An Upstream Truncation of the furA-katG Operon Confers High-Level Isoniazid Resistance in a Mycobacterium tuberculosis Clinical Isolate with No Known Resistance-Associated Mutations

Abstract

Although the major causes of isoniazid (INH) resistance in Mycobacterium tuberculosis are confined to structural mutations in katG and promoter mutations in the mabA-inhA operon, a significant proportion of INH-resistant strains have unknown resistance mechanisms. Recently, we identified a high-level INH-resistant M. tuberculosis clinical isolate, GB005, with no known resistance-associated mutations. A comprehensive study was performed to investigate the molecular basis of drug resistance in this strain. Although no mutations were found throughout the katG and furA-katG intergenic region, the katG expression and the catalase activity were greatly diminished compared to those in H37Rv (P < 0.01). Northern blotting revealed that the katG transcript from the isolate was smaller than that of H37Rv. Sequencing analysis of furA and upstream genes discovered a 7.2-kb truncation extended from the 96th base preceding the initiation codon of katG. Complementation of the M. tuberculosis Δ(furA-katG) strain with katG and different portions of the truncated region identified a 134-bp upstream fragment of furA that was essential for full catalase activity and INH susceptibility in M. tuberculosis. The promoter activity of this fragment was also shown to be stronger than that of the furA-katG intergenic region (P < 0.01). Collectively, these findings demonstrate that deletion of the 134-bp furA upstream fragment is responsible for the reduction in katG expression, resulting in INH resistance in GB005. To our knowledge, this is the first report showing that deletion of the upstream region preceding the furA-katG operon causes high-level INH resistance in a clinical isolate of M. tuberculosis.

INTRODUCTION

Isoniazid (INH) has been used as a first-line chemotherapeutic and prophylactic drug for tuberculosis (TB) treatment since its introduction in 1952. However, a recent national TB survey in China reported that about 16% of the new cases and 39% of previously treated cases were resistant to INH (1). Since INH resistance was always considered the first step for development of multidrug-resistant TB (MDR-TB) (2), there is considerable interest in understanding the molecular basis of INH resistance in Mycobacterium tuberculosis, particularly for the development of a rapid diagnostic assay for identification of INH-resistant strains.

The most predominant resistance mechanism is associated with the katG gene, which encodes a mycobacterial enzyme, the catalase peroxidase (KatG), that is involved in INH activation (3). Our previous study indicated that >50% of INH-resistant M. tuberculosis isolates in Hong Kong harbored resistance-associated mutations in codon 315 of the katG gene (4). Additionally, about 8% to 30% of INH-resistant M. tuberculosis isolates were shown to carry mutations at the promoter region of the mabA-inhA operon, which induces overproduction of the drug target, InhA, and results in INH resistance via a titration mechanism (5).

For better patient management and infection control, our team previously described a systematic cascade for rapid molecular diagnosis of drug-resistant M. tuberculosis, first using single-tube nested real-time PCR for detection of M. tuberculosis DNA directly from respiratory specimens (6), followed by identification of INH, rifampin (RIF), and olfoxacin (OFX) resistance-associated mutations by PCR sequencing (7, 8) or by the high-resolution melting (HRM) test (9, 10). For INH resistance, we further developed a multiplex allele-specific PCR (MAS-PCR) assay targeting two hot spot mutations (codon 315 of the katG gene and the 15th nucleotide preceding the mabA-inhA operon) (11). The assay has been implemented for routine diagnostic service in our hospital since 2011 and has successfully identified about 80% of the INH-resistant strains in our region (11). Consistent with studies reported elsewhere (12, 13), about 20% of the phenotypically confirmed INH-resistant M. tuberculosis isolates did not carry these two mutations and, therefore, were missed by the current assays. Only a few (<1%) of them were found to harbor mutations in other genetic regions, such as the structural region of mabA-inhA and katG, that have been validated to be associated with INH resistance in previous allelic-exchange experiments (3, 5, 14), whereas the resistance mechanism of the remaining isolates remained unknown. Previous studies suggested that in addition to chromosomal mutation, INH resistance may be due to epigenetic mechanisms such as efflux pump-driven resistance (15, 16).

