HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Primer details Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via Google Scholar Google Scholar Articles by Adékambi, T. Articles by Drancourt, M. Search for Related Content PubMed PubMed Citation Articles by Adékambi, T. Articles by Drancourt, M. Agricola Articles by Adékambi, T. Articles by Drancourt, M.

Dissection of phylogenetic relationships among 19 rapidly growing Mycobacterium species by 16S rRNA, hsp65, sodA, recA and rpoB gene sequencing

Unité des Rickettsies, CNRS UMR 6020 IFR 48, Faculté de Médecine, Université de la Méditerranée, 27, Boulevard Jean Moulin, 13385 Marseille Cedex 05, France

Correspondence
Michel Drancourt
Michel.Drancourt{at}medecine.univ-mrs.fr

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION REFERENCES

 
The current classification of non-pigmented and late-pigmenting rapidly growing mycobacteria (RGM) capable of producing disease in humans and animals consists primarily of three groups, the Mycobacterium fortuitum group, the Mycobacterium chelonae–abscessus group and the Mycobacterium smegmatis group. Since 1995, eight emerging species have been tentatively assigned to these groups on the basis of their phenotypic characters and 16S rRNA gene sequence, resulting in confusing taxonomy. In order to assess further taxonomic relationships among RGM, complete sequences of the 16S rRNA gene (1483–1489 bp), rpoB (3486–3495 bp) and recA (1041–1056 bp) and partial sequences of hsp65 (420 bp) and sodA (441 bp) were determined in 19 species of RGM. Phylogenetic trees based upon each gene sequence, those based on the combined dataset of the five gene sequences and one based on the combined dataset of the rpoB and recA gene sequences were then compared using the neighbour-joining, maximum-parsimony and maximum-likelihood methods after using the incongruence length difference test. Combined datasets of the five gene sequences comprising nearly 7000 bp and of the rpoB+recA gene sequences comprising nearly 4600 bp distinguished six phylogenetic groups, the M. chelonae–abscessus group, the Mycobacterium mucogenicum group, the M. fortuitum group, the Mycobacterium mageritense group, the Mycobacterium wolinskyi group and the M. smegmatis group, respectively comprising four, three, eight, one, one and two species. The two protein-encoding genes rpoB and recA improved meaningfully the bootstrap values at the nodes of the different groups. The species M. mucogenicum, M. mageritense and M. wolinskyi formed new groups separated from the M. chelonae–abscessus, M. fortuitum and M. smegmatis groups, respectively. The M. mucogenicum group was well delineated, in contrast to the M. mageritense and M. wolinskyi groups. For phylogenetic organizations derived from the hsp65 and sodA gene sequences, the bootstrap values at the nodes of a few clusters were <70 %. In contrast, phylogenetic organizations obtained from the 16S rRNA, rpoB and recA genes were globally similar to that inferred from combined datasets, indicating that the rpoB and recA genes appeared to be useful tools in addition to the 16S rRNA gene for the investigation of evolutionary relationships among RGM species. Moreover, rpoB gene sequence analysis yielded bootstrap values higher than those observed with recA and 16S rRNA genes. Also, molecular signatures in the rpoB and 16S rRNA genes of the M. mucogenicum group showed that it was a sister group of the M. chelonae–abscessus group. In this group, M. mucogenicum ATCC 49650^T was clearly distinguished from M. mucogenicum ATCC 49649 with regard to analysis of the five gene sequences. This was in agreement with phenotypic and biochemical characteristics and suggested that these strains are representatives of two closely related, albeit distinct species.

The GenBank/EMBL/DDBJ accession numbers of the sequences reported in this study are given in Figs 1–5.

Details of the primers used for sequencing are available as supplementary material in IJSEM Online.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION REFERENCES

 
Mycobacteria are aerobic, non-mobile, mycolic acid-containing bacteria that are characteristically acid-fast and contain DNA with a G+C content of 62–72 mol%. They all lie within the genus Mycobacterium, the only genus in the family Mycobacteriaceae. They are classified into slowly growing and rapidly growing, and pigmented and non-pigmented species. The rapidly growing mycobacteria (RGM), which require less than 7 days to produce easily seen colonies on solid media, comprise 56 environmental species (Pfyffer et al., 2003). Fifteen species commonly encountered in human and animals belong primarily to the Mycobacterium chelonae–abscessus, Mycobacterium fortuitum and Mycobacterium smegmatis groups (Brown-Elliott & Wallace, 2002; Lopez-Marin et al., 1993). Indeed, M. fortuitum, M. abscessus, M. chelonae, M. smegmatis, Mycobacterium immunogenum, Mycobacterium peregrinum, Mycobacterium houstonense (formerly M. fortuitum third biovariant sorbitol-positive), Mycobacterium neworleansense (formerly M. fortuitum third biovariant sorbitol-negative), Mycobacterium mageritense, Mycobacterium septicum, Mycobacterium mucogenicum (formerly the M. chelonae-like organism), Mycobacterium wolinskyi and Mycobacterium goodii were isolated from patients (Brown-Elliott & Wallace, 2002) whereas Mycobacterium senegalense, Mycobacterium farcinogenes and Mycobacterium porcinum are pathogens of veterinary importance (Chamoiseau, 1979; Lopez-Marin et al., 1993). The latter species has been recently shown to be indistinguishable from M. fortuitum third biovariant sorbitol-negative human isolates on the basis of PCR-restriction fragment length polymorphism analysis (PRA), partial 16S rRNA gene sequence and biochemical profile (Schinsky et al., 2004).

As for other bacterial genera, 16S rRNA gene-based phylogeny has had a major influence on our perception of taxonomic relationships among the pathogenic RGM (Tortoli, 2003) in addition to updates brought by HPLC of mycolic acid esters, fluorescence-HPLC and PRA of a 439-bp fragment of the 65-kDa heat-shock protein-encoding gene (hsp65) (Brown-Elliott & Wallace, 2002).

Based on nucleotide differences in the 16S rRNA gene sequence, RGM were classified into three groups: the M. fortuitum group included M. fortuitum, M. peregrinum, M. houstonense, M. neworleansense, M. septicum, M. mageritense, M. mucogenicum and M. senegalense; the M. chelonae–abscessus group comprised M. abscessus, M. chelonae and M. immunogenum; and the M. smegmatis group comprised M. smegmatis, M. wolinskyi and M. goodii (Brown-Elliott & Wallace, 2002). However, there has been controversy regarding the taxonomic position of M. mucogenicum, since its biochemical profile and antibiotic-susceptibility pattern were more closely related to those of members of the M. chelonae–abscessus group (Wallace et al., 1993). A 16S rRNA gene-based phylogenetic tree revealed that M. mageritense was more closely related to species of the M. smegmatis group (Brown et al., 1999) than to the M. fortuitum group, despite its antibiotic susceptibility and biochemical patterns (Brown-Elliott & Wallace, 2002; Wallace et al., 2002). Furthermore, little is known about the phylogenetic relationships of the three RGM of veterinary importance (Hamid et al., 2002; Tortoli, 2003). Indeed, variations in the datasets, in terms of the strains and the number of species used and analytical methods, have made it difficult to compare and evaluate the proposed phylogenetic relationships.

