A Binding Site for Thrombin in the Apple 1 Domain of Factor XI*

Previously we defined a binding site for high molecu- lar weight kininogen (HK) in the A1 domain of factor XI (FXI). Since thrombin can activate FXI and HK inhibits the activation of FXI by thrombin, we have identified a thrombin binding site in FXI. Both the recombinant A1 domain (Glu 1 –Ser 90 ) and a synthetic peptide (Phe 56 – Ser 86 ) containing the HK binding site inhibited FXI activation by thrombin. Both a monoclonal antibody, 5F7, recognizing the A1 domain, and the rA1 domain were shown to be competitive inhibitors of thrombin-cata- lyzed FXI activation. The peptides Ala 45 –Arg 54 and Val 59 –Arg 70 acted synergistically to inhibit FXI activation by thrombin. Mutant rA1 domain constructs (Val 64 3 Ala and Ile 77 3 Ala), which do not inhibit FXI binding to HK, retain full capacity to inhibit FXI activation by thrombin. The peptide Ala 45 –Arg 54 inhibited thrombin-catalyzed FXI activation, whereas it had no effect on FXI binding to HK. In contrast, the peptide Asn 72 –Leu 83 (which inhibited FXI binding to HK) did not inhibit FXI activation by thrombin. Thus, a thrombin binding site exists in the A1 domain of FXI spanning residues Ala 45 – Arg 70 that is contiguous with but separate and distinct from the HK binding site. These sites may regulate which ligand is bound to FXI and through which pathway FXI is activated. 740 Data Module, model 46K Universal Injector, and Lambda-Max model 481 Detector; Milford, MA). Reverse phase chromatography was performed using a Waters C8 (cid:109) Bondapak Column, whereas gel filtra- tion was carried out using a Waters Protein-Pak 60 column as described previously (20, 22, 29). Characterization of Synthetic Peptides— All the peptides utilized in this study were examined by HPLC (both reverse phase and gel filtra- tion), and all demonstrated a single homogeneous peak (data not shown). When the refolded peptides were examined by HPLC (both reverse phase and gel filtration), single homogeneous peaks with identical retention times to the original mixtures were observed, demon- strating the presence of a single homogeneous mixture of refolded peptides. The results were the same after reduction and alkylation of these same peptides. All reduced and alkylated or refolded peptides were examined for free SH groups using the Ellman reagent (5,5 (cid:57) - dithiobis[2-nitrobenzoic acid]). It was determined (30) that there was less than 0.02 mol of free SH/mol of peptide, which further verifies that these refolded peptides were homogeneous preparations consisting of intramolecular disulfide-bonded peptide.

Previously we defined a binding site for high molecular weight kininogen (HK) in the A1 domain of factor XI (FXI). Since thrombin can activate FXI and HK inhibits the activation of FXI by thrombin, we have identified a thrombin binding site in FXI. Both the recombinant A1 domain (Glu 1 -Ser 90 ) and a synthetic peptide (Phe 56 -Ser 86 ) containing the HK binding site inhibited FXI activation by thrombin. Both a monoclonal antibody, 5F7, recognizing the A1 domain, and the rA1 domain were shown to be competitive inhibitors of thrombin-catalyzed FXI activation. The peptides Ala 45 -Arg 54 and Val 59 -Arg 70 acted synergistically to inhibit FXI activation by thrombin. Mutant rA1 domain constructs (Val 64 3 Ala and Ile 77 3 Ala), which do not inhibit FXI binding to HK, retain full capacity to inhibit FXI activation by thrombin. The peptide Ala 45 -Arg 54 inhibited thrombincatalyzed FXI activation, whereas it had no effect on FXI binding to HK. In contrast, the peptide Asn 72 -Leu 83 (which inhibited FXI binding to HK) did not inhibit FXI activation by thrombin. Thus, a thrombin binding site exists in the A1 domain of FXI spanning residues Ala 45 -Arg 70 that is contiguous with but separate and distinct from the HK binding site. These sites may regulate which ligand is bound to FXI and through which pathway FXI is activated.
