Unique in Vivo Modifications of Coagulation Factor V Produce a Physically and Functionally Distinct Platelet-derived Cofactor

Platelet- and plasma-derived factor Va (FVa) serve essential cofactor roles in prothrombinase-catalyzed thrombin generation. Platelet-derived FV/Va, purified from Triton X-100 platelet lysates was composed of a mixture of polypeptides ranging from ∼40 to 330 kDa, mimicking those visualized by Western blotting of platelet lysates and releasates with anti-FV antibodies. The purified, platelet-derived protein expressed significant cofactor activity such that thrombin activation led to only a 2–3-fold increase in cofactor activity yet expression of a specific activity identical to that of purified, plasma-derived FVa. Physical and functional differences between the two cofactors were identified. Purified, platelet-derived FVa was 2–3-fold more resistant to activated protein C-catalyzed inactivation than purified plasma-derived FVa on the thrombin-activated platelet surface. The heavy chain subunit of purified, platelet-derived FVa contained only a fraction (∼10–15%) of the intrinsic phosphoserine present in the plasma-derived FVa heavy chain and was resistant to phosphorylation at Ser692 catalyzed by either casein kinase II or thrombin-activated platelets. MALDI-TOF mass spectrometric analyses of tryptic digests of platelet-derived FV peptides detected an intact heavy chain uniquely modified on Thr402 with an N-acetylglucosamine or N-acetylgalactosamine, whereas Ser692 remained unmodified. N-terminal sequencing and MALDI-TOF analyses of platelet-derived FV/Va peptides identified the presence of a full-length heavy chain subunit, as well as a light chain subunit formed by cleavage at Tyr1543 rather than Arg1545 accounting for the intrinsic levels of cofactor activity exhibited by native platelet-derived FVa. These collective data are the first to demonstrate physical differences between the two FV cofactor pools and support the hypothesis that, subsequent to its endocytosis by megakaryocytes, FV is modified to yield a platelet-derived cofactor distinct from its plasma counterpart.

Factor Va (FVa), 1 a heterodimeric protein composed of heavy chain (105 kDa) and light chain (74 kDa) subunits, is formed by limited proteolysis of factor V (FV) (1). In normal hemostasis, FVa functions as a non-enzymatic cofactor of the prothrombinase complex, which consists of a 1:1 stoichiometric and Ca 2ϩdependent complex of the serine protease factor Xa and FVa, bound to the membrane of appropriately activated platelets, and catalyzes the proteolytic conversion of prothrombin to thrombin (2). When incorporated into the prothrombinase complex, the catalytic activity of factor Xa is increased by approximately 5 orders of magnitude, and FVa contributes substantially to this increase (3). Removal of FVa from the prothrombinase complex results in a 10,000-fold decrease in the rate of thrombin generation (3), the physiologic effect of which is demonstrated in the bleeding diatheses expressed by FV-deficient individuals (4 -8) Factor V circulates in two pools in whole blood. The majority (75-80%) is found in the plasma as an inactive, single chain procofactor of 330 kDa, whereas the remainder (20 -25%) is found in the ␣-granules of platelets (9), where it is stored in a partially cleaved and activated state (9,10). The unique activation state of platelet-derived FV/Va and its proximity to the procoagulant surface of platelets make it an excellent source of cofactor activity for assembly and function of the prothrombinase complex at a site of vascular injury. Its paramount importance in the maintenance of normal hemostasis, when compared with that of plasma-derived FVa, has been demonstrated in several clinical studies. Individuals expressing substantially decreased levels of FV in their platelets exhibit a life-threatening bleeding diathesis despite their expression of relatively normal levels of plasma-derived FV (4). An individual with an antibody that neutralized greater than 99% of the plasmaderived FV pool endured surgical challenge as the antibody was without effect on platelet-derived FV/Va, indicating that the platelet stores of FV/Va were sufficient to maintain normal hemostasis (11). This latter observation is consistent with the ability of platelet transfusions to correct bleeding episodes in individuals who have expressed acquired FV inhibitors for several months (5).
Even though platelet-derived FV originates from the plasma pool of FV (12)(13)(14)(15), several studies have demonstrated distinct differences in the physical nature and proteolytic susceptibility of platelet-and plasma-derived forms of FV. Previous studies (16,17) have shown that platelet-derived FV/Va is composed of several peptides, some of which are distinct and have molecular weights unlike any intermediate species observed during the activation of the plasma-derived molecule. Whereas plasmaderived FV exists as a single chain procofactor expressing little if any cofactor activity (3), the platelet-derived FV/Va molecule, which is stored in the ␣-granule, has undergone partial proteolysis and expresses substantial cofactor activity upon platelet secretion (17)(18)(19). Furthermore, platelet-derived FV/Va is a unique substrate for several proteases when compared with its plasma counterpart. Kinetic analyses have demonstrated that plasma-derived FV is the preferred substrate for thrombin, whereas platelet-derived FV/Va is activated by factor Xa 50 -100 times more effectively than by thrombin (17). Plasmincatalyzed cleavage of platelet-derived FVa bound to activated platelets results in a 4-fold increase in cofactor activity that is sustained for as long as 4 h, whereas plasmin-catalyzed cleavage of plasma-derived FVa bound to activated platelets results in a rapid loss of cofactor activity (20). An intriguing observation was made in work published by Rand et al. (21) who demonstrated that the heavy chain of platelet-derived FV/Va is totally resistant to phosphorylation by a platelet-associated kinase, under circumstances where the plasma-derived molecule is readily phosphorylated. Finally, several studies have demonstrated that platelet-derived FVa is more resistant to activated protein C (APC)-catalyzed inactivation when compared with plasma-derived FVa (22,23). This apparent insensitivity to APC results from a reduced rate of cleavage at Arg 506 and is preferentially displayed when platelet-derived FVa is associated with the activated platelet membrane (22). Thus, platelet-derived FV/Va appears to be a unique cofactor when compared with its plasma counterpart.
