Characterization of the Initial α-Thrombin Interaction with Glycoprotein Ibα in Relation to Platelet Activation*

We have evaluated the properties of α-thrombin interaction with platelets within 1 min from exposure to the agonist, a time frame during which most induced activation responses are initiated and completed. Binding at 37 °C was rapidly reversible and completely blocked by a monoclonal antibody, LJ-Ib10, previously shown to be directed against the α-thrombin interaction site on glycoprotein (GP) Ibα. By 2–5 min, however, binding was no longer fully reversible and was only partially inhibited by the anti-GP Ibα antibody. Results were similar at room temperature (22–25 °C), whereas the initial characteristics of α-thrombin interaction with platelets were preserved for at least 20 min at 4 °C. Equilibrium binding isotherms obtained at the latter temperature were compatible with a two-site model, but the component ascribed to GP Ibα, completely inhibited by LJ-Ib10, had “moderate” affinity (k d on the order of 10−8 m) and relatively high capacity, rather than “high” affinity (k d on the order of 10−10 m) and low capacity as currently thought. The parameters of α-thrombin binding to intact GP Ibα on platelets at 4 °C corresponded closely to those measured with isolated GP Ibα fragments regardless of temperature. Blocking the α-thrombin-GP Ibα interaction caused partial inhibition of ATP release and prevented the association with platelets of measurable proteolytic activity. These results support the concept that GP Ibα contributes to the thrombogenic potential of α-thrombin.

Platelet deposition at sites of vascular injury is thought to be enhanced by ␣-thrombin during normal hemostasis as well as pathological arterial thrombosis (1-3), but the mechanisms responsible for this effect have yet to be fully understood. There is evidence that ␣-thrombin-induced platelet activation is initiated by coupling of the agonist to specific receptors (4 -6), whose nature is still a topic of debate. In this context, it is recognized that glycoprotein (GP) 1 Ib␣, a component of the GP Ib-IX-V complex (7,8), binds ␣-thrombin possibly with affinity higher than other sites on platelets (6,9,10). This interaction has been variably judged as functionally relevant (10 -12) or irrelevant (13) or even as a negative regulator acting through sequestration of the enzyme (14). It is also known that ␣-thrombin cleaves GP V (15,16) but with no apparent relation to platelet activation (17,18).
The agonist activity of ␣-thrombin depends on proteolysis, a fact explained with the identification of a seven-transmembrane domain receptor stimulated by a tethered ligand sequence exposed as the new amino terminus of the molecule after cleavage of an internal Arg-Ser bond (19). This proteaseactivated receptor, PAR1, exemplifies an effector mechanism common to a family of related proteins exhibiting distinct specificities as substrates for different proteases (20). Because the function of PAR1 seemed to explain many of ␣-thrombin effects on platelets, it was surprising that deletion of the homologous mouse gene, albeit lethal in many homozygous embryos, failed to result in decreased thrombogenic potential in the animals born alive (21). The subsequent demonstration on platelets of PAR3, another member of the family with specificity similar to PAR1 (22), provided a reasonable solution to the puzzle and reinforced the concept that a protease-activated receptor pathway is crucially involved in mediating responses to ␣-thrombin. Yet the participation of GP Ib␣ in these processes remains a possibility that must be addressed conclusively.
There are apparent contradictions in the reported characteristics of ␣-thrombin binding to platelets. Only few hundred high affinity sites have been ascribed to GP Ib␣ (6,10), but the latter is present in greater number on the membrane (23). Moreover, a specific anti-GP Ib␣ antibody has been shown to block the interaction of approximately 5,000 ␣-thrombin molecules with platelets, abolishing higher affinity sites but also decreasing markedly the moderate affinity ones (10). Finally, the apparent k d of the highest affinity sites on platelets, 0.25-1.3 nM (10), is substantially lower than that reported for ␣-thrombin interaction with the isolated amino-terminal domain of GP Ib␣, approximately 20 nM (24). Platelet binding parameters have been typically deduced from experiments with relatively long incubation times, in contrast with the rapidity of platelet responses to ␣-thrombin stimulation (25)(26)(27), and may reflect events not relevant for activation. Indeed, the results presented here indicate that GP Ib␣ accounts for most of the initial ␣-thrombin binding to platelets but apparently with the same "moderate" affinity assigned to the corresponding isolated functional domain (24). Different conclusions in this regard may be explained by time-and temperature-dependent deviations from equilibrium conditions. The fully reversible ␣-thrombin interaction with GP Ib␣ supports the association with platelets of a proteolytically active enzyme that may contribute to activation.

