Evidence that both exosites on thrombin participate in its high affinity interaction with fibrin.

Exosite 1 on thrombin mediates low affinity binding to sites on the NH2 termini of the alpha- and beta-chains of fibrin. A subpopulation of fibrin molecules (gammaA/gamma'-fibrin) has an alternate COOH terminus of the normal gamma-chain (gammaA/gammaA-fibrin) that binds thrombin with high affinity. To determine the roles of exosites 1 and 2 in the high affinity interaction of thrombin with gammaA/gamma'-fibrin, binding studies were done with thrombin variants and exosite 1- or 2-directed ligands. alpha-Thrombin bound gammaA/gamma'-fibrin via high and low affinity binding sites. A peptide analog of the COOH terminus of the gamma'-chain that binds alpha-thrombin via exosite 2 blocked the high affinity binding of alpha-thrombin to gammaA/gamma'-fibrin, suggesting that the interaction of alpha-thrombin with the gamma'-chain is exosite 2-mediated. In support of this concept, (a) gamma-thrombin, which lacks a functional exosite 1, bound to gammaA/gamma'-fibrin, but not to gammaA/gammaA-fibrin; (b) thrombin R93A/R97A/R101A, an exosite 2-defective variant, bound only to gammaA/gamma'-fibrin via low affinity sites; and (c) exosite 2-directed ligands reduced alpha-thrombin binding to gammaA/gamma'-fibrin. However, several lines of evidence indicate that exosite 1 contributes to the high affinity interaction of thrombin with gammaA/gamma'-fibrin. First, the affinity of gamma-thrombin for gammaA/gamma'-fibrin was lower than that of alpha-thrombin. Second, removal of a low affinity binding site on the beta-chain of gammaA/gamma'-fibrin reduced its affinity for alpha-thrombin. Third, exosite 1-directed ligands reduced alpha-thrombin binding to gammaA/gamma'-fibrin. Taken together, these data suggest that, although exosite 2 mediates the interaction of thrombin with the gamma'-chain of gammaA/gamma'-fibrin, simultaneous ligation of exosite 1 by low affinity binding sites is essential for the high affinity interaction of thrombin with gammaA/gamma'-fibrin.

Thrombin is a serine protease that plays an essential role in hemostasis, effecting procoagulant, anticoagulant, and profibrinolytic responses. As a procoagulant, thrombin activates platelets, converts fibrinogen to fibrin, and promotes its own generation by activating factors V, VIII, and XI. After binding to thrombomodulin, its cellular receptor, thrombin triggers anticoagulant and antifibrinolytic pathways by activating protein C and thrombin-activable fibrinolysis inhibitor, respectively (1,2).
The activity of thrombin is primarily regulated through inhibition by antithrombin in a reaction that is enhanced by heparin (3). Although the heparin-antithrombin complex readily inhibits fluid-phase thrombin, thrombin that remains bound to fibrin after clotting is protected from inactivation by the heparin-antithrombin complex. This protection reflects, at least in part, heparin-mediated bridging of thrombin onto fibrin to form a ternary thrombin-heparin-fibrin complex (4,5).
Fibrin-bound thrombin retains its ability to cleave fibrinogen (4, 6), a phenomenon that provides a plausible explanation for the in vitro observation that thrombi induce activation of platelets and trigger coagulation (7). The concept that fibrin-bound thrombin contributes to the procoagulant activity of thrombi is supported by the recent demonstration that human thrombi obtained from pathological specimens harbor active thrombin (8). Sequestered thrombin may be generated by activated factor X (factor Xa) bound to platelets within the thrombus. Like fibrin-bound thrombin, factor Xa within the platelet-bound prothrombinase complex also is protected from inactivation by the heparin-antithrombin complex (9,10). Consequently, thrombus-associated factor Xa may further contribute to the procoagulant activity of thrombi by triggering local thrombin generation (11).
Thrombin binds to fibrin via a domain distinct from the active site of the enzyme. Thrombin possesses two electropositive domains (termed exosites) that bracket the catalytic site. Exosite 1 serves as the initial docking site that orients substrates and inhibitors within the active site cleft (12). The second electropositive domain (exosite 2) binds heparin, other glycosaminoglycans, and prothrombin fragment 2 (13). In addition to acting as binding sites, the two exosites on thrombin may also modulate thrombin enzymatic activity. Thus, ligand binding to either exosite alters the active site environment (14), effecting conformational changes that alter thrombin substrate specificity (2,6). Moreover, there is evidence for allosteric linkage between the two exosites because ligand binding to one exosite can influence the binding properties of the other (15).
