Structural and functional characterization of platelet receptor-mediated factor VIII binding.

Optimal rates of factor X (FX) activation require occupancy of receptors for factor IXa (FIXa), factor VIII (FVIII), and FX on the activated platelet surface. The presence of FVIII and FX increases 5-fold the affinity of FIXa for the surface of activated platelets, and the presence of FVIII or FVIIIa generates a high affinity, low capacity specific FX-binding site on activated platelets. We have now examined the effects of FX and active site-inhibited FIXa (EGR-FIXa) on the binding of both FVIII and FVIIIa to activated platelets and show the following: (a) von Willebrand factor inhibits FVIII binding (K(i) = 0.54 nM) but not FVIIIa binding; (b) thrombin and the thrombin receptor activation peptide (SFLLRN amide) are the most potent agonists required for FVIII-binding site expression, whereas ADP is inert; (c) FVa does not compete with FVIIIa or FVIII for functional platelet-binding sites; and (d) Annexin V is a potent inhibitor of FVIIIa binding (IC(50) = 10 nM) to activated platelets. The A2 domain of FVIII significantly increases the affinity and stoichiometry of FVIIIa binding to platelets and contributes to the stability of the FX-activating complex. Both FVIII and FVIIIa binding were specific, saturable, and reversible. FVIII binds to specific, high affinity receptors on activated platelets (n = 484 +/- 59; K(d) = 3.7 +/- 0.31 nM) and FVIIIa interacts with an additional 300-500 sites per platelet with enhanced affinity (K(d) = 1.5 +/- 0.11 nM). FVIIIa binding to activated platelets in the presence of FIXa and FX is closely coupled with rates of F-X activation. The presence of EGR-FIXa and FX increases both the number and the affinity of binding sites on activated platelets for both FVIII and FVIIIa, emphasizing the validity of a three-receptor model in the assembly of the F-X-activating complex on the platelet surface.

The importance of factor VIII (FVIII) 1 and FIX in hemostatic reactions is evident by the fact that hemophilia A (FVIII deficiency) and hemophilia B (FIX deficiency) are the two most serious congenital coagulation defects, both producing severe, life-threatening and life-long hemorrhagic disease. FVIII is synthesized as a single polypeptide chain containing 2,351 amino acids (1) and shows a discrete domain structure, A1-a1-A2-a2-B-a3-A3-C1-C2; the domains are separated by short spacers (a1, a2, and a3) composed of acidic regions that contain clusters of Asp and Glu residues (2). The A domains display ϳ30% homology to each other, and they also share sequence homology with ceruloplasmin and FV. The B domain is unique to FVIII. The C domains share homology with FV and discoidin 1, a lectin (1). During proteolytic activation of pro-cofactor (FVIII), the B domain that does not contribute to cofactor function is excised. The heavy chain (HC) and light chain (LC) remain noncovalently associated through the A1 and A3 domains in a metal ion-dependent manner and participate as an active cofactor in tenase complex formation (1). The following terminology, suggested by Lenting et al. (2), for FVIII subunits was used: A1 ϭ HC-derived fragment containing residues 1-372; A2 ϭ HC-derived fragment containing residues 373-740; A1-A2 ϭ HC subunit of FVIII containing residues 1-740 (without the B domain); B ϭ HC subunit residues 741-1648; LC ϭ LC of FVIII; A3-C1-C2 ϭ LC-derived fragment arising from the cleavage at position 1689. Thrombin cleaves FVIII at one specific site within the LC, Arg 1689 , and at two sites in the HC, Arg 372 and Arg 750 , resulting in a heterodimer of 50 kDa (A1 domain), 43 kDa (A2 domain), and 73 kDa (A3-C1-C2 dimer); all of these subunits are required for the procoagulant activity (3). More specifically, the cleavages at Arg 372 and Arg 1689 are required to exert full cofactor activity (4,5), whereas cleavage of the LC at Arg 1689 is also responsible for dissociation of FVIII from von Willebrand factor (vWF) (6).
FVIII in the circulation forms a tight, noncovalent complex with vWF, and both the amino-terminal and the carboxylterminal regions of FVIII are involved in this interaction (6,7). The formation of the FVIII-vWF complex is essential for the survival of FVIII in the circulation. Another functional consequence of FVIII-vWF complex formation is to prevent binding of FVIII to components of the FX-activating complex, e.g. the binding of FVIII LC to FIXa has been shown to be inhibited by vWF (8). Furthermore, FVIII-vWF complex is less susceptible to proteolytic degradation by proteases like activated protein C (9) and FXa (10). vWF also prevents FVIII binding to phospholipids (11) and platelets (12). In contrast, vWF does not protect FVIII from thrombin cleavage (13,14). Recently, it has been shown (15) that FIXa displays similar affinity for pro-cofactor (FVIII) and active cofactor (FVIIIa), but as mentioned above, the FIXa-binding site is not accessible when FVIII is in com-plex with vWF (8). This clearly indicates that there is a delicate balance between FVIII binding in complex with vWF and FVIII binding to membrane surfaces. However, proteolytic activation of FVIII to FVIIIa causes a dramatic decrease (1,600-fold) in affinity for vWF (6), thus shifting the balance toward FVIIIa binding to the membrane surface. This phenomenon has recently been studied by using electrophoretic quasi-elastic light scattering, which demonstrated the physical exchange of FVIII between vWF and activated platelets (16).
Our laboratory provided the first evidence that platelets have an essential role in the activation of FX by the intrinsic pathway (17,18). Subsequent studies (19) indicated that the contribution of platelets to F-X activation by FIXa in the presence of activated FVIII requires stimulation of platelets by specific agonists such as collagen, thrombin, the calcium ionophore (A23187), or a combination of thrombin and collagen. Differences in the ability of specific platelet agonists to promote binding of the components of FX-activating complex (FIXa, FX, and FVIII) are not fully understood or worked out; although it is clear that only activated platelet membranes provide a procoagulant surface for the assembly and expression of a variety of coagulation protease complexes (20). Recent studies from our laboratory and others are focused on the molecular mechanisms involved in the assembly of the intrinsic F-X-activating complex on platelet membranes, showing that optimal rates of F-X activation require occupancy of receptors for FIXa (21), FVIII (12,22), and FX (23). The presence of cofactor (FVIIIa) and substrate (FX) increases 5-fold the affinity of FIXa (active site-inhibited Glu-Gly-Arg-FIXa, i.e. EGR-FIXa) for the surface of activated platelets (21). Recently, we have reported that the presence of FVIIIa generates a high affinity, low capacity specific FX-binding site on activated platelets (24).
