Evaluation of the MicroSeq System for Identification of Mycobacteria by 16S Ribosomal DNA Sequencing and Its Integration into a Routine Clinical Mycobacteriology Laboratory

Abstract

An evaluation of the MicroSeq 500 microbial identification system by nucleic acid sequencing and the Mayo Clinic experience with its integration into a routine clinical laboratory setting are described. Evaluation of the MicroSeq 500 microbial identification system was accomplished with 59 American Type Culture Collection (ATCC) strains and 328 clinical isolates of mycobacteria identified by conventional and 16S ribosomal DNA sequencing by using the MicroSeq 500 microbial identification system. Nucleic acid sequencing identified 58 of 59 (98.3%) ATCC strains to the species level or to the correct group or complex level. The identification results for 219 of 243 clinical isolates (90.1%) with a distance score of <1% were concordant with the identifications made by phenotypic methods. The remaining 85 isolates had distance scores of >1%; 35 (41.1%) were identified to the appropriate species level or group or complex level; 13 (15.3%) were identified to the species level. All 85 isolates were determined to be mycobacterial species, either novel species or species that exhibited significant genotypic divergence from an organism in the database with the closest match. Integration of nucleic acid sequencing into the routine mycobacteriology laboratory and use of the MicroSeq 500 microbial identification system and Mayo Clinic databases containing additional genotypes of common species and added species significantly reduced the number of organisms that could not be identified by phenotypic methods. The turnaround time was shortened to 24 h, and results were reported much earlier. A limited number of species could not be differentiated from one another by 16S ribosomal DNA sequencing; however, the method provides for the identification of unusual species and more accurate identifications and offers the promise of being the most accurate method available.

During the past two decades, clinical mycobacteriology has enjoyed several important advances that have improved the means for the detection and identification of mycobacteria from clinical specimens. Ultimately, all have improved patient care by shortening turnaround times in the laboratory, and the advances continue. For many years traditional phenotypic identification methods have been used to identify mycobacteria. All are slow, many provide equivocal results, and identifications are not always timely and accurate. High-performance liquid chromatography (HPLC) was used by state laboratories and some reference laboratories. This provided for the rapid identification of an expanded list of species of mycobacteria. Gas-liquid chromatography was also developed and proved to be a useful identification method; however, the number of species that could be identified by gas-liquid chromatography was less compared to the number that could be identified by HPLC. The development of nucleic acid probes for confirmation of the identity of the isolate had a major impact on the work flow in many laboratories. Probes were developed for the identification of members of the Mycobacterium tuberculosis complex, M. avium, M. intracellulare, members of the M. avium-M. intracellulare complex, M. gordonae, and M. kansasii. These allowed the clinical laboratory to accurately identify the aforementioned species of mycobacteria within 1 h of detection and significantly shortened the time to the reporting of results.

Molecular assays for the detection of the nucleic acids of the M. tuberculosis complex directly in clinical specimens were developed. At present, several are commercially available and exhibit good sensitivities and specificities. This has allowed the laboratory to detect patients with tuberculosis in a much shorter time than ever before, even those with acid-fast smear-negative specimens.

Nucleic acid sequencing has been available for several years and has been used for the identification of mycobacteria (1, 3, 5, 9, 17). Interest in using sequencing in the routine clinical laboratory has increased to a point where instrumentation and sequence databases have been developed and are marketed commercially. Many of the newly described species have been identified with the aid of 16S ribosomal DNA (rDNA) sequencing (6, 22), and it seems reasonable for large clinical and reference laboratories to consider this technology.

The MicroSeq 500 16S rDNA microbial identification system (Applied Biosystems, Foster City, Calif.) contains sequence libraries for bacteria, fungi, and mycobacteria. The most recent version of the mycobacterial sequence library (version 1.4.2) contains at least one sequence each for 82 of the 93 accepted species of mycobacteria listed by the Bacterial Nomenclature Up-to-Date website (http://www.dsmz.de/bactnom/bactname.htm), which is sponsored by the Deutsche Sammlung von Mikroorganismen und Zeitkulturen GmbH. The additional species included in version 1.4.2 were M. cookii, M. goodii, M. heidelbergense, M. hodleri, M. madagascariense, M. mageritense, M. murale, M. porcinum, M. septicum, and M. tusciae.

