Evidence for Allosteric Linkage between Exosites 1 and 2 of Thrombin*

Investigations to date have demonstrated that ligand binding to exosites 1 or 2 on thrombin produces conformational changes at the active site. In this study, we directly compared the effect of ligand binding to exosites 1 and 2 on the structure and function of the active site of thrombin and investigated functional linkage between the two exosites. Binding studies were performed in solution with fluorescein-Phe-Pro-Arg-CH2Cl (FPR)-thrombin. Hirudin-(54–65) and sF2, a synthetic peptide corresponding to residues 63–116 of prothrombin fragment 2, were used as ligands for exosites 1 and 2 of thrombin, respectively. The two ligands produce diametric changes in the fluorescence of fluorescein-FPR-thrombin and also have opposing effects on the rate of thrombin hydrolysis of a number of chromogenic substrates. These results indicate that sF2 and hirudin-(54–65) differentially affect the conformation of the active site. Experiments then were performed to investigate whether both ligands can bind to thrombin simultaneously. When thrombin-bound fluorescein-sF2 is titrated with hirudin-(54–65), complete displacement of fluorescein-sF2 is observed. Likewise, when thrombin-bound fluorescein-hirudin-(54–65) is titrated with sF2, complete displacement occurs. Additional support for reciprocal binding was obtained in fluorescence experiments where both probes were labeled and in experiments monitoring ligand binding to agarose-immobilized thrombin. This mutually exclusive binding of either ligand can be explained by reciprocal, allosteric modulation of ligand affinity between the two exosites. Thus, not only do the two exosites differentially influence the active site, they also affect the binding properties of the opposing exosite.

Thrombin is a trypsin-like enzyme that plays a major role in hemostasis by regulating the procoagulant, anticoagulant, and fibrinolytic pathways (1,2). A distinguishing feature of thrombin is the presence of two positively charged patches found on opposite sides of the thrombin molecule (3). These regions, termed anion binding exosites 1 and 2, contribute to the specificity of thrombin by serving as binding sites for substrates, cofactors, and other ligands that modulate thrombin activity (4).
Exosite 1 was initially recognized as the fibrinogen binding site, but more recent studies indicate that it also binds hirudin, heparin cofactor II, and the thrombin receptor (5)(6)(7)(8)(9). The docking interaction at exosite 1, used by both substrates and inhibitors, precedes the reaction at the active site. The importance of exosite 1 is demonstrated by the reduced reactivity of ␥-thrombin, a proteolytic derivative of thrombin lacking exosite 1, with fibrinogen, hirudin, and heparin cofactor II (9,10). A second role of exosite 1, revealed by crystallographic and fluorescence studies with hirudin or thrombomodulin, is to confer structural changes in the active site environment that facilitate subsequent binding interactions (11)(12)(13)(14)(15). In the case of hirudin, this serves to optimize alignment of the amino-terminal inhibitory domain with the active site of thrombin (11,15,16). With thrombomodulin, the allosteric changes serve to alter the substrate specificity of thrombin. This converts thrombin from a procoagulant to a potent anticoagulant enzyme by virtue of its ability to activate protein C (17). Thus, exosite 1 serves both as a substrate/inhibitor binding site and an allosteric regulatory site. Exosite 2, which has a stronger positive electrostatic field than exosite 1, is known mainly as a glycosaminoglycan binding site (3,16,18). Heparin binds to this site to catalyze antithrombin-mediated inhibition of thrombin (19). The chondroitin sulfate moiety of thrombomodulin also binds exosite 2, thereby increasing the affinity of the interaction (13,14,20). In addition to binding glycosaminoglycans, exosite 2, together with exosite 1, appears to be important in the interaction of thrombin with factors V and VIII, key cofactors in coagulation (21,22). Prothrombin fragment 2 (F2), 1 a kringle-containing activation fragment of prothrombin, binds to exosite 2 on thrombin (18). This interaction induces conformational changes at the active site, influences the regulation of thrombin activity by calcium, and reduces the rate of thrombin inhibition by antithrombin (20,(23)(24)(25). Like exosite 1, therefore, ligand binding to exosite 2 also can induce allosteric changes in the active site that modulate thrombin activity.
