|Home | About | Journals | Submit | Contact Us | Français|
Matrix metalloproteases (MMPs) play many important roles in normal and pathological remodeling processes including atherothrombotic disease, inflammation, angiogenesis and cancer. Traditionally, MMPs have been viewed as matrix-degrading enzymes, but recent studies have shown that they possess direct signaling capabilities. Platelets harbor several MMPs that modulate hemostatic function and platelet survival, however their mode of action remains unknown. We demonstrated that platelet MMP-1 activates protease-activated receptor-1 (PAR1) on the surface of platelets. Exposure of platelets to fibrillar collagen converts the surface-bound proMMP-1 zymogen to active MMP-1 which promotes aggregation through PAR1. Unexpectedly, we found that MMP-1 cleaved PAR1 at a novel site which strongly activated Rho-GTP pathways, cell shape change and motility, and MAPK signaling. Blockade of MMP1-PAR1 greatly curtailed thrombogenesis under arterial flow conditions and inhibited thrombosis in animals. These studies provide a link between matrix-dependent activation of metalloproteases and platelet-G protein signaling and identify MMP1-PAR1 as a new target for the prevention of arterial thrombosis.
Myocardial infarction due to rupture of atherosclerotic plaques is a leading contributor to morbidity and mortality in the United States, Europe, and other industrialized nations. Acute plaque rupture exposes subendothelial collagen which promotes platelet activation and formation of a potentially occlusive thrombus at the site of vascular damage (Ruggeri, 2002). Following their initial tethering to subendothelial collagen and matrix proteins, activation of transiently adhered platelets by autocrine mediators is critical for propagation of the platelet thrombus. Reinforcement of the transient adhesive contacts by activating G protein-dependent shape change, granule release, and integrins permits growth of a stable thrombus that is resistant to the high shear stress of arterial blood flow (Jackson et al., 2003; Moers et al., 2003). Drugs that target the secondary autocrine mediators of platelet thrombus formation such as aspirin and thienopyridines have proven to be beneficial, however, many patients taking these drugs still sustain thrombotic events and might benefit from new therapeutics that interfere with matrix-dependent platelet activation (Bhatt and Topol, 2003).
Matrix metalloproteases have recently emerged as important mediators of platelet function and vascular biology. Initially described as extracellular matrix remodeling enzymes involved in tissue repair and cancer invasion, a renewed focus has centered on MMPs and the related metalloprotease disintegrins because of their prominence in vascular wall inflammation (Dollery and Libby, 2006) and thrombotic thrombocytopenic purpura (Levy et al., 2001). Endogenous platelet metalloproteases have been shown to damage platelet function by cleaving cell surface receptors and broad-spectrum metalloprotease inhibitors improve post-transfusion recovery of platelet concentrates (Bergmeier et al., 2003; Stephens et al., 2004). Platelets express several metalloproteases including MMP-1, MMP-2, MMP-3, and MMP-14 on their surface (Chesney et al., 1974; Galt et al., 2002; Kazes et al., 2000; Sawicki et al., 1997). Notably, endogenous MMP-1 and MMP-2 can actually promote platelet aggregation but the cell surface target(s) and mechanism of activation have not been elucidated (Galt et al., 2002; Sawicki et al., 1997). A recent study that examined the effects of MMP-1 promoter polymorphisms in 2000 patients, found a significantly increased risk of myocardial infarction in patients with high promoter activity haplotypes and a significantly decreased risk in patients with low promoter activity haplotypes (Pearce et al., 2005). Moreover, serum levels of MMP-1 and MMP-13 were found to be highly elevated in the culprit coronary artery relative to peripheral blood of patients with acute myocardial infarction (Suzuki et al., 2008), raising the question of whether MMPs are directly involved in arterial thrombosis.
It was recently shown that the G protein-coupled receptor, PAR1, is directly cleaved and activated on the surface of cancer cells by fibroblast-derived MMP-1 (Boire et al., 2005). PAR1 is the major thrombin receptor of human platelets (Leger et al., 2006b) and is an important mediator of platelet aggregation following tissue factor (TF)-dependent generation of thrombin (Mackman, 2004). However, under pathophysiologic conditions of acute plaque rupture, exposed collagen is the most efficient stimulus of the critical early events of platelet recruitment and propagation under arterial flow which could trigger metalloprotease activation on the platelet surface.