In this study we have demonstrated that both pro-cofactor (FVIII) and active cofactor (FVIIIa) bind to platelets with enhanced affinity in the presence of the enzyme (EGR-FIXa) and the substrate (FX), thus emphasizing the validity of a threereceptor complex on the platelet surface (20). In order to be more specific about the location and stability of FVIII-interactive site on platelet membranes, we have studied the effect of isolated subunits of FVIII on the binding of FVIIIa to activated platelets. Recent studies indicate that it is the spontaneous dissociation of A2 domain from the metal-linked A1/A3-C1-C2 dimer that causes FVIIIa activity to be unstable (for recent reviews, see Refs. 2 and 25). Our results clearly show that the presence of the A2 domain increases the affinity and stoichiometry of FVIIIa binding. These studies support a model in which the A2 subunit of FVIIIa contributes to the catalysis and stability of the F-X-activating complex both in vitro (26) and in vivo (27). Thus, our studies further emphasize that F-X activation is a platelet receptor-mediated process tightly coupled to receptor occupancy by FIXa, FVIIIa, and FX.

EXPERIMENTAL PROCEDURES
Reagents-Phenalanyl-prolyl-arginine chloromethyl ketone (PPACK) and calcium ionophore A23187 were from Calbiochem. Prostaglandin E1 and apyrase were from Sigma. All other reagents were of analytical quality.
Proteins-Human FIX, FX, vWF, and prothrombin were purchased from both Enzyme Research Laboratories, Inc. (South Bend, IN) and Hematologic Technologies, Inc. Other highly purified human coagulation factors (vWF, FIXa␤, RVV-X, and antithrombin III) were from Enzyme Research Laboratories, Inc. Human ␣-thrombin (4,165 units/ mg) was purchased from Sigma. Highly purified recombinant FVIII (rFVIII, Ͼ4,000 units/mg) was the generous gift of Genetics Institute (Cambridge, MA) and Baxter Health Care Corp. (Duarte, CA). Human prothrombin fragment 1 was liberally supplied by Dr. S. Krishnaswamy (University of Pennsylvania, Philadelphia) and human placenta annexin V (28) was given to us by Dr. Kazuo Fujikawa (University of Washington, Seattle). The murine monoclonal antibody, R8B12, which reacts with the carboxyl-terminal region of FVIII A2 domain (3), was obtained from Dr. Philip J. Fay of the University of Rochester School of Medicine, Rochester, NY. All proteins were Ͼ95% pure as judged by SDS-PAGE (29) and protein staining by Coomassie Brilliant Blue.
Radiolabeling of Proteins-For binding assays, 125 I-labeled FIX was prepared by the IODO-GEN method (30) and had a specific radioactivity of ϳ2.5 ϫ 10 6 cpm/g. Human FX was radioiodinated via lactoperoxidase (23). Attempts were made to label FVIII by various procedures: IODO-GEN (30), IODO-BEADS (31), transfer technique (32), lactoperoxidase (33), Enzymobeads (34), and NHS-[ 3 H]propionate (35). Finally, we had success in developing a highly reproducible method utilizing the Bolton-Hunter reagent (36) to radiolabel FVIII to high specific radioactivity while retaining virtually 100% of its specific biological activity. In this procedure, 20 -50 g of rFVIII was precipitated with 25% PEG 8000, washed, and then resuspended in 50 l of 0.5 M NaCl, 20 mM HEPPS, pH 8.0, 5 mM CaCl 2 , 0.1% Tween 80 just before use. The protein was then added to a siliconized glass vial containing 0.25-0.5 mCi of iodinated Bolton-Hunter reagent that had been previously dried onto the surface. The mixture was incubated for 45 min at 0°C and then quenched by the addition of 5 l of 0.2 M glycine, 0.1 M borate, pH 8.5, followed by further incubation for 45 min at 0°C. The quenched reaction mixture was then precipitated with 25% PEG 8000, and the supernatant containing most of the unbound reagent was discarded. An aliquot was removed just before precipitation to determine percent incorporation (20 -30%) by trichloroacetic acid precipitation. It was often necessary to precipitate the labeled protein more than once to remove the noncovalently associated reagent. Percent incorporation and percent bound radioactivity were determined by trichloroacetic acid precipitation. The concentration of rFVIII protein was either determined by enzyme-linked immunosorbent assay (if bovine serum albumin was added as a carrier) or a BCA protein assay if there were no other proteins present. Usually 85-95% of the original protein was recovered after labeling. Finally, the functional activity of the labeled rFVIII was compared with unbound rFVIII in order to determine whether the protein was damaged during the radiolabeling procedure.
Determination of Functional Activity of 125 I-Factor VIII-We developed a chromogenic assay based on the modification of an assay developed by Wagenvoord et al. (37) that depends on the ability of FVIII to enhance the catalytic efficiency of surface-bound FIXa in F-X activation. In this assay the concentration of FVIII is the limiting factor in activation of FX by FIXa in the presence of phospholipid vesicles and calcium ions. The velocity of FXa generation is determined from the rate of the cleavage of FXa-sensitive chromogenic substrate. Briefly, three reagents were prepared in assay buffer (0.5 mg/ml bovine serum albumin, 175 mM NaCl, 20 mM HEPPS, pH 8.0, at 37°C) and lyophilized for storage. Reagent 1 contains 300 nM thrombin, 300 nM FIXa, 60 M phospholipid vesicles (25% phosphatidylserine; 75% phosphatidylcholine, mol/mol), and 15 mM CaCl 2 . Reagent 2 contains 1 M FX, and reagent 3 was a substrate solution containing 20 mM EDTA, 700 M Kabi S-2765 (FXa-sensitive substrate), and 2 M Kabi I-2581 (thrombin inhibitor). In the assay, appropriate samples of ϳ20 l were added to 20 l of reagent 2 in 1.5-ml Eppendorf tube, then 20 l of reagent 1 was added quickly, vortexed, and placed at 37°C water bath. After 100 s, 50 l of the reaction mixture was removed and added to a second Eppendorf tube containing 450 l of 20 mM EDTA at 4°C to stop further activation of FX. Seventy five l of this solution was then placed in a microtiter plate, warmed at 37°C, and mixed with 75 l of substrate solution (reagent 3). The plates were monitored kinetically at 405 nM for 10 min (Molecular Devices, ThermoMax Reader). The averaged, background subtracted, velocity measurements were plotted as the velocity of substrate hydrolysis versus 1/titer. Slopes were fit to each series of sample dilutions using a Levenberg-Marquardt algorithm (38). The activation of FX by FIXa was also determined at 37°C in the presence of thrombin-stimulated platelets, FVIIIa and CaCl 2 , as described previously (21). Simultaneously FVIII-specific activities (4,000 -5,000 units/mg) were also measured by the ability to clot hemophilic plasma in a clotting assay (39).
