The Antituberculosis Drug Ethionamide Is Activated by a Flavoprotein Monooxygenase*

Ethionamide (ETA), a prodrug that must undergo metabolic activation to exert its cytotoxic effects, is a second line drug against tuberculosis, a disease that infects more than a third of the world's population. It has been proposed, on the basis of genetic experiments, that ETA is activated in Mycobacterium tuberculosis by the protein encoded by the gene Rv3854c (DeBarber, A. E., Mdluli, K., Bosman, M., Bekker, L.-G., and Barry, C. E., III (2000) Proc. Natl. Acad. Sci. U. S. A. 97, 9677–9682; Baulard, A. R., Betts, J. C., Engohang-Ndong, J., Quan, S., McAdam, R. A., Brennan, P. J., Locht, C., and Besra, G. S. (2000) J. Biol. Chem.275, 28326–28331). We report here the expression, purification, and characterization of the protein encoded by this gene. Our results establish that the enzyme (EtaA) is an FAD-containing enzyme that oxidizes ETA to the corresponding S-oxide. TheS-oxide, which has a similar biological activity as ETA, is further oxidized by EtaA to 2-ethyl-4-amidopyridine, presumably via the unstable doubly oxidized sulfinic acid intermediate. This flavoenzyme also oxidizes thiacetazone, thiobenzamide, and isothionicotinamide and thus is probably responsible, as suggested by the observation of crossover resistance, for the oxidative activation of other thioamide antitubercular drugs.

Tuberculosis continues to be a major worldwide epidemic with approximately one-third of the world population infected with Mycobacterium tuberculosis, 7 million people each year developing the active disease, and 2 million deaths per annum (1,2). Drugs such as isoniazid (INH) 1 and rifampicin have historically been successful in the treatment of tuberculosis infections. In recent history, however, poor compliance with the prolonged and complicated chemotherapeutic regimens currently used to treat the disease (3), in conjunction with the advent of the AIDS epidemic and the increased mobility of human populations, has led to the emergence of numerous multidrug-resistant M. tuberculosis strains (4). Resistance to frontline therapeutics, most notably INH and rifampicin, results in treatment of patients with "second-line" agents that are less effective and/or more toxic. Among these second tier drugs for the treatment of multidrug-resistant tuberculosis, one of the most effective is ethionamide (ETA) (5).
ETA (1, Fig. 1), like INH, is thought to be a prodrug that must be converted to its active form by the bacterial cell. Both ETA and INH, when activated, appear to disrupt cell wall biosynthesis and have at least one common cellular target, the enoyl-acyl carrier protein reductase InhA (6,7). Support for a common site of action can be deduced from gene array studies demonstrating that both ETA and INH induce similar patterns of gene expression in M. tuberculosis (8). Despite this evidence for a common site of action, INH and ETA are activated by different mechanisms as resistance to INH does not confer resistance to ETA (9). INH is now known to be oxidized by the bacterial catalase-peroxidase KatG to a reactive species, probably an acyl free radical, that is eventually responsible for its bacterial toxicity (10 -13). Mutations in KatG that diminish or annihilate its activity toward INH are responsible for a large proportion of the INH-resistant M. tuberculosis strains. Until recently, however, the enzyme that activates ETA has remained obscure (14,15). Sequence homology studies using the Rv3854c gene as a template indicate that the protein responsible for ETA activation may be a flavin monooxygenase (14,15). In vivo mammalian and bacterial studies suggest that the first metabolite of ETA is the thioamide S-oxide (ETA-SO, 2, Fig. 1), which retains the biological activity of the parent drug (16 -19). Other metabolites identified in whole cell bacterial systems include the nitrile (3), amide (4), and alcohol (6) derivatives of ETA ( Fig. 1) (14).
