|Home | About | Journals | Submit | Contact Us | Français|
In serine proteases, Gly193 (chymotrypsin numbering) is conserved with rare exception. Mutants of blood coagulation proteases have been reported with Glu, Ala, Arg or Val substitutions for Gly193. To further understand the role of Gly193 in protease activity, we replaced it with Ala or Val in coagulation factor XIa (FXIa). For comparison to the reported FXIa Glu193 mutant, we prepared FXIa with Asp (short side chain) or Lys (opposite charge) substitutions. Binding of p-aminobenzamidine (pAB) and diisopropylfluorphosphate (DFP) were impaired 1.6–36-fold and 35–478-fold, respectively, indicating distortion of, or altered accessibility to, the S1 and oxyanion-binding sites. Val or Asp substitutions caused the most impairment. Salt bridge formation between the amino terminus of the mature protease moiety at Ile16 and Asp194, essential for catalysis, was impaired 1.4–4-fold. Mutations reduced catalytic efficiency of tripeptide substrate hydrolysis 6–280-fold, with Val or Asp causing the most impairment. Further studies were directed toward macromolecular interactions with the FXIa mutants. kcat for factor IX activation was reduced 8-fold for Ala and 400–1100-fold for other mutants, while binding of the inhibitors antithrombin and amyloid β-precursor protein Kunitz domain (APPI) was impaired 13–2300-fold and 22–27000-fold, respectively. The data indicate that β-branching of the side chain of residue 193 is deleterious for interactions with pAB, DFP and amidolytic substrates, situations where no S2′–P2′ interactions are involved. When an S2′–P2′ interaction is involved (factor IX, antithrombin, APPI), β-branching and increased side chain length are detrimental. Molecular models indicate that the mutants have impaired S2′ binding sites and that β-branching causes steric conflicts with the FXIa 140-loop, which could perturb the local tertiary structure of the protease domain. In conclusion, enzyme activity is impaired in FXIa when Gly193 is replaced by a non-Gly residue, and residues with side chains that branch at the β-carbon have the greatest effect on catalysis and binding of substrates.
Serine proteases play an essential role in many biologic processes including digestion of dietary proteins, blood coagulation, the complement cascades, fibrinolysis, bone resorption and remodeling, cell differentiation, and fertilization (1-7). The cleft shaped active site of the chymotrypsin/trypsin family of serine proteases contains a catalytic triad comprising residues His57, Asp102, and Ser195 (chymotrypsin numbering system)1 (8). These residues are located at the entrance to the substrate-binding pocket, where the N2 of His57 is poised to accept a proton from the Oγ of Ser195 during the nucleophilic attack by this oxygen on the carbonyl carbon of the substrate. The negatively charged carboxyl group of Asp102 serves to stabilize the positively charged form of His57 by making an H-bond with N1 of His57. This charge relay system makes Ser195-Oγ an unusually strong nucleophile and permits peptide bond hydrolysis (9).
Amino acids from the amino terminus to carboxy terminus of the substrate are designated Pn, …, P3, P2, P1, P1′, P2′, P3′, …, Pn’, with peptide bond cleavage occurring between residues P1 and P1′. The corresponding binding sites on the enzyme are designated Sn, …, S3, S2, S1, S1′, S2′, S3′, …, Sn’ (10). In trypsin-like proteases, Asp189 is at the bottom of the S1 site, and forms a salt bridge with the guanidinium or the ammonium group of the P1 Arg or Lys, respectively, of the substrate (9). Another defining feature of serine proteases, including those containing divergent folds from chymotrypsin, subtilisin, and serine carboxypeptidase families, is the presence of an oxyanion-binding site (9). In the chymotrypsin/trypsin family, the oxyanion binding site is formed by the backbone amide nitrogens of Gly193 and Ser195. The nucleophilic attack by the Oγ of Ser195 on the carbonyl carbon atom of the substrate changes the geometry around this carbon from trigonal planar to tetrahedral. The tetrahedral reaction intermediate is intrinsically unstable due to the negative charge on the peptide carbonyl oxygen atom. However, in serine proteases this charged oxyanion is stabilized by hydrogen bonds with the amide NH groups of Gly193 and Ser195 in a location referred to as the oxyanion hole. These interactions result in preferential binding of the substrate in the transition state, a necessary requirement for enzyme catalysis.
