AFJ and PHL contributed equally to this work and share first authorship.
Cystathionine γ-synthase (CGS) catalyzes the condensation of O-succinyl-L-homoserine (L-OSHS) and L-cysteine (L-Cys), to produce L-cystathionine (L-Cth) and succinate, in the first step of the bacterial transsulfuration pathway. In the absence of L-Cys, the enzyme catalyzes the futile α,γ-elimination of L-OSHS, yielding succinate, α-ketobutyrate, and ammonia. A series of 16 site-directed variants of Escherichia coli CGS (eCGS) was constructed to probe the roles of active-site residues D45, Y46, R48, R49, Y101, R106, N227, E325, S326, and R361. The effects of these substitutions on the catalytic efficiency of the α,γ-elimination reaction range from a reduction of only ∼2-fold for R49K and the E325A,Q variants to 310- and 760-fold for R361K and R48K, respectively. A similar trend is observed for the kcat/K of the physiological, α,γ-replacement reaction. The results of this study suggest that the arginine residues at positions 48, 106 and 361 of eCGS, conserved in bacterial CGS sequences, tether the distal and α-carboxylate moieties, respectively, of the L-OSHS substrate. In contrast, with the exception of the 13-fold increase observed for R106A, the K is not markedly affected by the site-directed replacement of the residues investigated. The decrease in kcat observed for the S326A variant reflects the role of this residue in tethering the side chain of K198, the catalytic base. Although no structures exist of eCGS bound to active-site ligands, the roles of individual residues is consistent with the structures inhibitor complexes of related enzymes. Substitution of D45, E325, or Y101 enables a minor transamination activity for the substrate L-Ala.
Cystathionine γ-synthase (CGS) is the first enzyme of the transsulfuration pathway, which converts L-cysteine (L-Cys) to L-Hcys, the immediate precursor of L-methionine (Fig. 1). The branch-point between L-methionine (L-Met) and L-threonine biosynthesis is L-homoserine in bacteria and O-phospho-L-homoserine (L-OPHS) in plants. Therefore, the bacterial and plant CGS enzymes condense distinct forms of activated L-homoserine, O-succinyl-L-homoserine (L-OSHS), and L-OPHS, respectively, with L-Cys to produce L-cystathionine (L-Cth).1 Structures are available for Escherichia coli CGS (eCGS) and Nicotiana tabacum CGS (nCGS), as well as for the latter in complex with 3-(phosphonomethl)pyridine-2-carboxylic acid (PPCA), 5-carboxymethylthio-3-(3'-chlorophenl)-1,2,4-oxadiazol (CTCPO) and DL-E-2-amino-5-phosphono-3-pentenoic acid (APPA) (Fig. 1).2, 3 The structural similarity of the eCGS (1CS1) and nCGS (1I41) is demonstrated by the r.m.s. deviation of only 2.04 Å for ∼350 Cα atoms, the majority of the peptide backbone.3, 4 Additionally, the r.m.s. deviation is only 0.3 Å between the native and inhibitor-bound forms of nCGS.2 The bacterial and plant CGS enzymes are attractive targets for the development of novel anti-microbial compounds and herbicides because they are not present in mammals.
