Coordinate Activation of Human Platelet Protease-activated Receptor-1 and -4 in Response to Subnanomolar α-Thrombin*

We previously demonstrated that human platelets activated with SFLLRN release PAR-1 activation peptide, PAR-1-(1–41), even in the presence of hirudin. This observation suggests that during their activation, platelets generate a protease that activates PAR-1. In this study, PAR-1 and -4 activation peptides were detected 10 s after ≤1.0 nm α-thrombin, 10 μm SFLLRN, or 100 μm AYPGKF were added to platelets. When SFLLRN or AYGPKF were added to platelets, generation of PAR-1 and -4 activation peptides was complete at 10 s. Generation of both PAR-1 and -4 activation peptides in response to 1 nm α-thrombin was significantly inhibited by affinity-purified anti-PAR-1-(35–62) IgY, anti-PAR-4-(34–54) IgY, and by the specific PAR-1 antagonist BMS 200261, but not by the PAR-4 antagonist YD3. Effective inhibition of platelet aggregation in response to 1.0 nm α-thrombin occurred only in the presence of both anti-PAR span antibodies. We conclude that platelet activation initiated with ≤1.0 nm α-thrombin proceeds via simultaneous PAR-1 and -4 activation. Inhibiting the activation of either PAR inhibits activation of the other. Both PAR-1 and -4 activation must be inhibited to prevent platelet activation subsequent to thrombin binding to platelets. The more efficient generation of PAR activation peptides by platelets activated with SFLLRN or AYGPKF, compared with α-thrombin, indicates that a platelet-derived serine protease that is inactivated by soybean trypsin inhibitor propagates PAR-1 and -4 activation.

The function of protease-activated receptor 1 (PAR-1) 2 in directing the responses of human platelets to various concen-trations of human ␣-thrombin has been well documented (1)(2)(3). Activation of human platelet PAR-1 arising from cleavage at Arg 41 -Ser 42 in response to ␣-thrombin and the simultaneous exposure of the previously cryptic tethered ligand domain (beginning with the sequence 42 SFLLRN), were first reported by Vu and colleagues (4). Two studies have confirmed cleavage at Arg 41 -Ser 42 by quantifying platelet PAR-1 activation peptide release (5,6). Monoclonal antibodies directed against the hirudin-like ␣-thrombin-binding domain of PAR-1 or against residues spanning the reported ␣-thrombin cleavage site (PAR-1 span antibodies) practically eliminated all responses of human platelets to Յ1 nM ␣-thrombin. However, Ն10 nM ␣-thrombin overcame inhibition by either PAR-1 monoclonal antibody, allowing human platelets to respond normally (5,7).
A second thrombin receptor on human platelets, namely PAR-4, lacks the hirudin-like ␣-thrombin-binding domain of PAR-1, and apparently requires Ն10.0 nM ␣-thrombin for its activation and participation in human platelet activation (8 -10). Cleavage of human platelet PAR-4 at Arg 47 -Gly 48 and the release of PAR-4-(1-47) as a direct index of PAR-4 activation have not been reported. Whether Ն10 nM ␣-thrombin only activates PAR-4, or simultaneously activates PAR-1 when human platelets are activated in the presence of the monoclonal anti-PAR-1 antibodies above has not been reported (5,7,9). Dual protease-activated receptors, namely PAR-3 and -4, also govern the responses of mouse platelets to thrombin in vivo and in vitro (11)(12)(13). PAR-3 serves as a co-factor for PAR-4 activation on mouse platelets (11,12) by altering exosite 1 of murine ␣-thrombin in a manner that promotes the diffusion of PAR-1 into the active center of murine ␣-thrombin (14).
PAR-1 and -4 on human platelets are activated and signal inter-dependently. A recent study has reported that dimerization of PAR-1 and -4 on human platelets facilitates PAR-4 cleavage by ␣-thrombin (15). We have reported previously that SFLLRN propagates PAR-1 activation measured as the [PAR-1- ] that is released into activated platelet supernatants (16). The present study investigates whether PAR-1 activation also facilitates activation of PAR-4 on human platelets. Using a new ELISA developed for quantifying PAR-4 activation peptide, PAR-4-(1-47), and our previous ELISA for quantifying PAR-1 activation peptide, PAR-1-(1-41) (5,16), the primary goal of this study was to compare the relative contributions of PAR-1 and -4 activation to the responses of human platelets to ␣-thrombin. This study also determined whether activation of either PAR-1 or -4 alone could sustain platelet activation in response to Յ1.0 nM ␣-thrombin.
