Role of the Gla and first epidermal growth factor-like domains of factor X in the prothrombinase and tissue factor-factor VIIa complexes.

Factor X (FX) has high structure homology with other proteins of blood coagulation such as factor IX (FIX) and factor VII (FVII). These proteins present at their amino-terminal extremity a gamma-carboxyglutamic acid containing domain (Gla domain), followed by two epidermal growth factor-like (EGF1 and EGF2) domains, an activation peptide, and a serine protease domain. After vascular damage, the tissue factor-FVIIa (TF-FVIIa) complex activates both FX and FIX. FXa interacts stoichiometrically with tissue pathway inhibitor (TFPI), regulating TF-FVIIa activity by forming the TF-FVIIa-TFPI-FXa quaternary complex. Conversely, FXa boosts coagulation by its association with its cofactor, factor Va (FVa). To investigate the contribution of the Gla and EGF1 domains of FX in these complexes, FX chimeras were produced in which FIX Gla and EGF1 domains substituted the corresponding domains of FX. The affinity of the two chimeras, FX/FIX(Gla) and FX/FIX(EGF1), for the TF-FVIIa complex was markedly reduced compared with that of wild-type-FX (wt-FX) independently of the presence of phospholipids. Furthermore, the association rate constants of preformed FX/FIX(Gla)-TFPI and FX/FIX(EGF1)-TFPI complexes with TF-FVIIa were, respectively, 10- and 5-fold slower than that of wt-FXa-TFPI complex. Finally, the apparent affinity of FVa was 2-fold higher for the chimeras than for wt-FX in the presence of phospholipids and equal in their absence. These data demonstrate that FX Gla and EGF1 domains contain residues, which interact with TF-FVIIa exosites contributing to the formation of the TF-FVIIa-FX and TF-FVIIa-TFPI-FXa complexes. On the opposite, FXa Gla and EGF1 domains are not directly involved in FVa binding.

Factor X (FX) has high structure homology with other proteins of blood coagulation such as factor IX (FIX) and factor VII (FVII). These proteins present at their aminoterminal extremity a ␥-carboxyglutamic acid containing domain (Gla domain), followed by two epidermal growth factor-like (EGF1 and EGF2) domains, an activation peptide, and a serine protease domain. After vascular damage, the tissue factor-FVIIa (TF-FVIIa) complex activates both FX and FIX. FXa interacts stoichiometrically with tissue pathway inhibitor (TFPI), regulating TF-FVIIa activity by forming the TF-FVIIa-TFPI-FXa quaternary complex. Conversely, FXa boosts coagulation by its association with its cofactor, factor Va (FVa). To investigate the contribution of the Gla and EGF1 domains of FX in these complexes, FX chimeras were produced in which FIX Gla and EGF1 domains substituted the corresponding domains of FX. The affinity of the two chimeras, FX/FIX(Gla) and FX/FIX(EGF1), for the TF-FVIIa complex was markedly reduced compared with that of wild-type-FX (wt-FX) independently of the presence of phospholipids. Furthermore, the association rate constants of preformed FX/FIX(Gla)-TFPI and FX/FIX(EGF1)-TFPI complexes with TF-FVIIa were, respectively, 10-and 5-fold slower than that of wt-FXa-TFPI complex. Finally, the apparent affinity of FVa was 2-fold higher for the chimeras than for wt-FX in the presence of phospholipids and equal in their absence. These data demonstrate that FX Gla and EGF1 domains contain residues, which interact with TF-FVIIa exosites contributing to the formation of the TF-FVIIa-FX and TF-FVIIa-TFPI-FXa complexes. On the opposite, FXa Gla and EGF1 domains are not directly involved in FVa binding.
The blood coagulation cascade consists of a series of enzymatic conversions driven by the formation of complexes between serine proteases and cell membrane-bound cofactors. Human factor X (FX) 1 is one of the serine protease zymogens playing a central role in coagulation processes leading to the formation of a fibrin clot. This is illustrated by the behavior of FX as a substrate or as an enzyme in three essential blood coagulation complexes. First, FX is a natural substrate, as well as factor IX (FIX), of the tissue factor-factor VIIa (TF-FVIIa) complex (1) considered as the initial enzyme complex in the cascade following vascular damage. FX activation by TF-FVIIa results from specific cleavage and release of a 52-residue activation peptide. Activated FX (FXa) can generate a tiny amount of thrombin from prothrombin in an extremely inefficient reaction (2). Tissue factor pathway inhibitor (TFPI) binds to TF-FVIIa-FXa to limit the production of FXa and FIXa by TF-FVIIa (3,4). Nevertheless, once produced, thrombin and the initially formed FXa activate small quantities of factor V (FV) to FVa and factor VIII (FVIII) to FVIIIa (5)(6)(7)(8). The activation of these two cofactors leads to the formation of two other essential procoagulant complexes, both involving FX, at the surface of procoagulant phospholipids in the presence of calcium ions (9), FIXa-FVIIIa and FXa-FVa complexes, which convert FX to FXa and prothrombin to thrombin, respectively. These complexes are 10 5 -10 6 -fold more active than the serine proteases devoid of their respective cofactors (10 -12) and promote the formation of a fibrin clot.
