Dissociation of Thrombin's Substrate Interactions Using Site-Directed Mutagenesis

Ungerer M, Kessebohm K, Kronsbein K, et al.: 1996. Activation of beta-adrenergic receptor kinase during myocardial ischemia. Circ Res 79:455–460. Winstel R, Freund S, Krasel C, et al.: 1996. Protein kinase cross-talk: membrane tar-

geting of the beta-adrenergic receptor kinase by protein kinase C. Proc Natl Acad Sci USA 93:2105–2109.

PII S1050-1738(00)00053-0

TCM

Dissociation of Thrombin’s Substrate Interactions Using Site-Directed Mutagenesis Lawrence L.K. Leung* and Scott W. Hall

Thrombin is an allosteric enzyme that interacts with multiple procoagulant substrates such as specific clotting factors and cell surface thrombin receptors, as well as the anticoagulant substrate protein C. Functional mapping of thrombin’s interactions with its various substrates has been carried out using a collection of thrombin mutants generated by systematic alanine scanning mutagenesis. A thrombin mutant, E229K, has been identified that has essentially lost all of its procoagulant properties while retaining its ability to activate protein C, thus functioning as an anticoagulant in vitro and in vivo. It is also found that specific and distinct domains are involved in thrombin’s interaction with thrombomodulin (TM) and the subsequent activation by the thrombin/TM complex of protein C and the thrombin-activatable fibrinolysis inhibitor (TAFI). (Trends Cardiovasc Med 2000;10:89–92). © 2000, Elsevier Science Inc.

At the site of vascular injury, tissue factor is exposed which initiates the activation of the clotting cascade, culminating in the formation of thrombin. Thrombin is a serine proteinase that can interact with a large number of macromolecular substrates that have both procoagulant and anticoagulant effects. Procoagulant activities include the conversion of fibrinogen to fibrin, activation of the factors V, VIII, and XI to accelerate the rate of thrombin generation, activation of the

Lawrence L.K. Leung and Scott W. Hall are at the Division of Hematology, Stanford University Medical School, Stanford, California, USA. *Address correspondence to: Lawrence L.K. Leung, Division of Hematology, CCSR 1155, Stanford University Medical School, Stanford, CA 94305-5156. © 2000, Elsevier Science Inc. All rights reserved. 1050-1738/00/$-see front matter

TCM Vol. 10, No. 2, 2000

transglutaminase factor XIII which mediates cross-linking of fibrin monomers, and the cleavage of the platelet thrombin receptors (protease-activated receptors PAR-1 and PAR-4) leading to platelet activation (Kahn et al. 1999). On the other hand, when thrombin binds to thrombomodulin (TM), an integral membrane protein found on endothelial cells especially in the microcirculation, it undergoes a remarkable transformation. Thrombin undergoes conformational changes such that it loses its procoagulant enzyme activities, and acquires the ability to activate protein C, a serine protease zymogen in plasma (Esmon and Schwartz 1995). Recent studies by Esmon and coworkers show that the activation of protein C by the thrombin-TM complex is enhanced approximately 10-fold by binding of protein C to an endothelial protein C receptor (EPCR) (Esmon et al. 1999). Activated