In this study, we set out to explore a novel metalloprotease-dependent pathway of platelet thrombogenesis through PAR1. We found that exposure of platelets to collagen caused activation of MMP-1 which in turn directly cleaved PAR1 on the surface of platelets. Unexpectedly, MMP-1 cleaved the N-terminal extracellular domain of PAR1 at a distinct site from the thrombin cleavage site. This cleavage event generated a longer tethered peptide ligand which was an agonist of platelet activation and PAR1 signaling. Blocking the MMP1-PAR1 pathway inhibited collagen-dependent thrombogenesis, arterial thrombosis and clot retraction, suggesting that therapeutics that target this metalloprotease-receptor system could be an orthogonal strategy in treatment of patients with acute coronary syndromes.
Studies from the 1970’s (Chesney et al., 1974) showed that human platelets contain significant amounts of collagenase activity which could be released upon exposure to various agonists. More recent studies (Galt et al., 2002) identified the major platelet collagenase as MMP-1 and found that MMP-1 could prime the aggregatory response to other agonists and cause redistribution of the β3 integrins to the cell periphery, however, the mechanism of action of MMP-1 remained unknown. We set out to determine whether the pro-aggregatory effects of platelet MMP-1 were mediated by the PAR1 receptor.
We measured the amount of in situ activation of endogenous proMMP-1 on the platelet surface following stimulation with the primary agonist collagen versus the secondary mediators ADP and thromboxane. Stimulation of platelets with collagen led to the release of the platelet collagenase activity into the supernatant (Figure 1A). Cleavage of a fluorogenic collagen substrate was completely blocked by the MMP-1 inhibitor, FN-439. Blocking antibodies against MMP-1 also completely inhibited the platelet collagenase activity released by collagen, whereas blocking antibodies against the two other collagenases MMP-8 and MMP-13 or IgG control had no effect. Stimulation of gel-filtered platelets with ADP or the thromboxane mimetic, U-46619, however, resulted in a majority of the MMP-1 collagenase activity remaining bound to the platelet. Treatment of washed platelets with collagen but not ADP or U-46619 led to efficient release of the proMMP-1 domain (and/or proMMP-1) as assessed by ELISA using antibodies that recogized the pro domain of MMP-1 (Figure 1B). FACS analysis confirmed that MMP-1 is expressed on the surface of resting platelets which could be released by exposure to collagen (Figure 1C). The lectin convulxin, which ligands specifically with the GPVI/FcγR collagen receptor (Nieswandt and Watson, 2003), also caused full release of the proMMP-1 domain from the platelet surface (Figure 1B). Thus, collagen fibrils per se are not necessary for the release of proMMP-1 from the platelet surface and that other strong platelet agonists may trigger the release mechanism.
The binding site(s) for the platelet-associated proMMP-1 are not known. One candidate is the α2β1 collagen receptor as previous studies showed specific interactions between proMMP-1 and α2β1 on human keratinocytes and astrocytes (Conant et al., 2004; Dumin et al., 2001). Co-immunoprecipitation experiments indicated that proMMP-1 forms a stable complex with the α2β1 integrin on platelets (Figure 1D). MMP-1 (predominantly the pro form) was also found to associate with the αIIbβ3 integrin, as suggested by previous con-focal microscopy studies (Galt et al., 2002). Conversely, proMMP-1 did not associate with GPIbα or GPVI. Therefore, proMMP-1 is likely to be preassociated with both collagen and fibrinogen integrins in resting platelets.
The ability of collagen to activate significant amounts of endogenous MMP-1 collagenase activity on the surface of platelets would lead to the prediction that PAR1 may be cleaved in either an autocrine or paracrine manner following exposure to collagen. Using a monoclonal antibody raised against the amino-terminal thrombin-cleavage peptide region of PAR1, residues 32–46, we assessed the relative abilities of thrombin, MMP-1 and collagen to cause cleavage of platelet PAR1. As shown in Figure 1E, incubation of platelets with thrombin or MMP-1 was able to cause release of the N-terminal cleavage peptide of PAR1 into the supernatant, which was blocked by hirudin or FN-439, respectively. MMP-3 and MMP-7 were not able to cause release of the PAR1 N-terminal peptide (Supplementary Figure 1A). Treatment of the resting platelets with collagen led to the release of the N-terminal peptide which was blocked by FN-439 or a MMP1-blocking Ab, but not by hirudin (Figure 1E, Supplementary Figure 1B). Furthermore, ADP or U-46619 were also able to cause release of the N-terminal peptide of PAR1 albeit at lower efficiency (Supplementary Figure 1C). Blockade of either the P2Y12 ADP receptor with AR-C69931MX (ARC) or thromboxane with aspirin (ASA) had no effect on the collagen-dependent release of the PAR1 N-terminal peptide (Supplementary Figure 1D). Together, these data provide direct evidence that the endogenously-generated MMP-1 collagenase activity is able to cleave PAR1 on the surface of human platelets independently of thrombin.