Taxonomic and phylogenetic studies of mycobacterial species were for many years based on 16S rRNA gene analysis (Tortoli, 2003). The suggestion that bacterial strains belong to the same species if they have fewer than 5 to 15 base differences in their 16S rRNA gene sequences, proposed for other micro-organisms (Fox et al., 1992), is not applicable to the RGM, whose members are more closely related to each other. Since it is doubtful whether phylogenetic relationships should be based solely on the 16S rRNA in cases where sequence identities are 99 % (Drancourt et al., 2000), it was desirable to analyse genes other than the 16S rRNA gene in order to assess phylogenetic relationships further among the pathogenic RGM and to help to clarify their controversial taxonomy. In RGM, hsp65 gene analysis was previously limited to PRA of a 439-bp fragment (Brunello et al., 2001; Steingrube et al., 1995). The partial sequence of the sodA gene encoding the superoxide dismutase was used to define the taxonomic status of M. mageritense within nine RGM under study (Domenech et al., 1997). Recently, PRA of the 16S–23S internal transcribed spacer (ITS) was shown to be a suitable tool for identifying most mycobacterial species and to segregate them into five clusters (Roth et al., 2000).

In this study, we examined the sequences of four genes in addition to the 16S rRNA gene in 19 pathogenic RGM species and we utilized a combined approach to summarize common support and to reconcile discrepancies among datasets: sodA, hsp65, the recA gene, encoding part of the DNA recombination and repair system, and the rpoB gene, encoding the subunit of the RNA polymerase. These genes were previously proposed as suitable phylogenetic markers for the classification of mycobacteria (Adékambi et al., 2003; Blackwood et al., 2000; Domenech et al., 1997; Kim et al., 1999; Rinquet et al., 1999).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION REFERENCES

 
Bacterial strains and molecular methods.
The type strains used in this study are listed in Figs 1–7. They were preserved at –20 °C in skimmed milk until used. Each strain was inoculated into Middlebrook 7H9 liquid medium and subcultured onto Middlebrook and Cohn 7H10 agar (Becton Dickinson) at 30 °C. Purity was confirmed by naked-eye examination of colonies and microscopic examination after Ziehl–Nielsen staining. Colonies were scraped and genomic DNA was extracted using the FAST DNA kit according to the instructions of the manufacturer (Q.biogene). PCRs were carried out in a Biometra thermocycler (BIOLABO). PCR mixtures (50 µl) contained 5 µl 10x Taq buffer, 200 µM of each dNTP, 2·5 mM MgCl₂, 1 U Taq DNA polymerase (Invitrogen), 10 mmol of each appropriate pair of primers (Eurogentec), 33 µl sterile water and 2 µl of the purified DNA. PCR conditions are described below. Purified amplicons (QIAquick PCR purification kit; Qiagen) were sequenced using the ABI Prism d-Rhodamine dye terminator cycle sequencing ready reaction kit according to the manufacturer's instructions (Perkin Elmer Applied Biosystems) and the following program: 30 cycles of denaturation at 94 °C for 10 s, primer annealing at 50 °C for 15 s and extension at 60 °C for 1 min. Products of sequencing reactions were recorded with an ABI Prism 3100 DNA sequencer following the standard protocol of the supplier (Perkin Elmer Applied Biosystems).

View larger version (31K): [in this window] [in a new window]   Fig. 1. Phylogenetic tree of 16S rRNA gene sequences of 19 RGM using the neighbour-joining method with K2P distance correction. The support of each branch, as determined from 1000 bootstrap samples, is indicated by percentages at each node. M. tuberculosis and M. leprae were used as the outgroups. Bar, 0·05 % nucleotide sequence difference.

View larger version (29K): [in this window] [in a new window]   Fig. 2. Phylogenetic tree of hsp65 gene sequences of 19 RGM using the neighbour-joining method with K2P distance correction model. See Fig. 1 for further details. Bar, 2 % nucleotide sequence difference.

View larger version (32K): [in this window] [in a new window]   Fig. 3. Phylogenetic tree of sodA gene sequences of 19 RGM using the neighbour-joining method with K2P distance correction model. See Fig. 1 for further details. Bar, 2 % nucleotide sequence difference.

View larger version (30K): [in this window] [in a new window]   Fig. 4. Phylogenetic tree of recA gene sequences of 19 RGM using the neighbour-joining method with K2P distance correction model. See Fig. 1 for further details. Bar, 5 % nucleotide sequence difference.

View larger version (30K): [in this window] [in a new window]   Fig. 5. Phylogenetic tree of rpoB gene sequences of 19 RGM using the neighbour-joining method with K2P distance correction model. See Fig. 1 for further details. Bar, 2 % nucleotide sequence difference.

View larger version (25K): [in this window] [in a new window]   Fig. 6. Phylogenetic tree of the combined rpoB+recA+sodA+hsp65+16S rRNA gene sequences of 19 RGM using the neighbour-joining method with K2P distance correction model. See Fig. 1 for further details. Bar, 2 % nucleotide sequence difference.

View larger version (25K): [in this window] [in a new window]   Fig. 7. Phylogenetic tree of the combined rpoB+recA gene sequences of 19 RGM using the neighbour-joining method and K2P distance correction model. See Fig. 1 for further details. Bar, 2 % nucleotide sequence difference.

hsp65.
The hsp65 gene sequence was available in GenBank for 15/19 RGM under study at the initiation of this work. All the type strains were reanalysed by partial 477-bp hsp65 gene sequencing using primers –21M13F and M13R described by Rinquet et al. (1999) (Supplementary Table A). Additional primers Tb11 and Tb12 described by Telenti et al. (1993) (Supplementary Table A) were used for sequencing reactions. Conditions for hsp65 gene amplification were 2 min at 95 °C followed by 35 cycles of 94 °C for 30 s, 60 °C for 30 s and 72 °C for 2 min, with a final extension step at 72 °C for 5 min.

sodA.
Nine sodA sequences were available in GenBank at the initiation of this work. They were partial 388-bp sequences including identical sequence for certain type strains (Domenech et al., 1997). The only complete sequence available was that of M. fortuitum. The type strains were reanalysed by partial 541-bp sodA gene sequencing using primer pair SodlgF–SodlgR (Supplementary Table A) designed after alignment of the complete sodA sequences of M. fortuitum, Mycobacterium tuberculosis and Mycobacterium leprae (GenBank/Entrez accession nos X70914, NC_000962 and AL450380). Additional primers SodF–SodR described by Domenech et al. (1997) (Supplementary Table A) were used as internal sequencing primers. Conditions for sodA gene amplification were 2 min at 95 °C followed by 35 cycles of 94 °C for 30 s, 60 °C for 30 s and 72 °C for 2 min, with a final extension step at 72 °C for 5 min.

