Triabin, a highly potent exosite inhibitor of thrombin.

Triabin, a new thrombin inhibitor, has been purified from the saliva of Triatoma pallidipennis, a blood-sucking triatomine bug. It forms a noncovalent complex with thrombin at a molar ratio of 1:1, inhibits thrombin-induced platelet aggregation, and prolongs thrombin clotting time and activated partial thromboplastin time. However, it only minimally suppresses the amidolytic activity of thrombin, as measured by a chromogenic peptide substrate assay. It completely blocks trypsin-catalyzed cleavage of thrombin, probably via protection of the anion-binding exosite and inhibits the effect of thrombomodulin on thrombin in a dose-dependent fashion. These results indicate that the inhibitor is directed toward the anion-binding exosite of thrombin. The protein was partially sequenced and the information used to isolate cDNA clones from a T. pallidipennis salivary gland library. Four slightly polymorphic variants coding for mature proteins of 142 amino acids preceded by a putative leader sequence were obtained. The recombinant protein expressed in the periplasmic space of Escherichia coli has a biological activity similar to that of salivary triabin, as tested in a thrombin-induced platelet aggregation assay. In addition, recombinant triabin inhibits thrombin-catalyzed hydrolysis of fibrinogen with a K of about 3 pM.

Triabin, a new thrombin inhibitor, has been purified from the saliva of Triatoma pallidipennis, a blood-sucking triatomine bug. It forms a noncovalent complex with thrombin at a molar ratio of 1:1, inhibits thrombin-induced platelet aggregation, and prolongs thrombin clotting time and activated partial thromboplastin time. However, it only minimally suppresses the amidolytic activity of thrombin, as measured by a chromogenic peptide substrate assay. It completely blocks trypsincatalyzed cleavage of thrombin, probably via protection of the anion-binding exosite and inhibits the effect of thrombomodulin on thrombin in a dose-dependent fashion. These results indicate that the inhibitor is directed toward the anion-binding exosite of thrombin. The protein was partially sequenced and the information used to isolate cDNA clones from a T. pallidipennis salivary gland library. Four slightly polymorphic variants coding for mature proteins of 142 amino acids preceded by a putative leader sequence were obtained. The recombinant protein expressed in the periplasmic space of Escherichia coli has a biological activity similar to that of salivary triabin, as tested in a thrombin-induced platelet aggregation assay. In addition, recombinant triabin inhibits thrombin-catalyzed hydrolysis of fibrinogen with a K i of about 3 pM.
Evolution has provided hematophageous animals with specific mechanisms to interfere with blood coagulation. Factor Xa inhibitors were isolated from the leech Haementeria officinalis (1) and the tick Ornithodoros moubata (2). Likewise, several proteins which inhibit thrombin have been described. The most widely known is hirudin, which was identified almost 40 years ago in the leech Hirudo medicinalis (3). Thrombin is a key enzyme in the blood coagulation cascade and a strong inducer of platelet aggregation. Three functional domains of thrombin have been described: an active site, a heparin-binding site, and a so-called anion-binding exosite mediating the binding to fibrinogen (4), the platelet receptor (5), and thrombomodulin (6).
The triatomine bug Rhodnius prolixus has been reported to contain a Kazal-type inhibitor which seems to be specifically directed to the active site of thrombin (7). We investigated the related triatomine bug Triatoma pallidipennis, which also belongs to the family of Reduviidae or assassin bugs, and recently described the purification, cloning, and expression of a protein called pallidipin which specifically inhibits collagen-induced platelet aggregation (8). Now we have identified a thrombin inhibitor in the saliva of this insect and report its purification and characterization, and the cloning and expression of the corresponding cDNA. In contrast to the inhibitor from R. prolixus, the protein we isolated from T. pallidipennis and called triabin appears to be directed at the anion-binding exosite of thrombin.

EXPERIMENTAL PROCEDURES
Collection of Saliva-The saliva was obtained from T. pallidipennis bugs reared in captivity. Adult insects were hand stimulated to eject saliva which was collected in siliconized capillary tubes and lyophilized.
