rpoB sequence-based identification of Mycobacterium avium complex species

The Mycobacterium avium complex (MAC) comprises slowly growing mycobacteria (Tortoli et al., 1999; Turenne et al., 2007; Yajko et al., 1995) responsible for opportunistic infections and zoonoses (Karakousis et al., 2004; Lauzi et al., 2000). M. avium has been further divided into M. avium subsp. avium, M. avium subsp. paratuberculosis (Lauzi et al., 2000), M. avium subsp. silvaticum (Thorel et al., 1990) and M. avium subsp. hominissuis (Mijs et al., 2002). It has recently been shown that M. avium subsp. hominissuis comprises a heterogeneous group of organisms, with M. avium subsp. avium/silvaticum and M. avium subsp. paratuberculosis being two independent, specialized clones (Turenne et al., 2007). Mycobacterium intracellulare has been isolated more frequently in HIV-negative patients (Devallois et al., 1996). Two novel species, Mycobacterium chimaera (formerly sequevar MAC-A) (Tortoli et al., 2004) and Mycobacterium colombiense (formerly sequevar MAC-X) (Murcia et al., 2006), have recently been described (Table 1).

Table 1. Specific host and infectious diseases related to seven MAC reference strains

Since phenotypic tests do not easily discriminate between these closely related species and subspecies, molecular identification has been developed. 16S rDNA sequencing has demonstrated low variability when compared with the Lipav2 assay (Lebrun et al., 2005). M. avium serotype 2 SalI-EcoRI DT6 and SalI-BamHI DT1 fragment probing (Thierry et al., 1993) and sequencing of the sodA (Liu et al., 2001), dnaJ (Morita et al., 2004), gyrB (Kasai et al., 2000), recA (Blackwood et al., 2000), the 32 kDa protein gene (Soini et al., 1996) or a combination thereof (Devulder et al., 2005) have been further used to identify MAC species. These studies, however, did not include all of the four MAC species and used restricted panels of reference strains, thus limiting the significance of the data. The hsp65 gene has been thoroughly investigated (Devallois et al., 1996; Leao et al., 1999; Swanson et al., 1997; Telenti et al., 1993) and sequence analysis of this gene of almost 1600 bp has been used to distinguish between M. avium subsp. avium and M. avium subsp. paratuberculosis strains and disclosed six M. avium subsp. hominissuis sequavars (Turenne et al., 2006). Intergenic spacer (ITS) sequencing (Abed et al., 1995; Barry et al., 1991; Glennon et al., 1994; Roth et al., 1998) further subdivided MAC into 32 sequevars and recently delineated M. chimaera (Tortoli et al., 2004) and M. colombiense (Murcia et al., 2006).

Herein, we studied partial rpoB gene sequencing as a new method for MAC species identification. This single-copy chromosomal gene encoding the bacterial RNA polymerase β-subunit has been previously used for the identification of several bacterial groups, including rapidly growing mycobacteria (RGM) and MAC (Kim et al., 1999). However, only M. avium and M. intracellulare were included in earlier studies, which were based on the analysis of a 342 bp (Kim et al., 2001) or 705 bp (Gingeras et al., 1998) region of rpoB comprising only 8.5–17 % of the entire rpoB gene length and did not ensure that the most variable rpoB region for identification was targeted.

Herein, we gained the opportunity afforded by genome sequence data to further investigate rpoB as a tool for molecular identification of MAC species, i.e. M. avium, M. intracellulare, M. chimaera and M. colombiense.

Mycobacterial strains.
M. avium subsp. avium ATCC25291^T, M. avium subsp. silvaticum ATCC49884^T, M. avium subsp. paratuberculosis ATCC19698^T and M. intracellulare ATCC15985 were purchased from ATCC. M. chimaera DSM44623^T was purchased from DSMZ. M. avium subsp. hominisssuis IWGMT49 (originally isolated from a pig) was provided by Dick van Soolingen (National Institute of Public Health and the Environment, Bilthoven, The Netherlands). M. colombiense CIP 108962^T was purchased from the Institut Pasteur Collection. In this study, 100 MAC clinical isolates collected in our clinical microbiology laboratory in 1997–2007 were also studied (Table 2). Every isolate was inoculated into Middlebrook 7H9 liquid medium and subcultured onto Middlebrook and Cohn 7H10 agar (Becton Dickinson) at 30 °C. Clinical isolates coated on beads were inactivated as previously described (Djelouagji & Drancourt, 2006), and the DNA extracted using a Qiagen kit was used as the template for PCR amplification of the rpoB, 16S rDNA, ITS and hsp65 genes.