This report describes both our evaluation of the commercially available MicroSeq 500 microbial identification system for the identification of clinically important mycobacteria and our experience with its integration into a clinical laboratory.

MATERIALS AND METHODS

Isolates for evaluation of MicroSeq 500 16S rDNA microbial identification system.

Fifty-nine isolates of mycobacteria were purchased from the American Type Culture Collection (ATCC) and included isolates of the species M. abscessus, M. agri, M. aichiense, M. alvei, M. austroafricanum, M. avium, M. branderi, M. chitae, M. chubuense, M. duvalii, M. flavescens, M. fortuitum, M. gadium, M. gilvum, M. goodii, M. gordonae, M. immunogenum, M. interjectum, M. intracellulare, M. komossense, M. lentiflavum, M. mageritense, M. malmoense, M. moriokaense, M. neoaurum, M. nonchromogenicum, M. parafortuitum, M. peregrinum, M. phlei, M. porcinum, M. porferiae, M. pulveris, M. rhodesiae, M. senegalense, M. shimoidei, M. simiae, M. smegmatis, M. szulgai, M. thermoresistible, M. tokaiense, M. triviale, M. tuberculosis, M. bovis (BCG), M. vaccae, M. wolinskyi, and M. xenopi.

Three hundred twenty-eight clinical isolates of mycobacteria were used in the study; many of the isolates were considered difficult to identify, and some could not be identified to the species level by conventional methods. All were grown on Middlebrook 7H10 agar (Becton Dickinson, Sparks, Md.) and checked for purity prior to nucleic acid sequencing. Cultures were grown on sheep blood agar to ensure purity.

Identification of isolates by phenotypic methods.

All nonchromogenic mycobacterial isolates were subjected to nucleic acid probe testing for confirmation of the identities of members of the M. tuberculosis complex and members of the M. avium-M. intracellulare complex. Hybridization assays were performed according to the instructions of the manufacturer (AccuProbe; Gen-Probe, Inc., San Diego, Calif.). Probe testing was performed for M. kansasii and M. gordonae when pigmented colonies grew on solid media.

All isolates that could not be identified with nucleic acid probes were grown on Middlebrook 7H10 agar and examined for growth rate, macroscopic and microscopic morphological features, and pigmentation. Identification to the species level was made by gas-liquid chromatography of short-chain cell wall fatty acids (Microbial Identification system; Microbial ID, Newark, Del.) and a battery of biochemical tests including Tween 80 hydrolysis, nitrate reduction, aryl sulfatase and urease production, tellurite reduction, salt tolerance, and semiquantitative catalase production (10, 15, 24). The selection of appropriate biochemical tests was made on an individual basis after the growth rate, pigment production, and colonial morphological features were considered. Isolates that could not be definitively identified were give the designation “most closely resembles” a particular species.

16S rDNA sequencing.

The MicroSeq 500 16S rDNA bacterial identification kit, composed of PCR and cycle sequencing modules, analysis software, and a 500-bp 16S rDNA library of bacterial nucleic acid sequences, was used for identification.

DNA was extracted from mycobacterial cells by placing one large colony or several small colonies into a tube containing 0.5 ml of alkaline wash solution composed of 0.05 M sodium citrate and 0.5 M NaOH. The tubes were vortexed, allowed to stand for 5 min, and centrifuged at 20,800 × g for 5 min; and the cell pellet was washed in 0.5 ml of 0.5 M Tris-HCl (pH 8.0) and resuspended in 100 μl of RNase-free water. The tubes were placed at 95°C for 15 min.

Two microliters of genomic DNA was amplified in 50 μl of a master mixture consisting of 0.4 μM TGGAGAGTTTGATCCTGGCTCAG and TACCGCGGCTGCTGGCAC primers, 200 mM deoxynucleoside triphosphates, PCR buffer, 0.3 U of AmpliTaq DNA polymerase, and 10% glycerol. PCR conditions were 95°C for 5 min; 40 cycles each of 95°C for 30 s, 55°C for 30 s, and 72°C for 45 s; and a final step at 72°C for 10 min.

