Single Nucleotide Polymorphisms in Genes Associated with Isoniazid Resistance in Mycobacterium tuberculosis

Abstract

Isoniazid (INH) is a central component of drug regimens used worldwide to treat tuberculosis. Previous studies have identified resistance-associated mutations in katG, inhA, kasA, ndh, and the oxyR-ahpC intergenic region. DNA microarray-based experiments have shown that INH induces several genes in Mycobacterium tuberculosis that encode proteins physiologically relevant to the drug's mode of action. To gain further insight into the molecular genetic basis of INH resistance, 20 genes implicated in INH resistance were sequenced for INH resistance-associated mutations. Thirty-eight INH-monoresistant clinical isolates and 86 INH-susceptible isolates of M. tuberculosis were obtained from the Texas Department of Health and the Houston Tuberculosis Initiative. Epidemiologic independence was established for all isolates by IS6110 restriction fragment length polymorphism analysis. Susceptible isolates were matched with resistant isolates by molecular genetic group and IS6110 profiles. Spoligotyping was done with isolates with five or fewer IS6110 copies. A major genetic group was established on the basis of the polymorphisms in katG codon 463 and gyrA codon 95. MICs were determined by the E-test. Semiquantitative catalase assays were performed with isolates with mutations in the katG gene. When the 20 genes were sequenced, it was found that 17 (44.7%) INH-resistant isolates had a single-locus, resistance-associated mutation in the katG, mabA, or Rv1772 gene. Seventeen (44.7%) INH-resistant isolates had resistance-associated mutations in two or more genes, and 76% of all INH-resistant isolates had a mutation in the katG gene. Mutations were also identified in the fadE24, Rv1592c, Rv1772, Rv0340, and iniBAC genes, recently shown by DNA-based microarray experiments to be upregulated in response to INH. In general, the MICs were higher for isolates with mutations in katG and the isolates had reduced catalase activities. The results show that a variety of single nucleotide polymorphisms in multiple genes are found exclusively in INH-resistant clinical isolates. These genes either are involved in mycolic acid biosynthesis or are overexpressed as a response to the buildup or cellular toxicity of INH.

Isoniazid (INH) is one of the primary chemotherapeutic and prophylactic drugs used against Mycobacterium tuberculosis, the causative agent of tuberculosis, which remains the leading single cause of death due to an infectious agent throughout the world. Recent studies indicate that the median rate of primary resistance to INH is 7.3% (range, 1.5 to 32%) and that the rates of acquired resistance range from 5.3 to 70% globally (9, 45). The overall rate of resistance to INH is 8.4% in the United States and has remained relatively stable in the last decade (23). Global reports of clusters of tuberculosis cases caused by drug-resistant strains together with the emergence and dissemination of multidrug-resistant tuberculosis have underscored the need for research into the mechanisms of drug resistance and the design of more effective antituberculous agents. Despite the use of INH for several decades, the molecular basis for its bactericidal action and the mechanisms by which INH resistance evolves in M. tuberculosis are only beginning to be understood.

INH has a simple chemical structure consisting of a hydrazide group attached to a pyridine ring, but its mode of action is very complex (8). It is proposed that INH enters M. tuberculosis as a prodrug by passive diffusion and is activated by catalase-peroxidase, encoded by katG, to generate free radicals, which then attack multiple targets in the cells (6). Recent studies have shown that an NADH-dependent enoyl acyl carrier protein (ACP) reductase, encoded by inhA, and a β-ketoacyl ACP synthase, encoded by kasA, are two potential intracellular enzymatic targets for activated INH; and both of these enzymes are involved in the biosynthesis of mycolic acids (4, 19, 20). Resistance-associated amino acid substitutions have been identified in the katG, inhA, and kasA genes of INH-resistant clinical isolates of M. tuberculosis (7, 20, 24, 26, 29). In addition, mutations in the oxyR-ahpC intergenic region have been identified in INH-resistant isolates (36). Additional genetic and biochemical studies have shown that certain promoter mutations of alkylhydroperoxide reductase, encoded by ahpC, in INH-resistant isolates result in overexpression of ahpC as a compensatory mechanism for the loss of catalase activity due to katG mutations (15, 32). Recently, missense mutations were identified in ndh, a gene encoding NADH dehydrogenase, which is an essential respiratory chain enzyme that regulates the NADH/NAD⁺ ratio in cells (18, 22). The molecular mechanism by which mutations in ndh confer INH resistance in M. tuberculosis is poorly understood. In addition, low-level INH resistance in mycobacteria has been shown to be associated with enhanced expression of arylamine N-acetyltransferase (NAT), which is believed to inactivate the prodrug INH by acetylating the molecule. However, no INH resistance-associated mutations have been described in the NAT gene (25, 37).

Publication of the complete genome sequence of M. tuberculosis has made possible the use of a DNA microarray to monitor changes in gene expression in response to INH (10, 43). Research has been shown that INH induces several genes, including ahpC; genes that encode type II fatty acid synthase enzymes involved in mycolic acid biosynthesis; fbpC, which encodes trehalose dimycolyl transferase (antigen 85-C), which is involved in mycolate maturation; fadE23 and fadE24, which are presumably involved in fatty acid β-oxidation; and several other genes with unknown functions (43). There have been no studies of the resistance-associated polymorphisms in the genes induced by INH. In an effort designed to better understand the molecular genetic basis of INH resistance, all genes implicated in conferring INH resistance as well as genes induced by INH were sequenced in their entirety to identify resistance-associated mutations in clinical isolates of M. tuberculosis.

MATERIALS AND METHODS

Bacterial isolates and DNA extraction.

Thirty-eight epidemiologically unrelated INH-monoresistant and 86 INH-susceptible isolates of M. tuberculosis cultured from patients with pulmonary and extrapulmonary tuberculosis were studied. The resistant isolates were obtained from 35 residents of Texas and 3 residents of Mexico. The Texas residents lived in 11 different counties. Foreign birth was verified for 11 Texas residents and included El Salvador, Honduras, India, the Philippines, and Zaire (n = 1 each), Vietnam (n = 2), and Mexico (n = 4). INH-susceptible isolates were also obtained from residents of Texas and Mexico. The INH-resistant and INH-susceptible isolates were chosen to represent all three major genetic groups of M. tuberculosis. Furthermore, on the basis of IS6110 profiling and genetic group designation, the genotypes of the susceptible control isolates were judged to be closely similar to those of the INH-resistant organisms. Genomic DNA was isolated from bacteria grown on Lowenstein-Jensen (LJ) medium. All bacterial growth and DNA extraction procedures were conducted in a biosafety level 3 laboratory.

