A Subpopulation of Platelets Responds to Thrombin- or SFLLRN-stimulation with Binding Sites for Factor IXa*

Strong agonists cause platelets to expose a procoagulant surface supporting the assembly of two important coagulation enzyme complexes. Equilibrium binding has determined the density of high affinity saturable factor IXa binding sites to be 500–600 sites/platelet. We have now used flow cytometry to visualize the binding of factor IX and IXa to thrombin- or SFLLRN-activated platelets. Concentrations of these agonists that are half-maximal or maximal in kinetic studies resulted in only a small subpopulation (4–20%) of platelets binding factor IX or IXa with the density of binding sites for factor IX being about half of that for factor IXa, consistent with previous equilibrium binding studies. A small subpopulation (5 (cid:1) 1.5%) of platelets stimulated with either agonist also exposed annexin V binding sites, and this subpopulation of platelets also bound factor IXa. Annexin V decreased factor IXa binding in the presence or absence of factor VIIIa, and factor IXa could also decrease annexin V binding on some platelets indicating a common binding site in agreement with previous studies. All platelets binding factor IXa were positive for glycoprotein IX, at the same glycoprotein IX surface density as seen in platelets negative for factor IXa binding. These studies refine the results from equilibrium binding studies and suggest that, on average, only a small subpopulation ( (cid:1) 10%) of PAR 1-stimulated platelets expose (cid:1) 6000 factor IXa binding sites/platelet.

Although normally circulating in a quiescent state, when activated with strong agonists, platelets provide an important surface for assembly of two major coagulation enzyme complexes, the intrinsic factor X-activating complex and the prothrombinase complex (1). In vitro kinetic assays have demonstrated that in the presence of activated platelets (or artificial phospholipid vesicles) factor Xa generation is enhanced several million-fold, and thrombin generation is enhanced 300,000-fold over solution phase activity (2)(3)(4)(5).
It is known that not all agonists are equivalent in their ability to stimulate all functional end points of platelet activation. Weak agonists (e.g. ADP and epinephrine) require feedback enhancement from secreted molecules or synthesized agonists such as thromboxane A2 to achieve stimulation of platelet aggregation and are unable to stimulate a procoagulant surface capable of supporting assembly of these enzyme complexes (6,7). Strong agonists (e.g. thrombin and collagen) are capable of generating all platelet functional endpoints (8).
Are all platelets equivalent in their response to these agonists? Visualizations of in vivo clot formation show most platelets rolling, adhering, aggregating, and serving as the foundation for fibrin clot formation (9). However, circulating platelets may constitute a heterogeneous population. Their generation from megakaryocytes packages them in different sizes, with different granule loads, differing numbers of mitochondria, and other vestiges of the parent cells (10,11). The circulation contains platelets of different ages as well as sizes. Flow cytometry analyses detect a wide range of forward and side scatter of blood platelets, necessitating the viewing of platelets on log scales (12). Flow cytometry provides an opportunity to view the homogeneity or heterogeneity of the incidence and density of platelet markers as well as the appearance of activation-dependent markers providing a clinical tool to probe for platelet activation in whole blood samples (13,14).
We have formerly used both equilibrium binding studies and in vitro kinetic studies to investigate the nature of the factor IXa-factor VIIIa-catalyzed factor X-activating complex on thrombin-or SFLLRN-stimulated platelets. We determined that platelets upon activation expose 500 -600 binding sites for the enzyme factor IXa (15), 1000 -1200 binding sites for the cofactor factor VIIIa (6), and 1000 -1200 specific binding sites for the substrate factor X (as well as 30,000 sites shared with prothrombin) (16). The affinity of each protein for its binding site as well as complete occupancy of sites are both optimal when all components of the enzyme complex are present. It has been assumed that the saturable binding sites determined from equilibrium binding studies are generated on all platelets as a result of agonist stimulation. We now present data from flow cytometry studies visualizing the binding of the enzyme and the zymogen that result in the surprising conclusion that although all platelets respond to the PAR 1 agonists with markers of activation, only a small subpopulation of platelets stimulated with thrombin or SFLLRN-amide responds with procoagulant surface changes allowing binding of factor IXa.
