studies specifically designed for elderly PCNSL patients are scarce.4-6 Furthermore, the few available, somewhat outdated series mostly included relatively small numbers of patients (range, 10-107). These studies congruently showed that the prognosis of elderly patients remained poor and unchanged over the past decades, with overall survival (OS) ranging between 14-37 months. Collectively, apart from omitting consolidation radiotherapy after chemotherapy, the optimal treatment for elderly PCNSL patients is ill-defined.1,6,7 Population-based studies can complement prospective trials, especially in settings where data from prospective trials are scarce. At present, contemporary populationbased studies with detailed data regarding primary therapy specifically among >70-year old PCNSL patients to inform clinical practice are lacking. Therefore, in this contemporary, nationwide, population-based study, we assessed primary therapy and OS among >70 year old PCNSL patients diagnosed in the Netherlands. Established in 1989, the nationwide Netherlands Cancer Registry (NCR) has an overall coverage of >95% of all malignancies in the Netherlands.8 We identified all >70-year old PCNSL patients diagnosed confirmed with cytology, histology, and/or flow cytometry between January 1st 2014 and December 31st 2017 from the NCR. Diffuse large B-cell PCNSL was defined using the International Classification of Diseases for Oncology morphology and topography codes (Online Supplementary Methods). Two patients diagnosed at autopsy were excluded. We included patients diagnosed from 2014 because the NCR has collected data on the therapeutic regimen from that year onwards. The NCR is based on comprehensive case notifications through the Nationwide Network of Histopathology and Cytopathology, and the National Registry of Hospital Discharges (i.e., outpatient and inpatient discharges). Information on dates of birth and diagnosis, sex, disease stage, topography, and morphology, performance score, and primary therapy was available for individual patients. This information is collected by trained registrars of the NCR through retrospective medical records review. Primary therapy was categorized into chemotherapy, radiotherapy only, and supportive care only. Corticosteroids are not standardly registered in the NCR and may have been given in all treatment groups. The category of chemotherapy was broken down by the exact therapeutic regimen. The Privacy Review Board of the NCR approved the use of anonymous data for this study. The primary survival endpoint was OS, defined as the time from diagnosis until death. Patients were censored at emigration or end of follow-up (1st February 2019). OS haematologica | 2021; 106(2)

Table 1. Patients’ characteristics.

Characteristic Total n. of patients Sex Male Female Age at diagnosis, years Median (range) 71-74 75-79 ≥80 Performance score 0-1 2-4 Unknown Prior malignancy No Yes Vital status Alive Death Median follow-up, months (range) Overall Alive Death

(%)

145 73 72

(50) (50)

75 (71-87) 55 58 32

(38) (40) (22)

24 52 69

(17) (36) (48)

113 32

(78) (22)

27 118

(19) (81)

4.1 (0.0-60.0) 31.7 (15.2-60.0) 2.6 (0.0-41.4)

was calculated for three age groups (71-74, 75-79, and ≥80 years) and according to primary treatment (chemotherapy, radiotherapy only, and supportive care only) using the Kaplan-Meier method. Survival distributions were compared with the log-rank test. Multivariable Cox regression was conducted to assess covariates (sex, age at diagnosis, a prior malignancy before PCNSL diagnosis, receipt of rituximab, and type of primary therapy) associated with OS. P<0.05 was considered statistically significant. Further details about the statistical analyses are available in the Online Supplementary Appendix. A total of 145 PCNSL patients >70 years old (50% males) were included in the study. Median age was 75 years (range, 71-87), with 55 (38%), 58 (40%), and 32 (22%) patients aged 71-74, 75-79, and ≥80 years at diagnosis, respectively (Table 1). Overall, 43% of patients received chemotherapy, 20% radiotherapy only, and 37% supportive care only (Table 2). Numbers of patients receiving chemotherapy decreased with older age (58%, 40%, and 22% respectively across the three age groups), while radiotherapy only or supportive care only increased (P=0.002) (Table 2 and Online Supplementary Table S1). All 62 (43%) chemotherapy-treated patients except one were treated with either methotrexate (MTX) monotherapy (n=25) or a variety of MTX-based regimens (n=36) (Table 2). MTX with teniposide, carmustine, and prednisolone (MBVP) was the most commonly applied MTX-based regimen (25 of 36; 69%). Rituximab was added to chemotherapy in 17 of 62 (27%) patients (Table 2). Of note, 6 of 7 chemotherapy-treated patients aged ≥80 years were treated with MTX-monotherapy. During follow-up, 118 (81%) patients died. The medi597

Letters to the Editor

an follow-up of patients still alive was 31.7 months (range, 15.2-60.0). Overall, median OS was 4.1 months (95% confidence interval [CI]: 2.8-6.8) and 2-year OS was 25% (95%CI: 18%-32%) (Figure 1A). There were no significant differences in OS between the three age groups (P=0.185) (Figure 1B). The difference in OS between 71-74-year olds (7.7 months, 95%CI: 2.6-16.3) and those â&#x2030;Ľ75 years (3.9 months, 95%CI: 2.4-5.5) was also not statistically significant (P=0.08) (Online Supplementary Figure S1). OS according to primary treatment did show significant differences, with chemothera-

py-treated patients having a superior median OS (16.3 months 95%CI: 7.8-35.2) compared with those who received radiotherapy only (7.7 months, 95%CI: 4.613.2) or supportive care only (1.4 months, 95%CI: 1.11.7; P<0.001) (Figure 1C). Two-year OS was 45% (95%CI: 32-57%) in those receiving chemotherapy, whereas it was exceedingly low in the other two treatment groups (Figure 1C). Multivariable Cox regression analysis revealed that primary treatment was the only factor associated with OS, whereas sex, age, a prior malignancy before PCNSL diagnosis, and receiving ritux-

Figure 1. Overall survival (OS) among patients over 70 years of age with primary central nervous system lymphoma in the Netherlands: 2014-2017. OS is shown for the total cohort (A), and according to age at diagnosis (B), treatment group (C), and the type of therapy with methotrexate (MTX) (D). The tables below (B-D) show the median OS, and the projected 1- and 2-year OS with associated 95% confidence intervals. CI: confidence interval.

598

haematologica | 2021; 106(2)

Letters to the Editor

imab were not associated with OS (Online Supplementary Table S2). Excluding the four patients in the chemotherapy group who subsequently received whole-brain radiotherapy did not change survival estimates. Within the chemotherapy group, median OS for recipients of MTXmonotherapy was 5 months (95%CI: 2.6-41.4) and for MTX-based regimens 27 months (95%CI: 10.3-not reached) (Figure 1D) but was was not statistically significant (P=0.170). Also, and more importantly, the number of patients was too small to allow a meaningful comparison to be made. Therefore, a multivariable analysis of MTX only versus MTX-based regimens was not performed. In this contemporary, nationwide, population-based study among newly diagnosed PCNSL patients >70 years old we observed that the prognosis of these patients remains poor. This finding agrees with prior populationbased studies spanning the past decades.1 Age is a strong prognostic factor in adult PCNSL patients.9,10 However, within our study population of patients >70 years old, there was no clear prognostic gradient with increasing age, although with larger patient numbers, a statistically significant association between age and OS might be seen. Instead, despite the small patient numbers, treatment was a strong prognostic factor. Although only 22% of patients aged â&#x2030;Ľ80 years received chemotherapy (which possibly hints towards selection bias or confounding by indication) this finding suggests that treatment, more than age, influences survival in elderly patients judged fit enough to receive therapy. Selection bias might also have impacted MTXmonotherapy versus MTX-based chemotherapy; performance status and comorbidity, in particular renal insufficiency, might have influenced the choice of chemotherapeutic regimen. Prior prospective studies provided evidence that highdose MTX, especially when combined with alkylating chemotherapy, is the most efficacious treatment for elderly PCNSL patients.11 Although there are conflicting data

on the therapeutic value of chemoradiation over chemotherapy alone in elderly PCNSL patients,11,12 it is unquestionable that consolidation with radiotherapy in this population carries a high risk of neurotoxicity and severe cognitive decline.13 Controversy remains regarding the therapeutic value of rituximab in PCNSL. Findings from the current study and a recent randomized phase III trial among PCNSL patients aged 18-70 years showed no added therapeutic value of rituximab on survival outcomes.3 However, our results should be interpreted with caution given the low number of rituximab-treated patients. Similarly, a metaanalysis of 343 patients with PCNSL aged 50-67 years showed a possible positive effect of rituximab on PFS but not on OS.14 In contrast, a recent population-based study among 164 adult PCNSL patients diagnosed between 2005-2010 in Austria, of whom 40% were >70 years old, after a short follow-up (median 12 months) suggested that rituximab might increase survival.15 The strength of the current study is the use of a nationwide population-based cancer registry. As such, our findings are not compromised by selection and/or referral biases to the extent encountered in clinical trials. Therefore, our study represents the general population of elderly PCNSL patients. Limitations of our study mainly concern the lack of data throughout most of the registry on comorbidities, the use of corticosteroids, and the dose of steroids and chemotherapeutic agents, relapse rates, and salvage treatment. In addition, the performance score is poorly documented in medical records, thereby hampering its inclusion in the regression analyses due to the high percentage of unknown values (48%) (Table 1); this limits its contribution to the clinical decision-making process based on performance status. In summary, in this nationwide, population-based study, survival among PCNSL patients >70 years old remains poor in contemporary clinical practice. Nevertheless, our data demonstrate that MTX-based multi-agent chemotherapy, as compared with radiothera-

Table 2. Detailed information on primary therapy in over 70-year old patients with PCNSL.

Age at diagnosis, years 75-79

71-74 Primary therapy Total n. of patients Supportive care only Radiotherapy alone Chemotherapy MTX-based MBVPa,e MPb,f MCPMc MCPc R-CHOP + MTX MAc MTX onlyd Other PC

N 55 17 6 32 20 15 2 1 1 0 1 12 0 0

(%) (31) (11) (58) (36) (27) (4) (2) (2) (2) (22) -

N 58 22 13 23 15 10 3 1 0 1 0 7 1 1

(%) (38) (22) (40) (26) (17) (5) (2) (2) (12) (2) (2)

Total

â&#x2030;Ľ80 N 32 15 10 7 1 0 1 0 0 0 0 6 0 0

(%) (47) (31) (22) (3) (3) (19) -

N 145 54 29 62 36 25 6 2 1 1 1 25 1 1

(%) (37) (20) (43) (25) (17) (4) (1) (1) (1) (1) (17) (1) (1)

MTX: methotrexate; MBVP: MTX, teniposide, carmustine, and prednisolone; MP: methotrexate and procarbazine; MCPM: methotrexate, lomustine, procarbazine, and cytarabine; MCP: methotrexate, lomustine, and procarbazine; R-CHOP: rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone; MA: MTX and cytarabine; PC: procarbazine and lomustine. aCytarabine and rituximab were applied in 15 and 2 patients, respectively. bRituximab was applied in 5 patients. cRituximab was applied in 1 patient. dRituximab was applied in 6 patients. eWhole brain radiotherapy was administered after chemotherapy in 3 patients. fWhole brain radiotherapy was administered after chemotherapy in 1 patient.

haematologica | 2021; 106(2)

599

Letters to the Editor

py only and supportive care only, results in the best outcome in elderly patients judged eligible to receive such treatment, with a 2-year OS of approximately 50%. The challenge remains to balance the benefits and risks of intensive chemotherapy in this patient group. Therefore, future prospective intervention studies are needed to assess which elderly patients can benefit from intensive chemotherapy or less intensive approaches. Matthijs van der Meulen,1,2 Jacoline E.C. Bromberg,1 Marcel Nijland,3 Otto Visser,4 Jeanette K. Doorduijn5 and Avinash G. Dinmohamed5,6,7 1 Department of Neuro-Oncology, Erasmus MC Cancer Institute, University Medical Center Rotterdam, Rotterdam; 2Department of Neurology, Medical Spectrum Twente, Enschede; 3Department of Hematology, University Medical Center Groningen, Groningen; 4 Department of Registration, Netherlands Comprehensive Cancer Organisation (IKNL), Utrecht; 5Department of Hematology, Erasmus MC Cancer Institute, University Medical Center Rotterdam, Rotterdam; 6Department of Research and Development, Netherlands Comprehensive Cancer Organisation (IKNL), Utrecht and 7 Department of Public Health, Erasmus MC Cancer Institute, University Medical Center Rotterdam, Rotterdam, the Netherlands Correspondence: M. VAN DER MEULEN - m.vandermeulen.2@erasmusmc.nl doi:10.3324/haematol.2020.247536 Disclosures: no conflicts of interests to disclose. Contributions: JECB and AGD designed the study; AGD analyzed the data; OV collected the data; MvdM wrote the manuscript with contributions from JECB, MN, OV, JKD and AGD; and all authors interpreted the data, and read, commented on, and approved the final version of the manuscript. Acknowledgments: the authors would like to thank the registration clerks of the Netherlands Cancer Registry (NCR) for their dedicated data collection. The nationwide population-based NCR is maintained and hosted by the Netherlands Comprehensive Cancer Organisation (IKNL).

References 1. van der Meulen M, Dinmohamed AG, Visser O, Doorduijn JK, Bromberg JEC. Improved survival in primary central nervous system lymphoma up to age 70 only: a population-based study on incidence, primary treatment and survival in the Netherlands, 1989-

600

2015. Leukemia. 2017;31(8):1822-1825. 2. Ferreri AJM, Cwynarski K, Pulczynski E, et al. Whole-brain radiotherapy or autologous stem-cell transplantation as consolidation strategies after high-dose methotrexate-based chemoimmunotherapy in patients with primary CNS lymphoma: results of the second randomisation of the International Extranodal Lymphoma Study Group-32 phase 2 trial. Lancet Haematol. 2017;4(11):e510-e523. 3. Bromberg JEC, Issa S, Bakunina K, et al. Rituximab in patients with primary CNS lymphoma (HOVON 105/ALLG NHL 24): a randomised, open-label, phase 3 intergroup study. Lancet Oncol. 2019; 20(2):216-228. 4. Omuro A, Chinot O, Taillandier L, et al. Methotrexate and temozolomide versus methotrexate, procarbazine, vincristine, and cytarabine for primary CNS lymphoma in an elderly population: an intergroup ANOCEF-GOELAMS randomised phase 2 trial. Lancet Haematol. 2015;2(6):e251-e259. 5. Fritsch K, Kasenda B, Schorb E, et al. High-dose methotrexate-based immuno-chemotherapy for elderly primary CNS lymphoma patients (PRIMAIN study). Leukemia. 2017;31(4):846-852. 6. Siegal T, Bairey O. Primary CNS lymphoma in the elderly: the challenge. Acta Haematol. 2019;141(3):138-145. 7. Mendez JS, Ostrom QT, Gittleman H, et al. The elderly left behindchanges in survival trends of primary central nervous system lymphoma over the past 4 decades. Neuro Oncol. 2018;20(5):687-694. 8. Schouten LJ, Hoppener P, van den Brandt PA, Knottnerus JA, Jager JJ. Completeness of cancer registration in Limburg, The Netherlands. Int J Epidemiol. 1993;22(3):369-376. 9. Ferreri AJ, Blay JY, Reni M, et al. Prognostic scoring system for primary CNS lymphomas: the International Extranodal Lymphoma Study Group experience. J Clin Oncol. 2003;21(2):266-272. 10. Abrey LE, Ben-Porat L, Panageas KS, et al. Primary central nervous system lymphoma: the Memorial Sloan-Kettering Cancer Center prognostic model. J Clin Oncol. 2006;24(36):5711-5715. 11. Kasenda B, Ferreri AJ, Marturano E, et al. First-line treatment and outcome of elderly patients with primary central nervous system lymphoma (PCNSL)--a systematic review and individual patient data meta-analysis. Ann Oncol. 2015;26(7):1305-1313. 12. Korfel A, Thiel E, Martus P, et al. Randomized phase III study of whole-brain radiotherapy for primary CNS lymphoma. Neurology. 2015;84(12):1242-1248. 13. Thiel E, Korfel A, Martus P, et al. High-dose methotrexate with or without whole brain radiotherapy for primary CNS lymphoma (GPCNSL-SG-1): a phase 3, randomised, non-inferiority trial. Lancet Oncol. 2010;11(11):1036-1047. 14. Schmitt AM, Herbrand AK, Fox CP, et al. Rituximab in primary central nervous system lymphoma- a systematic review and metaanalysis. Hematol Oncol. 2019;37(5):548-557. 15. Neuhauser M, Roetzer T, Oberndorfer S, et al. Increasing use of immunotherapy and prolonged survival among younger patients with primary CNS lymphoma: a population-based study. Acta Oncol. 2019;58(7):967-976.

haematologica | 2021; 106(2)

Letters to the Editor

Heterogenous mutation spectrum and deregulated cellular pathways in aberrant plasma cells underline molecular pathology of light-chain amyloidosis Light-chain (AL) amyloidosis (ALA) is a rare but fatal monoclonal gammopathy (MG) causing organ and tissue damage resulting from the deposition of misfolded immunoglobulin free light chains in the form of amyloid fibrils.1 In some cases, ALA coexists with multiple myeloma (MM) (ALA+MM), which is the second most common blood cancer and is caused by the proliferation of clonal plasma cells (PC).2 Due to insufficient knowledge of ALA and ALA+MM biology, therapeutic options have mirrored treatment regimens of MM, which focus on the elimination of clonal PC.3,4 We investigated the mutation and gene expression profiles in clonal aberrant PC (aPC) in order to better understand ALA and ALA+MM etiology and to clarify the molecular differences between individual MG diagnoses. In order to address this, we analyzed 14 newly diagnosed (untreated) histologically proven ALA samples, 11 ALA+MM samples and 37 MM samples. All ALA+MM and MM samples manifested at least one myeloma-defining event. We isolated DNA and/or RNA from clonal bone marrow PC sorted using CD45-PB, CD38-FITC, CD19-PECy7 and CD56-PE fluorescent antibodies. Samples of frozen aPC from 12 ALA, 10 ALA+MM and 29 MM were subjected to whole genome amplification (WGA) using REPLI-g Mini Kit (Qiagen) employing multiple displacement amplification. Amplified DNA and DNA from peripheral blood (for exclusion of germline variants) served for exome library preparation and sequencing (median coverage 56x, Online Supplementary Table S3). RNA was transcribed from six ALA, four ALA+MM and eight MM newly diagnosed patients and hybridized on GeneChip Human Gene 1.0 ST Array. Detailed methods and patient’s characteristics are presented in the Online Supplementary Appendix and the Online Supplementary Tables S1, S8. Four ALA and three ALA+MM samples were used for both the exome and the transcriptome analysis in parallel. We evaluated only single-nucleotide variants (SNV) due to the potential bias in indels and copy number variations (CNV) introduced by the WGA method.5 Median numbers of non-synonymous exonic SNV that passed “effect” filters for ALA, ALA+MM, MM cohorts were 16.5, 20.5 and 23, respectively, and the mutation burden was similar across all three cohorts, median 1.36 SNV per megabase (Mb) (range: 0.28-5.86) (Figure 1A, Online Supplementary Table S2). The mutation burden did not significantly correlate with age in ALA or MM. For ALA+MM the test was not performed due to the lack of age information of some patients. The intra-sample analysis of clonality was performed in samples exceeding 30 unfiltred non-synonymous SNV and coverage ≥10X in copy number neutral regions. The results were available for 10 ALA, nine ALA+MM and 29 MM patients. Of those, only one clone was observed in one ALA, two ALA+MM, and three MM samples (Online Supplementary Table S1). MM samples in our analysis were composed of a higher number of subclones when compared to ALA and ALA+MM, though the difference was not statistically significant (Figure 1B). The median number of sublones for ALA and ALA+MM remained four subclones per sample, while the median for MM was five subclones per sample. The total pool of mutated genes consisted of 209 genes for ALA, 191 for ALA+MM and 682 for MM (Online haematologica | 2021; 106(2)

Supplementary Tables S4-S6). An overlap of gene sets among diagnoses is provided in Figure 1C. Heterogeneity of mutated gene profiles could be observed among all studied cohorts. Pairwise comparisons showed that only six mutated genes (FAT3, MUC3A, MUC6, PABPC3, RYR3, ZDHHC11) were present in at least one sample in all three diagnoses. These genes code for large proteins and possess a medium or high gene damage according to the gene damage index score, which points to their polymorphic nature in the normal population. Thus, variants in those genes are unlikely to cause disease,6 however, the role of some those genes in MM, e.g., FAT3, is still debated.7 We identified more genes shared between ALA+MM versus MM (25) than in ALA versus MM (14) or ALA versus ALA+MM.6 This suggests a slightly more similar mutation profile between ALA+MM and MM. From the list of all 209 ALA mutated genes, only 26 genes were shared with a previous sequencing project of Boyle et al. 20188 and four mutated genes were in common with the original study by Paiva et al. 20169 (Online Supplementary Table S7). Such low gene set overlaps are in line with the assumed mutational heterogeneity in ALA. Our datasets were not large enough to perform analysis of significantly mutated genes. Within the ALA, ALA+MM and MM cohort, genes mutated in more than one patient represented only 2, 6 and 47 genes (1%, 3% and 6.9%), respectively (Figure 1C). Such a marked heterogeneity was also detected in previous ALA exome studies.8,9 However, Boyle’s work reported that 16% of the genes in ALA were mutated more than once.8 The observed difference can be explained by the different approach for the separation of target populations of cells and by different variant calling algorithms. We performed comparison of all mutated genes with the 63 known MM drivers obtained from Walker et al. 2018.10 The results revealed that the average number of drivers per patient was lower in ALA and ALA+MM versus MM, even though the differences were not significant (Figure 1D). The total number of drivers present in the entire set of samples (ALA, ALA+MM and MM) was 28 (Figure 1E, Online Supplementary Table S2). The only shared mutated drivers among the diagnoses were NRAS found in ALA+MM and MM, and DIS3 present in ALA and MM. Interestingly, ALA and ALA+MM did not possess any SNV in the same driver gene (Figure 1E). Previously identified ALA drivers DIS3 and EP300 overlap with our ALA dataset and NRAS and TRAF3 overlap with our ALA+MM dataset8 (Online Supplementary Table S7). The most frequent functional driver categories were epigenetic regulators in ALA, NF-κB pathway in ALA+MM and MEK/ERK pathway in MM (Figure 1E). Interestingly, the NF-κB pathway was previously suggested to be one of the main affected pathways for ALA.8 Surprisingly, the differences at the mutational level did not manifest at the gene expression level. Our analysis did not reveal any differentially expressed genes between ALA and ALA+MM despite using several thresholds of significance. The expression analysis yielded 783 deregulated genes (837 probe sets) on the level of 0.05 P-value and fold change above 2 or below 0.5 in ALA or ALA+MM compared to MM (Online Supplementary Table S9-S10). Genes uniquely upregulated in ALA fall into the regulation of B-cell activation, phagocytosis or regulation of protein localization pathway gene ontology (GO) terms (Figure 2A, Online Supplementary Table S11), while genes that were exclusively downregulated in ALA belonged to the mitochondrial translation and ribosome biogenesis 601

Letters to the Editor

Figure 1. Profiling of somatic variants. (A) Distribution of numbers of filtered single nucleotide variants (SNV) per sample in light-chain amyloidosis (ALA), ALA plus multiple myeloma (ALA+MM) and MM. (B) Distribution of subclones in ALA, ALA+MM and MM. (C) Overlap of sets of mutated genes in ALA, ALA+MM and MM (large circles); numbers of genes mutated in more than two samples within ALA, ALA+MM and MM (small circles). (D) Distribution of MM drivers per sample in ALA, ALA+MM and MM. (E) Heat-map of occurrence of mutated drives in each cohort. Cohort sizes are given in brackets. The assignment of drivers into functional categories was performed according to Walker et al.8; ST kinase: serine/threonin kinase.

602

haematologica | 2021; 106(2)

Letters to the Editor

Figure 2. Gene Ontology (GO) analysis of genes differentially expressed in light-chain amyloidosis (ALA) and ALA plus multiple myeloma (ALA+MM) versus MM diagnoses and assigned into the category “biological process”. (A) The overlap of upregulated genes in ALA and ALA+MM. The top upregulated pathways are provided for the colored intersections. (B) The overlap of downregulated genes in ALA and ALA+MM. The top downregulated pathways are provided for the colored intersections. 1P=0.000002, 2P=0.000283, 3P=0.00078, 4P=0.000011, 5P=0.001867, 6P=0.04333 and 7P=0.02736.

GO terms (Figure 2B, Online Supplementary Table S11). The small number of deregulated genes between MM and ALA+MM only, did not allow for identification of pathways uniquely affected in these entities (Figure 2AB, Online Supplementary Table S11). Although our dataset was relatively small (18 samples), we detected a specific pattern of gene expression common in ALA and ALA+MM versus MM. Interestingly, the expression profiles and pathways specific for ALA/ALA+MM pointed to downregulation of genes involved in mitochondrial RNA metabolism and translation (Figure 2, Online Supplementary Table S11). Ribosomal deregulation in ALA was previously indicated by Kryukov et al.11 in 2016, however, our set of ribosomal proteins is involved in the function of mitochondrial rather than cytoplasmic ribosomes. We can speculate that downregulation of mitochondrial translation is a compensatory mechanism for increased stress. Downregulation of mitochondrial RNA to avoid oxidative stress was described by Crawford et al.12 in 1997 and oxidative stress as well as endoplasmtic reticulum stress were found to be elevated in ALA versus MM PC.13 One of the most important aims and biological questions of our study was to characterize the mutation and subclonal profile of ALA+MM and determine whether it is more similar to ALA or MM. In order to answer this question, we first defined similarities at the level of the somatic variants. More mutated genes were shared between ALA+MM and MM. On the other hand, there were more similarities in the number of subclones between ALA and ALA+MM compared to MM. This observation is supported by a clonality study using cytohaematologica | 2021; 106(2)

genetic methods.14 Results of Bochtler et al.14 demonstrated that all PC dyscrasias containing amyloid deposits share less clones compared to non-ALA counterparts. Furthermore, we compared the gene expression levels to gain additional insights into ALA+MM. Surprisingly, transcription profiles of ALA and ALA+MM were indistinguishable, while MM was a clearly separate entity, but still closer to ALA+MM than to ALA. Based on these results, we conclude that ALA+MM share a mutation profile which is more similar to MM, but these changes were not manifested on the gene expression level, or on the level of plasma cell infiltration. The typical myeloma symptoms present in ALA+MM may thus be caused by mechanisms other than the global expression profile of aPC. Our detailed study of ALA diseases represents an important step towards improved understanding of their genetic and transcriptomic background, which is a perquisite for development of optimal treatment strategies in the future. All sequencing reads mapped on the reference genome are deposited in European Genome-phenome Archive (EGA) with accession code EGAS00001004214. The gene expression data-matrix is available in the Online Supplementary Table S10. Zuzana Chyra,1,2,3* Tereza Sevcikova,1,2,4* Petr Vojta,1,2,5 Janka Puterova,6,7 Lucie Brozova,8 Katerina Growkova,1,2,4 Jana Filipova,1,2,4 Martina Zatopkova,1,2,4 Sebastian Grosicki,9 Agnieszka Barchnicka,9 Wieslaw Wiktor-Jedrzejczak,10 Anna Waszczuk-Gajda,10 Alexandra Jungova,11 Aneta Mikulasova,12 Marian Hajduch,5 Martin Mokrejs,13° 603

Letters to the Editor

Ludek Pour,14 Martin Stork,4 Lubica Harvanova,15 Martin Mistrik,15 Gabor Mikala,16 Pawel Robak,17 Anna Czyz,16 Jakub Debski,18 Lidia Usnarska-Zubkiewicz,18 Artur Jurczyszyn,19 Lukas Stejskal,20 Gareth Morgan,21 Fedor Kryukov,1,2°° Eva Budinska,6 Michal Simicek,1,2,4 Tomas Jelinek,1,2,4 Matous Hrdinka1,2,4 and Roman Hajek1,2 1 Faculty of Medicine, University of Ostrava, Ostrava, Czech Republic; 2Department of Hematooncology, University Hospital Ostrava, Ostrava, Czech Republic; 3LeBow Institute for Myeloma Therapeutics and Jerome Lipper Multiple Myeloma Center, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA; 4Department of Biology and Ecology, Faculty of Science, University of Ostrava, Ostrava, Czech Republic; 5Institute of Molecular and Translational Medicine, Faculty of Medicine and Dentistry, Palacky University in Olomouc, Olomouc, Czech Republic; 6RECETOX, Faculty of Science, Masaryk university in Brno, Brno, Czech Republic; 7Brno University of Technology, Faculty of Information Technology, Center of Excellence IT4Innovations, Brno, Czech Republic; 8Institute of Biostatistics and Analyses, Faculty of Medicine, Masaryk University, Brno, Czech Republic; 9Department of Hematology and Cancer Prevention, Facility of Health Sciences, Medical University of Silesia in Katowice, Poland; 10Department of Hematology, Oncology and Internal Diseases, Medical University of Warsaw, Warsaw, Poland; 11Charles University Hospital Pilsen, Pilsen, Czech Republic; 12Biosciences Institute, Newcastle University, Newcastle upon Tyne, UK; 13IT4Innovations, VŠB – Technical University of Ostrava, Ostrava, Czech Republic; 14Department of Internal Medicine, Hematology and Oncology, University Hospital Brno, Brno, Czech Republic; 15Department of Hematology and Transfusiology, Faculty of Medicine Comenius University, University Hospital Bratislava, Bratislava, Slovakia; 16Department of Hematology and Stem Cell Transplantation, South Pest Central Hospital, National Institute of Hematology and Infectious Diseases, Budapest, Hungary; 17Department of Hematology, Medical University of Lodz, Copernicus Memorial Hospital, Łódž, Poland; 18Department and Clinic of Hematology, Blood Neoplasms and Bone Marrow Transplantation, Wroclaw Medical University, Wroclaw, Poland; 19Jagiellonian University Medical College, Cracow, Poland; 20Silesian Hospital Opava, Opava, Czech Republic and 21Department of Medicine, Multiple Myeloma Research Perlmutter Cancer Center, NYU School of Medicine, New York, NY, USA *ZC and TS contributed equally as co-first authors °current address: Biomedical Center, Faculty of Medicine, Charles University of Pilsen, Pilsen & Institute of Experimental Medicine, Czech Academy of Sciences, Prague, Czech Republic °°current address: JSC “BIOCAD” Company, Saint-Petersburg, Russia Correspondence: ROMAN HAJEK - roman.hajek@osu.cz doi:10.3324/haematol.2019.239756 Disclosures: no conflicts of interests to disclose. Contributions: ZC and TS contributed equally to this manuscript , processed nucleic acid samples, designed experiments, prepared the figures and wrote the manuscript; PV, JP, LB, EB, AM and MMo performed bioinformatic and statistical analysis; KG, JF and MZ

604

performed cell sorting and nucleic acid isolation; TJ, LP, MS, SG, WWJ, AWG, AJ, AB, MH, LH, MMi, GM, PR, JD, AJ, JS and AC diagnosed patients, collected samples of peripheral blood and bone marrow and provided patient’s clinical data; GM provided the multiple myeloma exome data; FK designed the study and organized the initial sample collection and experiments; RH designed the study and managed the team; MS, MH and TJ designed experiments and wrote the manuscript. All authors have commented on and proofread the manuscript. Acknowledgments: we thank the anonymous reviewers for their valuable comments on the manuscript. We thank the Biobank in Brno and Ostrava for providing the archived samples of bone marrow and peripheral blood. We thank Mgr. Lucie Cerna for help with cell sorting and nucleic acid isolation. We are grateful to Shira Timilsina, MD for English language editing. Funding: this work was supported by the Ministry of Health of the Czech Republic (15-29667A).