Platelet Preparation and Characterization-Platelets were isolated from fresh citrated blood from healthy volunteers and resuspended at 5 ϫ 10 11 platelets/liter in a modified Tyrode buffer (0.136 M NaCl, 2.5 mM KCl, 12 mM NaHCO 3 , 0.42 mM NaH 2 PO 4 , 1 mM MgCl 2 , 2 mM CaCl 2 , 5 mM HEPES, 0.1% glucose, and 0.35% bovine serum albumin, pH 7.15) using the procedures of Mustard et al. (20). Washed platelets were incubated at 37°C in the absence (control) or presence of synthetic PAR-1 or PAR-4 antagonists or anti-PAR-1 or anti-PAR-4 antibodies for 30 min prior to activation. The volunteer blood donors had not taken any medication known to affect platelet function in the previous 14 days and provided informed consent. Ethics approval was obtained from the Research Ethics Board, Faculty of Health Sciences, McMaster University.
For serotonin-release studies, washed platelets were preloaded with [ 14 C]serotonin (1 Ci/40 ml; Amersham Biosciences Inc.) at 37°C followed by treatment with 1 M imipramine to prevent re-uptake of serotonin. Release of [ 14 C]serotonin and aggregation of washed human platelets in response to ␣-thrombin were quantified using procedures described previously (5). Expression of CD62 and CD63 on washed human platelets were quantified by flow cytometry using FITC-labeled CD63 antibody (Immunotech, Beckman Coulter Co., Fullerton, CA) and phycoerythrin-labeled CD62 antibody (BD Biosciences) as described previously (5). PAR-1 cleavage at Arg 41 -Ser 42 and release of PAR-1 activation peptide, PAR-1-(1-41), in the supernatants of washed platelets was quantified by an ELISA as described previously (16). PAR-4 cleavage at Arg 47 -Gly 48 was also measured by an ELISA as the release of PAR-4-(1-47) as described below.

Measurement of Changes in Intracellular [Ca 2ϩ
]-For intracellular calcium mobilization studies, 10 M Fura-PE3 (AM) (TEF Labs, Austin, TX) was added to the platelet-rich plasma, incubated for 45 min at room temperature in the dark, followed by the washing procedure outlined above for isolating washed platelets. Platelets were preincubated with inhibitors as described. Aliquots (190 l) of washed platelets (5 ϫ 10 11 /liter) loaded with 10 M Fura-PE3 (AM) were added in duplicate to black 96-well plates. Using the Fluoroskan Ascent microplate fluorometer (Thermo Labsystems, Thermo Electron Corp., Milford, MA), the Fura-PE3 (AM)-labeled platelets were sub-jected to excitation wavelength of 340 nm while measuring emission at 510 nm. The extracellular [CaCl 2 ] was 2 mM. The intensity of fluorescence was measured for 120 s after addition of 10 l of agonist. After each experiment the platelets were treated with 0.2% Triton X-100 followed by the addition of 10 mM EGTA to obtain the maximum and minimum fluorescence, respectively. The [Ca 2ϩ ] mobilized was calculated according to the equation [Ca 2ϩ ] ϭ K d (F Ϫ F min )/(F max Ϫ F) (22) using a dissociation constant for Fura-PE3 (AM) and Ca 2ϩ of 264 nM. After 2 min, the total area under the curve was calculated.
