Sequence Analysis of Fluoroquinolone Resistance-Associated Genes gyrA and gyrB in Clinical Mycobacterium tuberculosis Isolates from Patients Suspected of Having Multidrug-Resistant Tuberculosis in New Delhi, India

Abstract

Fluoroquinolones (FQs) are broad-spectrum antibiotics recommended for the treatment of multidrug-resistant tuberculosis (MDR-TB) patients. FQ resistance, caused by mutations in the gyrA and gyrB genes of Mycobacterium tuberculosis, is increasingly reported worldwide; however, information on mutations occurring in strains from the Indian subcontinent is scarce. Hence, in this study, we aimed to characterize mutations in the gyrA and gyrB genes of acid-fast bacillus (AFB) smear-positive sediments or of M. tuberculosis isolates from AFB smear-negative samples from patients in India suspected of having MDR-TB. A total of 152 samples from patients suspected of having MDR-TB were included in the study. One hundred forty-six strains detected in these samples were characterized by sequencing of the gyrA and gyrB genes. The extracted DNA was subjected to successive amplifications using a nested PCR protocol, followed by sequencing. A total of 27 mutations were observed in the gyrA genes of 25 strains, while no mutations were observed in the gyrB genes. The most common mutations occurred at amino acid position 94 (13/27 [48.1%]); of these, the D94G mutation was the most prevalent. The gyrA mutations were significantly associated with patients with rifampin (RIF)-resistant TB. Heterozygosity was seen in 4/27 (14.8%) mutations, suggesting the occurrence of mixed populations with different antimicrobial susceptibilities. A high rate of FQ-resistant mutations (17.1%) was obtained among the isolates of TB patients suspected of having MDR-TB. These observations emphasize the need for accurate and rapid molecular tests for the detection of FQ-resistant mutations at the time of MDR-TB diagnosis.

INTRODUCTION

Treatment of multidrug-resistant tuberculosis (MDR-TB) patients, with strains resistant to rifampin (RIF) and isoniazid (INH), is further complicated by the presence of fluoroquinolone (FQ) resistance, due to prolonged, limited, and expensive treatment options (1, 2). A recent meta-analysis of the responses to treatment of 6,724 MDR-TB patients from 26 centers revealed that the frequency of treatment success was 64%, while that for patients with MDR-TB plus FQ resistance was only 48% (3). This clearly indicates the need for routine molecular screening for FQ resistance-associated molecular markers so that the treatment of such patients can be optimized without delay. Many such patients develop extensively drug resistant tuberculosis (XDR-TB), defined as MDR-TB plus resistance to any one FQ and any of the aminoglycosides/cyclic peptides (A-CP) (4). This poses an even more serious threat to TB management, since only 40% of XDR-TB patients have successful treatment outcomes (2). Previous studies have suggested that 5.4% (95% confidence interval [95% CI], 3.4 to 7.5%) of MDR-TB cases may actually be XDR-TB (2).

FQs are oral antibacterial agents that have activity against Mycobacterium tuberculosis (5, 6). Hence, FQs are recommended for the treatment of MDR-TB patients and patients with intolerance to RIF (2). FQs are also being evaluated in conjunction with first-line regimens for newly diagnosed drug-susceptible TB (7). Since FQs are broad-spectrum antibiotics, they are often overprescribed by clinicians for diverse infections (8). In many resource-limited countries, FQs are readily available as over-the-counter medications, exacerbating their misuse (9, 10). Such indiscriminate use has contributed to the increasing emergence of FQ-resistant M. tuberculosis strains in India and worldwide (2, 11). For effective exploration of new FQ-based regimens, now under way, knowledge of FQ resistance data from countries with high TB burdens is necessary (12).

Traditionally, culture-based methods have been used for determining antimicrobial susceptibility. These procedures usually take several weeks. Thus, a patient harboring a drug-resistant M. tuberculosis strain may continue to spread the infection undetected in the community (13). Molecular methods, however, have decreased the time to the detection of drug resistance, especially for MDR-TB (14). Among these methods, reverse hybridization or line probe assays (LPAs) have been implemented as part of national TB programs in high-burden countries for rapid detection of MDR-TB (15). These assays are directed toward the detection of the most common mutations leading to RIF resistance (14). Fewer molecular tests are available for diagnosing FQ resistance than RIF resistance, and the existing methods are limited by lower levels of association with phenotypic drug resistance (16–18).

