Platelet SR-PSOX/CXCL16–CXCR6 Axis Influences Thrombotic Propensity and Prognosis in Coronary Artery Disease

Platelets express the transmembrane chemokine SR-PSOX/CXCL16, proteolytic cleavage of which generates the sCXCL16 soluble-(s) chemokine. The sCXCL16 engages CXCR6 on platelets to synergistically propagate degranulation, aggregation and thrombotic response. Currently, we have investigated the pro-thrombotic and prognostic association of platelet CXCL16–CXCR6 axis in CAD-(n = 240; CCS n = 62; ACS n = 178) patients. Platelet surface-associated-CXCL16 and CXCR6 surface expression ascertained by flow cytometry correlated significantly with platelet activation markers (CD62P denoting degranulation and PAC-1 binding denoting α2bβ3-integrin activation). Higher platelet CXCL16 surface association (1st quartile vs. 2nd–4th quartiles) corresponded to significantly elevated collagen-induced platelet aggregation assessed by whole blood impedance aggregometry. Platelet-CXCL16 and CXCR6 expression did not alter with dyslipidemia, triglyceride, total cholesterol, or LDL levels, but higher (>median) plasma HDL levels corresponded with decreased platelet-CXCL16 and CXCR6. Although platelet-CXCL16 and CXCR6 expression did not change significantly with or correlate with troponin I levels, they corresponded with higher Creatine Kinase-(CK) activity and progressively deteriorating left ventricular ejection fraction (LVEF) at admission. Elevated-(4th quartile) platelet-CXCL16 (p = 0.023) and CXCR6 (p = 0.030) measured at admission were significantly associated with a worse prognosis. However, after Cox-PH regression analysis, only platelet-CXCL16 was ascertained as an independent predictor for all-cause of mortality. Therefore, the platelet CXCL16–CXCR6 axis may influence thrombotic propensity and prognosis in CAD patients.


Platelets may Contribute to Circulatory sCXCL16 and Engage Circulatory sCXCL16
We enrolled n = 62 CCS, and n = 178 ACS patients in the current clinical cohort of CAD (n = 240) patients. Baseline characteristics of the enrolled patients are presented in Table 1.   Platelet surface-associated CXCL16 (platelet transmembrane-SR-PSOX/CXCL16 and platelet surface-associated circulatory sCXCL16 combined, as detected by flow cytometry and hereafter referred to as platelet-CXCL16) showed a strong and significant positive correlation with platelet surface expression of CXCR6 (ρ = 0.800, p < 0.001). This suggested that much of platelet surface-associated CXCL16 may be circulatory sCXCL16 sequestered by ligation to platelet CXCR6, its cognate receptor. We also evaluated serum levels of sCXCL16 (as assessed in the PLATO study [22] and HUNT cohort [23]) in CAD patients. However, neither platelet-CXCL16 nor platelet CXCR6 showed evident correlation with serum sCXCL16, denoting that serum or circulatory sCXCL16 may be derived from other cellular sources like leukocytes and the vascular endothelium, in addition to circulating platelets. Platelet-CXCL16 and CXCR6 predictably correlated significantly with mean platelet volume (MPV) (ρ = 0.357, p < 0.001 and ρ = 0.373, p < 0.001, respectively) ( Figure 1A,B). Moreover, we observed a significant correlation between MPV and sCXCL16 (ρ = 0.134, p = 0.043), which might suggest platelets as a potential source of circulatory sCXCL16; but at the same time, a significant inverse correlation (ρ = −0.149; p = 0.025) between platelet count and sCXCL16 ( Figure 1C) could suggest that platelets may substantially sequester sCXCL16 available in circulation.
Results from experimental in vitro studies suggest the involvement of a positive feedback loop mediated through ADP in sCXCL16-platelet CXCR6 exerted pro-thrombotic effects, which are counteracted in the presence of apyrase and P2Y 12 antagonist [8]. Therefore, we ascertained the influence of P2Y 12 antagonists on platelet-CXCL16, CXCR6 and serum sCXCL16 levels in the current clinical cohort. Neither ticagrelor nor clopidogrel administration showed any significant impact on platelet-CXCL16 (p = 0.443, p = 0.224, respectively) and platelet CXCR6 (p = 0.132, p = 0.070, respectively) surface expression. Among ticagrelor-administered patients, platelet-CXCL16 was median MFI (25th; 75th percentile) = 45. 15   . Correlation between serum sCXCL16 and circulatory platelet count (Plts = Platelet count × 10 3 /µL) in CAD patients. Correlations were assessed by Spearman's rank correlation coefficient (ρ); p: level of significance; MFI: mean fluorescence intensity. (C). Schematic diagram depicting platelets as a potential source of sCXCL16 generated by the proteolytic cleavage of the transmembrane-SR-PSOX/CXCL16 by ADAM10. The sCXCL16 may engage its cognate receptor CXCR6 on platelets to elicit pro-thrombotic effects.

