Detection for the Sensitive Determination of Specific Mycobacterial ...

3 downloads 0 Views 665KB Size Report
of Mycobacterium leprae isolated from a naturally infected armadillo. Two-dimensional gas chromatography with electron capture detection may in some ...
JOURNAL OF CLINICAL MICROBIOLOGY, OCt. 1989, p. 2230-2233 0095-1137/89/102230-04$02.00/0 Copyright C 1989, American Society for Microbiology

Vol. 27, No. 10

Two-Dimensional Gas Chromatography with Electron Capture Detection for the Sensitive Determination of Specific Mycobacterial Lipid Constituents LENNART LARSSON,1* JULIO JIMENEZ,1 ANDERS SONESSON,2 AND FRANCOISE PORTAELS3 Department of Medical Microbiology, Lund University Hospital, Solvegatan 23, S-223 62 Lund,' and Department of Technical Analytical Chemistry, Chemical Centre, Lund University, S-221 00 Lund,2 Sweden, and Institute of Tropical Medicine, B-2000 Antwerp, Belgium3 Received 24 April 1989/Accepted 7 July 1989

A method was developed for determining two characteristic mycobacterial lipid constituents, tuberculostearic acid (as its pentafluorobenzyl ester) and 2-eicosanol (as its pentafluorobenzoyl ester), by using gas chromatography with electron capture detection. A microprocessor-controlled column-switching system (two-dimensional gas chromatography) facilitated sample preparation and increased specificity. The usefulness of the technique was illustrated by its ability to reveal picogram amounts of tuberculostearate in a suspension of Mycobacterium leprae isolated from a naturally infected armadillo. Two-dimensional gas chromatography with electron capture detection may in some instances provide a convenient alternative to gas chromatographymass spectrometry for use in demonstrating the presence of mycobacteria in a complex environment.

Gas chromatography (GC) can be used for determining mycobacterial lipid constituents in trace amounts. The technique has been applied to the analysis of suspensions of Mycobacterium leprae prepared from experimentally infected armadillos in order to detect any possible contamination by in vitro-cultivable mycobacteria (4, 9). In addition, the presence of mycobacteria (intact as well as disintegrated cells) can be directly demonstrated in clinical specimens (3, 5, 7). Selected ion monitoring (SIM) by using mass spectrometry, particularly chemical-ionization negative-ion mass spectrometry, is preferred as the detection mode (5). Alternatively, electron capture detection (ECD) may be used; this detection method is considerably less expensive than mass spectrometry and provides almost as good sensitivity, but it is also much less selective. A variety of different postderivatization sample cleanup procedures have been developed for removing excess amounts of reagents and other interfering components in attempts to utilize ECD for detecting trace amounts of halogenated derivatives of microbial metabolites (1, 2, 6, 11, 12). We reported recently that two-dimensional GC (columnswitching technique) provides extremely efficient on-line purification of samples, thus allowing considerably simplified sample preparation prior to GC-ECD analysis (13). With column switching, sample components are first separated on a pre-column, after which selected fractions are switched to a cold-trap unit and then injected into the analytical column. The present report describes a method of using this novel technique to sensitively and selectively determine tuberculostearic acid (TSA), a cellular component present in all pathogenic species of mycobacteria, and 2-eicosanol, a secondary alcohol found in some nontuberculous mycobacterial species, including the M. avium complex. The method is illustrated by detection of TSA in a preparation of M. leprae isolated from an infected armadillo.

*

MATERIALS AND METHODS Microorganisms and chemicals. We used three clinical isolates belonging to the M. avium complex, cultivated on Middlebrook 7H10 agar for three weeks, and a suspension of M. leprae prepared from the liver of a naturally infected armadillo as described previously (10). Analytical-grade 2,3,4,5,6-pentafluorobenzoyl (PFBO) chloride was purchased from Fluka Chemie AG (Buchs, Switzerland). 2,3,4,5,6-Pentafluorobenzyl (PFB) bromide and triethylamine were obtained from Sigma Chemical Co. (St. Louis, Mo.), and pyridine was obtained from E. Merck AG (Darmstadt, Federal Republic of Germany). TSA was from our laboratory collection of reference bacterial fatty acids. Sample preparation. The mycobacterial preparations were freeze-dried and heated in 1 ml of 15% (wt/vol) sodium hydroxide in 50% (vol/vol) aqueous methanol for 1 h at 80°C. After hexane extraction (2 ml), the aqueous phase was acidified by using 0.2 ml of dilute sulfuric acid and then reextracted. Both hexane phases were evaporated to dryness. The first extract was subjected to 2-eicosanol derivatization, and the second was subjected to TSA derivatization. The PFBO alcohol derivative was prepared by adding acetonitrile (30 ,ul) to the dried sample followed by PFBO chloride (10 pul) and pyridine (10 ,ul). Correspondingly, the PFB acid derivative was prepared by adding a 30% (vol/vol) solution of PFB bromide in acetonitrile (10 ,ul), a 2% (vol/vol) solution of trietylamine in acetonitrile (10 ,ul), and acetonitrile (30 ,ul). Both the PFB and PFBO reactions were allowed to proceed at room temperature for 30 min; thereafter, 1.5 ml each of hexane and phosphate-buffered saline buffer (pH 7.2) were added. After shaking, the organic phase was evaporated to dryness and made up with hexane before analysis. The detection sensitivity for TSA was evaluated by using the synthesized compound. GC. A model 3700 (Varian Los Altos, Calif.) gas chromatograph equipped with a 63Ni ECD and a multipleswitching intelligent controller (MUSIC) column-switching system (Chrompack, Middelburg, The Netherlands) was used. A fused-silica wide-bore column (10 m by 0.53 mm) with cross-linked CP-SIL-8 (Chrompack) as the stationary

