A Thrombin-based Peptide Corresponding to the Sequence of the Thrombomodulin-binding Site Blocks the Procoagulant Activities of Thrombin*

Thrombomodulin, a cofactor in the thrombin-cata- lyzed activation of protein C, blocks the procoagulant activities of thrombin such as fibrinogen clotting, Fac- tor V activation, and platelet activation. The binding site for thrombomodulin within human thrombin has been localized at a region comprising residues Thr14?-Ser'" of the B-chain of thrombin. The dodecapeptide sequence, TWTANVGKGQPS, corresponding to these residues inhibits thrombin binding to thrombomodulin with an apparent Ki = 94 WM (Suzuki, K., J., and Hayashi, T. (1990) J. Biol. Chem. 265, 13263- 13267). We have found that the inhibitory effect of the dodecapeptide on the thrombin-thrombomodulin inter- action is sequence-specific, and that residues and Gln15' are essential for thrombomodulin binding. The dodecapeptide was also found to directly block thrombin procoagulant activities, fibrinogen clotting (concentration for half-maximum inhibition, 385 p ~ ) . Factor V activation (concentration for half-maximum inhibition, 33 p ~ ) , and platelet activation (concentration for half-maximum inhibition, 645 p ~ ) . This peptide did not block thrombin inhibition by anti- thrombin 111, but blocked thrombin inhibition by hirudin. These findings suggest that the binding site for thrombomodulin coupled to the cu-amino deprotected resin. The side chain protection used was: Asp(0 Bzl), Lys(ClZ), Ser(Bzl), Trp(CHO), Tyr(BrZ), and Glu(OBz1). Trp deformylation was performed using 1 M piperidine in dimethylformamide for 90 min at 0 "C, and the peptide was cleaved from the resin using a mixture of HF:anisole:dimethyl sulfide (1Ol:l) for 2 h at 0 "C. The peptide was purified by reverse-phase preparative HPLC on a YMC-R-ODs-5 120A (4.9 X 300 mm) column (Yamamura Chemical Laboratories, Tokyo) using a gradient of 0-70% acetonitrile in 0.1% trifluoroacetic acid. The peptides were at least 95% pure as judged by reverse-phase HPLC and amino acid analysis, which was performed after hydrolysis in HC1 for 2 h at 165 "C, using the Waters Pic0 Tag System (Millipore). The concentrations of peptides were determined by quantitative amino acid analysis of concentrated pep- tide solutions prior to dilution for assay. thrombin Tris-HC1, 7.5, 5 CaC12, p1 hirudin (1.72 pg/ml) added, incubated at 37 "C 15 min. Subse- quently, 200 pl of 200 p~ H-D-Phe-Pip-Arg-pNA added, and, after 10 min, liberated p-nitroaniline was determined at 405 nm. of thrombin by antithrombin 111 in the presence or absence of peptide was determined with H-D-Phe-Pip-Arg-pNA using a modification of the method described previously (8).

. We have found that the inhibitory effect of the dodecapeptide on the thrombin-thrombomodulin interaction is sequence-specific, and that residues and Gln15' are essential for thrombomodulin binding. The dodecapeptide was also found to directly block thrombin procoagulant activities, fibrinogen clotting (concentration for half-maximum inhibition, 385 p~) . Factor V activation (concentration for halfmaximum inhibition, 33 p~) , and platelet activation (concentration for half-maximum inhibition, 645 p~) . This peptide did not block thrombin inhibition by antithrombin 111, but blocked thrombin inhibition by hirudin. These findings suggest that the binding site for thrombomodulin in thrombin is shared with the sites for fibrinogen, Factor V, platelets, and hirudin, and that, therefore, the inhibition of thrombin procoagulant activities by thrombomodulin in part results from blocking of the interaction between thrombin and the procoagulant protein substrates by thrombomodulin.