recA.
Partial recA sequences were available in GenBank for 8/19 RGM under study at the initiation of this work, as well as the complete recA gene sequence of M. smegmatis. The type strains were characterized by complete recA gene sequencing using the primers described by Blackwood et al. (2000) and additional primers recF2b, rec755R, rec3288F, rec3335R and rec3575R designed after alignment of the complete recA sequences of M. smegmatis, M. tuberculosis and M. leprae (GenBank/Entrez accession nos X99208, NC_000962 and AL450380). Primer pairs recF2–recR1 (or recF2b–recR1), recG1–recR2 and rec3288F–rec3575R (Supplementary Table A) were used to amplify three overlapping fragments of the complete recA gene. Conditions for these PCRs were 2 min at 95 °C followed by 35 cycles of 94 °C for 30 s, 54 °C for 30 s and 72 °C for 2 min, with a final extension step at 72 °C for 5 min. Products were gel-purified and extracted with the QIAquick gel extraction kit (Qiagen) when necessary.

rpoB.
The type strains were identified by complete rpoB gene sequence analysis into seven overlapping fragments using primer pairs Smeg7F–Smeg601R, Smeg529F–Smeg1485R, MF–Smeg2333R, Fort623F–Smeg2649R, Smeg2426F–Smeg3288R and Smeg2835F–Smeg3668R and Smeg2885F–Fort4260R as previously described (Adékambi et al., 2003; Kim et al., 1999). Additional primers allowed completion of rpoB gene sequencing (Supplementary Table B). Conditions for these PCRs were 2 min at 95 °C followed by 35 cycles of 94 °C for 30 s, 64 °C for 30 s and 72 °C for 2 min, with a final extension step at 72 °C for 5 min.

Sequence analyses and phylogenetic comparisons.
Percentages of similarity between sequences were determined using the CLUSTAL program with weighted residue weight table in the MegAlign package (Windows version 4.10e; DNASTAR). For phylogenetic analyses, sequences were trimmed in order to start and finish at the same nucleotide position for all the strains under study. Multisequence alignment was performed by using the CLUSTAL X program, version 1.81, in the PHYLIP software package (Thompson et al., 1997). Phylogenetic trees were obtained from DNA sequences by using the neighbour-joining method with Kimura's two-parameter (K2P) distance correction model with 1000 bootstrap replications (in MEGA version 2.1; Kumar et al., 2001), maximum-parsimony method with min-mini heuristic search option (MEGA version 2.1) and maximum-likelihood method (DNAml software in PHYLIP). The incongruence length difference (ILD) test or partition homogeneity test (1000 randomizations of datasets) was performed using PAUP4.0b program (Swofford, 1998). A bootstrap analysis (1000 repeats) using M. tuberculosis and M. leprae as the outgroups was performed to evaluate the robustness of the phylogenetic trees and bootstrap values above 90 % were considered as significant. These trees were rooted by using M. tuberculosis and M. leprae, which are the most closely related species to RGM for which the five genes under study were available in GenBank/Entrez (accession nos NC_000962 and AL450380).

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION REFERENCES

 
In our approach, we sequenced a battery of five genes previously proposed as suitable tools for studying RGM phylogeny. The 16S rRNA (1483–1489 bp), hsp65 (420 bp), sodA (441 bp), recA (1041–1056 bp) and rpoB (3486–3495 bp) genes were sequenced in 19 species and showed interspecies sequence similarity of 95·5–100 %, 87·8–100 %, 86·7–100 %, 84·4–100 % and 87·7–99·5 %, respectively. We first analysed the five genes separately and then used different combined databases to create a single alignment dataset. Identical phylogenetic trees were inferred from multiple sequence alignments of the nucleotide sequences using the neighbour-joining, maximum-parsimony and maximum-likelihood methods. The three methods gave almost similar bootstrap values. The bootstrap values given below were derived from the neighbour-joining method only using the K2P distance correction model.

Separate analyses
16S rRNA gene phylogenetic trees (Fig. 1).
The phylogenetic trees were based on 1490 unambiguously aligned positions, 71 of which were informative under the parsimony criterion. Five separate clusters were characterized. M. chelonae CIP 194535^T, M. chelonae ATCC 19237, M. abscessus and M. immunogenum formed a monophyletic cluster I with a 100 % bootstrap value. The three strains of M. mucogenicum formed a second monophyletic cluster II with a 96 % bootstrap value. M. peregrinum, M. septicum, M. neworleansense, M. porcinum, M. farcinogenes, M. senegalense, M. houstonense and M. fortuitum formed cluster III, with a 72 % bootstrap value. M. mageritense and M. wolinskyi formed cluster IV, albeit with a low bootstrap value (65 %). M. smegmatis and M. goodii formed a monophyletic cluster V with a 100 % bootstrap value, a sister group of cluster IV. Sister clusters I, II and III were supported by non-significant bootstrap values of 46–60 %. Two subclusters, IIIa (M. peregrinum, M. septicum, M. neworleansense and M. porcinum) and IIIb (M. farcinogenes, M. senegalense, M. houstonense and M. fortuitum), were supported by bootstrap values of 89 and 62 %, respectively. The bootstrap values at the nodes of a few clusters obtained were too low to induce much confidence. The 16S rRNA gene sequence did not discriminate M. houstonense, M. senegalense and M. farcinogenes or M. mucogenicum ATCC 49650^T from M. mucogenicum ATCC 49651. These data indicate that the 16S rRNA gene sequence alone did not resolve the phylogenetic relationships between all currently recognized RGM species as indicated by the low bootstrap values obtained at the nodes of different sister groups and subgroups (72 %). The percentage similarity of the 16S rRNA gene sequence in 19 RGM examined was 95·5–100 %, explaining the poor phylogenetic information obtained from this sequence.

hsp65 phylogenetic trees (Fig. 2).
The phylogenetic trees were based on 420 unambiguously aligned positions (omitting the primer sequences used for the amplification), 59 of which were informative under the parsimony criterion. The deduced amino acid sequences comprised 140 residues [G₅₂ to M₁₉₁ (M. tuberculosis numbering)]. M. chelonae CIP 194535^T, M. chelonae ATCC 19237, M. abscessus and M. immunogenum grouped together with a non-significant bootstrap value of 51 %. The three M. mucogenicum strains formed a monophyletic cluster II with a 99 % bootstrap value. Subclusters IIIa (M. peregrinum, M. septicum, M. porcinum and M. neworleansense) and IIIb (M. fortuitum, M. houstonense, M. senegalense and M. farcinogenes) were supported by bootstrap values of 61 and 84 %, respectively. Subcluster IIIb was found to be a sister group of cluster IV (M. mageritense and M. wolinskyi), which was found to be a poorly monophyletic cluster (55 %). M. smegmatis and M. goodii formed a monophyletic cluster V with a bootstrap value of 99 %. However, the hsp65 gene sequence did not discriminate either M. senegalense from M. farcinogenes or M. mucogenicum ATCC 49650^T from M. mucogenicum ATCC 49651.