Gly193 is conserved in nearly all serine proteases (9-13) and is part of a type II β-turn. In the FXIa2 structure, the -ψ conformation parameters for residue 193 are similar to those in many serine protease crystal structures [RCSB Protein Data Bank], and are in a region of the Ramachandran plot that is compatible only with a Gly residue. The β-carbon of a non-Gly residue would have a steric conflict with the carbonyl O of residue 192, resulting in a type I β-turn that causes the amide nitrogen of residue 193 to point away from the oxyanion hole, disrupting the oxyanion binding site and the S1 site (14). In addition, the side chain of a non-Gly 193 residue is expected to occupy the S2′ site, interfering with S2′–P2′ interactions. Therefore, the side chain length, charge, and conformation at residue 193 are expected to play a significant role in determining the extent of dysfunction for a mutant protein.
Naturally occurring mutations at position 193 have been identified in plasma coagulation proteases, including Gly → Glu1 in factor XI (FXI)>2, Gly → Ala, Arg, Val or Glu in factor IX (FIX), and Gly → Glu or Arg in factor VII (FVII) (15-22). FXIa, the active form of the homodimeric protease zymogen FXI (23), is a component of a pathway involved in sustaining thrombin generation at a wound site (24). FXIa has considerably higher amidolytic activity toward tripeptide chromogenic substrates when compared to FVIIa and FIXa, and FXIa interactions with macromolecular substrates and inhibitors are reasonably well-defined. For these reasons, we chose FXIa as a model coagulation enzyme to study the effects of amino acid substitutions for Gly193. The data indicate that β-branching and side chain length play a significant role in determining the degree of enzyme impairment.
H-D-Ile-Pro-Arg-p-nitroanilide (S-2288) and pyroGlu-Pro-Arg-p-nitroanilide (S-2366), were from DiaPharma (West Chester, OH). Sodium boro[3H]-hydride was from Perkin-Elmer. Enhanced chemiluminescence (ECL) detection reagents were from Amersham Pharmacia Biotech. Diisopropylfluorophosphate (DFP) was from Calbiochem. Activated partial thromboplastin reagent was from Beckman Coulter, and normal plasma was from George King (Overland Park, KS). Unfractionated heparin was from Pharmacia Hepar, Inc. (Franklin, OH). Fatty acid free bovine serum albumin (BSA), p-aminobenzamidine (pAB), polyethylene glycol 8000 (PEG), Polybrene, and other chemicals of the highest grade available were from Sigma. Corn trypsin inhibitor (CTI), human FXIIa, FIX, and antithrombin were from Enzyme Research Laboratories (South Bend, IN). The Kunitz domain of protease Nexin-2/ amyloid β Protein Precursor Kunitz protein inhibitor (APPI) was a gift from Dr. Alvin Schmaier (Case-Western Reserve University, Cleveland).
The GGA triplet coding for Gly555 (corresponds to chymotrypsin Gly193) in the FXI cDNA (nucleotides 1760–1762) (25) was changed to GCA (Ala), GTA (Val), GAA (Glu), GAC (Asp), or AAA (Lys) using a QuikChange mutagenesis kit (Stratagene). The cDNAs were ligated into mammalian expression vector pJVCMV, which contains a cytomegalovirus promoter, as described (26). 293 fetal kidney fibroblasts (5 × 106, ATCC CRL 1573) were transfected by electroporation (Electrocell Manipulator 600 BTX, San Diego) with 40 μg of pJVCMV-FXI construct and 2 μg of pRSVneo that contains a cDNA conferring neomycin resistance. Cells were grown in DMEM with 5% fetal bovine serum and 500 μg/mL G418, and supernatants from G418 resistant clones were tested for protein expression by ELISA using goat anti-human FXI polyclonal antibodies (Affinity Biologicals, Hamilton, Ontario). Expressing clones were expanded in 175 cm2 flasks using Cellgro Complete media (Mediatech, Herndon, VA). Conditioned medium was collected every 48–72 h, supplemented with benzamidine (5 mM), and stored at −20 °C.
