All available type strains of M. chelonae-abscessus complex were obtained from American Type Culture Collection (ATCC), Collection of Institut Pasteur, or Culture Collection, University of Göteborg, Sweden. Previously identified clinical isolates of M. abscessus, and M. chelonae by partial 16S rRNA gene sequencing and internal transcribed spacer (ITS) PCR were retrieved for comparative analysis. A subset of these isolates was described in a prior study (8). CVs were defined as isolates that were cefoxitin susceptible or showed intermediate resistance, and identified as M. chelonae by partial 16S rRNA gene sequencing and ITS PCR (9) by Associated Regional and University Pathologists Laboratories and the Hospital of the University of Pennsylvania or by failure to amplify hsp65 by PCR restriction fragment length polymorphism analysis (10,11) by the University of Texas Health Science Center. Microbiologic and medical records were reviewed for clinical information for select isolates. Clinical case reviews were conducted under institutional review board approved protocols at Associated Regional and University Pathologists Laboratories, Children’s Hospital of Philadelphia, University of Pennsylvania, and the University of Texas Health Science Center. Study isolates CV002 (ATCC [pending] and DSMZ 45524) and CV005 (ATCC [pending]) have been deposited into culture collections.

DNA extractions, PCR, and sequencing reactions were performed as described (8). Amplifications and sequencing reactions were performed by using primers specific for ≈1,400 bp of the 16S rRNA 5F (5′-TTGGAGAGTTTGATCCTGGCTC-3′) and 1492R (5′-ACGGITACCTTGTTACGACTT-3′), ≈700 bp of rpoB, ≈400 bp of sodA (12), ≈240 bp of the ITS (13), and ≈400 bp of hsp65 (11) genes or region. Sequence alignments and phylogenetic trees were constructed by using neighbor-joining method with Kimura 2-parameter distance correction model and 1,000 bootstrap replications in MEGA4 (14). Only unique sequences (sequevars) were included in the trees.

Final species identifications were based on comparisons of sequences for the full 16S rRNA and the partial rpoB genes to GenBank references of type strains. Full-length 16S rRNA sequences were used in this study, and we used 99.5% shared identity for identification to a type strain sequence. Species identification for rpoB gene was based on an identity of 98.0%–100% as outlined by Adekambi et al. (12,15).

Purified DNA of the type strains and the patient isolates CV002, CV004, CV005, CV006, and CV005 was prepared as described (16). CV002 and CV005 strains were labeled with [³²P] dCTP using the Nick Translation Kit (Invitrogen, Carlsbad, CA, USA). Labeled DNA from the CV002 was hybridized with unlabeled DNA from isolates CV002, CV004, CV005, CV006, and CV015 and with unlabeled DNA from the type strains. Labeled DNA from patient isolate CV005 was then hybridized with unlabeled DNA from isolates CV002 and CV005. The reciprocal experiment was performed because of the nearness of CV005 to the 70% cut-off designated by Wayne (17) and the 0% divergence obtained in the first experiment. Hybridization was performed as previously described (18). All reactions were performed in duplicate at 70°C. The relative binding ratio (RBR) was calculated by using the method of Brenner et al. (19). The percentage divergence (calculated to the nearest 0.5%) was determined by assuming that each degree of heteroduplex instability, when compared with the melting temperature of the homologous duplex, was caused by 1% unpaired bases (19).

Single-end DNA libraries of M. bolletii CIP 108541^T, M. chelonae ATCC 35752^T, M. immunogenum CIP 106684^T, M. massiliense CCUG 48898^T, and CV002 were prepared using Illumina DNA Sample Kit (Illumina, Inc., San Diego, CA, USA) according to manufacturer’s recommendations. Sequencing was performed in individual flow cell lanes on the Illumina Genome Analyzer (Illumina, Inc.) at the University of Utah Huntsman Cancer Institute Core Sequencing Facility.

De novo assembly of raw Illumina sequence data was achieved by using Velvet software (20). Velvet was run in 2 parts, velveth and velvetg. For velveth, the hash length was set at 23 (value was selected by calculations in the software manual); default settings were used for all other parameters. In velvetg, the –cov cutoff value was set to auto (setting allows software to automate appropriate coverage cutoff), and –min_contig_lgth was set to 100; all other settings were default.

The genome of M. abscessus CIP 104536^T (21) was used as the source reference set of 123 genes that were identified as likely core genome components for the phylum Actinobacter by Ventura et al. (22). The set of reference genes was randomly divided into 5 similarly sized sets to facilitate analysis. SeqMan (DNASTAR Inc., Madison, WI, USA) was used to align the assembled contigs from the sequenced species against each of the sets of reference genes. DNA and inferred amino acid sequences were aligned using MEGA. Only near full-length genes were used in future comparisons. Confirming the translation of the gene was in the correct reading frame relative to the different isolates substantiated quality of each assembled gene. The DNA and amino acid sequences were concatenated for each isolate and sequences alignments and phylogenetic trees were constructed in MEGA using the neighbor-joining method. Kimura 2-parameter distance correction was used for DNA and Poisson correction model was used for amino acid trees; 1,000 bootstrap replications were used for each tree constructed.

We determined antimicrobial susceptibility by broth microdilution using the recommended Clinical and Laboratory Standards Institute guidelines for rapidly growing Mycobacterium spp (23). Some isolates were not tested for all antimicrobial agents and the concentrations of antimicrobial agents that were tested varied in the panels. MICs of clarithromycin were assessed at 3 days.

