ABSTRACT
Elucidation of the molecular basis of isoniazid (INH) resistance in Mycobacterium tuberculosis has led to the development of different genotypic approaches for the rapid detection of INH resistance in clinical isolates. Mutations in katG, in particular the S315T substitution, are responsible for INH resistance in a large proportion of tuberculosis cases. However, the frequency of the katG S315T substitution varies with population samples. In this study, 52 epidemiologically unrelated clinical INH-resistant M. tuberculosis isolates collected in Australia were screened for mutations at katG codon 315 and the fabG1-inhA regulatory region. Importantly, 52 INH-sensitive isolates, selected to reflect the geographic and genotypic diversity of the isolates, were also included for comparison. The katG S315T substitution and fabG1-inhA −15 C-to-T mutation were identified in 34 and 13 of the 52 INH-resistant isolates, respectively, and none of the INH-sensitive isolates. Three novel katG mutations, D117A, M257I, and G491C, were identified in three INH-resistant strains with a wild-type katG codon 315, fabG1-inhA regulatory region, and inhA structural gene. When analyzed for possible associations between resistance mechanisms, resistance phenotype, and genotypic groups, it was found that neither the katG S315T nor fabG1-inhA −15 C-to-T mutation clustered with any one genotypic group, but that the −15 C-to-T substitution was associated with isolates with intermediate INH resistance and isolates coresistant to ethionamide. In total, 90.4% of unrelated INH-resistant isolates could be identified by analysis of just two loci: katG315 and the fabG1-inhA regulatory region.
Isoniazid (INH) is one of the most effective antimycobacterial agents available for the treatment of tuberculosis (TB). However, soon after its introduction in 1952, strains of Mycobacterium tuberculosis, the causative agent of TB, resistant to INH emerged (17). INH inhibits the synthesis of mycolic acids, which are integral components of the mycobacterial cell wall, by inhibiting a key enzyme involved in mycolic acid biosynthesis, InhA (3). INH is a prodrug (11, 12) that is converted to its active form in vivo by the katG-encoded M. tuberculosis catalase-peroxidase KatG (35). Resistance to INH is predominantly associated with mutations in katG, particularly at codon 315. The genes fabG1 and inhA both encode mycolic acid biosynthetic enzymes and are cotranscribed (4). Mutations within the inhA structural gene and the fabG1-inhA regulatory region have also been identified and associated with INH resistance (3, 7, 13). There remain a proportion of isolates resistant to INH that are not accounted for by mutations in either katG or inhA. The frequency distributions of INH resistance-associated mutations have been shown to vary according to geographic region, population, and genotype (20, 22).
The introduction of molecular epidemiological typing methods such as IS6110 restriction fragment length polymorphism, spoligotyping, and more recently mycobacterial interspersed repetitive units (MIRUs), has led to the identification and characterization of groups of related M. tuberculosis strains described as families, lineages, or clades, including the Beijing, Haarlem, East-African-Indian (EAI), Latino-American and Mediterranean (LAM), X, T, and Central Asian (CAS) families (9). These families are likely to have originally been endemic within specific geographical areas, and while some remain circumscribed to the particular regions, others such as the Beijing family have become widely disseminated (27). An alternative scheme for classifying isolates is based on the likely evolutionary pathway of the M. tuberculosis complex. According to this scheme, strains of M. tuberculosis fall into three major genetic groups: genetic group 1 is evolutionarily the oldest and the precursor of groups 2 and 3 (28). More recent phylogenetic analysis has led to the proposal of “ancestral” and “modern” strains of M. tuberculosis (6). Genetic group 1 comprises both ancestral and modern strains, including isolates representative of the Beijing, CAS, and EAI families, whereas groups 2 and 3 contain only modern strains, including isolates belonging to the Haarlem, LAM, and X families (6).
Evaluation of the frequency distribution of resistance-associated mutations in different clinical settings is a prerequisite for the large-scale implementation of genotypic approaches for the rapid detection of INH resistance. This paper presents the results of a study of 52 INH-resistant (INHr) and 52 INH-susceptible (INHs) epidemiologically unrelated M. tuberculosis clinical strains of diverse geographic origins and genotypes isolated in Victoria, Australia. The aims of the study were twofold: first, to investigate the prevalence of mutations in the two most commonly reported loci associated with INH resistance, katG codon 315 and the fabG1-inhA regulatory region; and second, to analyze whether INH resistance and/or mutations in these regions were associated with a particular geographic origin, genotypic group, or drug resistance profile.