Studies have demonstrated that serine proteases such as thrombin, plasmin and APC directly hydrolyze PAR1 at LDPR41↓S42 FL (P4P3P2P1↓P1′P2′P3′) to generate the S42FLLRN – tethered ligand (TRAP), which activates PAR1 in an intramolecular mode (Kuliopulos et al., 1999; Seeley et al., 2003; Vu et al., 1991). However, matrix metalloproteases such as MMP-1 generally prefer a hydrophobic amino acid at the P1′ site, a basic or hydrophobic amino acid at P2′, and a small residue (alanine, glycine or serine) at P3′ (Netzel-Arnett et al., 1991; Turk et al., 2001). Therefore, MMP-1 may not efficiently cleave at the R41↓S42FL thrombin site. To identify the MMP-1 cleavage site, we synthesized a 26 amino acid peptide (TR26, PAR1 residues 36–61) corresponding to the N-terminal domain of PAR1 which contained the thrombin cleavage site and flanking regions (Figure 2A–B). Incubation of the TR26 peptide with thrombin yielded the expected cleavage peptide, TR20 (residues 42–61), as determined by mass spectrometry. In contrast, incubation of the TR26 peptide with MMP-1 yielded TR22, which corresponds to PAR1 residues 40–61 (Figure 2A–B). This indicates that MMP-1 cleaves the PAR1 exodomain at LD39↓P40RSFL, a site which is located 2-amino acid residues to the N-terminal side of the thrombin cleavage site at R41-S42.
To verify the location of this putative MMP-1 cleavage site in the full-length receptor, we mutated the critical P1′ residues of both the MMP-1 and thrombin cleavage sites and measured the effects of these mutations on cleavage rates of T7-tagged receptor. To inhibit cleavage by MMP-1, we replaced the putative P1′ proline with asparagine (P40N PAR1), a substitution which had previously been shown to reduce cleavage of α1 collagen peptides to less than 10% (Berman et al., 1992). To inhibit proteolysis by thrombin, we mutated the P1′ serine of the thrombin cleavage site to aspartate (S42D PAR1), a mutation which was anticipated to suppress cleavage by thrombin (Chang, 1985). The P40N PAR1 mutant was fully cleaved by thrombin but was poorly cleaved by MMP-1 using two independent sources of MMP-1 (Figure 2C–D, Supplementary Figure 2A). Conversely, the S42D PAR1 mutant was substantially cleaved by MMP-1 but was poorly cleaved by thrombin. Identical results were seen for cleavage of a mutant TR26-P40N peptide which was cleaved at the R41-S42 bond by thrombin but was not cleaved by MMP-1 (Figure 2A). Functional studies validated the relative cleavage specificities of the P40N and S42D mutants for thrombin and MMP-1. Thrombin was able to fully activate Rho signaling and chemotactic migration in MCF-7 cells expressing the P40N mutant, but had negligible activity for the S42D mutant (Figure 2E–F). Conversely, MMP-1 was able to induce Rho signaling and chemotaxis in MCF-7 cells expressing the S42D mutant, but had little activity towards the P40N mutant. MMP-1 (0.3–10 nM) was not able to detectably cleave T7-tagged PAR2, PAR3, nor PAR4 expressed on COS7 cells (Supplementary Figure 2B–D). Together, these cleavage and signaling data indicate that MMP-1 specifically activates PAR1 by cleaving at LD39↓P40RSFL rather than at the LDPR41↓S42FL thrombin cleavage site and does not cleave the other PARs.