Ten microliters of amplicon was loaded onto a 2% E-Gel, subjected to electrophoresis, and viewed according to the instructions of the manufacturer (Invitrogen Life Technologies, Carlsbad, Calif.) to determine whether PCR products were present.

Purification of the PCR product to remove excess primers and nucleotides was performed with shrimp alkaline phosphatase (2.0 U/μl) and exonuclease I (10.0 U/μl) (U.S. Biochemical Corporation, Cleveland, Ohio). Briefly, 1.0 μl of each enzyme was added to 5.0 μl of amplicon. The enzymes were activated for 15 min at 37°C, followed by inactivation at 80°C for 15 min.

Cycle sequencing was performed with the sequencing module, and after removal of excess dyes, the labeled amplicon was placed on an ABI 3100 16 capillary genetic analyzer (Applied Biosystems).

Sequence data analysis.

All sequence files for the samples were assembled, edited, and compared to those in the MicroSeq 500 bacterial database (version 1.4.2, February 2002), which contained 1,434 entries, including entries for 82 species of mycobacteria. A distance score of 0.00% to less than 1.00% was used as the criterion for species identity. The organism choice that gave the closest match was considered the most likely correct identification. Organisms with distance scores greater than 1.00% were considered unique isolates that were most closely related to organisms in the database with the closest match or novel species.

Resolution of discrepant results for isolates.

In 34 instances the phenotypic and sequencing identifications disagreed, and cultures were sent to the Centers for Disease Control and Prevention, where the organisms were identified by HPLC. If the results between the identifications obtained by HPLC and sequencing were concordant, the identification was considered correct. The HPLC and sequencing identifications disagreed in 10 instances, and the cultures were referred to the University of Texas Health Sciences Center at Tyler, where they were subjected to PCR restriction analysis (PRA) (2, 19, 20) of the 65-kDa heat shock protein region. When the 16S rDNA sequencing and PRA results were concordant, the identification was accepted as correct.

Integration of nucleic acid sequencing into the clinical mycobacteriology laboratory.

After the evaluation of the MicroSeq 500 database was completed, the mycobacteriology laboratory routinely used sequencing to identify all isolates that could not be identified with nucleic acid probes (during 2000 and 2001). A significant number could not be identified by phenotypic methods or DNA sequencing. These isolates were sent to the Centers for Disease Control and Prevention and were reported as unclassified mycobacteria or variants of a common species or of a group or complex.

A Mayo Clinic database of 223 mycobacterial sequences was constructed; some represented species not included in the MicroSeq 500 database, some represented unclassified species, and others represented species identified by the Centers for Disease Control and Prevention and/or The University of Texas Health Center at Tyler. The sequences that were determined to be different genotypes of common organisms were also included in the database. Identifications made in the Mayo Clinic experience part of the evaluation were based on sequences found in the MicroSeq 500 database and/or the Mayo Clinic database, and these sequences were not included in the verification of the MicroSeq 500 database.

Culture identifications made during 1998 and 1999, when sequencing was not routinely performed, were compared to those made during 2000 and 2001, when sequencing was used as a routine identification method for the identification of mycobacteria. A cost analysis of phenotypic methods and nucleic acid sequencing was performed.

Nucleotide sequence accession numbers.

The nucleotide sequences of the mycobacteria in the Mayo Clinic database described in this report were deposited in the GenBank sequence database. The accession numbers are AY215201 to AY215380.

RESULTS

Comparison of phenotypic methods to nucleic acid sequencing using the MicroSeq 500 database. (i) ATCC strains.

Forty-one of 59 (69.5%) ATCC strains of mycobacteria were identified to the species level, and 28.8% were placed into the correct group or complex; however, when our selection of organisms is considered, these results were not unexpected.

Four strains belonging to the M. fortuitum complex shared the same sequence and could not be differentiated from the species within the complex. M. immunogenum, a unique species closely related to M. chelonae-M. abscessus, had a distance score of 1.0%. It was not included in the MicroSeq 500 database; however, it was placed into the correct complex by sequencing. M. lentiflavum (which was also not included in the MicroSeq 500 database) was incorrectly placed into the M. simiae-M. genavense group. The sequences of two isolates of M. triviale had distance scores of 2.19% compared to the sequence of the type strain. This indicates that the two strains may be related to the type strain but more likely represent unique strains that are not in the database.

