Role of Regulatory Exosite I in Binding of Thrombin to Human Factor V, Factor Va, Factor Va Subunits, and Activation Fragments*

The blood coagulation proteinase, thrombin, converts factor V into factor Va through a multistep activation pathway that is regulated by interactions with thrombin exosites. Thrombin exosite interactions with human factor V and its activation products were quantitatively characterized in equilibrium binding studies based on fluorescence changes of thrombin covalently labeled with 2-anilinonaphthalene-6-sulfonic acid (ANS) linked to the catalytic site histidine residue byN α-[(acetylthio)acetyl]-D-Phe-Pro-Arg-CH2Cl ([ANS]FPR-thrombin). Exosite I was shown to play a predominant role in the binding of factor V and factor Va from the effect of the exosite I-specific ligand, hirudin54–65, on the interactions. Factor V and factor Va bound to exosite I of [ANS]FPR-thrombin with similar dissociation constants of 3.4 ± 1.3 and 1.1 ± 0.4 μm and fluorescence enhancements of 182 ± 41 and 127 ± 17%, respectively. Native thrombin and labeled thrombin bound with similar affinity to factor Va. Among factor V activation products, the factor Va heavy chain was shown to contain the site of exosite I binding, whereas exosite I-independent, lower affinity interactions were observed for activation fragments E and C1, and no detectable binding was observed for the factor Va light chain. The results support the conclusion that the factor V activation pathway is initiated by exosite I-mediated binding of thrombin to a site in the heavy chain region of factor V that facilitates the initial cleavage at Arg709 to generate the heavy chain of factor Va. The results further suggest that binding of thrombin through exosite I to factor V activation intermediates may regulate their conversion to factor Va and that similar binding of thrombin to the factor Va produced may reflect a mode of interaction involved in the regulation of prothrombin activation.

Blood coagulation factor V is proteolytically processed by thrombin into factor Va through a multistep pathway that is essential for accelerating blood coagulation to the rate required for normal hemostasis (1)(2)(3)(4)(5). The single-chain, 330,000 molecular weight human factor V molecule has a homology domain structure of A1-A2-B-A3-C1-C2 and is activated by thrombin cleavage at Arg residues 709, 1018, and 1545 within the B domain (4, 6 -9). This generates the factor Va heavy chain (A1-A2) and light chain (A3-C1-C2) subunits and releases the B domain as two activation fragments: fragment E (residues 710 -1018) and fragment C1 (residues 1019 -1545) (4, 6 -9). Factor Va is a dimer of the heavy and light chain subunits, which are associated noncovalently in a calcium-dependent interaction that is required for activity (4,10). Activation of factor V results in an increase in its affinity for factor Xa and prothrombin necessary for efficient assembly on phospholipid membranes of a factor Xa-factor Va-prothrombin catalytic complex that generates thrombin at a ϳ300,000-fold faster rate than factor Xa alone (4,(11)(12)(13). Both associated subunits of factor Va are required for factor Xa binding, whereas prothrombin binds to the factor Va dimer and to the isolated heavy chain subunit (14 -17).
The pathway of factor V activation by thrombin follows a kinetically preferred order of bond cleavage, in which products of cleavage at Arg 709 appear first, followed closely by cleavage at Arg 1018 , and generation of the factor Va light chain by cleavage at Arg 1545 as the slowest reaction (4,8,9,18,19). The interactions responsible for specific binding of factor V to thrombin as a substrate and the mechanisms that regulate the specificity of the reactions on the activation pathway are not completely understood. Recent studies indicate that exosites I and II on thrombin participate in the mechanism of factor V activation (20 -22). Exosites I and II are distinct electropositive sites on thrombin that contribute to the specificity of the enzyme by mediating the binding of certain protein substrates and inhibitors, as well as the binding of macromolecular effectors (23,24). Exosite I has been implicated in factor V activation from the inhibitory effect of the specific peptide ligands, hirudin 54 -65 (Hir 54 -65 ) 1 (Gly-Asp-Phe-Glu-Glu-Ile-Pro-Glu-Glu-Tyr-Leu-Gln) or N-acetyl-hirudin 53-64 (hirugen) (N-acetyl-Asn-Gly-Asp-Phe-Glu-Glu-Ile-Pro-Glu-Glu-Tyr(SO 3 )-Leu) (20 -22). Proteolytic derivatives of thrombin that have lost exosite I function, ␤and ␥-thrombin, also show decreased activity in factor V activation (22). The role of exosite II in factor V activation is less clear. Although a decreased rate of bovine factor V activation was observed with an exosite II-disabled thrombin mutant (20), the expected inhibitory effects of the exosite II ligands, heparin and prothrombin fragment 2, have not been observed (20,22).