Proteins were purified by monoclonal antibody affinity chromatography from 1 L of conditioned medium using an anti-FXI IgG 1G5.12 monoclonal affinity column (26), and stored in Tris-HCl 50 mM pH 7.5, 150 mM NaCl (TBS). Protein concentration was determined by dye binding assay (BioRad, Hercules, CA), and purity by SDS–PAGE. Purified proteins were homogeneous on SDS–PAGE with the correct molecular mass for a FXI homodimer. To prepare FXIa, FXI (300 μg/mL) was incubated with 5 μg/mL FXIIa at 37 °C, and complete conversion to FXIa was confirmed by reducing SDS-PAGE. FXIIa was inhibited with 20-fold molar excess of CTI. Wild type FXIa is designated hereafter as FXIaWT, while mutant FXIa is designated FXIaG193A, FXIaG193V, FXIaG193E, FXIaG193D, or FXIaG193K.
Each reaction contained TBS with 5 mM Ca2+, 100 μg/mL BSA, 0.5 nM FXIaWT or 2–5 nM FXIaG193 mutant proteins, and different amounts of S-2288 or S-2366. The rate of pNA release was measured using a Beckman DU800 spectrophotometer with kinetics module at 405 nm for 15 min. The amount of pNA released was calculated using an extinction coefficient of 9.9 mM−1, cm−1 at 405 nM (27). The initial rate was converted to μM substrate hydrolyzed/min. The program GraFit was used to determine Km and Vmax, using the Enzyme Kinetics Program from Erithacus Software.
Each reaction mixture contained 250 nM FXIa in TBS with 100 μg/mL BSA and 5 mM CaCl2. Increasing amounts of DFP (2 μM to 0.8 mM) were added to each reaction and incubated at room temperature. At various time points, 5 μL aliquots were removed and added to 155 μL of TBS with 100 μg/mL BSA containing 625 μM S-2288. Changes in absorbance at 405 nm were measured, and residual FXIa activity was determined at each time point. Residual activity was plotted as a percent of initial activity, and the first-order rate constants (kobs) for each concentration of DFP were obtained using eq 1,
where At and A0 are the percent FXIa activity at time t and 0 s, respectively. Values for kobs were plotted against DFP concentration to obtain second-order rate constants.
The Kd(app) was determined by analyzing competitive inhibition of S-2288 hydrolysis. Reactions were carried out in TBS with 100 μg/mL BSA and 5 mM CaCl2 and either 100 μM S-2288 (for FXIaWT) or 500 μM S-2288 (for FXIaG193 mutants). Increasing amounts of pAB were added, and reactions were initiated by addition of either 1 nM FXIaWT or 2–5 nM FXIaG193 mutant. Initial rates of pNA release were measured on a SpectraMax 190 microplate reader (Molecular Devices, Sunnyvale, CA) at 405 nm, and converted to micromolar substrate released per minute. The IC50 (concentration of pAB required for 50% inhibition) was determined by fitting results with the IC50 four-parameter logistic equation of Halfman (eq 2) (28),
where y is the rate of pNA release in the presence of a given concentration of pAB designated by x, a is the maximum rate of pNA release in the absence of pAB, and s is the slope factor. Each point was weighted equally and data was fitted to eq 2 using the nonlinear regression analysis program GraFit from Erithacus Software. To obtain Kd(app) values for the interaction of pAB with FXIa, we used eq 3, as described by Cheng and Prusoff (29) and elaborated on by Craig (30),
where [S] is the S-2288 concentration. The Km values obtained for each protease for S-2288 hydrolysis were used to obtain Kd(app).
Carbamylation experiments were performed as described by Camire (31). Each reaction contained 1 μM FXIa in 20 mM Hepes, 0.15 M NaCl, 0.1% PEG, 2 mM CaCl2, pH 7.5 (HBSP). Each experiment was performed in the absence or presence of pAB (10 times the Kd pAB). Reactions were started by addition of 0.2 M NaNCO. The pH after the addition of NaNCO was 7.5. Every 30 min, 5 μL aliquots were removed into 145 μL of HBSP containing 500 μM S-2288. Residual FXIa activity was determined from rates of hydrolysis using a Beckman DU 800 spectrophotometer, and plotted as a percent of initial activity. The kobs for carbamylation were determined using eq 1.
