Heterogeneity of RNA Polymerase Gene (rpoB) Sequences of Mycobacterium gordonae Clinical Isolates Identified with a DNA Probe Kit and by Conventional Methods

Abstract

In a previous study, we have evaluated genetic identification by using the rpoB gene, which was recently introduced by Kim et al. (J. Clin. Microbiol. 39:2102-2109, 2001; J. Clin. Microbiol. 37:1714-1720, 1999). In this process, we examined the rpoB gene heterogeneity of clinical isolates identified as Mycobacterium gordonae with the conventional biological and biochemical tests and/or a commercially available DNA probe kit. Sequencing of the rpoB gene of 34 clinical isolates revealed that M. gordonae clinical isolates were classified into four major clusters (A, B, C, and D). Interestingly, organisms belonging to cluster D (15 isolates) did not hybridize with M. gordonae ATCC 14470 and specifically possessed urease activity. Therefore, it could be considered to be a novel mycobacterium. The identification of M. gordonae is known to have ambiguous results sometimes. On the other hand, identification of clinical isolates seems to be inconvenient and unsuitable because of a more than 99% 16S rRNA gene similarity value between clusters. These findings suggest that the existence of M. gordonae-like mycobacteria that share similar biochemical and biological characteristics with the 16S rRNA gene of an M. gordonae type strain but less similarity at the genomic DNA level may have complicated the identification of M. gordonae in many laboratories. Furthermore, compared with hsp65 PCR restriction analysis (PRA), rpoB PRA would have the advantage of producing no ambiguous results because of the intracluster homogeneity of the rpoB gene. In this case, rpoB would provide clearer results than hsp65, even if PRA analysis was used. We demonstrated that these M. gordonae-like mycobacteria were easily distinguished by PRA of the rpoB sequence. Additionally, the significance of this M. gordonae-like cluster may help to establish the comparison between the M. gordonae isolates from a clinical specimen and an infectious process in a given patient and to determine the true incidence of infection with this microorganism.

Diseases caused by mycobacteria other than Mycobacterium tuberculosis (MOTT), such as Mycobacterium avium and Mycobacterium kansasii, are becoming a new health problem in the countries or areas where epidemics of tuberculosis are slowing down. The rapid identification of causative mycobacteria is important for treatment and epidemiology because they show distinctive drug susceptibilities and epidemiological distributions. Identification based on the conventional biological and biochemical characteristics is time-consuming and laborious. In the last decade, various genetic typing methods, including 16S rRNA-, hsp65-, DNA gyrase-, and dnaJ-based methods for rapid identification, have been developed and partly applied to the identification of mycobacterial species. However, they do not have enough ability to distinguish among mycobacterial species for identification; there has been no perfect method yet. For example, although 16S rRNA is the most widely accepted method, it could not distinguish among certain species.

Recently, the rpoB gene, which encodes the β subunit of RNA polymerase, was used for the identification of mycobacterial species (4). In addition, PCR-based restriction analysis of rpoB can be used for identification of mycobacterial species (9, 10). In order to evaluate this method, we sequenced 94 mycobacterial strains and a number of clinical isolates. In this process, we discovered the diversity of the rpoB gene among certain mycobacterial species such as Mycobacterium simiae and Mycobacterium gordonae (unpublished data).

M. gordonae, known as Mycobacterium aquae or mycobacterium tap water scotochromogen, ubiquitously exists in the environment and is generally considered to be saprophytic and nonpathogenic to humans. However, it is rarely isolated from humans and causes pulmonary and cutaneous infection, particularly in immunocompromised hosts but even in immunocompetent hosts (3, 8, 19). Recently, M. gordonae was recognized as one of the opportunistic pathogens of human immunodeficiency virus-positive patients (1, 14).

The heterogeneity of M. gordonae has been shown on the genetic level. Kirschner and Bottger reported the microheterogeneity of 5′ 16S ribosomal DNA intraspecies for M. gordonae (11). Telenti et al. demonstrated that M. gordonae isolates were differentiated into 5 subspecies by PCR restriction enzyme pattern analysis of the hsp65 gene (16). The variation of dnaJ was also reported (17). There was no report which determined the difference in virulence or characteristics among each subspecies or cluster. Therefore, the significance of these genetic variations remains unclear.