Affinity chromatography of factor V activation products on a matrix of thrombin immobilized through its active site was used in previous studies to demonstrate that thrombin binds to factor Va in a previously undescribed, exosite I-dependent interaction with the heavy chain subunit (21). This observation may have significance with respect to the function of factor Va in prothrombin activation as well as the mechanism of factor V activation, as suggested by kinetic studies of prethrombin 2 as a substrate analog of prothrombin (25,26). Thrombin acts as a competitive inhibitor of bovine prethrombin 2 activation through exosite interactions with the factor Xa-factor Va-membrane complex (25,26), suggesting that thrombin binding to factor Va could be involved.
The roles of thrombin exosite interactions in factor V activation and factor Va activity have been investigated further in the present studies, where the interactions with human factor V, factor Va, the factor Va subunits, and the activation fragments were characterized in quantitative equilibrium binding studies for the first time. The unique properties of a fluorescent thrombin derivative labeled at the catalytic site with 2-anilinonaphthalene-6-sulfonic acid as a reporter of exosite I interactions have been used to resolve the role of this regulatory site in binding of factor V/Va species. Exosite I is shown to be principally responsible for binding of factor V to active sitelabeled thrombin, with a small contribution from exosite I-independent, lower affinity interactions. Examination of thrombin binding to each of the purified factor V activation products identified only one exosite I-mediated interaction of thrombin with a site on the heavy chain subunit of factor Va as the predominant mode of binding, again with small contributions from lower affinity interactions. These studies support a mechanism of factor V activation in which factor V is recognized as a specific thrombin substrate by exosite I-mediated binding at a site on the heavy chain region of factor V, which enhances specific cleavage at Arg 709 to initiate the activation pathway. The observation that factor Va retains the affinity for the exosite I interaction indicates that similar thrombin binding to factor V activation intermediates may regulate other steps in the activation pathway. The finding that thrombin binding to factor Va has similar characteristics to those described previously for prothrombin binding to bovine factor Va (15,16,27) suggests that these interactions may be related, reflecting a common mode of binding of the substrate and product of prothrombin activation.
Factor V was purified from normal human plasma and activated by incubation at 2-6 mg/ml with 50 nM thrombin in 50 mM Hepes, 0.11 M NaCl, 5 mM CaCl 2 , 1 mg/ml polyethylene glycol 8000, pH 7.4, for 40 min at 37°C as described previously (21). The thrombin was inactivated by incubation for 15 min at room temperature with 1 M FPR-CH 2 Cl, and the factor Va was stored at Ϫ70°C. Factor V activation products were isolated from thrombin-activated factor V by chromatography on (2-mercaptoethanol)-(3,3Ј-iminobispropylamine)-agarose and on thrombin immobilized onto iodoacetyl-(3,3Ј-iminobispropylamine)-agarose through its active site with N ␣ -[(acetylthio)acetyl]-FPR-CH 2 Cl, following the methods developed previously (21). Briefly, fragment C1 was separated from factor Va and fragment E by chromatography of thrombin-activated factor V on (2-mercaptoethanol)-(3,3Ј-iminobispropylamine)-agarose and further purified by passage through a 5-ml column of S-Sepharose in 25 mM Hepes, 50 mM NH 4 Cl, 5 mM CaCl 2 , pH 6.5, at 4°C. Following removal of fragment C1, fragment E and the two-subunit form of factor Va (Va-hl) were separated on the immobilized thrombin matrix in buffer containing 5 mM CaCl 2 . The factor Va heavy chain (Va-h) and light chain (Va-l) subunits were isolated by chromatography of EDTA-dissociated Va-hl on the thrombin matrix in buffer containing 1 mM EDTA. Proteins were concentrated by ultrafiltration with YM 30 membranes and dialyzed against the buffer used for the experiments. Polyethylene glycol 8000 was omitted from the buffers used for elution of the factor Va species to prevent its co-concentration during ultrafiltration. Proteins were characterized by SDS-gel electrophoresis on 4-15% gradient gels stained with Coomassie Blue or periodic acid-Schiff reagent. Concentrations of factor V/Va species were determined from the 280 nm absorbance using the following absorption coefficients ((mg/ml) Ϫ1 cm Ϫ1 ) and molecular weights: factor V and thrombin activated-factor V, 0.89 and 330,000 (2); Va-hl, 1.50 and 179,000; Va-h, 1.03 and 105,000; Va-l, 2.17 and 74,000; fragment C1, 0.22 and 150,000; and fragment E, 0.52 and 71,000. With the exception of the value for factor V, the absorption coefficients were calculated from the amino acid compositions (7,32).