It should be kept in mind that many of the ATCC isolates likely have not been subjected to nucleic acid sequencing, and the identification provided by phenotypic methods might not be the same as that provided by sequencing.

(ii) Clinical isolates.

Of the 328 mycobacteria sequenced during the evaluation of the MicroSeq 500 database, 243 (74.1%) had distance scores of less than 1% (Table 1). Two hundred three isolates were identified to the species level or the group or complex level by phenotypic methods; 138 (56.8%) were identified to the species level. In this study, results were considered concordant if the sequencing and the phenotypic methods both identified the organism to the correct species level or complex or group level; 40 isolates were identified as most closely resembling a particular species. Two hundred nineteen (90.1%) isolates had concordant phenotypic identifications to the appropriate species level or group or complex level.

TABLE 1.

Comparison of mycobacterial identification by conventional methods and nucleic acid sequencing

Organism(s) identified by phenotypic methods	No. of isolates	Result obtained when organisms were identified to species level or group or complex level by using <1% match score and MicroSeq 500 database
		% Concordant resultsa	Distance score range (%)	Sequencing-based identification (no. of isolates)
M. abscessusb	1	100	0.0	M. chelonae-M. abscessus (1)
M. avium-M. intracellulare complex	6	66.7	0-0.97	M. avium (1), M. malmoense (1), M. paratuberculosis (3), M. simiae-M. genavense (1)
M. avium-M. intracellulare- M. scrofulaceum complex	11	54.5	0-0.79	M. scrofulaceum (6), M. gilvum (1), M. intermedium (1), M. kansasii-M. gastri (1), M. novocastrense (2)
M. alvei	1	100	0.2	M. alvei (1)
M. asiaticum	11	100	0-0.78	M. asiaticum (11)
M. aurum	7	85.7	0-0.2	M. neoaurum (6), M. obuense (1)
M. celatum	3	66.6	0.19	M. celatum (2), M. simiae-M. genavense (1)
M. chelonae	7	85.7	0-0.60	M. chelonae-M. abscessus (6), M. porcinum (1)
M. chelonae-M. abscessus	18	100	0-0.6	M. chelonae-M. abscessus (18)
M. flavescens	1	100	0.38	M. novocastrense (1)
M. fortuitum-M. peregrinum	8	100	0-0.2	M. farcinogenes (3), M. septicum-M. peregrinum (3), M. porcinum (2)
M. fortuitum	10	80.0	0-0.6	M. fortuitum-M. fortuitum subsp., M. acetamidolyticum (6), M. peregrinum (2), M. mageritense (1), M. alvei (1)
M. gastri	1	100	0	M. kansasii-M. gastri (1)
M. gordonae	5	100	0-0.78	M. gordonae (5)
M. haemophilum	8	100	0-0.58	M. haemophilum (8)
M. hassiacum	1	100	0	M. hassicum (1)
M. interjectum	7	57.1	0-0.80	M. interjectum (4), M. simiae-M. genavense (3)
M. intermedium	1	100	0.20	M. intermedium (1)
M. kansasii	6	100	0-0.39	M. kansasii-M. gastri (6)
M. mageritense	5	100	0-0.20	M. mageritense (5)
M. malmoense	5	80	0	M. malmoense (4) M. novocastrense (1)
M. marinum	7	100	0	M. marinum (7)
M. mucogenicum	18	100	0-0.97	M. mucogenicum (18)
M. nonchromogenicum	9	100	0.97	M. nonchromogenicum (9)
M. obuense	1	100	0	M. obuense (1)
M. peregrinum	2	100	0	M. peregrinum-M. septicum (1), M. farcinogenes (1)
M. porcinum	1	100	0	M. porcinum (1)
M. scrofulaceum	11	42.8	0.30-0.59	M. scrofulaceum (4), M. pulveris (1), M. neoaurum (1), M. gilvum (1), M. novocastrense (4)
M. simiae	15	100	0.0-0.80	M. simiae-M. genavense (14)
M. shimodei	2	100	0	M. shimodei (2)
M. smegmatis	10	100	0-0.20	M. smegmatis (3), M. mageritense (1), M. goodii (6)
M. szulgai	9	100	0-0.19	M. szulgai (9)
M. tuberculosis	8	100	0-0.58	M. tuberculosis (8)
M. terrae	8	100	0-0.97	M. terrae (8)
M. thermoresistibile	3	100	0	M. thermoresistibile (3)
M. triplex	6	100	0-0.20	M. triplex (6)
M. wolinskyi	1	100	0.20	M. wolinskyi (1)
M. xenopi	9	100	0.1-0.91	M. xenopi (9)

