Haematologica, Volume 105, Issue 10 by Haematologica

haematologica Journal of the Ferrata Storti Foundation

Editor-in-Chief Luca Malcovati (Pavia)

Deputy Editor Carlo Balduini (Pavia)

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), Francesco Rodeghiero (Vicenza), Andreas Rosenwald (Wuerzburg), Davide Rossi (Bellinzona), Jacob Rowe (Haifa, Jerusalem), Wyndham Wilson (Bethesda), Swee Lay Thein (Bethesda)

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); Francesco Lo Coco (Rome); 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 2019 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 105, Issue 10: October 2020

About the cover 2345

100-year-old Haematologica images: Hodgkin and Reed cells in needle aspiration of the spleen Carlo L. Balduini

Editorials 2346

Making fish a little more human: a zebrafish hematopoietic xenotransplant model is improved by the expression of human cytokines Owen J. Tamplin

2348

Pyridoxamine: another vitamin for sickle cell disease? Marilyn J. Telen

2350

ARQ531: the therapy that targets multiple pathways in acute myeloid leukemia Charlotte Hellmich et al.

2352

A step ahead toward precision medicine for chronic lymphocytic leukemia Andrea Patriarca and Gianluca Gaidano

2355

From weakly adhesive to highly thrombogenic: the shear gradient switch Yathreb Asaad and Netanel Korin

Centenary Review Article 2358

Multiple myeloma: the (r)evolution of current therapy and a glance into the future Annamaria Gulla' and Kenneth C. Anderson

Review Articles 2368

The molecular basis for the prothrombotic state in sickle cell disease Arun S. Shet et al.

2380

Thromboinflammatory mechanisms in sickle cell disease â&#x20AC;&#x201C; challenging the hemostatic balance Nicola Conran and Erich V. De Paula

Articles Hematopoiesis

2391

Humanized zebrafish enhance human hematopoietic stem cell survival and promote acute myeloid leukemia clonal diversity Vinothkumar Rajan et al.

Iron Metabolism & its Disorders

2400

Dietary intake of heme iron is associated with ferritin and hemoglobin levels in Dutch blood donors: results from Donor InSight Tiffany C. Timmer et al.

Red Cell Biology & its Disorders

2407

Repurposing pyridoxamine for therapeutic intervention of intravascular cell-cell interactions in mouse models of sickle cell disease Jing Li et al.

Hematologic Neoplasms

2420

The new small tyrosine kinase inhibitor ARQ531 targets acute myeloid leukemia cells by disrupting multiple tumor-addicted programs Debora Soncini et al.

Haematologica 2020; vol. 105 no. 10 - October 2020 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

2432

Myeloproliferative and lymphoproliferative malignancies occurring in the same patient: a nationwide discovery cohort Johanne M. Holst et al.

Chronic Lymphocytic Leukemia

2440

Prognostic and predictive role of gene mutations in chronic lymphocytic leukemia: results from the pivotal phase III study COMPLEMENT1 Eugen Tausch et al.

Plasma Cell Disorders

2448

ATR addiction in multiple myeloma: synthetic lethal approaches exploiting established therapies Oronza A. Botrugno et al.

Platelet Biology & its Disorders

2457

Dominant negative Gfi1b mutations cause moderate thrombocytopenia and an impaired stress thrombopoiesis associated with mild erythropoietic abnormalities in mice Hugues Beauchemin et al.

Hemostasis

2471

Shear rate gradients promote a bi-phasic thrombus formation on weak adhesive proteins, such as fibrinogen in a von Willebrand factor-dependent manner Nicolas Receveur et al.

Coagulation & its Disorders

2484

Murine tissue factor disulfide mutation causes a bleeding phenotype with sex specific organ pathology and lethality Susanna H. M. Sluka et al.

Letters to the Editor e488

Germline biallelic PIK3CG mutations in a multifaceted immunodeficiency with immune dysregulation Marini Thianet al. http://www.haematologica.org/content/105/10/e488

e493

Hemochromatosis proteins are dispensable for the acute hepcidin response to BMP2 Alessia Pagani et al. http://www.haematologica.org/content/105/10/e493

e497

Safe and efficient peripheral blood stem cell collection in patients with sickle cell disease using plerixafor Naoya Uchida et al. http://www.haematologica.org/content/105/10/e497

e502

Elevated TWIST1 expression in myelodysplastic syndromes/acute myeloid leukemia reduces efficacy of hypomethylating therapy with decitabine Hongjiao Li et al. http://www.haematologica.org/content/105/10/e502

e507

High expression of the stem cell marker GPR56 at diagnosis identifies acute myeloid leukemia patients at higher relapse risk after allogeneic stem cell transplantation with the CD34+/CD38- population Madlen Jentzsch et al. http://www.haematologica.org/content/105/10/e507

e512

Brentuximab vedotin in the treatment of elderly Hodgkin lymphoma patients at first relapse or with primary refractory disease: a phase II study of FIL ONLUS Vittorio Stefoni et al. http://www.haematologica.org/content/105/10/e512

Haematologica 2020; vol. 105 no. 10 - October 2020 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation e515

High-throughput analysis of the T-cell receptor gene repertoire in low-count monoclonal B-cell lymphocytosis reveals a distinct profile from chronic lymphocytic leukemia Andreas Agathangelidis et al. http://www.haematologica.org/content/105/10/e515

e519

Impact of idelalisib on health-related quality of life in patients with relapsed chronic lymphocytic leukemia in a phase III randomized trial Paolo Ghia et al. http://www.haematologica.org/content/105/10/e519

e523

Validated single-tube multiparameter flow cytometry approach for the assessment of minimal residual disease in multiple myeloma Sandra Maria Dold et al. http://www.haematologica.org/content/105/10/e523

e531

HLA-DRB1*11 is a strong risk factor for acquired thrombotic thrombocytopenic purpura in children BĂŠrangĂ¨re S. Joly et al http://www.haematologica.org/content/105/10/e531

Case Reports e535

First description of revertant mosaicism in familial platelet disorder with predisposition to acute myelogenous leukemia: correlation with the clinical phenotype Ana C. Glembotsky et al. http://www.haematologica.org/content/105/10/e535

e540

Dramatic presentation of acquired thrombotic thrombocytopenic purpura associated with COVID-19 Marco Capecchi et al. http://www.haematologica.org/content/105/10/e540

Haematologica 2020; vol. 105 no. 10 - October 2020 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

The origin of a name that reflects Europe’s cultural roots.

Ancient Greek

Scientific Latin

Modern English

aÂma [haima] = blood a·matow [haimatos] = of blood lÒgow [logos]= reasoning haematologicus (adjective) = related to blood haematologica (adjective, plural and neuter, used as a noun) = hematological subjects The oldest hematology journal, publishing the newest research results. 2019 JCR impact factor = 7.116

ABOUT THE COVER 100-year-old Haematologica images: Hodgkin and Reed-Sternberg cells in needle aspiration of the spleen Carlo L. Balduini Ferrata-Storti Foundation, Pavia, Italy E-mail: CARLO L. BALDUINI - carlo.balduini@unipv.it doi:10.3324/haematol.2020.265546

he color drawings reproduced on the cover of this issue of Haematologica have been taken from a 1932 article published by Paolo Introzzi, at that time a young assistant of Adolfo Ferrata at the Department of Internal Medicine of the University of Pavia and subsequently Editor in Chief of Haematologica from 1960 to 1973.1 In this paper, Introzzi describes four patients with splenomegaly and without superficial adenomegalies in whom the diagnosis of Hodgkin disease had been made through the cytological study of spleen aspirates. The procedure was without side effects. A few weeks later there was involvement of superficial lymph nodes and the diagnosis of Hodgkin disease was confirmed by their histological examination. At that time, spleen puncture was considered a bold procedure to carry out. However, Introzzi concluded that it does

not present undue risk to the patient in the face of its high diagnostic value. This idea has been recently confirmed by the analysis of large case series of patients undergoing ultrasoundguided fine needle aspiration of the spleen. The conclusion was that the procedure is safe and effective in the diagnosis of lymphoproliferative diseases.2 In addition to reporting the pioneering use of the procedure, the article is notable for the beautiful designs that describe the cytological morphology of Hodgkin and Reed-Sternberg cells in their finest detail.

References 1. Introzzi P. [La puntura della milza]. Haematologica. 1932;13;571-586. 2. Gochhait D, Dey P, Rajwanshi A, et al. Role of fine needle aspiration cytology of spleen. APMIS. 2015;123(3):190-193.

Figures 1 and 2. Hodgkin and Reed-Sternberg cells. These beautiful hand-drawn color plates illustrate an article published in Haematologica in 1932 entitled “La puntura della milza nel granuloma maligno” (Spleen puncture in malignant granuloma).

haematologica | 2020; 105(10)

2345

EDITORIALS Making fish a little more human: a zebrafish hematopoietic xenotransplant model is improved by the expression of human cytokines Owen J. Tamplin Department of Cell & Regenerative Biology, University of Wisconsin-Madison, Madison, WI, USA E-mail: OWEN J. TAMPLIN - tamplin@wisc.edu doi:10.3324/haematol.2020.256909

or decades the mouse xenotransplant model has allowed us to track the behavior of transplanted human cells. Whether it is transplantation of hematopoietic stem and progenitor cells (HSPC) or leukemic cells, immunodeficient mice provide a means of measuring homing, engraftment, and stem cell dynamics.1-3 However, there are limitations to these mouse models that have made it difficult to advance certain aspects of HSPC and leukemia research. For example, immunodeficient mouse models such as NSG (NOD-scid IL2Rγnull)4 are expensive to maintain, making them prohibitive to scaling for screening purposes. Direct intravital imaging can be performed on mouse transplant models,5,6 but because of its technical challenges, this technique is not suitable for use in a large number of animals. To tackle the diversity of human HSPC and leukemias, there is a need for a model that can be housed at high density, easily treated with therapeutics, and directly imaged if necessary. The zebrafish has become well-established as an alternate model that can satisfy these additional research needs.7 Although there are immunodeficient adult zebrafish models available,8 working with embryos [0-3 days post-fertilization (dpf)] and larvae (3-30 dpf) has its own advantages. The zebrafish does not fully develop adaptive immunity until juvenile stage (>30 dpf), making it ideal as a xenotransplant model. Also, the smaller and more transparent embryos are particularly amenable to live imaging. As one adult female can spawn up to 200 embryos per week, and larvae can be maintained at high density, it is a truly scalable model. It is easy to deliver cells into the circulation by microinjection. If injected cells are fluorescent, made so by either genetic modification or dye label, they can be immediately tracked from the injection site as they home to and interact with the niche. One caveat of this model is that the ideal temperature for maintenance of zebrafish is 28.5°C, so a compromise temperature of 35°C must be used in xenotransplants,9 allowing mammalian cells to be compatible with the host environment. Even with these advantages, there has still been room for optimization of the zebrafish as a recipient for human HSPC and leukemic cells. The hematopoietic ontogeny, genetic programs, and cell lineages are highly conserved among vertebrates, making results from zebrafish translate exceptionally well to humans.10,11 However, the cytokines that are essential as regulators of hematopoiesis are not well conserved.12 This led researchers to “humanize” mouse models by introducing human cytokines, either by injection or transgenesis.13,14 Following this approach, but with the added advantages of the zebrafish model, Berman’s group developed a humanized zebrafish that expresses factors critical for support of human HSPC and leukemia cells.15 To develop this model, they selected the human cytokines GM-CSF, SCF/KITLG, and SDF1-α/CXCL12 (named “GSS”) as top candidates to support human cells in the zebrafish model. They expressed these factors in vivo by gen2346

erating a triple transgenic zebrafish line to support transplanted human hematopoietic cells. Human CXCL12 is driven by the zebrafish cxcl12 promoter, and GM-CSF and SCF were doxycycline inducible. Strikingly, expression of all three cytokines had the effect of promoting human cell survival and differentiation. Previous work has been done in Berman’s laboratory to develop zebrafish leukemia xenotransplant models.9,16 Other groups have successfully transplanted adult human CD34+ HSPC17 and mouse HSPC18 into zebrafish larvae. Interestingly, these human cells were found to trigger similar cellular behaviors in the niche as were seen during endogenous zebrafish HSPC lodgement,19 highlighting the similarity between mammalian and zebrafish HSPC. To evaluate their new GSS model over previously developed models, they applied a number of metrics to test its function: (i) migration; (ii) proliferation; (iii) chemotherapy response; (iv) clonality; and (v) host survival. Ultimately, their goal was to make a better microenvironment for human HSPC and leukemic cells. First, to test migration, they injected Jurkat human T-acute lymphoblastic leukemia cells into the yolk of single SDF1-α/CXCL12-overexpressing larvae (S fish) because this cell line expresses high levels of CXCR4. There was little initial response and they reasoned that the injection site was not optimal (yolk and not circulation), and SDF1-α/CXCL12 expression levels of the transgenics were low. To compensate they irradiated recipient larvae and were able to induce a migratory response of transplanted cells in SDF1-α/CXCL12expressing transgenics. Next, to test proliferation, they transplanted a human Down syndrome acute myeloid leukemia (AML) cell line (CMK) into transgenic larvae carrying the doxycycline-inducible GMCSF/CSF2 and SCF/KITLG transgenes (GS fish). Expression of these human factors in the zebrafish larvae proved effective, as CMK cells were more proliferative 3 days post-injection (dpi). However, as in the experiment above, xenotransplanted cells were injected into the yolk, which can have the effect of trapping the cells and reducing their access to circulation. Following these results, the authors proceeded to perform all injections directly into the circulation. The authors then combined transgenics to establish the triple transgenic GM-CSF, SCF/KITLG, and SDF1-α/CXCL12 (GSS) line. As proof-of-concept for drug screening, the authors first chose the chemotherapy medication cytarabine. The CMK cell line they used was derived from a 10-year old Down syndrome patient with AML who was responsive to this drug. Zebrafish embryos and larvae are well-suited for mediumthroughput drug screening because the entire organism can be soaked in drug. These treatments produce an in vivo phenotypic read-out of drug effects. Interestingly, the drug was only effective on xenografts in the GSS line but not in control fish. This response was not fully explained in the study; however, the increased proliferation of leukemic cells from GSS haematologica | 2020; 105(10)

Editorials

cytokines may have made their reduction by chemotherapy treatment more obvious. Alternatively, the GSS cytokines may directly affect the CMK cells, allowing them to have a more physiological response to the drug. Following the promising treatment of CMK xenograft larvae with cytarabine, the authors injected four different primary AML cells into the triple GSS fish at 3 dpf. Consistently, AML cells in GSS fish reduced survival of the host, presumably because cells were able proliferate more in the presence of human cytokines. Diminished survival at 3 dpi was not simply because of the xenotransplant procedure, as introduction of human HSPC into the circulation of zebrafish larvae did not decrease survival. Another measure of these xenografts was clonality, which was determined by error-corrected RNA sequencing. Twice as many individual clones from heterogenous human AML were preserved in the GSS model compared to controls. Together, the above experiments demonstrated that expression of human pro-hematopoietic cytokines in the GSS transgenic model provided a superior microenvironment for xenografts of human HSPC and leukemic cells. This was assessed by increased human cell migration with human CXCL12, increased proliferation with GMCSF/CSF2 and SCF/KITLG, decreased survival of the recipient, better response to chemotherapy, and increased clonality of xenotransplant cells. Despite the many advantages of the zebrafish, there are still some limitations. For example, xenograft human cells only survive for a few days at most. This prevents the long-term tracking of human cell engraftment and disease progression in the host that is possible in mammalian transplant models, such as the mouse1-3 or non-human primate.20 Overall this novel xenotransplant model has many potential applications, from live imaging of leukemia interaction with the niche, to the rapid optimization of patient-specific chemotherapy.

References 1. 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. 2. Lapidot T, Pflumio F, Doedens M, Murdoch B, Williams DE, Dick JE. Cytokine stimulation of multilineage hematopoiesis from immature human cells engrafted in SCID mice. Science. 1992;255(5048):11371141.

haematologica | 2020; 105(10)

3. Larochelle A, Vormoor J, Hanenberg H, et al. Identification of primitive human hematopoietic cells capable of repopulating NOD/SCID mouse bone marrow: implications for gene therapy. Nat Med. 1996;2(12):1329-1337. 4. Ohbo K, Suda T, Hashiyama M, et al. Modulation of hematopoiesis in mice with a truncated mutant of the interleukin-2 receptor gamma chain. Blood. 1996;87(3):956-967. 5. Duarte D, Hawkins ED, Akinduro O, et al. Inhibition of endosteal vascular niche remodeling rescues hematopoietic stem cell loss in AML. Cell Stem Cell. 2018;22(1):64-77. 6. Hawkins ED, Duarte D, Akinduro O, et al. T-cell acute leukaemia exhibits dynamic interactions with bone marrow microenvironments. Nature. 2016;538(7626):518-522. 7. Jing L, Zon LI. Zebrafish as a model for normal and malignant hematopoiesis. Dis Model Mech. 2011;4(4):433-438. 8. Yan C, Yang Q, Do D, Brunson DC, Langenau DM. Adult immune compromised zebrafish for xenograft cell transplantation studies. EBioMedicine. 2019;47:24-26. 9. Corkery DP, Dellaire G, Berman JN. Leukaemia xenotransplantation in zebrafish--chemotherapy response assay in vivo. Br J Haematol. 2011;153(6):786-789. 10. North TE, Goessling W, Walkley CR, et al. Prostaglandin E2 regulates vertebrate haematopoietic stem cell homeostasis. Nature. 2007;447(7147):1007-1011. 11. Cutler C, Multani P, Robbins D, et al. Prostaglandin-modulated umbilical cord blood hematopoietic stem cell transplantation. Blood. 2013;122(17):3074-3081. 12. Brocker C, Thompson D, Matsumoto A, Nebert DW, Vasiliou V. Evolutionary divergence and functions of the human interleukin (IL) gene family. Hum Genomics. 2010;5(1):30-55. 13. Wunderlich M, Chou FS, Link KA, et al. AML xenograft efficiency is significantly improved in NOD/SCID-IL2RG mice constitutively expressing human SCF, GM-CSF and IL-3. Leukemia. 2010;24(10): 1785-1788. 14. Coughlan AM, Harmon C, Whelan S, et al. Myeloid engraftment in humanized mice: impact of granulocyte-colony stimulating factor treatment and transgenic mouse strain. Stem Cells Dev. 2016;25 (7):530-541. 15. Rajan V, Melong N, Wong WH, et al. Humanized zebrafish enhance human hematopoietic stem cell survival and promote acute myeloid leukemia clonal diversity. Haematologica. 2020;105(10:2391-2399. 16. Bentley VL, Veinotte CJ, Corkery DP, et al. Focused chemical genomics using zebrafish xenotransplantation as a pre-clinical therapeutic platform for T-cell acute lymphoblastic leukemia. Haematologica. 2015;100(1):70-76. 17. Hamilton N, Sabroe I, Renshaw SA. A method for transplantation of human HSCs into zebrafish, to replace humanised murine transplantation models. F1000Res. 2018;7:594. 18. Parada-Kusz M, Penaranda C, Hagedorn EJ, et al. Generation of mouse-zebrafish hematopoietic tissue chimeric embryos for hematopoiesis and host-pathogen interaction studies. Dis Model Mech. 2018;11(11):dmm034876. 19. Tamplin OJ, Durand EM, Carr LA, et al. Hematopoietic stem cell arrival triggers dynamic remodeling of the perivascular niche. Cell. 2015;160(1-2):241-252. 20. Koelle SJ, Espinoza DA, Wu C, et al. Quantitative stability of hematopoietic stem and progenitor cell clonal output in rhesus macaques receiving transplants. Blood. 2017;129(11):1448-1457.

2347

Editorials

Pyridoxamine: another vitamin for sickle cell disease? Marilyn J. Telen Wellcome Professor of Medicine, Duke University, Durham, NC, USA E-mail: MARILYN J. TELEN - marilyn.telen@duke.edu doi:10.3324/haematol.2020.257998

he last few years have seen a flurry of activity in the development of pharmacotherapeutics for sickle cell disease (SCD). These efforts have involved the invention and discovery of novel compounds, the development of biologics, and the repurposing of previously available drugs and compounds (reviewed by Telen et al.1). Against this backdrop, Li et al. now present data regarding the possible utility of pyridoxamine, a form of vitamin B6, in preventing vasoocclusion and inflammation in SCD.2 Vitamin B6 is a term referring to any one or more of 6 vitamers alone or in combination, including pyridoxal and its phosphorylated ester pyridoxal 5â&#x20AC;&#x2122; phosphate (and its monohydrate salt), pyridoxine and pyridoxamine, and both of their phosphorylated esters. These six forms are readily interconverted in the body, and pyridoxal 5â&#x20AC;&#x2122; phosphate is an essential cofactor in numerous enzymatic reactions in various tissues. When any phosphorylated form of vitamin B6 is ingested, it is usually hydrolyzed by intestinal phosphatases, and the non-phosphorylated forms are then rapidly absorbed. After absorption, the vitamin can then be phosphorylated and converted into the active form. Historically, pyridoxamine was marketed as a dietary supplement, often as the hydrochoride salt, pyridoxamine dihydrochloride. Pyridoxamine became of interest to diabetologists and those studying complications of diabetes because it inhibits the formation of advanced glycation end products (AGE) from glycated proteins and traps pathogenic reactive carbonyl compounds (Amadori reaction products), which are intermediates in the formation of AGE. In 2009, the US Food and Drug Administration designated pyridoxamine as a pharmaceutical, when it became the active ingredient in Pyridorin, a drug designed by Biostratum Inc., and investigated for possible utility in diabetic nephropathy due to its ability to decrease the production of advanced glycation end products (AGE) in animal models of diabetes.3 However, a clinical study produced disappointing results,4 and studies of vitamin B6 for diabetic nephropathy registered at clinicaltrials.gov appear no longer active. A Cochrane review of available data in 2015 failed to find evidence for improvement of kidney function with vitamin B6 or its derivatives,5 despite several animal studies suggesting benefit. Nonetheless, one report associated plasma levels of advanced glycation end products (pentosidine, N(epsilon) -(carboxymethyl)lysine (CML) and N(epsilon) -(carboxyethyl)lysine (CEL) with SCD organ complications thought to be hemolysis-related.6 At steady state, both pentosidine and CML levels correlated significantly to hemolytic rate, and pentosidine was significantly related to the number of organ complications. Those investigators suggested that increased plasma AGE levels in sickle cell patients might be implicated in the pathophysiology 2348

of the hemolytic phenotype of SCD, with its attendant organ damage.

Vitamins for sickle cell disease Clinical trials currently listed on the clinicaltrials.gov website include 27 studies involving vitamins and nutritional supplements for patients with SCD. Of these, nine list a form of vitamin D as the active (but not always the only active) subject of study. Other trials focus on vitamin A, folic acid (with and without other supplements), vitamin E, niacin, nicotinamide, omega-3 fatty acids, sodium bicarbonate, zinc, Îą-lipoic acid, L-carnitine, and ready to use supplementary food (RUSF), which may contain various protein sources, multiple vitamins, calcium, arginine and citrulline. In many cases, the goal of these therapies is to improve response to oxidative stress, which is markedly increased in SCD.

Multiple targets When viewing the drug development landscape for SCD, most notable, perhaps, is the wide range of therapeutic targets of the newly approved drugs and drugs still in development. Of the three recently approved drugs for SCD, each has a different therapeutic target. One, L-glutamine,7 is a nutritional supplement thought to act by reducing the sequelae of oxidative stress. The second, voxelotor, is an anti-polymerization and anti-sickling agent8 that increases the oxygen affinity of hemoglobin and, by hampering deoxygenation, forestalls sickling and hemolysis and usually raises hematocrit. The third, crizanlizumab,9 inhibits P-selectin interactions and is thought thereby to reduce adhesive and inflammatory cell-cell interactions. Of these three, two, L-glutamine and crizanlizumab, were approved as a consequence of their efficacy in reducing the frequency of vaso-occlusive events, while the third, voxelotor, was approved due to its ability to improve anemia by reducing hemolysis, without regard to any effect on vaso-occlusive events. Disappointingly, none of the studies leading to approval of these drugs involved extensive exploration of the pathophysiologic mechanism affected. In contradistinction, the current paper by Li et al. has explored many mechanistic aspects of potential pyridoxamine effect.2

Targeting oxidative stress An increasing body of literature supports the concept that oxidative stress is an important contributor to the pathophysiology of SCD, including the phenomena of hemolysis, vaso-occlusion and multi-organ damage.10-13 There are multiple mechanisms that tie vaso-occlusion in SCD to oxidative stress and oxidative damage (Figure 1). Pyridoxamine has been shown to facilitate reduction in oxidative stress parameters and reactive oxygen species (ROS) production.14 At physiological pH, pyridoxamine haematologica | 2020; 105(10)

Editorials

Figure 1. Pathophysiology of vaso-occlusion in sickle cell disease. Erythrocytes (RBC) containing predominantly hemoglobin (Hb) S (or HbS with another hemoglobin variant that participates in hemoglobin polymerization) circulate as biconcave discs as well as deformed “sickled” RBC; in addition, there is an abnormally increased number of RBC-derived microparticles in the circulation. Oxidative damage occurs both within the sickle RBC, due to HbS denaturation, instability and auto-oxidation, as well as in tissues in contact with the circulation, due to release of pro-oxidant substances (free Hb and reactive oxygen species [ROS]) by sickle RBC undergoing cell lysis (hemolysis). In the RBC, auto-oxidation of HbS promotes oxidation of βCys93, Hb dimerization and hemichrome formation. Sickle RBC also have retained mitochrondria, which can contribute oxidants that lead to membrane damage, and a diminished complement of anti-oxidant compounds, such as glutathione. Oxidative membrane changes and effect of oxidants on signaling pathways lead to both protein and lipid changes, including exposure of phosphatidylserine (PS) and upregulation of adhesion receptor activity, among other effects. In turn, hemolysis (with release of both Hb and ROS) leads to activation of endothelial cells as well as of cellular blood elements (leukocytes and platelets). In the post-capillary circulation, both sickle RBC and neutrophils adhere to endothelial cells as well as to each other. Platelets also participate in heterocellular aggregate formation. These interactions lead to vaso-occlusion in post-capillary venules. Obstruction of blood flow then results in further HbS deoxygenation,

can most avidly scavenge the •OCH3 radical, in both aqueous and lipidic media, and also has weaker but physiologically relevant ability to trap •OOH and •OOCH3 radicals.14 Pyridoxamine also inhibits AGE formation due to its ability to bind to important enzymes responsible for oxidative reactions in the advanced stages of the protein glycation pathway. Li et al. first determined whether sickle mice who receive pyridoxamine demonstrate clinically relevant beneficial effects. They showed that pyridoxamine reduced neutrophil recruitment to the cremaster venular wall of SCD mice after they were exposed to either hypoxia/reoxygenation or tumor necrosis factor-α (TNFα). Such treatment also improved survival of challenged mice. However, these effects did not appear to be via reduction in AGE, as pyridoxamine reduced endothelial cell and blood cell (neutrophil, platelet) activation states or adhesiveness without affecting the plasma levels of AGE or nitric oxide (NO). Moreover, pyridoxamine appeared equally effective in the presence or absence of hydroxyurea. In vitro studies also confirmed that pyridoxamine appears to have quite selective effects on neutrophils and platelets. It reduced neutrophil degranulation and decreased the surface amount of αMβ2 integrin, a receptor required for the interaction of neutrophils with endothelial cells and platelets, while it did not affect neutrophil L-selectin shedding and ROS production. haematologica | 2020; 105(10)

Furthermore, treatment of SCD mice or mouse platelets in vitro with pyridoxamine inhibited platelet aggregation and ATP secretion after exposure to thrombin or a collagen-related peptide (CRP, a glycoprotein VI-specific agonist) without affecting P-selectin. In vitro experiments with human neutrophils also showed that pyridoxamine reduced ROS generation but did not affect the surface amount of αMβ2 and L-selectin on neutrophils following agonist stimulation.

What more needs to be done? Based on their results, Li et al. suggest that pyridoxamine should be investigated in human SCD to prevent both vaso-occlusion and possibly hemolysis-related organ damage. Although pyridoxamine did not seem to have its effect via reduction of AGE in sickle mice, its protective effect against oxidative challenges may be relevant. In diabetic rats, treatment with pyridoxamine produced a decline in oxidative stress parameters and ROS production. Given that oxidative damage is believed to be critical to the organ damage seen in both diabetes and SCD, further exploration of pyridoxamine in SCD is reasonable, despite the disappointing results in human diabetic patients despite promising data in animal models of diabetes. The article by Li et al. also offers interesting insights into where pyridoxamine may affect the many pathways leading to vaso-occlusion. It is disappointing, 2349

Editorials

however, that they did not report effects of pyridoxamine on the root cause of vaso-occlusion, i.e., the sickle red blood cells (RBC). Especially at this time, when we know clearly that sickle RBC are deficient in anti-oxidant capacity,15 and that oxidative pathways in RBC promote vasoocclusion,13 understanding the effects of anti-oxidant therapies on sickle RBC biology, the root cause of SCD, should be prioritized.

References 1. Telen MJ, Malik P, Vercellotti GM. Therapeutic strategies for sickle cell disease: towards a multi-agent approach. Nat Rev Drug Discov. 2019;18(2):139-158. 2. Li J, Jeong SY, Xiong B, et al. Repurposing pyridoxamine for therapeutic intervention of intravascular cell-cell interactions in mouse models of sickle cell disease. Haematologica. 2020;105(10):2407-2419. 3. Chen JL, Francis J. Pyridoxamine, advanced glycation inhibition, and diabetic nephropathy. J Am Soc Nephrol. 2012;23(1):6-8. 4. Lewis EJ, Greene T, Spitalewiz S, et al. Pyridorin in type 2 diabetic nephropathy. J Am Soc Nephrol. 2012;23(1):131-136. 5. Raval AD, Thakker D, Rangoonwala AN, Gor D, Walia R. Vitamin B and its derivatives for diabetic kidney disease. Cochrane Database Syst Rev. 2015;1:CD009403.

6. Nur E, Brandjes DP, Schnog JJ, et al. Plasma levels of advanced glycation end products are associated with haemolysis-related organ complications in sickle cell patients. Br J Haematol. 2010;151(1):62-69. 7. Niihara Y, Smith WR, Stark CW. A phase 3 trial of L-glutamine in sickle cell disease. N Engl J Med. 2018;379(19):1880. 8. Vichinsky E, Hoppe CC, Ataga KI, et al. A phase 3 randomized trial of voxelotor in sickle cell disease. N Engl J Med. 2019;381(6):509-519. 9. Ataga KI, Kutlar A, Kanter J, et al. Crizanlizumab for the prevention of pain crises in sickle cell disease. N Engl J Med. 2017;376(5):429-439. 10. Belcher JD, Chen C, Nguyen J, et al. Control of oxidative stress and inflammation in sickle cell disease with the Nrf2 activator dimethyl fumarate. Antioxid Redox Signal. 2017;26(14):748-762. 11. Morris CR, Suh JH, Hagar W, et al. Erythrocyte glutamine depletion, altered redox environment, and pulmonary hypertension in sickle cell disease. Blood. 2008;111(1):402-410. 12. Nur E, Biemond BJ, Otten HM, Brandjes DP, Schnog JJ. Oxidative stress in sickle cell disease; pathophysiology and potential implications for disease management. Am J Hematol. 2011;86(6):484-489. 13. Thamilarasan M, Estupinan R, Batinic-Haberle I, Zennadi R. Mn porphyrins as a novel treatment targeting sickle cell NOXs to reverse and prevent acute vaso-occlusion in vivo. Blood Adv. 2020;4(11): 23722386. 14. Ramis R, Ortega-Castro J, Caballero C, et al. How does pyridoxamine inhibit the formation of advanced glycation end products? The role of its primary antioxidant activity. Antioxidants (Basel). 2019;8(9):344. 15. Sangokoya C, Telen MJ, Chi JT. microRNA miR-144 modulates oxidative stress tolerance and associates with anemia severity in sickle cell disease. Blood. 2010;116(20):4338-4348.

ARQ531: the therapy that targets multiple pathways in acute myeloid leukemia Charlotte Hellmich,1,2 Kristian Bowles1,2 and Stuart Rushworth1 1

Norwich Medical School, University of East Anglia, Norwich Research Park, Norwich and 2Department of Haematology, Norfolk and Norwich University Hospitals NHS Trust, Norwich, UK E-mail: STUART RUSHWORTH - s.rushworth@uea.ac.uk doi:10.3324/haematol.2020.257022

o far this century we have witnessed the introduction of a number of targeted therapies, developed through rational drug design, which have changed cancer treatment and resulted in improved outcomes for many patients, including those with a spectrum of chronic lymphoid and myeloid malignancies.1,2 However, despite improved understanding of the biology of acute myeloid leukemia (AML), similar scale benefits by targeting kinases and other intracellular and surface proteins have yet to be realized, and the prognosis for patients with AML remains poor. Moreover, cytotoxic drugs and therapies developed in the last century currently remain the backbone of AML treatment, and as AML primarily affects the elderly, many of whom are therefore frail with multiple co-morbidities, the clinical application of such curative therapies is somewhat limited.3 Furthermore, even in those fit enough for intensive chemotherapy, both relapse and treatment resistance are common, due to the aggressive nature of the disease. The search therefore continues for biology-driven targeted treatments for patients with AML which can be delivered to all, and at the same time increase remission rates, reduce relapses and prevent treatment resistance. The expectation is that these therapies will come from advances in the understanding of the biology of AML. 2350

Tyrosine kinases are known to play a role in the tumorigenesis of both solid tumors and hematological malignancies and they therefore present a potential treatment target.4 In particular, Bruton tyrosine kinase (BTK) inhibitors have been shown to be effective in the treatment of a number of hematologic malignancies including chronic lymphocytic leukemia and lymphomas.5-7 BTK is a non-receptor tyrosine kinase with an important role in both normal and malignant hematopoiesis.8 Its expression and phosphorylation are high in AML and BTK inhibition with ibrutinib has been shown to have an anti-leukemic effect.9 Moreover, many other tyrosine kinases have been shown to be activated in AML and hematologic malignancies including SYK, FLT3, MAPK, PI3K and AKT.10,11 This knowledge supports further exploration of tyrosine kinase inibitors in the treatment of AML. In this issue of Haematologica, Soncini et al. explore the potential role of ARQ531, a reversible small molecule inhibitor of BTK and a number of other kinases, in the management of AML12 (Figure 1). BTK was shown to be constitutively active in a range of different subtypes of AML, suggesting that targeting it could have a broad clinical application in all patients with AML. ARQ531 was observed to have dose-dependent anti-leukemic activity by inducing apoptosis in both AML cell lines and primary AML cells. haematologica | 2020; 105(10)

Editorials

Figure 1. Schematic representation of the mechanism of action of ARQ531 in acute myeloid leukemia. BTK: Bruton tyrosine kinase; ERK: extracellular-signal-related kinase; MAPK: mitogen-activated protein kinase; AML: acute myeloid leukemia.

The effect of ARQ531 was superior to that of treatment with ibrutinib and this was likely due to the broader mode of action of ARQ531. While ARQ531 has anti-BTK activity its action does not appear to be dependent on BTK and it was shown to inhibit a number of other oncogenic pathways as well. Thus, ARQ531 treatment reduced cell viability even after knock-down of BTK in AML cells. We know that AML is heavily reliant on its tumor microenvironment13 and disrupting this relationship is a crucial step in improving AML treatments and therefore outcomes for patients. When considering effectiveness of new treatments, we must therefore assess whether they are able to target leukemic cells within this supportive environment. Soncini et al. were able to demonstrate that the antileukemic activity of ARQ531 was preserved when AML cells were co-cultured with mesenchymal stromal cells. Moreover, the viability of the mesenchymal stromal cells as well as that of other non-malignant cells, including hematopoietic stem cells, was not affected by treatment with ARQ531. This is important when considering the potential toxicity of a drug as there is a need to protect nonmalignant cells in order to prevent side effects and reduce treatment-related morbidity and mortality. A number of oncogenic pathways have been shown to drive leukemogenisis, disease progression and treatment resistance. In addition to upregulation of BTK, these include activation of the mitogen-activated protein kinase (MAPK) pathway and dysregulation of the hematopoietic transcriptional factor MYB, which have been implicated in the haematologica | 2020; 105(10)

pathogensis of AML.14,15 The mode of action of ARQ531 is complex and genetic analysis revealed that it inhibits a number of different oncogenic pathways in AML. The data published in this issue show that ARQ531 disrupts the oncogenic MAPK pathway by inhibiting ERK and AKT activation as well as downregulating MYC. Furthermore, ARQ531 was shown to deregulate and promote degradation of the oncogenic driver MYB. Interestingly the action on all of these pathways, including BTK, MAPK and MYB, as well, potentially, as others not yet identified, appears to have a cumulative effect on cell viability. Inhibition of each pathway individually is less effective in reducing cell viability than using ARQ531, suggesting that additional pathways are important in this drugâ&#x20AC;&#x2122;s mode of action. It is clear that leukemogenesis relies on the dysregulation of multiple oncogenic pathways and therefore targeting only one of these is not sufficient to achieve cell death and reduce tumor growth. A drug that has the ability to target a number of these pathways does, therefore, have greater potential to effectively eliminate malignant cells and at the same time has little or no lasting effect on non-malignant cells. To demonstrate this Soncini et al. designed an in vivo experiment using a patient-derived AML mouse model. Following engraftment of immunocompromised mice with primary AML cells they demonstrated that treatment with ARQ531 impaired AML engraftment, reduced tumor burden and improved the animalsâ&#x20AC;&#x2122; survival. Furthermore, there were no signs of toxicity, suggesting that ARQ531 could be an effective and well-tolerated treatment for AML in the 2351

Editorials

future. However, a note of caution is that many potential therapies have been shown to have in vivo efficacy in AML, but when tested clinically have had little or no effect on this disease. In conclusion, by targeting a number of different oncogenic pathways, in vitro and in vivo treatment with ARQ531 results in reduced AML cell viability, reduced tumor growth and improved survival of animals. The research by Soncini et al. suggests that a multi-targeted inhibitor such as ARQ531 is required to impair AML survival effectively; since this drug does not rely specifically on high expression of BTK or other tyrosine kinases it could be widely applicable to different subtypes of AML.

References 1. Longo DL. Imatinib changed everything. N Engl J Med. 2017;376(10):982-983. 2. Yosifov DY, Wolf C, Stilgenbauer S, Mertens D. From biology to therapy: the CLL success story. HemaSphere. 2019;3(2):e175. 3. Juliusson G, Antunovic P, Derolf A, et al. Age and acute myeloid leukemia: real world data on decision to treat and outcomes from the Swedish Acute Leukemia Registry. Blood. 2009;113(18):4179-4187. 4. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011;144(5):646-674. 5. 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. 6. Wang ML, Rule S, Martin P, et al. Targeting BTK with ibrutinib in relapsed or refractory mantle-cell lymphoma. N Engl J Med. 2013;369(6):507-516. 7. Treon SP, Tripsas CK, Meid K, et al. Ibrutinib in previously treated Waldenström's macroglobulinemia. N Engl J Med. 2015;372(15): 1430-1440. 8. Rushworth SA, Pillinger G, Abdul-Aziz A, et al. Activity of Bruton's tyrosine-kinase inhibitor ibrutinib in patients with CD117-positive acute myeloid leukaemia: a mechanistic study using patient-derived blast cells. Lancet Haematol. 2015;2(5):e204-211. 9. Rushworth SA, Murray MY, Zaitseva L, Bowles KM, MacEwan DJ. Identification of Bruton's tyrosine kinase as a therapeutic target in acute myeloid leukemia. Blood. 2014;123(8):1229-1238. 10. Chalandon Y, Schwaller J. Targeting mutated protein tyrosine kinases and their signaling pathways in hematologic malignancies. Haematologica. 2005;90(7):949-968. 11. Fernandez S, Desplat V, Villacreces A, et al. Targeting tyrosine kinases in acute myeloid leukemia: why, who and how? Int J Mol Sci. 2019;20(14). 12. Soncini D, Orecchioni S, Ruberti S, et al. The new small molecule tyrosine-kinase inhibitor ARQ531 targets acute myeloid leukemia cells by disrupting multiple tumor-addicted programs. Haematologica. 2020; 105(10):2420-2431. 13. Shafat MS, Gnaneswaran B, Bowles KM, Rushworth SA. The bone marrow microenvironment - home of the leukemic blasts. Blood Rev. 2017;31(5):277-286. 14. Milella M, Kornblau SM, Estrov Z, et al. Therapeutic targeting of the MEK/MAPK signal transduction module in acute myeloid leukemia. J Clin Invest. 2001;108(6):851-859. 15. Frech M, Teichler S, Feld C, et al. MYB induces the expression of the oncogenic corepressor SKI in acute myeloid leukemia. Oncotarget. 2018;9(32):22423-22435.

A step ahead toward precision medicine for chronic lymphocytic leukemia Andrea Patriarca and Gianluca Gaidano Division of Hematology, Department of Translational Medicine, University of Eastern Piedmont and Maggiore Charity Hospital, Novara, Italy E-mail: GIANLUCA GAIDANO - gianluca.gaidano@uniupo.it doi:10.3324/haematol.2020.257048

he concept of precision medicine applied to human tumors implies the personalized tailoring of clinical management and treatment choices according to the status of an array of molecular biomarkers, in conjunction with other patient features.1 In chronic lymphocytic leukemia (CLL), the extensive body of genetic data that have been accumulated in recent years has led to the identification of many new molecular biomarkers with prognostic value. However, only a few of these serve the role of true predictors for choosing the most appropriate treatment for any given patient.1,2 The active search for molecular predictors in CLL is becoming increasingly more important in the current therapeutic landscape of the disease, that ranges from chemo-immunotherapy with both old and newer monoclonal antibodies (mAb) to chemo-free options based on B-cell receptor (BCR) inhibitors, targeting either Bruton tyrosine kinase or phosphatidyilinositol-3-kinase, and BCL2 inhibitors.3,4 In this issue of Haematologica, Tausch et al. have analyzed the prognostic and, more importantly, the predictive role of a panel of gene mutations in the randomized, 2352

phase III COMPLEMENT1 trial comparing chlorambucil with ofatumumab-chlorambucil in treatment-naïve CLL patients not eligible for intensive therapy because of age or comorbidities.5 The COMPLEMENT 1 trial had documented that addition of the type 1 anti-CD20 mAb ofatumumab to chlorambucil leads to clinically significant improvement in progression-free survival (PFS) (22.4 months in the arm treated with ofatumumab chlorambucil vs. 13.1 months in the arm treated with single agent chlorambucil), with a manageable side effect profile.6 But whether ofatumumab provided an advantage to all molecular subgroups of CLL remains unexplored. Remarkably, in the genetic analysis performed by Tausch et al., mutations of NOTCH1 were seen to predict weak benefit from the addition of ofatumumab to the chlorambucil backbone.5 The NOTCH1 signaling pathway is a key feature in CLL growth and survival, and is deregulated by mutations in a sizable fraction of CLL7 (Figure 1). NOTCH1 mutations in CLL may target either the autoregulatory PEST domain, or the non-coding 3’untranslated region (3’-UTR) sequence.7 In the context of haematologica | 2020; 105(10)

Editorials

the COMPLEMENT1 trial, the addition of ofatumumab to chlorambucil provided a significant benefit in PFS to NOTCH1 wild-type patients, whereas no statistically significant benefit was achieved in NOTCH1 mutated cases, including patients whose mutations disrupted the NOTCH1 PEST autoregulatory domain as well as patients with NOTCH1 mutations affecting the 3â&#x20AC;&#x2122;-UTR of the gene.5 The refractoriness to ofatumumab imparted by NOTCH1 mutations is reminiscent of the refractoriness to another type 1 anti-CD20 mAb, namely rituximab, that had been observed in the CLL8 trial comparing fludarabine-cyclophosphamide with fludarabine-cyclophosphamide-rituximab (FCR) in young and fit CLL patients.8 In fact, in the CLL8 trial, rituximab failed to improve response and survival in patients carrying NOTCH1

mutations.8 The fact that NOTCH1 mutations behave as a predictor of reduced benefit from type 1 anti-CD20 mAb in two prospective, randomized trials with different anti-CD20 antibodies (ofatumumab in COMPLENT1; rituximab in CLL8), different chemotherapy backbones (chlorambucil in COMPLEMENT1; fludarabinecyclophosmide in CLL8), and different target CLL populations (patients not eligible to intensive therapy in COMPLEMENT1; patients eligible to fludarabine-containing regimens in CLL8) contributes further to the robustness of the predictive significance of NOTCH1 mutations in CLL treated with chemo-immunotherapy containing anti-CD20 type 1 mAb.5,8 The obvious question is whether the novel, type 2 antiCD20 mAb in use for CLL, namely obinutuzumab, may overcome the refractoriness imparted by NOTCH1 muta-

Figure 1. NOTCH1 signaling pathway and effects of NOTCH1 mutations on CLL susceptibility to anti-CD20 mAb. In the context of a wild-type NOTCH1 gene (left panel), ligands (DLL -1, -3, -4 belonging to the Delta-like family or JAGGED -1, -2 belonging to the Serrate family) expressed by stromal cells and by antigen presenting cells (APC) bind to the extracellular portion of the NOTCH1 receptor on CLL cells. Ligand-receptor binding triggers sequential cleavages of the NOTCH1 receptor mediated by the ADAM10 metalloprotease and the S3 Îł-secretase. As a consequence, the IntraCellular NOTCH1 (ICN) domain is free to translocate to the nucleus, where it interacts with RBPJ and other co-activators to induce transcription of target genes promoting cell growth and survival and other cellular programs. The signaling cascade is terminated by ubiquitinylation of the NOTCH1 autoregulatory PEST domain, that is mediated by the FBW7 complex and leads to ICN degradation in the proteasome. In CLL cells with wild type NOTCH1 genes, type 1 anti-CD20 antibodies (rituximab, ofatumumab) induce cell death in vitro and, in vivo contribute to better patient outcomes in patients treated with chemo-immunotherapy. NOTCH-1 mutations occur in a sizeable fraction of CLL (right panel), upregulate NOTCH1 signaling and lead to increased expression of target genes. Most mutations in CLL disrupt the PEST domain, reducing proteasomal degradation of ICN and stabilizing ligand-triggered NOTCH1 signaling. Type 1 anti-CD20 mAb are less efficacious against NOTCH1 mutated CLL cells both in vivo and in vitro. The exact mechanism of anti-CD20 refractoriness associated with NOTCH1 mutations is not fully understood, but has been suggested to be linked, at least in part, to downregulation of CD20 expression.

haematologica | 2020; 105(10)

2353

Editorials

tions to anti-CD20 therapy. This may be possible, since the glycoengineered type 2 anti-CD20 obinutuzumab exploits a different mode of action, based on enhanced antibody-dependent cell-mediated cytotoxicity and increased direct cell death compared to the type 1 antiCD20 mAb rituximab and ofatumumab.9 Preliminary data seem to suggest that obinutuzumab might be able to overcome such refractoriness in the CLL11 trial comparing obinutuzumab-chlorambucil with rituximab-chlorambucil.10 Guidelines for CLL still recommend chemoimmunotherapy as a therapeutic option despite the advent of BCR and BCL2 inhibitors.11 In this context, knowledge of NOTCH1 mutation status might be important in clinical decision-making whenever a chemoimmunotherapy regimen containing an anti-CD20 mAb is being offered to patients. The evidence acquired so far on anti-CD20 refractoriness and NOTCH1 mutations would support the concept that, in the presence of a mutated NOTCH1 gene, the use of a chemo-immunotherapy regimen containing a type 1 anti-CD20 mAb may not be the most appropriate choice and might be replaced by one of the many other therapeutic options that are currently available for CLL.5,7,11 Recommendations by guidelines on this specific issue are desirable at this stage. The use of anti-CD20 mAbs in CLL is not limited to chemo-immunotherapy regimens both in treatmentnaïve and in relapsed/refractory patients. For example, the MURANO trial has shown the superiority of venetoclax-rituximab compared to bendamustine-rituximab in relapsed/refractory CLL.12 The CLL14 trial has documented that venetoclax-obinutuzumab associates with longer PFS compared to chlorambucil-obinutuzumab in treatment-naïve CLL.13 The iLLUMINATE trial has shown the advantage of ibrutinib-obinutuzumab over chlorambucil obinutuzumab as first-line therapy.14 Ibrutinib-rituximab is superior to chemo-immunotherapy in an Eastern Cooperative Oncology Group (ECOG) trial devoted to treatment-naïve CLL.15 At present, it is not known whether the reduced efficacy of type 1 anti-CD20 mAbs observed in NOTCH1 mutated patients treated with chemo-immunotherapy would also be a feature of novel chemo-free regimens based on BCR or BCL2 inhibitors in combination with an anti-CD20 mAb. The precise molecular mechanism through which NOTCH1 mutations confer resistance to anti-CD20 type 1 mAb remains, to a certain extent, elusive (Figure 1). Though the biological relationship between NOTCH1 mutation expression and CD20 cell surface expression was not a specific focus of the report by Tausch et al., measuring CD20 levels by flow cytometry in the COMPLEMENT1 trial population failed to reveal differences between NOTCH1 mutated and wild-type cases.5 Conversely, in a wide CLL series of almost 700 cases, CLL cells from cases harboring mutations of the NOTCH1 PEST domain showed lower CD20 expression compared to NOTCH1 wild-type cases.16 Reduced surface expression of CD20 appears to be a feature also of CLL cases harboring a different type of NOTCH1 mutations affecting the 3’-UTR of the gene.17 Lower CD20 expression on the cell surface of CLL cells has been shown to be coupled to lower mRNA levels of the MS4A1 gene that encodes 2354

the CD20 antigen.16 As a consequence, cell lysis induced by anti-CD20 type 1 antibodies, namely rituximab and ofatumumab, appears to be also lower in NOTCH1 mutated cases compared to CLL without this genetic lesion.16 Consistent with these observations, pharmacological inhibition of the NOTCH1 protein or siRNA silencing of the NOTCH1 gene have been shown to induce upregulation of the CD20 molecule on CLL cells.16 It is well known that several epigenetic and transcription factors regulate expression of the MS4A1 gene and of the CD20 antigen.18 Interestingly, mutations of the NOTCH1 intracellular domain lead to accumulation of mutated NOTCH1 in the nucleus and may alter the fine epigenetic regulation of MS4A1 and CD20 expression through interactions with the RBPJ transcription factor that is involved in the NOTCH1 signaling pathway.16,18 Overall, the biological relationship between NOTCH1 signaling, its deregulation by mutations and expression of CD20 requires further investigation, ideally in study designs aimed at comparing different type 1 and type 2 anti-CD20 mAb in order to understand not only the mechanisms of resistance, but also the strategies to overcome such refractoriness. It should also be considered that NOTCH1 belongs to a molecular pathway and that mutations in B-cell malignancies may also target other players of the pathway.6 Because these genetic alterations either potentiate positive signals or compromise negative regulators of NOTCH1, it would be interesting to understand whether alterations of other NOTCH1 pathway genes, in addition to NOTCH1 itself, might have an effect on anti-CD20 mAb response in vitro and in vivo. The clinical management and therapeutic landscape of CLL have changed substantially over the last few years and continue to evolve. The availability of a variety of treatment options, ranging from chemo-immunotherapy to molecular inhibitors of the BCR and BCL2 pathways, has generated the need to search for robust biomarkers that may assist clinicians in choosing the most suitable and sustainable treatment strategy for every patient. Guidelines recommend TP53 disruption and IGHV mutation status as molecular predictors and these are commonly used when choosing treatment.19 Tausch et al. now consolidate NOTCH1 mutation status as a novel potential biomarker for optimizing anti-CD20 treatment when a chemo-immunotherapy option is offered to patients.5 Other predictive biomarkers are also emerging, and include loss of function mutations of BIRC3, that deregulate the NFkB pathway and confer resistance or reduced efficacy with chemo-immunotherapy regimens,20,21 as well as use of specific stereotyped BCR subsets, in particular subset #2, as observed in the correlative analysis of multicentric clinical trials.22 Step by step, precision medicine is becoming a solid reality in the field of CLL for the benefit of patients and to optimize allocation of resources in clinical practice. At present, the available therapeutic options for CLL that are recommended by guidelines have not always been subjected to rigorous and multiple head-to-head prospective comparisons, thus leaving several unanswered questions when physicians and patients need to make a treatment choice. Choosing wisely, based on robust molecular predictors, coupled to the patient’s fitness and comorbidities, haematologica | 2020; 105(10)

Editorials

might represent a viable and clinically meaningful strategy for achieving the best therapeutic outcome for the individual patient and to satisfy the need to optimize resources for the patient community.

References 1. Moia R, Patriarca A, Schipani M, et al. Precision medicine management of chronic lymphocytic leukemia. Cancers. 2020;12(3):642. 2. Condoluci A, Rossi D. Genetic mutations in chronic lymphocytic leukemia: impact on clinical treatment. Expert Rev Hematol. 2019;12(2):89-98. 3. Awan FT, Al-Sawaf O, Fischer K, Woyach JA. Current perspectives on therapy for chronic lymphocytic leukemia. Am Soc Clin Oncol Educ Book. 2020;40:1-10. 4. Moia R, Patriarca A, Deambrogi C, et al. An update on: molecular genetics of high-risk chronic lymphocytic leukemia. Expert Rev Hematol. 2020;13(2):109-116. 5. Tausch E, Beck P, Schlenk RF, et al. Prognostic and predictive role of gene mutations in chronic lymphocytic leukemia: results from the pivotal phase III study COMPLEMENT1. Haematologica. 2020;105(10):2440-2447. 6. Hillmen P, Robak T, Janssens A, et al. Chlorambucil plus ofatumumab versus chlorambucil alone in previously untreated patients with chronic lymphocytic leukaemia (COMPLEMENT 1): a randomised, multicentre, open-label phase 3 trial. Lancet. 2015;385(9980):18731883. 7. Arruga F, Vaisitti T, Deaglio S. The NOTCH pathway and its mutations in mature B cell malignancies. Front Oncol. 2018;8:550. 8. Stilgenbauer S, Schnaiter A, Paschka P, et al. Gene mutations and treatment outcome in chronic lymphocytic leukemia: results from the CLL8 trial. Blood. 2014;123(21):3247-3254. 9. Prica A, Crump M. Improving anti-CD20 antibody therapy: obinutuzumab in lymphoproliferative disorders. Leuk Lymphoma. 2019;60(3):573:582. 10. Estenfelder S, Tausch E, Robrecht S, et al. Gene mutations and treatment outcome in the context of chlorambucil (Clb) without or with the addition of rituximab (R) or obinutuzumab (GA-101, G) - results of an extensive analysis of the phase III study CLL11 of the German CLL Study Group. Blood. 2016;128(22):3227.

11. Wierda WG, Byrd JC, Abramson JS, et al. Chronic lymphocytic leukemia/small lymphocytic lymphoma, version 4.2020, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw. 2020;18(2):185-217. 12. Seymour JF, Kipps TJ, Eichhorst B, et al. Venetoclax-rituximab in relapsed or refractory chronic lymphocytic leukemia. N Engl J Med. 2018;378(12):1107-1120. 13. Fischer K, Al-Sawaf O, Bahlo J, et al. Venetoclax and obinutuzumab in patients with CLL and coexisting conditions. N Engl J Med. 2019;380(23):2225-2236. 14. Moreno C, Greil R, Demirkan F, Tedeschi A, 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. 15. Shanafelt TD, Wang XV, Kay NE, et al. Ibrutinib-rituximab or chemoimmunotherapy for chronic lymphocytic leukemia. N Engl J Med. 2019; 381(5): 432-443. 16. Pozzo F, Bittolo T, Arruga F, et al. NOTCH1 mutations associate with low CD20 level in chronic lymphocytic leukemia: evidence for a NOTCH1 mutation-driven epigenetic dysregulation. Leukemia. 2016;30(1):182-189. 17. Bittolo T, Pozzo F, Bomben R, et al. Mutations in the 3' untranslated region of NOTCH1 are associated with low CD20 expression levels chronic lymphocytic leukemia. Haematologica. 2017;102(8):e305e309. 18. Pavlasova G, Mraz M. The regulation and function of CD20: an "enigma" of B-cell biology and targeted therapy. Haematologica. 2020;105(6):1494-1506. 19. 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. 20. Diop F, Moia R, Favini C, et al. Biological and clinical implications of BIRC3 mutations in chronic lymphocytic leukemia. Haematologica. 2020; 105(2):448-456. 21. Tausch E, Schneider C, Robrecht S, et al. Prognostic and predictive impact of genetic markers in patients with CLL treated with obinutuzumab and venetoclax. Blood. 2020;135(26):2402-2412. 22. Jaramillo S, Agathangelidis A, Schneider C, et al. Prognostic impact of prevalent chronic lymphocytic leukemia stereotyped subsets: analysis within prospective clinical trials of the German CLL Study Group (GCLLSG). Haematologica. 2019 Dec 26. [Epub ahead of print]

From weakly adhesive to highly thrombogenic: the shear gradient switch Yathreb Asaad and Netanel Korin Faculty of Biomedical Engineering, Technion-Israel Institute of Technology, Haifa, Israel E-mail: NETANEL KORIN - korin@bm.technion.ac.il doi:10.3324/haematol.2020.257030

ormation of a blood clot within an artery, is a complex process orchestrated by numerous chemical and physical factors, including: platelets, endothelium, subendothelial matrix, soluble blood proteins involved in hemostasis e.g., fibrinogen and von Willebrand factor (VWF) and blood flow.1 The pivotal role of flow characteristics in thrombosis and hemostasis has been well recognized in the field, as blood flow regulates the physical environment of the clotting process and the transport of molecules and blood cells.2, 3 More specifically, in vivo and in vitro studies under constant flow highlighted wall shear rate, the spatial rate of change in velocity near the wall which affects transport and friction forces near the wall, as a key parameter controlling the thrombosis processes.4â&#x20AC;&#x201C; 7 Under physiological conditions, wall shear is tightly reghaematologica | 2020; 105(10)

ulated in the arterial vascular system. However, under pathological conditions, such as arterial stenosis, wall shear rate can increase significantly above its physiological level.8 Thus, the study of thrombosis under pathological high wall shear rates has received considerable attention and has uncovered important shear dependent processes such as platelet shear activation and VWF unfolding.9,10 However, unlike constant wall shear conditions, in stenotic sites the flow is complex and the wall shear rate changes dramatically at the flow acceleration and deceleration zones.11 Several studies have investigated platelet aggregation mechanisms under complex shear gradient to emphasize the key role of disturbed hemodynamics in thrombus cascade.12 One important study in this field was conducted by Nesbitt et al., Nature 2355

Editorials

Figure1. Bi-phasic model of thrombus formation cascade at stenotic arteries under a shear gradient. (A)Wall shear rate gradually elevates within an arterial stenosis. The shear gradient alters the mechanosensitive von Willebrand factor (VWF) conformational state from globular to elongated. In atherosclerotic lesions, the extracellular matrix becomes exposed, possibly displaying weakly-adhesive proteins like fibrinogen. At these sites, thrombus is initiated when the elongated VWF adheres to the exposed fibrinogen thus creating a thrombogenic surface. Thenceforth, activated platelets from the blood stream are recruited to the vessel wall, to adhere and accumulate. (B)Thrombus formation under gradient shear is bi-phasic. Phase I: individual platelets slowly adhere to VWF fibers bound to a weakly adhesive protein. Phase II: platelets rapidly aggregate and VWF elongated fibers are formed which facilitate thrombus formation.

Medicine 2009, where platelets aggregation was investigated under shear gradient hemodynamics both in vivo and in vitro.1 The study showed that local shear micro-gradients as a consequence of vessel geometry or thrombus formation, encourage the formation of platelets aggregation.1 Despite the important findings of this study, only type 1 fibrillar collagen and VWF-coated surfaces were examined, thus the possible role of other proteins, such as fibrinogen, under disturbed flow is yet to be deciphered. Thenceforth, another prominent study conducted by Westein et al., PNAS 2012, aimed to investigate stenosisdependent platelet aggregation both in vitro and in vivo. Westein et al. designed stenotic flow channels with VWF/fibrinogen-coated surfaces to represent the surface of growing thrombi and the shear rate gradient at stenotic atherosclerotic plaque sites. The combination of the physical and the biological recapitulation of the stenotic microenvironment, showed that activated platelets, VWF/fibrinogen surface combined with shear alterations at a stenosis outlet is sufficient for a strong pro-aggregatory effect.9 Despite the valuable findings of this work, it centrally highlights the role of the mechano-sensitive VWF in arterial thrombosis while less emphasis was 2356

given to the weakly-adhesive proteins role in the thrombosis cascade e.g., fibrinogen. More generally, there is still a lack of mechanistic understanding of the physical interaction between VWF and fibrinogen at stenotic sites where complex flow patterns exist. Published in this issue of Haematologica, the study by Receveur et al., addresses this important gap of knowledge and highlights the role of weakly adhesive surfaces, e.g., fibrinogen, on thrombus formation under a shear gradient in stenotic vessels.13 Unlike most of the studies in this field, which focused on platelets behavior on an immobilized collagen or collagen/VWF surface, Receveur et al. comprehensively describe platelets adhesion mechanism under gradient shear rate on a fibrinogen surface. Surprisingly, unlike constant flow/shear conditions, under a shear gradient, weakly-adhesive protein surfaces become highly thrombogenic, see Figure 1. Moreover, the authors describe the two phases of thrombus formation under these conditions where a slow single platelet adhesion phase is followed by a rapid aggregation phase, as shown in Figure 1. In their study, Receveur et al. used symmetric stenotic channels with a 90% lumen reduction to create a gradient haematologica | 2020; 105(10)

Editorials

shear stress profile with ascending shear stress values. The channels were coated with fibrinogen to simulate the early stages of thrombus formation when an extracellular matrix protein array is exposed. Whole blood with fluorescently labeled platelets was perfused over these channels while visualizing their spatial adhesion in real-time. To validate the results, platelets were perfused on straight channels that are characterized by a uniform flow regime and also on passivated channels by albumin. To validate the results in vivo, an experimental thrombosis approach was performed. An iron chloride solution and a micromanipulator were used to injure the murine carotid artery vessel wall and to create a severe stenosis which produced a shear stress gradient, respectively. Using real-time video-microscopy, Receveur et al. succeeded in revealing several interesting observations. Under a uniform low wall shear rate, (<1,000 s-1), single platelets were recruited to fibrinogen-coated channel. However, when the shear rate was elevated, (>1,000 s-1), platelets failed to adhere. This indicates that fibrinogen has a limited affinity to its platelet ligand, integrin ÎąIIbÎ˛3, which under high wall shear rate cannot support platelets adhesion. On the other hand, when platelets were perfused in the stenotic channel, they adhered to the immobilized fibrinogen. This phenomenon was not observed, notwithstanding, when platelets experienced a similar wall shear rate in a straight channel. This unique behavior proves that platelets adhesion to immobilized fibrinogen is mediated by gradient shear due to the stenotic geometry. Another pivotal finding of this paper is the unique interaction of VWF and fibrinogen under flow. Receveur et al. show that VWF-fibrinogen interaction is shear-gradient related. In the absence of disturbed flow, the VWF-fibrinogen complex has minimum bonds to support platelets adhesion. However, when shear gradient exists, the mechano-sensitive VWF unfolds, exposes its shielded domains, and adheres to the immobilized exposed fibrinogen. Thenceforth, the sticky VWF recruits more platelets from the blood stream. This indicates that the critical step in thrombus escalation is shear gradient dependent, which forms due to a geometrical alteration at stenotic vessels. This study also introduces a model of bi-phasic thrombus formation under shear rate gradients. The model suggests the existence of two different stages in the thrombus formation cascade: phase I and II, see Figure 1. Phase I is characterized by the slow and heterogeneous recruitment of individual platelets while phase II is characterized by long VWF sticky fibers that very rapidly contribute to platelet aggregation. More importantly, the biphasic thrombus cascade occurs only for elevated values of shear rate gradient while VWF-self association may play a crucial rule in it. Despite the important contribution of this paper in terms of deciphering arterial thrombosis, there are some limitations. Studies have shown that parallel flow chambers and microfluidics do not give accurate results as they fail to precisely recapitulate real-scale hemodynamic forces in human arteries, important flow parameters and mass transport characteristics.14 Additionally, it is worth noting that experiments were conducted under constant

haematologica | 2020; 105(10)

flow disregarding the physiological pulsatile blood flow that exists in human arteries. Therefore, the microfluidicscale observations do not necessarily reflect the human macro-scale phenomena. Accordingly, it would be interesting to validate these observations under physiological hemodynamic conditions. Moreover, hemodynamic parameters which are highly dependent on the vessel diameter and the flow pulse, prevent us from extrapolating data from in vivo mouse experimental conditions which are different from humans not only in their biology but also in their underlying physics.15 To conclude with, this study offers a novel perspective regarding the reciprocal interactions along the axis: platelets-VWF-fibrinogen. A comprehensive outlook of these interactions along the thrombosis axis, may pave the way toward major implications for arterial thrombosis treatment and platelets adhesion down-regulation in disease conditions, which have not been examined so far. Moreover, better understanding of the processes affected by complex flow can also be valuable for the proper design of extracorporeal systems and vascular implantable devices.

References 1. Nesbitt W, Westein E, Tovar-Lopez F, et al. A shear gradientâ&#x20AC;&#x201C;dependent platelet aggregation mechanism drives thrombus formation. Nat Med. 2009;15(6):665-673. 2. Virchow R. Cellular pathology. As based upon physiological and pathological histology. Lecture XVI-Atheromatous affection of arteries. 1858. Nutr Rev. 1989;47(1):23-25. 3. Nesbitt WS, Mangin P, Salem HH, Jackson SP. The impact of blood rheology on the molecular and cellular events underlying arterial thrombosis. J Mol Med (Berl). 2006;84(12):989-995. 4. Holme PA, Orvim U, Hamers MJ, et al. Shear-induced platelet activation and platelet microparticle formation at blood flow conditions as in arteries with a severe stenosis. Arterioscler Thromb Vasc Biol. 1997;17(4):646-653. 5. Jackson SP. The growing complexity of platelet aggregation. Blood. 2007;109(12):5087-5095. 6. Wootton DM, Ku DN. Fluid mechanics of vascular systems, diseases, and thrombosis. Annu Rev Biomed Eng. 1999;1:299-329. 7. Bark DL Jr, Para AN, Ku DN. Correlation of thrombosis growth rate to pathological wall shear rate during platelet accumulation. Biotechnol Bioeng. 2012;109(10):2642-2650. 8. Korin N, Gounis MJ, Wakhloo AK, Ingber DE. Targeted drug delivery to flow-obstructed blood vessels using mechanically activated nanotherapeutics. JAMA Neurol. 2015;72(1):119-122. 9. Westein E, van der Meer AD, Kuijpers MJ, Frimat JP, van den Berg A, Heemskerk JW. Atherosclerotic geometries exacerbate pathological thrombus formation poststenosis in a von Willebrand factor-dependent manner. Proc Natl Acad Sci U S A. 2013;110(4):1357-1362. 10. Ruggeri ZM. The role of von Willebrand factor in thrombus formation. Thromb Res. 2007:120(Supplement 1):5-7. 11. Epshtein M, Korin N. Shear targeted drug delivery to stenotic blood vessels. J Biomech. 2017;50:217-221. 12. Jackson SP, Nesbitt WS, Westein E. Dynamics of platelet thrombus formation. J Thromb Haemost. 2009:7(Supplement 1):17-20. 13. Receveur N, Nechipurenko D, Knapp Y, et al. Shear rate gradients promote a bi-phasic thrombus formation on weak adhesive proteins, such as fibrinogen in a VWF-dependent manner [published online ahead of print, 2019 Nov 14]. Haematologica. 2020;105(10):24712483. 14. Khoury M, Epshtein M, Zidan H, Zukerman H, Korin N. Mapping deposition of particles in reconstructed models of human arteries. J Control Release. 2020;318:78-85. 15. Winkel LC, Hoogendoorn A, Van der Heiden K, et al. Animal models of surgically manipulated flow velocities to study shear stressinduced atherosclerosis. Atherosclerosis. 2015;241(1):100-110.

2357

CENTENARY REVIEW ARTICLE Ferrata Storti Foundation

Multiple myeloma: the (r)evolution of current therapy and a glance into the future Annamaria Gulla' and Kenneth C. Anderson

Division of Hematologic Neoplasia, Department of Medical Oncology, Jerome Lipper Multiple Myeloma Center, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA

ABSTRACT

Haematologica 2020 Volume 105(10):2358-2367

ver the past 20 years, the regulatory approval of several novel agents to treat multiple myeloma (MM) has prolonged median patient survival from 3 to 8-10 years. Increased understanding of MM biology has led to advances in diagnosis, prognosis, and response assessment, and has informed the development of targeted and immune agents. Here we provide an overview of the recent progress in MM, and highlight the most promising research areas to further improve patient outcome in the future.

Introduction Remarkable progress in our understanding of the pathobiology of myeloma (MM) has transformed the treatment paradigm and patient outcome. Preclinical studies have guided the discovery of more effective targeted therapies and informed clinical management. However, constitutive and ongoing genetic complexity and instability, coupled with the tumor promoting, immunosuppressive bone marrow (BM) microenvironment, remain an obstacle to cure. An estimated 32,270 new MM cases and 12,830 deaths in 2020 in the USA,1 coupled with a worldwide 126% increase in MM cases from 1990 to 2016,2 highlight the urgent need for novel therapies.

Definition of disease and precursor stages Correspondence: KENNETH C. ANDERSON kenneth_anderson@dfci.harvard.edu Received: June 10, 2020. Accepted: July 16, 2020. Pre-published: July 23, 2020. doi:10.3324/haematol.2020.247015 ©2020 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.

2358

Multiple myeloma is characterized by malignant plasma cells (PC) in the BM associated in most cases with monoclonal protein in serum and or urine; PC can also be detected in extramedullary sites and/or peripheral blood during progression of disease.3,4 Examination for MM-defining events allows for the discrimination between MM and its precursor stages, namely monoclonal gammopathy of undetermined significance (MGUS) and smoldering MM (SMM).5 Specifically, diagnosis of MM requires 10% or more PC in the BM plus one or more signs of end-organ damage including hypercalcemia, renal dysfunction, anemia, or bone disease (CRAB criteria).4 Even without CRAB features, patients who manifest MM-defining events including clonal BM PC >60%, serum : ratio >100 fold, and/or more than one bone focal lesion on magnetic resonance imaging (MRI) or positron emission tomography (PET)/computed tomography (CT) scan are also treated, as their risk of progression to symptomatic disease is approximately 80% at 2 years.5,6 Clinical manifestations of MM result from excessive production of monoclonal immunoglobulin protein by malignant PC in blood and/or urine, infiltration of BM by neoplastic clone, and aberrant cytokine secretion.4 MGUS patients are monitored for progression off all therapy, as their risk of progression overall is 1% yearly. The standard of practice is also to follow SMM patients expectantly off treatment,7,8 as the risk of progression is 10% per year in the first 5 years, dropping to 3% per year thereafter. Recently, the new “20-20-20” Mayo Clinic criteria have identified a high-risk (HR)-SMM subgroup (patients with two or more features including: BM PC infiltration >20%, monoclonal protein >20g/L and FLC ratio >20) with a median time to progression of 29 months.9 The QuiRedex study showed that lenalidomide+dexamethasone treatment prolonged time to progression and overall survival (OS) in HR-SMM.10 More recently lenalidomide alone has been shown to delay progressions of HR-SMM;11 however, there was a high rate of treatment discontinuation and secondary cancers in the lenalidomide cohort.11 Ongoing haematologica | 2020; 105(10)

MM: current therapy and future prospects

clinical trials are also evaluating alternative treatment strategies to delay progression of HR-SMM.12 Most recently, next-generation sequencing (NGS) analysis of MGUSSMM-MM patients has proven to be a useful tool to decipher the timing and chronology of disease initiation events.13,14 In the near future, the combination of genomic signatures and markers of disease burden will likely enable identification of those SMM patients who may benefit from early intervention, and definition of the optimal time to initiate treatment to avoid the development of clinical sequelae. Assessment of the value of early intervention must balance the benefit of delaying/preventing symptomatic MM against the risk of adverse events, and such early interventions should be of finite duration.

Prognostic factors and risk stratification Clinical and laboratory factors including disease stage, cytogenetic abnormalities, and depth of response to therapy can impact survival of MM patients.15 Cytogenetic analysis and fluorescence in situ hybridization (FISH)based genetic profiling should be routinely performed to evaluate disease biological behavior and prognosis.16 Among the poor prognostic markers, del(17p) and t(4;14) are the most informative;17,18 concomitant secondary cytogenetic abnormalities may impact prognosis.4 The International Staging System (ISS), based on albumin and β2-microglobulin levels, is most widely used,19 and has been revised (R-ISS) to incorporate lactate dehydrogenase (LDH) and HR-FISH abnormalities.20 Given the genomic complexity of MM, more sophisticated techniques including gene expression profiling, mutational status, and copy number abnormalities have been used, alone or in combination with FISH-based approaches, to more deeply characterize disease biology and prognosis. For example, newly diagnosed MM (NDMM) patients carrying HR del(17p) may be further stratified using subclonal analysis.21 Targeted sequencing has been used as an alternative to whole exon sequencing to specifically analyze fractions of the genome and provide more accurate risk stratification.22 Although not widely incorporated into clinical practice, these approaches will help to define future personalized treatment strategies in MM.

Assessment of response: minimal residual disease The high rate of complete response (CR) observed with the introduction of novel agents has led to the need for metrics capable of detecting even deeper responses to be developed. Response criteria are based on assessment of monoclonal protein in serum and urine, as well as BM evaluation. However, these parameters alone are not sensitive enough to detect low levels of residual tumor cells in the BM.23 More recently, both retrospective meta-analyses and prospective clinical trials have demonstrated the values of measuring minimal residual disease (MRD) within the BM using next-generation flow (NGF) or NGS, and at extramedullary sites using imaging such as PET/CT.24,25 The International Myeloma Working Group (IMWG) updated response criteria now include MRD status defined by absence of BM PC by NSG or NGF with a minimum sensitivity of 1 in 105 nucleated cells in patients with CR, providing guidelines that can be uniformly interpreted and applied in the context of clinical trials.26 MRD should be evaluated over the course of the disease, informing disease biology and treatment.26 For example, the DFCI/IFM clinical trial comparing lenalidomide-bortehaematologica | 2020; 105(10)

zomib-dexamethasone followed by early versus late autologous stem cell transplant (ASCT) showed that MRD negativity at the level of 10-6 was associated with prolonged progression-free survival (PFS) and OS.27,28 Moreover, those patients with MRD-BM who were also imaging (PET/CT scan) MRD negativity had the best outcome.27,28 Whether MRD-negativity should represent the goal of therapy for all patients with NDMM or relapsed/refractory (RRMM), or whether treatment decisions should be predicated on MRD status, is still the focus of ongoing clinical trials.

Biologically-based treatments High-dose chemotherapy plus ASCT remains the standard of care for NDMM patients of physiologic age 70 years or younger who have adequate cardiac, pulmonary, hepatic and renal function.4 Patients who are ineligible for transplant receive induction regimens dependent upon their frailty status.4 In both groups, the integration of scientifically-informed combinations of novel agents including immunomodulatory drugs (IMiD), proteasome inhibitor (PI), dexamethasone, and more recently monoclonal antibodies (mAbs), has transformed the treatment paradigm and patient outcome. However, genomic and clonal evolution in the tumor-promoting BM milieu underlies relapse of disease in most patients, and novel therapies are urgently needed.

Direct targeting of multiple myeloma cell dependencies - Multiple myeloma “lineage” dependencies Tumor cells may crucially rely on survival mechanisms that are imprinted during lineage development, namely lineage-dependency.29 For example, the clinical success of first-in-class PI bortezomib in MM has validated the heightened dependency of MM cells on the protein quality control pathway as a therapeutic target.30-32 The ubiquitin-proteasome system (UPS) is the primary mechanism for maintaining protein homeostasis.33 In normal PC, high protein turnover due to immunoglobulin production requires intact proteasome function, and this dependency is even higher in MM PC with aberrant protein turnover which further increases proteasome load. PI can overwhelm the imbalance between proteasome degradative capacity and proteasome load,34-36 leading to endoplasmic reticulum stress due to accumulation of misfolded and unfolded proteins, activation of the unfolded protein response, and cell death. Since proteins involved in cell proliferation and apoptosis, cell-cycle, DNA repair, and metabolism are substrates of the proteasome, PI inhibitor bortezomib has broad effects.37,38 It triggers both intrinsic and extrinsic MM cell apoptosis and MM cell cycle arrest, and modifies bone turnover and osteoclast activity in the BM.38 Bortezomib inhibits the NFκB pathway by blocking degradation of its inhibitor, IκB.33,38 Importantly, NFκB is a major oncogenic pathway in MM, which mediates MM survival and DNA repair, promotes interactions of MM cells-BM accessory cells via the transcription of adhesion molecules, as well as modulating transcription of cytokines (such as IL-6, VEGF, IGF-1), which in turn mediate MM growth and drug resistance, and confer immunosuppression in the BM.33,38 Over the disease course, MM cells acquire resistance to bortezomib via genetic and non-genetic mechanisms.33 Extensive preclinical research has delineated mechanisms 2359

A. Gulla’ and K.C. Anderson

of PI resistance and informed strategies to overcome resistance. Second-generation PI have been generated to overcome bortezomib resistance. The irreversible covalent epoxyketone PI carfilzomib, either alone or in combination with lenalidomide, has been approved to treat RRMM.39 Ixazomib, the first oral boronic acid-based PI, has been approved, alone and in combination with lenalidomide, to treat RRMM;40 this all-oral regimen showed low toxicity profile and improved patient quality of life. The pan-PI marizomib, which penetrates the central nervous system (CNS), also demonstrates anti-MM activity in the setting of bortezomib resistance.41 Given the multitude of available PI, the side effect profiles and identification of biomarkers of PI resistance/sensitivity will determine their optimal and rational use. Targeting upstream components of the ubiquitin proteasome system (UPS) has recently emerged as a promising strategy to overcome PI resistance. Therapeutic targeting

of deubiquitylating enzymes (DUB) and the 19S proteasome-associated ubiquitin receptor Rpn13 overcame PI resistance in preclinical studies.42-45 However, the first-inhuman trial of USP14/UCHL5 DUB inhibitor for RRMM has been stopped due to dose-limiting toxicity. Targeted therapies against Rpn13 have been developed for evaluation in setting of PI resistance.46 An alternative approach to overcome PI resistance is the concomitant block of the aggresome/autophagy pathway using an inhibitor of histone deacetylase 6 (HDAC6), which is recruited to maintain proteostasis balance as an adaptive response mechanism.38 Lineage vulnerabilities in MM also include aberrant transcription factor (TF) regulatory networks controlling lineage factor IRF4.47 Although direct targeting of TF represents an attractive strategy, there are no available inhibitors for clinical application. However, we found that aberrant regulatory KDM3A-IRF4-KLF2 loop may be effi-

Figure 1. Overview of the different anti-multiple myeloma (MM) strategies discussed in the review. Purple: strategies designed to directly target MM cell vulnerabilities; we can distinguish those exploiting “lineage dependencies” or “clonal dependencies”. (Center) Strategy targeting “epigenetic modifications” that may broadly affect both lineage and clonal vulnerabilities. (Bottom, yellow) Strategies aiming to disrupt MM-bone marrow microenvironment (BMM) interplay and restore host immunosurveillance. Purple double pointed arrows: MM-related approaches; yellow double pointed arrows: BMM-related approaches. These highlight the fact that one specific treatment, even in case of target therapy, may also affect multiple cellular components/interactions thus amplifying the therapeutic effects. OB: osteoblast; OC: osteoclasts; BMEC: bone marrow endothelial cells; ECM: extracellular matrix; BMSC: bone marrow stromal cells; DC: dendritic cell; pDC: plasmacytoid DC; MDSC: myeloid derived suppressor cells; Treg: regulatory T cell.

2360

haematologica | 2020; 105(10)

MM: current therapy and future prospects

ciently targeted by KDM3A inhibitor, which restores IRF4 and KLF2 promoter methylation and suppresses their transcription, thereby resulting in decreased MM cell homing to the BM and direct anti-MM toxicity.48 Moreover, repression of IRF4 transcription is observed after lenalidomide treatment, which triggers cereblon (CRBN)-mediated degradation of IRF4 transcriptional activator IKZF3.49,50 From these findings, a new platform technology has been developed to trigger selective protein degradation. Specifically, degronimids, also known as proteolysis-targeting chimeras (PROTAC), are designed by conjugating the small-molecule binder of the target protein to an E3 ubiquitin ligase binding scaffold, such as the analogs of thalidomide which bind CRBN.51 This approach will allow for the therapeutic degradation of protein substrates that are otherwise challenging to target.

- Multiple myeloma “clonal” dependencies Large inter-patient and intra-patient genetic heterogeneity limits the identification of universal drivers of MM. However, several oncogenic dependencies are primary events related to mutations in driver genes and primary translocations.52 Translocations or gains of MYC locus (along with dysregulation of upstream signaling pathways, such as IRF4) support an oncogenic role of MYC in MM, especially in the context of MGUS-MM transition.53 Alteration of the transcriptional program of MYC, and of its functional collaborators such as E2F1, promotes oncogenic signaling and PC survival.54,55 Frequent gene mutations in MM include RAS, either KRAS or NRAS, with subsequent activation of the MAPK pathway, BRAF, DIS3, and FAM46C.56 Their role as prognostic factors has not been completely defined, as only TP53 mutation (6-8% of patients at diagnosis) clearly confers worse patient outcome.57 Mutation-targeted treatments in MM are often compromised by intra-clonal heterogeneity. Specifically, deep sequencing has identified a complex subclonal structure in MM with different patterns of clonal evolution impacted by BM, immune response, and therapy.58 In this complex scenario, MM cells may share common mutations, but they may also express additional subclones which compromise mutation-targeting therapies.58,59 The MyDRUG trial is enrolling patients with relapsed MM based upon genomic sequencing; patients receive a specific treatment targeting their unique tumor mutations, along with standard-of-care treatment.60 This trial will reveal whether the abnormal clone can be targeted, and provide the rationale for further derived clinical trials of targeted therapies. Genomic complexity in MM is due to genomic instability and ongoing DNA damage.61 MM cells display hyperactivation of DNA repair mechanisms which confer a survival advantage and drug resistance with increasing numbers of new mutations over time.62 These aberrant processes may reveal new vulnerabilities. For example, in MM patients, in whom ongoing DNA damage occurs concurrently with low Hippo co-transcription factor (YAP1) levels, MM cell apoptosis is prevented. Conversely, inhibition of STK4 rescues YAP1 and triggers DNA-damageinduced apoptosis, providing the framework for clinical evaluation of STK4 inhibition.63 A second example is the induction of “BRCAness” status in MM cells by bortezomib, thereby increasing their sensitivity to PARP inhibitors.37 Finally, MYC amplification in MM can induce DNA response pathway and reactive oxygen species; the haematologica | 2020; 105(10)

former can be blocked by ATR inhibitors and the latter can be increased by bortezomib, together triggering MM cell apoptosis in a synthetic lethal mechanism.61

- Multiple myeloma epigenetic modifications Epigenetic alterations affect regulation of gene activity and expression, without altering gene sequence. Such alterations are associated with MM onset and progression, and modulate several important biological processes.64 Among epigenetic changes, DNA methylation, histone modification, and non-coding RNA (ncRNA) deregulation are the best characterized.64 Global hypomethylation of the genome characterizes the transition from MGUS to MM, whereas pervasive genome re-methylation occurs in the transition from MM to more aggressive leukemic stage (PCL).64 Universal overexpression of histone methyltransferase MMSET is detected in patients carrying t(4;14) translocation and promotes MM cell survival by activating oncogenic MAPK pathway, increasing MYC and IRF4 transcription, and inducing chemo-resistance through enhancing DNA repair mechanisms.65,66 Therefore, development of MMSET inhibitor represents a promising therapeutic strategy for this subset of MM. We have demonstrated the oncogenic role and prognostic relevance of type II arginine methyltransferase PRMT5 in MM, whose inhibition results in MM cell killing via NFκB inhibition, thus providing the rationale for clinical trials targeting PRMT5 in MM.67 Histone deacetylases (HDAC) are generally hyperactive in MM, and HDAC inhibitors are the most investigated epigenetic drugs.65 Preclinical studies have led to clinical trials and the approval of non-selective HDAC inhibitor panobinostat in combination with bortezomib in RRMM.68 However, increased toxicity observed with panobinostat prompted the development and translation of selective HDAC6 inhibitors (ricolinostat and ACY 241), which showed promising results and lower toxicity in combination with bortezomib and dexamethasone.69,70 Over the last decade, extensive studies have also highlighted the contribution of the ncRNA compartment in MM pathogenesis and progression. Specifically, microRNAs (miRNAs) are key regulators of gene expression at the post-transcriptional level, as they can induce either translational repression or degradation of target mRNAs upon total or partial complementary binding with 3′ untranslated region (3′ UTR).71 Given the multitude of targets for a single miRNA, these molecules harbor the potential to concomitantly regulate multiple biological processes. Preclinical data have defined their oncogenic (miR-221/222,-21,-17-92 cluster) or tumor suppressive (miR-29b,-34a,-125b,-15,-16) roles in MM associated with repression or overexpression, respectively, of genes involved in essential pro-survival pathways.72-77 The role of miRNAs has been similarly investigated in the context of drug resistance, BM-PC interaction, and bone disease.78-81 Although miRNA-based therapeutics have not yet translated into US Food and Drug Administration (FDA)approved drugs, several candidates are being tested in other diseases and will soon be evaluated in MM,82 using miRNA replacement or inhibition strategies. As they are endogenous antisense of mRNAs, their replacement is likely to induce a “natural” effect on the targets, with less off-target effects compared to siRNAs; moreover, the recent availability of in vivo delivery systems now allows for clinical trials.83 Likewise, miRNA inhibition strategies take advantage of new antisense oligonucleotide tech2361

A. Gullaâ&#x20AC;&#x2122; and K.C. Anderson

nologies, and ongoing early trials in several cancers will likely pave the way for their investigation in MM.83 Finally, long ncRNAs (lncRNA) represent major regulators of gene expression and chromatin dynamics by interacting with DNA and proteins.84 LncRNA genes outnumber protein-coding genes, with a partner of expression often restricted to specific cell types or conditions.84 With few exceptions, however, their functional role is still largely obscure in MM.85 We recently described the lncRNA landscape in MM, and their role as independent predictors of clinical outcome;86 ongoing and future studies will define their role in disease pathogenesis and as potential therapeutic targets.

