If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
O-acetyl serine sulfhydrylase (OASS), referred to as cysteine synthase (CS), synthesizes cysteine from O-acetyl serine (OAS) and sulfur in bacteria and plants. The inherent challenge for CS is to overcome 4 to 6 log-folds stronger affinity for its natural inhibitor, serine acetyltransferase (SAT), as compared with its affinity for substrate, OAS. Our recent study showed that CS employs a novel competitive-allosteric mechanism to selectively recruit its substrate in the presence of natural inhibitor. In this study, we trace the molecular features that control selective substrate recruitment. To generalize our findings, we used CS from three different bacteria (Haemophilus, Salmonella, and Mycobacterium) as our model systems and analyzed structural and substrate-binding features of wild-type CS and its ∼13 mutants. Results show that CS uses a noncatalytic residue, M120, located 20 Å away from the reaction center, to discriminate in favor of substrate. M120A and background mutants display significantly reduced substrate binding, catalytic efficiency, and inhibitor binding. Results shows that M120 favors the substrate binding by selectively enhancing the affinity for the substrate and disengaging the inhibitor by 20 to 286 and 5- to 3-folds, respectively. Together, M120 confers a net discriminative force in favor of substrate by 100- to 858-folds.
). A number of studies have reported that enzymes employ induced-fit mechanism to form catalytically competent “enzyme·substrate complex” as compared with short-lived, nonproductive “enzyme·competitor·ligand” complexes (
Cys-Gly specific dipeptidase Dug1p from S. cerevisiae binds promiscuously to di-, tri-, and tetra-peptides: peptide-protein interaction, homology modeling, and activity studies reveal a latent promiscuity in substrate recognition.
). In a recent study, we showed that substrate-binding-induced active site conformational changes allowed CS to selectively recruit its substrate, O-acetyl serine (OAS), in the presence of high-affinity natural inhibitor, the C-terminal of serine acetyl transferase (SAT). Our extensive structural and analytical work revealed that only substrate binding can induce the active site of CS to a closed state, not the binding of the high-affinity inhibitor. We proposed a novel “competitive-allosteric” mechanism by which CS selectively recruits its substrate while bound to high-affinity natural inhibitors (
). Binary and ternary complexes of enzyme·substrate, enzyme·inhibitor, and enzyme·inhibitor·substrate complexes, in combination with fast kinetics, showed that CS achieves selective substrate recruitment in the presence of high-affinity inhibitor by employing a novel “competitive-allosteric” mechanism (
). Glucose-binding-induced ATPase activity of hexokinase, steric switch mechanism of ribosome, correct substrate base-binding-induced organization of DNA polymerase active site are classical examples of substrate-induced functional specificity (
). However, mechanism of substrate selectivity achieved during cysteine synthesis step by CS is different because, the active site-bound high-affinity inhibitor must be removed before substrate enters.
The most important difference is the mode of recognition of inhibitor and substrate by CS (
Structural insights into catalysis and inhibition of O-acetylserine sulfhydrylase from Mycobacterium tuberculosis. Crystal structures of the enzyme alpha-aminoacrylate intermediate and an enzyme-inhibitor complex.
). We traced the molecular origins of substrate selectivity in this study. We analyzed structural, activity, and ligand recognition properties of CS from three different species (Haemophilus, Salmonella, and Mycobacterium) and ∼13 CS mutants of these three enzymes. Our results show that mutation of either M120 or M92 significantly reduces the substrate binding and catalytic efficiency of CS, whereas M96 shows limited or insignificant contribution to substrate recruitment. We present a generalized model of gated substrate-recruitment mechanism in which M120 acts as a gate sensor, which triggers allosteric conformational changes that disengage inhibitor (SAT) and allow OAS to enter.
Analyses of active site channel features for identification of substrate discriminative residues
CS from both bacteria and plants switch to closed-state structure only upon binding to the substrate, not the inhibitor, SAT C-terminal. The substrate-induced structural change is of very significant magnitude, evidenced by multiple crystal structures of CS in complex with OAS (
). As shown in the Figure 1A, the α-helix5 moves ∼8 Å toward the active site channel after OAS enters the channel and reacts with active site pyridoxal 5′phosphate (PLP) (shown in yellow). With α-helix5, the N-terminal domain constituting residues from 65 to 125 moves as one unit to close the active site channel. Therefore, it is reasonable to expect that front-line residues of movable domain that line the active site entrance should be the first set of residues that come in contact with the incoming OAS. Indeed, a ternary complex crystal structure ((PDB Code: 4ORE) resolved in our recent study showed that M120 present in the N-terminus of α-helix5 is in contact with the substrate, OAS (Fig. S1A) (
). Therefore, we performed multiple sequence alignment and comparative structural analyses to check whether this M120 is conserved and also searched for residues that are highly conserved but undergo large structural changes when the active site of CS switches from open to closed state. Among the many residues that are highly conserved within the movable domain, we noticed a network of three conserved methionine residues, which move in tandem but exhibit large structural movements during the transition to the closed state (Fig. 1, B–C). In the open state, these three methionine residues, M120, M96, and M92, form a triangle, and they are structurally disposed at ∼20 Å from each other (Fig. 1D). M120 is located at the end of α-helix 5, which is connected to the α-β5 loop, a mobile loop that has no electron density in the closed state. Based on comparative sequence and structural analyses, we decided to investigate the role of this “methionine trio-network” in substrate recruitment.