^aSequencing identification concordant with identification to species level or complex or group level.

^bKnown isolate confirmed by PRA.

As shown in Table 1, the isolates that were consistently identified to the correct species level included M. alvei, M. asiaticum, M. gordonae, M. haemophilum, M. hassiacum, M. intermedium, M. mageritense, M. marinum, M. mucogenicum, M. nonchromogenicum, M. obuense, M. shimoidei, M. szulgai, M. terrae, M. thermorestible, M. triplex, M. triviale, M. wolinskyi, and M. xenopi.

The remaining 85 mycobacteria (Table 2) had distance scores greater than 1.00%. Forty-four (51.8%) of these had distance scores between 1.00 and 1.50% (data not shown), while 41 had distance scores ranging from 1.56 to 4.60%. Thirty-five (41.1%) isolates were identified to the appropriate species level or group or complex level, and 26 (15.3%) isolates were identified to the appropriate species level. All 85 sequences were determined to be those of mycobacterial species, either unique species or, perhaps in some instances, those that exhibited significant divergence from an organism in the database with the closest match.

TABLE 2.

Comparison of mycobacterial identification by conventional methods and nucleic acid sequencing

Organism(s) identified by phenotypic methods	No. of isolates	Result obtained when organisms were identified to species level or group or complex level by using ≥1% match score and MicroSeq 500 database
		% Concordant resultsa	Distance score (% [avg range])	Sequence-based identification (no. of isolates)
M. avium-M. intracellulare complex	5	20	1.0-1.79	M. avium-M. intracellulare-M. paratuberculosis (1), M. gastri-M. kansasii (1), M. simiae-M. genavense (3)
M. avium-M. intracellulare- M. scrofulaceum complex	10	30	1.2-1.96	M. scrofulaceum (2), M. simiae-M. genavense (3), M. intracellulare-M. szulgai-M. conspicuum (1), M. celatum (1), M. malmoense (2), M. moriokaense (1)
M. aurum	8	50	1.2-2.59	M. neoaurum-M. diernhoferi (4), M. mucogenicum-M. farcinogenes (1), M. tokaiense-M. porcinum (1), M. flavescens-M. austroafricanum (1), M. madagascariense (1)
M. celatum	3	100	1.1-4.6	M. celatum (3)
M. chelonae	1	100	1.0	M. chelonae-M. abscessus (1)
M. flavescens	3	0	1.19-2.65	M. moriokaense (2), M. tokaiense-M. murale (1)
M. fortuitum	1	0	1.79	M. simiae/M. genavense (1)
M. gordonae	4	100	1.17-2.54	M. gordonae (4)
M. interjectum	2	0	1.0	M. simiae-M. genavense (2)
M. kansasii	3	100	1.56-3.50	M. kansasii-M. gastri (3)
M. lentiflavum	9	0	1.0-1.29	M. simiae-M. genavense (9)
M. malmoense	2	0	1.0-1.36	M. terrae (1), M. simiae-M. genavense (1)
M. mucogenicum	3	100	1.0-1.40	M. mucogenicum (3)
M. nonchromogenicum	12	75	1.16-2.91	M. nonchromogenicum (9), M. hiberniae (2), M. neoaurum-M. diernhoferi (1)
M. scrofulaceum	4	0	1.39-2.18	M. pulveris (1), M. goodii (1), M. thermoresistibile (1), M. simiae- M. genavense (1)
M. smegmatis	2	0	2.20-2.59	M. madagascariense (1), M. chitae (1)
M. terrae	10	40	1.16-2.37	M. terrae (4), M. nonchromogenicum (5), M. tokaiense-M. murale (1)
M. triplex	1	0	2.79	M. simiae-M. genavense (1)
M. triviale	1	100	2.40	M. triviale (1)
M. xenopi	1	0	4.06	M. celatum (1)

^aSequence identification concordant with identification to species level or complex or group level.

