Platelet function studies in myeloproliferative neoplasms patients with Calreticulin or JAK2V617F mutation

Background JAK2V617F and Calreticulin (CALR) mutations are the most frequent molecular causes of Phi-negative myeloproliferative neoplasms (MPN). Patients with CALR mutations are at lower risk of thrombosis than patients with JAK2V617F. We hypothesized that CALR-mutated blood platelets would have platelet function defects that might explain why these patients are at lower risk of thrombosis. Objectives Our main objective was to explore and compare platelet function depending on the MPN molecular marker. Methods We analyzed platelet function in 16 patients with MPN with CALR mutations and 17 patients with JAK2V617F mutation and compared them with healthy controls. None of these patients was taking antiplatelet therapy. We performed an extensive analysis of platelet function and measured plasmatic soluble P-selectin and CD40L levels. Results We observed significant defects in platelet aggregation, surface glycoprotein expression, fibrinogen binding, and granule content in platelets from patients with MPN compared with that in controls. Moreover, soluble CD40L and P-selectin levels were elevated in patients with MPN compared with that in controls, suggesting an in vivo platelet preactivation. Comparison of platelet function between patients with CALR and JAK2V617F MPN revealed only minor differences in platelets from patients with CALR. However, these results need to be interpreted within the context of absence of an inflammatory environment that could impact platelet function during MPN. Conclusions These results do not support the hypothesis that calreticulin-mutated platelets have platelet function defects that could explain the lower thrombotic risk of patients with CALR.

domains. The N-and P-domains contain the lectin and ERp-57 binding sites, allowing calreticulin to play its role of chaperone [3,4] because it contributes to better folding of neosynthesized glycoproteins. Calreticulin is also a major protein in calcium homeostasis-thanks to the C-domain that contains many calcium-binding sites, allowing its storage/buffering within the ER. More than 50 CALR mutations have been previously described, with 2 of them, ie, CALRdel52 and CALRins5, representing more than 80% of cases [1,2]. All CALR mutations affect the C-domain of the protein resulting in the same 36 amino acid terminal sequence and the loss of several calcium-binding sites. Interestingly, it has been shown that CALR mutants can activate the JAK/STAT pathway by binding to the Nglycosylated residue of the thrombopoietin receptor MPL [5][6][7][8].
Disease evolution of MPN is marked by the onset of several complications, among which thrombosis is the most frequent. It occurs in approximately 30% of PV [9] and 20% of ET [10] at diagnosis and in more than 10% during the follow-up of the disease [9,10]. The 2 main risk factors for thrombosis are age older than 60 years and a history of thrombosis. Surprisingly, although CALR and JAK2 V617F mutations act on the same intracellular signaling pathways, the thrombotic risk is significantly lower in patients carrying CALR mutations compared with JAK2 V617F in essential thrombocythemia (ET) [1,11]. The reason for such a difference is yet to be investigated.
Many studies have sought to explore the roles of the different blood cells in the pathogenesis of thrombosis during MPN [12].
Because platelets play a critical role in thrombus formation, they require particular attention when exploring the mechanisms of thrombogenicity in MPN. Knowing the key role of calcium in platelet activation, we hypothesized that the potential dysregulation of calcium homeostasis induced by CALR mutations could disturb platelet function, explaining the relative decreased risk of thrombotic complications in these patients. In this study, our objective was to explore and compare platelet functions in patients with MPN carrying CALR or JAK2 mutations without taking any antiplatelet therapy.

| Study population
Sixteen patients with CALR and 17 patients with JAK2 V617F MPN without ongoing antiaggregatory drugs were prospectively enrolled in the University Hospital of Bordeaux between 2018 and 2020 after having obtained their informed consent. In total, 11 patients were treated using hydroxyurea. All investigations were approved by the local ethics committee. Diagnoses of MPN were made according to WHO 2008 criteria.
Because we were concerned that a sufficient proportion of platelets would be mutated for CALR or JAK2 V617F , we chose to only include patients with an allele burden greater than 10%. The results were compared with those obtained from 10 different healthy volunteers.

| Collection of blood samples
For platelet count, blood was collected on sodium ethylene diamine tetraacetate-anticoagulated tubes. Platelet counts were obtained using the Beckman DXH 800 hematology analyzer (Beckman Coulter), which is used routinely. For platelet function assays and electron microscopy, blood was collected on citrate tubes containing 0.109 molar sodium citrate and tubes containing acid citrate dextrose. To avoid platelet activation, samples were analyzed within 2 hours after sampling. Dry tubes for serum collection were also used.

| Light transmission aggregometry
Platelet aggregation was tested in citrated platelet-rich plasma (PRP) aggregometer (Elitech) according to standard procedures [13]. Native PRP concentration was adjusted if the platelet count was higher than 600G/L to reach a platelet count of 500 G/L.