MATERIALS AND METHODS
Mycobacterial isolates.The Victorian Infectious Diseases Reference Laboratory (VIDRL) serves as the reference laboratory for all mycobacterial infections in Victoria, Australia. Six hundred forty-five new laboratory-confirmed cases of TB caused by M. tuberculosis were collected during the study period (2001 to 2003). All isolates were tested for susceptibility to first-line drugs INH (0.1 μg/ml), ethambutol (EMB) (5 μg/ml), and rifampin (RIF) (1 μg/ml), using the BACTEC MGIT 960 system. Resistance to pyrazinamide (PZA) was initially tested using Wayne's method (33), with confirmation of resistance in the BACTEC 460 (100 μg/ml). Those strains resistant to one or more first-line drugs were tested for their susceptibility to second-line drugs amikacin, ciprofloxacin, kanamycin, ethionamide (ETH) (5 μg/ml), and rifabutin in the BACTEC 460 system. All strains resistant to RIF were also resistant to rifabutin. As defined by the BACTEC MGIT 960 manual, 42 isolates were resistant to INH at 0.4 μg/ml and 10 isolates were resistant to INH at 0.1 μg/ml but sensitive at 0.4 μg/ml (defined as having intermediate resistance to INH and denoted as INHI). For the purposes of this study, unless specified, those isolates displaying intermediate resistance have been classified as INHr. Each matched pair of INHr and INHs isolates belonged to the same genotypic group, had similar MIRU-exact tandem repeat (ETR) profiles, and was isolated from patients born in the same country or region. DNA was extracted from M. tuberculosis cultures by the method described by Ross et al. (24). Patient country-of-origin data were kindly provided by the Department of Human Services, Victoria, Australia.
MIRU-ETR typing.ETR loci A, B, and C were amplified using the primers and amplification conditions described by Frothingham and Meeker-O'Connell (10). MIRU loci 2, 4, 10, 16, 20, 23, 24, 26, 27, 31, 39, and 40 were amplified using the primers and amplification conditions described by Supply et al. (30). PCRs were performed singly rather than using the multiplex format. The number of tandem repeats at each locus was determined by estimating the amplicon sizes after electrophoresis on 2% (wt/vol) agarose Tris-acetate-EDTA gels.
Classification of strains into major genetic groups and ancestral and modern strains.For the purpose of association studies, the isolates were each assigned to either genetic group 1 or groups 2/3 according to the classification scheme described by Sreevatsan et al. (28). (Genetic groups 2 and 3 were not separated due to the small number of isolates.) The presence of katG463 CTG (Leu) (group 1) or CGG (Arg) (groups 2/3) was determined by PCR-restriction fragment length polymorphism as described by Sreevatsan et al. (28). Genetic group 1 isolates were divided into “ancestral” and “modern” strains based on the number of repeats at MIRU locus 24 (29). Ancestral isolates have more than one repeat at locus 24, while modern strains have one repeat.
Identification of the Beijing and CAS genotype.Strains belonging to the Beijing family were identified based on the scheme proposed by Ferdinand et al. (8), by comparing the MIRU profiles of the study isolates with those of previously published Beijing isolates (18) or by comparison of IS6110 profiles to those described by Kremer et al. (15; data not shown). The aphC −46A polymorphism has been reported to be associated with the Delhi/CAS family of strains (2). The presence of either G or A at aphC −46A was determined by nucleotide sequencing of the 200-bp PCR product obtained using primers FSQ AphC-67 (5′-GTCGACTGGCTCATATCGAGA-3′) and R4OxyR-AhpC (5′-GGTTAGCAGTGGCATGACTCT-3′) (M. Hazbon; personal communication). PCR products were sequenced using BigDye terminator 3.1 cycle sequencing kit (Applied Biosystems) according to the manufacturer's instructions, and precipitated reaction products were run on an Applied Biosystems 3730 DNA analyzer.
