Action Mechanism of Antitubercular Isoniazid

Activation of the antitubercular isoniazid (INH) by the Mycobacterium tuberculosis KatG produces an inhibitor for enoyl reductase (InhA). The mechanism for INH activation remains poorly understood, and the inhibitor has never been isolated. We have purified the InhA-inhibitor complex generated in the M. tuberculosis KatG-catalyzed INH activation. The complex exhibited a 278-nm absorption peak and a shoulder around 326 nm with a characteristicA 326/A 278 ratio of 0.16. The complex was devoid of enoyl reductase activity. The inhibitor noncovalently binds to InhA with a K d < 0.4 nm and can be dissociated from denatured InhA for chromatographic isolation. The free inhibitor showed absorption peaks at 326 (ε326 6900 m −1cm−1) and 260 nm (ε260 27,000m −1 cm−1). The inactive complex can be reconstituted from InhA and the isolated inhibitor. The InhA inhibitor from the KatG-catalyzed INH activation was identical to that from a slow, KatG-independent, Mn2+-mediated reaction based on high pressure liquid chromatography analysis and absorption and mass spectral characteristics. By monitoring the formation of the InhA-inhibitor complex, we have found that manganese is not essential to the INH activation by M. tuberculosis KatG. Furthermore, the formation of the InhA inhibitor in the KatG reaction was independent of InhA.

M ؊1 cm ؊1 ). The inactive complex can be reconstituted from InhA and the isolated inhibitor. The InhA inhibitor from the KatG-catalyzed INH activation was identical to that from a slow, KatG-independent, Mn 2؉ -mediated reaction based on high pressure liquid chromatography analysis and absorption and mass spectral characteristics. By monitoring the formation of the InhA-inhibitor complex, we have found that manganese is not essential to the INH activation by M. tuberculosis KatG. Furthermore, the formation of the InhA inhibitor in the KatG reaction was independent of InhA.
Tuberculosis due to Mycobacterium tuberculosis infection is the leading cause of death worldwide among known infectious diseases. A sizeable increase of tuberculosis cases in the United States since 1985 (1) is followed by a decrease in more recent years. However, ϳ13% of recent cases involve M. tuberculosis strains that are resistant to one or more frontline antitubercular drugs such as isoniazid (isonicotinic acid hydrazide, INH), 1 rifampicin, and streptomycin (2)(3)(4).
INH has been the cornerstone in tuberculosis chemotherapy for almost half a century since its discovery as a potent antituberculosis drug in 1952 (5)(6)(7). INH is a prodrug, and its antituberculosis function requires in vivo activation by KatG, an enzyme with dual activities of catalase and peroxidase. The involvement of KatG in the INH action was first implied by an apparent correlation between the loss of KatG catalase activity and INH resistance (8) and confirmed by a genetic study (9).
Clinical M. tuberculosis isolates resistant to INH were subsequently revealed to have various alterations in the katG gene (10 -12). INH activation leads to inhibition of the synthesis of mycolic acid, a long chain fatty acid-containing component of the mycobacterial cell wall (13,14). Two enzymes involved in the elongation cycle of the fatty acid biosynthesis, namely an enoyl-acyl carrier protein reductase (InhA) (15,16) and ␤-ketoacyl-acyl carrier protein synthase (17), are believed to be targets of the activated inhibitor(s).
Progress made thus far notwithstanding the mechanisms of INH action and resistance are still poorly understood. Purified KatG from either M. tuberculosis (18) or Mycobacterium smegmatis (19) catalyzes the in vitro inactivation of InhA by INH in the presence of NADH and Mn 2ϩ . However, the molecular nature for the INH activation by KatG and the functional role of Mn 2ϩ remain unclear. InhA is also inactivated in a slow nonenzymatic, Mn 2ϩ -dependent activation of INH. The crystal structure of the resulting InhA-inhibitor complex has been determined, which shows that the bound inhibitor is an isonicotinic acyl NADH (20). It is, however, uncertain whether or not the inhibitor generated by this nonenzymatic activation is identical to that formed in the KatG-dependent process. Moreover, the inhibitor derived from INH by either the nonenzymatic or the KatG-dependent activation has never been isolated, and no simple method has been developed for the detection and quantification of the inhibitor. Consequently, biochemical or biophysical characterizations of the nature and consequences of the inhibitor binding by InhA have been greatly hindered by these limitations.