The principal target of FQ drugs in M. tuberculosis is the DNA gyrase protein complex, which catalyzes the ATP-dependent introduction of negative supercoils into closed circular DNA molecules. The gyrA and gyrB genes encode the two subunits of DNA gyrase, which form the catalytically active GyrA2–GyrB2 complex. Resistance to FQs in M. tuberculosis is caused primarily by mutations in the quinolone resistance-determining region (QRDR), a conserved 320-bp region within the gyrA gene. FQ resistance has also been shown to be associated with, but less commonly attributable to, mutations in a 375-bp region of the gyrB gene (5).

Mutations in the gyrA and gyrB genes of clinical isolates of M. tuberculosis have been described for many populations in many studies (5, 17–23); however, information on mutations in the gyrA and gyrB genes from the Indian subcontinent is scarce (24, 25), although India has one of the highest percentages of MDR-TB cases in the world (2). There is evidence that different mutations are associated with different levels of phenotypic resistance to FQs, or different MICs; moxifloxacin (MOX) has been suggested to be more active, with lower MICs for all isolates, than ofloxacin (OFL) (23, 25, 26). Thus, knowledge of the specific mutation may help reveal the level of FQ resistance, which can, in turn, inform decisions regarding the selection of OFL or newer FQs, such as MOX or levofloxacin (LVX). This information can also be used to determine whether phenotypic testing should be extended from OFL to MOX and LVX. In the present study, we therefore aimed to characterize mutations in the gyrA and gyrB genes of M. tuberculosis isolates from patients from the Indian subcontinent suspected of having MDR-TB.

MATERIALS AND METHODS

Sample selection.

Samples from presumptive MDR-TB patients from districts within New Delhi, India, sent to the Mycobacteriology Laboratory in New Delhi over a period from 1 to 30 November 2013, were analyzed in this study. The presumptive MDR-TB patients were identified according to the national guidelines (programmatic management of drug-resistant tuberculosis [PMDT]); criteria include failures, with positive smears, on a category I or category II regimen; a history of contact with MDR-TB; all retreatment cases (with positive or negative smears); and HIV-TB coinfection (27). A category I regimen is prescribed for new patients (an intensive phase for 2 months, consisting of INH, RIF, pyrazinamide [PZA], and ethambutol [EMB] given under observation thrice weekly, followed by a continuation phase for 4 months consisting of INH and RIF given thrice weekly) (27). A category II regimen is prescribed for TB patients who have had a history of at least 1 month of antituberculosis treatment. Patients with relapses, default, or failures are treated with a category II regimen (an intensive phase for 2 months consisting of INH, RIF, PZA, EMB, and streptomycin [SM] given under observation thrice weekly, followed first by 1 month of INH, RIF, PZA, and EMB given under observation thrice weekly and then by a continuation phase of INH, RIF, and EMB given thrice weekly for 5 months) (27).

Sample testing.

Testing was performed using the GenoType MTBDRplus assay (Hain Lifescience, Nehren, Germany), a line probe assay (LPA), and a WHO-endorsed molecular test. This diagnostic tool may be used for the detection of MDR-TB (resistance to both RIF and INH) directly on smear-positive sputum samples or, if samples are smear negative, on culture isolates. Smear-positive sputum samples were identified as MDR-TB directly by an LPA, with a turnaround time of approximately 5 days. Smear-negative sputum samples were inoculated for culture and, if positive for M. tuberculosis, were subjected to an LPA (27) according to the PMDT. The site of testing was the Mycobacteriology Laboratory located in the Department of Microbiology at the National Institute of Tuberculosis and Respiratory Diseases (NITRD) in New Delhi, India. This is one of the National Reference Laboratories (NRL) under the Revised National Tuberculosis Control Program (RNTCP) in India and a WHO center of excellence.

Processing of samples.

The samples were processed by a standard N-acetyl-l-cysteine–sodium hydroxide (NALC-NaOH) method (28), with a final concentration of 1.0% NaOH (28). The samples were stained for acid-fast bacilli (AFB) by the Ziehl-Neelsen (ZN) method. All processed AFB smear-positive samples were considered for DNA extraction and an LPA, and all AFB smear-negative samples were inoculated into Bactec MGIT (Becton Dickinson, Sparks, MD, USA) tubes and, when positive for M. tuberculosis, were considered for an LPA. The cultures were incubated for 42 days before being considered negative.