Nonsulfated hirudin 54 -65 (Sigma or Bachem) was dissolved in water, and its concentration was determined from the tyrosine absorbance at 293 nm in 0.1 M NaOH with an absorption coefficient of 2381 M Ϫ1 cm Ϫ1 (33).
Equilibrium Binding Studies-All experiments were performed in 50 mM Hepes, 0.11 M NaCl, 5 mM CaCl 2 , 1 mg/ml polyethylene glycol 8000, pH 7.4, and at 25°C. FPR-CH 2 Cl (1 M) was added to the buffer for all titrations except those containing native thrombin. Fluorescence measurements were made with an SLM 8100 fluorometer in the ratio mode. Acrylic cuvettes were coated with polyethylene glycol 20,000 to minimize protein adsorption (34) for measurements of ANS fluorescence, and uncoated cuvettes were used for tryptophan fluorescence. Binding of Hir 54 -65 to thrombin was measured from the changes in tryptophan fluorescence with 295 nm excitation (4 or 8 nm band pass) and 360 nm emission (8 nm band pass). Titrations were done by measuring the fluorescence after successive additions of small volumes of titrant. The observed fluorescence was corrected for dilution (Յ13%) and for background by subtraction of measurements on solutions lacking the fluorescent species. Fluorescence data were expressed as the fractional change in the initial fluorescence (⌬F/F o ϭ(F obs Ϫ F o )/F o ). Titrations of thrombin with Hir 54 -65 were fit by the quadratic equilibrium binding equation to obtain the maximum fluorescence change (⌬F max /F o ) and dissociation constant (K Hir ), assuming one binding site for Hir 54 -65 on thrombin.
Binding of factor V/Va species to [ANS]FPR-thrombin was measured from changes in ANS fluorescence emission at 450 nm with excitation at 325 nm and using 8 nm band passes. Fixed concentrations of [ANS] FPR-thrombin were titrated with factor V/Va species in the absence of Hir 54 -65 and in the presence of a fixed, near-saturating Hir 54 -65 concentration. Mixtures of [ANS]FPR-thrombin and a fixed level of factor V/Va species were also titrated with Hir 54 -65 under the same conditions. The relatively high protein concentrations required for the titrations necessitated corrections for dilution of Յ 35%. Dilution of [ANS] FPR-thrombin was included in analysis of the results. However, such corrections had little effect because the concentration of labeled thrombin was below the dissociation constants for the interactions. Corrections of the ANS fluorescence data for background were typically Յ13%, with the exception of the lowest fluorescence enhancement in the fragment C1 titrations (Յ20%) and in the native thrombin competition experiments at the highest protein concentrations (Յ27%).
Binding of factor V/Va species to [ANS]FPR-thrombin and the effect of Hir 54 -65 were analyzed as a special case of the general situation in which two ligands bind competitively to a fluorescent protein probe and the interactions are accompanied by unequal fluorescence changes. In this model, ligands L and C bind mutually exclusively to n equivalent and independent sites on the probe (P) with dissociation constants K L and K C and maximum fluorescence changes relative to the fluorescence of the probe of ⌬F max L /F o and ⌬F max C /F o (Scheme I).
The observed fluorescence change is given by the sum of the contributions from the two probe complexes, weighted by the maximum fluorescence changes associated with their formation (35), The fraction of sites on P occupied by L or C can be expressed in terms of the free ligand concentrations as follows.
Equations 1-3 would be adequate for analysis of the fluorescence changes as a function of [L] and [C] if the approximation could be made that the free concentrations of the ligands were equivalent to their total concentrations. However, this assumption is not always justified, and typically only the total concentrations are known in spectroscopic studies. The free concentrations of L and C can be expressed in terms of the total concentrations and the concentration of PL as shown in Equations 4 and 5.
Substitution of these expressions into Equation 2 and rearranging gives the cubic Equations 6 -10 for [PL]/n[P] o in terms of the total ligand and probe concentrations, the dissociation constants, and the stoichiometric factor. , which is obtained from the mass conservation equation for P and the definition of the equilibrium constant.
