Binding of protein S to factor Va associated with inhibition of prothrombinase that is independent of activated protein C.

Since plasma protein S serves an anticoagulant function by mechanisms which are not completely understood, its possible interaction with Factor Va was investigated. Human protein S bound to immobilized human Factor Va in a calcium-dependent, saturable, and reversible manner and Factor Va bound similarly to immobilized protein S. Binding of protein S to immobilized Factor V was greatly enhanced by pretreatment of the surface-bound Factor V with increasing doses of thrombin up to 1 unit/ml. Binding of protein S to Factor Va was also demonstrated in fluid phase with a Kd of 33 +/- 9 nM. Biotin-labeled heavy chain of Factor Va bound to immobilized protein S, and this binding was reversed by a 17-fold molar excess of intact unlabeled Factor Va. Protein S competed efficiently with prothrombin for binding to immobilized Factor Va. The prothrombinase activity in a reaction mixture of purified clotting factors was inhibited by protein S and exhibited a pattern of mixed inhibition. The concentration of protein S needed for 50% inhibition of the prothrombinase activity of a mixture containing 1 nM Factor Xa, 20 pM Factor Va, and 50 microM phospholipids was about 16 nM. Since not all protein S preparations exhibited this degree of prothrombinase inhibitory activity, extensive control experiments were performed to verify that the inhibitory activity was associated with protein S during immunoaffinity chromatography and was not caused by traces of activated protein C in the protein S preparations. These data show that protein S has an anticoagulant function which is independent of activated protein C and, at least in part, that this is because of its competition with prothrombin for direct binding to Factor Va.


Binding of Protein S to Factor Va Associated with Inhibition of Prothrombinase That Is Independent of Activated Protein C*
Mary J. Heeb, Rolf M. Mesters, Guido Tans$, J a n Rasing$, and John H. Griffins .

Maastricht, The Netherlands
Since plasma protein S serves an anticoagulant function by mechanisms which are not completely understood, its possible interaction with Factor Va was investigated. Human protein S bound to immobilized human Factor Va in a calcium-dependent, saturable, and reversible manner and Factor Va bound similarly to immobilized protein S. Binding of protein S to immobilized Factor V was greatly enhanced by pretreatment of the surface-bound Factor V with increasing doses of thrombin up to 1 unitlml. Binding of protein S to Factor Va was also demonstrated in fluid phase with a K d of 33 f 9 nM. Biotin-labeled heavy chain of Factor Va bound to immobilized protein S, and this binding was reversed by a 17-fold molar excess of intact unlabeled Factor Va. Protein S competed efficiently with prothrombin for binding to immobilized Factor Va. The prothrombinase activity in a reaction mixture of purified clotting factors was inhibited by protein S and exhibited a pattern of mixed inhibition. The concentration of protein S needed for 50% inhibition of the prothrombinase activity of a mixture containing 1 nM Factor Xa, 20 PM Factor Va, and 50 N M phospholipids was about 16 nM. Since not all protein s preparations exhibited this degree of prothrombinase inhibitory activity, extensive control experiments were performed to verify that the inhibitory activity was associated with protein S during immunoaffinity chromatography and was not caused by traces of activated protein C in the protein S preparations. These data show that protein S has an anticoagulant function which is independent of activated protein C and, at least in part, that this is because of its competition with prothrombin for direct binding to Factor Va.

