Exosite binding tethers the macromolecular substrate to the prothrombinase complex and directs cleavage at two spatially distinct sites.

The prothrombinase complex, composed of the proteinase, factor Xa, bound to factor Va on membranes, catalyzes thrombin formation by the specific and ordered proteolysis of prothrombin at Arg(323)-Ile(324), followed by cleavage at Arg(274)-Thr(275). We have used a fluorescent derivative of meizothrombin des fragment 1 (mIIaDeltaF1) as a substrate analog to assess the mechanism of substrate recognition in the second half-reaction of bovine prothrombin activation. Cleavage of mIIaDeltaF1 exhibits pseudo-first order kinetics regardless of the substrate concentration relative to K(m). This phenomenon arises from competitive product inhibition by thrombin, which binds to prothrombinase with exactly the same affinity as mIIaDeltaF1. As thrombin is known to bind to an exosite on prothrombinase, initial interactions at an exosite likely play a role in the enzyme-substrate interaction. Occupation of the active site of prothrombinase by a reversible inhibitor does not exclude the binding of mIIaDeltaF1 to the enzyme. Specific recognition of mIIaDeltaF1 is achieved through an initial bimolecular reaction with an enzymic exosite, followed by an active site docking step in an intramolecular reaction prior to bond cleavage. By alternate substrate studies, we have resolved the contributions of the individual binding steps to substrate affinity and catalysis. This pathway for substrate binding is identical to that previously determined with a substrate analog for the first half-reaction of prothrombin activation. We show that differences in the observed kinetic constants for the two cleavage reactions arise entirely from differences in the inferred equilibrium constant for the intramolecular binding step that permits elements surrounding the scissile bond to dock at the active site of prothrombinase. Therefore, substrate specificity is achieved by binding interactions with an enzymic exosite that tethers the protein substrate to prothrombinase and directs cleavage at two spatially distinct scissile bonds.

The prothrombinase complex, composed of the proteinase, factor Xa, bound to factor Va on membranes, catalyzes thrombin formation by the specific and ordered proteolysis of prothrombin at Arg 323 -Ile 324 , followed by cleavage at Arg 274 -Thr 275 . We have used a fluorescent derivative of meizothrombin des fragment 1 (mIIa⌬F1) as a substrate analog to assess the mechanism of substrate recognition in the second half-reaction of bovine prothrombin activation. Cleavage of mIIa⌬F1 exhibits pseudo-first order kinetics regardless of the substrate concentration relative to K m . This phenomenon arises from competitive product inhibition by thrombin, which binds to prothrombinase with exactly the same affinity as mIIa⌬F1. As thrombin is known to bind to an exosite on prothrombinase, initial interactions at an exosite likely play a role in the enzyme-substrate interaction. Occupation of the active site of prothrombinase by a reversible inhibitor does not exclude the binding of mIIa⌬F1 to the enzyme. Specific recognition of mIIa⌬F1 is achieved through an initial bimolecular reaction with an enzymic exosite, followed by an active site docking step in an intramolecular reaction prior to bond cleavage. By alternate substrate studies, we have resolved the contributions of the individual binding steps to substrate affinity and catalysis. This pathway for substrate binding is identical to that previously determined with a substrate analog for the first half-reaction of prothrombin activation. We show that differences in the observed kinetic constants for the two cleavage reactions arise entirely from differences in the inferred equilibrium constant for the intramolecular binding step that permits elements surrounding the scissile bond to dock at the active site of prothrombinase. Therefore, substrate specificity is achieved by binding interactions with an enzymic exosite that tethers the protein substrate to prothrombinase and directs cleavage at two spatially distinct scissile bonds.
Prothrombinase is an archetypal enzyme complex of blood coagulation (2). The enzyme complex assembles through well characterized, reversible, protein-protein and protein-membrane interactions between the serine protease, factor Xa, the cofactor, factor Va, and membranes in the presence of calcium ions (2)(3)(4). The resulting complex catalyzes the conversion of prothrombin to thrombin at a greatly enhanced rate, compared with the reaction rate catalyzed by factor Xa alone (2).
Prothrombin is activated by proteolytic cleavage at two sites, Arg 274 -Thr 275 and Arg 323 -Ile 324 , which yields the fragment 1.2 activation peptide and thrombin 1 (5,6). The reaction catalyzed by prothrombinase proceeds almost exclusively via the initial cleavage at Arg 323 -Ile 324 , yielding meizothrombin as an intermediate, followed by cleavage at Arg 274 -Thr 275 to yield the final products of the reaction (7,8). Single turnover kinetic studies indicate that the overall process is likely the sum of two consecutive enzyme-catalyzed reactions (8). Consequently, steady state kinetic constants derived from measurements of the conversion of prothrombin to thrombin are difficult to interpret and are unlikely to provide valid mechanistic insights into this process. This problem can be circumvented by the use of proteolytic derivatives of prothrombin as analog substrates for the individual half-reactions of prothrombin activation (8 -11). Prethrombin 2, generated by preparative cleavage at Arg 274 -Thr 275 , has been shown to be a valid substrate analog for kinetic studies of the cleavage at Arg 323 -Ile 324 , which represents the first cleavage reaction in the activation of prothrombin by prothrombinase (8,12). Prior cleavage at Arg 274 -Thr 275 was shown to have no effect on the recognition and cleavage of the Arg 323 -Ile 324 site by prothrombinase (8). Considerable advances in the understanding of substrate-prothrombinase interactions have been gained by studies with prethrombin 2 in the bovine system (12)(13)(14). The results support a model in which the substrate, prethrombin 2, binds to prothrombinase via a multistep reaction (13,14). The affinity of the enzyme for the substrate is determined by a bimolecular reaction between prethrombin 2 and extended macromolecular recognition sites (exosites) on the enzyme complex. This step is followed by interactions between elements surrounding the scissile bond with the active site of the enzyme followed by bond cleavage. The resulting product, thrombin, remains bound to the exosite and requires dissociation for subsequent rounds of catalysis. Since active site interactions between substrate and enzyme were found to be unfavorable, it has been suggested that bind-ing specificity for cleavage at Arg 323 -Ile 324 is largely determined by exosite interactions with the enzyme (13).
