Residue Met(156) contributes to the labile enzyme conformation of coagulation factor VIIa.

Serine protease activation is typically controlled by proteolytic cleavage of the scissile bond, resulting in spontaneous formation of the activating Ile(16)-Asp(194) salt bridge. The initiating coagulation protease factor VIIa (VIIa) differs by remaining in a zymogen-like conformation that confers the control of catalytic activity to the obligatory cofactor and receptor tissue factor (TF). This study demonstrates that the unusual hydrophobic Met(156) residue contributes to the propensity of the VIIa protease domain to remain in a zymogen-like conformation. Mutation of Met(156) to Gln, which is found in the same position of the highly homologous factor IX, had no influence on the amidolytic and proteolytic activity of TF-bound VIIa. Furthermore, the mutation did not appreciably stabilize the labile Ile(16)-Asp(194) salt bridge in the absence of cofactor. VIIa(Gln156) had increased affinity for TF, consistent with a long range conformational effect that stabilized the cofactor binding site in the VIIa protease domain. Notably, in the absence of cofactor, amidolytic and proteolytic function of VIIa(Gln156) were enhanced 3- and 9-fold, respectively, compared with wild-type VIIa. The mutation thus selectively influenced the catalytic activity of free VIIa, identifying the Met(156) residue position as a determinant for the zymogen-like properties of free VIIa.

Allosteric regulation of catalytic activity of the serine protease factor VIIa (VIIa) 1 is utilized as a mechanism to control the initiation of the coagulation pathways (1). VIIa circulates in the blood plasma as zymogen as well as the cleaved two-chain enzyme (2), but proteolytic function only ensues upon binding to its cell surface receptor and catalytic cofactor tissue factor (TF). TF has two distinct effects that regulate proteolysis by VIIa. First, TF provides affinity for macromolecular substrate by contributing to an extended exosite together with the bound VIIa ␥-carboxyglutamic acid-rich (Gla) domain. This exosite is a binding site for the factor X Gla-domain (3)(4)(5). Second, TF enhances catalytic activity by allosteric effects on the VIIa protease domain. In the absence of cofactor, VIIa has only very low catalytic activity toward small peptidyl substrate mimetics and TF stimulates the amidolytic of VIIa up to 100-fold (1). However, macromolecular substrate factor X scissile bond catalysis is enhanced Ͼ1000-fold (6), indicating that cofactorinduced conformational changes may influence extended macromolecular substrate recognition regions in addition to the S1-S3 subsite 2 that is probed by the small substrates.
The low catalytic activity of free VIIa results from a zymogen-like conformation of the enzyme. Upon zymogen cleavage, serine protease domains typically undergo a conformational ordering of loop segments, termed the activation domain (7), resulting in the formation of an activating canonical salt bridge of Asp 194 with the newly generated amino-terminal Ile 16 . In the absence of cofactor, VIIa shows an increased susceptibility of the amino terminus to chemical modification (8), indicating exposure of Ile 16 that can result from an alternative conformation or increased flexibility and disorder in the activation pocket of free VIIa. The structural determinants for the propensity of VIIa to stay in a zymogen-like conformation have not been investigated. Available structures of free and TF-bound VIIa (9 -13) did not provide mechanistic insight, because in each case the active site of VIIa was occupied with inhibitors that are known to stabilize the Ile 16 -Asp 194 salt bridge (14) and restrict conformational flexibility in the VIIa protease domain (15). Mutational studies also failed to elucidate the basis for the labile enzyme conformation, because the approaches taken mainly probed the active enzyme in the TF⅐VIIa complex (16).
This study investigates the role of residue Met 156 in maintaining the zymogen-like conformation of VIIa. This residue is located within the activation pocket, covering Ile 16 upon aminoterminal insertion (9). The conformation of the 156 side chain can influence the catalytic activity of serine protease domains. In the case of tissue plasminogen activator (tPA) (17,18) and vampire bat plasminogen activator (19), Lys 156 can substitute for Ile 16 to form an activating salt bridge with Asp 194 , resulting in efficient catalysis in the absence of zymogen cleavage. However, Lys at this position is found in a large number of serine proteases without conferring catalytic activity in the zymogen precursors, indicating that multiple interactions within the activation pocket determine the activation state of serine protease domains. Although Lys or other hydrophilic side chains are predominant in serine proteases that undergo spontaneous ordering of the activation pocket upon zymogen cleavage, VIIa has a Met residue in the 156 position. We hypothesized that the side-chain property of Met 156 is one of the determinants that interfere with the acquisition of full catalytic activity of VIIa upon zymogen cleavage. This study demonstrates that replacement of Met 156 with Gln, the side chain found in factor IX, had little effect on the activity of TF-bound VIIa. However, free VIIa Gln156 had enhanced catalytic function toward macromolecular and small peptidyl substrates. These experiments thus identify the first residue side chain that is one of the determinants for the zymogen-like conformation of the VIIa protease domain.