For the present case, P was [ANS]FPR-thrombin, L was factor V, Va or Va subunit, and C was Hir 54 -65 . The fluorescence change accompanying Hir 54 -65 binding to [ANS]FPR-thrombin did not contribute directly to the observed fluorescence change because the maximum fluorescence increase was Յ3% of the change accompanying factor V/Va binding. This allowed the second term of Equation 1 to be set to zero. In the factor V/Va experiments, there was also an exosite I-independent increase in fluorescence that was a linear function of factor V/Va concentration. This was included in the binding model as a term linear in factor V/Va concentration, representing weak interactions of factor V/Va species with labeled thrombin. Simplification of Equation 1 for the present studies and incorporating the exosite I-independent fluorescence change gave Equation 12, where ANS-T represents [ANS]FPRthrombin and V represents the factor V/Va species.
Fluorescence changes were measured in titrations of [ANS]FPR-throm-bin as a function of the total concentration of factor V/Va species in the absence of Hir 54 -65 and at near-saturating concentrations of Hir 54 -65 , and as a function of Hir 54 -65 concentration at fixed concentrations of factor V/Va species. The results were fit simultaneously by Equation 12 with [ANS-T⅐V]/n[ANS-T] o calculated by solution of the cubic Equations 6 -10 using the Newton-Raphson algorithm incorporated into the BA-SIC programs Dnrp53 or Dnrpeasy (36,37). The fitted parameters were the dissociation constant for factor V/Va binding to [ANS]FPR-thrombin (K V/Va ), the maximum fluorescence change for this interaction, the dissociation constant for Hir 54 -65 binding, and the slope of the exosite I-independent fluorescence increase (⌬F exo-ind /F o (where "exo-ind" indicates "exosite I-independent")), with a value of 1 assumed for the stoichiometric factor. The effect of native thrombin on binding of factor Va to [ANS]FPRthrombin was measured in titrations of [ANS]FPR-thrombin with factor Va in the absence and presence of fixed concentrations of native thrombin and in the absence and presence of near-saturating concentrations of Hir 54 -65 . The results were analyzed by nonlinear least squares fitting of all of the titration data with Equation 12. For this case, which differs from the situation described above, the cubic equation described previously (38,39) for competitive binding of a ligand (factor Va) to a labeled protein probe ([ANS]FPR-thrombin) and an alternate unlabeled acceptor (native thrombin) was used to calculate Equation 12. The dissociation constants for factor Va binding to [ANS]FPR-thrombin and to native thrombin, the maximum fluorescence change, and the slope of the exosite I-independent fluorescence change were the fitted parameters.

Characterization of 2-Anilinonaphthalene-6-sulfonic Acid-labeled Thrombin as a Probe of Exosite I-dependent
Interactions with Factor V/Va Species-Binding of thrombin to factor V and its activation products was characterized quantitatively from fluorescence changes of thrombin that was specifically labeled with ANS covalently linked to the catalytic site histidine residue through N ␣ -[(acetylthio)acetyl]-D-Phe-Pro-Arg-CH 2 Cl (28,30,31). To determine the role of exosite I in the factor V/Va interactions, specific binding of the nonsulfated hirudin peptide, Hir 54 -65 , to exosite I of [ANS]FPR-thrombin was first characterized. Analysis of intrinsic tryptophan fluorescence changes accompanying Hir 54 -65 binding to [ANS]FPR-thrombin and native thrombin gave indistinguishable dissociation constants of 0.94 Ϯ 0.14 and 1.1 Ϯ 0.2 M and maximum fluorescence enhancements of 12.7 Ϯ 0.4 and 13.4 Ϯ 0.6%, respectively ( Fig. 1). Under the same conditions, binding of the peptide to [ANS]FPR-thrombin increased the ANS fluorescence by only 2.5 Ϯ 0.2% (Fig. 1). These results indicated that Hir 54 -65 bound to exosite I of ANS-labeled thrombin and native thrombin with the same affinity but that the ANS probe did not report this interaction with a significant change in fluorescence. The large differential response of [ANS]FPR-thrombin to binding of factor V and its activation products compared with binding of Hir 54 -65 was used to establish the role of exosite I in the factor V/Va interactions.