Recently two laboratories independently reported identification of a gene, Rv3854c, in the M. tuberculosis genome that codes for a protein that activates ETA (14,15). Genetic and transfection experiments provided strong evidence for the biological role of the Rv3854c encoded protein (EtaA) in the activation of ETA. However, the actual enzyme was not isolated, and its nature remains speculative. Herein we report cloning, heterologous expression, purification, and characterization of the Rv3854c gene product. Our results establish that the enzyme responsible for ETA activation is an FAD-containing enzyme, provide information on the catalytic and physical properties of this enzyme, and demonstrate that it catalyzes two steps rather than one step in the activation of ETA.

Instrumentation
UV-visible spectra were obtained on either a Hewlett Packard model 8254A diode array instrument or a Varian Cary 1E spectrophotometer. NMR spectra were collected on a Varian Unity 400 MHz instrument. Mass spectra were obtained using a Thermoquest Finnigan LCQDECA electrospray ionization (ESI) ion trap spectrometer. HPLC was performed on a Hewlett Packard Series 1090 liquid chromatograph equipped with either a Rheodyne model 7125 manual sample injector or an autosampler.

Kinetic Assays
All activity and steady state kinetic assays were conducted by monitoring the formation of the product ETA-SO (2). Product formation was measured at 350 nm over a time of 5-10 min using the Cary 1E split beam spectrophotometer. The molar extinction coefficient of ETA-SO in the reaction system was experimentally determined to be 6,000 (Ϯ4%) M Ϫ1 cm Ϫ1 . The reaction medium contained 50 mM Tris buffer (pH ϭ 7.5), 100 mM KCl, catalase (100 units/ml), superoxide dismutase (100 units/ ml), bovine serum albumin (0.1 mg/ml), and an NADPH-regenerating system consisting of glucose-6-phosphate dehydrogenase (2 units/ml), glucose 6-phosphate (250 mM), and either NADP ϩ or NADPH. The final concentration of enzyme was typically 500 nM in FAD. ETA stock solutions (25 mM) were freshly prepared in acetonitrile, and the final added volume never exceeded 1% of the total reaction volume. All stock solutions were kept on ice prior to mixing within the cuvette.
The enzymatic reaction was typically initiated by adding substrate (ETA) to the sample cuvette containing the reaction medium plus the enzyme and an equivalent volume of acetonitrile to the reference cuvette. The reference cell contained all solution components except for enzyme and substrate (ETA), allowing for accurate correction of any NADPH absorption in the region of interest. Prior to addition of substrate, the reaction and reference mixtures were thermally equilibrated in the cuvettes at 37°C for 2 min.

Synthesis of Proposed Metabolites
ETA-SO (2)-ETA-SO was prepared from ETA following a modification of a published protocol (20). ETA (1.45 g, 8.72 mmol) was suspended in absolute ethanol (7 ml) immersed in a room temperature water bath. Hydrogen peroxide (30% (v/v), 922 l, 9.59 mmol) was added slowly while monitoring the solution temperature. The temperature was never allowed to exceed 35°C. After all of the H 2 O 2 was added, the solution was stirred at room temperature for 45 min. The reaction was monitored by TLC (silica on plastic, eluent: 90% ethyl acetate, 10% methanol) and quenched with 3 ml of water when the reaction was ϳ90% complete. The solvent was removed in vacuo resulting in an orange oil. Recrystallization at Ϫ20°C from hot chloroform layered with diethyl ether yielded 172 mg (0.944 mmol) of a bright yellow microcrystalline solid. This solid material is stored in the dark at Ϫ20°C: 1

2-Ethyl-4-cyanopyridine (3)-