Measurement of ATP Release-Washed platelets (5 ϫ 10 11 / liter) were incubated with PAR-1 and -4 antagonists as previously described. Control and treated washed platelet preparations were stimulated with 0.5 or 1.0 nM ␣-thrombin, 10 M SFLLRN, or 100 M AYGPKF without stirring. After 180 s, 50-l aliquots were added to 200 l of a solution containing 1 M FPR-chloromethyl ketone (Calbiochem), 1 M D-EGR-chloromethyl ketone (Calbiochem), and 6.25 mM EDTA, and mixed well. The samples were then centrifuged for 2 min at 10,000 ϫ g and the supernatants removed. ATP levels in the supernatants were determined in duplicate using the ATP determination kit (A-22066; Molecular Probes, Eugene, OR) according to the manufacturer's instructions. Luminescence of the samples was measured using a Tropix TR717 microplate luminometer (Perkin-Elmer Applied Biosystems, Foster City, CA). The [ATP] in the experimental samples were interpolated from a standard curve. The effect of each PAR antagonist on the concentration of ATP measured in the absence of platelets or platelet supernatants was insignificant (data not shown).
Flow Cytometric Quantification of the Density of PAR-1 on Washed Platelets-Washed platelet preparations were incubated in the absence or presence of 1.0 nM ␣-thrombin, 10 M SFLLRN, or 100 M AYPGKF without stirring for 180 s, and 5 mM EDTA and 1 M hirudin (Berlex, Canada, Inc. Pointe Claire, QB, Canada) were then added. The platelets were divided and incubated for 1 h with: 1) buffer alone; 2) 12.5 g/ml affinity purified anti-PAR-1 span-IgY, washed, then incubated for 1 h with 0.25 g/ml FITC-conjugated goat anti-chicken IgG (Abcam, Cambridge, MA); or 3) 1 g/ml phycoerythrin-conjugated WEDE15 (Immunotech, Marseille, France). Platelets incubated with FITC-conjugated goat anti-chicken IgG alone were indistinguishable from unlabeled platelets, as determined by flow cytometry. The expression of PAR-1 on the surfaces of 10,000 platelets was quantified by flow cytometry, as previously described (16).
Data Analysis-Statistical comparisons were made using t tests, paired t tests, and one-way ANOVA, as appropriate. p values Ͻ 0.05 were considered to be significant. The data were analyzed using GraphPad Prism version 4.02 for Windows (GraphPad Software, San Diego, CA).
In initial experiments, platelets were incubated for up to 10 min with 0.5, 1.0, or 10 nM ␣-thrombin. At 10 s after ␣-thrombin addition, both PAR activation peptides were detected in all platelet supernatants indicating that activation of both PAR-1 and -4 begins rapidly even in response to 0.5 nM ␣-thrombin (supplementary materials). The concentration of PAR peptide released depended on the initial concentration of ␣-thrombin but only for the first 10 s. Less PAR-1 and -4 activation peptides were generated at 10 min in response to 10 nM ␣-thrombin, than were generated in response to the two lower thrombin concentrations. The observation that more PAR-1 and -4 were activated over 10 min in response to Յ1.0 nM than 10 nM ␣-thrombin supports previous reports that a platelet-derived protease, sensitive to inhibition by SBTI and Pefabloc SC, helps to propagate PAR-1 activation (16) and that PAR-1-(1-41) can activate platelets (24). Furthermore, PAR-1 activation peptide generation by this protease is not inhibited by hirudin, FPRchloromethyl ketone, or tissue factor pathway inhibitor. Platelet aggregation responses to SFLLRN, AYPGKF, or ␣-thrombin were abrogated by soybean trypsin inhibitor or 4 mM Pefabloc SC. In contrast, platelet aggregation responses to SFLLRN or AYPGKF were unaffected by 1 M FPR-chloromethyl ketone or 1 M D-EGR-chloromethyl ketone (data not shown).
To determine whether other platelet agonists also elicit PAR-1 and -4 activation peptide generation, platelets were incubated with 500 M arachidonic acid, 2 g/ml collagen, or 10 M ADP plus 40 g/ml fibrinogen. PAR-1 and -4 activation peptides were not released in response to collagen or ADP/ fibrinogen. However, PAR-1 and -4 peptides were released in response to arachidonic acid. The profiles describing the concentrations of the two activation peptides generated were similar to those obtained with SFLLRN and AYPGKF (data not shown). The platelets aggregated as expected in response to arachidonic acid, collagen, and ADP/fibrinogen (58 Ϯ 6, 69 Ϯ 4, and 40 Ϯ 10%, respectively; mean Ϯ S.E., for 4 determinations). Therefore, generation of PAR-1 and -4 activation peptides is not a general feature of all agonist-mediated platelet activation.