Purification of Triabin-The saliva was applied onto a Superose 12 column (Pharmacia Biotech Inc.) equilibrated in 10 mM Tris-HCl, pH 7.4, 0.2 mM CHAPS. 1 The fractions with inhibitory activity were pooled and loaded onto a thrombin-Sepharose 4B column prepared by coupling thrombin to CH-activated Sepharose as described by the manufacturer (Pharmacia). The column was washed with Tyrode's buffer (137 mM NaCl, 2.7 mM KCl, 0.5 mM MgCl 2 , 3.3 mM Na 2 HPO 4 , 5.5 mM glucose, 3.8 mM Hepes, pH 7.4) followed by 10 mM sodium acetate, pH 4.5. The inhibitor was eluted with 10 mM glycine, pH 2.5. The fractions were titrated to pH 7 and pooled. The purity of the preparation was determined by SDS-polyacrylamide gel electrophoresis according to Laemmli (9).
Lys-C Cleavage-45 g of triabin were incubated with 1 g of Lys-C (Boehringer Mannheim) for 6 h at 37°C. The generated peptides were separated by reversed phase high performance liquid chromatography using a Superspher RP column (MZ-Analysentechnik) with a gradient from 0.1% trifluoroacetic acid in water to 0.08% trifluoroacetic acid in 70% acetonitrile. Selected peaks were subjected to amino acid sequencing.
Protein Sequence Analysis-Triabin, modified with vinylpyridine (10), was subjected to automated Edman degradation in a gas-phase sequenator as described by the manufacturer (Applied Biosystems Inc.).
Gel Electrophoresis-SDS-polyacrylamide gel electrophoresis and native electrophoresis in the absence of detergent were performed according to Laemmli (9) and Maurer (11), respectively. Proteins were stained with Coomassie Brilliant Blue.
Determination of Protein Concentration-The method of Bradford (12) was used to determine protein concentration.
Partial Proteolysis of Thrombin-60 pmol of thrombin (Sigma) were preincubated either with 70 pmol of triabin or buffer (100 mM Tris-HCl, pH 7.8, 150 mM NaCl) for 5 min at 37°C followed by the addition of 40 ng of trypsin. After 1 h incubation at 37°C the reaction was stopped with 0.5 g of Pefabloc (E. Merck). The samples were boiled in sample buffer and electrophoresed according to Laemmli (9).
Activity Assays-Fibrinogen clotting activity was measured as described previously (13). This assay was used for the identification of active fractions during the purification procedure.
Aggregation of washed platelets was measured as described earlier (8). Briefly, thrombin (human from Sigma, 1 nM) and triabin or hirudin preincubated for 1 min at 37°C were added to washed platelets and aggregation was monitored in an aggregometer (Micron).
To assay amidolytic activity, 20 l of thrombin (human from Sigma, 45 pM final concentration) were incubated with 80 l of 10 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 100 l of inhibitor or buffer for 2 min at 37°C * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EMBL Data Bank with accession number(s) X80246.
For thrombin clotting time assays, 100 l of thrombin (Behring, 3 IU/ml) and 100 l of inhibitor or diethylbarbiturate acetate buffer (Behring) were incubated for 1 min at 37°C. After addition of 100 l of plasma (control plasma N from Behring), the time until a clot formed was measured. The same method was used to determine reptilase activity but 200 l of reptilase (Sigma, 5 g/ml) was added instead of 100 l of thrombin.
To determine activated partial thromboplastin time (APTT), 100 l of pooled citrated human plasma, 10 l of inhibitor, and 100 l of APTT reagent (Pathromtin, Behring, Germany) were incubated for 3 min at 37°C. After addition of 100 l of prewarmed 25 mM CaCl 2 , the time until a clot formed was measured in a fibrometer (Sarstedt Biomatic B10).
