MMP‐13 binds to platelet receptors αIIbβ3 and GPVI and impairs aggregation and thrombus formation

Abstract Essentials MMP‐13 has the potential to influence platelet function and thrombus formation directly. We sought to elucidate whether MMP‐13 is able to bind to specific platelet receptors. MMP‐13 is able to bind to platelet alphaIIbbeta3 (αIIbβ3) and glycoprotein (GP)VI. These interactions are sufficient to inhibit platelet aggregation and thrombus formation. Background Acute thrombotic syndromes lead to atherosclerotic plaque rupture with subsequent thrombus formation, myocardial infarction and stroke. Following rupture, flowing blood is exposed to plaque components, including collagen, which triggers platelet activation and aggregation. However, plaque rupture releases other components into the surrounding vessel which have the potential to influence platelet function and thrombus formation. Objectives Here we sought to elucidate whether matrix metalloproteinase‐13 (MMP‐13), a collagenolytic metalloproteinase up‐regulated in atherothrombotic and inflammatory conditions, affects platelet aggregation and thrombus formation. Results We demonstrate that MMP‐13 is able to bind to platelet receptors alphaIIbbeta3 (αIIbβ3) and platelet glycoprotein (GP)VI. The interactions between MMP‐13, GPVI and αIIbβ3 are sufficient to significantly inhibit washed platelet aggregation and decrease thrombus formation on fibrillar collagen. Conclusions Our data demonstrate a role for MMP‐13 in the inhibition of both platelet aggregation and thrombus formation in whole flowing blood, and may provide new avenues of research into the mechanisms underlying the subtle role of MMP‐13 in atherothrombotic pathologies.


| INTRODUCTION
Platelet-extracellular matrix and platelet-platelet adhesions are central to the formation of thrombi. MMP-13 is up-regulated in inflammation, and is elevated in the atherosclerotic plaque, contributing to its vulnerability. 1 It is also implicated in the progression and remodelling of cerebral tissue in stroke. 2 Plaque rupture releases MMP-13 into the local environment where it has direct access to plasma proteins, blood cells, and platelets. Following injury to the blood vessel wall, specific platelet receptors mediate platelet-collagen and platelet-platelet interactions. GPIbα binds to immobilized von Willebrand factor (VWF) in the vessel wall, initiating platelet capture, 3 and glycoprotein (GP)VI binds directly to collagen and activates platelets.
Our work identifies potential roles for MMP-13 in modulating the recruitment or activation of platelets in thrombotic pathologies.

| Washed platelet preparation and platelet adhesion assays
Plates were coated with 10 μg/ml MMP-13 variants in Tris buffered saline (TBS) for 1 h at 24°C. Plates were then blocked with 5% BSA in TBS for 20 minutes at 24°C and washed with TBS prior to the addition of washed platelets. Platelets were purified and adhesion assays conducted as previously described. 17,18 Glanzmann thrombasthenic blood was kindly provided by  ated as previously described 7 along with the anti-GPVI scFvs   10B12 and 1C3 and the non-GPVI-binding scFv 2D4 19-23 which were a kind gift from Dr. P. Smethurst. Human fibrinogen type I was purchased from Sigma, UK. Anti-α2β1 antibody 6F1 was a kind gift from Prof. B. Coller (Mount Sinai Hospital, New York, NY, USA).

| Aggregometry
Washed platelet aggregation was performed using a Chrono-Log turbidimetric aggregometer (Labmedics, Abingdon on Thames, UK). 250 μL aliquots of platelets, 2 × 10 8 /mL in calcium-free Tyrodes buffer (CFT), were pre-incubated for 1 h with 80 nmol L −1 MMP-13 or vehicle control prior to the addition of receptor agonists in a maximum volume of 5 μL. Thrombin, calcium ionophore A23187 (San Diego, CA, USA), bovine collagen I fibers (Ethicon Corp, Somerville, NJ, USA), HORM ® and CRP-XL were prepared and employed to activate platelets as previously described. 25,26 Aggregations were allowed to proceed for 5 minutes.

