Regions Remote from the Site of Cleavage Determine Macromolecular Substrate Recognition by the Prothrombinase Complex*

The proteolytic formation of thrombin is catalyzed by the prothrombinase complex of blood coagulation. The kinetics of prethrombin 2 cleavage was studied to delineate macromolecular substrate structures necessary for recognition at the exosite(s) of prothrombinase. The product, α-thrombin, was a linear competitive inhibitor of prethrombin 2 activation without significantly inhibiting peptidyl substrate cleavage by prothrombinase. Prethrombin 2 and α-thrombin compete for binding to the exosite without restricting access to the active site of factor Xa within prothrombinase. Inhibition by α-thrombin was not altered by saturating concentrations of low molecular weight heparin. Furthermore, proteolytic removal of the fibrinogen recognition site in α-thrombin only had a modest effect on its inhibitory properties. Both α-thrombin and prethrombin 2 were cleaved with chymotrypsin at Trp148 and separated into component domains. The C-terminal-derived ζ2 fragment retained the ability to selectively inhibit macromolecular substrate cleavage by prothrombinase, while the ζ1 fragment was without effect. As the ζ2 fragment lacks the fibrinogen recognition site, the P1-P3 residues or the intact cleavage site, specific recognition of the macromolecular substrate by the exosite in prothrombinase is achieved through substrate regions, distinct from the fibrinogen recognition or heparin-binding sites, and spatially removed from structures surrounding the scissile bond.

A major fraction of the increased catalytic efficiency observed upon assembly of factor Xa into the prothrombinase complex likely arises as a result of the influence of factor Va on the catalyst (6). However, the molecular basis for the ability of factor Va to enhance the catalytic efficiency of factor Xa within prothrombinase is poorly understood as is the basis for the narrow and distinctive macromolecular substrate specificity of factor Xa, despite its high degree of homology with trypsin (6,9). These two aspects of enzymic function appear closely related since the increased rate of prothrombin activation that results from the incorporation of factor Xa into the prothrombinase complex is not accompanied by changes in the rate of cleavage of synthetic peptidyl substrates, in the reaction with active site-directed reagents or even in the rate constant for inhibition by macromolecular inhibitors such as antithrombin III (10 -12).
Suggestions for the importance of specific macromolecular recognition, by prothrombinase, through interactions at extended recognition sites removed from the active site of factor Xa (exosites) were initially derived from studies with tick anticoagulant peptide (TAP) 1 (13). Work with a mutant derivative of TAP suggested that the selective modulation of such exosite interactions following the assembly of factor Xa into prothrombinase could lead to large changes in affinity and kinetic mechanism in the interaction of the enzyme with macromolecules (14). However, the significance of these findings toward prothrombin activation by prothrombinase has required documentation by appropriate functional studies.
The interpretation of kinetic studies of prothrombin activation is complicated by the fact that the conversion of prothrombin to thrombin involves the cleavage of two peptide bonds and is the sum of two consecutive enzyme-catalyzed reactions (6,15). However, kinetic interpretations are simplified by the use of prethrombin 2 as a substrate analog, which requires cleavage at a single site to yield thrombin (16). The kinetics of recognition and cleavage of this bond in prethrombin 2 are established to be indistinguishable from the cleavage of the same site in intact prothrombin (15).
Studies with reversible inhibitors targeting the active site of factor Xa have provided evidence for a significant contribution from exosite interactions within prothrombinase in the recognition of prethrombin 2 (17). Active site-directed reversible inhibitors as well as oligopeptidyl alternate substrates are classical noncompetitive inhibitors of macromolecular substrate cleavage by prothrombinase despite their established ability to compete for substrate binding to the active site (17). In contrast, thrombin is a competitive product inhibitor of prethrombin 2 activation but does not interfere with oligopeptidyl substrate cleavage by prothrombinase (17). These findings indicate that the affinity and binding specificity for prethrombin 2 is determined by interactions at exosites rather than by interactions between elements surrounding the scissile bond and the active site of the protease (17). Thus, competitive inhibition of macromolecular substrate cleavage by thrombin is achieved by competition for the initial exosite interaction between the substrate and prothrombinase.
It therefore follows that the structural features of prethrombin 2 that determine substrate affinity, through interactions with the exosite, are spatially distinct from residues surrounding the scissile bond. Following cleavage, the polypeptide sequence N-terminal to the scissile bond is not released, but retained in the two chain product, thrombin, through a disulfide bond (18). Since thrombin competes for prethrombin 2 binding without obscuring access to the active site of prothrombinase, it also follows that the interaction between thrombin and prothrombinase is achieved by product domains spatially distinct from the P1-P3 2 residues found in the A-chain of thrombin. We have used proteolytic derivatives of thrombin and prethrombin 2 as well as ligands established to bind to specific sites in the substrate and product, to test these predictions and delineate the regions of the substrate that contribute to binding specificity through interactions with the prothrombinase exosite.