We are interested in the mechanisms of the INH action and resistance. This work was carried out to isolate for the first time the InhA inhibitors generated by the nonenzymatic and the M. tuberculosis KatG-dependent processes and to characterize the free inhibitors and the InhA-inhibitor complexes. Evidence is also presented to show that neither Mn 2ϩ nor InhA is essential to the M. tuberculosis KatG-mediated activation of INH.
The polymerase chain reaction product was first cloned into pKK223-3 at the EcoRI site to generate the pKAG1 plasmid. However, the clone so obtained did not express KatG efficiently. Consequently, the katG was excised with EcoRI from pKAG1 and inserted into the EcoRI site of a modified pET20b. The modified pET20b had a deletion so that the CATATG of the NdeI site was immediately followed by the AATTC of the EcoRI site. The resulting construct, pKAG2, can be efficiently expressed in E. coli BL21 to produce a recombinant KatG with three extra amino acid residues, Met-Asn-Ser, fused to the initial N-terminal Met residue of the wild type KatG. In order for the recombinant KatG to be expressed in E. coli UM262, the XbaI-HindIII fragment containing the katG of pKAG2 was blunted at the XbaI end and then inserted into the SmaI and HindIII sites of pKK223-3 to get pKAG3. The recombinant KatG obtained from pKAG3-transformed E. coli UM262 cells was used throughout this study.
Purification of Enzymes-To obtain InhA, E. coli JM109 harboring pINA was first cultured in 6 flasks, each containing 1 liter of Luria-Bertani (LB) medium with 100 mg/l ampicillin, at 37°C to reach about 1.0 A 600 and then incubated for 8 h with 50 mg/l isopropyl-␤-D-thiogalactopyranoside. To 30 g of wet cell paste, 200 ml of 25 mM P i was added and the suspension was sonicated for 20 min. The lysate obtained by centrifugation of the lysed cells at 8000 ϫ g for 15 min was loaded on a 2.5 ϫ 10-cm DEAE-Sepharose column equilibrated with 25 mM P i . The column was eluted with 50 ml of 35 mM P i followed by 150 ml of 50 mM P i . InhA, identified by SDS-polyacrylamide gel electrophoresis, was associated with the major A 280 peak obtained in the 50 mM P i wash. The InhA pool was diluted 1:2.5 with water and loaded on to a 2.5 ϫ 10-cm DEAE-Sepharose column. The column was sequentially eluted with 60 ml of 40 mM P i , 50 ml of a linear gradient from 40 -50 mM P i , and lastly 50 mM P i . InhA was recovered under a single major peak, and the constituent fractions were pooled. Four grams of ammonium sulfate were added to 100 ml of the InhA sample so obtained. The sample was loaded on a 1 ϫ 10-cm phenyl-Sepharose column preequilibrated with 0.5 M P i . The column was washed first with 50 ml of 0.5 M P i followed by 10 ml of 50 mM P i . InhA was recovered, with Ͼ 95% purity based on SDS-polyacrylamide gel electrophoresis pattern, by elution with water. The identity of InhA was confirmed by N-terminal amino acid sequencing.
For the expression of KatG, pKAG3-transformed E. coli UM262 cells were first grown overnight on an LB plate and were resuspended in 60 ml of LB medium supplemented with 100 mg of ampicillin for inoculation into 6 flasks each containing 1 liter of the same medium. Cells was cultured at 37°C for at least 24 h. The level of catalase activity was followed throughout the growth, and cells were harvested when the catalase activity in cells from 1 ml of culture reached about 0.5 ⌬A 240 / min in 1 ml of assay solution. About 30 g of the wet cell paste were sonicated in 200 ml of 25 mM P i and cooled on ice for 20 min. The lysate obtained by centrifugation at 8000 ϫ g for 15 min was directly loaded on a 2.5 ϫ 10-cm Q-Sepharose column and washed with 200 ml of 25 mM P i . KatG was recovered by elution with 100 mM P i . Ammonium sulfate was added to the KatG pool for a final concentration of 0.4 M. The sample was then loaded on a 2.5 ϫ 8-cm phenyl-Sepharose column preequilibrated with 0.4 M P i . The column was sequentially eluted with 50 ml of 0.3 M P i , 200 ml of a linear gradient of 0.3 M P i -water, and 100 ml of water. The KatG pool was diluted with water to a conductivity of Ͻ 3 mS/cm at room temperature and loaded on a 2.5 ϫ 10-cm DEAE-Sepharose column. The column was eluted with 200 ml of a linear gradient of 30 -300 mM P i . Fractions with A 407 /A 280 of Ͼ 0.65 were pooled. The KatG sample so obtained was apparently Ͼ 95% pure based on SDS-polyacrylamide gel electrophoresis.