Proteins-Factors IX, X, and XIa were obtained from Enzyme Research Laboratories, Inc. (South Bend, IN) or Hematologic Technologies, Inc. (Burlington, VT). Coagulation proteins were obtained in or dissolved in 20 mM Tris-hydroxymethane, 150 mM NaCl and dialyzed against the same buffer to remove inhibitors. Concentrations were determined by bicinchoninic acid assay (Pierce), and purity was assessed by SDS-PAGE visualized with Coomassie Brilliant Blue staining or by Western blot using specific antibodies. Baxter Healthcare Corporation (Duarte, CA) generously supplied recombinant factor VIII. Human ␣-thrombin was purchased from Sigma and was active-site titrated against human antithrombin III, the residual activity was assayed chromogenically with S-2238. The monoclonal antibody C10D raised against the catalytic domain of factor IX (17) was fluoresceinated using manufacturer's instructions (Molecular Probes) or was biotinylated using manufacturer's instructions (Pierce). Protein modifications did not interfere with the ability of the antibody to inhibit the clotting of normal human plasma.
Platelet Preparation-Washed, aspirin-treated, gel-filtered platelets were prepared from human whole blood as described previously (18) with the following modifications. Platelet-rich plasma was incubated with 1 mM acetylsalicylic acid for 30 min to inhibit feedback activation through prostaglandin synthesis, with prostaglandin E1 (2 M) added 5 min before layering over the bovine serum albumin gradient for the wash step.
Factor Xa Generation-Factor Xa was generated as described previously (18) with the following modifications. Washed, aspirin-treated, gel-filtered platelets (10 8 /ml) were added to wells of a microtiter plate. Factor IXa and varying concentrations of agonists (thrombin or SFLLRN-amide) were added, and activation proceeded for 10 min before the addition of the cofactor and substrate. Factor VIII (200 units/ ml) was incubated 1 min at 37°C with thrombin (0.1 units/ml) and an additional minute with hirudin (5 units/ml) immediately before addition to reaction mixtures. Reaction mixtures were stopped after 3 min and were analyzed for factor Xa generated by chromogenic substrate S-2765 cleavage as described (19).
Data Analysis-Results from enzyme assays were converted to factor Xa formed/min by comparison to a standard curve of factor Xa cleavage of S-2765 and plotted using KaleidaGraph software (Synergy, Reading, PA) to derive kinetic parameters. Results from multiple experiments were pooled and analyzed for means and standard errors.
Flow Cytometry-Washed, aspirin-treated, gel-filtered platelets (10 6 ) were incubated with FITC-labeled CD42b against platelet GPIb␣ to define the platelet population. Platelets were incubated with FITC-or PE-labeled annexin V at specified concentrations or with various concentrations of factor IXa or IX, and half of each sample was activated with thrombin (0.2 or 2 units/ml) or with SFLLRN-amide (30 -150 M) for 10 min. Some samples contained thrombin-activated factor VIIIa at 10 units/ml. Samples containing factor IX or IXa were incubated with either FITC-labeled C10D or biotinylated C10D for 15 min, and the latter was detected with streptavidin-PE-cy5. Some samples were duallabeled to simultaneously detect PE-annexin V binding and factor IXa binding (through FITC-C10D). Other samples were dual-labeled to simultaneously detect platelet GPIX and factor IXa binding. All samples were diluted 5-fold and analyzed on a Facscan flow cytometer (Pharmingen) equipped with a 488-nm-emitting laser. Data were collected using either E00 or E01 voltage settings for forward scatter sensitivity, and a threshold of 2 arbitrary fluorescence units was established to eliminate debris. Compensations were set to minimize fluorescence channel 1 (FL-1) spillover into fluorescence channel 2 or 3 (FL-2 and -3) and FL-3 spillover into FL-1 or -2. Using CellQuest Pro software (Pharmingen), platelets were defined by their characteristic forward versus side scatter pattern as well as by positive staining for fluorescein-CD42b (anti-GPIb␣). Negative controls included excess unlabeled CD42a for FITC-CD42a, 10 mM EDTA for FITC-or PE-annexin V, and incubation with FITC-C10D (or the combination of biotinylated C10D and streptavidin-PE-cy5) in the absence of factor IX(a). Gated platelets were analyzed for fluorescein as FL-1 emission, PE as FL-2 emission, or PE-cy5 as FL-3 emission and displayed on log scales as density plots or histograms.