To measure factor Xa and trypsin activities, triabin was incubated with factor Xa (American Diagnostica Inc.) or with trypsin (Sigma) up to a molar ratio of 43:1 for 2 min at 37°C followed by the addition of N-benzoyl-Ile-Glu-Gly-Arg-p-nitroaniline (S2222, Kabi Vitrum, final concentration 0.167 mM). After 20 min at 37°C, the absorption at 405 nm was measured and compared to the control value without triabin.
Cloning of Triabin cDNA-A tag was first generated by polymerase chain reaction, using T. pallidipennis salivary gland cDNA (8) as template. Sense (5Ј-GCNGARGGNGAYGAYTGYTCNCTNGARAARGCN-3Ј) and antisense (5Ј-TTRTTRTANCCRATYTCNACRATYTGNGT-RAA-3Ј) primers deduced from the NH 2 -terminal and an internal region of triabin were used in 40 cycles of 2 min denaturation at 94°C, 1 min 30 s annealing at 52°C, and 2 min elongation at 72°C. The amplified 200-base pair long fragment was subcloned and sequenced using the dideoxy chain termination method (15). A screening probe was derived from it using the Taq polymerase (Perkin Elmer) in 5 cycles of 2 min 30 s at 94°C, 2 min at 40°C, and 3 min 30 s at 72°C, in the presence of labeled dCTP (Amersham) and using the antisense primer. A T. pallidipennis cDNA library (8) was screened after plating on nylon filters (Biodyne A, Pall) under high stringency conditions (16) and positive clones were analyzed after in vivo excision. The complete DNA sequence of both insert strands was determined for the selected cDNAs.
Expression of Recombinant Triabin-The construction of the pKK/ cph expression vector, which is derived from pKK233-2 into which a sequence derived from the signal peptide coding part of secreted bacterial cyclophilin (17) was introduced, has been described (18). The region coding for mature triabin (clone Tr28) was amplified by polymerase chain reaction (10 cycles of 2 min at 94°C, 2 min at 53°C, and 2 min 30 s at 72°C) while introducing new flanking NheI and HindIII sites, using appropriate primers (5Ј-GCGATAGCTAGCAGAAGGTGACGAC-3Ј; 5Ј-GCGATAGGATCCAAGCTTACTAACAAATTTCATTAGCAT-CAGG-3Ј). The polymerase chain reaction product was subcloned into pKK/cph, downstream of the cyclophilin signal peptide coding region and the resulting construct checked by DNA sequencing. Transformation was into Escherichia coli JM105 and induction was carried out on a bacterial culture grown to an A 600 of 0.6 by adding 1 mM isopropyl-1thio-␤-D-galactopyranoside and shaking for 6 h at 37°C. The bacteria were centrifuged, resuspended in one-tenth volume of 0.5 M saccharose, and equilibrated for 10 min at room temperature. They were then centrifuged and the pellet was resuspended in ice-cold water and left for 10 min on ice. Following the osmotic shock, the cells were centrifuged and the supernatant which corresponds to the periplasmic fraction was recovered.
Purification of Recombinant Triabin-The periplasmic fraction was first loaded on a DE-52 cellulose column for ion-exchange chromatography. The column was equilibrated and washed with 10 mM Tris-HCl, pH 7.5, 0.075 M NaCl and the elution carried out in 10 mM Tris-HCl, pH 7.5, with a 0.6 M and a 1 M NaCl step. Active fractions were further purified by affinity chromatography on a thrombin-Sepharose column as described above.
