Mutations Found in embCAB, embR, and ubiA Genes of Ethambutol-Sensitive and -Resistant Mycobacterium tuberculosis Clinical Isolates from China

To better understand the molecular mechanisms of Ethambutol (EMB) resistance, the mutant hot spot region of five genes (embB, embA, embC, embR, and ubiA) was amplified and sequenced in 109 EMB-resistant and 153 EMB-susceptible clinical isolates from China. Twenty-seven EMB-susceptible isolates were found to have nonsynonym mutations, 23 of which were in embB. The mutations occurred most frequently in embB (85.3%, 93) and were seldom in embC (2.8%, 3), embA (3.7%, 4), embR (3.7%, 4), and ubiA (8.3%, 9) in EMB-resistant isolates. For the embB gene, 63 isolates showed mutations at embB306, 20 at embB406, nine at embB497, and five at embB354 in EMB-resistant isolates. In addition, the particular mutants at embB406 and embB497 indicated both high levels of EMB resistance (MICs > 5 μg/mL) and broad anti-TB drug resistance spectrums. Our data supported the facts that embB306 could be used as a marker for EMB resistance with a sensitivity of 57.8% and a specificity of 78.8%.


Introduction
Ethambutol (EMB) is an antituberculosis drug that is widely used for treating drug resistance, and it is also commonly used for treating multidrug-resistant tuberculosis [1]. The collective results of the EMB drug susceptibility test (DST) of clinical M. tuberculosis strains, which has been extensively reviewed in many countries, indicate that many of the strains are resistant to EMB [2][3][4]. Due to the numerous EMBresistant strains, the mechanisms underlying EMB resistance, namely, mutations related to EMB target genes, have been both investigated and summarized [4][5][6][7][8].
EMB appears to inhibit arabinosyl transferases encoded by the embCAB operon, which is involved in polymerizing arabinose into the arabinan components of arabinogalactan and lipoarabinomannan. The mutations in the emb-CAB operon are responsible for its resistance, especially the "canonical" mutations in codons 306, 406, or 497 of embB [4,9,10]. Belanger et al. (1996) reported that embR modulates the level of arabinosyltransferase activity in vitro, which might confer EMB resistance [11]. embR may control arabinosyltransferase activity in M. tuberculosis in a phosphorylation-dependent fashion, acting downstream of the Ser/Thr-kinase PknH [11]. Recently, Safi et al. found that mutations on the gene ubiA were associated with high-level resistance and had multiplicative effects with embB mutations on minimum inhibitory concentrations (MICs) [4]. The ubiA gene encoding 5-phospho-alpha-d-ribose-1-diphosphate:decaprenyl-phosphate 5phosphoribosyltransferase is known to be essential for the growth of M. tuberculosis, and EMB was found to inhibit other steps in arabinan biosynthesis [12]. Although EMB does not directly inhibit ubiA, ubiA mutations have been shown to increase DPA synthesis, causing the MICs to increase in both a wild-type background and an embB codon 306 mutant background [4].
The embB mutations that are related to the EMB target genes have been extensively studied, but studies regarding embA, embC, embR, and ubiA mutations are lacking. Moreover, less data has been generated on the simultaneous presence of these particular gene mutations in a large amount of clinical isolates. In this study, we sequenced the five genes to find the concomitant existence of the mutations in 109 clinical isolates. This study was an important step towards gaining a full understanding of the molecular mechanisms of EMB resistance and the mutation patterns in clinical isolates from China.

Bacterial Strains and Susceptibility
Testing. M. tuberculosis H37Rv (ATCC 27294), which was used as the control for the antibiotic susceptibility test, was obtained from the Beijing Bio-Bank of Clinical Resources on Tuberculosis for the isolates. From January 1, 2009, to December 31, 2009, a total of 109 EMB-resistant clinical M. tuberculosis isolates were collected from 1,048 isolates. The absolute two-concentration method was conducted twice in order to determine the low or high resistance levels on Lowenstein-Jensen (L-J) slants [13]. The 109 EMB-resistant and 153 randomly selected EMBsusceptible isolates included in this study were also subjected to susceptibility testing for isoniazid (INH), rifampicin (RFP), streptomycin (SM), EMB, ofloxacin (OFX), Capreomycin (CPM), para-aminosalicylic acid sodium (PAS), and amikacin (AMK) by the absolute concentration method [9]. The INH, RFP, OFX, SM, EMB, CPM, PAS, and AMK were purchased from Sigma-Aldrich (Beijing, China). They were dissolved to 100-fold concentrated stock solutions according to the manufacturer's instructions.

DNA Extraction and PCR Amplification.
The genomic DNA from the samples was isolated from the mycobacterial cultures using the QIAmp DNA Mini Kit (Qiagen, CA, USA). PCR was performed for various gene loci of embC, embA, embR, ubiA, and the enlarged embB primers. The primer sets that were used are described in Table 1 (Table 1), were the ones that had been previously reported. The DNA templates for the PCR products were purified using a QIAquick PCR Cleanup Kit (QIAGEN, CA, USA) as per the manufacturer's instructions and were subjected to DNA sequencing.