All of the studies detailed above were done with plateletderived FV/Va present in platelet releasates prepared with a variety of agonists. To date, functional studies with purified, platelet-derived FV/Va have been precluded by an inability to isolate the platelet-derived cofactor pool in its native, functional state. Therefore, this work outlines for the first time a protocol to purify platelet-derived FV/Va in order to define its physical and functional characteristics allowing the discovery of molecular characteristics unique to platelet-derived FV/Va.
For the isolation of platelet-derived FV/Va, human platelets were obtained from either outdated platelet-pheresis packs or single unit donor packs (10 -12 days post-draw) purchased from Haematologic Technologies Inc. (Essex Junction, VT). Platelets from ϳ50 to 100 units of plasma were pooled and washed in the presence of PGE 1 (5 M) by centrifugation according to the procedure of Mustard et al. (37) with minor modifications (22). Platelets were centrifuged at 1,500 ϫ g in 250-ml conical polypropylene tubes for 15 min at ambient temperature. The supernatant was decanted, and the platelets were resuspended in a total of 1 liter of modified Tyrode's buffer (MTB; 0.14 M NaCl, 2.7 mM KCl, 12 mM NaHCO 3 , 0.42 mM NaH 2 PO 4 , 1 mM MgCl 2 , 2 mM CaCl 2 , 5 mM dextrose and 5 M PGE 1 , pH 7.35) with the addition of 5 units/ml heparin. Red blood cells and those platelets that were not readily resuspended were discarded, and the suspended platelets were centrifuged at 1,500 ϫ g. Platelets were washed two additional times in MTB buffer, and the final platelet pellet was resuspended in 20 mM Tris, 150 mM NaCl, pH 7.4, containing 5 mM CaCl 2 (TBS/Ca 2ϩ ). The final platelet suspension was counted on a model Z1 Coulter Counter (Coulter Corp., Miami, FL).
Isolation of Platelet-derived FV/Va-Platelet-derived FV/Va was isolated by a modification of the procedure for the purification of plasmaderived FV (30). Washed platelets were diluted to a final concentration 4 ϫ 10 9 platelets/ml in TBS/Ca 2ϩ containing 5 M PGE 1 yielding ϳ250 -350 ml of platelet suspension. Subsequently, the protease inhibitors benzamidine (10 mM), phenylmethylsulfonyl fluoride (0.5 mM), leupeptin (0.5 mM), and soybean trypsin inhibitor (20 g/ml) were added to prevent protein cleavage. Platelets were lysed with the addition of Triton X-100 (1% v/v) for 5 min at 25°C with occasional mixing yielding ϳ560 units of FV. The platelet lysate was then incubated at 0°C for 90 min to generate a cryoprecipitate, which was removed by centrifugation at 16,000 ϫ g for 40 min at 4°C with greater than 98% recovery of platelet-derived FV/Va. An anti-human FV monoclonal antibody (30) coupled to Sepharose CL-4B resin (50 ml resin; 3-4 mg antibody/ml resin) was added to the supernatant, and the suspension was slowly stirred at 4°C for 2 h to adsorb the FV onto the resin. The resin was then transferred to a sintered glass funnel, and the supernatant was removed by vacuum. The resin was then slurried with ambient temperature TBS/Ca 2ϩ containing 1 mM benzamidine (TBS/Ca 2ϩ /Benz), transferred to a 2.5 ϫ 20 cm chromatography column, and washed at ambient temperature with TBS/Ca 2ϩ /Benz until the corrected A 280 (A 280C ) was below 0.01. This was followed by an identical wash with TBS/Ca 2ϩ /Benz containing 0.35 M NaCl until an A 280C Ͻ 0.01 was reached in the flow through. The bound FV was subsequently eluted with TBS/Ca 2ϩ /Benz containing 5.0 M NaCl at a flow rate of ϳ0.2 ml/min. Collected fractions (5 ml) were assayed for FV cofactor activity using a prothrombin timebased clotting system employing FV-deficient plasma. Fractions containing detectable FV cofactor activity were pooled and generally exhibited nearly 10% of the starting FV/Va present in the platelet lysate. Buffer exchange of the pool into TBS/Ca 2ϩ was accomplished at 4°C using an Amicon ultrafiltration cell with a YM30 filter that had been soaked for 30 min in TBS/Ca 2ϩ containing 1% PEG-8000 to reduce FV binding. The pool was then concentrated down to a volume of ϳ0.15 ml, followed by the addition of glycerol/CaCl 2 (50%, 2 mM final, respectively), and stored at Ϫ20°C. The concentration of platelet-derived FV/Va was determined by western blot quantitation of the light chain fragment by densitometry using the NIH Image software package. Additionally, these calculated concentrations were confirmed by comparison of the fully activated platelet-derived cofactor to a standard curve of plasma-derived FVa. Typically, the purified platelet-derived FV/Va exhibited a 4 -8% yield and a specific cofactor activity of ϳ120 units/mg with greater than 140-fold purification.
