Mycobacterium tuberculosis Dihydrofolate Reductase Is Not a Target Relevant to the Antitubercular Activity of Isoniazid

Cloning, expression, and purification.The M. tuberculosis dfrA and katG genes were cloned as previously described (2). The plasmids of M. tuberculosis katG and dfrA were singly and doubly transformed into E. coli BL21(DE3) (EMD Bioscience, catalog no. 69387-3). The strain containing plasmids of katG and dfrA was cultured in LB-Miller medium containing 50 μg of kanamycin/ml and 50 μg of carbenicillin/ml at 37°C until the optical density (OD) at 600 nm reached 0.5. Expression of both genes was carried out by induction for 20 h at 18°C by the addition of 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside). At the same time of induction, 100 μg of INH/ml was also added to the culture. The same protocol was used for the strain containing just the dfrA plasmid. Recombinant M. tuberculosis DHFR was purified according to a previously described method (2).

DHFR and KatG enzymatic activity assay.All assays were carried out on a Cary 100 Bio Spectrophotometer at 25°C. DHFR assays are performed by monitoring the oxidation of NADPH and reduction of dihydrofolate (DHF) at 340 nm. Reactions were initiated by adding DHFR (10 nM) to assay mixtures containing NADPH (10 μM), DHF (4.5 μM), and phosphate buffer (pH 7.5, 50 mM). The KatG activity was assayed as previously described (31).

Mass spectroscopy analysis.Purified DHFR was heated for 60 s at 100°C. After the heat treatment, denatured enzyme was separated by filtration, using a Centricon (cutoff, 3 kDa). The filtrate was used for mass analysis. The matrix-assisted laser desorption ionization (MALDI) mass spectrometry experiment was carried out on an ABI Voyager-DE STR.

Crystallization of InhA in complex with INH-NAD adduct.Crystallization was accomplished by the hanging-drop vapor diffusion method. M. tuberculosis InhA in complex with INH-NAD was cocrystallized in hanging droplets containing 2 μl of protein solution at 10 mg/ml and with 2 μl of buffer (12% 2-methyl-2,4-pentanediol, 4% dimethyl sulfoxide, 0.1 M HEPES, and 0.025 M sodium citrate) at 16°C in Linbro plates against 1 ml of the same buffer. Diamond shaped protein crystals formed ∼4 days later.

T7-based expression system for mycobacteria. (i) Construction of M. smegmatis expression strain, mc²4517.pT7Pol26 was digested with BamHI to excise the RNA polymerase gene of bacteriophage T7. The 2.7-kb BamHI DNA fragment was ligated to BamHI-digested pSD26 plasmid, and the orientation was confirmed by digestion with the appropriate restriction enzymes. The DNA fragment of an acetamidase promoter fused to T7 polymerase was then cloned into pMV306, to generate pYUB1232. The expression strain mc²4517 was obtained after electroporating pYUB1232 into M. smegmatis mc²155.

(ii) Construction of expression plasmids.pYUB1062, the parental expression plasmid, was constructed by combining the NarI/SspI DNA fragment (948 bp) from pET30a plasmid and the NarI/EcoRV DNA fragment (4,026 bp) of pMV206. Shuttle vector pYUB1062 contains an oriE for replication in E. coli, oriM for replication in mycobacteria, a hygromycin cassette for antibiotic selection, a T7 promoter/terminator, and a histidine tag region for the purification of overproduced protein. The various M. tuberculosis genes were cloned into pYUB1062 using NdeI or NcoI as the upstream site and an appropriate downstream site.

(iii) Expression of INH-NAD- or INH-NADP-binding proteins from M. tuberculosis. M. smegmatis mc²4517 cells were transformed with plasmids expressing INH-NAD- or INH-NADP-binding protein of M. tuberculosis. A single bacterial colony was grown in all cases at 37°C to an A₆₀₀ of 0.6 to 0.8 in Middlebrook 7H9 broth medium supplemented with 0.2% Tween 80, 0.05% glycerol, kanamycin (20 μg/ml), and hygromycin B (100 μg/ml). Cell cultures were diluted 1:100 in 50 ml of the same liquid medium and cells were grown to mid-exponential phase (A₆₀₀ = 0.5 to 0.6) at 37°C at 120 rpm before induction with acetamide. The bacterial cultures were divided into equal volumes, and 0.2% (wt/vol) acetamide was added to the one of the aliquots (the other aliquot was used as a control). Expression of protein was determined after 6 h of induction by analyzing the cell-free protein extract by 10 to 20% SDS-PAGE.

