Identification of Mycobacterial Species by PCR Sequencing of Quinolone Resistance-Determining Regions of DNA Gyrase Genes

Abstract

The determination of the amino acid sequence of quinolone resistance-determining regions (QRDRs) in the A and B subunits of DNA gyrase is the molecular test for the detection of fluoroquinolone resistance in mycobacteria. We looked to see if the assignment of mycobacterial species could be obtained simultaneously by analysis of the corresponding nucleotide sequences. PCR sequencing of gyrA and gyrB QRDRs was performed for 133 reference and clinical strains of 21 mycobacterial species commonly isolated in clinical laboratories. Nucleotide sequences of gyrA and gyrB QRDRs were species specific, regardless of fluoroquinolone susceptibility.

During the last few years, the emergence of infections caused by nontuberculous mycobacteria and the report of tuberculosis outbreaks has brought considerable interest in clinical mycobacteriology. It led to the development of new diagnosis tools based on molecular technology for antibiotic susceptibility testing and identification (24). Rapid identification of mycobacteria to species level is recommended in clinical laboratories for assessing the diagnosis of mycobacteriosis and also for making decisions for effective therapy (2).

In the clinical laboratory, the differentiation of closely related species of mycobacteria by phenotypic and biochemical tests remains difficult for some very common species. The phenotypic methods are slow, require expertise, and often use nonstandardized reagents (8). New biochemical methods such as high-performance liquid chromatography of mycolic acids are implemented only in specialized laboratories (4). Molecular methods were developed with enthusiasm because they are rapid (no need to subgrow bacteria) and require a small quantity of bacteria. The reference molecular method for identification is the determination of sequences of 16S ribosomal DNA (rDNA) (21), but identical sequences were reported for some common species. Other DNA sequences or genes have been described for the differentiation of mycobacterial species, such as the internal transcribed spacer (ITS) 16S-23S (28), recA (3), dnaJ (34), hsp65 encoding the 65-kDa heat shock protein (27), rpoB encoding the B subunit of RNA polymerase (20), the gene of the 32-kDa protein (30), sod encoding the superoxide dismutase (41), and gyrB encoding the B subunit of DNA gyrase (18). None of these genes can presently differentiate all the mycobacterial species commonly isolated in the clinical laboratory.

Fluoroquinolones are active against mycobacteria (23) and are recommended for the treatment of drug-resistant tuberculosis, drug-resistant leprosy, and infections caused by some nontuberculous mycobacteria (2, 9, 14). Fluoroquinolone resistance is mainly due to alterations in DNA gyrase, the unique type II topoisomerase of mycobacteria (1, 6, 7, 22, 35). These alterations are substitutions in the quinolone resistance-determining regions (QRDRs) in the A subunit (region 67 to 106) (numbering system used for Escherichia coli) and in the B subunit (region 426 to 464), as described for other bacteria (11). The detection of missense mutations at positions 83, 84, and 87 in GyrA and positions 426, 447, and 464 in GyrB is a rapid and efficient test for molecular detection of fluoroquinolone resistance in mycobacteria (5, 22, 26). This test is complementary to the antibiotic susceptibility testing that is not standardized yet for quinolones.

In the National Reference Center laboratory, we implemented a few years ago the detection of fluoroquinolone resistance in pathogenic mycobacteria by PCR sequencing of the gyrA and gyrB QRDRs. Therefore, we were able to compare the corresponding nucleotide sequences of different mycobacterial species. In a previous work, it was shown, for a few mycobacterial species, that gyrA and gyrB sequences can differentiate between some species and help in phylogenetic analysis (15). In the present study, PCR sequencing of gyrA and gyrB QRDRs was tested to differentiate 21 mycobacterial species commonly isolated in the clinical laboratory, out of which 17 were pathogenic and 4 were nonpathogenic. Reference and clinical strains, wild-type strains for fluoroquinolone susceptibility, and strains with acquired resistance to fluoroquinolones were studied.

