Université Paris Descartes

Ecole doctorale Bio-Sorbonne Paris Cité Biologie Cellulaire et Moléculaire, Physiologie, Physiopathologie INSERM U970, Centre de recherche cardiovasculaire de Paris, Equipe 1

Physiopathologie des évènements cardiovasculaires chez les malades atteints de syndrome myéloprolifératif Bcr/Abl-négatif

Pathophysiology of cardiovascular events in Bcr/Abl- negative myeloproliferative neoplasms

Par Johanne Poisson

Thèse de doctorat de physiologie et physiopathologie

Dirigée par Pr. Pierre-Emmanuel Rautou

Présentée et soutenue publiquement le 27 Septembre 2018

Devant un jury composé de : Pr. Chloé James Rapporteur Pr. Jonel Trebicka Rapporteur Pr. Caroline Le Van Kim Examinateur Dr. Yacine Boulaftali Examinateur Pr. Pierre-Emmanuel Rautou Directeur de thèse

REMERCIEMENTS

Aux membres de mon jury, c’est un immense honneur que vous me faites, Professeur Chloé James, merci d’avoir pris le temps d’évaluer mon travail en temps que rapporteur. Avoir eu la chance de vous rencontrer et d’échanger avec vous a été un vrai honneur. Professor Jonel Trebicka, I would like to sincerely thank you for accepting to evaluate my work and to be part of the jury for my PhD defense. Professeur Caroline Le Van Kim, je vous suis extrêmement reconnaissante d’avoir accepté de faire partie de mon jury de thèse. Docteur Yacine Boulaftali, merci d’avoir accepté de faire partie de mon jury. C’est un réel plaisir de vous avoir à mes côtés ce jour là et de pouvoir échanger avec vous sur mon travail.

Professeur Pierre-Emmanuel Rautou, alias « PE », difficile de savoir par où commencer pour te remercier. J’ai lu quelque part « que l’être humain possède deux qualités : le pouvoir et le don. Le pouvoir conduit l’homme à la rencontre de son destin ; le don l’oblige à partager avec les autres ce qu’il a de meilleur en lui ». Je ne suis pas sûr que tous les êtres humains possèdent ces qualités mais par contre toi c’est sûr. Ta force et ton énergie incroyable pour aller de l’avant et construire ton avenir est absolument inspirante (et épuisante juste à te regarder faire) et ton incroyable passion à transmettre le meilleur de toi est une chance inouïe pour les gens qui t’entourent. Ce n’est que le début de notre chemin, mais d’ores et déjà, merci.

Aux gériatres, Professeur Olivier Saint-Jean, merci de m’avoir dirigé vers ce centre de recherche pour un master 2, qui s’est tranformé en thèse et m’a ouvert à un nouveau monde passionnant. Professeur Elena Paillaud, Hayat, Julien, Mathilde, Elise, Tanguy, Victoire et les autres, merci.

Docteur Chantal Boulanger, merci de m’avoir accueillie dans votre équipe sans quoi rien n’aurait été possible. Pour vos conseils et votre bienveillance tout au long de mon parcours dans votre équipe, merci.

2 Docteur Alain Tedgui, Merci de m’avoir accueillie dans votre centre de recherche. Il y règne un air familial, bienveillant et extrêmement stimulant intellectuellement, qui est insufflé directement par la manière dont vous dirigez ce centre.

A tous les membres de l’équipe 1, merci pour votre soutien scientifique, technique et émotionnel. Marion, merci pour ton aide sans faille. J’ai énormément appris à tes côtés. Aux étudiants de passage qui m’ont énormément aidé pendant ma thèse, Hortense évidemment, Fatou, Mendel, merci. Xavier, Cécile, Adel, Juliette, Michael, PM, Shruti, Stephan et bien d’autres, merci pour votre bonne humeur.

A tous les membres du 2ième étage et du PARCC, merci pour votre bonne humeur.

Aux membres de la plateforme administrative, merci de réconcilier tous le monde avec le mot « administratif ». Votre gentillesse et votre efficacité n’ont pas d’égal.

A mes amis, merci d’être toujours là…

A mon encadrant de M2, merci de m’avoir formé et de m’avoir donné le goût de la recherche. Merci pour toutes ces discussions scientifico-philosophiques et pour ton soutien lors des bons et des mauvais moments.

A ma famille, merci pour votre soutien sans faille.

3 Quelques proverbes, dictons et autres pensées provenant du « placard du laboratoire » en guise de souvenirs et remerciements Together Everyone Achieves More 1

« A chaque problème sa solution et à chaque solution son problème ». Adel Hammoutene

« Si les résultats ne sont pas à la hauteur de tes espérances, dis toi que le grand chêne aussi, un jour, a été un gland ! ». Le grand sage des pistes de ski…

« Si tu arrives à exploiter les gens et qu’ils t’aiment encore c’est que tu es une bonne chef ». Hortense Davy (Je ne sais pas bien à qui elle fait allusion)

« L’oisiveté est mère de tous les vices ». Pierre-Emmanuel Rautou

« Impossible n’est pas Français (ou Algérien apparemment)». Adel Hammoutene

« 100% des gagnants ont tenté leur chance ». Team 1 (quand on tente de soumettre au NEJM)

« Faites ce que je dis et pas ce que je fais ». Anonyme

« Un beau schéma vaut mieux qu’un long discours ». Pierre-Emmanuel Rautou

« Ce qui ce conçoit bien s’énonce clairement ». Pierre-Emmanuel Rautou

« Errare humanum est, perseverare diabolicum ! ». Pierre-Emmanuel Rautou

« Ce n’est pas le doute, c’est la certitude qui rend fou ! ». Nietzsche

« La confiance n’exclut pas le contrôle ». Pierre-Emmanuel Rautou

« Il vaut mieux viser l’excellence et échouer que la médiocrité et réussir ». Pierre-Emmanuel Rautou

« Je ne râle pas je m’exprime ». Marion Tanguy

« L’avenir appartient aux gens qui se lèvent tôt ». Vincent Poisson

« Medecine is a science of uncertainty and an art of probability ». Osler

« Quelle est la différence entre un tour de force et un tour de con ? ». Pierre-Emmanuel Rautou

« La flemme n’est jamais un bon conseiller ». Pierre-Emmanuel Rautou

« Le (foie) gras c’est la vie ». Marion Tanguy – « Et notre gagne pain ». Adel Hammoutene

Et pour finir : « Johanne a toujours raison ». Adel Hammoutene… No comment

4 I. LIST OF ABBREVIATIONS ...... 6 II. INTRODUCTION ...... 7 A. Myeloproliferative neoplasms ...... 7 1. Definitions ...... 7 a) Polycythaemia Vera ...... 9 b) Essential thrombocythemia ...... 10 c) Pre primary myelofibrosis and primary myelofibrosis ...... 11 2. Pathophysiology ...... 13 a) JAK/STAT signalling ...... 13 b) Mutational landscape ...... 16 3. Non vascular complications ...... 23 a) Secondary myelofibrosis ...... 24 b) Acute myeloid leukaemia ...... 25 4. Treatments ...... 26 a) Polycythaemia Vera ...... 26 b) Essential thrombocythemia ...... 28 c) Pre primary myelofibrosis and primary myelofibrosis ...... 31 B. Myeloproliferative neoplasms and cardiovascular complications ...... 32 1. Risk factors for cardiovascular complications in MPNs ...... 34 a) Driver mutations ...... 34 b) Leukocytes ...... 35 a) Platelets ...... 36 b) Red blood cells ...... 36 2. Venous thrombosis and myeloproliferative neoplasms ...... 37 a) Pathophysiology of venous thrombosis in myeloproliferative neoplasms ...... 37 b) Site specificity in MPNs patients ...... 50 3. Arterial vascular events and myeloproliferative neoplasms ...... 58 a) Atherosclerosis and MPNs ...... 58 b) Beyond atherosclerosis ...... 71 III. Thesis work ...... 83 A. Aims ...... 83 B. JAK2V617F in arterial events ...... 84 1. Article 1: Erythrocyte microvesicles increase arterial contraction in JAK2V617F myeloproliferative neoplasms ...... 84 C. JAK2V617F in splanchnic vein thrombosis ...... 120 1. Endothelial JAK2V617F and Budd-Chiari syndrome ...... 120 a) Background and aims ...... 120 b) Materials and methods ...... 120 c) Article 2: Endothelial JAK2V617F does not enhance liver lesions in mice with Budd-Chiari syndrome ...... 125 2. Calreticulin mutations and splanchnic vein thrombosis ...... 128 a) Article 3: Selective testing for calreticulin gene mutations in patients with splanchnic vein thrombosis: A prospective cohort study ...... 128 IV. DISCUSSION AND PROSPECTS ...... 137 A. JAK2V617F in arterial events ...... 137 B. JAK2V617F in Budd-Chiari syndrome ...... 139 C. CALR and splanchnic vein thrombosis ...... 141 V. CONCLUSION ...... 142 VI. REFERENCES ...... 143 VII. APPENDIX ...... 170 A. Appendix 1: Review: Liver sinusoidal endothelial cells: physiology and role in liver diseases ...... 170 B. Appendix 2: Reply to “Calreticulin mutations and their importance in Budd-Chiari syndrome” ... 187 C. Appendix 3: Curriculum vitae and list of publications ...... 191

I. LIST OF ABBREVIATIONS

AML: Acute myeloid leukaemia PV: Polycythaemia vera BCS: Budd-Chiari syndrome PVT: Portal vein thrombosis BM: Bone marrow RBCs: Red blood cells CALR: Calreticulin SH2: Src homology 2 CML: Chronic myelogenous leukaemia SOCS: Induction of suppressor of cytokine COX: Cyclooxygenase signalling proteins CV: Cardiovascular STAT: Signal transducers and activators CVT: Cerebral venous thrombosis SVT: Splanchnic vein thrombosis EC: Endothelial cells t-PA: tissue plasminogen EPCR: Protein C receptor TAFI: Thrombin activated fibrinolysis EPO: erythropoietine inhibitor EPOR: Erythropoetin receptor TF: Tissue factor ERK: Extracellular signal-regulated kinase TFPI: Tissue factor pathway inhibitor ET: Essential thrombocythemia TM: Thrombomodulin FERM: Four-point-one ezrin radixin moesin TNFα : Tumor necrosis factor alpha G-CSF: Granulocyte colony-stimulating TPO: Thrombopoietin factor receptor TYK: Tyrosine kinases HU: Hydroxyurea u-PA: urokinase HUVECs: Human umbilical vein endothelial vWF: Von Willebrand factor cells WBC: White blood cells IFN-α: Interferon-alpha WHO: World health organization IL: Interleukin WT: Wild-type IPSET: International prognostic score for thrombosis in essential thrombocythemia IPSS: International prognostic scoring system JAK: Janus Kinase JH1: Jak homology 1 MAPK: Mitogen-activated protein kinases MDS: myelodysplastic syndrome MPL: Myeloproliferative leukaemia virus / Thrombopoietin receptor MPNs: Myeloproliferative neoplasms NETs: Neutrophil extracellular traps NO: Nitric oxide Pa: Pascal PAD4: Petidyl-arginine deiminase PAI-1: Type I plasminogen activator inhibitor Ph: Philadelphia PI3K: Phosphatidylinositol-3’-kinase PIAS: Protein inhibitor of activated STAT PMF: Primary myelofibrosis PrePMF: Prefibrotic/early primary PSGL-1: P-selectin glycoprotein ligand-1 PTP: Protein tyrosine phosphatases

II. INTRODUCTION

A. Myeloproliferative neoplasms

1. Definitions

Myeloproliferative neoplasms (MPNs) are clonal hematopoietic diseases characterized by an overproduction of differentiated hematopoietic cells. The first report of

« myeloproliferative disorders » in 1951 described a group of disease without a clear separation between polycythaemia vera, thrombocythemia chronic granulocytic leukaemia, splenic myeloid metaplasia and leuko-erythroblastic anaemia with bone marrow fibrosis [1].

The 2001 world health organization (WHO) classification separated four categories of chronic myeloid neoplasms: 1/Chronic myeloproliferative diseases, including what was called

"classical myeloproliferative diseases" - i.e. chronic myelogenous leukaemia (CML), polycythaemia vera (PV), essential thrombocythemia (ET) and primary myelofibrosis (PMF)-, chronic neutrophilic leukaemia, chronic eosinophilic leukaemia/hyper-eosinophilic syndrome and unclassified chronic myeloproliferative disease -, 2/myelodysplasic syndromes,

3/myeloproliferative/myelodysplasic syndromes, and 4/mast cell disease [2].

In the 2008 WHO classification, the term “myeloproliferative disease” was replaced by

“myeloproliferative neoplasm”. In addition, mast cell disease was included in the MPNs group and myeloid neoplasms associated with hyper-eosinophilia and abnormalities of PDGFRA,

PDGFRB or FGFR1 were isolated in a new category [3].

The elucidation of the molecular mechanisms underlying MPNs started in 1960 with the discovery of the Philadelphia (Ph) chromosome in chronic myelogenous leukaemia [4], later characterized as a reciprocal translocation between the chromosome 9 and 22 and identified as the BCR-ABL1 transcript [5]. This transcript allowed separating chronic myeloproliferative

7 neoplasms group into 2 major entities: Ph positive (i.e. chronic myelogenous leukaemia) and Ph negative chronic myeloproliferative neoplasms (Figure 1).

In 2016, thanks to the discovery of several other molecular mechanisms and to the better characterization of morphological features, particularly in Ph negative MPNs, an updated 2008

WHO classification was published [6]. Clear boundaries between the 3 entities, PV, ET and PMF could not be well established, as a certain continuum in the progression of ET and PV to secondary myelofibrosis is observed. Thus, a new transitional entity was described: the prefibrotic/early primary myeloproliferative (PrePMF), distinct from “true” essential thrombocythemia [6].

My thesis work focuses on Ph negative MPNs, i.e. PV, ET and PrePMF/PMF, thereafter referred as MPNs.

MPNs annual incidence and prevalence widely varies between studies and geographic areas. In western countries, PV, ET and PMF incidence ranges from 0.5 to 3, from 0.4 to 2 and from 0.1 to 1 per 100 000 individuals per year, respectively, while their prevalence ranges from

5 to 57, from 4 to 57 and from 0.5 to 2.7 per 100 000 inhabitants respectively [7,8].

8 Myeloproliferative neoplasms

Philadelphia chromosome + Philadelphia chromosome -

Chronic myeloid leukemia

Atypical Classical

Chronic neutrophilic leukemia Polycythemia vera Chronic eosinophilic leukemia Essential thrombocytosis Hypereosinophilic syndrome MPNs unclassifiable Early (Prefibrotic) and Overt Primary myelofibrosis

Figure 1: Myeloproliferative neoplasms 2016 classification [6].

a) Polycythaemia Vera

PV is the most common MPN. PV was first described by a French physician, Henri

Vaquez in 1892 [9]. In 1903, William Osler, distinguished PV from both relative and secondary polycythaemia [10]. PV is characterized by erythrocytosis, with a progressive increased erythropoiesis, thrombopoiesis and granulopoiesis. Patients also display splenomegaly due to extra-medullar haematopoiesis and can evolve to myelofibrosis and acute leukaemia. The diagnostic criteria of the 2016 update of the WHO classification are presented in Table 1.

9 Table 1. WHO criteria for PV [6] Major criteria 1. Haemoglobin > 16.5 g/dL in men and 16 g/dL in women or, Haematocrit > 49% in men and 48% in women or, increased red cell mass 2. Bone marrow (BM) biopsy showing hyper-cellularity for age with trilineage growth including prominent erythroid, granulocytic, and megakaryocytic proliferation with pleomorphic, mature megakaryocytes 3. Presence of JAK2V617F or JAK2 exon 12 mutation Minor criteria Subnormal serum erythropoietin level Diagnosis of PV requires meeting either all 3 major criteria, or the first 2 major criteria and the minor criterion Note: Criterion number 2 (BM biopsy) may not be required in cases with sustained absolute erythrocytosis: haemoglobin levels > 18.5 g/dL in men (haematocrit, 55.5%) or >16.5 g/dL in women (haematocrit, 49.5%) if major criterion 3 and the minor criterion are present. However, initial myelofibrosis (present in up to 20% of patients) can only be detected by performing a

BM biopsy; this finding may predict a more rapid progression to overt myelofibrosis (post- PV myelofibrosis) [6].

b) Essential thrombocythemia

Essential thrombocythemia was described for the first time in 1934, by Emil Epstein and

Alfred Goedel [11]. It is the most indolent MPN and is characterized by thrombocytosis alone.

However, isolated thrombocytosis may be the first manifestation of PV or PMF/PrePMF. In addition to thrombocytosis, patients can display mild splenomegaly, leucocytosis and can evolve to myelofibrosis and acute leukaemia. 2016 WHO criteria for ET diagnosis are described in Table 2.

Table 2. WHO criteria for ET [6] Major criteria 1. Platelet count ≥450 X 109/L 2. BM biopsy showing proliferation mainly of the megakaryocyte lineage with increased numbers of enlarged, mature megakaryocytes with hyperlobulated nuclei. No significant increase or left shift in neutrophil granulopoiesis or erythropoiesis and very rarely minor (grade 1) increase in reticulin fibers 3. Not meeting WHO criteria for BCR-ABL1+ CML, PV, PMF, myelodysplastic syndrome (MDS) or other myeloid neoplasms 4. Presence of JAK2, CALR or MPL mutation Minor criteria Presence of a clonal marker or absence of evidence for reactive thrombocytosis Diagnosis of ET requires meeting either all 4 major criteria, or the first 3 major criteria and the minor criterion

c) Pre primary myelofibrosis and primary myelofibrosis

PMF was described for the first time in 1879 by Gustav Heuck [12]. It is the least common and most aggressive MPN. PMF manifestations include de novo bone marrow fibrosis, splenomegaly due to extra-medullary haematopoiesis, increase in circulating CD34+ cells, anaemia, variable changes in platelet and leukocyte counts, and constitutional symptoms

(>10% weight loss in 6 month, night sweats and unexplained fever (>37.5°C)). PMF can result in bone marrow failure and transformation to acute leukaemia [13].

PrePMF was a controversial entity [14] that has been distinguished in the 2016 update of the

WHO classification from “true ET” and overt PMF (Tables 3 and 4) [6]. A clear distinction between PrePMF “true ET” and overt PMF is however not always obvious, since there is a continuum between these 3 entities [15,16]. Yet, PrePMF has clearly a worse prognosis than

“true ET” [17–20].

Table 3. WHO criteria for Pre-PMF [6] Major criteria 1. Megakaryocytic proliferation and atypia, without reticulin fibrosis >grade 1*, accompanied by increased age-adjusted BM cellularity, granulocytic proliferation, and often decreased erythropoiesis 2. Not meeting the WHO criteria for BCR-ABL1+ CML, PV, ET, MDS, or other myeloid neoplasms 3. Presence of JAK2, CALR, or MPL mutation or in the absence of these mutations, presence of another clonal marker, or absence of minor reactive BM reticulin fibrosis Minor criteria Presence of at least 1 of the following, confirmed in 2 consecutive determinations: a. Anaemia not attributed to a comorbid condition b. Leucocytosis ≥11 X 109/L c. Palpable splenomegaly d. LDH increased to above upper normal limit of institutional reference range Diagnosis of prePMF requires meeting all 3 major criteria, and at least 1 minor criterion * See table 5

Table 4. WHO criteria for PMF [6] Major criteria 1. Presence of megakaryocytic proliferation and atypia, accompanied by either reticulin and/or collagen fibrosis grades 2 or 3* 2. Not meeting WHO criteria for ET, PV, BCR-ABL1+ CML, PV, ET, MDS, or other myeloid neoplasms 3. Presence of JAK2, CALR, or MPL mutation or in the absence of these mutations, presence of another clonal marker, or absence of reactive myelofibrosis Minor criteria Presence of at least 1 of the following, confirmed in 2 consecutive determinations: a. Anaemia not attributed to a comorbid condition b. Leucocytosis ≥11 X 109/L c. Palpable splenomegaly d. LDH increased to above upper normal limit of institutional reference range e. Leuco-erythroblastosis Diagnosis of PMF requires meeting all 3 major criteria, and at least 1 minor criterion *See table 5

Table 5. Grading of myelofibrosis [6] MF-0 Scattered linear reticulin with no intersections (crossovers) corresponding to normal BM MF-1 Loose network of reticulin with many intersections, especially in perivascular areas MF-2 Diffuse and dense increase in reticulin with extensive intersections, occasionally with focal bundles of thick fibres mostly consistent with collagen, and/or focal osteosclerosis MF-3 Diffuse and dense increase in reticulin with extensive intersections and coarse bundles of thick fibres consistent with collagen, usually associated with osteosclerosis

2. Pathophysiology

a) JAK/STAT signalling

Haematopoiesis is the cumulative result of highly regulated signal transduction cascades that are mediated by cytokines (interleukins, interferons, colony-stimulating factors, thrombopoietin and erythropoietin [21]) and their cognate receptors (composed of at least two single membrane-spanning chains) [22]. However, most hematopoietic cytokine receptors lack a cytoplasmic kinase domain and transmit their signals via a family of cytoplasmic tyrosine kinases (TYK) termed Janus Kinase (JAK), composed of 4 members, JAK1, JAK2, JAK3 et TYK2

[22]. While JAK3 and TYK2 are mainly important for immune responses, JAK1 and JAK2 have broad functions that range from immune response, neural development and haematopoiesis

[22]. Homodimeric receptors such as erythropoetin receptor (EPOR), thrombopoietin receptor

(Myeloproliferative leukemia virus - MPL), granulocyte colony-stimulating factor receptor

(GCSFR) are coupled with JAK2, whereas heteromeric receptors are coupled with JAK1 and

JAK2/TYK2 or JAK3 [23]. The complete Knock-out of Jak1 or Jak2 is lethal in mice. Following cytokine engagement, receptor chains oligomerize and transactivate the two associated JAKs.

Activated JAKs phosphorylate the tyrosine residues in the cytoplasmic part of the cytokine

13 receptors, which allow the selective binding of DNA-binding proteins, the signal transducers and activators (STAT-1, -2, -3, -4, -5a, -5b and -6) family and in a lesser degree other pathway such as mitogen-activated protein kinases (MAPK) and phosphatidylinositol-3’-kinase (PI3K)

[24]. Tyrosine-phosphorylation of STATs allows them to dimerize, translocate to the nucleus and regulate the expression of a broad number of genes involved for example in cell differentiation, proliferation and survival [25]. Several mechanisms negatively regulate JAK-

STAT pathway, such as ubiquitin-mediated receptor internalization, dephosphorylation of tyrosines in the JAK activation loop by protein tyrosine phosphatases (PTP), induction of suppressor of cytokine signalling proteins (SOCS) and protein inhibitor of activated STAT

(PIAS) that inhibit the transcriptional activity of STATs (Figure 2) [24].

Figure 2: JAK-STAT signalling [26]. “Republished with permission of Springer

Nature, from [26] ; permission conveyed through Copyright Clearance Center, Inc. ”

14 Janus Kinases are composed of (Figure 3) [27]:

• A Jak homology 1 (JH1) domain, a tyrosine kinase domain located at the

carboxy terminus of the protein

• A JH2 domain, a pseudo-kinase domain directly adjacent to the JH1 domain,

which prevent JH1 inappropriate kinase activity

• A receptor binding module at the N-terminal of the protein, including an

atypical Src homology 2 (SH2) domain (JH3) and a FERM (four-point-one

ezrin radixin moesin) domain (JH4-JH7)

JAK2

JH3 JH2 JH1 N FERM (JH4-JH7) C Atypical SH2 Pseudokinase kinase

Figure 3: JAK2 structure

JAK2 plays an essential role in the maintenance of hematopoietic stem cells, via thrombopoietin

(TPO) and stem cell factor signal transduction and in myelopoiesis through erythropoietin

(EPO), TPO, GCSF, granulocyte-macrophage colony-stimulating factor, interleukin-3 (IL-3) and

IL-5 [24]. JAK2 has also been shown to act as an endoplasmic reticulum chaperone protein that facilitates EPO and TPO receptors expression at the cell surface [28]. JAK signalling regulates a wide range of cellular function and is also involved in the pathogenesis of non-hematologic diseases including rheumatoid arthritis, inflammatory bowel diseases, cardiac ischemic stress, solid malignancy and cirrhosis [22,29–31].

15 b) Mutational landscape

All MPNs entities arise from a single somatically mutated hematopoietic stem cell that clonally expands and gives rise to virtually all clonal myeloid cells. Such mutations, called driver mutations, are located in JAK2, CALR (Calreticulin) and MPL genes. Additional genetic and/or epigenetic anomalies are also involved in MPNs phenotype and progression [23,32–34].

(1) Driver mutations

Driver mutations have been described as mutually exclusive, but recent data have shown their possible coexistence, such as JAK2V617F with JAK2 exon 12 or JAK2V617F with CALR mutations [35].

JAK2 gene mutations

JAK2 mutations are the most common MPN drivers. Before 2005, the molecular pathophysiology of MPNs was largely elusive. In 2005, a G to T somatic mutation at nucleotide

1849, in exon 14 of JAK2, was described. This mutation results in the substitution of a valine to a phenylalanine at codon 617 (JAK2V617F) in the JH2 pseudo-kinase domain, impairing the JH2 pseudo-kinase domain physiological inhibitory action on JH1 kinase domain and resulting in

JAK2 constitutive activation [36–39] (Figure 4 et 5). In the heterozygous state, receptors coupled to JAK2V617F are still responsive to growth factors. Only with JAK2V617F homozygosity, usually due to 9p uniparental disomy, do these receptors become autonomous with respect to growth factors (Figure 4) [35–37]. It results in an increased activation of STAT 1, 3 and 5,

MAPK and phosphoinositide 3-kinase (PI3K) [35,40–42]. In addition, JAK2V617F may allow escaping from negative regulators, such as the suppressor of cytokine signalling 3 [43]. The exact mechanism how JH2 mutation prevents this inhibition is still unclear.

Figure 4: JAK2V617F [44]. Reproduced with permission from [44], Copyright

Massachusetts Medical Society

JAK2V617F can be found in around 70% of MPNs: 95% in PV and 50% to 60% in ET and

PrePMF/PMF [44]. JAK2V617F activates signalling through EPOR, MPL and G-CSFR. JAK2V617F appears in pluripotent hematopoietic progenitor and is present in all myeloid lineages (Figure

4). JAK2V617F has been more rarely detected in B, NK cells and later in disease progression in T cells [23]. JAK2V617F is supposedly absent in non-haematopoietic cells. However, 2 independent teams described the presence of JAK2V617F in endothelial cells in the liver and the spleen of patients with splanchnic vein thrombosis [45,46] and in endothelial progenitors [47].

Although the reason why the same mutation, JAK2V617F is responsible for several phenotypes is still unclear, some mechanisms have been put forward. First, additional genetic and/or

17 epigenetic changes are involved in the disease presentation (see chapter “other genetic aspects”) [23,32,33]. Second, JAK2V617F mutation often undergoes a transition from heterozygosity to homozygosity due to occurrence of mitotic recombination resulting in copy- neutral loss of heterozygosity along a variable size region on the short arm of the chromosome

9 (9pLOH) [37,48–50]. Third, the variant allele frequency is highly variable, with a continuum between the threshold of detection (around 1%) to 100% (around 25% in ET and > 50% in PV)

[49]. Fourth, JAK2V617F has also been shown to act independently from the JAK-STAT signalling, on chromatin phosphorylation, impairing histone methylation [51,52], but the implication of this effect on chromatin in MPNs pathophysiology is unknown.

An insertion or a deletion in JAK2 Exon 12 (located at the interface between the SH2-JH2 linker region) is detected in approximately 3% of patient with PV, representing the majority of

JAK2V617F negative PV patients. This JAK2 Exon 12 mutations probably alters the interface between JH1 and JH2 domains [53] (Figure 5). Patient carrying JAK2 Exon 12 mutations are younger than those with JAK2V617F [53,54]. Several JAK2 Exon 12 mutations have been described, the most frequent being the N542-E543del (23%), E543-D544del (11%) and F537-

K539delinsL and K539L (10%) [53]. JAK2 Exon 12 is usually not found in ET and PMF patients, but can be detected in PMF secondary to PV [54]. JAK2 Exon 12 mutations are associated with a more benign phenotype than JAK2V617F, ranging from isolated erythrocytosis to a complete PV phenotype [35]. The variant allele frequency can be low and homozygosity can also appear

[55].

18 V617F G to T substitution in exon 14

A JAK2

JH3 JH2 JH1 N FERM (JH4-JH7) C Atypical SH2 Pseudokinase kinase

Insertion/Deletion in exon 12

B Wild type CALR

N domain P domain C domain KDEL

Insertions/deletions in exon 9 C Mutated CALR

N domain P domain C domain Novel C domain

Figure 5: JAK2 and CALR mutations

CALR gene mutations

CALR gene, encoding for calreticulin, is the second most common MPN driver gene. CALR is a multifunctional protein involved in glycoprotein folding and calcium homeostasis in endoplasmic reticulum. CALR phenotype depends on the presence of MPL [56,57]. Indeed,

CALR binds to MPL in the endoplasmic reticulum by interaction of its lectin domain with MPL

N-glycosylation, controls the quality of the proteins, and then dissociates from the receptor, which traffics to the cell surface. The completely glycosylated MPL is expressed at the cell surface where it can bind TPO to be activated [23,35].

In 2013, frame shift mutations in the CALR gene were discovered in patients without JAK2 or

MPL mutations (50%-60% in ET and 75% in PMF) [58,59]. CALR mutations consist in a wide range of insertions or deletions, inducing a +1 (-1+2) frame shift, in exon 9, which removes

19 KDEL, a canonical endoplasmic reticulum retrieval motif important for protein retention, together with a switch from a negatively charged to a positively charged peptide sequence in the CALR C terminal domain (Figure 6). Mutations preserving the original reading frame are not known to be pathogenic [23,35,60–62].

Mutant CALR binds to MPL in the ER by the same domain as the wild-type (WT) CALR, but the new C terminus reinforces the interaction. Thus, CALR protein remains attached to MPL, which traffics to cell surface and is expressed at the membrane in an immature form and attached to

CALR (Figure 6). The immature MPL is activated by the new C terminus (autocrine activation) resulting in a JAK2 dimerization and downstream STAT5 and ERK (extracellular signal- regulated kinase) phosphorylation independently of TPO presence and can only be slightly activated by TPO [23,56,63–65].

The two most frequent mutations are the so-called type 1 mutation, i.e. a 52-bp deletion (del52 responsible for the lost of the WT exon 9 sequence and calcium-binding sites), and the type 2, i.e. a 5-bp insertion (ins5 which is closer to the WT sequence with 50% of conserved negative charges). The other pathogenic mutations are classified as type 1-like or type 2-like based on the conservation of an α helix close to the WT in type 2-like mutations [61]. Type 1 mutations represent 55% of CALR mutations in ET and 75% in PMF, while type 2 represent 35% in ET and

15% in PMF [66].

CALR mutations in hematopoietic stem cells mainly activate MPL, and at a low level the G-CSFR, but not the EPOR, explaining the thrombocytosis associated with CALR mutations [23,64]. CALR mutations give a greater proliferative advantage compared to JAK2 mutations [67]. Indeed,

CALR mutations are usually heterozygous, but can be homozygous as a result of 19 uniparental disomy [68]. Variant allele frequency for CALR is usually high (around 40%) [58,59].

ET driven by CALR mutations affect predominantly men, with a younger age, a higher platelets count and a lower haematocrit level than ET driven by JAK2 or MPL [49,69]. This reflects a

20 higher level of MPL activation induced by CALR mutations than by MPL heterozygous mutations. This elevated MPL activation might account for the high risk of myelofibrosis transformation associated with CALR mutations [70], as confirmed by MPL mouse model in which excessive MPL signalling results in MF [71] and by other mouse models where homozygous MPL increases BM reticulin deposition [72].

Figure 6: CALR mutations [23] “Republished with permission of American

society of hematology, from [23] ; permission conveyed through Copyright Clearance

Center, Inc. ”

MPL gene mutations

MPL mutations, resulting in a truncated form of the thrombopoietin receptor gene, have been associated with MPNs. Activation of MPL by TPO induces JAK2 and TYK2 activation, but only JAK2 is indispensable for cell proliferation [23]. MPL mutations are the least frequent

21 driver mutations in MPNs and occur in ET (around 3%) and PMF (around 5%) [73,74]. The most frequent MPL mutations, on the tryptophan W515, are located at the boundary of the trans-membrane and cytosolic domains of MPL, located in exon 10 (MPLW515WL and K) [75].

MPL mutations are responsible for receptor conformational change, activating JAK2 in the absence of thrombopoietin and increase STAT3, STAT5, ERK and AKT signalling [75,76]. MPL mutations are usually heterozygous, but can be homozygous [72].

(2) Other genetic aspects

The same driver mutations can lead to very different phenotypes in patients and in mice, especially concerning JAK2V617F, implying other associated mechanisms [23,35,77]. Different genetic, epigenetic and environmental factors influence MPN phenotype:

• A specific driver mutation can orientate towards a specific phenotype, such as JAK2

Exon 12 towards erythrocytosis while JAK2V617F is more pleiotropic, and CALR and

MPL towards thrombocytosis. The difference could be due to a differential coupling

of JAK2 proteins with specific cytokine receptors. Indeed, JAK2 Exon12 preferentially

binds EPOR and JAK2V617F MPL [78].

• The loss of heterozygosity, which is possible for all driver mutations, but particularly

for JAK2V617F, is a major determinant of the degree of erythrocytosis both in patients

and in mouse models [33,79,80]. The exact differential mechanism in the JAK-STAT

pathway activation under the homozygous or heterozygous mutation remains

unclear, but it can be hypothesized that homozygosity favours EPOR binding,

because most of homozygous JAK2V617F are found in PV patient [77].

• Germline predispositions can be separated in 2 groups: one with common variants

that predispose to MPNs, and one with rare variants that can be found in familial

MPNs. These germline variants involve sensitivity of megakaryocyte progenitors to

22 TPO, cell senescence, JAK-STAT signalling, myeloid differentiation, DNA damage

repair and epigenetic regulation [77,81–87]

• Other additional somatic mutations have been shown to be present in approximately

one third of patients with MPNs, involving DNA methylation, DNA repair, chromatin

modifications and mRNA splicing, as reviewed elsewhere [23,59,88].

• The order of mutations apparition also seems to play a role in the phenotypic

presentation. For example, in patients carrying JAK2V617F mutation, a group so-called

“JAK2-first patients” is more likely to present PV and to develop thrombosis than the

so-called “TET2-first patients” [89].

• Patient-specific factors, such as age, sex, renal function, EPO level and iron status are

also involved in the clinical presentation of MPNs [77].

In conclusion, the inappropriate activation of JAK-STAT signalling is common to the three main driver mutations and plays an important role in the disease pathogenesis. No genetic cause is found in around 10% of patients with MPNs. There are also clearly other actors involved in

MPNs, but their exact and relative implication remains to be clarified.

3. Non vascular complications

The natural history of MPNs is highly variable and ranges from slow protracted course to a progressive course with transformation to secondary myelofibrosis and/or acute myeloid leukaemia. However, the first cause of mortality of patients with MPNs is represented by vascular complications, which are discussed below (section “myeloproliferative neoplasms and vascular complications”). The median survival in a Mayo Clinic cohort including 826 patients was approximately for the all population 20 years for ET, 14 years for PV and 6 years for PMF and for patients under 60 years old 33, 24 and 15 years respectively [90].

23 a) Secondary myelofibrosis

All MPNs have the propensity to progress towards a late myelofibrotic stage characterized by splenomegaly and decreased circulating cells count due to BM failure. Criteria for the diagnosis of post-PV or post-ET MF published by the international working group for

MPNs research and treatment (IWG-MRT) are detailed in Table 6 and Table 7 respectively [91].

The frequency of secondary MF differs between PV and ET, ranging from 20% 10 years or more after original diagnosis in PV patients, to less than 1% after 10 years and less than 10% after 15 years of follow-up in ET patients [17,92]. The distinction of late PMF from post-PV or post-ET

MF requires disease history, but megakaryocyte morphology can be also helpful. Indeed, in case of post-PV MF, megakaryocytes retain their typical PV morphological features (pleomorphic nuclei with minimal maturation defects and without significant dysmegakaryopoiesis and severe alteration in the nucleus-to-cytoplasm ratio) [93]. The pathophysiology of PMF and post-PV MF are most likely different: post-PV MF results from a slowly evolving clonal disease with the accumulation of many genetic alterations over the time, while PMF shows a propensity to fibre deposition at an early stage [93,94].

Table 6. IWG-MRT criteria for post-PV myelofibrosis [91] Required 1. Documentation of a previous diagnosis of PV as defined by the criteria WHO criteria* 2. Bone marrow fibrosis grade 2-3** Additional a. Anaemia or sustained loss of requirement of either criteria phlebotomy or cytoreductive treatment b. A leucoerythroblastic peripheral blood picture c. Increasing splenomegaly d. Development of ≥ 1 of three constitutional symptoms: >10% weight loss in 6 month, night sweats, unexplained fever (>37.5°C) Diagnosis of post-PV myelofibrosis requires meeting all 2 required criteria, and at least 2 additional criterion * cf Table 1; ** cf Table 5

Table 7. IWG-MRT criteria for post-ET myelofibrosis [91] Required 1. Documentation of a previous diagnosis of ET as defined by the criteria WHO criteria* 2. Bone marrow fibrosis grade 2-3** Additional a. Anaemia and a ≥ 2mg.ml-1 decreased from baseline criteria haemoglobin level b. A leucoerythroblastic peripheral blood picture c. Increasing splenomegaly d. Increased LDH (above reference level) e. Development of ≥ 1 of three constitutional symptoms: >10% weight loss in 6 month, night sweats, unexplained fever (>37.5°C) Diagnosis of post-ET myelofibrosis requires meeting all 2 required criteria, and at least 2 additional criterion * cf Table 2; ** cf Table 5

b) Acute myeloid leukaemia

The incidence of acute myeloid leukaemia (AML) range from 1.5% in patient with ET,

5% in patient with PV to 8%-10% in those with PMF [35,95,96]. Average duration between diagnosis of MPN and AML development is also highly variable, ranging from approximately 30 months to > 80 months, with a shorter duration in PMF than in PV and ET [97,98]. Age and chemotherapy increases the incidence of AML [99,100]. AML is also associated with 9p uniparental disomy, 1q amplification and additional cytogenetic mutations [35]. AML more

25 frequently involves a sub-clone (de novo AML, like in patients without MPNs), but can also involve the founding hematopoietic stem-cell clone [35]. AML diagnosis is based on blasts count (>20% in the bone marrow or in the blood) and is based on the classical FAB (French-

American-British) classification of AML and the more recent WHO criteria focusing on significant cytogenetic and molecular genetic subgroups [6].

4. Treatments

The only curative treatment would be allogenic stem cell transplantation, but this approach remains limited due to the age of the patients, the comorbidities and the high mortality associated with this procedure. Because survival of ET patients is close to-normal and

10-year survival of PV patients is >75% with a 10-year risk of leukemic transformation < 5% and fibrotic transformation < 10% [90], the goal of current treatment in patients with PV and

ET is mainly to prevent cardiovascular complications, which exceed 20% [101]. Accordingly, current treatment strategies are based on thrombotic risk stratification.

a) Polycythaemia Vera

(1) Risk stratification

PV patients with a history of thrombosis and an advanced age are considered at higher risk of thrombosis, with a distinction regarding the treatment between arterial and venous events. In addition, in 2018 treatment algorithm, hypertension and leucocytosis are also taken into account [102].

(2) First line treatment

In all patients, phlebotomy, to keep haematocrit bellow 45%, and low dose aspirin therapy once a day, which may be increased to twice a day in case of hypertension and leucocytosis, are recommended [102]. In high-risk patients, i.e. those aged 60 years or more, or

26 with a history of vascular event, aspirin twice a day is recommended in case of arterial event and systemic anticoagulation in case of venous thrombosis together with cytoreductive drug

[102–104]. The most commonly used cytoreductive agent is hydroxyurea (HU). HU, also known as hydroxycarbamide, is a ribonucleotide reductase inhibitor, which inhibits DNA synthesis and stops cell cycle in phase S, resulting in cell death. HU reduces cardiovascular events, in comparison with phlebotomy alone and lowers leukemic transformation in comparison with historical cohorts treated with chlorambucil [105]. There was a controversy about a potential increased risk of leukemic transformation with HU, but several uncontrolled studies did not confirm this risk [106]. HU only allows clinical and haematological response, but has no influence on molecular aspects. The other option for first line treatment is pegylated interferon-alpha (IFN-α), which is relatively safe and can allow clinical, haematological and molecular response [102,107–110]. However, HU is in most cases the first-line treatment in PV, pegylated IFN-α being used for HU intolerant patient or as second line therapy [102].

(3) Second line treatment

Three drugs can be considered for second line treatment in case of HU intolerance or inefficacy: namely pegylated IFN-α (see previous chapter), busulfan and ruxolitinib. Pegylated

IFN-α and busulfan display a better activity against clonal proliferation and a better haematological and molecular response [102,109,111–113]. In addition, pegylated IFN-α and busulfan being old drugs, the long-term safety is available and acceptable. The latest proposed treatment is Ruxolitinib, a JAK2-JAK1 inhibitor and is recommended for HU resistant or intolerant patients [32,102,103]. The use of ruxolitinib in HU resistant patients is based on 2 large, randomized controlled trials, RESPONSE and RESPONSE-2, demonstrating a higher rate of haematocrit control, of reduction in spleen size, and of improvement in general symptoms as compared with standard therapies. However, ruxolitinib had limited disease-modifying activity, with less than 24% of complete hematologic remission and less than 2% of molecular

27 remission [114–116]. Ruxolitinib studies where not designed to evaluate the impact on cardiovascular events, nor on leukaemia or myelofibrosis free survival. In addition, long term and immunosuppressive consequences of ruxolitinib are still unknown. Future therapies should focus on controlling cardiovascular events, which are the first cause of death in this mostly indolent disease, but because of a lack of clear pathophysiological understanding of these complications, new options remain limited.

b) Essential thrombocythemia

(1) Risk stratification

Life expectancy in patients with ET is close to hat of the general population [90]. Driver mutations do not influence overall survival. Because of the overall good survival of patients with ET, prevention of cardiovascular events is the main goal of the treatment [117]. The two features classically taken in consideration to base therapeutic decisions are age > 60 years old and history of thrombosis [118]. In addition, JAK2-MPL mutations are independent risk factors for thrombosis [69,119]. More specifically, JAK2V617F was identified as a risk factors for cardiovascular events, along with thrombosis history, age > 60 years old, leucocytosis and cardiovascular CV risk factors in a cohort of 891 patient with ET [120]. Extreme thrombocytosis (> 1000 x 109/L) and CALR mutations were associated with a lower risk of thrombosis [120–122]. A recent thrombotic risk factor stratification, the IPSET Score based on

1019 patients [123], validated in an independent cohort of 585 patients with ET [124], defines

4 groups, described in Table 8. The recommended treatment for each group of patients is based on expert opinions and not based on controlled studies.

Table 8. Revised international prognostic score for thrombosis in essential thrombocythemia (IPSET) [123] Very low risk No thrombosis history and ≤ 60 years old JAK2 wild type Low risk No thrombosis history and ≤ 60 years old With JAK2 mutation Intermediate No thrombosis history and > 60 years old risk JAK2 wild type High risk Thrombosis history or > 60 years old With JAK2 mutation CALR mutations does not modify this score [121]

(2) First line treatment

In “very low risk” patients, a simple surveillance is recommended. Low-dose aspirin is prescribed in patients with cardiovascular risk factors, without extreme thrombocytosis or acquired von Willebrand syndrome. In case of extreme thrombocytosis with symptoms or bleeding complications, HU is recommended to decrease platelet count [118]. For “low risk” patients it is now recommended to treat with low dose aspirin, once or twice a day depending on the presence of cardiovascular risk factors and with the previously mentioned precaution regarding extreme thrombocytosis [118,125]. For “intermediate risk” patients, HU can be discussed as a first line treatment, but is not mandatory [123]. The recommendation for aspirin treatment is the same as in “low-risk” patients. For “high-risk” patients, HU is recommended as first line treatment. Indeed, HU decreases significantly cardiovascular events in this population.

It should however be noted that in the study basing this recommendation, a different definition of high risk patients was used and the goal was to decrease platelets count below 600x109/L

[126]. We now know that extreme thrombocytosis is not associated with an increase in cardiovascular risk [120,127]. In addition, aspirin twice daily in case of arterial thrombosis history is recommended. While in case of venous thrombosis history systemic anticoagulation,

29 with or without aspirin once a day in case of associated cardiovascular risk factor, is now recommended [104,118,128,129].

(3) Second line treatment

In case of HU intolerance or resistance [130], several treatments can be considered:

The first one recommended, which is yet off label in France, is pegylated IFN-α, especially in young patients and pregnant women [117,118]. Indeed, pegylated IFN-α in “high risk” patients has been shown to be relatively safe with good clinical and haematological response and few molecular remissions, especially in the presence of CALR mutations [131–133].

Anagrelide has been evaluated in first line treatment for ET. Even though it was not inferior to

HU in terms of haematological efficacy and thrombotic complications prevention [134], there was an increased risk of bleeding complications and myelofibrosis [135,136]. Thus, anagrelide should be considered as a second line treatment.

Busulfan can be also used, with good haematological response, but with only few molecular responses. There was a previous concern about the potential risk of leukaemic transformation with Busulfan. However, a large international study including more than 1500 patients did not confirm this risk [111–113,137].

Ruxolitinib was tested in a non comparative phase I/II trial including HU resistant ET patients where it decreased platelet count, WBC count, ET-related symptoms and achieved molecular response in approximately 60% of patients after 312 weeks of treatment [138]. The impact of ruxolitinib on thrombotic complications in ET patients is however still unclear [32,118,139]. A randomized controlled trial compared ruxolitinib with best available therapy (HU, anagrelide and interferon) in “high risk” ET patients intolerant or resistant to HU and showed greater amelioration in ET-related symptoms with ruxolitinib, but no significant difference in term of haematological response, cardiovascular events and leukemic transformation [140].

30 Ruxolitinib is currently tested as second line therapy in patients with high-risk ET in 2 trials: the RESET 272 study (NCT03123588) in the United States comparing ruxolitinib with anagrelide, and a French study (NCT02962388) comparing ruxolitinib with anagrelide or IFN-α

[139]. The RUXO-BEAT trial (NCT02577926) in Germany is comparing ruxolitinib with best available therapy in patients with high-risk ET who may be treatment-naive or previously treated [139]. The remaining question is the place of a high cost new drug in an indolent disease such as ET, with possible long-term infectious complications. The real challenge in ET treatment is the adequate distinction of “true” ET from pre-PMF to decide the appropriate and safer treatment and the prevention of cardiovascular complications.

c) Pre primary myelofibrosis and primary myelofibrosis

The only curative treatment for PMF is allogenic stem cell transplantation. Other treatments remain palliative to control anaemia, splenomegaly and general symptoms.

(1) Risk stratification

Unlike PV and ET, risk stratification in PMF has been develop to predict death from any cause, using the IPSS (International prognostic scoring system) at the time of diagnosis or during the course of the disease (Table 9) [17,117,141,142].

Table 9. International prognostic scoring system for primary myelofibrosis [141,142] 1. Age > 65 years 2. Constitutional symptoms 3. Haemoglobin < 10 g/dL 4. (White blood cells) WBC count > 25x109/L 5. Blood blast ≥ 1% The presence of zero (low risk), one (intermediate risk-1), two (intermediate risk-2), or three (high risk) of these five variables defines four risk groups.

(2) Allogenic stem cell transplantation

In “intermediate risk-2” and “high risk” patients with PMF and depending on the patient performance status, age, comorbidities and donor availability, allogenic stem cell

31 transplantation must be considered, because it is the only available curative treatment.

Reduced intensity conditioning regiment can be proposed but has not been evaluated against standard conditioning [117,142–144].

(3) Other available treatments

For patients not eligible to allogenic stem cell transplantation, symptomatic treatments and ruxolitinib are recommended.

Ruxolitinib is the first and only approved agent for PMF since 2012 [145,146]. Only

“intermediate risk-2” and “”high risk patients or “low risk” and “intermediate risk-1” with troublesome symptoms and/or splenomegaly are eligible for Ruxolitinib. Indeed, Ruxolitinib significantly improves splenomegaly and symptoms, but molecular response and improved bone marrow fibrosis and survival remain moderate [147–149]. Other JAK inhibitors are being tested in clinical trial for PMF patients, but none have been approved yet and several have been withdrawn due to toxicity [32,117].

Other symptomatic treatments include HU for splenomegaly if ruxolitinib is not available, transfusion, EPO, corticosteroids, immunomodulating drugs (thalidomide, lenalidomide), danazol and splenectomy for anaemia management [117].

B. Myeloproliferative neoplasms and cardiovascular complications

The prevalence of major thrombosis at the time of diagnosis in ET and PV patients, range from 10% to 29% and 34% to 39% respectively, and at follow-up from 8% to 31% in both ET and PV, the most frequent events being myocardial infarction, stroke and venous thrombosis

[150–155]. In PV, cardiovascular mortality account for 41% of all deaths (15% myocardial infarction, 8% congestive heart failure and 8% pulmonary embolism), which represent the first cause of death in this patients [150]. Arterial events are the most common one, indeed it represent 60-70% of cardiovascular events in patients with MPNs [152–154,156].

32 According to Virchow’s triad several factors contributes to thrombosis pathophysiology: vascular wall abnormalities (endothelial dysfunction or breach), blood components and blood flow. Indeed each of them participates to vascular homeostasis and pathology. Arterial and venous thrombosis mechanisms are not exactly similar, the actors are the same but their roles are different and some molecular mechanisms are vascular bed type-specific. However, the classical view of separates pathophysiology between arterial thrombi involving platelets and inflammation and venous thrombi fibrin and red blood cells (RBCs) is now being challenged

[157,158]. Indeed several lines of evidence suggest that arterial and venous thrombosis have some common pathophysiology: 1/ The existing risk of venous thrombosis in patient with arterial pathology and vice versa 2/ The proof of inflammation and platelets activation involved in venous thrombosis and fibrin in arterial thrombosis 3/ The endothelial dysfunction being key in both settings 4/ The anticoagulant therapy efficiency in both settings [158,159]. Still some specific factors remain different between arterial and venous thrombosis, mainly the blood flow condition and the presence of atherosclerosis in arterial beds. In most MPNs studies and reviews, thrombosis is used as a generic term. However, this term usually encloses all cardiovascular events resulting in a reduction or cessation of blood flow to a tissue, ranging from arterial ischemia with or without underlying atherosclerosis and thrombosis, to venous thrombosis from the limb to the liver. For the clarity and to underline specific factors I will focus on 1/risk factors for cardiovascular events in the context of MPNs with specific factors such as genetic abnormalities, 2/ the venous thrombosis, generalities and site specificity in

MPNs context and 3/ the arterial vascular events in the context of atherosclerosis and beyond it in patients with MPNs.

33 1. Risk factors for cardiovascular complications in MPNs

a) Driver mutations

(1) JAK2 mutations

JAK2V617F is associated with an increased risk of arterial and venous events in ET and PV patients. Indeed, one retrospective study in ET patients, showed an increased risk of CV events associated with JAK2V617F for arterial and venous events (RR 1.41 in general patients and 1.87 in patients younger than 60 years), but more specifically for arterial events, splanchnic vein thrombosis and cerebral venous thrombosis and not for common venous site [160].

In another retrospective series [161] of 1282 patients, JAKWT ET patients had an decreased risk of CV events compare to JAK2V617F ET and PV patients. The five years follow-up showed an occurrence of CV events in 9% of JAKWT ET, 11 % in JAK2V617F ET and 14% in JAK2V617F PV patients [161]. At diagnosis, leukocytes and not JAK2V617F were independently associated with an increased risk of CV events in PV patients, while after 5 years from diagnosis JAK2V617F allele burden became the only independent factor associated with thrombotic events. Indeed leukocytes decreased in follow-up due to HU treatment. Interestingly, the frequency of CV events progressively increased according to the amount of mutated allele burden both in ET and in PV patients [161].

Several meta analysis including more than 2000 patients, consistently showed that JAK2V617F ET patients have a two fold higher risk of developing arterial and venous events than JAK2WT ET patients [162–164].

In patients with pre-PMF or PMF, JAK2V617F was not associated with CV events [153,163]

(2) CALR mutation

ET patients with CALR mutations have a decreased risk of CV events, but are also younger, with higher platelets count and lower haematocrit level than ET patients with

34 JAK2V617F mutation, which are confusing parameter that could account for the decreased thrombotic risk [49,69,165]. One meta analysis of 435 CALR mutated patients and 1116

JAK2V617F patients with PMF confirmed that CALR mutated patients display a lower risk of CV events compare to JAK2V617F patients [166].

In conclusion, JAK2V617F is associated with an increased risk of arterial and venous cardiovascular events in ET and PV, but not in PMF patients. However, confusing factors such as the allele burden, blood cells count and the treatment response, are also implicated in this association. CALR mutation is associated with a lower risk of CV events than JAK2V617F patients.

b) Leukocytes

In PV, ET and PMF leucocytosis has been consistently associated with cardiovascular events, especially arterial events [153,167–171]. In addition, leucocytosis is associated with worse survival in ET, PV and PMF patients, reflecting a more active and aggressive disease

[137,172], but there is still no clear evidence that lowering white blood cells might have an effect on CV risk even if it is recommended [173,174]. One prospective study found a lower thrombosis risk with lowering leukocytes secondary to HU treatment, however HU decreases also other blood cells such as red blood cells and acts on disease activity [175].

The time point of when leukocytes count has the highest predictive value for thrombosis and the cut off are still unknown [137]. Indeed, in ET patients, some studies found that the count at diagnosis was a significant risk factor and not at follow-up usually under treatment [176], while some other found the opposite [177]. In addition, leucocytosis seems to have a higher impact in low risk patients (young, without JAK2V617F mutation) than high risk one [168,178]. Suggesting that several mechanisms could be involved in cardiovascular events related to MPNs depending of the type of disease and the evolution.

35 a) Platelets

Most studies did not find an association between increased platelets count and thrombotic risk [167,169,178–180] or survival [137,174]. The matter of the cut-off remains central. Indeed, beyond a certain cut-off (>1000 to 1500 G/L), extreme thrombocytosis is associated with less arterial events and more bleeding complications, in particular in association with anti-platelets therapy [181–184]. The issue is that acquired Von Willebrand disease, with potential bleeding complications, can be also seen without extreme thrombocytosis [118,185]. An other confounding factor is that highest platelets count correlates with less acute leukaemia and myelofibrosis transformation because thrombocytosis in contrary to leukocytosis, is a sign of less aggressive disease [180].

b) Red blood cells

Increased haematocrit with hyper-viscosity syndrome has been linked to CV events for a long time, particularly arterial one [186–188]. A randomized trial designed to evaluate the intensity of treatment needed in PV patients has shown that a target haematocrit less than

45%, decreased significantly CV death and major CV events in comparison with a target between 45 and 50 % [189]. However the ECLAP study, a prospective observational study, failed to prove that an haematocrit level > 50% in follow-up was associated with CV events

[180]. Two important points have to be taken into consideration: (a) the fact that in the first study treatment not only decreased haematocrit level but also leucocyte count and (b) the fact that the differences observed could be secondary to treatment intensity and disease severity.

Retrospective studies also showed contradictory results, with the issue of the cut-off remaining and the differences in treatment intensity used to achieved the recommended goal of an haematocrit < 45% [153,177,178].

36 In conclusion, the quantitative analysis of blood cells count has been debated wildly in the literature, but enclose major artefacts. Blood cell counts are probably a mirror of disease severity and treatment efficacy, but no conclusion can be made regarding CV events pathophysiology.

2. Venous thrombosis and myeloproliferative neoplasms

a) Pathophysiology of venous thrombosis in myeloproliferative

neoplasms

Blood flow is permanently in contact with the endothelium and remains fluid. Indeed, the luminal surface of quiescent endothelial cells is anticoagulant and non-thrombogenic. By contrast, macromolecules of the basal lamina are strongly thrombogenic. In case of endothelial activation or injury, with an exposition of the sub-endothelial matrix, primary haemostasis results in platelets adherence and initiation of the coagulation cascade leading to fibrin formation (secondary haemostasis) to stop the bleeding, before mechanisms of tissue repair are put in place in parallel of fibrinolysis.

(1) Activation of endothelial cells (EC) (Figure 7)

The endothelium is a monolayer of ECs, which constitutes the inner cellular lining of vessels (arterial and venous vessels, capillaries and lymphatic system). It was first thought to be a physical barrier containing blood inside vessels, but its pleiotropic function in vascular homeostasis and organ pathophysiology is now recognized [190].

The first step of the haemostatic process is the loss of the endothelial anticoagulant function, by activation or injury. The endothelium has a double pivotal role in this process: 1/ promoting the thrombotic process at the site of injury by expression of adhesion molecules (E-

Selectin and P-Selectin) and Von willebrand factor (vWF) and 2/ limiting the thrombotic process by healthy endothelium surrounding the site of thrombosis, by releasing prostacyclin

37 and nitric oxide (NO) [191]. vWF is a multimeric protein, which is mainly synthesized by endothelial cells and is stored within the Weibel-Palade bodies [192,193]. vWF plays a major role in the initial platelets recruitment, particularly when the shear rate is high [191,192]. vWF is also associated with collagen VI in the sub-endothelium and acts as a bridge between matrix and platelet via the GPIb / IX / V glycoprotein [194]. Damaged or activated ECs release in the circulation specific soluble markers including selectins, thrombomodulin, and vWF [195], which can be used as index of endothelial damage in cardiovascular disorders [196,197].

Several indirect observations argue in favour of EC activation and injury in MPNs patients. First, soluble E-Selectin (CD62) [198–201], and thrombomodulin (TM) levels are increased in the blood of ET and PV patients [202,203]. However their direct implications in the pathophysiology of CV events in MPNs are still unknown. vWF levels have been consistently found increased in MPNs patients [203–205]. Moreover, MPNs patients with past history of thrombosis have a higher level of vWF than those without it [204]. Circulating endothelial cells,

CD146+, that are known markers of vascular injury [206] are also increased in the blood of

MPNs patients compare to healthy controls [201,207–209] and in MPNs patients with a past history of thrombosis compare to those without it [210]. In addition, one study reported in

2009 the existence of JAK2V617F in hepatic veins endothelial cells from 2 Budd-Chiari syndrome patients [45]. Subsequently, JAK2V617F was also detected in splenic endothelial cells from patients with myelofibrosis [46]. Later, JAK2V617F was found in circulating endothelial progenitor cells (ECFC) in 5 out of 17 JAK2V617F patients [47], a finding confirmed by an independent group [211,212]. Interestingly, patients harbouring JAK2V617F in ECFC were those with a history of thrombosis (splanchnic vein thrombosis, deep vein thrombosis or stroke)

[47,211].

38 In conclusion, indirect evidence suggests that ECs in MPNs patients are activated.

However, their direct implication in the pathophysiology of venous thrombosis in MPN has been only recently highlighted. Indeed the group of C. James [213], demonstrated that JAK2V617F in endothelial cells induces the exposure of P-selectin at the surface of endothelial cells, which increased neutrophils and mononuclear cells adhesion in vitro. Interestingly, in vivo, small concentrations of tumor necrosis factor alpha (TNFα) were required to uncover this effect.

Importantly, mice expressing JAK2V617F in ECs (Pdgfb-icreERT2+/-) developed more thrombi in the lungs. The only other team who looked previously at the implication of JAK2V617F in ECs, used an arterial FeCl3 assay, which induces ECs damage, likely masking changes in endothelial phenotype due to the mutation [214].

In conclusion, JAK2V617F in EC seems to favour endothelial activation and venous thrombosis, as shown by C. James team.

(2) Circulating blood cells recruitment (Figure 7)

Platelets recruitment and activation

After endothelial activation or injury, circulating platelets adhere and begin the haemostatic process. This platelet recruitment depends on several integrins present on their surface that mediate adhesion to the sub-endothelium (collagen, fibronectin, laminin).

Activated platelets undergo morphological changes and degranulation which release ADP, ATP, thromboxane A2, serotonin, vWF and fibrinogen, activating neighbouring platelets and amplifying the aggregation process [192]. Activated platelets express at their surface and release in the circulation the adhesive molecule, P-Selectin [192].

In MPNs patients, platelets P-Selectin expression and soluble P-Selectin level are increased compared to healthy control in basal condition and after stimulation with several agonists

[200,215–221]. In addition the level of soluble P-Selectin is higher in patients with past-history

39 of thrombosis than without [200,215,217,222]. This increased platelets activation is supported by the elevated circulating platelet aggregates in MPNs patients compared to healthy controls

[217–219,221,223,224], particularly in the presence of JAK2V617F when the allele burden is high

[225].