Protein S is a vitamin K-dependent anticoagulant protein
(1,2) which can serve as a nonenzymatic cofactor for activated protein C (APC)' in its inactivation of Factors Va and VIIIa (FVa and FVIIIa) (3, 4). Its biological importance is inferred * This study was supported in part by National Institutes of Health Research Grant HL-21544, by a fellowship (to R. M. M.) from the Deutsche Forschungsgemeinschaft, and by the Nederlandse Hartstichting. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The abbreviations used are: APC, activated protein C; F, Factor; MES, 4-morpholineethanesulfonic acid; p-APMSF, para-amidinophenylmethylsulfonyl fluoride; TBS, Tris-buffered saline. from the association of recurrent venous thrombosis with hereditary deficiency of protein S (5-8). The mechanism of action of protein S is not completely understood. Evidence has been presented that protein S increases the affinity of APC for phospholipid vesicles (9), endothelial cell surfaces (lo), platelets (11, 12), and platelet microparticles (13). It has been proposed that protein S localizes APC to a cell surface on which APC can inactivate FVa in the prothrombinase complex or FVIIIa in the "tenase" complex. One report described an APC-independent anticoagulant activity of a protein S preparation made using a unique monoclonal antibody coupled to Sepharose (14). Using an independent approach, we were unable to demonstrate that protein S increased the binding of APC to immobilized vesicles composed of various mixtures of phosphatidylserine and phosphatidylcholine.* Moreover, protein S deficiency but not protein C deficiency is associated with arterial as well as venous thrombosis in some patients (15), suggesting that the modes of anticoagulant activity of protein S may be more complex than those of APC. Therefore, we investigated whether protein S might directly interact with FVa and thereby exhibit anticoagulant activity independent of APC.

MATERIALS AND METHODS
Proteins-Protein S was purified by slight variations of published methods (16). Briefly, the steps for protein S purification were: 1) barium adsorption of 1-day-old citrated human plasma containing 10 mM benzamidine, 1 mM diisopropylfluorophosphate, 1 mM PMSF, 100 units/ml aprotinin, and 0.02% NaN3; 2) elution of the barium citrate pellet with EDTA 3) chromatography on DEAE-Sephacel (Pharmacia LKB Biotechnology Inc.) in 5 mM MES, 0.15 M NaCl, 1 mM benzamidine, 5 mM EDTA, 0.02% NaN3, pH 6.0, with a linear gradient of 0-0.4 M NaCl in the same buffer; 4) chromatography on Blue Sepharose (Pharmacia) in 0.05 M Tris-HC1, 0.05 M NaC1, 1 mM benzamidine, 2 mM EDTA, pH 7.4, with a linear gradient of 0-0.4 M NaCl in the same buffer; 5) chromatography on heparin-Sepharose (Pharmacia) in 0.05 M MES, 2 mM CaC12, 1 mM benzamidine, 0.01% Tween 20, 0.02% NaN3, pH 6.0, with a linear gradient of 0-0.2 M NaCl in the same buffer. The yield for the two preparations used in most of the experiments was about 46% of the free protein S or 19% of the total protein S in the starting plasma. Following purification, protein S was dialyzed against TBS and stored in aliquots at -70 "C. The two protein S preparations used had APC cofactor activities varying by no more than 20% when compared with each other or with commercially available protein S (Enzyme Research Laboratories) as judged by their prolongation of a FXa one-stage clotting time in protein S-depleted plasma with APC (17). Immediately preceding binding assays, protein S was pretreated for 20 min with 400 PM p-APMSF (Chemicon, Temecula, CA). Prothrombin (1) and FV (18) were purified as described previously, and protein C (19) was purified with 20 nM thrombin in 50 mM Hepes, 175 mM NaCI, pH 7.5, for 30 and activated as described previously. FV (0.2 mg/ml) was activated min at 37 "C. The thrombin was subsequently neutralized by adding  Goat polyclonal antibodies to protein S and to prothrombin were prepared as described previously (5). Anti-protein S antibodies were affinity purified using a column of protein S coupled to cyanogen bromide-activated Sepharose (Pharmacia) according to the manufacturer's instructions. Anti-prothrombin antibodies were adsorbed using the same column to remove any anti-protein S antibodies. The IgG fraction of rabbit polyclonal antibodies to FV was obtained from Dako, Carpinteria, CA. Non-calcium-dependent murine monoclonal antibodies (S7 to protein S, and C3 to protein C) were prepared as described for protein C monoclonal antibodies (20) and purified as described (21). Monoclonal antibody S7 had at least 1,000-fold higher affinity for protein S than for protein C, FIX, FX, or pr~thrombin.~ Ascites fluid containing monoclonal antibody to the heavy chain of FV was kindly donated by Drs. Dario Altieri and Tom Edgington, Scripps Clinic, La Jolla, CA, and this antibody was purified as described (21). A sample of rabbit anti-tissue factor pathway inhibitor antiserum and a positive control were generously provided by Drs. Bonnie Warn-Cramer and Samuel Rapaport of University of California at San Diego. Antibodies to a,-antitrypsin and antithrombin 111 were obtained from Behring, Marburg, Germany; antibody to heparin cofactor I1 was obtained from Diagnostica Stago, Asnieres, France; and antibody to human serum amyloid P component was from Atlantic Antibodies, Stillwater, MN.