In the present study, we examined the rpoB sequences of 34 clinical isolates which had been identified as M. gordonae by a biochemical test and by a DNA probe kit (Gen-probe).

MATERIALS AND METHODS

Bacterial strains.

M. gordonae (ATCC 14470) was used for the reference strain. Thirty-four clinical isolates of M. gordonae, including one strain from tap water, were obtained from eight hospitals in Japan and were stored between 1988 and 2001 in our laboratory (Table 1). The strains were identified as M. gordonae by the Gen-probe test (Gen-Probe Inc., San Diego, Calif.) or the conventional test.

TABLE 1.

Biological and biochemical characteristics of 34 M. gordonae clinical isolates and the type strain

Strain	Source of specimena	Hospitalh	Pigmen- tationc	Morphol- ogyd	Growth rate	Growth at temp (°C):			Result for:				rpoB cluster
						28	37	42	Nitrate reduction	Tween hydrolysis	Urease test	Gen probe
ATCC 14470	GLg		Y	S	2 wk	+	+	−	−	+	−	+	A
KK33-02	None	F	O	S	7 days	+	+	−	+	+	−	+	A
KK33-06	None	E	Y	Sr	7 days	+	+	−	+	+	−	+	A
KK33-12	None	D	OY	S	10 days	+	+	−	+ (weak)	+	−	+	A
KK33-40	Sputum	A	Y	S	2 wk	+	+	−	−	+	−	+	A
KK33-49	Sputum	A	OY	S	2 wk	+	+	−	−	+	−	+	A
KK33-50	Sputum	A	Y	S	2 wk	+	+	−	+	+	−	+	A
KK33-51	Sputum	A	Y	S	2 wk	+	+	−		+	−	+	A
NT241	Sputum	I	OY	S	NDf	+	+	−	+	+	−	+	A
KK33-08	None	D	O	Sr	10 days	+	+	−	−	+	+	+	B
KK33-44	Sputum	A	OY	S	10 days	+	+	−	+	+	−	+	B
KK33-54	Sputum	A	O	S	2 wk	+	+	−	+	+	−	+	B
NT193	Sputum	A	O	S	ND	+	+	−	−	+	ND	+	B
KK33-39	Sputum	A	OY	Sr	2 wk	+	+	−	+ (weak)	+	−	+	C
KK33-48	Sputum	A	OY	S	10 days	+	+	−	−	+	−	+	C
KK33-53	Sputum	A	OY	S	10 days	+	+	−	−	+	−	+	C
KK33-05	None	E	OY	S	7 days	+	+	−	+ (weak)	+	+	+	D
KK33-13	None	B	O	S	10 days	+	+	−	+ (weak)	+	±i	+	D
KK33-14	None	B	OY	S	10 days	+	+	−	+ (weak)	+	−	+	D
KK33-15	None	B	O	Sr	10 days	+	+	−	−	+	−	+	D
KK33-41	Sputum	A	OY	S	2 wk	+	+	−	−	+	+	+	D
KK33-42	Sputum	A	O	S	10 days	+	+	−	−	+	−	+	D
KK33-46	TWb	A	O	S	2 wk	+	+	−	−	+	+	+	D
KK33-52	Sputum	A	OY	S	10 days	+	+	−	+	+	−	+	D
NT215	Sputum	H	OY	S	10 days	+	+	−	−	+	+	+	D
NT219	Sputum	A	OY	S	2 wk	+	+	−	−	+	+	+	D
NT224	Sputum	A	O	S	2 wk	+	+	−	−	+	+	+	D
NT328	Sputum	A	OY	S	2 wk	+	+	−	−	+	+	+	D
NT330	Sputum	A	O	Sr	10 days	+	+	−	+ (weak)	+	+	+	D
NT60	Sputum	G	Y	S	ND	+	+	−	+	+	+	ND	D
No. 668	Sputum	J	O	S	2 wk	+	+	−	−	+	+	+	D
KK33-01	None	C	O	S	2 wk	+	+	−	−	+	−	+
KK33-09	None	D	OY	Sr	2 wk	+	+	−	−	+	±	+
KK33-10	None	D	Y	S	2 wk	+	+	−	+ (weak)	+	−	+
NT51	None	A	O	S	ND	+	+	−	−	+	−	+e

^aEach isolate was isolated from a different patient.

