Factor VII central. A novel mutation in the catalytic domain that reduces tissue factor binding, impairs activation by factor Xa, and abolishes amidolytic and coagulant activity.

Factor VII is a vitamin K-dependent zymogen of a serine protease that participates in the initial phase of blood coagulation. A factor VII molecular variant (factor VII Central) was identified in a 24-year-old male with severe factor VII deficiency and whose plasma factor VII antigen was 38% of normal, but expressed <1% factor VII procoagulant activity. DNA sequence analysis of the patient's factor VII gene revealed a thymidine to cytidine transition at nucleotide 10907 in exon VIII that results in a novel amino acid substitution of Phe328 to Ser. The patient was homozygous for this mutation, whereas each parent of the patient was heterozygous for this mutation. To investigate the molecular properties of this variant, a recombinant F328S factor VII mutant was prepared and analyzed in relation to wild-type factor VII. F328S factor VII exhibited <1% factor VII procoagulant activity and a 2-fold decreased affinity for tissue factor and failed to activate factor X or IX in the presence of tissue factor following activation by factor Xa. In addition, F328S factor VIIa exhibited no detectable amidolytic activity in the presence of tissue factor. The rate of F328S factor VII activation by factor Xa was markedly decreased relative to the rate of wild-type factor VII activation as revealed by densitometry scanning of SDS gels. Temporal analysis of this reaction by SDS-polyacrylamide gel electrophoresis also revealed the formation of two novel F328S factor VII degradation products (40 and 9 kDa) resulting from factor Xa proteolysis of the Arg315-Lys316 peptide bond in intact F328S factor VII. Computer modeling and molecular dynamics simulations of the serine protease domain of factor VIIa suggested that the inability of F328S factor VIIa to cleave substrates may result from the apparent formation of a hydrogen bond between Tyr377 and Asp338, a residue at the bottom of the substrate-binding pocket important for the interaction of substrate arginine side chains with the enzyme. These findings suggest that Phe328, which is conserved in prothrombin, factor IX, factor X, factor VII, and trypsin, is important for factor VIIa catalysis.

Factor VII is a vitamin K-dependent zymogen of a serine protease that participates in the initial phase of blood coagulation. A factor VII molecular variant (factor VII Central) was identified in a 24-year-old male with severe factor VII deficiency and whose plasma factor VII antigen was 38% of normal, but expressed <1% factor VII procoagulant activity. DNA sequence analysis of the patient's factor VII gene revealed a thymidine to cytidine transition at nucleotide 10907 in exon VIII that results in a novel amino acid substitution of Phe 328 to Ser. The patient was homozygous for this mutation, whereas each parent of the patient was heterozygous for this mutation. To investigate the molecular properties of this variant, a recombinant F328S factor VII mutant was prepared and analyzed in relation to wild-type factor VII. F328S factor VII exhibited <1% factor VII procoagulant activity and a 2-fold decreased affinity for tissue factor and failed to activate factor X or IX in the presence of tissue factor following activation by factor Xa. In addition, F328S factor VIIa exhibited no detectable amidolytic activity in the presence of tissue factor. The rate of F328S factor VII activation by factor Xa was markedly decreased relative to the rate of wild-type factor VII activation as revealed by densitometry scanning of SDS gels. Temporal analysis of this reaction by SDSpolyacrylamide gel electrophoresis also revealed the formation of two novel F328S factor VII degradation products (40 and 9 kDa) resulting from factor Xa proteolysis of the Arg 315 -Lys 316 peptide bond in intact F328S factor VII. Computer modeling and molecular dynamics simulations of the serine protease domain of factor VIIa suggested that the inability of F328S factor VIIa to cleave substrates may result from the apparent formation of a hydrogen bond between Tyr 377 and Asp 338 , a residue at the bottom of the substrate-binding pocket important for the interaction of substrate arginine side chains with the enzyme. These findings suggest that Phe 328 , which is conserved in prothrombin, factor IX, factor X, factor VII, and trypsin, is important for factor VIIa catalysis.
