Incorporation of a single His residue by rational design enables thiol-ester hydrolysis by human glutathione transferase A1-1

Materials and Methods

All UV spectroscopic measurements, with one exception, were carried out by using a Cary-100 UV-visible spectrophotometer and submicrocells (Varian). The pK_a of Y9 was determined by using a UV spectrophotometer (Shimadzu). HPLC experiments were performed by using a ProStar 230 delivery system and a ProStar 330 photodiode detector (Varian). HPLC gradients of acetonitrile and 0.1% trifluoroacetic acid in water and C-18 reversed-phase columns (Hypersil ODS, Supelco) were used to detect and quantify GSB and benzoic acid (kinetics, 22→46%), GSB and methylbenzoate (MeOH trapping, 22→82%), and N-acetyl GSB, N-acetyl GSH, and benzoic acid (background reaction, 22→70%) with p-nitrophenol as an internal standard.

Site-Directed Mutagenesis and Protein Expression. GST A1-1 mutants were prepared by using inverted PCR with Pfu DNA polymerase (Promega), custom-synthesized oligonucleotide primers (DNA Technology, Aarhus, Denmark), and the template pGNdeA1 (25). Mutants were identified by sequence analysis, and the coding sequence was subcloned into the expression vector pET-21a(+) (Novagen) by using the NdeI and SalI restriction sites and expressed in Escherichia coli BL21(DE3) (Novagen). The recombinant proteins were purified on a HiTrap SP cation-exchange column (Amersham Biosciences) as described (26), and purity was confirmed by SDS/PAGE (27). Protein concentrations were determined spectrophotometrically by using ε₂₈₀ = 24,700 M^–1·cm^–1 (26), and the mutants were checked for activity with a standard assay (28) by using 1 mM 1-chloro-2,4-dinitrobenzene/1 mM GSH in 100 mM sodium phosphate (pH 6.5) at 298 or 303 K.

Synthesis of GSB. GSH was prepared by solid-phase peptide synthesis using standard 9-fluorenylmethoxycarbonyl (Fmoc) protocols and Fmoc-Gly-Wang resin (0.7-mmol scale) as described (29). Cys was orthogonally protected with a 4-methoxytrityl group, and the thiol was selectively deprotected with 1% trifluoroacetic acid in dichloromethane to enable thiol-ester coupling with benzoic acid. The coupling was performed in dimethylformamide by using benzoic acid/1-hydroxybenzotriazole/N,N′-diisopropylcarbodiimide/diisopropylethylamine (2:3:3:6 dilution). Global deprotection and cleavage from the solid support was enabled by trifluoroacetic acid/H₂O (97.5:2.5 dilution), followed by precipitation of the peptide with diethyl ether. GSB was purified by reverse-phase HPLC (29). GSB concentrations were determined spectrophotometrically by using an extinction coefficient of ε₂₆₆ = 7,889 M^–1·cm^–1.

Kinetic Measurements. Kinetic measurements with the exception of measurements used to determine saturation-kinetics parameters were performed at protein concentrations of 5 or 25 μM and a GSB concentration of 75 μM in 100 mM sodium phosphate at pH 7 and 298 K. Saturation kinetics were run with a concentration of A216H of 5 μM and GSB concentrations in the range of 5–80 μM. The results were analyzed by using the Michaelis–Menten equation in the form in which no assumption has been made that [E] « [S]. The buffers used to record the pH profile were 100 mM sodium acetate (pH 5–5.6) and 100 mM sodium phosphate (pH 5.7–8). Reaction rates were determined by monitoring the concentration of GSB spectroscopically at 266 nm. In the trapping of the acyl intermediate, various concentrations of MeOH (0–1 M) were mixed with A216H or wild type (final concentration 5 μM) in 100 mM sodium phosphate (pH 7), and GSB was added to a final concentration of 75 μM. The reactions were followed spectrophotometrically at 266 nm and 298 K. The reaction product methylbenzoate was identified by reverse-phase HPLC. Typical measurements for the saturation-kinetics plot and the pH-dependence determination were carried out in duplicate, with maximum deviations from the mean value of <10%. The rate constants in the presence of MeOH were estimated from single measurements only. Liberated GSH eventually reacts with a solvent exposed Cys residue to form a GSH–GST conjugate by disulfide formation. Such modifications have been shown not to affect the catalytic activity of human GST A1-1 (30).