Proteins
Prothrombin and factor X were purified from bovine plasma as described previously (23). Bovine factor X was converted to factor Xa by using the purified activator from Russell's viper venom and further purified by chromatography using benzamidine Sepharose (24,25). Kinetic titration of factor Xa with p-nitrophenyl-pЈ-guanidinobenzoate (26), typically yielded 1.12-1.15 mol of active sites/mol of factor Xa. Bovine factor Va was purified using an established procedure (27) and recombinant TAP was produced and purified as described (28). Prethrombin 2 and ␣-thrombin were prepared as described previously (16,18). Gel-filtration chromatography of prethrombin 2 on Sephadex G-50 in 20 mM Hepes, 2.5 M sodium chloride, pH 7.4 (16), resolved prethrombin 2 from a polymeric form eluting at the void volume. The aggregated species was discarded as it was found to be a potent inhibitor of prethrombin 2 cleavage by prothrombinase. Fibrin-Sepharose was prepared as described and extensively precycled prior to use (29). Buffer components of the hirugen preparation were removed by preparative reversed phase HPLC (Aquapore C-18, ABI, San Jose, CA) and elution with a gradient of increasing CH 3 CN in 0.1% (v/v) trifluoroacetic acid. Multiple peaks were identified by monitoring absorbance at 215 nm. Heterogeneity in the sample was confirmed by microbore reversed phase HPLC. However, amino acid analysis was consistent with the primary structure of hirugen and the basis for sample heterogeneity could not be determined by further analysis using either laser desorption (MALDI) or electron spray mass spectrometry. As the individual species identified by HPLC could not be adequately resolved on a preparative scale, a single pool was prepared, lyophilized, and dissolved in assay buffer. Peptide content was established by amino acid analysis.
Fragments of thrombin and prethrombin 2 were analyzed by MALDI mass spectrometry and by automated Edman degradation at the Emory University Microchemical facility. For N-terminal sequence analysis, fragments were separated by SDS-PAGE following disulfide bond reduction with dithiothreitol and transferred to Immobilon PSQ membranes (Millipore, Bedford, MA) by electroblotting in a semidry apparatus (Hoeffer, San Francisco, CA) as described previously (30). Protein sequence was determined from excised membrane fragments.

Preparation of Thrombin Derivatives
Inactivation of ␣-thrombin with APMSF to yield ␣-IIa i was performed as described previously (17).
For the preparation of bovine ␥ T -thrombin, ␣-thrombin in assay buffer (22 M, 45 ml) was incubated with 0.7 M trypsin for 3 h at room temperature. The reaction was quenched with 10 M soybean trypsin inhibitor, dialyzed against 20 mM Tris-PO 4 , pH 5.8, 40 mM NaCl, 0.1% (w/v) PEG for 4 h at 4°C and applied to a column (1.5 ϫ 12.5 cm) of S-Sepharose equilibrated in the same buffer. Bound protein was eluted (4 ml/min, 120 min) with a linear gradient of increasing NaCl (40 -700 mM) in 20 mM Tris-PO 4 , 0.1% (w/v) PEG, pH 5.8. All fractions containing protein exhibited equivalent specific activities to ␣-thrombin in the cleavage of S2238. Fractions from the leading peak, containing ␥ Tthrombin with an estimated contamination of 5% undigested material determined by SDS-PAGE were pooled, dialyzed against 20 mM Hepes, 40 mM NaCl, 0.1% (w/v) PEG, pH 7.4, and subject to affinity chromatography using a 4.5 ϫ 17-cm column of fibrin-Sepharose equilibrated in the same buffer. Bound protein was eluted with a linear gradient of increasing NaCl (40 -600 mM) in 20 mM Hepes, 0.1% (w/v) PEG, pH 7.4. The ␥ T -thrombin containing fractions were pooled and reapplied to a second fibrin-Sepharose column with isocratic elution to remove traces of remaining ␣-thrombin. The flow-through fractions were characterized by the same specific activity as ␣-thrombin toward S2238 with ϳ1% of the specific activity of ␣-thrombin in a fibrinogen clotting assay (18). Pooled material was concentrated by ultrafiltration in a stirred cell (YM10, Amicon, Danvers, MA) to a concentration of ϳ100 M, inactivated by the addition of 1 mM APMSF followed by brief incubation at room temperature and dialyzed against assay buffer. The resulting preparation of inactivated ␥ T -thrombin (␥ T -IIa i ) possessed Ͻ0.01% catalytic activity when compared with ␣-thrombin. Protein sequencing of the fragments resolved by SDS-PAGE yielded the expected sequence for bovine ␥ T -thrombin (3)(4)(5)31). In addition, the mass of ␥ T -thrombin was determined by MALDI mass spectrometry and found to be consistent with removal of the undecapeptide (Ile 68 -Arg 77A ). 3 Bovine -thrombin was prepared by treatment of ␣-thrombin (27 M, 2 Nomenclature of Schechter and Berger (1). 3 The residues of ␣-thrombin have been numbered after the corresponding amino acids in chymotrypsin using the convention of Bode et al. (2). Cleavage of human ␣-thrombin by trypsin at Arg 67 , Arg 77A , and Lys 149E leads to the loss of the Ile 68 -Arg 77A peptide and yields ␥ T -15 ml) in 0.25 M sodium phosphate buffer, pH 6.5, with 4.2 nM chymotrypsin for 4 h at room temperature. The digest, terminated by addition of 10 M TPCK, was dialyzed against 20 mM MES, 80 mM NaCl, pH 6.5, and applied to a column (1.5 ϫ 8 cm) of S-Sepharose equilibrated in the same buffer. Elution with a linear gradient (4 ml/min, 130 min) of increasing NaCl (80 -800 mM) in 20 mM MES, pH 6.5, resulted in the separation of chymotrypsin, undigested thrombin, and two peaks of cleaved material. Protein in the two resolved peaks of cleaved material appeared identical by SDS-PAGE, protein sequencing, and specific activity measured with S2238. The cleaved material was pooled, concentrated, inactivated with AMPSF, and dialyzed into assay buffer as described above. Protein sequence analysis indicated a single cleavage at Trp 148 consistent with the expected result for the action of chymotrypsin on human thrombin to yield -thrombin (32). On this basis, the resulting inactivated preparation was designated -IIa i .