MMP-1 cleavage of PAR1 at LD↓P40RS will generate a longer tethered ligand, P40RSFLLRN–, than that produced by thrombin (Seeley et al., 2003). To provide further evidence that MMP1-generated tethered ligand could activate PAR1, we tested the ability of the synthetic peptide, PR-SFLLRN (PR-TRAP) to stimulate PAR1 signaling. PR-TRAP was a full agonist of PAR1-dependent Rho and p38 MAPK signaling in platelets (Figure 3A–B). Addition of the PAR1 antagonist, RWJ-56110, completely blocked signaling induced by PR-TRAP. We also determined that the PR-TRAP ligand activated platelet shape change, a critical early event in platelet activation which is mediated by G12/13-Rho signaling (Huang et al., 2007; Offermanns et al., 1994). PR-TRAP-induced platelet shape change was completely blocked by the PAR1 antagonist, RWJ-56110 (Figure 3C). We tested two other peptides for agonist activity which would be generated by putative cleavage at the flanking peptide bonds: the R-TRAP peptide corresponding to cleavage at LDP40↓R41SFLLRN and the DPR-TRAP peptide corresponding to cleavage at L38↓D39PRSFLLRN (Figure 2B). The R-TRAP peptide retained partial agonist activity for PAR1-dependent platelet shape change, whereas the DPR-TRAP peptide had nearly no activity (Figure 3C). Likewise, the control peptide, RP-TRAP, in which the first two amino acid residues were reversed, did not stimulate Rho or p38 MAPK, nor platelet shape change (Figure 3A–C).
We then confirmed that exogenously-added MMP-1 was able to activate PAR1-dependent signaling in platelets. MMP-1 (3 nM) was able to stimulate Rho-GTP activity to the same extent as equimolar thrombin (Figure 4A). MMP-1 was also able to elicit platelet shape change, calcium mobilization, and aggregation which was inhibited by the PAR1 antagonist, RWJ-56110 (Figure 4B–D). Exogenously added MMP-1 also activated phospho-p38 MAPK and its substrate, MAPKAP-K2, in an activity profile similar to thrombin (Figure 4E–F). MAPKAP-K2 phosphorylates the small heat shock protein HSP27 involved in cytoskeletal reorganization, further suggesting that MMP-1 may play a role in the initial events leading to platelet shape change and help prime platelets for aggregation.
We tested whether pharmacologic blockage of metalloproteases or PAR1 would have an effect on collagen-dependent platelet aggregation. Soluble type I fibrillar collagen stimulated platelet aggregation with an EC50 of 5 μg/ml. Inhibition of metalloproteases with the zinc-chelating agent 1,10-phenanthroline, resulted in 80% loss of aggregation to 5 μg/ml collagen (Figure 5A). Likewise, the broad spectrum metalloprotease hydroxamate inhibitor, MMP-200 (IC50 = 7 nM for MMP-1, 2.3 nM for MMP-2, 135 nM for MMP-3, 10–100 nM for MMP-7, 1–10 nM for MMP-13) caused a significant 50–60% inhibition of collagen-initiated aggregation. Treatment with the MMP-1 inhibitor, FN-439, inhibited collagen-induced aggregation to the same extent as MMP-200. Conversely, inhibitors against MMP-8, MMP-9 and MMP-13 had no effect on collagen-induced aggregation (data not shown). The specific thrombin inhibitor hirudin or the broad-spectrum serine protease inhibitor, PPACK, had no effect on collagen aggregation (Figure 5A). PAR1 was inhibited by three orthogonal approaches to evaluate its contribution to collagen-dependent aggregation. The small-molecule inhibitor RWJ-56110 or a PAR1-blocking antibody, attenuated 50% of collagen (5 μg/ml)-induced aggregation, the same extent as the MMP-1 inhibitor. Likewise, the cell-penetrating PAR1 pepducins, P1pal-12 and P1pal-7, which inhibit PAR1 signaling to intracellular G proteins (Boire et al., 2005; Covic et al., 2002a; Kaneider et al., 2007), gave identical levels of inhibition as blocking MMP-1 (Figure 5A). Inhibition of the PAR4 thrombin receptor with the P4pal-10 pepducin (3 μM) had only a slight (~10%) effect on collagen-induced aggregation.
