Inhibition of Prothrombinase by Human Secretory Phospholipase A2 Involves Binding to Factor Xa*

Human group II secretory phospholipase A2 (hsPLA2) exhibits significant anticoagulant activity that does not require its enzymatic activity. We examined which coagulation factor was targeted by hsPLA2and analyzed which region of the protein may be involved in this inhibition. Prothrombin time coagulation assays indicated that hsPLA2 did not inhibit activated factor V (FVa) activity, whereas activated factor X (FXa) one-stage coagulation assays suggested that FXa was inhibited. The inhibitory effect of hsPLA2 on prothrombinase activity of FXa, FV, phospholipids, and Ca2+complex was markedly enhanced upon preincubation of hsPLA2with FXa but not with FV. Prothrombinase activity was also strongly inhibited by hsPLA2 in the absence of PL. High concentrations of FVa in the prothrombinase generation assay reversed the inhibitory effect of hsPLA2. By using isothermal titration calorimetry, we demonstrated that hsPLA2 binds to FXa in solution with a 1:1 stoichiometry and a K d of 230 nm. By using surface plasmon resonance we determined the rate constants, k on andk off, of the FXa/hsPLA2 interaction and analyzed the Ca2+ effect on these constants. When preincubated with FXa, synthetic peptides comprising residues 51–74 and 51–62 of hsPLA2 inhibited prothrombinase assays, providing evidence that this part of the molecule, which shares similarities with a region of FVa that binds to FXa, is likely involved in the anticoagulant interaction of hsPLA2 with FXa. In conclusion, we propose that residues 51–62 of hsPLA2 bind to FXa at a FVa-binding site and that hsPLA2 decreases the prothrombinase generation by preventing FXa·FVa complex formation.

The human group II secretory phospholipase A 2 (hsPLA 2 ) 1 has been detected in various cellular types including macrophages, eosinophils, and blood platelets (1). hsPLA 2 is associated with the ␣-granules of platelets and is released into the extracellular medium upon activation (2). The hsPLA 2 shares common characteristics with other group II secretory phospholipase A 2 (sPLA 2 ), including polypeptide sequence homology, and its catalytic mechanism of action is identical (3)(4)(5)(6). The hsPLA 2 has attracted particular attention with respect to inflammatory diseases since its level in body fluids correlated with the severity of the pathological states (7,8). hsPLA 2 may be involved in the degradation of bacteria (9), in exocytosis/ degranulation processes (10,11), and in the production of eicosanoids by stimulated inflammatory cells (12)(13)(14).
The hsPLA 2 does not participate in the production of eicosanoids during platelet activation and does not interfere with platelet stimulation once secreted (15,16). Platelet activation plays a central role during hemostasis, leading to primary plug formation and increasing the efficiency of the coagulation process (17)(18)(19). Blood platelets are also a source of factor V (20). It was thus tempting to suggest a role for hsPLA 2 on blood coagulation, once it is secreted by activated platelets. The prothrombinase complex, composed of FVa, FXa, phospholipids (PL), and Ca 2ϩ , plays a central role in the coagulation cascade (17,21). hsPLA 2 exerts a moderate anticoagulant effect on plasma (22) and inhibits prothrombinase activity (23,24). The anticoagulant effect of hsPLA 2 is phospholipid-independent and not observed when FV-deficient plasma is used for coagulation assays (24). This observation leads to the hypotheses that the molecular target of hsPLA 2 is either the FV/FVa protein, the FXa protein, or both proteins.
All sPLA 2 s have a common active site but may possess distinct pharmacological sites, explaining the diversity of pharmacological effects of venom sPLA 2 s, such as neurotoxicity, myotoxicity, cardiotoxicity, inhibition or potentiation of platelet aggregation, and anticoagulant action (25,26). In particular, all venom anticoagulant sPLA 2 s have a basic pI correlated with the presence of basic amino acids located between residues 50 and 80 which may be involved in the anticoagulant effect of venom sPLA 2 in addition to the lipolytic activity (27,28).
The aim of this study was to identify the human plasma coagulation factor target for the anticoagulant effect of hsPLA 2 in biochemical studies using purified proteins. Isothermal titration calorimetry affords simultaneous determination of all thermodynamic parameters relevant to protein-protein binding, including N, the stoichiometry of the association reaction; K a , the association constant (K a ϭ 1/K d ); and the enthalpic (⌬H 0 ) and entropic (⌬S 0 ) contributions to the Gibbs free energy of association (⌬G 0 ) (29,30). By using this technique, we demonstrated here that hsPLA 2 binds to FXa in solution. By using surface plasmon resonance, we investigated the effect of calcium ion on the kinetic rate constants of the FXa/hsPLA 2 interaction. Furthermore, by using overlapping synthetic peptides corresponding to residues 51-74, we determined which part of the hsPLA 2 is likely to be involved in its anticoagulant action.