The allosteric changes evoked by ligand binding to exosite 1 or 2 may affect different regions of the active site. Binding to exosite 1 is thought to influence the autolysis loop near the south rim of the active site, whereas residues within exosite 2 form part of the S2 subsite of the substrate binding cleft (3,4). This raises the possibility that ligand binding to one exosite may elicit different effects on the structure and function of the active site than ligand binding to the other exosite. In addition to their effects on the active site, ligand binding to one exosite may also alter the affinity of the opposing exosite for its ligands. To explore these possibilities we used a synthetic peptide encompassing the carboxyl-terminal half of human F2 as a ligand for exosite 2 (sF2) and a synthetic peptide analogue of the carboxyl terminus of hirudin (hirudin-(54 -65)) as a ligand for exosite 1. Herein we report that ligand binding to one exosite directly influences the binding properties of the other, revealing an allosteric linkage between the two regulatory sites on thrombin. We also demonstrate that the two ligands have very different effects on the structure and function of the active site.

Proteins
Active Site Blocked Thrombin-Thrombin (1-3 mg/ml) was incubated with a 2-fold molar excess of FPRCK or fluorescein-FPRCK at 23°C. After 15 min, the residual chromogenic activity of 200 nM FPR-thrombin with 0.2 mM tGPR-pNA was determined. Additional equimolar aliquots were added until no chromogenic activity was detected. The sample was dialyzed against 500 ml of 20 mM Tris-HCl, pH 7.4, 150 mM NaCl (TBS) at 4°C, with three changes of buffer over 18 h. After dialysis, the sample was recovered and centrifuged in a microcentrifuge for 2 min. The concentration was determined by measuring the absorbance using ⑀ 0.1% 280 of 1.8 (28) after correction for light scatter at 320 nm using the relationship A 280 corr ϭ A 280 Ϫ 1.7 ϫ A 320 (26). Residual chromogenic activity of 200 nM FPR-thrombin was monitored as above. In addition, the presence of residual FPRCK was determined by incubating 10 l of 5-10 M FPR-thrombin with 10 l of 100 nM active thrombin for 10 min. The chromogenic activity with 0.2 mM tGPR-pNA was then compared with that of a control lacking FPR-thrombin.
F2-F2 was isolated as a product of a prothrombin activation reaction as described by Stevens and Nesheim (29). Human prothrombin (20 mg) was activated in 20 ml of TBS containing 2 mM CaCl 2 , 23 M phosphatidylcholine:phosphatidylserine, 4 nM factor Va, and 5 nM factor Xa. After 2 h the reaction was stopped by the addition of EDTA to 5 mM. The sample was diluted with an equal volume of H 2 O and applied to a sulfopropyl C-50 (5 ml) column to adsorb the thrombin. Unbound protein was applied to a benzamidine-agarose column (2 ml) to remove residual factor Xa and thrombin. The flow-through was made 1 M in FPRCK and, after the addition of 0.025 volume of 20 ϫ concentrated TBS, the sample was applied to a Mono Q (HR 5/5) column on an FPLC System (Pharmacia Biotech Inc.). After washing the sample with 20 ml of TBS, a 30-ml linear gradient from 150 to 500 mM NaCl in 20 mM Tris-HCl, pH 7.4, was run at a flow rate of 1 ml/min. Fractions of 1 ml were collected and the A 280 was determined. Typically, the elution profile demonstrated base-line separation of two peaks that were identified as prothrombin fragments 1 and 2 by comparison of their mobilities with molecular weight standards on SDS-polyacrylamide gel electrophoresis (30). Consistent with previous reports (20,24), two closely migrating bands were observed for F2. The concentration of F2 was determined using ⑀ 0.1% 280 of 1.1 (28). Synthetic F2-A 54-amino acid peptide corresponding to residues 218 -271 of human prothrombin (sF2) was synthesized by Chiron Mimotopes P/L (Victoria, Australia). Cys 231 and Cys 248 were replaced by Ser whereas Lys 236 was changed to Gln, the residue present in bovine F2 (18). A scrambled variant of sF2, retaining the original positions of the two Cys residues (Cys 219 and Cys 243 ), also was synthesized. Peptides were synthesized on polyethylene pins that had been radiationgrafted with hydroxyethylmethacrylic acid as described (31). The crude peptides were lyophilized after deprotection, cleaved from the solid support, and purified by reverse phase high pressure liquid chromatography. To cyclize the peptides, they were dissolved at a final concentration of 0.2 mg/ml in 50 mM ammonium bicarbonate, pH 8.05, 30% acetonitrile, covered in aluminum foil, mixed on an orbital shaker for 2 days at 23°C, and then lyophilized. The cyclized peptides were purified by preparative reverse phase high pressure liquid chromatography. The peptides displayed Ͼ90% purity by high pressure liquid chromatography and yielded the expected composition on quantitative amino acid analysis, and the identities were confirmed by mass spectrometry. Peptides were reconstituted in TBS to a concentration of about 10 mg/ml, and 1-l aliquots of 1 N NaOH were added until a neutral pH was obtained.