Thrombin utilizes exosite 1 to dock fibrinogen as it is converted to fibrin. Fibrinogen is composed of duplicated A␣-, B␤-, and ␥-chains held together by disulfide bonds. Two chromatographically distinct forms of fibrinogen, distinguished by the structure of their ␥-chains, can be isolated from human plasma (16). Most fibrinogen molecules contain two ␥ A -chains, each composed of 411 amino acids, and are thus designated ␥ A /␥ A . About 10% of circulating fibrinogen molecules contain a variant of the native ␥-chain (termed ␥Ј) composed of 427 amino acids. This longer ␥Ј-chain results from alternative mRNA polyadenylation and has an acidic hirudin-like 20-amino acid sequence at its COOH terminus (17,18). Fibrinogen het-erodimers containing one ␥ A -and one ␥Ј-chain are designated ␥ A /␥Ј.
Conversion of ␥ A /␥ A -or ␥ A /␥Ј-fibrinogen to fibrin requires sequential thrombin-mediated release of fibrinopeptides A and B from the NH 2 termini of the A␣-and B␤-chains of fibrinogen, respectively. Current thinking is that once thrombin-mediated fibrinopeptide release is effected, some of the thrombin remains bound to the resultant fibrin via exosite 1 (16). Functionally, ␥ A /␥ A -and ␥ A /␥Ј-fibrin are distinguished by their thrombinbinding properties (16). Binding of thrombin to ␥ A /␥ A -fibrin occurs through a single class of low affinity binding sites that have been localized to the NH 2 termini of the ␣and ␤-chains, proximal to the fibrinopeptide cleavage sites (19). Thrombin binds to these low affinity sites with K d values that range from 1 to 3.4 M (20). In addition to low affinity thrombin binding, ␥ A /␥Ј-fibrin also displays high affinity thrombin binding that has been attributed to the COOH terminus of the ␥Ј-chain. The K d value for this interaction is 0.18 M (16).
To further explore the high affinity interaction of thrombin with ␥ A /␥Ј-fibrin, we measured the affinity of thrombin for a synthetic 20-amino acid peptide analog of the COOH terminus of the ␥Ј-chain. Using thrombin variants or DNA aptamers directed against the exosites, we demonstrated that thrombin binding to this ␥Ј-chain peptide is mediated by exosite 2. Building on this observation, we then explored the role of exosites 1 and 2 in the interaction of thrombin with ␥ A /␥ A -and ␥ A /␥Јfibrin. These studies suggest that both exosites on thrombin contribute to the high affinity interaction of thrombin with ␥ A /␥Ј-fibrin. In contrast, the low affinity interaction of thrombin with ␥ A /␥Ј-or ␥ A /␥ A -fibrin is mediated solely by exosite 1.
Isolation of ␥ A /␥ A -and ␥ A /␥Ј-Fibrinogen-Human fibrinogen was fractionated on a DEAE-Sepharose column (1.5 ϫ 11.3 cm) (31). After diluting fibrinogen with water and 39 mM Tris and 5 mM phosphoric acid (pH 8.8) (buffer A) at a 1:1:1 ratio, 180 mg of fibrinogen was applied to the column at a flow rate of 5 ml/min. A concave pH gradient from buffer A to 500 mM Tris and 500 mM phosphoric acid (pH 4) was used to separate ␥ A /␥ A -from ␥ A /␥Ј-fibrinogen (32). 10-ml fractions were collected, and their absorbance at 280 nm was determined using a Beckman Gold 168 detector. ␥ A /␥ A -and ␥ A /␥Ј-fibrinogen fractions were precipitated by addition of ammonium sulfate to 50%, incubated for 3 h at 4°C, and subjected to centrifugation at 2000 ϫ g for 15 min at 23°C. Pellets were resuspended in Tris-buffered saline (TBS; 20 mM Tris-HCl and 150 mM NaCl (pH 7.4)) containing 0.01% Tween 20 (TBS/Tween) and incubated for 60 min at 37°C prior to dialysis against TBS/Tween. Fibrinogen concentrations were determined using a Beckman DU 7400 spectrophotometer by measuring absorbance at 280 and using M r ϭ 340,000 and ⑀ 280 1% ϭ 15 (33). Aliquots were stored at Ϫ70°C. The integrity of isolated ␥ A /␥ A -and ␥ A /␥Ј-fibrinogen fractions