Vol. 22, No. 5 JOURNAL OF CLINICAL MICROBIOLOGY, Nov. 1985, p. 761-767 0095-1137/85/110761-07$02.00/0 Copyright C) 1985, American Society for Microbiology Rapid and Sensitive Identification of Mycobacterium tuberculosis CATHY V. KNISLEY,'t JAMES J. DAMATO,lt* J. KENNETH McCLATCHY,2§ AND PATRICK J. BRENNAN' Department of Microbiology, Colorado State University, Fort Collins, Colorado 80523,' and Department of Clinical Laboratories, National Jewish Hospital and Research Center, Denver, Colorado 802062 Received 1 April 1985/Accepted 1 August 1985 The fatty acid constituents of 14 species of Mycobacterium (14 isolates) and one isolate each of Corynebac- terium xerosis, Nocardia asteroides, and Streptomyces albus were examined with the purpose of distinguishing Mycobacterium tuberculosis from other acid-fast bacilli. Combined thin-layer chromatography (TLC) of methyI mycolates and gas-liquid chromatography (GVC) of shorter-chain fatty acid esters provided an unequivocal identification of M. tuberculosis in a matter of 2 to 3 days. The methodology included rapid and simplified procedures for methanolysis and extraction of bacterial lipids with equally facilitated GLC and TLC analyses. These studies were performed with 0.5 to 1.0 mg of dry bacterial cells (approximately 2.5 x 107 CFU). When applied to 100 unknown cultures, the methodology with combined TLC-GLC correctly identified all 49 of the M. tuberculosis-Mycobacterium bovis cultures and a variety of other mycobacterium taxa. It was also interesting to note that 28 of 39 (72%) of the nontuberculous mycobacteria were correctly identified. An additional five species were tentatively identified as belonging to either of two species (Mycobacterium malmoense, Mycobac- terium terrae), but in all cases, the two species belonged to the same Runyon group. All six nonmycobacterial species were differentiated from the mycobacteria studied. There are now available many excellent drug regimens for the treatment of tuberculosis (3). A primary factor which thwarts the implementation of suitable drug programs is the protracted period that is required to isolate and identify Mycobacterium tuberculosis. The reason for this is its slow growth. Admittedly, treatment for tuberculosis can be started based on clinical symptoms and the microscopic detection of acid-fast bacilli in sputa (12). However, acid- fastness may also apply to Nocardia spp., Actinomyces spp., certain fungi, and a variety of nontuberculous mycobacteria (4). Likewise, clinical symptoms which per- tain to tuberculosis may also apply to mycobacterioses other than tuberculosis and to other diseases, and the drug thera- pies recommended for tuberculosis are distinct from those recommended for infections due to nontuberculous myco- bacteria (12). Another important reason for rapid identifica- tion of M. tuberculosis is that individuals infected with this organism may be contagious and need to be isolated from other persons until an effective chemotherapy program can be established. The primary procedures used in our work for the identification of M. tuberculosis and its differentiation from other bacteria are chromatographic, based on the presence of species-specific long-chain, alpha-alkyl, beta- hydroxy fatty acids, the mycolic acids (1), and simpler, shorter-chain fatty acids in mycobacteria and related taxa. MATERIALS AND METHODS Bacterial cultures. The 14 species of mycobacteria used in this study were obtained from the Trudeau Mycobacterial * Corresponding author. t Present address: Program in Infectious Disease and Clinical Microbiology, University of Texas Health Science Center at Hous- ton, Houston, TX 77030. t Present address: Clinical Microbiology Laboratories, Depart- ment of Pathology and Area Laboratory Services, Walter Reed Army Medical Center, Washington, DC 20307-5001. § Present address: Colorado Clinical Laboratories, Denver, CO 80206. Culture Collection at National Jewish Hospital and Research Center (Table 1). One strain each of Nocardia