KatG Assays-The catalase activity of KatG was assayed by adding the enzyme into a 1-ml H 2 O 2 solution in 50 mM P i , pH 7.0, and monitoring the decrease in A 240 associated with the decomposition of H 2 O 2 (⑀ 240 ϭ 0.0436 mM Ϫ1 cm Ϫ1 ) (22). The peroxidase activity of KatG was measured as the oxidation rate of 0.1 mM O-dianisidine by monitoring the increase at A 460 (⑀ 460 ϭ 11.3 mM Ϫ1 cm Ϫ1 ) in the presence of 23 mM t-butyl hydroperoxide at 23°C in 50 mM P i , pH 7.0 (23). One unit of peroxidase activity is defined as the oxidation of 1 mol of O-dianisidine/min at 23°C.
InhA Assay-Enoyl reductase activity of InhA was assayed by monitoring the oxidation of NADH by octenoyl CoA. Reactions were initiated by adding InhA or InhA-inhibitor complex into 1 ml of 50 mM P i containing 50 M NADH and 34 M octenoyl CoA, and changes in A 340 were monitored immediately thereafter. The concentration of octenoyl CoA was determined on the basis of ⑀ 295.5 ϭ 16,800 M Ϫ1 cm Ϫ1 . Octenoyl CoA was synthesized according to the method of Goldman and Vagelos (24). The quality of the octenoyl CoA so obtained was essentially the same as that reported earlier (18).
INH Activation-The formation of InhA-inhibitor complex was used to assess the activation of INH by KatG. A 1-ml 50 mM P i solution containing 2 mM NADH, 1 mM INH, 0.5 M KatG, and 60 M InhA was incubated for 150 min at 23°C under aerobic condition. The reaction solution was loaded on a 1 ϫ 40-cm Sephadex G-25 column preequilibrated and eluted with 50 mM P i . The gel filtration was repeated once more for the pool of the protein fractions. The UV-visible spectrum of the protein peak was measured. In comparison with the absorption spectrum of InhA, increases in absorbance at 326 nm and the A 326 /A 278 ratio were used as indicators for the InhA-inhibitor complex formation.
Isolation of InhA Inhibitor-A 10-ml 50 mM P i solution containing 180 M InhA, 2.4 mM INH, 6.4 mM NADH, and 3 M KatG was shaken under aerobic condition at room temperature for 150 min. The reaction solution was then loaded 2 ml/run on a Sephadex G-25 column (1 ϫ 40 cm) preequilibrated with 10 mM P i and eluted with the same buffer. The protein peak containing the InhA-inhibitor complex and a trace amount of KatG was loaded on a 1 ϫ 10-cm DEAE-Sepharose column preequilibrated with 10 mM P i . The column was first washed with 10 mM P i until no absorbance in the range of 250 to 650 nm was present in the filtrate. The InhA-inhibitor complex, freed from KatG, was then recovered by elution with 50 ml of a linear gradient from 50 -200 mM P i . The sample was heated in boiling water for 40 s. Denatured protein was separated from the inhibitor by centrifugation of the sample through Microcon 3 with a molecular weight cut-off of 3000 (Millipore). The inhibitor was concentrated by lyophilization. Alternatively, urea was added to the InhA-inhibitor complex sample to 8 M, and the sample was loaded on a Sephadex G-25 column preequilibrated and eluted with 8 M urea. Two peaks in the A 280 profile of the gel filtration were observed. The first peak was denatured InhA, and the second one was the released inhibitor. The concentrated and urea-free inhibitor could be obtained by DEAE-Sepharose chromatography.