The phenotypic characteristics used in our laboratory were unable to differentiate the species belonging to groups that contained M. abscessus and M. chelonae; M. avium, M. intracellulare, and M. scrofulaceum; M. aurum, M. neoaurum, and M. diernhoferi; M. farcinogenes, M. fortuitum, M. peregrinum, M. porcinum, and M. septicum; M. smegmatis and M. goodii; M. interjectum, M. simiae, and M. genavense; and members of the M. tuberculosis complex. Of 19 selected isolates of rapidly growing mycobacteria that were further identified by PRA, all were previously identified as M. chelonae-M. abscessus, 18 were previously identified as M. abscessus, and 1 was previously identified as M. chelonae.

Those isolates that gave distance scores of greater than 1.00% may represent (i) organisms that were misidentified by phenotypic methods, (ii) species not present in the database, (iii) species exhibiting exceptional divergence from the organism in the database giving the closest match, or (iv) a novel species. When the organism in the database with the closest match was compared to the phenotypic identification, 41.2% of the isolates in this group had concordant results. The organisms in this group of 85 isolates were not identified further; however, the sequences and organisms will be kept for further evaluation. Our results are not unexpected. Tortoli et al. (21) reported that of 72 (6.9%) of 1,035 isolates screened in a reference laboratory, none hybridized with any of the four AccuProbes, nor did they belong to any of the officially recognized mycobacterial species.

Integration of nucleic acid sequencing into the clinical mycobacteriology laboratory.

During 1998 and 1999, 4,047 mycobacteria were identified in the clinical mycobacteriology laboratory at Mayo Clinic; 3,183 (78.9%) organisms were identified by use of nucleic acid probes; the remaining 864 isolates were identified by previously described phenotypic methods. Two hundred fifteen (5.3%) organisms could not be identified.

During 2000 and 2001, 4,217 mycobacteria were identified; nucleic acid probes identified 3,398 (80.6%), sequencing identified 773 (18.3%), and 78 (1.8%) could not be identified. Overall, nucleic acid sequencing was used to identify 16.5% of the total isolates, and it dramatically reduced the number of isolates that could not be identified.

The turnaround time required for sequencing was reduced from 2 days to 24 h by beginning the DNA extraction step at 8:00 a.m., performing PCR at 9:00 a.m., conducting amplicon purification at 12:00 p.m., and performing cycle sequencing at 1:00 p.m. An additional person was used on the evening shift to expedite the sequencing process. Samples were placed on the genetic analyzer at 6:00 p.m., and sequencing data were analyzed and given to the laboratory at 8:00 a.m. the following morning; the results were reported immediately.

After consideration of the limitations, the accuracies of the identifications, and the shortened turnaround time, the clinical mycobacteriology laboratory made the decision to use nucleic acid sequencing as a routine tool to replace phenotypic methods. Sequencing was limited to isolates that did not hybridize with the four available AccuProbes.

Throughout 2000 and 2001 the laboratory identified 773 isolates of mycobacteria using nucleic acid sequencing. Table 3 shows that 84% were more commonly encountered species and the unusual species accounted for only 4.8% of those identified. Data from 2000 and 2001 were compared to those from 1998 and 1999, when sequencing was not done. Sequencing reduced the number of organisms that could not be identified by approximately one-third. These organisms represented novel species; all were sent to the Centers for Disease Control and Prevention and were identified as unclassified mycobacteria. Most of the organisms that could not be identified to the species level were rapidly growing mycobacteria; the taxonomic classification is not yet completed.

TABLE 3.