asteroides, Corynebacterium xerosis, and Streptomyces albus were also examined. In addition, 100 previously identified isolates from the Centers for Disease Control (CDC) were tested blind by us. All strains were grown in 7H11 agar medium for 2 weeks, autoclaved for 20 min at 121°C, harvested by centrifugation, washed free of medium, lyophilized, and kept at 0°C until ready for use. Methanolysis. Some initial effort was spent in selecting conditions which would result in maximum release of fatty acids of all classes. Undoubtedly, saponification (refluxing 10% methanolic KOH) (5) gave the greatest yields in terms of quantity and number of different components; however, saponification is a protracted procedure involving extraction of fatty acids, dehydration of preparations, and esterifica- tion. Direct acid methanolysis on dry bacilli is the most rapid procedure; however, in our hands, several protocols in current use (e.g., I to 6 N methanolic HCl, 87°C, 48 h) (5, 16) resulted in minor release of methyl mycolates and inconsis- tent recovery. We found that the procedure described by Minnikin et al. (7) resulted in the optimum release of fatty acid esters and methyl mycolates in the shortest period of time. Accordingly, lyophilized organisms (usually 1 to 2 mg) were placed in a culture tube (13 by 100 mm) with a Teflon-lined screw cap. Dry methanol-toluene-concentrated sulfuric acid (30:15:1) (1 ml) was added, and the contents were sonicated (Bransonic 12; SmithKline Beckman Corp.) for 20 min to disrupt the cells. This preparation was then held at 75°C for 18 to 24 h. The methanolysates were extracted twice with 1 ml of redistilled hexanes (bp 69°C), and the extracts were combined. Approximately 100 mg of ammonium bicarbonate was added to the extracts for neu- tralization. The mixture was filtered through glass wool in a Pasteur pipette to remove excess ammonium bicarbonate, evaporated to dryness under a stream of nitrogen, and redissolved in 500 ,ul of hexane. This was then divided two ways: one half for direct thin-layer chromatography (TLC) 761 on January 26, 2020 by guest http://jcm.asm.org/ Downloaded from
Transcript
Vol. 22, No. 5JOURNAL OF CLINICAL MICROBIOLOGY, Nov. 1985, p. 761-7670095-1137/85/110761-07$02.00/0Copyright C) 1985, American Society for Microbiology
Rapid and Sensitive Identification of Mycobacterium tuberculosisCATHY V. KNISLEY,'t JAMES J. DAMATO,lt* J. KENNETH McCLATCHY,2§ AND PATRICK J. BRENNAN'Department of Microbiology, Colorado State University, Fort Collins, Colorado 80523,' and Department of Clinical
Laboratories, National Jewish Hospital and Research Center, Denver, Colorado 802062
Received 1 April 1985/Accepted 1 August 1985
The fatty acid constituents of 14 species of Mycobacterium (14 isolates) and one isolate each of Corynebac-terium xerosis, Nocardia asteroides, and Streptomyces albus were examined with the purpose of distinguishingMycobacterium tuberculosis from other acid-fast bacilli. Combined thin-layer chromatography (TLC) of methyImycolates and gas-liquid chromatography (GVC) of shorter-chain fatty acid esters provided an unequivocalidentification of M. tuberculosis in a matter of 2 to 3 days. The methodology included rapid and simplifiedprocedures for methanolysis and extraction of bacterial lipids with equally facilitated GLC and TLC analyses.These studies were performed with 0.5 to 1.0 mg of dry bacterial cells (approximately 2.5 x 107 CFU). Whenapplied to 100 unknown cultures, the methodology with combined TLC-GLC correctly identified all 49 of theM. tuberculosis-Mycobacterium bovis cultures and a variety of other mycobacterium taxa. It was also interestingto note that 28 of 39 (72%) of the nontuberculous mycobacteria were correctly identified. An additional fivespecies were tentatively identified as belonging to either of two species (Mycobacterium malmoense, Mycobac-terium terrae), but in all cases, the two species belonged to the same Runyon group. All six nonmycobacterialspecies were differentiated from the mycobacteria studied.