Materials
Human prothrombin was purified according to diScipio and Davie (31). Human FV was purified as described (32) with minor modifications (33). Activation of human FV was performed as described previously (33). FIXa, FX, and FXa were from Enzyme Research Laboratories (South Bend, IN). Recombinant FVIII was a kind gift of Dr. Roger Lundblad of Baxter Hyland, Duarte, CA. The chromogenic substrate for thrombin amidolytic activity, CBS 34.47, and the chromogenic substrate for FXa amidolytic activity, S-2222, were from Diagnostica Stago (Asnières, France) and Chromogenix (Cincinnati, OH). FV-deficient plasma was from George King Bio-Medical (Overland Park, KS). Innovin (recombinant tissue factor reagent) was from Dade Diagnostics (Miami, FL); BSA (fraction V) was from Sigma. The synthetic scrambled peptide containing hsPLA 2 residues 51-74 as well as peptides 51-62, 59 -70, 63-74, 62-51 (reverse) and D-51-62 (all residues in the Dconfiguration) were from Neosystem, Isochem SA (Strasbourg, France). The synthetic peptide containing residues 51-74 of the hsPLA 2 was synthesized by the organic chemistry unit of Pasteur Institute (Paris, France). The peptides were purified by high pressure liquid chromatography (purity Ն95%), and the predicted sequence was verified by the observed mass using mass spectroscopy. All these peptides have an N-terminal acetyl group and a C-terminal amide group.

Preparation of Recombinant hsPLA 2
The recombinant Leu 8 -hsPLA 2 was expressed as a fusion protein in Escherichia coli. This phospholipase A 2 species is referred to as hsPLA 2 throughout this paper. Active hsPLA 2 was obtained following tryptic cleavage of the fusion protein and then purified as described (24,34).
FV, FXa, and PL-In FXa/FV preincubation conditions, 20 pM FXa was incubated with 120 pM FV for 4 min, and then the reaction was started with 5 M PL and 200 nM prothrombin. In FV preincubation conditions, 10 pM FV was incubated for 4 min, and then the reaction was started with 1 nM FXa, 5 M PL, and 200 nM prothrombin. In FXa preincubation conditions, 10 pM FXa was incubated for 4 min, and then the reaction was started with 1 nM FV, 5 M PL, and 200 nM prothrombin.
FVa and FXa-FXa was incubated for 4 min with FVa at the indicated concentrations, and then the reaction was started with 1 M prothrombin.
FXa and PL-2 nM FXa was incubated for 4 min, and then the reaction was started with 5 M PL and 1 M prothrombin.
FVa, FXa, and PL-The same experimental procedure was followed as for FXa/FV preincubation conditions, except that 120 pM FV was replaced by 20 pM FVa.
After the addition of prothrombin, aliquots were taken at various time points, and prothrombin activation was immediately stopped by the addition of 50 mM EDTA (final concentration). The level of activated prothrombin activity was determined by hydrolysis of the chromogenic substrate CBS (300 M, final concentration), monitored at 405 nm, and expressed in terms of thrombin concentration, using a calibration curve established with purified ␣-thrombin. To determine the effect of hsPLA 2 or synthetic peptides on prothrombinase activity under these different conditions, these compounds (or control buffer) were added to the various reaction mixtures at the beginning of the 4-min preincubation period.

Addition of FVa during Prothrombinase Assay
FXa/FV preincubation conditions were used as described above. Briefly, 20 pM FXa was incubated with 120 pM FV for 4 min, and then the reaction was started with 5 M PL and 200 nM prothrombin. After 6 min, the indicated amounts of FVa were added. The formation of activated prothrombin was followed as a function of time, as described above.

Intrinsic Tenase Assay
Intrinsic tenase purified protein components were mixed at 37°C using purified components in Tris-buffered saline under the following conditions. 25 nM FIXa was incubated for 5 min with 1 nM FVIII in the absence or in the presence of defined hsPLA 2 amounts, and then FX activation was started by addition of 1 M FX (final concentrations). Aliquots were then taken at various time points, and the reaction was immediately stopped by the addition of aliquots containing 50 mM EDTA (final concentration). The level of FXa activity was determined by hydrolysis of the chromogenic substrate S-2222 (200 M, final concentration) monitored at 405 nm, in comparison to a standard curve using purified FXa.

FXa One-stage Coagulation Assay
FXa (20 nM) was incubated at 37°C in Hepes-buffered saline (50 mM Hepes, pH 7.4, 0.1% BSA, 0.1 M NaCl) for 2 min with various concentrations of hsPLA 2 , in the absence or presence of 5 mM CaCl 2 . Then, FXa one-stage coagulation assays were performed as follows. 20 l of the incubation mixture was added to a pre-warmed mixture of 25 l of FV-deficient plasma and 30 l of PL (166 M). After 1 min, coagulation was started by the addition of 50 l of CaCl 2 (20 mM). The coagulation time was recorded using an ST4 coagulometer (Diagnostica Stago, Asnières, France).