FX circulates in blood as a two-chain molecule and has the same modular structure as other vitamin K-dependent blood coagulation proteins such as FVII and FIX (13). The light chain has 11 amino-terminal glutamyl residues that are post-translationally modified in a vitamin K-dependent reaction to form a ␥-carboxyglutamic acid-containing domain or "Gla domain" critical for the binding of calcium ions and phospholipids (14). The Gla domain is followed by two domains homologous to the epidermal growth factor (EGF) precursor, considered important for protein-protein interactions (15). The heavy chain is joined to the light chain by a single disulfide bond and contains a 52-amino acid peptide and a trypsin-like serine protease domain (16), which forms the carboxyl-terminal end of the molecule. Although FXa has substantial sequence similarities and high homologous three-dimensional structure with FVIIa and FIXa (17)(18)(19), it displays significant differences in substrate specificity and catalytic activity. Furthermore, each protease requires a specific cofactor to express enhanced catalytic activity within the procoagulant complexes of blood coagulation.
It is clear that the catalytic specificities of blood coagulation proteases are supported by the carboxyl-terminal half of the enzymes, a trypsin-like serine protease domain (20). The prime role of the serine protease domain has also been demonstrated in the interaction of the blood coagulation proteases with their cofactors. For instance, site-directed mutagenesis revealed that the serine protease domain is the most important part of FIXa and FXa in the FIXa-FVIIIa and FXa-FVa interactions, respectively (21)(22)(23). However, there is also evidence that Gla-and EGF-like domains may also mediate protein-protein interactions and may consequently be implicated in the assembly of the protein complexes of blood coagulation. For instance, FVIIa multiple residues, distributed over the entire light chain, are in contact with TF (19). The identification of the FIXa light chain residues directly involved in the interaction with FVIIIa has to be performed, but three different approaches have suggested that the region around residues 85-90 in the linker area between EGF1 and EGF2 might contact residues 1804 -1818 of FVIIIa (24 -26). Concerning FX, there is evidence that the light chain of FXa contributes in the interaction of the enzyme with its cofactor (27,28). However, it is unknown whether the Gla and the first EGF domains interact directly with FVa in the prothrombin activation complex or whether they position the catalytic domain at a correct distance above the phospholipid membrane. Several investigations have suggested by studying the effect of mutations in TF that TF-FVIIa complex binds to the Gla domain of FX (29,30), and there is evidence that the first EGF-like domain of FX is required for the activation of the substrate by the TF-FVIIa complex (31). From all these analyses it has been suggested that the same residues from FX and FIX within the Gla and EGF1 domains are involved in these interactions.
In the present study, the contribution of the Gla and first EGF domains in FX-specific reconnaissance by the TF-FVIIa complex and in the prothrombin activation complex was addressed using chimeric recombinant FX containing either the Gla domain or the first EGF domain of FIX. There are two main reasons for the choice of FIX for exchanging FX homologous domains, the structural similarities between the two molecules and their respective cofactors, as well as the high similitude of the blood coagulation reactions in which they participate. The purified recombinant proteins were then compared with normal FX with regard to their activation by TF-FVIIa complex and to their properties as activated serine proteases in the TF-FVIIa-TFPI-FXa and FXa-FVa complexes.
Proteins-Mouse monoclonal anti-FX antibody KB-FX008 was prepared in our laboratory as described previously (32,33). KB-FX008 was found as being directed against FX protease domain by Western blot. Polyclonal antibodies against FX conjugated or not with horseradish peroxidase were obtained from Dako (Dakopatts, Glostrup, Denmark). Purified human plasma-derived FX (pd-FX), FX-activating enzyme from Russell's viper venom (RVV-X), bovine antithrombin III, human prothrombin, human thrombin, and recombinant human TF were obtained from Kordia (Leiden, The Netherlands). TFPI was obtained from American Diagnostica (Andresy, France). Human FVa was purchased from Hematologic Technologies, Inc. (Essex Junction, VT). High purity recombinant human FVIIa was from Novo Nordisk A/S (NovoSeven, Bagsvaerd, Denmark).
