Conformational Changes in Thrombin When Complexed by Serpins*

Thrombin possesses two positively charged surface domains, termed exosites, that orient substrates and inhibitors for reaction with the enzyme. Because the exosites also allosterically modulate thrombin’s activity, we set out to determine whether the structure or function of the exosites changes when thrombin forms complexes with antithrombin, heparin cofactor II, or (cid:1) 1 - antitrypsin (M358R), serpins that utilize both, one, or neither of the exosites, respectively. Using a hirudin-derived peptide to probe the integrity of exosite 1, no binding was detected when thrombin was complexed with heparin cofactor II or (cid:1) 1 -antitrypsin (M358R), and the peptide exhibited a 55-fold lower affinity for the thrombin-antithrombin complex than for thrombin. Bound peptide or HD-1, an exosite 1-binding DNA aptamer, was displaced from thrombin by each of the three serpins. Thrombin binding to fibrin also was ab-rogated when the enzyme was complexed with serpins. These data reveal that, regardless of the initial mode of interaction, the function of exosite 1 is lost when thrombin is complexed by serpins. In contrast, the integrity of exosite 2 is largely retained when thrombin is complexed by serpins, because interaction with heparin or an exosite 2-directed DNA aptamer was only modestly altered. The disorganization of exosite 1 that occurs when thrombin is complexed by serpins densitometry

Thrombin activates platelets, converts fibrinogen to fibrin, and amplifies its own generation by activating factors V, VIII, and XI, highlighting its central role in coagulation. To prevent excessive clotting, the activity of thrombin must be tightly regulated (1). Down-regulation of thrombin activity is mediated by two serpins, antithrombin and heparin cofactor II, which form 1:1 stoichiometric complexes with thrombin (2). Another inhibitory mechanism involves thrombomodulin, a thrombin receptor found on the surface of vascular endothelial cells. Once bound to thrombomodulin, thrombin undergoes a specificity change that converts it from a procoagulant enzyme into one that initiates an anticoagulant pathway by activating protein C (3).
Selective interactions of thrombin with its substrates, cofactors, and inhibitors reflect structural characteristics unique to thrombin. These structural features include two surface loops that limit access to the active site by protruding over the active site cleft. In addition, thrombin possesses two positively charged domains, termed exosites, located on opposite poles of the thrombin molecule (4 -6). The principal role of exosite 1 is to bind substrates, cofactors, and inhibitors and orient them for optimal interaction with the active site. This exosite, which also is known as the fibrinogen recognition site (7), interacts with negatively charged domains on fibrinogen, the thrombin receptor, hirudin, thrombomodulin, and heparin cofactor II. In contrast, the other exosite, which is designated exosite 2, binds heparin, dermatan sulfate, and chondroitin sulfate, glycosaminoglycans that promote thrombin's interactions with serine protease inhibitors and thrombomodulin.
In addition to mediating thrombin's interactions with its substrates and inhibitors, the exosites also modulate thrombin's activity. By binding to exosite 1, in a concerted interaction that also involves exosite 2, thrombomodulin abolishes the procoagulant activity of thrombin by hindering its reaction with fibrinogen, factors V and VIII, and the thrombin receptor (8 -10). Coincidentally, thrombin's interaction with thrombomodulin induces allosteric changes at the active site of the enzyme (11,12) that promote its ability to activate protein C and the procarboxypeptidase B-like enzyme, thrombin-activable fibrinolysis inhibitor (TAFI) 1 (13). Interactions at exosite 2 also modulate the structure and function of the active site of thrombin (14,15). Furthermore, ligand binding to either exosite can influence the other exosite in a reciprocal fashion (14).
Exosite 2 contributes to the inactivation of thrombin by antithrombin and heparin cofactor II by serving as a heparinbinding site, thereby promoting heparin-mediated bridging of the enzyme to the inhibitor (16,17). Exosite 1 also plays an important role in the interaction of thrombin with heparin cofactor II. Binding of heparin or dermatan sulfate to heparin cofactor II disrupts the intramolecular interaction of the NH 2terminal domain, allowing it to bridge to exosite 1 on thrombin (17,18). The exosites are not essential for thrombin inactivation, however, because ␣ 1 -antitrypsin Pittsburgh, a naturally occurring variant with its reactive site P 1 residue, Met-358, replaced with Arg (M358R), has no requirement for interaction with either exosite (19,20).