References 1. Merlini G, Palladini G. Light chain amyloidosis: the heart of the problem. Haematologica. 2013;98(10):1492-1495. 2. Desikan KR, Dhodapkar MV, Hough A, et al. Incidence and impact of light chain associated (AL) amyloidosis on the prognosis of patients with multiple myeloma treated with autologous transplantation. Leuk Lymphoma. 1997;27(3–4):315-319. 3. Jelinek T, Kryukova E, Kufova Z, Kryukov F, Hajek R. Proteasome inhibitors in AL amyloidosis: focus on mechanism of action and clinical activity. Hematol Oncol. 2017;35(4):408-419. 4. Jelinek T, Kufova Z, Hajek R. Immunomodulatory drugs in AL amyloidosis. Crit Rev Oncol Hematol. 2016;99:249-260. 5. Babayan A, Alawi M, Gormley M, et al. Comparative study of whole genome amplification and next generation sequencing performance of single cancer cells. Oncotarget. 2016;8(34):56066-56080. 6. Itan Y, Shang L, Boisson B, Patin E, et al. The human gene damage index as a gene-level approach to prioritizing exome variants. Proc Natl Acad Sci U S A. 2015;112(44):13615-13620. 7. Bolli N, Biancon G, Moarii M, et al. Analysis of the genomic landscape of multiple myeloma highlights novel prognostic markers and disease subgroups. Leukemia. 2018;32(12):2604-2616. 8. Boyle EM, Ashby C, Wardell CP, et al. The genomic landscape of plasma cells in systemic light chain amyloidosis. Blood. 2018; 132(26):2775-2777. 9. Paiva B, Martinez-Lopez J, Corchete LA, et al. Phenotypic, transcriptomic, and genomic features of clonal plasma cells in light-chain amyloidosis. Blood. 2016;127(24):3035-3039. 10. Walker BA, Boyle EM, Wardell CP, et al. Mutational spectrum, copy number changes, and outcome: results of a sequencing study of patients with newly diagnosed myeloma. J Clin Oncol. 2015; 33(33):3911-3920. 11. Kryukov F, Kryukova E, Brozova L, et al. Does AL amyloidosis have a unique genomic profile? Gene expression profiling meta-analysis and literature overview. Gene. 2016;591(2):490-498. 12. Crawford DR, Wang Y, Schools GP, Kochheiser J, Davies KJ. Downregulation of mammalian mitochondrial RNAs during oxidative stress. Free Radic Biol Med. 1997;1;22(3):551-559. 13. Oliva L, Orfanelli U, Resnati M, et al. The amyloidogenic light chain is a stressor that sensitizes plasma cells to proteasome inhibitor toxicity. Blood. 2017;129(15):2132-2142. 14. Bochtler T, Merz M, Hielscher T, et al. Cytogenetic intraclonal heterogeneity of plasma cell dyscrasia in AL amyloidosis as compared with multiple myeloma. Blood Adv. 2018;2(20):2607-2618.

haematologica | 2021; 106(2)

Letters to the Editor

Selectively targeting FLT3-ITD mutants over FLT3wt by a novel inhibitor for acute myeloid leukemia Wild-type FLT3 (FLT3-wt) kinase is expressed in immature hematopoietic cells, placenta, gonads, and brain.1 It plays important roles in the differentiation and survival of hematopoietic stem cells in bone marrow.2 In a normal hematopoietic environment, FLT3 is predominantly expressed in CD34 positive cells and integrally involved in early hematopoiesis, reconstitution of multilineage myeloid precursors,3 and dendritic cell maturation.4,5 In acute myeloid leukemia (AML), the internal tandem duplications in the juxtamembrane domain of FLT3 kinase (FLT3-ITD), which displays heterogeneity in amino acid sequences in different patients, is the most prevalent mutation of FLT3 kinase and observed in approximately 30-40% of patients. A number of FLT3 kinase inhibitors have been investigated in clinical trials, such as gilteritinib,6 crenolanib,7 quizartinib8 and midostaurin,9 etc. However, most of the current FLT3 kinase inhibitors cannot distinguish the structurally similar cKIT kinase and FLT3-wt kinase, which might lead to myelosuppression toxicity.10 Here, we report the discovery of a novel FLT3-ITD mutant selective inhibitor, CHMFL-FLT3-362 (abbreviated as compound 362), which achieves high selectivity over both FLT3-wt and cKIT kinases. It also displays impressive in vitro and in vivo efficacies against the preclinical models of FLT3ITD+ AML. We first examined the activities of compound 362 against FLT3-ITD and FLT3-wt using the Z’-LYTE (Invitrogen) biochemical assay with purified FLT3 wt/ITD mutant proteins. The data showed that compound 362 (see Figure 1A for chemical structure) achieved over 30-fold selectivity between FLT3-ITD and FLT3-wt (Figure 1B). Kinetic study of the binding mode revealed that compound 362 was an ATP competitive inhibitor (Figure 1C). We then tested the anti-proliferative effects of compound 362 with a panel of engineered BaF3 cells which were transformed with different FLT3 wt/ITD mutants (Figure 1D and Online Supplementary Table S1). Interestingly, the compound displayed potent inhibitory activity against all ITD mutants with different ITD lengths ranging from 6 to 33 amino acids and achieved 7~30-fold selectivity over FLT3-wt. However, it was much less potent against drug resistant mutants of FLT3-ITD including FLT3-ITD-G697R/D835(del/I/V) /Y824(R/H), and primary gain-of-function mutations including FLT3-D835V/H/N/Y and FLT3-K663Q. All these data suggested that compound 362 was a FLT3ITD mutant selective inhibitor. As expected, this selectivity was recapitulated in leukemia cell lines evidenced by the selective inhibition against FLT3-ITD-dependent AML cells (MV4-11, MOLM-13, and MOLM-14) versus FLT3 wt-expressing cells (U937, CMK, OCI-AML-2, and HL-60) (Figure 1D and Online Supplementary Table S1). To further show the kinome-wide selectivity of compound 362, we examined it with the DiscoverX’s KINOMEscanTM technology at the concentration of 1 mM. The results demonstrated that compound 362 exhibited a good selectivity profile (S score 35 = 0.02). Besides FLT3, compound 362 also displayed strong binding against cKIT, CSF1R, FLT1, VEGFR2, PDGFRα and PDGFRβ kinases (Figure 1E and Online Supplementary Table S2). Since the KINOMEscanTM is a binding-based assay and may not really reflect the actual inhibitory activity of the kinase, we then examined compound 362 against these targets with the Z’-LYTE haematologica | 2021; 106(2)

biochemical assay. The results showed that it potently inhibited CSF1R, moderately inhibited PDGFRα and PDGFRβ, but much less potently inhibited cKIT, VEGFR2 and FLT1. This is not surprising since FLT3, CSF1R, and PDGFR kinases all belong to the type III receptor tyrosine kinase family and their ATP binding pockets are structurally highly similar. In addition, since cKIT kinase is off-target of most FLT3 kinase inhibitors and it is directly correlated with myelosuppression toxicity, we further tested compound 362 in the TEL-cKIT transformed BaF3 cells; the GI value was 5.6 mM which indicated a much weaker inhibition to cKIT kinase. This result was further verified in a more physiologically relevant M-07e cell line which overexpresses cKIT wt kinase.11 Again, compound 362 did not affect the phosphorylation of cKIT kinase in this cell line up to 1 mM (Online Supplementary Figure S1). We then investigated the effects of compound 362 on the signaling pathways mediated by FLT3 kinase. The results displayed that the FLT3 autophosphorylation in FLT3-ITD-dependent cell lines (MV4-11, MOLM-13, and MOLM-14) were significantly decreased and the phosphorylation of downstream signaling mediators including STAT5, AKT, and ERK were almost completely inhibited between 0.1 and 0.3 mM (Figure 2A). However, in FLT3-wt-expressing cell lines (OCI-AML-3, NOMO-1), FLT3 autophosphorylation and downstream signaling were not affected up to 3 mM of compound 362, while midostaurin could potently inhibit the FLT3 auto-phosphorylation (Online Supplementary Figure S2). These data further confirmed the selectivity of compound 362 between FLT3-ITD mutants and FLT3-wt. We next evaluated the apoptosis-related proteins in FLT3-ITD+ cell lines upon compound 362 treatment. Dose-dependent increase of the expression of cleaved caspase-3 and cleaved PARP were observed in these cell lines, which indicated apoptotic cell death (Figure 2B). In addition, dose-dependent cell cycle arrest was observed in FLT3-ITD positive cell lines but not FLT3wt cell lines (Online Supplementary Figures S3 and S4). Furthermore, compound 362 exhibited remarkable antiproliferative effects against FLT3-ITD positive AML patient primary cells but not FLT3-wt primary cells or peripheral blood mononuclear cells (PBMC) (Figure 2C and Online Supplementary Tables S3 and S4). Dosedependent inhibition of the phosphorylation of FLT3 and downstream signaling mediators STAT5, ERK and AKT in these primary cells confirmed the on-target inhibition effects of compound 362 (Figure 2C). We next evaluated the pharmacokinetics properties of compound 362 in rats, mice, and Beagle dogs following intravenous (1 mg/kg) and oral (5-10 mg/kg) administration (Online Supplementary Table S5). The data revealed a good drug-like PK profile including acceptable bioavailability (> 60%), suitable half-life (T1/2 = 1-4 hours [h]), and good Cmax (>3,000 ng/mL). We then used the zebrafish model10,12 to examine the safety of compound 362 by testing the effects at 72 h post fertilization (hpf) of embryos. The tail curvature caused by compound 362 was much less than that caused by midostaurin at 1 mM (Online Supplementary Figure S5), indicating a relatively lower toxicity of compound 362. In order to examine the myelosuppression effects, we further examined the effect of compound 362 on WT granulopoiesis using myeloperoxidase (mpx)+ positive cell numbers as a readout of cKIT-related myelopoiesis/myelosuppression. As expected, compound 362 showed no apparent difference of cell num50

605

Letters to the Editor

Figure 1. Characterization of CHMFL-FLT3-362 as a highly selective FLT3-ITD mutants kinase inhibitor. (A) Chemical structure of CHMFL-FLT3-362. (B) Invitrogen (Madison, WI, USA) Z’LYTE biochemical assay determination of the IC s of CHMFL-FLT3-362 against FLT3 wild-type (wt)/ITD kinases. Error bars, mean±standard error of mean (SEM), n=3. (C) Kinetic study of CHMFL-FLT3-362 against FLT3-wt/ITD kinase with varied concentrations (10 nM, 30 nM and 100 nM, respectively) of ATP. (D) Determination of the growth inhibition effects of CHMFL-FLT3-362 against a panel of 22 BaF3 engineered cell lines (including different ITD sequences and mutants), three FLT3-ITD+ human leukemia cell lines and three FLT3-wt human leukemia cell lines. Cells were treated with CHMFL-FLT3-362 (maximum concentration 10 mM) for 72 hours (h), and cell viability was measured using the Cell Titer-Glo assay. Error bars, mean±SEM, n=3. (E) (Left) Kinome-wide selectivity profiling of CHMFL-FLT3-362 with DiscoverX’s KINOMEscanTM technology (http://www.kinomescan.com). Red circles indicate kinases bound, and circle size indicates relative binding affinity compared to DMSO (Ctrl%). The complete dataset is shown in Online Supplementary Table S2. (Right) Inhibition effects of CHMFLFLT3-362 against FLT3, FLT3/ITD, KIT, PDGFRα, PDGFRβ, CSF1R, FLT1 and VEGFR2. In vitro kinase activity was measured by Invitrogen Z’-LYTE assay (IC : mM, mean±SEM, n=3) and engineered BaF3 cell lines (GI : mM, mean±SEM, n=3). 50

606

haematologica | 2021; 106(2)

Letters to the Editor

Figure 2. Characterization of CHMFL-FLT3-362 in cell and in vivo in preclinical acute myeloid leukemia (AML) models with FLT3-ITD mutants. (A) The phosphorylation levels of FLT3 (Tyr 589/591), STAT5 (Tyr 694), ERK1/2 (Tyr 202/204), and AKT (Ser 308) were detected by western blot in MOLM13, MOLM14 and MV411 cell lines. These cells were incubated with the indicated concentrations of CHMFL-FLT3-362 for 4 hours (h) before lysis. (B) Western blot analysis for the expression of PARP, cleaved PARP, caspase-3, and cleaved caspase-3 in MOLM-13, MOLM-14, and MV4-11 cells treated with CHMFL-FLT3-362 for 24 h. (C) Growth inhibition effects of CHMFL-FLT3-362 against FLT3-ITD+ AML patient primary cells, FLT3-wild-type (wt) AML patient primary cells, and peripheral blood mononuclear cells using cell Titer-Glo assay after 72 h drug treatment (maximum concentration 10 mM, mean±SEM, n=3). (D) Effects of CHMFL-FLT3-362 on zebrafish normal myelopoiesis and the FLT3/ITD-induced myeloid cells expansion. WT granulopoiesis were measured by the number of myeloperoxidase positive cells using in situ hybridization assay with mpx specific RNA probe. Inject-the injection of FLT3-ITD gene mRNA during the fertilized egg period, uninject-the FLT3ITD gene mRNA was not injected during the fertilized egg period; DMSO uninject vs. CHMFL-FLT3-362 uninject, P=0.4165; DMSO uninject vs. DMSO inject, P=0.0067; CHMFL-FLT3-362 uninject vs. CHMFL-FLT3-362 inject, P=0.4385; DMSO inject vs. CHFML-FLT3-362 inject, P=0.0039, **P<0.01, ns: not significant. (E) Bone marrow (BM) smear of mouse 14-day sub-drug test. (F) Anti-leukemia effects of CHMFL-FLT3-362 with once daily (QD) dosing at 50, 100, and 150 mg/kg. Total study length was 28 days. Each treatment group=5 animals. (Left) Representative graphs of relative tumor size are shown. (Right) Representative graphs of tumor weight are shown of different groups. Error bars, mean±SEM, n=5. *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001. (G) Anti-tumor efficacy of CHMFL-FLT3-362 in a BM engrafted mouse model. Kaplan-Meier plots of survival. (Left) The disseminated NOD/SCID mice were intravenously inoculated with MV4-11 cells. Mice received daily oral administration of CHMFL-FLT3-362 dosing at 50, 100, and 150 mg/kg. Total study length was 174 days. Each treatment group = 5 animals. (Right) Disseminated NOD/SCID mice were intravenously inoculated with MOLM-13 cells. Mice received daily oral administration of CHMFLFLT3-362 dosing at 50, 100, and 150 mg/kg. Total study length was 32 days. Each treatment group=5 animals.

haematologica | 2021; 106(2)

607

Letters to the Editor ber at 1 mM after treatments at 30 hpf (DMSO Uninject vs. CHMFL-FLT3-362 Uninject) (Figure 2D), indicating no apparent myelosuppression toxicity. In FLT3-ITDtransduced embryos, compound 362 effectively rescued the abnormal proliferation of mpx+ myeloid cells caused by overexpression of the FLT3-ITD gene (DMSO Inject vs. CHFML-FLT3-362 Inject) (Figure 2D) and inhibited the spread of FLT3-ITD-injected leukemic blasts at 1 mM (Online Supplementary Figure S6) indicating that compound 362 exerted its effects through FLT3-ITD on-target inhibition. In addition, oral administration of 300, 600, and 1,200 mg/kg/day dosages for 14 days did not result in significant toxicity and weight loss in the mice (Online Supplementary Figure S7). Furthermore, bone marrow (BM) smear analysis also showed that compound 362 had no effect on the proliferation and activity of mouse BM cells (Figure 2E). To examine the inhibitory effects of compound 362 on tumor growth, different dosages of compound 362 were administered orally every day for 28 days in the subcutaneous MV4-11 cells xenograft mice model. It displayed dose-dependent anti-tumor efficacy and achieved the tumor growth inhibition of 95% at a dosage of 150 mg/kg/day (Figure 2F). No weight loss or any other obvious signs of toxicity was observed. Immunohistochemistrystaining of the tumor tissue also confirmed that the cell proliferation was inhibited (Ki67 staining) and the apoptosis was induced (TUNEL staining) (Online Supplementary Figure S8A-C). We then further confirmed the anti-tumor efficacy of compound 362 in an orthotopic model of BM engraftment using MV4-11 and MOLM-13 cells, which physiologically differs from the subcutaneous MV4-11 xenograft model. Compound 362 dose-dependently extended the survival of mice at 50, 100, and 150 mg/kg/day dosages with no apparent weight loss at all dosages (Figure 2G and Online Supplementary Figure S8D and E). Flow cytometry analysis revealed significant reduction of the MV4-11 cells in the BM in this in vivo model (Online Supplementary Figure S9). In this study, we describe a novel FLT3-ITD mutant selective inhibitor CHMFL-FLT3-362, which achieved 30-fold selectivity between FLT3-ITD mutants and FLT3-wt in biochemical assays and 10-fold selectivity in cellular context. Considering that FLT3-wt is essential for the proliferation of normal primitive hematopoietic cells, this selectivity indicates that it might provide better safety profiles. In addition, compound 362 was potent against different ITD mutants which are more relevant to the clinically observed heterogenicity. Furthermore, it also achieved great selectivity over cKIT kinase which would help to avoid the myeloid suppression toxicity due to the FLT3/cKIT dual inhibition. The unique selectivity profile combined with acceptable in vivo PK/PD properties in the preclinical models makes compound 362 a valuable research tool for FLT3 mediated pathological study as well as a novel potential antiFLT3-ITD+ AML drug candidate. Aoli Wang,1,2* Chen Hu,1,2* Cheng Chen,1,3* Xiaofei Liang,1,2* Beilei Wang,1,2* Fengming Zou,1,2 Kailin Yu,1 Feng Li,1,3 Qingwang Liu,2,4 Ziping Qi,1,2 Junjie Wang,1,3 Wenliang Wang,1,3 Li Wang,1,2 Ellen L. Weisberg,5 Wenchao Wang,1,2,3 Lili Li,6 Jian Ge,6 Ruixiang Xia,6 Jing Liu1,2,3# and Qingsong Liu1,2,3,4,7# 1 High Magnetic Field Laboratory, Key Laboratory of High Magnetic Field and Ion Beam Physical Biology, Hefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei, Anhui, P.R. China; 608

Precision Medicine Research Laboratory of Anhui Province, Hefei, Anhui, P.R. China; 3University of Science and Technology of China, Hefei, Anhui, P.R. China; 4Precision Targeted Therapy Discovery Center, Institute of Technology Innovation, Hefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei, Anhui, P.R. China; 5Department of Medical Oncology, Dana Farber Cancer Institute, Harvard Medical School, Boston, MA, USA and 6Department of Hematology, The First Affiliated Hospital of Anhui Medical University, Hefei, Anhui, P.R. China and 7Institute of Physical Science and Information Technology, Anhui University, Hefei, Anhui, P.R. China *AW, CH, CC, XL and BW contributed equally as co-first authors. # Jing Liu and Qingsong Liu contributed equally as co-senior authors. Correspondence: QINGSONG LIU - qsliu97@hmfl.ac.cn JING LIU - jingliu@hmfl.ac.cn doi:10.3324/haematol.2019.244186 Disclosures: no conflicts of interests to disclose. Contributions: ALW, CH, CC, XFL, and BLW designed and performed the experiments; FMZ, KLY, FL, QWL, ZPQ, JJW, WWW, and LW aquired the data; ELW provided materials and reviewed the data. WCW reviewed the data and revised the manuscript. LLL, JG, and RXX collected the patient samples. JL and QSL designed the project and wrote the manuscript. Acknowledgments: we are also grateful for the Youth Innovation Promotion Association of CAS support (n. 2016385) for XL and the support of Hefei leading talent for FZ. Part of this work was supported by the High Magnetic Field Laboratory of Anhui Province. Funding: this work was supported by the National Natural Science Foundation of China (grant ns. 81773777, 81803366, 81872745, 81872748, 81673469), the “Personalized MedicinesMolecular Signature-Based Drug Discovery and Development”, Strategic Priority Research Program of the Chinese Academy of Sciences (grant n. XDA12020114), the National Science & Technology Major Project “Key New Drug Creation and Manufacturing Program” of China (grant n. 2018ZX09711002), the China Postdoctoral Science Foundation (grant ns. 2018T110634, 2018M630720, 2019M652057), the Postdoctoral Science Foundation of Anhui Province (grant ns. 2018B279, 2019B300, 2019B326), the Frontier Science Key Research Program of CAS (grant n. QYZDB-SSW-SLH037), the CASHIPS Director's Fund (grant n. BJPY2019A03), and the Key Program of 13th five-year plan of CASHIPS (grant n. KP-2017-26).

References 1. deLapeyriere O, Naquet P, Planche J, et al. Expression of Flt3 tyrosine kinase receptor gene in mouse hematopoietic and nervous tissues. Differentiation. 1995;58(5):351-359. 2. Markovic A, MacKenzie KL, Lock RB. FLT-3: a new focus in the understanding of acute leukemia. Int J Biochem Cell Biol. 2005;37(6)1168-1172. 3. Mackarehtschian K, Hardin JD, Moore KA, et al. Targeted disruption of the flk2/flt3 gene leads to deficiencies in primitive hematopoietic progenitors. Immunity.1995;3(1):147-161. 4. Maraskovsky E, Brasel K, Teepe M, et al. Dramatic increase in the numbers of functionally mature dendritic cells in Flt3 ligandtreated mice: multiple dendritic cell subpopulations identified. J Exp Med. 1996;184(5):1953-1962. 5. McKenna HJ, Stocking KL, Miller RE, et al. Mice lacking flt3 ligand have deficient hematopoiesis affecting hematopoietic progenitor cells, dendritic cells, and natural killer cells. Blood. 2000;95(11):3489-3497. 6. Lee LY, Hernandez D, Rajkhowa T, et al. Preclinical studies of gilteritinib, a next-generation FLT3 inhibitor. Blood. 2017;129(2): 257-260. 7. Smith CC, Lasater EA, Lin KC, et al. Crenolanib is a selective

haematologica | 2021; 106(2)

Letters to the Editor

type I pan-FLT3 inhibitor. Proc Nat Acad Sci U S A. 2014;111(14):5319-5324. 8. Zarrinkar PP, Gunawardane RN, Cramer MD, et al. AC220 is a uniquely potent and selective inhibitor of FLT3 for the treatment of acute myeloid leukemia (AML). Blood. 2009;114(14):29842992. 9. Weisberg E, Boulton C, Kelly LM, et al. Inhibition of mutant FLT3 receptors in leukemia cells by the small molecule tyrosine kinase inhibitor PKC412. Cancer Cell. 2002;1(5):433-443. 10. Warkentin AA, Lopez MS, Lasater EA, et al. Overcoming myelo-

haematologica | 2021; 106(2)

suppression due to synthetic lethal toxicity for FLT3-targeted acute myeloid leukemia therapy. Elife. 2014;3. 11. Smolich BD, Yuen HA, West KA, et al. The antiangiogenic protein kinase inhibitors SU5416 and SU6668 inhibit the SCF receptor (c-kit) in a human myeloid leukemia cell line and in acute myeloid leukemia blasts. Blood, 2001;97(5):1413-1421. 12. He BL, Shi X, Man CH, et al. Functions of flt3 in zebrafish hematopoiesis and its relevance to human acute myeloid leukemia. Blood, 2014;123(16):2518-2529.

609

Letters to the Editor

miR-939 acts as tumor suppressor by modulating JUNB transcriptional activity in pediatric anaplastic large cell lymphoma Anaplastic large cell lymphoma (ALCL) is a distinct subtype of aggressive non-Hodgkin lymphoma, mainly affecting children and young adults. The vast majority of the pediatric cases are anaplastic lymphoma kinase-positive (ALK+) due to the chromosomal translocation t(2;5)(p23;q35), which leads to the constitutive expression and activation of NPM-ALK fusion protein.1 ALK downstream signaling includes multiple oncogenic pathways and has been directly implicated in ALCL pathogenesis.2 Current therapeutic approaches can cure 70% of children affected by ALCL. However, despite this, the

prognosis of patients experiencing failure of first-line treatment is very poor.3 The use of biological risk factors such as minimal disseminated disease (MDD), minimal residual disease (MRD), and anti-ALK antibody titer in the forthcoming ALCL trials aims to reduce toxicity for low-risk patients, while making it easier to identify those at higher risk of relapse, who might benefit from more intense regimes and new targeted therapies.4,5 We recently identified two clearly distinct subgroups of ALCL patients who were characterized by different endogenous NPM-ALK expression levels at diagnosis and were defined ALK-high and ALK-low.6 These patients displayed a bimodal distribution for NPM-ALK expression and the median expression values were 2,500 NPM-ALK copies/10,000 ABL1 in ALK-low cases and 125,000 NPM-

Figure 1. miR-939 is differentially expressed according to anaplastic lymphoma kinase (ALK) expression in tumor tissues. Box plots of differential miR-939 expression in anaplastic large cell lymphoma (ALCL) tumor biopsies detected by (A) human microRNA microarray v.2 (n=16: 11 ALK-high and 5 ALK-low; **P=0.0016) and (B) quantitative reverse transcription PCR (qRT-PCR) (n=49: 39 ALK-high and 10 ALK-low; **P=0.0017). qRT-PCR data have been calculated according to the comparative Î´Ct method, using miR-16 as endogenous control. The horizontal line in the box indicates the median expression level of miR-939 (ALK-low median value: 3.45; ALK-high median value: 2.62). (C) Heterogeneous expression levels of miR-939 in 49 ALCL cases. Data have been measured by qRT-PCR according to the comparative Î´Ct method and compared to reactive lymph nodes tissue (LN, n=11). Blue and red bars are ALK-high and ALK-low tumor biopsies, respectively. miR-939 median expression level=2.83.

610

haematologica | 2021; 106(2)

Letters to the Editor

ALK copies/10,000 ABL1 in ALK-high patients. Interestingly, most of the patients who relapsed belonged to the ALK-high subgroup. The median time of relapse of these ALK-high patients was 5 months, in striking contrast to the time of relapse of ALK-low cases (30 months). By means of a gene set enrichment analysis and in vitro functional studies, we showed a differential gene expression signature in ALK-high versus -low patients. Moreover, we identified signaling pathways preferentially associated to the different phenotype and clinical course of ALK-high and ALK-low cases. In particular, the less aggressive phenotype observed in the ALK-low group was characterized by IL-2 signaling retention, while higher ALK expression levels were shown to induce cyclins and Aurora Kinases overexpression in the ALK-high subgroup. These data also suggest that cells expressing high ALK levels could acquire a selective growth advantage as compared to ALK-low cells.6 To further elucidate the molecular mechanisms responsible for the heterogeneous aggressiveness of ALK+ ALCL, and to better characterize these two previously uncovered disease entities, we performed array-based miRNA profiling in 16 pediatric cases previously classified according to their ALK endogenous expression (n=11 ALK-high, n=5 ALK-low) (Online Supplementary Table S1). By supervised analysis, we identified 19 microRNAs significantly upregulated in the ALK-low subgroup (False Discovery Rate <0.05) (Online Supplementary Table S2).