Collagen is known to induce p38 stress-activated protein kinase (MAPK) pathways in human platelets though the mechanism remains unclear (Kuliopulos et al., 2004). Addition of collagen caused robust phosphorylation of p38 MAPK (Figure 5B). The collagen (5 μg/ml)-induced phospho-p38 MAPK signal was effectively blocked by the PAR1 and MMP-1 inhibitors, but not with inhibitors against MMP-8, MMP-9/13, or thrombin. We also found that collagen-dependent activation of the p38 MAPK substrate, MAPKAP-K2 was dependent on both PAR1 and MMP-1. Collagen-activation of phospho-MAPKAP-K2 was blocked by the PAR1 antagonists, RWJ-56110 and P1pal-7, and by FN-439 but not by MMP8 inh (data not shown). Collagen caused robust activation of Rho-GTP which was attenuated by 75% with antagonists against PAR1 and MMP-1, but not by inhibitors or blocking antibodies against MMP-8, MMP-9/13, or thrombin (Figure 5C–D). However, at saturating levels of collagen (20 μg/ml) sufficient to elicit full aggregation of platelets, none of the PAR1 nor MMP-1 inhibitors had a major effect (≤25%) on collagen-dependent aggregation, the phospho-p38 MAPK signal, or Rho-GTP activity (Supplementary Figure 3A–D), indicating that the MMP1-PAR1 pathway can be bypassed at super-EC50 levels of collagen. To address whether the observed MMP-1 effects were due to secondary secretion of ADP or thromboxane after collagen stimulation, we inhibited the P2Y12 ADP response with ARC and thromboxane with aspirin. Treatment of platelets with either ARC or aspirin had no effect on the ability of 5 or 20 μg/ml collagen or 10 nM MMP-1 to activate Rho-GTP (Figure 5D, Supplementary Figure 3D) or phospho-p38 MAPK, but the inhibitors could still suppress aggregation to collagen (Supplementary Figure 3A,C). In contrast, blockade of the MMP1-PAR1 pathway nearly completely inhibited activation of p38 and Rho-GTP to 5 μg/ml collagen (Figure 5B–D). This would indicate that at EC50 collagen exposure, MMP1-PAR1 is essential for activation of p38 and Rho-GTP, and important for aggregation, whereas the secondary ADP and thromboxane pathways do not activate p38 and Rho-GTP at any range of collagen concentration. At saturating collagen, the ADP and thromboxane contributions appear to compensate for the MMP1-PAR1 pathway in platelet aggregation via mechanisms that do not involve p38 or Rho-GTP signaling.
Activation of platelets in ruptured atherosclerotic plaques occurs under high shear-stress conditions on subendothelial surfaces enriched in collagen fibrils. We examined whether initial formation and propagation of platelet-platelet thrombi on collagen surfaces was affected by blocking MMP-1 or PAR1. Whole human blood was anti-coagulated with heparin and perfused over collagen-coated surfaces at arterial flow rates of 1000 s−1. Treatment with either MMP-1 or PAR1 inhibitors did not affect the primary adhesion of platelets to the immobilized collagen fibrils. However, the growth rate and size of platelet aggregate “strings” was significantly attenuated by ~75% with the MMP-1 inhibitor, FN-439, or the PAR1 blocking agents P1pal-12, P1pal-7 or RWJ-56110 (Figure 6A,C). By comparison, aspirin pre-treatment had little or no effect on the growth of the platelet thrombi. This result is consistent with thromboxane playing a relatively minor role in thrombogenesis under arterial shear stress conditions (Jackson et al., 2003) as compared to the MMP1-PAR1 pathway.
Collagen-activated platelets also provide a pro-coagulant surface and produce tissue factor which aids in the production of thrombin (Giesen et al., 1999; Mackman, 2004). To evaluate whether MMP1-PAR1 activation of early platelet thrombi formation is also relevant under conditions in which thrombin activity is not inhibited, we conducted the arterial flow experiments in the presence of corn trypsin inhibitor (CTI) which blocks factor XIIa and the contact pathway of coagulation but does not inhibit thrombin generation in whole blood (Mann et al., 2007). The results using the CTI anti-coagulant were very similar to those conducted with heparin. Inhibition of MMP-1 or PAR-1 significantly attenuated the size of the platelet micro-thrombi on the collagen surfaces, whereas addition of the thrombin inhibitor, hirudin, had no effect (Figure 6B,D). Again, aspirin pre-treatment of the CTI-whole blood did not affect the extent of platelet-thrombi formation on the collagen surfaces. Thus, under conditions of arterial shear stress, MMP1-PAR1 significantly promotes early thrombogenesis on collagen surfaces.
