The Mechanism of Mycobacterium tuberculosis Alkylhydroperoxidase AhpD as Defined by Mutagenesis, Crystallography, and Kinetics*

AhpD, a protein with two cysteine residues, is required for physiological reduction of the Mycobacterium tuberculosis alkylhydroperoxidase AhpC. AhpD also has an alkylhydroperoxidase activity of its own. The AhpC/ AhpD system provides critical antioxidant protection, particularly in the absence of the catalase-peroxidase KatG, which is suppressed in most isoniazid-resistant strains. Based on the crystal structure, we proposed recently a catalytic mechanism for AhpD involving a proton relay in which the Glu 118 carboxylate group, via His 137 and a water molecule, deprotonates the catalytic residue Cys 133 (Nunn, C. M., Djordjevic, S., Hillas, P. J., Nishida, C., and Ortiz de Montellano, P. R. (2002) J. Biol. Chem. 277, 20033–20040). A possible role for His 132 in subsequent formation of the Cys 133 -Cys 130 disulfide bond was also noted. To test this proposed mechanism, we have expressed the H137F, H137Q, H132F, H132Q, E118F, E118Q, C133S, and C130S mutants of AhpD, determined the crystal structures of the H137F and H132Q mutants, estimated the p

Mycobacterium tuberculosis infects an estimated 8 million per year and kills 2-3 million people in the same time span (1). It is estimated that ϳ2 billion people have been exposed to this lethal pathogenic organism and thus are at risk of developing the active disease. The majority of infected individuals reside in the third world, but the rates of infection in other areas that are undergoing rapid social change, such as the Soviet Union, are increasing at an alarming rate (2). Furthermore, partially as a result of the symbiotic relationship between human immunodeficiency virus and tuberculosis, the incidence of multidrug-resistant tuberculosis is rapidly increasing (3). The resurgence of tuberculosis as a world-wide phenomenon, in conjunction with the recent determination of the M. tuberculosis genome, has fuelled a renewed search for agents that are active against drug-resistant strains, completely sterilize the infection, and/or shorten the duration of drug therapy and thus promote drug compliance.
Mutations in the M. tuberculosis catalase-peroxidase KatG result in resistance to the prodrug isoniazid, because KatG is required to oxidize this drug to its biologically active form (4 -9). In a similar manner, mutations in the flavoprotein monooxygenase EtaA result in resistance to ethionamide, because it is also a prodrug that must be oxidized by EtaA to its active form (10,11). Interestingly, even though the activating enzymes differ, the mutations of KatG and EtaA that cause resistance to isoniazid and ethionamide, respectively, result in elevated expression of AhpC (4,(12)(13)(14)(15)17). AhpC is thought to provide protection against the oxidative stress associated with both of these mutations, particularly against loss of the KatG peroxidase activity (12). AhpC is a member of the ubiquitous peroxiredoxin family. In the peroxiredoxins, a cysteine residue reacts with a peroxide or other oxidant to give a sulfenic acid (-SOH) intermediate (19,20). This sulfenic acid is then converted to a disulfide bond by intramolecular or intermolecular reaction with a second sulfhydryl group. Finally, the disulfide bond is reduced to regenerate the cysteine thiol groups. The disulfide bond is reduced by different mechanisms in different organisms; in yeast, the reduction is mediated by thioredoxin and thioredoxin reductase (21) and in Salmonella typhimurium by a flavoprotein known as AphF (22). The M. tuberculosis thioredoxin and thioredoxin reductase, however, do not reduce the corresponding AhpC (23), and no homologue of AhpF is detected in the genome of this organism by BLAST searches. However, a gene (Rv2429) coding for AhpD, a protein with no sequence identity to AhpC or AhpF, is located immediately adjacent to the AhpC gene. Recent work has shown that AhpD functions as the reducing partner for AhpC. AhpD itself is reduced by a novel system consisting of dihydrolipoamide succinyltransferase (SucB), a lipoamide-containing protein, and dihydrolipoamide dehydrogenase (25). SucB can be replaced in this system by dihydrolipoamide itself. Interestingly, AhpD, in addition to being the reducing partner for AhpC, has independent alkylhydroperoxidase activity of its own when AhpF from S. typhimurium is used as a surrogate reducing partner (24).