Sialyl-3H-factor IX (specific activity 2.1 × 108 cpm/mg) was prepared as described (32). Labeled protein was a single band of appropriate molecular mass on reducing and nonreducing SDS–PAGE, and has ~85% of the activity of unlabeled FIX as measured in an activated partial thromboplastin time assay (32). The rate of FIX activation by FXIa was measured by determining the amount of 3H-labeled FIX activation peptide released over time. Each reaction was carried out at 37 °C in TBS containing 0.5 mg/mL BSA, 5 mM CaCl2, and FIX (2.0 μg/mL to 25 μg /mL). Reactions were initiated by addition of 0.25–25 nM FXIa. Aliquots (100 μL) were removed at various times into 100 μL of cold stop buffer (TBS, 50 mM benzamidine, 50 mM EDTA, and 5 mg/mL BSA). Equal volumes (200 μL) of 6% trichloroacetic acid (TCA) were added, followed by centrifugation to precipitate protein. Aliquots (100 μL) of TCA supernatant, which contain 3H-labeled free activation peptide, were added to Aquasol 2 (4 mL), and 3H was measured on a Beckman LS 5000CE β-counter. In the absence of FXIa, counts in the TCA supernatant were ≤1% of total counts (background), while complete FIX activation results in ~40% of the total counts in supernatant. Background counts (~1% of total) were subtracted from results for each sample to determine the amount of activity released by FXIa. The amount of FIXa formed was obtained from the average for three experiments, and rates of activation were determined by least-squares fitting of the data points to a straight line. Rates of activation were then plotted against FIX concentration, and Km and kcat were obtained using the Enzyme Kinetics program from Erithacus Software.
Reactions were carried out in 25 mM Tris-HCl, 100 mM NaCl pH 7.4, plus 0.1% Tween-20 (TBS-T) at 37 °C. FXIa (6 nM) was incubated with antithrombin (30 nM to 0.9 μM) in the presence of 1 unit/mL heparin. At various times, 20 μL samples were removed into 80 μL of TBS-T containing 500 μM S-2288 and 5 μg/mL Polybrene (to dissociate FXIa and heparin) in 96-well microtiter plates. Changes in absorbance at 405 nm were measured on a microplate reader, and residual FXIa activity was determined for each time point. Residual activity was plotted as a percent of initial activity, and the first-order rate constants (kobs) were obtained for each concentration of antithrombin using eq 1, where At and A0 are percent FXIa activity at time t and 0 s, respectively. The values of kobs were plotted against the antithrombin concentration to obtain second-order rate constants.
All reactions were carried out in TBS with 0.1 mg/mL BSA and 5 mM CaCl2. Wild type FXIa or FXIaG193A (1 nM) or other FXIa mutants (10 nM) were incubated with various concentrations of APPI (10−1 to 105 nM) for 1 h at room temperature in a 96-well microtitration plate (total volume 100 μL). S-2288 (5 μL) was then added to a final concentration of 1 mM, and residual amidolytic activity was measured in a Molecular Devices Vmax kinetic microplate reader. The apparent inhibition constant, , was determined using the nonlinear regression data analysis program GraFit. Data for wild type FXIa and the FXIaG193A were analyzed with an equation for a tight-binding inhibitor (eq 4) where vi and v0 are the inhibited and uninhibited rates, respectively, and [I]0 and [E]0 are the total concentrations of inhibitor and enzyme, respectively (33, 34).
Inhibition data for FXIaG193V, FXIaG193E, FXIaG193D, or FXIaG193K, where Ki [E]0, were analyzed using eq 5.
Ki values were obtained by correcting for the effect of substrate according to Beith et al. (33), using eq 6, where [S] is S-2288 concentration and Km values were those for S-2288 hydrolysis (see above).
The X-ray crystal structures of zymogen FXI (pdb code 2F83 (35)), FXIa with benzamidine (pdb code 1ZHM (36)), FXIa in complex with ecotin mutants containing amino acid replacements to mirror the activation peptide region of FIX (pdb codes 1XXF, 1XXD, and 1XX9 (37)), and FXIa in complex with APPI (pdb code 1ZJD (38)) were used for modeling. In each structure, Gly193 was changed to Ala, Asp, Glu, Lys, or Val and the local areas were energy minimized with slight adjustments. In modeling the 193 mutants, all residues could be fitted into the pocket occupied by Gly193, and all bad contacts were relieved by minor adjustments. The FXIa–ecotin structure containing the amino acid sequence surrounding the 180–181 FIX activation peptide region (pdb code 1XX9 (37)) was used to model the effect of Gly193 substitution on FIX binding, and the FXIa:APPI structure was used to model interactions and effects of the Gly193 substitutions on APPI binding (38). The trimolecular complex of antithrombin:heparin:thrombin (pdb code 1TB6 (39)) was used to model the FXIa: antithrombin complex and effects of Gly193 substitutions on antithrombin binding. Thrombin and FXIa were aligned and minor adjustments were made to correctly position FXIa with antithrombin.
