Inactivation of Thrombin by Antithrombin Is Accompanied by Inactivation of Regulatory Exosite I*

Exosite I of the blood clotting proteinase, thrombin, mediates interactions of the enzyme with certain inhibitors, physiological substrates and regulatory proteins. Specific binding of a fluorescein-labeled derivative of the COOH-terminal dodecapeptide of hirudin ([5F] Hir54–65) to exosite I was used to probe changes in the function of the regulatory site accompanying inactivation of thrombin by its physiological serpin inhibitor, antithrombin. Fluorescence-monitored equilibrium binding studies showed that [5F]Hir54–65 and Hir54–65 bound to human α-thrombin with dissociation constants of 26 ± 2 nm and 38 ± 5 nm, respectively, while the affinity of the peptides for the stable thrombin-antithrombin complex was undetectable (≥200-fold weaker). Kinetic studies showed that the loss of binding sites for [5F]Hir54–65 occurred with the same time-course as the loss of thrombin catalytic activity. Binding of [5F] Hir54–65 and Hir54–65 to thrombin was correlated quantitatively with partial inhibition of the rate of the thrombin-antithrombin reaction, maximally decreasing the bimolecular rate constants 1.7- and 2.1-fold, respectively. These results support a mechanism in which thrombin and the thrombin-Hir54–65 complex can associate with antithrombin and undergo formation of the covalent thrombin-antithrombin complex at modestly different rates, with inactivation of exosite I leading to dissociation of the peptide occurring subsequent to the rate-limiting inactivation of thrombin. This mechanism may function physiologically in localizing the activity of thrombin by allowing inactivation of thrombin that is bound in exosite I-mediated complexes with regulatory proteins, such as thrombomodulin and fibrin, without prior dissociation of these complexes. Concomitant with inactivation of thrombin, the thrombin-antithrombin complex may be irreversibly released due to exosite I inactivation.

The information available indicates that the catalytic site of thrombin is accessible to antithrombin when the proteinase is bound in exosite I complexes with hirudin peptide, fibrin I and II, or thrombomodulin, and the reaction rate is not greatly affected (7,19,20,24,26,28). On the basis of the mechanism of the thrombin-antithrombin reaction (see Refs. 29 and 30 for reviews), this implies that thrombin can associate with antithrombin and undergo irreversible formation of the covalent thrombin-antithrombin complex with exosite I occupied. This suggests that exosite I may remain accessible in the proteinase-inhibitor complex, which could inhibit further exosite I-dependent reactions by accumulation of this bound product. Results of other studies, however, suggest that this is not the case.
The thrombin-antithrombin complex formed in the presence of thrombomodulin does not inhibit the rate of subsequent thrombomodulin-enhanced inactivation of thrombin by antithrombin or activation of protein C, and it has been proposed that loss of affinity of the thrombin-antithrombin complex for thrombomodulin allows the cofactor protein to recycle following thrombin inactivation (22,27).
Collectively, the observations available present a paradox concerning the mechanism that can accommodate the relatively small, nonessential effects of exosite I interactions on the rate of thrombin inactivation by antithrombin, and also result in loss of affinity of the thrombin-antithrombin complex for exosite I ligands. The present studies were undertaken to investigate this question by determining the fate of exosite I in the thrombin-antithrombin reaction. A fluorescent analog of hirudin 54 -65 (Hir 54 -65 ), similar to the derivative of hirudin [53][54][55][56][57][58][59][60][61][62][63][64] described previously by Liu et al. (31), was prepared and used as a specific probe of the function of the exosite. The affinities of fluorescein-labeled Hir 54 -65 and unlabeled Hir 54 -65 for exosite I were decreased Ͼ200-fold in the thrombin-antithrombin complex. Results of kinetic studies indicate that thrombin with Hir 54 -65 bound to exosite I can associate with antithrombin, and that effectively irreversible inactivation of exosite I and dissociation of the peptide occur subsequent to the rate-limiting formation of the covalent thrombinantithrombin complex. These observations support a mechanism in which thrombin bound to regulatory macromolecules through exosite I may be inactivated by antithrombin without the requirement for dissociation of these complexes. Formation of the thrombin-antithrombin stable complex may drive the release of the inactive complex, thereby providing a mechanism for regulatory macromolecules to act as catalytic effectors of thrombin inactivation. This mechanism may function physiologically in clearing thrombin from regulatory macromolecules and participate in localizing thrombin activity.