Protein Concentrations-ELISA using anti-FX polyclonal antibodies assayed pd-FX protein. FX was expressed in units, where 1 unit represents the amount in 1 ml of normal human plasma. Recombinant FX proteins and pd-FX were assayed by ELISA employing coated mouse monoclonal antibody KB-FX008. Bound FX proteins were detected using peroxidase-conjugated polyclonal FX antibodies. Proteins were quantified by the method of Bradford (34), using BSA as a standard. Molar concentrations of FXa, and ␣-thrombin were determined by active site titration (35).
Recombinant FX Derivatives-Plasmids encoding wt-FX, FX/FIX-(Gla), and FX/FIX(EGF1) have been described previously (36) as well as the domain borders of the chimeras. The FIX and FX domains are expressed by the chimera names, i.e. FX/FIX(Gla) contains the Gla domain and the hydrophobic stack of FIX and FX/FIX(EGF1) contains the EGF1 domain of FIX. All DNA constructions were expressed by Madin-Darby canine kidney cells maintained in cell factories with Dulbecco's modified Eagle's medium supplemented with 2,5% fetal calf serum, 100 units/ml penicillin, 100 g/ml streptomycin, 5 g/ml vitamin K 1 , and 1 g/ml amphotericin B/0.8 g/ml deoxycholate. FX containing medium was harvested every 48 h. Benzamidine was added to a final concentration of 10 mM, and medium was centrifuged (6,000 ϫ g), passed over cellulose acetate membranes (0.22 m) to eliminate cell debris, and stored at Ϫ80°C. Conditioned medium was thawed at 37°C. EDTA was added to a concentration of 5 mM. The medium was diluted to bring the final NaCl concentration to 60 mM. The mixture was then stirred at room temperature for 30 min with QAE-Sephadex A-50 beads to achieve a final concentration of 0.25% (w/v). Beads were washed before eluting with 50 mM Tris (pH 7.4), 500 mM NaCl, and 10 mM benzamidine. Recombinant FX (ELISA) contained in eluted fraction was then concentrated by precipitation with two successive steps of 40 and 70% saturated ammonium sulfate. Proteins obtained after the second precipitation step were diluted in 50 mM Tris (pH 7.4) and 100 mM NaCl and immediately dialyzed against 50 mM Tris (pH 7.4), and 100 mM NaCl, containing 10 mM benzamidine, and 5 mM EDTA. Insoluble proteins were discarded by centrifugation (6,000 ϫ g for 30 min). Recombinant FX was purified from the soluble proteins by immunoaffinity chromatography using monoclonal anti-human FX antibody KB-FX008. After adsorption, the antibody column was washed with 50 mM Tris (pH 7.4), 100 mM NaCl, containing 10 mM benzamidine followed by elution of bound recombinant FX with 0.1 M glycine (pH 2.5), and 10 mM benzamidine. The protein solution was immediately neutralized with 2 M Tris (pH 7.4). Recombinant FX containing fractions were combined and dialyzed against 50 mM Tris (pH 7.4), 100 mM NaCl, containing 10 mM benzamidine. If present, residual contaminants were removed by Q-Sepharose Fast Flow chromatography in 50 mM Tris (pH 7.4), 100 mM NaCl, containing 10 mM benzamidine, using a linear NaCl gradient (0.05-1 M) for elution. The eluate was dialyzed extensively against 50 mM Tris (pH 7.4), 100 mM NaCl and stored at Ϫ80°C. Before any analysis, a final pass over a benzamidine-Sepharose column equilibrated with 50 mM Tris (pH 7.4) and 100 mM NaCl was used to eliminate trace contaminants of FXa that may have been generated during production or purification of the recombinant protein.
Amino Acid Sequence Analysis-Purified recombinant FX derivatives were reduced and loaded onto a 15% SDS-polyacrylamide gel. The resolved proteins were transferred to an Immobilon membrane and stained with Ponceau S. The light chains were excised and sequenced using an Applied Biosystem Procise model 494 sequencer in the sequencing facility of the Institut de Biologie et Chimie des Protéines (Lyon, France).
Activation of Recombinant FX Derivatives by RVV-X-Recombinant FX derivatives and wt-FX were activated by RVV-X (37). RVV-X (1 mg) was coupled to 1 ml of CNBr-activated Sepharose 4B according to manufacturer's instructions. Recombinant FX derivatives (1 M) were incubated with coupled RVV-X (30 nM) in 50 mM Tris (pH 7.4), 100 mM NaCl containing 10 mM CaCl 2 . After 2 h, the reaction was stopped by addition of 15 mM EDTA. Activated recombinant FX derivatives were dialyzed against 50 mM Tris (pH 7.4) and then with 50 mM Tris (pH 7.4), 100 mM NaCl and loaded on a benzamidine-Sepharose column equilibrated in the same buffer. After washing, bound activated FX was eluted with 50 mM Tris (pH 7.4), 100 mM NaCl containing 5 mM benzamidine. Fractions containing FXa were pooled and precipitated by the addition of solid ammonium sulfate to 80% saturation. The precipitated proteins were diluted in 50 mM Tris (pH 7.4) and 100 mM NaCl, and the protein solution was immediately dialyzed against the same buffer containing 50% glycerol (v/v) and stored at Ϫ20°C until use. The active site concentrations of activated recombinant FX derivatives were determined by titration with known concentration of antithrombin in the presence of heparin, and an active site-specific assay using biotinyl-⑀aminocaproyl-D-glutamic acid glycylarginine chloromethyl ketone as described previously (28,38,39). Concentrations of activated recombinant FX derivatives and activated wt-FX were found to correlate between the two methods.