The enzymes of the γ-subfamily of fold-type I of pyridoxal 5′-phosphate (PLP)-dependent enzymes display remarkable structural similarity, as exemplified by the ∼1.5 Å r.m.s. deviation between ∼350 Cα atoms observed on superposition of the structures of eCGS (1CS1), Trichomonas vaginalis methionine γ-lyase (tMGL; 1E5F), yeast (Saccharomyces cerevisiae) cystathionine γ-lyase (yCGL; 1N8P) and E. coli CBL (eCBL; 1CL1), the second enzyme of the bacterial transsulfuration pathway.5 These enzymes catalyze α,β and α,γ-elimination and replacement reactions on similar amino acid substrates and share common active-site features including the residues corresponding to Y46, R48, Y101, S326, and R361 of eCGS, as well as K198, the catalytic base. In contrast, D45, R106, N227, and E325 of eCGS are shared by CGL sequences, but not by CBL and eCGS-R49 is common to bacterial CGS and CBL enzymes, but is replaced by a serine in yeast CGL and a tyrosine in the plant enzyme nCGS. Interestingly, an arginine at position 106 of eCGS is conserved in bacterial and plant CGS enzymes, but while this residue is proposed to bind the distal carboxylate of L-OSHS, it does not interact with the phosphonate moiety of APPA, an L-OPHS analog, in the structure of the nCGS complex.2, 4
The crystal structures of the plant nCGS enzyme in complex with inhibitors (Figs. 1 and 2) provides valuable insight into active site interactions.2 However, the difference in the activated L-homoserine substrates and the active sites of the plant and bacterial enzymes complicates the use of these structures as a model for eCGS.2–4 Additionally, no information is available concerning the binding site of L-Cys. The side chains of R48, R106, Y101, S326, and R361 and those of E45, R49, and E325 were proposed to bind the L-OSHS and L-Cys substrates, respectively, of eCGS.4 Residues R58 and R372 of eCBL, which correspond to eCGS-R48 and R361, bind the distal and α-carboxylate moieties of the L-Cth substrate, the product of CGS.6, 7 A key determinant of β-elimination reaction specificity in the context of the eCBL active site is S339, which is not involved in substrate binding but tethers the catalytic base following its release from the PLP cofactor upon formation of the external aldimine.6 A series of 16 site-directed variants of eCGS residues D45, Y46, R48, R49, Y101, R106, N227, E325, S326, and R361 was constructed to probe the specific roles of these residues in the γ-elimination and replacement reactions catalyzed by this enzyme. With the exception of R106A, the of all variants is within 3-fold of the wild-type enzyme, indicating that the other nine targeted residues do not interact directly with the L-Cys substrate. In contrast, the observed increases in the values of the R48K, R106A,K, and R361K variants identify these residues as interacting with L-OSHS. The 40 and 360-fold decreases in the kcat of the α,γ-elimination and replacement activities of the alanine replacement variant of S326 distinguish this residue, which guides the ε-amino group of K198, the catalytic base. A novel role for residues D45, Y101, and E325 as determinants of reaction promiscuity and specificity is described.
Site-directed variants of E. coli CGS (eCGS) were constructed and characterized to probe the roles of residues D45, Y46, R48, R49, Y101, R106, N227, E325, S326, and R361 in the active site. The effects of the substitutions on the catalytic efficiency of the physiological α,γ-replacement and minor α,γ-elimination reactions, as well as the reaction with L-Ala, were investigated. The wild-type and 16 site-directed variants of eCGS were purified to homogeneity. All enzymes are soluble and, with the exceptions of R48A and R361A, possess detectable levels of the physiological α,γ-replacement, and minor α,γ-elimination, activities and are substrate inhibited by L-Cys but not L-OSHS. The hydrolysis of O-acetylserine (OAS) is undetectable for the wild-type enzyme and the variants investigated, demonstrating their lack of α,β-elimination activity, in contrast with the closely related eCBL.6 The L-Cth hydrolysis activity of the variants was also assayed (data not shown) and follows a similar trend to the α,γ-elimination of L-OSHS (Table II). The low catalytic efficiency of the wild-type enzyme for hydrolysis of L-Cth (kcat/ = 23.3 ± 0.8 M−1s−1), two orders of magnitude lower than the α,γ-elimination of L-OSHS (kcatE/ = 2000 ± 200 M−1s−1), precludes meaningful comparison for variants with decreased kcat or increased .
Table II. Kinetic Parameters of the α,γ-Elimination of OSHS by Wild-Type Site-Directed Variants of eCGSa
kcatE/ (M−1 s−1)
Kinetic parameters reported are for hydrolysis of L-OSHS. Reaction conditions: 0.1–50 mM L-OSHS and 1–26.25 μM wild-type or mutant eCGS, depending on the activity of the enzyme, in assay buffer at 25°C. The data were fit to the Michaelis–Menten equation to obtain kcat and and Eq. (6) to obtain kcat/ .
