Cofactor residues lysine 165 and 166 are critical for protein substrate recognition by the tissue factor-factor VIIa protease complex.

High affinity binding of factor VIIa (VIIa) to its cellular receptor tissue factor (TF), as well as association of factor X with phospholipid are required for optimal assembly of the extrinsic activation complex. In addition to the interactions of substrate with phospholipid and enzyme, we here provide evidence that cofactor residues Lys-165 and Lys-166 specifically contribute to the recognition of macromolecular substrate. Ala for Lys replacement in TFA165A166 was compatible with high affinity binding of VIIa when analyzed on cell surfaces as well as in the absence of phospholipid. Dissociation of TFA165A166.VIIa did not occur with a faster rate compared to TF.VIIa, further supporting unaltered VIIa binding function of TFA165A166. Cleavage of chromogenic peptidyl substrate by TFA165A166.VIIa complexes was not diminished, demonstrating that TFA165A166 supported enhancement of catalytic function of the VIIa protease domain. In contrast, factor X activation was reduced in the presence and absence of phospholipid. Further, TFA165A166 effectively competed with wild-type TF in the cleavage of factor X at limited VIIa concentrations. Selective reduction in macromolecular substrate hydrolysis combined with normal VIIa binding by TFA165A166 indicates that the cofactor TF does contribute, either directly or indirectly via specific interactions with VIIa, to factor X recognition.

whether T F is inserted into a phospholipid bilayer or is in solution suggesting that protein-protein interactions are sufficient for tight binding of the protease to its receptor (4). High affinity binding of VIIa to TF is dependent on a properly conformed and Ca2+-saturated Gla domain of VIIa (4, 5), as well as, other sites of VIIa (4, 6,7) and is a requisite for optimal recognition and proteolytic activation of protein substrate by the TF.VIIa complex (4). Substrate assembly with cofactor-enzyme complexes in the coagulation pathways may require either substrate-phospholipid or substrate-cofactor interactions. The former is exemplified by the prothrombinase complex assembly (8) which is enhanced by local concentration of the reactants on the phospholipid surface. In contrast, protein-protein interactions of the substrate protein C with the cofactor thrombomodulin are critical for substrate recognition by the thrombomodulin-thrombin complex, as demonstrated by the dissection of enzyme-binding sites and critical regions for cofactor function using proteolytic fragmentation (9)(10)(11) or recombinant deletion proteins based on the domain structure (12, 13). Phospholipid is required for full functional activity of the TF.VIIa complex. This reflects, at least in part, the contribution of substrate-phospholipid interactions, as demonstrated by the preferential activation of phospholipid-associated factor X by soluble TF. VIIa (14). In addition to these structural requirements for assembly and substrate recognition by TF . VIIa, we now identify 2 amino acid residues that are required for optimal recognition and cleavage of protein substrate by TF. VIIa, but are not required for binding of VIIa or the induced enhancement of function of the catalytic site of VIIa.

EXPERIMENTAL PROCEDURES
Reagents-Coagulation proteins and anti-TF mAbs were purified as described (14). Cross-linking reagents were obtained from Pierce Chemical Co. and Sequenase from U. S. Biochemical. VIIa was from Novo Nordisk (Gentofte, Denmark) and the purification and characterization of this recombinant protein has been previously described (15,16). The binding constants of this recombinant VIIa for TF in the absence of phospholipid have been shown to be similar to those of plasma-derived VI1 ( 4 ) .