A total of 133 strains representing 21 mycobacterial species have been studied: Mycobacterium tuberculosis (n = 21; H37Rv strain, 20 clinical strains of which 10 were fluoroquinolone-resistant strains), M. bovis (n = 7; ATCC 2001, five clinical strains and M. bovis BCG), M. africanum (n = 4; ATCC 30007 and three clinical strains), M. xenopi (n = 6; ATCC 19250 and five clinical strains), M. avium (n = 8; ATCC 25291, six clinical strains and one in vitro fluoroquinolone-resistant mutant), M. intracellulare (n = 5; ATCC 13950 and four clinical strains), M. gordonae (n = 5; ATCC 14470 and four clinical strains), M. kansasii (n = 4; ATCC 12478 and three clinical strains), M. gastri (n = 3; ATCC 15754 and two clinical strains), M. malmoense (n = 2; two clinical strains), M. szulgai (n = 4; NCTC 10831 and three clinical strains), M. simiae (n = 4; ATCC 25275 and three clinical strains), M. leprae (n = 12; 12 clinical strains, of which one is a fluoroquinolone-resistant strain), M. marinum (n = 5; ATCC 927 and four clinical strains), M. ulcerans (n = 6; ATCC 14188 and five clinical strains), M. chelonae (n = 3, ATCC 14472 and two clinical strains), M. abscessus (n = 6; ATCC 19977 and five clinical strains), M. fortuitum (n = 10; ATCC 6841 and one in vitro fluoroquinolone-resistant mutant and eight clinical strains, of which one is a fluoroquinolone-resistant strain), M. peregrinum (n = 7; ATCC 14467 and one in vitro fluoroquinolone-resistant mutant and five clinical strains), M. smegmatis (n = 8; ATCC 19420, mc²155 and three in vitro fluoroquinolone-resistant mutants, and NCTC 53 and two in vitro fluoroquinolone-resistant mutants), and M. aurum (n = 3; ATCC 23366, CIPT 141210005 and one in vitro fluoroquinolone-resistant mutant). Fluoroquinolone-resistant strains were described previously (5, 6, 7, 15, 16). All the clinical strains were identified by classical phenotypic and biochemical tests (8) and molecular reference tests. These molecular tests were DNA probes (13) (Accuprobe and Genprobe; Biomérieux, Marcy L'Etoile, France) for M. tuberculosis complex, M. avium, M. intracellulare, M. gordonae, and M. kansasii and sequencing of the 16S rDNA (21) or of the hsp65 gene (27) for the other species. Extraction of mycobacterial DNA and amplification of the DNA fragments corresponding to the gyrA and gyrB QRDRs were performed as previously described (15, 16) for gyrA by using the degenerated oligonucleotides Pri9 (5′-CGCCGCGTGCTG/CATGCA/GATG-3′) and Pri8 (5′-C/TGGTGGA/GTCA/GTTA/GCCC/TGGCGA-3′) and for gyrB by using GyrbA (5′-GAGTTGGTGCGGCGTAAGAGC-3′) and GyrbE (5′-CGGCCATCAA/GCACGATCTTG-3′). The amplification reactions consisted of the following steps: one denaturation cycle at 94°C for 10 min and 40 cycles of amplification at 94°C for 1 min, 55°C for 1 min, and 72°C for 1 min, followed by one elongation cycle at 72°C for 10 min. Sequencing of the gyrA and gyrB QRDRs was performed as previously described (16).

Amino acid sequences of GyrA and GyrB QRDRs were identical for all the mycobacterial species, except in two cases. In the first case, 1 amino acid was different between the GyrA sequences of M. fortuitum, M. peregrinum, and M. aurum, which harbored a serine at position 83, and that of the other mycobacterial species, which harbored an alanine at position 83 (16). This difference, which we reported previously, has been related to the intrinsic quinolone susceptibility of the former three species. In the second case, 1 amino acid was different between the GyrA sequences of different strains of M. tuberculosis, with either a serine or a threonine at position 88 (35). The Ser88-Thr natural polymorphism has been related to the phylogenetic origin of the strains. The strains with Thr88 are ancestral to those with Ser88 and are more frequent (31).

Nucleotide sequences of the gyrA QRDR (120 bp) and of the gyrB QRDR (117 bp) were, overall, highly conserved among all the strains tested, the similarity values ranging between 75 and 100% for the gyrA QRDR and 79 to 100% for the gyrB QRDR (Table 1). Such similarity values between mycobacterial species have been reported for the genes used for molecular identification to species level (31). The gyrA and gyrB nucleotide sequences were compared for all the species (interspecies similarity) and for all the strains within each species (intraspecies similarity).

TABLE 1.

Highest similarity values (%) between the sequences of gyrA QRDR (lower left) and of gyrB QRDR (upper right) from species of the Mycobacterium genus