The Mn 2ϩ -mediated INH activation was carried out under the previously described condition (20) with slight modifications. InhA (400 M) was incubated with 3 mM INH, 1 mM Mn 2ϩ , and 14.7 mM NADH in 1 ml of 50 mM P i at room temperature for 5 days. The sample was passed through a Sephadex G-25 column (1 ϫ 20 cm) preequilibrated and eluted with 50 mM P i , and the gel filtration was repeated once for the protein fraction pool. The InhA-inhibitor complex in a glass tube was heated in boiling water for 40 s and then put on ice. The treated sample was transferred into a Microcon 3 centricom filter unit and spun to recover the released inhibitor in the filtrate, which was used for spectral measurements and HPLC analysis.
Assay and Time Course of Inhibitor Formation-The reactions were performed in triplicate in 4 ml of 50 mM P i containing 0.5 M KatG, 1 mM NADH, and 1 mM INH without or with 1 M Mn 2ϩ at room temperature with gentle shaking. After 30, 60, 120, and 180 min, 1 ml of the reaction mixture was withdrawn each time and freed from KatG by spinning for 20 min through two Micron 30 filters. To 0.8 ml of the filtrate, InhA was added (ranging from 30 to 60 l of a 200-M InhA stock), and the mixture was incubated for at least 10 min. The InhAinhibitor complex was isolated by two-time gel filtration on a 1 ϫ 40-cm Sephadex G-25 column. A 291 and A 326 of the isolated complex were used to calculate the amounts of the inhibitor formed using 7.1 and 7.2 mM Ϫ1 cm Ϫ1 of ⑀ 291 and ⑀ 326 , respectively, for the bound inhibitor and 26.4 mM Ϫ1 cm Ϫ1 of ⑀ 291 for InhA. The A 326 of InhA was negligible.
Mn 2ϩ -mediated INH Oxidation-The oxidation of INH to isonicotinic acid was carried out in 1 ml of 50 mM P i containing 100 M INH and 0 -1 M Mn 2ϩ at 37°C for 4 h. Aliquots (20 l) of the reaction were subject to HPLC analysis using Waters HPLC system with a Bondapak C18 HPLC column. Isonicotinic acid and INH had retention times of 4.50 and 5.56 min, respectively, when the column was eluted isocratically with 22% acetonitrile aqueous solution containing 40 mM ammonium acetate (pH 7.0). A 260 was used for the detection.
Analyses of INH Inhibitor-Inhibitor samples from the KatG-and the Mn 2ϩ -mediated activation of INH were each analyzed by HPLC using the same system as described above. The column was eluted isocratically with 10% methanol aqueous solution at a rate of 0.6 ml/ min, and A 330 was monitored. Both inhibitor samples were also analyzed by mass spectrometry. For each purified inhibitor, 10 l of the inhibitor sample with A 260 of about 0.5 was injected into a Hewlett Packard 1100 MSD mass spectrometer and carried by a mixture of 2% of acetonitrile and 98% of 0.05% trifloroacetic acid at a rate of 0.5 ml/min. The spectra were obtained by scanning the m/e range of 120 -1000 after ionization with atmospheric pressure electrospray and fragmentation voltage of 200 V. The spectra were analyzed in the negative mode.
Other Measurements-Protein concentrations were determined by the method of Lowry et al. (25) using bovine serum albumin as a standard. A Milton Roy Spectronic 3000 absorption spectrophotometer was used for absorption spectra and single-wavelength measurements. The N-terminal sequence of InhA was determined by automated Edman degradation with a gas sequencer. Automated DNA sequencing was performed using a cycle sequencing kit and an automated DNA sequencing instrument (model 373A, Applied Biosystems). Heme contents were determined by the pyridine hemochrome assay using ⑀ 418 ϭ 191.5 mM Ϫ1 cm Ϫ1 (26). Inductively coupled plasma atomic emission (27) was used to detect Mn 2ϩ in the purified KatG using a Perkin-Elmer inductively coupled plasma/5500 emission spectrometer.