Site-directed Mutagenesis and Generation of Stable Cell Lines-Site-directed mutagenesis was performed with the T F coding sequence inserted in CDM8 using the uracil substitution method described by Kunkel(l7) with modifications as previously reported (18). To generate TFA16SA166, the oligonucleotide AAGTTCAGGCm GCAACAGCCAAAA was synthesized which introduced the triplets AAGTTCAGGCgAAAACAGCC and TGGAAATCTTCCTCG-(underlined) for Ala in place of Lys-165 and Lys-166. Similarly, AGCGGAAAGCAACAGCCAAA were used to generate the single Ala substitutions for Lys-165 and Lys-166, respectively. Purified plasmid DNA was sequenced using Sequenase and transfected into Chinese hamster ovary cells using the calcium phosphate method. In transient transfection experiments, cells were harvested after 2 days for analysis. To establish stable cell lines, pMAMneo was cotrans-6375 fected with the T F coding sequence, and clones were selected for resistance to the neomycin analogue G418 (GIBCO). Stable clones were tested for epitope expression of a panel of anti-TF mAbs using flow cytometry as described (18).
Chemical Cross-linking of TF-Chemical cross-linking of T F on viable cell surfaces was performed as described by Roy et al. (19) with the following modifications. The water-soluble, non-reducible, homobifunctional, and amino-reactive cross-linking reagent Bis(sulfosuccinimidy1)suberate (BS3) and the reducible reagent dithiobis(sulfosuccinimidy1propionate) (DTSSP) were used at 5 mM in HEPES-buffered saline (HBS, 10 mM HEPES, 140 mM NaC1, pH 7.4). Cells (5-10 X IO6 cells/ml) were incubated with cross-linker or buffer control for 1 h on ice followed by washing with 100 mM Tris, pH 7.4, to quench the reaction. Samples which were not cross-linked and those which were cross-linked with DTSSP were applied to sodium dodecyl sulfate-polyacrylamide gels under non-reducing conditions; samples cross-linked with BS3 were reduced prior to gel electrophoresis on 8% sodium dodecyl sulfate-polyacrylamide gels (20). TF was detected using a polyclonal primary antibody which had been affinity purified on immobilized TF, followed by alkaline phosphatase-conjugated secondary antibody and color development as described previously (21).
Extraction of TF from Cells-TF or TFA166A166 were extracted from cell lysate. Frozen cells were subjected to the delipidation and solubilization protocol which had previously been used for the purification of TF from brain (22). This included the preparation of an acetone powder by extracting the cells five times with acetone followed by heptanebutanol (21, v/v) extraction of the acetone powder to remove phospholipid. The protein pellet was then extracted with 0.1% Triton X-100 to reduce contaminating proteins, and TF was solubilized in a final extraction with 10 mM CHAPS in TBS (20 mM Tris, 140 mM NaCI, pH 7.4). The concentration of T F and TF~16~~166 in this extraction was determined by ELISA (14) using two mAbs which have been shown to be reactive with both molecules by flow cytometry.
Functional Characterization of TF Mutants-Initial characterization of the TF mutants was performed in a one-stage clotting assay. Briefly, cells were lysed at 2 X lo6 cells/ml with 15 mM octyl-6-Dglucopyranoside in HBS at 37 "C for 20 min followed by a %fold dilution with HBS (23). Clotting times were determined for these cell lysate in a one-stage clotting assay (equal volumes of sample, plasma, and 20 mM CaC12) and converted to units based on a calibration curve using purified and phospholipid reconstituted TF. The clotting activity of viable cells which stably expressed T F A~~~~~B~ was also compared with cells expressing wild-type TF to exclude artifacts introduced by the lysis method. The clotting times obtained in the one-stage clotting assay were converted into units of T F activity using a calibration curve based on serial dilutions of the cell line which expressed wildtype TF, essentially as previously described in detail (18).
