Activation and Active Site Occupation Alter Conformation in the Region of the First Epidermal Growth Factor-like Domain of Human Factor VII*

The first epidermal growth factor-like domain (EGF-1) of factor VII (FVII) provides the region of greatest contact during the interaction of FVIIa with tissue factor. To understand this interaction better, the conformation-sensitive FVII EGF-1-specific monoclonal antibody (mAb) 231-7 was used to investigate the conformational effects occurring in this region upon both FVII activation and active site occupation. The binding affinity of mAb 231-7 was approximately 3-fold greater for the zymogen state than for the active state; a result affected by the presence of both calcium and the adjacent Gla domain. Once activated, active site inhibition of FVIIa with a variety of chloromethyl ketone inhibitors resulted in a 10-fold range of affinities of FVIIai molecules to mAb 231-7. Gla domain removal eliminated this variation in affinity, suggesting the involvement of a Gla/EGF-1 interaction in this conformational effect. In addition, the binding of mAb 231-7 to FVIIa EGF-1 stimulated the amidolytic activity of free FVIIa. Taken together, these results imply an allosteric interaction between the FVIIa active site and the EGF-1 domain that is sensitive to variation in active site occupant structure. Thus, these present studies indicate that the conformational change associated with FVII activation and active site occupation involves the EGF-1 domain and suggest potential functional consequences of these changes.

The first epidermal growth factor-like domain (EGF-1) of factor VII (FVII) provides the region of greatest contact during the interaction of FVIIa with tissue factor. To understand this interaction better, the conformation-sensitive FVII EGF-1-specific monoclonal antibody (mAb) 231-7 was used to investigate the conformational effects occurring in this region upon both FVII activation and active site occupation. The binding affinity of mAb 231-7 was approximately 3-fold greater for the zymogen state than for the active state; a result affected by the presence of both calcium and the adjacent Gla domain. Once activated, active site inhibition of FVIIa with a variety of chloromethyl ketone inhibitors resulted in a 10-fold range of affinities of FVIIai molecules to mAb 231-7. Gla domain removal eliminated this variation in affinity, suggesting the involvement of a Gla/ EGF-1 interaction in this conformational effect. In addition, the binding of mAb 231-7 to FVIIa EGF-1 stimulated the amidolytic activity of free FVIIa. Taken together, these results imply an allosteric interaction between the FVIIa active site and the EGF-1 domain that is sensitive to variation in active site occupant structure. Thus, these present studies indicate that the conformational change associated with FVII activation and active site occupation involves the EGF-1 domain and suggest potential functional consequences of these changes.
Blood coagulation is initiated when circulating FVII 1 in plasma binds to its essential cofactor, the trans-membrane lipoprotein receptor TF. Upon binding to cell surface TF, zymogen FVII can be activated to FVIIa by factor IXa (FIXa) (1), factor Xa (FXa) (2), or in an autocatalytic manner by endogenous FVIIa (3), thus allowing propagation of the coagulation cascade. Activation of FVII occurs upon proteolytic cleavage of the Arg 152 -Ile 153 bond, giving rise to a 152-amino acid light chain (ϳ20 kDa) linked by a disulfide bridge to a 254-amino acid heavy chain (ϳ30 kDa) (4). It is the light chain of FVII that contains the first epidermal growth factor-like domain (EGF-1), a stretch of 37 amino acids with characteristic structure that has been implicated as the principal site of FVII interaction with TF (5)(6)(7)(8). The FVII molecule also requires calcium for the expression of optimum activity, one molecule of which is bound in each of the protease domains (9,10) and the EGF-1 domain at a high affinity site (11), with seven more Ca 2ϩ molecules bound with variable affinity by the Gla domain (12).
