C-terminal Residues 621–635 of Protein S Are Essential for Binding to Factor Va*

Protein S is anticoagulant in the absence of activated protein C because of direct interactions with coagulation Factors Xa and Va. Synthetic peptides corresponding to amino acid sequences of protein S were tested for their ability to inhibit prothrombinase activity. The peptide containing the C-terminal sequence of protein S, residues 621–635 (PSP14), reversibly inhibited prothrombinase activity in the presence but not in the absence of Factor Va (K i ∼2 μm). PSP14 inhibition of prothrombinase was independent of phospholipids but could be competitively overcome by increasing Factor Xa concentrations, suggesting that the C-terminal region of protein S may compete for a Factor Xa binding site on Factor Va. Studies using peptides with amino acid substitutions suggested that lysines 630, 631, and 633 were critical residues. PSP14 inhibited Factor Va activity in Factor Xa-one-stage clotting assays. PSP14 inhibited protein S binding to immobilized Factor Va. When preincubated with protein S, antibodies raised against PSP14 inhibited binding of protein S to Factor Va and blocked inhibition of prothrombinase activity by protein S. These results show that the C-terminal region of protein S containing residues 621–635 is essential for binding of protein S to Factor Va and that this interaction contributes to anticoagulant action.

Protein S is anticoagulant in the absence of activated protein C because of direct interactions with coagulation Factors Xa and Va. Synthetic peptides corresponding to amino acid sequences of protein S were tested for their ability to inhibit prothrombinase activity. The peptide containing the C-terminal sequence of protein S, residues 621-635 (PSP14), reversibly inhibited prothrombinase activity in the presence but not in the absence of Factor Va (K i ϳ2 M). PSP14 inhibition of prothrombinase was independent of phospholipids but could be competitively overcome by increasing Factor Xa concentrations, suggesting that the C-terminal region of protein S may compete for a Factor Xa binding site on Factor Va. Studies using peptides with amino acid substitutions suggested that lysines 630, 631, and 633 were critical residues. PSP14 inhibited Factor Va activity in Factor Xa-one-stage clotting assays. PSP14 inhibited protein S binding to immobilized Factor Va. When preincubated with protein S, antibodies raised against PSP14 inhibited binding of protein S to Factor Va and blocked inhibition of prothrombinase activity by protein S. These results show that the C-terminal region of protein S containing residues 621-635 is essential for binding of protein S to Factor Va and that this interaction contributes to anticoagulant action.
Protein S is a vitamin K-dependent protein that can act as a cofactor for the anticoagulant functions of activated protein C (1)(2)(3). Heterozygous deficiency of protein S or protein C is associated with increased risks for both venous thrombosis (4 -8) and ischemic stroke (9 -11). Homozygous deficiency of protein S, like that of protein C (12,13), can lead to potentially fatal purpura fulminans in infancy unless aggressively treated (14). The physiological importance of protein S is underscored by the finding that infusions of protein S concentrates without activated protein C showed antithrombotic activity in a rabbit thrombosis model (15). Modes of antithrombotic action of protein S are, however, incompletely understood. Protein S can be anticoagulant even in the absence of activated protein C by inhibiting prothrombinase activity of the FVa-FXa complex in fluid phase and on phospholipid vesicles, endothelial cells, or platelets (16 -18). This inhibition presumably involves binding of protein S to Factor Va (FVa) 1 and Factor Xa (FXa) as well as binding to the lipid membrane (16 -20). Protein S binds FVa and FXa with apparent K d values of 33 and 18 nM, respectively (16,17). In addition, protein S inhibits thrombin generation when subendothelial matrix is exposed to plasma under flow conditions (19). The relative anticoagulant potency of protein S has been correlated with its affinity for phospholipid vesicles, and protein S competes with prothrombinase components particularly at limiting phospholipid concentrations, e.g. 0.2 M (20). An assay has been developed for the activated protein C-independent activity of protein S in individual plasma samples (21). FV in combination with protein S enhances activated protein C anticoagulant activity (22), and it was suggested that this complex phenomenon may involve binding of the sex hormone binding globulin region of protein S residues 270 -635 to FV (23). Studies presented here identify an essential binding site for FVa at the C terminus of protein S that contributes to the ability of protein S to inhibit prothrombinase activity.