Some of these, such as miR-155, miR-146a and miR-497, had previously been described as differentially expressed in ALK+ compared to ALK-ALCL.7 Since high ALK expression was previously associated with a more aggressive ALCL phenotype,6 miRNAs upregulated in the ALK-low cases could probably act as tumor suppressors. In line with this concept, HoareauAveilla et al. found that the overexpression of miR-497 in ALK+ cell lines (corresponding to ALK-high patients) inhibited ALCL cell growth and induced cell cycle arrest.8 In this study, we focused our attention on miR-939, that was the most differentially expressed miRNA between ALK-high and ALK-low tumor cells (Figure 1A). We first evaluated miR-939 expression by quantitative reverse transcription PCR (qRT-PCR) in an extended panel of 49 ALCL primary tumor samples collected in the last 15 years in our lab from the Italian Association of Pediatric Hematology and Oncology (Associazione Italiana di Ematologia e Oncologia Pediatrica, AIEOP) centers. Unfortunately, despite a long period of recruitment, this larger cohort included only 20% of ALK-low cases, and miR-939 expression difference, observed with the array-based miRNA profiling, was less appreciable. Of note, three ALK-low cases had very high levels of miR-939 (Figure 1B). Although whether the expression of miR-939 is actually higher in ALK-low profile patients could only be confirmed by an analysis of a very large cohort of ALCL pediatric cases, it does seem evident that

Figure 2. miR-939 impairs tumor cell migration and invasion. (A) miR-939 basal expression levels in anaplastic large cell lymphoma (ALCL) cell lines. Data have been calculated according to the comparative δCt method and compared to reactive lymph nodes tissue (LN, n=11). miR-16 was used as endogenous control. Karpas299 and SUP-M2 cells were transiently transfected with miR-939 mimic, or negative control (NC) or with the inhibitor and used to analyze migration (B) and clonogenicity (C). The results are shown as fold change of negative control values. Two or three independent experiments were performed in triplicate and mean results are shown. *P<0.05; **P<0.001. (D) Western blotting analysis of JUNB and JUN proteins in Karpas299 and SUP-M2 cell lines transfected with miR-939 mimic or negative control at indicated post-transfection time points. γ-tubulin was used as loading control. Vertical lines have been inserted to indicate repositioned gel lanes. (E) The JUNB 3’UTR sequence containing the miR-939 putative binding site has been subcloned into pmirGLO reporter plasmid. Karpas299 and SUP-M2 cells transfected with miR-939 mimic and the reporter plasmid containing wild-type or mutated miR-939 binding site were tested by luciferase assay 24 hours post transfection. **P<0.001; ****P<0.0001; ns: non-significant. (F) miR-939 binding site on 3’-UTR of JUNB. The putative binding sequences have been predicted using the on-line tool available at http://www.microrna.org.

haematologica | 2021; 106(2)

611

Letters to the Editor

ALCL tumors express different levels of miR-939 (Figure 1C and Online Supplementary Table S3). To decipher the functional role of miR-939 in ALCL, we used Karpas299 and SUP-M2 cells, characterized by different basal levels of the miRNA (Figure 2A). We transiently overexpressed miR-939 or inhibited its expression in these cell lines (Online Supplementary Figure S1) that are more similar to ALK-high ALCL cells than to the ALKlow counterpart.6 The overexpression of miR-939 significantly reduced the migratory ability of both ALCL cell lines as compared to negative control cells, while its inhibition was shown to significantly increase ALCL cell migration (Figure 2B). Moreover, miR-939 overexpression reduced both Karpas299 and SUP-M2 clonogenic growth capacity (Figure 2C). In contrast, its inhibition significantly increased Karpas299 clonogenicity, while a less noticeable effect was observed on SUP-M2, which might be in line with the lower basal level of miR-939 in SUP-M2 compared to Karpas299 cells (Figure 2A). Of note, NPMALK expression was not affected by miR-939 overexpression, thus suggesting that miR-939 does not contribute mechanistically to the ALK-high or ALK-low status (Online Supplementary Figure S2). In silico prediction of miR-939 targets and Gene Ontology analysis were performed by using miRTarBase 7.0 (http://mirtarbase.mbc.nctu.edu.tw/php/index.php) and Panther 14.1 (http://pantherdb.org/data/) online tools. Among transcription factors and nucleic acid binding proteins, which were highly represented in the miR-939 putative target list, we focused on the transcriptional factor JUNB as the most promising candidate for our analy-

ses, since its knockdown in NPM-ALK expressing cells was previously shown to impair cell proliferation.9 Indeed, transient overexpression of miR-939 decreased JUNB protein expression both in Karpas299 and SUP-M2 cells (Figure 2D). Moreover, direct modulation of JUNB by miR-939 was confirmed by 3’UTR reporter assay in both Karpas299 and SUP-M2 cell lines transfected with miR-939 or negative control and reporter vectors containing wild-type and mutated miR-939 binding site in JUNB 3’-UTR sequence (Figure 2E). The miR-939 recognition site was identified by using the online tools at http://www.microrna.org (Figure 2F). JUNB is an activator protein-1 (AP-1) transcription factor, together with JUN, JUND and also members of the Fos/Fra, ATF and Maf subfamilies. AP-1 proteins exert their activity by forming homo- and hetero-complexes, and mainly regulate the expression of genes involved in cell proliferation.10 JUNB overexpression was shown in CD30+ lymphomas and its transcriptional activity was directly implicated in CD30 promoter induction in both Hodgkin lymphoma and ALCL.11 NPM-ALK itself contributes to JUNB activation and CD30 expression in ALK+ALCL via both ERK1/2-MAPK and mTOR pathways.9,12 JUNB and JUN activities were also shown to sustain PI3K-associated activation of AKT1, further suggesting the existence of a synergistic crosstalk between NPM-ALK signaling and AP-1 transcription factors.13 More recently, platelet-derived growth factor receptor B (PDGFRB) was confirmed as a JUNB transcriptional target in NPM-ALK-driven lymphomas and the combination of ALK and PDGFR

Figure 3. PDGFRB expression according to miR-939 levels in anaplastic lymphoma kinase-positive (ALK+) anaplastic large cell lymphoma (ALCL) cell lines and ALCL patients’ samples. (A) PDGFRB protein immunoprecipitation in Karpas299 and SUP-M2 cell lines transfected with miR-939 mimic or negative control at indicated post-transfection time points; western blotting analysis of γ-tubulin was used as immunoprecipitation input control. Blk: blank. (B) Box plot of PDGFRB mRNA expression in patients expressing high or low levels of miR-939 (n=25 miR-939_low [≤ median value]; n=24 miR-939_high [> median value]) detected by quantitative reverse transcription polymerase chain reaction (qRT-PCR). *P=0.03. Data have been calculated according to the comparative dCt method and compared to reactive lymph node tissues (LN, n=11). ABL1 was used as endogenous control. (C) Dot plot representing PDGFRB protein expression assessed by RPPA in ten available ALCL tumor tissue samples (5 miR-939_low, 5 miR-939_high; *P=0.01). (D) Schematic representation of miR-939-mediated JUNB modulation in ALK-low/-high ALCL. In the presence of lower levels of NPM-ALK transcript, miR-939 is able to inhibit JUNB transcriptional activity, resulting in PDGFRB downregulation (black arrows). Gray arrows are referred to previous data from the literature.12

612

haematologica | 2021; 106(2)

Letters to the Editor

inhibitors was suggested as a valuable strategy to reduce relapse rates in ALCL.14 In the presence of lower NPM-ALK levels, we hypothesized that miR-939 overexpression could contribute to fine-tuning the JUNB-mediated oncogenic signaling, further explaining the favorable outcome of ALK-low ALCL cases. To assess if miR-939 was able to modulate PDGFRB expression through JUNB modulation, we evaluated PDGFRB expression in Karpas299 and SUP-M2 cells after transfection with miR-939 mimic or negative control. PDGFRB was reduced in both the cell lines after 48 hours of transfection with miR-939 mimic (Figure 3A and Online Supplementary Figure S3). Even though JUN is also necessary to regulate PDGFRB transcription in ALCL,14 our results suggest that miR-939 modulates PDGFRB expression by specifically targeting JUNB, since JUN protein was not affected by miR-939 overexpression (Figure 2D). While JUNB mRNA did not show any significant difference between ALK-high and ALK-low groups (Online Supplementary Figure S4), PDGFRB transcript levels were significantly reduced in miR-939_high compared to miR-939_low ALCL cases (Figure 3B). By performing reverse phase protein arrays on ten ALCL tumor tissue samples, PDGFRB protein was also found to be significantly downregulated in miR-939_high cases (Figure 3C), suggesting that miR-939 could act as an oncosuppressor by reducing PDGFRB expression. In conclusion, we previously showed that tumors with high or low NPM-ALK levels are characterized by a different gene expression profile and a different aggressiveness. In particular, when expressed at lower levels, ALK is unable to render the transformed lymphocytes completely unresponsive to pathways normally expressed in T cells, such as those involved in interleukin signaling.6 Here we demonstrated that miR-939 expression could contribute to PDGFRB inhibition, a crucial driver for ALCL lymphomagenesis, via JUNB downregulation (Figure 3D). Further investigations will clarify if miR-939 expression could affect anti-CD30 therapies or imatinib treatment efficacy in ALK-low ALCL patients. Anna Garbin,1,2 Federica Lovisa,1,2 Antony B. Holmes,3 Carlotta C. Damanti,1,2 Ilaria Gallingani,1,2 Elisa Carraro,1 Benedetta Accordi,1,2 Giulia Veltri,1,2 Marco Pizzi,4 Emanuele S.G. d’Amore,5 Marta Pillon,1 Alessandra Biffi,1,2,6 Katia Basso,3,7 and Lara Mussolin1,2 1 Division of Pediatric Hematology, Oncology and Stem Cell Transplant, Department of Women and Child Health, University of Padua, Padua, Italy; 2Istituto di Ricerca Pediatrica Città della Speranza, Padova, Italy; 3Institute for Cancer Genetics, Columbia University, New York, NY, USA; 4Surgical Pathology and Cytopathology Unit, Department of Medicine, University of Padova, Padova, Italy; 5Department of Pathological Anatomy, San Bortolo Hospital, Vicenza, Italy; 6Gene Therapy Program, Dana Farber/Boston Children’s Cancer and Blood Disorders Center, Boston, MA, USA and 7Department of Pathology and Cell Biology, Columbia University, New York, NY, USA Correspondence: LARA MUSSOLIN - lara.mussolin@unipd.it doi:10.3324/haematol.2019.241307 Disclosures: no conflicts of interests to disclose. Contributions: AG performed the experiments, analyzed results, prepared figures and wrote the manuscript; FL analyzed results and wrote the manuscript; ABH performed microarray analyses and revised

haematologica | 2021; 106(2)

the manuscript; CCD and IG processed clinical samples and performed qRT-PCR analyses; EC and MP collected clinical data, performed statistical analyses and revised the manuscript; BA and GV performed Reverse Phase Protein Array experiments and commented on the manuscript; MP and ESGdA performed hemato-pathological revision and commented on the manuscript; AB revised the manuscript; KB supervised microarray analyses and revised the manuscript; LM conceived and designed research, supervised experimental work and revised the manuscript. All the authors read and approved the final version of the manuscript. Funding: this work was supported by Fondazione Città della Speranza, Padova, Italy; Fondazione CA.RI.PA.RO, Padova, Italy (grant 17/03 to LM); Camera di Commercio Venezia, Venezia, Italy; Fondazione Roche, Roma, Italy (grant to FL) and Fondazione Umberto Veronesi, Milano, Italy (fellowship to FL); Associazione Italiana contro le Leucemie, sezione provinciale di Treviso (grant to BA).

References 1. Morris SW, Kirstein MN, Valentine MB, et al. Fusion of a kinase gene, ALK, to a nucleolar protein gene, NPM, in non-Hodgkin’s lymphoma. Science. 1994;263(5151):1281-1284. 2. Chiarle R, Voena C, Ambrogio C, Piva R, Inghirami G. The anaplastic lymphoma kinase in the pathogenesis of cancer. Nat Rev Cancer. 2008;8(1):11-23. 3. Woessmann W, Zimmermann M, Lenhard M, et al. Relapsed or refractory anaplastic large-cell lymphoma in children and adolescents after Berlin-Frankfurt-Muenster (BFM)-type first-line therapy: a BFMGroup study. J Clin Oncol. 2011;29(22):3065-3071. 4. Mussolin L, Damm-Welk C, Pillon M, et al. Use of minimal disseminated disease and immunity to NPM-ALK antigen to stratify ALKpositive ALCL patients with different prognosis. Leukemia. 2013;27(2):416-422. 5. Damm-Welk C, Mussolin L, Zimmermann M, et al. Early assessment of minimal residual disease identifies patients at very high relapse risk in NPM-ALK-positive anaplastic large-cell lymphoma. Blood. 2014;123(3):334-337. 6. Pomari E, Basso G, Bresolin S, et al. NPM-ALK expression levels identify two distinct subtypes of paediatric anaplastic large cell lymphoma. Leukemia. 2017;31(2):498-501. 7. Hoareau-Aveilla C, Meggetto F. Crosstalk between microRNA and DNA methylation offers potential biomarkers and targeted therapies in ALK-positive lymphomas. Cancers (Basel). 2017;9(8). 8. Hoareau-Aveilla C, Quelen C, Congras A, et al. miR-497 suppresses cycle progression through an axis involving CDK6 in ALK-positive cells. Haematologica. 2019;104(2):347-359. 9. Staber PB, Vesely P, Haq N, et al. The oncoprotein NPM-ALK of anaplastic large-cell lymphoma induces JUNB transcription via ERK1/2 and JunB translation via mTOR signaling. Blood. 2007;110(9):3374-3383. 10. Pearson JD, Lee JKH, Bacani JTC, Lai R, Ingham RJ. NPM-ALK and the JunB transcription factor regulate the expression of cytotoxic molecules in ALK-positive, anaplastic large cell lymphoma. Int J Clin Exp Pathol. 2011;4(2):124-133. 11. Watanabe M, Sasaki M, Itoh K, et al. JunB induced by constitutive CD30-extracellular signal-regulated kinase 1/2 mitogen-activated protein kinase signaling activates the CD30 promoter in anaplastic large cell lymphoma and Reed-Sternberg cells of Hodgkin lymphoma. Cancer Res. 2005;65(17):7628-7634. 12. Hsu FYY, Johnston PB, Burke KA, Zhao Y. The expression of CD30 in anaplastic large cell lymphoma is regulated by nucleophosminanaplastic lymphoma kinase-mediated JunB level in a cell type-specific manner. Cancer Res. 2006;66(18):9002-9008. 13 Atsaves V, Zhang R, Ruder D, et al. Constitutive control of AKT1 gene expression by JUNB/CJUN in ALK+ anaplastic large-cell lymphoma: a novel crosstalk mechanism. Leukemia. 2015;29(11):21622172. 14. Laimer D, Dolznig H, Kollmann K, et al. PDGFR blockade is a rational and effective therapy for NPM-ALK-driven lymphomas. Nat Med. 2012;18(11):1699-1704.

613

Letters to the Editor

Myeloid/lymphoid neoplasms with eosinophilia/basophilia and ETV6-ABL1 fusion: cell-of-origin and response to tyrosine kinase inhibition ETV6-ABL1 rearrangements have been reported in a spectrum of hematologic malignancies, including B- or T-acute lymphoblastic leukemia (ALL), acute myeloid leukemia (AML), and myeloproliferative neoplasms (MPN).1-11 Mostly reported as single cases, ETV6-ABL1 rearranged MPN shows clinical features mimicking chronic myeloid leukemia (CML) and empirically responds to tyrosine kinase inhibitors (TKI). Therefore, these cases are commonly diagnosed as Philadelphia chromosome negative CML (Phâ&#x20AC;&#x201C; CML), CML with atypical ABL1 fusions, or even atypical CML. Transformation to AML, B-ALL and T-ALL has also been observed in a subset of cases.3-8 In addition, eosinophilia is a hallmark in nearly all cases, proportionally much greater than seen in CML. On the other hand, a few de novo AML and ALL cases with this fusion also present with eosinophilia,9,10 raising the possibility of a progression from an underlying chronic phase of myeloid/lymphoid neoplasm, similar to those seen in PDGFRA, PDGFRB, FGFR1 and PCM1JAK2 rearrangements. Here we report six patients with myeloid/lymphoid proliferation and ETV6-ABL1 fusions and review the literature. Our findings support the classification of such cases as myeloid/lymphoid neoplasms with eosinophilia/basophilia and ETV6-ABL1 fusion, similar to the category of myeloid/lymphoid neoplasms with eosinophilia and rearrangements of PDGFRA, PDGFRB, or FGFR1, or with PCM1-JAK2 listed in the World Health Organization classification.12 A search for cases at Memorial Sloan Kettering Cancer Center from January 2014 to December 2019 was performed and identified five patients with ETV6-ABL1 fusions and we also include a case from University Hospital Cleveland Medical Center (Cleveland, OH, USA). The clinicopathologic features including laboratory findings, pathologic evaluation, and cytogenetic and molecular results are summarized in Table 1. Two female and four male patients were included, with a median age of 49.5 years (range 23-88 years). All six patients presented with myeloid proliferation and eosinophilia: four patients were diagnosed as CML, atypical CML or CML with atypical ABL1 fusion, one as essential thrombocythemia (ET) based on morphologic findings and peripheral blood counts, and one as myeloid/lymphoid neoplasm with eosinophilia and ETV6-ABL1 fusion. Five patients (with the exception of Patient 4) were treated with either first- or second-generation TKI (imatinib, dasatinib and nilotinib) and showed complete cytogenetic response a few months (range 2-6 months) after initiation of treatment. Patient 1 had cytogenetic and morphologic relapse after imatinib treatment for 10 years (ABL1 mutational analysis failed) but again achieved cytogenetic remission 2 months after switching to dasatinib treatment. This patient continued to have cytogenetic remission 5 years after dasatinib. Patient 4 had a cryptic rearrangement not detected by routine karyotyping and was initially managed as ET. The patient failed multiple lines of treatment (hydroxyurea, Heat Shock Protein 90 inhibitor, ruxolitinib, anagrelide, and Îą-interferon), and progressed four years later to AML with marked basophilia. RNA-based sequencing studies revealed ETV6-ABL1 fusion, confirmed by fluorescence in situ hybridization (FISH) analysis. Combined imatinib and cytarabine treatment was initiated; howev614

er, the patient died shortly after due to comorbid complications. Patient 5 responded to nilotinib treatment for 2 years then progressed to B-ALL (ABL1 mutational analysis failed) and obtained complete remission after HyperCVAD. Patient 6 presented as myeloid sarcoma and T-lymphoblastic lymphoma (TLL) in two separate foci of the same neck node with no increased blasts in the marrow. She was treated with dasatinib for 8 weeks. Her lymphadenopathy and eosinophilia both resolved. All patients had peripheral eosinophilia, ranging from 1.7-44.5x109/L. Patient 1 had 11% eosinophils in peripheral blood (PB) at relapse. Three (Patients 2-4) had eosinophilia in the marrow at the time of diagnosis. Peripheral basophilia was documented in Patients 3 and 5 at presentation, in Patient 1 at relapse (2% basophils in PB) and in Patient 4 at transformation (10% basophils in PB). Leukocytosis was widely variable, ranging from 9374 x109/L. Patient 3 had anemia and thrombocytopenia. Patient 4 had marked thrombocytosis but no splenomegaly. Patient 5 had anemia. Patients 2-6 had diagnostic marrow biopsy for review that showed 90100% cellularity and markedly increased M:E ratio. Blasts were not increased in any of the six patients at diagnosis. Megakaryocyte morphology was highly variable: both small and large forms in Patient 2, predominantly large forms in Patient 4, increased hypolobated forms in Patients 5 and 6, unremarkable in the other two cases (Figure 1A). There was no overt dysplasia in myeloid or erythroid lineages observed in any of the cases (data not shown). FISH analysis using ETV6 and/or ABL1 break-apart probes detected the presence of the ETV6 rearrangement in metaphase cells in four patients (Patients 1, 2, 4 and 5): ABL1 rearrangement in Patient 2, an extra normal fusion signal in Patients 1 and 6 (ABL1 gain), and normal signal pattern in both Patients 4 and 5. Patient 6 had no ETV6 FISH testing but only ABL1. FISH was not performed on the diagnostic sample from Patient 3. Next generation RNA sequencing (RNAseq NGS) with a customized 199gene panel (Archer FusionPlex) identified the ETV6-ABL1 transcripts involving the same breakpoints with ETV6 exon 5 and ABL1 exon2 in all six cases (Figure 1D and Online Supplementary Table S1). Next generation DNA sequencing was performed using FoundationOne Heme (Foundation Medicine 406 gene panel, Patients 1, 4 and 6) and MSK-IMPACT Heme (400 gene panel, Patients 2, 3 and 5). Patients 1 and 2 were positive for ARID2 truncating mutations. In addition, Patient 1 had TP53 point mutation while Patient 2 had CDKN1B truncating mutations. Patient 5 was positive for SETD2 mutation. Patients 3, 4 and 6 were negative for additional mutations (Online Supplementary Table S2). While ARID2 defect has been associated with megakaryocytic dysplasia,13 in our study, one patient harboring an ARID2 mutation had no megakaryocytic atypia (Patient 1) whereas the other showed variable megakaryocyte morphology (Patient 2), suggesting that the functional significance of ARID2 mutations in such cases needs further investigation. Although CDKN1B expression level was reported to be a potential biomarker for prognostication in acute myeloid leukemia,14 the biological role of this mutation in this entity remains unclear. To investigate the downstream signaling pathway activation of ETV6-ABL1 fusion, phosphorylation levels of STAT3, STAT5 and ERK were evaluated by immunochemistry using antibodies specific for phosphorylated proteins on the bone marrow biopsy from Patient 4. Although phospho-STAT3 was not increased, phospho-STAT5 showed a markedly increased signal, suggestive of a spehaematologica | 2021; 106(2)

Letters to the Editor

cific STAT5 activation as a downstream target (Figure 1B). ERK1/2 phosphorylation was also increased (Figure 1C). In order to study the cell-of-origin of the ETV6-ABL1 fusions, various cell populations from two patients (Patients 1 and 4) were sorted based on immunophenotype by flow cytometry, including CD34+CD38â&#x20AC;&#x201C; (hematopoietic stem cells and early progenitors, HSPC),

CD34+CD38+ (late committed progenitors), monocytes, granulocytes and lymphocytes. FISH studies were performed on the sorted cells. ETV6 rearrangements were observed in CD34+CD38â&#x20AC;&#x201C;, CD34+CD38+, monocytes, and granulocytes but not in mature lymphocytes (Figure 1FK), supporting the view that ETV6-ABL1 fusions originate from HSPC.

Table 1. Clinicopathologic features of myeloid neoplasms with ETV6-ABL1 fusions.

Patient ID

Age/Sex Splenomegaly WBC (x109/L) Hb (g/dL) PLT (x109/L) Abs Eo(x109/L) at presentation Abs Baso(x109/L) at presentation BM cellularity Megakaryocytes

34 yrs/M Yes 27 13.9 236 1.7 0.5 Normal* Unremarkable*

23 yrs/M Yes 374 6 76 44.5 3.7 100% Unremarkable

M:E ratio Eo (%) on aspirate Blast (%) Initial diagnosis

7.3* 3* Not increased* CML with atypical fusion

45 yrs/M Yes 17.2 11.6 203 1.7 0.2 100% Ranging from small to large, clustering 10:1 32 Not increased Atypical CML

Karyotype

46,XY,t(9;12) (q34;p13)[9]/46, XY[11]

ETV6 FISH Mutations

Positive ARID2 TP53 ETV6-ABL1 Imatinib, Dasatinib

Fusions Treatment

Response to TKI

55 yrs /F No 9 12.5 818 9.2 0.2 100% Increased, large, clustering >10:1 10:1 12 10 Not increased Not increased CML Essential with atypical thrombocythemia fusion (ET)

46,XY,t(9;12) 46,XY (q34.1;p13)[16]/ 46,idem,del(7) (q22q36)[1]/46,XY[11] Positive NA ARID2 Absent CDKN1B ETV6-ABL1 ETV6-ABL1 Dasatinib Dasatinib

Diagnosed in 2005 Diagnosed achieved cytogenetic in 02/2018, remission on imatinib; achieved relapsed in 2015, cytogenetic/ achieved cytogenetic/ molecular molecular remission remission on dasatinib. on dasatinib; Alive Alive

46,XX

Positive Absent

54 yrs/M Yes 217 7.8 191 2.0 2.1 >90% Increased, hypolobated

88 yrs/F No 14.5 11.7 417 2.6 0.6 75% Increased, hy-polobated

10:1 6 2 CML

3.4:1 7 4 Myeloid/ lymphoid neoplasm with eosinophilia and gene rearrangement 45,XY,-7[17]/ 53,XX,+X,+8, 46,idem,+11 [4] +10,+11,+14, +18,+19[7]/ 46,XX[13] Positive NA SETD2 Absent

ETV6-ABL1 ETV6-ABL1 ETV6-ABL1 Hydroxyurea, Nilotinib Dasatinib Heat shock protein Inhibitor, ruxolitinib, Anagrelide, interferon Imatinib, cytarabine Diagnosed in Treated as ET Leukocytosis Marked 10/2018, in 2011-2016; in 2016, improvement achieved progressed to AML diagnosed in lymphadenopathy; cytogenetic/ in 2016 and treated and treated resolution of molecular with induction with nilotinib eosinophilia, remission and imatinib until 2018, alive on dasatinib; died in a month progressed Alive from complications to BALL, achieved remission with HyperCVAD, alive

*At relapse in 2015. WBC: white blood cell count; PLT: platelets; yrs: years; mons: months; M: male; F: female; NA: not available; CML: chronic myeloid leukemia; TKI: tyrosine kinase inhibitor; MPAL: mixed phenotype acute leukemia.

haematologica | 2021; 106(2)

615

Letters to the Editor

All six ETV6-ABL1 rearranged patients showed myeloid proliferation and eosinophilia, a common characteristic of ETV6 rearranged myeloid neoplasms, such as ETV6-PDGFRA, ETV6-PDGFRB, and ETV6-JAK2.15 Review of the literature identified 21 additional cases of MPN and four cases of AML with ETV6-ABL1 fusions

(median age 48 years, 18 male, 7 female) (Online Supplementary Table S3). PB absolute eosinophilia was reported in 13 of 14 patients who had available data for evaluation. Interestingly, 8 of 14 patients also had peripheral basophilia (defined by >1% basophils). Six out of 21 MPN cases with ETV6-ABL1 rearrangements progressed

Figure 1. Myeloid neoplasms with ETV6-ABL1 fusions. (A) Hematoxylin & Eosin staining of a bone marrow biopsy specimen (Patient 4) showed hypercellularity, increased M:E ratio, highly variable megakaryocytes morphology, predominantly large in size. (B) Immunohistochemistry showed markedly increased phosphoSTAT5 signals but not phospho-STAT3 (inset). (C) Immunohistochemistry showed mildly increased phospho-ERK1/2 signals. (D) Schematic illustration of ETV6ABL1 fusions, bidirectional RNA sequencing reads, and transcript sequence of the in-frame fusion product detected by Archer FusionPlex with exons 1-5 of ETV6 fused to exons 2-11 of ABL1. (F-K) Flow cytometry sorted cell populations, including CD34+CD38– (enriched for hematopoietic stem cells, HSC), CD34+CD38+ (hematopoietic progenitors/blasts), monocytes, granulocytes and lymphocytes (Lym) (F and I). Fluorescence in situ hybridization (FISH) analysis with an ETV6 break-part probe set (Abbott Molecular) shows a split ETV6 signal pattern. The 5′ and 3′ ETV6 were labeled with green and red, respectively. ETV6 rearrangement was observed in CD34+CD38– (G), CD34+CD38+ (H), monocytes (J), and granulocytes (K) but not in mature lymphocytes (inset in I). (L) A summary of ETV6 FISH results on flow sorted cell populations from two patients. n/a: not available.

616

haematologica | 2021; 106(2)

Letters to the Editor

to acute leukemia (3 AML, 2 B-ALL, and 1 T-ALL) and all died shortly after transformation despite the addition of TKI to the standard chemotherapy.3-8 In contrast, four other MPN patients (with no increased blasts) who received TKI achieved long-term survival (longest survival 9 years),1,2,11 similar to the experience at our institution. Therefore, it appears to be of paramount importance to identify this fusion and incorporate TKI treatment at an early stage of disease. High index of suspicion is critical to initiate proper testing in patients presenting with signs of MPN and eosinophilia. With disease progression, TKI response may be limited. Considering the clinicopathologic similarities (myeloid proliferation, eosinophilia, basophilia, and transformation to AML, B-ALL and T-ALL) between these cases and sensitivity to TKI treatment, we propose classifying them as one group: myeloid/lymphoid neoplasms with eosinophilia/basophilia and ETV6-ABL1 fusion. Our findings do not favor a diagnosis of Ph- CML or atypical CML based on the pathologic features, eosinophilia and potential transformation to T-ALL. Due to the cryptic nature of t(9;12), the incidence of ETV6-ABL1 fusion might have been underestimated in the literature. The presence of additional ABL1 signal by FISH studies in the absence of BCR-ABL1 fusion is a clue to search for other ABL1 fusions in a subset of cases. FISH analysis using ETV6 and ABL1 break-apart probes and RNA-based NGS assay are able to detect the fusion with high sensitivity. RNAseq NGS assay has recently been developed for fusion detection in hematologic malignancies, which can overcome the technical barriers in the identification of ETV6-ABL1 fusion by traditional methods. Our laboratory’s panel has extensive ETV6 and ABL1 coverage, facilitating the detection of both ETV6-ABL1 and any other novel ETV6 or ABL1 fusions. With the wide application of RNAseq assay, we expect that more fusions will be detected in hematologic malignancies. RNAseq assay can also elucidate transcript type, critical in determining “A versus B”.16 Type A involves ETV6 exon 4, whereas type B involves exon 5, joining to ABL1 exon 2; type B is the prevalent transcript form (17 of 18 cases in Online Supplementary Table S3), detected in five patients with ETV6-ABL1 fusions in our institution. As a member of Ets family of transcription factors, when ETV6 fuses to a receptor tyrosine kinase, the fusion protein displays an elevated tyrosine kinase activity. Type B has higher kinase activity since ETV6 exon 5 includes a direct binding site for the SH2 domain of the GRB2, which enhances the PI3-kinase and MAP Kinase pathways.17 Consistently, ERK phosphorylation was increased in ETV6-ABL1 positive marrow cells. In addition, STAT5 phosphorylation was also upregulated similar to observations in BCR-ABL1 CML, although this needs to be confirmed in more patients. In summary, our approach combining flow cytometry sorting technology and downstream FISH studies clearly demonstrated ETV6-ABL1 fusions in sorted HSC, myeloid progenitors/blasts, granulocytes and monocytes but not lymphocytes, suggesting an HSC origin. The mechanism of this fusion driving myeloid/lymphoid differentiation remains to be investigated. Patients with these fusions typically present initially as myeloid/lymphoid proliferation with eosinophilia/basophilia and show hypersensitivity to TKI treatment. Early identification by FISH (particularly by ETV6 probes) and/or RNA sequencing and potential for such therapy is advantageous based on the current and published experience with this rare neoplasm. We propose to classify this group of disease as myeloid/lymphoid neoplasms with haematologica | 2021; 106(2)

eosinophilia/basophilia and ETV6-ABL1 fusion to facilitate and unify the clinico-pathologic and molecular diagnosis and subsequent clinical management. Development of a quantitative polymerase chain reaction assay similar to BCR-ABL1 fusion is needed for future disease and therapy monitoring. Jinjuan Yao,1,* Lianrong Xu,1,* Umut Aypar,1,* Howard J. Meyerson,2 Dory Londono,1 Qi Gao,1 Jeeyeon Baik,1 James Dietz,1 Ryma Benayed,1 Allison Sigler,1 Mariko Yabe,1 Ahmet Dogan,1 Maria E. Arcila,1 Mikhail Roshal,1 Yanming Zhang,1 Michael J. Mauro3and Wenbin Xiao1 1 Department of Pathology, Memorial Sloan-Kettering Cancer Center, New York, NY; 2Department of Pathology, University Hospitals of Cleveland, Cleveland, OH and 3Department of Medicine, Memorial Sloan-Kettering Cancer Center, New York, NY, USA *JY, LX and UA contributed equally as co-first authors. Correspondence: WENBIN XIAO - xiaow@mskcc.org doi:10.3324/haematol.2020.249649 Disclosures: WX has received research support from Stemline Therapeutics. MJM has served as a consultant/advisor for Novartis, Pfizer, Bristol-Myers Squibb, Takeda/Millennium, and receives institutional research support from Bristol-Myers Squibb, Novartis, and Sun Pharma/SPARC. The other authors have no conflicts of interest to disclose. Contributions: JY, LX, YZ, MJM and WX conceived the study, collected and analyzed the data, and wrote the manuscript; UA, DL, and YZ performed cytogenetic studies and interpreted the data; HM contributed critical clinical materials; QG, JB, JD, RB, MY, NS and AEM collected data; AD and MR interpreted the data. All the authors approved the final version of the manuscript. Funding: This study was supported in part through the NIH/NCI Cancer Center Support Grant P30 CA008748.

