Proteolytic Formation of Either of the Two Prothrombin Activation Intermediates Results in Formation of a Hirugen-binding Site*

Hirugen, a synthetic dodecapeptide corresponding to the carboxyl-terminal amino acids 53-64 of hirudin, binds within a deep groove in thrombin that contains a cationic region referred to as the anion-binding exosite. This region is important in many of the binary interactions of thrombin with macromolecular substrates and cofactors. Fluorescein-labeled hirugen was used to probe which steps in the prothrombin activation process generate this anion-binding exosite. Two activation cleavage sites exist in bovine prothrombin. Cleavage at Arg”74-Thr“7“ releases the activation frag- ments to generate the thrombin precursor, prethrombin 2. Cleavage of prothrombin within a disulfide loop leads to formation of meizothrombin with no loss of peptide material but with formation of amidolytic activity. Cleavage of the same bond in prethrombin 2 generates thrombin. Hirugen, labeled at the amino terminus with fluorescein isothiocyanate, does not bind to prothrombin but does bind to thrombin (Kd = 9.6 f 1.2 X lo-’ M), prethrombin 2 (& = 1.3 f 0.1 X M), thrombin-fragment-2

characteristics (1,2). Prethrombin 2, a single chain precursor of thrombin with no proteolytic or amidolytic activity, results from proteolytic cleavage at Arg"'-Thr''" to release the activation fragments that constitute nearly one-half of the prothrombin molecule. Prethrombin 2 is converted to thrombin by cleavage of the Arg'2'i-Ile'i24 bond within a disulfide loop (3). Both thrombin and prethrombin 2 interact noncovalently with activation fragment-2. This interaction augments prethrombin 2 activation (4) and enhances thrombin's esterolytic activity ( 5 ) . Meizothrombin is formed from prothrombin by the cleavage of the Arg""-Ile:'"' bond, but because the cleavage occurs within a disulfide loop, no activation fragments are released. Although meizothrombin has activity toward low molecular weight substrates, it has little activity toward fibrinogen (6). One explanation for the lack of clotting activity is that the extended fibrinogen-binding pocket is not yet available in meizothrombin, perhaps because the activation fragment masks the extended fibrinogen-binding pocket, referred to as the anion-binding exosite. No direct studies have been performed to evaluate this hypothesis.
A useful reagent to probe this issue was suggested by recent studies on the function of different domains of the leech thrombin inhibitor hirudin (7,8). Of particular importance to the present study, the carboxyl-terminal 12-residue portion of hirudin, referred to as hirugen, binds to thrombin and inhibits fibrinogen clotting activity without inhibiting hydrolytic activity toward low molecular weight substrates (9,10). Thus, the hirugen-thrombin complex shares many properties in common with meizothrombin. X-ray crystallographic data of the thrombin-hirudin complex indicate that the hirugen portion of the hirudin molecule binds to the deep groove that appears to include the anion-binding exosite (11).
We have used hirugen to determine which proteolytic events in prothrombin activation are associated with formation of the anion-binding exosite.
Bovine and human prothrombin, thrombin, and the activation intermediates and fragments were isolated by published methods (1). Meizothrombin and meizothrombin des fragment-1 were generated in the presence of PPACK to block the active site during their formation and were subsequently isolated as described previously (12). Hirugen was a generous gift from Dr. J. M. Maraganore at Biogen, Cambridge, MA. Except where specifically stated, bovine clotting factors were used in this study.
