The Gla Domain of Factor IXa Binds to Factor VIIIa in the Tenase Complex

During blood coagulation factor IXa binds to factor VIIIa on phospholipid membranes to form an enzymatic complex, the tenase complex. To test whether there is a protein-protein contact site between the (cid:1) -carboxyglu-tamic acid (Gla) domain of factor IXa and factor VIIIa, we demonstrated that an antibody to the Gla domain of factor IXa inhibited factor VIIIa-dependent factor IXa activity, suggesting an interaction of the factor IXa Gla domain with factor VIIIa. To study this interaction, we synthesized three analogs of the factor IXa Gla domain (FIX 1–47 ) with Phe-9, Phe-25, or Val-46 replaced, respec- tively, with benzoylphenylalanine (BPA), a photoactivatable cross-linking reagent. These factor IX Gla domain analogs maintain native tertiary structure, as demonstrated by calcium-induced fluorescence quenching and phospholipid binding studies. In the absence of phospholipid membranes, FIX 1–47 was able to inhibit factor IXa activity. This inhibition is dependent on the presence of factor VIIIa, suggesting a contact site between the factor IXa Gla domain and factor VIIIa. To demonstrate a direct interaction we did cross-linking experiments with FIX 1–47 9BPA, FIX 1–47 25BPA, and FIX 1–47 46BPA. Covalent cross-linking

Factor IX is composed of a single polypeptide comprising an N-terminal ␥-carboxyglutamic acid (Gla 1 )-rich domain, an aromatic amino acid stack domain, two epidermal growth factor domains, and a C-terminal serine protease domain. The Gla domain contains 12 ␥-carboxyglutamic acid residues a posttranslationally modified glutamic acid residue. ␥-Carboxyla-tion occurs in the liver where the vitamin K-dependent carboxylase converts glutamic acid to ␥-carboxyglutamic acid in an enzymatic reaction that requires vitamin K (1). In the presence of Ca 2ϩ , the Gla domain folds to become a membrane-binding structure (2,3). In the absence of metal ions the Gla domain is largely unfolded and does not bind to membranes. Mg 2ϩ ions, in the absence of Ca 2ϩ , support folding of residues 12-47 of factor IX but not folding of the membrane-interactive omega loop included in residues 1-11 (4), whereas Ca 2ϩ ions support folding of the membrane-binding omega loop and are necessary for function of factor IXa (5). The metal ion-specific conformational changes have provided the basis for developing antibodies that recognize the Ca 2ϩ -dependent 11-amino acid membrane binding region (6).
Factor VIIIa is a cofactor for the factor IXa-mediated cleavage of factor X to factor Xa. Factor VIII is synthesized as a single polypeptide chain composed of three A domains (A1, A2, and A3), one B domain, and two C domains (C1 and C2) in the sequence, A1-A2-B-A3-C1-C2 (7). The A domains are homologous to ceruloplasmin, and the C domains are homologous to discoidin I, a phospholipid binding lectin (8), and to a murine milk fat globule membrane protein (9). The B domain has no known homology to any other protein. Factor VIII is variably cleaved, prior to its secretion, between the B and A3 domains to yield a heterodimeric protein composed of a variably sized heavy chain (A1-A2-B) and a light chain (A3-C1-C2). Thrombin activates factor VIII to factor VIIIa via several cleavages within both the heavy and light chains to yield a heterotrimeric protein (A1, A2, A3-C1-C2) that is non-covalently associated. The C2 domain of factor VIII/VIIIa contains the major membrane binding motif (10,11). Properties of the membrane-binding interaction include high affinity (12), high specificity for membrane binding sites that include phosphatidylserine (13), and stereoselective interaction with phosphatidyl-L-serine (14).