Initially we examined FXIa Gly193 mutants for perturbation of the S1, S2, and S4 substrate binding subsites and the oxyanion binding site, using two tripeptide substrates differing at the S4 site. The chromogenic substrate S-2288 has an isoleucyl side chain at S4, while S-2366 contains a pyroglutamyl group. These data are presented in Table 1. The Km for both substrates was most affected when residue 193 had side chains that branch at the β-carbon (Val and Asp) and with a negative charge (Glu). It should be noted that the Km and kcat values for FXIaG193V and FXIaG193D for the substrate S-2366 are approximations because the value exceeded the highest substrate concentration used (10 mM). The kcat values for both substrates were most affected by Val and Asp substitutions. The overall catalytic specificity constant (kcat/Km) was also most affected (~200-fold) with these two amino acid substitutions.
The impairment in kcat noted for S-2288 and S-2366 hydrolysis could in part stem from the inability of the mutants to stabilize the tetrahedral intermediate containing the oxyanion formed during catalysis (8, 9). DFP specifically reacts with the active site serine (Ser195) in these proteases and contains an oxyanion, making it a useful probe to test the integrity of the oxyanion hole (8, 9, 40, 41). Results for DFP inhibition of FXIaWT and FXIa 193 mutants are shown in Figure 1 and summarized in Table 2. The oxyanion binding site is most impaired in FXIaG193V and FXIaG193D, and the DFP binding data are consistent with the kcat values for synthetic substrate hydrolysis by the FXIaG193V and FXIaG193D.
The Km values for synthetic substrate hydrolysis reflect perturbation in the S1, S2, and/or S4 subsites. To ascertain the contributions of the S1 site toward impairment of synthetic substrate binding, we investigated the binding of pAB, which specifically interacts with Asp189, located at the bottom of the S1 site. These data are presented in Figure 2 and summarized in Table 2. pAB binding was most impaired for FXIaG193V (36-fold) and FXIaG193D (14-fold), reflecting the increase in Km values for synthetic substrate hydrolysis by these mutants.
Data in the previous sections indicate that the oxyanion hole and S1 site are not properly formed in FXIa with a non-Gly residue at position 193. Development of the S1 site and oxyanion hole in serine proteases requires formation of a salt bridge between the amino group of Ile16 and the carboxylate group of Asp194. Covalent modification of the amino group of Ile16 by carbamylation results in an inactive enzyme, and a faster rate of carbamylation is indicative of a destabilized salt bridge. As summarized in Table 3, the rates of carbamylation for FXIaG193D, FXIaG193E and FXIaG193V were 2.5–4-fold faster than for FXIaWT. The rates of carbamylation of FXIaG193A and FXIaG193K were less impaired. Interestingly, these are the two amino acids that we have tested that do not have branched side chains. Occupancy of the S1 site by pAB fully corrects the impairment (Table 3), consistent with earlier observations that occupancy of the S1 site stabilizes the Ile16–Asp194 salt bridge in mutants of factor Xa and factor IXa (31, 42).
FIX activation to the active proteases FIXaβ involves cleavage after Arg145 and Arg180 to liberate an 11 kDa activation peptide (43). FIX activation by FXIa was examined using an assay that measures liberation of the FIX activation peptide (32). These data are presented in Figure 3 and summarized in Table 4. The values for Km were relatively similar for all proteases (167–268 nM), consistent with observations that exosite interactions distant from the protease active site are important in initial recognition and binding of FIX to FXIa (44-46). However, kcat was decreased ~8-fold for FXIa G193A and ~400–1100-fold for the other mutants. The values for kcat are influenced by the interaction of FXIa with both FIX activation cleavage sites and by catalysis of the FIX peptide bonds by FXIa. Therefore, values for kcat represent cumulative effects of restricted binding and catalysis at the two activation sites.
Antithrombin and APPI are physiologic inhibitors of FXIa. Antithrombin belongs to the serpin family of inhibitors, and its inhibition of FXIa is potentiated by heparin (47, 48). The first-order rate constants, kobs, for binding of antithrombin to FXIaWT and to FXIa mutants were plotted against antithrombin concentration (Figure 4). These data were used to obtain second-order rate constants (k2), which are summarized in Table 5. The k2 value for FXIaG193A inhibition was ~13-fold lower than for FXIaWT, and ~1400–2300-fold lower for the other mutants. APPI, the protease inhibitory domain of the β-amyloid precursor protein, is a member of the Kunitz family of inhibitors (38, 49). The overall binding constant (Ki) for inhibition by APPI was 22-fold greater for FXIaG193A and ~6500–27,000-fold greater for the other mutants (Figure 5, and summarized in Table 5).