^bIsolated from tap water (TW).

^cO, orange; Y, yellow; OY, orange-yellow.

^dS, smooth; Sr, smooth (rough).

^eDetermined with DDH.

^fND, not determined.

^gGL, gastric lavage fluid.

^hLetters correspond to individual hospitals in different prefectures of Japan.

ⁱ±, most experiments produced positive results, but a few yielded negative results.

Preparation of DNA.

A loopful of bacteria grown on 1% Ogawa medium, including 100 ml of salt solution-1% potassium phosphate, 1% sodium glutamate, 6% glycerol, and 0.12% malachite green, added to 200 ml of whole-egg homogenate, was suspended in 500 μl of lysis buffer (10 mM Tris-HCl, 1 mM EDTA, 10% sodium dodecyl sulfate [pH 8.0]) in a 2.0-ml screw-cap microcentrifuge tube containing 0.2 g of grass beads (diameter, 0.1 mm). Bacteria were heat inactivated (95°C, 10 min), and then an equal volume of phenol-chloroform-isoamyl alcohol (50:49:1) was added to them. Bacteria were mechanically disrupted by a mini-bead beater (Biospec products) for 80 s, and then the tube was centrifuged (3,000 × g; 5 min) to separate the aqueous phases. The extraction procedure was repeated more than two times. Chromosomal DNA was purified by ethanol precipitation. The resulting DNA was resuspended with 100 μl of Tris-EDTA buffer (pH 8.0).

DNA amplification of the rpoB, 16S rRNA, and hsp65 genes.

The amplification of the rpoB sequence (342 bp) was performed as follows. Five microliters of template DNA was added to each reaction tube. The composition of the PCR mixture (100 μl) was 1× PCR buffer, 1.5 mM MgCl₂, 20 pmol of primer MF (5′-CGACCACTTCGGCAACCG-3′) and of primer MR (5′-TCGATCGGGCACATCCGG-3′), 200 μM concentrations of each deoxynucleotide triphosphate, and 1.25 U of ampliTaq Gold (Applied Biosystems). The reaction mixture was subjected to 40 cycles of amplification (1 min at 94°C, 1 min at 66°C, and 1 min at 72°C) preceded by a preactivation step (10 min at 94°C) and followed by an extension step (10 min at 72°C) with the GeneAmp PCR system 9600 thermal cycler (Applied Biosystems). To amplify the 16S rRNA gene, primers 263 (5′-TGC ACA CAG GCC ACA AGG GA-3′) and 285 (5′-GAG AGT TTG ATC CTG GCT CAG-3′) were used (11). The amplification was performed for 40 cycles (1 min at 94°C, 1 min at 60°C, and 1 min at 72°C). To amplify the hsp65 gene (439 bp), primers Tb11 (5′-ACC AAC GAT GGT GTG TCC AT-3′) and Tb12 (5′-CTT GTC GAA CCG CAT ACC CT-3′) were used (16). The amplification was performed for 40 cycles (1 min at 94°C, 1 min at 60°C, and 1 min at 72°C).

Sequencing of the rpoB and 16S rRNA genes.