Mycobacterial species no. (name)	Similarity (% with species no.)
	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21
1. (M. tuberculosis)		100	100	85.5	88.0	87.2	88.9	85.5	83.8	88.9	88.9	83.8	87.2	86.3	81.2	84.6	88.0	84.6	84.6	87.2	87.2
2. (M. bovis)	100		100	85.5	88.0	87.2	88.9	85.5	83.8	88.9	88.9	83.8	87.2	86.3	81.2	84.6	88.0	84.6	84.6	87.2	87.2
3. (M. africanum)	100	100		85.5	88.0	87.2	88.9	85.5	83.8	88.9	88.9	83.8	87.2	86.3	81.2	84.6	88.0	84.6	84.6	87.2	87.2
4. (M. xenopi)	90.8	90.8	90.8		91.4	94.0	90.6	91.4	93.2	93.2	94.9	89.7	88.9	88.9	82.9	87.2	94.0	92.3	95.7	90.6	95.7
5. (M. avium)	90.8	90.8	90.8	90.0		95.7	93.2	92.3	90.6	94.0	94.0	88.0	89.7	88.9	80.3	91.4	94.9	94.0	94.0	90.6	92.3
6. (M. intracellulare)	90.0	90.0	90.0	89.2	96.7		90.6	95.7	94.9	94.0	93.2	89.7	90.6	88.9	80.3	91.4	94.9	94.0	94.0	89.7	94.0
7. (M. gordonae)	92.5	92.5	92.5	90.8	90.8	93.3		92.3	91.4	92.3	95.7	88.0	88.9	88.9	80.3	85.5	92.3	89.7	89.7	95.7	91.4
8. (M. kansasii)	92.5	92.5	92.5	90.8	92.5	95.8	91.7		96.6	91.4	92.3	89.7	87.2	87.2	79.5	88.0	93.2	90.6	91.4	91.4	91.4
9. (M. gastri)	92.5	92.5	92.5	88.3	90.0	90.8	90.8	93.3		91.4	94.0	88.9	87.2	87.2	79.5	85.5	90.6	89.7	91.4	89.7	90.6
10. (M. malmoense)	90.0	90.0	90.0	90.8	91.7	93.3	94.2	92.5	90.8		94.9	87.2	90.6	89.7	80.3	90.6	94.0	94.0	93.2	90.6	94.0
11. (M. szulgai)	88.3	88.3	88.3	90.0	92.5	92.5	91.7	91.7	90.0	88.3		86.3	88.9	88.0	82.9	88.0	93.2	92.3	92.3	91.4	93.2
12. (M. simiae)	94.2	94.2	94.2	90.8	90.0	92.5	93.3	95.0	89.2	90.0	90.0		89.7	89.7	80.3	85.5	89.7	88.0	90.6	92.3	91.4
13. (M. marinum)	88.3	88.3	88.3	90.0	91.7	92.5	94.2	91.7	89.2	92.5	90.8	91.7		100	83.8	87.2	93.2	90.6	92.3	88.9	90.6
14. (M. ulcerans)	88.3	88.3	88.3	89.2	91.7	92.5	94.2	91.7	89.2	91.7	90.8	91.7	100		82.9	86.3	92.3	89.7	91.4	88.9	89.7
15. (M. leprae)	80.8	80.8	80.8	81.7	84.2	86.7	80.8	85.0	83.3	83.3	85.0	80.8	82.5	81.7		82.0	79.5	78.6	82.0	82.0	80.3
16. (M. chelonae)	90.8	90.8	90.8	86.7	86.7	88.3	90.0	90.0	86.7	89.2	86.7	88.3	87.5	87.5	80.0		93.2	93.2	92.3	88.0	89.7
17. (M. abscessus)	90.8	90.8	90.8	87.5	85.8	86.7	91.7	87.5	87.5	87.5	85.8	87.5	88.3	88.3	75.0	89.2		97.4	96.6	91.4	94.9
18. (M. fortuitum)	89.2	89.2	89.2	89.2	90.0	90.8	92.5	91.7	90.0	90.0	90.0	90.0	91.7	91.7	77.5	84.2	88.3		94.9	89.7	92.3
19. (M. peregrinum)	89.2	89.2	89.2	88.3	92.5	94.2	90.8	91.7	87.5	89.2	90.0	91.7	90.0	90.0	80.0	85.0	85.8	92.5		89.7	94.0
20. (M. smegmatis)	93.3	93.3	93.3	90.8	89.2	90.0	92.5	90.8	90.0	88.3	90.8	91.7	90.0	90.0	80.0	87.5	90.0	90.8	92.5		93.2
21. (M. aurum)	90.8	90.8	90.8	87.5	88.3	91.7	91.7	90.0	90.0	87.5	88.3	90.8	89.2	89.2	80.8	87.5	88.3	91.7	93.3	94.2

