Human protein S cleavage and inactivation by coagulation factor Xa.

Human factor Xa specifically cleaves the anticoagulant protein S within the thrombin-sensitive domain. Amino-terminal amino acid sequencing of the heavy chain cleavage product indicates cleavage of protein S by factor Xa at Arg60, a site that is distinct from those utilized by alpha-thrombin. Cleavage by factor Xa is unaffected by the presence of hirudin and is completely blocked by tick-anticoagulant-peptide and D-Glu-Gly-Arg-chloromethyl ketone, the latter two being specific inhibitors of factor Xa. The cleavage requires the presence of phospholipid and Ca2+, and is markedly inhibited by the presence of factor Va. Factor Xa-cleaved protein S no longer possesses its activated protein C-dependent or -independent anticoagulant activity, as measured in a factor VIII-based activated partial thromboplastin time clot assay. The apparent binding constant for protein S binding to phospholipid (Kd approximately 4 nM +/- 1.0) is unaffected by factor Xa or thrombin cleavage, suggesting that the loss of anticoagulant activity resulting from cleavage is not primarily due to the loss of membrane binding ability. Cleavage and inactivation of protein S by factor Xa may be an additional way in which factor Xa exerts its procoagulant effect, after the initial stages of clot formation.

Protein S is a vitamin K-dependent, 635-amino acid, nonenzymatic glycoprotein (M r Х 78,000; Ref. 1) that acts as an anticoagulant in blood (2)(3)(4). The domain organization of protein S is similar to other plasma vitamin K-dependent proteins in that it contains an amino-terminal ␥-carboxyglutamate-rich domain and several (four) epidermal growth factor-like domains (5,6). Unique to protein S, however, is a 29-amino acid thrombin-sensitive domain, located between the ␥-carboxyglutamate and first epidermal growth factor-like domains. The thrombin-sensitive domain corresponds to exon IV of the protein S gene (7). Protein S also contains a sex steroid binding protein-like domain (8) at the carboxyl-terminal end of the protein, in place of the serine protease domain found in the vitamin K-dependent plasma proteases. The sex steroid bind-ing protein-like domain has been identified as being composed of two tandem repeat units homologous to the Cys-poor laminin A globular (G) domain found in a number of extracellular ligand binding basement membrane proteins involved in cellular growth and differentiation (9). Schneider and co-workers have described a cultured cell growth arrest-specific protein (Gas6) from NIH 3T3 mouse cells (10) and human IMR90 fibroblasts (11), which, except for the absence of a thrombinsensitive domain, is homologous throughout to protein S.
Thrombin cleavage converts protein S into a two-chain, disulfide-linked protein that no longer possesses anticoagulant activity (12). The thrombin cleavage sites were first established for bovine protein S (at Arg 52 and Arg 70 ) by Dahlback et al. (13) and recently at the corresponding residues (Arg 49 and Arg 70 ) for human protein S (1).
The defined mechanism(s) by which protein S exhibits its anticoagulant activity is presently unknown. Walker and coworkers were the first to show an acceleration by protein S of the APC 1 -mediated inactivation of factor Va (3, 14) and factor VIIIa (15). It is well established that factor Va inactivation by APC can be prevented by factor Xa in the bovine system and that protein S functions by abrogating this protection (16). Initially, Kalafatis and Mann in the bovine system (17) and, subsequently, Rosing et al. (18) and Egan et al. (19) in the human system have reported that protein S stimulates APCmediated factor Va inactivation by selectively promoting the cleavage of factor Va at Arg 306 . A kinetic study of factor Va inactivation by APC revealed that protein S increases the K cat of inactivation by 2-fold without changing the apparent K m for factor Va (16). Shen and Dahlback in a purified component system were the first to observe a synergistic effect of protein S and factor V on APC inactivation of factor VIIIa (20). This synergistic effect of protein S has been recently confirmed using human factor VIII (21,22).