Factor XI (FXI) 1 is a homodimeric plasma glycoprotein that circulates as a complex with its cofactor high molecular weight kininogen (HK) (1,2) and is proteolytically activated on negatively charged surfaces by FXIIa to give rise to FXIa (3)(4)(5)(6)(7)(8)(9)(10). The mechanism, involving interactions of FXII, prekallikrein (PK), and HK, by which contact activation is initiated and its significance in vivo have yet to be established, since individuals congenitally deficient in any one of these contact factors (FXII, HK, and PK) do not experience abnormal bleeding, suggesting that these proteins are not required for coagulation in vivo (11,12). In contrast, a deficiency of FXI can result in excessive bleeding after trauma or surgery (13,14). These observations suggest that FXI may be activated in vivo by a protease(s) other than FXIIa.
The ability of thrombin, an enzyme generated late in the coagulation cascade, to activate FXI has been demonstrated (15,16). The site at which FXI is cleaved by thrombin is identical to that cleaved by FXIIa (16,17). Determination of the kinetic parameters of FXI activation by thrombin and FXIIa indicate that at a physiological concentration of FXI, in the presence of dextran sulfate, thrombin would be the more potent activator (16). Although FXI is readily activated by thrombin in a purified system with dextran sulfate present, the reaction may not proceed as readily in plasma (15,16,18), since although HK promotes the FXIIa-mediated reaction it inhibits thrombin-catalyzed activation of FXI (15,16,18). These observations raise the following two related questions. Is thrombin a physiological activator of FXI in plasma? What is the mechanism by which HK can inhibit thrombin-catalyzed FXI activation?
The present study was undertaken to determine the sequence of amino acids in FXI that mediate its interaction with thrombin. Clarification of the mechanism of interaction of these two proteins might also help to elucidate the physiological importance of thrombin-catalyzed FXI activation. Four tandem repeat sequences (designated A1, A2, A3, and A4 or Apple domains) are present in the heavy chain of FXI (7). We have previously reported evidence for the presence of an HK binding site in the A1 domain (19,20), a binding site for FXIIa in the A4 domain (21), a substrate binding site for FIX in the A2 domain (22), and recently, a specific binding site for platelets in the A3 domain (23). Evidence for a binding site in the A1 domain of FXI that is important for interaction with thrombin is reported in the present study.
Peptide Synthesis-Peptides were synthesized on an Applied Biosystems 430A peptide synthesizer by a modification (29) of the procedure described by Kent and Clark-Lewis (28). The sequences of the synthetic peptides utilized in this study are given in Table I. All the peptides utilized in this work were rationally designed, conformationally constrained synthetic peptides based upon a previously published (20) molecular model for the A1 domain of FXI. Each peptide was separately modeled using energy minimization calculations (20) that confirmed that the modeled peptides assumed a conformation similar to that of the A1 domain.
Refolding and Reduction and Alkylation of Peptides-A previously published method (24) was used to refold peptides containing cysteine residues. Alternatively, peptides were reduced with dithiothreitol and alkylated with iodoacetamide as described previously (24).
High Performance Liquid Chromatography (HPLC)-The HPLC system employed was from Waters (Waters 600 Gradient Module, model 740 Data Module, model 46K Universal Injector, and Lambda-Max model 481 Detector; Milford, MA). Reverse phase chromatography was performed using a Waters C8 Bondapak Column, whereas gel filtration was carried out using a Waters Protein-Pak 60 column as described previously (20,22,29).
Characterization of Synthetic Peptides-All the peptides utilized in this study were examined by HPLC (both reverse phase and gel filtration), and all demonstrated a single homogeneous peak (data not shown). When the refolded peptides were examined by HPLC (both reverse phase and gel filtration), single homogeneous peaks with identical retention times to the original mixtures were observed, demonstrating the presence of a single homogeneous mixture of refolded peptides. The results were the same after reduction and alkylation of these same peptides. All reduced and alkylated or refolded peptides were examined for free SH groups using the Ellman reagent (5,5Јdithiobis[2-nitrobenzoic acid]). It was determined (30) that there was less than 0.02 mol of free SH/mol of peptide, which further verifies that these refolded peptides were homogeneous preparations consisting of intramolecular disulfide-bonded peptide.
Assay of Factor XI Activation-Activation of FXI (60 nM) by thrombin (1.25 nM) was measured by chromogenic assay. Incubations were carried out in 200 l of 50 mM Tris, 150 mM NaCl, pH 7.3, with 1% bovine serum albumin (TBSA) and dextran sulfate (average molecular weight ϭ 500,000) (Sigma) (1 g/ml) at 37°C. After diluting to a final volume of 1 ml with TBSA containing 600 M S-2366 (Glu-Pro-Arg-pNA, Chromogenix, Mölndal, Sweden), the amount of free paranitroaniline was determined by measuring the change in absorbance at 405 nm (A405). The amount of FXIa generated was assayed by reference to a standard curve constructed using purified FXIa.