Semiquantitative catalase assay.

INH-resistant isolates with mutations in katG encoding the catalase-peroxidase enzyme were tested for catalase activity, with strain H37Rv used as a positive control. A suspension of the isolates was prepared from freshly growing LJ agar slants in Middlebrook 7H9 broth to a density equivalent to that of a McFarland 1.0 standard. The suspensions were allowed to stand for 30 min so that large particles would settle to the bottoms of the tubes. One hundred microliters of the supernatant was inoculated onto an LJ deep (an agar tube with a horizontal surface) and incubated at 35°C for 2 weeks. One milliliter of a 1:2 mixture of Tween and 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.

MIC determination by E-test.

All isolates were initially classified as INH monoresistant or susceptible in routine diagnostic laboratories by the BACTEC radiometric method (0.1 μg/ml). The MICs for 32 INH-resistant organisms and H37Rv as a positive control were determined by the E-test (41). The E-test compares favorably with the proportion method in quantifying INH susceptibility results for M. tuberculosis (13, 14). Briefly, a suspension of M. tuberculosis was prepared from a freshly growing LJ agar slant in Middlebrook 7H9 broth to a density equivalent to that of a McFarland 3.0 standard. Four to five sterile glass beads were added, and the suspension was vigorously vortexed. The suspension was then allowed to stand for 30 min so that large particles would settle to the bottom of the tube. One 90-mm Middlebrook agar plate was inoculated by swabbing the suspension in three directions, taking care to evenly cover the plate. The plates were incubated at 35°C in 5% CO₂ for 24 h, at which time an INH E-test strip (AB Biodisk, Solna Sweden) was placed on each plate. The plates were sealed with CO₂-permeable tape or shrink seals and were reincubated under the same conditions until growth was visible (5 to 10 days). The MIC was read and was the point where the growth of the organism intersected the E-test strip.

IS6110 profiling.

All isolates were assessed for epidemiologic independence by IS6110 restriction fragment length polymorphism (RFLP) profiling by an internationally standardized method (39). The hybridizing DNA fragments were visualized by enhanced chemiluminescence, and the band patterns were analyzed with the BioImage (Ann Arbor, Mich.) Whole Band Analysis program (version 3.2). The number of hybridizing bands ranged from 1 to 21. Six isolates had five or fewer IS6110 bands (low-copy-number isolates). Except for two isolates that shared the same print pattern, designated Z (13 IS6110 copies), all other isolates had unique fingerprints, a result indicating epidemiologic independence.

Spoligotyping and major genetic group assignment.

Low-copy-number isolates were subjected to a secondary typing method by use of spoligotyping. This method was performed with a commercially available kit (Isogen Bioscience BV, Maarssen, The Netherlands) according to the instructions of the manufacturer. The spoligotype patterns were compared with those in the database maintained at the Houston Tuberculosis Initiative (34). All six low-copy-number isolates had unique spoligotype patterns.

All isolates were assigned to one of three major genetic groups on the basis of the polymorphisms present at gyrA codon 95 and katG codon 463 (35, 37). The distributions of the INH-resistant isolates on the basis of major genetic groups were as follows: group 1, n = 11; group 2, n = 18; and group 3, n = 7. The group designation was not given for two isolates because of katG deletions. The distributions of the INH-susceptible isolates on the basis of major genetic groups were as follows: group 1, n = 28; group 2, n = 35; and group 3, n = 23.

PCR amplification and sequencing strategy.

The 20 genes analyzed for nucleotide sequence variation are listed in Table 1. The oligonucleotide primers and PCR conditions used to amplify katG, inhA, ahpC, the oxyR-ahpC intergenic region, and the gyrA codon 95 region have been described previously (24, 36). The primers used to amplify all other genes are listed in Table 2. DNA was amplified with a GeneAmp system 9700 thermocycler (Applied Biosystems, Inc., Foster City, Calif.). Unincorporated nucleotides and primers were removed by filtration with Microcon 100 microconcentrators (Amicon Inc., Beverly, Mass.). DNA sequencing reactions were performed with a BigDye terminator cycle sequencing kit (Applied Biosystems, Inc.). The sequencing reaction products were purified by using Centrisep columns (Princeton Separations, Adelphia, N.J.). The sequence data generated with an ABI 377 automated instrument (Applied Biosystems, Inc.) were assembled and edited electronically with the ALIGN and EDITSEQ programs (DNASTAR, Madison, Wis.), and the sequences were compared with the corresponding sequences in the published sequence of strain H37Rv (10). All nonsynonymous mutations identified in the INH-resistant isolates were confirmed by resequencing.

TABLE 1.

Genes analyzed for nucleotide sequence diversity in INH-resistant M. tuberculosis

Gene	Product or function	Gene size (bp)	Positions of nucleotides sequenceda	Codons with nonsynonymous substitutionsb	Codons with synonymous substitutions
furA	Ferric uptake regulator	453	−70 to 570	5r, 115s	0
katG	Catalase-peroxidase	2,223	1 to 2311	(90, 91, 110, 138, 315, 336, 397, 434, 438, 529)r 463rs	322, 411, 436
mabA	3-Ketoacyl reductase	744	−199 to 781	21r	34, 203, 241
inhA	Enoyl ACP reductase	810	−75 to 898	(21, 194)r	3
ahpC	Alkylhydroperoxide reductase	588	−355 to 608	44s	6, 7
Rv0340	Unknown	540	−231 to 619	(101, 143, 149)s, 163r	30, 170, 173
iniB	INH-inducible gene	1,440	−192 to 1440	192s	0
iniA	INH-inducible gene	1,923	−36 to 1923	(3, 537)r, 481rs	178
iniC	INH-inducible gene	1,482	1 to 1566	83r	60
srmR homolog	Regulatory gene	1,245	−127 to 1509	3r, 323rs	108, 112, 296
fabD	Malonyl-CoA ACP transacylase	909	−109 to 962	(199, 275)rs	138
kasA	β-Ketoacyl ACP synthase	1,251	−29 to 1264	(77, 269)r, 312rs	6, 60
accD6	Acetyl-CoA carboxylase	1,422	−15 to 1460	229rs	143, 200
fbpC	Trehalose dimycolyl transferase	1,023	−250 to 1084	158rs	103, 205, 322
fadE24	Fatty acyl-CoA dehydrogenase	1,404	−168 to 728	0	0
efpA	Efflux protein	1,590	−92 to 718	73rs	0
Rv1592c	Unknown	1,338	−232 to 1357	(42, 430)r, (321, 322)rs (60, 355)s	56, 321
Rv1772	Unknown	309	−333 to 388	4r, 49s	0
ndh	NADH dehydrogenase	1,392	−186 to 1464	18rs	0
nhoA	Arylamine N-acetyltransferase	999	−539 to 1030	(67, 207)rs	304