Binding Analy~is--'*~I-VII binding to cell surface T F was analyzed on a cell monolayer, and Scatchard analysis was performed as previously described (23). Binding of VI1 to T F or T F A I~~A~~~ in the absence of phospholipid was determined using the previously described binding assay based on a noninhibitory anti-TF capture mAb (4). Briefly, "'I-VIIa and T F (2 nM) solubilized by 0.4 or 4 mM CHAPS as a detergent were assembled in the presence of 5 mM Ca2+ in a microtiter well which had been coated with the anti-TF mAb TF9-lOH10. After equilibrium, free and bound radioactivity was separated by rapid washes, and bound VIIa was determined from the captured radioactivity in the well, whereas the free ligand was determined from the radioactivity in the supernatant. Scatchard analysis was based on these experimental data. Dissociation of VIIa from T F and TF~165~166 was studied on cell surfaces, with variations from previously described procedures (24). VIIa (20 nM) was assembled with T F or T F A I~~A I~~ at 5 mM CaCl,, 0.5% bovine serum albumin for 2 h at 37 "C. Displacement of VIIa bound to T F was analyzed by two different procedures. First, displacement was initiated by dilution jump which included removal of the reaction mixture (400 pl) followed by addition of 4 ml of buffer containing 5 mM CaC12, 0.5% bovine serum albumin at 37 "C t o each well. Second, displacement was studied by addition of the inhibitory anti-TF mAb TF9-6B4 in a small volume (100-fold molar excess over IZ5I-VIIa in the reaction) in order to prevent the reassociation of VIIa after dissociation from TF. The residual 12511-VIIa bound after various times was determined based on control reactions which were incubated in parallel.
Substrate Hydrolysis-Chromogenic substrate (Spectrozyme FXa, American Diagnostica) hydrolysis was determined with VIIa (50 nM) bound to T F or TFA165A166 (1.5-25 nM) at 5 mM Ca" in a kinetic plate reader as described (14). TF or T F A~~~A~~~ were from delipidated cell extracts solubilized by 10 mM CHAPS. The detergent concentration in the assay was adjusted to 4 and 0.4 mM. Factor Xa generation by TF.VIIa was determined with T F expressed on viable cells in a coupled amidolytic assay, and data were fitted to the Michaelis-Menten equation as previously described (14,18). For fluid-phase analysis, TF and TF~165~166 from 10 mM CHAPS extractions of cell lysate were analyzed with various concentrations of VIIa for the activation of factor X (0.07-10 pM) at 0.4 and 4 mM CHAPS using an experimental procedure as for the analysis on viable cells. This included preincubation of the detergent-solubilized TF (4 nM) with VIIa for 10 min in the presence of CaZ+ followed by addition of factor X and removal of aliquots for determination of the rate of factor Xa formation. Similarly, competition experiments of TFA165A166 with wildtype TF were performed by preincubating the competitor (TFA166AIs) and T F with VIIa in the presence of Ca2+, followed by the addition of substrate and determination of initial rates of factor Xa formation.

Alanine Exchange for Lys-165 and Lys-166-Recombinant
site-directed mutants of TF in which the charged residues Lys-165 and Lys-166 were replaced by Ala were generated and expressed. Chinese hamster ovary cells (CHO-K1) were transiently transfected with plasmid DNA for the mutants T F A~~~A~~, T F A~~~, and T F A~~, .
The cells were harvested after 2 days, and cell lysate was analyzed immunologically for TF expression by ELISA and for function in a one-stage clotting assay. From these data, specific activities were calculated. Single alanine substitutions for Lys-165 resulted in a 70 f 27% ( n = 4) and for Lys-166 in a 88 f 4% (n = 3) loss of specific functional activity in comparison to wild-type TF. The specific activity of the double mutant T F A~~~A~~I was reduced by 98 f 2% (n = 4) indicating that the phenotype of this mutant reflects the additive effect of mutations in both Lys residues. This double mutant was further characterized after generation of a stable cell line which expressed this mutant.