EXPERIMENTAL PROCEDURES
Purification and Iodination of ␣-Thrombin-Purified human ␣-thrombin with specific clotting activity between 2,180 and 2,800 NIH units/mg (28) (a gift of Dr. John W. Fenton II, Wadsworth Center for Laboratories and Research, New York Department of Health, Albany) was radiolabeled with 125 I (Amersham Corp.) using IODO-GEN (Pierce) (29). The radiolabeled ligand, with specific radioactivity between 5 and 7 mCi/mg, was characterized and stored as described (10) and was used within 2 weeks of iodination.
Preparation of Washed Platelets-Blood was drawn from normal volunteers, who had denied ingestion of drugs known to interfere with platelet function for at least 1 week and given their informed consent to these experimental studies according to the declaration of Helsinki, and was collected into one-sixth final volume of citric acid/citrate/dextrose, pH 4.5, containing 25 nM prostaglandin E 1 (Sigma). Platelets were washed free of plasma constituents using the albumin density gradient method (30), with modifications previously described (31).
Antibodies for Inhibition Studies-The epitope of LJ-Ib10 lies between residues 238 and 290 of the amino-terminal domain of GP Ib␣. This monoclonal antibody inhibits ␣-thrombin binding to platelets (10,32) without effect on von Willebrand factor binding (33,34). The rabbit polyclonal antibody, anti-TR 1-160 , binds to one or more epitopes within the 160 amino-terminal residues of PAR1 and abolishes platelet activation induced by low doses of ␣-thrombin and by the SFLL peptide ligand (35). Purified IgG and divalent F(abЈ) 2 fragments, prepared as reported (36), were stored at Ϫ80°C in 20 mM Tris, 150 mM NaCl, pH 7.4, until used.
Binding of ␣-Thrombin to Platelets-Binding was measured according to a procedure described previously in detail (6,9), in the presence of a binding buffer composed of 25 mM Tris-HCl and 136 mM CH 3 CO 2 Na, pH 7.3, containing 0.6% polyethylene glycol (average molecular weight 6,000 -7,000; Serva, Heidelberg, Germany) and 1% bovine serum albumin (Sigma). Washed platelets were kept at the temperature selected for any given experiment for 15 min before use. Binding as a function of ligand concentration was measured by adding platelets (2.8 ϫ 10 8 /ml, final count) to a mixture composed of a constant concentration (0.1 nM) of 125 I-labeled ␣-thrombin and increasing concentrations (0.1-1000 nM) of nonlabeled ␣-thrombin. Each experimental mixture had a total volume of 125 l. After incubation, platelet-bound and free ␣-thrombin were separated by centrifugation through a 20% sucrose layer at 12,000 ϫ g for 4 min. Binding isotherms, each consisting of 20 experimental points, were analyzed using the COLD option of the computer-assisted program LIGAND, calculating nonsaturable binding as a fitted parameter (37,38). A concentration of 1 nM 125 I-␣thrombin was employed in time course and dissociation assays. In the latter, a 1,000-fold excess of nonlabeled ␣-thrombin was added to platelets after incubation with the labeled ligand.
