Hydrophobic Contact between the Two Epidermal Growth Factor-like Domains of Blood Coagulation Factor IX Contributes to Enzymatic Activity*

The three-dimensional structure of activated factor IX comprises multiple contacts between the two epidermal growth factor (EGF)-like domains. One of these is a salt bridge between Glu78 and Arg94, which is essential for binding of factor IXa to its cofactor factor VIII and for factor VIII-dependent factor X activation (Christophe, O. D., Lenting, P. J., Kolkman, J. A., Brownlee, G. G., and Mertens, K. (1998) J. Biol. Chem. 273, 222–227). We now addressed the putative hydrophobic contact at the interface between the EGF-like domains. Recombinant factor IX chimeras were constructed in which hydrophobic regions Phe75-Phe77 and Lys106-Val108 were replaced by the corresponding sites of factor X and factor VII. Activated factor IX/factor X chimeras were indistinguishable from normal factor IXa with respect to factor IXa enzymatic activity. In contrast, factor IXa75–77/factor VII displayed ∼2-fold increased factor X activation in the presence of factor VIII, suggesting that residues 75–77 contribute to cofactor-dependent factor X activation. Activation of factor X by factor IX106–108/factor VII was strongly decreased, both in the absence and presence of factor VIII. Activity could be restored by simultaneous substitution of the hydrophobic sites in both EGF-like domains for factor VII residues. These data suggest that factor IXa enzymatic activity requires hydrophobic contact between the two EGF-like domains.

enriched in ␥-carboxyglutamic acid residues, followed by a short hydrophobic stack and two epidermal growth factor (EGF)-like domains (denoted as EGF1 and EGF2 domain). The heavy chain comprises the protease domain with the catalytic center (3).
FIXa functions as activator of the zymogen factor X (FX). Activation of FX by FIXa is enhanced by several orders of magnitude in the presence of phospholipid membrane, Ca 2ϩ ions, and the cofactor, activated factor VIII (FVIIIa) (2,6,7). Factor VIII (FVIII) is a heterodimer consisting of a light chain and a heavy chain. FVIII is activated upon cleavage of the heavy and light chain by thrombin or activated FX (FXa) (8). The activated forms of FVIII and FIX are known to assemble into the FX-activating complex (for review see Refs. 9 -11). In the FIXa protease domain, regions 301-303 and 333-339 have been identified as potential binding sites for FVIIIa (12,13). Most likely, these sites are involved in the binding of the FVIII heavy chain (10).
The crystal structure of porcine FIXa revealed that the relative orientation of the EGF-like domains is fixed by a variety of contacts between residues located at the interface between the EGF1 and EGF2 domain (14). The EGF-like domains define an angle of 110° (14), and this particular orientation may be of importance for FIXa function. This view is supported by the observation that disruption of a salt bridge between Glu 78 in the EGF1 domain and Arg 94 in the EGF2 domain is associated with strongly reduced FVIIIa-dependent FX activation by FIXa and decreased affinity of FIXa for FVIII light chain (15). These results provide support to the idea that the specific orientation of both EGF-like domains is a prerequisite for proper interaction between FIXa and FVIIIa and that residues near Glu 78 and Arg 94 may be involved in a direct interaction with FVIIIa. The structure of the interface between the EGF-like domains of porcine FIXa is further characterized by the presence of an exposed hydrophobic site in the EGF1 domain, comprising residues Val 75 -Phe 77 , that is captured in a hydrophobic pocket formed by multiple residues, including Val 107 and Cys 109 in the EGF2 domain (14).
In this study, we focused on the functional role of the regions Phe 75 -Gly 76 -Phe 77 and Lys 106 -Val 107 -Val 108 in human FIX. In our approach, we constructed recombinant FIX mutants in which these regions were replaced for corresponding regions of homologous serine proteases FX and FVII. Our results demonstrate that substitution of the FIX residues predominantly affects FIXa enzymatic activity and the extent of stimulation by its natural cofactor, FVIIIa. Our data support a model in which residues 75, 76, 107, and 108 form a hydrophobic contact between the EGF1 and EGF2 domain that is essential for FIXa function.