Values for wild-type eCGS are from Aitken et al.17
The value of kcatE/ of R48K was determined via linear regression based on the assumption that >> [L-Cth], because this variant showed no evidence of saturation within the solubility limit of the substrate. Therefore, the estimates of kcatE and are provided only as estimates to illustrate the degree to which substrate binding is altered by the conservative substitution of R48 by lysine.
The value of is unchanged by site-directed substitutions targeting R49 and E325 and is increased only ∼3-fold by replacement of D45 with alanine or asparagine (Table I). Although the kcatR of the α,γ-replacement activity is reduced 20-fold by the R49A substitution (Table I), the kcatE for the α,γ-elimination of L-OSHS is unchanged (Table II) and the kinetic parameters of R49K are identical to those of wild-type eCGS, within experimental error (Tables I and II). The 5- to 10-fold reductions in the kcatR/ and kcatR/ of the alanine and asparagine replacement variants of D45 are the result of 3- to 12-fold increases in and as kcatR is decreased ≤2-fold by these substitutions (Table I). Interestingly, although the kinetic parameters of the α,γ-elimination of L-OSHS are unaffected by the alanine and glutamine substitutions of E325, the , and values, for the α,γ-replacement activity, of these variants are decreased ∼5-fold (Tables I and II, Fig. 3).
Table I. Kinetic Parameters for the Condensation of L-OSHS and L-Cys by Wild-Type Site-Directed Variants of eCGSa
kcatR/ (M−1 s−1)
kcatR/ (M−1 s−1)
Kinetic parameters reported are for condensation of L-OSHS and L-Cys. Reaction conditions: 0.1–50 mM L-OSHS, 0.025–20 mM L-Cys, 1.3 mM NADH, 1.0 μM eCBL, 1.0 μM LDH, and 0.094–21.25 μM wild-type or mutant eCGS, depending on the activity of the enzyme, in assay buffer at 25°C. The data were fit to the Eqs. (1)–(3).
Kinetic parameters of eCGS reported by Aitken et al.17
n.s. indicates that exceeds the solubility limit of the L-OSHS, such that kcatR/ was determined via linear regression, at the L-Cys concentration for which maximal activity was observed, and kcatR, and were determined from the fit of velocity versus L-Cys, at the highest concentration of L-OSHS used, to Eq. (4).
Upon reaction of wild-type eCGS with 200 mM L-alanine, the 421-nm absorbance, corresponding to the ketoenamine tautomer of the internal aldimine of the PLP cofactor, is decreased with the concomitant increase in a 320–330 nm shoulder [Fig. 4(A)], corresponding to the enolimine tautomer of the external aldimine of L-Ala or to a ketimine intermediate formed upon subsequent deprotonation of Cα and protonation of C4′. The behavior of the R49A, R49K, and E325Q enzymes is identical to that of native eCGS [Fig. 4(A)]. In contrast, upon reaction of the D45A, D45N, and E325A substitution variants with L-Ala, the 421-nm peak of the internal aldimine decreases in intensity, with a concomitant formation of a 320-nm peak, which is distinct from the 320–330 nm shoulder observed for the wild-type enzyme (Fig. 4).
The R48K, R106A,K, R361K, and N227A variants
Residue eCGS-N227 corresponds to Y238 of eCBL, which binds the distal carboxylate of the substrate.6 However, in the context of the eCGS active site, replacement of N227 with alanine results in modest, <4-fold changes in kinetic parameters, compared with the wild-type enzyme (Tables I and II). The 100, 40, 760, and 310-fold decreases in the catalytic efficiency of the α,γ-elimination activity of the R106A, R106K, R48K, and R361K variants are dominated by the 20- to 25-fold and ∼230-fold increases in the of the R106 variants and the R48 and R361 lysine-substitution variants, respectively, as the kcatE is decreased only 2- to 7-fold (Table II). The and values, for the physiological, α,γ-replacement activity, of the R106 variants are both increased, by 9- and 3-fold, respectively (Fig. 3), and the 80- and 20-fold decreases in the kcatR of R106A and R106K are of similar magnitude (Table I). The data for the R48K and R361K variants could not be fit to Eqs. (1)–(3) as saturation was not observed within the solubility limit of the L-OSHS substrate. Therefore, with the assumption that >> [L-OSHS], the Michaelis-Menten equation was modified to obtain kcatR/ , at the concentration of L-Cys at which each enzyme displays maximal activity (Table I). Values of kcatR, , , and kcatR/ were determined for the R48K and R361K variants from the fit of velocity versus L-Cys, at the highest concentration of L-OSHS used, to Eqs. (4) and (5). Therefore, the values of kcatR, , and for these two enzymes should be considered as lower limits for the values of these kinetic parameters as the L-OSHS concentration was limited by its solubility under assay conditions. The and values of these enzymes are within 5-fold of wild-type eCGS. Although the 750- and 220-fold decreases in kcatR dominate the 12,000 and 4600-fold reductions in the apparent kcatR/ values of R48K and R361K (Table I), the degree to which is increased cannot be accurately determined. The behavior of the R48, R106, and R361 targeted substitution variants in the presence of 200 mM L-Ala is identical to that of wild-type eCGS.