It has been shown recently that the activation of FVII to FVIIa, as well as occupation of the active site by pseudosubstrate inhibitors, results in conformational changes within the heavy chain with implications on cofactor and substrate interaction (13)(14)(15). To date, the conformational effects of these events on the TF-binding EGF-1 domain of the native FVII molecule have not been reported. Relevantly, experiments by Ambrosini et al. (16) have described conformational changes in the region of the EGF-1 domain of the highly homologous FX molecule upon activation which appears to be needed for binding to effector cell protease receptor-1. Chang and co-workers (17) have also observed conformational change in the Gla/ EGF-1 region of a FVII/FIX chimera upon proteolytic activation. As well, Persson and others (12,19,20) have indicated the presence of an interaction between the Gla/EGF-1 region and the active site of FVII, indicating that active site occupation may also affect conformation within this region.
Conformation-specific monoclonal antibodies (mAbs) with well characterized binding epitopes have been used as effective probes to study the conformational structure of proteins (for a review see Ref. 21), as well as to identify functional consequences of binding to epitopes (22,23). Recently our laboratory has used a conformation-sensitive mAb, 231-7, to help characterize the nature of the naturally occurring FVII EGF-1 structural variant N57D (24). The epitope specificity of mAb 231-7, within amino acid residues 51-88 of FVII (5), makes it a unique tool to investigate conformational changes specifically involving the FVII EGF-1 domain. An understanding of the mechanisms by which the EGF-1 domain of FVII interacts with TF is of particular relevance, as the interaction of FVII with TF appears to play a critical role not only in coagulation, but in other biological processes such as tumor metastasis and angiogenesis (25,26). The recent evidence that the interaction of FVIIa with TF elicits signal transduction indicates that these processes likely involve intracellular signaling events (27 1 The abbreviations used are: FVII, coagulation factor VII; FVIIa, activated coagulation factor VII; FVIIai, active site-inhibited coagulation factor VII; dGla-FVII, factor VII with the Gla domain proteolytically removed; TF, tissue factor; BFPRck, biotinylated D-Phe-Pro-Arg chloromethyl ketone; FPRck, D-Phe-Pro-Arg chloromethyl ketone; D⅐EGRck FVII, 1,5-dansyl-Glu-Gly-Arg chloromethyl ketone; FFRck, D-Phe-Phe-Arg chloromethyl ketone; EGF-1, first epidermal growth factor-like domain; Gla, the ␥-carboxylated domain; FXa, activated coagulation factor X; FIXa, activated coagulation factor IX; mAb, monoclonal antibody; ELISA, enzyme-linked immunosorbent assay; BSA, bovine serum albumin; AP, alkaline phosphatase; DTT, dithiothreitol.
In this report, we utilize the mAb 231-7 to investigate the conformational ramifications to the FVII EGF-1 domain upon the activation and active site occupation of the FVII molecule. We provide evidence demonstrating that both modifications are associated with conformational changes involving the FVII EGF-1 region, as well as providing evidence of functional consequences for these changes on cofactor binding and the catalytic activity of factor VII.

EXPERIMENTAL PROCEDURES
Reagents-Recombinant human FVIIa was purified from 293 cell culture supernatant 2 using a protocol adapted from Ref. 28. Recombinant human FVII(R152Q) was purified from 293 cell culture supernatant using a protocol slightly modified from that described previously (29). The FVII(R152Q) mutant was chosen because of its inability to become activated during the course of analysis, thus ensuring the zymogen conformation is maintained, as indicated by others (30). Plasma-derived zymogen FVII was obtained from Enzyme Research Labs (South Bend, IN). Isolated EGF-1 peptide, expressed and purified from E. coli and comprising amino acids 45-87 of human factor VII, was the generous gift of Dr. Geoffrey Lee (Genentech, San Francisco, CA). mAb 231-7 was prepared as described previously (31). Human monocytederived cathepsin G was obtained from Calbiochem. FPRck, BFPRck, D⅐EGRck, and FFRck were obtained from Calbiochem. Spectrozyme® FVIIa (CH 3 SO 2 -D-CHA-But-Arg-pNA.AcOH) was obtained from American Diagnostica (American Diagnostica Inc., Montreal, Quebec, Canada). Bovine serum albumin (BSA) was purchased from Sigma. Recombinant re-lipidated human TF (Ortho RecombiPlasTin®) was purchased from Hemoliance (Ortho Diagnostics Systems, Toronto, Ontario, Canada). All FVII protein concentrations were determined by antigen assay as described previously (24). Goat anti-mouse IgG conjugated to alkaline phosphatase (AP) and goat anti-rabbit IgG conjugated to AP was obtained from Jackson ImmunoResearch Laboratories (Bio/Can Scientific, Inc., Mississauga, Ontario, Canada). Monospecific rabbit antihuman FVII polyclonal sera were obtained from Diagnostica Stago (Wellmark Diagnostics, Guelph, Ontario, Canada). Sigma 104 phosphatase substrate (AP substrate, p-nitrophenyl phosphate, disodium, hexahydrate) was obtained from Sigma. Protein purification spin columns were obtained from Millipore (Millipore Corp., Bedford, MA). Immulon II HB 96-well microtiter plates were obtained from Dynex Technologies Inc. (Chantilly, VA). Tween 20 detergent (Surfact-Amps) was obtained from Pierce. All other reagents were of the finest quality available.