Binding of ␣-Thrombin to Immobilized Glycocalicin-The extracytoplasmic domain of GP Ib␣ was purified from fresh platelet concentrates as reported (39). The glycoprotein was immobilized onto Sepharose CL 4B beads (Sigma) bearing covalently bound anti-GP Ib␣ monoclonal antibody, LJ-P3 (33), by incubating 800 l of packed beads with 2 ml of glycocalicin solution (0.5-2 mg/ml) for 1 h at 22-25°C with constant mixing. The beads were then washed twice with a buffer composed of 100 mM Tris, 500 mM LiCl 2 , 1 mM EDTA, pH 7.4, and 1 volume of packed beads was resuspended into 6 volumes of binding buffer containing 0.6% polyethylene glycol and 4.1% bovine serum albumin. This suspension was used immediately. The presence of purified glycocalicin on the beads was confirmed by SDS-polyacrylamide gel electrophoresis of protein eluted at pH 2.9 (39) and by measuring the binding of two different 125 I-labeled anti-GP Ib␣ monoclonal antibodies, LJ-P19 and LJ-Ib10 (24). The binding of 125 I-labeled ␣-thrombin was evaluated by mixing 20 l of bead suspension (corresponding to 3 l of packed beads) with 65 l of binding buffer, or other appropriate reagent, and 40 l of the desired ligand concentration. After incubation at the desired temperature, the radiolabeled ligand bound to the beads was separated from free ligand by centrifugation at 12,000 ϫ g for 4 min through a 20% sucrose layer. Binding isotherms were analyzed with the computer-assisted program LIGAND (37,38).
Flow Cytometric Analysis-Surface membrane expression of P-selectin (40,41) and GP Ib␣ was determined by using, respectively, an anti-CD62P monoclonal antibody (Becton-Dickinson) labeled with phycoerythrin (PE) and the anti-GP Ib␣ monoclonal antibody, LJ-Ib1 (33), labeled with fluorescein isothiocyanate (42). Washed platelets were stimulated with increasing concentrations of ␣-thrombin (0.01-100 nM) at a desired temperature, until recombinant hirudin (Iketon, Milan, Italy) was added at the final concentration of 200 NIH units/ml to neutralize the proteolytic activity. Platelets were then fixed with 1% paraformaldehyde for 30 min at 4°C, washed twice in Tris buffer, incubated for 15 min at 22-25°C with the specific antibodies or, in control experiments, with mouse IgG-labeled with the same fluorochromes (43), and analyzed in a flow cytometer (Becton-Dickinson). Fluorescence intensity was measured on an arbitrary scale, and platelets were considered positive for a given marker when their level of fluorescence was at least twice that of background or control platelets.
Measurement of Platelet ATP Secretion-The release of ATP from the dense granules of platelets was measured by the luciferin-luciferase assay (44). Washed platelets were resuspended in calcium-free Tyrode buffer (31) at a count of 2 ϫ 10 8 /ml, and 0.4 ml were mixed with 150 g/ml F(abЈ) 2 fragment of the anti-GP Ib␣ monoclonal antibody, LJ-Ib10, or the rabbit polyclonal anti-TR 1-160 antibody. In control experiments, the same F(abЈ) 2 fragment concentration of the anti-GP Ib␣ monoclonal antibody, LJ-Ib1, and of preimmune rabbit IgG was used as control. The mixtures were placed in a glass cuvette and stirred at 1,200 revolutions per min (rpm) with a Teflon-coated magnetic bar for 5 min at 37°C in a lumiaggregometer (Chrono-log Corp.). At the end of the incubation, 50 l of luciferin-luciferase (Chrono-lume reagent, Chronolog Corp.) was added, and platelet release was induced by the addition of ␣-thrombin at final concentration between 0.5 and 3 nM. Luminescence was recorded to monitor ATP release, measured by comparing peak height with that generated by known standard amounts of ATP.
Amidolytic Activity of ␣-Thrombin-Washed platelets at a count of 2.8 ϫ 10 8 /ml, treated with control buffer or test antibodies, were mixed with 3 nM ␣-thrombin and binding buffer to a volume of 0.6 ml. After incubation for 1 min at 37°C, platelets were sedimented by centrifugation through a 20% sucrose layer at 12,000 ϫ g for 2 min. The supernatant containing free ␣-thrombin was removed, the sucrose layer was discarded, and the platelet pellet was resuspended with 600 l of binding buffer. The chromogenic substrate S-2238 (Kabi) (45) was added into the supernatant as well as the resuspended platelets at the concentration of 0.4 mM, and the incubation was continued for 5 min at 37°C. The hydrolysis reaction was then stopped with 4% acetic acid, and the release of p-nitroaniline was measured at 405 nm in a spectrophotometer (Beckman DU-65) after removing the platelets by centrifugation at 12,000 ϫ g for 2 min.