DNA extraction.

For DNA extraction, 500 μl of NALC-NaOH-processed samples or 1 ml of a culture from Bactec MGIT tubes was used. DNA was extracted using the GenoLyse kit (version 1.0; Hain Lifescience, Nehren, Germany). Supernatants containing DNA were transferred to a fresh tube and were stored at −20°C.

GenoType MTBDRplus assay.

Amplification and hybridization using the GenoType MTBDRplus kit, version 2.0, were performed according to the manufacturer's instructions and as described previously (29). For quality control for each assay, M. tuberculosis H37Rv was used as a positive control and molecular-grade water as a negative control in every run.

Primer design for sequencing of FQ resistance regions of gyrA and gyrB.

Molecular amplification, sequencing, and detection of polymorphisms were carried out at National Jewish Health, Denver, CO. A consensus sequence of the target regions was created using alignments of approximately 10 to 15 M. tuberculosis gyrA or gyrB sequences present in the National Center for Biotechnology Information (NCBI) database (http://www.ncbi.nlm.nih.gov/). Two pairs of primers each for gyrA and gyrB were designed using Primer3 (http://bioinfo.ut.ee/primer3-0.4.0/primer3/). The gyrA primer sets were gyrARS-Forward (F)/gyrAPRR-Reverse (R) (5′-CAGCTACATCGACTATGCGA-3′/5′-ATTTCCCTCAGCATCTCCA-3′), with a 358-bp amplicon, and a nested set, gyrAPRR-F/gyrARS-R (5′-GACTATGCGATGAGCGTGAT-3′/5′-GGGCTTCGGTGTACCTCAT-3′), with a 310-bp amplicon. The two primer sets designed for gyrB were gyrBPRR-F/gyrBPRR-R (5′-AACAGCTGACCCACTGGTTT-3′/5′-CGCTGCCACTTGAGTTTGTA-3′), with a 556-bp amplicon, and gyrBPRS-F/gyrBPRS-R (5′-CGCAAGTCCGAACTGTATGTCGTAG-3′/5′-GTTGTGCCAAAAACACATGC-3′), with a 346-bp amplicon. The annealing temperature and magnesium concentrations were optimized prior to sample analysis.

Amplification.

Initial amplification of gyrA and gyrB using 40 cycles with a single pair of primer sets for each gene gave poor amplification, likely due to low DNA quantity in the GenoLyse-extracted DNA. Secondary purification and concentration of DNA also failed to yield consistent amplification. Thus, nested amplification was performed with primer sets as described above.

gyrA amplification.

PCR was performed with the Advantage GC Genomic LA polymerase mix (catalog no. 639153; Clontech, Mountain View, CA, USA) using 400 μM deoxynucleoside triphosphates (dNTP) and 400 nM primers. The first amplification employed gyrARS-F/gyrAPRR-R in a 10-μl reaction mixture under the following conditions: one cycle for 2 min at 94°C; 15 cycles of 15 s at 95°C, 20 s at 60°C, and 75 s at 72°C; one cycle for 2 min at 72°C; and a hold at 4°C. The second amplification was performed with gyrAPRR-F/gyrARS-R for 35 cycles under the same conditions, using 2 μl of the first PCR mixture as the template. Figure S1 in the supplemental material illustrates the positioning of forward and reverse primers for gyrA.

gyrB amplification.

The first PCR was performed with AmpliTaq Gold DNA polymerase (Thermo Fisher Scientific, Carlsbad, CA) using the gyrBPRR primer set with 2.5 mM Mg²⁺, 200 μM dNTP, and 200 nM primers under the following conditions: one cycle at 95°C for 10 min; 20 cycles of 95°C for 15 s, 59.1°C for 20 s, and 72°C for 75 s; one cycle at 72°C for 2 min; and a 4°C hold. The second amplification used the gyrBRS primer set with 2 mM Mg²⁺, 200 μM dNTP, and 200 nM primers under the following conditions: one cycle at 95°C for 10 min; 40 cycles of 95°C for 15 s, 60°C for 20 s, and 72°C for 75 s; one cycle at 72°C for 2 min; and a 4°C hold. Figure S1 in the supplemental material illustrates the positioning of forward and reverse primers for gyrB.