Platelet Isolation and Binding Studies-Albumin density gradient washed gel-filtered platelets (wGFP) were isolated from human venous blood by the modification of the method of Walsh et al. (40), as recently reported by us (23). Platelets obtained in this manner contained no detectable levels of FII, FX, or vWF. Therefore, in all our experiments we used activated platelets prepared by incubation of platelets at 37°C with 1.0 nM human ␣-thrombin for 5 min, followed by the addition of 2.5 M PPACK. Similarly, binding studies of activated platelets with active cofactor (FVIIIa) were performed by activating both FVIII and platelets by the addition of 0.1 unit/ml human ␣-thrombin for 5 min followed by the addition of 2.5 M PPACK just prior to the binding assay. In a typical binding experiment, wGFP platelets (3.5 ϫ 10 8 /ml) in Ca 2ϩ -free HEPES/Tyrode's buffer, pH 7.4, were incubated at 37°C in a 1.5-ml Eppendorf plastic centrifuge tube with mixtures of unlabeled and radiolabeled rFVIII or FVIIIa, CaCl 2 , platelet agonists, and other proteins. After incubation, aliquots (100 l) were removed and centrifuged through a mixture of silicon oils as described previously (41). The data were analyzed, and the number of binding sites and K d values were calculated from the means of six independent determinations each done in duplicate using a Macintosh Quadra 900 computer (Apple Computer, Cupertino, CA) and the Ligand Program as modified by G. A. McPherson (Elsevier Science Publishers BV, The Netherlands).
Measurements of Rate of Factor Xa Formation-The activation of FX by platelet-bound FIXa in the presence of bound FVIIIa was determined in the presence of CaCl 2 and FIXa at concentrations of the reactants indicated in the figure legends and as described previously by us (23,41).
Purification of Factor VIII Subunits-Subunits of FVIII were prepared by the method described by Fay et al. (3). Briefly, human recombinant FVIII was first dialyzed into 0.1 M NaCl, 2.5 mM CaCl 2 , 10 mM HEPES, 0.01% Tween 20, pH 7.2, then diluted to 0.2 mg/ml, and activated with 20 nM human ␣-thrombin at room temperature. After 3 min 50 nM FPR-CK was added to inhibit thrombin activity, and aliquots were removed and assayed for FVIII activity using a one-stage clotting assay. Activated FVIII was applied to a Mono-S column (equilibrated with 0.1 M NaCl, 2.5 mM CaCl 2 , 10 mM HEPES, 0.01% Tween 20, pH 6.0). Unbound material contained B domain, whereas the bound material was eluted by 30-ml linear gradient, i.e. A2 domain was eluted at ϳ0.33 M NaCl and A1/A3C1C2 dimer was eluted at ϳ0.64 M NaCl with some residual contamination of A2 subunit. Both A2 subunit fraction and A1/A3C1C2 were subsequently purified to homogeneous proteins by chromatography over R8B12 immunoabsorbent. Dimer eluted in the unbound fraction, whereas A2 was eluted from the R8B12 antibody column (equilibrated with 0.7 M NaCl, 5 mM CaCl, 10 mM HEPES, 0.01% Tween 20, pH 7.2) with buffer containing 50% ethylene glycol. Purification of A1 subunit from A3C1C2 subunit was carried out by chromatography on Mono-Q column equilibrated and eluted in 0.15 M NaCl, 0.02 M HEPES, 0.01% Tween 20 plus 0.04 M EDTA (42). A3C1C2 subunit was eluted in the unbound fraction whereas A1 subunit was absorbed and subsequently batch-eluted with buffer containing ϳ0.8 M NaCl. Protein was determined by the Coomassie dye binding method of Bradford (43) and SDS-PAGE analysis of Mono-S fast protein liquid chromatographic column; R8B12 immunoabsorbent column and Mono-Q columns were performed using the buffer system of Laemmli (29) as shown in Fig. 1.  (35) worked very well except the low energy emission from tritium was not sensitive enough for use. In comparison to other procedures, the Bolton-Hunter reagent (36) for labeling FVIII was relatively mild and resulted in retention of Ͼ95% of specific functional activity. Therefore, in all our experiments we used the Bolton-Hunter reagent to radiolabel FVIII. To characterize the bound FVIII and FVIIIa structurally, platelets were incubated with thrombin (0.1 u/ml), CaCl 2 (5 mM), EGR-FIXa (45 nM), and FX (1.5 M) for 20 min at 37°C in the presence of either 125 I-FVIII or 125 I-FVIIIa and centrifuged through 20% sucrose to separate the bound from free ligand as described under "Experimental Procedures." Platelet pellets were solubilized in SDS and were analyzed by polyacrylamide gel electrophoresis and autoradiography as described previously (21,23). The bound radioligand migrated under reduced conditions as two major bands, i.e. HC, molecular mass ϭ 90 -200 kDa, and LC, molecular mass ϭ 80 kDa, respectively (Fig. 2, lane 1). Thrombin activated FVIII, i.e. FVIIIa (Fig. 2, lane 2) migrated under reduced conditions as several bands of molecular mass ϭ ϳ90, 72, 50, and 43 kDa representing A1-A2 domain; A3 plus C1, C2; A1 and A2 respectively. Fig. 2 (lane 2) clearly indicates the apparent retention of the A2 domain in platelet-bound FVIIIa and the complete absence of B domain. Both bound FVIII and FVIIIa were indistinguishable from free FVIII and FVIIIa (data not shown), and the data provided in Fig. 2 provide no evidence for the formation of high molecular weight covalent complexes or for proteolytic degradation of FVIII or FVIIIa by platelets thus confirming that the bound radioactivity consists entirely of FVIII or FVIIIa and not a radiolabeled contaminant. The fact that the bound cofactor (FVIII or FVIIIa) is structurally intact and indistinguishable from free cofactor (FVIII or FVIIIa) provides strong evidence that the functional F-X-activating complex consists of bound cofactor (FVIII or FVIIIa) together with bound enzyme (FIXa) (21) and bound substrate (FX) (23) on the surface of platelets as will be further discussed later.