Binding Assays-FV or FVa (2.5 pg/ml) was coated to the wells of Nunc (Naperville, IL) Maxisorp microtiter plates in 50 pl/well in 0.1 M sodium carbonate, pH 9.0, for 1 h at 37 'C in a humidified chamber. The plates were washed once with Tris-buffered saline, 0.02% NaN3, pH 7.4 (TBS), and blocked for 1 h at room temperature with 0.5% porcine skin gelatin (Sigma) in TBS. Control wells with no FVa were also blocked. In some cases, FV on the plate was treated for 20 min at room temperature with thrombin in blocking solution containing 5 mM CaC12 followed by four washes with the same buffer. The plates were incubated for 50 min at 37 "C with 50 pl of various concentrations of protein S in binding buffer consisting of 0.5% gelatin, 0.05 M Tris, 0.2 M NaCl, 5 mM CaC12, 0.1 mM MnC12, 0.02% NaN3, pH 7.4. The plates were washed three times with wash buffer, which was identical to binding buffer except that the gelatin content was 0.1%. The plates were incubated sequentially at room temperature with 50 &l of each of the following: 10 pg/ml antibody in wash buffer, 50 min; biotin-secondary antibody (Pierce Chemical Co.) diluted 1:1,000 in wash buffer, 50 min; and streptavidin-alkaline phosphatase conjugate (Pierce) diluted 1:500 in wash buffer, 30 min. Each step was followed by three washes, except the last step, which was followed by five washes. The plates were developed with a p-nitrophenyl phosphate substrate kit (Bio-Rad). The initial change in absorbance at 405 nm was recorded over a 5-10-min interval on a Bio-Tek (Winooski, VT) microtiter plate reader using a Kineti-calc program. The reverse assay was performed in a similar manner except that protein S was coated to the plate at 5 pg/ml, and FVa was used in the binding step. For some experiments, FVa heavy chain (22) was biotin labeled (23), and bound biotin-FVa heavy chain was detected, omitting the antibody steps. Binding of prothrombin to FVa was monitored in a manner similar to the binding of protein S, except that detection employed 20 pg/ml IgG fraction of goat anti-prothrombin which had been adsorbed on a column of protein S-Sepharose.
For determination of the dissociation constant ( K d ) for binding of protein S to FVa in the fluid phase, a method recently applied to determine the Kd for protein S binding to C4b-binding protein was used (24,25). Protein S at various concentrations was incubated with a fixed concentration of FVa at 37 "C for 2 h in the wells of a gelatinblocked microtiter plate (Stockwell Scientific, Walnut, CA). Standards of Va at the fixed concentration and at several lower concentrations were incubated in a similar manner. Free FVa in solution was determined by transferring the mixtures to a protein S-coated plate and incubating at room temperature for brief periods of 10, 15, or 20 min. FVa bound to the protein S-coated plate was detected as described above. Free FVa in the fluid phase mixtures was calculated from a standard curve made with different concentrations of FVa standards without protein S present, and this curve was linear in the range used. Bound FVa in the mixtures was calculated by subtracting free FVa from the initial concentration of FVa in each mixture. It R. M. Mesters, M. J. Heeb, and J. H. Griffin, unpublished data. was then assumed as a first approximation that 1 mol of protein S was bound to 1 mol of FVa. The data were subjected to Scatchard and linear regression analysis to determine Kd and stoichiometry.