^aThe primers used for the katG region amplify furA and katG plus 70 bp upstream from the start codon of furA and 88 bp downstream from the end of katG. The primers used for the inhA region amplify 199 bp upstream from the start codon of mabA, the mabA and inhA genes, and 88 bp downstream from the end of the inhA gene. The primers used for the ahpC region amplify 355 bp upstream from the start codon of ahpC, which includes the oxyR-ahpC intergenic regulatory region, the entire ahpC gene, and 20 bp downstream from the end of the gene. The primers used for the ini locus amplify 231 bp upstream of the start codon of Rv0340, the entire Rv0340 gene, the Rv0340-iniB intergenic region, iniB, the iniB-iniA intergenic region, iniA, iniC, and 84 bp downstream from the end of iniC. The primers used for the srmR homolog amplify 127 bp upstream from the start codon of the srmR homolog gene, the entire srmR homolog gene, and 264 bp downstream from the end of the gene. The primers used for the fabD region amplify 109 bp upstream from the start codon of fabD, the entire fabD gene, and 53 bp downstream from the end of the gene. The primers used for the kasA region amplify 29 bp upstream from the start codon of kasA, the entire kasA gene, and 13 bp downstream from the end of the gene. The primers used for the accD6 region amplify 15 bp upstream from the start codon of accD6, the entire accD6 gene, and 38 bp downstream from the end of the gene. The primers used for the fbpC region amplify 250 bp upstream from the start codon of fbpC, the entire fbpC gene, and 61 bp downstream from the end of the gene. The primers used for the fadE region amplify 168 bp upstream from the start codon of fadE24 and 728 bp of the fadE24 gene. The primers used for the efpA region amplify 92 bp upstream from the start codon of efpA and 718 bp of the coding region of the gene. The primers used for the Rv1592c region amplify 232 bp upstream from the start codon of Rv1592c, the entire Rv1592c gene, and 19 bp from the end of the gene. The primers used for the Rv1772 region amplify 333 bp upstream from the start codon of Rv1772, the entire Rv1772 gene, and 79 bp from the end of the gene. The primers used for the ndh region amplify 186 bp upstream from the start codon of ndh, the entire ndh gene, and 72 bp from the end of the gene. The primers used for the nhoA region amplify 539 bp upstream from the start codon of nhoA, the entire nhoA gene, and 31 bp from the end of the gene.

^br, nonsynonymous sites found exclusively in INH-resistant isolates; rs, nonsynonymous sites found in both INH-resistant and INH-sensitive isolates; s, nonsynonymous sites found in INH-sensitive isolates only.

TABLE 2.

PCR primers and conditions used to amplify the designated M. tuberculosis gene regions^a

Gene region	Forward primer	Reverse primer	Size (bp)	Length of D(s)	A (temp [°C], time [s])	Length of E(s)
furA	5′-GCGATCGGGTCCTAGCAG-3′	5′-TTCATATGACCCACGACGG-3′	641	40	61.0, 30	40
Rv0340	5′-ATGCGTCGTATGCTTGG-3′	5′-CCAAACACCTATCGGGATC-3′	850	40	58.5, 30	40
iniB	5′-ATAAGTTCCGGACCGGCG-3′	5′-CGACAGATGAGGCATAGCAG-3′	1,053	40	56.0, 30	45
iniB-iniA	5′-TTGAACGGCGCTGCTATG-3′	5′-GTGCTGATGTCATCGACGG-3′	1,070	40	62.0, 30	45
iniA	5′-CAACCGCAGCGGTTGACAT-3′	5′-CCGCATGCCGATAATCATT-3′	1,079	40	61.0, 30	45
iniA-iniC	5′-GGAATCGAAACCGCTGCG-3′	5′-CCAGCCCACCGATCTGTTTGA-3′	1,090	40	64.0, 30	45
iniC	5′-TCCTGTTGCGCACCCTGAAC-3′	5′-AACATGTTCCACCCGGTGGC-3′	1,040	40	65.0, 30	45
srmR homolog	5′-GCCAGTACCGGATCGACG-3′	5′-GTGTCGCGGACATCCTGG-3′	819	30	63.5, 30	40
	5′-CCAGGATGTCCGCGACACC-3′	5′-TGTCCGGGTGCGAGCAAC-3′	835	30	67.5, 30	40
fabD	5′-AAAAACATAGCTTACAGGCCCG-3′	5′-GTTGTGTACAAATCGAACTGACG-3′	1,071	40	59.5, 30	45
kasA	5′-GTTCAGGCGAGGCTTGAG-3′	5′-GCGATGTCGTGCTTCAGTAA-3′	1,293	45	60.5, 30	50
accD6	5′-GACAGGAGACCTGCGATGAC-3′	5′-GGCAGAACAATCCGACCA-3′	655	30	61.5, 30	30
	5′-CGGTTGGTCGGATTGTTC-3′	5′-CCTCGCGCGTGCGATTCTG-3′	842	35	61.0, 30	35
fadE24	5′-GCTATCAGATGCCTGGCG-3′	5′-GTGGGATCGAATAGTGGCTG-3′	890	35	61.0, 30	35
fbpC	5′-GCATGGGTCTCTCCTCTG-3′	5′-GCTGATACCAGTCGGTGTAG-3′	635	30	56.0, 30	30
	5′-GCCAATCCAGTTTCTACACCG-3′	5′-ATGTTCCACACATCGGC-3′	732	35	57.5, 30	35
efpA	5′-ACCTCCCGCGATCATCG-3′	5′-CGTCGAGCTTCATCCGTTC-3′	810	35	64.5, 30	35
Rv1592c	5′-AACTCGGCGTACCCAACC-3′	5′-ACACGCTCGGAATTCAAGG-3′	804	35	61.5, 30	35
	5′-GAGCGTGTCGGGTTGTCC-3′	5′-CGAGGTTGTGTGCCAGGTC-3′	834	35	64.0, 30	35
Rv1772	5′-CGGGTGTTTCTCAACGAC-3′	5′-GGACTGGACTCGCTGATTG-3′	633	30	57.5, 30	30
ndh	5′-ATCACCACCGCCGCTGAAGC-3′	5′-GTTCGGGTACCCGGGAATG-3′	1,134	45	65.0, 30	50
	5′-CATTCCCGGGTACCCGAAC-3′	5′-GTCGACCGTTTTGGCGTTGG-3′	535	30	65.0, 30	30
nhoA	5′-CTAGTGGCGCGAGCAGAC-3′	5′-CCGACGTCGACGAGATAGC-3′	927	40	61.5, 30	40
	5′-GCTATCTCGTCGACGTCG-3′	5′-GCATTCTACGTCTACGCCG-3′	662	30	59.0, 30	30