Besides APC cofactor activity, protein S has also recently been reported by several groups to exhibit APC-independent anticoagulant activity. The first report of APC-independent protein S anticoagulant activity was that of Mitchell et al. (23), in which protein S appeared to act as a competitive inhibitor in prothrombin activation. Heeb et al. (24) have also reported the inhibition of the prothrombinase complex by protein S. They also observed direct, reversible binding of protein S to factor Va, which was competitive with prothrombin binding (24). In a follow-up study, Heeb et al. also observed the direct, phospholipid-independent reversible binding and slow inhibition of factor Xa by protein S (25).
Bouma's laboratory (26) has also reported the inhibition by protein S of prothrombinase activity on the surface of cultured human umbilical vein endothelial cells or unstimulated platelets. They have also observed inhibition by protein S of factor Xa generation by the "tenase" complex (purified factors X, IXa, and VIIIa; Ca 2ϩ ; and synthetic phospholipid or endothelial cells or platelets) (27). A correlation between protein S binding to synthetic phospholipid vesicles and inhibition of both the purified component prothrombinase and tenase complexes suggests that the APC-independent anticoagulant effect of protein S is at least in part due to its competition for phospholipid surface sites (28).
Despite the lack of a clear understanding of the mechanism by which protein S acts as an anticoagulant and its modest cofactor activity, the importance of in vivo protein S anticoagulant activity has been inferred from the correlation of inherited familial protein S deficiency and predisposition to recurrent venous thrombosis (29 -31). The common clinical manifestations of early age, recurrent superficial thrombophlebitis, deep vein thrombosis, and pulmonary embolism are similar to those associated with familial protein C deficiency, consistent with, but not proving, the believed cofactor role of protein S in the protein C anticoagulation pathway (32). The incidence of heterozygous protein S deficiency in patients with thrombotic disease has been estimated to be between 5% (29) and 8% (33), roughly comparable with those for protein C and antithrombin-III deficiency. In many cases, it is likely that other factors contribute to disease. For example, APC resistance resulting from Arg 506 3 Gln mutation in factor V greatly increases the risk of thrombosis in protein S-deficient individuals (34).
The procoagulant factor Xa, another member of the vitamin K-dependent family of plasma proteins, plays a crucial role in the amplification phase of the blood clotting cascade by enzymatically converting its target, prothrombin, to its active form, thrombin (for reviews, see Refs. [35][36][37]. Factor Xa also plays a pivotal role in blood clotting as a result of its generation by both the intrinsic (factors IXa and VIIIa) and the extrinsic (factor VIIa and tissue factor) pathways (35)(36)(37). We report here the specific cleavage and inactivation of human protein S by factor Xa. This cleavage and inactivation of protein S may represent a second important mechanism by which factor Xa exerts its procoagulant effect.

Materials
Plasma-derived HPS (pHPS) was purified as described elsewhere (16) or purchased from Enzyme Research Laboratories Inc. (South Bend, IN) and was judged by SDS-PAGE to be 15-20% cleaved. Purified human APC and factor Xa and dansylated Glu-Gly-Arg-chloromethyl ketone (DEGR) were generous gifts from Dr. Paul Haley, Hematologic Technologies, Inc. (Essex Junction, VT). Synthetic phospholipid vesicles composed of 75% L-palmitoyl-2-oleoyl phosphatidylcholine and 25% Lpalmitoyl 2-oleoyl phosphatidylserine (PCPS) and ␣-thrombin were prepared as described previously (38,39). Human factor V was purified and converted to the active form of the cofactor (Va) as described by Katzmann et al. (40). Human recombinant protein S (rHPS) was produced in human kidney 293 cells, purified, and chemically characterized as described elsewhere (1). Hirudin was obtained from Genentech (South San Francisco, CA). Tick anticoagulant peptide was a gift from Dr. Sriran Krishnaswamy (Emory University, Atlanta, GA). D-Glu-Gly-Arg-chloromethyl ketone and D-Phe-Pro-Arg-chloromethyl ketone (PPACK) were purchased from Calbiochem. Benzamidine was purchased from Aldrich. Human plasma-derived factor VIII (outdated Hemofil M, Baxter Healthcare Corp.) was a gift from Dr. Katherine High, Children's Hospital (Philadelphia, PA).