Kinetic Analysis of Recombinant Triabin-The concentration of the fibrinogen stock solution and the triabin solution were calculated from the absorbance at 280 nm using A 280 1% ϭ 15.1 (19) and ⑀ ϭ 12,450 M Ϫ1 cm Ϫ1 . For the determination of the K m , 1.5-40 M fibrinogen were incubated with 75 pM thrombin in 50 mM phosphate, pH 7.4, 100 mM NaCl, 0.1% polyethylene glycol 6000. To measure the K i , 25-200 pM triabin were incubated with 75 pM thrombin and the reaction was started with 10 M fibrinogen. In both studies aliquots were taken after 0, 1, 1.5, 2, 2.5, and 4 min, and the reaction was stopped by the addition of H-D-Phe-Pro-Arg-chloromethylketone (Bachem, final concentration 170 M) followed by heating at 56°C for 5 min. The amount of fibrinopeptide A released was measured in a competition enzyme-linked immunosorbent assay, as stated by the manufacturer (Accurate Chemical and Scientific Corp.). Briefly, fibrinogen was removed after precipitation with ethanol. The supernatants or standards (52-829 nM) were incubated with peroxidase-labeled antibodies (8C2-5) against fibrinopeptide A and applied to a microtiter plate coated with antigen. After 30 min incubation, peroxidase activity was determined using o-phenylenediamine/H 2 O 2 as substrate. The velocities of fibrinogen cleavage were determined from the gradient of fibrinopeptide A concentration between 0 and 4 min. The K m was determined by nonlinear regression analysis of velocity versus substrate concentration according to Duggleby (20) using the computer program k⅐cat (Europa Scientific Software Corp.). The K i was graphically determined by the method of Dixon (21).

RESULTS
The saliva of T. pallidipennis was found to inhibit not only platelet aggregation induced by collagen but also aggregation mediated by thrombin. These two inhibitory activities were separated by gel filtration although the two proteins have similar molecular masses. This was achieved by using a Superose 12 column with a buffer containing only low salt concentrations which led to a retardation of pallidipin but not of the thrombin inhibitor. The thrombin inhibitor was further purified by affinity chromatography on thrombin Sepharose. 11 g of purified inhibitor, which we named triabin, were recovered from 10 mg of dried saliva. The purity of this preparation was ascertained by SDS-polyacrylamide electrophoresis, which showed a single band with an apparent molecular mass of 21 kDa under reducing conditions and 18 kDa under nonreducing conditions (Fig. 1, lanes 7 and 8).
In order to investigate the triabin-thrombin complex, we incubated triabin with thrombin at different molar ratios and analyzed the probes by nondenaturing polyacrylamide gel electrophoresis. As shown in Fig. 1, the inhibitor was totally bound in a complex with thrombin when the molar ratio of triabin to thrombin was 1:1, whereas free triabin was detected at a ratio of 1. 5:1 (lanes 3 and 4). This complex was not covalently linked since it could be separated by SDS-polyacrylamide gel electrophoresis (lane 6).
Triabin inhibited thrombin-induced platelet aggregation in a dose-dependent fashion (Fig. 2). If platelet aggregation was induced by 1 nM thrombin, total inhibition was observed at 1 nM triabin. The dose-response curve was almost identical to that of hirudin. In contrast, the COOH-terminal tail of hirudin, hir(54 -65), also called hirugen, exhibited a nearly 500 times lower activity (Fig. 2). No effect of triabin was noted on the aggregation induced by collagen or ADP (not shown). Triabin inhibited not only platelet aggregation but also interfered with the other known activities of thrombin. A 4-fold prolongation of thrombin-clotting time was measured with 22 nM triabin (Fig.  3). Clotting of plasma can also be catalyzed by reptilase, an enzyme from the snake venom of Bothrops atrox (22). Triabin did not inhibit this enzyme (data not shown). Activated partial thromboplastin time was prolonged 4-fold by about 600 nM triabin and about 400 nM hirudin. While hirudin already showed an effect at 10 nM, triabin was only active at concentrations higher than 500 nM (Fig. 3). In contrast to its strong inhibitory effect in the assays described above, triabin hardly affected thrombin activity in an amidolytic assay using a synthetic peptide substrate (Fig. 4). Only a 35% inhibition was achieved with a 3-fold molar excess of triabin. Higher concentrations up to a 67-fold molar excess did not lead to an increase of this inhibitory effect. This is in contrast to hirudin, which totally blocked the amidolytic activity of thrombin at low concentrations (Fig. 4). Other serine proteases, e.g. trypsin and factor Xa, were not inhibited in an amidolytic assay even when using a 45-fold molar excess of triabin (data not shown).