There are now available many excellent drug regimens forthe treatment of tuberculosis (3). A primary factor whichthwarts the implementation of suitable drug programs is theprotracted period that is required to isolate and identifyMycobacterium tuberculosis. The reason for this is its slowgrowth. Admittedly, treatment for tuberculosis can bestarted based on clinical symptoms and the microscopicdetection of acid-fast bacilli in sputa (12). However, acid-fastness may also apply to Nocardia spp., Actinomycesspp., certain fungi, and a variety of nontuberculousmycobacteria (4). Likewise, clinical symptoms which per-tain to tuberculosis may also apply to mycobacterioses otherthan tuberculosis and to other diseases, and the drug thera-pies recommended for tuberculosis are distinct from thoserecommended for infections due to nontuberculous myco-bacteria (12). Another important reason for rapid identifica-tion of M. tuberculosis is that individuals infected with thisorganism may be contagious and need to be isolated fromother persons until an effective chemotherapy program canbe established. The primary procedures used in our work forthe identification of M. tuberculosis and its differentiationfrom other bacteria are chromatographic, based on thepresence of species-specific long-chain, alpha-alkyl, beta-hydroxy fatty acids, the mycolic acids (1), and simpler,shorter-chain fatty acids in mycobacteria and related taxa.
MATERIALS AND METHODSBacterial cultures. The 14 species of mycobacteria used in
this study were obtained from the Trudeau Mycobacterial
* Corresponding author.t Present address: Program in Infectious Disease and Clinical
Microbiology, University of Texas Health Science Center at Hous-ton, Houston, TX 77030.
t Present address: Clinical Microbiology Laboratories, Depart-ment of Pathology and Area Laboratory Services, Walter ReedArmy Medical Center, Washington, DC 20307-5001.
Culture Collection at National Jewish Hospital and ResearchCenter (Table 1). One strain each of Nocardia asteroides,Corynebacterium xerosis, and Streptomyces albus were alsoexamined. In addition, 100 previously identified isolates fromthe Centers for Disease Control (CDC) were tested blind byus. All strains were grown in 7H11 agar medium for 2 weeks,autoclaved for 20 min at 121°C, harvested by centrifugation,washed free of medium, lyophilized, and kept at 0°C untilready for use.
Methanolysis. Some initial effort was spent in selectingconditions which would result in maximum release of fattyacids of all classes. Undoubtedly, saponification (refluxing10% methanolic KOH) (5) gave the greatest yields in termsof quantity and number of different components; however,saponification is a protracted procedure involving extractionof fatty acids, dehydration of preparations, and esterifica-tion. Direct acid methanolysis on dry bacilli is the most rapidprocedure; however, in our hands, several protocols incurrent use (e.g., I to 6 N methanolic HCl, 87°C, 48 h) (5, 16)resulted in minor release of methyl mycolates and inconsis-tent recovery. We found that the procedure described byMinnikin et al. (7) resulted in the optimum release of fattyacid esters and methyl mycolates in the shortest period oftime. Accordingly, lyophilized organisms (usually 1 to 2 mg)were placed in a culture tube (13 by 100 mm) with aTeflon-lined screw cap. Dry methanol-toluene-concentratedsulfuric acid (30:15:1) (1 ml) was added, and the contentswere sonicated (Bransonic 12; SmithKline Beckman Corp.)for 20 min to disrupt the cells. This preparation was thenheld at 75°C for 18 to 24 h. The methanolysates wereextracted twice with 1 ml of redistilled hexanes (bp 69°C),and the extracts were combined. Approximately 100 mg ofammonium bicarbonate was added to the extracts for neu-tralization. The mixture was filtered through glass wool in aPasteur pipette to remove excess ammonium bicarbonate,evaporated to dryness under a stream of nitrogen, andredissolved in 500 ,ul of hexane. This was then divided twoways: one half for direct thin-layer chromatography (TLC)
of methyl mycolates, and the remaining half for gas-liquidchromatography (GLC) of fatty acid methyl esters (FAME).TLC of methyl mycolates. Conventional commercial plates