Plasma-based Coagulant Activities-The activity of non-activated derivatives of FX and wt-FX were functionally characterized in a prothrombin time assay as described previously (40) with slight modifications. Recombinant FX derivatives or wt-FX were preincubated 5 min at 37°C in a fibrometer with FX immunodepleted human plasma. Clotting was initiated by the addition of rabbit brain thromboplastin-C reagent and calculated from a standard curve generated with the clotting times versus the dilutions of pooled normal plasma. Recombinant FX derivatives and wt-FX clotting activities were expressed as a percentage of normal activity.
Relipidation of TF Apoprotein-Recombinant human TF (2.5 g) in 100 l of 50 mM Tris (pH 7.4), 100 mM NaCl containing 10 mM CHAPS was mixed with an equal volume of phospholipid vesicles preparation (PC/PS, 3:1, 2 mM). Phospholipid vesicles (PC/PS, 3:1) of nominal 200 nm diameter were synthesized by the method of membrane extrusion (41) in 10 mM Hepes (pH 7.5), 100 mM NaCl, 5 mM CaCl 2 . Phospholipid concentrations were determined by phosphate analysis. After a 1-h incubation at 37°C, the relipidated TF was dialyzed at 4°C against 50 mM Tris (pH 7.4), 100 mM NaCl. The preparation was kept under N 2 at 4°C and used within 2 weeks. The functional concentration of the TF preparation in these vesicles was considered to be 300 nM, half of its total concentration (42). Throughout this paper, unless specified, TF refers to this reconstituted preparation.
Amidolytic Activity-The steady-state kinetics of hydrolysis of S-2765 by pd-FXa, wt-FXa, and recombinant FXa derivatives were assayed in 50 mM Tris (pH 7.4), 100 mM NaCl, containing 2 mg/ml BSA, and 5 mM CaCl 2 . Kinetic parameters of substrate hydrolysis were determined employing an enzyme concentration of 2 nM and various substrate concentrations ranging from 0 to 5 mM. The release of paranitroanilide was monitored at 405 nm at 37°C in a kinetic microplate reader (Bio-Tek Instruments, Winooski, VT). The apparent K m and k cat values for substrate hydrolysis were calculated from the Michaelis-Menten equation, and the catalytic efficiencies were expressed as the ratio of k cat /K m .
Thrombin Formation-The rate at which activated recombinant FX derivatives, wt-FXa, or pd-FXa can activate prothrombin to thrombin in the presence of phospholipids as a function of FVa concentrations was compared as described previously with slight modifications (23). Briefly, phospholipid vesicles (PC/PS, 3:1, 30 M), and 20 pM FXa were incubated for 5 min with various concentrations of FVa (0 -1 nM). The reaction was started by the addition of 1 M prothrombin. The assay was performed in 50 mM Tris (pH 7.4), 100 mM NaCl, containing 5 mM CaCl 2 , and 0.2% (w/v) BSA at 37°C. Aliquots were taken at specified times, and the reaction was stopped in EDTA (10 mM final). Thrombin formation in each was then determined by measuring the amidolytic activity of the samples toward the synthetic substrate S-2238 (250 M), in the presence of 15 g/ml soybean trypsin inhibitor to inhibit amidase activity of FXa. During the assay, less than 5% of prothrombin was converted to thrombin, and thrombin formation was linear. Conversion of substrate was monitored at 405 nm. Concentrations of thrombin generated in the activation reactions were determined from a standard curve prepared from the cleavage rate of S-2238 by known concentrations of thrombin under the same conditions.