MIC determination. M. smegmatis cells expressing the M. tuberculosis proteins were grown to an OD of 0.6 and then induced with 0.2% acetamide for ∼6 h until the OD reached ∼1.0. The cells were then diluted by 10⁵-fold and incubated in 200 μl of 7H9 medium containing INH (0, 0.6, 1.2, 4.7, 9.4, 18.6, 37.5, 75, 150, and 300 μg/ml) in a 96-well plate. The plates were incubated at 37°C and observed after 72 h. The MIC was determined as the concentration of drug at which no visible growth was observed. MICs for M. tuberculosis strains were determined by using an MTT assay (14). The concentrations of INH tested were 0, 0.015, 0.03, 0.06, 0.12, 0.24, 0.48, 0.96, 1.92, and 3.84 μg/ml.

Sequencing of clinical isolates.Six INH-resistant clinical isolates (strains 5071, 5072, 5297, 5324, 5358, and 5400) were obtained for whole-genome sequencing. The clinical isolates were collected from Mexico between 1991 and 1996 by the Laboratory of Clinical Microbiology, Instituto Nacional de Ciencias Médicasy Nutrición Salvador Zubirán, Mexico City, under an IRB-approved protocol. Five of the six strains were genotyped as belonging to SNP cluster group SCG_3b, whereas strain 5358 belongs to the M. bovis family based on spoligotyping. MICs for INH ranged from 0.2 to 8.0 μg/ml compared to 0.03 μg/ml for H37Rv.

Genomic DNA was extracted by using a CTAB (cetyltrimethylammonium bromide) protocol as previously described (11). The DNA library was constructed by using a genomic DNA sample preparation kit (Illumina). The sample was first fragmented by using a nebulization technique. Then, the double-stranded DNA fragments comprised of 3′ or 5′ overhangs were converted into blunt ends, using T4 DNA polymerase and Klenow enzyme. Klenow 3′-to-5′ exonuclease minus was used to add an “A” base to the phosphorylated 3′ blunt end of the DNA fragments so that the fragments could be ligated to the adaptors, which have a single “T” base overhang at their 3′ end. The ligated DNA was then size selected on a 2% agarose gel. DNA fragments of ∼300 bp were excised from the preparative portion of the gel. DNA was then recovered by using a Qiagen gel extraction kit and PCR amplified to produce the final DNA library. A total of 5 pmol of DNA from each strain was loaded onto a different lane of the sequencing chip (eight lanes total), and the clusters were generated on the cluster generation station of a Genome Analyzer II using the Illumina Cluster generation kit. Bacteriophage φX174 DNA was used as a control.

The sequencing reaction was run for 36 cycles (tagging, imaging, and cleavage of one terminal base at a time), and four images of each tile on the chip were taken in different wavelengths for exciting each base-specific fluorophore. Image analysis and base-calling were done by using v0.3 of the Illumina GA Pipeline software.

The 36 bp reads that were generated for each strain were mapped (aligned) against H37Rv as a reference sequence (or M. bovis in the case of strain 5358, which was the closest reference strain by spoligotyping). Apparent differences (at sites where the consensus base from overlapping reads differed from the expected base in the reference sequence), along with sites where coverage was low or observed bases were heterogeneous, were identified, and local contig-building was used to resolve them into SNPs versus indels, based on alignment to the corresponding region in the reference genome.