Prothrombinme Assay-Prothrombinase assays were performed in binding buffer containing bovine serum albumin (Calbiochem) in place of gelatin. A mixture of phospholipids, FXa, and FVa was preincubated for 15 min at room temperature with or without protein S in 135 pl in the wells of a low binding microtiter plate (Stockwell Scientific). Final concentrations for most experiments were 20 pM FVa, 5-100 nM protein S, 0.07-1.2 pM prothrombin, 50 p M phospholipids, and 1 nM FXa. Following the addition of 15 p1 of prothrombin, 20-4 aliquots were removed at 1-min intervals and quenched in 80 pl of 0.05 M Tris, 0.1 M NaCI, 10 mM EDTA, 0.02% NaN3, 0.1% bovine serum albumin, pH 8.2, in separate wells. After sample collection, thrombin formed was assessed by the addition of 50 p1 of 1 mM S-2238 substrate (Kabi, Stockholm, Sweden) to each well. The change in absorbance at 405 nm was recorded on a microplate reader. In some experiments, protein S was preincubated with 200 pg/ml C3 monoclonal antibody to protein C as well as with 400 p M p-APMSF to exclude the possibility of inhibition caused by traces of APC or other serine proteases. Phospholipid vesicles used in the assays were prepared by mixing 0.2 mol fraction of phosphatidylserine with 0.8 mol fraction of phosphatidylcholine in chloroform (Sigma), drying under nitrogen in a siliconized tube, and adding TBS to a phospholipid concentration of 1.28 mM. The mixture was then sonicated in a cooling ice bath at 30 watts for 6 X 30 s, with I-min intervals. The resultant vesicles were stored at 4 "C and used within 2 weeks. A more detailed report of the prothrombinase assay appears elsewhere (17).

RESULTS
Binding of Protein S to FV or FVa-The ability of purified protein S to bind to purified FV or FVa was tested using microtiter plates that contained immobilized protein. Protein S bound to immobilized FVa in a dose-dependent manner (Fig. 1, A and B ) . Protein S bound less well to FV than to FVa; however, binding to FV was increased to levels seen for FVa when FV on the plate was pretreated with increasing doses of thrombin up to 1 unit/ml (Fig. 1B). Pretreatment of FVa on the plate with APC resulted in significant loss of ability of FVa to support binding to protein S (Fig. 1A). In different experiments, 75-100% less binding was detected when the protein S binding step was performed in the presence of 10 mM EDTA, as compared with in the presence of divalent metal ions (Fig. lA), suggesting that the binding of protein S to FVa requires calcium ions. Nonspecific binding to control wells without immobilized FVa varied from 5-25% of the total binding in different experiments, and the signal for nonspecific binding was subtracted from values for corre-Protein S Addad, w/ml FIG. 1. Binding of protein S to immobilized FV or FVa. The wells of a microtiter plate were coated with FV or FVa as described under "Materials and Methods." Some wells were pretreated as indicated with 5 pg/ml APC, with 0.2 unit/ml thrombin (panel A), or with varying amounts of thrombin (panel B ) as described. Various concentrations of protein S as indicated were incubated in the wells, and then protein S binding was detected as described under "Materials and Methods," using affinity-purified goat anti-protein S antibody. Each point is the average of duplicates. In one series of wells (panel A, closed triangles), 10 mM EDTA was included in the protein S binding step, but not in the detection steps.

Protein S Binds Factor Vu and
Inhibits Prothrombinase sponding wells coated with FVa. Similar results were obtained when bound protein S was detected with polyclonal or monoclonal antibody.