Time-dependent Protein S Cleavage by Factor Xa
Recombinant HPS (300 nM) was digested with factor Xa (6 nM) in the presence of 20 M PCPS in 20 mM HEPES (pH 7.4), 150 mM NaCl, 5 mM CaCl 2 , 20 nM hirudin for various times (0 -120 min) at 37°C. The samples were then mixed with 2% SDS, 2% ␤-mercaptoethanol and heated for 5 min at 90°C prior to electrophoresis on a 10% SDS-PAGE gel (41) and visualized by silver staining. For amino-terminal sequencing, the contents of the gel were electroblotted onto a polyvinylidene difluoride membrane and stained with Coomassie Brilliant Blue as previously detailed (17). The transferred protein S heavy chain 2-h digestion product (see Fig. 1) was submitted to conventional Edman degradation and product resolution on an Applied Biosystems 475A sequencer in the laboratory of Dr. Alex Kurosky (University of Texas, Medical Branch of Galveston).
To test the effect of factor Va, rHPS was also digested in the context of the "prothrombinase" complex (20 M PCPS, 6 nM factor Xa, and 20 nM factor Va). The possibility of cleavage caused by any contaminating thrombin was excluded by the presence of 20 nM hirudin.

Protein S Functional Assay
A modified three-stage assay was developed to measure functional protein S activity, in which the end point is a factor VIII(a)-dependent activated partial thromboplastin time clot time (42).
Stage 1-HPS (300 nM) was incubated with 30 nM factor Xa in the presence of 100 M PCPS in 20 mM HEPES, 150 mM NaCl, 5 mM CaCl 2 (pH 7.4) (HBS/Ca 2ϩ buffer), for 2 h at 37°C. Following incubation, DEGR was added to all samples (100 nM final concentration), and they were kept at room temperature for 30 min. The samples were then dialyzed twice in Slide-A-Lyzer cassettes (Pierce) versus HBS-Ca 2ϩ buffer at room temperature for 3 h to remove unreacted DEGR and tested in a factor Xa chromogenic assay for remaining active factor Xa and DEGR. Portions of the dialyzed samples were run on reduced SDS-PAGE gels to confirm complete cleavage (see Fig. 3B). Factor Xa was determined to be completely and irreversibly inhibited by DEGR, and unreacted DEGR was effectively removed by dialysis, (based upon amidolytic assays with synthetic substrate Spectrozyme® factor Xa according to the manufacturer (American Diagnostica Inc., Greenwich, CT) (data not shown).
Stage 2-Human plasma-derived factor VIII (outdated Hemofil M, Baxter Healthcare Corp.) at 30 nM concentration (1 unit Х 0.07 pmol) was incubated for various times (0 -120 min) at 37°C in HBS/Ca 2ϩ , 50 M PCPS, in the absence or presence of 5 nM APC, or in 5 nM APC plus 100 nM factor Xa-cleaved HPS or undigested HPS from stage 1.
Stage 3-Equal volumes (100 l) of HBS buffer, activated partial thromboplastin time reagent (Organon Teknika Corp., Durham, NC), and factor VIII-deficient human plasma (George King Bio-Medical Inc., Overland Park, KS) were combined and preincubated for 1 min at 37°C. One-l volumes (10 fmols of initial factor VIII) of samples from stage 2 were then added, followed immediately by initiation of clot formation at 37°C by the addition of 100 l of 25 mM CaCl 2 .
In control experiments, the cleavage and inactivation of protein S by human ␣-thrombin was also measured. The procedure was the same as that described above except 15 nM ␣-thrombin, no PCPS or Ca 2ϩ , and a 1-h incubation was used in stage 1. The ␣-thrombin was completely and irreversibly inhibited by incubation with 50 nM PPACK, and dialysis removed all active PPACK, as determined with amidolytic assays employing synthetic substrate Spectrozyme® TH (American Diagnostica Inc.) (data not shown). Stages 2 and 3 were identical to those described above.