Construction of Recombinant FIX-The plasmid encoding wild-type FIX (wt-FIX) has been described previously (16). Site-directed mutagenesis was performed using the plasmid encoding wt-FIX as a template to construct plasmids coding for FIX 75-77 /FX, FIX 106 -108 /FX, FIX 75-77 /FVII, FIX 106 -108 /FVII, and FIX-Phe 77 3 Ser. Plasmids were constructed using the Overlap Extension polymerase chain reaction mutagenesis method (17). Plasmids encoding FIX 75-77 /FX and FIX 75-77 / FVII were used as template in the polymerase chain reactions for construction of plasmids encoding FIX 75-77, 106 -108 /FX and FIX 75-77, 106 -108 /FVII, respectively. Sequence analysis was performed to verify the presence of the mutations in the plasmids.
Proteins-Monoclonal antibody CLB-FIX 14 and polyclonal antibodies against FIX have been described (18). Antibodies were purified employing protein A-Sepharose as recommended by the manufacturer. FIX, FX, and prothrombin were purified from a concentrate of human prothrombin, FIX, and FX (7,19) and converted into their active forms (FIXa, FXa, and thrombin, respectively) as described (19,20). Human FVIII was purified as described previously (20). FVIII was activated by thrombin (molar ratio 30:1) for 10 min, and activated FVIIIa was purified by CM-Sepharose according to a previously described method (21). Purified FVIIIa preparations were obtained with a specific procoagulant activity of Ͼ100,000 units/mg, as determined in the one-stage clotting assay (22). Final FVIIIa preparations were stored at Ϫ80°C.
Expression and Culturing of FIX Variants-Recombinant FIX was expressed in Madine-Darby canine kidney cells as described previously (15). Cells were transfected with plasmid DNA employing the calcium precipitation method as outlined previously (16) and grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin. After 2 days, Geneticin (800 g/ml) was added to the medium for selection of transfected cells. After 2 weeks, individual clones were isolated and grown in selective medium to obtain stable cell lines. Production of FIX was monitored by measuring FIX antigen by enzyme-linked immunosorbent assay as described previously (20). Stable cell lines producing appropriate amounts of FIX were selected and used for large scale production of FIX in cell factories (6320 cm 2 ). Medium (Dulbecco's modified Eagle's medium, supplemented with 2.5% fetal calf serum, 100 units/ml penicillin, 5 g/ml vitamin K 1 , 100 g/ml streptomycin, 1 g/ml amphotericin B, and 0.8 g/ml desoxycholate) containing secreted FIX was harvested three times a week and stored at Ϫ20°C.
Purification of Recombinant FIX Variants-Cultured medium was filtered through a 0.22 m membrane (Plasmaflux P1, Fresenius, Bad Homburg, Germany) to remove cell debris. wt-FIX and FIX mutants were purified by immunoaffinity chromatography using monoclonal antibody CLB-FIX 14 according to an established procedure (12,15,23). After purification, one single FIX band for all FIX variants was present after SDS-polyacrylamide gel electrophoresis followed by Coomassie Brilliant Blue staining. FIX mutants displayed the same electrophoretic mobility as purified plasma-derived FIX, indicating that no propeptide-containing unprocessed FIX was present. Recombinant FIX produced by this expression system is indistinguishable from plasmaderived FIX with regard to binding to barium citrate and to the Ca 2ϩdependent monoclonal antibody CLB-FIX 2 (23). Furthermore, Ca 2ϩdependent activity of wt-FIXa toward a synthetic substrate and toward FX is similar to that of plasma-derived factor IXa (15,16,23). FIX was converted into FIXa by incubation of 1.4 M FIX with 7 nM FXIa in 100 mM NaCl, 5 mM CaCl 2 , and 50 mM Tris (pH 7.4) for 2 h. Activation was stopped by the addition of EDTA (10 mM) and benzamidine (10 mM). FIXa was loaded on Q-Sepharose and washed with 150 mM NaCl, 5 mM benzamidine, and 50 mM Tris (pH 7.4) to remove FXIa. FIXa was eluted from the column by the addition of 500 mM NaCl, 5 mM benzamidine, and 50 mM Tris (pH 7.4). FIXa preparations were dialyzed against 100 mM NaCl, 5% glycerol, 50 mM Tris (pH 7.4), and subsequently against the same buffer containing 50% glycerol. Final FIXa preparations were stored at Ϫ20°C.