Table 2. List of clinical MAC and seven atypical clinical isolates investigated in this study

Construction of a MAC partial rpoB gene database.
Consensus PCR primers were designed after alignment of rpoB gene sequences of M. avium subsp. paratuberculosis K-10 (GenBank accession no. NC-002944), M. avium subsp. hominissuis strain 104 (GenBank accession no. NC-008595) and Mycobacterium tuberculosis strain H37Rv (GenBank accession number L27989). Interspecies and intraspecies rpoB gene sequence variability was analysed using the in-house VARiability Analysis Program (SVARAP) (Colson et al., 2006). The SVARAP analysis allowed us to design consensus PCR primers Myco-F (5' GGCAAGGTCACCCCGAAGGG 3'; base positions 2479–2498 with reference to the M. paratuberculosis K-10 rpoB sequence) and Myco-R (5' AGCGGCTGCTGGGTGATCATC 3'; base positions 3219–3239) in two conserved regions flanking the most variable rpoB region (Adékambi et al., 2003). This primer pair was used for partial amplification of the rpoB gene in the seven MAC reference strains. PCR was carried out in a 2720 Thermal Cycler (Applied Biosystems) in a 50 µl final volume containing 25 µl H₂O, 5 µl 10x buffer (Qiagen), 2.5 µl 25 µM MgCl₂, 5 µl 100 µM dNTP, 5 µM of each primer (Eurogentec), 2.5 U Taq DNA polymerase (Invitrogen) and 10 µl mycobacterial DNA using the following program: 5 min at 95 °C, followed by 35 cycles consisting of 94 °C for 30 s, 64 °C for 30 s and 72 °C for 90 s, and a 10 min elongation step at 72 °C. PCR mix was used as the negative control. Amplicons were purified by adding 50 µl distilled H₂O to a purification plate (Millipore) and agitated for 10 min. Forward and reverse sequencing mixtures contained 3 µl buffer (Big Dye V1, Applied Biosystems), 10 µl distilled H₂O and 1 µl 3.2 pmol µl^–1 primer in a final volume of 16 µl. The sequencing reaction comprised an initial denaturation step of 1 min at 95 °C, followed by 25 cycles of denaturation at 96 °C for 10s, annealing at 50 °C for 5 s and elongation at 60 °C for 3 min. Sequencing products were purified using a Sephadex plate (Amersham Biosciences) that was centrifuged at 720 g for 3 min and deposited on a MicroAmp Optical 96-well reaction plate (Applied Biosystems). Sequencing electrophoresis was performed on a 3100 Genetic Analyzer (Applied Biosystems). Accurate correction of the rpoB nucleotide sequence was assessed by its translation into the amino acid sequence (). The sequences have been deposited in GenBank (Table 3) and the percentage of similarity between nucleotide sequences was determined by BioEdit () (Tables 4 and 5).

Table 3. GenBank accession numbers of original sequences determined in this study

Table 4. Sequence similarity values for rpoB sequences (lower left) and hsp65 sequences (upper right) of MAC reference strains and seven atypical clinical isolates M. a. a, M. avium subsp. avium; M. a. p, M. avium subsp. paratuberculosis; M. a. s, M. avium subsp. silvaticum; M. a. h, M. avium subsp. hominissuis.