sodA phylogenetic trees (Fig. 3).
The phylogenetic trees were based on 441 unambiguously aligned positions, 86 of which were informative under the parsimony criterion. The deduced amino acid sequences comprised 147 residues [Q₂₃ to Q₁₆₉ (M. tuberculosis numbering)]. M. chelonae CIP 194535^T, M. chelonae ATCC 19237, M. abscessus and M. immunogenum formed a monophyletic cluster I with a 100 % bootstrap value. The three M. mucogenicum strains formed a monophyletic cluster II (100 %). M. peregrinum, M. septicum, M. neworleansense, M. porcinum and M. fortuitum grouped into subcluster IIIa (42 %), whereas M. houstonense, M. senegalense and M. farcinogenes formed a monophyletic subcluster IIIb with a bootstrap value of 100 %. M. mageritense appeared to be the only species of cluster IV, forming a sister group of subcluster IIIb with a 64 % bootstrap value at the node. M. smegmatis and M. goodii formed a monophyletic cluster V with a 99 % bootstrap value. M. wolinskyi seemed to be the only species of cluster VI, forming a sister group of cluster V with a 92 % bootstrap value at the node. The sodA gene sequence did not discriminate M. senegalense from M. farcinogenes.

recA phylogenetic trees (Fig. 4).
The phylogenetic trees were based on 1056 unambiguously aligned positions, including 236 informative positions under the parsimony criterion. M. chelonae CIP 194535^T, M. chelonae ATCC 19237, M. abscessus and M. immunogenum formed a monophyletic cluster I with a 100 % bootstrap value. The three M. mucogenicum strains formed a monophyletic cluster II with a 100 % bootstrap value. M. peregrinum, M. septicum, M. neworleansense, M. porcinum and M. fortuitum formed subcluster IIIa (88 %) and M. houstonense, M. senegalense and M. farcinogenes formed a monophyletic subcluster IIIb (100 %). M. smegmatis and M. goodii grouped into cluster V with a 100 % bootstrap value. M. mageritense and M. wolinskyi were the only species of clusters IV and VI, respectively; they grouped with 30 and 73 % bootstrap values with the respective sister group at the node. The recA gene sequence did not discriminate M. senegalense from M. farcinogenes.

rpoB phylogenetic trees (Fig. 5).
The phylogenetic trees were based on 3533 bp unambiguously aligned positions, comprising 650 informative positions under the parsimony criterion. M. chelonae CIP 194535^T, M. chelonae ATCC 19237, M. abscessus and M. immunogenum formed a monophyletic cluster I with a 100 % bootstrap value. The three M. mucogenicum strains formed a monophyletic cluster II with a 100 % bootstrap value. Subclusters IIIa (M. peregrinum, M. septicum and M. neworleansense) and IIIb (M. porcinum, M. fortuitum, M. houstonense, M. senegalense and M. farcinogenes) were supported by bootstrap values of 97 and 72 %, respectively, and formed cluster III, supported by a 100 % bootstrap value. M. wolinskyi was the only representative of cluster IV, with a 96 % bootstrap value at the node with cluster III. M. smegmatis and M. goodii formed a monophyletic cluster VI with a 100 % bootstrap value. M. mageritense was the only species in cluster V, with a 90 % bootstrap value at the node with cluster VI.

Combination in a simultaneous analysis (Fig. 6)
The combined 16S rRNA, hsp65, sodA, recA and rpoB phylogenetic trees were compared with individual phylogenies. M. chelonae CIP 194535^T, M. chelonae ATCC 19237, M. abscessus and M. immunogenum formed a monophyletic cluster I with a 100 % bootstrap value. The three M. mucogenicum strains formed a monophyletic cluster II with a 100 % bootstrap value. Subclusters IIIa (M. peregrinum, M. septicum, M. neworleansense and M. porcinum) and IIIb (M. fortuitum, M. houstonense, M. senegalense and M. farcinogenes) were supported by bootstrap values of 95 and 100 %, respectively. M. wolinskyi and M. mageritense formed two independent branches. However, M. wolinskyi (cluster IV) and M. mageritense (cluster V) and their respective sister groups were supported by low bootstrap values of 44 and 35 %, respectively. M. smegmatis and M. goodii formed a monophyletic cluster VI with a 100 % bootstrap value.

Conditional combination approach (Fig. 7)
In this approach, the ILD test was used as a preliminary step before choosing either to combine congruent data to increase the accuracy of the phylogenetic reconstruction or to analyse the data separately to discover the reasons for incongruence. Despite the incongruence observed within the five genes, rpoB, recA and 16S RNA gene sequencing trees had globally the same topology. The ILD test was applied to check whether trees for the different genes were sufficiently similar in rates of divergence and branching order for it to be legitimate to combine the data (Cunningham, 1997). The test calculates how much longer the combined parsimony tree is than the sum of the separate trees and compares this difference with the expected distribution estimated by allocating the data randomly into partitions of the same size as the gene. A three-partition test was conducted on the rpoB, recA and 16S rRNA gene sequences. Invariant characters were removed for all analyses (Gaunt et al., 2001). The ILD was supported for rpoB+recA (P=0·73) but rejected for rpoB+recA+16S rRNA (P<0·001). These results allowed us to combine the rpoB and recA gene sequences, but suggested that the 16S rRNA gene should be treated separately, probably because of the small number of informative sites (Darlu & Lecointre, 2002). A neighbour-joining phylogenetic tree derived from the combined rpoB+recA gene sequences (Fig. 7) was compared with the corresponding five gene sequences and 16S rRNA gene sequence. The two protein-encoding genes rpoB+recA led to the same topology as observed with the simultaneous analyses but they improved the bootstrap values meaningfully. The bootstrap values at the nodes of M. wolinskyi (cluster IV) and M. mageritense (cluster V) with their respective sister groups increased from 44 to 81 % and from 35 to 53 %, respectively. The bootstrap values at the nodes of the M. chelonae–abscessus and M. mucogenicum groups increased from 88 to 99 % (Fig. 7). Furthermore, the phylogenetic tree derived from the combined rpoB+recA gene sequences supported the 16S rRNA gene classification of the 19 RGM (Figs 1 and 7). Thus, our rpoB+recA data reinforce the view that the 16S rRNA gene-based phylogeny holds for the majority of the genome.

Evidence from insertions and deletions (molecular signatures)
Several molecular signatures due to insertions and deletions in the sequence alignment under study provided additional phylogenetic evidence. Relative to all the other rpoB gene sequences (including outgroups), the M. chelonae–abscessus and M. mucogenicum groups have 3- and 9-bp deletions at positions 37–39 and 2830–2838, respectively, and a 6-bp insertion at positions 967–972. Moreover, the M. mucogenicum group has an additional 3-bp deletion at positions 34–36 (M. fortuitum numbering). These deletions and insertions suggested that these two groups were sister groups. M. smegmatis and M. goodii have a 3-bp deletion at positions 28–30. Within the M. fortuitum group, subcluster IIIa with the exception of M. neworleansense exhibited a 3-bp deletion at positions 22–24, whereas M. porcinum, a member of subcluster IIIb, had the same deletion.

There were also length variants in the 16S rRNA gene sequence resulting from regions too variable to ascertain the exact alignment. In M. smegmatis and M. goodii, there were 2-bp insertions at positions 82–83 and 185–186, a 1-bp insertion at position 202 and a 1-bp deletion at position 178 (Escherichia coli numbering). A 1-bp deletion was found at positions 818 and 1267 in M. mucogenicum strains and the species of cluster I (M. chelonae, M. abscessus, M. immunogenum). Similarly, M. mageritense and M. wolinskyi shared two 1-bp insertions at positions 1433 and 1440 also noted in the M. goodii 16S rRNA gene sequence.