The Y46F, Y101F, and S326A mutants
The 350-fold decrease in the kcatR/ of the α,γ-replacement activity of eCGS-Y46F is comprised of equal, ∼20-fold changes in kcatR and , whereas the is increased by only 3-fold. Although the kcatE of the minor α,γ-elimination activity of Y46F is unchanged, compared to the wild-type enzyme, a 12-fold increase is observed for , similar to that of . The kcatE and kcatR of eCGS-Y101F are reduced 2- and 12-fold, respectively, and values of both the γ-elimination and replacement reactions are within 5-fold of the wild-type enzyme. The 40- and 360-fold decreases in kcatE and kcatR of eCGS-S326A are unique among the site-directed variants investigated and the and values of this variant are increased by only 5-fold (Tables I and II, Fig. 3). Although the behavior of S326A in the presence of 200 mM L-Ala is identical to that of eCGS, the Y46F enzyme displays a minor 320-nm peak following a 4-h incubation. In contrast, complete conversion of the 421-nm species to one with a distinct 320-nm absorbance, identical to D45A, D45N, and E325A, is observed for Y101F [Figs. 4(B) and 5]. To investigate the nature of the 320-nm species, the Y101F variant was incubated with 200 mM L-Ala for 4 hours followed by dialysis, to remove the excess substrate, and a subsequent, 4-h incubation with 50 mM pyruvate to return any pyridoxamine (PMP), produced on reaction with L-Ala, to the internal aldimine of PLP. Incubation with pyruvate resulted in a 3.8-fold increase in the intensity of the 421-nm band of Y101F, suggesting that PMP was formed upon reaction with L-Ala and that it reacted with the α-ketoacid pyruvate to produce alanine and regenerate the PLP form of the cofactor, thereby completing a transamination catalytic cycle (Fig. 5).
The enzymes of the γ-subfamily of fold-type I, including those of the transsulfuration pathways, comprise an ideal model system for investigation of the structure-function relationships underlying substrate and reaction specificity in PLP-dependent enzymes. The structural similarity of CGS, CBL, and CGL extends to the active site as five of the residues proposed to participate in substrate binding and catalysis are conserved in these three enzymes, from both prokaryotic and eukaryotic species.5–7 These conserved positions include a pair of arginines (eCGS-R48 and R361), two tyrosines (eCGS-Y46 and Y101), and a serine residue (eCGS-S326). An additional five active-site residues demonstrate conservation with either CGL (eCGS-D45, R106, N227, and E325) or CBL (eCGS-R49). The conformation and orientation of substrate(s) within the active site as well as the degree of freedom of rotation about the Cα-Cβ bond, of the substrate covalently bound to the cofactor, have been proposed to be determinants of reaction specificity among the enzymes of the γ-subfamily of fold-type I of PLP-dependent enzymes.7, 8 However, these factors are subtle and not evident from comparison of the available crystal structures of the unliganded or inhibitor-bound complexes of these enzymes, from a mixture of bacterial and plant species, as a conserved residue may play different roles dependent on the context of the specific active site. Additionally, interconversion of the two residues most distinct between the active sites of the α,γ-elimination catalyzing eCGS (D45 and E325) and yCGL (E48 and E333) and eCBL (F55 and Y338), which catalyzes an α,β-elimination reaction, does not result in a corresponding increase in α,γ or α,β-elimination activity.9 Therefore, exploration of the roles of active-site residues in these enzymes, dependent on the versatile PLP cofactor, is necessary to decipher the subtle and complex structure-function relationships of these structurally similar but mechanistically distinct enzymes. The characterization of a series of 16 site-directed variants of 10 eCGS active-site residues (D45, Y46, R48, R49, Y101, R106, N227, E325, S326, and R361) is the focus of this study.