Comparable information on the second half-reaction, in which the Arg 274 -Thr 275 peptide bond is cleaved by prothrombinase, is lacking. Although prior cleavage at Arg 274 -Thr 275 does not influence the kinetics of cleavage at Arg 323 -Ile 324 , the reverse is not true (8). Initial cleavage at Arg 323 -Ile 324 has been suggested to enhance the rate of cleavage at Arg 274 -Thr 275 by a factor of ϳ10 (11). The rate of cleavage at Arg 274 -Thr 275 , which converts meizothrombin to thrombin, is only modestly stimulated by factor Va (9,15). Structural models based on x-ray diffraction studies indicate that the two cleavage sites in prothrombin are spatially distinct and separated by as much as 36 Å (16,17). Finally, rapid kinetic studies support the possibility that the two cleavage reactions catalyzed by prothrombinase derive from two distinct types of substrate-enzyme interactions (8). Taken together, these observations suggest that there may be significant differences in the mechanisms underlying the recognition and cleavage of the two bonds in prothrombin by prothrombinase.
Meizothrombin (mIIa), 2 produced as the intermediate of prothrombin activation by prothrombinase following initial cleavage at Arg 323 -Ile 324 , is the relevant substrate analog for kinetic studies of the action of prothrombinase on the Arg 274 -Thr 275 cleavage site (8,9,15). Single turnover kinetic studies have established that proteolytic removal of the membrane-binding fragment 1 domain from mIIa to yield meizothrombin des fragment 1 (mIIa⌬F1) does not affect the kinetics of substrate recognition and cleavage by prothrombinase (8). Thus, although mIIa can bind membranes (18,19), this interaction does not play an obvious enhancing role in its ability to be recognized and cleaved by prothrombinase. Therefore, mIIa⌬F1 is a valid substrate analog for the second half-reaction of prothrombin activation that permits studies of the enzyme-substrate interaction in the absence of the obscuring effects of membrane-mediated substrate delivery steps (8,20,21). We have therefore pursued steady state kinetic studies of the cleavage of bovine mIIa⌬F1 by bovine prothrombinase to further investigate the mechanisms underlying macromolecular substrate recognition by prothrombinase.
Three types of catalytically inactivated derivatives of mIIa⌬F1 were prepared by preparative cleavage of prethrombin 1 by ecarin in the presence of the appropriate peptidyl chloromethyl ketone. The nonfluorescent, inactivated derivative mIIa⌬F1 I was prepared by cleavage in the presence of FPR-CH 2 Cl. Dansyl-modified product, mIIa⌬F1 D , was obtained by cleavage in the presence of DEGR-CH 2 Cl as described previously (8). The fluorescein derivative, mIIa⌬F1 F , was prepared by cleavage in the presence of ATA-FPR-CH 2 Cl followed by thioester hydrolysis and modification with 6-(iodoacetamido)fluorescein. Typically, mIIa⌬F1 I was prepared by cleavage of prethrombin 1 (12.7 M, 35 mg) in 20 mM Hepes, 50 mM NaCl, pH 7.5, by ecarin (8 g/ml) at 25°C, in the presence of 50 M FPR-CH 2 Cl. After 25 min, the reaction mixture was quenched by the addition of 10 mM EDTA, and excess inhibitor was removed by gel filtration using a column (2.5 ϫ 120 cm) of Sephadex G-25 equilibrated in 20 mM Hepes, pH 7.5. Protein eluting in the void volume was pooled and applied directly to a Poros HS/M column (4.6 ϫ 100 mm) to rapidly adsorb ecarin and side products. Unbound protein was pooled and subjected to high resolution anion exchange using a Poros HQ/M 10 ϫ 100-mm column (PerSeptive Biosystems). Bound mIIa⌬F1 derivatives were eluted at 6.4 ml/min with a gradient of increasing NaCl (0.05-0.2 M, 96 ml) in 20 mM Hepes, pH 7.5. Fractions containing mIIa⌬F1 I were pooled, precipitated with (NH 4 ) 2 SO 4 , collected by centrifugation (100,000 ϫ g, 45 min), dissolved in 50% (v/v) glycerol, and stored at Ϫ20°C.
Steady State Fluorescence Measurements-All fluorescence measurements were performed in stirred quartz cells (1 ϫ 1 cm) in 20 mM Hepes, 0.15 M NaCl, 2 mM CaCl 2 , 0.1% (w/v) polyethylene glycol with average M r ϭ 8000, pH 7.5 (assay buffer), maintained at 25°C. The cleavage of mIIa⌬F1 F was studied using a LS50B fluorescence spectrophotometer (PerkinElmer Life Sciences) using ex ϭ 494 nm and em ϭ 525 nm. Scattered light was minimized using a long pass filter (KV-500, Schott, Dureay, PA) in the emission beam and slits were adjusted to minimize photobleaching. The cleavage of mIIa⌬F1 D was measured in a modified SLM8000 fluorescence spectrophotometer (37), using ex ϭ 280 nm and monitoring broadband fluorescence ( em Ն 500 nm) isolated with a long pass filter (Schott KV-500) in the emission beam. Experiments with both substrates were also conducted using a customized rapid scanning fluorescence spectrophotometer (RSM-1000, On-Line Instrument Systems, Bogart, GA). Spectral-time data sets were obtained using the appropriate excitation wavelength for the probe studied and rapidly scanning the emission monochromator. Data collection times, time constants, and integration times were adjusted to permit the collection of 400 or 1000 spectra over 8 -10 half-lives of the measured reaction.