Table 5. Sequence similarity values for 16S rDNA sequences (lower left) and 16S–23S rDNA spacer (ITS) sequences (upper right) of MAC reference strains and seven atypical clinical isolates M. a. a, M. avium subsp. avium; M. a. p, M. avium subsp. paratuberculosis; M. a. s, M. avium subsp. silvaticum; M. a. h, M. avium subsp. hominissuis.

rpoB gene sequence-based identification of clinical isolates.
The partial rpoB gene sequence was determined for every one of the 100 clinical isolates studied using the primer pair Myco-F/Myco-R and the PCR/sequencing protocol presented above. rpoB sequence similarity for the clinical isolates was determined in comparison with the reference sequences in the rpoB database by using the CLUSTAL_X program with a weighted residue table in the MEGALIGN package (Windows version 4.10e; DNASTAR). Isolates exhibiting ≥99.3 % sequence similarity with a reference strain were identified at the species level. Isolates exhibiting <99.3 % similarity with a reference strain were not identified at the species level (see below).

Additional tests for molecular identification.
All reference strains, clinical isolates identified as M. intracellulare, M. chimaera and M. avium subsp. paratuberculosis, as well as isolates exhibiting an original, non-identified rpoB gene sequence, were submitted to further 16S rDNA, ITS and hsp65 amplification and sequencing. 16S rDNA amplification and sequencing were done using the primer pair fd1–Rp2 (Woese, 1987). The conditions of amplification were 5 min at 95 °C, followed by 35 cycles of 94 °C for 30 s, 52 °C for 30 s and 72 °C for 1 min. For ITS amplification and sequencing, we used primers Ec16S.1390p and Mb23S.44n (Frothingham & Wilson, 1993) and 38 cycles of an initial 5 min denaturation at 95 °C, followed by 30 s denaturation at 94 °C, 30 s annealing at 62 °C and 90 s extension at 72 °C, with final extension for 7 min. For amplification and sequencing of the 3' end of hsp65, we initially used the primer pair hsp65-574F and hsp65-R (Turenne et al., 2006); since these primers failed to amplify the homologous sequence in M. colombiense, we further designated primers hsp65-F106 (5' AACGTCGTCCTGGAGAAGAA 3') and hsp65-R1558 (5' GCCTTCTCCGGCTTGTC 3'), which covered almost the entire hsp65 gene in this species with hsp65-574F (5' GGTTCGACAAGGGYTACATC 3') as an additional primer for sequencing the 3' region of the gene. The conditions of PCR were 5 min at 95 °C, followed by 35 cycles of 95 °C for 45 s, 60 °C for 45 s and 72 °C for 90 s, with a final extension for 7 min. PCR mixtures (50 µl) contained 5 µl 10x Taq buffer, 200 µM each dNTP, 2.5 µM MgCl₂, 2.5 U Taq DNA polymerase, 10 mM of each primer, 25 µl sterile water and 10 µl mycobacterial DNA. Sequencing reactions were performed as described above.

Phylogenetic analyses.
M. tuberculosis was used as an outgroup in all phylogenetic analyses. Sequences were trimmed to start and finish at the same nucleotide position for all reference strains and clinical isolates. Multisequence alignment was performed with the CLUSTAL_X program, version 1.81, in the PHYLIP software package. Five different phylogenetic trees were obtained from DNA sequences using the neighbour-joining method with the Jukes–Cantor parameter in the MEGA.3 program (Kumar et al., 2004). Trees were based on the 16S rDNA, ITS, hsp65 and rpoB sequences and concatenation of these four sequences. A bootstrap analysis (1000 repeats) was performed to evaluate the topology of each phylogenetic tree; bootstrap values above 75 % were considered significant.

Construction of a partial rpoB gene sequence database of MAC species
The primer pair Myco-F/Myco-R amplified a 711 bp fragment in every MAC reference strain studied, whereas negative controls remained negative. Each amplicon yielded a deduced amino acid sequence that comprised 237 residues. Further analysis revealed the presence of nucleotide substitutions among the species tested with interspecies nucleotide sequence similarity ranging from 94.24 % between M. intracellulare and M. avium subsp. silvaticum to 99.3 % between M. intracellulare and M. chimaera (Table 4). Intraspecies similarity among M. avium subspecies ranged from 99.58 % among M. avium subsp. silvaticum, M. avium subsp. hominissuis and M. avium subsp. paratuberculosis to 99.86 % among M. avium subsp. avium and M. avium subsp. silvaticum (Table 4). We found no deletion or insertion within this rpoB gene region in the MAC reference strains.