Protein Concentrations-The amount of protein was quantified by the method of Bradford (24) using human serum albumin as a standard. FVIII activity was assayed by a spectrophotometric assay employing bovine coagulation factors and a chromogenic substrate specific for FXa (Coatest FVIII, Chromogenix, Mölndal, Sweden). One unit of FVIII activity was assumed to correspond to 0.35 nM and represents the amount of FVIII activity present in 1 ml of pooled normal plasma. FIXa concentrations were determined by active site titration employing antithrombin as the active site titrant and CH 3 SO 2 -LGR-pNA as chromogenic substrate (20). Plots of the FIXa activity versus the antithrombin concentration were linear for all FIXa variants.
Amidolytic Activity-Hydrolysis of CH 3 SO 2 -LGR-pNA was assayed as described previously (23). Briefly, 50 l of a 5 mM solution of CH 3 SO 2 -LGR-pNA was added to 50 l of a solution containing 300 nM FIXa and various concentrations of CaCl 2 in a microtiter plate (Corning Costar, Badhoevedorp, The Netherlands). Initial rates of substrate hydrolysis were measured by monitoring the absorbance at 405 nm in time. Absorbance values were converted into molar concentrations using a molar extinction coefficient of 9.65 ϫ 10 3 M Ϫ1 cm Ϫ1 for p-nitroanilide and a path length of 0.35 cm for a 100-l volume.
FX Activation-In the absence of FVIIIa, FX activation was assayed as described previously (15) with the exception that Pefachrome Xa was used as substrate. In experiments using FVIII (0 -1.5 nM), unactivated FVIII was activated by 5 nM thrombin in 100 mM NaCl, 2 mM CaCl 2 , 0.2 mg/ml ovalbumin, and 50 mM Tris (pH 7.4) for 0.5 min at 37°C. Preliminary experiments revealed that maximal activity of FVIIIa was reached after 0.5 min of activation by thrombin. The reaction was stopped by the addition of hirudin (10 units/ml), and FVIIIa was added to siliconized tubes containing preincubated phospholipid vesicles (0.1 mM, L-␣-phosphatidyl-L-serine:L-␣-phosphatidylcholine, 50:50), CaCl 2 (5 mM), FIXa (0.1 nM), 100 mM NaCl, 0.2 mg/ml ovalbumin, and 50 mM Tris (pH 7.4). The reaction was initiated by the addition of FX (0.2 M). The amount of FXa generated was quantified as described previously (25). The same procedure was followed for assessing FX activation in the absence of phospholipids, except FVIIIa, FIXa, and FX were increased and FXa formation was allowed to proceed for 30 min before terminating the reaction with EDTA.

RESULTS
Recombinant FIX Variants-In the present study, we have investigated the functional role of hydrophobic residues that are located at the interface between the EGF1 and EGF2 domain in the FIXa light chain. To this end, recombinant FIX variants were constructed in which residues were replaced by corresponding residues of FX and FVII. Substitutions were based upon alignment of the human FIX and human FVII sequences and the position of the hydrophobic residues in a three-dimensional structure of human FIX based on the crystal structure of porcine FIXa (14) (Fig. 1). This analysis revealed that in the EGF1 domain a hydrophobic loop, comprising residues Phe 75 , Gly 76 , and Phe 77 , is positioned opposite of hydrophobic residues Val 107 , Val 108 , and Lys 106 in the EGF2 domain. Two FIX chimeras were constructed in which the separate sites in the EGF1 domain and the EGF2 domain were replaced by corresponding residues of FX, whereas a third chimera comprised a combination of these replacements. The same procedure was used for construction of FIX/FVII chimeras. In addition to these chimeric variants, one FIX variant was constructed in which Phe 77 was substituted for Ser 77 . Residue Phe 77 is conserved both in FX and FVII, and Phe 77 3 Ser mutation has been reported to be associated with mild hemophilia B (28). Recombinant FIX variants and wt-FIX were expressed in Madin-Darby canine kidney cells and purified from conditioned medium by immunoaffinity chromatography (see "Experimental Procedures"). All mutants were activated by FXIa under conditions similar to those for wt-FIX and plasma-derived FIXa. All FIX chimeras could be completely converted into the active form, FIXa (data not shown). The concentration of active FIXa in the purified preparations was determined by active site titration using antithrombin.