Amplicon purification and sequencing.

The amplicons were prepared for sequencing by treatment with exonuclease I and shrimp alkaline phosphatase (Exo-SAP). Ten microliters of 1× PCR buffer–0.375 mM Mg²⁺ containing 0.05 U SAP and 2 U exonuclease I was added to each PCR mixture. The mixture was incubated at 37°C for 30 min, followed by 95°C for 10 min. Sequencing was performed with the BigDye Terminator cycle sequencing kit, version 3.1 (Applied Biosystems, Inc. [ABI], Foster City, CA). Each 10-μl reaction mixture contained 1.5 μl 5× buffer, 0.875 μl BDX64 (Molecular Cloning Laboratories [MCLAB], San Francisco, CA), 0.35 μl BigDye Terminator (version 3.1), 1 μl 5 μM primer, and 2 μl Exo-SAP-treated amplicon. Cycling conditions consisted of 2 min at 96°C, followed by 35 cycles of 10 s at 96°C, 5 s at 52°C, and 2 min at 60°C. Sequencing products were purified with spin columns (EdgeBio, Gaithersburg, MD) according to the manufacturer's directions, dried, and loaded onto an ABI 3730 genome analyzer (Applied Biosystems, Foster City, CA).

Sequence analysis.

Sequence traces were imported into a BioNumerics database (Applied Maths, Austin, TX) and were trimmed if necessary. Sequences were exported in FASTA format. The genetic polymorphisms of gyrA and gyrB were compared to the reference sequence from M. tuberculosis strain H37Rv (GenBank accession numbers NC_000962 and L27512). The polymorphisms obtained were annotated according to the M. tuberculosis genome numbering system.

Structure analysis.

Gyrase A mutations were mapped onto the X-ray crystal structure of the M. tuberculosis gyrase A protein and were color coded using the PyMol structure visualization software package (https://www.pymol.org/). The coordinates of the gyrase A X-ray crystal structure (PDB reference code 3ILW) were obtained from the Protein Data Bank (http://www.rcsb.org/pdb/home/home.do).

RESULTS

A total of 152 samples from patients suspected of having MDR-TB from the Delhi region of North India were considered for the GenoType MTBDRplus assay. The history of antituberculosis drugs in a category I or category II regimen or the history of contact was collected for these patients. Overall, 104/152 patients (68.4%) were at the start of a retreatment/category II regimen due to default, incomplete treatment, treatment failure, or relapse. Six of 152 patients (10.5%) had failures of a new TB/category I regimen. Thirty of 152 patients (19.7%) had failures of a retreatment/category II regimen. Of 152 patients, 1 had a history of contact with MDR-TB and 1 was HIV positive. LPAs were performed on 16 M. tuberculosis isolates from cultures of AFB smear-negative sputum samples, whereas the remaining LPAs were performed directly on AFB smear-positive sputum samples.

The 179-bp region between bp 148 and bp 327 of the gyrA gene, encoding amino acids 49 to 109, was studied for the identification of mutations. Six strains yielded poor gyrA sequences even on repeat sequencing and were excluded from the analysis. Among 146 strains studied for the gyrA gene, 27 mutations in 25 strains (17%) were obtained; of these, 2 strains had more than one mutation. Table 1 shows clinical details, susceptibility to INH and RIF, and mutations observed in gyrA. The substitution at amino acid 94 was the most common mutation (13/27 [48.1%]). The nucleotide (amino acid) mutations observed include A281G (D94G) (8/27 [29.6%]) and C269T (A90V) (8/27 [29.6%]), followed by A281C (D94A) (2/27 [7.4%]), G280T (D94Y) (2/27 [7.4%]), C269Y (A90V*) (2/27 [7.4%]), G280A (D94N) (1/27 [3.7%]), A283G (S95A) (1/27 [3.7%]), T271C (S91P) (1/27 [3.7%]), and A281R (D94G*) (2/27 [7.4%]), where asterisks indicate heterozygosity. Strains ID12 and ID18 had the double mutations A90V D94G and A90V D94A, respectively. The locations of the mutations in the codons and the respective amino acid changes in the primary sequence are shown in Fig. 1. The locations of mutations in the GyrA protein structure are shown in Fig. 2.

TABLE 1.