Because antithrombin, heparin cofactor II and ␣ 1 -antitrypsin M358R have different requirements for access to the exosites on thrombin, we set out to elucidate the structural integrity of exosites 1 and 2 when thrombin is complexed by these serpins. This study extends a previous report that exosite 1 is no longer accessible to a hirudin-derived peptide once thrombin is complexed by antithrombin (21).
Preparation of 125 I-Thrombin-1 mCi of Na 125 I (Perkin-Elmer Life Sciences, Markham, Ontario) was incubated with one IODO-BEAD (Pierce Chemical Co., Rockford, IL) in phosphate-buffered saline for 10 min. The bead was removed, and 200 g of thrombin was added to the buffer containing Na 125 I. After 10-min incubation, the reaction was stopped by the addition of sodium metabisulfite to 40 mM, and the sample was applied to a PD-10 column equilibrated with 20 mM Tris-HCl, 150 mM NaCl, pH 7.4 (TS), containing 0.01% Tween 20 and 0.6% polyethylene glycol 8000. The column was eluted under gravity, and 0.5-ml fractions were collected. Based on absorbance at 280 nm and radioactivity, fractions containing 125 I-thrombin were identified and pooled, and the concentration was determined spectrophotometrically (14). The specific activity of thrombin radiolabeled in this fashion was 5 ϫ 10 8 cpm/mg.
Preparation of Thrombin-Serpin Complexes-Complexes of thrombin with antithrombin or heparin cofactor II were prepared by incubating a 10 -20% molar excess of inhibitor over enzyme in the absence or presence of 1-5 nM heparin. Complexes of thrombin with ␣ 1 -antitrypsin M358R were prepared in the absence of heparin. After 60-min incubation of thrombin with serpin, no residual thrombin activity was detected using the thrombin-directed chromogenic substrate, tGPR-pNA. Aliquots of reaction mixtures also were subjected to PAGE analysis on 4 -15% polyacrylamide gels (Ready Gels, Bio-Rad, Mississauga, Ontario) in the presence of SDS to confirm the formation of thrombininhibitor complexes (25).
Complexes of 125 I-thrombin with FPRck or serpins were prepared in the absence of glycosaminoglycans by incubating 2-5 M 125 I-thrombin with 2-to 5-fold molar excess FPRck, antithrombin, heparin cofactor II, or ␣ 1 -antitrypsin M358R. After incubation for 60 min at 23°C, the residual activity of thrombin with tGPR-pNA was determined. The complex with heparin cofactor II showed 98% inhibition of chromogenic activity, whereas the others demonstrated greater than 99.5% inhibition. Concentrations of thrombin-serpin complexes were based on starting concentration of thrombin in each reaction. SDS-PAGE analysis followed by autoradiography revealed reduced mobility of 125 I-thrombin in the presence of serpins, but not FPRck (Fig. 1).
Activation of Fluorescein-Prothrombin-Lys residues in exosites 1 and 2 of thrombin are sensitive to chemical modification (26). Consequently, fluorescein-thrombin was prepared by activating fluoresceinprothrombin, with the expectation that the incompletely exposed exosites on prothrombin (27,28) would be protected from modification. Prothrombin was labeled with 20-fold molar excess of FITC in 100 mM sodium phosphate buffer, pH 9.0, for 1 h in the dark. Unincorporated FITC was removed by chromatography on a PD-10 column. Activation of 5 mg of fluorescein-prothrombin was carried out in the presence of 3 nM factor Xa, 4 nM factor Va, 20 M phosphatidylcholine/phosphatidylserine vesicles, and 2 mM CaCl 2 until the activity of aliquots with tGPR-pNA reached a plateau. The reaction was made 5 mM in EDTA and applied to a Q-Sepharose column. Fluorescein-thrombin, which eluted in the flow-through fraction, was precipitated with 80% ammonium sulfate, resuspended in TS, and dialyzed against TS overnight. By protein absorbance and chromogenic activity, the concentration was determined to be 16 M. Fluorescein-thrombin, prepared in this manner, demonstrated clotting activity, chromogenic activity, and inhibition by antithrombin comparable to native thrombin.