Platelet CXCL16-CXCR6 Axis may Influence Pro-Thrombotic Disposition in CAD Patients
Previously we have documented a synergistic impact of recombinant-sCXCL16 on platelet-driven thrombotic response by acting through CXCR6 in experimental studies with human (in vitro) and murine systems (arterial thrombosis model in vivo) [8]. In the current cohort of CAD patients, we observed a strong correlation between platelet- Correlations between platelet-CXCL16, platelet-CXCR6, serum levels of soluble CXCL16 in CAD patients, and Mean platelet volume (MPV). (Biv). Correlation between serum sCXCL16 and circulatory platelet count (Plts = Platelet count × 10 3 /µL) in CAD patients. Correlations were assessed by Spearman's rank correlation coefficient (ρ); p: level of significance; MFI: mean fluorescence intensity. (C). Schematic diagram depicting platelets as a potential source of sCXCL16 generated by the proteolytic cleavage of the transmembrane-SR-PSOX/CXCL16 by ADAM10. The sCXCL16 may engage its cognate receptor CXCR6 on platelets to elicit pro-thrombotic effects.

Influence of Dyslipidemia and Plasma Lipids on Platelet CXCL16-CXCR6 Axis in CAD Patients
Although sCXCL16 is a chemokine, transmembrane-SR-PSOX/CXCL16 is a scavenger receptor for oxLDL [13,16]; the expression of this in cellular sources like endothelium [14,26], smooth muscle cells [14,15], macrophages [17] and platelets [8,9] might be influenced by circulating lipids or lipoproteins. Contrary to our expectations, in the current CAD-cohort, dyslipidemia ( Figure   . Box plots depicting collagen-induced platelet aggregation response in CAD patients corresponding to relative levels of lower (1st quartile) and higher (2nd-4th quartile) platelet-CXCL16, platelet-CXCR6, and serum levels of sCXCL16. Data represent median with 95% CI and statistical significance calculated by the Mann-Whitney U test; p shows the significance level.

Association of Platelet CXCL16-CXCR6 Axis and Cardiovascular Risk Factors in CAD
Next, we verified the influence of additional major cardiovascular risk factors like the inflammatory mediator hsCRP ( Figure 4A), diabetes ( Figure 4B), eGFR ( Figure 4C) and arterial hypertension ( Figure 4D) on platelet CXCL16-CXCR6 axis as compared to serum sCXCL16. Platelet-CXCL16 showed no significant difference (p = 0.824) between hs-CRP > median and hs-CRP ≤ median subgroups (Figure 4Ai)  Diabetic patients in the current cohort ( Previous studies have shown that serum sCXCL16 concentration negatively correlates with glomerular filtrate rate (eGFR), the rate of creatinine clearance and levels of blood albumin; however, it correlates positively with proteinuria, blood urea nitrogen, and is therefore indicative of renal injury in T2DM patients [27]. Therefore, we evaluated the effect of deteriorating eGFR on the platelet CXCL16-CXCR6 axis as compared to serum sCXCL16 (Figure 4Ci (Figure 4Ci). Confirming previous reports [27] and adding the information on platelet-CXCL16 and CXCR6, we observed a significant correlation of all three with eGFR ( Table 2).

Prognostic Association of Platelet CXCL16-CXCR6 in CAD Patients
Since we observed an association between platelet-CXCL16, platelet CXCR6 and CK levels, we validated their association with LVEF in patients (Table 3). Higher levels of both platelet-CXCL16 and CXCR6 corresponded with a decline in LVEF measured at admission among CAD patients. For platelet-CXCL16 ( Figure 6Ai) the values were median platelet-  (Table 2). Similarly, platelet CXCR6 levels were also significantly increased (Figure 5Dii) (Figure 5Diii).