Data Analysis-Quadrants (density plots) or markers (histograms) were established with negative control samples to determine the statistics of positive and negative populations. Positive regions were defined so that density plots displaying total events from defined regions could subsequently be analyzed as platelet and non-platelet events and for forward scatter statistics. Events measured with buffer alone were subtracted from events measured with platelet samples. Data from different experiments collected under the same conditions were averaged and statistically analyzed.

RESULTS
Platelets activated with low levels of thrombin or PAR 1-activating peptide (SFLLRN) support intrinsic factor Xa generation (5) and expose high affinity saturable binding sites for factor X-activating complex components (6,7,15,20). We have now used flow cytometry studies to visualize the effect of PAR 1 stimulation on both factor IXa binding and on exposure of procoagulant lipids and compared this with the ability of PAR 1 agonists to stimulate platelet support for factor Xa generation.
Platelets stimulated with increasing amounts of thrombin or SFLLRN-amide for 10 min at 37°C supported increasing factor Xa generation saturating at 14.4 Ϯ 0.7 nM/min in response to thrombin at 1 units/ml and at 9.7 Ϯ 0.2 nM/min in response to 150 M PAR 1 agonist peptide (Fig. 1). The EC 50 for thrombin was 0.14 Ϯ 0.02 units/ml (1.4 nM) and for SFLLRN-amide was 34 Ϯ 2.5 M. When incubated with an FITC-labeled antibody to P-selectin (CD62P), platelets stimulated with 35 M SFLLRN-FIG. 1. Agonist stimulation of platelet-supported factor IXacatalyzed factor X activation. Platelets prepared as described under ''Experimental Procedures'' were combined with calcium (5 mM) and factor IXa (1 nM) and activated with varying amounts of either thrombin (a) or SFLLRN-amide (b) before the addition of thrombin-activated factor VIII and factor X (250 nM). Reactions were stopped after 3 min at 37°C, and aliquots were assayed for factor Xa with chromogenic substrate S-2765 in which hydrolysis was compared with that of standard factor Xa. Data from nine experiments were analyzed by KaleidaGraph.
amide showed a 5-fold fluorescence shift of the entire population of platelets indicating that all platelets had responded to the agonist peptide with secretion from ␣-granules and incorporation of membrane-bound P-selectin into the plasma membrane. The same effect on P-selectin exposure was found in response to thrombin at 0.2 units/ml, and these platelets aggregated fully without exogenous fibrinogen when stimulated with as little as 10 M SFLLRN or thrombin at 0.2 units/ml (data not shown) indicating that all platelets are capable of aggregation as well as secretion in response to PAR 1 stimulation.
The assembly of the factor X-activating complex on platelets was probed by flow cytometry for factor IXa binding. The monoclonal antibody C10D to the heavy chain of factor IX was either fluoresceinated with FITC for detection in the FL-1 window or was biotinylated and detected with streptavidin conjugated to PE-cy5, a tandem dye absorbing at 488 nm and fluorescing by resonance energy transfer from PE to cy5 in the FL-3 window.