Chloromethyl Ketone Inhibition of FVIIa-The chloromethyl ketone inhibitors were reconstituted according to the manufacturers instructions and diluted in reaction buffer immediately prior to reaction. FVIIa samples were diluted to 1 ml in reaction buffer (50 mM Tris, pH 6.8, 150 mM NaCl, 10 mMCaCl 2 , 100 g/ml BSA) to a final FVIIa concentration of 1 M and reacted with a 25 M excess of chloromethyl ketone inhibitor (1 l addition) for 30 min. At this time another 25 M inhibitor was added (50 M total), and the reaction mix incubated at room temperature for 2 h total. The reaction mixture underwent purification using a protein spin column (Millipore Ultrafree®, molecular mass cut-off at 5 kDa) with 3ϫ buffer exchange into TBS, pH 8.0. The completeness of FVIIa inhibition was verified to be ϳ99% using the prothrombin time assay, essentially as described previously (24). Chloromethyl ketone labeling was specific for the protease domain of factor VIIa, as judged by Western blot analysis.
Generation of Gla-domainless FVII Molecules-10 g of the appropriate FVII sample was diluted to a concentration of 1 M in buffer (10 mM Tris, pH 8.6, 75 mM NaCl, 5 mM EDTA, 100 g/ml BSA) and the Gla-domainless FVIIa, -FVII(R152Q), -FPRckFVIIa, -D⅐EGRckFVIIa molecules were generated by cathepsin G-mediated proteolytic cleavage, according to the method of Nicoliasen et al. (32), shown to be specific for removal of the Gla domain. The samples were judged by Western blot to have achieved complete cleavage. No nonspecific cleavage patterns were seen. Reaction mixtures were stored frozen (Ϫ20°C) before direct use in the binding affinity analyses.
Sensitivity of mAb 231-7 Binding to FVII Conformation-Relative bindings of native, heat-denatured, and DTT-reduced FVII zymogen to both mAb 231-7 and rabbit anti-human FVII polyclonal antibody were performed using an ELISA technique. Briefly, to prepare reduced FVII, plasma-derived FVII zymogen was pretreated overnight with 10 mM DTT, and to prepare heat-denatured FVII plasma-derived FVII zymogen was boiled for 5 min in 1% SDS followed by snap cooling. Native, DTT-reduced and heat-denatured FVII samples were then incubated overnight at 4°C in carbonate antigen coating buffer (15 mM Na 2 CO 3 , 35 mM NaHCO 3 , 0.02% NaN 3 , pH 9.6) in microtiter wells on an Immulon II 96 well microtiter plate. Microtiter wells were washed 3 times with TBS-T/Ca 2ϩ buffer (10 mM Tris, pH 8.0, 150 mM NaCl, 10 mM CaCl 2 , 0.025% Tween 20,) followed by blocking of nonspecific binding sites with blocking buffer (5 mg/ml BSA in TBS-T/Ca 2ϩ ) for 3 h at room temperature. Serial dilutions of mAb 231-7 or rabbit anti-human FVII polyclonal sera in TBS-T/Ca 2ϩ buffer, containing 5 mg/ml BSA, were incubated in triplicate in the sample wells for 3 h at room temperature. Sample wells were washed (3 times) with TBS-T/Ca 2ϩ buffer followed by incubation with either goat anti-mouse IgG-conjugated AP for mAb 231-7 or goat anti-rabbit IgG-conjugated AP for the rabbit anti-FVII polyclonal sera, for 1 h at room temperature. Sample wells were washed with TBS-T/Ca 2ϩ buffer (3 times) followed by addition of AP substrate (100 l of p-nitrophenyl phosphate, disodium, and hexahydrate) at 1 mg/ml in diethanolamine buffer. The absorbance of each sample was measured at 405 nm on an Automated Microplate Reader (model EL 808, Bio-Tek Instruments, Winooski, VT) and plotted versus antibody concentration.