FVa Activity Measured in a Prothrombin Time Coagulation Assay
FVa (1 nM) was incubated for 2 min at 37°C in Hepes-buffered saline containing 5 mM CaCl 2 in the presence of various concentrations of hsPLA 2 . Then, a prothrombin time assay was performed as follows. 5 l of the incubation mixture was added to a pre-warmed mixture of 50 l of FV-deficient plasma and 45 l of Hepes-buffered saline. After 1 min, coagulation was started by the addition of 50 l of Innovin (recombinant tissue factor reagent). Clotting times were recorded using an ST4 coagulometer.

Measurement of FV Activation by Thrombin or FXa
FV (300 nM) was incubated for 30 min in Hepes-buffered saline containing 5 mM CaCl 2 at 37°C with or without 3.5 M hsPLA 2 , and then the activation was started by the addition of 1 nM ␣-thrombin. During 60 min, the amount of generated FVa at different times was determined by performing a prothrombin time coagulation assay as described above, and FV proteolysis was also analyzed by SDS-polyacrylamide gel electrophoresis. In experiments using PL, FV (3 nM) was incubated for 30 min in Hepes-buffered saline containing 5 mM CaCl 2 at 37°C in the presence of 25 M PL, and with or without 3.5 M hsPLA 2 , then the activation was started by the addition of 6 nM FXa. During 120 min, the amount of generated FVa was determined at different times by performing a prothrombin time coagulation assay as described above.

Isothermal Titration Calorimetry (ITC)
Experiments were carried out on the MicroCal MCS ultrasensitive titration calorimeter (MicroCal Inc., Northampton, MA) using the OBSERVER software provided by the manufacturer for instrument control and data acquisition (36). To improve base-line stability, the temperature of the system was kept at 5°C below the temperature of the actual experiment with a water bath, and temperature was equilibrated for 12 h. During a titration experiment, the FXa sample was thermostated at 37.0 Ϯ 0.1°C in a stirred (410 rpm) reaction cell (1.3514 ml), and 31 injections, each of 8-l volume and 5-s duration, with a 3.5-min interval between injections, were carried out using a 250-l syringe filled with hsPLA 2 solution. An injection series was preceded by a 2-l calibration injection. The reference cell of the calorimeter contained water plus 0.01% sodium azide. Data points were averaged and stored at 2-s intervals. All buffer solutions (Tris-HCl 20 mM, pH 7.4, NaCl 0.125 M, and CaCl 2 5 mM) were thoroughly degassed by stirring under vacuum before use. Protein samples were prepared in buffer of the same batch to minimize artifacts due to any differences in buffer composition. Titration experiments were performed with 1.5, 3.5, and 4.1 M FXa and corresponding concentrations of hsPLA 2 in the syringe, ensuring a final hsPLA 2 :FXa mole ratio of 2:1 in the reaction cell. Raw calorimetric data, i.e. heats absorbed or released accompanying the addition of aliquots of the hsPLA 2 solution into the FXa solution, were processed using the software package ORIGIN (36,37). The area under the resulting peak following each injection is proportional to the heat of interaction Q. When corrected for the titrant dilution heat and normalized to the concentration of added titrant, Q is equal to the binding enthalpy ⌬H' b at that particular degree of binding. The calorimetric binding isotherm was fitted by an iterative nonlinear least squares algorithm (Marquardt method) to a binding model employing a single set of independent sites. The association (K a ) and dissociation (K d ) constants, molar binding stoichiometry (N), and molar binding enthalpy (⌬H 0 ) were determined directly from the fitted curve. The Gibbs free energy and molar entropy of binding were calculated using the equations ⌬G 0 ϭ -RT In K a and ⌬S 0 ϭ (⌬H 0 -⌬G 0 )/T, respectively, where R is the gas constant, and T is the absolute temperature in degrees Kelvin.

Surface Plasmon Resonance (SPR) Experiments
Studies were performed using a BIACORE ® 2000 system (Biacore AB, Uppsala, Sweden). Reagents, including surfactant P20, the amine coupling kit containing N-hydroxysuccinimide, N-ethyl-NЈ-(3-dimethylaminopropyl)-carbodiimide, ethanolamine hydrochloride, and Sensor Chips CM5 were supplied by Biacore. The immobilization of FXa on the sensor chip surface was performed as follows: 30 l of FXa (14 g/ml in 10 mM sodium acetate, pH 4.8) was covalently coupled via primary amino groups on a CM5 sensor chip surface according to the manufacturer's description (Biacore). The immobilization run was performed at a flow of 5 l/min at 25°C. The SPR signals for immobilized FXa (three different flow cells with three different quantities of FXa) were found to be 4,400 resonance units (RU), 2,000 RU, and 1,000 RU, where 1 RU corresponds to an immobilized protein concentration of ϳ1 pg/mm 2 . Unreacted moieties on the surface were blocked by ethanolamine. One independent flow cell of the same sensor chip, used as a control flow cell, was subjected to a "blank immobilization," i.e. with no FXa added. All experiments were carried out in 10 mM Hepes, pH 7.4, 0.005% surfactant P20, and 150 mM NaCl. hsPLA 2 , varying from 0 to 25 g/ml, was injected in the same buffer in the presence or in the absence of CaCl 2 at 5 mM, with a flow of 10 l/min. Between each injection, surfaces were regenerated with 10 l of 1 M NaCl. Analyses were performed at 25°C. Kinetic constants, k on (association rate constant) and k off (dissociation rate constant), for the interaction of hsPLA 2 with immobilized FXa were calculated using BIA evaluation 2.1 software using curve fitting to a simple two-component model of interaction (A ϩ B ϭ AB) for a titration of the solution hsPLA 2 -immobilized FXa (38). Values, for a series of FXa⅐hsPLA 2 complexes, were determined after subtraction of control signals obtained from the injection of various hsPLA 2 concentrations on the control flow cell.