Prognostic Association of Platelet CXCL16-CXCR6 in CAD Patients
Since we observed an association between platelet-CXCL16, platelet CXCR6 and CK levels, we validated their association with LVEF in patients (Table 3). Higher levels of both platelet-CXCL16 and CXCR6 corresponded with a decline in LVEF measured at admission among CAD patients. For platelet-CXCL16 ( Figure 6Ai)  We could evaluate the deterioration course of LVEF in ACS patients after a follow-up period of 3 years (Figure 6Bi (Figure 6Bii). However, we did not observe a direct correlation between platelet-CXCL16, platelet-CXCR6 measured at baseline, and LVEF% monitored after 3 years (Table 2). No significant differences (p = 0.443) or correlations (  (Figure 6Biii). Furthermore, we ascertained the prognostic relevance of platelet CXCL16-CXCR6 axis as compared to serum sCXCL16 levels on all-cause death in CAD patients (n = 217). Higher baseline expression of platelet-CXCL16, platelet-CXCR6, and serum sCXCL16 were all associated with poor prognosis of CAD patients ( Figure 6C), as seen in the Kaplan-Maier curves. Baseline characteristics of the complete study cohort stratified according to prognosis are presented in Table 3.
As shown in Cox PH regression analysis in Table 4, platelet-CXCL16 (p = 0.039) was validated as an independent risk factor in influencing prognosis for all-cause of mortality. Initially, we performed a multivariable analysis of the covariables with age and LG troponin I remained independently associated with all-cause mortality. A further variable selection step was performed for the three CXC (serum sCXCL16, platelet-CXCL16, platelet-CXCR6) chemokine-chemokine receptor variables, keeping age and LG troponin I in the analysis.
The only variable significant was platelet CXCL16. However, the power of the study was not large enough to establish that platelet CXCL16 is superior to CXCR6.