Direct Binding of FVII(R152Q), FVIIa, FPRckFVIIa, and D⅐EGRckFVIIa to Immobilized mAb 231-7-Briefly, mAb 231-7 (100 ng/well) was immobilized in microtiter wells on a 96-well microtiter plate overnight at 4°C in carbonate antigen-coating buffer. Nonspecific binding sites were subsequently blocked with 2 mg/ml BSA in TBS-T/ Ca 2ϩ buffer for Ն3 h at room temperature. After washing 5 times with TBS-T/Ca 2ϩ buffer, serial dilutions of FVII(R152Q), FVIIa, FPRckF-VIIa, and D⅐EGRckFVIIa molecules in TBS-T/Ca 2ϩ buffer containing 2 mg/ml BSA were incubated in duplicate in the microtiter wells for 2 h at room temperature. Samples to be analyzed in the absence of Ca 2ϩ were incubated in TBS-T/E buffer, replacing CaCl 2 with 10 mM EDTA. Sample wells were then washed 5 times with TBS-T/Ca 2ϩ buffer and incubated with biotinylated rabbit anti-human FVII polyclonal antibody (100 l of 1 g/ml antibody concentration) in TBS-T/Ca 2ϩ buffer containing 2 mg/ml BSA for 1 h at room temperature. Sample wells were washed again 5 times with TBS-T/Ca 2ϩ buffer, followed by incubation with streptavidin-conjugated AP in TBS-T/Ca 2ϩ buffer containing 2 mg/ml BSA for 1 h at room temperature. After washing 5 times with TBS-T/Ca 2ϩ buffer, AP substrate was added (100 l, 1 mg/ml in diethylanolamine buffer), and the color reaction was allowed to develop for 30 min. Absorbance was measured at 405 nm on an automated microplate reader, and the results were plotted as absorbance versus FVII concentration.
Determination of Equilibrium Dissociation Constants (K D ) for the Binding to mAb 231-7 of Various FVII Molecules and the EGF-1 Peptide-The K D of mAb 231-7 for FVII(R152Q), FVIIa, BFPRckFVIIa, FPRckFVIIa, FFRckFVIIa, D⅐EGRckFVIIa, dGla-FVII(R152Q), dGla-FVIIa, dGla-FPRckFVIIa, dGla-FFRckFVIIa, dGla-D⅐EGRckFVIIa and the EGF-1 peptide were determined using the homogeneous solution phase method described by Friguet et al. (33). Briefly, 200 pM mAb 231-7-binding sites were incubated in buffer solution (50 mM Tris, pH 8.0, 150 mM NaCl, 10 mM CaCl 2 , 1 mg/ml BSA) with serial dilutions of FVII molecules overnight with agitation at 4°C. Samples to be measured in the absence of Ca 2ϩ were incubated in buffer containing 10 mM EDTA instead of CaCl 2 . The amount of mAb 231-7 that remained unbound at equilibrium for each sample was determined using an ELISA. Briefly, for the ELISA, 25 ng/well of FVII zymogen was immobilized in 96-well microtiter plates by incubation overnight at 4°C in carbonate antigen-coating buffer. After washing 2 times with TBS-T buffer, nonspecific binding sites were subsequently blocked with 2 mg/ml BSA in either TBS-T/Ca 2ϩ or TBS-T/EDTA buffer for Ն3 h at room temperature. Following blocking, samples were incubated in duplicate in the blocked microtiter wells for 45 min at room temperature. Sample wells were then washed 5 times with TBS-T/Ca 2ϩ buffer and incubated with goat anti-mouse IgG-conjugated AP in TBS-T/Ca 2ϩ buffer containing 2 mg/ml BSA for 1 h at room temperature. Sample wells were washed again 5 times with TBS-T/Ca 2ϩ buffer followed by the addition of AP substrate (100 l, 1 mg/ml in diethylanolamine buffer), and the color reaction was allowed to develop for 30 min. Absorbance at 405 nm was measured using an automated microplate reader, with the raw ELISA data transformed and analyzed using Scatchard analysis with linear regression. Dissociation constants (K D ) were determined from the reciprocal of the slope of the linear regression curves of the Scatchard plots (33).