The reaction was carried out under a dry argon atmosphere with flame-dried glassware. Following a published procedure (21), CH 2 Cl 2 (10 ml) was transferred by syringe to a round bottom flask charged with ETA (333.5 mg, 2.01 mmol), and the suspension was cooled to 0°C. First triethylamine (635 l, 4.50 mmol) and then 2,2,2-trichloroethyl chloroformate (315 l, 2.29 mmol) were added dropwise. The reaction was monitored by silica TLC (eluent: 30% ethyl acetate in hexanes) for product formation. The solution was stirred under argon for 1.5 h, allowing it to warm to room temperature. The reaction was quenched with water (10 ml) and extracted with CH 2 Cl 2 (2 ϫ 15 ml). The organic layer was then washed with brine and dried with Na 2 SO 4 . The solvent was then removed in vacuo, and the resultant residue was purified on a silica column (12.5 ϫ 2.3 cm, l ϫ diameter) packed in a mixture of 5% ethyl acetate in hexanes. The product was eluted using a gradient of 10 -15% ethyl acetate in hexanes (5%/100 ml). This procedure yielded 192 mg (1.45 mmol) of pure product (72% purified yield) as a pale yellow oil: IR (neat) 2238 cm Ϫ1 (CN asymmetrical stretch); 1  2-Ethyl-4-amidopyridine (4)-Following a published procedure (22), the amide was synthesized from the nitrile under phase transfer catalytic conditions. A solution of 3 (84 mg, 0.635 mmol) and tetrabutylammonium tetrafluoroborate (62 mg, 0.188 mmol) in CH 2 Cl 2 (0.318 ml) was cooled to 0°C in an ice bath. To the cooled solution, 30% aqueous H 2 O 2 (0.424 ml) was slowly added followed by 20% (w/w) aqueous NaOH (0.339 ml). The solution was vigorously stirred for 2 h, allowing the solution to warm to room temperature. The reaction mixture was extracted with CH 2 Cl 2 (10 ml), and the organic layer washed with an equal volume of brine and dried over Na 2 SO 4 . The solvent was removed in vacuo, and the residue was purified on an alumina column (8.5 ϫ 2.3 cm, l ϫ diameter) packed in ethyl acetate. The product was eluted using a gradient of 0 -20% CH 2 Cl 2 in ethyl acetate (5%/100 ml). This procedure yielded 57 mg (0.38 mmol) of pure product (50% purified yield) as a white solid: 1  2-Ethyl-4-carboxypyridine (5)-The nitrile, 3 (529 mg, 4 mmol), was dissolved in water (10 ml). Concentrated H 2 SO 4 (4 eq) was then added, and the solution was refluxed for 4 h. The reaction mixture was extracted with CH 2 Cl 2 to remove the residual starting material before the aqueous layer was basified to pH ϭ 10 with aqueous NaOH. The water was evaporated in vacuo, and the residue was redissolved in methanol and filtered. The sodium salt of the acid (153 mg, 0.884 mmol) was obtained as a yellowish powder following evaporation of the methanol: 1

Protein Expression and Purification
The gene was obtained as an insert in the pMH29 vector from Clifton Barry III and Andrea E. DeBarber (National Institutes of Health). PCR was used to amplify the insert with NdeI and HindIII restriction sites at the 5Ј and 3Ј termini, respectively. The primers used for the PCR were: 5Ј-end, 5Ј-GCGTGGACATATGACCGAGCACCTCG; 3Ј-end, 5Ј-CTAAAGCTTCGCTAAAGCTAAACCCCC (the restriction sites are underlined). The insert was ligated into the pCWori expression vector containing a poly-His insert at the 5Ј-end of the coding sequence just prior to the start codon. Sequencing was used to confirm that the full-length gene without alterations in the sequence was inserted into the expression vector. The expression vector was transformed into commercial (Invitrogen) competent DH5␣ Escherichia coli cells, and protein expression was induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside once the cells grew to A 600 ϭ 0.9. Cells were grown for 24 h postinduction at 22°C and were harvested by centrifugation at 4,000 ϫ g.