Generation of a Stable cell Line Expressing T F A~~~A~~~ TF protein coding sequence in the vector CDM8 was mutated to replace both Lys-165 and Lys-166 by Ala. The mutated sequence was confirmed by DNA sequencing of the plasmid which was then used for transfection of Chinese hamster ovary cells. Stable clones were obtained by G418 selection for the cotransfected neomycin resistance plasmid pMAMneo. Based on flow cytometry, mutant TFA165A166 was expressed at 2-fold higher levels on the cell surface compared to the wildtype TF cell line used in this study (Fig. lA). This is consistent with antigen determination in cell lysate from the two lines which demonstrated 1.50 f 0.31 pmol/1o6 cells for TF~165~166 and 0.77 & 0.07 pmol/106 cells for TF (mean f standard deviation, n = 4). The viable cells expressing TF~165~166 demonstrated a 99 f 1% (n = 3) decrease of functional activity compared to the wild-type cell line when analyzed in a plasma clotting assay at concentrations from 3 X 1 0 ' to 1 X IO5 cells/ ml. The functional defect observed in the initial screening with cell lysate was therefore reproduced with T F~~6 5~~6 6 expressed on a viable cell suggesting that the reduced function is not due to inactivation of TFA165~166 during the detergent cell-lysis procedure.
No Evidence for Global Conformational Alterations of TFA165A166-The epitopes for three non-overlapping mAbs (21) on the TF extracellular domain were equivalently expressed on both TF and TF~166~166, as demonstrated by flow cytometry (Fig. lA). The TF9-6B4 epitope was mapped to residues 40-83, and the TF8-5G9 epitope is in the carboxyl-terminal aspect of TF (residues 106-219). The latter epitope is discontinuous and conformation-dependent (21). Similarly, TF9-9C3 could not be assigned to a linear epitope suggesting a discontinuous epitope for this mAb. Reactivity with these mAbs therefore indicates a proper overall conformation of the A suspension of cells expressing T F or TFA16sA166 was incubated without (no) and with the cross-linking reagent BS:' or DTSSP. Cells (2 X lo5) were lysed in reducing (+) or non-reducing (-) sample buffer before gel electrophoresis. After transfer, T F was detected by a polyclonal antibody which was visualized by a secondary, alkaline phosphatase conjugated antibody followed by a color reaction.
T F~~6 5~~6 6 extracellular domain. Based on electrophoretic mobility, T F A~~~A~~ appeared to be glycosylated to a similar extent as the wild-type protein (Fig. 1B) suggesting normal cellular processing as an indication of proper global folding of the molecule. In addition, the high levels of cell surface expression of TFA165A166 indicate lack of intracellular degradation due to a folding defect of T F A~~~A~~~ and thus support unaltered global fold (25,26).
Dimerization of TF and TF~165~166-We used chemical cross-linking to analyze whether TFA165A166 is expressed in a dimeric organization on the cell surface as described for the wild-type molecule (19). TFA165A1m cross-linked with the reducible reagent DTSSP demonstrated dimer formation to an extend similar to wild-type TF, whereas the noncross-linked samples lacked a band migrating at a position corresponding to a dimer of TF (Fig. 1B). We also used the non-reducible cross-linker BS3 to demonstrate that the dimerization observed in the presence of DTSSP is not due to disulfide bond formation via the Cys residues in the cytoplasmic domain of T F after exposure of the cells to the cross-linking reagent. When samples cross-linked with BS3 were reduced to disrupt cystines, the analysis by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Western blotting demonstrated a band corresponding to the dimer of T F (Fig. 1B). This crosslinking analysis thus demonstrates that pairs of TF, as well as TFA165A166, are in sufficient proximity on viable cell surfaces, consistent with the idea of a dimeric cell surface expression. However, the previously demonstrated cross-linking of wildtype TF suggested a much higher percentage (up to 50%) of dimer formation (19) than our analysis. Variations in the expression levels and differences in the cell lines used for the expression of the proteins may account for the quantitative differences. The cross-linking of TFA165A166 to a similar extent as TF suggests that lysine residues 165 and 166 are not required for association with adjacent T F molecules and in-termolecular cross-linking with the hornobifunctional, amino group-reactive reagents.