Titration of Fluorescein-Hirudin 54 -65 with Thrombin-Serpin
Complexes-Fluorescein-hirudin 54 -65 , prepared as described previously (14), was added to a quartz cuvette at a concentration of 10 nM in 900 l of TS. The sample was stirred with a mini-stirbar and maintained at 23°C using a circulating water bath. Fluorescence of the sample (I o ) was monitored at 1-s intervals using a Perkin-Elmer LS 50B luminescence spectrometer at excitation and emission wavelengths of 492 and 535 nm, respectively, with a 515-nm filter in the emission beam and slit widths of 15 and 20 nm, respectively. The sample was then titrated with active thrombin or thrombin-serpin complexes. After addition of each aliquot, the signal was allowed to stabilize for 1 min. After the titration, fluorescence intensity (I) values were obtained from the time-drive profile. Plots of I/I o versus titrant concentration were analyzed by non-linear regression of the binding isotherm equation, as described previously (14).
Displacement of Fluorescein-Hirudin 54 -65 from Thrombin in the Presence of Serpins-The fluorescence of 10 nM fluorescein-hirudin 54 -65 was monitored as described above. When heparin was added to the cuvette to a concentration of 400 nM, there was no change in fluorescence intensity. Subsequent addition of 200 nM thrombin produced a ϳ15% decrease in fluorescence intensity. Antithrombin, heparin cofactor II, or ␣ 1 -antitrypsin M358R was then added to a concentration of 400 nM, and fluorescence was monitored until the signal reached a plateau.
Displacement of Fluorescein-HD-1 from Thrombin in the Presence of Serpins-The affinity of fluorescein-HD-1 for thrombin was determined by titrating 20 nM fluorescein-HD-1 in 2 ml of TS with active thrombin. Fluorescence was monitored, and the plot of I/I o versus thrombin concentration was analyzed as described above to determine K d . The effect of serpin addition on the binding of fluorescein-HD-1 to thrombin was then examined in the absence or presence of 4 nM heparin. Addition of 100 nM thrombin to 20 nM fluorescein-HD-1 resulted in a ϳ4% increase in fluorescence intensity. Once the signal equilibrated, antithrombin, heparin cofactor II, or ␣ 1 -antitrypsin M358R was added to 500 nM, and the fluorescence monitored until the signal reached a plateau.
Binding of HD-22 to Fluorescein-Thrombin-Fluorescein-thrombin or fluorescein-thrombin-serpin complexes were titrated with HD-22 for determination of K d . A 2-ml sample containing 100 nM fluorescein- thrombin was incubated with 200 nM ␣ 1 -antitrypsin M358R, 1 M antithrombin, or 1 M heparin cofactor II for more than 15 min, a time sufficient to achieve complete inhibition of thrombin activity. Fluorescence was monitored while fluorescein-thrombin or fluorescein-thrombin-serpin complexes were then titrated with HD-22 and K d values were determined as described above.
Binding of Pentasaccharide to Antithrombin and Thrombin-Antithrombin Complex-Binding of synthetic pentasaccharide to antithrombin was assessed by monitoring the intrinsic protein fluorescence of antithrombin in the absence and presence of pentasaccharide. Antithrombin at 100 nM, in the absence or presence of 300 nM thrombin, was equilibrated in 800 l of TS in a cuvette for 1 h. The sample was monitored with excitation and emission wavelengths of 280 and 340 nm, respectively, with slitwidths of 5 nm and a 290-nm cutoff filter in the emission beam. Aliquots of pentasaccharide were added, and the fluorescence was monitored. I/I o values were determined and K d values calculated as described above.
Heparin-Sepharose Chromatography-Heparin-Sepharose affinity chromatography was used for qualitative assessment of the affinities of thrombin-and thrombin-serpin complexes for heparin. Samples containing 1-2 g of 125 I-thrombin or 125 I-thrombin-serpin complexes were loaded onto a 0.6 ϫ 10 cm column containing 4 ml of heparin-Sepharose equilibrated with 10 mM Tris-HCl, pH 7.4, at a flow rate of 1 ml/min using a Beckman System Gold high performance liquid chromatography system. After washing, a 50-min linear gradient to 1 or 2 M NaCl was initiated. Fractions (2 ml) were collected and counted for radioactivity. In supplemental experiments, unlabeled thrombin or thrombinserpin complexes (50 -400 g) were subjected to chromatography, and the elution was monitored at 280 nm with an online UV detector. Fractions of 1 ml were collected, and peak protein-containing fractions were lyophilized, reconstituted in water, and subjected to SDS-PAGE analysis. Gels were stained with Fast Stain (Zoion Research, Shrewsbury, MA), and the bands were quantified by densitometry with an ImageMaster VDS (Amersham Pharmacia Biotech).