The D45A,N, R49A,K, and E325A,Q variants
Clausen et al. proposed that the acidic residues D45 and E325 and basic R49 interact with the α-amino and α-carboxylate groups, respectively, of the L-Cys substrate.4 The near-native values of the alanine substitution variants of these residues, which remove their hydrogen-bonding capacity, do not substantiate this mode of L-Cys binding (Table I). The observation that, with the exception of R106A, was increased, ≤3-fold, by only D45A,N and Y46F, of the 16 site-directed variants of eCGS active-site residues investigated, suggests that L-Cys may bind via backbone or water-mediated interactions or that the succinate leaving group of L-OSHS may be involved in the positioning of this substrate. The 20-fold reduction in the kcatR and the 4-fold increase in the , of the α,γ-replacement and elimination reactions, respectively, suggest that replacement of R49 with alanine subtly alters the architecture of the active site. The wild-type kinetic parameters of the R49K variant, unaltered despite the ∼1 Å shorter side-chain of lysine, confirm that this residue does not play a direct role in substrate binding or catalysis. In contrast with the corresponding R59 of eCBL, which forms a salt bridge with residue E235 of the neighboring subunit of the catalytic dimer, the side chain of eCGS-R49 does not interact with other amino acids.4, 10 The 4- to 9-fold reductions in the kcatE/ and kcatR/ of the D45A and D45N variants reflect minor, 2- to 4-fold changes in kcat and resulting from subtle alterations in active-site architecture or the positioning of water molecules. In contrast, replacement of residue D45 with phenylalanine, to mimic the corresponding F55 of eCBL, decreases kcat by 16-fold and causes a 6-fold increase in . 9 These results suggest that, although this residue does not participate directly in substrate binding, it is situated in proximity to the L-Cys binding site. Similarly, the unique 4- to 6-fold decreases in the and values of the alanine and glutamine substitutions of E325 (Table I, Fig. 3) indicate that this residue is located in the L-OSHS binding site and that its negative charge reduces the binding of both substrates to this site but does not contribute to the ability of the enzyme to discriminate between them. Residue E325 of eCGS is conserved as a glutamate in CGL but replaced by tyrosine in bacterial CBL (eCBL-Y338). Messerschmidt et al. proposed a role for the corresponding E333 of yCGL as a determinant of α,γ versus α,β-elimination specificity via interaction with the sulfur atom of the L-Cth substrate, common to CGL and CBL.5 Replacement of eCBL-Y338 with phenylalanine has negligible effect on the kinetic parameters of eCBL and no discernible effect on reaction specificity.6 In the context of the eCGS active site, residue E325 acts as a determinant of reaction promiscuity as the ratio of the γ-replacement/γ-elimination catalytic efficiencies, for L-OSHS, is increased ∼5-fold by both the alanine and glutamine substitution variants. A possible evolutionary role for the minor γ-elimination activity is the production of propanoyl-CoA, the cellular precursor of C5 branched dibasic acids. Alternatively, the minor γ-elimination activity of eCGS may prevent alternative substrates from reacting with the ketimine intermediate in the situation that the cellular pool of the L-Cys substrate is diminished. A role for this residue as a determinant of reaction specificity is also implied by the observation that the E325A variant displays the 320-nm PMP absorption, suggestive of transamination, upon reaction with L-Ala.