Fluorescein-Labeling of Hirugen-Hirugen, 5 X lo-* mol in 10 11 of water, was mixed with 2 p l of 1 M NaHCO:,, pH 9.0, and 6 p1 of a FITC solution (10 mg/ml in dimethyl sulfoxide) and incubated at room temperature for 1 h in the dark. The reaction was then terminated by addition of an excess (5 pl) of 1 M NH,CI, pH 7.0. A Sephadex G-10 column (0.5 X 10 cm) was used to separate the labeled peptide from free FITC. For the equilibrium binding studies, the FITC-labeled hirugen (FI-hirugen) was further purified on a Vydac C18 218TP54 reverse phase column (4.6 X 300 mm, 5 pm/300 A, Separations Group, Hesperia, CA) attached to a Gilson high pressure liquid chromatography system (Gilson Medical Electronics, Inc., Middleton, WI). FI-hirugen was eluted with a 60-ml linear gradient in 10 mM sodium phosphate, pH 7.0, from 0 to 80% acetonitrile. FI-hirugen was stored at -80 "C. The concentration of FI-hirugen was estimated assuming a molar absorptivity of 6.8 X lo4 M-' cm" at 492 nm at pH 8.0 for covalent FITC conjugates (13). Uncorrected spectra of FIhirugen had an excitation maximum of 492 nm and a fluorescence emission maximum at 515 nm in 0.1 M NaC1, 20 mM HEPES, pH 7.5. All fluorescence measurements were performed at these wavelengths with an excitation bandpass of 4 nm and an emission bandpass of 8 nm.
formed on a TSK BioSil-250 column (7.5 X 300 mm) (Bio-Rad) Gel Filtration Studies-Gel filtration chromatography was perattached to an fast protein liquid chromatography system (Pharmacia LKB Biotechology Inc.). In experiments with prethrombin 2, the column was equilibrated and eluted in 0.1 M NaC1, 20 mM HEPES, pH 7.5. In all other experiments, the column was equilibrated and eluted in 0.15 M NaCl, 10 mM sodium phosphate, pH 7.5. Two different buffer systems were selected, since the latter buffer dissociated the prethrombin 2-hirugen complex but substantially decreased thrombin adsorption to the matrix, allowing improved separation of thrombin and other components. The flow rate was 1 ml/min. Samples of 100 p1 were applied through a preprogramed injection loop that concomitantly activated the UV strip chart tracing. Fractions (200 pl) were collected, and the fluorescence intensities from the FIhirugen were detected on a SLM-8000 C spectrofluorimeter (SLM Instruments Inc., Urbana, IL). Fluorescence intensities were determined by sequential addition of 25-p1 aliquots of successive fractions into a 1 X 1-cm cuvette. The fluorescence intensity of an individual fraction was determined by the difference between successive readings after correction for volume changes. Visualization of Complex Formation by Gel Electrophoresis-To determine qualitatively the influence of hirugen on fragment-2thrombin complex formation, nondenaturing gel electrophoresis was employed. The migration patterns were virtually identical with those seen previously (4). Polyacrylamide gel (12% homogeneous, acrylamide:bisacrylamide = 37.5:1) electrophoresis was run without detergent and without the stacking gel. The upper tank buffer was Tris/ glycine buffer (37.6 mM Tris, 40 mM glycine, pH 8.9). The lower tank buffer was Tris-HCI buffer (63 mM Tris base, 50 mM HCI, pH 7.5). Samples (50-pl total volume) were preincubated with hirugen at room temperature for 30 min before loading onto the gel. In samples that contained fragment-2, the fragment-2 was added immediately prior to electrophoresis. This sequence of addition was designed to facilitate hirugen competition for the fragment-2 site. Incubations contained the following amounts of material. Equilibrium Binding Studies-The fluorescence emission intensity of FI-hirugen decreased upon binding to thrombin, prethrombin 2, meizothrombin, and the thrombin-fragment-2 complex. This spectral change was utilized to determine the binding affinity of FI-hirugen for these proteins. Titrations were performed by the sequential ad-dition of 2-10 p1 of t.hrombin or other protein solutions into 140 pl of FI-hirugen in a 0.5 X 0.5-cm cuvette in 0.1 M NaC1, 0.02 M Tris-HCI, pH 7.5 (Tris-buffered saline, TBS). FI-hirugen titration with TBS resulted in less than a 1% change in fluorescence emission intensity. The F I E , data were corrected for volume changes. After each addition, two 20-s readings were performed. The percent bound was assumed to equal Fc, -F/F(, -F,,,, where F,, is the fluorescence intensity before titrant addition, F is the fluorescence intensity at a particular titrant input, and F,,,, is the extrapolated fluorescence intensity at infinite titrant. The Kd was derived from the fluorescence data by assuming one hirugen-binding site per thrombin and fitting the binding data using nonlinear regression analysis, ENZFITTER (Imperial College of Science and Technology, London, United Kingdom).