The crystal structure of AhpD was independently determined by two laboratories (25,26). AhpD is a homotrimeric protein in which the individual subunits consist of 177 amino acids with a molecular mass of 18,781 Da (25,26). Two cysteines, Cys 130 and Cys 133 , are present in the AhpD sequence (numbering based on inclusion of the initial methionine) in a Cys-His-Ser-Cys motif within a novel protein fold. Site-specific mutagenesis of the cysteines has shown that both cysteines, but particularly Cys 133 , are critical for the alkylhydroperoxidase activity supported by the S. typhimurium AhpF (24,25).
Despite the absence of structural homology to other proteins, analysis of the detailed structure of AhpD identified several residues that could be involved in the catalytic activity of the enzyme (26) based on local structural analogy to the active site of a functional analogue, thioredoxin. Specifically, the carboxylate anion of Glu 118 is 2.6 Å from the nitrogen of His 137 and is hydrogen-bonded to it (Fig. 1). His 137 is hydrogen-bonded to a water molecule 2.5 Å away that, in turn, interacts with Cys 133 at a distance of 3.3 Å. These hydrogen bonding interactions provide a reasonable mechanism for deprotonation of Cys 133 , the residue thought to react with the O-O bond of peroxides or the disulfide bond present in the oxidized form of AhpC. The location of His 132 4.8 Å from Cys 130 and 3.7 Å from Cys 133 in the reduced enzyme suggested that it might also play a role in the catalytic mechanism. These mechanistic proposals have been tested in the present study by mutating the residues, determining the crystal structures of two of the mutants, and determining the effects of the mutations on the catalytic activities in both the AhpF and AhpD/AhpC/lipoamide/dihydrolipoamide dehydrogenase assay systems.

EXPERIMENTAL PROCEDURES
Materials-All chemical reagents were purchased from Sigma. Escherichia coli strain BL21(DE3) was from Novagen, and strain DH5␣ was from Invitrogen. Q-Sepharose Fast Flow was purchased from Amersham Biosciences, and polyethyleneimine was from Research Biotechnologies, Inc. (Natick, MA). LB medium was obtained from Invitrogen. Chloramphenicol, NaCl, NaOH, SDS, and MOPS 1 were from Fisher. Protein molecular weight standards were from Invitrogen. Purified proteins were concentrated using Millipore YM10 regenerated cellulose ultrafiltration membranes. Isopropyl-1-thio-␤-D-galactopyranoside was from Promega (Madison, WI), and polyethyleneimine (10% solution) was from Research Biochemicals International (Natick, MA). Lipoamide, D,L-6,8-thioctic acid amide, and dihydrolipoamide dehydrogenase (EC 1.8.1.4) from bovine intestinal mucosa, 100 units/mg protein were purchased from Sigma. Oligonucleotide synthesis and DNA sequencing were performed by the Biomolecular Resource Center of the University of California, San Francisco, CA. A PerkinElmer Life Sciences 480 DNA thermal cycler was used for PCR experiments. An Amersham Biosciences Sephadex 200 column, connected to an Amersham Biosciences PCC-500 fast performance liquid chromatography system, was used for the final stage of protein purification. A Hewlett-Packard HP-8452 UV-visible spectrophotometer was used for all spectroscopic measurements. S. typhimurium AhpF was generously provided by Patrick Hillas, as were the C130S and C133S constructs (24). All expression plasmids were introduced into competent BL21(DE3) E. coli cells.
Bacterial Cell Growth-Bacterial growth was carried out at 37°C in LB medium containing 50 g/ml chloramphenicol. One colony was used to inoculate 50 ml of LB medium containing the antibiotic, and the culture was incubated for 12 h. The culture was used to inoculate a 1-liter culture of LB containing the antibiotic at a ratio of 10 ml/liter. When the A 600 value of the culture reached 0.7-0.8, isopropyl-␤-Dthiogalactopyranoside was added to a final concentration of 0.2 mM. Incubation was continued for 12 h at 20°C. Cells were harvested by centrifugation at 5000 ϫ g for 45 min at 4°C and stored at Ϫ20°C overnight.