Protein S Binding to Phospholipid
The effect of digestion by factor Xa on protein S binding to phospholipid was determined with a solid phase microtiter plate procedure described by van Wijnen et al. (43). Protein S was first digested with factor Xa in the presence of PCPS vesicles as described for stage 1 of the functional assay, above. EDTA was then added to 10 mM, and the digested protein S was loaded onto a Fast-Q-Flow column (0.2-ml bed volume) and washed with 10 column volumes of 20 mM Tris, pH 7.4, 150 mM NaCl, 5 mM benzamidine, 10 mM EDTA (to disrupt the calcium-dependent interaction of protein S with phospholipid and to remove the phospholipid by elution. The column was then washed with 10 volumes of the above wash solution lacking EDTA, followed by elution of the purified protein S with 400 l of 20 mM Tris, pH 7.4, 150 mM NaCl, 20 mM CaCl, as described previously (1). Digestion of protein S was confirmed by SDS-PAGE of the eluted protein as described above, and protein concentration was determined by the micro-BCA protein assay (Pierce) following the supplier's instructions and using purified, untreated human rHPS as a standard. Protein S was also digested with ␣-thrombin as described for stage 1 of the functional assay (above), followed by inhibition of thrombin by the addition of 40 mM PPACK, and directly added to the phospholipid-coated wells without column purification.
Increasing concentrations of digested protein S and undigested protein S (submitted to identical treatment without the addition of factor Xa or ␣-thrombin) were used in the solid phase lipid binding assay, as described in detail elsewhere (43). Briefly, PCPS (2.5 M, 100 l) was used to coat microtiter wells overnight at 4°C in 50 mM NaHCO 3 , pH 9.6, followed by blocking with 1% bovine serum albumin in 50 mM Tris, pH 7.4, 150 mM NaCl, 3 mM CaCl 2 for 1 h at room temperature. Different amounts of protein S (in 100 l) were then added in the above buffer containing 0.3% bovine serum albumin (binding buffer) and incubated for 2 h at room temperature, followed by three 10-min washes with 150 l of binding buffer. Bound protein S was then released by the addition of 100 l of 50 mM Tris, pH 7.4, 150 mM NaCl, 10 mM EDTA (elution buffer), and 2-h incubation at room temperature.
The eluted protein S was then measured in an enzyme-linked immunosorbent assay. Microtiter wells were coated with rabbit anti-human protein S IgG (P4555 Sigma), 100 l at 5 g/ml in the above binding buffer overnight at 4°C, followed by three 10-min washes with binding buffer. Eluted protein S from the phospholipid titer wells (100 l of elution buffer) was then added to the antibody-coated wells and incubated for 3 h at room temperature, followed by three 10-min washes with binding buffer. Mouse anti-human protein S monoclonal antibody 2a (44) was then added (100 l, 2 g/ml) and incubated overnight at 4°C. After washing, 100 l of a 1:4000 dilution of horse anti-mouse IgG monoclonal antibody-peroxidase conjugate (PI-2000, Vector Laboratories, Burlingame CA) was applied and incubated for 1 h at room temperature. The plates were then washed six times with binding buffer, followed by the addition of enzyme substrates (100 l of 50 mM citrate buffer, pH 5.0, containing 0.1 mg of o-phenylenediamine and 0.04 l of 30% H 2 O 2 ) and a 5-min incubation at room temperature. The reactions were quenched by the addition of 50 l of 4 M H 2 SO 4 , and the absorbance at 490 nm was measured on a V max spectrophotometer (Molecular Devices, Menlo Park, CA). Absorbance at 490 nM versus initial protein S concentration was plotted, and apparent K d values were calculated with the computer software PRIZM (GraphPad , San Diego, CA), assuming that protein S has one global binding site for phospholipid. Fig. 1A shows the time-dependent cleavage of rHPS by factor Xa. The results indicate that a heavy chain cleavage product is generated, which is indistinguishable from that resulting from ␣-thrombin cleavage (lane 10). The digestion of protein S, shown in Fig. 1, was performed in the presence of 20 nM hirudin, demonstrating that the cleavage is not due to contaminating ␣-thrombin in the purified plasma-derived factor Xa. The addition of tick anticoagulant peptide (90 nM) or D-Glu-Gly-Arg-chloromethyl ketone (150 M), both specific factor Xa inhibitors, completely blocked the digestion of rHPS by factor Xa under identical conditions as that described in Fig. 1 (2-h incubation), indicating that the cleavage is indeed factor Xaspecific (data not shown). Also, no cleavage by factor Xa is observed in the presence of 5 mM EDTA or in the absence of CaCl 2 or phospholipid (data not shown), indicating that the cleavage is both phospholipid-and Ca 2ϩ -dependent.