Mutation of methionine 120 significantly reduces cysteine synthesis activity
The catalytic cycle of CS comprises three distinct phases; substrate recruitment, reaction with active PLP, and product (cysteine) formation. As a first step, we examined the role of methionines in the first phase, i.e., the recruitment of substrate into the active site channel. We generated 13 mutants (variants) of CS enzymes from three different bacteria (Haemophilus influenzae, Salmonella typhimurium, and Mycobacterium tuberculosis) as described in the methods. We targeted three methionine sites, M92, M96, and M120, and created these 13 mutants, either as point mutants or as combination mutants (double and triple mutants). For H. influenzae CS, we generated approximately seven mutants; HiM92A, HiM96A, HiM120A (three single mutants), HiM92 AM96A, HiM92 AM120, HiM96 AM120 (three double mutants, DM), and HiM92 AM96 AM120A (one triple mutant, TM). For S. typhimurium CS, three single mutants StM92A, StM96A and StM120A were generated. Similarly, for M. tuberculosis CS, three single mutants MtM92A, MtM96A, and MtM120A were created. All 16 proteins were purified by affinity and size-exclusion chromatography methods. Size-exclusion profiles of mutants as compared with that of their respective wild-type CS show that these mutations do not alter the oligomeric state of the protein, and all mutants elute as homodimers (Figs. S2, A–C and S3, A–B). CS and 13 mutants were further characterized by circular dichroism (CD) spectroscopy method, which indicated that secondary structural content CD signatures of mutants are similar to that of wild-type (Fig. S2D).
All three methionines, M92, M96, and M120, are noncatalytic residues, located 10 to 20 Å away from the reaction center. To assess their impact on cysteine synthesis activity of CS, we examined catalytic properties of 13 methionine mutants. We performed single-point activity assays at saturating substrate concentrations to examine the effect of mutations on cysteine synthesis. As shown in (Fig. S4, A–C), mutation of M120 with alanine either as single mutation or in background (HiM120A, StM120A, Mt120A, HiM92 AM120A, HiM96 AM120A) significantly reduces the cysteine synthesis activity by ∼65 to 78%. HiM96A, StM96A, and MtM96A mutants exhibit almost similar or slightly more cysteine synthesis activity than that of WT protein, suggesting that the M96 plays very limited role in directly controlling the catalytic property of CS. Cysteine synthesis activities of M92A mutants were lower than the wild-type, but the extent of decrease was species-specific and varied from 25 to 75%. Both StM92A and HiM92A exhibited ∼75% activity with reference to their respective wild-type enzymes, but MtM92A displayed quite less, only ∼25% activity (Fig. S4B). In summary, mutation of M120 significantly reduced cysteine synthesis activities of CS from all three species (HiCS, StCS, and MtCS) to ∼35 to 22%.
Catalytic turnover rates of M120A, M92A, and M120A background mutants are significantly reduced
To quantify and compare the catalytic efficiencies of mutants with that of their wild-type, we performed detailed steady-state kinetic analysis of all 16 enzymes in triplicates. As expected from single-point activity assays, amplitudes of M120A, M92A, and M120A background mutants are reduced significantly (Fig. 2, A–C). Therefore, KM (apparent substrate affinity) determined from such low-amplitude kinetics data is not reliable as substrate concentration-dependent initial velocities saturate fast with very few data points in the presaturation phase. However, turnover rates (kcat) of these mutants can be determined more accurately as there are enough data points at the saturation phase. Kinetics of three wild-type CS and M96A mutants display high amplitude and consist of enough data points at both presaturation and saturation phases. Therefore, both KM and kcat for the wild-type CS and M96A mutants can be determined reliably. Initial velocities from triplicate experiments were averaged, and data were fit to Michaelis–Menten model to obtain steady-state kinetic parameters, KM and kcat. Any parameter estimated with high uncertainty will not be sensitive to functional or structural changes. To further check the sensitivity of KM to predict structural outcome, we plotted kcat and KMversus mutant/WT. The plot shows that values of KM exhibit random behavior, whereas there is a clear dichotomous or categorical distribution of kcat values (Fig. 2, D–E). The horizontal lines representing mean values of KM and kcat show that kcat values of all three wild-type CS and two of M96A mutants are much higher than the average (kcat-AV), whereas kcat values of M120A, M92A, and M120A background mutants are below the kcat-AV. Such categorical distribution is absent in the plot of KM distribution. We tabulated the values along with fold changes for each mutant (Table 1). To estimate the (kcat/KM) for comparative analyses, we divided kcat values of wild-type CS and mutants by their respective KM, but for M120A, M92A, and M120A background mutants, we divided their respective kcat by mean of their KM as the mean is better representation of KM for the mutants due to high uncertainty.
Table 1Steady-state kinetic parameters for CS and mutants
M120A mutation has significantly reduced the turnover rate (kcat, 25–107 fold) and catalytic efficiencies (kcat/KM-AV, 11–53 fold) across CS. Similarly, M92A mutant also shows significantly lower catalytic efficiencies, consistent with results of single-end point assay. Turnover rates (kcat) of StM96A are very similar to that of its WT, but the (kcat) for HiM96A is 1.6-fold more than the WT. On the contrary MtM96A displays 25-fold reduced turnover (kcat) compared with its WT, whereas the catalytic efficiencies (kcat/KM) for HiM96A are similar to that of the WT. But the catalytic efficiencies for StM96A and MtM96A are 5- to 18-fold lesser than that of their respective WT. Results of double and triple mutants also show that M120A background mutants display significantly reduced catalytic turnover. Consistent with results of single-end point assay, mutation of noncatalytic residues M120 or M92 reduces the cysteine synthesis ability of CS significantly. Statistical analyses show that differences observed between each set of wild-types and their respective mutants are statistically significant. For example, the mean value and error determined for kcat of HiCS wild-type is 884 ± 35 s−1, which gives a range from 849 (lower bound) to 919 (upper bond). Comparison of the mean values and error range associated with kcat or kcat/KM of wild-type CS and their mutants clearly shows that the error range of mutants do not overlap with that of wild-type (Table 1). Also, we estimated p-value, which indicates that the observed differences between respective wildtype and mutants are statistically significant (Table 1).