protein C, in the presence of its cofactor protein S, cleaves and inactivates the activated factors V and VIII, thereby dampening the clotting cascade. The in vivo importance of this natural anti-thrombotic pathway is amply supported by the clinical hypercoagulable state found in patients who are heterozygous deficient in either protein C or S, or have the factor V Leiden mutation (R506Q), in which one of the major aPC cleavage sites is mutated, rendering activated factor V Leiden relatively resistant to aPC inactivation. Recently, a new physiologic substrate for the thrombin-TM complex has been identified and termed plasma procarboxypeptidase B or U, or thrombin-activatable fibrinolysis inhibitor (TAFI) (Bajzar et al. 1995, 1996, Redlitz et al. 1995). Activated TAFI functions as a fibrinolysis inhibitor by cleavage of C-terminal lysines from partially digested fibrin, thus diminishing the incorporation and activation of plasminogen, leading to delayed clot lysis. From a physiologic standpoint, one can envisage that upon thrombin binding to TM, the thrombinTM complex will activate protein C to dampen the clotting cascade, and will also activate TAFI, thereby protecting the clot already formed at the vascular wound from premature degradation (Figure 1). Of note, the concentration of thrombin required for the activation of TAFI is substantially higher than that for fibrinogen clotting. Thus the feedback activation of factors V, VIII, and XI by the initial small amount of thrombin to accelerate and amplify thrombin generation through the clotting cascade appears to be critical for TAFI activation. The bleeding diathesis encountered in hemophiliacs and patients deficient in factor XI may be partly related to the suboptimal activation of TAFI, leading to enhanced clot lysis (Broze and Higuchi 1996, Minnema et al. 1998). Similarly to protein C, the activation of TAFI by the thrombin-TM complex is approximately 1000-fold faster than by free thrombin alone. Unlike protein C, TAFI is a metalloprotease, has a short half-life (approximately 10 min), and its instability may be a significant regulator of its activity in vivo (Boffa et al. 1998). TAFI has been “knocked-out” in the mouse by homologous recombination, but its phenotype has not yet been described in detail. Patients with either TAFI deficiency with an expected bleeding pheno-

89

Figure 1. Activation of protein C and TAFI by the thrombin-TM complex on endothelial cell surface at the site of vascular injury.

type or excess amount of TAFI with a thrombotic phenotype have not been described to date. The potential in vivo importance of TAFI in fibrinolysis has recently been reported by two groups (Bajzar et al. 1999, Zhao et al. 1999). Using rabbit thrombosis models, both groups demonstrated enhanced clot lysis by tPA in the presence of potato carboxypeptidase inhibitor (PCI), an inhibitor of endogenous TAFI. In both venous and arterial thrombi, enhanced clot lysis could be obtained with lower doses of tPA in the presence of PCI than with high dose tPA alone. There was no apparent increase in bleeding in these animals receiving tPA and PCI, as opposed to animals receiving high dose tPA. These data suggest that in vivo inhibition of TAFI may enhance the thrombolytic efficacy of tPA in patients with acute coronary syndrome or stroke, allowing for dose reduction and improvement in bleeding complications. Insights into the diverse specificity of thrombin have come from structural analysis of crystallized thrombin-inhibitor complexes (Rydel et al. 1990). After cleavage by the prothrombinase complex, prothrombin is converted to a-thrombin, which consists of an A and B chain linked by a disulfide bond. Thrombin is a compact globular protein, with a deep, narrow canyon-like cleft, termed the active-site cleft, on one side of the molecule. The active site of the enzyme, consisting of the classic serine proteinase catalytic triad, H43, D99, and S205 (human thrombin B chain numbering), is located deep within this cleft (Figure 2). Access to the active site is limited by two “insertion loops,” so-called since these

90

are sequences unique to thrombin relative to chymotrypsin. One insertion loop (L45-N57) with a prominent tryptophan residue (W50) is located on the upper border of the active-site cleft, while the other one (L144-G155, “autolysis loop”) protrudes from the lower rim of the cleft. Additional enzyme specificity is conferred by a positively charged domain, termed exosite I, located right next to the active-site cleft. It has a cluster of cationic residues that are important in mediating the docking of thrombin to many of its physiologic substrates, including fibrinogen, TM, thrombin receptor, and heparincofactor II. Exosite II is a second posi-

tively charged domain on thrombin, which is located on the other side of the thrombin molecule and is important in thrombin binding to heparin and glycosaminoglycans. In addition to these domains, there is a sodium-binding site on thrombin as defined by DiCera et al. (1995). It is mediated by the carbonyl oxygens of Y190, R233, and K236 and is located on the right lower rim of the active-site cleft. Sodium binds to thrombin with a Kd approximating the plasma sodium concentration, suggesting that thrombin will equilibrate between the sodium-bound and sodium-free forms, with the former having a higher specificity for fibrinogen and the latter being more protein C selective. • Functional Mapping of Thrombin and Identification of Protein C Activators Functional mapping of thrombin has been carried out initially by selective sitedirected mutagenesis and more recently by systematic alanine scanning mutagenesis (Tsiang et al. 1995). In the latter strategy, 77 charged or polar surface residues determined to be accessible to a solvent probe were replaced by alanine, resulting in about 55 thrombin mutants that carry either single, double or triple