Recently, we identified an M. tuberculosis strain, known as GB005, with high-level INH resistance (8 μg/ml) without the common katG315 and inhA promoter mutations. A complete sequencing analyses of all candidate resistance genes, including the entire katG gene, furA-katG intergenic region, mabA-inhA operon, kasA, ndh, and dfrA, revealed no known INH resistance-associated mutations. A comprehensive study of this isolate was therefore performed to unveil its resistance mechanism against INH.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The M. tuberculosis clinical strain GB005 was obtained from a sputum specimen of a patient with relapsed pulmonary tuberculosis at Queen Mary Hospital, Hong Kong, in 2012. Laboratory strain H37Rv was used as the reference control for the drug susceptibility test, the semiquantitative catalase assay, real-time PCR quantification of katG expression study, and Northern blotting of the katG transcript. Additionally, an INH-resistant clinical isolate, GA031, with deletion of the entire furA-katG operon (from position 2,153,201 to 2,156,995; GenBank accession no. NC_000962) was used as a host for transformation experiments to determine the promoter activities of different katG upstream regions and their associations with INH resistance in M. tuberculosis. Unless otherwise specified, all M. tuberculosis strains were isolated from Löwenstein-Jensen (LJ) medium (bioMérieux, France) and were then subcultured to Middlebrook 7H9 broth supplemented with oleic acid-albumin-dextrose (OADC) in a shaking incubator at 37°C until they reached the mid-log growth phase (optical density [OD], 0.6 to 0.8). The Escherichia coli strain and plasmids used in this study are listed in Table 1.

TABLE 1.

E. coli strain and plasmids used in this study

E. coli strain or plasmid	Genotype or description	Source or reference
Strain DH5α	fhuA2 lac (del)U169 phoA glnV44 ϕ80′ lacZ (del)M15 gyrA96 recA1 relA1 endA1 thi-1 hsdR17	ATCC, United Kingdom
Plasmids
pOLYG	E. coli-Mycobacterium shuttle vector; nonexpressive; pAL5000 origin of replication; Hygr	6
pOLYG::katG-1	pOLGY carrying H37Rv katG gene only	This study
pOLYG::katG-2	pOLGY carrying H37Rv katG gene plus 96-bp upstream region (katG plus 38-bp intergenic region plus 58-bp 3′ end of furA)	This study
pOLYG::katG-3	pOLGY carrying H37Rv katG gene plus 490-bp upstream region (entire furA-katG operon)	This study
pOLYG::katG-4	pOLGY carrying H37Rv katG gene plus 616-bp upstream region (entire furA-katG operon plus 134-bp upstream region)	This study
pOLYG::katG-5	pOLGY carrying H37Rv katG gene plus 1,369-bp upstream region (entire furA-katG operon plus Rv1910c plus 98-bp 3′ end of lppC)	This study
pOLYG::katG-GB005	pOLGY carrying GB005 katG gene plus 96-bp upstream region	This study
pEGFP-N1	pEGFP-N1 carries a red-shifted variant of wild-type GFP which has brighter fluorescence and higher expression; Kanr	Clontech, USA
pOLYG::egfp	pOLYG carrying an egfp gene without promoter region; serving as a null control for normalization of the GFP study	This study
pOLYG::furA−134-egfp	pOLYG carrying a 134-bp upstream fragment of furA preceding an egfp gene	This study
pOLYG::katG−96-egfp	pOLYG carrying a 96-bp upstream fragment of katG preceding an egfp gene	This study
pOLYG::katG−616-egfp	pOLYG carrying a 616-bp upstream fragment of katG preceding an egfp gene	This study

Phenotypic and genotypic drug susceptibility tests.

Phenotypic drug susceptibility for clinical isolates was determined by the 1% standard proportion method according to the Clinical and Laboratory Standards Institute (CLSI) guideline (17). For genotypic testing, the presence of mutations inside the rifampin resistance-determining region (RRDR) of the rpoB gene and the quinolone resistance-determining region (QRDR) of the gyrA gene were detected by our in-house PCR-sequencing assays described previously (7, 8), whereas the katG315 mutation and mabA-inhA −15 promoter mutation associated with INH resistance were identified by our MAS-PCR assays (11).