Amidolytic Activity-The EGF1 domain of FIXa contains a high affinity Ca 2ϩ binding site. Binding of Ca 2ϩ to this site has been demonstrated to modulate activity of FIXa toward a small synthetic substrate (23). To investigate the potential role of the amino acid replacements in the FIXa light chain on amidolytic activity, FIXa variants were tested employing the synthetic substrate CH 3 -SO 2 -LGR-pNA. All FIXa chimeras tested were indistinguishable from plasma-derived FIXa with respect to Ca 2ϩ -dependent stimulation of substrate hydrolysis, displaying rates of substrate hydrolysis between 2.5 and 2.7 M pnitroanilide/min in the presence of 5 mM CaCl 2 . Apparently, the various amino acid substitutions in the EGF1 and EGF2 domain leave FIXa amidolytic activity unaffected.
FX Activation in the Absence of FVIIIa-Subsequently, enzymatic activity of the FIXa chimeras toward the macromolecular substrate FX was explored. Activation of FX by FIXa was investigated in the presence of phospholipids and Ca 2ϩ ions but in the absence of FVIIIa. Most FIXa chimeras displayed normal (FIXa/FX chimeras, FIXa 75-77 /FVII) or slightly reduced (FIXa-Phe 77 3 Ser) FX activation ( Fig. 2 and Table I). Apparently, substitution of residues Phe 75 , Gly 76 , and Lys 106 for corresponding residues of FX and FVII does not affect FIXa enzymatic activity toward FX, whereas substitution of Phe 77 3 Ser has only a minor effect. In contrast, FX activation by FIXa 106 -108 /FVII was strongly reduced. This was mainly because of a decrease in the apparent catalytic rate constant (k cat ) resulting in a 10-fold reduction in the apparent catalytic efficiency (k cat / K m ) compared with that of wt-FIXa (Table I). In view of the severe defect of FIXa 106 -108 /FX, it is remarkable that activation of FX by FIXa 75-77, 106 -108 /FX proved only a minor reduction compared with normal FIXa (Fig. 2 and Table I). Apparently, the deleterious effect caused by substitution of Val 107 3 Arg and Val 108 3 Ser in the EGF2 domain is compensated for by simultaneous substitution of Phe 75 3 Pro and Gly 76 3 Ala in the EGF1 domain.
FX Activation in the Presence of FVIIIa-The contribution of FVIIIa as cofactor for FIXa to the activation of FX was assessed in a kinetic system containing phospholipids, Ca 2ϩ ions, FIXa, FX, and various concentrations of FVIIIa. For wt-FIXa, FX activation was enhanced by FVIIIa in a dose-dependent manner (Fig. 3A). FIXa/FX substitution variants were identical to wt-FIXa (Fig. 3B). In contrast, FIXa/FVII chimeras and FIXa-Phe 77 3 Ser displayed different cofactor-dependent FX activation. One chimera, FIXa 75-77 /FVII, displayed an increased rate of FX activation compared with wt-FIXa (Fig. 3A). The enhanced activity is FVIII-dependent, because no elevated activation of FX was observed in the absence of FVIIIa (Fig. 2). Reduced activity was observed for FIXa-Phe 77 3 Ser and the other two FIXa/FVII chimeras. Like in the absence of FVIIIa (Fig. 2), FIXa 75-77, 106 -108 /FVII activity was increased compared with that of FIXa 106 -108 /FVII. These data suggest that a functional interaction exists between residues 75 and 76 in the EGF1 domain, and residues 107 and 108 in the EGF2 domain, which is important for regulation of FIXa activity both in the presence and absence of the cofactor FVIIIa.