The comparison of the initial rates of prothrombin activation by FXa derivatives in the presence of subsaturating FVa concentration was performed as follow. In the absence of phospholipids, thrombin formation was initiated by the addition of 10 M prothrombin to a reaction mixture containing 0.5 nM FXa and 75 nM FVa in 50 mM Tris (pH 7.4), 100 mM NaCl, containing 5 mM CaCl 2 , and 0.2% (w/v) BSA at 37°C. In the presence of phospholipids, thrombin formation was initiated by the addition of 1 M prothrombin to a reaction mixture containing 20 pM FXa, 250 pM FVa, and 30 M phospholipids in 50 mM Tris (pH 7.4), 100 mM NaCl, containing 5 mM CaCl 2 , and 0.2% (w/v) BSA at 37°C. In all experiments, aliquots were removed from the reaction mixtures at specified times and diluted into 10 mM EDTA buffer to stop the reaction. Thrombin concentration was calculated as described above.
Activation of Recombinant FX Derivatives by FVIIa and Human TF-Activation of recombinant FX derivatives by FVIIa were measured and compared directly to wt-FX activation. The assay was performed in 50 mM Tris (pH 7.4), 100 mM NaCl, containing 0.2% (w/v) BSA, and 5 mM CaCl 2 at 37°C. Initial rates of activation by FVIIa and relipidated human TF were determined as described previously with little modification (43). Briefly, 60 pM FVIIa was added to 60 pM relipidated TF in the presence of 30 M phospholipid vesicles (PC/PS, 3:1). The mixture was incubated at 37°C for 20 min in order to establish a FVIIa-TF complex, and the reaction was started by the addition of various concentrations of FX (0 -3 M). After diverse incubation times, aliquots were taken, diluted in stop solution containing EDTA (10 mM final), and assayed for FXa formation employing the synthetic substrate S-2765 (250 M). Conversion of substrate was monitored at 405 nm. The concentration of FXa generated in the activation reaction was determined from a standard curve prepared from the cleavage rate of S-2765 by known concentrations of active site titrated FXa under the exact same conditions.
Comparison of the initial rates of FXa derivatives activation by TF-FVIIa was performed as follow. In the absence of phospholipids, the reaction was initiated by the addition of 250 nM FX to a reaction mixture containing 50 nM FVIIa and 250 nM TF solubilized at 37°C with 5 mM CHAPS in 50 mM Tris (pH 7.4), 100 mM NaCl, containing 5 mM CaCl 2 , and 0.2% (w/v) BSA. The mixture was incubated at 37°C for 5 min prior to the addition of the substrate to establish a FVIIa-TF complex. In the presence of phospholipids, the reaction was initiated at 37°C by addition of 1 M FX to a mixture containing 8 nM FVIIa and 0.5 nM TF in the presence of 1 mM phospholipids in 50 mM Tris (pH 7.4), 100 mM NaCl, containing 5 mM CaCl 2 , and 0.2% (w/v) BSA. As in the absence of phospholipids, the mixture was incubated at 37°C for 5 min prior to the addition of the substrate to establish a FVIIa-TF complex. In all experiments, aliquots were removed from the reaction mixture at specified times and diluted into 10 mM EDTA buffer to stop the reaction. FXa concentration was calculated as described above.
Inhibition of FX Activation by TFPI-FXa-Inhibition of pd-FX activation by TFPI-recombinant FXa derivatives was measured as described previously (44,45) with minor modifications. Briefly, 25 pM of relipidated TF was incubated with 8 nM FVIIa for 5 min. FXa generation was started by the addition of 400 nM pd-FX followed by immediate addition of preformed TFPI-recombinant FXa derivatives (25, 50, and 100 pM). After various incubation times, aliquots were removed, diluted in a stop solution containing EDTA (10 mM final), and assayed for FXa formation using the synthetic substrate S-2765 (500 M). FXa was then quantified as described above. The association and dissociation rate constants for the inhibition of TFPI-recombinant FXa derivatives of TF-FVIIa were estimated by fitting experimental data on FX activation by TF-FVIIa to Equation 1, where [FXa] max is the maximal concentration of FXa; [FXa] t is FXa concentration at a given time point t, and k obs is the observed first order rate constant. The apparent second order association rate constant for TFPI-recombinant FXa derivatives in the inhibition of the TF-FVIIa complex and the dissociation rate constant were determined as the slope and the intercept of the y axis, respectively, of the plot k obs versus TFPI-recombinant FXa derivatives concentration (44).

RESULTS
Recombinant Proteins-As reported previously, the employed expression system produces fully processed recombinant zymogen of the coagulation system (46) with normal Ca 2ϩdependent properties (47). Indeed, NH 2 -terminal sequence analysis revealed the expected sequence of the mature purified recombinant proteins, ANSFLEEMKK for wt-FX and FX/FIX-(EGF1) and YNSGKLEEFV for FX/FIX(Gla), indicating that the signal sequence and the propeptide have been accurately and efficiently removed before secretion. In addition, sequence analysis disclosed that the average yield for the two glutamic acid residues at the NH 2 terminus was less than 5% of the average yield of the two subsequent residues. Because ␥-carboxylation reduces the yield of Glu residues, these data demonstrate that the glutamic residues were appropriately modified. All recombinant FX proteins were activated by RVV-X under conditions similar to those of pd-FX. All FX derivatives could be completely converted into active form, and the final activated preparations were more than 90% active as determined by active site titration.