Overexpression of dfrA does not confer resistance to INH.The most direct evidence to support that DHFR is a target of INH comes from the observation that overexpression of dfrA in M. smegmatis conferred a 2-fold increase in resistance to INH (2). Therefore, we first reexamined whether overexpression of this protein using a strong promoter in M. smegmatis would lead to increased resistance to INH. Use of cosmids harboring a gene on a multicopy plasmid along with its native promoter might be a desired approach to overexpress putative targets. However, this can result in a failure to observe overexpression whether the expression of the protein is from a tightly regulated promoter. To eliminate the promoter-specific effect, a novel T7-based expression system was developed for the efficient expression of proteins in mycobacteria. In brief, an M. smegmatis strain (mc²4517) was generated by integrating a plasmid expressing the T7 polymerase under the acetamidase-inducible promoter into the genome at the mycobacteriophage L5 attachment site. M. tuberculosis genes were cloned under the T7 promoter on a separate episomal plasmid pYUB1062. Protein expression was induced by the addition of 0.2% acetamide. To ensure overexpression of dfrA and inhA, protein lysates were analyzed on SDS gels, and a high level of expression was observed within 6 h of induction as a distinct band corresponding to the expected size compared to the uninduced control (Fig. 1). The strains overexpressing dfrA (mc²5097) and inhA (mc²5089) were then tested for their susceptibility to INH and ETH and the results obtained are summarized in Table 1. The MIC for the inhA overexpression strain mc²5089 increased more than 60-fold to 300 μg/ml for INH, compared to the wild-type strain. In contrast to InhA, we did not observe any increase in MIC for mc²5097, the strain overexpressing DHFR, which had been reported to confer 2-fold resistance to INH when induced in M. smegmatis with an acetamidase promoter (2). Other than DHFR and InhA, 16 additional proteins from M. tuberculosis were identified in a previous study (1) that bound to an INH-NAD or INH-NADP adduct in vitro. We applied the same T7 promoter system to successfully overexpress 14 of these genes in M. smegmatis, but none of them could confer resistance to INH or ETH (Table 1). Since this analysis was done in liquid culture, we also tested the ability of the strains to grow on plates with two times the MIC for INH. As shown in Fig. 2, the M. smegmatis strain containing the overexpressed DHFR (mc²5072) failed to grow on an INH-containing plate, while the strain overexpressing InhA (mc²5071) showed complete resistance. To find out whether DHFR was indeed expressed, the strain containing the overexpressed DHFR was plated on medium containing the known DHFR inhibitor trimethoprim. Strain mc²5072 (DHFR overexpresser) was able to grow on medium containing 50 or 150 μg of trimethoprim/ml. In contrast, the parental strain mc²5069 or the strain overexpressing InhA (mc²5071) failed to grow on trimethoprim-containing plates. One possible explanation for the difference between our findings and those reported previously (2) might be that the growth inhibition experiment in the previous study was carried out at 30°C, which is not optimal for growth, whereas the growth inhibition experiments in our study were performed at 37°C. Our observation that the overexpression of M. tuberculosis InhA but not DHFR confers resistance to INH and ETH in M. smegmatis using this new expression system is consistent with InhA being the primary target of INH and ETH. Even though DHFR has been shown to bind an INH-NADP adduct in vitro, this phenomenon is not sufficient to confer INH or ETH resistance in mycobacterial cells.

• Open in new tab
• Download powerpoint

FIG. 1.

Overexpression of the proteins in M. smegmatis confirmed by SDS-PAGE analysis. Lanes: M, marker; 1, vector; 2, Rv0155 (PntAA, 40 kDa); 3, Rv1484 (InhA, 29.5 kDa); 4, Rv2763c (DfrA, 17.4 kDa); 5, Rv1996 (34.9 kDa); 6, Rv2623 (32.7 kDa); 7, Rv2766c (FabG5, 30.8 kDa); 8, Rv2858 (AldC, 50 kDa); 9, uninduced Rv2763c (DfrA); 10, induced Rv2763c (DfrA).

• Open in new tab
• Download powerpoint

FIG. 2.

Growth of the wild-type M. smegmatis strain and transformed strains overexpressing DHFR or InhA on media containing the known DHFR inhibitor trimethoprim (50 or 150 μg/ml) or INH (10 or 120 μg/ml).