References 1. Xie W, Wang SA, Hu S, Xu J, Medeiros LJ, Tang G. Myeloproliferative neoplasm with ABL1/ETV6 rearrangement mimics chronic myeloid leukemia and responds to tyrosine kinase inhibitors. Cancer Gen. 2018;228-229:41-46. 2. Zaliova M, Moorman AV, Cazzaniga G, et al. Characterization of leukemias with ETV6-ABL1 fusion. Haematologica. 2016;101(9): 1082-1093. 3. O'Brien SG, Vieira SA, Connors S, et al. Transient response to imatinib mesylate (STI571) in a patient with the ETV6-ABL t(9;12) translocation. Blood. 2002;99(9):3465-3467. 4. Kelly JC, Shahbazi N, Scheerle J, et al. Insertion (12;9)(p13;q34q34): a cryptic rearrangement involving ABL1/ETV6 fusion in a patient with Philadelphia-negative chronic myeloid leukemia. Cancer Gen Cytogenet. 2009;192(1):36-39. 5. Barbouti A, Ahlgren T, Johansson B, et al. Clinical and genetic studies of ETV6/ABL1-positive chronic myeloid leukaemia in blast crisis treated with imatinib mesylate. Br J Haematol. 2003;122(1):85-93. 6. Tirado CA, Siangchin K, Shabsovich DS, Sharifian M, Schiller G. A novel three-way rearrangement involving ETV6 (12p13) and ABL1 (9q34) with an unknown partner on 3p25 resulting in a possible ETV6-ABL1 fusion in a patient with acute myeloid leukemia: a case report and a review of the literature. Biomark Res. 2016;4(1):16. 7. Kakadia PM, Schmidmaier R, Volkl A, et al. An ETV6-ABL1 fusion in a patient with chronic myeloproliferative neoplasm: Initial response to imatinib followed by rapid transformation into ALL. Leuk Res Rep. 2016;6:50-54. 8. Yamamoto K, Yakushijin K, Nakamachi Y, et al. Extramedullary Tlymphoid blast crisis of an ETV6/ABL1-positive myeloproliferative neoplasm with t(9;12)(q34;p13) and t(7;14)(p13;q11.2). Ann Hematol. 2014;93(8):1435-1438. 9. La Starza R, Trubia M, Testoni N, et al. Clonal eosinophils are a morphologic hallmark of ETV6/ABL1 positive acute myeloid leukemia. Haematologica. 2002;87(8):789-794. 10. Park J, Kim M, Lim J, et al. Variant of ETV6/ABL1 gene is associated

617

Letters to the Editor

with leukemia phenotype. Acta Haematol. 2013;129(2):78-82. 11. Perna F, Abdel-Wahab O, Levine RL, Jhanwar SC, Imada K, Nimer SD. ETV6-ABL1-positive "chronic myeloid leukemia": clinical and molecular response to tyrosine kinase inhibition. Haematologica. 2011;96(2):342-343. 12. Arber DA, Orazi A, Hasserjian R, Thiele J, Borowitz MJ, Le Beau MM, et al. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood. 2016; May 19;127(20):2391-405. 13. Sakai H, Hosono N, Nakazawa H, et al. A novel genetic and morphologic phenotype of ARID2-mediated myelodysplasia. Leukemia. 2018;32(3):839-843. 14. Haferlach C, Bacher U, Kohlmann A, et al. CDKN1B, encoding the cyclin-dependent kinase inhibitor 1B (p27), is located in the mini-

618

mally deleted region of 12p abnormalities in myeloid malignancies and its low expression is a favorable prognostic marker in acute myeloid leukemia. Haematologica. 2011;96(6):829-836. 15. Haferlach C, Bacher U, Schnittger S, et al. ETV6 rearrangements are recurrent in myeloid malignancies and are frequently associated with other genetic events. Genes Chromosomes Cancer. 2012; 51(4):328-337. 16. Choi SI, Jang MA, Jeong WJ, et al. A case of chronic myeloid leukemia with rare variant ETV6/ABL1 rearrangement. Ann Lab Med. 2017;37(1):77-80. 17. Million RP, Harakawa N, Roumiantsev S, Varticovski L, Van Etten RA. A direct binding site for Grb2 contributes to transformation and leukemogenesis by the Tel-Abl (ETV6-Abl) tyrosine kinase. Mol Cell Biol. 2004;24(11):4685-4695.

haematologica | 2021; 106(2)

Letters to the Editor

SMAD1 promoter hypermethylation and lack of SMAD1 expression in Hodgkin lymphoma: a potential target for hypomethylating drug therapy Hodgkin lymphoma (HL) is an immunologically active lymphoid neoplasm composed of a few (usually 1-10%) neoplastic Hodgkin and Reed-Sternberg (HRS) cells or lymphocyte-predominant (LP) cells and >90% non-neoplastic cells, mainly T- and B-lymphocytes, plasma cells, macrophages, eosinophils and fibroblasts. The substantial amount of reactive cells in HL is supposed to be the net effect of a complex signaling network of cytokines and chemokines secreted by either the HRS cells or nonneoplastic cells.1 One component of this network is transforming growth factor beta (TGF-β), which is produced by HRS cells and cancer-associated fibroblasts. TGF-β unfolds its immunosuppressive impact by stimulating tumor-infiltrating T-lymphocytes (TIL) to differentiate into anergic, tumor-promoting, regulatory T cells (Treg).2 Additionally, TGF-β inhibits natural killer cells one of the key components of the innate anticancer immunity.3 Interestingly and still poorly understood, the HRS cells themselves seem to remain unaffected by the tumor-suppressive properties of TGF-β.4 Recent studies on diffuse large B-cell lymphoma (DLBCL) revealed a previously unknown tumor-suppressive signaling axis involving SMAD1 as a downstream messenger of TGF-β.5 SMAD1 functions as an intracellular signal transducer between extracellular TGF-β and the nucleus, where it modulates the transcription of target genes. This signaling cascade was shown to be recurrently inactivated in DLBCL, mainly by hypermethylation of five promoter regions surrounding the SMAD1 transcription start site, which finally generates a significant growth advantage for lymphoma cells.5 In the course of these investigations, we noted that SMAD1 was not expressed in HRS cells of screened HL cases. This led us to hypothesize that the absence of SMAD1 expression in HRS cells may mechanistically be linked to their resistance to the tumor-suppressive effects of TGF-β.4 In order to further elucidate this finding, we analyzed 132 well-characterized archival tissue-microarrayed cases,6 and 11 conventional routine lymphadenectomy

specimens from patients suffering from all subtypes of classic HL (77 nodular sclerosis [NS]; 48 mixed cellularity [MC]; 7 lymphocyte-rich [LR]; 5 lymphocyte-depleted; and 6 unclassifiable classic HL) and 14 routine samples from patients suffering from nodular lymphocyte-predominant HL (NLPHL). We analyzed all these instances for immunohistochemical expression of SMAD1. Importantly, to guarantee retained antigenicity, only cases containing (physiologically) SMAD1-positive endothelia were considered. We found that all NLPHL (14/14 cases; 100%) and the great majority of classic HL (138/143 cases; 97%) displayed SMAD1-negative LP and HRS cells, respectively (Figure 1A and B). Single HRS cells stained faintly for SMAD1 in five cases only (2 NS; 2 MC; and 1 LR classic HL). With respect to non-neoplastic cells, 65/143 classic HL (45%) showed moderate (15-49% of TIL) up to abundant (≥50% of TIL) amounts of SMAD1 positive surrounding TIL, thus being potentially susceptible to the suppressive influence of TGF-β (Figure 1A and B); in NLPHL, 11/14 (79%) cases displayed abundant SMAD1-expressing TIL, including TIL involved in rosetting around LP cells (Online Supplementary Figure S1). The presence of abundant SMAD1-expressing TIL did not correlate with disease stage, patients’ age, gender, presence of B symptoms, association with Epstein-Barr virus (EBV) or outcome, while showing significant correlations with the NS subtype (45/77 NS cases, i.e. 58%, compared to 20/66 non-NS cases, i.e. 30%, P=0.025 χ2 test) and with the amount of FOXP3-positive Treg (Rho=0.351, P=0.000053 Spearman correlation), which both, in turn, may be directly linked to the effects of TGF-β, promoting sclerosis and a shift towards Treg differentiation.2 In contrast, surrounding plasma cells seemed to lack SMAD1 expression, potentially rendering them insensitive to the pro-apoptotic and anti-proliferative signals of TGF-β.7 With regard to plasma cells, this largely fits with the newly described negative prognostic impact of their increased numbers in classic HL.8 To strengthen our hypothesis, we investigated the promoter methylation status of the SMAD1 gene in six different HL cell lines, including one NLPHL cell line (DEV) exactly as described elsewhere.5 Methylation analysis by bisulfite sequencing was successful for three regions of

Figure 1. Expression of SMAD1 in classic Hodgkin lymphoma. (A) Tissue microarrayed archival mixed cellularity classic Hodgkin lymphoma with moderate numbers of SMAD1-positive tumor-infiltrating lymphocytes (TIL) and a few strongly staining endothelia. Note that all Hodgkin and Reed-Sternberg (HRS) cells are negative. (B) Diagnostic lymphadenectomy of a nodular sclerosis cHL with abundant SMAD1-positive TIL and a few strongly staining endothelia. Note that all HRS cells are negative. Immunoperoxidase staining, original magnification 400x.

haematologica | 2021; 106(2)

619

Letters to the Editor

the SMAD1 promoter and yielded four hypermethylated cell lines (L428, KMH-2, DEV, HDLM-2), that differed substantially, particularly regarding their methylation of region A4(3), from cell lines showing no evidence of promoter hypermethylation (L1236, L540) (Figure 2A). Importantly, KMH-2 is known to be SMAD1 p.ADTP220fs mutant (https://portals.broadinstitute.org/ ccle/page?cell_line=KMH2_haematopoietic_and_lymphoid_tis sue), which additionally points towards a potential role of this gene silencing in lymphomagenesis. The impact of the promoter methylation status of SMAD1 on protein expression was further addressed by western blot analysis, comparing one hypermethylated cell line (DEV) with a cell line without hypermethylation (L1236). Concordantly, the expression of SMAD1 differed clearly, without detectable protein in the DEV cell line (Figure 2B). When treated with the DNA methyltransferase inhibitor decitabine, which reverses, among others, the hypermethylation of SMAD1 promoters, the SMAD1-negative cell line DEV died immediately after exposure. As expected, the expression of SMAD1 was not affected by treatment in the L1236 cell line, which is not hypermethylated. To obtain further clinical evidence, the promoter methylation status of SMAD1 was assessed in samples from three patients with classic HL. To do this, we used our newly developed, flow sorting-assisted technique for HRS cell enrichment from formalin-fixed and paraffinembedded tissues, allowing for targeted genetic analysis of DNA isolated from classic HL tumor cells.9 In this collective, not only the A4(3) promoter region of SMAD1, but also the A1(1) region was substantially hypermethylated (Figure 2C). Furthermore, SMAD1 promoter hypermethylation was identified in sorted tumor-infiltrating plasma cells, fitting with the immunohistochemically 620

Figure 2. The SMAD1 promoter region is hypermethylated in Hodgkin lymphoma cell lines and patients’ samples. (A) Methylation analysis by bisulfite sequencing of regions A1(1), A4(3) and B2(5) within the SMAD1 promoter in a panel of Hodgkin lymphoma (HL) cell lines. Each circle represents one CG dinucleotide; black circles indicate methylated, white circles indicate unmethylated cytosines. Each line represents one clone. Two or three clones were sequenced per sample. X indicates aligned mismatches between genomic and bisulfite sequences. The cell line KMH-2 has a known frameshift mutation at the SMAD1 locus. (B) SMAD1 expression at the protein level after no treatment (-) or treatment with 1 mM decitabine (+) for 96 h of two HL cell lines as determined by western blotting. Results from two independent experiments are exemplified. Decitabine-treated samples of the DEV cell line are not shown because of lack of protein after demethylating treatment. (C) Methylation analysis by bisulfite sequencing of regions A1(1), A4(3) and B2(5) within the SMAD1 promoter in samples from three patients with classic HL. Each circle represents one CG dinucleotide; black circles indicate methylated, white circles indicate unmethylated cytosines. Each line represents one clone. Two or three clones were sequenced per sample. X indicates aligned mismatches between genomic and bisulfite sequences.

noted lack of SMAD1 in plasma cells. Although risk-adjusted standard treatment of HL is successful in over 90% of patients, relapses after salvage therapy and refractory cases represent oncological challenges and run an unfavorable clinical course with limited therapeutic options. In this context, demethylating agents such as decitabine and azacytidine may be of potential therapeutic interest and have already shown promising effects. At clinically relevant concentrations, decitabine has been documented to inhibit the growth of classic HL cell lines in vitro and a single case observation of regressing relapsed classic HL as an unexpected “side effect” of azacytidine has been reported in a patient suffering from concomitant myelodysplastic syndrome.10,11 Decitabine and azacytidine inhibit DNA methyltransferases, and thereby reverse promoter hypermethylation of SMAD1.12 Importantly, promising results with decitabine were also obtained in DLBCL cell lines lacking SMAD1 expression due to promoter hypermethylation.5 Four days of treatment were sufficient to restore SMAD1 transcription and protein expression in a subset of initially SMAD1-negative DLBCL cell lines, an effect corroborated by observations in a patient-derived xenograft DLBCL mouse model with proven SMAD1 promoter hypermethylation.5 Analogous to this, we treated the SMAD1-negative HL cell line DEV with decitabine. The cells died immediately after exposure, possibly due to marked responsiveness to decitabine. Since therapeutic reversion of SMAD1 promoter hypermethylation would be of potential relevance only in the presence of TGF-β receptors (TGFBR), we analyzed publically available (Gene Expression Omnibus [GEO] accession n. GSE12453) and our own (GEO accession n. GSE147387) gene expression data from primary HRS and LP cells and cell lines to estimate whether HRS and LP haematologica | 2021; 106(2)

Letters to the Editor

cells express TGFBR and associated proteins (ACVR1, ACVR1B, ACVR1C, ACVR2A, ACVR2B, ACVRL1, AMHR2, BMPR2, TGFBR1, TGFBR2, TGFBR3, TGFBRAP1). The classic HL cell line KMH-2 contained relevant transcript levels of all TGFBR types, HDLM-2 expressed TGFBR1 and TGFBR3, and the NLPHL cell line DEV expressed TGFBR1; the classic HL cell lines L1236 and L428 exhibited TGBRAP1 transcripts. In all these cell lines SMAD1 transcripts were decreased. In summary, our data suggest a likely important, not yet described role of SMAD1 hypermethylation in HL, potentially causing an imbalance of TGF-β signaling axis responses in involved tissues. SMAD1 has been demonstrated to be part of the TGF-β-mediated anti-proliferative pathway in different B-cell lymphomas. Intriguingly, lymphomas with mutated or knocked-out SMAD1 were protected from the tumor-suppressive effects of TGF-β.13 Lack of SMAD1 expression in HRS and LP cells due to promoter hypermethylation or gene mutation may analogously contribute to their resistance towards the proapoptotic and anti-proliferative effects of TGF-β, despite the presence of TGFBR transcripts. This hypothesis is further supported by observations in EBV-positive classic HL, in which decreased SMAD2 levels due to EBNA1mediated increased protein turnover14 disable TGF-β signaling, being congruent with our data regarding SMAD1 downregulation in HRS cells. In contrast, retained SMAD1 expression in surrounding TIL may contribute to immune escape, as intact TGF-β signaling promotes T-cell differentiation into tumor-supporting Treg,2 which is reflected by the observed correlation between higher numbers of FOXP3positive Treg and expression of SMAD1 in TIL. The tumor-suppressive effects of TGF-β on TIL have recently been challenged by a promising clinical study in which the infusion of TGF-β-insensitive T cells was successfully used in patients with EBV-positive relapsed classic HL.15 Our data suggest a possible rationale for the application of a more tailored treatment with hypomethylating agents in HL, which may be worth of prospective investigations, as agents such as decitabine have already shown promising results in SMAD1 hypermethylated DLBCL5 and in classic HL cell lines.10 Magdalena M. Gerlach,1 Anna Stelling-Germani,2 Cheuk Ting Wu,2 Sebastian Newrzela,3 Claudia Döring,3 Visar Vela,1 Anne Müller,2 Sylvia Hartmann3 and Alexandar Tzankov1 1 Institute of Medical Genetics and Pathology, University of Basel, Basel, Switzerland; 2Institute of Molecular Cancer Research, University of Zurich, Zurich, Switzerland and 3Institute of Pathology, Department of Pathology, Goethe University Frankfurt on Main, Germany Correspondence: ALEXANDAR TZANKOV - alexandar.tzankov@usb.ch doi:10.3324/haematol.2020.249276 Disclosures: no conflicts of interests to disclose. Contributions: MMG wrote the manuscript and evaluated the histology and immunohistochemical stains; AS-G supervised the

haematologica | 2021; 106(2)

SMAD1 promoter methylation assessment, performed cell line viability experiments, corrected the manuscript and wrote the legend to Figure 2; CTW assessed SMAD1 promoter methylation; SN provided gene expression data of Hodgkin lymphoma cell lines; CD provided and analyzed gene expression data of Hodgkin lymphoma cell lines; VV enriched Hodgkin and Reed-Sternberg cells from archived clinical samples and isolated DNA from them; AM supervised SMAD1 promoter methylation assessment; SH provided cell lines; AT designed the study, supervised histopathological assessment, performed statistics, analyzed gene expression data of Hodgkin lymphoma cell lines, partially wrote and completely edited the manuscript Funding: this study was supported by the Swiss National Science Foundation.

References 1. Poppema S. Immunobiology and pathophysiology of Hodgkin lymphomas. Hematology Am Soc Hematol Educ Program. 2005:231238. 2. Liu Y, Zhang P, Li J, Kulkarni AB, Perruche S, Chen W. A critical function for TGF-beta signaling in the development of natural CD4+CD25+Foxp3+ regulatory T cells. Nat Immunol. 2008; 9(6):632-640. 3. Chiu J, Ernst DM, Keating A. Acquired natural killer cell dysfunction in the tumor microenvironment of classic Hodgkin lymphoma. Front Immunol. 2018;9:267. 4. Menter T, Tzankov A. Lymphomas and their microenvironment: a multifaceted relationship. Pathobiology. 2019;86(5-6):225-236. 5. Stelling A, Wu CT, Bertram K, et al. Pharmacological DNA demethylation restores SMAD1 expression and tumor suppressive signaling in diffuse large B-cell lymphoma. Blood Adv. 2019;3(20):3020-3032. 6. Tzankov A, Matter MS, Dirnhofer S. Refined prognostic role of CD68-positive tumor macrophages in the context of the cellular micromilieu of classical Hodgkin lymphoma. Pathobiology. 2010;77(6):301-308. 7. Lebman DA, Edmiston JS. The role of TGF-beta in growth, differentiation, and maturation of B lymphocytes. Microbes Infect. 1999;1(15):1297-1304. 8. Gholiha AR, Hollander P, Hedstrom G, et al. High tumour plasma cell infiltration reflects an important microenvironmental component in classic Hodgkin lymphoma linked to presence of B-symptoms. Br J Haematol. 2019;184(2):192-201. 9. Juskevicius D, Jucker D, Dietsche T, et al. Novel cell enrichment technique for robust genetic analysis of archival classical Hodgkin lymphoma tissues. Lab Invest. 2018;98(11):1487-1499. 10. Swerev TM, Wirth T, Ushmorov A. Activation of oncogenic pathways in classical Hodgkin lymphoma by decitabine: a rationale for combination with small molecular weight inhibitors. Int J Oncol. 2017;50(2):555-566. 11. D'Alò F, Leone G, Hohaus S, et al. Response to 5-azacytidine in a patient with relapsed Hodgkin lymphoma and a therapy-related myelodysplastic syndrome. Br J Haematol. 2011;154(1):141-143. 12. Foulks JM, Parnell KM, Nix RN, et al. Epigenetic drug discovery: targeting DNA methyltransferases. J Biomol Screen. 2012;17(1):2-17. 13. Munoz O, Fend F, de Beaumont R, Husson H, Astier A, Freedman AS. TGFbeta-mediated activation of Smad1 in B-cell non-Hodgkin's lymphoma and effect on cell proliferation. Leukemia. 2004;18(12):2015-2025. 14. Wood VH, O'Neil JD, Wei W, Stewart SE, Dawson CW, Young LS. Epstein-Barr virus-encoded EBNA1 regulates cellular gene transcription and modulates the STAT1 and TGFbeta signaling pathways. Oncogene. 2007;26(28):4135-4147. 15. Bollard CM, Tripic T, Cruz CR, Dotti G, Gottschalk S, Torrano V. Tumor-specific T cells engineered to overcome tumor immune evasion induce clinical responses in patients with relapsed Hodgkin lymphoma. J Clin Oncol. 2018;36(11):1128-1139.

621

Letters to the Editor

Efficacy of recombinant erythropoietin in autoimmune hemolytic anemia: a multicenter international study Autoimmune hemolytic anemia (AIHA) is due to autoantibody mediated hemolysis with or without complement (C) activation.1,2 Increasing attention has been paid to the role of bone marrow compensation,3,4 since inadequate reticulocytosis is observed in 20-40% of cases and correlates with poor prognosis.3,5 Moreover, features resembling bone marrow failure syndromes have been described in patients with chronic and relapsed/refractory AIHA, including dyserythropoiesis and fibrosis.6,7 Recombinant erythropoietin (rEPO) is an established treatment of myelodysplastic syndromes, which has been used in a limited number of AIHA cases.8,9 In this study, we systematically evaluated the safety and efficacy of EPO treatment in a multicenter, international European cohort of AIHA patients. The protocol was approved by the Ethics Committees of Human Experimentation, and patients gave informed consent in accordance with the Declaration of Helsinki. We included 51 patients with primary or secondary AIHA treated with recombinant EPO, either alone or concomitantly with other therapies. Patients were treated from June 2007 until October 2019 at nine centers in Italy, France, Norway, Austria, UK, Denmark, and the Netherlands. Efficacy of EPO was evaluated at 15 and 30 days, and then at 3, 6 and 12 months. Response was defined as complete response (CR) (hemoglobin [Hb]>12 g/dL and normalization of all hemolytic markers), partial response (PR) (Hb>10 g/dL or at least 2 g/dL increase in Hb, and no transfusion requirement). Finally, retrospective data on AIHA diagnosis and course were obtained from clinical charts. Statistical analysis was performed using Student’s t-test for continuous variables and χ2 or Fisher’s exact tests for categorical ones. Table 1 shows the patients’ characteristics at diagnosis: 66% of cases were aged >60 years and the male to female ratio was 1. Main AIHA types (warm, cold, mixed, and direct antiglobulin test [DAT] negative) were represented. Five cases were secondary to a lymphoproliferative disorder (two chronic lymphocytic leukemia, two Waldenstrom macroglobulinemia, and one marginal zone lymphoma), without active disease at the time of commencement of EPO. At commencement of EPO, the majority of patients (90%) had been pretreated with at least one therapy line, and the median time from diagnosis to EPO was 24 months (range: 0.03-187 months). Sixty-seven % of cases received EPO treatment because of non-response to a concomitant therapy, including steroids (n=24), rituximab (n=9), cytotoxic immune-suppressor (n=8), or sutimlimab (n=1). Patients who had been treated with rituximab received this drug within a median of 1 month (range: 0-5 months) before EPO. The hematologic parameters were similar to those observed at diagnosis, with 37% of cases displaying severe anemia and 53% lactate dehydrogenase levels >1.5xU/L. Inadequate reticulocytosis was observed in the majority of patients (31of 42). Regarding bone marrow characteristics, erythroid hyperplasia was present in the majority (hypercellularity of all lineages in 53%), with dyserythropoietic features in 40% and reticulin fibrosis in 9% of patients. A lymphoid infiltrate was demonstrated in 24 cases, being greater than 10% only in patients with underlying lymphoproliferative disease. In the primary forms, lymphoid infiltrate was polyclonal (T-cells in seven cases, B-cells in two, and mixed in six), but for five 622

CAD patients who showed typical monoclonal CADassociated lymphoid cells. The median endogenous EPO was 32 U/L (9.3-1,328) greater than the normal range, but inadequate in 88% of AIHA subjects considering Hb levels. Renal function was normal in all patients but two Table 1. Patients' characteristics at diagnosis.

Data at diagnosis

All patients (N=51)

Age years, median(range) M/F Secondary AIHA, N(%) Type of AIHA CAD, N(%) WAIHA IgG, N(%) WAIHA IgG+C, N(%) MIXED, N(%) DAT neg, N(%) Hematologic parameters at diagnosis Hb g/L, median(range) LDH U/L, median (range) Ret x109/L, median(range) BMRI, median(range) Previous therapy lines Treated , N(%) N of lines, mean+SD steroids, N(%) rituximab, N(%) splenectomy, N(%) immunosuppressor, N(%) Time from diagnosis to EPO, months, median (range)

Bone marrow features

Hematologic parameters Hb g/L, median(range) LDH U/L, median (range) Ret x109/L, median(range) BMRI, median(range) Concomitant therapy, N(%) Time on EPO days, median (range)

21 (41) 11 (22) 15 (29) 3 (6) 1 (2) 72 (40-118) 477 (174-6,000) 137 (10-310) 77 (4-193) 46 (90) 2.3+1.4 41 (80) 35 (68) 5 (9) 23 (45) 24 (0.03-187)

N=32

Cellularity %, median (range) Hypercellularity, N(%) Dyserythropoiesis, N(%) Lymphoid infiltrate %, median (range) Reticulinic fibrosis MF1, N(%)

Data at EPO start

68 (25-92) 24/27 5 (9)

50 (20-95) 17 (53) 13 (40) 5 (0-90) 3 (9)

N=51 85 (37-109) 344 (193-1,030) 117 (11-310) 85 (5-222) 34 (67) 188 (11-1,550)

Primary autoimmune hemolytic anemia (AIHA) was defined by hemolytic anemia and positive direct antiglobulin test (DAT), in the absence of associated overt lymphoproliferative, infectious, autoimmune, or neoplastic diseases. None of the patients had a drug-induced AIHA. Patients were classified as warm (wAIHA; DAT positive for IgG or IgG+C), cold agglutinin disease (CAD; DAT positive for C only, with high titer cold agglutinins), mixed (DAT positive for IgG+C with high titer cold agglutinins) and atypical (DAT negative, DAT positive for IgA only, warm IgM, mitogen stimulated -DAT positive only).The efficacy of the compensatory erythroblastic response was expressed as absolute reticulocyte count as well as bone marrow responsiveness index (BMRI: absolute reticulocyte count x patient’s Hb/normal Hb)[Russo R, Gambale A, Langella C, et al. Am J Hematol. 2014;89(10):e169–e175]. M: male; F: female. Hb: hemoglobin; LDH: lacatae dehydrogenase; Ret: reticulocytes; EPO: erythropoietin.

haematologica | 2021; 106(2)

Letters to the Editor

(for whom endogenous EPO levels were not available). As shown in Figure 1A, an expected negative correlation was present between EPO and Hb levels (r=-0.64, P=0.0004). However, EPO levels in AIHA were reduced compared with expected values of controls with other types of anemia.10 Moreover, EPO levels were significantly reduced in transfusion dependent patients versus transfusion independent ones (30 U/L, range: 8-91, vs. 44 U/L, range: 10-1,328; P=0.05). Finally, bone marrow compensatory response (BMRI) negatively correlated with EPO levels (r=-0.42, P=0.03), and positively with Hb (r=0.28, P=0.05). Regarding the efficacy of rEPO treatment (Figure 1B), the overall response rate (ORR) was 55% at 15 days, 71% at month+1, 73% at month+3, 76% at month+6, and 78% at month+12 with a progressive increase of the CR rate. The median Hb and reticulocyte increase from baseline was 24 (range: 2-83) g/L (P<0.001) and 25 (range: 0-220) x109/L at month+1; 30 (range: 0-94) g/L (P<0.001) and 33 (range: 0-352) x109/L at month+3; and 42 (9-94) g/L (P=0.01) and 21 (range: 0-218) x109/L at month+6. Hb increased independently of the AIHA type and number of previous therapy lines (see the Online Supplementary Figure S1). Response rates were greater in patients who started rEPO within the first year from AIHA diagnosis compared with patients with a longer AIHA history (85% vs. 64%, P=0.06). Likewise, a better response was observed in primary versus secondary AIHA patients (77% vs. 44%), although the difference was not statistically significant. Concerning rEPO discontinuation and outcome, 23 patients were still on EPO and 28 had discontinued treatment at the last follow-up. Reasons for discontinuation were long standing CR (n=14), subopti-

mal response (n=12, Hb increase <20 g/L or Hb<100 g/L), and AIHA relapse (n=2). During treatment, two patients experienced a thrombotic event (one thrombosis of a peripheral inserted central vein catheter and one pulmonary embolism concomitant to AIHA relapse in a splenectomized patient). The occurrence of thrombotic episodes is a concern in AIHA, being observed in up to 20% of patients,1-4,11 and associated with intravascular hemolysis, complement activation, and previous splenectomy.1-4 On the other hand, rEPO treatment has been associated with thrombotic diathesis although in our patients, other risk factors were also present. In summary, EPO therapy was able to increase Hb levels (median Hb increase greater than 2 g/dL) in more than 70% of patients, both frontline and in relapsed/refractory chronic disease. Most responses were complete and longlasting, allowing EPO discontinuation in about one third of responders. Almost all responses were observed between month+1 and +3, with more than half as soon as day+15. EPO efficacy was higher in patients treated within the first year from diagnosis, suggesting that the duration of the immune-suppressive therapy is particularly detrimental for bone marrow responsiveness. Accordingly, a worse response was observed in secondary forms, where both underlying condition and cytotoxic therapy may have affected bone marrow function. Additional possible confounders of EPO efficacy may be concomitant treatments or recent therapy with rituximab. However, EPO was started because of persistent non-response to these agents and an early response promptly observed, suggesting a primary role of EPO stimulation.

haematologica | 2021; 106(2)

Figure 1. Relationship between endogenous erythropoietin and hemoglobin levels and response evaluation after recombinant erythropoietin treatment. (A) Continuous line shows the relationship in patients with autoimmune hemolytic anemia. Black circles represent patients responding to recombinant erythropoietin (rEPO) and white circles non responders; no significant differences were observed between these two groups. As controls, 49 aplastic anemia patients (white triangles) are shown, with the corresponding correlation (dotted-dashed line); dashed line represent patients with other types of anemia, including iron and vitamin deficiency [log(Epo)=4.478(0.284xHb)] [Bergamaschi et al., Haematologica 2008 Dec;93(12):1785-91]. (B) Overall response rate to rEPO at different time points. Patients were treated for a median of 7 months and received mainly epoetin α 40,000 U/week (n=17, 33%) or darbepoetin α 20-300 mcg/week (n=20, 39%); a minority of cases received epoetin z 30,000 U/week (n=6, 12%) or β 30,000 U/week (N=1, 2%). Two patients received epoetin α 4,000 U/week because of co-existent chronic kidney disease. CR: complete response: PR: partial response; NR: no response).

623

Letters to the Editor

Figure 2. Cytokine serum levels in patients with autoimmune hemolytic anemia and relationship with hematologic parameters and endogenous erythropoietin levels. Upper panel: serum cytokine individual values of autoimmune hemolytic anemia (AIHA) patients before therapy with recombinant erythropoietin (rEPO). Grey areas indicate mean+1 standard deviation of 40 age and sex matched healthy controls. TNF-α level was lower in AIHA than in controls (P<0.001), whereas IL10, IL6, IL17, and TGF-β levels were all higher (P<0.001, P<0.001, P=0.014, and P=0.002, respectively); IFN-γ level was comparable between patients and controls. Cytokines were evaluated using commercialeEnzyme-linked immunosorbent assay kits. Lower panel: correlation between endogenous EPO, bone marrow reticulocytes index (BMRI), hemoglobin (Hb) and cytokines. TNF-α positively correlated with endogenous EPO (r=0.77, P=0.005), and negatively with BMRI (r=-0.052, P=0.05). Moreover, IL6 and IL17 positively correlated with BMRI (r=0.45, not significant; r=0.53, P=0.04, respectively). Similar correlations were observed for reticulocytes (TNF-α: r=-0.051, P=0.05; IL6: r=0.53, P=0.04; IL17: r=0.61, P=0.02). Finally, a negative correlation was observed between TGFβ and Hb values (r=-0.63, P=0.04). Dashed line: negative correlation; continuous line: positive correlation.

