Differential Inhibition of Human Atherosclerotic Plaque–Induced Platelet Activation by Dimeric GPVI-Fc and Anti-GPVI Antibodies

Methods

Atherosclerotic plaques were obtained from patients undergoing endarterectomy for high-grade carotid artery stenosis. Patient informed consent was obtained, as approved by the Ethics Committee of the Faculty of Medicine of the University of Munich in accordance with the ethical principles for medical research involving human subjects as set out in the Declaration of Helsinki.

The carotid plaque tissue was endarterectomized, processed, and preserved as described ^{3, 25}. Plaque homogenates from 5 patients were mixed to obtain plaque pools that were kept in aliquots at −80°C. Plaque homogenates were used for platelet aggregation studies or coated onto glass coverslips for flow studies ^{3, 26, 27}.

We stimulated blood with plaque homogenates containing all potential thrombogenic compounds or Horm collagen. Plaque contains mainly type I and III collagens ^{5, 6}. Horm collagen consists of collagen type I, as suggested by the supplier, but also collagen type III (W. Siess, unpublished observations, October, 2013). In addition to experiments with blood pre-incubated with GPVI-Fc and subsequent plaque stimulation, we performed experiments with plaque pre-incubated with 35-fold or 50-fold higher concentrations of GPVI-Fc than finally present in blood to maximally saturate the collagen-binding sites for GPVI.

For additional details, see the Online Appendix.

Results

Aggregation measurements showed that GPVI-Fc, but not Fc lacking the external GPVI domain, delayed plaque- and collagen-stimulated platelet aggregation in blood (Figure 1A). In plaque-stimulated samples, the lag time until the start of aggregation increased from 65 ± 19 s with Fc control protein to 119 ± 23 s (n = 6; p < 0.001) with GPVI-Fc (50 μg/ml, 300 nM), when the proteins were pre-incubated with blood and from 72 ± 20 s with Fc control protein to 134 ± 25 s (n = 6; p < 0.001) with GPVI-Fc, when the proteins were mixed with plaque before blood stimulation. Inhibition was specific because GPVI-Fc did not affect platelet aggregation when stimulated with adenosine diphosphate (ADP) and thrombin receptor-activating peptide (Figure 1B). Collagen-stimulated platelet aggregation was dose dependently reduced (maximally by −45%) (Figure 1C) as was plaque-induced platelet aggregation by blood pre-treatment with GPVI-Fc (maximally by −51%) (Figure 1D). Surprisingly, inhibition was not significantly enhanced if plaque was pre-incubated with a 35-fold higher GPVI-Fc concentration before stimulation of the blood samples (Figure 1E), even if pre-incubation time was extended to 30 min (Online Figure 1).

The anti–GPVI antibodies BLO8-1 (10 μg/ml, 833 nM) and 5C4 (1.25 μg/ml, 25 nM) almost completely inhibited plaque- and collagen-induced platelet aggregation in a concentration-dependent manner (Online Figures 2A and 2B, and not shown). The highest concentration of BLO8-1 decreased aggregation to 12% of control (n = 9) after plaque stimulation and to 16% (n = 8) after collagen stimulation. Residual aggregation after pre-incubation with the highest 5C4 concentration was 7% on plaque stimulation (n = 5) and 18% on collagen stimulation (n = 5). Inhibition was specific because neither BLO8-1 nor 5C4 affected platelet aggregation when stimulated by ADP and thrombin receptor–activating peptide (Online Figure 2C).

Because dimeric GPVI on resting platelets is essential for collagen binding and platelet activation (16), we performed experiments with m-Fab-F directed against dimeric GPVI ^{16, 18} and compared it with 5C4, which blocks monomeric and dimeric GPVI. The m-Fab-F inhibited plaque-induced platelet aggregation less effectively than 5C4 (−64 ± 11% vs. −86 ± 8%; p < 0.05). Inhibition of plaque-induced platelet aggregation by dimeric GPVI-Fc was −53 ± 17% (Online Figure 3). 5C4 inhibits platelet aggregation with a half maximal inhibitory concentration (IC₅₀) of ∼0.2 μg/ml, corresponding to a dissociation constant (K_D) of ∼1 nM, whereas m-Fab-F has a reported K_D for GPVI dimer of ∼10 nM. However, although m-Fab-F binding to GPVI dimer is saturable, lower maximal available binding sites (B_max) were reached using m-Fab-F than using other antibodies (16), indicating that m-Fab-F does not bind to all GPVI dimers present on the platelet surface.

To simulate plaque rupture and subsequent platelet activation, human whole blood was perfused in a parallel plate flow chamber over human plaque homogenate at different arterial shear rates: 550/s and 1,100/s are within the range of physiological mean and peak wall shear rates of carotid and coronary arteries ^{28, 29}, and shear rates of ∼1,500/s prevail over mildly stenotic coronary lesions.