EXPERIMENTAL PROCEDURES
Protein Purification and Characterization-Human ␣-thrombin was purified as described previously (32) or obtained from Dr. John Fenton (New York State Department of Health, Albany, NY). Thrombin concentration was determined by active site titration and absorbance at 280 nm with an absorption coefficient of 1.83 (mg/ml) Ϫ1 cm Ϫ1 and a molecular weight of 36,600 (33). Preparations of thrombin used in these studies were Ͼ95% active. Thrombin inactivated with FPR-CH 2 Cl (FPR-thrombin) was prepared by incubation of 30 M enzyme with a 2-fold excess of FPR-CH 2 Cl for 30 min. at 25°C, and the residual activity (Ͻ0.02%) was determined by chromogenic substrate assay, as described below. Thrombin was labeled at the active site with 4Ј-{[(iodoacetyl)amino]methyl}fluorescein and N ␣ -[(acetylthio)acetyl]-D-Phe-Pro-Arg-CH 2 Cl ([4Ј-AF]FPR-T) and characterized as described previously (34,35). The preparation used in these studies contained 0.87 mol of fluorescein/mol of thrombin. Antithrombin was purified from human plasma as described previously (36), and its concentration was determined by absorbance at 280 nm with an absorption coefficient of 0.65 (mg/ml) Ϫ1 cm Ϫ1 and a molecular weight of 58,000. Preparations of antithrombin were fully active, as measured by thrombin reaction stoichiometries of 1.08 Ϯ 0.08 mol/mol and a second-order rate constant for thrombin inhibition of 0.89 ϫ 10 4 M Ϫ1 s Ϫ1 under the conditions used previously (37). Thrombin-antithrombin complex (T-AT*) was prepared by reaction of 28 -39 M thrombin with a 1.7-2.2-fold excess of antithrombin for 10 -120 min at 25°C in 50 mM Hepes, 0.125 M NaCl, 1 mM EDTA, 1 mg/ml polyethylene glycol 8000, pH 7.40. The concentration of T-AT* was taken as the initial thrombin concentration, consistent with the essentially complete disappearance of thrombin observed by SDSgel electrophoresis. The residual thrombin concentration (Ͻ0.19%) was calculated from the residual chromogenic substrate activity.
Synthesis and Characterization of [5F]Hir 54 -65 -Solutions of the Tyr 63 -sulfated hirudin dodecapeptide (Sigma or Bachem) were prepared in water and their concentrations calculated from the purity specified by the manufacturers. The fluorescent derivative, [5F]Hir 54 -65 , was prepared by modifications of the procedure described previously for a similar derivative (31). 5-Carboxyfluorescein succinimidyl ester (Molec-ular Probes) in Me 2 SO was added to 0.4 -0.5 mM peptide to give a 5-9-fold molar excess in 50 mM sodium phosphate buffer, pH 8.0, and incubated for 2 h in the dark at 22°C. The reaction was stopped by addition of 1 M NH 4 Cl, pH 7.4, to 0.1 M and the labeled peptide was separated from excess dye by chromatography on Sephadex G10 (1 ϫ 119 cm) in water. The labeled peptide was purified by reverse-phase HPLC on a Beckman Ultrasphere 5-m C 18 -silica column (4.6 ϫ 150 mm) equilibrated with 0.1% trifluoroacetic acid in water and developed at 0.5 ml/min with 0.1% trifluoroacetic acid in CH 3 CN. The column was washed with trifluoroacetic acid/H 2 O for ϳ20 min after application of the sample, increased to 20% CH 3 CN over 10 min and washed for ϳ50 min, and eluted finally with a linear gradient from 20% to 50% CH 3 CN over 60 min. Fractions containing the major fluorescent peak eluting at ϳ36% CH 3 CN were neutralized by addition of 1 M NH 4 HCO 3 to 50 mM and lyophilized. The purified peptide was dissolved in water, and its concentration was determined from the fluorescein absorbance at 490 nm in 10 mM NaOH with an absorption coefficient of 89,125 M Ϫ1 cm Ϫ1 (38). Purified [5F]Hir 54 -65 eluted from the above column with a linear gradient as a single peak at 39% CH 3 CN. Solutions of [5F]Hir 54 -65 were stored frozen in water. The labeled peptide was stable in water for at least 2 days at room temperature in the dark, as shown by the unchanged HPLC elution profile.