EXPERIMENTAL PROCEDURES
Proteins and Peptides-Human protein S, FVa, prothrombin, monoclonal antibody S7 to protein S, and phospholipid vesicles (80% phosphatidylcholine and 20% phosphatidylserine) were prepared as described (16,17,24). Antibodies to PSP14 were prepared and immunoaffinity purified using protein S-Sepharose as described for anti-protein C peptide antibodies (24). Peptides were prepared under the supervision of Dr. Richard Houghten of Torrey Pines Institute using the simultaneous multiple synthesis method (25), were in the C-terminal amide form, and were analyzed by reverse-phase high pressure liquid chromatography and mass spectral analysis to verify purity and composition (24). Unless otherwise specified, peptides were pretreated with excess iodoacetamide before use to block any free sulfhydryl groups that might be present. Controls showed that iodoacetamide did not affect the various assays.
Prothrombinase Assays-Prothrombinase activity assays were described in detail elsewhere (16,24) and were performed in the same binding buffer as described under "Binding Assays" below except that bovine serum albumin was substituted for gelatin and NaCl was 0.
Solving Equations 3 and 5 for FVa free yields which can be rearranged to Thus, plots of 1/v versus the inhibitor concentration I should yield straight lines for each FXa concentration chosen, the intercepts of which cross each other at [PSP14] ϭ -K i (see also "Results" and Fig. 4).
Binding Assays-Binding assays were described in detail elsewhere (16). Detection of bound protein S was with murine monoclonal antibody S7, and detection of bound FVa was with rabbit antibodies (purified IgG from Dako, Carpenteria, CA) to FV. Bound antibodies in turn were detected with biotin-secondary antibody, steptavidin-alkaline phosphatase, and phosphatase substrate as described (16). Noncoated wells served as nonspecific controls, the values of which were subtracted from corresponding coated wells. Nonspecific binding ranged from 10 to 30% in various experiments. Binding assays were performed in microtiter plates using a buffer consisting of 0.05 M Tris, 0.2 M NaCl, 5 mM CaCl 2 , 0.1 mM MnCl 2 , 0.02% NaN 3 , and 0.5% porcine skin gelatin, pH 7.4 (binding buffer). Human FXa (Enzyme Research Labs, South Bend, IN) was biotin labeled as described (26). Binding of biotin-FXa to immobilized protein S was detected with streptavidin-alkaline phosphatase and phosphatase substrate as for other binding assays. Microtiter V-well plates were from Nunc (Naperville, IL), binding plates (Immulon-2) were from Dynatech (Chantilly, VA), and low binding plates for thrombin assays were from Stockton Scientific (Scottsdale, AZ).

Inhibition of Prothrombinase Activity by Peptide PSP14 -To
elucidate structural requirements for protein S interaction with FVa, 14 pentadecapeptides were synthesized, representing sequences in the protein S molecule. These peptides were tested for their ability to inhibit prothrombin activation by the complete prothrombinase complex, i.e. FXa, FVa, phospholipid vesicles, and calcium ions ( Table I). The strongest inhibition was observed with peptide PSP14, representing residues 621-635 (RAHSCPSVWKKTKNC, with the substitution of Cys for Ser 635 of the C terminus) of protein S. Mass spectral analysis of PSP14 revealed that it had the correct mass for a monomer with an intramolecular disulfide bond. Prothrombinase inhibition by PSP14 was dose-dependent (IC 50 ϭ ϳ28 M), and a control peptide with a scrambled sequence of PSP14 (CK-WTAKSNVRSHPSK) was significantly less inhibitory to prothrombinase activity (Fig. 1). The data suggested that peptide PSP14 at micromolar concentrations could mimic the independent anticoagulant activity of protein S (17). Maximal inhibition by PSP14 was observed after a 1-min preincubation of peptide with FVa, and inhibition was reversed within 1 min when a FVa-peptide PSP14 mixture was diluted 50-fold (data not shown). When the prothrombinase assay was performed in the absence of FVa, no inhibition by PSP14 was observed ( Fig.  1), suggesting that the peptide interacts with either FVa or with a FVa binding site on FXa or prothrombin. In control experiments, a PSP14 peptide with D-amino acids instead of L-amino acids inhibited prothrombinase by less than 10% at concentrations up to 100 M, showing that the anticoagulant action of PSP14 was stereospecific and not simply based on the amino acid composition of PSP14.