Regarding platelets integrins expression, Jensen et al [215], found reduced levels of CD41

(GPIIbIIIa) and CD42b (GP1b) but increased level of CD36 (GPIV) compared to healthy controls in basal condition. Interestingly, after stimulation there was less increased expression of surface integrins than in healthy controls [215]. A lower expression of platelets CD41 and

CD42b was confirmed in ET patients compared to healthy controls in basal and stimulated conditions [220]. The decreased functionality is in line with abnormal aggregation described in several studies [182,215,220,226–228]. Platelet function in MPNs remain poorly understood and several factors have to be taken into consideration: 1/ Platelets aggregation function tests are highly variable even in healthy individuals 2/ Shedding adhesion molecules mirror platelets activation but also correlates with decreased platelets expression of integrins (GPIV and GP1b) and platelets adhesiveness as a negative feed-back [229,230] and 3/ There is different population and therapy (specific for MPNs and anti-platelets) in these studies.

To overcome these limitations inherent to patients’ sample analyses, murine models of MPNs have been used. Hobbs et al used heterozygous Jak2 Floxed/1 Mx1Cre1+ mice as a ET model and found that platelets from Jak2V617F knock-in mice had an increased response to agonists that translated into increased aggregation in vitro and a decrease tail bleeding time, independently of the platelet count [231]. Contradictory data have been found in a PV-like disease model (irradiated WT transplanted with VavCre/Jak2V617F KI bone marrow), which showed a decreased platelets activation (decreased GPVI) and responsiveness to different

40 agonists. Time to occlusion after FeCl3 treatment was surprisingly shorter, but formed thrombi were unstable in Jak2V617F mice [232].

The role of mutated platelets has been also studied in a Pf4-Cre/FF1 mice, with Jak2V617F only expressed in megakaryocytes (increased CFU-MK, but no increased circulating blood cells) and platelets aggregation in response to different agonists was similar to WT [214]. However, Pf4-

Cre activity is not megakaryocyte lineage-specific but extends to other myeloid and lymphoid lineages [233,234]. In an ET-like model (Tie2-Cre/FF1 mice, with Jak2 mutation in myeloid and endothelial cells) the tail bleeding time was increased in comparison of control mice. However, when platelets number was normalized in this model there was no differences anymore [214].

This study suggests that it is the number of platelets and not the mutation that is responsible for this abnormal haemostasis. Strassel et al compared a murine model of thrombocytosis without Jak2 mutation (Yall;Mpl with misbalance between thrombopoietin and its receptor) and thrombocytosis secondary to Jak2V617F mutation in a context of ET-like disease (irradiated

WT transplanted with VavCre/Jak2V617F KI bone marrow) and PV-like disease (irradiated WT transplanted with MxCre;FF1/Jak2V617F KI bone marrow) [235]. Interestingly, they used different thrombotic challenges: 1/ a microvascular one (collagen-adrenaline injection) and found an increased mortality in the 3 models compared to WT; 2/ an arterial thrombosis assay with FeCl3 injury, which was not modified in the 3 group compare to WT; and 3/a vena cava stasis model (complete ligation), where clots were similar in term of size in the 3 groups (with a higher platelet to fibrin ratio) compared to WT. In term of haemorrhagic tendency (tail bleeding assay), Yall;Mpl and VavCre;FF1 displayed an increased bleeding time with lower proportion of the more reactive high-molecular-weight forms of vWF in their plasma but without clear defects in platelet activation, which were normal or only weakly decreased. This study highlights the fact that it is more the number of platelets that matters than the mutational status itself on platelets function, which is concordant with one human study that showed that

41 soluble P-Selectin expression was similar between MPNs patients and secondary thrombocytosis patients, both higher than healthy controls [236].

The important discrepancies between murine studies are probably multifactorial, 1/different strains and phenotype, 2/ different platelets isolation procedures for in vitro aggregation and activation which are known to alter the results [237], 3/ different non physiological thrombotic models and 4/different thrombocytosis level which are known to affect haemostasis.

In conclusion, platelets function in MPNs is still unclear, but there is no evident prothrombotic phenotype.

Neutrophils recruitment

Activated platelets potentiate several neutrophil functions, and neutrophils, in turn, enhance platelet adhesiveness and aggregation. P-selectin appears to be a key endothelial cell receptor that captures circulating leukocytes expressing P-selectin glycoprotein ligand-1

(PSGL-1) [238]. In the absence of endothelial injury or activation, the initiation of thrombosis can also be supported by platelets-neutrophils interactions [239].

Leucocytes activation in MPNs patients is supported by several lines of evidence, namely increased level of adhesive molecules, such as L-selectin [198], increased neutrophil proteases level (elastase and myeloperoxidase) [203,240] and increased number of platelets-neutrophil aggregates [241]. Platelets from ET patients contribute to this activation since they do not inhibit anymore neutrophils activation [242]. Activated neutrophils release superoxide anion promoting in return platelets aggregation and adhesiveness [242]. Furthermore an increased expression of Mac1 (also known as surface receptor integrin CD18/CD11b) by leukocytes could play a role in favouring platelets’ activation in MPNs patients [243] and CD18 hypermethylation has been shown to be correlated with a higher risk of thrombotic complications in PMF patients [244].

However, leukocytes activation analysis according to the JAK2V617F mutation have shown that

42 some markers are altered in the presence of JAK2V617F (i.e., CD14 and leukocyte alkaline phosphatase), whereas others are not dependant of JAK2V617F (i.e., CD11b and plasma elastase)

[220,240]. In conclusion, leucocytes seem to be activated in MPNs patients. However mechanisms dependent and independent of the JAK2V617F mutation coexist and probably contribute to the pathophysiology of cardiovascular events in the context of MPNs, but their direct role has not yet been proven.

Another mechanism by which neutrophils can participate to thrombosis is by the formation of neutrophil extracellular traps (NETs). Indeed, in the presence of pathogens, neutrophils can release chromatin and their granular components, creating fibrous traps with anti-microbial properties [245]. In the absence of pathogens, activated neutrophils can also produce NETs, which are involved in venous thrombosis by several mechanisms: 1/NETs can bind platelets and red blood cells in the thrombi, 2/ the released neutrophil elastase inactivates tissue factor pathway inhibitor, thus resulting in a pro-coagulant activity and 3/ NETs can activate endothelial cells and platelets and increase vWF release [246]. Interestingly, NETs are increased in a CML mouse model [247]. A recent study showed also an increased NETs formation in vitro both from JAK2V617F human and mouse neutrophils in response to ionomycin stimulation [248]. To support the increased NETosis in MPNs patients and its role in venous thrombosis, experiments carried out in Jak2V617F mice (Jak2V617F/WT;Vav-Cre mice) found that increased NET formation was associated with increased venous thrombosis (lung thrombosis and inferior vena cava stenosis), which was abrogated by ruxolitinib administration [248]. In addition petidyl-arginine deiminase (PAD4) is increased in leucocytes from Jak2V617F mouse model and is required for NETs formation and thrombosis in vivo. However, opposite results were obtained previously by another group reporting an increased neutrophil activation but surprisingly no differences in NETs formation in response to TNFα and IL-8 and impaired NETs formation in response to PMA stimulation in patients compare to healthy controls [249]. Such

43 apparently discrepant results may be the result of the different stimuli used.

In conclusion, the leucocytes activation in patients and in MPNs mouse model, in addition to the demonstration of the implication of NETs in venous thrombosis in vivo by

Wolach et al, clearly support the fact that NETosis is implicated in venous thrombosis in the context of MPNs

RBC and venous thrombosis

RBCs are the most abundant blood cells, however their role in vascular homeostasis and cardiovascular events has been overlooked until major works on sickle cells disease [250]. It is now proven that RBCs play an important role in vascular homeostasis and in the haemostatic process in both arterial and venous system. Indeed, RBCs can directly adhere to the activated endothelium or subendothelial matrix and to other blood cells, including neutrophils and platelets [250]. In addition, haemoglobin released from haemolysed or damaged RBCs can scavenge NO, which modifies vascular tone and favours platelets adhesion and activation [188].

Under in vivo arterial flow conditions, axial migration of RBCs occurs with displacement of platelets to the mural plasmatic zone, exposing them to maximal vessel wall shearing forces.

With increased haematocrit, the width of the plasmatic zone becomes narrower, allowing greater platelet-endothelial cell as well as platelet–platelet interactions [186]. Finally, RBCs within thrombi are compressed into shapes (“polyhedrocytes”) that permit tight packing and reduce clot permeability, which may delay access of thrombolytic enzymes to the clot and consequently delay thrombus resolution. This mechanism was thought to be involved only in venous haemostasis, however more recent data suggest that RBCs are also present in arterial thrombi [188,251]. In MPNs patients, we can speculate that the higher haematocrit will promote platelets activation. However it has never been proven in this context.

Unlike sickle cells disease, MPNs are not considered as a haemolytic pathology. Indeed,

44 haptoglobin is catabolized more rapidly in PV patients but without overt hemolysis [252] and a more recent study found a low level of haptoglobin only in approximately 30% of PMF patients, which was associated with high JAK2V617F allele burden and Ruxolitinib use without any sign of acute hemolysis [253]. Nonetheless, even if there is no obvious clinical evidence of intravascular haemolysis in MPNs patients, Rusak et al have shown the presence of an increased level of free haemoglobin in PV patients in comparison of control [254], but its implication in cardiovascular events in MPNs has not been tested.

Based on previous work on sickle cells disease, Wautier et al [255] showed that RBCs from PV patients adhere more to HUVECs under static and venous flow conditions (post-capillary venule shear equivalence around 0.1 Pascal (Pa)) than RBC from healthy volunteers. This difference disappeared in higher shear stress conditions, equivalent to arteries (around 1

Pascal (Pa)). This increased endothelial adhesion was mediated by erythroid Lu/BCAM

(receptor for laminin α5 chain-a major component of extracellular matrix) and was potentiated after Lu/BCAM phosphorylation by PKA. Phosphorylation of Lu/BCAM is increased in RBCs from PV patients, and is mediated by JAK2V617F mutation, through Rap1/Akt signalling pathway, independently of the classical EPOR/JAK2/PI3K/Akt pathway [256]. These results prove that

JAK2V617F is still active in mature RBCs. However, these phenotypic modifications of RBCs have not yet been directly linked to cardiovascular events in vivo.

Figure 7: Venous thrombosis and MPNs

(3) Formation of fibrin-rich thrombus (Figure 8)

The second step after platelets activation and adhesion is the activation of the coagulation cascade to eventually produce fibrin.

The blood coagulation cascade can be divided into three parts: the extrinsic (TF, FVIIa), intrinsic (FXIIa, FXIa, FIXa), and common (FXa and thrombin) pathways (reviewed in ref [238]).

Tissue factor (TF) is responsible for the initiation of the coagulation cascade. In healthy blood vessels, TF is located in the vessel wall underlying the inactivated endothelium. In disease conditions, TF present in the subendothelial is exposed to circulating blood following endothelial damage; TF is also expressed by activated endothelium and leucocytes and by microvesicles (MVs), i.e. small membrane vesicles released from activated cells [257,258].

Coagulation cascade activation leads to production of thrombin (FIIa), which cleaves fibrinogen into fibrin, allowing fibrin polymerization [238]. Fibrin network combined with the aggregated activated platelets, bind leukocytes and RBC and forms the blood clot [192]. Endothelium actively participates in the regulation of blood coagulation by its anticoagulant properties.

Endothelium produces tissue factor pathway inhibitor (TFPI), an inhibitor of TF-factor VIIa complex [259] and thrombomodulin (TM), which enhances the anticoagulant activity of the protein C (inhibitor of factors V and VIII) in association with its specific cofactor, protein S

[192]. Endothelium also expresses at its surface protein C receptor (EPCR), which increases the activity of the protein C and heparin-like sulphated glycosaminoglycan that binds and activates anti-thrombin [192].

Patients with ET and PV display features suggesting an activation of coagulation cascade, since they have increased levels of prothrombin fragment, thrombin-antithrombin complexes and D-Dimers [220,260], but also increased thrombin generation using thrombin

47 generation assays and particularly in patients with JAK2V617F mutation [219,241,260–263]. The following changes reported in patients with MPN could account for this activation of the coagulation: (a) an acquired activated protein C resistance, even higher in patients with a past history of thrombosis [219,225,262]; (b) increased levels of TF in the plasma and on MVs, which was higher in case of JAK2V617F and correlated with allele burden [219,225,264]; (c) TM resistance, which was influenced by JAK2V617F allele burden and the presence of MVs [261]; (d) decreased levels of protein S [202,225,262]; (e) increased level of heparanase (TF activator) in the bone marrow of MPNs patients as compared to CML patients and in JAK2V617F transfected human glioma cells as compared to control, which was dependent of the EPO-JAK2 pathway

[265].

Phosphatidylserine exposure (negatively-charged membrane phospholipid) facilitates the binding of coagulation proteins (FII, FVII, FIX, and FX) with positively-charged γ- carboxyglutamic acid domains and the assembly of the cofactor/protease complexes [266]. In newly diagnosed and untreated ET patients (JAK2V617F, CALR and triple negative) phosphatidylserine expression is increased on circulating blood cells (leucocytes, platelets and erythrocytes) and on serum-cultured Human umbilical vein endothelial cells (HUVECs) compared to healthy controls [260]. Among ET patients, those with JAK2V617F mutation displayed higher level of phosphatidylserine exposure on circulating blood cells than CALR and triple-negative patients [260].

When the blood clot is no longer needed, fibrinolytic system allows its dissolution. An inactive pro-enzyme, plasminogen, is converted into active plasmin by two different activators, the activator of the tissue plasminogen (t-PA) and urokinase (u-PA) and degrades the fibrin

[192]. The main plasma inhibitor of t-PA and u-PA is the type I plasminogen activator inhibitor

(PAI-1)[267].

48 Several lines of evidence point out to a less efficient fibrinolytic system in MPNs, with higher

PAI-1 level and activity as well as a lower thrombin activated fibrinolysis inhibitor (TAFI) in ET and PV patients than in healthy controls and patients with secondary thrombocytosis [268–

272]. Moreover, patients with ET and PV have reduced clot permeability and a longer clot lysis time than controls [273,274].

Although these coagulation cascade modifications constitute an indirect evidence of an imbalance toward a procoagulant status, no direct link has been made between these changes and JAK2V617F, the JAK-STAT signalling (except the one from Kogan et al [265]) or the risk of thrombosis in the context of MPNs in a prospective cohort or in vivo.

Activated Platelets Leucocytes Platelets ì Elastases MVs Endothelial cells

PS EPCR TF Polyphosphate TFPI î TM Anticoagulant ì Elastases Factors ìTF APC FXIIa î FVIIa î Prot S Extrinsic pathway ì PAI-1 FXIa FIXa FXa Fibrinolysis FVIIIa FVa Plasminogen u-PA Prothrombin Thrombin TAFI ì t-PA Plasmin Common pathway

Fibrinogen Fibrin Polymerization Intrinsic pathway Clot formation Figure 8: Coagulation cascade and MPNs

49 b) Site specificity in MPNs patients

Most of systemic hypercoagulability states are associated with local thrombotic phenotypes, which is supported by mice experiments showing different thrombosis pattern depending of the mice genetic background and the breeding conditions [275,276]. The suggested major actor responsible for this site specificity is the endothelium [276]. Importantly, endothelial-derived anticoagulant and pro-coagulant molecules described in the previous chapter are unevenly expressed between the different vasculature as illustrated in Figure 9 and in the same site, between endothelial cell, because of phenotypic mosaicism [277].

In addition, to the different factors that could participate to venous thrombosis, detailed in the previous chapter, site-specific properties that could contribute to specific localization of thrombotic lesions are detailed in the following paragraphs. Indeed, in patients with MPNs, venous thrombosis generally occurs in limb deep veins and/or pulmonary veins (embolism), but the involvement of unusual venous sites is characteristic, with splanchnic vein and less frequently cerebral vein thrombosis.

50 Vein Capillary Artery

TM TM t-PA t-PA EPCR EPCR TFPI TFPI vWF vWF

TM t-PA EPCR TFPI vWF

Figure 9: Site specificity of anticoagulant and procoagulant factors (adapted from Aird et al [275])

(1) Limb deep veins thrombosis

The specificity of the lower extremities’ veins is the presence of venous valves, where thrombosis preferably initiates. Indeed, inside the valves, endothelial cells are exposed to low blood flow velocity, decreased oxygen tension (hypoxemia) and increased circulating blood cells adhesion [275]. The previously detailed increased adhesive protein expression by myeloid cells in the context of MPNs (cf chapter above) could be enhance in the venous valves and participate to deep vein thrombosis. In addition, in patients with MPNs, hypoxemia might be more severe in the venous valves. Indeed HIF-1a (hypoxia-inducible factor) seems to be increased in the bone marrow of patients with newly diagnosed MPNs and correlates with

JAK2V617F allele burden [278]. PV patients exposed to hypoxemia in altitude have almost 4 fold more vascular events than to those living near the sea [279]. Hypoxia is also implicated in the generation of NETs (cf chapter above), a well known pro-thrombotic phenomenon increased in

MPNs patients [246,248]. Thus, hypoxemia could participate to the pathophysiology of venous

51 thrombosis in the context of MPNs, but the direct link between hypoxemia in MPNs and venous thrombosis and particularly limb deep vein thrombosis has not yet been proven.

(2) Splanchnic vein thrombosis

Splanchnic vein thrombosis (SVT) includes Budd-Chiari syndrome (BCS) and portal vein thrombosis (PVT). Primary BCS is a rare disorder defined as a hepatic venous outflow obstruction at various levels from small hepatic veins to the terminal portion of the inferior vena cava [280]. Non-malignant non-cirrhotic extra-hepatic PVT is characterized by thrombus development in the main portal vein and/or its right or left branches and/or splenic or mesenteric veins, or by the permanent obliteration that results from a prior thrombus [280]

(Figure 10). Estimated incidence of BCS and of PVT in the absence of cirrhosis and of cancer is

0.35-2.5 cases per million per year and 1 per 100 000 million per year, respectively [281,282].

The pathogenesis of SVT is largely dependent on the presence of systemic pro-thrombotic conditions that promote thrombus formation in the respective splanchnic veins [280,283]. If patients with BCS or PVT commonly have risk factors for thrombosis, BCS and PVT represent only ≈1% of all venous thromboembolism suggesting that patients developing BCS and PVT have additional local factors [284].

52 Inferior vena cava

Budd-Chiari syndrome

Portale vein thrombosis

Figure 10: Splanchnic vein thrombosis

MPNs are the leading cause of SVT and are diagnosed in 25 to 50% of patients with SVT [285].

Importantly, PVT and BCS are 2’000 and 10’000 fold more common in patients with MPNs than in the general population [286]. The reason for this special link between MPNs and SVT is still unclear. Potential mechanisms, developed below, are summarized in Figure 11.

JAK2V617F mutation involvement

JAK2V617F is detected in most patients with SVT and MPNs, since only 10% to 20% have evidence for MPN at bone marrow biopsy or at assessment of endogenous erythroid colonies formation and no JAK2V617F [287]. JAK2 exon 12, CALR or MPL mutations are rarely identified in patients with SVT [287].

Over the last ten years, several studies have shed a new light on this close relationship between

MPNs and BCS or PVT. A study reported in 2009 the existence of JAK2V617F in hepatic veins endothelial cells from 2 Budd-Chiari syndrome patients [45]. Subsequently, JAK2V617F was also detected in splenic endothelial cells from patients with myelofibrosis [46]. Later, JAK2V617F was

53 found in circulating endothelial progenitor cells (ECFC) in 5 out of 17 JAK2V617F patients [47], a finding confirmed by independent groups [211,212]. Interestingly, patients harbouring

JAK2V617F in ECFC were those with a history of thrombosis (splanchnic vein thrombosis, deep vein thrombosis or stroke) [47,211]. JAK2V617F -mutated ECFC showed significantly higher adhesion proficiency to mononuclear cells than normal ECFC [47]. Nevertheless, the exact mechanism of how alterations in the endothelium lead to BCS or PVT remains unclear.

The previously cited work of C James [213] showing increased P-Selectin expression in transduced JAK2V617F cultured endothelial cells and the related increased neutrophils adhesion and in vivo thrombosis could be part of the explanation. However, mice developed thrombi in the lungs and there was no information about potential splanchnic vein thrombosis. In addition, this explanation stands only if the presence of JAK2V617F mutation is specific of the digestive vascular bed. If endothelial JAK2V617F mutation is ubiquitous, additional local factors are needed. Interestingly, in the study by C James and colleagues, in vivo, small concentrations of TNFa were required to uncover the pro-thrombotic effect of the endothelial JAK2V617F mutation [213]. This could explain the splanchnic specificity of thrombi, considering that potential inflammatory mediators derived from the gut might trigger P-selectin exposure.

However, against the hypothesis that JAK2V617F mutation is responsible for this splanchnic vein thrombosis specificity in MPNs, is the also high association of SVT with paroxysmal nocturnal hemoglobinuria, which is a rare acquired hematologic disorder of hematopoietic stem cells without JAK2 mutation involvement, suggesting mechanisms beyond JAK2V617F.

ECFCs

Increased ECFCs frequency was independently associated with previous SVT in a MPNs cohort and was absent in SVT patients without MPNs, thus increased ECFCs could be specific of

SVT in the context of MPNs [211]. In addition, as described previously ECFCs carrying JAK2V617F display higher adhesion capacity to mononuclear cells than normal ECFC [47]. We do not

54 known where ECFCs are capted, but they could go to the liver vasculature. There was also more

ECFCs in female and young patients which correspond to the subpopulation of higher risk of

SVT in MPNs, while there was no differences in ECFC frequency between sexe and age in healthy controls [211]. Thus even if the mechanism underneath is unknown, ECFCs could explain this association between MPNs and SVT.

Increased ECFCs frequency was also independently associated with non-active MPNs (normal blood cell count) [211]. The authors also conclude that patients with non active MPNs disease associated with SVT are a separate entity, based on several other papers pointing out the association of « indolent » MPNs and SVT [288–290]. However this concept of « indolent »

MPNs and SVT has been challenged by hepatologists, arguing that the « normal » blood formula is not due to an indolent MPN but to the portal hypertension in these patients. Indeed, patients with portal hypertension and hyper-splenism usually present hemodilution and thrombocytopenia, thus a « normal » blood count with SVT suggests an underlying MPN [291–

293]. However, up to 70% of patients with SVT do not carry a prior diagnosis of MPNs [294] while it only represents 40% maximum of patients with all thrombosis, and in a small cohort 6 out of 8 patients with JAK2V617F negative results at the time of SVT diagnosis were found

JAK2V617F positive after 21 month of follow-up [295]. This suggests probably a very low allele burden at the time of diagnosis. We could speculate that maybe JAK2V617F allele burden in ECFC and endothelial cells and not in circulating myeloid cells are the crucial points for SVT. In addition, aside from the specific case of MPNs and SVT, « indolent »MPN has also been seen with other type of cardiovascular events, such as myocardial infarction and cerebral venous thrombosis [288,290] thus supporting the reality of this entity.

Liver endothelial cells specificity

Endothelium morphology and function varies depending of the type of vessels and organs. Indeed, arterial and venous endothelium is a continuous layer, coupled by tight junction

55 and anchored to a continuous basal membrane. By contrast, the endothelium can be fenestrated with a basal membrane in endocrine and exocrine glands, gastrointestinal tract, choroid plexus, kidney glomeruli and subpopulations of renal tubules or without basal membrane in venous sinusoidal vascular beds in the liver and in the glomeruli, spleen and bone marrow [296]. The liver contains different types of endothelial cells, including arterial endothelial cells, venous endothelial cells in the portal system and the hepatic veins, and liver sinusoidal endothelial cells. As reviewed in Appendix 1, the liver sinusoidal endothelial cells have a very specific phenotype [297]. One of the characteristics of liver sinusoidal endothelial cells is an immune tolerance, which prevents a response to these gut factors [298]. There is no evidence that the endothelial cells of the portal and hepatic veins are similarly protected

[275,298], which could create a vulnerability to these possibly pro-thrombotic factors.

The possible involvement of liver vasculature, implied by the increased ECFC in patients with

SVT is further supported by the work that was carried out on pulmonary hypertension and

MPNs. Indeed, a link exists between pulmonary hypertension and MPNs, with multiple mechanisms suggested such as myeloid metaplasia, angiogenesis, pulmonary venous occlusive disease, endothelial dysfunction and fibrosis [299–301]. The increased proliferation seen in pulmonary endothelial cells in the context of pulmonary hypertension is dependant on the JAK-

STAT signalling pathway [302]. We could speculate that the same mechanisms could take place in the liver and participate to thrombotic events because of endothelial activation and proliferation. The link between sinusoid vessels and hematopoietic niche is now well described in the bone marrow of patients with MPNs [303–310], thus the possible extra-medullar haematopoiesis taking preferentially place in the liver and the spleen and the related endothelial activation and proliferation could participate to this association between splanchnic system and MPNs.

56 Other factors

Phagocytosis of abnormal RBC by the liver and the spleen, as shown in secondary erythrocytosis mouse model (EPO overexpression) could be an interesting additional pro- thrombotic mechanism to evaluate [311]. In addition, extravascular haemolysis seen in the liver and the spleen of patients with MPNs could participate to this site-specific thrombosis, by releasing free haemoglobin that can generate oxidative stress, and endothelial cells activation

[299].

Figure 11: Potential mechanisms involved in splanchnic vein thrombosis and MPNs

(3) Cerebral vein thrombosis

Cerebral venous thrombosis (CVT) is a rare condition. It includes thrombosis of cerebral veins and dural sinuses. It represents 0.5% of all strokes and affect 4 to 5 people per million in the general population [312]. In MPNs patients, the prevalence of CVT is around 1%, which is approximately 2000 times higher than in the general population [160,313]. MPNs is revealed by CVT in 3 to 7% of patients with CVT. JAK2V617F is detected in most of them, half of them without overt MPN [314–317].

57 No other factor than those described in the previous chapters can yet explain this association.

Whether JAK2V617F is present in cerebral endothelial cells and could participate to this association is still unknown.

3. Arterial vascular events and myeloproliferative neoplasms

The following paragraphs review the mechanisms by which MPNs might favour arterial events. In human, most arterial vascular events happen in the context of atherosclerosis. In some particular contexts, such as auto-immune and auto-inflammatory diseases or genetics abnormalities, arterial events can happen without underlying atherosclerosis.

a) Atherosclerosis and MPNs

Atherosclerosis is a progressive disease involving medium and large size arteries and causing several cardiovascular diseases, including coronary artery disease, stroke and peripheral arterial disease. MPNs patients are supposedly predispose to premature atherosclerosis, because of chronic inflammation and high risk of arterial cardiovascular events

[318,319], however the direct link has yet not be proven in patients or in animal model. Clonal haematopoiesis has recently been linked to atherosclerosis but outside the context of overt

MPNs. Only one study look at carotid plaque in 40 ET patients, free of overt atherosclerosis and there was no differences compare to healthy controls but the effectif was low and the patients treatment for MPNs was not mentioned [320]. However, there was significantly more high carotid calcification score than healthy control, which is known to predict major cardiac events

[320]. Several known factors in MPNs could favour atherosclerosis and are detailed below.

(1) Clonal haematopoiesis and atherosclerosis

Recent studies opened a new field in cardiovascular research regarding the link between clonal haematopoiesis and cardiovascular disease, in particular atherosclerosis. Indeed,

58 carrying somatic mutations in haematopoiesis-related genes increased by 2-fold the risk of developing atherosclerosis, an effect that was comparable or even greater than that of conventional risk factors [321]. Hematopoietic stem/progenitor cells accumulate somatic mutations during life. Some of these random mutations confer a competitive advantage to the mutant cells, leading to its clonal expansion, which without haematological abnormalities is called clonal haematopoiesis of indeterminate potential (CHIP) [322]. Most somatic mutations associated with clonal haematopoiesis occur in 3 genes (DNA methyltransferase 3α [DNMT3A], tet methylcytosine dioxygenase 2 [TET2], and additional sex combs like 1 [ASXL1]) that encode for epigenetic regulators that are involved in the control of haematopoiesis. Jaiswal et al. published a case–control studies that together enrolled 4726 participants with coronary heart disease and 3529 controls and by whole-exome sequencing detected the presence of CHIP.

JAK2 mutation without “overt MPNs” is referred as CHIP in this study [323] while this definition is contested by others [324]. However, 5 out of 6 patients with JAK2V617F mutation and myocardial infarction in this study displayed a normal blood count [323,325]. In addition, the presence of JAK2V617F mutation in patients without overt MPNs and with cerebral venous thrombosis [316], stable coronary heart disease [326], or peripheral artery disease [327] support the fact that JAK2V617F can be found before a possible complete MPNs diagnosis.

JAK2V617F and TET2 mutations conferred a 12 and 2 fold increased risk of coronary heart disease respectively compared to patients without mutations [323]. In two independent studies

TET2 loss-of-function–driven clonal haematopoiesis accelerates atherosclerosis in hyperlipidemic mice [323,328], which in the first study was mainly because of exacerbated expression of proinflammatory cytokine IL-1β [328]. In this context, the CANTOS trial, the first successful immunotherapy trial in preventing secondary atherosclerotic cardiovascular disease in high-risk patients exhibiting evidence of systemic inflammation, using a neutralizing antibody (canakinumab) against IL-1β could be interesting in the context of clonal

59 haematopoiesis and MPNs [329]. However, the exact mechanism underlying accelerated atherosclerosis in the context of clonal haematopoiesis is still unknown. And the direct role of

JAK2V617F in atherosclerosis development has yet not been explored.

(2) Initiation phase (Figure 12)

Shear stress

Cells lining the circulation are exposed to fluid forces that generate a wide range of hemodynamic stresses that vary greatly depending on the area and the pathophysiological context. Forces acting on vessels due to blood flow can be resolved into two principal vectors, perpendicular to the wall (representing blood pressure) and parallel to the wall, creating a frictional force, at the surface of the endothelium, called shear stress [330]. Shear stress evaluated by Poiseuille’s law is classically expressed in dyne/cm2 (or sometimes in Pascal (Pa)) and is proportional to the flow and viscosity and inversely proportional to the radius: τ = 4 μ Q

/ π r3 (μ=viscosity, Q=flow, r=radius) [331]. In a normal, straight portion of a artery, shear stress is between 10 to 70 dyne/cm2, while it is only between 1 and 6 dyne/cm2 in a normal vein [331]. At bifurcations and curvatures, flow is bidirectional, decreasing shear stress [331].

A constant physiological shear stress is required to maintain a healthy endothelial phenotype, with anticoagulant and anti-inflammatory properties, and is protective against oxidative stress and cell death [331,332]. Atherosclerotic plaques appear in these specific areas, namely intersections and curvatures, where blood flow is disturbed.

In MPNs patients, increased haematocrit and hyper-viscosity have been the oldest aspect incriminated in the associated cardiovascular events [189,333–335]. Lowering haematocrit decreases thrombotic events [189]. However, therapeutics used not only decrease shear stress but also act on other factors, such as pathological circulating blood cells, NO pathway and inflammation. We could argue that this increased hyper-viscosity increases shear

60 rate and could actually compensate endothelial dysfunction. However, excessive shear stress can be deleterious for the endothelium and activate platelets [336,337]. Indeed, paradoxical hypotension following moderate increased haematocrit (<19% from baseline) and blood viscosity, due to the subsequent increase in shear stress and the induction of the eNOS-NO pathway has been described, while this effect was lost for higher haematocrits [338].

Endothelial dysfunction

In normal conditions, the endothelium is able to generate vasodilator factors in response to increased shear stress in order to attenuate blood pressure. The loss of this property is called endothelial dysfunction [339]. In the previously described low shear stress areas, endothelial cells display increased permeability, activation, senescence and apoptosis. Low density lipoproteins (LDLs) cross the endothelial barrier and accumulate in the sub- endothelium in these areas [340]. Then LDLs undergo several modifications, such as oxidation after exposure to reactive oxidative species from endothelial cells and macrophages. Oxidized

LDLs also inhibit the production of NO in endothelial cells, favouring endothelial dysfunction and vasoconstriction [341]. Modified LDLs are removed by macrophages, which in turn become foam cells and accumulate in the intima [342].

In MPNs patients, only one study suggested a role of endothelial dysfunction [343].

Indeed, endothelial dependant flow mediated vasodilatation was markedly decreased in PV patients even in the absence of overt arterial disease; whereas endothelium-independent, nitroglycerine-induced vasodilatation was not impaired compared to control. In addition this impaired flow mediated vasodilatation was not correlated to the haematocrit level and platelets count [343].

61 Endothelial to mesenchymal transition

Accumulating evidence suggests that endothelial to mesenchymal transition (phenotypic switch by which ECs lose their characteristics and acquire mesenchymal traits) represents a key link in the complex interactions between inflammatory stress and endothelial dysfunction

[344,345]. The degree of endothelial to mesenchymal transition correlates with coronary atherosclerosis in patients, is enhanced in disturbed shear stress areas and as been involved in atherosclerosis initiation (through endothelial dysfunction), progression (through plaque remodelling) and complications (through plaque instability) [344,346]. In patients with PMF more than 30% of endothelial cells in the spleen and the bone marrow vessels exhibit mesenchymal phenotype [347]. In addition, endothelial TGFβ signalling correlates with the degree of endothelial to mesenchymal transition in atherosclerotic mouse model [346] and increased TGFβ signalling is well described in MPNs patients [348–350]. Thus, we could speculate that the activated TGFβ signalling in MPNs patients could promotes endothelial to mesenchymal transition and atherosclerosis development. However, even if endothelial to mesenchymal transition is present in MPNs patients, its implication in atherosclerosis development and arterial CV events has never been study in this context.

(3) Progression phase (Figure 12)

Recruitment of Inflammatory Cells

Considerable evidence supports the early involvement of the monocyte/macrophage, the most prominent cellular component of the innate immune response, during atherogenesis.

Leukocytes adhesion to the endothelium is essential in the development of atherosclerosis.

This interaction is facilitated by the coordinated expression of various adhesion molecules

(selectins and integrins) and chemo-attractant factors [351]. Secondary homotypic interaction between leukocytes is possible via PSGL-1/L-selectin binding [351]. Then, leukocytes cross the

62 endothelium via trans-endothelial migration (diapedesis) [352]. In the context of atherosclerosis modified LDLs in the intima trigger an early inflammatory response, which favour monocyte and macrophages recruitment into the intima via activated endothelial cells expressing adhesive receptors and releasing cytokines, such as monocyte chemotactic protein-

1, IL-6, and IL-8 [353–355].

In MPNs patients leucocytes and endothelial activation, in addition to increase adhesive molecules expression (as described above) could participate to an increase tissue inflammation [151,220,223,243,356]. In addition, MPNs patients display a low grade chronic inflammation, with significantly higher level of several cytokines compared to healthy control, such as MCP-1, IL-6, IL-8 [357] and hsCRP [356] that could participate to premature atherosclerosis, as described in other chronic inflammatory diseases, such as rheumatoid arthritis, systemic lupus erythematosus or diabetes [358–360]. The study from C James team also suggests a pro-inflammatory endothelial phenotype with increased P-Selectin in the presence of JAK2V617F mutation in endothelial cells [213]. The in vitro experiments were done on HUVEC, which mimic arterial endothelial cells, however in mouse model only venous thrombosis was evaluated. In addition the presence of the mutation in arterial endothelial cells has never been proven yet.

Monocytes-macrophages transition / Foam cells formations

The transformation of recruited monocytes into lipid-laden macrophages or “foam cells” by modified LDL is central to the development of atherosclerotic lesions [361]. Deregulated uptake of modified LDL via scavenger receptors such as SR-A and CD36 determines foam cells formation in vivo [362,363]. Scavenger receptors expression is regulated by PPAR Gamma

(transcription factor) and by cytokines such as TNF-α and IFN-γ [340]. During progression, macrophages and foam cells then undergo cell death leading to the deposition of extracellular lipids and apoptotic cells forming the lipid core [340].

63 The increased number of activated macrophages and monocytes with increased proinflammatory cytokine production in patients with MPNs [357] could favour atherosclerosis development, but it has never been directly proven. In MPNs patients, foam cell formation could be altered by 1/increased phagocytic signal on mutated blood cells or by 2/an increased phagocytic capacity by macrophages, within the plaque.

First, hypocholesterolemia has been described a long time ago in MPNs patients [357].

Non-receptor-mediated removal of LDL is increased in MPNs patients compared to healthy controls [364,365] and this catabolism takes place in the bone marrow and the spleen of patients [366]. We could speculate that this increased LDL removal probably by activated monocytes-macrophages would increased foam cell formation or that hypocholesterolemia would prevent atherosclerosis. Scavenger receptors and foam cells formation has never been studied in the context of MPNs and atherosclerosis.

Secondly, CALR expressed at the cell surface is known to transduce a phagocytic signal to macrophages, by binding to the LDL receptor on phagocytes and is counterbalanced by anti- phagocytic signal by CD47 [367]. One group looked at phagocytic signal from blood cells in

MPNs patients [368]. They have shown that Haematopoietic stem cells and mature blood cells from PV/ET patients have a similar surface expression of CALR and CD47 compared to healthy control and that JAK2 and CALR mutations do not alter this expression. In addition, haematopoietic stem cells and mature blood cells from MPNs patients and healthy controls displayed a similar sensitivity to healthy macrophages engulfment [368].

In conclusion, mutated blood cells don’t seem to trigger increased phagocytic response to healthy macrophages in MPNs patients, however macrophages themselves could display an increased phagocytic capacity, thus possibly favouring foam cell and atherosclerosis formation, even if it has never been proven.

64 Adaptive immune cells

It is now clear that adaptative immune response is also implicated in atherosclerosis development. Indeed, an array of plaque antigens is presented to effector T-lymphocytes, such as NK cells by antigen-presenting cells, such as dendritic cells [353]. NK cells can promote atherosclerosis by secreting cytokines that activate immune cells present in the atherosclerotic lesion and by the induction of apoptosis by cytotoxic proteins [369]. In addition, regulatory T cells can induce the downregulation or tolerance of the immune system leading to limitation of atherosclerotic plaque progression and complications in an antigen-specific and non-specific manner [370,371]. Patients with clinically stable atherosclerotic plaque reported higher levels of Tregs than patients with recurrent myocardial infarction [372,373].

Several evidences point out the involvement of lymphoid lineage in MPNs. Indeed, MPL and JAK2V617F mutations have been found in lymphocytes, mainly B and natural killer cells

(more rarely and later in T cells) [23,77,357]. The hypothesis is that some driver mutations could occur in a more primitive lympho-myeloid progenitor. One group have shown in 54 ET patients than compare to healthy controls, B-cells were significantly more activated (CD69 and

CD86, increased intracellular IL-6 and IL-1β levels, and higher expression of TLR4) regardless of the mutational statut. However, without clear explanations, these increased B-Cells activation was not found in PV patients [374]. Regarding NK cells, sanchez et al have shown that patients with PV presented an increase number of NK cells, which can carry JAK2V617F in comparison with healthy donor [375]. However, the NK cells phenotype was similar between

PV patients and healthy controls [375]. Regulatory T cells have been studied by several teams with contradictory results [376]. Zhao et al [377] described an increased number of regulatory

T cells in 21 PV patients compare to healthy controls, while Keohane et al [378] studied 50 patients with MPNs (ET, PV and PMF) in which regulatory T cells were lower than healthy controls.

65 In conclusion, lymphoid cells modifications related to MPNs exist, but need to be confirmed and their implications in disease progression and complications are still unknown [357].

(4) Neovessel formation

In a normal vessel wall, microvasculature is limited to the adventitia and outer media

[379]. In the presence of atherosclerosis, the intima thickens and intimal neoangiogenesis appears [380]. This network of leaky neovessels is involved in the development of atherosclerosis, in intra-plaque haemorrhage and its complications (acute rupture) and is an independent predictor of systemic cardiovascular outcome [379,381]. The molecular mechanisms responsible for neovessels formation relate predominantly to hypoxia in the thickened plaque through HIF-1 activation, which stimulate eNOS and VEGF-A. In addition HIF-

1 can be activated by inflammatory pathway (Toll like receptor, TNF, NF-kB and other growth factors (IGF-1, IGF-2, FGF-2 and TGF-1) favoured by inflammatory cells infiltration [379].

To approach the question of angiogenesis in the context of MPN, we must first adjust the definition of endothelial progenitors cells (EPC), which have been extensively studied in the context of MPNs, but without consistent results because of variable definitions and identifications methods [382,383]. Progenitor cells in adults are highly heterogeneous and present different stages of differentiation, thus making their identification difficult. Asahara in

1997, described from circulating mononuclear cells circulating progenitors, CD34+, VEGFR+,

Tie2+, which part of them can differentiate in ECs in vitro (CD34, CD31, Flk-1, Tie-2, and E selectin), that he called EPC [384]. Confusion exists because all cell type involved in adult vasculogenesis are called « EPC ». Prater et al [206] clarified the different types of the so called

“endothelial progenitor cells” and concluded that the more appropriate way to identify endothelial progenitor cells is based on their ability to form endothelial colonies in long outgrowth culture (ECFC) [385,386], the other so called endothelial progenitors cells are

66 probably not from endothelial origin but participate can also in cardiovascular disease [387–

390].

Furthermore, endothelial cells (only proved in venous endothelium) and ECFC can carry the

JAK2V617F mutation and is correlated with previous thrombotic events [47,208,211,212]. Some studies found mutational abnormalities related to MPNs only in the other types of EPC, which are suspected to be of myeloid origin and not in ECFC [386,391]. Anyway, non-mutated ECFC from patients showed an increased proliferative potential compared to healthy controls, which could participate to increased angiogenesis in MPNs patients regardless of the mutational status [386]. Indeed, an increased micro-vascular density has been described in the bone marrow of MPNs patients, especially in the presence of myelofibrosis and high JAK2V617F allele burden, with increased level of VEGF/VEGFR [304–310]. Angiogenesis inversely correlates with survival, but no studies with anti-angiogenic treatment have been shown to be efficient in

MPNs patients on disease activity [392]. However, in contrast with bone marrow angiogenesis and its role in the development of MPNs [207], the implication of endothelial progenitor cells and angiogenesis in the context of cardiovascular events is still unclear in MPNs patients.

We could speculate that this increased angiogenesis in MPNs patients could participate to atherosclerotic development and complications. In addition, as described in the previous section, hypoxia pathways are enhanced in MPNs patients and could play a part not only in venous thrombosis but also in neovessel formation in the context of atherosclerosis [278,310].

(5) Plaque Rupture and Thrombosis (Figure 12)

Physical disruption of plaques may trigger thrombosis and promote downstream ischemic events. Different types of physical disruption may occur [340,353] (figure 13):

• The most common mechanism of plaque disruption is a fracture of the plaque’s fibrous

cap (60-80%).

67 • Second, superficial erosion or microscopic areas of endothelial cells desquamation

account for approximately one-quarter of fatal coronary thrombosis (20-40%).

• Rarely, disruption of the micro-vessels that form within atherosclerotic plaques

furnishes another scenario for sudden plaque progression. The neovessels in the plaque

may be particularly fragile and prone to micro-haemorrhage [381].

2 other mechanisms can participate in atherosclerosis ischemic complications (Figure 13):

First thrombi embolism: as the thrombus begins to protrude into the lumen the fibrin component increases, but any surface exposed to the blood in the lumen will be covered by activated platelets. While antegrade flow continues over this exposed thrombus, clumps of activated platelets are swept down into the distal intramyocardial arteries as microemboli

[393].

Secondly, arterial spasm is an under diagnosed phenomenon that can happen outside of atherosclerosis context (cf next chapter), but is also favoured by underlying non stenotic atherosclerotic plaque, because of local endothelial dysfunction responsible for local intense vasoconstriction [393,394].

Thrombus-mediated complications of atherosclerosis depends more on plaque composition and vulnerability than on the degree of stenosis [340]. Lesions thrombogenicity depend on tissue factor expression within the plaque [257]. Vulnerable plaques are mostly characterized by a large lipid core rich in tissue factor with increased number of inflammatory cells infiltrations covered by a thin fibrous cap (low vascular smooth muscle cells density and collagen disorganization) prone to rupture [340]. Weakening of the fibrous cap results from extracellular matrix proteins degradation and from endothelial and smooth muscle cell disappearance likely after cell apoptosis [395,396].

Figure 13: Different mechanisms for arterial ischemia [394]. “Republished with permission of Wolters Kluwer Health, Inc, from [394]. ; permission conveyed through Copyright Clearance

Center, Inc. ”

In MPNs patients, the phenomenon described in the venous section, with activated leucocytes and platelets in addition to a pro-coagulant status can participate to increased atherosclerotic complications. Indeed, several mechanisms can lead to complications: 1/The activated leucocytes and NET’s infiltration associated with increased inflammation might increase plaque vulnerability, 2/the increased neutrophils proteases release can favour endothelium erosion and 3/the enhanced angiogenesis and intra-plaque neovessel formation could lead to plaque haemorrhage and disruption. In addition, endothelial to mesenchymal transition, which has been described in MPNs is associated with plaque instability [344].

Figure 13: Potential mechanisms in atherosclerosis and atherothrombosis in MPNs

b) Beyond atherosclerosis

Arterial ischemia with or without obstruction may occur in the absence of atherosclerosis. Indeed, thrombosis and/or ischemia without atherosclerosis can happen in

the presence of vasculitis (in systemic lupus erythematosus and Takayasus’s or giant cell

arteritis for example) or in apparently normal arterial vessel, like in MPNs [275,397–399].

The other possible mechanism for arterial ischemia without atherosclerosis is by

microvasculature dysfunction and/or thrombosis [400,401].

In patients with MPNs, a multicentric European study including 1051 patients with PV

without past history of thrombosis showed an incidence of arterial cardiovascular events of

2% per patient per year (3 years follow-up) [402], while in patients without PV and with

similar cardiovascular risk factor the incidence was 0.2% per patient per years [403].

Interestingly, an other study showed in patients with PV and ET that 21% of those with

myocardial infarction did not have significant coronary stenosis (>50%) by angiography

[404], while the prevalence of myocardial infarction with normal coronary angiography is

estimated to range from 1 to 10% maximum [400,405]. These data suggest that other

mechanisms beyond atherosclerosis are involved in MPNs patients and prompted the

European society of cardiology to recommend searching for MPNs in case on myocardial

infarction without obstructive coronary disease [400].

The absence of significant atherosclerosis stenosis, or normal coronary artery does not

mean the absence of atherosclerosis involvement. Indeed, plaque disruption and

thromboembolism can appear with non significant stenosis and non significant and/or not

71 visible atherosclerosis plaque can favour other mechanisms, such as vasospastic angina and microvasculature dysfunction (Figure 14) [401].

Atherothrombosis Vasospam

Microvascular ischemia

Figure 14: Arterial ischemia triad (adapted from [401])

(1) Arterial vasospasm (Figure 15)

Arterial vasospasm can occur in normal or atherosclerotic arteries and has been described to participate to myocardial infarction and ischemia with non obstructive coronary disease [400,401,406]. The mechanisms responsible for vasospastic angina are still not elucidated, but endothelial dysfunction and vascular smooth cell hyper-reactivity have been described [400,401,406,407]. Vasoconstriction abnormalities are classically described in the coronary heart disease, but can also occur in brain arteries [408].

72 The endothelium plays a major role in the control of vasomotor tone. Indeed, under the

influence of various physical and humoral stimuli derived either from the blood or the

perivascular compartment, endothelium controls vascular smooth muscle cells contraction, by releasing various molecules with vasodilator or vasoconstrictor actions, or by physically interacting with them.

Endothelium-derived relaxing factors Nitric oxide

In 1980, Furchgott and Zawadzki demonstrated that vasodilatation induced by

acetylcholine was dependent on the presence of an intact endothelium and appeared

depend on a powerful unknown factor, referred to as a relaxing factor derived from

endothelium (EDRF) [409]. Several studies have shown that EDRF is released into the

blood vessels under basal conditions, as well as after stimulation with acetylcholine, and its

biological effects were mediated by stimulating soluble guanylate cyclase resulting in

elevated levels of cyclic GMP in vascular smooth muscle cells [410]. Based on the

similarities between the pharmacological properties of EDRF and NO, EDRF has

subsequently been identified as NO [411] and L-arginine as the precursor of NO synthesis by endothelial cells [412]. Endothelial cells metabolize L-arginine through endothelial NO synthase (eNOS) to form NO and L-citrulline [192]. NO synthesis can be stimulated by

receptor-dependent agonists (acetylcholine and bradykinin), or independent receptors

(calcium ionophores), but also by shear stress exerted by the blood flow (Figure 15) [413].

Once NO is produced by eNOS, it diffuses quickly through cell membranes to act as a

powerful paracrine mediator, with a half-life of a few seconds in tissues and physiological

fluids since it reacts with superoxide anion to form peroxynitrite or is rapidly inactivated

73 by oxyhaemoglobin to form nitrate and methaemoglobin [192]. Although the majority of

NO physiological effects are due to the activation of soluble guanylate cyclase, other actions are the consequence of a particular reaction, the S-nitrosylation of various proteins, which

modulates a number of physiological processes, including cell proliferation [411],

apoptosis [414,415], and ion channel activity [416].

In addition to NO, three other gaseous mediators have been described: carbon monoxide

(CO), hydrogen sulfide (H2S) and possibly sulfur dioxide (SO2) [192].

In MPNs patients, plasma NO level has been measured in several studies

[198,199,254]. Cella et al found that NO plasma level in 18 ET patients was decreased as

compared to 19 healthy controls, but in the 14 PV patients NO level was increased as

compared to the healthy controls group, potentially secondarily to increased shear stress

(mean haematocrit at 58%). Hydroxyurea increased significantly the NO level in ET

patients and did not modify it in PV patients in this study [198]. In line with this finding,

Piccin et al [199], found a decreased NO level in 11 untreated ET patients compare to 18

healthy controls and Rusak et al [254], increased NO level in 73 PV patients in comparison

of 38 healthy controls.

However, all these studies used the nitrite/nitrates evaluation methods. Although

Nitrite/nitrates can be byproducts of eNOS-derived NO oxidation [417], they are not

completely specific of the NO pathway [418]. Thus, what these study looked at, are the

oxidize form of NO and not its active form. The best method to assess NO is Electron

Paramagnetic Resonance Spectroscopy [418]. However, this technique is not easily

available and complicated and has not been used, in MPNs patients.

74 The only study available to evaluate NO active form in MPNs patients is the previously

mentioned work done by Neunteufl et al, that showed that endothelial dependant flow

mediated vasodilatation was markedly decreased in PV patients even in the absence of

overt arterial disease [343], which argues for an impaired NO pathway in PV patients.

Phosphorylation of eNOS by Akt represents a Ca2+-independent regulatory mechanism for

activation of eNOS [419]. Thus, we could extrapolate that the activation of Akt by JAK2V617F

mutation, could up-regulate eNOS activity and that the decreased active NO level in MPNs

patients could be the result of an increased scavenging or an inactive eNOS by different

mechanisms such as uncoupling [420]. Indeed, Rusak et al [254] found in his cohort of 73

PV patients that plasmatic nitrites/nitrates level moderately correlated with increased free

haemoglobin. In addition, the increased free haemoglobin was correlated with the blood

pressure, supporting the idea of increased NO scavenging in MPNs patients.

Prostacyclin

In the vascular system, prostacyclin is produced preferentially by the

endothelium, by several enzymes including phospholipase and cyclooxygenase (COXs) and is the main metabolite of arachidonic acid. Prostacyclin is a powerful inhibitor of platelet

adhesion to the surface of endothelial cells and platelet aggregation and acts as a

vasodilator derived from the endothelium and an inhibitor of vascular smooth muscle cells

migration and proliferation [192].

In 1985, a superficial vein was surgically removed from 18 patients with newly diagnosed

MPNs (with an old definition), and the prostacyclin generation by these vessel were

analysed and found normal [421]. Prostanoid metabolites were then extensively study in

patients under aspirin treatment to evaluate its efficacy in MPNs patients. Indeed, Cavalca

75 et al showed that the prostanoid metabolism was similar in the serum and urine from

MPNs patients treated with aspirin than healthy controls [422], but the platelets functional response was abnormal as compared to controls, shifting towards increased platelets aggregation, with a resistance to anti-aggregating prostaglandin [423–425]. However, this study did not include a control group of ET patients without aspirin treatment.

Prostacyclin analogues have been used successfully only in case reports in patients with

MPNs, in severe mesenteric ischemia in a context of primary thrombocytemia [426] and in coronary vasospasm in a patient with CML [427].

Endothelium-derived contracting factors Endothelin-1

Endothelin-1 (ET-1) interacts with two G protein-coupled receptors called ETA and ETB. ETA and ETB receptors are both localized on vascular smooth muscle cells where they exert vasoconstrictor, proliferative and hypertrophic role. Endothelial cells express

ETB receptors, which activation is associated with the release of NO and prostacyclin.

These endothelial effects of ET-1 counterbalance the effects of ETA and ETB receptor stimulation on vascular smooth muscles cells [428,429].

ET-1 levels were only measured in serum from treated patients with MPNs, where it was

lower than in healthy controls [199], but no data on basal ET-1 production in MPN patients

was available. One very interesting case report from a patient with JAK2V617F positive PMF

showed an abundant ET-1 staining in the bone marrow, which could participate to

osteosclerosis, via osteoblast stimulation [430]. But we could speculate that if the bone

marrow endothelium produces abundantly ET-1, it could be the same in other sites and in

the blood circulation.

76 Thromboxane A2

Thromboxane A2 is an arachidonic acid-derived vasoconstrictor agent, mainly

derived from platelet COX-1, but that can also be produced by other cell types, including

endothelial cells. Thromboxane A2 stimulates not only causes platelet aggregation and

contraction of vascular smooth muscle cells, but also the expression of adhesion molecules

and leucocytes infiltration [431]. Thromboxane A2 levels have been measured by numerous groups since platelets were thought to be the major cells involved in cardiovascular complications in MPN. An increased thromboxane A2 level in the plasma,

excreted in the urine and released ex-vivo by platelets has been described in PV and ET patients [204,432–441]. One prospective study found that the level of thromboxane A2 excreted in the urine was associated with the development of arterial micro-vascular

phenomenon after aspirin withdraw in PV and ET patients [442]. Thromboxane A2 was

also extensively studied to understand the apparent aspirin resistance in MPNs patients.

Indeed, in MPNs patients, because of the abnormal megakaryopoiesis, with accelerated

platelet regeneration, low-dosing aspirin incompletely inhibits platelets thromboxane A2.

This impaired platelet inhibition can be rescued by modulating the aspirin dosing interval

(Twice a day) rather than the dose [422,437,443,444]

Reactive oxygen species

The oxidative stress plays an important role in the regulation of vascular tone. In

response to different stimuli, several endothelial enzymes, including xanthine oxidase,

NADPH oxidase, decoupled NO synthase or cyclo-oxygenations, are able to produce reactive oxygen species, especially hydrogen peroxide and anion superoxide. The superoxide anion effectively inactivates NO (reaction giving rise to peroxynitrite), thus

77 reducing its bioavailability and favouring endothelium dependant contractions. In addition,

peroxynitrites cause nitration of proteins, in particular prostacyclin synthase, thereby

inhibiting their activity. The metabolism of arachidonic acid is then diverted to vasoconstrictor prostaglandin synthesis pathways [192].

If the role of ROS in bone marrow and stem cells in clonal expansion is well established

[358,445–449], its implication in vascular events in patients with MPNs has never been evaluated. However, several lines of evidence suggest a general unbalanced oxidative stress status in MPNs patients.

First, in vitro, JAK2V617F leukemic cell lines produce an higher level of oxidative stress products with increased DNA damage from aberrant PI3K signalling, concomitantly with repressed apoptosis than cells transfected with control siRNA [450].

Second, ex-vivo, neutrophil-derived reactive oxygen species has been measured in several

MPN cohorts. MPN neutrophils display increased basal ROS production [249,451], which

was not restricted to JAK2V617F patients, but also in JAK2WT and CALR patients [249]. In one

cohort, neutrophil derived ROS levels were significantly higher in PV patients than in ET

and PMF patients [249].

Finally, in vivo, only few studies analysed the oxidative status, with discrepancies due to

population heterogeneity. Most of them found higher serum ROS products levels in MPNs

patients [452–454] and lower antioxidant status [453] than in healthy individuals. This

finding was confirmed by whole blood transcriptional profiling from ET, PV and PMF

patients, with a significant up-regulation of several oxidative stress genes in concert with

down-regulation of important anti-oxidative defence genes [455]. However, one recent

78 study selected 30 ET patients without a history of thrombosis or on-going cytoreductive treatments, to avoid confounding factors and found no statistical differences in ROS level between patients and healthy controls [456]. Two explanations can be put forward to account for this discrepancy: (a) there was a tendency towards increased antioxidant system in MPN patients, maybe as a mechanism compensating increased ROS production

[456]; (b) MPN patients included were at low or intermediate risk for thrombosis, which might reduce the probability of finding significant differences in ROS level between patients and controls.

If oxidative status of MPNs patients has been analysed in stem cells or the neutrophils, it has not been assessed in other cells participating in vascular homeostasis and pathophysiology. Indirect data could suggest an involvement in addition to neutrophil of red blood cells, because of an possible correlation between ROS and haemoglobin value in

PV patients [452]. In retrospective studies, ROS products were significantly higher in patients with thrombosis compared to those without thrombotic events [452,453], but this finding does not give causality arguments of ROS for thrombotic events.

Angiotensin II

Kidney-derived renin cleaves liver-derived angiotensinogen to form angiotensin I in circulating blood, which is then converted into angiotensin II, the main effector peptide of the renin–angiotensin system, by angiotensin-converting enzyme predominantly located at the luminal side of the endothelium. Angiotensin-II activates its receptor in the surface of vascular smooth muscle cells, which mediates most of its effects, including vasoconstriction, increased proliferation of vascular smooth muscle cells and extracellular

79 matrix synthesis. The vasoconstrictor effect of angiotensin-II is essentially due to Gq

protein activation resulting in intracellular calcium mobilization, but also by an increased

reactive oxygen species production following NADPH oxidase activation. Angiotensin II

action on its endothelial receptors induces a vasodilator effect secondary to increase NO

production [192].

In MPNs patients, increased local bone marrow synthesis of ACE and other angiotensin

components has been identified in PV and ET patient, particularly in the presence of

JAK2V617F mutation [457–461]. Therefore, up-regulation of the local bone marrow renin- angiotensin system has been suggested to play a role in the development of MPNs [457]. Its participation in cardiovascular complications is not yet proved. Indeed, the RAS is not only implicated in vascular tone, but also in cell adhesion and migration through MMP2/MMP9 and ICAM1 regulation [461,462]. However, in the ECLAP study patients under ACE inhibitor or not had the same risk of thrombosis [457].

In conclusion, MPNs patients seem to display a tendency toward endothelial dysfunction, with decreased vasodilator agents and increased vasoconstrictive one, which could play a part in arterial disease.

Figure 15: Potential mechanisms in vasospam and MPNs

(2) Microvasculature ischemia

An other mechanisms possibly involved in arterial events beyond atherosclerosis is microvascular dysfunction [406] Coronary microvascular dysfunction is defined as epicardial, microvascular endothelial, or non endothelial dysfunction that limits myocardial perfusion, most often detected as reduced coronary flow reserve [406].

Traditional atherosclerosis risk factors such as aging, hypertension, diabetes mellitus, and dyslipidemia are associated with coronary microvascular dysfunction [406]. Other risk factors are also associated with coronary microvascular dysfunction, such as systemic inflammation, abnormal adrenergic nerve function, platelets function and platelets- leukocytes aggregates [406]. Like macrovascular spam, the microvascular bed can also suffer from unbalanced vasodilatation-vasoconstriction [394]. Inflammation, platelets- leukocytes abnormalities and endothelial dysfunction are found in MPNs patients and could contribute to microvascular dysfunction and consequently to arterial events.

Interestingly, perivascular fibrosis in the coronary microvasculature can contribute to

81 myocardial ischemia [394] and one team found in a PV mouse model, a major heart dysfunction with extensive fibrosis [463].

MPNs patients display frequently dermal microvascular dysfunction and thrombosis, characterized by erythromelalgia (painful red or blue extremities), that by pathology examination demonstrated fibromuscular hyperplasia, narrowing of the lumen, swelling of

ECs, and platelet-rich thrombi in arterioles [275,397,464]. Thus, we could speculate that other sites than the derm could display microvascular dysfunction and/or thrombosis in

MPNs patients.

82 III. Thesis work A. Aims

This work focuses on the pathophysiology of cardiovascular events in the context of myeloproliferative neoplasms.

First, arterial cardiovascular events are the most common one in patients with

MPNs. As detailed in the introduction, other mechanisms beyond atherosclerosis can be

suspected. Indeed, patients with MPNs display a high frequency of myocardial infarction

with normal coronary angiography. The mechanism underlying the link between

myocardial infarction without obstructive coronary disease and MPNs is unknown, but

vasoactive phenomenon (local intense vasoconstriction) can be suspected. Therefore, the

purpose of the first study was to examine the consequences of JAK2V617F on arterial vascular

reactivity.

Secondly, myeloproliferative neoplasms are the leading cause of BCS and JAK2V617F has been detected in endothelial cells in patients with SVT. The consequences of JAK2V617F in liver endothelial cells are still unknown. Therefore, second study focused on the consequences of endothelial JAK2V617F in the context of BCS.

Finally, the aim of the last study was to identify the subgroup of patients with SVT at the highest risk of harbouring CALR mutations and thus requiring this genetic testing.

B. JAK2V617F in arterial events

1. Article 1: Erythrocyte microvesicles increase arterial contraction in JAK2V617F myeloproliferative neoplasms Erythrocyte microvesicles increase arterial

contraction in JAK2V617F myeloproliferative

neoplasms

Authors: Johanne. Poisson1, Marion. Tanguy1, Hortense. Davy2, Fatoumata. Camara1,

Marie-Belle. El MDawar1, Cécile. Devue1, Marouane. Kheloufi1, Juliette. Lasselin1, Aurélie.