The E. coli cells were resuspended in 50 mM Tris (300 mM NaCl, 1% (v/v) Triton X-100, 6 mM imidazole, pH ϭ 7.4), lysed, and sonicated. The cytosolic fraction was removed after centrifugation at 25,000 ϫ g. This supernatant was loaded directly onto a Ni-NTA affinity column preequilibrated with 50 mM Tris, 500 mM NaCl, 10% glycerol, pH ϭ 7.4. After loading, the column was washed with 10 column volumes of the same buffer, and the protein was eluted with a gradient over 10 column volumes of 0 -200 mM imidazole in 50 mM Tris (pH ϭ 7.4) containing 500 mM NaCl and 10% glycerol. The fractions containing the pure Rv3854c gene product (EtaA), as determined by sodium dodecylsulfate polyacrylamide gel electrophoresis, were pooled and concentrated. Imidazole was removed from the protein solution by passing it through a Sephadex G-25M size exclusion column (elution buffer: 50 mM Tris, 100 mM KCl, 10% glycerol, pH ϭ 7.4) with a final yield of ϳ15 mg of pure protein/liter of culture.
Fast protein liquid chromatographic analysis of the protein was done on a Superdex 75 FPLC column eluted at a rate of 0.5 ml/min with 50 mM Tris buffer, pH ϭ 7.4, containing 100 mM KCl and 10% glycerol. Thin layer chromatographic comparison of the prosthetic group extracted from the recombinant flavoprotein with authentic FMN and FAD was done on silica gel using water as the elution solvent. The flavins could be detected visually due to their bright yellow/orange color.

HPLC Identification of Metabolites
The enzymatic reactions with ETA, thiobenzamide, isothionicotinamide, and thiacetazone were analyzed by HPLC, and the metabolites were identified, where possible, by comparison with authentic standards. The enzymatic reaction was run for 3 h at 37°C using conditions identical to those described for the kinetic assays. The crude reaction mixture was filtered through a 0.22-m syringe filter (Millex ® -GV low protein binding Durapore membrane) and analyzed, without further modification, by HPLC on a Whatman Partisil ® C8 reverse-phase column (particle size, 5 m; 250-ϫ 4.6-mm inner diameter) equipped with a C8 guard column (particle size, 5 m; 7.5-ϫ 4.6-mm inner diameter). The solvent system consisted of 0.01% formic acid in water (A) and 0.01% formic acid in acetonitrile (B). The mixture was eluted with a gradient of 1-8.3% B over 30 min. The eluent was monitored at 284 and 350 nm, and complete UV-visible spectra of the metabolite peaks were collected for comparison with authentic standards.

RESULTS
Cloning and Expression of the Protein-Genetic experiments, including expression in intact cells, have implicated the protein encoded by Rv3854c in the bioactivation of ETA to a form of the drug with direct antimycobacterial activity (14,15). The metabolic products of ETA suggest that the protein in question is likely to be a monooxygenase (15). The transformations are consistent with the involvement of either a cytochrome P450 or a flavin monooxygenase, but the protein sequence encoded by Rv3854c is more consistent with a flavoprotein (14,15). To elucidate the precise nature of the enzyme involved in the activation of ETA and the nature of the transformation(s) that it catalyzes, we have cloned and expressed Rv3854c in E. coli. The protein was expressed with a poly-His tag to facilitate purification of the protein. Expression of two constructs, one with the poly-His tag at the amino terminus and the other with the tag at the carboxyl terminus, gives protein with similar catalytic properties, indicating that the poly-His tag does not interfere with catalytic function (all subsequent experiments were conducted with the amino-terminally His-tagged protein). An exploration of expression condi-tions led to production of the enzyme in yields of 15 mg of purified protein/liter of medium. The protein was highly purified by Ni-NTA affinity chromatography as shown by the fact that even at high concentrations the protein gives rise to a single band on SDS-PAGE (Fig. 2). The specific activity of the protein product at different stages of the purification procedure, measured as indicated under "Experimental Procedures" and as discussed below, is shown in Table I. Herein the recombinant Rv3854c gene product is denoted as EtaA.