Equilibrium Binding Studies-All experiments were performed in 50 mM Hepes, 0.125 M NaCl, 1 mM EDTA, 1 mg/ml polyethylene glycol 8000, pH 7.4, and at 25°C. Fluorescence measurements were made with an SLM 8000 fluorometer in the ratio mode, using acrylic cuvettes (Sarstedt) coated with polyethylene glycol 20,000 to minimize protein adsorption (39). Measurements of [5F]Hir 54 -65 binding to thrombin were performed with 491 nm excitation (4 nm band pass) and 515 nm emission (8 nm band pass), corresponding to the difference maxima from spectra of 0. ) to obtain the maximum fluorescence change (⌬F max /F o ), dissociation constant (K P ), and stoichiometric factor (n).
For analysis of less extensive data obtained with the same materials, to determine the dissociation constant (K C ) and stoichiometric factor (m) for competitive binding of Hir 54 -65 to thrombin, as well as K P and ⌬F max /F o (40,41).
Conditions of these experiments were chosen with [[5F]Hir 54 -65 ]ϽK P to reduce the dependence of the fit on n, which was fixed at its separately determined value. The same methods were used to analyze binding of Residual thrombin concentration ([T] t ) was expressed as the fraction of the initial activity measured for an identical control reaction mixture lacking AT, to account for small effects of transferred hirudin peptide on the assay rate. Reaction progress curves were fit by Equation 3, for a single exponential decay, to obtain the observed pseudo-first-order rate constant (k obs ), the reaction amplitude ([T] o ), and end point (B).
For analysis of reactions predicted to deviate from initially exponential behavior, model-independent estimates of the initial k obs (k obs o ) were obtained by fitting Equation 4 to the data, where the time dependence of the deviation from exponential behavior was approximated by a second degree polynomial.
The data were truncated at 90% of completion for this analysis, and k obs o was taken as the least-squares fit value of the coefficient, b. The apparent bimolecular rate constants were calculated as k obs o /[AT] o . Kinetic studies of the dissociation of [5F]Hir 54 -65 from T upon its reaction with AT were done by first measuring the decrease in fluorescence on addition of thrombin to the fluorescent peptide, and then the increase with time subsequent to initiation of the reaction with AT. The fluorescence changes measured with time (⌬F t ) were transformed into the total concentrations of thrombin binding sites for the peptide probe (P) at the corresponding times ([T] t ) with Equation 5, which is a rearrangement of the binding equation (Equation 1), using the parameters determined in the binding studies.
The data were expressed as the fraction of the initial total thrombin concentration remaining ([T] t /[T] o ), and the progress curves were analyzed as described for the activity experiments.
Least-squares fitting, numerical integration, and simulation were performed with SCIENTIST software (MicroMath Software). All reported estimates of error represent Ϯ 2 S.E. 54 -65 that was modified at the amino terminus with 5-carboxyfluorescein ([5F]Hir 54 -65 ) was synthesized, and its binding to thrombin was characterized from changes in fluorescence intensity. As shown in Fig. 1A, thrombin titrations of fixed concentrations of [5F]Hir 54 -65 from 10 nM to 500 nM were well described by binding of the peptide to 0.91 Ϯ 0.04 sites on thrombin with a dissociation constant of 26 Ϯ 2 nM, and a maximum decrease in fluorescence of 26 Ϯ 0.3% (Table I).

