Fibrin and D-dimer bind to monomeric GPVI

Abstract

Fibrin has recently been shown to activate platelets through the immunoglobulin receptor glycoprotein VI (GPVI). In the present study, we show that spreading of human platelets on fibrin is abolished in patients deficient in GPVI, confirming that fibrin activates human platelets through the immunoglobulin receptor. Using a series of proteolytic fragments, we show that D-dimer, but not the E fragment of fibrin, binds to GPVI and that immobilized D-dimer induces platelet spreading through activation of Src and Syk tyrosine kinases. In contrast, when platelets are activated in suspension, soluble D-dimer inhibits platelet aggregation induced by fibrin and collagen, but not by a collagen-related peptide composed of a repeat GPO sequence or by thrombin. Using surface plasmon resonance, we demonstrate that fibrin binds selectively to monomeric GPVI with a K_D of 302 nM, in contrast to collagen, which binds primarily to dimeric GPVI. These results establish GPVI as the major signaling receptor for fibrin in human platelets and provide evidence that fibrin binds to a distinct configuration of GPVI. This indicates that it may be possible to develop agents that selectively block the interaction of fibrin but not collagen with the immunoglobulin receptor. Such agents are required to establish whether selective targeting of either interaction has the potential to lead to development of an antithrombotic agent with a reduced effect on bleeding relative to current antiplatelet drugs.

Graphical abstract

Figure 1.

GPVI-deficient patients are unresponsive to fibrin. Glass coverslips were coated with collagen or fibrin as described in "Materials and methods." (Ai) Human platelets (ptl; 20 × 10⁹/L) were allowed to spread on coated coverslips, followed by staining of actin with Alexa-488 phalloidin. The pictures are representative results of a control and 2 GPVI-deficient patients. Scale bar, 5 μm. (Aii) The histograms illustrate the quantification of the surface area of platelets and the number of platelets calculated per millimeter squared. In each independent experiment, 5 random pictures were analyzed (100 platelets in total). The results are shown as mean ± standard deviation. ***P < .001. (B) Representative aggregation traces form a GPVI-deficient patient are shown (the control aggregation trace is representative of >20 donors).

Figure 2.

Fibrin induces Syk-dependent platelet activation and aggregation. (A) Platelets (500 × 10⁹/L), pretreated where indicated with dasatinib (dasat; 3 μM) or PRT-060318 (10 μM), were stimulated with fibrin dispersed as a suspension. (B) The histogram represents 3 independent experiments. The results are shown as mean ± standard deviation. **P < .01. (C) Stimulations were stopped with addition of 2× lysis buffer. A sample of the whole-cell lysate (WCL) was removed, and the remaining lysate was used to immunoprecipitate (IP) Syk. WCLs and IPs were separated by SDS-polyacrylamide gel electrophoresis and western blotted (WB) for pTyr and Syk. The results are shown as a representative of 3 independent experiments.

Figure 3.

Monomeric GPVI but not dimeric GPVI binds to fibrin. SPR was performed using OneStep titration as described in "Materials and methods." (A) Representative SPR binding curve using 1 μM of GPVI. The calculated K_D was 302 ± 5 nM. Solid-based binding assay was performed in nunc maxisorb 96-well plates coated overnight with 10 μM of bovine serum albumin (BSA), collagen, fibrinogen, or fibrin. (B) Monomeric (red bars) or dimeric (blue bars) GPVI (100 nM) was incubated as described. (C) Revacept, a dimeric GPVI, was allowed to bind to 10 μM of BSA, collagen, human fibrinogen, fibrin, D-dimer, or fragment E (Frag E). (D) Monomeric GPVI was allowed to bind to 10 μM of human fibrinogen, fibrin, D-dimer, or fragment E. Bound GPVI was detected using HRP coupled to an anti-6×His monoclonal antibody for monomeric GPVI or an anti-human immunoglobulin for dimeric GPVI and Revacept. The histograms (mean ± standard deviation) show the results from 5 independent experiments. **P < .01, ***P < .001 compared with a control (BSA or fibrinogen). OD, optical density.

Figure 4.

GPVI-dependent platelet spreading on D-dimer. (A) Glass coverslips were coated overnight with fibrinogen, fibrin, or D-dimer. Mouse platelets (ptl; 20 × 10⁹/L) were allowed to spread on coated coverslips followed by staining for actin with Alexa-488 phalloidin. Scale bar, 5 μm. The figure is representative of 3 similar experiments. (B) Quantification of platelet surface area and platelet count per millimeter squared (n = 3). In each independent experiment, 5 random pictures were analyzed (100 platelets in total). The results are shown as mean ± standard deviation. *P < .05, ***P < .001 compared with fibrinogen; ^#P < .05, ^###P < .001 compared with wild type (WT).

Figure 5.

D-dimer–induced spreading is dependent on GPVI signaling. (A) Glass coverslips were coated overnight with D-dimer as described in "Materials and methods." Human platelets (ptl; 20 × 10⁹/L) were preincubated with dasatinib (3 μM), PRT-060318 (10 μM), or vehicle before being allowed to spread on coated coverslips. Actin staining was performed on fixed platelets with Alexa-488 phalloidin. Scale bar, 5 μm. The figure is representative of 3 similar experiments. (B) Quantification of platelet surface area and platelet count per millimeter squared (n = 3). In each independent experiment, 5 random pictures were analyzed (100 platelets in total). The results are shown as mean ± standard deviation. *P < .05, ***P < .001 compared with control.

Figure 6.

D-dimer in suspension reduces platelet spreading on collagen and fibrin. (Ai) Glass coverslips were coated with collagen or fibrin as described in "Materials and methods." Human platelets (ptl; 20 × 10⁹/L) were preincubated with D-dimer (100 μg/mL) before being allowed to spread on coated coverslips. Actin staining was performed on fixed platelets with Alexa-488 phalloidin. Scale bar, 5 μm. (Aii) Quantification of the surface area of platelets and the number of platelets per millimeter squared of 3 independent experiments. In each independent experiment, 5 random pictures were analyzed (100 platelets in total). The results are shown as mean ± standard deviation (SD). *P < .05, **P < .01. (B) The graph represents the competition binding assay between monomeric GPVI (100 nM) and D-dimer on collagen- or fibrin-coated surfaces. Binding in the absence of D-dimer is represented as 100%. The results are shown as mean ± SD and are representative of 3 experiments. **P < .01, ***P < .001.

Figure 7.

D-dimer inhibits platelet activation by collagen and fibrin in washed platelets. Platelets (500 × 10⁹/L) in the presence of D-dimer (30 μg/mL) were stimulated with collagen (Ai-ii) or CRP, PAR1-peptide, or fibrin (Bi-ii). Aggregation and secretion were monitored by lumiaggregometry (Chrono-Log). The figures show representative aggregation traces with histograms summarizing 3 independent experiments. The results are shown as mean ± standard deviation. *P < .05, **P < .01, ***P < .001.