These results demonstrate that triabin is specific for thrombin. The difference in its inhibitory activity using either fibrinogen or a small synthetic peptide as substrate indicates that it may exert its inhibitory effect by binding to the anion-binding exosite rather than to the catalytic site of thrombin.
Limited trypsin proteolysis converts ␣-thrombin into ␤-thrombin through cleavage of the Arg 73 -Asn 74 bond within the anion-binding exosite (23). Preincubation of thrombin with triabin prevented the cleavage by trypsin as analyzed by SDSpolyacrylamide gel electrophoresis (data not shown). Since triabin did not inhibit trypsin, the effect was probably achieved by binding of triabin to the exosite of thrombin.
Thrombomodulin is known to exhibit its effect on thrombin by binding to the anion-binding exosite (6). The resulting change of substrate specificity enables thrombin to activate protein C. This activity can be tested by incubating thrombomodulin and thrombin with the inactive preform of protein C. This leads to activation of protein C which is detectable by its amidolytic activity. Triabin neither showed a thrombomodulinlike activity, i.e. it did not modulate thrombin, nor did it acti-vate protein C. However, it was able to antagonize the effect of thrombomodulin in the protein C activation assay in a dose-dependent manner (Fig. 5). This observation further corroborates our hypothesis that triabin binds to this exosite of thrombin.
Twenty amino-terminal amino acids of triabin were sequenced and additional internal sequences were determined after cleavage with endoproteinase Lys-C and separation of the fragments by high performance liquid chromatography. Degenerate primers were made based on the protein sequence information and used to amplify a specific tag from T. pallidipennis salivary gland first-strand cDNA. The probe was employed to screen approximately 1 ϫ 10 6 primary phage clones from a T. pallidipennis salivary gland cDNA library. DNA sequencing revealed that four different types of triabin cDNA had been isolated (Tr5, Tr12, Tr28, and Tr45). The clone with the longest insert is depicted in Fig. 6. It exhibited an open reading frame potentially coding for a 160-amino acid long protein. Hydrophilicity analysis showed that the 18 NH 2 -terminal residues displayed the hallmarks characteristic of a signal peptide (24). The following position was occupied by the alanine residue representing the NH 2 terminus of triabin isolated from T. pallidipennis saliva. Mature triabin is therefore a 142-amino acid long protein with a predicted relative molecular mass of 15,948, slightly less than that of the saliva protein which is probably glycosylated. Indeed a potential N-glycosylation site was detected in the deduced amino acid sequence. The stretches previously determined by protein sequencing were retrieved in the amino acid sequence deduced from the cDNA but two discrepancies were noted at positions 8 and 77 where leucine and aspartic acid found in the protein were replaced by isoleucine and glycine, respectively. The untranslated region downstream of the open reading frame contained a sequence almost exactly matching a polyadenylation signal (25) in the appropriate position upstream of a poly(A) tail.
In addition to Tr5, cDNAs coding for three variant forms of triabin were also isolated. The deduced mature proteins had the same length but differed slightly in their sequence. The degree of conservation was, however, very high since there was over 94% amino acid sequence identity between the mature forms. In total, there were only 9 positions where all four proteins were not identical and in four of these positions the amino acid exchange was conservative (Tr12: 8L, 77D, 86G, 114I, 127F, and 139N; Tr28: 8L 32D, 77D, 86G, 114I, 127F and 139N; Tr45: 8L, 32D, 72A, 74K, 77D, 86G, 127F, and 139N). It can therefore be assumed that the biological function of the variants is identical. Unlike the product of Tr5, the three additional forms exactly matched the partial sequences directly determined on the natural salivary protein. A search in the SwissProt data base did not reveal any significant homology between triabin and other known proteins (the best score was 19% identity for an 83-amino acid long overlap).