Discussion
The current clinical investigation was undertaken to explore the pathophysiological significance of the platelet-CXCL16-CXCR6 axis [8] in influencing thrombotic disposition and its prognostic association in CAD patients (Figure 7). This investigation might be considered a translational extension of our previous experimental study, which documented a synergistic impact of recombinant-sCXCL16 on platelet-driven thrombotic responses that are executed through CXCR6 [8] and delineated the molecular mediators downstream. We validated these observations in experiments with human platelets (in vitro) and a murine model of arterial thrombosis in vivo. The currently observed correlation between platelet-CXCL16 and CXCR6 surface expression with markers of platelet activation validates our experimental finding that ligation of CXCR6 by sCXCL16 may support a pro-thrombotic mode of action in pathophysiological settings like CAD. Since serum levels of free sCXCL16 did not correlate with platelet activation or aggregatory potential, the surface association of sCXCL16 on platelets seems critical in mediating the pro-thrombotic effects through CXCR6 ligation. In potential pathological settings, elevated circulatory sCXCL16 or CXCL16 deposited at atherosclerotic lesions and plaques [6,7,28,29] may exaggerate platelet aggregation by engaging platelet CXCR6 [28].
A significant correlation between platelet-CXCL16 and CXCR6 and a negative correlation between platelet count and platelet-free serum levels of sCXCL16 as currently observed also suggests that platelets may sequester circulatory sCXCL16, levels of which are elevated in ACS patients [23,28]. Platelets in their activated state may also shed transmembrane-SR-PSOX/CXCL16, thereby adding to the circulatory levels of sCXCL16 derived from other vascular cells and leukocytes. We compared the levels of platelet surface-associated CXCL16, surface-expressed CXCR6 and serum sCXCL16 with respect to treatment with P2Y 12 antagonists clopidogrel and ticagrelor. Although none of the P2Y 12 antagonists showed any impact on either platelet-CXCL16 or CXCR6, ticagrelor significantly reduced serum sCXCL16 levels, which might be attributed to the counteraction of transmembrane-SR-PSOX/CXCL16 shedding from activated platelets during blood clotting. It is worth considering that the pro-thrombotic effects of sCXCL16 mediated through platelet CXCR6 are antagonized by ADP degrading apyrase and ADP receptor P2Y 12 antagonists [8]. The currently observed effects of ticagrelor add to the potential benefits of P2Y 12 antagonists in reducing pro-thrombotic sCXCL16 levels in CAD patients.
Contrary to our expectations, platelet-CXCL16, CXCR6 or serum sCXCL16 levels were not influenced by dyslipidemia to a significant extent or even by individual lipid (total cholesterol, triglyceride) or lipoprotein (LDL) profiles, all of which are major risk factors for ACS [30]. On the other hand, as HDL exerts some regulatory impact on platelet activation and aggregation [31,32], higher plasma HDL levels corresponded with significantly reduced platelet-CXCL16 and CXCR6 surface expression. Among other cardiovascular risk factors, only sCXCL16 levels were affected considerably by inflammatory mediator hsCRP, the concurrence of diabetes, and a decline in eGFR. Hypertension seemingly did not affect platelet-CXCL16, serum sCXCL16 or CXCR6 surface expression on platelets.
Platelet-CXCL16 and CXCR6 did not alter significantly between CCS and ACS patients. However, our data confirmed previous findings that serum sCXCL16 levels are significantly enhanced in ACS patients [23]. Both TnI and CK are indicators of cardiac injury. Platelet-CXCL16 and CXCR6 did not show any association with TnI, but higher levels of both corresponded with increased CK. Elevated serum sCXCL16 levels in the current cohort corresponded with peak TnI levels in accordance with previous reports [19,23,28]. Platelet-CXCL16 and CXCR6 levels were relatively higher in patients with worsened LVEF ascertained during the hospital stay and a follow-up measurement after three years, suggesting its potential association with myocardial dysfunction. Kaplan-Meier analyses of data from the current cohort showed a significant association between platelet-CXCL16, platelet CXCR6, sCXCL16, and all-cause of mortality in the three years of the follow-up period, but not explicitly for MI. We further analyzed the differences in clinical and analytical/test variants between alive and deceased patients in our cohort. Based on multivariable analyses after adjusting for clinical variables, i.e., age, NYHA class, LVEF% upon admission, hs-CRP and Troponin I, which were significantly different between alive and deceased patients, only platelet-CXCL16 remained as an independent prognostic indicator for all causes of mortality in the current cohort. Our investigation does not undermine the effectiveness of sCXCL16 as a predictive marker of MI, but it puts forth the association between platelet-CXCL16 and prognosis for all-cause of mortality in CAD patients. In the subgroup analysis of the 5142 patients randomized in the PLATO trial [22], clinical outcome was specifically ascertained in ACS patients as compared to CAD (including CCS and ACS) patients in our cohort. The PLATO investigators considered sudden MI or stroke and composite of cardiovascular death as clinical endpoints. The investigators had also concluded that serum CXCL16 measured upon admission was independently associated with cardiovascular death but not explicitly with ischemic events [22]. Investigators of the HUNT2 cohort in Norway estimated the prognostic effectiveness of sCXCL16 as a potential risk factor for MI in a large population of 58,761, who were followed for the first incidence of MI over a period of 11.3 years, and registered 1587 incidents. Comparing sCXCL16 values in MI patients to 3959 age-and sex-matched controls, they observed only subtle differences in sCXCL16 levels; nevertheless, there was a substantially increased risk of MI amongst those in the highest quartile [23].
Conclusions: The major findings of the present study are that: (i) Platelet surfaceassociated CXCL16 and CXCR6 surface expression are significantly associated with activation status (α-granule degranulation and α 2b β 3 -integrin activation) of platelets in circulation; (ii) Higher platelet-CXCL16 and CXCR6 surface expression also corresponded with a higher degree of activation potential in platelets ascertained by collagen-induced platelet aggregation response ex vivo. Engagement of circulatory sCXCL16 to platelet surface expressed CXCR6 is seemingly essential to trigger the activatory signaling cascade to drive platelet activation [8]. Therefore, platelet-free serum levels of sCXCL16 did not correlate with platelet activation markers or aggregatory potential; (iii) Although platelet-CXCL16, CXCR6 expression did neither change significantly between ACS and CCS patients, nor correlated significantly with Tn I levels, increased platelet-CXCL16 and CXCR6 corresponded with higher CK and progressively deteriorated LVEF; (iv) Consequently, elevated baseline platelet-CXCL16 and CXCR6 were significantly associated with time to all-cause of mortality/death (Kaplan-Meier curve); (v) In multivariable regression analysis, platelet-CXCL16 was found to be independently associated with time to all-cause of mortality.
Study limitations: The current investigation has several limitations: (i) We could measure the laboratory parameters, i.e., platelet-CXCL16, CXCR6 and sCXCL16, only upon admission, which leaves potential alterations in these parameters over time (3 years of follow-up period) and the impact of subsequent anti-platelet therapy to speculation. We could, however, evaluate LVEF in ACS patients after the follow-up period; (ii) 23 out of 240 patients (9.6%) in the overall CAD cohort were lost to follow-up; (iii) The size of the current cohort was relatively small (n = 240), being a single-center study, as compared to a large number of patients enrolled in the PLATO subgroup analysis [22] and HUNT2 cohort [23]; (iv) A relative comparison of platelet-CXCL16, platelet CXCR6 and serum sCXCL16 levels in an age-matched non-CAD control group is currently lacking; (v) We did not offer any mechanistic basis behind the observations from the clinical cohort, but have previously reported the molecular mediators in the platelet CXCL16-CXCR6 axis that can regulate thrombotic response [8,9]. Nevertheless, the results from this cohort present a novel translational aspect on the relevance of the platelet CXCL16-CXCR6 axis in influencing thrombotic propensity and prognosis in CAD patients; this will encourage more extensive clinical investigations in the future. the current cohort was relatively small (n = 240), being a single-center study, as compared to a large number of patients enrolled in the PLATO subgroup analysis [22] and HUNT2 cohort [23]; (iv) A relative comparison of platelet-CXCL16, platelet CXCR6 and serum sCXCL16 levels in an age-matched non-CAD control group is currently lacking; (v) We did not offer any mechanistic basis behind the observations from the clinical cohort, but have previously reported the molecular mediators in the platelet CXCL16-CXCR6 axis that can regulate thrombotic response [8,9]. Nevertheless, the results from this cohort present a novel translational aspect on the relevance of the platelet CXCL16-CXCR6 axis in influencing thrombotic propensity and prognosis in CAD patients; this will encourage more extensive clinical investigations in the future. Figure 7. Schematic diagram showing that platelet surface associated CXCL16 levels are enhanced in CAD patients with a decline in LVEF, and so is the surface expression of CXCR6. Proteolytic cleavage of transmembrane-SR-PSOX/CXCL16 from activated platelets generates circulatory sCXCL16, levels of which are also elevated in CAD patients. This effect is apparently counteracted following ticagrelor administration. Circulatory sCXCL16 may engage CXCR6 on the platelet surface and trigger the activatory signaling cascade involving PI3K-Akt, leading to dense granule release of ADP and a profound pro-thrombotic drive. ADP engagement of P2Y12 and a subsequent positive feedback loop is therefore counteracted by ADP degrading apyrase and P2Y12 antagonist clopidogrel, suggesting the clinical significance of platelet CXCL16-CXCR6 axis.