The two chymotryptic fragments of thrombin were isolated by HPLC separation of the products generated following chymotryptic cleavage of ␣-thrombin. Thrombin (30 mg) treated with chymotrypsin (as above) was dialyzed against 20 mM Tris, 30% (v/v) CH 3 CN, pH 9.0, for 4 h at room temperature. Following clarification by centrifugation (50,000 ϫ g, 20 min), aliquots (3 ml) were fractionated by cation exchange HPLC (Aquapore Cation 7 m, 0.46 ϫ 22.2 cm, ABI). Elution (1 ml/min, 30 min) with a linear gradient of increasing NaCl (0 -350 mM) in 20 mM Tris, 30% (v/v) CH 3 CN, pH 9.0, resolved two peaks corresponding to the two thrombin fragments. Material from each of the peaks accumulated from successive runs was pooled, dialyzed against 0.1% (v/v) trifluoroacetic acid, concentrated by lyophilization, and further purified by reversed phase HPLC. Each of the pools was dialyzed against Buffer A (20 mM NEt 3 -PO 4 , pH 2.5) and fractionated in ϳ1-mg aliquots using an Aquapore Phenyl column (0.46 ϫ 22.2 cm, ABI). Bound protein was eluted (1 ml/min) with a biphasic gradient of increasing Buffer B (20 mM NEt 3 -PO 4 , 80% (v/v) CH 3 CN, pH 2.5) of 0 -24% Buffer B in 25 min followed by 30 -37% Buffer B in 80 min. Analysis by SDS-PAGE confirmed quantitative separation of the two fragments which were identified as 1-thrombin and 2-thrombin based on N-terminal sequence analysis and mass spectrometry. Pools were prepared for each of fragments, lyophilized, dialyzed against assay buffer, and clarified by centrifugation (50,000 ϫ g, 20 min). Dialysis resulted in variable amounts of protein loss due to precipitation. The final concentrations were in the range of 15 M for 1-thrombin and 40 M for 2-thrombin.

Preparation of Prethrombin 2 Derivatives
The two chymotryptic fragments of prethrombin 2 were prepared by treating prethrombin 2 (27 M, 5 ml) in 0.25 M sodium phosphate, pH 6.5, with 1 nM chymotrypsin for 3 h at room temperature. The reaction mixture was dialyzed against Buffer A (above) and fragments were separated by reversed phase HPLC as described above for the thrombin fragments. The resulting peptides were lyophilized, dialyzed against assay buffer, and clarified by centrifugation (50,000 ϫ g, 20 min). The fragments were designated 1-prethrombin 2 and 2-prethrombin 2 on the basis of SDS-PAGE, N-terminal sequence analysis, and mass spectrometry.

Preparation of a Fluorescent Derivative of Factor Xa
Factor Xa with a fluorescent probe tethered to the active site via a peptidyl chloromethyl ketone was prepared using the procedure developed by Bock (39). In the first step, the acetothioester derivative of EGR-CH 2 Cl was prepared by minor modifications to described procedures (39). Purification of the acetothioacetyl adduct of EGR-CH 2 Cl (ATA-EGR-CH 2 Cl) was performed by application to a column (1.5 ϫ 120 cm) of S-Sepharose equilibrated in 50 mM sodium phosphate, pH 2.5, with isocratic elution. The resulting ATA-EGR-CH 2 Cl was identified, concentrated, and characterized as described (39). Factor Xa in 0.5 M Tris, 0.1% (w/v) PEG, pH 7.5, was inactivated by sequential additions of ATA-EGR-CH 2 Cl and the resulting modified factor Xa (ATA-EGR-Xa) was separated from excess unreacted inhibitor by chromatography on a G-25 column (40). Traces of uninhibited factor Xa were depleted by affinity chromatography using soybean trypsin inhibitor Sepharose (41). In the final step, a fluorophore was incorporated into the active site by reacting ATA-EGR-Xa (60 M, 3 mg) with 680 M Oregon Green 488 iodoacetamide in 0.1 M Hepes, 0.15 M NaCl, 87 mM NH 2 OH, pH 7.0, for 75 min at room temperature. Unreacted dye and buffer components were separated from the protein by gel filtration as described previously (40), to yield the fluorescent adduct of factor Xa ([OG 488 ]-EGR-Xa). The concentration of the fluorescent adduct was determined with a colorimetric protein assay (BCA assay, Pierce, Rockford, IL) using a standard curve prepared with known concentrations of ATA-EGR-Xa (40).