Statistical Analyses
The significance of the data was evaluated with the Student's t test for unpaired data.

Effect of hsPLA 2 on FXa and FVa Activities Measured in
Coagulation Assays-FXa one-stage coagulation assays, performed with FV-deficient plasma and fixed amounts of FXa that had been preincubated with varying amounts of hsPLA 2 , showed a dose-dependent inhibition of FXa activity by hsPLA 2 (Fig. 1). The inhibition was strongly reduced in the absence of Ca 2ϩ in the preincubation mixture, even though Ca 2ϩ was later present during FXa one-stage assays (Fig. 1). This implies that the inhibitory effect of hsPLA 2 does not involve a Ca 2ϩ -dependent action on PL, but rather an interaction with FXa that requires the presence of Ca 2ϩ . The effect of hsPLA 2 on FVa activity was investigated with a prothrombin time coagulation assay using FV-deficient plasma. Fixed amounts of FVa were preincubated with hsPLA 2 before the assay, in which coagulation efficiency was related to the activity of FVa. Fig. 1 shows that preincubation of hsPLA 2 with FVa did not reduce the activity of FVa. Thus, hsPLA 2 did not inhibit FVa activity, whereas FXa was inhibited by hsPLA 2 .
Inhibition of Prothrombinase Activity by hsPLA 2 -Prothrombinase complex formation analyses can be performed under optimal conditions (FXa, FVa, PL, and Ca 2ϩ ) and under suboptimal conditions (i.e. using FV instead of FVa, as well as in the absence of FVa or PL). The effect of hsPLA 2 on prothrom-binase complex formation was first analyzed under different conditions that were suboptimal for prothrombinase activity. Using 20 pM FXa, 120 pM FV, 5 M PL, and 5 mM Ca 2ϩ , we observed a strong inhibition of generation of prothrombinase activity by low concentrations of hsPLA 2 when hsPLA 2 was preincubated with FXa and FV ( Fig. 2A). When the PL concentration was decreased to 1 M, or increased to 50 M, the inhibitory activity of hsPLA 2 was essentially the same (data not shown). When hsPLA 2 was preincubated with FXa ( Fig.  2B), the inhibitory action of hsPLA 2 was much stronger than for preincubation with FV ( Fig. 2C), consistent with the above observations made using coagulation assays.
The effect of hsPLA 2 on prothrombinase activity was then tested in the absence of PL. We observed a potent inhibitory effect of hsPLA 2 in the presence of FVa (Fig. 3, A and B). In contrast, the action of FXa on prothrombin in the absence of both PL and FVa was unaffected by hsPLA 2 (Fig. 3, A and B), suggesting that hsPLA 2 does not interfere with direct prothrombin activation by FXa. Moreover, the amidolytic activity of FXa on its chromogenic substrate, S-2222, was not inhibited by hsPLA 2 (data not shown). The inhibition of prothrombinase (20 nM FXa, 120 pM FVa, and 5 mM Ca 2ϩ ) by hsPLA 2 was never complete in the absence of PL, even at high concentrations of hsPLA 2 (10 M), and we observed that the rate of prothrombin activation decreased until a value corresponding the same as that obtained with FXa alone, i.e. 0.16 Ϯ 0.008 nM thrombin⅐ min Ϫ1 compared with 0.14 Ϯ 0.004 nM thrombin⅐min Ϫ1 (Fig.  3A). Under these suboptimal conditions, only a fraction of FXa is bound to FVa, and the prothrombinase activity of free FXa is not inhibited by hsPLA 2 . When assays were performed in the absence of PL under conditions that increase the ratio of FXa bound to FVa (1 nM FXa, 1 nM FVa, and 5 mM Ca 2ϩ ), the percent of inhibition by hsPLA 2 was much higher, and at high hsPLA 2 levels the prothrombin activation rate approached that seen for FXa alone (Fig. 3B). Then, 20 l of the incubation mixture was used to perform an FXa one-stage assay as described under "Experimental Procedures." The amount of active FXa was determined using a calibration curve of purified FXa. The activity of FVa was measured using prothrombin time coagulation assays, performed as described under "Experimental Procedures," in the presence of 5 mM CaCl 2 and the indicated amounts of hsPLA 2 (OE). The FVa activity was determined using a calibration curve of purified FVa. The means Ϯ S.E. of three independent experiments are shown.
