Molecular Characterization of Multidrug- and Extensively Drug-Resistant Mycobacterium tuberculosis Strains in Jiangxi, China

Abstract

In this study, a total of 77 multidrug-resistant and extensively drug-resistant (MDR and XDR, respectively) isolates of Mycobacterium tuberculosis were characterized among samples from patients living in Jiangxi province, China. The following two approaches were used: (i) genotyping all drug-resistant isolates by the 15-locus MIRU-VNTR (mycobacterial interspersed repetitive-unit–variable-number tandem-repeat) method and identifying the Beijing family genotype using the RD105 deletion targeted multiplex PCR and (ii) determining the mutation profiles associated with the resistance to the first-line antituberculous drugs rifampin (RIF) and isoniazid (INH) and the second-line drugs ofloxacin (OFX), kanamycin (KAN), amikacin (AMK), and capreomycin (CAP) with DNA sequencing. Six loci were examined: rpoB (for resistance to RIF), katG and mabA-inhA (INH), gyrA and gyrB (OFX), and rrs (KAN, AMK, and CAP). It is shown that the Beijing genotype was predominant (80.5%) among these strains and that the selected drug-resistant strains were genetically diverse, suggesting that they probably had independently acquired drug resistance. In comparison to the phenotypic data, the sensitivities for the detection of RIF, INH, OFX, and KAN/AMK/CAP resistance by DNA sequencing were 94.8, 80.5, 84.6, and 78.9%, respectively. The most prevalent mutations involved in RIF, INH, OFX, and KAN/AMK/CAP resistance were Ser531Leu in rpoB (44.2%), Ser315Thr in katG (55.8%) and C-15T in mabA-inhA (11.7%), Asp94Gly in gyrA (48.7%), and A1401G in rrs (73.7%), respectively. Five novel katG mutants (Trp191Stop, Thr271Pro, Trp328Phe, Leu546Pro, and Asp695Gly) and six new alleles (Ile569Val, Ile572Met, Phe584Ser, Val615Met, Asp626Glu, and Lys972Thr) in the rpoB gene were identified.

INTRODUCTION

With an estimated 9.4 million new cases and 1.7 million deaths in 2009, tuberculosis (TB) remains one of the major life-threatening diseases worldwide (49). The emergence of drug-resistant strains of Mycobacterium tuberculosis, especially those that are multidrug resistant (MDR) and extensively drug resistant (XDR), has posed a serious threat to global TB control. MDR TB is defined as resistance to both isoniazid (INH) and rifampin (RIF), XDR is defined as MDR TB with additional resistance to any fluoroquinolone (FQ) and at least one of the three second-line injectable drugs: kanamycin (KAN), amikacin (AMK), and capreomycin (CAP) (36). Comparing to drug-susceptible TB, the management of MDR is more complex, costly, time-consuming, and less effective (23). However, the regimen for XDR is more expensive and difficult than MDR, and the prognosis is much worse, particularly in those infected with human immunodeficiency virus (24). As of February 2011, a total of 69 countries worldwide have reported identifying at least one case of XDR TB (47).

The accurate and rapid detection of drug resistance is imperative for the prompt implementation of effective regimens to interrupt the transmission of MDR and XDR isolates. Any delay in the identification of resistant strains jeopardizes the efforts to control the disease. Exploring the molecular genetic basis of drug resistance will contribute to the establishment of efficient methods for the rapid identification of drug-resistant M. tuberculosis strains. However, the frequency, location, and type of resistance-associated mutations vary in different geographical areas (1, 30, 43, 52, 54).

Jiangxi Province, a resource-limited area with about 44.5 million inhabitants in 2010, is situated in the southeast part of China. In the province, the prevalence of bacteriological positive pulmonary TB was recorded to be 203.6 cases per 100,000 inhabitants (http://jiangxi.jxnews.com.cn) in 2000, which is higher than the national average (160 cases per 100,000 inhabitants in China in 2000) (17). Unfortunately, thus far, little information has been obtained with regard to the molecular mutations of MDR and XDR strains isolated from this province.

To characterize the prevalence and pattern of mutations occurred to the drug target loci, a collection of MDR and XDR strains were isolated from patients in Jiangxi Province. We screened the drug target genes for RIF, INH, FQ, and KAN, AMK, or CAP (KAN/AMK/CAP), which are the first and second-line drugs commonly used for treating TB in China. The genetic loci studied were rpoB (RNA polymerase B subunit), katG (catalase-peroxidase), mabA-inhA (inhA regulatory region), rrs (16S rRNA), gyrA, and gyrB (DNA gyrase). In addition, 15-locus MIRU-VNTR (mycobacterial interspersed repetitive-unit–variable-number tandem-repeat) genotyping was performed to identify the genetic lineage of these TB isolates.

MATERIALS AND METHODS

M. tuberculosis isolates.

From January 2010 to June 2011, a total of 97 M. tuberculosis strains were isolated as monocultures from single patients (69 males and 28 females; aged 14 to 78 years) with pulmonary TB. These patients were epidemiologically unlinked, and all were living in Jiangxi Province. Of the 97 isolates, 77 were drug-resistant strains, which included 35 simple MDR strains (resistance to INH and RIF, but not an FQ, and one of three second-line injectable drugs), 26 pre-XDR strains (defined as resistance to INH and RIF and either an FQ or one of three second-line injectable drugs, but not both) (55), and 16 XDR strains. However, the remaining 20 isolates were pan-susceptible strains and therefore used as negative controls. Meanwhile, the M. tuberculosis H37Rv (ATCC 27294) strain was used as a reference. All of the isolates were cultured on Lowenstein-Jensen solid medium, and grown colonies were identified using p-nitrobenzoic acid and thiophene carboxylic acid hydrazine resistance tests. The first passage of all of the clinical strains was used in testing for the first-line drug susceptibility, while the second passage was used in two testings for the second-line drug susceptibility and for the mutations in the target genes. These were done in the Jiangxi Chest Hospital, a tertiary care hospital and the sole specialized center for the diagnosis and treatment of TB in Jiangxi province. This study was approved by the Institutional Review Board of the Third Affiliated Hospital of Sun Yat-Sen University.