coated with Silica Gel G (CaSO4 as binder; Analtech, Inc.) orhigh-performance thin-layer plates (HP-KF; Whatman) werespotted 1 cm apart with approximately one half of the methylester preparation which had been reserved for TLC (i.e., onequarter of the total material). The solvent combination washexane-diethyl ether (78:22) (17). After a single develop-ment, the plates were air dried and lightly sprayed with 0.6%potassium dichromate in 55% sulfuric acid (11) and thenheated in an oven to 150°C for 15 to 20 min. Clean glassplates were clamped over the developed chromatograms tofacilitate handling and photography.GLC of fatty acid esters. One half of the fatty acid ester
preparation obtained from 1 to 2 mg of bacilli was evapo-rated to dryness under nitrogen and redissolved in 5 ,ul ofhexanes. A sample (about 1 to 2 ,ul) was then injected intoone of two 1.8-m glass columns (4 mm, inner diameter)packed with 3% OV-1 on 80/100 Supelcoport (Supelco, Inc.).The other column was used as a reference to adjust the baseline to zero and to compensate for temperature drift. Peri-odically, the two columns were switched. Except whenotherwise indicated, injector and detector port temperaturesfor both columns were 330 and 350°C, respectively. Atemperature program was used with an initial temperature of190°C held for 5 min, then increasing by 4°C/min to a finaltemperature of 310°C which was maintained for 10 min. Theflow rates producing the greatest sensitivity were: air, 300ml/min; H2, 30 ml/min (detection gas); N2, 30 ml/min (carriergas). Detector sensitivity was set at 256 x 10-l1. A VarianAerograph 3700 gas chromatograph equipped with two flameionization detectors was used in conjunction with a Varian9176 recorder.
Several GLC stationary-phase support combinations weretested, seeking maximum resolution for a wide range of fattyacid esters and maximum thermal stability. Among thosetested were SP-2340, OV-1 on Supelcoport, and DEGS onChromosorb W. Although OV-1 was not ideal for resolvingsome unsaturated fatty acid ester combinations, its highthermal stability (350°C) made it suitable for examination ofesters of long chain length (ca. C26) which otherwise takeprotracted periods to elute. Peak identifications were tenta-tive, being based on the retention time (tR) value of knownFAME. FAME reference standards (Supelco, Analabs, Inc.,
Alltech Associates, Inc.) were used for tentative identifica-tion and are listed in Table 2.
RESULTS
TLC of methyl mycolates. Analyses of TLC of wholemethanolysates of different mycobacterial species gave re-producible patterns. Some of the results are shown in Fig. 1.The majority of strains showed a characteristic array ofmycolic acid methyl esters (MAME in Fig. 1), and this wasthe region primarily used to identify M. tuberculosis andother acid-fast bacilli.GLC of shorter-chain FAME. GLC proved to be an invalu-
able adjunct to TLC for the identification of M. tuberculosis.The conditions used were selected to enhance the areas ofdemarcation between the different mycobacterial species.The most typical peaks, and those upon which identificationwere primarily based, had retention times indicative ofmethyl tetracosanoate, hexacosanoate, and other medium-chain fatty esters. Most probably, these were the result ofon-column pyrolysis of methyl mycolates (9, 10). To test'thispossibility and to arrive at optimal conditions for the pro-duction of these indicator esters, we examined the effects ofincreasing the injector-port temperature on the M. tubercu-losis products (Fig. 2). These were compared with in-tubepyrolysis conducted as described by Stodola et al. (13).There was a dramatic increase in the production of themedium-chain esters when the temperature was increasedfrom 250 to 330°C, and beyond that temperature there waslittle increase (Fig. 2A and C). Moreover, the effects of330°C, 350°C, and in-tube pyrolysis were about the same(Fig. 2C, D, and E). The injector-port temperature of 330°Cwas preferred because the majority of injector septa wereunstable at higher temperatures. Accordingly, 330°C on-column pyrolysis was used throughout this work, thereby
a The number following the colon denotes the number of double bondsfound in that methyl ester. Methyl esters with the same number of doublebonds but having different bond locations are grouped together.