Complete sequencing analysis of candidate genes associated with INH resistance.

The INH-resistant M. tuberculosis clinical isolate GB005, without katG315 and mabA-inhA −15 mutations, was subjected to complete sequencing analysis of candidate genes associated with INH resistance, including the entire katG gene, furA-katG intergenic region, mabA-inhA operon, kasA gene, ndh gene, and dfrA gene. The PCR mixture consisted of 1× PCR buffer (Applied Biosystems, USA), 1.5 mM MgCl₂ (Applied Biosystems), 0.2 mM (each) deoxynucleoside triphosphate (dNTP; Fermentas, USA), 5% dimethyl sulfoxide (DMSO; Stratagene, USA), 0.35 μM forward and reverse primers for each gene (see Table S1 in the supplemental material), and 2.5 U AmpliTaq Gold (Applied Biosystems). The PCR was carried out under the following conditions: initial denaturation at 96°C for 8 min, 5 cycles at 95°C for 1 min, 65°C for 1 min, and 72°C for 3 min, 5 cycles at 95°C for 1 min, 63°C for 40 s, and 72°C for 3 min, and 30 cycles at 94°C for 1 min, 61°C for 30 s, and 72°C for 3 min, followed by a final 10-min extension at 72°C. The amplicons were then subjected to cycle sequencing using an ABI 3130 genetic analyzer (Applied Biosystems). All the sequencing primers are listed in Table S1 in the supplemental material. The nucleotide numbering and amino acid numbering were based on the H37Rv reference sequence (GenBank accession no. NC_000962).

Semiquantitative catalase assay.

A suspension of M. tuberculosis isolate was prepared by adding freshly grown LJ culture into Milli-Q water and vortexing with glass beads to a density equivalent to McFarland 1.0 standard. One hundred microliters of the suspension was added into an LJ medium tube with a horizontal surface and incubated at 37°C for 2 weeks. One milliliter of a 1:2 mixture of Tween 20 and 30% hydrogen peroxide was then added, and the mixture was incubated at room temperature for 5 min. The height of the bubbles (in millimeters) was then measured (12).

Extraction of total RNA.

Mid-log-phase M. tuberculosis culture was incubated with 5 M guanidium thiocyanate solution for 1 h with agitation, followed by centrifugation at 3,500 rpm for 10 min at 4°C. Harvested bacterial cells were resuspended in 1 ml TRIzol (Invitrogen, USA) and transferred to a Lysing matrix B tube (MP Biomedicals, USA). The mycobacterial cell wall was mechanically disrupted by the matrix beads under the vigorous shaking carried out by the FastPrep instrument (MP Biomedicals) for 45 s at a setting of 6.5. Bacterial lysates were treated with chloroform, and the aqueous phase was collected. Extracted total RNA was then cleaned and extracted by DNase I (New England BioLabs, USA) and an RNeasy MiniElute cleanup kit (Qiagen, Germany) for complete removal of residual genomic DNA.

Quantification of katG transcripts using two-step real-time PCR.

For each isolate, 100 ng of total RNA was reverse transcribed into cDNA using SuperScript III reverse transcriptase (Invitrogen, USA) with 60 μM random hexamer according to the manufacturer's instructions. Two quantitative real-time PCRs (qPCRs) were then performed on each cDNA sample using the LightCycler TaqMan master kit (Roche, USA). One reaction was for measurement of katG transcripts using forward primer katGFq (5′-GCGTCGCAGGAACAAACC-3′), reverse primer katGRq (5′-TGCCCTTTCCGAGGTAGTTTC-3′), and TaqMan probe katGPq (5′-FAM-AATCCTTTGCCGTGCTGGAGCC-BHQ1-3′ [FAM, 6-carboxyfluoresein; BHQ1, black hole quencher 1]). The other qPCR targeted the rrs gene for normalization with the following primer/probe set: forward primer RRS-taqfwd (5′-TCCCGGGCCTTGTACACA-3′), reverse primer RRS-taqrev (5′-CCACTGGCTTCGGGTGTTA-3′), and TaqMan probe RRS-taqprobe (5′-FAM-CGCCCGTCACGTCATGAAAGTCG-BHQ1-3′). Both qPCRs were performed in the same run with the following thermal cycling conditions: initial heat activation at 95°C for 10 min followed by 50 cycles of 95°C for 10 s, 50°C for 20 s, and 72°C for 15 s, with an acquisition of fluorescent signal at 530 nm.