Binding of Hirudin Peptides to Thrombin-A fluorescent analog of Hir
Binding of unlabeled Hir 54 -65 to thrombin was examined from its effect on the [5F]Hir 54 -65 interaction. Addition of Hir 54 -65 to mixtures of thrombin and the labeled peptide resulted in return of the quenched fluorescence toward the initial value (Fig. 1B). Simultaneous nonlinear least-squares analysis of titrations with Hir 54 -65 at fixed thrombin concentrations ranging from 20 to 460 nM indicated a good fit with the equation for high affinity competitive binding of the labeled and unlabeled peptides, with Hir 54 -65 binding to 0.89 Ϯ 0.07 sites on thrombin with a dissociation constant of 38 Ϯ 5 nM (Table I). These results indicated that Hir 54 -65 and the fluorescein-labeled analog bound competitively and with similar, high affinity to exosite I of thrombin.
Binding of Hirudin Peptides to the Thrombin-Antithrombin Complex-In contrast to the results for thrombin, the thrombin-antithrombin stable complex (T-AT*) or antithrombin (AT) decreased the fluorescence of [5F]Hir 54 -65 by Ͻ3% at concentrations up to 3 M T-AT* or 10 M AT (Fig. 2). This indicated either an undetectably low affinity of the peptide for T-AT*, or that the peptide bound but there was no significant change in fluorescence resulting from the interaction. To investigate this further, the effect of the T-AT* complex on binding of [5F]Hir 54 -65 to thrombin that had been active site-blocked with  Table I. Titrations were performed and the data were analyzed as described under "Experimental Procedures." D-Phe-Pro-Arg-CH 2 Cl (FPR-thrombin) was examined. Active site-blocked thrombin was used in these experiments to prevent reaction with excess antithrombin present in the mixture of thrombin and antithrombin used to form the T-AT* complex. As shown by the titrations of 11 nM and 110 nM [5F]Hir 54 -65 with FPR-thrombin in Fig. 3, binding of the peptide to active site-blocked thrombin was accompanied by a slightly larger, 30 Ϯ 0.2% change in fluorescence when compared with native thrombin, and the same dissociation constant (Table I). The presence of 4.2 M T-AT* complex had no significant effect on titrations of the labeled peptide with FPR-thrombin (Fig. 3). Simulation of the effect of competitive binding of [5F]Hir 54 -65 to T-AT* and FPR-thrombin indicated that an affinity of the labeled peptide for T-AT* complex up to 200-fold lower than the affinity for thrombin would have been detected (Fig. 3).
To determine if the low affinity of the T-AT* complex for [5F]Hir 54 -65 was due to the presence of the fluorophore on the peptide, binding of unlabeled Hir 54 -65 was studied using an active site-labeled fluorescent thrombin derivative as a binding probe. To identify a fluorescent derivative with suitable properties, six derivatives were prepared by active site-specific inactivation of thrombin with either of the two thioester tripeptide chloromethyl ketones, ATA-FPR-CH 2 Cl and ATA-FFR-CH 2 Cl (32,34). The thiol generated from the thioester on the incorporated peptides was labeled with each of three structural isomers of fluorescein-iodoacetamide in which the iodoacetamide group was linked through the 5-, 6-, or 4Ј-positions of the fluorescein ring system (34). Results of screening these thrombin derivatives for those that signaled binding of Hir 54 -65 showed, remarkably, that five of the six derivatives exhibited small, Ϫ1 to  (Fig. 4). Simulation of the effect of competitive binding indicated that, similar to the previous results with [5F]Hir 54 -65 , the affinity of Hir 54 -65 for T-AT* was at least 200-fold lower than the affinity for native thrombin (Fig. 4). Titrations of the active site-labeled thrombin in the absence of Hir 54 -65 with up to 2 M native thrombin, 4 M T-AT*, or 10 M AT produced Ϯ1% changes in fluorescence (data not shown), indicating no significant nonspecific interactions.