A 377A automatic sequencer and a BigDye Terminator cycle sequencing kit were used for the sequencing of the PCR product. For the sequencing reaction, purified PCR product (10 ng for rpoB or 50 ng for 16S rRNA), 3.2 pmol of the appropriate primer, and 8 μl of BigDye terminator PCR mix were mixed and made up to 20 μl. For the rpoB sequencing, primers P3 (5′-CCA GAA CCA GAT CCG CGT CGG-3′) and MR were used. For the 16S rRNA sequencing, the following set of primers was used: primer 285, primer 264, r325-305 (5′-CCC CAC TGC TGC CTC CCG TAG-3′), r655-635 (5′-GCA TTC CAC CGC TAC ACC AG-3′), p305-325 (5′-CAT CGG GAG GCA GCA GTG GGG-3′), and p635-655 (5′-CTG GTG TAG CGG TGG AAT GC-3′). The sequences were aligned by using GENETYX-WIN (Software Development, Tokyo, Japan). A similarity search of 16S rRNA sequences was also performed by RIDOM2 (http://www/ridom.de). Phylogenetic trees of the rpoB gene were drawn by using the TreeView program (version 1.6.5; Taxonomy and Systematics, Glasgow, Scotland) (http://taxonomy.zoology.gla.ac.uk/rod/treeview.html).

PRA of hsp65 and rpoB.

PCR restriction analysis (PRA) of hsp65 was performed as described by Brunello et al. (2). Briefly, 10 μl of the amplified reaction mixture was transferred to a fresh tube, and 0.5 μl of HaeIII (10 U/μl) or BstEII (12 U/μl), 2 μl of appropriate restriction buffer (10×), and 7.5 μl of distilled water were added to the tube. The mixtures were incubated for 120 min at 37°C for HaeIII and at 60°C for BstEII. After digestion, the samples were electrophoresed on a 3% agarose gel (at 100 V for 50 min), and DNA bands were visualized by ethidium bromide staining and photographed. For PRA of rpoB and hsp65, the amplified products were digested with HaeIII.

DDH.

DNA-DNA hybridization (DDH) was performed by using colorimetric microdilution plate hybridization with a kit that is commercially available in Japan (DDH-mycobacteria; Kyokuto Pharmaceuticals, Tokyo, Japan) (13). This kit is based on analyzing the hybridization between sample DNA and immobilized DNA of 18 standard strains (Mycobacterium bovis BCG, M. kansasii, Mycobacterium marinum, M. simiae, M. gordonae, Mycobacterium szulgai, M. avium, Mycobacterium intracellulare, Mycobacterium gastri, Mycobacterium xenopi, Mycobacterium nonchromogenicum, Mycobacterium terrae, Mycobacterium triviale, Mycobacterium fortuitum, Mycobacterium chelonae, Mycobacterium abscessus, and Mycobacterium peregrinum). The extent of hybridization was determined by colorimetric measurement. The criteria for identification using this kit are as follows: when the ratio of maximum color intensity and the color intensity of the negative-control well (value for Escherichia coli) was higher than 1.9 and the relative relatedness of the well which emitted the second-strongest color intensity was lower than 70% of the maximum color intensity, identity to the standard strain is assured.

RESULTS

rpoB sequence-based phylogenetic analysis of M. gordonae clinical isolates.

Sequencing analysis revealed the heterogeneity of the rpoB sequences in these clinical isolates (Fig. 1 and 2). They were divided into four major clusters: cluster A (8 isolates), which shared a similar rpoB sequence with M. gordonae type strain ATCC 14470, cluster B (4 isolates), cluster C (3 isolates), and cluster D. Four isolates were not clustered. The rpoB sequence similarities within each cluster were more than 99% (99.7% in cluster A, 99.3% in cluster B, 100% in cluster C, and 99.3% in cluster D), and the divergences between each cluster were less than 97.1% (95.1% to 97.1%). Clusters A, B, and D consist of isolates from different hospitals and patients. Although the isolates in cluster C were from the same hospital, they were isolated at different times from different patients.

FIG. 1.

Phylogenetic tree of 34 M. gordonae clinical isolates, the M. gordonae type strain, M. tuberculosis, M. szulgai, and M. asiaticum based on the rpoB gene sequence. M. gordonae clinical isolates were grouped into four clusters.

FIG. 2.

rpoB sequence alignment of M. gordonae ATCC 14470 (cluster A), KK33-08 (cluster B), KK33-53 (cluster C), and isolate No. 668 (cluster D).

16S rRNA sequence analysis of each rpoB cluster.