For each M. tuberculosis isolate, the test was repeated 3 times by using cDNA prepared at different occasions. The original total RNA samples were also subjected to real-time PCR amplification to confirm that there was no residual DNA. The final results were presented as a relative expression ratio (katG/rrs ratio), calculated by dividing the average reading of the katG transcript by that of the rrs transcript.

Northern blotting to visualize the size of katG transcript.

Prior to Northern blotting, a digoxigenin (DIG)-labeled probe complementary to the katG transcript was synthesized by PCR amplification of a 594-bp katG fragment using forward primer katG-DIG-probeF (5′-GGCACCTACCGCATCCAC-3′) and reverse primer katG-DIG-probeR (5′-GGTTCCGGTGCCATACGA-3′), followed by DIG labeling using the DIG High Prime kit (Roche).

The Northern blotting procedures followed the formaldehyde/formamide protocol (18). In brief, at least 5 μg of denatured mycobacterial RNA was loaded onto the morpholinepropanesulfonic acid (MOPS)/formaldehyde gel and separated by electrophoresis at 80 V for 45 min using 1× MOPS buffer as the electrophoresis buffer. The gel was subsequently washed with diethyl pyrocarbonate (DEPC)-treated Milli-Q water and saline-sodium citrate (SSC) buffer before the RNA was transferred onto a nylon membrane, Hybond-N+ (GE Health Care, USA). UV cross-linking (70,000 μJ/cm²) was used to fix the RNA onto the membrane, which was then incubated with DIG Easy Hyb hybridization solution (Roche) containing 125 ng of denatured DIG-labeled katG probe at 60°C for 18 h with agitation. After washing with SSC buffer, the membrane was incubated with 75 mU/ml antidigoxigenin conjugated with alkaline phosphatase at room temperature for 30 min. The chemiluminescent signal was developed after the addition of chloro-5-substituted adamantyl-1,2-dioxetane phosphate (CSPD) solution (Roche). The signal was finally captured by radiograph developed on Lumi-Film chemiluminescent detection film (Roche) after a 25-min exposure.

Determination of the size and location of the truncated region.

Several PCRs targeting upstream genetic regions of katG were performed on GB005 to determine the size and location of the truncation. The PCR mixtures consisted of 1× PCR buffer (Applied Biosystems), 1.5 mM MgCl₂ (Applied Biosystems), 0.2 mM (each) dNTP (Fermentas), 4% DMSO (Stratagene), 0.5 μM forward and reverse primers (see Table S2 in the supplemental material), and 2.5 U AmpliTaq Gold (Applied Biosystems). The temperature profiles included initial heat activation at 96°C for 8 min, 40 cycles at 95°C for 1 min, 65°C for 1 min, and 72°C for 2 min, and final elongation at 72°C for 10 min.

If the target regions were located inside the truncation, no PCR amplification was obtained. Conversely, successful PCRs implicated the ends of a truncated region. The genetic nature of the truncated region was further determined by Sanger sequencing using the appropriate primers (see Table S2 in the supplemental material).

Transformation of katG gene with differential upstream regions into catalase-negative INH-resistant M. tuberculosis.

DNA fragments containing the katG gene with different upstream regions were amplified from wild-type strain H37Rv and clinical strain GB005 using cloning primers with the XbaI and HindIII restriction sites (see Table S3 in the supplemental material). The amplified DNA fragments were digested with XbaI and HindIII (New England BioLabs) and then cloned to the corresponding sites of the promoterless Escherichia coli-mycobacteria shuttle vector, pOLYG, resulting in the plasmids listed in Table 1 (pOLYG::katG-1 to pOLYG::katG-5 and pOLYG::katG-GB005). The plasmids were then transformed into DH5α for amplification. Finally, a total of 1 μg plasmid was electroporated (2.5 kV, 25 μF, 1,000 Ω) into a catalase-negative INH-resistant M. tuberculosis strain, GA031, using a gene pulser (Bio-Rad, USA). The cells were immediately resuspended in 7H9 broth with OADC, incubated for 4 to 8 h at 37°C, and plated onto 7H10 agar containing 50 μg/ml hygromycin (7). The INH MIC and the catalase activities of the transformants were determined as described previously (12, 17).