Plessier2, Villeval. Team3, Olivier. Blanc-Brude1, Michèle. Souyri4, Jean-Luc Villeval3,

Chantal. M. Boulanger1, Pierre-Emmanuel. Rautou1,2,5*

Affiliations:

1Inserm, UMR-970, Paris Cardiovascular Research Center, PARCC, Paris Descartes

University, Paris, France.

2 Service d'Hépatologie, Centre de Référence des Maladies Vasculaires du Foie, DHU Unity,

Pôle des Maladies de l’Appareil Digestif, Hôpital Beaujon, AP-HP, Clichy, France

3Inserm U1170, Institut Gustave Roussy, University Paris IX, Villejuif, France

4Inserm UMR-S1131, IHU, University Paris Diderot, Paris, France.

5 University Paris Diderot, Paris, France.

Original article (Submitted)

One Sentence Summary:

In JAK2V617F myeloproliferative neoplasms, red blood cells microvesicles increase arterial contraction by promoting endothelial oxidative stress, an effect prevented by simvastatin.

Abstract:

Arterial cardiovascular events are the leading cause of death in patients with JAK2V617F

myeloproliferative neoplasms. Their mechanisms are poorly understood. A significant

proportion of these events occur in the absence of atherosclerosis. JAK2V617F is present in

myeloid and endothelial cells. Consequences of JAK2V617F on vascular reactivity are

unknown. Using myography experiments, we demonstrated that the aorta of mice carrying

Jak2V617F in myeloid cells displayed a strong increase in arterial contraction in response to

vasocontrictive agents. This effect required the presence of endothelial cells, but not of

Jak2V617F in endothelial cells. This effect was not observed in mice with polyglobulia

induced by epoietin. This increased arterial contraction was reproduced by circulating

microvesicles isolated from patients carrying JAK2V617F. We tested different subpopulations

of microvesicles and only microvesicles derived from Jak2V617F red blood cells induced this

increased arterial contraction. This effect implicated nitric oxide. We observed that the

endothelium of the aorta of mice with Jak2V617F myeloproliferative neoplasms displayed a

high level of oxidative stress. We thus treated these mice with N-acetyl-cysteine and

showed normalization of arterial contraction. These data prompted us to test simvastatin,

which has antioxidant effects, in mice with Jak2V617F myeloproliferative neoplasms and we

observed a strong improvement in arterial contraction. In conclusion, our results show that

JAK2V617F myeloproliferative neoplasms induce a potent increase in arterial contraction,

85 mediated by red blood cells microvesicles and endothelial oxidative stress. This could

account for the high frequency of arterial events associated with JAK2V617F myeloproliferative neoplasms. Simvastatin appears as promising strategy in this setting.

(250/250)

86 [Main Text: ]

Introduction

Bcr/Abl-negative myeloproliferative neoplasms (MPNs) are clonal hematopoietic stem cell disorders characterized by the proliferation of particular hematopoietic lineages without blockage in cell maturation. They include polycythaemia vera, essential thrombocythemia, and primary myelofibrosis (1, 2). JAK2 is the most common MPN driver gene. JAK2V617F is a gain of function mutation leading to growth factors hypersensitivity, detected in around 70% of MPNs (95% in polycythaemia vera and 50% to 60% in essential thrombocythemia and pre-primary myelofibrosis / primary myelofibrosis) (3). JAK2V617F appears in pluripotent hematopoietic progenitor cells and is present in all myeloid lineages

(2). In addition, several independent groups described JAK2V617F in endothelial cells in the liver and the spleen of patients with splanchnic vein thrombosis (4, 5) and in circulating endothelial progenitor cells (6–8).

Cardiovascular diseases (CVD) reveal MPNs in about 30% of the patients and are the first cause of morbidity and mortality in these patients (9). Arterial events represent 60-

70% of these cardio-vascular events (10–13). Interestingly, myocardial infarction without significant coronary stenosis by angiography was observed in 21% of patients with MNP

(14) versus only 3% in a similar population without MPN (15). This observation prompted the European society of cardiology to recommend searching for MPNs in case of myocardial infarction without obstructive coronary artery disease (16). The mechanism underlying this link between myocardial infarction without obstructive coronary artery disease and

MPNs is unknown, but vasoactive phenomenon (local intense vasoconstriction) can be suspected (17, 18).

87 Therefore, the purpose of the present study was to examine the consequences of the

JAK2V617F on arterial vascular reactivity.

Results

Increased arterial contraction in mice with Jak2V617F both in myeloid and endothelial cells

As JAK2V617F is present in both myeloid and endothelial cells in patients with MPN (4,

5), we first investigated vasoactive response in a mouse model mimicking the human

disease. We generated mice expressing Jak2V617F in myeloid and in endothelial cells by

crossing Jak2V617F Flex/WT mice with VE-Cadherin-cre mice. VE-Cadherin being expressed

during early embryonic life in a precursor of both endothelial and myeloid cells (19),

Jak2V617F Flex/WT;VE-Cadherin-cre, thereafter referred to as Jak2V617F BM-EC, developed as

expected a MPN, attested by higher spleen weight (2.3 to 5.7 % of body weight vs. 0.3 to 0.6

% for littermate controls, p<0.0001), higher haemoglobin level, platelet and white blood

cell counts than littermate controls (Figure 1A to D). The endothelial recombination was

verified by crossing VE-Cadherin-cre with mTmG mice (Fig. S1).

By performing myography assay, we observed that aortas from Jak2V617F BM-EC mice

displayed a major increased response to phenylephrine (Figure 1E), but also to potassium

chloride (Figure 1F) and angiotensin II (Figure 1G), as compared with littermate controls.

Removing endothelium suppressed this increased arterial contraction (Figure 1H). Thus,

Jak2V617F in endothelial and myeloid cells strongly increases response to vasoconstrictors in

an endothelial-dependent manner.

88 ABCD *** 25 4000 40 WT V617F *** ***

Jak2 Jak2 /L) 9 BM-EC 20 /L)

9 3000 30

15 1cm 2000 20 10

1000 10

5 Platelets count (10 Haemoglobin level (g/dL) White blood cells count (10 0 0 0 Jak2wt Jak2V617F Jak2wt Jak2V617F Jak2wt Jak2V617F BM-EC BM-EC BM-EC EFGH

wt 20 20 2.0 20 Jak2 wt wt Jak2 Jak2 Jak2V617F BM-EC * ns Jak2V617F BM-EC Jak2V617F BM-EC *** 1.5 15 15 *** *** 15 *** *** 10 *** *** 10 1.0 10

*** Contraction (mN) Contraction (mN) Contraction (mN) 5 5 Contraction (mN) 5 0.5 * * * 0 0 0.0 0 -9 -8 -7 -6 -5 -4 10-9 10-8 10-7 10-6 10-5 10-4 Jak2wt Jak2V617F 10-9 10-8 10-7 10-6 10 10 10 10 10 10 Log (Phenylephrine mol/L) BM-EC Log (Angiotensin II mol/L) Log (Phenylephrine mol/L) Fig. 1. Increased arterial contraction in a Jak2V617F myeloproliferative mouse model Representative picture of the spleen (A). Haemoglobin (B), platelet (C) and white blood cell count (D) of 8 to 12 weeks old control mice (Jak2WT; n=13) and Jak2V617F Flex/WT ;VE-Cadherin-cre mice (Jak2V617F BM-EC; n=13). Data are expressed as median with interquartile range. Cumulative dose-response curves to phenylephrine (E) (Jak2WT, n=13; Jak2V617F BM-EC, n=13), to angiotensin II (G) (Jak2WT, n=3; Jak2V617F BM-EC, n=4) and contraction response to potassium chloride (80 mmol/L) (F) (Jak2WT, n=13; Jak2V617F BM-EC, n=13) of aortas with endothelium. Cumulative dose-response curves to phenylephrine of aortas without endothelium (H) (Jak2WT, n=6; Jak2V617F BM-EC, n=6). Data are expressed as mean with standard error of the mean. Jak2WT mice are in blue and Jak2V617F BM-EC mice in red. Abbreviations: ns, not significant; * p<0.05, *** p< 0.001. Increased arterial contraction in mice with Jak2V617F restricted to myeloid cells but not to endothelial cells

To determine if this increased arterial contraction was due to Jak2V617F in endothelial cells or in myeloid cells, we first generated mice expressing Jak2V617F only in

endothelial cells. We crossed Jak2V617F Flex/WT mice with inducible VE-Cadherin-cre-ERT2 mice expressing the cre recombinase after tamoxifen injection only in endothelial cells. As

expected, Jak2V617F Flex/WT; VE-Cadherin-cre-ERT2 (thereafter referred to as Jak2V617F EC) mice

did not develop MPN (Figure 2A-D). Endothelial recombination was verified by crossing

VE-Cadherin-cre-ERT2 mice with mTmG mice (Fig. S1). We observed no difference in the arterial response to phenylephrine between the Jak2V617F EC mice and littermate controls

(Jak2WT ) (Figure 2E).

To assess the implication of Jak2V617F in myeloid cells, we generated mice expressing

Jak2V617F only in myeloid cells, by transplanting lethally irradiated C57BL/6 mice with a

Jak2V617F bone marrow obtained from Jak2V617F BM-EC mice. Irradiated C56BL/6 mice transplanted with Jak2WT BM were used as controls. Myeloid expression of Jak2V617F induced a MPN (Figure 2F to I) and an increased arterial response to phenylephrine

(Figure 2J). Taken altogether, these findings demonstrate that Jak2V617F in myeloid, but not in endothelial cells, is responsible for an increase arterial contraction. ABCDE 25 2000 20 ns ns * 15

/L) wt

9 Jak2 V617F

/L) Jak2 EC

9 15 JAK2WT JAK2V617F 20 1500 EC 10 ns 1000 10 15 1cm 5

500 5 Contraction (mN)

10 Platelets count (10 Haemoglobin level (g/dL) White blood cells count (10 0 0 0 Jak2wt Jak2V617F Jak2wt Jak2V617F Jak2wt Jak2V617F 10-9 10-8 10-7 10-6 10-5 10-4 EC EC EC Log (Phenylephrine mol/L)

FGHIJns 20 25 2000 ** V617F ** Jak2 BM * /L)

9 15 * Jak2wt * Jak2WT Jak2V617F 20 * /L)

9 1500 15 * BM * 15 10 1000 10 1cm 10 5 *

500 5 Contraction (mN)

5 Platelets count (10 Haemoglobin level (g/dL) White blood cells count (10 0 0 0 0 Jak2wt Jak2V617F Jak2wt Jak2V617F Jak2wt Jak2V617F 10-9 10-8 10-7 10-6 10-5 10-4 BM BM BM Log (Phenylephrine mol/L)

Fig. 2. Jak2V617F specifically expressed in endothelial cells (A-E) or in myeloid cells (F-J) Representative picture of the spleen (A and F). Blood cell count of 10 to 13 week old control mice (Jak2WT, n=7, blue) and Jak2V617F Flex/WT; VE-Cadherin-cre-ERT2 mice (Jak2V617F EC, n=7, red) (B to D) and of 13 to 15 week old chimeric C57BL/6 mice transplanted with a bone marrow of wild-type mice (Jak2WT, n=5, blue) or of Jak2V617F BM-EC mice (Jak2V617F BM, n=5, red) (G) (H) (I). Data are expressed as median with interquartile range. Cumulative dose-response curve to phenylephrine of aortas from Jak2WT (n=7, blue) and Jak2V617F EC (n=7, red) (E) and from Jak2WT (n=5, blue) and Jak2V617F BM (n=5, red) (J). Data are expressed as mean and standard error of the mean. Abbreviations: ns, not significant; * p<0.05, ** p< 0.01. Increased arterial contraction induced by microvesicles from JAK2V617F patients

We then sought to identify the mediators responsible for the increased response to vasoconstrictors when Jak2V617F was present on myeloid cells and tested the hypothesis that circulating blood might convey biological information from myeloid cells to the vascular wall. Circulating microvesicles, i.e. extracellular vesicles having a size ranging from

0.1 to 1 μm, are now recognized as triggers of various types of vascular dysfunction (20). We therefore examined the effect of circulating microvesicles isolated from the blood of patients with MPN on vascular responses to vasoactive agents. We isolated plasma microvesicles from 7 patients carrying Jak2V617F , (2 males, 5 females; blood drawn before introduction of cytoreductive therapy), and from 5 healthy controls (2 males, 3 females; age not significantly different from patients) We incubated these microvesicles at their plasma concentration with aortic rings from wild type mice and observed that plasma microvesicles from patients carrying Jak2V617F reproduced the increased response to phenylephrine (Figure 3A).

Increased arterial contraction induced by red blood cell, but not by platelet or leukocyte derived microvesicles from Jak2V617F mice

We then sought to determine the subpopulation of microvesicles responsible for this increased arterial contraction. We generated microvesicles from each type of blood cells from Jak2V617F BM-EC mice or littermate controls and incubated these microvesicles, at the same concentration, with aortic rings from wild type mice. As shown in Figure 3, red blood cell derived microvesicles generated from Jak2V617F BM-EC mice reproduced the increased response to phenylephrine while platelet, peripheral blood mononuclear cell and

polynuclear neutrophil microvesicles did not (Figure 3B-E).

To determine whether the increased number of red blood cells could in itself explain

this effect, we generated a mouse model of polycythaemia without Jak2V617F, by epoietin

injections. After 3 weeks of epoietin treatment, haemoglobin reached a level similar to that

of Jak2V617F BM-EC mice (18.5 g/dL, interquartile range 16.5-19.5, vs. 17.6 g/dL interquartile range 15.7-19.7, respectively; n=5 and n=13 respectively; p=0.67). As shown in Figure 3F, this high number of red blood cells did not reproduce the increased response to phenylephrine observed in Jak2V617F BM-EC mice.

Thus, microvesicles derived from Jak2V617F red blood cells are responsible for this increased arterial contraction.

93 AB 20 20 ** ** ** ** ** 15 15 ** * ** ** ** ** ** * * ** 10 10 Contraction (mN) Contraction (mN) 5 5 MVs from patients MVs RBC Jak2V617F MVs from controls MVs RBC Jak2WT 0 0 10-9 10-8 10-7 10-6 10-5 10-4 10-9 10-8 10-7 10-6 10-5 10-4 Log (Phenylephrine mol/L) Log (Phenylephrine mol/L) CD E 20 20 20

ns ns 15 15 15 ns

10 10 10 Contraction (mN) Contraction (mN) 5 MVs Platelets Jak2WT 5 MVs PBMC Jak2WT Contraction (mN) 5 MVs PMNC Jak2WT MVs Platelets Jak2V617F MVs PBMC Jak2V617F MVs PMNC Jak2V617F 0 0 0 10-9 10-8 10-7 10-6 10-5 10-4 10-9 10-8 10-7 10-6 10-5 10-4 10-9 10-8 10-7 10-6 10-5 10-4 Log (Phenylephrine mol/L) Log (Phenylephrine mol/L) Log (Phenylephrine mol/L) F 20

C57BL/6 EPO 15 C57BL/6 Vehicle

ns 10

Contraction (mN) 5

0 10-9 10-8 10-7 10-6 10-5 10-4 Log (Phenylephrine mol/L)

Fig. 3. Myeloid cells derived microvesicles Cumulative dose-response curves to phenylephrine of aortas from WT mice incubated with microvesicles isolated from Jak2V617F patients (n=7, red) and controls (n=5, blue) at their circulating concentration (A), and with microvesicles generated from red blood cells (n=9 and n=4, respectively) (B), platelets (n=5 and n=5, respectively) (C), PBMC (n=5 and n=6, respectively) (D) and PMNC (n=5 and n=5) (E) from Jak2V617F BM-EC mice (Jak2V617F, red) or littermate control mice (WT, blue). Cumulative dose- response curve to phenylephrine of aortas from WT mice injected with vehicle (blue, n= 5) or epoietin (red, n=8) (F). Data are expressed as mean with standard error to the mean. Abbreviations: ns, not significant; ** p< 0.01; MVs, microvesicles; RBC, Red blood cells; PBMC, peripheral blood mononuclear cells; PMNC, polymorphonuclear cells; EPO, epoietin; WT, wild type.

NO pathway inhibition and endothelial increased oxidative stress status

We then investigated how microvesicles derived from Jak2V617F red blood cells

increase response to vasoconstricting agents. We tested first the nitric oxide (NO) pathway.

We observed that aortas from Jak2V617F BM-EC mice displayed an impaired dilatation to

acetylcholine (Figure 4A). This impaired dilatation capacity was not due to decreased

sensitivity to NO of vascular smooth muscle cells, since dilatation in response to a direct

NO-donor (SNAP) was not different between Jak2V617F BM-EC mice and littermate controls

(Figure 4B). We also observed that after pre-incubation with the NO synthase (NOS)

inhibitor L-Name, aortas from Jak2V617F BM-EC mice had a similar response to

phenylephrine as littermate controls (Figure 4C). Therefore, these results demonstrate that

the increased arterial contraction observed in Jak2V617F BM-EC mice results from a

dysfunctional endothelial NO pathway.

As reactive oxygen species can regulate NOS activity and NO bioavailability (21), we

assessed oxidative stress using a fluorogenic probe designed to reliably measure reactive

oxygen species in living cells (CellROX®). We observed that reactive oxygen species generation was 4 times higher in aorta endothelium from Jak2V617F BM-EC mice than in littermate controls (Figure 4D, 5E). Conversely, reactive oxygen species generation was normal in the aorta endothelium from Jak2V617F EC mice, expressing Jak2V617F only in endothelial cells (Figure 4F, G). To ascertain the implication of oxidative stress in the increased arterial contraction in Jak2V617F BM-EC mice, we treated these mice with N-

Acetyl-Cysteine for 14 days intraperitoneally. We observed that this anti-oxidative therapy had no effect on blood cell count and spleen weight (Figure 4H-K), but normalized arterial contraction to phenylephrine (Figure 4L). Altogether, these results show that the increased arterial contraction in Jak2V617F

BM-EC mice is induced by excessive oxidative stress in endothelial cells leading to a decreased availability of NO.

96 ABC 20 0 0 Jak2wt -10 -10 Jak2V617F BM-EC Jak2V617F BM-EC V617F ns -20 Jak2 BM-EC -20 Jak2wt 15 Jak2wt -30 -30 -40 -40 * -50 -50 10 -60 -60 ** -70 Relaxation (%) Relaxation (%) -70 * Contraction (mN) 5 * * -80 -80 -90 -90 * -100 ns -100 0 -10 -9 -8 -7 -6 -5 -9 -8 -7 -6 -5 -4 10-9 10-8 10-7 10-6 10-5 10-4 10 10 10 10 10 10 10 10 10 10 10 10 Log (Acetylcholine mol/L) Log (SNAP mol/L) Log (Phenylephrine mol/L) L-Name 10-4 mol/L

X 4 DE*** FG 0.003 0.003 Jak2WT Jak2V617F Jak2V617F Jak2WT EC Jak2V617F EC ns 0.002 0.002

0.001 0.001 Positive surface / total cells Positive surface / total cells Positive surface / total cells Positive surface / total cells

0.000 0.000 Jak2wt Jak2V617F Jak2wt Jak2V617F BM-EC EC

H ns IJKLns ns ** ** ns ** ** Jak2wt Vehicle 8 25 4000 40 20 V617F ** ns ** ns ** ns ** ns Jak2 BM-EC Vehicle /L) /L) wt 9 Jak2 NAC 20 Jak2V617F BM-EC NAC

6 /uL) 3000 30 15 9 * 15 *** 4 2000 20 10 10 ns ** 2 1000 10 Contraction (mN) 5 5 Platelets count (10 Haemoglobin level (g/dL) White blood cells count (10 Spleen weight/ body weight (%) Spleen weight/ body weight (%) 0 0 0 0 0 Jak2wt Jak2V617F Jak2wt Jak2V617F Jak2wt Jak2V617F Jak2wt Jak2V617F Jak2wt Jak2V617F Jak2wt Jak2V617F Jak2wt Jak2V617F Jak2wt Jak2V617F 10-9 10-8 10-7 10-6 10-5 10-4 Vehicle BM-EC NAC BM-EC Vehicle BM-EC NAC BM-EC Vehicle BM-EC NAC BM-EC Vehicle BM-EC NAC BM-EC Log (Phenylephrine mol/L) Vehicle NAC Vehicle NAC Vehicle NAC Vehicle NAC Fig. 4. Nitric oxide pathway and oxidative stress status. Cumulative dose-response curve of aortas from Jak2V617F Flex/WT ;VE-Cadherin-cre mice (Jak2V617F BM-EC mice in red) and littermate controls (Jak2WT in blue) to acetylcholine (n=11 and n=11, respectively) (A), to S-Nitroso-N-Acetylpenicillamine (SNAP) (n=5 and n=6, respectively) (B) and to phenylephrine after L-NAME incubation (n=11 and n=7, respectively) (C). Quantification of reactive oxygen species (ROS) generation (Red surface) per endothelial cell in control mice (Jak2WT) in blue and Jak2V617F BM-EC mice in red (D) and from control mice (Jak2WT in blue) and Jak2V617F Flex/WT; VE- Cadherin-cre-ERT2 mice (Jak2V617F EC in red) (F). Representative images of “en face” endothelial staining with CellRox® (Red fluorogenic probes for ROS generation) of aortas (E and G). Spleen weight/body weight ratio (H), haemoglobin level (I), platelet count (J) and white blood cell count (K) and cumulative dose-response curve to phenylephrine of aortas from Jak2WT mice treated with vehicle (n=6, blue) and NAC (n=5, purple) and from Jak2V617F BM-EC mice treated with vehicle (n=5, red) and with NAC (n=7, orange) (L). Data are expressed as mean with standard error of the mean for cumulative curve and median with interquartile range for spleen weight and blood cell count. Abbreviations: ns, not significant; NAC, N-Acetyl-cysteine; * p<0.05, ** p< 0.01, *** p< 0.001.

Statins as a potential new drug in myeloproliferative neoplasms

We then tested if available treatments for MPNs, namely hydroxyurea and ruxolitinib, affect this increased arterial contraction. In Jak2V617F BM-EC mice, hydroxyurea

for 10 consecutive days decreased spleen weight, haemoglobin level and white blood cells

(Figure 5A, B, D). However, platelet count was not affected by this short time hydroxyurea

treatment (Figure 5C). Hydroxyurea significantly improved contraction in response to

phenylephrine as compared to vehicle (Figure 5E).

We then treated Jak2V617F BM-EC mice with ruxolitinib for 21 consecutive days and

observed a significant decrease in the spleen weight and white blood cell count (Figure 5F,

I), but no effect on the haemoglobin level and on platelet count (Figure 5G, H). Ruxolitinib

had no effect on arterial response to phenylephrine (Figure 5J).

Beyond its lowering cholesterol effect, simvastatin also improves endothelial

function through NO pathway and by preventing oxidative stress damage (22, 23). Thus, we

tested its effect on arterial response to phenylephrine in Jak2V617F BM-EC mice. Fourteen

days of treatment with simvastatin did not change spleen weight, haemoglobin level or

platelet count (Figure 5K, 5L, 5M). There was only a slight decrease in white blood cells

count following simvastatin treatment (Figure 5N). Interestingly, simvastatin significantly

improved aortic response to phenylephrine as compared to vehicle (Figure 5O). A BCDE Jak2V617F BM-EC Vehicule 4 25 4000 30 25 ** * ns ** Jak2V617F BM-EC Hydroxyurea /L) 9 20 20 3 /L) 9 3000 20 * 15 15 2 2000 10 10 10

1 1000 Contraction (mN)

5 Platelets count (10 5 Haemoglobin level (g/dL) White blood cells count (10 Spleen weight/ body weight (%) Spleen weight/ body weight (%) 0 0 0 0 0 Jak2V617F Jak2V617F Jak2V617F Jak2V617F Jak2V617F Jak2V617F Jak2V617F Jak2V617F 10-9 10-8 10-7 10-6 10-5 10-4 BM-EC BM-EC BM-EC BM-EC BM-EC BM-EC BM-EC BM-EC Log (Phenylephrine mol/L) Vehicle HU Vehicle HU Vehicle HU Vehicle HU F GHIJ 25 V617F 4 * 24 4000 30 Jak2 BM-EC Ruxolitinib ns ns * V617F /L) 9 20 Jak2 BM-EC Vehicle ns

3 22 /uL) 3000 9 20 15

2 20 2000 10 10 Contraction (mN) 1 18 1000 5 Platelets count (10 Haemoglobin level (g/dL) White blood cells count (10 Spleen weight/ body weight (%) Spleen weight/ body weight (%) 0 16 0 0 0 Jak2V617F Jak2V617F Jak2V617F Jak2V617F Jak2V617F Jak2V617F Jak2V617F Jak2V617F 10-9 10-8 10-7 10-6 10-5 10-4 BM-EC BM-EC BM-EC BM-EC BM-EC BM-EC BM-EC BM-EC Log (Phenylephrine mol/L) Vehicle Ruxolitinib Vehicle Ruxolitinib Vehicle Ruxolitinib Vehicle Ruxolitinib KLMNO 25 JAK2V617F BM-EC Vehicle 4.0 ns ns 4000 ns 30 * 20 JAK2V617F BM-EC Simvastatin

/L) * 9 20 * /L)

3.5 9 3000 * 15 * * 20 * 15 * 3.0 2000 10 10 10 2.5 1000 Contraction (mN) 5 Platelets count (10 5 Haemoglobin level (g/dL) White blood cells count (10 Spleen weight/ body weight (%) Spleen weight/ body weight (%) 2.0 0 0 0 Jak2V617F Jak2V617F Jak2V617F Jak2V617F Jak2V617F Jak2V617F Jak2V617F Jak2V617F 0 -9 -8 -7 -6 -5 -4 BM-EC BM-EC BM-EC BM-EC BM-EC BM-EC BM-EC BM-EC 10 10 10 10 10 10 Vehicule Simvastatin Vehicule Simvastatin Vehicule Simvastatin Vehicule Simvastatin Log (Phenylephrine mol/L)

Fig. 5. Statins as a potential new drug in myeloproliferative neoplasms Spleen to body weight ratio (A), haemoglobin level (B), platelet (C) and white blood cell count (D) in Jak2V617F Flex/WT ;VE-Cadherin-cre mice (Jak2V617F BM-EC) treated with vehicle (Jak2V617F BM-EC vehicle, red, n=4) or with hydroxyurea (Jak2V617F BM-EC HU, orange, n=7). Spleen to body weight ratio (F), haemoglobin level (G), platelet (H) and white blood cell count (I) in Jak2V617F BM-EC mice treated with vehicle (Jak2V617F BM-EC vehicle, red, n=5) or with ruxolitinib (Jak2V617F BM-EC ruxolitinib, orange, n=4). Spleen to body weight ratio (K), haemoglobin level (L), platelet (M) and white blood cell count (O) in Jak2V617F BM-EC mice treated with vehicle (red, n=10) or with simvastatin (Jak2V617F BM-EC simvastatin, orange, n=7).Cumulative dose-response curves to phenylephrine of aortas from Jak2V617F BM-EC mice treated with vehicle or hydroxyurea (Jak2V617F BM-EC vehicle in red, n=4; and Jak2V617F BM-EC HU, in orange, n=7) (E), with vehicle or ruxolitinib (Jak2V617F BM-EC vehicle in red, n= 5; and Jak2V617F BM-EC Ruxolitinib in orange, n=4) (J), and with vehicle or simvastatin (Jak2V617F BM-EC vehicle in red, n=10; and Jak2V617F BM-EC simvastatin in orange, n=7) (O). Data are expressed as mean with standard error of the mean for cumulative curve and median with interquartile range for spleen weight and blood cell count. Abbreviations: ns, not significant; * p<0.05, ** p< 0.01, HU hydroxyurea. Discussion

This study demonstrated that JAK2V617F red blood cell derived microvesicles strongly increased arterial contraction in response to vasoconstrictive agents, possibly accounting for arterial events associated with MPNs. This effect was due to an augmented oxidative stress in arterial endothelium and a decreased availability of NO. Simvastatin, a drug with anti-oxidant properties, improved arterial contraction.

10 times higher incidence of arterial events in patients with polycythaemia vera than in the general population (25, 26) and the high prevalence of myocardial infarction without significant coronary stenosis by angiography in patients with MPN (14). Arterial spasm is an underdiagnosed phenomenon that can happen in patients without atherosclerosis, but is also favoured by underlying non-stenotic atherosclerotic plaques. This suggests that the effect we observed might not only account for myocardial infarction without significant coronary stenosis reported in patients with MPN, but might also more widely favour arterial events in patients with atherosclerotic plaques and MPN (17, 18). Moreover, arterial spasm not only occurs in coronary arteries, but also in brain arteries (16, 27–29). The mechanisms underlying arterial spasm are not completely elucidated, but arterial contraction plays a central role (16, 28–30), which is

concordant with our findings, since we observed a pronounced increase in contraction in

response to different vasoconstrictive agents. We found only a mild impairment in arterial

dilatation, which is in line with the altered endothelial dependant flow mediated vasodilatation

reported in patients with polycythaemia vera, in the absence of overt arterial disease (31).

The second major finding of our work is the identification of JAK2V617F red blood cell derived microvesicles as responsible for the increased arterial contraction associated with MPN.

Importantly, we observed this effect with JAK2V617F red blood cell microvesicles from mice, but also with microvesicles isolated from patients carrying JAK2V617F. We thus highlight here a crucial vascular role of microvesicles in MPNs, beyond their so far described implication in coagulation in this setting (32–36). Although patients with MPNs have higher circulating levels of microvesicles than healthy individuals, we assessed vascular reactivity using the same concentrations of microvesicles for both groups, suggesting that microvesicle composition, and not concentration, accounts for the observed vascular effect (32, 33, 37–40). A specificity of

JAK2V617F red blood cells derived microvesicles is also supported by our observation that

JAK2V617F platelets or white blood cells derived microvesicles did not increase arterial contraction, and that polycythaemia induced by epoietin without JAK2V617F had no effect on arterial contraction. These results are reminiscent of epidemiological studies showing that MPN patients with JAK2V617F have higher haematocrit level (41, 42) and a higher risk of cardiovascular events than MPN patients without JAK2V617F (42–46).

Finally, in our work we demonstrated that NO pathway inhibition and increased endothelial oxidative stress are implicated in this increased arterial contraction in MPN. Several groups reported high levels of circulating reactive oxygen species products (47–49) and low antioxidant status in MPN (48) (50), but endothelial oxidative stress had never been investigated.

Besides explaining how MPN induce this increased arterial contraction, this observation opens new potential therapeutic perspectives to prevent MPN. Indeed, statins being known to play a protective role on endothelial function and on oxidative stress, we tested this drug and observed a strong improvement in arterial response to vasocontricting agent in our MPN mouse model (22,

101 23). Simvastatin being a well-known and easily accessible drug, these results thus pave the way

for testing simvastatin to prevent arterial events in patients with MPN. We also tested available

treatments for MPN and observed that hydroxyurea, but not ruxolitinib, improved arterial

contraction. This difference might be explained by the fact that hydroxyurea decreased red blood

cell count in our mouse model whereas ruxolitinib did not (51, 52). Another explanation could be that hydroxyurea has been shown to enhance NO release by endothelial cells while such an effect has not been reported with ruxolitinib (53).

In conclusion, our study showed that red blood cell derived microvesicles are responsible for an increased arterial contraction in Jak2V617F MPNs. This effect implicated NO and endothelial oxidative stress. Simvastatin appears as an original new approach to prevent arterial events in MPN that warrants further studies.

Materials and Methods

Experimental design

The objective of our study was to analyse endothelial reactivity in MPN. We first noticed a major increase in arterial contraction in Jak2V617F BM-EC mice, a model with Jak2V617F expression both in myeloid and endothelial cells that mimics the human disease. We then created mouse models specifically mutated in endothelial or in myeloid cells. We then searched for the mediators responsible for the increased response to vasoconstrictors

when Jak2V617F was present in myeloid cells and tested the hypothesis that circulating blood might convey biological information from myeloid cells to the vascular wall. Sample size was chosen based on previous works using the same technique (myography) and microvesicles, published by our team (54–56).

102 Mouse breeding occurred in our animal facility in accordance with the local

recommendations. Control mice were littermate, appropriate, age-, sex and genetic background, matched to account for any variation in data. Institutional animal care and use committee at INSERM (Descartes university, Paris, France) approved all animal experiments (CEEA-17053).

Number of experimental replicates is provided in each figure legend and included at least 3

independent experiments. For each myography experiment, duplicates with the same aorta

were used, averaged and counted as n=1. There was no randomization in these

experiments. We did not exclude any other sample than those not fulfilling the quality

criteria detailed in the corresponding methods section. Only aortas with a viable

endothelium were used for myography (see corresponding methods section for criteria).

For human samples, inclusion and exclusion criteria were defined prior to sample

collection (see corresponding methods section for criteria). No outliers were excluded.

Investigators were not blinded to group allocation during collection and analysis of the

data. All patients (carrying JAK2V617F with a past history of splanchnic vein thrombosis, not receiving any specific treatments other than vitamin-k antagonists) and healthy volunteers gave writing consent to the study. Human study was performed in accordance with the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the institutional review board Bichat-Claude-Bernard (Paris; France).

Murine models

All mice were on a C57BL/6 background. Mice carrying constitutive JAK2V617F mutation in endothelial and myeloid cells were obtained by crossing VE-cadherin-Cre transgenic mice provided by M. Souyri (19) with Jak2V617F Flex/WT mice provided by M.

Villeval (57). Mice carrying inducible JAK2V617F mutation specifically in endothelial cells

103 were obtained by crossing VE-Cadherin-cre-ERT2 transgenic mice provided by R. Adams

(58) with Jak2V617F Flex/WT mice provided by M. Villeval (57). In all experiments, male and

female mice were used.

For organ chamber experiments, mice were euthanized between the ages of 8 and

17 weeks. For induction of Cre recombinase expression in Jak2V617F Flex/WT;VE-Cadherin-cre-

ERT2 mice, mice were injected intraperitoneally with tamoxifen (Sigma, T5648), 1

mg/mouse/day for 5 consecutive days over 2 consecutive weeks (10 mg in total per

mouse) between the ages of 5 to 7 weeks. Experiments were performed between 4 to 6

weeks after the last tamoxifen injection.

Experiments were conducted according to the French veterinary guidelines and

those formulated by the European community for experimental animal use (L358-

86/609EEC) and were approved by the French ministry of agriculture (n° A75-15-32).

Patient's inclusion

All patients fulfilling inclusion criteria were prospectively included at the

Hepatology department, Beaujon Hospital, Clichy, France, between May 2016 and July

2016. Only patients carrying JAK2V617F without specific treatment for MPNs were included.

All patients had a past history of Budd-Chiari syndrome or portal vein thrombosis and

were receiving vitamin K antagonists. Controls were healthy voluntaries. All patients and

controls gave written consent to the study. This study was performed in accordance with

the ethical guidelines of the 1975 Declaration of Helsinki and was approved by our

institutional review board (CPP Ile de France IV, Paris; France).

Organ Chamber Experiments

104 Thoracic aortas from adult mice were isolated after animal sacrifice under 2% isoflurane anaesthesia. The aortic rings were mounted in organ chambers (Multi

WireMyograph system, model 610 M; Danish Myo Technology, Aarhus, Denmark) filled with Krebs–Ringer solution (NaCl 118.3 mmol/L, KCl 4.7 mmol/L, MgSO4 1.2 mmol/L,

KH2PO4 1.2 mmol/L, CaCl2 1.25 mmol/L, NaHCO3 25.0 mmol/L and glucose 5.0 mmol/L).

The presence of functional endothelial cells was confirmed by the relaxation to

acetylcholine chloride (Sigma, A6625) (10-5 mol/L) following a contraction evoked by

phenylephrine (10-7 mol/L) and was defined as a relaxation ≥ 70% of the precontraction as

previously described (56, 59). After extensive washout and equilibration, contraction to

phenylephrine hydrochloride (concentration-response curve, 10−9 to 10−4 mol/L) (Sigma,

P1250000) or angiotensin II (concentration-response curve, 10−9 to 10−6 mol/L) (Sigma,

A9525) or KCL (80 mmol/L) and relaxation to acetylcholine chloride (concentration-

response curve, 10−9 to 10−4 mol/L) or SNAP (S-Nitroso-N-acetyl-DL-penicillamine, Sigma,

N3398) (concentration-response curve, 10−10 to 10−5 mol/L) were studied. For NO

synthase inhibition, aorta rings were preincubated for 45 min with L-NAME 10-4 mol/L

(Cayman, 80210) prior to concentration-response curve to phenylephrine without

washout. In some experiments, the endothelium was mechanically removed by inserting

the tip of a pair of forceps within the lumen and by gently rubbing the ring back and forth

on a piece of wet tissue. For the N-Acetyl-Cysteine experiment (NAC, commercial

HIDONAC®, Zambon), NAC was added to the Krebs-Ringer solution at a final concentration

of 20 mmol/L.

Isolation and characterization of patients’ circulating microvesicles

Circulating microvesicles from patients or healthy control were isolated from

platelet-free plasma obtained by successive centrifugations of venous blood, as reported

105 previously (60). Briefly, citrated venous blood (15 mL) was centrifuged twice at 2500g for

15 minutes (at room temperature) to remove cells and cell debris and to obtain platelet- free plasma (PFP). A portion of this PFP was then aliquoted and stored at -80°C. The rest was centrifuged at 20500g for 2 hours and 30 minutes (4°C). Supernatant of this 20500g centrifugation was then discarded and the resulting microvesicles pellet was resuspended in a minimal volume of supernatant, aliquoted and stored at -80°C. For each patient, concentrations of annexin V positive microvesicles were analysed in the PFP and the resuspended pellet of microvesicles.

Circulating levels of annexin V+ microvesicles were determined on a Gallios flow cytometer

(Beckman Coulter, Villepinte, France) using a technique previously described in detail (56,

60). Regions corresponding to microvesicles were identified in forward light scatter and side-angle light scatter intensity dot plot representation set at logarithmic gain. MV gate was defined, using calibration beads (Megamix plus FSC, Biocytex, France), as events having a 0.1 - 1 µm diameter. Events were then plotted on a fluorescence / forward light scatter dot plot to determine MV counts positively labelled by annexin V in presence of calcium. Annexin V fluoroisothiocyanate was purchased from Beckman-Coulter.

Microvesicles concentration was assessed by comparison with a known amount of flowcount calibrator beads (AccuCount Fluorescent Particles, Spherotech, Chicago, 20 µL) added to each sample just before performing flow cytometry analysis.

Generation of microvesicles from mice

Blood samples were collected from the inferior vena cava of Jak2V617F BM-EC mice or littermate controls using a 25 gauge x 1’ needle in a 1 mL syringe pre-coated with 3.8% sodium citrate. PFPs were generated as described above for patients and used to measure plasma annexin V positive microvesicles in mice. The pelleted cells obtained following the

106 first 2500g centrifugation were resuspended in PBS to a final volume of 5 mL for control

mice and 15 mL for Jak2V617F BM-EC mice. PBMC, PMNC and red blood cells were separated using a double percoll gradient (63% and 72% for control mice and 63% and 66% for

Jak2V617F BM-EC mice) using a 700g centrifugation for 25 min, without brake. The slight differences between the protocols used for control and Jak2V617F BM-EC mice are the results of the preliminary experiments we did to obtain pure isolation of each cell type despite the very different blood cell count between the two strains. Cells were subsequently washed with PBS, then incubated with 5 µmol/L ionomycin TBS for 2 hrs at 37°C to induce microvesicles generation. 5 mmol/L EDTA was then added to chelate free calcium. Cells were then discarded by centrifugations at 15000g for 1 min and the supernatants were collected. Microvesicles were isolated for myography assays, as described above using a

20500g centrifugation. Concentrations of annexin V positive microvesicles (as described above) were analysed in the PFP and the 20500g microvesicles pellet for each mouse.

To isolate platelets, 500 µL of whole blood were diluted in 10 mL of PBS. A 1.063 g/mL density barrier was created by combining 5 mL of 1.320 g/mL 60% iodixanol stock solution

(OptiPrep density gradient medium, Sigma-Aldrich, Saint Louis, MO, USA) with 22 mL of diluent (0.85% NaCl, 20 mM HEPES-NaOH, pH 7.4, 1 mM EDTA). For platelet separation, 10 mL of each diluted blood were layered over 10 mL of density barrier and centrifuged at 350 g for 15 minutes at 20°C with the brake turned off. The interface between the density barrier and the blood contained platelets. Residual contaminating red blood cells were removed by magnetic sorting. Briefly, the cell suspension was labelled with Anti-Ter-119

MicroBeads (Miltenyi Biotec ref 130-049-901) and red blood cell (Ter-119+) were

negatively sorted using a MACS® Separator. The remaining cells (platelets) were

subsequently washed with PBS and exposed to 5 μmol/L ionomycin in TBS for 30 minutes

at 37°C. 5 mM EDTA was then added to chelate free calcium. Finally, cells were discarded

107 by centrifugation at 15000g for 1 minute, the supernatant was collected and microvesicles

isolated as previously described.

Vascular reactivity following exposure to microvesicles

Thoracic aortas from adult C57BL/6 mice (8 to 10 week-old) were isolated after

sacrifice under isoflurane anaesthesia. Mouse aortic rings were incubated for 24 hrs; 37°C

in a 5% CO2 incubator, with filtered DMEM supplemented with antibiotics (100 IU/mL

streptomycin, 100 IU/mL penicillin (Gibco, Invitrogen, Paisley, Scotland), and 10 μg/mL

polymyxin B (Sigma, St Louis, MO) in the presence of microvesicles. Microvesicles from

patients and healthy controls were incubated at their respective individual plasma

concentration (Annexin-V positive microvesicles). Microvesicles generated from mice were

incubated at the same concentration for Jak2V617F BM-EC mice and control mice, namely

7000 Annexin V positive microvesicles / μL for red blood cell and platelets derived microvesicles and 700 Annexin V positive microvesicles / μL for PBMC and PMNC derived microvesicles. We chose these concentrations since we found in preliminary experiments that the majority of mice have concentrations of Annexin V positive microvesicles between

1000 and 10000 / μL, and because PBMC and PMNC derived microvesicles are consistently found less abundant in the blood than red blood cell and platelets derived microvesicles

(32, 37). Aortic rings were then mounted in organ chambers and concentration-response curves to pharmacological agents were performed.

Bone marrow transplantation

We subjected 6 to 8 week-old C57Bl/6J mice to medullar aplasia following 9.5 gray

lethal total body irradiation. We repopulated the mice with an intravenous injection of

bone marrow cells isolated from femurs and tibias of age matched Jak2V617F BM-EC and of

108 littermate control mice. Medullar reconstitution was allowed for 8 weeks before

experiments.

Treatments

Hydroxyurea (Sigma, H8627) or the same volume of vehicle (NaCl 0.9%), was

administrated for 10 consecutive days (100 mg/kg/day BID) by intra-peritoneal injections

Ruxolitinib (Novartis) was administered for 21 consecutive days (30 mg/kg 2 times per day) by oral gavage (61). Ruxolitinib was prepared from 15-mg commercial tablets in

PEG300/5% dextrose mixed at a 1:3 ratio, as previously reported (62). Control mice were

administered the same volume of vehicle (PEG300/dextrose 5%).

Simvastatin (Sigma S6196) was administered for 14 days (20 mg/kg/day, once a

day) by intra-peritoneal injections. Activation by hydrolysis was first achieved by

dissolving 50 mg in 1 mL of pure ethanol and adding 0.813 ml of 1 mol/L NaOH. pH was

adjusted to 7.2 by adding small quantities of 1 mol/L HCl and dilution was then performed

in PBS (63). Control mice were injected with the same volume of vehicle.

Human recombinant Epoietin alfa (5000 UI/kg, diluted in 0.2% BSA in PBS) or

vehicle (0.2% BSA in PBS) was administered to wild type mice every 2 days for 3 weeks by

intraperitoneal injection, as previously described (64).

N-Acetyl-Cysteine (commercial HIDONAC, Zambon) diluted in NaCl 0.9% or the

same volume of vehicle (NaCl 0.9%), was administrated for 14 consecutive days (500

mg/kg/day) by intraperitoneal injections.

Blood Cell count analysis

109 Blood was collected on the day of sacrifice from the inferior vena cava using a 25 gauge x 1’

needle in a 1 mL syringe pre-coated with 3.8% sodium citrate. Blood counts analyses were

performed using a Hemavet 950FS analyser (Drew scientific).

Quantification of reactive oxygen species generation

Thoracic aortas from adult mice were isolated after animal sacrifice under 2% isoflurane anaesthesia, longitudinally opened and placed directly in HBSS (Hanks' balanced salt solution, Sigma, 14025-092). For positive and negative controls, 2 pieces of wild type aortas were incubated with H2O2 (100 µmol/L final concentration) for 20 min at 37°C. For negative controls, N-Acetyl-Cysteine (5 mmol/L final concentration) was incubated together with H2O2 for 20 min at 37°C. All aortas were then incubated with 5 µmol/L

CellROX® (Fisher scientific, C10422) for 30 min at 37°C. CellROX® Deep Red Reagent is a fluorogenic probe designed to reliably measure reactive oxygen species in living cells. The cell-permeable CellROX® Deep Red dye is nonfluorescent while in a reduced state and, upon oxidation, exhibits excitation/emission maxima at 640/665 nm. After rinsing and fixation (Paraformaldehyde 4%, 20 min), samples were costained with DAPI (0.1 μg/mL,

Sigma) in order to identify cell nuclei. After staining, aortas were washed with PBS, mounted “en face” on glass slides and imaged using a bright field Zeiss Axio Imager Z1

(Zeiss) microscope. Images were acquired in the 2 hours following staining at 400 X magnification. CellROX® positive surface (in red) and the number of cells were quantified using Image J Software.

Statistics

For cumulative dose response curves, data were expressed as mean with standard error of the mean and compared using an analysis of variance for repeated measures. Other

110 data were expressed as median with interquartile range (blood cell count and spleen weight) and compared using the Mann-Whitney U-test. All tests were 2 sided and used a significance level of 0.05. Data handling and analysis were performed with GraphPad

Softwar, Inc.

111 Supplementary Materials

Materials and Methods

Verification of the efficient endothelial recombination in mouse models

All mice were on a C57BL/6 background. Mice with the mTmG reporter provided by C.

James (Inserm 1034) were crossed with VE-cadherin-Cre transgenic mice provided by M.

Souyri (49) or VE-Cadherin-cre-ERT2 transgenic mice provided by R. Adams (58). For

induction of mTmG;VE-Cadherin-cre-ERT2 model, mice were injected intraperitoneally with tamoxifen (Sigma, T5648), 1 mg/mice/day for 5 consecutive days over 2 consecutive

weeks (10 mg in total per mice) between the ages of 5 to 7 weeks, and experiments were performed 2 weeks after the last injection of tamoxifen. Aortas were harvested under isoflurane anaesthesia, mounted “en face” on glass slides and imaged using a Leica SP5 confocal microscope (Leica) at 400 X magnification.

112

VE-Cadherin-Cre-ERT2 VE-Cadherin-Cre A B C D 100 R26R td-tomato Gfp ✗ stop Active Non active VE-Cadherin-Cre VE-Cadherin-Cre or or Cadherin5-Cre-ERT2 Cadherin5-Cre-ERT2 50

R26R Gfp R26R td-tomato Gfp ✗ stop

Green fluorescence Red fluorescence Endothelial recombination (%) 10μm 10μm 0 = Recombination = No recombination VECre VECre ERT2

Fig. S1. Endothelial recombination. Mechanism of mTmG reporter strategy (A). Representative “en face” images of the endothelium of aortas from inducible mTmG;VE- Cadherin-Cre-ERT2 (n=5) mice (B) and from constitutive mTmG;VE-Cadherin-Cre (n=3) mice (C). Quantification (D). Abbreviations:VECreERT2, mTmG;VE-Cadherin-Cre-ERT2; VECre, mTmG;VE-Cadherin-Cre,.

References and Notes:

1. J. L. Spivak, D. L. Longo, Ed. Myeloproliferative Neoplasms, N. Engl. J. Med. 376, 2168– 2181 (2017).

2. W. Vainchenker, R. Kralovics, Genetic basis and molecular pathophysiology of classical myeloproliferative neoplasms, Blood 129, 667–679 (2017).

3. P. J. Campbell, A. R. Green, The myeloproliferative disorders, N. Engl. J. Med. 355, 2452– 2466 (2006).

4. S. Sozer, M. I. Fiel, T. Schiano, M. Xu, J. Mascarenhas, R. Hoffman, The presence of JAK2V617F mutation in the liver endothelial cells of patients with Budd-Chiari syndrome, Blood 113, 5246–5249 (2009).

5. V. Rosti, L. Villani, R. Riboni, V. Poletto, E. Bonetti, L. Tozzi, G. Bergamaschi, P. Catarsi, E. Dallera, F. Novara, M. Massa, R. Campanelli, G. Fois, B. Peruzzi, M. Lucioni, P. Guglielmelli, A. Pancrazzi, G. Fiandrino, O. Zuffardi, U. Magrini, M. Paulli, A. M. Vannucchi, G. Barosi, Associazione Italiana per la Ricerca sul Cancro Gruppo Italiano Malattie Mieloproliferative (AGIMM) investigators, Spleen endothelial cells from patients with myelofibrosis harbor the JAK2V617F mutation, Blood 121, 360–368 (2013).

6. V. Rosti, E. Bonetti, G. Bergamaschi, R. Campanelli, P. Guglielmelli, M. Maestri, U. Magrini, M. Massa, C. Tinelli, G. Viarengo, L. Villani, M. Primignani, A. M. Vannucchi, F. Frassoni, G. Barosi, on behalf of the A. Investigators, High Frequency of Endothelial Colony Forming Cells Marks a Non-Active Myeloproliferative Neoplasm with High Risk of Splanchnic Vein Thrombosis, PLOS ONE 5, e15277 (2010).

7. Helman Ricardo, Pereira Welbert de O., Marti Luciana C., Campregher Paulo V., Puga Renato Davi, Hamerschlak Nelson, Chiattone Carlos S., Santos Fabio Pires de S., Granulocyte whole exome sequencing and endothelial JAK2V617F in patients with JAK2V617F positive Budd‐ Chiari Syndrome without myeloproliferative neoplasm, Br. J. Haematol. 180, 443–445 (2018).

8. L. Teofili, M. Martini, M. G. Iachininoto, S. Capodimonti, E. R. Nuzzolo, L. Torti, T. Cenci, L. M. Larocca, G. Leone, Endothelial progenitor cells are clonal and exhibit the JAK2(V617F) mutation in a subset of thrombotic patients with Ph-negative myeloproliferative neoplasms, Blood 117, 2700–2707 (2011).

9. R. Marchioli, G. Finazzi, R. Landolfi, J. Kutti, H. Gisslinger, C. Patrono, R. Marilus, A. Villegas, G. Tognoni, T. Barbui, Vascular and Neoplastic Risk in a Large Cohort of Patients With Polycythemia Vera, J. Clin. Oncol. 23, 2224–2232 (2005).

10. P. J. Campbell, L. M. Scott, G. Buck, K. Wheatley, C. L. East, J. T. Marsden, A. Duffy, E. M. Boyd, A. J. Bench, M. A. Scott, G. S. Vassiliou, D. W. Milligan, S. R. Smith, W. N. Erber, D. Bareford, B. S. Wilkins, J. T. Reilly, C. N. Harrison, A. R. Green, Definition of subtypes of essential thrombocythaemia and relation to polycythaemia vera based on JAK2 V617F mutation status: a prospective study, The Lancet 366, 1945–1953 (2005). 11. Buxhofer‐Ausch Veronika, Gisslinger Heinz, Thiele Jürgen, Gisslinger Bettina, Kvasnicka Hans‐Michael, Müllauer Leonhard, Frantal Sophie, Carobbio Alessandra, Passamonti Francesco, Rumi Elisa, Ruggeri Marco, Rodeghiero Francesco, Randi Maria L., Bertozzi Irene, Vannucchi Alessandro M., Antonioli Elisabetta, Finazzi Guido, Gangat Naseema, Tefferi Ayalew, Barbui Tiziano, Leukocytosis as an important risk factor for arterial thrombosis in WHO‐defined early/prefibrotic myelofibrosis: An international study of 264 patients, Am. J. Hematol. 87, 669– 672 (2012).

12. Montanaro Marco, Latagliata Roberto, Cedrone Michele, Spadea Antonio, Rago Angela, Giandomenico Jonny, Spirito Francesca, Porrini Raffaele, Muro Marianna, Leonetti Sabrina Crescenzi, Villivà Nicoletta, Gregoris Cinzia, Breccia Massimo, Montefusco Enrico, Santoro Cristina, Cimino Giuseppe, Majolino Ignazio, Mazzucconi Maria Gabriella, Alimena Giuliana, Andriani Alesssandro, Thrombosis and survival in essential thrombocythemia: A regional study of 1,144 patients, Am. J. Hematol. 89, 542–546 (2014).

13. V. D. Stefano, I. Martinelli, Splanchnic vein thrombosis: clinical presentation, risk factors and treatment, Intern. Emerg. Med. 5, 487–494 (2010).

14. É. Pósfai, I. Marton, Z. Borbényi, A. Nemes, Myocardial infarction as a thrombotic complication of essential thrombocythemia and polycythemia vera, Anatol. J. Cardiol. 16, 397– 402 (2016).

15. A. I. Larsen, P. D. Galbraith, W. A. Ghali, C. M. Norris, M. M. Graham, M. L. Knudtson, Characteristics and outcomes of patients with acute myocardial infarction and angiographically normal coronary arteries, Am. J. Cardiol. 95, 261–263 (2005).

16. S. Agewall, J. F. Beltrame, H. R. Reynolds, A. Niessner, G. Rosano, A. L. P. Caforio, R. De Caterina, M. Zimarino, M. Roffi, K. Kjeldsen, D. Atar, J. C. Kaski, U. Sechtem, P. Tornvall, ESC working group position paper on myocardial infarction with non-obstructive coronary arteries, Eur. Heart J. 38, 143–153 (2017).

17. M. J. Davies, The pathophysiology of acute coronary syndromes, Heart 83, 361–366 (2000).

18. F. Crea, P. Libby, Acute Coronary Syndromes: The Way Forward From Mechanisms to Precision Treatment, Circulation 136, 1155–1166 (2017).

19. E. Oberlin, B. El Hafny, L. Petit-Cocault, M. Souyri, Definitive human and mouse hematopoiesis originates from the embryonic endothelium: a new class of HSCs based on VE- cadherin expression, Int. J. Dev. Biol. 54, 1165–1173 (2010).

20. C. M. Boulanger, X. Loyer, P.-E. Rautou, N. Amabile, Extracellular vesicles in coronary artery disease, Nat. Rev. Cardiol. 14, 259–272 (2017).

21. P. M. Vanhoutte, Y. Zhao, A. Xu, S. W. S. Leung, Thirty Years of Saying NO: Sources, Fate, Actions, and Misfortunes of the Endothelium-Derived Vasodilator Mediator, Circ. Res. 119, 375–396 (2016).

22. S. Sen-Banerjee, S. Mir, Z. Lin, A. Hamik, G. B. Atkins, H. Das, P. Banerjee, A. Kumar, M. K. Jain, Kruppel-Like Factor 2 as a Novel Mediator of Statin Effects in Endothelial Cells, Circulation 112, 720–726 (2005).

31 23. R. P. Mason, M. F. Walter, R. F. Jacob, Effects of HMG-CoA reductase inhibitors on endothelial function: role of microdomains and oxidative stress, Circulation 109, II34-41 (2004).

24. T. Barbui, G. Finazzi, A. Falanga, Myeloproliferative neoplasms and thrombosis, Blood 122, 2176–2184 (2013).

25. R. Landolfi, L. Di Gennaro, T. Barbui, V. De Stefano, G. Finazzi, R. Marfisi, G. Tognoni, R. Marchioli, European Collaboration on Low-Dose Aspirin in Polycythemia Vera (ECLAP), Leukocytosis as a major thrombotic risk factor in patients with polycythemia vera, Blood 109, 2446–2452 (2007).

26. B. M. Kaess, S. R. Preis, A. Beiser, D. B. Sawyer, T. C. Chen, S. Seshadri, R. S. Vasan, Circulating vascular endothelial growth factor and the risk of cardiovascular events, Heart Br. Card. Soc. 102, 1898–1901 (2016).

27. O. Y. Bang, K. Toyoda, J. F. Arenillas, L. Liu, J. S. Kim, Intracranial Large Artery Disease of Non-Atherosclerotic Origin: Recent Progress and Clinical Implications, J. Stroke 20, 208–217 (2018).

28. C. N. Bairey Merz, C. J. Pepine, M. N. Walsh, J. L. Fleg, Ischemia and No Obstructive Coronary Artery Disease (INOCA): Developing Evidence-Based Therapies and Research Agenda for the Next Decade, Circulation 135, 1075–1092 (2017).

29. J. F. Beltrame, F. Crea, J. C. Kaski, H. Ogawa, P. Ong, U. Sechtem, H. Shimokawa, C. N. Bairey Merz, Coronary Vasomotion Disorders International Study Group (COVADIS), The Who, What, Why, When, How and Where of Vasospastic Angina, Circ. J. Off. J. Jpn. Circ. Soc. 80, 289–298 (2016).

30. P. Ong, A. Aziz, H. S. Hansen, E. Prescott, A. Athanasiadis, U. Sechtem, Structural and Functional Coronary Artery Abnormalities in Patients With Vasospastic Angina Pectoris, Circ. J. Off. J. Jpn. Circ. Soc. 79, 1431–1438 (2015).

31. T. Neunteufl, S. Heher, T. Stefenelli, I. Pabinger, H. Gisslinger, Endothelial dysfunction in patients with polycythaemia vera, Br. J. Haematol. 115, 354–359 (2001).

32. A. Charpentier, A. Lebreton, A. Rauch, A. Bauters, N. Trillot, O. Nibourel, V. Tintillier, M. Wemeau, J.-L. Demory, C. Preudhomme, B. Jude, T. Lecompte, N. Cambier, S. Susen, Microparticle phenotypes are associated with driver mutations and distinct thrombotic risks in essential thrombocythemia, Haematologica 101, e365-368 (2016).

33. X. Tan, J. Shi, Y. Fu, C. Gao, X. Yang, J. Li, W. Wang, J. Hou, H. Li, J. Zhou, Role of erythrocytes and platelets in the hypercoagulable status in polycythemia vera through phosphatidylserine exposure and microparticle generation, Thromb. Haemost. 109, 1025–1032 (2013).

34. H. Baccouche, M. Ben Jemaa, A. Chakroun, S. Chadi, S. Mahjoub, I. Sfar, Y. Gorgi, N. Ben Romdhane, The evaluation of the relevance of thrombin generation and procoagulant activity in thrombotic risk assessment in BCR-ABL-negative myeloproliferative neoplasm patients, Int. J. Lab. Hematol. 39, 502–507 (2017).

32 35. J. Duchemin, V. Ugo, J.-C. Ianotto, L. Lecucq, B. Mercier, J.-F. Abgrall, Increased circulating procoagulant activity and thrombin generation in patients with myeloproliferative neoplasms, Thromb. Res. 126, 238–242 (2010).

36. M. Marchetti, C. J. Tartari, L. Russo, M. Panova-Noeva, A. Leuzzi, A. Rambaldi, G. Finazzi, B. Woodhams, A. Falanga, Phospholipid-dependent procoagulant activity is highly expressed by circulating microparticles in patients with essential thrombocythemia, Am. J. Hematol. 89, 68–73 (2014).

37. M. C. Trappenburg, M. van Schilfgaarde, M. Marchetti, H. M. Spronk, H. ten Cate, A. Leyte, W. E. Terpstra, A. Falanga, Elevated procoagulant microparticles expressing endothelial and platelet markers in essential thrombocythemia, Haematologica 94, 911–918 (2009).

38. W. Zhang, J. Qi, S. Zhao, W. Shen, L. Dai, W. Han, M. Huang, Z. Wang, C. Ruan, D. Wu, Y. Han, Clinical significance of circulating microparticles in Ph− myeloproliferative neoplasms, Oncol. Lett. 14, 2531–2536 (2017).

39. M.-P. Moles-Moreau, C. Ternisien, A. Tanguy-Schmidt, F. Boyer, M. Gardembas, M. Dib, A. Ponthieux, P. Guardiola, N. Ifrah, M. Hunault-Berger, Flow cytometry-evaluated platelet CD36 expression, reticulated platelets and platelet microparticles in essential thrombocythaemia and secondary thrombocytosis, Thromb. Res. 126, e394-396 (2010).

40. J. Kissova, P. Ovesna, A. Bulikova, J. Zavřelova, M. Penka, Increasing procoagulant activity of circulating microparticles in patients with Philadelphia-negative myeloproliferative neoplasms: a single-centre experience, Blood Coagul. Fibrinolysis Int. J. Haemost. Thromb. 26, 448–453 (2015).

41. E. Rumi, D. Pietra, V. Ferretti, T. Klampfl, A. S. Harutyunyan, J. D. Milosevic, N. C. Them, T. Berg, C. Elena, I. C. Casetti, JAK2 or CALR mutation status defines subtypes of essential thrombocythemia with substantially different clinical course and outcomes, Blood 123, 1544– 1551 (2014).