E F GH l J K L M NOFIG. 1. TLC of methyl mycolates from standard cultures. Solvent, hexane-diethyl ether (78:22); stain, 0.6% K2CrO4 in 55% H2SO4. Lanes:
A, M. tuberculosis; B, M. bovis BCG; C, M. intracellulare serovar 8; D, M. xenopi; E, M. kansasii; F, M. fortuitum; G, Mycobacteriumsimiae serovar I; H, M. simiae serovar II; I, Mycobacterium szulgai; J, Mycobacterium smegmatis; K, Mycobacterium phlei; L, M.marinum; M, N. asteroides; N, C. xerosis; 0, S. albus.
eliminating the need for a separate pyrolysis procedure andeffectively shortening identification time.
In general, fatty acid esters of chain length C14 to C17which had a tR of 2 to 4 min were not markedly dissimilarfrom one species to another (Fig. 3). The earliest peaks,which showed consistent differences among species, corre-
sponded to a range of C18 fatty acid esters (OR, 5 to 6.5 min).These peaks (11, 12, and 13 in Fig. 3) differed principally inrelative heights and were consistently reproduced with re-
peat analyses on the same and different lots of mycobacteria.Several other peaks which were of paramount importance inidentification were in the C24 (tetracosanoate) and C26(hexacosanoate) region (tR, 17 to 18 and 19 to 20 min),respectively. Those peaks were designated 19 and 20, re-
spectively (Fig. 3), and showed both qualitative and quanti-tative differences. For example, C26 (peak 20) was present inM. tuberculosis and Mycobacterium bovis but absent fromMycobacteriumfortuitum and Mycobacterium kansasii (Fig.3A, B, E, and F), and, for a quantitative example, C24 (peak19) in serovar 8 of the Mycobacterium avium-Mycobacte-rium intracellulare-Mycobacterium scrofulaceum (MAIS)complex was found in greater amounts than that observed inM. tuberculosis (Fig. 3D and A). Differentiation can also bebased on other distinctive peaks. For instance, in the case ofMycobacterium xenopi, there was a fatty acid ester at about3.5 min (peak 9) which was not present in those othermycobacteria which contained a large C26 peak. Moreover,M. kansasii had a large characteristic peak (peak 2) at 2.5min which was present in only two other species tested thusfar. Identification, therefore, can be based on those peaksdesignated 1 through 20. Figure 4 is a decision tree based on
these characteristic peaks.Application of combined TLC-GLC to unmarked isolates
from CDC. Dixie Snider and Robert Good of the CDC,Atlanta, Ga., provided four lots of 25 cultures with theprovision that growth and identification be completed within
3 weeks. Since the combined TLC-GLC procedure allowedidentification of as few as 2.5 x 107 CFU, it was usuallypossible to use a portion of the supplied inoculating cultureto accomplish identification. However, cultures were alsogrown on 7H11 agar plates (courtesy of Leonid Heifets andBarbara Bassett), and the harvested organisms were identi-fied a second time. Our instructions were to classify theisolates into one of several categories which included thefollowing: M. tuberculosis-M. bovis, mycobacteria otherthan tuberculosis, other microorganisms, and no organisms(no growth). The results are shown in Table 3. Table 4demonstrates the potential of the methodology for successfulidentification of mycobacteria.
DISCUSSION
Owing to its sensitivity, GLC of fatty acid esters showedpromise for becoming a feasible method for differentiatingmycobacteria. For instance, Thoen et al. (14, 15) examinedfatty acid esters from 35 strains of M. kansasii and Myco-bacteriurn marinum. All of the M. kansasii isolates were
similar, characterized by an alpha-methyl branched fattyacid ester of medium chain length. This substituent was
absent from lipid extracts of M. avium and other non-photochromogens, which gave hope that fatty acid profilescould be used generally for mycobacterial identification.Indeed, the vittual absence of this branched fatty acid in M.marinum, like M. kansasii a photochromogen, gave cre-
dence to this possibility; therefore, the idea of identificationbased on fatty acid composition was further pursued. Larr-son and Mardh (5) claimed reproducible, albeit slight, differ-ences in the fatty ester profiles of M. tuberculosis, M. bovis,M. avium, and M. kansasii. Ohashi et al (8) demonstratedthat M. tuberculosis was distinguished by one relativelypolar ester which was absent from a range of othermycobacteria.