Discontinuous Measurements of the Initial Rate of Thrombin Formation
The initial rate of thrombin formation by the action of prothrombinase on prethrombin 2 was determined as described previously (16). Reaction mixtures (290 l) were prepared in assay buffer containing 25 nM factor Va, 60 M PCPS, the indicated concentrations of prethrombin 2 and inhibitor at 25°C. Thrombin formation was initiated by the addition of 5 nM factor Xa (10 l, 150 nM). Aliquots (10 l) were withdrawn either before initiation (t ϭ 0 datum) or serially at t ϭ 0.5, 1, 1.5, 2, and 3 min following the addition of factor Xa, and mixed with 90 l of 20 mM Hepes, 0.15 M NaCl, 0.1% (w/v) PEG, 50 mM EDTA, 2 M TAP, pH 7.4, to quench thrombin formation. Aliquots (10 l) of each quenched sample were diluted further in the same buffer but lacking TAP in wells of a 96-well plate and the initial velocity of S2238 hydrolysis was determined by continuously monitoring the change in absorbance at 405 nm in a kinetic plate reader (Molecular Devices, Sunnyvale, CA) following the addition of 200 M S2238 in 20 mM Hepes, 0.15 M NaCl, 0.1% (w/v) PEG, 50 mM EDTA, pH 7.4. The concentration of thrombin formed as a function of time was determined by interpolation from the linear dependence of initial velocity of S2238 hydrolysis on known concentrations of thrombin performed as a control with each experiment. The initial, steady state rate of thrombin formation in the initial reaction mixture was determined from the slope of the linear appearance of thrombin as a function of time. Control experiments established that inhibitors added to the initial stage of the assay were sufficiently dilute to have no noticeable effect on the activity of thrombin toward S2238. When inhibition studies were performed using fragments purified by HPLC and dialysis, control experiments using appropriate volumes of the dialysate ruled out trace solvent contamination as a trivial explanation for the results. For kinetic measurements in the presence of hirugen, the final concentration of factor Xa was increased to 10 nM and aliquots were quenched at 0, 1, 2, 3, 4, and 6 min following initiation.
The concentrations of factor Va and PCPS were chosen to be saturating relative to the concentration of factor Xa and to the measured equilibrium dissociation constants for the binary interactions that lead to the assembly of the prothrombinase complex (42). It is therefore valid to normalize reaction rates by dividing by the concentration of factor Xa which determines the concentration of prothrombinase. The validity of this assumption was established in separate control experiments documenting the linear dependence of reaction rate on the concentration of factor Xa under these conditions.

Steady State Measurements of Fluorescence Anisotropy
Fluorescence anisotropy was measured in T-format using a SLM8000C fluorescence spectrophotometer (SLM Instruments, Urbana, IL). Anisotropy was measured in 1 ϫ 1-cm 2 quartz cuvettes using ex ϭ 490 nm and em ϭ 520 nm with long pass filters (Schott KV-520) in the emission beam. Reaction mixtures (2 ml) in assay buffer containing the indicated concentration of [OG 488 ]-EGR-Xa, 50 M PCPS and the indicated concentration of ␣-IIa i were titrated with microliter additions of a stock solution of factor Va prepared in assay buffer. Following each addition, the reaction mixture was gently mixed and anisotropy was measured at 25°C by manually rotating the excitation polarizer, integrating the signal at each of the two positions for 10 s and averaging 5-8 successive readings. Anisotropy was calculated (43), following the subtraction of a scattering blank from a reaction mixture containing identical reactant concentrations except that factor Xa previously inactivated with APMSF (Xa i ) was used instead of [OG 488 ]-thrombin (3)(4)(5). Tryptic digestion of bovine thrombin results in cleavage at Arg 67 and Arg 77A but not at Glu 149E . We have designated this derivative of bovine thrombin, which lacks a significant part of the fibrinogen recognition site, as ␥ T -thrombin. EGR-Xa. When necessary, ⌬r at each concentration of titrant was calculated by subtracting the anisotropy observed in the absence of factor Va. Displacement experiments were performed by preforming prothrombinase in assay buffer using 28.8 nM [OG 488 ]-EGR-Xa, 50 M PCPS and either 20 or 60 nM factor Va. Anisotropy was measured following titration with increasing concentrations of Xa i . Following corrections for scattering, ⌬r at each addition of titrant was calculated by subtraction of the anisotropy from that of parallel reaction mixtures lacking factor Va.

Data Analysis
Data were analyzed according to the indicated equations by nonlinear least squares regression analysis using the Marquardt algorithm (44). Alternative models were eliminated on the basis of poorer fits on the basis of criteria described previously (45). The fitted constants are presented Ϯ 95% confidence limits and one of at least two similar experiments performed using different protein preparations is presented in each case.
Inhibition of Prethrombin 2 Activation by Thrombin and Derivatives-Initial velocity measurements of prethrombin 2 activation by prothrombinase using increasing concentrations of substrate at different fixed concentrations of thrombin derivatives or determined at one fixed concentration of prethrombin 2 and increasing concentrations of inhibitor were analyzed according to the rate expression for linear competitive inhibition (46), to yield fitted values for K m , V max , and K i .
Inhibition of Prethrombin 2 Activation by Hirugen-Initial rates of thrombin formation by prothrombinase at increasing concentrations of prethrombin 2 in the presence of different fixed concentrations of hirugen were adequately described using Equation 1 (46) which assumes that 1 mol of hirugen (I) reversibly binds per mole of prethrombin 2 (S) and the resulting SI complex cannot bind to prothrombinase, where I and S refer to the total concentrations of substrate and inhibitor, K dS,I is the equilibrium dissociation constant for the interaction of I with S and V max and K m represent the Michaelis constants for the cleavage of free (unliganded) S. Analysis according to Equation 1 yielded fitted values for K dS,I , V max , and K m .