42. A. Tefferi, E. A. Wassie, T. L. Lasho, C. Finke, A. A. Belachew, R. P. Ketterling, C. A. Hanson, A. Pardanani, N. Gangat, A. P. Wolanskyj, Calreticulin mutations and long-term survival in essential thrombocythemia, Leukemia 28, 2300–2303 (2014).

43. Y. C. Elala, T. L. Lasho, N. Gangat, C. Finke, D. Barraco, M. Haider, A. K. A. Hussein, C. A. Hanson, R. P. Ketterling, A. Pardanani, A. Tefferi, Calreticulin variant stratified driver mutational status and prognosis in essential thrombocythemia, Am. J. Hematol. 91, 503–506 (2016).

44. A. Carobbio, J. Thiele, F. Passamonti, E. Rumi, M. Ruggeri, F. Rodeghiero, M. L. Randi, I. Bertozzi, A. M. Vannucchi, E. Antonioli, H. Gisslinger, V. Buxhofer-Ausch, G. Finazzi, N. Gangat, A. Tefferi, T. Barbui, Risk factors for arterial and venous thrombosis in WHO-defined essential thrombocythemia: an international study of 891 patients, Blood 117, 5857–5859 (2011).

45. G. Finazzi, A. Carobbio, P. Guglielmelli, C. Cavalloni, S. Salmoiraghi, A. M. Vannucchi, M. Cazzola, F. Passamonti, A. Rambaldi, T. Barbui, Calreticulin mutation does not modify the IPSET score for predicting the risk of thrombosis among 1150 patients with essential thrombocythemia, Blood 124, 2611–2612 (2014).

33 46. G. Rotunno, C. Mannarelli, P. Guglielmelli, A. Pacilli, A. Pancrazzi, L. Pieri, T. Fanelli, A. Bosi, A. M. Vannucchi, Impact of calreticulin mutations on clinical and hematological phenotype and outcome in essential thrombocythemia, Blood 123, 1552–1555 (2014).

47. C. Musolino, A. Allegra, A. Saija, A. Alonci, S. Russo, G. Spatari, G. Penna, D. Gerace, M. Cristani, A. David, S. Saitta, S. Gangemi, Changes in advanced oxidation protein products, advanced glycation end products, and s-nitrosylated proteins, in patients affected by polycythemia vera and essential thrombocythemia, Clin. Biochem. 45, 1439–1443 (2012).

48. A. Durmus, A. Mentese, M. Yilmaz, A. Sumer, I. Akalin, C. Topal, A. Alver, Increased oxidative stress in patients with essential thrombocythemia, Eur. Rev. Med. Pharmacol. Sci. 17, 2860–2866 (2013).

49. C. Vener, C. Novembrino, F. B. Catena, N. S. Fracchiolla, U. Gianelli, F. Savi, F. Radaelli, E. Fermo, A. Cortelezzi, S. Lonati, M. Menegatti, G. L. Deliliers, Oxidative stress is increased in primary and post-polycythemia vera myelofibrosis, Exp. Hematol. 38, 1058–1065 (2010).

50. H. C. Hasselbalch, M. Thomassen, C. H. Riley, L. Kjær, T. S. Larsen, M. K. Jensen, O. W. Bjerrum, T. A. Kruse, V. Skov, Whole blood transcriptional profiling reveals deregulation of oxidative and antioxidative defence genes in myelofibrosis and related neoplasms. Potential implications of downregulation of Nrf2 for genomic instability and disease progression, PloS One 9, e112786 (2014).

51. A. Angona, A. Alvarez-Larrán, B. Bellosillo, R. Longarón, C. Fernández-Rodríguez, C. Besses, Dynamics of JAK2 V617F allele burden of CD34+ haematopoietic progenitor cells in patients treated with ruxolitinib, Br. J. Haematol. 172, 639–642 (2016).

52. W. Vainchenker, E. Leroy, L. Gilles, C. Marty, I. Plo, S. N. Constantinescu, JAK inhibitors for the treatment of myeloproliferative neoplasms and other disorders, F1000Research 7, 82 (2018).

53. V. P. Cokic, B. B. Beleslin-Cokic, M. Tomic, S. S. Stojilkovic, C. T. Noguchi, A. N. Schechter, Hydroxyurea induces the eNOS-cGMP pathway in endothelial cells, Blood 108, 184– 191 (2006).

54. C. M. Boulanger, A. Scoazec, T. Ebrahimian, P. Henry, E. Mathieu, A. Tedgui, Z. Mallat, Circulating Microparticles From Patients With Myocardial Infarction Cause Endothelial Dysfunction, Circulation 104, 2649–2652 (2001).

55. N. Amabile, Circulating Endothelial Microparticles Are Associated with Vascular Dysfunction in Patients with End-Stage Renal Failure, J. Am. Soc. Nephrol. 16, 3381–3388 (2005).

56. P.-E. Rautou, J. Bresson, Y. Sainte-Marie, A.-C. Vion, V. Paradis, J.-M. Renard, C. Devue, C. Heymes, P. Letteron, L. Elkrief, D. Lebrec, D. Valla, A. Tedgui, R. Moreau, C. M. Boulanger, Abnormal plasma microparticles impair vasoconstrictor responses in patients with cirrhosis, Gastroenterology 143, 166-176.e6 (2012).

57. C. Marty, C. Lacout, N. Droin, J.-P. Le Couédic, V. Ribrag, E. Solary, W. Vainchenker, J.-L. Villeval, I. Plo, A role for reactive oxygen species in JAK2V617F myeloproliferative neoplasm progression, Leukemia 27, 2187–2195 (2013).

34 58. Y. Wang, M. Nakayama, M. E. Pitulescu, T. S. Schmidt, M. L. Bochenek, A. Sakakibara, S. Adams, A. Davy, U. Deutsch, U. Lüthi, A. Barberis, L. E. Benjamin, T. Mäkinen, C. D. Nobes, R. H. Adams, Ephrin-B2 controls VEGF-induced angiogenesis and lymphangiogenesis, Nature 465, 483–486 (2010).

59. C. Heymes, A. Habib, D. Yang, E. Mathieu, F. Marotte, J. Samuel, C. M. Boulanger, Cyclo- oxygenase-1 and -2 contribution to endothelial dysfunction in ageing, Br. J. Pharmacol. 131, 804–810 (2000).

60. A. Payancé, G. Silva-Junior, J. Bissonnette, M. Tanguy, B. Pasquet, C. Levi, O. Roux, O. Nekachtali, A. Baiges, V. Hernández-Gea, C. Laouénan, D. Lebrec, M. Albuquerque, V. Paradis, R. Moreau, D. Valla, F. Durand, C. M. Boulanger, J.-C. Garcia-Pagan, P.-E. Rautou, Hepatocyte microvesicle levels improve prediction of mortality in patients with cirrhosis, Hepatol. Baltim. Md (2018), doi:10.1002/hep.29903.

61. L. Kubovcakova, P. Lundberg, J. Grisouard, H. Hao-Shen, V. Romanet, R. Andraos, M. Murakami, S. Dirnhofer, K.-U. Wagner, T. Radimerski, others, Differential effects of hydroxyurea and INC424 on mutant allele burden and myeloproliferative phenotype in a JAK2- V617F polycythemia vera mouse model, Blood 121, 1188–1199 (2013).

62. S. Maschalidi, F. E. Sepulveda, A. Garrigue, A. Fischer, G. de Saint Basile, Therapeutic effect of JAK1/2 blockade on the manifestations of hemophagocytic lymphohistiocytosis in mice, Blood 128, 60–71 (2016).

63. R. Kou, T. Shiroto, J. L. Sartoretto, T. Michel, Suppression of Gαs synthesis by simvastatin treatment of vascular endothelial cells, J. Biol. Chem. 287, 2643–2651 (2012).

64. L. Lamrani, C. Lacout, V. Ollivier, C. V. Denis, E. Gardiner, B. Ho Tin Noe, W. Vainchenker, J.-L. Villeval, M. Jandrot-Perrus, Hemostatic disorders in a JAK2V617F-driven mouse model of myeloproliferative neoplasm, Blood 124, 1136–1145 (2014).

35 Acknowledgment: We thank members of the INSERM UMR-970 animal facility (ERI) for

animal handling and breeding. We thank R. Adams for having provided Cadherin5Cre-ERT2 mice.

We also thank A. Payancé, K. Zekrini and D. Rezigue for their help in identifying the patients

and M. Salama for his help with myography experiments.

Author Contributions: J.P., and P-E.R. designed the experiments and wrote the manuscript.

J.P., H.D., F.C. performed myography experiments. J.P. performed oxidative stress experiments,

M.T. generated microvesicles. J-L.V., C.J. and M.S. provided transgenic mice. M-B.E-M., C.D.

and S.H. characterized the first mouse model. J.L. managed the mouse colony. M.K. performed

mTmG experiments. A.P. included the patients and healthy controls. P-E.R. obtained funding for

the project. All authors discussed and critically revised the manuscript.

Financial support: This work was supported by the Agence Nationale pour la Recherche (ANR

14 CE35 0022 03/ JAK-POT) and J.P by the “poste accueil INSERM”.

Conflict-of-interest disclosure: The authors declare no competing financial interest.

C. JAK2V617F in splanchnic vein thrombosis

1. Endothelial JAK2V617F and Budd-Chiari syndrome

a) Background and aims

Budd-Chiari syndrome (BCS) is defined as hepatic venous outflow obstruction in the absence of congestive or restrictive heart disease. The obstacle causing BCS can be located in hepatic veins, or in the suprahepatic portion of inferior vena cava. Primary BCS is related to thrombosis as opposed to secondary BCS caused by invasion or compression by a tumour [465]. The clinical manifestations of BCS include abdominal pain, ascites, liver and spleen enlargement, and portal hypertension. About 15% of the patients display an asymptomatic form.

Myeloproliferative neoplasms are the leading cause of BCS, diagnosed in 25–50% of such patients [287]. In most patients with BCS and myeloproliferative neoplasms, JAK2V617F mutation is found in myeloid cells. JAK2V617F has also been detected in liver endothelial cells of patients with BCS [45]. The consequences of JAK2V617F in liver endothelial cells are still unknown. In BCS, JAK2V617F is associated with poorer prognostic features at presentation and earlier need for hepatic decompression procedures [287].

This observation leads to the hypothesis that JAK2V617F enhances liver injury and fibrosis induced by hepatic venous outflow obstruction, thus worsening BCS.

b) Materials and methods Murine model

All mice were on a C57BL/6 background. Mice carrying inducible JAK2V617F mutation specifically in endothelial cells were obtained by crossing Cadherin5-Cre-ERT2 transgenic mice provided by R. Adams [466] with Jak2V617F Flox/WT mice provided by M. Villeval [467]. For induction of Jak2V617F Flox/WT;Cadherin5-Cre-ERT2 model, mice were injected intraperitoneally

120 with tamoxifen (Sigma, T5648), 1mg/mice/day for 5 consecutive days over 2 consecutive weeks (10mg in total per mice) between 5 to 7 weeks old. Experiments were conducted according to the French veterinary guidelines and those formulated by the European community for experimental animal use (L358-86/609EEC) and were approved by the French agriculture minister (n° A75-A 15-32).

0 5 7 12 18 Age (weeks) Tamoxifen injections (IP) 2x 5 days Birth Begining End Surgery Sacrifice of tamoxifen of tamoxifen

Budd-Chiari mouse model by partial inferior vena cava ligation The partial inferior vena cava ligation was done in collaboration with the team of Vijay Shah and the technic has been described previously [468]. Briefly, under anesthesia the suprahepatic

IVC was circumferentially isolated (A) and a sterile steel wire of 0.6 mm in diameter was placed on the anterior surface of the IVC (B). Then a 6.0 silk thread was tightly tied around both the

IVC and the wire (C), which was subsequently gently removed (D). The sham operation included all the steps above except for the ligation. Mice were sacrificed 6 weeks postoperatively.

121

Portal pressure measurements

Portal pressure was directly measured as previously described [468] just before euthanasia using a 16-gauge catheter attached to a pressure transducer (digital blood pressure analyzer

(Digi-Med)) that was inserted into the portal vein and sutured in place. The pressure was continuously monitored, and the average portal pressure was recorded.

Euthanasia and organs preparations

Prior to sacrifice, the liver was perfused with 10 mL of 1X PBS through the portal vein. The spleen and liver were harvested and parts of the organs were snap frozen in liquid nitrogen for mRNA extraction and other parts fixed in 10% formaldehyde for histology.

Serum analyses

Serum was collected from whole blood of each animal at time of sacrifice. Specimens were send for serum transaminases measurement in Bichat, Hospital (clinical chemistry laboratory,

Paris), using a chemical chemistry analyser “Olympus AU400.

Quantitative Real-Time Polymerase Chain Reaction

Total RNA was extracted from snap frozen mouse liver samples with Trizol reagent using a polytron (T25 basic, IKA, Labortechnik). Chloroform was then added, before centrifugation. The

122 hydrous phase containing RNAs was then recuperate and washed with 70% ethanol. RNAs concentrations were measured by nanodrop and then stored at -80°C. Reverse transcription was done following manufacturer instruction (kit QuantiTect Reverse Transcription (Qiagen)).

Real-time fluorescence monitoring was performed with the Applied Biosystems, Step One Plus

Real-Time PCR System with Power SYBR Green PCR Master Mix (Eurogentec). Collagen1 α1, α-

SMA and TNFα mRNA were normalized to Gapdh, Hprt and Ppia as recommended [469].

Relative expression was calculated using the 2-delta-delta CT method followed by geometric average, as recommended [469].

T Gene Amorce Sequence (°C)

Gapdh Sens 5’ CGT CCC GTA GAC AAA ATG GTG AA 3’ 60,9 Anti-sens 5’ GCC GTG AGT GGA GTC ATA CTG GAA CA 3’ HPRT Sens 5’ TGT GCT CAA GGG GGG CTA TAA GTT 3’ 57,4 Anti-sens 5’ ACT TTT ATG TCC CCC GTT GAC TGA 3’ PPIA Sens 5’ CAC-CGT-GTT-CTT-CGA-CAT-CA 3’ 60 Anti-sens 5’ CAG-TGC-TCA-GAG-CTC-GAA-AGT 3’ α SMA Sens 5’ GAA CCC TAA GGC CAA CCG GGA GAA A 3’ 59.3 Anti-sens 5’ CCA CAT ACA TGG CGG GGA CAT TGA 3’ Collagen 1α1 Sens 5’ TGA CCG ATG CAT TCC CGT TCG AGT A 3’ 64.5 Anti-sens 5’ CCC CAA GTT GCG GTG TGA CTC 3’ TNF α Sens 5’ GAT GGG GGG CTT CCA GAA CT 3’ 62.6 Anti-sens 5’ CGT GGG CTA CAG GCT TGT CAC 3’

Liver fibrosis staining and quantification

Livers were fixed in 10% phosphate buffered formalin for 48 h at 4°C, washed twice with water, stored in 70% ethanol at 4°C, and then embedded in paraffin. After deparaffinization and hydration, sections (5 μM) were stained with picro-sirius red (Sigma). Sirius red was quantitated in sections (X30; 12 fields each from sample) using Archimed software, around centro-lobular veins. Images were then analysed using « Image J » software. Two independent

123 examinators who were blind for genotype at the time did the quantifications.

Statistics

Data are expressed as median (interquartile range) for quantitative data (blood cell count, body and spleen weight) and comparisons between groups were performed using the Mann-

Whitney U-test. All tests were 2 sided and used a significance level of 0.05. Data handling and analysis were performed with GraphPad Softwar, Inc.

124 c) Article 2: Endothelial JAK2V617F does not enhance liver lesions in mice with Budd-Chiari syndrome

Endothelial JAK2V617F does not enhance liver lesions in mice with Budd-Chiari syndrome

Johanne Poisson1,2,y Moira B. Hilscher3,y Marion Tanguy1,2 Adel Hammoutene1,2 Chantal M. Boulanger1,2 Jean-Luc Villeval4,5 Douglas A. Simonetto3 Dominique Valla6,7 Vijay H. Shah3 Pierre-Emmanuel Rautou1,2,6,7

1INSERM, UMR-970, Paris Cardiovascular Research Center – PARCC, Paris, France 2Université Paris Descartes, Sorbonne Paris Cité, Paris, France 3Gastroenterology Research Unit, Division of Gastroenterology and Hepatology, Mayo Clinic, Rochester, MN, USA 4INSERM, Institut Gustave Roussy, INSERM U1170, Villejuif, France 5University Paris XI, Villejuif, France 6Service d’hépatologie, DHU Unity Hôpital Beaujon, APHP, Clichy, France 7Université Denis Diderot-Paris 7, Sorbonne Paris Cité, 75018 Paris, France

Letter published in J Hepatol. 2018 May;68(5):1086-1087

125 Letters to the Editor JOURNAL OF HEPATOLOGY

Endothelial JAK2V617F does not enhance liver lesions in mice with Budd-Chiari syndrome

promotes major vasodilation and hemostasis impairment, 7 To the Editor: making surgery extremely challenging in these animals. Budd-Chiari syndrome is defined as hepatic venous outflow Accordingly, we analyzed the endothelial component using mice V617F obstruction in the absence of congestive or restrictive heart expressing JAK2 specifically in endothelial cells. We gener- V617F disease. Myeloproliferative neoplasms are the leading cause ated these transgenic mice by crossing conditional Jak2 Cre-ERT2 of Budd-Chiari syndrome, diagnosed in 25–50% of such knock-in mice with inducible Cadherin5 mice. Recombina- V617F Cre-ERT2 patients.1,2 In most patients with Budd-Chiari syndrome and tion was induced in Jak2 knock-in – Cadherin5 (here- V617F myeloproliferative neoplasms, Janus kinase 2 gene (JAK2) after referred to as JAK ) mice by tamoxifen injection V617F mutation is found in myeloid cells. JAK2V617F has also (1 mg/day/mice intraperitoneously for five consecutive days, been detected in liver endothelial cells of patients with two consecutive weeks) at the age of five weeks. Littermate WT Budd-Chiari syndrome, attributed to a common cell of origin controls (hereafter referred to as JAK2 ) received the same for myeloid and endothelial cells, called hemangioblast.3–5 In treatment. Partial inferior vena cava ligation (pIVCL), or sham Budd-Chiari syndrome, JAK2V617F is associated with poorer surgeries were performed at the age of 12 weeks and mice were 6 prognostic features at presentation and earlier need for hepatic sacrificed six weeks postoperatively (Fig. 1). Based on previous decompression procedures.1 This observation leads to the experiments using this surgical model, we included 8 to 10 mice 6 hypothesis that JAK2V617F enhances liver injury and fibrosis per group. All experiments were performed in accordance with induced by hepatic venous outflow obstruction, thus worsen- the European Community guidelines for the care and use of ° ing Budd-Chiari syndrome. laboratory animals (N 07,430) and were approved by our insti- In order to test this hypothesis, we applied a recently tutional ethical committee (17-053). WT described surgical model of Budd-Chiari syndrome to mice As expected, JAK2 mice undergoing pIVCL had higher expressing JAK2V617F.6 JAK2V617F expression in myeloid cells portal pressure, liver expression of profibrogenic genes and liver

A D n.s. F n.s. 10 n.s. n.s. 20 ** n.s. 0 5 7 12 18 Age 8 15 ld change) (weeks) mRNA

o 6

Tamoxifen 1

injections (IP) α 10 2x 5 days 4 Birth Begining End Surgery Sacrifice (fold change)

mRNA (f 5 2

of tamoxifen of tamoxifen α Collagen1 0 0 TNF JAK2 WT JAK2 WT JAK2 V617F JAK2 WT JAK2 WT JAK2 V617F Sham pIVCL Sham pIVCL n.s. n.s. BC8 E 15 G *** n.s. n.s. ** 3 800 ** n.s. *** n.s. *** n.s.

6 /L)

U 600 10 2

4 LT (I 400 A 5 1

2 200 (fold change) mRNA sredpositivearea(%) Serum iriu Portal pressure (mmHg) 0 SMA S 0 α 0 0 JAK2 WT JAK2 WT JAK2 V617F JAK2 WT JAK2 WT JAK2 V617F JAK2 WT JAK2 WT JAK2 V617F JAK2 WT JAK2 WT JAK2 V617F Sham pIVCL Sham pIVCL Sham pIVCL Sham pIVCL

Fig. 1. Endothelial JAK2V617F does not enhance liver injury in mice after partial inferior vena cava ligation. Partial inferior vena cava ligation was performed in 12-week aged male and female Jak2V617F knock-in - Cadherin5Cre-ERT2 (JAKV617F) mice and in littermate controls (JAK2WT). Sham surgery was also performed. All mice were on a C57BL/6 background. (A) Mice were sacrificed 6 weeks postoperatively. (B) Portal pressure, (C) serum ALT level, (D) liver TNFa, (E) aSMA and (F) collagen1a1 gene expressions were determined (supplementary CTAT Table) and (G) liver fibrosis was quantified (Sirius red positive areas). Data are given as median (horizontal bar) and interquartile range (error bar). Comparisons between groups of mice were performed using the Mann–Whitney U test. *p <0.05; **p <0.01; and ***p <0.001. ALT, alanine aminotransferase; JAK2, Janus kinase 2; pIVCL, partial inferior vena cava ligation; n.s., no significant difference.

Keywords: Myeloproliferative neoplasm; Liver ﬁbrosis; Portal hypertension; Hepatic venous outﬂow obstruction; Splanchnic thrombosis.

Journal of Hepatology 2018 vol. 68 j 1086–1106 JOURNAL OF HEPATOLOGY

fibrosis than sham mice, while showing no change in serum References aminotransferases or in liver expression of proinflammatory Author names in bold designate shared co-first authorship genes (Fig. 1).6 However, as shown (Fig. 1), the expression of JAKV617F in liver endothelial cells did not affect any of these [1] Kiladjian J-J, Cervantes F, Leebeek FW, Marzac C, Cassinat B, Chevret S, parameters. et al. The impact of JAK2 and MPL mutations on diagnosis and prognosis of splanchnic vein thrombosis: a report on 241 cases. Blood In conclusion, we found no evidence in an animal model 2008;111:4922–4929. V617F that endothelial JAK2 can explain the more severe pre- [2] European Association for the Study of the Liver. EASL clinical practice sentation of patients with Budd-Chiari syndrome and guidelines: vascular diseases of the liver. J Hepatol 2016;64:179–202. JAK2V617F. The explanation for increased severity of these [3] Sozer S, Fiel MI, Schiano T, Xu M, Mascarenhas J, Hoffman R. The presence patients should therefore be sought mostly in myeloid of JAK2V617F mutation in the liver endothelial cells of patients with V617F Budd-Chiari syndrome. Blood 2009;113:5246–5249. JAK2 . Thus, future therapeutic strategies to improve the [4] Rosti V, Villani L, Riboni R, Poletto V, Bonetti E, Tozzi L, et al. Spleen management of patients with Budd-Chiari syndrome and endothelial cells from patients with myelofibrosis harbor the JAK2V617F myeloproliferative neoplasms might focus on myeloid cells mutation. Blood 2013;121:360–368. rather than on endothelial cells. Beside the cytoreductive [5] Teofili L, Martini M, Iachininoto MG, Capodimonti S, Nuzzolo ER, Torti L, et al. Endothelial progenitor cells are clonal and exhibit the JAK2(V617F) agent hydroxyurea, treatments for myeloproliferative neo- mutation in a subset of thrombotic patients with Ph-negative myelopro- plasms now also include the JAK2/1 inhibitor ruxolitinib. liferative neoplasms. Blood 2011;117:2700–2707. One phase II trial recently reported that ruxolitinib is safe [6] Simonetto DA, Yang H, Yin M, de Assuncao TM, Kwon JH, Hilscher M, et al. in patients with splanchnic vein thrombosis.8 Whether ruxoli- Chronic passive venous congestion drives hepatic fibrogenesis via sinusoidal thrombosis and mechanical forces. Hepatology 2015;61: tinib is useful in this setting to improve patient outcomes 648–659. should be evaluated in larger studies. [7] Lamrani L, Lacout C, Ollivier V, Denis CV, Gardiner E, Ho Tin Noe B, et al. Hemostatic disorders in a JAK2V617F-driven mouse model of myeloproliferative neoplasm. Blood 2014;124:1136–1145. Financial support [8] Pieri L, Paoli C, Arena U, Marra F, Mori F, Zucchini M, et al. Safety and This work was supported by the Agence Nationale pour la efficacy of ruxolitinib in splanchnic vein thrombosis associated with Recherche (ANR-14-CE12-0011 and ANR-14-CE35-0022) and myeloproliferative neoplasms. Am J Hematol 2017;92:187–195. J.P by the ‘‘poste d’accueil INSERM”. Johanne Poisson1,2,y Moira B. Hilscher3,y Conflict of interest Marion Tanguy1,2 The authors declare no conflicts of interest that pertain to this Adel Hammoutene1,2 work. Chantal M. Boulanger1,2 Please refer to the accompanying ICMJE disclosure forms for Jean-Luc Villeval4,5 further details. Douglas A. Simonetto3 Dominique Valla6,7 3 Vijay H. Shah⇑ Authors’ contributions Pierre-Emmanuel Rautou1,2,6,7, J.P. and M.B.H contributed equally to the work. M.B.H. and D.A.S. 1INSERM, UMR-970, Paris Cardiovascular Research Center – PARCC, performed mouse surgeries. J.P., A.H. and M.T. analyzed liver Paris, France samples. J-L.V. generated Jak2V617F knock-in mice. J.P. and P-E. 2Université Paris Descartes, Sorbonne Paris Cité, Paris, France R. wrote the manuscript. C.M.B, D.V. and V.H.S discussed and 3Gastroenterology Research Unit, Division of Gastroenterology and analyzed the results. All authors critically revised the Hepatology, Mayo Clinic, Rochester, MN, USA manuscript. 4INSERM, Institut Gustave Roussy, INSERM U1170, Villejuif, France 5University Paris XI, Villejuif, France 6Service d’hépatologie, DHU Unity Hôpital Beaujon, APHP, Clichy, Acknowledgements France We thank the members of the INSERM UMR-970 animal facility 7Université Denis Diderot-Paris 7, Sorbonne Paris Cité, 75018 Paris, (ERI), Fatoumata Camara for superb technical assistance, and the ⇑ France Hôpital Bichat biochemistry core facility. We also thank R. Adams Corresponding author. Address: Service d’Hépatologie, Hôpital for having provided Cadherin5Cre-ERT2 mice. Beaujon, 100 boulevard du Général Leclerc, 92110 Clichy, France. Tel.: +33 1 40 87 52 83; fax: +33 1 40 87 54 87. Supplementary data E-mail address: [email protected] Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.jhep.2018.01. 010. y J.P. and M.B.H. contributed equally to this study.

Journal of Hepatology 2018 vol. 68 j 1086–1106 1087 2. Calreticulin mutations and splanchnic vein thrombosis

a) Article 3: Selective testing for calreticulin gene mutations in patients with splanchnic vein thrombosis: A prospective cohort study

Selective testing for calreticulin gene mutations in patients with splanchnic vein thrombosis : A prospective cohort study

Johanne Poisson1, Aurélie Plessier2, Jean-Jacques Kiladjian3, Fanny Turon4, Bruno Cassinat3,

Annalisa Andreoli3, Emmanuelle De Raucourt5, Odile Goria6, Kamal Zekrini3, Christophe

Bureau7, Florence Lorre8,9, Francisco Cervantes13, Dolors Colomer14, François Durand2,10,11,

Juan-Carlos Garcia-Pagan4,12, Nicole Casadevall8,9, Dominique-Charles Valla2,10,11, Pierre-

Emmanuel Rautou1,2,11, Christophe Marzac8,9, for the French national network for vascular liver diseases

1 Inserm, U970, Paris Cardiovascular Research Center - PARCC, Université Paris Descartes, Sorbonne Paris Cité, Paris, France; 2 DHU Unity, Pôle des Maladies de l’Appareil Digestif, Service d’Hépatologie, Centre de Référence des Maladies Vasculaires du Foie, Hôpital Beaujon, AP-HP, Clichy, France; 3 Centre d’Investigations Cliniques, Hôpital St Louis, AP-HP, Clichy, France; 4 Barcelona Hepatic Hemodynamic Laboratory, Liver Unit, Hospital Clínic, IDIBAPS, Barcelona, Spain; 5 Service d’Hématologie Biologique, Hôpital Beaujon, AP-HP, Clichy, France; 6 Service d’Hépato-gastroentérologie, CHU Rouen, Rouen, France; 7 Liver-Gastroenterology Department, University Hospital and Paul Sabatier University, Toulouse, France; 8 UPMC, Univ Paris 06, Groupe de Recherche Clinique sur les Myéloproliférations Aiguës et Chroniques MYPAC,Paris, France; 9 Laboratoire d’Immunologie et Hématologie Biologique, Hôpital Saint-Antoine, AP-HP, Paris, France; 10 Inserm U1149, Centre de Recherche sur l’Inflammation (CRI), Paris, Université Paris 7-Denis-Diderot, Clichy, UFR de Médecine, Paris, France; 11 Université Paris Diderot, Sorbonne Paris cité, Paris, France; 12 Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBERehd), Spain; 13Hematology Department, Hospital Clínic, IDIBAPS, University of Barcelona, Spain; 14 Hematopathology Unit, Hospital Clínic, IDIBAPS, CIBERONC, Spain

Original article published in J Hepatol. 2017 Sep;67(3):501-507

128 Research Article

Selective testing for calreticulin gene mutations in patients with splanchnic vein thrombosis: A prospective cohort study

Graphical Abstract Authors

Johanne Poisson, Aurélie Plessier, Jean-Jacques Kiladjian, ..., Dominique-Charles Valla, Pierre-Emmanuel Rautou, Christophe Marzac

Correspondence [email protected] (P.-E. Rautou)

Lay summary Mutations of the CALR gene are detected Highlights in 0 to 2% of patients with SVT, thus the CALR mutations are detected in 2% of patients with splanch- utility of systematic CALR mutation test- nic vein thrombosis. ing to diagnose MPN is questionable. This study demonstrates that CALR mutations CALR mutations should not be tested in patients with testing can be restricted to patients with JAK2V617F. SVT, a spleen height 16 cm, a platelet 9 V617F CALR mutations should be tested in patients with spleno- count >200Â10 /L, and no JAK2 . This megaly & platelets >200Â109/L. strategy avoids 96% of unnecessary CALR mutations testing. This strategy avoids 96% of unnecessary CALR mutations testing.

129 Research Article

Selective testing for calreticulin gene mutations in patients with splanchnic vein thrombosis: A prospective cohort study

Johanne Poisson1, Aurélie Plessier2, Jean-Jacques Kiladjian3, Fanny Turon4, Bruno Cassinat3, Annalisa Andreoli3, Emmanuelle De Raucourt5, Odile Goria6, Kamal Zekrini3, Christophe Bureau7, Florence Lorre8,9, Francisco Cervantes13, Dolors Colomer14, François Durand2,10,11, Juan-Carlos Garcia-Pagan4,12, Nicole Casadevall8,9, Dominique-Charles Valla2,10,11, ⇑ Pierre-Emmanuel Rautou1,2,11, ,y, Christophe Marzac8,9,y, for the French national network for vascular liver diseases

1Inserm, U970, Paris Cardiovascular Research Center - PARCC, Université Paris Descartes, Sorbonne Paris Cité, Paris, France; 2DHU Unity, Pôle des Maladies de l’Appareil Digestif, Service d’Hépatologie, Centre de Référence des Maladies Vasculaires du Foie, Hôpital Beaujon, AP-HP, Clichy, France; 3Centre d’Investigations Cliniques, Hôpital St Louis, AP-HP, Clichy, France; 4Barcelona Hepatic Hemodynamic Laboratory, Liver Unit, Hospital Clínic, IDIBAPS, Barcelona, Spain; 5Service d’Hématologie Biologique, Hôpital Beaujon, AP-HP, Clichy, France; 6Service d’Hépato-gastroentérologie, CHU Rouen, Rouen, France; 7Liver-Gastroenterology Department, University Hospital and Paul Sabatier University, Toulouse, France; 8UPMC, Univ Paris 06, GRC n°7, Groupe de Recherche Clinique sur les Myéloproliférations Aiguës et Chroniques MYPAC, Paris, France; 9Laboratoire d’Immunologie et Hématologie Biologique, Hôpital Saint-Antoine, AP-HP, Paris, France; 10Inserm U1149, Centre de Recherche sur l’Inﬂammation (CRI), Paris, Université Paris 7-Denis-Diderot, Clichy, UFR de Médecine, Paris, France; 11Université Paris Diderot, Sorbonne Paris cité, Paris, France; 12Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBERehd), Spain; 13Hematology Department, Hospital Clínic, IDIBAPS, University of Barcelona, Spain; 14Hematopathology Unit, Hospital Clínic, IDIBAPS, CIBERONC, Spain

Background and Aims: Myeloproliferative neoplasms (MPN) are criteria had a positive predictive value of 56% (5/9) and a negative the leading cause of splanchnic vein thrombosis (SVT). Janus predictive value of 100% (0/233) for the identiﬁcation of CALR kinase 2 gene (JAK2)V617F mutations are found in 80 to 90% of mutations. In the validation cohort, these criteria had a positive patients with SVT and MPN. Mutations of the calreticulin (CALR) predictive value of 33% (2/6) and a negative predictive value of gene have also been reported. However, as their prevalence 99% (1/96). ranges from 0 to 2%, the utility of routine testing is questionable. Conclusion: CALR mutations should be tested in patients with This study aimed to identify a group of patients with SVT at high SVT, a spleen height P16 cm, platelet count [200Â109/L, and risk of harboring CALR mutations and thus requiring this genetic no JAK2V617F. This strategy avoids 96% of unnecessary CALR muta- testing. tions testing. Methods: CALR, JAK2V617F and thrombopoietin receptor gene Lay summary: Mutations of the CALR gene are detected in 0 to 2% (MPL) mutations were analysed in a test cohort that included of patients with SVT, thus the utility of systematic CALR mutation 312 patients with SVT. Criteria to identify patients at high risk testing to diagnose MPN is questionable. This study demonstrates of CALR mutations in this test cohort was used and evaluated in that CALR mutations testing can be restricted to patients with a validation cohort that included 209 patients with SVT. SVT, a spleen height P16 cm, a platelet count [200Â109/L, and Results: In the test cohort, 59 patients had JAK2V617F, ﬁve had no JAK2V617F. This strategy avoids 96% of unnecessary CALR CALR and none had MPL mutations. Patients with CALR mutations mutations testing. had higher spleen height and platelet count than patients with- Ó 2017 European Association for the Study of the Liver. Published out these mutations. All patients with CALR mutations had a by Elsevier B.V. All rights reserved. spleen height P16 cm and platelet count [200Â109/L. These

Keywords: Myeloproliferative neoplasms; Budd-Chiari syndrome; Portal vein Introduction thrombosis; JAK2V617F; MPL mutation; CALR mutations; Platelets count; Splenomegaly; DNA mutational analysis; Genetic testing. Splanchnic vein thrombosis (SVT) indicates Budd-Chiari syndrome Received 14 February 2017; received in revised form 19 April 2017; accepted 22 April 2017; available online 5 May 2017 (BCS) and portal venous system thrombosis (PVT). Primary BCS is a ⇑ Corresponding author. Address: Service d’Hépatologie, Hôpital Beaujon, Assis- rare disorder deﬁned as a blocked hepatic venous outﬂow tract at tance Publique-Hôpitaux de Paris, Clichy, France. Tel.: +33 171114679; fax: +33 various levels from small hepatic veins to the terminal portion of 140875530. the inferior vena cava.1 Non-malignant non-cirrhotic extrahepatic E-mail address: [email protected] (P.-E. Rautou). y These authors contributed equally to this work. PVT is characterized by thrombus development in the main portal

Journal of Hepatology 2017 vol. 67 j 501–507 Research Article vein and/or its right or left branches and/or splenic or mesenteric Bone marrow biopsy and/or endogenous erythroid colonies formation were veins, or by the permanent obliteration that results from a prior performed when considered relevant by the physician according to French recommendations.18 thrombus.1 The pathogenesis of SVT is largely dependent on the presence of systemic prothrombotic conditions that promote Investigations for other thrombotic risk factors thrombus formation in the respective splanchnic veins.2,3 Myeloproliferative neoplasms (MPNs) are the leading cause of 4 Patients were tested according to previously reported methods for the following SVT and are diagnosed in 25 to 50% of patients with SVT. In most thrombotic risk factors:19 factor V R506Q mutation (factor V Leiden); G20210A V617F patients with SVT and MPN, Janus kinase 2 gene (JAK2) muta- factor II gene mutation; deficiencies in protein C, protein S, or antithrombin tion is found. In 10% to 20% of patients with SVT this specific (regarded as primary deficiencies only in conjunction with a prothrombin index mutation is absent, whereas bone marrow biopsy or assessment P80%); paroxysmal nocturnal hemoglobinuria; and anti-phospholipid antibodies.1 Oral contraceptive use was considered a thrombotic risk factor when taken of endogenous erythroid colonies formation provide evidence within the three months preceding diagnosis of SVT.20 for MPN.5 Mutations across JAK2 exon 12 or the thrombopoietin 5 receptor gene (MPL) are rarely identified in patients with SVT. Imaging analyses Two independent groups described heterozygous calreticulin

(CALR) mutations as the second most prevalent acquired genetic All abdominal multidetector computed tomography or magnetic resonance imag- alteration in essential thrombocythemia and primary myelofibro- ing performed within six months of SVT diagnosis were reviewed to measure the sis.6,7 CALR mutations are mutually exclusive of JAK2 and MPL greatest spleen height in coronal view. mutations. Thereafter, CALR mutations have been found in 0 to 2% of patients with SVT.8–17 Although CALR mutations appear to Statistical analysis be rare in patients with SVT and their detection not readily accessible to all centers, their identification influences patients’ clinical Quantitative variables were expressed as median (interquartile range), and cate- management. This prompted us to take advantage of a large gorical variables as absolute and relative frequencies. Comparisons between groups of quantitative and qualitative variables were performed using Mann prospective cohort of SVT patients to identify the subgroup at Whitney and the Fisher exact tests, respectively. All tests were two-sided and the highest risk of harboring CALR mutations and thus requiring used a significance level of 0.05. Data handling and analysis were performed with this genetic testing. SPSS 17.0 (SPSS Inc., Chicago, IL). For further details regarding the materials used, please refer to the CTAT table.

Patients and methods

Inclusion criteria Results

This study prospectively included patients with BCS or PVT seen between 2005 Patient characteristics and 2013 at the French Reference Center for Vascular Disorders of the Liver (Clichy, France) and for whom peripheral blood DNA was available for mutation Three hundred twelve patients were enrolled, including 99 (32%) screening (Supplementary CTAT Table). The protocol was performed in accor- with BCS and 213 (68%) with PVT. Patients’ characteristics are dance with the ethical guidelines of the 1975 Declaration of Helsinki and was shown in Table 1. approved by the institutional review board (CPP Ile de France IV, Paris; France). Informed consent was obtained from all patients included in the study. In patients with BCS, hepatic venous outflow obstruction was We looked for criteria characterizing patients at high risk of having CALR due to occlusion of one, two and three hepatic veins in 10, 19 and mutations in this French cohort, thereafter referred to as ‘‘test cohort” and then 67 patients, respectively, and due to obstruction of the suprahep- tested these criteria for validation in a previously reported cohort from Hospital atic segment of the inferior vena cava in two patients. The last Clinic, Barcelona, thereafter referred to as ‘‘validation cohort”.8 patient had small hepatic veins BCS. Thirteen out of the 99 patients with BCS also had a PVT. Out of the 213 patients with Definitions PVT without BCS, 39% had a portal cavernoma and 61% an acute PVT. Portal, splenic and mesenteric veins were involved in 199 BCS was defined as hepatic outflow obstruction regardless of the cause or level of (93%), 79 (37%) and 118 (55%) of these 213 patients, respectively. obstruction, from the small hepatic veins to the entrance of the inferior vena cava into the right atrium. BCS was confirmed by ultrasonography and/or multidetec- Risk factors for thrombosis are detailed in Table 1. The most V617F tor computed tomography and/or magnetic resonance imaging, and/or common cause was MPN. JAK2 was detected in 81% of the venography. Sinusoidal obstruction syndrome, as well as outflow obstruction MPN patients. No MPL mutation was found. occurring in the setting of heart failure, orthotopic liver transplantation and hepatobiliary cancer were excluded from this definition. Diagnostic criteria for PVT included recent portal, and/or splenic and/or mesenteric venous thrombosis or CALR mutation portal cavernoma. PVT patients with cirrhosis or abdominal malignancies were excluded. CALR mutations were detected in five patients (1.6%), a proportion in agreement with previous studies. Their individual charac- Hematologic studies teristics are presented in Table 2. None of the patients with CALR mutations had JAK2V617F or MPL mutations. Out of the patients JAK2V617F, MPL and CALR mutations were tested in all patients. JAK2V617F and MPL without CALR mutations, 59 had JAK2V617F, nine had a triple neg- 5 mutation analyses were performed as previously described. ative MPN (absence of JAK2V617F, CALR or MPL mutations but pos- For CALR mutations, we used DNA extracted by an automated standardized procedure (Qiasymphony, Qiagen) from blood samples collected by venipuncture itive bone marrow biopsy in eight patients, or endogenous in tubes containing 0.11 mol/L trisodium citrate and stored at À80 °C until anal- erythroid colonies formation in one patient) and 57 did not have ysis. The mutational status of CALR was determined using previously described MPN after bone marrow biopsy and/or endogenous erythroid high-resolution sizing of fluorescent dye-labeled PCR amplification of exon 9, colonies formation examination. We compared the clinical and with Sanger sequencing controls.7 laboratory features of patients with CALR mutations with those

502 Journal of Hepatology 2017 vol. 67 j 501–507 JOURNAL OF HEPATOLOGY

Table 1. Characteristics and risk factors for thrombosis in patients with BCS or PVT. BCS PVT (n = 99) (n = 213) Age, years 35 (25–45) 43 (32–56) Males 28 (28%) 119 (56%) Inherited thrombophilia Protein C deficiency 4/81 (5%) 14/185 (8%) Protein S deficiency 4/78 (5%) 19/188 (10%) Antithrombin deficiency 1/84 (1%) 4/197 (2%) Factor V gene mutation 11/96 (11%) 9/210 (4%) Factor II gene mutation 4/99 (4%) 15/213 (7%) Acquired thrombophilia Myeloproliferative neoplasms 30/99 (30%) 44/213 (21%) Polycythemia vera 13 (13%) 19 (9%) Essential thrombocythemia 10 (10%) 13 (6%) Primary myelofibrosis 3 (3%) 3 (1%) Unclassifiable 4 (4%) 9 (4%) JAK2V617F 28/99 (28%) 31/213 (15%) MPL mutation 0/99 0/213 CALR mutation 1/99 (1%) 4/213 (2%) Antiphospholipid antibody syndrome 9/96 (9%) 23/210 (11%) Paroxysmal nocturnal haemoglobinuria 4/83 (5%) 1/167 (0.5%) Hormonal (OC and/or pregnancy) 33/96 (34%) 45/212 (21%) Systemic disordery 7/99 (7%) 11/213 (5%) Local risk factorà 3 (3%) 16 (8%) Single risk factor 44 (44%) 91 (43%) Multiple risk factors 26 (26%) 49 (23%) No risk factor 29 (29%) 73 (34%) Values are n (%) or median (interquartile range). BCS, Budd-Chiari syndrome; OC, oral contraception; PVT, portal venous system thrombosis. y Behçet disease, sarcoidosis, vasculitis, connective tissue disease or lymphoide hemopathy. à Intra-abdominal inflammation, infection, or abscess. of patients with JAK2V617F, triple negative MPN and without MPNs SVT previously reported.8 As shown in Fig. 2, two out of the three (Table 3). Patients with CALR mutations had significantly higher patients with CALR mutation fulfilled these criteria. The third platelet counts than patients with triple negative MPNs and patient with CALR mutations had a platelet count of 477Â109/L, patients without MPNs. Patients with CALR mutations also had but spleen height of 11 cm. A fourth patient with a CALR mutation a higher spleen height than patients with JAK2V617F, with triple in this cohort had a platelet count [200Â109/L and an enlarged negative MPNs and patients without MPNs (p = 0.05, p = 0.05, spleen based on the radiology report. However, images could p = 0.001, respectively). Patients with CALR mutations also had not be retrieved so that precise spleen size could not be significantly lower haemoglobin and haematocrit levels than measured. Patients’ features are detailed in Table 2. In the valida- patients with JAK2V617F. There was no difference between patients tion cohort, the criteria ‘‘spleen height P16 cm and platelet count with CALR mutations and other groups regarding other clinical [200Â109/L” thus had a positive predictive value of 33% (2/6) characteristics, frequency of inherited and other acquired risk and negative predictive value of 99% (1/96) for the identification factors for thrombosis or laboratory features. of CALR mutations. Given the high platelet count and spleen height observed in patients with CALR mutations, we tested the hypothesis that criteria derived from those proposed in 2005, before JAK2 and CALR Discussion mutations discovery, namely spleen height P16 cm and platelet [ Â 9 count 200 10 /L, could identify a group at high risk of CALR This study of more than 500 patients with SVT has allowed for 18 mutation among patients with SVT. As shown in Fig. 1, all five the characterization of patients in whom CALR mutations should patients with CALR mutations fulfilled these criteria. These crite- be tested. ria thus had a positive predictive value of 56% (5/9) and a nega- Indeed, the main finding of this study was that CALR muta- tive predictive value of 100% (0/233) for the identification of tions were almost exclusively found in patients with SVT, with- CALR mutations. Out of the other four patients with a spleen out JAK2V617F, when spleen height was P16 cm and platelet 9 P [ Â 9 height 16 cm and platelet count 200 10 /L, two had a histo- count [200Â10 /L. CALR mutations are driving essential throm- logically proven MPN, one had a histiocytosis and one had an bocythemia and primary myelofibrosis, but not polycythemia antiphospholipid antibody syndrome. vera.21 Indeed, these mutations induce activation of the thrombopoietin receptor, MPL, resulting in the proliferation of the Validation cohort megakaryocytic lineage. Clinical manifestations of CALR mutated MPNs include high platelet count and enlarged spleen.21 Thus, We then tested the criteria ‘‘spleen height P16 cm and platelet identification in our study of these two parameters as markers 9 count [200Â10 /L” in an independent cohort of patients with of the presence of CALR mutations is not surprising. Because of

Journal of Hepatology 2017 vol. 67 j 501–507 503 Research Article

Table 2. Characteristics of patients with SVT and CALR mutations. Patient Cohort Age Gender JAK2V617F Haematologic Type of Other risk factor for Spleen height Platelet count (yr) status diseases SVT thrombosis (cm) (109/L) 1 Test 24 Female Negative PMF BCS OC and Gastroenteritis 20.0 417 2 Test 30 Female Negative ET PVT Sarcoidosis 16.6 436 3 Test 39 Male Negative PMF PVT None 19.0 453 4 Test 32 Female Negative PMF PVT None 17.0 476 5 Test 36 Male Negative PMF PVT None 18.0 477 6 Validation 34 Male Negative ET BCS None 18.0 300 7 Validation 57 Female Negative PMF PVT None 18.0 607 8 Validation 57 Female Negative ET BCS None 11.0 477 9 Validation 73 Male Negative ET PVT None n.a.* 337 BCS, Budd-Chiari syndrome; ET, essential thrombocythemia; PMF, primary myeloﬁbrosis; n.a., not available; OC, oral contraception; PVT, Portal venous system thrombosis; yr, years. * No spleen height available but the patient was known to have an enlarged spleen.

Table 3. Characteristics associated with the CALR mutations in patients with SVT from the test cohort. With CALR mutations JAK2V617F Triple negative MPN No MPN (n = 5) (n = 59) (n = 9) (n = 57) Age, years 32 (27–37) 42 (31–49) 47 (30–55) 35 (26–47) Males 2 (40%) 15 (25%) 4 (44%) 20 (35%) Liver disease BCS 1 (20%) 28 (48%) 1 (11%) 25 (44%) PVT 4 (80%) 31 (52%) 8 (89%) 32 (56%) Inherited thrombophilia Protein C deficiency 0/5 3/43 (7%) 1/9 (11%) 2/50 (4%) Protein S deficiency 0/5 6/43 (14%) 0/9 7/51 (14%) Antithrombin deficiency 0/5 1/45 (2%) 0/9 0/57 Factor V gene mutation 0/5 7/58 (12%) 0/9 1/55 (2%) Factor II gene mutation 0/5 5/59 (8%) 0/9 4/55 (7%) Acquired thrombophilia MPNs 5 (100%) 59 (100%) 9 (100%) 0*** Antiphospholipid antibody syndrome 0/5 5/56 (9%) 1/9 (11%) 5/57 (10%) Paroxysmal nocturnal haemoglobinuria 0/5 0/46 0/7 1/42 (2%) Hormonal (OC and/or pregnancy) 1/5 (20%) 15/56(25%) 2/9 (22%) 20/56 35%) Systemic disordery 1/5 (20%) 0/59 0/9 2/57 (3%) Local risk factorà 1/5 (20%) 1/55 (2%) 2/9 (22%) 1/54 (2%) Single risk factor 3/5 (60%) 30/59 (51%) 5/9 (56%) 23/57 40%) Multiple risk factors 2/5 (20%) 29/59 (49%) 4/9 (44%) 9/57 (16%) No risk factor 0/5 0/59 0/9 25/57 (44%) Haemoglobin, g/dl 11.9 (10.0–13.0) 14.2 (12.7–15.5)* 13.2 (12.1–15.4) 12.7 (11.6–14.6) Haematocrit, % 36.5 (31.3–40.4) 42.8 (38.8–48.0)* 40.0 (35.5–45.5) 38.5 (35.5–43.1) WBC count, Â109/L 7 (5.3–12.1) 10 (7.6–13.5) 7.2 (6.3–8.7) 6.8 (4.9–8.1) ANC, Â109/L 5.2 (3.4–10) 7.2 (5.3–10.8) 4.7 (3.9–5.3) 3.9(2.6–5.5) Platelet count, Â109/L 453 (427–477) 377 (286–500) 234 (225–431)* 227 (142–314)** Spleen height, cm 18.0 (16.8–19.5) 15.0 (13.2–18.0) 14.2 (7–18) 11.0 (9.0–14.0)** AST, U/L 71 (21–114) 44 (28–91) 37 (21–53) 28 (22–42) ALT, U/L 73 (33–169) 56 (32–90) 51 (32–100) 35 (21–49) Serum bilirubin, mmol/L 16 (11–31) 16 (10–33) 11 (8–24) 13 (7–20) Serum creatinine, mmol/L 54 (51–86) 68 (58–77) 71 (68–84) 71 (64–84) Serum albumin, g/L 34 (29–43) 38 (33–43) 41 (33–42) 38 (33–45) Factor V, % 86 (53–111) 62 (52–82) 88 (43–117) 90 (71–106) Values are n (%) or median (interquartile range). Triple negative correspond to patient without JAK2V617F, CALR and MPL mutation and with confirmed MPN. All patients from the group confirmed MPN had a BM biopsy and/or a EEC. *p \0.05, **p \0.01, ***p \0.001 vs. patients with CALR mutations (only differences with the CALR mutated group are indicated). Comparisons between groups of quantitative and qualitative variables were performed using Mann Whitney and the Fisher exact tests, respectively. ALT, alanine transaminase; ANC, absolute neutrophil count; AST, aspartate transaminase; BCS, Budd-Chiari syndrome; BM, bone marrow; EEC, endogenous erythroid colonies formation; MPN, myeloproliferative neoplasm; OC, oral contraception; PVT, portal venous system thrombosis; vs., versus; WBC, white blood cell. y Behçet disease, sarcoidosis, vasculitis, connective tissue disease or lymphoide hemopathy. à Intra-abdominal inflammation, infection, or abscess.

504 Journal of Hepatology 2017 vol. 67 j 501–507 JOURNAL OF HEPATOLOGY

312 patients with splanchnic vein thrombosis (99 Budd-Chiari syndrome; 213 portal venous system thrombosis)

With JAK2 V617F Without JAK2 V617F (n = 59) (n = 253)

Spleen height ≥16 cm and Spleen height <16 cm or Unavailable data platelet count >200x109/L platelet count ≤200x109/L (n = 11) (n = 9) (n = 233)

Without JAK2 V617F BM and/or EEC No BM and/or Without JAK2 V617F CALR mutation or CALR (n = 2) EEC (n = 9) CALR mutation or CALR or (n = 5) or MPL mutation (n = 0) MPL mutation (n = 4) No MPN (n = 233) (n = 2) No BM BM No BM BM and/or and/or (n = 3) (n = 1) EEC EEC (n = 61) (n = 172)

Triple negative No MPN Triple negative No MPN MPN (n = 2) (n = 1) MPN (n = 7) (n = 54)

Fig. 1. Flow chart for the test cohort. Unavailable data correspond to patients with platelet count [200Â109/L but without spleen height available. Triple negative MPN patients are patients without JAK2V617F, CALR and MPL mutation, but with a MPN proven by a BM biopsy and/or an EEC. BM, bone marrow; EEC, endogenous erythroid colonies formation; MPN, myeloproliferative neoplasm. hypersplenism and haemodilution related to portal hypertension, when CALR mutations are absent. Six patients with SVT and CALR we chose a lower threshold for platelet count (200Â109/L plate- mutations have been reported so far in the literature, in addition lets) than in patients without SVT (450Â109/L).18,21 to those included in the present study.9,10,12,15 Spleen size and When combining the test and validation cohorts together, the platelet count were mentioned for three of them: all had enlarged criteria ‘‘spleen height P16 cm and platelet count [200Â109/L” spleens and a platelet count [200Â109/L, which provide further had a negative predictive value of 99.7%. Out of the 344 patients evidence supporting the relevance of our findings.9,12 without JAK2V617F, these criteria would have avoided 329 unnec- Due to the rarity of CALR mutations in patients with SVT, we essary CALR mutations tests with only one false negative result. were not able to determine whether these mutations are associ- Combining the two cohorts, four out of the eight patients with ated with a specific pattern of splanchnic vessels involvement or spleen height P16 cm and a platelet count [200Â109/L and no a particular outcome. CALR mutation had a triple negative MPN. These results suggest Our results have several implications. Firstly, identification of that a bone marrow biopsy should be performed in patients with CALR mutations allows the diagnosis of an underlying MPN and, both a spleen height P16 cm and a platelet count [200Â109/L in some cases, can remove the need for bone marrow biopsy,

209 patients with splanchnic vein thrombosis (69 Budd-Chiari syndrome; 140 portal venous system thrombosis)

With JAK2 V617F Without JAK2 V617F (n = 61) (n = 148)

≥ and or Spleen height 16 cm Unavailable data Spleen height <16 cm 9 ≤ 9 platelet count >200x10 /L (n = 46) platelet count 200x10 /L (n = 6) (n = 96)

Without JAK2 V617F Without JAK2 V617F CALR mutation or CALR or CALR mutation CALR mutation or CALR or (n = 2) MPL mutation (n = 1) (n = 1) MPL mutation (n = 4) (n = 95)

Fig. 2. Flow chart for the validation cohort. Unavailable data correspond to patients with platelet count [200Â109/L but without spleen height available. MPN, Myeloproliferative neoplasm.

Journal of Hepatology 2017 vol. 67 j 501–507 505 Research Article

JAK2 V617F testing

Absent Present

Spleen height <16 cm Spleen height ≥16 cm or platelets ≤200x109/L and platelets >200x109/L 20-30% MPN

Consider bone marrow biopsy CALR mutations 2% and/or endogenous erythroid testing 3% colonies formation

Absent Present No 55-65% MPN

Bone marrow biopsy

Fig. 3. Proposed algorithm for the identiﬁcation of myeloproliferative neoplasms (MPNs) in patients with primary splanchnic vein thrombosis (SVT). The ﬁrst step consists of JAK2V617F testing. Patients without JAK2V617F but with spleen height P16 cm and platelet count [200Â109/L should be tested for CALR mutations; if CALR mutations are absent, a bone marrow biopsy should be performed. In the remaining patients, MPNs are extremely uncommon and a bone marrow biopsy and/or endogenous erythroid colonies formation can be considered on a case by case basis.

an invasive procedure. Secondly, the criteria we proposed here Conflict of interest (spleen height was P16 cm and platelet count [200Â109/L) are readily accessible and identify a patient population at high risk The authors who have taken part in this study declared that they of having a MPN (with or without CALR mutations). These do not have anything to disclose regarding funding or conflict of patients should undergo rapid haematological investigations, to interest with respect to this manuscript. consider cytoreductive therapy. Indeed, data from the French network on vascular liver diseases suggest that early introduction of cytoreductive therapy in patients with SVT and MPN reduces Authors’ contributions severe liver-related complications and improves event free sur- 22 vival. Thirdly, by avoiding 96% of unnecessary CALR mutations J.P., and P-E.R. wrote the paper. C.M., and N.C. performed the testing, this strategy will have economic consequences. For mutational screening. A.P., O.G., C.B. and K.Z. collected the clinical instance, in France, based on an incidence of primary SVT of 22 data. F.T., F.C., D.C., and J.C.G.P., collected and analysed the data per million inhabitants (around 2 per million for BCS and 2 per from the Spanish cohort. All authors discussed and critically 23 100, 000 for PVT ), on a population of 67 million inhabitants revised the manuscript. and on a cost of CALR of 124 euros/test, this strategy would save approximately 200,000 euros per year. In conclusion, this study provides the rationale for a new algo- Supplementary data rithm to diagnose MPNs in patients with SVT (Fig. 3). Given its high frequency in this setting, JAK2V617F must be tested first. Supplementary data associated with this article can be found, in Thereafter, patients without JAK2V617F but with spleen height the online version, at http://dx.doi.org/10.1016/j.jhep.2017.04. P16 cm and platelet count [200Â109/L should be tested for 021. CALR mutations. A bone marrow biopsy should be proposed in patients without CALR mutations when the spleen is enlarged and platelets counts are normal or increased. In the remaining patients, namely those without JAK2V617F, when spleen height is References \16 cm and platelet count 6200Â109/L, MPNs are extremely uncommon and further studies are needed to identify those Author names in bold designate shared co-first authorship requiring a bone marrow biopsy. [1] Plessier A, Rautou P-E, Valla D-C. Management of hepatic vascular diseases. J Hepatol 2012;56:S25–S38. [2] Plessier A, Darwish-Murad S, Hernandez-Guerra M, Consigny Y, Fabris F, Financial support Trebicka J, et al. Acute portal vein thrombosis unrelated to cirrhosis: a prospective multicenter follow-up study. Hepatology 2010;51:210–218. [3] Darwish Murad S, Plessier A, Hernandez-Guerra M, Fabris F, Eapen CE, Bahr This work was supported by the Agence Nationale pour la MJ, et al. Etiology, management, and outcome of the Budd-Chiari syndrome. Recherche (ANR 14 CE35 0022 03/JAK-POT) and J.P by the ‘‘poste Ann Intern Med 2009;151:167–175. accueil INSERM”.