Immunoprecipitation of Thrombin-Serpin Complexes-Autoantibody D, an IgG that inhibits clotting, binds to exosite 1 on thrombin (29). The antibody was affinity-purified from a crude IgG fraction on a thrombinagarose column, prepared with biotin-FPRck and streptavidin-agarose (14). The IgG fraction was applied to the column and the flow-through material reapplied three times. After the column was washed with TS, bound IgG was eluted with Gentle Elution Buffer (Pierce). The eluate was dialyzed versus TS and concentrated by ultrafiltration with a Centricon 30 (Amicon Corp., Beverly, MA). For binding studies, 0.5 M 125 I-thrombin or preformed 125 I-thrombin-serpin complexes was incubated with 2.4 M autoantibody D for 1 h. An equal volume of protein G-agarose was added to the tubes, and the samples were mixed for 15 min on an end-over-end rotator. The samples were centrifuged, and the supernatants were removed and counted for radioactivity. The agarose was washed two times with TS, mixed with an equal volume of SDS gel sample buffer, boiled, and aliquots were subjected to SDS-PAGE. Autoradiographs of the dried gels were used to identify radioactive bands, which were excised and counted in a ␥-counter. The amount of material immunoprecipitated by autoantibody D/protein G-agarose was calculated as a fraction of the radioactivity of aliquots of the reactions taken prior to autoantibody D addition.
Binding of Thrombin-Serpin Complexes to Fibrin-Complexes of 125 I-thrombin with serpins or FPRck were incubated in microcentrifuge tubes at a concentration of 50 nM in the presence of 0 -10 M fibrinogen and 10 mM CaCl 2 . Atroxin was added to each sample to a final concentration of 5% (v/v), and samples were incubated for 60 min at 23°C. Resultant fibrin was compacted by centrifugation at 13,000 ϫ g for 10 min, and duplicate 30-l aliquots of the supernatant were counted for radioactivity. Plots of concentration of 125 I-thrombin-complexes bound versus the fibrin concentration were subjected to rectangular hyperbola analysis to calculate K d values (30).

Integrity of Exosite 1 on Thrombin Complexed by Serpins-
To determine whether the integrity of exosite 1 on thrombin is compromised when the enzyme is complexed by serpins, we first compared the affinity of hirudin 54 -65 , a peptide that binds exclusively to exosite 1 on thrombin, for active thrombin with that for thrombin complexed by the various serpins. Titration of a fluorescein derivative of hirudin 54 -65 with thrombin produced a saturable decrease in fluorescence intensity (Fig. 2). Fluorescein-hirudin 54 -65 bound thrombin with a K d of 41 nM, a value consistent with that reported previously (15,21). Although saturable binding also was observed when fluorescein-hirudin 54 -65 was titrated with thrombin-antithrombin complexes, the affinity of the interaction was 55-fold weaker than that with thrombin (K d value of 2270 nM). Bock et al. (21) also demonstrated reduced affinity of fluorescein-hirudin 54 -65 for the thrombin-antithrombin complex. In their study, saturable binding of fluoresceinhirudin 54 -65 to the thrombin-antithrombin complex could not be demonstrated, suggesting that the K d value was greater than 5 M.
In contrast to the results obtained with thrombin-antithrombin complexes, when thrombin-heparin cofactor II or thrombin-␣ 1 -antitrypsin M358R complexes were titrated with fluorescein-hirudin 54 -65 , no binding was detected (Fig. 2). These findings suggest that the structure of thrombin may differ depending on the serpin with which it is complexed. To further explore this concept, the binding of fluoresceinhirudin 54 -65 to thrombin was monitored in real time as the sample was titrated with each of the three serpins. As illustrated in Fig. 3, addition of thrombin to fluoresceinhirudin 54 -65 produced a ϳ12% decrease in fluorescence intensity, and this value was unchanged upon the addition of 400 nM heparin. When 100 nM antithrombin was added, the fluorescence intensity rapidly returned to a value approaching that obtained with fluorescein-hirudin 54 -65 alone, suggesting that antithrombin reduced the amount of fluorescein-hirudin 54 -65 bound to thrombin. When parallel experiments were performed with heparin cofactor II and ␣ 1 -antitrypsin M358R, similar results were obtained. Based on fluorescence intensity values recorded after serpin addition, heparin cofactor II or ␣ 1 -antitrypsin M358R displaced 99% of fluorescein-hirudin 54 -65 from thrombin, whereas only 92% displacement was observed with antithrombin. Reduced displacement with antithrombin relative to the other two serpins is consistent with the observation that fluorescein-hirudin 54 -65 retains affinity for the thrombinantithrombin complex, albeit much less than that for thrombin (Fig. 2).