Effect of Active Site Occupation on Tissue Factor Binding-Measurements of the inhibitory constant for 50% inhibition (IC 50 ) of the interaction between biotinylated FVIIa and recombinant, relipidated human TF for the FVIIa, FPRckFVIIa, D⅐EGRckFVIIa, and FFRckFVIIa mol-2 R. Kelley, unpublished results. ecules were performed using a competitive ELISA, based on a method described previously by our laboratory (34).
Effect of Binding of mAb 231-7 on FVIIa Amidolytic Activity-100 nM FVIIa was incubated in amidolytic buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1 mM EDTA, 2 mg/ml BSA) with increasing concentrations (see Fig. 3) of mAb 231-7, or a mAb not specific for FVII, for 3 h at room temperature (final volume, 100 l). A 96-well microtiter plate was blocked for 3 h with amidolytic buffer ϩ 1 mg/ml BSA. Spectrozyme FVIIa in amidolytic buffer (100 l), at a final concentration of 200 M, was added to the mAb/FVIIa mix and agitated. Aliquots of the mAb 231-7/FVIIa/substrate mix (100 l) were incubated in triplicate in the 96-well microtiter plate, and the color was allowed to develop for 6 h at room temperature, at which time the absorbance at 405 nm was determined using the automated microplate reader.

RESULTS
Conformational Sensitivity of mAb 231-7 Binding to FVII-To verify the sensitivity of mAb 231-7 for perturbations in factor VII structure, the relative bindings of mAb 231-7 and polyclonal rabbit anti-human FVII antibody to native, heatdenatured, and DTT-reduced FVII zymogen were compared by ELISA. mAb 231-7 exhibited significantly reduced binding affinity to heat-denatured FVII zymogen compared with that for native FVII zymogen. In addition, the binding of mAb 231-7 to antigen was rendered negligible upon reduction of the FVII zymogen with 10 mM DTT. In contrast, the binding of native, heat-denatured, and DTT-reduced FVII zymogen to polyclonal rabbit anti-human FVII antibody was unaffected by either heat denaturation or DTT reduction treatment of the antigen, indicating that the observed reduction was specific for mAb 231-7 (data not shown).