Factor IX is proteolytically activated to factor IXa, a serine protease. Factor VIII is activated to its cofactor form, factor VIIIa, via a proteolytic mechanism. Factor IXa and factor VIIIa assemble on phosphatidyl-L-serine-containing phospholipid membranes in the presence of Ca 2ϩ to form an enzymatic complex known as the tenase complex (2,3). The tenase complex converts factor X to factor Xa, the enzyme that converts prothrombin to thrombin leading to the conversion of fibrinogen to fibrin and the formation of a fibrin clot. Although much is known about the structures of both factor IXa and factor VIIIa from x-ray crystallography and NMR spectroscopy (5,(15)(16)(17), details about the interaction of factor IXa with factor VIIIa have been limited to the analysis of naturally occurring mutations in hemophilic patients or site-specific mutagenesis studies (18 -21). Analyses of factor IXa-factor VIIIa binding have revealed domain interactions between the serine protease domain of factor IXa and the A2 domain of factor VIIIa as well as the first epidermal growth factor domain of factor IXa and the A3 domain of factor VIIIa (22). Furthermore, a naturally occurring mutation of the Gla domain of factor IX leads to loss of factor VIIIa-mediated tenase activity (23). In this report, we use a synthetic analog of the factor IX Gla domain, FIX 1-47 ( Fig. 1), which we modify by replacing a specific hydrophobic amino acid with the photoactivatable amino acid, benzoylphenylalanine. With these Gla domain peptides, we provide direct evidence using a cross-linking strategy for the binding of the factor IX Gla domain to the light chain of factor VIIIa.

EXPERIMENTAL PROCEDURES
Reagents-Factor IXa and factor X were purchased from Hematologic Technologies Inc. Recombinant human factor VIII formulated in sucrose (Kogenate FS) was received as a gift from Bayer Canada and activated to factor VIIIa as described previously (24). Bovine brain phosphatidylserine (PS), egg yolk phosphatidylcholine (PC), and dansylated phosphatidylethanolamine (PE) were obtained from Avanti Polar Lipids. Chromogenic substrate S-2765 was purchased from Diapharma Group Inc. ESH8 antibodies were purchased from American Diagnostica. Antibodies to the Gla domain of factor IXa were prepared as previously described (6). An antibody to ␥-carboxyglutamic acid was a gift from Dr. Johan Stenflo (25). The FIX 1-47 BPA analogues were synthesized as previously described (4).
Enzymatic Assays-The activation of factor X by factor IXa was measured with a two-step amidolytic substrate assay. For inhibition experiments with the FIX 1-47 BPA peptides or with anti-FIX CaCl 2specific antibodies, factor IXa (20 nM) was incubated with 500 nM factor X, 4 nM factor VIIIa, and the specified concentrations of the FIX 1-47 peptides or anti-FIX CaCl 2 -specific antibodies in 150 mM NaCl, 20 mM HEPES, pH 7.4, 5 mM CaCl 2 , and 0.1% bovine serum albumin for 60 min at 25°C. The reaction was then stopped with 150 mM NaCl, 20 mM HEPES, pH 7.4, 5 mM EDTA. The amount of factor Xa generated was determined immediately using the chromogenic substrate S-2765 (0.3125 mg/ml) using a Molecular Devices enzyme-linked immunosorbent assay plate reader. Experiments conducted without factor VIIIa were incubated for 120 min. Measurements were carried out in duplicate, and IC 50 values for FIX 1-47 BPA peptide inhibition of factor IXa activity in the presence of factor VIIIa were determined by non-linear least-square fitting of the values of factor Xa generation versus the logarithm of FIX 1-47 peptide concentration using SigmaPlot 8.0 for Windows (SPSS).
Preparation of Phospholipid Vesicles-Small unilamellar phospholipid vesicles (PC:PS, 75:25) were prepared by the method of Barenholz (26). Phospholipids in chloroform were dried at 45°C under N 2 , washed three times with methylene chloride, resuspended in 150 mM NaCl, 50 mM Tris-HCl, pH 7.4, and sonicated in a bath sonicator until the solution cleared. The suspension of phospholipid vesicles was centrifuged, first at 160,000 ϫ g for 30 min and then at 250,000 ϫ g for 90 min in a Beckman L8 -80M ultracentrifuge using a Ti-70.1 rotor. The supernatant contained the small unilamellar phospholipid vesicles. Phospholipid concentrations were determined by phosphorus analysis (27).
Fluorescence Measurements-The fluorescence energy transfer experiments and fluorescence quenching experiments were performed as previously described (28). Dissociation constants (K D ) were calculated as described previously (29).