Binding of FVa to immobilized protein S was also demonstrated as seen in Fig. 2 A . The FVa was detected with either polyclonal antibodies or monoclonal antibody to FVa heavy chain, although the color intensity observed using a monoclonal antibody was approximately %s of that observed using the polyclonal antibodies. EDTA at 10 mM reduced the binding of FVa by 50-75%, showing that this binding required calcium ions (data not shown). Fig. 2B demonstrates that biotinylated FVa heavy chain also bound to immobilized protein S and that this binding was time-dependent, saturable, and reversed by the addition of a 17-fold molar excess of unlabeled FVa.
In ligand blotting experiments, biotinylated FVa heavy chain ligand blotted to nonreduced protein S which had been electrophoresed on a denaturing SDS-polyacrylamide gel and transferred to an Immobilon membrane (Millipore, Marlborough, MA) (data not shown). No band with an apparent M , other than that of protein S (75,000) was detected by this ligand blotting, suggesting that the FVa binding observed using microtiter plates was to intact protein S and not to any contaminant or to a proteolytic fragment of protein S.
To show that the data for the binding of protein S to FVa observed using immobilized protein S or immobilized FVa were not artifacts of the microtiter plate methods, the binding of protein S to FVa in the fluid phase (24,25) was determined as described under "Materials and Methods" (Fig. 3). The binding of protein S to FVa was saturable, and the K d was 33 4 9 (mean +standard deviation for five experiments), using a fixed concentration of 3-12 nM FVa in various experiments. The intercept on the abscissa of the Scatchard plot as shown in the insets was consistent with a stoichiometry of 1.0-1.3 mol of protein S/mol of FVa in various experiments. The data from Fig. 3, A and B, were also fit to hyperbolas using the Enzfitter program (not shown). By this method, the respective calculated K d values were 26 and 50 nM, and the ratios of protein S to FVa were 1.1 and 1.4. Similar K d values were obtained whether the 2-h incubation mixtures were incubated in the protein S-coated detection plate for 10, 15, or 20 min, suggesting that equilibrium was not significantly disturbed during incubation on the detection plate.
Competition for Prothrombin Binding to FVa by Protein nm. mln

FIG. 2. Binding of FVa or biotin-FVa heavy chain to immobilized protein S . In panel A, various concentrations of protein
S were coated to the wells of a microtiter plate. Protein S coated to the plate was detected directly in some wells using monoclonal antibody to protein S (triangles). In other wells, 2 pg/ml FVa was incubated, and bound FVa was detected as described under "Materials and Methods," using a polyclonal antibody (closed circles) or a monoclonal antibody (squares). In panel B, 5 pg/ml (66 nM) protein S was coated to the wells of a microtiter plate. Biotin-FVa heavy chain (1 pg/ml, 9 nM) was incubated in the wells for various times as indicated on the abscissa (closed circles). For some wells, unlabeled FVa (50 gg/ ml, 150 nM) was added at 15, 30, or 45 min of incubation and the incubation continued until 60 min (open symbols). Biotin-FVa heavy chain was detected as described under "Materials and Methods."  and open circles), and the incubation was continued until 60 min. Prothrombin binding was detected as described under "Materials and Methods." Closed circles indicate incubations of prothrombin alone for the indicated times. In panel B, protein S in binding buffer at several concentrations as indicated or binding buffer alone was preincubated in FVa-coated wells for 1 h at room temperature. One set of wells (open circles) was washed, and protein S was allowed to remain in another set of wells (closed circles) prior to the addition of prothrombin to a final concentration of 15 pg/ml and further incubation at 37 "C for 60 min. Bound prothrombin was detected as described under "Materials and Methods." S-Experiments were performed to test whether protein S competes with prothrombin for binding to immobilized FVa. When 15 pg/ml (214 nM) prothrombin was incubated in the wells of FVa-coated microtiter plates and bound prothrombin was measured at various times, 3 pg/ml (39 nM) protein s that was added at 15, 30, or 45 min after the addition of prothrombin abolished all prothrombin binding to FVa by the end of 60-min incubation (Fig. 4-4, dashed lines). In other experiments, wells containing immobilized FVa were preincubated with various concentrations of protein S for 1 h followed by the addition of 15 pg/ml prothrombin; after 1 additional h, bound prothrombin was determined (Fig. 4B). Prothrombin binding to FVa was reduced by preincubation of FVa with 1 or 2 pg/ml protein S, and it was completely abolished by preincubation with 3 pg/ml protein S (Fig. 4B). This effect was because of competition of protein S and prothrombin for binding to FVa, rather than an interaction of protein S with prothrombin in fluid phase since results were similar whether the protein S preincubated with FVa was washed from the wells (Fig. 4B, open circles) or left in the wells (Fig. 4B, closed circles) when prothrombin was added.