Thrombin is the most important procoagulant serine protease, converts fibrinogen to fibrin, and activates Factor V, Factor VIII, Factor XIII, and platelets (1). In binding to its receptor on the vascular endothelium, thrombomodulin, thrombin also plays a role as an anticoagulant enzyme. Thrombin bound to thrombomodulin activates protein C, which inhibits coagulation by inactivation of Factors Va and VIIIa (2, 3). Furthermore, thrombomodulin directly blocks the procoagulant activities of thrombin such as fibrinogen clotting, Factor V activation (4), and platelet activation ( 5 ) .
* This study was supported in part by Grants-in-Aid for Scientific Research from the Ministry of Education, Science and Culture of Japan, the Mochida Memorial Foundation for Medical and Pharmaceutical Research, the Inamori Foundation, and the CIBA-GEIGY Foundation (Japan) for the Promotion of Science. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore he hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ T o whom correspondence and reprint requests should be addressed.
The change in substrate specificity brought about by thrombomodulin is presumed to arise as a result of a conformational change around the active center of thrombin (6).
No6 et al. (7) suggested that a region comprising residues Arg6' to Arg73 of the B-chain of thrombin interacts with thrombomodulin as well as with fibrinogen and hirudin from evidence that a monospecific antibody raised to a peptide composed of residues Arg6* to Arg73, RIGKHSRTRYER, inhibits thrombin-catalyzed activation of protein C in the presence of thrombomodulin. We previously localized the thrombomodulin-binding site within thrombin to a region that corresponds to residues ThrI4? to SerI5' of the B-chain, and a dodecapeptide synthesized from this region, TWTANVGKGQPS, was found to block the interaction between thrombin and thrombomodulin with an apparent Ki = 94 /IM (8). In the present study, we have examined the effects of several homologous dodecapeptides on the thrombinthrombomodulin interaction. We also examined the effects of the peptides on the procoagulant activities of thrombin to elucidate how the binding of thrombomodulin modulates the procoagulant activities of thrombin. The results imply that thrombomodulin directly blocks the interaction between thrombin and the procoagulant protein substrates.

EXPERIMENTAL PROCEDURES
Preparation of Proteins-All chemicals used were of the highest commercial grade available. Protein C (2), a-thrombin (2,500 units/ mg) (9), Factor V (lo), prothrombin ( l l ) , Factor X ( l l ) , and antithrombin I11 (12) were all purified from human plasma as described previously. Recombinant soluble thrombomodulin, which is composed of an NH2-terminal domain, six epidermal growth factor-like structural domains, and an 0-glycosylation site-rich domain, was prepared as described (13). Human fibrinogen was obtained from Daiichi Pure Chemicals, Tokyo. Recombinant hirudin was provided from Mitsui Toatsu Chemicals, Tokyo. The phospholipid suspension was prepared using bovine brain extract (type 111) from Sigma. The chromogenic substrate for thrombin, H-D-Phe-Pip-Arg-pNA' (S-2238) was obtained from Kabi, Sweden. Chemicals for peptide synthesis were obtained from Applied Biosystems.
Preparation of Peptides-Synthetic peptides FRKSPQELL, RIG-KHSRTRYER, TWTANVGKGQPS, DSTRIRI, EGDSGGP, and PheI9 to Leu", Arg to Arg", Thr'47 to Ser'", Asp"s to Ile'"', Glu2"' SWGEGCDRDGK, respectively corresponding to sequence residues to Pro'", and SerXLti to LYS~'~ of the B-chain of thrombin, and SPQGKGVNATWT which is the inverted sequence between residues ThrI4' and Ser'", were produced by the solid phase method using an Applied Biosystems Model 431A peptide synthesizer. Other dodecapeptides, TWTADVGKGQPS, TWTANVGEGQPS, TWTANVGK-GEPS, TWTPNVGKGQPS, and TWTANGGKGQPS, which were homologous to TWTANVGKGQPS, were also synthesized by the same method. t-Butoxycarbonyl-protected amino acids were converted to the 1-hydroxybenzotriazole active ester and sequentially ' The abbreviations used are: Pip, pipecoryl; pNA, p-nitroaniline; HPLC, high performance liquid chromatography. coupled to the cu-amino deprotected resin. The side chain protection used was: Asp(0 Bzl), Lys(ClZ), Ser(Bzl), Trp(CHO), Tyr(BrZ), and Glu(OBz1). Trp deformylation was performed using 1 M piperidine in dimethylformamide for 90 min a t 0 "C, and the peptide was cleaved from the resin using a mixture of HF:anisole:dimethyl sulfide (1Ol:l) for 2 h at 0 "C. The peptide was purified by reverse-phase preparative HPLC on a YMC-R-ODs-5 120A (4.9 X 300 mm) column (Yamamura Chemical Laboratories, Tokyo) using a gradient of 0-70% acetonitrile in 0.1% trifluoroacetic acid. The peptides were at least 95% pure as judged by reverse-phase HPLC and amino acid analysis, which was performed after hydrolysis in HC1 for 2 h at 165 "C, using the Waters Pic0 Tag System (Millipore). The concentrations of peptides were determined by quantitative amino acid analysis of concentrated peptide solutions prior to dilution for assay.