RESULTS
As shown in Fig. 1B, we also observed that factor Va nearly abolishes this cleavage under conditions where all the factor Xa is in the prothrombinase complex. These results suggest that factor Va inhibits the cleavage by interacting directly with factor Xa and preventing factor Xa accessibility to protein S.
The results of NH 2 -terminal amino acid sequencing are shown in Table I. Comparison of the observed Edman degradation amino acid products with the published sequence of human protein S reveals that cleavage occurs on the carboxylterminal side of Arg 60 and is distinct from the two ␣-thrombin cleavage sites (shown schematically in Fig. 4).
The effect of protein S cleavage by factor Xa on functional activity was measured in a factor VIII-dependent clotting assay (42). As a control, the effect of protein S cleavage by factor Xa was compared with that by ␣-thrombin. The functional consequences of rHPS digestion by factor Xa are presented in Fig.  2A. Under the conditions employed and as previously demonstrated, APC alone has impaired capabilities in inactivating factor VIII (filled squares) (22). The effect of APC alone is equal to the effect of HPS alone (filled triangles). Thus, as described previously, HPS has an APC-independent effect on the intrinsic tenase. In the presence of both APC and HPS, there is a significant increase in factor VIII inactivation (filled circles). This increase in inactivation rates may be the result of increased rate of cleavages at Arg 336 and Arg 562 . It is also possible that the increase in the inactivation rate seen in the pres-  a The number in parenthesis indicates pmol of amino acid at the given cycle.
b No predominant amino acid was observed, which is consistent with Cys at this point based on the sequence shown in Fig. 4. ence of HPS is the result of the two effects together (i.e. APCdependent and -independent) for intrinsic tenase formation. Following cleavage by factor Xa, there is loss of the APCindependent effect of HPS on factor VIII. Further, no effect on factor VIII inactivation by APC could be observed when factor Xa-treated HPS was added to the mixture. Thus, most likely, HPS acts synergistically with APC for the inactivation of intrinsic tenase assembly. APC acts by slow cleavage at Arg 336 and Arg 562 of factor VIII, whereas HPS acts by displaying factor VIII from the surface as described (27).
Essentially identical time-dependent curves for factor VIII(a) inactivation were obtained for thrombin-digested rHPS and pHPS digested with factor Xa or thrombin. Fig. 2B depicts the results obtained following a 60-min incubation of factor VIII with APC with the various forms of HPS (plasma or recombinant) shown in Fig. 3C. A loss in the APC-dependent and -independent activity of both HPS and rHPS was observed under the conditions employed following treatment with either factor Xa or thrombin. These data are apparently in contradiction with previously published observations (27). The apparent discrepancy between our study and the study of Koppelman et al. (27) can, however, be explained by the low concentration of HPS and the high lipid concentration used in our experiments.