To determine whether formation of a covalent complex between thrombin and a non-serpin, active-site-directed inhibitor is sufficient to alter the integrity of exosite 1, FPRck (500 nM) was added to a cuvette containing fluorescein-hirudin 54 -65 and thrombin (not shown). In contrast to the results with the serpins, chloromethyl ketone addition had no effect on fluores- cence intensity, indicating that it did not displace fluoresceinhirudin 54 -65 from exosite 1. These results suggest that reduced binding of exosite 1-directed ligands to thrombin complexed by serpins cannot be explained solely by covalent complex formation at the active site of the enzyme.
Displacement experiments also were performed using an exosite 1-directed DNA aptamer, HD-1, in place of the hirudin 54 -65 peptide (22). The fluorescent derivative of HD-1 binds thrombin with a K d value of 235 nM, producing an ϳ11% increase in fluorescence intensity (Fig. 4, inset). Under the conditions of the displacement experiment, addition of 100 nM thrombin to 20 nM fluorescein-HD-1 resulted in a ϳ3.5% increase in fluorescence intensity (Fig. 4). When 500 nM antithrombin, heparin cofactor II, or ␣ 1 -antitrypsin M358R was added in the presence of 4 nM heparin, the fluorescence intensity rapidly returned to a value similar to that of unbound fluorescein-HD-1. These results are in agreement with those obtained with hirudin 54 -65 , suggesting that exosite 1 function is compromised when thrombin is complexed by serpins. The observation that antithrombin addition results in complete displacement of fluorescein-HD-1, but not of fluoresceinhirudin 54 -65 , may reflect differing binding interactions between the two ligands and thrombin.
To determine whether the interaction of exosite 1 with physiological ligands also was altered when thrombin was complexed with serpins, we compared the binding of thrombinserpin complexes to fibrin with that of thrombin (Fig. 5). Samples containing 50 nM 125 I-thrombin or 125 I-thrombin-serpin complexes and 0 -10 M fibrinogen were clotted with Atroxin. After 60-min incubation, clots were compacted by centrifugation and unbound 125 I-thrombin or 125 I-thrombin-serpin complex in the supernatants was quantified by counting aliquots for radioactivity. Under these conditions, thrombin bound fibrin with a K d value of 1.6 M, a value comparable to that reported previously (31,32). In contrast, no binding of thrombin-serpin complexes was observed. Thus, the loss of function of exosite 1 observed with synthetic ligands extends to physiological ligands such as fibrin.
As a second macromolecular probe of exosite 1, binding of a thrombin-specific antibody to thrombin and thrombin serpin complexes was examined. Autoantibody D is a human-derived antibody that binds to thrombin to exosite 1, as demonstrated by the lack of binding to thrombin variants with mutations in this region (29). Using protein G-agarose to extract thrombinantibody complexes, autoantibody D bound 66% of 125 I-thrombin (not shown). The antibody recovered 11, 10, and 13% of the radioactivity of thrombin-antithrombin, thrombin-heparin cofactor II, or thrombin-␣ 1 -antitrypsin M358R complexes, respectively. Thus, reactivity with the exosite 1-directed antibody was greatly reduced when thrombin was complexed with serpins.