Mycobacteria identified at Mayo Clinic by nucleic acid sequencing over 2-year period by using an in-house database and the MicroSeq 500 database

Organism	No. of isolates identified
M. alvei	1
M. asiaticum	5
M. aurum-M. neoaurum	7
M. aurum-M. neoaurum (MCRa)	2
M. avium	16
M. avium-M. intracellulare	3
M. avium-M. intracellulare-M. scrofulaceum group	7
M. avium-M. intracellulare-M. scrofulaceum group (MCR)	1
M. celatum	1
M. chelonae-M. abscessus	244
M. chelonae-M. abscessus (MCR)	18
M. flavescens	2
M. fortuitum	101
M. fortuitum-M. peregrinum-M. farcinogenes	11
M. fortuitum-M. peregrinum-M. porcinum	4
M. fortuitum-M. peregrinum-M. septicum	38
M. gadium	1
M. goodii	6
M. gordonae	50
M. gordonae (MCR)	3
M. haemophilum	4
M. hassicum	1
M. immunogenum	6
M. interjectum	1
M. intermedium	4
M. intracellulare	10
M. kansasii	15
M. lentiflavum	11
M. mageritense	3
M. malmoense	5
M. malmoense (MCR)	1
M. marinum	35
M. mucogenicum	34
M. mucogenicum (MCR)	2
M. nonchromogenicum	3
M. nonchromogenicum (MCR)	1
M. obuense	1
M. peregrinum	6
M. pulveris	1
M. scrofulaceum	4
M. scrofulaceum (MCR)	2
M. simiae	8
M. simiae (MCR)	1
M. smegmatis	1
M. species	46
M. szulgai	8
M. terrae	5
M. terrae (MCR)	1
M. tuberculosis complex	6
M. wolinskyi	1
M. xenopi	25

^aMCR, most closely resembles.

DISCUSSION

The identification of mycobacteria can be a complicated, expensive, and difficult process; many laboratories are now referring uncommon organisms to laboratories that have the capability of using additional technology. Nucleic acid probes have offered laboratories the ability to rapidly and accurately identify four of the most common mycobacterial species, and they have rarely misidentified an organism (3).

A more important issue is the inaccuracy of phenotypic methods in providing a reliable and timely identification of the other mycobacteria (14). Nucleic acid sequencing of 16S rDNA has been investigated as a definitive method for the identification of many microorganisms, including mycobacteria (3, 5, 8, 11, 12, 16-18, 23), and its use is becoming more extensive. Most of the studies used a small number of organisms for evaluation or included only common species and/or the type species from culture collections.

We sought to determine whether 16S rDNA sequencing with a commercially available system would be the ultimate method for the identification of all mycobacteria. Our concern related to those organisms that are phenotypically difficult or impossible to identify in the routine clinical mycobacteriology laboratory. We used a mixture of common species and other clinical isolates that presented diagnostic challenges to determine how much the MicroSeq 500 database would help our laboratory. Furthermore, we wanted to know if the turnaround time for these identifications could be reduced to a minimum so that results would be clinically useful when they were reported to the clinical staff.

Overall, nucleic acid sequencing identified 95.1% of 328 clinical isolates to the appropriate species level or complex or group level, and of these, 62.5% were correctly identified to the species level. The inability of the system to provide a species-level identification for all isolates was attributed to the large number of organisms that belong to a complex or a group whose members cannot be differentiated from one another. In a few instances, the MicroSeq 500 database did not contain a sequence for a particular species included in this evaluation and an erroneous identification was made. Also, the database did not provide for genotypic variation since a single sequence was used for each species.

Version 1.4.2 of the MicroSeq 500 analysis software contained one sequence each for all mycobacteria recognized at present except M. botniense, M. doricum, M. elephantis, M. frederiksbergense, M. heckeshornense, M. immunogenum, M. kubicae, M. lentiflavum, M. leprae, M. lepraemurium, and M. ulcerans.

As noted by others (3, 12, 23), a limitation of 16S rDNA sequencing is that it cannot differentiate between M. avium and M. paratuberculosis; M. chelonae and M. abscessus; M. flavescens and M. novocastrense; M. fortuitum and M. fortuitum subsp. acetamidolyticum; M. gastri and M. kansasii; M. peregrinum and M. septicum; M. tuberculosis and M. bovis, M. bovis BCG, M. africanum, and M. microti; and M. murale and M. tokaiense. This was well substantiated by our evaluation.