In order to determine whether the loss of protein S anticoagulant activity following cleavage by factor Xa or ␣-thrombin is due to a loss of its ability to bind to the lipid membrane surface, direct binding of protein S to PCPS was measured. As shown in Fig. 3, the binding curves for uncleaved and cleaved protein S, either by factor Xa or thrombin, are indistinguishable within the error of this assay. In all cases, an apparent K d of approximately 4 nM was derived. This low value, relative to other vitamin K-dependent plasma proteins, is in good agreement with those reported by others (7-70 nM) for human protein S from different sources and using various purification procedures (43,45,46) and initially observed semiquantitatively by Nelsestuen et al. (47). Our results suggest that cleav-FIG. 2. Effect of rHPS with or without APC on factor VIII clotting activity. A, time course of factor VIII inactivation. Plasma-derived human factor VIII was incubated with dialyzed undigested or digested (factor Xa or thrombin) rHPS (with or without APC) for varying lengths of time, and the percentage of functional activity was measured in a clot-based assay as described under "Experimental Procedures." Percentage of activity is based upon a standard curve using different amounts of untreated factor VIII. Zero time incubation factor VIII alone clot time in the assay was 67 Ϯ 2 s, and no added factor VIII gave a clot time of 124 Ϯ 1 s. ࡗ, factor VIII alone; f, factor VIII plus APC; OE, factor VIII plus rHPS; q, factor VIII plus APC and rHPS; ϫ, factor VIII plus factor Xa-digested rHPS; E, factor VIII plus factor Xa-digested rHPS and APC. A decrease in percentage of factor VIII activity reflects functional HPS and APC anticoagulant activity. B, comparison of functional activity for different forms of protein S. 60-min incubation time points from panel A (rHPS (r) Ϯ factor Xa digestion, with or without APC) are plotted on a bar graph. Also presented are 60-min time points from similar plots of remaining factor VIII activity after incubation with undigested or factor Xa (Xa)digested plasma-derived HPS (p) and with thrombin (Th)-digested HPS, in the presence or absence of APC. age of protein S by either factor Xa or thrombin has no apparent effect on its ability to bind to phospholipid and suggest that the loss of anticoagulant activity is not due to the protein's inability to bind to the lipid bilayer.

DISCUSSION
In this paper, we report the specific cleavage of protein S by the procoagulant factor Xa. This cleavage is both phospholipidand Ca 2ϩ -dependent and is blocked by tick anticoagulant peptide and D-Glu-Gly-Arg-chloromethyl ketone, both specific inhibitors of factor Xa. As measured by a factor VIII-based clotting assay, factor Xa destroys both the APC-dependent and -independent anticoagulant activity of protein S under the conditions used. These results suggest a second or alternative proteolytic mechanism by which the anticoagulant activity of protein S may be physiologically regulated, involving factor Xa.
Cleavage of protein S by factor Xa results in a heavy chain cleavage product that is indistinguishable by SDS-PAGE analysis from that generated by thrombin, suggesting that factor Xa cleavage is also in the thrombin-sensitive disulfide loop of protein S. This conclusion was confirmed by direct NH 2 -terminal sequencing of the heavy chain product, which indicates cleavage by factor Xa at Arg 60 . It is notable that this site (shown in Fig. 4) is different from that for ␣-thrombin (Arg 49 and Arg 70 ) (1). Furthermore, it does not resemble the canonical substrate cleavage site in the primary target for factor Xa on prothrombin, -Ile-Glu/Asp-Gly-Arg-X- (48), and therefore represents a novel cleavage site for factor Xa Our data demonstrate that cleavage of protein S by factor Xa at Arg 60 inhibits both its APC-dependent and -independent inhibitory activity with respect to the intrinsic tenase. Our data are in apparent contradiction with the data published by Koppelman et al. (27). However, the latter study used very high amounts of protein S in a system where limited membrane surface is available. In contrast, our study uses low amounts of protein S and saturating amounts of phospholipids. Thus, the APC-independent effect of factor Xa-cleaved or thrombincleaved protein S on the intrinsic tenase activity can be only visible under conditions where limited membrane surface is available. It is also possible that the discrepancy between our results and the results of Koppelman et al. is due to the use of two different assays (activated partial thromboplastin time assay in the present work as compared with an Xa generation assay in the study of Koppelman et al.).