SDS-PAGE and Western Blotting of Purified Plasma-and Plateletderived FV/Va-Purified platelet-derived FV/Va (0.5 M) in 20 mM HEPES, 150 mM NaCl, 5 mM CaCl 2 , pH 7.4 (HBS/Ca 2ϩ ), containing 0.1% PEG-8000 was treated with thrombin (20 nM) at 37°C for 15 min to ensure full activation in the presence of residual glycerol (ϳ6%) from the FV storage buffer. Additionally, a sample of purified plasma-derived FV was identically thrombin-activated for 10 min, sufficient to ensure full activation. The activated FVa solution was then treated with hirudin (30 nM) to terminate the reaction. Protein samples (50 l; 7 g) were diluted with 12.5 l of 312.5 mM Tris-HCl, 10% SDS, 50% glycerol, 10% 2-mercaptoethanol, 0.005% bromphenol blue, pH 6.8 (5ϫ SDS-sample preparation buffer), heated at 100°C for 2 min, and subjected to SDS-PAGE on 4 -12% polyacrylamide gels by the method of Laemmli (33). Gels were either silver-stained by the method of Merril et al. (38), or protein was transferred to nitrocellulose as described by Towbin et al. (39) and analyzed by western blotting techniques using either mouse anti-human FV monoclonal antibodies ␣HFVa LC 9 (26), ␣HFVa HC 17 (26), or a burro anti-human FV polyclonal antibody (9), as described previously (12). . The initial rate of thrombin generation in the various prothrombin activation mixtures was calculated by linear regression analysis of the data obtained from the subsamples removed over time. Binding constants were determined by non-linear regression analysis to equations described previously (40,41).
Analyses of Post-translational Modification of Platelet-and Plasmaderived FVa-For in-gel trypsin digestion and extraction of FVa peptides, FVa 105-kDa heavy chain and 74-kDa light chain bands were isolated by excision from a 5-15% gradient gel following SDS-PAGE (33) and digested with sequencing grade trypsin as described (42). Briefly, gel slices containing either FVa heavy chain or light chain peptides were minced into 1-mm 3 pieces and washed three times with 50% acetonitrile, 25 mM ammonium bicarbonate, pH 8.0, for 15 min. The samples were dehydrated with 100% acetonitrile for 5 min and dried under vacuum by rotary evaporation for 25 min. A 75-l solution of trypsin (0.01 g/l trypsin in 0.2 M ammonium bicarbonate, pH 8.0) was added to the dried gel pieces, and digestion proceeded for 21.5 h at 37°C. To elute the resulting FVa peptides from the gel pieces, the liquid phase was removed to a second siliconized Eppendorf tube and replaced with 50 l of 50% acetonitrile, 5% trifluoroacetic acid followed by incubation at ambient temperature for 60 min. The resulting supernatant was removed, added to the sample collected previously, replaced with an additional equal volume of the same solution, and subsequently incubated with the gel pieces an additional 30 min. The supernatants were then pooled and dried under vacuum by rotary evaporation. Samples for MALDI-TOF analyses were prepared using a ZipTip C18 micro column as per the manufacturer's instructions and spotted on a 100well stainless steel sample plate in 10 mg/ml ␣-cyanohydroxycinnamic acid as matrix. Post-translational modifications present on the peptides were identified by comparison of unmatching peak masses to all known eukaryotic post-translational modifications using the FindMod tool of the ExPASy system (www.expasy.ch/tools/findmod) from the Swiss Institute of Bioinformatics. Identification of N-linked glycoforms was obtained by input of unmatching peak masses into the GlycoMod Tool (www.expasy.ch/cgi-bin/glycomod_form.html) of the same system.

Phosphorylation of Purified Plasma-and Platelet-derived FVa by
Casein Kinase II-In order to compare the relative susceptibility of the plasma-and platelet-derived FVa heavy chains to phosphorylation by casein kinase II, purified plasma-derived FV (1 M) was activated to FVa with thrombin (40 nM) for 10 min at 37°C followed by the addition of hirudin (60 nM) to inhibit further thrombin function. Purified platelet-derived FV/Va (1 M) was activated to FVa with thrombin (40 nM) for 40 min at 37°C to ensure full activation in the presence of residual glycerol (ϳ12%) from the FV storage buffer, and the reaction was terminated by the addition of hirudin (60 nM). Both reactions were performed in HBS/Ca 2ϩ containing 0.1% PEG-8000. Samples of plasmaderived FVa (8 -500 nM) or platelet-derived FVa (125 and 250 nM) were treated with casein kinase II (1.25 ng/l) in HBS/Ca 2ϩ containing 0.1% PEG-8000 in the presence of ATP (50 M), [␥-32 P]ATP (0.15 Ci/l), and MgCl 2 (1 mM) for 3 h at 30°C. To control for residual glycerol (ϳ3%) present in the 250 nM platelet-derived FVa sample, all samples were made 3% in glycerol. Reactions (40 l) were terminated by the addition of 5ϫ Laemmli SDS-sample preparation buffer (10 l) followed by boiling at 100°C for 2 min. Samples were subjected to SDS-PAGE (33) on 5-15% linear polyacrylamide gradient gels followed by autoradiography at Ϫ80°C.

Phosphorylation of Purified Plasma-and Platelet-derived FVa by an Activated Platelet-associated Casein Kinase II-like Enzyme-Samples
(20 l) of purified plasma-or platelet-derived FV/Va (1 M) in HBS/ Ca 2ϩ containing 0.1% PEG-8000 were added to suspensions (1 ml) of washed platelets (1 ϫ 10 9 platelets/ml) containing RGDS (1 mM) and [␥-32 P]ATP (0.20 Ci/l) in HTA. Thrombin was added to 1 nM, and the suspension was rocked for 5 min at 25°C, followed by the addition of hirudin (1.5 nM) to inactivate the thrombin. The platelets were removed by centrifugation (1000 ϫ g, 5 min, 25°C), and the supernatant (64 l) was diluted in 5ϫ SDS sample preparation buffer (16 l) and heated at 100°C for 2 min. Samples were subjected to SDS-PAGE (33) on 5-15% linear polyacrylamide gels followed by autoradiography as described above. Identical samples (40 l) were diluted in 5ϫ Laemmli SDS sample preparation buffer (SPB) (10 l) and subjected to SDS-PAGE (33) and western blotting with a mouse monoclonal antibody (␣HFVa HC 17).