Plasma-based Assay of Function-The clotting activity of wt-FX, FX/FIX(Gla), and FX/FIX(EGF1) was estimated in a prothrombin time assay. Recombinant FX derivatives were incubated with FX immunodepleted human plasma and clotting initiated by addition of rabbit brain thromboplastin. Clotting activity of wt-FX was 75 Ϯ 10% of the activity of pooled normal plasma. In contrast, FX/FIX(Gla) and FX/ FIX(EGF1) clotting activities were less than 1 and 6 Ϯ 2% of the activity of pooled normal plasma, respectively. Because this global measurement of the dysfunction of the chimera could be due to abnormal activation and/or catalytic activity, these enzymatic steps were studied separately using purified components.
Amidolytic Activity-To explore whether substitutions of the Gla and EGF1 domains of FX by the corresponding domains of FIX affect the amidolytic activity of FXa, hydrolysis of various concentrations of synthetic substrate S-2765 was monitored as described under "Experimental Procedures." As shown in Table  I, all chimeras display similar rates of substrate hydrolysis compared with that of wt-FXa or pd-FXa. This indicates that Gla or EGF1 substitution does not adversely affect the reactivity of the catalytic triad in the activated FX chimeras.
Prothrombin Activation-As the chimeric FX variants displayed a normal reactivity of the catalytic triad, it was of interest to investigate the influence of the Gla and EGF1 domain substitutions in prothrombin activation by the FXa-FVa complex. Therefore, FXa chimeras were compared with wt-and pd-FXa in their ability to activate prothrombin. The rate of thrombin generation was studied as a function of increasing concentrations of FVa and in the presence of an excess of phospholipids. For all FXa tested, thrombin formation was dependent on FVa concentrations and was saturable (Fig. 1). The apparent K d values were 420 Ϯ 38, 405 Ϯ 26, 230 Ϯ 11, and 225 Ϯ 11 pM for pd-FXa, wt-FXa, FX/FIX(Gla), and FX/FIX-(EGF1), respectively, whereas the apparent k cat values were between 15.5 and 19.9 s Ϫ1 . These experiments show that activated chimeras have full catalytic activity toward prothrombin in the prothrombinase complex. In addition, substitution of the Gla or EGF1 domains by the corresponding domains of FIX even has a positive effect on the apparent affinity of FVa. To evaluate whether the slight increased affinity of FVa for the chimeras was related to the presence of phospholipids, the initial rates of thrombin formation by the chimeras was com-pared with that of wt-FXa in the presence of subsaturating concentrations of FVa with phospholipids ( Fig. 2A) or without phospholipids (Fig. 2B). The same initial rates of prothrombin activation were obtained for the chimeras and wt-FXa in the absence of phospholipids (Fig. 2B). A similar rate of activation was obtained with pd-FXa (data not shown). Thus, the increased affinity of FVa observed in the experiment depicted in Fig. 1 was due to the presence of phospholipids. In conclusion, substitution of the Gla or EGF1 domains by the corresponding domains of FIX has no effect on the apparent affinity of FVa for FXa, and even a slight increase of affinity was observed in the presence of phospholipids.
FX Activation by TF-FVIIa Complex-FX is one natural substrate of the TF-FVIIa complex, which is considered as the initial enzyme complex in the cascade following vascular damage. Because several studies have implicated the Gla domain (28,48) and the first EGF domain (31) of FX in the interaction with TF-FVIIa, FX chimeras were compared with wt-FX for their ability to serve as substrates for the complex in the presence of phospholipid vesicles and calcium ions. The apparent k cat of the TF-FVIIa complex toward all chimeras was slightly decreased compared with that of wt-FX (Table II). Conversely, the apparent affinity of FX/FIX(Gla) and FX/FIX-(EGF1) for the enzymatic complex was greatly decreased (Table II). These data indicate that substitution of the Gla or the EGF1 domains by the corresponding domains of FIX impairs the binding capacity of FX to the TF-FVIIa. Because Gla domains of FIX and FX bind to phospholipids which provide a surface for the assembly of the FVIIa-TF-substrate complex, it is possible that the substitutions of the Gla and EGF1 domains alter the alignment between the substrate and TF-FVIIa complex. This possibility was tested by using detergent-solubilized TF in an assay comparing the initial rates of chimeras and wt-FX activation by TF-FVIIa. Similarly to the phospholipidscontaining system (Fig. 3A), activation rates of the chimeras were impaired in the presence of solubilized TF (Fig. 3B). The initial rate was 3.5 Ϯ 1.1% for FX/FIX(Gla) and 10.8 Ϯ 3.2% for FX/FIX(EGF1) when compared with the rate obtained for wt-FX (Fig. 3). Wt-FX was activated at a similar rate of pd-FX (data not shown).