Effect of Varying Concentrations of Prothrombinase Components on Inhibition by Peptide PSP14 -Peptide PSP14 displayed a pattern of mixed inhibition of prothrombinase when both PSP14 and prothrombin concentrations were varied, and data were analyzed by a Lineweaver-Burk plot (Fig. 2). This is the same pattern of inhibition of prothrombinase seen for native protein S (16) and indicates that PSP14 does not simply compete with the substrate, prothrombin, but affects both the K m and V max of prothrombin activation. Peptide PSP14 inhibited prothrombinase activity also in the absence of phospholipids, and in the presence of phospholipids, the inhibition was insensitive to varying phospholipid concentrations from 25 to 150 M (data not shown). Inhibition of prothrombinase activity by PSP14 was relatively insensitive to varying concentrations of FVa from 0.02 to 1 nM (not shown), but inhibition was ablated by increasing the FXa concentration from 0.19 to 8 nM (Fig. 3). This suggests that peptide PSP14 competes with FXa for a binding site on FVa. To further test this idea, rates of prothrombin activation were determined with varying FXa and peptide PSP14 concentrations, and the results were analyzed in a model in which PSP14 and FXa compete for the same binding site on FVa as described in detail under "Experimental Procedures." Consistent with this model straight lines were obtained when 1/prothrombinase (FXaFVa) activity (i.e. 1/v) at varying concentrations of FXa was plotted versus inhibitor concentration (Fig. 4). From the point where the lines intersect with one another the K i was found to be 1.8 M. The peptide RAHSCPSVWKKTKNC Ͼ98 a Underlined residues are those not in the native sequence. E was substituted for ␥-carboxyglutamic acid, and S or G was substituted for some C residues. In many cases a C was added to the end of a sequence to facilitate coupling to keyhole limpet hemocyanin and production of antibodies.
solution used in Fig. 4 experiments was freshly prepared and therefore somewhat more active than the solutions used in experiments presented in other figures.
Essential Amino Acids in Peptide PSP14 -To identify residues essential for inhibition of prothrombinase by PSP14, peptide analogs were studied. Because the number of Cys residues varied, all peptides were reduced with dithiothreitol and alkylated using iodoacetamide to facilitate comparisons, except as noted (Table II). The linearized, reduced, and alkylated PSP14 inhibited prothrombinase activity by 25% at 400 M, suggesting that the structural constraints conferred by the intramolecular disulfide bond between residues 625 and 635 in PSP14 enhanced its activity because the cyclized PSP14 at 300 M gave Ͼ98% inhibition (Tables I and II and Figs. 1 and 2). Reduced and alkylated peptide PSP14A with the naturally occurring Ser rather than Cys at position 635 gave 55% inhibition at 400 M (Table II). Lys residues 630, 631, and 633 were essential for inhibitory activity because substitution of each one of these residues by Ala resulted in complete loss of inhibitory activity (PSP14B, PSP14C, and PSP14D, Table II). Substitution of Ala for Pro at 626 had no significant effect on inhibitory activity. The short sequence 628 -635 (PSP628) that contained the critical Lys residues 630, 631, and 633 in a linear peptide was not sufficient for significant inhibition of prothrombinase, although introduction of a disulfide bond to cyclize the peptide (PSP629) markedly enhanced its activity, i.e. giving 76% versus 6% inhibition (Table II). An unrelated basic pentadecapeptide (MQAYIDIKNCPKKTR) with three Lys and one Arg residues did not significantly inhibit prothrombinase (data not shown). Similarly, an unrelated pentadecapeptide (CGPLLICYKESVDQR) with a cyclic conformation did not significantly inhibit prothrombinase (data not shown), showing that neither a high content of basic residues nor an internal disulfide bond per se confer inhibitory activity.