Binding of Factor V to [ANS]FPR-thrombin-Addition of single-chain factor V to [ANS]FPR-thrombin resulted in a large, saturable enhancement in the probe fluorescence, signaling factor V binding (Fig. 2). Titration of a mixture of labeled thrombin and factor V with Hir 54 -65 resulted in a partial return of the enhanced fluorescence toward the initial value, consistent with displacement of thrombin from factor V by the peptide. Titration of [ANS]FPR-thrombin with factor V in the presence of a concentration of Hir 54 -65 sufficient to essentially saturate exosite I (ϳ100 ϫ K Hir ) produced a linear increase in ANS fluorescence, which represented 21% of the enhancement seen with factor V alone at the highest concentration measured (Fig. 2). These results indicated that the fluorescence enhancement associated with factor V binding was due primarily to an exosite I-dependent interaction, with a smaller contribution from a lower affinity, exosite I-independent process. The possibility that the exosite I-independent process was due to free fluorescence probe in the labeled thrombin preparations was ruled out by the observations that (a) free probe was not detected in the dialysis buffer from [ANS]FPR-thrombin, and (b) more exhaustive gel filtration or dialysis in the preparation of [ANS]FPR-thrombin did not affect the fluorescence change. The effect of unrelated proteins on the fluorescence of [ANS] FPR-thrombin was examined to evaluate the possibility that the linear fluorescence change was due to nonspecific interactions of factor V with the labeled protein. Ovalbumin and bovine serum albumin also increased the fluorescence of [ANS] FPR-thrombin linearly up to 30 -50 M, with slopes of 0.5 and 0.2% M Ϫ1 , respectively (results not shown). These results suggested that nonspecific interactions could be responsible, although the magnitude of the fluorescence change of 7.1% M Ϫ1 seen with factor V was significantly larger. Results of further studies with factor Va described below showed that native thrombin had no detectable effect on the exosite I-independent fluorescence change, indicating that it was due either to nonspecific interactions only exhibited by the probe-labeled protein or to an interaction of factor V/Va at a second site on thrombin with an affinity too low to be determined.
On the basis of these observations, the binding data were analyzed by computer fitting of a model in which factor V and Hir 54 -65 bound competitively to [ANS]FPR-thrombin, with a large fluorescence change associated with factor V binding and no change for Hir 54 -65 binding. The model also incorporated weak binding of factor V to thrombin in exosite I-independent interactions characterized by a linear increase in fluorescence. An equation was derived for analysis of the general situation in which two ligands bind competitively to a probe with unequal fluorescence changes accompanying the interactions (see under "Experimental Procedures"). This allowed all of the fluorescence titration results obtained as a function of the total concentrations of both ligands to be fit by nonlinear least squares analysis without the need for the assumption of equivalence of the free and total concentrations. As shown in Fig. 2 (Table I). The good agreement between the dissociation constant for Hir 54 -65 binding to [ANS]FPR-thrombin obtained from its displacement of factor V and the value of 0.94 Ϯ 0.14 M measured directly for binding of the peptide to exosite I supported the validity of the binding model.
In this analysis and that described below for other factor V/Va species, the exosite I-independent interactions were included in the model as a linear binding process. As expected, a more detailed model in which these additional interactions were included as individual factor V/Va binding steps did not fit the data with unique parameters for these interactions because of the linearity of the fluorescence increase and the large number of unknowns. However, if equivalent affinity and fluorescence enhancements were assumed for the exosite I-independent interactions, acceptable fits of the results could be obtained when the exosite I-independent interactions had an affinity Ͼ5-fold lower than exosite I binding, as judged by the agreement between the fitted values of K Hir and the upper error limit of the experimentally determined dissociation constant. These findings confirmed that analysis of the data with more complex models was not justified and supported further the validity of representation of the exosite I-independent interactions in the model as a single, low affinity process.

Binding of Factor Va to [ANS]FPR-thrombin-Titration of [ANS]
FPR-thrombin with thrombin-activated factor V, consisting of a mixture of the two-subunit form of factor Va (Va-hl), fragments E and C1, also produced a fluorescence enhancement (Fig. 3). The enhancement was partially reversed by Hir 54 -65 , and a residual exosite I-independent fluorescence increase was again observed in the presence of a near-saturat-  Table I. Titrations were performed and analyzed as described under "Experimental Procedures." ing level of the peptide (Fig. 3). Analysis of these results gave a dissociation constant of 1.1 Ϯ 0.4 M for factor Va binding to [ANS]FPR-thrombin, representing a 3-fold higher affinity compared with factor V and a 1.4-fold smaller fluorescence enhancement of 127 Ϯ 17% (Table I). The adequacy of the binding model was again supported by the agreement between the dissociation constant obtained for Hir 54 -65 and that determined directly in the absence of factor Va ( Table I).
Binding of the Isolated Two-subunit Form of Factor Va to [ANS]FPR-thrombin-To investigate further the exosite I-dependent interactions with factor V activation products, factor Va-hl, activation fragments E and C1, and the factor Va subunits were isolated by affinity chromatography. Preferential binding of fragment C1 to (2-mercaptoethanol)-blocked iodoacetyl-(3,3Ј-iminobispropylamine)-agarose and binding of factor Va and the factor Va heavy chain to thrombin specifically immobilized onto this matrix through its active site (21) enabled each of the proteins to be obtained in highly purified form, as was shown by SDS gel electrophoresis (Fig. 4). Titration of [ANS]FPR-thrombin with purified factor Va-hl demon-strated an exosite I-dependent interaction similar to that seen with unfractionated factor Va, with a dissociation constant of 0.6 Ϯ 0.2 M and maximum fluorescence enhancement of 113 Ϯ 15% (Fig. 5). These parameters were indistinguishable from those obtained for unfractionated factor Va (Table I), indicating that the exosite I-dependent interaction was accounted for by binding of the factor Va heterodimer to [ANS]FPR-thrombin.