Materials and Methods
Materials: Mouse anti-human CXCL16-PE, mouse anti-human CXCR6-PE, and human CXCL16 Quantikine ELISA kit were procured from R&D systems; anti-human CD62P-FITC and CD42b-FITC were from Beckman Coulter; anti-human CD42b-PE, PAC-1-FITC were procured from BD Biosciences. Reagents for platelet aggregation tests in the Figure 7. Schematic diagram showing that platelet surface associated CXCL16 levels are enhanced in CAD patients with a decline in LVEF, and so is the surface expression of CXCR6. Proteolytic cleavage of transmembrane-SR-PSOX/CXCL16 from activated platelets generates circulatory sCXCL16, levels of which are also elevated in CAD patients. This effect is apparently counteracted following ticagrelor administration. Circulatory sCXCL16 may engage CXCR6 on the platelet surface and trigger the activatory signaling cascade involving PI3K-Akt, leading to dense granule release of ADP and a profound pro-thrombotic drive. ADP engagement of P2Y 12 and a subsequent positive feedback loop is therefore counteracted by ADP degrading apyrase and P2Y 12 antagonist clopidogrel, suggesting the clinical significance of platelet CXCL16-CXCR6 axis.

Materials and Methods
Materials: Mouse anti-human CXCL16-PE, mouse anti-human CXCR6-PE, and human CXCL16 Quantikine ELISA kit were procured from R&D systems; anti-human CD62P-FITC and CD42b-FITC were from Beckman Coulter; anti-human CD42b-PE, PAC-1-FITC were procured from BD Biosciences. Reagents for platelet aggregation tests in the Multiplate ® analyzer were procured from Roche Diagnostics. Human metabolic panel 1 Legendplex ® cytometric bead array kit for adipokines was from Biolegend.