Cluster analysis.The unweighted-pair group method using average linkages (UPGMA) tree was constructed using the Sequence Type Analysis and Recombinational Tests (START) program (http://outbreak.ceid.ox.ac.uk/software.htm). This algorithm, while not appropriate for the generation of a phylogenetic tree, grouped the isolates according to similarities in their profiles based on the 15 MIRU-ETR loci.
Statistical analyses.Tests for association were performed using the chi-square distribution.
katG315 MAS-PCR.Screening for mutations at katG codon 315 was carried out using the multiplex allele-specific PCR (MAS-PCR) assay developed and described by Mokrousov et al. (20). In this assay, the codon 315 region is PCR amplified with two outer primers and an inner reverse primer specific for the katG315 wild-type allele (AGC). Isolates with a wild-type codon 315 produce two PCR products, while those isolates with a mutation at codon 315 produce only one PCR product. All 52 INHr and 52 INHs isolates were screened using this method. Those isolates identified as not having a wild-type katG codon 315 (AGC) were then screened using a modified MAS-PCR assay, which, instead of the inner reverse primer specific for the katG315 wild-type sequence (20), used an inner reverse primer designed to detect the presence of either the katG315 ACC mutation, primer 523 (5′-ATACGACCTCGATGCCGG-3′), or the ACA mutation, primer 534 (5′-ATACGACCTCGATGCCTG-3′).
Sequencing of the fabG1-inhA regulatory region, inhA ORF, and katG ORF.The fabG1-inhA regulatory region was PCR amplified using primers 519 (5′-CCTCGCTGCCCAGAAAGGGA-3′) and 520 (5′-ATCCCCCGGTTTCCTCCGGT-3′). The inhA open reading frame (ORF) was amplified using primers 534 (5′-TCCGGTGCGGTCATCCCG-3′) and 535 (5′-AACGGCCGCACCTGCTCG-3′). The katG ORF was amplified in three overlapping segments: segment 1 was amplified using primers 527 (5′-ACACTTCGCGATCACATCCG-3′) and 528 (5′-ACCTCGATGCCGCTGGTG-3′), segment 2 with primers 540 (5′-CGGTCACACTTTCGGTAAGA-3′) and 541 (5′-GGCGAAGGACACTTTGATGT-3′), and segment 3 using primers 542 (5′-GCCAGCCTTAAGAGCCAGAT-3′) and 543 (5′-ACGCGGGGTCTGACAAAT3′). PCR products were sequenced as described previously.
RESULTS
Detection of mutations at katG codon 315 and in the fabG1-inhA regulatory region.The 52 INHr and 52 matched INHs sensitive control strains were screened for mutations at katG codon 315 and the fabG1-inhA regulatory region by MAS-PCR and DNA sequence analysis, respectively. A complete list of specific mutations identified is provided in Table 1. Thirty-four (65.4%) INHr isolates had mutations at katG codon 315. The wild-type codon, AGC (Ser), was altered to ACC (Thr) in 31 strains and ACA (Thr) in three strains. Mutations in the fabG1-inhA regulatory region were identified in 13 (25.0%) of the 52 INHr isolates (Table 1). All but one of these was the substitution of C for T 15 nucleotides upstream from the fabG1 start codon. The one exception was a G-to-C substitution at position −17. None of the INHr isolates had mutations in both katG315 and the fabG1-inhA regulatory region. No mutations at either katG codon 315 or in the fabG1-inhA regulatory region were identified in any of the INHs isolates.
Mutations detected in M. tuberculosis study isolates
Identification of mutations in the katG and inhA ORFs.Five INHr isolates had a wild-type sequence at both katG315 and in the fabG1-inhA regulatory region. The entire katG and inhA ORFs of these isolates were sequenced together with their matched INHs control strains. All five INHr and five INHs isolates had a wild-type inhA structural gene. Mutations in katG were identified in three of the five INHr isolates (Table 1). Isolate 2285 had an Asp117Ala substitution, isolate 2562 a Gly491Cys substitution, and isolate 2938 a Met257Ile substitution. These mutations were not present in the three INHs control strains tested; however, a more extensive survey of INHr and INHs isolates would need to be performed for an association between these mutations and INH resistance to be determined. No mutations in katG, the fabG1-inhA regulatory region, or the inhA structural gene were identified in isolate 2919 or 2956. Thus, resistance in these isolates is unlikely to be associated with either of these enzymes.