Drug susceptibility testing.

The first-line drug susceptibility testing (DST) was routinely performed on Lowenstein-Jensen medium using 1% indirect proportion method. For the MDR isolates, four second-line drugs were chosen to test. These were done according to the WHO guidelines (48) based on the following drug concentrations: isoniazid (0.2 μg/ml), RIF (40.0 μg/ml), streptomycin (STR; 4 μg/ml), ethambutol (EMB; 2.0 μg/ml), ofloxacin (OFX; 2.0 μg/ml), kanamycin (KAN; 30.0 μg/ml), capreomycin (CAP; 40.0 μg/ml), and amikacin (AMK; 40.0 μg/ml) (48, 55). Periodic external quality assessment of the performance of DST results was conducted by the Tuberculosis Reference Laboratory at the Beijing Research Institute for Tuberculosis Control.

Genomic DNA isolation, PCR amplification, and DNA sequencing.

Genomic DNA was extracted from samples using the CTAB (cetyltrimethylammonium bromide)-NaCl method as described previously (44). Expected fragment were amplified each in a standard 50-μl reaction volume, containing ca. 20 to 50 ng of genomic DNA, 1.25 U of Taq DNA polymerase (TaKaRa Bio, Dalian, China), 1 μl of forward and reverse primers (20 μM each), 5 μl of 10× Taq PCR buffer, 4 μl of deoxynucleotide triphosphates (2.5 mM each), and 36.5 μl of distilled H₂O. The amplification was performed on a Veriti 96-well Thermal cycler (Applied Biosystems) under the following condition: initial denaturation at 94°C for 5 min that was followed by 35 cycles each consisting of denaturation at 94°C for 1 min, annealing at the primer-dependent temperature for 45 s, and extension at 72°C for 1 min. The final extension was at 72°C for 10 min. The primer sequences and amplicon sizes are presented in Table 1.

Table 1.

Primers used by the PCR method in the hot spots of drug target genes

Primera	Nucleotide sequence (5′–3′)	Annealing temp (°C)	Amplicon positionb	Product size (bp)	Source or reference
rpoB-F	TCAAGGAGTTCTTCGGCACC	60.2	761068–761624	557	This study
rpoB-R	CTGCATGTTTGCCCCCAT
katG-F	TGGGCGGACCTGATTGTT	59.5	2459–3051c	593	This study
katG-R	CCGTCCTTGGCGGTGTATT
(mabA-inhA)-F	AAGGCAGAAGCCGAGTAG	57.4	1673100–1673619	520	This study
(mabA-inhA)-R	ACATTCGACGCCAAACAG
gyrA-F	CCGGATCGAACCGGTTGAC	57.2	7340–7766	427	20
gyrA-R	GTTAGGGATGAAATCGACTG
gyrB-F	GTCGTTGTGAACAAGGCTGTG	57.1	6452–6864	413	21
gyrB-R	GTGGAAATATGTTGGCCGTC
rrs-F	GTGAGATGTTGGGTTAAGTCC	55.3	1472913–1473393	481	20
rrs-R	TGGTGCTCCTTAGAAAGGAG

^aF, forward; R, reverse.

^bAccording to the complete sequences of the M. tuberculosis H37Rv genome (accession no. NC_000962), except as noted for katG.

^cAccording to the sequence of the M. tuberculosis H37Rv gene for catalase-peroxidase (accession no. X68081).

Amplified products were detected by 1.5% agarose gel electrophoresis and sent to Beijing Genomics Institute (Shenzhen, China) for sequencing. The amplified products were sequenced in both directions using the same forward and reverse primers as for the PCR amplification.

Nucleotide and amino acid sequences of the amplified fragments were aligned with the corresponding sequences of the reference M. tuberculosis H37Rv strain using the BLAST 2 software on the National Center for Biotechnology Information website (http://blast.ncbi.nlm.nih.gov/Blast.cgi). All of the mutations found were compared to those included in the TB Drug Resistance Mutation Database (www.tbdreamdb.com) (41).

In addition, we performed complete sequencing of rpoB, katG, gyrA, gyrB, and rrs (whole) genes, respectively, for the corresponding resistant strains in which no mutations were identified within the specific regions of these genes. The primers, used in complete sequencing of the five target genes, were presented in Table 2. Long fragments were amplified by using a high-fidelity DNA polymerase (PrimeSTAR HS; TaKaRa).

Table 2.