MATERIALS AND METHODS
Proteins-Wild-type and mutant VIIa were expressed in Chinese hamster ovary cells grown in suspension culture in serum-free medium supplemented with vitamin K 3 . Mutant and wild-type recombinant VII was immunoaffinity purified with a calcium-dependent monoclonal antibody, autoactivated to VIIa at 4°C, followed by ion exchange chromatography on MonoQ, as described previously (20). Factor X was purified from plasma with a final monoclonal antibody affinity step to reduce contamination by plasma VII (16). Full-length recombinant human TF, produced from insect cells, was reconstituted into 30% phosphatidylserine/70% phosphatidylcholine (PCPS), as described previously (21). The soluble extracellular domain of TF (TF-(1-218)) was expressed in Escherichia coli and refolded from inclusion bodies (22). The Kunitztype inhibitor 5L15 selected for VIIa specificity by phage display (23) was kindly provided by Dr. George Vlasuk (Corvas International, San Diego, CA).
Functional Assays-Kinetic parameters for factor X activation were determined at fixed concentration (200 pM) of TF reconstituted in PCPS (TF/PCPS) and excess wild-type or mutant VIIa (1 nM) in Hepes buffer saline (HBS, 10 mM Hepes, 150 mM NaCl, pH 7.4), 5 mM CaCl 2 , 0.2% bovine serum albumin. After a brief incubation at 37°C to allow TF⅐VIIa complex formation, factor X (8 nM to 1 M) was added, and factor Xa was quantified with the chromogenic substrate Spectrozyme FXa (American Diagnostica, Greenwich, CT) in samples quenched with 100 mM EDTA. Initial rates of factor Xa generation, based on calibration curves made with purified Xa, were fitted to the Michaelis-Menten equation using least squares regression analysis. For the determination of the proteolytic activity of free VIIa, 250 nM VIIa was incubated with 1 M factor X at 37°C in the presence or absence of 100 M PCPS, followed by determination of factor Xa generation by chromogenic assay. Kinetic parameters for chromogenic substrate hydrolysis (Chromozym tPA, Roche Molecular Biochemicals) were determined at fixed enzyme concentration (60 nM in the absence of TF or 30 nM in the presence of 120 nM TF-(1-218)) with varying concentrations of substrate (0.02-2 mM) in Tris-buffered saline (TBS, 20 mM Tris, 150 mM NaCl, pH 8.0), 5 mM CaCl 2 , and 0.2% bovine serum albumin at ambient temperature. Initial rate data were fitted to the Michaelis-Menten equation using least squares regression analysis.
VII mutants were expressed in transient transfection experiments, and expression levels were determined by immunoassay. Proteolytic activities of mutants were analyzed in a functional assay at 37°C in HBS, 5 mM CaCl 2 , and 0.2% bovine serum albumin. A fixed concentration of TF/PCPS (5 pM) was saturated with increasing concentrations of mutant or wild-type VIIa, followed by addition of 50 nM factor X and determination of factor Xa generation by amidolytic assay. The maximum rate of Xa generation as a measure of the proteolytic function of the mutant VIIa in complex with TF was calculated based on calibration curves made with purified Xa.
Carbamylation of Ile 16 in VIIa-Chemical modification of the aminoterminal Ile 16 of wild-type or mutant VIIa was performed at ambient temperature according to Higashi et al. (8). Free VIIa (4 M) or VIIa (1 M) in complex with TF-(1-218) (4 M) was reacted with 0.2 M KCNO in HBS, 5 mM CaCl 2 for various times. Samples were withdrawn and diluted 25-or 60-fold for free or TF-bound VIIa, respectively, and the residual amidolytic activity was determined with 0.7 mM Chromozym tPA. Rates of inactivation were calculated from a plot of the residual activity (in percent of the initial activity) versus incubation time.
Inhibition of VIIa by Antithrombin III/Heparin-To analyze the time dependence of inhibition of free VIIa (100 nM) or VIIa (50 nM) in complex with TF-(1-218) (250 nM) by antithrombin III/heparin, wildtype and mutant VIIa were reacted with 0.5 M antithrombin III (Hematologic Technologies) in the presence of 5 units/ml unfractionated heparin (Elkins-Sinn), at 37°C in TBS, 5 mM CaCl 2 , and 0.2% bovine serum albumin. After defined times (2-60 min) samples were diluted into the chromogenic substrate Chromozym tPA (1.5 mM), and the residual amidolytic activity was immediately determined in a kinetic microplate reader.