Targeting the tumor-bone marrow microenvironment interface Disrupting the interactions of MM cells with the BM represents an ideal therapeutic strategy in MM, as shown by agents such as IMiD which remain active against MM even in the BM milieu. Cellular and non-cellular components of the BM niche support MM cell proliferation, migration, survival and drug resistance, while also conferring immunosuppression, and therefore represent targets for novel therapeutics.4,87,88

- Immunomodulatory drugs Extensive preclinical and clinical studies have led to the FDA approval of the IMiD thalidomide and its more potent derivatives lenalidomide and pomalidomide for treatment of both NDMM and RRMM.89-92 IMiD induce direct cytotoxic effects on MM cells including growth arrest and caspase-8-mediated apoptosis, associated with CRBN-dependent degradation of IKZF1/3 followed by IRF4 downregulation.49,50,93 In the BM microenvironment they abrogate MM cell adhesion to the BM, modulate cytokine and growth factor secretion, inhibit angiogenesis, and most importantly, upregulate T, NK, and NKT cells, while downregulating regulatory T cells.3,93 Mechanistically, binding to CRBN has also been implicated in mediating the immune-related effects of IMiD, as IKZF1/3 degradation in T cells increases their secretion of cytokines including IL-2.94 This mechanism is also associated with increased natural killer (NK) and NK-T-cell cytotoxicity and antibody-dependent cellular cytotoxicity (ADCC) (i.e., anti-CD38, daratumumab and -CD20, rituximab) observed after IMiD treatment.95,96 Furthermore, IMiD can also enhance NK and T-cell cytotoxicity by triggering granzyme-B via either CRBN- or ZAP-70-dependent mechanisms.95 Preclinical and clinical studies have already confirmed the strong synergism of IMiD with PI and with mAb.97-102 Recently, iberdomide, a higher affinity CRBN E3-ligase modulator (CELMoD), showed significant preclinical activity against IMiD-resistant MM cells,103 and ongoing clinical trials are examining its efficacy in RRMM resistant to lenalidomide and pomalidomide. Promising results have also been recently reported with the more potent CELMoD CC-92480, currently under investigation to treat IMiD-resistant RRMM.104

- Immune-based therapies Loss of immune surveillance supports MM growth and resistance, and is associated with alterations in accessory and immune cells in the BM.105 Moreover, there is increasing evidence that evolving immune dysfunction is an important determinant of progression from MGUS/SMM 2362

to symptomatic MM.105 For example, functional interaction of plasmacytoid dendritic cells (DC) with MM cells promotes their survival and drug resistance,106 providing the framework for targeting this interaction in novel therapies.107-109 Similarly, myeloid derived suppressor cells, regulatory T cells, Th17 cells, tumor-associated macrophages, mesenchymal stromal cells, and osteoclasts significantly contribute to tumor immune escape and immunocompromised clinical status.110 Immune escape is also mediated, at least in part, by increased expression of immune checkpoints, i.e., PD-1/PD-L1, in T cells and MM cells, associated with disease progression from MGUS and SMM to MM.111 Although preclinical data have suggested the therapeutic utility of PD-1/PD-L1 blockade, early clinical trials have been discouraging.111 No singleagent activity of pembrolizumab has been shown, and importantly, two randomized clinical trials evaluating pembrolizumab in combination with IMiD to treat RRMM were closed due to excessive mortality.111 Ongoing studies are characterizing the role of the other immune checkpoint or agonist proteins (i.e., LAG 3 or TIGIT and OX40, respectively) as potential therapeutic targets, alone and in combination with MM targeted and immune therapies.112 Multiple mAb targeting MM surface antigens can trigger ADCC, antibody-dependent cellular phagocytosis, complement activation, and direct effect on MM cells.113 Elotuzumab and daratumumab target SLAMF7 and CD38, respectively.113 Elotuzumab also directly activates NK cells and is FDA approved in combination with lenalidomide or pomalidomide in RRMM.99,100 Daratumumab has shown remarkable extent and frequency of response, leading to its FDA approval as a single agent or in combination with IMiD and PI in both RRMM and NDMM.97,98,114,115 The recent GRIFFIN trial compared standard lenalidomide-bortezomib-dexamethasone with or without daratumumab in transplant eligible patients, and showed deeper responses and MRD negativity rate in the daratumumab-treated patient cohort.116 Moreover, recent approval of subcutaneous formulation of daratumumab will dramatically reduce patient treatment times.117 More recently, a new CD38directed mAb, isatuximab, has been approved in combination with pomalidomide-dexamethasone to treat RRMM;101 whether it is effective even in daratumumab refractory MM remains to be determined. Monoclonal antibody technologies have also provided the framework for the development of Ab-drug conjugates (ADC) and bispecific T-cell engagers (BiTE). In the former, the conjugation with cytotoxic chemicals (such as auristatin) via synthetic linkers provides for direct tumor killing, and the mAb provides for selective tumor cell targeting, as well as immune-mediated cytotoxicity.118-122 Bcell maturation antigen (BCMA)-directed ADC are currently under investigation in both preclinical and clinical settings, and represent a promising approach due to the highly specific expression on BCMA on MM cells and late memory B cells, as well as the role of the BCMA/APRIL pathway in supporting MM cell survival in the BM.113,118,123 The bi-specificity of the BiTE (mainly for CD3 on T cells and several MM-associated antigens, such as BCMA and GPRC5D) allows for engagement of T cells with tumor cells, resulting in formation of cytolytic synapses and tumor lysis.124-126 Although results from early trials look promising, longer follow-up in larger studies are needed to assess the clinical benefit and potential toxicity. haematologica | 2020; 105(10)

MM: current therapy and future prospects

Importantly, BCMA ADC and BiTE have the advantage for “off the shelf” availability and universal use. Cellular therapies represent an additional strategy to boost MM-specific immunity using either adoptive T-cell (ACT) or engineered T-cell approaches.3 Clinical experience with ACT using marrow-infiltrating lymphocytes (MIL) in MM has shown promise in achieving memory immune responses and stable disease.127 Importantly, advances in engineering technologies have allowed for both T-cell receptor (TCR) and chimeric antigen receptor (CAR) T-cell approaches.128 CAR are chimeric proteins that bring together the signaling moieties of TCR complexes and the variable domains of Ab which recognize a tumor-associated antigen.128,129 Co-stimulatory molecules have been included in the second-generation CAR-T cells to enhance T-cell activation by mimicking a physiological T-cell response.128,129 After genetic modification, a patient’s T cells expressing the chimeric protein can be expanded ex vivo, and then activate a specific T-cell response once reinfused to the patient.128,129 This allows CAR-T cells, in contrast to TCR-T cells, to recognize unprocessed tumor antigen in an MHC-independent manner.128,129 A major determinant of successful CAR-T therapy is the identification of a target uniquely and highly expressed by MM cells, thus limiting the occurrence of “off-target” effects. Among a variety of antigen targets, BCMA is the most frequently used due to its selectivity for normal plasma and MM cells. Several CAR-T products have been clinically tested in heavily pre-treated (PI-IMiD-CD38 mAb) RRMM, and have demonstrated remarkable deep (MRD) responses.129-132 Clinical experience has also helped to improve management of the most commonly observed toxicities of CAR-T cells, including cytokine release syndrome and neurotoxicity.128,129 To date, however, most patients have relapsed, and ongoing research is assessing mechanisms of resistance to CAR-T, utilizing combination immune approaches with CAR-T, and using CAR-T earlier in the disease course in order to achieve more durable responses. Lastly, vaccination strategies have been developed to improve antigen-specific memory anti-MM immunity. Specifically, multi-peptide-based vaccines induce effective and durable memory peptide-specific CTL in SMM patients, providing the rationale for their clinical evaluation to delay progression from SMM to active disease.133,134 More recently, a novel engineered heteroclitic BCMA peptide has been used to induce a BCMA-specific memory anti-MM immunity, suggesting its potential use in vaccination and/or ACT strategies to generate longlasting immunity against MM.135 As alternative vaccination strategy employs MM cell/DC fusion vaccines to generate anti-MM immunity in the post-ASCT setting, this vaccine induces anti-MM immunity and enhances depth of response.136 The most significant obstacle to successful vaccination therapy in MM is the disease- and treatment-related immune dysfunction, which may limit the immune responses in vivo. As such, a randomized trial comparing lenalidomide versus lenalidomide plus MM cell/DC fusion vaccine post-transplant is ongoing (clinicaltrials.gov identifier: NCT02728102) to determine whether combination of vaccination with IMiD may improve its efficacy. Overall, future treatment approaches will likely rely on the optimal combination of targeted and immunebased strategies to obtain a durable anti-MM response and restore the host immune-surveillance. haematologica | 2020; 105(10)

Future directions Despite tremendous advances, the clinical management of MM patients remains challenging, since acquisition of resistance underlies relapse of disease in most patients. Correlative science studies on patient samples are delineating mechanisms of resistance to both targeted and immune agents in order to inform clinical strategies to overcome resistance and improve patient outcome. Development of second-generation more potent drugs of the same class has overcome both PI and IMiD resistance, as have combination therapies with agents targeting pathways mediating resistance. Identification of biomarkers of patient MM resistance/sensitivity may further inform sequential and combination therapies in the future. Importantly, agents targeting novel MM vulnerabilities are urgently needed. For example, selinexor, a selective inhibitor of nuclear export of tumor suppressor proteins, growth factors, and mRNAs of oncogenic proteins, has recently been approved in triple-class (PI-IMiD-anti-CD38 mAb) refractory MM.137 A similar scenario of resistance is now beginning to appear for immune-based approaches, with several possible explanations. Loss of targeted antigens (such as BCMA, CD38) is a common event, either due to loss with tumor evolution or to suppression in the face of immune pressure. Multi-antigen targeting may potentially overcome this obstacle, and several trials are evaluating this strategy.129 Similarly, circulating antigen in a soluble form may potentially interfere with immune-targeted approaches. For example, high levels of soluble BCMA have been detected in MM patients, and anti-BCMA CAR-T therapy is being combined with γ-secretase inhibitor to block BCMA cleavage from the MM cell surface (clinicaltrials.gov identifier: NCT03502577). Additional resistance may be intrinsic to the technology or modality, and ongoing efforts to increase CAR-T cell expansion and persistence in vivo include enriching for early memory Tcell phenotype, optimizing CAR design to avoid antigenindependent tonic signaling, and/or intensifying lymphodepletion to promote CAR-T cell persistence.129 Lastly, T-cell exhaustion and the immunosuppressive BM may contribute to both targeted- and immune-therapy resistance.128 Restoration of host anti-MM immunity represents an important unmet need in MM. Several models, such as SCID-hu and SCID-synth-hu, have been developed to recapitulate the in vivo growth on patient MM cells in the context of BM.138,139 However, understanding the role of the immune system in disease pathobiology requires the use of immunocompetent models (such as the 5T and Vk*MYC).53,140 This is critical for evaluating not only immune therapies, but also targeted MM agents, such as bortezomib, which induce immunogenic cell death in vivo. A recent area of investigation in MM is assessing the role of gut microbiome in shaping the immune system response, including anti-tumor immunity. For example, a commensal bacterium Prevotella heparinolytica promotes progression of MM by favoring differentiation of Th17 cells in the gut, which migrate to the BM of Vk*MYC mice and activate eosinophils; targeting IL-17-eosinophil immune axis may, therefore, represent a potential treatment for HR-SMM.141 Abundance of Eubacterium limosum bacteria in the intestinal flora has been associated with relapse after allogeneic stem cell transplantation.142 2363

A. Gullaâ&#x20AC;&#x2122; and K.C. Anderson

Similarly, high presence of Eubacterium hallii in the intestinal microbiota correlates with achievement of MRD negativity.143 Although still in an early stage, studies of the MM microbiome may identify future biomarkers or therapeutic agents to improve MM patient outcome. Finally, identification of specific biomarkers predictive of therapy response within a patientâ&#x20AC;&#x2122;s heterogeneous MM has been a major focus of research. The first example of biomarker-driven anti-MM treatment in MM is the Bcl-2 inhibitor venetoclax, whose safety and efficacy are predicated upon occurrence of t(11;14) or presence of high levels of BCL2.144 Several trials in RRMM are showing efficacy of venetoclax, as monotherapy and in combination, restricted to this patient subset.145,146 Integration of current and future technologies may further guide disease management and allow for precision medicine. For example, encouraging results have recently been reported in a trial using a multi-omics approach integrating DNA and RNA sequencing to inform drug treatment for RRMM.147 The ongoing MyDRUG trial profiles relapsed MM, and is then examining whether targeting genomic abnormalities in combination with standard relapse MM therapy can delete the abnormal MM clone.60 This and other trials will inform the utility of precision medicine in MM, especially in the presence of concomitant genetic abnormalities. Given the multiple available treatment options, welldesigned randomized clinical trials are necessary to assess the superior efficacy of specific regimens with head-tohead comparison. For example, interim analysis of the phase III ENDURANCE trial did not show superior PFS of carfilzomib versus bortezomib in combination with lenalidomide-dexamethasone for NDMM. Importantly, regulatory randomized trial results require real-world validation, since patient age, frailty status, and comorbidity frequently do not reflect trial patients. High-dose melphalan with ASCT remains a standard of care, and its role in the era of novel therapies is under evaluation in the IFM/DFCI 2009 DETERMINATION and FORTE clinical trials. However, the recent use of quadruplet therapies including daratumumab in the CASSIOPEIA and GRIFFIN trials shows that the addition of mAb can achieve increased extent and depth of response to induction ther-

References 1. National Cancer Institute. Cancer stat facts: myeloma. Available from: https://seer.cancer.gov/statfacts/html/mulmy.html. Accessed 20 July 2020. 2. Cowan AJ, Allen C, Barac A, et al. Global burden of multiple myeloma: a systematic analysis for the Global Burden of Disease Study 2016. JAMA Oncol. 2018;4(9):1221-1227. 3. Anderson KC. Progress and paradigms in multiple myeloma. Clin Cancer Res. 2016;22(22):5419-5427. 4. Kumar SK, Rajkumar V, Kyle RA, et al. Multiple myeloma. Nat Rev Dis Primers. 2017;3:17046. 5. Rajkumar SV, Dimopoulos MA, Palumbo A, et al. International Myeloma Working Group updated criteria for the diagnosis of multiple myeloma. Lancet Oncol. 2014;15 (12):e538-548. 6. Mateos MV, Hernandez MT, Giraldo P, et al. Lenalidomide plus dexamethasone for highrisk smoldering multiple myeloma. N Engl J

2364

apy,116,148 and whether high-dose melphalan and ASCT improves outcome of quadruplet therapy remains to be determined. Nonetheless, research continues to improve alkylating agents as well. For example, melflufen is a prodrug which is digested to melphalan by high levels of aminopeptidase in MM cells, thereby improving its therapeutic index.149,150 Ultimately, the future use of novel targeted and immune therapies, as well as the role of conventional therapies, will be defined by vulnerabilities within individual patients and/or patient subgroups.

Conclusions Over the past decades, a deeper understanding of the complex MM pathobiology has informed drug development and clinical practice, resulting in significant improvements in patient outcome. Combination approaches targeting MM cells, disrupting MM cell/BM interactions, and enhancing anti-MM immune responses, have remarkably improved response extent and frequency. Remaining obstacles to cure include constitutive and evolving genomic heterogeneity in MM cells, as well as the immunosuppressive BM milieu. In the future, integration of advanced sequencing technologies profiling both the MM cell and BM accessory/immune cells will identify novel targets and inform more potent, selective, and well tolerated targeted and immune therapies. Long-term disease-free survival and potential cure in MM will require both achieving MRD negativity and restoring host antiMM immunity. Such patients can then be free of disease while off all therapies. Funding This work is supported by NIH/NCI grants SPOREP50CA100707 (KCA), R01-CA050947 (KCA), R01CA207237 (KCA), P01CA155258 (KCA) and R01CA178264 (KCA); and the Sheldon and Miriam Medical Research Foundation (KCA). KCA is an American Cancer Society Clinical Research Professor. AG is a Fellow of The Leukemia & Lymphoma Society and a Scholar of the American Society of Hematology (ASH).

Med. 2013;369(5):438-447. 7. Kyle RA, Greipp PR. Smoldering multiple myeloma. N Engl J Med. 1980;302(24):13471349. 8. Kapoor P, Rajkumar SV. Smoldering multiple myeloma: to treat or not to treat. Cancer J. 2019;25(1):65-71. 9. Lakshman A, Rajkumar SV, Buadi FK, et al. Risk stratification of smoldering multiple myeloma incorporating revised IMWG diagnostic criteria. Blood Cancer J. 2018;8(6):59. 10. Mateos MV, Hernandez MT, Giraldo P, et al. Lenalidomide plus dexamethasone versus observation in patients with high-risk smouldering multiple myeloma (QuiRedex): long-term follow-up of a randomised, controlled, phase 3 trial. Lancet Oncol. 2016;17(8):1127-1136. 11. Lonial S, Jacobus S, Fonseca R, et al. Randomized trial of lenalidomide versus observation in smoldering multiple myeloma. J Clin Oncol. 2020;38(11):1126-1137. 12. D'Agostino M, Bertamini L, Oliva S, Boccadoro M, Gay F. Pursuing a curative approach in multiple myeloma: a review of

new therapeutic strategies. Cancers (Basel). 2019;11(12). 13. Rustad EH, Yellapantula V, Leongamornlert D, et al. Timing the initiation of multiple myeloma. Nat Commun. 2020;11(1):1917. 14. Aktas Samur A, Minvielle S, Shammas M, et al. Deciphering the chronology of copy number alterations in multiple myeloma. Blood Cancer J. 2019;9(4):39. 15. Rajkumar SV. Multiple myeloma: 2016 update on diagnosis, risk-stratification, and management. Am J Hematol. 2016;91(7): 719-734. 16. Sonneveld P, Avet-Loiseau H, Lonial S, et al. Treatment of multiple myeloma with highrisk cytogenetics: a consensus of the International Myeloma Working Group. Blood. 2016;127(24):2955-2962. 17. Avet-Loiseau H, Hulin C, Campion L, et al. Chromosomal abnormalities are major prognostic factors in elderly patients with multiple myeloma: the Intergroupe Francophone du Myelome experience. J Clin Oncol. 2013;31(22):2806-2809. 18. Kumar S, Fonseca R, Ketterling RP, et al.

haematologica | 2020; 105(10)

MM: current therapy and future prospects Trisomies in multiple myeloma: impact on survival in patients with high-risk cytogenetics. Blood. 2012;119(9):2100-2105. 19. Greipp PR, San Miguel J, Durie BG, et al. International staging system for multiple myeloma. J Clin Oncol. 2005;23(15):34123420. 20. Palumbo A, Avet-Loiseau H, Oliva S, et al. Revised International Staging System for multiple myeloma: a report from International Myeloma Working Group. J Clin Oncol. 2015;33(26):2863-2869. 21. Thakurta A, Ortiz M, Blecua P, et al. High subclonal fraction of 17p deletion is associated with poor prognosis in multiple myeloma. Blood. 2019;133(11):1217-1221. 22. 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. 23. Paiva B, van Dongen JJ, Orfao A. New criteria for response assessment: role of minimal residual disease in multiple myeloma. Blood. 2015;125(20):3059-3068. 24. Perrot A, Lauwers-Cances V, Corre J, et al. Minimal residual disease negativity using deep sequencing is a major prognostic factor in multiple myeloma. Blood. 2018;132(23): 2456-2464. 25. Munshi NC, Avet-Loiseau H, Rawstron AC, et al. Association of minimal residual disease with superior survival outcomes in patients with multiple myeloma: a meta-analysis. JAMA Oncol. 2017;3(1):28-35. 26. Kumar S, Paiva B, Anderson KC, et al. International Myeloma Working Group consensus criteria for response and minimal residual disease assessment in multiple myeloma. Lancet Oncol. 2016;17(8):e328e346. 27. Moreau P, Attal M, Caillot D, et al. Prospective evaluation of magnetic resonance imaging and [(18)F]fluorodeoxyglucose positron emission tomography-computed tomography at diagnosis and before maintenance therapy in symptomatic patients with multiple myeloma included in the IFM/DFCI 2009 trial: results of the IMAJEM study. J Clin Oncol. 2017;35(25):29112918. 28. Attal M, Lauwers-Cances V, Hulin C, et al. Lenalidomide, bortezomib, and dexamethasone with transplantation for myeloma. N Engl J Med. 2017;376(14):1311-1320. 29. Garraway LA, Sellers WR. Lineage dependency and lineage-survival oncogenes in human cancer. Nat Rev Cancer. 2006;6(8): 593-602. 30. Richardson PG, Sonneveld P, Schuster M, et al. Extended follow-up of a phase 3 trial in relapsed multiple myeloma: final time-toevent results of the APEX trial. Blood. 2007;110(10):3557-3560. 31. Richardson PG, Sonneveld P, Schuster MW, et al. Bortezomib or high-dose dexamethasone for relapsed multiple myeloma. N Engl J Med. 2005;352(24):2487-2498. 32. Richardson PG, Barlogie B, Berenson J, et al. A phase 2 study of bortezomib in relapsed, refractory myeloma. N Engl J Med. 2003;348 (26):2609-2617. 33. Gandolfi S, Laubach JP, Hideshima T, Chauhan D, Anderson KC, Richardson PG. The proteasome and proteasome inhibitors in multiple myeloma. Cancer Metastasis Rev. 2017;36(4):561-584. 34. Cenci S, Mezghrani A, Cascio P, et al. Progressively impaired proteasomal capacity during terminal plasma cell differentiation. EMBO J. 2006;25(5):1104-1113.

haematologica | 2020; 105(10)

35. Bianchi G, Oliva L, Cascio P, et al. The proteasome load versus capacity balance determines apoptotic sensitivity of multiple myeloma cells to proteasome inhibition. Blood. 2009;113(13):3040-3049. 36. Obeng EA, Carlson LM, Gutman DM, Harrington WJ Jr, Lee KP, Boise LH. Proteasome inhibitors induce a terminal unfolded protein response in multiple myeloma cells. Blood. 2006;107(12):49074916. 37. Neri P, Ren L, Gratton K, et al. Bortezomibinduced "BRCAness" sensitizes multiple myeloma cells to PARP inhibitors. Blood. 2011;118(24):6368-6379. 38. Hideshima T, Anderson KC. Biologic impact of proteasome inhibition in multiple myeloma cells-from the aspects of preclinical studies. Semin Hematol. 2012;49(3):223-227. 39. Stewart AK, Rajkumar SV, Dimopoulos MA, et al. Carfilzomib, lenalidomide, and dexamethasone for relapsed multiple myeloma. N Engl J Med. 2015;372(2):142-152. 40. Moreau P, Masszi T, Grzasko N, et al. Oral ixazomib, lenalidomide, and dexamethasone for multiple myeloma. N Engl J Med. 2016;374(17):1621-1634. 41. Richardson PG, Zimmerman TM, Hofmeister CC, et al. Phase 1 study of marizomib in relapsed or relapsed and refractory multiple myeloma: NPI-0052-101 Part 1. Blood. 2016;127(22):2693-2700. 42. Song Y, Li S, Ray A, et al. Blockade of deubiquitylating enzyme Rpn11 triggers apoptosis in multiple myeloma cells and overcomes bortezomib resistance. Oncogene. 2017;36(40):5631-5638. 43. Chauhan D, Tian Z, Nicholson B, et al. A small molecule inhibitor of ubiquitin-specific protease-7 induces apoptosis in multiple myeloma cells and overcomes bortezomib resistance. Cancer Cell. 2012;22(3):345-358. 44. Tian Z, D'Arcy P, Wang X, et al. A novel small molecule inhibitor of deubiquitylating enzyme USP14 and UCHL5 induces apoptosis in multiple myeloma and overcomes bortezomib resistance. Blood. 2014;123(5): 706-716. 45. Du T, Song Y, Ray A, Chauhan D, Anderson KC. Proteomic analysis identifies mechanism(s) of overcoming bortezomib resistance via targeting ubiquitin receptor Rpn13. Leukemia. 2020 May 18. doi: 10.1038/s41375-020-0865-2. [Epub ahead of print] 46. Song Y, Park PMC, Wu L, et al. Development and preclinical validation of a novel covalent ubiquitin receptor Rpn13 degrader in multiple myeloma. Leukemia. 2019;33(11):2685-2694. 47. Shaffer AL, Emre NC, Lamy L, et al. IRF4 addiction in multiple myeloma. Nature. 2008;454(7201):226-231. 48. Ohguchi H, Hideshima T, Bhasin MK, et al. The KDM3A-KLF2-IRF4 axis maintains myeloma cell survival. Nat Commun. 2016;7:10258. 49. Kronke J, Udeshi ND, Narla A, et al. Lenalidomide causes selective degradation of IKZF1 and IKZF3 in multiple myeloma cells. Science. 2014;343(6168):301-305. 50. Lu G, Middleton RE, Sun H, et al. The myeloma drug lenalidomide promotes the cereblon-dependent destruction of Ikaros proteins. Science. 2014;343(6168):305-309. 51. Winter GE, Buckley DL, Paulk J, et al. Drug development. Phthalimide conjugation as a strategy for in vivo target protein degradation. Science. 2015;348(6241):1376-1381. 52. Walker BA, Mavrommatis K, Wardell CP, et al. Identification of novel mutational drivers

reveals oncogene dependencies in multiple myeloma. Blood. 2018;132(6):587-597. 53. Chesi M, Robbiani DF, Sebag M, et al. AIDdependent activation of a MYC transgene induces multiple myeloma in a conditional mouse model of post-germinal center malignancies. Cancer Cell. 2008;13(2):167-180. 54. Jovanovic KK, Roche-Lestienne C, Ghobrial IM, Facon T, Quesnel B, Manier S. Targeting MYC in multiple myeloma. Leukemia. 2018;32(6):1295-1306. 55. Fulciniti M, Lin CY, Samur MK, et al. Nonoverlapping control of transcriptome by promoter- and super-enhancer-associated dependencies in multiple myeloma. Cell Rep. 2018;25(13):3693-3705 e6. 56. Bolli N, Avet-Loiseau H, Wedge DC, et al. Heterogeneity of genomic evolution and mutational profiles in multiple myeloma. Nat Commun. 2014;5:2997. 57. Perrot A, Corre J, Avet-Loiseau H. Risk stratification and targets in multiple myeloma: from genomics to the bedside. Am Soc Clin Oncol Educ Book. 2018;38:675-680. 58. Corre J, Cleynen A, Robiou du Pont S, et al. Multiple myeloma clonal evolution in homogeneously treated patients. Leukemia. 2018;32(12):2636-2647. 59. Ledergor G, Weiner A, Zada M, et al. Single cell dissection of plasma cell heterogeneity in symptomatic and asymptomatic myeloma. Nat Med. 2018;24(12):1867-1876. 60. Study tests targeted drugs for multiple myeloma. Cancer Discov. 2019;9(4):459. doi:10.1158/2159-8290.CD-NB2019-014. 61. Cottini F, Hideshima T, Suzuki R, et al. Synthetic lethal approaches exploiting DNA damage in aggressive myeloma. Cancer Discov. 2015;5(9):972-987. 62. Shammas MA, Shmookler Reis RJ, Koley H, Batchu RB, Li C, Munshi NC. Dysfunctional homologous recombination mediates genomic instability and progression in myeloma. Blood. 2009;113(10):2290-2297. 63. Cottini F, Hideshima T, Xu C, et al. Rescue of Hippo coactivator YAP1 triggers DNA damage-induced apoptosis in hematological cancers. Nat Med. 2014;20(6):599-606. 64. Amodio N, D'Aquila P, Passarino G, Tassone P, Bellizzi D. Epigenetic modifications in multiple myeloma: recent advances on the role of DNA and histone methylation. Expert Opin Ther Targets. 2017;21(1):91101. 65. Ohguchi H, Hideshima T, Anderson KC. The biological significance of histone modifiers in multiple myeloma: clinical applications. Blood Cancer J. 2018;8(9):83. 66. Mirabella F, Wu P, Wardell CP, et al. MMSET is the key molecular target in t(4;14) myeloma. Blood Cancer J. 2013;3:e114. 67. Gulla A, Hideshima T, Bianchi G, et al. Protein arginine methyltransferase 5 has prognostic relevance and is a druggable target in multiple myeloma. Leukemia. 2018;32(4):996-1002. 68. San-Miguel JF, Hungria VT, Yoon SS, et al. Overall survival of patients with relapsed multiple myeloma treated with panobinostat or placebo plus bortezomib and dexamethasone (the PANORAMA 1 trial): a randomised, placebo-controlled, phase 3 trial. Lancet Haematol. 2016;3(11):e506-e515. 69. Hideshima T, Qi J, Paranal RM, et al. Discovery of selective small-molecule HDAC6 inhibitor for overcoming proteasome inhibitor resistance in multiple myeloma. Proc Natl Acad Sci U S A. 2016;113(46):13162-13167. 70. Vogl DT, Raje N, Jagannath S, et al. Ricolinostat, the first selective histone

2365

A. Gullaâ&#x20AC;&#x2122; and K.C. Anderson deacetylase 6 inhibitor, in combination with bortezomib and dexamethasone for relapsed or refractory multiple myeloma. Clin Cancer Res. 2017;23(13):3307-3315. 71. Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell. 2004;116 (2):281-297. 72. Leone E, Morelli E, Di Martino MT, et al. Targeting miR-21 inhibits in vitro and in vivo multiple myeloma cell growth. Clin Cancer Res. 2013;19(8):2096-2106. 73. Morelli E, Biamonte L, Federico C, et al. Therapeutic vulnerability of multiple myeloma to MIR17PTi, a first-in-class inhibitor of pri-miR-17-92. Blood. 2018;132(10):10501063. 74. Amodio N, Stamato MA, Gulla AM, et al. Therapeutic targeting of miR-29b/HDAC4 epigenetic loop in multiple myeloma. Mol Cancer Ther. 2016;15(6):1364-1375. 75. Di Martino MT, Leone E, Amodio N, et al. Synthetic miR-34a mimics as a novel therapeutic agent for multiple myeloma: in vitro and in vivo evidence. Clin Cancer Res. 2012;18(22):6260-6270. 76. Morelli E, Leone E, Cantafio ME, et al. Selective targeting of IRF4 by synthetic microRNA-125b-5p mimics induces antimultiple myeloma activity in vitro and in vivo. Leukemia. 2015;29(11):2173-2183. 77. Caracciolo D, Di Martino MT, Amodio N, et al. miR-22 suppresses DNA ligase III addiction in multiple myeloma. Leukemia. 2019;33(2):487-498. 78. Gulla A, Di Martino MT, Gallo Cantafio ME, et al. A 13 mer LNA-i-miR-221 inhibitor restores drug sensitivity in melphalan-refractory multiple myeloma cells. Clin Cancer Res. 2016;22(5):1222-1233. 79. Zhao JJ, Chu ZB, Hu Y, et al. Targeting the miR-221-222/PUMA/BAK/BAX pathway abrogates dexamethasone resistance in multiple myeloma. Cancer Res. 2015;75(20): 4384-4397. 80. Botta C, Cuce M, Pitari MR, et al. MiR-29b antagonizes the pro-inflammatory tumorpromoting activity of multiple myelomaeducated dendritic cells. Leukemia. 2018;32(4):1003-1015. 81. Pitari MR, Rossi M, Amodio N, et al. Inhibition of miR-21 restores RANKL/OPG ratio in multiple myeloma-derived bone marrow stromal cells and impairs the resorbing activity of mature osteoclasts. Oncotarget. 2015;6(29):27343-27358. 82. Hong DS, Kang YK, Borad M, et al. Phase 1 study of MRX34, a liposomal miR-34a mimic, in patients with advanced solid tumours. Br J Cancer. 2020;122(11):16301637. 83. Rupaimoole R, Slack FJ. MicroRNA therapeutics: towards a new era for the management of cancer and other diseases. Nat Rev Drug Discov. 2017;16(3):203-222. 84. Morelli E, Gulla A, Rocca R, et al. The noncoding RNA landscape of plasma cell dyscrasias. Cancers (Basel). 2020;12(2). 85. Amodio N, Stamato MA, Juli G, et al. Drugging the lncRNA MALAT1 via LNA gapmeR ASO inhibits gene expression of proteasome subunits and triggers anti-multiple myeloma activity. Leukemia. 2018;32(9): 1948-1957. 86. Samur MK, Minvielle S, Gulla A, et al. Long intergenic non-coding RNAs have an independent impact on survival in multiple myeloma. Leukemia. 2018;32(12):2626-2635. 87. Chesi M, Mirza NN, Garbitt VM, et al. IAP antagonists induce anti-tumor immunity in multiple myeloma. Nat Med. 2016;22(12): 1411-1420.

2366

88. Fan F, Bashari MH, Morelli E, et al. The AP1 transcription factor JunB is essential for multiple myeloma cell proliferation and drug resistance in the bone marrow microenvironment. Leukemia. 2017;31(7): 1570-1581. 89. Dimopoulos M, Spencer A, Attal M, et al. Lenalidomide plus dexamethasone for relapsed or refractory multiple myeloma. N Engl J Med. 2007;357(21):2123-2132. 90. McCarthy PL, Owzar K, Hofmeister CC, et al. Lenalidomide after stem-cell transplantation for multiple myeloma. N Engl J Med. 2012;366(19):1770-1781. 91. Benboubker L, Dimopoulos MA, Dispenzieri A, et al. Lenalidomide and dexamethasone in transplant-ineligible patients with myeloma. N Engl J Med. 2014;371 (10):906-917. 92. Miguel JS, Weisel K, Moreau P, et al. Pomalidomide plus low-dose dexamethasone versus high-dose dexamethasone alone for patients with relapsed and refractory multiple myeloma (MM-003): a randomised, open-label, phase 3 trial. Lancet Oncol. 2013;14(11):1055-1066. 93. Davies F, Baz R. Lenalidomide mode of action: linking bench and clinical findings. Blood Rev. 2010;24(Suppl 1):S13-S19. 94. Gandhi AK, Kang J, Havens CG, et al. Immunomodulatory agents lenalidomide and pomalidomide co-stimulate T cells by inducing degradation of T cell repressors Ikaros and Aiolos via modulation of the E3 ubiquitin ligase complex CRL4(CRBN). Br J Haematol. 2014;164(6):811-821. 95. Hideshima T, Ogiya D, Liu J, et al. Immunomodulatory drugs activate NK cells via both Zap-70 and cereblon-dependent pathways. Leukemia. 2020 Apr 1. doi.org/10.1038/s41375-020-0809-x. [Epub ahead of print] 96. Luptakova K, Rosenblatt J, Glotzbecker B, et al. Lenalidomide enhances anti-myeloma cellular immunity. Cancer Immunol Immunother. 2013;62(1):39-49. 97. Facon T, Kumar S, Plesner T, et al. Daratumumab plus lenalidomide and dexamethasone for untreated myeloma. N Engl J Med. 2019;380(22):2104-2115. 98. Dimopoulos MA, Oriol A, Nahi H, et al. Daratumumab, lenalidomide, and dexamethasone for multiple myeloma. N Engl J Med. 2016;375(14):1319-1331. 99. Dimopoulos MA, Dytfeld D, Grosicki S, et al. Elotuzumab plus pomalidomide and dexamethasone for multiple myeloma. N Engl J Med. 2018;379(19):1811-1822. 100. Lonial S, Dimopoulos M, Palumbo A, et al. Elotuzumab therapy for relapsed or refractory multiple myeloma. N Engl J Med. 2015;373(7):621-631. 101. Attal M, Richardson PG, Rajkumar SV, et al. Isatuximab plus pomalidomide and lowdose dexamethasone versus pomalidomide and low-dose dexamethasone in patients with relapsed and refractory multiple myeloma (ICARIA-MM): a randomised, multicentre, open-label, phase 3 study. Lancet. 2019;394(10214):2096-2107. 102. Martin T, Baz R, Benson DM, et al. A phase 1b study of isatuximab plus lenalidomide and dexamethasone for relapsed/refractory multiple myeloma. Blood. 2017;129(25): 3294-3303. 103. Bjorklund CC, Kang J, Amatangelo M, et al. Iberdomide (CC-220) is a potent cereblon E3 ligase modulator with antitumor and immunostimulatory activities in lenalidomide- and pomalidomide-resistant multiple myeloma cells with dysregulated CRBN.

Leukemia. 2020;34(4):1197-1201. 104. Hansen JD, Correa M, Nagy MA, et al. Discovery of CRBN E3 ligase modulator CC-92480 for the treatment of relapsed and refractory multiple myeloma. J Med Chem. 2020;63(13):6648-6676. 105. Zavidij O, Haradhvala NJ, Mouhieddine TH, et al. Ghobrial single-cell RNA sequencing reveals compromised immune microenvironment in precursor stages of multiple myeloma. Nature Cancer. 2020 Apr 27. doi.org/10.1038/s43018-020-0053-3. [Epub ahead of print] 106. Chauhan D, Singh AV, Brahmandam M, et al. Functional interaction of plasmacytoid dendritic cells with multiple myeloma cells: a therapeutic target. Cancer Cell. 2009;16(4): 309-323. 107. Ray A, Das DS, Song Y, et al. A novel agent SL-401 induces anti-myeloma activity by targeting plasmacytoid dendritic cells, osteoclastogenesis and cancer stem-like cells. Leukemia. 2017;31(12):2652-2660. 108. Ray A, Tian Z, Das DS, et al. A novel TLR-9 agonist C792 inhibits plasmacytoid dendritic cell-induced myeloma cell growth and enhance cytotoxicity of bortezomib. Leukemia. 2014;28(8):1716-1724. 109. Ray A, Song Y, Chauhan D, Anderson KC. Blockade of ubiquitin receptor Rpn13 in plasmacytoid dendritic cells triggers antimyeloma immunity. Blood Cancer J. 2019;9(8):64. 110. Holthof LC, Mutis T. Challenges for immunotherapy in multiple myeloma: bone marrow microenvironment-mediated immune suppression and immune resistance. Cancers (Basel). 2020;12(4). 111. Costello C. The future of checkpoint inhibition in multiple myeloma? Lancet Haematol. 2019;6(9):e439-e440. 112. Costa F, Das R, Kini Bailur J, Dhodapkar K, Dhodapkar MV. Checkpoint inhibition in myeloma: opportunities and challenges. Front Immunol. 2018;9:2204. 113. Wudhikarn K, Wills B, Lesokhin AM. Monoclonal antibodies in multiple myeloma: current and emerging targets and mechanisms of action. Best Pract Res Clin Haematol. 2020;33(1):101143. 114. Chari A, Martinez-Lopez J, Mateos MV, et al. Daratumumab plus carfilzomib and dexamethasone in patients with relapsed or refractory multiple myeloma. Blood. 2019;134(5):421-431. 115. Spencer A, Lentzsch S, Weisel K, et al. Daratumumab plus bortezomib and dexamethasone versus bortezomib and dexamethasone in relapsed or refractory multiple myeloma: updated analysis of CASTOR. Haematologica. 2018;103(12):2079-2087. 116. Voorhees PM, Kaufman JL, Laubach JP, et al. Daratumumab, lenalidomide, bortezomib, and dexamethasone for transplant-eligible newly diagnosed multiple myeloma: GRIFFIN. Blood. 2020 Apr 23. doi: 10.1182/blood.2020005288. [Epub ahead of print] 117. Mateos MV, Nahi H, Legiec W, et al. Subcutaneous versus intravenous daratumumab in patients with relapsed or refractory multiple myeloma (COLUMBA): a multicentre, open-label, non-inferiority, randomised, phase 3 trial. Lancet Haematol. 2020;7(5):e370-e380. 118.Cho SF, Anderson KC, Tai YT. Targeting B cell maturation antigen (BCMA) in multiple myeloma: potential uses of BCMAbased immunotherapy. Front Immunol. 2018;9:1821. 119. Xing L, Lin L, Yu T, et al. A novel BCMA

haematologica | 2020; 105(10)

MM: current therapy and future prospects PBD-ADC with ATM/ATR/WEE1 inhibitors or bortezomib induce synergistic lethality in multiple myeloma. Leukemia. 2020 Feb 14. doi: 10.1038/s41375-020-0745-9. [Epub ahead of print] 120. Trudel S, Lendvai N, Popat R, et al. Targeting B-cell maturation antigen with GSK2857916 antibody-drug conjugate in relapsed or refractory multiple myeloma (BMA117159): a dose escalation and expansion phase 1 trial. Lancet Oncol. 2018;19(12):1641-1653. 121. Kinneer K, Meekin J, Tiberghien AC, et al. SLC46A3 as a potential predictive biomarker for antibody-drug conjugates bearing noncleavable linked maytansinoid and pyrrolobenzodiazepine warheads. Clin Cancer Res. 2018;24(24):6570-6582. 122. Tai YT, Mayes PA, Acharya C, et al. Novel anti-B-cell maturation antigen antibodydrug conjugate (GSK2857916) selectively induces killing of multiple myeloma. Blood. 2014;123(20):3128-3138. 123. Tai YT, Acharya C, An G, et al. APRIL and BCMA promote human multiple myeloma growth and immunosuppression in the bone marrow microenvironment. Blood. 2016; 127(25):3225-3236. 124. Frerichs KA, Broekmans MEC, Marin Soto JA, et al. Preclinical activity of JNJ-7957, a novel BCMAxCD3 bispecific antibody for the treatment of multiple myeloma, is potentiated by daratumumab. Clin Cancer Res. 2020;26(9):2203-2215. 125. Seckinger A, Delgado JA, Moser S, et al. Target expression, generation, preclinical activity, and pharmacokinetics of the BCMA-T cell bispecific antibody EM801 for multiple myeloma treatment. Cancer Cell. 2017;31(3):396-410. 126. Hipp S, Tai YT, Blanset D, et al. A novel BCMA/CD3 bispecific T-cell engager for the treatment of multiple myeloma induces selective lysis in vitro and in vivo. Leukemia. 2017;31(8):1743-1751. 127. Noonan KA, Huff CA, Davis J, et al. Adoptive transfer of activated marrow-infiltrating lymphocytes induces measurable antitumor immunity in the bone marrow in multiple myeloma. Sci Transl Med. 2015;7 (288):288ra78. 128. Rodriguez-Otero P, Paiva B, Engelhardt M, Prosper F, San Miguel JF. Is immunotherapy here to stay in multiple myeloma? Haematologica. 2017;102(3):423-432. 129. D'Agostino M, Raje N. Anti-BCMA CAR Tcell therapy in multiple myeloma: can we do

haematologica | 2020; 105(10)

better? Leukemia. 2020;34(1):21-34. 130. Yan Z, Cao J, Cheng H, et al. A combination of humanised anti-CD19 and anti-BCMA CAR T cells in patients with relapsed or refractory multiple myeloma: a single-arm, phase 2 trial. Lancet Haematol. 2019;6(10):e521-e529. 131. Raje N, Berdeja J, Lin Y, et al. Anti-BCMA CAR T-cell therapy bb2121 in relapsed or refractory multiple myeloma. N Engl J Med. 2019;380(18):1726-1737. 132. Xu J, Chen LJ, Yang SS, et al. Exploratory trial of a biepitopic CAR T-targeting B cell maturation antigen in relapsed/refractory multiple myeloma. Proc Natl Acad Sci U S A. 2019;116(19):9543-9551. 133. Bae J, Smith R, Daley J, et al. Myeloma-specific multiple peptides able to generate cytotoxic T lymphocytes: a potential therapeutic application in multiple myeloma and other plasma cell disorders. Clin Cancer Res. 2012;18(17):4850-4860. 134. Bae J, Prabhala R, Voskertchian A, et al. A multiepitope of XBP1, CD138 and CS1 peptides induces myeloma-specific cytotoxic T lymphocytes in T cells of smoldering myeloma patients. Leukemia. 2015; 29(1):218-229. 135. Bae J, Samur M, Richardson P, Munshi NC, Anderson KC. Selective targeting of multiple myeloma by B cell maturation antigen (BCMA)-specific central memory CD8(+) cytotoxic T lymphocytes: immunotherapeutic application in vaccination and adoptive immunotherapy. Leukemia. 2019; 33(9):2208-2226. 136. Rosenblatt J, Avivi I, Vasir B, et al. Vaccination with dendritic cell/tumor fusions following autologous stem cell transplant induces immunologic and clinical responses in multiple myeloma patients. Clin Cancer Res. 2013;19(13):3640-3648. 137. Chari A, Vogl DT, Gavriatopoulou M, et al. Oral selinexor-dexamethasone for tripleclass refractory multiple myeloma. N Engl J Med. 2019;381(8):727-738. 138. Tassone P, Neri P, Carrasco DR, et al. A clinically relevant SCID-hu in vivo model of human multiple myeloma. Blood. 2005;106 (2):713-716. 139. Calimeri T, Battista E, Conforti F, et al. A unique three-dimensional SCID-polymeric scaffold (SCID-synth-hu) model for in vivo expansion of human primary multiple myeloma cells. Leukemia. 2011;25(4):707711.

140. Radl J, Punt YA, van den Enden-Vieveen MH, et al. The 5T mouse multiple myeloma model: absence of c-myc oncogene rearrangement in early transplant generations. Br J Cancer. 1990;61(2):276-278. 141. Calcinotto A, Brevi A, Chesi M, et al. Microbiota-driven interleukin-17-producing cells and eosinophils synergize to accelerate multiple myeloma progression. Nat Commun. 2018;9(1):4832. 142. Peled JU, Devlin SM, Staffas A, et al. Intestinal microbiota and relapse after hematopoietic-cell transplantation. J Clin Oncol. 2017;35(15):1650-1659. 143. Pianko MJ, Devlin SM, Littmann ER, et al. Minimal residual disease negativity in multiple myeloma is associated with intestinal microbiota composition. Blood Adv. 2019;3(13):2040-2044. 144. Touzeau C, Maciag P, Amiot M, Moreau P. Targeting Bcl-2 for the treatment of multiple myeloma. Leukemia. 2018;32(9):18991907. 145.Kumar S, Kaufman JL, Gasparetto C, et al. Efficacy of venetoclax as targeted therapy for relapsed/refractory t(11;14) multiple myeloma. Blood. 2017;130(22):2401-2409. 146. Moreau P, Chanan-Khan A, Roberts AW, et al. Promising efficacy and acceptable safety of venetoclax plus bortezomib and dexamethasone in relapsed/refractory MM. Blood. 2017;130(22):2392-2400. 147. Lagana A, Beno I, Melnekoff D, et al. Precision medicine for relapsed multiple myeloma on the basis of an integrative multiomics approach. JCO Precis Oncol. 2018;2018. 148. Moreau P, Attal M, Hulin C, et al. Bortezomib, thalidomide, and dexamethasone with or without daratumumab before and after autologous stem-cell transplantation for newly diagnosed multiple myeloma (CASSIOPEIA): a randomised, open-label, phase 3 study. Lancet. 2019;394(10192):2938. 149. Richardson PG, Bringhen S, Voorhees P, et al. Melflufen plus dexamethasone in relapsed and refractory multiple myeloma (O-12-M1): a multicentre, international, open-label, phase 1-2 study. Lancet Haematol. 2020;7(5):e395-e407. 150. Schjesvold F, Robak P, Pour L, Aschan J, Sonneveld P. OCEAN: a randomized phase III study of melflufen + dexamethasone to treat relapsed refractory multiple myeloma. Future Oncol. 2020;16(11):631-641.

2367

REVIEW ARTICLE Ferrata Storti Foundation

The molecular basis for the prothrombotic state in sickle cell disease Arun S. Shet,1 Maria A. Lizarralde-Iragorri1 and Rakhi P. Naik2

Laboratory of Sickle Thrombosis and Vascular Biology, National Heart, Lung, and Blood Institute, NIH, Bethesda and 2Division of Hematology, Department of Medicine, Johns Hopkins University, Baltimore, MD, USA

Haematologica 2020 Volume 105(10):2368-2379

ABSTRACT

he genetic and molecular basis of sickle cell disease (SCD) has long been characterized but the pathophysiological basis has not been entirely defined. How a red cell hemolytic disorder initiates inflammation, endothelial dysfunction, coagulation activation, and eventually leads to vascular thrombosis, is yet to be elucidated. Recent evidence has demonstrated a high frequency of unprovoked/recurrent venous thromboembolism (VTE) in SCD, with an increased risk of mortality among patients with a history of VTE. Here, we provide an in-depth review of the molecular basis for the prothrombotic state in SCD, specifically highlighting emerging evidence for activation of overlapping inflammation and coagulation pathways that predispose to venous thromboembolism. We share perspectives in managing venous thrombosis in SCD, highlighting innovative therapies with the potential to influence the clinical course of disease and reduce thrombotic risk, while maintaining an acceptable safety profile.

Introduction

Correspondence: ARUN S. SHET arun.shet@nih.gov Received: April 6, 2020. Accepted: July 22, 2020. Pre-published: August 13, 2020. doi:10.3324/haematol.2019.239350 ©2020 NIH (National Institutes of Health)

2368

Renewed interest in sickle cell disease (SCD) as a significant global health problem by academia, industry and policy makers is leading the resurgence of efforts to treat or cure patients with this disease. Drug development has recently led to US Food and Drug Administration (FDA) approval of three new agents (L-glutamine, crizanlizumab, voxelotor) adding much needed diversity to the hitherto lone disease-modifying therapy, hydroxyurea (HU).1 Moreover, advanced phase clinical trials of molecules targeting diverse disease mechanisms, if efficacious, could ameliorate the protean manifestations of SCD. In addition, curing SCD either through stem cell transplantation or gene therapy has become a reality for a small number of patients.2 Given the October 2019 joint National Institutes of Health (NIH)-Bill and Melinda Gates Foundation funding declaration, inclusion of patients from high disease burden resource-scarce settings in such trials of novel curative therapies seems plausible.3 Prospects appear particularly promising for SCD patients in resource-scarce settings worldwide, where the need is greatest. Despite these advances, many patients with SCD continue to experience severe complications, and understanding the pathophysiology of “downstream” events remains important for the development of therapies to target specific complications. One of these phenomena is the sickle prothrombotic state that is largely believed to be responsible for both arterial and venous thrombosis in SCD. In its most devastating form, thrombosis occurs in arteries leading to overt stroke and silent cerebral infarction in SCD patients as early as childhood.4,5 In adulthood, SCD patients develop deep venous thrombosis (DVT) and pulmonary embolism (PE), collectively termed venous thromboembolism (VTE).6-8 Thrombotic vasculopathy in SCD is accompanied by significant organ dysfunction, morbidity from diminished quality of life, and mortality.6,7 Yet, how the complex pathobiology initiated by sickle RBC-mediated endothelial inflammation/dysfunction and coagulation activation leads to vessel injury, leakage, and vascular thrombosis remains to be clarified. At the mechanistic level, there is a scientific gap in our understanding of coagulation-mediated pathologies of SCD, a fact noted in the National Heart, Lung, and Blood Institute (NHLBI) evidence-based guidelines,9 which makes this an active area of research. Therefore, gaining insight into the pathohaematologica | 2020; 105(10)

Prothrombotic state in sickle cell disease

physiological basis for the prothrombotic state in patients with SCD and how one might attenuate thrombotic risk is of the utmost importance. Studies of vascular pathobiology using in vitro and animal models of SCD have identified both insightful and discrepant findings, when compared with human SCD, as noted in the accompanying review in this issue of the Journal by Conran and De Paula.10 Here we examine the molecular basis for the prothrombotic state and establish the scientific rationale to target dysregulated coagulation pathways in patients with SCD. For a more comprehensive overview of arterial thrombosis and/or SCD management, the reader is referred to recent publications in the literature.1,11-13

Sickle cell disease: an acquired hypercoagulable state The underlying mechanism(s) for the prothrombotic tendency in SCD has been the subject of speculation for many decades.14,15 Several clinical observations suggest that the presence of sickle hemoglobin (HbS) is either directly or indirectly associated with the development of thrombotic complications in SCD. The early age of onset of arterial complications such as stroke and silent cerebral infarction in children with hemoglobin SS disease, in particular, is a clear example of the thrombotic potential associated with hemoglobin S. Furthermore, in adults, a wide range of sickling hemoglobinopathies have been associated with the development of VTE, from hemoglobin SS to compound heterozygous states to sickle cell trait (SCT). Numerous epidemiological studies of VTE in SCD have verified an increased risk of this complication (Table 1). A retrospective analysis of the Cooperative Study of SCD demonstrated a 11.3% cumulative incidence of VTE by the age of 40,16 and a similarly high incidence was found in a large California discharge database.7 These estimates are particularly striking when compared to population studies of patients with thrombophilia, as the incidence rates of VTE observed in SCD are similar to those found in families with high-risk thrombophilia such as protein C and S deficiency.16 In addition, after controlling for confounding factors such as hospitalization, studies have found that VTE rates are high even among SCD with a lower number of hospitalizations,7 and that VTE prevalence in pregnant women with SCD increases approximately 5-fold compared to non-SCD pregnant women.17,18 Furthermore, the recurrence rate of VTE in SCD has also been noted to be exceedingly high (>35% at 5 years),8,16 another indication of the persistent thrombogenic potential in SCD. Further confirmation of the relationship between the presence of HbS and a prothrombotic state are observations noted in healthy individuals with SCT. In two large population-based studies, VTE risk in individuals with SCT was almost 2-fold higher than ethnic-matched counterparts, despite the absence of either clinical sickling symptoms or HbS percentages encountered in patients with SCD (Table 1).19,20 As with the known pathophysiology of SCD complications, the presence of HbS alone is unlikely to be the sole contributor to thrombotic risk. Additional mechanisms, described further below, are likely to contribute to the acquired prothrombotic state in SCD. haematologica | 2020; 105(10)

Clinical and genetic risk factors for the prothrombotic state Clinical and biological factors provide insight into why a genetic disorder primarily involving red cells sets off a cascade of events leading to an acquired prothrombotic state. Risk factors for VTE in SCD include increased disease severity (as measured by averaging ≥3 hospital admissions a year for VOC), exposure to erythropoiesis stimulating agents or blood transfusion, insertion of central venous catheters (CVC), surgical splenectomy, and hospitalization.21-24 As mentioned above, not only are SCD patients at a high risk for developing early onset VTE, but their risk for VTE recurrence after an index VTE has been noted to be high. Clinical risk factors for VTE recurrence include averaging >3 hospital admissions a year, lower extremity DVT as the index event, and a prior history of pneumonia/acute chest syndrome (ACS).8,23 Incomplete adherence to anticoagulation is also likely to contribute to higher recurrence rates, though this has not been formally evaluated. In a single center retrospective study, use of direct oral anticoagulants (DOAC) was associated with lower recurrence rate.23 Genotype may also modify thrombotic risk in SCD. Certainly, children with HbSS/Sβ0 are at highest risk for arterial thrombosis such as stroke.5,23 In terms of VTE, the influence of genotype has not been fully elucidated, with some studies demonstrating an increased risk of VTE among sickle variant syndromes and others showing the highest risk among HbSS/Sβ0.6,16 The influence of co-inheritance of either α-thalassemia or gene modifiers regulating human fetal hemoglobin expression25-27 or both, has not been studied for VTE but may attenuate stroke risk.5 The influence of heritable thrombophilia mutations on VTE risk in SCD remains to be completely determined. Genetic variants, factor V Leiden and prothrombin FII G20210A in particular explain up to 50% of unprovoked VTE among Caucasians.28 Notably, VTE risk is 5-fold higher in individuals reporting African descent compared with those reporting Asian descent, whereas white individuals have only an intermediate-level risk.29 However, factor V Leiden and prothrombin FII G20210A have low allele frequencies among individuals reporting African descent and are not associated with venous thrombosis in SCD.30,31 Thus, other heritable or acquired thrombophilia-associated mutations could explain thrombotic risk in SCD. One recent study identified two known thrombomodulin gene variants (THBD rs2567617 and rs1998081) that were associated with arterial and venous thrombosis in SCD patients.23 A recent genome-wide association study conducted among African Americans in the general population identified three novel intronic gene variations (LEMD3, LY86, LOC100130298) associated with higher odds of developing VTE.32 LEMD3 encodes for a nuclear membrane protein that interacts with transforming growth factor (TGF)-β to downregulate the activation of TGF-β target genes33,34 and LY86 encodes for MD-1, which regulates expression of a cell surface protein homologous to toll-like receptor (TLR) 4.35,36 Given the involvement of the innate immune system in both VTE37 and the vascular pathobiology of SCD,10 genetic dysregulation of these pathways in SCD patients could influence thrombophilia. Acquired mutations also influence thrombotic risk; JAK2V617F the most frequent age-induced clonal hematopoietic mutation is associated with increased 2369

A.S. Shet et al.

venous thrombotic events.38 Future studies may clarify the role of these heritable or acquired mutations.

Thrombo-inflammatory processes in sickle cell disease that alter hemostatic balance A triad of thrombosis risk factors first described by Virchow that includes increased blood coagulability, altered blood flow (stasis) and endothelial dysfunction, are all evident in SCD.39 The endothelium, a critical regulator of thrombo-inflammatory processes maintains vascular health by exerting anti-coagulant, anti-inflammatory, and anti-platelet actions. The sickle proinflammatory state leads to endothelial dysfunction, thereby shifting the hemostatic balance towards a prothrombotic state (Figure 1). Accumulated evidence suggests that SCD is a thrombo-inflammatory disorder as reflected by alterations in

components of coagulation and inflammatory pathways.40-43 This is most clearly demonstrated in the acute exaggerated thrombo-inflammatory response that accompanies ischemia-reperfusion (IR) injury in SCD patients.44 Endothelial inflammation leads to surface expression of adhesion molecules (P-selectin and E-selectin) and release of prothrombotic granule contents (von Willebrand factor and FVIII), both effects enhancing leukocyte/platelet adhesion (Figure 2). Repeated VOC episodes amongst other pathophysiology leads to hemolysis and subsequent release of cell-free heme/hemoglobin. By activating converging inflammatory pathways, such as TLR signaling,45 NETosis/neutrophil extracellular trap (NET) formation46 and priming the inflammasome,47 cell-free heme amplifies inflammation (Figures 1 and 2).48 Inflammation, shear stress and hypoxia, which are common phenomena in VOC, under experimental conditions can induce abnormal endothelial TF gene and protein expression.49-51 SCD

Table 1. Epidemiological studies of venous thromboembolism (VTE) in sickle cell disease (SCD) and sickle cell trait (SCT).

Author and ref. 65

Stein et al.

Austin et al.137

Naik et al.16

Seaman et al.18

Study aim, design and setting

Main findings

Aim: Determine frequency and prevalence of PE and DVT. Retrospective analysis of National Hospital Discharge Survey comparing SCD patients ≤40 years vs. African-American patients. Aim: Evaluate the incidence of VTE in individuals with SCT. Case-control study of N=515 hospitalized patients vs. 555 controls. Aim: Define frequency and characteristics of VTE in SCD. Retrospective single center study of N=404 SCD patients (Sickle Cell Center for Adults-Johns Hopkins). Aim: Evaluate rates and risk factors related with VTE in pregnant women with SCD. Retrospective using Pennsylvania Health Care Database.

SCD patients have a high prevalence of PE (0.44%) compared with African-Americans without SCD (0.12%). DVT prevalence was similar in both groups, 0.44% and 0.40%, respectively.

Naik et al.6

Aim: Determine the incidence of first VTE. Retrospective cohort study using prospective data from the co-operative study of SCD. N= 1,523 SCD patients ≥15 years.

Folsom et al.19

Aim: Risk of VTE in individuals with SCT. Prospective population-based cohort (1987-2011). N=268 SCT middle-aged patients. Aim: Determine risk of VTE in individuals with SCT. Prospective cohort study with nested case-control design. N=6,758 Aim: Evaluate VTE incidence in SCD. Retrospective study using California administrative database. N=6,237 SCD patients. Aim: Evaluate the VTE incidence in children with SCD. Pediatric health information database. N=181. Aim: Determine VTE recurrence and bleeding risk in SCD patients with index VTE. Retrospective study using California administrative database. N=877 SCD patients with an index VTE. Aim: Investigate the genetic and clinical predictors of arterial and venous thrombosis in SCD. Longitudinal single center cohort study. N=1,193 pediatric and adult SCD patients.

Little et al.20

Brunson et al.7

Kumar et al.22

Brunson et al.8

Srisuwananukorn et al.23

The risk of VTE is increased approx. 2-fold among SCT patients compared with controls. VTE is common (25%), occurring at a mean age of 30 years. Sickle variants genotypes and tricuspid regurgitant jet velocity ≥2.5 m/s were associated with non-catheter-related VTE. Pregnancy-related VTE in women with SCD appeared to be 1.5-5 times greater than pregnancy-related VTE in the general population. N=212 SCD pregnant women The cumulative incidence for first VTE was 11.3% by age 40 years. Incidence of PE exceeded that of isolated DVT. SCD patients with VTE had a higher mortality than those without VTE. SCT carries a 2-fold increased risk of PE but did not appear to be associated with elevated DVT risk. SCT individuals had a higher risk of VTE, particularly PE, compared with non-carriers. The cumulative incidence of VTE was high (11.2%) in SCD patients. The occurrence of VTE was associated with higher mortality. 1.7% developed VTE, use of a central venous catheter was associated with VTE development, and VTE was associated with mortality. The cumulative incidence of VTE recurrence was 13.2% and 24.1% at 1-year and 5-year follow-up. The cumulative incidence of bleeding was 4.9% and 7.9% at 6 months and 1 year following an incident VTE. VTE risk was independently associated with HbSS/Sβ0 genotype, HU use, lower estimated glomerular filtration rate, and higher Hb and WBC count. THBD variants, rs2567617 and rs1998081 were associated with thrombosis.

PE: pulmonary embolism; DVT: deep vein thrombosis; N: number; HU: hydroxyurea; THBD: thrombomodulin; Hb: hemoglobin; WBC: white blood cell.

2370

haematologica | 2020; 105(10)

Prothrombotic state in sickle cell disease

patients demonstrate abnormally elevated levels of intravascular TF that is believed to trigger activation of coagulation and pathological thrombosis.52-55 Critically, intravascular TF binding with factor VIIa activates the extrinsic pathway of coagulation by converting coagulation factor X to Xa and generating thrombin (Figure 3), which, unchecked, leads to vascular fibrin deposition. From this perspective, defining the molecular mechanisms regulating these thrombo-inflammatory processes in SCD and identifying interventions to counter vascular thrombosis is of major relevance.

Cellular components of blood that facilitate thromboinflammation Sickle red cells That red cells in SCD are likely to be involved in thrombus formation is supported by the relationship between hematocrit and VTE,24 possibly resulting from alterations in viscosity, adhesive cellular interactions, and microvascular stasis.56,57 Studies of sickle red cells have identified numerous receptors and ligands that mediate adhesive interactions with the vessel wall, implicating their role in

Figure 1. Sickle hemoglobin (HbS) polymerization, hemolysis and ischemia/reperfusion injury induce chronic inflammation. (A) Primary to sickle cell disease pathology is polymerization of HbS during red cell deoxygenation that results in sickled red cells and frequent painful vaso-occlusive crises. Hypoxia in the venous vasculature and valve pockets may worsen sickling mediated hemolysis and venous endothelial injury/inflammation. (B) Repeated sickling and unsickling episodes lead to intravascular hemolysis and release of free heme that consumes nitric oxide (NO). Sickle RBC also activate neutrophils among other cells, forming hetero and homotypic aggregates with blood cells that lead to vaso-occlusion in the post capillary venule. Endothelial damage and endothelial surface expression of adhesion and procoagulant molecules leads to transit delays increasing the potential for stasis and further sickling. Damage-associated molecular pattern (DAMP) molecules (free heme and high mobility group box 1 [HMGB1]) activate of various inflammatory pathways, e.g. NETosis, toll-like receptor (TLR) signaling, innate immune response and production of reactive oxidative species (ROS) lead to chronic inflammation. Repeated episodes of vaso-occlusive crisis (VOC) leads to ischemia followed by reperfusion mediating a well characterized injury response in the vascular endothelium. Figure created with BioRender.com. VCAM1: vascular cell adhesion molecule 1; EV TF: extracellular vesicle tissue factor; NET: neutrophil extracellular trap.

haematologica | 2020; 105(10)

2371

A.S. Shet et al.

post capillary venule occlusion (Figure 2).58,59 Elevated numbers of phosphatidylserine (PS) positive sickle red cell and red cell derived extracellular vesicles (EV) are observed in SCD patients, which correlate with markers of thrombin generation.55,60,61 In non-SCD models, venous thrombus size is impacted by localization of red cells within blood clots, which in turn is affected by fibrin network density and fibrin Îą-chain crosslinking activity.62,63 Recent studies in SCD mice have shown higher amounts of fibrin deposition and red cell entrapment within the developing venous thrombus64 that are likely to impact thrombus structure and stability. These findings imply that venous clots in SCD patients and individuals with SCT are more friable and prone to embolization, possibly explaining why individuals with sickling hemoglobinopathies appear to have a higher risk of PE compared to DVT.6,7,19,65 However, in situ pulmonary thrombosis in SCD patients66

suggests heterogenous mechanisms, that may include abnormal pulmonary vascular endothelial TF expression.67

Platelets Due to the physiological role platelets play in primary hemostasis, platelet activation probably contributes to thrombosis in SCD. For example, activated platelets in SCD can recruit leukocytes to sites of inflammation through surface P-selectin, and release prothrombotic granule contents. Besides, activated platelets form homotypic and heterotypic cell-aggregates, and platelet-neutrophil aggregates contribute to pulmonary arteriolar micro emboli.68-71 In recent studies, both murine models and SCD patients have consistently demonstrated activation of the platelet NLRP3 inflammasome, suggesting an autocrine feedback loop for IL-1Î˛ driven priming of innate immune and vascular endothelial cells.68,72 Moreover, solu-

Figure 2. Inflammation and adhesion propagate stasis and contribute to fibrin deposition. Persistent inflammation leads to endothelial cell activation and endothelial dysfunction. Increased tissue factor (TF) and TF+ extracellular vesicle (EV) can generate thrombin and activate coagulation. Thrombin also activates cell surface protease activating receptors (PAR), worsening endothelial inflammation and dysfunction. Abnormal surface expression of adhesion molecules (P-selectin, E-selectin and V-CAM) and platelet activation facilitate heterotypic cellular interactions. Similarly, increased heterotypic and homotypic cellular adhesion interactions mediated via cell surface ligands E-selectin/CD44, Laminin/LuBCAM, Îą-4/VCAM-1, PSR/PS etc., promote vascular stasis and favor thrombosis. Overall this prothrombotic milieu favors red cell entrapment, vascular fibrin deposition, and thrombus formation. Figure created with BioRender.com. PSR: phosphatidylserine receptor; PAR1: protease activated receptor 1; PSGL-1: P-selectin glycoprotein ligand 1; EV TF: extracellular vesicle tissue factor; ICAM-4: intercellular adhesion molecule; Lu/BCAM: Lutheran/basal cell adhesion molecule; PS: phosphatidylserine.

2372

haematologica | 2020; 105(10)

Prothrombotic state in sickle cell disease

ble CD40L and thrombospondin-1, two platelet-derived molecules are reportedly elevated in SCD patients with a history of ACS,73,74 which, in the light of dense platelet-rich thrombi found on autopsy in the pulmonary vasculature of over half of ACS patients,75 provides evidence for platelet-mediated thrombosis.

Leukocytes Innate immune cells mediate SCD pathophysiology, as recently reviewed in this Journal.76 Amongst these, monocytes and neutrophils facilitate thrombotic vasculopathy by directly inducing endothelial injury and activating coagulation. Neutrophils undergo NETosis, a neutrophil defense mechanism that involves the release of NET and initiates venous thrombosis.77 In vitro studies suggest that actually cell-free DNA (cfDNA) and histones, rather than intact NET, activate coagulation,78 and while SCD patients have elevated plasma levels of cfDNA, histones and NET,79,80 no direct evidence links them with VTE development. Neutrophil-endothelial and neutrophil-platelet cross-talk, mediated via P-selectin-PSGL1 interactions are key aspects of SCD pathobiology, as evidenced by the clinical efficacy of the anti-P-selectin antibody crizanlizumab (Figure 2).81 Monocytes regulate important aspects of blood coagulation, innate immune response, reticuloen-

dothelial function, and phagocytosis. Circulating TF+ monocytes and TF+ monocyte EV form the major fraction of measurable blood borne TF and thus contribute meaningfully to coagulation abnormalities observed in SCD patients.54,55 Monocyte activation by heme and placental growth factor (PLGF) released from sickle red cells can stimulate the production of proinflammatory cytokines and chemokines (e.g., IL-1β, TNF-α, MCP-1 and MIP1β),45,82,83 that in turn are capable of inducing TF gene expression.84 Lastly, frequent VOC in SCD could reduce the number of patrolling monocytes (CD14lowCD16+) responsible for restoration of endothelial function85 thereby augmenting endothelial injury and dysfunction.

Molecular components in blood facilitating thromboinflammation Blood borne tissue factor As described above, intravascular TF expression in the setting of SCD-related endothelial damage is likely a major contributor to the hypercoagulable state of SCD; however, the role of “blood borne” TF in SCD-related thrombosis is largely unknown.86-89 Studies have shown

Figure 3. The TF and contact pathways activate coagulation and generate thrombin. Abnormal expression of intravascular TF in sickle cell disease (SCD), the so called “blood borne TF”, triggers intravascular blood coagulation and pathological thrombosis. After binding with plasma factor VIIa, TF activates the extrinsic pathway of coagulation, generating thrombin, which mediates fibrin deposition, thrombosis and vascular remodeling. Besides lowered natural anticoagulant factor levels (thrombomodulin, protein C and S) also favor thrombosis. Activation of the contact pathway of coagulation by RBC phosphatidylserine (PS) exposure, cell-derived extracellular vesicles (EV), platelet polyphosphates, cfDNA and NET also sustain thrombin generation. TF and thrombin cause chronic inflammation possibly leading to endothelial injury, vascular permeability, angiogenesis and vascular remodeling, all of which are reflected by vasculopathy. Figure created with BioRender.com.

haematologica | 2020; 105(10)

2373

A.S. Shet et al.

that SCD patients have elevated blood borne TF procoagulant activity.52-55,90 Specifically, TF+ EV, which are derived from monocytes and endothelial cells, have been found to be elevated in SCD patients during acute VOC.54 Although, frequent VOC in SCD has been associated with an increased risk of VTE,7 no studies have evaluated whether cell-surface TF expression or TF+ EV play a direct role in initiating venous thrombosis. Several other regulatory mechanisms could modify VTE risk in SCD, as not all SCD patients with elevated levels of blood borne TF experience pathological thrombosis. For example, inactivation of blood borne TF by tissue factor pathway inhibitor (TFPI) probably protects against pathological thrombosis,91 but TFPI antigen levels in SCD patients are found to be normal.53 Besides, the procoagulant activity of TF is under post-translational regulatory control; these mechanisms are beyond the scope of this discussion but are reviewed elsewhere.92 Regulation of TF procoagulant activity could limit overt clinical thrombosis in SCD but does not appear sufficient to inhibit TF and thrombin-mediated inflammatory vasculopathy.41,93

Thrombin generation Activation of coagulation either through the TF or contact factor pathways leads to generation of thrombin. Thrombin, one of the most potent biological proinflammatory molecules, activates both cellular and plasmatic components of blood. These actions primarily include cleavage of factors comprising the plasma coagulation, complement and fibrinolytic cascades. In addition to converting soluble fibrinogen to insoluble fibrin, thrombin activates complement, innate immune cells, the endothelium and platelets, which in SCD mediates vascular thrombosis and vessel wall remodeling (Figure 3).41,93 Moreover, overwhelming evidence for thrombin-mediated vasculopathy is derived from elegant studies in sickle mouse models.40,41,94 In the aftermath of the initial thrombin burst generated by the TF-VIIa complex, a continuous supply of thrombin generation is required for pathological thrombosis. Thrombin generation is sustained in SCD patients even during the steady state, as demonstrated by the observation of elevated plasma levels of thrombin-anti thrombin complexes, D-dimers, and prothrombin fragment 1.295-97 and elevated in vitro thrombin generation potential.98 The contact pathway (plasma kallikrein-kinin system) is also implicated in sustaining thrombin generation in SCD patients,99,100 and endogenous activators of the contact pathway, e.g., PS on the surface of red cells and EV,99,100 polyphosphates (polyP) released by platelets, and nucleic acids (NET or cfDNA)80 are elevated in both SCD patients and those with SCT.101 Thus, both TF and the contact pathway activation contribute to thrombin generation in SCD patients.15

Canonical pathways critical to thrombosis pathophysiology Several concurrent inflammatory processes increase thrombotic risk in SCD. In the absence of pathogenic organisms, cell death and release of damage-associated molecular patterns (DAMP) trigger a vascular response termed “sterile inflammation”. In SCD, DAMP prime the innate immune system through converging inflammatory 2374

pathways, such as TLR signaling, NETosis (see above) and activation of the inflammasome.43 Murine studies that delineate how prototypic DAMP molecules (e.g., cell-free hemoglobin and high-mobility group box 1 [HMGB1]) play a critical role in VTE pathophysiology offer major insight into the occurrence of thrombosis in SCD.45,80,102-105 As noted above, cell-free hemoglobin is proinflammatory and prothrombotic due to nitric oxide (NO) consumption,106 induction of endothelial surface TF107 and TLR4induced monocyte priming.45 Similarly, disulfide HMGB1, derived from platelets, prime the leukocyte inflammasome and induce NET, facilitating VTE development in a murine model.102 Plasma HMGB1 levels are elevated in patients with SCD,72,103 but their contribution to NET formation and VTE pathophysiology in SCD is uncertain. Murine models of intravascular hemolysis demonstrate a co-operative role for complement activation and P-selectin in mediating thrombotic injury of hepatic and renal vascular endothelium, and provide insight into how sickle hemolysis might mediate organ injury.104,108 Taken together with the findings that HU attenuates complement activation in SCD patients109 and the efficacy of P-selectin blockade in preventing VOC,81 it is apparent that modulating these pathways may offset microvascular thrombosis. Advancing our understanding of these complex interactions between dysregulated inflammation and coagulation processes in SCD might offer additional insights and identify new therapeutic targets to limit thrombosis, particularly VTE.

Thromboinflammation, vascular injury and vasculopathy In addition to pathological thrombosis occurring as a result of the TF/VIIa complex (as described above), cell surface TF expression leads to inflammation. TF-mediated inflammation occurs either via the effects of downstream coagulation proteases (see above) on other vascular endothelial and blood cells94 or intracellularly, via its cytoplasmic tail.88 Thrombin’s subsequent interaction with endothelial cell surface protease activating receptors (PAR) accentuates vascular endothelial inflammation in SCD.110 Continued activation of coagulation, initially triggered by TF, likely occurs via EV and PS positive red cells in circulating blood, that by virtue of their surface PS content support the assembly of tenase and prothrombinase complexes111 (Figures 1 and 3). Moreover, in addition to procoagulant effects exerted by surface TF and PS, EV display membrane surface antigens, e.g., P-selectin, that engage with ligands to enhance production of TF+ EV112,113 and reactive oxygen species (ROS).105 Splenic hypofunction and diminished reticuloendothelial clearance in SCD lead to accumulation of prothrombotic mediators, i.e., PS+ red cells, red cell derived EV and leukocyte/endothelial derived TF+ EV. In venous vascular beds prone to stasis and hypoxia, accumulation of procoagulant factors may overwhelm anticoagulant defenses and lead to pathological thrombosis. A likely sequence of events may include the following: (i) sluggish cell transit through the post capillary venules which may increase “delay time”114 augmenting the fraction of polymerized HbS; and (ii) formation of hetero and homotypic cellular aggregates that, coupled with haematologica | 2020; 105(10)

Prothrombotic state in sickle cell disease

increased endothelial cell adherence promote stasis, lead to further red cell sickling, vaso-occlusion, and ischemic crisis. These sickling cycles promote repeated vascular injury, vasomotor dysfunction, chronic inflammation and vasculopathy in patients with SCD.115,116 Moreover, the cumulative effects of stasis, altered rheology, vascular fibrin deposition and chronic inflammation ultimately leads to chronic vasculopathy47 and characterizes the multiorgan failure observed in SCD patients.117 Vasculopathy in patients experiencing DVT is reflected by the onset of post-thrombotic syndrome, valvular venous insufficiency or leg ulceration and, in patients experiencing PE, by the onset of chronic thromboembolic pulmonary hypertension (CTEPH).118 The sickle prothrombotic state and resulting VTE, therefore, results in considerable cardiopulmonary morbidity and mortality.

Managing venous thrombosis in sickle cell disease Because prospective trials of anticoagulation in VTE have not, to our knowledge, included subjects with SCD, recommendations on VTE management in SCD patients generally rely on clinical guidelines developed for the general population.119 It should be noted that, while clinical probability scores and laboratory biomarkers help determine the pretest probability of VTE in the general population,120 there is no evidence to support this approach in SCD patients. In addition, the lack of prospective primary prevention and/or management studies of VTE in either pregnant women with SCD or high-risk SCD patients leads to over-reliance on data from prospective studies of VTE in individuals with inherited prothrombotic states.39 Patients with SCD who are suspected of having VTE should undergo compression ultrasound Dopplers for DVT and multidetector computerized tomographic pulmonary angiography and/or radionuclide scanning (ventilation-perfusion [V/Q] scanning) for PE.39 Subsequent management of confirmed VTE occurs in two phases: (i) “active treatment” consisting of therapeutic dose anticoagulation for three months to suppress the acute episode of thrombosis; and (ii) “secondary prevention” consisting of therapeutic or prophylactic dose anticoagulation for an indefinite duration to prevent new VTE episodes unrelated to the index event.39 According to ACCP guidelines, in patients with a low risk of bleeding, a risk of recurrent VTE of >13% in the first year results in a strong recommendation and a risk of 8-13% in the first year results in a weak recommendation for indefinite anticoagulation therapy.119 Prior cohort studies have suggested that the VTE recurrence rate in SCD ranges from approximately 25-40%,8,16 although anticoagulation adherence data are generally lacking in these studies. Nonetheless, the high VTE recurrence rate in SCD patients appears to justify indefinite anticoagulation as an efficacious secondary prevention strategy, especially in patients with unprovoked or recurrent provoked thrombosis.121 Even SCD patients with less severe disease have high VTE recurrence rates (18% at 5 years) with no difference according to whether the incident event occurred within or >90 days after hospitalization,6,8 suggesting that most patients are likely to benefit from secondary prevention. However, caution may be warranted, as clinical and plashaematologica | 2020; 105(10)

ma biomarkers that reliably predict VTE recurrence in SCD patients have not been identified to guide decision making in this subgroup, unlike in the general population with unprovoked VTE.120 When considering indefinite anticoagulation therapy, the perceived risk/benefit ratio influences both patients and physicians alike. Population-based studies reveal a non-linear bleeding risk associated with indefinite anticoagulation, which increases with age, disease comorbidity, polypharmacy, and renal insufficiency.28 In addition, a high incidence of bleeding, particularly gastrointestinal bleeding, has been noted in SCD patients exposed to anticoagulation.122 Thus, evaluating common risk factors, such as prior bleeding episodes, severe anemia, thrombocytopenia, renal and hepatic failure, use of antiplatelet and NSAID, can inform decisions about both anticoagulant choice and duration of therapy in SCD. Moreover, assessing for central nervous system (CNS) vasculopathy (Moya Moya disease and aneurysmal dilatation of cranial vessels), which can increase the risk of intracranial hemorrhage even in the absence of anticoagulation, is important. The primary goal of anticoagulation is to limit and resolve thrombosis with minimal perturbation of hemostasis. Among all the available anticoagulants, the DOAC come closest to achieving this goal, although the risk of bleeding is not completely eliminated. For instance, in patients with VTE and no active cancer, a meta-analysis revealed that DOAC were at least as effective as warfarin but reduced the risk of major bleeding only by 40% and did not completely eliminate it.123 Prospective evidence from a meta-analysis of two randomized trials of DOAC in cancer-associated VTE reported 6-month outcomes of improved efficacy with DOAC compared to dalteparin (relative risk [RR] of recurrent VTE: 0.65; 95% confidence interval [CI]: 0.42-1.01), but the risk of bleeding was greatly increased (major bleeding RR: 1.74; 95%CI: 1.05-2.88 and non-major bleeding RR: 2.31; 95%CI: 0.85-6.28).124 In patients with SCD-associated VTE anticoagulation associated bleeding is particularly evident. Retrospective analysis of anticoagulation for SCD-associated VTE reveals a cumulative bleeding incidence of 4.9% (95%CI: 3.5-6.4%) at 6 months and 7.9% (95%CI: 6.2-9.8%) at 1 year.8 In patients with severe SCD, the bleeding risk was greatest (HR: 1.61; 95%CI: 1.11-2.35).8 Similarly, smaller retrospective studies indicate that all anticoagulants have an increased bleeding risk, with DOAC having the least risk.23,125,126 In spite of these advantages, the exact role for DOAC in the management of VTE in SCD patients is not clear.

Perspectives Dysregulation of inflammatory and coagulation pathways, both during the steady state and during VOC, are an important hallmark of the sickle prothrombotic state. They appear to be directly associated with the development of thrombotic complications, such as VTE, which is known to increase SCD morbidity and mortality. Given the important role of coagulation in SCD, it is valid to expect that sickle specific therapies seeking to reduce inflammation and coagulation biomarkers would have an effect on lowering thrombosis incidence. Although some sickle specific therapies (HU and transfusion) are associated with reduced arterial thrombosis risk,13 surprisingly few studies have evaluated the effect of these therapies 2375

A.S. Shet et al.

on venous thrombosis. HU use was associated with reduced biochemical indicators of coagulation activation in SCD patients127 when compared with non-users, and led to lowered cfDNA levels.79 In myeloproliferative neoplasms, prospective studies of HU have shown a reduction in thrombosis biomarkers and thrombosis risk,128 but the lack of similar studies in SCD makes this area a research priority. For instance, targeting P-selectin can diminish heterotypic cellular adhesive interactions (see above) and may prevent vascular endothelial cell activation and injury. Considering its mild adverse event profile,81,129 crizanlizumab could possibly lower the incidence of thrombosis in SCD without compromising hemostasis. Given the diverse pathophysiological processes involved in SCD, it is important to narrow down key thrombo-inflammatory pathways involved in VTE pathobiology and design interventions specifically targeting venous thrombosis. From this viewpoint, several trials evaluating the effects on anticoagulants on modulating the vascular pathobiology of SCD are either ongoing or completed (Table 2). However, concerns regarding study design and implementation limit the interpretability of several of these studies, emphasizing the challenges faced by clinical investigation in this field. Nonetheless, implementation of a double-blinded, randomized, placebo controlled trial of prasugrel in children with SCD130 demonstrates feasibility of rigorous scientific experimentation in this population, and provides hope for future efficacy studies of thrombosis endpoints. Because traditional anticoagulation for VTE in the general population and in SCD patients is associated with increased bleeding, the development of safer anticoagulant treatments is of considerable importance. Besides, as indicated above, even DOAC fail to suppress TF-mediated inflammation or prevent chronic organ dysfunction in SCD,41 suggesting the need for novel therapies. Targeting coagulation factors involved in SCD pathobiology (specifically TF and contact pathway components, e.g.,

FXII and FXI) would appear to have a lowered bleeding risk, especially those that are not involved in physiological hemostasis, i.e., contact pathway factors.131 For example, anti-XI therapy is both safe and efficacious for VTE prevention in the general population132 but this approach is untested in SCD. Moreover, as TF plays an important role in physiological hemostasis, therapeutic agents that inhibit TF-mediated inflammation while sparing TF-procoagulant activity are worth developing. Finally, drugs that target post-translational mechanisms regulating TF procoagulant activity, e.g., annexins or thiol-disulfide exchange inhibitors,92 could prevent pathological thrombosis. Selecting drugs with a demonstrably lower propensity for bleeding is another approach to maximize anti-thrombotic efficacy in SCD without altering hemostasis. Aspirin, with its low bleeding risk, antithrombotic efficacy for secondary prevention of VTE,133 and proven safety in SCD may have an important thromboprophylaxis role. Statins reduce markers of hypercoagulability in subjects with unprovoked VTE after cessation of anticoagulation treatment,134 reduce abnormal pulmonary vascular TF expression in SCD mice,67 and are generally safe in SCD patients,135 providing a rationale for their further investigation in VTE thromboprophylaxis. Canakinumab, an IL-1Î˛ antagonist, is associated with a reduction in all-cause mortality from atherothrombotic coronary artery disease, largely via its anti-inflammatory effects.136 Elevated IL-1Î˛ levels and dysregulated inflammatory pathways in SCD patients, along with the relative safety of canakinumab in children with SCD (clinicaltrials.gov identifier: NCT02961218), provide a compelling rationale for testing this agent. Evaluating these therapies in the setting of controlled clinical trials would demonstrate their potential role for secondary VTE prevention in SCD. In spite of the advances in treatment and prevention, venous thrombosis profoundly impacts chronic organ dysfunction and mortality in patients with SCD. Major scientific advances have furthered our understanding of

Table 2. Recently conducted or ongoing trials of anticoagulant therapies in sickle cell disease.

Trial identifier

Study design

NCT01419977

Phase II randomized parallel assignment N=34

NCT01036802

Phase II randomized N=20

NCT02580773

Phase III randomized parallel assignment N=200 Double blind, parallel group, N=25*

NCT02179177

NCT02072668

NCT02098993

Phase II randomized cross over with 2 weeks wash out period N=14 Phase II randomized pilot feasibility study N=7**

Intervention

Primary outcome

Status

Dalteparin 5,000 IU SC OD x 7 days vs. placebo Warfarin x 12 months vs. placebo

Change in value between D1 and D3 of D-dimer, visual analog pain score and thrombin generation tests

Completed 2015 Results awaited

Pulmonary artery systolic pressure measurements on doppler echocardiography

Tinzaparin 175 IU/kg/day x 7 days vs. 4,500 IU/kg/day Apixiban 2.5 mg BID x 6 months vs. placebo Rivaroxaban 20 mg OD x 4 weeks vs. placebo Unfractionated heparin therapeutic dose vs. standard care

Time to ACS resolution and number of major bleeding episodes

Terminated 2016, due to poor enrollment Recruiting

Effects on hospitalization days, daily pain scores, and pain crisis frequency

Terminated 2019, due to lack of funds

Biomarkers of inflammation, endothelial activation and thrombin generation; microvascular flow

Completed 2019 Results awaited

Duration of hospitalization for acute chest syndrome

Terminated 2019, due to poor enrollment

SC: subcutaneous; OD: once a day; D: day; ACS: acute chest syndrome. *N=16 recruited at study termination. **Actual enrollment prior to termination.

2376

haematologica | 2020; 105(10)

Prothrombotic state in sickle cell disease

the vascular pathobiology of SCD and led to the development of novel therapeutic agents optimal for clinical investigation. Conducting effective preventative and treatment studies to reduce VTE in SCD and establishing the scientific evidence base to guide appropriate management has therefore become a research priority. The design of these studies should include endpoints that

References 1. Tisdale JF, Thein SL, Eaton WA. Treating sickle cell anemia. Science. 2020;367 (6483):1198-1199. 2. Benz EJ Jr, Mondoro TH, Gibbons GH. Accelerating the science of SCD therapies-is a cure possible? JAMA. 2019 Aug 8. [Epub ahead of print] 3. Collins FS. Curing HIV and sickle cell falls short if the most vulnerable populations are left out. Fortune. New York: Fortune Media IP Limited; 2020. 4. Adams RJ, McKie VC, Hsu L, et al. Prevention of a first stroke by transfusions in children with sickle cell anemia and abnormal results on transcranial Doppler ultrasonography. N Engl J Med. 1998;339(1):5-11. 5. Ohene-Frempong K, Weiner SJ, Sleeper LA, et al. Cerebrovascular accidents in sickle cell disease: rates and risk factors. Blood. 1998;91(1):288-294. 6. Naik RP, Streiff MB, Haywood C Jr, et al. Venous thromboembolism incidence in the Cooperative Study of Sickle Cell Disease. J Thromb Haemost. 2014;12(12):2010-2016. 7. Brunson A, Lei A, Rosenberg AS, et al. Increased incidence of VTE in sickle cell disease patients: risk factors, recurrence and impact on mortality. Br J Haematol. 2017; 178(2):319-326. 8. Brunson A, Keegan T, Mahajan A, et al. High incidence of venous thromboembolism recurrence in patients with sickle cell disease. Am J Hematol. 2019;94(8):862-870. 9. Yawn BP, Buchanan GR, Afenyi-Annan AN, et al. Management of sickle cell disease: summary of the 2014 evidence-based report by expert panel members. JAMA. 2014;312(10):1033-1048. 10. Conran N, De Paula EV. Thromboinflammatory mechanisms in sickle cell disease-challenging the hemostatic balance. Haematologica. 2020;105(10):23802390. 11. Telen MJ, Malik P, Vercellotti GM. Therapeutic strategies for sickle cell disease: towards a multi-agent approach. Nat Rev Drug Discov. 2019;18(2):139-158. 12. Matte A, Cappellini MD, Iolascon A, et al. Emerging drugs in randomized controlled trials for sickle cell disease: are we on the brink of a new era in research and treatment? Expert Opin Investig Drugs. 2020;29(1):23-31. 13. Kassim AA, Galadanci NA, Pruthi S, et al. How I treat and manage strokes in sickle cell disease. Blood. 2015;125(22):3401-3410. 14. Wun T, Brunson A. Sickle cell disease: an inherited thrombophilia. Hematology Am Soc Hematol Educ Program. 2016;2016(1): 640-647. 15. Noubouossie D, Key NS, Ataga KI. Coagulation abnormalities of sickle cell disease: relationship with clinical outcomes and the effect of disease modifying therapies.

haematologica | 2020; 105(10)

account for simultaneous effects on anti-thrombotic efficacy and harm due to bleeding. Acknowledgement This work was supported by the Intramural Research Program of the National Heart, Lung, and Blood Institute, National Institutes of Health.