For the first time, we showed that EPO levels were inappropriately low compared to hemoglobin levels in the majority of patients. Even if reduced, endogenous EPO maintained a negative correlation with Hb values, reticulocyte counts and bone marrow responsiveness index. As expected, patients who were transfused showed lower levels of endogenous EPO, possibly indicating that the feed-back loop (anemia- hypoxia – EPO production) is intact in AIHA. Interestingly, patients with AIHA showed significantly reduced EPO values compared to other types of anemia. These findings are similar to what has been described for immune thrombocytopenia, where low or inappropriately normal levels of endogenous thrombopoietin have been demonstrated12 and have led the way to TPO-receptor agonists use. The question is why AIHA patients show reduced EPO levels compared to other anemias. It may be speculated that reticulocyte response to hemolysis, by reversing the relative hypoxia (of the anemic state), may give a negative feed-back to the kidney. This may result in inhibition of EPO production, similar to what has been observed in transfused patients. Another possible mechanism may be due to the quick instauration of anemia in AIHA where massive erythrocyte destruction may occur in a few 624

hours, whilst bone marrow compensation requires more time.1-4 Finally, the existence of a “stunned bone marrow” unable to build up a prompt response to anemia as observed in patients with septic state13 may be hypothesized. In this setting, a temporary EPO stimulation may be preferred to additional immunosuppression. In a fraction of cases, serum levels of TNF-α, interleukin 10 (IL10), IL6, IL17, TGF-β, and IFN-α were evaluated (Figure 2) and showed a different pattern in patients versus matched controls. As already reported, a shift towards T-helper 2 (Th2) and T-helper 17 phenotype was observed, with reduced TNF-α and increased levels of IL6, IL10, TGF-β, and IL17, consistently with a prevalent humoral autoimmune response. Despite the limited number of cases and the known variability of cytokine levels, we found interesting correlations with endogenous erythropoietin levels. In particular, TNF-α, a known negative regulator of erythropoiesis, positively correlated with EPO levels and negatively with reticulocyte response. The Th2 cytokines IL6 and IL17 were positively related to reticulocytosis mirroring the degree of autoimmune attack. Finally, TGF-β has a negative impact on Hb levels, confirming its well-known inhibitory and detrimental effect.14 All these cytokine abnormalities may haematologica | 2021; 106(2)

Letters to the Editor

play a role in the inflammatory milieu and bone marrow dysregulation of AIHA. An interesting hypothesis is that an autoimmune attack may also occur against bone marrow precursors, resulting in peripheral reticulocytopenia and accounting for the severity of AIHA.6,15 In this context, therapy with recombinant EPO may interrupt the vicious circle, by priming the reticulocyte compensation, sustaining erythropoiesis and leaving time for the resolution of the autoimmune flare avoiding excessive immune-suppression. Bruno Fattizzo,1 Marc Michel,2,3 Anna Zaninoni,1 Juri A. Giannotta,1 Stephanie Guillet,2,3 Henrik Frederiksen,4 Josephine M.I. Vos,5 Francesca R. Mauro,6 Bernd Jilma,7 Andrea Patriarca,8 Francesco Zaja,9 Anita Hill,10 Sigbjørn Berentsen11 and Wilma Barcellini1 1 Hematology, Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico di Milano and University of Milan, Milan, Italy; 2Department of Internal Medecine, Henri Mondor University Hospital, Assistance Publique Hôpitaux de Paris, Creteil, France; 3 Internal Medecine, National Reference Center for Adult Immune Cytopenias, Henri Mondor University Hospital, Université ParisEst Créteil, France; 4Department of Haematology, Odense University Hospital, Odense, Denmark; 5Amsterdam UMC, Location AMC, Department of Hematology, Amsterdam, the Netherlands; 6Hematology, Department of Translational and Precision Medicine, Sapienza University, Rome, Italy; 7 Department of Clinical Pharmacology, Medical University of Vienna, Vienna, Austria; 8UO Ematologia, AOU Maggiore della Carità, Novara, Italy; 9S.C. Ematologia, Azienda Sanitaria Universitaria Integrata, Trieste, ITA; 10Hematology, Leeds Teaching Hospitals, Leeds, UK and 11Department of Research and Innovation, Haugesund Hospital, Haugesund, Norway Correspondence: WILMA BARCELLINI - wilma.barcellini@policlinico.mi.it doi:10.3324/haematol.2020.250522 Disclosures: all authors declare that they have no financial nor personal relationships with other people or organisations that could inappropriately influence (bias) their work. Contributions: B F and W B designed the study, followed patients, collected and analyzed data, wrote the paper and revised the paper for important intellectual content. MM, JG, SG, HF, JMIV, FRM, BJ, AP, FZ, AH, and SB followed patients, collected data, and revised the paper for important intellectual content. AZ performed cytokines studies and revised the paper for important intellectual content.

haematologica | 2021; 106(2)

References 1. Brodsky RA.Warm Autoimmune hemolytic anemia. N Engl J Med. 2019; 381(7):647-654. 2. Berentsen S. How I manage patients with cold agglutinin disease. Br J Haematol. 2018;181(3):320-330. 3. Barcellini W, Fattizzo B, Zaninoni A, et al. Clinical heterogeneity and predictors of outcome in primary autoimmune hemolytic anemia: a GIMEMA study of 308 patients. Blood. 2014;124(19):2930-2936. 4. Barcellini W, Zaninoni A, Fattizzo B, et al. Predictors of refractoriness to therapy and healthcare resource utilization in 378 patients with primary autoimmune hemolytic anemia from 8 Italian Reference Centers. Am J Hematol. 2018;93(9):E243-E-246. 5. Aladjidi N, Leverger G, Leblanc T, et al. New insights into childhood autoimmune hemolytic anemia: a French national observational study of 265 children. Haematologica. 2011;96(5):655-663. 6. Fattizzo B, Zaninoni A, Gianelli U, et al. Prognostic impact of bone marrow fibrosis and dyserythropoiesis in autoimmune hemolytic anemia. Am J Hematol. 2017;93(4):E88-E91. 7. Barcellini W. The relationship between idiopathic cytopenias/dysplasias of uncertain significance (ICUS/IDUS) and autoimmunity. Expert Rev Hematol. 2017;10(7):649-657. 8. Arbach O, Funck R, Seibt F, Salama A. Erythropoietin may improve anemia in patients with autoimmune hemolytic anemia associated with reticulocytopenia. Transfus Med Hemother. 2012;39(3):221-223. 9. Salama A, Hartnack D, Lindemann HW, Lange HJ, Rummel M, Loew A.The effect of erythropoiesis-stimulating agents in patients with therapy-refractory autoimmune hemolytic anemia. Transfus Med Hemother. 2014;41(6):462-468. 10. Bergamaschi G, Markopoulos K, Albertini R, et al. Anemia Of chronic Disease And Defective Erythropoietin Production In Patients With Celiac Disease. Haematologica. 2008;93(12):1785-1791. 11. Yusuf HR, Hooper WC, Grosse SD, Parker CS, Boulet SL, Ortel TL. Risk of venous thromboembolism occurrence among adults with selected autoimmune diseases: a study among a U.S. cohort of commercial insurance enrollees.Thromb Res. 2015;135(1):50-57. 12. Kosugi S, Kurata Y, Tomiyama Y, et al. Circulating thrombopoietin level in chronic immune thrombocytopenic purpura. Br J Haematol. 1996;93(3):704-706. 13. Cho JN, Avera S, Iyamu K. Pancytopenia as a consequence of sepsis and intravenous antibiotic drug toxicity. Cureus. 2019;11(2):e3994. 14. Barcellini W, Zaja F, Zaninoni A, et al. Low-dose rituximab in adult patients with idiopathic autoimmune hemolytic anemia: clinical efficacy and biologic studies. Blood. 2012;119(16):3691-3697. 15. Barcellini W. New insights in the pathogenesis of autoimmune hemolytic anemia. Transfus Med Hemother. 2015 ;42(5):287-293.

625

Letters to the Editor

Heme induces human and mouse platelet activation through C-type-lectin-like receptor-2 Intravascular hemolysis is a serious complication developed in many severe infections such as malaria, sepsis,1 typical hemolytic uremic syndrome2 and congenital diseases such as sickle cell disease.3 Hemolysis is accompanied by thrombocytopenia, platelet activation, plateletleukocyte aggregates, inflammation, vascular obstruction and organ damage. During hemolysis, red blood cell destruction leads to the release of molecules such as ADP, hemoglobin and heme which exert potent prothrombotic and proinflammatory actions. Under physiological conditions, free heme is scavenged by the plasma protein hemopexin, and is subsequently catabolized by heme oxygenase-1 into carbon monoxide, biliverdin and ferrous iron (Fe2+). However, acute or/and chronic hemolysis exhausts the scavenging system leading to an increase in free heme in the blood. Upon release, reduced heme is rapidly and spontaneously oxidized in the blood into ferric (Fe3+) form, hemin, with increased levels observed in hemolytic diseases. It was recently shown that hemin induces human platelet activation and platelet death through ferroptosis, an intracellular iron-mediated cell death.4 However, the mechanisms and receptors involved in platelet activation by hemin are not known. In this study, we investigated hemin-mediated human and mouse platelet activation in vitro. An increase in labile (i.e., weakly bound and free) heme/hemin is detected in patients with hemolytic diseases with plasma concentrations ranging between 2 mM and 50 mM.5,6 Similar concentrations of heme are detected in mice with chronic or acute hemolysis such as sickle cell mice or Escherichia coli-induced hemolysis.1,7 In this study, we show that hemin (2-12 mM) induces rapid aggregation of human washed platelets, while aggregation is slower and reduced in magnitude at higher (≥ 25 mM) concentrations (Figure 1A, B). Platelet aggregation was associated with increased P-selectin expression and GPIIbIIIa activation as measured by flow cytometry (Figure 1C, D). High (≥25 μM) but not lower concentrations of hemin significantly increased phosphatidylserine (PS) exposure on platelets as assessed using Annexin V staining (Figure 1E). Moreover, high concentrations of hemin induced toxicity as measured by increased levels of lactate dehydrogenase (LDH) in the supernatant of activated platelets (Figure 1F). Recent data have shown that a high concentration of hemin (25 mM) triggers lipid peroxidation and human platelet death through ferroptosis but not apoptosis or necroptosis.4 Whether the increase in PS exposure modulates the contribution of platelets to the coagulopathies observed in hemolytic diseases requires further investigation. Hemin is known to bind to Toll-like receptor (TLR) TLR4 on endothelial cells resulting in endothelial cell activation including secretion of Weibel-Palade bodies.3 However, blocking TLR4 signaling in platelets using TAK-242 did not alter activation by hemin, suggesting a TLR-4-independent pathway of activation (Figure 1H). At low concentrations, human platelet aggregation by hemin was blocked by inhibitors of Syk (PRT-060318), Src family kinases (PP2) and Btk (Ibrutinib), and partially by GPIIbIIIa inhibitor eptifibatide (Integrilin) (Figure 1G, H). These results suggest that hemin induces human platelet activation via an immunoreceptor-tyrosine-based activation-motif (ITAM) receptor-based pathway.8 Indeed, low concentrations of hemin provoked phospho626

rylation of Syk and PLCγ2 (Figure 1I). In order to identify the receptor triggering platelet activation by hemin, we used recombinant dimeric forms of C-type-lectin-like receptor-2 (CLEC-2) and collagen receptor glycoprotein VI (GPVI) (hFc-CLEC-2 and hFc-GPVI, respectively) as a blocking strategy. Pre-incubation of hemin with hFcCLEC-2 but not hFc-GPVI abolished platelet aggregation by hemin identifying CLEC-2 as the receptor mediating human platelet activation by hemin (Figure 1G, H). These results also suggest a direct interaction between hemin and CLEC-2. However, while we believe that the inhibition is likely the result of competition between recombinant and platelet CLEC-2, we cannot rule out that recombinant CLEC-2-hemin complex may interfere with CLEC-2 clustering thereby inhibiting platelet activation. Platelet aggregation was not altered by AYP1, 9, an antibody that blocks the interaction between CLEC-2 and podoplanin, providing indirect evidence that hemin binds to a distinct site on CLEC-2 to podoplanin (Figure 1H). Protoporphyrin IX, the precursor of heme, and cobalt hematoporphyrin were shown to bind to CLEC-2 and inhibit podoplanin-CLEC-2 interaction without inducing platelet activation.10 The type of interactions by which different porphyrins selectively bind to CLEC-2 and induce platelet activation requires further investigation. Platelet aggregation mediated by low concentrations of hemin (6.25 mM) was not altered by cyclooxygenase (COX) (indomethacin) or P2Y12 (cangrelor) inhibitors, showing that, in contrast to podoplanin, secondary mediators are not required for hemin-mediated platelet activation (Figure 1G, H). This may reflect the ability of hemin to induce higher oligomers of CLEC-2 increasing signal transduction. Platelet aggregation does not depend on oxidative stress as hemin-mediated platelet aggregation was not altered by the antioxidant N-acetyl cysteine (NAC) (Figure 1H; Online Supplementary Figure S1). At higher concentrations (>25 mM), platelet aggregation was not inhibited by integrilin or inhibitors of Src, Syk and Btk, suggesting that higher concentrations of hemin induce agglutination (Online Supplementary Figure S2). Increasing the concentrations of inhibitors did not alter platelet agglutination/aggregation (not shown). The distinct mechanisms of platelet activation by low and high concentrations of hemin might be related to formation of hemin aggregates at high concentrations, which might exert different activities as compared to monomeric or dimeric hemin present at low concentrations. Hemin induces mouse platelet aggregation, albeit at a slightly higher concentration (Figure 2A). Similar to human platelets, aggregation by low concentrations of hemin was inhibited by inhibitors of Src, Syk and Btk tyrosine kinases, recombinant mouse Fc-mCLEC-2 and GPIIbIIIa blockade (Figure 2B, C). In contrast, inhibitors of TLR4, COX and P2Y12 had no effect on platelet aggregation by hemin (Figure 2C). Platelet aggregation by hemin was significantly reduced in CLEC-2-deficient platelets confirming a key role for CLEC-2 (Figure 2D, E). However, deletion of CLEC-2 did not inhibit platelet shape change suggesting that one or more other receptors support platelet activation by hemin in mice. The fact that recombinant CLEC-2 inhibits heminmediated platelet aggregation suggests a direct interaction between hemin and CLEC-2. Using surface plasmon resonance-based technique, we demonstrated that hemin binds to both murine and human recombinant CLEC-2 with KD values of ~200 nM in both cases (Figure 3A, C). Hemin binding to mouse CLEC-2 was characterized by kinetic rate association (ka) and dissociation (kd) constants of ka=2.77±0.08×104 mol-1 s-1 and 5.76±0.09×10-3 s-1. haematologica | 2021; 106(2)

Letters to the Editor

Similar values of kinetic rate constants were determined for hemin binding to human CLEC-2, ka=2.65±0.09×104 mol-1 s-1 and kd=5.06±0.05×10-3 s-1. The interaction of hemin with human and mouse CLEC-2 was further confirmed by UV-vis absorbance spectroscopy (Figure 3B, D). The differential absorbance spectra profiles revealed significant shifts of the spectra of hemin towards higher wavelength by human and mouse CLEC-2 demonstrating a direct binding. The differences in the spectral

changes of hemin in the presence of human and mouse CLEC-2 can be explained by distinct residues coordinating heme’s iron. The blue shift in the Soret region observed with mCLEC-2 suggests the involvement of sulfur containing amino acid (cysteine)11 whereas the red shift observed with human CLEC-2 suggests histidine coordination. Altogether, these results show that hemin is an endogenous agonist for CLEC-2 leading to platelet activa-

Figure 1. Hemin mediates human platelet activation through C-type-lectin-like receptor-2. (A, B) Human washed platelets (2x108/mL) were incubated with increasing concentrations of hemin in the presence of Ca2+ (2 mM). Platelet aggregation was assessed for 6 minutes (min) using light transmission aggregometry (n=5). (C, D) For flow cytometry analysis, 1x106 platelets were incubated with different concentrations of hemin for 20 min at 37ºC. Platelet activation was determined by flow cytometry using (C) anti-P-selectin and (D) GPIIbIIIa PAC-1 antibodies (n=4). Data is shown as fold increase of median of fluorescence (MFI) of treated platelets over control (absence of hemin). (E) Phosphatidylserine exposure was assessed using Annexin V staining. (F) Lactate dehydrogenase (LDH) levels were measured in the supernatant following platelet aggregation (6 min). Platelet lysate was used to detect the total level of LDH in platelets. (G, H) Platelet aggregation by hemin was assessed using Btk inhibitor Ibrutinib (500 nM), Src family kinase inhibitor PP2 (20 mM), Syk inhibitor PRT-060318 (20 mM), TLR4 inhibitor TAK-242 (10 mM), P2Y12 inhibitor Cangrelor (10 mM) or cyclooxygenase inhibitor indomethacin (10 mM). Platelets were pre-incubated with different inhibitors for 5 min at 37 ºC prior to the addition of hemin. Dimeric C-type-lectin-like receptor-2 (CLEC-2) (hFc-CLEC-2, 20 mg/mL) or dimeric collagen receptor glycoprotein VI (GPVI) (hFc-GPVI, 20 mg/mL) were pre-incubated with hemin for 20 min at 37 ºC prior to addition to platelets (n=5). (H) Histogram data are shown as mean ± standard deviation. (I) Protein levels of the phosphorylation of Syk and PLCγ2, and total Syk and PLCγ2 in platelet lysates were determined by western blotting after 6 min aggregation. Western blots are representative of five independent experiments. The statistical significance was analyzed using a one-way ANOVA with Tukey’s multiple comparisons test using Prism 8 (GraphPad Software Inc, USA). Significance is shown compared to control (absence of hemin), *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

haematologica | 2021; 106(2)

627

Letters to the Editor

Figure 2. Hemin induces mouse platelet aggregation. (A) Mouse washed platelets (2x108/mL) were incubated with increasing concentrations of hemin in the presence of Ca2+ (2 mM). Platelet aggregation was assessed using light transmission aggregometry. (B, C) Hemin-mediated platelet aggregation (12.5 mM hemin) was assessed in the presence of Ibrutinib (2 mM), PP2 (20 mM), PRT-060318 (20 mM), TAK-242 (10 μM), Cangrelor (10 mM), recombinant mouse C-typelectin-like receptor-2 (CLEC-2) (Fc-mCLEC-2, 10 mg/mL) (n=4). (D, E) Washed platelets from wild-type (WT) or Clec-2 deficient (Clec-1bfl/fl PF4cre, Clec-2 KO) were incubated with hemin (12.5 mM) in the presence of Ca2+ (2 mM) (n=5). (E) Histogram data are shown as mean ± standard deviation. The statistical significance was analyzed using a one-way ANOVA with Tukey’s multiple comparisons test using Prism 8. Significance is shown compared to control ****P<0.0001.

628

Figure 3. Hemin binds to mouse and human recombinant C-typelectin-like receptor-2. (A, C) For surface plasma resonance, realtime interaction profiles were obtained after injection of increasing concentrations of hemin (19.5–625 nM) over recombinant mouse (A) and human (C) C-type-lectin-like receptor-2 (CLEC-2) immobilized on CM5 sensor chip. The association and dissociation phases were followed for 5 minutes, each. The black lines show the experimental data, the grey lines depict the fit obtained using Langmuir global analyses model. (B, D) For UV-vis absorbance spectroscopy, differential spectra were generated after titration of (B) mouse or (D) human dimeric CLEC-2 (2 µM) with increasing concentrations of hemin (0.5–16 mM). The differential spectra were obtained after subtraction of the spectra of hemin at a given concentration from the spectra of the same concentration of hemin in the presence of the protein. The measurements were done at 25°C in optical cell with 10 mm light path.

haematologica | 2021; 106(2)

Letters to the Editor

tion through activation of integrin GPIIbIIIa at low concentrations and agglutination at high concentrations. Recombinant CLEC-2 inhibits platelet activation by both mechanisms suggesting that this may represent a novel form of therapeutic intervention to inhibit platelet activation in hemolytic disease. Ethical approval for collecting blood from healthy volunteers was granted by Birmingham University Internal Ethical Review (ERN_11-0175). Mice experiments were performed in accordance with UK laws (Animal [Scientific Procedures] Act 1986) with approval of the local ethical committee and UK Home Office approval (PPL P0e98D513). Human and mouse washed platelets were prepared as previously described.12,13 Platelet aggregation was assessed using light transmission aggregometry for 6 minutes (ChronoLog, Havertown, PA, USA). Fresh synthetic hemin solution was freshly prepared (Frontiers Scientific, USA) for each experiment. PP2 (Tocris/Biotechne Ltd), PRT-060318 (Caltag medsystems), indomethacin (Sigma-Aldrich), cangrelor (The Medicines Company), Ibrutinib (Stratech Scientific), NAcetyl-L-cysteine (Sigma-Aldrich) and TAK-242 (SigmaAldrich) were preincubated for 5 minutes (min) at 37ºC with platelets prior to addition of hemin. Human (h) and mouse (m) CLEC-2 and GPVI were produced as Fc fusion proteins (hFc-CLEC-2, hFc-GPVI, Fc-mCLEC-2) in human embryonic kidney HEK293 cells and purified using chromatography affinity. Recombinant human and mouse CLEC-2 were preincubated with hemin for 15 min at 37ºC before addition to platelets. Following aggregation, platelets were centrifuged at 1,000g for 10 min and LDH was measured in the supernatant using CyQUANT™ LDH Cytotoxicity Assay (ThermoFisher Scientific). Clec2 deficient mice (Clec-1bfl/fl PF4cre) and littermate controls were previously described.14 Platelet activation (106 platelets) by different concentrations of hemin (20 min at 37ºC) was assessed by flow cytometry with antibodies against CD62P (PE/Cy5 antihuman CD62P antibody, biolegend) and activated GPIIbIIIa (Alexa Fluor® 647 anti-human CD41/CD61 antibody PAC1, Biolegend) using BD Accuri C6 Plus flow cytometer (BD Bioscience). For western blotting, Phospho-Syk (Tyr525/526) (C87C1) Rabbit mAb (#2710), Phospho-PLCγ2 (Tyr759) antibody (#3874), Syk antibody (#2712) and PLCγ2 antibody (#3872) were used (Cell signalling). Protein phosphorylation was assessed after 6 min of aggregation as previously described.12 The binding of human and mouse dimeric CLEC-2 to hemin was performed at 25°C and assessed using surface plasma resonance and UV-spectroscopy methods.15 Joshua H. Bourne,1* Martina Colicchia,1* Ying Di,1 Eleyna Martin,1 Alexander Slater,1 Lubka T. Roumenina,2 Jordan D. Dimitrov,2 Steve P. Watson,1,3 and Julie Rayes1,3 1 Institute of Cardiovascular Sciences, College of Medical and Dental Sciences, University of Birmingham, Birmingham, UK; 2 Centre de Recherche des Cordeliers, INSERM, Sorbonne Université, USPC, Université Paris Descartes, Université Paris Diderot, Paris, France and 3Center of Membrane Proteins and Receptors (COMPARE), Universities of Birmingham and Nottingham, The Midlands, UK *JHB and MC contributed equally as co-first authors

haematologica | 2021; 106(2)

Correspondence: JULIE RAYES - j.rayes@bham.ac.uk doi:10.3324/haematol.2020.246488 Disclosures: no conflicts of interests to disclose. Contributions: JHB, MC performed experiments and analyzed data; YD, AS, EM generated reagents; LTR contributed to data interpretation; JDD contributed to research design, performed experiments and interpreted data; SPW contributed to research and data interpretation and provided reagents; JR designed and performed research, collected data, analyzed and interpreted data and wrote the manuscript. All authors reviewed and approved the manuscript. Acknowledgments: the authors would like to thank Dr B. Grygielska for genotyping of mice. Funding: this work was supported by a BHF Accelerator Award (AA/18/2/34218), BHF programme grant (RG/13/18/30563), Wellcome Trust 4 year studentship (204951), Wellcome Trust Joint Investigator Award (204951/Z/16/Z) and the Centre of Membrane Proteins and Receptors. JDD hold an H2020 ERC starting grant (CoBABATI 678905). SPW holds a BHF Chair (CH/03/003).

References 1. Martins R, Maier J, Gorki AD, et al. Heme drives hemolysis-induced susceptibility to infection via disruption of phagocyte functions. Nat Immunol. 2016;17(12):1361-1372. 2. Frimat M, Tabarin F, Dimitrov JD, et al. Complement activation by heme as a secondary hit for atypical hemolytic uremic syndrome. Blood. 2013; 122(2):282-292. 3. Belcher JD, Chen C, Nguyen J, et al. Heme triggers TLR4 signaling leading to endothelial cell activation and vaso-occlusion in murine sickle cell disease. Blood. 2014;123(3):377-390. 4. NaveenKumar SK, SharathBabu BN, Hemshekhar M, Kemparaju K, Girish KS, Mugesh G. The Role of Reactive Oxygen Species and Ferroptosis in Heme-Mediated Activation of Human Platelets. ACS Chem Biol. 2018;13(8):1996-2002. 5. Muller-Eberhard U, Javid J, Liem HH, Hanstein A, Hanna M. Plasma concentrations of hemopexin, haptoglobin and heme in patients with various hemolytic diseases. Blood. 1968;32(5):811-815. 6. Gouveia Z, Carlos AR, Yuan X, et al. Characterization of plasma labile heme in hemolytic conditions. FEBS J. 2017;284(19):3278-3301. 7. Belcher JD, Chen C, Nguyen J, et al. Haptoglobin and hemopexin inhibit vaso-occlusion and inflammation in murine sickle cell disease: Role of heme oxygenase-1 induction. PLoS One. 2018; 13(4):e0196455. 8. Rayes J, Watson SP, Nieswandt B. Functional significance of the platelet immune receptors GPVI and CLEC-2. J Clin Invest. 2019;129(1):12-23. 9. Gitz E, Pollitt AY, Gitz-Francois JJ, et al. CLEC-2 expression is maintained on activated platelets and on platelet microparticles. Blood. 2014; 124(14):2262-2270. 10. Tsukiji N, Osada M, Sasaki T, et al. Cobalt hematoporphyrin inhibits CLEC-2-podoplanin interaction, tumor metastasis, and arterial/venous thrombosis in mice. Blood Adv. 2018;2(17):2214-2225. 11. Kuhl T, Wissbrock A, Goradia N, et al. Analysis of Fe(III) heme binding to cysteine-containing heme-regulatory motifs in proteins. ACS Chem Biol. 2013;8(8):1785-1793. 12. Nicolson PLR, Hughes CE, Watson S, et al. Inhibition of Btk by Btk-specific concentrations of ibrutinib and acalabrutinib delays but does not block platelet aggregation mediated by glycoprotein VI. Haematologica. 2018;103(12):2097-2108. 13. Hughes CE, Pollitt AY, Mori J, et al. CLEC-2 activates Syk through dimerization. Blood. 2010;115(14):2947-2955. 14. Finney BA, Schweighoffer E, Navarro-Nunez L, et al. CLEC-2 and Syk in the megakaryocytic/platelet lineage are essential for development. Blood. 2012;119(7):1747-1756. 15. Wiatr M, Merle NS, Boudhabhay I, et al. Anti-inflammatory activity of intravenous immunoglobulin through scavenging of heme. Mol Immunol. 2019;111:205-208.

629

Letters to the Editor

The clinical and biological characteristics of NUP98-KDM5A pediatric acute myeloid leukemia Pediatric acute myeloid leukemia (AML) is a rare, heterogeneous disease, characterized by recurrent cytogenetic and molecular aberrations.1,2 Genetic aberrations are the most important prognostic factor, besides early response to treatment.3,4 In pediatric AML, improvement of clinical outcome has reached a plateau, despite intensive chemotherapy and allogeneic hematopoietic stem cell transplantation (HSCT) in selected cases, leading to event-free survival rates of approximately 55-60% in the current era.1,2,5 Hence, identification of prognostic subgroups is important for risk-group stratification.1,5 Furthermore, novel therapeutic options should be explored for pediatric AML, with the objective of improving survival especially for patients with high-risk subtypes, such as those with the NUP98-NSD1 rearrangement (NUP98-KDM5A+).6,7

Here, we present the results of a collaborative study between the Children’s Oncology Group (COG) and European AML study groups, aimed to define the biological and clinical characteristics of AML patients carrying NUP98-KDM5A outside acute megakaryoblastic leukemia. The methodology and analysis are described in detail in the Online Supplementary Methods. In total 2,393 patients were included from various trials by the COG, AIEOP (Associazione Italiana di Ematologia e Oncologia Pediatrica), BFM (Berlin-Frankfurt-Münster) group, CPH (Czech Pediatric Hematology Working Group), DCOG (Dutch Childhood Oncology Group) and LAME (Leucémie Aiquë Myéloblastique Enfant). NUP98KDM5A rearrangements were detected in 47/2,393 pediatric AML cases (2.0%) with similar prevalence in the COG and European cohorts (Table 1). The median age of the NUP98-KDM5A+ patients was significantly lower than that of patients without this fusion gene (3.2 vs. 9.4 years; P=0.001) (Table 1, Online Supplementary Figure S1A). The median white blood cell count of NUP98-

Table 1. Clinical characteristics of pediatric patients with or without NUP98-KDM5A rearrangements.

Characteristic Total, n. Group COG (03p1/0531/1031), n (%) European, n (%) Gender Male, n (%) Female, n (%) Unknown, n Median age at diagnosis, years (range) Unknown, n Median WBC x109/L (range) FAB type, n (%) M0 M1 M2 M4 M5 M6 M7 Other Not otherwise specified Treatment response CR obtained, n (%) Refractory disease, n (%) Unknown, n HSCT in CR1 5-year survival, % (2SE) Event-free survival Overall survival Relapse risk Relapse-free survival Transplant-related mortality

NUP98-KDM5A+

NUP98-KDM5A –

2346

36 (77) 11 (23)

2003 (85) 343 (15)

0.093

24 (53) 21 (47) 2 3.2 (0.07-18.5) 2 11.7 (1.8-237.3)

1215 (52) 1114 (48) 17 9.4 (0-29.8) 13 23.9 (0.2-2730)

0.877

1 (3) 2 (7) 1 (3) 2 (7) 6 (21) 5 (17) 10 (34) 2 (7) 18*

41 (3) 161 (13) 280 (22) 317 (25) 275 (22) 16 (1) 99 (8) 87 (7) 1070*

0.617 0.568 0.017 0.026 0.911 <0.001 <0.001 1.000

41 (91) 4 (9) 2 7 (15)

2011 (90) 224 (10) 111 332 (14)

1.000

29.6 (14.6) 34.1 (16.1) 62.6 (16.7) 32.5 (15.7) 4.9 (6.8)

47.0 (2.1) 63.7 (2.1) 42.5 (2.3) 51.6 (2.1) 5.9 (1.1)

0.005 <0.001 0.002 0.001 0.913

0.001 0.006

0.885

*The AAML1031 study did not collect data on FAB types; however, we were able to determine acute megakaryoblastic leukemia status for AAML1031 patients according to the World Health Organization criteria. COG: Children’s Oncology Group; WBC: white blood cell count; FAB: French American British morphology classification; CR: complete remission; HSCT: hematopoietic stem cell transplant; CR1: first complete remission; SE, standard error.

630

haematologica | 2021; 106(2)

Letters to the Editor

Figure 1. Survival and gene expression of NUP98-KDM5A+ and NUP98-KDM5- acute myeloid leukemia. (A) Kaplan-Meier survival curve of event-free survival (EFS) of NUP98-KDM5A+ versus NUP98-KDM5– patients. (B) Kaplan-Meier survival curve of overall survival (OS) of NUP98-KDM5A+ versus NUP98-KDM5A– patients. (C) Relapse risk (RR) of NUP98-KDM5A+ versus NUP98-KDM5A– patients. (D) Unsupervised hierarchical clustering analysis by pairwise sample correlations (Pearson R). (E) HOX expression-based clustering using principal component analysis. The five distinct groups were determined using K-means clustering and depicted in convex hulls. (F) Venn diagram of differentially expressed genes in NUP98-KDM5A+ and NUP98-NSD1– cases as compared to other subtypes of acute myeloid leukemia (excluding those with NUP98 rearrangements).