Surface Plasmon Resonance Analysis-Binding constants for wildtype and mutant VIIa were analyzed using a BIAcore 2000 instrument (Amersham Pharmacia Biotech Biosensor). A noninhibitory anti-TF antibody (TF9 -10H10) was directly immobilized by amino-coupling to an activated dextran matrix for capture of full-length recombinant TF, as described previously (24). TF was injected to saturate the antibody, and association data were collected from injections of five concentrations (25 nM to 1 M) of VIIa in HBS, 5 mM CaCl 2 , 0.005% surfactant P20, and 3 mM CHAPS. Binding kinetics in the presence of the Kunitztype inhibitor 5L15 was determined by premixing VIIa with 10 M 5L15. Dissociation data were collected for 250 s after return to buffer flow, and the chip surface was regenerated with pulses of 0.1 M EDTA and 4 M MgCl 2 . Dissociation of TF from the antibody could not be detected over a 6-h period under the standard buffer conditions, and the measured dissociation upon injection of VIIa thus reflects dissociation of the TF⅐VIIa complex, rather than the release of TF from the immobilized antibody. Association and dissociation constants (k on and k off ) were determined using the software provided by the manufacturer.

RESULTS
Mutational Analysis of Position 156 in VIIa-As a first step to define the role of the 156 residue position in catalytic activity of VIIa, we characterized the proteolytic function of various side chain replacements in transient transfection experiments. Maximum rates of factor Xa generation were determined by saturating a fixed concentration of phospholipid-reconstituted TF with the goal of defining the permissive mutations that do not interfere with amidolytic and proteolytic function of TFbound VIIa. Changing the 156 position to Lys, as found in the highly homologous factor X, resulted in diminished function. In contrast, a Gln side chain, as found in factor IX of all species and in the majority of chymotrypsin-like serine proteases, allowed for normal or slightly enhanced proteolytic function, as compared with wild-type VIIa (Table I). Negatively charged residues, such as Glu and Asp, were consistently less well tolerated than their respective amide counterparts. In the case of Asn, ϳ60% of wild-type activity was retained. Only Gln at 156 produced a fully functional VII molecule, whereas a number of smaller side chain replacements, such as Ala, Val, or Ser, showed significantly reduced proteolytic function. Thus, replacement of the hydrophobic Met 156 side chain by the more hydrophilic side chain Gln is the only permissive mutation that allows for normal proteolytic function of the TF⅐VIIa complex.
Effect of the Met 156 to Gln Mutation on TF Binding-VII Gln156 was stably expressed, purified, and autoactivated at 4°C. Autoactivation of the mutant appeared to proceed somewhat faster than wild-type VII, but the purified protein showed electrophoretic mobility and Ͼ95% conversion to the two-chain enzyme indistinguishable from wild-type VIIa. Binding of mutant and wild-type VIIa to antibody-captured full-length TF was analyzed by surface plasmon resonance measurements. Although the mutation had only a minor Ͻ2-fold effect on the association rate, VIIa Gln156 dissociated from TF with a significantly lower rate (Table II), indicating that the cofactor binding site is in a conformation that allows for tighter binding to TF. Active site occupancy of wild-type VIIa by the Kunitz-type inhibitor 5L15 is known to tighten the binding with TF by slowing the dissociation rate (24). Although 5L15 increased affinity for wild-type VIIa 4-fold, only marginal changes were  (8,25), the decrease in the amidolytic activity can be used to determine the rate of modification of the amino terminus as measure for the stability of the Ile 16 -Asp 194 salt bridge. Carbamylation of free enzyme showed a similar rate of inactivation of VIIa Gln156 versus wild-type VIIa (6.3 Ϯ 0.4 versus 7.5 Ϯ 0.3% loss of initial activity/10 min), demonstrating that the residue replacement was not sufficient to completely order the activation pocket and to stabilize the Ile 16 -Asp 194 salt bridge. In addition, the rate of inactivation of the TF-bound mutant was also indistinguishable from wild-type VIIa (1.1 Ϯ 0.1 versus 1.4 Ϯ 0.1% loss of initial activity/10 min). Thus, the Gln replacement for Met 156 does not appreciable influence the stability of the Ile 16 -Asp 194 salt bridge in free or TF-bound enzyme.