Integrity of Exosite 2 on Thrombin Complexed with Serpins-
Having demonstrated that the function of exosite 1 was impaired when thrombin was complexed by serpins, it was of interest to examine the integrity of exosite 2 in the thrombininhibitor complexes. Initial experiments utilized an exosite 2-directed DNA aptamer, HD-22 (23). Titration of fluoresceinthrombin with HD-22 produced a saturable decrease in fluorescence intensity as HD-22 bound thrombin with a K d value of 227 nM (Fig. 6). When complexes of fluorescein-thrombin with antithrombin, heparin cofactor II, or ␣ 1 -antitrypsin M358R were titrated with HD-22, there was a saturable decrease in fluorescence intensity. Although the decreases in fluorescence intensity with the complexes were less than that observed with thrombin, the K d values of HD-22 for fluorescein-thrombin complexed with antithrombin, heparin cofactor II, and ␣ 1 -antitrypsin M358R were 370, 220, and 272 nM, respectively, values less than 2-fold higher than that of HD-22 for fluoresceinthrombin. These data suggest that, in contrast to exosite 1, the integrity of exosite 2 is minimally affected when thrombin is complexed by serpins. Addition of FPRck to a sample of HD-22 and fluorescein-thrombin caused no change in fluorescence intensity, indicating that low molecular weight active site adducts also have no effect on the integrity of exosite 2 (not shown).
To confirm these concepts, we used heparin, a more physiological ligand, to probe the integrity of exosite 2 on thrombin complexed by serpins. Elution from heparin-Sepharose was used to qualitatively monitor the heparin-thrombin interaction. 125 I-Thrombin or 125 I-thrombin-serpin complexes were applied to the column in 10 mM Tris-HCl, pH 7.4, washed, and eluted using a linear NaCl gradient (Fig. 7). Peak elution of uncomplexed 125 I-thrombin was at fraction 34 at ϳ450 mM NaCl. The three 125 I-thrombin-serpin complexes bound to the heparin-Sepharose but eluted at slightly lower concentrations of NaCl. 125 I-Thrombin complexes with antithrombin and heparin cofactor II eluted with peaks at about 340 mM NaCl, whereas the thrombin-␣ 1 -antitrypsin M358R complex eluted at 250 mM NaCl. These results demonstrate that the thrombinserpin complexes retain affinity for heparin. Because only the thrombin moiety of the complexes was radiolabeled (Fig. 1), binding of unreacted serpin to heparin-Sepharose was not detected by this method. To positively identify the species eluting from the column, 50 -400 g of unlabeled thrombin or thrombin-serpin complexes was applied to the column and elution was monitored spectrophotometrically at 280 nm (not shown). Peak fractions were concentrated and electrophoresed on SDS-PAGE gels. Under these conditions, antithrombin eluted at ϳ1200 mM NaCl, consistent with its higher affinity for heparin relative to thrombin (33). The thrombin-antithrombin reaction mixture eluted in two peaks, one at a position coincident with thrombin and a second eluting with a slightly lower concentration of NaCl. SDS-PAGE analysis revealed that these two peaks were thrombin and thrombin-antithrombin complex, respectively. The elution profile confirms that the thrombin-antithrombin complex retains affinity for heparin comparable to that of thrombin. Heparin cofactor II demonstrated lower affinity for heparin-Sepharose than thrombin, likely reflecting intramolecular interactions between its positively charged heparin binding domain and its anionic NH 2 -terminal domain (34). The thrombin-heparin cofactor II reaction mixture eluted in two peaks, one coincident with unreacted heparin cofactor II and the other, identified as thrombin-heparin cofactor II complex, at a NaCl concentration close to that of thrombin. ␣ 1 -Antitrypsin M358R demonstrated only weak heparin affinity, consistent with its lack of a heparin binding site (19). The thrombin-␣ 1 -antitrypsin M358R reaction mixture eluted from heparin-Sepharose in three peaks, two equivalent to free thrombin and ␣ 1 -antitrypsin M358R, and the other at an intermediate NaCl concentration. PAGE analysis identified the latter peak as thrombin-␣ 1 -antitrypsin M358R complex. Taken together, these results indicate that thrombin-serpin complexes retain their affinity for heparin, suggesting that the integrity of exosite 2 is not compromised when thrombin is complexed by serpins.
To demonstrate that the thrombin-antithrombin complex was not adsorbing to the heparin-Sepharose complex via its antithrombin moiety, binding of pentasaccharide to antithrombin and thrombin-antithrombin complex was assessed (not shown). Pentasaccharide binding to antithrombin, monitored by changes in intrinsic fluorescence of antithrombin, yielded a K d value of 25 nM and a 30% increase in fluorescence intensity.
