The cofactor function of the N-terminal domain of tissue factor.

Tissue factor (TF) is an integral membrane protein cofactor for factor VIIa (fVIIa) that initiates the blood coagulation cascade during vascular injury. TF has two fibrinonectin type III-like domains, both of which make extensive interactions with both the light and heavy chains of fVIIa. In addition to interaction with fVIIa, the membrane proximal C-terminal domain of TF is also known to bind the natural substrates factors IX and X, thereby facilitating their assembly and recognition by fVIIa in the activation complex. Both fVIIa and TF are elongated proteins, and their complex appears to be positioned nearly perpendicular to the membrane surface. It is possible that, similar to fVIIa, the N-terminal domain of TF also contacts the natural substrates. To investigate this possibility, we substituted all 23 basic and acidic residues of the N-terminal domain of TF with Ala or Asn and expressed the mutants as soluble TF(2-219) in a novel expression/purification vector system in the periplasmic space of bacteria. Following purification to homogeneity, the cofactor properties of mutants in promoting the amidolytic and proteolytic activity of fVIIa were analyzed in appropriate kinetic assays. The amidolytic activity assays indicated that several charged residues spatially clustered at the junction of the N- and C-terminal domains of TF are required for high affinity interaction with fVIIa. On the other hand, the proteolytic activity assays revealed that none of the residues under study may be an interactive site for either factor IX or factor X. However, it was discovered the Arg(74) mutant of TF was defective in enhancing both the amidolytic and proteolytic activity of fVIIa, suggesting that this residue may be required for the allosteric activation of the protease.

Tissue factor (TF) is an integral membrane protein cofactor for factor VIIa (fVIIa) that initiates the blood coagulation cascade during vascular injury. TF has two fibrinonectin type III-like domains, both of which make extensive interactions with both the light and heavy chains of fVIIa. In addition to interaction with fVIIa, the membrane proximal C-terminal domain of TF is also known to bind the natural substrates factors IX and X, thereby facilitating their assembly and recognition by fVIIa in the activation complex. Both fVIIa and TF are elongated proteins, and their complex appears to be positioned nearly perpendicular to the membrane surface. It is possible that, similar to fVIIa, the N-terminal domain of TF also contacts the natural substrates. To investigate this possibility, we substituted all 23 basic and acidic residues of the N-terminal domain of TF with Ala or Asn and expressed the mutants as soluble TF 2-219 in a novel expression/purification vector system in the periplasmic space of bacteria. Following purification to homogeneity, the cofactor properties of mutants in promoting the amidolytic and proteolytic activity of fVIIa were analyzed in appropriate kinetic assays. The amidolytic activity assays indicated that several charged residues spatially clustered at the junction of the N-and C-terminal domains of TF are required for high affinity interaction with fVIIa. On the other hand, the proteolytic activity assays revealed that none of the residues under study may be an interactive site for either factor IX or factor X. However, it was discovered the Arg 74 mutant of TF was defective in enhancing both the amidolytic and proteolytic activity of fVIIa, suggesting that this residue may be required for the allosteric activation of the protease.
Tissue factor (TF) 1 is an integral membrane cofactor that upon exposure to circulating blood binds with high affinity to factor VIIa (fVIIa) to catalyze the rapid activation of procoagulant zymogens factors IX and X, thereby initiating the blood clotting cascade (1)(2)(3). The structure of TF is composed of two fibrinonectin type III-like extracellular domains, a single membrane-spanning domain and a short cytoplasmic tail (4,5). Although all three domains of TF are required for the physiological function of the cofactor, previous studies have indicated that the two extracellular domains expressed by recombinant DNA methods as a soluble protein (sTF) can bind to fVIIa with a high affinity to enhance both the amidolytic and proteolytic activities of the protease (6 -9). The crystal structure of sTF either alone or in complex with fVIIa has been determined (4,10). The structural data have indicated that the extracellular domains of TF make extensive interactions with both the light and heavy chains of fVIIa (10). It is believed that these interactions allosterically change the conformation of the active-site pocket of fVIIa, leading to a dramatic improvement in the catalytic efficiency of the protease toward both synthetic and natural macromolecular substrates (1). Most of the functionally critical residues of TF have been mapped by the Ala-scanning mutagenesis approaches (7,9,11). The characterization of these mutants has indicated that, in addition to interaction with fVIIa, the C-terminal membrane proximal domain of TF also provides binding sites for the ␥-carboxyglutamic acid and/or the first epidermal growth factor-like domains of factors IX and X thereby facilitating the assembly and optimal recognition of these substrates by fVIIa in the activation complex (11,12). Thus, it has been demonstrated that the interaction of the ␥-carboxyglutamic acid domains of both substrates with the two basic residues Lys 165 and Lys 166 in the membrane proximal C-terminal domain of TF makes a significant contribution to the specificity of the zymogen activation by fVIIa (11).