Blood Rev. 2016;30(4):245-256. 16. Naik RP, Streiff MB, Haywood C Jr, et al. Venous thromboembolism in adults with sickle cell disease: a serious and under-recognized complication. Am J Med. 2013;126(5): 443-449. 17. James AH, Jamison MG, Brancazio LR, et al. Venous thromboembolism during pregnancy and the postpartum period: incidence, risk factors, and mortality. Am J Obstet Gynecol. 2006;194(5):1311-1315. 18. Seaman CD, Yabes J, Li J, et al. Venous thromboembolism in pregnant women with sickle cell disease: a retrospective database analysis. Thromb Res. 2014;134(6):12491252. 19. Folsom AR, Tang W, Roetker NS, et al. Prospective study of sickle cell trait and venous thromboembolism incidence. J Thromb Haemost. 2015;13(1):2-9. 20. Little I, Vinogradova Y, Orton E, et al. Venous thromboembolism in adults screened for sickle cell trait: a populationbased cohort study with nested case-control analysis. BMJ Open. 2017;7(3):e012665. 21. Ogunsile FJ, Naik R, Lanzkron S. Overcoming challenges of venous thromboembolism in sickle cell disease treatment. Expert Rev Hematol. 2019;12(3):173-182. 22. Kumar R, Stanek J, Creary S, et al. Prevalence and risk factors for venous thromboembolism in children with sickle cell disease: an administrative database study. Blood Adv. 2018;2(3):285-291. 23. Srisuwananukorn A, Raslan R, Zhang X, et al. Clinical, laboratory, and genetic risk factors for thrombosis in sickle cell disease. Blood Adv. 2020;4(9):1978-1986. 24. Yu TT, Nelson J, Streiff MB, et al. Risk factors for venous thromboembolism in adults with hemoglobin SC or Sbeta(+) thalassemia genotypes. Thromb Res. 2016;141:35-38. 25. Lettre G, Sankaran VG, Bezerra MA, et al. DNA polymorphisms at the BCL11A, HBS1L-MYB, and beta-globin loci associate with fetal hemoglobin levels and pain crises in sickle cell disease. Proc Natl Acad Sci U S A. 2008;105(33):11869-11874. 26. Raffield LM, Ulirsch JC, Naik RP, et al. Common alpha-globin variants modify hematologic and other clinical phenotypes in sickle cell trait and disease. PLoS Genet. 2018;14(3):e1007293. 27. Thein SL, Menzel S, Peng X, et al. Intergenic variants of HBS1L-MYB are responsible for a major quantitative trait locus on chromosome 6q23 influencing fetal hemoglobin levels in adults. Proc Natl Acad Sci U S A. 2007;104(27):11346-11351. 28. Kearon C, Akl EA. Duration of anticoagulant therapy for deep vein thrombosis and pulmonary embolism. Blood. 2014;123(12): 1794-1801. 29. Zakai NA, McClure LA, Judd SE, et al. Racial and regional differences in venous thromboembolism in the United States in 3 cohorts. Circulation. 2014;129(14):1502-

1509. 30. Zimmerman SA, Ware RE. Inherited DNA mutations contributing to thrombotic complications in patients with sickle cell disease. Am J Hematol. 1998;59(4):267-272. 31. Zimmerman SA, Howard TA, Whorton MR, et al. Thrombophilic DNA mutations as independent risk factors for stroke and avascular necrosis in sickle cell anemia. Hematology. 2001;6(5):347-353. 32. Heit JA, Armasu SM, McCauley BM, et al. Identification of unique venous thromboembolism-susceptibility variants in AfricanAmericans. Thromb Haemost. 2017;117(4): 758-768. 33. Lin F, Blake DL, Callebaut I, et al. MAN1, an inner nuclear membrane protein that shares the LEM domain with lamina-associated polypeptide 2 and emerin. J Biol Chem. 2000;275(7):4840-4847. 34. Bourgeois B, Gilquin B, Tellier-Lebegue C, et al. Inhibition of TGF-beta signaling at the nuclear envelope: characterization of interactions between MAN1, Smad2 and Smad3, and PPM1A. Sci Signal. 2013;6(280):ra49. 35. Nagai Y, Shimazu R, Ogata H, et al. Requirement for MD-1 in cell surface expression of RP105/CD180 and B-cell responsiveness to lipopolysaccharide. Blood. 2002;99(5):1699-1705. 36. Ortiz-Suarez ML, Bond PJ. The structural basis for lipid and endotoxin binding in RP105-MD-1, and consequences for regulation of host lipopolysaccharide sensitivity. Structure. 2016;24(1):200-211. 37. Reitsma PH, Rosendaal FR. Activation of innate immunity in patients with venous thrombosis: the Leiden Thrombophilia Study. J Thromb Haemost. 2004;2(4):619622. 38. Perner F, Perner C, Ernst T, et al. Roles of JAK2 in aging, inflammation, hematopoiesis and malignant transformation. Cells. 2019;8(8):854. 39. Shet AS, Wun T. How I diagnose and treat venous thromboembolism in sickle cell disease. Blood. 2018;132(17):1761-1769. 40. Ansari J, Gavins FNE. Ischemia-reperfusion injury in sickle cell disease: from basics to therapeutics. Am J Pathol. 2019;189(4):706718. 41. Sparkenbaugh E, Pawlinski R. Prothrombotic aspects of sickle cell disease. J Thromb Haemost. 2017;15(7):1307-1316. 42. Ataga KI, Moore CG, Hillery CA, et al. Coagulation activation and inflammation in sickle cell disease-associated pulmonary hypertension. Haematologica. 2008;93(1): 20-26. 43. Conran N, Belcher JD. Inflammation in sickle cell disease. Clin Hemorheol Microcirc. 2018;68(2-3):263-299. 44. Hebbel RP. Ischemia-reperfusion injury in sickle cell anemia: relationship to acute chest syndrome, endothelial dysfunction, arterial vasculopathy, and inflammatory pain. Hematol Oncol Clin North Am.

2377

A.S. Shet et al. 2014;28(2):181-198. 45. 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. 46. Zhang D, Xu C, Manwani D, et al. Neutrophils, platelets, and inflammatory pathways at the nexus of sickle cell disease pathophysiology. Blood. 2016;127(7):801809. 47. Sundd P, Gladwin MT, Novelli EM. Pathophysiology of sickle cell disease. Annu Rev Pathol. 2019;14:263-292. 48. Mendonca R, Silveira AA, Conran N. Red cell DAMPs and inflammation. Inflamm Res. 2016;65(9):665-678. 49. Lawson CA, Yan SD, Yan SF, et al. Monocytes and tissue factor promote thrombosis in a murine model of oxygen deprivation. J Clin Invest. 1997;99(7):17291738. 50. Yan SF, Zou YS, Gao Y, et al. Tissue factor transcription driven by Egr-1 is a critical mechanism of murine pulmonary fibrin deposition in hypoxia. Proc Natl Acad Sci U S A. 1998;95(14):8298-8303. 51. Houston P, Dickson MC, Ludbrook V, et al. Fluid shear stress induction of the tissue factor promoter in vitro and in vivo is mediated by Egr-1. Arterioscler Thromb Vasc Biol. 1999;19(2):281-289. 52. Solovey A, Gui L, Key NS, et al. Tissue factor expression by endothelial cells in sickle cell anemia. J Clin Invest. 1998;101(9):18991904. 53. Key NS, Slungaard A, Dandelet L, et al. Whole blood tissue factor procoagulant activity is elevated in patients with sickle cell disease. Blood. 1998;91(11):4216-4223. 54. Shet AS, Aras O, Gupta K, et al. Sickle blood contains tissue factor-positive microparticles derived from endothelial cells and monocytes. Blood. 2003;102(7):2678-2683. 55. Ragab SM, Soliman MA. Tissue factor-positive monocytes expression in children with sickle cell disease: clinical implication and relation to inflammatory and coagulation markers. Blood Coagul Fibrinolysis. 2016;27(8):862-869. 56. Vichinsky EP. Current issues with blood transfusions in sickle cell disease. Semin Hematol. 2001;38(1 Suppl 1):14-22. 57. Connes P, Alexy T, Detterich J, et al. The role of blood rheology in sickle cell disease. Blood Rev. 2016;30(2):111-118. 58. Colin Y, Le Van Kim C, El Nemer W. Red cell adhesion in human diseases. Curr Opin Hematol. 2014;21(3):186-192. 59. Goel MS, Diamond SL. Adhesion of normal erythrocytes at depressed venous shear rates to activated neutrophils, activated platelets, and fibrin polymerized from plasma. Blood. 2002;100(10):3797-3803. 60. Setty BN, Rao AK, Stuart MJ. Thrombophilia in sickle cell disease: the red cell connection. Blood. 2001;98(12):32283233. 61. van Beers EJ, Schaap MC, Berckmans RJ, et al. Circulating erythrocyte-derived microparticles are associated with coagulation activation in sickle cell disease. Haematologica. 2009;94(11):1513-1519. 62. Aleman MM, Byrnes JR, Wang JG, et al. Factor XIII activity mediates red blood cell retention in venous thrombi. J Clin Invest. 2014;124(8):3590-3600. 63. Byrnes JR, Wolberg AS. Red blood cells in thrombosis. Blood. 2017;130(16):1795-1799. 64. Faes C, Ilich A, Sotiaux A, et al. Red blood cells modulate structure and dynamics of

2378

venous clot formation in sickle cell disease. Blood. 2019;133(23):2529-2541. 65. Stein PD, Beemath A, Meyers FA, et al. Deep venous thrombosis and pulmonary embolism in hospitalized patients with sickle cell disease. Am J Med. 2006;119(10):897. 66. Adedeji MO, Cespedes J, Allen K, et al. Pulmonary thrombotic arteriopathy in patients with sickle cell disease. Arch Pathol Lab Med. 2001;125(11):1436-1441. 67. Solovey A, Kollander R, Shet A, et al. Endothelial cell expression of tissue factor in sickle mice is augmented by hypoxia/reoxygenation and inhibited by lovastatin. Blood. 2004;104(3):840-846. 68. Vats R, Brzoska T, Bennewitz MF, et al. Platelet extracellular vesicles drive inflammasome-IL-1beta-dependent lung injury in sickle cell disease. Am J Respir Crit Care Med. 2020;201(1):33-46. 69. Wun T, Paglieroni T, Field CL, et al. Plateleterythrocyte adhesion in sickle cell disease. J Investig Med. 1999;47(3):121-127. 70. Wun T, Cordoba M, Rangaswami A, et al. Activated monocytes and platelet-monocyte aggregates in patients with sickle cell disease. Clin Lab Haematol. 2002;24(2):81-88. 71. Polanowska-Grabowska R, Wallace K, Field JJ, et al. P-selectin-mediated platelet-neutrophil aggregate formation activates neutrophils in mouse and human sickle cell disease. Arterioscler Thromb Vasc Biol. 2010;30(12):2392-2399. 72. Vogel S, Arora T, Wang X, et al. The platelet NLRP3 inflammasome is upregulated in sickle cell disease via HMGB1/TLR4 and Bruton tyrosine kinase. Blood Adv. 2018;2(20):2672-2680. 73. Novelli EM, Little-Ihrig L, Knupp HE, et al. Vascular TSP1-CD47 signaling promotes sickle cell-associated arterial vasculopathy and pulmonary hypertension in mice. Am J Physiol Lung Cell Mol Physiol. 2019;316(6):L1150-L1164. 74. Garrido VT, Sonzogni L, Mtatiro SN, et al. Association of plasma CD40L with acute chest syndrome in sickle cell anemia. Cytokine. 2017;97:104-107. 75. Anea CB, Lyon M, Lee IA, et al. Pulmonary platelet thrombi and vascular pathology in acute chest syndrome in patients with sickle cell disease. Am J Hematol. 2016;91(2):173178. 76. Allali S, Maciel TT, Hermine O, et al. Innate immune cells, major protagonists of sickle cell disease pathophysiology. Haematologica. 2020;105(2):273-283. 77. Thalin C, Hisada Y, Lundstrom S, et al. Neutrophil extracellular traps: villains and targets in arterial, venous, and cancer-associated thrombosis. Arterioscler Thromb Vasc Biol. 2019;39(9):1724-1738. 78. Noubouossie DF, Whelihan MF, Yu YB, et al. In vitro activation of coagulation by human neutrophil DNA and histone proteins but not neutrophil extracellular traps. Blood. 2017;129(8):1021-1029. 79. Ulug P, Vasavda N, Kumar R, et al. Hydroxyurea therapy lowers circulating DNA levels in sickle cell anemia. Am J Hematol. 2008;83(9):714-716. 80. Chen G, Zhang D, Fuchs TA, et al. Hemeinduced neutrophil extracellular traps contribute to the pathogenesis of sickle cell disease. Blood. 2014;123(24):3818-3827. 81. Ataga KI, Kutlar A, Kanter J, et al. Crizanlizumab for the prevention of pain crises in sickle cell disease. N Engl J Med. 2017;376(5):429-439. 82. Kalra VK, Zhang S, Malik P, et al. Placenta growth factor mediated gene regulation in

sickle cell disease. Blood Rev. 2018;32(1):6170. 83. Belcher JD, Marker PH, Weber JP, et al. Activated monocytes in sickle cell disease: potential role in the activation of vascular endothelium and vaso-occlusion. Blood. 2000;96(7):2451-2459. 84. Mackman N. Regulation of the tissue factor gene. Thromb Haemost. 1997;78(1):747-754. 85. Liu Y, Zhong H, Bao W, et al. Patrolling monocytes scavenge endothelial-adherent sickle RBCs: a novel mechanism of inhibition of vaso-occlusion in SCD. Blood. 2019;134(7):579-590. 86. Furie B, Furie BC. Mechanisms of thrombus formation. N Engl J Med. 2008;359(9):938949. 87. Grover SP, Mackman N. Tissue factor: an essential mediator of hemostasis and trigger of thrombosis. Arterioscler Thromb Vasc Biol. 2018;38(4):709-725. 88. Witkowski M, Landmesser U, Rauch U. Tissue factor as a link between inflammation and coagulation. Trends Cardiovasc Med. 2016;26(4):297-303. 89. Giesen PL, Rauch U, Bohrmann B, et al. Blood-borne tissue factor: another view of thrombosis. Proc Natl Acad Sci U S A. 1999;96(5):2311-2315. 90. Solovey A, Somani A, Belcher JD, et al. A monocyte-TNF-endothelial activation axis in sickle transgenic mice: therapeutic benefit from TNF blockade. Am J Hematol. 2017;92(11):1119-1130. 91. Esmon CT. Role of coagulation inhibitors in inflammation. Thromb Haemost. 2001;86 (1):51-56. 92. 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. 93. Arumugam PI, Mullins ES, Shanmukhappa SK, et al. Genetic diminution of circulating prothrombin ameliorates multiorgan pathologies in sickle cell disease mice. Blood. 2015;126(15):1844-1855. 94. Nasimuzzaman M, Malik P. Role of the coagulation system in the pathogenesis of sickle cell disease. Blood Adv. 2019;3(20): 3170-3180. 95. Leslie J, Langler D, Serjeant GR, et al. Coagulation changes during the steady state in homozygous sickle-cell disease in Jamaica. Br J Haematol. 1975;30(2):159-166. 96. Francis RB. Platelets, coagulation, and fibrinolysis in sickle cell disease: their possible role in vascular occlusion. Blood Coagul Fibrinolysis. 1991;2(2):341-353. 97. Hagger D, Wolff S, Owen J, et al. Changes in coagulation and fibrinolysis in patients with sickle cell disease compared with healthy black controls. Blood Coagul Fibrinolysis. 1995;6(2):93-99. 98. Noubouossie DF, LĂŞ PQ, Corazza F, et al. Thrombin generation reveals high procoagulant potential in the plasma of sickle cell disease children. Am J Hematol. 2012;87 (2):145-149. 99. Noubouossie DF, Henderson MW, Mooberry M, et al. Red blood cell microvesicles activate the contact system, leading to factor IX activation via 2 independent pathways. Blood. 2020;135(10):755-765. 100. Whelihan MF, Lim MY, Mooberry MJ, et al. Thrombin generation and cell-dependent hypercoagulability in sickle cell disease. J Thromb Haemost. 2016;14(10):1941-1952. 101. Amin C, Adam S, Mooberry MJ, et al. Coagulation activation in sickle cell trait: an exploratory study. Br J Haematol. 2015;171

haematologica | 2020; 105(10)

Prothrombotic state in sickle cell disease (4):638-646. 102. Stark K, Philippi V, Stockhausen S, et al. Disulfide HMGB1 derived from platelets coordinates venous thrombosis in mice. Blood. 2016;128(20):2435-2449. 103. Xu H, Wandersee NJ, Guo Y, et al. Sickle cell disease increases high mobility group box 1: a novel mechanism of inflammation. Blood. 2014;124(26):3978-3981. 104. Merle NS, Grunenwald A, Rajaratnam H, et al. Intravascular hemolysis activates complement via cell-free heme and heme-loaded microvesicles. JCI Insight. 2018;3(12): e96910. 105. Camus SM, De Moraes JA, Bonnin P, et al. Circulating cell membrane microparticles transfer heme to endothelial cells and trigger vasoocclusions in sickle cell disease. Blood. 2015;125(24):3805-3814. 106. Reiter CD, Wang X, Tanus-Santos JE, et al. Cell-free hemoglobin limits nitric oxide bioavailability in sickle-cell disease. Nat Med. 2002;8(12):1383-1389. 107. Setty BN, Betal SG, Zhang J, et al. Heme induces endothelial tissue factor expression: potential role in hemostatic activation in patients with hemolytic anemia. J Thromb Haemost. 2008;6(12):2202-2209. 108. Merle NS, Paule R, Leon J, et al. P-selectin drives complement attack on endothelium during intravascular hemolysis in TLR4/heme-dependent manner. Proc Natl Acad Sci U S A. 2019;116(13):6280-6285. 109. Roumenina LT, Chadebech P, Bodivit G, et al. Complement activation in sickle cell disease: dependence on cell density, hemolysis and modulation by hydroxyurea therapy. Am J Hematol. 2020;95(5):456-464. 110. Sparkenbaugh EM, Chen C, Brzoska T, et al. Thrombin-mediated activation of PAR-1 contributes to microvascular stasis in mouse models of sickle cell disease. Blood. 2020;135(20):1783-1787. 111. Gilbert GE, Sims PJ, Wiedmer T, et al. Platelet-derived microparticles express high affinity receptors for factor VIII. J Biol Chem. 1991;266(26):17261-17268. 112. Hrachovinova I, Cambien B, HafeziMoghadam A, et al. Interaction of P-selectin and PSGL-1 generates microparticles that correct hemostasis in a mouse model of hemophilia A. Nat Med. 2003;9(8):10201025. 113. Gross PL, Furie BC, Merrill-Skoloff G, et al. Leukocyte-versus microparticle-mediated

haematologica | 2020; 105(10)

tissue factor transfer during arteriolar thrombus development. J Leukoc Biol. 2005;78(6):1318-1326. 114. Eaton WA, Bunn HF. Treating sickle cell disease by targeting HbS polymerization. Blood. 2017;129(20):2719-2726. 115. Ranque B, Menet A, Boutouyrie P, et al. Arterial stiffness impairment in sickle cell disease associated with chronic vascular complications: the multinational African CADRE study. Circulation. 2016;134(13) :923-933. 116. Belhassen L, Pelle G, Sediame S, et al. Endothelial dysfunction in patients with sickle cell disease is related to selective impairment of shear stress-mediated vasodilation. Blood. 2001;97(6):1584-1589. 117. Powars DR, Chan LS, Hiti A, et al. Outcome of sickle cell anemia: a 4-decade observational study of 1056 patients. Medicine (Baltimore). 2005;84(6):363-376. 118. Mehari A, Gladwin MT, Tian X, et al. Mortality in adults with sickle cell disease and pulmonary hypertension. JAMA. 2012;307(12):1254-1256. 119. Kearon C, Akl EA, Ornelas J, et al. Antithrombotic therapy for VTE disease: CHEST guideline and expert panel report. Chest. 2016;149(2):315-352. 120. Tritschler T, Kraaijpoel N, Le Gal G, et al. Venous thromboembolism: advances in diagnosis and treatment. JAMA. 2018;320 (15):1583-1594. 121. Liem RI, Lanzkron S, Coates TD, et al. American Society of Hematology 2019 guidelines for sickle cell disease: cardiopulmonary and kidney disease. Blood Adv. 2019;3(23):3867-3897. 122. Hariharan N, Brunson A, Mahajan A, et al. Bleeding in patients with sickle cell disease: a population-based study. Blood Adv. 2020;4(5):793-802. 123. van Es N, Coppens M, Schulman S, et al. Direct oral anticoagulants compared with vitamin K antagonists for acute venous thromboembolism: evidence from phase 3 trials. Blood. 2014;124(12):1968-1975. 124. Li A, Garcia DA, Lyman GH, et al. Direct oral anticoagulant (DOAC) versus low-molecular-weight heparin (LMWH) for treatment of cancer associated thrombosis (CAT): a systematic review and meta-analysis. Thromb Res. 2019;173:158-163. 125. Roberts MZ, Gaskill GE, Kanter-Washko J, et al. Effectiveness and safety of oral antico-

agulants in patients with sickle cell disease and venous thromboembolism: a retrospective cohort study. J Thromb Thrombolysis. 2018;45(4):512-515. 126. Patel A, Williams H, Baer MR, et al. Decreased bleeding incidence with direct oral anticoagulants compared to vitamin K antagonist and low-molecular-weight heparin in patients with sickle cell disease and venous thromboembolism. Acta Haematol. 2019:142(4):233-238. 127. Colella MP, De Paula EV, Conran N, et al. Hydroxyurea is associated with reductions in hypercoagulability markers in sickle cell anemia. J Thromb Haemost. 2012;10(9): 1967-1970. 128. Falanga A, Marchetti M. Thrombosis in myeloproliferative neoplasms. Semin Thromb Hemost. 2014;40(3):348-358. 129. Kutlar A, Kanter J, Liles DK, et al. Effect of crizanlizumab on pain crises in subgroups of patients with sickle cell disease: a SUSTAIN study analysis. Am J Hematol. 2019;94(1): 55-61. 130. Heeney MM, Hoppe CC, Abboud MR, et al. A multinational trial of prasugrel for sickle cell vaso-occlusive events. N Engl J Med. 2016;374(7):625-635. 131. Weitz JI, Chan NC. Novel antithrombotic strategies for treatment of venous thromboembolism. Blood. 2020;135(5):351-359. 132. Buller HR, Bethune C, Bhanot S, et al. Factor XI antisense oligonucleotide for prevention of venous thrombosis. N Engl J Med. 2015;372(3):232-240. 133. Brighton TA, Eikelboom JW, Mann K, et al. Low-dose aspirin for preventing recurrent venous thromboembolism. N Engl J Med. 2012;367(21):1979-1987. 134. Glynn RJ, Danielson E, Fonseca FA, et al. A randomized trial of rosuvastatin in the prevention of venous thromboembolism. N Engl J Med. 2009;360(18):1851-1861. 135. Hoppe C, Jacob E, Styles L, et al. Simvastatin reduces vaso-occlusive pain in sickle cell anaemia: a pilot efficacy trial. Br J Haematol. 2017;177(4):620-629. 136. Ridker PM, Everett BM, Thuren T, et al. Antiinflammatory therapy with canakinumab for atherosclerotic disease. N Engl J Med. 2017;377(12):1119-1131. 137. Austin H, Key NS, Benson JM, et al. Sickle cell trait and the risk of venous thromboembolism among blacks. Blood. 2007;110(3): 908-912.

2379

REVIEW ARTICLE Ferrata Storti Foundation

Thromboinflammatory mechanisms in sickle cell disease – challenging the hemostatic balance Nicola Conran and Erich V. De Paula

Hematology Center, University of Campinas, UNICAMP, Cidade Universitária, Campinas-SP, Brazil.

Haematologica 2020 Volume 105(10):2380-2390

ABSTRACT

ickle cell disease (SCD) is an inherited hemoglobinopathy that is caused by the presence of abnormal hemoglobin S (HbS) in red blood cells, leading to alterations in red cell properties and shape, as the result of HbS dexoygenation and subsequent polymerization. The pathophysiology of SCD is characterized by chronic inflammatory processes, triggered by hemolytic and vaso-occlusive events, which lead to the varied complications, organ damage and elevated mortality seen in individuals with the disease. In association with activation of the endothelium and leukocytes, hemostatic alterations and thrombotic events are welldocumented in SCD. Here, we discuss the role of inflammatory pathways in modulating coagulation and inducing platelet activation in SCD, due to tissue factor activation, adhesion molecule expression, inflammatory mediator production and the induction of innate immune responses, among other mechanisms. Thromboinflammatory pathways may play a significant role in some of the major complications of SCD, such as stroke, venous thromboembolism and possibly acute chest syndrome, besides exacerbating the chronic inflammation and cellular interactions that trigger vaso-occlusion, ischemia-reperfusion processes, and eventually organ damage.

Correspondence: NICOLA CONRAN conran@unicamp.br ERICH V. DE PAULA erich@unicamp.br Received: March 5, 2020. Accepted: April 29, 2020. Pre-published: May 21, 2020. doi:10.3324/haematol.2019.239343 ©2020 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.

2380

Introduction The sickle cell disorders are inherited hemoglobinopathies that occur with an estimated incidence of between 300,000 to 400,000 births every year worldwide.1 These disorders are caused by mutations in the HBB gene (encoding hemoglobin subunit β), resulting in an amino-acid substitution in the β-globin chain. Homozygosity of the mutated sickle hemoglobin gene incurs sickle cell anemia, while compound heterozygosity with another HBB gene mutation causes variations of sickle cell disease (SCD).2 The altered hemoglobin S (HbS) that is produced polymerizes in the red blood cell (RBC) when deoxygenated, provoking the complex pathophysiology of the disease. SCD, particularly sickle cell anemia, is characterized by a high burden of morbidity and reduced life expectancy,1 due to hemolytic anemia and vaso-occlusive processes, which result in the painful vasoocclusive episodes that can lead to the hospitalization of patients and contribute to chronic organ damage. Other complications of SCD vary widely from patient to patient, but include both acute and chronic complaints such as stroke, acute chest syndrome, retinopathy, renal disease, osteonecrosis, cardiovascular complications and leg ulcers.1

Overview of the pathophysiology and inflammatory mechanisms of sickle cell disease HbS polymerization induces RBC sickling and impairs cell deformability, altering blood rheology and the RBC membrane. Sickle RBC are dense, and tend to be more dehydrated and more adhesive to the endothelium; they also generate reactive oxygen species, release more pro-inflammatory microparticles containing heme haematologica | 2020; 105(10)

Thromboinflammation in SCD

and can activate the complement and coagulation systems.3 These fragile and rigid RBC lyse easily in the circulation, both extravascularly and intravascularly, and have a significantly shorter lifespan than that of normal RBC.2 The ensuing release of non-compartmentalized (or cellfree) hemoglobin from ruptured RBC has significant pathophysiological consequences in SCD, in which endogenous scavenging mechanisms are overwhelmed.4 Cell-free hemoglobin reacts rapidly with vascular nitric oxide (NO), disrupting vasodilation and favoring cellular activation. Furthermore, oxidative reactions with cell-free hemoglobin lead to the release of heme and free iron.5 Similarly, the ischemia-reperfusion events that result from vaso-occlusive processes in the microcirculation also activate inflammatory cells and initiate sterile inflammatory processes and oxidant generation.6 Endothelial activation, a hallmark of SCD pathophysiology, augments pro-inflammatory molecule generation, and in vitro and in vivo studies suggest that it is the adhesion and capture of activated and more adhesive red cells, leukocytes and platelets to the endothelium of the blood vessel wall that trigger vaso-occlusion.6-8 Endothelial cells and leukocytes also play major roles in generating the plethora of pro-inflammatory molecules that are upregulated in SCD, including cytokines (such as tumor necrosis factor and interleukin-1β), chemokines, growth factors, eicosanoids and peptides, all of which can further stimulate cells, and induce expression of surface adhesion molecules.9 Activation of platelets is another characteristic of SCD, in which these cells participate in the generation of inflammatory molecules and heterocellular interactions that propagate vaso-occlusive processes, as will be discussed later.10,11 Additionally, functional asplenia renders individuals with SCD (particularly children) more susceptible to bacterial infections,12 and gut permeability is reportedly augmented in individuals,13,14 leading to a further exacerbation of inflammatory processes, mediated by innate immune mechanisms generated in response to exposure to pathogens and gut microbiota-derived molecules.13 The inflammatory mechanisms of SCD have been comprehensively reviewed recently,9,15 with evidence from ex vivo and in vivo studies illustrating the complex pathophysiology of SCD which involves a vicious circle of RBC alterations, hemolytic events, vaso-occlusive processes and a chronic inflammatory state involving several molecular pathways.

Sickle cell disease as a prothrombotic state The characterization of SCD as a prothrombotic state is supported by extensive laboratory and clinical data. From a laboratory perspective, almost every compartment of hemostasis has been evaluated in patients with SCD, mainly at steady-state but also during vaso-occlusive crises, in several independent cohorts.16 Although there are some inconsistencies between results, possibly reflecting differences in study design (e.g. issues related to sample collection, processing and analysis) and patients’ heterogeneity (e.g. definition of vaso-occlusive crises), one can safely state that SCD is a condition in which the hemostatic balance is clearly tipped towards a prothrombotic state. Details about these mechanisms have been the subject of excellent recent reviews,17-19 so that in an attempt to convey a broader view of these studies, we will list this evihaematologica | 2020; 105(10)

dence according to whether studies were mainly descriptive, or whether they also provided mechanistic insights regarding this hypercoagulable state. Descriptive studies comprise those that measured markers that are normally changed in response to the activation of hemostasis, and most likely represent consequences thereof, such as studies showing that SCD patients have increased levels of thrombin-antithrombin complexes, Ddimer, prothrombin fragment 1.2, factor (F) VIII and fibrinogen.20-22 This group includes studies that showed that SCD is also associated with hypercoagulable profiles in global assays of hemostasis, such as the thrombin generation assay and thromboelastometry,23 which are relevant because they provide information on the net effect of the complex hemostatic abnormalities found in patients. The second group comprises studies that evaluated parameters which, in addition to illustrating that hemostasis is activated, could also represent mechanisms by which the prothrombotic state is either initiated and/or perpetuated. These include studies showing decreased levels of natural anticoagulants such as protein C and protein S,17,20 increased expression and/or activity of tissue factor (TF) in whole blood, monocytes and circulating endothelial cells,24-26 increased levels of von Willebrand factor (vWF) coupled with decreased levels of ADAMTS-13,27 increased numbers of TF- and phosphatidylserine-bearing microparticles,21,22,28 decreased levels of contact pathway factors (FXII, prekallikrein and high molecular weight kininogen),19 and increased markers of neutrophil extracellular trap (NET) formation.29 A non-exhaustive summary of these abnormalities is presented in Table 1. Importantly, most of these phenomena have also been

Table 1. Key hemostatic abnormalities in clinical studies of sickle cell disease.

Biomarkers of increased activation of hemostasis

↑ Thrombin antithrombin complexes ↑ Prothrombin fragment 1.2 ↑ D-dimer ↑ Direct markers of thrombin generation (thrombin generation assay) Hypercoagulable pattern in viscoelastic assays (thromboelastometry assay) ↑ Factor VIII ↑ Fibrinogen ↑ Platelet αiibβ3 integrin activation ↑ CD40L

Likely involved in the initiation and/or perpetuation of hemostasis activation

↑ Tissue factor expression and/or activation ↓ Natural anticoagulants (proteins C and S) ↑ Von Willebrand factor reactivity with relatively low ADAMTS13 ↑ Phosphatidylserine-bearing red blood cell microparticles ↑ Plasminogen activator inhibitor-1 ↑ Markers of neutrophil extracellular trap formation ↓ Contact pathway factors (factor XII, prekallikrein and high molecular weight kininogen) ↑ Platelet counts ↑ P-selectin expression 2381

N. Conran and E.V. De Paula

observed in the two most representative animal models of SCD, which have been critical to our understanding of the mechanisms of hypercoagulability in this condition. First, it was shown that ischemia-reperfusion injury, one of the key pathogenic elements of SCD, triggers endothelial TF expression,30 which was later shown to be involved mostly in the inflammatory spectrum of the pathogenesis of SCD, as we shall discuss later. The role of TF was demonstrated by a study in which inhibition of TF with anti-TF antibodies reduced both coagulation activation and inflammation, but that the former effect was not dependent on endothelial TF.31 The role of neutrophils in the formation of microvascular thrombi was also confirmed in SCD animal models,8,32 while further in vivo studies established that excess heme activates TF-mediated coagulation, and also confirmed the significance of leukocyte TF in this effect.33 The relevance of coagulation activation for the pathogenesis of SCD was nicely illustrated by demonstrations that downregulating thrombin generation by either genetic34 or pharmacological35 approaches reduced inflammation and improved clinical manifestations in SCD mice. The participation of fibrinogen in this cycle as more than a biomarker was also demonstrated by studies using SCD mice.36 Additionally, observations of clot formation in SCD mice, and in human SCD blood samples, show that the interaction of RBC with each other and with other cells inside venous thrombi may make an additional contribution to hypercoagulability in SCD.37 Finally, dense sickle cells and hemolysis are implicated in activation of the complement system, an innate immune defense cascade, in patients with SCD,38 and complement activation by hemolytic and ischemia-reperfusion mechanisms are reported to promote vaso-occlusive processes and prothrombotic responses in mice with SCD.39 From a clinical perspective, the long-described association of SCD with ischemic stroke, and the more recently characterized risk of silent cerebral infarcts and venous thromboembolism (VTE) provide definite evidence that SCD represents a prothrombotic state. Data from the Cooperative Study on Sickle Cell Disease revealed an incidence of 0.61/100 patient-years for stroke,40 and of 7.6/1,000 patient-years for VTE41 in patients with the most severe forms of SCD, yielding cumulative rates as high as 24% and 11.3% by the age of 40 and 45 years for stroke and VTE, respectively. For reasons that are yet to be determined, but that indicate a somewhat different pathogenesis compared to that of other conditions associated with VTE, only pulmonary embolism, but not deep vein thrombosis, is more frequent in SCD.41 Again, excellent reviews have been recently published on these manifestations and their relationship with coagulation abnormalities.42 Collectively, laboratory data encompassing almost every known biomarker of hemostasis, coupled with clinical and epidemiological findings provide robust evidence of the presence of a prothrombotic state in SCD. Although stroke and VTE represent the most straightforward thrombotic manifestations in SCD, other clinical manifestations of SCD, whose pathogenesis involves vaso-occlusive processes in the microcirculation. are normally associated with the prothrombotic phenotype described in these patients.18,19,43 These include acute chest syndrome, acute painful episodes, pulmonary hypertension and even chronic organ failure attributed to chronic ischemia. In fact, despite the intuitive association that can be established between these manifestations and a pro2382

thrombotic hemostatic balance, as well as the demonstration that downregulation of thrombin generation can improve these manifestations in mouse models of SCD,34,35 their association with classical markers of hemostasis activation in patients with SCD is not consistent,44-47 in what remains one of the most intriguing issues in this area of research. As we shall discuss in the next session, the application of the well-established concept that hemostasis activation cannot be dissociated from the immune response in the field of SCD15,18,43 has been instrumental to better understanding this issue, paving the way for novel therapeutic advances.

Thromboinflammation and sickle cell disease Hemostasis as part of the immune response Hemostasis and innate immunity are both involved in the immediate response to external and internal agents that threaten tissue/organ integrity, such as pathogens and trauma. Classically viewed as two independent systems, initial discussions on their inter-relationship derived mostly by teleological reasoning based on the example of the horseshoe crab, an extant organism with 450 million years of phylogenetic history in which breeches in tegument and invasion by pathogens are tackled by a single cellbased system that senses endotoxins and triggers a cascade of events that closely resembles the architecture of the hemostatic system.48 However, in the last two decades several lines of empirical evidence have emerged to confirm the intimate crosstalk between hemostasis and innate immune responses, and the terms thromboinflammation and immunothrombosis49,50 have been introduced to summarize the concept that the cellular programs driving the formation of microthrombi in a spatially- and temporallyorganized fashion can be elicited in response to pathogens and/or tissue damage, in order to facilitate pathogen removal and tissue repair. Evidence supporting the association between hemostasis and innate immunity include: (i) studies in animal models of sepsis and/or infection in which discrete elements of hemostasis are reportedly up- or down-regulated; despite some heterogeneity, a less effective hemostatic system is generally associated with increased pathogen multiplication and, in some cases, poorer outcomes (recently reviewed elsewhere51); (ii) demonstrations that critical elements of hemostasis such as TF, thrombin and activated protein C also signal through pathways that regulate the immune response, independently of their role in fibrin formation;52 (iii) examples of microorganisms whose virulence factors are proteins that disrupt local hemostatic balance, such as Yersinia pestis;53 (iv) findings that byproducts of coagulation activation, such as peptides derived from thrombin and fibrinogen, have antibacterial properties;51 (v) intravital imaging studies showing that leukocytes are not only found in arterial and venous thrombi, but that localized coagulation activation and fibrin formation also seem to direct their adhesion to the endothelium and migration into tissues;54 and more recently, (vi) the demonstration that downregulation of critical elements of the innate immune response can protect mice from experimental thrombosis, with the example of NET formation being most paradigmatic.55 Though less straightforward, studies using system biology approaches, capable of analyzing the complex interplay between several systems, haematologica | 2020; 105(10)

Thromboinflammation in SCD

have also provided indirect evidence that pathways involved in tissue repair, such as hemostasis, angiogenesis and complement, are associated with innate immune responses in inflammatory conditions such as sepsis and SCD.56 Taken together, these data support the concept that, in a disease long-recognized as a systemic inflammatory condition, such as SCD, the sustained activation of immunothrombosis underlies the consistent, yet heterogeneous, increase in biomarkers of coagulation activation that is observed in these patients.

Activation of hemostasis in inflammatory states and in sickle cell disease Understanding how hemostasis is activated as part of inflammatory responses is critical to our understanding of the pathogenesis of hypercoagulability in SCD, since the regulation of these mechanisms may be somewhat different from the classical pathways of hemostasis activation in non-inflammatory states, which consist mostly of αiibβ3 integrin-mediated platelet activation, and of the contact and activation of TF with activated FVII on cellular surfaces.52 According to recent reports, these elements can be regulated and activated by different pathways during inflammation, and new players may contribute to the activation of hemostasis. As previously mentioned, thrombin and activated protein C can also activate pro-inflammatory and pro-angiogenic pathways by signaling through protease-activated receptors (PAR).57 Of note, this pathway is thought to require much lower concentrations of thrombin than those required for overt conversion of fibrinogen into fibrin,58 consistent with a role for localized thrombin generation in the regulation of the immune response. In the context of SCD, the relevance of coagulation-independent functions of these proteins has been demonstrated in animal models in which low TF levels (~1%) in nonhematopoietic cells attenuated inflammation, but not activation of coagulation.31 Similarly, PAR-2 deficiency has been shown to attenuate lung inflammation and interleukin-6 production, without affecting classical markers of coagulation activation.35 More recently, PAR-1 deficiency or PAR-1 inhibition was found to decrease heme and lipopolysaccharide-induced vaso-occlusion.59 Another mechanism of coagulation activation and amplification by inflammation is mediated by components of NET (DNA and histones) released from neutrophils.60 The participation of NET in the pathogenesis of SCD was suggested by the association of increased biomarkers of NET formation with the severity of acute vasoocclusive crises in these patients,29 as well as by data from animal models of SCD.61 A third mechanism by which hemostasis can be activated during inflammatory states is through the intrinsic pathway involving FXII, prekallikrein, high molecular weight kininogen and FXI. Evidence that this pathway contributes to hypercoagulability led to a series of ongoing clinical trials with FXI inhibitors in patients with thrombosis.62 The role of the intrinsic pathway in coagulation activation in SCD has been demonstrated in animal models of FXII deficiency and consequently reduced thrombin generation.19 Interestingly, several molecules (e.g. DNA, polyphosphates), some of which are increased in SCD, have been shown to activate or amplify this pathway in vitro,63 although the relevant elements that regulate the intrinsic pathway in SCD patients remain to be determined. haematologica | 2020; 105(10)

More recently, two additional pathways have been added to the list of potential mediators of hemostasis activation during inflammation. The angiopoietin/Tie-2 axis has been shown to participate in the activation of hemostasis during sepsis,64 and the CLEC-2/podoplanin pathway, involved in securing hemostasis during inflammation, has also been associated with outcomes that can fit into the spectrum of immunothrombosis in both animal models65 and humans.66 Exploring their role in the context of SCD should be of interest. In fact, although it is tempting to speculate that mechanisms that drive hemostasis activation in models of inflammation, such as sepsis, might also be relevant to the pathogenesis of SCD, one should not ignore the recent evidence suggesting that, in addition to the inflammatory processes of ischemia /reperfusion that typify SCD,6 one very specific characteristic of SCD, namely sustained hemolysis, could well be central to the activation of thromboinflammatory pathways in this disease.

Hemolysis as a driver of thromboinflammation Intravascular hemolysis is now widely accepted as a major pro-inflammatory mechanism that significantly influences the pathophysiology of numerous diseases and disorders, including SCD.4 As previously mentioned, the reaction of cell-free hemoglobin, released during hemolysis, with vascular NO incurs immediate effects on endothelial and vascular function. Oxidative reactions with hydrogen peroxide and lipid peroxide also yield the Hb-Fe3+ product in the circulation and tissues,5 which readily releases its heme group, a hydrophobic molecule that has recently been designated as a major erythrocytic danger-associated molecular pattern (DAMP) that activates and amplifies inflammatory mechanisms.67 Evidence of the pro-inflammatory effects of heme continues to advance with both in vitro and in vivo studies demonstrating the role of this molecule in multiple inflammatory pathways, due to the induction of oxidant reactions, lipid peroxidation5 and activation of receptors involved in innate immune responses.68 Notably, an intricate association exists between free heme and thromboinflammation, which promotes activation of hemostasis and its components. Heme has a significant effect on the endothelium, which plays a primary role in hemostasis due to its expression of vWF, TF exposure and major thrombin generation, among other properties.69 Non-protein associated cell-free heme can directly activate the endothelium via binding to toll-like receptor (TLR)-468 or via the transfer of heme from heme- and hemoglobin-laden microparticles derived from erythrocytes.70 In addition to inducing oxidative stress, heme-mediated activation of the endothelium results in mobilization of vWF and P-selectin from the WeibelPalade bodies to the endothelial cell surface, as well as upregulation of other adhesion molecules, which may promote platelet and leukocyte recruitment to blood vessel walls.68,71 TLR-4-mediated activation of pro-inflammatory nuclear factor-κB transcription factor activity68 may also upregulate endothelial cytokine generation, as well as TF expression.72 Furthermore, heme can induce reactive oxygen species-dependent NLRP3 inflammasome assembly in endothelial cells, with consequent processing of interleukin-1β,73 a potent pro-inflammatory cytokine that 2383

N. Conran and E.V. De Paula

may amplify platelet activation and clot formation, among other effects.74 Importantly, heme exposure induces C3 deposits on glomerular endothelial cells, amplifying innate immune complement activation and local thrombotic lesions,75 while the stimulation of TLR-4-dependent endothelial P-selectin expression by experimental hemolysis has been reported to trigger complement activation via the non-covalent anchoring of C3 activation fragments in the murine liver, leading to liver damage.76 Platelet activation and aggregation are central to thrombus formation; importantly, the decrease in NO bioavailability that occurs during intravascular hemolysis, in addition to modulating vasodynamics, may repress the major inhibitory effect that NO has on the expression and activation of adhesion molecules on cell surfaces. In particular, NO inhibits the activities of the fibrinogen and collagen receptors (αiibβ3 and α2β1 integrins, respectively)77,78 on the platelet surface and, therefore, the hemolysis-induced reduction in NO may play a role in thrombus formation. Hemolysis also activates platelets due to the release of intra-erythrocytic adenosine 5'-diphosphate (ADP), via activation of P2Y1, P2Y12, and P2X1 ADP receptors on the platelet surface.79 Additionally, heme directly mediates the activation and death of platelets via lipid peroxidation and subsequent ferroptosis, an iron-dependent form of nonapoptotic cell death,80 and oxidized heme induces divalent action-dependent platelet aggregation, associated with thromboxane formation and fibrinogen binding.81 Hemolysis and products of hemolysis have long been associated with modulation of coagulation. The administration of hematin to patients with porphyria reportedly activated coagulation parameters in some patients, prolonging prothrombin and partial thromboplastin times, as well as increasing fibrinogen-degradation products, among other effects,69,82 although in vitro studies have shown the inhibition of coagulant factors and clot lysis by hematin.83 In addition to its aforementioned stimulation of TF expression by endothelial cells, in vivo administration of heme also stimulates TF production in leukocytes33 and may activate coagulation, at high concentrations, via stimulation of TF and the extrinsic pathway.33,84 In contrast, free heme binds to several sites on FVIII with high affinity and at high concentrations can inhibit its activity in vitro;85 FVIII-heme interactions can, however, be inhibited by vWF, whose release from endothelial cells is also stimulated by heme.68 In turn, multimeric vWF can be bound by extracellular hemoglobin and bilirubin (a degradation product of heme), impairing its enzymatic cleavage by the ADAMTS-13 protein, resulting in increased prothombotic activity. With regards to final clot formation, hemolysis has been suggested to exacerbate hyperfibrinolysis and facilitate uncontrolled bleeding, with red cell lysate accentuating tissue plasminogen activator-mediated fibrinolysis86 In contrast, hematin and the products of heme degradation, iron, bilirubin and carbon monoxide, are known to interact with fibrinogen, and induce fibrin formation or decrease fibrinolysis, possibly accelerating the formation of resistant clots.83,87,88 As such, heme has a clear role in triggering the initiation of coagulation, although its effects on the propagation of this system are not as well understood.69 Cell-free heme and heme-loaded microvesicles also modulate the complement system,89 which integrates closely with innate immunity and coagulation.90 This mechanism is particularly relevant given that paroxysmal nocturnal hemoglobinuria, a disease characterized by complement2384

driven chronic hemolysis, predisposes patients to thrombosis,90 among other complications. In particular, heme overactivates the alternative complement pathway by binding to the C3 molecule, and increasing its spontaneous hydrolysis; the anchoring of complement activation fragments to erythrocytes and endothelial cells may result in their destruction (and further hemolysis) or activation, respectively,69 and thromboinflammation. The classical complement pathway, in contrast, may be inhibited by heme due to heme binding to C1q and reduced C3 convertase formation and C3b deposition.69 Importantly, anti-inflammatory intravenous immunoglobulin scavenges heme and can inhibit heme-mediated activation of the complement system on the surface of human endothelial cells.91 Endogenous systems are in place to ensure that cell-free hemoglobin and heme are inactivated and eventually degraded in the organism during hemolytic processes. Haptoglobin and hemopexin are plasma proteins with very high binding affinities for hemoglobin and heme,92 respectively, rendering them relatively non-reactive and delivering them to macrophages and hepatocytes, respectively, for endocytosis and degradation of their heme moieties by the heme oxygenase (HO-1) enzyme, which is upregulated in response to the presence of heme.9 The anti-inflammatory properties of HO-1 are well documented, and HO-1 activity in SCD mice appears to offer some protection from thrombosis.93 Furthermore, HO-1, and the products of heme degradation, were recently shown to decrease clot size in a murine venous thrombosis model, while clot size is augmented in hemopexin-knockout mice.94 The equilibrium between free and protein-associated heme is vital for determining the pro-inflammatory capacity of heme and it has been suggested that, in fact, it is unlikely that equilibrium conditions that lead to inflammation and, therefore, thromboinflammation could ever occur in vivo.92 However, in diseases characterized by severe intravascular hemolysis, such as SCD, plasma haptoglobin and hemopexin levels are often depleted,95,96 meaning that the free and bound heme equilibrium may be tipped, leading to the thromboinflammatory consequences of hemolysis.

Ischemia-reperfusion injury and thromboinflammation Ischemia-reperfusion injury is a feature of multiple clinical disorders, including myocardial infarction and ischemic stroke.97 Processes of ischemia, followed by reperfusion, occur in SCD during vaso-occlusive episodes, although their preponderance in the microvasculature means that they generally have comparatively fewer acute consequences than those occurring in larger vessels; however, their recurrent nature means that these processes contribute significantly to local tissue and organ damage. The twophase character of ischemia-reperfusion injury results in the generation of danger signals, or DAMP, from damaged cells,9 and activation of components of the innate immune system, resulting both from the interruption of the tissue’s vascular supply during vaso-occlusion and from the reoxygenation of the vasculature upon resumption of blood flow.98 The generation of reactive oxygen species is a particularly damaging feature of the reoxygenation stage.98 As described, thromboinflammatory mechanisms are significant in triggering the cellular and molecular events haematologica | 2020; 105(10)

Thromboinflammation in SCD

that lead to vaso-occlusion, but ischemia and reperfusion events further exacerbate inflammatory responses including thromboinflammatory events, reflecting, once again, the cyclic nature of SCD pathophysiology. Further activation of platelet function, in association with altered endothelial hemostatic activity and TF-thrombin pathway activation are particular features of ischemiareperfusion.99,100 Mice subjected to arterial occlusion, followed by reperfusion, demonstrate platelet adhesion to the endothelium and the formation of aggregates,101 while depletion of platelets, or inactivation of their function, can significantly reduce ischemia-reperfusion in multiple organs.97 In turn, hypoxia-reoxygenation processes induce RBC sickling in mice with SCD, generating oxidative stress and the endothelial cell activation that contributes to thromoboinflammation and TF expression.98 As such, hemostatic alterations and thrombotic processes appear to be intimately associated with inflammatory mechanisms in SCD that are triggered by hemolysis and ischemiareperfusion injury. Activation of endothelial cells and leukocytes is well-documented in SCD and plays a role in eliciting alterations in coagulation pathways and the activation of platelets, due to TF release, adhesion molecule expression, inflammatory mediator production and the induction of innate immune responses (Figure 1), contributing to further inflammation and tissue damage.

Sickle cell disease thromboinflammation and platelet activation Beyond their role in thrombosis and hemostasis, platelets are increasingly recognized to be key regulators of inflammatory responses, in both sterile and infectious situations.102 The basis of platelet activation and abnormal platelet function in SCD is not clear; as mentioned, hemolysis may be a major factor in platelet activation due to ensuing NO depletion and the release of the platelet agonist, ADP, in addition to exposure of RBC phosphatidylserine and enhanced thrombin generation.21,79 Platelets also express TLR4, which can be activated by pathogen-derived lipopolysaccharide and, likely, heme.103 Platelet activation in SCD is characterized by differentiated gene expression,104 augmented adhesion molecule activity,105,106 and the increased circulation of platelet microparticles and platelet-derived proteins.107,108 Persistent platelet activation, while apparently leading to some refractory inhibition of platelet aggregation,109 plays an increasingly recognized role in SCD pathophysiology and appears to be further exacerbated during acute vaso-occlusive episodes,110 while probably representing a key mediator of inflammatory pathways. The increased expression and activity of adhesion molecules, such as αIIbβ3, P-selectin, and glycoprotein Ibα, on SCD platelets facilitate the cells’ adhesion to the endothelium and induce endothelial activation11,105,111 and also permit the formation of P-selectin-mediated heterocellular aggregates between platelets and leukocytes and/or erythrocytic cells, which play an increasingly recognized role in the initiation of vaso-occlusive processes.8,10,32,112 Additionally, platelets are major sources of inflammatory mediators and contribute to chronic inflammation in SCD by releasing inflammatory cytokines, such as platelet factor 4, interleukin-1β, LIGHT (TNFSF14), and CD40L;11,106,113 elevated haematologica | 2020; 105(10)

levels of circulating CD40L have been associated with the occurrence of acute chest syndrome in SCD. Activated platelets are also a major source of thrombospondin-1, another protein that has been related to the incidence of acute chest syndrome and vaso-occlusive episodes108 and that, interestingly, triggers shedding of RBC microparticles, causing vaso-occlusion, in mice with SCD.114 Furthermore, the platelet NLRP3 inflammasome is reportedly upregulated in SCD and may, therefore, play a major role in the processing of interleukin-1β and interleukin-18, which have important pro-inflammatory roles in the disease.115

Thrombotic manifestations of sickle cell disease in the spectrum of thromboinflammation Despite our current knowledge about the characteristics and mechanisms by which hemostasis is activated in SCD, the precise role of hemostasis in the pathogenesis of complications of the disease has not been completely clarified, as the association between classical hemostatic biomarkers and the occurrence of clinical manifestations of SCD is not straightforward,42 and results from clinical trials using antithrombotic agents have yielded heterogeneous results. On the other hand, significant clues about the relevance of hemostasis activation in SCD include data from animal models showing that reductions of prothrombin to levels normally observed in anticoagulated patients improved organ damage in SCD mice and even mortality,34 in association with attenuation of inflammation and vaso-occlusion after downregulation of PAR-1, PAR-2 and TF.35,59,116 Interestingly, in sickle cell disorders with genotypes that are generally regarded as being associated with a milder phenotype, such as HbSC disease and S/β+ thalassemia, patients have an increased risk of thromboembolic events, albeit lower than that in homozygous and S/β0 patients.117,41 Considering that hemolytic and inflammatory processes are apparently less exacerbated in the majority of this subset of patients,118,119 compared to homozygous sickle cell anemia patients, the mechanisms behind this observation require further elucidation. In a retrospective analysis of 147 patients with these two conditions, higher hemoglobin levels and a history of surgical splenectomy were independently associated with the risk of VTE,120 suggesting that increased viscosity and RBC membrane alterations might be involved.121 However, it should also be noted that relatively increased levels of parameters indicative of hemolytic activity and inflammation have been associated with greater activation of coagulation and risk of VTE in patients with HbSC disease.122,123 The concept of thromboinflammation encompasses a broad spectrum of human disorders with both thrombotic and inflammatory manifestations, and it has been proposed that the level of thrombin would be a key determinant of the relative contribution of inflammation or thrombosis to disease presentation.50 According to this model, thrombotic microangiopathies and antiphospholipid syndrome are located in that part of the spectrum in which thrombotic manifestations are predominant, while rheumatoid arthritis and atherosclerosis are located on the opposite side of the spectrum, dominated by inflammatory mechanisms. Interestingly, SCD has been proposed to be located in the middle part of this postulated spectrum,50 reflecting the difficulty in characterizing the prominence of hemostasis acti2385

N. Conran and E.V. De Paula

vation or inflammation in its pathogenesis. Additionally, the fact that infections represent a trigger for some SCD complications makes this characterization even more challenging.1 Accordingly, a recent review about the prothrombotic manifestations of SCD suggested that future clinical studies targeting coagulation activation should focus on a broader range of endpoints including end-organ damage.19 We believe that this suggestion could be further expanded in the future by attempting to classify SCD complications as parts of a similar thromboinflammatory spectrum, in which the

contributions of inflammation and hemostatic activation vary (Figure 2). Refinement of this model based on previous and future studies could allow a more precise definition of biomarkers and targets for the treatment of SCD.

Perspectives Thromboinflammatory pathways, involving the activation of hemostasis and platelets, and prompted by RBC

Figure 1. Thromboinflammatory pathways in sickle cell disease. Hemolytic and ischemia-reperfusion events (due to vaso-occlusion) drive inflammation and hemostatic alterations in sickle cell disease (SCD). Hemolysis leads to the release of cell-free hemoglobin, heme and other red blood cell contents into the vascular environment, causing nitric oxide depletion and contributing to platelet activation. Non-protein bound heme activates the endothelium (with some participation of tolllike receptor 4), induces tissue factor (TF) expression and activation in monocytes and endothelial cells, neutrophil extracellular trap formation, leukocyte and complement activation. Endothelial activation by platelets, leukocytes, ischemia-reperfusion events and heme/heme-laden microparticles results in the expression of adhesion molecules, including P-selectin and integrin αiibβ3, which recruit leukocytes and, in turn, red cells to the blood vessel walls. Activated endothelial cells, leukocytes and platelets also produce mature interleukin-1β, which further activates platelets and clot formation, in addition to having major inflammatory effects. Plateletderived proteins, such as CD40L and thrombospondin-1, activate the endothelium and may trigger red cell microparticle release. TF expression by monocytes leads to thrombin generation via the extrinsic (TF/FVIIa/FXa) pathway, while TF expression in activated endothelium signals through PAR-2 receptors (TF/FVIIa/FXa) triggering pro-inflammatory responses (interleukin-6 expression and leukocyte recruitment). Thrombin also mediates pro-inflammatory pathways through PAR-1 receptors, triggering von Willebrand factor release and P-selectin expression, which contribute to platelet adhesive mechanisms. Thrombin generation is further increased by phosphatidylserine-expressing red blood cells and microparticles, platelet polyPs and cell-free DNA derived from NET, via the intrinsic pathway of coagulation. Thrombin generation ultimately results in fibrin production and clotting, but also induces red cell adhesion to the vascular wall, in addition to inflammation through PAR-1 signaling. Oxidative stress contributes to NET formation and endothelial activation. These cellular responses are part of thromboinflammation, illustrated in the inset of figure 1. Leukocyte, platelet and red blood cell recruitment to the vascular wall, together with clotting processes, neutrophil extracellular trap components and the formation of heterocellular aggregates between platelets, leukocytes and red blood cells, with subsequent red cell trapping, results in the vaso-occlusive processes that characterize SCD. ADP: adenosine diphosphate; ICAM: intercellular adhesion molecule; IL: interleukin; NET: neutrophil extracellular trap; NO: nitric oxide; PAR: protease-activated receptor; TF: tissue factor; TLR4: toll-like receptor 4; TSP: thrombospondin; VWF: von Willebrand factor. Figure created with BioRender.com.

2386

haematologica | 2020; 105(10)

Thromboinflammation in SCD

alterations and destruction, and ensuing innate immune responses, play significant roles in some of the major complications of SCD, such as stroke, and exacerbate the chronic inflammation and cellular interactions that trigger vaso-occlusion, ischemia/reperfusion processes and eventually organ damage. As such, components of hemostasis and markers of platelet activation constitute therapeutic targets in SCD. Indeed, crizanlizumab, which was recently approved by the Food and Drug Administration for use in SCD, following clinical data from the phase II SUSTAIN trial124 that indicated that it may be effective for preventing painful vaso-occlusive crises, is an anti P-selectin therapy, which may benefit patients not just by diminishing endothelial receptivity to leukocyte adhesion, but by reducing the participation of platelets in the formation of heterocellular aggregates. The most frequently employed therapy for SCD is hydroxyurea; while hydroxyurea provides clinical benefits by increasing fetal hemoglobin production, and therefore partially inhibiting HbS polymerization, this drug is an important anti-inflammatory agent.9 In addition to reducing the incidence of hospitalization and acute pain in SCD, hydroxyurea reduces the occurrence of acute chest syndrome and may provide an alternative to chronic transfusion therapy in some subjects at risk of stroke,125,126 indicating benefits on the hemostatic and thromboinflammatory mechanisms that contribute to these complica-

tions. However, curiously, while an association of hydroxyurea therapy with reduced levels of hypercoagulability biomarkers has been reported in SCD,127 together with reductions in platelet-fibrinogen binding, leukocyte TF expression and complement activation,38,105,128 the effects of hydroxyurea on platelet-leukocyte aggregate formation129 and on the circulating concentrations of some platelet-derived pro-inflammatory molecules are less clear.10,108 128 Anti-platelet therapies, while producing promising results in pre-clinical studies have not, thus far, been successful in larger clinical trials; prasugrel hydrochloride, an inhibitor of platelet activation, was apparently well-tolerated and safe in children and adolescents with SCD, but failed to reduce vaso-occlusive complications significantly when administered for up to 24 months.130 While concerns regarding the use of anticoagulants in SCD patients have been raised, the long-term administration of rivaroxaban in SCD patients with VTE was not associated with major bleeding events,131 although the use of vitamin K antagonists and low-molecular-weight heparin in such patients has been associated with some bleeding events.132 Thus, given the aforementioned range of complications that may be triggered by thrombosis and inflammation (see Figure 1), it may be important to tailor prospective therapies to target thromboinflammatory mechanisms that could be more specific to certain complications of SCD, such as

Figure 2. Complications of sickle cell disease and thromboinflammation. Thromboinflammation, or immunothrombosis, can be viewed as a physiological response to pathogens or tissue damage in which hemostasis activation and localized activation of inflammatory pathways at the microvascular level (i.e. vascular inflammation) act in concert to facilitate pathogen removal or tissue repair. However, loss of localization or deregulated activation of these pathways underlies the pathogenesis of a multitude of immune-mediated diseases in which the relative contribution of hemostasis and/or vascular inflammation varies, determining disease presentation, as proposed by Jackson et al.50 Sickle cell disease (SCD) is a condition in which thromboinflammatory mechanisms have long been recognized as critical to its pathogenesis. However, the complex interplay between hemostasis and innate immunity activation makes it difficult to define precisely the relative contribution of each of these two processes to the pathogenesis of different complications, possibly explaining the absence of a straightforward association between classical biomarkers of hemostasis activation and the risk or the severity of some of these clinical manifestations. Accordingly, although it is likely that venous thromboembolism and avascular necrosis of the hip are complications in which hemostasis activation plays a dominant role (in orange), while pulmonary hypertension, priapism and leg ulcers can be more adequately placed in the vascular inflammation (in red) side of this thromboinflammatory spectrum, for most other complications, there is yet not enough evidence to precisely identify their localization in this cycle/spectrum, which could facilitate the identification of therapeutic targets, as well as biomarkers for each SCD complication.

haematologica | 2020; 105(10)

2387

N. Conran and E.V. De Paula

anticoagulants to avoid VTE, antiplatelet drugs for patients at risk of acute chest syndrome and anti-adhesive therapies to reduce vaso-occlusion and end-organ damage. Finally, given that thromboinflammation may be a major element in the pathogenesis of COVID-19, possibly contributing to both increased risk of VTE and to thrombosis in the lung microcirculation, pre-existing SCD thromboinflammation

References 1. Kato GJ, Piel FB, Reid CD, et al. Sickle cell disease. Nat Rev Dis Primers. 2018;4:18010. 2. Steinberg MH. Overview of sickle cell anemia pathophysiology. In: Costa FF, Conran N, eds. Sickle Cell Anemia: From Basic Science To Clinical Practice. Switzerland: Springer International, 2016:49-75. 3. Wandersee NJ, Hillery CA. Red blood cells and the vaso-occlusive process. In: Costa FF, Conran N, eds. Sickle Cell Anemia: From Basic Science To Clinical Practice. New York: Springer Int., 2016:49-74. 4. Reiter CD, Wang X, Tanus-Santos JE, et al. Cell-free hemoglobin limits nitric oxide bioavailability in sickle-cell disease. Nat Med. 2002;8(12):1383-1389. 5. Schaer DJ, Buehler PW. Cell-free hemoglobin and its scavenger proteins: new disease models leading the way to targeted therapies. Cold Spring Harb Perspect Med. 2013;3(6):a013433. 6. Hebbel RP, Belcher JD, Vercellotti GM. The multifaceted role of ischemia/reperfusion in sickle cell anemia. J Clin Invest. 2020;130 (3):1062-1072. 7. Turhan A, Weiss LA, Mohandas N, Coller BS, Frenette PS. Primary role for adherent leukocytes in sickle cell vascular occlusion: a new paradigm. Proc Natl Acad Sci U S A. 2002;99(5):3047-3051. 8. Bennewitz MF, Jimenez MA, Vats R, et al. Lung vaso-occlusion in sickle cell disease mediated by arteriolar neutrophil-platelet microemboli. JCI Insight. 2017;2(1):e89761. 9. Conran N, Belcher JD. Inflammation in sickle cell disease. Clin Hemorheol Microcirc. 2018;68(2-3):263-299. 10. Dominical VM, Samsel L, Nichols JS, et al. Prominent role of platelets in the formation of circulating neutrophil-red cell heterocellular aggregates in sickle cell anemia. Haematologica. 2014;99(11):e214-217. 11. Proenca-Ferreira R, Brugnerotto AF, Garrido VT, et al. Endothelial activation by platelets from sickle cell anemia patients. PLoS One. 2014;9(2):e89012. 12. Nickel RS, Hsu LL. Clinical manifestations of sickle cell anemia: infants and children. In: Costa FF, Conran N, eds. Sickle Cell Anemia: From Basic Science To Clinical Practice. Switzerland: Springer International, 2016:213-29. 13. Zhang D, Chen G, Manwani D, et al. Neutrophil ageing is regulated by the microbiome. Nature. 2015;525(7570):528-532. 14. Dutta D, Methe B, Amar S, Morris A, Lim SH. Intestinal injury and gut permeability in sickle cell disease. J Transl Med. 2019;17(1): 183. 15. Allali S, Maciel TT, Hermine O, de Montalembert M. Innate immune cells, major protagonists of sickle cell disease pathophysiology. Haematologica. 2020;105 (2):273-283. 16. Ataga KI, Key NS. Hypercoagulability in

2388

17.

18.

19. 20.

21.

22.

23. 24.

25.

26.

27.

28.

29.

30.

31.

32.

could represent a serious risk for complications from COVID-19 infection in this patient population.133 Funding The authors of this review receive research funding from FAPESP (Fundação de Amparo à Pesquisa do Estado de S. Paulo), grant number: 2014/00984-3.

sickle cell disease: new approaches to an old problem. Hematology Am Soc Hematol Educ Program. 2007:91-96. Faes C, Sparkenbaugh EM, Pawlinski R. Hypercoagulable state in sickle cell disease. Clin Hemorheol Microcirc. 2018;68(23):301-318. Nasimuzzaman M, Malik P. Role of the coagulation system in the pathogenesis of sickle cell disease. Blood Adv. 2019;3(20): 3170-3180. Sparkenbaugh E, Pawlinski R. Prothrombotic aspects of sickle cell disease. J Thromb Haemost. 2017;15(7):1307-1316. Nsiri B, Gritli N, Bayoudh F, Messaoud T, Fattoum S, Machghoul S. Abnormalities of coagulation and fibrinolysis in homozygous sickle cell disease. Hematol Cell Ther. 1996;38(3):279-284. Setty BN, Rao AK, Stuart MJ. Thrombophilia in sickle cell disease: the red cell connection. Blood. 2001;98(12):32283233. van Beers EJ, Schaap MC, Berckmans RJ, et al. Circulating erythrocyte-derived microparticles are associated with coagulation activation in sickle cell disease. Haematologica. 2009;94(11):1513-1519. Lim MY, Ataga KI, Key NS. Hemostatic abnormalities in sickle cell disease. Curr Opin Hematol. 2013;20(5):472-477. Key NS, Slungaard A, Dandelet L, et al. Whole blood tissue factor procoagulant activity is elevated in patients with sickle cell disease. Blood. 1998;91(11):4216-4223. Setty BN, Key NS, Rao AK, et al. Tissue factor-positive monocytes in children with sickle cell disease: correlation with biomarkers of haemolysis. Br J Haematol. 2012;157(3):370-380. Solovey A, Gui L, Key NS, Hebbel RP. Tissue factor expression by endothelial cells in sickle cell anemia. J Clin Invest. 1998;101(9): 1899-1904. Sins JWR, Schimmel M, Luken BM, et al. Dynamics of von Willebrand factor reactivity in sickle cell disease during vaso-occlusive crisis and steady state. J Thromb Haemost. 2017;15(7):1392-1402. Shet AS, Aras O, Gupta K, et al. Sickle blood contains tissue factor-positive microparticles derived from endothelial cells and monocytes. Blood. 2003;102(7):2678-2683. Schimmel M, Nur E, Biemond BJ, et al. Nucleosomes and neutrophil activation in sickle cell disease painful crisis. Haematologica. 2013;98(11):1797-1803. Solovey A, Kollander R, Shet A, et al. Endothelial cell expression of tissue factor in sickle mice is augmented by hypoxia/reoxygenation and inhibited by lovastatin. Blood. 2004;104(3):840-846. Chantrathammachart P, Mackman N, Sparkenbaugh E, et al. Tissue factor promotes activation of coagulation and inflammation in a mouse model of sickle cell disease. Blood. 2012;120(3):636-646. Polanowska-Grabowska R, Wallace K, Field

33.

34.

35.

36.

37.

38.

39. 40.

41.

42.

43.

44.

45.

46. 47.

JJ, et al. P-selectin-mediated platelet-neutrophil aggregate formation activates neutrophils in mouse and human sickle cell disease. Arterioscler Thromb Vasc Biol. 2010;30(12):2392-2399. Sparkenbaugh EM, Chantrathammachart P, Wang S, et al. Excess of heme induces tissue factor-dependent activation of coagulation in mice. Haematologica. 2015;100(3):308314. Arumugam PI, Mullins ES, Shanmukhappa SK, et al. Genetic diminution of circulating prothrombin ameliorates multiorgan pathologies in sickle cell disease mice. Blood. 2015;126(15):1844-1855. Sparkenbaugh EM, Chantrathammachart P, Mickelson J, et al. Differential contribution of FXa and thrombin to vascular inflammation in a mouse model of sickle cell disease. Blood. 2014;123(11):1747-1756. Nasimuzzaman M, Arumugam PI, Mullins ES, et al. Elimination of the fibrinogen integrin alphaMbeta2-binding motif improves renal pathology in mice with sickle cell anemia. Blood Adv. 2019;3(9):1519-1532. Faes C, Ilich A, Sotiaux A, et al. Red blood cells modulate structure and dynamics of venous clot formation in sickle cell disease. Blood. 2019;133(23):2529-2541. Roumenina LT, Chadebech P, Bodivit G, et al. Complement activation in sickle cell disease: dependence on cell density, hemolysis and modulation by hydroxyurea therapy. Am J Hematol. 2020;95(5):456-464. Vercellotti GM, Dalmasso AP, Schaid TR, Jr., et al. Critical role of C5a in sickle cell disease. Am J Hematol. 2019;94(3):327-337. Ohene-Frempong K, Weiner SJ, Sleeper LA, et al. Cerebrovascular accidents in sickle cell disease: rates and risk factors. Blood. 1998;91(1):288-294. Naik RP, Streiff MB, Haywood C Jr, Segal JB, Lanzkron S. Venous thromboembolism incidence in the Cooperative Study of Sickle Cell Disease. J Thromb Haemost. 2014;12 (12):2010-2016. Noubouossie D, Key NS, Ataga KI. Coagulation abnormalities of sickle cell disease: relationship with clinical outcomes and the effect of disease modifying therapies. Blood Rev. 2016;30(4):245-256. Sparkenbaugh E, Pawlinski R. Interplay between coagulation and vascular inflammation in sickle cell disease. Br J Haematol. 2013;162(1):3-14. Ataga KI, Moore CG, Hillery CA, et al. Coagulation activation and inflammation in sickle cell disease-associated pulmonary hypertension. Haematologica. 2008;93(1): 20-26. Ataga KI, Brittain JE, Desai P, et al. Association of coagulation activation with clinical complications in sickle cell disease. PLoS One. 2012;7(1):e29786. Green D, Scott JP. Is sickle cell crisis a thrombotic event? Am J Hematol. 1986;23(4):317321. van Beers EJ, Spronk HM, Ten Cate H, et al.

haematologica | 2020; 105(10)

Thromboinflammation in SCD

48.

49. 50.

51.

52. 53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

No association of the hypercoagulable state with sickle cell disease related pulmonary hypertension. Haematologica. 2008;93(5): e42-e44. Opal SM. Phylogenetic and functional relationships between coagulation and the innate immune response. Crit Care Med. 2000;28(9 Suppl):S77-80. Engelmann B, Massberg S. Thrombosis as an intravascular effector of innate immunity. Nat Rev Immunol. 2013;13(1):34-45. Jackson SP, Darbousset R, Schoenwaelder SM. Thromboinflammation: challenges of therapeutically targeting coagulation and other host defense mechanisms. Blood. 2019;133(9):906-918. Fiusa MM, Carvalho-Filho MA, AnnichinoBizzacchi JM, De Paula EV. Causes and consequences of coagulation activation in sepsis: an evolutionary medicine perspective. BMC Med. 2015;13:105. Versteeg HH, Heemskerk JW, Levi M, Reitsma PH. New fundamentals in hemostasis. Physiol Rev. 2013;93(1):327-358. Sodeinde OA, Subrahmanyam YV, Stark K, Quan T, Bao Y, Goguen JD. A surface protease and the invasive character of plague. Science. 1992;258(5084):1004-1007. Gaertner F, Massberg S. Blood coagulation in immunothrombosis - at the frontline of intravascular immunity. Semin Immunol. 2016;28(6):561-569. Fuchs TA, Brill A, Duerschmied D, et al. Extracellular DNA traps promote thrombosis. Proc Natl Acad Sci U S A. 2010;107 (36):15880-15885. Hounkpe BW, Fiusa MM, Colella MP, et al. Role of innate immunity-triggered pathways in the pathogenesis of sickle cell disease: a meta-analysis of gene expression studies. Sci Rep. 2015;5:17822. Posma JJ, Grover SP, Hisada Y, et al. Roles of coagulation proteases and PARs (proteaseactivated receptors) in mouse models of inflammatory diseases. Arterioscler Thromb Vasc Biol. 2019;39(1):13-24. Camerer E, Huang W, Coughlin SR. Tissue factor- and factor X-dependent activation of protease-activated receptor 2 by factor VIIa. Proc Natl Acad Sci U S A. 2000;97(10):52555260. Sparkenbaugh EM, Chen C, Brzoska T, et al. Thrombin-mediated activation of PAR-1 contributes to microvascular stasis in mouse models of sickle cell disease. Blood. 2020 Jan 23. [Epub ahead of print]. Noubouossie DF, Whelihan MF, Yu YB, et al. In vitro activation of coagulation by human neutrophil DNA and histone proteins but not neutrophil extracellular traps. Blood. 2017;129(8):1021-1029. Chen G, Zhang D, Fuchs TA, Manwani D, Wagner DD, Frenette PS. Heme-induced neutrophil extracellular traps contribute to the pathogenesis of sickle cell disease. Blood. 2014;123(24):3818-3827. Nickel KF, Long AT, Fuchs TA, Butler LM, Renne T. Factor XII as a therapeutic target in thromboembolic and inflammatory diseases. Arterioscler Thromb Vasc Biol. 2017;37(1):13-20. Baker CJ, Smith SA, Morrissey JH. Polyphosphate in thrombosis, hemostasis, and inflammation. Res Pract Thromb Haemost. 2019;3(1):18-25. Higgins SJ, De Ceunynck K, Kellum JA, et al. Tie2 protects the vasculature against thrombus formation in systemic inflammation. J Clin Invest. 2018;128(4):1471-1484. Hitchcock JR, Cook CN, Bobat S, et al. Inflammation drives thrombosis after

haematologica | 2020; 105(10)

66.

67.

68.

69.

70.

71.

72.

73.

74. 75.

76.

77.

78.

79.

80.

81.

82.

Salmonella infection via CLEC-2 on platelets. J Clin Invest. 2015;125(12):44294446. Lavallee VP, Chagraoui J, MacRae T, et al. Transcriptomic landscape of acute promyelocytic leukemia reveals aberrant surface expression of the platelet aggregation agonist podoplanin. Leukemia. 2018;32(6):13491357. Kato GJ, Steinberg MH, Gladwin MT. Intravascular hemolysis and the pathophysiology of sickle cell disease. J Clin Invest. 2017;127(3):750-760. 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. Roumenina LT, Rayes J, Lacroix-Desmazes S, Dimitrov JD. Heme: modulator of plasma systems in hemolytic diseases. Trends Mol Med. 2016;22(3):200-213. Camus SM, De Moraes JA, Bonnin P, et al. Circulating cell membrane microparticles transfer heme to endothelial cells and trigger vasoocclusions in sickle cell disease. Blood. 2015;125(24):3805-3814. Wagener FA, Feldman E, de Witte T, Abraham NG. Heme induces the expression of adhesion molecules ICAM-1, VCAM-1, and E selectin in vascular endothelial cells. Proc Soc Exp Biol Med. 1997;216(3):456-463. Setty BN, Betal SG, Zhang J, Stuart MJ. Heme induces endothelial tissue factor expression: potential role in hemostatic activation in patients with hemolytic anemia. J Thromb Haemost. 2008;6(12):2202-2209. Erdei J, Toth A, Balogh E, et al. Induction of NLRP3 inflammasome activation by heme in human endothelial cells. Oxid Med Cell Longev. 2018; 2018:4310816. Bester J, Pretorius E. Effects of IL-1beta, IL-6 and IL-8 on erythrocytes, platelets and clot viscoelasticity. Sci Rep. 2016; 6:32188. May O, Merle NS, Grunenwald A, et al. Heme drives susceptibility of glomerular endothelium to complement overactivation due to inefficient upregulation of heme oxygenase-1. Front Immunol. 2018;9:3008. Merle NS, Paule R, Leon J, et al. P-selectin drives complement attack on endothelium during intravascular hemolysis in TLR4/heme-dependent manner. Proc Natl Acad Sci U S A. 2019;116(13):6280-6285. Roberts W, Riba R, Homer-Vanniasinkam S, Farndale RW, Naseem KM. Nitric oxide specifically inhibits integrin-mediated platelet adhesion and spreading on collagen. J Thromb Haemost. 2008;6(12):2175-2185. Oberprieler NG, Roberts W, Riba R, Graham AM, Homer-Vanniasinkam S, Naseem KM. cGMP-independent inhibition of integrin alphaIIbbeta3-mediated platelet adhesion and outside-in signalling by nitric oxide. FEBS Lett. 2007;581(7):1529-1534. Helms CC, Marvel M, Zhao W, et al. Mechanisms of hemolysis-associated platelet activation. J Thromb Haemost. 2013;11(12):2148-2154. 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. Neely SM, Gardner DV, Reynolds N, Green D, Ts'ao CH. Mechanism and characteristics of platelet activation by haematin. Br J Haematol. 1984;58(2):305-316. Morris DL, Dudley MD, Pearson RD. Coagulopathy associated with hematin

83.

84.

85.

86.

87.

88.

89.

90.

91.

92.

93.

94.

95.

96.

97.

98.

99.

treatment for acute intermittent porphyria. Ann Intern Med. 1981;95(6):700-701. Green D, Reynolds N, Klein J, Kohl H, Ts'ao CH. The inactivation of hemostatic factors by hematin. J Lab Clin Med. 1983;102 (3):361-369. de Souza GR, Hounkpe BW, Fiusa MML, et al. Tissue factor-dependent coagulation activation by heme: a thromboelastometry study. PLoS One. 2017;12(4):e0176505. Repesse Y, Dimitrov JD, Peyron I, et al. Heme binds to factor VIII and inhibits its interaction with activated factor IX. J Thromb Haemost. 2012;10(6):1062-1071. Moore HB, Moore EE, Gonzalez E, et al. Hemolysis exacerbates hyperfibrinolysis, whereas platelolysis shuts down fibrinolysis: evolving concepts of the spectrum of fibrinolysis in response to severe injury. Shock. 2015;43(1):39-46. Nielsen VG, Pretorius E. Iron and carbon monoxide enhance coagulation and attenuate fibrinolysis by different mechanisms. Blood Coagul Fibrinolysis. 2014;25(7):695702. Gligorijevic N, Minic S, Robajac D, Nikolic M, Cirkovic Velickovic T, Nedic O. Characterisation and the effects of bilirubin binding to human fibrinogen. Int J Biol Macromol. 2019;128:74-79. Merle NS, Grunenwald A, Rajaratnam H, et al. Intravascular hemolysis activates complement via cell-free heme and heme-loaded microvesicles. JCI Insight. 2018;3(12). Keragala CB, Draxler DF, McQuilten ZK, Medcalf RL. Haemostasis and innate immunity - a complementary relationship: a review of the intricate relationship between coagulation and complement pathways. Br J Haematol. 2018;180(6):782-798. Wiatr M, Merle NS, Boudhabhay I, et al. Anti-inflammatory activity of intravenous immunoglobulin through scavenging of heme. Mol Immunol. 2019;111:205-208. Schaer DJ, Buehler PW, Alayash AI, Belcher JD, Vercellotti GM. Hemolysis and free hemoglobin revisited: exploring hemoglobin and hemin scavengers as a novel class of therapeutic proteins. Blood. 2013;121(8): 1276-1284. Wang H, Luo W, Wang J, et al. Paradoxical protection from atherosclerosis and thrombosis in a mouse model of sickle cell disease. Br J Haematol. 2013;162(1):120-129. Nath KA, Grande JP, Belcher JD, et al. Antithrombotic effects of heme-degrading and heme-binding proteins. Am J Physiol Heart Circ Physiol. 2020;318(3):H671-H81. 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. Santiago RP, Guarda CC, Figueiredo CVB, et al. Serum haptoglobin and hemopexin levels are depleted in pediatric sickle cell disease patients. Blood Cells Mol Dis. 2018;72:34-36. Maiocchi S, Alwis I, Wu MCL, Yuan Y, Jackson SP. Thromboinflammatory functions of platelets in ischemia-reperfusion injury and its dysregulation in diabetes. Semin Thromb Hemost. 2018;44(2):102113. Hebbel RP. Ischemia-reperfusion injury in sickle cell anemia: relationship to acute chest syndrome, endothelial dysfunction, arterial vasculopathy, and inflammatory pain. Hematol Oncol Clin North Am. 2014;28(2):181-198. Schanze N, Bode C, Duerschmied D. Platelet contributions to myocardial

2389

N. Conran and E.V. De Paula ischemia/reperfusion injury. Front Immunol. 2019;10:1260. 100. Mackman N. The role of the tissue factorthrombin pathway in cardiac ischemiareperfusion injury. Semin Vasc Med. 2003;3(2):193-198. 101. Senchenkova EY, Ansari J, Becker F, et al. Novel role for the AnxA1-Fpr2/ALX signaling axis as a key regulator of platelet function to promote resolution of inflammation. Circulation. 2019;140(4):319-335. 102. Rayes J, Bourne JH, Brill A, Watson SP. The dual role of platelet-innate immune cell interactions in thrombo-inflammation. Res Pract Thromb Haemost. 2020;4(1):23-35. 103. Vallance TM, Zeuner MT, Williams HF, Widera D, Vaiyapuri S. Toll-like receptor 4 signalling and its impact on platelet function, thrombosis, and haemostasis. Mediators Inflamm. 2017;2017:9605894. 104. Liu FF, Tu TT, Zhang HF, et al. Coexpression network analysis of platelet genes in sickle cell disease. Platelets. 2019;30(8):1022-1029. 105. Proenca-Ferreira R, Franco-Penteado CF, Traina F, Saad ST, Costa FF, Conran N. Increased adhesive properties of platelets in sickle cell disease: roles for alphaIIb beta3mediated ligand binding, diminished cAMP signalling and increased phosphodiesterase 3A activity. Br J Haematol. 2010;149(2):280288. 106. Lee SP, Ataga KI, Orringer EP, Phillips DR, Parise LV. Biologically active CD40 ligand is elevated in sickle cell anemia: potential role for platelet-mediated inflammation. Arterioscler Thromb Vasc Biol. 2006;26(7): 1626-1631. 107. Wun T, Paglieroni T, Rangaswami A, et al. Platelet activation in patients with sickle cell disease. Br J Haematol. 1998;100(4):741-749. 108. Novelli EM, Kato GJ, Ragni MV, et al. Plasma thrombospondin-1 is increased during acute sickle cell vaso-occlusive events and associated with acute chest syndrome, hydroxyurea therapy, and lower hemolytic rates. Am J Hematol. 2012;87(3):326-330. 109. Mehta P, Mehta J. Abnormalities of platelet aggregation in sickle cell disease. J Pediatr. 1980;96(2):209-213. 110. Villagra J, Shiva S, Hunter LA, Machado RF, Gladwin MT, Kato GJ. Platelet activation in patients with sickle disease, hemolysis-associated pulmonary hypertension, and nitric oxide scavenging by cell-free hemoglobin. Blood. 2007;110(6):2166-2172. 111. Li J, Kim K, Jeong SY, et al. Platelet protein disulfide isomerase promotes glycoprotein Ibalpha-mediated platelet-neutrophil inter-

2390

actions under thromboinflammatory conditions. Circulation. 2019;139(10):1300-1319. 112. Kim K, Li J, Barazia A, et al. ARQ 092, an orally-available, selective AKT inhibitor, attenuates neutrophil-platelet interactions in sickle cell disease. Haematologica. 2017;102 (2):246-259. 113. Garrido VT, Proenca-Ferreira R, Dominical VM, et al. Elevated plasma levels and platelet-associated expression of the prothrombotic and pro-inflammatory protein, TNFSF14 (LIGHT), in sickle cell disease. Br J Haematol. 2012;158(6):788-797. 114. Camus SM, Gausseres B, Bonnin P, et al. Erythrocyte microparticles can induce kidney vaso-occlusions in a murine model of sickle cell disease. Blood. 2012;120(25):50505058. 115. Vogel S, Arora T, Wang X, et al. The platelet NLRP3 inflammasome is upregulated in sickle cell disease via HMGB1/TLR4 and Bruton tyrosine kinase. Blood Adv. 2018;2(20):2672-2680. 116. Sparkenbaugh EM, Chantrathammachart P, Chandarajoti K, Mackman N, Key NS, Pawlinski R. Thrombin-independent contribution of tissue factor to inflammation and cardiac hypertrophy in a mouse model of sickle cell disease. Blood. 2016;127(10):13711373. 117. Pecker LH, Schaefer BA, Luchtman-Jones L. Knowledge insufficient: the management of haemoglobin SC disease. Br J Haematol. 2017;176(4):515-526. 118. da Guarda CC, Yahouedehou S, Santiago RP, et al. Sickle cell disease: a distinction of two most frequent genotypes (HbSS and HbSC). PLoS One. 2020;15(1):e0228399. 119. Nagel RL, Fabry ME, Steinberg MH. The paradox of hemoglobin SC disease. Blood Rev. 2003;17(3):167-178. 120. Yu TT, Nelson J, Streiff MB, Lanzkron S, Naik RP. Risk factors for venous thromboembolism in adults with hemoglobin SC or Sbeta(+) thalassemia genotypes. Thromb Res. 2016;141:35-38. 121. Lionnet F, Hammoudi N, Stojanovic KS, et al. Hemoglobin sickle cell disease complications: a clinical study of 179 cases. Haematologica. 2012;97(8):1136-1141. 122. Colella MP, de Paula EV, Machado-Neto JA, et al. Elevated hypercoagulability markers in hemoglobin SC disease. Haematologica. 2015;100(4):466-471. 123. Saad STO, Gilli SO. Hemoglobin Sbeta thalassemia, SC disease, and SD disease. In: Costa FF CN, ed. Sickle Cell Anemia: From Basic Science To Clinical Practice.