Kinetics of Thrombin Inactivation by Antithrombin and the Loss of Affinity for the Hirudin Peptide-Addition of antithrombin to a mixture of thrombin and the fluorescent hirudin peptide resulted in a time-dependent return of the quenched fluorescence of [5F]Hir 54 -65 to the initial value, consistent with dissociation of the peptide from thrombin upon its reaction with antithrombin. On the basis of the previous results, the changes in fluorescence were taken as measures of the concentrations of thrombin containing the peptide binding site, which were calculated using the binding parameters that had been determined (see "Experimental Procedures"). As shown in Fig.  5, the loss of thrombin binding sites for [5F]Hir 54 -65 closely paralleled the loss of thrombin activity following addition of antithrombin, over the full course of the reaction. Similar comparisons were done at other concentrations of [5F]Hir 54 -65 (see below). The measured fluorescence changes of 5-17% for these reactions corresponded to calculated changes in the concentration of thrombin able to bind the peptide that were indistinguishable from the initial thrombin concentrations (89 Ϯ 30%). Initial exponential analysis of activity loss kinetics The large error in this calculation reflected the diminishing reaction amplitude at higher concentrations of the fluorescent peptide, and a deviation of the fluorescence amplitudes of Յ 2% from the predicted maxima. For all of the reactions, the final fluorescence was within 1% of the initial fluorescence of the peptide alone, indicating complete dissociation of the peptide.
Effect of Hirudin Peptides on the Kinetics of Thrombin Inactivation by Antithrombin-Kinetic studies were undertaken to characterize the mechanism of the effect of the hirudin peptides on the reaction of antithrombin with thrombin, from the time courses of enzyme inactivation and peptide dissociation. The initial results, together with previous observations (7), suggested that the reactions under bimolecular conditions would follow the mechanism shown in Scheme I.
In this mechanism, thrombin (T) and the thrombin-peptide (P) complex in rapid equilibrium can react with AT with different second-order rate constants, k and kЈ, to form the stable T-AT* complex. The reaction of the thrombin-peptide complex with AT is accompanied by a large decrease in affinity for the peptide. This integrates to an exponential function characterized by an initial, pseudo-first-order rate constant (k obs o ) (shown in Equation 8). The fraction of thrombin sites occupied by the peptide is given by the binding equation (Equation 9).
To allow for the predicted deviation from exponential behavior following the initial reaction phase, the reaction progress curves were fit by the exponential function with a second degree polynomial in time as the exponent. This approximation was valid because the deviations from exponential behavior were small (Fig. 5), although significant enough to affect the analysis. This method allowed for deviation of the reactions from first-order behavior and permitted estimation of the pseudo-first-order rate constant for the initial portion of the reaction, k obs o , from the fitted coefficient of the first-order term of the polynomial. As shown in Fig. 6A, the apparent bimolecular rate constant, k obs o /[AT] o , decreased with increasing concentration of [5F]Hir 54 -65 , approaching a limiting value at high concentrations. The rate constant for the fluorescence-monitored peptide dissociation was indistinguishable from the rate constant for activity loss, for all of the reactions. The pseudo-firstorder rate constants increased linearly with the antithrombin concentration from 500 to 2000 nM in the presence of 45 nM and 150 nM [5F]Hir 54 -65 , or 400 nM Hir 54 -65 (data not shown), indicating that the loss of activity and dissociation of the peptide were both second-order reactions. For the reactions in the absence of the peptide, which were known to be pseudo-firstorder processes, there was good agreement (Ϯ6%) between k obs o and the rate constants obtained by fitting of single exponentials, which supported the validity of the analysis. In the pres-  k)), as given by Equation 8, and fit with the binding equation (Eqn. 9) to determine kЈ and K P , with n and k fixed at their determined values. This gave a limiting bimolecular rate constant at saturating peptide concentration (kЈ) that was 1.7-fold lower than the value in the absence of the peptide (k), and an apparent peptide dissociation constant (K P ) of 16 Ϯ 13 nM (Fig. 6A, Table I). The latter value was indistinguishable from the 26 Ϯ 2 nM dissociation constant determined for free thrombin in the binding studies, supporting the mechanism in Scheme I.