Factor VII is a single-chain, vitamin K-dependent plasma glycoprotein (50 kDa) that plays a key role in the initiation of the extrinsic pathway of blood coagulation (2). The gene for factor VII is located on chromosome 13 at q34-qter, spans 12.8 kilobase pairs, and contains nine exons (3,4). The gene organization and protein structure for factor VII are similar to other vitamin K-dependent coagulation proteins in that each is a multidomain glycoprotein containing a Gla domain, epidermal growth factor-like or kringle domains, and a C-terminal catalytic domain homologous to trypsin. Inherited factor VII deficiency is a rare autosomal recessive disorder and is phenotypically heterogeneous. The clinical features are variable with a rather poor correlation between reported procoagulant activity and bleeding tendency (5)(6)(7). Factor VII deficiency is classified as either CRM Ϫ (type 1) or CRM ϩ (type 2) based upon the absence or presence, respectively, of a disparity between the activity and antigen levels (7). In the majority of factor VIIdeficient patients, plasma levels of factor VII activity diminish in parallel with immunoreactive factor VII antigen. A discrepancy between factor VII coagulant activity and factor VII antigen levels has been found in a few patients, suggesting the presence of a dysfunctional factor VII (reviewed in Ref. 8). In this report, we describe the molecular basis of a severe factor VII deficiency in a 24-year-old male whose plasma factor VII activity was Ͻ1%, while his plasma factor VII antigen was 38% of normal. We have analyzed the factor VII gene sequence of this patient, designated as factor VII Central, and demonstrate that the patient was homozygous for a single point missense mutation in exon VIII in the catalytic domain, resulting in a Phe to Ser substitution at amino acid 328. We have expressed and purified this variant from baby hamster kidney cells transfected with the cDNA coding for this mutant and demonstrate that the isolated mutant exhibits Ͻ1% procoagulant activity, no demonstrable amidolytic or proteolytic activity following activation by factor Xa, and decreased affinity for tissue factor, thus recapitulating the patient's factor VII phenotype. In addition, the F328S factor VII mutant, as well as the isolated patient's factor VII, undergoes proteolytic degradation by factor Xa at Arg 315 -Lys 316 , suggesting that the Phe 328 to Ser substitution creates an exposed surface loop in this region readily accessible for cleavage by factor Xa, further reducing the catalytic activity of this clotting factor.

EXPERIMENTAL PROCEDURES
Materials-Recombinant Taq DNA polymerase was obtained from Perkin-Elmer. NuSieve GTG-agarose was a product of FMC Corp. Bio-Products. Bovine serum albumin (fatty acid-free) and peroxidase-conjugated goat anti-rabbit IgG were purchased from Boehringer Mannheim. Affi-Gel 10 was obtained from Bio-Rad. Microtitration plates (96-well) were obtained from Nunc. 125 I-Labeled protein A was purchased from DuPont NEN. H-D-Ile-Pro-Arg-p-nitroanilide (S-2288) was obtained from Kabi Pharmacia Hepar, Inc. Strepavidin-coated paramagnetic beads were purchased from Dynal, Inc.. All other reagents were of the highest grade commercially available.
Proteins-Human plasma factor Xa (9), recombinant human factor VII (10), soluble recombinant human tissue factor apoprotein TF 1-218 1 (11), human brain thromboplastin (12), mixed brain phospholipids (13), and affinity-purified rabbit anti-human factor VII IgG (14) were prepared by published methods. Calcium-dependent (CaFVII22) and calcium-independent, heavy chain-specific (AD-1) murine anti-human factor VII monoclonal antibodies were produced in Balb/cj mice essentially according to Kohler and Milstein (15) and purified from ascites fluid by either protein A-Sepharose or DEAE-Affi-Gel blue column chromatography. Monoclonal antibody CaFVII22 was coupled to Affi-Gel 10 according to the manufacturer's recommendation. Recombinant F328S factor VII 2 was expressed in stably transfected baby hamster kidney cells using mutagenesis, transfection, and culturing conditions essentially as described for S344A factor VII (16) and was purified to homogeneity in a single step from ϳ10 liters of serum-free baby hamster kidney conditioned medium by CaFVII22-Affi-Gel 10 immunoaffinity chromatography (16).