Characteristics of gyrA mutations and corresponding amino acid changes in various Mycobacterium tuberculosis strains isolated in clinically suspected multidrug-resistant tuberculosis cases

Strain no.	Age (yr)	Sexa	Category of treatment	Type of case	Stage of treatmentb	Acid-fast bacillus smear resultc	Susceptibilityd to:		Nucleotide change(s) (amino acid change[s])e
							INH	RIF
ID1	21	M	2	Retreatment	Entry	2+	S	S	C269T (A90V)
ID2	33	M	2	Retreatment	Entry	1+	S	S	A281G (D94G)
ID3	11	M	2	Retreatment	Entry	Neg.	R	R	G280A (D94N)
ID4	11	M	2	Retreatment	Entry	Neg.	R	R	A281G (D94G)
ID5	67	F	2	Retreatment	Entry	3+	S	S	G280T (D94Y)
ID6	26	F	2	Retreatment	FUP	3+	S	R	C269T (A90V)
ID7	50	M	2	Retreatment	Entry	3+	S	R	C269T (A90V)
ID8	13	F	2	Retreatment	Entry	1+	R	S	A281C (D94A)
ID9	26	F	2	Retreatment	Entry	3+	R	R	A281G (D94G)
ID10	28	F	2	Retreatment	Entry	1+	R	R	A281G (D94G)
ID11	20	F	2	Retreatment	Entry	3+	R	R	A281G (D94G)
ID12	35	M	2	Retreatment	Entry	Neg.	R	R	C269Y (A90V)*, A281R (D94G)*
ID13	30	M	2	Retreatment	Entry	3+	R	R	C269T (A90V)
ID14	28	F	2	Retreatment	Entry	3+	R	R	A281G (D94G)
ID15	20	M	2	Retreatment	Entry	3+	S	S	C269T (A90V)
ID16	24	M	2	Retreatment	Entry	3+	R	R	T271C (S91T)
ID17	51	M	2	Retreatment	Entry	2+	S	S	A281G (D94G)
ID18	22	M	2	Retreatment	Entry	3+	R	R	C269T (A90V), A281C (D94A)
ID19	17	F	2	Retreatment	FUP	1+	S	S	C269T (A90V)
ID20	23	M	2	Retreatment	Entry	3+	R	R	A281G (D94G)
ID21	21	F	2	Retreatment	FUP	3+	R	R	C269T (A90V)
ID22	22	M	2	Retreatment	FUP	2+	R	R	G280T (D94Y)
ID23	13	F	1	New case	FUP	Scanty	S	S	A283G (S95A)
ID24	70	M	2	Retreatment	Entry	1+	S	S	C269Y (A90V)*
ID25	30	M	2	Retreatment	Entry	3+	R	R	A281R (D94G)*

^aM, male; F, female.

^bFUP, acid-fast bacillus smear positive on follow-up samples.

^cNeg., negative; Scanty, 1 to 9 AFB/100 fields; 1+, 10 to 99 AFB/100 fields; 2+, 1 to 10 AFB/field after examination of 50 fields; 3+, >10 AFB/field after examination of 20 fields.

^dINH, isoniazid; RIF, rifampin; S, susceptible; R, resistant.

^e*, heteroresistance.

FIG 1.

Locations of the mutations in the strains analyzed. Sequence polymorphisms likely to be involved in fluoroquinolone resistance are given in the boxes above (nucleotide changes) and below (amino acid changes) the primary sequence. A common lineage-specific polymorphism, which is not likely to be involved in resistance, is circled.

FIG 2.

Locations of the M. tuberculosis gyrA mutations identified. The mutations identified in this study are indicated in red on the gyrA protein structure and cluster at the quinolone resistance-determining region.

Coexistence of wild-type and mutated bases (heterozygosity) was seen in 4/27 (14.8%) mutations, in strains ID12, ID24, and ID25. Two resistance-associated mutations were seen in ID12; details are given in Table 1 and in Fig. S2 in the supplemental material. The G284C polymorphism (encoding an S95T polymorphism) relative to the H37Rv gyrA reference sequence was found in 138/146 (94.5%) strains. The S95T polymorphism is a common lineage-specific polymorphism, not likely to be involved in fluoroquinolone resistance.