The R48K, R106A,K, R361K, and N227A variants
Residue eCGS-N227 was targeted in this study because the corresponding Y238 of eCBL has been shown to interact with the distal carboxylate of the substrate.6 The Nδ1 and Oδ1 atoms of N227 are 3.3 and 2.8 Å from the guanidinium groups of the side chains of R106 and R48, respectively, suggesting that this residue may play a role in positioning of one or both of these arginine residues.4 However, the negligible impact of removal of the hydrogen-bonding capacity of this residue on the kinetic parameters of eCGS (Tables I and II), resulting from its replacement with alanine, demonstrate that N227 does not interact with either substrate or play an important role in the positioning of other active-site residues in the context of the eCGS active site.
Docking studies with eCGS predicted roles for three arginine residues, R48/R106 and R361, in binding the distal and α-carboxylate moieties, respectively, of the L-OSHS substrate.4 This model is supported by the 130, ∼10, and 120-fold increases in for the α,γ-elimination activity of the R48K, R106A,K and R361K variants, respectively (Table II). Similarly, the value for the α,γ-replacement activity is increased 3- and 9-fold (Fig. 3), respectively, by substitution of R106 with alanine or lysine, whereas the R48K and R361K variants are not saturated within the solubility limit of L-OSHS (Table I). Residue R106 of eCGS is conserved as an arginine in prokaryotic and eukaryotic CGS and CGL sequences and replaced by an aspartate in bacterial CBL (eCBL-D116). Substitution of the side chain of D116, located 8.3 Å from the distal amino group of AVG, in the structure of the eCBL complex with this inhibitor, does not modify the kinetic parameters of eCBL.7, 8 Similarly, the side-chain of R167 of the plant nCGS enzyme, which corresponds to eCGS-R106 and eCBL-D116, is oriented away from the phosphonate moiety of the inhibitor in the nCGS-APPA complex (Fig. 2), suggesting that although this residue likely does not bind the distal phosphate group of the L-OPHS substrate, it may interact with the α-carboxylate of L-Cys.2 However, the 3- to 9- and ∼10-fold increases in the and values of the R106A,K variants, suggest that this residue interacts, directly or in a water-mediated manner, with the distal portion of the L-OSHS substrate of eCGS (Tables I and II, Fig. 1). The 13-fold increase in of the alanine substitution variant of R106 is unique among the 16 active-site variants investigated. Residue R106 may be situated such that it bridges the carboxylate groups of the succinate moiety of L-OSHS and of L-Cys.
Although residue nCGS-R110, which corresponds to eCGS-R48, does not interact with the phosphonate moiety of the inhibitor in the structure of the nCGS-APPA complex, the distal carboxylate of L-OSHS was proposed to form a salt bridge with the corresponding eCGS-R48.2, 4 This is supported by the 120-fold increase in and by comparison of the effect of alanine and lysine substitution of the corresponding eCBL-R58 on the Ki for inhibition by AVG, lacking the distal carboxylate moiety of L-Cth, and the , which demonstrates the interaction between the distal carboxylate of the substrate and R58. A pair of hydrogen bonds between the α-carboxylate group of the inhibitor and the side chain of R423, which corresponds to eCGS-R361, are apparent in the structure of the nCGS-APPA complex (Fig. 2).2 The 130-fold increase in the and lack of L-OSHS saturation observed for the α,γ-replacement activity of R361K (Tables I and II), support the proposed role of R361 in L-OSHS binding. A similar 80-fold increase in has been reported for the corresponding R372K variant of eCBL, which also interacts with the α-carboxylate moiety of the inhibitor in the eCBL-AVG complex.7, 8 The ≥220-fold decrease in the kcat of the α,γ-replacement activity of eCGS-R361K is in keeping with the 55-fold decrease in this parameter reported for the corresponding AATase-R386K but contrasts with the 4-fold decrease in the kcat of CBL-R372K.7, 11, 12 The observed reductions in the activity of the eCGS and eAAT variants reflects differences in the binding orientation and conformation of the substrates, such that they are not optimally positioned for catalysis.12 In contrast, the substantially smaller decrease in the kcat of eCBL-R372K likely reflects the facile nature of the α,β-elimination of L-Cth, which, in contrast with the transamination of eAATase and α,γ-replacement reaction of eCGS, does not require the generation of a ketimine intermediate or the binding of a second substrate. The observed 80- and 750-fold decreases in the kcatR of the R106A and R48K variants are likely also the result of suboptimal positioning and conformation of L-OSHS within the active site of eCGS. Additionally, as R48 is one of the primary ligands of the phosphate moiety of PLP, its substitution by lysine, the side chain of which is approximately 1 Å shorter, may result in a change in the positioning of the cofactor and corresponding alteration of active-site architecture.4
