The microvesicle/CD36 complex triggers a prothrombotic phenotype in patients with non‐valvular atrial fibrillation

Abstract The mechanisms responsible for platelet activation, the prothrombotic state, in non‐valvular atrial fibrillation (NVAF) are still obscure. Microvesicles (MVs) can transfer various messages to target cells and may be helpful for exploring the detailed mechanisms. We aimed to investigate the possible mechanisms by which proatherogenic factors of NVAF contribute to platelet activation. Two hundred and ten patients with NVAF were stratified as being at ‘low to moderate risk’ or ‘high risk’ for stroke according to the CHADS2 score. Levels of platelet‐derived MVs (PMVs) and platelet activation were examined. CD36‐positive or CD36‐deficient human platelets were stimulated by MVs isolated from NVAF patients with or without various inhibitors in vitro. Levels of PMVs and platelet activation markers enhanced significantly in high‐risk patients. The MVs isolated from plasma of NVAF patients bound to platelet CD36 and activated platelets by phosphorylating the mitogen‐activated protein kinase 4/Jun N‐terminal kinase 2 (MKK4/JNK2) pathways. However, CD36 deficiency protected against MV‐induced activation of platelets. We reveal a possible mechanism of platelet activation in NVAF and suggest that the platelet CD36 might be an effective target in preventing the prothrombotic state in NVAF.


| INTRODUC TI ON
Non-valvular atrial fibrillation (NVAF) is a major cause of stroke.
Although warfarin is the mainstay of antithrombotic therapy for NVAF patients in Asian countries, 3.06% of patients per year on warfarin still experience stroke or systemic embolism in Asians. 1 The mechanisms underlying thromboembolism in NVAF patients are still obscure. Platelets provide a catalytic surface for coagulation and bind coagulation factors. In addition, approximately 56% of NVAF patients have concomitant ischaemic heart disease. 2 Concomitant use of antiplatelet therapy and oral anticoagulation was recommended in current guidelines in NVAF patients with ischaemic heart disease. 3 However, antiplatelet therapy in NVAF patients has been challenged. Compared with anticoagulant monotherapy, the combination therapy of antiplatelet therapy and oral anticoagulation increases risk of bleeding without reduction in risk of thromboembolism. 4 Insufficiency of the present antiplatelet therapy may be associated with the diversity of platelet activation pathways. The molecular mechanisms that underlie platelet activation in NVAF are poorly defined but may involve the aetiology of NVAF.
The potential effects of inflammation, oxidative stress and metabolic disorders in the NVAF-related prothrombotic state have aroused much attention. 5 However, these risk factors have not been integrated together in targeting prevention of NVAF-related platelet activation. To determine a vector exhibiting all these proatherogenic factors in NVAF, we can study the potential mechanisms of platelet activation and explore new, effective targets for preventing thrombosis in NVAF. Recently, studies of microvesicles (MVs) have enlightened us. [6][7][8] Circulating microvesicles (MVs), submicrometer vesicles formed by activated cells, can act as messengers delivering a variety of cargos, including cell surface receptors, proinflammatory cytokines and signature proteomes, to target cells. 9,10 Circulating levels of procoagulant MVs, result from integrated actions of oxidative stress and inflammation, are increased in patients with NVAF. 6,11,12 In a recent study, investigators found that MVs isolated from plasma of polyangiitis patients can activate platelets. 13 Thus, we speculated that MVs may be important mediators between pathogenic factors and platelet activation in NVAF.
Phosphatidylserine (PS) on the surface of MVs could be a ligand of CD36, thus generating a MV-CD36 complex. 14 Chen et al 15 showed that platelet CD36 mediates mitogen-activated protein kinase (MAPK) activation induced by oxidized low-density lipoprotein (oxLDL). We previously showed that platelet-derived microvesicles (PMVs) generated in response to oxidative insult could activate CD36/mitogen-activated protein kinase 4/Jun N-terminal kinase 2 (CD36/MKK4/JNK2) signal pathway and lead to platelet activation. 16 Whether the same pathway was also activated in the context of NVAF is unclear.
In the study, we sought to determine whether the MVs, vectors of integrating all the proatherogenic factors in NVAF, could be endogenous CD36 ligands that transmit an activating signal to platelets to induce prothrombotic phenotype and whether the PMV-CD36 complex is an effective target for treating NVAFrelated thrombosis.

