Mycobacterium riyadhense sp. nov., a non-tuberculous species identified as Mycobacterium tuberculosis complex by a commercial line-probe assay

A 19-year-old male reported to the otolaryngology department of the King Faisal Hospital with pain and swelling of the left side of his face with protrusion of left eye after blunt trauma. A computed tomography scan of the sinuses was performed, which revealed a tumour in the left maxillary sinus. The tumour involved the nasal septum, extended into the left orbit and infiltrated the medial and inferior rectus as well as the optic nerve. A chest radiograph revealed no abnormalities. A lavage of both maxillary sinuses was performed for diagnosis. Mycobacterial cultures were performed on Lowenstein–Jensen medium and yielded an isolate with unusual colony morphology after 3 weeks of incubation at 36 °C.

The patient was presumed to have bone tuberculosis and started a 9-month anti-tuberculosis regimen. The patient improved both clinically and radiologically. He has not suffered a relapse since. The isolated strain was sent to the Dutch National Mycobacteria Reference Laboratory (RIVM, Bilthoven, Netherlands) for identification, as part of a second-line quality control programme. Identification of the strain was first attempted using three commercial line-probe assays, Hain GenoType MTBC and GenoType AS/CM (Hain Lifescience GmbH) and INNO-LiPA MYCOBACTERIA v2 (Innogenetics NV), all used according to the manufacturers' instructions.

To obtain identification to the species level, we sequenced the 16S rRNA gene, 16S–23S internal transcribed spacer (ITS) and rpoB and hsp65 genes, using previously described approaches (Springer et al., 1996; Roth et al., 1998; Kim et al., 1999; Telenti et al., 1993). The DNA sequences obtained were compared with the RIDOM (Ribosomal Differentiation of Medical Microorganisms; ) and GenBank (National Center for Biotechnology Information; ) databases.

To establish the presence of a region of difference 1 (RD1)-like element and the DNA sequences of the esat-6 and cfp-10 genes, we used primers from a previous study (Arend et al., 2005).

The most commonly investigated biochemical and phenotypic features [colony morphology, ability to grow at temperatures ranging from 24 to 45 °C, niacin accumulation, nitrate reduction, β-glucosidase, Tween 80 hydrolysis, 3 day arylsulfatase, urease, tellurite reduction, 68 °C and semiquantitative catalase, growth rate, pigmentation, growth on MacConkey agar and tolerance to NaCl, thiophene-2-carboxylic hydrazide (TCH), oleate and p-nitrobenzoic acid and to 10 μg thiacetazone ml⁻¹, 1 μg hydroxylamine ml⁻¹ and 500 μg isoniazid ml⁻¹, all three in Middlebrook 7H10 agar] as well as HPLC analysis of cell-wall mycolic acids were tested by the Regional Reference Center for Mycobacteria (Careggi Hospital, Florence, Italy) using standard procedures described previously (Kent & Kubica, 1985; CDC, 1996).

Drug susceptibility testing was performed using the 25-well agar-dilution method (van Klingeren et al., 2007). We included isoniazid, rifampicin, rifabutin, ethambutol, clarithromycin, ciprofloxacin, cycloserine, prothionamide, amikacin, clofazimine and streptomycin in the test panel.

Applying the Hain GenoType MTBC assay, a non-specific reaction was noted with hybridization of the M. tuberculosis complex band only (banding pattern 1, 2, 3). The InnoLipa MYCOBACTERIA v2 line-probe assay yielded a Mycobacterium genus probe reaction, though no species-specific result. The Hain GenoType CM (Common Mycobacteria) kit identified the strain as Mycobacterium tuberculosis complex based on a band 1, 2, 3, 10, 16 pattern. The supplementary AS kit identified the strain as a non-specified Mycobacterium species, with banding pattern 1, 2, 3, 12. Sequencing of the full 23S rRNA gene, the assay's target, established its identity as a non-tuberculous mycobacterium (Table 1⇓). Based on these findings, IS6110 and IS1081 RFLP were performed to confirm the identity of the strain as M. tuberculosis (van Soolingen et al., 1993). No IS6110 or IS1081 element copies could be demonstrated (results not shown).