For all kinetic studies, protein substrates and product inhibitors were exchanged into assay buffer either by dialysis or by column centrifugation using Sephadex G-25 prior to use.
Progress Curve Analysis of mIIa⌬F1 Cleavage-The reaction mixture contained 8.0 M mIIa⌬F1 (0.2 M mIIa⌬F1 F plus 7.8 M mIIa⌬F1 I ), 50 M PCPS, and 30 nM factor Va in assay buffer. Cleavage was initiated by the addition of 1 nM Xa, and fluorescence was monitored continuously. At the indicated times, aliquots (65 l) were withdrawn from a parallel reaction mixture and quenched by mixing with 35 l of 156 mM Tris, 29 mM EDTA, 25% (v/v) glycerol, 5% (w/v) SDS, 0.025% (w/v) bromphenol blue, pH 6.8, heated, and analyzed by SDS-PAGE (38). Following electrophoresis, fluorescent bands were detected and analyzed using a Storm 840 fluorescence scanner and ImageQuaNT software (Molecular Dynamics, Sunnyvale, CA). Protein bands were visualized by staining with Coomassie Brilliant Blue, imaged with transmitted light using a Kodak DC120 digital camera (Eastman Kodak Co.), and analyzed using ImageQuaNT. The exposure and aperture settings for the imaging step were selected to yield a linear densitometric response.
Steady State Kinetics of mIIa⌬F1 Cleavage-The dependence of cleavage rate on the concentration of mIIa⌬F1 was determined using reaction mixtures containing increasing concentrations of substrate Cleavage was initiated by the addition of factor Xa to a final concentration of 2, 3, or 6 nM, and reaction progress was inferred by continuously monitoring fluorescence intensity of the fluorescein probe.
Inhibition Kinetics of Peptidyl Substrate Cleavage-mIIa⌬F1 I used in alternate substrate studies was treated with an excess of amidinophenylmethanesulfonyl fluoride to inactivate traces of active proteinase followed by dialysis into assay buffer prior to use. Reaction mixtures (170 l), prepared in wells of a 96-well plate, contained increasing concentrations of SpXa, 50 M PCPS, 40 nM Va, and different fixed concentrations of mIIa⌬F1 I (0 -40 M) added shortly before initiation. The reactions were initiated by the addition of Xa (30 l) to achieve a final concentration of 0.5 nM. Following mixing by brief vibration, peptidyl substrate hydrolysis was measured by continuously monitoring absorbance at 405 nm in a kinetic plate reader (Spectramax 250, Molecular Devices, Sunnyvale, CA). Control experiments without added Xa indicated that the velocity contribution due to active proteinase contamination in the mIIa⌬F1 I preparation was negligible (Ͻ1%), at the highest concentrations of mIIa⌬F1 I .
Data Analysis-Reactant concentrations were chosen to ensure that the concentrations of PCPS and Va were saturating relative to the concentration of factor Xa. Based on measured equilibrium and rate constants (39,40), greater than 95% of the factor Xa, under these conditions, is expected to be incorporated into the prothrombinase complex rapidly (t 1 ⁄2 Х 30 ms) following initiation of the reaction with factor Xa. The concentration of enzyme (prothrombinase) was therefore considered equivalent to the limiting concentration of factor Xa present in the reaction mixture. This expectation was empirically verified by documenting a linear dependence of reaction rate on the concentration of factor Xa, provided its concentration was limiting.
Data were analyzed according to the indicated equations using nonlinear least squares analysis (41). The quality of the fit was assessed by the criteria described (42). Fitted parameters are reported Ϯ 95% confidence limits.
The initial, steady state rate of product formation was derived from the initial slope of the fluorescence traces and converted to concentration terms using the limits of the progress curve to signify 0% and 100% conversion of substrate to product. This approach is justified by the results of densitometry analysis following SDS-PAGE (below).
Images of fluorescent or Coomassie Blue-stained bands were analyzed by volume integration (pixel intensity integrated over band area) and normalized using the signal obtained by integration over the entire lane. The resultant normalized densitometry results were converted to concentrations of substrate and product at each time point using the considerations described in detail (8).
Time-resolved emission spectra, obtained by rapid scanning fluorescence measurements, were analyzed by singular value decomposition (43), to determine the number of kinetically and optically resolved species, followed by analysis of the eigenvectors to yield rate constant information. Global analysis was according to a single exponential, to yield the component spectra of the starting and limiting species, a normalized wavelength-independent kinetic trace for the spectral transition, and k obs for the transition.
Fixed wavelength kinetic traces were analyzed according to a single exponential.
The signal at time t (F t ) is determined by the limits of the signal at zero (F 0 ) and infinite (F ϰ ) time and the observed pseudo-first order rate constant (k obs ) for the process. Non-linear least squares analysis yielded fitted values for F 0 , F ϰ , and k obs . The dependence of k obs on the initial concentrations of substrate (S) and product (P) was analyzed according to the rate expression developed by Mihalyi (44), for the limiting case where substrate and product bind competitively to the enzyme with identical affinities.
E T , V max , and K m have the usual meaning with the constraint K m ϭ K p . K p refers to the equilibrium dissociation constant for the binding of P to E. Initial velocity data for the cleavage of mIIa⌬F1 (S) obtained at increasing concentrations of substrate in the presence of different fixed concentrations of PAB (I) were analyzed according to the rate expression for Scheme I previously developed using the rapid equilibrium assumption (13).