Effect of Peptides on the Rate of Activation of Factor XI by Thrombin-The assay procedure was the same as described above except thrombin (1.25 nM) was incubated for 5 min at 37°C with either peptide or buffer solution before the addition of FXI (60 nM).
To determine whether the A1-derived peptides inhibit the enzymatic activity of thrombin in the conversion of fibrinogen to fibrin, an established procedure (31) was employed in which 0.1 ml of human fibrinogen (Sigma), 1.5 mg/ml was diluted in 0.8 ml of 0.05 M Tris-HCl, 0.1 M NaCl, pH 7.4, to which was added 0.1 ml of ␣-thrombin (5 units/ml) that had previously been incubated with A1 domain peptides Glu 1 -Ser 90 , Phe 56 -Ser 86 , or Ala 45 -Arg 54 . The increase in turbidity was recorded for 10 min.
Kinetics of Activation of Factor XI by Thrombin-The assay was the same as described above except that the initial rates of activation were determined over a wide range of substrate concentrations. For determination of the effects of peptides or antibodies on the kinetics of the reaction, the FXI was incubated with various concentrations of peptide or antibody for 10 -15 min at 37°C before the addition of thrombin. Values for the Michaelis constant (K m ) and the maximum velocity (V max ) were obtained by the Lineweaver-Burk method (32) and were calculated using least-square fit as described previously (33,34).
Analysis of Kinetic Data for the Quantitation of the Inhibitor Constants-The I 50 method of Cha (35) was used to determine the inhibitor constants as described previously (14). In the case of classical competitive inhibition, I 50 (total inhibitor concentration at which the enzyme reaction velocity is 50% of the uninhibited reaction) is related to the substrate concentration as follows: where E t equals the total enzyme concentration and S equals the substrate concentration. Thus, from the plot of I 50 versus S, K i can be determined.

Effect of Monoclonal Antibodies on the Activation of Factor XI
by Thrombin-Previously, four monoclonal antibodies have been developed and characterized that have been useful in elucidating structure-function relationships of FXI (24,33,36). We initially determined the effects of binding of these antibodies to FXI on thrombin-catalyzed FXI activation. Fig. 1A shows that monoclonal antibody 5F7 can completely block thrombincatalyzed FXI activation with 50% inhibition at ϳ5 ϫ 10 Ϫ10 M 5F7, a concentration close to the K d (6.2 ϫ 10 Ϫ10 M) for 5F7 binding to FXI (36). Antibody 5F7 binds to the A1 domain of FXI (19) and blocks surface-mediated activation of FXI and its binding to HK (36). The specificities of the other four antibodies are indicated in the legend to Fig. 1. The concentration of 3C1 monoclonal antibody required to inhibit thrombin-catalyzed FXI binding was 100-fold greater (5 ϫ 10 Ϫ8 M) than that of the 5F7 antibody (Fig. 1A), whereas antibodies 1F1 and 5F4 had no effect. This experiment suggests that a binding site for thrombin exists in the A1 domain of FXI.
Effects of Heavy Chain Peptides on the Activation of Factor XI by Thrombin-It is well documented that in vitro FXI (60 nM) can be proteolytically activated by thrombin (1.25 nM) to generate FXIa (15,16). Since the 5F7 monoclonal antibody inhibited thrombin-catalyzed FXI activation (Fig. 1A), we determined whether peptides derived from amino acid sequences of the heavy chain of FXI could affect the activation of FXI by thrombin. The rA1 domain (Glu 1 -Ser 90 ) inhibited the activation of FXI by thrombin with an IC 50 of 5 ϫ 10 Ϫ6 M, whereas a conformationally constrained synthetic peptide (Phe 56 -Ser 86 ) containing the HK binding site (19,20) also inhibited thrombin-catalyzed FXI activation (IC 50 ϭ 10 Ϫ4 ) (Fig. 1B). In contrast, neither the rA2 domain (Ser 90 -Ala 181 , containing a sub-

FXI A1
Ala 45 -Ser 86 The numbers indicate the length of each peptide as reported by Fujikawa et al (7). A gap was inserted in the A1 domain at residue 72 to maintain maximal alignment of residues.