Interspecies comparison showed that nucleotide sequences of the gyrA and gyrB QRDRs were species specific; i.e., they were clearly different from one species to another (Fig. 1; Table 1). Species that are closely related by either phenotypic or biochemical characters or ribosomal sequences had different gyrA and gyrB QRDR sequences. For instance, M. kansasii and M. gastri, which have the same 16S and ITS rRNA sequences, had specific gyrA and gyrB QRDR sequences, with 8- to 10-nucleotide differences (highest similarity value of 93.3%) between the gyrA QRDRs and 4- or 5-nucleotide differences (highest similarity value of 96.6%) between the gyrB QRDRs. M. szulgai and M. malmoense, which are not differentiated by their 16S rDNA sequences, were differentiated by 14 or 15 nucleotides (highest similarity value of 88.3%) between the sequences of gyrA QRDRs and by 6 to 9 nucleotides (highest similarity value of 94.9%) between the sequences of gyrB QRDRs. M. avium and M. intracellulare, which belong to the same complex, were differentiated by 4 to 7 nucleotides between the gyrA QRDRs and 5 to 9 nucleotides between the gyrB QRDRs. Scotochromogen species, such as M. gordonae, M. szulgai, and M. aurum, were differentiated with regard to the gyrA QRDR sequences (10-, 14-, and 10-nucleotide differences, respectively) and with regard to the gyrB QRDR sequences (4- to 7-, 8- to 11-, and 10- or 11-nucleotide differences, respectively). Finally, gyrA and gyrB sequencing was efficient for the differentiation of rapidly growing mycobacterial species. Nucleotide sequences of gyrA and gyrB QRDRs were clearly different not only between the M. chelonae group (M. chelonae and M. abscessus) and the M. fortuitum group (M. fortuitum and M. peregrinum) but also between the two species within each group (Fig. 1). Precisely, the sequences of the gyrA QRDRs from M. chelonae and M. abscessus differed by 13 or 14 nucleotides and the gyrB QRDR sequences differed by 8 to 14 nucleotides. The sequences of the gyrA QRDRs from the strains of M. fortuitum and M. peregrinum differed by 9 nucleotides, and the gyrB QRDR sequences differed by 6 or 7 nucleotides. However, species within the M. fortuitum group appeared to form an homogeneous cluster and appeared well differentiated from the M. chelonae group (Fig. 1; Table 1), which is consistent with published data (39).

FIG. 1.

Alignment of the nucleotide sequences of the gyrA and gyrB QRDRs from the 21 mycobacterial species. Sequences of M. tuberculosis were taken as the reference sequences, and dashes represent identical nucleotides. The nucleotide polymorphisms, i.e., a nucleotide difference that has been observed between strains within the same species, were indicated by a letter in boldface type, and their meaning is the following: the letter R when A or G was observed, Y for C or T, M for A or C, K for G or T, S for G or C, and W for A or T.

The nucleotide sequences of gyrA and gyrB QRDRs did not discriminate with regard to the species that belong to the M. tuberculosis complex, i.e., M. tuberculosis, M. africanum, and M. bovis. This was expected, since so far the sequences of the genes used for identification to species level do not differentiate these species (31). The nucleotide sequences of gyrA and gyrB QRDRs did not differentiate M. marinum from M. ulcerans, either. Nearly all of the genes sequenced so far and used for identification to species level were identical for M. ulcerans and M. marinum (32).

Intraspecies similarity was studied by sequencing the gyrA and gyrB QRDRs of 3 to 10 wild-type strains of each species. Similarity ranged from 97.5 to 100%. Nucleotide differences were observed rarely between the strains of a same species, with the exception of fluoroquinolone-resistant mutants. The intraspecies differences were considered a natural polymorphism of the sequence and are indicated in Fig. 1. Since only some species were concerned, such as M. avium, M. intracellulare, M. kansasii, and M. abscessus, it might be due to the taxonomic heterogeneity of the species (38).

Molecular methods that were described for the identification of mycobacteria often require PCR sequencing of long DNA fragments. This can be circumvented by using PCR-restriction fragment length polymorphism, and some laboratories found it simple and inexpensive (10, 12, 17, 19, 29). However, they also reported disadvantages of PCR-restriction fragment length polymorphism, such as a lack of specificity, the need for a large panel of control species, and the fact that new species are undetected (37, 40). Hybridization to oligonucleotide probes (LiPA and Chip) is a simple and robust technique but has been so far applied only to 16S rDNA and ITS rDNA sequences (25, 33, 36). The work flow of the technique that we described herein is simple: one PCR at a unique hybridization temperature for the two genes followed by a short sequencing (120 bp only for each gene). Nowadays, amplification and sequence determination are implemented in most of the hospitals or can be done outside for a low price.

Analysis of the nucleotide sequences of gyrA and gyrB QRDRs can rapidly determine the mycobacterial species. Rapid identification of nontuberculous mycobacteria to species level is particularly useful, because the antibiotic susceptibility pattern and the clinical interest vary depending on the mycobacterial species (2). Moreover, for pathogenic mycobacteria, this test can simultaneously give an answer regarding susceptibility to quinolones. The mutations that we have observed in quinolone-resistant strains of M. tuberculosis complex, M. fortuitum, M. leprae, M. avium, M. smegmatis, and M. peregrinum did not result in a sequence specific to another species. Conversely, nucleotide differences observed between species were different from the mutations involved in fluoroquinolone resistance.