INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Identification to the species level of the strains belonging to the Mycobacterium terrae complex, i.e., M. terrae sensu stricto, M. nonchromogenicum, and M. triviale, has been regarded as unnecessary, due to the nonpathogenic nature of the complex and difficulties in separating the species. However, reports have indicated that M. nonchromogenicum is a potential pathogen to humans (9, 13, 19). This makes reliable separation among the three species necessary. In this study, a scheme is presented for species identification among strains of the M. terrae complex and also for their separation from other nonpigmented slowly growing mycobacteria which may cause misidentification (21). This scheme is based on analysis of cellular fatty acid methyl esters, fatty alcohols, and mycolic acid cleavage products (MACP).

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains. Clinical isolates consisted of 24 strains which were collected in Finland over a period of 20 years; they were initially identified as members of the M. terrae complex on the basis of biochemical and growth characteristics (Table 1). Seven of the strains were identified as M. terrae sensu stricto, three were identified as M. nonchromogenicum, and four were identified as M. triviale. Ten of the strains were classified only to the group level as members of the M. terrae complex. The conventional identification methods used were described earlier in detail (3). The following reference strains were also included in the study: M. terrae ATCC 15755^T, M. nonchromogenicum ATCC 19530^T, ATCC 19533, and ATCC 35783, M. triviale ATCC 23291, M. avium ATCC 15769, M. intracellulare ATCC 13950^T, M. branderi ATCC 51789^T, M. celatum ATCC 51131^T, M. gastri ATCC 15754^T, M. malmoense ATCC 29571^T, M. shimoidei ATCC 27962^T, and M. tuberculosis ATCC 25177. The strains were stored in Middlebrook 7H9 broth at 70°C in the strain collection of the Department of Clinical Microbiology, Kuopio University Hospital, Kuopio, Finland.

Characterization of strains. In this study, the strains were retested for growth at 37 and 42°C (2), Tween 80 hydrolysis, pyrazinamidase production (10), and urease and nitrate reduction by using commercial discs as recommended by the manufacturer (Rosco, Taastrup, Denmark). When necessary for differentiation, the strains were also tested by the AccuProbe M. avium complex identification test (GenProbe Inc., San Diego, Calif.).

Lipid analyses. For fatty acid and mycolic acid analyses, the strains were grown on Middlebrook 7H11 agar supplemented with Middlebrook OADC enrichment (Difco, Detroit, Mich.) at 37°C for 21 to 30 days. Fatty acid, methyl esters, and alcohols were prepared by acidic methanolysis (6) and analyzed by use of a Perkin-Elmer (Norwalk, Conn.) AutoSystem gas chromatograph equipped with a flame ionization detector and a fused-silica capillary column coated with methylpolysiloxane (NB-30; 25 m by 0.32 mm by 0.25 µm; HNU-Nordion, Helsinki, Finland). The column flow rate of the carrier gas, helium, was 5 lb/in². The injector and detector temperatures were held at 325°C, and the oven temperature was programmed to increase from 125 to 280°C at a rate of 10°C/min. For identification of fatty acid methyl esters and MACP, gas-liquid chromatography-mass spectrometry (GLC-MS) analysis was performed on a Hewlett-Packard (Palo Alto, Calif.) model G1800A GCD system equipped with an electron ionization detector, an HP-5 (30 m by 0.25 mm by 0.25 µm) column, and an HP 7673 automatic sampler. The flow rate of the carrier gas, helium, was approximately 1.0 ml/min in splitless injections. The injector and detector temperatures were 325°C. The oven temperature was programmed to hold at 125°C for 2 min and then to increase by 8°C/min to 280°C. The mass spectra were recorded at an electron energy of 70 eV and a trap current of 300 µA. The ion source temperature was 210°C, and the molecular separator temperature was 155°C. Trimethylsilyl (TMS) derivatives of alcohols were prepared by adding 55 µl of N,O-bis-(trimethylsilyl)trifluoroacetamide:pyridine:trimethylchlorosilane (5:5:1, vol/vol/vol) and heating at 70°C for 15 min. Excess pyridine was removed under a nitrogen stream, and 500 µl of hexane was added. The sample was washed twice with 500 µl of distilled water, dried with anhydrous Na₂SO₄, and analyzed with the GLC-MS as described above. Saturated derivatives of monoenes were prepared as described previously (16) and analyzed as presented above.