Direct Binding to Immobilized mAb 231-7 of FVII(R152Q), FVIIa, FPRckVIIa, and D⅐EGRckFVIIa-Immobilized mAb 231-7 specifically bound FVII(R152Q), FVIIa, FPRckFVIIa and D⅐EGRckFVIIa in both the presence and absence of Ca 2ϩ . In the presence of 10 mM Ca 2ϩ , mAb 231-7 bound FVII(R152Q), FVIIa, FPRckFVIIa, and D⅐EGRckFVIIa with distinctly different binding characteristics, indicative of different conformations in the region of the EGF-1 domain for each molecule. FVII(R152Q) bound immobilized mAb 231-7 with greatest affinity, followed by FPRckFVIIa, FVIIa, and D⅐EGRckFVIIa in descending order (Fig. 1A). In the absence of Ca 2ϩ , the binding differential between FVII(R152Q) and FVIIa could not be demonstrated, indicating the activation-dependent conformational change to be calcium-dependent. In the absence of Ca 2ϩ , FVIIa bound with greater affinity to immobilized mAb 231-7 than did any of the active site-inhibited FVIIa molecules analyzed (Fig.  1B). Fig. 2 and Table I show the equilibrium binding data of various full-length FVII molecules to mAb 231-7. In the presence of Ca 2ϩ , mAb 231-7 bound FVII(R152Q) with a K D of 3.8 nM, a 2.5-fold greater affinity than that determined for FVIIa (K D ϭ 9.7 nM). In the absence of calcium the affinity of FVIIa increased approximately 2.5-fold, such that there was no significant difference between the K D of FVII(R152Q) and FVIIa for mAb 231-7. In the presence of 10 mM Ca 2ϩ , FPRckFVIIa (4.0 nM) and BFPRck-FVIIa (4.8 nM) bound mAb 231-7 with significantly greater affinity than FVIIa, whereas D⅐EGRckFVIIa (42 nM) bound mAb 231-7 with significantly less affinity (4-fold less) than FVIIa. FFRckFVIIa did not bind mAb 231-7 with a significantly different affinity than uninhibited FVIIa in the presence of calcium. Upon removal of Ca 2ϩ , the affinities of FPRckFVIIa and FFRckFVIIa for mAb 231-7 were not significantly altered (5.5 and 11 nM, respectively), whereas the affinity of D⅐EGRckFVIIa (19 nM) was increased significantly by approx-imately 2-fold, a value similar to the increase in affinity for uninhibited FVIIa. Table II shows the equilibrium binding data for various FVII molecules with the Gla domain removed, as well for the isolated EGF-1 peptide. Removal of the Gla domain resulted in a 2-fold reduction in binding affinity for dGla-FVIIa for mAb 231-7 compared with that seen for FVIIa in the presence of calcium (19 versus 9.7 nM) (Table III). This contrasts with the apparent increase in mAb 231-7 binding affinity for dGla-FVII(R152Q) compared with that seen for FVII(R152Q) (1.7 versus 3.8 nM). The removal of the Gla domain thus increased the affinity differential between the zymogen and active form of FVII for mAb 231-7 from 2.5-to 10-fold in the presence of calcium and no difference to 10-fold difference in the absence of calcium. Gla removal increased the mAb 231-7 affinity of D⅐EGRckFVIIa almost 4-fold in the presence of calcium. Gla removal also increased the affinity of D⅐EGRckFVIIa in the absence of calcium to 12.3 nM, eliminating the calciumdependent difference. Gla removal decreased the mAb 231-7 affinity of FPRckFVIIa in both the presence (7.2 nM) and absence (10 nM) of calcium, showing no calcium-dependent difference in affinity for mAb 231-7. Gla removal had no significant effect on the mAb 231-7 affinity for FFRckFVIIa in either the presence or absence of calcium (data not shown). As well, the presence or absence of calcium had no significant effect on the mAb 231-7 binding affinity of the isolated EGF-1 peptide.  Table III shows Table I. ␥ represents the fraction of bound antibody and a the concentration of free antigen, at equilibrium. respectively, representing a 10-fold increase in inhibition over unmodified FVIIa. D⅐EGRckFVIIa had an IC 50 value of 0.53 nM, a 4-fold increase in inhibition over uninhibited FVIIa. Uninhibited FVIIa had an IC 50 value of 2.32 nM.