Figure 2. Localization of structural domains on the surface of thrombin. Space-filling model of human thrombin in the standard orientation, looking directly into the active-site cleft. Catalytic residue S205 (red) is located deep within the cleft. The W50 insertion loop L45-N57 (light blue) is located above the cleft while insertion loop L144-155 (autolysis loop) (purple) protrudes from the lower rim. Exosite I residues (green) lie to the right of the active site and exosite II residues, lying above and extending to the other hemisphere of the molecule, are depicted in royal blue.

TCM Vol. 10, No. 2, 2000

mutations. This collection of thrombin mutants has been used to map the binding sites required for fibrinogen clotting and TM-dependent protein C activation, the two prototypic procoagulant and anticoagulant functions of thrombin. All functional residues involved in these two activities were mapped to the hemisphere of the molecule containing the activesite cleft and exosite I, with considerable topographical overlap between the two functional domains. Upon closer analysis of the functional residues, it was discovered that four mutants (W50, K52, E229, and R233) displayed a significant substrate shift in favor of protein C activation. These mutants represent a novel class of anticoagulant, which acts by activating the endogenous protein C pathway and have thus been termed protein C activators. The best protein C activator, as defined by saturation mutagenesis of the four mutants, is E229K (Gibbs et al. 1995, Tsiang et al. 1996). It has lost almost all of its procoagulant functions, with a catalytic efficiency for fibrinopeptide A (FPA) release of 0.4% compared to wildtype thrombin, while retaining 50% efficiency for protein C activation, representing a 130-fold switch in substrate specificity. It has an 18-fold reduction in platelet aggregation, and produced reversible anticoagulation as demonstrated by prolongation of PTT when infused into Cynomolgus monkeys without concomitant consumption of platelets or clotting factors. It had also significant antithrombotic efficacy in reducing platelet and fibrin deposition in a well-established baboon AV shunt model (Hanson et al. 1997). E229 is situated near the sodiumbinding site, and its mutation resulting in such dramatic dissociation of thrombin’s fibrinogen clotting and protein C activation may be partly related to its disruption of sodium binding by thrombin.

Figure 3. Localization of residues in thrombin mediating its interaction with TM, protein C, and TAFI. Space-filling model of thrombin with the same orientation as in Figure 2. Residues implicated in interacting with TM are green, residues important in TAFI activation are royal blue, and residues involved in protein C activation are yellow. The light blue residue is Trp50 (W50) which is involved in both PC and TAFI activation.

tional analyses of thrombin–substrate interactions. Comparison of protein C activation in the presence and absence of TM identified 11 residues mediating the thrombin–TM interaction. These are all located in exosite I (Figure 3). Three mutants (E25A, D51A, R98A/R93A/E94A) had decreased ability to activate TAFI

yet retained normal protein C activation, indicating they represent a TAFI-binding domain, and are located above the activesite cleft. On the other hand, three other mutants (R178A/R180A/D183A, E229A, R233A) had decreased protein C activation but normal TAFI activation, and are located below the active-site cleft, clus-

• Specific Binding Domains on Thrombin for TM, Protein C, and TAFI The library of thrombin mutants was recently used to compare the activation of the TM-dependent substrates protein C and TAFI (Hall et al. 1999). The mutants were expressed in stable cell lines and purified to homogeneity to provide a collection of readily usable reagents that can broadly be used for structural-funcTCM Vol. 10, No. 2, 2000

Figure 4. Models of the thrombin-TM complex interacting with protein C and endothelial protein C receptor (EPCR) (left) and TAFI (right). The EGF-like domains 4 and 5 of TM bind to thrombin’s exosite I and the glycosaminoglycan moiety of TM binds to thrombin’s exosite site II.