KDM5A+ patients was significantly lower than that of NUP98-KDM5A− patients (11.7x109/L vs. 23.9x109/L, P=0.006). Previously described as a recurrent rearrangement in acute megakaryoblastic leukemia, this study identified NUP98-KDM5A in all French-American-British (FAB) types of AML, with the rearrangement being present most frequently in M7 (34%), M5 (21%) and M6 (17%) (Online Supplementary Figure S1B).8 Interestingly, these characteristics were different from those previously described for NUP98-NSD1-rearranged pediatric AML (n=37), in which patients had a median age of 10.4 years (range, 1.2-19.4), a median white blood cell count of 181.2x109/L, and the rearrangement was most frequent in FAB types M4/M5 (51%).7,9 In concordance with other studies, we found that NUP98-KDM5A+ patients lacked other common AML fusions.8,10 Mutations in genes that are recurrently mutated in other AML subtypes, such as RAS, WT1 and FLT3, occurred with very low frequency in NUP98-KDM5A+ cases, suggesting that the fusion itself may have a sufficient transforming effect (Online Supplementary Table S1). There was not a significant difference in complete remission rates between NUP98-KDM5A+ and NUP98KDM5A− patients (Table 1). Minimal residual disease (MRD) data were available for 31 NUP98-KDM5A+ patients; 17/31 (55%) NUP98-KDM5A+ patients were haematologica | 2021; 106(2)

MRD+ (>10-3 blasts detected) at the end of induction, compared to 471/1740 NUP98-KDM5A− patients (27%, P<0.001). MRD status did not clearly influence outcome, as survival rates of NUP98-KDM5A+ patients were similar between MRD+ patients (4-year event-free survival of 27% ± 25.4% and overall survival of 29.9% ± 31.7%), and MRD– patients (4-year event-free survival of 36% ± 25.6% [P=0.30] and overall survival of 42.9% ± 26.5% [P=0.65]) (Online Supplementary Figure S2A and B). These survival rates were comparable to those of NUP98KDM5A− but MRD+ patients (5-year event-free survival of 30.8% ± 3.4% and overall survival of 48.5% ± 3.7%), and significantly lower than survival rates of NUP98KDM5A− patients who were MRD– (5-year event-free survival of 58.3% ± 2.2% [P<0.001] and overall survival of 73.3% ± 2.0% [P<0.001]). However, as the numbers of NUP98-KDM5A+ patients with available MRD data were low, definitive conclusions cannot be drawn. Event-free and overall survival rates of NUP98KDM5A+ patients were nearly superimposable with 5-year survival rates of 29.6% ± 14.6%, and 34.1% ± 16.1%, respectively. This illustrates that NUP98KDM5A+ AML is more difficult to rescue than other AML subtypes. The relapse risk of NUP98-KDM5A+ patients was 62.6% ± 16.7%. These outcomes were significantly worse when compared to those of NUP98-KDM5A631

Letters to the Editor

patients who had a 5-year event-free survival of 47.0% ± 2.1% (P=0.005), overall survival of 63.7% ± 2.1% (P≤0.001), and a relapse risk of 42.5 ± 2.3 (P=0.002) (Figure 1A-C). Previously described NUP98-NSD1+ cases had similar outcomes to NUP98-KDM5A+ patients cases, with a 5-year event-free survival rate of 22.2% ± 6.9% and an overall survival rate of 45.7% ± 8.7%.7,9 Based on multivariable analysis, NUP98-KDM5A was an independent risk factor for poor overall survival and relapse, with hazard ratios of 1.77 for overall survival (P=0.009) and 1.62 for relapse risk (P=0.049), but not for event-free survival (hazard ratio 1.40, P=0.08) (Online Supplementary Table S2). Seven NUP98-KDM5A+ patients underwent HSCT in first complete remission. Among these patients, five relapsed and four subsequently died. Gene expression analysis using RNA-sequencing data from 1,035 patients, including 16 with NUP98-KDM5A+ and 49 with NUP98-NSD1+ , revealed 2,252 differentially

expressed genes between NUP98-KDM5A+ and NUP98KDM5A− cases, of which 1,542 were upregulated genes (Online Supplementary Table S3). However, hierarchical clustering did not cluster the majority of NUP98KDM5A+ cases together, but scattered them among samples with normal and other karyotypes (Figure 1D). The top upregulated genes included many HOXB genes (Online Supplementary Table S3). The upregulation of both HOXA and HOXB genes has been previously reported in NUP98-rearranged cases, and represents a common pathway to leukemia development, as it is also shared with NPM1-mutated AML, t(8;16)-rearranged AML, and others.7,11 HOXA/B-gene-based principle coordinate analysis, excluding patients with rare or unknown driving aberrations, revealed several subgroups (Figure 1E). NUP98-KDM5A clustered together with NPM1-mutated cases and DEK-NUP214, while NUP98-NSD1 clustered separately. The difference between NUP98-KDM5A and

Figure 2. Continued on next page. Legend on next page.

632

haematologica | 2021; 106(2)

Letters to the Editor

Figure 2 (continued from previous page). Top hits of the drug screening and validation of the most promising candidate drugs on primary NUP98-KDM5A+ and NUP98-NSD1- acute myeloid leukemia samples. (A) Heatmaps of the most effective hits from the drug library screens at 10 nM, 100 nM and 1,000 nM on primary samples from a case of NUP98-KDM5A+ acute myeloid leukemia (AML) and a case of NUP98-NSD1+ AML, ranked by difference in cell viability. Drugs occurring multiple times in the heatmap indicate that the drug was present in multiple screened drug libraries. (B-G) Dose-response curves of the selected candidate drugs on primary NUP98-KDM5A+ (n=2) and NUP98-NSD1+ (n=3) AML samples to the tubulin inhibitors vincristine (top) and fosbretabulin (bottom) (B), the PI3K inhibitor omipalisib (C), the MEK inhibitor trametinib (D), the BRD4 inhibitors OTX-015 (top) and ARV825 (bottom) (E), the CDK9 inhibitors dinaciclib (top) and CDKI-73 (bottom) (F), and the HSP90 inhibitor ganetespib (G). Data are based on a 4-day MTT assay and normalized to values in controls treated with dimethylsulfoxide.

NUP98-NSD1 was further underlined by comparing gene expression profiles with other AML cases. This revealed 2,176 differentially expressed genes in NUP98-KDM5A+ cases and 2,980 differentially expressed genes in NUP98NSD1+ cases (Figure 1F, Online Supplementary Tables S3 and S4). Among these differentially expressed genes, 810 were shared between the two groups: 68 genes were upregulated in both groups, 48 were downregulated and 694 had opposing expression profiles (Online Supplementary Table S5). Gene set enrichment analysis of the differentially expressed genes revealed upregulation of targets of E2F and FLT3 in both NUP98-rearranged subgroups (Online Supplementary Tables S6-S9). TP53 and HDAC targets were downregulated in both rearrangements. Interestingly, STAT5, NF1 and NOTCH1 pathways and targets were upregulated in NUP98-KDM5A cases and downregulated in NUP98-NSD1 cases, whereas the MYC pathway was upregulated in NUP98-NSD1 and downregulated in NUP98-KDM5A cases. Connectivity mapping using the 90th percentile absolute log foldchange of the differentially expressed genes indicated different potential targets for treatment of NUP98-KDM5A when compared to NUP98-NSD1 cases (Online Supplementary Figure S3). Both NUP98-KDM5A and NUP98-NSD1 cases had negative median tau scores for histone deacetylase inhibitors. Furthermore, NUP98KDM5A cases had a median tau score of -21.4 for microtubule inhibitors, indicating that these inhibitors could reverse the gene expression signature of the NUP98haematologica | 2021; 106(2)

rearranged patients. In search of treatment options, high-throughput screening of more than 4,000 compounds (Online Supplementary Table S10) was performed on primary samples from a patient with NUP98-KDM5A+ and a patient with NUP98-NSD1+ AML, and revealed an overall resistance profile to different drugs (Online Supplementary Figure S4). At a drug concentration of 1,000 nM, 146 unique compounds were identified that inhibited cell viability by >70% in one or both samples. At 100 nM, 41 compounds were found to inhibit cell viability by >60%, and at 10 nM eight drugs inhibited cell viability by >50% (Figure 2A). Etoposide and cytarabine, chemotherapeutics used in standard AML treatment, were not identified as top hits at the doses of 10 nM and 100 nM. A total of nine drugs, comprising six drug target classes, namely microtubule, PI3K, MEK, HSP90, CDK9 and BRD4 inhibitors, were selected for further validation on NUP98-KDM5A+ (n=2) and NUP98-NSD1+ (n=3) primary AML samples (Online Supplementary Table S11), as well as the CHRF-288-11 cell line harboring a NUP98-KDM5A fusion (Online Supplementary Figure S5). These validation experiments confirmed the effects observed in the highthroughput drug screening (Online Supplementary Figure S2B-G and S6). In concordance with the connectivity map analysis, tubulin inhibitors such as vincristine and fosbretabulin, decreased cell viability in vitro in NUP98-KDM5A+ cases, while showing limited cell toxicity in NUP98-NSD1+ cases (Online Supplementary Figure S7, Online 633

Letters to the Editor

Supplementary Table S12). In concordance with our pathway analysis, which showed upregulation of the STAT5 pathway in NUP98-KDM5A+ AML as compared to other types of AML and downregulation in NUP98-NSD1+ AML, the PI3K inhibitor omipalisib decreased cell viability in NUP98-KDM5A+ cases with a half maximal inhibitory concentration (IC50) of 1.6 nM - 2.2 nM), whereas it had little effect on NUP98-NSD1+ cases (IC50= 95.4 nM - 3000 nM) (Figure 2C). Trametinib, a MEK inhibitor with a manageable safety profile in pediatric patients, produced variable responses in both tested NUP98 fusions, but in all cases had an IC50 value of less than ~300 nM (Figure 2D)12. Both NUP98-fusion types responded to drugs targeting BRD4, CDK9 and HSP90, highlighting that the fusions do not solely have distinctive features but also common leukemia hallmarks (Figure 2E-G). Although these compounds produce promising in vitrro responses, implementation in pediatric AML first awaits testing in phase I/II trials in adults. Furthermore, in vivo drug testing is required as in vitro response does not always imply in vivo response.13 Overall, these data show the value of including screening for NUP98-KDM5A rearrangements as part of the standard survey in children with AML irrespectively of their AML FAB subtype. Although HSCT in first complete remission did not seem to prevent relapse in these cases, HSCT is currently the most effective post-remission therapy for preventing relapse. Therefore, we suggest that NUP98-KDM5A+ AML deserves stratification into the high-risk group, and that HSCT in first complete remission should be considered. We showed that NUP98-KDM5A+ and NUP98-NSD1+ cases of AML have different clinical and biological characteristics, and may benefit from different types of targeted treatment. Sanne Noort,1* Priscilla Wander,1,2* Todd A. Alonzo,3,4 Jenny Smith,5 Rhonda E. Ries,5 Robert B. Gerbing,3 M. Emmy M. Dolman,2 Franco Locatelli,6 Dirk Reinhardt,7 Andre Baruchel,8 Jan Stary,9 Jan J. Molenaar,2 Ronald W. Stam,2 Marry M. van den Heuvel-Eibrink,1,2 C. Michel Zwaan1,2,10 and Soheil Meshinchi5,11 1 Department of Pediatric Oncology/Hematology, Erasmus MC-Sophia Children’s Hospital Rotterdam, the Netherlands; 2 Princess Máxima Center for Pediatric Oncology, Utrecht, the Netherlands; 3Children's Oncology Group, Monrovia, CA, USA; 4 Keck School of Medicine, University of Southern California, Los Angeles, CA, USA; 5Fred Hutchinson Cancer Research Center, Seattle, WA, USA; 6IRCCS Ospedale Bambino Gesù, Sapienza, University of Rome, Rome, Italy; 7AML-BFM Study Group, Pediatric Hematology and Oncology, Essen, Germany; 8 Pediatric Hematology-Immunology Department, University Hospital Robert Debré and Paris Diderot University, Paris, France; 9Czech Pediatric Hematology/Oncology (CPH), University Hospital Motol and Charles University, Prague, Czech Republic; 10Dutch Childhood Oncology Group (DCOG), The Hague, the Netherlands and 11 Department of Pediatrics, Seattle Children's Hospital, University of Washington, Seattle, WA, USA. *SN and PW contributed equally as co-first authors. Correspondence: SOHEIL MESHINCHI - smeshinc@fhcrc.org doi:10.3324/haematol.2019.236745 Disclosures: no conflicts of interests to disclose.

634

Contributions: SN and PW contributed equally to this work; SN, PW, RWS, CMZ, MMvdH-E and SM designed the study; TAA., RBG, FL, DR, AB, JS, JJM, RWS and CMZ contributed materials and clinical data; PW, MEMD and SN performed the experiments; SN, PW, TAA, JS, RER, RBG, RWS, MMvdH-E, CMZ and SM analyzed data; SN, PW, TAA, JSRER and RBG performed the statistical analysis; SN, PW, RWS, CMZ, MMvdH-E and SM wrote the paper; SM, RWS, CMZ and MMvdH-E supervised the study; and all coauthors critically reviewed the manuscript and gave their final approval. Acknowledgments: the authors would like to thank Sandra Mimoso Pinhanços, Bianca Koopmans, Luke Jones, Susan Arentsen-Peters and Patricia Garrido Castro for their assistance in the in vitro drug screens and validation.

References 1. Zwaan CM, Kolb EA, Reinhardt D, et al. Collaborative efforts driving progress in pediatric acute myeloid leukemia. J Clin Oncol. 2015; 33(27):2949-2962. 2. Rubnitz JE, Gibson B, Smith FO. Acute myeloid leukemia. Hematol Oncol Clin North Am. 2010;24(1):35-63. 3. Balgobind BV, Hollink IH, Arentsen-Peters ST, et al. Integrative analysis of type-I and type-II aberrations underscores the genetic heterogeneity of pediatric acute myeloid leukemia. Haematologica. 2011;96(10):1478-1487. 4. Coenen EA, Raimondi SC, Harbott J, et al. Prognostic significance of additional cytogenetic aberrations in 733 de novo pediatric 11q23/MLL-rearranged AML patients: results of an international study. Blood. 2011;117(26):7102-7111. 5. de Rooij JD, Zwaan CM, van den Heuvel-Eibrink M. Pediatric AML: from biology to clinical management. J Clin Med. 2015;4(1):127149. 6. Niktoreh N, Walter C, Zimmermann M, et al. Mutated WT1, FLT3ITD, and NUP98-NSD1 fusion in various combinations define a poor prognostic group in pediatric acute myeloid leukemia. J Oncol. 2019;2019:1609128. 7. Hollink IH, van den Heuvel-Eibrink MM, Arentsen-Peters ST, et al. NUP98/NSD1 characterizes a novel poor prognostic group in acute myeloid leukemia with a distinct HOX gene expression pattern. Blood. 2011;118(13):3645-3656. 8. de Rooij JD, Hollink IH, Arentsen-Peters ST, et al. NUP98/JARID1A is a novel recurrent abnormality in pediatric acute megakaryoblastic leukemia with a distinct HOX gene expression pattern. Leukemia. 2013;27(12):2280-2288. 9. Bolouri H, Farrar JE, Triche T, Jr., et al. The molecular landscape of pediatric acute myeloid leukemia reveals recurrent structural alterations and age-specific mutational interactions. Nat Med. 2018;24(1):103-112. 10. Struski S, Lagarde S, Bories P, et al. NUP98 is rearranged in 3.8% of pediatric AML forming a clinical and molecular homogenous group with a poor prognosis. Leukemia. 2017;31(3):565-572. 11. Bisio V, Zampini M, Tregnago C, et al. NUP98-fusion transcripts characterize different biological entities within acute myeloid leukemia: a report from the AIEOP-AML group. Leukemia. 2017;31(4):974-977. 12. McCowage GB, Mueller S, Pratilas CA, et al. Trametinib in pediatric patients with neurofibromatosis type 1 (NF-1)–associated plexiform neurofibroma: A phase I/IIa study. J Clin Oncol. 2018;36(15_suppl): 10504-10504. 13. Aplenc R, Meshinchi S, Sung L, et al. The addition of bortezomib to standard chemotherapy for pediatric acute myeloid leukemia has increased toxicity without therapeutic benefit: a report from the Children's Oncology Group. Blood. 2016;128(22):899.

haematologica | 2021; 106(2)

Letters to the Editor

Down syndrome-related transient abnormal myelopoiesis is attributed to a specific erythro-megakaryocytic subpopulation with GATA1 mutation Down syndrome-related transient abnormal myelopoiesis (TAM) is a temporal preleukemic state presenting the marked elevation of blast cells at birth.1 Although somatic GATA1 mutations are known to be a cause of TAM, it is still unclear in which immature hematopoietic progenitor cell (HPC) subpopulation they most strongly evoke abnormal proliferation.2 Considering the different properties of hematopoietic cells from different species, it is necessary to establish an appropriate human model applicable for studying TAM and its embryonic definitive hematopoietic origin. Recent pluripotent stem cell (PSC)-based models have reported the recapitulation of TAM phenotypes and suggested the implication of a specific gene locus on chromosome 21 in GATA1 mutation-driven perturbated hematopoiesis.3-5 However, little is known about which progenitor cells the subpopulation with abnormal myelopoiesis are primarily associated with the abnormal myelopoiesis due to the lack of chronological analysis following hematopoietic differentiation. In order to overcome this point, we previously used our step-wise hematopoietic differentiation system to trace the progenitors from mesoderm to lineage-committed HPC via very immature multipotent progenitors.6,7 In this study, we first examined the GATA1-dependent hematopoietic property using two independent isogenic trisomy 21 (Ts21) PSC pairs (Figure 1A): one is a TAM patient-blast-derived induced PSC (iPSC) pair (Online Supplementary Figure S1)8 and the other is a human embryonic stem cells (ESC) pair with or without GATA1 mutation.9 We confirmed that the obtained genome-edited clones have no additional karyotype abnormality other than Ts21. No off-target mutations in the entire exonic sequence of GATA1 gene were observed by Sanger sequencing. GATA1-wild-type clones (WT-clones) expressed both full-length and short GATA1, while GATA1-mutated clones (G1s-clones) expressed only short GATA1. These G1s-clones showed abnormal hematopoiesis such as a decrease in CD71bright+CD42bCD235a+ erythroid and CD41+CD42b+CD235amegakaryocytic cells in step-wise hematopoietic differentiation (Online Supplementary Figure S2). On the other hand, CD34-CD235a-CD41-CD43+CD45+ myeloid-lineage cells from G1s-clones showed an increase not only in multi-lineage culture (Online Supplementary Figure 2AB) but also in myeloid-specific culture (data not shown). In order to identify the relevant HPC subpopulation inducing these differences, we traced the differentiation efficacy and expression profiles using an hematopoiesisfocused PCR array during step-wise hematopoiesis (Figure 1B; Online Supplementary Figure 3; Online Supplementary Table S1).10 In mesodermal and early hematopoietic differentiation, WT- and G1s-clones showed similar efficiencies in the patterns of phenotype transition. KDR+CD34+ hemoangiogenic progenitors were similarly observed in both clones on day 4, followed by CD34+CD43+ HPC on day 6 (Online Supplementary Figure S4). On the other hand, later phase progenitors on day 9 were significantly different between WTand G1s-clones. The frequency of CD34+CD43+CD235a- HPC was significantly higher in G1s-clones (Figure 1C). The day 9 HPC were further categorized into three subpopulations based on the expreshaematologica | 2021; 106(2)

sions of CD11b (a representative marker of myeloid cells), CD71 (erythroid and megakaryocytic progenitors), and CD41 (megakaryocytic progenitors) as CD34+CD43+CD235a- CD11b+CD41- (designated here as P-mye) (Figure 1D), CD34+CD43+CD235a-CD11bCD71+CD41- (P-erymk41(-)) (Figure 1E), and CD34+CD43+CD235a-CD11b-CD71+CD41+ (P-erymk41(+)) (Figure 1F), respectively. Regarding the P-mye frequency, no significant difference was observed between WT- and G1s-clones at this stage. On the other hand, the P-erymk41(-) and P-erymk41(+) frequencies showed a significant difference between WT- and G1sclones, but in opposite directions: P-erymk41(-) was increased in G1s-clones and P-erymk41(+) in WT-clones. These data suggested the presence of some pivotal pathways in these two immature subpopulations that caused significant differences in later development, that is, increased myeloid lineages and impaired megakaryocytic maturation in G1s-clones on day 16. Thus, our observations indicated that GATA1 mutation distinctly affected the properties of late-stage CD34+ P-erymk progenitors rather than earlier-stage CD34+ progenitors. In order to narrow down the responsible factors and their target subpopulations in day 9 CD34+CD43+CD235a- HPC, we performed correlation analysis using the first approximation to compare each subpopulation frequency on day 9 and the resultant lineage-committed cell number on day 16. As a result, only the P-erymk41(+) subpopulation on day 9 showed a significantly positive correlation with the number of both mature megakaryocytic cells on day 16 in WT-clones and immature megakaryoblastic cells in G1s-clones (Figure 1G, second and third panels from the left), and also showed a positive correlation with erythroid cells, which were observed only in WT-clones (Figure 1G, leftmost panel). These results suggested that P-erymk41(+) is most closely related to the normal erythroid and megakaryocytic differentiation of WT-clones and the impaired maturation of megakaryocytic lineage in G1s-clones. On the other hand, regarding myeloid lineages, no subpopulations showed a significant correlation in either WT- or G1s-clones (Figure 1G, fourth and fifth panels from the left), indicating that myeloid lineage cells on day 16 could not be attributed to a single HPC subpopulation on day 9 under erythroid-megakaryocytic differentiation condition. Post hoc analyses on targeted transcription profiles strongly indicated that the in vitro TAM phenotype in the G1s-clones reflected some perturbation of the gene expression profiles resulting from a GATA1 mutation in P-erymk41(+) (Online Supplementary Figure S5; Online Supplementary Table S2). We therefore explored differences between WT- and G1s-clones in the core pathway signatures of this subpopulation. Indeed, gene network analysis based on 103 genes extracted by the clustering algorithm (correlation index >0.8 in a factorial space given by principle component analysis)11,12 unveiled striking differences in the cellular pathways (Figure 2A-D; Online Supplementary Figure S6; Online Supplementary Table S3). We found that gene sets related to multi-lineage differentiation, including erythroid-, megakaryocytic-, and myeloid-lineages were significantly enriched in the WT-clones, whereas only myeloid-related gene sets were enriched in the G1s-clones (Figure 2C-D; Online Supplementary Table S3; P-values corrected with Bonferroni step down <0.1). Moreover, we found that pathways related to the cell cycle and DNA damage were highly and significantly enriched only in G1s-clones (Figure 2E; Online Supplementary Table S3; top ten P-val635

Letters to the Editor

ues corrected with Bonferroni step down <0.001). These findings characterize deviations in the cellular pathway profiles of this subpopulation that are dependent on GATA1 mutation. Taken together, these data suggest that the mutation brings about a decay of intracellular signal coordination in these cells at this stage, including an imbalance of signals that determine cell proliferation and

death, as well as in cells involved in cell fate decision. Those perturbations should contribute to the progenitor distribution shown in Figure 1, although more detailed analyses are required. In order to corroborate our findings, we performed progenitor assays by culturing sorted cells from both WT- and G1s-clones (Figure 3A-B). As depicted in Figure

Figure 1. GATA1-dependent differences in the CD34+CD43+CD235- progenitor stage. (A) Schema of the GATA1-wild-type (WT) and GATA1-mutated (G1s) Ts21pluripotent stem cell (PSC) line panels used in this study. (B) Protocol for sequential hematopoietic progenitor cell differentiation. (C) Percentages of CD34+CD43+CD235- cells on day 9. P-values between WT and G1s Ts21-embryonic stem cells (ESC): P=0.0079; and between transient abnormal myelopoiesis (TAM)-induced pluripotent stem cells (iPSC)(TAM-iPSC) WT and G1s: P<0.0001. (D) Percentages of P-mye (CD34+CD43+CD235-CD11b+CD41-) in CD34+CD43+CD235- cells on day 9. (E) Percentages of P-erymk41(-) (CD34+CD43+CD235-CD11b-CD71+CD41-) in CD34+CD43+CD235- cells on day 9. P-values between WT and G1s Ts21-ESC: P=0.0303; and between WT and G1s TAM-iPSC: P=0.0134. (F) Percentages of P-erymk41(+) (CD34+CD43+CD235-CD11bCD71+CD41+) in CD34+CD43+CD235- cells on day 9. P-values between WT and G1s Ts21-ESC: P=0.016; and between WT and G1s TAM-iPSC: P=0.0199. Studentâ&#x20AC;&#x2122;s t-test. Data are shown as the mean + standard deviation. *P<0.05, **P<0.01, ***P<0.001; n.s: not significant; n = 3-4. (G) Table of the correlation between the percentage of each subpopulation on day 9 (column headings, y-axis) and logarithmically-transformed cell number of the resultant lineage-committed cells on day 16 (row headings, x-axis) of the culture. Correlation coefficients and P-values (Pearson correlation in GraphPad Prism) for each square are described under each regression line. *P<0.05, **P<0.01; n=10-14.

636

haematologica | 2021; 106(2)

Letters to the Editor

Figure 2. P-erymk41(+) with GATA1 mutation display activation of pathways that cause abnormal myelopoiesis. (A) Principal component analysis (PCA) mapping P-erymk41(+) (CD34+CD43+CD235-CD11b-CD71+CD41+)of GATA1-wild-type clones (WT-clones) and GATA1-mutated clones (G1s-clones) from Ts21-embryonic stem cells (Ts21-ESC) and transient abnormal myelopoiesis (TAM)-induced pluripotent stem cell (TAM-iPSC) strains. (B-F) The results of a clustering algorithm for automated functional annotations. (B) An expanded map showing the clusters of genes extracted by the positive correlation (correlation index >0.8 in the PCA shown in (A)) with WT (blue) and G1s (-) (magenta). The closed circles depict genes originally included in the PCR array. Open circles depict genes replenished by GENEMANIA. Each gene is listed in the Online Supplementary Table S4. (C-E) The intracellular pathways specifically enriched in (C) WT and (D) G1s cell populations. GO terms with P<0.08 in Bonferroni step down analysis are listed. No significant GO terms for gene sets categorized to WT cluster D or G1s clusters D or E were enriched. The closed circles in (C) and (D) show the lineage specifications indicated by GO terms. The open triangle in (C) depicts the indicated negative relationship to myeloid differentiation. (E) The intracellular pathways specifically enriched in the G1s-clones P-erymk41(+) filtered by â&#x20AC;&#x153;biological processâ&#x20AC;?. The top 10 GO terms based on P-values corrected with Bonferroni step down <0.001 are listed.

haematologica | 2021; 106(2)

637

Letters to the Editor

Figure 3. P-erymk41(+) are an origin of predominant myelopoiesis. (A-B) Schematic process of the progenitor assay. (C) Morphology (May-Grunwald Giemsa staining) of sorted progenitor cells. Representative cells from GATA1-wild-type clones (WT-clones) and GATA1-mutated clones (G1s-clones) from transient abnormal myelopoiesis (TAM)-induced pluripotent stem cell (TAM-iPSC) lines are shown. Scale bars, 25 mm. (D-E) Number of lineage cells derived from 1x104 sorted (D) P-erymk41(+) (CD34+CD43+CD235-CD11b-CD71+CD41+) and (E) P-erymk41(-) of each clone under myeloid lineage-specific differentiation condition. (D) P-values compare Ery (erythrocytes), Meg (megakaryocytes), Megablast (magakaryoblasts), and Mye (myeloids) between WT and G1s Ts21-embryonic stem cells (Ts21-ESC): P=0.6777, P=0.0153, P=0.0121 and P<0.0001; and between WT and G1s TAM-iPSC: P=0.0603, P=0.8109, P=0.1407 and P=0.0018, respectively. (E) P-values compare Ery, Meg, Megablast, and Mye between WT and G1s Ts21-ESC: P=0.0806, P=0.6408, P=0.6860 and P=<0.0001; and between WT and G1s TAM-iPSC: P=0.0071, P=0.0087, P=0.5668 and P<0.0001, respectively. Studentâ&#x20AC;&#x2122;s t-test. Data are presented as the mean + standard devaition. *P<0.05, **P<0.01, ***P<0.001; no mark: not significant; n = 3-7. (F) Fold changes of the total cell number calculated from the ratio of each G1s-clone per WT-clone on day 16. P-values compare Ts21-ESC, TAM-iPSC between P-erymk 41(+) and P-erymk 41(-): P<0.001 and TAM-iPSC (-): P=0.021279, respectively. Multiple comparisons using the Holm-Sidak method. Data are presented as the mean + standard devaition. *P<0.05, **P<0.01 and ***P<0.001; n = 3-7. (G) Graphical schematic showing the impact of GATA1 mutation on P-erymk41(+) in our hematopoietic system.

638

haematologica | 2021; 106(2)

Letters to the Editor 3C, cells harvested from WT- and G1s-clones showed similar morphology in P-erymk41(+) and P-erymk41(-). However, when each subpopulation was separately cultured, both fractions from the G1s-clones presented significantly higher levels in cell number expansion under the myeloid-specific differentiation condition (Figure 3D, E). This change suggested we could use this assay to delineate the impact of the GATA1 mutation in abnormal cell proliferation. Moreover, the difference in cell number expansion revealed that P-erymk41(+) was significantly affected more by the GATA1 mutation only under myeloid-specific differentiation condition (Figure 3F; Online Supplementary Figure S7C) verifying the network analysis described above. On the other hand, under erythroid-megakaryocytic condition, P-erymk41(+) and P-erymk41(-) G1s-clones did not show any significant increase in total cell numbers compared with WT-clones (Online Supplementary Figure S7D). These data together demonstrated P-erymk41(+) is a progenitor intermediate that contributes strongly to abnormal myeloid cell proliferation (Figure 3G). In summary, the current study identified the diseaseresponsible progenitors and the impact of GATA1 mutation on these cells by dissecting the differentiation steps of PSC-based hematopoiesis. The abnormal cells observed in patients with TAM have long been thought as being derived from definitive hematopoiesis,13 and recent studies have shown that PSC-derived definitive hematopoiesis predominantly arises from KDR+ mesodermal progenitors via CD235a-CD34+ subpopulations.14 Given that our data are provided from those specified subpopulations, our Ts21-PSC-based model likely describes definitive hematopoiesis in vitro. Thus, our hematopoietic induction model adequately reflects the pathogenesis of TAM. P-erymk41(+) is an inflection stage for both impaired erythro-megakaryocyte maturation and abnormal cell proliferation in TAM specifically in response to erythromegakaryocytic and myeloid-directed stimulation, respectively. Together, we speculate that P-erymk41(+) can be a target of pre-emptive treatment for TAM. Our data also suggest that DNA damage responses may accumulate in P-erymk41(+) in the presence of GATA1 mutation. As TAM is a condition preceding Down syndrome acute megakaryoblastic leukemia (DS-AMKL) and as DS-AMKL is considered the consequence of acquired gene mutations,2 it is important to identify the most appropriate cell population from which the functional and molecular consequences of GATA1 mutation begins. From that viewpoint, our model should provide novel insights into the pathogenesis, prediction, and treatment of not only TAM, but also acute megakaryocytic leukemia in Down syndrome, though further analysis is needed. Yoko Nishinaka-Arai,1,2* Akira Niwa,1* Shiori Matsuo,1 Yasuhiro Kazuki,3,4 Yuwna Yakura,3 Takehiko Hiroma,5 Tsutomu Toki,6 Tetsushi Sakuma,7 Takashi Yamamoto,7 Etsuro Ito,6 Mitsuo Oshimura,3 Tatsutoshi Nakahata8 and Megumu K. Saito1 1 Department of Clinical Application, Center for iPS cell Research and Application, Kyoto University, Kyoto; 2Department of Human Health Sciences, Graduate School of Medicine, Kyoto University, Kyoto; 3Chromosome Engineering Research Center, Tottori University, Tottori; 4Division of Genome and Cellular Functions, Department of Molecular and Cellular Biology, School of Life Science, Faculty of Medicine, Tottori University, Tottori; 5 Perinatal Medical Center, Nagano Childrenâ&#x20AC;&#x2122;s Hospital, Nagano; 6 Department of Pediatrics, Hirosaki University Graduate School of haematologica | 2021; 106(2)

Medicine, Hirosaki; 7Division of Integrated Sciences for Life, Graduate School of Integrated Sciences for Life, Hiroshima University, Hiroshima and 8Drug Discovery Technology Development Office, Center for iPS cell Research and Application, Kyoto University, Kyoto, Japan *YNA and AN contributed equally as co-first authors Disclosures: No conflicts of interest to disclose. Contributions: YNA and AN designed and performed experiments and analyzed data; SM performed experiments; YK, YY and MO provided Ts21-KhES1 cells and contributed to scientific discussion; TS and TY designed and made plasmids for gene editing; TH, TT and EI obtained informed consent from a patient and harvested primary cells; TN and MKS supervised the project and experimental design; YNA, AN and MKS wrote the paper. Funding: this work was supported by grants from the Ministry of Health, Labour and Welfare of Japan, Ministry of Education, Culture, Sports, Science and Technology of Japan, Japan Society for the Promotion of Science (JSPS) Research Fellowships for Young Scientists [YNA], KAKENHI Grant Number 19K17358 [YNA], KAKENHI Grant Number 25221308 [MO] and KAKENHI Grant Number 16K10026 [AN], Regional Innovation Support Program from the Ministry of Education, Culture, Sports, Science and Technology of Japan [MO and YK], the Leading Project of the Ministry of Education, Culture, Sports, Science and Technology [TN], the Funding Program for World-Leading Innovative Research and Development on Science and Technology of the Japan Society for the Promotion of Science [TN and MKS], CREST [TN], the Core Center for iPS Cell Research of Research Center Network for Realization of Regenerative Medicine from the Japan Agency for Medical Research and Development (AMED) [TN and MKS], the Program for Intractable Diseases Research utilizing Disease-specific iPS cells of AMED (17935423) [YNA, TN and MKS], and Practical Research Project for Rare/Intractable Diseases of AMED (17930095) [MKS]. Correspondence: MEGUMU K. SAITO - msaito@cira.kyoto-u.ac.jp AKIRA NIWA - akiranw@cira.kyoto-u.ac.jp doi:10.3324/haematol.2019.242693

References 1. Gamis AS, Alonzo TA, Gerbing RB. Natural history of transient myeloproliferative disorder clinically diagnosed in Down syndrome neonates: A report from the Childrenâ&#x20AC;&#x2122;s Oncology Group Study A2971. Blood. 2011;118(26):6752-6759. 2. Yoshida K, Toki T, Okuno Y, et al. The landscape of somatic mutations in Down syndrome-related myeloid disorders. Nat Genet. 2013;45(11):1293-1299. 3. Chou ST, Byrska-Bishop M, Tober JM, et al. Trisomy 21-associated defects in human primitive hematopoiesis revealed through induced pluripotent stem cells. Proc Nat Acad Sci U S A. 2012;109(43):17573-17578. 4. Byrska-Bishop M, VanDorn D, Campbell AE, et al. Pluripotent stem cells reveal erythroid-specific activities of the GATA1 N-terminus. J Clin Invest. 2015;125(3):993-1005. 5. Banno K, Omori S, Hirata K, et al. Systematic cellular disease models reveal synergistic interaction of trisomy 21 and GATA1 mutations in hematopoietic abnormalities. Cell Rep. 2016;15(6):1228-1241. 6. Niwa A, Heike T, Umeda K, et al. A novel serum-free monolayer culture for orderly hematopoietic differentiation of human pluripotent cells via mesodermal progenitors. PLoS One. 2011;6(7):e22261. 7. Yanagimachi MD, Niwa A, Tanaka T, et al. Robust and highlyefficient differentiation of functional monocytic cells from human pluripotent stem cells under serum- and feeder cell-free conditions. PLoS One. 2013;8(4):e59243. 8. Sakuma T, Ochiai H, Kaneko T, et al. Repeating pattern of nonRVD variations in DNA-binding modules enhances TALEN activity. Sci Rep. 2013;3:3379. 9. Kazuki Y, Hoshiya H, Kai Y, et al. Correction of a genetic defect

639

Letters to the Editor

in multipotent germline stem cells using a human artificial chromosome. Gene Ther. 2008;15(8):617-624. 10. Psaila B, Barkas N, Iskander D, et al. Single-cell profiling of human megakaryocyte-erythroid progenitors identifies distinct megakaryocyte and erythroid differentiation pathways. Genome Biol. 2016;17:83. 11. Warde-Farley D, Donaldson SL, Comes O, et al. The GeneMANIA prediction server: biological network integration for gene prioritization and predicting gene function. Nucleic Acids Res. 2010;38(Web Server issue):W214-220.