Enhanced Amidolytic Activity of VIIa Gln156 in the Absence of Cofactor-The higher affinity binding of VIIa Gln156 to TF may reflect conformational changes in the cofactor binding site that are associated with increased catalytic function of the mutant in the absence of cofactor. To address this issue, the catalytic activities of wild-type or mutant VIIa were analyzed with small peptidyl substrates. The catalytic efficiency of hydrolysis of Chromozym tPA by free VIIa Gln156 was increased 3-fold compared with wild-type VIIa (Table III), indicating that the Gln side chain stabilizes a more active conformation of the VIIa protease domain. However, VIIa Gln156 in complex with TF cleaved the chromogenic substrate indistinguishable from wildtype VIIa. Thus, the side-chain replacement selectively influenced the catalytic function of the free enzyme.
Effect of the Met 156 to Gln Mutation on Inhibition by Antithrombin III-To test whether inhibitor binding to the active site of VIIa Gln156 was also enhanced in comparison to wild-type VIIa, inhibition of VIIa activity by a fixed concentration of antithrombin III in the presence of 5 units/ml heparin was studied. Consistent with previous studies (26,27), the amidolytic activity of free VIIa was inefficiently inhibited over time by antithrombin III/heparin (Fig. 1). VIIa Gln156 was inhibited at a rate that was approximately twice as fast as the rate of inhibition of wild-type VIIa (50% inhibition in 31 Ϯ 3 min for mutant versus 64 Ϯ 12 min for wild-type VIIa, n ϭ 3). Inhibition of TF⅐VIIa complexes occurred with significantly faster rates for both mutant and wild-type VIIa, reaching 50% inhi-bition at 4.5 Ϯ 1.8 min for mutant versus 6.0 Ϯ 1.9 min for wild-type VIIa (n ϭ 3). The somewhat more efficient inactivation of VIIa Gln156 in the presence of TF suggests improved interaction of the serpin with the mutant. This may involve exosite interactions, because mutation of the 156 position in tPA also influenced the interaction with plasminogen activator inhibitor 1 (18). Thus, the enhanced inhibition of free VIIa Gln156 by antithrombin III cannot solely be attributed to a stabilized active conformation of the mutant but may also involve direct inhibitor binding to the newly introduced Gln side chain.
Increased Proteolytic Function of Free VIIa upon Replacement of Met 156 with Gln-Factor X activation by free VIIa was analyzed in the presence and absence of a negatively charged phospholipid surface (PCPS). In both cases, activation of factor X by VIIa Gln156 was enhanced 9-fold compared with wild-type VIIa (Table IV). However, only subtle changes in the activation of factor X by VIIa Gln156 in complex with phospholipid-reconstituted TF were detected (Table IV), both in regard to K m and k cat . These data demonstrate that macromolecular substrate binding as well as scissile bond cleavage are not affected by the mutation after complex formation with the catalytic cofactor and after acquisition of full catalytic activity. In free VIIa, however, the catalytic function toward the macromolecular substrate is enhanced, consistent with the data for small substrate hydrolysis that also demonstrated selectively increased amidolytic activity of free VIIa Gln156 .

DISCUSSION
The zymogen-like conformation of VIIa is required to subject the regulation of VIIa's catalytic activity under strict control of the cell surface expression of the cofactor TF. The labile enzyme conformation of VIIa allows for the presence of significant amounts of the two-chain enzyme in the circulating blood and thus for the initiation of the coagulation pathways, as soon as the cofactor is exposed upon disruption of vascular integrity. This mechanism of triggering an enzyme cascade differs from the activation of the fibrinolytic system by tPA, which utilizes a catalytically active zymogen for the initial proteolytic events. Whereas an activating salt bridge of Lys 156 with Ile 16 has been defined as the structural basis for the active zymogen conformation of tPA (17,18), structural determinants for the labile enzyme conformation of VIIa remained unclear from previous structural and mutational studies. Here, we provide evidence that the hydrophobic Met 156 residue side chain within the activation pocket, in part, is responsible for the zymogen-like conformation of the VIIa protease domain.