The structural and fluorescence energy transfer studies data have further indicated that the binding of TF to fVIIa stabilizes the protease in a topological orientation in which the active site pocket of fVIIa is maintained far above the membrane surface in the activation complex (10,13). Thus, for effective activation by fVIIa, both factors IX and X are also required to assemble into the activation complex in a similar extended conformation to maintain the activation peptide of the substrates at the same distance above the membrane surface. Noting that several charged residues of the N-terminal domain of TF make extensive interactions with the protease domain of fVIIa, we hypothesized that these residues may also interact with the protease domains of the substrates in the activation complexes. To test this hypothesis, we substituted all 23 basic and acidic residues of the TF Nterminal domain with Ala and expressed the mutant constructs in soluble sTF 2-219 forms in a newly developed expression/ purification vector system in the periplasmic space of bacteria. The Ala substitution mutants of two constructs (Glu 56 and Asp 58 ) were not expressed to a high yield, thus these residues were substituted with Asn. All of the mutants were purified to homogeneity and characterized with respect to their ability to function as cofactors to enhance the amidolytic and proteolytic activities of fVIIa using standard kinetic assays. The results of the amidolytic activity assays indicated that a number of basic and acidic residues of the N-terminal domain, centered spatially on the middle section of the cofactor near its junction with the C-terminal domain, are required for the high affinity interaction of the cofactor with fVIIa. This finding is consistent with previous similar mutagenesis studies (7). On the other hand, detailed kinetic analysis in the presence of saturating concentrations of the sTF 2-219 mutants suggested that the N-terminal domain of the cofactor may have no interactive site for either factor IX or factor X in the related activation complexes. However, we identified Arg 74 as a unique residue that is required for the catalytic function of fVIIa in reaction with both synthetic and macromolecular substrates. The mutagenesis of Arg 74 markedly impaired both the amidolytic and proteolytic functions of fVIIa, suggesting that this residue may be required for the allosteric activation of the catalytic pocket of the protease.

EXPERIMENTAL PROCEDURES
Construction, Mutagenesis, and Expression of Soluble Tissue Factor in Bacteria-Construction and expression of soluble TF lacking both the trans-membrane and cytoplasmic domains (sTF 2-219 ) in the pINIII-pelB bacterial periplasmic expression/purification vector system has been described previously (6). To improve the expression yield of sTF, this vector was modified by inserting a promoterless neomycin cDNA lacking the Shine-Dalgarno sequence immediately downstream of the sTF 2-219 stop codon. This construction strategy translationally coupled the expression of the neomycin cDNA to that of sTF. To facilitate purification, the 12-residue epitope for the Ca 2ϩ -dependent monoclonal antibody, HPC4, was also linked in-frame to the pelB signal peptide as described previously (6). The Ala substitution mutants of sTF 2-219 were prepared by PCR mutagenesis methods in the same vector system as described previously (6). After confirmation of the accuracy of the mutagenesis by DNA sequencing, the mutant constructs were transformed into the BL21 strain of Escherichia coli and selection was carried out in kanamycin as described previously (6). Bacterial colonies harboring the wild-type and mutant sTF 2-219 derivatives were grown in LB medium containing 50 g/ml kanamycin in 1-liter flasks at 37°C to a cell density of A 600 ϭ 0.6 -1.0 and then were induced by adding isopropyl-␤-Dthiogalactopyranoside to a final concentration of 1 mM. The incubation was continued overnight (ϳ12 h) at room temperature (ϳ25°C). The supernatant was collected by centrifugation at 4000 ϫ g for 30 min, and the periplasmic extract was prepared by hypotonic shock of bacterial pellet in 100 ml of H 2 O for 30 min and centrifugation at 10,000 ϫ g for 15 min. The supernatant and the periplasmic extract were mixed and supplemented with 20 mM Tris-HCl (pH 7.5), 10 mM benzamidine, 0.02% NaN 3 , and 5 mM CaCl 2 and chromatographed on the HPC4 monoclonal antibody immobilized on Affi-Gel 10 (Bio-Rad) as described previously (6). The sTF 2-219 concentrations were determined from the absorbance at 280 nm assuming a molecular weight of 27,000 and extinction coefficient (E 1 cm 1% ) of 14.8 and by an amidolytic activity assay as described previously (6).