rpoB sequence-based identification of clinical isolates
All of the clinical isolates yielded a 711 bp amplicon when amplified using the Myco-F/Myco-R primer pair, whereas negative controls remained negative. Ninety-three clinical isolates (93 %) exhibited a partial rpoB gene sequence with ≥99.3 % similarity to one of the four MAC reference strains, and these isolates were regarded as identified at the species level (Table 4). They comprised 81 isolates exhibiting 100 % partial rpoB sequence similarity to M. avium subsp. hominissuis IWGMT49^T and two isolates with 99.7 % partial rpoB sequence similarity to M. avium subsp. paratuberculosis, eight isolates (8 %) exhibiting 99.2–99.7 % sequence similarity to M. intracellulare, and two blood isolates (2 %) exhibiting 100 % rpoB sequence similarity to M. chimaera. Seven isolates (7 %) exhibiting partial rpoB gene sequence <99.3 % similarity with any one of the reference strains were regarded as non-identified (Table 2). They comprised two sputum isolates (nos 3256799 and 5351974) exhibiting 97.89 % rpoB sequence similarity with M. chimaera, two isolates recovered from sputum (no. 62863) and alveolar washing (no. 5356591) exhibiting 96.7–97.8 % rpoB sequence similarity with M. chimaera, one sputum isolate (no. 68257) exhibiting 97.7 % rpoB gene sequence similarity with M. colombiense, one sputum isolate (no. 4355387) exhibiting 97.1 % rpoB sequence similarity with M. colombiense, and one sputum isolate (no. 27497) exhibiting 96.2 % rpoB sequence similarity with M. colombiense.

Reference molecular test results
The four M. avium subspecies yielded an identical 16S rDNA sequence, as did M. intracellulare and M. chimaera (Table 5). In the reference strains, ITS sequence similarity varied from 99.7 to 100 % among the four M. avium subspecies; this value was 62.5 % between M. avium subsp. avium and M. colombiense, 45.4 % between M. avium subsp. avium and M. chimaera, and 44.2 % between M. avium subsp. avium and M. intracellulare. The primer pair hsp65-574F and hsp65-R (Turenne et al., 2006) amplified hsp65 in M. avium, M. intracellulare and M. chimaera, but not in M. colombiense. Instead, we were able to amplify the 3' extremity of the hsp65 gene in all reference strains, including M. colombiense, using PCR primers hsp65F-106 and hsp65R-1558 and primer hsp65F-574 for sequencing. The analysis of the M. colombiense sequence disclosed a T/C mismatch at position 588 (M. avium subsp. paratuberculosis K-10 numbering). The 3' extremity of hsp65 had sequence similarity values ranging from 95.3 to 100 % among M. avium subspecies, 94.8 % between M. avium and M. intracellulare, 97.6 % between M. intracellulare and M. colombiense, and 99.8 % between M. intracellulare and M. chimaera. The same three genes, 16S rDNA, ITS and hsp65, were further amplified and sequenced in clinical isolates, including the seven non-identified isolates. Accordingly, PCR targeting the hsp65 gene using primer set hsp65-574F/hsp65-R (Turenne et al., 2006) failed to amplify this gene in reference strain M. colombiense CIP 108962^T and three out of seven non-identified isolates, numbers 68257, 5356591 and 62863, whereas primers hsp65-F106/hsp65-R1558 successfully amplified almost the entire 1462 bp fragment of the hsp65 gene for further sequencing. Isolates 3256799 and 5351974 yielded identical sequence similarity values of 99.6 % for 16S rDNA, 90.1 % for ITS and 98 % for hsp65 with M. chimaera (Tables 4 and 5); isolates 62863 and 5356591 yielded identical similarity values of 99.8 % for 16S rDNA, 86.3 % for ITS and 97.5 % for hsp65 with M. chimaera; isolate number 68257 yielded 100 % for 16S rDNA sequence similarity, 51.8 % for ITS sequence similarity and 99.6 % for hsp65 sequence similarity with M. colombiense; isolate number 4355387 yielded 99.9 % for 16S rDNA sequence similarity, 51.8 % for ITS sequence similarity and 97.8 % for hsp65 sequence similarity with M. colombiense; and isolate number 27497 yielded 100 % for 16S rDNA sequence similarity, 51.3 % for ITS sequence similarity and 95 % for hsp65 sequence similarity with M. colombiense.