RESULTS
Recombinant KatG-About 20 mg of homogeneous KatG were obtained from 6 liters of culture without isopropyl-␤-Dthiogalactopyranoside induction. The purified enzyme had an A 408 /A 280 ratio of 0.70 and, based on protein content, an extinction coefficient ⑀ 280 of 2 mg Ϫ1 cm Ϫ1 or 12.5 mM Ϫ1 cm Ϫ1 (for KatG dimer). In the determination of heme content by the pyridine hemochrome assay, the ratio of A 408 of the purified KatG over A 418 developed in the assay was found to be 0.60 Ϯ 0.04 yielding an ⑀ 408 of 115 Ϯ 8 mM Ϫ1 cm Ϫ1 for the heme. These measurements gave 1.9 Ϯ 0.1 mol heme/dimer. The KatG possesses catalase and peroxidase activities similar to those of KatG from E. coli expressing the wild type katG gene (28). The KatG catalase activity displayed a K m of 2.4 mM for H 2 O 2 and a k cat of 1.3 ϫ 10 4 s Ϫ1 , yielding a k cat /K m of 5.4 ϫ 10 6 M Ϫ1 s Ϫ1 . The specific peroxidase activity of KatG was 0.17 unit/mg under the conditions described under "Experimental Procedures." INH Activation by KatG-A limiting amount of KatG was incubated with InhA, INH, and NADH for 150 min, and the proteins in the reaction were then separated from free small molecules by gel filtration. The UV-visible spectrum of the protein sample so obtained displayed substantial absorption in the range of 315-350 nm and a smaller absorption peak at 408 nm due to the KatG-bound heme (Fig. 1, dashed curve). The 315-350 nm absorption was absent in control samples obtained the same way but without the addition of KatG (Fig. 1, dotted  curve) or INH (Fig. 1, solid curve). The extra absorption in the 315-350 nm range were, as will be shown later, associated with an InhA inhibitor formed in this INH activation. Therefore, A 320 provides a convenient indicator for following the INH activation and the inhibitor formation. The inhibitor was apparently bound by InhA very tightly. Repeated gel filtrations easily removed other small molecules such as NADH but not the inhibitor from the protein samples.
InhA-Inhibitor Complex-A series of experiments were carried out to establish the formation of an InhA-inhibitor complex and to characterize the absorption spectral properties of the free and bound inhibitor. First, the protein fraction from the KatG reaction was subject to DEAE-Sepharose chromatography to separate InhA from KatG. The InhA so obtained showed an absorption peak at 278 nm and a pronounced shoulder around 326 nm ( Fig. 2A, dashed curve), giving an A 326 /A 278 ratio of 0.16 Ϯ 0.03. Second, the inhibitor was released from InhA by heat treatment and was isolated. The free inhibitor exhibited absorption peaks at 260 and 326 nm with an A 326 / A 260 ratio of 0.255 ( Fig. 2A, solid curve). The bound inhibitor showed no significant absorption at Ն 350 nm, whereas the free inhibitor had a pronounced absorption at 350 nm. Apparently, there was a blue shift of the 326-nm peak of the inhibitor upon binding to InhA. The inhibitor was also obtained after dissociation from InhA by 8 M urea and purified by DEAE-Sepharose chromatography. The inhibitor so isolated showed an absorption spectrum identical to that of the free inhibitor obtained by the heat treatment.
Reconstitution of the complex was performed using InhA and the isolated inhibitor. Mixing equal molar of InhA and the isolated inhibitor resulted in a spectrum that was significantly different from the sum of the spectra of the two individual constituents but identical to that of the InhA-inhibitor complex. A difference spectrum ([A inhibitor ϩ A inhA ] Ϫ A inhibitor:InhA ) was obtained with peaks at 259 and 353 nm and troughs at 301 and 405 nm (Fig. 2B). The trough at 301 nm and the peak at 353 nm are consistent with the blue shift of a 326-nm peak of the free inhibitor upon binding to InhA as shown in Fig. 2A. Again, the inhibitor binding by InhA was apparently very tight. In a subsequent gel filtration of the mixture, the absorption spectrum of the reconstituted complex remained unchanged.