To examine catalytic parameters of FX activation in the presence of FVIIIa, various concentrations of FX were incubated with FIXa variants in the presence of FVIIIa and phospholipids. In this experiment, FIXa/FX chimeras were not studied in detail, because these chimeras were identical to wt-FIXa with respect to FVIII-dependent FX activation (Fig. 3B). FIXa 75-77 /FVII displayed about 2-fold increased activity compared with wt-FIXa (Fig. 4). The elevation in apparent k cat (Table I) (14). Porcine FIX residues Val 75 and Leu 108 were replaced by human FIX residues Phe 75 and Val 108 , and the human structure was obtained by molecular modeling employing the Swiss-Model Automated Comparative Protein Modeling Server (26,27). Part of the EGF1 domain is shown at the bottom right of the figure and is followed by the EGF2 domain. The backbone of FIXa is represented in ribbon format. Lys 106 , Val 107 , and Val 108 in the EGF2 domain are positioned opposite of the hydrophobic site in EGF1 comprising residues Phe 75 , Gly 76 , and Phe 77 . Below the sequence of human FX and human FVII is aligned to the human FIX sequence. FX and FVII amino acid residues, which have been substituted into the recombinant FIX variants, are indicated in bold. tion in the presence of phospholipid vesicles, whereas such differences may not occur in the absence of phospholipids (13). We therefore tested whether the FIXa/FVII chimeras and FIXa-Phe 77 3 Ser display the same discrepancy with wt-FIXa in the absence of phospholipids. Because of the low conversion rate of FX in the absence of phospholipids, relatively high concentrations of FX, FVIIIa, and FIXa were employed. Under these conditions, FIXa 75-77 /FVII stimulated FX activation to a higher extent compared with wt-FIXa (Fig. 5). The calculated apparent K d for the interaction between FVIIIa and FIXa 75-77 / FVII was 2-fold lower than the apparent K d for the interaction with wt-FIXa. FIXa-Phe 77 3 Ser, FIXa 106 -108 /FVII, and FIXa 75-77, 106 -108 /FVII displayed less effective stimulation by FVIIIa. For all FIXa variants the extent of FX activation compared with wt-FIXa was similar to that observed in the presence of phospholipids. These results demonstrate that alterations in FIXa variants are because of modified enzymatic properties and not to differences in phospholipid-dependent enzyme-cofactor or enzyme-substrate complex assembly. DISCUSSION The serine proteases of the blood coagulation process display a high level of structural similarity (2,29). Most of these enzymes comprise a protease domain containing the specific Ser/Asp/His catalytic triad and a light chain containing two or more EGF-like domains. EGF-like domains have been demonstrated to be important for a number of functional properties, including Ca 2ϩ binding and protein-protein interactions (30 -32). Structural data derived from NMR and crystallographic studies have shed light on the three-dimensional structure of these domains and the orientation thereof in the complete protein (14,(31)(32)(33)(34)(35). The crystal structure of porcine FIXa revealed that multiple interdomain contacts exist between the  EGF1 and EGF2 domain (14). In the present study, we focused on hydrophobic contacts between the two EGF-like domains employing recombinant FIX/X and FIX/VII chimeras (Fig. 1). All of the FIXa/FX chimeras we analyzed (Table I) displayed normal enzymatic activity compared with wt-FIXa, both in the absence and presence of FVIIIa (Figs. 2 and 3B). It seems relevant to note that amino acid replacements in the FIXa/FX chimeras were limited to residues 75 and 106, because residues 76, 77, 107, and 108 are identical in FX (see Fig. 1). The observation that substitution of residues 75 and 106 is not accompanied by major changes in enzymatic activity suggests that these residues are not essential for FIXa function. The fact that in the FIXa/FX chimeras the majority of the hydrophobic residues remains unaffected further implies that residues in position 76, 77, 107, and 108 may be more important in maintaining the hydrophobic contact between the two EGF-like domains.
Studies using other chimeric FIXa variants have demonstrated that the complete EGF1 domain can be replaced by that of FX without apparent effect on FIX function (36). In contrast, introducing the entire EGF1 domain from FVII leads to increased activity in the presence of FVIIIa, which presumably is because of an increased affinity for this cofactor (37). In our study, we observed a rate enhancement for FIXa 75-77 /FVII similar to that reported for FIXa EGF1 /FVII (Fig. 3A). This suggests that the effect of EGF1 domain replacement could be fully explained by the introduction of Pro 75 and Ala 76 in the FIX molecule. In this respect, our data are compatible with the suggestion made by Chang et al. (37) that elimination of the Phe 75 side chain may increase flexibility between the EGF1 and EGF2 domains, which might facilitate FVIIIa binding. An alternative explanation would be that the sequence Pro 75 -Ala 76 -Phe 77 in FIXa 75-77 /FVII directly binds to FVIIIa. However, this seems less likely because the presence of the same residues in the FIXa 75-77, 106 -108 /FVII chimera is not associated with increased response to FVIIIa (Fig. 3A). Our finding that cofactor stimulation is reduced in FIXa-Phe 77 3 Ser further supports the role of hydrophobic residues Phe 75 -Gly 76 -Phe 77 in regulation of FVIIIa-dependent FIXa activity (Fig. 3).