Time Course and Reversibility of ␣-Thrombin Binding to
Platelets-The binding of 125 I-␣-thrombin to washed platelets reached a maximum in 5 min at 37°C but required 10 min and was about 30% lower at 4°C (Fig. 1). Concurrent addition of 1000-fold excess of nonlabeled ␣-thrombin inhibited the binding of labeled ligand by greater than 90% at either temperature (not shown). In contrast, addition of nonlabeled ligand 20 min after 125 I-␣-thrombin, when binding of the latter was maximum, resulted in 80 -95% dissociation of bound ligand at 4°C but only about 50 -60% at room temperature (22-25°C) and 25-30% at 37°C (Fig. 2). At the latter temperature, dissociation was approximately 70% when nonlabeled ligand was added 1 min after 125 I-␣-thrombin, 60% when it was added after 2 min, and 45% when it was added after 10 min (Fig. 2). The dissociation of bound ligand at 4°C was not only greater in extent but occurred more rapidly than at higher temperatures, being almost maximal in 1 min as opposed to 5 min (Fig. 2).
Inhibitory Effect of Antibody LJ-Ib10 on ␣-Thrombin Binding to Platelets and Purified Glycocalicin as a Function of Incubation Time and Temperature-The anti-GP Ib␣ monoclonal antibody, LJ-Ib10, inhibited the maximum binding of 125 I-␣-thrombin to platelets, measured after incubation of 20 min, by approximately 75% at 4°C but only 60% at 22-25°C and less than 50% at 37°C (Fig. 3). The same antibody inhibited the maximum binding to glycocalicin, the isolated extracytoplasmic domain of GP Ib␣, by at least 90% at all temperatures tested (Fig. 3). In the latter case, the degree of inhibition was equivalent to that produced by a 1000-fold excess of nonlabeled ligand added concurrently with 125 I-␣-thrombin (not shown). At 37°C, the time of incubation between 125 I-␣-thrombin and platelets influenced the inhibitory effect of LJ-Ib10 on the interaction. Inhibition was essentially complete, i.e. equivalent to that caused by a 1000-fold excess of unlabeled ligand, during the first 2 min of incubation, when binding reached approximately 50% of maximum (Fig. 4). With continuing incubation, however, the inhibitory effect of the antibody progressively decreased as compared with that seen with excess unla-beled ligand (Fig. 4). Altogether, the results shown in Figs. 1-4 are compatible with the hypothesis that ␣-thrombin binding to GP Ib␣ on platelets becomes progressively less reversible as a consequence of changes related to activation, as they occur better at 37°C than at lower temperatures. The fact that ␣-thrombin binding to isolated glycocalicin was inhibited by the antibody LJ-Ib10 in identical manner at all the temperatures tested is in agreement with such a concept.
Markers of ␣-Thrombin-induced Platelet Activation-The following experiments were performed to evaluate the time course of platelet stimulation by ␣-thrombin and correlate the membrane expression of an activation marker, P-selectin, with changes in the accessibility of GP Ib␣ to antibody probes. Greater than 50% of platelets incubated with ␣-thrombin concentrations as low as 0.1 nM for 20 min at 37°C exhibited increased P-selectin membrane expression, and greater than 80% was positive when the agonist concentration was in the range of 1-10 nM; in contrast, at 4°C there was no significant change relative to nonstimulated platelets even with concentrations as high as 100 nM (Fig. 5). The number of platelets displaying surface expression of P-selectin increased rapidly after stimulation with ␣-thrombin, reaching a maximum in 20 -40 s at 37°C (Fig. 5).