The fluorescence micrographs in Figures 2A and 2B and diagrams in Figure 2C (quantifying the area covered with platelets over time) show inhibition of plaque-induced platelet deposition by GPVI-Fc, BLO8-1, and 5C4 at different arterial shear rates. Platelet coverage tested at full minutes for all treatments and shear rates of 550/s and 1,100/s and for GPVI-Fc versus control for shear rates of 550/s, 1,100/s, and 1,500/s by 3-way analysis of variance was significant for factors treatment (p < 0.001), shear (p < 0.05), time (p < 0.001), and the interaction of treatment with shear (p < 0.05) and time (p < 0.001). GPVI-Fc (50 μg/ml) significantly delayed and reduced plaque-induced platelet aggregation compared with controls (Figures 2A and 2C, Online Videos 1 and 2). Because the limited inhibition at the shear rate of 550/s might be explained by subsaturating blood concentrations of GPVI-Fc not blocking all tandem GPO motifs in plaque collagen, we pre-incubated plaque-coated coverslips with 50-fold higher GPVI-Fc concentrations (than reached after GPVI-Fc addition to blood) before low shear rate flow blood perfusion. As in the static experiments (Figure 1, Online Figure 1), inhibition by GPVI-Fc was not increased (Online Figure 4).

Interestingly, inhibition by GPVI-Fc increased with shear rate. At 1,500/s, GPVI-Fc effectively inhibited plaque-induced platelet aggregation (Figure 2C, bottom).

The anti-GPVI antibodies BLO8-1 (20 μg/ml) and 5C4 (1.25 μg/ml) almost completely inhibited platelet aggregate formation at shear rates of 550/s and 1,100/s. Only platelet adhesion was observed, which was predominantly transient (Figure 2, Online Video 3).

Discussion

The crucial role of GPVI-collagen interaction in plaque rupture- or erosion-triggered thrombotic occlusion stimulated the generation of specific inhibitors promising benefit beyond established dual-antiplatelet therapy. To help define the most effective and safest anti-GPVI strategy, we compared compounds targeting GPVI-collagen interaction in various tests of platelet response to plaque material under static and arterial flow conditions. We found compound-specific differences relevant to clinical trial design.

The antibodies recognizing platelet monomeric and dimeric GPVI (BLO8-1, 5C4) inhibited collagen- and plaque-induced platelet aggregation almost completely under static and flow conditions at low and high shear rates. Although preserved, platelet adhesion was transient rather than stable. Inhibition was less with GPVI-Fc. Dimeric GPVI is essential for binding to collagen and activation of platelets ^{16, 17, 18}, and GPVI-Fc was constructed to mimic dimeric GPVI. It binds to collagen with high affinity, thus concealing GPVI-binding sites from platelets. Our results of collagen- and plaque-induced static platelet aggregation, in which the integrin α₂β₁- and the vWF-binding sites of collagen do not play a role in platelet activation, indicate that not all tandem GPO motifs on collagen were occupied by GPVI-Fc, even after pre-incubation with very high GPVI-Fc concentrations. These few unoccupied GPO sites may suffice to trigger GPVI-induced signaling and subsequent ADP and TxA₂ release, which then autoamplifies platelet activation under static and low shear rate flow conditions. Indeed, we have previously shown that in static assays plaque-induced aggregation is largely dependent on dense granule secretion that is abolished by a combination of platelet cyclooxygenase inhibition and P2Y₁₂-receptor blockade ^{26, 27}.

How does soluble dimeric GPVI-Fc differ from platelet GPVI dimer? Plasmon surface resonance yielded a low dissociation constant of collagen GPVI-Fc binding (K_D = 1.17 × 10⁻⁷ M), as reported previously (17). In line with the high affinity of GPVI-Fc to collagen, fluorescent-labeled GPVI-Fc added to flowing blood rapidly bound to plaque. This binding was stable, indicating a very low off rate of GPVI-Fc. Differences between soluble GPVI-Fc and platelet GPVI dimer must be due to factors other than affinity. First, GPVI dimers are recruited from monomers during platelet activation and cluster in lipid rafts leading to a high local GPVI-dimer density on the platelet surface ^{15, 16, 31}. Second, spacing of the 2 collagen binding sites on dimeric GPVI-Fc lacking the mucin-like stems of the platelet GPVI dimer might be less flexible and may not always fit with the GPO repeats exposed on plaque collagen (Central Illustration A, left). This might explain why even pre-incubation of plaque with very high GPVI-Fc concentrations did not enhance platelet inhibition under static and low shear rate flow conditions (Figure 1, Online Figures 1 and 4). Third, the superior suppression of plaque-induced platelet aggregation by anti-GPVI antibodies recognizing both monomeric and dimeric platelet GPVI suggests that monomeric GPVI is functionally important. By also binding to monomers, these antibodies might impair dimer recruitment during platelet activation ^{15, 16} (Central Illustration B).