Replacement of Met 156 by Gln, the side chain found in factor IX, did not appreciably influence catalytic function of the TF-bound enzyme toward small chromogenic substrates or the macromolecular substrate factor X. Thus, the cofactorstabilized, active conformation of VIIa does not absolutely require a Met side chain at this position. Backbone-superimposition of the structures of porcine factor IXa (28) and VIIa  in the TF⅐VIIa complex (9) demonstrate that essentially the same space is occupied by the Met and Gln side chains in the respective proteases, providing a structural rational why the Gln replacement is permissive for function of the TF-bound VIIa. However, free VIIa Gln156 displayed increased function compared with wild-type VIIa. First, the mutant had higher affinity for TF and affinity was not appreciably influenced by active site inhibitor binding that typically slows dissociation for wild-type VIIa and for most of the previously characterized mutants (15). The cofactor binding site of VIIa Gln156 thus appears to be altered, indicating a long range conformational change of the residue replacement of Met 156 by Gln in the activation pocket. Second, VIIa Gln156 had higher amidolytic and proteolytic activity in the absence of cofactor. Amidolytic activity was enhanced 3-fold, although activation of factor X was increased 9-fold in the presence or absence of phospholipid. The difference in enhancement may result from an additional ordering of regions that influence macromolecular substrate scissile bond cleavage, but not the hydrolysis of p-nitroanilide chromogenic substrates, which bind primarily to the S1-S3 subsites of the catalytic cleft. Because stabilization of the amino-terminal Ile 16 -Asp 194 salt bridge is generally considered to be the major determinant for the activation state of the VIIa protease domain, it was surprising that chemical modification of the amino-terminal Ile 16 in free VIIa Gln156 was little different from wild-type VIIa. How-ever, we have found poor correlation of cofactor-mediated catalytic enhancement and susceptibility of the amino terminus to chemical modification in a number of other VIIa mutants (29). Most importantly, certain mutations in the TF binding interface of the VIIa protease domain resulted in a complete failure to show protection of the amino terminus upon cofactor binding, but TF was still able to enhance amidolytic activity of these VIIa mutants by Ͼ10-fold. Stabilization of the insertion of the amino terminus, as measured by the susceptibility to chemical modification, is thus not absolutely required for changes in catalytic function of VIIa to occur.
The local environment at the 156 position in VIIa differs from other homologous coagulation factors such as factor IXa. Although porcine factor IXa has a hydrophobic Ile residue in the 154 position along with a pair of charged residues at position 21 (Asn) and 156 (Gln), VIIa has a charged Glu 154 and a pair of hydrophobic residues, i.e. Val 21 and Met 156 ( Fig.  2A). One or two water molecules (not shown) are consistently resolved between the Glu 154 and Met 156 side chain in structures of the TF⅐VIIa complex (9, 10). Fig. 2B shows Met 156 replaced by Gln in one possible rotamer position that orients the amide of the Gln side chain toward Glu 154 . The low catalytic function of acidic side chain replacement mutants and the high functional activity of amide counterparts (Table  I) indirectly supports the notion that such an orientation of the introduced Gln 156 is most compatible with optimal activity. The charge complementarity of the Gln 156 and Glu 154 side chains may stabilize the local conformation of this region in free VIIa Gln156 , possibly involving the coordination of a water  molecule between the two side chains. The activating effect of the Gln 156 mutation may thus be mediated through an effect on the Glu 154 position rather than the adjacent amino-terminal insertion.
In wild-type VIIa, Ala mutation of Glu 154 reduces amidolytic function 3-fold and proteolytic function 6-to 10-fold, displaying some divergence of the mutational effect on amidolytic versus proteolytic function (30). The free Gln 156 mutant showed functional enhancement of similar magnitude in both assays, arguing in favor of a function linkage between Glu 154 and Gln 156 in VIIa Gln156 . Glu 154 together with the adjacent Leu 153 side chain are within a "hot spot" that is known to allosterically regulate VIIa's catalytic activity. First, Glu 154 is located within the epitope of an inhibitory monoclonal antibody and crucial for the inhibitory allosteric switch induced by this exosite inhibitor (30). Second, a small peptide inhibitor of VIIa binds to an epitope that includes Leu 153 (13). This exosite inhibitor influences catalytic function by a major reorientation of the 142-154 "autolysis" loop that also contains specific residues involved in extended macromolecular substrate recognition (20). The 142-154 loop is considered part of the activation domain, and stabilization and ordering of this loop is essential for the zymogen to enzyme transition. Structural studies on prothrombin derivatives (31) further suggest that the 142-154 loop can behave as a partially autonomous region of the activation domain. The enhanced function of VIIa Gln156 in the absence of a stabilization of the amino-terminal insertion is thus consistent with a conformational effect on the 142-154 loop through the proximity of Glu 154 with Gln 156 .
This study identifies the first residue side chain that contributes to the zymogen-like state of free VIIa. Although the single side chain replacement was insufficient to confer complete spontaneous ordering and full catalytic activity of the VIIa protease domain, the proposed stabilization of the 142-154 loop would achieve one of the necessary conformation orderings in the activation domain. The significantly improved function of free VIIa Gln156 argues that a limited number of side chains may be responsible for retaining VIIa in the zymogen-like conformation. Whether changing the local environment of the activation pocket is sufficient to achieve a fully active VIIa or whether changes in the unique loop segments that contact TF are also required remain important questions for further studies.