Gly193 has two functional roles in an active serine protease: (1) it is part of the type II β-turn which allows the amide nitrogen to participate in formation of the oxyanion binding site, and (2) it allows the S2′ site to remain open for occupancy by the P2′ residue of substrates/inhibitors. In contrast, in protease zymogens, Gly193 is not part of a type II β-turn and the S2′ site is not formed. Therefore, replacing Gly193 with another residue is unlikely to have a significant effect on zymogen structure, as was observed in modeling studies (data not shown) with Ala, Asp, Glu, Lys, and Val at position 193 using the zymogen structure of FXI (34). This was also the case when zymogen FIX was modeled with Val at position 193 (20). Consistent with this premise, all recombinant FXI proteins used in the current study interacted similarly with the antibody used for affinity purification, which binds to the FXI protease domain, and all were activated similarly by FXIIa (data not shown). Substitution of Gly193 with a non-Gly residue in FXIa, on the other hand, is predicted to reorient the peptide bond between amino acids 192 and 193, causing the amide nitrogen of residue 193 to point away from the oxyanion binding site. Furthermore, it is anticipated that the size and nature of the residue 193 side chain will cause varying degrees of impairment by occupying the S2′ site. To test these concepts, we replaced Gly193 in FXIa with several naturally occurring substitutions observed in the blood coagulation serine proteases FVIIa and FIXa and compared them to the previously described FXIaG193E mutant (50).
The studies with small tripeptide substrates (S-2288 and S-2366) and inhibitors (DFP and pAB) demonstrate that interactions with the S1 sites and the oxyanion binding sites are most impaired in FXIaG193D and FXIaG193V, to a lesser extent in FXIaG193E and FXIaG193K, and least impaired in FXIaG193A (Table 2). In general, these data agree with the impairment in formation of the salt bridge between Ile16 and Asp194. The side chain of Val branches at the β-carbon, and Asp has a bulky carboxylate group at the β-carbon. These side chains are likely to perturb the local environment to a greater extent than the short side chain of Ala, or the unbranched side chain of Lys, which points into the solvent. In modeling studies the side chains with branching at the β-carbons impinge on the protease 140-loop (chymotrypsin numbering) and could cause distortion of the local region including the S1 site, the oxyanion binding site, the S2′ site, and possibly other interaction sites. It is interesting to note that, in the absence of S1 site occupancy, residues 189 through 194 are somewhat disordered in the structure of mouse glandular kallikrein-13 (11), one of the few serine proteases with a non-Gly residue 193 (Asp). These structural data are consistent with the deleterious effect of Asp or Val on both the S1 site and oxyanion binding sites in FXIa. The Km for tripeptide substrates was most affected in FXIaG193V and FXIaG193D, and least affected for FXIaG193A and FXIaG193K (Table 1), suggesting that S1 site occupancy contributes significantly toward impairment in the Km. In interpreting these data, it must be recognized that the FXIa residue 193 side chains may adopt unanticipated conformations and/or interactions due to specific properties of the side chains and the local environment that are not accounted for in our analysis. Indeed, Asp193 in mouse glandular kallikrein-13 is part of a type II turn, in spite of the non-Gly residue at position 193, due to a unique salt bridge with Lys150 (11). Such interactions not only could influence formation of the oxyanion binding site, S1 site, and Ile16–Asp194 salt bridge but also may affect accessibility of substrates and inhibitors to the active site.
Consistent with results for S-2288 and S-2366 cleavage, the kcat for FIX activation was also most affected for FXIaG193V and FXIaG193D and least affected for FXIaG193A (Figure 3 and Table 3). However, the effects of mutations on kcat for activation of FIX were significantly greater than for tripeptide substrates. With the exception of FXIaG193A, where kcat was impaired a modest 4.7-fold more for factor IX than for S-2288, kcat for factor IX cleavage by the other mutants was impaired ~30–300-fold more than for the tripeptide substrates. Furthermore, again with the exception of FXIaG193A, the kcat for FIX cleavage did not vary as widely between mutants as did kcat for tripeptide substrate hydrolysis. These results are consistent with modeling (Figure 6A), which indicates that the side chains of Asp, Glu, Lys, and Val all cause severe steric conflicts with the FIX P2′ residues Glu147 and Val182. In the case of Lys or Glu, the long side chain causes significantly greater steric interference with the P2′ residue than do the side chains of Asp and Val, accounting for the relatively greater effect of Lys and Glu on FIX activation, relative to cleavage of tripeptide substrates or interactions with DFP or pAB.