506 Journal of Hepatology 2017 vol. 67 j 501–507 JOURNAL OF HEPATOLOGY

[4] Smalberg JH, Arends LR, Valla DC, Kiladjian J-J, Janssen HL, Leebeek FW. [14] Castro N, Rapado I, Ayala R, Martinez-Lopez J. CALR mutations screening Myeloproliferative neoplasms in Budd-Chiari syndrome and portal vein should not be studied in splanchnic vein thrombosis. Br J Haematol thrombosis: a meta-analysis. Blood 2012;120:4921–4928. 2015;170:588–589. [5] Kiladjian J-J, Cervantes F, Leebeek FW, Marzac C, Cassinat B, Chevret S, et al. [15] Roques M, Park J-H, Minello A, Bastie JN, Girodon F. Detection of the CALR The impact of JAK2 and MPL mutations on diagnosis and prognosis of mutation in the diagnosis of splanchnic vein thrombosis. Br J Haematol splanchnic vein thrombosis: a report on 241 cases. Blood 2015;169:601–603. 2008;111:4922–4929. [16] Iurlo A, Cattaneo D, Gianelli U, Fermo E, Augello C, Cortelezzi A. Molecular [6] Nangalia J, Massie CE, Baxter EJ, Nice FL, Gundem G, Wedge DC, et al. Somatic analyses in the diagnosis of myeloproliferative neoplasm-related splanchnic CALR mutations in myeloproliferative neoplasms with nonmutated JAK2.N vein thrombosis. Ann Hematol 2015;94:881–882. Engl J Med 2013;369:2391–2405. [17] Haslam K, Langabeer SE. To screen for CALR mutations in patients with [7] Klampﬂ T, Gisslinger H, Harutyunyan AS, Nivarthi H, Rumi E, Milosevic JD, splanchnic vein thrombosis? Clin Lab 2015;61:441–442. et al. Somatic mutations of calreticulin in myeloproliferative neoplasms. N [18] Chait Y, Condat B, Cazals-Hatem D, Rufat P, Atmani S, Chaoui D, et al. Engl J Med 2013;369:2379–2390. Relevance of the criteria commonly used to diagnose myeloproliferative [8] Turon F, Cervantes F, Colomer D, Baiges A, Hernández-Gea V, Garcia-Pagán disorder in patients with splanchnic vein thrombosis. Br J Haematol JC. Role of calreticulin mutations in the aetiological diagnosis of splanchnic 2005;129:553–560. vein thrombosis. J Hepatol 2015;62:72–74. [19] Denninger MH, Chaït Y, Casadevall N, Hillaire S, Guillin MC, Bezeaud A, et al. [9] Plompen EPC, Valk PJM, Chu I, Murad SD, Plessier A, Turon F, et al. Somatic Cause of portal or hepatic venous thrombosis in adults: the role of multiple calreticulin mutations in patients with Budd-Chiari syndrome and portal concurrent factors. Hepatology 2000;31:587–591. vein thrombosis. Haematologica 2015;100:e226–e228. [20] Valla D, Le MG, Poynard T, Zucman N, Rueff B, Benhamou JP. Risk of hepatic [10] Ho WK, Hong FS. CALR exon 9 mutations in idiopathic splanchnic vein vein thrombosis in relation to recent use of oral contraceptives. A case- thrombosis in an Australian cohort. Thromb Res 2016;150:51–52. control study. Gastroenterology 1986;90:807–811. [11] Zhang X, Zhou M, Chao H, Lu X, Cen L. CALR gene mutation detection and [21] Rumi E, Cazzola M. Diagnosis, risk stratiﬁcation, and response evaluation in clinical observation of 150 essential thrombocythemia patients. Zhonghua classical myeloproliferative neoplasms. Blood 2017;129:680–692. Xue Ye Xue Za Zhi 2015;36:378–382. [22] Chagneau-Derrode C, Roy L, Guilhot J, Gloria O, Ollivier-Hourmand I, Bureau [12] Sekhar M, Patch D, Austen B, Howard J, Hart S. Calreticulin mutations and C. Impact of cytoreductive therapy on the outcome of patients with their importance in splanchnic vein thrombosis. Br J Haematol myeloproliferative neoplasms and hepatosplanchnic vein thrombosis. Hepa- 2016;174:158–160. http://dx.doi.org/10.1111/bjh.13756. tology 2013;58:857A. [13] Colaizzo D, Amitrano L, Guardascione MA, Favuzzi G, Tiscia GL, D’Andrea G, [23] Ageno W, Dentali F, Pomero F, Fenoglio L, Squizzato A, Pagani G, et al. et al. Clinical utility of screening for CALR gene exon 9 mutations in patients Incidence rates and case fatality rates of portal vein thrombosis and Budd- with splanchnic venous thrombosis. Thromb Haemost Chiari Syndrome. Thromb Haemost 2017;117:794–800. 2015;113:1381–1382.

Journal of Hepatology 2017 vol. 67 j 501–507 507 IV. DISCUSSION AND PROSPECTS

A. JAK2V617F in arterial events

The incidence of arterial cardiovascular events are 10 times higher in patients with polycythaemia vera than in the general population [402,470]. As detailed in the introduction, other mechanisms beyond atherosclerosis can be suspected. Indeed, patients with MPNs display a high frequency of myocardial infarction with normal coronary angiography [404]. The mechanism underlying the link between myocardial infarction without obstructive coronary disease and MPNs is unknown, but vasoactive phenomenon (local intense vasoconstriction) can be suspected [393,394]. Therefore, the purpose of the first study was to examine the consequences of JAK2V617F on arterial vascular reactivity.

The first major finding of our study is the demonstration that JAK2V617F MPN induces a considerable increase in arterial contraction. This finding suggests a vasospastic phenomenon associated with MPN and thus represents a paradigm shift in MPNs where arterial events were only seen as a result of a thrombotic process [243]. This mechanism is relevant not only for coronary arteries, but also for brain arteries [400,401,406,408] and in general in any site with non-stenotic atherosclerotic plaques, which is known to favour arterial spasm [393,394]. The mechanisms underlying arterial spasm are not completely elucidated, but arterial contraction plays a central role [400,401,406,407], which is concordant with our findings, since we observed a pronounced increase in contraction in response to different vasoconstrictive agents.

The second major finding of our work is the identification of JAK2V617F RBCs derived microvesicles as responsible for the increased arterial contraction associated with MPNs.

Importantly, we observed this effect with JAK2V617F RBCs microvesicles from mice, but also with microvesicles isolated from patients carrying JAK2V617F. We wanted to focus on the qualitative

137 effect of microvesicles. Therefore we assessed vascular reactivity using the same concentrations of microvesicles for both groups, suggesting that microvesicle composition, and not concentration, accounts for the observed vascular effect. These results are reminiscent of epidemiological studies showing that MPN patients with JAK2V617F have higher haematocrit level [49,69] and a higher risk of cardiovascular events than MPN patients without JAK2V617F

[69,119–122]. In addition, as described in the introduction red blood cells have already been implicated in cardiovascular events in sickle cell disease [250] and more recently in PV patients

[255]. Indeed, under venous flow condition, RBCs from JAK2V617F PV patients display an increased adhesion to endothelial cells in vitro [255], which is mediated by JAK2V617F mutation

[256]. A recent study conducted by Zhao et al [471], demonstrated that the deletion of pleckstrin-2, which plays an important role in erythroid survival and proliferation, in a mouse model of MPNs (VavCre/Jak2V617F), prevents venous thrombosis, mainly by a reduction of

Jak2V617F RBCs mass. These results prove that RBCs are major actors in MPNs cardiovascular events and that JAK2V617F in RBCs is responsible for phenotypic modifications, which support our findings. However, their role has been study only in venous thrombosis and never in arterial events in the context of MPNs. In this matter, our work supports for the first time the importance of RBCs in arterial cardiovascular events in the context of JAK2V617F MPNs.

Finally, in our work we demonstrated that NO pathway inhibition and increased endothelial oxidative stress are implicated in this increased arterial contraction in MPN.

Several groups reported high levels of circulating reactive oxygen species products [452–454] and low antioxidant status in MPN [453] [455], but endothelial oxidative stress had never been investigated.

Our work opens new potential therapeutic perspectives to prevent MPNs’ cardiovascular events. Indeed, statins being known to play a protective role on endothelial function and on oxidative stress, we tested this drug and observed a strong improvement in

138 arterial response to vasocontricting agent in our MPN mouse model [472,473]. We also tested available treatments for MPN and observed that hydroxyurea, but not ruxolitinib, improved arterial contraction. This difference might be explained by the fact that hydroxyurea decreased red blood cell count in our mouse model whereas ruxolitinib did not [474,475]. Another explanation could be that hydroxyurea has been shown to enhance NO release by endothelial cells while such an effect has not been reported with ruxolitinib [476].

Additional works would be useful to further elucidate the pathophysiology of arterial cardiovascular events in MPNs. First, arterial vasospasm should be proven in vivo in our mouse model of MPNs. We could use a previously described model of provoked coronary vasospam using methylergometrine [477]. Secondly, the exact mechanism of how the RBCs’ microvesicles from MPNs increased ROS generation in endothelial cells remains not clearly elucidated.

Dysfunctional mitochondria carried by microvesicles could be an interesting hypothesis.

Indeed, dysfunctional mitochondria are well known ROS inducers [478], ROS have been implicated in cardiovascular disease and endothelial dysfunction for a long time [479] and microvesicles can carry organelles, such as mitochondria [480]. Mature RBCs do not carry mitochondria, but immature RBCs do. An other hypothesis, supported by previous work on sickle cell disease [481–483], could be a NO scavenging by RBCs’ microvesicles containing heme. Finally, Simvastatin is a well-known and easily accessible drug, thus our results could pave the way for testing simvastatin to prevent arterial events in patients with MPNs.

B. JAK2V617F in Budd-Chiari syndrome

Myeloproliferative neoplasms are the leading cause of BCS, diagnosed in 25–50% of such patients [287]. In most patients with BCS and myeloproliferative neoplasms, JAK2V617F mutation is found in myeloid cells. JAK2V617F has also been detected in liver endothelial cells of patients with BCS [45]. In BCS, JAK2V617F is associated with poorer prognostic features at

139 presentation and earlier need for hepatic decompression procedures [287]. Therefore, the second study focused on the consequences of endothelial JAK2V617F in a mouse model of BCS, by partial inferior vena cava ligation.

In our work, the operated mice developed as expected portal hypertension secondary to the stenosis of the inferior vena cava in its supra-hepatic portion. The presence of the JAK2V617F mutation in endothelial cells did not have an impact on the development of this portal hypertension. In addition, the partial inferior vena cava ligation was responsible for an increased hepatic fibrosis, quantified by Collagen 1α1 mRNA expression, αSMA mRNA expression and sirius red staining in comparison with sham mice. These results are consistent with the original article describing this model by Simonetto et al [468] and with the histology of

Budd-Chiari syndrome in humans. However, endothelial JAK2V617F did not influence liver fibrosis. Serum markers of hepatic injury (AST and ALT) and liver inflammation (TNF α mRNA expression) showed no significant increase secondary to surgery, which is also consistent with the original description of the model. Endothelial JAK2V617F had not influence on hepatic injury.

Efficient endothelial recombination in liver endothelial cells was verified in our mice using an mTmG approach in collaboration with the group of C James (Bordeaux, INSERM 1034).

In conclusion, we found no evidence in an animal model that endothelial JAK2V617F can explain the more severe presentation of patients with Budd-Chiari syndrome and JAK2V617F. The explanation for increased severity of these patients should therefore be sought mostly in myeloid JAK2V617F. Thus, future therapeutic strategies to improve the management of patients with Budd-Chiari syndrome and myeloproliferative neoplasms might focus on myeloid cells rather than on endothelial cells. The role of circulating myeloid cells JAK2V617F in the complications of BCS should be assessed in a mouse model of MPNs. However, because of bleeding tendency, the surgery in this model could be difficult. The question remaining

140 unanswered is the implication of endothelial JAK2V617F in the initial development of BCS.

However, no mouse model of spontaneous BCS has yet been described.

C. CALR and splanchnic vein thrombosis

The aim of the last study was to identify the subgroup of patients with SVT at the highest risk of harbouring CALR mutations and thus requiring this genetic testing.

Our work showed that patients without JAK2V617F mutation and SVT, who display a spleen height ≥16 cm and platelet count > 200x109/L should be tested for CALR mutations.

This algorithm had a negative predictive value of 99.7% and avoided 96% of unnecessary test in our study. In this group of patients, a bone marrow biopsy should be proposed when CALR mutations are negative. In the remaining patients, namely those without JAK2V617F, when spleen height is <16 cm or platelet count ≤ 200x109/L, MPNs are extremely uncommon and further studies are needed to identify those requiring a bone marrow biopsy.

Interestingly, Jain et al published a letter to the editor in response to our study in Journal of Hepatology. They suggested that the type of CALR mutation could influence the validity of our findings (see appendix 2). They based their response on the fact that in their cohort of 210 patients with Budd-Chiari syndrome, one patient, with a 3-bp deletion, had a low platelets count and a spleen height < 16cm. However, mutations involving indels occurring as multiples of 3-bp preserve the original reading frame and are not known to be pathogenic (3-6). Thus, the only other patient representing a CALR mutation in their cohort confirmed our algorithm. Our response was published in the journal (see Appendix 2).

Currently, still 2 to 5% of patients with SVT suffer from MPN without JAK2 or CALR mutations. Theoretically, BM biopsy should be performed to more than 50% of patients with

SVT, meaning those without JAK2 or CALR mutations. However, it is not easily feasible and

141 actually never done in real life. We are planning for future work to create a composite score, with morpho-biological parameters that could identify more precisely patients with SVT, who have a high risk of MPNs without JAK2 or CALR mutations to avoid unnecessary BM biopsy.

V. CONCLUSION

In conclusion this thesis work provides new insights in the pathophysiology of cardiovascular events in patients with myeloproliferative neoplasms.

On the arterial side, this work opens a new paradigm in arterial events in the context of

MPNs with possible vasospastic phenomenons. Indeed, JAK2V617F red blood cell derived microvesicles induce a major increase in arterial contraction that could contribute to arterial events associated with MPNs and simvastatin improves this arterial contraction.

On the venous side, this work focuses on splanchnic vein thrombosis in the context of endothelial JAK2V617F and CALR mutations. Endothelial JAK2V617F does not seem to modify presentation of Budd-Chiari syndrome in a mouse model of partial inferior vena cava ligation.

In addition, this work provides a new algorithm to detect CALR mutated patients in the context of splanchnic vein thrombosis.

142

VI. REFERENCES [1] Dameshek W. Editorial: Some Speculations on the Myeloproliferative Syndromes. Blood 1951;6:372–5. [2] Thiele J, Kvasnicka HM. [Chronic myeloproliferative disorders. The new WHO classification]. Pathol 2001;22:429–43. [3] Tefferi A, Vardiman JW. Classification and diagnosis of myeloproliferative neoplasms: the 2008 World Health Organization criteria and point-of-care diagnostic algorithms. Leukemia 2008;22:14–22. [4] Nowell PC, Hungerford DA. Chromosome studies on normal and leukemic human leukocytes. J Natl Cancer Inst 1960;25:85–109. [5] Groffen J, Stephenson JR, Heisterkamp N, Klein A de, Bartram CR, Grosveld G. Philadelphia chromosomal breakpoints are clustered within a limited region, bcr, on chromosome 22. Cell 1984;36:93–9. [6] Arber DA, Orazi A, Hasserjian R, Thiele J, Borowitz MJ, Beau MML, et al. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood 2016;127:2391–405. [7] Moulard Odile, Mehta Jyotsna, Fryzek Jon, Olivares Robert, Iqbal Usman, Mesa Ruben A. Epidemiology of myelofibrosis, essential thrombocythemia, and polycythemia vera in the European Union. Eur J Haematol 2014;92:289–97. [8] Titmarsh Glen J., Duncombe Andrew S., McMullin Mary Frances, O’Rorke Michael, Mesa Ruben, Vocht Frank, et al. How common are myeloproliferative neoplasms? A systematic review and meta-analysis. Am J Hematol 2014;89:581–7. [9] Vaquez H. Sur une forme spéciale de cyanose s’ accompagnant d’hyperglobulie excessive et persistante. CR Soc Biol Paris 1892:vol. 44, 384-388. [10] Osler W. Osler, W. (1903). Chronic cyanosis with polycythemia and enlarged spleen: a new clinical entity. Trans Amer Phycns 1903:vol. 18, 299. [11] Epstein E, Goedel A. Hämorrhagische thrombocythämie bei vasculärer schrumpfmilz. Virchows Arch Für Pathol Anat Physiol Für Klin Med 1934. [12] Heuck G. Zwei Fälle von Leukämie mit eigenthümlichem Blut- resp. Knochenmarksbefund. Arch Für Pathol Anat Physiol Für Klin Med 1879;78:475–96. [13] Tefferi A. Myelofibrosis with Myeloid Metaplasia. N Engl J Med 2000;342:1255–65. [14] Barbui T, Thiele J, Vannucchi AM, Tefferi A. Problems and pitfalls regarding WHO- defined diagnosis of early/prefibrotic primary myelofibrosis versus essential thrombocythemia. Leukemia 2013;27:1953–8. [15] Guglielmelli P, Pacilli A, Rotunno G, Rumi E, Rosti V, Delaini F, et al. Presentation and outcome of patients with 2016 WHO diagnosis of prefibrotic and overt primary myelofibrosis. Blood 2017;129:3227–36. [16] Mudireddy Mythri, Shah Sahrish, Lasho Terra, Barraco Daniela, Hanson Curtis A., Ketterling Rhett P., et al. Prefibrotic versus overtly fibrotic primary myelofibrosis: clinical, cytogenetic, molecular and prognostic comparisons. Br J Haematol 2017;0. [17] Barbui T, Thiele J, Passamonti F, Rumi E, Boveri E, Ruggeri M, et al. Survival and Disease Progression in Essential Thrombocythemia Are Significantly Influenced by Accurate Morphologic Diagnosis: An International Study. J Clin Oncol 2011;29:3179–84. [18] Thiele J, Kvasnicka HM, Müllauer L, Buxhofer-Ausch V, Gisslinger B, Gisslinger H. Essential thrombocythemia versus early primary myelofibrosis: a multicenter study to validate the WHO classification. Blood 2011;117:5710–8.

143 [19] Barbui T, Thiele J, Carobbio A, Passamonti F, Rumi E, Randi ML, et al. Disease characteristics and clinical outcome in young adults with essential thrombocythemia versus early/prefibrotic primary myelofibrosis. Blood 2012;120:569–71. [20] Rupoli S, Goteri G, Picardi P, Micucci G, Canafoglia L, Scortechini AR, et al. Thrombosis in essential thrombocytemia and early/prefibrotic primary myelofibrosis: the role of the WHO histological diagnosis. Diagn Pathol 2015;10. [21] Kishimoto T, Taga T, Akira S. Cytokine signal transduction. Cell 1994;76:253–62. [22] O’Shea JJ, Holland SM, Staudt LM. JAKs and STATs in Immunity, Immunodeficiency, and Cancer. N Engl J Med 2013;368:161–70. [23] Vainchenker W, Kralovics R. Genetic basis and molecular pathophysiology of classical myeloproliferative neoplasms. Blood 2017;129:667–679. [24] Springuel L, Renauld J-C, Knoops L. JAK kinase targeting in hematologic malignancies: a sinuous pathway from identification of genetic alterations towards clinical indications. Haematologica 2015;100:1240–53. [25] Carter-Su C, Schwartz J, Argetsinger LS. Growth hormone signaling pathways. Growth Horm IGF Res 2016;28:11–5. [26] Shuai K, Liu B. Regulation of JAK–STAT signalling in the immune system. Nat Rev Immunol 2003;3:900–11. [27] Babon JJ, Lucet IS, Murphy JM, Nicola NA, Varghese LN. The molecular regulation of Janus kinase (JAK) activation. Biochem J 2014;462:1–13. [28] Cleyrat Cédric, Darehshouri Anza, Steinkamp Mara P., Vilaine Mathias, Boassa Daniela, Ellisman Mark H., et al. Mpl Traffics to the Cell Surface Through Conventional and Unconventional Routes. Traffic 2014;15:961–82. [29] Walker JG, Smith MD. The Jak-STAT pathway in rheumatoid arthritis. J Rheumatol 2005;32:1650–3. [30] Ananthakrishnan R, Hallam K, Li Q, Ramasamy R. JAK-STAT pathway in cardiac ischemic stress. Vascul Pharmacol 2005;43:353–6. [31] Klein S, Rick J, Lehmann J, Schierwagen R, Schierwagen IG, Verbeke L, et al. Janus-kinase- 2 relates directly to portal hypertension and to complications in rodent and human cirrhosis. Gut 2017;66:145–55. [32] Bose P, Gotlib J, Harrison CN, Verstovsek S. SOHO State-of-the-Art Update and Next Questions: MPN. Clin Lymphoma Myeloma Leuk 2018;18:1–12. [33] Li J, Kent DG, Godfrey AL, Manning H, Nangalia J, Aziz A, et al. JAK2V617F homozygosity drives a phenotypic switch in myeloproliferative neoplasms, but is insufficient to sustain disease. Blood 2014;123:3139–3151. [34] Godfrey AL, Chen E, Pagano F, Ortmann CA, Silber Y, Bellosillo B, et al. JAK2V617F homozygosity arises commonly and recurrently in PV and ET, but PV is characterized by expansion of a dominant homozygous subclone. Blood 2012;120:2704–7. [35] Spivak JL. Myeloproliferative Neoplasms. N Engl J Med 2017;376:2168–81. [36] James C, Ugo V, Le Couédic J-P, Staerk J, Delhommeau F, Lacout C, et al. A unique clonal JAK2 mutation leading to constitutive signalling causes polycythaemia vera. Nature 2005;434:1144–1148. [37] Kralovics R, Passamonti F, Buser AS, Teo S-S, Tiedt R, Passweg JR, et al. A Gain-of- Function Mutation of JAK2 in Myeloproliferative Disorders. N Engl J Med 2005;352:1779–90. [38] Levine RL, Wadleigh M, Cools J, Ebert BL, Wernig G, Huntly BJP, et al. Activating mutation in the tyrosine kinase JAK2 in polycythemia vera, essential thrombocythemia, and myeloid metaplasia with myelofibrosis. Cancer Cell 2005;7:387–97. [39] Baxter EJ, Scott LM, Campbell PJ, East C, Fourouclas N, Swanton S, et al. Acquired mutation of the tyrosine kinase JAK2 in human myeloproliferative disorders. The Lancet

144 2005;365:1054–61. [40] Silvennoinen O, Hubbard SR. Molecular insights into regulation of JAK2 in myeloproliferative neoplasms. Blood 2015;125:3388–92. [41] Leroy E, Dusa A, Colau D, Motamedi A, Cahu X, Mouton C, et al. Uncoupling JAK2 V617F activation from cytokine-induced signalling by modulation of JH2 αC helix. Biochem J 2016;473:1579–91. [42] Grinfeld J, Nangalia J, Green AR. Molecular determinants of pathogenesis and clinical phenotype in myeloproliferative neoplasms. Haematologica 2017;102:7–17. [43] Hookham MB, Elliott J, Suessmuth Y, Staerk J, Ward AC, Vainchenker W, et al. The myeloproliferative disorder–associated JAK2 V617F mutant escapes negative regulation by suppressor of cytokine signaling 3. Blood 2007;109:4924–9. [44] Campbell PJ, Green AR. The myeloproliferative disorders. N Engl J Med 2006;355:2452– 66. [45] Sozer S, Fiel MI, Schiano T, Xu M, Mascarenhas J, Hoffman R. The presence of JAK2V617F mutation in the liver endothelial cells of patients with Budd-Chiari syndrome. Blood 2009;113:5246–9. [46] Rosti V, Villani L, Riboni R, Poletto V, Bonetti E, Tozzi L, et al. Spleen endothelial cells from patients with myelofibrosis harbor the JAK2V617F mutation. Blood 2013;121:360–8. [47] Teofili L, Martini M, Iachininoto MG, Capodimonti S, Nuzzolo ER, Torti L, et al. Endothelial progenitor cells are clonal and exhibit the JAK2(V617F) mutation in a subset of thrombotic patients with Ph-negative myeloproliferative neoplasms. Blood 2011;117:2700–7. [48] Kralovics R, Guan Y, Prchal JT. Acquired uniparental disomy of chromosome 9p is a frequent stem cell defect in polycythemia vera. Exp Hematol 2002;30:229–36. [49] Rumi E, Pietra D, Ferretti V, Klampfl T, Harutyunyan AS, Milosevic JD, et al. JAK2 or CALR mutation status defines subtypes of essential thrombocythemia with substantially different clinical course and outcomes. Blood 2014;123:1544–1551. [50] Spivak JL, Considine M, Williams DM, Talbot CC, Rogers O, Moliterno AR, et al. Two Clinical Phenotypes in Polycythemia Vera. N Engl J Med 2014;371:808–17. [51] Dawson MA, Bannister AJ, Göttgens B, Foster SD, Bartke T, Green AR, et al. JAK2 phosphorylates histone H3Y41 and excludes HP1α from chromatin. Nature 2009;461:819–22. [52] Liu F, Zhao X, Perna F, Wang L, Koppikar P, Abdel-Wahab O, et al. JAK2V617F-Mediated Phosphorylation of PRMT5 Downregulates Its Methyltransferase Activity and Promotes Myeloproliferation. Cancer Cell 2011;19:283–94. [53] Scott LM, Tong W, Levine RL, Scott MA, Beer PA, Stratton MR, et al. JAK2 Exon 12 Mutations in Polycythemia Vera and Idiopathic Erythrocytosis. N Engl J Med 2007;356:459–68. [54] Passamonti F, Elena C, Schnittger S, Skoda RC, Green AR, Girodon F, et al. Molecular and clinical features of the myeloproliferative neoplasm associated with JAK2 exon 12 mutations. Blood 2011;117:2813–6. [55] Li S, Kralovics R, Libero GD, Theocharides A, Gisslinger H, Skoda RC. Clonal heterogeneity in polycythemia vera patients with JAK2 exon12 and JAK2-V617F mutations. Blood 2008;111:3863–6. [56] Marty C, Pecquet C, Nivarthi H, El-Khoury M, Chachoua I, Tulliez M, et al. Calreticulin mutants in mice induce an MPL-dependent thrombocytosis with frequent progression to myelofibrosis. Blood 2016;127:1317–24. [57] Cazzola M. Mutant calreticulin: when a chaperone becomes intrusive. Blood 2016;127:1219–21. [58] Klampfl T, Gisslinger H, Harutyunyan AS, Nivarthi H, Rumi E, Milosevic JD, et al. Somatic Mutations of Calreticulin in Myeloproliferative Neoplasms. N Engl J Med 2013;369:2379–90. [59] Nangalia J, Massie CE, Baxter EJ, Nice FL, Gundem G, Wedge DC, et al. Somatic CALR

145 Mutations in Myeloproliferative Neoplasms with Nonmutated JAK2. N Engl J Med 2013;369:2391–405. [60] Vainchenker W, Delhommeau F, Constantinescu SN, Bernard OA. New mutations and pathogenesis of myeloproliferative neoplasms. Blood 2011;118:1723–35.. [61] Pietra D, Rumi E, Ferretti VV, Buduo CAD, Milanesi C, Cavalloni C, et al. Differential clinical effects of different mutation subtypes in CALR-mutant myeloproliferative neoplasms. Leukemia 2016;30:431–8. [62] Szuber N, Lamontagne B, Busque L. Novel germline mutations in the calreticulin gene: implications for the diagnosis of myeloproliferative neoplasms. J Clin Pathol 2016:jclinpath- 2016-203940. [63] Araki M, Yang Y, Masubuchi N, Hironaka Y, Takei H, Morishita S, et al. Activation of the thrombopoietin receptor by mutant calreticulin in CALR-mutant myeloproliferative neoplasms. Blood 2016;127:1307–1316. [64] Chachoua I, Pecquet C, El-Khoury M, Nivarthi H, Albu R-I, Marty C, et al. Thrombopoietin receptor activation by myeloproliferative neoplasm associated calreticulin mutants. Blood 2016;127:1325–35. [65] Elf S, Abdelfattah NS, Chen E, Perales-Patón J, Rosen EA, Ko A, et al. Mutant Calreticulin Requires Both Its Mutant C-terminus and the Thrombopoietin Receptor for Oncogenic Transformation. Cancer Discov 2016;6:368–81. [66] Cabagnols X, Defour JP, Ugo V, Ianotto JC, Mossuz P, Mondet J, et al. Differential association of calreticulin type 1 and type 2 mutations with myelofibrosis and essential thrombocytemia: relevance for disease evolution. Leukemia 2015;29:249–52. [67] Cavalloni C, Rumi E, Ferretti VV, Pietra D, Roncoroni E, Bellini M, et al. Sequential evaluation of CALR mutant burden in patients with myeloproliferative neoplasms. Oncotarget 2017;8:33416–21. [68] Theocharides APA, Lundberg P, Lakkaraju AKK, Lysenko V, Myburgh R, Aguzzi A, et al. Homozygous calreticulin mutations in patients with myelofibrosis lead to acquired myeloperoxidase deficiency. Blood 2016;127:3253–9. [69] Tefferi A, Wassie EA, Lasho TL, Finke C, Belachew AA, Ketterling RP, et al. Calreticulin mutations and long-term survival in essential thrombocythemia. Leukemia 2014;28:2300–3. [70] Assaf CA, Obbergh FV, Billiet J, Lierman E, Devos T, Graux C, et al. Analysis of phenotype and outcome in essential thrombocythemia with CALR or JAK2 mutations. Haematologica 2015;100:893–7. [71] Villeval J-L, Cohen-Solal K, Tulliez M, Giraudier S, Guichard J, Burstein SA, et al. High Thrombopoietin Production by Hematopoietic Cells Induces a Fatal Myeloproliferative Syndrome in Mice. Blood 1997;90:4369–83. [72] Rumi E, Pietra D, Guglielmelli P, Bordoni R, Casetti I, Milanesi C, et al. Acquired copy- neutral loss of heterozygosity of chromosome 1p as a molecular event associated with marrow fibrosis in MPL-mutated myeloproliferative neoplasms. Blood 2013;121:4388–95. [73] Chaligné R, James C, Tonetti C, Besancenot R, Couédic JPL, Fava F, et al. Evidence for MPL W515L/K mutations in hematopoietic stem cells in primitive myelofibrosis. Blood 2007;110:3735–43. [74] Beer PA, Campbell PJ, Scott LM, Bench AJ, Erber WN, Bareford D, et al. MPL mutations in myeloproliferative disorders: analysis of the PT-1 cohort. Blood 2008;112:141–9. [75] Pikman Y, Lee BH, Mercher T, McDowell E, Ebert BL, Gozo M, et al. MPLW515L Is a Novel Somatic Activating Mutation in Myelofibrosis with Myeloid Metaplasia. PLOS Med 2006;3:e270. [76] Varghese LN, Defour J-P, Pecquet C, Constantinescu SN. The Thrombopoietin Receptor: Structural Basis of Traffic and Activation by Ligand, Mutations, Agonists, and Mutated Calreticulin. Front Endocrinol 2017;8.

146 [77] Nangalia J, Green AR. Myeloproliferative neoplasms: from origins to outcomes. Blood 2017;130:2475–83. [78] Yao H, Ma Y, Hong Z, Zhao L, Monaghan SA, Hu M-C, et al. Activating JAK2 mutants reveal cytokine receptor coupling differences that impact outcomes in myeloproliferative neoplasm. Leukemia 2017;31:2122–31. [79] Tiedt R, Hao-Shen H, Sobas MA, Looser R, Dirnhofer S, Schwaller J, et al. Ratio of mutant JAK2-V617F to wild-type Jak2 determines the MPD phenotypes in transgenic mice. Blood 2008;111:3931–40. [80] Akada H, Akada S, Hutchison RE, Mohi G. Loss of wild-type Jak2 allele enhances myeloid cell expansion and accelerates myelofibrosis in Jak2V617F knock-in mice. Leukemia 2014;28:1627–35. [81] Rumi Elisa, Cazzola Mario. Advances in understanding the pathogenesis of familial myeloproliferative neoplasms. Br J Haematol 2017;178:689–98. [82] Saliba J, Saint-Martin C, Di Stefano A, Lenglet G, Marty C, Keren B, et al. Germline duplication of ATG2B and GSKIP predisposes to familial myeloid malignancies. Nat Genet 2015;47:1131–40. [83] Harutyunyan AS, Giambruno R, Krendl C, Stukalov A, Klampfl T, Berg T, et al. Germline RBBP6 mutations in familial myeloproliferative neoplasms. Blood 2016;127:362–5. [84] Loscocco Giuseppe Gaetano, Mannarelli Carmela, Pacilli Annalisa, Fanelli Tiziana, Rotunno Giada, Gesullo Francesca, et al. Germline transmission of LNKE208Q variant in a family with myeloproliferative neoplasms. Am J Hematol 2016;91:E356–E356. [85] Rumi E, Harutyunyan AS, Pietra D, Feenstra JDM, Cavalloni C, Roncoroni E, et al. LNK mutations in familial myeloproliferative neoplasms. Blood 2016;128:144–5. [86] Tapper W, Jones AV, Kralovics R, Harutyunyan AS, Zoi K, Leung W, et al. Genetic variation at MECOM, TERT, JAK2 and HBS1L-MYB predisposes to myeloproliferative neoplasms. Nat Commun 2015;6:6691. [87] Hinds DA, Barnholt KE, Mesa RA, Kiefer AK, Do CB, Eriksson N, et al. Germ line variants predispose to both JAK2 V617F clonal hematopoiesis and myeloproliferative neoplasms. Blood 2016;128:1121–8. [88] Delic Sabit, Rose Dominic, Kern Wolfgang, Nadarajah Niroshan, Haferlach Claudia, Haferlach Torsten, et al. Application of an NGS-based 28-gene panel in myeloproliferative neoplasms reveals distinct mutation patterns in essential thrombocythaemia, primary myelofibrosis and polycythaemia vera. Br J Haematol 2016;175:419–26. [89] Ortmann CA, Kent DG, Nangalia J, Silber Y, Wedge DC, Grinfeld J, et al. Effect of Mutation Order on Myeloproliferative Neoplasms. N Engl J Med 2015;372:601–12. [90] Tefferi A, Guglielmelli P, Larson DR, Finke C, Wassie EA, Pieri L, et al. Long-term survival and blast transformation in molecularly annotated essential thrombocythemia, polycythemia vera, and myelofibrosis. Blood 2014;124:2507–13. [91] Barosi G, Mesa RA, Thiele J, Cervantes F, Campbell PJ, Verstovsek S, et al. Proposed criteria for the diagnosis of post-polycythemia vera and post-essential thrombocythemia myelofibrosis: a consensus statement from the international working group for myelofibrosis research and treatment. Leukemia 2008;22:437–8. [92] Czader M, Orazi A. Acute Myeloid Leukemia and Other Types of Disease Progression in Myeloproliferative Neoplasms. Am J Clin Pathol 2015;144:188–206. [93] Boiocchi L, Mathew S, Gianelli U, Iurlo A, Radice T, Barouk-Fox S, et al. Morphologic and cytogenetic differences between post-polycythemic myelofibrosis and primary myelofibrosis in fibrotic stage. Mod Pathol 2013;26:1577–85. [94] Lundberg P, Karow A, Nienhold R, Looser R, Hao-Shen H, Nissen I, et al. Clonal evolution and clinical correlates of somatic mutations in myeloproliferative neoplasms. Blood

147 2014;123:2220–8. [95] Cervantes F, Tassies D, Salgado C, Rovira M, Pereira A, Rozman C. Acute Transformation in Nonleukemic Chronic Myeloproliferative Disorders: Actuarial Probability and Main Characteristics in a Series of 218 Patients. Acta Haematol 1991;85:124–7. [96] Björkholm M, Derolf ÅR, Hultcrantz M, Kristinsson SY, Ekstrand C, Goldin LR, et al. Treatment-Related Risk Factors for Transformation to Acute Myeloid Leukemia and Myelodysplastic Syndromes in Myeloproliferative Neoplasms. J Clin Oncol 2011;29:2410–5. [97] Abdulkarim Khadija, Girodon François, Johansson Peter, Maynadié Marc, Kutti Jack, Carli Paule-Marie, et al. AML transformation in 56 patients with Ph− MPD in two well defined populations. Eur J Haematol 2008;82:106–11. [98] Heaney ML, Soriano G. Acute Myeloid Leukemia Following a Myeloproliferative Neoplasm: Clinical Characteristics, Genetic Features and Effects of Therapy. Curr Hematol Malig Rep 2013;8:116–22. d [99] Berk PD, Goldberg JD, Silverstein MN, Weinfeld A, Donovan PB, Ellis JT, et al. Increased incidence of acute leukemia in polycythemia vera associated with chlorambucil therapy. N Engl J Med 1981;304:441–7. [100] Najean Y, Rain JD. Treatment of polycythemia vera: the use of hydroxyurea and pipobroman in 292 patients under the age of 65 years. Blood 1997;90:3370–7. [101] Tefferi Ayalew, Barbui Tiziano. Polycythemia vera and essential thrombocythemia: 2015 update on diagnosis, risk-stratification and management. Am J Hematol 2015;90:162–73. [102] Tefferi A, Vannucchi AM, Barbui T. Polycythemia vera treatment algorithm 2018. Blood Cancer J 2018;8:3. [103] Vannucchi AM. How I treat polycythemia vera. Blood 2014;124:3212–20. [104] Hernández-Boluda J-C, Arellano-Rodrigo E, Cervantes F, Alvarez-Larrán A, Gómez M, Barba P, et al. Oral anticoagulation to prevent thrombosis recurrence in polycythemia vera and essential thrombocythemia. Ann Hematol 2015;94:911–8. [105] Fruchtman SM, Mack K, Kaplan ME, Peterson P, Berk PD, Wasserman LR. From efficacy to safety: a Polycythemia Vera Study group report on hydroxyurea in patients with polycythemia vera. Semin Hematol 1997;34:17–23. [106] Finazzi G, Caruso V, Marchioli R, Capnist G, Chisesi T, Finelli C, et al. Acute leukemia in polycythemia vera: an analysis of 1638 patients enrolled in a prospective observational study. Blood 2005;105:2664–70. [107] Gisslinger H, Zagrijtschuk O, Buxhofer-Ausch V, Thaler J, Schloegl E, Gastl GA, et al. Ropeginterferon alfa-2b, a novel IFNα-2b, induces high response rates with low toxicity in patients with polycythemia vera. Blood 2015;126:1762–9. [108] Larsen TS, Bjerrum OW, Pallisgaard N, Andersen MT, Møller MB, Hasselbalch HC. Sustained major molecular response on interferon alpha-2b in two patients with polycythemia vera. Ann Hematol 2008;87:847–50. [109] Kiladjian J-J, Cassinat B, Chevret S, Turlure P, Cambier N, Roussel M, et al. Pegylated interferon-alfa-2a induces complete hematologic and molecular responses with low toxicity in polycythemia vera. Blood 2008;112:3065–72. [110] Mascarenhas JO, Prchal JT, Rambaldi A, Mesa RA, Berenzon D, Yacoub A, et al. Interim Analysis of the Myeloproliferative Disorders Research Consortium (MPD-RC) 112 Global Phase III Trial of Front Line Pegylated Interferon Alpha-2a Vs. Hydroxyurea in High Risk Polycythemia Vera and Essential Thrombocythemia. Blood 2016;128:479–479. [111] Kuriakose ET, Gjoni S, Wang YL, Baumann R, Jones AV, Cross NCP, et al. JAK2V617F allele burden is reduced by busulfan therapy: a new observation using an old drug. Haematologica 2013;98:e135–7. [112] Douglas G, Harrison C, Forsyth C, Bennett M, Stevenson W, Hounsell J, et al. Busulfan is

148 effective second-line therapy for older patients with Philadelphia-negative myeloproliferative neoplasms intolerant of or unresponsive to hydroxyurea. Leuk Lymphoma 2017;58:89–95. [113] Begna K, Abdelatif A, Schwager S, Hanson C, Pardanani A, Tefferi A. Busulfan for the treatment of myeloproliferative neoplasms: the Mayo Clinic experience. Blood Cancer J 2016;6:e427. [114] Barosi G, Birgegard G, Finazzi G, Griesshammer M, Harrison C, Hasselbalch H, et al. A unified definition of clinical resistance and intolerance to hydroxycarbamide in polycythaemia vera and primary myelofibrosis: results of a European LeukemiaNet (ELN) consensus process. Br J Haematol 2010;148:961–3. [115] Vannucchi AM, Kiladjian JJ, Griesshammer M, Masszi T, Durrant S, Passamonti F, et al. Ruxolitinib versus Standard Therapy for the Treatment of Polycythemia Vera. N Engl J Med 2015;372:426–35. [116] Passamonti F, Griesshammer M, Palandri F, Egyed M, Benevolo G, Devos T, et al. Ruxolitinib for the treatment of inadequately controlled polycythaemia vera without splenomegaly (RESPONSE-2): a randomised, open-label, phase 3b study. Lancet Oncol 2017;18:88–99. [117] Vannucchi AM, Barbui T, Cervantes F, Harrison C, Kiladjian J-J, Kröger N, et al. Philadelphia chromosome-negative chronic myeloproliferative neoplasms: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol 2015;26:v85–99. [118] Tefferi A, Vannucchi AM, Barbui T. Essential thrombocythemia treatment algorithm 2018. Blood Cancer J 2018;8:2. [119] Elala YC, Lasho TL, Gangat N, Finke C, Barraco D, Haider M, et al. Calreticulin variant stratified driver mutational status and prognosis in essential thrombocythemia. Am J Hematol 2016;91:503–6. [120] Carobbio A, Thiele J, Passamonti F, Rumi E, Ruggeri M, Rodeghiero F, et al. Risk factors for arterial and venous thrombosis in WHO-defined essential thrombocythemia: an international study of 891 patients. Blood 2011;117:5857–9. [121] Finazzi G, Carobbio A, Guglielmelli P, Cavalloni C, Salmoiraghi S, Vannucchi AM, et al. Calreticulin mutation does not modify the IPSET score for predicting the risk of thrombosis among 1150 patients with essential thrombocythemia. Blood 2014;124:2611–2. [122] Rotunno G, Mannarelli C, Guglielmelli P, Pacilli A, Pancrazzi A, Pieri L, et al. Impact of calreticulin mutations on clinical and hematological phenotype and outcome in essential thrombocythemia. Blood 2014;123:1552–5. [123] Barbui T, Vannucchi AM, Buxhofer-Ausch V, De Stefano V, Betti S, Rambaldi A, et al. Practice-relevant revision of IPSET-thrombosis based on 1019 patients with WHO-defined essential thrombocythemia. Blood Cancer J 2015;5:e369–e369. [124] Haider M, Gangat N, Lasho T, Hussein AKA, Elala YC, Hanson C, et al. Validation of the revised international prognostic score of thrombosis for essential thrombocythemia (IPSET- thrombosis) in 585 Mayo clinic patients. Am J Hematol 2016;91:390–4. [125] Barbui T, Finazzi G, Carobbio A, Thiele J, Passamonti F, Rumi E, et al. Development and validation of an International Prognostic Score of thrombosis in World Health Organization- essential thrombocythemia (IPSET-thrombosis). Blood 2012;120:5128–33; quiz 5252. [126] Cortelazzo S, Viero P, Finazzi G, D’Emilio A, Rodeghiero F, Barbui T. Incidence and risk factors for thrombotic complications in a historical cohort of 100 patients with essential thrombocythemia. J Clin Oncol 1990;8:556–62. [127] Latagliata R, Montanaro M, Cedrone M, Veroli AD, Spirito F, Santoro C, et al. High platelet count at diagnosis is a protective factor for thrombosis in patients with essential thrombocythemia. Thromb Res 2017;156:168–71. [128] Collaborative overview of randomised trials of antiplatelet therapy--III: Reduction in

149 venous thrombosis and pulmonary embolism by antiplatelet prophylaxis among surgical and medical patients. Antiplatelet Trialists’ Collaboration. BMJ 1994;308:235–46. [129] Becattini C, Agnelli G, Schenone A, Eichinger S, Bucherini E, Silingardi M, et al. Aspirin for preventing the recurrence of venous thromboembolism. N Engl J Med 2012;366:1959–67. [130] Barosi G, Besses C, Birgegard G, Briere J, Cervantes F, Finazzi G, et al. A unified definition of clinical resistance/intolerance to hydroxyurea in essential thrombocythemia: results of a consensus process by an international working group. Leukemia 2007;21:277–80. [131] Verger E, Cassinat B, Chauveau A, Dosquet C, Giraudier S, Schlageter M-H, et al. Clinical and molecular response to interferon-α therapy in essential thrombocythemia patients with CALR mutations. Blood 2015;126:2585–91. [132] Cassinat B, Verger E, Kiladjian J-J. Interferon alfa therapy in CALR-mutated essential thrombocythemia. N Engl J Med 2014;371:188–9. [133] Quintás-Cardama A, Abdel-Wahab O, Manshouri T, Kilpivaara O, Cortes J, Roupie A-L, et al. Molecular analysis of patients with polycythemia vera or essential thrombocythemia receiving pegylated interferon α-2a. Blood 2013;122:893–901. [134] Gisslinger H, Gotic M, Holowiecki J, Penka M, Thiele J, Kvasnicka H-M, et al. Anagrelide compared with hydroxyurea in WHO-classified essential thrombocythemia: the ANAHYDRET Study, a randomized controlled trial. Blood 2013;121:1720–8. [135] Harrison CN, Campbell PJ, Buck G, Wheatley K, East CL, Bareford D, et al. Hydroxyurea Compared with Anagrelide in High-Risk Essential Thrombocythemia. N Engl J Med 2005;353:33–45. [136] Birgegård G, Besses C, Griesshammer M, Gugliotta L, Harrison CN, Hamdani M, et al. Treatment of essential thrombocythemia in Europe: a prospective long-term observational study of 3649 high-risk patients in the Evaluation of Anagrelide Efficacy and Long-term Safety study. Haematologica 2018;103:51–60. [137] Tefferi A, Rumi E, Finazzi G, Gisslinger H, Vannucchi AM, Rodeghiero F, et al. Survival and prognosis among 1545 patients with contemporary polycythemia vera: an international study. Leukemia 2013;27:1874–81. [138] Verstovsek S, Passamonti F, Rambaldi A, Barosi G, Rumi E, Gattoni E, et al. Ruxolitinib for essential thrombocythemia refractory to or intolerant of hydroxyurea: long-term phase 2 study results. Blood 2017;130:1768–71. [139] Bose P, Verstovsek S. Ruxolitinib for essential thrombocythemia? Oncoscience 2017;4:148–9. [140] Harrison CN, Mead AJ, Panchal A, Fox S, Yap C, Gbandi E, et al. Ruxolitinib vs best available therapy for ET intolerant or resistant to hydroxycarbamide. Blood 2017;130:1889– 97. [141] Cervantes F, Dupriez B, Pereira A, Passamonti F, Reilly JT, Morra E, et al. New prognostic scoring system for primary myelofibrosis based on a study of the International Working Group for Myelofibrosis Research and Treatment. Blood 2009;113:2895–901. [142] Barbui T, Barosi G, Birgegard G, Cervantes F, Finazzi G, Griesshammer M, et al. Philadelphia-Negative Classical Myeloproliferative Neoplasms: Critical Concepts and Management Recommendations From European LeukemiaNet. J Clin Oncol 2011;29:761–70. [143] Ballen KK, Shrestha S, Sobocinski KA, Zhang M-J, Bashey A, Bolwell BJ, et al. Outcome of transplantation for myelofibrosis. Biol Blood Marrow Transplant J Am Soc Blood Marrow Transplant 2010;16:358–67 [144] Kröger N, Holler E, Kobbe G, Bornhäuser M, Schwerdtfeger R, Baurmann H, et al. Allogeneic stem cell transplantation after reduced-intensity conditioning in patients with myelofibrosis: a prospective, multicenter study of the Chronic Leukemia Working Party of the European Group for Blood and Marrow Transplantation. Blood 2009;114:5264–70.

150 [145] Harrison C, Kiladjian J-J, Al-Ali HK, Gisslinger H, Waltzman R, Stalbovskaya V, et al. JAK Inhibition with Ruxolitinib versus Best Available Therapy for Myelofibrosis. N Engl J Med 2012;366:787–98. [146] Verstovsek S, Mesa RA, Gotlib J, Levy RS, Gupta V, DiPersio JF, et al. A Double-Blind, Placebo-Controlled Trial of Ruxolitinib for Myelofibrosis. N Engl J Med 2012;366:799–807. [147] Harrison CN, Vannucchi AM, Kiladjian J-J, Al-Ali HK, Gisslinger H, Knoops L, et al. Long- term findings from COMFORT-II, a phase 3 study of ruxolitinib vs best available therapy for myelofibrosis. Leukemia 2016;30:1701–7. [148] Verstovsek S, Mesa RA, Gotlib J, Gupta V, DiPersio JF, Catalano JV, et al. Long-term treatment with ruxolitinib for patients with myelofibrosis: 5-year update from the randomized, double-blind, placebo-controlled, phase 3 COMFORT-I trial. J Hematol OncolJ Hematol Oncol 2017;10:55. [149] Verstovsek S, Gotlib J, Mesa RA, Vannucchi AM, Kiladjian J-J, Cervantes F, et al. Long-term survival in patients treated with ruxolitinib for myelofibrosis: COMFORT-I and -II pooled analyses. J Hematol OncolJ Hematol Oncol 2017;10:156 [150] Marchioli R, Finazzi G, Landolfi R, Kutti J, Gisslinger H, Patrono C, et al. Vascular and Neoplastic Risk in a Large Cohort of Patients With Polycythemia Vera. J Clin Oncol 2005;23:2224–32. [151] Tefferi A, Elliott M. Thrombosis in Myeloproliferative Disorders: Prevalence, Prognostic Factors, and the Role of Leukocytes and JAK2V617F. Semin Thromb Hemost 2007;33:313–20. [152] Campbell PJ, Scott LM, Buck G, Wheatley K, East CL, Marsden JT, et al. Definition of subtypes of essential thrombocythaemia and relation to polycythaemia vera based on JAK2 V617F mutation status: a prospective study. The Lancet 2005;366:1945–53. [153] Buxhofer-Ausch Veronika, Gisslinger Heinz, Thiele Jürgen, Gisslinger Bettina, Kvasnicka Hans-Michael, Müllauer Leonhard, et al. Leukocytosis as an important risk factor for arterial thrombosis in WHO-defined early/prefibrotic myelofibrosis: An international study of 264 patients. Am J Hematol 2012;87:669–72. [154] Montanaro Marco, Latagliata Roberto, Cedrone Michele, Spadea Antonio, Rago Angela, Giandomenico Jonny, et al. Thrombosis and survival in essential thrombocythemia: A regional study of 1,144 patients. Am J Hematol 2014;89:542–6. [155] Barbui T, Carobbio A, Cervantes F, Vannucchi AM, Guglielmelli P, Antonioli E, et al. Thrombosis in primary myelofibrosis: incidence and risk factors. Blood 2010;115:778–82. [156] Stefano VD, Martinelli I. Splanchnic vein thrombosis: clinical presentation, risk factors and treatment. Intern Emerg Med 2010;5:487–94. [157] Lippi G, Franchini M, Targher G. Arterial thrombus formation in cardiovascular disease. Nat Rev Cardiol 2011;8:502–12. [158] Franchini M, Mannucci PM. Venous and arterial thrombosis: Different sides of the same coin? Eur J Intern Med 2008;19:476–81. [159] Jerjes-Sanchez C. Venous and arterial thrombosis: a continuous spectrum of the same disease? Eur Heart J 2005;26:3–4. [160] De Stefano Valerio, Rossi Elena, Za Tommaso, Ciminello Angela, Betti Silvia, Luzzi Claudia, et al. JAK2 V617F mutational frequency in essential thrombocythemia associated with splanchnic or cerebral vein thrombosis. Am J Hematol 2011;86:526–8. [161] Carobbio A, Finazzi G, Antonioli E, Guglielmelli P, Vannucchi AM, Dellacasa CM, et al. JAK2V617F allele burden and thrombosis: A direct comparison in essential thrombocythemia and polycythemia vera. Exp Hematol 2009;37:1016–21. [162] Ziakas PD. Effect of JAK2 V617F on thrombotic risk in patients with essential thrombocythemia: measuring the uncertain. Haematologica 2008;93:1412–4. [163] Dahabreh IJ, Zoi K, Giannouli S, Zoi C, Loukopoulos D, Voulgarelis M. Is JAK2 V617F

151 mutation more than a diagnostic index? Leuk Res 2009;33:67–73. [164] Lussana F, Caberlon S, Pagani C, Kamphuisen PW, Büller HR, Cattaneo M. Association of V617F Jak2 mutation with the risk of thrombosis among patients with essential thrombocythaemia or idiopathic myelofibrosis: A systematic review. Thromb Res 2009;124:409–17. [165] Alvarez-Larran A, Pereira A, Guglielmelli P, Hernandez-Boluda JC, Arellano-Rodrigo E, Ferrer-Marin F, et al. Antiplatelet therapy versus observation in low-risk essential thrombocythemia with a CALR mutation. Haematologica 2016;101:926–31. [166] Pei Y-Q, Wu Y, Wang F, Cui W. Prognostic value of CALR vs. JAK2V617F mutations on splenomegaly, leukemic transformation, thrombosis, and overall survival in patients with primary fibrosis: a meta-analysis. Ann Hematol 2016;95:1391–8. [167] Landolfi R, Di Gennaro L, Barbui T, De Stefano V, Finazzi G, Marfisi R, et al. Leukocytosis as a major thrombotic risk factor in patients with polycythemia vera. Blood 2007;109:2446–52. [168] De Stefano Valerio, Za Tommaso, Rossi Elena, Vannucchi Alessandro M., Ruggeri Marco, Elli Elena, et al. Leukocytosis is a risk factor for recurrent arterial thrombosis in young patients with polycythemia vera and essential thrombocythemia. Am J Hematol 2009;85:97–100. doi:10.1002/ajh.21593. [169] Barbui T, Masciulli A, Marfisi MR, Tognoni G, Finazzi G, Rambaldi A, et al. White blood cell counts and thrombosis in polycythemia vera: a subanalysis of the CYTO-PV study. Blood 2015;126:560–1. [170] Gangat Naseema, Strand Jacob, Li Chin-Yang, Wu Wenting, Pardanani Animesh, Tefferi Ayalew. Leucocytosis in polycythaemia vera predicts both inferior survival and leukaemic transformation. Br J Haematol 2007;138:354–8. [171] Passamonti F, Rumi E, Pascutto C, Cazzola M, Lazzarino M. Increase in leukocyte count over time predicts thrombosis in patients with low-risk essential thrombocythemia. J Thromb Haemost JTH 2009;7:1587–9. [172] Passamonti F, Thiele J, Girodon F, Rumi E, Carobbio A, Gisslinger H, et al. A prognostic model to predict survival in 867 World Health Organization–defined essential thrombocythemia at diagnosis: a study by the International Working Group on Myelofibrosis Research and Treatment. Blood 2012;120:1197–201. [173] Rumi E, Cazzola M. Diagnosis, risk stratification, and response evaluation in classical myeloproliferative neoplasms. Blood 2017;129:680–92. [174] Tefferi A, Barbui T. Polycythemia vera and essential thrombocythemia: 2017 update on diagnosis, risk-stratification, and management. Am J Hematol 2017;92:94–108. [175] Caramazza D, Caracciolo C, Barone R, Malato A, Saccullo G, Cigna V, et al. Correlation between leukocytosis and thrombosis in Philadelphia-negative chronic myeloproliferative neoplasms. Ann Hematol 2009;88:967–71. [176] Buxhofer-Ausch Veronika, Gisslinger Bettina, Schalling Martin, Gleiss Andreas, Schiefer Ana-Iris, Müllauer Leonhard, et al. Impact of white blood cell counts at diagnosis and during follow-up in patients with essential thrombocythaemia and prefibrotic primary myelofibrosis. Br J Haematol 2016;179:166–9. [177] Campbell PJ, MacLean C, Beer PA, Buck G, Wheatley K, Kiladjian J-J, et al. Correlation of blood counts with vascular complications in essential thrombocythemia: analysis of the prospective PT1 cohort. Blood 2012;120:1409–11. [178] Carobbio A, Antonioli E, Guglielmelli P, Vannucchi AM, Delaini F, Guerini V, et al. Leukocytosis and Risk Stratification Assessment in Essential Thrombocythemia. J Clin Oncol 2008;26:2732–6. [179] Tefferi A, Gangat N, Wolanskyj AP. Management of extreme thrombocytosis in otherwise low-risk essential thrombocythemia; does number matter? Blood 2006;108:2493–4.

152 [180] Di Nisio Marcello, Barbui Tiziano, Di Gennaro Leonardo, Borrelli Giovanna, Finazzi Guido, Landolfi Raffaele, et al. The haematocrit and platelet target in polycythemia vera. Br J Haematol 2006;136:249–59. [181] Gugliotta L, Iurlo A, Gugliotta G, Tieghi A, Specchia G, Gaidano G, et al. Unbiased pro- thrombotic features at diagnosis in 977 thrombocythemic patients with Philadelphia-negative chronic myeloproliferative neoplasms. Leuk Res 2016;46:18–25. [182] Michiels JJ, Berneman Z, Schroyens W, Finazzi G, Budde U, Vliet HHDM van. The Paradox of Platelet Activation and Impaired Function: Platelet-von Willebrand Factor Interactions, and the Etiology of Thrombotic and Hemorrhagic Manifestations in Essential Thrombocythemia and Polycythemia Vera. Semin Thromb Hemost 2006;32:589–604. [183] Regev Arie, Stark Pinhas, Blickstein Dorit, Lahav Meir. Thrombotic complications in essential thrombocythemia with relatively low platelet counts. Am J Hematol 1998;56:168–72. [184] Falchi Lorenzo, Bose Prithviraj, Newberry Kate J., Verstovsek Srdan. Approach to patients with essential thrombocythaemia and very high platelet counts: what is the evidence for treatment? Br J Haematol 2017;176:352–64. [185] Castaman G, Lattuada A, Ruggeri M, Tosetto A, Mannucci PM, Rodeghiero F. Platelet von Willebrand factor abnormalities in myeloproliferative syndromes. Am J Hematol 1995;49:289– 93. [186] Turitto VT, Weiss HJ. Red blood cells: their dual role in thrombus formation. Science 1980;207:541–3 [187] Adams BD, Baker R, Lopez JA, Spencer S. Myeloproliferative disorders and the hyperviscosity syndrome. Hematol Oncol Clin North Am 2010;24:585–602. [188] Byrnes JR, Wolberg AS. Red blood cells in thrombosis. Blood 2017;130:1795–9. [189] Marchioli R, Finazzi G, Specchia G, Cacciola R, Cavazzina R, Cilloni D, et al. Cardiovascular events and intensity of treatment in polycythemia vera. N Engl J Med 2013;368:22–33. [190] Félétou M. Introduction. Morgan & Claypool Life Sciences; 2011. [191] Verhamme P, Hoylaerts MF. The pivotal role of the endothelium in haemostasis and thrombosis. Acta Clin Belg 2006;61:213–9. [192] Félétou M. The Endothelium. Morgan & Claypool Life Sciences Publisher; 2011. [193] Jaffe EA, Hoyer LW, Nachman RL. Synthesis of von Willebrand factor by cultured human endothelial cells. Proc Natl Acad Sci U S A 1974;71:1906–9. [194] Savage B, Saldívar E, Ruggeri ZM. Initiation of platelet adhesion by arrest onto fibrinogen or translocation on von Willebrand factor. Cell 1996;84:289–97. [195] Carlos TM, Harlan JM. Leukocyte-endothelial adhesion molecules. Blood 1994;84:2068– 101. [196] Spiel AO, Gilbert JC, Jilma B. von Willebrand factor in cardiovascular disease: focus on acute coronary syndromes. Circulation 2008;117:1449–59. [197] Lip GY, Blann A. von Willebrand factor: a marker of endothelial dysfunction in vascular disorders? Cardiovasc Res 1997;34:255–65. [198] Cella G, Marchetti M, Vianello F, Panova-Noeva M, Vignoli A, Russo L, et al. Nitric oxide derivatives and soluble plasma selectins in patients with myeloproliferative neoplasms. Thromb Haemost 2010;104:151–6. [199] Piccin A, Steurer M, Feistritzer C, Murphy C, Eakins E, Van Schilfgaarde M, et al. Observational retrospective study of vascular modulator changes during treatment in essential thrombocythemia. Transl Res 2017;184:21–34. [200] Karakantza M, Giannakoulas NC, Zikos P, Sakellaropoulos G, Kouraklis A, Aktypi A, et al. Markers of Endothelial and In Vivo Platelet Activation in Patients with Essential Thrombocythemia and Polycythemia Vera. Int J Hematol 2004;79:253–9. [201] Belotti Angelo, Elli Elena, Speranza Tiziana, Lanzi Eraldo, Pioltelli Pietro, Pogliani Enrico.

153 Circulating endothelial cells and endothelial activation in essential thrombocythemia: Results from CD146+ immunomagnetic enrichment—flow cytometry and soluble E-selectin detection. Am J Hematol 2012;87:319–20. [202] Jensen MK, de Nully Brown P, Thorsen S, Hasselbalch HC. Frequent occurrence of anticardiolipin antibodies, Factor V Leiden mutation, and perturbed endothelial function in chronic myeloproliferative disorders. Am J Hematol 2002;69:185–91. [203] Falanga A, Marchetti M, Evangelista V, Vignoli A, Licini M, Balicco M, et al. Polymorphonuclear leukocyte activation and hemostasis in patients with essential thrombocythemia and polycythemia vera. Blood 2000;96:4261–6. [204] Santilli F, Romano M, Recchiuti A, Dragani A, Falco A, Lessiani G, et al. Circulating endothelial progenitor cells and residual in vivo thromboxane biosynthesis in low-dose aspirin- treated polycythemia vera patients. Blood 2008;112:1085–90. [205] Lancellotti S, Dragani A, Ranalli P, Petrucci G, Basso M, Tartaglione R, et al. Qualitative and quantitative modifications of von Willebrand factor in patients with essential thrombocythemia and controlled platelet count. J Thromb Haemost JTH 2015;13:1226–37. [206] Prater DN, Case J, Ingram DA, Yoder MC. Working hypothesis to redefine endothelial progenitor cells. Leukemia 2007;21:1141–9. [207] Alonci A, Allegra A, Bellomo G, Penna G, D’Angelo A, Quartarone E, et al. Evaluation of circulating endothelial cells, VEGF and VEGFR2 serum levels in patients with chronic myeloproliferative diseases. Hematol Oncol 2008;26:235–9. [208] Oppliger Leibundgut E, Horn MP, Brunold C, Pfanner-Meyer B, Marti D, Hirsiger H, et al. Hematopoietic and endothelial progenitor cell trafficking in patients with myeloproliferative diseases. Haematologica 2006;91:1465–72. [209] Massa M, Rosti V, Ramajoli I, Campanelli R, Pecci A, Viarengo G, et al. Circulating CD34+, CD133+, and Vascular Endothelial Growth Factor Receptor 2–Positive Endothelial Progenitor Cells in Myelofibrosis With Myeloid Metaplasia. J Clin Oncol 2005;23:5688–95. [210] Torres C, Fonseca AM, Leander M, Matos R, Morais S, Campos M, et al. Circulating Endothelial Cells in Patients with Venous Thromboembolism and Myeloproliferative Neoplasms. PLOS ONE 2013;8:e81574. [211] Rosti V, Bonetti E, Bergamaschi G, Campanelli R, Guglielmelli P, Maestri M, et al. High Frequency of Endothelial Colony Forming Cells Marks a Non-Active Myeloproliferative Neoplasm with High Risk of Splanchnic Vein Thrombosis. PLOS ONE 2010;5:e15277. [212] Helman Ricardo, Pereira Welbert de O., Marti Luciana C., Campregher Paulo V., Puga Renato Davi, Hamerschlak Nelson, et al. Granulocyte whole exome sequencing and endothelial JAK2V617F in patients with JAK2V617F positive Budd-Chiari Syndrome without myeloproliferative neoplasm. Br J Haematol 2018;180:443–5. [213] Guy A, Gourdou-Latyszenok V, Le Lay N, Vieira Dias J, Kilani B, Peghaire C, et al. Endothelial cells are important in the pathogenesis of thrombosis in myeloproliferative neoplasms 2017. [214] Etheridge SL, Roh ME, Cosgrove ME, Sangkhae V, Fox NE, Chen J, et al. JAK2V617F- positive endothelial cells contribute to clotting abnormalities in myeloproliferative neoplasms. Proc Natl Acad Sci U S A 2014;111:2295–300. [215] Jensen MK, de Nully Brown P, Lund BV, Nielsen OJ, Hasselbalch HC. Increased platelet activation and abnormal membrane glycoprotein content and redistribution in myeloproliferative disorders. Br J Haematol 2000;110:116–24. [216] Blann A, Caine G, Bareford D. Abnormal vascular, platelet and coagulation markers in primary thrombocythaemia are not reversed by treatments that reduce the platelet count. Platelets 2004;15:447–9. [217] Arellano-Rodrigo E, Alvarez-Larrán A, Reverter JC, Villamor N, Colomer D, Cervantes F.