Cross-linking Experiments-Covalent cross-linking of BPA-labeled FIX 1-47 peptides to factor VIIIa was performed through photoactivation of p-benzoyl-L-phenylalanine substituted for specific hydrophobic residues in FIX  . Each of the three peptides (FIX 1-47 9BPA, FIX 1-47 25BPA, or FIX 1-47 46BPA), at a concentration of 1 M, was mixed with factor VIIIa (100 nM) in 50 mM Tris-HCl, pH 7.4, 50 mM NaCl, and 5 mM CaCl 2 . The samples were irradiated with a B-100 AP ultraviolet lamp (UVP) for 20 min at a distance of 5.5 cm. Non-covalent Ca 2ϩ -mediated protein-protein binding was reversed by the addition of SDS and EDTA and heating at 37°C for 5 min. The samples were run on a 7.5% SDS-PAGE gel and then transferred to a polyvinylidene difluoride membrane. The membrane was probed with a primary mouse monoclonal antibody to ␥-carboxyglutamic acid, followed by the addition of a secondary antibody, sheep anti-mouse IgG horseradish peroxidase-linked antibody. The blots were developed with an ECL reagent (Amersham Biosciences, UK) according to the manufacturer's protocol. The cross-linking protocol was carried out for each peptide in the presence and absence factor VIIIa, and calcium. In addition, crosslinking experiments were carried out in the presence of increasing concentrations of factor IXa.
Immunoprecipitation-Immunoprecipitation experiments were carried out with ESH8, a monoclonal antibody to the C2 domain of factor VIIIa. After irradiation and photo-induced cross-linking, the samples were immunoprecipitated with ESH8, subjected to SDS-PAGE, and analyzed by Western blotting as described for the cross-linking experiments.

RESULTS
The tenase complex, assembled on phospholipid membranes, is composed of the enzyme, factor IXa, and the cofactor, factor VIIIa. To test the hypothesis that the factor IXa Gla domain might bind to factor VIIIa, inhibition of factor IXa-catalyzed activation of factor X was performed in the presence or absence of factor VIIIa using an antibody to the calcium-stabilized conformation of factor IXa (anti-FIX⅐Ca(II)-specific antibodies). The epitope recognized by this antibody is targeted against residues 1-11 within the Gla domain of factor IXa. Because anti-FIX⅐Ca(II)-specific antibodies inhibit factor IX binding to phospholipid membranes (6), phospholipid membranes were omitted from this assay system. In the absence of phospholipid membranes, factor IXa can catalyze the conversion of factor X to factor Xa albeit at a significantly reduced rate. Anti-FIX⅐Ca(II)-specific antibodies inhibited the generation of factor Xa by factor IXa, if factor VIIIa was present ( Fig. 2A). In the absence of factor VIIIa, there was no inhibition of the conversion of factor X to factor Xa by factor IXa (Fig. 2B). These results are consistent with antibody-mediated inhibition of the interaction of the Gla domain of factor IXa with factor VIIIa.
A peptide comprising the Gla and aromatic amino acid stack domains of factor IXa, FIX  , was also used to study factor IXa Gla domain-factor VIIIa interactions. FIX  has been previously demonstrated to be a good model for studying the biochemical properties of the factor IXa Gla domain (28). To demonstrate direct binding between the Gla domain of factor IXa and factor VIIIa, analogs of FIX 1-47 containing benzoylphenylalanine (BPA) were synthesized. BPA forms covalent bonds with adjacent structures within a 3 Å radius when irradiated at 350 nm (30). It has been previously used to identify the residues that constitute a phospholipid binding site in factor IXa (4). Three analogs were chosen based upon the positioning of the BPA in the Gla domain, and each had different amino acids replaced with BPA (Fig. 1). Phe-9 and Val-46 were replaced, respectively, in two peptides (FIX 1-47 9BPA and FIX 1-47 46BPA) and have been previously characterized (4). A third peptide was prepared with Phe-25 replaced by BPA. All three peptides bind calcium similarly to unsubstituted FIX 1-47 (28), suggesting that the BPA substitution does not disrupt tertiary structure (Fig. 3A). Fifty percent fluorescence quenching occurred at 210, 100, and 317 M CaCl 2 for FIX 1-47 9BPA, FIX 1-47 25BPA, and FIX 1-47 46BPA, respectively, similar to 50 M CaCl 2 seen with FIX 1-47 (28), indicating that they all undergo the calcium-induced fluorescence transition. The ability of the BPA-containing FIX 1-47 peptides to inhibit the conversion of factor X to factor Xa by factor IXa in the presence and absence of factor VIIIa was determined in the absence of phospholipid membranes. A representative experiment, depicted in Fig. 4, shows inhibition of factor IXa-catalyzed activation of factor X by FIX 1-47 25BPA in the presence (Fig. 4A) but not in the absence of factor VIIIa (Fig. 4B). These results suggest that FIX 1-47 25BPA can bind directly to a specific site in factor