Fluorescein-sF2-The sF2 peptide, devoid of Lys residues, was labeled at the amino terminus with FITC. Five 4-l aliquots of 10 mg/ml FITC in dry Me 2 SO were added to 0.5 mg of sF2 in 300 l of 0.1 M phosphate, pH 7.5. The sample was kept in the dark for 4 h at room temperature prior to quenching the reaction with 15 l of 1.0 M NH 4 Cl, pH 7.0. After fractionation on Sephadex G-10 (1 ϫ 18 cm) and collecting 1-ml fractions, the fluorescent peptide was detected using a hand-held fluorescent lamp (302 nm). The concentration of the peptide was determined as described (12).
Mansyl-sF2-One mg of sF2 in 300 l of 0.1 M sodium phosphate, pH 7.4, was labeled with mansyl by the addition of 20 l of 50 mg/ml mansyl chloride in 60% methanol, 40% acetone. The reaction was left overnight and then fractionated as described above. The labeled peptide was located by A 324 and the concentration was determined using ⑀ 1 mM 324 of 24 (Molecular Probes).

Binding Studies
sF2 Displacement by F2-A 1 ϫ 1 cm quartz cuvette containing 100 nM fluorescein-sF2 in 2 ml of TBS was excited at 492 nm (5 nm slit) and continuously monitored at 522 nm (5 nm slit) in time drive using a Perkin-Elmer LS50B luminescence spectrometer. The contents of the cuvette were stirred with a micro stir bar and maintained at 25°C using a circulating water bath. The fluorescence intensity (I) of 280 nM FPR-thrombin was measured before and after the addition of 2-5-l aliquots of 92 M F2 (containing 100 nM fluorescein-sF2), allowing the signal to stabilize prior to each addition. Titration was continued until there was no change in the sample I with subsequent additions. After the experiment, intensity values were read from the time drive profile. I/I o values were calculated and plotted versus the F2 concentration. The data were analyzed by nonlinear regression of the binding isotherm Equation 1 (32).
using TableCurve (Jandel, San Rafael, CA) to solve for ␣, the maximum fluorescence change, and K d , the dissociation constant, given L, the ligand concentration, and P o , the concentration of the target protein, and assuming a stoichiometry of 1. sF2 and Hirudin-(54 -65) Binding to Thrombin-To compare the affinities of sF2 and hirudin-(54 -65) for ␣and ␥-thrombin, fluorescein-FPR derivatives of the enzymes were used. For sF2 binding studies, 2 ml of 150 nM fluorescein-FPR ␣or ␥-thrombin was added to a 1 ϫ 1-cm cuvette, and fluorescence was monitored as described above. The samples were then titrated with aliquots of a solution of 5 mM sF2 (containing 150 nM appropriate fluorescein-FPR-thrombin to prevent probe dilution). After the titration, the time drive profile was analyzed, and the binding parameters were determined as above. For hirudin-(54 -65) binding, the fluorescein-FPR-thrombin concentration was 250 nM and the stock hirudin-(54 -65) was 100 M.