was assessed by SDS-PAGE (34) on 4 -15% polyacrylamide gels (Ready-Gel, Bio-Rad, Mississauga, Ontario, Canada) under reducing conditions (Fig. 1A). The gels were then transferred to nitrocellulose and subjected to immunoblot analysis using an antibody directed against the ␥Ј-chain (Fig. 1B). Two bands are seen in the ␥-chain region of ␥ A /␥Ј-fibrinogen (Fig. 1A, lane 2), but only the higher molecular weight band of the doublet was recognized by the antibody against the ␥Ј-chain (Fig. 1B, lane 2). In contrast, the single ␥-chain band of ␥ A /␥ A -fibrinogen (Fig. 1A, lane 1) was not recognized by the antibody (Fig. 1B, lane 1). Preparation of Des-B␤1-42-␥ A /␥ A -and Des-B␤1-42-␥ A /␥Ј-Fibrinogen-␥ A /␥ A -or ␥ A/ ␥Ј-fibrinogen (8.4 M; in TBS containing 12 mM EDTA) was incubated with 2.3 M protease III fraction from C. atrox venom for 90 min at 23°C. Reactions were terminated by addition of FPRck to 10 M. After 10 min, 20% (v/v) ethanol was added, and the solution was maintained on ice for 30 min prior to centrifugation at 14,000 ϫ g for 4 min. Pellets were resuspended in 1 ml of TBS and dialyzed against TBS. Protein concentrations were determined as described above, and the material was stored in aliquots at Ϫ70°C. The integrity of des-B␤1-42-␥ A /␥ A -or des-B␤1-42-␥ A /␥Ј-fibrinogen was assessed by SDS-PAGE analysis. Immunoblot analysis using an antibody against the ␥Ј-chain was used to ensure that treatment with C. atrox venom had no effect on the COOH-terminal region of the ␥Ј-chain of ␥ A /␥Ј-fibrinogen. Upon SDS-PAGE, the ␤-chain lacking B␤1-42 migrated with the ␥-chain (Fig. 1A, lanes 3 and 4). Immunoblot analysis demonstrated that the ␥Ј-chain of des-B␤1-42-␥ A /␥Ј-fibrinogen was still recognized by the antibody against the ␥Ј-chain (Fig. 1B, lane 4). NH 2terminal sequence analysis of the ␤-chain of des-B␤1-42-␥ A /␥ A -or des-B␤1-42-␥ A /␥Ј-fibrinogen confirmed that the ␤-chains of both species were cleaved at the ␤42-43 bond. In contrast, the sequences of the NH 2 termini of the ␣and ␥-chains corresponded to those of native fibrinogen (data not shown).
Labeling of Proteins-Fluorescent derivatives of ␣-IIa, ␥-IIa, and RA-IIa (f-FPR-␣-IIa, f-FPR-␥-IIa, and f-FPR-RA-IIa, respectively) were prepared by incubation with a 5-fold molar excess of fluorescein-FPRck until there was no detectable chromogenic activity against tGPR-pNA. Samples were then dialyzed against TBS, and thrombin concentrations were determined using M r ϭ 37,000 and ⑀ 280 1% ϭ 18. A fluorescent derivative of the ␥Ј-peptide (f-␥Ј-peptide) was generated by incubating the peptide with a 10-fold molar excess of fluorescein isothiocyanate in 1 M NaHCO 3 (pH 9) for 90 min at 23°C. Unreacted fluorescein isothiocyanate was removed by passing the material over a 10-ml Sephadex G-10 column equilibrated with TBS, and the concentration was determined as described (15).
␣-IIa was radiolabeled with Na 125 I (PerkinElmer Life Sciences) using IODO-BEAD (Pierce) as described (35). Specific radioactivity ranged from 0.7 to 1.5 ϫ 10 9 cpm/mg. Radiolabeled ␣-IIa migrated with its unlabeled counterpart when subjected to SDS-PAGE on 4 -15% polyacrylamide gels under reducing conditions. Radiolabeled ␣-IIa bound to fibrin monomer-Sepharose (36) and to heparin-Sepharose with affinities similar to those of unlabeled ␣-IIa, thereby confirming the integrity of exosites 1 and 2, respectively. Only the fraction of 125 I-␣-IIa that bound to the columns was used. Active site-blocked 125 I-␣-IIa was prepared by incubating 125 I-␣-IIa with a 10-fold molar excess of FPRck, followed by dialysis against TBS. When prepared in this fashion, 125 I-FPR-␣-IIa had no activity against tGPR-pNA (37). For binding studies, trace amounts of 125 I-FPR-␣-IIa were combined with FPR-␣-IIa.