In contrast, addition of pentasaccharide to thrombin-antithrombin complex had no effect on intrinsic fluorescence, sug- gestive of no interaction. Because antithrombin within the thrombin-antithrombin complex has reduced affinity for pentasaccharide, the retention of the thrombin-antithrombin complex on heparin-Sepharose is mediated by the thrombin moiety. DISCUSSION This study demonstrates that the functional integrity of exosite 1 on thrombin is essentially lost when the enzyme is complexed by serpins. In contrast, exosite 2 on thrombin complexed by serpins retains its affinity for its ligands. Results were obtained using small, synthetic exosite-binding ligands as well as macromolecular targets of thrombin binding. The thrombin-antithrombin complex exhibited 55-fold lower affinity for fluorescein-hirudin 54 -65 than thrombin in a direct binding experiment (Fig. 1). This result corroborates a previous report that the affinity of fluorescein-hirudin 54 -65 for thrombin is reduced greater than 200-fold when the enzyme is complexed with antithrombin (21). The discrepancy in values may reflect differences in the methods used to label the hirudin 54 -65 peptide. Isothiocyanate was used in the present study, whereas a succinimidyl ester-labeling procedure was employed in the previous study. We have extended the results of the earlier study by examining other serpins. In contrast to the result with antithrombin, fluorescein-hirudin 54 -65 does not bind to thrombin-heparin cofactor II or thrombin-␣ 1 -antitrypsin M358R complexes. Similarly, peptidyl and oligonucleotidyl ligands bound to exosite 1 were displaced when thrombin was inhibited by serpins. Hirudin 54 -65 was quantitatively displaced when thrombin was complexed by heparin cofactor II or ␣ 1 -antitrypsin M358R, but not antithrombin. Thrombin-serpin complexes also have greatly reduced affinity for fibrin, an interaction mediated by exosite 1, and for an antibody directed against exosite 1. These observations indicate that the function of exosite 1 is compromised when thrombin forms complexes with serpins.
The integrity of exosite 2 is maintained when thrombin is complexed by serpins. Thus, an exosite 2-directed oligonucleotide binds thrombin-serpin complexes with affinities similar to that for thrombin. Studies using heparin, a physiological ligand for exosite 2, verified that this exosite retained measurable functional activity when thrombin was complexed by serpins. The use of a non-heparin binding serpin, ␣ 1 -antitrypsin M358R, confirmed that binding of thrombin-serpin complexes to heparin was not mediated by the heparin-binding domains of antithrombin or heparin cofactor II. In addition, we demonstrated that, once complexed with thrombin, antithrombin no longer binds pentasaccharide, a finding in agreement with a previous report (35). Therefore, when complexed by serpins, the integrity of exosite 2 on thrombin is retained, whereas that of exosite 1 is lost.
Three serpins with different modes of interaction with thrombin were used to examine whether impairment of exosite function was influenced by different use of the exosites during the inhibition reaction. The interactions of antithrombin and ␣ 1 -antitrypsin M358R with thrombin are exosite 1-independent, as evidenced by inhibition studies with ␥-thrombin or exosite 1 mutants (36, 37). 2 Despite this, when thrombin formed a complex with either serpin, exosite 1 function was lost. Heparin cofactor II differs from antithrombin and ␣ 1antitrypsin M358R, because its NH 2 -terminal domain interacts with exosite 1 on thrombin during the inhibition reaction. However, based on the results with the other two serpins, the NH 2 -terminal domain of heparin cofactor II is unlikely to remain bound to the impaired exosite. Indirect verification of this concept is provided by our observation that the thrombin-hep-arin cofactor II complex displaces thrombin bound to fibrin monomer-Sepharose (38). This suggests that the NH 2 -terminal domain of heparin cofactor II within the thrombin-serpin complex is capable of binding to other thrombin molecules. The fact that hirudin 54 -65 binds, albeit weakly, to the thrombin-antithrombin complex, but not to thrombin complexed by heparin cofactor II or ␣ 1 -antitrypsin M358R, may indicate subtle differences among the complexes. However, it also is possible that the fluorophore did not register interactions with the latter two inhibitors.