Product Identification. The benzoic acid produced in the hydrolysis reaction was identified under turnover conditions by reverse-phase HPLC and by NMR spectroscopy (Inova-600, Varian). The NMR experiments were carried out at 600 MHz and 298 K by using 10% D₂O in 90% H₂O as the solvent. The chemical shifts of the ring protons of benzoic acid (7.47, 7.54, and 7.86 ppm) and of those of GSB (7.56, 7.71, and 7.99 ppm) were determined in independent experiments.

Background Hydrolysis. N-acylated GSB was synthesized by mixing 75 μM GSB and excess N-hydroxysuccinimide acetate in 100 mM sodium phosphate (pH 7). The mixture was incubated at 298 K, and aliquots were collected for 14 days. The concentrations of remaining substrate and hydrolysis products were analyzed by reverse-phase HPLC.

The Determination of the pK_a of Y9. The pK_a of Y9 in the A216H and F220H mutants was determined by difference UV spectroscopy at 298 K using buffers in the range of pH 5–9 as described (31). The protein concentration was 15 μM, and spectra were recorded at 220–360 nm. The spectrum at pH 5.5 was subtracted from the other spectra, and the pK_a of Y9 was calculated by plotting the values of the peak arising at 250 nm against pH.

Digestion Experiments. GST variants from reaction mixtures were digested by trypsin or by Staphylococcus aureus protease V8 (Pierce). We digested 20 μM wild type or A216H mixed with 40 μM GSB by trypsin for 3 h in 100 mM NH₄HCO₃ (pH 8) at ambient temperature with a protein/trypsin ratio of 100:1. In the case of protease V8, 10 μM wild type or A216H was mixed with 20 μM GSB and digested in 50 mM NH₄CH₃COO (pH 4) at 310 K for 3 h with a protein/protease ratio of 40:1. The digests were desalted with ZipTips (C-18, Millipore) and analyzed by matrix-assisted laser desorption ionization–time of flight MS (Voyager 4212, Applied Biosystems) by using α-cyano-4-hydroxycinnamic acid matrix with detection in the positive mode.

Results and Discussion

The Single Mutant A216H Is an Enzyme. The GST A1-1 mutant A216H catalyzed the hydrolysis of GSB (5 μM A216H/75 μM GSB/100 mM sodium phosphate, pH 7, 298 K) to form GSH and benzoic acid, as demonstrated by HPLC and NMR spectroscopy under turnover conditions. The transformation of GSB into benzoic acid and GSH was readily followed by monitoring the gradual appearance of the resonances of the aromatic-ring protons of benzoic acid and the gradual disappearance of the ring protons of GSB, as well as by HPLC. The reaction followed saturation kinetics with a k_cat/K_M of 156 M^–1·min^–1 (k_cat = 7.8 × 10^–4·min^–1, K_M = 5 μM), Fig. 2. In contrast, the wild-type protein did not catalyze the hydrolysis of GSB but consumed one equivalent of GSB, without release of benzoic acid, as demonstrated by HPLC under reaction conditions. The incorporation of a single amino acid residue by rational design had, thus, converted the wild-type protein into a turnover system for thiol-ester hydrolysis.