b Designates a residue in which cysteine replaced for an amino acid in the normal factor XI sequence. c Designates a peptide in which cysteine replaced one or more amino acid(s) in the normal factor XI sequence. d Designates a peptide in which alanine replaced one or more amino acid(s) in the normal factor XI sequence. Previously we have identified specific amino acid residues within the A1 domain involved in binding HK (19,20,37). Utilizing mutational analysis we have determined that the binding of FXI to HK is mediated at least in part by Val 64 and Ile 77 in the A1 domain of FXI (37). Therefore, we examined the effects of mutations of these two residues on the capacity of the rA1 domain to inhibit thrombin-catalyzed FXI activation. We found that mutant rA1 domain constructs (Val 64 3 Ala and Ile 77 3 Ala), which have lost the capacity to inhibit FXI binding to HK (37), retain the full capacity of the rA1 domain (Glu 1 -Ser 90 ) to inhibit thrombin-catalyzed FXI activation (Fig. 2). Therefore, the binding sites for HK and thrombin in the A1 domain, although contiguous, are apparently separate and distinct. Another experiment that supports this conclusion is that after reduction and alkylation, the rA1 domain (Glu 1 -Ser 90 ) retains the capacity to inhibit FXI binding to HK (IC 50 ϳ10 Ϫ6 M) (19,20,37), whereas it is unable to inhibit thrombincatalyzed FXI activation (data not shown).
Effects of Conformationally Constrained, A1-derived Peptides on the Activation of Factor XI by Thrombin-Detailed structural information is not available for the A1 domain or any other part of the FXI molecule. However, we have constructed a molecular model of this region (20,37), which may or may not ultimately be found to accurately portray the true structure of the FXI A1 domain once it is determined from x-ray crystallography or nuclear magnetic resonance studies. Thus, the model was used as a guide to generate hypotheses about the structure of the A1 domain to be tested in functional studies. Using this hypothetical model, we have made testable predictions about the structure and function of the HK binding site in the A1 domain (20,37). Our experiments revealed that the sequence of amino acids, Val 59 -Lys 83 within the A1 domain of FXI, contains two antiparallel ␤-strands connected by ␤-turns that comprise a continuous surface utilized for the binding of HK (20). Since the rA1 domain peptide (Glu 1 -Ser 90 ) was more effective in inhibiting FXI activation by thrombin than the peptide (Phe 56 -Ser 86 ) by 2 orders of magnitude, we suspected that the amino acid sequence Glu 1 -Trp 55 might contain residues important in binding thrombin. Therefore, we examined a molecular model of the A1 domain that predicts the presence of three stem-loop structures (antiparallel ␤-strands connected by ␤-turns) defined by amino acid residues Ala 45 -Arg 54 , Val 59 -Arg 70 , and Asn 72 -Lys 83 (19,20). We therefore prepared conformationally constrained cyclic peptides compris- FXI was incubated with either buffer or various concentrations of the antibody solution or peptides prior to use in the assay. The rate of FXIa formation was determined as described under "Materials and Methods." A, data shown are those obtained with monoclonal antibodies, 5F7 (q--q), 3C1 (E--E), 1F1 (Ç--Ç), and 5F4 (É--É). Antibody 5F7 is immunologically reactive to the A1 domain of FXI (19). Antibody 3C1 is heavy chain-specific, is immunologically cross-reactive with the A2 domain (36), and is a competitive inhibitor of FIX activation by FXIa (24,33). Antibody 5F4 is light chain-specific (24,33). Antibody 1F1 is also directed against the heavy chain of FXI (24). ing these peptide loop structures to determine whether they might assume a conformation that comprises a thrombin binding site. These peptides were identical to those tested to delineate the HK binding surface (19,20,37). Cysteine residues were introduced at the amino and carboxyl terminus of each peptide so that the resulting disulfide bond might stabilize the loop-like structure (19,20). The peptide designated Ala 45 -Arg 54 (C) in which cysteines were substituted at positions 49 and 53 (see Table I) had no effect (refolded or reduced and alkylated) on FXI binding to HK at concentrations up to 10 mM (20). By comparison, this peptide (Ala 45 -Arg 54 ) when properly folded inhibited thrombin-catalyzed activation of FXI with an IC 50 ϳ2 ϫ 10 Ϫ5 M. This peptide was 1 order of magnitude more effective than peptide F56-S86 (IC 50 ϭ 1 ϫ 10 Ϫ4 M), which indicates that most of the binding energy for thrombin resides within this amino acid sequence (Ala 45 -Arg 54 ) of FXI (Fig. 3A). The ability of Val 59 -Arg 70 (refolded) to inhibit thrombin-catalyzed FXI activation (Fig. 3B) was identical to that of Phe 56 -Ser 86 (Fig. 3A) with an IC 50 of 10 Ϫ4 M. However, Asn 72 -Lys 83 (both refolded and reduced and alkylated) had no effect on thrombin-catalyzed activation of FXI (data not shown), whereas this peptide was a potent inhibitor of HK binding to FXI (IC 50 ϳ10 Ϫ4 M; Refs. 20 and 37). These results strongly suggest that Ala 45 -Arg 54 and Val 59 -Arg 70 form two stem-loop structures consisting of antiparallel ␤-strands connected by ␤-turns, which together form the thrombin binding site. The experiments in Fig. 3 demonstrate synergism in the effects of both peptides together compared with either one alone, since when the two peptides were added together their effect was greater than a simple additive effect, i.e. when used in combination at equimolar concentrations these two peptides were 1 order of magnitude more effective than either one alone in inhibiting thrombin-catalyzed FXI activation.