Next, we used clot retraction assays to examine the potential role of MMP-1 on the structure of large platelet-rich clots over time. Platelet receptors trigger clot retraction by activating myosin-dependent contraction of the cytoskeleton which is in turn connected to the extracellular matrix (fibrinogen) via focal adhesions. We found that the MMP-1 inhibitor, FN-439, completely inhibited clot formation and retraction induced by 2.5–5 μg/ml collagen (Supplementary Figure 4). Blockade of PAR1 with RWJ-56110 gave a nearly identical pattern of inhibition, whereas the negative control MMP9/13 inhibitor had negligible effects over the whole collagen titration. Therefore, MMP-1 and PAR1 play a significant role in the formation and retraction of large platelet-rich thrombi initiated by collagen.
We examined whether blocking the MMP1-PAR1 pathway would protect against collagen-mediated systemic platelet activation in vivo. Guinea pigs serve as an relevant model to test platelet function because like humans they also express PAR1 on their platelets (Leger et al., 2006a) and guinea pig MMP-1 shares 90% identity with human MMP-1 (Huebner et al., 1998). We first confirmed that guinea pig platelets express proMMP-1 on their surface by FACS analysis and that addition of collagen causes release of collagenase activity which is completely blocked by either FN-439 or a MMP1-neutralizing Ab (Supplementary Figure 5A–B). Likewise, inhibition of MMP-1 or PAR1 gave 35–50% suppression of aggregation and complete inhibition of Rho-GTP activity in response to 10 μg/ml collagen in guinea pig platelets (Supplemental Figure 5C–D), which was consistent with the previous results using human platelets. Intravascular platelet activation was then induced by an intravenous injection of collagen into the guinea pigs. The infused collagen caused a severe drop in mean systemic platelet counts from a baseline level of 309,000 ± 50,000/mL to 194,000 ± 20,000/mL. Strikingly, pre-administration of the PAR1 pepducin, P1pal-7, almost completely protected against collagen-induced thrombocytopenia in the guinea pigs (Figure 6E). The MMP-1 inhibitor, FN-439, also afforded significant protection against intravascular platelet activation.
To assess the efficacy of blockade of MMP-1 and PAR1 on arterial thrombosis, we used a standard carotid artery FeCl3 injury model in guinea pigs. FeCl3 causes denudation of the artery and exposure of type I collagen and other subendothelial matrix proteins. Intravenous administration of the PAR1 antagonist, P1pal-7, gave a significant 50% prolongation of the mean occlusion time (Figure 6F). Administration of the MMP-1 antagonist, FN-439, gave a similar prolongation of the mean arterial occlusion time. Co-administration of the PAR1 and MMP-1 inhibitors did not lead to further prolongation of the mean occlusion time, consistent with MMP-1 acting in the same pathway as PAR1. Lastly, collagen zymography revealed that the platelet-rich clot isolated from injured carotid arteries of vehicle-treated animals (veh clot) had significant MMP-1 activity which co-migrated with APMA-activated MMP-1 and with the MMP-1 activity from the supernatants of collagen-activated platelets (Figure 6G). Conversely, resting platelets (control) from whole blood or arterial thrombi from animals treated with the MMP1-inhibitor (FN439 clot) did not contain active MMP-1. These data, together with the previous results, indicate that inhibition of MMP1-PAR1 may provide substantial protection against collagen-dependent platelet activation and acute arterial thrombosis in animals.
Matrix metalloproteases have long been implicated in the chronic pro-inflammatory and tissue-remodeling events leading to cleavage of interstitial collagen and development of vulnerable atherosclerotic plaques (Sukhova et al., 1999). Although patho-anatomic studies of human atherosclerotic lesions suggest that large plaques can cause ischemic symptoms, the key contributing factor to the morbidity and mortality associated with atherosclerosis is excessive platelet thrombus formation on exposed collagen surfaces following acute plaque rupture (Ruggeri, 2002).
In the present study, we discovered a new collagen-initiated pathway of thrombogenesis which is mediated by the autocrine action of platelet MMP-1 on the PAR1 receptor. Exposure of platelets to collagen caused robust activation of MMP-1 on the platelet surface which in turn directly cleaved and activated PAR1 independently of thrombin. These studies provide a link between matrix-dependent activation of metalloproteases and platelet G protein signaling and identify MMP1-PAR1 as a potential new target for the prevention of arterial thrombosis.