| Study design
A case-control study was carried out. The study conformed to the Declaration of Helsinki and was approved by the institutional ethics committee. Written informed consent was obtained from each subject. We included 210 patients with persistent or paroxysmal NVAF (mean age 62.23 ± 11.47 years; 127 men) and 35 healthy controls (mean age 55.71 ± 7.43; 17 men). Patients were classified as being at 'low to moderate risk' or 'high risk' for stroke according to the CHADS2 risk scheme. 3 Patients with CHADS2 score 0 or 1 were considered at 'low to moderate risk' and those with score ≥ 2 at 'high risk'. To minimize the effect of medication, all the subjects had not taken aspirin or non-steroidal anti-inflammatory drugs (NSAIDs) for 1 week before blood samples were collected. Basic characteristics and blood examination of all patients and healthy controls were recorded. The expression of platelet integrin α IIb β 3 and CD36 was detected by flow cytometry. Measurement of oxidized low-density lipoprotein (oxLDL), interleukin-6 (IL-6), 8-iso-prostaglandin F2α (8-iso-PGF2α) and soluble P-selectin was tested by enzyme-linked immunosorbent assay (ELISA). Plasma levels of PMVs or PS-positive PMVs were detected by ELISA and flow cytometry. For MV stimulation experiments, CD36-positive or CD36-deficient resting platelet suspensions were exposed to MVs collected from AF patients.
Platelet integrin α IIb β 3 , CD40L and CD36 were evaluated by flow cytometry. P-selectin was tested by ELISA. Inhibitory strategies with Annexin V and specific antibodies were applied to investigate the role of PS, CD36 or MKK4/JNK2 signalling pathway in the interactions. Phosphorylation of MKK4 and JNK was analysed by Western blotting.
Recombinant human Annexin V was from Enzo Life Sciences. was originally raised against human membrane-bound liver alkaline phosphatase isolated from patients with liver cancer. 10,11 The hybridoma cells were cultured in RPMI 1640 medium containing 10% FBS to produce antibody.

| Laboratory analysis
The expression of platelet integrin α IIb β 3 (PAC-1) and CD36 of subjects was detected by flow cytometry. Citrated plasma was obtained by centrifugation at 3000 g for 10 minutes. Immediately after double centrifugation of whole blood for 10 minutes at 3000 g, the plasma was centrifuged for 3 minutes at 12 000 g to prepare platelet-free plasma (PFP). Measurement of oxLDL, IL-6, 8-iso-PGF2α and soluble P-selectin involved ELISA with commercial reagents (Uscn life science).

| ELISA quantification of PMVs
Quantification of PMVs in PFP involved use of the home-made ELISA as described. [16][17][18][19] Briefly, the MVs were captured on the surface of the ELISA plate coated with AD-1. Then, the PMV detection was achieved by estimating the amount of CD41-positive MVs.
The plate was incubated with biotin-conjugated mouse anti-human CD41 monoclonal antibody (Abcam) at 37°C for 2 hour and then with horseradish peroxidase-conjugated streptavidin (R&D Systems) for 30 minutes. The linear absorbance was read at 450 nm by a microplate reader (Thermo Scientific) after the addition of substrate.

| ELISA quantification of annexin V-positive PMVs
Quantification of Annexin V-positive PMVs involved use of a homemade ELISA as described with minor modification. 16,20 The PFP was recalcified by addition of 50 mmol/L CaCl 2 immediately before laying into the wells of the plate previously coated with Annexin V (for capture of PS present on MVs). Then, the PMV detection was achieved by estimating the amount of CD41-positive MVs as described above.

| Platelet isolation
Platelets were prepared as previously described. 16 The healthy volunteers had not taken any medication for 14 days. Venous blood was collected in 0.109 mol/L sodium citrate under minimal tourniquet pressure using a sterile 22-gauge needle. Platelet-rich plasma (PRP) was obtained by centrifugation at 120 g for 10 minutes. Washed platelets were isolated from PRP after centrifugation at 800 g for 10 minutes in the presence of 100 nmol/L prostaglandin E 1 (PG-E 1, Sigma-Aldrich) and were resuspended in modified Tyrode's buffer [137 mmol/L NaCl, 2.7 mmol/L KCl, 12 mmol/L NaHCO 3 , 0.4 mmol/L NaH 2 PO 4 , 5 mmol/L HEPES, 0.1% glucose and 0.35% bovine serum albumin (BSA), pH7.2] with 100 nmol/L PG-E 1. Platelet concentration was adjusted to 1 × 10 6 /mL and then used at once.

| Isolation of MVs from platelet supernatant and plasma
For isolation of MVs from the plasma, PFP of untreated NVAF patients was centrifuged at 15 000 g for 1 hour in 4°C to precipitate MVs. Then, the washed MVs from NVAF patients were characterized by flow cytometry by staining with PEcy5-conjugated anti-CD41a antibody and FITC-conjugated Annexin V in binding buffer (10 mmol/L Hepes/NaOH, 140 mmol/L NaCl, 2.5 mmol/L CaCl 2 , pH 7.4) for 15 minutes at room temperature in the dark.

| Platelet functional studies
For MV stimulation experiments, CD36-positive or CD36-deficient platelet suspension (1 × 10 6 /mL) was stimulated with 30 µg/mL MVs from NVAF patients for various times as indicated and used for subsequent experiments.

| Flow cytometry of platelet activation and CD36 expression
Expression of platelet integrin α IIb β 3 , CD40L and CD36 was detected

| Immunoassay for soluble P-selectin
After the above treatment, the platelet-MV mixtures were centrifuged at 3000 g for 20 minutes twice. The last supernatant was used for detection of soluble P-selectin.