␥Ј-Peptide Binding to f-FPR-IIa-f-FPR-␣-IIa
, f-FPR-␥-IIa, or f-FPR-RA-IIa (at a concentration of 100 nM in 1 ml of TBS) was added to a semi-micro quartz cuvette, and samples were stirred with a mini-stir bar and maintained at 23°C using a circulating water bath. Initial fluorescence (I o ) was monitored at 1-s intervals using a PerkinElmer LS 50B luminescence spectrometer at excitation and emission wavelengths of 492 and 535 nm, respectively, with a 515-nm cutoff filter in the emission beam and excitation and emission slit widths of 10 nm. Samples were then titrated with 1-10-l aliquots of 50 M ␥Ј-peptide containing 100 nM f-FPR-IIa to prevent fluorescent probe dilution. The signal was allowed to stabilize before each addition, and fluorescence intensity values (I) were obtained from the time drive profile. I/I o values were plotted against the concentration of ␥Ј-peptide, and binding was analyzed by nonlinear regression (Table Curve, Jandel Scientific, San Rafael, CA) in Equation 1, where L o is the concentration of ligand added, P o is the concentration of target protein, ␣ is the maximum change in emission intensity, and n is the stoichiometry. Experiments were then performed to examine the ability of DNA aptamers directed against exosite 1 or 2 (HD1 and HD22, respectively) to displace the ␥Ј-peptide from ␣-IIa. The fluorescence of a 1-ml sample containing 1 M thrombin was monitored before and after addition of 50 nM f-␥Ј-peptide. The sample was subsequently titrated with 1-2-l aliquots of HD1 or HD22 to a final concentration of 5.6 or 4.4 M, respectively. Data were analyzed as described above.
Thrombin Binding to Fibrin-Binding studies were performed with either radiolabeled or unlabeled thrombin. When labeled thrombin was used, 125 I-FPR-␣-IIa (0 -8 M) was added to microcentrifuge tubes containing 2 mM CaCl 2 in the absence or presence of 2 M ␥ A /␥ A -or ␥ A /␥Ј-fibrinogen. Clotting was initiated by addition of 10 nM ␣-IIa; and after a 60-min incubation at 23°C, fibrin was pelleted by centrifugation at 14,000 ϫ g for 4 min. Two 10-l aliquots of supernatant were removed and counted for radioactivity. The concentration of unbound thrombin in the supernatant was determined by comparison with a standard curve prepared in the absence of fibrinogen, and the concentration of the bound thrombin was then calculated by subtraction.
Binding of unlabeled thrombin to fibrin was assessed in 1.5-ml microcentrifuge tubes to which were added 2 mM CaCl 2 , 30 nM ␣-, ␥-, or RA-IIa, and 0 -10 M ␥ A /␥ A -or ␥ A /␥Ј-fibrinogen to a total volume of 50 l. Because ␥-IIa has minimal procoagulant activity, atroxin was added to 5% (v/v) to initiate clotting. After a 60-min incubation at 23°C, fibrin was pelleted by centrifugation at 14,000 ϫ g for 4 min, and two 10-l aliquots of supernatant were removed. The chromogenic activity of thrombin in each of these samples was assessed by their addition to the wells of a 96-well plate prefilled with 200 l of 200 M tGPR-pNA. Substrate hydrolysis was monitored at 405 nm in a plate reader (Molecular Devices, Sunnyvale, CA). The thrombin concentration was determined by comparison with a standard curve, and the concentration of bound thrombin was then calculated by subtraction. In all cases, values were divided by the fibrinogen concentration to determine the number of moles of thrombin bound per mol of fibrin (v). For each point in the titration, these values were then plotted against the concentration of unbound protein. Scatchard plots were also constructed, and depending on whether these appeared linear, reflecting a single class of binding sites, or curved downward, reflecting two classes of binding sites, data were analyzed by nonlinear regression analysis (Table Curve, Jandel Scientific) of a one-site (Equation 2) or two-site (Equation where L represents the concentration of unbound protein, n reflects stoichiometry, and K d is the dissociation constant (38). Statistical Methods-Unless otherwise indicated, experiments were performed at least three times in duplicate. Results are presented as the mean Ϯ S.E. Student's t test was used to evaluate significance.