Protein Purification-M. tuberculosis AhpD and mutants were expressed heterologously in E. coli and purified according to the protocol of Hillas et al. (24) with some modifications.
Cells were suspended in a 6-fold excess (with respect to the initial weight of cells) of lysis buffer. The solution was stirred for 60 min at 4°C. A polyethyleneimine supernatant was prepared and loaded onto the Q-Sepharose column. After loading, the resin was washed with the same buffer for 20 column volumes. The protein was eluted with a gradient from 0 to 0.75 M KCl in 50 mM potassium P i , pH 7.0, 1.0 mM dithiothreitol, 1.0 mM EDTA, and 5% glycerol. The protein eluted at ϳ0.2 M KCl. Fractions containing maximum amount of protein, as assessed by denaturing 20% polyacrylamide gels using the PhastSystem (Amersham Biosciences), were pooled and concentrated to 5 mg/ml or greater using a Millipore concentrator equipped with a YM10 (10,000 molecular weight cut-off) regenerated cellulose ultrafiltration membrane. Enzyme concentrations were determined from the molar absorption coefficients using the method of Pace et al. (27) and a molar extinction coefficient of 15,720 M Ϫ1 cm Ϫ1 . In contrast to AhpD a single ion-exchange chromatographic protocol was insufficient to purify the AhpD E118Q and H137Q mutants. Further purification by gel filtration on Superdex 200 using 50 mM MOPS, pH 7.2, 100 mM KCl, 20% glycerol, 5 mM dithiothreitol, and 0.2 mM EDTA was therefore carried out with all proteins. The proteins were judged to be Ͼ95% pure by denaturing SDS-PAGE. The proteins were finally aliquoted into 25-or 100-l volumes, frozen on dry ice, and stored at Ϫ80°C until used.
AhpF-dependent Activity Assays-Rates of hydroperoxide reduction were determined anaerobically in a coupled assay with AhpF, monitoring the decrease in absorbance at 340 nm because of NADH oxidation as reported previously (24). The assays typically contained 2 mM hydroperoxide substrate in 100 mM potassium P i , pH 7.0, 1 mM EDTA, 0.25 mM NADH, 20 M AhpD or mutant, and 10 M AhpF. Background NADH oxidation caused by AhpF was monitored, the hydroperoxide substrate was added, and the enzymatic rate was observed.
AhpD-dependent Peroxidase Activity AhpC Assays-The rate of NADH oxidation catalyzed by AhpC in the presence of AhpD, lipoamide, and bovine lipoamide dehydrogenase was measured as described by monitoring the change in absorbance at 340 nm (25). Typical conditions for the assays were as follows: 50 mM potassium P i , pH 7.0, 1 mM EDTA, 200 M NADH, 2.5 M AhpC, 2.5 M AhpD, 0.2 units of bovine dihydrolipoamide dehydrogenase, and 50 M lipoamide. This assay can be used to evaluate the activity of AhpD and AhpD mutants, because the AhpD-supported activity of AhpC directly depends on the activity of AhpD. For steady-state kinetic assays, the substrate concentration was varied, and data were fit to the equation Measurement of Thiol pK a Values by UV Absorption-The pH-dependent ionization of the cysteine thiols was monitored by the absorbance of the thiolate anion at 240 nm as described (28,29). All of the measurements were carried out at 25°C, with 10 M AhpD or AhpD mutant in 1 mM each of argon-purged citrate, borate, and phosphate buffer containing 0.2 mM KCl and 5 mM dithiothreitol. The solution was titrated with 0.2 M KOH.