The Influence of Hirugen on Thrombin Specificity-Peptide chromogenic substrate hydrolysis by thrombin, with or without hirugen present, was carried out in 96-well vinyl plates (Costar Corp., Cambridge, MA) in 0.1 M NaCl, 5 mM CaCI2, 1 mg/ml bovine serum albumin, and 50 mM MOPS, pH 7.5. The time course of hydrolysis was monitored with a V,,,., kinetic microplate reader in the kinetic mode (Molecular Devices Corp., Menlo Park, CA). Only the linear portion of the curve (less than 20% substrate consumption) was used to calculate the initial rates. Because the major goal of this study was to identify a hirugen-insensitive substrate, the K,, and k,.,, values for each of the substrates tested were not determined.
The Influence of Hirugen and Prothrombin Fragment-2 on Prethrombin 2 Actiuation-Human prethrombin 2 activation was performed at room temperature for 1 min in the MOPS buffer described above with factor Xa with or without factor Va, fragment-2, and hirugen. To ensure factor Xa complex formation with factor Va in the absence of phospholipid and still retain an activation rate slow enough to allow initial rate measurements conveniently, we included a molar excess of active site-blocked factor Xa in the reaction mixture (covalently modified by incubation with p-amidinophenylmethanesulfonyl fluoride (see the legend to Fig. 4). The reactions were terminated by diluting the sample 2-fold with 0.1 M NaCI, 10 mM EDTA, 50 mM MOPS, pH 7.5, and chromogenic substrate. Substrate hydrolysis was monitored immediately. Relative thrombin generation was determined by comparing A,,,:,/min from thrombin generated in the different reaction mixtures. Under the conditions employed, the AJIk-> was directly proportional to the thrombin concentration.

RESULTS
Gel filtration studies indicated that F1-hirugen binds specifically to thrombin. Fluorescence from the F1-hirugen CO-eluted with thrombin from the TSK-250 column (Fig. 1B). In this case and all other experiments shown in Fig. 1, the F1hirugen was displaced by a 20-fold molar excess of unlabeled hirugen (data not shown). Addition of F1-hirugen to protein samples did not alter the elution volume of the respective proteins.
Thrombin (Fig. 1A) and prethrombin 2 ( Fig. 1Z) appear to bind to the gel filtration matrix because their elution volumes are considerably greater than that of fragment-2 (Fig. 1D).
Thrombin and fragment-2 interact to form a noncovalent complex that can be identified by a decrease in elution volume of the complex relative to both thrombin and fragment-2 ( Fig.   1, E a n d F ) . Fluorescence from F1-hirugen was associated with the thrombin-fragment-2 complex, consistent with formation of a ternary complex. Addition of a 2-fold molar excess of fragment-2 to the thrombin before gel filtration shifted more of the thrombin into complex but did not prevent F1hirugen binding (Fig. 1F).
Whether the Arg''"-Thr''5 proteolytic cleavage in prethrombin 2 was necessary for formation of the hirugen-binding site was also investigated by gel filtration chromatography. Prethrombin 2 and fragment-2 also form a noncovalent complex, as indicated by a reduced elution volume of the complex.
When F1-hirugen was incubated with prethrombin 2 and fragment-2, fluorescence from the F1-hirugen was associated with this complex (Fig.   1 4 . No binding was detected to fragment-2 alone (data not shown).
The relationship between hirugen binding and cleavage of A.

G .
H. the Arg'2''-Ile'24 bond and subsequent release of the activation fragment was also investigated. F1-hirugen bound to PPACK meizothrombin or PPACK meizothrombin des fragment-1 (Fig. 1, G and H). Similar gel filtration experiments with prothrombin and prethrombin 1 failed to detect any Fl-hirugen binding (data not shown). Thus, neither the formation of catalytic activity nor the release of the covalently associated activation fragments are required for hirugen binding, but the binding site is cryptic in prothrombin.