RESULTS
Interaction of ␣-IIa with the ␥Ј-Peptide-To explore the interaction of ␣-IIa with the ␥Ј-peptide, the fluorescence of 100 nM f-FPR-␣-IIa was monitored as the sample was titrated with the ␥Ј-peptide. Addition of the ␥Ј-peptide elicited a saturable increase in the fluorescence intensity of f-FPR-␣-IIa (Fig. 2), consistent with binding, and nonlinear regression analysis of the data revealed a K d value of 2.2 M. This compares with the K d value of 0.68 M reported for interaction of the fluorescently labeled ␥Ј-peptide with ␣-IIa (24). To identify the exosite mediating the interaction of f-FPR-␣-IIa with the ␥Ј-peptide, the experiment was repeated using f-FPR-␥-IIa, a thrombin derivative lacking exosite 1, and f-FPR-RA-IIa, a thrombin variant with an impaired exosite 2. The ␥Ј-peptide bound to f-FPR-␥-IIa with a K d of 1.0 M (Fig. 2), an affinity similar to that for f-FPR-␣-IIa. In contrast, the ␥Ј-peptide did not bind to f-FPR-RA-IIa. These data suggest that exosite 2 mediates the interaction of ␣-IIa with the ␥Ј-peptide, in agreement with a recent report (24).
To further explore this concept, we performed reverse titrations to examine the ability of exosite 1-or 2-directed DNA aptamers to displace the f-␥Ј-peptide from ␣-IIa. The fluorescence intensity of 50 nM f-␥Ј-peptide decreased by 6% after addition of ␣-IIa (Fig. 3A). Titration with HD22, an aptamer directed against exosite 2, restored the fluorescence intensity to initial background levels, indicating displacement of the f-␥Ј-peptide from ␣-IIa. In contrast, HD1, an exosite 1-directed aptamer, had no effect (Fig. 3B). These data support the concept that the interaction of ␣-IIa with the ␥Ј-peptide is mediated by exosite 2.
Binding of 125 I-␣-IIa to ␥ A /␥ A -and ␥ A /␥Ј-Fibrin-The interactions of 125 I-FPR-␣-IIa with clots formed from ␥ A /␥ A -and ␥ A /␥Ј-fibrinogen were compared to examine the influence of the ␥Ј-chain on binding. Gel electrophoretic analysis verified that the ␥ A /␥ A -and ␥ A /␥Ј-fibrinogen preparations contained one and two ␥-chains, respectively (Fig. 1). 125 I-FPR-␣-IIa bound to both forms of fibrin in a concentration-dependent and saturable fashion. Scatchard analysis of the data for the interaction of 125 I-FPR-␣-IIa with ␥ A /␥ A -fibrin (Fig. 4A) reveals a straight line, consistent with a single class of binding sites. The K d of 2.3 Ϯ 0.32 M, determined by nonlinear regression analysis of the direct plot, is in good agreement with previously reported results (20). Binding to ␥ A /␥Ј-fibrin was also saturable, but the direct plot revealed a steeper increase in binding at lower 125 I-FPR-␣-IIa concentrations, indicating a higher affinity interaction. The Scatchard plot for the interaction of 125 I-FPR-␣-IIa with ␥ A /␥Ј-fibrin is nonlinear (Fig. 4B), indicating heterogeneous binding sites or negative cooperativity. A doublereciprocal plot of 1/bound versus 1/free yields a straight line, whereas a plot of bound 2 /free versus bound yields a sigmoidal curve (data not shown). These findings are indicative of binding site heterogeneity (39). Accordingly, the data were fit to a two-site model  (16).
To confirm the role of the ␥Ј-chain in the high affinity interaction of ␣-IIa with fibrin, we examined the influence of the ␥Ј-peptide on 125 I-FPR-␣-IIa binding to ␥ A /␥Ј-fibrin. 2 M ␥ A /␥Јfibrinogen was clotted in the presence of 125 I-FPR-␣-IIa (0 -8 M) in the absence and presence of 50 M ␥Ј-peptide (Fig. 5). The direct plot for the data in the presence of the ␥Ј-peptide closely resembles that for 125 I-␣-IIa binding to ␥ A /␥ A -fibrin. Scatchard analysis of the binding data in the presence of the ␥Ј-peptide yields a straight line, consistent with a single class of binding sites with a K d of 2.8 Ϯ 0.12 M, a value comparable to that for the low affinity interaction of ␣-IIa with ␥ A /␥ A -fibrin. These findings suggest that the ␥Ј-peptide eliminates the high affinity interaction of ␣-IIa with ␥ A /␥Ј-fibrin by competing with its ␥Ј-chain for access to exosite 2 on ␣-IIa. The data also support the concept that the ␥Ј-peptide serves as a good surrogate for the ␥Ј-chain of intact fibrin. The ␥Ј-peptide had no effect on the K d of 125 I-FPR-␣-IIa binding to ␥ A /␥ A -fibrin (data not shown).