Exposure of platelets to ␣-thrombin at 4°C had no significant effect on the binding of an anti-GP Ib␣ antibody, whereas progressively lower binding as a function of agonist concentration was seen at 37°C (Fig. 6). Identical results were observed whether or not 2 mM Ca 2ϩ and/or 1 mM Mg 2ϩ was present in the incubation mixtures (data not shown). The observed changes in anti-GP Ib␣ antibody binding started after a time interval approximately 10-fold longer (Fig. 6) than required for increase in P-selectin surface expression (Fig. 5) following ␣-thrombin stimulation.
Effects of Temperature on ␣-Thrombin Binding to Platelets and Immobilized Glycocalicin-The concentration-dependent binding of ␣-thrombin to platelets was different at 4°C as compared with 37°C (Fig. 7). Scatchard-type analysis of data generated at 37°C, performed for comparative purposes even  Fig. 5, upper panel, except that 100 g/ml of fluorescein isothiocyanate-labeled anti-GP Ib␣ antibody, LJ-Ib1, was used instead of the anti-CD62P antibody. a, platelets stimulated at 37°C; b, platelets stimulated at 4°C. Curve 1, platelets stimulated with 1 nM ␣-thrombin; curve 2, platelets stimulated with 100 nM ␣-thrombin. The curves obtained at other thrombin concentrations have been omitted for graphical clarity. Curve NS, fluorescence distribution of nonstimulated platelets. The curve at the extreme left corresponds to background fluorescence. Similar results were obtained in three separate experiments. c, platelets were incubated with 1 nM ␣-thrombin at 37°C for the indicated periods before measuring anti-GP Ib␣ antibody binding (see above). Each point is the mean Ϯ S.E. of three experiments in which the median fluorescence intensity of stimulated platelets was expressed as percentage of that of nonstimulated ones. though binding was not at equilibrium (see above), resulted in a plot with downward concavity in the range of ligand concentrations between 0.1 and 3.5 nM and upward concavity at higher concentrations. The data generated at 4°C, on the other hand, yielded an upward concave Scatchard plot (Fig. 7). The results obtained at 4°C could be represent by a two-site model, with fitted nonsaturable binding in good agreement with experimentally determined values on the order of 2 ϫ 10 Ϫ2 (Table  I). In contrast to the results with platelets, the binding of 125 I-labeled ␣-thrombin to immobilized glycocalicin was the same at 4 and 37°C, yielding a linear Scatchard plot indicative of a single class of binding sites (Fig. 7) with a k d of 4 Ϯ 1 ϫ 10 Ϫ8 M (mean Ϯ S.E. of four experiments at each temperature). The latter corresponded closely to the k d of the lower affinity sites detected on platelets at 4°C (Table I). Binding isotherms generated at 4°C in the presence of the antibody LJ-Ib10 (Fig.  8) revealed selective and complete inhibition of the lower affinity sites but no effect on the higher affinity ones (Table I). Of note, the function blocking anti-PAR1 rabbit polyclonal antibody, anti-TR 1-160 , did not interfere with ␣-thrombin binding to platelets at 4°C (Table I).
Inhibition of ␣-Thrombin-induced ATP Secretion by Anti-GP Ib␣ and Anti-PAR1 Antibodies-Washed platelets stimulated with 3 nM ␣-thrombin at 37°C exhibited rapid ATP release. This was inhibited approximately 50% by LJ-Ib10 and 80% by anti-TR 1-160 but greater than 90% when the two were used together (Fig. 9). The difference between the effect of anti-TR 1-160 alone and in combination with LJ-Ib10 was significant (t test: p ϭ 0.01). Since release of ATP always occurs in parallel with that of the platelet agonist, ADP (46), these results indicate that ␣-thrombin binding to GP Ib␣ is associated with a response capable of enhancing platelet pro-thrombotic functions.