Primers used to perform complete sequencing of drug target genes

Primera	Nucleotide sequence (5′–3′)	Positionb	Product size (bp)	Source or reference
PCR primers
rpoB-F1	GCACCGCTCCTCTAAGGGCTCT	–54 to –33	3,624	This study
rpoB-R1	GAGCACGTAACTCCCTTTCCCCTA	3569 to 3546
katG-F1	TCAGCGCACGTCGAACCTGTCGAG	1 to 24	2,223	This study
katG-R1	GTGCCCGAGCAACACCCACCCATTA	2223 to 2109
gyrA-F1	TTCCTGGATGTCTAACGCAAC	–48 to –28	2,616	This study
gyrA-R1	ATTCCTCCTCAGATCGCTACG	2567 to 2547
gyrB-F1	GACGCACCAGGAAGAAAGATGT	–41 to –20	2,237	This study
gyrB-R1	ACGTCGTGTCTGTCATCTAT	2195 to 2176
rrs-F1	TACCTTTGGCTCCCTTTTCC	–35 to –16	1,617	This study
rrs-R1	ACGGCCTACGCCCCACCAGT	1581 to 1562
Sequencing primersc
rpoB-F2	GTCACCGTGCTGCTCAA	682 to 698		This study
rpoB-R2	AGATGTTCGGGATGTCG (CS)d	2311 to2295		This study
rpoB-R3	CACAATGGCGTTCGGCT (CS)	2898 to2882		This study
katG-R2	CCGCCTTTGCTGCTTTC (CS)	1675 to 1659		This study
gyrA-F2	TATCCCGCCGCACAACC	564 to 580		This study
gyrA-R2	GCCCGAGCGATTGGAGT (CS)	1938 to 1922		This study
gyrB-F2	GTCACCGTGCTGCTCAA	486 to 502		This study

^aF, forward; R, reverse.

^bNumbers are based on the nucleotide position relative to the initiation codon of each gene.

^cSequencing primers for each gene also contain the same forward and reverse primers as PCR amplification for the corresponding gene.

^dCS, complementary strand.

Genotyping method.

Seventy-seven MDR isolates were genotyped using the 15 locus MIRU-VNTR technique as described previously (45). Identification of the Beijing family strains was carrying out using the RD105 deletion-targeted multiplex PCR (DTM-PCR) method (13). The MIRU-VNTR genotyping data, transformed into a distance matrix on the MIRU-VNTRplus website (http://www.miru-vntrplus.org) using the default setting, were treated as categorical variables, and phylogenetic analysis of the distance data was conducted with the neighbor-joining algorithm using MEGA version 5.05 (3, 46).

Nucleotide sequence accession numbers.

The nucleotide sequences for the mutant genes obtained in the present study were deposited in the GenBank database under the following accession numbers: JN037847, JN049495, and JN049496 for the katG Trp191stop, Trp328Phe, and Thr271Pro gene mutants, respectively, and JN037845, JN037846, JN210554, and JN210555 for the rpoB Ile572Met, Phe584Ser, Ile569Val, and Ala615Met/Asp626Glu gene mutants, respectively.

RESULTS

General profile of drug resistance and genotyping.

Of the 97 isolates obtained, 77 were drug resistant. These included 35 MDR, 26 pre-XDR, and 16 XDR isolates. However, the remaining 20 were susceptible to all of the drugs tested. Table S1 in the supplemental material shows the profiles of the 77 resistant strains against 8 drugs: INH, RIF, EMB, STR, OFX, KAN, AMK, and CAP. Of these isolates, 42 (54.5%), 51 (66.2%), 39 (50.6%), 18 (23.4%), 16 (20.8%), and 13 (16.9%) were resistant to EMB, STR, OFX, KAN, AMK, and CAP, respectively. They each showed a unique genotype pattern, but no cluster was formed, as analyzed using the 15-locus MIRU-VNTR method (Fig. 1). Most isolates (62/77; 80.5%) were identified as Beijing family strains by DTM-PCR.

Fig 1.

Phylogenetic analysis of drug-resistant isolates. A phylogenetic map was generated by the neighbor-joining method, based on the 15-locus MIRU-VNTR genotyping data for 77 drug-resistant strains. “JX” indicates that an M. tuberculosis clinical isolate originated from Jiangxi, China. The numbers indicate the strain identification.

Mutations identified in the rpoB, katG, mabA-inhA, gyrA, gyrB, and rrs genes.

Sequences were obtained for specific regions of the rpoB, katG, gyrA, gyrB, and rrs genes and for the mabA-inhA operon as well. Mutations identified in the 77 resistant strains are summarized in Tables 3 and 4.

Table 3.

Nucleotide and amino acid changes identified in the specific region of six M. tuberculosis drug-resistant associated loci for the corresponding antibiotic