Control experiments in reverse of the above showed that prothrombin competed weakly with protein S for binding to FVa. Preincubation of immobilized FVa with 20 pg/ml prothrombin resulted in a 17-25% decrease in the protein S bound to FVa when protein S was subsequently incubated at 2.5-5.0 pg/ml in FVa-coated wells (data not shown). Diisopropyl-FXa and FXa competed modestly with protein S for binding to FVa, since preincubation of immobilized FVa with 20 pg/ml diisopropyl-FXa resulted in a 16-3895 decrease, and 20 pg/ml FXa resulted in a 29-48% decrease, in the protein S bound to FVa when protein S was subsequently incubated a t 5-10 pg/ml in FVa-coated wells (data not shown). The inhibition of prothrombinase activity by protein S was not affected when protein S was preincubated with p-APMSF and with antibody to protein C under conditions shown to inhibit APC completely at concentrations of 2 ng/ml to 4 pg/ ml. Thus, the inhibitory activity of protein S was not because of contamination with APC or other protease sensitive to p -APMSF. Furthermore, immunoblotting of the inhibited prothrombinase mixtures revealed no apparent proteolytic cleavage of FVa. arin cofactor I1 could be found (data not shown). Since the last three of these inhibitors are heparin-dependent, the effect of 1 unit/ml heparin in combination with protein S was assessed, but no stimulation of inhibition of prothrombinase activity was observed (data not shown). Thus, no obvious protease inhibitor contamination or heparin dependence was found for protein S preparations that inhibited prothrombinase activity.
To show directly that the prothrombinase inhibitor was protein S, a preparation of purified protein S was adsorbed using an immobilized monoclonal anti-protein S antibody. A 100-pg aliquot of protein S in TBS was chromatographed on a 2-ml column of monoclonal antibody S7-Sepharose. The eluted fractions were analyzed for their ability to inhibit prothrombinase and for the distinctive protein S doublet near M, = 75,000 on SDS-polyacrylamide gels (data not shown).
Both the prothrombinase inhibitory activity and the protein S antigen were retained on the column after washing with 10 ml of TBS, followed by 10 ml of TBS containing 10 mM EDTA. Inhibitory activity and protein S could be recovered when the column was subsequently eluted with 0.1 M glycine, pH 2.5, followed by neutralization of the fractions with 1 M Tris base. A control showed that a test aliquot of neutralized glycine did not inhibit prothrombinase. This experiment supports the hypothesis that protein S itself is responsible for direct inhibition of prothrombinase activity. Fig. 6 shows a silver stain of a number of different protein S preparations following SDS-polyacrylamide gel electrophoresis under nonreduced or reduced conditions. The relative potencies in APC-independent inhibition of prothrombinase and in APC cofactor activity are indicated below the figure.