Among patients suspected of having MDR-TB whose isolates had gyrA mutations, 22/25 (88%) were AFB smear positive and 3/25 (12%) were AFB smear negative. Of these isolates, the majority (24/25 [96%]) were from previously treated pulmonary TB patients, and one (1/25 [4%]) was from a patient with failure of a category I regimen. Four previously treated patients (4/24 [16.7%]) had failures of a category II regimen.

Of 146 patients suspected of having MDR-TB, 21 (14.4%) were found to have MDR-TB, 10 (6.8%) to have RIF-monoresistant strains, 5 (3.4%) to have INH-monoresistant strains, and 110 (75.3%) to have strains susceptible to both antibiotics based on the well-characterized mutations. Fourteen of the 21 MDR-TB strains (66.7%) and 2 of the 10 RIF-monoresistant strains (20%) had FQ resistance mutations in the gyrA gene. Among the non-MDR-TB cases, FQ resistance was observed in 10/115 (8.7%) strains; the difference was statistically significant (P < 0.0001). Similarly, RIF resistance was found for 16/25 (64%) strains with gyrA mutations, in contrast to 15/121 (12.4%) strains without gyrA mutations. This difference was also found to be statistically significant (P < 0.0001). The 418-bp region between bp 1229 and bp 1647 of the gyrB gene, encoding amino acids 410 to 543, was also investigated for 109 strains, and no mutations were found in any of the samples.

DISCUSSION

This study aimed at the detection of mutations in gyrA and gyrB genes conferring FQ resistance. The study population included TB patients from North India suspected of having MDR-TB. Most of the patients had a history of irregular or defaulted TB treatment using first-line antituberculosis agents such as RIF, INH, EMB, and PZA. It has been shown previously that such patients are much more likely to develop MDR-TB than others. The treatment regimen for such patients includes FQ or aminoglycosides (2).

In the present study, 17.1% of the patients were found to have strains harboring FQ mutations. In India, with a population of more than 1 billion, this has significant implications, due to the high potential proportion of FQ resistance among individuals suspected of having TB and MDR-TB. This is likely due to the overprescription of FQs in certain regions of the world, especially in many resource-limited countries, where they are readily available over the counter (8). In the Indian subcontinent, the level of FQ resistance in M. tuberculosis has increased in Mumbai from 3% of laboratory cultures in 1996 to 35% in 2004 and in Pakistan from 17.4% in 2005 to 42.9% in 2009 (9, 10). FQ resistance is known to be associated with poor treatment outcomes among MDR-TB patients (30), and the increasing prevalence of FQ resistance among TB isolates is of great concern.

Resistance to FQ in M. tuberculosis is caused mainly by mutations in the QRDR, a highly conserved region in gyrA, encoding amino acids 74 to 113, located at the N-terminal portion of the GyrA protein. FQ resistance in M. tuberculosis is less commonly due to mutations in the gene encoding the second protein of the gyrase complex, GyrB, with QRDR mutations observed from codon 461 to codon 499 (31). In the review of mutations by Maruri et al., 59% and 41% of mutations were in gyrA and gyrB, respectively. In gyrA, 81% and 19% of mutations were inside and outside the QRDR, respectively, whereas in gyrB, 44% of mutations were inside the QRDR, 50% were outside the QRDR, and 1% were deletions (31).

In the present study, mutations were found only in the gyrA gene; among these, substitutions at codon 94 were the most common, at 48.1%. The review by Maruri et al. also found substitutions at codon 94 (37%) to be the most common (31). In another review, across 18 countries, by Avalos et al., mutation data for 3,846 clinical isolates with known profiles of phenotypic resistance to FQ were studied. The most frequent gyrA mutation was D94G, which ranged from 21 to 32% (32). The relative frequencies of QRDR mutations reported from different regions worldwide are detailed in Table 2. D94G is reported as the most common codon 94 substitution in most reports, as seen in studies from Pakistan, China, the Philippines, Thailand, Vietnam, Uzbekistan, Russia, Germany, France, and the United States (17, 33–41). All mutations at codon 94 have been associated with in vitro resistance, as seen in earlier studies (17, 24, 33, 36, 37, 40, 41). The codon 90 mutation A90V has also been found at a high frequency; studies from India, the Philippines, China, and Belgium have reported it as the most common mutation (23, 24, 35, 42). The review by Avalos et al. found that A90V accounted for 13 to 20% of mutations contributing to resistance (32). In the present study, D94G and A90V were the most commonly observed mutations. Interestingly, both these mutations have been reported in all gyrA mutation studies. However, other substitutions at amino acid 94, including D94A, D94Y, D94N, and D94H, have been reported with various frequencies, from 0 to 17.6%, and a rare D94F substitution has been reported in only one study (35).