Additional evidence that hirugen does not prevent fragment-2 thrombin and fragment 2-prethrombin 2 complex formation was obtained by electrophoresis studies. When subjected to polyacrylamide gel electrophoresis in the absence of detergents under the conditions described under "Experimental Procedures," thrombin and prethrombin 2 migrate slowly, fragment-2 migrates rapidly, and the complex is intermediate. A 70-fold molar excess of hirugen did not disrupt fragment-2 complex formation with either thrombin or prethrombin 2, and hirugen itself did not alter the electrophoretic mobility of either thrombin or prethrombin 2 in the absence of fragment-2 (data not shown). Attempts to provide direct evidence of a ternary complex of fragment-2-thrombin-himgen using F1-hirugen were negative, probably because the off rate was too high to even allow visualization of the thrombin-F1 hirugen complex alone.
Fluorescence Analysis of F1-hirugen Binding-To ascertain the affinities of FI-hirugen binding to thrombin, prethrombin 2, and PPACK-meizothrombin derivatives, we utilized the observation that F1-hirugen binding is associated with a decrease in fluorescein emission intensity. The titration curves are shown in Fig. 2, and the maximum emission intensity changes and calculated Kd values for each of the protein-F1hirugen interactions are summarized in Table I. It is apparent that all forms of the thrombin intermediates and thrombin itself bound F1-hirugen saturably and with moderately high affinity. In all cases, inclusion of a 60-100-fold molar excess of unlabeled hirugen over the thrombin or thrombin intermediate reversed greater than 94% of the emission intensity change, thereby indicating that F1-hirugen binding was specific (data not shown). The observation that hirugen binding was 10-fold weaker in the presence of fragment-2 suggested the possibility that there could be some displacement of fragment-2 during the titrations. If this were the case, then the F1-hirugen binding affinity would be different at different saturating concentrations of fragment-2. The binding was monitored a t 2-and 5-fold molar excesses of fragment 2. The titration curves were indistinguishable, indicating that displacement of fragment-2 is an unlikely explanation for the 10-fold reduction in binding affinity (data not shown). The observation that the maximum change in F1-hirugen fluorescence is greater with the fragment-2-thrombin complex than with thrombin alone also indicates that F1-hirugen can bind to form a ternary complex.  The influence of Hirugen on Chromogenic Substrate Hydrolysis-The amidolytic activity of human thrombin was modified upon interacting with hirugen. For example, the rate of cleavage of Spectrozyme TH at saturating substrate (4 x 10P M) was increased 57% by hirugen (Fig. 3). In contrast, hirugen inhibited the hydrolysis of several substrates (Fig. 3). T o facilitate the analysis of hirugen effects on prothrombin activation, it was particularly useful to identify substrates that were unaffected by hirugen. Under our experimental conditions, hirugen had almost no effect on the hydrolysis rates of tosyl-Gly-Pro-Arg-p-nitroanilide and benzoyl-Ile-Glu-Gly-Arg-p-nitroanilide. The results with bovine thrombin were similar, except that none of the substrates proved insensitive to the presence of hirugen (data not shown).
The Influence of Hirugen on the Actiuation of Prethrombin 2-The rate of activation of human prethrombin 2 by factor Xa and factor Va remained essentially constant up to about 1 PM hirugen (Fig. 4). At higher hirugen concentations, there was a dose-dependent decrease in the initial rate of prethrombin 2 activation. In the presence of fragment-2, inhibition by hirugen required higher concentrations (Fig. 4, line 2). Hirugen also inhibited prothrombin activation under these conditions with an IC,,, = 10 PM (data not shown). By using a synthetic substrate that is least affected by hirugen (N-ptosyl-Gly-Pro-Arg-p-nitroanilide), the observed effect of hirugen on thrombin was minimized (Fig.  4, line I ) . In the absence of factor Va in the activation mixture, hirugen failed to inhibit prethrombin 2 activation a t concentrations as high as 1 mM (Fig. 4, line 4 ) . DISCUSSION In prothrombin and prethrombin 1, the hirugen-binding site is not expressed, but the binding site is  of the two possible activation intermediates, as well as thrombin itself. This conclusion is based primarily on direct binding studies employing gel filtration chromatography and by equilibrium methods employing F1-hirugen. Fragment-2 can interact with either thrombin or prethrombin 2 without displacing F1-hirugen. The equilibrium methods are not unambiguous, however, since we have no independent means of monitoring fragment 2 binding in these experiments. The