Locus (no. of resistant isolates tested)	Drug	Codon(s)	Changea		No. (%) of isolates	Other mutation
			Nucleotide	Amino acid
rpoBb (77)	RIF	Nonec	None	None	4 (5.2)
		505	TTC→TTT	Phe→Phef	1 (1.3)
		511	CTG→CCG	Leu→Pro	3 (3.9)
		514	TTC→Inse	Phe→Ins	1 (1.3)
		516	GAC→TAC	Asp→Tyr	1 (1.3)
			GAC→GTC	Asp→Val	4 (5.2)
			GAC→TTC	Asp→Phe	1 (1.3)
		526	CAC→AAC	His→Asn	2 (2.6)
			CAC→TAC	His→Tyr	3 (3.9)
			CAC→CTC	His→Leu	2 (2.6)
			CAC→GAC	His→Asp	7 (9.1)
			CAC→CGC	His→Arg	2 (2.6)
			CAC→TGC	His→Cys	2 (2.6)
		529	CGA→AAA	Arg→Lys	1 (1.3)
		531	TCG→TTG	Ser→Leu	29 (37.6)
		511/515d	CTG→CCG	Leu→Pro	1 (1.3)
			ATG→GTG	Met→Val
		511/516	CTG→CCG	Leu→Pro	2 (2.6)
			GAC→GGC	Asp→Gly
		511/516	CTG→CCG	Leu→Pro	1 (1.3)
			GAC→GCC	Asp→Ala
		511/516	CTG→CCG	Leu→Pro	1 (1.3)
			GAC→AAC	Asp→Asn
		513/561	CAA→GAA	Gln→Glu	1 (1.3)
			ATC→GTC	Ile→Val
		516/572	GAC→GGC	Asp→Gly	1 (1.3)
			ATC→CTC	Ile→Leu
		516/572	GAC→TAC	Asp→Tyr	1 (1.3)
			ATC→ATG	Ile→Met*
		531/525	TCG→TTG	Ser→Leu	1 (1.3)
			ACC→ACA	Thr→Thrf
		531/569	TCG→TTG	Ser→Leu	1 (1.3)
			ATC→GTC	Ile→Val*
		531/584	TCG→TTG	Ser→Leu	2 (2.6)
			TTC→TCC	Phe→Ser*
		515/526/535	ATG→GTG	Met→Val	1 (1.3)
			CAC→AAC	His→Asn
			CCC→TCC	Pro→Ser
		531/615/626	TCG→TTG	Ser→Leu	1 (1.3)
			GTG→ATG	Val→Met*
			GAC→GAG	Asp→Glu*
katG (77)	INH	None	None	None	16 (20.8)
		None	None	None	5 (6.5)	–15C→T upstream of mabA
		None	None	None	1 (1.3)	–8T→A upstream of mabA
		191	TGG→CGG	Trp→Arg	1 (1.3)
			TGG→TGA	Trp→Stop*	1 (1.3)
		271	ACT→CCT	Thr→Pro*	2 (2.6)	–15C→T upstream of mabA
		299	GGC→TGC	Gly→Cys	1 (1.3)
			GGC→AGC	Gly→Ser	2 (2.6)
		315	AGC→ACC	Ser→Thr	36 (46.8)
			AGC→ACC	Ser→Thr	2 (2.6)	–15C→T upstream of mabA
			AGC→ACC	Ser→Thr	2 (2.6)	–8T→C upstream of mabA
			AGC→AAC	Ser→Asn	1 (1.3)	–8T→C upstream of mabA
			AGC→ACA	Ser→Thr	3 (3.9)
			AGC→CGC	Ser→Arg	1 (1.3)
			AGC→ATC	Ser→Ile	1 (1.3)
		327	AAA→AAG	Lys→Lysf	1 (1.3)
		328	TGG→TTT	Trp→Phe*	1 (1.3)
mabA-inhA (77)	INH	None	None	None	64 (83.1)
			–8T→A		1 (1.3)
			–8T→C		2 (2.6)	AGC315ACC (Ser→Thr) in katG
			–8T→C		1 (1.3)	AGC315AAC (Ser→Asn) in katG
			–15C→T		5 (6.5)
			–15C→T		2 (2.6)	ATC271CCT (Thr→Pro) in katG
			–15C→T		2 (2.6)	AGC315ACC (Ser→Thr) in katG
gyrA (39)	OFX	None	None	None	1 (2.6)
		21/95	GAG→CAG	Glu→Gln	5 (12.8)
			AGC→ACC	Ser→Thr
		21/95	GAG→CAG	Glu→Gln	1 (2.6)	GAC500CAC (Asp→His) in gyrB
			AGC→ACC	Ser→Thr
		21/95/90	GAG→CAG	Glu→Gln	3 (7.7)
			AGC→ACC	Ser→Thr
			GCG→GTG	Ala→Val
		21/95/91	GAG→CAG	Glu→Gln	2 (5.1)
			AGC→ACC	Ser→Thr
			TCG→CCG	Ser→Pro
		21/95/94	GAG→CAG	Glu→Gln	19 (48.7)
			AGC→ACC	Ser→Thr
			GAC→GGC	Asp→Gly
		21/95/94	GAG→CAG	Glu→Gln	5 (12.8)
			AGC→ACC	Ser→Thr
			GAC→AAC	Asp→Asn
		21/95/94	GAG→CAG	Glu→Gln	1 (2.6)
			AGC→ACC	Ser→Thr
			GAC→GCC	Asp→Ala
		21/95/94	GAG→CAG	Glu→Gln	1 (2.6)
			AGC→ACC	Ser→Thr
			GAC→TAC	Asp→Tyr
		21/95/94	GAG→CAG	Glu→Gln	1 (2.6)	GAA540GAC (Glu→Asp) in gyrB
			AGC→ACC	Ser→Thr
			GAC→GCC	Asp→Ala
gyrB (39)	OFX	None	None	None	37 (94.8)
		500	GAC→CAC	Asp→His	1 (2.6)
		540	GAA→GAC	Glu→Asp	1 (2.6)	GAC94GCC (Asp→Ala) in gyrA
rrs (19)	KAN, CAP	None	None	None	2 (10.5)
	CAP	None	None	None	1 (5.3)
	KAN, AMK, CAP	None	None	None	1 (5.3)
	KAN, AMK		1401A→G		6 (31.6)
	KAN, AMK, CAP		1401A→G		8 (42.1)
	KAN, AMK, CAP		1402C→T		1 (5.3)

^a*, Mutations not previously reported.

^bThe amino acid numbering is based on homologous mutations in E. coli.

^cNone, no mutation.

^dMultiple mutations in one isolate.

^eIns, insertion.

^fSilent mutation.

Table 4.