Physical Determination of the Site of Phosphate Incorporation into Plasma-derived FVa Heavy Chain-Plasma-derived FV (1 M) in HBS/ Ca 2ϩ PEG (20 mM HEPES, 0.15 M NaCl, 5 mM CaCl 2 , 10 mM MgCl 2 , 0.1% PEG-8000, pH 7.4) was activated with thrombin (2 units/ml, 10 min, 37°C) followed by the addition of hirudin (30 nM). Plasma-derived FVa (500 nM) was then incubated with casein kinase II (50 ng/l) and ATP (10 mM) for 3 h at 37°C. The reaction was quenched in 5ϫ SPB, and the phosphorylated plasma-derived FVa heavy chain was separated by SDS-PAGE (33) and visualized by Coomassie Blue staining. Samples were trypsin-digested, and the resulting fragments were analyzed by MALDI-TOF mass spectroscopy as described above.
APC-catalyzed Inactivation of Purified Plasma-and Platelet-derived FVa on an Activated Platelet or PCPS Vesicle Surface-Purified plasma-or platelet-derived FV/Va (20 nM) was added to washed platelets (1 ϫ 10 9 /ml) in the presence of RGDS peptide (1 mM) to inhibit thrombin-induced platelet aggregation. The mixture was treated with thrombin (1 nM) for 5 min at ambient temperature followed by the addition of hirudin (1.5 nM) to inhibit thrombin activity. APC (0.25 nM) was added to the mixture, and timed aliquots were removed and assayed for FVa cofactor activity in a clotting assay employing FV-deficient plasma according to the method of Nesheim et al. (31). Alternatively, samples of the reaction mixture were analyzed for cofactor activity in a purified system. Samples (50 l) of the reaction mixture were withdrawn at specific time intervals and assayed for remaining cofactor activity with saturating amounts of factor Xa (5 nM) and PCPS vesicles (20 M) as the membrane surface as described previously (3,22). In order to monitor the inactivation of purified platelet-and plasma-derived FVa on a synthetic phospholipid surface subsequent to activation in the presence of platelets, platelets and FV species were activated as described above, and the platelets were removed by centrifugation in a microcentrifuge for 5 min at 1000 ϫ g. PCPS vesicles (10 M) were added to the supernatant to provide a phospholipid surface, and the samples were treated with APC (0.25 nM) and assayed for FVa activity in a clotting assay as described above.

RESULTS
Purification and Characterization of Platelet-derived FV/ Va-Purification of platelet-derived FV/Va was achieved using an immunoaffinity chromatography protocol similar to that described for the purification of the plasma-derived cofactor (41). However, elution of the immunoadsorbed, platelet-derived FV/Va protein and cofactor activity required buffers containing 5.0 M NaCl, in marked contrast to the 1.2 M NaCl capable of eluting immunoadsorbed, plasma-derived FV, indicating that the two cofactors bound to the antibody-coupled resin with very different affinities. Although typical yields of the plasma-derived procofactor approached ϳ45% by using this protocol, final yields of purified platelet-derived FV/Va approached ϳ6% (n ϭ 10). Despite this low yield, the product obtained was representative of the entire platelet-derived FV/Va pool as detailed below.
Silver staining (Fig. 1B) of purified platelet-derived FV/Va samples indicated that unlike plasma-derived FV, which is defined as a single chain 330-kDa procofactor (30,(43)(44)(45), platelet-derived FV/Va contained only a small fraction of the procofactor and several lower molecular weight peptides (Fig.  1B, lane 2). This observation was anticipated as earlier studies indicate that platelet-derived FV/Va is stored in ␣-granules as a partially activated cofactor, due to limited proteolytic processing (17) (Fig. 1A). In fact, when compared with partially activated purified plasma-derived FVa (Fig. 1B, lane 1), these lower molecular weight peptides migrated similarly to the 220-kDa FVa activation intermediate (residues 1018 -2196), as well as the 105-kDa heavy chain fragment (residues 1-709), and the light chain fragment which appeared as a doublet of 72 and 74 kDa (residues 1546 -2196). A peptide of ϳ90 kDa, which resembled a thrombin-derived FVa heavy chain species previously reported by Hockin et al. (46), was also apparent. Treatment of purified platelet-derived FV/Va with thrombin (Fig.  1B, lane 3) resulted in the disappearance of the peptides larger than 105 kDa and increased the levels of the 105-kDa heavy chain and the 74-kDa light chain fragment, whereas little change was observed in the 72-kDa light chain fragment.
The samples depicted by silver staining in Fig. 1B were also analyzed by using western blotting techniques (Fig. 1C) that employed two monoclonal antibodies, one which recognized an epitope within the heavy chain region of FV and one which recognized an epitope within the light chain region of FV. Comparison of these data in Fig. 1, A and B, indicated that, with two exceptions, the silver-stained peptides represented platelet-derived FV/Va. The silver-stained peptide identified with an * represented albumin contamination that was also present in the purified plasma-derived FVa sample, and the peptide identified with ** most likely represented IgG heavy chain. Western blotting analyses of platelet-derived FV/Va present in a representative platelet lysate (Fig. 1A) obtained from a single donor, indicated that all the platelet-derived FV/Va components present in a platelet lysate were also present in the purified platelet-derived cofactor sample representative of as many as 120 donors. Thus, purified platelet-derived FV/Va is representative of the total platelet-derived cofactor pool.
Functional characterization of purified platelet-derived FV/Va indicated that the purified protein expressed significant FVa cofactor activity that could only be increased 2-3-fold following full activation with thrombin (Table I), an activity that mimicked the cofactor activity observed in freshly prepared platelet lysates. In addition, thrombin-activated, purified platelet-derived FVa expressed a specific activity identical to that of purified plasma-derived FVa when assayed in a clot-based system using FV-deficient plasma (data not shown). These combined data confirmed that the purification protocol did not affect either the intrinsic or total cofactor activity of the purified, platelet-derived FV/Va molecule.