91

tering around the sodium-binding pocket. One mutant (W50A) displayed decreased activation of both protein C and TAFI. Trp50 (W50) appears to be a critical residue for all of the activities of thrombin, as mutation or deletion of this residue results in diminished small molecule substrate hydrolysis, fibrinogen clotting, and antithrombin III inhibition as well. Thus, in contrast to the extensive overlap of residues mediating TM binding and fibrinogen clotting, there are specific and distinct domains involved in thrombin’s interactions with TM, protein C, and TAFI. TM is an integral membrane protein with its extracellular portion consisting of an N-terminal lectin domain followed by six EGF-like domains. Previous studies with EGF deletion mutants demonstrated that EGF5-6 of TM binds to thrombin with high affinity but does not enhance PC activation (Parkinson et al. 1992, Tsiang et al. 1992, Ye et al. 1992). Protein C binding to the thrombin-TM complex requires EGF-like domain 4, whereas TAFI binding requires EGF-like domain 3 (Nagashima et al. 1993, Kokame et al. 1998). These data, taken together with the current thrombin mapping data, suggests a model for formation of the ternary complexes (Figure 4). EGF-like domains 5 and 6 would mediate the binding of TM to thrombin exosite I and orient EGFlike domains 4 and 3 so that they are adjacent to the lower and upper lips of the active-site cleft. EGF-like domain 4 would then align protein C along the lower cleft, where it would interact with residues around the sodium-binding pocket. This assembly process is further augmented by the endothelial protein C receptor, such that thrombin and protein C would be bridged on opposite sides by TM and the endothelial protein C receptor in a quaternary complex. On the other hand, EGF-like domain 3 would align TAFI along the upper lip of the active-site cleft to contact the acidic residues in and around the W50 insertion loop. This model could be tested directly using TM alanine mutants (Nagashima et al. 1993) in conjunction with thrombin mutants or TAFI mutants to identify the critical sites of interaction between all these molecules. In summary, while thrombin is clearly an allosteric enzyme and interacts with a large number of procoagulant and anticoagulant substrates, studies to date using site-directed mutagenesis indicate

92

that one can successfully identify candidate residues that effectively dissociate one thrombin activity from another. Since knock-out of the prothrombin gene in mice results in either embryonic lethality or death soon after birth from massive hemorrhage (Xue et al. 1998, Sun et al. 1998), introduction of (pro)thrombin mutants with selected deficient activities may allow one to salvage these mice and examine the interplay of the numerous and complex macromolecular interactions of thrombin in vivo. • Acknowledgments This work was supported in part by NIH grant RO1 HL57530 and the Cheong Har Family Foundation.

References Bajzar L, Liao P, Vlasin P, et al.: 1999. The safety and efficacy of a TAFIa inhibitor in a rabbit arterial thrombolysis and bleeding model. Blood 94(Suppl 1):22a. Bajzar L, Manuel R, Nesheim M: 1995. Purification and characterization of TAFI, a thrombin-activable fibrinolysis inhibitor. J Biol Chem 270:14,477–14,484. Bajzar L, Morser J, Nesheim M: 1996. TAFI, or plasma procarboxypeptidase B, couples the coagulation and fibrinolytic cascades through the thrombin-thrombomodulin complex. J Biol Chem 271:16,603–16,608. Boffa MB, Wang W, Bajzar L, et al.: 1998. Plasma and recombinant thrombin-activable fibrinolysis inhibitor (TAFI) and activated TAFI compared with respect to glycosylation, thrombin/thrombomodulin-dependent activation, thermal stability, and enzymatic properties. J Biol Chem 273:2127–2135. Broze GJ, Higuchi DA: 1996. Coagulationdependent inhibition of fibrinolysis: role of carboxypeptidase U and the premature lysis of clots from hemophilic plasma. Blood 88:3815–3823. Di Cera E, Guinto ER, Vindigni A, et al: 1995. The Na1 binding site of thrombin. J Biol Chem 270:22,089–22,092. Esmon CT, Schwartz HP: 1995. An update on clinical and basic aspects of the protein C anticoagulant pathway. Trends Cardiovasc Med 5:141–148. Esmon CT, Xu J, Gu JM, et al.: 1999. Endothelial protein C receptor. Thromb Haemostas 82:251–258. Gibbs CS, Coutre SE, Tsiang M, et al.: 1995. Conversion of thrombin into an anticoagulant by protein engineering. Nature 378:413– 416. Hall SW, Nagashima M, Zhao L, et al.: 1999. Thrombin interacts with thrombomodulin, protein C, and thrombin-activatable fibrinolysis inhibitor via specific and distinct domains. J Biol Chem 274:25,510–25,516.