Phylogenetic analyses
We constructed phylogenetic trees (Figs 1 and 2) based on the 16S rDNA, ITS, partial hsp65 and rpoB sequences determined for the seven reference strains of the four MAC species, and for the seven original clinical isolates that were not assigned to a known species on the basis of rpoB sequencing. Trees derived from separated sequences and the tree derived from concatenated genes gave identical results (Fig. 2). In all trees, the four subspecies of M. avium grouped together with bootstrap values of 94–100 %, while the other reference strains were clearly separated, with the exception of the 16S rDNA sequence-based tree, which did not separate M. intracellulare and M. colombiense because of the 100 % sequence similarity (Fig. 1a). The rpoB sequence-based tree was most similar to that of the concatenated tree, as it indicated that isolates 68257 and 4355387 were most closely related to M. colombiense with a 100 % bootstrap value, and that isolates 5351974 and 3256799 on the one hand and 62863 and 5356591 on the other hand formed two clusters separated from known MAC species with 100 % bootstrap values in the concatenated tree. The phylogenetic position of isolate number 27497 remained uncertain because this isolate was more closely related to M. avium in the hsp65 sequence-derived tree and the concatenated tree with a 99 % bootstrap value, and more closely related to M. colombiense in the other trees with a 93 % bootstrap value in the 16S rDNA sequence-based tree, an 89 % bootstrap value in the ITS sequence-based tree and a 99 % bootstrap value in the rpoB sequence-based tree.

(65K): Fig. 1. Phylogenetic tree traced using the neighbour-joining method based on the 1442 bp 16S rDNA (a), the 438 bp ITS (b; indicating the sequavar for unidentified isolates, with MAC-X designating a previously unknown sequavar), the partial 995 bp hsp65 gene (c) and the partial 711 bp rpoB gene (d) sequences of 100 MAC clinical isolates and seven reference strains. The sequences were aligned using the neighbour-joining method and Kimura's two-parameter distance correction model. Bootstrap values ≥75 % are indicated by the value at each node (as a percentage). M. tuberculosis H37Rv was used as the outgroup. Bars: (a, b) 0.2 % difference in nucleotide sequence; (c, d), 0.1 %. T, type sequence.

(16K): Fig. 2. Phylogenetic tree based on concatenation of the 16S rDNA, ITS, partial rpoB and partial hsp65 sequences of 100 MAC clinical isolates and seven reference strains. The sequences were aligned by using the neighbour-joining method and Kimura's two-parameter distance correction model. Bootstrap values ≥75 % are indicated by the value at each node (as a percentage). M. tuberculosis H37Rv was used as the outgroup. Bar, 0.5 % difference in nucleotide sequence.

In this study, we found a unique partial rpoB sequence in each member of seven strains representative of the four MAC species and M. avium subspecies. Sequences determined herein were interpreted as authentic, since negative controls remained negative in all PCR experiments and the amplified fragments were of the predicted 711 bp size for the reference strains and clinical isolates. Moreover, we found original nucleotide sequences that were translated in silico into a peptide without any internal stop codon, further suggesting that these sequences were authentic and did not merely result from sequencing errors. All original nucleotide rpoB sequences have been deposited into GenBank (Table 3). In this study, partial rpoB sequences have been determined using the primer pair Myco-F/Myco-R, previously used for the identification of RGM (Adékambi et al., 2003). The present data therefore indicated that the usefulness of the Myco-F/Myco-R primer pair could be extended to the identification of slowly growing mycobacteria. This point may warrant further exploration in additional slowly growing mycobacteria species in order to confirm that the Myco-F/Myco-R primer pair is a broad-spectrum primer pair for partial rpoB gene amplification and sequencing in Mycobacterium species.