DNA Sequence Analysis of the Propositus Factor VII Gene and Detection of the Mutation in the Family-Collection of blood samples from the propositus and his parents was performed following appropriate consent at a time when the propositus had not been transfused with plasma for at least 2 weeks. Genomic DNA was extracted from peripheral blood samples using proteinase K lysis, phenol extraction, and ethanol precipitation (17). Based on published intron data for human factor VII (4), eight pairs of oligonucleotides (Table I) were synthesized and used to perform polymerase chain reaction amplification (18) of the seven coding regions (exons II-VIII) and exon-intron boundaries in the three DNA samples. Due to their small size, exons III and IV were amplified as a single fragment. In each reaction, one of the polymerase chain reaction primers was 5Ј-biotinylated to facilitate the subsequent preparation of single-stranded templates for sequence analysis. Inasmuch as exon VIII is relatively large, we used both forward and reverse 5Ј-biotinylated primers for separate amplifications. In the polymerase chain reaction, the biotinylated primer was incorporated into the amplification product strand complementary to the sequencing primer subsequently used. Target sequences were amplified essentially according to Chaing et al. (18) in a 100-l volume containing 0.5 g of genomic DNA, 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 0.01% gelatin, and variable concentrations of MgCl 2 (Table I).
General Methods-SDS-polyacrylamide slab gel electrophoresis (PAGE) was performed according to Laemmli (19) using 10% polyacrylamide separating gels. Following electrophoresis, the proteins were visualized either by staining with Coomassie Brilliant Blue or by immunoblotting and autoradiography following electrotransfer to nitrocellulose. The coagulant activities of recombinant wild-type human factor VII and F328S factor VII were assessed in a one-stage clotting assay using immunodepleted factor VII-deficient plasma (Ͻ1% factor VII antigen) and human brain thromboplastin essentially as described (20). Factor VII antigen concentrations were determined by an enzyme-linked immunosorbent assay method (21, 22) using a factor VIIa heavy chain-specific monoclonal antibody (AD-1) as the capture antibody. The Gla content of the recombinant wild-type and F328S factor VII preparations was determined according to Kuwada and Katayama (23). Biotinylated and unbiotinylated synthetic oligonucleotides were prepared by solid-phase phosphoramidite chemistry on an automated DNA/RNA synthesizer (Applied Biosystems Model 394) in the University of New Mexico School of Medicine Protein Chemistry Laboratory. Amino-terminal amino acid sequence was determined using a Beckman Model LF3000 gas-phase sequenator.