The effect of hsPLA 2 on intrinsic tenase activity in the absence of PL was examined. Intrinsic tenase-purified components included 25 nM FIXa, 1 nM FVIII, and 5 mM Ca 2ϩ . The substrate, FX, was used at 1 M. We did not observe any inhibitory effect of the hsPLA 2 on the ability of these components to activate FX under these conditions, even at hsPLA 2 concentrations up to 10 M (data not shown). Thus, the PLindependent anticoagulant action of hsPLA 2 was specific for the prothrombinase complex.
We observed an inhibition of the prothrombinase generation in the absence of FVa but in the presence of PL. This may result from an inhibition of the binding of FXa to PL vesicles due to an interaction of hsPLA 2 with FXa, or PL vesicles, or both (data not shown).
To investigate whether the observed inhibition of prothrombinase complex formation by hsPLA 2 was caused by an effect on the activation of FV by ␣-thrombin or by FXa, several studies of FV activation were performed. First, 300 nM FV was preincubated for 30 min with or without 3.5 M hsPLA 2 , and then activation was started by the addition of 1 nM thrombin. Second, 3 nM FV was incubated for 30 min with or without 3.5 M hsPLA 2 in the presence of 25 M PL, and activation was started by the addition of 6 nM FXa. In both cases, the forma-tion of FVa was determined as a function of time in a prothrombin time assay using FV-deficient plasma. We observed that hsPLA 2 was unable to inhibit FV activation by either thrombin or FXa/PL (data not shown). This suggested that hsPLA 2 may specifically inhibit prothrombin activation by inhibiting the formation of the prothrombinase complex.
Reversal of hsPLA 2 Inhibition of Prothrombinase Generation by Addition of FVa-When conditions were optimal for prothrombinase assembly (i.e. 20 pM FXa, 20 pM FVa, 5 M PL, and 5 mM Ca 2ϩ ), we observed a loss of the inhibitory action of hsPLA 2 (data not shown). To understand the lack of the inhibitory effect of hsPLA 2 on prothrombinase activity in the presence of FVa and PL, we reconstituted the prothrombinase complex with FV instead of FVa (20 pM FXa, 120 pM FV, 5 M PL, and 5 mM Ca 2ϩ ). 3.5 M hsPLA 2 was preincubated with FV, FXa, and PL for 4 min, and then prothrombin (200 nM) was added, and prothrombin activation was followed for 6 min. At this time, various amounts of FVa (4, 20, or 200 pM) were added to the prothrombinase reaction mixture. Consistent with previous experiments ( Fig. 2A), in the absence of added FVa, a strong inhibition of prothrombinase generation by hsPLA 2 was seen during the first 6 min (Fig. 4). However, addition of increasing amounts of FVa at 6 min led to a dose-dependent reversal of the anticoagulant effect of hsPLA 2 (Fig. 4). It should be noted in Fig. 4 that in the presence of hsPLA 2 and FV but in the absence of additional FVa, the maximal rate of thrombin generation indicated by the slope of the prothrombinase activity curve eventually reached that of the prothrombinase complex obtained in the absence of hsPLA 2 after a prolonged lag time. Thus, in the presence of hsPLA 2 , the prothrombin activation curve was shifted to the right along the x axis, leading to the suggestion that the assembly of the prothrombinase complex was delaying but eventually was formed and yielded the same final activity. Using the same rationale as described above, this observation can be easily explained as follows: whereas hsPLA 2 is effectively able to inhibit prothrombinase activity in the presence of FV, traces of prothrombin activation will provide enough thrombin to activate FV, yielding high enough levels of FVa to subsequently counteract the inhibitory effect of hsPLA 2 . To rule out the possibility that the lag was due to proteolytic destruction of hsPLA 2 , we used SDS-polyacrylamide gel electrophoresis to show that the hsPLA 2 was not cleaved by either FXa or thrombin during the prothrombinase complex activity measurements (data not shown).
The hsPLA 2 was thus able to inhibit the prothrombinase activity under conditions that were suboptimal for prothrombinase complex assembly, even when FVa was present. However, when prothrombinase complex assembly conditions were optimal, the presence of FVa reversed the inhibitory effect of hsPLA 2 . These results suggest that hsPLA 2 could either bind to FXa at the same site(s) as FVa or that binding of FVa to FXa may modify the structure of FXa leading to the dissociation of hsPLA 2 .