Fluorescence Measurements of the Factor Xa and Va Interaction-Fluorescence titrations of prothrombinase assembly obtained at one or more fixed concentrations of [OG 488 ]-EGR-Xa, a fixed and saturating concentration of PCPS with varying concentrations of factor Va were analyzed using the model and experimental considerations previously developed for the assembly of prothrombinase (42). In this case, titration curves describe the equilibrium dissociation constant for the interaction between [OG 488 ]-EGR-Xa and factor Va on the membrane surface (42). The dependence of fluorescence anisotropy on the concentration of factor Va was analyzed according to Equation 2,

Kinetics of Inhibition of Prothrombinase by ␣-Thrombin-
Previous studies have shown that ␣-thrombin acts as an exosite-directed product inhibitor of macromolecular substrate cleavage by prothrombinase (17). Thrombin, inactivated with APMSF (␣-IIa i ) was tested for its ability to inhibit the cleavage of either a tripeptidyl substrate (SpXa) or the macromolecular substrate analog, prethrombin 2, by prothrombinase (Fig. 1). Even though either substrate was present at approximately the same multiple of K m (ϳ0.7 ϫ K m ), increasing concentrations of ␣-IIa i yielded significant inhibition of prethrombin 2 activation (K i ϭ 2.02 Ϯ 0.11 M) with a minor effect (K i Ն 80 M) on the initial rate of SpXa hydrolysis (Fig. 1). Initial velocity studies using increasing concentrations of prethrombin 2 at different fixed concentrations of ␣-IIa i (not shown) yielded linear competitive inhibition with K i ϭ 2.15 Ϯ 0.3 M (Table I). Thus, ␣-IIa i and prethrombin 2 bind in a mutually exclusive fashion to prothrombinase. Because ␣-IIa i has a minor effect on peptidyl substrate hydrolysis by prothrombinase, competitive inhibition of prethrombin 2 cleavage is achieved without restricting access to the active site of factor Xa within the prothrombinase complex. Such observations, in part, form the basis for the previous suggestion that the affinity of the enzyme complex for macromolecular substrates such as prethrombin 2 is determined by binding interactions at exosites and not the active site of factor Xa within prothrombinase (17).
Influence of Thrombin on the Assembly of the Prothrombinase Complex-Factor Xa assembled into the ternary prothrombinase complex with saturating concentrations of factor Va and phospholipid membranes catalyzes prethrombin 2 cleavage with greatly increased catalytic efficiency compared with factor Xa in solution or saturated with membranes (16). Since the three enzyme species only exhibit minor differences in the kinetics of hydrolysis of SpXa and other peptidyl substrates (11), inhibition of the assembly of prothrombinase by ␣-IIa i could provide a trivial explanation for the selective inhibition of macromolecular substrate cleavage.
Equilibrium and kinetic measurements of the assembly of prothrombinase have previously relied on the use of a dansyl reporter group covalently tethered to the active site of factor Xa via a peptidyl chloromethyl ketone (48). Reliable assessment of the equilibrium constant for the Xa-Va interaction on the membrane surface requires nanomolar reactant concentrations and measurements of fluorescence intensity of the dansyl moiety with excitation at 280 nm (42). However, significant inner filter effects limit the use of this approach to assess the effect of ␣-IIa i on the assembly of prothrombinase. A derivative of factor Xa containing Oregon Green 488 tethered to the active site ([OG 488 ]-EGR-Xa) displayed a significant change in fluorescence anisotropy upon assembly into the ternary prothrombinase complex with appropriate spectral properties and sensitivity suitable for use at reactant concentrations comparable to those used in the kinetic studies.
Addition of PCPS at sufficient concentrations to saturate all the [OG 488 ]-EGR-Xa present (49) produced no change in anisotropy (not shown). Titration of reaction mixtures containing a fixed concentration of [OG 488 ]-EGR-Xa and saturating PCPS with increasing concentrations of factor Va yielded a saturable increase in fluorescence anisotropy ( Fig. 2A). The Va-dependent increase in anisotropy was not observed when excess EDTA was present to chelate the Ca 2ϩ (Fig. 2A). These observations are consistent with the requirements for the high affinity interaction between factors Xa and Va established in previous work (42,48). Because of the use of saturating concentrations of PCPS, such titration curves yield the equilibrium dissociation constant for the binding of membrane-bound [OG 488 ]-EGR-Xa to membrane bound factor Va (42). Analysis of titration curves obtained at two fixed concentrations of [OG 488 ]-EGR-Xa ( Fig. 2A) yielded K d ϭ 0.89 Ϯ 0.08 nM with a stoichiometry of 0.94 Ϯ 0.01 mol of Va bound per mol of [OG 488 ]-EGR-Xa at saturation. These values are indistinguishable from those previously determined using equivalent assumptions either by functional measurements or by fluorescence binding studies (42). Titration of reaction mixtures containing [OG 488 ]-EGR-Xa, PCPS, and factor Va (prothrombinase) with increasing concentrations of inactivated but nonfluorescent factor Xa derivative (Xa i ) decreased the anisotropy to near-baseline levels ( Fig. 2A, inset). Analysis of these displacement curves yielded comparable equilibrium dissociation constants for the binding of [OG 488 ]-EGR-Xa (K d ϭ 0.86 Ϯ 0.28 nM) or Xa i (K dcomp ϭ 1.74 Ϯ 0.52 nM) to factor Va on the membrane surface. Thus, the interaction between [OG 488 ]-EGR-Xa and factor Va on membranes is reversible. The near equivalence of the two K d values indicates that neither the fluorophore nor the tripeptidyl tether has an obvious effect on the ability of [OG 488 ]-EGR-Xa to assemble into prothrombinase. These data establish the validity of the use of [OG 488 ]-EGR-Xa as an appropriate probe for measurements of the assembly of the prothrombinase complex.
Fluorescence titrations using a single fixed concentration of [OG 488 ]-EGR-Xa, saturating PCPS, and increasing concentrations of factor Va were compared in the presence or absence of 12 M ␣-IIa i (Fig. 2B). The data were normalized using the anisotropy in the absence of factor Va because the presence of ␣-IIa i led to a small but reproducible increase (0.004) in the anisotropy of [OG 488 ]-EGR-Xa either in the presence or absence of PCPS (not shown). The two titration curves were indistinguishable from each other and could be adequately described by equilibrium parameters comparable to those established above. Thus, ␣-IIa i has no detectable effect on the assembly of prothrombinase at concentrations which yield substantial inhibition of prethrombin 2 cleavage. These findings exclude a potentially trivial explanation for the ability of ␣-IIa i to selectively inhibit macromolecular substrate cleavage by prothrombinase.