Inhibition of InhA Activity-As detailed under "Experimental Procedures," the InhA-inhibitor complex was isolated from a reaction solution containing InhA, NADH, INH, and KatG. Similarly, a control InhA sample was also isolated from the same reaction solution but without KatG. The control InhA, at 3.6 M, was similarly active in catalyzing the oxidation of NADH, with octenoyl CoA as a cosubstrate, in the absence and presence of KatG (Fig. 3A, E and q, respectively). No activity was observed in the assay without octenoyl CoA (Fig. 3A, ‚). In comparison with InhA, the InhA-inhibitor complex was only about 3% active in the presence of NADH and octenoyl CoA (Fig. 3A, OE). To test whether the InhA-inhibitor complex dissociated at lower concentrations in the presence of the substrates, 0.36 M of the complex and the control InhA were tested for their activities under conditions identical to that described for Fig. 3A. The rates of NADH oxidation were decreased by 10fold for both samples with the inhibition level of the diluted complex remained at 97% (Fig. 3B). No apparent increase of activity in the diluted InhA-inhibitor sample was observed after 10 min of standing. Therefore, the 3% activity of the InhA-inhibitor complex samples at 3.6 and 0.36 M was probably due to the presence of a trace amount of free InhA, and no significant dissociation was observed at 0.36 M InhA-inhibitor complex. Conservatively, the dissociation constant of the InhAinhibitor complex at 23°C could be estimated to be less than 0.4 nM. Reconstituted InhA-inhibitor complex was also inactive in the enoyl reductase assay.
The degree of inhibition increased with the increase of A 326 / A 278 of the isolated InhA-inhibitor samples and reached nearly 100% when the A 326 /A 278 was around 0.16 at which the InhA must be saturated by the inhibitor. Assuming a 1:1 stoichiom- etry for the binding of inhibitor by InhA and based on spectral differences between bound and free inhibitor, the ⑀ 326 and ⑀ 260 of the free inhibitor can be estimated to be 6900 and 27,000 M Ϫ1 cm Ϫ1 , respectively.
Identity of InhA Inhibitor-The InhA-inhibitor complex and the free inhibitor were also isolated from the Mn 2ϩ -mediated activation of INH as described under "Experimental Procedures" and compared with that generated by KatG activation. The inhibitors or their complexes with InhA from the two activation reactions had identical absorption spectra. In the HPLC analysis, the free inhibitors from both activation reactions also had the same retention time. As shown in Fig. 4 (20). Taken all analyses together, the KatG-generated inhibitor was apparently identical to the inhibitor, isonicotinic acyl NADH, from the Mn 2ϩ -mediated INH activation.
Nonrequirement of InhA for Inhibitor Production-Two possible modes for the inhibitor production were considered. First, KatG itself can catalyze the formation of the inhibitor, which then interacts with InhA. Second, InhA is involved in the addition to KatG in the inhibitor-producing reaction. To distinguish these two possibilities, INH was first incubated with NADH and KatG for 150 min in the presence and absence of InhA. The reaction mixture without InhA was freed from KatG, and InhA was added to the same level as that in the sample that contained InhA during the INH activation. After standing at room temperature for 10 min, both samples were twice passed through Sephadex G-25 columns, and the formation of the InhA-inhibitor complex was quantified by absorption measurements. We found that the KatG activation of INH produced the inhibitor at about the same level in the presence and absence of InhA.
Mn 2ϩ Is Not Essential for INH Activation by KatG-Manganese ion has been reported to be essential to the INH activation by the M. smegmatis KatG (19), but its requirement was not tested for the M. tuberculosis KatG. In this work, the time courses of the InhA inhibitor formation in reaction solutions containing M. tuberculosis KatG, NADH, and INH without and with the addition of 1 M Mn 2ϩ were compared. The assay for the inhibitor developed for such measurements was sufficiently sensitive to detect at least as low as 1 M inhibitor. As shown in Fig. 5, a time-dependent formation of the InhA inhibitor was observed over 180 min for the sample containing both KatG and 1 M Mn 2ϩ for INH activation. Importantly, the omission of Mn 2ϩ resulted in only about a 30% decreases in the amounts of inhibitor formed over the same period. As a control, activation of INH by 1 M Mn 2ϩ but without KatG generated very low levels of the inhibitor. These results indicated that Mn 2ϩ enhanced the efficiency of but was not essential to the activation of INH by the M. tuberculosis KatG.