Switzerland: Springer International, 2016: 319-338. 124. Ataga KI, Kutlar A, Kanter J, et al. Crizanlizumab for the prevention of pain crises in sickle cell disease. N Engl J Med. 2017;376(5):429-439. 125. Charache S, Barton FB, Moore RD, et al. Hydroxyurea and sickle cell anemia. Clinical utility of a myelosuppressive "switching" agent. The Multicenter Study of Hydroxyurea in Sickle Cell Anemia. Medicine (Baltimore). 1996;75(6):300-326. 126. Ware RE, Davis BR, Schultz WH, et al. Hydroxycarbamide versus chronic transfusion for maintenance of transcranial doppler flow velocities in children with sickle cell anaemia-TCD With Transfusions Changing to Hydroxyurea (TWiTCH): a multicentre, open-label, phase 3, non-inferiority trial. Lancet. 2016;387(10019):661-670. 127. Colella MP, De Paula EV, Conran N, et al. Hydroxyurea is associated with reductions in hypercoagulability markers in sickle cell anemia. J Thromb Haemost. 2012;10(9): 1967-1970. 128. Gumiero D, Di Gennaro L, Nicolazzi MA, Landolfi R. Hydroxyurea-mediated release of nitric oxide in myeloproliferative neoplasms patients: effects on platelet-leukocyte interaction. J Clin Pharmacol. 2015;55(10):1125-1130. 129. Trelinski J, Tybura M, Smolewski P, Robak T, Chojnowski K. The influence of low-dose aspirin and hydroxyurea on platelet-leukocyte interactions in patients with essential thrombocythemia. Blood Coagul Fibrinolysis. 2009;20(8):646-651. 130. Heeney MM, Hoppe CC, Abboud MR, et al. A multinational trial of prasugrel for sickle cell vaso-occlusive events. N Engl J Med. 2016;374(7):625-635. 131. Christen JR, Bertolino J, Jean E, et al. Use of direct oral anticoagulants in patients with sickle cell disease and venous thromboembolism: a prospective cohort study of 12 patients. Hemoglobin. 2019;43(4-5):296299. 132. Patel A, Williams H, Baer MR, Zimrin AB, Law JY. Decreased bleeding incidence with direct oral anticoagulants compared to vitamin K antagonist and low-molecular-weight heparin in patients with sickle cell disease and venous thromboembolism. Acta Haematol. 2019;142(4):233-238. 133. Connors JM, Levy JH. Thromboinflammation and the hypercoagulability of COVID-19. J Thromb Haemost. 2020 Apr 17. [Epub ahead of print]

haematologica | 2020; 105(10)

ARTICLE

Hematopoiesis

Humanized zebrafish enhance human hematopoietic stem cell survival and promote acute myeloid leukemia clonal diversity

Ferrata Storti Foundation

Vinothkumar Rajan,1 Nicole Melong,2 Wing Hing Wong,3 Benjamin King,4 R. Spencer Tong,5 Nitin Mahajan,3 Daniel Gaston,6 Troy Lund,7 David Rittenberg,8 Graham Dellaire,9 Clinton J.V. Campbell,10 Todd Druley,3 and Jason N. Berman1,2, 11*

Department of Microbiology and Immunology, Dalhousie University, Halifax, Nova Scotia, Canada; 2Department of Pediatrics, University of Ottawa, Ottawa, Ontario, Canada; 3Department of Pediatrics, Division of Hematology-Oncology, Washington University, St. Louis, MO, USA; 4Department of Ocean Sciences, Memorial University, St. John’s, Newfoundland and Labrador, Canada; 5MRC Weatherall Institute of Molecular Medicine, University of Oxford, John Radcliffe Hospital, Oxford, UK; 6Department of Pathology, Dalhousie University, Halifax, Nova Scotia, Canada; 7Department of Pediatrics, University of Minnesota, Minneapolis, MN, USA; 8Department of Obstetrics and Gynecology, IWK Health Science Center, Halifax, Nova Scotia, Canada; 9 Departments of Pathology and Biochemistry & Molecular Biology, Dalhousie University, Halifax, Nova Scotia, Canada; 10Stem Cell and Cancer Research Institute, McMaster University, Hamilton, Ontario, Canada and 11CHEO Research Institute, Ottawa, Ontario, Canada 1

Haematologica 2020 Volume 105(10):2391-2399

ABSTRACT

enograft models are invaluable tools in establishing the current paradigms of hematopoiesis and leukemogenesis. The zebrafish has emerged as a robust alternative xenograft model but, like mice, lacks specific cytokines that mimic the microenvironment found in human patients. To address this critical gap, we generated the first “humanized” zebrafish that expresses human hematopoietic-specific cytokines (granulocyte-monocyte colony-stimulating factor, stem cell factor, and stromal cell-derived factor 1α). Termed GSS fish, these zebrafish promote survival, self-renewal and multilineage differentiation of human hematopoietic stem and progenitor cells and result in enhanced proliferation and hematopoietic niche-specific homing of primary human leukemia cells. Using error-corrected RNA sequencing, we determined that patient-derived leukemias transplanted into GSS zebrafish exhibit broader clonal representation compared to transplants into control hosts. GSS zebrafish incorporating error-corrected RNA sequencing establish a new standard for zebrafish xenotransplantation which more accurately recapitulates the human context, providing a more representative cost-effective preclinical model system for evaluating personalized response-based treatment in leukemia and therapies to expand human hematopoietic stem and progenitor cells in the transplant setting.

Correspondence: JASON N. BERMAN jberman@cheo.on.ca Received: March 27, 2019. Accepted: December 5, 2019. Pre-published: December 19, 2019. doi:10.3324/haematol.2019.223040 ©2020 Ferrata Storti Foundation

Introduction The availability of xenograft models has dramatically influenced our current understanding of leukemogenesis and stem cell biology over the last decade. Patientderived xenografts provide a better microenvironmental and stromal context than any in vitro system because they maintain the clonal heterogeneity inherent in human cancers, which is of translational importance for assays that involve pharmacological interventions and responses.1,2 Current gold standard xenograft assays use small mammals, like the mouse, with a depleted immune system in models that have been refined over many years from their original derivation.3-6 However, findings from these murine xenografts may not be congruent with similar experimental results observed in human studies.7 Some human samples do not engraft in a foreign host, haematologica | 2020; 105(10)

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.

2391

V. Rajan et al.

while in other cases, following successful initial engraftment, the chimera disappears over time. Given that xenografts include both human tumor and host stroma (including immune cells), these discrepancies are accounted for, in part, by the lack of evolutionary conservation of microenvironmental signaling pathways between humans and rodents. Further, cytokines present in the microenvironment are essential for the differentiation and maintenance of individual cells but are not entirely conserved across species.8 For example, there is a lack of conservation of interleukin 3 (IL3) and granulocyte-macrophage colony stimulating factor (GM-CSF/CSF2) between humans and mice at the amino acid level, evidenced by the fact that mouse IL-3 and GMCSF do not react with their respective human receptors.9,10 Thus, to compensate for these limitations, efforts to “humanize” rodent model systems have led to the introduction of human microenvironmental factors along with human cell populations.11,12 Various efforts have been made to introduce human factors into model organisms, including the injection of recombinant proteins such as PIXY321 (a GM-CSF/IL-3 fusion protein)4 and a cost-efficient method to enable human cytokine expression using knock-in10,13 and transgenic technologies,11,14 where researchers have introduced various factors including erythropoietin and IL-3. The approach of humanizing mice has been successful to the extent that it permits enhanced engraftment and, depending on the cytokine introduced, differentiation and maintenance of specific cell lineages. For example, humanized transgenic SGM3 mice expressing human stem cell factor/KIT ligand (SCF/KITLG), GM-CSF and IL-3 showed a significant increase in the myeloid15 and mast cell compartments16 and improved engraftment efficiency of human acute myeloid leukemia (AML) cells.11 This modified murine xenograft model provides a unique advantage to enhance clonal heterogeneity and thereby enrich for more robust and meaningful responses to pharmacological interventions. However, the mouse model has significant limitations: it remains laborious, is limited to small numbers of animals, and human cells take months to engraft. As such, they are not amenable to high- or medium-throughput drug screening efforts and cannot provide results to inform patient management decisions in a clinically actionable time-frame. We previously pioneered a zebrafish larval xenograft assay to study human leukemia progression and demonstrated the feasibility of employing this platform for primary patient bone marrow-derived T-cell acute lymphoblastic leukemia (T-ALL) samples.17-19 The zebrafish xenograft platform offers several advantages, including a high degree of genetic conservation with humans at the protein level20 with the added benefit of visual tractability of human cells in an organism amenable to medium-throughput chemical screening.21,22 However, similar to mice, zebrafish express evolutionarily divergent cytokines (or lack them altogether) that are critical to the maintenance of human cell clonal heterogeneity. Previous publications have suggested that the receptors and ligands of the IL-3 subfamily that include IL-5, GM-CSF, and IL-3 are absent in zebrafish,23 and in silico analysis revealed that the critical cell migration chemokine, CXCL12/SDF1α, is conserved less than 50% at the amino acid level between humans and zebrafish. While zebrafish leukemia xenograft platforms have been successful,17,18 the survival of human hematopoietic stem 2392

and progenitor cells (HSPC) in zebrafish is uncertain. It was previously demonstrated that HSPC do not survive in zebrafish for more than 12 hours,24 while a recent study showed that they could survive only up to 13 hours postinjection,25 raising concerns whether the zebrafish host enable human HSPC survival and clonal expansion after transplantation. As such, zebrafish xenograft approaches to date share a flaw in lacking an optimal microenvironment to support the clonal evolution of human HSPC and leukemia cells, questioning the clinical transferability of findings from this model. To address this critical gap, we have generated a humanized zebrafish that expresses multiple human hematopoietic-specific cytokines. We subsequently transplanted primary human-derived HSPC and leukemia cells and performed clonal heterogeneity evaluation using error-corrected sequencing (ECS). Using these humanized zebrafish models, we reveal that transgenic fish expressing human cytokines prolong survival and differentiation of human HSPC. Furthermore, in the presence of these critical cytokines, transplanted leukemia cells exhibit hematopoietic niche homing that more accurately models the behavior of human leukemia. These results lay the foundation for a new paradigm in zebrafish xenograftbased drug discovery platforms for molecular targeting of human leukemia and the expansion of HSPC.

Methods Zebrafish studies All zebrafish studies reported were approved by the Dalhousie University Committee on Laboratory Animals (UCLA), under protocol #17-007. Briefly, the tol2 constructs were injected into singlecell stage zebrafish to make the human CXCL12/SDF1α and the human SCF/KITLG and GM-CSF/CSF2-expressing transgenic zebrafish. Fish were further crossed to make a humanized triple GSS (GM-CSF, SCF, and SDF1α) transgenic fish (see details in the Online Supplementary Methods). The casper26 strain of double pigment zebrafish mutants was used to generate all transgenic and control fish.

Human umbilical cord and bone marrow samples The use of human samples in the study was approved by the IWK Health Center Research Ethics Board (REB# 1007549 & REB# 1007549). Fresh human umbilical cord blood and human leukemia bone marrow samples were collected from patients at IWK Health Center (Halifax, Nova Scotia, Canada) after formal consent had been obtained. For leukemia samples, the buffy coat was isolated, and the white blood cells were cryopreserved directly. For samples from umbilical cord blood, post-buffy coat isolation, the samples underwent immunomagnetic enrichment to isolate lineage-depleted (lin-)HSPC.

Orthotropic xenograft experiments with primary samples Both transgenic and control (casper) larvae were treated with 10 mg/mL doxycycline hydrochloride (Sigma Aldrich) at 24 hours post-fertilization (hpf) to induce the expression of SCF/KITLG and GM-CSF/CSF2 in transgenic fish and as a control for any drug effects in casper larvae. All larvae were irradiated at 72 hpf to induce cxcl12 promoter activity and niche clearance of the organism for transplant. Human patient-derived samples were then labeled with a cytoplasmic green fluorescent dye to facilitate in vivo cell tracking, and approximately 150-250 cells were injected into the common cardinal vein. The larvae were screened immediately haematologica | 2020; 105(10)

Humanized fish enhance AML and HSPC engraftment.

following the injection to confirm that cells were present in the circulation and then moved to a 35°C incubator, a previously-established compromise between the normal developing temperature of zebrafish (28.5°C) and human cells (37°C).18,27 Detailed methods are provided in the Online Supplementary Methods.

Results

CXCL12 and GM-CSF/CSF2, we did not incorporate IL-3, which was previously used in mouse models. We developed two independent transgenic zebrafish models by coinjecting tol2 mRNA and plasmids (Figure 1A, C). One expresses human CXCL12 under the zebrafish cxcl12 promoter (Figure 1B) while the other expresses human KITLG and CSF2 under a tetracycline-inducible promoter (Figure 1D); expression was confirmed using immunoblotting (Figure 1E).

Generating “humanized” transgenic zebrafish To improve the current zebrafish platform for human leukemia and HSPC xenografts, we generated transgenic zebrafish expressing human hematopoietic cytokines. Cytokines that are poorly conserved between human and zebrafish but that have been demonstrated to be critical for normal hematopoiesis were chosen for this study. In this regard, the CXCR4 ligand, CXCL12/SDF1α, was our priority, given its functions in stem cell fate decisions such as expansion, homing, self-renewal, differentiation, control of stem cell exhaustion and protection against genotoxic stress.28-30 Both GM-CSF/CSF2 and SCF/KITLG were also determined to be essential candidates based on previous mouse experiments.11 Due to its redundant function with

Humanized zebrafish demonstrate enhanced human leukemia cell migration and proliferation We initially performed validation experiments in the CXCL12 and GM-CSF/CSF2-SCF/KITLG compound transgenic models separately. The CXCR4-CXCL12 axis is critical for cell migration and homing.31 We therefore selected migration as a mode of validation for the human CXCL12expressing transgenic zebrafish. Jurkat cells are a human TALL cell line that expresses high levels of CXCR4, the cognate CXCL12 receptor.32 Zebrafish cxcl12 promoter expression begins only at 72 hpf, so we injected Jurkat cells into the yolk sac of CXCL12-expressing and casper control larvae at 72 hpf and screened for migration at 3 days post-injection

Figure 1. Humanized transgenic zebrafish express human CXCL12, KITLG, and CSF2. Transgenic zebrafish expressing human cytokines were generated by co-injecting tol2 mRNA and plasmids (A and C). (A) A cartoon of the construct used to make the transgenic zebrafish expressing human CXCL12 (hCXCL12) along with tagBFP under the zebrafish cxcl12 promoter. (B) Representative image of a transgenic zebrafish expressing human CXCL12 in the posterior hemal arc near the tip of the tail at 3 days post-fertilization (dpf). CXCL12 expression continues to progress anteriorly through the hemal arc (representative image shows 8 dpf larvae). (C) A cartoon of the constructs used to make the tet-inducible human SCF/KITLG and GM-CSF/CSF2 expressing zebrafish. (D) Representative image of the human SCF/KITLG (hKITLG) and GM-CSF/CSF2 (hCSF2)-expressing zebrafish. Image specification: magnification = 5x, numerical aperture = 0.16. (E) Representative western blot showing expression of human GM-CSF/CSF2 (hCSF2) and SCF/KITLG (hKITLG) in transgenic zebrafish. S1 and S2 denote samples from the transgenic larvae, and C1 and C2 are samples from control casper larvae. SCF/KITLG: stem cell factor/KIT ligand; GM-CSF/CSF2: granulocyte-monocyte colony- stimulating factor/colony-stimulating factor 2.

haematologica | 2020; 105(10)

2393

V. Rajan et al.

(dpi). We categorized each injected fish into one of the following categories: "no migration," "local dissemination," (dissemination within the yolk sac), and "migration" (distant migration beyond the yolk sac). There was no migratory difference between the cells injected into control versus CXCL12 fish (top panels, Figure 2A, B). However, expression of CXCL12 is low at this time point and is restricted to the posterior hemal arc near the tail. Previous studies have shown that DNA double-stranded breaks caused by either gamma irradiation or chemical agents, such as 5-fluorouracil or etoposide, can cause an increase in CXCL12 expression.33,34 Thus, we gamma-irradiated zebrafish larvae with a sublethal dose of radiation (15 Gy) 2 hours before transplantation and repeated the assay. We observed a drastic increase in the number of CXCL12 larvae that exhibited human T-ALL cell migration compared to the controls (bottom panels, Figure 2A, B). From the cells that migrated out of the yolk sac, we also saw hematopoietic niche-specific homing to the caudal hematopoietic tissue (CHT) at 144 hpf and later to the kidney marrow at 216 hpf (Online Supplementary Figure S1). For the transgenic fish expressing GM-CSF/CSF2 and SCF/KITLG (GS fish), we used CMK, a human Down syndrome acute myeloid leukemia (ML-DS) cell line for validation. While CMK cells survive in culture without additional cytokines, previous experiments in our hands demonstrated drastic cell death in zebrafish xenograft assays, suggesting that the cells are susceptible to their microenvironment. It was recently demonstrated that GM-CSF/CSF2 enhances survival in Down syndrome transient abnormal myelopoiesis (TAM), suggesting a growth advantage for ML-DS under GM-CSF/CSF2-rich conditions.35 When CMK cells were injected into GS larvae, xenografts demonstrated increased cell proliferation at 3 dpi compared to casper controls (Figure 2C, D). Strikingly, this was preceded by a sudden decrease in the number of CMK cells in both control and transgenic larvae at 2 dpi (Figure 2D). These cells were injected into the yolk sac, an acellular environment, which may have resulted in delayed proliferation due to the restricted access of injected cells to circulating human cytokines. Thus, moving forward, all xenografts were performed by injection directly into the bloodstream of larval zebrafish, which is likely more relevant to adult human hematopoiesis. This strategy is in keeping with a recent emphasis on mechanochemical mechanisms, such as blood flow, leading to blood stem cell regulation.36

Humanized zebrafish larvae show improved response to drug administration compared to controls Following our previous observation, we evaluated whether xenotransplantation of human cells into the circulation of GSS fish improved proliferation compared to that following yolk-sac injection. We injected human ML-DS CMK cells into both the yolk sac and the circulation of the larvae and observed a trend to increased proliferation of cells injected into the circulation compared to that of cells injected into the yolk sac (Online Supplementary Figure S2A). We then wanted to evaluate the fitness of our model in a preclinical drug-testing scenario. The ML-DS CMK cells were established from a 10-year old patient who responded to cytarabine. We wanted to determine whether this response was conserved in the context of GSS larvae. We injected CMK cells into the circulation of both GSS and casper larvae. The larvae were divided into two groups, and one group was treated with 1 mM cytarabine 1 day-post 2394

injection. While there was no significant difference between the cytarabine administered and untreated casper groups (P=0.94), the GSS larvae treated with cytarabine did show a significant decrease in the number of cells compared to the untreated control (P=0.005) (Online Supplementary Figure S2B).

GSS transgenic larvae show increased mortality compared to controls when transplanted with primary acute myeloid leukemia cells Following these validation experiments, both transgenic zebrafish lines were crossed to create a GSS (GMCSF/CSF2, SCF/KITLG and SDF1Îą/CXCL12) triple transgenic fish. We wanted to compare engraftment and expansion of primary patient-derived leukemias in the GSS larvae to controls. These xenografts were performed orthotopically by injecting primary AML cells into the circulation (common cardinal vein) at 72 hpf. We xenografted four distinct pediatric patient-derived AML samples: a CBL exon 8 deletion with KMT2A-MLLT3 (MLL-AF9) fusion by karyotype (A23352); KRAS G12C point mutation (A23280), KMT2AMLLT3 fusion (AS12029811) and a ML-DS sample. Immediately post-injection, larvae were screened to select for a similar number of cells in both GSS and control groups of fish. We tracked the larvae until one of the groups reached 50% mortality and used the remaining larvae for targeted ECS) (Figure 3A). While the number of days required to reach 50% mortality varied across AML samples, the GSS fish consistently suffered greater mortality compared to control larvae (Figure 3A). This indicates increased cellular proliferation and leukemic burden, because when transplanted with human HSPC, both control and GSS larvae showed almost negligible death (Online Supplementary Figure S3). Since most leukemias are heterogeneous, they provide us with polymorphism or mutationspecific biomarkers that enable screening for clonal conservation. We prepared RNA from human AML xenografted control and GSS zebrafish and performed RNA-ECS to quantify clonal variability and conservation via the relative abundance of human leukemia-specific gene transcripts in the background of zebrafish transcripts. While some single nucleotide polymorphism variants were detected alternatively in the GSS or the control fish, overall the GSS fish retained a higher number of single nucleotide polymorphism variants representing more leukemic clones than the control fish. Mutations such as NOTCH (5094 C>T) and ALK (3375 C>A), which are silent and hence not pathogenic, were only present with a high allelic frequency (>0.5) in GSS xenografts, suggesting the elimination of some clones in the controls (Figure 3B). Altogether, the data from both the control and GSS xenografts yielded 46 high confidence nucleotide variants, of which only 23 (50%) were represented in controls compared to 42 (93.4%) variants represented in the GSS zebrafish. This finding demonstrates that the GSS zebrafish provides a superior microenvironment for survival and expansion of human AML clonal diversity.

The CXCR4-CXCL12 locus is dispensable for the migration of human leukemia cells to the caudal hematopoietic tissue but necessary for homing to the kidney marrow During validation of the CXCL12 transgenic zebrafish, we injected Jurkat cells into the yolk sac of 3 dpf larvae, and a proportion of the cells transplanted into the transgenic larvae showed migration to the CHT, a region equivalent to haematologica | 2020; 105(10)

Humanized fish enhance AML and HSPC engraftment.

the fetal liver in humans. However, when we performed primary AML xenografts and injected these cells into the circulation, cells migrated to CHT uniformly in both control and transgenic larvae. Since the kidney marrow (KM), the zebrafish equivalent of human bone marrow, plays the role of the hematopoietic niche after migration from the CHT, we were curious to determine whether there was any difference in the migratory behavior to the KM between the GSS and casper larvae. We observed a profound difference in KM homing, with primary AML samples injected into the GSS larvae showing a propensity to migrate to the KM, which was not seen in controls (Figure 3C). In CXCL12

knock-out mouse models, hematopoietic stem cells migrated from the aorta-gonad-mesonephros to the fetal liver, but fetal liver to bone marrow homing was impaired.37,38 However, previous in vitro studies showed that hematopoietic stem cells from both fetal liver and bone marrow could respond to CXCL12 stimuli in a Boyden chamber assay.39 Together, these findings highlight that while the CXCR4CXCL12 axis is critical in bone marrow homing, it may be dispensable for fetal liver homing. We wanted to see if the homing of human leukemia cells to the zebrafish CHT was dependent on CXCL12-CXCR4. To do this, we returned to T-ALL, in which we had initially seen differential homing

Figure 2. Irradiation of human CXCL12-expressing transgenic zebrafish dramatically increases cell migration, while myeloid leukemia cells exhibit enhanced proliferation in the presence of CSF2/GM-CSF and KITLG/SCF. (A) Human Jurkat cells (which highly express CXCR4, the receptor for CXCL12) were xenografted into control casper and CXCL12-expressing larvae, which were further divided into two groups: one which received 15 Gy irradiation and the other no irradiation at 72 hours postfertilization (hpf). Larvae were screened for cell migration at 144 hpf. Representative images of control and CXCL12-expressing larvae that were not irradiated uooer panel). Representative images of control and CXCL12-expressing larvae following 15 Gy gamma irradiation (lower panel). (B) Quantification of cell migration was classified into “no migration,” “local dissemination” (dissemination within the yolk sac) and “migration” (distant migration beyond the yolk sac). Results represent three independent experiments. Numbers on the bar denote the total number of larvae per classification. (C) Representative images of zebrafish injected with CMK cells, a myeloid leukemia of Down syndrome (ML-DS) cell line. (D) Cell proliferation was quantified in transgenic larvae expressing GM-CSF/CSF2 and SCF/KITLG (GS fish) and casper controls following enzymatic digestion and dissociation at 1 day post-injection (dpi) (baseline), 2 dpi and 3 dpi. The analysis included fluorescence microscopy and cell counting. At 2 dpi there was a slight decrease in cell numbers in both transgenic and control larvae. By 3 dpi there was an increase in cell numbers in GS larvae, whereas cell numbers in control larvae decreased. Data presented represents four replicates with each dot in the bar graph denoting a single replicate. SCF/KITLG: stem cell factor/KIT ligand; GM-CSF/CSF2: granulocyte-monocyte colony-stimulating factor/colony-stimulating factor 2.

haematologica | 2020; 105(10)

2395

V. Rajan et al.

using Jurkat cells, and this time employed a primary patient-derived T-ALL sample expressing very high levels of CXCR4. This sample was injected into the circulation of GSS and control larvae. The majority of the T-ALL cells in both control and GSS transgenic larvae migrated and stationed in the CHT, consistent with publications from other groups.40,41 However, in contrast to other zebrafish reports,40,41 following the addition of either a CXCR4-targeting antibody or isotype control, transplanted human T-ALL cells continued to migrate to the CHT (Online Supplementary Figure S4). These data are consistent with murine data and suggest that the CXCR4-CXCL12 axis does not contribute significantly to the migration of cells to the fetal liver.

Xenotransplanted of human hematopoietic stem cells and progenitor cells in GSS fish exhibit both enhanced self-renewal capacity and multilineage differentiation Despite the widespread success of human tumor engraftment in zebrafish, normal tissue xenografts, including hematopoietic cells, have not been reported to engraft suc-

cessfully.24 Given the enhanced proliferation observed for human leukemia samples transplanted into the GSS zebrafish, we wanted to determine if the presence of human cytokines could enhance human HSPC survival and differentiation in a zebrafish host. We collected umbilical cord blood from newly delivered infants and isolated lineage-depleted (lin-) cells, which are highly enriched for human HSPC. These cells were fluorescently labeled with a cytoplasmic dye and transplanted into the circulation of both casper control and GSS larvae at 72 hpf. Consistent with previous reports, human HSPC did not survive past 24 hours in control larvae,24 but HSPC transplanted into GSS larvae continued to survive past 48 hours after which the cells began losing the cytoplasmic dye (Figure 4A). We extracted RNA for targeted exon sequencing from both control and GSS-injected larvae between 20-24 hpi. Of the three samples sequenced, only one sample injected into the casper larva had a detectable level of the human transcripts, in contrast to the GSS larvae, for which all three HSPC xenografted samples were found to have detectable human

Figure 3. Patient-derived acute myeloid leukemia transplantation into GSS transgenic larvae increases leukemia-related disease mortality and shows increased clonal representation in comparison to control larvae. (A) Kaplan-Meier curve showing increased acute myeloid leukemia (AML)-related mortality in GSS larvae transplanted with each of four different patient-derived AML samples compared with casper control larvae transplanted with the same samples (P<0.0001). (B) Heatmap showing increased clonal representation in the GSS larvae compared to control larvae transplanted with the same sample as measured by RNA-error corrected sequencing. Different colors represent allele frequency from 0.002 (dark blue) to 1 (yellow). The white box represents an absence or allele frequency of less than 0.002. (C) Representative immunofluorescence images from sagittal zebrafish sections showing human CD33+ AML cells localized in the kidney marrow of the GSS transgenic fish. The left panel shows an overview of the fish section at 10x, and the white box highlights the region of interest. The top three right panels show images taken under normal exposure based on controls and the overexposed image shown in the middle of the lower three panels illustrates the kidney morphology. The parameters were kept constant between GSS and control sections during imaging (N=5). The yellow scale bar is equivalent to 100 mm, and the red scale bar is equivalent to 10 mm. The top panel shows images from kidney marrow of control larvae injected with human AML samples, where leukemia cells are not present, and the bottom shows GSS larvae with human CD33+ AML cells. Image specifications: 10x images: numerical aperture (NA) = 0.45; 63x images: NA = 1.4, Zoom = 1.5. GM-CSF/CSF: granulocyte-monocyte colony-stimulating factor/colony-stimulating factor; SCF/KITLG: stem cell factor/KIT ligand; CXCL12/ SDF1α:CXCL12/stromal cell derived factor 1α; GSS: GM-CSF/CSF, SCF/KITLG, CXCL12/SDF1α-expressing transgenic zebrafish.

2396

haematologica | 2020; 105(10)

Humanized fish enhance AML and HSPC engraftment.

transcripts present. To further determine if human transcripts were genuinely absent in the control fish, we performed another round of ECS-RNA library preparation and increased the starting total RNA from 100 ng to 300-500 ng (depending on RNA availability). The absence of a detectable level of human transcripts was confirmed. Upon targeted transcriptome sequencing, we found that multilineage differentiation occurred in both GSS and control larvae, but with a myeloid bias. In the lymphoid lineage, only B cells were sufficiently tractable, and transcripts of early Tcell markers, CD3E and PTCRA, were absent, suggesting an absence of T-cell differentiation (Figure 4B). In terms of selfrenewal capacity, HSPC transplanted into GSS larvae showed increased expression of both CD34 and GATA2 (Figure 4B). Specifically, there were 194 and 4,682 error-corrected transcripts of the human CD34 gene in 100 ng total RNA input for control and GSS samples, respectively. While CD34 is a bona fide HSPC marker in hematopoietic cells,

GATA2 is required for the maintenance, generation, and survival of HSPC.42 The increased expression of CD34 and GATA2 suggests that HSPC undergo self-renewal only in the cytokine-rich context found in GSS larvae, but not in controls.

Discussion While murine xenografts have provided essential insights into human leukemia pathogenesis,3,4,11 and despite even more excellent opportunities provided by ever more immunocompromised hosts, this model system continues to have limitations. Primary leukemia xenografts remain challenging in mice for many reasons, including intrinsic leukemic properties, absence or lack of bioactivity of some of the human factors found in the host microenvironment, and the presence of innate immune cells in the organism

Figure 4. Umbilical cord blood-derived hematopoietic stem and progenitor cells show engraftment, self-renewal and multilineage differentiation in GSS larvae. (A) Representative image of GSS larvae and casper control larvae transplanted with umbilical cord blood (UCB)-derived hematopoietic stem and progenitor cells (HSPC). Image shows near complete absence of HSPC in control larvae at 28 hours post-injection (hpi), whereas the HSPC continue to survive in the GSS larvae at 28 hpi and HSPC are seen until 72 hpi (N=20/genotype). Image specification: magnification =10x, numerical aperture = 0.3. White and black scale bars represent 100 μm and the yellow scale bar represents 50 mm. (B) Heatmap from RNA sequencing analysis of transplanted HSPC shows the increased expression level of self-renewal specific genes in HSPC transplanted into GSS larvae. Control and GSS larvae showed an identical expression of different lineage-specific genes. GM-CSF/CSF: granulocyte-monocyte colony-stimulating factor/colony-stimulating factor; SCF/KITLG: stem cell factor/KIT ligand; CXCL12/ SDF1α:CXCL12/stromal cell derived factor 1α; GSS: GM-CSF/CSF, SCF/KITLG, CXCL12/SDF1α−expressing transgenic zebrafish.

haematologica | 2020; 105(10)

2397

V. Rajan et al.

that eliminate transplanted cells. Moreover, the complexity of these murine xenograft experiments, including the time to engraftment (typically 3-6 weeks), renders studies challenging to conduct in a high-throughput setting, and thus not readily amenable to clinically actionable drug screening experiments. Our group has pioneered human leukemia xenografts in zebrafish larvae and shown that this approach is amenable to medium-throughput drug screening in an actionable time-frame of 1-2 weeks.17,18 However, the issue of conserved elements within the tumor microenvironment is also applicable, as illustrated by previous unsuccessful efforts to develop sustainable human HSPC zebrafish xenografts.24 To address the inherent limitations of the zebrafish microenvironment for sustaining human tumor xenografts, we created a humanized zebrafish model, the GSS fish, which expresses human CXCL12/SDF1α, SCF/KITLG, and GM-CSF/CSF2 to enhance human HSPC and patientderived leukemia engraftment to enable real-time preclinical therapeutic studies. As prior humanized mouse models informed our choice of cytokines, the GSS fish resembles the NSG-SGM3 mouse model,11,43 which expresses three cytokines with poor conservation between mouse and humans: GM-CSF, SCF, and IL-3. Zebrafish, being more evolutionarily distant, have only 20% conservation with the human CXCL12 ligand (Online Supplementary Figure S5A, B) and considerable alteration in the CXCR4 binding region in humans (Online Supplementary Figure S5C). IL-3 is a critical factor in the expansion and chemotaxis of hematopoietic cells, but hematopoietic cytokines often perform redundant functions. For its expansion and homing function, IL-3 activates Raf/MEK/ERK signaling and small GTPases such as Rac and Ras. CXCL12 follows a very similar mechanism of action and additionally controls the fate of hematopoietic stem cells by restricting differentiation and enhancing stemness.44 GM-CSF, IL-3, and IL-5 also share a common β chain that acts as a signaling subunit.45 GM-CSF also redundantly activates STAT and JAK2 pathways such as IL-3.46 So, we hypothesized that together SCF, GM-CSF, and CXCL12 would compensate for the absence of IL-3. Previous observations revealed that even though there is minor to low cross-reactivity of human GM-CSF and SCF between mouse and humans, overexpression in the NSGS mice increased the number of mouse myeloid cells at the expense of erythropoiesis.43 Normally cytokine genes are expressed transiently and at low levels. Due to the use of powerful promoters and multiple integrations of the transgene, the quantity of cytokines secreted would be high in transgenic organisms and might lead to increased stress in the animals. Using a weak expression system will help overcome this phenomenon. The effect of prolonged exposure to human GM-CSF and SCF had not previously been studied in zebrafish, providing another reason for our choice of an inducible promoter. As observed when cells from the ML-DS line (CMK) were transplanted into GSS fish compared with casper controls, we observed enhanced proliferation supported by the cytokine milieu (likely predominantly GM-CSF, to which CMK cells are known to be responsive47) and a concomitant enhanced inhibition of proliferation following treatment with cytarabine. Our results from primary AML cells regarding latency and mortality are consistent with findings from the NSGS mouse model,11 but the zebrafish model offers the benefit of medium- to high-throughput that can be applied in drug 2398

screening with survival as a readout. Furthermore, using ECS, we demonstrated that human AML transplanted into GSS larvae maintain better clonal representation than transplants into zebrafish lacking human cytokines. These findings are indicative of the GSS fish providing a more clinically representative microenvironment to that found in humans. Given the results observed with patient-derived AML transplantation into GSS zebrafish, we hypothesized that these humanized zebrafish might provide an improved host environment for engrafting human blood stem cells, which to date has been a challenge in the zebrafish field.24 As previously demonstrated, umbilical cord blood-derived HSPC did not survive in control larvae but survived until 72 hpi in the GSS larvae, marking a substantive improvement over the current limit of 13 hours reported in the literature.25 ECS performed from these two different transplant populations revealed the upregulation of human genes specific to self-renewal only in the GSS larvae. The maintenance of this key stem cell characteristic exclusively in the GSS larvae affirms the utility of the “GSS human HSPC model” as a platform for studying drugs that enhance stem cell expansion in vivo. By contrast, we saw significant levels of all transcripts associated with multilineage differentiation, except for T-cell differentiation, in both control and GSS larvae. While the myeloid bias in GSS fish might be expected due to expression of GM-CSF and SCF,11,43 an absence of lymphoid cell differentiation in control larvae may be accounted for by the natural timeline of zebrafish lymphocyte development (the thymus does not appear until 5 dpf) and the lack of endogenous lymphoid-specific cytokines at this experimental time point. With growing interest in and efficacy of T-cell-mediated cancer immunotherapy and the absence of a fully functional adaptive immune system in zebrafish larvae until 1 month of age, this model is poised for further manipulation with respect to T-cell differentiation as a future platform for the preclinical testing of novel immunotherapy approaches using human cancer and HSPC co-transplantation. In summary, through the generation of novel humanized zebrafish that express key hematopoietic cytokines, this study exploits the previously recognized imaging and higher-throughput screening advantages of the zebrafish model system to create a powerful new preclinical tool. The GSS fish, in conjunction with ECS bar-coding, can be used to screen for and also validate anti-leukemic and stem cell expanding therapeutics and contribute to the goal of providing biologically rational personalized treatment to patients. Acknowledgments The authors would like to thank David Malloy, Connor Booker, David Maley and Gretchen Wagner for zebrafish care and maintenance and Jennifer Curran for administrative support. This work was supported by a Canadian Cancer Society Research Innovation grant to JB, Kellsie’s Hope Foundation and the Eli Seth Matthews Leukemia Foundation through a grant to TD, and a Discovery Grant from the Natural Sciences and Engineering Research Council of Canada (NSERC) to GD. GD and JB are Senior Scientists of the Beatrice Hunter Cancer Research Institute (BHCRI), and VR was funded by the Cancer Research Training Program of the BHCRI, with funds provided by the Terry Fox Research Institute through the Dr. Linnea Veinotte Memorial Graduate Student Award. VR was also funded by a Nova Scotia Health Research Foundation (NSHRF) Scotia Scholar award. haematologica | 2020; 105(10)

Humanized fish enhance AML and HSPC engraftment.

References 1. Hidalgo M, Amant F, Biankin AV, et al. Patient-derived xenograft models: an emerging platform for translational cancer research. Cancer Discov. 2014;4(9):9981013. 2. Siolas D, Hannon GJ. Patient derived tumor xenografts: transforming clinical samples into mouse models. J Cancer Res. 2013;73(17):5315-5319. 3. Bonnet D, Dick JE. Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell. Nat Med. 1997;3(3):730-737. 4. Lapidot T, Pflumio F, Doedens M, Murdoch B, Williams DE, Dick JE. Cytokine stimulation of multilineage hematopoiesis from immature human cells engrafted in SCID mice. Science. 1992;255 (5048):1137-1141. 5. Shlush LI, Zandi S, Mitchell A, et al. Identification of pre-leukaemic haematopoietic stem cells in acute leukaemia. Nature. 2014;506(7488):328-333. 6. Larochelle A, Vormoor J, Hanenberg H, et al. Identification of primitive human hematopoietic cells capable of repopulating NOD/SCID mouse bone marrow: implications for gene therapy. Nat Med. 1996;2(12):1329-1337. 7. Voskoglou-Nomikos T, Pater JL, Seymour L. Clinical predictive value of the in vitro cell line, human xenograft, and mouse allograft preclinical cancer models. Clin Cancer Res. 2003;9(11):4227-4239. 8. Brocker C, Thompson D, Matsumoto A, Nebert DW, Vasiliou V. Evolutionary divergence and functions of the human interleukin (IL) gene family. Hum Genomics. 2010;5(1):30-55. 9. Manz MG. Human-hemato-lymphoid-system mice: opportunities and challenges. Immunity. 2007;26(5):537-541. 10. Willinger T, Rongvaux A, Takizawa H, et al. Human IL-3/GM-CSF knock-in mice support human alveolar macrophage development and human immune responses in the lung. Proc Natl Acad Sci USA. 2011;108(6): 2390-2395. 11. Wunderlich M, Chou FS, Link KA, et al. AML xenograft efficiency is significantly improved in NOD/SCID-IL2RG mice constitutively expressing human SCF, GM-CSF and IL-3. Leukemia. 2010;24(10):1785-1788. 12. Morton JJ, Bird G, Keysar SB, et al. XactMice: humanizing mouse bone marrow enables microenvironment reconstitution in a patient-derived xenograft model of head and neck cancer. Oncogene. 2015;35(3):290300. 13. Rongvaux A, Willinger T, Takizawa H, et al. Human thrombopoietin knockin mice efficiently support human hematopoiesis in vivo. Proc Natl Acad Sci. 2011;108(6):23782383. 14. Traggiai E, Chicha L, Mazzucchelli L, et al. Development of a human adaptive immune system in cord blood cell-transplanted mice. Science. 2004;304(5667):104-107. 15. Coughlan AM, Harmon C, Whelan S, et al. Myeloid engraftment in humanized mice: impact of granulocyte-colony stimulating factor treatment and transgenic mouse strain. Stem Cells Dev. 2016;25(7):530-541. 16. Bryce PJ, Falahati R, Kenney LL, et al.

haematologica | 2020; 105(10)

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

Humanized mouse model of mast cell mediated passive cutaneous anaphylaxis and passive systemic anaphylaxis. J Allergy Clin Immunol. 2016;138(3):769-779. Bentley VL, Veinotte CJ, Corkery DP, et al. Focused chemical genomics using zebrafish xenotransplantation as a preclinical therapeutic platform for T-cell acute lymphoblastic leukemia. Haematol. 2015.100(1):70-76 Corkery DP, Dellaire G, Berman JN. Leukaemia xenotransplantation in zebrafish–chemotherapy response assay in vivo. Br J Haematol. 2011;153(6):786-789. Rajan V, Dellaire G, Berman JN. Modeling leukemogenesis in the zebrafish using genetic and xenograft models. Methods Mol Biol. 2016;1451:171-189. Howe K, Clark MD, Torroja CF, et al. The zebrafish reference genome sequence and its relationship to the human genome. Nature. 2013;496(7446):498-503. North TE, Goessling W, Walkley CR, et al. Prostaglandin E2 regulates vertebrate haematopoietic stem cell homeostasis. Nature. 2007;447(7147):1007-1011. Liu Y, Asnani A, Zou L, et al. Visnagin protects against doxorubicin-induced cardiomyopathy through modulation of mitochondrial malate dehydrogenase. Sci Transl Med. 2014;6(266):266ra170. Stachura DL, Svoboda O, Campbell CA, et al. The zebrafish granulocyte colony stimulating factors (Gcsfs): two paralogous cytokines and their roles in hematopoietic development and maintenance. Blood. 2013;122(24):3918-3928. Pruvot B, Jacquel A, Droin N, et al. Leukemic cell xenograft in zebrafish embryo for investigating drug efficacy. Haematologica. 2011;96(4):612-616 Hamilton N, Sabroe I, Renshaw SA. A method for transplantation of human HSCs into zebrafish, to replace humanised murine transplantation models. F1000Res. 2018;7: 594. White RM, Sessa A, Burke C, et al. Transparent adult zebrafish as a tool for in vivo transplantation analysis. Cell Stem Cell. 2008;2(2):183-189. Haldi M, Ton C, Seng WL, McGrath P. Human melanoma cells transplanted into zebrafish proliferate, migrate, produce melanin, form masses and stimulate angiogenesis in zebrafish. Angiogenesis. 2006;9(3):139-151. Zhang Y, Dépond M, He L, et al. CXCR4/CXCL12 axis counteracts hematopoietic stem cell exhaustion through selective protection against oxidative stress. Sci Rep. 2016;6:37827. Greenbaum A, Hsu Y-MS, Day RB, et al. CXCL12 in early mesenchymal progenitors is required for haematopoietic stem-cell maintenance. Nature. 2013;495(7440):227230. Sugiyama T, Kohara H, Noda M, Nagasawa T. Maintenance of the hematopoietic stem cell pool by CXCL12-CXCR4 chemokine signaling in bone marrow stromal cell niches. Immunity. 2006;25(6):977-988. Peled A, Kollet O, Ponomaryov T, et al. The chemokine SDF-1 activates the integrins LFA-1, VLA-4, and VLA-5 on immature human CD34+ cells: role in transendothelial/stromal migration and engraftment of NOD/SCID mice. Blood. 2000;95(11):32893296.

32. Barretina J, Caponigro G, Stransky N, et al. The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature. 2012;483(7391):603. 33. Ponomaryov T, Peled A, Petit I, et al. Induction of the chemokine stromal-derived factor-1 following DNA damage improves human stem cell function. J Clin Invest. 2000;106(11):1331-1339. 34. Glass TJ, Lund TC, Patrinostro X, et al. Stromal cell–derived factor-1 and hematopoietic cell homing in an adult zebrafish model of hematopoietic cell transplantation. Blood. 2011;118(3):766-774. 35. Labuhn M, Perkins K, Papaemmanuil E, et al. Modelling the progression of a preleukemic stage to overt leukemia in children with Down syndrome. Blood. 2018; 132(Suppl 1):543. 36. Theodore LN, Lundin V, Wrighton PJ, et al. YAP regulates hematopoietic stem cell formation in response to the biophysical forces of blood flow. Blood 2017;130(Suppl 1):1147. 37. Zou Y-R, Kottmann AH, Kuroda M, Taniuchi I, Littman DR. Function of the chemokine receptor CXCR4 in haematopoiesis and in cerebellar development. Nature. 1998;393(6585):595-599. 38. Nagasawa T, Hirota S, Tachibana K, et al. Defects of B-cell lymphopoiesis and bonemarrow myelopoiesis in mice lacking the CXC chemokine PBSF/SDF-1. Nature. 1996;382(6592):635-638. 39. Christensen JL, Wright DE, Wagers AJ, Weissman IL. Circulation and chemotaxis of fetal hematopoietic stem cells. PLOS Biol. 2004;2(3):e75. 40. Tulotta C, Stefanescu C, Beletkaia E, et al. Inhibition of signaling between human CXCR4 and zebrafish ligands by the small molecule IT1t impairs the formation of triple-negative breast cancer early metastases in a zebrafish xenograft model. Dis Model Mech. 2016;9(2):141-153. 41. Sacco A, Roccaro AM, Ma D, et al. Cancer cell dissemination and homing to the bone marrow in a zebrafish model. Can Res. 2016;76(2):463-471. 42. de Pater E, Kaimakis P, Vink CS, et al. Gata2 is required for HSC generation and survival. J Ex Med. 2013;210(13):2843-2850. 43. Nicolini FE, Cashman JD, Hogge DE, Humphries RK, Eaves CJ. NOD/SCID mice engineered to express human IL-3, GM-CSF and Steel factor constitutively mobilize engrafted human progenitors and compromise human stem cell regeneration. Leukemia. 2003;18(2):341-347. 44. Arai A, Jin A, Yan W, et al. SDF-1 synergistically enhances IL-3-induced activation of the Raf-1/MEK/Erk signaling pathway through activation of Rac and its effector Pak kinases to promote hematopoiesis and chemotaxis. Cell Signal. 2005;17(4):497506. 45. Rossjohn J, McKinstry WJ, Woodcock JM, et al. Structure of the activation domain of the GM-CSF/IL-3/IL-5 receptor common βchain bound to an antagonist. Blood. 2000;95(8): 2491-2498. 46. Ihle JN. Cytokine receptor signalling. Nature. 1995;377(6550):591-594. 47. Komatsu N, Suda T, Moroi M, et al. Growth and differentiation of a human megakaryoblastic cell line, CMK. Blood. 1989;74(1): 42-48.

2399

ARTICLE Ferrata Storti Foundation

Iron Metabolism & its Disorders

Dietary intake of heme iron is associated with ferritin and hemoglobin levels in Dutch blood donors: results from Donor InSight Tiffany C. Timmer,1,2,3 Rosa de Groot,1,4 Judith J.M. Rijnhart,4 Jeroen Lakerveld,4 Johannes Brug,5,6 Corine W.M. Perenboom,7 A. Mireille Baart,7 Femmeke J. Prinsze,1 Saurabh Zalpuri,1 C. Ellen van der Schoot,3,8 Wim L.A.M. de Kort1,2 and Katja van den Hurk1

Haematologica 2020 Volume 105(10):2400-2406

Sanquin Research, Department of Donor Medicine Research - Donor Studies, Amsterdam; 2Amsterdam UMC, University of Amsterdam, Department of Public Health, Amsterdam Public Health, Amsterdam; 3Landsteiner Laboratory, Amsterdam UMC, University of Amsterdam, Amsterdam; 4Amsterdam UMC, Location VU University Medical Center, Department of Epidemiology and Biostatistics, Amsterdam Public Health Research Institute, Amsterdam; 5National Institute for Public Health and the Environment, Bilthoven; 6 University of Amsterdam, Amsterdam School of Communication Research (ASCoR), Amsterdam; 7Wageningen University and Research, Division of Human Nutrition and Health, Wageningen and 8Sanquin Research, Department of Experimental Immunohematology, Amsterdam, the Netherlands 1

ABSTRACT

W Correspondence: KATJA VAN DEN HURK k.vandenhurk@sanquin.nl Received: June 13, 2019. Accepted: November 12, 2019. Pre-published: November 14, 2019. doi:10.3324/haematol.2019.229450

hole blood donors, especially frequently donating donors, have a risk of iron deficiency and low hemoglobin (Hb) levels, which may affect their health and eligibility to donate. Lifestyle behaviors, such as dietary iron intake and physical activity, may influence iron stores and thereby Hb levels. We aimed to investigate whether dietary iron intake and questionnaire-based moderate-to-vigorous physical activity (MVPA) were associated with Hb levels, and whether ferritin levels mediated these associations. In Donor InSight-III, a Dutch cohort study of blood and plasma donors, data on heme and non-heme iron intake (mg/day), MVPA (10 minutes/day), Hb levels (mmol/L) and ferritin levels (mg/L) were available in 2,323 donors (1,074 male). Donors with higher heme iron intakes [regression coefficients (β) in men and women: 0.160 and 0.065 mmol/L higher Hb per 1 mg of heme iron, respectively] and lower nonheme iron intakes (β: -0.014 and -0.017, respectively) had higher Hb levels, adjusted for relevant confounders. Ferritin levels mediated these associations [indirect effect (95% confidence interval) in men and women, respectively: 0.074 (0.045; 0.111) and 0.061 (0.030; 0.096) for heme and -0.003 (-0.008;0.001) and -0.008 (-0.013;-0.003) for non-heme]. MVPA was negatively associated with Hb levels in men only (β: -0.005), but not mediated by ferritin levels. In conclusion, higher heme and lower non-heme iron intake were associated with higher Hb levels in donors, via higher ferritin levels. This indicates that donors with high heme iron intake may be more capable of maintaining iron stores to recover Hb levels after blood donation.

©2020 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.

2400

Introduction A whole blood donation results in the loss of approximately 225-250 mg of iron.1 Therefore, frequent whole blood donations may lead to iron depletion and a subsequent decline in hemoglobin (Hb) levels.2,3 In order to ensure donor health and blood product quality, donor eligibility criteria are set.4 In many countries, including the Netherlands, minimum Hb levels are mandated at each donation: in the Netherlands these are 8.4 mmol/L (135 g/L) for men and 7.8 mmol/L (125 g/L) for women. A study among blood donors has shown that donors differ in Hb level recovery after blood donation, with some donors showing relatively stable Hb trajectories over time, while other donors show declining trajectories.5 This latter may haematologica | 2020; 105(10)

Lifestyle, ferritin and hemoglobin in donors

be due to the fact that for those donors a donation interval of 56 days, which is the minimum interval in many countries including the Netherlands, is too short to restore Hb levels.2,6,7 Several factors, including sex, age, season and number of donations are established determinants of Hb levels.8-12 It may be speculated that differences in lifestyle behaviors between donors, such as dietary iron intake and physical activity, may influence Hb levels as well. Iron homeostasis is tightly regulated and maintained by recycling iron from old erythrocytes, by replacing lost iron with dietary iron, and by mobilizing stored iron when necessary.1,13,14 In blood donors, dietary iron intake may be even more important in order to maintain iron homeostasis and thereby Hb levels given the iron loss associated with blood donation. A diet generally contains heme iron (present in animal foods) with high bioavailability (1535%) and non-heme iron (especially present in plantbased foods) with 1-20% bioavailability.14,15 Heme iron generally constitutes only about 15% of the total dietary iron intake.14,15 Two previous studies among blood donors did not find associations between intake of iron-rich food items and iron stores or Hb levels,16,17 while one study among blood donors found mainly meat intake to be associated with iron stores.18 To our knowledge, it is unknown whether dietary heme and non-heme iron intake are positively associated with Hb levels and iron stores in blood donors. Physical activity may influence Hb levels as well. Available literature suggests two general hypotheses with regard to this relation. First, physical activity may decrease Hb levels through iron loss via sweat, urine, and the gastrointestinal tract, as well as by exercise-induced hemolysis or hemodilution.19-21 Second, physical activity may increase Hb levels as physical activity requires increased amounts of oxygen to be transported throughout the body by Hb.22-24 The number of studies investigating the effect of physical activity on ferritin levels (i.e. a measure representing iron stores)14,25 are limited, particularly in blood donors, and the results of these studies are inconclusive.2629

Insights into associations between lifestyle behaviors and Hb levels are valuable for blood supply organizations. Lifestyle behaviors can potentially be taken into account in order to prevent Hb deferrals, for example through tailored donation intervals or lifestyle advice. In addition, studying the mediating role of ferritin levels in the associations between lifestyle behaviors and Hb levels will help to gain insight into whether iron stores could indeed be the limiting or enabling factor that links lifestyle behavior to Hb level recovery after donation. In a Swiss study, donors who were low in Hb or ferritin levels could choose one or more of the three following strategies: (i) iron supplementation; (ii) extension of the donation interval; and/or (iii) suggestions of dietary changes.30 This study found that these measures contributed to an increase in Hb level.30 The Dutch Donor InSight-III (DIS-III) study provides both questionnaire-based and accelerometryderived data on physical activity, as well as data on both heme and non-heme iron intake using validated questionnaires.31 This, in combination with measurements of Hb and ferritin levels, provides a unique opportunity to study how lifestyle behaviors are related to Hb levels in blood donors. Hence, we investigated: (i) associations between dietary iron intake and physical activity with Hb levels and Hb trajectories; and (ii) to what extent these associations are mediated by ferritin levels. We hypothesized haematologica | 2020; 105(10)

that a higher intake of heme iron, and to a lesser extent of non-heme iron, is associated with higher ferritin and Hb levels. Additionally, we hypothesized a potential positive association between MVPA and ferritin and Hb levels.

Methods Study population Data were collected as part of DIS-III (2015-2016), a cohort study among blood and plasma donors in the Netherlands. DIS-III aimed at gaining insight into donor characteristics, health and behavior. Details of DIS-III have been described elsewhere and information about Hb trajectories is available in the Online Supplementary Methods.32 Participants completed a general questionnaire and food frequency questionnaire (FFQ) one week prior to providing blood samples for DIS-III. Blood samples were either taken from the sampling pouch of a blood bag or, if not combined with a regular donation, through venipuncture. These blood samples were used to do a full blood count and to store samples to measure ferritin at a later moment (see Measurements section). A total of 2,552 (42% response rate) donors provided blood samples and completed the general questionnaire. For the current analyses, donors with self-reported diagnosis of hemochromatosis (n=6), who used iron supplements medication (n=221) and who were pregnant during DIS-III (n=5) were excluded (n=229, 9% in total), resulting in 2,323 participants. The Medical Ethical Committee of the Academic Medical Center (AMC) in the Netherlands and Sanquin’s Ethical Advisory Board approved DIS-III and all participants gave their written informed consent.

Measurements Hemoglobin levels and erythrocyte parameters [red blood cell count (RBC), hematocrit, mean cell volume (MCV), mean cell hemoglobin (MCH), mean cell hemoglobin concentration (MCHC) and red cell distribution width (RDW)] were measured for DIS-III using a hematology analyzer (XT‐2000, Sysmex, Kobe, Japan) in an EDTA whole blood sample within 24 hours (h) after blood donation.33 Collected lithium heparin tubes were centrifuged within 24 h after DIS-III blood collection and resulting plasma was subsequently stored at -80°C.34 Ferritin levels were measured within a year after blood collection, using the stored plasma sample from lithium heparin tubes (Architect Ci8200, Abbott Laboratories, IL, USA).34 Dietary heme and non-heme iron intake (mg/day) were measured with a FFQ31 adapted to assess iron intake. The FFQ assessed usual dietary consumption in the past four weeks. Physical activity and sedentary behavior were questionnaire-based as well as accelerometry-derived. Questionnaire-based assessments were done by using the validated International Physical Activity Questionnaire - Short Form (IPAQ-SF).35,36 Sedentary behavior was checked for confounding (see Online Supplementary Methods). Time spent in moderate-to-vigorous physical activity (MVPA) and sedentary behavior was expressed in minutes/day. In a subset of DIS-III participants (n=654), these were also objectively measured with accelerometers (wGT3X-BT and GT3X Actigraph, Pensacola, FL, USA) and data were handled using Troiano (2008) cut-off points.37 See Online Supplementary Methods for details on possible confounders.

Statistical analysis Descriptive statistics are presented as mean±standard deviation (SD), or in case of a skewed distribution as median and interquartile range (IQR). Associations between lifestyle behaviors 2401

T.C. Timmer et al. Table 1. Characteristics of the study population. Age at DIS-III, years Hb level, mmol/L RBC, x104 Hct, % MCV, fL MCH, amol MCHC, mmol/L RDW, % Subgroup Stable Hb trajectory Declining Hb trajectory Random sample Ferritin level, mg/L Heme iron intake, mg/day Non-heme iron intake, mg/day MVPA (questionnaire), min/day MVPA (accelerometer), min/day Sedentary behavior (questionnaire), min/day Sedentary behavior (accelerometer), min/day Initial Hb level*, mmol/L Number of donations in 2 years before DIS-III Donation interval, months Current smoker Yes No Menstruation in past 6 months Yes No

Males (n=1,074)

Females (n=1,249)

51.1 ± 13.0 9.3 ± 0.6 497.1 ± 36.2 44.9 ± 2.7 90.5 ± 4.6 1875.9 ± 106.5 20.7 ± 0.6 13.6 (13.1 – 14.2)

47.0 ± 13.0 8.4 ± 0.6 452.8 ± 34.1 41.2 ± 2.9 91.2 ± 4.8 1862.7 ± 252.2 20.4 ± 2.0 13.6 (13.1 – 14.4)

232 (22%) 468 (44%) 374 (35%) 56.8 (31.2 – 95.5) 1.1 (0.8 – 1.5) 9.7 (7.9 – 11.7) 64.3 (31.8 – 139.6) 32.4 (19.9 – 49.3) 480.0 (300.0 – 720.0) 575.3 (509.6 – 632.3) 9.5 ± 0.6 4 (0 – 7) 6 (3 – 25)

433 (35%) 438 (35%) 378 (30%) 35.9 (19.3 – 59.3) 0.9 (0.6 – 1.2) 7.9 (6.4 – 9.5) 51.4 (26.1 – 107.7) 26.7 (17.3 – 40.3) 420.0 (265.0 – 615.0) 532.7 (485.9 – 580.6) 8.5 ± 0.6 2 (0 – 4) 9 (5 – 35)

86 (8%) 917 (85%)

94 (8%) 1,060 (85%)

NA NA

563 (45%) 671 (54%)

Continuous variables: mean±standard deviation or median (interquartile range) if skewed; dichotomous variables: n (%); NA: not applicable; DIS-III: Donor InSight-III; Hb: hemoglobin; RBC: red blood cell count; Hct: hematocrit; MCV: mean cell volume; MCH: mean cell Hb; MCHC: mean cell Hb concentration; RDW: red cell distribution width; MVPA: moderate-to-vigorous physical activity. Note: due to missing data, numbers might not add up to total for dichotomous variables. Percentages might not add up to 100 because of rounding. *First capillary Hb measurement available in the blood bank information system.

(heme/non-heme iron intake and questionnaire-based MVPA) and Hb levels, and mediation analyses of ferritin levels as mediator of this association were studied using multiple linear regression analyses. Complete case analyses were performed, and in case of non-linear associations with skewed variables, the dependent variables were log-transformed. All models were constructed for men and women separately and adjusted for relevant confounders (Online Supplementary Methods and Online Supplementary Table S1).

intake were higher in men than in women (heme: 1.1 vs. 0.9 mg/day; non-heme: 9.7 vs. 7.9 mg/day). Higher medians were seen for questionnaire-based compared with accelerometry-derived MVPA and these medians were higher in men compared with women (questionnaire: 64.3 vs. 51.4 minutes/day; accelerometer: 32.4 vs. 26.7 minutes/day).38,39 In total, 232 males and 433 females had a stable Hb trajectory and 468 males and 438 females had a declining Hb trajectory.

Results

Associations of lifestyle behaviors with hemoglobin and ferritin levels

A total of 1,074 males and 1,249 females were included with a mean (SD) age of 51.1 (13.0) and 47.0 (13.0) years, respectively (Table 1). In total, 1,016 males and 1,171 females provided information on heme and non-heme iron intake, 795 males and 962 females provided information on self-reported MVPA, and 313 males and 357 females had information on accelerometry-derived MVPA. Men had higher median ferritin and mean Hb levels than women: 56.8 versus 35.9 mg/L and 9.3 versus 8.4 mmol/L, respectively. Median heme and non-heme iron

Associations of heme and non-heme iron intake, and MVPA with Hb levels are presented in Table 2. Age was found not to be an effect modifier. Adjustments were made for: (i) age, smoking, menstruation (in women only); (ii) number of donations in the previous two years, time since last donation; (iii) sedentary behavior, heme and non-heme iron intake or MVPA; and (iv) initial Hb levels. A higher intake of one mg of heme iron per day was associated with 0.160 and 0.065 mmol/L higher Hb levels in men and women, respectively. Higher intake of non-heme iron, however, was associated with slightly

2402

haematologica | 2020; 105(10)

Lifestyle, ferritin and hemoglobin in donors Table 2. Associations between lifestyle behaviors (heme and non-heme iron intake and moderate-to-vigorous physical activity (MVPA)] and hemoglobin (Hb) levels.

Independent variable

Crude model β (95% CI)

Model 1 β (95% CI)

Model 2 β (95% CI)

Model 3 β (95% CI)

Model 4 β (95% CI)

Heme Non-heme MVPA Heme Non-heme MVPA

0.125 (0.057; 0.193) -0.016 (-0.026; -0.007) -0.005 (-0.009; -0.002) 0.106 (0.033; 0.178) -0.020 (-0.032; -0.008) -0.000 (-0.004; 0.004)

0.123 (0.055; 0.191) -0.017 (-0.026; -0.008) -0.005 (-0.009; -0.002) 0.102 (0.030; 0.174) -0.021 (-0.033; -0.010) -0.002 (-0.006; 0.002)

0.126 (0.058; 0.194) -0.017 (-0.026; -0.008) -0.005 (-0.009; -0.001) 0.106 (0.034; 0.178) -0.022 (-0.033; -0.010) -0.002 (-0.006; 0.002)

0.188 (0.103; 0.272) -0.021 (-0.033; -0.008) -0.006 (-0.009; -0.002) 0.093 (0.005; 0.181) -0.022 (-0.036; -0.007) -0.003 (-0.007; 0.001)

0.160 (0.083; 0.238) -0.014 (-0.025; -0.003) -0.005 (-0.008; -0.001) 0.065 (-0.018; 0.148) -0.017 (-0.031; -0.003) -0.003 (-0.007; 0.001)

M: males; F: females; β: regression coefficient, 95%CI: 95% confidence interval; MVPA: moderate-to-vigorous physical activity in 10 minutes/day; Hb: hemoglobin in mmol/L. Heme and non-heme iron in mg/day. Model 1: adjusted for age, smoking, and menstruation (women only). Model 2: additionally adjusted for number of donations in the 2 years before DIS-III and donation interval. Model 3: additionally adjusted for sedentary behavior and MVPA in models with heme and non-heme iron intake as independent variables or sedentary behavior, heme and non-heme iron intake in models with MVPA as independent variable. Model 4: additionally adjusted for initial Hb level. Over 10% of participants excluded due to missing data in males model 3-4 for heme and non-heme iron intake, and models 1-4 for MVPA, and in females model 1-4 for heme and non-heme iron intake and MVPA.

Table 3. Associations between lifestyle behaviors [heme and non-heme iron intake and moderate-to-vigorous physical activity (MVPA)] and hemoglobin (Hb) levels and mediation by ferritin levels.

Lifestyle behaviors

Mediator

Dependent variable

Total effect (c path)† β (95%CI)

M Heme Non-heme MVPA F Heme Non-heme MVPA

Ferritin*

Hb levels

Effect of lifestyle behaviors on ferritin levels (a path) LN(β (95% CI))

0.160 (0.083; 0.238) 0.288 (0.192; 0.383) -0.014 (-0.025; -0.003) -0.012 (-0.027; 0.002) -0.005 (-0.008; -0.001) -0.000 (-0.005; 0.004) 0.065 (-0.018; 0.148) 0.222 (0.115; 0.328) -0.017 (-0.031; -0.003) -0.028 (-0.046; -0.010) -0.003 (-0.007; 0.001) 0.001 (-0.004; 0.006)

Effect of ferritin levels on Hb levels (b path) β (95% CI)

Direct effect (c’ path)

Indirect effect (a path * b path)

β (95% CI)

0.256 (0.198; 0.314)

0.090 (0.014; 0.165) 0.074 (0.045; 0.111) -0.011 (-0.022; -0.000) -0.003 (-0.008; 0.001) -0.005 (-0.009; -0.002) -0.000 (-0.001; 0.001)

0.276 (0.225; 0.327)

0.002 (-0.077; 0.080) 0.061 (0.030; 0.096) -0.010 (-0.023; 0.004) -0.008 (-0.013; -0.003) -0.003 (-0.007; 0.000) 0.000 (-0.001; 0.002)

M: males; F: females; β: regression coefficient, 95%CI: 95% confidence interval; BCI: bootstrapped confidence interval; MVPA: moderate-to-vigorous physical activity in 10 minutes/day; heme and non-heme iron intake in mg/day, ferritin levels in µg/L; Hb level in mmol/L. *Residuals of ferritin levels were not normally distributed and therefore log transformed; this table presents log transformed data. †These results are identical to the results of model 4 in Table 2. Total effect (c path): association between heme and non-heme iron intake or MVPA and Hb level; a path: association between heme and non-heme iron intake or MVPA and mediating variable ferritin levels; b path: association between mediating variable ferritin levels and Hb level; direct effect (c’ path): association between heme and non-heme iron intake or MVPA and Hb level adjusted for mediating variable ferritin levels; indirect effect: indirect effect of heme and non-heme iron intake or MVPA on Hb level through mediating variable ferritin levels. Adjusted for age, smoking, menstruation (in models with women only), number of donations, donation interval, sedentary behavior, MVPA in models with heme and non-heme iron intake as independent variables or heme and non-heme iron intake in models with MVPA as independent variable, and initial Hb level.

lower Hb levels in men and women (-0.014 and -0.017 mmol/L, respectively). Spending more time per day on MVPA was associated with lower Hb levels, but these results were only statistically significant in men [β (95%CI): -0.005 (-0.008 to -0.001)] (Table 2). Both, heme and non-heme iron intake, showed positive associations with ferritin levels. Table 3 (a path) shows log transformed results, as ferritin levels were not normally distributed. Back-transformation of these results showed 1.334 and 1.249 mmol/L higher ferritin levels per mg higher heme iron intake in men and women, respectively. For non-heme iron intake, these values were 0.988 and 0.972 for males and females, respectively. MVPA showed no statistically significant association with ferritin levels in either men [β (95%CI): 1.000 (0.995 to 1.004)) nor women (β (95%CI): 1.001 (0.996 to 1.006)]. Lifestyle behaviors and Hb trajectories showed similar, but less pronounced, associations as those with Hb levels as outcome (Online Supplementary Table S2). haematologica | 2020; 105(10)

Mediation by ferritin levels Associations between dietary iron intake and Hb levels were mediated by ferritin levels (Table 3). In both men and women, higher intake of heme iron was significantly associated with higher ferritin levels, and higher ferritin levels with higher Hb levels [b path: 0.256 (0.198; 0.314) in men and 0.276 (0.225; 0.327) in women]. The association between heme iron intake was largely mediated by ferritin, showing indirect effects of 0.074 (0.045; 0.111) in men and 0.061 (0.030; 0.096) in women. The direct, nonmediated effect of heme iron intake on Hb levels was only statistically significant in men. Higher intake of non-heme iron was associated with lower ferritin and Hb levels. The association between non-heme iron intake and Hb levels was also mediated by ferritin, but only significantly in women [indirect effect: -0.003 (-0.008; 0.001) in men and -0.008 (-0.013; -0.003) in women]. Ferritin levels did not mediate the association between MVPA and Hb levels (Table 3). MVPA was not associated with ferritin levels (Table 3, a path) and accordingly, the 2403

T.C. Timmer et al. Table 4. Associations between erythrocyte parameters and lifestyle behaviors [heme and non-heme iron intake and moderate-to-vigorous physical activity (MVPA)].

Independent variable M Heme Non-heme MVPA F Heme Non-heme MVPA

RBC β (95% CI)

Hct β (95% CI)

MCV β (95% CI)

MCH β (95% CI)

MCHC β (95% CI)

RDW* LNβ (95% CI)

0.043 (-0.036; 0.122) 0.004 (-0.008; 0.015) 0.000 (-0.004; 0.003)

-0.004 (-0.014; 0.006) -0.001 (-0.002; 0.001) 0.000 (-0.001; 0.000)

0.099 (-0.624; 0.822) 1.786 (-14.487; 18.058)† -0.019 (-0.109; 0.071)† -0.073 (-0.194; 0.048) -2.076 (-4.799; 0.647)† -0.005 (-0.020; 0.010)† -0.021 (-0.054; 0.012) -0.236 (-0.977; 0.506)† 0.002 (-0.002; 0.006)†

0.003 (-0.010; 0.016) -0.001 (-0.003; 0.001) 0.000 (-0.001; 0.000)

4.242 (-0.397; 8.881) -0.484 (-1.166; 0.198) -0.248 (-0.457; -0.040)

0.681 (0.330; 1.033) 0.612 (0.005; 1.220) 16.284 (1.833; 30.736) -0.076 (-0.128; -0.025) -0.072 (-0.162; 0.017) -1.137 (-3.260; 0.987) -0.023 (-0.039; -0.007) -0.002 (-0.029; 0.025) -0.094 (-0.744; 0.556)

4.600 (-0.791; 9.991) -0.573 (-1.477; 0.330) -0.083 (-0.329; 0.163)

0.465 (0.022; 0.908) -0.082(-0.156; -0.008) -0.016 (-0.037; 0.004)

M: males; F: females; β: regression coefficient, 95%CI: 95% confidence interval; MCH: mean cell hemolgobin (Hb) in mmol; MVPA: moderate-to-vigorous physical activity in 10 minutes/day; heme and non-heme iron intake in mg/day; RBC: red blood cell count in x104; Hct: hematocrit in %; MCV: mean cell volume in fL; MCH: mean cell Hb in amol; MCHC: mean cell Hb concentration in mmol/L; RDW: red cell distribution width in %; ferritin levels in µg/L. *Residuals of ferritin levels and RDW were not normally distributed and therefore log transformed; this table presents log transformed data. Adjusted for age, smoking, menstruation (in models with women only), number of donations, donation interval, sedentary behavior, MVPA in models with heme and non-heme iron intake as independent variables or heme and non-heme iron intake in models with MVPA as independent variable, and initial Hb. †Results with outlier removed, without outlier removed: heme [-29.812 (-77.210 to 17.585)], non-heme [-0.534 (-8.474 to 7.407)] and MVPA [-0.864 (3.025 to 1.297)] for MCH and heme [-0.292 (-0.688 to 0.103)], non-heme [0.009 (-0.058 to 0.075)] and MVPA [-0.003 (-0.021 to 0.015)] for MCHC.

indirect effects were close to zero. The direct effect (c’ path) of MVPA on Hb level was statistically significant in men only [β (95%CI): -0.005 (-0.009 to -0.002)]. With regard to Hb trajectories, a significant indirect effect was found for heme and non-heme iron intake in women only (0.094 (0.013; 0.203) for heme iron and -0.011 (-0.026; -0.002) for non-heme iron). Ferritin did not mediate the other associations between lifestyle behaviors and Hb trajectories (Online Supplementary Table S3).

Sensitivity and post-hoc analyses Table 4 shows the results of post-hoc analyses on associations between lifestyle behaviors and erythrocyte parameters. Heme iron intake, non-heme iron intake and MVPA were mainly associated with hematocrit. In men, associations were also found between heme iron intake and MCV and MCH, and between MVPA and RBC. In sensitivity analyses in a subset of the study population with accelerometry-derived MVPA, the direct effect of MVPA on Hb levels in men was no longer statistically significant in any model (Online Supplementary Table S4). Post-hoc analyses in which additional adjustments for phytate-rich and polyphenol-rich food items were made showed similar associations with heme iron intake [0.145 (0.066; 0.225) and 0.079 (-0.007; 0.165) in men and women, respectively], while associations with non-heme iron intake diminished [β (95%CI): 0.000 (-0.018; 0.018) and 0.015 (-0.036; 0.006) in men and women, respectively].

Discussion In this study among Dutch blood donors, we found that dietary iron intake was associated with Hb levels of blood donors via ferritin levels. Heme iron intake showed a positive and non-heme iron intake a negative association with Hb levels. To put this amount of iron into perspective, 1 mg higher heme iron intake, equivalent to 58 grams of prepared beef or 700 grams of prepared chicken filet,40 was associated with 0.160 mmol/L higher Hb levels in men. With regard to non-heme iron, 1 mg higher non-heme iron intake, equivalent to 60 grams of cooked whole wheat 2404

pasta or 2.5 salty herring (187.5 grams),40 was associated with -0.014 mmol/L lower Hb levels in men. A statistically significant, but rather small (-0.015 mmol/L for 30 min/day MVPA in men), negative association between questionnaire-based but not accelerometry-derived physical activity and Hb levels was found in men only. This association was independent of ferritin levels. Results were independent of frequency of previous donations as we adjusted for number of donations in the 2 years before DIS-III and donation interval. As hypothesized, a positive association was found between heme iron intake and Hb and ferritin levels. It seems that donors who consume more heme iron can restore their iron stores better, resulting in higher ferritin and Hb levels. Further analyses in which lifestyle behaviors were associated with erythrocyte parameters showed that heme iron intake increases the volume of blood that is occupied by red blood cells (higher Hct) and vice versa (lower Hct) for non-heme iron intake and MVPA. In men, higher heme iron intake was also associated with higher MCV and MCH, indicating that, in men, in addition to a larger volume of red blood cells, these cells also contain more Hb. Interestingly, higher intake of non-heme iron was associated with lower Hb levels, independent of heme iron intake. An explanation might be that with higher intake of non-heme iron, more phytate-rich and polyphenol-rich foods and beverages (e.g., legumes, grains and coffee) are consumed, preventing absorption of nonheme iron.41-43 Indeed, post-hoc analyses with additional adjustments for phytate-rich and polyphenol-rich food items (i.e., legumes, bread, pasta, cereals, nuts and coffee) diminished the negative associations between non-heme iron intake and Hb levels. More precise measurements of total phytate and polyphenol intake, rather than the consumption of food items, would enable more accurate adjustments for these substances. With regard to MVPA, the negative association with Hb levels may be due to exercise-induced hemolysis; however, this has mainly been found in studies investigating endurance athletes.17,19,44 In the additional analyses of MVPA with erythrocyte parameters, we did find that more MVPA was associated with lower numbers of erythrocytes. Another haematologica | 2020; 105(10)

Lifestyle, ferritin and hemoglobin in donors

potential explanation could be hemodilution, caused by exercise-induced plasma volume expansion.20 This phenomenon is often seen in athletes and is also known as sports anemia.20,21 However, based on the results of this study, no firm conclusions on mechanisms behind the association of MVPA with Hb levels can be drawn. In contrast to our findings on associations with Hb levels, there was only one statistically significant association between lifestyle behaviors and Hb trajectories (Online Supplementary Table S2). An explanation could be the loss of power to detect an association due to the dichotomization of the Hb level measurements into Hb trajectories (stable/declining) and the lower number of participants with a known Hb trajectory.45 Our finding that iron intake was associated with Hb and ferritin levels is in contrast with previous studies conducted among blood donors.16,17 However, these previous studies assessed consumption of iron-containing food items. Since the majority of iron in food is consumed as nonheme iron, and we found this to be negatively associated with Hb levels, heme iron needs to be distinguished from non-heme iron in order to recognize the positive effect on Hb levels of heme iron intake. Research among a general population of Dutch adults supports our findings; they also found that heme iron intake was positively, and nonheme iron intake negatively associated with iron status.46 With regard to physical activity, our results are in agreement with another study among Danish blood donors that also found a negative association for questionnairebased physical activity (hours/week) with Hb levels in men only [Î˛ (95%CI) of -0.09 (-0.11 to -0.06) for nonsmokers and -0.11 (-0.18 to -0.05) for smokers].17 Strengths of this study include the large study population and the detailed assessment of lifestyle behaviors, erythrocyte parameters and ferritin levels. As the FFQ enabled us to calculate both heme and non-heme iron intake rather than assessing total iron intake, we were able to show that there are important differences in the direction and magnitude of the associations between heme and non-heme iron intake and Hb levels. A limitation of the FFQ, however, is that it does not measure when and in which combination food items are consumed.47 Another limitation of this study is the use of questionnaire-based MVPA, which is prone to social desirability and recall bias, and the validity is known to differ across respondents from different socio-economic strata, but it is also the most cost-effective way to measure physical activity in a large sample and enables differentiation between types of activity.39,48,49 We did, however, use a validated questionnaire,35,36 and were able to perform sensitivity analyses with accelerometry-derived data in a subgroup of the participants. Results were similar but not significant in the

References 1. O'Brien SF, Goldman M. Understanding iron depletion and overload in blood donors. ISBT Sci Ser. 2017;12(1):11-18. 2. Niittymaki P, Arvas M, Larjo A, et al. Retrospective analysis of capillary hemoglobin recovery in nearly 1 200 000 blood donor returns. Blood Adv. 2017;1(14):961967. 3. Schotten N, Pasker-de Jong PC, Moretti D,

haematologica | 2020; 105(10)

accelerometry sub-group, probably due to the smaller study population. Next, it could be argued that including menstruating women in the analyses could have altered associations between heme and non-heme iron intake and Hb levels in this subpopulation. However, analyses showed that menstrual status was not an effect modifier of the association between iron intake and Hb levels, indicating that the association between iron intake and Hb levels is similar for menstruating versus non-menstruating women. Last, analyses of this study were cross-sectional, and we are therefore unable to infer causation. The results from this observational study do not implicate that blood donors will benefit from dietary advice. A review regarding solutions to iron deficiency in young women living in industrialized countries showed that dietary advice was not associated with an increase in Hb levels in two studies, but these studies showed conflicting results with regard to the effect of dietary advice on ferritin levels.50 A study among Swiss donors found that iron supplementation, extension of the donation interval, and/or suggestions of dietary changes resulted in a decrease in the prevalence of anemia and iron deficiency.30 However, this study was limited by the fact that donors were not randomized to one of the three interventions and no standardized information was handed out to the donor.30 Lastly, a study among blood donors conducted by our research group found that dietary advice did not reduce the risk of low-Hb deferral.51 Tailored donation intervals might be useful in preventing low Hb levels in blood donors;3,6,7 however, the usefulness of dietary information in tailoring these intervals should first be investigated further. In conclusion, blood donors with a higher intake of heme iron and lower intake of non-heme iron generally had higher Hb levels, and this was mediated by higher ferritin levels. In men, more time spent in MVPA was associated with lower Hb levels, independent of ferritin. Giving separate consideration to heme iron intake and non-heme iron intake may be useful in the prevention of low Hb levels in blood donors. Funding This work was supported by a Product and Process Development Grant (PPOC-14-028) from the Sanquin Blood Supply Foundation. Acknowledgments The authors would like to thank all participants and blood bank personnel for their contribution to the study. The authors also would like to thank J. van Rosmalen and K. Nasserinejad for fitting the growth mixture models and EMJ Huis in â&#x20AC;&#x2DC;t Veld for her contribution to the plan for analyses in an early stage.

et al. The donation interval of 56 days requires extension to 180 days for whole blood donors to recover from changes in iron metabolism. Blood. 2016; 128(17):2185-2188. 4. Council of Europe. Guide to the preparation, use and quality assurance of blood components, European Committee (partial agreement) on Blood Transfusion (CD-PTS): Recommendation no. r(95) 15. 15 ed. Strasbourg: Council of Europe 2010. 5. Nasserinejad K, van Rosmalen J, van den

Hurk K, et al. Prevalence and determinants of declining versus stable hemoglobin levels in whole blood donors. Transfusion. 2015;55(8):1955-1963. 6. Di Angelantonio E, Thompson SG, Kaptoge S, et al. Efficiency and safety of varying the frequency of whole blood donation (INTERVAL): a randomised trial of 45 000 donors. Lancet. 2017; 390(10110):2360-2371. 7. Baart AM, van den Hurk K, de Kort WLAM. Minimum donation intervals

2405

T.C. Timmer et al.

10. 11.

12. 13. 14. 15.

16.

17.

18.

19. 20. 21.

22.

2406

should be reconsidered to decrease low hemoglobin deferral in whole blood donors: an observational study. Transfusion. 2015;55(11):2641-2644. Baart AM, de Kort WLAM, Atsma F, Moons KG, Vergouwe Y. Development and validation of a prediction model for low hemoglobin deferral in a large cohort of whole blood donors. Transfusion. 2012; 52(12):2559-2569. Murphy WG. The sex difference in haemoglobin levels in adults - mechanisms, causes, and consequences. Blood Rev. 2014; 28(2):41-47. Kelly A, Munan L. Haematologic profile of natural populations: red cell parameters. Br J Haematol. 1977;35(1):153-160. Hoekstra T, Veldhuizen I, van Noord PA, de Kort WLAM. Seasonal influences on hemoglobin levels and deferral rates in wholeblood and plasma donors. Transfusion. 2007;47(5):895-900. Kiss JE. Laboratory and genetic assessment of iron deficiency in blood donors. Clin Lab Med. 2015;35(1):73-91. Ganz T. Systemic iron homeostasis. Physiol Rev. 2013;93(4):1721-1741. Strain JJ, Cashman KD. Minerals and trace elements. Human nutrition: The Nutrition Society. 2009. Cao C, Thomas CE, Insogna KL, O'Brien KO. Duodenal absorption and tissue utilization of dietary heme and nonheme iron differ in rats. J Nutr. 2014;144(11):17101717. Cable RG, Glynn SA, Kiss JE, et al. Iron deficiency in blood donors: the REDS-II Donor Iron Status Evaluation (RISE) study. Transfusion. 2012;52(4):702-711. Kotze SR, Pedersen OB, Petersen MS, et al. Predictors of hemoglobin in Danish blood donors: results from the Danish Blood Donor Study. Transfusion. 2015; 55(6):1303-1311. Rigas AS, Sorensen CJ, Pedersen OB, et al. Predictors of iron levels in 14,737 Danish blood donors: results from the Danish Blood Donor Study. Transfusion. 2014;54(3 Pt 2):789-796. Hinton PS. Iron and the endurance athlete. Appl Physiol Nutr Metab. 2014;39(9):10121018. Eichner ER. Sports anemia, iron supplements, and blood doping. Med Sci Sports Exerc. 1992;24(9 Suppl):S315-S318. Garvican-Lewis LA, Schumacher YO, Clark SA, et al. Stage racing at altitude induces hemodilution despite an increase in hemoglobin mass. J Appl Physiol (1985). 2014;117(5):463-472. Otto JM, Montgomery HE, Richards T. Haemoglobin concentration and mass as determinants of exercise performance and of surgical outcome. Extrem Physiol Med.

2013;2(1):33-40. 23. Gim MN, Choi JH. The effects of weekly exercise time on VO2max and resting metabolic rate in normal adults. J Phys Ther Sci. 2016;28(4):1359-1363. 24. Fogelholm M, Alopaeus K, Silvennoinen T, Teirila J. Factors affecting iron status in non-pregnant women from urban south Finland. Eur J Clin Nutr. 1993;47(8):567574. 25. Lynch S. The rationale for selecting and standardizing iron status indicators. Geneva: World Health Organization; 2012. 26. Schumacher YO, Schmid A, Grathwohl D, Bultermann D, Berg A. Hematological indices and iron status in athletes of various sports and performances. Med Sci Sports Exerc. 2002;34(5):869-875. 27. Murray-Kolb LE, Beard JL, Joseph LJ, Davey SL, Evans WJ, Campbell WW. Resistance training affects iron status in older men and women. Int J Sport Nutr Exerc Metab. 2001;11(3):287-298. 28. Milman N, Kirchhoff M. Relationship between serum ferritin and risk factors for ischaemic heart disease in 2235 Danes aged 30-60 years. J Intern Med. 1999;245(5):423433. 29. Bourque SP, Pate RR, Branch JD. Twelve weeks of endurance exercise training does not affect iron status measures in women. J Am Diet Assoc. 1997;97(10):1116-1121. 30. O'Meara A, Infanti L, Stebler C, et al. The value of routine ferritin measurement in blood donors. Transfusion. 2011;51(10):2183-2188. 31. Streppel MT, de Vries JH, Meijboom S, et al. Relative validity of the food frequency questionnaire used to assess dietary intake in the Leiden Longevity Study. Nutr J. 2013;12:75. 32. Timmer TC, de Groot R, Habets K, et al. Donor InSight: characteristics and representativeness of a Dutch cohort study on blood and plasma donors. Vox Sang. 2019;114(2):117-128. 33. Hill VL, Simpson VZ, Higgins JM, et al. Evaluation of the Performance of the Sysmex XT-2000i Hematology Analyzer With Whole Bloods Stored at Room Temperature. Lab Med. 2009;40(12):709-718. 34. Abbott. Ferritin. 2010 [cited; Available from: http://www.ilexmedical.com/ files/PDF/Ferritin_ARC.pdf 35. Mader U, Martin BW, Schutz Y, Marti B. Validity of four short physical activity questionnaires in middle-aged persons. Med Sci Sports Exerc. 2006;38(7):12551266. 36. Craig CL, Marshall AL, Sjostrom M, et al. International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc. 2003;35(8):13811395.

37. Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, McDowell M. Physical Activity in the United States Measured by Accelerometer. Med Sci Sports Exerc. 2008; 40(1):181-188. 38. Sabia S, van Hees VT, Shipley MJ, et al. Association between questionnaire- and accelerometer-assessed physical activity: the role of sociodemographic factors. Am J Epidemiol. 2014;179(6):781-790. 39. Skender S, Ose J, Chang-Claude J, et al. Accelerometry and physical activity questionnaires - a systematic review. BMC Public Health. 2016;16:515. 40. The Netherlands Nutrition Centre. IJzer. Voedingscentrum Encyclopedie [cited 0506-2019]; Available from: https://www. voedingscentrum.nl/encyclopedie/ ijzer.aspx 41. Conrad ME, Umbreit JN. Iron absorption and transport-an update. Am J Hematol. 2000;64(4):287-298. 42. Lim KH, Riddell LJ, Nowson CA, Booth AO, Szymlek-Gay EA. Iron and zinc nutrition in the economically-developed world: a review. Nutrients. 2013;5(8):3184-3211. 43. Kim EY, Ham SK, Shigenaga MK, Han O. Bioactive dietary polyphenolic compounds reduce nonheme iron transport across human intestinal cell monolayers. J Nutr. 2008;138(9):1647-1651. 44. Smith JA. Exercise, training and red blood cell turnover. Sports Med. 1995;19(1):9-31. 45. Altman DG, Royston P. The cost of dichotomising continuous variables. BMJ. 2006;332(7549):1080. 46. Brussaard JH, Brants HA, Bouman M, Lowik MR. Iron intake and iron status among adults in the Netherlands. Eur J Clin Nutr. 1997;51 Suppl 3:S51-S58. 47. Ahmadi A, Enayatizadeh N, Akbarzadeh M, Asadi S, Tabatabaee SH. Iron status in female athletes participating in team ballsports. Pak J Biol Sci. 2010;13(2):93-96. 48. Adams SA, Matthews CE, Ebbeling CB, et al. The effect of social desirability and social approval on self-reports of physical activity. Am J Epidemiol. 2005;161(4):389398. 49. Winckers AN, Mackenbach JD, Compernolle S, et al. Educational differences in the validity of self-reported physical activity. BMC Public Health. 2015; 15:1299. 50. Beck KL, Conlon CA, Kruger R, Coad J. Dietary determinants of and possible solutions to iron deficiency for young women living in industrialized countries: a review. Nutrients. 2014;6(9):3747-3776. 51. Baart AM, van den Hurk K, de Kort WLAM, et al. Impact of risk-dependent interventions on low haemoglobin deferral rates in whole blood donors. Vox Sang. 2020;115(3):171-181.

haematologica | 2020; 105(10)

ARTICLE

Red Cell Biology & its Disorders

Repurposing pyridoxamine for therapeutic intervention of intravascular cell-cell interactions in mouse models of sickle cell disease

Ferrata Storti Foundation

Jing Li,1* Si-yeon Jeong,1* Bei Xiong,1,2* Alan Tseng,1 Andrew B. Mahon,3 Steven Isaacman,3 Victor R. Gordeuk4,5 and Jaehyung Cho1

Department of Pharmacology, University of Illinois at Chicago College of Medicine, Chicago, IL, USA; 2Department of Hematology, Zhongnan Hospital of Wuhan University, Wuhan, Hubei, P.R. China; 3PHD Biosciences, New York, NY, USA; 4Section of Hematology/Oncology, University of Illinois at Chicago College of Medicine, Chicago, IL, USA and 5Comprehensive Sickle Cell Center, University of Illinois at Chicago College of Medicine, Chicago, IL, USA 1

Haematologica 2020 Volume 105(10):2407-2419

*Jing Li, Si-yeon Jeong, and Bei Xiong contributed equally as co-first authors.