Sequencing and Data
Analysis. All PCR products that were utilized in this research were sequenced by Sangon Co. Ltd. in China. The sequencing data was assembled by SeqMan Pro (version 7.1, DNAstar Lasergene), and the mutations that were uncovered were identified by comparison with the H37Rv sequences (NC 000962.3) of embB, embA, embC, embR, and ubiA from the GenBank database (http://www.ncbi.nlm.nih.gov/nuccore/NC 000962.3) using the MegAlign (version 7.1, DNAstar Lasergene). Both the frequency calculations and the association analyses were performed using GraghPad 5 for Windows (GraghPad, Inc., USA). To further investigate the drug-resistant spectrum, the 109 EMB-resistant isolates and 153 randomly selected EMBsensitive isolates were also subjected to susceptibility testing for INH, RFP, SM, EMB, OFX, CPM, PAS, and AMK. Results showed that the EMB-resistant isolates were resistant to an average of 3.49 ± 1.59 (mean ± SD) of the eight tested anti-TB drugs, while the EMB-sensitive isolates were resistant to an average of 0.72 ± 0.41 (mean ± SD) of the eight tested anti-TB drugs. The highly EMB-resistant isolates were resistant to an average of 4.58 ± 1.96, whereas the isolates with low EMB resistance were resistant to an average of 2.94 ± 1.02. The 153 EMB-sensitive isolates were resistant to an average of 0.73 ± 0.42 of the eight tested anti-TB drugs.

Mutations in the Tested Genes.
Of all the 153 EMBsusceptible isolates, only one or two were found to have nonsynonym mutations in embC, embA, embR, and ubiA. Among the 109 EMB-resistant isolates, there were one, four, four, and six isolates with nonsynonym mutations in the embC, embA, embR, and ubiA, respectively. The mutation pattern in embA included V343L, L105V, and R380P, for EMBresistant, and V343L, R380P for EMB-susceptible isolates. In embR, nonsynonym mutations occurred at P49A, S104N, and P243S in EMB-resistant isolates and at L125S and R230W in EMB-susceptible isolates. Nonsynonym mutations occurred at S244T, I179T, E149D, and A38T in EMB-resistant isolates and none in EMB-susceptible isolates in ubiA. Only one nonsynonym mutation was found in embC at E305D, which was found in EMB-resistant isolates, and a synonym mutation occurred at E305E, which was found in both EMB-resistant and EMB-susceptible isolates.

Correlation between Mutations and Drug Resistance.
Mutations at embB497, embB354, and embB328 were found only in EMB-resistant clinical isolates, and they were considered to correlate to EMB resistance. Mutations at embB406 and embB306 were also found mainly in EMB-resistant isolates, and they were correlated to EMB resistance with an odd ratio (OR) of 50.7 ( < 0.001) and 46.5 ( < 0.001), respectively.
Of all the 109 EMB-resistant isolates, more than 18 isolates were found to have mutations (including synonym mutations) in at least two of the five tested genes. When mutations occurred in more than two of the five tested genes, high levels of EMB resistance occurred (OR = 6.2; = 0.001); isolates with mutations in two or more of the tested genes were resistant to more anti-TB drugs (5.87 ± 1.60) than those with mutations in only one of the tested genes (3.03 ± 1.37). Some strains with certain mutation patterns showed broad anti-TB drug-resistant spectrums. The average number of resistant anti-TB drugs for the mutant at embB306, embB328, embB354, embB406, and embB497 was 2.96 ± 1.07, 4.33 ± 1.53, 2.60 ± 0.89, 5.19 ± 1.23, and 5.75 ± 0.88, respectively.