Alternate substrate effects of the protein substrate (S) on the cleavage of the peptidyl substrate (SpXa) were analyzed according to Scheme I, assuming that the binding of SpXa to the enzyme occurs in steps equivalent to those illustrated for the binding of active site-directed ligand, I, followed by cleavage to yield the chromophoric product measured by monitoring absorbance at 405 nm.
SpXa refers to the concentration of the peptidyl substrate, k cat(SpXa) and K SpXa are the kinetic constants for the cleavage of SpXa by prothrombinase, S is the concentration of the protein substrate, and the constants K s and K s * refer to the equilibrium dissociation constants illustrated in Scheme I. Initial velocity measurements of the rate of SpXa cleavage at increasing concentrations of SpXa in the presence of different fixed concentrations of S, were analyzed according to Equation 6 to yield fitted values of k cat(SpXa) , K SpXa , K s , and K s *.

RESULTS
Experimental Design-Because of autocatalytic degradative reactions, mIIa and mIIa⌬F1 can only be produced in stable form in the presence of proteinase inhibitors (9,45). Stable derivatives of mIIa and mIIa⌬F1 have previously been produced by inactivation with DEGR-CH 2 Cl, wherein the dansyl moiety serves as a reporter group for subsequent cleavage by prothrombinase (10). Measurements with mIIa⌬F1 D have relied on energy transfer from aromatic side chains to the dansyl moiety to yield a fluorescence intensity change associated with cleavage at Arg 274 -Thr 275 (8). In this approach, the useful substrate concentration range is limited by probe sensitivity and the inner filter effect. Some of these limitations could be overcome by the preparation of a fluorescent derivative, mIIa⌬F1 F , containing 6-(iodoacetamido)fluorescein incorporated into the active site with a peptidyl chloromethylketone linker.
Cleavage of mIIa⌬F1, monitored using the fluorescein probe in mIIa⌬F1 F , yielded a useful change in fluorescence intensity. Following initiation of cleavage with prothrombinase, fluorescence intensity increased with time to yield a limiting enhancement of ϳ9% (Fig. 1). SDS-PAGE analysis and imaging of the fluorescent bands indicated that the increase in fluorescence intensity correlated with the appearance of the expected products following cleavage of the Arg 274 -Thr 275 bond in mIIa⌬F1 F (Fig. 1, inset). Densitometry analysis of product formation either from the fluorescence gel image or following staining for total protein with Coomassie Brilliant Blue yielded progress curves that were essentially indistinguishable from each other and from the continuously measured fluorescence trace (Fig. 1). The substrate solution (0.2 M mIIa⌬F1 F plus 7.8 M mIIa⌬F1 I ) contained an excess of the inactivated but nonfluorescent derivative (mIIa⌬F1 I ) relative to the concentration of mIIa⌬F1 F . The densitometry results (Fig. 1) indicate that mIIa⌬F1 F and mIIa⌬F1 I are cleaved at identical rates. Within experimental error, the change in fluorescence intensity of mIIa⌬F1 F is coincident with the kinetics of cleavage of the Arg 274 -Thr 275 bond in the substrate. Thus, mIIa⌬F1 F can be utilized as a valid tracer to report the cleavage of the Arg 274 -Thr 275 site in the bulk substrate. This was further validated by showing that the rate of product formation, inferred from the fluorescence change, was independent of the ratio of the fluorescent to non-fluorescent substrate species at a fixed total concentration of mIIa⌬F1 (data not shown). Therefore, the total substrate concentration (mIIa⌬F1) was considered equal to the sum of the concentrations of mIIa⌬F1 F and mIIa⌬F1 I and initial velocities of mIIa⌬F1 cleavage were calculated using the limits of the fluorescence trace to signify 0 and 100% conversion of bulk substrate to product.
The amplitude of the fluorescence change, observed upon substrate cleavage, was influenced by the total concentration of mIIa⌬F1, even when the concentration of mIIa⌬F1 F was held constant. The limiting enhancement varied from 20% at 0.1 M mIIa⌬F1 F to 6% at 0.1 M mIIa⌬F1 F plus 14.9 M mIIa⌬F1 I , despite the fact that quantitative cleavage of the substrate could be established by SDS-PAGE. Further studies indicated that equivalent fluorescence quenching of the products could be achieved with increasing concentrations of fragment 2 (data not shown). This observation is consistent with the reported finding that fragment 2 binds with modest affinity to thrombin (K d Х 5 M) and quenches the fluorescence of 6-(iodoacetamido)fluorescein incorporated into the active site using an ATA-FPR-CH 2 Cl tether (46). Thus, at least part of the fluorescence change upon cleavage of mIIa⌬F1 F by prothrombinase likely arises from the dissociation of fragment 2 from the fluorescent thrombin species following cleavage at Arg 274 -Thr 275 . The equivalence between kinetics of bond cleavage and the fluorescence change (Fig. 1) implies that such steps which may contribute to the increase in fluorescence are not rate-limiting and do not compromise the use of the fluorescence signal to infer rates of bond cleavage.
Rapid Scanning Measurements of mIIa⌬F1 Cleavage-The possibility that the change in intensity upon cleavage of mIIa⌬F1 F arises from discrete steps was further investigated by rapid scanning fluorescence measurements. Spectral-time data sets, obtained upon cleavage of 0.3 M mIIa⌬F1 F by prothrombinase, were globally analyzed by singular value decomposition (Fig. 2). The wavelength-independent kinetic trace could be adequately described by a single exponential transition from a less fluorescent species to one with ϳ20% greater fluorescence intensity as denoted by the deconvoluted component spectra (Fig. 2, inset). Thus, the change in probe fluorescence is coincident with cleavage at Arg 274 -Thr 275 in the substrate with no evidence to suggest the presence of additional complicating steps represented by spectrally or kinetically resolved intermediates on the steady state timescale.