VIIIa, preventing factor VIIIa from binding to factor IXa and serving as a cofactor in the activation of factor X to factor Xa. The IC 50 of this inhibition was 3.8 Ϯ 0.35 M. The IC 50 of the other peptides were generally similar as shown in Table I. Although somewhat higher activity was observed in the presence of very low concentrations of the FIX 1-47 25BPA peptide, as shown in the representative experiment depicted in Fig. 4A (ϳ10% on average; four experiments), the basis for this effect is unclear. It could reflect some stabilization of the normally very labile factor VIIIa during the 1-h assay period, because it was not seen in the absence of factor VIIIa (Fig. 4B). This observation notwithstanding, the overall effect of this peptide was to prevent tenase complex formation by inhibiting the binding of factor VIIIa to factor IXa To demonstrate direct binding between the factor IXa Gla domain and factor VIIIa and to identify the region on FIX 1-47 that is near or in the contact site that binds to factor VIIIa, cross-linking experiments were performed to determine which FIX 1-47 BPA species would bind covalently to factor VIIIa. The photoactivatable peptides, FIX 1-47 9BPA, FIX 1-47 25BPA, or FIX 1-47 46BPA, were incubated with factor VIIIa in the presence of CaCl 2 . After irradiation, the protein mixtures were analyzed by Western blot analysis using an anti-␥-carboxyglutamic acid antibody. As shown in Fig. 5, FIX 1-47 25BPA was covalently cross-linked to factor VIIIa, whereas reaction mixtures containing FIX 1-47 9BPA showed no cross-linking. There is a faint band in the FIX 1-47 46BPA cross-linking experiment suggesting cross-linking is present albeit to a very minor degree. Furthermore, cross-linking was observed only when the reaction mixture included FIX 1-47 25BPA or FIX 1-47 46BPA peptides, calcium ions, and factor VIIIa and when the reaction was irradiated at 350 nm. The complex of FIX 1-47 25BPA coupled to factor VIIIa migrated on SDS gels in a position just above that of the light chain of factor VIIIa, suggesting a covalent complex between FIX 1-47 25BPA and the light chain of factor VIIIa. To substantiate this result, we immunoprecipitated the cross-linked material with ESH8, monoclonal antibody to the C2 domain of factor VIIIa. The immunoprecipitate was run on an SDS-PAGE and Western blotted with an anti-Gla antibody. This result, shown in Fig. 6, substantiates the conclusion that the region on factor VIIIa to which FIX 1-47 25BPA binds is the light chain (A3-C1-C2 domain).
To establish the specificity of the cross-linking of FIX 1-47 25BPA to the light chain of factor VIIIa, full-length factor IXa was employed at different concentrations to inhibit cross-linking. As shown in Fig. 7, the addition of increasing amounts of factor IXa led to complete inhibition of the crosslinking of FIX 1-47 25BPA to factor VIIIa. DISCUSSION In the present work, we demonstrate that the Gla domain of factor IXa interacts with the light chain of factor VIIIa. A   FIG. 4. Inhibition of the catalytic function of the factor VIIIafactor IXa complex but not factor IXa alone by FIX 1-47 25BPA. A, factor IXa (20 nM) and factor X (500 nM) were incubated with factor VIIIa (4 nM) (q) for 60 min at room temperature in 5 mM CaCl 2 and the indicated concentrations of FIX 1-47 25BPA. The line of best fit was calculated with a non-linear least-square fit regression curve using SigmaPlot 8.0 for Windows (SPSS). B, factor IXa (20 nM) and factor X (500 nM) were incubated without factor VIIIa (E) for 120 min at room temperature in 5 mM CaCl 2 and the indicated concentrations of FIX 1-47 25BPA. Shown is a representative experiment of four performed in duplicate.   peptides of the factor IXa-factor VIIIa complex in the absence of phospholipid membranes For each peptide, the IC 50 values were calculated for FIX 1-47 BPA peptide inhibition of factor IXa activity in the presence of factor VIIIa but absence of phospholipid membranes from a non-linear least-square analysis of data plotted on a semi-log plot of concentration of peptide versus factor Xa generation (cf. Fig. 4A  schematic model of this complex, assembled on phospholipid membranes, is presented in Fig. 8. Two contact sites on factor IXa have previously been implicated in this interaction. Using fluorescence anisotropy, Fay et al. demonstrated a contact site between the A2 domain of factor VIIIa and the serine protease domain of factor IXa. This interaction was characterized by an affinity of ϳ300 nM (31)(32)(33). Another contact site of higher affinity (ϳ15 nM) has been identified between the A3 domain of factor VIIIa and the first epidermal growth factor domain of factor IXa (34). In the current study, we show that the Gla domain of factor IXa and light chain of factor VIIIa, both phospholipid-binding structures, also bind to each other.