Discussion
EMB is an important antimycobacterial drug and is recommended to treat tuberculosis as well as opportunistic infections by M. avium in patients with acquired immunodeficiency syndrome [14]. However, EMB resistance has been reported frequently in many countries. The traditional views of the mechanisms for EMB resistance mainly focus on the mutations of the embB gene, which creates resistance by altering drug-protein interaction. Including the embCAB operon, the transcriptional regulators embR and ubiA have also been associated with EMB-resistant M. tuberculosis [15]. In the present study, we sequenced the embCAB operon, embR, and ubiA in 109 EMB-resistant and 153 EMB-sensitive M. tuberculosis isolates to find the relationships between the mutations and drug resistance.
Our data supported that mutations in codon embB were the predominant mechanism associated with EMB resistance, since 85.3% (93/109) were found to be mutated in EMBresistant isolates and 15% (23/153) were found to be mutated in EMB-susceptible isolates. High mutation frequencies in     BioMed Research International Table 3: Mutants in embB, embA, embC, embR, and ubiA within the EMB-sensitive isolates.
In this study, mutations occurred at the embB codon in 55% of the EMB-resistant clinical isolates. Our data supported the facts that embB306 could be used as a marker for EMBresistance with a sensitivity of 57.8% and a specificity of 78.8%. A different frequency of the mutation patterns in the embB gene was reported in India. Of all the 52 different positions that were investigated, the most commonly found mutations were located at codon 378 (11), followed by mutations at codons 368 (9), 306 (8), 380 (7), and 406 (6) [21]. This discrepancy may be due to heterogeneity in the methodologies used (e.g., drug susceptibility testing methods) or to the intrinsic molecular variability between isolates from diverse geographical regions.
In this study, all the EMB-resistant isolates with embB497 or embB406 mutations were MDR-TB, which was consistent with the facts reported by Shi et al. and Srivastava et al. [17,21]. Moure et al. also reported that the percentage of multidrug resistance among isolates with at least one embB406 substitution was significantly higher than that found in the group of isolates without mutations in this codon (100% versus 73.1%, = 0.035). In our report, both higher drug resistance level and broader anti-TB drug spectrum were found in EMBresistant isolates with embB497 or embB406 mutations than in those with embB306, embB328, or embB354 mutations.
Mutation in embB gene showed lots of patterns in different countries or regions, but quite different for embC, embA, and embR. In this study, the mutations were found mostly in embB (85.3%) and less in ubiA (8.26%), embA (3.7%), embC (4.6%), and embR (3.7%) in EMB-resistant isolates. In congruence with similar studies conducted in Taiwan, nonsynonymous mutations in embC (1), embA (4), and embR (3) were only rarely encountered in this study [24]. Ramaswamy et al. first reported two nonsynonymous nucleotide substitutions in embR resulting in C110W and Q379D replacements [23]. Later, several EMB resistanceassociated polymorphisms in embR (16/44; 36.3%) were found in India [21]. In mainland China, 2 of 77 EMB-resistant MDR isolates and 4 of 56 EMB-sensitive isolates were found to have mutations in embC, and 5 of 74 EMB-resistant MDR-TB and 6 of 54 EMB-sensitive MDR-TB were found to have mutations in embA [25]. In New York, USA, embC had only 2 EMB resistance-associated nonsynonymous, N394D and R738E, in 75 EMB-resistant samples, and 8 EMB resistanceassociated amino acid replacements were identified in embA [23]. No mutation was identified in the embA gene isolated in India [26], but novel mutations at A254, L251R, T270I, and 297 (11/44) were found in embC [21,27,28], which we did not find in this study. Some evidence also supported that T270I changed on its own and plays no role in EMB resistance in embC [28] and that T270I is not a marker for EMB resistance in the M. tuberculosis complex [29]. Mutations in ubiA were reported in 19 of 63 that were randomly selected from the World Health Organization Special Programme for Research and Training in Tropical Disease strain bank and in 17 of the 89 isolates from China [30].
To confirm the mutations in the various genes described in the aforementioned literature, allelic exchange experiments were carried out. Safi et al. confirmed that mutations of M306V, M306L, M306I (ATA), and M306I (ATC) all caused EMB resistance (MIC = 4 mg/L) when incorporated into wild-type strains 210 and 5310 [31]. The fold increase in EMB MIC was also investigated for M306V, M306I (ATA), and M306I (ATC) that had been introduced into H37Rv by Starks et al. [26] and Plinke et al. [32]. Safi et al. also looked at the role of common mutations found in clinical strains with high-level EMB resistance at the embB 406 and 497 codons [10]. By introducing the point mutation in embC, Goude et al. verified that the mutations D294G, M300L, and M300V increased susceptibility to EMB and that mutation M300I had no resistance effect [28]. The introduction of Rv3806c mutations into either codon 18, 188, 237, 240, 249, 174, 176, or 175 caused the increase of EMB MIC [4,30], but not into codon 149 [4]. Newly found mutations at codons 38,254,198, and 249 of Rv3806c must be studied further.
In this study, no mutations were found at the tested sites in 14 of the total 109 isolates and 17 EMB sensitive isolates were found with the mutations of embB306. The discrepancy in drug susceptibility between the phenotype and the genotype was multifactorial. Those factors included the overlapping of the MIC distributions between the wildtype and mutant strains [33], the heteroresistance from the bacterial population [22,34], the limitation of the current DST [35,36], and the bacterial itself changes in the cell wall thickness, the efflux pump activity and mutations at other genes not included in this study [25,38].

Conclusion
In conclusion, we have demonstrated that mutations were frequently found in the embB gene, especially in EMBresistant M. tuberculosis strains. The embB306, embB497, and embB406 mutation patters were ranked as the top three in mutation frequency and were found to be associated with EMB resistance. In addition, the particular mutants at embB406 and embB497 indicated both high levels of EMB resistance (MICs > 5 g/mL) and broad anti-TB drug resistance spectrums. The features of EMB resistance revealed in this study will increase our understanding of the distribution and frequency of mutations in M. tuberculosis isolates with EMB resistance in TB patients from China.