Physical Characterization of the Protein-EtaA, as expected from the predicted length of 488 amino acids, migrates with a molecular mass of ϳ55 kDa (Fig. 2). FPLC analysis of EtaA on a Superdex column shows that the protein elutes as a single peak but at a molecular weight that corresponds to that of an oligomeric (3-4-mer) species, suggesting that EtaA aggregates in solution. UV-visible spectra of solutions of EtaA exhibit maxima at 365 and 440 nm in agreement with its identification as a flavoprotein (Fig. 3). Furthermore, extraction of the prosthetic group by boiling EtaA for 5 min and sedimenting the denatured protein by centrifugation yields a supernatant with a spectrum identical to that of authentic FAD (Fig. 3). Quantitation of the flavin released from denatured EtaA yields a ratio of flavin to protein of 1:1.3, indicating that the protein binds one flavin group. Thin layer chromatographic comparison of the extracted prosthetic group (R F ϭ 0.9) with authentic FAD (R F ϭ 0.9) and FMN (R F ϭ 0.1) showed that the prosthetic group has the retention time of FAD and not FMN. Rv3854c clearly codes for a flavoprotein with a single FAD prosthetic group.
Catalytic Oxidation of ETA by EtaA-The identity of the metabolite(s) produced in the reaction of purified EtaA with ETA and NADPH was determined by letting the reaction proceed for 3 h, separating the products by reverse-phase HPLC, and comparing the peaks with authentic (synthetic) standards. The standards that were available were the S-oxide (2), 2-ethyl-4-cyanopyridine (3), 2-ethyl-4-amidopyridine (4), and 2-ethyl-4-carboxypyridine (5), all of which have been reported as metabolites from the in vivo mycobacterial oxidation of ETA (14,16). The two major metabolic products observed with this method had retention times of 9.1 and 14.9 min. The two peaks were identified as the amide (4) and ETA-SO (2), respectively, by comparison with the elution time and spectrum of the synthetic standards (t R (4) ϭ 9.1 and t R (2) ϭ 14.9 min) (Fig. 4).
Since ETA-SO exhibits similar antimycobacterial activity as ETA (14), it was important to determine whether the amide metabolite (4) is a product of EtaA metabolism of ETA-SO or a product of an alternate enzymatic oxidation of ETA. The synthetic ETA-SO standard was therefore incubated with EtaA and NADPH as described above for incubations with ETA. A further oxidative transformation was detected by HPLC (Fig.  4). The second oxidation product co-elutes with the 9.1-min metabolite observed in the HPLC analyses of the ETA incuba- tions, indicating that the enzyme is capable of metabolizing ETA-SO to the amide (4) (Fig. 4).
EtaA was further implicated in the direct metabolism of ETA and ETA-SO by analysis of 3-h (37°C) incubations of either ETA or ETA-SO with and without enzyme in the presence of NADPH and with and without NADPH in the presence of the enzyme. In both sets of experiments no metabolism or decomposition of either ETA or ETA-SO was observed by HPLC when either the enzyme or NADPH was omitted. As might be expected for a flavoprotein, both NADPH and O 2 are required for formation of both the ETA sulfoxidation product (2) and the amide (4) by EtaA. Omission of NADPH, or running the reaction under anaerobic conditions, suppressed product formation. Furthermore, NADH could not be substituted for NADPH in the reaction (data not shown).
The catalytic activity of purified EtaA with respect to ETA was evaluated by measuring the initial rate of ETA-SO formation. The rate of ETA sulfoxidation was monitored at 350 nm (see "Experimental Procedures" for specific assay conditions). The K m , V max , and k cat values for the oxidation of ETA were found, from analysis of Lineweaver-Burke double-reciprocal plots, to be 194 (Ϯ3%) M, 1.46 (Ϯ3%) nmol of ETA-SO min Ϫ1 , and 7.73 (Ϯ5%) mol of ETA-SO min Ϫ1 mol Ϫ1 , respectively (inset, Fig. 5). Spectroscopic analysis of the reaction of EtaA with ETA in the presence of an NADPH-regenerating system demonstrates a clean initial conversion of ETA to ETA-SO as evidenced by a clear isosbestic point at 309 nm (Fig. 5).