Radiolabeling and Structural Characterization of Platelet
Binding of Pro-cofactor (Factor VIII) and Active Cofactor (Factor VIIIa) to Activated Platelets-The specific binding of 125 I-FVIII and 125 I-FVIIIa to wGFP activated with 0.1 units/ml thrombin or with 25 M SFLLRN-amide, in the presence of 5 mM CaCl 2 is shown in Fig. 3, A and B (open circles), respectively. There was a progressive increase in the ligand binding (FVIII and FVIIIa) to activated platelets reaching the maximum in 15 min (data not shown). The maximum amounts of specific binding observed for FVIII and FVIIIa (after subtracting the nonspecific binding, which was measured in the presence of 100-fold molar excess of unlabeled ligands) were 3.8 and 6.0 pmol/10 10 platelets, respectively. In reversibility experiments the dissociation of 125 I-labeled FVIII and FVIIIa was determined when high concentrations (40 nM or ϳ100-fold) of either unlabeled FVIII or FVIIIa were added to the reaction mixture. When the additions were made at different time intervals (1, 5, 7.5, and 10 min), the dissociation of bound radioligand was observed resulting in the removal of 20 -30% of the total bound radioligand during the 1st min followed by slower further dissociation. We observed Ͼ75% of the bound FVIII or FVIIIa removed in 2 min representing a decrease in specific binding equal to nonspecific binding (data not shown). Our results indicate that freely dissociable equilibrium binding of both FVIII and FVIIIa to platelets occurs under our experimental conditions and that this binding does not become irreversible within the time frame of our experiments. Since binding of both FVIII and FVIIIa was at equilibrium and saturable (Fig.  3, A and B), these data were subjected to Scatchard analysis. Straight lines were obtained (data not shown) indicating the presence of a single class of binding sites for both FVIII and FVIIIa. The affinity and stoichiometry of binding for both ligands under these experimental conditions was determined in four separate experiments, the mean (Ϯ S.E.) of which are given in Table I. It is apparent that the number of binding sites and the affinity of binding for active cofactor FVIIIa (n ϭ 750 Ϯ 146 per platelet; K d ϭ 1.5 Ϯ 0.15 nM) were both significantly higher than that for pro-cofactor FVIII (n ϭ 484 Ϯ 59 per platelet; K d ϭ 3.7 Ϯ 0.31 nM) when determined in the absence  (21). Conversely, one might expect the binding of FVIII or FVIIIa would demonstrate a similar increase in affinity when EGR-FIXa and FX are present. This expectation was borne out by increased affinity with which we found 125 I-FVIII and 125 I-FVIIIa binds to activated platelets (Fig. 3, A and B, closed circles) in the presence of both enzyme (FIXa) and substrate (FX). Our data suggest that the enzyme (FIXa) and the cofactor (FVIII/FVIIIa) both contribute to the formation and stability of the F-X-activating complex on the platelet surface ( Table I). As shown in Fig. 3, A and B (closed circles), and in Table I Table I). These results demonstrate that the active cofactor (FVIIIa) interacts with an additional 300 -500 sites per platelet with enhanced affinity, and the presence of both the enzyme (EGR-FIXa) and substrate (FX) increases both the number (by ϳ200 -450 sites per platelet) and affinity of binding sites on activated platelets for both FVIII and FVIIIa (Table I).
To explore further the relationship between pro-cofactor (FVIII) and active cofactor (FVIIIa)-binding sites on thrombinactivated platelets and the physiological significance of the cofactor binding to platelets, we also carried out competition studies with unlabeled FVIII and FVIIIa. Thrombin-stimulated platelets were incubated in the presence of CaCl 2 for 20 min with either 125 I-FVIII or 125 I-FVIIIa and various concentrations of unlabeled FVIIIa and FVIII, respectively, in the absence and presence of EGR-FIXa or FX. When the residual binding was determined (data not shown), it was apparent that excess FVIII and FVIIIa prevented approximately 95% of 125 I-FVIII and 125 I-FVIIIa respectively, whereas unlabeled procofactor (FVIII) had no effect on binding of 125 I-FVIIIa, and unlabeled active cofactor (FVIIIa) had no effect on the binding of 125 I-FVIII. These studies clearly demonstrate that pro-cofactor (FVIII) and active cofactor (FVIIIa) bind to separate, discrete sites on the activated platelets. Furthermore, the presence or absence of EGR-FIXa and FX had no significant effect on these distinct and separate cofactor-binding sites.
Functional Characterization of Bound Factor VIIIa-To characterize bound FVIIIa functionally, platelets (5 ϫ 10 7 /ml) were incubated with FVIIIa (0.1-10 nM), thrombin (0.1 units/ ml), and CaCl 2 (5 mM) and subsequent neutralization of thrombin with PPACK (50 nM), and examined for their capacity to support F-X activation in the presence of FIXa (0.45 nM) and FX (1.5 M) as described under "Experimental Procedures." The purpose of the functional characterization was to determine whether the apparent K d value of FVIIIa binding from our equilibrium binding experiments corresponds with the concentration of FVIIIa required for half-maximal rates of F-X activation in the presence of thrombin-activated platelets. Our results showing the rate of FXa formed versus the concentration of FVIIIa added indicated that the concentration of FVIIIa required for half-maximal rates of FXa formation is 0.65 nM (Fig. 4, closed circles). This value is close to the K d value reported from equilibrium binding of FVIIIa, i.e. 0.80 nM. To study further the relationship between FVIIIa binding and F-X activation, we carried out experiments in which both FVIII binding and F-X activation were examined in the same reaction mixture, consisting of thrombin-activated platelets, 125 I-labeled FVIIIa, FIXa, FX, and CaCl 2 . Fig. 4 demonstrates a close correspondence between the amount of FVIIIa bound and the rate of F-Xa formation both showing a hyperbolic relationship to the concentration of FVIIIa added. The close correlation between FVIIIa binding and F-X activation reported herein suggests that the specific FVIIIa-binding site induced on the surface of thrombin-activated platelets in the presence of enzyme (FIXa) and substrate (FX) is functionally active in F-X activation.
Effect of von Willebrand Factor on Platelet-Cofactor Interaction-FVIII circulates in plasma as a noncovalently bound complex (K d ϭ 0.2 nM) with vWF, preventing FVIII binding to phospholipids or to a platelet surface (12). It also prevents activation of FVIII by FXa (3) or its inactivation by activated protein C (44). Therefore, it is important to address the effect of vWF on both the pro-cofactor (FVIII) and the active cofactor (FVIIIa) binding to activated platelets, both in the absence and presence of EGR-FIXa and FX. For this purpose we carried out competition studies with vWF by including thrombin-stimulated platelets in the presence of CaCl 2 for 20 min at 22°C with 125 I-labeled FVIII or 125 I-FVIIIa and various concentrations of vWF in the presence or absence of EGR-FIXa (45 nM) and FX (1.5 M). When the residual binding of 125 I-labeled FVIII was determined, it was apparent that excess vWF prevented Ͼ85% of 125 I-FVIII binding. From the results presented in Fig. 5, it is estimated that the concentration of vWF required for halfmaximal inhibition of FVIII in the presence or absence of EGR-FIXa and FX was ϳ0.5 nM. This IC 50 value is consistent with the K d value reported previously for the pro-cofactor (FVIII) inhibition by vWF (7,45), i.e. K d ϭ 0.46 nM with a stoichiometry of one FVIII molecule/vWF monomer. When we calculated the inhibition constant (K i ) for vWF both in the presence and absence of EGR-FIXa and FX using a computer fit displacement curve as described by Rodbard (46) and modified to calculate the K i value using the formula of Cheng and Prasoff (47), we obtained a K i value of 0.54 nM for vWF. As expected and in contrast to pro-cofactor (FVIII), active vWF was unable to compete with FVIIIa for binding to activated platelets, either in the absence or presence of EGR-FIXa and FX (Fig. 5), that is no active cofactor (FVIIIa) was displaced from the surface of activated platelets by vWF in the range studied (0.1-30 nM).