Steady State Kinetics of mIIa⌬F1 Cleavage-Steady state kinetic constants for the cleavage of mIIa⌬F1 by prothrombinase were determined from initial velocity measurements using mixtures of mIIa⌬F1 F and mIIa⌬F1 I . Initial velocities determined at increasing concentrations of total substrate (mIIa⌬F1) could be adequately described by the Henri-Michaelis-Menten equation (Fig. 3) yielding K m ϭ 4.2 Ϯ 0.3 M and V max /E T ϭ 91 Ϯ 3 s Ϫ1 . Similar experiments with mIIa⌬F1 D (data not shown), where the total substrate concentration was equal to the fluorophore-labeled species yielded K m ϭ 2.0 Ϯ 0.3 M and V max /E T ϭ 105 Ϯ 6 s Ϫ1 . Thus, kinetic constants in tolerable agreement are obtained, regardless of the fluorophore or peptidyl tether incorporated into the active site of mIIa⌬F1. However, the finding that bond cleavage proceeds with pseudofirst order kinetics at a substrate concentration well above the determined K m (Fig. 1) questions the physical significance of these constants and implies a violation of assumptions underlying the derivation of the Henri-Michaelis-Menten equation (47).
Further analysis revealed that pseudo-first order kinetics were observed for mIIa⌬F1 cleavage by prothrombinase, regardless of the substrate concentration relative to K m . This unusual observation was sustained in measurements with mIIa⌬F1 F and mIIa⌬F1 D , as well as by direct measurements of bond cleavage by SDS-PAGE (e.g. Fig. 1). Similar observations were also made when the reversible fluorescent inhibitor, dansylarginine-N-(3-ethyl-1,5-pentanediyl)amide, was used to infer reaction progress (9). This is illustrated by the adequate fit of progress curves to a single exponential rise at total substrate concentrations corresponding to ϳ0.07 ϫ K m or ϳ3 ϫ K m (Fig.  4). The observed rate constant was found to decrease with increasing substrate concentration.
As described by Mihalyi (44), these circumstances are expected when there is competitive product inhibition, with the constraint that substrate and product bind to the enzyme with exactly equal affinity i.e. K m ϭ K p . The rate expression developed by Mihalyi (44) for these limiting conditions (Equation 2), predicts a linear dependence of the pseudo-first order rate constant (k obs ) on the concentration of enzyme and a hyperbolic decrease in k obs with increasing initial substrate concentra-tions. These criteria were met for the cleavage of mIIa⌬F1 (Fig.  5). The data were adequately described by Equation 2, to yield K m ϭ K p ϭ 2.6 Ϯ 0.2 M and V max /E T ϭ 70.2 Ϯ 3.5 s Ϫ1 . Similar experiments with mIIa⌬F1 D yielded comparable fits with the constants K m ϭ K p ϭ 2.0 Ϯ 0.1 M and V max /E T ϭ 107 Ϯ 5 s Ϫ1 . The linear dependence of k obs on enzyme concentration (Fig. 5,  inset), further ensures that the peculiarity of the observed kinetics does not arise from a rate-limiting unimolecular process that follows the cleavage of mIIa⌬F1 F .
Product Inhibition of mIIa⌬F1 Cleavage-Cleavage of the Arg 274 -Thr 275 bond in mIIa⌬F1 by prothrombinase yields fragment 2 and thrombin (45). The suggestion of a significant role for product inhibition in the cleavage of mIIa⌬F1 was further pursued by inhibition studies with the individual products. Increasing concentrations of fragment 2 did not detectably decrease k obs for mIIa⌬F1 F cleavage at concentrations as high as 12 M (Fig. 6, upper panel). In contrast, increasing concentrations of IIa i yielded a hyperbolic decrease in k obs (Fig. 6,  lower panel). Progress curve analysis of bond cleavage by SDS-PAGE verified these findings (data not shown). The data indicate that it is thrombin and not fragment 2 that is responsible  for competitive product inhibition, with equal affinity as the substrate, during mIIa⌬F1 cleavage by prothrombinase. This finding is somewhat surprising since the COOH terminus of the fragment 2 domain retains the residues preceding the cleaved scissile bond that are expected to interact with the active site of factor Xa within the prothrombinase complex.
Further verification of the initial conclusions was obtained by the analysis of data from all experiments conducted in the presence or absence of IIa i as a product inhibitor. These data, combined from separate experiments at different concentrations of mIIa⌬F1 and IIa i , could be adequately described by Equation 2 to yield K m ϭ K p ϭ 3.6 Ϯ 0.26 M and V max /E T ϭ 84.3 Ϯ 4.9 s Ϫ1 (Fig. 6, inset). The fitted value for K p is in agreement with the documented equilibrium dissociation constant for the binding of IIa i to prothrombinase (13,14). Thus, the anomalous kinetics of mIIa⌬F1 cleavage by prothrombinase arises from the fact that the substrate and thrombin bind in a mutually exclusive manner to the enzyme with identical apparent affinities.
Kinetic Scheme for mIIa⌬F1 Cleavage-Competitive product inhibition by thrombin, in this one-substrate/two-product system, implies that mIIa⌬F1 and thrombin bind to free enzyme in the catalytic cycle. Since fragment 2 does not significantly inhibit the reaction nor does it further enhance the inhibitory properties of thrombin at the concentrations tested, it follows that product release is either ordered (fragment 2 release, followed by thrombin release) or that product is released as the fragment 2-thrombin complex, wherein fragment 2 does not significantly enhance the ability of thrombin to bind the enzyme. These possibilities are incorporated into the product release steps illustrated in Scheme I.