The first-order rate constant of the uncatalyzed reaction was 1.1 × 10^–5·min^–1 (100 mM phosphate, pH 7, 298 K) and the observed catalytic proficiency (32) (k_cat/K_M)/k_uncat of the A216H-catalyzed reaction was, thus, >10⁷ M^–1. A catalytic efficiency k_cat/k_uncat of almost two orders of magnitude, in addition to the proximity effect provided by the enzyme substrate binding, accounted for the observed catalysis. The mutant A216H provides a rare example of a successful incorporation of a chemical function based on the common amino acid residues into a protein scaffold, resulting in a new enzyme with nonnatural activity. Because of the competing intramolecular acyl transfer reaction of GSB to form N-acylated GSH, N-acetylated GSB was used as the substrate to determine the rate of the uncatalyzed background reaction, but N-acylation was not expected to affect the rate of thiol-ester hydrolysis. Although evolutionary pressure has made it unlikely that single-point mutations can lead to enhanced efficiencies in natural enzymes catalyzing their natural reactions, these results show that single mutations can lead to altered chemical functions, suggesting that searching for new chemistries, rather than new specificities, may be a viable route to new enzymes.

The Benzoate Ester Formed at the Side Chain of Y9 Is an Intermediate on the Reaction Pathway. The fate of the benzoyl group in the wild-type protein was determined by matrix-assisted laser desorption ionization–time of flight MS by subjecting the reaction mixture to cleavage by trypsin or S. aureus protease V8 and comparing the molecular masses of the peptide fragments to those obtained in the absence of substrate. The fragment containing Y9 (residues 7–13 for trypsin) was readily observed in the wild-type enzyme and was acylated to a large degree by benzoic acid under reaction conditions (29). Under the same conditions, A216H showed only minor acylation at the side chain of Y9. It appeared that, in the A216H-catalyzed reaction, the acylated form of Y9 was an intermediate that reacted further to form a reaction product. As a by-product, the released GSH was found by matrix-assisted laser desorption ionization–time of flight MS to form a disulfide bridge with C112, the only Cys residue in the GST A1-1 protein scaffold. Conjugation with GSH does not affect the enzymatic activity of human GSH transferase A1-1 (30).

The Mechanism of Hydrolysis of GSB. Key residues for thiol-esterase activity were identified by site-directed mutagenesis. F220H showed almost identical behavior to that of the wild type and consumed one equivalent of substrate but did not turnover. The double mutant A216H/F220H had a catalytic efficiency that was approximately equal to that of A216H and demonstrated turnover kinetics, suggesting that H216 was the critical residue in both mutants. We concluded that H216 is essential for catalyzing the hydrolysis of the acyl intermediate and that H220 is inactive. The mutants Y9H and Y9F/A216H/F220H were inactive and remained unacylated, showing that Y9 also was essential for the catalytic function. Therefore, the roles of H216 and Y9 were further elucidated by determining their dependence on pH. Attempts to determine the pK_a values of H216 by NMR spectroscopy failed because of the lack of spectral assignments. The pK_a value of Y9 in the wild type (8.2 ± 0.1) had been determined in ref. 31, and the pK_a values of Y9 in A216H (7.8 ± 0.3) and F220H (7.6 ± 0.2) were determined by UV spectroscopy in a similar way and were found to be significantly lower than the pK_a of Tyr residues in small peptides (9.7) (33). The pH dependence of k_cat was determined for the A216H-catalyzed reaction (Fig. 3), and it suggested that the catalysis depended on a residue in its unprotonated form with a pK_a of ≈6. Clearly, this residue could not be Y9, but the observed pH dependence was instead in agreement with the expected pK_a value of a His residue, although the identity of the catalytically active residue responsible for the observed pH profile has not been unequivocally established. The data suggest that the His residue is involved in the rate-limiting step.

Further mechanistic information about the His residue and the identity of the rate-limiting step was obtained from experiments that were designed to trap the reaction intermediate by MeOH (Fig. 4). The rate of the hydrolysis reaction catalyzed by A216H was increased by the addition of MeOH under conditions in which the structure of the enzyme was not adversely affected, as evidenced by tests of activity toward the standard GST substrate 1-chloro-2,4-dinitrobenzene. Methylbenzoate, the reaction product obtained upon MeOH trapping, was identified by HPLC in the A216H-catalyzed reaction, whereas no methylbenzoate was observed with the wild-type enzyme. Therefore, the acyl intermediate is on the reaction pathway and the breakdown is the rate-limiting step.