Assays for Effects of Peptides on Thrombin Functions-Effects of peptides on several functions of thrombin were determined by the following methods.
Binding of thrombin to thrombomodulin fixed in microwell plates and apparent inhibition constants ( K , ) of peptides in the thrombinthrombomodulin interaction were determined as described previously (8).
Thrombin-induced fibrinogen clotting time was determined using a n Amelung KC-IO instrument as follows. Fifty pl of thrombin (3.3 pg/ml) in 50 mM Tris-HC1, p H 7.5, containing 150 mM NaCl and 5 mM CaCI2, was mixed with 10 pl of various concentrations of peptide.
After a 2-min incubation a t 37 "C, 100 pl of fibrinogen (160 pg/ml) in the same buffer was added to determine the clotting time.
Thrombin-catalyzed Factor V activation in the presence or absence of peptide was determined as follows. Fifty p1 of a mixture of Factor V (1.0 pg) and various concentrations of peptide in 50 mM Tris-HC1, p H 7.5, containing 5 mM CaC12, were incubated with 5 pl of thrombin (0.1 pg/ml) a t 37 "C for 5 min. One pl of the reaction mixture was then mixed with 10 pl of a mixture of prothrombin (165 pg/ml) and bovine brain phospholipids (500 pglml) in 50 mM Tris-HC1, pH 7.5, containing 5 mM CaCl?. Thereafter, 5 p1 of Factor Xa (1.1 ng/ml) in t.he same buffer was added to the mixture and incubated for 3 min. T h e reaction was terminated by the addition of 5 p1 of 250 mM EDTA, then 500 pl of 200 p M H-D-Phe-Pip-Arg-pNA was added and incubated a t 37 "C for 10 min. p-Nitroaniline liberated by the generated thrombin was determined spectrophotometrically a t 405 nm.
Thrombin-induced platelet aggregation was determined using a Sepharose 2B gel-filtered platelet suspension, prepared according to the method of Tangen et al. (14), in a SIENCO platelet aggregometer.
Two hundred pl of platelet suspension (3.2 X lo5 cells/pl) in 50 mM Tris-HC1, pH 7.5, containing 100 mM NaCl in a cuvette was mixed with 20 pI of various concentrations of peptide. Platelet aggregation was then started by the addition of 5 pl of thrombin (5 pg/mI) in 50 mM Tris-HC1, p H 7.5, containing 100 mM NaCl and 50 mM CaCle, and changes in transmittance (%) were measured for 1 min after thrombin addition. Zero and 100% transmittance were assigned as the transmittance of a platelet suspension before and after the addition of thrombin, respectively.
Inhibition of thrombin by hirudin in the presence or absence of peptide was determined as follows: to 20 pl of a mixture of thrombin (0.1 pg) and various concentrations of peptide in 50 mM Tris-HC1, p H 7.5, containing 5 mM CaC12, 10 p1 of hirudin (1.72 pg/ml) was added, and the mixture was incubated at 37 "C for 15 min. Subsequently, 200 pl of 200 p~ H-D-Phe-Pip-Arg-pNA was added, and, after 10 min, liberated p-nitroaniline was determined a t 405 nm.
Inhibition of thrombin by antithrombin 111 in the presence or absence of peptide was determined with H-D-Phe-Pip-Arg-pNA using a modification of the method described previously (8).