Reciprocal Exosite Binding-Reciprocal exosite binding was investigated by observing the influence of increasing ligand occupation at one site on the amount of ligand bound at the other exosite. In the first case, the fluorescence of 10 nM fluorescein-hirudin-(54 -65) was monitored before and after the addition of FPR-thrombin to 25 nM, and the sample was then titrated with 3.8 mM sF2 (containing 10 nM fluoresceinhirudin-(54 -65)) and changes in fluorescence were monitored in time drive. The intensity values and the K d were determined as described above. In the second case, the fluorescence of 100 nM fluorescein-sF2 was monitored before and after the addition of FPR-thrombin to 280 nM, and the sample was then titrated with 100 M hirudin-(54 -65) (containing 100 nM fluorescein-sF2).
Analysis of Binding at Both Exosites-Rhodamine-hirudin-(54 -65) and fluorescein-sF2 were used to monitor simultaneous binding to both exosites. A 2-ml sample containing 150 nM FPR-thrombin and 100 nM fluorescein-sF2 was titrated with 18.3 M rhodamine-hirudin-(54 -65). The fluorescein and rhodamine fluorescence of the sample was monitored at ex of 492 nm and em of 522 nm for fluorescein and ex of 550 nm and em of 575 nm for rhodamine at each step in the titration. A blank titration lacking FPR-thrombin was used to correct the data for spectral overlap of the two probes. The displacement of fluorescein-sF2 was quantified by analysis of the I/I o values as described above. Binding of rhodamine-hirudin-(54 -65) to thrombin was determined by calculating the difference in intensity of rhodamine-hirudin-(54 -65) in the absence and presence of FPR-thrombin, which was taken to represent the thrombin-bound fraction of rhodamine-hirudin-(54 -65). Binding of both sF2 and hirudin-(54 -65) was analyzed by nonlinear regression as described above. To graphically display displacement of fluorescein-sF2 by rhodamine-hirudin-(54 -65) titration, the amount of fluorescein-sF2 bound was converted to a percent of that initially bound to thrombin, prior to rhodamine-hirudin-(54 -65) addition. To display rhodaminehirudin-(54 -65) binding, the amount bound was calculated as the percent of maximal rhodamine-hirudin-(54 -65) binding at saturation.
Thrombin-Agarose Binding Studies-Ligand binding to immobilized thrombin was used to supplement the fluorescence studies. FITC-labeled peptides could not be used in these experiments because high salt concentrations needed in the final wash step influenced the fluorescence intensity. Instead, the effect of hirudin-(54 -65) on sF2 binding to thrombin-agarose was assessed by first binding mansyl-sF2 to immobilized thrombin and then monitoring mansyl-sF2 fluorescence in the suspension supernatant as the sample was titrated with hirudin-(54 -65). To accomplish this, the fluorescence of 500 l of mansyl-sF2 was first determined ( ex of 324 nm and em of 463 nm). The material was then mixed with 40 l of thrombin-agarose (ϳ2 M) for 15 min. The suspension was centrifuged for 2 min, the supernatant removed, and the agarose washed two times with TBS to remove unbound mansyl-sF2. The agarose pellet was suspended in 500 l of TBS, re-centrifuged, and the I of the supernatant determined. The supernatant was returned to the tube containing thrombin-agarose and was then titrated with aliquots of hirudin-(54 -65). The fluorescence of the supernatant was determined after each addition of hirudin-(54 -65). At the end of the titration, NaCl was added to 2 M to observe the total amount of mansyl-sF2 that was bound to the thrombin-agarose. In the complementary experiment, the effect of sF2 on 125 I-hirudin-(54 -65) binding to thrombin-agarose was assessed. To a 250-l suspension of thrombinagarose, 125 I-hirudin-(54 -65) was added to 200 nM and the tube was mixed for 5 min. The sample was centrifuged for 2 min and the agarose washed two times with TBS. The agarose was suspended in 250 l of TBS, and 50 l of supernatant was counted in a ␥-counter. The sample was returned to the suspension, and an aliquot of 25 mM fluorescein-sF2 was added. The process was repeated throughout the sF2 titration. The percent of 125 I-hirudin-(54 -65) bound was calculated relative to the input hirudin-(54 -65), and the K d was determined by nonlinear regression analysis as described above with the exception that the fraction of labeled ligand displaced was used in place of I/I o .