Amidolytic Activity of ␣-Thrombin Associated with GP Ib␣-Measurable amidolytic activity could be partitioned with platelets after incubation with ␣-thrombin, and this association was selectively inhibited by LJ-Ib10 (Fig. 10). The appearance of platelet-associated ␣-thrombin activity corresponded to an equivalent decrease measured in the suspension medium after platelet separation, and such a decrease was also blocked selectively by LJ-Ib10 (Fig. 10). Thus, ␣-thrombin that had become associated with GP Ib␣ retained similar amidolytic activity as the free enzyme. DISCUSSION The effects of ␣-thrombin on platelets are rapidly manifest (27), and most of the induced responses may reach completion in a time frame of seconds (25,26). Here we show that within 1 min from exposure to the agonist at 37°C, a time sufficient for maximal activation as judged by surface expression of Pselectin, the initial ␣-thrombin interaction with platelets is  I Parameters of ␣-thrombin binding to platelets Control indicates normal platelets; LJ-Ib10 and anti-TR 1-160 indicate normal platelets treated with saturating amounts (150 g/ml) of F(abЈ) 2 fragment of the anti-GP Ib␣ or anti-PAR1 antibody, respectively. All parameters and the corresponding standard error of estimated values were calculated with LIGAND (37,38 rapidly reversible and completely blocked by a specific anti-GP Ib␣ antibody. By 5-10 min, however, ␣-thrombin binding to platelets is less promptly reversible and only partially blocked by the anti-GP Ib␣ antibody. These changes occur in temporal relationship with decreased accessibility of GP Ib␣ to antibody probes (43,(47)(48)(49), an event that is also known to correlate with decreased von Willebrand factor binding (50), and both are prevented at 4°C. The latter temperature, therefore, appears to preserve in time the characteristics of ␣-thrombin binding to GP Ib␣ as during the initial interaction with platelets, thus providing appropriate conditions to obtain equilibrium binding isotherms. Accordingly, in the time frame relevant for agonist-induced activation, GP Ib␣ may bind ␣-thrombin with "intermediate" or moderate affinity (k d on the order of 10 Ϫ8 M), rather than being the "highest" affinity site (k d on the order of 10 Ϫ10 M) as currently thought (6,10,12). Our findings also indicate that selective blockade of ␣-thrombin interaction with GP Ib␣ dampens responses to the agonist and prevents the association with platelets of a proteolytically active, thus potentially procoagulant, enzyme. Such conclusions are in agreement with recent evidence showing that kininogens inhibit ␣-thrombin-induced platelet aggregation because they interfere with agonist binding to GP Ib␣ (51). Indeed, owing to its relatively high membrane density, the function of GP Ib␣ may be relevant in localizing ␣-thrombin at sites of vascular injury, thus facilitating its action on specific substrates. Application of Scatchard-type analysis to ␣-thrombin-plate-let binding isotherms has usually resulted in curvilinear, upwardly concave plots indicative of a deviation from the simplest model of reversible ligand interaction with a homogeneous class of noninteracting receptors (52). This lack of uniformity has been interpreted as evidence for the presence of more than one type of receptor (6,10,12), leading to the proposed existence of three ␣-thrombin binding sites with high, intermediate, and low affinity (k d of approximately 0.3, 30, and 3000 nM, respectively) without nonspecific binding (6). Even considering the more likely possibility that the low affinity site represents nonspecific binding with respect to physiologic significance (32), accepting the existence of the other two sites with the reported characteristics requires prior exclusion of alternative explanations for the observed curvilinear Scatchard plots and more direct experimental evidence. Regardless of the method used for analysis of experimental results, the validity of estimated binding parameters, such as dissociation constant and receptor density, depends on the assumption that ideal thermodynamic conditions, including reversibility of ligand-receptor coupling (37,38), are satisfied in the assay. We show here that, using intact platelets and active ␣-thrombin, this condition is met at 4°C but not at 22-25 or 37°C, owing to partly irreversible ligand binding at 37°C as well as room temperature (usually 22-25°C). Notably, previous studies proposing the concept that GP Ib␣ is the high affinity ␣-thrombin binding site were performed at room temperature (6,10,12) and, thus, may have resulted in incorrect estimates for the parameters of interaction.