Inhibition of the TF-FVIIa Complex by TFPI-FXa-Because the chimeric variants FX/FIX(Gla) and FX/FIX(EGF1) displayed a reduced affinity for TF-FVIIa, kinetics of TF-FVIIa inhibition by variable concentrations of preformed TFPI-FXa chimeras was investigated. Conditions of a limited amount of  relipidated TF (25 pM) and excess of FVIIa (8 nM) were optimal for TF-FVIIa complex formation. TF-FVIIa complex was incubated with substrate pd-FX (400 nM) and preformed TFPI-FXa variants (25-100 pM). The results of inhibition of TF-FVIIa-dependent pd-FXa by increasing concentrations of TFPI-FXa complexes are presented in Fig. 4. The progress curves show a similar increasing inhibition of TF-FVIIa by increasing concentrations of TFPI-pd-FX (Fig. 4A) and of TFPI-wt-FX (Fig. 4B). Conversely, a significantly reduced inhibition was observed in the presence of the same concentrations of TFPI-FX/FIX(Gla) (Fig. 4C) or TFPI-FX/FIX(EGF1) (Fig. 4D). From the linear fit of k obs versus TFPI-FXa concentrations, apparent second order rate association constants and dissociation rate constants were determined and reported in Table III. No significant differences were observed between all the dissociation rate constants. Conversely, the rate association constants of the TFPI-FX/FIX(Gla) and FX/FIX(EGF1) complexes with TF-FVIIa were about 5-10-fold reduced compared with that of the TFPIwt-FXa. Therefore, substitution of the Gla or the EGF1 domains by the corresponding domains of FIX reduced the capacity of FXa, in complex with TFPI, to inhibit TF-FVIIa activity toward FX by reducing the capacity of the inhibitor complex to bind to the enzymatic complex. DISCUSSION In this study, the role of the Gla and the first EGF-like domains of FX has been explored. To this end, FX chimeras containing the Gla or the EGF1 domains of FIX substituting the corresponding domains of FX were produced. Their properties, regarding different functions of FX, were compared with those of normal FX.
In the prothrombinase complex FVa displays an apparent affinity for FX/FIX-Gla and FX/FIX-EGF1 chimeras ϳ2-fold higher than for normal FXa (Fig. 1). One possibility is that Gla and EGF1 domains of FXa interact with FVa. Therefore, the apparent increased affinity of FVa for the FX chimeras is due to the presence of appropriate binding sites for the cofactor on the Gla and EGF1 domains of FIX. So far, no interactions have been described between FVa and FIXa. A second possibility is that FVa does not interact with the Gla and EGF1 domains of FXa. Hence, in FX the Gla and EGF1 domains would act as spacers between phospholipids and the rest of the molecule to match the cofactor binding sites. Therefore, the presence of the Gla or EGF1 domains of FIX modifies the inclination toward the phospholipid surface of the EGF2 and protease domains of the chimeras compared with that of normal FXa. This mechanism is supported by experimental data showing that in the absence of phospholipids the initial rate of thrombin formation by the chimeras associated to FVa is identical to those observed with normal FXa (Fig. 2B). A similar spacer role has been described previously for FIX in studies using FIX mutated in the EGF1 domain or in which the corresponding domain of protein C substituted the EGF1 domain. The mutated FIX molecule displays a reduced interaction with FVIIIa in the presence of phospholipids but not in its absence (21). Observation that the FX chimeras displayed a normal catalytic activity in the presence of FVa (Fig. 2B) demonstrates that the FVabinding sites on the chimeras are identical to those of normal FXa. Thus, because no interactions have been described between FVa and FIXa, these data imply that the Gla and the EGF1 domains of FXa do not contain FVa-binding sites and are only involved in the interaction with the cofactor in the presence of phospholipids to conform the enzyme toward FVa.