^aAbbreviations: D, denaturation at 94°C; A, primer annealing conditions; and E, extension at 72°C. All PCR were run for 25 cycles, were preceded by a denaturation step at 95°C for 15 min, and included a final extension step at 72°C for 7 min.

RESULTS

Overall sequence analysis.

Of the 2.6 megabases sequenced for the 124 M. tuberculosis isolates, 28 polymorphic synonymous nucleotide sites were identified (Table 1). Synonymous substitutions do not result in amino acid replacements and, hence, are unlikely to participate in drug resistance. These changes will not be discussed further when the sequencing results for the 20 genes studied are described. Of the 47 polymorphic nonsynonymous sites identified, 24 were present exclusively in INH-resistant isolates, 14 were found in both INH-resistant and INH-susceptible isolates, and 9 were found only in INH-susceptible isolates. The observation of small numbers of polymorphic sites in the structural genes is in agreement with previous observations (35). Seventeen (44.7%) of the 38 INH-resistant isolates had a resistance-associated mutation in only 1 of the 20 genes sequenced (Table 3). Seventeen isolates (44.7%) had resistance-associated mutations in more than one gene (Table 4). Six of the 20 genes (fabD, accD6, efpA, fbpC, ndh, and nhoA) sequenced did not have any resistance-associated mutations.

TABLE 3.

INH resistance-associated mutations in M. tuberculosis

Gene or gene region	No. of isolates with mutations at:		% of isolates with mutation at single locus	MIC rangeb (μg/ml)
	Single locusa	Multiple loci
furA	0	1	0
katG	15	14	39.5	0.1->256
mabA	1	8	2.6	2
inhA	0	2	0
ahpC	0	2	0
Rv0340	0	1	0
iniB	0	1	0
iniA	0	3	0
iniC	0	1	0
srmR homolog	0	1	0
fabD	0	0	0
kasA	0	3	0
accD6	0	0	0
fbpC	0	0	0
fadE24	0	1	0
efpA	0	0	0
Rv1592c	0	2	0
Rv1772	1	0	2.6	0.125
ndh	0	0	0
nhoA	0	0	0
Total	17		44.7

^aThe 15 katG mutations were Trp90STOP (TGG→TAG), Asn138Ser (AAC→AGC), Ser315Ile (AGC→ATC), Trp397Tyr (TGG→TAC) and Asn529Asp (AAC→GAC) (same isolate), Trp438Arg (TGG→CGG), and deletion of katG (n = 1 each) and Ser315Thr (AGC→ACC; n = 9). One isolate had two mutations in mabA. They were at position −15 (C→T) and Thr21A1a (ACC→GCC). The mutation in Rv1772 was Thr4A1a (ACA→GCA; n = 1). Seventeen isolates had INH resistance-associated mutations in two or more genes.

^bThe MIC range shown is for the isolates with single-locus mutations. The MICs for isolates with multiple mutations are shown in Table 4.

TABLE 4.

INH-resistant M. tuberculosis isolates with mutations in two or more genes

Isolate	MIC (μg/ml)	Mutationa
		furA	katG	mabA	inhA	oxyR-ahpC	srmR homolog	kasA	Rv0340	iniB	iniA	iniC	Rv1592c	fadE24
UT739	4		S315T									W83G
UT1885	12		S315T										V430A
UT2090	>256		124; G ins			9 UPS; G→A
UT2204	2		W91R				D3G
UT3174	>256		DEL			48 UPS; G→A
UT3230	ND		S315T											64 UPS; 2 bp ins
UT3306	>256		Q434STOP						V163I
UT3315	ND		L336P					G269S
UT3437	ND		S315T								5-bp DEL at position 94
UT3740	ND			147 UPS; C→T							P3A
HN1984	>256		S315N	15 UPS; C→T							R537H
HN1993	0.19		A110V	15 UPS; C→T				G269S
HN2029	0.19			15 UPS; C→T	I194T								P42L
NHN670	1.5		A110V	15 UPS; C→T
NHN673	0.25		A110V	15 UPS; C→T
NHN674	<0.016	S5P	A110V	8 UPS; T→A				M77I		12-bp DEL at position 222
NHN681	ND			15 UPS; C→T	I21V

^aThe single-letter amino acid designations used are as follows: A, alanine; D, aspartic acid; G, glycine; H, histidine; I, isoleucine; L, leucine; M, methionine; N, asparagine; P, proline; Q, glutamine; R, arginine; S, serine; T, threonine; V, valine; and W, tryptophan. The single-letter nucleotide designations used are as follows: A, adenine; C, cytosine; G, guanine, and T, thymidine. The other abbreviations are as follows: ins, insertion of bases; UPS, nucleotide position (base pairs) upstream of the start codon; DEL, deletion of the bases; STOP, termination mutation; ND, not determined.

Polymorphisms in the katG region.