Analyses of the kinetics of prothrombin activation were used to define the apparent K D value governing the interaction of platelet-derived FVa with the other constituents of the prothrombinase complex assembled on a membrane provided by PCPS vesicles. The apparent K D values obtained for platelet-or plasma-derived FVa were nearly identical (0.76 Ϯ 0.1 nM and 0.72 Ϯ 0.1 nM, respectively) (Fig. 2), indicating that the incorporation of platelet-derived FVa into prothrombinase mimicked its purified plasma-derived counterpart both in the required binding interactions, as well as in its expression of cofactor activity.
APC-catalyzed Inactivation of Purified Plasma-and Plateletderived FVa on Both the PCPS Vesicle Membrane and the Thrombin-activated Platelet Membrane-Although purified platelet-and plasma-derived FVa expressed similar cofactor activities in both clot-based and purified assay systems, several unique characteristics have been ascribed to the platelet-derived cofactor. Notably, in experiments using platelet-released and -bound FVa, platelet-derived FVa was demonstrated to be more resistant to APC-catalyzed inactivation when compared with added plasma-derived FVa (23). To confirm that this unique characteristic was retained by purified, platelet-derived FVa, purified platelet-or plasma-derived FV was subjected to APC-catalyzed inactivation on either the thrombin-activated platelet membrane or that of PCPS vesicles. Under the conditions used, endogenous platelet-derived FVa (Ϸ0.25 nM) would have represented a mere 1% of the total FV pool. The rate and extent of the APC-catalyzed inactivation of purified plasma-or platelet-derived FVa bound to PCPS vesicles were identical (Fig. 3, inset). In marked contrast, when activated platelets remained in the reaction mixtures to provide the required membrane surface, purified plasma-derived FVa was inactivated by APC Ϸ2-3 times more rapidly than was purified, platelet-derived FVa (Fig. 3). Thus, the resistance of plateletderived FVa to APC-catalyzed inactivation observed previously in in situ assays, was also exhibited by the purified plateletderived cofactor indicating that the purification protocol allowed for the retention of characteristics unique to plateletderived FVa. Rand et al. (21) demonstrated that, in contrast to purified, plasma-derived FVa, FVa released during thrombin-mediated platelet activation is resistant to phosphorylation on its heavy chain by a platelet-associated casein kinase II-like enzyme. Thus, to determine whether purified platelet-derived FVa retained this characteristic, its susceptibility to phosphorylation in the presence of activated platelets was determined. Samples of purified plasma-or platelet-derived FV/Va (1 M) were added to resting platelets (1 ϫ 10 9 platelets/ml) in the presence of [␥-32 P]ATP followed by the addition of thrombin to fully activate both the platelets and FV present in the mixture. Although radioactivity was incorporated into the heavy chain region of the plasmaderived FVa molecule (Fig. 4A, left panel, lane 2), the purified, platelet-derived FVa was completely resistant to phosphorylation (Fig. 4A, left panel, lane 3), even though equal amounts of cofactor protein were present as indicated by western blotting analyses of the reaction mixtures (Fig. 4A, right panel, lanes 2 and 3). These data were supported by an inability to phosphorylate the intrinsic platelet-derived cofactor pool (Fig. 4A, left panel, lane 1), even though sufficient protein was present for detection ( Fig. 4A, right panel, lane 1).

Susceptibility of Platelet-and Plasma-derived FV to Phosphorylation Catalyzed by Purified Casein Kinase II and a Platelet-associated Kinase-Previous studies by
Identical results were obtained in a purified system in which thrombin-activated plasma-and platelet-derived FVa were subjected to phosphorylation by human casein kinase II in the presence of [␥-32 P]ATP. Autoradiographic visualization of the incorporated phosphate demonstrated that the heavy chain of plasma-derived FVa stoichiometrically incorporated radioactivity in a concentration-dependent manner (Fig. 4B, lanes  1-7). In marked contrast, minimal radioactivity was incorporated into the platelet-derived FVa heavy chain even at concentrations as high as 250 and 500 nM (Fig. 4B, lanes 8 and 9,  respectively).
As unmodified platelet-derived FVa exhibits resistance to inactivation by APC, experiments were performed to determine the native levels of phosphate present in purified platelet-derived FV. Samples of purified plasma-or purified platelet-derived FV/Va were activated to FVa and subjected to SDS-PAGE (33) followed by western blotting with an anti-phosphoserine polyclonal antibody. When compared with the heavy chain bands observed in plasma-derived FV (Fig. 4C, lane 1), very little phosphoserine was detected in the heavy chain bands of platelet-derived FVa (Fig. 4C, lane 2). However, the anti-phosphoserine antibody displayed similar reactivity with the light chain bands of the two FVa species. These data demonstrate that the site of phosphate incorporation in plasma-derived FVa is most likely not occupied by phosphate in the platelet-derived cofactor and further demonstrate different levels of intrinsic phosphate present on the two cofactors.  Casein Kinase II Phosphorylates Plasma-derived FVa at Ser 692 -Studies by Kalafatis (58) using peptides mimicking discrete FVa heavy chain regions suggested that Ser 692 was phosphorylated by casein kinase II on plasma-derived FVa. As the addition of phosphate is blocked on the heavy chain of platelet-derived FVa, and no intrinsic phosphate was detected on platelet-derived FVa heavy chain, experiments were performed employing MALDI-TOF mass spectroscopy to determine whether this residue is indeed the site of casein kinase II-catalyzed phosphorylation on plasma-derived FVa. MALDI-TOF analyses of platelet-and plasma-derived FVa heavy chain tryptic fragments allowed recognition of ϳ40% of the total FVa heavy chain sequence. Control, unphosphorylated plasma-derived FVa HC analyses yielded a peak at m/z of 2087.8 corresponding to residues 685-701 (Fig. 5A). Following phosphorylation by casein kinase II, this peak disappeared, and a peak of m/z ϭ 2167.6 appeared indicating a shift of m/z of 79.9 and corresponding to the addition of a phosphate on residues 685-701 (Fig. 5B) on which Ser 692 is the only potential site of phosphate incorporation by casein kinase II. In agreement with the above results, similar spectra of native platelet-derived FVa heavy chain demonstrated no intrinsic phosphate at Ser 692 (Fig. 6A). Furthermore, the absence of phosphate at Ser 692 in plasma-derived FVa heavy chain prior to phosphorylation by casein kinase II indicates that the intrinsic phosphate unique to plasma-derived FVa is located at an alternate site that was not observed in the current study.