Evaluation of the K d(app) Values-The affinity of the sTF 2-219 mutants for binding to fVIIa was evaluated based on their ability to enhance the catalytic activity of fVIIa toward cleavage of Spectrozyme VIIa. Increasing concentrations of the cofactor (0.78 -100 nM for wildtype and up to 12 M for the mutants) were incubated with fVIIa (5 nM) in 100 mM NaCl, and 20 mM Tris-HCl (pH 7.5) containing 0.1 mg/ml bovine serum albumin, 0.1% polyethylene glycol 8000, and 5 mM CaCl 2 (TBS/Ca 2ϩ ) in 96-well assay plates. After 1 min of incubation at room temperature, Spectrozyme VIIa was added to a final concentration of 0.5 mM and the rate of hydrolysis was measured at 405 nm on a kinetic microplate reader (Molecular Devices, Menlo Park, CA). The background activity of fVIIa in the absence of sTF 2-219 was subtracted from the measured values. The maximal activity and K d(app) values were calculated from the non-linear curve fitting of amidolytic activity data to the Langmuir isotherm and/or the quadratic binding equation as described previously (15).
Factor X Activation-The steady-state kinetics of factor X activation by fVIIa in complex with wild-type or mutant sTF 2-219 derivatives were studied in TBS/Ca 2ϩ at room temperature as described previously (16). The generation of product (factor Xa) was monitored by a two-stage discontinuous assay. In the first stage, fVIIa (0.3 nM), PC/PS vesicles (0.3 mM), and saturating concentrations of sTF 2-219 mutants (at least 10ϫ K d(app) values) were incubated for 10 min in 96-well assay plates and the reactions then were initiated by adding increasing concentrations of factor X (20 -2630 nM). After 10 min of incubation at room temperature, the reactions were terminated by adding 50 mM EDTA. In the second stage, the amidolytic activity of each sample was determined by the subsequent addition of Spectrozyme FXa (a final concentration of 0.1 mM). The absorbance at 405 nm was monitored over 5 min using a V max kinetic microplate reader, and the initial rates of chromogenic substrate hydrolysis (⌬A 405 /min) were converted to the nanomolar of product by reference to a standard curve prepared with purified human factor Xa. The apparent K m and k cat values for substrate hydrolysis were calculated from the Michaelis-Menten equation, and the catalytic efficiencies were expressed as the ratio of k cat /K m .
Factor IX Activation-Steady-state kinetics of factor IX activation by fVIIa in complex with wild-type and mutant sTF 2-219 derivatives was performed by the same methods described above for factor X. fVIIa (5 nM), PC/PS vesicles (0.3 mM), and saturating concentrations of sTF 2-219 mutants (at least 10ϫ K d(app) values) were incubated for 10 min, and the reactions then were initiated by adding increasing concentrations of factor IX (27-3510 nM). After 50 min of activation at an ambient temperature, the reactions were terminated by adding 50 mM EDTA. The rate of factor IXa generation was determined by an amidolytic activity assay using CBS 31.39 containing 33% ethylene glycol as described previously (17). The initial rates of chromogenic substrate hydrolysis (⌬A 405 /min) were converted to nanomolar of product by reference to a standard curve prepared with purified human factor IXa. The apparent K m and k cat values for substrate hydrolysis were calculated as described above.