The effect of Hir 54 -65 on the kinetics of the thrombin-antithrombin reaction was studied similarly from the loss of thrombin catalytic activity. Results of these experiments showed similar behavior (Fig. 6B), with a decrease in the rate constant of 2.1-fold and an apparent dissociation constant of 31 Ϯ 24 nM, in agreement with the value of 38 Ϯ 5 nM determined for binding of the unlabeled peptide to thrombin (Table I).
The above analysis of the kinetics supported the mechanism of Scheme I but resulted in rather large uncertainties in the parameters because of the small effect of the peptides on the rate, and error inherent in the measurements. To test the mechanism further, all of the full time course fluorescence and activity data collected as a function of [5F]Hir 54 -65 concentration, represented by the results in Fig. 6A, were fit without approximation by numerical integration of the differential equation (Equation 6) to solve for the thrombin concentration as a function of time, with changes in the concentration of the thrombin-peptide complex with time calculated with Equation 9. This analysis gave second-order rate constants indistinguishable from those determined by the initial exponential analysis (Table I) and an apparent peptide dissociation constant of 19 Ϯ 6 nM, in closer agreement with the dissociation constant obtained in the equilibrium binding studies. Application of the same analysis to all of the progress curve data for Hir 54 -65 gave a dissociation constant of 37 Ϯ 7 nM and rate constants that were similarly in good agreement with the values obtained in the preceding analysis and binding studies (Table I). These results supported the adequacy of the mechanism in Scheme I as a quantitative description of the effect of the hirudin peptides on the kinetics of thrombin inactivation and peptide dissociation.
The effect of the peptides on the inhibition reaction stoichiometry was examined to determine if binding of the peptides to thrombin decreased the rate by enhancing an alternative substrate reaction of antithrombin with thrombin (42,43). The stoichiometry of 1.01 Ϯ 0.03 mol of AT/mol of T obtained in the absence of peptide was unchanged by 5.  Table I. B, the dependence of the apparent bimolecular rate constant for inactivation of thrombin by antithrombin on the concentration of unlabeled hirudin peptide ([Hir 54 -65 ] o ) is shown, along with the least-squares fit of the data (solid line) with the parameters listed in Table I. Kinetics were measured and analyzed as described under "Experimental Procedures." Scheme 2.
A rapidly reversible Michaelis-type of enzyme-inhibitor complex (T⅐AT) is formed initially between the proteinase active site and the reactive-site sequence of the inhibitor, exposed on a flexible loop (45)(46)(47)(48). The reactive-site bond is subsequently engaged chemically by the proteinase as a normal protein substrate, in an irreversible reaction that results in a covalent intermediate complex (T-AT) (44,49). A conformational change in the T-AT complex results finally in irreversible trapping of thrombin in the stable product complex (T-AT*). In parallel with this reaction, the covalent intermediate can undergo completion of the substrate reaction, ultimately deacylating to yield reactive-site cleaved inhibitor (AT m ) and active proteinase (42-44, 50, 51). The rate-limiting step of thrombin inactivation is the formation of the covalent T-AT intermediate, which is thought to trigger the subsequent, faster serpin conformational change (52,53).