Drug resistance pattern, geographic origin, and strain differentiation.The drug resistance phenotype, MIRU-ETR profile, genotypic group, geographic origin, and mutation identified for each of the 52 INHr isolates are presented in Table 2. Twenty-three isolates were INH monoresistant; 16 isolates were resistant to both INH and streptomycin; 5 isolates were multidrug resistant (resistant to INH and rifampin); 4 isolates were resistant to both INH and ETH; 3 isolates were resistant to INH, pyrazinamide, and streptomycin; and 1 isolate was resistant to INH and ethambutol. Ten isolates displayed intermediate resistance to INH.
Characteristics of INH-resistant M. tuberculosis isolates
Although the isolates were collected from patients resident in Australia, only 7.7% (n = 4) of patients were born in Australia. The majority of overseas-born patients originated from Vietnam (n = 16), India (n = 8), China (n = 4), and the Philippines (n = 4), with 11 other countries represented. To determine whether INH resistance was associated with isolates from particular geographic origins, the proportion of strains resistant to INH from each country was compared to the proportion in the rest of the population (data not shown). Patients born in Vietnam were more likely to be infected with INHr isolates than non-Vietnamese-born patients (P = 0.001). This finding cannot be attributed to patient-to-patient transmission, as the isolates were epidemiologically unrelated.
The UPGMA tree (Fig. 1) based on MIRU-ETR patterns (Table 2) illustrates the genetic relationship, genotypic group, drug resistance profile, and mutations identified in the 52 INHr M. tuberculosis isolates. The 16 INHr isolates identified as ancestral genetic group 1 represented 9.5% (16 of 168) of the total ancestral strains collected during the study period. In comparison, 11.6% (25 of 215) of modern group 1 and only 4.2% (11 of 262) of group 2/3 strains were found to be resistant to INH. Group 2/3 isolates were therefore less likely to be resistant to INH than strains belonging to group 1 (P = 0.008). All five multidrug-resistant isolates were modern strains: one isolate belonged to groups 2/3, and four belonged to group 1. Beijing family strains accounted for 30.8% (n = 16) of the INHr isolates (Fig. 1), which was slightly higher than the proportion of Beijing strains in the total collection (22.6%). With the exception of strain 2817, the remaining modern group 1 strains (n = 8) were of the Delhi/CAS genotype (Fig. 1).
Genetic relationship, genotypic group, drug resistance profile, and mutations identified in the 52 INH-resistant M. tuberculosis isolates. This UPGMA tree was generated using the START program based on the 50 distinct MIRU-ETR patterns (Table 2) that represented the 52 isolates. I, isolate identification number; II, drug resistance phenotype; INHI, INH intermediate resistance; III, inhAP, fabG1-inhA regulatory region mutation; katG117, katG D117A substitution; katG257, katG M257I substitution; katG315, katG S315T substitution; katG491, katG G491C substitution.
Association between mutations at katG codon 315 and in the fabG1-inhA regulatory region, genotypic group, and drug resistance phenotype.The proportions of isolates within each genotypic group with the katG S315T substitution were as follows: ancestral, 12/16; group 1, 15/25; and groups 2/3, 7/11. The proportion of Beijing isolates harboring the S315T substitution (68.8%) was slightly higher than that of non-Beijing strains (63.9%). The numbers of isolates with the fabG1-inhA −15C→T mutation in each of the genotypic groups were as follows: ancestral, 4/16; group 1, 6/25; and groups 2/3, 2/11. Thus, neither the katG S315T substitution nor the fabG1-inhA −15C→T mutation clustered with any one genetic group. However, there was an association between the −15C→T substitution and those isolates with intermediate resistance to INH, with the mutation identified in eight of the 10 INHI isolates. In addition, the −15C→T substitution was strongly associated with resistance to ETH, with four of the five ETH-resistant isolates having the mutation.
DISCUSSION
From 2001 to 2003, 52 (8.1%) of the 645 new laboratory-confirmed M. tuberculosis cases in Victoria were resistant to INH. Isolates belonging to genetic groups 2/3 were less likely to be resistant to INH than genetic group 1 strains (P = 0.008). This is consistent with a number of studies that reported higher rates of INH or multidrug resistance in Beijing strains (which belong to genetic group 1) than isolates from other families (1, 16, 31, 32).