RETENTION TIME (MIN)FIG. 2. GLC of M. tuberculosis comparing on-column pyrolysis with different injector temperatures with complete (in-tube) pyrolysis.
Injector temperatures are indicated on the chromatogram along with peak 20 (hexacosanoate, C26:0). Column, 3% OV-1 on 80/100Supelcoport; temperature program, 190°C/5 min, then increasing 4°C/min to 310°C.
Perhaps the most impressive case for identification ofmycobacterial species based on fatty acid profiles was pre-sgnted by Tisdall et al. (16). Of 81 clinical isolates, 64% wereidentified to species level by GLC alone and an additional35% were differentiated to the same groups of two or threeorganisms. Their paper represented an important study sinceit clearly demonstrated that GLC of fatty acid derivatives,when conducted with care, will allow identification of spe-cies and sensitive identification of mycobacteria. A featureof the work of Tisdall et al. (16) was that the most charac-teristic fatty acids were generally greater than C24. There-fore, it occurred to us that if, through the use of pyrolysis ofthe species-specific mycolates, the larger fatty acids could beaccentuated, then differentiation and identification would beenhanced. While applying this idea of pyrolysis GLC tomycobacterial isolates, Wayne Moss kindly supplied us withunpublished information on the application of the sameprincipal to a few mycobacterial species. The results (2)were similar to those recorded above and further substanti-ated the evidence that this approach can provide consistent,
readily obtainable profiles upon which AM. tuberculosis canbe identified.A number of the GLC peaks were tentatively identified
based on the reference standards available, but many of theunusual branched-chain unsaturated fatty acids found inbacterial cell walls were not available commercially. Toidentify these components, mass spectrophotometry wasnecessary, but this methodology is not practical for use inthe clinical laboratory where speed, accuracy, and cost arecrucial elements of any implemented technology. Using thedescribed GLC and TLC methodologies in conjunction witheach other provided a clinically oriented system for the rapididentification of mycobacteria.These procedures have been used with as few as 2.5 x 107
CFU (ca. 25 ,ug of dry bacilli) of M. tuberculosis. Such smallquantities of bacteria were estimated while still in suspen-sion (i.e., before lyophilization) by using a McFarland stan-dard no. 1 with an optical density of 0.10, which corre-sponded to 150 x 106 CFU. Knowing that 1 mg of dry bacilliwas equivalent to about 109 organisms, weights were calcu-
RETENTION TIME (MIN)FIG. 3. GLC of shorter-chain (less than C26:0) FAME from standard cultures. Injection temperature, 330°C; other conditions are described
in the legend to Fig. 2. M. TB, M. tuberculosis; BCG, M. bovis BCG; M. XEN., M. xenopi; MAIS 8, M. intracellulare serovar 8; M. FORT.,M. fortuitum; M. Kans., M. kansasii.
lated accordingly. Special vials (Reacti-Vials, 10 .l1, catalogno. 13100; Pierce Chemical Co.) were useful in concentratingthe small amount of fatty acid esters derived from suchquantities of bacteria and allowing the implementation ofboth TLC and GLC.There was little doubt from the pioneering work of
Minnikin and colleagues (6) that TLC of methyl mycolatesprovides an excellent means for the differentiation of M.tuberculosis from other acid-fast bacteria. However, themore recent proposal on the use of two-dimensional chro-matography or multidevelopment (7) posed an obstacle forthe ready implementation of the technique in a clinicalsetting. Moreover, it was commonly held that TLC is ofinsufficient sensitivity for clinical purposes. Clearly, one-dimensional TLC of methyl mycolates (Fig. 1) provides aconsiderable element of the type of specificity required forpresent needs and, when combined with GLC, provides thedegree of accurate identification (Table 3) that is necessaryin the clinical laboratory. Moreover, TLC, particularly
microparticulate (so-called high-performance) TLC, seemsto be of sufficient sensitivity to warrant its retention as aroutine diagnostic tool. We were able to recognize a definitepattern with as few as 2.5 x 107 CFU and still have sufficientmaterial for GLC purposes. To more fully evaluate thismethodology, it will be necessary to challenge the test witha greater number of sputum specimens.The region of the plate above the mycolic acid methyl
esters contained the methyl esters of shorter-chain fattyacids, including the short-chain (C26 to C34) methylmycolates of C. xerosis, and was only of value whendistinguishing between mycobacteria, corynebacteria, andN. asteroides. Of all the strains tested thus far, M. tubercu-losis and M. bovis BCG were frequently the most difficult todifferentiate (Fig. 1, lanes A and B). However, in the contextof human disease, it is not usually necessary to distinguishbetween these. The mycolates of M. bovis BCG are separa-ble into three major components, with the middle mycolatepresent in much greater abundance, while M. tuberculosis
M. chOlonoFIG. 4. Decision tree for identification of mycobacteria based on GLC of fatty acid esters. These decisions are based on the profiles shown
in Fig. 3 and other chromatograms not reproduced. MAIS, M. aviiu,n-M. intracelluilare-M. scrofulaceum complex.