Binding of hsPLA 2 on FXa-The direct association of hsPLA 2 and FXa was demonstrated using isothermal titration calorimetry (ITC). Fig. 5 (top panel) shows original data from the calorimetric titration of a 1.4-ml solution containing 4.1 M FXa with a solution containing 45.7 M hsPLA 2 at 37°C. Two kinetic phases were associated with each injection. Immediately following injection, an initial exothermic phase (negative numbers) which we will refer to as site binding was observed. This was followed by a smaller and slower endothermic phase (positive numbers) suggesting that a slow conformational rearrangement takes place after binding. As the number of injections increases, binding sites become saturated since the final exothermic peaks decrease in size and remain nearly constant with further injections of hsPLA 2 . The corresponding binding isotherm in Fig. 5 (filled squares, bottom panel) shows the association transition. A least squares estimate of association parameters using a single-site binding model gave N ϭ 0.95 Ϯ 0.10, ⌬H 0 ϭ -4.56 Ϯ 0.08 kcal/mol, ⌬S 0 ϭ 15.8 Ϯ 0.8 cal/mol/K (⌬H 0 /⌬G 37°C ϭ 48%), and K d ϭ 230 Ϯ 30 nM. These results (Fig.  5) were confirmed by titration experiments performed at lower concentrations of FXa (1.5 and 3.5 M; data not shown). Therefore, these experiments show that hsPLA 2 and FXa bind with a small change in enthalpy (⌬H 0 ) in a 1:1 stoichiometry. The K d value is between the K d values reported for the FVa⅐FXa complex in the absence of PL (0.8 M) and in the presence of PL (1 nM) (39,40).
Effect of Ca 2ϩ on k on and k off Rate Constants of FXa/hsPLA 2 Interaction-The inhibitory effect of hsPLA 2 on FXa activity, measured in FXa one-stage assays, was mainly observed in the presence of Ca 2ϩ (Fig. 1). Based on this observation, the influence of Ca 2ϩ on FXa/hsPLA 2 binding affinity was tested using surface plasmon resonance (SPR), studies that require less protein material than ITC. SPR allows measurements of the association rate constant (k on ) and of the dissociation rate constant (k off ) of hsPLA 2 to immobilized FXa. The ratio of the rate constants measured by SPR provides an apparent dissociation constant (K d app ϭ k off /k on ) that is an estimation of the equilibrium dissociation constant in solution (K d ) measured by ITC. The K d app value is usually smaller than the equilibrium K d value as SPR rate constant measurements are performed far from equilibrium, and with one immobilized protein which decreases the overall entropy of the association reaction with respect to free protein association in solution.
Effects of Ca 2ϩ on k on and k off for FXa/hsPLA 2 association are given in Table I. Table I shows that hsPLA 2 binds to FXa immobilized onto the sensor chip surface in the absence of Ca 2ϩ . However, 5 mM Ca 2ϩ in the screening buffer decreases k off nearly 10-fold and increases k on nearly 3-fold, leading to a 30-fold increase in K d app . Therefore, data in Table I demonstrate that Ca 2ϩ allows the formation a higher affinity FXa⅐hsPLA 2 complex.
Inhibition of Prothrombinase Activity by Peptides Related to Residues 51-74 of hsPLA 2 -We investigated the anticoagulant effect of the synthetic peptide 51-74, corresponding to the basic region of the hsPLA 2 (see Table II for the amino acid sequence of the whole molecule and Table III for three smaller 12-mer peptides related to peptide 51-74). A scrambled peptide containing a randomized sequence comprising the amino acids of residues 51-74, as well as a reversed peptide 62-51 and a control peptide 51-62 that contained all amino acids in the D-configuration were used to characterize the specificity of the inhibitory effects observed with peptides 51-75 and 51-62, respectively.
When peptide 51-74 was preincubated with FXa and FV in the presence of PL, we observed a dose-dependent inhibition of prothrombinase generation that progressively disappeared as described in Fig. 4 for the inhibitory action of hsPLA 2 (data not shown). The IC 50 value (where IC 50 is the concentration of inhibitor producing 50% inhibition) for the peptide 51-74 was 8 Ϯ 1 M (Table III). Among the three smaller overlapping 12-mer peptides, only peptide 51-62 inhibited prothrombinase activity with an IC 50 value of 22 Ϯ 3 M, close to that observed for the larger peptide 51-74 (Table III). The control scrambled peptide, which did not inhibit in 0.22 M NaCl, was weakly inhibitory in 0.1 M NaCl (IC 50 of 70 Ϯ 5 M) (Table III). This supports the idea that the inhibitory effect of the scrambled peptide in 0.1 M NaCl concentration was nonspecific due to its basic nature (net charge ϩ6) rather than due to a specific sequence of residues. In contrast, the inhibition of prothrombinase activity by peptides 51-74 and 51-62 was maintained with similar efficiency in 0.22 M NaCl (Table III). A 12-mer reversed peptide 62-51 was devoid of inhibitory effect, as was the control peptide D-51-62 with all amino acid residues in a D-configuration. These results indicate that the basic region (residues 51-74) of hsPLA 2 , and particularly residues 51-62, specifically inhibits prothrombinase activity and is most likely responsible for the inhibition of FXa⅐FVa complex formation. Furthermore, the stereospecific L-conformation of amino acids in peptide 51-62 was required for the inhibition of FXa⅐FVa complex formation, since control peptides with either a reversed sequence or with D residues were inactive. DISCUSSION hsPLA 2 , secreted during platelet activation, exhibits anticoagulant activity and may exert a negative feedback regulation on coagulation, which would prevent an excessive procoagulant effect of activated platelets (24). This inhibitory effect does not require the enzymatic activity of the enzyme, indicating that hsPLA 2 may interact with non-phospholipid targets (24). It was suggested that some venom sPLA 2 s are able to affect blood coagulation by mechanisms that do not involve their catalytic activity, although the targets of the venom enzymes were not identified (41)(42)(43). We investigated which plasma coagulation factors may be affected by hsPLA 2 . Prothrombin time coagulation assays indicated that hsPLA 2 did not inhibit FVa activity, whereas FXa one-stage coagulation assays suggested an inhibition of FXa by hsPLA 2 particularly in the presence of Ca 2ϩ . An inhibitory action of hsPLA 2 on prothrombinase activity has also been observed (23), although the molecular mechanism was not investigated.