Crystallization-Crystallization of the AhpD H137F and H132Q mutants was carried out by the hanging drop vapor diffusion method reported previously (26). Hanging drops were prepared by mixing equal volumes of protein solution (4.5 mg/ml in 25 mM MOPS buffer, pH 7.2, containing 50 mM KCL, 10% glycerol, 0.1 mM EDTA, and 5.0 mM dithiothreitol) with a reservoir solution containing 100 mM sodium citrate buffer at pH 5.6 containing 200 mM ammonium acetate and 26% polyethylene glycol 4000. Crystals grew as rhombohedral prisms over a 1 The abbreviation used is: MOPS, 4-morpholinepropanesulfonic acid. period of days. The crystals belong to space group C2 for both mutant structures with cell dimensions a ϭ 96.60 Å, b ϭ 62.18 Å, c ϭ 89.48 Å, ␤ ϭ 121.81°for H137F and a ϭ 100.12 Å, b ϭ 58.65 Å, c ϭ 88.91 Å, ␤ ϭ 120.47°for the H132Q mutant. A Matthews coefficient of 1.96 corresponds to 37% (v/v) solvent for the crystallographic asymmetric unit of H137F, and a value of 1.93 corresponds to 34% (v/v) solvent in the crystallographic asymmetric unit of H132Q (30).
Data Collection and Processing-Data collection parameters are shown in Table I. All data processing, integration, and scaling was carried out using HKL (31) and the CCP4 suite of programs (32).
Model Building and Refinement-Throughout the refinement 10% of the reflections were used for cross-validation analysis (33), and the behavior of R free was used to monitor the refinement strategy. All refinement was carried out using the program CNS v.1.1 (34), and model building was carried out using programs O (35) and TURBO-FRODO (36).
A protein trimer exists within the crystallographic asymmetric unit. One AhpD trimer from the native crystal structure (Protein Data Bank code 1GU9) was used as a model for molecular replacement. Crossrotation and translation functions were calculated using data from 15 to 4 Å resolution, and a clear solution was obtained for both mutant structures. Following molecular replacement rigid body refinement was carried out prior to full positional refinement using simulated annealing with torsion angle dynamics, anisotropic scaling, energy minimization, individual isotropic B-factor refinement, and bulk-solvent correction against the maximum-likelihood target. Refinement progressed well with further rounds of positional and individual B factor refinement. Toward the end of the structure refinement electron density maps were used to locate and include solvent water within the refinement. Refinement parameters for both mutant structures are given in Table II.
Preparation of Figures-The structural figures were prepared using DINO (www.dino3d.org).
Coordinates-The atomic coordinates and structure factors have been deposited in the Protein Data Bank with accession codes 1LW1 (H137F) and 1ME5 (H132Q).

RESULTS
Catalytic Activities of AhpD Mutant Proteins-The recently determined crystal structure of AhpD suggests important roles for Glu 118 , Cys 130 , Cys 133 , His 132 , and His 137 in the catalytic mechanism of AhpD (25,26). To ascertain the roles of these residues, single mutants were constructed in which each of the cysteines was mutated to a serine, and each of the two histidines and the glutamic acid was mutated to either a phenylalanine or a glutamine. The mutant proteins were expressed and purified by minor modifications of the protocol reported earlier for purification of wild-type AhpD (24,26). The two AhpD Cys 3 Ser mutants and the two His 3 Phe mutants were expressed in yields comparable with those of the wild-type protein (ϳ25 mg/L), but the AhpD H137Q, E118F, and E118Q mutants were expressed at significantly lower levels (ϳ20% of wild-type). The glutamine mutants do not appear to bind to Q-Sepharose as strongly as native AhpD or the other mutants and tend to unfold and degrade relatively easily. The glutamine substitution may destabilize the protein fold or may force the protein into a conformation that exposes proteolytically sensitive sites.