Twenty-nine (76.3%) of the 38 INH-resistant isolates had a resistance-associated mutation in katG. Fifteen of those 29 isolates had a resistance-associated mutation in katG only. Two isolates had complete katG gene deletions, based on the reproducible absence of PCR products; two isolates had different termination mutations in codons 90 (TGG→TAG) and 434 (CAG→TAG); one isolate had an insertion of a guanine nucleotide at the third base in codon 124, which resulted in a frameshift mutation; and eight codons had substitution mutations in katG. One isolate had two substitution mutations in codons 397 (TGG→TAC, Trp→Tyr) and 529 (AAC→GAC, Asn→Asp) of katG and carried no other mutations in the other 19 genes sequenced. The most common substitution mutation occurred in codon 315 (Ser→Thr [AGC→ACC], n = 13; Ser→Ile [AGC→ATC], n = 1); and (Ser→Asn [AGC→AAC], n = 1). None of the polymorphisms identified in katG were observed in the susceptible organisms, suggesting that these mutations probably contribute to INH resistance.

It has recently been shown that katG is cotranscribed with furA from a common regulatory region and that furA is a negative regulator of katG (27, 46). So, we sequenced furA and its promoter region and found two polymorphisms. One of the polymorphisms was located at position 30 upstream of the start site of furA, resulting in a T→C change. This mutation was also identified in susceptible organisms and, thus, may not contribute to INH resistance. The second mutation was located in codon 5 (TCC→CCC, Ser→Pro) of a single INH-resistant isolate. This isolate also had other mutations in four different genes (Table 4).

Polymorphisms in the inhA region.

The inhA region consists of two genes, mabA and inhA, both of which encode enzymes involved in mycolic acid biosynthesis (4, 5). One isolate had two mutations in mabA: a C→T change at position 15 upstream of the start site of mabA and Thr→Ala substitution at codon 21. This isolate did not have mutations in any of the other genes. Eight isolates had nucleotide changes upstream of the start site of mabA at positions 15 (n = 6) and positions 8 and 147 (n = 1 each) (Table 4). Two isolates with a C→T change at position 15 upstream of the start site of mabA also had nonsynonymous substitutions in inhA, resulting in Ile194Thr (ATC→ACC) and Ile21Val (ATC→GTC) replacements. One of these isolates also contained a Pro42Leu (CCT→CTT) substitution in Rv1592c, and the other isolate did not have any other changes.

Polymorphisms in the oxyR-ahpC intergenic region.

Two isolates had nucleotide substitutions in the oxyR-ahpC intergenic region, which has previously been shown to be involved in INH resistance (29). The sites were at positions 9 (G→A) and 48 (G→A) upstream of the transcriptional start site of ahpC. The MICs were high for both isolates, and the isolates had reduced levels of catalase activity, suggesting that these mutations were probably selected as a consequence of the katG mutations. One isolate had a complete deletion of katG, and the other isolate had an insertion of a guanine nucleotide in codon 124 of katG, resulting in a frameshift mutation (Table 4). Three isolates had a substitution mutation at position 100 (G→A) upstream of the transcriptional start site of ahpC, but this change was also identified in susceptible organisms.

Polymorphisms in the kasA region.

The kasA genomic region consists of five genes (fabD, acpM, kasA, kasB, and accD6) in an operon preceded by a regulatory gene (srmR homolog) encoding enzymes involved in the fatty acid biosynthetic (type II fatty acid synthase [FAS-II] system) pathway (10, 12, 20, 33). fabD and acpM encode a malonyl coenzyme A (CoA) ACP transacylase and an ACP, respectively. Both kasA and kasB encode a β-ketoacyl ACP synthase involved in the production of long-chain mycolic acids. The fifth gene in the cluster, accD6, is an acetyl-CoA carboxylase (β-subunit) that is involved in the production of malonyl-CoA. It has previously been shown (43) that these genes are overexpressed in M. tuberculosis in the presence of activated INH. The open reading frame immediately upstream of the FAS-II gene cluster encodes a homolog of the regulatory protein, SrmR, that controls the production of polyketide in Streptomyces ambofaciens (12). Except for kasB and acpM, all genes in the kasA region were sequenced.

Sequencing analysis showed that two isolates had a substitution mutation in codon 3 (GAC→GGC, Asp→Gly) and codon 323 (ATG→ACG, Met→Thr) of the srmR homolog. The nonsynonymous change at codon 323 was also found in susceptible organisms, and the isolate with the codon 3 substitution also contained a mutation in codon 91 (TGG→CGG, Trp→Arg) of katG (Table 4). Three isolates had substitution mutations in kasA at codon 269 (GGT→AGT [Gly→Ser], n = 2) and codon 77 (ATG→ATA [Met→Ile], n = 1) that were not found in the susceptible organisms. However, these mutations were associated with mutations in katG and other changes in different genes (Table 4). In addition, one isolate had a Gly312Ser (GGC→AGC) replacement that was also observed in susceptible organisms. No resistance-associated mutations were identified in fabD or accD6. However, four isolates had Ser275Asn (AGC→AAC, n = 2) and Ala199Thr (GCG→ACG, n = 2) replacements in fabD that were also observed in susceptible organisms. Seven group 1 isolates had a GAC→GGC (Asp→Gly) replacement in codon 229 of accD6 which was also present in group 1 susceptible organisms, indicating that this polymorphism is a surrogate marker for group 1 M. tuberculosis isolates and does not participate in INH resistance.

Polymorphisms in the iniBAC region.

The ini chromosomal region has four genes designated Rv0340, iniB, iniA, and iniC, of which the last three genes are organized as an operon (2, 3, 10). The Rv0340 gene is located upstream of the iniBAC operon and is transcribed in the same orientation (10). The ini genes were originally identified on the basis of induction by INH and ethambutol treatment in vitro (2). On the basis of sequence annotation, the iniB gene encodes a protein with weak homology to alanine- and glycine-rich cell wall structural proteins, iniA encodes a protein with a phosphopantetheine attachment site motif characteristic of ACPs, and iniC encodes a protein that is 34% identical to IniA (2, 3). We have previously demonstrated ethambutol resistance-associated mutations in all four genes and also elected to sequence these genes in this study (30).