The sequencing results are listed in Table 1⇑; the 16S rRNA gene, ITS and rpoB and hsp65 gene sequences were all unique. Sequencing of the full 16S rRNA gene identified the bacterium as a Mycobacterium szulgai-like species (Table 1⇑). The 16S rRNA gene sequence was aligned with those of reference strains of the closest related mycobacteria using clustal_x software (Thompson et al., 1997). The resulting topology and tree, inferred by neighbour-joining and visualized using the mega software package (Tamura et al., 2007), were evaluated by bootstrap analyses based on 1000 resamplings (Fig. 1⇓). The tree was rooted with Nocardia abscessus ATCC BAA-279^T as an outgroup. Similar multisequence alignments and trees were created based on the ITS and hsp65 sequences (Supplementary Fig. S1, available in IJSEM Online).

We were able to amplify an RD1 region, including esat-6 (270 bp) and cfp-10 (251 bp) genes. Results of sequence comparisons for both genes are recorded in Table 1⇑.

On Middlebrook 7H10, Ogawa and Stonebrink media, the strain produced small, rough, non-pigmented colonies after 28 days of incubation at 36 °C. After 3 days of exposure to light at ambient temperature, no pigmentation was observed. Growth on Middlebrook 7H10 agar was only observed at 24, 30 and 36 °C. Optimal growth occurred at 36 °C. Colony morphology on Middlebrook 7H10 agar was similar at all temperatures.

Generally, the biochemical profile of the isolate is unique, although it shares characteristics with M. szulgai and Mycobacterium malmoense (Marks et al., 1972). The isolate was negative for niacin accumulation, heat-stable catalase (pH 7, 68 °C), β-glucosidase, tellurite reduction, growth on MacConkey agar and tolerance to p-nitrobenzoic acid, hydroxylamine and oleic acid, but positive for nitrate reduction, semiquantitative catalase, Tween 80 hydrolysis, 3 day arylsulfatase and urease activity and tolerance to TCH, thiacetazone and isoniazid. A comparison with profiles of M. szulgai and M. malmoense is recorded in Table 2⇓.

HPLC revealed a pattern characterized by a single, narrow, late-emerging cluster of peaks. A similar profile is presented by a limited number of mycobacteria (Mycobacterium brumae, M. fallax, M. triviale and M. tuberculosis complex), with none of them fully overlapping the pattern of the strain characterized here (Fig. 2⇓). We used the HPLC mycobacterium library () for this comparison.

Drug susceptibility testing (minimal inhibitory concentrations given in μg ml⁻¹) revealed in vitro resistance to amikacin (10) and p-aminosalicylate (>1), intermediate susceptibility to isoniazid (1) and susceptibility to rifampicin (0.2), rifabutin (<0.2), ethambutol (5), clarithromycin (<2), ciprofloxacin (2), cycloserine (20), prothionamide (<1), clofazimine (<0.5) and streptomycin (5).

Although identification of M. tuberculosis complex is not the main use of the GenoType CM test, false-positive results may lead to incorrect diagnoses of tuberculosis and unwarranted treatment. Aside from the identification as M. tuberculosis by a line-probe assay, which sparked this study, both the molecular and HPLC analyses of the novel strain suggest a phylogenetic relationship with M. tuberculosis; the presence of an RD1 region strengthens this assumption. Its identity as a non-tuberculous mycobacterium is easily proven by sequencing of the 16S rRNA, rpoB, hsp65 or RD1 genes and the absence of IS6110 and IS1081 elements. The novel strain seems closely related to M. szulgai, based on 16S rRNA gene and ITS sequences (Fig. 1⇑ and Supplementary Fig. S1a), although it is related more distantly to M. szulgai based on rpoB and hsp65 sequences (Table 1⇑ and Supplementary Fig. S1b). Its non-chromogenicity, however, is distinct from the photochromogenic M. szulgai.

Retrospectively, this isolate seems clinically relevant, based on its isolation from a normally sterile body site and the symptomatic improvement of the patient after 9 months of tuberculosis therapy (Griffith et al., 2007). This suggests that the novel strain was the causative agent of this patient's disease.

Both the close phylogenetic relationship with species such as M. szulgai and Mycobacterium kansasii, which are among the most pathogenic non-tuberculous mycobacteria (Griffith et al., 2007; van Ingen et al., 2008), and the presence of an RD1 region with esat-6 and cfp-10 genes, which is a virulence factor in M. tuberculosis (Lewis et al., 2003), add to our view that this strain is a human pathogen. On the basis of the data presented here, we describe a novel species to accommodate strain NLA000201958^T, for which we propose the name Mycobacterium riyadhense sp. nov.