Effect of Active Site Occupation on Tissue Factor Binding-
Effect of Binding of mAb 231-7 on FVIIa Amidolytic Activity-mAb 231-7 increased the amidolytic activity of human FVIIa toward the tri-peptide chromogenic substrate Spectrozyme FVIIa reproducibly and in a dose-dependent fashion (Fig. 3). FVIIa/mAb 231-7 activity reached a maximal increase of 41 Ϯ 11% (n ϭ 5) over FVIIa alone in the presence of approximately a 3-fold molar excess of IgG-binding sites to FVIIa molecules. An increase in FVIIa activity was detectable with as little as a 1:5 molar ratio of monoclonal binding sites to FVIIa. The stimulatory ability of mAb 231-7 also appeared to increase with increasing pH (9.9, data not shown), an effect also known to occur for TF (35). Incubation of FVIIa with a nonspecific monoclonal antibody of the same IgG subclass did not result in any increase in the amidolytic activity of FVIIa toward Spectrozyme FVIIa (Fig. 3) nor did mAb 231-7 alone elicit any amidolytic activity (data not shown).

DISCUSSION
Previous studies from our laboratory have shown that mAb 231-7 is specific for the EGF-1 domain of factor VII (5) and is sensitive to structural alterations involving this domain (24). The conformation specificity of mAb 231-7 was thus used to characterize the conformational behavior of the FVII EGF-1 domain upon zymogen activation, as well as the occupation of the active site of FVIIa with various active site-specific chloromethyl ketones.
The decrease in affinity for mAb 231-7 to FVIIa compared with that observed for the zymogen FVII(R152Q) in the presence of calcium was observable in both the direct binding assay and the homogenous solution phase assay. This effect was reproducibly quantified to give ϳ3-fold change in magnitude of binding affinity. Thus, these data indicate that the zymogen conformation of the EGF-1 domain differs from that of the catalytically active form, FVIIa. As the binding epitopes of mAb 231-7 and TF are known to overlap within EGF-1 (5), it is likely that this activation-dependent conformational change affects the manner in which FVII interacts with TF. This hypothesis is consistent with the findings of Chang and co-workers (17), who showed an activation-dependent conformational change in a protein chimera containing the Gla/EGF-1 domains of FVII that affected TF binding ability ϳ4-fold. Such domain-specific conformational changes may go undetected when one measures the overall affinity of factor VII for TF, due to the complexity of the interaction (7), an effect we overcome through the use of the EGF-1-specific mAb 231-7. Our observations thus extend the findings of Chang et al. (17) to the native FVII molecule. We have further localized the activation-dependent conformational change occurring in the light chain of FVII to involve amino acids 51-88 of the EGF-1 domain.
The observation that the activation-dependent conformational change occurring between FVII and FVIIa is dependent on the presence of calcium (Table I) indicates that there is clearly involvement of a calcium-binding region. The fact that the isolated EGF-1 domain did not show any calcium dependence in its interaction with mAb 231-7 indicates that the pres-    ence of another domain is needed to facilitate this interaction. Given the proximity of the calcium-binding Gla region and its effect on calcium binding to the EGF-1 domain in FVII (36), we investigated the possible role of the Gla domain in the observed activation-dependent conformational change. Proteolytic removal of the Gla domain eliminated any change in the FVIIa affinity to mAb 231-7 due to calcium, indicating that the calcium-associated binding was facilitated in the presence of the Gla domain and that the protease domain was not responsible for the observed calcium dependence. We feel that the Gla/Ca 2ϩdependent EGF-1 conformational changes occurring upon FVII activation may be important for the development of optimum activity of the factor VIIa molecule, as is hypothesized for the calcium-mediated Gla/EGF-1 conformational change seen in factor Xa (37).
Analysis of the impact of occupation of the active site of FVIIa on the binding of mAb 231-7 allowed us to investigate conformational changes in the EGF-1 region once the catalytic activity of the enzyme had been established. Incorporation of FPRck into the active site of FVIIa in the presence of Ca 2ϩ is associated with an apparent 3-fold increase in mAb 231-7 affinity compared with that observed upon FFRck inhibition. The fact that the two inhibitors bind in an identical manner to the active site (7) would suggest that the conformational changes propagated to the EGF-1 domain from the active site likely involved at least the S2 sub-site. (S2, P2, etc. nomenclature used in this discussion is that of Schecter and Berger (44).) In comparison, D⅐EGRck inhibition of FVIIa resulted in a 10-fold decrease in mAb 231-7 affinity compared with that observed upon FPRck inhibition. This effect may also be mediated by the different P2 residues present in the two inhibitors, although concomitant differences in P3 and the amino-terminal cap leave open the possibility of propagation through other interactions, such as at the aryl binding site. Importantly, there is precedence for allosteric linkages at the S2 and S3 sub-sites in the serine proteases, as these sites appear to be involved in the modulation of the thrombin active site by thrombomodulin (38,39).