We obtained 6 type strains representing all members of the M. chelonae-abscessus complex. All type strains and all 127 archived isolates underwent multilocus sequencing. For the 127 archived clinical isolates, we identified 64 M. abscessus, 58 M. chelonae, and 5 M. massiliense isolates. We designated an additional 26 isolates as CV on the basis of our case definition. All 26 CV isolates underwent rpoB gene sequencing, and a subset (n = 11) underwent multilocus sequencing. Unique sequevars are designated among the species M. abscessus, M. massiliense, M. bolletii, M. chelonae, and CV isolates.

Full-length genes were successfully assembled for 43 genes from 5 type strains and 1 representative clinical CV isolate. Gene names and corresponding GenBank accession numbers are provided in the supplementary tables (Technical Appendix Table 2). Pair-wise alignments of the DNA and amino acid sequences were performed for each isolate (Technical Appendix Tables 3, 4). The 43 concatenated genes ranged from 41,580 to 41,619 bp and created an alignment of 41,792 nt including gaps. Comparisons of DNA percent identity ranged from 89.1% (M. chelonae to M. abscessus) to 98.3% (M. abscessus to M. bolletii). M. massiliense, M. abscessus, and M. bolletii showed a close association with percent identity range of 98.2%–98.3%. The novel CV isolate shared no greater than 90.5% identity with any type strain. Phylogenetic analysis shows closest relationship between M. abscessus, M. massiliense, and M. bolletii (Figure 6, panel A). The CV isolate is on a separate branch with nearly identical distances from M. immunogenum and M. chelonae.

Percent identity of amino acid comparisons ranged from 95.7% (M. chelonae to M. abscessus) to 99.6% (M. abscessus to M. bolletii). M. massiliense, M. abscessus, and M. bolletii showed a close association with percent identity range of 99.5%–99.6%. The novel CV isolate shared no greater than 96.1% identity with any type strain. Phylogenetic analysis shows closest relationship between M. abscessus, M. massiliense, and M. bolletii (Figure 6, panel B).

DNA-DNA hybridization studies of clinical isolates CV004, CV005, CV006, and CV015 with labeled patient isolate CV002 showed RBRs of 66 to 96% and %D values of 0.0 to 2.0%. The reciprocal DNA-DNA hybridization of patient isolate CV002 with labeled isolate CV005 showed an RBR of 100% and a %D of 0.5%. The DNA-DNA hybridization studies of the phylogenetically (16S rRNA gene) related type strains M. abscessus, M. bolletii, M. chelonae, M. immunogenum, M. massiliense, and M. salmoniphilum performed with the patient isolate CV002 showed RBRs of 15%–69% and %D values of 5.0%–8.0%. DNA-DNA hybridization results are summarized in Technical Appendix Table 5.

Susceptibilities by broth microdilution were available for most M. chelonae, M. abscessus, M. massiliense, and CV isolates (Table A1). The MICs for the first 3 taxa were comparable those in to previous reports (24). With the exception of 1 isolate, all isolates of M. chelonae were cefoxitin resistant (MIC >128 µg/mL), and M. abscessus and M. massiliense were cefoxitin intermediate (MIC 32–64 µg/mL). All 3 taxa were minocycline resistant. Unlike M. chelonae, CV isolates were intermediate (88%) or susceptible (12%) to cefoxitin.

The most common source for the CV isolates was respiratory (n = 20) (Technical AppendixTable 6). The remaining 6 CV isolates had clinical sources that included skin (n = 2), granulomatous liver lesion (n = 1), central line infections (n = 2), and an unspecified body fluid. Most isolates (n = 15) were recovered from patients seen in 4 different hospitals or clinics in Pennsylvania. Eight patients with CV infection were seen at the Hospital of the University of Pennsylvania and 7 of 8 patients were adult females (ages 41–74 years) who acquired the mycobacterial infection as outpatients. Six charts were available for review, and partial information was available for 2 additional patients (patients CV007, CV008, CV010, CV012–CV014, CV034, and CV036).

The medical histories of the 6 patients with complete information fell into 2 groups: those with chronic sinusitis (2 patients) and those with lower respiratory symptoms (4 patients). All patients with lower respiratory symptoms and cultures positive for CV had underlying pulmonary disease (cystic fibrosis, primary ciliary dyskinesia, lung cancer, chronic obstructive pulmonary disease, recurrent pneumonia/bronchiectasis). No patients with lower respiratory symptoms received specific antibiotic therapy aimed at treating rapidly growing mycobacterial infection, although 1 patient received long term antimicrobial drug therapy for concomitant M. avium infection. Two patients with sinusitis were not treated with antimicrobial drugs, but both had sinus surgery with symptomatic improvement. One patient had 3 positive sputum cultures for a rapidly growing Mycobacterium over a 3-year period, 2 of which were shown by genetic sequencing to be CV organisms (CV014 was selected for further study). No other patient had >1 positive culture for CV, if follow-up cultures were performed. Three patients had sequential or concomitant infections with M. abscessus, M. avium-intracellulare, or both bacteria.

The first isolate identified as a CV was discovered in 2005 from a patient in New York. The next isolate was not identified until 2007 from Pennsylvania. Overall, 23 (88%) of 26 were isolated in patients in the northeastern United States, and the remaining 3 isolates were recovered from patients in Minnesota, Oregon, and Colorado.