Competitive Binding Studies-The competitive effect of the Arg 74 to Ala mutant of sTF 2-219 , exhibiting impaired cofactor activity toward macromolecular substrates, on factor X activation by the wild-type fVIIa⅐sTF 2-219 complex was studied. In this case, the activation of factor X (500 nM) by a limiting fixed concentration of fVIIa (2 nM) in complex with PC/PS vesicles (100 g/ml) and 100 nM sTF 2-219 was monitored in TBS/Ca 2ϩ in the presence of increasing concentrations the sTF 2-219 mutant (0.156 -10 M). Following incubation for 5 min at room temperature, the reactions were terminated by adding 50 mM EDTA and the rate of factor Xa generation was determined as described above.

RESULTS AND DISCUSSION
Expression and Purification of sTF  Derivatives-All of the sTF 2-219 derivatives were expressed in the periplasmic space of bacteria using the pINIII-pelB-neomycin expression/purification vector system as described under "Experimental Procedures." This vector system, which incorporates a 12-residues epitope for the Ca 2ϩ -dependent monoclonal antibody HPC4 to the Nterminal of the expressed protein, is an improved version of pINIII-pelB, which we previously used to express and purify sTF 2-219 from bacterial culture (6). In the current study, we inserted a promoterless neomycin cDNA immediately downstream from the sTF 2-219 stop codon. This strategy improved the expression yield to greater than 20 mg/liter sTF 2-219 . Wild-type and mutant proteins were separated from the bacterial culture supernatants and periplasmic extracts by a single step immunoaffinity chromatography on immobilized HPC4 as described previously (6). With the exception of Glu 56 and Asp 58 , all of the basic and acidic residues of the N-terminal domain of sTF 2-219 were substituted with Ala in individual constructs. The Ala substitution mutants of both Glu 56 and Asp 58 did not express to high yield; thus, these residues were substituted with Asn. SDS-PAGE analysis of the purified proteins under non-reducing conditions indicated that the isolated proteins were essentially homogenous (Fig. 1). We previously demonstrated that the bacterial sTF 2-219 with or without an N-terminal HPC4 epitope exhibits identical cofactor activity as the mammalian sTF 1-219 (6). Thus, the cofactor activities of all of the sTF 2-219 derivatives were evaluated without cleaving the HPC4 epitope from the N termini of the recombinant proteins.
Evaluation of the K d(app) of sTF  Mutants for Binding to fVIIa-TF is known to enhance the amidolytic activity of fVIIa toward tripeptidyl chromogenic substrates 30 -100-fold (1). This property of the cofactor was utilized to evaluate the affinity of sTF 2-219 derivatives for binding to fVIIa. In agreement with previous results, wild-type sTF 2-219 interacted with fVIIa with a K d(app) of 5.3 nM (Table I). On the basis of their importance for interaction with fVIIa and improving the catalytic efficiency of the protease, the sTF 2-219 mutants can be functionally classified into three general categories: 1) mutants with impaired affinity but with normal cofactor activity; 2) mutants with both impaired affinity and cofactor activity; and 3) mutants with improved cofactor activity. Among the first class, at least five mutants (Lys 20 , Glu 24 , Asp 44 , Lys 48 , and Asp 58 ) exhibited the greatest loss of affinity for fVIIa, suggesting that these residues play a critical role either directly or indirectly in interaction with fVIIa (Table I). However, at saturating concentrations of the mutant cofactors, fVIIa exhibited nearly similar maximal amidolytic activity. It should be noted that, with the D58N mutant, no K d(app) could be measured because the amidolytic activity of fVIIa in the presence of the mutant cofactor remained linear for up to 12 M sTF 2-219 mutant. Thus, the results with this mutant cannot be analyzed with certainty. The functionally important residues of the TF N-terminal domain were found to be spatially clustered at the junction of the N-and C-terminal domains of TF (Fig. 2). Mutagenesis of these residues and the use of synthetic peptides based on this region have been reported to dramatically impair the cofactor activity of fVIIa (7, 18 -20) Results of this study suggest that fVIIa has a normal amidolytic activity in the presence of saturating concentrations of these sTF 2-219 mutants, suggesting that the catalytic defect with these mutants is solely due to their loss of interaction with the protease. In the case of Asp 58 , substitution with Asn dramatically reduced the affinity of sTF for fVIIa. Interestingly, when Lee and Kelley (21) used phage-display technology to create an altered version of sTF that retained wild-type binding affinity for fVIIa but which failed to enhance the proteolytic activity of fVIIa, they found that their mutants contained exclusively a substitution of Trp for Asp at position 58 of sTF.