VII Binding to Cell Surface TF-Binding of '2sI-VII to TF and TFA165A166 was similar (Fig. 2) and saturable with twice the number of binding sites on cells expressing T F A~~~A compared to cells expressing TF, consistent with the higher surface expression on the TFA165A166 cell line demonstrated by flow cytometry. The dissociation constants ( K d ) were comparable with 13.8 nM for T F and 14.3 nM for TFA165A166 (Fig.  2). The number of sites was 0.81 pmol/106 cells for TF and 1.62 pmol/1o6 cells for T F A~~~A~~ consistent with the antigen determinations in cell lysates by ELISA (see above). This is consistent with a 1:l stoichiometry of VI1 binding to TF and TF~16~~166, as previously demonstrated for wild-type TF (23,27).
VZZ Binding in the Absence of Phospholipid-Similar affinity of VIIa binding to TF and TF~165~166 was further demonstrated by binding analysis in the absence of phospholipid. Crude TF and T F A~~~A~~~ were obtained from the stable cell lines by organic extractions to remove lipids followed by solubilization of T F with 10 mM CHAPS as a detergent. Binding of VIIa to 2 nM detergent-solubilized TF or T F A~~~A was analyzed in a mAb-based binding assay at high (4 mM CHAPS) and low (0.4 mM CHAPS) detergent concentrations. The affinities of VIIa binding to T F and TFA165A166 were similar and were not influenced by the detergent concentrations (Table I). The binding constants for binding to cell surface TF and detergent-solubilized TF were similar, consistent with our previous analysis (4). This analysis demonstrates that CHAPS at the indicated concentrations did not demonstrably influence the TF-VI1 interaction.   Dissociation of VIIa from Cell Surface TF-To further substantiate the unaltered binding of VIIa to T F A~~~A~~, the dissociation of the TF VIIa and TF~165~166 .VIIa complex was studied. First, dissociation was induced by dilution jump after VIIa was assembled with TF or T F A~~~~~~. The rate of dissociation from T F A~~~A~~ was similar to the reaction with the wild-type TF (Fig. 3). A 100-fold molar excess of an inhibitory mAb was used in a second experimental setup to prevent the reassociation of VIIa which had dissociated from the complex. The mAb epitope was similarly expressed on both TF and TF~~65~166 based on flow cytometry (Fig. u). The rate of dissociation of the TF'VIIa complex in the presence of a competitive inhibitor was consistent with results from experiments which used the TF+ human bladder carcinoma cell line 582 and demonstrated a 30-min dissociation half-time for the TF .VIIa complex (24). The rate of dissociation with inhibitor present was substantially faster compared to the dilution experiments (Fig. 3). This difference in dissociation rates is typical for ligand-receptor interactions which are characterized by positive cooperativity (28), and VIIa has been shown to bind to cell surface TF (29) as well as to phospholipid reconstituted TF (27) in a positive cooperative manner. The dissociation of VIIa from T F A~~~A~~ occurred with a slightly slower rate compared to TF, thus demonstrating no evidence for a lower affinity of the V I I B -T F A~~~A~~~ interaction. The analysis of displacement therefore provides additional support for high affinity VIIa binding by TFA165~166.
Substrate Cleavage by the TFA165A166. VIIa Complex-We analyzed the hydrolysis of a small peaidyl chromogenic substrate (Spectrozyme FXa) as well as the proteolytic activation of the protein substrate factor X by complexes of VIIa formed with either TF or T F A~~~A~~~. Peptide hydrolysis by VIIa (50 nM) was analyzed in the presence of various concentrations of T F or T F A~~~A~~~ solubilized with detergent. High concentrations (4 mM) of CHAPS did not influence the TF-or TF~165~166-enhanCed catalysis of chromogenic peptidyl substrates compared to 0.4 mM CHAPS (Fig. 4A). There was no difference in the rate of peptide hydrolysis between the TF.