The results obtained with platelets at 4°C are still best fitted with a two-site model represented by an upwardly concave Scatchard plot but are compatible with the conclusion that the binding of ␣-thrombin to GP Ib␣ occurs with a k d of between 4 and 9 ϫ 10 Ϫ8 M, as shown by obliteration of this class of sites by the monoclonal antibody LJ-Ib10 without any influence on the putative higher affinity sites. The latter observation is relevant, since previous evaluation of the effects of this antibody at room temperature had shown apparent inhibition of both high and intermediate affinity receptors, as well as appearance of a new class of binding sites, not present on control platelets, with affinity halfway between high and intermediate (10). This finding, later confirmed independently (12), could be taken to reflect the existence of negative cooperativity between two distinct receptors but, in view of the above considerations on equilibrium binding, is more likely to indicate that the parameters estimated at 22-25°C were erroneous. It is also relevant to note that the sum of presumed high and intermediate affinity sites inhibited by LJ-Ib10 at room temperature (10, 32) is of the same order of magnitude as the number of homogeneously intermediate affinity sites inhibited at 4°C, in agreement with the notion that the sites may be the same and the estimated affinities at room temperature may be misleading.
The conclusion that ␣-thrombin interaction with GP Ib␣ on platelets has a k d on the order of 10 Ϫ8 M is substantiated both by results obtained with isolated glycocalicin, as shown here and in agreement with independent data reported elsewhere (24,32), and by previous findings with the recombinant aminoterminal domain of GP Ib␣ (24). In the case of isolated receptor fragments containing the ␣-thrombin binding site, interactions with the ligand are fully reversible at 4°C as well as 37°C and occur with k d between 1 and 5 ϫ 10 Ϫ8 M, reflecting the initial attributes of ␣-thrombin pairing with GP Ib␣ on platelets. These results cannot support the proposed alternative possibility that high and intermediate affinity sites are both expressed on GP Ib␣ (9). They also indicate that other components of the GP Ib-IX-V complex have no direct influence on the function of the ␣-thrombin binding site on GP Ib␣, suggesting that data to the contrary obtained in heterologous expression systems (53) may depend on specific experimental conditions. The nature and physiologic significance of the ␣-thrombin binding sites not inhibited by LJ-Ib10, of higher affinity than GP Ib␣ sites on the basis of the results obtained at 4°C, remain undetermined at present. Candidates for their identification may include PAR1 (13,19), PAR3 (22), and protease nexin 1 (54).
Despite evidence to the contrary from independently performed experiments (10,12), others have reached the conclusion that antibodies against the proposed ␣-thrombin binding site on GP Ib␣, such as LJ-Ib10, have no inhibitory effect on platelet interaction with the agonist nor on activation (13). As in the case of studies aimed at determining binding characteristics, the methodology used may have influenced the conclusions reached. Lack of inhibitory effect by LJ-Ib10 was reported in experiments in which platelets were fixed after incubation with ␣-thrombin, then processed for indirect detection of bound ligand after repeated washing steps. It may be that, after such a procedure, the ␣-thrombin remaining associated with platelets, not defined in terms of quantity or binding characteristics, is interacting with PAR1 rather than GP Ib␣, as suggested (13); such a conclusion is compatible with the two-site model discussed here. On the other hand, the present studies provide direct evidence that the reversible ␣-thrombin binding to platelets at 37°C can be blocked by LJ-Ib10 within the first 60 s of incubation. The same antibody reduces dense granule ATP release, acting with an anti-PAR1 antibody to yield more efficient inhibition. Moreover, it is apparent that the ␣-thrombin associated with platelets through a GP Ib␣-dependent mechanism remains available as an active enzyme. One function of this relatively high capacity site, therefore, may be that of increasing the concentration of ␣-thrombin onto or in proximity of the platelet membrane for subsequent proteolytic cleavage of appropriate substrates. This latter event may take place not while the enzyme is bound to GP Ib␣, assuming that the association prevents catalytic function (14), but after dissociation from the receptor that may occur rapidly during the time frame of interaction relevant for platelet activation and clotting. Therefore, even without considering the possibility of coupling to a distinct signaling pathway that remains to be proven directly (12), our present findings add evidence to the previously proposed concept (10, 12) that ␣-thrombin interaction with GP Ib␣ has a net prothrombotic effect.