Preparations H89-4C and H89-5 were used for all of the experiments described above, but preparation H89-4A was equally potent in inhibition of prothrombinase. Two different commercial preparations of protein S (C410 and C450) and one of our own preparations (H89-3A, B, and C) only weakly or moderately inhibited prothrombinase activity and bound

Protein S Binds Factor V u and
Inhibits Prothrombinase weakly to FVa, yet they were equally active as the protein S preparations described here when they were tested for APC cofactor activity in a FXa one-stage clotting assay. Preparations H89-5, H90, and C450 had the same NHn-terminal sequence of 5 amino acids. Preparations H89-4A and H89-5 contained a trace contaminant of mobility M, = 125,000-130,000. This contaminant was separated on a small scale trial chromatography from the protein S without loss of the protein S ability to inhibit prothrombinase or to bind to FVa (data not shown). We examined differences in purification procedures for the different protein S preparations. H89-5, H89-4A, and H89-4C were purified as described under "Materials and Methods," except that heparin-Sepharose chromatography was omitted for H89-4A and H89-4C and replaced with chromatography on monoclonal anti-protein C-Sepharose to remove traces of protein C. Preparation H89-3 began with frozen cryosupernatant of plasma, included an ammonium sulfate fractionation following barium elution, and omitted the heparin-Sepharose chromatography. Different pools A, B, and C of preparation H89-3 were collected based on the proportion of upper and lower band of the protein S doublet at M, = 75,OO-85,000 on reduced SDS-polyacrylamide gel electrophoresis (Fig. 6), but none of the pools was very inhibitory toward prothrombinase activity. Preparation H90 was partially purified in another laboratory and was not well studied because it had low APC cofactor activity and modest ability to inhibit prothrombinase. The commercial preparations were purified using ammonium sulfate fractionation following barium adsorption and included hydroxylapatite chromatography instead of heparin-Sepharose chromatography. No differences were noted in the procedures for preparation C410, which inhibited prothrombinase moderately, as compared with preparation C450, which inhibited weakly or not at all? A third commercial preparation (C570) purified with heparin-Sepharose in place of hydroxylapatite inhibited prothrombinase by 32% at 25 pg/ml (data not shown). Several other preparations made in our laboratory by the procedure given under "Materials and Methods" and by slight variations (such as using 3day-old plasma) inhibited prothrombinase by 50-75% at approximately 20 pg/ml (267 nM) and bound FVa, but less well than preparations H89-4A, H89-4C, and H89-5. Several treatments were tested in an attempt to convert a less active protein S preparation to a more active one, or vice versa.
These included treatment with thrombin, neuraminidase, 0.01% Tween 20, 0.1 M glycine, pH 2.5, f EDTA, and repeated freeze-thaws. None of these treatments was effective. Treatment with N-glycanase in SDS did not change the ability of protein S to bind biotin-FVa heavy chain on a ligand blot.
In summary, the less active protein S preparations were derived from older plasma or cryosupernatant or were prepared using hydroxylapatite chromatography, but we were unable to determine with certainty whether or how the various protein S preparations differed in molecular form or conformation or in the presence of a trace protein S-binding component. Nonetheless, the data suggest that some form of protein S binds directly to FVa and has anticoagulant functional activity that is independent of APC.

DISCUSSION
Evidence from clinical studies strongly demonstrates that protein S is an important natural antithrombotic factor. One study of 141 unrelated patients with venous thrombotic dis-M. Morris, Enzyme Research Laboratories, personal communication.
ease under the age of 45 revealed that 6% were deficient in protein S (27), and numerous other studies documented familial heterozygous deficiency of protein S associated with venous thrombosis (6,8,28,29) or arterial thrombosis (15). Yet the potency of protein S as a cofactor for APC in purified systems or systems employing platelets or platelet microparticles has been rather modest, typically increasing the rate of inactivation of FVa in the presence of phospholipid by about 2-fold (18, 30). The APC cofactor activity of protein S seems greater when FXa is present (30, 31). Therefore, we investigated whether protein S might have an alternate role by directly binding to FVa in a manner that directly inhibits the activity of the prothrombinase complex.