The Y46F, Y101F, and S326A variants
Based on the structure of eCGS, a role for residues Y101 and S326 in binding the distal and α-carboxylate groups, respectively, of L-OSHS was proposed; a model which is not supported by the minor, 2- and 5-fold changes in of the Y101F and S326A variants, respectively (Table I).4 The tyrosine at position 46 of eCGS was targeted for investigation in this study because it is conserved in many members of fold-type I and phenylalanine substitution variants of the corresponding residue in diverse enzymes, including Y56 of eCBL, Y70 of E. coli aspartate aminotransferase, Y121 of murine erythroid aminolevulinate synthase (meALAS), Y71 of Citrobacter freundii tyrosine phenol lyase and Y64 of Treponema denticola cystalysin, have been reported to impact the kinetic parameters of the target enzyme.6, 13–16 The 20-fold decrease in kcatR and 20 and 12-fold increases in and of eCGS-Y46F (Table I) are of common magnitude with the 6-fold decrease in kcat and 20-fold increase in of the corresponding Y56F variant of eCBL. Based on the similar, 15-fold increase in the Km for glycine of the corresponding meALAS-Y121F variant, Tan et al. proposed that the observed change in Km is due to the increase in of the cofactor, rather than a direct interaction with the substrate, as glycine does not possess a side chain.15 Given that the hydrogen bond between the phosphate moiety of the cofactor and residue Y121 of meALAS, is conserved in eCBL and eCGS, the observed similar increases in the Km values of the corresponding Y56F and Y46F variants, may also reflect the absence of this link rather than a direct interaction with the substrate.
The 2- to 12-fold changes in the kcat and Km values of eCGS-Y101F demonstrate that Y101 is not directly involved in catalysis, demonstrating that the role proposed for this residue in proton transfer between the α-amino group of L-OSHS and the succinate leaving group is not supported.4 However, the transamination of L-Ala, not observed for the wild-type enzyme, is most apparent for Y101F (Figs. 4 and 5), among the 16 variants investigated, demonstrating that the hydroxyl moiety of this residue acts as a minor determinant of reaction specificity. This contrasts with eCBL for which the corresponding Y111F variant does not demonstrate transamination activity, but is uniquely inhibited by DTNB, illustrating that conserved residues play distinct roles in the active sites of the three γ-subfamily enzymes of the transsulfuration pathways. Among the corresponding substitution variants of eCBL active-site residues that have been investigated, a novel transaminase activity was observed for S339A. Interestingly, no 320-nm absorbance, indicative of PMP was observed upon reaction of the corresponding eCGS-S326A.
The 40- and 360-fold decreases in kcatE and kcatR, respectively, of eCGS-S326A are unique among the site-directed variants investigated and in agreement with the effect of the corresponding S339A variant of the closely related eCBL (Tables I and II, Fig. 3). Therefore, a role in guiding and tethering the ε-amino group of K198, the catalytic base, proposed for eCBL-S339, is also predicted for eCGS-S326. The presence of a serine residue at this position is unique to the enzymes of the γ-subfamily of fold-type I. A 5600-fold decrease in kcat for the hydrolysis of L-Cth has been observed for the corresponding S339A variant of eCBL.6 The 140-fold greater effect of the S339A substitution on the kcat of α,β-elimination activity of eCBL, compared to the α,γ-elimination activity of eCGS is insightful. The ε-amino group of eCBL-K210 must be more strictly restrained than the corresponding eCGS-K198 because protonation of C4′ of the cofactor is required in the α,γ-elimination reaction of eCGS but must be prevented in eCBL.