TABLE 2.

Frequencies of gyrA mutations in QRDRs of Mycobacterium tuberculosis isolates in different geographic regions

Mutated codon(s)a	Frequency (%) inb:														Phenotypic antimicrobial susceptibility result(s) for fluoroquinolonec (reference[s])
	New Delhi, India (n = 28) (present study)	India (n = 8) (Siddiqi et al. [24])	Pakistan (n = 41) (Ali et al. [33])	China, 2012 (n = 56) (Chen et al. [34])	China, 2011 (n = 63) (Huang et al. [18])	Philippines (n = 7) (Bravo et al. [35])	Thailand (n = 21) (Pitaksajjakul et al. [36])	Vietnam (n = 30) (Kiet et al. [37])	Uzbekistan (n = 23) (Feuerriegel et al. [38])	Russia (n = 40) (Mokrousov et al. [39])	Belgium (n = 28) (Von Groll et al. [23])	France, 2010 (n = 25) (Brossier et al. [17])	Germany, 2009 (n = 29) (Hillemann et al. [40])	USA (n = 17) (Devasia et al. [41])
T80A											7.1*	8.0			S (23), R or S (17)
M81T														5.9	R (41)
G88A							4.8			2.5		8.0			R (17, 36, 39)
G88C					9.5							4.0			R (17, 18)
D89N					1.6						3.6				R (18, 23)
A90V	28.6	37.5	26.8	17.9	18	42.9	23.8	20.0	21.7	17.5	39.3	20.0	10.3	5.9	R (17, 18, 23, 24, 33–41)
A90E												4.0			R (17)
A90G											3.6	4.0			R (17, 23)
A90V*	7.1												3.4		R (40)
S91T	3.6			1.8											R (34)
S91P		12.5	4.9	7.1	3.2		9.5		4.3				3.4	5.9	R (18, 24, 33, 34, 36, 38, 40, 41)
D94G	28.6	25	39	46.4	47.6	42.9	33.3	23.3	43.5	32.5	14.3	28.0	44.8	41.2	R (17, 18, 23, 33–41)
D94A	7.1	25	4.9	5.4	9.5		9.5	13.3	17.4	7.5	10.7	8.0	13.8		R (17, 18, 23, 33, 34, 36–40)
D94Y	7.1		12.2	7.1	11.1			3.3	4.3	17.5	10.7			11.8	R (18, 33, 34, 37–39, 41)
D94N	3.6		4.9	10.7	9.5		9.5		4.3	2.5	7.1	8.0	3.4	17.6	R (17, 18, 23, 33, 34, 36, 38–41)
D94H				3.6			9.5		4.3		3.6	4			R (17, 23, 34, 36, 38)
D94F						14.3									R (35)
D94G*	7.1							16.7		5		4			R (17, 37, 39)
D94N*								3.3		2.5					R (37, 39)
D94Y*										2.5					R (39)
D94A*								3.3					3.4		R (37, 40)
S95A	3.6
L96P			2.4												R (33)
L109P														5.9	R (41)
D111N			4.9												R (33)
Q113L														5.9	R (41)
A90V*, D94G*								10		2.5			3.4		R (37, 39, 40)
A90V*, D94A*								3.3							R (37)
S91P*, D94G*													3.4		R (40)
D94N*, D94G*										5			6.9		R (39, 40)
D94A*, D94G*								3.3							R (37)
D94N*, D94Y*													3.4		R (40)
D94N*, D94G*, D94Y*										2.5					R (39)

^aAsterisks indicate heteroresistance.

^bAfter the geographical region, the total number of mutations reported (n) and the source or reference are given in parentheses.

^cR, resistant; S, susceptible.

The frequency of A90V mutations has been found to be lower in some studies from Germany (10.3%), and the United States (5.9%), suggesting that the frequency of mutations in a gene tends to differ by geographic location (40, 41, 43). Two strains in the present study had double mutations: ID18, with A90V and D94A, and ID12, with A90V and D94G. The presence of double mutations has been related to high-level drug resistance, as reported previously (17, 31, 39, 40, 42). Aubry et al. reported high-level drug resistance in a strain with A90V and D94G mutations; the MIC for this strain was very high, at >160 μg/ml (5).