M120A and background mutants display reduced reaction intermediate formation
Next, we examined the roles of methionine network in the second phase, PLP-reactive phase, of the catalytic cycle. Since PLP is buried at the deep end of the active site tunnel, the substrate has to travel the ∼20 Å channel to reach reaction center to enter into phase II. To dissect out which one of these three phases is most affected by methionine mutations, we examined reactabilities of mutants by monitoring the extent of reaction intermediate formation. The PLP, linked to active site lysine, is present in the form of internal aldimine and interacts with the in-coming OAS to form a number of reaction intermediates, including a more stable reaction intermediate, α-amino acrylate (
). Distinct absorption properties of α-amino acrylate and other intermediates such as geminal diamine and external aldimine (λmax ∼ 340 nm, geminal diamine; λmax ∼ 418 nm, external aldimine; λmax ∼ 330 nm and λmax ∼ 470 nm, α-amino acrylate) allow us to monitor the extent of reaction between PLP and OAS (
). The peak intensity at 470 nm should be proportional to the extent of reaction intermediate formation, and mutants with compromised substrate recruitment should show reduced signal at 470 nm.
In the case of HiCS, mutation of either M92 or M120 completely abrogates the formation of reaction intermediate (as both 330 nm and 470 nm peaks are not observed), but mutation of M96 has subdued effect (Fig. 3A). However, mutation of 96A in the M120A background (double mutant and triple mutant) of HiCS displays insignificant or no reaction intermediate peaks (Fig. 3B). In the case of MtCS, both M92A and M96A show similar but reduced intermediate formation as the magnitudes of 470 nm peaks are decreased by ∼45%. Even though MtM92A mutant displays significantly reduced catalytic turnover (Table 1), the extent of formation of reaction intermediate indicated by the magnitude of 470 nm peak is higher. Mutation of either M92 or M96 affects only partially as both spectra show more or less similar signatures in MtCS. The 470 nm peak of M120A of MtCS is lower by 70% suggesting that mutation of M120 has more pronounced negative effect on the reaction intermediate formation (Fig. 3C). Similarly, mutation of M120 in StCS reduces the reactivity very significantly but magnitudes of 470 nm peaks of M96A or M92A are more than that of wild-type StCS suggesting that effect of mutations on reaction intermediate formation is not significant (Fig. 3D). It is interesting to note that negative effect of M96A mutation in the MtCS is more pronounced, compared with that of in the StCS, as the relative magnitude of 470 nm peak of MtM96A is reduced by ∼45%. In summary, mutation of M120 in all three species has drastically reduced the ability of CS to react with OAS, consistent with results of reduction in the cysteine synthesis activities observed above. Since reaction intermediate formation depends on the substrate flux into the channel, it is possible that mutation of M120 compromises substrate supply.
M120 mutation affects OAS binding and recruitment
Binding of OAS to all three isoforms, HiCS, StCS, and MtCS, is accompanied by change in the active site PLP fluorescence. Monitoring the extent of PLP fluorescence change at 507 nm when the protein is excited at 412 nm would allow us to quantify the extent of substrate binding and recruitment (
). As expected, PLP fluorescence at 507 nm decreases after each addition of OAS until all free enzyme molecules are saturated with OAS (Fig. 4, A–C). To find the minimum OAS concentration and equilibration time necessary to saturate fluorescence quenching at a given CS concentration (0.2 μM), we performed concentration and time range exploration experiments. Relative changes of the fluorescence quenching were scanned as a function of OAS concentrations and mixing times (Fig. S5, A–B). We determined that when ∼2.0 μM of OAS is mixed and equilibrated with 0.2 μM of CS/mutants for ∼2 min, the PLP fluorescence quenching saturates with no systematic change beyond this point. As shown in Figure 4, the fluorescence of free enzyme in the absence of OAS is set to 100%, and relative quenching is estimated by adding predetermined levels of OAS (2.0 μM) and recording the fluorescence after ∼2 min.
The relative quenching ratio (Fobs/F0) is calculated as the ratio of absolute fluorescence intensity of wild-type/mutant saturated with OAS (Fobs), normalized to the fluorescence of free wild-type/mutant CS (F0) at the same concentration. The percentage of relative quenching can be determined from the percentage of unquenched fraction reported on top of each bar (Fig. 4, A–C). While three wild-type CS display quenching in the range of ∼75 to 77% (23–26% unquenched), M120A mutants show very low quenching in the range of ∼3 to 12% (>88% unquenched). The ∼75% quenching achieved for wild-type under fixed experimental conditions suggests that the ratio of enzyme–substrate complex to free enzyme is similar for three wild-type CS. Similar to wild-type CS, M96A mutants also show high percentage of quenching in the range of 73 to 94%. However, OAS binding-induced fluorescence quenching of M92A mutants shows species dependency with StM92A showing 62% quenching as compared with 1 to 5% quenching of HiM92A and MtM92A mutants. The percentage of quenching, Fobs/F0, is proportional to the fraction of enzyme·substrate (CS·OAS) complex, and higher quenching percentage corresponds to better ability of that mutant to recruit OAS and form enzyme·substrate (CS·OAS) complex. Consistent with results of single-point activity and steady-state kinetic studies, M120A mutants across the species showed very low quenching (12% or less, ∼90% unquenched) as compared with that of their respective wild-types and other mutants. Similarly, double and triple mutants of HiCS (M92 AM120A, M96 AM120A, and M92 AM96 AM120A) with M120A mutation also showed significantly reduced quenching (Fig. 4C). Both M96A and M92A mutants display species-dependent quenching, but M96A behaves oppositely to M92A. M96A fluorescence quenching is very sensitive to OAS like that of wild-type, whereas the quenching of M92A is insensitive to OAS, like that of M120A mutant. Together, fluorescence quenching results suggest that mutation of M120A almost abolishes the OAS binding (>90%) and M92A may also plays a role in OAS binding although to a lesser extent.