CGS is an attractive target for the development of novel antimicrobial compounds because it catalyzes the first reaction in the transsulfuration pathway, which is unique to plants and bacteria. The design of effective inhibitors of the bacterial CGS enzyme requires an in-depth characterization of the active-site residues participating in substrate binding. This study has identified R48/R106 and R361 as residues interacting with the distal and α-carboxylate groups, respectively, of the L-OSHS substrate, interactions which can be targeted for the design of inhibitors. However, with the exception of the 13-fold increase observed for R106A, the negligible effect of the 16 site-directed variants on demonstrates that the residues investigated do not directly participate in L-Cys binding. A role for Y101, D45, and E325 as determinants of reaction specificity is proposed as substitution of these residues resulted in a minor L-Ala transaminase activity not observed for the wild-type enzyme. Given the ∼9 and 8 Å separation between each of these acidic residues and Y101, a direct interaction is not anticipated as the underlying structural feature preventing the transamination side reaction in the wild-type enzyme.
Materials and Methods
L-Cth, L-Cys, OSHS, and L-lactate dehydrogenase (LDH) were purchased from Sigma. Ni-NTA resin was obtained from Qiagen. Oligonucleotide primers were synthesized by Integrated DNA Technologies and mutants were sequenced by BioBasic before expression and purification. The 6-His-tagged eCBL and d-2-Hydroxyisocaproate dehydrogenase (HO-HxoDH) coupling enzymes used in the eCGS γ-replacement and γ-elimination assays, respectively, were expressed and purified as described previously.17–19
Construction, expression, and purification of site-directed mutants
Site-directed mutants of eCGS, constructed via the overlap-extension polymerase chain reaction method, were inserted into the pTrc-99aAF plasmid. The amino-terminal, 6-His tag, and linker encoded by this vector enables affinity purification of the expressed enzymes but, as demonstrated by Farsi et al., it does not alter the kinetic parameters of eCGS.9 Wild-type and site-directed variants of eCGS were expressed in the E. coli ER1821 metB::aadA strain, in which the gene encoding eCGS is replaced by aadA, encoding resistance to streptomycin, to prevent contamination with the wild-type E. coli enzyme. The wild-type and site-directed mutants of eCGS were expressed and purified, via Ni-nitrilotriacetic acid affinity chromatography, as described by Farsi et al.9
Determination of steady-state kinetic parameters
Enzyme activity was measured in a total volume of 100 μL at 25°C on a Spectramax 340 microtiter plate spectrophotometer (Molecular Devices). The assay buffer was comprised of 50 mM Tris, pH 7.8, with 20 μM PLP. The formation of L-Cth, via the condensation of L-OSHS and L-Cys, was detected using the CBL-LDH coupled assay.17 A background reading was recorded, before initiation of the reaction by the addition of the experimental enzyme, in all assays. The data for the γ-replacement activity of eCGS were fitted to Eqs. (1)–(3) for the modified ping pong mechanism described by Aitken et al., in which the E and R subscripts denote the γ-elimination and γ-replacement activities, respectively, to obtain kcatR, , , , kcatR/ , and kcatR/ . The independently determined values of kcatE and , for the γ-elimination activity, were substituted into Eqs. (1)–(3) to reduce the number of kinetic parameters to be determined.17 Data were fit by nonlinear regression with the SAS software package (SAS Institute, Cary, NC).
Values of kcatR, , and for the R48K and R361K variants, for which exceeds the solubility of L-OSHS, were determined from the fit of velocity versus L-Cys, at the highest concentration of L-OSHS used, to Eq. (4), which incorporates the term for substrate inhibition by L-Cys. The parameter kcatR/ was obtained independently from Eq. (5).
The hydrolysis of L-OSHS was detected via the continuous γ-elimination assay, in which α-ketobutyrate is reduced, with concomitant oxidation of NADH (ε340 = 6,200 M−1s−1), by HO-HxoDH.17 Values of kcatE and were obtained by fitting of the data to the Michaelis-Menten equation and kcatE/ was obtained independently from Eq. (6)
The authors thank the reviewers of this manuscript for their insightful comments and suggestions.