Nucleotide and amino acid changes found in complete sequencing of target genes

Gene (no. of resistant isolates tested)	Isolatea	Mutationb
		Nucleotide (codon)c	Amino acid
rpoB (5)	JX015	A2672C (AAG891ACG)*	K972Td*
katG (23)	JX044	C329T (GCC110GTC)	A110V
	JX084	C insertion at position 133	Frameshift
		G1388T (CGG463CTG)	R463L
	JX085	A2084G (GAC695GGC)*	D695G*
		G1388T (CGG463CTG)	R463L
	JX099	T1637C (CTC546CCC)*	L546P*
		G1388T (CGG463CTG)	R463L
	JX110	A141C (GTA47GTC)	V47V
		G1388T (CGG463CTG)	R463L
gyrA (7)	JX001	G2003A (GGC668GAC)	G668D
	JX002	G2003A (GGC668GAC)	G668D
	JX003	G2003A (GGC668GAC)	G668D
	JX008	G2003A (GGC668GAC)	G668D
	JX018	G2003A (GGC668GAC)	G668D
	JX064	G2003A (GGC668GAC)	G668D
gyrB (37)		None	None
rrs (4)		None	None

^aJX, isolates from Jiangxi province. The numbers indicate the strain identification.

^b*, Mutation not previously reported. None, no mutation.

^cThe mutation coordinates are based on the start nucleotide (codon) of the target genes.

^dThe mutation coordinate is in reference to E. coli and not M. tuberculosis.

Rifampin and rpoB.

Since most RIF-resistant strains harbor mutations within the 81-bp RIF resistance-determining region (RRDR) (9, 39), a 557-bp fragment of the rpoB gene comprising this region and its flanking sequences was analyzed. In the present study, the majority (72/77; 93.5%) of MDR strains harbored at least one mutation within the RRDR of rpoB, while 5 of the RIF-resistant strains lacked this mutation (Table 3). Moreover, a mutation Lys972Thr was detected outside the RRDR of rpoB in one pre-XDR isolate with complete sequencing of rpoB gene for the five strains (Table 4). A total of 26 missense mutations, one 3-bp insertion, and two silent mutations were observed (Table 3 and 4). One MDR isolate had the silent mutation Phe505Phe. Another MDR isolate showed the silent mutation Thr525Thr plus the substitution Ser531Leu. Eleven isolates had two missense mutations represented by nine unique base changes. Five double mutants each showed a mutation within and a mutation outside the RRDR. Two isolates had triple mutations: Ser531Leu, Val615Met, and Asp626Glu were observed in one MDR isolate, while Met515Val, His526Asn, and Pro535Ser were found in one pre-XDR isolate. Of the 26 missense mutations, 6 (Ile569Val, Ile572Met, Phe584Ser, Val615Met, Asp626Glu, and Lys972Thr) were first reported here (Tables 3 and 4). Among them, Ile569Val was found in one MDR isolate, while Ile572Met and Phe584Ser were observed in two XDR isolates, respectively. In addition, the most highly changed codons were 531, 526, 516, and 511, which had frequencies of 44.2, 24.7, 13.0, and 10.4%, respectively (Table 3). When all of the mutations were considered, regardless they are single, double, and triple, a total of 27 rpoB genotype patterns were identified (Tables 3 and 4). The three most frequently observed mutations accounted for 58.4% of the RIF-resistant isolates (Table 5). In contrast, 20 pan-susceptible strains and the reference H37Rv strain contained no mutations within the target fragment of rpoB gene. Detection of nonsynonymous single nucleotide polymorphisms (nSNPs) in the rpoB gene exhibited a sensitivity of 94.8% and a specificity of 100% for predicting the phenotypic RIF results (Table 6).

Table 5.

Most frequently identified mutations within six M. tuberculosis drug resistance-associated loci among isolates resistant to that antibiotic for which each locus serves as a resistance marker

Drug(s)	Locus	Mutation	Frequency (no. of isolates)	Relative frequencya (%)
RIF	rpoB	Ser531Leu	34	44.2
		His526Asp	7	9.1
		Asp516Val	4	5.2
INH	katG	Ser315Thr	43	55.8
	inhA	–15C→T	9	11.7
		–8T→C	3	3.9
OFX	gyrA	Asp94Gly	19	48.7
		Asp94Asn	5	12.8
		Ala90Val	3	7.7
	gyrB	Asp500His	1	2.6
		Glu540Asp	1	2.6
KAN/AMK/CAP	rrs	A1401G	14	73.7

^aCompared to the total number of isolates resistant to the drug of interest.

Table 6.

Number of clinical M. tuberculosis isolates stratified by resistance to antibiotics, mutation, and validity values^a

Drug	Locus	No. of isolates				Accuracy values
		Resistant		Susceptible
		With mutation	Without mutation	With mutation	Without mutation	Sensitivity (%)	Specificity (%)
RIF	rpoB	73	4	0	20	94.8	100
INH	katG	58	19	0	20	75.3	100
	inhA	13	64	0	20	16.9	100
	katG and inhA	62	15	0	20	80.5	100
OFX	gyrA	32	7	0	20	82.1	100
	gyrB	2	37	0	20	5.1	100
	gyrA and gyrB	33	6	0	20	84.6	100
KAN, AMK, and CAP	rrs	15	4	0	20	78.9	100
MDR (RIF and INH)	rpoB, katG, and/or inhA	60	17	0	20	77.9	100

^aThe diagnostic performance of the DNA sequencing method in comparison to the drug susceptibility was determined after the resolution of natural polymorphisms and silent mutations.

Isoniazid and katG and inhA.