Dissection of Interaction Sites of Thrombin and Prethrombin 2 with Prothrombinase-A schematic representation of prethrombin 2, ␣-thrombin, and known proteolytic derivatives is provided in Fig. 3A. The corresponding products isolated on a preparative scale were analyzed by SDS-PAGE (Fig. 3B). Activation of the single chain zymogen, prethrombin 2, converts it to the two chain ␣-thrombin, which retains the P1-P3 residues (Fig. 3A). Functional and structural studies have established the presence of sites in ␣-thrombin, removed from the P1-P3 residues, that play a role in the diverse macromolecular interactions of the protease (50). The fibrinogen recognition site and the heparin-binding site in ␣-thrombin (Fig. 3A), are at least partially expressed in prethrombin 2 as well (51).
Limited proteolysis of ␣-thrombin by trypsin in the fibrinogen recognition site releases a peptide comprising residues Ile 68 -Arg 77A (3)(4)(5). The resulting ␥ T -thrombin (Fig. 3A) is unable to cleave fibrinogen, shows dramatically reduced affinity for hirudin but retains full catalytic activity toward oligopeptidyl substrates (3,5). ␥ T -Thrombin, with these expected properties was isolated (Fig. 3B) and inactivated with APMSF to yield ␥ T -IIa i . Product inhibition studies of prethrombin 2 activation by prothrombinase using ␥ T -IIa i yielded linear competitive inhibition (Table I). In contrast to the large reduction in fibrinogen cleavage following proteolytic removal of the fibrinogen recognition site, the K i for the inhibition of prothrombinase by ␥ T -IIa i was only modestly increased in comparison to ␣-IIa i (Table I). However, some variability was noted in the inhibitory properties of different preparations of ␥ T -IIa i , illustrated by the representative K i values in Table I. These effects are likely related to difficulties associated with the reproducibility of partial protease digests, stability of the proteolysed product, and variable trace contamination (Ͻ Ͻ5%) with undigested material. These points lead to the qualified conclusion that tryptic removal of the fibrinogen recognition site may have a small effect on but does not eliminate the ability of thrombin to act as an exosite-directed product inhibitor of macromolecular substrate cleavage by prothrombinase.
Cleavage of thrombin in the autolysis loop at Trp 148 has also been shown to modulate the interaction of thrombin with macromolecules such as fibrinogen and antithrombin III (5). Thrombin was digested with chymotrypsin under conditions comparable to those previously established (52) to yield -thrombin (Fig. 3A). Purification of the derivative yielded two fragments evident by SDS-PAGE of the expected size (Fig. 3B)  Fig. 3.
b Initial steady state velocities were determined using increasing concentrations of prethrombin 2 (8 -14 values) at different fixed concentrations of the indicated inhibitor (3-5 values). Kinetic constants were determined by analysis according to the rate expression for linear competitive inhibition. Kinetic constants are presented Ϯ 95% confidence limits. c V max divided by the total concentration of prothrombinase. The resulting value is not equivalent to the first order rate constant for catalysis (17).
d Two sets of kinetic constants are presented to illustrate the extent of variability in the data with this derivative. The lower K i value is representative of results with three different preparations of ␥ T -IIa i and the larger K i was obtained with one preparation. and cleavage at Trp 148 was confirmed by N-terminal sequencing and mass spectrometry. The resulting -thrombin, inactivated with APMSF to yield -IIa i , retained the ability to act as a competitive inhibitor of prethrombin 2 hydrolysis with a K i that was slightly lower than that of ␣-IIa i (Table I). Therefore, the binding of macromolecular product to the exosite on prothrombinase is unaffected by prior cleavage in the autolysis loop.
Potential contributions from the heparin-binding site were evaluated by examining the effects of saturating concentrations of LMW heparin on the ability of ␣-IIa i to selectively inhibit prethrombin 2 activation by prothrombinase. Since both thrombin and prethrombin 2 are known to bind heparin with high affinity, LMW heparin concentrations were chosen to be high relative to the published K d for binding thrombin as well as to provide sufficient binding sites for both prethrombin 2 and thrombin at the highest concentrations used (53). In preliminary experiments, LMW heparin was found to substantially enhance prethrombin 2 activation by increasing both the K m and V max (Table I). While the basis for these effects was not extensively investigated, they appear related to the ability of prethrombin 2 to bind heparin since LMW heparin had no effect on the kinetics of SpXa cleavage by prothrombinase (not shown). However, the presence of saturating concentrations of LMW heparin had no detectable effect on the kinetics of inhibition by ␣-IIa i (Table I). Comparable experiments with unfractionated heparin yielded equivalent results except that initial velocity measurements at high substrate concentrations were compromised by protein precipitation possibly due to interactions between thrombin and long chain heparin at multiple sites (53). Since occupation of the heparin-binding site(s) in ␣-IIa i does not affect its ability to act as a product inhibitor, it is unlikely that this macromolecular interaction site in the product is either directly or indirectly involved in mediating recognition by the exosite in prothrombinase.