Inhibition of FXIaG193D, FXIaG193V, FXIaG193E and FXIaG193K by antithrombin was impaired ~2000-fold, whereas inhibition of FXIaG193A was impaired only ~10-fold. The degree of impairment is somewhat greater than that seen for factor IX activation by these respective mutants. The P2′ residue in antithrombin is Leu, which is spatially somewhat larger than either Glu or the Val, the P2′ residues at the two activation cleavage sites in FIX. Thus, the P2′ Leu of antithrombin is expected to involve a larger area of the S2′ site and have steric conflicts with all the FXIa mutants with the exception of FXIaG193A (Figure 6B). This provides a rational explanation for the observed differences in impairment in FIX activation and antithrombin inhibition for various mutants. APPI has Met at the P2′ residue, the side chain of which is longer than the Leu in antithrombin. Thus, one would expect that the degree of impairment in the interaction between the FXIa mutants and APPI would be greater than was observed either for FIX or antithrombin. Indeed, such is the case in both the experimental and modeling (Figure 6C) studies.
The impairment in Km and kcat values in the FXIa mutants is consistent with our previous work with FXIaG193E (50) and with subsequent studies from other investigators (51, 52). As Km reflects ground state binding of substrate and kcat reflects transition state binding, it appears that both native and transition states of the enzyme are affected by a non-Gly residue at position 193. This conclusion is supported by evidence for impairment of both the oxyanion binding site (Figure 1) and the S1 binding site (Figure 2). A substitution for Gly193 likely affects the environment in the vicinity of the 192–193 peptide bond, as suggested by the structure of mouse glandular kallikrein-13 with Asp at position 193 (11).
Based on previous observations with FVIIa, occupancy of the S1 site by pAB, which does not contain an oxyanion group, likely does not restore the integrity of the oxyanion hole in non-Gly mutants (53). Furthermore, the energy required to restore the oxyanion binding site also likely depends on the nature of the 193 residue, with Ala having the least effect and Val and Asp (present study), and Pro (52) having the greatest effects. This may reflect the energy required to break interactions between the larger side chains of Asp, Val, Glu, and Lys and surrounding residues to allow rotation of the 192–193 bond so that the amide nitrogen of residue 193 points toward the oxyanion hole. In the case of Pro193, the residue lacks an amide nitrogen preventing its participation in hydrogen bonding with the oxyanion of the substrate/inhibitor (52). The effects of residue 193 side chains on FXIa activity should apply to proteases such as FVIIa and FIXa, which have similar hydrophobic S2′ sites (36-38, 53, 54). Side chain branching at the β-carbon of residue 193 would be anticipated to be most detrimental in these proteases.
The side chain of residue 193 also contributes to specificity toward macromolecular substrates and inhibitors, depending on the P2′ residue of the substrate/inhibitor, as has been observed for human brain trypsin with Arg at position 193 (pdb code 1h4w (55)) and Trimeresurus stejnejeri plasminogen activator with Phe at position 193 (pdb code 1bqy (56)). The results for FXIa support this concept in that different amino acids at position 193 differentially affect interactions depending on whether Glu or Val (FIX), Leu (antithrombin), or Met (APPI) is the P2′ residue. In general, a serine protease with a non-Gly residue at position 193 would likely be restricted to cleaving substrates possessing Gly or short side chains residues at the P2′ position.
The authors thank Dr. Alvin Schmaier for providing the APPI protein used in these studies.
†This work was supported by Awards HL36365 and HL70369 to S.P.B. and HL58837 to D.G. from the National Heart, Lung, and Blood Institute.
1The chymotrypsin amino acid numbering system is used to assign numbers to amino acids in the FXIa catalytic domain, and is used throughout the manuscript. Gly193 in chymotrypsin corresponds to Gly555 in FXIa, Gly363 in factor IXa and Gly342 in factor VIIa. Chymotrypsin Ile16 and Asp194 correspond to FXIa Ile370 and Asp556, respectively.