Essentials
• Patients with MPN with CALR mutations are at lower risk of thrombosis than patients with JAK2 V617F .
• Platelet function analyses were conducted in 16 patients with CALR MPN.
• Analysis revealed platelet function defects in patients with MPN compared with that in controls.
• We observed minor differences in platelets between patients with CALR and patients with JAK2 V617F .

| Platelet flow cytometry studies
Surface expressions of the main platelet glycoproteins (GPs) GPIb, α IIb β 3 , and P-selectin were measured in PRP (250 × 10 9 /L) using a platelet calibrator kit (Biocytex) with specific monoclonal antibodies, according to the manufacturer's instructions. Briefly, after sample stabilization at room temperature for 1 hour, platelets were stained for 20 minutes before cytometry analysis. Readings were taken before and after activation with 60 μM TRAP. Results were expressed as number of sites derived from mean fluorescence intensities.
Platelet fibrinogen binding was measured using PRP adjusted at 250 × 10 9 /L and incubated with Alexa Fluor 488-labeled human fibrinogen (Molecular probes) after activation by 10-μM ADP.

| Soluble markers of platelet activation
Soluble P-selectin was measured in patients' sera using the Quantikine Human P-selectin ELISA kit (R&D systems), and soluble CD40L with the Invitrogen Human CD40L ELISA Kit (ThermoFischer Scientific), according to the manufacturer's instructions.

| Statistical analysis
The presence or absence of normality was tested for all the parameters analyzed. Comparisons between the 3 groups were made using 1-way analysis of variance (followed by analysis of multiple comparisons with Tukey's test) or the Kruskal-Wallis test (followed by analysis of multiple comparisons with Dunn's test). Owing to the absence of normality distribution for the groups analyzed, comparisons between 2 groups required Mann-Whitney U-tests. All results, comparisons, and graphs were performed with the statistical software GraphPad Prism 9. A P value < .05 was considered statistically significant. Male-to-female ratio was similar between the 2 groups. All patients and controls were Caucasians. Allele burden was similar between patients with CALR+ (mean: 31.2%, min: 11%, max: 48%) and JAK2 V617F + (mean: 45%, min: 15%, max: 88%). Twelve patients (37.5%; 6 patients with CALR+ and 6 patients with JAK2 V617F + MPN) were treated with cytoreduction. Patients with CALR+ MPN had an increased platelet count (793 × 10 9 /L ± 104) compared with patients with JAK2 V617F + MPN (PLT = 555 × 10 9 /L ± 75.8) ( Table 1). The results were compared with those of the controls with the analysis of 10 healthy volunteers. Seven men and 3 women were studied with a mean age of 34 years (SD: 6.9) and a mean platelet count of 248 G/L (SD: 24).

| Patients with MPN have moderate platelet aggregation defects
We first assessed platelet function using light transmission aggregometry with 6 different agonists. We did not observe any difference in response to 5-μM ionophore, 2.5-μM ADP, and 1 μg/mL collagen, whereas in response to arachidonic acid, platelet aggregation was significantly lower in patients with MPN compared with controls ( Figure 1A). Interestingly, epinephrine-and TRAP-induced platelet aggregation was significantly lower in patients with CALR+ compared with control and patients with JAK2 V617F + ( Figure 1A). This result was GUY ET AL.

| Platelet surface glycoprotein expression is impaired in patients with MPN
We then wondered whether there were differences in surface glycoprotein expression. At baseline, α IIb β 3 and GPIb expressions were significantly reduced in blood platelets from patients with CALR+ and JAK2 V617F +, without difference between blood platelets from patients with CALR+ and JAK2 V617F + MPN ( Figure 1B). After activation, we

| Fibrinogen-binding defects in MPN platelets
Fibrinogen binding is a marker of α IIb β 3 activation with a switch to a high-affinity conformation, allowing blood platelets to aggregate. We explored fibrinogen binding to α IIb β 3 at different time points after stimulation with 10-μM ADP. Fibrinogen binding was similarly and significantly impaired in both MPN groups compared with controls, at all time points analyzed ( Figure 1D), which may be related to decreased expression of α IIb β 3 expression ( Figure 1B).