The catalytic activities of the mutants were evaluated in two assay systems. In the first system, the ability of the AhpD mutants to reduce cumene hydroperoxide in the presence of the S. typhimurium AhpF and NADH was evaluated (24). This assay system tests the ability of AhpD to directly reduce a hydroperoxide in a reaction supported by a surrogate electron donor partner. The AhpF-supported catalytic activities of wildtype AhpD and its mutants are shown in Fig. 2. As reported previously (24), the two cysteine mutations impair the alkyl peroxidase activity of AhpD, with mutation of Cys 133 completely suppressing the activity and mutation of Cys 130 decreasing it to less than 5% of the wild-type activity (Fig. 2). Likewise, the two His to Phe mutations decrease the catalytic activity, but mutation of His 137 causes a greater decrease than mutation of His 132 . Mutation of the two His residues to glutamines also decreases the level of activity but to a lower extent than mutation to phenylalanines. As found for the phenylalanine mutants, the H137Q mutation decreased the activity more severely than the H132Q mutation. Replacement of Glu 118 by a  glutamine only modestly lowered the catalytic activity, whereas replacement by a phenylalanine decreased the activity to a much higher extent. In the second assay system, reduction of cumene hydroperoxide was measured in the fully reconstituted system consisting of AhpC, the AhpD mutant, lipoamide, bovine dihydrolipoamide dehydrogenase, and NADH (25). The lipoamide-dihydrolipoamide dehydrogenase-AhpC assay (Fig. 3) gives higher absolute alkylhydroperoxidase activities than the AhpF-dependent assay (Fig. 2). Nevertheless, as shown in Fig. 3, the impairment of the catalytic activities caused by the mutations follows the same trends as seen for direct reduction of the peroxide by AhpD/AhpF. The ratios among the activities of the different mutants are similar to those found in the AhpF assay. Thus, the two cysteine mutants are completely inactive in this assay, in agreement with an earlier report (25), the H137F mutation causes a greater impairment than the His 132 mutation, and the E118Q mutation has a relatively minor effect when compared with the E118F mutation.
The pH Dependence of Wild-type and Mutant AhpD Proteins in the Lipoamide-Lipoamide Dehydrogenase-AhpC Assay-The pH dependence of the activities of native AhpD and three of its mutants are shown in Fig. 4. If either AhpC or AhpD is assayed in the absence of the other, the level of activity is several times lower than when both enzymes are present. The pH profiles of AhpD and its mutants are bell shaped. The pH dependence of AhpD alone may be considered as the background pH-dependent activity, as the activity of AhpC alone does not appear to be sensitive to pH in the pH 5-10 range. The optimum activity of native AhpD occurs at pH 7.2. Below pH 7.0 and above pH 7.5, there is a gradual decrease in the activity. Substitution of either His 132 or His 137 by a Gln shifts the pH optimum to a value of 7.5. This shift to a higher pH optimum indicates that His 137 and His 132 help to determine the pK a values of critical catalytic residues, as implied by the postulated catalytic mechanism (26).
The pK a Values of the Cysteine Residues in AhpD-To evaluate the pK a of the two cysteine residues in AhpD the thiolate anion absorbance at 240 nm (29) was monitored between pH 6 and 9. The increase in the 240-nm absorption with pH indicated it was caused by deprotonation of a thiol with pK a ϭ 7.2 Ϯ 0.1 (Fig. 5). In an effort to identify the cysteine with the indicated pK a value, the thiol titration experiment was carried out with the C130S and C133S mutants of AhpD. Measurements of the increase in the absorbance at 240 nm indicated that Cys 133 in the C130S mutant had a pK a of 6.9 (Fig. 5). In contrast, similar measurements with the C133S mutant indicated that the remaining cysteine, Cys 130 , had a pK a of 7.5. The pK a of ϳ7.2 determined for the wild-type enzyme is thus an average of the values for the two cysteines, one with a pK a of 6.9 and the other of 7.5. These results are based on the assumption that the two cysteine thiol pK a values are independent of each other, so that mutation of one cysteine does not influence the pK a of the second cysteine. No direct hydrogen bonding connection is apparent in the crystal structure between the two cysteines, so this is likely to be a reasonably valid assumption.
Crystal Structures of the AhpD H137F and H132Q Mutants-The crystal structures of the two mutant AhpD proteins display a homotrimeric arrangement with a similar overall conformation for each subunit to that observed in the crystal structure of native AhpD (26). The overall root mean square deviation between the H132Q and H137F mutant structures is 1.2 Å, whereas the root mean square deviations between those of the H137F and H132Q mutants relative to that of native AhpD are 1.1 and 1.2 Å, respectively.