The early differences in PAR-1 and PAR-4 activation peptide release in response to 0.5 or 1.0 nM ␣-thrombin are reflected in differences in all the platelet responses measured. As summarized in Table 1, platelets incubated with 1 nM ␣-thrombin aggregated significantly more, released more [ 14 C]serotonin and ATP, and mobilized 2-fold more intracellular calcium, than did platelets stimulated with 0.5 nM ␣-thrombin. Furthermore, more platelets expressed CD62 and CD63 on their surfaces in response to 1.0 nM than in response to 0.5 nM ␣-thrombin at all 3 time points, with expression of both markers approaching their maxima at 60 s.
To explore possible connections between PAR-1 and PAR-4 activation and the subsequent platelet aggregation, the Spearman correlation coefficients relating percent platelet aggregation after 180 s with the levels of the peptides in platelet supernatants at 10, 60, and 180 s were calculated (Table 2). (The Spearman correlation coefficient (r) quantifies the degree to which the ranks of two variables vary together, with an r ϭ 1 indicating a perfect correlation and r ϭ 0 indicating that the two variables do not vary together.) The percent platelet aggregation achieved with 0.5 or 1.0 nM ␣-thrombin best correlated with the concentrations of PAR-1 and -4 activation peptides in the platelet supernatants at 10 and 60 s. Thus, in the first 60 s after ␣-thrombin addition, activation of PAR-1 and -4 partially determines the platelet aggregation response. There was no such correlation between the release of PAR-1 and -4 activation peptides and platelet aggregation in response to either SFLLRN or AYPGKF.
Simultaneous Inhibition of PAR-1 and -4 Blocks Platelet Responses to 1.0 nM ␣-Thrombin-The involvement of both PAR-1 and -4 in propagating the responses of human platelets to ␣-thrombin was explored using anti-PAR-1 span and anti-PAR-4 span antibodies. Because anti-PAR-1 span IgY eliminated release of PAR-1 and -4 activation peptides and aggregation of most washed platelet preparations in response to 0.5 nM ␣-thrombin, we used 1 nM ␣-thrombin for these studies. Anti-

TABLE 1
Platelets respond more effectively to 1.0 nM than to 0.5 nM ␣-thrombin Washed human platelets (5 ϫ 10 11 /liter) were incubated with 0.5 or 1.0 nM human ␣-thrombin and platelet responses were followed as described under "Experimental Procedures." The data presented are mean Ϯ S.E. (n). The p values were calculated using Student's t test, comparing 0.5 to 1.0 nM ␣-thrombin as platelet agonists.  PAR-1 span IgY reduced levels of both PAR-1 and -4 activation peptides in platelet supernatants at 10 (Table 3) and 60 s (not shown), demonstrating that inhibition of PAR-1 cleavage also delays and inhibits PAR-4 cleavage. To a lesser degree, anti-PAR-4 span IgY also inhibited both PAR-1 and PAR-4 cleavage at 10 (Table 3) and 60 s (not shown). By 180 s, however, inhibition of PAR activation by either antibody had become less marked (not shown). When both antibodies were added together, PAR-1 cleavage was eliminated and PAR-4 cleavage was greatly reduced. When platelets were stimulated with 1 nM ␣-thrombin, elimination of PAR-1 activation peptide generation by the combination of anti-PAR-1 span and anti-PAR-4 span was not associated with the elimination of PAR-4 activation peptide generation. Thus, PAR-4 activation alone in this case was sufficient to allow 30% aggregation of the platelets (Table 3). When anti-PAR-1 span IgY was present, aggregation was significantly inhibited, whereas release of [ 14 C]serotonin and ATP, changes in intra-platelet [Ca 2ϩ ] i , and expression of CD62 and CD63 initiated by 1.0 nM ␣-thrombin were all significantly reduced. Anti-PAR-4 span IgY inhibited these platelet responses less effectively than did anti-PAR-1 span IgY. In the presence of both antibodies, however, platelet responses to 1.0 nM ␣-thrombin were inhibited more effectively than by either antibody alone. Thus, simultaneous PAR-1 and -4 inhibition effectively blocks platelet signaling initiated by 1 nM ␣-thrombin.