Loss of function of exosite 1 on thrombin in complex with serpins occurs in the presence or absence of heparin. When fluorescein-hirudin 54 -65 or fluorescein-HD-1 was displaced from thrombin upon addition of serpins, heparin was added to catalyze the inhibition reactions. However, the preformed thrombin-serpin complexes, which were used to titrate fluorescein-hirudin 54 -65 , were prepared in the absence of heparin. Therefore, the similarity of results obtained with exositedependent and -independent inhibitors in the absence and presence of heparin demonstrates that impairment of exosite 1 function is a generalized response to thrombin inhibition by serpins.
Disruption of protease structure is not simply a function of the formation of a covalent adduct at the active site, because exosite 1 function is not impaired when thrombin is complexed with a chloromethyl ketone derivative. Thus, in contrast to serpins, inhibition by FPRck does not displace fluoresceinhirudin 54 -65 from thrombin. Likewise, FPR-thrombin retains its affinity for fibrin, whereas thrombin-serpin complexes do not bind to fibrin. These data are consistent with the suggestion that the interaction of proteases with macromolecular inhibitors produces greater conformational changes in the enzyme than interaction with inhibitors of lower molecular mass (39).
The failure of thrombin-serpin complexes to bind hirudin 54 -65 or fibrin is consistent with previous observations that the thrombin-antithrombin complex does not bind to thrombomodulin. Consequently, thrombomodulin recovers its cofactor activity for protein C activation when thrombin is inhibited by antithrombin (40). Although this observation is complicated by the fact that thrombomodulin also binds exosite 2 on thrombin via its chondroitin sulfate moiety, only the exosite 1-mediated interaction is sufficient to endow thrombin with the capacity to activate protein C. Previous studies have demonstrated that FPRck modification of thrombin does not affect its interaction with thrombomodulin (41), further supporting the contention that loss of exosite 1 function only occurs when thrombin interacts with macromolecular ligands (39).
Numerous studies have focused on the structural changes that occur in serpins when they complex their target proteases (42,43). Covalent complex formation produces considerable movement of the enzyme relative to the serpin, as the portion of the reactive site loop to which the enzyme is covalently attached inserts itself into the body of the serpin. Insertion moves the enzyme from one pole of the serpin to the other. Crystallographic analysis of the covalent trypsin-␣ 1 -antitrypsin complex indicates that the enzyme moiety also undergoes structural alterations that extend beyond the active site to involve a significant portion of the body of the enzyme (43). Thus, these data suggest that the loss of exosite function that occurs when thrombin is complexed by serpins may be a universal phenomenon common to all serpin-protease interactions.
Based on the crystal structure of a representative proteaseserpin complex, the loss of exosite 1 function that occurs when thrombin is complexed by serpins is more likely to reflect structural changes than steric inhibition (43). This is supported by our observation that even small ligands, such as hirudin 54 -65 and HD-1, exhibit reduced or absent binding to the thrombin-antithrombin complex. Analysis of the trypsin-␣ 1 -antitrypsin crystal structure reveals disorganization of about a third of the enzyme moiety, including the regions that correspond to exosite 1 on thrombin. Thus, the loss of function of exosite 1 may be a generalized phenomenon reflecting denaturation of the protease once it is complexed by a serpin. Selective disorganization of this region renders the enzyme more susceptible to proteolysis. Numerous studies have demonstrated increased sensitivity of the protease portion of the protease-serpin complex to proteolysis (39,44). Protease denaturation and subsequent proteolysis may serve a role in facilitating clearance of protease-serpin complexes.
The targeted inactivation of exosite 1 leaves open the question as to why the function of exosite 2 is spared when thrombin is complexed by serpins. Consistent with the functional studies presented here, the trypsin-␣ 1 -antitrypsin crystal structure reveals less disorganization in the exosite 2 region of the molecule (43). Retention of the heparin-binding domain in the thrombin moiety of the thrombin-antithrombin complex may promote clearance via low density lipoprotein receptor-related protein (45). However, the lack of heparin binding sites in other serine proteases argues against this being a universal clearance mechanism for enzyme-serpin complexes.
As a general phenomenon, the role of enzyme denaturation upon interactions with serpins is unclear. In the case of thrombin, however, this process ensures that, once thrombin is inhibited, it no longer remains bound to thrombomodulin, thereby preventing thrombin-antithrombin complexes from acting as competitive inhibitors of thrombin-mediated protein C or TAFI activation. Much like the release of heparin from antithrombin within the thrombin-antithrombin complex, recovery of thrombomodulin upon inhibition of thrombin permits cofactor recycling.