Prekallikrein, a protein with 58% sequence identity to FXI, also binds HK in the A1 domain within the homologous amino acid sequence Phe 56 -Gly 86 (38). This stretch of amino acids displays 65% homology with a comparable sequence in FXI. Therefore, we tested the PK Phe 56 -Gly 86 peptide for its ability to inhibit thrombin-catalyzed FXI activation. Unlike the FXI Phe 56 -Ser 86 peptide, the PK Phe 56 -Gly 86 peptide did not inhibit thrombin-catalyzed FXI activation (Fig. 3A). It has been reported that this sequence of amino acids in PK (Phe 56 -Gly 86 ) binds HK (38) as does the homologous sequence of FXI (19,20). Thus, the amino acid sequences involved in FXI and PK interaction with HK are not involved in binding thrombin.
Fine Mapping of the Thrombin Binding Site in Factor XI-In order to gain information about which specific amino acids comprise the thrombin binding site we prepared synthetic peptides with amino acid substitutions, determined by examining our molecular model for residues that project their side chains into a predicted contact surface (37). One such candidate in FXI was a glutamic acid residue at position 66 (37), which was therefore changed to an alanine in the conformationally constrained synthetic peptide Val 59 -Arg 70 . This altered peptide failed to inhibit thrombin-catalyzed FXI activation, whereas the native peptide with a glutamic acid present at position 66 (Val 59 -Arg 70 ) inhibited thrombin-catalyzed FXI activation with an IC 50 of ϳ10 Ϫ4 M (Fig. 3B). However, Glu 66 is apparently not involved in binding HK since Val 59 -Arg 70 (Glu 66 3 Ala) had inhibitory activity equal to that of the native peptide in FXI binding to HK (37). When the linear peptide Ala 45 -Arg 54 (Fig. 3C) was compared with the conformationally constrained cyclic peptide Ala 45 -Arg 54 (C) (Fig. 3B) they both inhibited thrombin-catalyzed FXI activation with the same IC 50 (2 ϫ 10 Ϫ5 M; see Table II), indicating 1) that Ser 49 and Thr 53 (which were substituted with cysteines in the cyclic peptide) are not part of the contact surface and 2) that cyclization and conformational constraint are not required for inhibitory activity of this particular peptide. However, when alanine was substituted for Asp 51 in peptide Ala 45 -Arg 54 (C) the altered peptide (IC 50 ϭ 1. the altered peptide compared with 2 ϫ 10 Ϫ5 M for the native peptide ( Fig. 3C and Table II). Thus, these results indicate that the side chain of Asp 51 might be directly involved in binding thrombin, whereas Glu 50 might also make a minor contribution to thrombin binding. When the amino acids comprising part of the putative thrombin binding surface of Ala 45 -Arg 54 , consisting of amino acids Glu 50 -Asp 51 -Pro 52 , were tested as a conformationally constrained cyclic peptide, Ser 49 -Thr 53 (C), for its ability to inhibit thrombin-catalyzed FXI activation, it was 2 orders of magnitude less effective (IC 50 ϭ 1 ϫ 10 Ϫ3 M; Fig. 3C and Table II) than the Ala 45 -Arg 54 peptide (IC 50 ϭ2 ϫ 10 Ϫ5 M; Fig. 3A and Table II). This result suggests that the full effect of the peptide Ala 45 -Arg 54 on thrombin-catalyzed FXI activation requires a contribution from the amino-terminal portion in addition to Ser 49 -Thr 53 .