Platelet-derived FV/Va Is Specifically Modified by the Addition of an O-Linked Sugar Residue-MALDI-TOF mass spec-troscopy was used to identify the presence of unique characteristics intrinsic to platelet-derived FVa that may prevent phosphorylation at Ser 692 . MALDI-TOF analyses of plateletderived FVa heavy chain tryptic fragments revealed a peptide of m/z ϭ 2087.8 corresponding to residues 685-701 indicating no native modification at Ser 692 in the platelet-derived FVa heavy chain (Fig. 6A). However, one additional and significant peak at m/z ϭ 679.5 was observed (Fig. 6, A and B). Consideration of the known post-translational modifications in eukaryotic proteins indicated that this mass corresponds to residues 401-404 in the heavy chain with the addition of an O-linked GlcNAc or GalNAc at Thr 402 . A similar modification could not be observed in plasma-derived FVa from several different purified preparations (n ϭ 3) (Fig. 6C).
MALDI-TOF analyses also allowed the determination of the locations of other post-translational modifications within platelet-and plasma-derived FVa. The observation that residues 685-701 were unmodified in human platelet-and plasma-derived FVa is in contrast to evidence indicating the presence of a sulfate at Tyr 696 and Tyr 698 as was hypothesized previously (47). Further analyses also indicated that Tyr 665 did not contain sulfate.
Additionally, no differences were observed in the light chains of platelet-and plasma-derived FVa even though greater than 70% of the residues could be identified (data not shown). A peak at m/z ϭ 2082.8 was observed representing residues 1981-1985 within the light chain containing a biantennary monosialylated non-fucosylated glycosylation at Asn 1982 . Peaks were also observed at m/z ϭ 2651.4 and 2099.8 indicating a triantennary-

FIG. 3. APC-catalyzed inactivation of purified plasma-and platelet-derived FVa on the thrombin-activated platelet surface or PCPS vesicles.
Purified plasma-or platelet-derived FV/Va (20 nM) was added to washed platelets (1 ϫ 10 9 /ml) and subsequently treated with thrombin (1 nM) for 5 min. APC (0.25 nM) was added to the mixture, and timed aliquots were removed and assayed for FVa cofactor activity in a prothrombin time-based clotting assay. Inset, the purified FV species and platelets were activated as described above. Subsequent to removal of the platelets by centrifugation (1000 ϫ g, 5 min), PCPS vesicles (10 M) were added. APC (0.25 nM) was added, and FVa cofactor activity was assayed as described above. The data are expressed as the mean Ϯ S.D. (n ϭ 3) of the plasma-derived (f) or platelet-derived (E) FVa cofactor activity remaining at each time interval subsequent to the addition of APC. monosialylated, non-fucosylated and a triantennary core-fucosylated modification at this site, respectively. These N-linked modifications corresponded well to the carbohydrate profile established by Kumar et al. (48) and to published modifications observed on factor VIII (49). We also observed a peak of m/z ϭ 3091.3 corresponding to a triantennary-monosialylated corefucosylated modification at Asn 2181 . Glycosylation at this site was documented by other investigators and gives rise to the observed doublet in the FVa light chain following SDS-PAGE (33,50,51). Two other potential sites of N-linked glycosylation exist on the FVa light chain; however, peptides corresponding to these sites were not observed.
Identification of a Unique Activating Cleavage Site in Platelet-derived FV-Platelet-derived FV/Va contains several proteolytic cleavage products, two of which comigrate with the heavy and light chains of thrombin-activated plasma-derived FVa. The presence of these two fragments suggests intrinsic levels of cofactor activity are present in unactivated plateletderived FV/Va. Thus, to identify these fragments, also present in purified platelet-derived FV/Va, analyses of the C-and Nterminal regions were performed. N-terminal sequencing of the band migrating at 105 kDa produced a sequence corresponding to the N terminus of the FVa heavy chain (Fig. 7). Identical analyses of the 72-kDa band within platelet-derived FV/Va produced a sequence representing residues 1543-1552 corresponding to the recognized FVa light chain with the addition of two extra N-terminal residues (Fig. 7). These data indicated that the processing of platelet-derived FV/Va results in a cleav-age at Tyr 1543 by an as yet unidentified protease, in contrast to the normally recognized thrombin cleavage site at Arg 1545 . Additional N-terminal sequence analyses following thrombin activation of platelet-derived FV/Va indicated that this extended light chain is not further processed during activation to platelet-derived FVa, and two distinct sequences were produced. One sequence corresponded to the cleavage at Tyr 1543 defined above, and the other sequence mimicked that produced by thrombin cleavage at Arg 1545 of the intact single chain FV and some of the high molecular weight activation intermediates present in preparations of platelet-derived FV/Va (Fig. 7).