Interactions of the Isolated Factor Va Subunits and Activation Fragments with [ANS]FPR-thrombin-Comparison of titrations of [ANS]
FPR-thrombin with the separated factor Va subunits showed a large fluorescence enhancement for the heavy chain and no significant change (Յ3% up to 4 M) for the light chain (Fig. 6). Fragments E and C1 produced small increases in fluorescence that were linear with concentration, reaching levels of 12% (fragment E) and 30% (fragment C1) at 4 M (Fig. 6). Addition of 34 -50 M Hir 54 -65 to mixtures of [ANS]FPR-thrombin and Va-l, fragment E, or fragment C1 decreased the fluorescence by Յ6%, indicating that these fluorescence changes were exosite I-independent (Fig. 6). Extension of the titrations with fragment C1 to higher concentrations showed some evidence of curvature, suggestive of a weak interaction with a ϳ35 M dissociation constant (results not shown). Analysis of the binding of factor Va-h to [ANS]FPRthrombin as a function of heavy chain and Hir 54 -65 concentration (Fig. 7) Table I. Titrations were performed and analyzed as described under "Experimental Procedures. "   FIG. 4. Preparations of factor V and factor V activation products. SDS gels of reduced samples of preparations of the proteins used in these studies are shown: factor V (ϳ7 g), factor Va (ϳ9 g), factor Va-hl (ϳ4 g), Va-h (ϳ11 g), Va-l (ϳ7 g), fragment E (ϳ14 g), and fragment C1 (ϳ40 g). The migration positions of molecular weight markers are indicated with the molecular weights in thousands. Gels of fragment C1 were stained with periodic acid-Schiff reagent. Proteins were purified and electrophoresis was performed as described under "Experimental Procedures." thrombin accounted substantially for the exosite I-dependent interaction seen with the factor Va heterodimer ( Table I).
Effect  Fig. 8, in which the fitted linear, exosite I-independent fluorescence change has been subtracted from all of the results to illustrate the effects more clearly. Native thrombin decreased the affinity of exosite I-dependent binding of factor Va, whereas it had no detectable effect on the exosite I-independent fluorescence change (Fig. 8). In the absence of factor Va, native thrombin increased the fluorescence of [ANS]FPR-thrombin by Յ 3.5% at concentrations up to 10 M, indicating no significant nonspecific interaction. The effect of native thrombin on the exosite I-dependent interaction was well described by competitive binding of native thrombin and [ANS]FPR-thrombin to factor Va, with a ϳ2.5-fold higher dissociation constant of 2.9 Ϯ 0.7 M for native thrombin (Fig. 8 and Table I). Simulation of the effect that competitive binding would have on the slope of the exosite I-independent fluorescence change indicated that a Յ15 M dissociation constant would have been detected. Thus, these results indicated that the linear fluorescence increase either did not represent an interaction of native thrombin at all or represented a low affinity interaction with a dissociation constant of Ͼ15 M.

DISCUSSION
The results of quantitative characterization of the binding of active site-labeled thrombin to human factor V/Va species support the conclusion that binding of thrombin to a site on the factor V/Va heavy chain through exosite I is the predominant mode of nonenzymatic interaction. The unique properties of thrombin labeled at the catalytic site with ANS linked by N ␣ -[(acetylthio)acetyl]-FPR-CH 2 Cl in signaling binding of factor V/Va but not Hir 54 -65 allowed the contribution of exosite I to the fluorescence changes to be resolved from exosite I-independent changes. The difference in the fluorescence changes for factor V/Va and Hir 54 -65 reflects the dependence of the pertur-  Table I. Titrations were performed and analyzed as described under "Experimental Procedures." bations of the microenvironment of the probe on the structures of the ligands. This is consistent with the properties of [ANS] FPR-thrombin in reporting other exosite I (35,40,41) and exosite II (31,35,40) interactions with fluorescence changes that depend in sign and magnitude on the structure of the ligand as well as the site of binding. For several of these interactions, the fluorescence changes are correlated with changes in catalytic specificity for tripeptide substrates (42)(43)(44)(45). On this basis, the present results indicate that binding of factor V and Va are accompanied by changes in the active site of thrombin that are distinct from those produced by Hir 54 -65 and that may be associated with changes in substrate specificity. The fluorescence enhancements may be due to conformational changes induced by factor V and Va or may be the result of proximity of part of the much larger structures of factor V and Va to the probe in the complexes.