Sequence determination. Clinical isolates belonging to the M. terrae complex, except for strain 21236, which was lost to contamination, were also identified by partial sequencing of the 16S rRNA gene by use of an automated ALF DNA sequencer (Pharmacia, Uppsala, Sweden) as described earlier (8).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

On the basis of partial sequencing of the 16S rRNA gene of the clinical isolates, three strains could be assigned to M. triviale, four could be assigned to M. terrae, seven could be assigned to M. nonchromogenicum sensu stricto, and seven could be assigned to the newly described Mycobacterium sp. strain MCRO 6 (15) (Fig. 1). One strain (strain 31958) was found to be a hybrid (Mycobacterium sp. strain 31958) of three genetic entities, M. terrae, M. nonchromogenicum sensu stricto, and Mycobacterium sp. strain MCRO 6. Its sequence differed from their sequences by 7, 9, and 11 nucleotides, respectively. These numbers exceeded the natural point mutation frequencies (range, 0 to 5) detected in the analyzed regions in each of the genetic groups.

View larger version (49K): [in this window] [in a new window]   FIG. 1.   Alignment of 16S rRNA genes within hypervariable region A comprising helix 10 (A) and hypervariable region B comprising helix 18 (B). The numbering is according to the Escherichia coli 16S rRNA gene sequence. Dots represent identical nucleotides. The accession numbers for M. nonchromogenicum, Mycobacterium sp. strain MCRO 6, and M. terrae are X52928, X93032 and X52925, respectively, M. sp., Mycobacterium sp.

The major GLC fatty acids detected in the analyzed strains were those typical of the mycobacteria (22), i.e., hexadecanoate (16:0), octadecenoate (18:1), and 10-methyloctadecanoate (10-Me-18:0; tuberculostearic acid). The main MACP was tetracosanoate (24:0). Secondary alcohols, 2-octadecanol (2-OH-18:0alc) and 2-eicosanol (2-OH-20:0alc), were present in the ATCC strains of M. terrae and M. nonchromogenicum, in all strains identified by conventional tests as M. nonchromogenicum, M. terrae, and members of the M. terrae complex, and also in one strain (strain 10339) initially misidentified as M. triviale (Table 2). In the other M. triviale strains, neither secondary alcohols nor methyl-branched fatty acids other than 10-methyloctadecanoate were detected. Another strain (strain H33165) misidentified as M. nonchromogenicum in the initial testing was verified as a member of the M. avium complex upon retesting. It was negative for Tween 80 hydrolysis and hybridized with the M. avium complex probe. In addition to what has earlier been described for the fatty acid composition of M. nonchromogenicum (12), two earlier unknown peaks were detected in the GLC profiles of all M. nonchromogenicum strains sequenced and also all three ATCC strains (ATCC 19530^T, ATCC 19533, and ATCC 35783). The peaks were also detected in the strains assigned to Mycobacterium sp. strains MCRO 6 and 31958 by gene sequencing. These peaks were located at relative retention times of 1.75 and 1.97 (tetradecanoate = 1.0), as can be seen from the profile presented in Fig. 2. The first of the peaks was identified as 2-(8,15-dimethyl)docosenol (2-OH-8,15-di-Me-22:1alc) (Fig. 3E), and the second was identified as 2-(8,17-dimethyl)tetracosenol (2-OH-8,17-di-Me-24:1alc) (Fig. 3B). In addition, 2-(8,16-dimethyl)tricosenol (2-OH-8,16-di-Me-23:1alc) was often detected in trace amounts. The mass spectra of the components contained an ion at m/z 45, which is a common fragment of secondary alcohols. Further, the most abundant ion in the mass spectra of monounsaturated and saturated TMS-derivatized alcohols was at m/z 117 [CH₃CHOSi(CH₃)₃], in addition to prominent ions at m/z 73 [(CH₃)₃Si] and m/z 75 [(CH₃)₂Si=OH)] (Fig. 3C, D, and F), indicating the presence of typical fragments of secondary alcohols (1). After the catalytic hydrogenation, mass peaks of 2-(8,15-dimethyl)docosenol and 2-(8,17-dimethyl)tetracosenol at m/z 352 (Fig. 3E) and m/z 380 (Fig. 3B) were altered to peaks at m/z 354 and m/z 382 (Fig. 3A), respectively, and their positions in the GLC-MS chromatogram were shifted closer to 22:0 and 24:0, respectively, showing that each of the components contained one double bond. The numbers of carbons in the main carbon chains were determined to be 22 and 24, since 2-(8,15-dimethyl)docosenol and 2-(8,17-dimethyl)tetracosenol eluted between peaks labeled 20:0 and 22:0, and between peaks labeled 22:0 and 24:0, respectively, similarly to OH-18:0alc and 2-OH-20:0alc, which eluted between peaks 16:0 and 18:0 and between peaks 18:0 and 20:0, respectively. After these determinations were made, the calculation of the molecular masses of components revealed the presence of two methyl branches. The positions of methyl groups in 2-(8,15-dimethyl)docosenol and 2-(8,17-dimethyl)tetracosenol were at carbons 8 and 15 and at carbons 8 and 17, due to the presence of mass peaks at m/z 143 and m/z 253 and at m/z 143 and m/z 281, respectively (Fig. 3B and E). Finally, the positions of double bonds in both compounds appeared to be between two methyl branches, since the fragments at m/z 253 and m/z 281 of 2-(8,15-dimethyl)docosenol and 2-(8,17-dimethyl)tetracosenol were altered to m/z 255 and m/z 283, respectively, after the catalytic hydrogenation (Fig. 3A, B, and E). The same shift can be seen in TMS derivatives of alcohols before and after the catalytic hydrogenation; e.g., for TMS-derivatized 2-(8,17-dimethyl)tetracosenol, the position of the methyl branch shifted from m/z 339 to m/z 341 (Fig. 3C and D).