Prior work has established that IIa i binds to prothrombinase through an enzymic exosite, without restricting access of small molecules to the catalytic site of factor Xa with the enzyme complex (13,14). The observation that IIa i and mIIa⌬F1 bind to the enzyme in a mutually exclusive fashion, with apparently identical affinity, raises the possibility that the initial interaction between mIIa⌬F1 (S) and prothrombinase (E) to form ES, may also occur through exosite interactions with the enzyme. Initial interactions at an enzymic exosite must be followed by docking interactions between substrate structures surrounding the scissile bond and the active site of the enzyme prior to catalysis. This sequence of substrate binding steps, previously delineated for the recognition and cleavage of the Arg 323 -Ile 324 bond in prethrombin 2 by prothrombinase (13), is illustrated in Scheme I as one possible explanation for the steps underlying the productive interaction between mIIa⌬F1 and the enzyme complex.
Exosite-dependent Recognition of mIIa⌬F1 by Prothrombinase-Previous studies have established a facile approach using active site-directed reversible inhibitors to infer the relative contributions of active site versus exosite interactions toward protein substrate recognition by coagulation enzymes (13,48). A hallmark of ordered, exosite-driven, protein substrate recognition is that active site-directed reversible inhibitors that exclude peptidyl substrate binding at the active site of the enzyme fail to act as competitive inhibitors of protein substrate cleavage (13,48).
The speculated sequence of substrate binding events (Scheme I) was further investigated by steady state inhibition studies with PAB. PAB is expected to bind reversibly to the S 1 specificity site 3 of arginine-specific serine proteinases (49) and is established to act as a well behaved, classical competitive inhibitor of peptidyl substrate cleavage by prothrombinase (13). 3 The nomenclature is that of Schechter and Berger (1). . The line is drawn using the fitted parameters: K m ϭ K P ϭ 3.6 Ϯ 0.26 M and V max /E T ϭ 84.3 Ϯ 4.9 s Ϫ1 . SCHEME I Inhibition of mIIa⌬F1 cleavage by PAB obviously deviated from classical competitive inhibition (Fig. 7). This observation indicates that PAB and mIIa⌬F1 do not bind in a mutually exclusive way to the enzyme. Analysis according to the rate expression for classical noncompetitive inhibition yielded a reasonable fit, suggesting that PAB binding to the active site of factor Xa within prothrombinase minimally alters mIIa⌬F1 binding and vice versa. This observation provides empirical support for the hypothesis that the initial binding of S to E (Scheme I), likely involves interactions at sites removed from the active site of the enzyme.
Small but systematic deviations were observed between the data and the lines fitted according to classical noncompetitive inhibition (Fig. 7). The K i for PAB, determined by this analysis, was approximately 2-fold greater than the measured K d for PAB binding to prothrombinase (50), further implying a systematic error in the description of the data by this rate expression. Classical noncompetitive inhibition is expected for the mechanism depicted in Scheme I for the limiting case when the formation of ES* is unfavorable (K s * Ͼ Ͼ 1) (13,48). However, deviation from classical noncompetitive inhibition is expected when this criterion is not satisfied. The complete initial velocity expression for Scheme I (Equation 3) provided a superior fit to the data (Fig. 7), and yielded fitted values for K i in agreement with the measured K d for the binding of PAB to prothrombinase (50). Provided Scheme I adequately accounts for mIIa⌬F1 binding to prothrombinase, the fitted estimates K s ϭ 17.4 Ϯ 4.4 M and K s * ϭ 0.37 Ϯ 0.14 yield the tentative conclusion that interactions between the substrate and the active site of the enzyme, depicted by the equilibrium distribution between ES and ES*, are favorable.
These preliminary conclusions were further pursued by initial velocity studies of the ability of mIIa⌬F1 I to inhibit oligopeptidyl substrate cleavage by prothrombinase. This approach was feasible because the products of mIIa⌬F1 I cleavage do not exhibit catalytic activity. The experimental strategy relies on the previous observation that protein substrate or product interactions at the exosite of prothrombinase have no detectable affect on the ability of the enzyme to bind and cleave peptidyl substrates (13,14). Thus, in initial velocity studies of peptidyl substrate cleavage in the presence of the protein substrate, E and E⅐S (Scheme I) are expected to exhibit equivalent kinetic constants while E⅐S* is not expected to bind and cleave the peptidyl substrate. The resulting rate expression (Equation 6) describes the dependence of the initial velocity of peptidyl substrate hydrolysis on the concentration of peptidyl substrate (SpXa) in the presence of increasing concentrations of the protein substrate (S). Equation 6 predicts very little inhibition by S when K s * Ͼ Ͼ 1 and inhibition equivalent to classical competitive inhibition when K s * Ͻ Ͻ 1. Partial inhibition by S at nonsaturating concentrations of SpXa is expected within these boundary conditions.
The initial velocity for SpXa cleavage by prothrombinase was decreased in the presence of mIIa⌬F1 I (Fig. 8). However, mIIa⌬F1 I was a partial inhibitor of SpXa hydrolysis, as illustrated by the residual activity of ϳ50% at saturating concentrations of mIIa⌬F1 I and ϳK m concentrations of SpXa (Fig. 8,  inset). The family of curves obtained at different fixed concentrations of mIIa⌬F1 I could be adequately described by Equation 6 to yield fitted values of K s ϭ 11.7 Ϯ 1.4 M and K s * ϭ 0.26 Ϯ 0.02 (Fig. 8).
Collectively, the inhibition studies yield results consistent with the substrate binding steps illustrated in Scheme I. We therefore propose that the reaction between mIIa⌬F1 and prothrombinase results from interactions at an enzymic exosite governed by K s (Scheme I). The formation of ES is followed by an intramolecular binding step, governed by K s * (Scheme I), which permits structures about the scissile to interact with the active site of the enzyme in a modestly favorable step, prior to catalysis.