Reference strains of the four MAC species studied exhibited a 4.5–5.7 % partial rpoB sequence divergence with the notable exception of M. intracellulare and M. chimaera, which shared a high, 99.3 % partial rpoB sequence similarity. The latter value was of the same order of magnitude as that observed among the four M. avium subspecies (≤0.4 % sequence divergence). The close molecular proximity of M. intracellulare and M. chimaera disclosed herein by the rpoB sequence analysis was in agreement with the analysis of 16S rDNA, ITS and hsp65 gene sequences in this study and the description of M. chimaera (Tortoli et al., 2004). 16S rDNA has been regarded as the gold standard for molecular identification in Mycobacterium (Tortoli et al., 2003), but the present data indicated that this may not be the case for M. intracellulare and M. chimaera, in agreement with the description of the latter species, which found only one nucleotide mismatch between M. chimaera and M. intracellulare sequavar i (Tortoli et al., 2004). Another 16S rDNA sequence-based phylogenetic tree incorporating the four MAC species also showed the almost identical 16S rDNA sequences in M. intracellulare and M. chimaera (Murcia et al., 2006). This situation reminded us of what we previously observed when using the same rpoB region for the identification of RGM (Adékambi et al., 2003). For RGM, a 2 % cut-off value was suitable for all of the tested RGM species, except for distinguishing Mycobacterium fortuitum and Mycobacterium houstonense, which exhibited 98.2 % sequence homology. Based on results obtained with reference strains of the currently acknowledged four MAC species, a 0.7 % rpoB sequence divergence cut-off was used for the identification of MAC clinical isolates.

Using this cut-off value, the vast majority of the clinical isolates (93 %) were identified at the species level. All rpoB sequence-based identifications have been further confirmed by polyphasic molecular identification including 16S rDNA, ITS and hsp65 sequencing. Whereas the prevalence of M. intracellulare was 8 %, the vast majority of these isolates were M. avium subsp. hominissuis, a subspecies recently delineated among M. avium (Mijs et al., 2002) and regarded as the ancestor group for the highly adapted M. avium subsp. avium/silvaticum and M. avium subsp. paratuberculosis organisms (Turenne et al., 2007). Two isolates identified as M. chimaera represent the first isolates of this novel species to be reported in addition to the initial M. chimaera report (Tortoli et al., 2004). Interestingly, these two isolates have been recovered from the blood of two different patients, whereas the initial description of this novel species mentioned almost exclusively respiratory tract isolates from Italian patients in five different hospitals. We did not have additional clinical data to confirm the clinical significance; however, these data suggest the possibility of disseminated infection due to M. chimaera. We did not identify M. colombiense, a newly described MAC species isolated from the blood and sputum specimens of HIV-infected patients in Colombia (Murcia et al., 2006) and in one case of lymphadenopathy in a 3-year-old Spanish girl (Esparcia et al., 2008).

Seven clinical isolates were not identified at the species level based on partial rpoB sequencing, and the uniqueness of these isolates was further confirmed by sequence analysis of the 16S rDNA, ITS and 3'-end hsp65 genes. A new primer pair was therefore reported in this study for the successful partial amplification of the hsp65 gene in all MAC species. This polyphasic molecular approach suggested that these seven isolates could be representative of four new MAC species. Although three of the seven isolates exhibited 16S rDNA sequences indistinguishable from that of M. intracellulare and M. chimaera, they exhibited a unique genetic (ITS and partial hsp65 and rpoB genes) pattern. While five of the seven isolates exhibited previously described ITS sequences (Fig. 1b), two of the seven isolates featured novel ITS sequences. Moreover, phylogenetic analyses supported the uniqueness of these isolates with significant >75 % bootstrap values. These data suggest that the seven isolates may be representative of novel MAC species.

In conclusion, partial rpoB gene sequence analysis was effective in the identification of MAC isolates at the species level. The primer pair Myco-F/Myco-R allowed such identification as previously reported for RGM (Adékambi et al., 2003), suggesting that this primer pair could be used for broad-spectrum Mycobacterium species identification. Moreover, partial rpoB sequencing identified a few unique isolates that warrant further phenotypic and molecular characterization to confirm that they represent novel MAC species.

Edited by: M. Daffé

Footnotes

†Present address: Mycobacteriology Laboratory Branch, Division of Tuberculosis Elimination, Centers for Disease Control and Prevention, Atlanta, GA, USA.

The GenBank/EMBL/DDBJ accession numbers for the Mycobacterium avium complex sequences determined in this study are listed in Table 3.