Verification of promoter activities of furA or katG upstream regions with green fluorescent protein (GFP).

Three katG upstream fragments of different lengths from H37Rv, including positions −134 to −1 upstream of the furA initiation codon (termed furA⁻¹³⁴), positions −96 to −1 upstream of the katG initiation codon (termed katG⁻⁹⁶), and position −134 upstream of furA to position −1 upstream of katG (termed katG⁻⁶¹⁶), were amplified by PCR with the corresponding cloning primer sets (see Table S3 in the supplemental material). The PCR products were digested with BamHI and XbaI (New England BioLabs) and ligated to the pOLYG vector digested with the same restriction enzymes. Subsequently, the egfp gene was amplified from a vector plasmid, pEGFP-N1 (Clontech, USA), using corresponding cloning primers (see Table S3), digested with NdeI and HindIII, and cloned into the corresponding sites of the intact promoterless pOLYG vector (null control) and the 3 plasmids carrying different furA or katG upstream regions. The resulting plasmids (Table 1) were then transformed first to E. coli DH5α for amplification, followed by M. tuberculosis strain GA031 as described above. The M. tuberculosis transformants were cultivated in 7H9 broth supplemented with OADC in a shaking incubator at 37°C until they reached the mid-log growth phase (OD, 0.6 to 0.8). The green fluorescent signal of the transformants was quantified by a DTX800 multichannel detector (Beckman Coulter, USA).

Statistical analysis.

Student's t test was used to compare the katG expression level and the GFP signal obtained from different M. tuberculosis isolates or transformants.

RESULTS

Patient details.

Retrospective analysis of the treatment history indicated that the patient, a 56-year-old female, was a recent immigrant from mainland China who had a history of pulmonary tuberculosis. She had incomplete treatment with first-line antituberculosis agents and was lost to follow-up in 2009. She was diagnosed with MDR-TB in March 2012 in Hong Kong, followed by 1 year of second-line antituberculosis treatment with ethionamide, pyrazinamide, levofloxacin, and cycloserine. The patient had a good prognosis, and sputum cultures yielded no growth of mycobacteria in subsequent samples over 1 year. GB005 was recovered from a sputum specimen when her disease relapsed in 2012.

Drug susceptibility profiles of M. tuberculosis isolates.

The results of phenotypic and genotypic drug susceptibility testing for M. tuberculosis isolates are listed in Table 2. The conventional agar proportional method indicated that GB005 was resistant to both INH and RIF and was therefore considered an MDR isolate. The RRDR-rpoB PCR-sequencing assay identified a RIF resistance-associated mutation, S531L, in the rpoB of GB005, which was concordant with the phenotypic result. However, neither the katG315 nor the mabA-inhA −15 promoter mutation was detected by our in-house MAS-PCR assays. The INH MIC value was 8 μg/ml, which was considered high-level resistance. A complete sequencing analysis of all candidate genes associated with INH resistance was then performed. With the exception of the missense mutations R463L in katG and G312S in kasA, which are well documented to be unrelated to INH resistance in previous studies (12, 19), no known resistance-associated mutation was found in these genes, including the entire katG structural region and furA-katG intergenic region.

TABLE 2.

Phenotypic and genotypic characteristics of M. tuberculosis isolates used in this study