Here we report binding of protein S to immobilized FVa and vice versa. Binding of protein S was dependent on the presence of calcium ions and discriminated between FV and FVa. A solid phase binding assay was modified to quantitate the amount of free and bound FVa and protein S in various fluid phase mixtures of protein S and FVa, and a Kd of 33 nM was determined. Since the plasma concentration of free protein S is about 120 nM (32), this binding should occur under physiologic conditions when FVa is generated. Indeed, protein S at concentrations of 5-39 nM inhibited prothrombinase activity by 50% in different experiments. The pattern was that of mixed inhibition rather than that of classical competitive inhibition. However, in binding studies, protein S competed very effectively with prothrombin for binding to immobilized FVa. Protein S bound to biotinylated FVa heavy chain, and heavy chain of FVa is known to bind prothrombin (22). It remains possible that protein S may also bind to the light chain of FVa.
To determine whether the inhibitory activity of the protein S preparations may have been caused by traces of APC or another protease or known inhibitors of FXa or thrombin, control experiments were performed. Pretreatment of protein S with p-APMSF and antibody to protein C had no effect on the protein S anticoagulant activity. Immunoblotting of FVa preincubated with protein S revealed no detectable proteolytic cleavage of FVa. No heparin dependence was found for the inhibition by protein S, making the presence of the heparindependent inhibitors antithrombin 111, heparin cofactor 11, or tissue factor pathway inhibitor unlikely. Furthermore, antibodies to these inhibitors failed to detect their presence or the presence of cY1-antitrypsin or of serum amyloid P component (33) on immunoblots of protein S preparations. Biotinylated heavy chain of FVa ligand blotted to protein S at Mr = 75,000 but not to any other band of a different M, in the protein S preparations. The finding of a similar extent of inhibition of prothrombinase activity by protein S at a wide range of phospholipid concentrations indicates that competition of protein S for phospholipid binding of other prothrombinase components cannot explain the inhibitory activity of protein S. Since prothrombinase inhibitory activity was retained when protein S preparations were passed over a column of anti-protein S monoclonal antibody-Sepharose and then eluted after treatment with glycine at pH 2.5, the inhibitory activity is indeed associated with protein S rather than with an unrelated contaminant. We cannot exclude at this point that a trace component binds to protein S with high affinity and renders it anticoagulantly active. A protein S-binding protein has been reported in bovine plasma that enhances the APC cofactor activity of protein S (34), but such a protein has not yet been demonstrated to exist in human plasma. It remains to be determined whether 1) all protein S in plasma has the ability to inhibit prothrombinase but loses some of this activity during purification; or 2) protein S can be "acti-vated" in some way that increases its ability to inhibit prothrombinase; or 3) a subpopulation of protein S possesses a subtle difference in structure or conformation that enables it to bind FVa and inhibit prothrombinase.
FXa has the property of protecting FVa from inactivation by APC (35), and protein S negates this protective effect (30,31). Both FX and FXa apparently block an important binding site for APC (31). The finding here that protein S binds to FVa is consistent with those findings and suggests that protein S may negate the protective effect of FXa by displacing FXa or FX from its binding site on FVa, thereby making or keeping available a FVa site for APC binding. Our findings may be also related to another recent study, in which protein S purified on an immunoaffinity column that contained a unique monoclonal antibody had anticoagulant properties independent of APC. A novel conformational form of protein S, designated PSM, appeared to act as a competitive inhibitor of prothrombin in clotting assays (14). In a related study, Mitchell and Salem (36) showed that prothrombin could inhibit the APC cofactor activity of protein S. Our protein S preparations used in the studies reported here were not prepared using an immunoaffinity column yet displayed anticoagulant activity as great or greater than that described by Mitchell et al. (14). Thus, there exist molecular forms of protein S, as demonstrated here and previously by Salem's group (14), that directly bind to FVa and that have an anticoagulant activity which is independent of APC. Presumably this activity, at least in part, is caused by competition with prothrombin, but future studies will investigate possible interactions of protein S with FXa or thrombin. Since platelets contain protein S that is secreted upon platelet activation (37), the anticoagulant activity of protein S may be important in regulating coagulation reactions in the locus of activated platelets. If this is an important activity of protein S, its deficiency could contribute to risks of arterial thrombosis as recently described (15).