VIIa and the TF~165~166.VIIa complex over the range of TF concentrations tested. This demonstrates that both cofactor proteins support the catalytic enhancement of the serine protease domain of VIIa to a similar extent. However, the analysis of factor X cleavage by cell surface complexes of T F plex had a considerable slower rate of factor X activation compared to TF.VIIa (Fig. 4B). Kinetic  of VIIa which were required to achieve half-maximal rates were similar for TF and TFA1aA1m at 4 and 0.4 mM CHAPS (Fig. 5). However, the maximal rate of factor X activation in the presence of T F A~~~~~~~ was reduced 3-fold compared to the reaction with an identical concentration of wild-type TF, when analyzed at 0.4 mM CHAPS. This is consistent with the difference in reaction rates observed on cell surfaces and further supports the proposed selective reduction in protein substrate cleavage by TFA165~166.VIIa. In addition, the maximal rates were influenced by the detergent concentration. The rate of factor X activation was slightly reduced by 30% when wild-type TF at 4 mM CHAPS was compared with reactions at 0.4 mM detergent. This indicates that detergent interferes with the transient protein substrate assembly and cleavage by TF, VIIa, and the reduction appears to be selective, since cleavage of chromogenic peptidyl substrates was not altered at high detergent concentrations (Fig. 4A). The reaction rate for factor X cleavage by TFA165A166. VIIa decreased even more, and a 70% reduction was observed when reactions at 4 mM CHAPS were compared to the low detergent conditions at saturating concentrations of VIIa (Fig. 5 B ) . The slight alteration of the cleavage of protein substrate by TF . VIIa due to detergent thus appeared to be exaggerated with the mutant T F consistent with the idea that protein substrate recognition was further diminished in the presence of detergent. The differences between TF and T F A~~~A~~~ were also apparent when the concentration of factor X was varied at a fixed concentration of VIIa (Fig. 6 A ) . Consistently, high concentrations of detergent inhibited TF.VIIa complex to a lesser extent than the TFA165A166.VIIa complex (Fig. 6B), and increasing concentrations of factor X did not compensate for the reduced rate of protein substrate hydrolysis.
Competition of TFA165A166 with TF-The hypothesis of selective reduction in protein substrate cleavage without loss of VIIa binding function by T F A~G~A~G~ was further tested by competition analysis. At both, 4 and 0.4 mM CHAPS, TFA165A16G effectively competed with TF for a limited amount of VIIa (2 nM) in the assay (Fig. 7). A 10-fold excess of the mutant over wild-type TF reduced the rate of factor X activation to levels which were observed at 2 nM TFA166A166, and a 50% reduction of the rate of factor X activation by TF. VIIa was observed at approximately equal concentrations of TF (2 nM) and the mutant (Fig. 7). These data are consistent with unaltered VIIa binding by T F A~~~~~~~ even in the presence of substrate and at different detergent concentrations and provide further support that the T F A~~~A~~~. VIIa complex exhibits a selective defect in protein substrate recognition.

DISCUSSION
Assembly of a protease with a cofactor receptor and subsequent oriented association of substrate is an efficient mechanism to generate proteolytic activity on cell surfaces. Forma-tion of the TF . VIIa complex by high affinity binding of VIIa to TF is a requisite for full catalytic and proteolytic activity of VIIa. A functional Gla domain in VIIa appears to be required for high affinity binding of VIIa to TF and for the resulting efficient recognition of protein substrates (4). We here characterize a TF mutant which demonstrates normal VIIa binding, but a selective defect in the support of macromolecular substrate recognition in the TFA165A166. VIIa complex suggesting a contribution of the cofactor to substrate recognition which is independent of the high affinity binding of VIIa.
TFA~~F,AS~ appeared to adopt a proper global structure based on the lack of defects in the expression of three non-overlapping mAb epitopes, in apparent glycosylation and in proper cellular processing including high levels of cell surface expression. Proper global folding of TFA1eA1m is also suggested by unaltered high affinity binding of VIIa and by efficient chromogenic peptidyl substrate cleavage by the TF~165~166 + VIXa complex. Normal VI1 binding function of T F A~~~~~~~ was substantiated by radioligand binding analysis using TF expressed on viable cells as well as TF extracted from the cells and solubilized with detergent. Dissociation of VIIa bound to tively competed with TF in functional assays. The discrete functional defect resulting from the deletion of the side chains of Lys-165 and Lys-166 was a diminished rate of proteolytic activation of the protein substrate factor X by the TF~165~166.