Our imaging studies demonstrate that GPVI-Fc bound equally well to plaque collagen at high and low shear rates, both up- and downstream of plaque fragments. However, plaque collagen sites not occupied by GPVI-Fc arrested platelets mainly under low shear rate flow, when flow niches downstream of plaque fragments might create nearly static conditions. On the basis of our imaging studies and ADP/TxA₂ inhibition experiments, we suggest that GPVI signaling in a few platelets attaching downstream of the plaque induces stable adhesion and recruitment of circulating platelets into aggregates by the local release of TxA₂ and ADP (Central Illustration A, top). Our results with low threshold-inhibitory concentrations of 5C4 further support such a mechanism. At high flow velocities, ADP and TxA₂ will be flushed away, contributing to the increased effectiveness of GPVI-Fc at higher shear rates (Central Illustration A, bottom). An additional mechanism may be GPVI-Fc occupying vWF binding sites of collagen, which would inhibit vWF-mediated platelet adhesion relevant at high shear (32).

Surprisingly, the subtle (<8 μm) roughness of the plaque surface in our model created local differences in dynamics and extent of plaque-induced platelet aggregate formation at low and high shear rates in the presence of GPVI-Fc. Coronary thrombosis mostly arises from rupture of thin-capped (<65 μm) fibroatheroma, exposing the plaque necrotic core containing collagenous structures to circulating blood ^{6, 33, 34}. This creates a new thrombogenic and rough luminal surface that will influence the dynamics of platelet adhesion and aggregation, probably similar to our ex vivo plaque model. Although nonstenotic coronary lesions with physiological wall shear rates (500/s to 600/s) may rupture, coronary atheromas with the highest risk to cause myocardial infarction are at least 50% stenotic with shear rates of ∼1,500/s (35). Rupture frequently occurs at the inflamed macrophage-rich plaque shoulder or at the most stenotic part, where endothelial shear stress is highest (36). The potent platelet inhibition by GPVI-Fc at high shear suggests locally enhanced antithrombotic efficacy of GPVI-Fc at ruptured high-risk lesions.

Given its action at the ruptured plaque and its flow-dependent platelet inhibition, GPVI-Fc is expected to be safer than GPVI antibodies for systemic bleeding risk. GPVI antibodies target all circulating platelets as do established antiplatelet drugs such as aspirin and P2Y₁₂ antagonists, which increase bleeding. In fact, GPVI-Fc alone showed no increased bleeding in a phase I study (21). Under static and low shear rate flow conditions, GPVI-Fc, unlike GPVI antibodies, inhibited plaque- and collagen-initiated platelet aggregation only moderately. Because major bleeding complications arise mainly from vessels with lower flow velocities and shear rates than coronary arteries, safety concerns of triple platelet inhibition would appear lower with GPVI-Fc than with GPVI antibodies. Indeed, in mice, GPVI-Fc did not worsen bleeding time when combined with established common antithrombotic drugs such as aspirin, P2Y₁₂ antagonists, and heparin alone or in combination (37).

Anti-GPVI antibodies causing systemic platelet inhibition could increase bleeding. Patients with rare genetic or acquired GPVI defects show a variable bleeding diathesis depending on clinical background (14). Although the tail bleeding time of GPVI-deficient mice is only moderately increased (21), depletion of platelet GPVI by antibody treatment severely compromised hemostasis in mice with concomitant aspirin therapy (38). Additionally, anti-GPVI antibodies can cause immune responses and long-lasting platelet GPVI depletion through GPVI shedding or other mechanisms (14). It is not yet known whether BLO8-1 and 5C4 have these effects. BLO8-1 is a human domain antibody (the smallest functional binding units of human Igs) that is less likely to be immunogenic and more likely to resist proteolytic degradation.

Conclusions

Antibodies targeting platelet dimeric and monomeric GPVI are more effective inhibitors of plaque-induced platelet aggregation in static and dynamic models than GPVI-Fc masking GPVI-binding motifs on exposed plaque collagen. However, GPVI-Fc inhibition is flow dependent and increases with shear rate. This suggests a focused antithrombotic efficacy at rupture sites of high-risk lesions. Under low shear rate flow, GPVI-Fc synergizes with aspirin and P2Y₁₂ antagonists in platelet inhibition. Thus, compounds targeting the GPVI-collagen axis have antiatherothrombotic potential beyond standard dual-antiplatelet therapy, and GPVI-Fc might be safer than antibodies directed against GPVI.

Appendix