Tissue Factor Binding Assay-The affinity of F328S factor VII and recombinant wild-type factor VII for immobilized soluble recombinant tissue factor apoprotein TF-  was assessed in an enzyme-linked immunosorbent assay using a modification of the procedure described by Chaing et al. (18). In this procedure, 100 l of TF-1-218 (10 g/ml in 50 mM sodium carbonate (pH 9.6)) was added to the wells of a 96-well microtitration plate (MaxiSorp Immunoplate) and incubated at 4°C overnight. After washing the plate with TBS and 0.1% Tween 20, each well was treated with 200 l of 0.1% gelatin in TBS for 2 h to block nonspecific binding sites and subsequently washed with TBS/Tween 20. Aliquots (100 l) of various concentrations of either wild-type factor VII or F328S factor VII (dissolved in TBS, 5 mM CaCl 2 , 0.05% Tween 20, and 1% BSA) were added to the plate and allowed to incubate at 37°C for 2 h. The plate was then washed six times with TBS and 0.1% Tween 20 containing 5 mM CaCl 2 , and 100 l of affinity-purified rabbit antihuman factor VII IgG (5 g/ml in TBS, 5 mM CaCl 2 , and 1% BSA) was added to each well. Following a 2-h incubation at 37°C, the plate was washed six times with TBS/Tween 20/CaCl 2 and subsequently treated for 2 h at 37°C with 100 l of peroxidase-conjugated goat anti-rabbit IgG (5 g/ml in TBS/CaCl 2 /BSA). After a final washing, 100 l of o-phenylenediamine (1 mg/ml in 0.1 M sodium citrate (pH 4.5) and 0.5% hydrogen peroxide) was added to each well. After a 2-3-min incubation at room temperature, 100 l of 2.5 M sulfuric acid was added to each well, and the A 490 was measured at room temperature in a UV max kinetic microtiter plate reader (Molecular Devices). The A 490 was assumed to be directly proportional to factor VII binding to immobilized TF-  . Factor VII specific binding (A 490 s ) was obtained by subtracting the A 490 of the gelatin control from the apparent A 490 . The apparent dissociation constant (K d(app) ) was estimated from the slope of a plot relating A 490 s /offered factor VII concentration versus A 490 s .
Purification of the Patient's Factor VII-The patient's factor VII was isolated by a combination of barium citrate adsorption, elution, and immunoaffinity chromatography as follows. Citrated patient plasma (ϳ20 ml) was treated with 2.4 ml of 1 M BaCl 2 and 100 l of 1 M benzamidine and mixed for 1 h at 4°C. The mixture was centrifuged (8000 rpm, 15 min), and the precipitate was washed twice in TBS, 0.1 M BaCl 2 , and 1 mM benzamidine. The washed precipitate was then dissolved in 10 ml of TBS containing 30 mM EDTA and 1 mM benzamidine, and the eluate was dialyzed overnight at 4°C against 4 liters of TBS and 1 mM benzamidine. The dialyzed sample was made 5 mM in CaCl 2 and 10 mM in benzamidine and incubated with 1 ml of CaFVII22-Affi-Gel 10 at 4°C for 90 min. After washing the CaFVII22-Affi-Gel 10 twice with 50 mM Tris-HCl (pH 8), 0.1% BSA, and 5 mM CaCl 2 , the patient's factor VII was eluted from the resin with 50 mM Tris-HCl (pH 8), 0.1% BSA, and 30 mM EDTA and subsequently dialyzed against TBS.
Analysis of Factor VII Activation by Factor Xa-The proteolytic ac- tivation of wild-type, mutant, and patient factor VII by factor Xa in the presence of 5 mM calcium and mixed phospholipids was assessed by clotting activity changes as well as SDS-PAGE and immunoblotting essentially as described (22). Reactions were performed in 1.5-ml snapcap polypropylene tubes at 37°C. Incubation mixtures consisted of wild-type factor VII (200 nM), F328S factor VII (200 nM), or the patient's factor VII (200 nM) along with mixed brain phospholipids (ϳ0.5 mM final phospholipid concentration) and human factor Xa (1 nM) in a total volume of 400 l of TBS, 0.1% BSA, and 5 mM CaCl 2 . The reaction was initiated by the addition of factor Xa, and at selected intervals, 40-l aliquots were removed from the incubation mixture and added to 2 l of 0.5 M EDTA to stop the reaction. An aliquot (20 l) of this mixture was used for clotting activity measurements, while the remaining 20 l was subjected to SDS-PAGE following reduction with 10% ␤-mercaptoethanol. Following electrophoresis, the proteins were electrophoretically transferred to nitrocellulose membranes, and factor VI/VIIa and degradation products were visualized by incubation with affinity-purified rabbit anti-factor VII IgG followed by incubation with 125 I-labeled protein A and autoradiography. Structural Modeling-The Homology module within Insight II (Biosym/MSI Technologies) was used on an SGI R8000 workstation to construct a homology-based three-dimensional model of factor VII utilizing the crystal structure of factor Xa (24) as the reference template. Sets of highly conserved regions were initially identified. The nonidentical residues of the template were mutated to the corresponding residues of factor VII, followed by the necessary insertions and deletions in order to generate the sequence of the model. Areas that required loop generation or in which the lengths of the model loops differed from those in the reference protein were computationally constructed by searching the Brookhaven Data Bank for regions of proteins that meet similarity criteria. An alternative approach was taken in cases where the selected loops appeared to have steric overlaps with the newly built regions (25). To refine the final model-built structures, a few hundred steps of steepest descent energy minimization were performed using the Discover force field, and the resultant model was evaluated for sensible conformations and physicochemical properties (26).