Pre-steady-state kinetics of OAS binding
Results of equilibrium experiments clearly show that M120 is directly involved in OAS binding and recruitment. In order to understand the mechanism by which M120 facilitates OAS recruitment, we used pre-steady-state approaches for monitoring the kinetics of OAS binding (Figs. 5 and 6). Similar to equilibrium approach, a decrease in the intrinsic PLP fluorescence upon OAS addition was continuously monitored for determining the rates of OAS binding to different mutants. Rates measured as a function of OAS concentrations under similar solution conditions were used for estimating on-rate (kon) constants (Table 2). All experiments were performed under pseudo-first-order conditions (excess of OAS). All pre-steady-state kinetic time traces showed single exponential decay that approaches same plateau value corresponding to approximately >95% of quenching of initial PLP fluorescence. The extent of quenching suggests that the kinetic time course reflects the formation of closed-state CS·reaction-intermediate complex. As expected, rates of PLP fluorescence quenching signal increase with increasing OAS concentration. Traces in Figure 5 were fit to single exponential decay model using Equation 6 to obtain values of rates (kobs) at each OAS concentration. Kinetic time traces and fits to traces for wild-type HiCS and three methionine mutants, HiM120A, HiM92A, and HiM96A, are shown (Fig. 5, A–D). Rates of OAS binding to HiM120A and HiM92A are much lower than that of wild-type at any given concentration of OAS.
Table 2Summary of pre-steady-state kinetics data for the OAS binding to the CS
Rapid kinetics (association rate constant) kon (M−1 s−1)
Fold change (WT/Mutant)
(4.0 ± 0.08) × 104
(1.4 ± 0.08) × 102
(1.4 ± 0.08) × 105
(5 ± 0.08) × 102
(2.0 ± 0.08) × 102
(1.0 ± 0.08) × 103
HiM92 AM96 AM120A
(1.9 ± 0.08) × 103
(2.1 ± 0.08) × 104
(1.0 ± 0.08) × 103
(6.8 ± 0.17) × 102
(8.8 ± 0.5) × 104
(2.9 ± 0.05) × 104
NA, data couldn’t be fitted; NP, could not be performed.
Similarly, kinetics of OAS binding by StCS, MtCS, and methionine mutants of these two proteins were also performed under same solution conditions. Observed rates estimated from single exponential fit were plotted as a function of OAS concentration. As shown in (Fig. 6), under pseudo-first-order conditions ([OAS] >> [CS]), the observed rates (kobs) increase linearly with increasing substrate (OAS) concentration. The linear least squares fit to the rates plotted versus OAS concentrations yields the slope (on-rate constant, kon) and the intercept, (off-rate constant, koff) (Equation 6) (Fig. 7). On-rate binding constants of all three wild-type enzymes and that of other ten mutants are shown (Table 2). All three wild-type CS show similar rate of substrate binding with one- to twofold changes as compared with low on-rates of substrate binding displayed by M120A and M92A mutants of HiCS and StCS. Fast kinetics data for MtCS120A could not be fitted as there was no signal change when mixed with OAS. Pre-steady-state data for Mt96A and Mt92A could not be performed due to their high propensity to aggregate at higher concentrations. As seen in the gel filtration purification profile (Fig. S2C), the MtCS and its mutants were purified at lower concentrations. Both mutants aggregate upon concentrating above 15.0 μM.
Upon comparison of reduced on-rates of two M120 mutants, the StM120A binds approximately seven- to eightfold faster rate than the HiM120A, suggesting a species-specific effect of mutation. As shown in Table 2, on-rate binding constants of mutants of HiM120A, HiM92A, and HiM120 AM92A are ∼286-fold, 80-fold, and 200-fold lower than that of HiCS wild-type. The estimated error ranges (∼95% confidence interval) of respective WT CS and mutants do not overlap, suggesting that the observed differences are statistically significant. As expected, on-rates of both HiM96A and StM96A are three- and fivefold higher than that of their respective wild-type CS. StM120A and StM92A mutants show 21- to 31-fold reduction in the substrate recruitment rate. These results clearly demonstrate that the very first step in substrate recruitment is controlled by these noncatalytic surface residues, M120 and M92. Mutation of M120 reduces the recruitment rate by ∼20- to 286-fold, severely affecting the first step of contact between OAS and CS. The negative intercepts of wild-types indicate that kon[OAS] >> koff (Fig. 7). This is real because, the OAS is trapped as α-aminoacrylate within the active site of CS and its dissociation from the active site is very slow. Therefore, M120 plays a crucial role in the very first step of substrate binding.
Role of methionine-trio in inhibitor binding
The active site of CS is also the binding site for C-terminal of SAT, which is the natural binding partner as well as inhibitor of CS. Both equilibrium and knietic approaches provided direct evidence for the role of M120 in recruiting the substrate, and this result also validates our previous structural observation (PDB code: 4ORE) in which M120 makes the contact with the substrate. In two ternary complexes resolved in that study (PDB code: 4ORE and PDB code: 4ZU6), M120–substrate contact at the active site entrance has resulted in subtle but very conspicious conformational changes deep within the active site, ∼17 to 20 Å away from the site of M120 contact with the substrate. For example, curical hydrogen bonds between the last C-terminal ILE residue of the inhibitor and peptide binding loop of CS are broken, weakening the interaction between the enzyme and inhibitor (
). Therefore, we examined the contribution of M120 to the inhibitor affinity even though, M120 does not interact with inhibitor peptide in crystal structure resolved. Since the contact between M120 and substrate disengages the inhibitor, the contribution of M120 in favor of substrate can be estimated from the contribution of M120 to inhibitor binding in the absence of substrate.