A second overexpression experiment was conducted using the H37Rv strain of M. tuberculosis, which is more disease relevant, compared to M. smegmatis. The M. tuberculosis H37Rv strain was transformed with the pMV261::dfrA plasmid. The MIC of INH for the M. tuberculosis strain overexpressing dfrA is identical to that of the wild-type H37Rv strain (0.06 μg/ml). On plates with three times the MIC for INH, no growth of the M. tuberculosis strain overexpressing dfrA was observed. In contrast, the M. tuberculosis strain overexpressing inhA survived (Fig. 3), which is consistent with the previous report that the MIC for M. tuberculosis overexpressing inhA (1.0 μg/ml) increased ∼20-fold compared to the wild-type H37Rv strain (MIC = 0.06 μg/ml) (12). Therefore, similar to what we observed in our M. smegmatis experiment, overexpression of dfrA did not prevent growth inhibition of M. tuberculosis by INH, which genetically demonstrates that M. tuberculosis DHFR is not a relevant target of INH.

• Open in new tab
• Download powerpoint

FIG. 3.

Growth of M. tuberculosis H37Rv strains overexpressing DHFR or InhA on media containing INH (0.2 μg/ml) or kanamycin (100 μg/ml). Spots: A, pMV261::inhA (InhA overexpresser); B, pMV261 (plasmid only); C to F, four independent transformants of pMV261::dfrA (DHFR overexpresser).

INH-NADP adduct formation was not observed in an E. coli-based activation system coexpressing katG and dfrA.In the previous study, a synthetic INH-NADP adduct derived from INH demonstrated strong inhibition of M. tuberculosis DHFR in vitro (K_iapp = 1 nM) (2). However, the INH-NADP adduct was synthesized by using an inorganic catalyst, Mn(III). Thus, the yield of adduct generated from this approach might not truly reflect an enzyme-mediated process inside the cell. To better mimic the in vivo activation of INH, a cell-based activation system was designed to examine the KatG-catalyzed adduct formation and the inhibition of DHFR by the adduct. This E. coli-based activation system is similar to the one used previously to activate the prodrugs ETH and protionamide (PTH) (26). In this system, katG and dfrA were coexpressed in E. coli in the presence of INH to investigate whether the activated drug would inhibit DHFR. To construct this system, katG and dfrA were cotransformed into the E. coli BL21(DE3) strain and selected on 50 μg of kanamycin and carbenicillin/ml. The E. coli strain containing katG and dfrA genes was grown and induced in the presence and absence of INH, respectively. After the coexpression of both genes was confirmed by SDS-PAGE, recombinant KatG and DHFR proteins were readily purified.

Mass spectroscopy was used to determine whether DHFR purified from the experimental sample was bound with any inhibitor. We have previously applied mass spectroscopy to identify the INH-NAD, ETH-NAD, and PTH-NAD adducts bound to InhA (18, 26). Before the mass spectroscopic analysis, the purified DHFR was concentrated and then denatured, followed by filtration to separate small molecules from the denatured protein. The MALDI mass spectrum of the filtrate, which ranged between 200 and 1,200 Da, was carefully analyzed. We were not able to identify any compound that has a molecular mass corresponding to an INH-NADP adduct. The peaks shown on the spectrum most likely resulted from the cofactor NADP and some small fragments of the protein.

An enzyme assay was performed to determine the activity of purified DHFR. DHFR isolated from the experimental sample was found to be fully active (specific activity of 12 μmol mg⁻¹ min⁻¹) compared to the enzyme purified from expression in the absence of INH. Since the acyclic 4R INH-NADP adduct is extremely potent against DHFR in vitro, it would tightly bind to DHFR if the adduct is indeed generated by KatG catalysis inside the cell. However, both the activity assay result and the mass analysis indicated that no detectable amount of the INH-NADP adduct had bound to DHFR. The coexpressed KatG from the same experimental sample was purified and assayed for its activity in vitro. The specific catalase activity of isolated KatG was 17 mol mg⁻¹ min⁻¹, a finding comparable to published data (21 mol mg⁻¹ min⁻¹) (13), which confirmed that the lack of the INH-NADP adduct did not result from the absence of KatG activity. Therefore, the acyclic 4R INH-NADP adduct is not an activated INH product generated by KatG catalysis inside the E. coli cell-based system.