Isolate	Phenotypic drug susceptibility testinga	Catalase activity (mm)	INH MIC (μg/ml)	Genotypic drug susceptibility testing
				rpoB	mabA-inhA operon	kasA	ndh	dfrA	katG	furA-katG intergenic region	furA	furA-Rv1910c intergenic region	Rv1910c and upstream
Clinical
H37Rv	RIF-S, INH-S, EMB-S, PZA-S, STR-S, OFX-S	45	<0.2	wtb	wt	wt	wt	wt	wt	wt	wt	wt	wt
GB005	RIF-R, INH-R, EMB-S, PZA-S, STR-R, OFX-S	3	8	S531L	wt	G312S	wt	wt	R463L	wt	delc (1–386 bp)	del	del
GA031	RIF-S, INH-R, EMB-S, PZA-S, STR-S, OFX-S	0	>256	wt	wt	wt	wt	wt	del	del	del	del	del
Transformants (host strain/plasmid)
GA031/pOLYG::katG-1	ND	0	>256			ND			wt	del	del	del	del
GA031/pOLYG::katG-2	ND	5	8			ND			wt	wt	del (1–386 bp)	del	del
GA031/pOLYG::katG-3	ND	3	8			ND			wt	wt	wt	del	del
GA031/pOLYG::katG-4	ND	>90	<0.2			ND			wt	wt	wt	wt	del
GA031/pOLYG::katG-5	ND	>90	<0.2			ND			wt	wt	wt	wt	wt
GA031/pOLYG::katG-GB005	ND	7	8			ND			R463L	wt	del (1–386 bp)	del	del

^aS, susceptible; R, resistant; ND, none done.

^bwt, wild type.

^cdel, deletion.

Semiquantitative catalase assay.

INH-susceptible strain H37Rv, which produced intact and fully functional KatG, was used as a positive control for the catalase activity assay. After the addition of H₂O₂, GB005 produced bubbles 3 mm in height, which were much smaller than those of H37Rv (45 mm) but stronger than those of the KatG-negative strain GA031 (0 mm) (Table 2), indicating that the catalase activity of GB005 was diminished but not completely lost. This inspired us to further study the katG expression level of the isolate.

Characterization of katG transcripts from GB005.

Two-step RT real-time PCR was used to assess and compare the katG expression level among H37Rv, GB005, and GA031. No katG transcript was detected from the known katG deletion strain GA031, whereas GB005 showed positive expression of katG; however, the expression level was almost 60-fold lower than that of H37Rv (P < 0.01) (Fig. 1), indicating that katG expression was downregulated in GB005.

FIG 1.

Real-time PCR analysis of katG transcript level in M. tuberculosis H37Rv, GB005, and GA031. Levels of katG transcript were normalized relative to those of rrs transcript. Values are means ± standard errors of the means.

The sizes of the katG transcripts were visualized by the Northern blot technique. According to Pym et al. (20), katG can be cotranscribed with its upstream gene, furA, in wild-type M. tuberculosis isolates, and the expected size of the transcript is about 2,703 bp. Compared to an RNA molecular weight marker (Fig. 2), we found that the katG transcript from H37Rv was about 2,700 to 2,800 bp, which matched with the expected size. Interestingly, the size of the GB005 katG transcript was much smaller and had only 2,200 to 2,300 bp. This resembled a single gene the size of katG (2,223 bp). This finding suggested that katG is likely transcribed alone instead of cotranscribed with furA in GB005.

FIG 2.

Northern blot analysis of katG transcript from H37Rv and GB005.

Identification of a 7.2-kb truncation encompassing furA and its upstream genes.

Further investigation of the genetic makeup of katG and its upstream region was then performed on GB005. Up to the 96th nucleotide preceding the katG initiation codon, which included the entire furA-katG intergenic region plus the 58 bp of the 3′ end of furA, was detectable and had no mutation. However, the remaining region (386 bp) of furA together with the 7 upstream genes, including Rv1910c, lppC, fadB5, Rv1913, Rv1914c, aceAa, and aceAb, were absent in GB005. Positive PCR amplification reappeared from the 8th upstream gene, PPE34 (Fig. 3a). A PCR-sequencing assay with the primers complementary to katG and PPE34 showed the truncation was about 7.2 kb, extending from position 2,156,203 to position 2,163,397 (GenBank accession no. NC_000962). This was predicted to result in the loss of 386 bp of the 5′ region of furA; the entire region Rv1910c, lppC, fadB5, Rv1913, Rv1914c, aceAa, and aceAb; and 468 bp of the 3′ region of PPE34. The schematic diagram of the gene structure of the truncated region is shown in Fig. 3b.

FIG 3.