Effect of Agonist-induced Platelet Activation on the Exposure of the Cofactor-binding Site-Our initial experiments indicated that resting platelets (without added agonist) do not bind either pro-cofactor (FVIII) or active cofactor (FVIIIa) ( Table II). Thrombin (1 nM) and thrombin receptor peptide (SFLLRNamide, 25 M) were equally effective in their ability to induce cofactor-binding sites on the platelet membrane (Table II). In addition, platelets were exposed to ADP (10 M), collagen type I (10 g/ml), epinephrine (10 M) or ionophore A23187 (1 M). Platelets exposed to ADP showed no expression of FVIII-binding sites. In contrast, platelets exposed to 10 M epinephrine expressed 300 sites per platelet (K d ϳ10 nM) for pro-cofactor (FVIII) and 500 sites per platelet (K d ϳ10 nM) for active cofactor (FVIIIa), whereas platelets exposed to 10 g/ml collagen (type I) exhibited very low levels of binding for active cofactor (FVIIIa) only, which saturated at approximately 250 molecules per platelet (K d ϳ7.5 Ϯ 2.0 nM). A combination of thrombin and collagen showed no significant increase in the affinity or the number of binding sites as compared with thrombin alone.
Effect of Cofactor-Cofactor Interaction on the Assembly of the Factor X Activating Complex-The cofactor involved in tenase complex (FVIII) and the cofactor of the prothrombinase complex (FV) are homologous in primary (amino acid) and domain structures and have been shown to partially compete with one another for binding sites on phospholipid vesicles and plasma phospholipid bilayers (31, 48 -50). We used FVa to compete with either FVIII or FVIIIa to resolve whether platelet-binding sites express specificity for FVIII and/or FVIIIa. Fig. 6A shows that FVa (1-500 nM) was unable to displace the pro-cofactor FVIII from the platelet membrane, thus suggesting that there exists a specific receptor for FVIII on the platelet membrane. This result is consistent with that observed by Nesheim et al. (22). In contrast, FVa had greater ability to displace the active cofactor (FVIIIa) with ϳ35-40% of bound FVIIIa at ϳ50 nM FVa (Fig. 6A). No further inhibition was observed by up to 50-fold greater excess of FVa. When we examined the effect of added FVa on the intrinsic FXase activity (i.e. in a functional assay) in the presence of FIXa (0.25 nM) and FX (150 nM) to thrombin-activated platelets (5 ϫ 10 7 /ml), we found that the titration with increasing amounts of FVa had no significant effect on the FX-activating complex (Fig. 6B).
Effect of Annexin V on the Factor VIIIa Binding Complex-In order to define further the nature of the putative receptors (i.e. phospholipid and/or protein) for the components of the FXactivation complex, we carried out competition studies with annexin V. Annexin V is an anionic phospholipid-binding protein that binds to human platelets with K d ϳ7.0 nM (51, 52). Previously we have shown that annexin V is a very potent inhibitor of both FX and prothrombin binding to activated platelets with IC 50 of 3.1 and 2.6 nM, respectively (23). We have also shown that annexin V affects the affinity and stoichiometry of FIXa interaction with platelets and phospholipids, and we have concluded that the interaction of both the enzyme (FIXa) and the substrate (FX) is mediated in a complex manner by both phospholipids and perhaps a protein receptor (53). Studies from other laboratories (52,54) have reported that annexin V interferes with FV and FVa binding and prothrombinase activity on both the platelet membrane and phospholipid vesicles. Therefore, we determined whether annexin V has the same effect on the cofactor (FVIII and FVIIIa) binding as we have reported for the other components of F-X-activating complex, i.e. the enzyme (FIXa) and the substrate (FX).
We carried out equilibrium binding studies with FVIII and FVIIIa in the absence and presence of saturating concentrations of EGR-FIXa (45 nM) and FX (1.5 M) with or without 5 nM annexin V. Table III summarizes the affinity and stoichiometry of FVIII and FVIIIa binding in both the absence and presence of 5 nM annexin V and clearly shows that annexin V (5 nM) has only a minor effect on pro-cofactor (FVIII)-binding sites, i.e. there is virtually no effect on the stoichiometry or affinity of FVIII binding in the absence of EGR-FIXa and FX, whereas in the presence of EGR-FIXa and FX the number of FVIII-binding sites decreased from 666 to 410 sites per platelet, whereas the K d values increased from 1.9 to 3.5 nM. In contrast, the effect of annexin V on active cofactor (FVIIIa) was apparent both in the absence and presence of EGR-FIXa and FX; in both instances, the number of binding sites was decreased (from 750 to 580 sites per platelet and from 1,200 to 510 sites per platelet, respectively) with concomitant decreases in the affinity (K d from 1.5 to 15 nM and from 0.8 to 19.5 nM, respectively). In Fig.  7 we demonstrate the ability of annexin V to compete with pro-cofactor (FVIII) and active cofactor (FVIIIa) by incubating thrombin-stimulated platelets with 125 I-FVIII or 125 I-FVIIIa and with various concentrations of either unlabeled FVIII, FVIIIa, or annexin V. When the residual binding of pro-cofactor ( 125 I-FVIII) was determined, it was apparent that excess unlabeled FVIII and annexin V prevented Ͼ95% of 125 I-FVIII binding. As calculated from the competition curves presented in Fig. 7A, the concentration of pro-cofactor (FVIII) required for the half-maximal inhibition of 125 I-FVIII binding in the presence of EGR-FIXa and FX was 1.9 Ϯ nM, and the IC 50 value of annexin V inhibition of pro-cofactor interaction was 65 nM (Fig.  7A). In contrast, when the residual binding of active cofactor ( 125 I-FVIIIa) was determined, it was apparent that excess FVIIIa or annexin V prevented Ͼ95% 125 I-FVIIIa binding; the concentration of FVIIIa required for half-maximal inhibition of 125 I-FVIIIa binding in the presence of EGR-FIXa and FX was 0.8 nM, whereas the IC 50 value of annexin V was 10 nM (Fig. 7B) which is close to the reported K d value (7 nM) for annexin V binding to activated platelets (52).
The Effects of Isolated Subunits of Factor VIII on the Binding  of Factor VIIIa to Activated Human Platelets-We used isolated subunits of FVIII to determine their effects on the binding of the cofactor to activated human platelets. The purity of these isolated subunits is shown in Fig. 1. The effects of the A1 domain and the A2 domain on the specific binding of FVIIIa to activated platelets in the presence of EGR-FIXa (45 nM) and FX (1.5 M) were compared. As shown in Fig. 8 and Table IV, the A1 domain (250 nM) of FVIII did not influence the affinity or the stoichiometry of FVIIIa binding. Similar results were obtained when the LC (A3-C1-C2) domain (250 nM) was supplemented in place of A1 domain (data not shown). However, the supplemental A2 domain (250 nM) enhanced the affinity of FVIIIa binding (K d ϳ0.6 from 0.9 nM) while also increasing the number of FVIIIa-binding sites (ϳ1,200 to 1,500 sites per platelet) ( Fig. 8 and Table IV). Addition of a 20-fold excess of A1 domain (250 nM) together with A2 (250 nM) subunit further enhanced the high affinity binding site observed in the presence of A2 domain only ( Fig. 8 and Table IV). These subunits had no significant influence on the binding of the pro-cofactor (FVIII).