INH-NAD adduct was detected in the E. coli based activation system coexpressing katG and inhA.It has been shown that KatG activates INH and catalyzes the formation of an INH-NAD adduct in vitro (23, 27). In order to demonstrate that this KatG-catalyzed INH-NAD adduct formation and its inhibition of InhA can be reproduced in the E. coli-based system, both katG and inhA were transformed into E. coli and coexpressed in the presence of INH. InhA was rapidly purified by a Ni-NTA affinity column, and an in vitro enzyme assay was performed. InhA isolated from the experimental sample had <15% of the specific activity of InhA purified without the addition of INH under the same assay condition. The bound inhibitor was isolated by denaturing the InhA purified from the experimental sample, and a 1 μM concentration of the inhibitor led to complete inhibition of native InhA. The crystal structure of InhA in complex with the inhibitor was solved to 2.4-Å resolution. In the active site of InhA, an unbiased electron density map clearly indicated the presence of a modified NAD with an isonicotinic-acyl group covalently attached to the 4-position of the nicotinamide ring in a 4S configuration, which is consistent with the previously identified INH-NAD adduct (18). This is the first time that the activation of INH by KatG and the formation of the INH-NAD adduct has been demonstrated in a whole-cell environment. This confirms that our E. coli cell-based system is capable of activating the prodrug INH.

Absence of polymorphisms in dfrA in INH-resistant clinical isolates.As an alternative approach to determining whether DHFR might be a relevant target of INH, we used whole-genome sequencing to determine whether mutations in dfrA could be observed in INH-resistant clinical isolates. We sequenced the genomes of six clinical isolates that were INH resistant but known not to have relevant mutations in inhA, katG, or their promoters. The six strains were sequenced by using an Illumina Genome Analyzer II (see Materials and Methods). The depth of coverage (i.e., the number of reads overlapping each site, averaged over the whole genome) was 36.9x to 49.4x for each strain, providing high confidence in the final base calls. This method has been used successfully to identify known and novel mutations associated with resistance to INH, rifampin, pyrazinamide, kanamycin, ofloxacin, streptomycin, etc., in many other strains sequenced in our lab.

Coding regions for the 17 putative INH-NADH-binding proteins identified earlier (1) were searched for polymorphisms. The 100-bp regions upstream of the operon containing each gene were also searched for possible SNPs in the promoter region that might affect expression level. None of the six clinical isolates harbored any mutations in dfrA or its promoter (Table 2). Furthermore, few nonsynonymous mutations were found in any of the other 16 genes identified as potential proteins that can bind an INH-NAD(P)H adduct. Strain 5358 showed the most mutations, due to its evolutionary distance from the others (5358 belongs to the M. bovis family), but most of the mutations in strain 5358 were also observed in the genome for INH-sensitive M. bovis BCG, implying that they are not responsible for INH resistance. The only sites where there are potentially relevant differences (i.e., nonsynonymous and not in BCG) are a 931-bp deletion of mtn in 5400, aldC(T21A) in 5297, the a-64c promoter mutation upstream of Rv1059, the 1-bp frameshift mutation in Rv1059 in 5297, and Rv3777:P101A in 5297.

For completeness, other genes traditionally associated with INH resistance were also searched, including katG, iniC, ndh, ahpC, etc. (8), but no known mutations responsible for INH resistance could be identified (Table 3). Two strains showed mutations in katG (F129S in strain 5400 and G285V in strain 5324), and one strain has a mutation in efpA (Q513R in strain 5324). None of these mutations has been previously reported in INH-resistant strains. The katG F129S mutation is located at the dimer interface, and the katG G285V mutation is located at the mouth of the active site entrance, so their effect on INH activation is unclear. Mutations in mshA, which catalyzes the first step in mycothiol biosynthesis, have also been linked with INH and ETH resistance, but the N111S allele appears in the Erdman strain, which is as susceptible to these drugs as H37Rv (25a). Although the true cause of INH resistance in these six strains remains undetermined, the lack of mutations in dfrA and the 16 other putative INH-NAD(P)-binding proteins suggests that they are unlikely to be targets of INH. We also have sequenced the dfrA gene in six additional M. tuberculosis strains known to be INH-resistant but with no known relevant mutations. None of these had a mutation in the dfrA gene. Consistent with our findings, a recent large-scale screening of 127 INH-resistant M. tuberculosis isolates from Singapore by PCR sequencing also revealed no mutations within dfrA (9).