Characterization of the 7.2-kb truncation located upstream of katG in GB005. (a) The results of PCR amplification of katG and its 13 consecutive upstream genes on H37Rv and GB005. katG was positive for both isolates. furA to AceAb were positive for H37Rv but negative for GB005. PPE34 to FadD31 were positive for both isolates again. Lanes M, DNA size markers. (b) Map of the truncated region in GB005. Bold arrows indicate the open reading frames annotated in the H37Rv reference sequence (GenBank accession number NC_000962). The light-gray bold arrows correspond to the truncated regions, with the end sequences and H37Rv genome coordinates given below. Underlined sequences are possible substrates for recombination.

Association of different katG upstream regions with INH susceptibility and catalase activity.

To determine the essentiality of the truncation in INH susceptibility in M. tuberculosis, katG with differential sizes of upstream regions was cloned and transformed into the KatG-negative strain GA031. The resulting MIC values and catalase activities of the transformants are listed in Table 2. The transformation of katG with no promoter region (pOLYG::katG-1) resulted in no improvement in MIC and catalase activity (MIC, >256 μg/ml; catalase, 0 mm) (Fig. 4A), whereas introduction of katG with a 96-bp upstream region (pOLYG::katG-2 and pOLYG::katG-GB005) into GA031 successfully reduced the MIC from 256 μg/ml to 8 μg/ml and increased the catalase activity to 5 to 7 mm (Fig. 4B). There were no further changes in the MIC and catalase activity until the insert was extended to include the 134-bp upstream region of the furA (pOLYG::katG-4), which completely restored the wild-type phenotypes (MIC, <0.2 μg/ml; catalase, >90 mm) for GA031 (Fig. 4C). No additional enhancement of catalase activity or INH susceptibility was observed by transforming further upstream regions (pOLYG::katG-5) into GA031.

FIG 4.

MIC results, catalase activity, and schematic diagram of the cloned inserts of the transformants (left to right). N, undiluted bacterial suspension (∼1 × 10⁵ CFU/ml) on a drug-free agar well; 1:99, 100-fold-diluted bacterial suspension on drug-free agar well; 0.2, 0.5, 1, 2…1,024, media supplemented with 1, 2…1,024 μg/ml of INH. (A) GA031 transformed with pOLYG::katG-1 (GA031/pOLYG::katG-1). (B) GA031 transformed with pOLYG::katG-2 or pOLYG::katG-GB005 (GA031/pOLYG::katG-2 or GA031/pOLYG::katG-GB005). (C) GA031 transformed with pOLYG::katG-4 (GA031/pOLYG::katG-4).

Comparison of promoter activities of the furA or katG upstream region by GFP assay.

The average green fluorescent signals of the transformants with plasmids containing DNA fragments furA⁻¹³⁴, katG⁻⁹⁶, and katG⁻⁶¹⁶ were 98,886, 43,497, and 75,911 units, respectively. To eliminate the influence of background fluorescence, the fluorescent signals of the three transformants were normalized by the strain transformed with pOLYG::egfp. The normalized fluorescent signals are shown in Fig. 5. The results indicated that both furA⁻¹³⁴ and katG⁻⁹⁶ had promoter activity and were capable of driving the expression the gfp gene. The promoter activity of fragment furA⁻¹³⁴, however, was significantly stronger than that of fragment katG⁻⁹⁶ (P < 0.01). Interestingly, fragment katG⁻⁶¹⁶, which contained both furA⁻¹³⁴ and katG⁻⁹⁶, resulted in a lower fluorescent signal than that of furA⁻¹³⁴ alone. Although the difference was not statistically significant (P = 0.128), the result indicated that the coexistence of furA⁻¹³⁴ and katG⁻⁹⁶ did not have a synergistic effect on gene expression.

FIG 5.

Analysis of promoter activities of three upstream fragments, furA⁻¹³⁴, katG⁻⁹⁶, and katG⁻⁶¹⁶, in a GFP expression assay. The fluorescent signal of GA031 transformed with the upstream fragments preceding the egfp gene were normalized with the null control (GA031 transformed with pOLYG::egfp). The normalized fluorescent signals of the transformants with furA⁻¹³⁴, katG⁻⁹⁶, and katG⁻⁶¹⁶ were 3.55, 1.55, and 2.73, respectively.