In the present study, a novel mutation in gyrA, S95A, within the QRDR, was found in ID23. Figure 2 reveals the X-ray crystal structure of the M. tuberculosis gyrase A protein (PDB reference code 3ILW), with amino acids at positions A90, S91, D94, and S95 within the QRDR shown in red. Although the S95A mutation is in close proximity to well-characterized fluoroquinolone resistance mutation sites, including D94, further microbiological investigation would be necessary to assess the impact of this specific mutation on fluoroquinolone resistance.

Heteroresistance, defined as both the wild type and a mutation at specific loci, was seen in 14.8% of FQ mutations identified by sequencing. The presence of heteroresistant mutations suggests mixed susceptible and resistant bacteria in the same clinical sample. Two different mechanisms for heteroresistance have been postulated: the coexistence of two genotypically different strains, i.e., one susceptible and one resistant, or a single strain evolving from susceptible and resistant organisms. It has been shown previously that infection with two different strains was responsible for heteroresistance in 71% of patients and could serve as a quality indicator for TB control programs in various countries (44). Heteroresistance may be present at higher frequencies in regions of hyperendemicity (15). The evolution of a susceptible strain into a resistant strain could result from poor treatment in terms of dose and/or duration. Other studies have also reported heteroresistance with frequencies ranging from 4 to 37.1% (17, 37, 39, 40).

An interesting observation in the QRDR of M. tuberculosis gyrA is the presence of a G-to-C polymorphism at position 284 (encoding S95T) relative to the H37Rv reference gene. This codon encodes either serine (in M. tuberculosis H37Rv and H37Ra) or threonine (in M. tuberculosis Erdman, Mycobacterium bovis BCG, and Mycobacterium africanum) (19). In the present study, this polymorphism was found in 94.5% of isolates. The S95T polymorphism has not been implicated in FQ resistance, since it has been found in strains for which ofloxacin MIC values are less than the critical concentration (<2 μg/ml) (6, 33, 35, 42). An exception was an earlier Indian study, which found S95T in 88.2% of ofloxacin-resistant isolates, with no other mutation in the QRDR of gyrA or gyrB (24).

We looked for gyrB mutations in the QRDR, since several worldwide studies have reported mutations associated with in vitro resistance, such as D472H, R485C, D495N, D500N, D500H, N510D, N533T, N538D, N538I, T539N, and 546M (5, 23, 36, 45). However, in our study, we found no gyrB mutations among our samples, a result that agrees with those of other studies performed in India and China (24, 34).

Interestingly, we found a significant association of MDR-TB with the gyrA mutations. More than half of the MDR-TB strains examined in this study had mutations in gyrA (66.7%). The rates of FQ resistance among MDR-TB patients also have been found to be high in other countries in the region as well: 51.4% in the Philippines (46), 22.2% in Taiwan (47), and 25.1% in China (Shanghai) (20). The rates in countries on other continents are lower, e.g., 4.1% in the United States and Canada (48) and 4.3% in MDR-TB patients in Russia (49). It has been hypothesized that resistance to other anti-TB drugs could potentially influence resistance to FQ (13).

In summary, our study revealed a high rate of FQ mutations—17.1%—among the isolates of Indian TB patients suspected of having MDR-TB. The proportion of FQ resistance-associated mutations in sequence-verified MDR strains was 66.7% whereas in non-MDR strains, it was 18.7%. All the mutations were found exclusively in the gyrA gene; A90V and D94G were the most common. One novel mutation, S95A, was also observed. The data strongly suggest that expanded studies, examining both phenotypic and genotypic data for a larger number of isolates from India and surrounding countries, should be conducted. Such studies would contribute to the development, refinement, and improvement of rapid molecular diagnostic tests for determining resistance (50). They would also aid in the control and identification of pre-XDR-TB and XDR-TB in a more timely and effective manner, preferably at the time of detection of MDR-TB.

Funding Statement

This work was supported by the NIH/FIC AIDS and TB International Training and Research Program (D43TW001409) at NYU School of Medicine (principal investigator, Suman Laal). Michael Strong acknowledges support from the Boettcher Foundation Webb-Waring program.