View larger version (25K): [in this window] [in a new window]   FIG. 2.   Typical GLCs of M. nonchromogenicum and M. terrae. Peak designations indicate the number of carbon atoms: number of double bonds. FID, flame ionization detector.

View larger version (50K): [in this window] [in a new window]   FIG. 3.   Mass spectra of 2-(8,17-dimethyl)tetracosanol (A), 2-(8,17-dimethyl)tetracosenol (B), TMS derivative of 2-(8,17-dimethyl)tetracosanol (C), TMS derivative of 2-(8,17-dimethyl)tetracosenol (D), 2-(8,15-dimethyl)docosanol (E), and TMS derivative of 2-(8,15-dimethyl)docosanol (F).

Both M. terrae ATCC 15755 and four clinical strains identified as M. terrae sensu stricto (Fig. 2) had only trace amounts of 2-OH-8,17-di-Me-24:1alc and totally lacked 2-OH-8,15-di-Me-22:1alc (Table 2). On the basis of data presented here and earlier (15), we regard M. nonchromogenicum sensu stricto, and Mycobacterium sp. strains MCRO 6 and 31958 as most closely related and call them members of the M. nonchromogenicum complex.

After reclassification of the clinical strains by gene sequencing, all members of M. terrae sensu stricto were pyrazinamidase production negative, did not grow at 42°C, and could also be separated from the others by GLC fatty acid profile (Table 2). Among strains of the M. nonchromogenicum complex, with the same GLC fatty acid profile, M. nonchromogenicum sensu stricto grew at 42°C and was positive for pyrazinamidase production. In contrast, the other members, Mycobacterium sp. strains MCRO 6 and 31958, were unable to grow at 42°C and were also, with one exception, positive for pyrazinamidase production. All M. triviale strains were pyrazinamidase production negative and did not grow at 42°C, and they also had a distinct fatty acid profile.