| Platelet granule defects
To study α-granules, we measured P-selectin (CD62P) on resting and activated platelets. Platelets from patients with CALR-positive mutation showed a significant reduction in baseline P-selectin expression compared with platelets from controls and patients with JAK2 V617F -positive MPN. The difference between CALR and JAK2 V617F patients' platelets was not significant when we restricted our analysis to patients with ET ( Supplementary Fig. S1). After   Data are presented as boxes with 5% and 95% percentiles. Boxes show the first and third quartiles. Whiskers indicate 5% and 95% percentiles. Statistical significance assessed by using 1-way analysis of variance followed by posthoc Dunn's test. *P < .05; **P < .01; ***P < .001; ****P < .0001. (B) α IIb β 3 and GPIbα expression at resting state; (C) Ratios of α IIb β 3 and GPIbα expressed after platelet activation with TRAP 60 μM. Data are presented as boxes with 5% and 95% percentiles. Statistical significance was assessed using the 1-way analysis of variance followed by posthoc Dunn's test. *P < .05; **P < .01; ***P < .001; ****P < .0001. Analysis realized in platelets from 31 patients. (D) Fibrinogen binding in platelets from patients with CALR-and JAK2 V617F mutation. Fibrinogen binding was evaluated using flow cytometry at baseline and at different time points after ADP activation. Data are presented as boxes with 5% and 95% percentiles. Statistical significance was assessed using 1-way analysis of variance followed by posthoc Dunn's test. **P < .01; ***P < .001; ****P < .0001. Analysis realized in platelets from 33 patients. (E) CD62P expression in platelets from CALR-and JAK2 V617F -mutated patients at resting state and after platelet activation GUY ET AL. with TRAP 60 μM. Data are presented as boxes with 5% and 95% percentiles. Statistical significance was assessed using the 1-way analysis of variance followed by posthoc Dunn's test. *P < .05; **P < .01; ****P < .0001. Analysis realized in platelets from 31 patients. (F) Evaluation of δ-granules content in platelets from patients with CALR and JAK2 V617F mutation by analysis of dense body content by using electronic microscopy. Data are presented as mean and SEM. Analysis realized in platelets from 30 patients. (G) Evaluation of δ-granules content in platelets from patients with CALR and JAK2 V617F mutation ATP secretion measurement in the presence of different agonists: ADP 5 μM, collagen 1 μg/mL, and TRAP 10 μM. Data are presented as boxes with 5% and 95% percentiles. Statistical significance was assessed using 1-way analysis of variance followed by posthoc Dunn's test. *P < .05; **P < .01; ****P < .0001. Analysis realized in platelets from 33 patients.

| Platelets from patients with JAK2 V617F + mutation have increased procoagulant activity compared with controls
During platelet activation, the intracellular release of calcium activates the enzymes responsible for phospholipid externalization.
Thereby, phosphatidylserine is translocated to the outer membrane surface, allowing binding of coagulation factors. We thus measured phosphatidylserine exposure by using annexin V, before and after platelet activation with 5 μM ionophore. At baseline, there was no difference in phosphatidylserine exposure between the 3 groups ( Figure 2A), whereas after platelet activation, phosphatidylserine exposure was significantly higher in the JAK2 V617F group in comparison with controls ( Figure 2B), suggesting an increased procoagulant activity in this group. We did not observe differences in PS exposure between platelets from patients with CALR+ and JAK2 V617F +.
To note, we did not observe any significant differences between patients with low-and high-allele burdens, either in the JAK2 V617F or CALR group. We also found no difference in platelet aggregation, glycoprotein expression, fibrinogen binding, dense body content, and ATP secretion (Supplementary Fig. S2).

| Patients with MPN have increased soluble markers of platelet activation
To evaluate in vivo platelet activation, P-selectin and CD40L amounts were measured in patient sera. This method offers a reliable evaluation of the platelet activation state in vivo, whereas the upper functional tests explore ex vivo platelet reactivity. Soluble CD40L concentration ( Figure 3A) was significantly higher in CALR and JAK2 V617F groups, suggesting a previous platelet activation state in our patients. We also observed increased soluble P-selectin ( Figure 3B) concentration in patients with JAK2 V617F +. In patients with CALR+, soluble P-selectin concentration was also increased but without statistical significance (P = .06). To analyze the platelet activation status and because platelet count differs between controls and patients with MPN, we then analyzed the ratio between CD40L or P-selectin and the platelet count. The soluble CD40L/platelet ratio was increased in patients with CALR and JAK2 V617F , confirming the in vivo platelets activation ( Figure 3C). We did not observe a statistically significant increase of soluble P-selectin/platelet ratio, possibly owing to the low number of controls analyzed for this marker ( Figure 3D). Finally, we did not observe any difference between CALR and JAK2 V617F groups regarding CD40L or soluble P-selectin concentration.