Binding of ␣-IIa, ␥-IIa, or RA-IIa to ␥ A /␥ A -and ␥ A /␥Ј-Fibrin-Thrombin and thrombin variants were used to further explore the contribution of exosites 1 and 2 to the interaction of throm- bin with ␥ A /␥ A -or ␥ A /␥Ј-fibrin. These experiments were performed with unlabeled active thrombin species where unbound thrombin was detected by measuring chromogenic activity in clot supernatants. Active ␣-IIa bound to ␥ A /␥ A -fibrin with a K d value (2.25 Ϯ 0.05 M) comparable to that of 125 I-FPR-␣-IIa (Table I). RA-IIa, whose exosite 1 is intact, bound to ␥ A /␥ Afibrin with a K d value of 3.11 Ϯ 0.02 M, an affinity similar to that of ␣-IIa. In contrast, ␥-IIa, a variant with a cleaved exosite 1, did not bind to ␥ A /␥ A -fibrin. These findings are consistent with the concept that the low affinity interaction of ␣-IIa with ␥ A /␥ A -fibrin is mediated solely by exosite 1.
Experiments were then repeated using ␥ A /␥Ј-fibrin in place of ␥ A /␥ A -fibrin. ␣-IIa bound to ␥ A /␥Ј-fibrin with a K d of 80 Ϯ 30 nM, reflecting the high affinity interaction with the ␥Ј-chain. Because the affinity was measured by titrating ␣-IIa with fibrinogen, two-site binding of ␣-IIa to ␥ A /␥Ј-fibrin could not be discriminated. RA-IIa bound to ␥ A /␥Ј-fibrin with a K d of 2.02 Ϯ 0.03 M, an affinity similar to that of ␣-IIa for ␥ A /␥ A -fibrin. These findings suggest that without a functional exosite 2, RA-IIa cannot bind to the ␥Ј-chain and can interact only with low affinity binding sites on ␥ A /␥Ј-fibrin.
To further investigate the importance of exosite 2 in the interaction of thrombin with ␥ A /␥Ј-fibrin, the binding of ␥-IIa to ␥ A /␥Ј-fibrin was examined. ␥-IIa bound to ␥ A /␥Ј-fibrin with a K d of 5.54 Ϯ 0.02 M. Because ␥-IIa lacks a functional exosite 1, its interaction with ␥ A /␥Ј-fibrin must be mediated by exosite 2. Furthermore, ␥-IIa must bind to the ␥Ј-chain of ␥ A /␥Ј-fibrin because ␥-IIa does not bind to ␥ A /␥ A -fibrin. These concepts are supported by the observation that the ␥Ј-peptide inhibited ␥-IIa binding to ␥ A /␥Ј-fibrin in a concentration-dependent fashion (Fig. 6).
Effect of Exosite 1-or 2-directed Ligands on ␣-IIa Binding to ␥ A /␥ A -or ␥ A /␥Ј-Fibrin-To verify the contribution of the two exosites on ␣-IIa to its interaction with fibrin, binding assays were performed in the absence or presence of ligands directed at either exosite 1 or 2 of ␣-IIa, and the ability of these ligands to block ␣-IIa binding was determined. When ligands directed to exosite 1, hirudin-(54 -65), HD1, and heparin cofactor II-(54 -75), were present at saturating concentrations (50, 1, and 100 M, respectively), ␣-IIa binding to ␥ A /␥ A -or ␥ A /␥Ј-fibrin was reduced by 70 -80% (Fig. 8). These findings highlight the importance of ␣-IIa exosite 1 binding to either form of fibrin. To address the role of exosite 2, the experiment was then repeated with saturating concentrations of ␥Ј-peptide, h-F2, and HD22 (236, 64, and 0.25 M, respectively). The ␥Ј-peptide produced almost 100% reduction in ␣-IIa binding to ␥ A /␥Ј-fibrin. In contrast, the ␥Ј-peptide reduced ␣-IIa binding to ␥ A /␥ A -fibrin by 50%. Because the interaction of ␣-IIa with ␥ A /␥ A -fibrin is not expected to involve exosite 2 directly, these results suggest an allosteric linkage between the two exosites. Similar results were obtained with h-F2 and HD22, although the extent of reduction was less with both forms of fibrin. DISCUSSION Fibrin-bound thrombin has been implicated as an important trigger of thrombus growth (11,37). Consequently, it is of interest to define the mode of thrombin interaction with fibrin. Two classes of thrombin-binding sites have been identified on  For studies with ␥-IIa, 5% (v/v) atroxin was used to induce clotting. The concentration of bound ␣-IIa, ␥-IIa, or RA-IIa was calculated by comparing the thrombin chromogenic activity in the clot supernatants with that in control reactions containing all reagents except the thrombin variant and, in the case of ␥-IIa, atroxin. Data were analyzed by nonlinear regression analysis to determine the K d values. Each value represents the mean Ϯ S.E. of three experiments, each done in duplicate.