DISCUSSION

Although the major cause of INH resistance in M. tuberculosis is confined to mutation(s) in katG and the promoter region of the mabA-inhA operon, there are still a significant number of INH-resistant clinical isolates with unknown resistance mechanisms, limiting the sensitivities of existing molecular assays for rapid diagnosis of INH-resistant M. tuberculosis isolates. Mutations in other genes, such as kasA, dfrA, ndh, and mshC, have been found in INH-resistant clinical isolates (12, 21–25), but their contributions to INH resistance have not been confirmed by allelic-exchange studies. Using transformation experiments, our team recently validated the association of two novel mutations outside the RRDR of rpoB with RIF resistance in M. tuberculosis (7). With a similar approach, we extended our study to investigate the association of an upstream truncation of the furA-katG operon with INH resistance in a M. tuberculosis clinical isolate.

Although no resistance-associated mutation was found throughout the entire katG gene, significantly lower katG expression with diminished catalase activity was detected in an INH-resistant clinical isolate, GB005. The association of downregulation of katG with INH resistance in M. tuberculosis was described by Ando et al. (26), who reported that base substitutions in the furA-katG intergenic region caused decreased katG expression and conferred low-level INH resistance (<2 μg/ml). In contrast, GB005 did not harbor these mutations and had a high level of INH resistance. Interestingly, a 7.2-kb truncation encompassing two-thirds of the furA gene and 7 additional upstream genes was identified in this study. To validate its role in INH resistance, katG with different portions of truncated regions was introduced into the M. tuberculosis Δ(furA-katG) strain GA031. Transformation of GA031 with katG and the 96-bp upstream fragment from GB005 successfully reproduced the phenotype (MIC 8 μg/ml; catalase, 5 mm), indicating that the 96-bp DNA fragment harbored a promoter sequence which was responsible for katG expression, although the level was not sufficient to restore the transformant into a wild-type strain. Extension of the transformed insert to encompass the entire furA gene did not result in improvements in the MIC value and catalase activity, revealing no additional promoter inside the furA gene. However, with the addition of a 134-bp upstream fragment of the furA-katG operon, the transformant became fully susceptible to INH (MIC, <0.2 μg/ml), and the catalase activity was completely recovered (>90 mm). This result indicated that the 134-bp upstream fragment has strong promoter activity that was capable of expressing the katG gene located 616-bp downstream from it, and the expression level of katG was able to fully compensate for the loss of KatG activity in the Δ(furA-katG) strain. The two-step compensation of catalase activity and INH susceptibility demonstrated that there are two regulatory regions responsible for katG expression. The hypothesis is consolidated by cloning both putative promoter sequences upstream to a gfp gene. Both fragments were able to express the gfp gene to produce a sustainable green fluorescent signal. The furA upstream fragment demonstrated significantly stronger promoter activity than did the furA-katG intergenic region. Similarly, Pym et al. (20) reported that the furA upstream region was responsible for the cotranscription of furA and katG in wild-type M. tuberculosis isolates. However, no promoter activity was demonstrated in the furA-katG intergenic region in their study. The discrepancy may have been due to a difference in methodology for the measurement of catalase activity. We studied the direct catalase activities of the transformants, whereas Pym et al. measured the catalase activity from protein extract in polyacrylamide gel, which may not be sensitive to detection of low-level catalase activity. In our Northern blotting, only the large transcript encompassing both furA and katG, but not the one with katG alone, was detected from H37Rv, revealing that katG expression from a downstream promoter (furA-katG intergenic region) was relatively low and was masked by the transcriptional activity of an upstream promoter (furA upstream region) in wild-type M. tuberculosis isolates. Deletion of the furA upstream region would result in a severe reduction in katG expression and failure of complete INH activation, thereby conferring INH resistance in GB005.

To our knowledge, this is the first report showing that deletion of the upstream region preceding the furA-katG operon caused high-level INH resistance in a clinical isolate of M. tuberculosis. Although little attention has been paid to the association of katG expression with INH susceptibility in M. tuberculosis, the results of this study and those of Ando et al. (26) showed that mutations in regulatory regions of katG can be responsible for INH resistance. Genotypic studies of both regions should be performed on more clinical isolates, especially INH-resistant strains with no known mutations, to determine their prevalence in INH-resistant M. tuberculosis strains.