It has been well established that most INH-resistant isolates possess mutations in katG or mabA-inhA or the both (9, 39–40, 54). Therefore, a 593-bp region of katG encompassing codon 315 and a 520-bp region of mabA-inhA were sequenced. A total of 77 INH-resistant isolates were identified, 61 (79.2%) of which were found to contain mutation(s) in katG and/or mabA-inhA (Table 3). Of the 61 mutants, 7 (11.5%) showed mutations in both katG and mabA-inhA, 48 (78.7%) showed mutations only in katG, and 6 (9.8%) showed mutations only in mabA-inhA. In contrast, the remaining 16 (20.8%) isolates had no mutations within the target fragments of both loci. As anticipated, the most prevalent katG alteration was the substitution of threonine for serine at amino acid 315. This amino acid substitution resulted from a codon change from AGC to ACC in 40 (51.9%) isolates and to ACA in 3 isolates (3.9%). Other codon 315 mutations of katG included a substitution with asparagine (AAC), a substitution with arginine (CGC), and a substitution with isoleucine (ATC), each found in one isolate. Two resistant strains showed a single katG missense mutation at codon 271, resulting in the substitution of proline (CCT) for threonine (ACT). Three resistant isolates had a single katG missense mutation at codon 299, leading to the substitution of cysteine (TGC; n = 1) and serine (AGC; n = 2) for glycine (GGC). Moreover, one resistant isolate carried an alteration from TGG to TGA at codon 191, resulting in a pseudogene formation. In all, 11 different kinds of missense mutations were detected in five distinct codons (191, 271, 299, 315, and 328), and a silent mutation (Lys327Lys) was found within the 593-bp region of katG gene examined. Furthermore, complete sequencing of katG gene was performed for the 23 INH-resistant strains, in which no mutations were found within the specific region of katG gene. Consequently, one silent mutation (Val47Val) and five missense mutations were detected (Table 4). One MDR isolate harbored the mutation Ala110Val, and the C insertion at nucleotide position 133 of katG gene was detected in one XDR strain, Leu546Pro and Asp695Gly were found in one MDR isolate and one XDR strain, respectively, whereas Arg463Leu, a natural polymorphism, was identified in these strains (Table 4).

Collectively, five novel missense mutations were detected in katG: Trp191Stop, Thr271Pro, Trp328Phe, Leu546Pro, and Asp695Gly (Tables 3 and 4). Thr271Pro was found in one pre-XDR isolate and one XDR strain, respectively, while Trp191Stop and Trp328Phe were observed in two pre-XDR isolates, respectively. However, no katG mutations were found in the 20 pan-susceptible isolates. Detection of an nSNP within the region of katG analyzed exhibited a sensitivity of 75.3% and a specificity of 100% (Table 6).

Thirteen of the 77 INH-resistant isolates had a mutation within the mabA-inhA regulatory region. Among them, 9 (69.2%) had a C-to-T transition at position 15 upstream of the start site of mabA (Table 3), 3 had a transition from T to C at position 8 upstream of the mabA start site, and one had a transition from T to A at the nucleotide positioned 8 bases upstream of the mabA initiation codon. Of the 9 isolates with a C−15T mutation, 2 also had a katG Ser315Thr substitution, and 2 also had a katG Thr271Pro substitution. Of the three isolates with a T−8C mutation, one also had a katG Ser315Asn substitution, and 2 also had a katG Ser315Thr substitution. In contrast, none of the 20 pan-susceptible isolates possessed a mutation in the target fragment of mabA-inhA. Detection of an inhA promoter mutation was 16.9% sensitive and 100% specific. When the results for both katG and inhA are considered together, the assay sensitivity improves from 75.3 to 80.5%, and the specificity remains 100% (Table 6).

Ofloxacin and gyrA and gyrB.

Mutations in the quinolone resistance-determining region (QRDR) of gyrA, largely clustered in codons 90, 91, and 94, are the most important mechanism to confer FQ resistance (12, 15, 25). In the present study, we analyzed a 427-bp region of gyrA comprising the QRDR. Twenty pan-susceptible and 38 of 39 OFX-resistant isolates (23 pre-XDR and 16 XDR strains) were found to harbor the Glu21Gln and Ser95Thr mutations in gyrA (Table 2). Thirty-two OFX-resistant isolates displayed six different types of single-point mutations at codons 94, 90, and 91, with frequencies of 69.2, 7.7, and 5.1%, respectively (Table 3). The most common mutations were observed at codon 94 (n = 27), where the wild-type aspartic acid (GAC) was replaced with a glycine (Asp94Gly; n = 19), an alanine (Asp94Ala; n = 2), an asparagine (Asp94Asn; n = 5), or a tyrosine (Asp94Tyr; n = 1) (Table 2). In addition, three OFX-resistant isolates harbored the Ala90Val mutation. Ser91Pro was another mutation found to associate with the OFX-resistant isolates (n = 2). Seven OFX-resistant and 20 pan-susceptible isolates were determined to be wild type for the QRDR of gyrA. Further, a mutation Gly668Asp was detected in six of the seven OFX-resistant strains by means of complete sequencing for gyrA gene (Table 4). When the genotypic data for gyrA were compared to the drug susceptibility results for the OFX, the sensitivity and specificity values were determined to be 82.1 and 100%, respectively (Table 6).

One pre-XDR strain carried the Asp500His mutation, and another XDR isolate carried the Glu540Asp mutation, both in gyrB. The isolate with the latter mutation also had an Asp94Ala mutation in gyrA (Table 3). In contrast, other 37 OFX-resistant and 20 pan-susceptible isolates did not have any mutations in the examined 413-bp gyrB segment. In addition, no mutations were detected with complete sequencing of gyrB gene for the 37 OFX-resistant strains (Table 4). To determine the OFX phenotypes, the nSNPs of gyrB was analyzed, which showed a sensitivity of 5.1% and a specificity of 100% (Table 6). When the gyrA and gyrB data were combined, the overall sensitivity increased from 82.1 to 84.6% (Table 6).

Amikacin, kanamycin, capreomycin, and rrs.