Mutations at katG codon 315 were identified in 65.4% of the isolates studied. The S315T substitution was observed in 68.8% of Beijing strains compared with 63.9% of non-Beijing strains. This difference in frequency of the S3125T substitution between Beijing and non-Beijing strains is smaller than those in previous studies (19, 21). The prevalence of the S315T substitution among this set of diverse isolates highlights the selective advantage conferred by the S315T substitution, which provides the basal catalase-peroxidase activity to protect the cell from oxidative stress while reducing the conversion of the INH prodrug to its active form by KatG (25, 26, 34). The three other mutations identified in katG, D117A, M257I, and G491C, represent novel mutations that to our knowledge have not previously been reported in the literature. None of these mutations maps near either the active site or the proposed INH binding site of KatG (5). Therefore, these mutations may cause resistance to INH by causing conformational changes to KatG, rather than by direct inhibition of INH binding.
Thirteen of the 52 INHr isolates had mutations flanking the putative fabG1 ribosome binding site. While the absence of mutations in this region in the INHs control strains indicates that fabG1-inhA regulatory region mutations may play a role in INH resistance, the effect(s) of these mutations on FabG1 and/or InhA protein synthesis and in turn the mechanism of resistance conferred by these mutations are yet to be elucidated. Since experimental studies have shown that overexpression of the M. tuberculosis inhA gene confers resistance to INH and ETH in Mycobacterium smegmatis, Mycobacterium bovis, and M. tuberculosis (3, 4), it is possible that the mutations identified in the fabG1-inhA regulatory region in clinical isolates could lead to the increased expression of the InhA protein, producing INH resistance via a titration mechanism.
The strong association between mutations in the fabG1-inhA regulatory region and intermediate INH resistance supports previous studies in which mutations within the inhA promoter and structural regions were associated with low-level resistance to INH (14, 23, 36). In their 2003 study, Ramaswamy et al. (23) identified one INHr isolate with a fabG1-inhA regulatory region mutation that also had a S315N substitution in katG and had a very high MIC, suggesting that in certain isolates with multiple mutations, INH resistance may develop in a stepwise fashion. Therefore, it is possible that those isolates in the study by Ramaswamy et al. with either the −15C→T or −17G→T mutation that were resistant to INH at 4.0 μg/ml have an additional mutation in katG, which cumulatively causes the higher level of resistance to INH.
There was also an association between mutations in the fabG1-inhA regulatory region and resistance to ETH, with four of the five ETH-resistant isolates having the −15C→T mutation. Other authors have reported that mutations in the fabG1-inhA regulatory region occurred in a disproportionately higher percentage of strains resistant to both INH and ETH than in isolates resistant to INH but susceptible to ETH. However in our study, 9 of the 13 isolates with the −15C→T mutation were susceptible to ETH, indicating that this mutation is not predicative of ETH resistance. Furthermore, the observation that a number of ETH-resistant isolates do not have mutations in the fabG1-inhA regulatory region indicates that other, ETH-specific mechanisms of resistance are involved in ETH resistance.
Understanding the nature and frequency of mutations associated with drug-resistance in M. tuberculosis in different clinical settings is important for the development and large-scale implementation of rapid, genetics-based assays for the detection of drug resistance. In this study, mutations at katG codon 315 or the fabG1-inhA regulatory region were identified in 90.4% of the INHr isolates and in none of the 52 INHs control strains selected to reflect the genotypic and geographic diversity of the isolates. Therefore, mutations in these regions are highly predictive of INH resistance in this population. Nevertheless, the identification of three novel katG mutations in three of the five isolates with wild-type katG codon 315 and inhA-fabG1 regulatory regions highlights the importance of bacteriological susceptibility testing methods in achieving comprehensive identification of resistance among clinical isolates.
ACKNOWLEDGMENTS
This work was supported by a grant from the National Health and Medical Research Council, Australia (NH&MRC Program grant ID 215 201).
The authors wish to thank Lynne Brown of the Victorian Department of Human Services and Paul Vinton of VIDRL for providing the country-of-birth data and Heath Kelly of VIDRL for performing the statistical analyses.
FOOTNOTES
- Received 6 April 2005.
- Returned for modification 7 May 2005.
- Accepted 2 August 2005.
- American Society for Microbiology