Unexpectedly, MMP-1 cleaved PAR1 at a distinct site in its extracellular domain which generated a longer tethered ligand than that produced by thrombin. The MMP1-cleaved receptor or soluble peptide analog strongly stimulated G12/13-Rho-dependent pathways, chemotaxis and MAPK signaling in platelets and other cells. The MMP-1 cleavage site on PAR1 aligned with an optimized MMP-1 cleavage site motif determined from mixture-based oriented peptide libraries (Turk et al., 2001) and by substrate cleavage studies (Berman et al., 1992; Netzel-Arnett et al., 1991). Mutation of respective P1′ residues uncoupled MMP-1 from thrombin cleavage and generated PAR1 receptors that exhibited protease-specific activity.
Collagen signaling in human platelets through the α2β1 and GPVI/FcγR collagen receptors is not well understood, but has been shown to be partly dependent on G protein signaling through autocrine stimulation of ADP and thromboxane receptors (Jackson et al., 2003). Blockade of the P2Y12 Gi-coupled ADP receptor inhibits collagen-dependent thrombogenesis under arterial flow conditions, thus establishing an important role for downstream ADP-Gi signaling. Thromboxane activates the Gq and G12/13-coupled TXA2 receptor, however, aspirin failed to prevent thrombogenesis on collagen surfaces under arterial shear stress and does not prevent occlusive thrombus formation in patients with severe arterial stenosis (Veen et al., 1993). The current studies show that MMP-1 is a potent activator of PAR1-G12/13 pathways involved in platelet shape change and Rho activation and thus would synergize with P2Y12-Gi signaling.
Blockade of MMP-1 or PAR1 with pharmacologic inhibitors significantly attenuated thrombogenesis on collagen surfaces under arterial shear stress conditions and thrombosis in animals. As compared to MMP-1 inhibition, antagonism of thrombin had little effect on early thrombogenesis on the collagen surfaces under high arterial flow rates. Indeed, several studies have previously shown that thrombin may be more important for later propagation and stability of platelet thrombi, and is not involved in initiating early thrombus growth at high arterial shear stress (Fressinaud et al., 1992; Gast et al., 1994; Inauen et al., 1990) unless tissue factor levels are extremely high (Okorie et al., 2008). Likewise, thrombin inhibitors such as heparin have incomplete effects on platelet thrombus formation at high arterial flow rates, but have a more prominent inhibitory effect on the growth and overall stability of platelet thrombi at low and intermediate shear rates (Inauen et al., 1990). Unlike direct blockade of MMP-1 or thrombin, downstream inhibition of PAR1 might impact both the initial MMP1-dependent events of platelet thrombi propagation on blood vessel collagen, along with later thrombin-dependent propagation and stabilization and could prove beneficial in preventing arterial thrombosis in acute settings.
ProMMP-1 (≥90% purity, from human synovial fibroblasts), proMMP-3, proMMP-7, FN-439 (MMP inh-1), MMP8 inh, and MMP9/13 inh were from Calbiochem. MMP-200 was obtained from Enzyme Systems Products. RWJ-56110 was a gift from Johnson & Johnson and AR-C69931MX was from Astra Zeneca. The pepducins, P1pal-12, P4pal-10 and P1pal-7, and other peptides were synthesized with C-terminal amides and purified by RP-HPLC as before (Covic et al., 2002a). The IIaR-A monoclonal antibody which reacts to the amino-terminal thrombin-cleavage peptide of PAR1, was from Biodesign (Kennebunk, ME). A solid phase proMMP-1 ELISA system from R&D Systems (Quantikine DMP100) detects proMMP-1 and soluble pro domain but does not detect active MMP-1. The MMP-1 blocking Ab (AB8105) recognizes both pro and active forms of MMP-1 but do not cross react with other MMP family members (Chemicon). The MMP-8 (IM38L) and MMP-13 (IM44L) blocking Abs were from Oncogene, the anti-α2 (Gi9 or AK7), β1 (MAB1987), β3 (MAB1957) were from Chemicon, GPVI (SC20149) was from Santa Cruz, GPIBα (MM2/174) was from AbD Serotec.
Phlebotomy was performed on 20 healthy volunteer donors following informed consent procedures established by the IRB of Tufts Medical Center. Platelets from platelet-rich plasma (PRP) were isolated by gel filtration using a Sepharose 2B (Pharmacia) in modified PIPES buffer in the presence of 1 mM EDTA and 0.1 U/ml of apyrase as described (Kuliopulos et al., 1999). Alternatively, whole blood was obtained from Hartley Sprague guinea pigs (drawn from the vena cava) into 3% citrate plus 10 U/ml heparin. Washed platelets from the guinea pigs were prepared in PIPES buffer. Platelet aggregation was measured with a Chronolog 560VS/490-2D aggregometer as before (Kuliopulos et al., 1999).