Kinetics of Exosite-dependent Product Inhibition in the Presence of Hirugen-Hirugen binds to the fibrinogen recognition site and thereby competitively inhibits interactions between macromolecules and ␣-thrombin at this site (22). Kinetic studies with hirugen were used to further test the conclusions derived from inhibition studies with ␥ T -IIa i and -IIa i .
In initial experiments, hirugen was found to directly inhibit prethrombin 2 cleavage by prothrombinase. Increasing concentrations of hirugen were found to increase the K m for prethrombin 2 activation without affecting the V max (Fig. 4). Prethrombin 2 has been documented to bind hirugen (54). Thus, inhibition could arise from an interaction between hirugen and prethrombin 2 which precludes substrate binding to prothrombinase. The data could be adequately described by the rate expression for this type of inhibition (Equation 1, Fig. 4). However, the data could also be described equally well by rate expressions for alternate mechanisms, including partial inhibition resulting from a 5-7-fold decrease in substrate affinity following hirugen binding to prethrombin 2 or even by competitive inhibition resulting from an interaction between hirugen and prothrombinase.
Initial velocity studies illustrating the inhibition of prethrombin 2 activation by 30 M ␣-IIa i in the absence or presence of 241 M hirugen are provided in Fig. 5. The presence of ␣-IIa i at ϳ10 ϫ K i yielded the expected and substantial inhibition of prethrombin 2 activation in the absence of hirugen. Inhibition by ␣-IIa i was alleviated in the presence of saturating concentrations of hirugen. These data suggest that the binding of hirugen to ␣-IIa i attenuates or eliminates the ability of the product to bind to prothrombinase and inhibit prethrombin 2 activation. However, the direct inhibitory effect of hirugen on prethrombin 2 activation by prothrombinase suggests a complex mechanism of action and precludes definitive interpretation of these observations.
Localization of a Domain That Imparts Binding Specificity for Prethrombin 2-Cleavage of thrombin by chymotrypsin at Trp 148 (-thrombin, Fig. 3A) yields two peptides that remain noncovalently associated (32,52). The cleavage site separates thrombin into approximate hemispheres, each bearing elements of the catalytic triad, that can be reassociated following dissociation and separation to reconstitute enzymatic activity (52). The N-terminal domain (1-thrombin) bears the P1-P3 sites of the product, the intact fibrinogen recognition site as well as some residues involved in heparin binding. The Cterminal domain (2-thrombin) bears elements of the heparinbinding site (Fig. 3A). By analogy, cleavage at the same site in prethrombin 2 yields similar species denoted as 1and 2prethrombin 2 (Fig. 3A). While the 2 fragments from prethrombin 2 and thrombin are chemically identical, 1-prethrombin 2 retains the intact scissile bond acted upon by prothrombinase.
The resulting fragments from -thrombin and -prethrombin 2 were dissociated, preparatively purified by HPLC (Fig. 3B), and identified by mass spectrometry and N-terminal sequencing. As described previously (52), thrombin activity could be near-quantitatively recovered by mixing stoichiometric concentrations of the 1and 2-thrombin fragments in assay buffer (not shown).
Purified 1 and 2 fragments prepared from thrombin and prethrombin 2 were separately tested for their ability to inhibit SpXa cleavage and prethrombin 2 activation by prothrombinase (Fig. 6). Results from several experiments using different fragment preparations indicated that only the 2 fragment, from either thrombin or prethrombin 2, retained the ability to inhibit prethrombin 2 activation by prothrombinase. The K i for the inhibition of prethrombin 2 activation by the 2 fragment was equivalent to that observed with undissociated -IIa i (Fig.  6, Table I). In parallel experiments, the 2 fragment had a relatively small effect on the rate of hydrolysis of SpXa by prothrombinase (Fig. 6). In contrast, the 1 fragment derived from either prethrombin 2 or thrombin had no detectable effect (K i Ͼ Ͼ 23 M) on prethrombin 2 activation by prothrombinase (Fig. 6, inset) or on the rate of SpXa cleavage by the enzyme complex (not shown).
The data indicate that the inhibitory properties of ␣-IIa i or -IIa i can be completely reproduced with 2-thrombin or with 2-prethrombin 2. Since the fibrinogen recognition site is not present in the 2 fragment (Fig. 3A) and heparin has no effect on the ability of ␣-IIa i to inhibit prethrombin 2 activation (Table I), the data suggest that binding to the exosite in prothrombinase is mediated by substrate/product regions that are distinct from the fibrinogen recognition site or the heparinbinding site. The lack of significant inhibition by 1-thrombin and 1-prethrombin 2 despite the presence of an intact cleavage site in the latter fragment supports the conclusion that the domain of the substrate and product that mediates recognition at the exosite of prothrombinase and determines binding specificity is spatially removed from sites immediately surrounding the scissile bond in the substrate. DISCUSSION Productive recognition of prethrombin 2 by prothrombinase proceeds through interactions at exosites on the enzyme complex followed by binding of substrate structures surrounding the scissile bond to the active site of factor Xa prior to cleavage and product release as illustrated in Scheme I (17). Substrate affinity is determined by the bimolecular interaction between the substrate and enzymic exosite(s) rather than by binding interactions at the active site which are unfavorable and instead contribute to maximum catalytic rate (17). One hallmark for this type of mechanism is that thrombin acts as a linear competitive inhibitor of macromolecular substrate cleavage without obscuring access of small molecule inhibitors and oligopeptidyl substrates to the active site of factor Xa within prothrombinase (17). The present results bear out predictions that necessarily follow from these conclusions and indicate that binding to the exosite(s) in prothrombinase is mediated by a domain of prethrombin 2 and thrombin distinct from the residues immediately surrounding the cleavage site. As illustrated in Scheme I, binding specificity is conferred by structures present in the C-terminal region of prethrombin 2 or thrombin, physically separable from the domain containing the P1-P3 sites or the intact scissile bond and distinct from the fibrinogen recognition or heparin-binding sites.