Neither anti-PAR-1 span nor anti-PAR-4 span IgY alone inhibited aggregation or [ 14 C]serotonin release by platelets in response to SFLLRN (not shown). However, when both antibodies were added together, aggregation of and [ 14 C]serotonin release by platelets in response to SFLLRN were slightly (ϳ10%), but consistently inhibited (not shown). Based on the inability of the anti-PAR antibodies to significantly alter platelet responses to SFLLRN, we conclude that generation of the tethered ligand domain of PAR-1 in situ may attenuate the ability of either anti-PAR-1 span or anti-PAR-4 span IgY to inhibit platelet responses to ␣-thrombin.
Platelet  (Table 4). BMS 200261 did not inhibit platelet responses to AYGPKF, but partially inhibited platelet responses to ␣-thrombin. Therefore, BMS 200261, at a concentration that completely inhibits aggregation in response to SFLLRN, does not fully block platelet responses to 1 nM ␣-thrombin.
YD3 (2.8 M), a specific PAR-4 antagonist (19), blocked ATP release and aggregation in response to AYPGKF, without entirely preventing the expression of CD62, or the generation of PAR-1 and -4 activation peptides. YD3 had no effect on platelet responses to SFLLRN, and only slightly inhibited ATP release in response to ␣-thrombin. Therefore, YD3, at a concentration that completely inhibits platelet aggregation in response to AYPGKF, has little effect on platelet responses to ␣-thrombin. When PAR-1 signaling is blocked, generation of PAR-1 and -4 activation peptides in response to ␣-thrombin are dampened by about 50%. In contrast, when PAR-4 signaling is blocked, generation of the activation peptides is unaffected. We conclude that responses to ␣-thrombin are primarily mediated through PAR-1, but under conditions where PAR-1 signaling is blocked, for example, by BMS 200261 or the combination of anti-PAR-1 span IgY and anti-PAR-4 span IgY, signaling through PAR-4 allows the platelets to respond to 1 nM ␣-thrombin.
AYPGKF Increases the Density of PAR-1 on Platelets-We used two antibodies to probe the effects of ␣-thrombin, SFLLRN, and AYPGKF on the density of PAR-1 on platelets. The monoclonal anti-PAR-1 antibody, WEDE15, is directed against the hirudin-like domain of PAR-1 to which ␣-thrombin

TABLE 3 Anti-PAR-1 span IgY and anti-PAR-4 span IgY together block PAR-1 and PAR-4 activation and other platelet responses to 1.0 nM ␣-thrombin
Washed human platelets were resuspended in buffer (5 ϫ 10 11 /liter), incubated for 30 min in the presence or absence (control) of ϳ170 nM (25 g/ml) anti-PAR-1 span IgY, ϳ340 nM (50 g/ml) anti-PAR-4 span IgY, or 25 g/ml anti-PAR-1 span IgY plus 50 g/ml anti-PAR-4 span IgY, then activated with 1 nM ␣-thrombin. Aliquots were removed at 10, 60 (not shown), and 180 s (not shown), and ͓PAR-1-(1-41)͔ and ͓PAR-4-(1-47)͔ were measured in the platelet supernatants by specific ELISAs. ͓ATP͔ and ͓ 14 C͔serotonin released into the platelet supernatants were measured after 180 s. Aliquots were removed and fixed at 180 s, and the percentages of platelets expressing CD62 and CD63 were determined by flow cytometry. Changes in intracellular ͓Ca 2ϩ ͔ were measured fluorometrically for 120 s after the addition of agonist, and the area under the curve was calculated. The data are expressed as % inhibition (mean Ϯ S.E.), for at least 3 platelet preparations. binds and detects both activated (i.e. cleaved at Arg 41 -Ser 42 ) and intact forms of the receptor (7). To ensure that continuing occupancy of this site by ␣-thrombin did not influence the binding of any of the antibodies to platelets, the ␣-thrombin bound to platelets was displaced by a 1000-fold molar excess of hirudin. As summarized in Table 5, about 35% of resting platelets bound WEDE15 and neither ␣-thrombin nor SFLLRN affected this percentage. Therefore, as expected, generation of the PAR-1 activation peptide did not influence the binding of WEDE15 to platelets. However, AYPGKF almost doubled the proportion of platelets that bound WEDE15, suggesting that initiation of signaling essentially through PAR-4 increases PAR-1 density on the surface of the platelets, presumably through externalization of intra-platelet PAR-1.