A decisive conclusion would require the proof of a lack of significant amount of Mn 2ϩ in the so-called Mn 2ϩ -free reaction solution. In the absence of KatG and InhA, INH can be oxidized to isonicotinic acid (INA) in a Mn 2ϩ -dependent reaction. This reaction provides a test of the levels of Mn 2ϩ in our reaction solutions. INH was incubated in 50 mM P i buffer with 0 -1 M exogenously added Mn 2ϩ at 37°C for 4 h under aerobic conditions, and the reaction solutions were subject to HPLC analysis. As the Mn 2ϩ level increased from 0.1 to 1 M, increasing amounts of INH were converted to INA (Fig. 6). However, INA was formed in the 50 mM P i buffer without Mn 2ϩ addition in an amount about 30% of that with 0.1 M Mn 2ϩ added (Fig. 6). This suggested a possible maximum contamination of 0.044 M Mn 2ϩ in 50 mM P i . If this were the case, more efficient INA formation should be obtained using a higher concentration of the P i buffer. However, no increase in INA formation was observed when the P i concentration was raised from 50 to 200 mM (Fig. 6), thus arguing against any significant contamination of Mn 2ϩ in the P i buffer. Because the InhA inhibitor was formed with similar efficiencies with and without InhA during the INH activation by KatG, InhA samples could not be a source of Mn 2ϩ contamination. The KatG sample was also tested for Mn 2ϩ content. To 0.3 ml of KatG containing 0.14 mM heme, 0.3 ml of distilled 100% HNO 3 was added. The sample was gently heated until the precipitates formed upon the addition of HNO 3 disappeared. Water was then added to a total volume of 4.0 ml. This KatG sample and a 4.0-ml standard containing 10 M MnCl 2 in 10% HNO 3 were each examined by inductively coupled plasma atomic emission at 257.61 nm for the detection of Mn 2ϩ . Three independent series of measurements were made, and seven readings were recorded for each set of measurements. The averages were calculated after excluding the highest and lowest data from each set of measurements. The standard of 10 M MnCl 2 gave an emission of 0.233 Ϯ 0.001 in comparison with a blank for Ϫ0.0005 Ϯ 0.0005. Our KatG sample contained 10.5 M heme at a ratio of 1.9 hemes/KatG homodimer and was expected to produce an emission of 0.245 if KatG contained tightly bound Mn 2ϩ at a 1:1 molar ratio with the heme. Rather, the KatG sample yielded an emission of only 0.005 Ϯ 0.009, indicating no significant content of Mn 2ϩ .

DISCUSSION
An InhA-inhibitor complex can be obtained by a slow INH activation reaction that is Mn 2ϩ -mediated but KatG-independent; the identity of the bound inhibitor as isonicotinic acyl NADH has been established from the determination of the crystal structure of such a complex (20). An InhA inhibitor can also be obtained by a rapid KatG-dependent activation of INH. However, it is not clear whether these two inhibitors are the same. Moreover, the inhibitor derived from either activation process has never been isolated in solution. In this work, procedures were developed for the isolation and quantification of the InhA-inhibitor complex and the free inhibitor using either the M. tuberculosis KatG-dependent or the Mn 2ϩ -mediated process for the INH activation. The InhA-inhibitor complex obtained by the KatG-dependent activation of INH was apparently identical to its counterpart derived from the Mn 2ϩ -mediated activation with respect to absorption spectra and the lack of enoyl reductase activity. The free inhibitors obtained from these two complexes were also identical in their absorption spectra, HPLC retention times, and mass spectra. These results indicate that the same InhA inhibitor was generated by either the slow Mn 2ϩ -mediated or the fast KatG-catalyzed INH activation.