ABSTRACT

dherent neutrophils on vascular endothelium positively contribute to cell-cell aggregation and vaso-occlusion in sickle cell disease (SCD). In the present study, we demonstrated that pyridoxamine, a derivative of vitamin B6, might be a therapeutic agent to alleviate intravascular cell-cell aggregation in SCD. Using real-time intravital microscopy, we found that one oral administration of pyridoxamine, in a dose-dependent manner, increased the rolling influx of neutrophils and reduced neutrophil adhesion to endothelial cells in cremaster microvessels of SCD mice challenged with hypoxia-reoxygenation. Short-term treatment also mitigated neutrophil-endothelial cell and neutrophil-platelet interactions in the microvessels and improved the survival of SCD mice challenged with tumor necrosis factor-α. The inhibitory effects of pyridoxamine on intravascular cell-cell interactions were enhanced by co-treatment with hydroxyurea. We observed that long-term (5.5 months) oral treatment with pyridoxamine significantly reduced the adhesive function of neutrophils and platelets, and down-regulated the expression of E-selectin and intercellular adhesion molecule-1 on the vascular endothelium in tumor necrosis factor-α-challenged SCD mice. Ex vivo studies revealed that the surface amount of αMβ2 integrin was significantly decreased in stimulated neutrophils isolated from SCD mice treated with pyridoxamine-containing water. Studies using platelets and neutrophils from SCD mice and patients suggested that treatment with pyridoxamine reduced the activation state of platelets and neutrophils. These results suggest that pyridoxamine may be a novel therapeutic and a supplement to hydroxyurea to prevent and treat vaco-occlusion events in SCD.

Correspondence: JAEHYUNG CHO thromres@uic.edu Received: May 15, 2019. Accepted: October 29, 2019. Pre-published: October 31, 2019. doi:10.3324/haematol.2019.226720

Introduction Sickle cell disease (SCD) is an inherited autosomal recessive disorder caused by a Glu6Val mutation in β-globin, resulting in hemolysis of red blood cells (RBC), oxidative stress, and chronic inflammation.1 Recurrent vaso-occlusion (VOC) is the hallmark of SCD and is mediated by intravascular cell-cell adhesion and aggregation. VOC is a crucial trigger for severe pain crisis and acute chest syndrome, a common complication and cause of death in SCD patients.1,2 Hydroxyurea (HU) and L-glutamine are the only US Food and Drug Administration-approved medications for treating SCD. However, SCD patients undergoing HU therapy still suffer from VOC-mediated events, and 25-50% of patients do not respond to this drug with an adequate increase in fetal hemoglobin (HbF).3,4 A recent phase III study has shown that, compared to the placebo control, treatment with L-glutamine reduces the number of pain crises in SCD patients.5 In this trial, however, patients on both haematologica | 2020; 105(10)

2407

J. Li et al.

L-glutamine therapy and HU still experienced pain crises, and this underscores the need for new therapies for SCD. Adherent neutrophils on endothelial cells (EC) are a key contributor to cell-cell aggregation and VOC in SCD.6-8 Clinical studies with crizanlizumab (a blocking antibody against P-selectin) demonstrated that inhibition of leukocyte-EC interactions prevents vaso-occlusive pain crises in SCD patients.9,10 Nevertheless, complete blockade of leukocyte-EC contact might impair the immune response. Because of the contribution of activated platelets to cellcell aggregation and VOC, antiplatelet drugs including P2Y12 antagonists (e.g. prasugrel and ticagrelor) have been tested in clinical trials for the treatment of VOC in SCD patients. A phase III study, however, revealed that prasugrel does not decrease the rate of VOC-mediated pain crises in children and adolescents with SCD.11 Furthermore, antiplatelet drugs increase the risk of major bleeding.12 Thus, there remains a need for safer and more efficacious therapeutics that ameliorate intravascular cell-cell aggregation and VOC-mediated conditions in SCD. Pyridoxamine, a derivative of vitamin B6, scavenges reactive oxygen species (ROS) and toxic carbonyls and inhibits the formation of advanced glycation end products (AGE).13,14 Preclinical studies suggest that long-term treatment with pyridoxamine inhibits the progression of renal disease and hyperlipidemia in diabetic rats15 and reduces adherence of uropathogenic E. coli to the bladder in diabetic mice.16 Furthermore, a phase II study demonstrated that oral administration of pyridoxamine has benefits for creatinine clearance and the level of urinary transforming growth factor-β1 (TGF-β1) in patients with type 1 and type 2 diabetic nephropathy.17 In contrast, another clinical study did not show any benefit of pyridoxamine treatment on serum creatinine concentration in patients with type 2 diabetic nephropathy after 1 year of therapy.18 No or minimal adverse events of pyridoxamine were observed in clinical studies.17,18 Intriguingly, the plasma concentration of AGE is elevated under oxidative stress conditions and is associated with organ complications in SCD patients.19 We therefore tested the effect of pyridoxamine on intravascular cell-cell interactions and VOC in SCD. We found that short- and long-term oral administration of pyridoxamine reduces neutrophil recruitment to the cremaster venular wall of SCD mice challenged with hypoxia/reoxygenation (H/R) or tumor necrosis factor-α (TNF-α) and improved survival. The beneficial effects of pyridoxamine were enhanced when HU was co-administered to SCD mice. We observed that pyridoxamine reduced the activation state and/or adhesiveness of neutrophils, platelets, and EC in SCD mice without affecting the plasma levels of AGE and nitric oxide (NO). These results suggest that pyridoxamine could be beneficial as a stand-alone therapeutic agent and in combination with HU to prevent and treat VOC-mediated events in SCD patients.

In vivo intravital microscopy Intravital microscopy was performed as previously described.20,21 For short-term treatment, pyridoxamine was given to SCD mice 3 hours (h) before in vivo imaging, given a previous report showing that the half-life of pyridoxamine is around 1.5 h in rats after oral administration.15 SCD mice were placed into a hypoxic chamber (8% O2) for 3 h, followed by oral administration of vehicle (saline) or different doses of pyridoxamine (10-100 mg/kg BW). After 3 h of reoxygenation at room temperature, mice were anesthetized with ketamine and xylazine, and the cremaster muscle was exposed for intravital microscopy. In some experiments, vehicle or pyridoxamine was given orally to SCD mice before intraperitoneal (ip) injection of TNF-α (500 ng). Three hours after TNF-α injection, mice underwent intravital microscopy. To assess the effect of long-term treatment with pyridoxamine, mice were given acidic water (pH 4.0) with or without pyridoxamine (2 g/L) for 5.5 months as previously described,15,22 starting at 2 weeks after bone marrow transplantation (BMT). Body weight and complete blood counts were measured every other week. After 5.5-months of treatment, the mice were challenged with H/R or ip injection of TNF-α as described above. Platelets and neutrophils were monitored by infusion of DyLight 488-conjugated anti-CD42c (0.1 mg/g BW) and Alexa Fluor 647-conjugated anti-Ly-6G antibodies (0.1 mg/g BW), respectively, through a jugular cannula. Fluorescence and bright-field images were recorded using an Olympus BX61W microscope with a 60x1.0 NA water immersion objective lens and a high-speed camera (ORCAFlash4.0 V3, C13440-20CU, Hamamatsu). Data were analyzed using SlideBook 6 (Intelligent Imaging Innovations). Since the cremaster venules with a diameter of <10 mm were mostly occluded, as assessed by absence of circulating RBC, the dynamics of intravascular cell-cell adhesion/accumulation were monitored in the venules with a diameter of 25-40 mm. Five to eight vessels were randomly chosen in each mouse. The numbers of rolling (cells per minute) and adherent neutrophils (cells per 5 minutes) were counted in the field of view (221.45 mm x 221.45 mm). The rolling velocity of neutrophils was also measured. Time “0” was set as the image capture was initiated at each vessel. To determine the kinetics of platelet accumulation, the integrated median fluorescence intensity values of the anti-CD42c antibody were normalized to the number of adherent neutrophils and the length of vessels, and plotted as a function of time. The experiments were performed in a single-blind fashion in which the investigators did not know the treatment allocation of the mice. Further details of the methods used are provided in the Online Supplementary Appendix.

Statistical analysis Data were analyzed using GraphPad Prism 7 by Student’s t-test, Mann-Whitney U test, ANOVA with Dunnett’s or Tukey’s test, Kruskal-Wallis test with post-hoc Dunn correction, and MantleCox log-rank test (survival curve). P<0.05 was considered statistically significant.

Results Methods The University of Illinois Institutional Animal Care and Use Committee approved all animal care and experimental procedures. All patients enrolled in this study provided informed consent. The collection and use of blood samples for laboratory analysis were approved by the Institutional Review Board of the University of Illinois at Chicago. 2408

One oral administration of pyridoxamine reduces neutrophil-endothelial cell interactions in the cremaster venules of sickle cell disease mice challenged with hypoxia/reoxygenation Red blood cell sickling under hypoxic conditions leads to hemolysis, cell-cell aggregation, and VOC in SCD patients.23 Similarly, H challenge results in RBC sickling and sickle hemoglobin (HbS) polymerization in Berkeley haematologica | 2020; 105(10)

Effect of pyridoxamine on VOC in SCD

mice, and subsequent R conditions induce acute VOC events.20,24 We and others have demonstrated that adherent neutrophils on activated EC support adhesion of other blood cells, including sickle RBC and platelets,6,20,25,26 and that H/R challenge induces intravascular cell-cell interactions and VOC events in SCD mice.20,27 To test the effect of short-term treatment with pyridoxamine on cell-cell interactions in SCD, different doses of pyridoxamine (10-100

mg/kg BW) were given orally to SCD mice after H challenge (8% O2 for 3 h), followed by R (at room temperature for another 3 h) (Figure 1A). Using intravital microscopy, we found that compared to the vehicle control, pyridoxamine, in a dose-dependent manner, enhanced the rolling influx of neutrophils and decreased neutrophil adhesion to EC in H/R-challenged SCD mice (Figure 1B-D and Online Supplementary Videos 1-4). Pyridoxamine at the highest

***P<0.001, saline vs. 100

Figure 1. A single oral administration of pyridoxamine inhibits neutrophil-endothelial cell (EC) interactions in the cremaster venules of hypoxia/reoxygenation (H/R)-challenged sickle cell disease (SCD) mice. SCD mice were placed into a hypoxic chamber (8% O2) for 3 hours (h) and treated with oral administration of vehicle or different doses of pyridoxamine (10, 30 or 100 mg/kg BW). After reoxygenation, intravital microscopy was performed as described in the Methods section. Neutrophils (red) and platelets (green) were visualized by infusion of Alexa Fluor 647-conjugated anti-Ly-6G and DyLight 488-conjugated anti-CD42c antibodies, respectively. (A) Timeline for H/R challenge, pyridoxamine treatment, surgery, in vivo imaging, and recording survival times. (B) Representative images. Time “0” was set as when image capture began on each vessel. White dotted lines indicate the vessel wall; large arrows indicate direction of blood flow. Different colored small arrows indicate individual rolling neutrophils over 20 seconds. Bar = 10 mm. (C and D) The number of rolling and adherent neutrophils. (E) Cumulative frequency of the rolling velocity of neutrophils. (F) Number of platelets adherent to neutrophils and EC was counted. Data represent the mean±standard deviation (SD) (n=36-45 venules in 6-7 mice per group). (G) Survival curves of SCD mice during or after intravital microscopy. *P<0.05, **P<0.01, and ***P<0.001 versus vehicle control, ANOVA with Tukey’s test (C and D) and Kruskal-Wallis test with post-hoc Dunn correction (E).

haematologica | 2020; 105(10)

2409

J. Li et al.

P=0.05

Figure 2. A single oral administration of pyridoxamine reduces intravascular cell-cell interactions in the cremaster venules of tumor necrosis factor (TNF)-α-challenged sickle cell disease (SCD) mice and improves their survival. SCD mice were treated with oral administration of vehicle or pyridoxamine (100 mg/kg BW) before intraperitoneal injection of TNF-α. Three hours later, intravital microscopy was performed as described in Figure 1. (A) Timeline for pyridoxamine treatment, TNF-α injection, imaging, and recording survival times in SCD mice. (B) Representative images. Neutrophils and platelets are shown in red and green, respectively. Time “0” was set as when image capture began on each vessel. Large arrows indicate direction of blood flow; different colored small arrows indicate individual rolling neutrophils over 30 seconds. Bar = 10 mm. (C and D) The number of rolling and adherent neutrophils. (E) Integrated median fluorescence intensity values of anti-CD42c antibodies (F platelets) were plotted over time. Data represent the mean±standard deviation (n=33-47 venules in 6 mice per group). (F) Survival curves of SCD mice during or after intravital microscopy. ****P<0.0001 versus vehicle control, Student’s t-test (C and D) and Mantel-Cox log-rank test (F).

dose (100 mg/kg BW) markedly enhanced the cumulative frequency of the rolling velocity of neutrophils (***P<0.001 vs. saline) (Figure 1E). Platelets adhered to both neutrophils and EC in the microvessels of H/R-challenged SCD mice, and pyridoxamine treatment did not affect the number of adherent platelets (Figure 1F). Most SCD mice in all groups survived for the duration of the study (Figure 1G). Since aged neutrophils, defined by upregulated CXCR4/αMβ2 and down-regulated L-selectin on the surface, positively contribute to inflammatory conditions in SCD,8 we tested whether pyridoxamine affects neutrophil aging in SCD mice. As assessed by CXCR4high and L-selectinlow in flow cytometry, the percentage of aged neutrophils was maximal around 3 pm with a 14-hour light and 10-hour dark cycle, and one oral administration of pyridoxamine (100 mg/kg BW) had no significant effect on neutrophil aging (Online Supplementary Figure S1). These results indicate that a single oral administration of pyridoxamine attenuates neutrophil-EC interactions in the microvessels of H/R-challenged SCD mice without affecting neutrophil aging.

One oral administration of pyridoxamine reduces interactions of neutrophils with endothelial cells and platelets in the venules of sickle cell disease mice challenged with tumor necrosis factor-α and improves survival We have demonstrated that blocking neutrophil-EC and neutrophil-platelet interactions mitigates acute VOC events in microvessels and improves the survival of 2410

TNF-α-challenged SCD mice.20,21,28 Compared to H/Rchallenge, TNF-α challenge in SCD mice induces a more severe inflammatory condition under which most neutrophils stably adhere to the vessel wall, supporting platelet adhesion.20 Thus, we use the TNF-α model to assess the effect of pyridoxamine on neutrophil rolling and adhesion and neutrophil-platelet interactions (Figure 2A). Compared to the vehicle control, a single oral administration of pyridoxamine (100 mg/kg BW) significantly increased the rolling influx of neutrophils and reduced the number of adherent neutrophils on the inflamed EC (Figure 2B-D and Online Supplementary Videos 5 and 6). Platelets mainly attached to adherent neutrophils but not inflamed EC in TNF-α-challenged SCD mice, and pyridoxamine treatment inhibited adhesion and accumulation of platelets on adherent neutrophils as assessed by the fluorescence intensity of an anti-CD42c antibody (Figure 2E). Importantly, survival times for TNF-α-challenged SCD mice were prolonged after a single oral administration of pyridoxamine, compared to the control (P=0.05) (Figure 2F). Median survival time for mice treated with vehicle and pyridoxamine was 4.7 and 6 h after TNF-α injection, respectively. No beneficial effects were observed when 30 mg/kg BW of pyridoxamine was administered orally (data not shown). These results suggest that a single oral administration of pyridoxamine effectively mitigates neutrophil-EC and neutrophil-platelet interactions in microvessels and improves the survival of SCD mice under severe inflammatory conditions. haematologica | 2020; 105(10)

Effect of pyridoxamine on VOC in SCD

Co-treatment with hydroxyurea enhances the inhibitory effect of pyridoxamine on neutrophil rolling and adhesion in the venules of tumor necrosis factor-α-challenged sickle cell disease mice We reported that intravenous and oral administration of HU and an AKT inhibitor, compared to each drug alone, has immediate benefits for acute VOC events and the survival of TNF-α-challenged SCD mice.20,28,29 Therefore, we sought to test whether co-administration of HU and pyridoxamine potentiates the inhibitory effect on intravascular cell-cell interactions in SCD. Since a single oral administration (250 mg/kg BW) of HU reduces leukocyte adhesion to the venular wall of SCD mice,28,30 we tested the combined effect of 250 mg/kg BW of HU and 100 mg/kg BW of pyridoxamine in TNF-α-challenged SCD mice (Figure 3A). Compared to vehicle or HU alone, pyridoxamine or the combination of both treatments increased the rolling influx of neutrophils on the venules of SCD mice (Figure 3B). In contrast, the rolling velocity of neutrophils was faster in the groups treated with HU alone or both HU and pyridoxamine than those treated with vehicle or pyridoxamine alone (Figure 3C). These results suggest that pyridoxamine and HU are likely to inhibit neutrophil rolling on EC through distinct mechanisms. Treatment with either HU or pyridoxamine significantly reduced neutrophil adhesion to the vessel wall, and the inhibitory effect was enhanced when both were administered (Figure 3D). We found that platelet-neutrophil interactions were abrogated by treatment with each drug or in combination, compared to the vehicle control (Figure 3E). Furthermore, compared to the vehicle control, treatment with HU, pyridoxamine, or both significantly improved the survival of TNF-α-challenged SCD mice (Figure 3F). Median survival time for mice treated with vehicle, HU, pyridoxamine, and both drugs was 4.4, 5.6, 5.8, and 6 h after TNF-α injection, respectively. The interactions between EC E-selectin and neutrophil receptors initiate neutrophil rolling, whereas binding of neutrophil β2 integrins, such as αLβ2 and αMβ2, to EC ICAM-1 results in stable neutrophil adhesion and crawling during vascular inflammation.31 Therefore, we determined whether HU and pyridoxamine affect the expression of these molecules on the vessel wall of cremaster muscles taken from the SCD mice after intravital microscopy. We found that compared to the vehicle control, short-term treatment with HU alone or both HU and pyridoxamine, but not pyridoxamine alone, significantly decreased the expression of E-selectin (Figure 3G-I). ICAM-1 expression was further reduced in SCD mice treated with both drugs, compared to HU or pyridoxamine alone. Since the increased levels of soluble EC-derived adhesion molecules, including selectins and ICAM-1, are associated with organ dysfunction and mortality in SCD patients,32 we also measured the plasma levels of soluble E-selectin and ICAM-1. Oral administration of pyridoxamine or both HU and pyridoxamine significantly reduced the level of soluble ICAM-1 but not E-selectin in SCD mice (Figure 3J and K). Treatment with HU alone had no inhibitory effect, as reported previously.33 In addition, we evaluated the plasma levels of inflammatory cytokines, IL-1β and IL-6. We found that HU or pyridoxamine alone, or the combination of the two markedly decreased the level of IL-1β but not IL-6 (Figure 3L and M), supporting the anti-inflammatory effect of each drug. These results suggest that the combination of pyridoxamine and HU has synergistic haematologica | 2020; 105(10)

effects in reducing neutrophil-EC interactions in TNF-αchallenged SCD mice.

Long-term treatment with pyridoxamine attenuates neutrophil-endothelial cell interactions in the venules of tumor necrosis factor-α-orhypoxia/ reoxygenation-challenged sickle cell disease mice To test the effect of long-term oral administration of pyridoxamine on VOC events, SCD mice were given acidic water or pyridoxamine supplemented in drinking water for 5.5 months, starting at 2 weeks after BMT (Figure 4A). Water consumption was equivalent between the two groups, and long-term treatment with pyridoxamine did not influence body weight, the number of RBC or leukocytes, hematocrit, or the amount of Hb (Online Supplementary Figure S2A-E). Although in a normal range, platelet counts were significantly lower in SCD mice drinking pyridoxamine-containing water, compared to the control, during the first 2-2.5 months (Online Supplementary Figure S2F), and there was no difference in numbers between the two groups thereafter. Also, there was no difference in urine osmolality between the two groups (Online Supplementary Figure S3). As assessed by intravital microscopy, compared to the control, long-term treatment with pyridoxamine significantly increased the rolling influx and velocity of neutrophils and decreased the number of adherent neutrophils on the inflamed venules in TNF-α-challenged SCD mice (Figure 4B-D). Platelet-neutrophil interactions were also slightly reduced with pyridoxamine treatment (Figure 4E). Compared to the control, long-term treatment with pyridoxamine improved the survival of TNFα-challenged SCD mice (P=0.063) (Figure 4F). The median survival time for mice treated with acidic water and pyridoxamine-containing water was 5.3 and 6 h after TNF-α injection, respectively. Endothelial dysfunction, activated coagulation, and proinflammatory conditions contribute to organ damage in SCD.1 To assess the preventive effect of long-term treatment with pyridoxamine on tissue damage, liver, spleen, and kidney were obtained from SCD mice after intravital microscopy. There was no noticeable organ damage in unchallenged or TNF-α-challenged SCD mice treated with acidic water or pyridoxamine-containing water for 5.5 months after BMT (Figure 4G and H). This may result from the age of SCD mice (6-7 months after BMT) since a recent paper shows that severe organ damage in SCD mice is detected 12 months after BMT.34 Nevertheless, TNF-α-enhanced leukocyte recruitment to the hepatic vessel wall was significantly reduced by long-term treatment of SCD mice with pyridoxamine (Figure 4I). When intravital microscopy was performed in H/Rchallenged SCD mice, long-term treatment with pyridoxamine exhibited a significant increase in the rolling influx of neutrophils and a decrease in neutrophil adhesion to the venule wall when compared to the control (Online Supplementary Figure S4A and B). However, there was no difference in the number of platelets adherent to neutrophils and EC or in the percentage of RBC sickling induced by H/R challenge between the two groups (Online Supplementary Figure S4C and D). These results suggest that long-term treatment with pyridoxamine, like short-term treatment, reduces neutrophil adhesiveness in TNF-α- or H/R-challenged SCD mice. 2411

J. Li et al.

Long-term treatment with pyridoxamine down-regulates the expression of E-selectin and ICAM-1 on endothelial cell and reduces the surface amount of αMβ2 integrin on activated neutrophils To determine whether long-term treatment with pyridoxamine affects the expression of EC E-selectin and ICAM-1 and their levels in circulation, immunohistochemistry was conducted using the cremaster muscle and an enzyme-linked immunosorbent assay was performed with the plasma from TNF-α-challenged SCD mice after intravital microscopy. We found that the expression of E-selectin and ICAM-1 was slightly but significantly down-regulated on the vascular endothelium in SCD mice

drinking pyridoxamine-containing water compared to those drinking the control water (Figure 5A-C). However, long-term treatment with pyridoxamine did not affect the levels of soluble E-selectin and ICAM-1 in SCD mice (Figure 5D and E). In addition, the plasma levels of IL-1β and IL-6 were not reduced by long-term treatment with pyridoxamine (Figure 5F and G). We further tested whether long-term treatment with pyridoxamine alters the activation state of neutrophils in SCD mice. Neutrophils were isolated from female SCD mice treated with control or pyridoxamine-containing water, followed by flow cytometry to assess activation markers, including membrane translocation of αMβ2 inte-

****P<0.0001 ****P<0.0001 *P<0.05

C P=0.2310

2412

P=0.043 P=0.026 P=0.032

K P=0.0991

haematologica | 2020; 105(10)

Effect of pyridoxamine on VOC in SCD

Figure 3 (previous page). Coadministration of hydroxyurea (HU) and pyridoxamine effectively inhibits intravascular cell-cell interactions in the venules and improves the survival of tumor necrosis factor (TNF)-α-challenged sickle cell disease (SCD) mice. Intravital microscopy with TNF-α-challenged SCD mice was performed as described in Figure 2. (A) Timeline for HU (250 mg/kg BW) and pyridoxamine (100 mg/kg BW) treatment, TNF-α injection, imaging, and recording survival times. (B) Number of rolling neutrophils. (C) Cumulative frequency of the rolling velocity of neutrophils. (D) Number of adherent neutrophils. (E) Integrated median fluorescence intensity values of anti-CD42c antibodies (F platelets) were plotted over time. (F) Survival curves of SCD mice during or after intravital microscopy. Data represent the mean±standard deviation (SD) (n=40-59 venules in 7-8 mice per group). (GI) Following intravital microscopy, the cremaster muscle was taken and fixed for immunohistochemistry. Sections of the muscle were stained for E-selectin, ICAM-1 and PECAM-1. Blue: DAPI. (G) Representative images. Bar = 10 mm. (H and I) Mean fluorescence intensity (MFI) values of E-selectin and ICAM-1 expression (n=10-19 venules in 3-4 mice per group). (J-M) Levels of E-selectin, ICAM-1, IL-1β, and IL-6 were measured in the plasma isolated from SCD mice after recording survival times. Data represent the mean±SD. *P<0.05, **P<0.01, ***P<0.001, and ****P<0.0001, ANOVA with Tukey’s test (B, D, H-M), Kruskal-Wallis test with post-hoc Dunn correction (C), and Mantel-Cox log-rank test (F).

grin, L-selectin shedding, and reactive oxygen species (ROS) production.6,35 Compared to the control group, pyridoxamine treatment significantly decreased the surface amount of αMβ2 integrin, a receptor required for the interaction of neutrophils with EC and platelets, after agonist stimulation, whereas it did not affect L-selectin shedding and ROS production (Figure 5H-J). These results imply that pyridoxamine attenuates degranulation of αMβ2 integrin from activated neutrophils, reducing cell adhesiveness.

Long-term treatment with pyridoxamine does not affect the plasma levels of advanced glycation end products and nitric oxide It is reported that pyridoxamine inhibits AGE formation and scavenges ROS.13,14 To measure the plasma levels of AGE, we isolated the plasma from TNF-α-challenged male SCD mice after intravital microscopy and from female SCD mice treated with control or pyridoxaminecontaining water. We found that pyridoxamine treatment did not affect the plasma AGE levels between male and female mice and no difference was observed between SCD and non-transplanted Berkeley mice (Hbb-/-, 6-7 months old) and between hemizygous control (Hbb+/-) and non-transplanted Berkeley mice (Figure 5K). These results suggest that the inhibitory effects of pyridoxamine are not caused by the inhibition of AGE formation in SCD mice. Enhanced oxidative stress reduces NO availability and is a key trigger to induce tissue damage in SCD patients.1 Thus, we measured the plasma level of nitrites/nitrates (NOx) in SCD mice after long-term treatment with pyridoxamine. Compared with the control mice, pyridoxamine-treated SCD mice did not enhance NOx levels (Figure 5L). We observed that the NOx levels in SCD mice were significantly higher than hemizygous control and Berkeley mice.20 These results suggest that the beneficial effects of pyridoxamine is not derived from scavenging ROS and imply that the result of NOx levels in SCD mice should be interpreted with caution. haematologica | 2020; 105(10)

Pyridoxamine reduces the activation state of platelets and neutrophils from sickle cell disease mice and patients in vitro To test whether pyridoxamine affects platelet adhesive function, we performed an in vitro aggregation assay with platelets isolated from SCD mice. Compared to the vehicle control, treatment of platelets with 1 mM pyridoxamine significantly inhibited aggregation and adenosine triphosphate (ATP) secretion induced by thrombin or a collagenrelated peptide (CRP), a glycoprotein VI-specific agonist (Figure 6A-D). Similar inhibitory effects were observed in platelets isolated from WT mice (Online Supplementary Figure S5A and B). As a control, vitamin B6 did not display any inhibitory effect (Online Supplementary Figure S5C and D). We found that compared to vehicle and vitamin B6, treatment of SCD mouse platelets with pyridoxamine inhibited αIIbβ3 integrin activation and ROS production without affecting P-selectin exposure following thrombin stimulation (Figure 6E-G). Similar results were obtained with WT mouse platelets (Online Supplementary Figure S6). To evaluate the clinical significance of the beneficial effect, we further tested pyridoxamine in platelets from SCD patients. Compared to the vehicle control, pretreatment with 1 mM pyridoxamine significantly inhibited P-selectin exposure and αIIbβ3 activation but not ROS generation in activated platelets (Figure 6H-J). These results suggest that pyridoxamine reduces the activation state and adhesive function of platelets in SCD. The different inhibitory effects in SCD mouse and human platelets, however, remain to be determined. We then tested whether pyridoxamine treatment affects neutrophil activation. Consistent with the results from the ex vivo studies (Figure 5H), pretreatment of SCD mouse neutrophils with 1 mM pyridoxamine reduced the surface amount of αMβ2 integrin after fMLP stimulation compared with vehicle and vitamin B6 controls (Figure 6K). Lselectin shedding on activated neutrophils was inhibited by both pyridoxamine and vitamin B6 as compared with the vehicle control (Figure 6L). However, ROS production during neutrophil activation was not affected by pyridox2413

J. Li et al.

amine (Figure 6M). In contrast, pretreatment of neutrophils isolated from SCD patients with pyridoxamine, compared to the vehicle control, reduced ROS generation but did not affect the surface amount of αMβ2 and L-selectin following agonist stimulation (Figure 6N-P). Although the different effect of pyridoxamine in mouse and human neutrophils should be further investigated,

these results suggest that pyridoxamine reduces neutrophil activation in SCD.

Discussion Intravascular cell-cell aggregation directly contribute to ****P<0.001

P=0.063

Figure 4. Long-term treatment with pyridoxamine impairs neutrophil adhesive function in venules of tumor necrosis factor (TNF)-α-challenged sickle cell disease (SCD) mice. Male (A-P and T-U) and female (Q-U) SCD mice were given acidic water or pyridoxamine (2 g/L) in the drinking water, starting at 2 weeks after bone marrow transplantation and continued for 5.5 months. Intravital microscopy with TNF-α-challenged SCD mice was performed as described in Figure 2. (A) Timeline for the treatment with acidic or pyridoxamine-containing water, TNF-α injection, imaging, and recording survival times. (B) Rolling influx of neutrophils. (C) Cumulative frequency of the rolling velocity of neutrophils. (D) Number of adherent neutrophils. (E) Integrated median fluorescence intensity values of anti-CD42c antibodies (F platelets) were plotted over time. (F) Survival curves of SCD mice during or after intravital microscopy. Data represent the mean±standard deviation (SD) (n=58-62 venules in 7-8 mice per group). (G-I) Kidney, spleen, and liver were obtained from unchallenged or TNF-α-challenged SCD mice treated with acidic water or pyridoxamine-containing water after recording the survival time. (G) Representative hematoxylin and eosin staining. Bar=50 mm. (H) Glomerular size. (I) Number of adherent leukocytes on the hepatic vessel wall. Data represent the mean±SD (n=7 mice per group). **P<0.01, ***P<0.001, and ****P<0.0001, Student’s t-test (B, D, H, and I), Mann-Whitney U test (C), and Mantel-Cox log-rank test (F).

2414

haematologica | 2020; 105(10)

Effect of pyridoxamine on VOC in SCD

VOC in SCD.20,25,36,37 In this study, we demonstrate that short- and long-term oral administration of pyridoxamine significantly reduces intravascular cell-cell interactions in the microvessels of H/R- or TNF-α-challenged SCD mice and improves survival. The beneficial effects are enhanced when pyridoxamine is co-administered with HU. Mechanistically, pyridoxamine reduces the activation

state of neutrophils and platelets from SCD mice and patients. These results suggest that pyridoxamine might be repurposed to prevent and treat acute VOC events in SCD. Nevertheless, we are aware that although statistically significant, the inhibitory effect of pyridoxamine itself on cell-cell interactions is weak to moderate after shortand long-term treatment in SCD mice. Therefore, its effi-

J P=0.09

P=0.1241

P=0.46

L P=0.260

P=0.17

Figure 5. Long-term treatment with pyridoxamine reduces the expression of E-selectin and ICAM-1 on endothelial cells (EC) and decreases the surface amount of neutrophil αMβ2 integrin. (A-C) Immunihistochemistry was performed as described in Figure 3. Bar = 10 μm. Data represent the mean±standard deviation (SD) (n=29-36 venules in 5 mice per group). (D-G) Levels of E-selectin, ICAM-1, IL-1β, and IL-6 were measured in the plasma from SCD mice after recording the survival time. (H-J) Neutrophils were isolated from female SCD mice treated with control or pyridoxamine-containing water. Surface amounts of αMβ2 integrin and L-selectin and ROS generation were assessed by flow cytometry. (K-L) Plasma was isolated from TNF-α-challenged male SCD mice after intravital microscopy and from female SCD mice treated with control or pyridoxamine-containing water. Hbb-/- : non-transplanted Berkeley mouse; Hbb+/- : hemizygous control. Plasma levels of advanced glycation end products (AGE) and nitrogen oxides (NOx) were measured as described in the Methods section. Data represent the mean±SD (n=4-9 mice per group). *P<0.05, **P<0.01, and ***P<0.001, Student’s t-test (B-G) and ANOVA with Tukey’s test (H-L).

haematologica | 2020; 105(10)

2415

J. Li et al.

2416

P=0.43

haematologica | 2020; 105(10)

Effect of pyridoxamine on VOC in SCD

P=0.307

(MFI)

Figure 6 (previous page). Pyridoxamine treatment inhibits the activation state of platelets and neutrophils isolated from sickle cell disease (SCD) mice and patients. Platelets isolated from SCD mice (A-G) and patients (H-J) were pretreated with 1 mM pyridoxamine, followed by stimulation with 0.05 U/mL thrombin or 3.5 mg/mL collagen related peptide. (A and C) Representative traces of platelet aggregation and quantitative graphs. (B and D) Adenosine triphosphate secretion was monitored with a luciferin/luciferase reagent and are shown as % of control at the end of the time point. (E-J) Flow cytometry was performed to assess P-selectin exposure, αIIbβ3 activation and reactive oxygen species (ROS) generation. Neutrophils isolated from SCD mice (K-M) and patients (N-P) were pretreated with 1 mM pyridoxamine, followed by incubation with fMLP. Flow cytometry was carried out to determine the surface amounts of αMβ2 and L-selectin and ROS production. Data represent the mean±standard deviation (n=3-5). *P<0.05 and **P<0.01 versus vehicle control, Student’s t-test.

cacy should be carefully tested in a future study with SCD patients. Although short-term pyridoxamine treatment significantly increases the rolling influx of neutrophils in H/R- or TNF-α-challenged SCD mice, the rolling velocity of neutrophils was faster in H/R- but not in TNF-α-challenged mice with pyridoxamine treatment (Figures 1 and 3). This discrepancy might result from the difference in severity of the inflammatory conditions between the two models. Since our in vivo studies used (transplanted) SCD mice that would be less inflammatory compared to (non-transplanted) Berkeley mice with lifelong inflammation, the TNF-α model may better represent the inflammatory condition in SCD. Furthermore, our finding that short-term treatment with pyridoxamine does not affect E-selectin expression but inhibits ICAM-1 expression on the TNF-α-inflamed venular wall (Figure 3G-I) implies that pyridoxamine treatment increases neutrophil rolling due to decreased adhesiveness on EC. It should be noted that co-administration of pyridoxamine and HU enhances the inhibitory effect of each drug on the adhesive function of neutrophils. HU exhibits its clinical benefit through multiple mechanisms, including inducing HbF production, lowering the number of circulating leukocytes and reticulocytes, increasing NO production, and reducing phosphatidylserine exposure.38,39 Given the synergistic effect, pyridoxamine is likely to act distinctly from HU without compromising the beneficial effect, providing evidence that pyridoxamine could be a supplement to HU therapy for patients with SCD. After oral administration, pyridoxamine is converted to other forms of vitamin B6, including pyridoxal phosphate,40,41 which has been reported to inhibit platelet function.42 However, there is a limit to the conversion of haematologica | 2020; 105(10)

pyridoxamine to other forms of vitamin B6. For example, when 40-140 nmol of pyridoxamine was given to mice, the major B6 vitamer in the blood (>40%) was pyridoxamine itself.40,41 Since the saturation doses used in the pharmacokinetic studies are significantly lower than the doses used in our in vivo study (10-100 mg/kg BW, approximately 1-10 mmol per 25 g mouse), the dose-dependent effects (Figure 1) are likely due to increasing concentrations of pyridoxamine rather than other vitamin B6 derivatives. Pyridoxamine can trap reactive carboxyl intermediates that lead to the formation of AGE, which contribute to the pathophysiology of retinal and renal lesions in diabetes.43 Previous studies demonstrated that pyridoxamine inhibits the progression of diabetic nephropathy in mice and rats.15,44,45 In support of the preclinical studies, a phase II study with patients with type 1/2 diabetic nephropathy reveals that compared to placebo, oral administration of pyridoxamine (50 or 250 mg) twice daily for 24 weeks significantly reduces the levels of serum creatinine and urinary TGF-β1.17 However, another study with patients with proteinuric type 2 diabetic nephropathy shows that pyridoxamine (150 or 300 mg twice daily) does not significantly decrease the concentration of serum creatinine after 1 year of therapy,18 making the beneficial effect of pyridoxamine in diabetic nephropathy controversial. Our intravital microscopy with SCD mice demonstrates that long-term treatment with pyridoxamine attenuates neutrophil and platelet adhesive functions and improves mice survival (Figure 4). Although the increased levels of plasma and skin AGE are reported to be associated with oxidative stress and organ complications in SCD patients,19,46 our ex vivo studies reveal there is no difference in plasma AGE 2417

J. Li et al.

levels between 6-month old hemizygous control and Berkeley mice, and that these remain unchanged after pyridoxamine treatment (Figure 5K). These results imply that the beneficial effects of pyridoxamine in SCD result from a mechanism distinct from inhibiting AGE formation. Furthermore, there were no signs of tissue damage in the liver, kidney, and spleen of SCD mice 6-7 months after BMT. Thus, further studies with older Berkeley and SCD mice are required to investigate the plasma level of AGE and organ damage, and to examine the effect of pyridoxamine. Previous studies showed that pyridoxamine inhibits platelet aggregation induced by ADP or both collagen and Isoketal.42,47,48 However, whether it affects platelet activation remains to be explored. Furthermore, the effect of pyridoxamine on neutrophil function remains unknown. Our studies have shown that pyridoxamine treatment partially impairs the activation state and/or adhesiveness of platelets and neutrophils from SCD mice and patients (Figure 6). Since platelet αIIbβ3 integrin plays a role in platelet-neutrophil interactions in vivo,21 decreased αIIbβ3 activation by pyridoxamine in both SCD mouse and human platelets could account for the decreased cell-cell interaction in TNF-α-challenged SCD mice (Figures 2-4). In addition, our finding that pyridoxamine treatment reduces the surface amount of neutrophil αMβ2 integrin in mouse neutrophils (Figures 5H and 6K),49 is consistent with the impaired platelet-neutrophil interactions in SCD mice (Figures 2-4). Our results suggest that pyridoxamine may block a specific signaling pathway(s) during activation of platelets and neutrophils in SCD. In support of this speculation, we found that pretreatment of platelets and neutrophils isolated from SCD mice with pyridoxamine inhibits phosphorylation of AKT following agonist stimulation (Online Supplementary Figure S7). Nevertheless, we cannot fully explain why pyridoxamine has a different effect on platelets and on neutrophils isolated from SCD patients and mice. It is noteworthy that unlike SCD patients, Berkeley mice do not have mural thrombi, large

References 1. Zhang D, Xu C, Manwani D, Frenette PS. Neutrophils, platelets, and inflammatory pathways at the nexus of sickle cell disease pathophysiology. Blood. 2016;127(7):801809. 2. Paul RN, Castro OL, Aggarwal A, Oneal PA. Acute chest syndrome: sickle cell disease. Eur J Haematol. 2011;87(3):191-207. 3. Rodgers GP, Dover GJ, Noguchi CT, Schechter AN, Nienhuis AW. Hematologic responses of patients with sickle cell disease to treatment with hydroxyurea. N Engl J Med. 1990;322(15):1037-1045. 4. Steinberg MH, Lu ZH, Barton FB, et al. Fetal hemoglobin in sickle cell anemia: determinants of response to hydroxyurea. Blood. 1997;89(3):1078-1088. 5. Niihara Y, Miller ST, Kanter J, et al. A Phase 3 trial of l-glutamine in sickle cell disease. N Engl J Med. 2018;379(3):226-235. 6. Li J, Kim K, Hahm E, et al. Neutrophil AKT2 regulates heterotypic cell-cell interactions during vascular inflammation. J Clin Invest. 2014;124(4):1483-1496. 7. Belcher JD, Chen C, Nguyen J, et al. The fucosylation inhibitor, 2-fluorofucose,

2418

8. 9.

10.

11.

12.

13.

vessel vasculopathy, cerebral infarcts, or hemorrhagic strokes.50 These results suggest that compared to patients, SCD mice are less inflammatory and their blood cells might be less activated. Therefore, the different activation state of blood cells might influence the pyridoxamine effect. Furthermore, a different source of neutrophils used in this study (blood neutrophils from patients vs. bone marrow neutrophils from mice) might be the reason for the discrepancy. Drugs targeting the activation and adhesion of leukocytes and platelets, coagulation, and inflammation are being tested in clinical studies on the prevention and treatment of VOC-mediated pain crises in SCD patients.1,51 In recent phase II studies, the leading drugs, rivipansel and crizanlizumab, were reported to reduce time to resolution of VOC events and requirement for opioid analgesia in SCD patients.9,10 However, complete inhibition of leukocyte-EC contacts might disrupt the innate immune response. For this reason, the phase II clinical trial of rivapansel excluded patients with fever.10 Our results demonstrate the inhibitory effects of short- and long-term treatment with pyridoxamine on the adhesive function of neutrophils and platelets in the microvessels of SCD mice. Since previous clinical studies have shown no or minimal adverse events of pyridoxamine in patients with diabetic nephropathy,17,18 our results suggest that further investigations into the use of pyridoxamine to prevent and treat acute VOC-mediated crises in SCD patients undergoing HU therapy are warranted. Funding This work was supported by (R01HL130028, R01HL148280, and R43HL142402) and the University of Illinois at Chicago Center for Clinical and Translational Science award UL1TR002003. JL is a recipient of American Heart Association Career Development Award. BX is a recipient of a scholarship from the China Scholarship Council. AT is a recipient of NIH Ruth L. Kirschstein National Research Service Award Individual Predoctoral Fellowship (F30HL134296).

inhibits vaso-occlusion, leukocyteendothelium interactions and NF-kB activation in transgenic sickle mice. PLoS One. 2015;10(2):e0117772. Zhang D, Chen G, Manwani D, et al. Neutrophil ageing is regulated by the microbiome. Nature. 2015;525(7570):528-532. Ataga KI, Kutlar A, Kanter J, et al. Crizanlizumab for the prevention of pain crises in sickle cell disease. N Engl J Med. 2017;376(5):429-439. Telen MJ, Wun T, McCavit TL, et al. Randomized phase 2 study of GMI-1070 in SCD: reduction in time to resolution of vaso-occlusive events and decreased opioid use. Blood. 2015;125(17):2656-2664. Heeney MM, Hoppe CC, Abboud MR, et al. A Multinational Trial of Prasugrel for sickle cell vaso-occlusive Events. N Engl J Med. 2016;374(7):625-635. Becker RC, Sexton T, Smyth SS. Translational Implications of Platelets as Vascular First Responders. Circ Res. 2018; 122(3):506-522. Onorato JM, Jenkins AJ, Thorpe SR, Baynes JW. Pyridoxamine, an inhibitor of advanced glycation reactions, also inhibits advanced lipoxidation reactions. Mechanism of action of pyridoxamine. J Biol Chem. 2000;

275(28):21177-21184. 14. Voziyan PA, Khalifah RG, Thibaudeau C, et al. Modification of proteins in vitro by physiological levels of glucose: pyridoxamine inhibits conversion of Amadori intermediate to advanced glycation end-products through binding of redox metal ions. J Biol Chem. 2003;278(47):46616-46624. 15. Degenhardt TP, Alderson NL, Arrington DD, et al. Pyridoxamine inhibits early renal disease and dyslipidemia in the streptozotocindiabetic rat. Kidney Int. 2002; 61(3):939-950. 16. Ozer A, Altuntas CZ, Izgi K, et al. Advanced glycation end products facilitate bacterial adherence in urinary tract infection in diabetic mice. Pathog Dis. 2015;73(5). 17. Williams ME, Bolton WK, Khalifah RG, et al. Effects of pyridoxamine in combined phase 2 studies of patients with type 1 and type 2 diabetes and overt nephropathy. Am J Nephrol. 2007;27(6):605-614. 18. Lewis EJ, Greene T, Spitalewiz S, et al. Pyridorin in type 2 diabetic nephropathy. J Am Soc Nephrol. 2012;23(1):131-136. 19. Nur E, Brandjes DP, Schnog JJ, et al. Plasma levels of advanced glycation end products are associated with haemolysis-related organ complications in sickle cell patients.

haematologica | 2020; 105(10)

Effect of pyridoxamine on VOC in SCD

Br J Haematol. 2010;151(1):62-69. 20. Barazia A, Li J, Kim K, Shabrani N, Cho J. Hydroxyurea with AKT2 inhibition decreases vaso-occlusive events in sickle cell disease mice. Blood. 2015;126(22): 2511-2517. 21. Li J, Kim K, Jeong SY, et al. Platelet protein disulfide isomerase promotes glycoprotein Ibα-mediated platelet-neutrophil interactions under thromboinflammatory conditions. Circulation. 2019;139(10):1300-1319. 22. Murakoshi M, Tanimoto M, Gohda T, et al. Pleiotropic effect of pyridoxamine on diabetic complications via CD36 expression in KK-Ay/Ta mice. Diabetes Res Clin Pract. 2009;83(2):183-189. 23. Sun K, Xia Y. New insights into sickle cell disease: a disease of hypoxia. Curr Opin Hematol. 2013;20(3):215-221. 24. Rubin EM, Witkowska HE, Spangler E, et al. Hypoxia-induced in vivo sickling of transgenic mouse red cells. J Clin Invest. 1991;87(2):639-647. 25. Hidalgo A, Chang J, Jang JE, et al. Heterotypic interactions enabled by polarized neutrophil microdomains mediate thromboinflammatory injury. Nat Med. 2009;15(4):384-391. 26. Bennewitz MF, Jimenez MA, Vats R, et al. Lung vaso-occlusion in sickle cell disease mediated by arteriolar neutrophil-platelet microemboli. JCI Insight. 2017;2(1): e89761. 27. Polanowska-Grabowska R, Wallace K, Field JJ, et al. P-selectin-mediated plateletneutrophil aggregate formation activates neutrophils in mouse and human sickle cell disease. Arterioscler Thromb Vasc Biol. 2010;30(12):2392-2399. 28. Kim K, Li J, Barazia A, et al. ARQ 092, an orally-available, selective AKT inhibitor, attenuates neutrophil-platelet interactions in sickle cell disease. Haematologica. 2017;102(2):246-259. 29. Li J, Cho J. Ser/Thr protein kinase BbetaNADPH oxidase 2 signaling in thromboinflammation. Curr Opin Hematol. 2017; 24(5):460-466. 30. Almeida CB, Scheiermann C, Jang JE, et al. Hydroxyurea and a cGMP-amplifying

haematologica | 2020; 105(10)

31. 32.

33.

34.

35.

36.

37.

38. 39.

40.

41.

agent have immediate benefits on acute vaso-occlusive events in sickle cell disease mice. Blood. 2012;120(14):2879-2888. Phillipson M, Kubes P. The neutrophil in vascular inflammation. Nat Med. 2011; 17(11):1381-1390. Kato GJ, Martyr S, Blackwelder WC, et al. Levels of soluble endothelium-derived adhesion molecules in patients with sickle cell disease are associated with pulmonary hypertension, organ dysfunction, and mortality. Br J Haematol. 2005;130(6):943-953. Saleh AW, Duits AJ, Gerbers A, de Vries C, Hillen HF. Cytokines and soluble adhesion molecules in sickle cell anemia patients during hydroxyurea therapy. Acta Haematol. 1998;100(1):26-31. Nasimuzzaman M, Arumugam PI, Mullins ES, et al. Elimination of the fibrinogen integrin αMβ2-binding motif improves renal pathology in mice with sickle cell anemia. Blood Adv. 2019;3(9):1519-1532. Kim K, Li J, Tseng A, Andrews RK, Cho J. NOX2 is critical for heterotypic neutrophilplatelet interactions during vascular inflammation. Blood. 2015;126(16):1952-1964. Jimenez MA, Novelli E, Shaw GD, Sundd P. Glycoprotein Ibalpha inhibitor (CCP-224) prevents neutrophil-platelet aggregation in sickle cell disease. Blood Adv. 2017; 1(20):1712-1716. Li J, Kim K, Barazia A, Tseng A, Cho J. Platelet-neutrophil interactions under thromboinflammatory conditions. Cell Mol Life Sci. 2015;72(14):2627-2643. Singh PC, Ballas SK. Emerging drugs for sickle cell anemia. Expert Opin Emerg Drugs. 2015;20(1):47-61. Piccin A, Murphy C, Eakins E, et al. Insight into the complex pathophysiology of sickle cell anaemia and possible treatment. Eur J Haematol. 2019;102(4):319-330. Sakurai T, Asakura T, Mizuno A, Matsuda M. Absorption and metabolism of pyridoxamine in mice. I. Pyridoxal as the only form of transport in blood. J Nutr Sci Vitaminol (Tokyo). 1991;37(4):341-348. Sakurai T, Asakura T, Mizuno A, Matsuda M. Absorption and metabolism of pyridoxamine in mice. II. Transformation of pyri-

42.

43. 44.

45.

46.

47.

48.

49.

50.

51.

doxamine to pyridoxal in intestinal tissues. J Nutr Sci Vitaminol (Tokyo). 1992; 38(3):227-233. Subbardo K, Kuchibhotla J, Kakkar VV. Pyridoxal 5'-phosphate--a new physiological inhibitor of blood coagulation and platelet function. Biochem Pharmacol. 1979;28(4):531-534. Brownlee M. Advanced protein glycosylation in diabetes and aging. Annu Rev Med. 1995;46:223-234. Alderson NL, Chachich ME, Frizzell N, et al. Effect of antioxidants and ACE inhibition on chemical modification of proteins and progression of nephropathy in the streptozotocin diabetic rat. Diabetologia. 2004;47(8):1385-1395. Zheng F, Zeng YJ, Plati AR, et al. Combined AGE inhibition and ACEi decreases the progression of established diabetic nephropathy in B6 db/db mice. Kidney Int. 2006;70(3):507-514. Kashyap L, Alsaheel A, Ranck M, et al. Sickle cell disease is associated with elevated levels of skin advanced glycation endproducts. J Pediatr Hematol Oncol. 2018; 40(4):285-289. Chang SJ, Chang CN, Chen CW. Occupancy of glycoprotein IIb/IIIa by B-6 vitamers inhibits human platelet aggregation. J Nutr. 2002;132(12):3603-3606. Bernoud-Hubac N, Alam DA, Lefils J, et al. Low concentrations of reactive gammaketoaldehydes prime thromboxane-dependent human platelet aggregation via p38MAPK activation. Biochim Biophys Acta. 2009;1791(4):307-313. Wang Y, Sakuma M, Chen Z, et al. Leukocyte engagement of platelet glycoprotein Ibalpha via the integrin Mac-1 is critical for the biological response to vascular injury. Circulation. 2005;112(19):29933000. Manci EA, Hillery CA, Bodian CA, et al. Pathology of Berkeley sickle cell mice: similarities and differences with human sickle cell disease. Blood. 2006;107(4):1651-1658. Tran H, Gupta M, Gupta K. Targeting novel mechanisms of pain in sickle cell disease. Blood. 2017;130(22):2377-2385.

2419

ARTICLE Ferrata Storti Foundation

Haematologica 2020 Volume 105(10):2420-2431

Hematologic Neoplasms

The new small tyrosine kinase inhibitor ARQ531 targets acute myeloid leukemia cells by disrupting multiple tumor-addicted programs

Debora Soncini,1 Stefania Orecchioni,2 Samantha Ruberti,1 Paola Minetto,1,3 Claudia Martinuzzi,1 Luca Agnelli,4 Katia Todoerti,4 Antonia Cagnetta,1,3 Maurizio Miglino,1,3 Marino Clavio,1,3 Paola Contini,5 Riccardo Varaldo,6 Micaela Bergamaschi,1 Fabio Guolo,1 Mario Passalacqua,7 Alessio Nencioni,5 Fiammetta Monacelli,5 Marco Gobbi,1,3 Antonino Neri,4 Giovanni Abbadessa,8 Sudharshan Eathiraj,8 Brian Schwartz,8 Francesco Bertolini,2 Roberto M. Lemoli1,3 and Michele Cea1,3

Chair of Hematology, Department of Internal Medicine and Specialities (DiMI), University of Genoa, Genoa, Italy; 2Laboratory of Hematology-Oncology, European Institute of Oncology IRCCS, Milan, Italy; 3IRCCS Ospedale Policlinico San Martino, Genoa, Italy; 4Department of Oncology and Hemato-Oncology, University of Milan, Milan, Italy; 5Department of Internal Medicine and Specialities (DiMI), University of Genoa, Genoa, Italy; 6Division of Hematology and Hematopoietic Stem Cell Transplantation Unit, Ospedale Policlinico San Martino, Genoa, Italy; 7Department of Experimental Medicine (DIMES), University of Genoa, Genoa, Italy and 8ArQule, Burlington, MA, USA 1

ABSTRACT

T Correspondence: MICHELE CEA michele.cea@unige.it Received: April 23, 2019. Accepted: October 10, 2019. Pre-published: November 7, 2019. doi:10.3324/haematol.2019.224956 ©2020 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.

2420

yrosine kinases have been implicated in promoting tumorigenesis of several human cancers. Exploiting these vulnerabilities has been shown to be an effective anti-tumor strategy as demonstrated for example by the Bruton tyrosine kinase (BTK) inhibitor, ibrutinib, for treatment of various blood cancers. Here we characterize a new multiple kinase inhibitor, ARQ531, and evaluate its mechanism of action in preclinical models of acute myeloid leukemia. Treatment with ARQ531, by producing global signaling pathway deregulation, resulted in impaired cell cycle progression and survival in a large panel of leukemia cell lines and patient-derived tumor cells, regardless of the specific genetic background and/or the presence of bone marrow stromal cells. RNA-sequencing analysis revealed that ARQ531 constrained tumor cell proliferation and survival through BTK and transcriptional program dysregulation, with proteasome-mediated MYB degradation and depletion of shortlived proteins that are crucial for tumor growth and survival, including ERK, MYC and MCL1. Finally, ARQ531 treatment was effective in a patient-derived leukemia mouse model, causing significant impairment of tumor progression and survival, at tolerated doses. These data justify the clinical development of ARQ531 as a promising targeted agent for the treatment of patients with acute myeloid leukemia.

Introduction Acute myeloid leukemia (AML) is an aggressive disease characterized by uncontrolled clonal proliferation of abnormal myeloid progenitor cells in the bone marrow and blood. Despite recent advances in its treatment, as many as 70% of patients aged 65 or older will die within 1 year of diagnosis. The efficacy of standard high-dose chemotherapy and stem cell transplantation is limited by treatment-related morbidity and mortality, especially in elderly patients.1-3 Cancer treatment is undergoing a significant revolution from “one-size-fits-all” cytotoxic therapies to tailored approaches that target molecular alterations precisely. Notably, precision medicine, by linking specific genetic anomalies of tumors with available targeted therapies, is emerging as an innovative approach for AML treatment, with development of breakthrough drugs targeting specific molecular features (e.g., haematologica | 2020; 105(10)

Preclinical activity of ARQ531 in AML

FLT3 and IDH1/2 inhibitors).4-6 However, identification of patients who will benefit from targeted therapies is more complex than simply identifying patients whose tumors harbor the targeted aberration. A rational combination of therapeutic agents may prevent the development of resistance to therapy, with molecular strategies aimed at targeting multiple pathways resulting in a more effective treatment across cancer subtypes. The Bruton tyrosine kinase (BTK), a member of the TEC family kinases, is a critical terminal kinase enzyme in the B-cell antigen receptor signaling pathway.7,8 Its activation leads to BTK phosphorylation which in turn results in downstream events such as proliferation, immune function alteration and survival through multiple signaling cascades.9 Chronic activation of BTK-mediated signaling represents a key driver for a number of types of cancers,10-14 including AML.15-22 Therefore, new inhibitors are needed to target tyrosine kinases better in these patients. Recent studies have shown that oncogenic cellular dysregulation is critical for the activity of the anti-BTK targeting agent ibrutinib,23,24 and that co-treatment with BET protein bromodomain antagonists or BCL-2 inhibitors may enhance the efficacy of ibrutinib in tumor cells.25,26 Herein we characterize ARQ531, a reversible small molecule inhibitor of BTK and several additional kinases, in preclinical models of AML. We provide evidence that ARQ531 greatly compromises survival of AML cells by inducing a “one shot” inhibition of multiple oncogenic transcriptional pathways. This resulted in potent antiAML activity in a patient-derived xenograft AML mouse model, providing the rationale for future clinical trials.

of human cells in NSG mice was evaluated using the following anti-human antibodies: anti-CD117-PeCy7 (IMMU 103.44), anti-CD45-APC (J.33), anti-CD34-APCCy7 (D3HL60.251) from Beckman-Coulter (Irving, TX, USA) and anti-mouse CD45-PE (30-F11) from BD Biosciences to exclude murine cell contamination. Cell suspensions were evaluated by a three-laser, ten-color flow cytometer (Navios, Beckman Coulter, Brea, CA, USA) using analysis gates designed to exclude dead cells, platelets, and debris. Percentages of stained cells were determined and compared to appropriate negative controls. Seven-aminoactinomycin D (7AAD) from SigmaAldrich was used to enumerate viable, apoptotic, and dead cells.

Statistical analyses All in vitro experiments were repeated at least three times and performed in triplicate; a representative experiment is shown in each figure. All data are shown as mean ± standard deviation (SD). The Student t test was applied to compare two experimental groups using Graph-Pad Prism software (http://www.graphpad.com). The minimal level of significance was specified as P<0.05. Survival analysis was performed by the Kaplan-Meier method, and the log-rank test was used to compare survival differences. Drug interactions were assessed by CalcuSyn 2 software (Biosoft), which is based on the Chou-Talalay method. A combination index (CI) of 1 indicates an additive effect; a CI<1 indicates synergism and a CI>1 indicates antagonism.

Results Methods Reagents ARQ531 was provided by ArQule, Inc (Burlington, MA, USA). The compound was dissolved in dimethylsulfoxide (Sigma-Aldrich) and stored at 10 mM at -80°C for experiments. Ibrutinib, daunorubicin, cytarabine and MG132 were purchased from Selleck Chemicals LLC (Houston, TX, USA). ZVAD-FMK was purchased from Promega (catalog n. G7232).

Patient-derived xenograft acute myeloid leukemia cells Experiments were carried out on 6- to 8-week old, nonobese diabetic severe combined immunodeficient (NOD/SCID) interleukin-2 receptor γ (IL-2Rg)-null (NSG) mice. The NSG mice were bred and housed under pathogen-free conditions in the animal facilities at the European Institute of Oncology–Italian Foundation for Cancer Research Institute of Molecular Oncology (IEOIFOM, Milan, Italy). All animal experiments were carried out in strict accordance with Italian laws (Legislative Decree 26/2014 and subsequent amendments) and were approved by the institutional committee. NSG mice were engrafted with 300,000 primary human AML cells (M4, acute myelomonocytic leukemia with wild-type FLT3). On day 19 after introduction of the AML cells, once a systemic xenograft had been confirmed, mice were randomized into three groups: vehicle-treatment group (n=5), low-dose ARQ531 treatment group (25 mg/kg; n=5) and high-dose ARQ531 treatment group (37.5 mg/kg; n=5) and the percentage of human leukemic cells in peripheral blood was measured weekly until day 42. The phenotype haematologica | 2020; 105(10)

ARQ531 shows strong anti-acute myeloid leukemia activity but preserves normal hematopoietic stem cells In line with previously reported data,15,16 we observed that BTK is frequently dysregulated in AML, with mRNA levels being significantly higher than in other cancer types (Online Supplementary Figure S1). To confirm its relative abundance, we screened a representative panel of human AML cell lines and primary blasts for BTK expression and activity by western blot (Figure 1A). Protein was detectable in all AML-screened cells (15/15) and, more importantly, independent of specific mutational profiling. Similarly, BTK activity (measured by Y223 phosphorylation) was observed in FLT3 wild-type and FLT3 mutated cells as well. An analogous investigation was applied to a larger cohort of AML patients derived from The Cancer Genome Atlas database, which showed uniform expression of BTK transcript in different AML subtypes. Overall, these data, by confirming the presence of BTK in AML, support targeting this kinase in this hematologic malignancy, as previously reported.14,15 ARQ531 is a recently described, reversible BTK inhibitor with promising activity in mouse models of chronic lymphocytic leukemia and lymphomas.27 Based on constitutively active BTK levels observed in AML cells, we evaluated the therapeutic activity of ARQ531 on these cells, using ibrutinib as a control. In vitro efficacy screening was performed on cultured (n=8) and primary (n=13) AML cells, comparing the efficacy of both drugs. As shown in Figure 1B, exposure to ARQ531 reduced in vitro viability more than ibrutinib did (Figure 1C). Analysis of the half maximal inhibitory concentration (IC50) at 48 h 2421

D. Soncini et al.

after treatment showed that the cells were more sensitive to ARQ531 than to ibrutinib, which exhibited 10-fold lower activity. (Figure 1D) A significant anti-AML effect of ARQ531 was also observed on blasts from AML patients (n=13) regardless of mutational status, European LeukemiaNet risk, and surface expression of CD117 (Figure 1E, Table 1). Consistent with these data, a dosedependent increase in the percentage of apoptotic and dead cells measured by annexin V and propidium iodide staining was also observed after ARQ531 treatment, together with several apoptotic features including caspase 3 and poly(AD-ribose) polymerase cleavage as well as reduction of anti-apoptotic MCL-1 and BCL-2 protein expression (Figure 1F, G and Online Supplementary Figure S2A). Viability was completely restored by pre-incubation with the pan-caspase inhibitor Z-VAD (Online Supplementary Figure S2B) In contrast, ibrutinib treatment resulted in weaker effects on apoptosis, thus suggesting that ARQ531 is more effective than ibrutinib, probably because it induces downregulation of additional survival mechanisms. It is well known that the bone marrow microenvironment has a role in the promotion of tumor growth, survival and drug-resistance.28 We therefore treated AML cells in the presence of normal or leukemic mesenchymal stromal cells. As expected, normal and AML stroma both protected tumor cells from spontaneous apoptosis; however, the efficacy of ARQ531 was preserved, with no significant effect on the viability of mesenchymal stromal cells (data not shown). Indeed, compared to spontaneous apoptosis of blast cells, ARQ531 increased cell death in AML cells cultured alone, and preserved its activity in the presence of normal or AML mesenchymal stromal cells, suggesting that ARQ531 abrogates the survival benefit from stromal cells (Figure 2A). Overall, our data indicate that ARQ531 is a potent anti-AML drug even in the presence of a tumorsupportive microenvironment, and irrespective of FLT3 mutational status.21 Finally, ARQ531 activity on normal cells was also investigated by employing clonogenic and viability assays in order to measure the impact of treatment on both CD34+ cells and mononuclear cells isolated from the bone marrow and peripheral blood of healthy donors. As shown in Figure 2B-D, all of these cells were largely unaffected by exposure to ARQ531 at dose levels toxic to tumor cells, proving that ARQ531 targets cancer cells without off-target effects on hematopoietic stem cells, resulting in a favorable therapeutic index.

BTK signaling inhibition partially contributes to the anti-leukemic activity of ARQ531 Based on its reported activity, we first studied the effect of ARQ531 on BTK signaling by analyzing tumor cell migration.27,29-31 A transwell assay system was employed to investigate the role of the stromal cell-derived factor 1 (SDF-1)/CXCR4 axis in the anti-AML activity of ARQ531. As shown in Online Supplementary Figure S3, ARQ531 reduced tumor cell migration in response to SDF-1 by 66% (P=0.001), suggesting similar activity to ibrutinib (71% reduction; P<0.001). Next, to confirm the role of BTK in the anti-AML activity of ARQ531, we investigated its effect on BTK signaling in AML cells over a range of concentrations. As shown in Figure 2E, ARQ531 treatment completely abrogated the activity of BTK, as measured by Y223 phosphorylation, similar to the effects seen with ibrutinib treatment. 2422

However, as shown in Figure 1B, ARQ531 has anti-AML activity even on cells expressing low levels of BTK, suggesting that BTK targeting might not be critical for the activity of ARQ531. To corroborate this hypothesis, we treated BTK-silenced (BTK knocked down, BTK-KD) AML cells with increasing doses of ARQ531. ARQ531 treatment reduced the viability of both BTK-KD and BTK wild-type cells to about 50% of control, demonstrating the importance of alternative targets for ARQ531 activity in AML (Figure 2F).

ARQ531 treatment suppresses transcriptional oncogenic activity in acute myeloid leukemia cells To identify ARQ531-induced global perturbations in transcriptional profiling, we generated RNA-sequencing data and performed functional annotation analysis of drug- versus dimethylsulfoxide-treated AML cells. As shown in Online Supplementary Figure S4A, principal component analysis segregated samples based on treatment, suggesting a coherent transcriptional result rather than global, non-specific transcriptional silencing in response to this drug. Indeed, analysis of differential expression identified 377 and 852 genes that were significantly upregulated and downregulated, respectively, with a ratio greater than 2-fold and P<0.05. (Figure 3A, B) As a measure of the specificity of this effect, gene set enrichment analysis was performed on the entire set of signatures available from the Molecular Signatures Database (MSigDB). Biological modules associated with oncogenic transcriptional programs (e.g., ribosomal biogenesis and assembly, unfolded protein response stress and MYC) were significantly enriched in ARQ531-suppressed genes (Figure 3C and Online Supplementary Figure S4B). In line with these findings, although treatment did not exert significant suppression of gene sets for factors linked to the pathophysiology of AML, such as C/EBPÎą-Î˛, RUNX1, PU.1, ERG and FLI1, a significant reverse correlation was observed for transcriptional signatures of MYC-upregulated target genes, which in turn reflects the selective suppression of its transcriptional networks. (Figure 3D) Indeed, reverse transcriptase polymerase chain reaction analysis of MYC and its target, CDC2, showed consistent downregulation following short-term exposure to the drug (Online Supplementary Figure S4C), pointing to ARQ531 as a selective suppressor of the MYC-regulated transcriptional pathway. To further support these data, we used the ARQ531 expression signature to query the Library of Integrated Network-Based Cellular Signatures (LINCS) Program (www.lincscloud.org, web interface available at http://amp.pharm.mssm.edu/L1000CDS2/#/index). As shown in Figure 3E, the most significant ARQ531-correlated signatures were those of oncogenic transcription factor inhibitors (such as fluvastatin, gefitinib and histone deacetylase inhibitors) as well as those related to knockdown of ribosome subunits and translation initiation factors. Together, these data indicate that ARQ531 inhibits oncogenic transcriptional pathways in AML cells.

ARQ531 interferes with the pro-survival MAPK pathway in OCI-AML3 cells As already reported, ARQ531 is a potent, ATP-competitive, reversible inhibitor of BTK and several additional kinases important to the viability, proliferation, activation, and motility of tumor cells.27 Among the most intriguing additional targets of ARQ531 are RAF1 and haematologica | 2020; 105(10)

Preclinical activity of ARQ531 in AML

haematologica | 2020; 105(10)

Figure 1. ARQ531 shows strong anti-tumor activity by inducing apoptosis of acute myeloid leukemia cells. (A) Immunoblot for phosphoBTK, BTK and GAPDH (loading control) in the indicated human acute myeloid leukemia (AML) cell lines and primary AML samples regardless of the specific genomic landscape. (B, C) Viability of AML cell lines after treatment with ARQ531 (B) or ibrutinib (C), as measured by MTS assay. The mean Âą standard deviation (SD) from at least three independent experiments is shown. (D) Half maximal inhibitory concentration (IC50) values, measured for each tested cell line as in (B) and (C). (E) Drug effects on primary AML patient-derived samples (n=13) treated with increasing doses of ARQ531 or ibrutinib (0-30 mM for 48 h). IC50 values are visualized for each tested primary AML cell line. (F) HL60, OCI-AML2 and primary AML-002 cells were treated with ARQ531 or dimethylsulfoxide (CTR) in a dosedependent manner for 48 h. Apoptotic cells were detected by annexin V/propidium iodide staining. Representative dot plots are shown. (G) Immunoblots for PARP, caspase 3, MCL-1, BCL-2 and tubulin on indicated AML cell lines and primary blast cells following treatment with a BTK inhibitor (ARQ531 versus ibrutinib) at 24 h.

2423

D. Soncini et al.

MEK1, constituents of the ERK signaling pathway that is frequently dysregulated in tumor cells.32-34 To confirm this activity in AML, cells were treated with increasing doses of ARQ531. As expected, activation of AKT and ERK was inhibited in a dose-dependent manner, likely due to predicted inhibition of RAF1 and MEK1 (Figure 4A). Subsequent experiments confirmed this hypothesis by revealing specific impairment of these kinases after ARQ531 treatment (Figure 4B). To support the pivotal role played by ERK, we exposed cells to the mitogenic effects of a higher serum concentration (20%). As shown in Online Supplementary Figure S5A, B, this strategy resulted in enhanced phosphorylation of ERK, which rescued the anti-AML activity of the drug, thus providing evidence of the relevance of ERK in the observed anti-tumor effect. Moreover, consistent with RNA-sequencing analysis, drug exposure resulted in prominent and specific downregulation of the oncogenic transcription factor MYC at the protein level (Figure 4B). Since the MAPK pathway enhances MYC protein stability by inducing its phosphorylation at serine 62,35 we assessed p-MYC S62

Patient AML-013 AML-012 AML-011 AML-009 AML-007 AML-006 AML-005 AML-004 AML-003 AML-001 AML-008 AML-010 AML-002

NPM FLT3-ITD ELN risk group

FAB Karyotype classification AML M2 AML M1 AML M6 AML M2 AML M3 AML M4 AML M2 AML M4 AML M2 AML M3 AML M3 AML M4 AML M2

normal normal complex 50XX t(15;17) normal normal normal normal t(15;17) t(15;17) n.a normal

mut mut wt wt / mut mut mut wt / / wt mut

wt mut wt wt wt mut mut wt wt mut wt wt wt

low int. high high M3 low low low int M3 M3 int. high

FAB: French-American-British; ITD: internal tandem duplication; ELN: European LeukemiaNet; AML: acute myeloid leukemia; mut: mutated; wt: wild-type; int: intermediate; n.a.; not available

Table 1. Characteristics of the patients with acute myeloid leukemia.

Figure 2. ARQ531 triggers anti-acute myeloid leukemia toxicity regardless of BTK activity and presence of stromal cells but preserves normal hematopoietic stem cells. (A) Viability of OCI-AML2 GFP/luc+ cells treated with ARQ531 for 48 h, alone and in the presence of normal mesenchymal stem cells (MSC) (blank) or acute myeloid leukemia (AML)-MSC (gray) stroma, measured by a luciferase-based luminescence assay. Data are represented as mean ± standard deviation (SD) in all histograms (n=3). 0.02<*P<0.03; **P<0.05. (B-D) Healthy donor (HD)-derived hematopoietic precursor (BM-CD34+) and peripheral blood mononuclear cells (PBMC) were exposed to increasing doses of ARQ531, and clonogenic abilities (C) or viability (B, D) were calculated. Colony formation of ARQ531-treated cells (CFC) was measured after 2 weeks. Viability was calculated as propidium iodide (PI)-negative cells among the CD34+ population. Data are represented as mean ± SD (n=3); unpaired t test, ***P<0.001, ****P<0.0001. (E) Western blot showing that ARQ531 treatment effectively abrogates the BTK signaling cascade in three different human AML cell lines (HL60, OCI-AML3 and MOLM14) following 24 h of treatment. The effect of ibrutinib is also shown as a positive control. (F) Viability of BTK-silenced (nucleofected with specific siRNA targeting BTK) or control HL60 cells (siRNA scramble) treated with increasing doses of ARQ531 for 48 h. Data are data represented as mean ± SD in all (n=3).

2424

haematologica | 2020; 105(10)

Preclinical activity of ARQ531 in AML

haematologica | 2020; 105(10)

Figure 3. Molecular perturbation triggered by ARQ531 in acute myeloid leukemia cells. (A) Heatmap for the highest 50 down- and up-regulated genes (P<0.0001) following ARQ531 treatment of OCI-AML3 cells. (B) Volcano plot of RNA-sequencing of drug- versus dimethylsulfoxide-treated cells in OCI-AML3 showing that 377 and 852 genes were significantly up- and down-regulated, respectively, with a fold change (FC) >2. (C) Table of the ten most significantly enriched gene sets, from the Hallmark collection, enriched with genes downregulated by ARQ531 in acute myeloid leukemia (AML) cells. Number of genes in each set (n), the normalized enrichment score (NES), and the test of statistical significance of the false discovery rate (FDR) q value are highlighted. (D) Enrichment plots of the top four most significantly enriched gene sets (MYC-related) in transcriptional profiles of AML cells treated (right) or untreated (left) with ARQ531. (E) Connectivity score generated by the LINCS L1000 Characteristic Direction Signature Search Engine tool, which compared the ARQ531-derived transcriptional profile against 10,000 â&#x20AC;&#x153;perturbagenâ&#x20AC;? signatures (corresponding to shorthairpin RNA, open reading frame and compounds). Top-ranked scores of relevant results are indicated by arrows. HDACi: histone deacetylase inhibitor.

2425

D. Soncini et al. A

Figure 4. BTK inhibition and MYC/MYB degradation represent molecular bases for the anti-leukemic activity of ARQ531. (A) Western blot showing that 24 h of treatment with ARQ531 (0.3-1 mM) abrogates ERK and AKT activation in HL60 and OCI-AML2 cells. (B) Western blot analysis shows that 24 h of ARQ531 treatment (0.3-1 mM) affects kinases in the RAF/MEK/ERK pathway of acute myeloid leukemia (AML) cell lines, resulting in MYC downregulation. (C) Western blot showing time-dependent effects of ARQ531 exposure on p-MYC S62 and total MYC in the HL60 cell line. (D) Western blot showing deregulation of c-MYCcontrolled signals in AML cells following treatment with ARQ531 at the indicated doses after 24 h. (E) Viability of MYC-silenced or control HL60 cells treated with increasing doses of ARQ531 for 48 h. The mean Âą standard deviation (SD) are shown (n=3). (F) Protein and mRNA expression in AML cells after 24 h treatment with dimethylsulfoxide (DMSO) or the indicated ARQ531 concentrations, normalized to DMSO controls. Bars and error bars are means and SD of three independent experiments. *P<0.05; **P=0.01; ***P<0.001; n.s. not significant (relative to DMSO controls), one sample t-test. Western blots below graphs show examples of MYB protein expression. (G) Western blot analysis of MYB protein expression in AML cells after 24 h treatment with DMSO, ARQ531 (0.3-1 mM) or ARQ531 and 10 mM MG132, a proteasome inhibitor.

changes in AML-treated cells. As shown in Figure 4C, ARQ531 exposure resulted in a prompt decrease of phosphorylation, followed by a reduction of MYC protein. Accordingly, numerous MYC-addicted oncogenic cellular pathways, such as protein folding machinery, metabolic dependency and genome integrity, were compromised following this treatment, as highlighted by phosphoeukaryotic translation initiation factor 4E (eIF4e), ASCT2 and GLUT1 downregulation and ÎłH2AX enhancement, respectively (Figure 4D). Combined drug screening revealed synergistic activity of ARQ531 with compounds affecting these programs, such as DNA damaging agents (Online Supplementary Figure S6A, B). Overall, these data support the existence of a mechanism of action that begins with MAPK signaling dysregulation and results in ARQ531-induced cytotoxicity in AML cells. Among MYC-controlled programs, protein synthesis is emerging as the limiting step for tumor cell growth,36 so we focused on this pathway. As shown in Figure 4D, AML cells treat2426

ed with ARQ531 showed marked increases of eukaryotic translation initiation factor 4E-binding protein (4EBP1) with concomitant de-phosphorylation of p70 ribosomal S6 kinase (p70-S6K) and eIF4e which result in blocking of mRNA recruitment to ribosomes for protein translation.37 These data suggest that ARQ531 is a modulator of several hubs controlling translation initiation in AML cells, providing evidence of marked protein synthesis inhibition specifically triggered by this treatment. Although MYC activation resulting from multiple tumor-driven genetic aberrations has been recognized as a major factor of leukemogenesis, its targeting did not show significant clinical benefit in AML. Thus, by using a small interference RNA (siRNA) strategy, we investigated the role of MYC in the anti-leukemic activity of ARQ531. As shown in Figure 4E, MYC-silenced HL-60 cells (MYC knocked down) were treated with increasing doses of ARQ531. Surprisingly, despite their sensitivity to this treatment, the cells were quite resistant to the loss of haematologica | 2020; 105(10)

Preclinical activity of ARQ531 in AML

Figure 5. ARQ531 treatment results in oncogenic program dysregulation in acute myeloid leukemia cells. (A) Western blot analysis of cells treated with ARQ531 (1 mM) after the indicated hours. Time-dependent effects demonstrate early inhibition of BTK activity and MYC downregulation followed by a reduction of MYB with associated PARP and caspase 3 cleavage. (B) Treatment of HL60 cells with BET bromodomain inhibitor JQ1 (400 nM), ibrutinib (30 mM) or their combination, which resulted in a synergistic effect. Bars and error bars are means and standard deviation of three independent experiments. ***P<0.001 (relative to dimethylsulfoxide controls), one-sample t-test. (C) Western blot showing that ibrutinib, JQ1, and their combination result in appearance of apoptotic features, including caspase 3 and PARP cleavage in HL60 cells.

MYC protein expression, indicating that additional targets are implicated in anti-AML activity of this small molecule. Modulation of transcriptional regulatory machinery is an innovative strategy to treat AML.38,39 The oncogenic driver MYB, which is essential in hematopoiesis, is now emerging as a new target for anti-AML therapies.40-45 We hypothesized that ARQ531 treatment of AML cells may inhibit this pathway. To validate this hypothesis, we measured MYB protein levels in ARQ531-treated cells. Exposure to ARQ531 resulted in marked MYB deregulation (Figure 4F), suggesting an important contribution to ARQ531 anti-tumor activity. To gain further insights into MYB reduction triggered by ARQ531, we tested the proteasome contribution, as previously reported for other MYB-targeting agents.43 As shown in Figure 4G, co-treatment with the proteasome inhibitor MG132 preserved MYB protein levels, suggesting that, in addition to its supposed effects on protein synthesis, ARQ531 affects MYB degradation. Our findings therefore suggest that ARQ531 interferes with many pro-survival pathways, such as MAPK, in AML cells.

ARQ531 dysregulates multiple oncogenic transcription factors in acute myeloid leukemia cells To gain insights into the molecular mechanisms of ARQ531, we analyzed treated HL-60 cells over time. As shown in Figure 5A, BTK signaling deregulation occurred early, after 2 h of treatment, followed by MYC downregulation. Importantly, apoptotic cell features, including PARP and caspase 3 cleavage, were seen after the decrease in MYB, suggesting that these events are crucial for ARQ531 anti-tumor activity. Published data show that small molecule BET inhibitors, by downregulating hematopoietic transcription factors, lead to potent therapeutic effects in several cancer models, including AML.46,47 We therefore tested the anti-AML activity of the BTK inhibitor ibrutinib combined with the BET bromodomain haematologica | 2020; 105(10)

inhibitor JQ1.48 As was seen in other cell types,25 the BET inhibitor enhanced the anti-tumor activity of the BTK inhibitor (Figure 5B). Western blot analysis of AML-treated cells confirmed these findings, further supporting the pivotal role of transcription factor deregulation in the anti-AML activity of ARQ531 (Figure 5C). Based on these data, we investigated the role of MYB in ARQ531 antiAML activity by challenging BTK-silenced cells with the repurposed drug mebendazole, recently described as a drug that induces MYB degradation.44 As expected, mebendazole reduced cell viability of BTK-depleted cells more than a control (Online Supplementary Figure S7A). We then performed several genetic studies to confirm these findings. As shown in Figure 6A, reduced viability was observed in MYC/MYB-depleted cells compared with the control, but more importantly, viability was significantly dampened in triple MYB/MYC/BTK-silenced cells (reduction by 64.7% to 38.5%), suggesting that such inhibition is detrimental to AML cells. Consistently, simultaneous silencing of MYC, MYB and BTK resulted in PARP cleavage together with impairment of ERK phosphorylation (Figure 6B). Similar data were observed in BTK-KD cells (Figure 6C and Online Supplementary Figure S7B). Nonetheless, the effect of triple knockdown was not quite equal to that of ARQ531 treatment, suggesting that other covalent or noncovalent targets are involved in the drugâ&#x20AC;&#x2122;s mechanism of action. Since MYB is reported to be crucial for leukemogenesis,42,49 we assessed the relationship between BTK and MYB in AML cells. Molecular data analysis of different publicly available AML cohort databases revealed higher expression of BTK and MYB in AML cells compared to normal HSC, with a positive correlation (Online Supplementary Figure S8A-C). These data support the notion that several oncogenic pathways, including BTK, MYB and MYC, are essential for leukemia cell maintenance, supporting the concept that ARQ531 could be an effective multi-targeted agent for the treatment of AML. 2427

D. Soncini et al. A

Figure 6. ARQ531 affects BTK, MYC and MYB in acute myeloid leukemia cells. (A) BTK, MYC, MYB, and especially their simultaneous silencing considerably reduced viability of HL60 cells as measured by Trypan blue staining. The mean of triplicate experiments are shown. (B, C) Triple BTK/MYC/MYB-silenced HL60 cells demonstrate diminished phosphorylation of ERK and PARP full-length form compared with cells depleted either transiently (B) or stably (C) of each gene, individually.

ARQ531 shows potent activity in a patient-derived xenograft mouse model of acute myeloid leukemia Based on in vitro data, we next assessed whether ARQ531 treatment would be effective and tolerable in animal models by using our established AML patient-derived xenograft model. NSG mice (n=20) were engrafted with 300,000 primary human AML cells (M4, acute myelomonocytic leukemia). Successful engraftment was documented by measuring circulating human CD45+ (hCD45+) cells in the animals’ peripheral blood with flow cytometry weekly for 2 months. At day 19 after infusion of the cells, once a systemic xenograft had been confirmed, mice were dosed orally with vehicle or ARQ531 (25 or 37.5 mg/kg; 5 mice/group) daily for 2 weeks. The percentage of human cells in peripheral blood samples was measured once a week up to day 42 (Figure 7A). ARQ531-treated mice had significant reductions in the numbers of hCD45+ cells despite very rapid growth of the aggressive leukemic cells (Figure 7B). At day 42 after starting treatment, there were 66.5±0.1% and 69.5±0.2% hCD45+ cells after ARQ531 treatment at the doses of 37.5 and 25 mg/kg, respectively; in contrast, vehicle-treated mice had 85% hCD45+ cells (**0.005<P<0.008) (Figure 7C). Analyses of bone marrow and spleen also showed reductions in tumor burden (hCD45+), although they were not statistically significant (Online Supplementary Figure S9). In addition, ARQ531 treatment was found to improve mouse survival significantly. As shown in Figure 7D, Kaplan-Meier analyses indicated that mice treated with the higher dose of ARQ531 survived significantly longer than those treated with the vehicle control (P<0.001). Overall, treatment was well tolerated as suggested by the maintenance of body weight and the lack of signs of toxicity, such as lethargy, ruffled fur, respiratory distress and hunchback posture (data not shown). Together these data indicate that, in vivo, ARQ531 administration was well tolerated and efficiently reduced leukemia cell growth, providing impetus for clinical evaluation of this novel small molecule.