| Hydroxyurea does not affect most platelet function tests but reduces procoagulant activity in patients with MPN
Treatment with hydroxyurea (HU) is known to decrease the thrombotic risk in patients with MPN, mainly through decreased production of blood cells. To test whether treatment with HU could influence platelet function, we analyzed separately platelet function results in the 11 patients with MPN taking HU and in 21 patients without cytoreductive treatment whatever their molecular defects. These 2 groups were similar in JAK2 V617F /CALR status, age, gender repartition, MPN type, or allele burden. As expected, patients with HU had a significantly lower platelet count ( Table 2). In most platelet function assays, we did not observe any difference between treated and untreated patients (Figures 4A-G). Only phosphatidylserine exposure was reduced in the group of patients with HU, in favor of a lower procoagulant activity with this treatment ( Figure 4H). Finally, there was no difference in P-selectin or CD40L concentrations between patients treated with HU and untreated patients (Figures 4I-J). sCD62P concentration (ng/mL) (mean ± SD) 377.8 ± 45 340.9 ± 54 .66 CD40-ligand concentration (ng/mL) (mean ±SD) 12.7 ± 4.9 9.7 ± 1.6 .18 Statistical significance assessed by using Student's t-tests or Mann-Whitney U-tests for age, allele burden, platelet count, and sCD62P and CD40-L concentrations. Statistical significance assessed by using Fisher exact tests for gender and mutation type repartition. Statistically significant P values are indicated in bold. CALR compared with those with JAK2 V617F . We also aimed to compare platelets from CALR and JAK2 V617F + patients with platelets from healthy controls.
We here observed that in vitro platelet function was impaired in patients with MPN, with arguments for in vivo platelet preactivation.
The comparison of patients with CALR and JAK2 V617F + highlighted moderate differences in platelet function, not suggesting a crucial role for platelets in the difference in thrombotic risk.
Our results revealed that most of platelet function studies were defective in patients with MPN compared with healthy controls. These defects include storage pool deficiency, impaired aggregation in response to agonists, and decrease in glycoprotein expression and fibrinogen binding after platelet activation. We did not observe any statistically significant differences between patients with low-and high-allele burden, either in the JAK2 V617F or CALR group. Nevertheless, allele burden was quantified in leukocytes but not in platelets.
As a recent study reported that the allele burden can be different between platelets and granulocytes [16], we cannot definitively conclude that the allele burden in platelets does not correlate with the platelets function defect we observed. Our observations of defective platelet function in patients with MPN may seem counterintuitive, given the prothrombotic state during MPN, where one would have expected hyperreactive platelets. These in vitro defects were also reported in a number of previous works. For instance, impairment of platelet aggregation [17] and the defects of membrane glycoprotein expression [18,19] and granule content [20,21] have been previously described. A study reported abnormalities in arachidonic acid metabolism [20], which is consistent with our results that showed decreased aggregation in MPN platelets after activation with arachidonic acid. Recently, a study also revealed platelet dysfunction in JAK2 V617F transgenic mice [22].
The reasons for the in vitro MPN platelet function defects observed in our study, although consistent with the literature, deserve to be investigated. One hypothesis is that these platelets could be exhausted platelets that have already been activated and therefore have already released their content. The defect we observed in vitro would thus be an after-effect of an in vivo platelet preactivation. In line with that, platelet activation products were shown to be increased during MPN, such as platelet factor 4, β-thromboglobulin [23], urinary thromboxane metabolites [24], and soluble P-selectin or CD40 [25]. Our study confirms these observations as we report higher concentrations of soluble CD40L and P-selectin within CALR and JAK2 V617F groups than in controls. This in vivo platelet activation may be due to intrinsic platelet activation because of JAK-STAT pathway activation in megakaryocytes.
It is known that the JAK2-STAT3 pathway is involved in collageninduced platelet activation [26], and it has been demonstrated that a constitutive Src kinase preactivation is involved in platelet hyperreactivity in patients with MPN [27]. It has also been shown, using a mouse model allowing the expression of the human JAK2V617F, that JAK2 V617F megakaryocytes (MKs) display increased migration and proplatelet formation. In the same study, the authors demonstrated significant changes in RNA expression in JAK2 V617F MKs, notably in the cytoskeleton assembly and apoptosis pathways [28]. This could also be a consequence of platelet activation by inflammatory molecules or increased thrombin generation. Finally, it suggests that CALR and JAK2 V617F platelets are equally activated in vivo.