Chromogenic Activity-The chromogenic activity of 20 nM thrombin in TBS was determined with 0.5 mM various thrombin substrates in the presence of 80 M sF2 or 10 M hirudin- (54 -65). The rates of substrate hydrolysis, in A 405 /min, were determined and related to the activities of thrombin in the absence of hirudin-(54 -65) or sF2.

RESULTS
To monitor binding interactions at exosite 2, we prepared a synthetic peptide corresponding to a portion of F2. The 54amino acid peptide (sF2) is derived from residues 218 to 271 of human prothrombin (residues 63-116 of F2) in which the Cys 219 -Cys 243 disulfide bond is maintained, whereas Cys 231 and Cys 248 have been replaced by Ser. In addition, Lys 236 is changed to Gln, the residue found in bovine F2, to facilitate specific FITC labeling at the amino terminus. This peptide corresponds to the inner kringle loop and the contiguous 23residue acidic carboxyl-terminal connecting peptide of F2. To confirm its similarity to F2, binding studies were performed using a FITC derivative, fluorescein-sF2, and FPR-thrombin (Fig. 1). Addition of FPR-thrombin to fluorescein-sF2 resulted in a 1.6 Ϯ 0.2% increase (mean Ϯ S.D. of six determinations) in fluorescence intensity (I). When the sample was then titrated with plasma-derived F2, the I returned to slightly beyond the base-line value corresponding to unbound fluorescein-sF2. This indicates that sF2 and plasma-derived F2 bind to the same site on thrombin. Analysis of fluorescein-sF2 displacement by F2 revealed a K d value of 6.8 M, comparable to the value of 4 M obtained in direct binding studies with F2 (not shown; see Ref. 24). To further confirm the specificity of sF2 binding, FPRthrombin was added to a FITC-labeled scrambled homologue of sF2, containing an intact disulfide bond. There was no change in fluorescence of fluorescein in the presence of FPR-thrombin suggesting that there was no binding of the scrambled homologue to thrombin.
To assess interactions at exosite 1, FITC-hirudin-(54 -65) was titrated with FPR-thrombin. The experiment yielded a ␥-thrombin was performed to identify the sF2 binding site on thrombin. SF2 binding to both ␣and ␥-fluorescein-FPR-thrombin was characterized by a saturable decrease in I ( Fig. 2A), and K d values of 46 and 20 M were calculated for sF2 interaction with ␣and ␥-FPR-thrombin, respectively. Thus, sF2 binds to a site found on both ␣and ␥-thrombin. However, the sF2 peptide has about a 10-fold lower affinity for thrombin than plasma-derived F2 (K d values of 46 versus 4 M, respectively). Binding studies with a probe for exosite 1 also were performed with fluorescein-FPR derivatives of ␣and ␥-thrombin. As a ligand for exosite 1, a 12-amino acid peptide corresponding to a carboxyl-terminal segment of hirudin, hirudin-(54 -65), was used. This peptide binds to thrombin with an affinity similar to that of hirudin-(53-64), a peptide also known as Hirugen (7,33). Hirudin-(54 -65) binds to FPR-␣-thrombin with high affinity (K d of 62 nM and I/I o of 1.07, Fig. 2B) and shows no binding to FPR-␥-thrombin. These results are consistent with the concept that hirudin-(54 -65) binds to exosite 1, whereas sF2 binds to exosite 2, in agreement with the known specificity of exosites 1 and 2. Furthermore, the FITC derivatives of hirudin-(54 -65) and sF2 each report binding interactions through changes in I revealing that they are responsive probes that can be used as specific ligands for thrombin exosites 1 and 2, respectively.