Factor IXa binding results were equivalent using either detection system. Table I shows binding results determined in the presence and absence of the cofactor factor VIIIa (6 units/ml) when platelets were stimulated with half-maximal concentrations of SFLLRN-amide (35 M). Increasing factor IXa resulted in an increasing percentage of platelets binding more factor IXa/platelet in both the presence and absence of the cofactor. The presence of the cofactor did not affect the percentage of platelets responding for a given factor IXa concentration but did increase the density of binding sites at lower than saturating concentrations of ligand. At 10 nM factor IXa, the densities of binding sites in the presence and absence of the cofactor were equivalent. The mean percentage of platelets binding Factor IXa in the presence of 10 nM factor IXa in the absence of the cofactor was 9 Ϯ 2.4 with a mean fluorescein emission of 1000 Ϯ 147 (Table I). Activated platelets incubated with biotinylated C10D and streptavidin-PE-cy5 in the absence of factor IXa showed no positive population. Less than 1% of unstimulated platelets displayed low density binding in the presence of 10 nM factor IXa. Experiments comparing the binding of the enzyme with that of the zymogen found that 9% of the platelets were positive for factor IXa, and 6.5% were positive for factor IX, with the density of binding sites for factor IXa 2-to 3-fold higher than for factor IX (Table I). This is in approximate agreement with the equilibrium binding studies (15) that found half the number of sites for factor IX as were found for factor IXa.
When saturating (10 nM) factor IXa was incubated with platelets in the presence of half-maximal or saturating levels of thrombin (0. When probed for exposure of annexin V binding sites after stimulation with thrombin or SFLLRN at either their EC 50 or at saturation levels, only a small subpopulation of platelets was positive with a high density of annexin V binding sites/platelet (Fig. 2). Annexin V binding occurred in few (1.8%) unstimulated platelets (Fig. 2b) and increased as platelets were activated with half-maximal to maximal levels of both PAR 1 agonists, thrombin (4 -7%) and SFLLRN-amide (3.8 -5.7%). In five experiments with PE-annexin V and seven experiments with FITC-annexin V performed with the EC 50 concentrations of the agonists, the mean (ϮS.E.) percentage of platelets positive with thrombin stimulation of annexin V binding was 5% Ϯ 1.5 and with SFLLRN-amide stimulation was 4.2% Ϯ 0.5, whereas maximal levels of SFLLRN-amide (150 M) resulted in 7% Ϯ 1.8. Therefore, activation with PAR 1 agonists was accompanied by a 3-fold increase in the population positive for annexin V binding with a 2-fold increased density of binding sites.
Dual-labeling studies were performed to investigate the relationship between annexin V binding and factor IXa binding to SFLLRN-activated platelets. Platelets were stimulated with saturating SFLLRN-amide (150 M) and exposed to 10 nM factor IXa (Ϯ thrombin-activated factor VIII, 6 units/ml) in the presence of varying concentrations of annexin V. Increasing annexin V concentrations from 1 to 5.7 nM resulted in increased density of bound annexin V/platelet (data not shown). Fig. 3a shows that the density of bound factor IXa decreased with increasing annexin V concentration in the absence (gray bars) or presence (black bars) of factor VIIIa. The simultaneous presence of both ligands had no effect on the percentage of platelets positive for either annexin V (6 -7%) or for factor IXa (6 -8%). It was noted, however, that of eight experiments all showing a decrease in factor IXa binding in the presence of annexin V, only five showed no change in annexin V binding in the presence of 10 nM factor IXa. Fig. 3b presents the data for samples incubated with 3.8 nM annexin V in the absence (open bars) or presence (black bars) of factors IXa/VIIIa. The bars on the left represent all data analyzed together for annexin V binding. Five experiments (Fig. 3b, middle) showed no change in annexin V bound in the presence of factors IXa/VIIIa. Three experiments showed that annexin V binding could be substantially decreased by the presence of factors IXa/VIIIa (Fig. 3b,  right). These results were found using PE-annexin V with FITC detection of factor IXa and using FITC-annexin V with PE-cy5 detection of factor IXa. Factor IXa in the absence of factor VIIIa showed the same effect. The ability of factor IXa to reduce binding sites for annexin V was more noted with annexin V concentrations at or below the previously determined kinetic IC 50 (19).