640

12. Wu G, Dawson E, Duong A, et al. ReactomeFIViz: a cytoscape app for pathway and network-based data analysis. F1000Res. 2014;3:146. 13. Bhatnagar N, Nizery L, Tunstall O, et al. Transient abnormal myelopoiesis and AML in Down syndrome: an update. Curr Hematol Malig Rep. 2016;11(5):333-341. 14. Sturgeon CM, Ditadi A, Awong G, et al. Wnt signaling controls the specification of definitive and primitive hematopoiesis from human pluripotent stem cells. Nat Biotechnol. 2014;32(6):554561.

haematologica | 2021; 106(2)

CASE REPORTS Emapalumab treatment in an ADA-SCID patient with refractory hemophagocytic lymphohistiocytosis-related graft failure and disseminated bacillus Calmette-Guérin infection Interferon gamma (IFNγ)-targeted immunotherapy with emapalumab, a fully human monoclonal antibody, was recently approved by the US Food and Drug Administration for the treatment of adult and pediatric patients with primary hemophagocytic lymphohistiocytosis (HLH) who have refractory, recurrent or progressive disease or intolerance with conventional HLH therapy.1,2 Moreover, emapalumab has shown promising efficacy in the treatment of patients with graft failure (GF) requiring a second allogeneic hematopoietic stem cell transplantation (HSCT).3 Interestingly, there is growing evidence to

support common pathophysiolological mechanisms between HLH and immune-mediated GF, highlighting the key role of IFNγ in both conditions.3 IFNγ is a cytokine produced by macrophages and lymphocytes that plays a critical role in both innate and adaptive immune responses. In patients with complete IFNγ receptor deficiency or developing anti-IFNγ autoantibodies, the absence of IFNγ biological activity leads to an increased susceptibility to specific infections, such as mycobacterial infections.4,5 For this reason, latent tubercolosis (TB) infection represented an exclusion criterion in clinical trials investigating the therapeutic role of emapalumab in primary (clinicaltrials.gov identifiers: NCT03312751; NCT01818492) and secondary (clinicaltrials.gov identifiers: NCT03311854) HLH. Here we report a case of secondary HLH-related GF in

Table 1. Characteristics of the three HLA-haploidentical (haplo)-hematopoietic stem cell transplantations performed.

III

Donor Conditioning regimen

Father Treosulfan 14 g/m2 (days -6, -5, -4) Fludarabine 40 mg/m2 (days -6, -5, -4, -3) ATG grafalon 4 mg/kg (days -5, -4, -3) Rituximab 200 mg/m2 (day-1)

Father Fludarabine 30 mg/m2 (days -5;-4,-3) Cyclophosphamide 500 mg/m2 (days -5, -4, -3) ATG thymoglobulin 2 mg/kg (days -3, -2) Rituximab 200 mg/m2 (day -1)

GvHD prophylaxis Cell dose CD34+ cells, x106kg

TCD

Mother VP-16 150 mg/m2 (days -5,-4,-3) Cyclophosphamide 500 mg/m2 (days -5,-4,-3) ATG grafalon 2 mg/kg (days -3, -2) Cyclosporine-A 3 mg/kg (from day -2 to +36) TCD

9.52

10.5

TCR-αβ+ CD3+x104/kg TCR-γd+ CD3+, ×106/kg CD20+ ×104/kg CD3-CD19-CD20-, ×106/kg Engraftment's kinetic Neutrophils* Platelets** PB chimerism#

- αβ+CD19+-depleted CD34+=8.09; - positively selected CD34+=15.01 Total CD34+ =23.1 9

13.1 12 37

14.6 13 41

6.7 9 13

NA NA -

NA NA Donor >90% (day +10)§ Donor >70% (day +13)§ Host >90% (day +18)§ Donor 0%; host >99% (day +28)

Donor 0% (day +20)

Donor >70%; host 2.3% (day +13) Donor 0%; host >99% (day +20)

+10 days +14 days 2nd donor 86%; host 1.2%; 1st donor absent (father) (day +12) 2nd donor 89%; host 0.4%; 1st donor absent (father) (day +18) 2nd donor >95%; host 0%; 1st donor absent (father) (day +100) 2nd donor >99%; host 0.7%; 1st donor absent (father) (day +21) 2nd donor >90%; host 0%; (day +100)

BM chimerism

Haploidentical donors were mobilized with granulocyte-colony stimulating factor (G-CSF) (father) and G-CSF + Plerixafor (mother). *Defined as the time needed to reach an absolute neutrophil count (ANC) >0.5x109/L. **Defined as the time needed to reach an unsupported platelet count >20x109L. Primary graft failure was defined as: ANC <0.5x109/L by day +28, hemoglobin <80 g/L and platelets <20x109/L. Cord blood transplantation: up to day +42. Secondary graft failure was defined as: ANC <0.5x109/L after initial engraftment not related to relapse, infection, or drug toxicity. #Chimerism was assessed on total and/or sorted PB and BM aspirate samples by real-time polymerase chain reaction and relative quantification. At day +100 after the third haplo-HSCT, chimerism analysis was performed also on PB lymphoid and myeloid subpopulations including CD3+ (>90%), CD15+ (>90%) and CD14+ (>85%) cells. §Presence of host/donor cells not evaluated due to insufficient material. HSCT: hematopoietic stem cell transplantation; ATG: anti-thymocyte globulin; VP-16: etoposide; GvHD: graft-versus-host disease; TCD: T-cell depletion; TCR: T-cell receptor; NA: not achieved; PB: peripheral blood; BM: bone marrow.

haematologica | 2021; 106(2)

641

Case Reports

Figure 1. Clinical, brain magnetic resonance imaging (MRI) and chest computed tomography (CT) images of tubercolosis (TB) and fungal infections, and schematic summary of treatments. (A) Localized abscesses on lower limbs at the sites of polyethylene glycol-modified adenosine deaminase injection before surgical incision. (B) Localized abscesses on lower limbs at day +100 after the third haplo-HSCT. Surgical incision was performed before HSC-gene therapy (GT). (C) Brain MRI at diagnosis of intracerebral TB granuloma. Sagittal and coronal post-contrast T1W brain MRI images show a hypothalamic contrast enhancing lesion suggestive for tuberculous granuloma. (D) Brain MRI at day +100 after the third haplo-HSCT showing marked reduction of the tuberculous granuloma. (E) CT images of the lungs at time of aspergillosis diagnosis after the second haplo-HSCT. Axial chest CT images with lung window (left) and mediastinal window (right) show a left lower lobe pulmonary mass compatible with pulmonary aspergillosis. (F) CT images of the lungs at day +100 after the third haplo-HSCT showing marked improvement. (G) Schematic representation of the treatment given before and during the third haplo-HSCT to control secondary HLH and prevent graft failure. MPD: methylprednisolone; VP-16: etoposide; CTX: cyclophosphamide; ATG: anti-thymocyte globulin; Cs-A: cyclosporine-A.

642

haematologica | 2021; 106(2)

Case Reports

the context of HLA-haploidentical HSCT treated with emapalumab in the presence of concomitant life-threatening infections including disseminated bacille CalmetteGuérin disease (BCGitis). The patient is a 4-year old girl with severe combined immunodeficiency caused by adenosine deaminase deficiency (ADA-SCID) referred to our institution for ex vivo hematopoietic stem cell (HSC)-gene therapy (GT) with Strimvelis® in the absence of a human leukocyte antigen (HLA)-identical sibling donor.6 She was in poor clinical condition at presentation with bilateral abscesses on lower limbs, corresponding to the sites of polyethylene glycol-modified adenosine deaminase (PEG-ADA) injections (Figure 1A). Presence of Mycobacterium bovis was identified by direct next-generation sequencing performed on the material drained from the limb abscesses using the Deeplex® Myc-TB, an all-in-one test for specieslevel identification, genotyping and prediction of antibiotic resistance in Mycobacterium tuberculosis complex. Results were further confirmed by standard mycobacteriology procedure and whole genome sequencing of the strain. Moreover, magnetic resonance imaging (MRI) documented an intracerebral granuloma (Figure 1C) and vertebral osteolytic lesions, that, along with the limb abscesses, led to the diagnosis of disseminated BCGitis, as reactivation of the BCG vaccine strain received at birth. She was treated with surgical incision of the abscesses and anti-TB treatment (four-drug regimen [isoniazid, rifampicin, ethambutol, moxifloxacin] for 12 months as intensive phase; two-drug regimen [isoniazid, rifampicin] for 6 months as continuation phase). After 7 months of anti-TB treatment, at resolution of cutaneous abscesses and with residual encephalic mycobacterial lesions, the patient was considered eligible and treated with Strimvelis®. Failure of engraftment of gene-corrected HSC was declared at day +90 and enzyme replacement therapy (ERT) was resumed. Subsequently, due to the lack of a matched unrelated donor and after ERT withdrawal, the patient received a first HLA-haploidentical HSCT after αβ+ T-cell and CD19+ B-cell depletion (TCD) from the father after reduced toxicity conditioning regimen (Table 1).7 However, HSCT failed due to primary GF, likely related to concomitant adenovirus reactivation in the periengraftment phase. A second paternal haplo-HSCT was performed after reduced intensity conditioning employing exceeding HSC cryopreserved from the first transplant and infused on d+31 post first HSCT (Table 1). On day +13 after the second haplo-HSCT, the patient showed persistent fever, hepatosplenomegaly, high levels of triglycerides (383 mg/dL) and markedly elevated inflammatory markers such as ferritin (18,000 mg/dL) and soluble IL2 receptor (16,809 pg/mL; reference values 600-2,000) (Figure 2A and B). Donor chimerism on both peripheral blood (PB) and bone marrow (BM) was documented on days +10 and +13, respectively; however, it was followed by secondary GF with complete loss of donor engraftment (day +18). BM morphology showed hypocellularity with features of active hemophagocytosis. A secondary HLH was diagnosed based on 6 out of 8 HLH-2004 criteria,8 likely triggered by concurrent infections, including Stenotrophomonas maltophilia bacteremia, invasive pulmonary aspergillosis (Figure 1E) and adenovirus reactivation. Treatment with methylprednisolone (2 mg/kg/day) and high-dose immunoglobulins was started. In order to control HLH and reduce the possibility of GF after a third HSCT, compassionate use of emapalumab was requested and approved for this severely haematologica | 2021; 106(2)

immunocompromised patient unable to tolerate standard HLH immunochemotherapy. At time of emapalumab initiation, adenovirus reactivation and invasive pulmonary aspergillosis were active: adenovirus was detected both in plasma and stool with 1,940 copies/mL and >1,000,000 copies/mL, respectively; while galactomannan levels were above the upper limit of detection (index >6). Intensive antimicrobial treatment included antivirals (intravenous cidofovir, later switched to oral brincidofovir) and antifungals (voriconazole plus anidulafungin, later switched to liposomial amphotericine-B plus anidulafungin to minimize drugs interactions). Conversely, rifampicin and isoniazid were continued as secondary prophylaxis to avoid the risk of reactivation of TB, which was regularly monitored through blood cultures, fecal polymerase chain reaction (PCR) for Mycobacterium bovis and brain MRI. Emapalumab was administered intravenously twice a week for a total of 15 infusions with the objective of prompt tapering of glucocorticoids. After the first dose at 1 mg/kg, emapalumab dose was increased to 3 mg/kg: the laboratory parameters, while not worsening, did not show any satisfactory improvement. Thereafter, emapalumab dose was increased to 6 mg/kg, based on deterioration of inflammatory parameters (e.g., ferritin, C-reactive protein) (Figure 2A). IFNγ levels were not particularly elevated in this patient, as documented by CXCL9 values around 260 pg/mL at start of emapalumab treatment (Figure 2B). Nonetheless, the pharmacokinetics (PK) of emapalumab (Figure 2C) was affected by target-mediated drug disposition, documenting high IFNγ production and requiring emapalumab dose increase. CXCL9 progressively decreased to levels below 80 pg/mL, documenting complete neutralization of IFNγ. By the time of the third haplo-HSCT, glucocorticoids dose was reduced to approximately 50% of the starting dose, while maintaining a good clinical control of HLH. The patient, despite the occasional temporary worsening of a few HLH laboratory parameters, did not progress into overt HLH, likely due to the neutralization of IFNγ. The patient received the third TCD haplo-HSCT from the mother after a total of six emapalumab doses. Conditioning regimen included chemotherapeutic agents active against HLH,8 while cyclosporine-A (Cs-A) was added for graft-rejection prevention (Table 1). Anti-HLA antibodies were undetectable before and after HSCT. Neutrophil and platelet engraftment occurred on days +10 and +14, respectively. BM aspirate at day +21 was normo-cellulated with no evidence of hemophagocytosis and showed full donor chimerism. Emapalumab was administered until achievement of sustained donor engraftment (day +28) (Figure 1G). No adverse events occurred. HLH clinical and laboratory parameters progressively improved (Figure 2A and B) allowing Cs-A and steroids tapering and ultimately discontinuation (days +36 and +59, respectively) (Figure 1G). Remarkably, during blockade of IFN-γ with emapalumab, infections remained stable or improved with antimicrobial medications. At the end of treatment, no sign of reactivation of cutaneous TB lesions was observed (Figure 1B) and the brain imaging showed improvement of the lesions documented prior to emapalumab (Figure 1D). Bacteremia resolved and invasive pulmonary Aspergillosis improved with favorable radiological evolution (Figure 1F) and reduction of galactomannan up to negativity at day +132 post HSCT. After 8-week treatment with emapalumab, at day +39 post HSCT, adenovirus became undetectable in plasma. Treatment with brincidofovir was continued until negativity also in stool 643

Case Reports

Figure 2. Significant inflammatory markers from the hemophagocytic lymphohistiocytosis (HLH) diagnosis up to the end of treatment with emapalumab and pharmacokinetics (PK) of the drug. (A) Serum ferritin (normal values [nv]: 15-150 ng/mL) and C-reactive protein (CRP) (nv <6 mg/L); trends are reported in red and blue lines, respectively. (B) IL2 receptor (nv 600-2000 pg/mL) and CXCL9 levels are reported in purple and green, respectively. Normal ranges are reported in light yellow. (C) Concentration-time profile of emapalumab in the patient. Green dots and solid lines represent observed emapalumab concentrations. Black solid lines represent simulated concentrations for the specific patient (i.e., taking into consideration the dosage schedule and measured total interferon gamma (IFNÎł) concentrations) based on the population pharmacokinetic model of emapalumab in HLH patients. Gray area surrounded by orange dotted lines represents the 90% prediction interval of the simulated concentrations. Dotted red line represents the limit of quantification of the bioanalytical assay (62.5 ng/mL). Ticks and numbers on the top line represent times of administration and dose in mg/kg. IL2: interleukin 2; CXCL9: chemokine (C-X-C motif) ligand 9; HSCT: hematopoietic stem cell transplantation.

644

haematologica | 2021; 106(2)

Case Reports

and therefore withdrawn one month later. At day +100 post HSCT the patient was clinically well with full donor chimerism on total BM and PB, as well as on lymphoid and myeloid subpopulations (Table 1). She is currently nine months after the third haplo HSCT and the patient remains in good clinical conditions and is infection-free. Based on emapalumab half-life, the patient has remained on anti-TB prophylaxis to mitigate the risk of reactivation until measurable levels of the drug were present. In conclusion, we report the case of a very fragile, heavily immunosuppressed patient affected by ADASCID who experienced GF after multiple HSCT in the presence of life-threatening infections including disseminated TB, who was safely treated with emapalumab. The activation of the IFNγ pathway has a well-documented double role both in controlling mycobacterial infections and, also, in sustaining HLH hyperinflammatory response.10,11 In our patient, we had to face both challenges: on the one hand to prevent TB reactivation and on the other to inhibit the hyperinflammation responsible for both HLH and GF. Upon review of emapalumab safety and efficacy data reported in primary2,12,13 and secondary14 HLH, and despite the potential risk of TB reactivation, the benefit/risk ratio of treating with emapalumab was deemed favorable. Interestingly, during emapalumab treatment, the initial TB abscesses remained inactive and brain TB findings improved. In this context, neutralization of IFNγ might have contributed to control HLH without the prolonged use of additional myelosuppressive drugs. Moreover, since a third GF was not observed, our findings suggest that, in association with other lines of immunosuppressive/chemotherapic agents, emapalumab might have played a role in reducing the risk of graft rejection, as already shown in both murine models and humans.3,15,16 In addition, while an intrinsic defect of the mesenchymal/osteblast compartment in ADA-SCID patients with reduced capacity to support in vitro and in vivo hematopoiesis9 may have contributed to the repeated GF, the concomitant use of Cs-A in the peri-transplant phase and the mega-dose of CD34+ cells infused after the third haplo-HSCT may have played a role in preventing rejection. This seminal case suggests the feasibility and safety of emapalumab administration to manage secondary HLH and repeated GF also in patients bearing multiple active infections, including TB. Francesca Tucci,1,2* Vera Gallo,1,2* Federica Barzaghi,1,2 Francesca Ferrua,1,2 Maddalena Migliavacca,1,2 Valeria Calbi,1,2 Matteo Doglio,1 Elena Sophia Fratini,1,3 Zeynep Karakas,4 Sukru Guner,5 Matilde Zambelli,6 Cristina Parisi,6 Raffaella Milani,6 Salvatore Gattillo,6 Benedetta Mazzi,7 Chiara Oltolini,8 Maurizio Barbera,9 Cristina Baldoli,9 Daniela Maria Cirillo,10 Veronica Asnaghi,11 Cristina de Min,11 Maria Pia Cicalese,1,2 Fabio Ciceri,3,12 Alessandro Aiuti1,2,3 and Maria Ester Bernardo1,2 1 Pediatric Immunohematology Unit and BMT Program, IRCCS San Raffaele Scientific Institute, Milan, Italy; 2San Raffaele Telethon Institute for Gene Therapy (SR-Tiget), IRCCS San Raffaele Scientific Institute, Milan, Italy; 3Vita-Salute San Raffaele University, Milan, Italy; 4Department of Pediatrics, Division of Hematology and Oncology, Istanbul University, Faculty of Medicine, Istanbul, Turkey; 5 Department of Pediatrics, Division of Immunology and Allergy, Necmettin Erbakan University Faculty of Medicine, Konya, Turkey; 6 Immunohematology and Transfusion Medicine Unit, IRCCS San Raffaele Scientific Institute, Milan, Italy; 7Immunogenetics, HLA and Chimerism Laboratory, IRCCS San Raffaele Scientific Institute, Milan, haematologica | 2021; 106(2)

Italy; 8Clinic of Infectious Diseases, Division of Immunology, Transplantation and Infectious Diseases, IRCCS San Raffaele Scientific Institute, Milan, Italy; 9Neuroradiology Unit, Head and Neck Department, IRCSS San Raffaele Scientific Institute, Milan, Italy; 10 Emerging Bacterial Pathogens Unit, Division of Immunology, Transplantation and Infectious Diseases, IRCCS San Raffaele Scientific Institute, Milan, Italy; 11Swedish Orphan Biovitrum AG (Sobi), Basel, Switzerland and 12Hematology and Bone Marrow Transplantation Unit, IRCCS San Raffaele Scientific Institute, Milan, Italy *FT and VG contributed equally as co-first authors.

Correspondence: MARIA ESTER BERNARDO - bernardo.mariaester@hsr.it doi:10.3324/haematol.2020.255620 Disclosures: MPC is PI of the long-term follow up study sponsored by Orchard Therapeutics for patients treated with Strimvelis and AA is PI of clinical trials sponsored by Orchard Therapeutics. VA is an employee of Sobi; CdM is consultant for Sobi. Contributions: FT, VG followed the clinical course of the patient, collected data, analyzed results and wrote the paper; FB, FF, MM, VC, MD, MPC, ESF, CO followed the clinical course of the patient and collected data; ZK and SGu followed the patient before referral to our Institution; SGa collected peripheral blood hematopoietic stem cells from the donors; MZ, CP and RM performed graft manipulation; BM performed chimerism analyses; CB and MB performed radiological evaluations; DMC performed fecal PCR for detection of Mycobacterium bovis; VA and CdM provided emapalumab as per CU, performed PK/PD emapalumab evaluations, critically read the manuscript and helped with scientific discussion; AA and FC critically revised the manuscript and helped with scientific discussion; MEB advised on the design the study, analyzed data and contributed to the final writing of the paper. Acknowledgments: we are grateful to Ambra Corti for support in regulatory aspects of the study, and Manuela Gavina, Francesca Dionisio, Stefania Giannelli and Claudia Sartirana for technical support. We thank Emanuele Borroni for support with the mycobacteriology investigations.

References 1. Al-Salama ZT. Emapalumab: first global approval. Drugs. 2019; 79(1):99-103. 2. Locatelli F, Jordan MB, Allen C, et al. Emapalumab in children with primary hemophagocytic lymphohistiocytosis. N Engl J Med. 2020; 382(19):1811-1822. 3. Merli P, Caruana I, De Vito R, et al. Role of interferon-γ in immunemediated graft failure after allogeneic hematopoietic stem cell transplantation. Haematologica. 2019;104(11):2314-2323. 4. Dorman SE, Picard C, Lammas D, et al. Clinical features of dominant and recessive interferon gamma receptor 1 deficiencies. Lancet. 2004;364(9451):2113-2121. 5. Lammas DA, Casanova JL, Kumararatne DS. Clinical consequences of defects in the IL-12-dependent interferon-gamma (IFN-γ) pathway. Clin Exp Immunol. 2000;121(3):417-425. 6. Kohn DB, Hershfield MS, Puck JM, et al. Consensus approach for the management of severe combined immune deficiency caused by adenosine deaminase deficiency. J Allergy Clin Immunol. 2019; 143(3):852-863. 7. Bertaina A, Merli P, Rutella S, et al. HLA-haploidentical stem cell transplantation after removal of αβ+ T and B cells in children with nonmalignant disorders. Blood. 2014;124(5):822-826. 8. Henter JI, Horne A, Aricò M, et al. HLH-2004: diagnostic and therapeutic guidelines for hemophagocytic lymphohistiocytosis. Pediatr Blood Cancer. 2007;48(2):124-131. 9. Sauer AV, Mrak E, Jofra Hernandez R, et al. ADA-deficient SCID is associated with a specific microenvironment and bone phenotype characterized by RANKL/OPG imbalance and osteoblast insufficiency. Blood. 2009;114(15):3216-3226. 10. Zhang SY, Boisson-Dupuis S, Chapgier A, et al. Inborn errors of interferon (IFN)-mediated immunity in humans: insights into the respective roles of IFN-α/β, IFN-γ, and IFN-l in host defense. Immunol Rev. 2008;226:29-40.

645

Case Reports

11. Jordan MB, Hildeman D, Kappler J, Marrack P. An animal model of hemophagocytic lymphohistiocytosis (HLH): CD8+ T cells and interferon gamma are essential for the disorder. Blood. 2004; 104(3):735-743. 12. Locatelli F, Jordan MB, Allen CE, et al. Safety and efficacy of emapalumab in pediatric patients with primary hemophagocytic lymphohistiocytosis. Blood. 2018;132(Suppl 1):LBA-6. 13. De Benedetti F, Brogan P, Grom P, et al. Emapalumab, an interferon gamma (IFNγ) blocking monoclonal antibody, in patients with macrophage activation syndrome (MAS) complicating systemic juvenile idiopathic arthritis (sJIA). Ann Rheum Dis. 2019;78 (Suppl

646

2):178. 14. Lounder DT, Bin Q, de Min C, Jordan MB. Treatment of refractory hemophagocytic lymphohistiocytosis with emapalumab despite severe concurrent infections. Blood Adv. 2019;3(1):47-50. 15. Prencipe G, Caiello I, Pascarella A, et al. Neutralization of IFN-γ reverts clinical and laboratory features in a mouse model of macrophage activation syndrome. J Allergy Clin Immunol. 2018;141(4):1439-1449. 16. Buatois V, Chatel L, Cons L, et al. Use of a mouse model to identify a blood biomarker for IFNγ activity in pediatric secondary hemophagocytic lymphohistiocytosis. Transl Res. 2017;180:37-52.

haematologica | 2021; 106(2)

Case Reports

Detectable mutations precede neoplasia in aplastic anemia

late

myeloid

Aplastic anemia (AA) is bone marrow failure characterized by a hypocellular marrow and peripheral pancytopenia. Immunosuppressive therapy (IST) results in hematologic recovery in the majority, decreased short term cytopenia related complications, and increased survival. However, clonal evolution to a secondary myeloid malignancy is a major complication in long-term survivors.1 Clonal evolution occurs in 10-15% of severe aplastic anemia (SAA) patients after IST.2 Most high risk clonal evolution is associated with complete or partial loss of chromosome 7, which is also frequent in other marrow failure syndromes.3 Previous studies have found older age, multiple rounds of IST, and severity of cytopenia as risk factors for clonal evolution.2,3 Shorter telomere length at diagnosis of AA and accelerated telomere attrition preceding the malignant transformation are associated with chromosome 7 clonal evolution, suggesting genomic instability as a possible mechanism.1 Frank myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML) without chromosome 7 involvement is much less common.3-5 Mutations in myeloid neoplasia related genes are present in 20-30% of patients with SAA.6,7 Excluding PIGA mutations, the most frequently affected genes are epigenetic regulators: ASXL1, DNMT3A and BCOR. The presence of unfavorable somatic mutations as a group is associated with a higher risk of clonal evolution and poorer survival.7,8 Surprisingly, somatic mutations were not frequently detected in clonal evolution to monosomy 7,9 including in patients treated with eltrombopag (EPAG).10 We retrospectively assessed all subjects with SAA treated at the NIH Clinical Center on Hematology Branch protocols, with either horse- anti thymocyte globulin (ATG), rabbit-ATG or alemtuzumab based regimens for treatment-naĂŻve or relapsed/refractory disease from 1989 to 2019. A total of 666 subjects were identified, of whom 96 had clonal evolution. The definition of clonal evolution was consistent across protocols. Clonal evolution was defined as an acquisition of new cytogenetic abnormalities with or without morphologic evidence of a myeloid malignancy. It was considered a high-risk clonal evolution when there was a diagnosis of a myeloid malignancy, an isolated acquisition of chromosome 7 abnormality or complex karyotype. High-risk evolution was

noted in 58 of 96 subjects (Figure 1). Fifteen of these 58 subjects had evolved at 5 years or later after initial IST treatment. Among these 15 patients, eight subjects evolved with cytogenetic abnormalities and seven had MDS/AML with normal karyotype. We further investigated whether the late evolution to MDS/AML in the absence of cytogenetic changes was associated with somatic mutations in myeloid genes. Three subjects evolved at outside institutions and were excluded from the analysis due to insufficient data. One subject with loss of Y, a common age-related cytogenetic abnormality, was included. In this case series, we report four cases of late occurring MDS/AML without chromosome 7 abnormalities. Patients were screened for mutations in genes recurrently mutated in myeloid malignancies by next generation sequencing (NGS) at the time of clonal evolution. Detected variants were confirmed and tracked back in earlier serial samples by digital droplet PCR (ddPCR). All patients were evaluated with clinic visits, bone marrow examinations, and cytogenetics, as dictated by treatment protocols No NCT number: Rosenfeld, Blood. 1995: NCT00001964 and NCT01623167. All patients gave written informed consent to inclusion in the respective studies. Telomere length (TL) was measured by flowfluorescence in situ hybridization (FISH) at a commercial laboratory (Repeat Diagnostics) or by Southern blotting at the time of SAA diagnosis. The details about NGS methods are included in the Online Supplementary Appendix. The median age of SAA diagnosis in the four included subjects was 38 years (range: 29-59; Table 1). All four subjects had normal TL at SAA diagnosis. At 6 months, three subjects had achieved partial response and one a complete response (CR) to IST. Only unique patient number (UPN)-4 relapsed 7 months after treatment and required 2 years of additional cyclosporine and EPAG until CR was achieved. This subject also received eculizumab treatment for a large glycosylphosphatidylinositol (GPI) deficient granulocyte clone. At evolution to myeloid malignancy, median time from initial IST was 5.7 years (range: 5-7.3 years). UPN-1 had pancytopenia and UPN-2 had progressive thrombocytopenia at clonal evolution, but UPN-3 and UPN-4 were diagnosed on protocol-mandated bone marrow evaluations with minimal to no change in their blood counts. Cytogenetics were normal in all except UPN-4, who had loss of chromosome Y present a year prior to evolution.