The binding studies with the fluorescein-FPR derivatives of thrombin ( Fig. 2) suggest that ligand binding to either exosite causes conformational changes at the active site. This is supported by numerous studies demonstrating alterations in chromogenic activity of thrombin in the presence of hirudin-(54 -65) or F2 (7,12,15,34). In the current study, the influence of these ligands on thrombin chromogenic activity was compared directly (Fig. 3). Hirudin-(54 -65) and sF2 produced opposing effects on thrombin chromogenic activity with S-2238, Chz-Xa, Chz-tPA, S-2251, and S-2444. Increased activity was induced by sF2 on two of these substrates and by hirudin-(54 -65) on three substrates. With tGPR-pNA, the two ligands produced similar effects, whereas activity with S-2222 was promoted more by hirudin-(54 -65) than by sF2. Although these data confirm previous reports that ligand binding to either exosite 1 or 2 affects thrombin chromogenic activity, they also demonstrate that the active site can be differentially regulated depending on the exosite specificity of the ligand.
Interactions between Exosites-To determine whether the two exosites are allosterically linked, the fluorescence of an FITC-labeled ligand bound to FPR-thrombin was monitored throughout a titration with an unlabeled ligand specific for the opposing exosite. In Fig. 4A, the influence of sF2 on thrombinbound fluorescein-hirudin-(54 -65) was determined. Addition of 25 nM FPR-thrombin to 10 nM fluorescein-hirudin-(54 -65) resulted in a ϳ10% decrease in I. Subsequent titration of the sample with sF2 returned the I to the original value obtained with unbound fluorescein-hirudin-(54 -65). This indicates complete displacement of fluorescein-hirudin-(54 -65) from thrombin. The inset shows I/I o values for the sF2-induced displacement of thrombin-bound fluorescein-hirudin-(54 -65). The line represents nonlinear regression analysis of the sF2 binding data revealing a K i of 202 M. This corresponds to a 4-fold reduction in the affinity of thrombin for sF2 in the presence of fluorescein-hirudin-(54 -65). In the complementary experiment, thrombin-bound fluorescein-sF2 was titrated with hirudin-(54 -65) (Fig. 4B). The fluorescence of 100 nM fluorescein-sF2 increased 1.7% upon addition of 250 nM FPR-thrombin, an increase similar to that observed in Fig. 1. Subsequent   FIG. 2. Binding of sF2 and hirudin-(54 -65)  By having demonstrated that fluorescent derivatives of hirudin-(54 -65) and sF2 report binding interactions with thrombin, it was of interest to determine whether the two ligands could bind simultaneously. To accomplish this, rhodamine-hirudin-(54 -65) and fluorescein-sF2 were used. In-cubation of FPR-thrombin (150 nM) with 100 nM fluorescein-sF2 resulted in a 1.7% increase in I, as observed in Fig. 1. Subsequent stepwise titration with rhodamine-hirudin-(54 -65) resulted in a progressive decrease in the fluorescence of fluorescein-sF2 to a value approaching that obtained for unbound fluorescein-sF2. Concomitantly, the fluorescence of rhodamine-hirudin-(54 -65) increased in a saturable manner (not shown), as observed in control titrations performed in the absence of fluorescein-sF2. Fig. 5 shows the percent displacement of thrombin-bound fluorescein-sF2, determined by the fractional return to base-line fluorescence of fluorescein-sF2, as observed in Fig. 1. As well, the plot reveals the percent rhodamine-hirudin-(54 -65) bound to thrombin exosite 1, determined from the approach to saturation for the increase in rhodamine fluorescence. Each set of data was analyzed by nonlinear regression and revealed K d values of 180 and 300 nM for rhodamine-hirudin-(54 -65) binding and fluorescein-sF2 displacement, respectively. Therefore, as rhodamine-hirudin-(54 -65) bound to exosite 1 on thrombin there was concomitant displacement of fluorescein-sF2 from exosite 2. As noted above, the affinity of thrombin for hirudin-(54 -65) was reduced in the presence of sF2.