Substrate Specificity-To determine the degree of substrate specificity of EtaA, we have examined the oxidation of the ETA analogues thiobenzamide (7) and isothionicotinamide (9) (Fig.  6). HPLC analyses of EtaA incubations with thiobenzamide and isothionicotinamide reveal the formation in each instance of two observable metabolites (not shown). The products of the metabolism of thiobenzamide by EtaA have been confirmed to be the corresponding S-oxide (8) and benzamide by comparison of their absorption spectra and retention times with those of synthetic S-oxide and benzamide standards. One metabolite of the isothionicotinamide reaction was identified as the amide by comparison with commercially available material. The second metabolite was tentatively identified as the S-oxide, but an authentic standard was not available to confirm this assignment. It is nevertheless clear that both compounds are substrates for EtaA and undergo an initial oxidation similar to that observed with ETA. DISCUSSION Cloning experiments in M. tuberculosis, Mycobacterium smegmatis, and Mycobacterium bovis BCG have shown that Rv3854c codes for a protein that is critical for the antitubercular activity of ETA (14,15). These studies also demonstrated that the expression level of Rv3854c is regulated by a suppressor gene, Rv3855, and that resistance to ETA increases with increased expression levels of Rv3855. Conversely increases in the expression levels of Rv3854c confer heightened sensitivity of the bacilli to ETA (14,15). The work presented here demonstrates that the protein encoded by Rv3854c, EtaA, is a flavoprotein monooxygenase with a single FAD prosthetic group. The identity of EtaA as a flavoprotein, a classification suggested by the presence of a consensus NX 5 DX 3 GXGXXG flavin binding domain in the predicted protein sequence (23), is clearly established by the demonstration that the protein prosthetic group is a molecule of FAD. The UV-visible spectrum of the pure protein obtained after a final size-exclusion chromatographic step (Fig. 3) is typical of a flavoprotein. The Rv3854c protein is most likely membrane-associated when expressed in the E. coli system as evidenced by its presence predominantly in the pellet fraction of the first centrifugation step following a The flavin appears to partially dissociate from the protein during the last purification step. Preliminary results suggest that the protein can be reconstituted with exogenous flavin. cell lysis. The need for resolubilization of precipitated or membrane-associated protein can be avoided by introducing Triton X-100 detergent in the lysis buffer. A poly-His tag has been engineered into the amino terminus of the protein, allowing for efficient purification to homogeneity via Ni-NTA affinity chromatography. The expression and purification procedures delineated in this report allow for the isolation of highly purified, stable, and active recombinant EtaA that can be utilized in kinetic and metabolic studies.
The first step in the bioactivation of ETA is an NADPH-and O 2 -dependent reaction that yields the S-oxide metabolite ETA-SO. The K m , V max , and k cat for the initial oxidation of ETA by purified EtaA indicate that it is a reasonably good substrate for the enzyme. However, ETA-SO is not so reactive that it cannot be synthesized, stored, and physically characterized. These properties of ETA-SO, along with the observation that it is as lethal to the tubercule bacilli as ETA, suggest that ETA-SO itself requires further activation to a final cytotoxic species (15). Incubation of authentic ETA-SO with EtaA under turnover conditions shows, indeed, that it is further metabolized by the enzyme to another metabolite. Chromatographic comparison of this second metabolite with synthesized standards identify it as 2-ethyl-4-amidopyridine (Fig. 4). This second metabolite is thus a product of the enzymatic action of EtaA on ETA-SO and not a branching product stemming from an alternative reaction with ETA. However, 4 is likely not the actual enzymatic product of EtaA and ETA-SO but rather is a more stable molecule resulting from decomposition of this putative cytotoxic metabolite, possibly a sulfinic acid species (Fig. 1, 2a). Since the final metabolite, 4, has no antitubercular activity, the key species is the reactive intermediate formed by EtaA that is the precursor of the amide.