Figure 1. Cytogenetic abnormalities in high-risk clonal evolution stratified by time to evolution from initial immunosuppressive treatment for severe aplastic anemia. There were 58 high-risk clonal evolutions in this large cohort of 666 subjects with severe aplastic anemia (SAA) diagnosis treated with immunosuppressive treatment (IST). Majority occurred before 5 years and the most common cytogenetic abnormalities were in chromosome 7 including aneuploidy, partial loss/deletion, or inversion. Chr7: chromosome 7; NA: not available.

haematologica | 2021; 106(2)

647

Case Reports

Except for UPN-2, all subjects had less than 10% blasts at evolution. In order to determine the presence of mutations within the malignant clone, we performed NGS on UPN-1, UPN-2, and UPN-4 samples from the time of myeloid malignancy diagnosis that had been enriched by flow cytometry sorting based on blast immunophenotype (Table 1; Online Supplementary Table S1A). For UPN1, NGS of CD117+ peripheral blood mononuclear cells

(PBMC) at clonal evolution demonstrated mutations in NPM1 (Type A insertion) and NRAS. A custom ddPCR assay for the NPM1 mutation detected the variant in unsorted bone marrow taken 128 and 864 days prior to clonal evolution but not from samples collected prior to that (Figure 2A). In UPN-2, two variants in RUNX1 and one in ASXL1 were detected in sorted CD117+ from bone marrow mononuclear cells at clonal evolution. A custom

Table 1. Patient characteristics and details of clonal evolution.

Patient Age at SAA diagnosis Severity of SAA Sex Ethnicity Initial Treatment IST response at 6 months Relapse PNH Clone* Telomere length

UPN-1

UPN-2

UPN-3

UPN-4

35 years VSAA Female White h-ATG/CsA/MMF Partial No Normal

29 years SAA Male Hispanic h-ATG/CsA Partial No No Normal

41 years SAA Male White h-ATG/CsA/EPAG Complete No No Normal

59 years SAA Male White h-ATG/CsA/EPAG Partial Yes Yes Normal

At clonal evolution Time to clonal evolution Diagnosis Blood counts ANC (/mL) Platelet (/mL) Hemoglobin (g/dL) ARC (/mL) Bone marrow Cellularity (%) Blasts (%) Dysplasia Immunophenotype by flow cytometry Cytogenetics Treatment Outcomes molecular testing

7.3 years MDS EB-1

6.3 years AML

5.0 years MDS/MPN

5.1 years MDS EB-1

4,450 49,000 10.6 21,700

510 28,000 13.2 43,900

9,360 175,000 14.7 182,100

1,920 115,000 12.5 116,100

90-100 8 Megakaryocyte and erythrocyte NA; IHC (CD34 neg, CD117+, MPO+, TdT–, rare CD14+)

40-60 31 No

60-70 3-4 Megakaryocyte

40-50 5-6 Megakaryocyte

CD34+, CD117+, MPO+, TdT+

46, XY [20] MSD BMT Alive

CD34+, CD117+, HLADR+, CD33+, CD13+ 46, XY [20] Observation Alive

CD34+, CD117+, HLADR+, CD33+, CD13+, TDT+, MPO+ 45,X,-Y[20] MUD BMT Alive

Detected variants in enriched blasts or unsorted bone marrow at clonal evolution

RUNX1 NRAS NPM1

ASXL1

c.35G>C; p.G12A c.860_863dup; p.W288CfsTer12 -

SETBP1 PHF6

c.292del; p.L98SfsTer24 c.374del; p.P125QfsTer8 -

c.319C>T; p.R107C

c.2067dup; p.D690RfsTer28 -

c.1900_1922del; p.E635RfsTer15 c.2602G>A; D868N c.821G>A; p.R274Q

c.768del; p.T257LfsTer54 c.507_508+5dup -

SAA: severe plastic anemia; IST: immunosuppressive treatment; PNH: paroxysmal nocturnal hemoglobinuria; ANC: absolute neutrophil count; ARC: absolute reticulocyte count; h-ATG: horse anti-thymocyte globulin; CsA: cyclosporine; EPAG: eltrombopag; MMF: Mycophenolate mofetil; UPN: unique patient number; IHC: immunohistochemistry; MDS EB-1: myelodysplastic syndrome excess blast 1; AML: acute myeloid leukemia; MPN: myeloproliferative neoplasm; MSD: matched sibling donor; MUD: matched unrelated donor; BMT: bone marrow transplantation. VSAA (very severe aplastic anemia) was defined as ANC <200/mL. Complete response was defined as absolute ANC ≥1,000/mL, platelet≥100,000/mL, hemoglobin ≥10 g/dL. Partial response was defined as blood counts no longer meeting the standard “Camitta” criteria – ANC ≥500/mL, platelet ≥ 20,000/mL, absolute reticulocyte count ≥ 60,000/mL. *PNH clones >1% in granulocytes.

648

haematologica | 2021; 106(2)

Case Reports

ddPCR assay for the RUNX1 c.292del mutation detected the mutation in unsorted bone marrow from clonal evolution at variant allele frequency (VAF) of 7% and also 497 days prior to clonal evolution at VAF <1%, but not in a bone marrow sample 2163 days prior to clonal evolution (Figure 2B). UPN-3 and UPN-4 had clinical NGS testing at clonal evolution. For UPN-3, clinical NGS reported an ASXL1 variant at 6 months post-IST with successive acquisition of RUNX1, SETBP1, and PHF6 variants at 2, 5, and 7 years post-IST respectively (Online Supplementary Table S1B). In order to elucidate the clonal architecture in this subject, we employed a microfluidics-based, single-cell DNA sequencing assay on a sample from the time of malignancy diagnosis which demonstrated that these mutations were all present within the same clone derived from an initial ASXL1 mutated clonal population (Figure 2C). While no somatic variants were reported by clinical NGS analysis in UPN-4 at the time of clonal evolution, research NGS of CD34+ sorted PBMC from this timepoint demonstrated two RUNX1 mutations (Table 1). A custom ddPCR assay for one of these variants (p.Thr257LeufsTer54) confirmed detection at VAF <2% in unsorted bone marrow collected at the time of clonal evolution but not in a marrow sample from 1 year prior.

UPN-1, UPN-2 and UPN-4 underwent allogeneic bone marrow transplantation and all three were alive and disease-free at last follow-up. UPN-3 was followed expectantly with stable counts but a repeat bone marrow 2 years after clonal evolution and 7 years from initial IST revealed morphologic progression with increased dysplasia, presence of trisomy 21, and acquisition of PHF6 somatic mutation. Clonal evolution after IST for SAA has a varied clinical course. Although high risk evolution with chromosome 7 aneuploidy with or without morphologic evidence of myeloid malignancy is the most common, fewer SAA patients evolve to MDS/AML with normal cytogenetics or without chromosome 7 abnormalities. Isolated chromosome abnormalities, such as deletion 13q and trisomy 8, are considered low risk because patients appear to have better overall survival and less progression to AML.11 Late clonal evolution to MDS/AML without typical cytogenetic abnormalities after IST for SAA occurs in a very small proportion of patients. Much investigation has been done to understand the underlying etiology of clonal evolution associated with monosomy 7 but a deeper understanding of other forms of high-risk evolution is lacking. We report here the first case series to our knowledge describing clinical and

Figure 2. Detection of mutations in three severe aplastic anemia subjects prior to clonal evolution. (A) Presence of NPM1 mutation detected by digital droplet PCR (ddPCR) at, and 2 years prior to, the diagnosis of clonal evolution in unique patient number (UPN)-1; (B) RUNX1 variant was found using ddPCR at, and 1 year before, the diagnosis of clonal evolution in UPN-2; (C) single-cell DNA sequencing at the time of clonal evolution in UPN-3 demonstrates that the successive acquisition of mutations over a 7-year period, including 2 years after the diagnosis of clonal evolution to a myeloid malignancy were all within the same clonal population. SAA: severe aplastic anemia; VAF: variant allele frequency.

haematologica | 2021; 106(2)

649

Case Reports

genetic features of SAA patients developing MDS/AML without chromosome 7 abnormalities, arising more than 5 years after IST. In all four patients, we identified molecular alterations in genes recurrently mutated in myeloid malignancies: RUNX1, NRAS, NPM1, ASXL1, SETBP1, PHF6.12 Three subjects had at least one RUNX1 variant, for which the reported prevalence is 10% in de novo AML/MDS or therapy related myeloid neoplasm.12 Somatic RUNX1 mutations have also been described in myeloid malignancies arising from inherited marrow failure including Fanconi anemia (20%) and congenital neutropenia (64%).13,14 Accelerated telomere attrition correlates with monosomy 7 transformation,9 but acquisition of somatic mutations, specifically RUNX1, may be implicated in late high risk clonal evolution without chromosome 7 aberration. Previously, ASXL1 has been associated with higher risk of myeloid malignancy with abnormal cytogenetics.8 Whether clonal evolution to MDS/AML is a result of genomic instability present at diagnosis or developing after IST has not been elucidated. We were able to detect somatic variants present in bone marrow samples from 1-2 years prior to clonal evolution but not at initial SAA diagnosis. Given the rare nature of late evolution to myeloid malignancy without chromosomal abnormalities, and the current limitations of NGS for de novo discovery of MDS/AML variants,15 it is not known if sequential molecular monitoring would be clinically useful for early detection of evolving myeloid malignancy. The utility of early interventions designed to prevent development of myeloid malignancy in high-risk individuals is currently unknown. Approximately 10% of SAA patients treated with IST in this large cohort developed high-risk clonal evolution, supporting the importance of regular long-term clinical follow-up. Bhavisha A. Patel,* Jack Ghannam,* Emma M. Groarke,* Meghali Goswami, Laura Dillon, Fernanda GutierrezRodrigues, Olga Rios, Diego Quinones Raffo, Jennifer Lotter, Neal S. Young# and Christopher S. Hourigan# Hematology Branch, National Heart, Lung, and Blood Institute (NHLBI),Bethesda, MD, USA *BAP, JG, and EMG contributed equally as co-first authors. # CSH and NSY contributed euqually as co-senior authors

Correspondence: NEAL S. YOUNG - youngns@nhlbi.nih.gov doi:10.3324/haematol.2020.263046 Disclosures: Eltrombopag was provided by GlaxoSmithKline and Novartis under a Clinical Trials Agreement with NIH, which included research funding for additional experimental laboratory studies.

650

Contributions: BAP, EMG, NSY designed the study and performed clinical analyses; JG, MG, LD, DQ performed laboratory studies and analyses; BAP, EMG, NSY, OR, JL provided clinical care; BAP, EMG JG, FGR, CSH and NSY wrote and edited the manuscript. Funding: this research was supported in part by the Intramural Research Program of the NIH and NHLBI.

References 1. Young NS. Aplastic anemia. N Engl J Med. 2018;379(17):1643-1656. 2. Scheinberg P. and Young NS. How I treat acquired aplastic anemia.

Blood. 2012;120(6):1185-1196. 3. Li Y, Li X, Ge M, et al. Long-term follow-up of clonal evolutions in

802 aplastic anemia patients: a single-center experience. Ann Hematol. 2011;90(5):529-537. 4. Maciejewski JP and Balasubramanian SK. Clinical implications of somatic mutations in aplastic anemia and myelodysplastic syndrome in genomic age. Hematology Am Soc Hematol Educ Program. 2017; 2017(1):66-72. 5. de Planque MM, Bacigalupo A, Wursch A, et al. Long-term followup of severe aplastic anaemia patients treated with antithymocyte globulin. Severe Aplastic Anaemia Working Party of the European Cooperative Group for Bone Marrow Transplantation (EBMT). Br J Haematol. 1989;73(1):121-126. 6. Ogawa S. Clonal hematopoiesis in acquired aplastic anemia. Blood. 2016;128(3):337-347. 7. Kulasekararaj AG, Jiang J, Smith AE, et al., Somatic mutations identify a subgroup of aplastic anemia patients who progress to myelodys plastic syndrome. Blood. 2014;124(17):2698-2704. 8. Huang J, Ge M, Lu S, et al. Mutations of ASXL1 and TET2 in aplastic anemia. Haematologica. 2015;100(5):e172-e175. 9. Dumitriu B, Feng X, Townsley DM, et al. Telomere attrition and candidate gene mutations preceding monosomy 7 in aplastic anemia. Blood. 2015;125(4):706-709. 10. Winkler T, Fan X, Cooper J, et al. Treatment optimization and genomic outcomes in refractory severe aplastic anemia treated with eltrombopag. Blood. 2019;133(24):2575-2585. 11. Maciejewski JP, Risitano A, Sloand EM, Nunez O, Young NS, et al. Distinct clinical outcomes for cytogenetic abnormalities evolving from aplastic anemia. Blood. 2002;99(9):3129-3135. 12. Haferlach T, Nagata Y, Grossman V, et al. Landscape of genetic lesions in 944 patients with myelodysplastic syndromes. Leukemia. 2014;28(2):241-247. 13. Quentin S, Cuccuini W, Ceccaldi R, et al. Myelodysplasia and leukemia of Fanconi anemia are associated with a specific pattern of genomic abnormalities that includes cryptic RUNX1/AML1 lesions. Blood. 2011;117(15):e161-e170. 14. Skokowa J, Steinemann D, Katsman-Kuipers JE, et al. Cooperativity of RUNX1 and CSF3R mutations in severe congenital neutropenia: a unique pathway in myeloid leukemogenesis. Blood. 2014; 123(14):2229-2237. 15. Ghannam J, Dillon LW, Hourigan CS. Next-generation sequencing for measurable residual disease detection in acute myeloid leukaemia. Br J Haematol. 2020;188(1):77-85.

haematologica | 2021; 106(2)

Case Reports

A deep molecular response of splenic marginal zone lymphoma to front-line checkpoint blockade Splenic marginal zone lymphoma (SMZL) is an uncommon form of B-cell non-Hodgkin lymphoma (B-NHL) that frequently presents with prominent splenomegaly, often with circulating malignant lymphocytes but with modest lymphadenopathy. As SMZL is generally considered an incurable disease, treatment is often deferred until the manifestation of cytopenias, symptomatic splenomegaly, or constitutional symptoms warrant intervention. The current approach to frontline treatment typically involves rituximab given with or without chemotherapy, treatment of associated conditions such as hepatitis C infection, and less commonly splenectomy. Immune checkpoint blockade with agents such as antiPD-1 antibodies have been effective in certain subtypes of B-NHL, but the use of these agents in patients with previously untreated SMZL has not yet been explored. A 77 year-old man with a history of stage IIc melanoma treated with wide local excision with curative intent presented to the hematology clinic in June 2016 for evaluation of thrombocytopenia (Figure 1). An initial workup revealed what was thought to be an incidental IgA l and IgG κ monoclonal gammopathy of undetermined significance (MGUS), and the decision was made to proceed with observation. However, after developing anemia and progressive thrombocytopenia, further workup was initiated, including imaging that revealed mild mediastinal and bilateral axillary adenopathy and prominent splenomegaly. A bone marrow biopsy was performed in June 2017 that showed an atypical lymphoid population positive for B-lymphoid markers CD19 and CD20, negative for CD5 and CD23, and l light chain restricted. Next-generation sequencing of the bone marrow aspirate using an in-house gene panel revealed a TP53 I195S mutation with a variant allele frequency of 0.06 and loss of PTEN.1 Though uncommon, adenopathy can occur in SMZL, and while splenic B-cell lymphoma unclassifiable was also considered,2,3 the combination of an aberrant immunophenotype, cytopenias, and significant splenomegaly made SMZL the most likely diagnosis. However, given the absence of symptoms, therapy was deferred at that time. Approximately 3 months later, surveillance imaging for his melanoma suggested metastatic disease, which was subsequently confirmed by biopsy of a right lung nodule. Pembrolizumab monotherapy every 3 weeks was recommended to treat his melanoma. At the time, he had also developed worsening and symptomatic splenomegaly (22.9 cm), intermittent drenching night sweats, fatigue, and ongoing cytopenias, suggesting the need for SMZL therapy. However, given that the metastatic melanoma was thought to be the more immediately threatening malignancy, treatment of the SMZL was deferred in order to initiate melanoma therapy. On the day he initiated treatment with pembrolizumab in September 2017 his laborarory results were notable for a white blood cell count (WBC) of 26 K/uL, of which 69% were lymphocytes, hemoglobin of 11.1 g/dL, and a platelet count of 60 K/uL. Within 3 months of starting pembrolizumab, the melanoma metastases had significantly decreased in size. Interestingly, a substantial response was also observed in his SMZL: spleen size was reduced to 14.0 cm, WBC was 5.9 K/uL, absolute lymphocyte count was 0.84 K/uL, hemoglobin was 13.8 g/dL, and the B-symptoms had largely disappeared, although his platelets remained at 60. His pembrolizumab treatment course was complicathaematologica | 2021; 106(2)

ed by type I diabetes mellitus, which was considered an immune-related adverse event requiring an inpatient admission for diabetic ketoacidosis that was effectively managed. Later in his course he was admitted for sepsis, which was thought to be unrelated to treatment. At the last documented follow-up, he had received 35 cycles of pembrolizumab, the melanoma was in complete remission, the spleen size was normal, and his blood counts had remained stable, with the only abnormality being persistent thrombocytopenia, which had improved, but remained in the 90-105 range. To date he has not received any therapy specifically directed to the SMZL. After obtaining Instituational Review Board approval, genomic DNA was isolated from peripheral blood mononuclear cells (DNEasy Kit, Qiagen, Hilden, Germany). Hybrid capture was performed on the genomic DNA samples followed by library preparation with unique molecular identifiers using a custom bait set from Twist Bioscience (San Francisco, CA, USA). Sequencing was performed on the Illumina platform (Illumina, San Diego, CA, USA) and after deduplication and consensus sequence calling, mutations were identified using Varscan 2.2.3, and annotated using Annovar. Mutations were scored based on allele frequency, strand bias differential, local noise and mapping quality, and frequency in known single nucleotide polymorphism (SNP) databases. These variants were visually inspected in Integrated Genome Viewer (Broad Institute, Cambridge, MA, USA). Bone marrow staining was performed using antibodies to PD-L1 (Clone E1L3N, Cell Signaling Technology, Danvers, MA, USA) and BSAP (Clone PAX5, BD Biosciences, San Jose, CA, USA). We had banked a pre-treatment peripheral blood mononuclear cell (PBMC) sample just prior to pembrolizumab initiation, and after observing this dramatic response of therapy-naive SMZL to PD-1 blockade, we obtained another PBMC sample approximately 6 months later. Genomic DNA was extracted from the samples and error-corrected deep sequencing using unique molecular identifiers was performed on a panel of genes that are recurrently mutated in hematologic malignancies. With this technology we are able to detect the presence of mutations to a variant allele frequency (VAF) as low as 0.003.4 Prior to pembrolizumab initiation, the same TP53 I195S mutation identified in the diagnostic bone marrow biopsy was again observed, this time with an elevated VAF of 0.78, consistent with the predominance of lymphocytes at the time of sample acquisition and the enrichment of lymphoid DNA from PBMC.5 Remarkably, after 6 months of pembrolizumab treatment, the mutation was undetectable in the blood, suggesting a deep molecular response of the SMZL. The raw mutation calls visualized in Integrated Genome Viewer are shown in Figure 2A. We also examined the clinical sequencing that was performed on the melanoma biopsy specimen using another custom panel.6 The sequencing data reported two separate TP53 mutations: I195S (VAF 0.29) and G105V (VAF 0.13). The TP53 I195S mutation was therefore bloodderived, whereas the G105V was tumor derived. This conclusion is also consistent with the finding that the average VAF among all mutations identified in the tumor was 0.11, close to the VAF of the G105V mutation but less than half of that of the I195S mutation (Figure 2B). The “contamination” of solid tumor sequencing by somatic mutations present in blood cells is an increasingly recognized phenomenon, can complicate interpretation of these data, and may have important prognostic and therapeutic implications.7-10 651

Case Reports

Figure 1. Blood counts and spleen size changes over time. Shown are the trends of white blood cell count, hemoglobin, platelet count, and spleen size across the clinical course of the patient. The red arrows at the top indicate significant events that occurred during this time.

In addition to eliminating his symptoms, normalizing the spleen size, lymphocytosis, and anemia, the response to pembrolizumab was both rapid and durable. This striking response was associated with the disappearance of the TP53 I195S mutation and raises parallels to the concept of minimal residual disease and its implications in hematologic diseases. While we cannot rule out that sequencing of a bone marrow sample may have identified the mutation, the SMZL cells clearly were circulating (as evidenced by the lymphocytosis and sequencing data in the pre-treatment sample), and there is evidence to suggest high correlation between clonal mutations identified in the bone marrow and bloodstream.11 The persistent thrombocytopenia after treatment was of uncertain etio-logy but presumed to be immune-mediated in nature due to the pembrolizumab therapy. Finally, to gain insight into the mechanism of activity against the SMZL, we stained the bone marrow biopsy specimen for BSAP to highlight the B-cell population and PD-L1. Consistent with the striking response to pembrolizumab, the abnormal B-cell population uniformly expressed PD-L1 (Figure 2C). In contrast, the surrounding 652

myeloid and erythroid cells did not appear to be PD-L1 positive and there was no obvious local inflammatory infiltrate. There are limited data on the use of checkpoint inhibitors for the treatment of MZL, and we are not aware of any study reporting the use of these agents either as frontline therapy or specifically in SMZL. One patient with MZL was treated in a phase Ib study of nivolumab for relapsed or refractory hematologic malignancies but the results for this individual were grouped with other lymphoma subtypes and are therefore unavailable.12 There is an ongoing trial of pembrolizumab alone or with idelalisib or ibrutinib for relapsed or refractory lymphomas that allows for MZL, but the full results have not been reported.13 Furthermore, patients enrolled in clinical trials have generally already received numerous therapies and are thus likely more immune suppressed, in contrast to this case where the patient was therapynaive. While our data suggest that frontline PD-1 blockade may be efficacious for the treatment of SMZL, we recognize that established frontline therapy options in this disease, in particular rituximab, are already effective haematologica | 2021; 106(2)

Case Reports

Figure 2. Identification and quantification of TP53 I195S mutation. (A) The raw mutation calls visualized in Integrated Genome Viewer (IGV, Broad Institute, Cambridge, MA) are shown for the pretreatment (top) and posttreatment (bottom) blood samples. The T584G DNA mutation that corresponds to the protein change I195S was present in 78% of the reads in the pretreatment sample but absent in the posttreatment sample. (B) Variant allele frequency of missense, nonsense, and frameshift mutations identified in the metastatic melanoma biopsy. The TP53 I195S mutation identified in the blood is highlighted in red. (C) Diagnostic bone marrow biopsy specimen stained with B-cell marker BSAP (red) and PD-L1 (brown).

and well-tolerated. Our data do suggest that studying PD-1 blockade in earlier lines of therapy for relapsed or refractory disease, particularly prior to utilizing chemotherapy, may be more likely to affect a therapeutic response due to the potential to harness a more intact immune system. In this patient, the SMZL population expressed PD-L1, suggesting a direct T-cell mediated effect of pembrolizumab. The role of the tumor microenvironment in response to PD-1 blockade remains an area of active study, and despite the relative paucity of publications looking at checkpoint blockade in MZL, there does seem to be significant biological variability between and within different types of MZL.14,15 For example, in one study of 54 SMZL, PD-L1 positive histiocytes and dendritic cells were found in 75% of the tumors but the Pax-5 tumor cells themselves were uniformly PD-L1 negative.16 However, in clinical studies of patients with NHL, expression of PD-L1 does not always predict response to PD-1 blockade.12,13 These observations further raise the potential role of the tumor microenvironment in mediating the response to these and other immunotherapies. Our findings and these data provide biological rationale haematologica | 2021; 106(2)

for the use of PD-1 blockade in this setting. TP53 missense mutations are common in B-NHL, including at the known hotspots R175H and R248Q. The TP53 I195S missense mutation identified in this case occurs in the DNA binding domain of TP53 where most of the hotspot mutations occur.17 In a prior publication from our group we showed that mutations at position I195 generally, and I195S specifically, confer dominant negative activity on P53.18 In one study that extracted SMZL cases from 14 different studies, TP53 was the third most commonly mutated gene and present in 15% of cases.19 While the identification of a TP53 mutations does not provide diagnostic information, TP53 mutations in SMZL are associated with a worse prognosis and overall survival, further highlighting the dramatic response in this patient.20 Taken together, this case report highlights the potential for PD-1 directed therapy in the frontline setting for marginal zone lymphoma, the potential value of molecular analysis in identifying residual disease after treatment, and the importance of considering hematopoietic mutations when interpreting solid tumor sequencing data. We believe that a prospective study of PD-1 blockade early in 653

Case Reports

the therapeutic paradigm for marginal zone lymphoma is warranted. Peter G. Miller,1 Adam S. Sperling,1 Christopher J. Gibson,1 Olga Pozdnyakova,2 Waihay J. Wong,2 Michael P. Manos,1 Elizabeth I. Buchbinder,1 F. Stephen Hodi,1 Benjamin L. Ebert,1,3 and Matthew S. Davids1 1 Department of Medical Oncology, Dana-Farber Cancer Institute; 2 Department of Pathology, Brigham and Womenâ&#x20AC;&#x2122;s Hospital and 3 Howard Hughes Medical Institute, Dana-Farber Cancer Institute, Boston, MA, USA Correspondence: MATTHEW S. DAVIDS matthew_davids@DFCI.HARVARD.EDU doi:10.3324/haematol.2020.258426 Disclosures: PGM has received consulting fees from Foundation Medicine, Inc. BLE has received research funding from Celgene, Deerfield, consulting fees from GRAIL, and is on the scientific advisory boards for Exo Therapeutics and Skyhawk Therapeutics. EIB.has received consulting fees from Novartis and Bristol Myers Squibb and clinical trial support from Merck and Bristol Myers Squibb. MSD has received research funding from Astra-Zeneca, Bristol Myers Squibb, Genentech, MEI Pharma, Pharmacyclics, Surface Oncology, TG Therapeutics and Verastem, and consulting fees from AbbVie, Adaptive Biotechnologies, Ascentage Pharma, Astra-Zeneca, BeiGene, Celgene, Genentech, Gilead Sciences, Janssen, MEI Pharma, Merck, Pharmacyclics, Syros Pharmaceuticals, TG Therapeutics, Verastem, and Zentalis. FSH has received research support from Novartis, consulting fees from Novartis, Genentech/Roche, Aduro, Sanofi, Kairos, and Pieris, sits on the advisory board for Takeda, Surface, Compass Therapeutic, Apricity, Pionyr, Verastem, Torque, Rheos, Bicara, and Amgen, and has equity in Apricity, Pionyr, Torque, and Kairos. Full ICMJE disclosure forms are available upon request. Contributions: PGM, ASS, CJG, and MSD designed and conceived the study. MM, SH, EB, OP and WW collected the blood and bone marrow samples and processed the samples. PGM, ASS, CJG, and MSD.analyzed and interpreted the data. PGM, ASS and MSD drafted the manuscript. MSD and BLE obtained funding and supervised the study. Funding: PGM was supported by an Evans Foundation Young Investigator Award and American Society of Hematology Research Training Award for Fellows. ASS was supported by Ruth L. Kirschstein National Research Service Award and a Conquer Cancer Foundation Young Investigator Award. This work was supported by the NIH (R01HL082945 and P01CA108631), the Howard Hughes Medical Institute, the Edward P. Evans Foundation, and the Leukemia & Lymphoma Society to BLE. MSD is a Scholar in Clinical Research from the Leukemia & Lymphoma Society.

References 1. Kluk MJ, Lindsley RC, Aster JC, et al. Validation and implementation of a custom next-generation sequencing clinical assay for hmatologic malignancies. J Mol Diagn. 2016;18(4):507-515.

654

2. Matutes E, Oscier D, Montalban C, et al. Splenic marginal zone lymphoma proposals for a revision of diagnostic, staging and therapeutic criteria. Leukemia. 2008;22(3):487-495. 3. Mollejo M, Lloret E, Menarguez J, Piris MA, Isaacson PG. Lymph node involvement by splenic marginal zone lymphoma: morphological and immunohistochemical features. Am J Surg Pathol. 1997; 21(7):772-780. 4. Kennedy SR, Schmitt MW, Fox EJ, et al. Detecting ultralow-frequency mutations by Duplex Sequencing. Nat Protoc. 2014;9(11):25862606. 5. Bondar G, Cadeiras M, Wisniewski N, et al. Comparison of whole blood and peripheral blood mononuclear cell gene expression for evaluation of the perioperative inflammatory response in patients with advanced heart failure. PLoS One. 2014;9(12):e115097. 6. Garcia EP, Minkovsky A, Jia Y, et al. Validation of OncoPanel: a targeted next-generation sequencing assay for the detection of somatic variants in cancer. Arch Pathol Lab Med. 2017;141(6):751-758. 7. Lui YY, Chik KW, Chiu RW, Ho CY, Lam CW, Lo YM. Predominant hematopoietic origin of cell-free DNA in plasma and serum after sexmismatched bone marrow transplantation. Clin Chem. 2002; 48(3):421-427. 8. Ptashkin RN, Mandelker DL, Coombs CC, et al. Prevalence of clonal hematopoiesis mutations in tumor-only clinical genomic profiling of solid tumors. JAMA Oncol. 2018;4(11):1589-1593. 9. Razavi P, Li BT, Brown DN, et al. High-intensity sequencing reveals the sources of plasma circulating cell-free DNA variants. Nat Med. 2019;25(12):1928-1937. 10. Hu Y, Ulrich BC, Supplee J, et al. False-positive plasma genotyping due to clonal hematopoiesis. Clin Cancer Res. 2018;24(18):44374443. 11. Mohamedali AM, Gaken J, Ahmed M, et al. High concordance of genomic and cytogenetic aberrations between peripheral blood and bone marrow in myelodysplastic syndrome (MDS). Leukemia. 2015;29(9):1928-1938. 12. Lesokhin AM, Ansell SM, Armand P, et al. Nivolumab in patients with relapsed or refractory hematologic malignancy: preliminary results of a phase Ib study. J Clin Oncol. 2016;34(23):2698-2704. 13. Ding W, LaPlant BR, Call TG, et al. Pembrolizumab in patients with CLL and Richter transformation or with relapsed CLL. Blood. 2017;129(26):3419-3427. 14. Panjwani PK, Charu V, DeLisser M, Molina-Kirsch H, Natkunam Y, Zhao S. Programmed death-1 ligands PD-L1 and PD-L2 show distinctive and restricted patterns of expression in lymphoma subtypes. Hum Pathol. 2018;71:91-99. 15. Vranic S, Ghosh N, Kimbrough J, et al. PD-L1 status in refractory lymphomas. PLoS One. 2016;11(11):e0166266. 16. Vincent-Fabert C, Soubeyran I, Velasco V, et al. Inflamed phenotype of splenic marginal zone B-cell lymphomas with expression of PDL1 by intratumoral monocytes/macrophages and dendritic cells. Cell Mol Immunol. 2019;16(6):621-624. 17. Kastenhuber ER, Lowe SW. Putting p53 in context. Cell. 2017;170(6):1062-1078. 18. Boettcher S, Miller PG, Sharma R, et al. A dominant-negative effect drives selection of TP53 missense mutations in myeloid malignancies. Science. 2019;365(6453):599-604. 19. Jaramillo Oquendo C, Parker H, Oscier D, Ennis S, Gibson J, Strefford JC. Systematic review of somatic mutations in splenic marginal zone lymphoma. Sci Rep. 2019;9(1):10444. 20. Parry M, Rose-Zerilli MJ, Ljungstrom V, et al. Genetics and prognostication in splenic marginal zone lymphoma: revelations from deep sequencing. Clin Cancer Res. 2015;21(18):4174-4183.

haematologica | 2021; 106(2)