Inhibition of Clotting Activity by PSP14 -When PSP14 was preincubated with FVa and the FVa was subsequently tested in FXa-one-stage clotting assays using FV-deficient plasma, the clotting time (89 s in absence of peptide) was markedly prolonged (Ͼ999 s at 200 M PSP14). Using FVa standards to calculate the percentage of FVa activity, 50% inhibition was observed at a final concentration of ϳ9 M peptide (Fig. 5). The scrambled sequence control peptide (PSP14SCR) had little effect under the same conditions. PSP14 was also a potent inhibitor of the clotting of plasma in APTT tests. As little as 12.5 M PSP14 resulted in more than a doubling of the APTT, and the clotting time at 200 M final peptide concentration was Ͼ999 s as compared with a base-line clotting time of 45-58 s (data not shown). This effect was seen in normal plasma, protein Cdepleted plasma, and protein S-depleted plasma, demonstrating that the effect of peptide was not dependent on endogenous protein C or protein S in the plasma.
Inhibition of Binding of Protein S to FVa by Synthetic Peptides-Each synthetic protein S peptide was tested for its ability to inhibit binding of protein S to immobilized FVa or vice versa. PSP14 was capable of inhibiting 95% of the binding of protein S to immobilized FVa, with IC 50 estimated to be Ͻ8 M peptide (Fig. 6). A trilysine peptide and peptides such as PSP13 (residues 603-616) were not inhibitory. In an alternatively configured binding assay, PSP14 also inhibited FVa binding to immobilized protein S (IC 50 ϳ20 M, data not shown). In this same binding assay, reduction and alkylation of PSP14 increased its IC 50 to 200 M, whereas the control peptide with a scrambled sequence of PSP14 (PSP14SCR) did not exhibit significant inhibition of binding (data not shown). The linear PSP14 analog with Ser rather than Cys at position 635 (PSP14A), which was primarily in dimeric form by mass spectral analysis, displayed inhibition of binding of protein S to FVa similar to that of reduced and alkylated PSP14.
Effects of Antibody to Peptide PSP14 -Rabbit antibodies were prepared against a conjugate of PSP14 and keyhole limpet hemocyanin, and the antibodies were immunoaffinity purified on a column of protein S coupled to Sepharose. The purified antibodies effectively inhibited protein S binding to immobilized FVa (Fig. 7). Nonimmune rabbit IgG or purified antibodies to another protein S peptide, PSP11 (residues 408 -422), did not inhibit binding of protein S to FVa.
When protein S was preincubated with purified anti-PSP14 IgG prior to preincubation with FVa, the inhibition of prothrombin activation by protein S was effectively blocked (Fig.  8). These results suggest that residues 621-635 of protein S are essential for its inhibition of FVa cofactor activity in the pro- CWKKTKNC 76 a Peptides at 400 M were preincubated with FVa for 15 min in binding buffer prior to addition of other prothrombinase components and measurement of the rate of thrombin formation as described under "Experimental Procedures." Percentage of inhibition is relative to a control with no peptide, and each value is the average of two or more determinations.
b Reduced and alkylated peptides were treated for 30 min with a 1.5 times molar excess of dithiothreitol (except for PSP14, a 3 times molar excess) and then for 30 min with iodoacetamide at 2 times molar excess over dithiothreitol. Cyclized (oxidized) peptides that contained disulfide bonds that spontaneously formed were treated with a 1.5 times molar excess of iodoacetamide. Reagent controls were performed.
c An underlined amino acid denotes a substitution for the naturally occurring amino acid.

FIG. 5. Inhibition of FVa clotting activity by peptide PSP14.
FXa one-stage clotting assays were performed as follows: 40 l of FV-deficient plasma was preincubated for 1 min at 37°C with 40 l of cephalin (rabbit brain, Sigma). Clotting was initiated with a mixture of FXa (5 l of 13 nM), 50 l of 0.1 nM FVa that had been preincubated 20 min with or without peptide PSP14 in binding buffer with 0.5% bovine serum albumin (37°C), and 25 l of 50 mM CaCl 2 (37°C). The clotting time was determined in a Stago ST4 coagulometer. Clot times for standards with varying FVa concentrations were compared with FVa mixtures preincubated with peptide PSP14 or to a control peptide with the scrambled sequence of PSP14 (PSP14 SCR) to determine the percentage of FVa activity remaining.