The C termini of these fragments were characterized by MALDI-TOF analyses. Tryptic digests of the 105-kDa fragment yielded identification of a peak corresponding to the last eight residues of the well characterized heavy chain, residues 702-709 (m/z ϭ 784.7) (Fig. 8A). Similar analyses of the 72-kDa band identified a fragment corresponding to residues 2188 -2196 (m/z ϭ 1143.7), the final amino acids in the C terminus of the FVa light chain peptide (Fig. 8B). However, an additional tryptic fragment within the FVa light chain (1622-1631) would have an identical mass; thus it is possible that the observed peak represents the sum of both regions in FV. Regardless, these data demonstrate that platelet-derived FV/Va contains intact heavy chain FVa and a uniquely produced FVa light chain. DISCUSSION This report details the first direct analyses of the physical properties of platelet-derived FV/Va, which were achieved through the development of a method for the purification of FV from platelets. As anticipated, several peptides ranging in molecular mass from 330 to 72 kDa were present in the purified product as indicated by silver staining. With two exceptions, the peptides were detected by one or more antibodies specific for FV and mimic those peptides attributable to FV observed in   (33) and visualized by Coomassie Blue staining. The stained bands were excised from the gel and subjected to trypsin digestion, and the resulting fragments were analyzed by MALDI-TOF mass spectroscopy as described under "Experimental Procedures." The peptide peak representing the naturally circulating unmodified residues 685-701 in A is shifted 79.9 Da higher in B, the mass of a single phosphate moiety, following phosphorylation of the plasma-derived FVa by casein kinase II. platelet lysates. When fully cleaved and activated by thrombin, the purified platelet-derived FVa expresses specific activity identical to that of purified plasma-derived FVa. Furthermore, the purified, platelet-derived FV/Va possesses all of the defining characteristics observed previously in the FV/Va molecule released from platelets as follows: 1) a resistance to phosphorylation by either purified casein kinase II or a casein kinase II-like enzyme associated with platelets (21); 2) an activation quotient similar to that observed in FV released from platelets upon activation with collagen and ADP (17,18); and 3) an apparent resistance to APC (22,23). These observations, along with the use of extensively washed platelets, dismiss the possibility that the purified protein is contaminated with substantial quantities of plasma-derived FV that remained adsorbed to the platelet surface.
An additional defining characteristic of platelet-derived FV/Va is the expression of significant levels of cofactor activity prior to exposure to its well characterized activators such as FIG. 6. MALDI-TOF mass spectroscopic analyses of platelet-and plasma-derived FVa heavy chains demonstrate glycosylation specific to the platelet-derived cofactor. Tryptic digests of the heavy chains of platelet-(A and B) and plasma (C)-derived FVa were analyzed by MALDI-TOF mass spectroscopy. Detected peaks represented ϳ40% sequence recognition in each of the two heavy chains. The full spectra of both platelet-and plasma-derived FVa heavy chains displayed identical peak recognition (data not shown) with the exception of a significant peak at m/z ϭ 679.5 present only in spectra of platelet-derived FVa heavy chain, which corresponds to residues 401-404 with the addition of an Olinked sugar residue on Thr 402 . Peak labels represent amino acid numbering in FV. Several peak labels have been removed for clarity. The platelet-derived FVa heavy chain produces a unique peak at m/z ϭ 679.9 indicating the presence of a ␤-O-N-acetylglucosamine or N-acetylgalactosamine residue incorporated at Thr 402 . The spectra of the platelet-derived FVa (B) and plasma-derived FVa (C) heavy chains are expanded to illustrate the region between m/z ϭ 650 and m/z ϭ 700. For reference, the peak at m/z ϭ 684.2 (**) is a nonspecific peak representative of background spectra. thrombin. The purified platelet-derived cofactor expressed substantial intrinsic activity, mimicking that observed in FV/Va obtained from unfractionated platelet lysates, and which demonstrated the typical 2-3-fold increase in cofactor activity subsequent to thrombin-catalyzed activation. This is in sharp contrast to the 20-fold increase in cofactor activity observed following thrombin-catalyzed activation of plasma-derived FV even though the fully activated platelet-derived products clearly display indistinguishable levels of activity. This demonstrates that the multicomponent purified platelet-derived FV/Va fully represents the FV present in platelets. Furthermore, MALDI-TOF mass spectroscopic and N-terminal sequencing data indicate that the bands present as components of platelet-derived FV/Va that comigrate with the heavy chain and light chain FVa are indeed full-length heavy chain and slightly extended light chain FVa.
Although several studies have characterized multiple proteolytic cleavage sites in FV mediated by several proteases, cleavage at Tyr 1543 has not been identified previously. Neither thrombin nor factor Xa has demonstrated a propensity for cleavage at this site. Although the megakaryocyte protease(s) that produce these active fragments remains elusive, calpains, specifically the abundant -calpain present in megakaryocytes (52), have been demonstrated to cleave at tyrosine residues (53), and calpains present in platelets have also been demonstrated to cleave FV (54). Thus, proteases are present that may lead to this unique cleavage. Although cleavage at Tyr 1543 has not been specifically demonstrated to produce cofactor activity, recombinant FV following protease-mediated removal of residues 822-1492 expresses 30% cofactor activity (55,56), and further cleavage at Arg 1545 produces full cofactor activity (56). Thus, the presence of two residues C-terminal to the recognized fully activating cleavage site of Arg 1545 , as was observed in platelet-derived FV/Va, most likely imparts full levels of cofactor activity. Even though additional data are required to determine how the many peptide species that comprise plateletderived FV/Va are produced within the megakaryocyte, these combined data provide compelling evidence that subsequent to their adhesion and activation at the site of injury, platelets release a cofactor molecule capable of interacting with factor Xa to generate thrombin and obviating the need to activate the plasma-derived cofactor.