AMK, KAN, and CAP are key drugs for treating MDR TB, and the most common mutations associated with a resistance of these drugs are located within the rrs gene coding for 16S rRNA (4, 19, 20). We therefore analyzed a 481-bp region of this gene. Fourteen of 19 resistant isolates (3 pre-XDR and 16 XDR strains) showed an A-to-G transition at nucleotide position 1401, one pre-XDR isolate resistant to AMK, KAN, and CAP had a C-to-T transition at nucleotide position 1402, while the remaining four strains had no mutation within the specific region (Table 3). In addition, no mutations were detected in complete sequencing of rrs gene for the four resistant isolates (Table 4). Generally, when the genotype data were compared to the susceptibility results, the sensitivity and specificity values were determined to be 78.9 and 100% (Table 6).

DISCUSSION

In this study, a detailed genetic analysis of the drug resistance of M. tuberculosis isolates was performed using the MIRU-VNTR method, which identified 77 resistant strains among patients in the Jiangxi province of China. The 77 strains showed a great genetic diversity compared to each other and to those reported elsewhere, suggesting that these strains were epidemically unrelated and had evolved under varied conditions of selection. They could have independently acquired drug resistance, and the acquisition could have been more dependent on the selective pressure of treatment failure. The latter varied individually, as a result of either inappropriate chemotherapy or poor adherence to treatment or inadequate monitoring (23), although transmission is now thought to play an increasing role (16, 55). Of note, the Beijing genotype was found to be predominant (80.5%) among these 77 strains.

Of the 77 resistant isolates, a high number (55.8%) had the Ser315Thr mutation in katG (Table 5). This mutation has occurred at different rates in different countries, correlating with the prevalence of TB (5). Where the prevalence is high, the rate is also high. This mutation has occurred among 59, 71, and 91% of isolates from India (34), Vietnam (10), and Russia (1), respectively. In contrast, it accounted for INH resistance only in 22% of the strains in Japan (5) and 23% of the strains in Singapore (29). In addition, five novel katG mutants were identified. Among them, the Trp191Stop mutation confers an INH resistance since it results in a pseudogene formation in katG. However, it remains to be ascertained whether the Thr271Pro, Trp328Phe, Leu546Pro, and Asp695Gly mutations are associated with INH resistance. Further studies are needed to assess their specific effects on katG function. The present study showed that 15 resistant strains lacked missense mutation in the target fragments of katG and mabA-inhA. The INH resistance of the 15 strains may have involved additional mechanism(s), possibly the mutations in inhA structural gene or in other loci, such as ndh, mshA, and ahpc (4, 9, 54).

For rapid identification of the MDR isolates, the detection of mutations within the RRDR of rpoB is of prime importance. This is based on the following lines of evidence. First, among the available anti-TB drugs, the molecular mechanism of RIF resistance is the most completely understood. Second, mutations within the RRDR of rpoB have been reported to occur in 95% or higher of the RIF-resistant isolates (9, 36, 54). Third, RIF resistance is an excellent surrogate marker for MDR M. tuberculosis (26, 37). In the present study, six novel rpoB mutations were identified, and they all fell outside of the RRDR region of the rpoB gene. Nevertheless, it is not clear whether these mutations are involved in the RIF resistance. Among them, five mutations (Ile572Met, Phe584Ser, Ile569Val, Ala615Met, and Asp626Glu) were detected in the four resistant strains, which each had another mutation within the RRDR of rpoB (Table 3). Do they represent compensatory mutations (6, 22), or are they related to RIF resistance? Further studies are therefore required to characterize their roles. In addition, rpoB His526Asp was identified in pre-XDR (5/26, 19.2%) and XDR (2/16, 12.5%) strains but absent in simple MDR strains (0/35, 0%). It is unclear whether rpoB His526Asp becomes a marker for pre-XDR and XDR M. tuberculosis strains from Jiangxi province, because a relatively smaller number of samples were analyzed in the present study; more samples are required to validate this.

Fluoroquinolones are the mainstay of treatment for patients with MDR and XDR TB since their inclusion in therapeutic regimens improves treatment outcome (11). In contrast, resistance to FQ increases the risk of treatment failure and death (8). Hence, FQ resistance among patients means a poor prognosis. In the present study, 82.1% of the OFX-resistant isolates were identified by screening the QRDR region of the gyrA gene. However, by analysis of nSNPs in gyrB, only 5.1% of these isolates can be verified. The latter result indicates that the inclusion of gyrB may not considerably change the performance of the genotyping assay. This finding is in agreement with previous observations (21, 51) that gyrB mutations are less common than gyrA mutations, and it may reflect the fact that the phenotypic resistance to FQ is mainly due to mutations in the QRDR of gyrA rather than gyrB, as suggested in a recent review and meta-analysis (12). In addition, the correlation between FQ resistance and the gyrB mutations (Asp500His and Glu540Asp), which was first reported by Duong et al. in 2009 (18), has not yet been validated. Given this reason, we should preferentially screen the gyrA gene rather than gyrB for FQ resistance in resource-constrained settings with high burdens of TB. Consistent with other reports (18, 50), we found here that 84.6% of the OFX-resistant isolates had gyrA or gyrB mutations in the QRDR region, whereas 6 strains did not. The latter finding suggests that an additional mechanism likely accounts for the OFX resistance. Possibilities include decreased cell wall permeability to the drug, increased expression of efflux pump, drug sequestration, or drug inactivation (25). Of particular note, two rarely reported mutations in gyrA—Glu21Gln (42) and Gly668Asp (28)—were detected in the present study. Similar to the gyrA Ser95Thr mutation, the two mutations are not associated with FQ resistance, and they are characterized as a natural polymorphism and marker for evolutionary genetics, which has been validated by Lau et al. recently (28). Therefore, we identified here three natural polymorphisms in the gyrA gene: Glu21Gln, Ser95Thr, and Gly668Asp.