observation that A F,,, for F1-hirugen binding to thrombin is greater in the presence than in the absence of fragment 2 further supports the conclusion that a ternary complex can form. The nearly equivalent affinity of meizothrombin and thrombin for F1-hirugen indicate that the fragment-2-and hirugen-binding sites are essentially nonoverlapping. Since fragment-2 does not block hirugen binding, fragment-2 and hirugen cannot bind in the same location either in meizothrombin or the thrombin-fragment 2 complex. These results suggest that formation of the hirugen-binding site that accompanies either of the two activation cleavage events in prothrombin results from conformational changes rather than unmasking the hirugen-binding site by alleviating steric hindrance due to fragment-2 interaction. The latter conclusions are predicated on the assumption that reversible association of fragment-2 with thrombin or prethrombin 2 and the interaction of the fragment-2 and thrombin domains within meizothrombin involve the same site as in prothrombin. In drawing these conclusions, it is important to note that the affinity of meizothrombin for F1-hirugen was slightly higher than that of thrombin, whereas the affinity of prethrombin 2 for F1hirugen was slightly lower. These differences probably reflect differences in the conformation of the hirugen-binding site in these species. Noncovalent association of the activation peptide with thrombin and prethrombin 2 (data not shown) decrease the affinity of F1-hirugen binding, indicating that the conformation of the hirugen-binding cleft is different when fragment 2 associates noncovalently or covalently with thrombin and prethrombin 2, although limited overlap of the binding sites cannot be totally excluded.
Fluorescein labeling of hirugen allowed quantitative assessment of binding affinity, but created the potential problem of altering the binding specificity (14,15). The ability of unlabeled hirugen to displace F1-hirugen strongly argues against this possibility. The affinity for F1-hirugen observed in this study (Kd = 9.8 X lo-' M using bovine thrombin) was somewhat higher than the previously reported values for hirugen. K, = 1.44 X M in blocking human thrombin fibrinogen clotting (16); Ki = 5.4 x M in human thrombin-catalyzed release of fibrinopeptide A from fibrinogen (9). This difference is somewhat greater than it would appear since the affinity of human thrombin for hirugen is approximately 10-fold greater than that of bovine thrombin (17,18).
Not only does the hirugen-binding site result from conformational changes that occur during prothrombin activation, but interaction of hirugen with thrombin also elicits a conformational change that alters substrate specificity (9,19). Our data indicate that hirugen can function as either a positive or negative effector, depending on the substrate analyzed. Since many of the substrates were analyzed at near saturation, hirugen must influence k,,,, as well as K,, as recently reported (

19) *
At concentrations of hirugen at or greater than the prethrombin 2 concentration, in the presence of factor Va, inhibition of the initial rate of activation was observed. Higher hirugen concentrations were required to obtain inhibition of prethrombin 2 activation in the presence of fragment 2 or when prothrombin was the substrate. In the absence of factor Va, inhibition does not occur at hirugen concentrations up to 1 mM. The hirugen concentration dependence for inhibition of prethrombin 2 activation was somewhat similar to that observed for binding to prethrombin 2. Thus, the observed inhibition could reflect either that hirugen inhibits factor Va-Xa interaction or that the hirugen-prethrombin 2 complex is a poor substrate for the factor Xa-Va complex. In many experiments, there was a very small (10%) increase in the prethrombin 2 activation rate at hirugen concentrations somewhat lower than the prethrombin 2 concentration. This increase occurred only in the presence of factor Va and the absence of fragment 2 (data not shown). The latter experiments favor the concept that the hirugen may dissociate the factor Xa-Va complex, although definitive evidence will require direct equilibrium measurements.
The demonstration in the present study that the hirugenbinding site is present in prethrombin 2 complements the finding (3) that the active site titrant, dansyl-arginine-N-(3-ethyl-1,5-pentanediyl)amide, can interact with prethrombin 2. The presence of the hirugen site in prethrombin 2 helps to explain the observation that prethrombin 2 binds to thrombomodulin, albeit approximately 10-fold weaker than thrombin (20). Thus, several of the ligand-binding sites of thrombin are already expressed on prethrombin 2, although its catalytic triad is not yet competent.