| Cleavage of platelet receptors and their substrates by MMP-13
Recombinant human (rh)GPVI, purified αIIbβ3 (100 μg/mL, R&D Systems) and human fibrinogen type I (1 mg/mL) were incubated with MMP-13 or MMP-13(E204A) (8 μmol L −1 final concentration) for 2 h at 37°C. An equal volume of Tris buffer was used as a negative control. Reducing sample buffer was then added to the mixture in preparation for electrophoresis and Western blotting.

| In vitro sheddase activity assays
Dialysed MMP-13 at a final concentration of 130 nmol L -1 was incubated with washed platelets for 60 minutes at 37°C. Positive controls for shedding included thrombin (1 U/mL, Sigma, UK) combined with fibrous type I collagen (1 mg/mL), the calcium ionophore A23187 (1 μg/mL). The platelets were then pelleted at 1500 g for 1 minute. The supernatants were aspirated and centrifuged again to ensure platelet depletion. This new supernatant was retained for analysis. Where indicated, platelet lysate was resuspended in reducing sample buffer.

| Electrophoresis and Western blotting
Protein samples in reducing sample buffer were boiled for 5 minutes and applied to 4-12% NuPage Gels and separated by electrophoresis using the Xcell SureLock system (Invitrogen, Paisley, UK) under reducing conditions. Proteins were then transferred on to nitrocellulose membrane (Millipore, Bedford, UK) at 40 V overnight at 4°C using a Mini Protean II system (Bio-Rad, Hemel Hempstead, UK).
Following transfer, the PVDF was blocked (5% nonfat dry powdered F I G U R E 1 Washed platelet adhesion assays. (A) Platelets adherent to 10 μg/ml coated MMP-13 variants in the presence of 2 mmol L −1 Mg 2+ (red bars) or 2 mmol L −1 EDTA (orange bars) where stated. Where appropriate, platelets were preincubated with anti-GPVI, αIIbβ3 or α2β1 antagonists. BSA and GPP 10 were used as Mg 2+ -independent negative controls. The platelet α2β1 binding-peptide GFOGER was included as an Mg 2+ dependent positive control. *P < .05; **P < .01; † (oneway anova and Holm multiple comparison test) relative to untreated platelets in either the presence of Mg 2+ or EDTA, as appropriate. (B) Inhibition of platelet adhesion to CRP by anti-GPVI scFvs as described above (C) Platelets adherent to MMP-13(E204A) and MMP-13 CAT and HPX domains. CRP-XL was used as a positive control. **P < .01 (one-way anova and Holm multiple comparison test) relative to adhesion to MMP-13(E204A). Data represent mean A 405 ± SE of three experiments Anti-human GPVI was a kind gift from Dr. P. Smethurst, and anti-β3 was obtained from Abcam, Cambridge, UK. Following washes with TBST, the membrane was incubated with HRP conjugated secondary antibody (1:10000 dilution/TBST) for 1 h at 24°C. The PVDF was developed using a chemiluminescent substrate (GE Healthcare, Amersham, Bucks, UK).

| Whole blood perfusion experiments
Whole blood was pre-incubated with either carrier (TBS) or 80 nmol L −1 MMP-13 for 1 h prior to perfusion over 10 μg/mL type I fibrous collagen as previously described. 17,25 Where indicated, slides were coated with MMP-13(E204A) alone as a (negative) control.