Molecular dynamics simulations were also used in order to investigate possible conformational transitions in the wild-type and mutant proteins, in particular those associated with local conformational changes regarding the catalytic triad. Structures generated in previous model-building exercises were used as starting conformations for the wild-type and mutant proteins, and two sets of molecular dynamics simulations were performed in vacuo on both proteins with a distancedependent dielectric constant. In all simulation runs, an integration step of 1 fs was used, and nonbonded interaction pair lists were updated every 20 steps. Each system was equilibrated for 10 ps followed by 50 ps of data collection, and the simulation results were analyzed visually using interactive molecular graphics within Insight.

RESULTS
Case Report-The propositus is a 24-year-old Hispanic male from Central, New Mexico who experienced recurrent epistaxis as a child and joint and soft tissue hemorrhage that was usually related to trauma. He has a markedly prolonged prothrom-bin time (85 s; control ϭ 9 -13 s), and his plasma contains Ͻ1% factor VII coagulant activity as measured by a specific onestage factor VII clotting assay. The factor VII antigen in the patient's plasma is 190 ng/ml (normal ϭ 400 -600 ng/ml) as measured by an enzyme-linked immunosorbent assay (21,22). The factor VII activity levels in plasma samples obtained from the patient's mother, father, and sister, all clinically asymptomatic, are 71, 76, and 123% of normal, respectively.
DNA Sequence Analysis-All exons and exon-intron boundaries of the patient's factor VII gene were enzymatically amplified using biotin-labeled primers and by directly sequencing the isolated amplified biotinylated strand. When compared with the normal sequence, the sequence of the propositus factor VII gene indicated a single mismatch in exon VIII with a thymidine to cytidine transition at nucleotide 10907. This mutation results in the substitution of Ser for Phe at amino acid 328. By direct sequencing of polymerase chain reaction products, the patient was homozygous for this mutation (Fig. 1), whereas each of his parents was heterozygous for this mutation (data not shown). Furthermore, this mutation neither abolished nor created any restriction endonuclease site.
Characterization of Recombinant F328S Factor VII-Recombinant F328S factor VII was purified to homogeneity by immunoaffinity chromatography from the serum-free conditioned medium of baby hamster kidney cells stably transfected with a plasmid containing the Phe to Ser mutation. Purified F328S factor VII migrated as a single band in SDS-PAGE with essentially the same mobility as recombinant wild-type factor VII in the presence or absence of reducing agent (Fig. 2). Analysis of F328S factor VII for ␥-carboxyglutamic acid content following alkaline hydrolysis indicated that the preparation was fully ␥-carboxylated (data not shown). In a one-stage clotting assay, F328S factor VII exhibited a specific clotting activity of Ͻ20 units/mg, while the specific activity of wild-type factor VII was ϳ2000 units/mg.