The FX chimeras containing the Gla or the EGF1 domains of FIX are different from normal FX as substrate for the TF-FVIIa complex in the presence of phospholipids (Table II). The results indicate that substitution of one of the two amino-terminal regions of FX by the corresponding region of FIX affects the apparent affinity of the substrate for the enzymatic complex. Therefore, these two regions are involved in the formation of the TF-FVIIa-FX complex, and it can be suggested that within this ternary complex both domains of FX are directly or indirectly implicated in the binding to TF-FVIIa. The Gla domain involvement in the interaction with TF-FVIIa exosites is supported by several studies. For instance, by studying the effect of mutations in the carboxyl-terminal domain of TF, an interaction of this membrane-proximal region with the Gla domain of the substrates FIX and FX has been proposed (29,30,49). Recently, it has been revealed (31) that FIX and FX interact with TF through, in part, their EGF1 domains. It is noteworthy that the decreased affinity of the FX/FIX(Gla) and FX/FIX-(EGF1) compared with normal FX for the TF-FVIIa complex in the presence of phospholipids is associated with a normal k cat (Table II). These data can be integrated to the kinetic model for FX activation by TF-FVIIa proposed by Krishnaswamy and co-workers (50). In their model, substrate recognition by the TF-FVIIa complex is achieved through two sequential steps. Initial interactions between TF-FVIIa exosites and complementary sites on the substrate remote from structures surrounding the scissile bond are followed by an intramolecular binding step that allows the cleavage of adjacent structures to dock with the active site of the enzyme prior to bond cleavage. In the present study, it is observed that in the absence of phospholipids, the initial rates of activation of the chimeras were markedly decreased compared with the activation of wt-FX (Fig. 3). As with the previous data in presence of phospholipids, the defective activation in their absence is due to defective interaction of the chimeras with exosites of the TF-FVIIa complex. Taken together, these results suggest a direct role of the Gla and EGF1 domains in the interaction with the TF-FVIIa. Thus, despite sequence and structural homologies of the Gla and the EGF1 domains of FIX and FX and the probable involvement of the same residues to interact with TF-FVIIa, the severe defective interactions of the chimeras with TF-FVIIa, in the presence or absence of phospholipids (Figs. 2 and 3), demonstrate that these domains cannot be exchanged without altering the substrate affinity for the enzymatic complex. Two possibilities arise from these observations. First, the defective interaction of the chimeras with the enzymatic complex could be due to the involvement of different residues between FIX and FX domains with TF-FVIIa. However, a docking approach that proposed a model for the ternary complex TF-FVIIa-FIX has revealed that residues of FIX interacting with TF-FVIIa are rather conserved in FX (51). A second possibility is that the substitutions of Gla or EGF1 domains introduce an alteration of the favorable orientation of the substrate toward the enzymatic complex. The same structural modification is likely the cause of a slight increase of the apparent affinity of FVa for the chimeras in the presence of phospholipids (Fig. 1).
Kinetics of TF-FVIIa inhibition by variable concentrations of preformed TFPI-FXa show that the substitution of the Gla or the EGF1 domains markedly reduces the association constants of the FT-FVIIa-TFPI-FXa quaternary complexes (Table III). This suggests that the defective interaction of TFPI associated with the activated chimeras with TF-FVIIa is probably due to an alteration of the favorable orientation of the inhibitor complex toward the enzymatic complex. These data indicate that similar interactions contribute to the assembly of FX and FXa, after complex formation with TFPI, to the TF-FVIIa complex. This notion is in agreement with a previous study (52) showing that TF residues, which are important for the activation of FX by the TF-FVIIa complex, are required for the accelerated inhibition of the TF-FVIIa complex by TFPI mediated by FXa. These TF residues have been proposed to interact with the Gla domain of the substrates FIX and FX (29,30,49). It has been shown previously (53) that FXa devoid of the Gla domain is not able to support the TFPI-mediated inhibition of TF-FVIIa, suggesting the importance of FXa Gla domain in the formation of the TF-FVIIa-TFPI-FXa complex. However, proteolytic fragments of FX-containing multiple domains are not always reliable for the identification of binding sites because it has been observed that EGF domains must be covalently attached to the Gla domain of FX to maintain their properties (53). Therefore, the approach using FX/FIX chimeras carrying FIX regions is more appropriate to identify the role of the Gla and EGF1 domains.
In conclusion, this study demonstrates that the Gla and first EGF-like domains of FX are not directly involved in the interaction of FXa with FVa. Experimental data indicate that the FX Gla domain interacts with the FT-FVIIa complex. Moreover, this domain, together with the first EGF-like domain, is directly involved in the association of FXa after complex for- with the TF-FVIIa complex Association rate (k ϩ1 ) and dissociation rate (k Ϫ1 ) constants are calculated from the kinetic (time course) data of one representative experiment depicted in Fig. 4  mation with TFPI to the TF-FVIIa complex. This study also reveals that in a blood coagulation protein a loss of function can be counterbalanced by the gain of another one. Thus, a mutation within a FX molecule, which is responsible for a reduced rate of activation by the TF-FVIIa complex, could be compensated for by its beneficial effect on the activated molecule in the presence of its cofactor. There are other models in nature confirming this paradox. Therefore, it can be suggested that an individual with no bleeding disorders could possess such mutations, which would only be detected during coagulation investigation.