The most likely mechanism of the A216H-catalyzed reaction enabled by the introduction of H216 included the formation of a Michaelis complex between GSB and the enzyme, acylation of Y9, and hydrolysis of the acyl intermediate. In the wild-type reaction, Y9 was terminally acylated because the leaving group was not reactive enough to ensure turnover by spontaneous hydrolysis. In contrast, in the A216H-catalyzed reaction, the His residue catalyzed the deacylation of Y9 to form benzoic acid with regeneration of the free enzyme. The pH dependence excluded the possibility that His is a general-acid catalyst, but whether it catalyzes the breakdown of the intermediate by nucleophilic or general-base catalysis could not be determined based on the kinetic data. A crystal structure was determined of the complex between A216H and S-benzylglutathione to 2.1 Å resolution (see supporting information, which is published on the PNAS web site), to estimate whether the His residue was close enough to the carbonyl carbon of the benzoyl residue of the acyl intermediate to form a covalent bond, or whether a water molecule would have to be inserted in between the His nitrogen atom and the carbonyl carbon (Fig. 5). The observed distance in the complex (7 Å between the Tyr oxygen and the His nitrogen, Nε) suggests general-base catalysis, but because it is not a representation of the structure of the true enzyme–substrate complex, it does not constitute rigorous proof. Nucleophilic catalysis by His residues is uncommon in native enzymes, probably because the efficiency in nucleophilic catalysis decreases with the difference in pK_a between the leaving group and the nucleophile, and realistic substrates have poorly reactive leaving groups in comparison with His. However, in the present case, the pK_a of Y9 has been reduced in the binding pocket to a level at which it most likely is comparable with that of H216, and therefore, Y9 can, in principle, be efficiently expelled by H216. Nucleophilic catalysis might also be compatible with the observation that MeOH enhances the reaction rate but only if the His benzoate is higher in free energy than the Tyr benzoate, which is improbable because the pK_a of H216 is most likely <7.8 (the pK_a of Y9). Although no definitive proof has been obtained, the available evidence supports a reaction mechanism in which H216 acts as a general-base catalyst. It is noteworthy that the attack by MeOH appears also to be catalyzed by general base because no methylbenzoate is formed in the reaction with the wild-type enzyme. The crystal structure reveals further that the distance between the phenolic oxygen of Y9 and the ε-nitrogen of H216 is too large (≈7 Å) to permit general-acid catalysis. GST A1-1 from the rat has been shown to be capable of catalyzing the hydrolysis of glutathionyl ethacrynate but by a different reaction mechanism from the one introduced here (34). There was no detectable acylation of Y9, although it was required for catalysis, and it was concluded that the most likely function of Y9 in rat GST was to act as a general acid or general base in the hydrolysis of the bound substrate. The GSH thiol ester of ethacrynic acid is hydrolyzed also by other GSH transferase isoforms (35), including human GSH transferase A1-1. In some of these enzymes, Tyr is not required for catalysis and in contrast to what is the case for A216H, Y9 does not appear to function as a nucleophilic catalyst in any of them.

Conclusion

Thiol-ester hydrolytic activity was achieved by the introduction of a single His residue in the active site of GST A1-1 that enabled a new reaction pathway from the irreversibly acylated Y9. It is likely that dead-end reaction pathways similar to that of the wild-type GST A 1–1 exist in other natural proteins and can be accessed by minimal modifications to the protein scaffold to engineer enzymes with nonnatural functions. The limited invasion of the folded protein is expected to minimize the risk for structural rearrangements and to increase the probability that rational incorporation of new catalytic functions will be successful. The affinity of GST A1-1 for GSH, the natural substrate of GST A1-1, was exploited in binding GSB in a position in which Y9 could attack the carbonyl carbon to form the acyl intermediate that was subsequently hydrolyzed by H216. The covalent binding of the acyl intermediate in a promiscuous enzyme such as GST suggests that various nucleophiles can be used to trap a multiplicity of acyl groups and form a catalytic machinery for alternative chemical transformations.

Supplementary Material