Several signatures due to small insertions and deletions in the recA sequence alignment provided additional phylogenetic evidence. Relative to all the other sequences (including outgroups), M. fortuitum, M. mucogenicum, M. smegmatis, M. mageritense and M. wolinskyi groups exhibited insertions of 3 and 6 bp at positions 7–9 and 1012–1017. Likewise, M. tuberculosis, M. fortuitum and M. peregrinum have an additional 6-bp insertion at positions 1018–1023. These related deletions and insertions in sequence provided good support for the monophyly of these trees as shown in Figs 1, 4 and 5. No deletions or insertions were found in the partial sequences of hsp65 (420 bp) and sodA (441 bp).

Comparison between rpoB, recA, sodA, hsp65 and 16S rRNA gene phylogenetic trees
rpoB-, recA-, sodA-, hsp65- and 16S rRNA gene-based phylogeny inferred from neighbour-joining was compared. The topologies of the rpoB-based phylogenetic trees (Fig. 5) were almost the same as those derived from 16S rRNA and recA gene sequence analysis (Figs 1 and 4). However, the trees derived from rpoB gene sequence showed higher bootstrap values and more divergence than the recA and 16S rRNA gene-based trees. Bootstrap support for each cluster was higher than 90 % in the rpoB sequence analysis. The non-pigmented M. mageritense and M. wolinskyi groups were not well defined, with lower bootstrap values in the recA and 16S rRNA gene-based trees. All tested species showed good discrimination with regard to the rpoB gene. On the other hand, M. abscessus and M. chelonae, poorly discriminated by the 16S rRNA gene sequence (Adékambi et al., 2003; Kirschner et al., 1992), were clearly delineated as two closely related species, like M. senegalense and M. farcinogenes. Similarly, M. mucogenicum ATCC 49649 was clearly separated from M. mucogenicum ATCC 49650^T and M. mucogenicum ATCC 49651 despite the fact that they belong to the same species (Springer et al., 1995). As for the hsp65 gene, M. mageritense and M. wolinskyi formed a monophyletic group with a non-significant bootstrap value (55 %). In the sodA gene sequence analysis, M. mageritense formed a sister cluster with that formed by M. houstonense, M. senegalense and M. farcinogenes with a non-significant bootstrap value (64 %). Indeed, the bootstrap values at the nodes of a few clusters were too low to induce much confidence with regard to the hsp65 and sodA trees (Figs 2 and 3) and the suggestion that the relationships among the species are not the same in comparison to the 16S rRNA, recA and rpoB genes. These discrepancies can arise from inadequate sample sizes (Bull et al., 1993), since it was recently suggested that long sequences of several kilobases should be used in studies that aim to use clock tests to select sequences that approximate rate constancy (Bromham et al., 2000). These findings suggested that the rpoB- and recA-based phylogenies of RGM can be additional phylogenetic tools that globally support the 16S rRNA gene-based phylogeny of RGM.

However, separate analyses of individual datasets ignore hidden character support within datasets that emerges in combined analysis, and this support can be substantial (Gatesy & Arctander, 2000). Separate analyses of different character sets are not necessary to detect conflict among datasets and can distort interpretations of common character support (Gatesy et al., 1999). Therefore, it was suggested that the distribution of conflicts/support among datasets in a comprehensive combined analysis should be used to assess dataset congruence and to question the strength of support for different clusters. Keeping account of potentially dependent characters in simultaneous analysis offers the researcher a better understanding of the global evidence. In our study, nearly 7000 bp were analysed, corresponding to roughly 0·05 % of the genomic sequence and 0·1 % of the predicted genes. To our knowledge, this is the first reported example of so many nucleotides incorporated into a combined dataset analysis of a bacterial genus. When using the ILD test to quantify the conflicts that can occur between sets of characters from different data sources, combining rpoB and recA genes (nearly 4600 bp) broadly supported the expected phylogeny and improved the bootstrap values. Also, removal of two not-well-delineated groups, M. mageritense and M. wolinskyi, increased the mean bootstrap value to 99 % at the node of each group. However, the lower bootstrap value of certain species within a group for combined genes relative to randomly resampled nucleotides is consistent with the hypothesis that nucleotides within genes have not evolved independently (Averof et al., 2000; Cummings et al., 1995).

According to these observations, the two combined datasets obtained allowed us to discriminate six groups, as described below (Figs 6 and 7).

The M. chelonae–abscessus group (cluster I) was created to accommodate species with distinctive properties including positive 3-day arylsulfatase, better growth at 30 than at 35 °C, negative nitrate reductase, negative iron uptake and resistance to polymyxin B (Brown-Elliott & Wallace, 2002). They formed a defined cluster in the 16S rRNA and hsp65 gene phylogenies (Tortoli, 2003; Wilson et al., 2001). The recently described M. immunogenum was added to this group according to its phenotypic characteristics, its unique hsp65 sequence pattern and its 15–18 % DNA–DNA relatedness to other species of this group (Wilson et al., 2001). Our analyses of the single and combined datasets fully supported both the existence of this group and its taxonomic relationships. Further evidence was provided by molecular signatures: a 3-bp insertion at positions 34–36 of the rpoB gene, 3- and 6-bp deletions at positions 7–9 and 1012–1017 of the recA gene and a 1-bp deletion at position 1267 of the 16S rRNA gene. A recent study of ITS sequences (Hamid et al., 2002) confirmed that the M. chelonae–abscessus group was a separate group. With the data from the five combined genes, M. immunogenum appears to be a hybrid between M. abscessus and M. chelonae, with a 55 % bootstrap value. This phenomenon was previously described using PRA band patterns from a 439-bp segment of the hsp65 gene digested with BstEII and HaeIII (Wilson et al., 2001). Further studies are warranted to assess better the taxonomic relationships of this emerging taxon.