Discussion AML cells often demonstrate constitutive activation of tyrosine kinase signaling resulting from specific genomic 2428

aberrations.16 These aberrations are attractive therapeutic targets, as demonstrated by the pharmacological inhibitor of BTK, ibrutinib, which blocks AML blast proliferation, migration, and leukemic cell adhesion to bone marrow stromal cells.15 However, BTK-based treatment of AML patients has been unsuccessful to date,22 with only a few preclinical, ex vivo studies suggesting that ibrutinib is effective against FLT3(ITD) and CD117 harboring cells, unlike the clinical benefit seen in patients with chronic lymphocytic leukemia and lymphoma.18,21 Adding inhibitory pressure on the BTK pathway might enhance the efficacy of this strategy, as previously reported.19,20,50-54 In this study, using a combination of genetic and biochemical approaches, we extensively characterized ARQ531, a novel, reversible, orally bioavailable, ATP-competitive inhibitor of BTK and associated kinases. ARQ531 greatly compromises AML cell survival by modulating transcriptional regulatory machinery coordinated by MYC, demonstrating activity both in vitro and in a patient-derived xenograft AML mouse model. Thus, our study provides the rationale for developing clinical trials using ARQ531 as a new treatment for patients with AML. Since ibrutinib does not directly inhibit components of the MAPK pathway, it is possible that the superior activity of ARQ531 in AML may be due to its modulation of additional targets, including kinases related to ERK signaling.27 Although screening analysis of Src-family kinases (including Lyn and Syk) did not show any effect on AML cells,5557 (Online Supplementary Figure S10) we assume that targeting of additional kinases is responsible for the marked anti-AML activity of ARQ531. By combining computational models and whole transcriptional analysis, we observed that ARQ531 treatment induces dysregulation of several transcription-addicted programs, including MYC and MYB. The combination of BTK inhibition and MYC/MYB downregulation explains the improved antiAML activity of ARQ531 compared to single agent tyrosine kinase inhibitors such as ibrutinib. Since ARQ531 simultaneously inhibits different cellular functions such as folding machinery, metabolic dependency, and genome integrity, it may provide deeper and more durable remissions, while delaying the emergence of resistance. Additionally, based on reports that degrading MYB eradihaematologica | 2020; 105(10)

Preclinical activity of ARQ531 in AML

Figure 7. ARQ531 inhibits tumor growth and extends survival in a patient-derived xenograft mouse model of acute myeloid leukemia. (A) Experimental outline for the analysis of the anti-leukemic activity of ARQ531 against primary human acute myeloid leukemia (AML) cells. A patient-derived xenograft mouse model of human primary AML cells was used to assess the efficacy of ARQ531 against AML cells isolated from patients with AML M4. (B) Representative flow cytometric dot plots representing tumor engraftment evaluated at day 35 after treatment. In the right panel, the histogram represents the percentage of human CD45+ cells in mice. Data are represented as mean Âą standard deviation; **P=0.006; ****P<0.001. (C) Circulating human CD45+ cells were measured in peripheral blood by flow cytometry weekly for 2 months. At day 19, a systemic xenograft was confirmed (tumor engraftment) and mice were randomized to receive vehicle control, a low dose of ARQ531 (25 mg/kg) or a high dose of ARQ531 (37.5 mg/kg). The percentage of human leukemic cells in peripheral blood of mice was measured weekly, up to day 42. 0.005<**P<0.008. (D) Kaplan-Meier curve of the patient-derived xenograft AML model following treatment with vehicle, a low dose of ARQ531 (25 mg/kg) or a high dose of ARQ531 (37.5 mg/kg). The higher drug dose led to significantly longer overall survival compared to that of the vehicle-treated, control mice (5 mice/group; P<0.001).

cates AML cells in mice without impairing normal myelopoiesis,46 ARQ531 treatment may be safe for hematopoietic precursor cells, supporting its clinical relevance. We also provide experimental evidence that the bone marrow stroma is not affected by treatment and, more importantly, does not affect the anti-tumor activity of ARQ531. Preliminary phase I studies confirm the safety profile of ARQ531, adding to the data that support its clinical development. Recent studies suggest that modulating transcriptional regulatory machinery is an innovative strategy to treat blood malignancies, including AML.14,38 An example of this strategy is all-trans retinoic acid treatment which, by modulating the transcriptional target PML-RARÎą, induces differentiation of leukemic blasts resulting in improved survival of patients.58 However, most transcription factors remain notoriously difficult to target, with siRNA-mediated silencing of gene expression being one of the few feasible approaches.59 Other oncoproteins, including MYC and MYB, are emerging as compelling targets for drug haematologica | 2020; 105(10)

development in AML, due to their ability to influence tumor proliferation.40-44,60,61 In this context, the new small molecule ARQ531, by affecting multiple oncogenic pathways simultaneously, results in perturbation of the transcriptional regulatory machinery which maintains AML cell integrity. Therefore, targeting BTK, MYC and MYB with ARQ531 represents an innovative strategy for improving the efficacy of AML therapy. In summary, we have demonstrated that ARQ531, a new reversible tyrosine kinase inhibitor, suppresses AML cell viability in vitro and in vivo by abrogating different oncogenic targets including BTK, MYC and MYB. Gene silencing of BTK, MYC and MYB in AML cells was not as effective as ARQ531, suggesting that other covalent or noncovalent targets are involved in its mechanism of action. Based on our preclinical data, we provide the rationale to explore the effects of this multi-targeted agent on hematologic malignancies as well as solid tumors, beyond investigating its clinical benefit in AML patients. 2429

D. Soncini et al.

Acknowledgments This work was supported in part by the Associazione Italiana per la Ricerca sul Cancro (AIRC, MYFG #18491 to MC and #21552 to AC; I.G. to FB), Italian Ministry of

References 1. Sant M, Allemani C, Tereanu C, et al. Incidence of hematologic malignancies in Europe by morphologic subtype: results of the HAEMACARE project. Blood. 2010;116 (19):3724-3734. 2. Ferrara F, Schiffer CA. Acute myeloid leukaemia in adults. Lancet. 2013;381 (9865):484-495. 3. Esposito MT, So CW. DNA damage accumulation and repair defects in acute myeloid leukemia: implications for pathogenesis, disease progression, and chemotherapy resistance. Chromosoma. 2014;123(6):545-561. 4. Stein EM. FLT3 inhibitors for relapsed or refractory acute myeloid leukaemia. Lancet Oncol. 2018;19(7):849-850. 5. El Fakih R, Rasheed W, Hawsawi Y, Alsermani M, Hassanein M. Targeting FLT3 mutations in acute myeloid leukemia. Cells. 2018;7(1):4 6. DiNardo CD, Stein EM, de Botton S, et al. Durable remissions with ivosidenib in IDH1-mutated relapsed or refractory AML. N Engl J Med. 2018;378(25):2386-2398. 7. Buggy JJ, Elias L. Bruton tyrosine kinase (BTK) and its role in B-cell malignancy. Int Rev Immunol. 2012;31(2):119-132. 8. Hendriks RW, Bredius RG, Pike-Overzet K, Staal FJ. Biology and novel treatment options for XLA, the most common monogenetic immunodeficiency in man. Expert Opin Ther Targets. 2011;15(8):1003-1021. 9. Herman SEM, Montraveta A, Niemann CU, et al. The Bruton tyrosine kinase (BTK) inhibitor acalabrutinib demonstrates potent on-target effects and efficacy in two mouse models of chronic lymphocytic leukemia. Clin Cancer Res. 2017;23(11):2831-2341. 10. Wilson WH, Young RM, Schmitz R, et al. Targeting B cell receptor signaling with ibrutinib in diffuse large B cell lymphoma. Nat Med. 2015;21(8):922-926. 11. Advani RH, Buggy JJ, Sharman JP, et al. Bruton tyrosine kinase inhibitor ibrutinib (PCI-32765) has significant activity in patients with relapsed/refractory B-cell malignancies. J Clin Oncol. 2013;31(1):8894. 12. Treon SP, Tripsas CK, Meid K, et al. Ibrutinib in previously treated Waldenstrom's macroglobulinemia. N Engl J Med. 2015;372(15):1430-1340. 13. Rushworth SA, Bowles KM, Barrera LN, Murray MY, Zaitseva L, MacEwan DJ. BTK inhibitor ibrutinib is cytotoxic to myeloma and potently enhances bortezomib and lenalidomide activities through NF-kappaB. Cell Signal. 2013;25(1):106-112. 14. 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): 10181029. 15. Rushworth SA, Murray MY, Zaitseva L, Bowles KM, MacEwan DJ. Identification of Bruton's tyrosine kinase as a therapeutic target in acute myeloid leukemia. Blood. 2014;123(8):1229-1238. 16. Oellerich T, Mohr S, Corso J, et al. FLT3-

2430

Health (5 x 1000 Funds of IRCCS San Martino-IST 2014 and 2015, to MC and AC), Associazione Italiana Leucemie Linfomi e Mieloma (AIL sezione di Genova) and University of Genoa, Italy.

ITD and TLR9 use Bruton tyrosine kinase to activate distinct transcriptional programs mediating AML cell survival and proliferation. Blood. 2015;125(12):1936-1947. 17. Wu H, Hu C, Wang A, et al. Ibrutinib selectively targets FLT3-ITD in mutant FLT3positive AML. Leukemia. 2016;30(3):754757. 18. Rushworth SA, Pillinger G, Abdul-Aziz A, et al. Activity of Bruton's tyrosine-kinase inhibitor ibrutinib in patients with CD117positive acute myeloid leukaemia: a mechanistic study using patient-derived blast cells. Lancet Haematol. 2015;2(5):e204-211. 19. Rotin LE, Gronda M, MacLean N, et al. Ibrutinib synergizes with poly(ADP-ribose) glycohydrolase inhibitors to induce cell death in AML cells via a BTK-independent mechanism. Oncotarget. 2016;7(3):27652779. 20. Li X, Yin X, Wang H, et al. The combination effect of homoharringtonine and ibrutinib on FLT3-ITD mutant acute myeloid leukemia. Oncotarget. 2017;8(8):1276412774. 21. Pillinger G, Abdul-Aziz A, Zaitseva L, et al. Targeting BTK for the treatment of FLT3ITD mutated acute myeloid leukemia. Sci Rep. 2015;5:12949. 22. Cortes JE, Estey E, Stein AS, Graef T, et al. A multicenter, open-label phase 2a study of ibrutinib with or without cytarabine in patients with acute myeloid leukemia (PCYC-1131). J Clin Oncol. 2015;33 (15_suppl):TPS7096-TPS7096. 23. Chong IY, Aronson L, Bryant H, et al. Mapping genetic vulnerabilities reveals BTK as a novel therapeutic target in oesophageal cancer. Gut. 2018;67(10):17801792. 24. Moyo TK, Wilson CS, Moore DJ, Eischen CM. Myc enhances B-cell receptor signaling in precancerous B cells and confers resistance to Btk inhibition. Oncogene. 2017;36(32):4653-4661. 25. Sun B, Shah B, Fiskus W, Qi J, Rajapakshe K, Coarfa C, et al. Synergistic activity of BET protein antagonist-based combinations in mantle cell lymphoma cells sensitive or resistant to ibrutinib. Blood. 2015;126(13):1565-1574. 26. Sasi BK, Martines C, Xerxa E, et al. Inhibition of SYK or BTK augments venetoclax sensitivity in SHP1-negative/BCL-2positive diffuse large B-cell lymphoma. Leukemia. 2019;33(10):2416-2428. 27. Reiff SD, Mantel R, Smith LL, et al. The BTK inhibitor ARQ 531 targets ibrutinibresistant CLL and Richter transformation. Cancer Discov. 2018;8(10):1300-1315. 28. Lane SW, Scadden DT, Gilliland DG. The leukemic stem cell niche: current concepts and therapeutic opportunities. Blood. 2009;114(6):1150-1157. 29. Bam R, Ling W, Khan S, et al. Role of Bruton's tyrosine kinase in myeloma cell migration and induction of bone disease. Am J Hematol. 2013;88(6):463-471. 30. Chang BY, Francesco M, De Rooij MF, et al. Egress of CD19(+)CD5(+) cells into peripheral blood following treatment with the Bruton tyrosine kinase inhibitor ibrutinib in

mantle cell lymphoma patients. Blood. 2013;122(14):2412-2424. 31. Zaitseva L, Murray MY, Shafat MS, et al. Ibrutinib inhibits SDF1/CXCR4 mediated migration in AML. Oncotarget. 2014;5(20): 9930-9938. 32. Dhillon AS, Hagan S, Rath O, Kolch W. MAP kinase signalling pathways in cancer. Oncogene. 2007;26(22):3279-3290. 33. Lamba S, Russo M, Sun C, et al. RAF suppression synergizes with MEK inhibition in KRAS mutant cancer cells. Cell Rep. 2014;8(5):1475-1483. 34. McCubrey JA, Steelman LS, Franklin RA, et al. Targeting the RAF/MEK/ERK, PI3K/AKT and p53 pathways in hematopoietic drug resistance. Adv Enzyme Regul. 2007;47:64-103. 35. Sears R, Nuckolls F, Haura E, Taya Y, Tamai K, Nevins JR. Multiple Ras-dependent phosphorylation pathways regulate Myc protein stability. Genes Dev. 2000;14(19): 2501-2514. 36. Sonenberg N, Hinnebusch AG. Regulation of translation initiation in eukaryotes: mechanisms and biological targets. Cell. 2009;136(4):731-745. 37. Zucal C, D'Agostino VG, Casini A, et al. EIF2A-dependent translational arrest protects leukemia cells from the energetic stress induced by NAMPT inhibition. BMC Cancer. 2015;15:855. 38. Brien GL, Valerio DG, Armstrong SA. Exploiting the epigenome to control cancerpromoting gene-expression programs. Cancer Cell. 2016;29(4):464-476. 39. Li S, Vallet S, Sacco A, Roccaro A, Lentzsch S, Podar K. Targeting transcription factors in multiple myeloma: evolving therapeutic strategies. Expert Opin Investig Drugs. 2019;28(5):445-462. 40. Coulibaly A, Haas A, Steinmann S, Jakobs A, Schmidt TJ, Klempnauer KH. The natural anti-tumor compound Celastrol targets a Myb-C/EBPbeta-p300 transcriptional module implicated in myeloid gene expression. PLoS One. 2018;13(2):e0190934. 41. Ramaswamy K, Forbes L, Minuesa G, et al. Peptidomimetic blockade of MYB in acute myeloid leukemia. Nat Commun. 2018;9 (1):110. 42. Xu Y, Milazzo JP, Somerville TDD, et al. A TFIID-SAGA Perturbation that targets MYB and suppresses acute myeloid leukemia. Cancer Cell. 2018;33(1):13-28 e8. 43. Walf-Vorderwulbecke V, Pearce K, Brooks T, et al. Targeting acute myeloid leukemia by drug-induced c-MYB degradation. Leukemia. 2018;32(4):882-889. 44. Liu W, Wu M, Huang Z, et al. c-myb hyperactivity leads to myeloid and lymphoid malignancies in zebrafish. Leukemia. 2017;31(1):222-233. 45. Uttarkar S, Dasse E, Coulibaly A, et al. Targeting acute myeloid leukemia with a small molecule inhibitor of the Myb/p300 interaction. Blood. 2016;127(9):1173-1182. 46. Roe JS, Mercan F, Rivera K, Pappin DJ, Vakoc CR. BET bromodomain inhibition suppresses the function of hematopoietic transcription factors in acute myeloid

haematologica | 2020; 105(10)

Preclinical activity of ARQ531 in AML

leukemia. Mol Cell. 2015;58(6):1028-1039. 47. Zuber J, Shi J, Wang E, et al. RNAi screen identifies Brd4 as a therapeutic target in acute myeloid leukaemia. Nature. 2011;478 (7370):524-528. 48. Filippakopoulos P, Qi J, Picaud S, et al. Selective inhibition of BET bromodomains. Nature. 2010;468(7327):1067-1073. 49. Pattabiraman DR, Gonda TJ. Role and potential for therapeutic targeting of MYB in leukemia. Leukemia. 2013;27(2):269-277. 50. Linley A, Krysov S, Ponzoni M, Johnson PW, Packham G, Stevenson FK. Lectin binding to surface Ig variable regions provides a universal persistent activating signal for follicular lymphoma cells. Blood. 2015;126(16):1902-1910. 51. Cea M, Cagnetta A, Acharya C, et al. Dual NAMPT and BTK targeting leads to synergistic killing of Waldenstrom macroglobulinemia cells regardless of MYD88 and CXCR4 somatic mutation status. Clin Cancer Res. 2016;22(24):6099-6109.

haematologica | 2020; 105(10)

52. Deng J, Isik E, Fernandes SM, Brown JR, Letai A, Davids MS. Bruton's tyrosine kinase inhibition increases BCL-2 dependence and enhances sensitivity to venetoclax in chronic lymphocytic leukemia. Leukemia. 2017;31(10):2075-2084. 53. Hing ZA, Mantel R, Beckwith KA, et al. Selinexor is effective in acquired resistance to ibrutinib and synergizes with ibrutinib in chronic lymphocytic leukemia. Blood. 2015;125(20):3128-3132. 54. de Rooij MF, Kuil A, Kraan W, et al. Ibrutinib and idelalisib target B cell receptorbut not CXCL12/CXCR4-controlled integrin-mediated adhesion in Waldenstrom macroglobulinemia. Haematologica. 2016;101(3):e111-115. 55. Puissant A, Fenouille N, Alexe G, et al. SYK is a critical regulator of FLT3 in acute myeloid leukemia. Cancer Cell. 2014;25(2):226-242. 56. Hahn CK, Berchuck JE, Ross KN, et al. Proteomic and genetic approaches identify

Syk as an AML target. Cancer Cell. 2009;16(4):281-294. 57. Dos Santos C, Demur C, Bardet V, PradeHoudellier N, Payrastre B, Recher C. A critical role for Lyn in acute myeloid leukemia. Blood. 2008;111(4):2269-2279. 58. Lo-Coco F, Orlando SM, Platzbecker U. Treatment of acute promyelocytic leukemia. N Engl J Med. 2013;369(15):1472. 59. Yan C, Higgins PJ. Drugging the undruggable: transcription therapy for cancer. Biochim Biophys Acta. 2013;1835(1):7685. 60. Mertz JA, Conery AR, Bryant BM, et al. Targeting MYC dependence in cancer by inhibiting BET bromodomains. Proc Natl Acad Sci U S A. 2011;108(40):16669-16674. 61. Zuber J, Rappaport AR, Luo W, et al. An integrated approach to dissecting oncogene addiction implicates a Myb-coordinated self-renewal program as essential for leukemia maintenance. Genes Dev. 2011;25(15):1628-1640.

2431

ARTICLE Ferrata Storti Foundation

Haematologica 2020 Volume 105(10):2432-2439

Hematologic Neoplasms

Myeloproliferative and lymphoproliferative malignancies occurring in the same patient: a nationwide discovery cohort

Johanne M. Holst,1,2 Trine L. Plesner,3 Martin B. Pedersen,1 Henrik Frederiksen,4 Michael B. Møller,5 Michael R. Clausen,1,2 Marcus C. Hansen,1,2 Stephen Jacques Hamilton-Dutoit,6 Peter Nørgaard,7 Preben Johansen,8 Tobias Ramm Eberlein,9 Bo K. Mortensen,10 Gustav Mathiasen,11 Andreas Øvlisen,12 Rui Wang,13 Chao Wang,14 Weiwei Zhang,15 Hans Beier Ommen,1 Jesper Stentoft,1 Maja Ludvigsen,1,2 Wayne Tam,13 Wing C. Chan,14 Giorgio Inghirami13 and Francesco d’Amore1,2

Department of Hematology, Aarhus University Hospital, Aarhus, Denmark; 2Department of Clinical Medicine, Aarhus University, Aarhus, Denmark; 3Department of Pathology, Rigshospitalet, Copenhagen, Denmark; 4Department of Hematology, Odense University Hospital, Odense, Denmark; 5Department of Pathology, Odense University Hospital, Odense, Denmark; 6Institute of Pathology, Aarhus University Hospital, Aarhus, Denmark;7Department of Pathology, Herlev Hospital, Copenhagen, Denmark; 8 Department of Pathology, Aalborg University Hospital, Aalborg, Denmark; 9Department of Hematology, Regional Hospital West Jutland, Holstebro, Denmark; 10Department of Hematology, Herlev Hospital, Herlev, Denmark; 11Department of Hematology, Roskilde Hospital, Roskilde, Denmark; 12Department of Hematology, Aalborg University Hospital, Aalborg, Denmark; 13Department of Pathology and Laboratory Medicine, Weill Cornell Medicine, New York City, NY, USA; 14Department of Pathology, City of Hope Medical Center, Duarte, CA, USA and 15Department of Pathology and Microbiology, University of Nebraska Medical Center, Omaha, NE, USA 1

ABSTRACT

Correspondence: FRANCESCO D’AMORE frandamo@rm.dk Received: May 1, 2019. Accepted: November 26, 2019. Pre-published: November 28, 2019. doi:10.3324/haematol.2019.225839 ©2020 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.

2432

yeloid and lymphoid malignancies are postulated to have distinct pathogenic mechanisms. The recent observation that patients with a myeloproliferative neoplasm have an increased risk of developing lymphoproliferative malignancies has challenged this assumption. We collected a nationwide cohort of patients with both malignancies. Patients diagnosed between 1990 and 2015 were identified through the national Danish Pathology Registry. We identified 599 patients with a myeloproliferative neoplasm and a concomitant or subsequent diagnosis of lymphoma. Histopathological review of the diagnostic samples from each patient led to a final cohort of 97 individuals with confirmed dual diagnoses of myeloproliferative neoplasm and lymphoma. The age range at diagnosis of these individuals was 19-94 years (median: 71 years). To avoid the inclusion of cases of therapy-induced myeloproliferative neoplasm occurring in patients previously treated for lymphoma, only patients with myeloproliferative neoplasm diagnosed unequivocally before the development of lymphoma were included. The average time interval between the diagnoses of the two malignancies was 1.5 years. In the majority of patients (90%) both diagnoses were established within 5 years of each other. Among the lymphoma entities, the frequency of peripheral T-cell lymphomas was markedly increased. Interestingly, all but one of the T-cell lymphomas were of angioimmunoblastic type. These findings suggest that a myeloproliferative neoplasm and lymphoproliferative malignancy developing in the same patient may have common pathogenic events, possibly already at the progenitor level. We believe that the molecular characterization of the newly developed biorepository will help to highlight the mechanisms driving the genesis and clonal evolution of these hematopoietic malignancies. haematologica | 2020; 105(10)

Myeloid and lymphoid neoplasms in the same patient

Introduction The diagnosis of multiple clonal hematologic neoplasms in the same patient is considered to be rare. Furthermore, when it does occur, it is unclear whether the individual disorders are pathogenetically related, e.i., sharing common driver mutations, or whether they simply reflect independently developed random events. Studies of tumor DNA from patients with angioimmunoblastic T-cell lymphoma (AITL) have identified genomic changes which can be detected in the early hematopoietic stem cell precursors.1â&#x20AC;&#x201C;5 These alterations include changes in IDH2, TET2, and DNMT3A, and are commonly seen in myeloid malignancies. Acquired somatic mutation in the Janus kinase 2 (JAK2) gene plays an essential role in the development of myeloproliferative neoplasms (MPN). Notably, this mutation has also been detected in some lymphoid malignancies (LM).6,7 These findings have prompted the hypothesis that some genomic changes in early hematopoietic stem cell precursors may predispose to and could drive the development of both MPN and LM. MPN are clonal hematopoietic stem cell disorders characterized by proliferation of one or more of the myeloidderived cell lineages. They include essential thrombocythemia (ET), polycythemia vera (PV), primary myelofibrosis (PMF), chronic myeloid leukemia (CML) and MPNunclassifiable (MPN-U).8 According to the database of cancer statistics for Nordic countries (NORDCAN; wwwdep.iarc.fr/NORDCAN/English/frame.asp), the age-adjusted incidence rate of CML in Denmark is 0.95/100,000 and that for the other chronic myeloproliferative entities (PV, ET, PMF, MPN-U; taken as one group) 4.05/100,000. In general, MPN are slowly progressing diseases which are, however, capable of transformation to severe bone marrow failure or acute leukemia.8 Recent epidemiological studies have shown an increased risk of developing other types of malignancy in patients with MPN.9,10 In particular, the risk of also developing LM is significantly increased compared with the risk in a sex- and age-matched background population. Dual diagnoses of MPN and LM in the same patient have previously been described in both individual case reports and in small case series.11 However, a more substantial, population-based evaluation of this phenomenon, together with a biorepository of tumor specimens from such patients, has not yet been reported. The aim of our study was to identify and characterize a nationwide cohort of Danish patients with dual diagnoses of myeloproliferative and lymphoproliferative malignancies. We describe what we believe to be the largest series of patients with diagnoses of both MPN and LM, with particular emphasis on the establishment of the cohort, the histopathological classification of the malignancies, and description of the clinical characteristics of the patients.

Methods Cohort identification Patients diagnosed with both MPN and LM within the period 1977-2015 were identified through the national Danish Pathology Registry (DPR).12 The DPR is a nationwide register that records data on all pathology specimens from patients in Denmark. The haematologica | 2020; 105(10)

registry and its associated database are updated daily. Since essentially all pathology investigations in Denmark are performed within the tax-funded public health system, the coverage of the DPR is close to 100%. The registry includes detailed variables related to patients, specimens and workload. All Danish citizens and residents are assigned a unique identifier, the civil personal registration number, at birth or immigration.13 Using this registration number, a patientâ&#x20AC;&#x2122;s data in the DPR can be linked to the many other Danish clinical databases. In addition, each specimen identified via the DPR can be linked to the available formalin-fixed, paraffin-embedded tissue biopsies stored in the diagnostic archives of the Danish pathology departments, allowing identification, location and retrieval of relevant primary diagnostic tissue specimens from the cohort patients. The MPN diagnoses include PV, ET, PMF, CML, and MPN-U. In order to exclude secondary myelodysplasias/MPN occurring as a result of previous treatment for LM, only patients diagnosed with either both diseases concomitantly (i.e., diagnosed no more than 6 months apart) or with MPN first and LM subsequently, were selected for further histopathological revision. Because of the low number of samples available for the period 1977 to 1990 and the poor tissue quality of the older samples, only specimens from patients diagnosed in 1990 or later were included in the final cohort. The study was approved by The Central Denmark Region Committees on Health Research Ethics (record n. 1-10-72-161-15) and the Danish Data Protection Agency (record n. 1-16-02-42015), and it was conducted in compliance with the principles of the Helsinki Declaration.

Data sources All data sources were linked using the unique patient-specific civil personal registration number. Clinical data were obtained from the population-based Danish National Lymphoma Registry (LYFO)14 and supplemented by medical records. For comparison of overall survival, a diffuse large B-cell lymphoma (DLBCL) reference cohort matched for age, sex, and the International Prognostic Index (IPI) was identified and randomly selected from LYFO (n=100). For AITL patients, a reference cohort previously described by Pedersen et al. was used (n=25).15

Tissue collection and histological revision All specimens were pre-therapeutic biopsies from patients diagnosed with both MPN and LM. Formalin-fixed, paraffin-embedded tissue specimens were collected from the archives at 15 different Danish pathology departments. After careful evaluation of the original pathology reports, new hematoxylin and eosin-stained sections were cut from the study paraffin blocks and the histopathological diagnoses were reviewed. If necessary, supplementary immunohistochemical stains were assessed, in addition to those originally performed. Samples were reviewed by an experienced hematopathologist (TLP) at a tertiary referral center according to the 2017 revision of the 2008 World Health Organization Classification of Tumours of Haematopoietic and Lymphoid Tissues.8 AITL tumors were specifically tested for wellknown recurrent mutations in IDH2, TET2, DNMT3A, and RHOA genes.

Statistical methods Outcomes were described by overall survival, defined as the time interval from LM diagnosis to last follow-up or death from any cause. Overall survival estimates were calculated according to the Kaplan-Meier method and compared using the log-rank test. All statistical analyses were performed using STATA version IC 14.1 (StataCorp, College Station, TX, USA). 2433

J.M. Holst et al.

Results Establishment of the cohort The algorithm leading to the establishment of the cohort is shown in Figure 1. In total, 26,736 patients with MPN were identified. Of these, 1,524 had an additional registered diagnosis of concomitant or subsequent LM. If multiple biopsies from the MPN and/or LM diagnosis of a given patient were taken at different times the patient would end up with multiple registrations in the DPR. These redundant registrations were omitted (n=925). Patients with diagnoses of MPN and/or LM, which were unconfirmed after histopathological review (n=344), were excluded. A group of 67 patients diagnosed with LM prior to the MPN were also excluded, as were 18 patients diagnosed before 1990. Thus, diagnostic tissue specimens from 170 patients were retrieved for histopathological validation. Of these, 97 patients were confirmed to have both MPN and LM and represent the final cohort of the study.

Demographic and clinico-pathological features Table 1 summarizes the demographic characteristics of the patients. No major differences in sex distribution were observed, either overall or at subtype-specific level. The age at MPN diagnosis was for most patients within the sixth and seventh decades of life with the exception of patients with lymphoblastic lymphoma (LBL), who were significantly younger (median 33 years; interquartile range, 22-57 years). The overall average time between the diagnoses of MPN and LM was 1.5 years and in the majority of the patients (90%) both diagnoses were established within 5 years of each other. Table 2 shows a cross tabulation between MPN and LM diagnoses. The most frequent LM diagnoses were chronic lymphocytic leukemia (CLL, 32%) and DLBCL (21%), in line with these being two of the more common LM entities in the general population. Peripheral T-cell lymphomas (PTCL) usually account for approximately 10% of all lymphomas, which is in line with the relative frequencies seen in our study.

Figure 1. Flowchart illustrating the establishment of the cohort. DPR: the Danish pathology register; MPN: myeloproliferative neoplasms; LM: lymphoproliferative malignancy.

2434

haematologica | 2020; 105(10)

Myeloid and lymphoid neoplasms in the same patient

Interestingly, AITL patients accounted for 89% (8 out of 9) of the MPN-associated PTCL, an unexpectedly high frequency (5- to 7-fold) for this specific entity. AITL tumors frequently harbored mutations of TET2 (60%), IDH2 (60%), DNMT3A (60%), and RHOA (80%) genes. All five patients with LBL (B-cell lineage, n=2; T-cell lineage, n=2; and unclassifiable, n=1) were associated with a pre-existing CML, probably being cases of lymphoid blast crises on a CML background. Overall, 64% (n=62) of the MPN patients were diagnosed with a concurrent LM, while the remaining 36% (n=35) developed LM subsequently. Figure 2 illustrates the chronological occurrence of MPN and LM in each patient with PTCL, DLBCL, and CLL. The majority (n=23; 77%) of CLL patients were diagnosed with both malignancies concurrently. In contrast, most DLBCL (n=11; 55%) and, even more strikingly, most PTCL patients (n=6; 67%) were diagnosed at a mean time interval of between 1 to 3 years after the MPN diagnosis (1.5 years for DLBCL and 2.8 years for PTCL).

Outcome in selected lymphoma subtypes In a survival analysis, patients with both MPN and DLBCL had an inferior outcome compared with an age-, sex-, and IPI-matched DLBCL reference cohort [hazard ratio (HR)=1.9, 95% confidence interval (95% CI): 1.1-3.3;

P<0.02] (Figure 3A). The 5-year overall survival rate of the patients with MPN+DLBCL was 19% (95% CI: 5-39%) as compared with 34% (95% CI: 24-43%) for the DLBCL patients of the matched reference cohort. In contrast, no difference was found in outcomes between patients with both MPN and PTCL and patients with ‘AITL only’ (Figure 3B). The 5-year overall survival of the patients with MPN+CLL was 65% (95% CI: 45-79%). Historical data from the Danish CLL group showed a corresponding value for ‘CLL only’ patients diagnosed in the period 20122017 of 79% (95% CI: 77-81%).16

Discussion In establishing this population-based cohort, we identified a substantial number of patients diagnosed with myeloid malignancies, who were also diagnosed, either at the same time or later, with a lymphoid neoplasia. Among these lymphoid neoplasias, AITL were found at a much higher frequency than expected. We believe that this represents the largest reported cohort of patients with dual MPN and LM diagnoses, and that the associated biorepository is unique for its potential to foster the recognition of driver pathogenetic aberrations as well as the hierarchical

Table 1. Demographic features of the study population. Number Male/female ratio Age, years At MPN diagnosis, range At MPN diagnosis, median At MPN diagnosis, IQR Time between the MPN and LM diagnoses (mean), years

All

CLL

DLBCL

PTCL

LBL

97 1.1

31 0.9

20 1.0

9 0.8

10 2.3

5 4.0

19-94 71 63-79

57-88 72 63-81

52-88 71 66-79

60-82 67 64-73

59-81 71 64-77

19-57 33 22-57

1.5

1.4

2.4

2.8

1.4

1.7

CLL: chronic lymphocytic leukemia/small lymphocytic lymphoma; DLBCL: diffuse large B-cell lymphoma; PTCL: peripheral T-cell lymphoma; WM: Waldenström macroglobulinemia; MPN: myeloproliferative neoplasms; IQR: interquartile range; LBL: lymphoblastic lymphoma.

Table 2. Overview of the associated myeloproliferative and lymphoproliferative malignancies. Chronic lymphocytic leukemia Diffuse large B-cell lymphoma Low grade B-cell lymphoma - NOS Peripheral T-cell lymphoma Anaplastic large cell lymphoma Angioimmunoblastic T-cell lymphoma Waldenström macroglobulinemia Lymphoblastic lymphoma Marginal zone lymphoma Hodgkin lymphoma Follicular lymphoma Mantle cell lymphoma Primary CNS lymphoma Total

PMF

CML

MPN-U

Total

8 8 4

6 2 -

2 2 3

1 2 -

14 6 4

31 20 11

2 22

2 1 1 1 13

1 1 4 1 14

2 5 1 1 1 13

3 3 4 1 35

1 8 10 5 5 2 2 1 1 97

PV: polycythemia vera; ET: essential thrombocythemia; PMF: primary myelofibrosis; CML: chronic myeloid leukemia; MPN-U, myeloproliferative neoplasms – unclassifiable; NOS: not otherwise specified; CNS: central nervous system..

haematologica | 2020; 105(10)

2435

J.M. Holst et al.

relationship within the clonal evolution of hematologic malignancies. Epidemiological studies have revealed that patients with MPN have an increased standardized risk of developing LM.9,17â&#x20AC;&#x201C;20 Recently, a Swedish population-based study confirmed an increased risk of second malignancies in MPN patients with a hazard ratio of 2.6 (2.0-3.3) for developing lymphoma.10 However, none of the published studies indicated the frequency of different LM subtypes. Our data revealed a diverse range of different lymphoma entities. In the general population, AITL is, along with PTCL-NOS, the most common PTCL, accounting for 20-35% of all cases of PTCL in Caucasian populations.21,22 Notably, in our cohort, eight of nine PTCL cases were of angioimmunoblastic type, i.e., approximately 4-fold the expected number.

AITL has a complex clinical picture and is often diagnosed at advanced stage.23 Histopathological examination of AITL tissue usually shows a microenvironment of nonmalignant bystander cells, together with a minor population of neoplastic follicular helper T cells (TFH), which are believed to be the cell of origin of AITL.24 Recently, molecular alterations unique to AITL and to PTCL of TFH-origin have been described.4,25 Among the most frequent genomic alterations are recurrent mutations of RHOA5,26 and epigenetic modifier genes such as DNMT3A, TET2, and IDH2.1,3,27,28 The latter are wellknown genetic lesions and were originally identified in myeloid malignancies including myelodysplastic syndromes and MPN. DNMT3A and TET2 mutations have been predominantly found in progenitors prior to T-cell and B-cell commitment, whereas RHOA and IDH2 muta-

Figure 2. Swimmer plots showing, for each patient, for three selected lymphoid diagnoses the chronological occurrence of the myeloproliferative and lymphoproliferative malignancies. Each bar represents one patient in the study. Dark gray parts of the bars represent the time between the diagnosis of the myeloproliferative neoplasm and the lymphoma diagnosis. Light gray parts of the bars represent time with both diagnoses to death or last follow-up. PTCL: peripheral T-cell lymphoma; DLBCL: diffuse large B-cell lymphoma; CLL: chronic lymphocytic leukemia/small lymphocytic lymphoma.

2436

haematologica | 2020; 105(10)

Myeloid and lymphoid neoplasms in the same patient

tions are usually found more downstream in cells that have undergone lineage specification.29 TET2 and IDH2 mutations are mutually exclusive in myeloid malignancies, but often co-exist in AITL.5,25,29 Interestingly, the combination of TET2 deletion together with RHOA mutation has been shown to lead to the development of AITL in mice.30-32 In spite of an increasing number of anecdotal reports of MPN associated with the development of a variety of LM, no definitive relationship between the conditions has been established.33 However, the relatively high frequency of mutations in epigenetic modifier genes, found in both myeloid and lymphoid malignancies, could suggest a possible pathogenetic relevance of these mutations for the dual malignant transformation. Moreover,

they may represent evidence of shared pathogenetic mechanisms related to hierarchical mutation steps occurring as early events in the hematopoietic neoplastic process. Development of second malignancies in MPN patients may influence survival.34 Our study indicates that patients with both MPN and DLBCL have a worse prognosis compared with a reference cohort of patients with DLBCL alone matched for age, sex, and IPI score. This observation cannot be readily explained, but may be due to added morbidity as a consequence of multiple treatment courses related to both the myeloid and lymphoid malignancies and/or added deleterious genomic alterations at both stem cell and committed lineage-specific cell level.

Figure 3. Survival analyses. Kaplan Meier estimates of overall survival in (A) patients with diffuse large B-cell lymphoma with and without a previous diagnosis of a myeloproliferative neoplasm and (B) patients with peripheral T-cell lymphoma with and without a previous diagnosis of a myeloproliferative neoplasm. DLBCL: diffuse large B-cell lymphoma; MPN_ myeloproliferative neoplasm; PTCL: peripheral T-cell lymphoma; AITL: angioimmunoblastic T-cell lymphoma.

haematologica | 2020; 105(10)

2437

J.M. Holst et al.

Conversely, no significant outcome difference was seen between patients with both MPN and AITL and a reference cohort diagnosed with AITL alone. AITL is a rare disease with a poor prognosis and with frequent relapses. Hence, our outcome analysis was expectedly hampered by the limited size of the compared cohorts, and additional studies are needed to verify this observation. Recently, an overrepresentation of B-cell lymphomas in patients with PMF treated with JAK1/2 inhibitors has been observed.35 None of the patients with myelofibrosis in our cohort had received JAK inhibition therapy. Observational studies based on archival material have inherent limitations regarding the completeness of available clinical information and the validity of the histopathological diagnoses. Our study is a nationwide collaboration between departments of pathology and hematology in Denmark, one major aim being to provide specimens and clinico-pathological data from a population-based cohort of patients diagnosed with both a myeloid and a lymphoid malignancy. Furthermore, all cases underwent histopathological validation including a diagnostic update according to the most recent revision of the World Health Organization classification.17 A biorepository of tissue specimens has been established and DNA extracted for genomic analysis from both the MPN and LM samples.

References 1. Dobay M, Lemonnier F, Missiaglia E, et al. Integrative clinicopathological and molecular analyses of angioimmunoblastic T-cell lymphoma and other nodal lymphomas of follicular helper T-cell origin. Haematologica. 2017;201(4):e148-e151. 2. Cairns RA, Iqbal J, Lemonnier F, et al. IDH2 mutations are frequent in angioimmunoblastic T-cell lymphoma. Blood. 2012;119(8):1901-1903. 3. Lemonnier F, Couronne L, Parrens M, et al. Recurrent TET2 mutations in peripheral Tcell lymphomas correlate with TFH-like features and adverse clinical parameters. Blood. 2012;120(7):1466-1469. 4. Couronne L, Bastard C, Bernard OA. TET2 and DNMT3A mutations in human T-cell lymphoma. N Engl J Med. 2012;366(1):9596. 5. Sakata-Yanagimoto M, Enami T, Yoshida K, et al. Somatic RHOA mutation in angioimmunoblastic T cell lymphoma. Nat Genet. 2014;46(2):171-175. 6. Kodali S, Chen C, Rathnasabapathy C, Wang JC. JAK2 mutation in a patient with CLL with coexistent myeloproliferative neoplasm (MPN). Leuk Res. 2009;33(12): 236-239. 7. Roncero AM, López-Nieva P, CobosFernández MA, et al. Contribution of JAK2 mutations to T-cell lymphoblastic lymphoma development. Leukemia. 2016;30 (1):94-103. 8. Swerdlow S, Campo E, Harris N, et al. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues (revised 4th edition). Lyon. 2017. 9. Frederiksen H, Farkas DK, Christiansen CF, Hasselbalch HC, Sørensen HT. Chronic myeloproliferative neoplasms and subsequent cancer risk: a Danish populationbased cohort study. Blood. 2011;118(25):

2438

In conclusion, we describe here the frequency, demographics and epidemiology of the largest reported cohort of patients diagnosed with both MPN and LM. Patients with both MPN and DLBCL have poorer outcomes than those with DLBCL only. AITL is the lymphoma entity with by far the highest relative frequency of associated MPN. We hypothesize that shared genomic abnormalities may predispose to the combined or sequential development of MPN and LM in the same host. To further investigate this hypothesis, we have established a biorepository of all available specimens from this nationwide cohort. This unique material will be an invaluable resource for the performance of genomic studies to investigate the pathogenetic relationship between MPN and LM with possible novel therapeutic implications. Acknowledgments The authors would like to thank the collaborating pathology and hematology departments for retrieval of tissue samples and collection of clinical data. Special thanks for technical assistance to Kristina Lystlund Lauridsen, Laboratory of Molecular Pathology at Aarhus University Hospital and Joelle Racchumi, Department of Pathology and Laboratory Medicine at Weill Cornell Medicine. This work was supported by grants from Aarhus University, iLymph, SEB pension, the Danish Lymphoma Group, and the Ølufgaard P. N. Kristensen Foundation.

6515-6520. 10. Landtblom AR, Bower H, Andersson TM, et al. Second malignancies in patients with myeloproliferative neoplasms: a population-based cohort study of 9379 patients. Leukemia. 2018;32(10):2203-2210. 11. Marchetti M, Carobbio A, Capitoni E, Barbui T. Lymphoproliferative disorders in patients with chronic myeloproliferative neoplasms: a systematic review. Am J Hematol. 2018;93(5):698-703. 12. Erichsen R, Lash TL, Hamilton-Dutoit SJ, Bjerregaard B, Vyberg M, Pedersen L. Existing data sources for clinical epidemiology: the Danish national pathology registry and data bank. Clin Epidemiol. 2010;2:5156. 13. Schmidt M, Pedersen L, Sørensen HT. The Danish civil registration system as a tool in epidemiology. Eur J Epidemiol. 2014;29(8): 541-549. 14. Arboe B, Josefsson P, Jørgensen J, et al. Danish national lymphoma registry. Clin Epidemiol. 2016;8:577-581. 15. Pedersen M, Hamilton-Dutoit S, Bendix K, et al. DUSP22 and TP63 rearrangements predict outcome of ALK-negative anaplastic large cell lymphoma: a Danish cohort study. Blood. 2017;130(4):554-557. 16. Danish Lymphoma Group - Annual Report 2017. http://www.lymphoma.dk/arsrapporter/ 17. Rumi E, Passamonti F, Elena C, et al. Increased risk of lymphoid neoplasms in patients with myeloproliferative neoplasms: a study of 1,915 patients. Haematologica. 2011;96(3):454-458. 18. Vannucchi AM, Masala G, Antonioli E, et al. Increased risk of lymphoid neoplasms in patients with Philadelphia chromosomenegative myeloproliferative neoplasms. Cancer Epidemiol Biomarkers Prev. 2009;18(7):2068-2073. 19. Khanal N, Giri S, Upadhyay S, Shostrom VK, Pathak R, Bhatt VR. Risk of second pri-

20.

21.

22.

23.

24.

25.

26.

27.

mary malignancies and survival of adult patients with polycythemia vera: a United States population-based retrospective study. Leuk Lymphoma. 2016;57(1):129-133. Shrestha R, Giri S, Pathak R, Bhatt VR. Risk of second primary malignancies in a population-based study of adult patients with essential thrombocythemia. World J Clin Oncol. 2016;7(4):324-330. Vose J, Armitage J, Weisenburger D. International peripheral T-cell and natural killer/T-cell lymphoma study: pathology findings and clinical outcomes. J Clin Oncol. 2008;26(25):4124-4130. Laurent C, Baron M, Amara N, et al. Impact of expert pathologic review of lymphoma diagnosis: study of patients from the French Lymphopath Network. J Clin Oncol. 2017;35(18):2008-2017. Federico M, Rudiger T, Bellei M, et al. Clinicopathologic characteristics of angioimmunoblastic T-cell lymphoma: analysis of the International Peripheral Tcell Lymphoma Project. J Clin Oncol. 2013;31(2):240-246. de Leval L, Rickman DS, Thielen C, et al. The gene expression profile of nodal peripheral T-cell lymphoma demonstrates a molecular link between angioimmunoblastic T-cell lymphoma (AITL) and follicular helper T (TFH ) cells. Blood. 2007;109(11): 4952-4963. Odejide O, Weigert O, Lane AA, et al. A targeted mutational landscape of angioimmunoblastic T-cell lymphoma. Blood. 2014;123(9):1293-1296. Palomero T, Couronné L, Khiabanian H, et al. Recurrent mutations in epigenetic regulators, RHOA and FYN kinase in peripheral T cell lymphomas. Nat Genet. 2014;46(2): 166-170. Vallois D, Dobay MPD, Morin RD, et al. Activating mutations in genes related to TCR signaling in angioimmunoblastic and other follicular helper T-cell-derived lym-

haematologica | 2020; 105(10)

Myeloid and lymphoid neoplasms in the same patient

phomas. Blood. 2016;128(11):1490-1502. 28. Rosenquist R, Rosenwald A, Du M-Q, et al. Clinical impact of recurrently mutated genes on lymphoma diagnostics: state-ofthe-art and beyond. Haematologica. 2016;101(9):1002-1009. 29. Nguyen TB, Sakata-Yanagimoto M, Asabe Y, et al. Identification of cell-type-specific mutations in nodal T-cell lymphomas. Blood Cancer J. 2017;7(1):1-10. 30. Zang S, Li J, Yang H, et al. Mutations in 5methylcytosine oxidase TET2 and RhoA cooperatively disrupt T cell homeostasis. J

haematologica | 2020; 105(10)

Clin Invest. 2017;127(8):2998-3012. 31. Ng SY, Brown L, Stevenson K, et al. RhoA G17V is sufficient to induce autoimmunity and promotes T-cell lymphomagenesis in mice. Blood. 2018;132(9):935-947. 32. Cortes JR, Ambesi-Impiombato A, CouronnĂŠ L, et al. RHOA G17V induces T follicular helper cell specification and promotes lymphomagenesis. Cancer Cell. 2018;33(2):259-273. 33. Lemonnier F, Dupuis J, Sujobert P, et al. Treatment with 5-azacytidine induces a sustained response in patients with

angioimmunoblastic T-cell lymphoma. Blood. 2018;132(21):2305-2309. 34. Frederiksen H, Farkas DK, Christiansen CF, et al. Survival of patients with chronic myeloproliferative neoplasms and new primary cancers: a population-based cohort study. Lancet Haematol. 2015;2(7):e289e296. 35. Porpaczy E, Tripolt S, Hoelbl-Kovacic A, et al. Aggressive B-cell lymphomas in patients with myelofibrosis receiving JAK1/2 inhibitor therapy. Blood. 2018;132(7):694706.

2439

ARTICLE Ferrata Storti Foundation

Haematologica 2020 Volume 105(10):2440-2447

Chronic Lymphocytic Leukemia

Prognostic and predictive role of gene mutations in chronic lymphocytic leukemia: results from the pivotal phase III study COMPLEMENT1

Eugen Tausch,1 Philipp Beck,1 Richard F. Schlenk,1,2 Billy M.C. Jebaraj,1 Anna Dolnik,1,3 Deyan Y. Yosifov,1,4 Peter Hillmen,5 Fritz Offner,6 Ann Janssens,7 K. Govind Babu,8 Sebastian Grosicki,9 Jiri Mayer,10 Panagiotis Panagiotidis,11 Astrid McKeown,12 Ira V. Gupta,13 Alexandra Skorupa,14 Celine Pallaud,15 Lars Bullinger,1,3 Daniel Mertens,1,4 Hartmut Döhner1 and Stephan Stilgenbauer1,16

1 Department of Internal Medicine III, Ulm University, Ulm, Germany; 2NCT-Trial Center, National Center for Tumor Diseases, German Cancer Research Center, Heidelberg, Germany; 3Klinik für Innere Medizin mit Schwerpunkt Hämatologie, Onkologie und Tumorimmunologie, Charité, Berlin; 4Mechanisms of Leukemogenesis, German Cancer Research Center (DKFZ), Heidelberg, Germany; 5Department of Haematology, St. James's University Hospital, Leeds, UK; 6Universitair Ziekenhuis Gent, Gent, Belgium; 7 Universitair Ziekenhuis Leuven, Leuven, Belgium; 8Kidwai Memorial Institute of Oncology, Bangalore, India; 9Department of Hematology and Cancer Prevention, School of Public Health, Silesian Medical University in Katowice, Katowice, Poland; 10 Department of Haematology-Oncology, University Hospital Brno, Brno, Czech Republic; 11 University of Athens, Laikon General Hospital, Athens, Greece; 12Oncology Global Medicines Development, AstraZeneca, Melbourn, UK; 13GSK Oncology, GlaxoSmithKline, London, UK; 14Novartis Pharma GmbH, Nürnberg, Germany; 15Novartis AG, Basel, Switzerland and 16Department for Hematology, Oncology and Rheumatology, Saarland University Medical School, Homburg/Saar, Germany

ABSTRACT

Correspondence: STEPHAN STILGENBAUER stephan.stilgenbauer@uniklinik-ulm.de Received: June 10, 2019. Accepted: January 7, 2020. Pre-published: January 9, 2020. doi:10.3324/haematol.2019.229161 ©2020 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.

2440

ext generation sequencing studies in chronic lymphocytic leukemia (CLL) have revealed novel genetic variants that have been associated with disease characteristics and outcome. The aim of this study was to evaluate the prognostic value of recurrent molecular abnormalities in patients with CLL. Therefore, we assessed their incidences and associations with other clinical and genetic markers in the prospective multicenter COMPLEMENT1 trial [treatment naive patients not eligible for intensive treatment randomized to chlorambucil (CHL) vs. ofatumumab-CHL (O-CHL)]. Baseline samples were available from 383 patients (85.6%) representative of the total trial cohort. Mutations were analyzed by ampliconbased targeted next generation sequencing (tNGS). In 52.2% of patients we found at least one mutation; the incidence was highest in NOTCH1 (17.0%), followed by SF3B1 (14.1%), ATM (11.7%), TP53 (10.2%), POT1 (7.0%), RPS15 (4.4%), FBXW7 (3.4%), MYD88 (2.6%), and BIRC3 (2.3%). While most mutations lacked prognostic significance, TP53 (HR2.02, P<0.01), SF3B1 (HR1.66, P=0.01), and NOTCH1 (HR1.39, P=0.03) were associated with inferior progression-free survival (PFS) in univariate analysis. Multivariate analysis confirmed the independent prognostic role of TP53 for PFS (HR1.71, P=0.04) and overall survival (OS) (HR2.78, P=0.02), and of SF3B1 for PFS only (HR1.52, P=0.02). Notably, NOTCH1 mutation status separates patients with a strong from those with a weak benefit from addition of ofatumumab to CHL (NOTCH1wt: HR0.50, P<0.01; NOTCH1mut: HR0.81, P=0.45). In summary, TP53 and SF3B1 were confirmed as independent prognostic factors and NOTCH1 as a predictive factor for reduced ofatumumab efficacy in a randomized chemo/immunotherapy CLL trial. These results validate NGS-based mutation analysis in a multicenter trial and provide a basis for expanding molecular testing in the prognostic workup of patients with CLL. (Trial registered at clinicaltrials.gov identifier: NCT00748189.)

haematologica | 2020; 105(10)

Gene mutations in COMPLEMENT1

Introduction Chronic lymphocytic leukemia (CLL) is a heterogeneous disease. Many prognostic factors have been identified in the last decades, but only a few have found their way into clinical practice.1 In a variety of different CLL trials, genomic aberrations, particularly 17p deletion and mutation status for TP532-7 had the strongest relation to clinical outcome. Their assessment before institution of therapy has therefore been recommended for every CLL patient.1 The availability of new CLL treatment modalities have raised the question of the influence of these genomic aberrations as predictive factors for treatment response. Recent insights from unbiased next generation sequencing (NGS) approaches describe more than 40 recurrently mutated cancer driver genes in CLL, the clinical significance of which often remains undefined.8-11 The clinical value of single mutations was mostly studied in heterogeneous patient cohorts and assessed outside of the context of clinical and biologic features.12-16 Therefore, there is a need to perform mutation screening in large clinical trials, as they provide a homogeneous cohort of patients, standardized assessment of clinical and laboratory parameters, and valid outcome data. The UK CLL4 trial and the GCLLSG CLL8 trials served to determine the prognostic impact of recurrent mutations among the minority of young/fit patients. Both cohorts were screened for mutations in TP53, NOTCH1 and SF3B1, as mutations in these genes have a high incidence (10-15%) and typically occur in hotspots, which allows the use of Sanger sequencing for mutation screening. While mutated TP53 (TP53mut) was validated as an independent prognostic factor for progression-free survival (PFS) and overall survival (OS) in both studies, results for the other genes were less clear: mutated SF3B1 (SF3B1mut) was associated with decreased PFS in CLL8 and decreased OS in UK CLL4 but not vice versa.12,13,17 Notably, and in contrast to previous reports, mutated NOTCH1 (NOTCH1mut) was not an independent prognostic factor for PFS in both trials, but was identified as predicting the lack of efficacy of the addition of CD20 antibody (rituximab) in the CLL8 trial;13 this provocative finding needs confirmation in additional data sets. The COMPLEMENT1 trial evaluated chlorambucil (CHL), a less intense chemotherapy backbone for which the majority of CLL patients are eligible, with or without the addition of the CD20 antibody ofatumumab.15,16 Ofatumumab binds to a different epitope of CD20 compared to rituximab and is more efficacious in triggering complement-dependent cytotoxicity (CDC) in vitro. The major result of COMPLEMENT1 was a significantly prolonged PFS by the addition of ofatumumab (22.4 months; 95%CI: 19.0-25.2) to chlorambucil18,19 (13.1 months; 95%CI: 10.6-13.8) in a population who cannot tolerate more intensive therapy. The well characterized patient cohort and mature follow up of COMPLEMENT1 provided an ideal background to study incidence, associations, and prognostic as well as the predictive value of gene mutations in a typical, elderly, front-line CLL patient population.

Methods Patients The GlaxoSmithKline-sponsored phase III trial COMPLEMENT1 (OMB110911) enrolled 447 untreated patients ineligible haematologica | 2020; 105(10)

for fludarabine-based therapy and in need of treatment. Subjects were randomly assigned to receive CHL or O-CHL with a ratio of 1:1. The primary study end point was PFS as determined by an independent review committee. Biomarker analysis was implemented in the study protocol and was approved by the institutional review board or ethics committee of each participating institution. Each patient provided written informed consent before enrolment. For 383 of 447 patients (85.7%), a baseline sample with sufficient DNA and informed consent for research purposes were available. This subset was representative of the intent-to-treat population with regard to clinical, laboratory and genetic baseline characteristics (Online Supplementary Table S1).

Sequencing We designed a customized Illumina™ Truseq amplicon panel for all coding regions of TP53, ATM, BIRC3, MYD88, FBXW7, POT1, and for the most commonly affected exons 14, 15, 16 and 18 in SF3B1 and exon 34 in NOTCH1. The cumulative target size was 24,161 basepairs (bp) covered by 221 amplicons with a length of up to 250 bp each. Adjacent 10 intron bp were included to cover splice site mutations. Input of 250 ng DNA from peripheral blood mononuclear cells isolated by Ficoll gradient centrifugation were sufficient for libraries according to the Illumina TruSeq Custom Amplicon (TSCA) protocol. Sequencing was performed on an Illumina MiSeq™ with the 500-cycle MiSeq Reagent Kit v2. Mutations in the 3’UTR of NOTCH1 and in RPS15 exon 4 were analyzed via Sanger sequencing.

Statistical analysis Statistical analyses were performed on an intent-to-treat basis including all patients with samples available. Analysis was performed using R studio 1.1.447 with survival package (RStudio Inc., Boston, MA, USA). Categorical variables were compared using the Fisher Exact test, and continuous variables were compared using non-parametric rank-sum tests. Statistical tests were two-sided; P<0.05 was considered statistically significant. There were no adjustments for multiple testing, so that all reported Pvalues have an exploratory character for all analysis except for Online Supplementary Figure S2 for which we used false discovery rate (FDR) as an adjustment. Time to event was analyzed by Kaplan-Meier estimates and log-rank. To identify independent prognostic factors, we included treatment arm, del11q, del17p and IGHV mutation status as factors associated with PFS in univariate analysis in addition to all genetic subgroups defined by gene mutations. These 13 parameters were tested in 383 patients without forward or backward selection procedures to explore the independent prognostic character. In addition, we performed a multivariate treatment-gene mutation interaction analysis to explore treatment dependent and therefore predictive value. We used R studio 1.1.447 (RStudio Inc., Boston, MA, USA) with survival package for all statistical analyses.

Results Spectrum and associations of mutations Sequencing of 383 patient samples resulted in a mean coverage of 2,852 reads and 98.2% of target reads above 100x. We identified a total of 304 mutations in the selected gene set. Mutation incidences were NOTCH1 (17.0%), SF3B1 (14.1%), ATM (11.7%), TP53 (10.2%), POT1 (7.0%), RPS15 (4.4%), FBXW7 (3.4%), MYD88 (2.6%), and BIRC3 (2.3%) (Figure 1A). When looking at the muta2441

E. Tausch et al. A

*P<0.05

Figure 1. Incidence and distribution of genetic parameters (gene mutations, genomic aberrations and IGHV status). (A) Cluster diagram of patients (columns) with data for all genetic parameters (rows) (right) and overall incidence (left). Distribution of markers is ordered by rows. (B) Circos plots of the co-occurrence of gene mutations with each other (left) and pairwise with chromosomal aberrations (right). Lengths of arcs correspond to total incidences of respective markers while the width of each ribbon corresponds to the proportion of co-occurrence with a respective second marker. (C) Distribution of gene mutations in the IGHV mutation status subgroups.

tion type, we observed 195 missense mutations, 3 insertions, 56 deletions, 27 nonsense mutations, 11 splice site and 12 3â&#x20AC;&#x2122;UTR mutations. Of all variants, 122 were selected for validation via Sanger sequencing and all of them were confirmed, so that further validation procedures were omitted (Online Supplementary Table S2). While the number of mutations was slightly higher in CHL versus O-CHL (178 vs. 126, not significant), the number of patients with at least one mutation was similar in both arms: 55% for CHL and 53% for O-CHL (see Online Supplementary Figure S1 for a detailed overview). In the total cohort, only 11.5% of patients had neither a mutation nor a chromosomal aberration in the analyzed targets. As previously reported, mutations in SF3B1 and MYD88 were exclusively single nucleotide variants (SNV). Of 66 NOTCH1 mutations, 52 caused stop codons via frameshift or nonsense mutation and 12 affected the 3â&#x20AC;&#x2122;UTR (Online Supplementary Table S2). TP53 and BIRC3 showed various patterns of mutations including insertions, deletions and exonic as well as splice site SNV. Regarding associations, all but one case with MYD88 mutation had mutated IGHV, while TP53, RPS15 and BIRC3 mutations were found predominantly in patients with unmutated IGHV (Figure 1). As to cytogenetics, we found deletion of 11q associated with mutations of ATM and BIRC3 and trisomy 12q associated with mutations in 2442

NOTCH1, BIRC3 and FBXW7. Interestingly, of the nine BIRC3 mutated cases, six had an aberration in 11q and four in 12q. As expected, high concordance was found between 17p deletion and mutation in TP53, but also with FBXW7mut. These associations hold true when considering correction for multiple testing with false discovery rate (Online Supplementary Figure S2). Analyzing mutations in nine different genes and correlation with genomic aberrations and IGHV mutation status allowed us to derive an interaction network based on significant correlations considering incidence, co-occurrence, or mutual exclusivity (Figure 2). Interestingly, two dichotomies emerge from the complex network. Firstly, there is a mutual exclusivity between del13q and +12q. While del13q does not significantly associate with any of the gene mutations, cases with +12q cluster with mutated NOTCH1, BIRC3 and FBXW7. The second dichotomy is found in mutation of TP53 and del17p on one side and a cluster including del11q, +12q and ATM on the other side. Interestingly, mutations of SF3B1 are not associated with either group, suggesting that these mutations represent an independent pathogenic mechanism. Concerning clinical associations, SF3B1 mutations were more common in male patients (Online Supplementary Table S3) and associated with high absolute lymphocyte count (ALC) and CD19+CD5+ fraction by flow cytometry (P<0.01 and haematologica | 2020; 105(10)

Gene mutations in COMPLEMENT1 Table 1. Prognostic associations of gene mutations and progression-free survival (PFS) / overall survival (OS) in the full trial cohort (both treatment arms combined) in univariate analysis (log rank test).

Gene mutation

PFS 95%CI

OS 95%CI

N (events PFS/OS)

TP53 NOTCH1 SF3B1 ATM POT1 RPS15 FBXW7 MYD88 BIRC3

2.02 1.39 1.66 1.16 1.25 1.08 1.51 0.59 1.63

1.18-3.45 1.04-1.86 1.12-2.47 0.84-1.60 0.73-1.92 0.62-1.87 0.67-3.39 0.31-1.11 0.69-3.87

<0.01 0.03 <0.01 0.42 0.33 0.81 0.22 0.19 0.23

4.25 0.84 1.39 0.89 1.15 2.06 2.19 NA 1.29

1.65-10.92 0.49-1.44 0.67-2.87 0.48-1.68 0.50-2.58 0.72-5.93 0.52-9.33 NA 0.34-4.92

<0.01 0.62 0.32 0.78 0.77 0.11 0.12 NA 0.72

39 (27/16) 65 (50/10) 54 (43/11) 45 (34/7) 29 (21/5) 17 (10/5) 13 (9/4) 10 (6/0) 9 (6/2)

NA: not applicable due to no events in mutated patients. CI: Confidence Interval; N: number. Statistical significance in bold.

P=0.03), whereas ALC in MYD88 mutated patients was low. WBC >50x109/L were found more often in patients with mutated TP53, RPS15 or ATM (P=0.05, P=0.02 and P=0.03, respectively).

Clinical outcome and prognostic impact of gene mutations in the full cohort A significantly lower overall response rate was found only for TP53 mutated cases (HR 5.20, P<0.01). In univariate analysis, significantly decreased PFS was found for patients with mutations in TP53 (HR 2.02, P<0.01), SF3B1 (HR 1.66, P<0.01), and NOTCH1 (HR 1.39, P=0.03), but not for patients with mutations in ATM (HR 1.16, P=0.42) or BIRC3 (HR 1.63, P=0.23) (Table 1 with 95%CI, Figure 3, and Online Supplementary Figure S3). OS was significantly shorter in patients with mutations of TP53, and this was observed both in the total cohort (HR 4.25, P<0.01) as well as in patients without deletion of 17p (HR 2.56, P=0.03) (data not shown). Remarkably, no death event was observed among MYD88 mutated patients. The number of mutated genes correlated with PFS, and patients harboring two or more mutated genes had shorter PFS (Online Supplementary Figure S4). Mutation in at least one of the target genes associated with significantly shorter OS, while the number of mutated genes did not play a significant role. Taking into account that a disruption of both alleles of BIRC3 or ATM could be required to observe any effect as described in the UK CLL420 trial, we considered the del11q status for the impact of mutations in both genes. However, in our patient cohort, mutation of ATM did not add significant prognostic value either in the 11q deleted subgroup or in the 11q disomic subgroup (Online Supplementary Figure S5). As BIRC3 was mutated in only nine patients, the subgroups were too small to address this question. Mutations in TP53 and ATM showed high variant allelic fractions (VAF) of the mutant allele, mainly explained by a high co-occurrence with deletion of the other allele (Online Supplementary Figure S6A). Also mutations with a variant allele fraction (VAF) â&#x2030;¤10%, usually undetectable by Sanger sequencing, were rare in TP53 (2 of 38) and absent in MYD88 in this trial, but present in a significant proportion of cases with mutation of SF3B1, NOTCH1 or FBXW7. As the role of mutations with minor allelic fraction is still unclear, we performed survival analyses including patients with wild-type (WT) SF3B1/NOTCH1 and haematologica | 2020; 105(10)

with major and minor (< 10% VAF) mutated fractions (only NGS data considered). Interestingly, patients affected by only minor SF3B1 mutated clones showed significantly shorter PFS in comparison to WT (HR 3.09, 95%CI: 0.67-14.30, P=0.01), while for NOTCH1, minor mutations had no significant impact on the full cohort, mainly due to low numbers (HR 1.54, 95%CI: 0.71-3.34, P=0.26) (Online Supplementary Figure S6B and C). We performed multivariable analyses including treatment arm, IGHV mutation status, del17p, del11q and gene mutations to examine the independent prognostic value of these parameters. Besides ofatumumab+chlorambucil, IGHV status, del17p and del11q, only mutations of TP53 (HR 1.71, P=0.04) and SF3B1 (HR 1.52, P=0.02) were identified as independent factors associated with decreased PFS. For OS, only presence of del17p and TP53 mutation (HR 2.78, P=0.02) retained a significant independent role (Table 2 with 95%CI).

Predictive value for the efficacy of ofatumumab addition Randomization of patients into treatment arms with and without ofatumumab allowed the evaluation of differential CD20 antibody efficacy in subgroups defined by mutations. In the total cohort of 383 patients, treatment with ofatumumab was beneficial with regard to PFS (HR 0.53, P<0.01) as published for the whole trial population.19 Analyses of both treatment arms separately generally reiterated the results obtained for the total cohort, with mutations in SF3B1 and TP53 being associated with shorter PFS. However, for NOTCH1, we observed an impact on PFS in the O-CHL treatment arm (HR 1.94, 95%CI: 1.253.92, P<0.01) but not with CHL alone (HR 1.01, 95%CI: 0.69-1.47, P=0.98) (Figure 4). Conversely, the addition of ofatumumab to chlorambucil in patients with NOTCH1wt status was strongly beneficial (mPFS 23.8 vs. 13.3 months, HR 0.50, 95%CI: 0.39-0.63, P<0.01), while the benefit in the NOTCH1mut group was not significant (17.2 vs. 13.1 months, HR 0.81, 95%CI: 0.50-1.31, P=0.45). Notably, the same analysis confirmed the addition of ofatumumab to be beneficial in SF3B1 and TP53 mutated subgroups (for SF3B1: mPFS 17.3 vs. 10.8 months, HR 0.53, 95%CI: 0.290.97, P=0.03; for TP53: mPFS 12.8 vs. 3.7 months, HR 0.49, 95%CI: 0.23-1.05, P=0.05). This impact was strongest with NOTCH1 mutations at a mutant allele fraction >40% and weaker with smaller NOTCH1 variant fraction (Online Supplementary Figure S7). To investigate the rela2443

E. Tausch et al.

Line width = P-value P=0.1-0.01 P=1x0-2-1x0-3 P<1x0-3

Figure 2. Visualization of co-occurrence of gene mutations and genomic aberrations based on pairwise Fisher exact test. Line length corresponds to â&#x2C6;&#x161;(1/odds ratio). Therefore, lines with a length >1 show mutual exclusivity (red) and lines with a length <1 co-occurrence (blue). Line width corresponds to stated P-value of pairwise comparison; when P>0.1, no line is depicted. Font size characterizes incidence of mutation/aberration; green indicates association with mutated IGHV, and yellow indicates association with unmutated IGHV (P<0.1 each). FBXW7 is depicted twice.

tion between treatment and NOTCH1 mutational status, we performed an interaction-focused multivariate test. This analysis attributed a predictive impact to NOTCH1 (P=0.05), while for all other parameters no interaction was observed (Online Supplementary Table S4). There were no differences in CD20 surface levels measured with flow cytometry between subgroups defined by mutation of NOTCH1 suggesting that differential CD20 expression is not the cause of the lower CD20 antibody efficacy (data not shown).

Discussion Over recent years, in an increasing number of laboratories, tNGS has replaced Sanger sequencing as a tool for mutation analysis. The current report represents a combined analysis of recurrent gene mutations studied by tNGS with a comprehensive dataset in a large multicenter phase III trial in CLL. In contrast to prior analyses, we used amplicon-based tNGS which allowed sequencing of 119 exons in eight different genes with a detection limit of 5% VAF. Although some tNGS approaches allow detection of subclones with only 1% VAF and below,21 the aim of this project was to evaluate the prognostic value of molecular abnormalities for a better refinement of risk stratification. Development of such a robust, reproducible and affordable assay, which can be easily adopted into a diagnostic setup, could help in the clinical management of CLL patients. In favor of a uniform coverage of all targets, we trimmed sensitivity to achieve reliable results with high specificity. Mutations with minor allelic fractions are not detectable via Sanger sequencing and there is little evidence of their clinical impact, apart from in TP53.21,22 Therefore, not surprisingly, councils like the European Initiative for Research in CLL (ERIC) caution against deriving clinical conclusions from small subclonal mutations.23 2444

The low number of detected minor mutations in TP53 and other genes in COMPLEMENT1 does not allow any conclusions to be drawn about their prognostic significance. However, for minor SF3B1 mutations, we found an impact similar to mutations with VAF>10%. Further studies on clinical trials focusing on minor mutations are necessary to explore the impact of such variants with a comprehensive validation setup. Our data support prior observations characterizing SF3B1, NOTCH1, ATM and TP53 as the most frequently mutated genes in decreasing order of frequency, while mutations in BIRC3 and MYD88 remain very rare events in untreated CLL.10,11 However, putting associations of these mutations and chromosomal aberrations in a correlation network uncovers two substantial associations. First, the mutual exclusivity between del13q and +12q and associated mutations, which is supported by prior clonality and lineage analysis showing that both aberrations are clonal events in CLL associating with tumor pathogenesis, but independently of each other.10 Second, intermediate events, namely del17p and TP53 on one side and different genomic abnormalities, mainly connected to 11q and ATM on the other side. As these abnormalities associate with growth advantage and accumulation in pretreated CLL, one can envisage that either mutation of TP53 or mutation of one of the other genes is sufficient to cause a clonal survival advantage and therefore make the respective other event redundant. Besides genetic interactions, analysis of the impact on clinical parameters, and especially on outcome after therapy, was the major aim of this study. In the current COMPLEMENT1 data set, TP53 was the only gene mutation associated with a decrease in PFS and OS independent of other prognostic factors in multivariable analyses. Notably, this correlation was observed despite the high degree of collinearity with 17p deletion, confirming the need of TP53 mutation testing in addition to 17p deletion haematologica | 2020; 105(10)

Gene mutations in COMPLEMENT1

Figure 3. Kaplan-Meier estimates of progression-free survival (PFS) (left) and overall survival (OS) (right) according to the status of selected gene mutations for the total patient cohort. Red lines: mutated (mut) subgroups; blue lines: wild-type (wt). Denoted P-values were calculated by log-rank test (mut vs. unmutated subgroup).

Table 2. Multivariate analysis based on Cox-regression for progression-free survival (PFS) / overall survival (OS).

Parameter present

PFS 95%CI

OS 95%CI

N (events PFS/OS)

O-CHL arm TP53 NOTCH1 SF3B1 ATM POT1 RPS15 FBXW7 MYD88 BIRC3 IGHVunmut del17p del11q

0.47 1.71 1.32 1.52 1.02 1.28 0.82 1.15 0.72 1.12 1.46 2.90 1.72

0.35-0.61 1.04-2.81 0.95-1.84 1.06-2.17 0.67-1.56 0.79-2.1 0.42-1.59 0.58-2.29 0.29-1.78 0.47-2.66 1.09-1.95 1.49-5.67 1.19-2.51

<0.01 0.04 0.10 0.02 0.93 0.32 0.56 0.69 0.47 0.79 0.01 <0.01 <0.01

0.76 2.78 0.88 1.31 1.01 1.65 1.16 1.64 NA 0.63 1.72 3.22 1.94

0.44-1.29 1.17-6.62 0.43-1.81 0.62-2.8 0.43-2.38 0.62-4.4 0.44-3.06 0.56-4.83 NA 0.15-2.69 0.95-3.13 1.18-8.76 0.98-3.83

0.31 0.02 0.74 0.48 0.98 0.31 0.76 0.37 NA 0.53 0.08 0.02 0.06

183 (116/28) 39 (27/16) 65 (50/10) 54 (43/11) 45 (34/7) 29 (21/5) 17 (10/5) 13 (9/4) 10 (6/0) 9 (6/2) 199 (141/44) 19 (12/10) 58 (46/14)

NA: not applicable due to no events in mutated patients. CI: Confidence Interval; N: number. Statistical significance in bold.

diagnostics in routine practice, as recommended by current ERIC and International Workshop on CLL (iwCLL) guidelines.1 Accordingly, as only about half of the TP53 mutated cases harbor an additional 17p deletion, mutation status for TP53 should be assessed in addition to del17p to allocate patients to treatment with novel agents. However, this subgroup accounts for just over 10% (here 11.7%) of front-line CLL cases, and does not identify all patients with poor response and short survival times. Therefore, the major question was whether other genetic factors characterize aggressive CLL to the same degree. In previous studies, this had been shown for disrupted BIRC3, that affected the response to therapy and PFS in an extent comparable to mutation of TP53.24 In COMPLEMENT1, only nine patients were mutated in BIRC3, having an overall response rate (ORR) of 75% in contrast to 47% responders among TP53 mutated cases. While TP53 significantly shortened PFS (HR 2.02, P<0.01) and OS (HR haematologica | 2020; 105(10)

4.25, P<0.01), BIRC3 did not associate with either (HR 1.63 for PFS and 1.29 for OS, both not significant). The same applies to mutations in ATM, which together with mutations in BIRC3 associate with del11q, but do not add any additional prognostic value to the impact of 11q on PFS and OS in this trial. This is in contrast to data published within the UK CLL4 trial,20 which may be explained by a different treatment and the lack of a CD20 antibody in UK CLL4, a different filtering approach for ATM variants, or just smaller patient numbers. However, also other groups did not find an adverse impact of a biallelic inactivation in ATM,25,26 and therefore a more comprehensive setup with a bigger patient cohort with matched germline samples for validation of the somatic origin of these variants is required to definitively answer this question. In contrast, for mutations of SF3B1, we observed a higher risk for early progression independently from other prognostic factors, but this was less pronounced as com2445

E. Tausch et al.

HR=1.94, P<0.01

HR=1.01, P<0.98

Figure 4. Kaplan-Meier estimates of progression-free survival (PFS) in the chlorambucil + ofatumumab (left) and chlorambucil alone (right) treatment arms according to NOTCH1 mutation status. Mutated subgroups are depicted by red lines, wild type by blue lines. Denoted P-values were calculated by log-rank test (mutated vs. unmutated subgroup).

pared to TP53 or del17p. These results conform to a number of previous observations.8,13,27,28 Some approaches integrated SF3B1 and additional novel mutations into the hierarchical classification for prognostication established by Döhner in 2000.3 This is challenging as coincidences like mutations of NOTCH1 and +12q or ATM and del11q are not easy to resolve in these models and currently available data are not powered to address correlations of small subgroups with clinical parameters. In a recent analysis, del11q and mutations of SF3B1 and NOTCH1 were categorized as intermediate risk.29 Although del11q and SF3B1 mutation showed similar hazard ratios for PFS (1.72 and 1.52) in the multivariate analysis of COMPLEMENT1, the former shows a trend to inferior OS (HR 1.94, P=0.06) while SF3B1mut did not (HR 1.31, P=0.48). In addition, the prognostic impact of NOTCH1 appeared to strongly depend on the type of therapy (see below and Stilgenbauer et al.13), indicating that its use as a prognostic marker must be approached with caution. As in previous genomic studies, we observed an adverse outcome of NOTCH1mut patients on the total trial cohort, which is more pronounced in cases with coding mutations. However, when analyzing both treatment arms separately, this prognostic impact turned out to be due to differential treatment effect, i.e. it identified NOTCH1 mutation as a predictive factor. This finding reiterates the result obtained with rituximab, confirming a lesser efficacy of type 1 CD20 antibodies in NOTCH1 mutated CLL13 compared to WT. The impact on PFS is strongest in patients with a high mutant allele burden in NOTCH1 and less pronounced in patients with minor mutations, but can only be observed with O-CHL and not with CHL treatment. This predictive value remains after adding 3’UTR mutations in NOTCH1 and persisted in a treatment interaction analysis in contrast to all other markers. This may explain the discrepancies in previously published data: mutation of NOTCH1 was an independent prognostic factor for PFS in heterogeneous cohorts of patients mainly treated with CD20 antibodies as the current standard of therapy but was not in the UK LRF CLL4 2446

trial that did not contain CD20 antibody treatment.12,14,15,30,31 Interestingly, in contrast to the previously published relation between NOTCH1 mutation and CD20 expression,32 NOTCH1mut cases showed no difference in CD20 surface expression as analyzed by flow cytometry. This is in line with the results from the CLL8 trial but differs from recently published data.13,32 Even though the underlying molecular interrelation of NOTCH1 mutations with CD20 remains unclear, the predictive impact of NOTCH1 mutation is now confirmed in two independent clinical phase III trial cohorts. Although O-CHL shows slightly longer mPFS than CHL in NOTCH1mut patients, it is noteworthy that both mutation subgroups, NOTCH1mut and TP53mut, have a similar initial impact in the setting of chemo-immunotherapy with CD20-targeted therapy outcome (HR on PFS 1.88 for TP53mut and 1.94 for NOTCH1mut). Moreover, the proportion of patients affected by mutated NOTCH1 in the front-line setting is bigger than the group defined by TP53 mutation. Based on current treatment guidelines, chemoimmunotherapy is still a valuable option for a number of patients.33 Our findings raise a note of caution on the use of such therapy in a significant subset of CLL patients with mutated NOTCH1. Furthermore, CD20 antibodies remain important elements in combination with novel compounds e.g. with venetoclax34 or ibrutinib,35 and it is still unclear whether specific subgroups have a particular benefit of such combinations in comparison to single agent. The major implication from our observation is the need to understand and circumvent the resistance against CD20 antibodies as they remain an important element in the treatment of CLL. Furthermore, our results underline the role of recurrent mutations also in trials with novel treatment principles such as BTK-, PI3K- and BCL2 inhibition. Acknowledgments The authors thank all patients, their families and their physicians for trial participation and donation of samples. The authors thank Simon Müller for supporting experiments. haematologica | 2020; 105(10)

Gene mutations in COMPLEMENT1

Funding This work was supported by the Else Kröner-FreseniusStiftung (2010_Kolleg24), EC (01KT1601, FIRE CLL), BMBF

References 1. 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. 2. Zenz T, Eichhorst B, Busch R, et al. TP53 mutation and survival in chronic lymphocytic leukemia. J Clin Oncol. 2010; 28(29):4473-4479. 3. Döhner H, Stilgenbauer S, Benner A, et al. Genomic Aberrations and Survival in Chronic Lymphocytic Leukemia. N Engl J Med. 2000;343(26):1910-1916. 4. Ghia P, Stamatopoulos K, Belessi C, et al. Geographic patterns and pathogenetic implications of IGHV gene usage in chronic lymphocytic leukemia: the lesson of the IGHV3-21 gene. Blood. 2005;105(4):16781685. 5. Grever MR, Lucas DM, Dewald GW, et al. Comprehensive assessment of genetic and molecular features predicting outcome in patients with chronic lymphocytic leukemia: results from the US Intergroup Phase III Trial E2997. J Clin Oncol. 2007; 25(7):799-804. 6. 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. 7. Catovsky D, Richards S, Matutes E, et al. Assessment of fludarabine plus cyclophosphamide for patients with chronic lymphocytic leukaemia (the LRF CLL4 Trial): a randomised controlled trial. Lancet. 2007; 370(9583):230-239. 8. Wang L, Lawrence MS, Wan Y, et al. SF3B1 and other novel cancer genes in chronic lymphocytic leukemia. N Engl J Med. 2011; 365(26):2497-2506. 9. Puente XS, Pinyol M, Quesada V, et al. Whole-genome sequencing identifies recurrent mutations in chronic lymphocytic leukaemia. Nature. 2011;475(7354):101105. 10. Landau DA, Tausch E, Taylor-Weiner AN, et al. Mutations driving CLL and their evolution in progression and relapse. Nature. 2015;526(7574):525-530. 11. Puente XS, Beà S, Valdés-Mas R, et al. Noncoding recurrent mutations in chronic lymphocytic leukaemia. Nature. 2015; 526(7574):519-524. 12. Oscier DG, Rose-Zerilli MJJ, Winkelmann N, et al. The clinical significance of NOTCH1 and SF3B1 mutations in the UK LRF CLL4 trial. Blood. 2013;121(3):468-475. 13. Stilgenbauer S, Schnaiter A, Paschka P, et al.

haematologica | 2020; 105(10)

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

(031L0076C PRECISe), and Deutsche Forschungsgemeinschaft (SFB 1074 projects B1, B2). Genetic analyses were supported by GlaxoSmithKline and Novartis.

Gene mutations and treatment outcome in chronic lymphocytic leukemia: results from the CLL8 trial. Blood. 2014;123(21):32473254. Baliakas P, Hadzidimitriou A, Sutton L-A, et al. Recurrent mutations refine prognosis in chronic lymphocytic leukemia. Leukemia. 2015;29(2):329-336. Fabbri G, Rasi S, Rossi D, et al. Analysis of the chronic lymphocytic leukemia coding genome: role of NOTCH1 mutational activation. J Exp Med. 2011;208(7):1389-1401. Weissmann S, Roller A, Jeromin S, et al. Prognostic impact and landscape of NOTCH1 mutations in chronic lymphocytic leukemia (CLL): a study on 852 patients. Leukemia. 2013;27(12):2393-2396. Larrayoz M, Rose-Zerilli MJJ, Kadalayil L, et al. Non-coding NOTCH1 mutations in chronic lymphocytic leukemia; their clinical impact in the UK CLL4 trial. Leukemia. 2017;31(2):510-514. Gupta IV, Jewell RC. Ofatumumab, the first human anti-CD20 monoclonal antibody for the treatment of B cell hematologic malignancies. Ann N Y Acad Sci. 2012; 126343-126356. Hillmen P, Robak T, Janssens A, et al. Chlorambucil plus ofatumumab versus chlorambucil alone in previously untreated patients with chronic lymphocytic leukaemia (COMPLEMENT 1): a randomised, multicentre, open-label phase 3 trial. Lancet. 2015;385(9980):1873-1883. Rose-Zerilli MJJ, Forster J, Parker H, et al. ATM mutation rather than BIRC3 deletion and/or mutation predicts reduced survival in 11q-deleted chronic lymphocytic leukemia: data from the UK LRF CLL4 trial. Haematologica. 2014;99(4):736-742. Rossi D, Khiabanian H, Spina V, et al. Clinical impact of small TP53 mutated subclones in chronic lymphocytic leukemia. Blood. 2014;123(14):2139-2147. Malcikova J, Stano-Kozubik K, Tichy B, et al. Detailed analysis of therapy-driven clonal evolution of TP53 mutations in chronic lymphocytic leukemia. Leukemia. 2015; 29(4):877-885. Pospisilova S, Sutton L-A, Malcikova J, et al. Innovation in the prognostication of chronic lymphocytic leukemia: how far beyond TP53 gene analysis can we go? Haematologica. 2016;101(3):263-265. Rossi D, Fangazio M, Rasi S, et al. Disruption of BIRC3 associates with fludarabine chemorefractoriness in TP53 wild-type chronic lymphocytic leukemia. Blood. 2012;119(12):2854-2862. Hernández JÁ, Hernández-Sánchez M, Rodríguez-Vicente AE, et al. A low frequency of losses in 11q chromosome is associated with better outcome and lower

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

rate of genomic mutations in patients with chronic lymphocytic leukemia. PLoS One. 2015;10(11):e0143073. Nadeu F, Delgado J, Royo C, et al. Clinical impact of clonal and subclonal TP53, SF3B1, BIRC3, NOTCH1, and ATM mutations in chronic lymphocytic leukemia. Blood. 2016;127(17):2122-2130. Jeromin S, Weissmann S, Haferlach C, et al. SF3B1 mutations correlated to cytogenetics and mutations in NOTCH1, FBXW7, MYD88, XPO1 and TP53 in 1160 untreated CLL patients. Leukemia. 2014;28(1):108117. Rossi D, Bruscaggin A, Spina V, et al. Mutations of the SF3B1 splicing factor in chronic lymphocytic leukemia: association with progression and fludarabine-refractoriness. Blood. 2011;118(26):6904-6908. Rossi D, Rasi S, Spina V, et al. Integrated mutational and cytogenetic analysis identifies new prognostic subgroups in chronic lymphocytic leukemia. Blood. 2013; 121(8):1403-1412. Bo MD, Del Principe MI, Pozzo F, et al. NOTCH1 mutations identify a chronic lymphocytic leukemia patient subset with worse prognosis in the setting of a rituximab-based induction and consolidation treatment. Ann Hematol. 2014;93(10): 1765-1774. Del Giudice I, Rossi D, Chiaretti S, et al. NOTCH1 mutations in +12 chronic lymphocytic leukemia (CLL) confer an unfavorable prognosis, induce a distinctive transcriptional profiling and refine the intermediate prognosis of +12 CLL. Haematologica. 2012;97(3):437-441. Pozzo F, Bittolo T, Arruga F, et al. NOTCH1 mutations associate with low CD20 level in chronic lymphocytic leukemia: evidence for a NOTCH1 mutation-driven epigenetic dysregulation. Leukemia. 2016;30(1):182189. Fürstenau M, Hallek M, Eichhorst B. Sequential and combination treatments with novel agents in chronic lymphocytic leukemia. Haematologica. 2019; 104(11):2144-2154. Kater AP, Seymour JF, Hillmen P, et al. Fixed duration of venetoclax-rituximab in relapsed/refractory chronic lymphocytic leukemia eradicates minimal residual disease and prolongs survival: post-treatment follow-up of the MURANO Phase III Study. J Clin Oncol. 2019;37(4):269-277. 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.

2447

ARTICLE Ferrata Storti Foundation

Plasma Cell Disorders

ATR addiction in multiple myeloma: synthetic lethal approaches exploiting established therapies

Oronza A. Botrugno,1 Silvia Bianchessi,2 Desirée Zambroni,3 Michela Frenquelli,1 Daniela Belloni,4 Lucia Bongiovanni,5 Stefania Girlanda,6 Simona Di Terlizzi,7 Marina Ferrarini,4 Elisabetta Ferrero,4 Maurilio Ponzoni,5 Magda Marcatti6 and Giovanni Tonon1,8

Haematologica 2020 Volume 105(10):2448-2456

Functional Genomics of Cancer Unit, Experimental Oncology Division, IRCCS San Raffaele Scientific Institute; 2Laboratory of Lymphoid Organ Development, Division of Experimental Oncology, IRCCS San Raffaele Scientific Institute; 3ALEMBIC, Advanced Light and Electron Microscopy Bio-Imaging Center, IRCCS San Raffaele Scientific Institute; 4B-Cell Neoplasia Unit, Division of Experimental Oncology, IRCCS San Raffaele Scientific Institute; 5Pathology Unit, IRCCS San Raffaele Scientific Institute; 6Hematology and Bone Marrow Transplantation Unit, IRCCS San Raffaele Hospital; 7FRACTAL, Flow Cytometry Resource Advanced Cytometry Technical Applications Laboratory, IRCCS San Raffaele Scientific Institute and 8Center for Omics Sciences, IRCCS San Raffaele Scientific Institute, Milan, Italy 1

ABSTRACT

Correspondence: GIOVANNI TONON tonon.giovanni@hsr.it Received: January 7, 2019. Accepted: November 13, 2019. Pre-published: November 14, 2019.

herapeutic strategies designed to interfere with cancer cell DNA damage response have led to the widespread use of PARP inhibitors for BRCA1/2-mutated cancers. In the haematological cancer multiple myeloma, we sought to identify analogous synthetic lethality mechanisms that could be exploited in established cancer treatments. The combination of ATR inhibition using the compound VX-970 with a drug eliciting interstrand cross-links, melphalan, was tested in in vitro, ex vivo, and most notably in in vivo models. Cell proliferation, induction of apoptosis, tumor growth and animal survival were assessed. The combination of ATM inhibition with a drug triggering double strand breaks, doxorucibin, was also analyzed. We found that ATR inhibition is strongly synergistic with melphalan, even in resistant cells. The combination was dramatically effective in targeting myeloma primary patient cells and cell lines by reducing cell proliferation and inducing apoptosis. The combination therapy significantly reduced tumor burden and prolonged survival in animal models. Conversely, ATM inhibition only marginally impacted on myeloma cell survival, even in combination with doxorucibin at high doses. These results indicate that myeloma cells extensively rely on ATR, but not on ATM, for DNA repair. Our findings postulate that adding an ATR inhibitor such as VX-970 to established therapeutic regimens may provide a remarkably broad benefit to myeloma patients.

doi:10.3324/haematol.2018.215210

Introduction ©2020 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.