FIG. 6. Inhibition of protein S binding to immobilized FVa by various peptides. FVa (7.6 nM) was coated to the wells of microtiter plates, the wells were blocked, and peptides at concentrations indicated were preincubated in duplicate for 30 min in the wells prior to addition of protein S (final concentration, 53 nM) and further incubation for 60 min. After washing, bound protein S was detected (see "Experimental Procedures"). Wells not coated with FVa served as controls for nonspecific binding, and values for specific binding were obtained by subtracting these control values from paired FVa-coated wells. KKK denotes a trilysine peptide, and PSP13 represents a protein S peptide with the sequence of residues 603-616. thrombinase complex. Control experiments showed that anti-PSP14 IgG and nonimmune rabbit IgG had no effect on prothrombinase activity in the absence of protein S and that nonimmune rabbit IgG in the presence of protein S did not block protein S anticoagulant activity.
Tests for Interactions of PSP14 with Proteins other than FVa-Besides interacting with FVa, protein S binds to FXa (17,18), C4b-binding protein (27), and activated protein C (17,28). To determine whether residues 621-635 of protein S might play a role in any of these other interactions, several tests were performed. Biotin-FXa was prepared and used to monitor its binding to immobilized protein S in the presence of possible inhibitors. Binding of biotin-FXa to protein S was inhibited by 73% by adding a 50-fold molar excess of unlabeled FXa but not by adding 200 M PSP14 or 560 nM anti-PSP14 antibodies (data not shown). Likewise, when studies of protein S binding to C4b-binding protein were performed as described previously (29,30), PSP14 did not interfere with binding of biotin-protein S to immobilized C4b-binding protein or with binding of biotin-C4b-binding protein to immobilized protein S (data not shown). PSP14 did not enhance or inhibit the anticoagulant activity of activated protein C in either FXa one-stage or APTT clotting assays (data not shown). These data suggest that residues 621-635 of protein S are not essential for normal interactions with FXa, C4b-binding protein, or activated protein C. DISCUSSION A novel mechanism of anticoagulant action of protein S involves its ability to function independently of activated protein C, via its binding to FVa, FXa, and/or phospholipids, thereby causing inhibition of prothrombinase activity (16 -21). This report explores the interaction of protein S with FVa and identifies an essential binding site for FVa on protein S involving the region of residues 621-635. Peptide PSP14 representing this region was a potent inhibitor of FVa binding to protein S and of FVa-dependent prothrombin activation and was anticoagulant in clotting assays. The inhibitory effect of peptide PSP14 was only observed in the presence of FVa, but the inhibition was not dependent on the presence or concentration of phospholipids. However, increasing FXa concentrations from 0.2 to 8 nM negated the inhibitory activity of PSP14, suggesting that PSP14 may compete with FXa for a binding site on FVa. This hypothesis was supported by a plot of 1/v versus [PSP14] at varying [FXa], which revealed the pattern for a competitive inhibitor with an apparent K i of ϳ2 M. Competition between FXa and protein S for binding to FVa has not been directly demonstrated but is consistent with the report that a FVa 15-mer peptide binds both protein S and FXa (31,32) and with reports that protein S negates the ability of FXa to protect FVa from proteolytic inactivation by activated protein C (33,34).
The inhibitory potency of peptide PSP14 was enhanced by an internal disulfide bond introduced by the substitution of Cys for Ser 635 , because reduced and alkylated PSP14 and a peptide analog with Ser in position 635 were less potent than the cyclized peptide. Single amino acid substitutions in PSP14 analogs showed that Lys 630, 631, and 633 were each critical for prothrombinase inhibition. Although a short peptide with the linear sequence of residues 628 -635 containing these Lys residues was not a potent inhibitor, cyclization of the peptide by addition of two Cys residues at the end of the peptide rendered it potently anticoagulant (Table II). Because the degrees of freedom that determine the relative populations of various conformations for a given peptide are greatly reduced in a cyclic peptide, it appears that some of the conformations of cyclized PSP14 and PSP629 may resemble those of native protein S.