Sequence analysis showed that all four genes harbored mutations that were not found in the INH-susceptible organisms. One isolate with a termination mutation in codon 434 of katG also had a Val→Ile (GTT→ATT) replacement mutation in codon 163 of the Rv0340 gene (Table 4). One isolate had a 12-bp deletion beginning with the second base of codon 222 of iniB, resulting in a frameshift mutation. This isolate had changes in other genes as well (Table 4). Three isolates had INH resistance-associated mutations in iniA. One isolate had an Arg537His (CGC→CAC) replacement, another isolate had a Pro3Ala (CCC→GCC) substitution, and a third isolate had a 5-bp deletion beginning with the third base of codon 94, resulting in a frameshift mutation. All three isolates with resistance-associated changes in iniA had mutations in other genes as well (Table 4). Three INH-resistant isolates had a His481Gln (CAT→CAG) substitution that was also found in susceptible organisms (Table 1). One isolate had a resistance-associated mutation in codon 83 (TGG→GGG, Trp→Gly) of iniC that was not represented in INH-susceptible organisms. This isolate also had a Ser315Thr replacement in the katG gene.

Polymorphisms in other genes induced by INH.

fbpC is another gene induced by INH. It encodes an exported antigen (antigen 85-C) that has trehalose dimycolyl transferase activity and mediates a terminal step of mycolate maturation by esterifying mycolates to specific carbohydrate moieties in the cell wall of M. tuberculosis (43). Sequencing analysis showed that one isolate had a Gly158Ser (GGC→AGC) replacement mutation that was also found in susceptible organisms. Two isolates had a C→T nucleotide substitution at position 63 upstream of the start site of fbpC that was also found in INH-susceptible isolates.

Rv1592c and Rv1772 are two genes with unknown functions and are also transcriptionally induced by INH (10, 43). Sequencing analysis showed that two isolates had Pro42Leu (CCT→CTT) and Val430Ala (GTG→GCG) amino acid substitution mutations in Rv1592c that were absent in susceptible isolates. Both isolates had mutations in other genes as well. A group 1 INH-susceptible isolate had a Glu→Asp (GAA→GAC) substitution in codon 60, and a group 3 INH-susceptible isolate had a Asp→Glu (GAC→GAA) replacement in codon 355 of Rv1592c. Interestingly, all INH-resistant and -susceptible group 1 isolates also had a synonymous change in codon 321 (GAA→GAG) and a nonsynonymous substitution in codon 322 (ATT→GTT, Ile→Val). A subset of group 2 INH-resistant and -susceptible isolates also had the same changes. These two polymorphisms were not observed in any group 3 organisms. In addition, six INH-resistant group 2 organisms and six INH-susceptible group 2 organisms had a G→A nucleotide substitution at position 29 upstream of the start site of the Rv1592c gene. One isolate had a Thr4Ala (ACA→GCA) amino acid replacement mutation in Rv1772 that was not present in INH-susceptible isolates. This mutation was not accompanied by changes in the other 19 genes sequenced. The MIC for this isolate was low (0.125 μg/ml).

faDE24 is another gene induced by INH which encodes a fatty acyl-CoA dehydrogenase that is presumed to degrade fatty acids into acetyl-CoA subunits by β-oxidation (43). Sequencing analysis showed that one isolate had a 2-bp insertion at position 64 upstream of the translational start site of the gene. This mutation was not found in susceptible isolates but was accompanied by a Ser315Thr mutation in katG. In addition, two isolates had an A→C nucleotide substitution 23 bp upstream of the start codon, but this change was also found in INH-susceptible organisms.

An additional gene, efpA, encodes an efflux protein whose role in INH metabolism is not known (43). No resistance-associated mutation was identified in this gene. However, one isolate had an Ile→Thr (ATC→ACC) amino acid replacement mutation in codon 73 which was also discovered in INH-susceptible organisms.

Polymorphisms in ndh and nhoA genes.

ndh encodes an NADH dehydrogenase which controls the oxidative state of mycobacterial cells by regulating the NADH/NAD⁺ ratio (22). Mutations in ndh that cause an increase in this ratio result in coresistance to both INH and ethionamide in Mycobacterium smegmatis (22). Recently, mutations in ndh were reported in INH-resistant M. tuberculosis isolates (18). Our sequencing analysis shows no resistance-associated mutations in ndh, but two isolates had a Val→Ala (GTG→GCG) substitution mutation in codon 18 which was also identified in INH-susceptible organisms.

A recent study showed that nhoA encodes an NAT which acetylates INH in vitro and in Mycobacterium bovis BCG (25, 37). It was also shown that this gene was polymorphic at codon 207 and that all isolates with the polymorphism had the same IS6110 RFLP profile (37). Our sequence data showed a Gly→Glu (GGG→GAG) amino acid replacement mutation in codon 207 of nhoA in two isolates, but the mutation was also detected in INH-susceptible organisms. In addition, one isolate had a Gly→Arg (GGC→GGT) substitution at codon 67 which was also found in an INH-susceptible isolate. All the isolates with the two polymorphisms belonged to major genetic group 2.

Relationship between INH MICs, katG mutations, and catalase activity.

It has been known for a long time that the loss of catalase activity is associated with a high level of INH resistance (21). Our study corroborates this premise by showing that the seven INH-resistant isolates for which the MICs are the highest (>256 μg/ml) also had very low levels of or no catalase activity (Table 5). Two of those isolates lacked katG completely, two other isolates had truncated KatG proteins, one isolate had a frameshift mutation at codon 123, and the other two isolates had substitution mutations. The MICs for INH-resistant isolates with single-locus nonsynonymous mutations in katG were in the range of 0.19 to 256 μg/ml. Interestingly, for isolates with the Ser315Thr mutation in katG (n = 13), the MICs were in the range of 0.38 to 12 μg/ml, and the catalase activities of those isolates ranged from 10 to 26 mm. The MICs for isolates with amino acid replacements other than threonine at codon 315 of katG were higher, and the isolates had much lower catalase activities (Table 5). The MICs for two other isolates with single-locus, nonsynonymous, non-katG mutations were low (range, 0.125 to 2 μg/ml). There was also no simple correlation between the MICs for isolates with single or multiple resistance-associated mutations.

TABLE 5.