% Inhibition
The second anti-PAR-1 antibody preparation used was affinity purified chicken anti-human PAR-1-(35-62) (i.e. PAR-1 span) IgY, which we hypothesized would bind both native and cleaved PAR-1. However, we expected the affinity of this antibody for cleaved PAR-1 to be reduced due to the loss of the amino-terminal third of the region of PAR-1 against which the antibody was raised. The percentage of platelets labeled by anti-PAR-1 span IgY decreased after incubation with ␣-thrombin or SFLLRN, probably reflecting the loss of affinity predicted for the cleaved form of the receptor. After incubation with AYPGKF the percentage of labeled platelets does not decrease, despite the normal release of PAR-1-(1-41) into the platelet supernatant.
These experiments demonstrate the following. (i) Activation of platelets with ␣-thrombin or SFLLRN results in PAR-1 cleavage at Arg 41 , as detected by the release of PAR-1-(1-41) into platelet supernatants, and a reduction in the percentage of platelets expressing native PAR-1 as detected with anti-PAR-1 span IgY. (ii) Activation of platelets with AYPGKF results in the release of PAR-1-(1-41) and simultaneously nearly doubles the percentage of platelets expressing PAR-1, as detected by WEDE15. (iii) Activation of the platelets with ␣-thrombin, SFLLRN, or AYPGKF results in the release of PAR-4-(1-47) into the platelet supernatants. We conclude, therefore, that both PAR-1 and -4-tethered ligand agonists facilitate the simultaneous propagation of PAR-1 and -4 activation.

DISCUSSION
Simultaneous activation of PAR-1 and PAR-4 on human platelets, estimated as generation of PAR-1 and -4 activation peptides, and observed Ն10 s after adding subnanomolar ␣-thrombin to platelets, refutes prior reports that activation of PAR-4 on human platelets requires Ն10 nM ␣-thrombin (9 -11, 25, 26). Unlike the previous studies, the current study directly measured ␣-thrombin-induced generation of PAR-1 and -4 activation peptides beginning 10 s after this potent platelet agonist was added to platelets. Our data concur with previous observations that some anti-PAR-1 antibodies can prevent platelets from responding to Յ1.0 nM ␣-thrombin. In addition, this study provides new information demonstrating that when the anti-PAR-1 span IgY prevented platelets from aggregating, prevention of PAR-1 and -4 activation peptide generation and elimination of other platelet responses were also observed. Furthermore, when the anti-PAR-1 span IgY was also present, generation of both PAR-1 and -4 activation peptides preceded the partial restoration of the other platelet responses achievable with 1 nM ␣-thrombin. Thus, simultaneous PAR-1 and -4 activation occur after platelets are incubated with any concentration of ␣-thrombin that can effect their activation. The physiological relevance of our observations is that the concentrations of ␣-thrombin endogenous to most normal plasmas (ϳ1 nM) (27,28) should be able to effectively activate platelets, if gener-

Contributions of PAR-1 and -4 signaling pathways to thrombin-mediated platelet responses
Washed human platelets were resuspended in buffer (5 ϫ 10 11 /liter), incubated for 30 min in the presence or absence (control) of 1 M BMS 200261 or 2.8 M YD3, then activated with 1 nM ␣-thrombin, 10 M SFLLRN, or 100 M AYPGKF. Aliquots were removed at 10 (not shown), 60, and 180 s (not shown), and ͓PAR-1-(1-41)͔ and ͓PAR-4-(1-47)͔ were measured in the platelet supernatants by specific ELISAs. ͓ATP͔ released into the platelet supernatants was measured after 180 s. Aliquots were removed and fixed at 180 s, and the percentages of platelets expressing CD62 and CD63 (not shown) were determined by flow cytometry. Changes in intracellular ͓Ca 2ϩ ͔ were measured fluorometrically for 120 s after the addition of agonist, and the area under the curve was calculated. The data are expressed as % inhibition (mean Ϯ S.E.), for at least 3 platelet preparations.