routinely yields a larger mycolate spot which cannot beresolved further with a variety of solvents. In the samplesreceived from the CDC, there were six each M. bovis BCGand M. bovis (no strain indicated). All six M. bovis BCG hadTLC patterns identical to those of the test standards whichinclude M. bovis BCG Ravenel (TMC 401) and M. bovisBCG Branch (TMC 407), but the six other M. bovis isolatesand all M. tuberculosis isolates had TLC patterns identicalwith the M. bovis and M. tuberculosis test standards. Itappeared that M. bovis BCG had a different methyl mycolatepattern from that of M. tuberculosis and M. bovis.Of the several species examined by this procedure, M.
kansasii was the only other one with a pattern similar to thatof M. tuberculosis and M. bovis (Fig. 1, lanes A and E);however, the mycolates of M. kansasii were sufficiently
TABLE 3. Results of identification by TLC-GLC on culturesprovided by CDC'
identificationOther microorganisms 6/6 6/6 6/6No microorganisms 6/6 6/6 Not required
a At the time the study was conducted, there were no TLC-GLC standardsavailable to facilitate identification of eight isolates.
resolved into two components to allow for an adequatedifferentiation of the two species.
In regard to our confidence in the accuracy of identifica-tion by these methods, the combination of TLC and GLCprovided an unambiguous identification for all of the M.tuberc ulosis-M. bovis strains. For mycobacteria other than
TABLE 4. Results of analysis of combined TLC-GLC onnontuberculous mycobacteria and other microorganisms provided
by CDC"
No. ofOur identification by TLC-GLC True identification (CDC) strains
(correct/total)'
MAIS complex M. aviium 9/9M. simiae M. simniae 4/4M. kansasii M. kansasii 3/4M. fortuitum-M. chelonei M. fortuitum-M. chelonei 4/4complex
M. szulgai-M. gordonae M. szulgai-M. gordonae 6/7complex
M. flavescens M. flavescens 2/3Mycobacterium sp." M. malmoense 0/3Mvcobacterium sp." M. terrae 0/2Mvcobacterium sp." M. marinum 0/2M. tubercullosis Runyon group III 0/1Nocardia sp. N. asteroides 4/4Nocardia sp. Rhodochrous complex 2/2
a It was determined by TLC-GLC that these samples did not correspond toany of the standard cultures used in this study. As a result, identification wasnot possible.
tubercle bacilli, it was possible to rapidly separate them fromthe tubercle bacilli. Although some nontuberculous bacteriamay look like M. tuber(culosis by one method or another, acombination of the two gives conclusive results. In addition,it was possible to identify 28 of 39 nontuberculousmycobacteria and to differentiate these organisms from theother microorganisms examined.
ACKNOWLEDGMENTSThis work was supported by contract 200-80-0554 from the CDC,
U.S. Department of Health and Human Resources. Atlanta. Ga.We acknowledge the participation of L. Heifets and Barbara
Bassett of the National Jewish Hospital and Research Center andRobert Good and Dixie Snider of the CDC in this program, andwe thank them for their help.