Similar to previous studies, we used ten residue SAT C-terminal peptides as the high-affinity inhibitor and kept the enzyme concentration in the range of 0.2 to 0.4 μM (
). The extent of complex formation was quantified from the changes in PLP fluorescence at 507 nm (Fig. 8). Binding isotherms were analyzed using two similar binding sites model and equilibrium dissociation constants Kd (μM) were determined (Table 3). M120A mutation reduced the affinity for inhibitor peptides by 3, 5, and 1.4-fold in HiCS, StCS, and MtCS, although the reduction in MtCS was statistically insignificant. On the contrary, both M92A and M96A background mutations seem to favor the substrate-antagonistic SAT C-terminal peptide binding, although in a species-dependent manner. In HiCS, mutation of M96 increases the affinity by approximately fivefold (3.4 μM versus 0.65 μM), but the affinity increases by 24- to 34-fold in the double and triple mutant backgrounds (HiM96 AM120A and HiM92 AM96 AM120A). As shown in Table 3, only M96A mutation is able to reverse the negative effect of M120A mutation (please compare the affinities of HiM92 AM120A versus HiM96 AM120A). HiM92A exhibits approximately fivefold increase but in the M120A background, M92A mutation fails to reverse the effect of M120A. In StCS, StM92A and StM96A increase the inhibitor binding affinity by 1.5- to 3.0-fold, and MtM92A and MtM96A mutants show two- to fourfold higher affinity toward peptide as compared with the wild-type MtCS. In summary, the effect of M120 mutation on inhibitor peptide binding is not as profound as compared with substrate binding. Nevertheless, M120 contributes to SAT C-terminal peptide binding, by three- to fivefold in HiCS and StCS.
Table 3The equilibrium binding constants Kd of peptide binding for the wild-type CS and mutants
Structural analyses of mutants reveal two different inhibitor binding conformations
To understand how M120 and other two methionine residues influence substrate selection at molecular level, we crystalized and resolved high-resolution structures of four mutants (M92A, M96A, M120A, M120 AM92A) (Table 4). Results show that point mutations do not alter the overall structure of the enzyme, but notable conformational changes observed at the mutated site as well as within the reaction center allow us to explain the effect of mutations on affinities of substrate and inhibitor. In four crystal structures resolved in this study, “TSGNT loop” assumes “pre-inhibitor” binding pose in M92A and M96A structures (PDB codes: 7CM8 and 7C35) and in the M120A and M92 AM120A structures (PDB codes: 5XCN and 5XCW), “TSGNT loop” assumes “post-inhibitor” binding pose. The loop moves in opposite directions (∼5.3 Å between pre- and postbinding conformations) with reference to the position of loop observed in the wild-type HiCS structure. Opposite movement of loop directly correlates with the opposite trends of inhibitor binding affinities of M120 mutants versus M92 and M96 mutants, with prebinding poses favoring CS·inhibitor stability and postbinding pose opposing the peptide binding.
Table 4Crystallographic data collection and refinement statistics
HiCS (PDB ID)
M92 AM120A (5XCW)
Unit cell (A, B, C,) Å
112.44 112.44 44.00
113.01 113.01 44.05
112.27 112.27 46.09
112.55 112.55 46.35
35.56 Å–1.9 Å (1.96–1.9) Å
35.74 Å–2.10 Å (2.17–2.1) Å
28.07 Å–1.69 Å (1.75–1.69) Å
34.09 Å–1.89 Å (1.95–1.89) Å
35.56 Å–1.9 Å
35.74 Å–2.10 Å
28.07 Å–1.69 Å
34.09 Å–1.89 Å
Number of nonhydrogen atoms
Ramachandran favored (%)
Ramachandran allowed (%)
Ramachandran outliers (%)
Rotamer outliers (%)
Statistics for the highest-resolution shell are shown in parentheses.
I/Avg sigma (I): ratio of average intensity and average uncertainty.
I/Avg sigma (I): overall average (I/sigma (I)) of the data set.
Conformations of both α5-β4 loop of active site entrance and substrate/inhibitor binding “TSGNT” loop are altered in the M120A and M92 AM120A structures. In the wild-type CS, side chain of M120 interacts with the main-chain amino group of S70 and main-chain carboxyl group of A69 of “TSGNT” loop at 3.8 Å and 3.7 Å, respectively (Fig. S6, A–B). Similarly, M92 interacts with main-chain carboxylic group of T69 with the distance of 3.5 Å and M96 interacts with main-chain carboxylic group of S70 at a distance of 3.7 Å. All these interactions hold the “TSGNT” loop in the “reference” conformation, that is in between “pre and post” binding conformations (Fig. 9, A–C). Mutation of M120 disrupts the interaction between S70 and A68 of the peptide loop and pushing the loop ∼3.8 Å further away to postbinding conformation. This reduces the chances of incoming inhibitor to latch onto the binding loop as compared with reference conformation present in the wild-type, thus reducing the affinity of inhibitor. In the case of M96 mutation, the interaction between M92 and the carboxylic group of T69 is broken, and the loop moves in the opposite prebinding conformation, more close toward the active site channel. This move increases the chances of inhibitor contacts with the binding loop, therefore leading to increased binding affinities of inhibitor for M96 and M92 mutants. Our study presents the first structural evidence to show that the substrate/inhibitor binding loop may adopt two different conformations.
We employed an integrated approach to provide a detailed view of how CS is able to selectively recruit its substrate, OAS, in the presence of a high-affinity natural inhibitor, SAT. SAT and CS catalyze two consecutive steps of cysteine biosynthesis pathway in plants and bacteria (
). Therefore, it remained elusive until we showed recently that CS employs a novel “competitive-allosteric” mechanism to recruit its substrate even when the active site of CS is bound with SAT C-terminal, referred as inhibitor of CS (
). This study was undertaken to trace and dissect out molecular features that allow CS to recognize and recruit its substrate in the presence of natural inhibitor. To generalize our findings, CS from three different species and ∼13 mutants of these three different versions were analyzed and compared for cysteine synthesis activity, substrate recruitment, inhibitor binding by using a combination of high-resolution approaches.