2448

Inducing DNA damage in cancer cells for treatment purposes has been one of the mainstay in oncology for the past decades, and arguably remains one of the most effective strategies to induce cell death of epithelial and haematological cancers alike, to this day.1 Despite their effectiveness, one major limitation of the compounds eliciting DNA damage is represented by their poor specificity.1 Indeed, their administration quickly reaches dose-limiting side effects that are associated with unbearable toxicity. A very active research field is therefore aiming to identify synthetic lethal approaches,2,3 whereby genes and pathways within the DNA repair network are targeted to specifically increase the sensitivity of cancer cells endowed with specific genetic lesions, or towards DNA damaging agents.4 This quest has culminated in the identification of PARP inhibition as a means to trigger apoptosis in cancer cells presenting somatic or hereditary mutations in the BRCA1 and BRCA2 genes,5,6 which has profoundly modified the treatment of several tumor types, haematologica | 2020; 105(10)

ATR dependancy in multiple myeloma

including breast and ovarian carcinomas.7 However, only a small subset of tumors, arising in specific tissues, present somatic mutations in BRCA1 or BRCA2 genes, where PARP inhibitors can be exploited. While cancers not bestowed with these mutations nevertheless may contain other genomic or molecular “BRCAness” signatures that make them sensitive to PARP inhibition,8 it is imperative to discover additional synthetic lethality strategies that can be deployed to improve the treatment and the outcome of cancer patients. Towards this goal, one of the most enticing paths calls upon the inhibition of specific genes implicated in DNA repair, to complement and synergize with established DNA damaging agents.9 The vast majority of therapeutic regimens for the treatment of cancer patients include DNA damaging agents. The hematological cancer multiple myeloma (MM), is a particular case as it exhibits a still incurable clonal proliferation of malignant plasma cells.10 The alkylating agent melphalan was introduced in 1958 for the treatment of MM11 (later in association with prednisone), a landmark event in the history of the treatment of this disease, since there was no effective treatment for this cancer up to then.12 This treatment has remained the benchmark therapy for myeloma patients ever since.13 With regards to the mechanism of action of melphalan, it elicits cancer cell death by triggering interstrand DNA crosslinks (ICL), like other nitrogen mustards including chlorambucil and cyclophosphamide, still widely used for the treatment of various haematological cancers.4 The phosphoinositide 3-kinase (PI3K)-related kinases ATM and ATR control and coordinate the entire DNA damage response.14 ATM primarily orchestrates the global response to double-strand breaks (DSB). On the other hand, ATR is essential in relieving DNA replicative stress. ATR is endowed with an additional, less explored role, related to the repair of ICL, thus engaging the Fanconi anemia (FA) pathway. Therefore, ATM and ATR represent ideal candidates for targeted therapies aiming to unravel DNA repair in the presence of induced DNA damage. To this end, several ATM and ATR inhibitors have been recently developed.15,16 In this study, we comprehensively assessed the role of DNA damage response inhibition, namely of ATR and ATM, in MM, and analyzed if drugs, commonly used to treat MM patients, engage these pathways. We also assayed whether synthetic lethal approaches could be exploited, combining drugs used in the clinic, with ATM and ATR inhibition.

Methods MM cell lines and patient samples MM cell lines MM1.S, H929, KMS20, RPMI 8226, LP1, OPM2, U266, were kindly provided by fellow scientists or purchased from American Type Culture Collection (ATCC). Cell lines were authenticated by short tandem repeat (STR) analysis (Cell ID™ System, Promega, Madison, WI, USA) and routinely tested for the presence of mycoplasma contamination. MM1.S-Luc and U266-Luc cells stably expressing luciferase were generated by transduction with a third generation lentiviral vector carrying the luciferase gene. pLenti PGK V5-LUC Neo (w623-2) was a gift from Eric Campeau (Addgene plasmid # 21471). Primary MM cells were collected from bone marrow (BM) aspihaematologica | 2020; 105(10)

rates through positive selection with anti-CD138 coated magnetic nanoparticles (Robosep, Stemcell Technologies, Vancouver, Canada).17 Samples from patients were obtained upon written informed consent. This study was carried out in accordance with protocols approved by the Institutional Review Board, and the procedures followed were in accordance with the Declaration of Helsinki of 1975, revised in 2000.

Cell viability assays Cell viability was measured with CellTiter-Glo (Promega) according to the manufacturer’s instructions. Luminescence reading was expressed as percentage relative to the DMSO or PBStreated control cells. The experimental data (percentage of viable cells compared to control) were analyzed independently using the Combenefit software.18

Mice, bioluminescent imaging and pharmacological treatments All mice were housed and bred in the institutional pathogenfree animal facility, treated in accordance with the European Union guidelines and with the approval of the San Raffaele Scientific Institute Institutional Ethical Committee. Rag2−/−γc−/− mice on a BALB/c background were kindly provided by CIEA (Central Institute for Experimental Animals, Kawasaki, Japan) and Taconic (Rensselaer, New York, NY, USA). Mice were injected intravenously with 5x106 luciferase expressing cells in 200 mL of PBS and monitored for myeloma progression by bioluminescent imaging (BLI) using the IVIS SpectrumCT System (Perkin Elmer, USA). Treatment commenced when the tumor burden became detectable. Mice were randomized into four treatment groups of five animals each. Treatment cycles consisted of 5 days of treatment followed by 2 days of rest. A total of three treatment cycles were given. VX-970, 60 mg/kg was administered by oral gavage once a day continuously for 5 days. Melphalan, 2 mg/kg was administered by intraperitoneal injection once a day on day 1, 3 and 5 of each treatment cycle. Mice were euthanized by CO2 inhalation when they became detectably ill and developed hind limb paralysis.

Immunohistochemistry on human BM biopsies and three-dimensional culture of primary cells

Four mm thick sections were obtained from Bouin-fixed, paraffin-embedded tissue blocks of BM biopsies of untreated patients with a diagnosis of MM. Three-dymensional (3D) dynamic culture was performed using the RCCSTM bioreactor RCCS-1 (Synthecon Inc., Houston, TX, USA).17 Scaffold discs populated with MM cells were treated with either VX-970 (0.3 mM) or melphalan (1.2 mM) or both for 72 hours (h). At the end of the culture period, cells were recovered and stained with PC7-conjugated anti-CD38 (#560677) and FITC-conjugated Annexin V (#556547) both from BD Pharmingen (San Diego, CA, USA) before flow cytometric (FACS) analysis (FC500, Beckman Coulter, Brea, CA, USA).17

Results ICL-inducing melphalan activates the ATM and ATR pathways We first tested whether established DNA damaging agents used to treat MM patients activate the pathways commonly engaged in the DNA damage response. We first treated MM cells with melphalan. This alkylating agent triggers DNA ICL, which require ATR for their 2449

O.A. Botrugno et al.

Figure 1. VX-970 selectively attenuates ATR-CHK1 signaling axis elicited by DNA damaging agents in multiple myeloma cells. Exponentially growing MM1.S, H929, KMS20 and OPM2 cells were either left untreated (NT) or treated with increasing concentrations of VX-970 (0.15 and 0.3 mM). Treatment with VX-970 was initiated 1 hour (h) before the addition of hydroxyurea (HU, 2 mM) or melphalan (50 μM). After 3 h cells were harvested and analyzed by immunoblotting for the expression of the indicated proteins. Vinculin for MM1.S and GAPDH for the other cell lines, were used as loading controls. The asterisks indicate unspecific bands.

resolution.4 As a control to assay the engagement of the ATR pathway, cells were also treated with hydroxyurea that elicits replicative stress, which also activates ATR (Figure 1). As cellular model systems, we tested several MM cell lines featuring ongoing DNA damage19 (Online Supplementary Figure S1A-B). The treatment with melphalan triggered, as anticipated, DNA damage, assayed with γH2AX, and elicited a DNA damage response, measured through the increase in total and phosphorylated P53 (Figure 1 and Online Supplementary Figure S2). Additionally, both the ATM and ATR pathways were engaged and activated, as demonstrated by the robust phosphorylation of their downstream main targets, pCHK2, and pCHK1 and pRAD17, respectively. All together, these results suggest that ATR and ATM are actively engaged in the DNA damage response elicited by ICL in MM cells.

ATR inhibition is strongly synergistic with melphalan We then assessed whether ATR inhibition impacted on the response of MM cells to melphalan. ATR inhibitors have been recently proposed as important cancer drug treatments.16 We used a novel derivative of the ATR inhibitor VE-821, VE-822 (also known as VX-970)20,21 in a panel of MM cell lines (Online Supplementary Figure S3 and Online Supplementary Table S1). We found that VX-970 was by far the most effective ATR inhibitor in reducing MM cells survival, when compared to VE-821, the compound that we previously tested22 and a recently reported structurally unrelated ATR inhibitor, AZD6738.23 We hence exposed several MM cell lines to increasing concentrations of VX-970, with or without melphalan. ATR inhibition reduced the overall levels of DNA damage and the DNA damage response triggered by melphalan. Along 2450

similar lines, CHK1 and RAD17 phosphorylation as well as γH2AX were markedly reduced when combining the ICL-inducing compounds with VX-970, while pCHK2 levels were unchanged (Figure 1 and Online Supplementary Figure S2). We then explored whether ATR inhibition could synergize with the ICL-inducing compound melphalan. VX-970 alone consistently reduced cellular proliferation (Figure 2 A, dotted vertical line). Strikingly, this effect was profoundly enhanced when VX-970 was combined with melphalan (Figure 2 A-B and Online Supplementary Table S2). Specifically, some cell lines (MM1.S, NCI-H929) were sensitive to either melphalan or VX-970, and yet the combination of VX-970 with melphalan was highly synergistic, as revealed by the Bliss and Loewe synergy scores calculated using the Combenefit software18 (Figure 2A-B, Online Supplementary Figure S4 and Online Supplementary Table S2). Other cell lines (KMS20 and RPMI 8226) were in general resistant to melphalan, as higher doses were required to obtain some reduction in proliferation (Figure 2A-B, middle panels, Online Supplementary Figure S4 and Online Supplementary Table S2). Notwithstanding, these cells responded well to VX-970. Furthermore, a strong synergy between melphalan at higher doses and VX-970 was also evident (Figure 2B and Online Supplementary Figure S4). Finally, MM cell lines LP1 and OPM2 were resistant to melphalan, even at very high doses (Figure 2A-B, right panels, Online Supplementary Figure S4 and Online Supplementary Table S2), but remained sensitive to VX-970 alone, which was thus able to overcome melphalan resistance. In a parallel set of experiments, we assessed whether this reduced proliferation was associated with increased haematologica | 2020; 105(10)

ATR dependancy in multiple myeloma

Figure 2. ATR inhibition by VX-970 sensitizes multiple myeloma cells to melphalan. (A) Multiple myeloma (MM) cells were seeded in 96-well plates and treated for 72 hours (h) with DMSO (as a control, untreated [NT]) or increasing concentrations of VX-970 either alone or in combination with the indicated doses of melphalan. Cell viability was assessed using CellTiter-Glo assay. Results are presented as the mean percentage of viable cells in treated samples, relative to DMSO control cells averaged from a minimum of three independent experiments (mean ± standard error of the mean [SEM]), each with three repetitions per condition. Proliferation curves for each cell line were generated using GraphPad Prism. The dotted vertical line indicates the response of MM cells to increasing concentrations of VX-970 alone. Filled black dots indicate the cellular response to melphalan alone. Results of the statistical analysis are reported in the Online Supplementary Table S2. We could not rule out the potential appearance of general toxicity at the highest concentrations used. (B) For each cell line the Bliss synergy matrices and the relative drug synergy scores were calculated using the Combenefit software. The colored areas in the matrix are indicative of the degree of synergy between the drug combinations. (A-B) Different color codes (black, blue and red) are indicative of the concentrations of melphalan used to treat the cells depending on their sensitivity to the drug. (C) MM1.S and H929-melphalan sensitive and OPM2-melphalan resistant cell lines were treated with the indicated concentrations of melphalan either alone or in combination with increasing concentrations of VX-970 (0.075, 0.15 and 0.3 mM). After 48 h, cells were harvested, and immunoblotted for the indicated antibodies. The levels of cleaved PARP and caspase-3 served as indicators of apoptosis. GAPDH was used as loading control.

apoptosis, as assayed with PARP and caspase-3 cleavage (Figure 2C) and quantification by FACS analysis of Annexin V and PI positive cells (Online Supplementary Figure S5). The combination of VX-970 with melphalan triggered a robust apoptotic response (Figure 2C and Online Supplementary Figure S5). Taken together, these results suggest that the combination of an ATR inhibitor alongside a DNA damaging agent eliciting ICL such as melphalan could represent a powerful, very effective drug combination to treat even resistant MM cells.

ATM inhibition does not enhance apoptosis triggered by DSB-inducing compounds We then explored whether ATM inhibition could similarly impact on the survival and proliferation of MM cells. We first assessed whether a compound triggering DSB, such as doxorubicin, was able to activate the ATM pathway in MM cells. Indeed, the treatment with this compound elicited a strong activation of the ATM network in MM cells, as shown by the phosphorylation of its downstream targets CHK2 and γH2AX (Figure 3A and Online Supplementary Figure S6). These results suggest that, as expected, DSB call upon ATM for assisting on DNA repair. Treatment of MM cells with both doxorubicin and a haematologica | 2020; 105(10)

broad ATM inhibitor, KU-55933,24 markedly reduced the activation of DNA damage response (as assessed with pCHK2 and γH2AX) (Figure 3A and Online Supplementary Figure S6). We then tested whether ATM inhibition, alone or in combination with doxorubicin, might restrain proliferation and elicit apoptosis. To our surprise, unlike VX-970, the treatment with KU-55933 was overall neither associated with a remarkable reduction in proliferation nor with apoptosis, even when used at very high concentrations (Figure 3B, Online Supplementary Table S3 and Online Supplementary Figure S7). Also, in all cell lines, the combination of doxorubicin and KU-55933 was not synergistic (Figure 3C and Online Supplementary Figure S8). These results suggest that in MM cells the inhibition of ATM is not crucial for the survival after ongoing or induced DSB, unlike ATR inhibition after ICL-inducing treatments. We also assayed MM cell lines for their sensitivity to both ATR and ATM inhibitors. In line with a recent report,25 we found that the combination of the ATR and ATM inhibition, that is, VX-970 and KU-55933, was synergistic, in some, but not all, MM tested cell lines (Online Supplementary Figure S9A-B and Online Supplementary Table S4). Specifically, ATM inhibition was able to further 2451

O.A. Botrugno et al. A

Figure 3. ATM inhibition by KU-55033 alone or in combination with doxorubicin does not restrain proliferation in multiple myeloma cells. (A) Exponentially growing MM1.S, H929 and RPMI 8226 cells were either left untreated (NT) or treated with increasing concentrations of KU-55933 (KU, 5 and 10 mM) and VX-970 (VX, 0.15 mM). Treatment with both compounds initiated 1 hour (h) before the addition of doxorubicin (0.5 μM). After 3 h, cells were harvested and and the expression of the indicated proteins was analysed by immunoblotting. GAPDH was used as loading control. (B) MM cells were seeded in 96-well plates and treated for 72 h with DMSO (as a control, NT) or increasing concentrations of KU-55933 either alone or in combination with the indicated doses of doxorubicin. Cell viability was assessed using CellTiter-Glo assay. Results are presented as the mean percentage of viable cells in treated samples, relative to DMSO control cells averaged from a minimum of three independent experiments (standard error of the mean (SEM), each with three repetitions per condition. Proliferation curves for each cell line were generated using GraphPad Prism. The dotted vertical line indicates the response of MM cells to increasing concentrations of KU-55933 alone. Filled black dots indicate the cellular response to doxorubicin alone. Results of statistical analysis are reported in the Online Supplementary Table S3. (C) For each cell line the Bliss synergy matrices and the relative drug synergy scores were calculated using the Combenefit software. The colored areas in the matrix are indicative of the degree of synergy between the drug combinations. (B-C) Different color codes (red, blue and black) are indicative of the concentrations of doxorubicin used to treat the cells depending on their sensitivity to the drug.

increase the already potent activity of VX-970, suggesting that ATR synergizes with ATM in repairing DSB in MM cells, despite having only a modest effect on the intensity and number of γH2AX foci per nucleus (Online Supplementary Figure S10A-B).

ATR inhibition enhances the therapeutic efficacy of melphalan in vivo To determine the effect of the combined treatment of VX-970 with melphalan in a more physiological setting, we assessed the activity of this combination in an in vivo mouse model, the Rag2−/−γc−/−, whereby injected human MM cells home to the mouse bone marrow.26 Treatment with melphalan, VX-970, or both drugs combined, was started at the fourth week after intravenous injection of the MM cells and continued for 3 weeks, using drug dosages reported in the literature21,27 (Figure 4A). No significant body weight loss was observed in the treatment groups, indicating that the treatment was well tolerated by the animals (data not shown). The tumor burden was assessed by in vivo imaging every week (Figure 4B). The combination of the two drugs was remarkably effective, profoundly restraining the rate of tumor development (Figure 4B). We next assessed the impact of the treatments on survival (Figure 4 C). Both melphalan or VX-970 were able by 2452

themselves to delay the progression of the disease. Of note, VX-970 was more effective than melphalan, with a median survival of 67 days, instead of 57 days for the mice treated with melphalan, and 50 days for mice treated with vehicle alone. Strikingly, combining melphalan and VX-970 exerted a profound effect on the survival of the mice, with a median 107 day survival of the mice examined. Taken together, these results suggest that the combination of an ATR inhibitor, VX-970, with a drug eliciting ICL, such as melphalan, is profoundly effective against MM cells presenting ongoing DNA damage.

The combination of ATR inhibitors and DNA damaging agents is effective also in MM cells with low levels of DNA damage We have recently shown how the tumor cells of a subset of MM patients present with ongoing replicative stress, DNA damage and enhanced chromosomal instability.22 These patients have poor prognosis. Also, our previous data have shown how MM cell lines presenting with enhanced replicative stress are sensitive to the ATR inhibitor VE-821.21,22 Of note, one MM cell line, U266, displays lower levels of ongoing DNA damage,19,25,28 replicative stress,22 and responds less to these drugs.22 In line with the experiments performed in MM cells haematologica | 2020; 105(10)

ATR dependancy in multiple myeloma

Figure 4. VX-970 is effective as monotherapy and enhances the therapeutic efficacy of melphalan in an orthothopic mouse model of multiple myeloma. (A) Rag2−/−γc−/− mice were injected intravenously with 5x106 MM1.S-Luc cells. The treatment started 4 weeks after injection, when the tumor burden became evident by bioluminescent imaging (BLI) (week 4). Mice were randomized into four treatment groups of five animals each: vehicle controls, melphalan 2 mg/kg, VX-970 60 mg/kg, or both. (B) BLI was performed every week during treatment (week 5, week 6 and week 7) to monitor tumor burden and one week after the stop of the treatment (week 8). On the left, representative IVIS SpectrumCT System images of the luciferase signals observed in mice of each treatment group taken at the indicated interval are shown. On the right, the graph shows the tumor burden increase quantified as total flux measured from bioluminescent images during the treatment period. Data represent the mean ± standard deviation (SD) of five mice in each treatment arm. Statistical analysis was performed by the two-way ANOVA and Tukey’s multiple comparison test (***P<0.0005, ****P<0.001). (C) Cumulative survival in each treatment arm was compared by Kaplan-Meier survival analysis. Statistical analysis was performed by log rank Mantel-Cox test (**P<0.01, ***P<0.001).

with rampant extent DNA damage, we first assessed whether melphalan and hydroxyurea treatment elicited DNA damage and the engagement of the DNA repair pathway also in the U266 cell line. We found that indeed upon treatment with hydroxyurea and melphalan, several markers of the DNA repair pathway, including phosphorylation of P53 and of CHK1 were activated in this cell line as well (Figure 5A). In line with the previous experiments, the treatment of U266 cells with VX-970, when combined with hydroxyurea or melphalan, hampered the ATR response, as revealed by the reduction of pCHK1, pRAD17 and γH2AX protein levels (Figure 5A-B). We then tested whether the combination of melphalan and VX-970 could be effective also in these cells. ATR inhibition by itself was moderately effective in reducing proliferation in U266 cells (Figure 5C and Online Supplementary Table S2), in line with our previous observations.22 The two drugs however were highly synergistic (Figure 5C and Online Supplementary Table S2). Along these lines, also in U266 cells the combination of VX-970 with melphalan triggered a robust apoptotic response as assayed through the analysis of PARP and caspase-3 cleavage (Online Supplementary Figure S11A) and quantification by FACS analysis of Annexin V and PI positive cells (Online Supplementary Figure S11B). In line with the experiments performed on MM cell lines presenting with intense ongoing DNA damage, also in U266 cells the combination of doxorubicin with the ATM inhibitor KU-55933 had neither a major effect on haematologica | 2020; 105(10)

survival (Online Supplementary Figure S12A-D and Online Supplementary Table S3), nor was there any synergy between VX-970 and the ATM inhibitor KU-55933 (Online Supplementary Figure S9A-B and Online Supplementary Figure S10A-B). We also assessed the effect of VX-970 and melphalan, alone or in combination, in the Rag2−/−γc−/− mouse model (Online Supplementary Figure S13A). These experiments confirmed that the combination was highly effective in hampering the growth of U266 MM cells in vivo (Figure 5D and Online Supplementary Figure S13B). Taken together, these data suggest that even MM cells not presenting with DNA damage are sensitive to the combination of an ICL-inducing drug such as melphalan and ATR inhibition. ATM inhibition again had no major impact on cell survival upon induction of DSB.

ATR inhibition synergizes with melphalan by inducing apoptosis in MM patient cells We then assessed the efficacy of this treatment strategy on MM patient cells, co-cultured with BMSC (BM stromal cells) in a 3D bioreactor system, which we recently reported as a sensitive and reliable method to assess drug response in MM (Figure 6A).17 While single treatments with either VX-970 or melphalan in primary MM cells was variably effective, the combination consistently elicited apoptosis, suggesting it may represent an promising, broadly active drug regimen in MM patients (Figure 6B-C). Moreover, we analyzed by immunohistochemistry 2453

O.A. Botrugno et al.

Figure 5. ATR inhibition by VX-970 sensitizes U266 cells to melphalan. (A) Exponentially growing U266 cells were either left untreated or treated with the indicated concentrations of VX-970. Treatment with VX-970 initiated 1 hour (h) before the addition of hydroyurea (HU, 2 mM) or melphalan (50 mM). After 3 h cells were harvested and analyzed by immunoblotting for the expression of the indicated proteins. Vinculin was used as a loading control. The asterisk indicates an unspecific band. (B) As in (A), cells were left untreated (NT) or treated with 0.15 mM VX-970 (VX) before the addition of HU (2 mM) or melphalan (50 mM) for 3 h. At the end of the treatment, cells were fixed and processed for imaging flow cytometry to detect the γH2AX signal for individual nucleus (left Y axis). The mean of γH2AX signal (right Y axis) for each treatment is reported. Results are representative of two independent experiments. (C) Cells were seeded in 96-well plates and treated for 72 h with DMSO (as a control, NT) or increasing concentrations of VX-970 either alone or in combination with the indicated doses of melphalan. Cell viability was assessed using CellTiter-Glo assay. Results are presented as the mean percentage of viable cells in treated samples, relative to DMSO control cells averaged from a minimum of three independent experiments (mean ± standard error of the mean [SEM]) each with three repetitions per condition. Proliferation curves on the left were generated using GraphPad Prism. The dotted vertical line indicates the response of the cells to increasing concentrations of VX-970 alone. Filled black dots indicate the cellular response to melphalan alone. The Bliss and the Loewe synergy matrices and the relative drug synergy scores calculated using the Combenefit software are reported on the right of the proliferation profile. The colored areas in the matrix are indicative of the degree of synergy between the drug combinations. Results of statistical analysis are reported in the Online Supplementary Table S2. (D) Rag2−/−γc−/− mice were injected intravenously with 5x106 U266-Luc cells. The treatment started 8 weeks after injection, when the tumor burden became evident by BLI (week 8). Mice were randomized into four treatment groups of five animals each: vehicle controls, melphalan 2 mg/kg, VX-970 60 mg/Kg, or both. Treatments were scheduled as reported in the Online Supplementary Figure S10A. BLI was performed every week during the treatment to monitor the tumor burden. Representative IVIS SpectrumCT System images of the luciferase signals observed in mice of the treatment groups taken before beginning of the treatment (week 8) and 1 and 5 weeks after the stop of the treatment (week 12 and week 16).

the BM biopsies of these MM patients for γH2AX and pCHK1 expression. Phosphorylated CHK1 was diffusely expressed in 80-90% of neoplastic cells, while 5-30% of tumor plasma cells expressed γH2AX ranged between 5-30% of tumor plasma cells, suggesting that the ATR pathway is predominantly activated in MM cells present pervasive activation of the ATR pathway and respond to the combination of VX-970 with melphalan independently from their basal levels of ongoing DNA damage.

Discussion In this study, we have shown that drugs that elicit DNA damage and are commonly used to treat MM patients trigger a high activation of DNA repair response, engaging both the ATM and the ATR pathways. Surprisingly, however, ATM inhibition had little, if any, effect on MM cells, either used alone, or in combination with doxorubicin, a drug that elicits DSB. Conversely, inhibition of ATR was strongly synergistic with melphalan, a nitrogen mustard that has been used to treat MM patients for the past 60 2454

years, and that induces ICL. Furthermore, ATR inhibition in combination with melphalan was highly effective also in MM cells lacking ongoing DNA damage. Taken together, these results suggest that MM cells rely extensively on ATR for the repair of DNA damage. We argue that ATR inhibition, alone but particularly in combination with ICL-inducing agents, may represent a novel, critically important tool to be included in the therapeutic tools used to treat MM. ATR relieves tumor cells from replicative stress. Indeed, epithelial cancers present enhanced replicative stress and genomic instability, as a result of oncogene-driven tumultuous growth.29 We recently found that pervasive DNA damage is present in haematological cancers as well, including MM, lymphoma and leukaemia.19 To withstand this intense replicative stress and the ensuing DNA damage, cancer cells rely heavily on ATR. Along these lines, we have recently identified a subset of myeloma patients, whose cancer cells display prominent replicative stress and chromosomal instability. These patients have a dismal prognosis. MM cells presenting with ongoing replicative stress are sensitive to ATR inhibition.22 Apart from its role haematologica | 2020; 105(10)

ATR dependancy in multiple myeloma

Figure 6. ATR inhibition by VX-970 synergizes with melphalan in inducing apoptosis in multiple myeloma patient cells independently of the level of DNA damage. (A) Schematic representation of the experimental procedure in the 3D bioreactor. (B-C) CD138+ primary cells from multiple myeloma (MM) patients were seeded in gelatin scaffolds pre-seeded with CD73+ bone marrow stromal cells (BMSC) and cultured in the 3D bioreactor system either in untreated conditions (NT) or in the presence of VX-970 (0.3 mM), melphalan (1.2 mM), or a combination of both drugs. (B) Seventy-two hours after the beginning of the treatment, cells retrieved from scaffolds were stained with Annexin V and anti-CD38 antibody before flow cytometric (FACS) analysis. The table underneath the graph summarizes the Bliss and Loewe synergy scores calculated for the combined treatments in each patient. Due to the limited number of cells that could be recovered from MM patient bone marrow (BM) samples, it was possible to assay just one concentration for each drug, in each patient. (C) Representative immunohistochemical (IHC) analyses performed on scaffolds populated with primary MM cells and retrieved at the end of the culture period, showing the distribution of CD138+ MM and CD73+ BMSC cells and the effect of the co-treatment specifically on MM cells. (D) Representative IHC images form the BM biopsies of MM patients analyzed for ÎłH2AX and pCHK1 expression (original magnification 40X; samples presented are the same used for experiment in (B)). The inset in each panel indicates the percentage of tumor cells positive for the indicated marker. The percentage of neoplastic plasma cells (PC) is shown above each panel.

in replicative stress, our results suggest that the role of ATR in relieving ICL is equally important, as the inhibition of ATR strongly potentiates the activity of melphalan, a compound that induces ICL and that has been, and still is, widely used for the treatment of MM patients. The reliance of MM cells on ATR, but not on ATM for their survival, is remarkable. ATM is pivotal for the whole cellular response to DSB.14 We and others have shown how MM cell lines as well as patient cells display exceedingly high levels of rampant DNA damage and DSB, in the absence of exogenous DNA damage.19,28,30 Against this backdrop, and based on our data, we posit that ATR, and not ATM, is the central hub in MM cells that regulates the DNA damage response and assures the proper repair of DNA. Based on this ATR dependency, we hence argue that MM cells might be particularly vulnerable to the effects of specific DNA lesions, such as ICL (where ATR is requested), while being resilient to others, such as DSB. In fact, tumors may be preferentially sensitive to compounds targeting specific branches of the DNA response. For example, drugs triggering ICL have found prominent therapeutic applications in specific cancer subtypes, which include other chronic haematological cancers such as chronic lymphocytic leukaemia and lymphomas, but also tumors of epithelial origin such as ovarian cancer, a subset of colon carcinoma and a few other selected cancers.31 In fact, haematologica | 2020; 105(10)

when we assayed VX-970, alone or in combination with melphalan, in other two commonly used cancer tumor cell lines, cervical cancer HeLa and osteosarcoma U2OS cells, we found widely different sensitivity to these treatments (Online Supplementary Figure S14). We would then argue that a yet-to-be-characterized subset (or subgroups within) of cancer types may rely extensively on ATR for their survival, potentially being dependant on it. Additional ad hoc studies would be required to demonstrate this potential dependency. Recently, the phosphatase CDC25A has been described as a determinant of the sensitivity to ATR inhibitors in several tumor cell lines and has been proposed as a potential biomarker to rationalize the use of ATR inhibitors in cancer therapy.32 However, it appears that CDC25A exerts a different role in MM, as the reduced expression of the protein did not impact on the sensitivity of MM cells to the treatment with VX-970 either alone or in combination with melphalan (Online Supplementary Figure S15). The development of resistance to ICL-inducing drugs is a pervasive plague.33 Our data confirm and expand on previous data on MM34 and other cancers,35 showing how ATR inhibition can overcome melphalan resistance in MM. Intriguingly, a synergy between melphalan and VX-970 was evident even in cases resistant to the treatment with melphalan alone, suggesting a potential role for ATR inhibition in patients overtly resistant to ICL-inducing drugs. 2455

O.A. Botrugno et al.

In MM, acquired melphalan resistance has been linked to a reduction of melphalan-induced ICL, and most notably to an overall increase in the gene expression levels of the FA pathway.36 Therefore, the inhibition of ATR might supress the compensatory mechanism elicited by the increase activity of the FA pathway. In this study we sought to identify additional synthetic tumor cells specific lethal approaches, beyond the PARP1 and BRCA1/2 axis, whereby the combination of DNA damaging agents commonly used in the clinic to treat MM could be associated with novel interventions, to prevent cells from repairing DNA and hence trigger apoptosis. We found that the inhibition of ATR was highly synergistic with ICL-inducing melphalan, strongly warranting the clinical exploitation of VX-970 in combination with melphalan as a therapy in MM patients, irrespectively of their resistance to melphalan, and the

References 1. Puigvert JC, Sanjiv K, Helleday T. Targeting DNA repair, DNA metabolism and replication stress as anti-cancer strategies. FEBS J. 2016; 283(2):232-245. 2. Iglehart JD, Silver DP. Synthetic lethality - a new direction in cancer-drug development. N Engl J Med. 2009; 361(2):189-191. 3. Kaelin WG. The concept of synthetic lethality in the context of anticancer therapy. Nat Rev Cancer. 2005;5(9):689-698. 4. Fu D, Calvo JA, Samson LD. Balancing repair and tolerance of DNA damage caused by alkylating agents. Nat Rev Cancer. 2012;12(2):104-120. 5. Bryant HE, Schultz N, Thomas HD, et al. Specific killing of BRCA2-deficient tumours with inhibitors of poly (ADPribose) polymerase. Nature. 2005; 434(7035):913-917. 6. Farmer H, McCabe N, Lord CJ, et al. Targeting the DNA repair defect in BRCA mutant cells as a therapeutic strategy. Nature. 2005;434(7035):917-921. 7. Lord CJ, Ashworth A. PARP inhibitors: synthetic lethality in the clinic. Science. 2017; 355(1):1152-1158. 8. Davies H, Glodzik D, Morganella S, et al. HRDetect is a predictor of BRCA1 and BRCA2 deficiency based on mutational signatures. Nat Med. 2017;23(4):517-525. 9. Helleday T, Petermann E, Lundin C, Hodgson B, Sharma RA. DNA repair pathways as targets for cancer therapy. Nat Rev Cancer. 2008;8(3):193-204. 10. Palumbo A, Anderson K. Multiple myeloma. N Engl J Med. 2011;364(11):1046-1060. 11. Blokhin N, Larionov L, Perevodchikova N, Chebotareva L, Merkulova N. Clinical experiences with sarcolysin in neoplastic diseases. Ann N Y Acad Sci. 1958;68(3):1128-1132. 12. Kyle RA, Rajkumar SV. Multiple myeloma. Blood. 2008;111(6):2962-2972. 13. Mateos MV, Dimopoulos MA, Cavo M, et al. Daratumumab plus bortezomib, melphalan, and prednisone for untreated myeloma. N Engl J Med. 2018;378(6):518528. 14. Blackford AN, Jackson SP. ATM, ATR, and DNA-PK: The trinity at the heart of the DNA damage response. Mol Cell. 2017; 66(6):801-817. 15. Manier S, Huynh D, Shen YJ, et al.

2456

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

replicative stress and ongoing DNA damage of their tumor cells. Acknowledgments The authors would like to thank all the members of Tononâ&#x20AC;&#x2122;s lab for collaborative and helpful discussion and critical reading of the manuscript; Antonello Spinelli e Laura Perani from the Experimental Imaging Center at San Raffaele scientific Institute for assisting with the in vivo imaging; Raffaella Di Micco, Lucrezia Della Volpe and Anastasia Conti for sharing protocols and reagents. Funding Funding for this research was provided by Associazione Italiana per la Ricerca sul Cancro (AIRC; Investigator Grants and Special Program Molecular Clinical Oncology, 5 per mille no. 9965 to GT).

Inhibiting the oncogenic translation program is an effective therapeutic strategy in multiple myeloma. Sci Transl Med. 2017; 9(389):1-13. Toledo LI, Murga M, Zur R, et al. A cellbased screen identifies ATR inhibitors with synthetic lethal properties for cancer-associated mutations. Nat Struct Mol Biol. 2011;18(6):721-727. Belloni D, Heltai S, Ponzoni M, et al. Modeling multiple myeloma-bone marrow interactions and response to drugs in a 3D surrogate microenvironment. Haematologica. 2018;103(4):707-716. Di Veroli GY, Fornari C, Wang D, et al. Combenefit: an interactive platform for the analysis and visualization of drug combinations. Bioinformatics. 2016; 32(18):28662868. Cottini F, Hideshima T, Xu C, et al. Rescue of Hippo coactivator YAP1 triggers DNA damage-induced apoptosis in hematological cancers. Nat Med. 2014;20(6):599-606. Rouzitalab M, Sani ZA, Parsaee M, Farzaneh M, Khalilipur E, Rahimi S. ATR inhibitors VE-821 and VX-970 sensitize cancer cells to topoisomerase I inhibitors by disabling DNA replication initiation and fork elongation responses. Cancer Res. 2014;74(23):6968-6979. Hall AB, Newsome D, Wang Y, et al. Potentiation of tumor responses to DNA damaging therapy by the selective ATR inhibitor VX-970. Oncotarget. 2014; 5(14):5674-5685. Cottini F, Hideshima T, Suzuki R, et al. Synthetic lethal approaches exploiting DNA damage in aggressive myeloma. Cancer Discov. 2015;5(9):972-987. Brown JS, Oâ&#x20AC;&#x2122;Carrigan B, Jackson SP, Yap TA. Targeting DNA repair in cancer: beyond PARP inhibitors. Cancer Discov. 2017;7(1):20-37. Hickson I, Zhao Y, Richardson CJ, et al. Identification and characterization of a novel and specific inhibitor of the ataxiatelangiectasia mutated kinase ATM identification and characterization of a novel and specific inhibitor of the ataxia-telangiectasia mutated kinase ATM. Cancer Res. 2004; 64(24):9152-9159. Herrero AB, Gutierrez NC. Targeting ongoing DNA damage in multiple myeloma. effects of different Inhibitors of the DNA damage response on plasma cell survival.

Front Oncol. 2017;19(7):98. 26. Rozemuller H, van der Spek E, Bogers-Boer LH, et al. A bioluminescence imaging based in vivo model for preclinical testing of novel cellular immunotherapy strategies to improve the graft-versus-myeloma effect. Haematologica. 2008;93(7):1049-1057. 27. Xia J, Xu H, Zhang X, et al. Multiple myeloma tumor cells are selectively killed by pharmacologically-dosed ascorbic acid. EBioMedicine. 2017;1841-1849. 28. Herrero AB, San Miguel J, Gutierrez NC. Deregulation of DNA double-strand break repair in multiple myeloma: implications for genome stability. PLoS One. 2015; 10(3):e0121581. 29. Luo J, Solimini NL, Elledge SJ. Principles of cancer therapy: oncogene and non-oncogene addiction. Cell. 2009;136(5):823-837. 30. Walters D, Wu X, Tschumper R. Evidence for ongoing DNA damage in multiple myeloma cells as revealed by constitutive phosphorylation of H2AX. Leukemia. 2011;25(8):1344-1353. 31. Deans AJ, West SC. DNA interstrand crosslink repair and cancer. Nat Rev Cancer. 2011;11(7):467-480. 32. Ruiz S, Mayor-Ruiz C, Lafarga V, et al. A genome-wide CRISPR screen identifies CDC25A as a determinant of sensitivity to ATR inhibitors. Mol Cell. 2016;62(2):307313. 33. Jin L, Chun J, Pan C, et al. MAST1 drives cisplatin resistance in human cancers by rewiring cRaf-independent MEK activation. Cancer Cell. 2018;34(2):315-330.e7. 34. Landau HJ, McNeely SC, Nair JS, et al. The checkpoint kinase inhibitor AZD7762 potentiates chemotherapy-induced apoptosis of p53-mutated multiple myeloma cells. Mol Cancer Ther. 2012; 11(8):17811788. 35. Kurmasheva RT, Kurmashev D, Reynolds CP, et al. Initial testing (stage 1) of M6620 (formerly VX-970), a novel ATR inhibitor, alone and combined with cisplatin and melphalan, by the Pediatric Preclinical Testing Program. Pediatr Blood Cancer. 2018;65(2):1-11. 36. Hazlehurst LA, Enkemann SA, Beam CA, et al. Genotypic and phenotypic comparisons of de novo and acquired melphalan resistance in an isogenic multiple myeloma cell line model. Cancer Res. 2003;63(6):79007906.

haematologica | 2020; 105(10)

ARTICLE

Platelet Biology & its Disorders

Dominant negative Gfi1b mutations cause moderate thrombocytopenia and an impaired stress thrombopoiesis associated with mild erythropoietic abnormalities in mice

Ferrata Storti Foundation

Hugues Beauchemin,1 Peiman Shooshtharizadeh,1 Jordan Pinder,2 Graham Dellaire2 and Tarik Möröy1,3,4

Institut de Recherches Cliniques de Montréal, IRCM, Montréal, Quebec; 2Departments of Pathology and Biochemistry and Molecular Biology, Dalhousie University, Halifax, Nova Scotia; 3Département de Microbiologie, Infectiologie et Immunologie, Université de Montréal, Montréal, Quebec and 4Division of Experimental Medicine, McGill University, Montréal, Quebec, Canada

Haematologica 2020 Volume 105(10):2457-2470

ABSTRACT

FI1B-related thrombocytopenia (GFI1B-RT) is a rare bleeding disorder mainly caused by the presence of truncated GFI1B proteins with dominant-negative properties. The disease is characterized by low platelet counts, the presence of abnormal platelets, a megakaryocytic expansion, and mild erythroid defects. However, there are no animal models that faithfully reproduce the GFI1B-RT phenotype observed in patients. We had previously generated mice with floxed Gfi1b alleles that can be eliminated by Cre recombinase, but those animals developed a much more severe phenotype than GFI1B-RT patients and were of limited interest in assessing the disease. Using CRISPR/Cas9 technology, we have now established three independent mouse lines that carry mutated Gfi1b alleles producing proteins lacking DNA binding zinc fingers and thereby acting in a dominant negative (DN) manner. Mice heterozygous for these Gfi1b-DN alleles show reduced platelet counts and an expansion of megakaryocytes similar to features of human GFI1B-RT but lacking the distinctively large agranular platelets. In addition, Gfi1b-DN mice exhibit an expansion of erythroid precursors indicative of a mildly abnormal erythropoiesis but without noticeable red blood cell defects. When associated with megakaryocyte-specific ablation of the remaining allele, the Gfi1b-DN alleles triggered erythroid-specific deleterious defects. Gfi1b-DN mice also showed a delayed recovery from platelet depletion, indicating a defect in stress thrombopoiesis. However, injecting Gfi1b-DN mice with romiplostim, a thrombopoietin receptor super agonist, increased platelet numbers even beyond normal levels. Thus, our data support a causal link between DN mutations in GFI1B and thrombocytopenia, and suggest that patients with GFI1B-RT could be treated successfully with thrombopoietin agonists.

Correspondence: TARIK MÖRÖY tarik.moroy@ircm.qc.ca Received: March 20, 2019. Accepted: November 21, 2019. Pre-published: November 21, 2019. doi:10.3324/haematol.2019.222596

Introduction GFI1B-related thrombocytopenia (GFI1B-RT), also known as bleeding disorder, platelet-type, 17 (BDPLT17) as listed in the Online Mendelian Inheritance in Man (OMIM) database (http://www.omim.org/entry/187900), is a rare dominant congenital platelet disorder caused by mutations in the GFI1B gene, which encodes two protein isoforms of 32 and 37 kDa. These proteins carry an N-terminal SNAG domain that mediates transcriptional repression, an intermediate domain and six C-terminal C2H2-type zinc finger domains.1 In GFI1B-RT, mutations in GFI1B lead to either the production of truncated proteins lacking the last two zinc fingers or to the disruption of the first zinc finger alone, and cause mild to moderate bleeding diathesis, macrothrombocytopenia, an α-granule deficiency in platelets for most but not all cases, and is also often associated with anisocytosis and poikilocytosis.2-13 GFI1B-RT was first associated with the gray platelet syndrome (GPS), or BDPLT4, caused by mutation in the NBEAL2 gene.14 Although these two haematologica | 2020; 105(10)

2457

H. Beauchemin et al.

forms of thrombocytopenia share a variety of features such as enlarged platelets with reduced α-granule content and abnormal megakaryocytes (MK),6,14,15 they also exhibit significant differences, such as a higher variability in αgranule deficiency in GFI1B-RT, differences in platelet aggregation function, autosomal recessive (NBEAL2) versus autosomal dominant (GFI1B) transmission and GFI1Bspecific red blood cell (RBC) defects.6,7,16,17 Unlike NBEAL2-related GPS, for which mouse models have been described,18-21 no experimental models have been created yet for GFI1B-RT that would faithfully recapitulate the disease phenotype. The rarity of cases makes it difficult to study the cellular mechanisms underlying this thrombocytopenia, and although several Gfi1b-knockout mouse models exist, none of them faithfully reproduce the human disease.22-25 For example, full ablation of Gfi1b is embryonic lethal,22,23 and the use of either inducible Cre or MK-specific Cre to generate homozygous deletions of Gfi1b cause extreme phenotypes such as total abrogation of thrombopoiesis and erythropoiesis that can hardly be correlated with the disease seen in humans.24-26 Here we report results obtained with three Gfi1b mutant mouse lines generated by CRISPR/Cas9 genome editing to alter the region encoding the fifth zinc finger of GFI1B, which mimic similar mutations found in the human GFI1B gene. As in GFI1B-RT patients, the three mutants act in a dominant negative (DN) manner and the mice exhibit a mild to moderate thrombocytopenia with increased sensitivity to thrombopoietic stresses. Importantly, we show that treatment with the clinical drug romiplostim, a thrombopoietin (TPO) analog, could rescue the phenotype and allow the thrombocytopenic mice to significantly increase their platelet count, indicating potential therapeutic avenues for GFI1B-RT patients.

Methods Mice Protocols used in this paper were reviewed by the Animal Care Committee (ACC, #2013-04) of the Clinical Research Institute of Montréal and all animals were cared for in compliance with the Canadian Council on Animal Care guidelines (www.ccac.ca).

Hematologic parameters Circulating platelet counts were measured either on an Advia 120 cell analyzer (Bayer) using the mouse archetype of multispecies software version 2.2.06 or by quantitative flow cytometry using PE-conjugated anti-CD41 (Biolegend), AF647-conjugated anti-CD61 (BioLegend) and 123count eBeads™ counting beads (Invitrogen) as reference. Hematocrit was measured either on the Advia 120 cell analyzer or by centrifugation of microcapillary on a Haematokrit 200 (Hettich). Other experimental procedures are described in the Online Supplementary Methods.

Results Introduction of Gfi1b germline mutations similar to those found in GFI1B-RT patients into mice by CRISPR/Cas9 A vector containing both the Cas9 gene and a gRNA targeting the seventh exon of Gfi1b at the position encoding 2458

the fifth zinc finger of GFI1B was injected into fertilized murine oocytes. Of 78 pups produced, six carried indel mutations at the target site for a rate of approximately 8%, all of them occurring at the exact Cas9 cutting site. Five founders showed at least some levels of mosaicism, with three founders carrying more than two alleles, indicating that the expression vector remained active for a few cell divisions and that mutations occurred fairly late (Online Supplementary Table S1). Although a total of ten indel alleles were produced, some mutations were identical between several founders. Because of the high level of mosaicism, only two founders were able to transmit the mutant alleles to their progeny, which in turn resulted in three Gfi1b mutant lines carrying alleles modified by deletion of two base pairs (Gfi1bdel2), deletion of seven base pairs (Gfi1bde7), and insertion of four base pairs (Gfi1bins4) (Figure 1A). As they all occurred at the cutting site of the Cas9 in the region encoding for the fifth zinc finger of Gfi1b, all mutations produced frameshifts that disrupt the fifth and the sixth zinc fingers. The Gfi1bdel2 and Gfi1bins4 mutations switch to frame 2 and generate truncated proteins similar to the proteins identified in some GFI1B-RT patients,7 whereas the Gfi1bdel7 mutation switches to frame 3 and produces an elongated protein with a different C-terminus (Figure 1B and Online Supplementary Figure S1).

The three Gfi1b mutations act in a dominant negative manner Expression of Gfi1b was first assessed in lineage-depleted bone marrow from Gfi1b mutant heterozygous mice and compared to that from wild-type (WT) animals (Figure 2A). All three Gfi1b mutant lines showed an increased Gfi1b expression with Gfi1bdel2 and Gfi1bins4 being slightly higher than Gfi1bdel7, which is consistent with a loss of GFI1B self-repression.27 Sequencing the cDNA from heterozygous Gfi1bWT/del2, Gfi1bWT/del7, Gfi1bWT/ins4 mice showed mixed chromatograms, demonstrating that the mutant alleles were expressed along with the WT allele (Figure 2B). Moreover, cloning the cDNAs in vectors that were then sequenced, allowed comparison of the ratio of mutant versus WT transcripts of the two isoforms 1 and 2, demonstrating that the mutant mRNAs are present at comparable levels to the transcripts generated by the WT allele, and this, for both isoforms (Figure 2C). Since it was not possible to distinguish the mutant from the WT GFI1B proteins by western blot in heterozygous mice, we cloned the cDNAs of both isoforms of the three mutants and the WT alleles as well as cDNA prepared from Gfi1bWT/KO mice in an expression vector and transfected 293 cells to detect the proteins (Figure 2D and E and Online Supplementary Figure S2A). As expected, no proteins were detected in cells transfected with either an empty vector or with vectors containing Gfi1bKO cDNAs, which produce mRNAs that do not encode a protein since they lack an open reading frame. Both murine isoforms 1 (37 kDa identified as p37) and 2 (predicted size of 40 kDa and identified as p40) of GFI1BWT and the two mutants GFI1Bdel2 and GFI1Bins4 were readily detectable, whereas the GFI1Bdel7 protein was consistently detected at much lower levels, suggesting possible protein and/or transcript instabilities of this variant. In all cases, the long isoform was expressed at higher levels than the shorter isoform, suggesting that isoform 2 might be more stable than isoform 1. haematologica | 2020; 105(10)

Thrombocytopenia in Gfi1b mutant mice

To assess the functional consequences of the mutated Gfi1b alleles, HEK-293T cells were co-transfected with a luciferase expressing vector under the control of the human GFI1B promoter, which contains GFI1B binding sites,27 along with expression vectors of either isoforms 1 or 2 of GFI1Bdel2, GFI1Bins4, GFI1Bdel7, GFI1BWT or Gfi1bKO, or the isoform 1 of the human GFI1B. As expected, the human and both WT isoforms of the murine GFI1B were

efficient in repressing the activity of the GFI1B promoter (Figure 2F). On the contrary, none of the mutants were able to repress the luciferase activity, leading to reads similar to those observed in the absence of GFI1B protein (Figure 2F). To assess if these mutant alleles exhibit dominant negative (DN) activity as described in human patients,6 the three Gfi1b mutants were transfected along with the WT

Figure 1. Generation of mice carrying mutant Gfi1b. (A) Mice carrying a mutated allele resembling that of the human mutations causing GFI1B-RT were generated by CRISPR/Cas9 using a gRNA targeting the fifth zinc finger of GFI1B. Three lines capable the mutated allele were produced: one carrying a deletion of two nucleotides (del2), a second one carrying a deletion of seven nucleotides (del7), and one harboring an insertion of four nucleotides (ins4). (B) Resulting C-termini of the proteins produced by two mutants found in human GFI1B-RT patients (green, top) compared to mouse WT gene and the three mutants carried by the Gfi1b mutant mice (blue, bottom). The fifth and sixth zinc fingers are highlighted in yellow and the extraneous peptide sequences generated by the frameshifts are identified in red.

haematologica | 2020; 105(10)

2459

H. Beauchemin et al.

†

Figure 2. The three GFI1B mutants are non-functional and as dominant negative. (A) Gfi1b expression analyzed by real-time polymerase chain reaction (RT-PCR) from Gfi1b-mutant lineage-depleted bone marrow. Results are presented as mean±standard deviation of Gfi1b over the housekeeping gene Rplp0. Statistical significance was calculated using a one-way ANOVA (P<0.0001) followed by a Tukey’s multiple comparisons test comparing the mean of each column with the mean of every other column. *Adjusted P<0.05; ***P<0.001; ****P<0.0001. (B) The Gfi1b cDNA derived from erythroblasts of mutant mice was PCR-amplified and sequenced. (C) PCR-amplified cDNAs from (B) were cloned into an expression vector and each clone obtained was sequenced to identify the isoform and the allele. (D) Schematic representation of the two isoforms of GFI1B produced in the mutant mice and cloned in (C). The white rectangle represents the SNAG domain and the white ovals the functional zinc fingers. (E) Detection by western blot of both isoform 1 (v.1; GFI1B-p37) and isoform 2 (v.2; GFI1B-p40) of GFI1B from 293 cells transfected with either the mutants or the wild-type (WT) Gfi1b cDNAs. (F) Box-and-whisker plot of a GFI1B-dependent luciferase assay. Results are presented as luciferase activity (RLU) of six different transfections. EV: empty vector; knockout (KO) is a cDNA cloned from knockout animal and lacking exon 2 to 4; v.1: isoform 1; v.2: isoform 2. (G) Box-and-whisker plot of a GFI1B-dependent luciferase assay measuring the capacity of the GFI1B mutants to inhibit repression by the WT GFI1B. Both isoforms of the three GFI1B mutants were co-transfected along the WT forms and the results are presented as the median of twelve separate experiments. Individual results for the isoforms are presented in Online Supplementary Figure S2B. (F and G) Statistical significance was calculated using a Kruskal-Wallis followed by a Dunn's multiple comparisons comparing the median of each column with the median of all other columns and the full results are presented in Online Supplementary Tables S2 and S3. †Indicates statistical significance.

2460

haematologica | 2020; 105(10)

Thrombocytopenia in Gfi1b mutant mice

†

Figure 3. Moderate thrombocytopenia in Gfi1b-DN mice. (A) Platelet counts, large platelet counts and mean platelet volume (MPV) of Gfi1b-DN and wild-type (WT) mice along hemizygous Gfi1bWT/0 mice. In the scatter dot plots, means±standard deviation are indicated. In the histogram, the numbers given indicate sample size. Statistical significance was calculated using a Brown-Forsythe ANOVA and when significance was found, was followed by a Dunnett's T3 multiple comparisons test comparing the mean of each column with the mean of every other column. (B) Platelet CD41 Mean Fluorescence Intensity (MFI) measured by flow cytometry. Statistical significance was calculated using a one-way ANOVA (P=0.0023) followed by a Holm-Sidak's multiple comparisons test comparing the mean of each column with the mean of every other column. (C) Transmission electron microscopy of thin sections of platelets and violin plot of the quantification of α-granules per platelet thin section. Median and upper/lower quartiles are indicated in blue. In the micrographs, arrows indicate α-granules. Significance was calculated using a KruskalWallis test. (D) Hematocrit of Gfi1b-DN and WT mice along hemizygous Gfi1bWT/0 mice. (E) May-Grünwald Giemsa staining of blood smears showing normal platelets and erythrocytes in both controls and Gfi1b-DN mice. The full results of the Dunnett's T3 (A) and the Holm-Sidak's (B) tests are presented in Online Supplementary Tables S4 and S5, respectively. †Indicates statistical significance.

haematologica | 2020; 105(10)

2461

H. Beauchemin et al.

†

† †

†

Figure 4. Megakaryopoiesis and erythropoiesis in Gfi1b-DN mice. (A) Representative FACS analysis of the bone marrow of wild-type (WT) (Gfi1bWT/WT), hemizygotes (Gfi1bWT/0) and Gfi1b-DN (Gfi1bWT/del2, Gfi1bWT/del7 and Gfi1bWT/ins4) mice at steady state. A summary of the gating strategy for all cells is presented in Online Supplementary Figure S3. (B) Quantification of bone marrow megakaryocyte progenitors (MKP) and megakaryocytes (MK). (C) Quantification of bone marrow preerythroid colony forming unit (PreCFUe) and proerythroblasts. All results are reported as mean±standard deviation and significance was calculated by one-way ANOVA tests (P<0.0001 for all graphs) followed by Holm-Sidak’s multiple comparisons tests comparing the mean of each column with the mean of all other columns individually. Full results of the Holm-Sidak’s tests are presented in the Online Supplementary Tables S6 and S7. †Indicates statistical significance.

2462

haematologica | 2020; 105(10)

Thrombocytopenia in Gfi1b mutant mice

E E

Figure 5. Characterization of Gfi1b-DN mice in the presence of Gfi1b knockout alleles. (A) Ratio of the mice obtained carrying the proper genotype on the expected number (1 representing a perfect Mendelian transmission). The numbers on top indicate the number of pups that survived at least until weaning. Statistical significance between the number obtained compared to the number expected was calculated using a Fisher's exact test. (B) Kaplan-Meier plot of survival of mice with different DN alleles along a megakaryocyte (MK)-specific knockout allele and compared to wild-type (WT) and Pf4-cre:Gfi1bflox/flox mice. Median survival: Pf4-cre:Gfi1bflox/flox (KO/KO) = 186 days; Pf4-cre:Gfi1bflox/del2 (del2/KO) = 32 days; Pf4-cre:Gfi1bflox/del7 (del7/KO) = 36 days; Pf4-cre:Gfi1bflox/Ins4 (ins4/KO) = 34 days. Log-rank Mantel-Cox tests show no significant differences between the three Gfi1bflox/mutant lines (P=0.22) and a significant difference (P<0.0001) between any Gfi1bflox/mutant lines and either Gfi1bWT/WT or Pf4-cre:Gfi1bflox/flox mice. (C) Platelet counts and hematocrits measured from Pf4-cre:Gfi1bflox/mutant (grouping all Pf4-cre:Gfi1bflox/del2, Pf4-cre:Gfi1bflox/del7 and Pf4-cre:Gfi1bflox/Ins4 mice analyzed) to Pf4-cre:Gfi1bflox/flox and Pf4-cre:Gfi1bflox/wt. (D) Spleen weight and quantification by flow cytometry of erythroid precursors from 6week old female mice with genotypes as in (C). (E) Quantification by flow cytometry of bone marrow of erythroid precursors and progenitors from the same mice as in (C). (C and D) All results are reported as meanÂąstandard deviation and significance was calculated by either a Brown-Forsythe ANOVA (BF-ANOVA) test or a oneway ANOVA (ANOVA) test followed by Dunnett's T3 or Holm-Sidakâ&#x20AC;&#x2122;s multiple comparisons tests respectively to compare each column with all other columns individually. The statistical significance measured by the post tests are as follow: *P<0.05; **P<0.01; ***P<0.001; ****P<0.0001.

haematologica | 2020; 105(10)

2463

H. Beauchemin et al. A

Figure 6. Gfi1b-DN mice can respond adequately to an acute erythropoietic stress. Time course analysis of the response to Phenylhydrazine (PHZ)-induced hemolytic anemia. 4-5 mice for each time point were injected twice with PHZ and sacrificed 2, 4 or 7 days after the first injection for analysis. (A) Measurement of the hematocrit given as the percentage of red blood cell (RBC) volume/total blood volume (with the time of injections indicated by the arrows). (B) Circulating platelet levels. (C and D) Quantification by flow cytometry of bone marrow megakaryocyte progenitors (MKP) and megakaryocytes (MK) (C) or pre-erythroid colony forming unit (PreCFUe) and proerythroblasts (D) and reported as meanÂąstandard deviation of % of cells per live cells (top) or as change relative to day 0 (bottom).

Gfi1b in the same luciferase assay and we found that, regardless of the isoform, the presence of a mutant allele significantly impaired the capacity of the WT protein to repress the human GFI1B promoter (Figure 2G and Online Supplementary Figure S2B). This indicated that all GFI1B mutants, including GFI1Bdel7, which seemed less stable in vitro than the other two mutants, act in a DN manner independently of the isoform.

Gfi1b-DN mice exhibit a moderate thrombocytopenic phenotype To test whether our Gfi1b-DN mice recapitulate the phenotype observed in GFI1B-RT patients, we measured their hematologic parameters and observed that all three 2464

lines exhibited lower levels of circulating platelets than WT littermates or hemizygous Gfi1bWT/0 mice that have been produced by crossing Gfi1bflox mice with a Credeleter EIIa-Cre, but only a fraction were thrombocytopenic, i.e., had >50% platelet reduction (Figure 3A). However, unlike GFI1B-RT patients, none of the mice showed signs of spontaneous bleeding; the Gfi1b-DN mice did not show a global increase of large circulating platelets and there was no significant difference in mean platelet volume (MPV) compared to the controls (Figure 3A). Interestingly, Gfi1b-DN mice exhibit higher levels of integrin IIb (CD41) at the surface of platelets than WT mice (Figure 3B), but the number of Îą-granules in platelets was not significantly decreased as in GFI1B-RT patients haematologica | 2020; 105(10)

Thrombocytopenia in Gfi1b mutant mice

Figure 7. Delay in thrombopoietic stress response in Gfi1b-DN mice. Time course analysis of the response to platelet depletion by intravenous (i.v.) injection of antiGPIba antibodies. 4-5 mice for each time point received one injection of anti-GPIbÎą and were either bled daily to measure absolute platelet counts by quantitative flow cytometry on a first experiment, or were sacrificed 2, 4 or 7 days later on a second experiment for analysis of the bone marrow and hematologic parameters. (A) Circulating platelet levels measured daily after the injection (arrows). (B) Measurement of the hematocrit of the mice from the second experiment and given as the percentage of red blood cell (RBC) volume/total blood volume. (C and D) Quantification by flow cytometry of bone marrow megakaryocyte progenitors (MKP) and megakaryocytes (MK) (C) or pre-erythroid colony forming unit (PreCFUe) and proerythroblasts (D) and reported as meanÂąstandard deviation of % of cells per live cells (top) or as change relative to day 0 (bottom).

(Figure 3C). Finally, none of the Gfi1b-DN mice showed overt erythropoietic defects as revealed by normal hematocrit and RBC shape and size (Figure 3D and E). However, consistent with the phenotype observed in human patients, the three Gfi1b-DN lines presented a moderate megakaryocytic hyperplasia characterized by an increase in both MK and megakaryocyte progenitors (MKP) compared to WT and hemizygous mice (Figure 4A and B). Although no RBC defects were observed in peripheral blood, an analysis of the bone marrow erythroid lineage revealed a strong increase in the pre-erythroid colony-forming unit (PreCFUe) and proerythroblast populations (Figure 4A and C) but no major difference in the earlier pre-megakaryocytic/erythroid progenitor haematologica | 2020; 105(10)

(PreMegE) and hematopoietic stem cell (HSC) populations or in the more differentiated erythroblast subpopulations (Online Supplementary Figure S4). Interestingly, the Gfi1b-DN mice exhibited a mild spleen enlargement compared to the WT and hemizygotes controls (Online Supplementary Figure S5A). However, there was no significant difference in splenic erythroid precursor subpopulations in Gfi1b-DN mice compared to controls (Online Supplementary Figure S5B).

Embryonic lethality and decreased life expectancy of Gfi1b-dominant negative mice in the context of a megakaryocyte-specific Gfi1b knockout To test whether the Gfi1b-DN mutant proteins retained 2465

H. Beauchemin et al. A

Figure 8. Gfi1b-DN mice respond efficiently to thrombopoietin (TPO) stimulation. Time course analysis of the response to the TPO analog romiplostim-induced thrombocytosis. 4-5 mice for each time point received a single intraperitoneal injection with romiplostim and sacrificed 2, 4 or 7 days later for analysis. (A) Circulating platelet levels (with the time of the injections indicated by the arrows). (B) Measurement of hematocrit given as the percentage of red blood cell (RBC) volume/total blood volume. (C and D) Quantification by flow cytometry of bone marrow megakaryocyte progenitors (MKP) and megakaryocytes (MK) (C) or pre-erythroid colony forming unit (PreCFUe) and proerythroblasts (D) and reported as meanÂąstandard deviation of % of cells per live cells (top) or as change relative to day 0 (bottom).

some residual functions in vivo, we first crossed Gfi1b-DN mice with Gfi1bEGFP animals in which the EGFP gene had been knocked into the Gfi1b locus to replace the coding sequence.23 Gfi1bEGFP/DN mice died in utero around day e14.5 and presented a phenotype similar to that of the full Gfi1b knockout (Online Supplementary Figure S6A). To measure the effect of the Gfi1b-DN mutants on platelet production in the context of an MK-specific knockout of the remaining Gfi1b allele, we bred Gfi1bDN mice with Pf4-cre:Gfi1bflox mice that lack GFI1B in MK and exhibit a severe thrombocytopenia (Online Supplementary Figure S6B). Mice with the desired genotype (Pf4-cre:Gfi1bflox/DN) were produced well below the mendelian rate, suggesting either in utero or perinatal lethality (Figure 5A and Online Supplementary Figure S6C). Moreover, the few mice that were born died even faster than Pf4-cre:Gfi1bflox/flox mice, suggesting a deleterious 2466

effect mediated by the presence of a Gfi1b-DN allele in this context (Figure 5B). The few Pf4-cre:Gfi1bflox/DN mice that developed to term, suffered from a severe form of thrombocytopenia similar to that of the Pf4-cre:Gfi1bflox/flox mice although some mice exhibited slightly higher platelet counts than the knockout (Figure 5C). These mice also showed an increase in frequency of bone marrow MKP and MK compared to WT controls and similar to the knockout (Online Supplementary Figure S6D), indicating that, in the megakaryocytic linage, Gfi1b-DN mutants are unable to rescue the knockout phenotype. Interestingly, several Pf4cre:Gfi1bflox/DN mice had a severely reduced hematocrit suggesting anemia as the probable cause of death (Figure 5C). Indeed, the spleen of the Pf4-cre:Gfi1bflox/DN mice were severely enlarged, and this splenomegaly was associated with a dramatic increase of proerythroblasts and haematologica | 2020; 105(10)

Thrombocytopenia in Gfi1b mutant mice

basophilic erythroblasts but with a decrease of splenic orthochromatophilic erythroblasts (Figure 5D and Online Supplementary Figure S6E). An analysis of the bone marrow revealed a similar erythroid pattern with an increase in PreCFUe, proerythroblasts and basophilic erythroblast along with a decrease in orthochromatophilic erythroblasts (Figure 5E). This differentiation block at the basophilic stage suggests an ineffective erythropoiesis (Figure 6F) in the context of a perpetual erythropoietic stress due to recurring spontaneous bleeding.

Gfi1b-dominant negative mice can respond properly to an acute erythropoietic stress To further explore the possibility that Gfi1b-DN alleles cause defects during erythropoietic stress, we treated WT and heterozygous Gfi1b-DN mice with phenylhydrazine (PHZ) that rapidly destroys circulating RBC. In a first experiment, mice were injected with PHZ, then bled over a period of 3 weeks every 2-4 days for hematocrit measurement to test how fast the RBC pool can be replenished. In this setting, the hematocrit recovered at the same rate in all mice to a full recovery by day 7 (Online Supplementary Figure S7A). In a second experimental setting, mice were grouped in cohorts that were sacrificed at given time points following PHZ injections to permit assessment of their hematologic parameters and bone marrow progenitors (Figure 6). Again, the Gfi1b-DN mice did not show any delay in hematocrit recovery compared to the controls (Figure 6A); similarly, platelets remained largely unaffected in all of the mouse treated with PHZ (Figure 6B). Analysis of both erythroid and megakaryocytic lineages in the bone marrow revealed that the Gfi1b-DN mutant mice were able to achieve a normal response to the PHZ-induced hemolytic anemia with the sole exception being the proerythroblasts (and to a lesser extent the PreCFUe) that showed a significantly stronger response in the WT controls than in the Gfi1b-DN mice (Figure 6C and D and Online Supplementary Figure S7B and C).

Gfi1b-dominant negative mice are less efficient at recovering from severe acute thrombocytopenia than controls Next, we induced an acute thrombocytopenic state in the three Gfi1b-DN lines by intravenous (i.v.) injection of anti-GPIbÎą (CD42b) antibodies that lead to a rapid and complete platelet depletion. In a first experiment, mice from the three Gfi1b-DN lines as well as WT and hemizygote controls were injected and minimally bled 1 hour after the injection, then every day for a week, for quantitative platelet counts by FACS. The three Gfi1b-DN lines exhibited a slower platelet recovery compared to controls (Figure 7A). To analyze hematopoietic precursors, a new cohort of mice was injected but this time, the animals were sacrificed at day 2, 4 and 7 after platelet depletion, and blood and bone marrow samples were analyzed. The hematocrit remained normal (Figure 7B), but a stronger megakaryopoietic response was seen in Gfi1b-DN mutant mice, which exhibited a stronger increase in both MKP and MK than WT controls (Figure 7C). The erythroid lineage was not affected by the loss of platelets except for a short burst in the orthochromatophilic erythroblast compartment seen in all mice including the controls (Figure 7D and Online Supplementary Figure S8). haematologica | 2020; 105(10)

Gfi1b-dominant negative megakaryocytes can respond normally to thrombopoietin stimulation Because the Gfi1b-DN mice exhibited mildly dysregulated thrombopoiesis, we tested their response to TPO stimulation. Mice were given romiplostim, a fusion peptibody analog to TPO with a greater bioavailability than recombinant TPO28 and were analyzed at day 2, 4 or 7 after injection (Figure 8 and Online Supplementary Figure S9). Although romiplostim had no impact on the erythroid lineage except for a short burst in the orthochromatophilic erythroblast population, all mice (controls and Gfi1b-DN) showed a strong thrombopoiesis response as reflected by their high platelet counts associated with an increase in MKP and MK. Although the rate of platelet increase was similar between control and Gfi1b-DN mice, the platelet level remained consistently lower in Gfi1b-DN mice than WT controls (Figure 8A), indicating that MK in Gfi1b-DN mice are less efficient to produce platelets, but are still able to respond to thrombopoietic stress and TPO stimulation. In contrast to Gfi1bDN mice, Pf4-cre:Gfi1bflox/flox animals treated with romiplostim did not respond at all even after 7 days (Online Supplementary Figure S10).

Discussion Several autosomal dominant mutations have been identified in the gene GFI1B that are linked to the bleeding disorder type 17 (BDPLT17), three of which cause the production of a truncated protein lacking the two last zinc fingers, one causes the disruption of the last three zinc fingers and the others being missense mutations disrupting non-DNAbinding zinc fingers.6-11,29 Because of the variety of mutations that produce truncated proteins with DN properties, we decided to use CRISPR/Cas9 technology to generate indel mutations in the fifth zinc finger to mimic those found in humans. Injecting the CRISPR vectors directly into oocytes proved to be fairly efficient, although most of the founders produced showed some level of mosaicism that made transmission to F1 progeny more difficult. Interestingly, although ten mutant alleles were generated, only six different alleles were generated as some mutations were obtained multiple times independently, suggesting that the generation of indels might not be entirely random. This result is consistent with other reports suggesting that Cas9mediated insertions and/or deletions may be context specific.30-32 When modeling human mutations, including those that produce DN effects, the phenotypic consequences of those alleles may differ when translated into the mouse.33,34 Here, we demonstrate that forms of GFI1B truncated at the fifth zinc finger had a DN effect in vitro similar to the analog human GFI1B mutations found in patients.6 We also demonstrate that both murine short isoform 1 (GFI1B-p37) and long isoform 2 (GFI1B-p40) are not only able to repress target gene expression with an equivalent efficiency, but that they also can both act as DN alleles in a reporter gene assay when carrying the DN mutations. In humans, splice variants 1 (GFI1B-p37) and 2 (GFI1B-p32) exhibit distinct functions, with the long isoform 1 being important for megakaryocytic differentiation and the short isoform 2 being required for proper erythropoiesis.11,13,35,36 Whether the two murine isoforms have a similar lineage-specific functionality remains to be investigated. As both the major human long isoform 1 and the murine short isoform 1 are 2467

H. Beauchemin et al.

identical, they are expected to play similar roles. However, the minor isoforms 2 are quite different between mouse and human, with the human version lacking the two first non-DNA-binding zinc fingers and the mouse protein still carrying these domains separated from the four last zinc fingers by a spacer peptide. It is, therefore, difficult to predict if the isoforms 2 behave similarly in mouse and human even if they are expressed at similarly low levels relative to isoform 1.36,37 The reduction in platelet numbers measured in Gfi1b-DN mice was highly variable but never strong enough to cause overt bleedings. This was expected as it has been shown that mice are resistant to bleeding until platelet numbers are decreased by more than 95%.38 The GFI1B-RT patients show variations in severity ranging from severe lifelong bleeding histories to rare manifestations of bleeding events.4-6 Thus, the discrepancy in severity with the Gfi1b-DN mice might be due to the fundamental difference in normal number of platelets between these two organisms, with mice having normally around three times the amount of platelets per volume of blood than humans.39,40 Indeed, a modest decrease in platelet number in humans may easily bring their count under the critical level of 150x109/L, whereas it would take a more drastic decrease to generate a thrombocytopenic state in mice. We noted that platelets in Gfi1b-DN mice were not significantly larger than controls, which is at variance with the macrothrombocytopenia seen in patients or in MK-specific Gfi1b knockout mice.6,10,25 Similarly, platelets produced in Gfi1b-DN heterozygous mice have normal amounts of α-granules unlike the Gray platelet syndrome reported in GFI1B-RT patients, indicating another difference between this model and the human disease,6,8,9,13,24,41,42 except in the cases of mutations affecting only non-DNA binding zinc fingers of GFI1B and that do not exhibit α-granule deficiency.10-12 Moreover, platelets from Gfi1b-DN mice exhibit increased expression of the fibrinogen receptor integrin αIIbβ3 (CD41/CD61) which also contrasts with GFI1B-RT patients who have normal CD41 expression, but is consistent with MK-specific Gfi1b knockout animals.6,25 On the other hand, all three Gfi1b-DN lines show a megakaryocytic hyperplasia that translates into an increase in MKP and mature MK, which is an important characteristic for GFI1B-RT patients.6 Some studies have reported that mutations in GFI1B lead to defects in erythroid differentiation, albeit infrequently,7,9 but other studies did not observe such a phenotype.6 We could not detect any anisocytosis or poikilocytosis in Gfi1bDN mice, but all three lines developed a consistent and important expansion of late erythroid progenitors (preCFUe) and early precursor (proerythroblasts) that cannot be explained by a consequence of the thrombocytopenic state of these mice, since such a correlation was not observed in Pf4-cre:Gfi1bflox/flox mice that have an even stronger thrombocytopenia. It has been shown that the loss of GFI1B in mice leads to a complete block of the definitive erythroid lineage, occurring mainly at an erythroid progenitor stage between the PreMegE and the PreCFUe,22,24,43,44 suggesting that these stages of differentiation are particularly dependent on GFI1B. One surprising finding was that the combination of a mutant Gfi1b-DN allele with a MK-specific knockout allele led to a higher lethality and a lower transmission rate than a homozygous MK-specific Gfi1b knockout, suggesting that Gfi1b-DN mutant alleles are not only non-functional, 2468

but have deleterious effects.25 However, as the Pf4cre:Gfi1bflox/flox already exhibit an extremely severe thrombocytopenia with spontaneous bleeding,25 a deleterious effect of the Gfi1b-DN mutant alleles on the MK lineage seems unlikely. Rather, the severe splenomegaly associated with an almost complete block at the basophilic erythroblast stage and the severe anemia seen in Pf4-cre:Gfi1bflox/mutant mice is indicative of an ineffective erythropoiesis that could be triggered by excessive spontaneous bleeding caused by the lack of platelets in the MK-specific Gfi1b knockout mice.45,46 Because the Pf4-cre gene is not expressed in erythroid cells, this phenotype, which is not observed in Pf4-cre:Gfi1bflox/flox mice, is therefore very likely a consequence of the action of the GFI1B-DN form in the presence of a functional (non-deleted) floxed allele in erythroid cells. It is thus likely that the deleterious effect of the DN forms in erythroid precursors severely impairs erythropoiesis under chronic stress as for instance in a context of frequent spontaneous hemorrhages. It was therefore surprising that Gfi1b-DN mice responded just as well as the WT controls to PHZ-induced erythropoietic stress to replenish their RBC pool. This discrepancy might be due to the chronic nature of the erythropoietic stress seen in Pf4-cre:Gfi1bflox/DN mice compared to the acute erythropoietic challenge triggered by the PHZ. However, despite their capacity to replenish their RBC pool, Gfi1bDN mice were unable to further increase their PreCFUe and proerythroblasts as well as WT animals upon PHZ treatment. This poor response may be explained by the fact that these precursor populations in the heterozygous mutant mice were already at a higher than WT level, suggesting that these mice are already in a “stress response” state, even at steady state.47 Considering that, at steady state, their platelet counts are lower than WT animals, Gfi1b-DN mutant mice were expected to have a poorer thrombopoietic stress response, and indeed, platelet depletion experiments showed that Gfi1b-DN mice had increased response times for platelet recovery despite the fact that the expansion of their MK was even more pronounced than in controls, suggesting an impaired but not entirely dysfunctional platelet production. Treating mice with the TPO analog romiplostim triggered a proper response and did significantly increase platelet counts in these animals, which could be due to the already increased number of MK and MKP that may give the Gfi1b-DN mice a starting advantage. The effect of romiplostim, which is a drug already approved for some patients suffering from chronic idiopathic thrombocytopenia purpura, but was also reported in off-label uses such as treatment of persistent thrombocytopenia associated with stem cell transplantation or congenital amegakaryocytic thrombocytopenia,48-50 confirmed that heterozygous Gfi1b-DN mutant MK are still functional. This finding is important because it suggests that patients suffering from GFI1B-RT could potentially respond to a treatment with thrombopoietic agonists such as romiplostim. Ultimately, the mouse models described here could play a possible future role in drug screening for treatments for GFI1B-RT and other related thrombocytopenias, despite some significant differences compared to the features that are typical for human patients with this disease. A future development of mouse models based on the missense mutations in non-DNA-binding zinc finger seen in some patients could also possibly widen our understanding of the disease. haematologica | 2020; 105(10)

Thrombocytopenia in Gfi1b mutant mice

Acknowledgments The authors would like to thank the animal health technicians of the IRCM animal facility for excellent animal care and skillful technical procedures on animals and Jeannie Mui at the Facility for Electron Microscopy Research of McGill University for help in thin-sections preparation and microscope operation.

References 1. Tong B, Grimes HL, Yang TY, et al. The Gfi1B proto-oncoprotein represses p21WAF1 and inhibits myeloid cell differentiation. Mol Cell Biol. 1998;18(5):2462-2473. 2. Quick AJ, Hussey CV. Hereditary thrombopathic thrombocytopenia. Am J Med Sci. 1963;245:643-653. 3. Seip MF. Hereditary Hypoplastic Thromobocytopenia. Sangre. 1964;41:382384. 4. Kurstjens R, Bolt C, Vossen M, Haanen C. Familial thrombopathic thrombocytopenia. Br J Haematol. 1968;15(3):305-317. 5. Ardlie NG, Coupland WW, Schoefl GI. Hereditary thrombocytopathy: a familial bleeding disorder due to impaired platelet coagulant activity. Aust N Z J Med. 1976; 6(1):37-45. 6. Monteferrario D, Bolar NA, Marneth AE, et al. A dominant-negative GFI1B mutation in the gray platelet syndrome. N Engl J Med. 2014;370(3):245-253. 7. Stevenson WS, Morel-Kopp MC, Chen Q, et al. GFI1B mutation causes a bleeding disorder with abnormal platelet function. J Thromb Haemost. 2013;11(11):2039-2047. 8. Ferreira CR, Chen D, Abraham SM, et al. Combined alpha-delta platelet storage pool deficiency is associated with mutations in GFI1B. Mol Genet Metab. 2017;120(3):288294. 9. Kitamura K, Okuno Y, Yoshida K, et al. Functional characterization of a novel GFI1B mutation causing congenital macrothrombocytopenia. J Thromb Haemost. 2016;14(7):1462-1469. 10. Uchiyama Y, Ogawa Y, Kunishima S, et al. A novel GFI1B mutation at the first zinc finger domain causes congenital macrothrombocytopenia. Br J Haematol. 2018;181(6):843-847. 11. Rabbolini DJ, Morel-Kopp MC, Chen Q, et al. Thrombocytopenia and CD34 expression is decoupled from alpha-granule deficiency with mutation of the first growth factor-independent 1B zinc finger. J Thromb Haemost. 2017;15(11):2245-2258. 12. van Oorschot R, Marneth AE, Bergevoet SM, et al. Inherited missense variants that affect GFI1B function do not necessarily cause bleeding diatheses. Haematologica. 2019;104(6):e260-e264. 13. Schulze H, Schlagenhauf A, Manukjan G, et al. Recessive grey platelet-like syndrome with unaffected erythropoiesis in the absence of the splice isoform GFI1B-p37. Haematologica. 2017;102(9):e375-e378. 14. Kahr WH, Hinckley J, Li L, et al. Mutations in NBEAL2, encoding a BEACH protein, cause gray platelet syndrome. Nat Genet. 2011;43(8):738-740. 15. Larocca LM, Heller PG, Podda G, et al. Megakaryocytic emperipolesis and platelet function abnormalities in five patients with

haematologica | 2020; 105(10)

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

Funding These studies were supported by a Canadian Institutes of Health Research (CIHR) operating grant (MOP-111247), a Canada Research Chair (Tier1) in Hematopoiesis and Immune Cell Differentiation, a grant from the Canadian Hemophilia Society and a CIHR Foundation grant (FDN-148372) to TM; as well as a CIHR Project Grant to GD (PJT-156017).

gray platelet syndrome. Platelets. 2015;26 (8):751-757. Nurden AT, Nurden P. Should any genetic defect affecting alpha-granules in platelets be classified as gray platelet syndrome? Am J Hematol. 2016;91(7):714-718. Gunay-Aygun M, Falik-Zaccai TC, Vilboux T, et al. NBEAL2 is mutated in gray platelet syndrome and is required for biogenesis of platelet alpha-granules. Nat Genet. 2011; 43(8):732-734. Tomberg K, Khoriaty R, Westrick RJ, et al. Spontaneous 8bp Deletion in Nbeal2 Recapitulates the Gray Platelet Syndrome in Mice. PLoS One. 2016;11(3):e0150852. Kahr WH, Lo RW, Li L, et al. Abnormal megakaryocyte development and platelet function in Nbeal2(-/-) mice. Blood. 2013; 122(19):3349-3358. Deppermann C, Cherpokova D, Nurden P, et al. Gray platelet syndrome and defective thrombo-inflammation in Nbeal2-deficient mice. J Clin Invest. 2013;123(8):3331-3342. Guerrero JA, Bennett C, van der Weyden L, et al. Gray platelet syndrome: proinflammatory megakaryocytes and alpha-granule loss cause myelofibrosis and confer metastasis resistance in mice. Blood. 2014; 124(24):3624-3635. Saleque S, Cameron S, Orkin SH. The zincfinger proto-oncogene Gfi-1b is essential for development of the erythroid and megakaryocytic lineages. Genes Dev. 2002; 16(3):301-306. Vassen L, Okayama T, Moroy T. Gfi1b:green fluorescent protein knock-in mice reveal a dynamic expression pattern of Gfi1b during hematopoiesis that is largely complementary to Gfi1. Blood. 2007;109(6):2356-2364. Foudi A, Kramer DJ, Qin J, et al. Distinct, strict requirements for Gfi-1b in adult bone marrow red cell and platelet generation. J Exp Med. 2014;211(5):909-927. Beauchemin H, Shooshtarizadeh P, Vadnais C, Vassen L, Pastore YD, Moroy T. Gfi1b controls integrin signaling-dependent cytoskeleton dynamics and organization in megakaryocytes. Haematologica. 2017;102(3):484-497. Vassen L, Beauchemin H, Lemsaddek W, Krongold J, Trudel M, Moroy T. Growth factor independence 1b (gfi1b) is important for the maturation of erythroid cells and the regulation of embryonic globin expression. PLoS One. 2014;9(5):e96636. Vassen L, Fiolka K, Mahlmann S, Moroy T. Direct transcriptional repression of the genes encoding the zinc-finger proteins Gfi1b and Gfi1 by Gfi1b. Nucleic Acids Res. 2005;33(3):987-998. Wang B, Nichol JL, Sullivan JT. Pharmacodynamics and pharmacokinetics of AMG 531, a novel thrombopoietin receptor ligand. Clin Pharmacol Ther. 2004;76(6):628-638.

29. van Oorschot R, Hansen M, Koornneef JM, et al. Molecular mechanisms of bleeding disorder-associated GFI1BQ287* mutation and its affected pathways in megakaryocytes and platelets. Haematologica. 2019;104(7):1460-1472 30. Lemos BR, Kaplan AC, Bae JE, et al. CRISPR/Cas9 cleavages in budding yeast reveal templated insertions and strand-specific insertion/deletion profiles. Proc Natl Acad Sci U S A. 2018;115(9):E2040-E2047. 31. van Overbeek M, Capurso D, Carter MM, et al. DNA Repair Profiling Reveals Nonrandom Outcomes at Cas9-Mediated Breaks. Mol Cell. 2016;63(4):633-646. 32. Taheri-Ghahfarokhi A, Taylor BJM, Nitsch R, et al. Decoding non-random mutational signatures at Cas9 targeted sites. Nucleic Acids Res. 2018;46(16):8417-8434. 33. Hara S, Takada S. Genome editing for the reproduction and remedy of human diseases in mice. J Hum Genet. 2018;63(2): 107-113. 34. Vandamme TF. Rodent models for human diseases. Eur J Pharmacol. 2015;759:84-89. 35. Polfus LM, Khajuria RK, Schick UM, et al. Whole-Exome Sequencing Identifies Loci Associated with Blood Cell Traits and Reveals a Role for Alternative GFI1B Splice Variants in Human Hematopoiesis. Am J Hum Genet. 2016;99(2):481-488. 36. Laurent B, Randrianarison-Huetz V, Frisan E, et al. A short Gfi-1B isoform controls erythroid differentiation by recruiting the LSD1-CoREST complex through the dimethylation of its SNAG domain. J Cell Sci. 2012;125(Pt 4):993-1002. 37. Osawa M, Yamaguchi T, Nakamura Y, et al. Erythroid expansion mediated by the Gfi1B zinc finger protein: role in normal hematopoiesis. Blood. 2002;100(8):27692777. 38. Morowski M, Vogtle T, Kraft P, Kleinschnitz C, Stoll G, Nieswandt B. Only severe thrombocytopenia results in bleeding and defective thrombus formation in mice. Blood. 2013;121(24):4938-4947. 39. Fukuda T, Asou E, Nogi K, Goto K. Evaluation of mouse red blood cell and platelet counting with an automated hematology analyzer. J Vet Med Sci. 2017;79(10):1707-1711. 40. Crowther MA, Cook DJ, Meade MO, et al. Thrombocytopenia in medical-surgical critically ill patients: prevalence, incidence, and risk factors. J Crit Care. 2005;20(4):348-353. 41. Stevenson WS, Morel-Kopp MC, Ward CM. Platelets are not all gray in GFI1B disease. Clin Genet. 2015;87(3):299. 42. Fixter K, Rabbolini DJ, Valecha B, et al. Mean platelet diameter measurements to classify inherited thrombocytopenias. Int J Lab Hematol. 2018;40(2):187-195. 43. Randrianarison-Huetz V, Laurent B, Bardet V, Blobe GC, Huetz F, Dumenil D. Gfi-1B controls human erythroid and megakary-

2469

H. Beauchemin et al. ocytic differentiation by regulating TGFbeta signaling at the bipotent erythromegakaryocytic progenitor stage. Blood. 2010;115(14):2784-2795. 44. Garcon L, Lacout C, Svinartchouk F, et al. Gfi-1B plays a critical role in terminal differentiation of normal and transformed erythroid progenitor cells. Blood. 2005;105 (4):1448-1455. 45. Sjogren U. Erythroblastic islands and ineffective erythropoiesis in acute myeloid leukaemia. Acta Haematol. 1975;54(1):1117. 46. Suragani RN, Zachariah RS, Velazquez JG,

2470

et al. Heme-regulated eIF2alpha kinase activated Atf4 signaling pathway in oxidative stress and erythropoiesis. Blood. 2012;119(22):5276-5284. 47. Singh RP, Grinenko T, Ramasz B, et al. Hematopoietic Stem Cells but Not Multipotent Progenitors Drive Erythropoiesis during Chronic Erythroid Stress in EPO Transgenic Mice. Stem Cell Reports. 2018;10(6):1908-1919. 48. Doobaree IU, Newland A, McDonald V, et al. Primary Immune Thrombocytopenia (ITP) Treated with Romiplostim in Routine Clinical Practice: Retrospective Study from

the United Kingdom ITP Registry. Eur J Haematol. 2019;102(5):416-423. 49. Bento L, Bastida JM, Garcia-Cadenas I, et al. Thrombopoietin Receptor Agonists for Severe Thrombocytopenia after Allogeneic Stem Cell Transplantation: Experience of the Spanish Group of Hematopoietic Stem Cell Transplant. Biol Blood Marrow Transplant. 2019;25(9):1825-1831. 50. Pecci A, Ragab I, Bozzi V, et al. Thrombopoietin mutation in congenital amegakaryocytic thrombocytopenia treatable with romiplostim. EMBO Mol Med. 2018;10(1):63-75.

haematologica | 2020; 105(10)

ARTICLE

Hemostasis

Shear rate gradients promote a bi-phasic thrombus formation on weak adhesive proteins, such as fibrinogen in a von Willebrand factor-dependent manner

Ferrata Storti Foundation

Nicolas Receveur,1* Dmitry Nechipurenko,2,3,4* Yannick Knapp,5,6 Aleksandra Yakusheva,2,3,4 Eric Maurer,1 Cécile V. Denis,7 François Lanza,1 Mikhail Panteleev,2,3,4 Christian Gachet1 and Pierre H. Mangin1

Université de Strasbourg, INSERM, EFS Grand-Est, BPPS UMR-S1255, FMTS, Strasbourg, France; 2Faculty of Physics, Moscow State University, Moscow, Russia; 3 Federal Research and Clinical Centre of Pediatric Hematology, Oncology and Immunology, Moscow, Russia; 4Center for Theoretical Problems of Physicochemical Pharmacology, Moscow Russia; 5CNRS, Université Aix-Marseille, Ecole Centrale Marseille, IRPHE UMR7342, Marseille, France; 6Université Avignon, LAPEC EA4278, Avignon, France and 7HITh, UMR_S1176, INSERM, Université Paris-Sud, Université Paris-Saclay, Le Kremlin-Bicêtre, Paris, France 1

Haematologica 2020 Volume 105(10):2471-2483

*NR and DN contributed equally as co-first authors

ABSTRACT

lood flow profoundly varies throughout the vascular tree due to its pulsatile nature and to the complex vessel geometry. While thrombus formation has been extensively studied in vitro under constant flow, and in vivo under normal blood flow conditions, increased attention has been paid to the impact of complex hemodynamics such as flow acceleration found in stenosed arteries. We investigated the effect of flow acceleration, characterized by shear rate gradients, on the function of platelets adhering to fibrinogen, a plasma protein which plays a key role in hemostais and thrombosis. While we confirmed that under constant flow, fibrinogen only supports single platelet adhesion, we observed that under shear rate gradients, this surface becomes highly thrombogenic, supporting efficient platelet aggregation leading to occlusive thrombus formation. Interestingly, this phenomenon is general as it occurs on other weak adhesive matrices including laminins and thrombospondin-1. This shear rate gradient-driven thrombosis is biphasic with an initial step of slow platelet recruitment supported by direct plasma von Willebrand factor (VWF) adsorption to immobilized fibrinogen and followed by a second phase of explosive thrombosis initiated by VWF fiber formation on platelet monolayers. In vivo experiments confirmed that shear rate gradients accelerate thrombosis in a VWF-dependent manner. Together, this study characterizes a process of plasma VWF-dependent accelerated thrombosis on weak adhesive proteins such as, fibrinogen in the presence of shear rate gradients.