The M. mucogenicum group (cluster II) formed a distinct cluster in our study. This was discrepant with previous suggestions of its incorporation into the M. chelonae–abscessus group (Wallace et al., 1993) or the M. fortuitum group (Brown-Elliott & Wallace, 2002). Some phenotypic characteristics support our analysis, notably the highly mucoid aspect of the colonies and the susceptibility of M. mucogenicum to polymyxin B (Springer et al., 1995). Species of the M. mucogenicum group utilize mannitol as a carbon source, in contrast to species of the M. chelonae–abscessus group, and also citrate, in contrast to species of the M. fortuitum group (Springer et al., 1995). Likewise, comparative thin-layer and gas chromatography of fatty esters and alcohols showed clear differentiation of M. mucogenicum isolates from isolates belonging to the M. fortuitum and M. chelonae–abscessus groups (Muñoz et al., 1997; Wallace et al., 1993). In addition, M. mucogenicum strain ATCC 49649 differs from M. mucogenicum ATCC 49650^T by exhibiting high level of semi-quantitative catalase activity, cephalothin resistance and negative acetamidase (Wallace et al., 1993). The 16S rRNA gene sequence of M. mucogenicum ATCC 49649 (GenBank accession no. AY457073) obtained in our laboratory showed a double C and T (Y) peak at position 448, suggesting the presence of two 16S rRNA gene alleles. This was in accordance with the fact that RGM have two 16S rRNA gene copies with the exception of M. abscessus and M. chelonae, which have only one copy (Domenech et al., 1994). Double peaks were not observed in other RGM 16S rRNA genes sequenced in our laboratory. Furthermore, M. mucogenicum ATCC 49650^T was the only strain to contain a lipooligosaccharide composed of glucose and galactose (Muñoz et al., 1998). The percentage rpoB gene sequence similarity between these two strains suggested that, like M. peregrinum and M. septicum, they may be representatives of two closely related species, as they occupy similar positions in the M. mucogenicum group of the phylogenetic tree (Fig. 6). Further evidence is provided by a 3-bp deletion at positions 34–36 of the rpoB gene and a 1-bp deletion at position 818 of the 16S rRNA gene.

The M. fortuitum group (cluster III) group included three RGM of veterinary interest, M. porcinum, M. senegalense and M. farcinogenes, as determined by 16S rRNA, hsp65, sodA, recA and rpoB gene sequencing. This result is in good agreement with DNA–DNA relatedness studies, which indicated a moderate level of relatedness between M. fortuitum, M. peregrinum, M. farcinogenes and M. senegalense (Lévy-Frébault et al., 1986; Schinsky et al., 2000). Numerical analyses (Ridell & Goodfellow, 1983; Tsukamura & Ichiyama, 1986), comparative immunology (Lopez-Marin et al., 1993; Lanéelle et al., 1996) and chemotaxonomic properties (Kirschner et al., 1992; Wallace et al., 2002) revealed a high degree of similarity within the species of the M. fortuitum group. The 99 % rpoB gene sequence similarity between M. houstonense and M. fortuitum suggested that these strains may be closely related subspecies, although M. houstonense showed resistance to pipemidic acid, biochemical differences such as mannitol, inositol, sorbitol and trehalose utilization (Kirschner et al., 1992) and three base differences in the 16S rRNA gene sequence. The combined datasets delineated two subclusters, IIIa (M. peregrinum, M. septicum, M. neworleansense and M. porcinum) and IIIb (M. fortuitum, M. houstonense, M. senegalense and M. farcinogenes), within this group. However, a minor discrepancy was observed within these subclusters. In contrast to the 16S rRNA, hsp65, sodA and recA gene sequence analyses and the combined datasets, where M. porcinum belongs to subcluster IIIa, rpoB gene sequence analysis placed it in subcluster IIIb with a 65 % bootstrap value, although it possessed a 3-bp deletion at positions 22–24 characteristic of the species of subcluster IIIa (Fig. 6). These data suggested an intraspecific horizontal rpoB gene fragment exchange, as recently demonstrated by Lorenz & Sikorski (2000). In this group, the two subclusters IIIa and IIIb can be differentiated in that species of subcluster IIIb grow at 42 °C (Ridell & Goodfellow, 1983; Tsukamura & Ichiyama, 1986; Schinsky et al., 2000) and utilize sorbitol as a carbon source. Species of this group are responsible for animal infections (Youssef et al., 2002). Furthermore, M. fortuitum was the most commonly isolated mycobacterium obtained from chronic non-healing skin lesions in dogs and cats (Jang & Hirsh, 2002). Youssef et al. (2002) proposed the name ‘M. fortuitum–peregrinum group’ on the basis of tissue culture, chromatography and histopathological examination of pyogranulomatous panniculitis in a 6-year-old cat. This name seems to be appropriate, since M. fortuitum and M. peregrinum were the first species described in subclusters IIIa and IIIb, respectively. A recent study of the ITS sequence showed that the M. fortuitum group was a separate group (Hamid et al., 2002). In this group, M. senegalense was indistinguishable from M. farcinogenes on the basis of all genes analysed except rpoB (0·5 % divergence). This divergence was not in agreement with the suggestion that isolates may have species status when they exhibit 3 % rpoB gene sequence divergence (Adékambi et al., 2003). However, these two mycobacteria can be differentiated on the basis of their DNA–DNA relatedness, growth rates, biochemical activity and histopathological behaviour (Chamoiseau, 1979) as well as sequencing of the ITS (Hamid et al., 2002).

The M. wolinskyi group (cluster IV) and M. mageritense group (cluster V) represent independent branches within the trees, although they appear together in the 16S rRNA and hsp65 gene sequence trees with low bootstrap values of 66 and 55 %, respectively. These two genes were found to be less informative than sodA, recA and rpoB. The relative branching order of these sequences was not well supported by bootstrap values. Determining the relationships between these organisms and other species under study will require sequencing of additional RGM isolates such as M. wolinskyi ATCC 70009, whose 16S rRNA gene sequence (GenBank accession no. Y12871) differs from that of the type strain (Tortoli, 2003). M. wolinskyi and M. mageritense formed two separate groups, separated from the M. smegmatis and M. fortuitum groups, respectively. This result is supported by DNA–DNA hybridization values of <28 % between M. mageritense and the M. fortuitum group and of <20 % between M. wolinskyi and M. smegmatis group species (Domenech et al., 1997; Brown et al., 1999). Furthermore, M. wolinskyi was distinguished from the M. smegmatis group due to the absence of pigmentation. Similarly, M. mageritense was distinguished from the M. fortuitum group by exhibiting an intermediate amikacin MIC, resistance to kanamycin, a cefoxitin MIC of 32 µg ml^–1 and the use of L-rhamnose as a carbon source (Wallace et al., 2002). Moreover, comparison of enzymic activities (M. mageritense but not M. wolinskyi had a positive 3-day arylsulfatase reaction) and DNA–DNA relatedness suggested that these two groups were intermediate between the members of the M. fortuitum and thermotolerant M. smegmatis groups (Domenech et al., 1997; Brown et al., 1999; Wallace et al., 2002).

Species of the M. smegmatis group (cluster VI) were thermotolerant (M. smegmatis and M. goodii) with the exception of M. wolinskyi, as previously described (Brown-Elliott & Wallace, 2002). This proposition is in good agreement in respect of DNA–DNA relatedness, general absence of susceptibility to new macrolides including clarithromycin and positive 3-day arylsulfatase reaction. However, M. smegmatis is highly susceptible to tobramycin (MIC1 µg ml^–1), whereas M. goodii has intermediate susceptibility (MIC 2–8 µg ml^–1) and M. wolinskyi is resistant to tobramycin (MIC>8 µg ml^–1) (Brown et al., 1999). Further evidence was provided by a 3-bp deletion at positions 28–30 in the rpoB gene sequence and 2-bp insertions at positions 82–83 and 185–186 and a 1-bp insertion at position 202 and a 1-bp deletion at position 178 of the 16S rRNA gene sequence.