Agonist
Inhibitor % Inhibition (mean ؎ S.E.) ͓PAR-1- (1-41) OCTOBER 3, 2008 • VOLUME 283 • NUMBER 40 ated, at sites of vascular injury in vivo. There is good evidence for both increased platelet and prothrombin activation in patients with diabetes and arterial thrombosis, conditions well known to be associated with the development of heart disease and stroke (29 -32). Whether some or all of the increased prothrombin and/or platelet activation seen in these cases occur in the extravascular or intravascular space is not known. Nonetheless, the parallel increases in platelet and prothrombin activation raises the possibility that the two events are related. Furthermore, the relatively high proportion of diabetic patients who are aspirin resistant (33)(34)(35)(36) is consistent with thrombin participating in platelet activation in vivo in these cases.

Simultaneous Generation of PAR-1 and -4 Activation Peptides
The maximum concentrations of PAR-1 and -4 activation peptides had been generated 10 s after platelets were incubated with 10 nM ␣-thrombin, 10 M SFLLRN, 100 M AYPGKF, or 500 M arachidonic acid. Furthermore, additions of each agent to platelets resulted in the generation at 10 s of similar concentrations of PAR-1 activation peptide and similar but lower concentrations of PAR-4 activation peptide. These observations suggest that release of the PAR activation peptides is mediated, at least in part, by an unidentified endogenous platelet protease. Generation of PAR-1-(1-41) in response to ␣-thrombin or SFLLRN resulted in a decreased density of PAR-1 on the platelets as estimated with anti-PAR-1 span IgY.
Also consistent with the idea that an endogenous trypsin-like platelet protease activates PAR-1 and -4, soybean trypsin inhibitor and Pefabloc SC eliminated platelet aggregation responses to ␣-thrombin, SFLLRN, or AYPGKF. Furthermore, 1 M hirudin did not inhibit platelet responses to SFLLRN or AYPGKF (Ref. 16 and this study). Added to platelets before thrombin, 1 M hirudin prevented binding of up to 100 nM thrombin to platelets or their activation (37). Platelets incubated for 3 min with 10 M SFLLRN released ϳ700 pM prothrombin, 20 pM factor VII, 40 pM factor IX, and 20 pM factor X. 3 Even if all of the 700 pM prothrombin thus made available was converted to thrombin, it is unlikely that the thrombin could escape inhibition in the presence of 1 M hirudin to participate in PAR-1 and -4 activation peptide generation. In addition, aprotinin, a Kunitz-type trypsin inhibitor that does not inactivate thrombin, inhibits platelet PAR-1 activation during cardiopulmonary bypass surgery (38,39). We had previously reported that the tissue factor pathway inhibitor, leupeptin (an inhibitor of serine/cysteine proteases), and several specific inhibitors of metalloproteases, cathepsin A and B, aminopeptidases, chymotrypsin, cysteine proteases, and aspartyl proteases did not inhibit platelet activation or PAR-1 activation peptide generation in response to ␣-thrombin or SFLLRN (16). Thus, only the two Kunitz-type trypsin inhibitors that do not inactivate ␣-thrombin can inhibit platelet responses to ␣-thrombin (16,38,39).
Measurement of PAR-1-(1-41) and PAR-4-(1-47) in platelet supernatants by the specific ELISAs provides direct evidence for the generation of PAR-1 and -4-tethered ligands in situ, and at concentrations equimolar to their respective activation peptides, during platelet activation. The early differences (at 10 s) in PAR-1 and PAR-4 activation peptide release in response to 0.5 or 1.0 nM ␣-thrombin are reflected in all the platelet responses measured (Table 1). Thus, early activation of PAR-1 and -4 after ␣-thrombin addition at least partially determines the platelet aggregation response.