DISCUSSION
Evidence for a major role of exosite interactions in the productive pathway for protein substrate recognition by prothrombinase has been previously developed in kinetic studies using prethrombin 2 as a substrate analog for cleavage at the Arg 323 -Ile 324 peptide bond in prothrombin (13,14). Data obtained in the present work using mIIa⌬F1 suggest that equivalent exosite interactions are relevant for protein substrate recognition and cleavage at the Arg 274 -Thr 275 site as well.
Evidence to support this conclusion derives from the observation that the product, thrombin (IIa i ), acts as a competitive inhibitor of the cleavage of either prethrombin 2 (Arg 323 -Ile 324 cleavage) or mIIa⌬F1 (Arg 274 -Thr 275 cleavage) by prothrombinase. In contrast, the binding of IIa i to prothrombinase has no obvious effect on the access of small ligands and peptidyl substrates to the active site of Xa within prothrombinase (13,14). Competitive inhibition of prethrombin 2 activation by IIa i oc- curs with a K i comparable to the affinity of prothrombinase for the substrate (13). In the case of mIIa⌬F1 cleavage, the K i for IIa i is exactly equal to the apparent affinity of the enzyme for this substrate. Thrombin and prethrombin 2 share a series of structural features (17), while thrombin represents the COOHterminal domain of mIIa⌬F1 (16). These points justify the reasonable conclusion that equivalent interactions with an enzymic exosite underlie the recognition and cleavage of both analog substrates for the two half-reactions of prothrombin activation.
Alternate substrate studies with mIIa⌬F1 have permitted a resolution of the thermodynamic contributions of the presumed exosite (K s , Scheme I) and active site interactions (K s *, Scheme I) to substrate affinity. Although the overall affinity of prothrombinase for mIIa⌬F1 is equal to the affinity for IIa i (i.e. K m ϭ K p ), this equivalence does not apply to the inferred thermodynamics of the exosite-binding step. The inferred equilibrium dissociation constant for the binding of mIIa⌬F1 to the exosite (K s Х 12 M) is approximately 4-fold greater than the values determined for exosite binding by prethrombin 2 or IIa i (13,14). However, mIIa⌬F1 contains the fragment 2 domain that is established to modulate interactions between the protease domain of the substrate and other macromolecular ligands (51)(52)(53). Thus, the data obtained in studies of mIIa⌬F1 cleavage require comparison with the kinetics of activation of prethrombin 2 saturated with fragment 2. The value for K s inferred for mIIa⌬F1 in the present work is strikingly similar to the K m previously determined for prethrombin 2 plus fragment 2 (12). For prethrombin 2 activation, the data are consistent with an unfavorable active site docking step, implying K s * Ͼ Ͼ 1 (estimated by simulations at K s * Ն 8) and K m Х K s in substrate binding steps equivalent to those illustrated in Scheme I (13).
The present results, and the application of Equations 4 and 5 to the steady state kinetic constants for prethrombin 2 plus fragment 2 (12), allow for a more appropriate comparison of the stepwise binding interactions that lead to the recognition of the two bonds in the substrate by prothrombinase followed by catalysis (Scheme II). Although prethrombin 2 plus fragment 2 and mIIa⌬F1 are individually cleaved by prothrombinase with different steady state kinetic constants, the dissociation constant for the exosite binding step (K s ) and the inferred rate constant for catalysis (k cat ) for these two substrates are equivalent (Scheme II). The major difference appears to lie in the equilibrium dissociation constant for the active site docking step (K s *) that precedes bond cleavage. Active site interactions that precede the cleavage of the Arg 274 -Thr 275 site in mIIa⌬F1 appear to be modestly favorable, whereas the comparable binding step that precedes cleavage of the Arg 323 -Ile 324 site in prethrombin 2 plus fragment 2 is an unfavorable step (Scheme II).
Comparable values for K s inferred for both protein substrate analogs (Scheme II) supports the contention that the initial interaction between both substrates and prothrombinase involves equivalent exosite binding steps. Since prethrombin 2 and thrombin appear to bind this enzymic exosite with greater affinity (K s Х 3 M) (13,14), it follows that the relatively high affinity interaction between fragment 2 and prethrombin 2 (12) modestly decreases the affinity of the resultant substrate for the enzymic exosite by a factor of 4. The binding of fragment 2 to prethrombin 2 has also been established to increase the V max for substrate cleavage by prothrombinase by approximately the same factor (12). Thus, while the rate-enhancing effects of fragment 2 on the cleavage of prethrombin 2 by prothrombinase are well established in the literature (12,54,55), the data are most consistent with the interpretation that the binding of fragment 2 to prethrombin 2 somehow alters the structure of the substrate, leading to a modest perturbation in the kinetic constants (12). Based on previous studies with proteolytic fragments of prethrombin 2 and thrombin (14), it seems probable that the reduced affinity of the fragment 2-prethrombin 2 complex for exosite binding to prothrombinase arises from linkage between distinct sites in the proteinase domain of the substrate that mediate fragment 2 binding and interactions with the enzymic exosite.
The interaction of fragment 2 with thrombin is also likely to reduce the affinity of the product for exosite binding. If the SCHEME II affinity changes observed with prethrombin 2 directly apply to thrombin, it follows that the equilibrium dissociation constant for the interaction between thrombin and the enzymic exosite is approximately 4-fold lower than that of the fragment 2-thrombin complex i.e. K p Х 3 M, ␤K p Х 12 M, ␤ Х 4 (Scheme I). Consequently, the previously measured equilibrium dissociation constant for the binding of fragment 2 to thrombin (K d ϭ 5 M) (46) is also expected to be increased by a factor of 4 in the presence of prothrombinase i.e. K F2,IIa Х 5 M, ␤K F2,IIa Х 20 M (Scheme I). These points provide a reasonable quantitative accounting for the inhibition of mIIa⌬F1 cleavage by IIa i as well as the modest increase in reaction rate observed in the presence of increasing concentrations of fragment 2 (Fig. 6).