| RE SULTS
Adhesion assays were performed in the presence of 2 mmol L −1 EDTA or Mg 2+ to ablate or support integrin-mediated adhesion.
Platelet adhesion to MMP-13 preparations was significantly reduced, but not abolished, by EDTA, suggesting both integrindependent and -independent contributions, whereas EDTA fully abolished binding to the collagen-binding integrin-specific peptide GFOGER ( Figure 1A).
Platelet pre-incubation with the αIIbβ3 antagonists, GR144053, cRGD, and RGDS, and with anti-GPVI scFv 10B12 and 1C3, all caused a substantial and significant reduction (P < .01) in platelet adhesion to proMMP-13 ( Figure 1A), with residual adhesion being observed in the presence of EDTA remaining above negative control levels (nonspecific substrates). This may indicate cooperative binding to αIIbβ3 F I G U R E 2 Competition, GPVI and Glanzmann platelet binding assays. (A) MMP-13(E204A) and either 10B12 or GR144053 were used to obtain IC 50 values for the inhibition of washed platelet adhesion to 10 μg/ml coated fibrinogen (i, iii) or CRP (ii, iv), respectively. (v) Adhesion of MMP-13(E204A) to recombinant human GPVI monomer and dimer. Plates were coated with 10 μg/ml GPVI or BSA as a negative control. MMP-13(E204A) at a concentration of 83 nM was allowed to adhere for 1 h at room temperature, then detected using an antibody directed at the MMP-13 linker region, as described in Methods. Data represent mean A 450 ± SE of three experiments. (B) Platelets from a healthy donor (red bars) and from a Glanzmann thombasthenic individual (orange bars) were allowed to adhere to MMP-13, fibrinogen and CRP-XL coated plates. Where appropriate, platelets were pre-incubated with anti-GPVI (1C3), or αIIbβ3 antagonists as described for Figure 1. Data represent mean A 405 ± SE of duplicate readings for one experiment due to the rarity of the Glanzmann donor and GPVI. Interaction between MMP-13 and integrin α2β1 was less prominent, since blocking antibody 6F1 had just a small effect, and was not studied further. Platelet pre-incubation with the fibrinogenderived peptide, KQAGDV, had no effect on platelet adhesion, indicating that MMP-13 binds αIIbβ3 closer to the primary RGD-binding site. Soluble fibrinogen also did not block platelet adhesion to MMP-13, in line with the need for platelet activation for soluble fibrinogen binding to αIIbβ3 to occur, whereas immobilized fibrinogen is already competent to bind. The GPVI-specific scFv, 1C3, does not target the collagen-binding site at the apex of GPVI, unlike 10B12, and was unable to inhibit the adhesion of washed platelets to CRP ( Figure 1B).
1C3 binding requires both GPVI Ig domains and is thought to reduce platelet activation by inhibiting receptor clustering; its epitope includes isoleucine 148, 21,23 located in strand E on the opposite face of D2 to the crystal structure dimerization interface located in strand G. 24   to bind weakly to GPVI monomer, but strongly to the GPVI dimer ( Figure 2A[v]). Similar assays of adhesion to recombinant αIIbβ3 revealed some binding of its native ligand, fibrinogen, but little or no binding of MMP-13, regardless of whether Mg 2+ , Mn 2+ , or Ca 2+ was present, nor could we detect binding of MMP-13 to purified αIIbβ3 (results not shown). Adhesion of αIIbβ3-null Glanzmann platelets to MMP-13, however, was markedly reduced ( Figure 2B[i]). Blockade of αIIbβ3 on healthy platelets resulted in the same adhesion level as seen for αIIbβ3-null platelets. As expected, binding of Glanzmann platelets to fibrinogen was abolished ( Figure 2B[ii]) and to CRP was unaffected ( Figure 2B[iii]). Our results indicate that, whilst MMP-13 appears able to bind to αIIbβ3 on the platelet surface, recombinant αIIbβ3 used here cannot reproduce this effect.
Whilst it was able to cleave the recombinant αIIbβ3 β-chain and GPVI in solution, as well as fibrinogen α and β chains ( Figure 3A),