Single-point enzyme activity, steady-state kinetics, fluorescence quenching, and pre-steady-state kinetic studies clearly show that mutation of M120 reduces the cysteine synthesis activity by significantly compromising substrate recruitment abilities of all three wild-type CS. We employed rapid kinetic binding approaches for comparing forward rate constants of binding with catalytic turnovers determined for mutants. The high-resolution approach allowed us to map the role of M120 to the first phase of substrate recruitment of CS. Results are consistent with earlier structural observation that M120 makes the first contact with the incoming substrate (
). Therefore, this study establishes the role of M120 in selective substrate recruitment unambiguously. M120 contributes to selective substrate recruitment by enhancing affinity for substrate and selectively dissociating the bound inhibitor. Our extensive structural analyses showed that the crucial hydrogen bonds between the active site residues and inhibitor peptide are broken when M120 makes the first contact with the incoming substrate (
). In this study, we quantified the contribution of M120 in favor of substrate by measuring the changes in the on-rate of substrate binding and inhibitor binding. Therefore, the net discriminative effect in favor of substrate can be estimated by combining the contribution from both. In HiCS, M120 favors substrate binding by a factor of ∼858-fold as M120A mutation reduces the rate of substrate binding by 286-fold and peptide binding by threefold. In StCS, the net discriminative force estimated from multiplying on-rate of substrate binding/catalytic turnover and peptide dissociation is ∼100 to 150 fold. Together, our study unravels that M120 acts as a selectivity filter against the inhibitor with net discriminative force of 100- to 858-fold in favor of substrate. Since the first step of substrate recruitment is affected, substrate flux into the ∼20 Å channel is significantly reduced, resulting in manifold reduction in reaction rate and product formation. Mutation of M92 also affected the substrate recruitment although to a lower extent. M120 emerges as the primary and the most important residue in substrate recruitment by CS.
High-resolution crystal structures of four mutants provide insights into how M120 modulates inhibitor affinity by “engagement and disengagement” mechanism in the absence and presence of substrate. In the absence of substrate, M120 interaction with inhibitor/substrate binding “TSGNT” loop stabilizes the CS·Inhibitor complex, but M120 disengages the “TSGNT” loop as soon as it senses the substrate at the active site entrance. The “engagement and disengagement” switch is a gate-like allosteric mechanism that empowers M120 to facilitate the dissociation of inhibitor from the active site when it engages with substrate at the entrance. The equilibrium constants of inhibitor binding to M120 and other mutants allowed us to estimate the contribution of M120 to inhibitor in the absence of substrate. The three- to fivefold reduction in the affinity of inhibitor is an indirect measure of contribution of M120 to substrate recruitment when the substrate makes contacts with M120, but if we compare the contribution of all M120 and M92. Results presented here describe the first systematic study to explore the features of substrate recruitment of CS in the presence of natural high-affinity inhibitor, SAT. Results obtained from multiple independent and orthogonal approaches map the role of noncatalytic residue M120 in facilitating the substrate recruitment. Significantly reduced rate of substrate binding may suggest that substrate binding has become the rate-limiting step in the M120A mutants. Formation of α-aminoacrylate is characterized as the reaction intermediate in CS. However, to conclude substrate binding as the rate-limiting step, rate of reaction of substrate with PLP, rate of conformational transition to closed state, rate of intermediate formation, and rate of product release of all mutants need to be determined and compared with rates of substrate binding determined in this study. Together, our results estimate that CS use M120 to discriminate in favor of substrate at least by ∼100- to 858-fold by employing the allosteric engage (substrate) and disengage (inhibitor) mechanism.
A detailed understanding of both catalytic and substrate selection mechanisms has many applications, ranging from tuning the substrate selectivity of enzymes toward the desired “substrate-product conversion” to identify critical steps in the whole catalytic cycle to design target-specific therapeutic molecules. Our data provides a framework for identifying key features of substrate selectivity, which enable CS to discriminate the high-affinity natural inhibitor protein against its substrate. CS and SAT associate to form a high-affinity CRC complex (
). However, selective targeting remains a major challenge to drug discovery groups due to the paucity of information about details of complex formation and dissociation. Enzymes of cysteine biosynthesis pathway, including CS (CysK), have been considered as potential drug targets (
). Results presented open prospects for designing selective inhibitors that bind through competitive-allosteric mechanism by exploiting M120-based selective recruitment mechanism.
A total number of 18-homologous protein sequences of CS were taken for the multiple sequence alignment from a wide range of bacteria and plants from the NCBI database. Redundancy was eliminated by taking only one protein sequence from one genus. The alignment was performed using CLUSTAL-W (
Mutations at M120, M92, and M96 were introduced into the wild-type sequences of CS from three different microorganisms (M. tuberculosis, Haemophilus influenzea, S. typhimurium) using quick-change site-directed mutagenesis protocol (Agilent technologies, Inc). The primers containing the mutations were synthesized by IDT, Inc, USA. Mutations in the CS genes were confirmed by DNA sequencing. We generated 13 mutants (variants) of CS enzymes from three different bacteria (H. influenzae, S. typhimurium, and M. tuberculosis). We created 13 mutants, either as point mutants or as combination mutants (double and triple mutants). For H. influenzae CS, we generated approximately seven mutants; HiM92A, HiM96A, HiM120A (three single mutants), HiM92 AM96A, HiM92 AM120, HiM96 AM120 (three double mutants, DM), and HiM92 AM96 AM120A (one triple mutant, TM). For S. typhimurium CS, three single mutants StM92A, StM96A, and StM120A were generated. Similarly, for M. tuberculosis CS, three single mutants MtM92A, MtM96A, and MtM120A were created.