Platelet-derived FV/Va is not only capable of initiating thrombin generation at the injury site but also ensures a sustained platelet procoagulant phenotype. Previous data (22,23) from our laboratory have demonstrated that, in contrast to plasma-derived FVa, a substantial fraction of the FVa released from thrombin-stimulated platelets is resistant to inactivation by APC on the activated platelet surface. The current study confirms this early work using purified platelet-derived FVa indicating that some physical characteristics leading to APC resistance are inherent to the platelet-derived FVa molecule. We have hypothesized that differences in post-translational modifications between the two cofactor pools could account for their differential susceptibility to APC; however, until now, absolute comparison of the two pools of FVa in a purified system has been difficult due to the lack of purified plateletderived FV/Va, which expressed appropriate cofactor activity. Additional studies indicate that the presence of the thrombinactivated platelet surface affects a greater degree of APC resistance in the platelet-derived FVa molecule than a synthetic phospholipid surface (23,57). Thus, qualities inherent to the platelet-derived FVa molecule are not solely responsible for the observed resistance to APC, since the thrombin-activated platelet surface plays a requisite role as well.
The physical and structural features of platelet-derived FV/Va that may impart its many unique functional properties are now beginning to be identified. In this study, we have demonstrated that purified platelet-derived FV/Va is characterized by a C-terminally derived light chain that has been cleaved at Tyr 1543 , rather than Arg 1545 . Platelet-derived FV/Va is also O-glycosylated at Thr 402 with either an N-acetylglucosamine or an N-acetylgalactosamine moiety and cannot be phosphorylated at Ser 692 even though this residue remains unmodified as determined by MALDI-TOF mass spectroscopic analyses. In contrast, we have provided unequivocal proof that this site can be modified in the plasma-derived cofactor. Studies by Kalafatis (58) indicate that phosphorylation at this site may lead to an increase in the rate of cofactor inactivation. Hence, the resistance to phosphorylation observed in plateletderived FVa may explain its resistance to APC inactivation Recent studies from our laboratory clearly indicate that the entire pool of human platelet-derived FV/Va originates from megakaryocyte endocytosis of plasma-derived FV in vivo (13), an observation that is in marked contrast to studies in guinea pigs (59) and mice (60), yet consistent with an inability to detect endogenous FV antigen in megakaryocytes (13). Although the endocytosed FV would initially contain the types and levels of post-translational modifications found in plasmaderived FV, subsequent retailoring and modification by the megakaryocyte is likely to occur based on the reduction of intrinsic levels of phosphorylation and cleavage by a megakaryocyte protease. We hypothesize that plasma-derived FV is endocytosed by megakaryocytes, post-translationally retailored (possibly via routing to the trans-Golgi network), packaged into ␣-granules, and processed proteolytically to yield a functional pool of the platelet-derived cofactor.
In support of this hypothesis, recent studies (14) from our laboratory indicate that FV, subsequent to its endocytosis by CD34 ϩ -derived megakaryocytes and megakaryocyte-like cell lines, undergoes proteolysis to yield FV fragments that are similar in size to those observed in platelet lysates. Likewise, modification of Thr 402 on endocytosed FV with a GalNAc residue would suggest that endocytosed FV is routed to the trans- Golgi network. Protein glycosylation, subsequent to endocytosis, is not unique. For example, ricin movement through the endoplasmic reticulum prior to cytosolic access results in its glycosylation (61). Alternatively, modification of Thr 402 on endocytosed FV with a GlcNAc residue would suggest that this event occurs in a megakaryocyte endosomal compartment since GlcNAc transferase is a neutral, membrane-bound enzyme (62)(63)(64)(65). GlcNAc modification of proteins is widespread and imparts unique function to the altered protein. This modification appears to block phosphorylation sites on Ser and Thr residues (66) and regulate nuclear targeting (67). GlcNAc modification of proteins can protect them from proteolysis (68), which may explain the resistance to inactivation by APC that is only observed with platelet-derived FVa (23).
The addition of a single O-linked sugar to a protein can have significant impact on its structure. The addition of a single ␣-galactosamine to the backbone of the HIV-1 IIIb GP120 peptide RP135 results in substantial structural changes both in the area immediate to the sugar, as well as at distant sites on the peptide backbone (69). Therefore, GlcNAc or GalNAc residue at Thr 402 in platelet-derived FV/Va may create a structurally altered cofactor resulting in its unique functional characteristics.
Additional characterization of platelet-derived FV in this study led to the identification of specific residues in the FVa light chain modified with N-linked carbohydrate structures. By using the published carbohydrate profile of Kumar et al. (48), three carbohydrate structures at Asn 1982 were assigned, as well as the specific N-linked glycoform present at Asn 2181 . This latter modification has important functional consequences and causes a decrease in the phospholipid binding affinity of FVa (51). Additionally, the consistent observation that Tyr 665 , Tyr 696 , and Tyr 698 were not sulfated in both platelet-and plasma-derived FVa is in contrast to observations made with recombinant FV (70). Tyrosine sulfation has also been observed in the heavy chain of FV endogenously produced by the liver cell line HepG2 (47). However, in both these studies the locations of the observed sulfate residues were hypothesized and not conclusively identified. Thus, the actual sites of sulfation may not have been observed in this study since only 40% of the heavy chain sequence could be identified by the mass spectroscopic analyses. In addition, the presence of subsets of FVa that do not ionize well in such analyses cannot be ruled out.
In summary, the characterization of purified, platelet-derived FV/Va confirms and extends previous studies done with platelet releasates and indicates that the platelet-derived cofactor possesses unique physical and functional properties that distinguish it from plasma-derived FV. The production of platelet-derived FV/Va requires addition of at least one O-linked sugar residue and proteolytic processing to express a unique light chain, to form a pool of a partially activated, mature platelet-derived cofactor. Additional events must occur that prevent its phosphorylation at Ser 692 even though the residue is unmodified. These cumulative observations document the modification of an endocytosed plasma protein that may alter its functional characteristics, allowing it to sustain procoagulant events subsequent to its release by platelets deposited at the site of injury.