Effects of Peptides on Thrombin
Binding to Thrombomoddin-Of the peptides derived from the B-chain of thrombin, FRKSPQELL, RIGKHSRTRYER, TWTANVGKGQPS, DSTRIRI, EGDSGGP, and SWGEGCDRDGK, regions corresponding to the sequences of the latter three peptides are estimated to be located near the active center pocket of thrombin in a three-dimensional model (15), only TWTANVGKGQPS inhibited thrombin binding to thrombomodulin (Table I). To elucidate the amino acids in the TWTANVGKGQPS sequence essential for thrombin-thrombomodulin interaction, the effects of five homologous syn-  thetic peptides, TWTADVGKGQPS, TWTANVGEGQPS, TWTANVGKGEPS, TWTPNVGKGQPS, and TWTANGG-KGQPS, and SPQGKGVNATWT, which is a peptide with the sequence of TWTANVGKGQPS inverted, on thrombin binding to thrombomodulin was examined. As shown in Table  11, replacement of a neutral amino acid residue (Ala'50 + Pro or Val'52 + Gly) in the original dodecapeptide did not change the inhibitory activity of the peptide. However, the activity was significantly reduced when a basic amino acid residue was replaced with an acidic residue + Asp, L y P 4 + Glu, or G1x-1'~~ * Glu). The peptide with the inverted sequence also displayed diminished activity on thrombin binding to thrombomodulin.
Effects of Peptides on Thrombin-induced Fibrinogen Clotting-We then examined the effects of the peptides, TWTANVGKGQPS, RIGKHSRTRYER, DSTRIRI, and EGDSGGP, on thrombin-induced fibrinogen clotting. As shown in Fig. 1, only EGDSGGP did not effectively prolong the clotting time. The concentration at which half-maximum inhibition occurred by the most effective peptide, TWTANV-GKGQPS, for thrombin-induced fibrinogen clotting was 385 PM, which was obtained from a double reciprocal plot for the same data in Fig. 1. Inhibition constants for RIG-KHSRTRYER and DSTRIRI were unobtainable, since inhibition by both peptides was unsaturable up to 1 mM.
Effects of Peptides on Thrombin-induced Factor V Actiuation-As shown in Fig. 2, activation of Factor V depended on the concentration of Factor V in the reaction mixture (inset), and all peptides inhibited thrombin-induced Factor V activation in proportion to the peptide concentration. TWTAN-VGKGQPS was the most effective of the peptides. The concentrations for half-maximum inhibition of thrombininduced Factor V activation of TWTANVGKGQPS, RIGKHSRTRYER, DSTRIRI, and EGDSGGP, which were were incubated with 5 pl of thrombin (0.1 pg/ml) a t 37 "C for 5 min. One pl of the reaction mixture was then mixed with 10 pl of a mixture of prothrombin (150 pg/ml) and bovine brain phospholipids (500 pg/ ml) containing 5 mM CaCI2. Thereafter, 5 pl of Factor Xa (1.1 ng/ ml) was added to the mixture. After a 3-min incubation a t 37 "C, the reaction was terminated by the addition of 5 pl of 250 mM EDTA, then 500 pl of 200 p~ H-D-Phe-Pip-Arg-pNA was added. p-Nitroaniline liberated by the generated thrombin was determined at 405 nm. One unit ( U ) of Factor V is assigned as the activity of Factor V in 1 rnl of human plasma. A calibration curve for Factor V activation by various concentrations of thrombin is shown as an inset in the figure.  Fig. 2, were 33, 55, 81, and 103 pM, respectively.
Effects of Peptides on Thrombin-induced Platelet Activation- Fig. 3 shows that the increase of transmittance during platelet aggregation depended on thrombin concentration (inset), and all peptides appeared to inhibit thrombininduced platelet activation in proportion to the peptide concentration. DSTRIRI was particularly effective, and RIG-KHSRTRYER and EGDSGGP were more effective than T W T A N V G K G QPS. Maximum inhibition for platelet aggregation by DSTRIRI, RIGKHSRTRYER, EGDSGGP, and TWTANVGKGQPS were estimated to be 27%, 22%, 17%, and 15%, respectively, under the present conditions. The concentrations for half-maximum inhibition for platelet activation by DSTRIRI, RIGKHSRTRYER, EGDSGGP, and TWTANVGKGQPS, which were obtained by plotting the reciprocal values of the peptide concentration in Fig. 3, were  370,515, 550, and 645 pM, respectively.