Whole cell incubations of M. tuberculosis with radiolabeled ETA have indicated that ETA-SO and the nitrile (3) are initially formed and accumulate to a maximum concentration over a couple of hours (14). Following this early increase of ETA-SO and 3, their cellular concentrations begin to decrease, and increases in the amide (4) and alcohol (6) are observed (14). However, the final metabolites produced in the intact bacillus may arise both by alternative metabolic pathways acting on ETA and secondary metabolism of the products generated by EtaA. The link between the ETA-resistant tuberculosis and mutations in the Rv3855 and Rv3854c genes (14,15) clearly indicates that the activation step is catalyzed by EtaA and therefore that the oxidation of ETA-SO catalyzed by this enzyme is critical for activation of the drug to its cytotoxic metabolite.
The natural substrate for the M. tuberculosis flavoprotein monooxygenase, EtaA, is not known. As bacteria containing mutated, presumably inactive, forms of EtaA are viable (14), the enzyme cannot be involved in the processing of an essential endogenous substrate. In agreement with the fact that patients infected with ETA-resistant M. tuberculosis strains present cross-resistance to thiacetazone (14), we have shown here that thiobenzamide (7) and isothionicotinamide (9) are also substrates for EtaA. Thiobenzamide is oxidized by EtaA to thiobenzamide S-oxide (8) and benzamide. The oxidation of isothionicotinamide (9) by EtaA generates two products, a minor product unambiguously identified as isonicotinamide by comparison with an authentic sample and a major product tentatively identified as the sulfoxide from its HPLC properties. Thus, EtaA, like the mammalian flavoprotein monooxygenases, exhibits a relatively broad substrate specificity with the primary constraint being that an appropriate oxidizable functionality be present.
Thiobenzamide, a substrate for the M. tuberculosis flavopro- tein monooxygenase (see above), is one of a variety of related compounds that are known to be substrates for the mammalian flavoprotein monooxygenases (24 -27). As shown here for the reaction with the mycobacterial enzyme, thiobenzamide is also oxidized to the corresponding S-oxide by the mammalian flavoprotein monooxygenases (17,18). The thiobenzamide S-oxide is hepatotoxic in rats (24,28), but the available evidence suggests that this hepatotoxicity requires further transformation of the S-oxide to a secondary reactive metabolite. Thus, rat liver microsomes oxidize thiobenzamide to the S-oxide plus a small amount of benzamide (29). Under the same conditions, liver microsomes convert the thiobenzamide S-oxide exclusively to benzamide. This second reaction, which was proposed to involve conversion of the thiobenzamide S-oxide to the unstable S,S-dioxide, appears to also be mediated by the hepatic flavoprotein monooxygenase because it was inhibited by inhibitors of that enzyme (29). Model studies show that thiobenzamide is oxidized to the S-oxide and subsequently to the S,S-dioxide by H 2 O 2 (30) or, much more rapidly, by a hydroperoxyflavin model of flavin monooxygenases (31). The second oxidation, which was much slower with H 2 O 2 than the first S-oxidation, produced the nitrile at pH Ͼ8, the benzamide at pH 4.0, and mixtures of the two products at intermediate pH values (30). Thus, EtaA, like the mammalian flavin monooxygenase, is capable of converting thiobenzamide to the corresponding S-oxide and of further oxidizing this intermediate to the amide.
In summary, Rv3854c codes for a flavoprotein containing a single FAD group that catalyzes the NADPH-and O 2 -dependent monooxygenation of ETA to the corresponding S-oxide. This sulfoxidation is a required step in the activation of ETA to the species directly responsible for the antimycobacterial activity of the drug. EtaA is also capable of further oxidizing ETA-SO to what is believed to be the final cytotoxic species. This reaction does not occur until ETA-SO has appreciably accumulated in the reaction mixture. We now plan to (a) explore the further activation of ETA-SO to elucidate the nature of the putative reactive metabolite of ETA and (b) express and characterize the Rv3854c mutants known to convey resistance to ETA and to evaluate their interaction with the drug. Elucidation of the role of the mutations will make possible a better understanding of the mechanism of action of flavoprotein encoded by Rv3854c.