| D ISCUSS I ON
We have previously shown that degradation by MMP-13 has the potential to modulate platelet adhesion to collagen. 17 MMPs are zymogens; proteolysis is required to expose their catalytic site.
Here we show that surprisingly, all forms of MMP-13, pro-and active wild type enzyme as well as their catalytically inactive mutant counterparts, were able to support a high level of platelet adhesion under static conditions. This adhesion was inhibited by the anti-GPVI scFvs 10B12 and 1C3 suggesting that the relatively large MMP-13 occludes the sites of both 10B12 and 1C3 binding on the receptor. MMP-13 was also able to bind strongly to the GPVI dimer.
Although GPVI dimerization increases upon platelet activation, dimeric GPVI is also present on resting platelets and is required for their initial interaction with exposed collagen. 26 Crystallography of the proMMP-13 structure in complex with pro-domain peptides revealed a dimeric form as an HPX-mediated dimer like some other metalloproteinases, although in this study, 27 MMP-13 was not dimeric in solution. Conceivably, interaction of MMP-13 with platelet surface GPVI dimer may provide a template for dimerization of the MMP. Platelet adhesion to MMP-13 was also inhibited by the anti-αIIbβ3 compound GR144053, and binding of Glanzmanns αIIbβ3null platelets to MMP-13 was significantly reduced. Following pre-incubation of washed platelets with MMP-13, neither GPVI nor αIIbβ3 was shed from the platelet surface. It would appear, therefore, that whilst able to bind to platelet αIIbβ3 and GPVI, the orientation of MMP-13 on the platelet surface does not allow access of its CAT domain to the cleavage site, which, for other sheddases, resides close to the transmembrane region and is regulated by membrane structure 28 or substrate phosphorylation. 29 Preincubation with MMP-13 did not result in platelet activation or aggregation. Here it is worth noting that MMP-13 has been reported F I G U R E 5 Activation of platelets in whole blood. Whole blood was mixed with anti P-Selectin and the agonists 2 mmol L −1 proMMP-13(E204A)/MMP-13, 100 μg/ml CRP-XL, 100 μg/ml HORM ® equine collagen I fibers, thrombin activating peptide (TRAP; 500 μmol L −1 ) calcium ionophore A23187 (100 μmol L −1 ) or HBS (negative control) added. Alexa 488 conjugated anti-mouse was then added and after 10 minutes at 24°C the volume was made up to 500 μl with isotonic solution. After 30 minutes fluorescence was measured using an Accuri C6 flow cytometer (BD Biosciences, Oxford, UK). Data represent mean A 450 ± SE of three separate donors. **P < .005; (one-way anova and Holm multiple comparison test) Until now, the role of MMP-13 in atherothrombosis has been considered to be restricted to collagen proteolysis and remodelling, rendering plaque more friable and prone to rupture. 1 However, MMPs are now emerging as important mediators of platelet function. 32,33 MMPs -1 and -2 are released from activated platelets where they colocalize with integrins at the sites of platelet-platelet interaction. 10,34 Active MMP-1 and -2 can stimulate platelet function, suggesting F I G U R E 6 Platelet adhesion and thrombus deposition on fibrillar type I collagen. Untreated whole blood and blood pre-incubated with 80 nmol L −1 MMP-13 or negative control where stated was drawn through a flow chamber for 5 minutes over (A) collagen type I fibers or (B) collagen type I fibers co-coated with MMP-13 using a syringe pump to generate a wall shear rate of 1000 s −1 , corresponding to arteriolar conditions. Surface coverage (i) mean height (ii) and (iii) ZV 50 are the mean taken from a minimum of three different donors as measured using confocal microscopy. *P < .05; **P < .01; (one-way anova and Holm multiple comparison test) relative to MMP-untreated platelets receptor engagement and proteolysis. 34,35 MMPs in atherosclerotic lesions, released from the injured vessel wall itself or from platelets and monocytes, and that can also interact with platelets, are likely to interfere with the progression of plaque rupture, subsequent thrombosis and its associated pathologies including stroke, reperfusion injury, and hemorrhagic transformation. Indeed, these processes are associated with an upregulation of MMP activity. 2,31,36 In mice, MMP-13 is the key mediator of collagen degradation in atheroma and confers instability onto the vulnerability plaque cap. [37][38][39] Disruption of the blood brain barrier (BBB) by MMPs is associated with hemorrhagic transformation following ischemic stroke, 36,40,41 whilst MMPs -9 and -13 are implicated in the early pathology of stroke progression, and plasma MMP-13 levels correlate with lesion volume. 2,31 In addition, the platelet collagen receptor GPVI has been identified in models of models of reperfusion injury, 42 is associated with increased risk of stroke development, and is also seen after ischemic stroke. 43 Here we demonstrate that MMP-13 can exert an antithrombotic effect; inhibiting platelet aggregation and thrombus formation in flowing whole blood. It may be that this metalloproteinase has multiple roles in the pathology of ischemic stroke; firstly by undermining the stability of the fibrous cap of atheroma and so promoting its rupture, then modulating the BBB to increase bleeding risk, and finally acting on platelets to impair the aggregatory interactions, by antagonising GPVI and αIIbβ3 which would normally protect against bleeding. MMP-13 would appear therefore to modulate the architecture around sites of infarction to increase both risk of stroke and its hemorrhagic complications. The effect of MMP-13 will depend upon its local level and the exposure of MMP-13-binding matrix components and warrants further investigation.