Kinetics of Thrombin-catalyzed Factor-XI Activation in the Presence of the Monoclonal 5F7 or the rA1 Domain Glu 1 -Ser 90 -The activation of FXI by thrombin in the presence of various concentrations of either monoclonal antibody 5F7 or the rA1 domain (Glu 1 -Ser 90 ) are shown in Fig. 4. The doublereciprocal (Lineweaver-Burk) plots yielded patterns consistent with classical competitive inhibition, i.e. V max is unaffected by the antibody or peptide, whereas progressively higher concentrations of the antibody or peptide yielded progressively higher values of apparent K m . These results suggest that the binding of the rA1 domain Glu 1 -Ser 90 to a putative substrate (FXI) binding site within the thrombin molecule reduces the effective concentration of this binding site for its substrate (FXI), whereas the catalytic site of thrombin remains unaltered. The 5F7 monoclonal antibody apparently binds to an epitope within the A1 domain close to the site that interacts with the putative thrombin substrate (FXI) binding site.
Calculation of Binding Constants from Kinetic Parameters-The inhibitor constant, K i (the dissociation constant of the enzyme-inhibitor complex), was determined from the doublereciprocal plots as described under "Materials and Methods." In the case of competitive inhibition, I 50 (the total inhibitor concentration at which the enzyme reaction velocity is 50% of the uninhibited reaction) is a linear function of the substrate concentration. From the slope of this line K i can be determined. The insets of Fig. 4, A and B, represent such a plot, with the monoclonal antibody 5F7 (Fig. 4A) or the rA1 domain (Glu 1 -Ser 90 ) peptide (Fig. 4B) used as inhibitors. The calculated val-ues of K i from the slopes are 4 ϫ 10 Ϫ10 M (5F7) and 1 ϫ 10 Ϫ6 M (rA1, Glu 1 -Ser 90 ).

DISCUSSION
It is possible to activate FXI in the absence of contact proteins in the presence of the serine protease thrombin (15,16,18). Consequently, several laboratories have attempted to determine the physiological conditions required for the activation of FXI by thrombin or other proteases (16,18,39). Although FXI is activated by thrombin in a purified system, it is suspected that this reaction may not proceed in plasma (18,39). It is well known that HK enhances FXIIa-mediated activation of FXI in vitro (1)(2)(3)(4)(5)(6)(7)(8)(9)(10). However, HK (500 nM) inhibits thrombinmediated activation of FXI (15,16,18,39). Thus, thrombin may not be a suitable activator of FXI in plasma. To understand the physiological importance of thrombin-mediated FXI activation,  we have examined the interaction of FXI with thrombin and identified the amino acid sequence in FXI that interacts with thrombin.
Our experiments support the conclusion that a sequence of amino acids (Ala 45 -Lys 70 ) in the A1 domain of FXI that contains two antiparallel ␤-strands connected by ␤-turns comprises a surface that interacts with a substrate (FXI) binding site within thrombin (Fig. 5). The evidence supporting this conclusion is as follows: 1) a monoclonal antibody (5F7) that binds to the A1 domain of FXI (19,20,36) can completely block thrombin-catalyzed FXI activation with a K i ϳ5 ϫ 10 Ϫ10 M (close to the K d for 5F7 binding to FXI; see Ref. 36 and Figs. 1  and 4); 2) the rA1 domain peptide (Glu 1 -Ser 90 ) inhibited the activation of FXI by thrombin with a K i of 5 ϫ 10 Ϫ6 M (Figs. 1  and 4); 3) a molecular model of the A1 domain (Fig. 5) predicts the presence of three peptide loop structures, Ala 45 -Arg 54 , Val 59 -Arg 70 , and Asn 72 -Lys 83 , that form a solvent-exposed surface (37); 4) based on this model, conformationally constrained peptides were synthesized, two of which (Ala 45 -Arg 54 , Val 59 -Arg 70 ) act synergistically to inhibit thrombin-catalyzed FXI activation (Fig. 3); and, 5) Lineweaver-Burk plots of the activation of FXI by thrombin in the presence of either the monoclonal antibody 5F7 or the rA1 domain peptide yielded patterns consistent with a classical competitive inhibition (Fig. 4).