Cleavage of Factor VII Central and F328S Factor VII by Factor Xa-Factor VII Central (ϳ1-2 g) was purified to homogeneity from 20 ml of the patient's plasma by a combination of barium citrate adsorption, elution, and immunoaffinity chromatography. Incubation of recombinant wild-type factor VII with factor Xa (1:200 enzyme/substrate molar ratio) in the presence of calcium and mixed phospholipids resulted in the complete conversion of wild-type single-chain factor VII to twochain factor VIIa within 60 min of incubation at 37°C (Fig. 3). Under identical conditions, incubation of F328S factor VII or factor VII Central with factor Xa-calcium-phospholipids yielded a 40-kDa intermediate in addition to the factor VIIa heavy and light chains that migrate with apparent molecular masses of 34 and 26 kDa, respectively (Fig. 3). Although not visible by autoradiography, preparative incubation mixtures containing F328S factor VII and factor Xa-calcium-phospholipids also produced a lower molecular mass fragment in Coomassie Blue-stained gels that migrated with an apparent molecular mass of 8 -9 kDa (data not shown). Subjecting these preparative incubation mixtures to reverse-phase high pressure liquid chromatography (C 4 column) following reduction with 5 mM dithiothreitol resulted in the purification to homogeneity of the 40-and 9-kDa peptides. Amino-terminal sequence analyses of these peptides indicated sequences of Ala-Asn-Ala-Phe-Leu and Lys-Val-Gly-Asp-Ser for the 40-and 9-kDa peptides, respectively, which coincided with the amino terminus of the parent protein and an internal sequence located at Lys 316 -Ser 320 . Thus, substitution of Ser for Phe 328 in factor VII Central appears to make the Arg 315 -Lys 316 peptide bond more solvent-accessible, leading to its cleavage by factor Xa. Densitometry scans of each incubation mixture revealed that F328S factor VII and factor VII Central were each converted to twochain factor VIIa by factor Xa at a rate ϳ10 -20% that observed for the conversion of wild-type factor VII to factor VIIa by factor Xa. In addition, cleavage of Arg 315 -Lys 316 in F328S factor VII or factor VII Central appeared to render the degraded molecule resistant to cleavage by factor Xa at the Arg 152 -Ile 153 activation peptide bond, as no additional bands coinciding with the factor Xa-mediated cleavage of the 40-kDa species were observed. Clotting assays of each incubation mixture indicated that while cleavage of wild-type factor VII by factor Xa resulted in an ϳ25-30-fold increase in factor VII clotting activity, no apparent increase in clotting activity was observed for the incubation mixtures containing F328S factor VII or factor VII Central. Furthermore, in contrast to wild-type factor VIIa (27), no tissue factor-enhanced amidolytic activity was detected in temporal aliquots of activation incubation mixtures containing F328S factor VII, factor Xa, calcium, and phospholipids.
Analysis of Factor VII Binding to Immobilized Tissue Factor Aproprotein-To determine whether the markedly reduced clotting and amidolytic activities of F328S factor VII/VIIa observed above was related to its inability to interact with its cofactor, tissue factor, we next compared the affinity of wildtype factor VII and F328S factor VII for soluble TF apoprotein TF-1-218 bound to polystyrene microtitration plates. In these studies, increasing concentrations of wild-type and F328S factor VII were separately incubated at 37°C in duplicate wells that had previously been coated with either TF-(1-218) or gelatin. The wells were then washed six times to remove unbound factor VII, and the amount of bound factor VII was determined using affinity-purified rabbit anti-human factor VII IgG. Total binding and nonspecific binding were determined as the amount of factor VII bound to either TF-1-218 or gelatin, respectively, and specific binding was determined by subtracting nonspecific binding from total binding. Fig. 4A demonstrates the specific binding of wild-type and F328S fac- Interaction of recombinant wild-type human factor VII and human F328S factor VII with immobilized soluble tissue factor apoprotein. A, various concentrations of recombinant wildtype factor VII (E) or F328S factor VII (q) were incubated in the presence of CaCl 2 with either soluble tissue factor-or gelatin-coated microtiter plates as a control. After a 2-h incubation, bound factor VII was detected by rabbit anti-factor VII IgG and peroxidase-conjugated