The algorithm used in the Mayo Clinic clinical mycobacteriology laboratory included AccuProbe testing with all primary positive MGIT tubes with probes for members of the M. tuberculosis complex and the M. avium-M. intracellulare complex. If cultures exhibited pigmentation, probes for M. gordonae and M. kansasii were used. Probes were selectively used for identification of the isolates recovered on solid medium, and their selection was based on morphological features and pigment production. Nucleic acid sequencing was limited to isolates that did not hybridize with the four available AccuProbes.

The advantages of using nucleic acid sequencing as a tool for microbial identification outweigh its limitations, in general. A commercially available system such as the MicroSeq 500 microbial identification system has the distinct advantage of being able to provide identifications for aerobic and anaerobic bacteria, mycobacteria, and fungi due to the extensive database of sequences. Thus, the tool is versatile and can be used in several areas of the clinical microbiology laboratory. Specifically for the mycobacteriology laboratory, it offers a rapid turnaround time for the identification organisms that do not hybridize with the AccuProbes available at present. This turnaround time can be minimized if an evening shift is used to complete the sequencing process. It offers the laboratory the ability to identify mycobacteria with a better accuracy than those that phenotypic methods provide; more importantly, it provides more objective data on which identifications can be based. Sequencing gives the laboratory the ability to identify many of the more recently described species that are difficult or impossible to identify on the basis of phenotypic characteristics. Regardless of the sequencing hardware used, 16S rDNA sequences can be compared to entries in World Wide Web-based databases such as those of the Ribosomal Differentiation of Medical Microorganisms (RIDOM) (http://www.ridom.com/) (7) site, the Ribosomal Database Project (http://rdp.cme.msu.edu/html/), European Molecular Biology Laboratory (http://www.ebi.ac.uk/embl/), and GenBank (http://www.ncbi.nlm.nih.gov/). The RIDOM site provides peer-reviewed entries, along with extensive phenotypic and sequence data, and it is recommended.

However, there are limitations that must be understood by a laboratory when it uses nucleic acid sequencing as a tool for the identification of mycobacteria. The major perception or, perhaps, reality for many laboratories is the excessive cost of sequencing compared to those of phenotypic methods. Phenotypic identification of mycobacteria often requires subculture of an organism onto different media, incubation of cultures at different temperatures, and the use of HPLC or a battery of biochemical tests. All of these procedures are time-consuming, and in many instances the results are not available for several weeks. Both sequencing and phenotypic methods are costly; however, sequencing requires only a short time before results are available to the physician for use while they are still clinically relevant; this can reduce patient hospital charges due to the rapid initiation of therapy. Our analysis showed that the costs of performing nucleic acid sequencing and phenotypic methods differed by only $1.20. Diggle and Clarke (4) state that after initial start-up costs, nucleotide sequencing costs are low. Benefits are high throughput, low consumable costs, and good interlaboratory and intralaboratory reproducibilities.

As discussed earlier, some organisms may not be differentiated from each other by 16S rDNA sequencing. M. chelonae and M. abscessus represent the major pathogens with which this occurs. The heat shock protein 65 target can be used to distinguish between these two species, and it is our recommendation that it be used along with 16S rDNA sequencing when this problem arises (13). Sequencing also cannot differentiate members of the M. tuberculosis complex.

There is no standard recommended cutoff value for the distance score used to interpret sequencing data; however, the reportable range is <0.8 to 2.0% (3, 23); this needs to be standardized. Our laboratory selected ≤1.0% (99% similarity) as a cutoff value; however, future studies are needed to determine if different values may be specific for certain species, depending on the genetic diversity.

We must acknowledge the limitations of nucleic acid sequencing and concentrate on the benefits of using this technology. When new organisms are described, sequences should be peer reviewed and deposited in a database that is available to the public. Furthermore, the isolates should be placed in a culture collection that can be maintained over the years. Sequences and cultures of organisms that cannot be identified by 16S rDNA sequencing should be deposited so that their frequency and clinical importance can be determined when newer technology becomes available. It is important that sequences be placed in a database such as that at the RIDOM site, where good quality control is ensured and the information is available to all laboratories.