Within experimental error, digestion by factor Xa or ␣-thrombin had no effect on the affinity of protein S for surface phospholipid (K d app Х 4 nM). The capacity for PCPS binding to thrombin-cleaved protein S appeared to be slightly reduced but not statistically significant. Failure of either factor Xa or ␣-thrombin cleavage of protein S to significantly alter the binding of protein S to phospholipid suggests that the loss of protein S anticoagulant activity by proteolysis is not the result of reduced membrane surface binding but is more likely due to the direct interactions of protein S with target protein components or with APC. van Wijnen et al. (43) observed no difference between thrombin-cleaved and uncleaved human protein S binding to phospholipid. Contradictory results have been reported for thrombin-cleaved protein S by others using different forms of protein S and different assay systems. Walker (45) observed a marked reduction in binding capacity (ϳ6-fold) for thrombin-cleaved bovine protein S but no difference in apparent K d , as measured by light scattering. Hackeng et al. (26) also observed a reduced capacity of thrombin-cleaved human protein S to bind to cultured human umbilical vein endothelial cells. In contrast, Dahlback et al. (46) observed no difference in binding of throm- FIG. 3. Effect of factor Xa or ␣-thrombin digestion on protein S binding to phospholipid. Absorbance at 490 nm due to enzyme-linked immunosorbent assay substrate conversion is plotted versus the concentration of rHPS present in the initial phospholipid binding assay (see "Experimental Procedures" for detail of the assay). Inset, apparent K d values were determined from the curves with the computer program PRISM. f, recombinant protein S undigested and not submitted to Fast Q-Flow Sepharose chromatography to remove PCPS prior to phospholipid binding (control); OE, undigested rHPS submitted to chromatography after the addition of PCPS; , factor Xa-digested rHPS submitted to chromatography; ‚, thrombin-cleaved rHPS not submitted to chromatography. bin-cleaved versus uncleaved human protein S to platelet microparticles. The basis of these discrepancies is unclear and is not commented upon by the above authors, and it may involve the type of protein S utilized, the extent of proteolysis, purified component versus cellular substrates, and/or the nature of the binding assay. It is possible that endothelial cells have a protein receptor for protein S; thus, binding to synthetic phospholipid may be different in nature from binding to endothelial cells and cannot be compared.
One very interesting finding presented here is that factor Va markedly inhibits cleavage of protein S by factor Xa. It is well established that factor Xa alone and factor Xa associated with factor Va (within prothrombinase) behave as two different enzymes. The most striking example of this is the activation of prothrombin to thrombin. Factor Xa alone activates prothrombin to thrombin with a first cleavage at Arg 284 followed by a second cleavage at Arg 322 . In contrast, the order of cleavage is inverted when using prothrombinase, and prothrombin activation proceeds through the active intermediate meizothrombin (49). The inability of factor Xa to readily cleave protein S when it is associated with factor Va in the prothrombinase complex is consistent with a change in the catalytic properties of the enzyme. Under the initiation stages of whole blood clotting, the limiting component in prothrombinase formation (and thereby active thrombin generation) is factor Xa (50). Also, under these initial conditions factor Va is maintained in excess of prothrombinase (50). Our data suggest that an additional way in which factor Va acts to promote an initial procoagulant state is by sequestering factor Xa and/or altering its specificity, thereby increasing its ability to generate thrombin and reducing its ability to cleave protein S, which would also be expected to be present. An added benefit of this inhibition is that during the initial and propagation phases of clotting, protein S is spared from cleavage so that it can function at a later time as an anticoagulant and at distal locations from the site of desired clot formation.
It has been known for many years that thrombin cleaves and abolishes the APC-dependent and -independent anticoagulant activity of protein S. However, this inactivation by thrombin may not be of physiological significance, as evidenced by the complete inhibition of protein S cleavage by thrombin at physiological concentrations of Ca 2ϩ (2.5 mM) (51). Furthermore, the endothelial surface protein, thrombomodulin, whose major role is to potentiate the activation of protein C via thrombin cleavage, also inhibits the cleavage of protein S by thrombin (51). Finally, at least in the context of protein S APC-cofactor function, cleavage and inactivation of protein S by thrombin would be inconsistent with thrombin's role as an anticoagulant via the protein C pathway.
Cleavage and inactivation of protein S by a second protease, factor Xa, supports the idea that the thrombin-sensitive disulfide loop is conformationally poised for cleavage. Our results suggest that the cleavage of protein S by factor Xa may be an important component in physiological regulation of anticoagulant activity and hemostasis.