The thrombin-antithrombin mechanism can be expanded to include all of the exosite I binding equilibria as shown in Scheme 2. The results of the binding studies indicate that the affinity of thrombin in the product, T-AT* complex, for the hirudin peptides is reduced at least 200-fold compared with free thrombin. Characterization of binding of the fluoresceinlabeled hirudin 54 -65 peptide to human ␣-thrombin gave results consistent with those reported previously for bovine thrombin and an analogous derivative of hirudin [53][54][55][56][57][58][59][60][61][62][63][64] in which the probe was coupled through slightly different chemistry (31). The ϳ4fold higher affinity of the peptide used here and slightly larger fluorescence change are probably accounted for by the structural differences between the peptides and the lower affinity of such peptides for bovine thrombin compared with the human enzyme (6). In the present studies, [5F]Hir 54 -65 and the unlabeled peptide bound competitively and with similar, high affinity to a single site on thrombin, while [5F]Hir 54 -65 had no detectable affinity for T-AT* in direct fluorescence titrations or in competitive binding experiments with FPR-thrombin. Evidence of a similar loss of affinity for the unlabeled peptide was obtained in experiments in which active site labeled thrombin was used as a macromolecular fluorescence probe of binding. These results demonstrated that the fluorescence changes of [5F]Hir 54 -65 reported the peptide binding equilibrium, rather than perturbations due to nonspecific interactions. The absence of detectable binding of the peptides to the T-AT* complex is concluded to be due to loss of specific interactions with exosite I.
Although the thrombin-antithrombin reaction intermediates in Scheme 2 were not resolved in the present studies, the results of the kinetic studies support the conclusion that the loss of affinity of exosite I for the hirudin peptide occurs in the rate-limiting chemical reaction or a subsequent step. This is supported by the correspondence between the time courses of peptide dissociation and loss of thrombin activity, and between the dissociation constants for the peptides determined kineti-cally and by direct binding. The relatively small inhibition of the reaction, with 1.7-and 2.1-fold lower bimolecular rate constants at saturating peptide, rules out mechanisms of inhibition in which peptide binding competes with antithrombin for formation of the T⅐AT complex, or prevents the subsequent chemical reaction (Scheme 2). These mechanisms would show complete inhibition at saturating peptide concentration. These findings are in agreement with the previously reported effect of Hirugen (7), and the conclusion that recognition of antithrombin by thrombin does not depend significantly on exosite I (7,(15)(16)(17). The significant residual rate at saturating peptide concentrations implies that the thrombin-peptide complex reacts with antithrombin and undergoes conversion to the covalent P⅐T-AT complex. On this basis, it is concluded that dissociation of the peptide occurs in a step following either the irreversible reaction that produces the covalent complex, or following a subsequent step that occurs at a faster rate (Scheme 2).
Because the kinetic studies were restricted to bimolecular reaction conditions, the 1.7-and 2.1-fold inhibition of the rate by binding of the peptides could be due to an increase in the dissociation constant for antithrombin binding to the thrombin-peptide complex, or to a decrease in the rate constant for covalent complex formation. Previous observations indicate that initial recognition of antithrombin by thrombin involves binding of the reactive-site sequence at the primary substrate specificity site and secondary subsites (45,48,(53)(54)(55). In other studies, binding of hirudin peptides to exosite I has been shown to change thrombin specificity toward tripeptide substrates, which interact at these sites (12,31,56). This linkage between interactions at exosite I and the active site is thought to account for the difference observed here in the affinity of the hirudin peptide for [4Ј-AF]FPR-thrombin and native thrombin (Table I). This linkage may be similarly responsible for the decrease in the bimolecular rate constant for thrombin reaction with antithrombin.
The results of the kinetic studies do not resolve whether the loss of affinity of thrombin for hirudin peptides occurs as a result of the formation of the covalent T-AT complex, or the subsequent serpin conformational change leading to the final T-AT* complex (Scheme 2). The apparently small effect of the peptide on steps prior to the chemical reaction implies that the exosite is not greatly affected by the initial proteinase-inhibitor association, and more likely is disrupted or blocked as a result of the major structural changes which accompany the conformational change. The conformational change is thought to involve incorporation of the reactive site loop into ␤-sheet A of antithrombin, accompanied by a significant movement of the proteinase relative to the inhibitor (30,47,48,57), and loss of specific binding affinity of antithrombin for heparin (52,58).