To exclude the possibility that the fluorescence data reflected environmental changes of the labeled ligand rather than true displacement, we used agarose-immobilized thrombin to examine the ability of the ligands for exosites 1 and 2 to displace each other. Binding to thrombin-agarose was assessed by measuring the free labeled ligand remaining in the supernatant after brief centrifugation. These measurements were performed throughout the titration of the sample with the other, unlabeled ligand. Mansyl-sF2 and 125 I-hirudin-(54 -65) were used in place of the fluorescein-labeled derivatives for this study because the final, high salt elution step altered the fluorescence intensity of the fluorescein. Thrombin was bound to the support via biotin-FPR such that all molecules were coupled through their active site, leaving the exosites free to bind their ligands. With mansyl-sF2, about 46% bound to a 500-l suspension of thrombin-agarose containing about 2 M thrombin. Titration of the sample with hirudin-(54 -65) resulted in displacement of the bound mansyl-sF2, as evidenced by almost

DISCUSSION
In this study we used specific ligands for exosites 1 and 2 on thrombin to demonstrate an allosteric linkage between the two exosites and to show that ligand binding to one exosite produces changes at the active site different from those induced by ligand binding to the other exosite. Hirudin-(54 -65), which has properties similar to hirudin-(53-64) (7,33), was used as a ligand for exosite 1 since there is crystallographic evidence of its specificity (11) and a fluorescein derivative has been used successfully in previous binding studies (12,20). As a probe for exosite 2, prothrombin fragment 2 (F2) was utilized because of its defined composition and the crystallographic evidence of its specificity for exosite 2 (18,24). To facilitate reproducible and unique fluorescent labeling, a synthetic peptide corresponding to the carboxyl-terminal half of human F2 (prothrombin residues 218 -271) was synthesized. Both probes displayed binding consistent with their expected specificities since hirudin-(54 -65) binds only to ␣-thrombin, whereas sF2 binds to both ␣and ␥-thrombin (Fig. 2). The integrity of exosite 2, but not exosite 1, in ␥-thrombin is consistent with its ability to bind heparin (10) but not hirudin (35).
Interactions between the Exosites and the Active Site-Previous studies have demonstrated that F2 (12,20,24,36) and hirudin-(53-64) (12,15,37) affect the structure and function of thrombin. In the present study, direct comparison reveals that these two ligands induce diametric changes in the active site of thrombin. Whereas sF2 caused a ϳ16% decrease in the intensity of fluorescein-FPR-thrombin, hirudin-(54 -65) induced a ϳ7% increase (Fig. 2). Furthermore, the two ligands also produced opposing effects on the activity of thrombin with five chromogenic substrates. The fluorescence results reveal changes in the environment of the fluorophore, which may be located up to 15-Å away from the catalytic serine residue (3,13). Since it is unlikely that the fluorophore is proximal to both exosites, these results suggest that the intensity changes are mediated by residues in or around the thrombin binding pocket and not by contact of the fluorophore with the ligand. The data with the chromogenic substrates also point to changes in the substrate binding pocket since hirudin-(53-64) has been observed to influence the K m of thrombin for low molecular weight substrates (7). Although the fluorescence and the functional data cannot be directly correlated, taken together they suggest that the active site is differentially modulated depending on which exosite is occupied. This points to the possibility that exosites 1 and 2 play distinct roles in regulating thrombin activity. This concept is supported by studies examining the role of exosite 2 in the activation of protein C by thrombin bound to thrombomodulin. Binding of chondroitin sulfate or F2 to exosite 2 or mutation of basic residues in exosite 2 reduces the rate of protein C activation by thrombin complexed with chondroitin sulfate-deficient thrombomodulin and alter the calcium dependence of the reaction (20,38). These data suggest that ligands for exosites 1 and 2 have different effects on the active site.
Crystallographic studies of the F2 or hirudin-(53-64) in complex with FPR-thrombin show no major changes in thrombin structure (11,18). However, the allosteric changes detected by solution-phase binding studies may reflect movement of flexible loops of thrombin that are poorly defined in the crystal structure (18,39). Two of these flexible loops reside in the vicinity of the active site and thus are good candidates for regulatory domains (3,11,18).