DISCUSSION
The interaction between blood platelets and coagulation is essential for normal coagulation and hemostasis (17,18). Previous studies from our laboratory have demonstrated that platelets possess specific, high affinity, and saturable receptors for FIXa (21) and FX (23), and Nesheim et al. (22) have shown that FVIII also binds to platelets in a saturable and specific manner. Although there is ample evidence that FVIII and platelets (2,(17)(18)(19)44) or phospholipids (19,55) are essential cofactors that promote the proteolytic activation of FX by FIXa, the molecular mechanisms involved in the assembly of the F-X-activating complex on the platelet membrane are poorly understood. Studies by several investigators (56 -58) show that the activity and stability of the F-X-activating complex appears to be regulated by the integrity of the cofactor. More specifically, the active cofactor (FVIIIa) is structurally labile and the dissociation of the A2 subunit results in the decay of F-Xactivating complex. Therefore, in this study we have focused on the interaction of both the pro-cofactor (FVIII) and active cofactor (FVIIIa) with platelets and examined the affinity and stoichiometry of binding of both pro-cofactor and active cofactor to platelets in the absence and presence of enzyme (FIXa) and substrate (FX). It has been shown that the active cofactor (FVIIIa) is a substrate for the proteolytic inactivation by several enzymes, specifically FIXa, the enzyme for which FVIIIa serves as a cofactor. More specifically, it has been shown that prolonged reaction of the cofactor (FVIII) with the enzyme (FIXa) results in the proteolysis of the A1 subunit (i.e. cleavage at Arg 336 ) and results in the subsequent inactivation of the cofactor (59,60). Cleavage at this site liberates the carboxylterminal acidic region responsible for the A2 subunit retention. This results in reduced affinity of the A2 subunit for the A1/ A3-C1-C2 dimer but is not responsible for the instability of FVIIIa (thrombin-activated FVIII). Cleavage by FIXa at another site (Arg 1719 ) is not associated with the inactivation of the cofactor (61). Therefore, in our studies we have used active site blocked FIXa molecules (EGR-FIXa).
The activation of blood coagulation FX by enzyme FIXa occurs at physiologically significant rates only in the presence of calcium ions, cofactor FVIIIa, and either activated platelets or negatively charged phospholipids. Previous studies from our laboratory (20,21) have demonstrated the presence of specific, saturable receptors for human FIXa on thrombin-activated platelets (ϳ600 sites per platelets) with a K d of 2.5 nM. Similarly, Nesheim et al. (22) have shown that recombinant FVIII (labeled intrinsically with [ 35 S]methionine) also binds to platelets in a saturable and specific manner (500 sites per activated platelet; K d ϳ3 nM). In order to determine the binding affinities and stoichiometries of both the pro-and active cofactors to activated platelets, we made attempts to label the cofactor with seven different iodination procedures (30 -36). Our results indicate that maximum success was achieved with the Bolton-Hunter reagent for labeling FVIII that was relatively mild and resulted in the retention of Ͼ95% of specific functional activity with complete structural integrity of platelet-bound FVIII or FVIIIa (Fig. 2). The data presented in Fig. 3, A and B (open circles) demonstrate that both pro-cofactor (FVIII) and active cofactor (FVIIIa) interact directly and in a saturable manner with the platelet surface. The binding of pro-cofactor (FVIII) to activated platelets was identical (K d 3.7 nM; 484 sites per platelet) to that observed by Nesheim et al. (22), thus further confirming their observation. In contrast to the binding of pro-cofactor (FVIII), the active cofactor (FVIIIa) binds platelets with significantly higher affinity (K d ϭ 1.5 nM; 750 sites per platelet).
The experiments presented in this paper comprise a detailed comparison of the interaction of pro-cofactor (FVIII) and active cofactor (FVIIIa) with human platelets in the presence of enzyme/active site-blocked FIXa (i.e. EGR-FIXa) and substrate FX. It is possible to conclude from our results that a specific, high affinity binding site is induced on thrombin-activated platelets for both FVIII and FVIIIa when both EGR-FIXa and FX are present. Although both FVIII and FVIIIa clearly bind specifically and reversibly and in a saturable manner to thrombin-activated platelets in the absence of EGR-FIXa and FX (Fig. 3, A and B), the affinities of both the pro-cofactor (FVIII)   (62) showed that at low FVIIIa (Ͻ1 units/ml) with low FIXa (1 nM) and in the presence of phospholipid, porcine FVIIIa can be stabilized from spontaneous decay. It is possible that one of the mechanisms by which the presence of EGR-FIXa and FX enhanced the affinity and stoichiometry of binding of FVIII to platelets would be due to opposing the dissociation of FVIIIa subunits. This contention is supported by a recent study (63) in which FIXa transiently enhanced the reconstitution of FVIIIa from isolated subunits by a mechanism consistent with reducing the intersubunit dissociation rate constant. More recently, Fay (25) has proposed that for FIXa to stabilize FVIIIa, it may involve interaction of the enzyme with sites in the A2 subunit and A3 domain to essentially tether the labile subunits. Interestingly, in our recent studies (24) we have shown that platelet receptor occupancy by the substrate (FX) is a direct determinant of the rates of F-X activation by platelet-bound FIXa/ FVIIIa. Our studies indicate that FVIIIa, by itself, provides a high affinity (K d ϳ30 nM) binding site on platelets which interacts almost as tightly with FX as does the complete plateletbound FXase complex. Therefore, we hypothesized that at least part of the tremendous kinetic advantage that FVIIIa confers upon FIXa results from its ability to present platelet-bound FX to platelet-bound FIXa. Taken together the presence of both EGR-FIXa and FX confers both specificity and high affinity for the active cofactor (FVIIIa)-binding site on platelets and further emphasizes the validity of a three-receptor model in the assembly of F-X-activating complex on the platelet membrane. Our results, as discussed above, further support our hypothesis that cofactor (FVIII) active in F-X activation consists of active cofactor (FVIIIa) tightly bound (K d ϳ0.8 nM) to activated platelets in the presence of both tightly bound enzyme FIXa (K d ϳ0.5 nM) (21) and substrate FX (K d ϳ10 nM) (23). When we tested this hypothesis a step further by characterizing the bound cofactor, both structurally and functionally, we observed a close correlation between FVIIIa binding and F-X activation (Fig. 4). Those observations confirm that the bound cofactor is partially structurally intact (Fig. 2) and indistinguishable from the free cofactor and provide strong evidence that the functional F-X-activating complex consists of FVIIIa bound to plate-  5 M). Excess thrombin was neutralized with 50 mM PPACK before addition of FVIIIa. Binding was determined as detailed under "Experimental Procedures." Nonspecific binding was determined in the presence of excess unlabeled FVIII and was subtracted from total binding to obtain specific binding. The results shown represent specific binding of FVIII in the absence (E) or presence (q) of EGR-FIXa and FX. B, specific binding of 125 I-labeled FVIIIa to thrombin-activated platelets in the absence and presence of EGR-FIXa and FX. wGFP (3.5 ϫ 10 8 /ml) was incubated for 20 min at 37°C with human ␣-thrombin (0.1 units/ml), CaCl 2 (5 mM), and 125 I-labeled FVIIIa in the absence or presence of EGR-FIXa (45 nM) and FX (1.5 M). Binding was determined as detailed under "Experimental Procedures." Nonspecific binding was determined in the presence of excess unlabeled FVIIIa and was subtracted from total binding to obtain specific binding. The results shown represent specific binding of FVIIIa in the absence (E) or presence (q) of EGR-FIXa and FX. lets in the presence of EGR-FIXa and FX. Presently, we are focusing our attention on the coordinate interactions of the cofactor with either the enzyme FIXa (64) or the substrate FX (65). Nevertheless, the studies presented have strongly supported our three-receptor model, i.e. F-X activation is a platelet receptor-mediated process tightly coupled to receptor occupancy by the enzyme (FIXa), the cofactor (FVIIIa), and the substrate (FX).