The investigations of the calcium dependence of the conformational changes observed in FVIIa EGF-1 upon active site inhibition indicates variation in the calcium effect, depending on the inhibitor used. The affinities to mAb 231-7 upon FPRck and FFRck inhibition were not significantly affected by calcium removal, whereas the mAb 231-7 affinity measured upon D⅐EGRck inhibition showed a similar 2-fold affinity increase as that observed for uninhibited FVIIa. This variation between these inhibitors may relate to differences in the mechanism of allosteric propagation, as the P2-P4 positions of D⅐EGRck likely interact differently in the active site of FVIIa compared with FPRck and FFRck (40). Regardless of calcium involvement, we chose to investigate the effect of the presence of the adjacent calcium-binding Gla domain, as data from Persson and others (19,20) have indicated interactions involving both the Gla domain and the EGF-1 calcium-binding site with the FVII active site. Removal of the Gla domain abolished the differences in affinity among the inhibited molecules to mAb 231-7, indicating that the Gla domain was involved in the observed allosteric linkage of the active site to the EGF-1 domain. The fact that the inhibited dGla-FVIIai molecules still maintained a slight but significantly different affinity compared with uninhibited dGla-FVIIa indicates that the allosteric linkage may not be totally Gla-dependent.
Once the presence of an allosteric linkage extending from the active site of FVIIa to the EGF-1 domain was established, we then sought to investigate the functional consequences of such an interaction on the FVII molecule. Active site inhibition increased the ability of FVIIa to inhibit the FVIIa/TF interaction, consistent with the observation of increased affinity of FVIIai molecules for TF when compared with uninhibited FVIIa reported by others (14,30,41). Thus conformational changes occurring in FVIIai upon active site occupation that affect the interaction with TF are consistent with the conformational changes we observed in the EGF-1 domain. Relevantly, recent evidence has indicated that at least some of the changes in TF binding affinity can be attributed to changes in TF-binding residues in the protease domain (14,15), although changes to TF-binding regions in the light chain were not precluded by these studies.
As the mAb 231-7 binding epitope overlaps with that of TF in the EGF-1 domain (5) and the binding of TF elicits a large increase in FVIIa catalytic activity (42), we sought to investigate whether mAb binding at EGF-1 would have any affect on FVIIa activity. We observed that the binding of mAb 231-7 to EGF-1 stimulated the catalytic activity of FVIIa toward a tripeptide substrate in a dose-dependent manner. This result verified the reciprocal nature of the allosteric linkage between EGF-1 and the active site of FVIIa. The relevance of this effect is not obvious; however, the variation in mAb 231-7 binding affinities toward the various active site inhibitors suggests a role for this linkage in substrate/inhibitor specificity. The involvement of the S2 sub-site of FVIIa is consistent with this hypothesis, as the S2 sub-site has also been shown previously to be involved in substrate/inhibitor specificity for the highly homologous activated protein C and FXa serine proteases (43) as well as for thrombin (39).
The results of our study support the hypothesis that activation of FVII causes calcium-dependent conformational changes that involve the region of the EGF-1 domain. Once FVII is activated, an allosteric linkage can be observed between EGF-1 and the active site that involves the adjacent Gla domain, affecting the catalytic properties of the active enzyme and possibly cofactor binding. These observations shed light on conformational effects during the development of FVII catalytic activity and provide further evidence of the existence of independent conformational linkages that regulate the catalytic function of FVIIa, such as those shown by others (15,18,22), extending this model specifically to include the region of the EGF-1 domain of FVII.