Hanson S, Harker L, Kelly A, et al: 1997. Antithrombotic effects of selective activation of endogenous protein C. Thromb Haemost 78(Abstract Suppl):419. Kahn ML, Nakanishi-Matsui M, Shapiro MJ, et al.: 1999. Protease-activated receptors 1 and 4 mediate activation of human platelets by thrombin. J Clin Invest 103:879–887. Kokame K, Zheng X, Sadler JE: 1998. Activation of thrombin-activable fibrinolysis inhibitor requires epidermal growth factor-like domain 3 of thrombomodulin and is inhibited competitively by protein C. J Biol Chem 273:12,135–12,139. Minnema MC, Friederich PW, Levi M, et al.: 1998. Enhancement of rabbit jugular vein thrombolysis by neutralization of factor XI. In vivo evidence for a role of factor XI as an anti-fibrinolytic factor. J Clin Invest 101:10–14. Nagashima M, Lundh E, Leonard JC, et al.: 1993. Alanine-scanning mutagenesis of the epidermal growth factor-like domains of human thrombomodulin identifies critical residues for its cofactor activity. J Biol Chem 268:2888–2892. Parkinson JF, Nagashima M, Kuhn I, et al.: 1992. Structure-function studies of the epidermal growth factor domains of human thrombomodulin. Biochem Biophys Res Comm 185:567–576. Redlitz A, Tan AK, Eaton DL, Plow EF: 1995. Plasma carboxypeptidases as regulators of the plasminogen system. J Clin Invest 96: 2534–2538. Rydel T, Ravichandran KG, Tulinsky A, et al.: 1990. The structure of a complex of recombinant hirudin and human alpha-thrombin. Science 249:277–280. Sun WY, Witte DP, Degen JL, et al.: 1998. Prothrombin deficiency results in embryonic and neonatal lethality in mice. Proc Natl Acad Sci USA 95:7597–7602. Tsiang M, Jain AK, Dunn KE, et al.: 1995. Functional mapping of the surface residues of human thrombin. J Biol Chem 270:16,854–16,863. Tsiang M, Lentz SR, Sadler JE: 1992. Functional domains of membrane-bound human thrombomodulin. J Biol Chem 267:6164– 6170. Tsiang M, Paborsky LR, Li WX, et al.: 1996. Protein engineering thrombin for optimal specificity and potency of anticoagulant activity in vivo. Biochemistry 35:16,449–16,457. Xue J, Wu Q, Westfield LA, et al.: 1998. Incomplete embryonic lethality and fatal neonatal hemorrhage caused by prothrombin deficiency in mice. Proc Natl Acad Sci USA 95:7603–7607. Ye J, Liu LW, Esmon CT, et al.: 1992. The fifth and sixth growth factor-like domains of thrombomodulin bind to the anion-binding exosite of thrombin and alters its specificity. J Biol Chem 267:11,023–11,028. Zhao L, Nagashima M, Werner M, et al.: 1999. Inhibitor of plasma carboxypeptidase B as a novel fibrinolytic enhancer for thrombolytic therapy. Thromb Haemost 82(Abstract Suppl):602. PII S1050-1738(00)00047-5 TCM

TCM Vol. 10, No. 2, 2000