Whereas the functional role of residues 75-77 is relatively minor, mutation of residues 106 -108 proved to be of major impact on FIXa function. The chimera FIXa 106 -108 /FVII displayed dramatically reduced activity toward FX, both in the absence (Fig. 2) and presence (Fig. 3A) of FVIIIa. Because the defect was not only cofactor-independent but also phospholipidindependent (Fig. 5), it is evident that introduction of FVII residues at positions 107 and 108 in the EGF2 domain does not affect assembly of the FX-activating complex, but that the defect is because of a decrease in enzymatic activity. This suggests that a functional link exists between the EGF2 domain and the FIXa heavy chain. An explanation for the apparent "cross-talk" between the FIXa heavy and light chain, may be derived from the current three-dimensional FIXa structures and the atomic contacts between the two chains therein. Indeed, these structures reveal that there is an intimate contact between the EGF2 domain and the heavy chain (14,38). In this respect FIXa is similar to FXa, wherein the EGF2 domain and the protease domain may be regarded as a single operational unit (39). Although residues 107 and 108 are not in direct contact with the FIXa heavy chain, it is conceivable that substitution of these light chain residues may alter the interdomain interaction via an allosteric mechanism. These structural alterations in FIXa light chain-heavy chain contact may affect the arrangement of substrate recognition and cleavage sites in the heavy chain. Analysis of the interface between light and heavy light chain reveals multiple interdomain interactions including the disulfide bridge between Cys 132 and Cys 289 , a salt bridge between Glu 113 and Lys 409 , hydrophobic contact between Phe 98 and residues Tyr 295 , Phe 299 , and Phe 302 (14,38), and a potential hydrogen bond between Asn 92 and Tyr 295 . Interestingly, mutation of Asn 92 3 His in FIX is associated with reduced FIXa enzymatic activity (40), which also might support our idea that contact between the EGF2 domain and the heavy chain contributes to enzymatic activity. In view of the assumption that residues 106 -108 may regulate enzymatic activity via contact between light and heavy chain, it is particularly striking that the detrimental effect of replacing residues 106 -108 in the EGF2 domain is counteracted by the replacement of the complementary hydrophobic site 75-77. Our finding that the FIXa 75-77, 106 -108 /FVII chimera has apparently normal enzymatic activity (Figs. 2 and 3A) suggests that these sites comprise a functional link between the two EGF-like domains in FIXa. As such the hydrophobic contact, which has been described as a "ball-and-socket" joint in the FIXa light chain (14), is the counterpart of the salt bridge between Glu 78 and Arg 94 , which also links the EGF-like domains (15). The difference between these contacts, however, is that the salt bridge supports FVIIIa binding, whereas the hydrophobic contact primarily seems to regulate FIXa enzymatic activity.
It is further remarkable that substitution of the FIX sequence Phe 75 -Gly 76 -Phe 77 for the FVII sequence Pro 75 -Ala 76 -Phe 77 had no effect on enzymatic activity in the absence of FVIIIa (Fig. 2). This might imply that the EGF1 residues 75-77 may not be involved in regulation of enzymatic activity and that only residues in the EGF2 domain (e.g. residues 92, 107-108) contribute to FIXa enzymatic activity. However, it should be noted that amino acid replacements in FIXa 75-77 /FVII still maintain the hydrophobic character of this region. It seems conceivable therefore, that these FVII residues also are able to support the hydrophobic interaction with the EGF2 domain and consequently remain without significant effect on the overall structure of the protease domain of this FIXa chimera. The fact that the FIXa 75-77, 106 -108 /FVII chimera displays apparently normal activity further implies that the proteolytic activity is influenced by structure elements in the EGF1 domain. A  Fig. 3 and were found to be 30 (Ϯ 2), 15 (Ϯ 1), and 64 (Ϯ 7) nM for wt-FIXa, FIXa 75-77 /FVII, and FIXa 75-77, 106 -108 / FVII, respectively. For FIXa-Phe 77 3 Ser and FIXa 106 -108 /FVII the apparent K d was Ͼ200 nM. similar effect has been observed for FIXa variants with mutations in position 64 in the EGF1 domain (23). Mutation at this site eliminates the high affinity Ca 2ϩ binding site in this light chain domain and is associated with a variety of molecular defects including enzymatic activity toward both FX and a small synthetic peptide substrate. The role of the light chain in regulation of FIXa activity is further demonstrated by the observation that FIXa amidolytic activity could be completely inhibited by a monoclonal antibody against the light chain of FIXa (20). Although the proposed effects of the amino acid substitutions on the FIXa structure can only be confirmed by crystallographic data, we propose that the structural integrity of the interface between both EGF-like domains serves as an allosteric hinge, which regulates proper orientation of light and heavy chain within the FIXa molecule.