IIa variant
Native fibrin des-␤15-42-fibrin  6. Effect of the ␥-peptide on the binding of ␥-IIa to ␥ A /␥fibrin. 100 nM ␥-IIa was added to microcentrifuge tubes containing 10 M ␥ A /␥Ј-fibrinogen and 2 mM CaCl 2 in the absence or presence of the ␥Ј-peptide at the concentrations indicated. After clotting the fibrinogen with 5% (v/v) atroxin, the concentration of ␥-IIa bound to fibrin was calculated by comparing ␥-IIa-mediated hydrolysis of tGPR-pNA with that in control titrations lacking fibrinogen and atroxin. The concentration of ␥-IIa bound to ␥ A /␥Ј-fibrin in the presence of each concentration of ␥Ј-peptide is expressed as a percentage of that bound in the absence of the ␥Ј-peptide. Bars represent the mean of three experiments, each done in duplicate, and error bars reflect S.E. fibrin. The low affinity thrombin-binding sites, which reside on the NH 2 termini of the ␣and ␤-chains of fibrin, bind thrombin with a K d value ranging from 1 to 3.4 M (16,20). Recently, high affinity thrombin binding to fibrin has been attributed to the COOH terminus of the ␥Ј-chain of fibrin. Thrombin binds to this site with a K d value of 0.18 M (16). Because hirudin-(54 -65) disrupts thrombin binding to ␥ A /␥Ј-fibrin, it was presumed that thrombin bound to both classes of binding sites on fibrin via exosite 1, its substrate-binding domain (16). Our studies provide an alternative explanation, suggesting that, although exosite 2 mediates interaction of thrombin with the ␥Ј-chain, exosite 1 is necessary for high affinity binding of thrombin. In contrast, thrombin binding to the low affinity sites on fibrin is solely exosite 1-mediated.
Consistent with the previous report (16), ␣-IIa binds to ␥ A / ␥Ј-fibrin via two classes of binding sites: high affinity binding with a K d of 0.107 M and low affinity binding with a K d of 1.5 M. Exosite 2 contributes to the interaction of thrombin with ␥ A /␥Ј-fibrin because RA-IIa binds to ␥ A /␥ A -and ␥ A /␥Ј-fibrin with K d values similar to those for the low affinity interaction of ␣-IIa with ␥ A /␥Ј-and ␥ A /␥ A -fibrin. These findings suggest that because its exosite 2 function is impaired, RA-IIa does not bind ␥ A /␥Ј-fibrin with high affinity. In contrast, the exosite 1 variant ␥-IIa binds only ␥ A /␥Ј-fibrin. The interaction of ␥-IIa with ␥ A /␥Ј-fibrin likely reflects exosite 2-mediated binding to the ␥Ј-chain because the ␥Ј-peptide attenuates binding.
Further evidence that exosite 2 mediates the interaction of thrombin with the ␥Ј-chain comes from the demonstration that ␣-IIa and ␥-IIa bind the ␥Ј-peptide with similar affinities, a finding consistent with recently reported data (24). In contrast, RA-IIa, an exosite 2 variant, does not bind. Furthermore, ␣-IIa bound to the ␥Ј-peptide is displaced by an exosite 2-directed aptamer, but not by an aptamer that binds to exosite 1.