Endogenously generated PAR-1-tethered ligand propagates platelet activation in response to ␣-thrombin (3,4,40). We used the specific SFLLRN antagonist BMS 200261 (18) to explore the contributions of the PAR-1-tethered ligand domain generated in situ to the propagation of both PAR-1 and PAR-4 activation peptide generation. Herein, we confirm that 1 M BMS 200261 prevented SFLLRN-mediated platelet aggregation as reported previously (9,18). Furthermore, this PAR-1 antagonist effectively inhibited [ 14 C]serotonin and ATP release, and CD62 and CD63 expression on platelets in response to 10 M SFLLRN. Simultaneously, BMS 200261 significantly inhibited PAR-1 and -4 activation peptide generation. Furthermore, BMS 200261 partially inhibited generation of both PAR-1 and -4 activation peptides and, with the exception of aggregation, all the other measured platelet responses to Յ1.0 nM ␣-thrombin. That BMS 200261 inhibits the generation of PAR-1 and -4 activation peptides suggests that the exposed tethered ligand domain of PAR-1, generated in situ in response to ␣-thrombin, propagates both PAR-1 and -4 activation. The reported K D for the binding of SFLLRN to platelets is 15 nM (41). The IC 50 describing the inhibition by BMS 200261 of SFLLRN binding to platelets is 7.5 nM (18). Furthermore, 21 nM BMS 200261 inhibited platelet aggregation in response to 3 M SFLLRN by 50% (18). Therefore, it is unlikely that the Յ1 nM of the endogenous tethered ligand domain of PAR-1 generated in situ in response to Յ1 nM ␣-thrombin would propagate any PAR-1-dependent platelet signaling in the presence of 1 M BMS 200261.
Concentrations of anti-PAR-1 span IgY that prevented PAR-1 activation in response to 0.5 nM ␣-thrombin, also prevented PAR-4-(1-47) generation and all the other platelet responses including aggregation, changes in intracellular calcium concentrations, ATP release, and CD62 and CD63 expression. The molar concentration of the anti-PAR-1 span IgY used was ϳ6-fold less than that of BMS 200261 used but this anti-PAR-1 antibody was more effective in preventing platelet responses to subnanomolar ␣-thrombin. Thus, to suppress platelet activation, prevention of PAR-1 and -4 activation peptide generation was more effective than blocking the intramolecular binding of the PAR-1-tethered ligand to its binding site on PAR-1. This may in part reflect the ability of PAR-1-(1-41) to activate platelets as effectively as equimolar SFLLRN (23,24).
We also investigated the effects of anti-PAR-4 span IgY on PAR-1 activation. Anti-PAR-4 span IgY inhibited both PAR-1 and -4 activation but did so less effectively than anti-PAR-1 span IgY. Inhibition of PAR-1 and -4 activation by anti-PAR-4 span IgY, however, only slightly impaired the other platelet responses to ␣-thrombin. When the anti-PAR-1 and -4 antibodies were combined, platelet responses to 1.0 nM ␣-thrombin were then inhibited more effectively than by either antibody alone. Thus, simultaneous PAR-1 and -4 inhibition attenuates platelet signaling initiated by 1 nM ␣-thrombin. Therefore, signaling reactions arising from both PAR-1 and -4 activation help to propagate platelet activation-related responses to Յ1.0 nM ␣-thrombin. The dual roles of PAR-1 and PAR-4 in directing the responses of human platelets to concentrations of thrombin found physiologically are consistent with the dual roles of PAR-3 and -4 for the effective hemostatic functions of mouse platelets (13,42).
Our results support a hypothesis that platelets stimulated with SFLLRN, AYPGKF, or ␣-thrombin make available in situ an endogenous trypsin-like platelet protease able to effectively activate both PAR-1 and -4. Because BMS 200261 specifically blocks PAR-1 tethered ligand-mediated responses (Ref. 18 and this study), propagation of ␣-thrombin-initiated platelet responses, in the presence of BMS 200261, probably proceed exclusively via PAR-4 signaling. The tethered ligand of PAR-1 propagates PAR-1 and -4 activation and the PAR-4-tethered ligand similarly propagates PAR-1 and -4 cleavages. Simultaneous activation of platelet PAR-1 and -4 observed in response to Յ1.0 nM ␣-thrombin may explain, in part, why highly potent PAR-1 antagonists do not fully eliminate the responses of human platelets to physiological concentrations of ␣-thrombin in vitro (18,43).