When we compare the analogous assembly of prothrombinase complex on human platelets with our F-X-activating complex on platelet membranes, we note major differences between the two systems. FV(a) binds directly to platelets (66 -68) and then provides a high affinity binding site for FXa (67,69,70). In the presence of excess human FVa, there are between 200 (68, 69) and 5,000 (71) sites for FXa on human platelets with a K d between 30 (68 -70) and 200 pM (71). The binding of human FVa to human platelets is not saturated (68 -71) at concentrations up to ϳ12 nM, whereas in the presence of FXa and prothrombin ϳ3,000 molecules are bound per platelet (68). Nevertheless, the kinetic effect of FVa binding, in terms of its ability to enhance the catalytic efficiency of FXa toward prothrombin, saturates between 200 and 500 pM (68,71). Prothrombin activation occurs 300,000-fold more rapidly with bound than with free FXa, and 15-fold more rapidly in the presence of platelets (72). The major problem in studying the interaction of human FVa and FXa with human platelets is the fact that human platelets contain large amounts of FV (18 -25% of the total FV in human blood) which interferes with the determination of exogenous ligand receptor concentration (73). However, by using alternative techniques it has been shown that FXa and FVa bind in an equimolar complex (4,100 -5,100 functional sites per platelets) with K d ϳ10 Ϫ10 M (74). Another major complication in determining the binding of FV and FVa to platelets in the assembly of the prothrombinase complex is that FV is activated to FVa, not only by thrombin but also by platelet proteases, by FXa, and by activated protein C. All these activation mechanisms result in the cleavage of FV and FVa into products distinct from those generated by the action of thrombin (33) and the functional significance of those proteolytic cleavages remains to be determined.
In Fig. 5 we have demonstrated the effect of vWF on plateletcofactor (FVIII and FVIIIa) interaction. The interaction be-tween pro-cofactor (FVIII) and vWF is necessary for the normal survival of the cofactor in circulating blood (75,76). To function as a cofactor in the F-X-activating complex on the platelet membrane, FVIII first should be delivered to platelet surface by vWF (2). Recent studies by Li and Gabriel (16) on the physical exchange of FVIII between vWF and platelets clearly indicates that the state of activation of FVIII affects the partitioning of FVIII between its transport protein vWF and activated platelets. Our results indicate that excess vWF effectively competes with platelets for the binding of pro-cofactor (FVIII) and confirms the previous finding by Nesheim et al. (12,22). In contrast to pro-cofactor (FVIII), active cofactor (FVIIIa) binding to activated platelets was not perturbed by vWF. It is also essential to determine the role of platelet vWF (released from the ␣-granules) in both pro-cofactor (FVIII) and active cofactor (FVIIIa) binding. This issue has been elegantly addressed in a recent study (16) using electrophoretic quasielastic light scattering, demonstrating that platelets saturated with anti-vWF antibody showed no displacement of pro-cofactor (FVIII) from the activated platelet surface. As further proof of the validity of these results, we incubated platelets with polyclonal anti-vWF antibody and carried out competition experiments to show that platelet vWF does not influence the affinity of binding of either pro-cofactor FVIII or active cofactor FVIIIa (data not shown). Of note, we have used albumin density gradient washed gel-filtered platelets that contain no detectable levels of vWF. Also it has been recently reported (16) that in normal platelets, the vWF concentration is too low (Ͻ1 ng of vWF/IU of FVIII) to be significant in affecting cofactor expression.
Platelet activation by physiological agonists (e.g. thrombin and/or collagen) is critical for normal hemostasis because anionic phospholipids are exposed on the outer membrane surface providing binding sites for both the tenase complex (16,(21)(22)(23) and the prothrombinase complex (2,33). To determine the specific requirements for platelet activation in relation to the exposure of cofactor-binding sites, we examined the effects of several agonists (Table II). Our studies indicate that unactivated platelets do not bind either pro-cofactor (FVIII) or active cofactor (FVIIIa). In contrast, thrombin and SFLLRN activated platelets exposed binding sites for both pro-cofactor (FVIII) and active cofactor (FVIIIa), whereas epinephrine (10 M), collagen (10 g/ml), and ionophore A23187 (1 M) expose suboptimal binding sites for active cofactor (FVIIIa) only. Thrombin (1 nM) and the thrombin receptor peptide (SFLLRN-amide; 25 M) were the most potent activators, whereas ADP and collagen were ineffective in exposing the cofactor (FVIII)-binding site. These observations are consistent with those reported by Li and Gabriel (16). The partial (suboptimal) exposure of binding sites for FVIIIa but not FVIII is not the result of trace contamination with thrombin (left over from the activation of FVIII) since not all agonists were equivalent in their abilities. In comparison, neither FV nor FVa requires platelet activation to bind to human (68) or bovine (66,67) platelets. Furthermore, in the presence of FVa, FXa is also capable of binding to unactivated platelets via bound FVa (66,69,70). Platelet activation is, however, required for optimal prothrombin converting activity by the FXa-FVa complex (33).