154 Increased platelet and leukocyte activation as contributing mechanisms for thrombosis in essential thrombocythemia and correlation with the JAK2 mutational status. Haematologica 2006;91:169–75. [218] Robertson B, Urquhart C, Ford I, Townend J, Watson HG, Vickers MA, et al. Platelet and coagulation activation markers in myeloproliferative diseases: relationships with JAK2 V6I7 F status, clonality, and antiphospholipid antibodies. J Thromb Haemost JTH 2007;5:1679–85. [219] Panova-Noeva M, Marchetti M, Spronk HM, Russo L, Diani E, Finazzi G, et al. Platelet- induced thrombin generation by the calibrated automated thrombogram assay is increased in patients with essential thrombocythemia and polycythemia vera. Am J Hematol 2011;86:337– 42. [220] Falanga A, Marchetti M, Vignoli A, Balducci D, Russo L, Guerini V, et al. V617F JAK-2 mutation in patients with essential thrombocythemia: relation to platelet, granulocyte, and plasma hemostatic and inflammatory molecules. Exp Hematol 2007;35:702–11. [221] Villmow T, Kemkes-Matthes B, Matzdorff AC. Markers of platelet activation and platelet- leukocyte interaction in patients with myeloproliferative syndromes. Thromb Res 2002;108:139–45. [222] Yao H-X, Huang L, Wu C-M, Lin L-E, Huang Z-Q, Wu J-F, et al. [Effects of sodium ozagrel in primary thrombocytosis combined with thrombosis]. Zhongguo Shi Yan Xue Ye Xue Za Zhi 2009;17:1360–2. [223] Falanga A, Marchetti M, Barbui T, Smith CW. Pathogenesis of Thrombosis in Essential Thrombocythemia and Polycythemia Vera: The Role of Neutrophils. Semin Hematol 2005;42:239–47. [224] Jensen MK, de Nully Brown P, Lund BV, Nielsen OJ, Hasselbalch HC. Increased circulating platelet-leukocyte aggregates in myeloproliferative disorders is correlated to previous thrombosis, platelet activation and platelet count. Eur J Haematol 2001;66:143–51. [225] Arellano-Rodrigo E, Alvarez-Larrán A, Reverter J-C, Colomer D, Villamor N, Bellosillo B, et al. Platelet turnover, coagulation factors, and soluble markers of platelet and endothelial activation in essential thrombocythemia: Relationship with thrombosis occurrence and JAK2 V617F allele burden. Am J Hematol 2009;84:102–8. [226] Landolfi R, Rocca B, Patrono C. Bleeding and thrombosis in myeloproliferative disorders: mechanisms and treatment. Crit Rev Oncol Hematol 1995;20:203–22. [227] Wehmeier A, Südhoff T, Meierkord F. Relation of platelet abnormalities to thrombosis and hemorrhage in chronic myeloproliferative disorders. Semin Thromb Hemost 1997;23:391– 402. [228] Le Blanc K, Berg A, Palmblad J, Samuelsson J. Stimulus-specific defect in platelet aggregation in polycythemia vera. Eur J Haematol 1994;53:145–9. [229] Hosseini E, Ghasemzadeh M, Nassaji F, Jamaat ZP. GPVI modulation during platelet activation and storage: its expression levels and ectodomain shedding compared to markers of platelet storage lesion. Platelets 2017;28:498–508. [230] Ghasemzadeh M, Hosseini E, Roudsari ZO, Zadkhak P. Intraplatelet reactive oxygen species (ROS) correlate with the shedding of adhesive receptors, microvesiculation and platelet adhesion to collagen during storage: Does endogenous ROS generation downregulate platelet adhesive function? Thromb Res 2018;163:153–61. [231] Hobbs CM, Manning H, Bennett C, Vasquez L, Severin S, Brain L, et al. JAK2V617F leads to intrinsic changes in platelet formation and reactivity in a knock-in mouse model of essential thrombocythemia. Blood 2013;122:3787–97. [232] Lamrani L, Lacout C, Ollivier V, Denis CV, Gardiner E, Ho Tin Noe B, et al. Hemostatic disorders in a JAK2V617F-driven mouse model of myeloproliferative neoplasm. Blood 2014;124:1136–45.

155 [233] Calaminus SDJ, Guitart A, Sinclair A, Schachtner H, Watson SP, Holyoake TL, et al. Lineage Tracing of Pf4-Cre Marks Hematopoietic Stem Cells and Their Progeny. PLoS ONE 2012;7. [234] Pertuy F, Aguilar A, Strassel C, Eckly A, Freund J-N, Duluc I, et al. Broader expression of the mouse platelet factor 4-cre transgene beyond the megakaryocyte lineage. J Thromb Haemost n.d.;13:115–25. [235] Strassel C, Kubovcakova L, Mangin PH, Ravanat C, Freund M, Skoda RC, et al. Haemorrhagic and thrombotic diatheses in mouse models with thrombocytosis. Thromb Haemost 2015;113:414–25. [236] Musolino C, Alonci A, Bellomo G, Tringali O, Spatari G, Quartarone C, et al. Markers of Endothelial and Platelet Status in Patients with Essential Thrombocythemia and Polycythemia Vera. Hematology 1999;4:397–402. [237] Grignani C, Noris P, Tinelli C, Barosi G, Balduini CL. In vitro platelet aggregation defects in patients with myeloproliferative disorders and high platelet counts: are they laboratory artefacts? Platelets 2009;20:131–4. [238] Mackman N. New insights into the mechanisms of venous thrombosis. J Clin Invest 2012;122:2331–6. [239] von Brühl M-L, Stark K, Steinhart A, Chandraratne S, Konrad I, Lorenz M, et al. Monocytes, neutrophils, and platelets cooperate to initiate and propagate venous thrombosis in mice in vivo. J Exp Med 2012;209:819–35. [240] Falanga A, Marchetti M. Thrombotic disease in the myeloproliferative neoplasms. ASH Educ Program Book 2012;2012:571–81. [241] Marchetti M, Falanga A. Leukocytosis, JAK2V617F mutation, and hemostasis in myeloproliferative disorders. Pathophysiol Haemost Thromb 2008;36:148–59. [242] Carulli G, Minnucci S, Gianfaldoni ML, Angiolini C, Azzarà A, Ambrogi F. Interactions between platelets and neutrophils in essential thrombocythaemia. Effects on neutrophil chemiluminescence and superoxide anion generation. Eur J Clin Invest 1995;25:929–34. [243] Barbui T, Finazzi G, Falanga A. Myeloproliferative neoplasms and thrombosis. Blood 2013;122:2176–84. [244] Augello C, Cattaneo D, Bucelli C, Terrasi A, Fermo E, Martinelli I, et al. CD18 promoter methylation is associated with a higher risk of thrombotic complications in primary myelofibrosis. Ann Hematol 2016;95:1965–9. [245] Brinkmann V, Reichard U, Goosmann C, Fauler B, Uhlemann Y, Weiss DS, et al. Neutrophil Extracellular Traps Kill Bacteria. Science 2004;303:1532–5. [246] Martinod K, Wagner DD. Thrombosis: tangled up in NETs. Blood 2014;123:2768–76. [247] Demers M, Krause DS, Schatzberg D, Martinod K, Voorhees JR, Fuchs TA, et al. Cancers predispose neutrophils to release extracellular DNA traps that contribute to cancer-associated thrombosis. Proc Natl Acad Sci 2012;109:13076–81 [248] Wolach O, Sellar RS, Martinod K, Cherpokova D, McConkey M, Chappell RJ, et al. Increased neutrophil extracellular trap formation promotes thrombosis in myeloproliferative neoplasms. Sci Transl Med 2018;10. [249] Marin Oyarzún CP, Carestia A, Lev PR, Glembotsky AC, Castro Ríos MA, Moiraghi B, et al. Neutrophil extracellular trap formation and circulating nucleosomes in patients with chronic myeloproliferative neoplasms. Sci Rep 2016;6:38738. [250] Colin Y, Van Kim CL, El Nemer W. Red cell adhesion in human diseases: Curr Opin Hematol 2014;21:186–92. [251] Cines DB, Lebedeva T, Nagaswami C, Hayes V, Massefski W, Litvinov RI, et al. Clot contraction: compression of erythrocytes into tightly packed polyhedra and redistribution of platelets and fibrin. Blood 2014;123:1596–603. [252] Krauss S. Haptoglobin metabolism in polycythemia vera. Blood 1969;33:865–76.

156 [253] Strati P, Masarova L, Bose P, Daver N, Pemmaraju N, Verstovsek S. Haptoglobin is frequently low in patients with myelofibrosis: Clinical relevance. Leuk Res 2017;57:85–8. [254] Rusak T, Misztal T, Piszcz J, Tomasiak M. Nitric oxide scavenging by cell-free hemoglobin may be a primary factor determining hypertension in polycythemic patients. Free Radic Res 2014;48:230–8. [255] Wautier M-P, Nemer WE, Gane P, Rain J-D, Cartron J-P, Colin Y, et al. Increased adhesion to endothelial cells of erythrocytes from patients with polycythemia vera is mediated by laminin α5 chain and Lu/BCAM. Blood 2007;110:894–901. [256] Grandis MD, Cambot M, Wautier M-P, Cassinat B, Chomienne C, Colin Y, et al. JAK2V617F activates Lu/BCAM-mediated red cell adhesion in polycythemia vera through an EpoR- independent Rap1/Akt pathway. Blood 2013;121:658–65. [257] Rautou P-E, Vion A-C, Amabile N, Chironi G, Simon A, Tedgui A, et al. Microparticles, Vascular Function, and Atherothrombosis. Circ Res 2011;109:593–606. [258] Boulanger CM, Loyer X, Rautou P-E, Amabile N. Extracellular vesicles in coronary artery disease. Nat Rev Cardiol 2017;14:259–72. [259] Wood JP, Ellery PER, Maroney SA, Mast AE. Protein S is a cofactor for platelet and endothelial tissue factor pathway inhibitor-α but not for cell surface-associated tissue factor pathway inhibitor. Arterioscler Thromb Vasc Biol 2014;34:169–76. [260] Tong D, Yu M, Guo L, Li T, Li J, Novakovic VA, et al. Phosphatidylserine-exposing blood and endothelial cells contribute to the hypercoagulable state in essential thrombocythemia patients. Ann Hematol 2018;97:605–16. [261] Duchemin J, Ugo V, Ianotto J-C, Lecucq L, Mercier B, Abgrall J-F. Increased circulating procoagulant activity and thrombin generation in patients with myeloproliferative neoplasms. Thromb Res 2010;126:238–42. [262] Marchetti M, Castoldi E, Spronk HMH, van Oerle R, Balducci D, Barbui T, et al. Thrombin generation and activated protein C resistance in patients with essential thrombocythemia and polycythemia vera. Blood 2008;112:4061–8. [263] Tripodi A, Chantarangkul V, Gianniello F, Clerici M, Lemma L, Padovan L, et al. Global coagulation in myeloproliferative neoplasms. Ann Hematol 2013;92:1633–9. [264] Trappenburg MC, van Schilfgaarde M, Marchetti M, Spronk HM, ten Cate H, Leyte A, et al. Elevated procoagulant microparticles expressing endothelial and platelet markers in essential thrombocythemia. Haematologica 2009;94:911–8. [265] Kogan I, Chap D, Hoffman R, Axelman E, Brenner B, Nadir Y. JAK-2 V617F mutation increases heparanase procoagulant activity. Thromb Haemost 2016;115:73–80. [266] Grover SP, Mackman N. Tissue Factor: An Essential Mediator of Hemostasis and Trigger of Thrombosis. Arterioscler Thromb Vasc Biol 2018;38:709–25. [267] Carmeliet P, Stassen JM, Schoonjans L, Ream B, van den Oord JJ, De Mol M, et al. Plasminogen activator inhibitor-1 gene-deficient mice. II. Effects on hemostasis, thrombosis, and thrombolysis. J Clin Invest 1993;92:2756–60. [268] Friedenberg WR, Roberts RC, David DE. Relationship of thrombohemorrhagic complications to endothelial cell function in patients with chronic myeloproliferative disorders. Am J Hematol 1992;40:283–9. [269] Sonmez M, Saglam F, Karahan SC, Erkut N, Mentese A, Sonmez B, et al. Treatment related changes in antifibrinolytic activity in patients with polycythemia vera. Hematology 2010;15:391–6. [270] Kaftan O, Balcik OS, Cipil H, Ozet G, Bavbek N, Koşar A, et al. Plasma levels of thrombin- activatable fibrinolysis inhibitor in primary and secondary thrombocytosis. Clin Appl Thromb Off J Int Acad Clin Appl Thromb 2005;11:449–54. [271] Birdane A, Haznedaroğlu IC, Bavbek N, Koşar A, Büyükaşik Y, Ozcebe O, et al. The plasma

157 levels of prostanoids and plasminogen activator inhibitor-1 in primary and secondary thrombocytosis. Clin Appl Thromb Off J Int Acad Clin Appl Thromb 2005;11:197–201. [272] Pósán E, Ujj G, Kiss A, Telek B, Rák K, Udvardy M. Reduced in vitro clot lysis and release of more active platelet PAI-1 in polycythemia vera and essential thrombocythemia. Thromb Res 1998;90:51–6. [273] Małecki R, Gacka M, Kuliszkiewicz-Janus M, Jakobsche-Policht U, Kwiatkowski J, Adamiec R, et al. Altered plasma fibrin clot properties in essential thrombocythemia. Platelets 2016;27:110–6. [274] Rusak T, Piszcz J, Misztal T, Brańska-Januszewska J, Tomasiak M. Platelet-related fibrinolysis resistance in patients suffering from PV. Impact of clot retraction and isovolemic erythrocytapheresis. Thromb Res 2014;134:192–8. [275] Aird WC. Vascular bed-specific thrombosis. J Thromb Haemost JTH 2007;5 Suppl 1:283– 91. [276] Rosenberg RD, Aird WC. Vascular-bed--specific hemostasis and hypercoagulable states. N Engl J Med 1999;340:1555–64. [277] Yuan L, Chan GC, Beeler D, Janes L, Spokes KC, Dharaneeswaran H, et al. A role of stochastic phenotype switching in generating mosaic endothelial cell heterogeneity. Nat Commun 2016;7:10160. [278] Cheng Z, Fu J, Liu G, Zhang L, Xu Q, Wang S. Angiogenesis in JAK2 V617F positive myeloproliferative neoplasms and ruxolitinib decrease VEGF, HIF-1 enesis in JAK2 V617F positive cells. Leuk Lymphoma 2018;59:196–203. [279] Zangari M, Fink L, Tolomelli G, Lee JCH, Stein BL, Hickman K, et al. Could hypoxia increase the prevalence of thrombotic complications in polycythemia vera? Blood Coagul Fibrinolysis Int J Haemost Thromb 2013;24:311–6. [280] Plessier A, Rautou P-E, Valla D-C. Management of hepatic vascular diseases. J Hepatol 2012;56:S25–S38. [281] Ageno W, Dentali F, Pomero F, Fenoglio L, Squizzato A, Pagani G, et al. Incidence rates and case fatality rates of portal vein thrombosis and Budd-Chiari Syndrome. Thromb Haemost 2017;117:794–800. [282] Ollivier-Hourmand I, Allaire M, Goutte N, Morello R, Chagneau-Derrode C, Goria O, et al. The epidemiology of Budd-Chiari syndrome in France. Dig Liver Dis Off J Ital Soc Gastroenterol Ital Assoc Study Liver 2018. [283] Darwish Murad S, Plessier A, Hernandez-Guerra M, Fabris F, Eapen CE, Bahr MJ, et al. Etiology, management, and outcome of the Budd-Chiari syndrome. Ann Intern Med 2009;151:167–75. [284] Heit JA. Epidemiology of venous thromboembolism. Nat Rev Cardiol 2015;12:464–74. [285] Smalberg JH, Arends LR, Valla DC, Kiladjian J-J, Janssen HL, Leebeek FW. Myeloproliferative neoplasms in Budd-Chiari syndrome and portal vein thrombosis: a meta- analysis. Blood 2012;120:4921–4928. [286] How J, Zhou A, Oh ST. Splanchnic vein thrombosis in myeloproliferative neoplasms: pathophysiology and molecular mechanisms of disease. Ther Adv Hematol 2017;8:107–18. [287] Kiladjian J-J, Cervantes F, Leebeek FWG, Marzac C, Cassinat B, Chevret S, et al. The impact of JAK2 and MPL mutations on diagnosis and prognosis of splanchnic vein thrombosis: a report on 241 cases. Blood 2008;111:4922–9. [288] De Stefano V, Fiorini A, Rossi E, Za T, Farina G, Chiusolo P, et al. Incidence of the JAK2 V617F mutation among patients with splanchnic or cerebral venous thrombosis and without overt chronic myeloproliferative disorders. J Thromb Haemost JTH 2007;5:708–14. [289] Primignani M, Barosi G, Bergamaschi G, Gianelli U, Fabris F, Reati R, et al. Role of the JAK2 mutation in the diagnosis of chronic myeloproliferative disorders in splanchnic vein

158 thrombosis. Hepatol Baltim Md 2006;44:1528–34. [290] Barosi G, Buratti A, Costa A, Liberato LN, Balduini C, Cazzola M, et al. An atypical myeloproliferative disorder with high thrombotic risk and slow disease progression. Cancer 1991;68:2310–8. [291] Chait Y, Condat B, Cazals-Hatem D, Rufat P, Atmani S, Chaoui D, et al. Relevance of the criteria commonly used to diagnose myeloproliferative disorder in patients with splanchnic vein thrombosis. Br J Haematol 2005;129:553–60. [292] Valla D, Casadevall N, Huisse MG, Tulliez M, Grange JD, Muller O, et al. Etiology of portal vein thrombosis in adults. A prospective evaluation of primary myeloproliferative disorders. Gastroenterology 1988;94:1063–9. [293] Lamy T, Devillers A, Bernard M, Moisan A, Grulois I, Drenou B, et al. Inapparent polycythemia vera: an unrecognized diagnosis. Am J Med 1997;102:14–20. [294] De Stefano V, Za T, Rossi E, Vannucchi AM, Ruggeri M, Elli E, et al. Recurrent thrombosis in patients with polycythemia vera and essential thrombocythemia: incidence, risk factors, and effect of treatments. Haematologica 2008;93:372–80. [295] Colaizzo D, Amitrano L, Guardascione MA, Tiscia GL, D’Andrea G, Longo VAC, et al. Outcome of patients with splanchnic venous thrombosis presenting without overt MPN: a role for the JAK2 V617F mutation re-evaluation. Thromb Res 2013;132:e99–104. [296] Aird WC. Phenotypic Heterogeneity of the Endothelium: I. Structure, Function, and Mechanisms. Circ Res 2007;100:158–73. [297] Poisson J, Lemoinne S, Boulanger C, Durand F, Moreau R, Valla D, et al. Liver sinusoidal endothelial cells: Physiology and role in liver diseases. J Hepatol 2017;66:212–27. [298] Smalberg JH, Kruip MJHA, Janssen HLA, Rijken DC, Leebeek FWG, de Maat MPM. Hypercoagulability and hypofibrinolysis and risk of deep vein thrombosis and splanchnic vein thrombosis: similarities and differences. Arterioscler Thromb Vasc Biol 2011;31:485–93. [299] Mathew R, Huang J, Wu JM, Fallon JT, Gewitz MH. Hematological disorders and pulmonary hypertension. World J Cardiol 2016;8:703–18. [300] García-Manero G, Schuster SJ, Patrick H, Martinez J. Pulmonary hypertension in patients with myelofibrosis secondary to myeloproliferative diseases. Am J Hematol 1999;60:130–5. [301] Tabarroki A, Lindner DJ, Visconte V, Zhang L, Rogers HJ, Parker Y, et al. Ruxolitinib leads to improvement of pulmonary hypertension in patients with myelofibrosis. Leukemia 2014;28:1486–93. [302] Masri FA, Xu W, Comhair SAA, Asosingh K, Koo M, Vasanji A, et al. Hyperproliferative apoptosis-resistant endothelial cells in idiopathic pulmonary arterial hypertension. Am J Physiol-Lung Cell Mol Physiol 2007;293:L548–54. [303] Inra CN, Zhou BO, Acar M, Murphy MM, Richardson J, Zhao Z, et al. A perisinusoidal niche for extramedullary haematopoiesis in the spleen. Nature 2015;527:466–71. [304] Kvasnicka HM, Thiele J. Bone marrow angiogenesis: methods of quantification and changes evolving in chronic myeloproliferative disorders. Histol Histopathol 2004;19:1245–60. [305] Ni H, Barosi G, Hoffman R. Quantitative evaluation of bone marrow angiogenesis in idiopathic myelofibrosis. Am J Clin Pathol 2006;126:241–7. [306] Medinger M, Skoda R, Gratwohl A, Theocharides A, Buser A, Heim D, et al. Angiogenesis and vascular endothelial growth factor-/receptor expression in myeloproliferative neoplasms: correlation with clinical parameters and JAK2-V617F mutational status. Br J Haematol 2009;146:150–7. [307] Gianelli U, Vener C, Raviele PR, Savi F, Somalvico F, Calori R, et al. VEGF Expression Correlates With Microvessel Density in Philadelphia Chromosome–Negative Chronic Myeloproliferative Disorders. Am J Clin Pathol 2007;128:966–73. [308] Panteli K, Bai M, Hatzimichael E, Zagorianakou N, Agnantis NJ, Bourantas K. Serum

159 levels, and bone marrow immunohistochemical expression of, vascular endothelial growth factor in patients with chronic myeloproliferative diseases. Hematology 2007;12:481–6. [309] Boiocchi L, Vener C, Savi F, Bonoldi E, Moro A, Fracchiolla NS, et al. Increased expression of vascular endothelial growth factor receptor 1 correlates with VEGF and microvessel density in Philadelphia chromosome-negative myeloproliferative neoplasms. J Clin Pathol 2011;64:226–31. [310] Subotički T, Mitrović Ajtić O, Beleslin-Čokić BB, Nienhold R, Diklić M, Djikić D, et al. Angiogenic factors are increased in circulating granulocytes and CD34+ cells of myeloproliferative neoplasms. Mol Carcinog 2017;56:567–79. [311] Bogdanova A, Mihov D, Lutz H, Saam B, Gassmann M, Vogel J. Enhanced erythro- phagocytosis in polycythemic mice overexpressing erythropoietin. Blood 2007;110:762–9. [312] Stam J. Thrombosis of the cerebral veins and sinuses. N Engl J Med 2005;352:1791–8. [313] Martinelli I, De Stefano V. Rare thromboses of cerebral, splanchnic and upper-extremity veins. A narrative review. Thromb Haemost 2010;103:1136–44. [314] Passamonti SM, Biguzzi E, Cazzola M, Franchi F, Gianniello F, Bucciarelli P, et al. The JAK2 V617F mutation in patients with cerebral venous thrombosis. J Thromb Haemost JTH 2012;10:998–1003. [315] Lamy M, Palazzo P, Agius P, Chomel JC, Ciron J, Berthomet A, et al. Should We Screen for Janus Kinase 2 V617F Mutation in Cerebral Venous Thrombosis? Cerebrovasc Dis Basel Switz 2017;44:97–104. [316] De T, Prabhakar P, Nagaraja D, Christopher R. Janus kinase (JAK) 2 V617F mutation in Asian Indians with cerebral venous thrombosis and without overt myeloproliferative disorders. J Neurol Sci 2012;323:178–82. [317] Dentali F, Ageno W, Rumi E, Casetti I, Poli D, Scoditti U, et al. Cerebral venous thrombosis and myeloproliferative neoplasms: results from two large databases. Thromb Res 2014;134:41–3. [318] Hasselbalch HC. Perspectives on the impact of JAK-inhibitor therapy upon inflammation- mediated comorbidities in myelofibrosis and related neoplasms. Expert Rev Hematol 2014;7:203–16. [319] Finazzi G, De Stefano V, Barbui T. Are MPNs vascular diseases? Curr Hematol Malig Rep 2013;8:307–16. [320] Vrtovec M, Anzic A, Zupan IP, Zaletel K, Blinc A. Carotid Artery Stiffness, Digital Endothelial Function, and Coronary Calcium in Patients with Essential Thrombocytosis, Free of Overt Atherosclerotic Disease. Radiol Oncol 2017;51:203–10. [321] Jaiswal S, Fontanillas P, Flannick J, Manning A, Grauman PV, Mar BG, et al. Age-related clonal hematopoiesis associated with adverse outcomes. N Engl J Med 2014;371:2488–98. [322] Fuster JJ, Walsh K. Somatic Mutations and Clonal Hematopoiesis: Unexpected Potential New Drivers of Age-Related Cardiovascular Disease. Circ Res 2018;122:523–32. [323] Jaiswal S, Natarajan P, Silver AJ, Gibson CJ, Bick AG, Shvartz E, et al. Clonal Hematopoiesis and Risk of Atherosclerotic Cardiovascular Disease. N Engl J Med 2017;377:111–21. [324] Cassinat B, Harrison C, Kiladjian J-J. Clonal Hematopoiesis and Atherosclerosis. N Engl J Med 2017;377:1400–1. [325] Jaiswal S, Natarajan P, Ebert BL. Clonal Hematopoiesis and Atherosclerosis. N Engl J Med 2017;377:1401–2. [326] Muendlein A, Gasser K, Kinz E, Stark N, Leiherer A, Rein P, et al. Evaluation of the prevalence and prospective clinical impact of the JAK2 V617F mutation in coronary patients. Am J Hematol 2014;89:295–301. [327] Muendlein A, Kinz E, Gasser K, Leiherer A, Rein P, Saely CH, et al. Occurrence of the JAK2 V617F mutation in patients with peripheral arterial disease. Am J Hematol 2015;90:E17-21.

160 [328] Fuster JJ, MacLauchlan S, Zuriaga MA, Polackal MN, Ostriker AC, Chakraborty R, et al. Clonal hematopoiesis associated with TET2 deficiency accelerates atherosclerosis development in mice. Science 2017;355:842–7. [329] Ridker PM, MacFadyen JG, Everett BM, Libby P, Thuren T, Glynn RJ, et al. Relationship of C-reactive protein reduction to cardiovascular event reduction following treatment with canakinumab: a secondary analysis from the CANTOS randomised controlled trial. Lancet Lond Engl 2018;391:319–28. [330] Davies PF. Flow-mediated endothelial mechanotransduction. Physiol Rev 1995;75:519– 60. [331] Malek AM, Alper SL, Izumo S. Hemodynamic shear stress and its role in atherosclerosis. JAMA 1999;282:2035–42. [332] Parmar KM, Larman HB, Dai G, Zhang Y, Wang ET, Moorthy SN, et al. Integration of flow- dependent endothelial phenotypes by Kruppel-like factor 2. J Clin Invest 2006;116:49–58. [333] Wade JP, Pearson TC, Russell RW, Wetherley-Mein G. Cerebral blood flow and blood viscosity in patients with polycythaemia secondary to hypoxic lung disease. Br Med J Clin Res Ed 1981;283:689–92. [334] Wells RE, Merrill EW. INFLUENCE OF FLOW PROPERTIES OF BLOOD UPON VISCOSITY- HEMATOCRIT RELATIONSHIPS*. J Clin Invest 1962;41:1591–8. [335] Pearson TC, Wetherley-Mein G. VASCULAR OCCLUSIVE EPISODES AND VENOUS HÆMATOCRIT IN PRIMARY PROLIFERATIVE POLYCYTHÆMLX. The Lancet 1978;312:1219– 22. [336] Golse N, Bucur PO, Adam R, Castaing D, Sa Cunha A, Vibert E. New paradigms in post- hepatectomy liver failure. J Gastrointest Surg Off J Soc Surg Aliment Tract 2013;17:593–605. [337] Yamanaka K, Hatano E, Narita M, Kitamura K, Yanagida A, Asechi H, et al. Olprinone attenuates excessive shear stress through up-regulation of endothelial nitric oxide synthase in a rat excessive hepatectomy model. Liver Transplant Off Publ Am Assoc Study Liver Dis Int Liver Transplant Soc 2011;17:60–9. [338] Martini J, Carpentier B, Negrete AC, Frangos JA, Intaglietta M. Paradoxical hypotension following increased hematocrit and blood viscosity. Am J Physiol-Heart Circ Physiol 2005;289:H2136–43. [339] Dekker RJ, van Soest S, Fontijn RD, Salamanca S, de Groot PG, VanBavel E, et al. Prolonged fluid shear stress induces a distinct set of endothelial cell genes, most specifically lung Krüppel-like factor (KLF2). Blood 2002;100:1689–98. [340] Lusis AJ. Atherosclerosis. Nature 2000;407:233–41. [341] Knowles JW, Reddick RL, Jennette JC, Shesely EG, Smithies O, Maeda N. Enhanced atherosclerosis and kidney dysfunction in eNOS(-/-)Apoe(-/-) mice are ameliorated by enalapril treatment. J Clin Invest 2000;105:451–8. [342] Moss JWE, Ramji DP. Nutraceutical therapies for atherosclerosis. Nat Rev Cardiol 2016;13:513–32. [343] Neunteufl T, Heher S, Stefenelli T, Pabinger I, Gisslinger H. Endothelial dysfunction in patients with polycythaemia vera. Br J Haematol 2001;115:354–9. [344] Li Y, Lui KO, Zhou B. Reassessing endothelial-to-mesenchymal transition in cardiovascular diseases. Nat Rev Cardiol 2018. [345] Cho JG, Lee A, Chang W, Lee M-S, Kim J. Endothelial to Mesenchymal Transition Represents a Key Link in the Interaction between Inflammation and Endothelial Dysfunction. Front Immunol 2018;9:294. [346] Chen P-Y, Qin L, Baeyens N, Li G, Afolabi T, Budatha M, et al. Endothelial-to-mesenchymal transition drives atherosclerosis progression. J Clin Invest 2015;125:4514–28. [347] Erba BG, Gruppi C, Corada M, Pisati F, Rosti V, Bartalucci N, et al. Endothelial-to-

161 Mesenchymal Transition in Bone Marrow and Spleen of Primary Myelofibrosis. Am J Pathol 2017;187:1879–92. [348] Zahr AA, Salama ME, Carreau N, Tremblay D, Verstovsek S, Mesa R, et al. Bone marrow fibrosis in myelofibrosis: pathogenesis, prognosis and targeted strategies. Haematologica 2016;101:660–71. [349] Chou JM, Li C-Y, Tefferi A. Bone marrow immunohistochemical studies of angiogenic cytokines and their receptors in myelofibrosis with myeloid metaplasia. Leuk Res 2003;27:499–504. [350] Campanelli R, Rosti V, Villani L, Castagno M, Moretti E, Bonetti E, et al. Evaluation of the bioactive and total transforming growth factor β1 levels in primary myelofibrosis. Cytokine 2011;53:100–6. [351] Simon SI, Green CE. Molecular Mechanics and Dynamics of Leukocyte Recruitment During Inflammation. Annu Rev Biomed Eng 2005;7:151–85. [352] Carman CV, Springer TA. A transmigratory cup in leukocyte diapedesis both through individual vascular endothelial cells and between them. J Cell Biol 2004;167:377–88. [353] Libby P, Ridker PM, Hansson GK, Leducq Transatlantic Network on Atherothrombosis. Inflammation in atherosclerosis: from pathophysiology to practice. J Am Coll Cardiol 2009;54:2129–38. [354] Tedgui A, Mallat Z. Cytokines in atherosclerosis: pathogenic and regulatory pathways. Physiol Rev 2006;86:515–81. [355] Ramji DP, Davies TS. Cytokines in atherosclerosis: Key players in all stages of disease and promising therapeutic targets. Cytokine Growth Factor Rev 2015;26:673–85. [356] Barbui T, Carobbio A, Finazzi G, Vannucchi AM, Barosi G, Antonioli E, et al. Inflammation and thrombosis in essential thrombocythemia and polycythemia vera: different role of C- reactive protein and pentraxin 3. Haematologica 2011;96:315–8. [357] Barosi G. An immune dysregulation in MPN. Curr Hematol Malig Rep 2014;9:331–9. [358] Hasselbalch HC. Chronic inflammation as a promotor of mutagenesis in essential thrombocythemia, polycythemia vera and myelofibrosis. A human inflammation model for cancer development? Leuk Res 2013;37:214–20. [359] Ross R. Atherosclerosis--an inflammatory disease. N Engl J Med 1999;340:115–26. [360] Hasselbalch HC. Perspectives on chronic inflammation in essential thrombocythemia, polycythemia vera, and myelofibrosis: is chronic inflammation a trigger and driver of clonal evolution and development of accelerated atherosclerosis and second cancer? Blood 2012;119:3219–25. [361] Tabas I. Macrophage death and defective inflammation resolution in atherosclerosis. Nat Rev Immunol 2010;10:36–46. [362] Febbraio M, Podrez EA, Smith JD, Hajjar DP, Hazen SL, Hoff HF, et al. Targeted disruption of the class B scavenger receptor CD36 protects against atherosclerotic lesion development in mice. J Clin Invest 2000;105:1049–56. [363] Silverstein RL. Inflammation, atherosclerosis, and arterial thrombosis: role of the scavenger receptor CD36. Cleve Clin J Med 2009;76 Suppl 2:S27-30. [364] Ginsberg H, Gilbert HS, Gibson JC, Le NA, Brown WV. Increased low-density-lipoprotein catabolism in myeloproliferative disorders. Ann Intern Med 1982;96:311–6. [365] Ginsberg H, Goldberg IJ, Wang-Iverson P, Gitler E, Le NA, Gilbert HS, et al. Increased catabolism of native and cyclohexanedione-modified low density lipoprotein in subjects with myeloproliferative diseases. Arterioscler Dallas Tex 1983;3:233–41. [366] Vallabhajosula S, Gilbert HS, Goldsmith SJ, Paidi M, Hanna MM, Ginsberg HN. Low- density lipoprotein (LDL) distribution shown by 99mtechnetium-LDL imaging in patients with myeloproliferative diseases. Ann Intern Med 1989;110:208–13.

162 [367] Gardai SJ, McPhillips KA, Frasch SC, Janssen WJ, Starefeldt A, Murphy-Ullrich JE, et al. Cell-surface calreticulin initiates clearance of viable or apoptotic cells through trans-activation of LRP on the phagocyte. Cell 2005;123:321–34. [368] Daitoku S, Takenaka K, Yamauchi T, Yurino A, Jinnouchi F, Nunomura T, et al. Calreticulin mutation does not contribute to disease progression in essential thrombocythemia by inhibiting phagocytosis. Exp Hematol 2016;44:817-825.e3. [369] Getz GS, Reardon CA. Natural killer T cells in atherosclerosis. Nat Rev Cardiol 2017;14:304–14. [370] Ait-Oufella H, Salomon BL, Potteaux S, Robertson A-KL, Gourdy P, Zoll J, et al. Natural regulatory T cells control the development of atherosclerosis in mice. Nat Med 2006;12:178– 80. [371] Pastrana JL, Sha X, Virtue A, Mai J, Cueto R, Lee IA, et al. Regulatory T cells and Atherosclerosis. J Clin Exp Cardiol 2012;2012:002. [372] Liuzzo G, Goronzy JJ, Yang H, Kopecky SL, Holmes DR, Frye RL, et al. Monoclonal T-cell proliferation and plaque instability in acute coronary syndromes. Circulation 2000;101:2883– 8. [373] Cheng X, Yu X, Ding Y-J, Fu Q-Q, Xie J-J, Tang T-T, et al. The Th17/Treg imbalance in patients with acute coronary syndrome. Clin Immunol Orlando Fla 2008;127:89–97. [374] Lim K-H, Chen CG-S, Chang Y-C, Chiang Y-H, Kao C-W, Wang W-T, et al. Increased B cell activation is present in JAK2V617F-mutated, CALR-mutated and triple-negative essential thrombocythemia. Oncotarget 2017;8:32476–91. [375] Sanchez C, Baier C, Colle JG, Chelbi R, Rihet P, Le Treut T, et al. Natural killer cells in patients with polycythemia vera. Hum Immunol 2015;76:644–50. [376] Choi DC, Tremblay D, Iancu-Rubin C, Mascarenhas J. Programmed cell death-1 pathway inhibition in myeloid malignancies: implications for myeloproliferative neoplasms. Ann Hematol 2017;96:919–27. [377] Zhao W, Li Y, Liu X, Zhang L, Wang X. Involvement of CD4+CD25+ regulatory T cells in the pathogenesis of polycythaemia vera. Chin Med J (Engl) 2008;121:1781–6. [378] Keohane C, Kordasti S, Seidl T, Abellan PP, Thomas NSB, Harrison CN, et al. JAK inhibition induces silencing of T Helper cytokine secretion and a profound reduction in T regulatory cells. Br J Haematol 2015;171:60–73. [379] Shoeibi S, Mozdziak P, Mohammadi S. Important signals regulating coronary artery angiogenesis. Microvasc Res 2018;117:1–9. [380] Moreno PR, Purushothaman K-R, Sirol M, Levy AP, Fuster V. Neovascularization in human atherosclerosis. Circulation 2006;113:2245–52. [381] Virmani R, Kolodgie FD, Burke AP, Finn AV, Gold HK, Tulenko TN, et al. Atherosclerotic plaque progression and vulnerability to rupture: angiogenesis as a source of intraplaque hemorrhage. Arterioscler Thromb Vasc Biol 2005;25:2054–61. [382] Rohde E, Malischnik C, Thaler D, Maierhofer T, Linkesch W, Lanzer G, et al. Blood monocytes mimic endothelial progenitor cells. Stem Cells Dayt Ohio 2006;24:357–67. [383] Case J, Mead LE, Bessler WK, Prater D, White HA, Saadatzadeh MR, et al. Human CD34+AC133+VEGFR-2+ cells are not endothelial progenitor cells but distinct, primitive hematopoietic progenitors. Exp Hematol 2007;35:1109–18. [384] Asahara T, Murohara T, Sullivan A, Silver M, van der Zee R, Li T, et al. Isolation of putative progenitor endothelial cells for angiogenesis. Science 1997;275:964–7. [385] Ingram DA, Mead LE, Tanaka H, Meade V, Fenoglio A, Mortell K, et al. Identification of a novel hierarchy of endothelial progenitor cells using human peripheral and umbilical cord blood. Blood 2004;104:2752–60. [386] Yoder MC, Mead LE, Prater D, Krier TR, Mroueh KN, Li F, et al. Redefining endothelial

163 progenitor cells via clonal analysis and hematopoietic stem/progenitor cell principals. Blood 2007;109:1801–9. [387] Eelen G, de Zeeuw P, Treps L, Harjes U, Wong BW, Carmeliet P. Endothelial Cell Metabolism. Physiol Rev 2017;98:3–58. [388] Silvestre J-S, Smadja DM, Lévy BI. Postischemic Revascularization: From Cellular and Molecular Mechanisms to Clinical Applications. Physiol Rev 2013;93:1743–802. [389] Hill JM, Zalos G, Halcox JPJ, Schenke WH, Waclawiw MA, Quyyumi AA, et al. Circulating Endothelial Progenitor Cells, Vascular Function, and Cardiovascular Risk. Http://DxDoiOrgGate2InistFr/101056/NEJMoa022287 2009. [390] Werner N, Kosiol S, Schiegl T, Ahlers P, Walenta K, Link A, et al. Circulating endothelial progenitor cells and cardiovascular outcomes. N Engl J Med 2005;353:999–1007. [391] Piaggio G, Rosti V, Corselli M, Bertolotti F, Bergamaschi G, Pozzi S, et al. Endothelial colony-forming cells from patients with chronic myeloproliferative disorders lack the disease- specific molecular clonality marker. Blood 2009;114:3127–30. [392] Medinger M, Mross K. Clinical trials with anti-angiogenic agents in hematological malignancies. J Angiogenesis Res 2010;2:10. [393] Davies MJ. The pathophysiology of acute coronary syndromes. Heart 2000;83:361–6. [394] Crea F, Libby P. Acute Coronary Syndromes: The Way Forward From Mechanisms to Precision Treatment. Circulation 2017;136:1155–66. [395] Casscells W, Naghavi M, Willerson JT. Vulnerable atherosclerotic plaque: a multifocal disease. Circulation 2003;107:2072–5. [396] Newby AC. Matrix metalloproteinases regulate migration, proliferation, and death of vascular smooth muscle cells by degrading matrix and non-matrix substrates. Cardiovasc Res 2006;69:614–24. [397] Michiels JJ, Berneman ZN, Schroyens W, Van Vliet HHDM. Pathophysiology and treatment of platelet-mediated microvascular disturbances, major thrombosis and bleeding complications in essential thrombocythaemia and polycythaemia vera. Platelets 2004;15:67– 84. [398] Petrides PE, Siegel F. Thrombotic complications in essential thrombocythemia (ET): clinical facts and biochemical riddles. Blood Cells Mol Dis 2006;36:379–84. [399] Ozben B, Ekmekci A, Bugra Z, Umman S, Meric M. Multiple coronary thrombosis and stent implantation to the subtotally occluded right renal artery in a patient with essential thrombocytosis: a case report with review. J Thromb Thrombolysis 2006;22:79–84. [400] Agewall S, Beltrame JF, Reynolds HR, Niessner A, Rosano G, Caforio ALP, et al. ESC working group position paper on myocardial infarction with non-obstructive coronary arteries. Eur Heart J 2017;38:143–53. [401] Beltrame JF, Crea F, Kaski JC, Ogawa H, Ong P, Sechtem U, et al. The Who, What, Why, When, How and Where of Vasospastic Angina. Circ J Off J Jpn Circ Soc 2016;80:289–98. [402] Landolfi R, Di Gennaro L, Barbui T, De Stefano V, Finazzi G, Marfisi R, et al. Leukocytosis as a major thrombotic risk factor in patients with polycythemia vera. Blood 2007;109:2446–52. [403] Kaess BM, Preis SR, Beiser A, Sawyer DB, Chen TC, Seshadri S, et al. Circulating vascular endothelial growth factor and the risk of cardiovascular events. Heart 2016:heartjnl-2015- 309155. [404] Pósfai É, Marton I, Borbényi Z, Nemes A. Myocardial infarction as a thrombotic complication of essential thrombocythemia and polycythemia vera. Anatol J Cardiol 2016;16:397–402. [405] Larsen AI, Galbraith PD, Ghali WA, Norris CM, Graham MM, Knudtson ML. Characteristics and outcomes of patients with acute myocardial infarction and angiographically normal coronary arteries. Am J Cardiol 2005;95:261–3.

164 [406] Bairey Merz CN, Pepine CJ, Walsh MN, Fleg JL. Ischemia and No Obstructive Coronary Artery Disease (INOCA): Developing Evidence-Based Therapies and Research Agenda for the Next Decade. Circulation 2017;135:1075–92. [407] Ong P, Aziz A, Hansen HS, Prescott E, Athanasiadis A, Sechtem U. Structural and Functional Coronary Artery Abnormalities in Patients With Vasospastic Angina Pectoris. Circ J Off J Jpn Circ Soc 2015;79:1431–8. [408] Bang OY, Toyoda K, Arenillas JF, Liu L, Kim JS. Intracranial Large Artery Disease of Non- Atherosclerotic Origin: Recent Progress and Clinical Implications. J Stroke 2018;20:208–17. [409] Furchgott RF, Zawadzki JV. The obligatory role of endothelial cells in the relaxation of arterial smooth muscle by acetylcholine. Nature 1980;288:373–6. [410] Rapoport RM, Murad F. Agonist-induced endothelium-dependent relaxation in rat thoracic aorta may be mediated through cGMP. Circ Res 1983;52:352–7. [411] Ignarro LJ, Buga GM, Wood KS, Byrns RE, Chaudhuri G. Endothelium-derived relaxing factor produced and released from artery and vein is nitric oxide. Proc Natl Acad Sci U S A 1987;84:9265–9. [412] Palmer RM, Ashton DS, Moncada S. Vascular endothelial cells synthesize nitric oxide from L-arginine. Nature 1988;333:664–6. [413] Sessa WC. eNOS at a glance. J Cell Sci 2004;117:2427–9. [414] Dimmeler S, Haendeler J, Nehls M, Zeiher AM. Suppression of apoptosis by nitric oxide via inhibition of interleukin-1beta-converting enzyme (ICE)-like and cysteine protease protein (CPP)-32-like proteases. J Exp Med 1997;185:601–7. [415] Kang-Decker N, Cao S, Chatterjee S, Yao J, Egan LJ, Semela D, et al. Nitric oxide promotes endothelial cell survival signaling through S-nitrosylation and activation of dynamin-2. J Cell Sci 2007;120:492–501. [416] Xu L, Eu JP, Meissner G, Stamler JS. Activation of the cardiac calcium release channel (ryanodine receptor) by poly-S-nitrosylation. Science 1998;279:234–7. [417] Shiva S. Nitrite: A physiological store of nitric oxide and modulator of mitochondrial function. Redox Biol 2013;1:40–4. [418] Hogg N. Detection of Nitric Oxide by Electron Paramagnetic Resonance Spectroscopy. Free Radic Biol Med 2010;49:122–9. [419] Dimmeler S, Fleming I, Fisslthaler B, Hermann C, Busse R, Zeiher AM. Activation of nitric oxide synthase in endothelial cells by Akt-dependent phosphorylation. Nature 1999;399:601–5. [420] Vanhoutte PM, Zhao Y, Xu A, Leung SWS. Thirty Years of Saying NO: Sources, Fate, Actions, and Misfortunes of the Endothelium-Derived Vasodilator Mediator. Circ Res 2016;119:375–96. [421] Cortelazzo S, Viero P, Morelli C, Barbui T, de Gaetano G. Normal vascular prostacyclin generation in patients with chronic myeloproliferative disorders. Thromb Res 1985;39:139–41. [422] Cavalca V, Rocca B, Squellerio I, Dragani A, Veglia F, Pagliaccia F, et al. In vivo prostacyclin biosynthesis and effects of different aspirin regimens in patients with essential thrombocythaemia. Thromb Haemost 2014;112:118–27. [423] Cortelazzo S, Galli M, Castagna D, Viero P, Gaetano G de, Barbui T. Increased Response to Arachidonic Acid and U-46619 and Resistance to Inhibitory Prostaglandins in Patients with Chronic Myeloproliferative Disorders. Thromb Haemost 1988;59:73–6. [424] Cooper B, Ahern D. Characterization of the Platelet Prostaglandin D2 Receptor: LOSS OF PROSTAGLANDIN D2 RECEPTORS IN PLATELETS OF PATIENTS WITH MYELOPROLIFERATIVE DISORDERS. J Clin Invest 1979;64:586–90. [425] Cooper B, Schafer AI, Puchalsky D, Handin RI. Platelet resistance to prostaglandin D2 in patients with myeloproliferative disorders. Blood 1978;52:618–26. [426] Nimmo GR, McInnes G, Turner AR, Evan-Wong A. Severe mesenteric ischemia due to

165 primary thrombocythemia treated with prostacyclin. Crit Care Med 1990;18:889–90. [427] Zeidman A, Dicker D, Mittelman M. Interferon-Induced Vasospasm in Chronic Myeloid Leukaemia. Acta Haematol 1998;100:94–6. [428] Rubanyi GM, Polokoff MA. Endothelins: molecular biology, biochemistry, pharmacology, physiology, and pathophysiology. Pharmacol Rev 1994;46:325–415. [429] Thorin E, Clozel M. The cardiovascular physiology and pharmacology of endothelin-1. Adv Pharmacol San Diego Calif 2010;60:1–26. [430] Yachoui R, Kristianto J, Sitwala K, Blank RD. Role of Endothelin-1 in a Syndrome of Myelofibrosis and Osteosclerosis. J Clin Endocrinol Metab 2015;100:3971–4. [431] Nakahata N. Thromboxane A2: physiology/pathophysiology, cellular signal transduction and pharmacology. Pharmacol Ther 2008;118:18–35. [432] Yao H-X, Huang L, Lin L-E, Wu C-M, Wang H, Chen W-T, et al. [The study of relationship between platelet function and thrombus in patients with essential thrombocythaemia]. Zhonghua Yi Xue Za Zhi 2010;90:253–5. [433] Schafer AI. Deficiency of platelet lipoxygenase activity in myeloproliferative disorders. N Engl J Med 1982;306:381–6. [434] Smith IL, Martin TJ. Platelet thromboxane synthesis and release reactions in myeloproliferative disorders. Haemostasis 1982;11:119–27. [435] Okuma M, Uchino H. Altered arachidonate metabolism by platelets in patients with myeloproliferative disorders. Blood 1979;54:1258–71. [436] Rocca B, Ciabattoni G, Tartaglione R, Cortelazzo S, Barbui T, Patrono C, et al. Increased thromboxane biosynthesis in essential thrombocythemia. Thromb Haemost 1995;74:1225–30. [437] Dragani A, Pascale S, Recchiuti A, Mattoscio D, Lattanzio S, Petrucci G, et al. The contribution of cyclooxygenase-1 and -2 to persistent thromboxane biosynthesis in aspirin- treated essential thrombocythemia: implications for antiplatelet therapy. Blood 2010;115:1054–61. [438] Landolfi R, Ciabattoni G, Patrignani P, Castellana MA, Pogliani E, Bizzi B, et al. Increased thromboxane biosynthesis in patients with polycythemia vera: evidence for aspirin- suppressible platelet activation in vivo. Blood 1992;80:1965–71. [439] Patrono C, Ciabattoni G, Patrignani P, Rocca B, Landolfi R. Eicosanoid biosynthesis and metabolism in myeloproliferative disorders. Ann N Y Acad Sci 1994;744:229–36. [440] Zahavi M, Zahavi J. Platelet function and thromboxane synthesis in myeloproliferative disorders. Blood 1993;82:1377–8. [441] Zahavi J, Zahavi M, Firsteter E, Frish B, Turleanu R, Rachmani R. An abnormal pattern of multiple platelet function abnormalities and increased thromboxane generation in patients with primary thrombocytosis and thrombotic complications. Eur J Haematol 1991;47:326–32. [442] van Genderen PJ, Prins FJ, Michiels JJ, Schrör K. Thromboxane-dependent platelet activation in vivo precedes arterial thrombosis in thrombocythaemia: a rationale for the use of low-dose aspirin as an antithrombotic agent. Br J Haematol 1999;104:438–41. [443] Pascale S, Petrucci G, Dragani A, Habib A, Zaccardi F, Pagliaccia F, et al. Aspirin- insensitive thromboxane biosynthesis in essential thrombocythemia is explained by accelerated renewal of the drug target. Blood 2012;119:3595–603. [444] Tefferi A. Overcoming “aspirin resistance” in MPN. Blood 2012;119:3377–8. [445] Karantanos T, Moliterno AR. The roles of JAK2 in DNA damage and repair in the myeloproliferative neoplasms: Opportunities for targeted therapy. Blood Rev 2018. [446] Bjørn ME, Hasselbalch HC. The Role of Reactive Oxygen Species in Myelofibrosis and Related Neoplasms. Mediators Inflamm 2015;2015:648090. [447] Kagoya Y, Yoshimi A, Tsuruta-Kishino T, Arai S, Satoh T, Akira S, et al. JAK2V617F+ myeloproliferative neoplasm clones evoke paracrine DNA damage to adjacent normal cells

166 through secretion of lipocalin-2. Blood 2014;124:2996–3006. [448] Marty C, Lacout C, Droin N, Le Couédic J-P, Ribrag V, Solary E, et al. A role for reactive oxygen species in JAK2V617F myeloproliferative neoplasm progression. Leukemia 2013;27:2187–95. [449] Plo I, Nakatake M, Malivert L, de Villartay J-P, Giraudier S, Villeval J-L, et al. JAK2 stimulates homologous recombination and genetic instability: potential implication in the heterogeneity of myeloproliferative disorders. Blood 2008;112:1402–12. [450] Ahn JS, Li J, Chen E, Kent DG, Park HJ, Green AR. JAK2V617F mediates resistance to DNA damage-induced apoptosis by modulating FOXO3A localization and Bcl-xL deamidation. Oncogene 2016;35:2235–46. [451] Hurtado-Nedelec M, Csillag-Grange M-J, Boussetta T, Belambri SA, Fay M, Cassinat B, et al. Increased reactive oxygen species production and p47phox phosphorylation in neutrophils from myeloproliferative disorders patients with JAK2 (V617F) mutation. Haematologica 2013;98:1517–24. [452] Musolino C, Allegra A, Saija A, Alonci A, Russo S, Spatari G, et al. Changes in advanced oxidation protein products, advanced glycation end products, and s-nitrosylated proteins, in patients affected by polycythemia vera and essential thrombocythemia. Clin Biochem 2012;45:1439–43. [453] Durmus A, Mentese A, Yilmaz M, Sumer A, Akalin I, Topal C, et al. Increased oxidative stress in patients with essential thrombocythemia. Eur Rev Med Pharmacol Sci 2013;17:2860 [454] Vener C, Novembrino C, Catena FB, Fracchiolla NS, Gianelli U, Savi F, et al. Oxidative stress is increased in primary and post-polycythemia vera myelofibrosis. Exp Hematol 2010;38:1058–65. [455] Hasselbalch HC, Thomassen M, Riley CH, Kjær L, Larsen TS, Jensen MK, et al. Whole blood transcriptional profiling reveals deregulation of oxidative and antioxidative defence genes in myelofibrosis and related neoplasms. Potential implications of downregulation of Nrf2 for genomic instability and disease progression. PloS One 2014;9:e112786. [456] Iurlo A, De Giuseppe R, Sciumè M, Cattaneo D, Fermo E, De Vita C, et al. Oxidative status in treatment-naïve essential thrombocythemia: a pilot study in a single center. Hematol Oncol 2017;35:335–40. [457] Barbui T, Masciulli A, Ghirardi A, Carobbio A. ACE inhibitors and cytoreductive therapy in polycythemia vera. Blood 2017;129:1226–7. [458] Aksu S, Beyazit Y, Haznedaroglu IC, Kekilli M, Canpinar H, Misirlioğlu M, et al. Enhanced expression of the local haematopoietic bone marrow renin-angiotensin system in polycythemia rubra vera. J Int Med Res 2005;33:661–7. [459] Haznedaroğlu IC, Büyükaşik Y. Current evidence for the existence of a local renin- angiotensin system affecting physiological and pathological haemopoiesis in the bone marrow. Br J Haematol 1997;99:471. [460] Vrsalovic MM, Pejsa V, Veic TS, Kolonic SO, Ajdukovic R, Haris V, et al. Bone marrow renin-angiotensin system expression in polycythemia vera and essential thrombocythemia depends on JAK2 mutational status. Cancer Biol Ther 2007;6:1434–6. [461] Haznedaroglu IC, Malkan UY. Local bone marrow renin-angiotensin system in the genesis of leukemia and other malignancies. Eur Rev Med Pharmacol Sci 2016;20:4089–111. [462] Rodrigues-Ferreira S, Abdelkarim M, Dillenburg-Pilla P, Luissint A-C, di-Tommaso A, Deshayes F, et al. Angiotensin II facilitates breast cancer cell migration and metastasis. PloS One 2012;7:e35667. [463] Shi K, Zhao W, Chen Y, Ho WT, Yang P, Zhao ZJ. Cardiac hypertrophy associated with myeloproliferative neoplasms in JAK2V617F transgenic mice. J Hematol OncolJ Hematol Oncol 2014;7:25.

167 [464] Tarach JS, Nowicka-Tarach BM, Matuszek B, Nowakowski A. Erythromelalgia--a thrombotic complication in chronic myeloproliferative disorders. Med Sci Monit Int Med J Exp Clin Res 2000;6:204–8. [465] Valla D-C. Primary Budd-Chiari syndrome. J Hepatol 2009;50:195–203. [466] Wang Y, Nakayama M, Pitulescu ME, Schmidt TS, Bochenek ML, Sakakibara A, et al. Ephrin-B2 controls VEGF-induced angiogenesis and lymphangiogenesis. Nature 2010;465:483– 6. [467] Marty C, Lacout C, Martin A, Hasan S, Jacquot S, Birling M-C, et al. Myeloproliferative neoplasm induced by constitutive expression of JAK2V617F in knock-in mice. Blood 2010;116:783–7. [468] Simonetto DA, Yang H, Yin M, de Assuncao TM, Kwon JH, Hilscher M, et al. Chronic passive venous congestion drives hepatic fibrogenesis via sinusoidal thrombosis and mechanical forces. Hepatol Baltim Md 2015;61:648–59. [469] Schaefer A, Jung M, Miller K, Lein M, Kristiansen G, Erbersdobler A, et al. Suitable reference genes for relative quantification of miRNA expression in prostate cancer. Exp Mol Med 2010;42:749–58. [470] Kaess BM, Preis SR, Beiser A, Sawyer DB, Chen TC, Seshadri S, et al. Circulating vascular endothelial growth factor and the risk of cardiovascular events. Heart Br Card Soc 2016;102:1898–901. [471] Zhao B, Mei Y, Cao L, Zhang J, Sumagin R, Yang J, et al. Loss of pleckstrin-2 reverts lethality and vascular occlusions in JAK2V617F-positive myeloproliferative neoplasms. J Clin Invest 2017. [472] Sen-Banerjee S, Mir S, Lin Z, Hamik A, Atkins GB, Das H, et al. Kruppel-Like Factor 2 as a Novel Mediator of Statin Effects in Endothelial Cells. Circulation 2005;112:720–6. [473] Mason RP, Walter MF, Jacob RF. Effects of HMG-CoA reductase inhibitors on endothelial function: role of microdomains and oxidative stress. Circulation 2004;109:II34-41. [474] Angona A, Alvarez-Larrán A, Bellosillo B, Longarón R, Fernández-Rodríguez C, Besses C. Dynamics of JAK2 V617F allele burden of CD34+ haematopoietic progenitor cells in patients treated with ruxolitinib. Br J Haematol 2016;172:639–42. [475] Vainchenker W, Leroy E, Gilles L, Marty C, Plo I, Constantinescu SN. JAK inhibitors for the treatment of myeloproliferative neoplasms and other disorders. F1000Research 2018;7:82. [476] Cokic VP, Beleslin-Cokic BB, Tomic M, Stojilkovic SS, Noguchi CT, Schechter AN. Hydroxyurea induces the eNOS-cGMP pathway in endothelial cells. Blood 2006;108:184–91. [477] Miki T, Suzuki M, Shibasaki T, Uemura H, Sato T, Yamaguchi K, et al. Mouse model of Prinzmetal angina by disruption of the inward rectifier Kir6.1. Nat Med 2002;8:466–72. [478] Zorov DB, Juhaszova M, Sollott SJ. Mitochondrial reactive oxygen species (ROS) and ROS- induced ROS release. Physiol Rev 2014;94:909–50. [479] Heitzer T, Schlinzig T, Krohn K, Meinertz T, Münzel T. Endothelial dysfunction, oxidative stress, and risk of cardiovascular events in patients with coronary artery disease. Circulation 2001;104:2673–8. [480] Ponomareva AA, Nevzorova TA, Mordakhanova ER, Andrianova IA, Rauova L, Litvinov RI, et al. Intracellular origin and ultrastructure of platelet-derived microparticles. J Thromb Haemost JTH 2017;15:1655–67. [481] Camus SM, Gausserès B, Bonnin P, Loufrani L, Grimaud L, Charue D, et al. Erythrocyte microparticles can induce kidney vaso-occlusions in a murine model of sickle cell disease. Blood 2012;120:5050–8. [482] Camus SM, De Moraes JA, Bonnin P, Abbyad P, Le Jeune S, Lionnet F, et al. Circulating cell membrane microparticles transfer heme to endothelial cells and trigger vasoocclusions in sickle cell disease. Blood 2015;125:3805–14.