Gel filtered platelets were exposed to different agonists in the presence or absence of 5 μM FN-439 or 0.01 U hirudin, for 10 mins at 37 °C in a reaction volume of 250 μl. Reactions were terminated by centrifugation to remove the platelets from the supernatant. Supernatants were collected and concentrated 20-fold by slow evaporation in a Speed Vac; concentrated supernatants were then applied as a 10 μL spot on nitrocellulose membranes. Membranes were dried at room temperature for 30 min and then probed with the IIaR-A monoclonal antibody.
Human PAR1 was cloned into pcDEF3 as described previously (Kuliopulos et al., 1999) and was used for generating all mutants. The PAR1 mutants P40N and S42D were generated using the Quick Change Site-Directed Mutagenesis kit (Stratagene) and sequenced to verify the fidelity of the mutagenesis.
Human or guinea pig platelets from PRP were concentrated four-fold by centrifugation at 700g for 25 min at room temperature, and then resuspended in 0.25 volume of PIPES containing 1 mM EDTA (final platelet count was 109/mL). Platelets were treated with PBS (buffer), 20 μM ADP, 20 μM U-46619, or 20 μg collagen in the presence of 2.5 mM CaCl2. The platelets were incubated for 15 min at 37 °C with occasional gentle mixing. Platelets were collected by centrifugation at 10,000g for 5 min at 4 °C and resuspended in lysis buffer (50 mM Tris HCl, 100 mM NaCl, 1 mM NaF, 5 mM EDTA, 0.1% (v/v) Triton X-100, 100 μM PMSF, pH 7.4) and then sheared with a 27 gauge needle. Supernatants and platelet pellets were separated by centrifuging the lysate at 12,000g for 2 min. DQ Collagen type I (Molecular Probes) was employed as fluorescent reporter of collagenase activity as described (Boire et al., 2005). Collagen zymography was performed as described previously (Gogly et al., 1998). In brief, samples were resolved in 8% polyacrylamide gels containing (1 mg/ml) calf skin type I collagen (Sigma). After electrophoresis, gels were washed (30 min) twice in regeneration buffer followed by incubation in developing buffer (48 hrs) and stained with 0.25% coomassie brilliant blue G-250 and then destained.
A flow chamber (Glycotech) with Type-I fibrillar collagen-coated glass slides was mounted on the stage of an IMT-2 inverted microscope (Olympus) equipped with Retiga 1300 digital camera (QImaging) and 40x objective. One of the flow chamber inlets was connected to a syringe pump (Harvard Apparatus) calibrated to create a shear rate of 1,000 s−1. Normal human blood was anticoagulated with 10 U/ml of heparin, or with corn trypsin inhibitor (CTI, 30 μg/ml final). Whole blood was pretreated with various pharmacologic inhibitors for 5 min and then perfused over the collagen-coated glass slide. After 2–15 min of perfusion, blood was removed from the flow chamber by gentle displacement with PIPES buffer and images of 8–10 fields were acquired using OpenLab software (Improvision) and analyzed using NIH Image 1.63 software.
Animal experiments were performed in accordance with the NIH guidelines and approved by the Tufts IACUC. 2–4 week old Hartley guinea pigs (170–260 g) were anesthetized by i.p. injection of xylazine (10 mg/kg) plus ketamine (50 mg/kg) and then catheterized via the left jugular vein and injected (200 μL) with either vehicle (20% DMSO/80% water), P1pal-7 or FN-439. For determination of collagen-dependent systemic platelet activation, 10 min after administration of inhibitors, 200 μg collagen in 200 μL of PBS was delivered via the jugular vein. Ten min after collagen injection, blood was collected into sodium citrate (0.3% v/v final) from the contralateral jugular vein and platelet counts were measured with a Hemavet850. For arterial thrombosis experiments, 10 min after i.v. administration of inhibitors, the right carotid arteries were injured for 20 min using a 24 mm2 piece of Bio-Rad Trans-Blot paper soaked in 20% FeCl3. Arterial flow 5 mm distal to the site of injury was measured with a 0.5 V Doppler probe (Transonic Systems). An arterial occlusion was defined as a flow rate of <0.01 V for ≥5 min.
Support from NIH R01 HL-57905, R01 HL-64701 and R01 CA-122992 (to A.K.) is gratefully acknowledged.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.