Both the free and the InhA-bound inhibitor were sufficiently stable to allow the isolation and the subsequent characterization of the inhibitor. The complexes derived from both activation processes were indistinguishable from those reconstituted from InhA and the isolated free inhibitor with respect to absorption spectra. The binding of the inhibitor to InhA was apparently very tight with a dissociation constant estimated to be lower than 0.4 nM. The free inhibitor has absorption peaks at 260 and 326 nm ( Fig. 2A). In comparison with NADH, the 326-nm peak of the inhibitor is stronger in absorptivity (⑀ 326 about 6900 M Ϫ1 cm Ϫ1 ) and in a shorter wavelength range than the NADH 340-nm peak (⑀ 340 6200 M Ϫ1 cm Ϫ1 ). Upon binding to InhA, the inhibitor 326-nm peak was blue-shifted, but the complex still showed a pronounced shoulder around 320 nm ( Fig. 2A). The characteristic absorption around 320 nm provides a very useful reporting signal for monitoring the formation and isolation of the inhibitor. The spectral changes resulting from the inhibitor binding by InhA, especially the ⌬A 353 (Fig. 2B), also provide a convenient means for investigating the binding of the inhibitor by InhA. Using this method, the KatG S315T mutant frequently encountered in INH-resistant M. tuberculosis isolates has been shown to fail to catalyze the formation of the InhA inhibitor. 2 We believe that the methodology developed in this report can also be applied to a recently identified INH target KasA (17) for further verification and investigation of the inactivation mechanism.
There are some debates about whether InhA is a primary target of activated INH in M. tuberculosis (29) One is that the selection pressure is not high enough for the mutation. M. smegmatis (which does not cause tuberculosis) and M. tuberculosis are sensitive to INH with minimum inhibitory concentrations of Ͼ 5 and 0.01-0.02 g/ml, respectively. The M. smegmatis InhA S92A mutant confers INH resistance of minimum inhibitory concentrations Ͼ 50 g/ml, whereas most of clinical INH-resistant M. tuberculosis isolates have minimum inhibitory concentrations of 1-5 g/ml. In clinical treatments, INH was absorbed by the gut to reach peak levels of 3-7 g/ml in 1-2 h after a usual oral dosage of 300 mg (30). At such an in vivo level of INH, the InhA S92A mutation of M. tuberculosis would not be effectively selected. The other possibility is that, because the InhA enzymes from M. tuberculosis and M. smegmatis have 11.7% nonidentity in their amino acid sequences, the M. tuberculosis InhA S92A mutant may be distinct from the corresponding M. smegmatis mutant in re-maining sensitive to the inhibitor. This latter possibility is under current investigation.
The M. smegmatis KatG has been shown to require Mn 2ϩ for the activation of INH for the inhibition of InhA (19), possibly by converting Mn 2ϩ to Mn 3ϩ , which in turn oxidizes INH (31). We found that, although Mn 2ϩ enhanced the INH activation by M. tuberculosis KatG, the M. tuberculosis KatG can efficiently activate INH without exogenously added Mn 2ϩ . We did not find any detectable amount of Mn 2ϩ in the P i buffer, the purified InhA, or the purified KatG used in the activation reaction. Therefore, Mn 2ϩ is apparently not essential to the activation of INH by M. tuberculosis KatG. The two KatG enzymes from M. tuberculosis and M. smegmatis are thus different in their modes of the INH activation. Such a difference might be related to the differential susceptibilities of M. tuberculosis and M. smegmatis to INH.
Previous in vitro experiments of INH activation all included InhA in the reaction. It is not clear whether InhA is required in addition to KatG for INH activation. A similar level of the InhA inhibitor was generated in the KatG reactions with or without InhA. Therefore, the simultaneous presence of InhA and KatG is not required for the inhibitor production.
KatG used in this report had a Met-Asn-Ser tripeptide fused to the first residue Met of the wild type KatG. In comparison with the wild-type KatG, the purified modified enzyme exhibited essentially the same A 408 /A 280 ratio, kinetic parameters, the ability to activate INH, and blue shift of the Soret band A 408 (32) upon INH binding (not shown). The extra peptide apparently does not significantly change the structure or function of KatG. This is in contrast with another fusion enzyme that had a Met-Glu-Phe-Val tetrapeptide fused to the second residue Pro. This latter modified KatG bound only about 0.5 heme/ dimer, in comparison with 2 heme/dimer by the wild type and our modified KatG, and thus had a much lower enzyme activity (33). A constitutive expression was adopted to slowly accumulate our enzyme while the other KatG fusion was overexpressed with an inducible system. The difference in the expression strategy could lead to the difference in the incorporation of the heme cofactor.