MICs, catalase activity, and mutations in katG for INH-resistant M. tuberculosis isolates^a

Isolate	katG mutationb	Catalase (activity height of bubbles [mm])c	MIC (μg/ml)
UT450	S315I	13	32
UT473	W397Y, N529D	15	>256
UT561	S315T	26	1
UT739	S315T	10	4
UT1885	S315T	25	12
UT1998	S315T	15	0.75
UT2007	S315T	15	6
UT2090	124; G Ins	5	>256
UT2204	L91R	5	2
UT2340	S315T	20	0.38
UT2419	S315T	20	4
UT2883	S315T	10	8
UT2931	S315T	25	0.5
UT3171	W90STOP	No bubbles	>256
UT3174	Del	No bubbles	>256
UT3181	Del	No bubbles	>256
UT3194	S315T	14	19
UT3306	Q434STOP	5	>256
HN1984	S315N	5	>256
HN1993	A110V	10	0.19
NHN670	A110V	9	1.5
NHN673	A110V	10	0.25
NHN674	A110V	10	<0.016
H37Rv	WT	25	<0.016

^aH37Rv was used as a positive control.

^bThe single-letter amino acid designations used are as follows: A, alanine; D, aspartic acid; I, isoleucine; L, leucine; N, asparagine; P, proline; Q, glutamine; R, arginine; S, serine; T, threonine; V, valine; W, tryptophan; and Y, tyrosine. The single-letter nucleotide designation used is as follows: G, guanine. The other abbreviations are as follows: Ins, insertion of bases; Del, deletion of the gene; STOP, termination mutation; WT, wild type.

^cNo bubbles indicates a lack of catalase activity. Only isolates with data on catalase activity are shown.

DISCUSSION

INH has been used as a primary drug to treat tuberculosis since the 1950s (8, 21). Shortly after the introduction of this drug for the treatment of tuberculosis, INH-resistant strains that had lost catalase activity were isolated from patients (21). The observed clinical relationship between the loss of catalase activity and INH resistance was finally explained in 1992 by the isolation and characterization of katG, which encodes the bifunctional catalase and peroxidase activities in M. tuberculosis (47). Biochemical studies have shown that the catalase-peroxidase activity of katG is also required for the activation of INH, which passively enters M. tuberculosis cells as a prodrug (6, 7, 11). The identity of the activated INH is unknown, but it has been suggested that it may be a free-radical intermediate (42). The stable end products of KatG oxidation of INH result in the production of isonicotinic acid, isonicotinamide, pyridine-4-carboxyaldehyde, and 4-pyridylmethanol, none of which have antimycobacterial activities (42). On the basis of the hyperfine splitting patterns of electron paramagnetic resonance spin trapping experiments, KatG-mediated oxidation of INH has been shown to produce acyl, acyl peroxo, and pyridyl radicals of INH (42). These free radicals can potentially disrupt various cellular processes, resulting in pleiotropic effects. The leading hypothesis suggests that KatG-activated INH primarily targets at least two components of the FAS-II system of M. tuberculosis. An acyl pyridine free-radical derivative of INH has been shown to bind to NAD⁺, and this adduct inhibits InhA, an enoyl reductase involved in mycolic acid biosynthesis (11, 19). In addition, a β-ketoacyl ACP synthase encoded by kasA has been shown to covalently bind to activated INH (20). Several studies have documented a multitude of mutations in katG, including insertions, partial and complete deletions, and termination and substitution mutations, in INH-resistant isolates (1, 24, 29, 31). Also, resistance-associated mutations have been found in the inhA and kasA genes of INH-resistant clinical isolates (7, 20, 24, 29). Knowledge of the spectrum of mutations in the katG, inhA, and kasA genes that may participate in INH resistance is not complete. Furthermore, several additional chromosomal loci have recently been identified that encode proteins that are induced in M. tuberculosis during INH treatment (43). The identification of these genes raises the possibility that mutations in them may confer INH resistance or participate in the survival mechanisms of M. tuberculosis due to INH treatment. Moreover, sequence variations in these induced genes have not been studied in either INH-resistant or -susceptible organisms. Information gained from sequencing analysis will be useful in the design of rapid molecular genetic methods that can be used to identify INH-resistant isolates as well as for the recognition of single nucleotide polymorphisms that can be useful in epidemiologic surveillance. To address these questions, we sequenced all 20 genes implicated in INH resistance in a sample of 38 INH-monoresistant isolates independently recovered from patients with tuberculosis in Texas and Mexico.

The results obtained from this study show that the genetic mechanisms of INH resistance in M. tuberculosis are highly complex and involve several genes. Twenty-nine of the 38 INH-resistant isolates had complete deletions, a frameshift mutation that caused by an insertion, and termination and point mutations in katG. Isolates with insertions and deletions have lost their catalase activities, and the MICs for these isolates are also the highest, confirming previous observations about the loss of catalase activity and a high level of INH resistance (21, 29). This finding also suggests that katG is not an essential gene and that M. tuberculosis cells have other means to combat oxidative stress. The most common mutation was the Ser315Thr mutation in katG, which is in agreement with the bulk of the epidemiologic data (29, 40). Biochemical evidence obtained from studies with purified wild-type and Ser315Thr KatG proteins suggest that the Ser315Thr mutation results in a competent catalase-peroxidase that has a reduced ability to metabolize INH (29). The high frequency with which variants with this mutation are recovered from patients all over the world suggests that M. tuberculosis cells achieve a good balance between the need to maintain catalase-peroxidase activity and the need to reduce conversion of the prodrug INH to its activated form without a significant loss of bacterial fitness (27, 28).

The catalase-peroxidase gene (katG) is under the positive control of oxyR, a global transcriptional regulator of oxidative stress in several bacteria. The oxyR sequence in M. tuberculosis contains multiple substitution mutations and partial deletions that render the gene inactive (29, 36). This gene is divergently transcribed with ahpC, which encodes the catalytic subunit of alkylhydroperoxide reductase. Biochemical evidence suggests that increased AhpC activity could compensate for the loss of KatG activity in the detoxification of organic peroxides (32, 44). In light of the prevailing view regarding the role of ahpC in INH resistance, two isolates that had lost their catalase activities also had compensatory mutations in ahpC (29). In M. tuberculosis the katG gene is adjacent to furA, a homolog of the ferric uptake regulator (10). Recently, it was shown that FurA is a negative regulator of katG, and both genes are cotranscribed from a common regulatory region (27, 46). In addition, FurA appears to be a dominant regulator of oxidative stress and other genes involved in intracellular survival in mycobacterial species and lacks a functional oxyR (27). If furA is a negative regulator of katG, promoter mutations might lead to decreased levels of expression of KatG, which should then result in INH resistance. Our results show that a promoter mutation at position 30 upstream of the start codon of furA was identified in both INH-resistant and -susceptible organisms, indicating that this nucleotide is probably not important for transcriptional activity. Also, another strain with a structural mutation in codon 5 of FurA had mutations in other genes. The MIC for this strain was equal to that for the control organism, suggesting that the furA and other mutations in this isolate may contribute to a low-level INH resistance phenotype.