VIIa complex. This reduced activation of factor X was observed both in the presence and in the absence of a phospholipid surface. High concentrations of detergent interfered with factor X activation by TFA165A1~ VIIa to a larger extent than with the activation by TF . VIIa. These observations converge upon the conclusion that macromolecular recognition of factor X by the binary complex is selectively altered by the mutations in Lys-165 and Lys-166 consistent with the possibility of a direct contribution of cofactor residues to interactions with substrate.
There appears to be a quantitative difference when the reduction in coagulation of plasma is compared with analysis in the purified protein system. Clotting activity of T F A~~~A~~ on viable cells was reduced almost lOO-fold, whereas the catalytic efficiency of TFA165A166 on viable cells was reduced only %fold compared to wild-type TF. This difference arises in part from the much lower factor X concentration (-50 nM) in the clotting assay compared to the high concentrations (up to 5 p~) in the purified system. Since there appears to be a slight increase in the KMapp for factor X activation by TFA16SA166. VIIa in comparison to TF. VIIa, the loss of function would be more pronounced at lower substrate concentrations. In addition, the activation of factor IX by TF .VIIa is enhanced in the presence of factor X suggesting interactions of the two substrates during the initiation of the extrinsic pathway of coagulation (30). The cooperative activation of factor IX may contribute to the overall procoagulant activity determined in a clotting assay, and the demonstrated defect in factor X recognition may lead to a greater functional loss due to an indirect effect on the factor IX activation. However, direct loss of factor IX substrate recognition cannot be excluded based on our analysis.
The present study demonstrates critical contributions of 2 lysine side chains in the carboxyl half of the extracellular domain to extended recognition of factor X by the TF.VIIa complex. The epitope mapping of an inhibitory anti-TF mAb also implicated structures in exon 4 and 5 in binding of factor VI1 (21), and residues 129-169 have been shown to be in proximity to VIIa by photoaffinity cross-linking (31). Lysine T F A~~I~A I~~ Was Similar to wild-type TF, and TF~165~1g effec-165 and 166 may be in some spatial proximity to VI1 interactive sites on TF. However, no alterations in the VI1 binding to T F A~~~A~~~ were observed suggesting that the side chains of lysine residues 165 and 166 do not significantly contribute to the proposed VI1 interactive site in the carboxyl half of the TF extracellular domain (18). Kinetic analysis has previously provided evidence that association of substrate in the ternary complex diminishes dissociation of the TF. VIIa complex (32) suggesting that substrate interaction with the binary complex either tightens the binding of VIIa to TF by allosteric alterations of VIIa or prevents dissociation of VIIa by binding to a site on TF which is spatially close to VIIa interactive sites and thereby sterically hinders VIIa mobility and dissociation of the complex. Lys-165 and Lys-166 could form specific contacts with VIIa which are critical for extended recognition of substrates and which do not significantly contribute to the binding energy in the absence of substrate. In the presence of substrate, however, these contacts may prevent dissociation of VIIa from the TF. VIIa complex. Alternatively, these 2 lysine residues in TF could provide critical contacts in an interactive site for substrate on TF. Macromolecular assembly of substrate with the TF. VIIa complex may involve residues in VIIa as well as in TF which, once associated, provide a contiguous region for extended recognition of substrate. This analysis provides evidence for multiple alignments during the assembly of substrate with an enzyme-cofactor complex. A model for efficient substrate assembly is suggested which involves both enhancement of function of the catalytic site and association of specific cofactor and enzyme residues in a site for extended recognition of substrate.