Protein expression and purification
Coding frames of CS, from S. typhimurium (StCS) strain LT2, H. influenzae (HiCS), and M. tuberculosis (MtCS), were cloned into N-terminal 6His-pET28a+ expression vector. Similarly, wild-type and mutant constructs were expressed in Escherichia coli BL21DE3 cells and purified. The N-terminal His-tag of all enzymes was removed by thrombin digestion and further purified by size-exclusion chromatography. Purified fractions were analyzed on a 12% SDS-PAGE gel and found to be >95% pure. We determined protein concentrations using molar extinction coefficients of HiCS, StCS, MtCS (21,555 M−1 cm−1, 19,940 M−1 cm−1, 11,500 M−1 cm−1 respectively) estimated at 280 nm. The enzyme assay for three wild-types and all mutants was carried out using the acid ninhydrin assay for cysteine formation as described (
Secondary structure analyses by circular dichroism
CD measurements were carried out with a JASCO spectropolarimeter (Jasco, Tokyo, Japan) equipped with a Peltier-type temperature controller (PTC-348W). Far-UV spectra were obtained in a quartz cuvette with a 10 mm light path length, and each spectrum obtained was an average of seven scans. The ellipticities of protein CD spectra are subtracted from reference buffer spectrum and reported as mean residue ellipticity (MRE) in degcm2/dmol units.
Single-point activity and steady-state kinetics of enzymes
The single-point activity and detailed steady-state kinetics assays were carried out to assess the effect of mutations. Both assays were performed using the acid ninhydrin method for quantifying the amount of cysteine (
). The substrate, O-acetylserine (OAS, manufactured by Sigma) was dissolved in 0.1 M HEPES pH 7.0. CS hydrolyzes OAS in the presence of Na2S and synthesizes cysteine. Formation of cysteine is monitored at 560 nm (extinction coefficient of cysteine is 28,000 M−1 cm−1). Both single-point activity and steady-state kinetics assays were performed at 30 °C in 0.1 M HEPES, at pH 7.0 in a volume of 150 μl.
For single-point activity assay, OAS and enzyme concentrations were fixed to 5 mM and 100 ng, respectively. The CS-to-substrate ratio is in the substrate saturating range, and therefore, CS is expected to catalyze with maximal velocity. Steady-state kinetics experiments were performed in triplicates with enzyme concentration fixed at 100 ng and OAS concentration varied from 0.1 mM to 8.0 mM. The Na2S concentration was fixed at 3 mM in both assays. Substrates were added to buffer and mixed before the reaction was initiated by the addition of CS/mutants. The reaction was allowed to proceed at 30 °C for 20 min. The reaction was terminated by addition of 5% TCA, centrifuged at 13,000 rpm, and 125 μl of supernatant was transferred to a new tube. To this tube 125 μl of glacial acetic acid and 125 μl of acid ninhydrin reagent (250 mg of ninhydrin dissolved in 2 ml of conc. HCl and 3 ml of glacial acetic acid) were added. After mixing, samples were boiled for 10 min at 99 °C in preheated water bath. Samples were cooled to room temperature and diluted with 625 μl of chilled 95% ethanol. Absorbance was recorded at 560 nm, and amount of cysteine produced was calculated from standard curve, which was estimated for various concentrations of cysteine. Initial velocities determined from triplicate steady-state kinetics experiments were fit to Michaelis–Menten model (Equation 1) using nonlinear least squares method.
where v is velocity at given substrate concentration, [S] is substrate concentration, Vmax, is maximum velocity at [E]T << [S]T KM, is apparent substrate affinity. The turnover rate, kcat, is calculated by normalizing Vmax by [E]T, total enzyme concentration. Errors of kinetic parameters were estimated and reported with 95% confidence intervals (1.98∗standard error). To calculate, the catalytic efficiency (kcat/KM) of M92A and M120A mutants, we divided kcat value of each mutant by average of KM, calculated from KM of single and double mutants of M92A and M120A. For wild-type and M96A, kcat of each mutant was normalized with corresponding KM of that enzyme. Errors for kcat/KM were calculated by propagating errors associated with kcat and KM using Equation 2.
where Δkcat/KM, error of catalytic efficiency, is expressed as square root of sum of squares of relative errors of kcat and KM. and are errors estimated with 95% confidence intervals. Further, we compared the significance of differences of catalytic turnover between wild-type and mutants within each group by estimating p-values (0.05).
Absorption spectroscopic analyses of α-aminoacrylate formation
UV-visible spectra of HiCS, StCS, MtCS, and their mutants in complex with the OAS were recorded with Agilent cary-win UV spectrophotometer for estimation of α-aminoacrylate reaction intermediate formation. The spectra were recorded in the buffer with 50 mM Tris-Cl (pH 7.5) 100 mM NaCl, at 25 °C, and 5% glycerol. Concentrations of HiCS and its mutants used were ∼ 8 to 15 μM and for StCS/MtCS and its mutants were ∼ 8 to 25.0 μM. The spectra of the covalent α-amino acrylate complex were obtained after the addition of 100 μM OAS to the enzyme solution. Each enzyme was added to the above buffer, mixed, and OAS dissolved in the same buffer was added and mixed to the final volume of ∼800 μl. PLP absorbs at 412 nm. Addition of OAS to CS results in the formation of intermediates with new absorption spectra at 321 nm and 470 nm.