In this report we have also addressed the possibility of the existence of specific platelet cofactor (FVIII and FVIIIa) receptor involvement in the assembly and expression of F-X-activating complex. There is a high degree of structural and functional similarity existing between FVIII and FV (1,2). Moreover, previous studies provide conflicting evidence for the existence of specific receptors for both FVa and FVIIIa. One group of investigators suggests that surface phospholipids provide the site for FVa and FVIIIa binding based on two important observations. First, it has been shown that activated platelets can both expose negatively charged phospholipids on their surface and release microparticles which apparently both expose phospholipids and bind FVa and FVIIIa (31,49,50). Second, the exposure of phospholipids correlates with the expression of enzymatic activities (48 -50). In contrast, other observations by several investigators suggest that a separate and specific receptor distinct from phospholipid may exist for both FVIIIa and FVa based on the following reasons. 1) Nesheim et al. (22) reported that FV and FVIII do not compete for binding sites on the platelet membrane suggesting that non-equivalent and separate binding sites may exist for these cofactors. 2) Competition studies with bovine platelets (66,68) indicate the existence of specific receptors for FVa which do not promote the binding of FV, thus indicating that there is a receptor on the platelet that is highly specific for FVa. 3) Recently, Tracy et al. (33) reported a monoclonal antibody that blocks the binding of FVa to platelets but not to phospholipid vesicles. 4) Functional studies on prothrombin activation (71) and our studies on the kinetics of F-X activation (55) show that the number of FVa and FVIIIa molecules bound to platelets at saturation differs by 10-fold. 5) Platelets from a patient with an unusual bleeding disorder (the Scott syndrome) appear to be deficient in both FVIIIa binding (77) and in FVa binding (78). All these observations strongly suggest the existence of platelet receptors for both cofactors, i.e. FV(a) and FVIII(a), as reviewed by Nesheim et al. (79). Our results shown in Fig. 6A further confirm the observations of Nesheim et al. (22) that FVIII and FV do not compete with one another for binding to activated platelets. Moreover, our results with active cofactor (FVIIIa) indicate the minimal (30%) competition between FVIIIa and FVa that we observed on the platelet surface does not affect the functional tenase complex and may represent a nonfunctional cofactorcofactor interaction that may be due to the greater number of FVa sites available relative to FV sites. Another alternative explanation for the existence of two populations of FVIIIabinding sites (one that is apparently non-functional (30%) and sensitive to exogenous FVa and another that is functional and not readily displaced by FVa) is that there may be structural differences in the radiolabeled FVIIIa molecules that allow displacement of denatured/slightly degraded, underfunctional FVIIIa by FVa but do not permit displacement of intact functional FVIIIa. However, these nonfunctional FVa sites have been shown to interact with platelets and PCPS (phosphatidylcholine (75%), phosphatidylserine (25%)) vesicles (32,33) in prothrombinase complex assembly. Recent studies indicate that FVa associates with phospholipid membranes through the A3 and C2 domains on the LC of FV (34), whereas FVIIIa binding has been reported to be mediated through only the C2 domain in the LC of FVIII (35).
In order to define further the specific membrane receptors or the nature of the putative receptors (i.e. phospholipid and/or protein) for the components of FX activation complex, we carried out annexin V-cofactor competition studies with both the pro-cofactor (FVIII) and active cofactor (FVIIIa). Annexin V is an anionic phospholipid-binding protein that binds to human platelets with K d ϳ7.0 nM (51, 52). Previously, we have shown that annexin V is a very potent inhibitor of binding both FX and prothrombin to activated platelets with IC 50 of 3.1 and 2.6 nM, respectively (23). We have also shown that annexin V affects the affinity and stoichiometry of F-IXa interaction with platelets and phospholipids, and we have concluded that the interaction of both the enzyme (FIXa) and substrate (FX) is mediated in a complex manner by both the phospholipids and a protein receptor (53). Studies from other laboratories have reported that annexin V interferes with FV and FVa binding and prothrombinase activity on both the platelet membrane and phospholipid vesicles (52,54). Therefore, we wished to determine if annexin V has the same effect on cofactor (FVIII and FVIIIa) binding as we have reported for the other components of the F-X-activating complex, i.e. the enzyme (FIXa) and the substrate (FX). A direct comparison of the data on the effect of annexin V on both the pro-cofactor (FVIII) and active cofactor (FVIIIa) as shown in Fig. 7, A and B, and in Table III reveals that the annexin V is not an effective inhibitor of pro-cofactor binding to platelets, whereas it inhibits active cofactor (FVIIIa) interaction both in the absence and the presence of EGR-FIXa and FX. The inhibition curves of FVIIIa in the presence of EGR-FIXa and FX are quite steep, and the slope of the curve is greater than that described by a simple competitive inhibition model. Thus, our equilibrium binding studies and competition studies further clarify that annexin V has an effect only on active cofactor (FVIIIa), similar to the effects we have reported on FIXa binding (53) and FX binding (23). We recently reported (53) on the kinetic effects of annexin V on FVIIIa enhancement of F-X activation on activated platelets and phospholipids. These results demonstrate a noncompetitive mechanism of inhibition without affecting the affinity of FVIIIa for platelets, whereas the inhibitory effects of annexin V on F-X activation on phosphatidylserine/phosphatidylcholine vesicles suggested that the enzyme (FIXa) is accommodated differently on platelets and artificial phospholipids. Taken together, these studies suggest that the nature of the receptor site for FVIIIa on the F-X-activating complex on activated platelets consist of both negatively charged phospholipids and a protein component.
To better understand the platelet membrane receptor-binding mechanisms of cofactor (FVIII and FVIIIa), we investigated the effects of isolated subunits of activated FVIII on the binding of FVIIIa to activated platelets. Earlier studies by Fay et al. (3,9,11,26) indicate that the A2 subunit of FVIIIa markedly increases the catalytic activity (i.e. k cat of FIXa-catalyzed F-X activation) by enhancing the reaction rate ϳ100-fold. Furthermore, reconstitution of heterotrimeric FVIIIa from the isolated A2 subunit and A1/A3-C1-C2 dimer is also enhanced severalfold in the presence of FIXa and phospholipid (57). Based on these observations, Fay et al. (57) suggested a primary role of the A2 domain in modulating the active site of FIXa in the F-X-activating complex. Our results confirm those of Fay et al. (26,57) that the presence of both A2 subunit and A1/A3-C1-C2 dimer stabilizes the F-X-activating complex. Our results clearly indicate that the presence of FVIII A2 subunit alone (and not A1 subunit) increases the affinity of active cofactor (FVIIIa) binding. Addition of both A1 and A2 domain further enhances the affinity (Fig. 8 and Table IV) of FVIIIa binding to activated platelets in the presence of EGR-FIXa and FX, thus further emphasizing that F-X activation is a platelet receptormediated process tightly coupled to receptor occupancy by FIXa, FVIIIa, and FX. More recent studies using functional assays by Fay et al. (80) further indicated A1 and A2 subunits of FVIIIa synergistically stimulate FIXa catalytic activity yielding an overall increase in k cat of over 1,000-fold, compared with FIXa alone. Taken together our results from equilibrium binding studies and those of Fay et al. (26,80) from F-Xactivating studies are consistent with the hypothesis that the primary mechanism for decay of F-X-activating complex under physiological conditions is the dissociation of the A2 subunit (58,61,62). In vivo the mechanism of spontaneous dissociation dominates the mechanism of proteolytic degradation. Further work is needed to understand the relative contribution of these mechanisms in FVIII inactivation.