goat anti-rabbit IgG. Factor VII specific binding was determined by subtracting the A 490 of gelatin-coated control plates from the A 490 of soluble tissue factor-coated plates. B, shown is a Scatchard plot of factor VII specific binding to soluble tissue factor. K d(app) was estimated from the slope of a plot relating A 490 /offered factor VII concentration versus A 490 . tor VII to immobilized TF-  . Each factor VII preparation bound to TF-1-218 in a concentration-dependent manner that approached saturation at 20 -40 nM factor VII. Binding of each factor VII preparation was calcium-dependent, as little, if any, binding was observed in the presence of 10 mM EDTA (data not shown). Scatchard plots of the binding data (Fig. 4B) indicated apparent dissociation constants (K d(app) ) of 7.6 Ϯ 0.4 and 15.9 Ϯ 1.2 nM for wild-type factor VII and F328S factor VII, respectively. The K d(app) value for wild-type factor VII was consistent with that observed in previous reports (28 -31). These findings indicate that F328S factor VII and presumably factor VII Central exhibit an ϳ2-fold decrease in affinity for TF-1-218 in relation to wild-type factor VII and suggest that substitution of Ser for Phe 328 results in a new conformation of factor VII in this region of the molecule that disrupts its local interaction with tissue factor apoprotein. DISCUSSION We describe the molecular basis underlying a severe factor VII deficiency designated as factor VII Central. The factor VII Central propositus was homozygous for a single point missense mutation in exon VIII in the catalytic domain of the molecule, resulting in a phenylalanine to serine substitution at residue 328. As a result of this mutation, the patient's plasma factor VII clotting activity was Ͻ1% of normal, while his plasma factor VII antigen concentration was ϳ38% of normal.
In an effort to functionally characterize the patient's factor VII molecule, we constructed the F328S factor VII variant, expressed this mutant in baby hamster kidney cells, and purified it from serum-free conditioned media on a single-step, calcium-dependent immunoaffinity column. The specific clotting activity of the purified F328S factor VII preparation was Ͻ20 units/mg, in comparison with recombinant wild-type human factor VII, which exhibited a specific activity of ϳ2000 units/mg. Incubation of F328S factor VII with a complex of factor Xa-calcium-phospholipids resulted in the cleavage of the Arg 315 -Lys 316 peptide bond, in addition to cleavage of the Arg 152 -Ile 153 bond, which leads to the activation of wild-type factor VII. An identical temporal cleavage pattern for F328S factor VII was observed when factor Xa was substituted with a purified preparation of a factor VII activator derived from Taipan snake venom (32), strongly suggesting that cleavage of the Arg 315 -Lys 316 peptide bond was not unique to factor Xa (data not shown). As revealed by densitometry scans of temporal aliquots of the incubation mixture, the rate of F328S factor VIIa formation was 10 -20% of that observed for wild-type factor VII. At present, it is unclear as to whether this decreased rate of F328S factor VII activation by factor Xa relates to the apparent inability of factor Xa to cleave the 40-kDa fragment consisting of residues 1-315, inasmuch as no band corresponding to residues 152-315 (ϳ18 kDa) was observed in reduced samples of this incubation mixture. In spite of the formation of some two-chain F328S factor VIIa, no increase in coagulant activity was detectable in these incubation mixtures, whereas comparable incubation mixtures containing wild-type factor VII and factor Xa-calcium-phospholipids generated a 25-30fold increase in coagulant activity that correlated with the conversion of zymogen factor VII to factor VIIa. Of additional importance, no amidolytic activity for S-2288 was observed in aliquots of F328S factor VIIa following incubation with soluble tissue factor apoprotein. Essentially identical results to those observed for F328S factor VII with respect to factor Xa-mediated cleavage and specific coagulant activity were observed using the factor VII (1-2 g) preparation purified from small amounts of the patient's plasma by a combination of barium citrate adsorption, EDTA elution, and immunoaffinity chromatography, providing strong evidence that our F328S factor VII preparation recapitulated the structure-function characteristics of the patient's factor VII molecule.