168 [483] Merle NS, Grunenwald A, Rajaratnam H, Gnemmi V, Frimat M, Figueres M-L, et al. Intravascular hemolysis activates complement via cell-free heme and heme-loaded microvesicles. JCI Insight 2018;

169 VII. APPENDIX

A. Appendix 1: Review: Liver sinusoidal endothelial cells: physiology and role in liver diseases

Liver sinusoidal endothelial cells: Physiology and role in liver diseases

Johanne Poisson1,2*, Sara Lemoinne3,4*, Chantal Boulanger1,2, François Durand5,6,7, Richard

Moreau5,6,7, Dominique Valla5,6,7, Pierre-Emmanuel Rautou1,2,5,6,7

* These authors contributed equally as joint first authors.

1 INSERM, UMR-970, Paris Cardiovascular Research Center – PARCC, Paris, France;

2 Université Paris Descartes, Sorbonne Paris Cité, Paris, France;

3 INSERM, UMRS 938, Centre de Recherche Saint-Antoine, Université Pierre et Marie Curie

Paris 6, Paris, France;

4 Service d’hépatologie, Hôpital Saint-Antoine, APHP, Paris, France;

5 Service d’hépatologie, DHU Unity, Hôpital Beaujon, APHP, Clichy, France;

6 INSERM, UMR-1149, Centre de Recherche sur l’inflammation, Paris-Clichy, France;

7 Université Denis Diderot-Paris 7, Sorbonne Paris Cité, 75018 Paris, France

Review published in J Hepatol. 2017 Jan;66(1):212-227

170 Review

Liver sinusoidal endothelial cells: Physiology and role in liver diseases

Johanne Poisson1,2,y, Sara Lemoinne3,4,y, Chantal Boulanger1,2, François Durand5,6,7, ⇑ Richard Moreau5,6,7, Dominique Valla5,6,7, Pierre-Emmanuel Rautou1,2,5,6,7,

Keywords: Liver sinusoidal endo- Summary thelial cells; Capillarization; End- othelial dysfunction; Cirrhosis; Liver regeneration; Angiogenesis; Liver sinusoidal endothelial cells (LSECs) are highly specialized endothelial cells representing Drug delivery system; the interface between blood cells on the one side and hepatocytes and hepatic stellate cells on Endothelium. the other side. LSECs represent a permeable barrier. Indeed, the association of ‘fenestrae’, Received 24 May 2016; received in absence of diaphragm and lack of basement membrane make them the most permeable revised form 5 July 2016; accepted 7 endothelial cells of the mammalian body. They also have the highest endocytosis capacity July 2016 of human cells. In physiological conditions, LSECs regulate hepatic vascular tone contributing to the maintenance of a low portal pressure despite the major changes in hepatic blood flow occurring during digestion. LSECs maintain hepatic stellate cell quiescence, thus inhibiting intrahepatic vasoconstriction and fibrosis development. In pathological conditions, LSECs play a key role in the initiation and progression of chronic liver diseases. Indeed, they become capillarized and lose their protective properties, and they promote angiogenesis and vasoconstriction. LSECs are implicated in liver regeneration following acute liver injury or partial 1INSERM, UMR-970, Paris Cardio- hepatectomy since they renew from LSECs and/or LSEC progenitors, they sense changes in vascular Research Center – PARCC, shear stress resulting from surgery, and they interact with platelets and inflammatory cells. Paris, France; 2Université Paris Descartes, Sor- LSECs also play a role in hepatocellular carcinoma development and progression, in ageing, bonne Paris Cité, Paris, France; and in liver lesions related to inflammation and infection. This review also presents a detailed 3 INSERM, UMRS 938, Centre de analysis of the technical aspects relevant for LSEC analysis including the markers these cells Recherche Saint-Antoine, Université Pierre et Marie Curie Paris 6, Paris, express, the available cell lines and the transgenic mouse models. Finally, this review provides France; an overview of the strategies available for a specific targeting of LSECs. 4Service d’hépatologie, Hôpital Ó 2016 European Association for the Study of the Liver. Published by Elsevier B.V. All rights Saint-Antoine, APHP, Paris, France; reserved. 5Service d’hépatologie, DHU Unity Review Hôpital Beaujon, APHP, Clichy, France; Introduction 6INSERM, UMR-1149, Centre de Recherche sur l’inflammation, Paris- The vascular endothelium, representing the inter- LSECs in the normal liver Clichy, France; face between blood and other tissues, is not only 7Université Denis Diderot-Paris 7, Formation of sinusoids during embryogenesis Sorbonne Paris Cité, 75018 Paris, a physical barrier, but contributes to different phys- France iological and pathological processes, including hemostasis/thrombosis, metabolites transporta- As illustrated in Fig. 1, an early structural differentiation, inflammation, angiogenesis and vascular tone tion of hepatic sinusoids occurs between gestational [1]. Liver sinusoidal endothelial cells (LSECs) form weeks 5 and 12 in human embryos [3]. During that y These authors contributed the wall of the liver sinusoids and represent period, LSECs gradually loose cell markers of continu- equally as joint first authors. approximately 15 to 20% of liver cells but only 3% ous endothelial cells including platelet endothelial of the total liver volume [2]. LSECs are highly spe- adhesion molecule-1 (PECAM-1, also called cluster cialized endothelial cells. They have a discontinu- of differentiation (CD)31), CD34 and 1F10 antigen, ⇑ Corresponding author. Address: ous architecture meaning that fusion of the and acquire markers of adult sinusoidal cells includ- Service d’Hépatologie, Hôpital Beaujon, 100 Boulevard du luminal and abluminal plasma membrane occurs ing CD4, CD32 and the intracellular adhesion Général Leclerc, 92110 Clichy, at other sites than cell junctions, in areas called molecule-1 (ICAM-1). This differentiation of LSECs France. Tel.: +33 1 40 87 52 83; ‘fenestrae’. This review focuses on the role of LSECs is regulated by hepatoblasts, both via the vascular fax: +33 1 40 87 54 87. in physiological conditions and their involvement endothelial growth factor (VEGF) they release and E-mail address: pierre-emmanuel. [email protected] (P.-E. Rautou). in liver diseases. via direct intercellular interactions [4,5].

Journal of Hepatology 2017 vol. 66 j 212–227 JOURNAL OF HEPATOLOGY

The embryological origin of LSECs is still a mat- circadian changes in hepatic blood flow due to ter of debate. Initial observational studies described digestion, hepatic venous pressure gradient remains capillaries progressively surrounded by growing at 4 mmHg or less in a normal individual, attesting a cords of hepatoblasts in the septum transversum, fine regulation of hepatic vascular tone [25]. Intra- suggesting that LSECs derive from the septum hepatic shear stress is recognized as a main driver transversum mesenchyme, a part of the mesoderm of hepatic blood flow regulation [26]. Shear stress [3,6,7]. However, recent cell lineage experiments is a frictional force applied by blood flow on performed in mice showed that the septum endothelial surface [26]. It is proportional to flow transversum gives rise to mesothelial cells, hepatic intensity and to blood viscosity and inversely pro- stellate cells, portal fibroblasts, and perivascular portional to the cubic radius of the vessel [26]. Intra- mesenchymal cells, but not to LSECs [8]. A part of hepatic shear stress has never been directly LSECs rather derives from a common progenitor measured in human or animal. Its evaluation is to endothelial and blood cells, called the ‘‘heman- indeed difficult since the radius of sinusoids is very gioblast”, as attested by overlapping expression of small and varies within the liver. Moreover, viscos- hematopoietic and endothelial cell markers by ity is hard to estimate in this specific area and also LSECs and by fate tracing experiments [9–14]. varies with hemodilution. In normal conditions, in These progenitor cells form veins crossing the sep- the liver like in other vascular beds, the endothe- tum transversum, i.e., vitelin veins [15], umbilical lium is able to generate vasodilator agents in veins or cardinal veins and then LSECs [16,17]. response to increased shear stress in order to atten- Another part of LSECs derives from the endo- uate the increase in blood pressure. The loss of this cardium of the sinus venosus, a compartment of property is called endothelial dysfunction. An the primitive cardiac tube [18]. These two embry- endothelial specific transcription factor induced by ological origins might explain the heterogeneity of prolonged shear stress, called Kruppel-like factor 2 the markers expressed by LSECs in adults. (KLF2) mediates this effect of shear stress [27]. KLF2 induces the endothelial upregulation of LSECs renewal vasodilating agents including nitric oxide (NO) [28] (Fig. 2). Shah and colleagues previously demon- Although specific data are lacking, we can specu- strated that LSECs are the main source of NO in late that in a physiological state LSECs are quies- the normal liver through endothelial nitric oxide cent, i.e., with a low proliferation rate and a long synthase (eNOS) activation by shear stress [29]. life span, similar to endothelial cells from large ves- KLF2 also induces the downregulation of vasocon- sels [19]. LSECs renewal differs in physiological and strictive molecules including endothelin-1 [28]. in pathological conditions. Three cell types con- Other molecules released by LSECs regulating blood tribute to LSEC renewal, namely mature LSECs, flow include the vasodilating agent carbon monox- intrahepatic or resident sinusoidal endothelial cell ide (CO) and the metabolites of the cyclooxygenase progenitors, and bone marrow derived sinusoidal (COX) pathway (thromboxane A2, Prostacyclin) endothelial cell progenitors [20]. Mature LSECs [30]. All these molecules act in a paracrine manner can self-proliferate in normal conditions, when on hepatic stellate cells localized in the space of Review stimulated with growth factors such as VEGF and Disse [31]. Healthy LSECs maintain hepatic stellate FGF (fibroblast growth factor) [20,21]. Resident cell quiescence, thus inhibiting their vasoconstric- sinusoidal endothelial cell progenitors represent 1 tive effect [34]. The concept that hepatic stellate cell to 7% of the LSECs of a normal rodent liver and activation induces sinusoid constriction is based on probably contribute to LSECs regeneration [20]. their expression of molecules found in smooth mus- Bone marrow derived sinusoidal endothelial cell cle cells including aSMA, on their position wrapped progenitors do not participate in LSEC turnover in around the exterior of LSECs and on the ex vivo a normal liver [22]. By contrast, after liver injury, observation of their ability to contract [32,33]. Key point these cells are the main drivers of liver regenera- Although still controversial, LSEC could also regulate In a normal liver, differenti- tion [20,22]. Indeed, a subtoxic dose of monocro- blood flow by swelling, thus creating an inlet and an ated LSECs are gatekeepers talin, a toxic agent for LSECs, elicits liver injury outlet sphincter [32]. Kupffer cells possess contrac- of fibrogenesis by maintain- only when bone marrow is suppressed. In addition, tile proteins as well, but their role in the regulation ing hepatic stellate cells in infusion of bone marrow cells after a toxic dose of of hepatic blood flow remains controversial [32]. In their inactivated state. LSECs contrast to most vascular beds where blood flow is regulate sinusoidal blood monocrotalin almost fully corrects liver lesions flow through their action [23]. mostly regulated by smooth muscle cells, in the on hepatic stellate cells and liver, smooth muscle cells play a limited role since, thus maintain a low portal Hepatic blood flow regulation although present in hepatic arterioles, they are only pressure. found in limited numbers in portal venules [32]. Liver sinusoids have a dual blood supply, receiving blood flow from the portal vein (70%) and the hep- LSECs, a selective barrier atic artery (30%) [24]. Blood pressure equalizes in the sinusoid and blood is then drained into the hep- LSECs are positioned at an interface. On their atic veins and the inferior vena cava. Despite major sinusoidal side, they are exposed to the highly

Journal of Hepatology 2017 vol. 66 j 212–227 213 Review A This distribution could be related to the progressive Liver primordium decrease in oxygen tension along the lobule accom- Foregut panied with an increasing need for oxygen exchange Septum transversum [36]. Alternatively, this distribution could be a mar- V D ker of LSEC maturation as they spread along the lob- Blood vessels ule [39]. Fenestrae are not static structures. Their number and size varies in physiological conditions like fasting that decreases the number but increases B the size of the fenestrae [40] and in pathological Hepatoblasts conditions [36,39,41–43]. Using super-resolution (liver primordium) Differentiated Future LSEC optical microscopy, Mönkemöller and colleagues LSEC Foregut recently showed that sieve plates are surrounded and separated by microtubuli and that each fenestrae within a sieve plate is surround by actin ﬁla- ments [38]. Cytoskeleton is thus of great importance for the LSECS fenestrations. Fifteen years ago, LSEC fenestrations were thought to be sort of caveolae [44]. Caveolae are uncoated plasma mem- Time brane invaginations found in lipid-ordered domains of cell membranes called lipid rafts. Caveolin is a 58 12(Weeks of gestation) major structural protein of caveolae. Although caveolin-1 has been observed in LSECs fenestrations Markers CD4; CD32; ICAM-1; Fenestrae [44], caveolin-1 knockout mice have normal fenes- expressed by LSECs CD31; CD34; 1F10 trations [45]. In addition, Svistounov and colleagues [46,47] described the ‘‘sieve-raft crosstalk”, where fenestrations are formed in reduced lipid-raft Fig. 1. Formation of sinusoids during human embryogenesis. (A) Frontal section of an embryo showing the formation of an outgrowth of the foregut (endoderm), called the liver primordium, which regions of endothelial cells. Thus, fenestrations are extends into the septum transversum (mesoderm), in which blood vessels are developing. V, ventral; D, not dependent on caveolin-1 and are different struc- dorsal. (B) Transversal section of the embryo showing the liver primordium (i.e., hepatoblasts arranged tures from caveolae. in thick cords separated by vascular spaces) growing into the septum transversum. The hepatic In a normal liver, LSECs retain blood cells in the sinusoids are progressively established. First, the endothelial lining is continuous with a basement membrane (pink region) and no fenestrations. Around gestation week 12, fenestrations appear initially vessels, while molecules, such as metabolites, with diaphragms. These diaphragms disappear during development [15,170,171]. plasma proteins, pharmaceutical drugs, lipoproteins and small chylomicron remnants, viruses (<200 nm) and exosomes can access the space of Disse to be oxygenated arterial blood mixed with the portal taken up by hepatocytes and hepatic stellate cells blood derived from the gut and the pancreas con- [2,38,48]. There is no signiﬁcant osmotic and hydro- taining nutrients, bile acids, and hormones includ- static pressure gradient across the normal liver sinu-

Review ing insulin and glucagon. On the abluminal side, soids [41,49]. Small molecules and gasses freely they interact with hepatic stellate cells and hepato- diffuse through the fenestrae, so that the space of cytes that are crucial for protein, lipid and glucose Disse contains a para-vascular part of the plasma metabolism. LSECs thus represent a permeable bar- volume. In addition, as blood cells squeeze into the rier allowing exchanges but also active uptake and sinusoids, they massage the endothelial cells and degradation of molecules [35]. further mix plasma and space of Disse fluids [49]. Larger molecules, may also cross LSEC by a process Key point Fluid exchange through fenestrae called permselectivity or ‘‘sieving”, namely the LSECs act as a selective bar- Like endothelial cells located in other exchange ter- restricted transport of large molecules due to their rier, since exchanges occur ritories, such as the glomeruli, the spleen and the deformation capacity through membrane pores through fenestrae as well as bone marrow, LSECs are highly permeable [36]. [41]. The fluid present in the space of Disse is by transcytosis and LSEC The association of fenestrae, absence of diaphragm drained into hepatic lymphatics, then into hepatic scavenging functions. and lack of basement membrane make them the hilum lymphatics, cisterna chili, thoracic duct and most permeable endothelial cells of the mam- eventually the central venous circulation, succes- malian body [24]. These fenestrae are organized sively [50]. The fluid formed in excess gains free in clusters termed sieve plates [37]. LSEC fenestrae access to the Glisson’s capsule on the liver surface have a diameter ranging from 50 to 150 nm [49]. Contrary to the mesentery, the liver is thus [2,37,38]. Their size and number varies depending leaky to large molecules including proteins. This on their localization in the liver, with larger but explains why ascites related to post-sinusoidal fewer fenestrae per sieve plate in the periportal obstruction, such as cardiac failure or Budd-Chiari region and smaller but more numerous fenestrae syndrome, is protein rich while ascites resulting per sieve plate in the centrilobular region [37,39]. from cirrhosis is not [50].

214 Journal of Hepatology 2017 vol. 66 j 212–227 JOURNAL OF HEPATOLOGY

VASODILATION VASOCONSTRICTION

E ↑Resistance E NO LSEC LSEC ↑Shear stress Shear stress Statins NO NO

↑KLF2 ↑KLF2 x Endothelial dysfunction Fenestrae BM sieve plate E HSC E Activated HSC Quiescent NO Collagen

NO Space of Disse Space of Disse NO

Hepatocyte Hepatocyte

NORMAL LIVER CIRRHOTIC LIVER

Fig. 2. Role of liver sinusoidal cells (LSECs) in chronic liver diseases. In normal conditions, LSECs maintain hepatic stellate cell quiescence through a NO-dependent pathway as long as they are differentiated [101]. Exposure of LSECs to a physiological shear stress activates the transcriptor factor KLF2 leading to the release of vasodilating agents including nitric oxide (NO) and to the downregulation of vasoconstrictive molecules including endothelin-1. In a cirrhotic liver, LSECs become capillarized, meaning that they lose their fenestrae and a basement membrane appears. Capillarized LSECs permit hepatic stellate cell activation and thus production of collagen and of ﬁbrosis. This change is associated with an endothelial dysfunction meaning that increased shear stress no longer leads to vasodilation but rather to vasoconstriction and thus to increased intrahepatic resistance. Simvastatin restores the vasoprotective effect of KLF2 and improves HSC phenotype through a NO-dependent pathway (the effect of simvastatin appears in red) [102]. BM, basement membrane; E, endothelin; HSC, hepatic stellate cell; KLF2, Kruppel-like factor 2; LSEC, liver sinusoidal cell; NO, nitric oxide. Review

Endocytic capacity acLDL), advanced glycation end products and waste LSECs have one of the highest endocytic capacity in products (hyaluronan, chondroitin sulfate or N- the human body [51]. This property combined with terminal propeptides of procollagen (I, III)). The a strong lysosomal activity give LSECs the ability to main SRs of LSECs are SR-H/stabilin-1 and SR-H/ clear waste from the blood, as part of the ‘‘dual-cell stabilin-2. Stabilin1/2 double-knockout mice show principle” of waste clearance. This principle states only a mild liver fibrosis without liver dysfunction that the mononuclear system represents the pro- but a severe renal glomerular fibrosis [53], suggest- fessional phagocyte, eliminating large particles, ing that stabilin-1 and 2 are major liver endocytic and that the scavenger endothelial cells, including receptors implicated in the clearance of molecules LSECs, represents the professional pinocyte, clear- toxic mainly for the kidney. The mannose receptors ing soluble macromolecules and small particles are not specific of LSECs and bind a wide range of through endocytic receptors [52]. This property glycoproteins and microbial glycans, such as colla- can be used to specifically target LSECs. LSEC endo- gen alpha chains (I, II, III, IV, V, XI), tissue plasmino- cytosis also contributes to the transfer of molecules gen activator regulating fibrinolytic activity, and from the sinusoids to the space of Disse, a process lysosomal enzymes that are recruited for further called transcytosis [35]. Endocytosis by LSECs use in LSEC [54]. Thus, they have a role both in implies different high affinity endocytosis recep- immunity and in glycoprotein homeostasis [52]. tors, including scavenger receptors (SR-A, SR-B The Fc gamma-receptor IIb2 is the only Fc gamma- and SR-H), mannose receptor and Fc gamma- receptor expressed by LSECs and mediates the clear- receptor IIb2 [51,52]. The SRs mediate endocytosis ance of small circulating immune complexes; LSEC of polyanionic molecules, such as oxidized and play a role in vascular immunity through this recep- acetylate low-density lipoproteins (oxLDL and tor [51,52].

Journal of Hepatology 2017 vol. 66 j 212–227 215 Review À Technical aspects for the study of LSECs cytes. LSECs are usually described as CD45 , and liver CD45+ cells are often considered as Kupffer Markers of LSEC cells. However, the reality may be more complex, as LSEC CD45 positivity appears to depend on the Identification and isolation of LSECs is a major localization and the differentiation of LSECs Key point challenge for the understanding of liver physiology [24,39]: bright CD45 positivity is found in periportal and diseases. However, technical barriers as well as area where LSECs have less fenestration, while CD45 There is no unique specific a lack of consensual specific LSEC markers explain negativity appears to predominate in centrilobular marker of LSECs, apart from that LSECs populations differ between research areas where LSECs are more differentiated with their fenestrae devoid of dia- groups, which limits the interpretation and the more fenestrae. phragm in the absence of comparison of the results. Knowledge of LSEC markers helps understanding basement membrane. A Ò combination of markers is Features used to identify LSECs include: (a) their some drug adverse effects. For instance, Mylotarg thus mandatory for their high and rapid endocytic capacity, using labeled (gemtuzumab ozogamicin), a drug used for acute identification. formaldehyde-treated serum albumin, collagen myeloid leukemia, consists of a humanized antibody alpha chains or acLDL. As other cells, including anti-CD33, linked to a potent antitumor antibiotic Kupffer cells, also have endocytic capacities, (calicheamicin). CD33 is expressed on the surface labeled molecules have to be incubated in small of acute myeloid leukemia cells, but also of LSECs amount and for a short period of time to be specific likely explaining the high prevalence of hepatic for LSECs [55]. (b) Fenestrae without diaphragm sinusoidal obstruction syndrome following this and organized in sieve plates, using electron micro- treatment [60]. scopy. Although this feature is the only one specific of LSECs, it has some limitations. First, the distribu- LSECs culture tion of the fenestration varies along the lobule [37]. Second, LSEC isolation methods, including liver per- As mentioned above, obtaining a pure culture of pri- fusion and cell preparation for electron microscopy, Key point mary LSECs is challenging because of the lack of dilate fenestrae and might even create holes in cell specific markers of these cells. LSEC isolation proto- When cultured, primary surface [55]. Third, fenestrae rapidly disappear cols are detailed elsewhere [61,62]. The culture of LSECs rapidly lose their when LSEC are cultured as a monolayer of cells, LSECs has at least four particularities. First, cultured specific phenotype. How- out of their environment [56]. This loss of fenestra- LSEC tend to lose their typical phenotype. In order to ever, human and murine tion associated with basement membrane synthe- immortalized LSECs lines prevent this dedifferentiation, several methods have have been successfully sis and modification of the expression of surface been developed. Co-culture with hepatocytes and developed. markers is called capillarization. Capillarization fibroblasts rather than with hepatocytes alone not only happens in cultured LSEC but also in vivo allows LSECs to maintain their phenotype for up to in most liver diseases [56]. (c) surface markers 2 weeks [63]. Extracellular matrix coating mimick- [24] (Table 1). Some markers are common to other ing the space of Disse and its modifications in endothelial cells and some to hematopoietic cells. pathology can also be used, e.g., low-density base- No single marker is specific for LSECs and a combi- ment membrane-like matrix imitating normal con-

Review nation is required. For instance, Ding and col- ditions, and interstitial type matrix (fibril-forming + À leagues considered that LSEC are VEGFR3 CD34 collagen) imitating cirrhosis [63]. The addition of + + + À VEGFR2 VE-Cadherin FactorVIII CD45 [57], VEGF to the medium or the use of hepatocyte- + while Lalor and colleagues selected CD31 , LYVE- conditioned medium can also prevent LSECs dedif- + + + À À 1 , L-SIGN , Stabilin-1 , CD34 , PROX-1 cells ferentiation [56,64,65]. Second, when cultured [56]. CD31, CD45 and CD33 deserve specific com- alone, LSECs undergo apoptosis within 2 days [63]; ments. CD31 (PECAM-1) is an intercellular adhe- methods preventing dedifferentiation also prevent sion molecule classically expressed at the surface cell death. Third, serum supplementation is toxic of endothelial cells, but also of several leukocytes for LSECs [55]. Fourth, in the normal liver, LSECs [58]. The expression of CD31 by LSECs is controver- are exposed to an oxygen pressure decreasing along sial. Several studies reported CD31 positivity of the liver lobule from 90 to 30 mmHg [66]; oxygen LSECs in liver slices analyzed by immunohisto- level is thus lower than in atmospheric conditions chemistry or in cultured cells permeabilized before where oxygen pressure is 160 mmHg; actually, LSEC staining [55]. Conversely, for the isolation of LSECs are particularly sensitive to hyperoxia and to the using flow cytometry, LSECs are considered as resulting oxidative stress [67]; survival of primary CD31 negative, CD31 positive cells being arterial LSECs is improved under 5% oxygen instead of the and venous endothelial cells as well as capillarized commonly used 20% [51,66]. LSECs. An electron microscopic analysis reconciled To overcome the difficulties of culturing primary these results by showing that CD31 is located intra- LSECs, several teams have developed human and cellularly shortly after establishing LSEC cultures, murine immortalized LSECs lines. However, the first but, when fenestrae disappear few days later, immortalized lines, obtained by viral transfection CD31 becomes expressed at the cell surface like such as M1LEC, had no fenestrae [68–72]. Subse- in other endothelial cells [59]. CD45 is a quently, several humans and murine immortalized hematopoietic cell marker, expressed by leuco- LSEC lines have been developed. As summarized in

216 Journal of Hepatology 2017 vol. 66 j 212–227 JOURNAL OF HEPATOLOGY

Table 1. Liver sinusoidal endothelial cell markers.

Common with EC markers Endocytic markers Antigen presentation Common with leucocytes Common with lymphatic EC CD34$ CD36 CD40$ CD4 VAP-1 CD105* DC-SIGN CD80$ CD11b CD146 L-SIGN CD86$ CD11c$ Cytoplasmic CD31 Lectins Fc Gamma R (CD32b**) CD33 ICAM-1 LYVE-1 Mannose R CD45$ Ulex Lectin binding SR-A/SR-B MHC I/MHC II$ Cytoplasmic CD31 vWf (Factor VIII)$ Stabilin-1 Stabilin-2 Uptake of acLDL or denatured alpha-collagen chain

⁄ ⁄⁄ Also expressed by hepatic stellate cells and myofibroblasts; correlates with fenestration and corresponds to SE-1 in rats [63]; $controversial [55]. AcLDL, acetylated low-density lipoprotein; Ag, antigen; CD, cluster of differentiation; DC-SIGN, dendritic cell-specific intercellular adhesion molecule-3 grabbing non- integrin; EC, endothelial cells; ICAM, intracellular adhesion molecule; L-SIGN, liver specific intercellular adhesion molecule-3 grabbing non-integrin; LDL, low-density lipoprotein; LYVE, lymphatic vessel endothelial hyaluronan acid receptor; MHC, major histocompatibility complex; R, receptor; SR, scavenger receptors; VAP, vascular adhesion protein-1; vWF, von Willebrand factor.

Table 2. Liver sinusoidal endothelial cell lines features.

Human lines Rodent lines Authors Parent et al. Cogger et al. Matsumura et Hering et al. Zhao et al. Huebert et al. Maru et al. [145] [146] al. [68] [69] [147] [148] [70] Name TRP3 SK Hep1 TMNK-1 iSEC n.a. TSEC NP11, NP26, NP31, and NP32 Origin Hereditary Ascitic fluid from Human liver Human fetal liver Mouse Mouse Rat hemorrhagic a patient with endothelial cells telangiectasia hepatocellular patient carcinoma Method of Lentivirus Spontaneous Lentivirus Transfection Spontaneous Lentivirus (SV40) Lentivirus (SV40) immortalization (hTERT) (SV40 and with polyoma hTERT) virus large tumor antigen Fenestration Few (data not Yes n.a. n.a. Yes Few n.a. organized in shown) sieve CD31 n.a. Not at the surface Yes (RNA) n.a. Yes surf. Yes surf. and n.a. cytop. Uptake of Yes Yes (uptake of Yes n.a. Yes Yes Yes

acLDL FITC-FSA) Review Tube forming Yes Yes Yes Incomplete Yes Yes Only in NP11 and NP26 vWf Yes perm. Yes surf. Yes (RNA) Yes cytop. Yes perm. Yes surf. and No cytop. CD34 Yes perm. n.a. Yes (RNA) n.a. n.a. n.a. n.a. Other CD32b, Stabilin-2, NOS, VEGFR2 Collagen IV, Chemotaxis VEGFR-1 LYVE-1 and fibronectine in response (low level) cytoplasmic to angiogenic L-SIGN growth factors acLDL, acetylate low-density lipoproteins; Cytop., cytoplasmic; L-SIGN, liver speciﬁc intercellular adhesion molecule-3 grabbing non-integrin; LYVE, lymphatic vessel endothelial hyaluronan acid receptor; FITC, ﬂuorescein isothiocyanate; FSA, formaldehyde-treated serum albumin; n.a., not available; Perm., Permeabilized; Surf., Surface; Spont., Spontaneous; VEGFR, vascular endothelial growth factor receptor; vWF, Von Willebrand factor.

Table 2, these cell lines display many characteris- Mouse models tics of LSEC. Each cell line has particular advantages making it more appropriate for specific studies. For Transgenic mice using the Cre/Lox system can be instance, TSECs are adequate for angiogenesis anal- very useful to study the properties of LSECs in vivo. yses and Sk Hep1 for fenestration. However, the Briefly, Cre-recombinase, which can be regulated fact that these cell lines are immortalized implies by a tissue-specific promoter, excises essential that they may react differently from primary cells loxP-flanked (‘‘floxed’’) genes via intrachromosomal in response to stress. Therefore, a confirmation of recombination to generate so called conditional the findings using primary cells is useful. knockouts, i.e., knockouts specifically affecting

Journal of Hepatology 2017 vol. 66 j 212–227 217 Review

Table 3. Characteristics of transgenic mice available to study the properties of liver endothelial cells in vivo.

Transgenic mice [Ref.] Constitutive/ Liver endothelial expression in adults Expression by hematopoietic Limitation inducible Portal vein Sinusoids Centri lobular cells in adults vein PECAM1 -Cre [149] Constitutive n.a. n.a. n.a. Likely Poorly described Tie1 -Cre [150] Constitutive n.a. Good Good Yes (20%) Hematopoietic cell expression Tie2 -Cre [151] Constitutive Good Good Good Yes (90%) Strong hematopoietic cell expression Flk1 -Cre [17,152] Constitutive n.a. Moderate n.a. Yes Hematopoietic cell expression Cdh5-Cre [12,14] Constitutive Good Good Good Yes (50%) Moderate hematopoietic cell expression Tie2 -CreERT2 [153] Inducible Good Absent n.a. No No expression in LSEC Endothelial-SCL-CreERT2 [154] Inducible Good Absent Absent No No expression in LSEC Cdh5-CreERT2 [155] Inducible n.a. Mild Mild No Less penetrant than Cdh5(PAC)-CreERT2 Cdh5(PAC)-CreERT2 [156] Inducible Good Good Good No Pdgfb-iCreERT2 [157] Inducible n.a. Absent Good No during the first month No expression in LSEC after induction of Cre- mediated recombination Bmx cre [158] Constitutive Absent Absent Absent n.a. Artery specific

Recombination was classiﬁed as good (>66%), moderate (33–66%), mild (5–33%); absent (<5%) based on data provided in the articles describing each model for all mouse lines but Tie2-Cre, Pdgfb-iCreERT2 and Cdh5 (PAC)-CreERT2. Indeed, these last 3 lines were independently and thoroughly analyzed and compared back to back using mT/ mG reporter mice by the group of C James, Pessac, France (Kilani et al., unpublished). Cdh5-CreERT2 mice were also analyzed using mT/mG reporter by our group (unpublished data). Regarding LSEC expression, caution is needed since in all cases LacZ staining was performed without immunohistochemistry. Cells considered as LSEC were thus sinusoidal cells. They may be LSEC but also may be Kupffer cells. LSEC, liver sinusoidal endothelial cell; n.a., information not available.

tissues where the promoter is expressed. Several sinusoid capillarization, angiogenesis, angiocrine models with an endothelial cell expression of the signals and vasoconstriction. Cre-recombinase have been developed and are sum- Capillarization of LSECs, also called dedifferentia- Key point marized in Table 3. Mice with a constitutive expres- tion, occurs following liver injury in animal models The loss of the specific phe- sion of the Cre-recombinase appeared first. as well as in patients [75–80]. Capillarization is an notype of LSECs, including However, the expression of the Cre-recombinase is early event since it precedes the activation of hep- the disappearance of the fen- not restricted to endothelial cells, especially in adult atic stellate cells and macrophages and the onset estrae, the development of a mice, as recombination also occurs in hematopoietic of liver fibrosis, suggesting that it could be a prelim- basement membrane, and cells. Indeed, early embryonic endothelial and inary step necessary for fibrogenesis [76,81,82]. The

Review the appearance of specific hematopoietic cells arise from a common embryonic mechanisms of capillarization and the cross talk markers is called capillarization and is an early even in precursor called the hemangioblast [14]. This limita- between LSECs and hepatic stellate cells has been chronic liver injury. When tion can be overcome by performing a transplanta- reviewed elsewhere [83]. Briefly, LSECs are able to capillarized, LSECs lose their tion of wild-type bone marrow together with a maintain hepatic stellate cells quiescent as long as capacity to inactivate hepatic clodronate mediated Kupffer cell depletion [73]. they are differentiated so that differentiated LSECs stellate cells, thus promoting Indeed, in the absence of clodronate treatment, are gatekeepers of fibrosis [34,84]. VEGF contributes fibrogenesis and intrahepatic 2 months after bone marrow transplantation, 85% to the maintenance of LSEC differentiation (Fig. 3). vasoconstriction. of the Kupffer cells are still derived from the recipi- The role of LSECs in fibrosis regression is less clear. ent [74]. Myeloablation conditionings required for Indeed, in experimental models, restoration of LSEC bone marrow transplantation might however alter differentiation in vivo promotes regression of mild LSEC function. Another way to overcome the con- fibrosis [34,85]. However, immunohistochemical comitant expression of the Cre-recombinase in analysis of paired liver biopsies from 38 hepatitis C endothelial and hematopoietic cells is to use trans- virus patients with cirrhosis, before and after antivi- genic lines where Cre expression is induced in adult ral treatment, revealed that sinusoid capillarization endothelial cells after tamoxifen administration. In persists despite the regression of cirrhosis. LSEC dif- that case, there is no expression of the transgene in ferentiation is thus not crucial for fibrosis regression hematopoietic cells. in this setting [86]. Angiogenesis is defined by the development of LSECs in liver diseases new vessels from preexistent vessels [87]. Hepatic angiogenesis occurs during liver fibrogenesis and Chronic liver diseases these two processes are closely linked [88,89]. Liver fibrosis enhances angiogenesis and, in turn, liver LSECs play a key role in chronic liver disease initi- angiogenesis aggravates liver fibrosis, as attested ation and progression, through four processes: by the anti-fibrotic effect of most anti-angiogenic

218 Journal of Hepatology 2017 vol. 66 j 212–227 JOURNAL OF HEPATOLOGY

Differentiated LSEC Capillarized LSEC

Fenestrae BM sieve plate Fibrosis Angiogenesis HSC HSC TGF-β1 Activated Quiescent

↑VEGF VEGF Space of Disse Space of Disse

Hepatocyte Cholangiocyte Hepatocyte

NORMAL LIVER CIRRHOTIC LIVER

Fig. 3. A dual role of VEGF in chronic liver disease progression. In physiological conditions, VEGF released by hepatocytes, cholangiocytes and HSC, maintains LSEC differentiation (blue arrow) and consequently HSC quiescence. VEGF is thus anti-fibrogenic. During fibrogenesis, liver expression of VEGF increases. These high VEGF levels have a pro-fibrogenic action (red arrows) by inducing liver angiogenesis and by activating HSC. The activation of HSC results from a direct action of VEGF on HSC and from the release of TGF-b1 by capillarized LSECs. BM, basement membrane; HSC, hepatic stellate cell; LSEC, liver sinusoidal cell; VEGF, vascular growth factor; TGF-b1, transforming growth factor b1.

agents in animal models of liver fibrosis [90,91]. Endothelial dysfunction occurs early in chronic However, analysis of the relationships between liver disease, even before fibrosis and inflammation angiogenesis and fibrogenesis is not straightfor- take place, and persists in advanced cirrhosis ward since most tools used to inhibit angiogenesis [84,99,100] (Fig. 2). The mechanisms of endothelial also act on fibrogenesis. For instance, VEGF, the dysfunction have been reviewed elsewhere and are master regulator of angiogenesis, is also implicated summarized in Fig. 2 [83,84]. Importantly, pharma- in fibrogenesis (Fig. 3) [87,92–95]. Besides LSECs, cologic strategies improving LSECs in chronic liver Review endothelial progenitor cell (EPC), i.e., endothelial diseases, including statins, decrease liver fibrosis, cells derived from bone marrow, also contribute endothelial dysfunction and portal pressure to liver angiogenesis, as reviewed elsewhere [101,103,104]. [96,97]. LSECs also regulate fibrosis by releasing angio- Role of LSECs in hepatocellular carcinoma crine signals. This latter term refers to the paracrine factors produced by endothelial cells that maintain Hepatocellular carcinoma (HCC) most often emerges organ homeostasis, balance the self-renewal and in the context of chronic liver disease. The develop- differentiation of stem cells and orchestrate organ ment of HCC is thought to be a multistep process regeneration and tumor growth. A recent landmark from precancerous lesions (low then high grade dys- study demonstrated that LSECs release divergent plastic nodule) to early and advanced HCC [105]. angiocrine signals balancing liver regeneration Dysplastic nodules receive blood supply preferen- and fibrosis. After acute liver injury, activation of tially via the portal vein similarly to regenerative CXCR7-Id1 pathway in LSECs stimulates production nodules of cirrhosis. A switch to prominent arterial of hepatic-active angiocrine factors leading to liver blood supply occurs at the stage of early HCC regeneration. By contrast, chronic injury causes [106]. Then, angiogenesis results in a highly vascu- persistent FGFR1 activation in LSECs that perturbs larized tumor and promotes tumorigenesis and the CXCR7-Id1 pathway and favors a CXCR4-driven development of metastasis. HCC is associated with pro-fibrotic angiocrine response, thereby provoking changes in endothelial cells within and around the liver fibrosis. Therefore, in response to injury, dif- tumor. ferentially primed LSECs deploy divergent angio- Endothelial cells present within HCC sequentially crine signals to balance liver regeneration and lose during tumor progression LSECs markers, fibrosis [98]. including stabilin-1, stabilin-2, LYVE-1 and CD32b,

Journal of Hepatology 2017 vol. 66 j 212–227 219 Review as observed both in murine HCC models and in [112,114]. Shear stress is thus a key inducer of liver human HCC [107]. Moreover, as compared to LSECs regeneration. However, when resection is excessive, from a healthy human liver, endothelial cells exaggerated shear stress can damage LSECs and lead derived from human HCC have a higher expression to hemorrhagic necrosis [112]. Limiting shear stress of integrins, lower expression of ICAM-1, and exhi- could be a potential strategy to prevent post- bit higher angiogenic, procoagulant and fibrinolytic hepatectomy liver failure as suggested by the bene- capacities [108]. ficial effect of portosystemic shunts, splenectomy or LSECs in the peritumoral tissue also undergo splenic artery embolization in murine models and in changes as HCC progresses including the loss of patients with large liver resections [112,115–121].A the LSEC markers stabilin-2 and CD32b [107]. In a less invasive surgical intervention is being tested in mouse tumor xenograft model, peritumoral liver a prospective trial (NCT02390713), using a pneu- tissue displays a higher microvascular density and matic ring to modulate the diameter of the portal expression of the proangiogenic genes, vein and thus the post-hepatectomy shear stress. interleukin-6 (IL-6) and interleukin-6 receptor (IL- New promising molecules decreasing shear stress 6R) than the model tumoral tissue [109]. In the to prevent post-hepatectomy liver failure and same line, peritumoral endothelial cells isolated small-for-size-syndrome have been proposed from patients with HCC proliferate more when cul- including the vasodilator olprinone, a phosphodi- tured with IL-6 and soluble IL-6R than tumoral esterase III inhibitor [122,123] currently tested in a endothelial cells. IL-6 is secreted by peritumoral prospective trial (NCT00966745). endothelial cells in response to hypoxia while IL- In the second phase following hepatectomy (at 6R is secreted by macrophages, present in large day 4), LSECs begin to proliferate, via the upregula- number in the peritumoral liver tissue during tion of angiopoietin-2 and VEGFR2/VEGFA pathways tumoral progression. These data suggesting a major [111]. VEGFR2 is a classical mediator of the mito- role of peritumoral endothelial cells in HCC pro- genic and the angiogenic effect of VEGFA. The role gression echo the previous observation that gene of VEFGA/VEGFR1 pathway is more controversial expression in the nontumoral liver from patients than that of the VEGFR2 pathway. Le Couteur et al. with HCC has a higher prognostic value of than described that VEGFR1 activation in LSECs after liver gene expression in HCC [110]. injury, can paracrinally induce hepatocyte proliferation, without LSEC proliferation and protects LSEC and liver regeneration following acute liver parenchymal cells from the injury [124]. Key point injury or partial hepatectomy Liver regeneration not only implicates liver cells but also circulating cells including sinusoidal pro- LSECs are implicated in liver genitor cells, platelets and inflammatory cells. The regeneration following acute role of sinusoidal progenitor cells in liver regenera- liver injury or partial hepa- Liver regeneration following acute liver injury tectomy since they renew or partial hepatectomy is a complex process where tion has been recently reviewed elsewhere and is from LSECs and/or LSEC pro- LSECs play a key role. LSECs sense the major summarized in Fig. 4 [20]. Briefly, liver injury genitors, they sense the changes in shear stress resulting from resection. induces increased hepatic VEGF expression, which

Review shear stress changes result- They proliferate, and orchestrate the harmonious drives recruitment of hepatocyte growth factor- ing from surgery and inter- regeneration of the different cell types by interact- rich bone marrow sinusoidal progenitor cells and act with platelets and promotes expression of HGF by resident sinusoidal inflammatory cells. ing with sinusoidal progenitor cells, platelets and inflammatory cells (Fig. 4). progenitor cells and LSECs. HGF in turn stimulates After an acute liver injury or a partial hepatec- the proliferation of hepatocytes in liver regenera- tomy, LSECs play a central role in liver regeneration tion. In addition, sinusoidal progenitor cells replace through a dynamic regulation of the balance LSECs that were lost during injury. The role of the between hepatocytes proliferation and vascular interaction between LSECs and platelets in liver proliferation. There is an asynchronism between regeneration is summarized in Fig. 4 [125]. Follow- hepatocyte and LSEC proliferation. In the early ing liver injury, platelets are recruited to and phase (at day 2), non-proliferative LSECs activate trapped within the liver, where they adhere to LSEC. hepatocytes proliferation by two complementary Subsequent platelet activation results in the release mechanisms: (a) the downregulation of the hepato- of platelet granules, which stimulate hepatocyte cyte growth inhibitor TGF-b, through the downreg- proliferation. Platelets activate LSECs, leading to ulation of the Tie2 receptor antagonist, the secretion of growth factors, such as IL-6 [125]. angiopoiteine-2 [111]; and (b) the secretion of hep- Finally, LSECs and hepatocytes can also internalize atotropic cytokines, Wnt and hepatocyte growth platelets, but the effects of this alternate process factor (HGF), through the upregulation of the tran- on liver regeneration remain to be explored. The scription factor Id1 via the VEGFR2/VEGFA path- improvement in survival following subtotal liver ways [57]. Following liver resection, the portal resection in rats and mice obtained by the induction flow per gram of tissue immediately increases, of thrombocytosis by thrombopoietin injection, enhancing the shear stress on LSECs [112,113]. In splenectomy or platelet-rich plasma transfusion response to this increased shear stress, LSECs illustrates importance of platelets in liver regenera- release NO that sensitizes hepatocytes to HGF tion [126–128]. The endothelial-monocyte interac-

220 Journal of Hepatology 2017 vol. 66 j 212–227 JOURNAL OF HEPATOLOGY

Hepatocytes

+ Space of Disse + HGF NO

LSEC

Platelets Shear stress VEGF A +

Engraftment Differentiation

Space of Disse

VEGF Proliferation Liver injury SPCs

+ Proliferation Release Stimulation Mobilization

Fig. 4. Liver sinusoidal cell (LSECs) and liver regeneration following acute liver injury or partial hepatectomy. Following liver injury, liver expression of VEGF increases, leading to the proliferation of bone marrow sinusoidal progenitor cells (SPC) (1), to their mobilization to the circulation (2), their engraftment in the sinusoids (3) and their differentiation in mature LSECs (4). VEGF A stimulates liver regeneration trough LSECs (a) leading to HGF production (b), hepatocyte proliferation (c) and LSECs proliferation (d) [20]. Increased shear stress associated with liver resection induces LSECs derived nitric oxide (NO) (I), which increase the effect of HGF on hepatocytes proliferation (II). Platelets are rapidly recruited in the liver after liver surgery (A). They adhere to LSECs and stimulate secretion of key molecules involved in hepatocytes (F) and LSECs (B) proliferation and survival. Platelets can also be endocytosed by LSECs (C), or trapped in the space of Disse (E), a migration facilitated by the increased size of fenestration associated with liver surgery (D). Abbreviations: HSC, hepatic stellate cell; NO, nitric oxide; LSEC, liver sinusoidal cell; HGF, hepatocyte growth factor; VEGF, vascular growth factor; VEGFR, vascular growth factor receptor; SPC, sinusoid progenitor cells.

tion is also implicated in liver regeneration. Indeed, increases and expression of vascular cell adhesion circulating monocytes are recruited in the injured molecule-1 (VCAM-1) and CD31 are induced, lead- Review liver and stimulate parenchymal but also endothe- ing to the transendothelial migration of leucocytes. lial regeneration. LSECs regulate the infiltration of Stabilin-1 has also been reported to promote monocyte in the liver through a destabilization of transendothelial migration of leucocytes, preferen- VE-Cadherin junction and through adhesive mole- tially regulatory T cells [132]. Second, LSECs can cule expression [129]. modulate lymphocytes behavior. In physiological Lastly, liver regeneration also depends on the conditions, antigen presentation by LSECs leads to existence of lesion related to ischemia reperfusion. tolerance induction in CD8+ cells [133]. LSECs can Mechanisms of ischemia reperfusion injury have also induce differentiation of T cells into immuno- been reviewed previously and will not be detailed suppressive regulatory T cells (Treg) that are func- here [130]. tional in vitro and in vivo [134]. As an application, the selective delivery of autoantigen peptides to Inflammation and infection LSECs in vivo using a polymeric nanoparticle carrier can efficiently prevent and treat an animal model of LSECs regulate liver inflammation in two manners. autoimmunity, by increasing the number of Treg First, LSECs are a barrier separating the blood from [135]. In inflammatory conditions, LSECs also tend the rest of liver, and thus restrict or enable the to have an anti-inflammatory action since they entry of circulating leucocytes into the liver tissue. increase the expression of the anti-inflammatory The detailed mechanisms of the interactions cytokine IL-10 in Th1 cells via the Notch pathway between leucocytes and LSECs have been previ- [136]. ously reviewed [131]. Briefly, LSECs express But LSECs can also be targeted by pathogens. Due ICAM-1 and vascular adhesion protein-1 (VAP-1), to their scavenging ability, LSECs can capture circu- allowing adhesion of leucocytes to the endothe- lating viruses via the expression of lectin at their lium. During inflammation, expression of ICAM-1 surface and in turn induce infection of hepatocytes,

Journal of Hepatology 2017 vol. 66 j 212–227 221 Review

Table 4. Drug delivery system to LSEC in vivo.

Carrier distribution in vivo Reference Carrier Size (nm) LSEC* KC* Hep* Other organs Experimental strategies and results Toxicity Sano et al. HA n.a. Yes n.a. n.a. n.a. Delivering of sphingosine-1- n.a. [159] phosphate Hepatic I/R injury model in rats →↓ALT level and hepatocytes and LSECs apoptosis Fraser et al. HA n.a. 90% 0 4% Low in spleen Bio-distribution in rats n.a. [160] Toriyabe et al. HA-SA-liposome 254 ± 19 Yes Yes (NQ) No (NQ) Low expression in Bio-distribution in mice n.a. [161] lungs Takei et al. PLL-g-HA 100 to 200 Yes No (NQ) No (NQ) >93% liver Delivering of DNA complexes n.a. [162] 2.5-1% spleen, Bio-distribution in rats intestine and kidney <1% heart, thymus, lung and blood$ Kren et al. HA-nanocapsule <50 Yes n.a. No (NQ) Not in lung, kidney, Delivering of transposon vectors No [163] (polyethyleneimine) spleen, heart and expressing FVIII toxicity gonads Hemophilia A mice model in 72 h →Normalization of plasmatic and 3 FVIII expression and activity mo up to 11 mo Carambia et Iron oxide <50 Yes n.a. No (NQ) 90% liver Delivering of auto-antigen peptide No al. [135] nanocrystals 10% spleen and Autoimmune encephalomyelitis mice toxicity kidney$ model (9 wk) →Controlled disease progression by Treg induction by LSECs Tanoi et al. STR-KLGR 80-120 Yes n.a. Lower (NQ) n.a. Delivering of BAX siRNA 100% [164] modified YSK05- Acute liver damage (anti-FAS Ab) alive at MEND in mice 24 h →↓Hepatocytes apoptosis. Preserve sinusoidal structure Akhter et al. STR-KLGR 80-120 Yes n.a. Low High in liver Delivering of Tie2 siRNA No liver [165] modified YSK05- Lower expression in Bio-distribution in mice toxicity MEND lung and kidney →80% knockdown in LSECs (24 h) Bartsch et al. Aco-HSA-CCLs 154 ± 12 60% 40% 1.30% 60% liver Delivering of ODN n.a. [166] 4% spleen Bio-distribution in rats <1% lungs, heart, kidneys£ Kamps et al. Aco-HSA 92.1 ± 10 65% 25% 10% 80% liver Bio-distribution in rats n.a. Review [167] liposomes 5% spleen£ Bartsch et al. Aco-HSA-PEG- 164 ± 45 75% 25% n.a. 80 % liver Delivering of anti-ICAM-1-ODN n.a. [168] SAPLs 5% spleen£ No efficiency analyze in vivo ⁄ % of expression in liver cells; $Express as % of body distribution; £Express as % of injected dose. Ab, antibody; Aco-HAS, cis-aconitylated human serum albumin; ALT, alanine aminotransferase; CCLs, lipid-coated cationic lipoplexes; Hep, hepatocytes; HA, hyaluronic acid; I/R, ischemia reperfusion; ICAM, intracellular adhesion molecule; KC, Kupffer cells; LSECs, liver sinusoidal endothelial cells; MEND, multifunctional type nano device; n.a., not available; NQ, not quantiﬁed; ODN, antisense oligodeoxynucleotides; PEG, polyethylene glycol; PLL-g-HA, hyaluronate-grafted poly(L-Lysine) copolymer; SA, stearylamine; SAPLs, stabilized antisense lipid particles; siRNA, small interfering RNA; STR-KLRG, sterylated killer cell lectin-like receptor subfamily G; Treg, regulatory T cells; YSK05, pH-sensitive cationic lipid [169].

as observed for hepatitis B and hepatitis C viruses transmigrating retain a suppressive phenotype, [137,138]. Lectin expressed by LSECs is not only favoring virus persistence [140]. This change in involved in regulation of the entry of viruses but endothelial cells towards a proinﬂammatory pheno- also in the regulation of their clearance by modu- type induced by CMV might explain why acute CMV lating functions of T cells as it has been shown for infection can trigger portal vein thrombosis [141]. adenovirus [139]. LSECs can also be infected with The effect of CMV on LSECs and lymphocytes may CMV (cytomegalovirus) which upregulates ICAM- also be of particular interest in the setting of liver 1 and CXCL10 expression, thus favoring CD4 T cell transplantation where CMV infection may favor transendothelial migration. Migration of effector acute rejection, a disease characterized by memory T cells through CMV-infected LSECs is endotheliitis. associated with a change in memory T cells pheno- LSECs can also be infected with bacteria as elec- type towards an activated phenotype facilitating tron microscopy studies revealed Bartonella bacilli in hepatic inﬂammation, while regulatory T cells LSECs associated with angiomatosis and peliosis

222 Journal of Hepatology 2017 vol. 66 j 212–227 JOURNAL OF HEPATOLOGY hepatitis [142]. The fact that LSECs can be targets LSECs are implicated in most liver diseases including for pathogens with an impact on the local environ- chronic liver disease initiation and progression, hep- ment might explain why nodular regenerative atocellular carcinoma development and progression, hyperplasia develops in patients with primary liver regeneration following acute liver injury or hypogammaglobulinemia, a condition frequently partial hepatectomy, liver ageing and liver lesions associated with intra-sinusoidal lymphocytic infil- related to inflammation and infection. This role in tration. Immunodeficiency might favor infection most liver diseases makes them an attractive thera- of LSECs with pathogens, leading to a change in peutic target. Data summarized in Table 4 suggest a their phenotype towards a proinflammatory and promising place to specific LSECs targeting. Such prothrombotic phenotype and eventually to sinu- cell-specific approaches may limit the adverse soid obstruction [143]. effects associated with systemic drug delivery.

LSEC and ageing Financial support Pseudocapillarization refers to changes in the liver sinusoidal endothelium related with ageing. Elec- This work was supported by the Agence Nationale tron microscopy analyses showed that ageing is pour la Recherche (ANR-14-CE12-0011 and ANR- associated with a 50% increase in the thickness of 14-CE35-0022) and by the Association Francaise LSECs, a 50% reduction in the number of LSEC fen- pour l’Etude du foie (AFEF 2014) and J.P by the estrae and the formation of a basement membrane ‘‘poste d’accueil INSERM”. with perisinusoidal fibrosis and central vein fibrosis [42,82]. These changes decrease porosity and endocytic capacity of LSECs. Consequently clearance of Conflict of interest chylomicron remnants is impaired leading to post prandial triglyceridemia, which could participate The authors declared that they do not have anything to atherosclerosis development in older individuals to disclose regarding funding or conflict of interest [42,51]. Moreover, these LSEC changes can induce with respect to this manuscript. hepatocytes hypoxemia decreasing oxidative drug metabolism and possibly promoting adverse drug reactions [42,51,144]. Authors’ contributions

JP, SL and PER drafted the manuscript. FD, CMB, RM Conclusion and DV discussed and critically revised the manuscript. In conclusion, LSECs have a unique highly permeable phenotype allowing the passage of certain but not all molecules and cells. They also have a Acknowledgments very special localization at the interface between Review blood cells on the one side and hepatocytes and We thank Servier medical art for providing some hepatic stellate cells on the other side. LSECs are images included in the ﬁgures. in constant interaction with other liver cells [83].

References morphogenesis and gene expression of sinusoidal endothelial cells. J Anat 2012;221:229–239. Author names in bold designate shared co-ﬁrst authorship [6] Gouysse G, Couvelard A, Frachon S, Bouvier R, Nejjari M, Dauge MC, et al. Relationship between vascular development and vascular differentiation during liver organogenesis in humans. J Hepatol 2002;37:730–740. [1] Chiu J-J, Chien S. Effects of disturbed ﬂow on vascular endothelium: [7] Shiojiri N, Sugiyama Y. Immunolocalization of extracellular matrix compo- pathophysiological basis and clinical perspectives. Physiol Rev nents and integrins during mouse liver development. Hepatology 2011;91:327–387. 2004;40:346–355. [2] Maslak E, Gregorius A, Chlopicki S. Liver sinusoidal endothelial cells (LSECs) [8] Asahina K, Zhou B, Pu WT, Tsukamoto H. Septum transversum-derived function and NAFLD; NO-based therapy targeted to the liver. Pharmacol Rep mesothelium gives rise to hepatic stellate cells and perivascular mesenchy- 2015;67:689–694. mal cells in developing mouse liver. Hepatology 2011;53:983–995. [3] Couvelard A, Scoazec JY, Dauge MC, Bringuier AF, Potet F, Feldmann G. [9] Jaffredo T, Nottingham W, Liddiard K, Bollerot K, Pouget C, de Bruijn M. From Structural and functional differentiation of sinusoidal endothelial cells hemangioblast to hematopoietic stem cell: an endothelial connection? Exp during liver organogenesis in humans. Blood 1996;87:4568–4580. Hematol 2005;33:1029–1040. [4] Walter TJ, Cast AE, Huppert KA, Huppert SS. Epithelial VEGF signaling is [10] Bollerot K, Pouget C, Jaffredo T. The embryonic origins of hematopoietic stem required in the mouse liver for proper sinusoid endothelial cell identity and cells: a tale of hemangioblast and hemogenic endothelium. APMIS hepatocyte zonation in vivo. Am J Physiol Gastrointest Liver Physiol 2005;113:790–803. 2014;306:G849–G862. [11] Bailey AS, Fleming WH. Converging roads: evidence for an adult heman- [5] Takabe Y, Yagi S, Koike T, Shiojiri N. Immunomagnetic exclusion of E- gioblast. Exp Hematol 2003;31:987–993. cadherin-positive hepatoblasts in fetal mouse liver cell cultures impairs

Journal of Hepatology 2017 vol. 66 j 212–227 223 Review

[12] Alva JA, Zovein AC, Monvoisin A, Murphy T, Salazar A, Harvey NL, et al. VE- [38] Mönkemöller V, Øie C, Hübner W, Huser T, McCourt P. Multimodal super- Cadherin-Cre-recombinase transgenic mouse: a tool for lineage analysis and resolution optical microscopy visualizes the close connection between gene deletion in endothelial cells. Dev Dyn 2006;235:759–767. membrane and the cytoskeleton in liver sinusoidal endothelial cell fenes- [13] Zovein AC, Hofmann JJ, Lynch M, French WJ, Turlo KA, Yang Y, et al. Fate trations. Sci Rep 2015;5:16279. tracing reveals the endothelial origin of hematopoietic stem cells. Cell Stem [39] Xie G, Wang L, Wang X, Wang L, DeLeve LD. Isolation of periportal, Cell 2008;3:625–636. midlobular, and centrilobular rat liver sinusoidal endothelial cells enables [14] Oberlin E, El Hafny B, Petit-Cocault L, Souyri M. Definitive human and mouse study of zonated drug toxicity. AJP Gastrointest Liver Physiol 2010;299: hematopoiesis originates from the embryonic endothelium: a new class of G1204–G1210. HSCs based on VE-cadherin expression. Int J Dev Biol 2010;54:1165–1173. [40] O’Reilly JN, Cogger VC, Fraser R, Le Couteur DG. The effect of feeding and [15] Martinez-Hernandez A, Amenta PS. The hepatic extracellular matrix. II. fasting on fenestrations in the liver sinusoidal endothelial cell. Pathology Ontogenesis, regeneration and cirrhosis. Virchows Arch 1993;423:77–84. (Phila) 2010;42:255–258. [16] Shiojiri N, Niwa T, Sugiyama Y, Koike T. Preferential expression of connex- [41] Fraser R, Cogger VC, Dobbs B, Jamieson H, Warren A, Hilmer SN, et al. The in37 and connexin40 in the endothelium of the portal veins during mouse liver sieve and atherosclerosis. Pathology (Phila) 2012;44:181–186. liver development. Cell Tissue Res 2006;324:547–552. [42] Le Couteur DG, Fraser R, Cogger VC, McLean AJ. Hepatic pseudocapillarisation [17] Sugiyama Y, Takabe Y, Nakakura T, Tanaka S, Koike T, Shiojiri N. Sinusoid and atherosclerosis in ageing. Lancet 2002;359:1612–1615. development and morphogenesis may be stimulated by VEGF-Flk-1 signaling [43] Arias IM. The biology of hepatic endothelial cell fenestrae. Prog Liver Dis during fetal mouse liver development. Dev Dyn 2010;239:386–397. 1990;9:11–26. [18] Zhang H, Pu W, Tian X, Huang X, He L, Liu Q, et al. Genetic lineage tracing [44] Yokomori H, Oda M, Ogi M, Kamegaya Y, Tsukada N, Ishii H. Endothelial nitric identifies endocardial origin of liver vasculature. Nat Genet oxide synthase and caveolin-1 are co-localized in sinusoidal endothelial 2016;48:537–543. fenestrae. Liver 2001;21:198–206. [19] Davies PF, Civelek M, Fang Y, Fleming I. The atherosusceptible endothelium: [45] Warren A, Cogger VC, Arias IM, McCuskey RS, Le Couteur DG. Liver sinusoidal endothelial phenotypes in complex haemodynamic shear stress regions endothelial fenestrations in caveolin-1 knockout mice. Microcirculation in vivo. Cardiovasc Res 2013;99:315–327. 1994;2010:32–38. [20] DeLeve LD. Liver sinusoidal endothelial cells and liver regeneration. J Clin [46] Svistounov D, Warren A, McNerney GP, Owen DM, Zencak D, Zykova SN, Invest 2013;123:1861–1866. et al. The Relationship between fenestrations, sieve plates and rafts in liver [21] Margreet De Leeuw A, Brouwer A, Knook DL. Sinusoidal endothelial cells of sinusoidal endothelial cells. PLoS One 2012;7 e46134. the liver: Fine structure and function in relation to age. J Electron Microsc [47] Rodriguez-Vita J, Morales-Ruiz M. Down the liver sinusoidal endothelial cell Tech 1990;14:218–236. (LSEC) hole. Is there a role for lipid rafts in LSEC fenestration? Hepatology [22] Wang L, Wang X, Xie G, Wang L, Hill CK, DeLeve LD. Liver sinusoidal 2013;57:1272–1274. endothelial cell progenitor cells promote liver regeneration in rats. J Clin [48] Géraud C, Evdokimov K, Straub BK, Peitsch WK, Demory A, Dörflinger Y, Invest 2012;122:1567–1573. et al. Unique cell type-specific junctional complexes in vascular endothelium [23] Harb R, Xie G, Lutzko C, Guo Y, Wang X, Hill CK, et al. Bone marrow of human and rat liver sinusoids. PLoS One 2012;7. progenitor cells repair rat hepatic sinusoidal endothelial cells after liver [49] Lautt WW. Hepatic circulation: physiology and pathophysiology. San Rafael, injury. Gastroenterology 2009;137:704–712. CA: Morgan & Claypool Life Sciences; 2009. [24] DeLeve LD, Garcia-Tsao G. Vascular liver disease – mechanisms and [50] Rodés J, Benhamou J-P, Blei A, Reichen J, Rizzetto M. Textbook of management. Springer Science+Business media, LLC; 2011. http://dx.doi. hepatology: from basic science to clinical practice. 3rd ed. Blackwell org/10.1007/978-1-4419-8327-5_2. Publishing; 2007. [25] Lee SS, Hadengue A, Moreau R, Sayegh R, Hillon P, Lebrec D. Postprandial [51] Smedsrød B, Le Couteur D, Ikejima K, Jaeschke H, Kawada N, Naito M, et al. hemodynamic responses in patients with cirrhosis. Hepatology Hepatic sinusoidal cells in health and disease: update from the 14th 1988;8:647–651. International Symposium: Hepatic sinusoidal cells in health and disease. [26] Davies PF. Flow-mediated endothelial mechanotransduction. Physiol Rev Liver Int 2009;29:490–501. 1995;75:519–560. [52] Sorensen KK, McCourt P, Berg T, Crossley C, Couteur DL, Wake K, et al. The [27] Gracia-Sancho J, Russo L, García-Calderó H, García-Pagán JC, García-Cardeña scavenger endothelial cell: a new player in homeostasis and immunity. Am J G, Bosch J. Endothelial expression of transcription factor Kruppel-like factor 2 Physiol Regul Integr Comp Physiol 2012;303:R1217–R1230. and its vasoprotective target genes in the normal and cirrhotic rat liver. Gut [53] Schledzewski K, Géraud C, Arnold B, Wang S, Gröne H-J, Kempf T, et al. 2011;60:517–524. Deficiency of liver sinusoidal scavenger receptors stabilin-1 and -2 in mice Review [28] Parmar KM, Larman HB, Dai G, Zhang Y, Wang ET, Moorthy SN, et al. causes glomerulofibrotic nephropathy via impaired hepatic clearance of Integration of flow-dependent endothelial phenotypes by Kruppel-like factor noxious blood factors. J Clin Invest 2011;121:703–714. 2. J Clin Invest 2006;116:49–58. [54] Elvevold K, Simon-Santamaria J, Hasvold H, McCourt P, Smedsrød B, Sørensen [29] Shah V, Haddad FG, Garcia-Cardena G, Frangos JA, Mennone A, Groszmann RJ, KK. Liver sinusoidal endothelial cells depend on mannose receptor-mediated et al. Liver sinusoidal endothelial cells are responsible for nitric oxide recruitment of lysosomal enzymes for normal degradation capacity. Hepa- modulation of resistance in the hepatic sinusoids. J Clin Invest tology 2008;48:2007–2015. 1997;100:2923. [55] Elvevold K, Smedsrød B, Martinez I. The liver sinusoidal endothelial cell: a [30] Fernandez M. Molecular pathophysiology of portal hypertension. Hepatology cell type of controversial and confusing identity. Am J Physiol Gastrointest 2015;61:1406–1415. Liver Physiol 2008;294:G391–G400. [31] Kawada N, Tran-Thi T-A, Klein H, Decker K. The contraction of hepatic stellate [56] Lalor PF, Lai WK, Curbishley SM, Shetty S, Adams DH. Human hepatic (Ito) cells stimulated with vasoactive substances. Eur J Biochem sinusoidal endothelial cells can be distinguished by expression of phenotypic 1993;213:815–823. markers related to their specialised functions in vivo. World J Gastroenterol [32] McCuskey RS. Morphological mechanisms for regulating blood flow through 2006;12:5429–5439. hepatic sinusoids. Liver 2000;20:3–7. [57] Ding B-S, Nolan DJ, Butler JM, James D, Babazadeh AO, Rosenwaks Z, et al. [33] Rockey D. The cellular pathogenesis of portal hypertension: stellate cell Inductive angiocrine signals from sinusoidal endothelium are required for contractility, endothelin, and nitric oxide. Hepatology 1997;25:2–5. liver regeneration. Nature 2010;468:310–315. [34] DeLeve LD, Wang X, Guo Y. Sinusoidal endothelial cells prevent rat stellate [58] Woodfin A, Voisin M-B, Nourshargh S. PECAM-1: a multi-functional molecule cell activation and promote reversion to quiescence. Hepatology in inflammation and vascular biology. Arterioscler Thromb Vasc Biol 2008;48:920–930. 2007;27:2514–2523. [35] Braet F, Wisse E. Structural and functional aspects of liver sinusoidal [59] DeLeve LD, Wang X, Hu L, McCuskey MK, McCuskey RS. Rat liver endothelial cell fenestrae: a review. Comp Hepatol 2002;1:1. sinusoidal endothelial cell phenotype is maintained by paracrine and [36] DeLeve LD. Liver sinusoidal endothelial cells in hepatic fibrosis. Hepatology autocrine regulation. Am J Physiol Gastrointest Liver Physiol 2004;287: 2015;61:1740–1746. G757–G763. [37] Wisse E, De Zanger RB, Charels K, Van Der Smissen P, McCuskey RS. The liver [60] Fan CQ, Crawford JM. Sinusoidal obstruction syndrome (hepatic veno- sieve: considerations concerning the structure and function of endothelial occlusive disease). J Clin Exp Hepatol 2014;4:332–346. fenestrae, the sinusoidal wall and the space of Disse. Hepatology [61] Smedsrød B. Protocol for preparation of mouse liver Kupffer cells and liver 1985;5:683–692. sinusoidal endothelial cells, 2012.