Equilibrium measurements of OAS and peptide binding to CS and mutants
Both OAS and SAT-C10 peptide binding to CS and its mutants were examined by monitoring fluorescence changes of the active site PLP. The excitation wavelength was set at 412 nm and fluorescence was monitored at 507 nm. All experiments were done in triplicates and at 23.0 °C ± 1 deg. C with excitation and emission bandpass set to 5.0 nm. For OAS experiments, the enzymes concentration was fixed at 0.2 μM and for peptide binding, enzyme concentration was kept between 0.5 and 1.0 μM, and enzyme solution was mixed after each addition of ligands (OAS/peptides). For OAS quenching experiments, the (F0) fluorescence of free enzyme is taken as the reference state, and (F) fluorescence of enzymes is taken in the presence of OAS. Initial fluorescence data of wild-type and mutants at a fixed enzyme concentration (0.2 μM) are shown in absolute fluorescence intensity units (Fig. S8). We performed exploratory experiments to determine the saturating OAS concentration and time required for saturation. Briefly, using HiCS as the reference, we monitored the quenching of PLP fluorescence as a function of OAS concentration after incubating for 15 min. Upon estimating the OAS concentration to achieve the maximum quenching, we performed another set of scouting experiments to determine the minimal equilibration time. We fixed CS and OAS concentrations (0.2 μM and 2.0 μM) and monitored the PLP signal as a function of time and found that PLP quenching saturates at 2 to 3 min of incubation. Fluorescence data at each titration point represents the average of 90 readings (30 for each experiment performed in triplicates), and associated errors were estimated with 95% confidence intervals. We performed all OAS binding experiments for both wild-type and mutants by keeping buffer conditions, concentrations, and incubation time fixed. The ratio of F/F0 is plotted versus enzyme type, and percentage of quenching is estimated from 1 − F/F0 for each enzyme type. Mean values of each measurement were shown with errors estimated at 95% confidence intervals.
In the case of peptide binding experiments, we used stocks of peptide concentration in the range of 0.1 to 0.8 mM. After each addition, the reaction mixture was equilibrated for 2 to 3 min before recording PLP fluorescence, and data points from five such measurements were averaged to obtain Fave,i. The relative fluorescence quenching upon ligand binding is defined as Fobs,i = (Fave,I − Fo)/Fo. Inhibitor–CS complex formation was analyzed to obtain the equilibrium binding constant, Kobs = [PL]/[P]∗[L], using two independent site-binding models (Equation 3). The fit parameter Kobs was determined with 95% confidence interval.
Where n is 2, Fobs is observed fluorescence quenching, and Fmax is the maximum fluorescence quenching at saturation. Equilibrium dissociation constant (Kd) was obtained from Kobs by taking the inverse Kd = 1/Kobs. Errors were propagated using Equation 4
where Δ Kd is error of equilibrium dissociation constant, ΔKobs is error of Kobs.
Crystallization, data collection, and structural determination
Mutant proteins were crystallized using sitting drop by vapor diffusion method. A total of 1.0 μl drop containing 10 to 12 mg/ml protein mixed with 1.3 M sodium citrate and 100 mM HEPES pH 7.5 was used for the crystallization. Good-quality crystals were obtained within a week time by incubating at 20 °C. All the mutant crystals were obtained at similar conditions. X-ray data was collected at home source on Rigaku micro focus HF beam equipped with MAR345dtb image plate detector. The data was reduced with HKL2000 software package (
Pre-steady-state kinetics experiments were performed with Biologic rapid kinetics instrument (SFM400, Biologics, France) equipped with four syringes (10 ml) set in a parallel fashion. Fluorescence data was collected by MOS-250 unit equipped with PMT450 (detector) fitted with long-pass filter (450 nm) (Semrock Inc). All the experiments were done in 20 mM Tris pH 7.5, 20 mM NaCl as the running buffer in the flow lines. All proteins and OAS stocks were dissolved in the same buffer. Samples were excited at 412 nm using slit width of 4 nm, and emission was collected after the passage through emission long-pass filter (450 nm). After initial rapid mixing, time course internal fluorescence intensity data was recorded from the PLP. Protein concentration was used in the range of ∼15 to 30 μM, and OAS concentration was used in the range of (10.0–60.00 μM). However, Mt92A and Mt96A mutants were excluded due to their high aggregation propensity at higher concentrations. We observed that these two mutants exist as stable dimers below <10.0 μM and show time-dependent aggregation at higher concentrations. Fast kinetic data was collected at millisecond timescale (∼2000 points/s) between 0 and 1 s to capture all fast events that occur before steady state is reached. After 1 s, data is collected at 15 points/s as the late events are generally slow. Four to five traces were averaged and data were analyzed using Biokine analysis program, provided by the manufacturer. The rate of binding (kobs) is obtained by fitting data to single exponential model (Equation 5). The errors estimated for the rates represent two standard deviations with 95% confidence interval.
F is the fluorescence at time t, n is the number of exponential terms, A and kobs, are the amplitude and the observed rates, respectively, and Ao is the fluorescence intensity at t = 0. kobs values obtained from the fit were plotted against the concentration of ligand. The on-rate (kon) and off-rate constants were estimated from slope and intercept of that plot (Equation 6)
All data are included within the article. Coordinates and structure factor amplitudes have been deposited into the Protein Data Bank (http://wwpdb.org) under accession numbers: 5XCP; 7CM8; 5XCN; 5XCW; 7C35. Raw data or further information is available upon request from the corresponding author.
Conflict of interest
The authors declare that they have no conflict of interest with the contents of this article.
The authors thank Dr R. P. Roy, Scientist VII, National Institute of Immunology for editing and providing valuable suggestions.
S. K .designed all the experiments and wrote the manuscript. A. K., R. R., N. S., and R. P. S. conducted the experiments and analyzed the data. R. K. and S. Koul generated the mutants used in this study.
Funding and additional information
This research was supported by CSIR India, in part by Department of Biotechnology, India. A. K., R. P. S., and N. S. received CSIR SRF fellowships, and R. R. was a DBT-SRF fellow.
Cys-Gly specific dipeptidase Dug1p from S. cerevisiae binds promiscuously to di-, tri-, and tetra-peptides: peptide-protein interaction, homology modeling, and activity studies reveal a latent promiscuity in substrate recognition.
Structural insights into catalysis and inhibition of O-acetylserine sulfhydrylase from Mycobacterium tuberculosis. Crystal structures of the enzyme alpha-aminoacrylate intermediate and an enzyme-inhibitor complex.