Our results indicate that hsPLA 2 is able to inhibit prothrombinase activity, especially when the conditions are not optimal for prothrombinase complex assembly. We demonstrated that, in the absence of PL and in the presence of FVa, preincubation of hsPLA 2 with FXa effectively down-regulated prothrombinase generation (Fig. 3). When FV was present instead of FVa in the presence of PL, hsPLA 2 inhibited generation of prothrombinase activity only until there was enough FVa generated by traces of thrombin/FXa to overcome hsPLA 2 (Fig. 4). The activity of the intrinsic tenase complex (FIXa, FVIII, and Ca 2ϩ ) was unaffected by hsPLA 2 , although this coagulation complex shares common characteristics with the prothrombinase complex (associated with the structural homology of FVa and FVIIIa, and of FIXa and FXa). Moreover, activation of FV by either FXa or thrombin was not inhibited by hsPLA 2 . These observations indicate that the inhibitory action of hsPLA 2 is likely to be specific for the prothrombinase complex or for its formation.
Our results suggest that hsPLA 2 may bind to FXa at the same site(s) as FVa, but with a lower affinity, and thereby may decrease the prothrombinase activity by inhibiting the formation of a FXa⅐FVa complex. This was supported by the demonstration that hsPLA 2 binds to FXa with a 1:1 stoichiometry and a K d value of 230 nM (Fig. 5). Although not necessary for the interaction, the Ca 2ϩ increases the k on rate constant and decreases the k off rate constant, leading to a higher affinity of hsPLA 2 for FXa ( Table I). The prothrombinase complex has a catalytic efficiency in the activation of prothrombin that is several orders of magnitude higher than FXa acting alone. In the absence of PL, the association of FVa with FXa is governed by a K d of 0.8 M and is dependent on the presence of Ca 2ϩ (39). In the presence of PL vesicles and Ca 2ϩ , the K d of FVa for the PL⅐FXa complex decreases to approximately 1 nM (40). This may explain why, under suboptimal conditions where FVa was added in the absence of PL, FVa was unable to eliminate the hsPLA 2 inhibitory effect, since in this case the affinity of FVa for FXa is lower than that of hsPLA 2 for FXa.
We (24) and others (22) have previously reported that hsPLA 2 possesses an anticoagulant activity in whole plasma, as demonstrated by the increase of its recalcification time, and that this anticoagulant activity in plasma is even observed when activated platelets are present. hsPLA 2 is thus able to produce an inhibitory effect on blood coagulation under experimental conditions occurring during clot formation. Moreover, it is well established that the level of hsPLA 2 in serum is strongly increased, from 0.35 nM up to 0.6 M, under various pathological states associated with inflammation, as in the case of acute pancreatitis, multiple organ failure, septic shock, or rheumatoid arthritis (44 -47). Blood platelets secrete large amounts of hsPLA 2 upon activation (2,15), and it is likely that high local concentrations of this enzyme are reached during stasis platelet activation and secretion or during local thrombolytic events, even if hsPLA 2 levels are more difficult to assess than in serum (48). We demonstrate in this study that the hsPLA 2 /FXa interaction is governed by a K d of 230 nM, a value fully consistent with a potential localized and/or systemic anticoagulant role of hsPLA 2 during various physiologic or physiopathological states.
Although it is more difficult to extrapolate these results to physiological situations, where additional components also contribute to the hemostasis process in vivo, our study suggests that hsPLA 2 released by activated platelets during primary hemostasis may act as a negative modulator of thrombin generation by inhibiting the assembly of the prothrombinase complex. We already mentioned that the K d value of hsPLA 2 /FXa interaction (230 nM) is higher than the one of FVa/FXa interaction (1 nM). For equal concentrations of hsPLA 2 and FVa, these K d values would favor FXa/FVa interactions. However, at the initiation stage of coagulation, when only very low amounts of FVa are present (the precursor form of FVa, FV, being the major form circulating in plasma or stored into ␣-granules of platelets), the advantage would favor hsPLA 2 /FXa interactions. In agreement, we pointed out the original negative role of hsPLA 2 , since its effect will be significant at early stages of the coagulation process until enough FVa is generated. The function of hsPLA 2 will be therefore to inhibit coagulation by delaying the formation of a fully active prothrombinase complex, rather than to inhibit the activity of the fully active prothrombinase complex on prothrombin.