Recently, more novel mutations were identified in M. tuberculosis clinical isolates (14, 38). However, some strains with these mutations were phenotypically sensitive. For example, a study (35) demonstrated that the strain harboring substitution Gly551Arg in gyrB is susceptible to FQ, and eight gyrB substitutions (Asp473Asn, Pro478Ala, Arg485His, Ser486Phe, Ala506Gly, Ala547Val, Gly551Arg, and Gly559Ala) are not responsible for quinolone resistance, although they have been observed previously in FQ-resistant strains. Therefore, the use of FQ should not be ruled out in the treatment of TB patients infected by stains harboring these mutations in gyrB. Thus, it is obviously important in adopting appropriate chemotherapy regimens to know which mutations confer phenotypic resistance and which do not.

Although amikacin and kanamycin are aminoglycoside antibiotics and capreomycin is a macrocyclic peptide antibiotic, cross-resistance among them has been well documented (9, 19, 27, 31). Common rrs mutations associated with resistance to these drugs include the A1401G, C1402T, and G1484T mutations (9, 31). In the present study, we observed two of these mutations (A1401G and C1402T), with A1401G (73.7%) being the more predominant (Table 5). It is worth noting that none of these isolates resistant to KAN, AMK, and/or CAP harbored the G1484T substitution, which in most cases is associated with high-level resistance to all three second-line injectable drugs (31). There are a few reasons that may explain this. First, a relatively smaller number of samples were analyzed in the present study, which may have limited the detection of the variety of variations. Second, different geographical regions may give rise to different pressure of artificial antibiotic selection on the corresponding strains. Third, the different regimes administered may have induced different mechanisms of drug resistance. In the present study, one pre-XDR isolate resistant to KAN, AMK, and CAP was also found to contain the C1402T mutation in rrs (Table 3), but the A1401G and G1484T mutations associated with AMK resistance in the rrs gene were not detected. Since the C1402T mutation has not been shown to confer resistance to AMK (19), this suggests at least another unknown mechanism for AMK resistance in the isolate. Similar to other studies (9, 21), no mutation was detected in the rrs gene in four resistant strains. The resistance phenotypes are probably attributed to the alterations within other loci, such as tlyA (for CAP resistance) (32) or the promoter region of eis (KAN), which confer low-level resistance to KAN (53), or an enhanced multidrug efflux pump (54).

Compared to the phenotypic data, the sensitivities for the detection of RIF, INH, OFX, and KAN/AMK/CAP resistance by DNA sequencing were 94.8, 80.5, 84.6, and 78.9%, respectively, which are similar to those demonstrated by the majority of studies elsewhere (2, 7, 9). The most common genetic alterations related to RIF, INH, OFX, and KAN/AMK/CAP resistance were rpoB Ser531Leu (44.2%), katG Ser315Thr (55.8%) and mabA-inhA C-15T (11.7%), gyrA Asp94Gly (48.7%), and rrs A1401G (73.7%), respectively (Table 5). In addition, RIF resistance is an excellent surrogate marker for MDR M. tuberculosis. On the basis of these facts, we believe that detecting the RRDR of rpoB alone is efficient for the accurate and rapid identification of MDR strains from Jiangxi province and that screening the RRDR of rpoB and the QRDR of gyrA is an economical and effective method for the prediction of pre-XDR strains. Moreover, we believe that examining three loci (rpoB, gyrA, and rrs) simultaneously is suitable for the rapid identification of XDR strains in the region.

To aid in rapid diagnosis of MDR and XDR TB strains, two commercially available DNA strip assays, the GenoType MTBDRplus and the GenoType MTBDRsl assay (Hain Lifescience, Nehren, Germany), have been recently developed to detect genetic mutations associated with resistance to the first-line drugs (INH and RIF), and second-line injectable drugs and fluoroquinolones as well (7, 33). According to the preliminary sequencing results obtained in the present study and the mutations involved in drug resistance identified by the two assays, it has been speculated that the GenoType MTBDRplus assay might have been able to identify ca. 90% of RIF-resistant isolates, ca. 70% of INH-resistant isolates, and ca. 65% of MDR isolates from Jiangxi region. Meanwhile, the GenoType MTBDRsl assay might have been able to detect ca. 80% of the OFX resistance and 78% of the resistance to KAN, AMK or CAP. In combination with the MTBDRplus assay, the GenoType MTBDRsl assay might have been able to identify 75% of XDR isolates (see Table S1 in the supplemental material). Therefore, we conclude that the GenoType MTBDRplus assay might have performed equally well for the detection of RIF resistance in Jiangxi province. However, because the mechanisms of drug resistance in M. tuberculosis are complicated and not yet fully understood and because resistance-associated mutations vary geographically, the GenoType MTBDRplus assay might have demonstrated a relatively poor prediction for INH resistance in this area. The performance of the two assays for detecting MDR and XDR strains from the region are needed to evaluate in a real-world setting.

In summary, our results indicate that most frequent mutations in the rpoB, katG, inhA, gyrA, gyrB, and rrs genes are consistent with those reported from other regions of the world (2, 9, 54), suggesting that these mutations have global ramifications. Five novel mutations in katG gene and six new alleles in rpoB gene were identified, which will broaden current knowledge of molecular basis of drug resistance in M. tuberculosis. Further studies are needed to elucidate their actual roles in the drug resistance of M. tuberculosis. These new mutations, together with existing mutations, can be used to design diagnostic tests utilizing other mutation detection technologies such as the line probe assay or DNA microarrays. The molecular information presented here will be of particular value in the rapid clinical detection of MDR and XDR M. tuberculosis strains in the Jiangxi province in the near future.