Next, to examine whether similar 16S rRNA sequences are shared in each rpoB cluster, 16S rRNA (about 440 bp) including hypervariable region A were sequenced and compared on the RIDOM database. All isolates except two strains were identified as M. gordonae on the RIDOM database. Although identical 16S rRNA was generally shared in each cluster, a few exceptions were observed. All the strains classified in cluster A shared a similar rpoB sequence with the type strain; six of eight isolates were identical in 16S rRNA sequence to type strain M. gordonae DSM44160, one strain (KK33-06) was identical to M. gordonae DSM 43212, and NT241 was identical to M. gordonae Borste 9411/99. Three of four isolates in cluster B were identical to M. gordonae Borste 11340/99, one isolate was identical to M. gordonae Borste 9411/99. All isolates in cluster C or D were identical to M. gordonae 10681/99 or M. gordonae Borste 9411/99, respectively. Two isolates, NT51 and KK33-10, had similar 16S rRNA sequences to Mycobacterium asiaticum rather than M. gordonae. KK33-01 and KK33-09 were similar to M. gordonae Borste 10681/99. It is noted that the isolates that have identical 16S rRNA sequences to M. gordonae Borste 9411/99 were shared among three clusters. The divergence among these 16S rRNA sequences was less than 1.14%.

We determined the 16S rRNA sequence (993 bp, corresponding to bases 9 to 1046 of E. coli 16S rRNA) of No. 668, which belonged to cluster D. No. 668 showed 99.5% similarity with the type strain (Fig. 3).

FIG. 3.

16S rRNA sequence alignment of clinical isolate No. 668 and the M. gordonae type strain (ATCC 14470). Both MGO16SSRN and MSGRR16SI are 16S rRNA sequences of the M. gordonae type strain (ATCC 14470) in GenBank. The first nucleotide of isolate No. 668 corresponds to E. coli position 41. The similarity value between isolate No. 668 and ATCC 14470 was 99.5%.

Comparison of classification by rpoB sequence with classification by PRA of hsp65.

Telenti et al. reported that M. gordonae was classified into 5 subspecies by PRA of hsp65 (16). In order to evaluate the classification of these clinical isolates by rpoB sequence, hsp65 PRA of the same clinical isolates was performed. PRA of hsp65 also showed the heterogeneity of M. gordonae clinical isolates. The rpoB cluster A consisted of six isolates of M. gordonae type I and two of M. gordonae type II. Cluster B consisted of two type IV and two type V isolates. Cluster D consisted of eight type III and five type IV isolates and two unique patterns. Although three isolates in cluster C showed identical PRA patterns, they could not be identified at the species level by hsp65. Among isolates which were not clustered, NT51 had an identical pattern to M. asicaticum, KK33-10 had an identical pattern to type V, and KK33-01 and KK33-09 had identical patterns to type IV. The PRA pattern which was identical to M. gordonae type IV was shared among rpoB clusters B and D and KK33-01 and KK33-09. The PRA pattern which was identical to M. gordonae type V was shared with cluster B and KK33-10.

Biological and biochemical characteristics.

The biochemical characteristics of the clinical isolates are summarized in Table 1. No significant difference among the clusters was observed in colony morphology, pigmentation, the rate and temperature of growth, and Tween hydrolysis, and these biochemical characteristics were identical to those of M. gordonae. A remarkable difference among the clusters was observed in urease activity. Ten of 11 urease-positive isolates belonged to cluster D, and 1 isolate was from cluster B. Cluster D consisted of 10 urease-positive, 1 weak-positive, and 4 negative isolates. Variable results were obtained for nitrite reduction, and nitrate reduction did not seem to be a consistent property.

DDH analysis of each cluster.

In bacterial taxonomy, mycobacterial species should be determined based on whole DNA-DNA homology. We examined whether these four clusters are identified as M. gordonae in the DDH assay by using DDH-mycobacteria (Kyokuto). KK33-08 (cluster B) was identified as M. gordonae by this kit. The species of KK33-53 (cluster C) and KK33-01 could not be identified by this kit because the relative similarities with other mycobacterial DNAs were higher than 70%. KK33-46 (cluster D) did not especially hybridize with any mycobacterial DNA.