Since the initiation of our study, two species were described as emerging human RGM pathogens, Mycobacterium boenickei and Mycobacterium brisbanense (Schinsky et al., 2004), which can be placed within existing clusters on the basis of their 16S rRNA gene sequences. It would be interesting to confirm these placements when using the molecular tools herein described. Among the genes we studied, the rpoB sequence-based relationships were in accordance with those published previously.

In conclusion, our study indicated that there was a robust and consistent phylogeny for the genomic backbone of housekeeping genes in bacteria. Because of the bootstrap values of the rpoB, recA and 16S rRNA gene sequences and their topology, all three genes could be used when describing a novel RGM species or when isolates are to be reported.

	ACKNOWLEDGEMENTS

 
We thank Christiane Bibard and Antoine Carlioz for technical assistance, Pierre Darlu (INSERM U535 Génétique épidémiologique et Structure des populations humaines) for access to PAUP4.0b and the ILD test and Babu V. A. K. Gundi for expert reviewing of the manuscript.

	REFERENCES

TOP ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION REFERENCES

 
Adékambi, T., Colson, P. & Drancourt, M. (2003). rpoB-based identification of nonpigmented and late-pigmenting rapidly growing mycobacteria. J Clin Microbiol 41, 5699–5708.[Abstract/Free Full Text]

Averof, M., Rokas, A., Wolfe, K. H. & Sharp, P. M. (2000). Evidence for a high frequency of simultaneous double-nucleotide substitutions. Science 287, 1283–1286.[Abstract/Free Full Text]

Blackwood, K. S., He, C., Gunton, J., Turenne, C. Y., Wolfe, J. & Kabani, A. M. (2000). Evaluation of recA sequences for identification of Mycobacterium species. J Clin Microbiol 38, 2846–2852.[Abstract/Free Full Text]

Bromham, L., Penny, D., Rambaut, A. & Hendy, M. D. (2000). The power of relative rates tests depends on the data. J Mol Evol 50, 296–301.[Medline]

Brown, B. A., Springer, B., Steingrube, V. A. & 10 other authors (1999). Mycobacterium wolinskyi sp. nov. and Mycobacterium goodii sp. nov., two new rapidly growing species related to Mycobacterium smegmatis and associated with human wound infections: a cooperative study from the International Working Group on Mycobacterial Taxonomy. Int J Syst Bacteriol 49, 1493–1511.[Abstract/Free Full Text]

Brown-Elliott, B. A. & Wallace, R. J., Jr (2002). Clinical and taxonomic status of pathogenic nonpigmented or late-pigmenting rapidly growing mycobacteria. Clin Microbiol Rev 15, 716–746.[Abstract/Free Full Text]

Brunello, F., Ligozzi, M., Cristelli, E., Bonora, S., Tortoli, E. & Fontana, R. (2001). Identification of 54 mycobacterial species by PCR-restriction fragment length polymorphism analysis of the hsp65 gene. J Clin Microbiol 39, 2799–2806.[Abstract/Free Full Text]

Bull, J. J., Huelsenbeck, J., Cunningham, C. W., Swofford, D. L. & Wadell, P. J. (1993). Partitioning and combining data in phylogenetic analysis. Syst Biol 42, 384–397.[CrossRef]

Chamoiseau, G. (1979). Etiology of farcy in African bovines: nomenclatures of the causal organisms Mycobacterium farcinogenes Chamoiseau and Mycobacterium senegalense (Chamoiseau) comb. nov. Int J Syst Bacteriol 29, 407–410.[Abstract/Free Full Text]

Cummings, M. P., Otto, S. P. & Wakeley, J. (1995). Sampling properties of DNA sequence data in phylogenetic analysis. Mol Biol Evol 12, 814–822.[Abstract]

Cunningham, C. W. (1997). Can three incongruence tests predict when data should be combined? Mol Biol Evol 14, 733–740.[Abstract]

Darlu, P. & Lecointre, G. (2002). When does the incongruence length difference test fail? Mol Biol Evol 19, 432–437.[Abstract/Free Full Text]

Domenech, P., Menendez, M. C. & Garcia, M. J. (1994). Restriction fragment length polymorphisms of 16S rRNA genes in the differentiation of fast-growing mycobacterial species. FEMS Microbiol Lett 116, 19–24.[CrossRef][Medline]

Domenech, P., Jimenez, M. S., Menendez, M. C., Bull, T. J., Samper, S., Manrique, A. & Garcia, M. J. (1997). Mycobacterium mageritense sp. nov. Int J Syst Bacteriol 47, 535–540.[Abstract/Free Full Text]

Drancourt, M., Bollet, C., Carlioz, A., Martelin, R., Gayral, J. P. & Raoult, D. (2000). 16S ribosomal DNA sequence analysis of a large collection of environmental and clinical unidentifiable bacterial isolates. J Clin Microbiol 38, 3623–3630.[Abstract/Free Full Text]

Fox, G. E., Wisotzkey, J. D. & Jurtshuk, P., Jr (1992). How close is close: 16S rRNA sequence identity may not be sufficient to guarantee species identity. Int J Syst Bacteriol 42, 166–170.[Abstract/Free Full Text]

Gatesy, J. & Arctander, P. (2000). Hidden morphological support for the phylogenetic placement of Pseudoryx nghetinhensis with bovine bovids: a combined analysis of gross anatomical evidence and DNA sequences from five genes. Syst Biol 49, 515–538.[CrossRef][Medline]

Gatesy, J., O'Grady, P. & Baker, R. (1999). Corroboration among data sets in simultaneous analysis: hidden support for phylogenetic relationships among higher level artiodactyls taxa. Cladistics 15, 271–313.[CrossRef]

Gaunt, M. W., Turner, S. L., Rigottier-Gois, L., Lloyd-Macgilp, S. A. & Young, J. P. W. (2001). Phylogenies of atpD and recA support the small subunit rRNA-based classification of rhizobia. Int J Syst Evol Microbiol 51, 2037–2048.[Abstract]

Hamid, M. E., Roth, A., Landt, O., Kroppenstedt, R. M., Goodfellow, M. & Mauch, H. (2002). Differentiation between Mycobacterium farcinogenes and Mycobacterium senegalense strains based on 16S-23S ribosomal DNA internal transcribed spacer sequences. J Clin Microbiol 40, 707–711.[Abstract/Free Full Text]

Jang, S. S. & Hirsh, D. C. (2002). Rapidly growing members of the genus Mycobacterium affecting dogs and cats. J Am Anim Hosp Assoc 38, 217–220.[Abstract/Free Full Text]

Kim, B. J., Lee, S. H., Lyu, M. A., Kim, S. J., Bai, G. H., Chae, G. T., Kim, E. C., Cha, C. Y. & Kook, Y. H. (1999). Identification of mycobacterial species by comparative sequence analysis of the RNA polymerase gene (rpoB). J Clin Microbiol 37, 1714–1720.[Abstract/Free Full Text]

Kirschner, P., Kiekenbeck, M., Meissner, D., Wolters, J. & Böttger, E. C. (1992). Genetic heterogeneity within Mycobacterium fortuitum complex species: genotypic criteria for identification. J Clin Microbiol 30, 2772–2775.[Abstract/Free Full Text]