| Western blot analysis of phosphorylation of MKK4 and JNK
Cell lysates were subjected to immunoblotting as previously described 16    Note: Values are expressed as mean ± SD or number (%). Analyses done by chi-squared test (for categorical data) or one-way ANOVA (for continuous data) and post hoc least significant differences t test where appropriate.
Intergroup differences with least significant differences t test

| Statistical analysis
Data are expressed as mean ± SD (SE). Comparison among groups involved chi-squared test (for categorical data) or ANOVA with post hoc least significant differences or Dunnett's t test (for continuous data). P < .05 was considered statistically significant. Analyses involved use of SPSS 16.0 (SPSS Inc).

| Characteristics of NVAF patients and healthy controls: relationship to stroke risk stratified according to the CHADS2 score
Of the 210 NVAF patients, 113 (53.81%) were at high risk of stroke.
By definition of CHADS2 score, the high-risk NVAF patients were older and more likely to have congestive heart failure, hypertension, diabetes mellitus or history of stroke than those at low to moderate risk or healthy controls (P < .001). Furthermore, systolic blood pressure, fasting blood glucose, creatinine and white blood cell counts were higher in patients at high risk compared with controls (P < .01) ( Table 1).
The application of warfarin, calcium channel blockers, beta-receptor blockers and statins had no significant differences in the high-risk NVAF patients compared with those with low to moderate risk (P > .05); However, the patients in the high-risk group received more angiotensin-converting enzyme inhibitor (ACEI) or angiotensin receptor blocker (ARB) compared with those with low to moderate risk (P < .01); the healthy controls were free of any medication (

| Patients in the high-risk group had significantly higher levels of oxidative stress, inflammation, PMVs and platelet activation
Patients at high risk showed significantly higher levels of oxLDL, 8-isoPGF2α and IL-6 (P < .05) ( Table 1), which suggested that these patients had higher levels of oxidative stress and inflammation.
Furthermore, patients at high risk showed higher level of PMVs and Annexin V-positive PMVs compared with those at low to moderate risk (P < .05) (Figure 1). Platelet activation in patients was assessed by surface detection of integrin α IIb β 3 and degranulation of P-selectin, both of which were enhanced in the high-risk group compared with those at low to moderate risk (P < .05). As well, CD36 mean fluorescent intensity (MFI) in platelets was significantly increased in highrisk patients (P < .01) ( Table 1).

| Effect of MVs isolated from NVAF patients on platelet activation
We isolated MVs from the plasma of untreated NVAF patients at high risk of thromboembolism (AF-MVs) and from NVAF patients at low to moderate risk as control (C-MVs). The majority of MVs from high-risk patients were CD41a-and Annexin V-positive ( Figure 2).
We found enhanced expression of integrin α IIb β 3 and CD40L in normal platelets exposed to AF-MVs (P < .05). Indexes of platelet activation were increased but not significantly in platelets exposed to C-MVs ( Figure 3A-C).

| Activation of platelets by AF-MVs depends on CD36 and PS
We screened 2 CD36-deficient donors in the healthy volunteers by flow cytometry. The CD36-deficient platelets were unable to bind PE-conjugated anti-CD36 antibody ( Table 2).

| AF-MVs upregulate phosphorylation of platelet JNK2 and its upstream activator MKK4
Inhibition of JNK signalling with SP600125 blocked the effect of AF-MVs on platelet activation, including the expression of integrin

| D ISCUSS I ON
We demonstrated that MVs promoted platelet activation in the setting of NVAF. The engagement of MVs with platelet CD36 plays a critical role in facilitating platelet activation by phosphorylating MKK4/JNK2. Furthermore, we found that the CD36 might be one potential target for preventing prothrombotic state in NVAF. Similarly, we used Annexin V to mask surface PS exposed on AF-MVs and found the activation function of AF-MVs to depend on PS to some extent. Our work and that of others indicated that engagement of PSbearing MVs with platelet CD36 triggers platelet activation signals.
Then, we explored whether the signal pathway mediated by MV-CD36 could be a potential target for preventing the prothrombotic state of NVAF. Inhibition of the MKK/JNK pathway may induce unexpected complications because the activation of the MKK/JNK signal pathway regulates a range of biological processes. 24 In real world of both developed and developing countries, the adherence of NVAF patients to standardized anticoagulant therapy has been a great challenge. 25

| CON CLUS IONS
Our findings suggest that engagement of the MVs triggers platelet activation in a CD36-and PS-dependent manner in the setting of NVAF, and MKK4/JNK2 signalling pathway might be involved in this procedure. The platelet CD36 may be a potential target for preventing the prothrombotic state of NVAF.

CO N FLI C T O F I NTE R E S T
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Zhang and Ming Zhong designed the study and revised the article.

Hua
All the co-authors finally contributed to the final approval of the version to be published. All authors read and approved the final manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used to support the findings of this study are available from the corresponding author upon request.