Differentiation by PRA of rpoB cluster.

Comparison of the rpoB sequences of four clusters and isolates demonstrated that the restriction site for HaeIII is useful for differentiation of these clusters by rpoB PRA (Fig. 4). All PRA patterns were shared intracluster but not shared intercluster. Cluster A produced 117-, 114-, and 79-bp fragments by the digestion of HaeIII. Clusters B, C, and D produced unique fragments (149, 114, and 79 bp; 149, 114, and 50 bp; and 149, 80, and 50 bp, respectively). The others which were not clustered into the above four clusters also produced unique PRA patterns.

FIG. 4.

Differentiation of rpoB clusters by PRA (HaeIII). Amplified rpoB DNAs (342 bp) were digested with HaeIII and electrophoresed on 3% agarose gels. Representatives of each cluster are shown.

DISCUSSION

Thirty of 34 clinical isolates were distinguished in four clusters by rpoB phylogenetic analysis (Fig. 1 and 2). Cluster A consisted of 8 isolates that shared more than 99.7% rpoB sequence similarity to M. gordonae type strain ATCC 14470. Cluster B consisted of 4 isolates that shared more than 96.4% similarity with the type strain. Cluster C consisted of 3 isolates sharing 95.1% similarity with the type strain. Cluster D consisted of 15 isolates that shared more than 94.1% similarity with the type strain. The rpoB sequence similarities within each cluster are more than 99% (99.7% in cluster A, 99.3% in cluster B, 100% in cluster C, and 99.3% in cluster D). In our recent study evaluating rpoB-based mycobacterial species differentiation, 35 of 94 mycobacterial species tested shared more than 97% similarity with other mycobacterial species (unpublished data). Therefore, the divergence among each cluster seems to be comparable to the divergence between two different species. The digestion of rpoB PCR products with HaeIII generates unique PRA patterns, and these clusters were easily and rapidly distinguished by PRA (Fig. 4). Compared with the hsp65 PRA, the rpoB PRA would have the advantage of producing no ambiguous results because of the intracluster homogeneity of the rpoB gene.

Each cluster shared more than 98.9% similarity in the 16S rRNA sequence, and each cluster was identified as M. gordonae on the RIDOM database (Fig. 3). Most clinical isolates (32 isolates) used in this study were positive in the 16S rRNA-based Gen-probe assay (Table 1). The high similarity of their 16S rRNA genes would correlate with their responsiveness in the Gen-probe assay. Although each cluster can be distinguished by 16S rRNA, it seems to be inconvenient and unsuitable because of the more than 99% 16S rRNA gene similarity value between clusters. Moreover, a few contradictions were observed between the classifications by 16S rRNA and rpoB: the 16S rRNA genes identical to Borste 9411/99 and Borste 10681/99 are shared among different rpoB clusters, and the 16S rRNA gene is heterogeneous in rpoB clusters A, B, and D.

It has been reported that M. gordonae can be distinguished into at least 5 subspecies by PRA of hsp65 (16). However, there were some differences and contradictions between the classification by rpoB and hsp65. First of all, hsp65 showed that these rpoB clusters included more than two hsp65 PRA patterns. Second, some hsp65 PRA patterns were shared among rpoB clusters, i.e., M. gordonae type IV was shared with clusters B and D and KK33-01 and KK33-09, and type V was shared with cluster B and KK33-10. Third, the isolates in cluster C, KK33-13 (cluster D), and KK33-14 (cluster D) showed atypical hsp65 PRA patterns. These would be regarded as novel M. gordonae subspecies because the PRA patterns of KK33-39, KK33-48, and KK33-53 were identical and the PRA patterns of KK33-13 and KK33-14 were also identical. In the case of rpoB, the PRA patterns were completely identical within the cluster (data not shown); therefore, rpoB would provide clearer results than hsp65 even if PRA analysis was used. These results suggest that the hsp65 gene would be more variable than rpoB and these two genes would evolve independently at some level.