We have previously characterized a binding site for HK in the A1 domain of FXI (19,20,36,37). To perform fine mapping of this site we prepared conformationally constrained synthetic peptides and rA1 domain constructs (20,37). To identify specific amino acid residues involved in HK binding, conformationally constrained peptides were synthesized containing conserv-ative amino acid substitutions at residues suspected to contain side chains involved in binding including Val 64 3 Ala, Glu 66 3 A, Arg 73 3 Ala, and Ile 77 3 Ala (37). Because abnormal results were obtained with two of these peptides, Val 64 3 Ala and Ile 77 Ala, which failed to compete normally with FXI for binding HK, we prepared two mutant rA1 domains (Val 64 3 Ala and Ile 77 Ala), both of which exhibited diminished capacity to inhibit FXI binding to HK (37). Since the thrombin binding site was localized to the A1 domain and found to contain amino acid sequences overlapping the HK binding site (Fig. 1), we attempted to identify specific amino acid residues in the A1 domain that might bind thrombin. Our results are consistent with the following conclusions: 1) Val 64 and Ile 77 , which are important as contact sites for HK (37), do not participate in the interaction of the A1 domain with thrombin (Fig. 2); 2) Glu 66 and Asp 51 , which are not important as contact sites for binding HK (37), are both apparently important residues for binding thrombin (Fig. 3); and 3) another important difference between the HK and thrombin binding sites in the A1 domain is that reduction and alkylation of the A1 domain virtually destroys the thrombin binding site while leaving the HK binding site intact (37).
We also examined the plasma protein PK, which shares a high degree (58%) of sequence identity with FXI (7), to determine whether homologous amino acid sequences can also inhibit thrombin-catalyzed FXI activation. Unlike the FXI Phe 56 -Ser 86 peptide, the PK Phe 56 -Gly 86 peptide did not inhibit thrombin-catalyzed FXI activation. A glutamic acid is replaced by a glycine at position 66 in PK, and this amino acid substitution Glu 66 3 Ala 66 in FXI Val 59 -Arg 70 rendered this peptide inactive (Fig. 3). It is possible that this and other amino acid replacements in PK render PK Phe 56 -Gly 86 unable to inhibit thrombin-catalyzed FXI activation. It has also been established that PK, like FXI, also binds HK in the A1 domain within the homologous amino acid residues Phe 56 -Gly 86 (38). This stretch of amino acids is 65% identical to the comparable sequence in FXI (Fig. 6). Moreover, a peptide with the amino acid sequence Pro 45 -Lys 54 of PK had no effect on thrombincatalyzed FXI activation, unlike Ala 45 -Arg 54 of FXI, which was the most effective inhibitor (Fig. 3 and Table II). These two regions of FXI and PK have only 18% identity (Fig. 6), and it seems reasonable that PK does not contain a binding site for thrombin since thrombin does not activate PK (40). We have also attempted to fine map the important contact sites of peptide Ala 45 -Arg 54 (C) that interacts with thrombin. Apparently, Asp 51 is important in this interaction. However, amino acid residues in the first half of the peptide Ala 45 -Pro 48 also appear to be important in its interaction with thrombin (Fig. 3). Using the primary structure of the A1 domain and its known disulfide linkages, a model was calculated demonstrating antiparallel ␤-strands connected by ␤-turns (20). The colors represent the peptide loop structures (␣carbon backbone) and amino acids postulated to comprise binding sites for thrombin (yellow) (residues 45-58), HK (blue) (residues 72-85) and a loop structure utilized for binding both (or either) protein (red) (residues 59 -71). The data presented in this paper support the conclusion that the thrombin and HK binding sites in the A1 domain, while contiguous, are separate and distinct. However, these two binding sites appear to overlap since they share a common sequence of amino acids (Val 59 -Arg 70 ). The relationship between the putative thrombin and HK binding sites is depicted in the molecular model shown in Fig. 5. The model and our results predict that if HK is bound to the A1 domain, thrombin-mediated activation of FXI would be blocked and FXIIa-mediated activation of FXI would be favored. The reverse may also occur, i.e. the binding of thrombin to the A1 domain should prevent HK binding and FXIIa-mediated activation of FXI. Therefore, these two contiguous partially overlapping sites could constitute a point of regulation to determine by which pathway (contact activation versus feedback activation) and by which protease (FXIIa or thrombin) FXI might be activated.