In support of this evidence for competition, it was clear that the platelet populations responding to PAR 1 stimulation with binding of either annexin V or factor IXa were almost totally overlapping. Fig. 4 shows that almost all positive events in the dual-labeled sample (Fig. 4d) appeared in the upper right quadrant that was set to depict dual-labeled platelets. This overlap in binding determinants was consistent with previously reported kinetic studies (19) of annexin V inhibition of factor IXa-catalyzed factor X activation, which predicted an overlap in platelet binding sites for both ligands. Thus, platelets re-TABLE I Detection of factor IXa binding to activated platelets Platelets prepared as described under "Experimental Procedures" were incubated with factor IXa or factor IX at the indicated concentrations and activated by SFLLRN-amide (35 M) for 10 min at 37°C before the addition, where indicated, of thrombin-activated factor VIII (6 units/ml). After incubation with a detection system, data were collected and analyzed as described under "Experimental Procedures." Results are presented as the mean of two determinations or as the mean (ϮS.E.) of the number (n) of observations designated in the

Flow Cytometry Detection of Factor IXa Binding to Platelets
sponding to thrombin with the appearance of aminophospholipids that bind annexin V are the same platelets capable of binding the enzyme factor IXa. Activation of platelets can lead to microvesiculation (21), resulting in procoagulant vesicles displaying exposed aminophospholipids that can bind coagulation proteins. To determine whether activation of platelets with the EC 50 concentration of agonists leads to extra platelet procoagulant surfaces, data from single and dual-label platelet samples from factor IXa/annexin V experiments were analyzed for positive events inside and outside of the defined platelet population. It was found that 95% of all events positive for factor IXa binding, 94% of all events positive for annexin V binding, and 97% of all events positive for both labels were within the platelet population. Previously published data (21) reporting barely detectable microvesicle formation in unstirred platelets stimulated with low thrombin agonist concentration support our observation that factor IXa binding stimulated by PAR 1 activation occurs on the activated platelet surface.
To confirm that all events positive for factor IXa binding were platelets, dual staining was performed with PE-cy5-labeled anti-factor IXa and FITC-labeled anti-GPIX. A represent-ative experiment is shown in Fig. 5. The quadrants were drawn to define the negative population of platelets incubated with both FITC-anti-GPIX in the presence of excess unlabeled anti-GPIX and with biotinylated C10D and streptavidin-PE-cy5 in the absence of factor IXa (Fig. 5a). 6% of activated platelets were positive for factor IXa (Fig. 5b), whereas all platelets were positive for GPIX (Fig. 5c). Fig. 5d shows that all platelets positive for factor IXa contained the same density of GPIX (upper right quadrant) as the activated platelet population that remained negative for factor IXa binding (upper left quadrant).
In summary, these experiments produced evidence that platelet activation with doses of PAR 1 agonists mid to maximally effective at eliciting platelet procoagulant activity resulted in only a subpopulation of platelets exposing binding sites for factor IXa and for annexin V. This population displayed the same surface density of GPIb-IX-V complex as seen in the population of activated platelets negative for factor IXa binding. DISCUSSION Biochemical techniques for determining specific ligand binding rely on statistical averaging over the entire population of cells studied. Using these techniques, it was determined that factor IXa binds to thrombin-or SFLLRN-amide-activated platelets with 500 -600 sites/platelet and that the inactivated zymogen binds to one-half that number of sites (15). Flow cytometry provides the opportunity to analyze the total cell population as individual events and thus can help determine whether all cells provide binding sites or whether activation events are specific to portions of the cell population. It can also determine how uniformly distributed are the binding events. Although it has been demonstrated that thrombin, a strong agonist for many platelet activation events, is a weak agonist for activation-dependent exposure of procoagulant phospholipids (22), we have previously used thrombin or SFLLRN-amide to stimulate robust platelet support of factor Xa generation at one-fifth to one-tenth the concentrations that had shown little ability to stimulate exposure of procoagulant phospholipids (23). Here we correlate dose-response curves for the stimulation of factor Xa generation with the development of factor IXa binding sites and exposure of procoagulant phospholipids detected by flow cytometry.