A periplasmic expression construct based on the pKK233 plasmid and harboring the isopropyl-1-thio-␤-D-galactopyranoside-inducible trc promoter was devised. The coding sequence of mature triabin was fused to the signal peptide coding part of secreted bacterial cyclophilin, modified so as to optimize the leader peptidase cutting site (18,26). The Tr28 cDNA was selected for these studies because the protein it encoded showed no difference to the protein sequence data and most closely corresponded to a consensus triabin sequence inferred from the four variants. The periplasmic fraction of transformed E. coli was prepared. The fibrinogen clotting test indicated the presence of an inhibitory activity in this fraction, which was not observed with mock-transformed bacteria.
Recombinant triabin was purified using a two-step procedure involving a separation on DE-52 cellulose and affinity chromatography on a thrombin-Sepharose column. The purity of the protein was verified by SDS-polyacrylamide gel electrophoresis and a single band corresponding to a molecular mass of about 18 kDa, comparable to that of saliva triabin was observed (not shown). The identity was further checked by determination of the NH 2 -terminal sequence and the protein was found to start with the expected alanine residue, indicating that correct processing had taken place.
The biological activity was evaluated in a platelet aggregation assay (Fig. 2). A dose-response study showed that recombinant triabin inhibited aggregation induced by 1 nM thrombin in a dose-dependent manner, the IC 50 being 0.82 nM. A kinetic analysis of thrombin-catalyzed fibrinogen cleavage was performed by measuring fibrinopeptide A release, and the K m was found to be 14.0 Ϯ 3.7 M. The inhibitory effect of recombinant triabin was assessed in this test, and the K i measured. Since in this analysis, the amount of inhibitor bound by the enzyme was no longer negligible, the graphical method of Dixon (21) was used as it avoids the assumption that the total concentration of inhibitor is equal to the theoretically required free concentration and is therefore also applicable for tight-binding inhibitors. The K i was found to be 3 pM (Fig. 7). DISCUSSION Blood-sucking insects employ various mechanisms to interfere with the blood coagulation of their host species. The saliva of T. pallidipennis inhibits collagen-induced platelet aggregation by pallidipin (8) and ADP-mediated aggregation by an FIG. 5. Protein C activation assay. Triabin, thrombomodulin, thrombin, and protein C were incubated for 1 h at 37°C. After quenching the reaction with hirudin, S2266 was added and the generation of p-nitroaniline was monitored. Results are expressed as percentages of protein C activity without inhibitor. apyrase. 2 Now we report a third anticoagulatory mechanism, thrombin inhibition by a protein which we named triabin.
Triabin forms a complex with thrombin, probably in a molar ratio of 1:1. This complex is noncovalent, hence it is formed via a mechanism different from the physiological heparin-antithrombin III-thrombin interaction. Other inhibitors from hematophageous animals also form noncovalent complexes with thrombin but they differ in their mechanism of action. Hirudin binds to the active site and with its COOH-terminal tail to the anion-binding exosite (27), whereas rhodniin seems to affect mainly the active site (7). Bothrojaracin, isolated from the venom of Bothrops jararaca (28), exhibits its inhibitory effect exclusively via binding to the anion-binding site. All these inhibitors interfere with thrombin's activity on fibrinogen and on the platelet receptor. Whereas triabin has the same high inhibitory activity as hirudin, it is about 500 times more potent than hirugen, the COOH-terminal tail of hirudin. Compared to bothrojaracin (28), the IC 50 values for inhibition of platelet activation are 7 times lower for triabin. In contrast to the active site inhibitors, proteins specific for the anion-binding exosite should not affect the cleavage of chromogenic peptide substrates, which only interact with the active site. Indeed, bothrojaracin and hirugen do not decrease the amidolytic activity of thrombin. However, triabin partially suppresses this activity. Since this effect is low compared to its inhibitory activity on fibrinogen clotting and platelet aggregation, it is conceivable that it binds only to the anion-binding exosite. One explanation for the only partial inhibition of amidolytic activity may be that triabin sterically hinders the access of small peptides to the active site. Another possibility is that the binding of triabin to thrombin changes the conformation of the enzyme, resulting in lower activity. Proteolysis of thrombin within the anion-binding exosite also causes an inhibition of fibrinogen cleavage and a small decrease of the affinity to peptide substrates (6). Hofsteenge et al. (6) suggested that the proteolysis leads to a small conformational change in the thrombin molecule which affects the peptide-binding residues.