Effects of Peptides on Inhibition of Thrombin by Antithrombin III or Hirudin-None of the peptides affected the inhibition of thrombin by antithrombin 111 in the presence or absence of heparin (data not shown). On the other hand, as shown in Fig. 4, the amidolytic activity of thrombin was inhibited by hirudin (inset), and two peptides, RIG-KHSRTRYER and TWTANVGKGQPS, blocked the inhibition of thrombin by hirudin. The inhibition constants of the two peptides in blocking hirudin-induced thrombin inhibition were unobtainable.

DISCUSSION
The modulation of the substrate specificity of thrombin from a procoagulant to an anticoagulant has been assumed on the one hand to be due to a conformational change in the active center or substrate binding sites of thrombin ( 6 ) , and, on the other, to be due to steric hindrance by thrombomodulin which prevents interaction between thrombin and procoagulant protein substrates (16). In a previous study, we localized the interaction site for thrombomodulin within human thrombin, namely Thr'47 to S~I-'~' of the thrombin B-chain (8).
In the present study, we found that the inhibitory effect of the peptide TWTANVGKGQPS corresponding to residues Thr'47 to Ser15' is sequence-specific and unrelated to the amino acid composition of the peptide, since a peptide with the sequence TWTANVGKGQPS reversed did not inhibit the thrombin-thrombomodulin interaction. We also found that residues AsnI5', LYS'~~, and Gln'" in this region presumably located in the outer loop structure on the surface of the thrombin molecule (15), are essential for binding of thrombin to thrombomodulin. As thrombin has been suggested to bind to the fifth epidermal growth factor-like structure of thrombomodulin (17,18), acidic amino acid residues in the fifth epidermal growth factor-like structure of thrombomodulin may contribute to the interaction with thrombin.
Furthermore, the present study using synthetic peptides corresponding to residues located on the surface and near the active center of thrombin, RIGKHSRTRYER, TWTANVG-KGQPS, DSTRIRI, and EGDSGGP, respectively, Arg'j2 to Arg7', Thr'47 to SerI5$, to Ile'$l, and G1uZo2 to Pro2"', of the B-chain, indicated that the sequence Thr'47 to Ser15' inhibited all the procoagulant activities of thrombin: fibrinogen clotting, Factor V activation, and platelet activation, in addition to inhibiting the thrombin-thrombomodulin interaction. These findings indicate that at least one of the binding sites on thrombin for fibrinogen, Factor V, or platelets is shared with the site for thrombomodulin. These results imply that the antithrombin activity of thrombomodulin is mainly due to steric hindrance which prevents the interaction of thrombin with procoagulant protein substrates.
We also found that other peptides in addition to TWTANVGKGQPS inhibit thrombin-induced Factor V activation and platelet activation. This suggests that there are several interaction sites for Factor V and platelets near the active center of thrombin. Recently, two interaction sites for thrombin in hirudin were revealed one is located in the carboxyl-terminal tail region of hirudin which interacts with the region involved in residues Arg6' to Arg73 of the B-chain of thrombin (19), and the other is located in the amino-terminal core region of hirudin which interacts with a distinct site in thrombin (20). In addition to the report by No6 et al. (7) in which they found the blocking by hirudin of the thrombin-thrombomodulin interaction and also that by Dodt et al. (20) in which they showed that cleavage of the peptide bond at Alalso-Asnlsl of the B-chain of athrombin by pancreatic elastase results in the loss of hirudin binding to thrombin, the present study suggests that another interaction site for hirudin in thrombin is probably located at the same or a proximal site for interaction with thrombomodulin. Recent studies on the crystal structure of the thrombinhirudin complex support these observations (21,22).