We have found that although thrombin or SFLLRN stimulates all platelets to aggregate and to expose P-selectin through externalization of ␣-granules, only 4 -20% of the platelets respond to thrombin or SFLLRN with exposure of factor IXa binding sites. These platelets responding with factor IXa binding sites overlap almost entirely with the entire population exposing aminophospholipids detected with annexin V binding. The subpopulation varies widely in single-donor platelets, the reasons for which are unknown. Because Feng and Tracy (24) found that factor Xa binding was confined to a subpopulation of stimulated platelets, it would be interesting to determine whether there is a correlation between a tendency toward thrombotic events and the percentage of platelets responding to thrombin stimulation with exposure of factor IXa or Xa binding sites. Previous flow cytometric detection of factor IXa binding to thrombin-stimulated platelets provided dose-response data supporting the existence of two sites, a high affinity site comparable with what was found here and a low affinity site found with higher factor IXa concentrations (25).
Factor IXa binding results were similar for two different probes. Both fluorescein modification of a monoclonal antibody to the factor IX catalytic domain and biotinylation of that antibody detected with streptavidin-PE-cy5 allowed robust detection of 0.5 nM factor IXa in the presence of the cofactor and a weak detection of that concentration in the absence of the cofactor. Predictable from equilibrium binding results, 10 nM factor IXa represented a saturating level where binding-site density was not changed by the presence of the cofactor. The K d was not measured in these flow cytometry studies, but it was noted that although the presence of both cofactor and substrate in equilibrium binding studies lowered the K d from 2.5-3 nM to 0.5 nM (20), an inclusion of the cofactor in the absence of substrate here did not result in half-maximal binding density at 0.5 nM factor IXa. This may reflect either a lack of sensitivity at low binding-site density or the requirement for the substrate to stabilize the complex. However, the presence of the cofactor increased the density of binding sites/platelet most at a ligand concentration around the equilibrium K d . As predicted from the titrations of agonists in kinetic assays, maximal concentrations of both SFLLRN-amide and thrombin elicited twice the factor IXa binding as seen with half-maximal concentrations. The additional binding resulted from recruitment of twice the number of platelets to bind factor IXa rather than from increases in the number of factor IXa binding sites/platelet. In these studies, factor IX bound to almost the same percentage of platelets with one-third to one-half of the density of binding sites/platelet. This is in agreement with equilibrium binding studies. Also in agreement with previous studies (19) is the overlap in binding determinants for annexin V and factor IXa. In kinetic studies, annexin V was found to inhibit factor Xa generation with a K i of 3.5 nM. In these binding studies, although the density of factor IXa was reduced by the presence of annexin V, especially with annexin V present at concentrations above its K i for kinetic studies, annexin V density could also be reduced by the presence of factor IXa. These results suggest that there is true competition for overlapping binding determinants, with factor IXa as a less effective competitor for annexin V binding sites than annexin V is for factor IXa binding sites.
Binding sites for factor IXa and for annexin V were located primarily within the defined platelet population, which is in agreement with the previous observation that Ͻ5% of total factor IXa binding was attributed to platelet microparticles (25). This was predictable from the weak ability of thrombin to generate microparticles from activated platelets (21). The duallabeling studies detecting both GPIb/IX and factor IXa binding revealed that all platelets binding factor IXa contain an equivalent surface density of GPIb/IX/V complexes as the bulk of the platelet population that remains negative for factor IXa binding.
These flow cytometry studies visualizing factor IXa binding to thrombin-or SFLLRN-stimulated platelets suggest a modification of the previous conclusions about intrinsic factor X-activating complex binding sites. It appears from the present results that instead of all platelets presenting ϳ600 factor IXa binding sites/platelet, an average of 10% of activated platelets presents an average ϳ6000 factor IXa binding sites/platelet. Flow cytometry detection of platelet binding of other factor Xa complex components should reveal interesting data to augment and refine information obtained from equilibrium binding studies.