Analysis of the binding of factor V and Va to [ANS]FPRthrombin indicates that the highest affinity interaction occurs through exosite I, in competition with Hir 54 -65 binding to this site. The observed competitive binding of native thrombin and [ANS]FPR-thrombin to factor Va confirmed that the exosite I interaction was a property of the native enzyme, and the ϳ2.5fold lower affinity for native thrombin indicated that the presence of the probe had only a small effect. The linear increases in [ANS]FPR-thrombin fluorescence observed in titrations with factor V and Va when exosite I was blocked by Hir 54 -65 represent exosite I-independent interactions that are of uncertain significance. The possibility that these changes represent nonspecific interactions with the covalently attached probe was demonstrated by the similar effect of high concentrations of unrelated proteins. The magnitudes of the fluorescence changes for factor V/Va species were larger, however, and the possibility that they represent significant interactions through other sites on thrombin cannot be excluded. The absence of an observable effect of native thrombin on the exosite I-independent fluorescence changes indicates that if such interactions occur with the native enzyme, they have an affinity at least 5-fold lower than exosite I-dependent binding.
Comparison of the interactions of each of the isolated factor V activation products with [ANS]FPR-thrombin demonstrated one exosite I-dependent interaction with the factor Va heavy chain, and exosite I-independent interactions of lower affinity for fragments E and C1. The light chain showed no detectable affinity for [ANS]FPR-thrombin, in agreement with previous affinity chromatography results (21). Binding of the isolated factor Va heterodimer to [ANS]FPR-thrombin accounted for the results obtained with the complete mixture of factor V activation products, and the heavy chain subunit bound with an indistinguishable fluorescence enhancement and only 2-4fold lower affinity than factor Va. It is concluded from these results that the binding site for thrombin on factor Va is contained in the heavy chain alone, with little or no significant interaction with the light chain or effect of association of the subunits in the factor Va dimer.
The factor Va heavy chain contains two acidic sequences within residues 659 -698 that are homologous to hirudin 54 -65 and include one or more sulfated tyrosine residues (7,46,47). Tyrosine sulfate occurs frequently in exosite I-binding sequences, and sulfation of Tyr 63 in hirudin 54 -65 results in a 5-10-fold increase in affinity of the peptide for thrombin (45,46). One of the hirudin 54 -65 -like sequences in the factor Va heavy chain likely represents the site that interacts directly with exosite I. However, the reported linkage between binding of certain ligands to exosites I and II (43) raises the alternative possibility that factor V/Va binds to a different site on thrombin and that binding of Hir 54 -65 to exosite I results in a conformational change that greatly reduces the affinity of factor V/Va binding at the linked site. As is generally the case, if the changes in affinity are large, this possibility cannot be easily distinguished from simple competitive binding. A second sequence homologous to hirudin 54 -65 precedes the Arg 1545 cleavage site in fragment C1 and has also been proposed as a possible thrombin binding site, whereas no similar sequence neighbors the Arg 1018 site in fragment E (7, 46 -48). The fluorescence results for fragment E agreed with the observation of no detectable affinity of this fragment for thrombin by affinity chromatography (21). Although no evidence for an exosite I interaction with fragment C1 was obtained, this fragment produced the largest exosite I-independent increase in fluorescence, which could represent a lower affinity interaction with a different site on thrombin.
Binding of thrombin to the heavy chain region of factor V through exosite I is concluded to play an important role in factor V activation by mediating factor V binding as a specific thrombin substrate. This interaction is thought to enhance the specificity of cleavage at Arg 709 that initiates the factor V activation mechanism by approximating thrombin to the cleavage site and possibly by inducing changes in the thrombin catalytic site that affect substrate specificity. The characteristics of factor V and factor Va binding to [ANS]FPR-thrombin were slightly different, with factor V producing a 1.4-fold larger fluorescence enhancement and binding with 3-fold lower affinity. These relatively small differences reflect the effects of the activation bond cleavages and dissociation of the activation fragments on the environment of the thrombin active site and its binding site in the complexes with the intact protein substrate and the final reaction product. The K m of 0.07 M reported for factor V activation (9) can be taken as an approximation of the overall dissociation constant for factor V binding as a substrate. The 49-fold and 16-fold higher dissociation constants for factor V and factor Va binding, respectively, represent estimates of only the affinity of the exosite I-mediated interactions because the catalytic site and S1-S3 2 substrate binding subsites are occupied in the [ANS]FPR-thrombin derivative. With the assumption that the overall affinity includes contributions from both exosite I and catalytic site interactions, exosite I-mediated binding would account for 76% of the free energy of factor V binding as a substrate.