Despite its intact exosite 2, ␥-IIa binds to ␥ A /␥Ј-fibrin with an affinity lower than that of ␣-IIa. These findings raise the possibility that exosite 1 also contributes to the high affinity in-teraction of ␣-IIa with ␥ A /␥Ј-fibrin. Several observations support this concept. First, the affinity of ␣-IIa for des-␤15-42-␥ A / ␥Ј-fibrin is 44-fold lower than that for intact ␥ A /␥Ј-fibrin. Thus, abolishing exosite 1-mediated interaction with the low affinity binding site on the NH 2 terminus of the ␤-chain markedly reduces the affinity of ␣-IIa for ␥ A /␥Ј-fibrin. In contrast, removal of these low affinity binding sites from ␥ A /␥ A -fibrin produces only a 3.4-fold reduction in its affinity for ␣-IIa. Second, ␥-IIa binds to des-␤15-42-␥ A /␥Ј-fibrin with an affinity similar to that for intact ␥ A /␥Ј-fibrin. Likewise, removal of the NH 2 termini of the ␤-chains from ␥ A /␥Ј-fibrin reduces the affinity of RA-IIa to the same extent as removal of these low affinity binding sites from ␥ A /␥ A -fibrin.
The concept that both exosites on thrombin are necessary for its high affinity interaction with ␥ A /␥Ј-fibrin provides an explanation for the results of thrombin binding studies with dysfibrinogen Naples I. When clotted, fibrinogen Naples I (B␤ A68T) exhibits reduced affinity of both high and low affinity binding interactions (40). These findings suggest that attenuation of exosite 1-mediated interaction with the mutant NH 2 terminus of the ␤-chain impairs high affinity binding mediated by the ␥Ј-chain.
The proposal that both exosites are necessary for the high affinity interaction of ␣-IIa with fibrin is supported by studies using synthetic exosite 1-or 2-directed ligands. Exosite 1-directed ligands reduce ␣-IIa binding to either ␥ A /␥ A -or ␥ A /␥Јfibrin by 80 -100%, thereby confirming that this exosite is important for both the high and low affinity interaction of ␣-IIa with fibrin. Exosite 2-directed ligands reduce ␣-IIa binding to ␥ A /␥Ј-fibrin by 60 -100%, supporting the concept that exosite 2 is important for the interaction of ␣-IIa with ␥ A /␥Ј-fibrin. However, exosite 2 ligands also produce modest reductions in ␣-IIa binding to ␥ A /␥ A -fibrin, despite the fact that this interaction is mediated by exosite 1. These findings can be rationalized by our previous observation that there is allosteric linkage between the two exosites such that binding of a ligand to one exosite alters the affinity of the opposite exosite for its ligands (15). A recent study has challenged this concept based on the observation that prothrombin fragment 2, which binds to exosite 2 on thrombin, does not influence the binding of hirudin-(54 -65) to exosite 1 (41). This discrepancy likely reflects differences in assay technique because our current findings indicate that exosite 2-directed ligands, including prothrombin fragment 2, attenuate the binding of ␣-IIa to ␥ A /␥ A -fibrin via exosite 1.
Both prothrombin fragment 2 and the ␥Ј-chain of fibrinogen possess clustered anionic residues that likely mediate its interaction with the electropositive exosite 2 domain on thrombin (22,42,43). Although no homology between the sequences that bind to exosite 1 or 2 can be discerned (22), the two exosites appear to have functionally distinct roles. In general, exosite 1 on thrombin is utilized by substrates, cofactors, and inhibitors to gain direct access to the active site of the enzyme. In contrast, exosite 2 binds heparin or prothrombin fragment 2, both of which are ligands that recruit tertiary reactants, antithrombin or factor V, respectively. This suggests that exosite 2 functions primarily as an anchoring site, whereas exosite 1 serves to align substrates or inhibitors. This is consistent with the observation that exosite 2 functions are retained when thrombin is complexed by antithrombin (35). The exosite 2-mediated anchoring of thrombin to ␥Ј-chain of fibrinogen enhances the capacity of fibrin to harbor catalytically active thrombin. Thrombin bound in this fashion would remain active and protected from inhibition by heparin-antithrombin because the heparin-binding site of the bound thrombin is ligated by the ␥Ј-chain. The observation of a correlation between plasma lev- Ligands were used at concentrations that had maximal effects in prior dose-ranging experiments. After clotting the fibrinogen with 5% (v/v) atroxin, the concentration of 125 I-FPR-␣-IIa bound to fibrin was calculated as described in the legend to Fig. 4. The concentration bound in the presence of exosite 1-or 2-directed ligands was then expressed as a percentage of that bound in the absence of the ligand. Bars represent the mean of three experiments, each done in duplicate, and error bars reflect S.E. *, significant (p Ͻ 0.05) difference between the results with ␥ A /␥ A -and ␥ A /␥Ј-fibrin.