Previous studies have shown that mutations in the upstream region of the inhA locus result in increased levels of InhA expression, thereby elevating the drug target levels and producing INH resistance via a titration mechanism (29). Also, the mutations identified in the structural part of inhA results in INH resistance due to a reduced binding affinity of the INH-NAD⁺ adduct for enoyl reductase (4, 7, 29). Our results show that the MICs for all isolates harboring mutations in the promoter region of the inhA locus as well as point mutations in the structural part of inhA are low. Four of these isolates also contained an Ala110Val (GCC→GTC) substitution in katG, implying that this mutation contributes to low-level INH resistance. Another isolate with the inhA promoter mutation contained a Ser315Asn (AGC→AAC) change in katG and two different mutations in iniA. The MIC for this isolate was very high. This finding suggests that in certain isolates with multiple mutations, INH resistance develops in a stepwise fashion. This study also identified for the first time a structural mutation in mabA in an INH-resistant clinical isolate (5, 29). INH has not been shown to bind to MabA and, hence, is not a direct target for activated INH (5, 11). This isolate, for which the MIC was low, also had the promoter mutation, making it difficult to explain the contribution of the structural mutation to INH resistance.

Differential protein analysis of INH-treated and control cells has shown that activated INH covalently binds to AcpM and β-ketoacyl synthase (19). kasA and other genes in the FAS-II system are also upregulated by INH treatment (43). Our results show that mutations in kasA were accompanied by changes in other genes. The MICs for isolates with kasA mutations were low. Piatek et al. (26) have previously shown that Gly269Ser (GGT→AGT) point mutations can also be found in INH-susceptible isolates. We did not observe this change in any of the 86 INH-susceptible isolates in this study. The Met77Ile (ATG→ATA) substitution in kasA has also not been observed before. In addition, Lee et al. (17) have observed point mutations in codons 121 and 387 (n = 1 each) of kasA from two different INH-resistant isolates. On the basis of molecular modeling of the KasA protein, it has been suggested that the known mutations in kasA might affect the KasA-INH-AcpM complex, and this issue remains to be investigated. Kremer et al. (16) have described two clinical isolates with the Gly269Ser (GGT→AGT) substitution that are still sensitive to INH, suggesting that this mutation is not clinically significant. It then appears that resistance-associated kasA changes occur at a low frequency in INH-resistant clinical isolates and likely reflect different geographical prevalences of specific genotypes (18). Therefore, the routine use of sequencing of kasA to identify INH-resistant clinical isolates is not useful. The srmR homolog is a gene found upstream of the FAS-II gene cluster and controls the production of polyketide in S. ambofaciens (12). Polyketide biosynthesis is mechanistically similar to fatty acid biosynthesis by the FAS-II system (12). Our sequencing analysis showed that an INH-resistant isolate for which the INH MIC was at a medium level contained a mutation in the srmR homolog, but it should be noted that this isolate also had a katG mutation (Table 4).

The iniBAC operon encodes genes that are induced by a broad range of antibiotics including INH and ethambutol (3). The functions of the proteins encoded by the genes of this operon are unknown, and these genes lack close homologs in nonmycobacterial species. It has been shown that the iniBAC promoter is specifically induced by a broad range of inhibitors of cell wall biosynthesis in M. tuberculosis (3). On the basis of the induction pattern, it is postulated that these genes may participate in the regulation of cell wall growth or have a protective role in cell death (2, 3). Our results show that six isolates contained mutations in the ini genes exclusively represented in INH-resistant isolates. However, all these isolates also had changes in other genes. We have previously identified mutations in the ini genes in ethambutol-resistant M. tuberculosis isolates (30). The mutations identified in ini genes in this study are completely different from the ones selected due to ethambutol treatment. Ethambutol inhibits arabinogalactan biosynthesis and brings about cell death, whereas INH treatment leads to the accumulation of mycolic acids and eventual cell death. Thus, it remains plausible that mutations in the ini genes are selected as a consequence of cell wall injury due to INH or ethambutol therapy.

The roles of Rv1592c and Rv1772 in the physiology of M. tuberculosis are not known (10, 43). Further research is warranted to understand the significance of mutations identified in these genes. The identification of a synonymous substitution and a nonsynonymous substitution (codons 321 and 322, respectively) in Rv1592c in all group 1 isolates and a subset of group 2 isolates indicates that these polymorphisms are ancient and conserved in the evolution of the M. tuberculosis complex and suggests that this gene may have a role in the typing or classification of group 1 and 2 organisms (35).

Our sequencing analyses showed no resistance-associated mutations in the following genes: fabD, accD6, efpA, fbpC, ndh, and nhoA. However, this does not rule out a role for these genes in causing INH resistance. Additional studies that include INH-resistant M. tuberculosis isolates from global populations are needed to see if mutations in these genes play a role in INH resistance. It should be noted that resistance-associated mutations in ndh have been reported in M. tuberculosis isolates from Singapore (18). Four INH-resistant isolates in this study did not have resistance-associated mutations in any of the 20 genes sequenced. The MIC for one of those isolates was high (48 μg/ml). This finding suggests that genetic mechanisms other than the ones described in this study are likely involved in the causation of INH resistance. The sequencing data from this limited data set also showed that certain polymorphisms are exclusively present in INH-resistant isolates, while other polymorphisms are found in both INH-resistant and -susceptible isolates (Table 1). With the exception of substitutions in codon 315 of katG and the mabA promoter substitution at position 15 upstream of the start codon, all other polymorphisms associated with INH resistance occurred at low frequencies. In the absence of knowledge regarding the functions of many of the genes, the low frequency of mutations observed in the isolates makes it difficult to reach firm conclusions about their role in the causation of INH resistance. To solve the problem of associating low-frequency polymorphisms with drug resistance by sequencing alone, screening of a large number of susceptible isolates is required, but this is fiscally constraining. Further biochemical and genetic studies are required to establish a role for the mutations identified in INH resistance. The sequence data provided by analysis of 20 genes from this study is an initial step toward gaining insight into the complex mechanism of action of INH in M. tuberculosis.