Mycolic acid analyses confirmed the presence of alpha- and carboxymycolates and trace amounts of ketomycolates in all strains originally identified as M. terrae sensu stricto and M. nonchromogenicum or grouped as the M. terrae complex. The strain (strain 10339) initially identified as M. triviale but found to contain secondary alcohols by fatty acid analysis and classified as Mycobacterium sp. strain MCRO 6 by gene sequencing contained alpha-, keto-, and carboxymycolates, as determined by mycolic acid analysis. The genuine M. triviale strains (Table 1) contained only alpha-mycolates.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The results of this study exposed the heterogeneity among strains of the M. terrae complex, initially created on the basis of conventional classification schemes. In addition to the accepted species, M. terrae, M. nonchromogenicum, and M. triviale, this complex also seems to comprise variants with less definitely verified status. A recent study (15) indicated that one distinct group, Mycobacterium sp. strain MCRO 6, was closely related to M. nonchromogenicum. The same potentially new species was also detected among our clinical strains. An additional genetic variant (Mycobacterium sp. strain 31958) was discovered in this study. On the basis of fatty acid and alcohol compositions, biochemical characteristics, and 16S rRNA sequencing, both Mycobacterium sp. strains MCRO 6 and 31958 seem to be closer to M. nonchromogenicum than to M. terrae, and we consider them members of the M. nonchromogenicum complex.

Difficulties in species identification within the M. terrae complex have been a recognized problem when only conventional biochemical methods are applied (13, 19, 21). When GLC analysis of cellular fatty acids and alcohols is used as the basis of the identification, secondary alcohols, 2-octadecanol and 2-eicosanol, have been found to be excellent markers in the separation of M. triviale from M. nonchromogenicum and M. terrae (11, 12). In the present work, we indicate that additional secondary alcohols, 2-(8,15-dimethyl)docosenol and 2-(8,17-dimethyl)tetracosenol, could be used as an aid in classification after acidic methanolysis. A combination of these two alcohols was not detectable in M. terrae. In contrast, it was a constant feature in all members of the M. nonchromogenicum complex. This basis of separation agreed well with the results of sequencing of the variable regions A and B of the 16S rRNA gene and also with results of selected biochemical tests, i.e., growth at 42°C and pyrazinamidase production.

The lipid analysis system used in the present study is based on the detection of fatty acids, fatty alcohols, and MACP by GLC. Several of the lipid markers important for the classification of mycobacteria contain carbon chains longer than 20 atoms. These compounds are outside the scope of the commercial GLC Microbial Identification System. Consequently, it seems to lack reliability in the identification of mycobacteria (14). In contrast, a lipid technique with a high separation power is high-performance liquid chromatography (17). M. terrae and M. nonchromogenicum can also be differentiated by high-performance liquid chromatography (13).

The M. terrae complex is an uncommon colonizer of human epithelia. The complex is generally regarded as nonpathogenic. However, M. nonchromogenicum may occasionally cause human disease (9, 13, 18). Hence, both reliable separation of members of this group from other slowly growing species and identification to the species level within the M. terrae complex are important. No commercial probes are available for their separation. The discriminating power of biochemical tests alone is limited, because distinct species may exhibit similar or identical results in these reactions (21). For routine identification at the species level, GLC analysis of fatty acids and MAPC, complemented with detection of growth rate and the Tween 80 test, provides a good basis. Among slowly growing nonpigmented isolates, M. triviale is easily separated by its unique GLC profile from the other nonpigmented potentially pathogenic species, e.g., M. terrae, M. nonchromogenicum, the M. avium complex, M. malmoense, and M. shimoidei (Table 3). The GLC profiles of both M. terrae and the M. nonchromogenicum complex closely resemble that of the M. avium complex. The latter is, however, easily separated by a negative Tween 80 test result or by use of commercial genetic probes. Finally, the two novel markers, 2-OH-8,15-di-Me-22:1alc and 2-OH-8,17-di-Me-24:1alc, offer separation of M. terrae from the M. nonchromogenicum complex. So far, we have not detected this alcohol combination in other nonpigmented slowly growing mycobacteria but have detected it in M. nonchromogenicum sensu stricto and in its variants, Mycobacterium sp. strains MCRO 6 and 31958.