In the second class of mutants, we identified a new functionally important residue, Arg 74 , which was not included in the previous Ala-scanning mutagenesis study of TF (7). The mutagenesis of this residue resulted in both impaired affinity and cofactor activity for the sTF 2-219 mutant. As shown in Table I, relative to wild-type, the K d(app) of fVIIa interaction with the R74A was elevated ϳ190-fold. Moreover, unlike a normal amidolytic activity with the other mutants, the maximal activity of fVIIa in the presence of a saturating concentration of R74A was also impaired ϳ2-fold (Table I). These results suggest that the interaction of the side chain of Arg 74 with fVIIa may be responsible for allosterically changing the conformation of the active site pocket of fVIIa, leading to improvement in the catalytic efficiency of the protease. Thus, a cofactor-mediated  conformational change in the active site pocket of fVIIa may be, at least partially, mediated through protease interaction with Arg 74 of TF. Indeed, from the crystal structure of the sTF⅐fVIIa complex, residue Arg 74 appears to make contact with residues in the protease domain of fVIIa (10). In contrast to the loss of affinity and cofactor function for most of the sTF 2-219 mutants, the mutagenesis of three acidic residues and one basic residue of sTF 2-219 around the 61-66 sequence resulted in an improved affinity for fVIIa and also improved maximal amidolytic activity (Table I). These residues are located on the opposite surface of the cofactor and thus are not expected to contact fVIIa (although Asp 61 might be able to make contacts with fVIIa). It is not known how the neutralization of the charges of these residues improves the cofactor function of the molecules, although interestingly, these residues are all located within one of the very few ␣-helical regions on sTF.
Proteolytic Function-The cofactor function of the sTF 2-219 mutants in enhancing the proteolytic activity of fVIIa toward the natural substrates factors IX and X was also studied. The kinetic parameters K m and k cat for the activation of both substrates by fVIIa in complex with the sTF 2-219 mutants are presented in Table II for factor IX and Table III for factor X. Note that K m and k cat values for the activation of factors IX or X by the fVIIa⅐sTF 2-219 complex are reported here as apparent values. This is because both K m and k cat change as a function of phospholipid concentration due to partitioning of enzyme and substrate between the solution and membrane-bound phases as described in detail by Fiore et al. (22). All of the cofactor mutants with the exception of D58N and R74A exhibited near normal or improved cofactor activity in promoting the fVIIa activation of both substrates. The kinetic basis for the defect with the D58N mutant cannot be readily analyzed, because the affinity of this mutant for fVIIa was dramatically impaired and the activation assay was carried out at a subsaturating concentration of the mutant due to a limited amount of the cofactor. However, a similar previous study with the same mutant of TF in the full-length form has indicated that the Asp 58 mutant has only impaired affinity for fVIIa with a normal cofactor function (23). This has been evidenced by the observation that fVIIa in complex with a saturating concentration of the Asp 58 mutant exhibits a normal maximal activity (23). On the other hand, as shown in Fig. 3, the proteolytic cofactor function of the R74A mutant in activation of both factors IX and X by fVIIa was markedly impaired. The extent of impairment in the cofactor function of the R74A mutant was ϳ2.5and ϳ6-fold for the activation of factors IX and X, respectively. In both cases, the primary defect in the cofactor function was due to impairment in the k cat of the activation reaction (Tables II and III). These results further support the hypothesis that the side chain of Arg 74 may be in contact with a region of fVIIa that is allosterically linked to the active site pocket of the protease. The Ala substitution mutants of an acidic region in most C-terminal end of the TF N-terminal domain resulted in the improvement in the cofactor function of the TF mutants in the factor X but not factor IX activation assay (Table III). Thus, the catalytic efficiency of fVIIa in complex with the E105A mutant of sTF 2-219 was improved 3.5-fold (Table III). It is likely that the charge neutralization of this residue eliminates a repulsive interaction of an acidic region of factor X (most likely ␥-carboxyglutamic acid and/or EGF1 domain) with the fVIIa⅐sTF complex.