It is well established that sPLA 2 s exert different pharmacological effects (26). The presence of distinct "pharmacological sites," which are not implicated in the catalytic activity, may help to explain the diversity of these pharmacological effects (26). In particular, Kini and Evans (28) have proposed that the anticoagulant activity of some venom sPLA 2 s involves a basic region located between residues 50 and 80. The hsPLA 2 has seven basic amino acids (Lys or Arg) located between residues 51 and 74. Three-dimensional studies have shown that this part of hsPLA 2 is exposed at the surface of the molecule and is not associated with the catalytic site (5, 10). Thus, this region appears to be a good candidate for involvement in the inhibition of FXa by hsPLA 2 . To test this hypothesis, we examined the inhibitory effect of synthetic peptides comprising residues 51-74 on prothrombinase generation. The peptide 51-74 showed an inhibitory effect with an IC 50 of 8 M. This inhibition was specific, since a control scrambled peptide was much less potent in 0.1 M NaCl and completely devoid of detectable activity in 0.22 M NaCl. The moderate inhibitory effect of the scrambled peptide, observed at low ionic strength, was likely nonspecific due to its basic nature (net charge ϩ 6). Furthermore, the smaller peptide 51-62 was also able to inhibit prothrombinase, suggesting that the region of residues 51-62 of hsPLA 2 is involved in the inhibition of prothrombinase generation by interacting with FXa. Reversed peptide 62-51 and peptide D-51-62 were unable to inhibit prothrombinase activity up to 200 M, supporting the specificity of the interaction of peptide 51-62 with FXa.
The importance of the 51-62 region in the binding of hsPLA 2 on FXa points out the presence of basic clustered residues that might be critical for the interaction. In agreement, it has been shown that substitution of Lys 56 by Gln in hsPLA 2 reduced the antiprothrombinase activity of the enzyme, whereas substituting Asp 59 by Arg in porcine pancreatic sPLA 2 increased this activity (23). The thermodynamic characterization of the hsPLA 2 -FXa association reaction reported here also agrees with this conclusion. hsPLA 2 binding to FXa in solution presents a favorable but small binding enthalpy (⌬H 0 ϭ -4.56 kcal/mol) and a favorable and significant entropic contribution to binding (T⌬S ϭ ϩ4.90 kcal/mol). The small enthalpic contribution may account for the short amino acid sequence size at the binding site, whereas the favorable entropic contribution to In the sequence of hsPLA 2 (2), the amino acids of the active site are indicated in bold. The synthetic peptide 51-74 represents the basic region of the hsPLA 2 (residues 51-74) with the cysteine 59 replaced by a serine (indicated by an asterisk), to avoid disulfide formation between peptide molecules. For the two peptides, basic residues are shown in bold, and acidic residues are underlined. binding may reflect binding induced desolvation and/or anion release upon binding at the level of the basic sequence of residues 51-62 of hsPLA 2 . Heeb et al. (49) suggested that one binding site for FXa involves residues 493-506 in FVa (GLL-LICKSRSLDRR), which shows some similarity (bold letters) to the peptide 51-62 (YKRLEKRGSGTK) when basic residues are compared. This remarkable homology supports our hypothesis that hsPLA 2 acts by competing with FVa for binding to FXa. It is not at all clear which parts of the FXa molecule are targeted by FVa, and thus it is also very difficult to establish which FXa regions interact with hsPLA 2 . The most probable hypothesis is that hsPLA 2 binds to the same site as FVa on FXa, but we cannot exclude a noncompetitive mechanism due to the binding of clustered basic residues of hsPLA 2 to specific negatively charged residues present in ␥-carboxyglutamaterich domain of FXa. However, whatever the part of FXa that interacts with hsPLA 2 , this interaction does not affect the active site of FXa, since the enzymatic activity of FXa alone is not affected. Site-directed mutagenesis studies, as well as chemically synthesized hsPLA 2 variants (50), will help to analyze which amino acids of the region 51-62 are crucial for the anticoagulant activity of hsPLA 2 , and a three tridimensional structure of the FXa⅐hsPLA 2 complex will allow the identification of the interacting region of FXa.
In conclusion, our observation that hsPLA 2 binds to FXa and delays fully active prothrombinase generation apparently by preventing FVa/FXa interactions provides a molecular mechanism explaining the nonenzymatic anticoagulant effect of hsPLA 2 . Based on this work and on the purification and cloning of a protein receptor for sPLA 2 (51,52), it is clear that sPLA 2 exerts biochemical effects on protein targets through pharmacological sites that are distinct from the sPLA 2 -active site.

TABLE III
Inhibition of prothrombinase activity by selected peptides with sequences of the basic region of residues 51-74 of hsPLA 2 Prothrombinase activity assays were performed as described under "Experimental Procedures." Reactants (20 pM FXa, 120 pM FV, 5 M PL, and 5 mM Ca 2ϩ ) were preincubated in the presence of various concentrations of the indicated peptides. After 8-min preincubations, the prothrombinase activity was determined and expressed in percent of the prothrombinase activity measured in the absence of peptide. Then the percent of residual prothrombinase activity was analyzed as a function of the peptide concentration, and the IC 50