Each cluster demonstrated the biochemical characteristics for M. gordonae except by the urease test (Table 1). Urease is a nickel-containing enzyme that catalyzes the hydrolysis of urea to form carbon dioxide and ammonia. It is likely that ammonia produced by urease acts as a neutralizer for the acidic environment of the phagolysosome because ammonia is thought to contribute to the survival of M. tuberculosis. It was reported that ammonium chloride prevents phagosome-lysosome fusion in macrophages (5, 6). Moreover, the other enzyme involved in nitrogen metabolism, glutamine synthetase (l-glutamate-ammonia ligase [ADP-forming], EC 6.3.1.2), has been reported as a virulence factor of M. tuberculosis, probably influencing the ammonia level within infected host cells. (7, 15). Therefore, it is likely that the presence of urease allows it to survive in host cells longer than urease-negative bacilli.

In bacterial taxonomy, mycobacterial species should be determined based on DDH. Two rpoB clusters (clusters A and B) were identified as M. gordonae in the DDH assay. Clusters C and D and KK33-01 were not identified in the DDH assay. It is likely that KK33-53 (cluster C) and KK33-01 are relatives or subspecies of M. gordonae because they reacted with M. gordonae, although relative similarities were too high to identify them as M. gordonae. Of interest, KK33-46 (cluster D) hardly reacted with M. gordonae DNA. This suggests that cluster D is different from the other three clusters in DNA similarity. Although further examinations would be required, it is likely that cluster D is a novel mycobacterial species. This is also supported by the fact that only cluster D showed urease activity.

In Japan, DDH is one of the most commonly performed genotyping methods. It is known that the coincidence of DDH with conventional methods is relatively low (about 40%) in the case of M. gordonae or M. szulgai. These findings and our present results suggest that the existence of M. gordonae-like mycobacteria that shared similar biochemical characteristics with the M. gordonae type strain but did not share DDH results would complicate the identification of M. gordonae in many laboratories using DDH. Itoh et al. have reported a case of bronchiectasis caused by abundant M. gordonae, which was scotochromogenic, positive for urease and Tween 80 hydrolysis, and negative for DDH (8). Koizumi et al. have reported a case of pulmonary infection caused by M. gordonae which was negative for DDH regardless of the fact that the isolate’s biochemical characteristics were consistent with those of M. gordonae (12). In these cases, M. gordonae-like mycobacteria seemed to be the pathogens. On the other hand, in the laboratories using a 16S rRNA-based method such as Accuprobe, these M. gordonae-like mycobacteria would have been misidentified as M. gordonae because M. gordonae-like mycobacteria are able to react with a 16S rRNA-based probe.

In this study, 15 of 34 isolates were identified as belonging to cluster D. The reason a particular genotype was dominant remains to be made clear; it may be due to the difference of virulence among each genotype or the rate of each genotype in the environment. There is the possibility that cluster D is the most virulent among these genotypes because of its urease activity. Although the contribution of urease in the pathogenesis of mycobacteria remains to be established, several reports have shown the involvement of ammonia and a nitrogen-metabolizing enzyme in the survival of M. tuberculosis in macrophages (5, 6, 7, 15). Therefore, slight virulence is possibly critical to pathogenesis in the case of a bacillus with very low pathogenicity such as M. gordonae. It is also likely that our results simply reflected the rate of each genotype in the environment. Moreover, the mycobacteria that are difficult to identify in other laboratories tended to be collected in our laboratory because of the position of our laboratory as a reference laboratory in Japan; consequently, the collection of isolates could be biased. Further studies of the virulence and incidence in the environment and clinical isolation of each genotype are required to clarify the clinical significance and characteristics of this genotype.

This study supports the significance of rpoB-based mycobacterial identification. In the past decade, various molecular markers have been developed and utilized for mycobacterial identification. However, their markers seemed to have advantages and disadvantages. For example, 16S rRNA-based identification is one rapid method which is more widely used than other methods but it has some limitations, because M. kansasii could not be distinguished from Mycobacterium gastri (18). These results indicate that the use of a couple of molecular markers in combination would be necessary for precise identification of mycobacterial species.