To provide further evidence that Arg 74 is critical for the proteolytic function of fVIIa, the R74A mutant was used as a competitive inhibitor of factor X activation by fVIIa in complex with wild-type sTF 2-219 . As shown in Fig. 4, R74A inhibited the activation of factor X by fVIIa in complex with wild-type sTF 2-219 in  a concentration-dependent manner, and at a high concentration of the cofactor mutant, which was sufficient to displace most of the wild-type cofactor, the rate of factor X activation declined ϳ2-fold, paralleling the rate of the substrate activation by the fVIIa in complex with the R74A mutant. These results suggest that the role of Arg 74 in the cofactor function of TF is distinct from other charged residues under study and may confirm the hypothesis that it interacts with a site of fVIIa that is in allosteric linkage with the active site pocket of the protease. Previously, a similar cofactor role for the Asp 44 of TF has also been postulated (8). In that study, an ϳ4-fold impaired cofactor function for the D44A mutant of sTF in factor X activation by fVIIa has been reported (8). However, in the current study, we noted an ϳ2-fold impairment in the k cat of factor X activation by fVIIa in complex with the D44A mutant of sTF 2-219 and this defect was partially compensated by an improvement in the K m(app) of the activation reaction (Table III). Moreover, unlike the defective cofactor function of the R74A mutant in the activation of both macromolecular substrates, the D44A mutant exhibited near normal cofactor function in the activation of factor IX (Table II). The differences in the results of two studies may be due to different experimental conditions used to evaluate the cofactor effect of this mutant. The previous study used a limiting concentration of TF in complex with excess fVIIa to evaluate the cofactor function of the mutant at 37°C, whereas our studies were carried out with a limiting concentration of fVIIa in complex with an excess of the sTF 2-219 mutant at room temperature. However, despite these differences, the observation that the k cat of factor X activation with D44A mutant was impaired ϳ2-fold is in agreement with the previous hypothesis that Asp 44 of TF is critical for the proteolytic function of fVIIa in the extrinsic factor Xase complex.
In summary, we have mapped the functionally important basic and acidic residues of the N-terminal domain of TF and demonstrated that most of the charged residues critical of fVIIa binding are spatially located at the junction of the N-and C-terminal domains of TF. The results further suggest that, unlike the C-terminal domain of TF, which contains direct binding sites for interaction with both factors IX and X, the N-terminal domain of the cofactor may not interact with either substrate of fVIIa in the activation complex. Finally, we have identified Arg 74 as a critical residue that appears to directly interact with a region of fVIIa that is in allosteric linkage with the active site pocket of the protease. The reaction was initiated by the addition of different concentrations of factor IX shown on the x axis. The reactions were stopped by the addition of 50 mM EDTA, and the rate of factor IXa generation was measured by an amidolytic activity assay described under "Experimental Procedures." B, the same as A with the exception that the rate of activation of factor X by fVIIa (0.3 nM) and PC/PS vesicles (0.3 mM) was measured in the presence of saturating concentrations of wild-type (E) or R74A sTF 2-219 (q) in TBS/Ca 2ϩ . Solid lines in both panels are non-linear regression analysis of data according to the Michaelis-Menten equation.
FIG. 4. Competitive effect of R74A sTF 2-219 on factor X activation by fVIIa in complex with wild-type sTF 2-219 . Factor X (500 nM) was incubated with fVIIa (2 nM) in complex with PC/PS vesicles (100 g/ml) and wild-type sTF 2-219 (100 nM) in TBS/Ca 2ϩ in the presence of increasing concentrations of wild-type (WT) (E) or the R74A mutant of sTF 2-219 (q) (0.156 -5 M). Following incubation for 5 min at room temperature, the reactions were terminated by adding 50 mM EDTA and the rate of factor Xa generation was determined as described under "Experimental Procedures."