The small changes in affinity for the exosite I interaction accompanying conversion of factor V to factor Va suggest that thrombin binds similarly to factor V activation intermediates and that these interactions may regulate other steps in the activation pathway. The activation cleavage reactions are not strictly sequential but follow the preferred order of Arg 709 first to generate the heavy chain, followed by Arg 1018 , and Arg 1545 last to produce the light chain (4,8,9,18,19). Cleavage at Arg 709 is important for rapid initiation of factor V activation, whereas development of full factor Va activity requires cleavage at Arg 1545 (8,18,19,48). Studies of recombinant factor V mutants indicate that the cleavages at Arg 709 and Arg 1018 generate partially active intermediates and that these steps enhance the rate of cleavage at Arg 1545 (8,18,19,48). The effect of inhibiting exosite I interactions with Hir 54 -65 or hirugen indicated that the Arg 709 and Arg 1545 reactions were exosite I-dependent, whereas the results for Arg 1018 were less clear (20,21). Cleavage at Arg 1545 has also been shown to depend on the presence of the region of fragment C1 containing the sequence that is a putative binding site for thrombin (48). The absence of evidence for an exosite I-dependent interaction of [ANS]FPR-thrombin with isolated fragment C1 suggests that an interaction of thrombin with this site may be expressed differently in factor V activation intermediates and that the affinity may be lost in the released fragment. An alternative possibility is that thrombin binding to the heavy chain could regulate cleavage at Arg 1545 in addition to Arg 709 . A precedent for this possibility is the mechanism of conversion of fibrinogen to fibrin, in which thrombin binding through exosite I facilitates fibrinogen binding and cleavage of two bonds to release fibrinopeptides A and B (50 -52). The exosite I interaction is maintained with essentially the same affinity for the substrate, fibrinogen; the intermediate, fibrin I; and the final product, fibrin II (50 -52).
The properties of thrombin binding to factor Va suggest that the binding site on factor Va may be related to the site of prothrombin binding, which functions in the cofactor-assisted interaction of prothrombin as a substrate of the factor Xa-factor Va complex. In parallel with the results for thrombin, bovine prothrombin has been shown to bind to a site on the factor Va heavy chain in a calcium-independent interaction, and it binds with essentially the same affinity to the isolated subunit and the factor Va dimer (15,16,27). Kinetic studies of prethrombin 2 activation by the bovine factor Xa-factor Va-membrane complex have shown that substrate recognition is mediated by exosites present on the enzyme complex (25,26). Active siteblocked thrombin inhibits prethrombin 2 activation through competitive exosite binding, and the inhibitory interaction is greatly diminished in affinity by hirugen (26). The present results with the human proteins suggest that the site of thrombin binding through exosite I on factor Va may overlap one of the exosites in the enzyme complex that contributes to productive substrate interactions. This explanation is apparently in conflict with the results of additional studies with proteolytic derivatives of thrombin, which indicated that exosites I and II were not required for inhibition of the factor Xa-factor Va-membrane complex (26). It has been proposed that thrombin binding as an inhibitor occurs through a site that is distinct from exosite I but is conformationally linked to hirugen binding, such that the affinity of the inhibitory interaction is greatly reduced (26). Another possibility is that the substrate and product interactions with exosites expressed in the factor Xa-factor Va-membrane complex include thrombin binding to factor Va in addition to other modes of binding. Additional support for the idea that thrombin-factor Va binding may be related to factor Va cofactor activity is given by the results of studies of recombinant human factor V in which sulfation of tyrosine residues in the factor Va heavy chain and fragment C1 were suppressed (47). Nonsulfated factor V showed a decreased rate of thrombin cleavage at Arg 709 and generation of factor Va activity, consistent with the postulated role of hirudin 54 -65 -like sequences in exosite I-mediated thrombin binding to the heavy chain. In addition, nonsulfated factor Va showed reduced procoagulant activity, implicating the sulfated tyrosine residues in the factor Va heavy chain in factor Va activity. Together with previous observations, the results of the present studies indicate that the site of thrombin binding to factor V and Va through exosite I plays a major role in directing the initial step of the factor V activation pathway and suggest that this site may also be involved in interactions of factor Va in the regulation of prothrombin activation.