Limited proteolysis of ␣-thrombin by trypsin results in the cleavage of the Arg 73 -Asn 74 bond, which is located in the anionbinding exosite, and generates ␤-thrombin (23). Triabin was able to protect this site against cleavage by trypsin. Since a direct inhibition of trypsin was excluded, it is probably the binding of triabin to this site which is responsible for this effect.
Similar experiments were performed with hirudin and its COOH-terminal fragment, which likewise inhibited the cleavage (29), and x-ray diffraction analysis showed that indeed these peptides bind to the anion-binding exosite (30). These results lend additional support to our hypothesis that triabin binds to thrombin in a similar manner.
Binding of thrombomodulin to thrombin causes a change in the specificity of the protease. Hofsteenge et al. (31) demonstrated that thrombomodulin inhibits the binding of fibrinogen to thrombin but hardly affects the binding of peptide substrates. Cleaving ␣-thrombin at the Arg 73 -Asn 74 bond reduces its affinity to thrombomodulin (6). Additionally, an antibody binding to residues 62-73 of thrombin inhibits the activation of protein C by ␣-thrombin in the presence of thrombomodulin (6). These results indicate that the region around Arg 73 , the anionbinding exosite, is necessary for the interaction with thrombomodulin. Therefore, it was of interest to investigate the effect of triabin on this interaction as well. Indeed, triabin inhibited in a dose-dependent manner the thrombomodulin-mediated activation of protein C by thrombin. This result further corroborates the hypothesis that triabin binds to the anion-binding exosite.
Four different cDNA clones with minor polymorphisms that probably do not affect the biological function of triabin were isolated. This diversity might reflect the existence either of a protein family or more likely of a genetic heterogeneity among the insects from which the salivary glands were collected. No sequence homology was found between triabin and the thrombin inhibitors hirudin and rhodniin. This is in keeping with the different modes of action of these proteins.
A vector for periplasmic expression was used for the production of triabin, in order to favor correct folding of the protein and facilitate the subsequent purification. Our results show that in both the fibrinogen clotting test and the thrombininduced platelet aggregation assay, the biological activity of recombinant triabin purified from the periplasm of E. coli was similar to that of saliva triabin, suggesting that the protein has assumed its natural configuration.
The inhibition constant of 3 pM, which was measured in the thrombin-catalyzed fibrinogen cleavage assay, indicates a very tight binding of triabin to thrombin. For the hirudin-thrombin complex, dissociation constants of 63 pM (32), 3 pM (33), and 20 fM (34) have been cited in the literature. All these values were determined with an amidolytic substrate which could not be used for this study because triabin does not seem to block the active site. However, the value for triabin agrees well with the K i found by Dodt et al. (33). In contrast, the COOH-terminal hirudin peptide Hir 53-64 , which, like triabin, recognizes the anion-binding site of thrombin, has been shown to be less active in the thrombin-catalyzed hydrolysis of fibrinogen, the K i being 144 nM (35).
In summary, we have purified a thrombin inhibitor with a potency comparable to hirudin but with a different mode of action. The availability of recombinant triabin will now allow us to further explore its mechanism of action and carry out structure-function studies using site-directed mutagenesis in order to identify the structural elements involved. X-ray diffraction analysis should provide detailed information on the structure of the triabin-thrombin complex. Furthermore, in vivo models will be tested to assess the effectiveness of triabin as an antithrombotic.
Acknowledgments-We are indebted to J. Milde, L. Dummann, and C. Schlegel for excellent technical assistance, and to D. Sauvageot for oligodeoxynucleotide synthesis. The participation of Dr. U. Hechler and S. Gruner in the early phases of the project is acknowledged. We thank Dr. J. Krä tzschmar for valuable comments and for supplying the T. pallidipennis cDNA library, Dr. John Morser for help with the protein C assay, and Felipe Olvera and Alejandro Olvera for saliva collection.