Each of the three protomers comprises eight ␣-helices. These three protomers are arranged around a positively charged central cavity that is filled with water molecules. The catalytically active Cys 130 and Cys 133 residues lie close to the surface of the protein, with Cys 130 close to the surface of the central cavity and Cys 133 at the surface on the outside edge of the structure. Fig. 6 shows a section through the protein trimer that clearly illustrates the positions of the Cys 130 and Cys 133 residues. Cys 133 lies at the base of a deep groove on the external protein surface close to His 137 and Glu 118 , both of which also line the wall of this external surface. This groove is proposed to be the substrate binding site (26).
Overlays of the active site residues of the mutant structures with those of the native structure are shown in Fig. 7. The mean S-S distances within the H132Q and H137F mutants are 3.6 and 3.5 Å, respectively, compared with 3.5 Å for the native AhpD structure (26). For the H137F mutant the shortest distance between a Glu 118 oxygen atom and Phe 137 has increased to a mean value of 3.8 Å, well in excess of the hydrogen bonding interactions observed in both the native AhpD and H132Q mutant structures, where the Glu 118 oxygen atom hydrogen bonds with the nitrogen of His 137 . Other interactions within the active site of the H137F mutant are similar to those observed within the native structure. In the case of the H132Q mutant the position of Gln 132 differs from that seen for His 132 within the native AhpD structure.
Water-mediated interaction between His 137 and Cys 133 was reported previously (26) in the structure of native AhpD. Similar solvent molecules were also located within the crystal structure of H132Q. For protomers A and B the distance between the solvent site and the NE2 atom of His 137 is 3.16 Å with a slightly longer distance (mean value 3.54 Å) to the SG atom of Cys 133 . The solvent site for protomer C is less well defined and lies 3.98 Å from NE2 of His 137 and 3.41 Å from SG of Cys 133 . The distance between the nitrogen atom of His 137 and Cys 133 sulfur atom has a mean value of 5.0 Å in the H132Q mutant structure as compared with a value of 4.9 Å in native AhpD (26). DISCUSSION The mechanism proposed for catalytic turnover of the M. tuberculosis AhpD, based on the structure of the wild-type protein, involves reaction of the Cys 133 thiolate with the substrate. The reaction could involve attack at either the dioxygen bond of a peroxide or the disulfide bond of oxidized AhpC (26). This reaction was postulated to be facilitated by a decrease in the pK a of Cys 133 because of a relay system in which the thiol proton is removed by a water molecule whose basicity is increased by hydrogen bonding to His 137 , which in turn is activated by hydrogen bonding to the carboxylate group of Asp 118 .
We have demonstrated here that the pK a of the Cys 133 thiol group is 6.9, a value significantly lower than that of 7.5 for Cys 130 . At a pH of 7.2, the optimum for wild-type AhpD activity (Fig. 4), the major fraction (ϳ66%) of the thiol group in Cys 133 would be in the deprotonated, thiolate, form. These results are consistent with the proposal that Cys 133 is involved in a nucleophilic attack on the substrate (26).
Mutagenesis shows that a C133S mutation suppresses the AhpF-supported alkylperoxidase activity of AhpD, and a C130S mutation is almost as effective (Fig. 2) (24). When the activity of AhpD is assayed in a system also containing AhpC, lipoamide dehydrogenase, and lipoamide, the two cysteine mutations completely eliminate catalytic activity (Fig. 3). These results are consistent with the crystal structure of the oxidized form of AhpD, which shows that Cys 130 is linked to Cys 133 via a disulfide bond (25).