The anticoagulant and binding inhibitory effects of PSP14 were not due simply to its net positive charge per se, because a trilysine peptide, peptides with the same scrambled L-amino acid sequence (PSP14SCR) or with the same sequence of Damino acids and unrelated pentadecapeptides with four basic residues had little or no ability to inhibit binding of protein S to FVa or to inhibit prothrombinase or clotting assays.
Other evidence for specificity of the inhibitory effects of peptide PSP14 was presented in several forms. Antibodies to PSP14 blocked protein S binding to FVa and neutralized the prothrombinase inhibitory activity of protein S. Neither PSP14 nor anti-PSP14 antibodies could block binding of biotin-FXa to protein S. Moreover, the ability of PSP14 to inhibit prothrombinase was entirely dependent on the presence of FVa. The fact that PSP14 was anticoagulant in protein C-depleted plasma is FIG. 7. Inhibition of binding of protein S to immobilized Factor Va by anti-PSP14 antibodies. Protein S was preincubated 1 h with and without various concentrations of immunopurified anti-peptide antibodies against PSP14 or PSP11 (representing protein S residues 408 -422) or with nonimmune rabbit IgG in the wells of a microtiter plate. The preincubation mixtures were transferred in duplicates of 50 l each to wells of a microtiter plate coated with 7.6 nM FVa. Following incubation at 37°C for 1 h, the wells were washed, and bound protein S was detected.
FIG. 8. Neutralization of activated protein C-independent anticoagulant activity of protein S by anti-PSP14 IgG. Protein S, anti-PSP14 IgG, protein S plus anti-PSP14 IgG, or buffer alone (Control) were preincubated in v-shaped well plates for 30 min at room temperature. Then FVa and phospholipid vesicles were added, and preincubation was continued for 30 min. FXa and prothrombin were added, aliquots were taken over time, and thrombin formed was measured (see "Experimental Procedures"). The final concentration of protein S was 67 nM, and the final concentration of IgG was 780 nM. consistent with the concepts that protein S is anticoagulant in the absence of protein C and that the region of residues 621-635 is involved in protein S-FVa interactions.
Residues 621-635 are not necessary for complexation of protein S with C4b-binding protein because peptide PSP14 did not inhibit binding of protein S to C4b-binding protein. Sites for interaction of protein S with C4b-binding protein have been reported in the steroid hormone binding globulin domain (35) of protein S at residues 413-433 (29,30) and 447-460 (36,37). Peptides overlapping each of these regions in protein S did not inhibit prothrombinase activity (Table I), indicating that this experimental approach fails to implicate these sequences as essential for sites of interaction of protein S with FVa or FXa. The region in the FVa molecule with which residues 621-635 of protein S should interact remains a matter of speculation. A stretch of 14 amino acid residues in FVa (residues 493-506, GLLLICKSRSLDRR) potently inhibits prothrombinase activity and has been implicated in the binding of FVa to protein S, FXa, DEGR-FXa, and DIP-activated protein C but not to activated protein C or prothrombin (31,32,38). It seems unlikely that this sequence in FVa interacts with residues 620 -635 in protein S because of the likely electrostatic repulsion of the net positive charges of these sequences (ϩ3 and ϩ4, respectively). However, each peptide also contains a region of uncharged residues. It is likely that protein S and FVa have more than one site of molecular interaction. In the case of activated protein C interaction with FVa, three sites of molecular interaction have been suggested in the light and heavy chains of activated protein C, and two sites have been suggested in the light and heavy chains of FVa (24, 38 -41). For interactions between FXa and FVa, sites in both heavy and light chains of FVa are suggested (31,(42)(43)(44)(45)(46)(47) with specific implication of FVa residues 493-506 and 311-325 (31,45) and FXa residues 263-274 (46,47).
It is of interest to elucidate direct anticoagulant modes of action of protein S, such as its interaction with FVa or FXa in the down-regulation of thrombin formation. The present identification of a binding site for FVa involving protein S residues 620 -635 that does not bind FXa, protein C, or C4b-binding protein and is not dependent on the presence of phosholipids extends our understanding of the multiple anticoagulant mechanisms of protein S.