Other studies indicate that a conformational change also occurs in the thrombin component of the thrombin-antithrombin complex, as evidenced by an increased susceptibility to proteolysis by free thrombin (59,60). An initial site of cleavage was identified as Arg 68 (Arg 73 in the chymotrypsin numbering system) (60), which is located in the ␤-loop that forms an integral part of exosite I involved in binding of hirudin peptides (1,4,5). Cleavage of this loop in ␣-thrombin to produce ␤-forms of thrombin, or mutation of Arg 68 are accompanied by loss of exosite I functions, including cleavage of fibrinogen, and binding of hirudin and thrombomodulin (18,61). These observations suggest that exosite I is accessible to thrombin cleavage in the stable thrombin-antithrombin complex and that this occurs at an enhanced rate, supporting the idea that conformational changes in this region of the proteinase are likely responsible for the loss of affinity. Significant levels of degradation of the T-AT* complex requires excess thrombin and is relatively slow SCHEME 2 (59,60). The little degradation observed by SDS-gel electrophoresis (data not shown), and the correspondence between peptide dissociation and enzyme inactivation showed that the loss of affinity observed here was not due to proteolysis.
The results obtained for the hirudin peptide are thought to mimic the behavior of regulatory proteins that interact with exosite I. The finding that the loss of affinity of exosite I for hirudin peptide occurs concomitant with or after the ratelimiting irreversible inactivation step implies that a binding cycle exists for the peptide that may also apply to regulatory proteins, as illustrated by the model in Fig. 7. Previous binding and kinetic studies suggest that this model applies to interactions of thrombin with thrombomodulin and fibrin. Results of kinetic studies showed that thrombin can react with antithrombin while remaining bound to these regulatory proteins (7,19,20,24,26,28). Thrombin is displaced from complexes with fibrin I and II (7,11), and thrombomodulin (12, 13) by hirudin peptides, indicating that these interactions overlap in specificity for exosite I and that this binding accounts for a large portion of the affinity. In the case of thrombomodulin, the situation is more complicated because of a form of thrombomodulin that contains a covalently attached chondroitin sulfate chain, which contributes to the thrombin binding affinity by interaction with exosite II (24,62). The results of the present studies suggest that inactivation of exosite I accompanying formation of the T-AT* complex will result in substantially reduced affinity for fibrin and thrombomodulin, and result in clearance of thrombin from these complexes as the inactivated antithrombin complex. This mechanism is thought to be responsible for the lack of inhibition by the T-AT* complex of the effects of thrombomodulin on thrombin reactions with antithrombin (27) and protein C (22). The finding that thrombin bound to the rabbit heart microvascular endothelium is released as the thrombin-antithrombin complex may similarly involve inactivation of exosite I (63). In the mechanism proposed here, a regulatory protein that interacts with exosite I can act catalytically as an effector of thrombin inactivation. This mechanism may play an important role physiologically in localizing the activity of thrombin by coupling the inactivation of the proteinase at the sites of its exosite I-mediated interactions to the obligatory release of the proteinase-inhibitor complex. Such coupling would maintain the competency of regulatory proteins to bind thrombin and prevent dissemination of the enzyme from sites of vascular injury. FIG. 7. Model for exosite I-dependent interactions of a regulatory protein accompanying thrombin inactivation by antithrombin. Binding of a regulatory protein (R) to thrombin (T) through exosite I is proposed to allow reversible binding of antithrombin to the thrombin active site to form a regulatory protein-thrombin-antithrombin complex (R⅐T⅐AT). Inactivation of exosite I is proposed to accompany either the thrombin-antithrombin chemical reaction or the subsequent serpin conformational change, which are shown as a single irreversible reaction. Dissociation of the regulatory protein is postulated to occur subsequent to the rate-limiting irreversible formation of the stable product complex (T-AT*).