Evidence for a high affinity interaction between fragment 2 and thrombin (K d ϭ 0.8 nM) has previously been obtained at an ionic strength much lower than those used in the present work (56). A strong ionic strength dependence of this interaction is implied by the substantially larger equilibrium dissociation constant measured by Bock at I ϭ 0.15 M (46), which seems to represent the most appropriate value for considerations of product inhibition in the present studies. However, a value of K F2,IIa substantially lower than 5 M would provide more compelling support for the conclusion that fragment 2 binding does not enhance the ability of IIa i to bind prothrombinase and would imply a far stronger destabilizing effect of fragment 2 on this interaction.
Significant differences in the rate constant for catalysis, inferred by division of V max by E T , have been previously noted for the two cleavage reactions in the protein substrate catalyzed by prothrombinase (9,10). An obvious explanation for this finding has not been forthcoming since identical P 1 -P 4 residues precede both cleavage sites in prothrombin (36). Assuming that Scheme II provides an adequate description of the binding steps in substrate recognition, the present results indicate that the rate constant for catalysis is essentially the same for the two cleavage reactions and is comparable to values observed for the cleavage of peptidyl substrates bearing the same P 1 -P 4 sequence encountered in the protein substrate (57).
Structural models for mIIa⌬F1 and prethrombin 2 from xray diffraction data indicate that the residues preceding the Arg 323 -Ile 324 bond are either disordered or require significant rearrangement to be successfully docked into the active site of factor Xa (17). Such features are not observed for the identical residues preceding the Arg 274 -Thr 275 bond in mIIa⌬F1 (16). These observations may provide a structural explanation for the large differences in the equilibrium constant for the active site docking step (K s *, Scheme II) inferred for the two protein substrates. Previous work has established that, although cleavage at Arg 323 -Ile 324 in prethrombin 2 is greatly accelerated by factor Va, the cofactor has a much smaller effect on the cleavage at the Arg 274 -Thr 275 bond in mIIa⌬F1 (9,15). If it is indeed true that exosite binding by the protein substrate is only significant following assembly of the prothrombinase complex (13,50,58), then this initial interaction, which serves to tether the substrate to the enzyme complex, is likely to disproportionately enhance cleavage at the disordered Arg 323 -Ile 324 site governed by an unfavorable active site docking step in contrast to cleavage at the Arg 274 -Thr 275 site, which results from a favorable interaction at the active site. This hypothesis implies that the rate-enhancing effects of factor Va at least partly arise from indirect or direct contributions toward exosite binding by the protein substrate. This initial tethering reaction could overcome inefficient catalysis at suboptimally configured cleavage sites in the protein substrate.
Structural studies of prothrombin derivatives indicate that the two scissile bonds in the substrate are separated by as much as 36 Å (16), yet the present data indicate that both cleavages derive from equivalent exosite interactions that initially tether the substrate to prothrombinase. Therefore, there must be a considerable rearrangement of the protein substrate to permit the structures surrounding spatially distinct scissile bonds to interact with the active site before cleavage (Scheme II). Prior work has established that prothrombin activation by prothrombinase proceeds by cleavage at Arg 323 -Ile 324 followed by cleavage at Arg 274 -Thr 275 (7,8). Within experimental error, cleavage in the opposite order has been undetectable despite the fact that both bonds appear accessible in the protein substrate (8). An explanation for these observations may lie in the geometric constraints imposed by exosite binding on the accessibility of the individual bonds to the active site of the enzyme. Such an explanation would suggest that geometric constraints on the substrate bound to prothrombinase by exosite interactions somehow restrict access of the Arg 274 -Thr 275 site to the active site of the enzyme until the Arg 323 -Ile 324 bond is cleaved. Rapid kinetic measurements of prothrombin cleavage suggest such potential substrate rearrangements that could follow cleavage at Arg 323 -Ile 324 and permit active site access to the Arg 274 -Thr 275 site may be related to the resultant zymogenproteinase transition and may be rate-limiting (8).
Extended interactions between enzyme and substrate are likely to be relevant to a variety of enzymic systems that act on macromolecular substrates. In such cases, the productive interaction between substrate and enzyme is likely to result from two or more binding steps that contribute differentially to the perceived affinity of the enzyme for the substrate and the inferred rate constant for catalysis. The kinetic concepts and approaches developed in this work and in prior studies with prethrombin 2 (13, 14) may prove generally useful in studies with such systems. Several other trypsin-like enzymes of coagulation act specifically to cleave their protein substrates at more than one site (2). Exosite-dependent substrate tethering may play a role in determining specificity and cleavage order in some of these reactions as well. In addition to specific proteolytic reactions such considerations may also be relevant to other reaction systems such as protein carboxylation (59), phosphorylation (60), or acetylation (61), for example, where enzymic catalysts act with defined specificity but at multiple sites on macromolecular substrates.
In summary, the results of the present study of the cleavage of mIIa⌬F1 by prothrombinase suggest that exosite binding by the substrate plays a central role in both cleavage reactions of prothrombin activation. Enzymic specificity is apparently achieved by stepwise interactions of the protein substrate with an exosite followed by an active site docking step prior to bond cleavage. The two cleavage reactions are characterized by equivalent exosite interactions and rate constants for catalysis but differ significantly in the inferred thermodynamics of the active site docking step. Thus, binding to the enzymic exosite tethers the protein substrate to prothrombinase and directs cleavage at two spatially distinct sites. These findings provide novel insights into the basis for protein substrate specificity, the reaction pathway for the conversion of prothrombin to thrombin, and the function of the prothrombinase complex.