Kumar, S., Tamura, K., Jakobsen, I. B. & Nei, M. (2001). MEGA2: molecular evolutionary genetics analysis software. Bioinformatics 17, 1244–1245.[Abstract/Free Full Text]

Lanéelle, M. A., Silve, G., Lopez-Marin, L. M. & Daffe, M. (1996). Structures of the glycolipid antigens of members of the third biovariant complex of Mycobacterium fortuitum. Eur J Biochem 15, 270–279.[CrossRef]

Lévy-Frébault, V., Grimont, F., Grimont, P. A. D. & David, H. L. (1986). Deoxyribonucleic acid relatedness study of the Mycobacterium fortuitum-Mycobacterium chelonae complex. Int J Syst Bacteriol 36, 458–460.[Abstract/Free Full Text]

Lopez-Marin, L. M., Laneelle, M. A., Prome, D. & Daffe, M. (1993). Structures of the glycopeptidolipid antigens of two animal pathogens: Mycobacterium senegalense and Mycobacterium porcinum. Eur J Biochem 215, 859–866.[Medline]

Lorenz, M. G. & Sikorski, J. (2000). The potential for intraspecific horizontal gene exchange by natural genetic transformation: sexual isolation among genomovars of Pseudomonas stutzeri. Microbiology 146, 3081–3090.[Abstract/Free Full Text]

Muñoz, M., Julian, E., Garcia-Barcelo, M., Ausina, V. & Luquin, M. (1997). Easy differentiation of Mycobacterium mucogenicum from other species of the Mycobacterium fortuitum complex by thin-layer and gas chromatography of fatty esters and alcohols. J Chromatogr B Biomed Sci Appl 689, 341–347.[CrossRef][Medline]

Muñoz, M., Raynaud, C., Lanéelle, M.-A., Julián, E., López Marín, L. M., Silve, G., Ausina, V., Daffé, M. & Luquin, M. (1998). Seroreactive species-specific lipooligosaccharides of Mycobacterium mucogenicum sp. nov. (formerly Mycobacterium chelonae-like organisms): identification and chemical characterization. Microbiology 144, 137–148.[Abstract]

Pfyffer, G. E., Brown-Elliot, B. A. & Wallace, R. J., Jr (2003). Mycobacterium: general characteristics, isolation, and staining procedures. In Manual of Clinical Microbiology, 8th edn, pp. 532–559. Edited by P. R. Murray, E. J. Baron, J. H. Jorgensen, M. A. Pfaller & R. H. Yolken. Washington, DC: American Society for Microbiology.

Ridell, M. & Goodfellow, M. (1983). Numerical classification of Mycobacterium farcinogenes, Mycobacterium senegalense and related taxa. J Gen Microbiol 129, 599–611.[Medline]

Rinquet, H., Akoua-Koffi, C., Honore, S., Varnerot, A., Vincent, V., Berche, P., Gaillard, J. L. & Pierre-Audigier, C. (1999). hsp65 sequencing for identification of rapidly growing mycobacteria. J Clin Microbiol 37, 852–857.[Abstract/Free Full Text]

Roth, A., Reischl, U., Streubel, A., Naumann, L., Kroppenstedt, R. M., Habicht, M., Fischer, M. & Mauch, H. (2000). Novel diagnostic algorithm for identification of mycobacteria using genus-specific amplification of the 16S-23S rRNA gene spacer and restriction endonucleases. J Clin Microbiol 38, 1094–1104.[Abstract/Free Full Text]

Schinsky, M. F., McNeil, M. M., Whitney, A. M., Steigerwalt, A. G., Lasker, B. A., Floyd, M. M., Hogg, G. G., Brenner, D. J. & Brown, J. M. (2000). Mycobacterium septicum sp. nov., a new rapidly growing species associated with catheter-related bacteraemia. Int J Syst Evol Microbiol 50, 575–581.[Abstract]

Schinsky, M. F., Morey, R. E., Steigerwalt, G. A. & 9 other authors (2004). Taxonomic variation in the Mycobacterium fortuitum third biovariant complex: description of Mycobacterium boenickei sp. nov., Mycobacterium houstonense sp. nov., Mycobacterium neworleansense sp. nov. and Mycobacterium brisbanense sp. nov. and recognition of Mycobacterium porcinum from human clinical isolates. Int J Syst Evol Microbiol 54, 1653–1667.[Abstract/Free Full Text]

Springer, B., Böttger, E. C., Kirschner, P. & Wallace, R. J., Jr (1995). Phylogeny of the Mycobacterium chelonae-like organism based on partial sequencing of 16S rRNA gene and proposal of Mycobacterium mucogenicum sp. nov. Int J Syst Bacteriol 45, 262–267.[Abstract/Free Full Text]

Steingrube, V. A., Gibson, J. L., Brown, B. A., Zhang, Y., Wilson, R. W., Rajagonpalan, M. & Wallace, R. J., Jr (1995). PCR amplification and restriction endonuclease analysis of a 65-kilodalton heat shock protein gene sequence for taxonomic separation of rapidly growing mycobacteria. J Clin Microbiol 33, 149–153.[Abstract]

Swofford, D. L. (1998). PAUP*. Phylogenetic analysis using parsimony (*and other methods), version 4. Sunderland, MA: Sinauer Associates.

Telenti, A., Marchesi, F., Balz, M., Bally, F., Böttger, E. & Bodmer, T. (1993). Rapidly growing mycobacteria to the species level by polymerase chain reaction and restriction enzyme analysis. J Clin Microbiol 31, 175–178.[Abstract/Free Full Text]

Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F. & Higgins, D. G. (1997). The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res 25, 4876–4882.[Abstract/Free Full Text]

Tortoli, E. (2003). Impact of genotypic studies on mycobacterial taxonomy: the new mycobacteria of the 1990s. Clin Microbiol Rev 16, 319–354.[Abstract/Free Full Text]

Tsukamura, M. & Ichiyama, S. (1986). Numerical classification of rapidly growing nonphotochromogenic mycobacteria. Microbiol Immunol 30, 863–882.[Medline]

Wallace, R. J., Jr, Silcox, V. A., Tsukamura, M., Brown, B. A., Kilburn, J. O., Butler, W. R. & Onyi, G. O. (1993). Clinical significance, biochemical features, and susceptibility patterns of sporadic isolates of the Mycobacterium chelonae-like organism. J Clin Microbiol 31, 3231–3239.[Abstract/Free Full Text]

Wallace, R. J., Jr, Brown-Elliot, B. A., Hall, L. & 9 other authors (2002). Clinical and laboratory features of Mycobacterium mageritense. J Clin Microbiol 40, 2930–2935.[Abstract/Free Full Text]

Wilson, R. W., Steingrube, V. A., Böttger, E. C. & 11 other authors (2001). Mycobacterium immunogenum sp. nov., a novel species related to Mycobacterium abscessus and associated with clinical disease, pseudo-outbreaks and contaminated metalworking fluids: an international cooperative study on mycobacterial taxonomy. Int J Syst Evol Microbiol 51, 1751–1764.[Abstract]

Woese, C. R. (1987). Bacterial evolution. Microbiol Rev 51, 221–271.