Given the participation of platelets in coagulation pathway activation, we studied platelet procoagulant activity and observed that the procoagulant activity (explored by the measurement of phosphatidylserine exposure) was higher in the JAK2 V617F group only in comparison with controls. Besides, the importance of investigating procoagulant activity was demonstrated in a recent work conducted in patients with ET that compared phosphatidylserine exposure on blood cells from different mutational subtypes. Patients with JAK2 V617F had a higher phosphatidylserine exposure than CALR and triple-negative groups, which was correlated with more hypercoagulability and could possibly explain the higher thrombotic risk in the JAK2 V617F group [29]. These results suggest that JAK2 V617F platelets could trigger coagulation activation in a greater extent than CALR platelets.
Because the discoveries of driver mutations are relatively recent, most of the studies that focuses on platelet function in patients with MPN have been performed regardless of the mutation type. It is essential to distinguish the different mutational subtypes because clinical studies revealed differences in thrombotic risk depending on the driver mutation [9,10]. Our investigations here report a defective platelet aggregation in the presence of epinephrine and TRAP and slight differences in glycoprotein expression in platelets from patients with CALR compared with platelets from patients with JAK2 V617F mutation. Analysis of all MPN patients showed a significant decrease in basal CD62P expression in platelets from patients with CALR compared with that from patients with JAK2 V617F mutation, and subanalysis in patients with ET showed a significant decrease in basal GpIIbIIIa expression and a significant decrease in Gp1b internalization after activation. However, all the other tests showed comparable in vitro defects between platelets from patients with CALR and JAK2 V617F mutation. The study of platelet characteristics from patients with CALR and JAK2 V617F -positive ET has been performed in a recent study. On the basis of activation and adhesion assays, they concluded to less reactivity of CALR-mutated platelets in comparison with JAK2 V617F platelets, with the defect of CD62P and α IIb β 3 expressions after ADP stimulation [30]. However, this study has a major limitation because more than 90% of patients were taking antiplatelet therapy. Contrary to the study by Hauschner et al. [30], we did not observe decreased CD62P expression or α IIb β 3 expression after stimulation, but we did not use the same agonist. Nevertheless, the differences we here report are minor, though significant, as the only tests that were different are maximal aggregation after stimulation with epinephrin and TRAP (but not with other inductors) and slight differences in glycoprotein expression. Altogether, we do not think that such slight differences could explain the difference in thrombotic risk observed in clinical studies between patients with CALR and JAK2 V617F mutation. To our knowledge, our study is first analyzing CALR-and JAK2 V617F -mutated platelets of patients free from antiplatelet drugs. This inclusion criterion appears essential because these drugs affect platelet functions. GUY ET AL.
Regarding cytoreductive treatments, few studies have explored their potential effect on platelet function. One compared patients receiving HU with untreated ones and did not find any difference in Pselectin expression [31], whereas the other used the markers of in vivo clotting formation (levels of prothrombin fragment and thrombinantithrombin complex) and described no differences as well [32]. A recent study analyzed platelet characteristics after aspirin uptake in HU or pegylated interferon alpha (peg-INF)-treated patients with ET.
The authors observed that patients treated with peg-IFN had an increased aspirin-mediated platelet inhibition than patients treated with HU [33]. Our comparison of patients with MPN with and without HU showed a reduction in platelet procoagulant activity in the treated group after platelet stimulation by ionophore. This result is consistent with another study that described a decrease in thrombin generation when patients suffering PV and ET were treated by HU [34].
Our study has some limitations. The first is the presence of hydroxyurea-treated patients with MPN (11/28). Indeed, hydroxyurea has potential anti-inflammatory effects that could thus modify platelet functions. Second, we did not analyze the extrinsic factors that may modify platelet functions, such as inflammatory cytokines. Indeed, it is known that patients with MPN do have a proinflammatory environment that participates in the physiopathology of the disease [35].
In conclusion, our study showed global in vitro defect of platelet function in patients with CALR and JAK2 V617F compared with controls.
Interestingly, we here provide an extensive analysis of platelets from patients with CALR mutation, highlighting the presence of in vitro defective function, in a same extent than patients with JAK2 V617Fpositive mutation. Finally, soluble markers of platelet activation were increased in both MPN groups, suggesting that the ex vivo platelet function defects observed could be due to previous in vivo activation, either at the megakaryocyte level or extrinsic factors such as inflammatory molecules or increased thrombin generation.