Interactions between the Exosites-Two independent experiments (Figs. 4 and 6) demonstrate displacement of ligand bound to one exosite of thrombin by titration with the ligand specific for the opposing exosite. These findings suggest reciprocal binding between exosites 1 and 2. When the binding is quantitatively determined, the affinity of thrombin for one ligand is reduced when the ligand for the other exosite is present suggesting that the apparent reciprocal binding between the exosites may more accurately be described as modulation of affinity of one exosite by ligand occupation of the other exosite. The allosteric linkage observed between the two exosites likely results from conformational changes associated with exosite occupation. Thus, the allosteric changes associated with ligand binding to either exosite are extensive since they not only involve the active site but also the other exosite.  6. Ligand binding to thrombin-agarose. A, a 500-l suspension of thrombin-agarose was incubated with 0.5 M mansyl-sF2. After the sample was washed twice, the supernatant was monitored for mansyl fluorescence and returned to the agarose. Aliquots of hirudin-(54 -65) were added, and the fluorescence of the supernatant was determined after each addition. The total amount of bound mansyl-sF2, determined after the addition of NaCl to 2 M, was used to calculate the percent bound at each step in the titration. The line shows the nonlinear regression analysis of the data. Hirudin-(54 -65) bound to thrombin-agarose, as monitored by mansyl-sF2 displacement, with a K i of 2 M. B, 125 I-hirudin-(54 -65), bound to thrombin-agarose as described in A, was titrated with fluorescein-sF2. The radioactivity in the supernatant was determined after each addition. The total amount of 125 Ihirudin-(54 -65) was determined and used to calculate the percent of hirudin-(54 -65) bound. The data were analyzed by nonlinear regression analysis (line), yielding a K i of 35 M.
Our findings are supported by the work of other investigators. F2 has been reported to reduce the affinity of thrombin for hirudin-(53-64) and thrombomodulin (12,20), to reduce thrombin clotting activity (36), and to inhibit thrombomodulin-dependent protein C activation (36,38). Since exosite 1 is involved in all these reactions, these results suggest that F2 has the capacity to modulate the function of exosite 1. Ligands other than F2 can also influence exosite 1 function. For example, heparin reduces the rate at which hirudin inhibits thrombin (40), and thrombin binding to heparin-agarose is abrogated by hirudin (41). These results suggest that the allosteric linkage between the exosites constitutes a general regulatory mechanism that is not solely confined to hirudin-(54 -65) and F2.
Although the data presented herein suggest that only one exosite can be occupied at a time, there are examples of simultaneous occupation of both exosites. Thrombomodulin binds to exosite 1 via its growth factor domains and to exosite 2 via the chondroitin sulfate moiety (20). Both exosites contribute to binding because intact thrombomodulin binds thrombin more tightly than chondroitin sulfate-deficient thrombomodulin (12). Another example of concomitant occupation of both exosites is the ternary thrombin-fibrin-heparin complex in which fibrin binds to exosite 1 and heparin to exosite 2 of thrombin (42). In both of these complexes, the glycosaminoglycan serves to augment affinity for thrombin by binding to exosite 2 as well as to the ligand occupying exosite 1. This demonstrates another mechanism by which thrombin utilizes exosites 1 and 2 and reveals that different ligands may play distinct roles in thrombin regulation.
Conclusions-The current study provides three major observations regarding the role of exosite 2 in the regulation of thrombin. The first observation is that exosite 2 serves as an allosteric modulator of thrombin structure and function. This highlights its similarity to exosite 1, where considerable support exists for an allosteric linkage between exosite 1 and the active site (7, 8, 12-15, 43, 44). The second observation is that ligands for exosites 1 and 2 have different effects on the active site of thrombin, as judged by both fluorescence and functional studies. This suggests that the two exosites play distinct regulatory roles, possibly by invoking unique conformational changes. The third observation is that there is direct allosteric linkage between the two exosites. Conformational linkage between the two exosites and between the exosites and the active site could be exploited by thrombin to evoke ligand-specific responses. These findings point to a greater complexity in modulation of thrombin function and suggest new avenues of physiological and pharmacological regulation of thrombin activity.