Because factor VIIa is homologous with factor IXa, the contact sites through which tissue factor activates factor VIIa provide useful comparisons to the contact sites through which factor VIIIa activates factor IXa. On-off rates using both fluorescence spectroscopy and surface plasmon resonance demonstrate three interactive sites between the enzyme, factor VIIa, and its cofactor, tissue factor. A site of interaction between the Gla domain of factor VIIa and tissue factor is one site of interaction analogous to that identified in the present work. This site is considered to be of lower affinity in comparison to the other sites of interaction. The similar properties that we describe for the factor VIIIa-factor IXa Gla domain complex are consistent with that report (35). Fluorescent labeling of the involved residues in tissue factor that contact factor VIIa confirm that these sites become protected from solvent upon binding factor VIIa (36).
Two independent lines of evidence suggest that the contact between the light chain of factor VIIIa and the factor IXa Gla domain is physiologically important. First, a naturally occurring mutation in the Gla domain of factor IXa leading to he-mophilia B affect function only in the presence of factor VIIIa. This propisitus, with a substitution of glycine 12 by arginine, has a severe bleeding phenotype (23). Second, factor IXa-factor VIIIa interactions can be inhibited by a peptide comprising the Gla domain of factor IXa, FIX  .
Our data suggest an interaction between the C-terminal half of the Gla domain of factor IXa and the light chain of factor VIIIa. In the folded Gla domain Phe-25 is located at the distal end from the omega loop and phospholipid binding site and not too distant from residue 46. The side chains of Phe-25 and Val-46 project from the backbone of the Gla domain at an angle greater than 90 degrees from one another. Nevertheless, it is possible that the side chains of both residues interact with the light chain of factor VIIIa, although from the cross-linking data Phe-25 may interact more effectively than Val-46 (5). The failure of FIX 1-47 9BPA to cross-link to factor VIIIa suggests that the face of the omega loop, including Phe-9, does not contact factor VIIIa; in the presence of membranes it is buried in the phospholipid bilayer. 2 In this study, we identify a region of the factor IXa Gla domain containing Phe-25 and Val-46 as a contact site with factor VIIIa. These contact sites are localized to the C-terminal portion of the Gla domain, which is similar to two homologous serine proteases, factor VIIa and protein C. The crystal structure of the factor VIIa-tissue factor complex demonstrates a direct interaction between the C-terminal region of the Gla domain of factor VIIa and tissue factor (37). In the case of the 2 M. Shenone, B. Furie, and B. C. Furie, unpublished data.
FIG. 6. Immunoabsorption of the covalent complex with a factor VIII light chain antibody. Immunoprecipitation was performed with ESH8, a monoclonal antibody to the C2 domain of factor VIIIa. Following FIX 1-47 25BPA covalent cross-linking to factor VIIIa in the presence of phospholipid membranes, the samples were incubated with ESH8 or a control IgG antibody coupled to protein G-Sepharose. The material bound to the beads was then subjected to 7.5% SDS-PAGE and probed with an antibody to ␥-carboxyglutamic acid.  (5). Residues Phe-9, Phe-25, and Val-46, sites of BPA substitution, are shown in yellow, red, and blue, respectively. B, schematic diagram of factor IXa-factor VIIIa interactions based on previously identified contact sites (32,34). The putative factor IXa Gla domain and factor VIIIa domain interaction is also shown (*3). protein C-protein S complex, replacement of the Gla domain of protein C with that of prothrombin renders protein C activity independent of protein S, its cofactor. This protein S dependence localizes to residues 23-46 of the Gla domain of protein C (38).
In conclusion, our studies demonstrate a specific interaction between the Gla domain of factor IXa and the light chain of factor VIIIa. Given that both the Gla domain of factor IXa and the C2 domain of factor VIIIa are membrane-binding structures, we envision these regions bound to membranes and to each other. Whether the occupancy of this contact site promotes allosteric activation of the factor VIIIa-factor IXa complex on phospholipid membranes remains to be determined.