The results obtained with ␥ T -IIa i , -IIa i , and the purified 2 fragment support the conclusion that the fibrinogen recognition site is not directly involved in the interaction of prethrombin 2 or thrombin with the prothrombinase exosite. The strong- Prothrombinase (E) catalyzes the conversion of prethrombin 2 (ƒ) to ␣-thrombin (OE) by a multistep mechanism (17). The binding of S to E results from interactions at an exosite in E, followed by interactions at the active site prior to bond cleavage and product release. Ks and Ks * reflect the equilibrium dissociation constants for the stepwise interaction of S with E, k cat refers to the rate constant for the catalytic step, and K p is the equilibrium dissociation constant for P binding to E. The relevant sites in E are the exosite (hatched bar) and the active site (cleft). Binding to the exosite is determined by a region in the Cterminal half of either S or P, distinct from the P1-P3 sites (f), the fibrinogen recognition site (q), and the heparin-binding site (ࡗ). est evidence to suggest this derives from the ability of the 2 fragment, which lacks the fibrinogen recognition site, to reproduce the inhibitory properties of ␣-IIa i . While ␥ T -IIa i does display reduced affinity for prothrombinase, the changes in K i are modest when compared with the large reduction in fibrinogen cleavage or hirudin binding which accompanies proteolysis in the fibrinogen recognition site (3)(4)(5). However, a role for the fibrinogen recognition site in exosite binding is suggested by the ability of hirugen to directly inhibit prethrombin 2 cleavage by increasing the K m and by saturating concentrations of hirugen to reduce the ability of ␣-IIa i to act as a product inhibitor. Given the observations with the thrombin derivatives, it seems unlikely that the fibrinogen recognition site is directly involved in the binding of the macromolecular substrate/product to the exosite(s) in prothrombinase. Instead, the data are more consistent with the possibility that hirugen binding to thrombin or prethrombin 2 elicits changes at sites distinct from the fibrinogen recognition site which in turn modulate binding to prothrombinase. Extensive allosteric linkage has been documented between the fibrinogen recognition site and the binding of ligands such as heparin, antithrombin III, fragment 2, and Na ϩ to other sites in thrombin (55)(56)(57). Recent binding measurements imply that hirugen and a fragment 2 peptide behave as competitive ligands even though they are known to bind to distinct sites on thrombin (56). It is also possible that the effects observed at high concentrations of hirugen in the present study are related to secondary interactions of the highly charged peptide. We are unable to distinguish between these possibilities.
Structure-function studies of substrate specificity, of the type described, are susceptible to interpretation problems arising from unanticipated effects of protein fragments on the stability of the prothrombinase complex or from inhibition derived from the ability of protein fragments containing Arg-X residues to act as alternate substrates. In the present study, direct equilibrium binding measurements of prothrombinase assembly, at reactant concentrations approaching those used in the kinetic measurements, were used to exclude inhibitory effects arising from the destabilization of the enzyme complex. In addition, all inhibitory proteolytic derivatives acted as selective inhibitors of prethrombin 2 activation with a minor effect on synthetic peptidyl substrate hydrolysis by prothrombinase. This fact rules out significant alternate substrate effects as a trivial explanation for the present findings.
The ability of thrombin to act as a linear competitive inhibitor of prethrombin 2 cleavage catalyzed by prothrombinase implies that prethrombin 2 and thrombin bind to the enzyme complex in a mutually exclusive fashion. The similarities in the known structures of prethrombin 2 and thrombin (51) and the substrate-product relationship between the two suggests that they are likely to compete for interactions at the same site in the enzyme complex as illustrated in Scheme I. While, alternate, more complicated explanations cannot be adequately excluded at present, this interpretation is supported by the fact that the 2 fragment from either species exhibits equivalent inhibitory properties toward prethrombin 2 activation.
The fragment 1 domain mediates the binding of prothrombin to membranes (58) and the fragment 2 region has been shown to be responsible for the interaction between the substrate and factor Va (59). Since prothrombin binding to factor Va and membranes does not involve the active site of factor Xa within prothrombinase, they represent potential exosite interactions. However, these domains are absent in purified prethrombin 2 and thrombin. Thus, it is possible that the exosite interactions inferred in this and previous work (17) involves specific recognition of the macromolecular substrate by sites in factor Xa itself that are removed from the active site. This possibility is supported by prior work documenting the inhibition of macromolecular but not oligopeptidyl substrate cleavage by prothrombinase with a monoclonal antibody specific for factor Xa (60). However, it remains possible that previously unidentified interactions between prethrombin 2 or thrombin and other sites in the prothrombinase complex also contribute to substrate/product binding.
Explanations for the narrow and distinctive substrate specificity of factor Xa or prothrombinase have, thus far, been sought from the active site geometry based on the x-ray structure of factor Xa, by mutagenesis studies of residues surrounding the active site and by studies with synthetic peptidyl substrates (9,61,62). The results of this and previous studies (17) suggest that while such approaches may describe the properties of factor Xa in solution, they are inadequate for assessing the basis for the substrate specificity of the prothrombinase complex. The affinity of prothrombinase for the macromolecular substrate is not determined by binding of the substrate to the active site of factor Xa but rather by interactions between extended macromolecular recognition sites, distinct from the active site, and substrate regions spatially distinct from structures immediately surrounding the scissile bond.