LITERATURE CITED1. Goren, M. B., and P. J. Brennan. 1979. Mycobacterial lipids:
chemistry and biological activities, p. 63-193. In G. P. Youman(ed.), Tuberculosis. The W. B. Saunders Co., Philadelphia.
2. Guerrant, G. O., M. A. Lambert, and C. W. Moss. 1981.Gas-chromatographic analysis of mycolic acid cleavage prod-ucts in mycobacteria. J. Clin. Microbiol. 13:899-907.
3. Hinshaw, H. C. 1979. Treatment of tuberculosis, p. 457-488. InG. P. Youman (ed.), Tuberculosis. The W. B. Saunders Co.,Philadelphia.
4. Lamanna, C., M. F. Mallette, and L. N. Zimmerman. 1973.Basic bacteriology, its biological and chemical background, p.162-163, 4th ed. The Williams & Wilkins Co., Baltimore.
5. Larsson, L., and P. A. Mardh. 1976. Gas chromatographiccharacterization of mycobacteria: analysis of fatty acids andtrifluoracetylated whole-cell methanolysates. J. Clin. Microbiol.3:81-85.
6. Minnikin, D. E., L. Alshamoany, and M. Goodfellow. 1975.Differentiation of M! cohbacteriutn, Nocardfia and related taxa bythin-layer chromatographic analysis of whole-organismmethanolysates. J. Gen. Microbiol. 88:200-204.
7. Minnikin, D. E., I. G. Hutchinson, A. B. Caldicott, and M.Goodfellow. 1980. Thin-layer chromatography of methanoly-sates of mycolic-acid containing bacteria. J. Chromatogr.188:221-233.
8. Ohashi, D. K., T. J. Wade, and R. J. Mandle. 1977. Character-ization of ten species of mycobacteria by reaction gas-liquidchromatography alone. J. Clin. Microbiol. 6:469-473.
9. Reiner, E., R. E. Beam, and G. P. Kubica. 1969. Pyrolysisgas-liquid chromatography studies for the classification ofmycobacteria. Am. Rev. Respir. Dis. 99:750-759.
10. Reiner, E., J. J. Hicks, R. E. Beam, and H. L. David. 1971.Recent studies of mycobacterial differentiation by means ofpyrolysis gas-liquid chromatography. Am. Rev. Respir. Dis.104:656-660.
11. Rouser, G., G. Kritchevsky, and A. Yamamato. 1967. Columnchromatographic and associated procedures for separation anddetermination of phosphatides and glycolipids, p. 99-162. InG. V. Marinetti (ed.), Lipid chromatographic analysis, vol. 1.Marcel Dekker, Inc., New York.
12. Sommers, H. M. 1979. The laboratory diagnosis of mycobacte-rial disease, p. 404-434. In G. P. Youman (ed.), Tuberculosis.The W. B. Saunders Co., Philadelphia.
13. Stodola, F. H., A. Lesuk, and R. J. Anderson. 1938. Thechemistry of the lipids of tubercule bacilli. LIV. The isolationand properties of mycolic acid. J. Biol. Chem. 128:505-513.
14. Thoen, C. O., A. G. Karlson, and R. D. Ellefson. 1971. Fattyacids of Mvcbacterilun kantsasii. Appl. Microbiol. 21:628-632.
15. Thoen, C. O., A. G. Karlson, and R. D. Ellefson. 1972. Differ-entiation between Mvcobacterium kansasii and Mvcohacterilumtnariniin by gas-liquid chromatographic analysis of cellularfatty acids. Appl. Microbiol. 24:1009-1010.
16. Tisdall, P. A., G. D. Roberts, and J. P. Anhalt. 1979. Identifi-cation of clinical isolates of mycobacteria with gas-liquid chro-matography alone. J. Clin. Microbiol. 10:506-514.
17. Toriyama, S., I. Yano, M. Masui, M. Kusunose, and E.Kusunose. 1978. Separation of C5,()6( and C7080 mycolic acidmolecular species and their changes by growth temperatures inMvcohbacteritmn philei. FEBS Lett. 95:111-115.