Correspondence: PIERRE MANGIN pierre.mangin@efs.sante.fr Received: August 26, 2019. Accepted: November 13, 2019. Pre-published: November 14, 2019. doi:10.3324/haematol.2019.235754

Introduction It has long been recognized that hemodynamics play important roles in hemostasis and thrombosis.1 A study coupling intravital microscopy to hydrodynamic studies provided evidence that shear gradients found at the site of vessel stenosis enhance thrombus formation in the post-stenotic area.2 The rheology-driven thrombi were essentially composed of discoid platelets as they formed independently of soluble agonists. Exacerbated platelet function and thrombosis in the poststenotic area was confirmed in another study, which proposed a role for von Willebrand factor (VWF) in this process.3 To date, the mechanism by which VWF exacerbates platelet activation and thrombosis under shear gradients has not been established. In addition to VWF, atherosclerotic plaque rupture exposes many other adhesive proteins including fibrinogen, which is known to support platelet adhesion and activation under low laminar blood flow conditions.4 However, whether haematologica | 2020; 105(10)

2471

N. Receveur et al.

this protein initiates and contributes to atherothrombosis under profoundly altered flow remains unknown. VWF is a well-studied adhesive protein for platelets that is central in both hemostasis and arterial thrombosis.5 VWF is found in the subendothelium, in Weibel-Palade bodies of endothelial cells, in α-granules of platelets and in the plasma. Upon vascular injury, plasma VWF becomes rapidly adsorbed to several adhesive glycoproteins found at the site of injury, including collagen, laminins, thrombospondin, fibrin and fibronectin.6 Immobilized VWF then unfolds under the action of shear forces and exposes

sites for the GPIb-IX-V complex supporting the recruitment of flowing platelets. VWF has the unique ability amongst all adhesive proteins to recruit platelets under elevated wall shear rate (WSR) conditions (>1,500 s-1).7-9 It has also been shown that unfolding of VWF is facilitated in the presence of shear gradients and elongational flow and can form a network of fibers which supports stable adhesion of several individual platelets.10-14 Fibrinogen is a soluble plasma glycoprotein which plays a pivotal role in hemostasis. In suspension, activation of platelets results in a change of the conformation of inte-

Figure 1. Schematic of a microfluidic flow chamber designed to generate shear gradients. (A) Schematic of the poly-dimethylsiloxane (PDMS)-based microfluidic flow chamber containing a straight rectangular channel with a width of 1 mm and a height of 0.1 mm, and a stenosed channel with similar dimensions, but presenting a 90% reduction in width in the central region. (B) Bright field microscope view of the microfluidic flow chamber: transition from zone 2 to zone 3 of the stenosed channel. (C) Velocity magnitude heat map at mid-height in the stenosed channel obtained by micro-particle image velocimetry: transition from zone 2 to zone 3. (D) Non dimensional velocity profiles U⁄Umax as a function of Y⁄Ymax or Z⁄Zmax (U is the velocity magnitude, Ymax is the half width of the stenosis, Zmax the half height) in the mid height and mid width of the microfluidic flow chamber in zone 3. (E) Computational fluid dynamic analysis presenting the wall shear rate (WSR) heat map at the chamber floor (z=0) throughout the whole chamber and in the zoomed region corresponding to the entrance of zone 3. Channel geometry in the computational fluid dynamic simulation (CFD) model corresponded to the stenosed version of the chamber shown in panel (A). (F) The streamlines for the central horizontal crosssection (50 µm above the chamber floor) of the zone 2 to zone 3 connection region were calculated using the CFD-derived velocity field.

2472

haematologica | 2020; 105(10)

Shear gradients a promote biphasic thrombosis

haematologica | 2020; 105(10)

2473

N. Receveur et al. I

Figure 2 (previous page). Shear rate gradients promote platelet aggregation, procoagulant activity and fibrin formation on a fibrinogen surface. Hirudinated human whole blood was perfused through channels of the microfluidic chamber coated with a solution of fibrinogen (300 mg/mL). (A) Representative differential interference contrast (DIC) microscopy images of platelets adhering to immobilized fibrinogen at 300 s-1 or 4,800 s-1 after 5 min. Scale bar: 10 mm. (B) The bar graphs represent the surface of adherent platelets obtained after 5 min of perfusion. The surfaces represent the mean ± standard error of the mean (SEM) in six random fields of five separate experiments performed with different blood donors. **P<0.005. (C) Representative DIC microscopy images of platelets adhering to immobilized fibrinogen in various zones of the 90% stenosed channel, after 5 min. Scale bar: 10 mm. Theoretical wall shear rate (WSR) in rectangular regions were indicated. (D) The bar graphs represents the surface of adherent platelets in various regions obtained after 5 min of perfusion. The surfaces represent the mean ± standard error of the mean (SEM) in one random field of six separate experiments performed with different blood donors. (E) Representative DIC microscopy images of platelets adhering to immobilized fibrinogen in zone 3 of the stenosed channel or in the central region of a straight rectangular channel (0.1 mm/0.1 mm), at 4,800 s-1 after 5 min. Scale bar: 10 mm. (F) The bar graphs represent the surface of adherent platelets obtained after 5 min of perfusion. The surfaces represent the mean ±SEM in six random fields of three separate experiments performed with different blood donors. *P<0.05. (G) Washed human platelets loaded with morphological and Ca2+ dyes were reconstituted with 50% (vol/vol) autologous packed red blood cells at a final concentration of 2.5x108 platelets/mL and perfused through the stenosed channel at 300 s-1 (in zone 1). Changes in fluorescence in platelets adhering in zone 1 and at the entrance of zone 3 were monitored for 5 min by confocal microscopy, and cytosolic Ca2+ concentrations were calculated. The dot plot of the maximal increase relative to the basal state in individual adherent platelets is shown (n=80 from four independent experiments). (H) Hirudinated blood was perfused over fibrinogen in the presence of Alexa 488-conjugatdd annexin-V (1/50) at 300 s-1 for 10 min. Procoagulant platelets were detected by their annexin-V positivity (greend platelets) using a epifluorescence microscopy. Representative DIC/epifluorescence images are shown. (I) The bar graph represents the surface of annexin-V labelling per mm² in zone 1 and zone 3 of six independent experiments, **P<0.01. (J) Recalcified citrated human whole blood was perfused over fibrinogen (300 mg/mL), in the presence of the specific Alexa 647-coupled anti-human fibrin antibody 59d8 (5 mg/mL). Representative DIC/fluorescence images represent fibrin formation in blue after 10 min of blood perfusion. (K) The bar graph represents the surface of fibrin formation per mm², in four independent experiments. Scale bar: 10 mm. *P<0.05. (L) Representative DIC microscopy images of platelets accumulated on immobilized fibrinogen in zone 3 at shear rate gradients (SRG) of 2.5 s-1/mm (corresponding to a wall shear rate (WSR) of 1,600 s-1), 4.9 s-1/mm (corresponding to a WSR of 3,200 s-1), 7.4 s-1/mm(corresponding to a WSR of 4,800s-1), 9.8 s-1/mm (corresponding to a WSR of 6,400 s-1) and 12.2 s-1/mm (corresponding to a WSR of 8,000 s-1) after 10 min or less. Scale bar: 10 mm. (M) The scatter plot represent the time to occlusion in the stenosed channel and limited to the 10 first min of perfusion (the value was set at 10 min if no occlusion occurred). Each dot represents the value of distinct flows performed with different blood donors, *P<0.05.

grin αIIbβ3 which supports fibrinogen binding, allowing the formation of inter-platelet bonds, i.e, platelet aggregation.15 When immobilized on a surface, fibrinogen supports attachment of resting platelets which is limited to low blood flow conditions (<1,000 s-1), as fibrinogen is not recognized to allow plasma VWF adsorption.7 The adhesive properties of fibrinogen under disturbed hemodynamic conditions remains completely unexplored. Understanding the effect of such flows on platelet adhe2474

sion to fibrinogen will help to better understand the mechanism underlying thrombosis since this pathological process is intimately linked to severely disturbed blood flow patterns. This study evaluated the effect of shear rate gradients (SRG) such as those found at the entrance of a vessel stenosis, on platelet function on fibrinogen. A major finding is that fibrinogen becomes a highly thrombogenic surface under elevated SRG allowing the formation of occluhaematologica | 2020; 105(10)

Shear gradients a promote biphasic thrombosis

sive thrombi. We identified and characterized a process of SRG-accelerated thrombosis that is bi-phasic with an initial step of slow platelet recruitment initiated through the ability of fibrinogen to adsorb plasma VWF, followed by a second stage during which VWF fibers of a couple of hundred microns form and initiate a very rapid accumulation of highly activated platelets to promote occlusive thrombosis. This process is general as it can be observed on other weak adhesive proteins such as laminins and thrombospondin-1.

Methods Preparation of microfluidic devices Microfluidic devices were made from poly-dimethylsiloxane (PDMS) by standard soft lithography techniques according to published methods.16 Briefly, a patterned photoresistant-based mold was made by standard soft lithography techniques. The liquid prepolymer PDMS and its curing agent were mixed thoroughly at a volumetric ratio of 10:1, placed in a vacuum chamber to degas and poured onto the mold followed by curing at 70° C for 2 hours. The polymerized PDMS channels were then carefully peeled away from the mold and holes were punched at both extremities. Finally, the irreversible bonding of PDMS to glass coverslips was performed upon exposure of both surfaces to an oxygen plasma. We made a chamber containing: 5 cm long straight rectangular channels with a 1 mm width and 0.1 mm height constant section (Figure 1A; straight channel) and; stenosed channels with identical dimensions in addition to a symmetric stenosis showing a 90% lumen reduction in the central part of the channel (Figure 1A; stenosed channel). This symmetric two sided stenosed channel is composed of a constricted central part (zone 3) connected on both sides to rectangular sections similar to straight channel (zones 1 and 5) through linearly varying sections (zones 2 and 4). Finally, we made a chamber containing square section channels with a width of 0.1 mm, a height of 0.1 mm and a length of 5 cm.

Computational modeling of VWF dynamics VWF dynamics was described with a model reported in.14 In order to reduce computational complexity, hydrodynamic correlations were not taken into account. The VWF dimer was treated as a sphere with a 70 nm radius. The polymer chain was represented as the chain of interconnected spheres, joined with Hookean springs. The interaction parameters were taken from.14 In order to infer conformational dynamics of VWF molecules within different flow lines of the chamber (flow lines lying within the central z-x plane were considered), the corresponding flow velocity profiles were taken from the computational fluid dynamic (CFD) simulation results. The local velocity gradient tensor was computed from polynomial interpolation of velocity profiles obtained from CFD simulations. In order to reduce the computational burden we considered VWF dynamics along the 1 mm of flow lines corresponding to stenosis entrance region (as far as this region possesses the highest values of elongational velocity gradients). VWF elongation d for a given moment of time was determined as maximum linear size of VWF molecule divided by the length of fully unfolded molecule. Values of VWF elongation exceeding unity correspond to flow-mediated extension of linear unfolded molecule. In case VWF elongation exceeded unity, the molecule was considered to be fully unfolded. In order to obtain a relative value of unfolding events we considered flow lines with different heights above fibrinogen surface: 1, 5, 10, 15 and 20 mm. For each height above the surface, 100 VWF trajectories were analyzed to infer the mean relative number of VWF unfolding events. The maximum relative haematologica | 2020; 105(10)

number of unfolding events were analyzed for each flow line. Analysis was performed for long (n=80 dimers) VWF multimers.

In vivo FeCl3 thrombosis and stenosis FeCl3-induced thrombosis was performed as previously described17–19 with some modifications. Briefly, 8-10 week-old mice were anesthetized by intraperitoneal injection of a mixture of xylasine (100 mg/kg) and ketamine (20 mg/kg). DiOC6 (0.4 mL of a 100 mM solution/g of body weight) was injected into the jugular vein to label the thrombus. The carotid artery was injured by lateral external application of a filter paper containing a 4% FeCl3 solution for 2.5 minutes (min). Four min after vessel injury, a 80% stenosis of the carotid or no stenosis (control) was obtained by compression with a micro-manipulator (Eppendorf Transferman, Hamburg, Germany). Thrombus formation was monitored with a fluorescent macroscope (Macrofluo®, Leica Microsystems, RueilMalmaison, France). Ethics approval for animal experimentation was obtained from the French Ministry of Research in accordance with the European Union Guidelines as defined by European laws. Detailed methods are described in the Online Supplementary Materials and Methods.

Results Development and characterization of a microfluidic device to study the effect of SRG on platelet adhesion to fibrinogen We developed a PDMS-based microfluidic chamber with a 90% narrowing in its central region (zone 3) in order to generate SRG, similar to those found in pathological settings (Figure 1A). Micro-particle imaging velocimetry (microPIV) measurements performed while perfusing a 40/60 glycerol/PBS mixture to mimic the blood viscosity, at a flow rate of 30 mL/min, showed terminal acceleration of the flow at the end of zone 2 and the entry of the stenosis (zone 3), (Figure 1B-C). The maximal flow velocity of 0.115 m/s measured in the centerline of the stenotic area (Figure 1C) is very close to the expected 0.113 m/s computed from the mass flow rate conservation through the model and to that of the straight channel (data not shown). In agreement, the value of the absolute average of WSR (4,750±300 s-1) of zone 3 is also very close to the 4,800 s-1 for which the stenosed channel was designed. Reynolds numbers (Re) based on maximum centreline velocity and hydraulic diameter were characteristic of a laminar flow in all sections of the flow chamber (zone 1-5: Re=0.46; zone 3: Re=3.36) indicating the absence of turbulence. This laminar flow behavior was confirmed by the velocity profiles that can be inferred from the flow field measurements in the different sections of zone 3 (Figure 1D). In agreement with the microPIV measurements, computational fluid dynamic analysis indicated that the WSR were symmetrically distributed within the stenosed channel and of similar values (for 30 mL/min: zone 3 of 4,800 s-1) (Figure 1E-F). Finally, analytical solution of Navier-Stokes equation for infinitely long squareshaped channels20 with a corresponding volumetric flow rate gives a WSR value in the center line of zone 3 of 4,792 s-1, which is in line with CFD and microPIV results.

Effect of SRG on the function of platelet adhering to fibrinogen and other weak adhesive proteins In agreement with previous reports,7 perfusion of hirudinated whole blood over immobilized fibrinogen in 2475

N. Receveur et al.

channels with a constant section results in a modest platelet adhesion under low WSR (300 s-1), and no platelet recruitment under high WSR (4,800 s-1), (Figure 2A-B). In sharp contrast, in zone 3 of the 90% narrowed channel, we observed that numerous platelets adhered at 4,800 s-1, and formed aggregates (Figure 2C-D). Control experiments indicated that this adhesion was specific as no platelet adhesion was observed in any region of the channel in the absence of coating or when the chamber was passivated with albumin (data not shown). The efficient platelet accumulation on fibrinogen in zone 3 did not rely on local geometry, since no platelet adhesion was observed at similar WSR (4,800 s-1) in a straight fibrinogen coated channel with identical dimensions as those of zone 3 (Online Supplementary Figure S1 and Figure 2E-F). We therefore concluded that the efficient adhesion on fibrinogen at high WSR (4,800 s-1) observed in zone 3, results from flow acceleration occurring upstream (exit of zone 2) which generates SRG (Figure 1C). At early time points, we observed a modest 34.3% increase in Ca2+ signaling of platelets adhering to zone 3 compared to those attaching in zone 1 (Figure 2G). In contrast, at later stages, numerous platelets were Annexin-V positive in zone 3 (4,800 s-1), but not in zone 1 (300 s-1), suggesting that the level of platelet activation increases over time and that SRG favor procoagulant platelet formation (Figure 2H-I). In agreement, a signal for fibrin was detected in zone 3 when recalcified citrated human blood was perfused (Figure 2J-K). When different SRG were used, we observed single platelets

adhering at 2.5 s-1/m and 4.9 s-1/m, platelet aggregates started to form at 7.4 s-1/ m and became massive resulting in channel occlusion at 9.8 s-1/m and 12.2 s-1/m (Figure 2LM and Online Supplementary Video S1). Of importance, this phenomenon appears general as it also occurred on other weak adhesive proteins including vascular laminins and thrombospondin-1 (Online Supplementary Figure S2). Together, these results indicate that SRG can enhance platelet function on weak adhesive proteins, such as fibrinogen resulting in occlusive thrombosis.

Thrombus formation on fibrinogen under SRG is bi-phasic To further characterize the SRG-mediated process of platelet aggregation on fibrinogen, video-microscopy experiments were conducted. Real-time analysis of the flow experiments revealed a stepwise process of in vitro thrombus formation on fibrinogen, with an initial phase characterized by a slow and heterogeneous recruitment of individual platelets, followed by a stage of very rapid platelet aggregation visualized after 180 seconds (s) (Online Supplementary Video S2 and Figure 3A). Importantly, platelet adhesion and thrombus growth occur at the entrance of zone 3. Quantification of the surface coverage confirmed the bi-phasic nature of platelet accumulation on fibrinogen in zone 3 (Figure 3B). This bi-phasic process contrasted with platelet adhesion in zone 1 within the same channel which was linear (Figure 3B). The second phase (180-360 s) led to a very marked acceleration of platelet accumulation,

Figure 3. Platelet accumulation in the stenosed region under shear rate gradients (SRG) is bi-phasic. Hirudinated human whole bloos was perfused through channels of microfluidic chambers coated with a fibrinogen solution (300 mg/mL). (A) Representative differential interference contrast (DIC) microscopy images of platelets accumulating on immobilized fibrinogen in the stenosed region at indicated time points. Scale bar: 10 mm. (B) Representative curves of platelet accumulation in zone 1 and zone 3 over time. (C) The bar graph represents the slope of platelet accumulation in zone 3 before and after 210 s. Values are the mean Âą standard error of the mean (SEM) in eight separate experiments, *P<0.05.

2476

haematologica | 2020; 105(10)

Shear gradients a promote biphasic thrombosis

haematologica | 2020; 105(10)

Figure 4 (previous page). Integrin αIIbβ3 flow and the GPIb/VWF axis support platelet adhesion under shear rate gradients (SRG). Hirudinated human whole blood was perfused through channels of microfluidic chambers coated with a fibrinogen solution (300 mg/mL) at indicated local wall shear rates. The blood was incubated with the anti-αIIbβ3 blockers, Reopro: 40 mg/mL or Integrilin: 40 mg/mL (A- B), the blocking anti-GPIbα antibody AK2 (10 mg/mL), (C-D), the antiGPIbα peptide OS-1 (800 nM), (C-D), the anti-VWF A1 domain antibody 701 (20 mg/mL), (E-F), or with PBS (A- B), an irrelevant antibody (C-F), or a scrambled peptide (C-D) as a control. The bar graph represents the surface of adherent platelets obtained after 5 min of perfusion. The surfaces represent the mean ± standard error of the mean (SEM) in six random field of four to six separate experiments performed with different blood donors. *P<0.05, **P<0.01. (G) Representative immunofluorescence images of zone 1 and zone 3 depicting von Willebrand factor (VWF) labelling after perfusion of platelet-poor plasma (PPP) reconstituted with red blood cells (RBC) over immobilized fibrinogen (300 mg/mL) or human serum albumin (HSA), (300 mg/mL) within the channels of the microfluidic chambers at 300 s-1. (H) The bar graph represents the surface of VWF labelling per mm² on fibrinogen or HSA after 5 min of perfusion in zone 1 or zone 3. ***P<0.001. (I) Microwells were coated overnight at 4°C with fibrinogen or HSA at 10 mg/mL. Increasing concentrations of purified VWF were then added, and bound VWF was detected with a peroxydase-conjugated anti-VWF antibody. Curves show the optical density (OD) at 490 nm, and values are the mean ± SEM in three separate experiments performed in duplicate. (J) Prediction model for the relative number of VWF unfolding events in entrance of zone 3 under indicated SRG. (K) PPP was mixed or not with RBC and perfused through stenosed channels of microfluidic chambers coated with fibrinogen (300 mg/mL) at 300 s-1. The channels were labeled with an anti-VWF antibody. The bar graph represents the surface of VWF staining per mm² in zone 3 after 5 min of perfusion. **P<0.01.

2477

N. Receveur et al. A

2478

haematologica | 2020; 105(10)

Shear gradients a promote biphasic thrombosis Figure 5 (previous page). The enhanced platelet aggregation under shear rate gradients occurring on fibrinogen results from plasma von Willebrand factor fiber formation. Hirudinated human whole blood was perfused through channels of the microfluidic flow chamber coated with a solution of fibrinogen (300 mg/mL). (A) Representative differential interference contrast (DIC) microscopy image obtained in zone 3. Scale bar: 20 mm. (B) Representative DIC/fluorescent images represent the platelet aggregates (DIC) and VWF fibers (orange) formed in zone 3 following shear rate gradients (SRG). Scale bar: 20 mm. (C) Representative confocal images and schematic of platelet aggregates formed in zone 3. Platelets appear in red (RAM.1-Cy3 at 2 mg/mL), von Willebrand factor (VWF) in green (anti-VWF-Alexa488 at 2 mg/mL), and the overlay in yellow. Scale bar: 10 mm. Schematic of platelet aggregates (red) and VWF fibers (green) formed in zone 3 following SRG. (D) Dot plot represents the size of aggregates formed in presence or absence of OS-1 that was perfused once a platelet monolayer on fibrinogen had already formed. The shading corresponds to the standard error of mean (SEM). Data were compared by 2-tailed Mann-Whitney tests, *P<0.0001. (E) Computational fluid dynamic (CFD)-derived heatmap of shear rate values calculated for the system of three adhered platelets in the stenosed part of the chamber. The central section of system is shown. Platelets were placed in the central z-x plane of the stenosed part of chamber, near the wall. Hydrodynamic conditions are the same as described in Figure 1E. (F) The dependence of maximum relative number of VWF unfolding events on the height above the fibrinogen surface. Computational modeling of VWF dynamics was performed using the CFD-derived values of shear and elongation rates at different heights above the surface. For each height 100 simulations of long VWF multimers dynamics were obtained. Given values correspond to zone 2-zone3 connection, where VWF unfolding was maximal (see peaks at Figure 4J). VWF molecule was considered unfolded only if its maximal size exceeded contour length of the multimer. Data correspond to the same hydrodynamic conditions as used for Figure 1E and Figure 5E (maximal SRG of 7.4 s-1/mm ). (G) The effect of flow line contraction illustrated with CFD simulation for system of three ellipsoidal platelets. The flow lines are shown in dark red color. Note the convergence of flow lines above the platelets. (H) Schematic illustration of the consequence of flow line contraction effect: objects of finite size (like flowing VWF depicted in the figure or platelet) traveling at some distance above the surface will eventually interact with the obstacle (platelet aggregate) due to flow line convergence above the obstacle (platelet) leading to much closer distance between the flowing object and the barrier (platelet).

with a 3-fold slope increase when compared to early time points (<210 s) (Figure 3B-C). This bi-phasic platelet accumulation is original since platelet aggregation under flow is recognized to be linear over a very wide range of WSR in the classical model of “in vitro thrombus formation” when whole blood is perfused over collagen (data not shown).

Characterization of the first step of platelet adhesion to fibrinogen in the stenosed region We next analyzed the mechanism involved in platelet adhesion to fibrinogen in zone 3. Two clinically used αIIβb3 blockers, ReoPro® and Integrilin®, completely prevented platelet adhesion in the straight control channel at 300 s-1, as well as in zone 3 of the stenosed channel (4,800 s-1), (Figure 4A-B). Blockade of GPIbα or VWF exhibited a similar effect in zone 3, but not in the control channel at 300 s-1 (Figure 4C-F). These results suggested that plasma VWF becomes adsorbed to fibrinogen under SRG. This was confirmed by VWF immunostaining in zone 3 after perfusion of platelet-poor plasma (PPP) through the channel (Figure 4G). VWF staining was also detected in zone 1 but the signal was much weaker when compared to zone 3 further demonstrating that VWF binds to fibrinogen (Figure 4H). Direct interaction between VWF and fibrinogen was demonstrated in a purified assay under static conditions showing a specific and dose-dependent binding of soluble VWF to immobilized fibrinogen (Figure 4I). The increased VWF adsorption observed in zone 3 might be the result of VWF unfolding. A computational modeling approach of VWF dynamics, indicated that SRG similar to those occurring in transition from zone 2 to zone 3, especially 9.8 s-1/ m and 12.2 s-1/ m, favored unfolding of VWF molecules (Figure 4J and Online Supplementary Video S3), while this was not the case under flow conditions found in zone 1 at 300 s-1 and zone 3 at 4,800 s-1 (Online Supplementary Figure S3). We observed a 27.5% reduction of VWF deposition in zone 3 when PPP was perfused in the absence of red blood cells, suggesting that these cells contribute to this process (Figure 4K). Together, these results demonstrate that platelet adhesion observed on fibrinogen in zone 3 relies on integrin αIIbβ3, the GPIb-IX complex, and on plasma VWF adsorption which is the result of its unfolding due to SRG.

Characterization of the second step of accelerated platelet aggregation in the stenosed region: a role for VWF fibers Real-time video-microscopy showed that aggregates presented an unusual extended shape in the flow direction haematologica | 2020; 105(10)

and contained VWF fibers (Figure 5A-B and Online Supplementary Video S2 and S4). The fibers were particularly long reaching several hundred microns. Confocal microscopy confirmed the presence of VWF fibers and showed that they did not form directly on the fibrinogen surface but rather on top of platelet monolayers (Figure 5C). Interestingly these fibers appeared concomitantly to the initiation of the accelerated thrombosis, suggesting a role for them in this process (Online Supplementary Video S4). This role was confirmed in an experiment in which blockade of the GPIbα/VWF axis with OS-1, once a platelet monolayer on fibrinogen had already formed, prevented the second phase of accelerated thrombosis (Figure 5D). Computational fluid dynamic simulation (CFD) of the flow around single adhered platelets in zone 3 demonstrates a very strong increase in shear at their surface (>10,000 s-1), suggesting that events of VWF unfolding are very likely to occur (Figure 5E). Moreover, computational analysis of VWF dynamics at different heights shows SRG-dependent increase of VWF unfolding with an increase of height above the fibrinogen surface, reaching a maximum at 10 mm above the surface (Figure 5F). This result provides an explanation for acceleration of thrombus growth due to SRG-mediated unfolding of VWF molecules, which might trigger VWF fiber formation. Besides, CFD also shows significant streamline contraction effects on a second layer of platelets (Figure 5G) which might be responsible for the observed increase in the rate of VWF and platelet transport to the surface of growing aggregates as illustrated in Figure 5H. Thus, enhanced VWF unfolding and fiber formation combined with streamline contraction effects might explain the observed acceleration of thrombus formation in zone 3.

SRG accelerate in vivo thrombus growth in a VWF-dependent manner To evaluate whether SRG accelerate thrombus growth in vivo, an experimental thrombosis approach was used. A 4% FeCl3 solution was applied to the carotid artery of adult mice to injure the vessel wall, and once a monolayer of platelets had formed (~270 s), the artery was severely stenosed (>80%) with a micromanipulator to generate SRG. Measurements of blood flows in the mouse carotid with a Doppler probe combined to CFD analysis indicated that the SRG generated in such a setting were very similar to those we applied in vitro as they were reaching 8.6 s-1/m (Figure 6A). Intravital microscopy showed that thrombus formation occurred much earlier in a stenosed vessel com2479

N. Receveur et al.

pared to the control (Figure 6B and Online Supplementary Video S5). Quantification confirmed that thrombus formation was more rapid in the presence of a stenosis (Figure 6C-D). As expected thrombus formation was severely impaired in VWF-deficient mice under normal conditions

(Figure 6E-F). Interestingly, even though a tendency of increased thrombosis could be observed in the presence of a stenosis in VWF-/- mice, no significant difference in the time of thrombus formation was observed in these mice in the presence or absence of a stenosis (Figure 6E-F). This

Figure 6. Thrombus growth is accelerated under shear rate gradients (SRG) in an injured mouse carotid artery. The common carotid arteries of adult mice were exposed and a ﬁlter paper saturated with FeCl3 was placed laterally on the left vessel for 2.5 minutes (min). After 4 min, the vessel was pinched with a micromanipulator to realize an 80% stenosis. (A) Computational fluid dynamic analysis presenting the wall shear rate (WSR) heat map at the middle of the stenosis carotid artery (y=0) with a radius of 250 µm containing a stenosis with a width of 400 mm and a length of 1.35 mm throughout the whole vessel. (B) Representative fluorescent images of a thrombus forming after vessel injury obtained by labeling the platelets with DiOC6. Scale bar: 100 mm. Arrows indicate direction of blood flow. (C) Tracings representing the mean surface area (± standard error of the mean [SEM], n = 9) of thrombus growth after application of FeCl3 in the presence or absence of a stenosis. (D) Bar graph represents the time needed to obtain half maximal thrombus area. The shading corresponds to the SEM. Data were compared by 2-tailed MannWhitney tests, *P<0.05. (E) Tracings representing the mean surface area (± SEM, n=5) of thrombus growth after application of FeCl3 and stenosis on a carotid of VWF-/- mice. (F) Bar graph represents the time in seconds (s) needed to obtain half maximal thrombus area. The shading corresponds to the SEM. Data were compared by 2-tailed Mann-Whitney tests, P>0.05.

2480

haematologica | 2020; 105(10)

Shear gradients a promote biphasic thrombosis

result confirms that VWF plays a central role in the accelerated thrombosis that occurs under conditions of blood flow acceleration and SRG.

Discussion In this study, we describe and characterize a process of bi-phasic thrombus formation on weak adhesive proteins such as fibrinogen taking place under blood flow acceleration which can be found in the vicinity of an atherosclerotic plaque. This process is original and differs from thrombus formation in perfusion assays occurring under physiological or pathological hemodynamic conditions by several features including: i) its occurrence on several weak adhesive surfaces such as fibrinogen, laminins or thrombospondin-1 which are well-known to support only single platelet attachment; ii) its bi-phasic nature with a first step of slow platelet recruitment followed by a second phase of very rapid platelet aggregation; iii) its high reactivity that results in rapid channel occlusion; iv) its crucial dependence on VWF, which is not only key in the first step of platelet recruitment to the surface, but is also required to initiate the second step of platelet aggregation through VWF fiber formation on pre-adherent platelets. The elevated reactivity of the thrombotic process occurring under SRG, was not only evidenced by the ability of platelets to adhere at high shear, but also by their increased activation, their pro-coagulant activity and by fibrin deposition resulting in occlusive thrombosis. This SRG-enhanced thrombus growth was also evidenced in vivo, after mechanically inducing a severe stenosis of a FeCl3-injured carotid artery. Our results are in line with the pioneer work of Nesbitt and co-authors who showed that shear rate gradients promote thrombus formation.2 While they proposed that this mechanism was largely independent of soluble agonists, we propose that the formation of highly reactive VWF fibers could promote enough activation to support platelet aggregation. Our results are in line with previous reports7 showing that under normal blood flow conditions, immobilized fibrinogen recruits single platelets under low WSR with a threshold around 1,000 s-1 above which platelets fail to attach.7 The reason for the lack of attachment above 1,000 s-1 relies in the limited affinity of integrin ÎąIIbÎ˛3 for its ligand.7 We were therefore surprised by the capacity of platelets to attach to immobilized fibrinogen at a high shear (4,800 s-1) in the restricted area of a 90% stenosed channel. This response is not due to the local square geometry of the microfluidic device (zone 3), since it is not observed in a straight channel with similar dimensions. Instead, this effect is most likely the result of profoundly altered flow conditions resulting from the stenosis which generates SRG. Numerical studies indicated that the SRG found in the transition from zone 2 to zone 3 favor unfolding of VWF multimers, a phenomenon known to increase its adsorption to adhesive substrates.9 This likely explains the significant increase of VWF binding to fibrinogen that we observed in zone 3. Moreover, VWF unfolding dissociates the A1/A2 region and thereby unblocks the binding site of GPIb-IX complex which explains platelet recruitment to immobilized fibrinogen observed in zone 3 at 4,800 s-1.21 A major observation of this work is the description of a physical and direct interaction between fibrinogen and haematologica | 2020; 105(10)

VWF in a purified system. We show that VWF binds to fibrinogen both under static and flow conditions, however, the binding is markedly increased in conditions of SRG, which could be explained by an increased VWF unfolding. The reason why this interaction has been overlooked so far most likely relies in the fact that in the absence of SRG, the amount of fibrinogen-bound VWF is insufficient to support platelet recruitment at high shear. We observed that under SRG the amount of VWF adsorption to fibrinogen increases (zone 3), which explains platelet recruitment at high shear. The underlying mechanism probably relies in SRG-mediated VWF unfolding which favors its adsorption. Moreover, we propose that the streamline contraction effects taking place after platelet adhesion may participate as these effects favor VWF transport to the surface. We also observed that red blood cells modestly contribute to this process, potentially by shear diffusion-mediated transport of VWF. A striking result of this work is the very rapid platelet aggregation occurring on platelet monolayers under SRG. Real-time video-microscopy showed that this event was concomitant to the formation of VWF fibers of several hundred microns long, which could represent a trigger. This was shown in vitro, since blockade of the GPIb/VWF interaction once the monolayer had already formed, completely prevented the rapid platelet aggregation process. Similar results were obtained in vivo as the acceleration of thrombosis occurring under SRG was also markedly reduced in VWF-deficient mice. Interestingly, the fibers did not form directly on the surface of the channels, but rather at a distance on top of adherent platelets or small aggregates. The reason why VWF fibers formed in the restricted region is the result of a combination of factors. Firstly, this process occurred under SRG which efficiently promote VWF unfolding that is a prerequisite for VWF self-association and fiber formation.14 Secondly, fiber formation took place several microns above the surface where numerical studies of the flow surrounding adherent platelets predicted extremely high shear rates (>10,000 s-1) which also favors VWF unfolding and fiber formation. Thirdly, simulation of VWF dynamics confirmed that VWF unfolding was maximal at 10 microns above the surface. Finally, CFD indicates streamline contraction effects at such sites which facilitate recruitment of plasma VWF and therefore fiber formation. This study identified a process of bi-phasic thrombosis on weak adhesive surfaces. Such a process was previously reported on fibrillar collagen in a stenosed channel.22 In both cases the bi-phasic process occurred when the blood flow is accelerated. Acceleration is found throughout the vasculature due to the pulsatile nature of the blood flow and the complex geometry of the vessels. However, the bi-phasic thrombosis leading to vessel occlusion described herein was achieved only for elevated values of SRG. Interestingly, despite simple hemodynamic conditions in our microfluidic system (laminar and constant blood flow), the disturbance of flow caused by the stenosis results in a complex phenomena, including accelerated thrombosis. Therefore, we propose that this process observed in the microfluidic-scale system is most likely relevant to pathological flows which can be found in the vicinity of an evolved atherosclerotic plaque presenting a severe stenosis. This has been shown by the accelerated thrombosis that occurred in an injured carotid artery when a 80% stenosis was applied with a micromanipula2481

N. Receveur et al.

Figure 7. Model of bi-phasic thrombus formation under shear rate gradients. Shear rate gradients (SRG) found at the site of vessel stenosis favor the adsorption of plasma von Willebrand factor (VWF) on immobilized weak adhesive proteins such as fibrinogen (1). This process allows recruitment of circulating platelets at elevated wall shear rates (2). VWF fibers are then forming on a monolayer of adherent platelets at distance of the surface (3). The VWF fibers initiate rapid platelet aggregation which can lead to occlusive thrombosis (4).

tor. Whether such thrombi contain large VWF fibers and whether such fibers represent a major trigger for occlusive thrombosis still needs to be demonstrated. Based on our observations it is tempting to speculate that blockade of VWF fiber formation could efficiently reduce the accelerated thrombosis occurring when the flow accelerates. In agreement with this view, it has recently been proposed that blockade of VWF self-association could represent a potential strategy to prevent arterial thrombosis, highlighting the therapeutic potential of interfering with SRG driven thrombosis.23 Marked flow acceleration also takes place in extracorporeal circulation systems notably at the site of tubing-connector junctions where sharp 90° angles are particularly prone to generate elevated SRG.24 Such devices were shown to adsorb plasma proteins including fibrinogen and to expose a surface to becoming thrombogenic under SRG, therefore explaining the occurrence of large thrombi notably in extracorporeal membrane oxygenation systems.24 Future studies should be conducted to determine the structure of these thrombi and evaluate whether they contain VWF fibers. In summary, we describe and characterize a mechanism

References 1. Nesbitt WS, Mangin PH, Salem HH, Jackson SP. The impact of blood rheology on the molecular and cellular events underlying arterial thrombosis. J Mol Med (Berl). 2006; 84(12):989-995. 2. Nesbitt WS, Westein E, Tovar-Lopez FJ, et al. A shear gradient-dependent platelet aggregation mechanism drives thrombus forma-

2482

of bi-phasic thrombus formation that is initiated on immobilized fibrinogen, relies on VWF and occurs specifically under SRG. This process is general as it can be triggered on several weak adhesive surfaces, such as fibrinogen, laminins or thrombospondin-1, which become thrombogenic when they are subjected to blood flow acceleration generating SRG and elongational flow. Under such hemodynamic conditions, plasma VWF molecules become adsorbed to the surface allowing a modest recruitment of circulating platelets. This process is amplified on adherent platelets forming reactive VWF fibers that enhance platelet adhesion and activation, promote the procoagulant function of platelets, allow fibrin deposition and result in intense platelet aggregation and occlusive thrombosis (Figure 7). Funding This work was supported by ARMESA (Association de Recherche et Développement en Médecine et Santé Publique). Acknowledgments We would like to thank Christophe Dubois for providing the anti-fibrin antibody.

tion. Nat Med. 2009;15(6):665-673. 3. Westein E, Van Der Meer AD, Kuijpers MJE, Frimat JP, Van Den Berg A, Heemskerk JWM. Atherosclerotic geometries exacerbate pathological thrombus formation poststenosis in a von Willebrand factor-dependent manner. Proc Natl Acad Sci U S A. 2013;110(4):1357-1362. 4. Bini A, Fenoglio JJJ, Mesa-tejada R, Kudryk B, Kaplan KL. Identification and distribution of use of monoclonal antibodies.

Arteriosclerosis. 1989;9(1):109. 5. Ruggeri ZM. Von Willebrand factor: looking back and looking forward. Thromb Haemost. 2007;98(1):55-62. 6. Bergmeier W, Hynes RO. Extracellular matrix proteins in hemostasis and thrombosis. Cold Spring Harb Perspect Biol. 2012;4(2):a005132. 7. Savage B, Saldivar E, Ruggeri ZM. Initiation of platelet adhesion by arrest onto fibrinogen or translocation on von Willebrand fac-

haematologica | 2020; 105(10)

Shear gradients a promote biphasic thrombosis tor . Cell . 1996;84:289-297. 8. Siedlecki CA, Lestini BJ, Kottke-Marchant KK, Eppell SJ, Wilson DL, Marchant RE. Shear-dependent changes in the three-dimensional structure of human von Willebrand factor. Blood. 1996;88(8):2939-2950. 9. Schneider SW, Nuschele S, Wixforth A, et al. Shear-induced unfolding triggers adhesion of von Willebrand factor fibers. Proc Natl Acad Sci U S A. 2007;104(19):7899-7903. 10. Barg A, Ossig R, Goerge T, et al. Soluble plasma-derived von Willebrand factor assembles to a haemostatically active filamentous network. Thromb Haemost. 2007;97(4):514-526. 11. Colace T V., Diamond SL. Direct observation of von Willebrand factor elongation and fiber formation on collagen during acute whole blood exposure to pathological flow. Arterioscler Thromb Vasc Biol. 2013; 33(1):105-113. 12. Fu H, Jiang Y, Yang D, Scheiflinger F, Wong WP, Springer TA. Flow-induced elongation of von Willebrand factor precedes tensiondependent activation. Nat Commun. 2017;8(1):324.

haematologica | 2020; 105(10)

13. Savage B, Sixma JJ, Ruggeri ZM. Functional self-association of von Willebrand factor during platelet adhesion under flow . Proc Natl Acad Sci USA . 2002;99:425-430. 14. Sing CE, Alexander-Katz A. Elongational flow induces the unfolding of von willebrand factor at physiological flow rates. Biophys J. 2010;98(9):L35-L37. 15. Bennett JS. Platelet-fibrinogen interactions. Ann N Y Acad Sci. 2001;936(1):340-354. 16. McDonald JC, Duffy DC, Anderson JR, et al. Fabrication of microfluidic systems in poly(dimethylsiloxane). Electrophoresis. 2000;21(1):27-40. 17. Eckly A, Hechler B, Freund M, et al. Mechanisms underlying FeCl3-induced arterial thrombosis. J Thromb Haemost. 2011;9(4):779-789. 18. Kurz KD, Main BW, Sandusky GE. Rat model of arterial thrombosis induced by ferric chloride. Thromb Res. 1990;60(4):269280. 19. LĂŠon C, Eckly A, Hechler B, et al. Megakaryocyte-restricted MYH9 inactivation dramatically affects hemostasis while preserving platelet aggregation and secre-

tion. Blood. 2007;110(9):3183-3191. 20. Bahrami M, Yovanovich MM, Culham JR. Pressure drop of fully-developed, laminar flow in microchannel of arbitrary cross-section. J Fluids Eng Trans ASME. 2006; 128(5):1036-1044. 21. Aponte-SantamarĂa C, Huck V, Posch S, et al. Force-sensitive autoinhibition of the von willebrand factor is mediated by interdomain interactions. Biophys J. 2015; 108(9):2312-2321.. 22. Bark DL, Para AN, Ku DN. Correlation of thrombosis growth rate to pathological wall shear rate during platelet accumulation Bark - 2012 - Biotechnology and Bioengineering - Wiley Online Library. Biotechnol Bioeng. 2012. 23. Chung DW, Chen J, Ling M, et al. High-density lipoprotein modulates thrombosis by preventing von Willebrand factor self-association and subsequent platelet adhesion. Blood. 2016;127(5):637-645. 24. Hastings SM, Ku DN, Wagoner S, Maher KO, Deshpande S. Sources of circuit thrombosis in pediatric extracorporeal membrane oxygenation. ASAIO J. 2017;63(1):86-92.

2483

ARTICLE Ferrata Storti Foundation

Haematologica 2020 Volume 105(10):2484-2495

Coagulation & its Disorders

Murine tissue factor disulfide mutation causes a bleeding phenotype with sex specific organ pathology and lethality Susanna H. M. Sluka,1* Simon F. Stämpfli,1,2,3* Alexander Akhmedov,1 Tanja Klein Rodewald,4 Adrián Sanz-Moreno,4 Marion Horsch,4 Paula Grest,5 Andrea S. Rothmeier,6 Birgit Rathkolb,4,7,8 Anja Schrewe,4 Johannes Beckers,4,8,9 Frauke Neff,4 Eckhard Wolf,7 Giovanni G. Camici,1 Helmut Fuchs,4 Valerie Gailus Durner,4 Martin Hrabě de Angelis,4,8,9 Thomas F. Lüscher,1,2 Wolfram Ruf6,10 and Felix C. Tanner1,2 *SHMS and SFS contributed equally as co-first authors

1 Center for Molecular Cardiology, University of Zurich, Zurich, Switzerland; 2Department of Cardiology, University Heart Center, University Hospital, Zurich, Switzerland; 3Cardiology Division, Heart Center, Luzerner Kantonsspital, Luzern, Switzerland; 4German Mouse Clinic, Institute of Experimental Genetics, Helmholtz Zentrum München and German Research Center for Environmental Health, Neuherberg, Germany; 5Institute of Veterinary Pathology, University of Zurich, Zurich, Switzerland; 6Department of Immunology and Microbiology, Scripps Research, La Jolla, CA, USA; 7Institute of Molecular Animal Breeding and Biotechnology, Gene Center, Ludwig-Maximilians-University München, Munich, Germany; 8German Center for Diabetes Research (DZD), Neuherberg, Germany; 9 Experimental Genetics, School of Life Science Weihenstephan, Technische Universität München, Freising, Germany and 10Center for Thrombosis and Hemostasis Johannes Gutenberg University Medical Center, Mainz, Germany

ABSTRACT

T Correspondence: FELIX C. TANNER felix.tanner@usz.ch Received: February 7, 2019. Accepted: August 30, 2019. Pre-published: September 5, 2019. doi:10.3324/haematol.2019.218818 ©2020 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.

2484

issue factor (TF) is highly expressed in sub-endothelial tissue. The extracellular allosteric disulfide bond Cys186–Cys209 of human TF shows high evolutionary conservation and in vitro evidence suggests that it significantly contributes to TF procoagulant activity. To investigate the role of this allosteric disulfide bond in vivo, we generated a C213G mutant Tf mouse by replacing Cys213 of the corresponding disulfide Cys190-Cys213 in murine Tf. A bleeding phenotype was prominent in homozygous C213G Tf mice. Pre-natal lethality of one third of homozygous offspring was observed between embryonic (E) day E9.5 and E14.5 associated with placental hemorrhages. After birth, homozygous mice suffered from bleedings in different organs and reduced survival. Homozygous C213G Tf male mice showed higher incidence of lung bleedings and lower survival rates than females. In both sexes, C213G mutation evoked a reduced protein expression (about 10fold) and severely reduced pro-coagulant activity (at least 100-fold). Protein glycosylation was impaired and cell membrane exposure decreased in macrophages in vivo. Single housing of homozygous C213G Tf males reduced the occurrence of severe bleeding and significantly improved survival, suggesting that inter-male aggressiveness might significantly account for the sex differences. These experiments show that the TF allosteric disulfide bond is of crucial importance for normal in vivo expression, post-translational processing and activity of murine TF. Although C213G Tf mice do not display the severe embryonic lethality of Tf knock-out mice, their postnatal bleeding phenotype emphasizes the importance of fully functional TF for hemostasis.

Introduction Tissue factor (TF) is the cellular activator of the extrinsic pathway of blood coagulation. It is expressed in the sub-endothelial wall of blood vessels and organ parenchyma, leading to activation of coagulation after vessel injury. Levels of TF haematologica | 2020; 105(10)

Sex specific TF bleeding phenotype

expression, however, differ significantly between organs: heart, brain, lung, and uterus exhibit very high TF expression, while TF is barely detectable in skeletal muscle and joints.1,2 Total Tf deletion in mice leads to embryonic death due to insufficient development of yolk sac vessels and vascular failure.3-5 A TF transgene inducing low expression of human full length TF (flTF) in mice (low-TF mice, ~1% of normal murine TF levels) is sufficient to rescue Tf knock-out (KO) mice from embryonic lethality6 but not to restore normal postnatal hemostatic function. These mice suffer from spontaneous hemorrhages particularly in organs which in wild-type (wt) mice express high levels of TF such as the heart, lung, brain, gastrointestinal tract, and testis.7-9 Fatal lung hemorrhages are very common in low-TF mice8,10 and are further enhanced when exposed to either intratracheal lipopolysaccharide11 or infected with influenza A virus.12 Similarly, fatal brain hemorrhages have been observed in low-TF mice.10 Furthermore, female low-TF mice exhibit fatal hemorrhages in the placenta and uterus during pregnancy and post partum, suggesting that TF is crucial for uterine hemostasis.13 In line with these observations, tail bleeding time in low-TF mice was increased.14 Due to its sub-endothelial localization, TF is separated from circulating clotting factors in the absence of an injury. Minor amounts of TF are however expressed on the surface of monocytes and come in contact with blood.15,16 To control pro-coagulant activity, TF activity is post-translationally suppressed by a mechanism called encryption.17 The mechanism of encryption is controversially discussed: several studies suppose that predominantly exposure of negatively charged phospholipids account for the decryption of TF activity,18,19 while others suggest activation via thiol disulfide exchange reactions20-25 involving complement activation. Human TF indeed contains two disulfide bonds (Cys49-Cys57 and Cys186-Cys209) which are conserved in mice (Cys47-Cys55 and Cys190-Cys213) and many other species. The C-terminal bond shares sequence features of other allosteric disulfide bonds and is important for TF pro-coagulant activity26 and thus crucial for regulation of TF activity in cell cultures.27,28 Release of procoagulant, TF-containing extracellular vesicles (EV) from bone marrow-derived macrophages (BMDM) and smooth muscle cells was shown to be regulated by ATP stimulation of the purinergic receptor P2X7,29 also depending on thiol-exchange reactions.29 We generated a C213G Tf knock-in (KI) mouse model to investigate the function of disulfide bond mutated TF in vivo and its contribution to hemostasis.

Methods Construction of the C213G Tf targeting vector The coding sequence of the murine full-length Tf isoform containing a T721G mutation as well as the flanking genomic regions were inserted into KpnI/NotI restriction sites of a LNTK vector to replace the genomic TF locus. Embryonic stem cell transfection (129P2/OlaHsd) and blastocyst injection was conducted by PolyGene (Ruemlang, Switzerland). Chimeric mice were bred with C57BL/6J mice.

Animals One mouse colony was bred of heterozygous animals on a mixed 129P2/OlaHsd-C57BL/6J background (50% from each haematologica | 2020; 105(10)

strain). Mice were backcrossed to C57BL/6J (B6). All tests performed were approved by the responsible authority of the local government.

Embryonic analysis For embryonic analyses heterozygous C213G/+ Tf mice were bred and females were controlled for plugs every morning (day of the plugs considered E0.5).

Pathological analysis All organs were fixed in 4% buffered formalin and embedded in paraffin for histological examination. Four-µm-thick sections were cut and stained.

Expression profiling Total RNA was isolated from organs just before microarray hybridization. Amplified RNA was hybridized and analyzed on Illumina MouseRef8 v2.0 Expression Bead Chips containing about 25,000 probes.

Bleeding time After onset of anesthesia, tails were pre-warmed in a 37°C water bath for 10 min. Then 0.5 mm of the tail tip was amputated, and the tail was immediately put back to 37°C PBS.

Tail-cuff blood pressure measurement Blood pressure was measured in non-anesthetized mice with a non-invasive tail-cuff method using the MC4000 Blood Pressure Analysis Systems.

TF expression analysis RNA was extracted using TRIzol Reagent (Molecular Research Center). Transcript levels were quantified by Real Time PCR using SyBr Green Master Mix on an Applied Biosystems 7300 System. Protein was extracted by grinding of organ tissue in lysis buffer. Samples were separated by 10% SDS-PAGE and transferred to a PVDF membrane (Immobilion®-FL, Millipore). The membrane was stained with anti-mouse TF rabbit antiserum (R8084)29 and anti-human GAPDH (cross-reactive to mouse) mouse monoclonal (MAB374, Millipore) primary antibodies, followed by secondary antibodies.

TF activity Tissue was lysed in HEPES-saline containing 0.02% sodium azide (HBS) and diluted in HBS containing 1 mg/mL BSA and 50 µM phospholipid vesicles to measure TF activity using a plasma clotting assay.

Macrophage experiments Femora and tibiae were dissected and flushed with RPMI. Cells were resuspended in 90% FCS, 10% DMSO. BMDM were generated from total BM cultures as described previously.30 Cell surface TF activity was determined with 0.5 nM factor VIIa (FVIIa) and 50 nM factor X (FX) in HBS by measuring a time course of FXa generation. TF EV activity was measured in HBS with 2 nM FVIIa and 100 nM FX. EV prothrombinase activity was measured in HBS with 10 nM FVa, 5 nM FXa and 500 nM prothrombin. RNA was isolated from macrophages using TRIzol Reagent. Transcript levels were quantified by real time PCR.

Statistical analyses Data are indicated as mean ± standard error of the mean (SEM). Unpaired Student’s t-test and two-way ANOVA were used as appropriate. For comparison of genotype distributions, a X2-test was applied. A P-value <0.05 was considered significant. 2485

S.H.M. Sluka et al.

Results

Partial developmental lethality in homozygous C213G Tf offspring

Generation of C213G Tf KI mice

Heterozygous mice were bred, and among 889 offspring on a C57BL/6J genetic background, 141 homozygous C213G/C213G (16.1%) Tf, 508 heterozygous C213G/wt Tf (57.5%), and 240 (26.4%) wt animals were found at weaning. At birth, among 320 offspring, 56 homozygous C213G/C213G (17.5%), 170 heterozygous C213G/wt (53.1%), and 94 wt (29.4%) were found, excluding significant post-natal lethality (Figure 1B). Thus, one third of homozygous C213G/C213G pups were missing at birth

To generate mice expressing C213G Tf under the control of the endogenous Tf promoter, a replacement-type targeting vector was cloned. This vector contained the coding sequence of C213G full length Tf, which was flanked by 3 kb 5’ and 5.2 kb 3’ arms of homology; thus, the endogenous Tf sequence was replaced by the C213G TF coding sequence by homologous recombination (Figure 1A).

Figure 1. Genetic targeting, genotype distributions and survival. (A) The murine Tf allele was targeted with a replacement-type vector containing the murine flTF coding sequence with T721G nucleotide exchange in exon 5, flanked by 3 kb 5’ and 5.2 kb 3’ homology arms. Diphtheria toxin (dt) was used for negative selection. After homologous recombination into 129P2/OlaHsd embryonic stem cells, a loxP flanked neomycin resistance cassette was removed by transfection with a Cre expression plasmid. (B) Genotype distribution of offspring from heterozygous breeding pairs on a B6 background were analyzed at different time points. Genotype distributions were compared to a mendelian 25%/50%/25% distribution as well as to the distributions at other time points. (C) Reduced survival of homozygous C213G Tf mice. Survival curve of homozygous male and female C213G/C213G Tf offspring on mixed and B6 genetic background (male mixed n=32, female mixed n=32, male B6 n=32, female B6 n=43). Dead and diseased mice meeting the termination criteria were both included in the analysis.

2486

haematologica | 2020; 105(10)

Sex specific TF bleeding phenotype

Figure 2. Macroscopic images of lung hemorrhages and iron detection (Prussian blue staining) in heart, lung and brain hemorrhages. (A) Representative macroscopic images of lung bleedings in male and female homozygous C213G/C213G Tf mice. (B) Lung, (C) heart, and (D) brain consecutive sections from wild-type (wt) and homozygous C213G/C213G Tf mice were stained with hematoxylin and eosin (H&E) for morphological analysis and Prussian blue to show ferric iron (Fe3+) in blood. The area of the higher magnification pictures (20x) is demarcated with a rectangle in the lower magnification ones (aprox. 1.45x in heart/brain and 5x in lung). Iron deposits within lung macrophages are particularly obvious in (B).

haematologica | 2020; 105(10)

2487

S.H.M. Sluka et al.

and both genotype distributions at weaning and at birth differed significantly from a mendelian distribution. Genetic deletion of the Tf gene in mice leads to embryonic lethality between E9.5 and E10.5.3 To determine whether homozygous C213G/C213G Tf embryos are generated at mendelian ratios, 70 E9.5 embryos on a B6 genetic background were genotyped. Eighteen homozygous C213G/C213G Tf (25.7%), 34 heterozygous C213G/wt Tf (48.6%), and 18 wt (25.7%) embryos were found, in accordance with mendelian segregation (Figure 1B). In contrast to Tf KO mice, histology did not show any impairment of yolk sac vessel integrity and normal levels of -SMA expression in homozygous C213G/C213G Tf yolk sacs at E9.5 (Online Supplementary Figure S1A).3 At E14.5, 11 vital homozygous C213G/C213G TF (16.4%), 34 heterozygous C213G/wt (50.8%) and 22 wt embryos (32.8%) were found, indicating that C213G/C213G Tf mice show partial developmental lethality between day E9.5 and E14.5. Embryonic growth required umbilical blood flow via the chorioallantoic placenta from day E10.5.31 Besides the 11 vital homozygous C213G/C213G Tf embryos, three necrotic homozygous embryos could be identified on day 14.5 showing retinal pigmentation which indicates development to at least E11.5.32 Additional resorbed implantation sites were present, which could not be genotyped. Placentas of both vital and necrotic homozygous C213G/C213G Tf embryos showed hemorrhages of variable number and size (Online Supplementary Figure S1A).

same age. At 17 weeks of age we found a similar incidence of hemorrhages in the heart (7 of 11 in females and 5 of 10 in males) and brain (2 of 11 in females and 2 of 10 in males) of male and female homozygous C213G/C213G Tf mice, while the incidence of lung hemorrhage was more pronounced in males (5 of 10) compared to females (1 of 11) (Online Supplementary Figure S1B). Lungs of male homozygous C213G/C213G Tf mice also exhibited macroscopically more hemorrhagic areas in comparison to female mice and these were accompanied by edema in severe cases (Figure 2A). Histologically, affected areas showed intra-alveolar hemorrhage and the presence of macrophages, eosinophilic fibrillar material in bronchial lumen, and peri-vascular and peri-bronchial inflammatory infiltrates (Figure 2B). In contrast to reported calcified testes in low-TF mice 9, we did not observe calcification of testes (Online Supplementary Figure S2A). Epi- and myocardial hemorrhages (Figure 2C) as well as inflammation representing myocarditis (Figure 3A) were observed in the hearts of homozygous C213G/C213G Tf mice. Furthermore, fibrotic changes were detectable in all the hearts of homozygous C213G/C213G Tf mice, even without acute hemorrhage or inflammation (Figure 3B). However, the blood pressure and heart rate of male and female C213G/C213G Tf mice were comparable to that of male and female wt mice (Online Supplementary Figure S2B).

Decreased survival of adult homozygous C213G/C213G Tf offspring

Using Illumina Bead array, we assessed gene expression profiles in hearts and lungs of male and female homozygous C213G/C213G Tf mice compared to the mean expression values of wt Tf mice (Figure 3C and Online Supplementary Table S1). 113 genes were significantly regulated in male and 178 genes in female C213G/C213G Tf mouse hearts. GO-terms which were both overrepresented in hearts of homozygous male and female C213G/C213G Tf mice were cellular movement, hematological system, immune cell trafficking, cancer, cardiovascular system, cellular growth and proliferation, skeletal and muscular disorders, cell-to-cell signaling and interaction, metabolic disease, cardiovascular disease, respiratory disease, and cell death. This is in line with inflammation and fibrotic remodeling. In male C213G/C213G Tf mouse lungs 69 genes were significantly upregulated and classified to the following overrepresented GO terms: inflammatory response, cellular movement, cell-to-cell signaling, hematological system, connective tissue, cancer, respiratory disease, cardiovascular disease, and lipid metabolism. In contrast, in female C213G/C213G Tf mouse lungs only 23 genes were weakly regulated (fold change of -3.662.10) while no GO term was overrepresented. Although the macroscopically least affected lungs were selected for the analysis, the differential gene regulation in male C213G/C213G Tf lungs compared to females suggests increased early pathological alterations and supports the sex difference in lung pathology.

Between the age of 3-4 weeks, about 10% homozygous C213G/C213G Tf mice on a B6 background developed a hydrocephalus, while this was not observed in C213G/C213G Tf animals on a mixed background (data not shown). This may be due to the tendency of wt B6 mice to develop spontaneous hydrocephalous.33 With progressing age some homozygous C213G/C213G Tf mice developed symptoms of disease, mostly dyspnea, apathy and weight loss, but sometimes also neurological symptoms or spontaneous death. Diseased animals were euthanized when they met defined termination criteria. Both euthanized animals and those which died were included in the survival analysis. We found a sex difference in survival with impaired survival of male homozygous C213G/C213G Tf mice. The sex difference was more pronounced in animals on a mixed background (P<0.001) compared to animals on a B6 background (P=0.17, Figure 1C). Dissection of dead and euthanized mice showed macroscopic bleedings to heart, lung, and brain, which are organs with high TF expression. Hemorrhages were confirmed microscopically in heart, lung, and brain (Figure 2A-D). They were accompanied by secondary changes like epi- and myocarditis and fibrosis in hearts as well as edema (Figure 3A-B), bronchiolitis obliterans, and pneumonia in lungs.

Molecular changes in the heart and lung of C213G/C213G Tf mice

Sex difference in organ pathology In order to quantify the incidence of organ bleedings, we analyzed a group of male and female homozygous C213G/C213G Tf mice on a B6 background without apparent disease symptoms. This analysis of asymptomatic mice provided a measure for the incidence of organ bleedings in a uniform group of mice of both sexes of the 2488

Tail bleeding time and blood cell counts Tail bleeding was measured to investigate whether primary differences exist in hemostatic capacity. Tail bleeding time was prolonged in homozygous C213G/C213G Tf mice of both sexes, but no difference was detected between the sexes (Figure 4A). Comparing blood counts haematologica | 2020; 105(10)

Sex specific TF bleeding phenotype

Figure 3. Histological images of myocarditis and myocardial fibrosis, gene profile in heart and lung tissue. (A) Heart sections of wild-type (wt) and homozygous C213G/C213G Tf mice were stained with hematoxylin and eosin (H&E). Immune infiltrates are visible in the myocardium of C213G/C213G Tf animals. The area of the higher magnification picture (40x) is shown with a rectangle in the lower magnification one (15x). (B) Consecutive sections from wt and homozygous C213G/C213G TF mice were stained with H&E for morphological analysis and Picro-sirius red for better visualization of collagen. The area of the higher magnification (20x) pictures is demarcated with a rectangle in the lower magnification ones (approx. 1.45x). (C) Heatmaps of significantly (FDR<10%) regulated genes between homozygous C213G/C213G Tf and wt mice in heart and lung tissues. Genes are shown, if they are regulated in either male or female animals per tissue. Yellow (blue) indicates down (up) regulation compared to the mean expression of wt mice.

haematologica | 2020; 105(10)

2489

S.H.M. Sluka et al.

of C213G/C213G Tf mice to wt mice revealed a mild decrease of cellular hemoglobin content (MCH) and concentration (MCHC), slightly increased reticulocyte numbers with more immature reticulocytes, and an increased mean platelet size due to a higher proportion of big platelets (>12fl) in homozygous C213G/C213G Tf mice (Figure 4B). Even though these findings are consistent with compensated chronic bleedings, the absolute change of the values was small.

Tf expression and activity Tf mRNA expression was measured in the heart, brain, and lung of male and female homozygous C213G/C213G

Tf and wt mice using flTF specific primers (Figure 5A). flTF expression was slightly elevated in the brain and lung of homozygous C213G/C213G Tf mice compared to wt mice. In the hearts of male C213G/C213G Tf mice, flTf mRNA was significantly increased 9.6-times compared to hearts of wt mice suggesting a compensatory transcriptional induction of a functionally insufficient protein that may involve recently demonstrated feedback loops of the TF cytoplasmic domain.34 In C213G/C213G Tf female hearts a similar induction was seen, however with a higher variability. Induction of Tf transcription is in line with inflammatory and fibrotic remodeling seen in all hearts of both gen-

Group: flTF (6 months) additional group

Control (A)

Mutant (B)

Figure 4. Bleeding time and blood counts. (A) Bleeding times after amputation of 5 mm tail tip; male wt n=19, female wt n=13, male C213G/C213G n=15, female C213G/C213G n=11; **P<0.01, ***P<0.001. (B) Blood count analysis in C213G/C213G Tf and wt mice in with Sysmex XT-2000iV. Some of the values where significantly different between C213G/C213G Tf mice and wt Tf mice. However, absolute differences were all minor.

2490

haematologica | 2020; 105(10)

Sex specific TF bleeding phenotype

ders. To assess the protein expression and activity of C213G TF, organ lysates were prepared from heart, brain and lung. Protein expression of C213G TF was 10-times lower than expression of wt TF in brain and lung. In hearts, protein expression of C213G TF and wt TF did not differ (Figure 5B), in line with the compensatory upregulation of gene transcription. Pro-coagulant activity was measured using a plasma clotting assay. Pro-coagulant activity of brain and lung lysates from C213G/C213G Tf mice was decreased about 1000-fold compared to wt Tf brains and lungs. In heart lysates, the difference in procoagulant activity was only about 100-fold (Figure 5C). A 10-fold reduction of protein expression in brains and lungs of C213G/C213G Tf mice is in line with a 10-fold stronger decrease in pro-coagulant activity compared to the heart. In both genotypes, there was no difference between the sexes, neither in expression nor in pro-coagulant activity.

The TF allosteric disulfide is required for Tf procoagulant function in macrophages We characterized the pro-coagulant activity of the C213G Tf mutant in primary bone marrow-derived macrophages. In order to assure unaltered induction of Tf in IFN primed and 4 hour (h) lipopolysaccharides (LPS) stimulated macrophages,29,30 we measured Tf mRNA levels with flTF-specific primers (Figure 6A). flTf mRNA levels were significantly higher in C213G/C213G TF macrophages which is expected as in these cells both alleles of endogenous Tf (yielding fully processed mRNA for flTF and as TF) were replaced with a gene that only yields the mutated flTf. Given the unimpaired mRNA induction, we next measured TF protein expression 90 minutes (min) and 4 h after LPS stimulation (Figure 6B). While TF induction was below the detection limit in the absence of IFNγ priming, a low molecular ~48 kDa form of TF was expressed after 90 min of stimulation in both IFNγ-primed wt and C213G/C213G Tf macrophages. After 4 h of stimulation, wt TF was predominantly expressed as a highly glycosylated protein which was not detected in macrophages from C213G/C213G Tf mice. Since glycosylation of TF is required for cell surface expression and procoagulant activity in intestinal epithelial cells,35 we next tested the procoagulant activity of intact macrophages. Macrophage cell surface TF is largely inactive and requires activation of the P2X7 receptor with ATP. C213G/C213G Tf macrophages displayed minimal procoagulant activity on their cell surface after ATP stimulation and no TF activity or antigen was detected on EV released from these cells (Figure 6C-D). Western blotting for other proteins typically released on ATP-induced EV36 revealed no quantitative abnormalities in EV release from C213G/C213G Tf macrophages, demonstrating that TF is selectively absent in EV derived from C213G/C213G Tf macrophages (Figure 6D). In addition, measurements of prothrombinase activity showed that the release of procoagulant lipids on EV was not different between wt and C213G/C213G Tf macrophages (Figure 6C). Thus, mutation of the allosteric disulfide of TF prevents glycosylation and normal cellular processing, leading to a severely impaired procoagulant activity in macrophages stimulated with inflammatory mediators.

Behavioral differences Having excluded that primary differences in protein expression and activity as well as general hemostatic haematologica | 2020; 105(10)

Figure 5. TF expression and activity in brain, lung and heart. (A) Semi-quantitative RT-PCR in brain, lung, and heart of male and female homozygous C213G/C213G TF mice and wt TF mice using flTF-specific primers and normalization for β-actin (n=6, *P<0.05). (B) Representative blots and quantification of TF protein expression in brain, lung and heart of male and female homozygous C213G/C213G TF mice and wt TF mice (n=4, **P<0.01, ***P<0.001). (C) Plasma clotting time measurements of TF activity in brain, lung and heart of male and female homozygous C213G/C213G TF and wt TF mice (n=4, **P<0.01, ***P<0.001).

2491

S.H.M. Sluka et al. A

Figure 6. TF expression and extracellular vesicle release in bone marrow-derived macrophages. (A) TF mRNA levels were determined by semi-quantitative RT-PCR using flTF-specific primers and normalization for β-actin. Bone marrow-derived macrophages cultured overnight without or with IFNγ (100 ng/mL) were stimulated with 1 mg/mL LPS for 4 hours (h); wild-type (wt) n=5, C213G/C213G n=4; ***P<0.001. (B) Western-blotting for TF with a polyclonal anti-mouse TF antibody of membrane fractions from macrophages with or without overnight IFNγ priming stimulated with 1 mg/mL lipopolysaccharides (LPS) for the indicated times (representing two independent experiments). (C) TF cellular and extracellular vesicle (EV) activity measured by FXa generation assay of INFγ primed and LPS stimulated macrophages with or without activation of the P2X7 receptor with 5 mM ATP for 30 minutes (min); n=3, *P<0.05, **P<0.01. Prothrombinase activity of the same EV preparations was determined as a measure for the release of procoagulant, phosphatidylserine (PS)-expressing MP. (D) Protein composition of EV released after ATP stimulation for 30 min was evaluated by Western blotting for TF, integrin β1, and β-actin. Densitometric quantification of protein levels from three independent experiments showed that levels on EV from C213G/C213G TF macrophages were 1.4±1.1 for integrin β1 and 1.4±1.7 for β-actin relative to wt-derived EV levels (representing two independent experiments).

capacity exist between the sexes, we investigated whether behavioral differences of male and female mice may underlie differences in disease severity in homozygous C213G/C213G Tf mice. At the age of 21 days, 12 homozygous C213G/C213G Tf males were weaned to individual cages and housed individually up to the age of 100 days to prevent aggressive fighting. All 12 males survived to the age of 100 days, which is a significant improvement compared to group-housed males (P=0.0326) (Figure 7A). After day 100, pulmonary hemorrhages were detected in 3 of 12 (25%) single-housed homozygous C213G/C213G Tf males (Figure 7B), in comparison to 50% lung hemorrhages in homozygous C213G/C213G Tf group housed males. This experiment demonstrates that lung hemorrhages develop spontaneously in homozygous C213G/C213G Tf males, but single housing reduces the risk for lung bleedings and lowers mortality.

Discussion After the description of the importance of the TF C-ter2492

minal disulfide bridge for TF pro-coagulant activity26 and evidence for its involvement in TF decryption20,21,27 it is crucial to understand its significance in an in vivo model. In the C213G Tf mouse model we observed changes similar to those previously reported in vitro: reduced protein expression, glycosylation, and surface translocation. In line with this, the pro-coagulant activity is reduced by at least a factor 100 independent of changes in protein expression. The C213G Tf mutant still exhibits sufficient pro-coagulant activity to ensure embryonic development in two thirds of the offspring and to allow early post-natal survival. The overall phenotype of C213G/C213G Tf mice shows impressive similarity to the phenotype of low-Tf mice, which express a human TF transgene in the absence of mouse Tf. Pro-coagulant activity in low-TF mice was estimated 1% of normal levels resulting from low expression of TF under the control of the human promoter.6 However, low-TF mice are born in accordance with Mendelian ratios,6 while C213G Tf mice displayed a reduced number of homozygous C213G/C213G Tf offspring at birth. Partial developmental lethality is observed when other clotting factors (FV, prothrombin) or the haematologica | 2020; 105(10)

Sex specific TF bleeding phenotype

P=0.0326

Figure 7. Single housing reduces incidence of lung bleedings and lethality in homozygous C213G/C213G male mice. (A) All 12 single-housed C213G/C213G Tf males survived up to 100 days, in contrast to C213G/C213G Tf males, which were group-housed (n=32). (B) Incidence of lung bleedings was reduced from 50% to 25% in homozygous C213G/C213G TF males.

thrombin receptor PAR1 are knocked out.37-39 Deficiency in PAR2, which is activated by the TF/FVIIa complex, results in a partial, but undefined developmental lethality,40 suggesting that thrombin generation and PAR1 signaling, but not PAR2 signaling, are crucial for development. Rescue experiments of murine TF (murine Tf) KO using different human TF transgenic lines suggested that the threshold of TF activity needed to support embryonic development is lower than that needed to maintain postpartum hemostasis, as some transgenic lines quantitatively rescue genetic deletion of murine Tf but rescued pups die days or weeks after birth.41 For C213G/C213G Tf embryos this either suggests that expression under the endogenous mTF promoter may lead to different spatiotemporal expression patterns associated with different thresholds or that certain signaling functions of C213G Tf are altered, whose importance for development has not been recognized yet. Given the placental hemorrhages observed in homozygous C213G/C213G Tf mice, placental insufficiency might contribute to the partial developmental lethality in the latter and is in line with the low-TF phenotype exhibiting blood pooling in placentas of lowTF embryos at E14.5.13 Reduced survival of low-TF mice was mainly attributed to impaired left ventricular function induced by myocardial hemorrhages and fibrosis.6 Development of cardiac fibrosis was found to be sex-dependent with lower severity in female mice and associated with downregulation of fibrinolytic urokinase plasminogen activator.42 Although there was a sex-dependent effect on the development of cardiac fibrosis, no difference in survival was described between the sexes. No changes in heart rate, systolic, diastolic, or mean arterial blood pressure was found in C213G/C213G Tf mice, suggesting that the myocardial fibrosis did not fully translate into cardiac dysfunction. At the mRNA level, cardiac Tf expression was about 10-times higher in C213G/C213G Tf mice compared to wt TF mice. This increase translated in a similar cardiac protein expression in C213G/C213G Tf mice and wt Tf mice â&#x20AC;&#x201C; in contrast to lung and brain, where protein expression was lower in C213G/C213G Tf mice. This observation is consistent with a compensatory elevation of transcription and haematologica | 2020; 105(10)

may in part explain the unaltered hemodynamic parameters. C213G/C213G Tf males showed lower survival compared to females; however, from autopsies of dead or terminally ill mice it was difficult to define the exact cause of death, particularly when all the three organs brain, heart, and lung showed hemorrhages. We therefore analyzed the incidence of sub-lethal hemorrhages in a uniform cohort of male and female C213G/C213G Tf and wt Tf mice, assuming that the incidence of sub-lethal hemorrhages correlates with the incidence of terminal hemorrhages. In this healthy appearing cohort, we found a considerable degree of cardiac fibrosis in 100% of C213G/C213G Tf mice, and the level of fibrosis did not differ between male and female mice. However, male C213G/C213G Tf mice showed a higher incidence of lung bleedings compared to females as well as a differential pulmonary gene expression pattern. This suggests that in C213G/C213G Tf mice, lung hemorrhages may represent the main reason for terminal disease. To address the question why male C213G/C213G Tf mice are more prone to lung bleedings than female C213G/C213G Tf mice, we investigated possible molecular differences in Tf expression between the sexes or functional differences in bleeding time. As neither of these parameters showed a difference between female and male homozygous C213G/C213G Tf mice, we hypothesized that behavioral differences, in particular aggressive fighting, leads to aggravated pulmonary hemorrhage in homozygous C213G/C213G Tf males. Inter-male aggression is a well known problem in housing of male laboratory mice, which can be modulated by genetic background as well as cage enrichment.43 Cases of exercise induced pulmonary hemorrhage exist in healthy humans44,45 and spontaneous pulmonary hemorrhages are a known complication of anticoagulant therapy.46-49 Single housing indeed reduced the incidence of pulmonary hemorrhages and significantly improved survival of homozygous C213G/C213G Tf male mice with low pulmonary TF activity, suggesting that fully active TF is protective against spontaneous and exercise induced pulmonary hemorrhages. A previous report suggested that fatal lung hemorrhages may also represent a major reason for 2493

S.H.M. Sluka et al. reduced survival of low-TF mice.8 In line with this, 17% of low-TF mice die of spontaneous hemorrhages in lung, brain, and gastrointestinal tract.7,10 In contrast to low-TF mice,9 we did not observe any testis calcification in C213G/C213G Tf male mice. The morphology of the seminiferous tubules of the testis and the general sperm amount in the cauda epididymis seemed normal. Since the corresponding single Cys replacement in human TF is compatible with TF-FVIIa signaling,21 one possible explanation for these differences is preservation of integrin-dependent TF signaling50 in the context of testes physiology. In conclusion, mutation of the allosteric disulfide bond of TF leads to some reduction in protein expression and a pronounced reduction in protein activity of at least 100-fold which goes along with impaired glycosylation and cell surface expression. This impairment results in a

References 1. Drake TA, Morrissey JH, Edgington TS. Selective cellular expression of tissue factor in human tissues. Implications for disorders of hemostasis and thrombosis. Am J Pathol. 1989;134(5):1087-1097. 2. Mackman N, Sawdey MS, Keeton MR, Loskutoff DJ. Murine tissue factor gene expression in vivo. Tissue and cell specificity and regulation by lipopolysaccharide. Am J Pathol. 1993;143(1):76-84. 3. Carmeliet P, Mackman N, Moons L, et al. Role of tissue factor in embryonic blood vessel development. Nature. 1996; 383(6595): 73-75. 4. Bugge TH, Xiao Q, Kombrinck KW, et al. Fatal embryonic bleeding events in mice lacking tissue factor, the cell-associated initiator of blood coagulation. Proc Natl Acad Sci U S A. 1996;93(13):6258-6263. 5. Toomey JR, Kratzer KE, Lasky NM, Stanton JJ, Broze GJ, Jr. Targeted disruption of the murine tissue factor gene results in embryonic lethality. Blood. 1996;88(5):1583-1587. 6. Parry GC, Erlich JH, Carmeliet P, Luther T, Mackman N. Low levels of tissue factor are compatible with development and hemostasis in mice. J Clin Invest. 1998;101(3):560569. 7. Pawlinski R, Fernandes A, Kehrle B, et al. Tissue factor deficiency causes cardiac fibrosis and left ventricular dysfunction. Proc Natl Acad Sci U S A. 2002;99(24):15333-15338. 8. Pedersen B, Holscher T, Sato Y, Pawlinski R, Mackman N. A balance between tissue factor and tissue factor pathway inhibitor is required for embryonic development and hemostasis in adult mice. Blood. 2005; 105(7):2777-2782. 9. Mackman N. Tissue-specific hemostasis in mice. Arterioscler Thromb Vasc Biol. 2005; 25(11):2273-2281. 10. Bode MF, Mackman N. A combined deficiency of tissue factor and PAR-4 is associated with fatal pulmonary hemorrhage in mice. Thromb Res. 2016;146:46-50. 11. Bastarache JA, Sebag SC, Clune JK, et al. Low levels of tissue factor lead to alveolar haemorrhage, potentiating murine acute lung injury and oxidative stress. Thorax. 2012;67(12):1032-1039. 12. Antoniak S, Tatsumi K, Hisada Y, et al. Tissue factor deficiency increases alveolar

2494

13.

14.

15. 16. 17. 18.

19.

20.

21.

22.

23.

24.

25.

bleeding phenotype causing partial developmental lethality and high susceptibility of the lung, brain, and heart to bleedings after birth. Male mice exhibit more frequent and severe lung hemorrhages, which can be reduced by single housing preventing inter-male aggression. Thus, this study emphasizes the importance of the allosteric disulfide for proper TF expression, modification and function in vivo and the importance of functional TF for developmental and post-natal hemostasis and survival. Funding GMC was supported by the German Federal Ministry of Education and Research (Infrafrontier grant 01KX1012 to MHA). WR is supported by the National Heart Lung Blood Institute (HL-60472). SHMS and FCT were supported by the Swiss Heart Foundation.

hemorrhage and death in influenza A virusinfected mice. J Thromb Haemost. 2016; 14(6):1238-1248. Erlich J, Parry GC, Fearns C, et al. Tissue factor is required for uterine hemostasis and maintenance of the placental labyrinth during gestation. Proc Natl Acad Sci U S A. 1999;96(14):8138-8143. Pawlinski R, Pedersen B, Erlich J, Mackman N. Role of tissue factor in haemostasis, thrombosis, angiogenesis and inflammation: lessons from low tissue factor mice. Thromb Haemost. 2004;92(3):444-450. Osterud B. Tissue factor expression in blood cells. Thromb Res. 2010;125 Suppl 1:S31-34. Swystun LL, Liaw PC. The role of leukocytes in thrombosis. Blood. 2016;128(6):753-762. Bach RR. Tissue factor encryption. Arterioscler Thromb Vasc Biol. 2006; 26(3):456-461. Pendurthi UR, Ghosh S, Mandal SK, Rao LV. Tissue factor activation: is disulfide bond switching a regulatory mechanism? Blood. 2007;110(12):3900-3908. Kothari H, Nayak RC, Rao LV, Pendurthi UR. Cystine 186-cystine 209 disulfide bond is not essential for the procoagulant activity of tissue factor or for its de-encryption. Blood. 2010;115(21):4273-4283. Chen VM, Ahamed J, Versteeg HH, Berndt MC, Ruf W, Hogg PJ. Evidence for activation of tissue factor by an allosteric disulfide bond. Biochemistry. 2006;45(39):1202012028. Ahamed J, Versteeg HH, Kerver M, et al. Disulfide isomerization switches tissue factor from coagulation to cell signaling. Proc Natl Acad Sci U S A. 2006;103(38):1393213937. Reinhardt C, von Bruhl ML, Manukyan D, et al. Protein disulfide isomerase acts as an injury response signal that enhances fibrin generation via tissue factor activation. J Clin Invest. 2008;118(3):1110-1122. Langer F, Spath B, Fischer C, et al. Rapid activation of monocyte tissue factor by antithymocyte globulin is dependent on complement and protein disulfide isomerase. Blood. 2013;121(12):2324-2335. Subramaniam S, Jurk K, Hobohm L, et al. Distinct contributions of complement factors to platelet activation and fibrin formation in venous thrombus development. Blood. 2017;129(16):2291-2302. Muller-Calleja N, Ritter S, Hollerbach A,

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

Falter T, Lackner KJ, Ruf W. Complement C5 but not C3 is expendable for tissue factor activation by cofactor-independent antiphospholipid antibodies. Blood Adv. 2018;2(9):979-986. Rehemtulla A, Ruf W, Edgington TS. The integrity of the cysteine 186-cysteine 209 bond of the second disulfide loop of tissue factor is required for binding of factor VII. J Biol Chem. 1991;266(16):10294-10299. van den Hengel LG, Osanto S, Reitsma PH, Versteeg HH. Murine tissue factor coagulant activity is critically dependent on the presence of an intact allosteric disulfide. Haematologica. 2013;98(1):153-158. van den Hengel LG, van den Berg YW, Reitsma PH, Bos MH, Versteeg HH. Evolutionary conservation of the tissue factor disulfide bonds and identification of a possible oxidoreductase binding motif. J Thromb Haemost. 2012;10(1):161-162. Furlan-Freguia C, Marchese P, Gruber A, Ruggeri ZM, Ruf W. P2X7 receptor signaling contributes to tissue factor-dependent thrombosis in mice. J Clin Invest. 2011;121(7):2932-2944. 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. Mu J, Adamson SL. Developmental changes in hemodynamics of uterine artery, uteroand umbilicoplacental, and vitelline circulations in mouse throughout gestation. Am J Physiol Heart Circ Physiol. 2006;291(3): H1421-1428. Pequignot MO, Provost AC, Salle S, et al. The retinal pigment epithelium undergoes massive apoptosis during early differentiation and pigmentation of the optic cup. Mol Vis. 2011;17:989-996. JAXÂŽ Hydrocephalus in laboratory mice.Issue 490, Summer 2003 (available from: https://www.jax.org/news-andinsights/2003/july/hydrocephalus-in-laboratory-mice). 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. Reinhardt C, Bergentall M, Greiner TU, et al. Tissue factor and PAR1 promote microbiota-

haematologica | 2020; 105(10)

Sex specific TF bleeding phenotype

36.

37.

38.

39.

40.

induced intestinal vascular remodelling. Nature. 2012;483(7391):627-631. 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. Connolly AJ, Ishihara H, Kahn ML, Farese RV Jr, Coughlin SR. Role of the thrombin receptor in development and evidence for a second receptor. Nature. 1996;381(6582): 516-519. Cui J, O'Shea KS, Purkayastha A, Saunders TL, Ginsburg D. Fatal haemorrhage and incomplete block to embryogenesis in mice lacking coagulation factor V. Nature. 1996; 384(6604):66-68. Sun WY, Witte DP, Degen JL, et al. Prothrombin deficiency results in embryonic and neonatal lethality in mice. Proc Natl Acad Sci U S A. 1998;95(13):7597-7602. Damiano BP, Cheung WM, Santulli RJ, et al. Cardiovascular responses mediated by protease-activated receptor-2 (PAR-2) and

haematologica | 2020; 105(10)

41.

42.

43.

44.

45.

thrombin receptor (PAR-1) are distinguished in mice deficient in PAR-2 or PAR-1. J Pharmacol Exp Ther. 1999;288(2):671-678. Parry GC, Mackman N. Mouse embryogenesis requires the tissue factor extracellular domain but not the cytoplasmic domain. J Clin Invest. 2000;105(11):1547-1554. Davis DR, Wilson K, Sam MJ, et al. The development of cardiac fibrosis in low tissue factor mice is gender-dependent and is associated with differential regulation of urokinase plasminogen activator. J Mol Cell Cardiol. 2007;42(3):559-571. Van Loo PL, Van Zutphen LF, Baumans V. Male management: coping with aggression problems in male laboratory mice. Lab Anim. 2003;37(4):300-313. Ghio AJ, Ghio C, Bassett M. Exerciseinduced pulmonary hemorrhage after running a marathon. Lung. 2006;184(6):331333. Ko YC, Dai MP, Ou CC. Playing saxophone induced diffuse alveolar hemorrhage: a case

report. Ir J Med Sci. 2010;179(1):137-139. 46. Santalo M, Domingo P, Fontcuberta J, Franco M, Nolla J. Diffuse pulmonary hemorrhage associated with anticoagulant therapy. Eur J Respir Dis. 1986;69(2):114-119. 47. Kok LC, Sugihara J, Druger G. First case report of spontaneous pulmonary hemorrhage following heparin therapy in acute myocardial infarction. Hawaii Med J. 1996;5 5(5):83-84. 48. Sitges M, Villa FP. Massive pulmonary hemorrhage in a patient treated with a platelet glycoprotein IIb/IIIa inhibitor. Int J Cardiol. 1997;62(3):269-271. 49. Finley TN, Aronow A, Cosentino AM, Golde DW. Occult pulmonary hemorrhage in anticoagulated patients. Am Rev Respir Dis. 1975;112(1):23-29. 50. 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.

2495