224 Journal of Hepatology 2017 vol. 66 j 212–227 JOURNAL OF HEPATOLOGY

[62] Meyer J, Gonelle-Gispert C, Morel P, Bühler L. Methods for isolation and regression of hepatic fibrosis in rats. Gastroenterology 2012;142:918–927 purification of murine liver sinusoidal endothelial cells: a systematic review. e6. PLoS One 2016;11 e0151945. [86] D’Ambrosio R, Aghemo A, Rumi MG, Ronchi G, Donato MF, Paradis V, et al. A [63] March S, Hui EE, Underhill GH, Khetani S, Bhatia SN. Microenvironmental morphometric and immunohistochemical study to assess the benefit of a regulation of the sinusoidal endothelial cell phenotype in vitro. Hepatology sustained virological response in hepatitis C virus patients with cirrhosis. 2009;50:920–928. Hepatology 2012;56:532–543. [64] Yokomori H, Oda M, Yoshimura K, Nagai T, Ogi M, Nomura M, et al. Vascular [87] Carmeliet P, Jain RK. Molecular mechanisms and clinical applications of endothelial growth factor increases fenestral permeability in hepatic sinu- angiogenesis. Nature 2011;473:298–307. soidal endothelial cells. Liver Int 2003;23:467–475. [88] Ehling J, Bartneck M, Wei X, Gremse F, Fech V, Möckel D, et al. CCL2- [65] Krause P, Markus PM, Schwartz P, Unthan-Fechner K, Pestel S, Fandrey J, et al. dependent infiltrating macrophages promote angiogenesis in progressive Hepatocyte-supported serum-free culture of rat liver sinusoidal endothelial liver fibrosis. Gut 2014;63:1960–1971. cells. J Hepatol 2000;32:718–726. [89] Thabut D, Shah V. Intrahepatic angiogenesis and sinusoidal remodeling in [66] Martinez I, Nedredal GI, Øie CI, Warren A, Johansen O, Le Couteur DG, et al. chronic liver disease: New targets for the treatment of portal hypertension? J The influence of oxygen tension on the structure and function of isolated Hepatol 2010;53:976–980. liver sinusoidal endothelial cells. Comp Hepatol 2008;7:4. [90] Taura K, De Minicis S, Seki E, Hatano E, Iwaisako K, Osterreicher CH, et al. [67] Motoyama S, Minamiya Y, Saito S, Saito R, Matsuzaki I, Abo S, et al. Hydrogen Hepatic stellate cells secrete angiopoietin 1 that induces angiogenesis in liver peroxide derived from hepatocytes induces sinusoidal endothelial fibrosis. Gastroenterology 2008;135:1729–1738. cell apoptosis in perfused hypoxic rat liver. Gastroenterology 1998;114: [91] Thabut D, Routray C, Lomberk G, Shergill U, Glaser K, Huebert R, et al. 153–163. Complementary vascular and matrix regulatory pathways underlie the [68] Matsumura T, Takesue M, Westerman KA, Okitsu T, Sakaguchi M, Fukazawa beneficial mechanism of action of sorafenib in liver fibrosis. Hepatology T, et al. Establishment of an immortalized human-liver endothelial cell line 2011;54:573–585. with SV40T and hTERT. Transplantation 2004;77:1357–1365. [92] Tarantino G, Conca P, Pasanisi F, Ariello M, Mastrolia M, Arena A, et al. Could [69] Hering S, Griffin BE, Strauss M. Immortalization of human fetal sinusoidal inflammatory markers help diagnose nonalcoholic steatohepatitis? Eur J liver cells by polyoma virus large T antigen. Exp Cell Res 1991;195:1–7. Gastroenterol Hepatol 2009;21:504–511. [70] Maru Y, Yamaguchi S, Takahashi T, Ueno H, Shibuya M. Virally activated Ras [93] Coulon S, Francque S, Colle I, Verrijken A, Blomme B, Heindryckx F, et al. cooperates with integrin to induce tubulogenesis in sinusoidal endothelial Evaluation of inflammatory and angiogenic factors in patients with non- cell lines. J Cell Physiol 1998;176:223–234. alcoholic fatty liver disease. Cytokine 2012;59:442–449. [71] Saito M, Matsuura T, Masaki T, Maehashi H, Braet F. Study of the [94] Cayón A, Crespo J, Guerra AR, Pons-Romero F. Gene expression in obese reappearance of sieve plate-like pores in immortalized sinusoidal endothelial patients with non-alcoholic steatohepatitis. Rev Esp Enferm Dig cells - Effect of actin inhibitor in mixed perfusion cultures. Comp Hepatol 2008;100:212–218. 2004;3:S28. [95] Sahin H, Borkham-Kamphorst E, Kuppe C, Zaldivar MM, Grouls C, Al-samman [72] Matsuura T, Kawada M, Hasumura S, Nagamori S, Obata T, Yamaguchi M, M, et al. Chemokine Cxcl9 attenuates liver fibrosis-associated angiogenesis in et al. High density culture of immortalized liver endothelial cells in the mice. Hepatology 2012;55:1610–1619. radial-flow bioreactor in the development of an artificial liver. Int J Artif [96] Kaur S, Tripathi D, Dongre K, Garg V, Rooge S, Mukopadhyay A, et al. Organs 1998;21:229–234. Increased number and function of endothelial progenitor cells stimulate [73] Seki E, De Minicis S, Osterreicher CH, Kluwe J, Osawa Y, Brenner DA, et al. angiogenesis by resident liver sinusoidal endothelial cells (SECs) in cirrhosis TLR4 enhances TGF-beta signaling and hepatic fibrosis. Nat Med through paracrine factors. J Hepatol 2012;57:1193–1198. 2007;13:1324–1332. [97] Rautou P-E. Endothelial progenitor cells in cirrhosis: the more, the merrier? J [74] Kennedy DW, Abkowitz JL. Kinetics of central nervous system microglial and Hepatol 2012;57:1163–1165. macrophage engraftment: analysis using a transgenic bone marrow trans- [98] Ding B-S, Cao Z, Lis R, Nolan DJ, Guo P, Simons M, et al. Divergent angiocrine plantation model. Blood 1997;90:986–993. signals from vascular niche balance liver regeneration and fibrosis. Nature [75] Schaffner F, Poper H. Capillarization of hepatic sinusoids in man. Gastroen- 2013;505:97–102. terology 1963;44:239–242. [99] Francque S, Laleman W, Verbeke L, Van Steenkiste C, Casteleyn C, Kwanten [76] Horn T, Christoffersen P, Henriksen JH. Alcoholic liver injury: defenestration W, et al. Increased intrahepatic resistance in severe steatosis: endothelial in noncirrhotic livers–a scanning electron microscopic study. Hepatology dysfunction, vasoconstrictor overproduction and altered microvascular 1987;7:77–82. architecture. Lab Invest 2012;92:1428–1439. Review [77] Xu B, Broome U, Uzunel M, Nava S, Ge X, Kumagai-Braesch M, et al. [100] Pasarín M, La Mura V, Gracia-Sancho J, García-Calderó H, Rodríguez- Capillarization of hepatic sinusoid by liver endothelial cell-reactive autoan- Vilarrupla A, García-Pagán JC, et al. Sinusoidal endothelial dysfunction tibodies in patients with cirrhosis and chronic hepatitis. Am J Pathol precedes inflammation and fibrosis in a model of NAFLD. PLoS One 2012;7 2003;163:1275–1289. e32785. [78] Martinez-Hernandez A, Martinez J. The role of capillarization in hepatic [101] Marrone G, Russo L, Rosado E, Hide D, García-Cardeña G, García-Pagán JC, failure: studies in carbon tetrachloride-induced cirrhosis. Hepatology et al. The transcription factor KLF2 mediates hepatic endothelial protection 1991;14:864–874. and paracrine endothelial-stellate cell deactivation induced by statins. J [79] Mori T, Okanoue T, Sawa Y, Hori N, Ohta M, Kagawa K. Defenestration of the Hepatol 2013;58:98–103. sinusoidal endothelial cell in a rat model of cirrhosis. Hepatology [102] Marrone G, Maeso-Díaz R, García-Cardena G, Abraldes JG, García-Pagán JC, 1993;17:891–897. Bosch J, et al. KLF2 exerts antifibrotic and vasoprotective effects in cirrhotic [80] Warren A, Bertolino P, Benseler V, Fraser R, McCaughan GW, Le Couteur DG. rat livers: behind the molecular mechanisms of statins. Gut 2014. http://dx. Marked changes of the hepatic sinusoid in a transgenic mouse model of acute doi.org/10.1136/gutjnl-2014-308338. immune-mediated hepatitis. J Hepatol 2007;46:239–246. [103] Abraldes JG, Albillos A, Bañares R, Turnes J, González R, García-Pagán JC, et al. [81] DeLeve LD, Wang X, Kanel GC, Atkinson RD, McCuskey RS. Prevention of Simvastatin lowers portal pressure in patients with cirrhosis and portal hepatic fibrosis in a murine model of metabolic syndrome with nonalcoholic hypertension: a randomized controlled trial. Gastroenterology steatohepatitis. Am J Pathol 2008;173:993–1001. 2009;136:1651–1658. [82] Miyao M, Kotani H, Ishida T, Kawai C, Manabe S, Abiru H, et al. Pivotal role of [104] Abraldes JG, Villanueva C, Aracil C, Turnes J, Hernandez-Guerra M, Genesca liver sinusoidal endothelial cells in NAFLD/NASH progression. Lab Invest J, et al. Addition of simvastatin to standard therapy for the prevention of 2015;95:1130–1144. variceal rebleeding does not reduce rebleeding but increases survival in [83] Marrone G, Shah VH, Gracia-Sancho J. Sinusoidal communication in liver patients with cirrhosis. Gastroenterology 2016. http://dx.doi.org/10.1053/ fibrosis and regeneration. J Hepatol 2016;65:608–617. http://dx.doi.org/ j.gastro.2016.01.004. 10.1016/j.jhep.2016.04.018. [105] Sakamoto M. Early HCC: diagnosis and molecular markers. J Gastroenterol [84] García-Pagán J-C, Gracia-Sancho J, Bosch J. Functional aspects on the 2009;44:108–111. pathophysiology of portal hypertension in cirrhosis. J Hepatol [106] Semela D, Dufour J-F. Angiogenesis and hepatocellular carcinoma. J Hepatol 2012;57:458–461. 2004;41:864–880. [85] Xie G, Wang X, Wang L, Wang L, Atkinson RD, Kanel GC, et al. Role of [107] Géraud C, Mogler C, Runge A, Evdokimov K, Lu S, Schledzewski K, et al. differentiation of liver sinusoidal endothelial cells in progression and Endothelial transdifferentiation in hepatocellular carcinoma: loss of Stabilin-

Journal of Hepatology 2017 vol. 66 j 212–227 225 Review

2 expression in peri-tumourous liver correlates with increased survival. Liver [130] Peralta C, Jiménez-Castro MB, Gracia-Sancho J. Hepatic ischemia and Int 2013;33:1428–1440. reperfusion injury: effects on the liver sinusoidal milieu. J Hepatol [108] Wu LQ, Zhang WJ, Niu JX, Ye LY, Yang ZH, Grau GE, et al. Phenotypic and 2013;59:1094–1106. functional differences between human liver cancer endothelial cells and liver [131] Lalor PF, Shields P, Grant A, Adams DH. Recruitment of lymphocytes to the sinusoidal endothelial cells. J Vasc Res 2008;45:78–86. human liver. Immunol Cell Biol 2002;80:52–64. [109] Zhuang P-Y, Wang J-D, Tang Z-H, Zhou X-P, Quan Z-W, Liu Y-B, et al. Higher [132] Shetty S, Weston CJ, Oo YH, Westerlund N, Stamataki Z, Youster J, et al. proliferation of peritumoral endothelial cells to IL-6/sIL-6R than Common lymphatic endothelial and vascular endothelial receptor-1 medi- tumoral endothelial cells in hepatocellular carcinoma. BMC Cancer ates the transmigration of regulatory T cells across human hepatic sinusoidal 2015;15:830. endothelium. J Immunol 1950;2011:4147–4155. [110] Hoshida Y, Villanueva A, Kobayashi M, Peix J, Chiang DY, Camargo A, et al. [133] Limmer A, Ohl J, Kurts C, Ljunggren HG, Reiss Y, Groettrup M, et al. Efficient Gene expression in fixed tissues and outcome in hepatocellular carcinoma. N presentation of exogenous antigen by liver endothelial cells to CD8+ T cells Engl J Med 2008;359:1995–2004. results in antigen-specific T-cell tolerance. Nat Med 2000;6:1348–1354. [111] Hu J, Srivastava K, Wieland M, Runge A, Mogler C, Besemfelder E, et al. [134] Carambia A, Freund B, Schwinge D, Heine M, Laschtowitz A, Huber S, et al. Endothelial cell-derived angiopoietin-2 controls liver regeneration as a TGF-b-dependent induction of CD4+CD25+Foxp3+ Tregs by liver sinusoidal spatiotemporal rheostat. Science 2014;343:416–419. endothelial cells. J Hepatol 2014;61:594–599. [112] Golse N, Bucur PO, Adam R, Castaing D, Sa Cunha A, Vibert E. New paradigms [135] Carambia A, Freund B, Schwinge D, Bruns OT, Salmen SC, Ittrich H, et al. in post-hepatectomy liver failure. J Gastrointest Surg 2013;17:593–605. Nanoparticle-based autoantigen delivery to Treg-inducing liver sinusoidal [113] Yamanaka K, Hatano E, Narita M, Kitamura K, Yanagida A, Asechi H, et al. endothelial cells enables control of autoimmunity in mice. J Hepatol Olprinone attenuates excessive shear stress through up-regulation of 2015;62:1349–1356. endothelial nitric oxide synthase in a rat excessive hepatectomy model. [136] Neumann K, Rudolph C, Neumann C, Janke M, Amsen D, Scheffold A. Liver Liver Transpl 2011;17:60–69. sinusoidal endothelial cells induce immunosuppressive IL-10-producing Th1 [114] Schoen JM, Wang HH, Minuk GY, Lautt WW. Shear stress-induced nitric cells via the Notch pathway. Eur J Immunol 2015;45:2008–2016. oxide release triggers the liver regeneration cascade. Nitric Oxide [137] Breiner KM, Schaller H, Knolle PA. Endothelial cell-mediated uptake of a 2001;5:453–464. hepatitis B virus: a new concept of liver targeting of hepatotropic microor- [115] Sugawara Y, Yamamoto J, Shimada K, Yamasaki S, Kosuge T, Takayama T, ganisms. Hepatology 2001;34:803–808. et al. Splenectomy in patients with hepatocellular carcinoma and hyper- [138] Cormier EG, Durso RJ, Tsamis F, Boussemart L, Manix C, Olson WC, et al. L- splenism. J Am Coll Surg 2000;190:446–450. SIGN (CD209L) and DC-SIGN (CD209) mediate transinfection of liver cells by [116] Sato Y, Kobayashi T, Nakatsuka H, Yamamoto S, Oya H, Watanabe T, et al. hepatitis C virus. Proc Natl Acad Sci U S A 2004;101:14067–14072. Splenic arterial ligation prevents liver injury after a major hepatectomy by a [139] Liu B, Wang M, Wang X, Zhao D, Liu D, Liu J, et al. Liver sinusoidal endothelial reduction of surplus portal hypertension in hepatocellular carcinoma cell lectin inhibits CTL-dependent virus clearance in mouse models of viral patients with cirrhosis. Hepatogastroenterology 2001;48:831–835. hepatitis. J Immunol 1950;2013:4185–4195. [117] Troisi R, Cammu G, Militerno G, De Baerdemaeker L, Decruyenaere J, Hoste E, [140] Bruns T, Zimmermann HW, Pachnio A, Li K-K, Trivedi PJ, Reynolds G, et al. et al. Modulation of portal graft inflow: a necessity in adult living-donor liver CMV infection of human sinusoidal endothelium regulates hepatic T cell transplantation? Ann Surg 2003;237:429–436. recruitment and activation. J Hepatol 2015;63:38–49. [118] Ito K, Ozasa H, Horikawa S. Effects of prior splenectomy on remnant liver [141] Plessier A, Darwish-Murad S, Hernandez-Guerra M, Consigny Y, Fabris F, after partial hepatectomy with Pringle maneuver in rats. Liver Int Trebicka J, et al. Acute portal vein thrombosis unrelated to cirrhosis: a 2005;25:438–444. prospective multicenter follow-up study. Hepatology 2010;51:210–218. [119] Ren Y-S, Qian N-S, Tang Y, Liao Y-H, Liu W-H, Raut V, et al. Beneficial effects [142] Leong SS, Cazen RA, Yu GS, LeFevre L, Carson JW. Abdominal visceral peliosis of splenectomy on liver regeneration in a rat model of massive hepatectomy. associated with bacillary angiomatosis. Ultrastructural evidence of endothe- Hepatobiliary Pancreat Dis Int 2012;11:60–65. lial destruction by bacilli. Arch Pathol Lab Med 1992;116:866–871. [120] Mogl MT, Nüssler NC, Presser SJ, Podrabsky P, Denecke T, Grieser C, et al. [143] Malamut G, Ziol M, Suarez F, Beaugrand M, Viallard JF, Lascaux AS, et al. Evolving experience with prevention and treatment of splenic artery Nodular regenerative hyperplasia: the main liver disease in patients with syndrome after orthotopic liver transplantation. Transpl Int primary hypogammaglobulinemia and hepatic abnormalities. J Hepatol 2010;23:831–841. 2008;48:74–82. [121] Yoshizumi T, Taketomi A, Soejima Y, Ikegami T, Uchiyama H, Kayashima H, [144] Brunt EM, Gouw ASH, Hubscher SG, Tiniakos DG, Bedossa P, Burt AD, et al. et al. The beneficial role of simultaneous splenectomy in living donor liver Pathology of the liver sinusoids. Histopathology 2014;64:907–920. Review transplantation in patients with small-for-size graft. Transpl Int [145] Parent R, Durantel D, Lahlali T, Sallé A, Plissonnier M-L, DaCosta D, et al. An 2008;21:833–842. immortalized human liver endothelial sinusoidal cell line for the study of the [122] Yamanaka K, Hatano E, Iguchi K, Yamamoto G, Sato M, Toriguchi K, et al. pathobiology of the liver endothelium. Biochem Biophys Res Commun Effect of olprinone on liver microstructure in rat partial liver transplantation. 2014;450:7–12. J Surg Res 2013;183:391–396. [146] Cogger VC, Arias IM, Warren A, McMahon AC, Kiss DL, Avery VM, et al. The [123] Iguchi K, Hatano E, Yamanaka K, Sato M, Yamamoto G, Kasai Y, et al. response of fenestrations, actin, and caveolin-1 to vascular endothelial Hepatoprotective effect by pretreatment with olprinone in a swine partial growth factor in SK Hep1 cells. Am J Physiol Gastrointest Liver Physiol hepatectomy model. Liver Transpl 2014;20:838–849. 2008;295:G137–G145. [124] LeCouter J, Moritz DR, Li B, Phillips GL, Liang XH, Gerber H-P, et al. [147] Zhao X, Zhao Q, Luo Z, Yu Y, Xiao N, Sun X, et al. Spontaneous immortal- Endothelial protection of liver: role of VEGFR-1. Cancer Cell 2002;1:229. ization of mouse liver sinusoidal endothelial cells. Int J Mol Med [125] Meyer J, Lejmi E, Fontana P, Morel P, Gonelle-Gispert C, Bühler L. A focus on 2015;35:617–624. the role of platelets in liver regeneration: Do platelet-endothelial cell [148] Huebert RC, Jagavelu K, Liebl AF, Huang BQ, Splinter PL, LaRusso NF, et al. interactions initiate the regenerative process? J Hepatol 2015;63: Immortalized liver endothelial cells: a cell culture model for studies of 1263–1271. motility and angiogenesis. Lab Invest 2010;90:1770–1781. [126] Myronovych A, Murata S, Chiba M, Matsuo R, Ikeda O, Watanabe M, et al. [149] Terry RW, Kwee L, Baldwin HS, Labow MA. Cre-mediated generation of a Role of platelets on liver regeneration after 90% hepatectomy in mice. J VCAM-1 null allele in transgenic mice. Transgenic Res 1997;6:349–356. Hepatol 2008;49:363–372. [150] Gustafsson E, Brakebusch C, Hietanen K, Fässler R. Tie-1-directed expression [127] Murata S, Hashimoto I, Nakano Y, Myronovych A, Watanabe M, Ohkohchi N. of Cre recombinase in endothelial cells of embryoid bodies and transgenic Single administration of thrombopoietin prevents progression of liver mice. J Cell Sci 2001;114:671–676. fibrosis and promotes liver regeneration after partial hepatectomy in [151] Kisanuki YY, Hammer RE, Miyazaki J, Williams SC, Richardson JA, Yanagi- cirrhotic rats. Ann Surg 2008;248:821–828. sawa M. Tie2-Cre transgenic mice: a new model for endothelial cell-lineage [128] Matsuo R, Nakano Y, Ohkohchi N. Platelet administration via the portal vein analysis in vivo. Dev Biol 2001;230:230–242. promotes liver regeneration in rats after 70% hepatectomy. Ann Surg [152] Licht AH, Raab S, Hofmann U, Breier G. Endothelium-specific Cre recombi- 2011;253:759–763. nase activity in flk-1-Cre transgenic mice. Dev Dyn 2004;229:312–318. [129] Melgar-Lesmes P, Edelman ER. Monocyte-endothelial cell interactions in the [153] Forde A, Constien R, Gröne H-J, Hämmerling G, Arnold B. Temporal Cre- regulation of vascular sprouting and liver regeneration in mouse. J Hepatol mediated recombination exclusively in endothelial cells using Tie2 regula- 2015;63:917–925. tory elements. Genes 2000;2002:191–197.

226 Journal of Hepatology 2017 vol. 66 j 212–227 JOURNAL OF HEPATOLOGY

[154] Göthert JR, Gustin SE, van Eekelen JAM, Schmidt U, Hall MA, Jane SM, et al. [164] Tanoi T, Tamura T, Sano N, Nakayama K, Fukunaga K, Zheng Y-W, et al. Genetically tagging endothelial cells in vivo: bone marrow-derived cells do Protecting liver sinusoidal endothelial cells suppresses apoptosis in acute not contribute to tumor endothelium. Blood 2004;104:1769–1777. liver damage. Hepatol Res 2015. http://dx.doi.org/10.1111/hepr.12607. [155] Monvoisin A, Alva JA, Hofmann JJ, Zovein AC, Lane TF, Iruela-Arispe ML. VE- [165] Akhter A, Hayashi Y, Sakurai Y, Ohga N, Hida K, Harashima H. Ligand density cadherin-CreERT2 transgenic mouse: a model for inducible recombination in at the surface of a nanoparticle and different uptake mechanism: two the endothelium. Dev Dyn 2006;235:3413–3422. important factors for successful siRNA delivery to liver endothelial cells. Int J [156] Wang Y, Nakayama M, Pitulescu ME, Schmidt TS, Bochenek ML, Sakakibara A, Pharm 2014;475:227–237. et al. Ephrin-B2 controls VEGF-induced angiogenesis and lymphangiogene- [166] Bartsch M, Weeke-Klimp AH, Meijer DKF, Scherphof GL, Kamps JAAM. sis. Nature 2010;465:483–486. Massive and selective delivery of lipid-coated cationic lipoplexes of [157] Claxton S, Kostourou V, Jadeja S, Chambon P, Hodivala-Dilke K, Fruttiger M. oligonucleotides targeted in vivo to hepatic endothelial cells. Pharm Res Efficient, inducible Cre-recombinase activation in vascular endothelium. 2002;19:676–680. Genes 2000;2008:74–80. [167] Kamps JA, Morselt HW, Swart PJ, Meijer DK, Scherphof GL. Massive targeting [158] Ehling M, Adams S, Benedito R, Adams RH. Notch controls retinal blood of liposomes, surface-modified with anionized albumins, to hepatic endothe- vessel maturation and quiescence. Development 2013;140:3051–3061. lial cells. Proc Natl Acad Sci U S A 1997;94:11681–11685. [159] Sano N, Tamura T, Toriyabe N, Nowatari T, Nakayama K, Tanoi T, et al. New [168] Bartsch M, Weeke-Klimp AH, Morselt HWM, Kimpfler A, Asgeirsdóttir SA, drug delivery system for liver sinusoidal endothelial cells for ischemia- Schubert R, et al. Optimized targeting of polyethylene glycol-stabilized anti- reperfusion injury. World J Gastroenterol 2015;21:12778–12786. intercellular adhesion molecule 1 oligonucleotide/lipid particles to liver [160] Fraser JR, Alcorn D, Laurent TC, Robinson AD, Ryan GB. Uptake of circulating sinusoidal endothelial cells. Mol Pharmacol 2005;67:883–890. hyaluronic acid by the rat liver. Cellular localization in situ. Cell Tissue Res [169] Sato Y, Hatakeyama H, Sakurai Y, Hyodo M, Akita H, Harashima H. A pH- 1985;242:505–510. sensitive cationic lipid facilitates the delivery of liposomal siRNA and gene [161] Toriyabe N, Hayashi Y, Hyodo M, Harashima H. Synthesis and evaluation of silencing activity in vitro and in vivo. J Control Release 2012;163:267–276. stearylated hyaluronic acid for the active delivery of liposomes to liver [170] Enzan H, Himeno H, Hiroi M, Kiyoku H, Saibara T, Onishi S. Development of endothelial cells. Biol Pharm Bull 2011;34:1084–1089. hepatic sinusoidal structure with special reference to the Ito cells. Microsc [162] Takei Y, Maruyama A, Ferdous A, Nishimura Y, Kawano S, Ikejima K, et al. Res Tech 1997;39:336–349. Targeted gene delivery to sinusoidal endothelial cells: DNA nanoassociate [171] Herrnberger L, Hennig R, Kremer W, Hellerbrand C, Goepferich A, Kalbitzer bearing hyaluronan-glycocalyx. FASEB J 2004;18:699–701. HR, et al. Formation of fenestrae in murine liver sinusoids depends on [163] Kren BT, Unger GM, Sjeklocha L, Trossen AA, Korman V, Diethelm-Okita BM, plasmalemma vesicle-associated protein and is required for lipoprotein et al. Nanocapsule-delivered Sleeping Beauty mediates therapeutic Factor passage. PLoS One 2014;9 e115005. VIII expression in liver sinusoidal endothelial cells of hemophilia A mice. J Clin Invest 2009;119:2086–2099. Review

Journal of Hepatology 2017 vol. 66 j 212–227 227

B. Appendix 2: Reply to “Calreticulin mutations and their importance in Budd-Chiari syndrome”

Reply to: ‘‘Calreticulin mutations and their importance in Budd-Chiari syndrome”

Johanne Poisson1, Fanny Turon2, Christophe Marzac3,4, Dominique-Charles Valla5,6,7

Juan-Carlos Garcia-Pagan2,8, Pierre-Emmanuel Rautou1,5,7

1 Inserm, U970, Paris Cardiovascular Research Center – PARCC, Université Paris Descartes,

Sorbonne Paris Cité, Paris, France

2 Barcelona Hepatic Hemodynamic Laboratory, Liver Unit, Hospital Clínic, IDIBAPS, Barcelona,

Spain

3 UPMC, Univ Paris 06, GRC n_7, Groupe de Recherche Clinique sur les Myéloproliférations

Aiguës et Chroniques MYPAC, Paris, France

4 Laboratoire d’Hématologie, Département de Biologie et Pathologie Médicales, Institut

Gustave Roussy, Villejuif, France

5 DHU Unity, Pôle des Maladies de l’Appareil Digestif, Service d’Hépatologie, Centre de

Référence des Maladies Vasculaires du Foie, Hôpital Beaujon, AP-HP, Clichy, France

6 Inserm U1149, Centre de Recherche sur l’Inflammation (CRI), Paris, Université Paris 7-Denis-

Diderot, Clichy, UFR de Médecine, Paris, France

7 Université Paris Diderot, Sorbonne Paris cité, Paris, France

8 Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas

(CIBERehd), Spain

Letter published in J Hepatol. 2017 Nov;67(5):1112-1113.

187 JOURNAL OF HEPATOLOGY

Supplementary data [7] Chen QY, Wang XY, Harrison TJ, He X, Hu LP, Li KW, et al. HBsAg may reappear following reactivation in individuals with spontaneous HBsAg seroclearance 8 years previously. Epidemiol Infect 2017;145:728–738. Supplementary data associated with this article can be found, in [8] AASLD/IDSA HCV Guidelines. People with Hepatitis C Should Be Tested for the online version, at http://dx.doi.org/10.1016/j.jhep.2017.07. Hepatitis B Before Starting Antiviral Therapies. September 16 2016 cited; 014. Available from: http://www.aasld.org/about-aasld/pressroom/people-hepatitis-c-should-be-tested-hepatitis-b-starting-antiviral-therapies. [9] FDA Drug Safety Communication: FDA warns about the risk of hepatitis B References reactivating in some patients treated with direct-acting antivirals for hepatitis C. October 4 2016 cited; Available from: http://www.fda.gov/ [1] Konstantinou D, Deutsch M. The spectrum of HBV/HCV coinfection: Drugs/DrugSafety/ucm522932.htm. epidemiology, clinical characteristics, viral interactions and management. [10] European Association for the Study of the Liver. EASL recommendations on Ann Gastroenterol 2015;28:221–228. treatment of hepatitis C. J Hepatol 2017;66:153–194. [2] Yu ML, Lee CM, Chen CL, Chuang WL, Lu SN, Liu CH, et al. Sustained hepatitis C virus clearance and increased hepatitis B surface antigen seroclearance in Tsuyoshi Suda patients with dual chronic hepatitis C and B during posttreatment follow-up. ⇑ Hepatology 2013;57:2135–2142. Tetsuro Shimakami [3] Belperio Pamela S, Shahoumian Troy A, Mole Larry A, Backus LI. Evaluation of Takayoshi Shirasaki hepatitis B reactivation among 62,920 veterans treated with oral hepatitis C Tatsuya Yamashita antivirals. Hepatology 2017. http://dx.doi.org/10.1002/hep.29135. Eishiro Mizukoshi [4] Yeh Ming Lun, Huang Chung Feng, Meng-HsuanHsieh Yu-Min Ko, Chen Ku- Yu, Liu Ta-Wei, et al. Reactivation of hepatitis B in patients of chronic Masao Honda hepatitis C with hepatitis B virus infection treated with direct acting Shuichi Kaneko antivirals. J Gastroenterol Hepatol 2017. http://dx.doi.org/10.1111/ Department of Gastroenterology, Kanazawa University Graduate jgh.13771. School of Medical Science, Kanazawa, [5] Wang C, Ji D, Chen J, Shao Q, Li B, Liu J, et al. Hepatitis due to reactivation of Ishikawa 920-8641, Japan hepatitis B virus in endemic areas among patients with hepatitis C treated ⇑ with direct-acting antiviral agents. Clin Gastroenterol Hepatol Corresponding author. Address: Department of Gastroenterology, 2017;15:132–136. Kanazawa University Hospital, 13-1 Takaramachi, Kanazawa, [6] Kamitsukasa H, Iri M, Tanaka A, Nagashima S, Takahashi M, Nishizawa T, Ishikawa 920-8641, Japan. et al. Spontaneous reactivation of hepatitis B virus (HBV) infection in Tel.: +81 76 265 2233; fax: +81 76 234 4250. patients with resolved or occult HBV infection. J Med Virol 2015;87:589–600. E-mail address: [email protected]

Calreticulin mutations and their importance in Budd-Chiari syndrome

To the Editor: yurea. At a follow up of 24 months, the patient is doing well with We have read your recent publications on Calreticulin (CALR) a patent stent. Another of our patients, a 30-year-old man with gene mutations in splanchnic vein thrombosis with great inter- type 3 CALR mutation (3-bp deletion) had an acute presentation est. Poisson et al.1 reported prevalence of myeloproliferative neo- of BCS with ascites. He had thrombosis of suprahepatic inferior plasms in a cohort of 99 patients with Budd-Chiari syndrome vena cava and all three hepatic veins and subsequently developed (BCS), out of which 28 had JAK2V617F gene mutations and one portal vein thrombosis. He had a history of deep vein thrombosis had a CALR gene mutation. The authors suggested that a combi- of the right leg. Baseline investigations revealed haemoglobin of nation of spleen height !16 cm and platelet count [200Â109/L 9 g/dl, total count of 5,000/mm3, platelet count of 63 Â 109/L be tested for CALR mutation, in the absence of JAK2 gene muta- and a spleen height of 11 cm. He has been treated with anti-coag- tions. Turon et al.2 reported 24 JAK2V617F gene mutations and ulation and hydroxyurea. two CALR gene mutations among 69 patients with BCS. We have Detection of CALR mutations may have significant impacts on summarized the data including three other studies in Table 1. disease severity, prognosis and treatment. A study by Klampfl We wish to highlight data on our patients with CALR related et al.3 showed that patients with CALR exon mutations have low BCS. Between 2011 and 2017, 38 of 210 (18.1%) patients were levels of haemoglobin and platelets, leading to decreased rates found to have a JAK2V617F mutation and two patients (0.9%) were of thrombosis. Andrikovics et al.4 reported a lower risk of throm- found to have CALR mutations (Table 1). CALR mutation testing bosis with CALR mutations compared to JAK2 mutations. Addi- was done using polymerase chain reaction-Sanger DNA sequenc- tionally, in primary myelofibrosis, patients with CALR mutations ing.3 One of our patients was a 39-year-old man with type 2 CALR have better survival rates compared to patients with JAK2 or mutation (5-bp insertion), presenting with chronic BCS with MPL mutations.3 Type 1 mutations (52-bp deletion) are seen in ascites. He had thrombosis of the inferior vena cava, right hepatic 56%, type 2 (5-bp insertion) in 28% and type 3 (others) in 16% vein and right inferior accessory hepatic vein. Baseline investiga- of patients with CALR mutations.5 Type 2-like CALR mutations tions revealed haemoglobin of 12.7 g/dl, total count of 16,210 / are mainly associated with an essential thrombocythemia (ET) mm3, a platelet count of 708Â109/L and spleen height of 16 cm. phenotype, low risk of thrombosis and indolent clinical course, He was treated with anticoagulation, an IVC stenting and hydrox- while type 1-like mutations are mainly associated with a

Journal of Hepatology 2017 vol. 67 j 1106–1121 1111 Letters to the Editor

Table 1. Summary of JAK2 and CALR gene mutations in BCS. Study Year of publication Country Total patients with BCS JAK2 gene mutation, n (%) CALR gene mutation, n (%) Haslam et al.7 2014 Ireland 9 4 (44.4) 0 Turon et al.2 2015 Spain 69 24 (34.7) 2 (2.9) Plompen et al.8 2015 Multicentric 70 19 (27.1) 0 (9 European countries) Zhang et al.9 2016 China 100 2 (2) 0 Poisson et al.1 2017 France 99 28 (28.2) 1 (1) Present report - India 210 38 (18.1) 2 (0.9) Total 557 115 (20.6) 5 (0.9)

BCS, Budd-Chiari syndrome. myelofibrosis phenotype and a high risk of progression from ET References to myelofibrosis.5 However, patients with CALR mutations and splanchnic vein thrombosis (SVT) are reported to have a signifi- [1] Poisson J, Plessier A, Kiladjian JJ, Turon F, Cassinat B, Andreoli A, et al. French cant disease burden.6 These patients may derive some benefit national network for vascular liver diseases. Selective testing for calreticulin 6 gene mutations in patients with splanchnic vein thrombosis: A prospective from JAK inhibitors like hydroxyurea and ruxolitinib. cohort study. J Hepatol 2017;67:501–507. We agree with the conclusion reached by the authors (Poisson [2] Turon F, Cervantes F, Colomer D, Baiges A, Hernández-Gea V, Garcia-Pagán JC. et al.1 and Turon et al.2), that CALR mutations must be considered Role of calreticulin mutations in the aetiological diagnosis of splanchnic vein for patients with SVT, without JAK2 mutation. However, one of thrombosis. J Hepatol 2015;62:72–74. [3] Klampfl T, Gisslinger H, Harutyunyan AS, Nivarthi H, Rumi E, Milosevic JD, ! our patients did not satisfy criteria of spleen height 16 cm et al. Somatic mutations of calreticulin in myeloproliferative neoplasms. N 9 and platelet count [200Â10 /L. The clinical features may vary Engl J Med 2013;369:2379–2390. with the type of CALR mutation. The prognosis also differs among [4] Andrikovics H, Krahling T, Balassa K, Halm G, Bors A, Koszarska M, et al. the different types of CALR mutation. CALR mutation is reported Distinct clinical characteristics of myeloproliferative neoplasms with calreti- in less than 1% of patients with BCS. There is a need to develop culin mutations. Haematologica 2014;99:184–190. 1 [5] Pietra D, Rumi E, Ferretti VV, et al. Differential clinical effects of different an approach, like that recommended by Poisson et al. for testing mutation subtypes in CALR-mutant myeloproliferative neoplasms. Leukemia CALR mutations in patients who are JAK2 negative and present 2016;30:431–438. with BCS. However, our second patient highlights the limitations [6] Sekhar M, Patch D, Austen B, Howard J, Hart S. Calreticulin mutations and of this approach. We would suggest that more studies are their importance in splanchnic vein thrombosis. Br J Haematol 2016;174:158–160. required before any recommendations can be made regarding [7] Haslam K, Langabeer SE. Incidence of CALR mutations in patients with the criteria for selective testing of CALR. More work is also splanchnic vein thrombosis. Br J Haematol 2015;168:459–460. required to understand the long-term impact of CALR mutations [8] Plompen EP, Valk PJ, Chu I, Darwish Murad SD, Plessier A, Turon F, et al. in patients with BCS. European Network for Vascular Disorders of the Liver (EN-Vie). Somatic calreticulin mutations in patients with Budd-Chiari syndrome and portal vein thrombosis. Haematologica 2015;100:e226–e228. Conflict of interest [9] Zhang P, Ma H, Min Q, Zu M, Lu Z. CALR mutations in Chinese Budd-Chiari syndrome patients. Eur J Gastroenterol Hepatol 2016;28:361–362. The authors declare no conflict of interest. Abhinav Jain Pratik Tibdewal Author contributions ⇑ Akash Shukla DM Gastro. Department of Gastroenterology, G.S. Medical College A Jain: A substantial contribution to analysis and interpretation and KEM Hospital, Mumbai 400012, India of data and critical writing. P Tibdewal: A substantial contribu- Corresponding author. Address: Department of Gastroenterology, tion to analysis and interpretation of data. A Shukla: A substantial Set G.S. Medical College and KEM Hospital, Parel, contribution to interpretation of data; critical writing, revising Mumbai 400012, India. the intellectual content and final approval of the version to be Tel.: +91 22 24103057; fax: +91 22 24103057. published. E-mail address: [email protected]

Reply to: ‘‘Calreticulin mutations and their importance in Budd-Chiari syndrome”

To the Editor: haematological phenotype in patients with splanchnic vein We thank Jain Abhinav et al.1 for their letter ‘‘Calreticulin muta- thrombosis. tions and their importance in Budd-Chiari syndrome”, raising the Out of the 521 patients included in our study, nine had CALR hypothesis that the type of CALR mutation may inﬂuence the mutations. None of them had JAK2V617F. Seven out of the eight

1112 Journal of Hepatology 2017 vol. 67 j 1106–1121 JOURNAL OF HEPATOLOGY

Table 1. Type of CALR mutations and characteristics of patients with Splanchnic Vein Thrombosis. Patient Cohort Age Gender Hematologic Type of Spleen height Platelet count CALR mutations (yrs) diseases SVT (cm) (109/L) (Type) 1 Test 24 Female PMF BCS 20.0 417 52-bp deletion (1) 2 Test 30 Female ET PVT 16.6 436 5-bp insertion (2) 3 Test 39 Male PMF PVT 19.0 453 52-bp deletion (1) 4 Test 32 Female PMF PVT 17.0 476 52-bp deletion (1) 5 Test 36 Male PMF PVT 18.0 477 52-bp deletion (1) 6 Validation 34 Male ET BCS 18.0 300 34-bp deletion (Other) 7 Validation 57 Female PMF PVT 18.0 607 52-bp deletion (1) 8 Validation 57 Female ET BCS 11.0 477 52-bp deletion (1) 9 Validation 73 Male ET PVT NA* 337 52-bp deletion (1)

BCS, Budd-Chiari syndrome; ET, essential thrombocythemia; PMF, primary myeloﬁbrosis; NA, not available; PVT, Portal venous system thrombosis; yrs, years. * No spleen height available but the patient was known to have an enlarged spleen.

patients with CALR mutations, and spleen height and platelet [2] Poisson J, Plessier A, Kiladjian J-J, Turon F, Cassinat B, Andreoli A, et al. count available, had both spleen height P16 cm and platelet Selective testing for calreticulin gene mutations in patients with splanchnic vein thrombosis: A prospective cohort study. J Hepatol count [200Â109 /L, suggesting that CALR mutations testing can 2017;67:501–507. 2 be restricted to patients fulfilling these criteria. [3] Vainchenker W, Delhommeau F, Constantinescu SN, Bernard OA. New Jain Abhinav et al. reported two patients with CALR mutations mutations and pathogenesis of myeloproliferative neoplasms. Blood out of 210 patients with Budd-Chiari syndrome. One patient, with 2011;118:1723–1735. a type 2 CALR mutation (5-bp insertion), fulfilled the criteria we [4] Pietra D, Rumi E, Ferretti VV, Buduo CAD, Milanesi C, Cavalloni C, et al. 9 Differential clinical effects of different mutation subtypes in CALR-mutant proposed. The other, with a 3-bp deletion, had 63Â10 /L platelet myeloproliferative neoplasms. Leukemia 2016;30:431–438. count and a spleen height of 11 cm, apparently questioning the [5] Spivak JL. Myeloproliferative Neoplasms. N Engl J Med 2017;376:2168–2181. validity of our criteria. However, mutations involving indels [6] Szuber N, Lamontagne B, Busque L. Novel germline mutations in the occurring as multiples of 3-bp preserve the original reading calreticulin gene: implications for the diagnosis of myeloproliferative neoplasms. J Clin Pathol 2016. http://dx.doi.org/10.1136/jclinpath-2016- frame and are not known to be pathogenic.3–6 203940. Consequently, the only patient with a pathogenic CALR mutation described by Jain Abhinav et al. meets our criteria, thus rein- Johanne Poisson1 forcing our conclusions. The hypothesis that the type of mutation Fanny Turon2 could influence the validity of our criteria remains an interesting Christophe Marzac3,4 question. We analyzed the types of CALR mutations in our study Dominique-Charles Valla5,6,7 according to platelet count and spleen height. As shown in the 2,8 Juan-Carlos Garcia-Pagan ⇑ Table 1, the patient not fulfilling our criteria (patient 8) had a Pierre-Emmanuel Rautou1,5,7, type 1 mutation, while the two patients with type 2 and type 1Inserm, U970, Paris Cardiovascular Research Center – PARCC, 1-like mutation (patients 2 and 6) had a spleen height of Université Paris Descartes, Sorbonne Paris Cité, Paris, France P [ 9 16 cm and a platelet count 200Â10 /L. In conclusion, the 2Barcelona Hepatic Hemodynamic Laboratory, Liver Unit, criteria we proposed do not seem to be influenced by the type Hospital Clínic, IDIBAPS, Barcelona, Spain of CALR mutation. 3UPMC, Univ Paris 06, GRC n°7, Groupe de Recherche Clinique sur les Myéloproliférations Aiguës et Chroniques MYPAC, Paris, France Financial support 4Laboratoire d’Hématologie, Département de Biologie et Pathologie Médicales, Institut Gustave Roussy, Villejuif, France This work was supported by the Agence Nationale pour la 5DHU Unity, Pôle des Maladies de l’Appareil Digestif, Recherche (ANR 14 CE35 0022 03/JAK-POT) and J.P by the ‘‘poste Service d’Hépatologie, Centre de Référence des Maladies Vasculaires accueil INSERM”. du Foie, Hôpital Beaujon, AP-HP, Clichy, France 6Inserm U1149, Centre de Recherche sur l’Inflammation (CRI), Conflict of interest Paris, Université Paris 7-Denis-Diderot, Clichy, UFR de Médecine, Paris, France 7 The authors who have taken part in this study declared that they Université Paris Diderot, Sorbonne Paris cité, Paris, France 8 do not have anything to disclose regarding funding or conflict of Centro de Investigación Biomédica en Red de Enfermedades interest with respect to this manuscript. ⇑ Hepáticas y Digestivas (CIBERehd), Spain Corresponding author. Address: Service d’Hépatologie, Hôpital Beaujon, Assistance Publique-Hôpitaux de Paris, References Clichy, France. Tel.: +33 171114679; fax: +33 140875530. [1] Jain A, Tibdewal P, Shukla A. Calreticulin mutations and their importance in E-mail address: [email protected] Budd-Chiari syndrome. J Hepatol 2017;67:1111–1112.

Journal of Hepatology 2017 vol. 67 j 1106–1121 1113 C. Appendix 3: Curriculum vitae and list of publications

Johanne Poisson Woman, 32 years old, French and Canadian [email protected] ; [email protected]

Education 2015-now PhD in cardiovascular pathophysiology. Paris Descartes University. Supervisor Pr. Pierre-Emmanuel Rautou (in progress) 2018 Diploma of animal experimentation. Level 1. Paris Descartes University 2015 Medical doctor. PhD in Internal Medicine. Pierre and Marie Curie University. Paris 2014-2015 Master of science. Vascular biology, atherosclerosis, thrombosis and haemostasis. Paris Descartes University 2013-now Speciality in Geriatric Medicine. Pierre and Marie Curie University. Paris (in progress) 2013 Master of science. Biological ageing. Pierre and Marie Curie University. Paris 2010-now Speciality in Internal Medicine. Pierre and Marie Curie university. Paris (in progress) 2010 French national residency examination. Ranked 134 on 6962 candidats 2003-2010 Medical education. Montpellier University. France 2004 Medical education entry examination. Montpellier University. Ranked 85 on 1438 candidats

Work experience 2015-now Paris Cardiovascular Center (PARCC). Inserm U970. Paris. France PhD Thesis: Pathophysiology of cardiovascular events in myeloproliferative neoplasms 2015 Paris Cardiovascular Center (PARCC). Inserm U970. Paris. France Traineeship: Shear stress regulates endothelial senescence: Role of autophagy 2010-now Paris university hospitals. Medical residency 2008 Addiction treatment and research unit, Penn university hospital, Presbyterian hospital, Hospital of the university of Pennsylvania, Philadelphia, USA. 3 month traineeship 2006-2010 Montpellier University hospitals. Medical education internships

Grants and Awards 2015-2018 PhD fellowship « Poste d’accueil INSERM » (3 years) 2014-2015 Master of science fellowship by « Fonds d’études et de recherche du corps médical »

191 List of publications Original articles 1. Meunier T, Francois A, Poisson J, Gisselbrecht M, Arlet J.B, Ducot L, Lahjibi-Paulet H, Le Guen J, Mercadier E, Pouchot J, Saint-Jean O. Blood management in geriatric hospitalized population. La Revue de Medecine Interne. 2017 IF 0,6 2. Vion AC*, Kheloufi M*, Hammoutene A, Poisson J, Lasselin J, Devue C, Pic I, Dupont N, Busse J, Stark K, Lafaurie-Janvore J, Barakat A, Loyer X, Souyri M, Viollet B, Julia P, Tedgui A, Codogno P, Boulanger CM, Rautou PE. Autophagy is required for endothelial cell alignment and atheroprotection under physiological blood flow. Proc Natl Acad Sci U S A. 2017 Oct 10. IF 9,7 3. Poisson J, Plessier A, Kiladjian JJ, Turon F, Andreoli A, De Raucourt E, Goria O, Zekrini K, Bureau C, Lorre F, Cervantes F, Colomer D, Durand F, Boulanger C, Garcia-Pagan JC, Casadevall N, Valla DC, Rautou PE, Marzac C. Selective testing for CALR mutations in patients with splanchnic vein thrombosis: a prospective cohort study in 312 patients. Journal of Hepatology. 2017. 2017 May 5. IF 12,5 4. Baudry E, Cotto E, Poisson J, Couderc AL, Bornand-Rouseelot A, Goehrs L, Chaibi P, Canoui-Poitrine F. Réponse et tolérance à la chimiothérapie dans le lymphome B à grandes cellules du sujet âgé. Le Journal d’OncoGériatrie. Avril 2016, Volume 7, Numéro 1. 5. Touzot M, Poisson J, Faguer S, Ribes D, Cohen P, Geffray L, Anguel N, François H, Karras A, Cacoub P, Durrbach A, Saadoun D. Rituximab in anti-GBM disease: A retrospective study of 8 patients. Journal of Autoimmunity. 2015 Jun;60:74-9. IF 7,6 6. Le Guen J, Lenain E, Poisson J, Chatellier G, Saint Jean O. Hospitalisation des centenaires en France de 2009 à 2012 au sein d’un échantillon représentatif de la population générale. La Revue Francophone de Gériatrie et de Gérontologie. Juin 2015. N°216. 7. Poisson J, Six M, Morin C, Fardet L. Glucocorticoid therapy: what is the information sought by patients? Traffic analysis of the website cortisone-info.fr. La Revue de Medecine Interne. 2013 May. IF 0,6

Literature reviews 1. Poisson J*, Lemoinne S*, Boulanger C, Durand F, Moreau R, Valla D, Rautou PE. Liver sinusoidal endothelial cells: physiology and role in liver diseases. Journal of Hepatology. 2016 Jul 13. IF 12,5 2. Poisson J, Chaoui D. Livre : L’oncogériatrie en pratique par le FROG. Chapitre 27. Prise en charge du lymphome non hodgkinien B diffus à grandes cellules. 3. Poisson J, Chaoui D. Soins Gérontologie. N°112 mars/avril 2015.10.1016/j.sger.2015.01.004. Malignant blood diseases in the elderly.

Others 1. Poisson J, Hilscher MB, Tanguy M, Hammoutene A, Boulanger CM, Villeval JL, Douglas DA, Valla D, Shah VH, Rautou PE. Endothelial JAK2V617F does not enhance liver lesions in mice model wth Budd-Chiari syndrome. Journal of Hepatology 2018. IF 12,5 2. Kheloufi M, Vion AC, Hammoutene A, Poisson J, Lasselin J, Devue C, Pic I, Dupont N, Busse J, Stark K, Lafaurie-Janvore J, Barakat A, Loyer X, Souyri M, Viollet B, Julia P, Tedgui A, Codogno P, Boulanger CM, Rautou PE. Endothelial autophagic flux hampers atherosclerotic lesion development. Autophagy. 2017 Nov 20. IF 8,6 3. Poisson J, Turon F, Marzac C, Valla DC, Garcia-Pagan JC, Rautou PE. Reply to: Letter "Calreticulin mutations and their importance in Budd-Chiari syndrome". Journal of Hepatology. 2017. 2017 Jun 27. IF 12,5 4. Poisson J, Aregui A, Maley K, Darnige L, Gisselbrecht M. Association of chylothorax and direct pleura involvement in a case of Waldenström’s macroglobulinemia. Age Ageing. 2014 Jul. IF 4,3

192 Résumé Les syndromes myéloprolifératifs Bcr/Abl-negatifs (SMP) sont des maladies hématopoïétiques clonales, secondaires dans 80% des cas à la mutation sporadique JAK2V617F. JAK2V617F a été récemment mise en évidence dans les cellules endothéliales. Les deuxièmes mutations en ordre de fréquence sont les mutations de Calréticulin (CALR). Les évènements cardiovasculaires sont la première cause de mortalité des malades atteints de SMP (2/3 artériels et 1/3 veineux). Les évènements veineux sont caractérisés par une fréquence particulièrement élevée de thromboses survenant dans des sites inhabituels, comme les veines splanchniques (veines hépatiques (syndrome de Budd-Chiari) ou veine porte). Les mécanismes responsables de ces évènements cardiovasculaires chez les malades atteints de SMP sont mal compris. Les conséquences phénotypiques et fonctionnelles de la mutation JAK2V617F sur les cellules endothéliales n’ont pas été évaluées. L’objectif global de ce travail est de mieux comprendre la physiopathologie des évènements cardio-vasculaires associés aux SMP, Sur le plan artériel, j'ai montré, grâce à des expériences de myographie, que les aortes de souris portant JAK2V617F à la fois dans leurs cellules hématopoïétiques et endothéliales ont une très forte augmentation de la réponse aux agents vasoconstricteurs alors que cet effet n’est pas observé lorsque la mutation est uniquement endothéliale. J’ai ensuite isolé des microvésicules plasmatiques de malades porteurs de JAK2V617F, non traités pour leur SMP, et ai observé que ces microvésicules reproduisent l’effet d’hyperréponse artérielle aux agents vasoconstricteurs. J'ai par la suite montré que seules les microvésicules de globules rouges portant la mutation JAK2V617F étaient responsables de cet effet. J’ai ensuite analysé les mécanismes impliqués et ai déterminé que cette hyperréactivité vasculaire est dépendante de l’endothélium et des NO synthases. De plus, j'ai aussi mis en évidence une forte augmentation du stress oxydant dans l'endothélium des aortes de souris portant la mutation JAK2V617F en comparaison aux souris sauvages, suggérant que les perturbations de la voie du NO résultent d’une génération de stress oxydant induite par les microvésicules de globules rouges. Ces données nous ont poussés à évaluer de nouvelles thérapeutiques, comme les statines, qui sont des molécules anti- cholestérolémiques ayant indépendamment du cholestérol un effet anti-oxydant bénéfique sur la fonction endothélial. J'ai montré que l'utilisation de simvastatine chez les souris portant la mutation JAK2V617F diminue significative cette hyperréactivité vasculaire en comparaison aux souris contrôles. Mes résultats suggèrent que chez les malades atteints de SMP une hyperréponse aux agents vasoconstricteurs induite par les microvésicules circulantes d'origine érythrocytaire pourrait participer aux accidents artériels et que de nouvelles thérapeutiques, comme les statines pourraient être prometteuses. Sur le plan veineux, en analysant une cohorte prospective de 312 patients atteints de thromboses splanchniques, j’ai pu déterminer la prévalence des mutations CALR dans cette population (2%) et identifier un groupe de malades (ceux sans JAK2V617F et ayant une taille de rate ≥ 16cm et des plaquettes > 200G/L) chez lesquels la recherche des mutations CALR doit être effectuée. Ces critères ont une excellente valeur prédictive négative (100%) et permettent d’éviter 96% de tests inutiles. J’ai confirmé ces critères grâce à une collaboration européenne dans une cohorte de validation espagnole comprenant 209 patients. Je me suis aussi attachée à déterminer le rôle de JAK2V617F endothélial dans les conséquences hépatiques des thromboses des veines hépatiques. Nous avons montré que la présence de la mutation JAK2V617F dans l'endothélium n'aggrave pas le développement des lésions hépatiques secondaires à un syndrome de Budd-Chiari, ni en termes de fibrose hépatique ni en termes d’hypertension portale.

191