The structure-function basis for the inability of F328S factor VIIa to express proteolytic and particularly amidolytic activity remains enigmatic. While direct tissue factor binding assays revealed a 2-fold decrease in affinity of F328S factor VII for tissue factor apoprotein in relation to wild-type factor VII, these differences could not account for the complete absence of proteolytic and amidolytic activity in F328S factor VIIa. Our experimental evidence is consistent with a conformational change in the vicinity of the Cys 310 -Cys 329 loop induced by the Phe to Ser mutation, resulting in the surface expression of the Arg 315 -Lys 316 peptide bond and a mild disturbance of the factor VII-tissue factor interaction. In the latter case, crystallographic studies by Banner et al. (33) revealed that Asp 309 in the heavy chain of factor VIIa directly interacts with Tyr 94 in tissue factor apoprotein through three hydrogen bonds. Thus, it is entirely conceivable that the Phe to Ser substitution and the putative surface expression of the Arg 315 -Lys 316 peptide bond impact on this particular molecular interaction and slightly decrease the affinity of F328S factor VII for tissue factor.
To obtain insight into the complete lack of proteolytic and amidolytic activity of F328S factor VIIa, we used homology modeling, based on the published x-ray structure of human factor Xa (24), and performed dynamics simulations on both wild-type and F328S factor VII. The disposition of the critical residues (His 193 , Asp 242 , and Ser 344 ) in the active site of factor Xa and VIIa structures appears to be virtually superimposable (data not shown). In wild-type factor VII, the catalytic triad is located immediately below a loop that separates it from Phe 328 . The aromatic ring of Phe 328 stacks between Tyr 377 and His 348 , while the aromatic hydroxyl of Tyr 377 forms a strong hydrogen bond with Ser 339 . Upon inspection of the mutant, Ser 328 is situated in a mostly hydrophobic environment without any hydrogen-bonding counterparts in the vicinity.
In molecular simulation runs, the catalytic triad domain of wild-type factor VIIa appeared quite stable. The hydrogen bond between Ser 344 and His 193 weakened or broke occasionally, but was re-formed with favorable distances and acceptable dihedral angles. The hydrogen bond distances increased up to 5 Å, but the bond breakage did not last for more than 1 ps. However, this was not observed for the hydrogen bond between Ser 344 and Asp 242 , as this interaction was retained throughout the simulations with very little deviation from the original distance and geometry. In addition, Tyr 377 appeared to be rotating away from Ser 339 in wild-type factor VII, thus resulting in a breakage of the hydrogen bond interaction with Ser 339 (Fig. 5A). This motion of Tyr 377 , however, strengthens its stacking interaction with Phe 328 , for the phenol chromophore is oriented roughly perpendicular to the aromatic Phe 328 , which is typical for interactions of this type. The mutant enzyme displayed a similar behavior throughout the simulations, with transient weakening or breakage followed by re-formation. However, examination of the environment surrounding Ser 328 provided some clues as to the local motions related to the single mutation. In this regard, Tyr 377 no longer formed a hydrogen bond with Ser 339 , but rather interacted with Asp 338 , forming a strong linear hydrogen bond, while Ser 328 rotated toward His 348 (Fig.  5B). Thus, based upon these types of analyses, we speculate that the Phe to Ser substitution at residue 328 in factor VII Central results in the formation of a new conformation in the molecule such that Tyr 377 interacts with Asp 338 , a critical residue at the bottom of the substrate-binding pocket of the enzyme (34), and thereby precludes substrate binding. Whether the interaction of F328S factor VII with tissue factor induces other changes in this conformation that reduce substrateenzyme interactions further is unknown and will require additional studies.