To explore the role of His 137 in activating Cys 133 for catalytic attack on the substrate, we have mutated it to a phenylalanine and a glutamine. As shown in Figs. 2 and 3, mutation to a phenylalanine, which is not able to hydrogen bond to the proposed catalytic water molecule, severely depresses the activity of the enzyme in both the AhpF and AhpC/lipoamide dehydrogenase, lipoamide catalytic systems. Mutation to a glutamine, a residue with a side chain that is still able to enter into hydrogen bonding interactions, depresses the activity relative to the wild-type but to a much lower extent than the H137F mutation. The crystal structure of the H137F mutant supports these results, as phenylalanine is unable to engage in a watermediated interaction with Cys 133 . Although aromatic hydrogen bonding interactions are possible between phenyl aromatic rings and hydrogen bond donors (37) no such interactions are observed for H137F. If protonated, Glu 118 could act as a hydrogen donor, in a hydrogen bond analogous to that of Glu 118 -His 137 in the wild-type enzyme. However, the mean Phe 137 -Glu 118 separation for the H137F mutant is 3.8 Å, and the geometry is inconsistent with hydrogen bonding between these side chains. The separation between His 137 and Cys 133 in both the native and H132Q AhpD structures (26) is spatially similar to that observed between the active site residues in thioredoxin ( Fig. 8) (38). In thioredoxin the catalytic mechanism proceeds via a mixed disulfide intermediate whose breakdown is enhanced by the involvement of the residue Asp 30 that acts as a base catalyst toward residue Cys 39 . The separation between Asp 30 OD1 and Cys 39 SG in the thioredoxin crystal structure is 5.9 Å and too large for direct proton transfer. It is proposed that proton transfer between Asp 30 and Cys 39 is mediated by a water molecule resulting in the subsequent nucleophilic attack of Cys 39 on the mixed disulfide bridge (39). The effect of replacing the Asp by Asn in the active site of thioredoxin (16,18) was of comparable magnitude to the effect on the catalytic activity created by replacement of His 137 by Gln providing further parallels in the mechanisms of the two enzymes.
Mutation of Glu 118 , the residue postulated to activate His 137 , to a glutamine decreases catalytic activity in both assay systems but only to a small degree, whereas mutation to a phenylalanine lowered the activity to a higher degree. It is clear from this that glutamic acid can be replaced by a glutamine in the catalytic mechanism. Efforts to make the E118F mutant were not successful because of the instability of the resulting protein.
A role was also postulated for His 132 in the catalytic mechanism (26), in this case in helping to deprotonate Cys 130 in the step that releases the reduced alcohol product or the reduced AhpC while concomitantly producing the oxidized (disulfide) form of AhpD. Mutation of His 132 to a phenylalanine, as predicted, decreased the activity of the enzyme in both catalytic assays to an extent approaching that of the H137F mutation. Replacement of His 132 by a glutamine also depressed the activity but less than replacement by a phenylalanine. Again, the effect of the mutation is less than the corresponding mutation of His 137 but is nevertheless significant. This result provides support for the proposed role of His 132 in the catalytic process.
Within the crystal structures of both native AhpD (26) and the H137F mutant, the nitrogen atom of His 132 lies at ϳ 5Å from the Cys 130 sulfur atom with the imidazole ring oriented away from Cys 130 . This orientation could not facilitate direct deprotonation without a change in the torsion angles of the His 132 side chain that produces a geometry between the imidazole nitrogen atom of His 132 and the sulfur atom of Cys 130 suitable for interaction of these two centers. Inspection of the crystal structure shows that reorientation of His 132 via a rotation about the CA-CB is possible, and therefore the flexibility about this bond could effect such a conformational change under solution conditions. Interestingly, significant conformational flexibility was observed for the side chain of the H132Q mutant within the trimeric crystal structure, in accord with the notion that the active site allows a torsional flexibility of either a His (wild-type) or Gln (mutant) side chain that enhances Cys 130 deprotonation.
AhpD, the first member of a new family of proteins unrelated by sequence to any other proteins, is shown by the present results to employ a variant of the mechanism used by thioredoxin to enhance the nucleophilicity of its central catalytic cysteine residue. In thioredoxin, a water molecule mediates the interaction of a carboxylate group with the cysteine (39), but in AhpD there is an additional player, a histidine, in the proton relay mechanism. This novel proton relay system apparently enables the cysteine to react effectively with either the oxygen-oxygen bond of a peroxide or the disulfide bond of the oxidized form of AhpC. A second histidine, His 132 , appears to facilitate the subsequent internal reaction of Cys 130 with the initial sulfenic acid or the disulfide cross-link between Cys 133 and a cysteine of AhpC. The resulting disulfide form of AhpD must then be reduced by a lipoamide-dependent process, but it is not known whether the catalytic residues play any role in this reductive reaction.