Corresponding author. 2230

VOL. 27, 1989

DETERMINATION OF LIPIDS BY TWO-DIMENSIONAL GC

phase was used as the pre-column attached to a flame ionization detector (FID), while a fused-silica capillary column (25 m by 0.25 mm) coated with SE-30 (SGE, Ringwood, Australia) was used as the analytical column attached to the ECD. The nitrogen carrier gas flow rate through the precolumn was 4 ml/min, and the inlet pressure of the analytical column was 0.9 x 105 Pa. The temperature of the injector was 250°C, while that of the detectors was 350°C. The temperature of both columns was programmed to rise 10°/ min from 180 to 260°C. The cold trap was chilled to -70°C with liquid carbon dioxide and heated to 250°C upon reinjection of the trapped fractions into the analytical column. The data were evaluated by using a Chrompack control and integration system with an IBM PS/2 model 30 and a Chrompack BD 70 printer-plotter. A Varian model 3500 gas chromatograph equipped with a split-splitless injector and an FID was used in comparative experiments. The column and the chromatographic conditions were identical to those used for the analytical column in the MUSIC system (see above).

2231

a

b RESULTS The methods used for preparing the PFB and PFBO ester derivatives were, in principle, the same as those described previously (5, 14). The retention times of the derivatives on the pre-column were 10.3 min (TSA) and 10.9 min (2eicosanol). Ninety-second fractions around these compounds were cold-trapped and subsequently automatically introduced onto the analytical column. No overloading of the ECD was observed even though appreciable amounts of the unreacted, halogenated reagents were present in the samples injected onto the pre-column. Gas chromatograms demonstrating the presence of TSA in 5 ng of M. avium complex cells (corresponding to approximately 25 pg of the PFB derivative) and 2-eicosanol in 20 ng of cells are shown in Fig. 1. Obviously, other mycobacterial constituents simultaneously injected onto the pre-column did not interfere with the analyses. Chromatograms representing the M. leprae suspension, derivatized for TSA detection, are shown in Fig. 2. The PFB derivative could not be detected by using conventional capillary column GC with FID (Fig. 2a); excessive background noise was observed when the attenuation of the detector signal was lowered (Fig. 2b). When the same preparation was analyzed by using the MUSIC-ECD system, however, TSA was readily detected (Fig. 2c). The identity of TSA was confirmed by negative-ion-SIM analysis, as previ-

ously reported (5). DISCUSSION Use of GC analysis for detecting mycobacteria in clinical and environmental specimens takes advantage of their unusual lipid constituents. Thus, TSA is found in all pathogenic species of Mycobacterium as well as in certain other members of the order Actinomycetales. 2-Eicosanol, present only in a limited number of mycobacterial species including the M. avium complex and M. Iepraemurium, is not produced, e.g., by members of the M. tuberculosis complex or M. leprae. These differences in lipid composition can be exploited in several ways. For example, the presence of 2-eicosanol in suspensions of M. leprae prepared from experimentally infected armadillos indicates contamination by certain in vitro-cultivable mycobacteria, which may confuse bacteriological-immunological studies of the suspen-

FIG. 1. Detection of tuberculostearic acid in approximately 5 ng (a) and 2-eicosanol in approximately 20 ng (b) of mycobacterial cells belonging to the M. avium complex.

sions (4, 9). To be able to reveal such contamination, e.g., in M. leprae preparations intended for vaccine production, is important. Furthermore, detection of TSA in sputum and cerebrospinal fluid specimens may be useful for making rapid diagnoses of mycobacterial infections (1-3, 5, 7). It should be understood that GC, regardless of the type of detector used, registers both viable and nonviable mycobacteria, as well as fragments thereof, with equal sensitivity. The need for a highly sensitive and selective means of GC detection in the above-mentioned applications is obvious. Although negative-ion-SIM is clearly superior in these respects (5), the instrumentation required is expensive and sophisticated. Since it has previously been shown that both ECD and negative-ion-SIM respond with high sensitivity to PFB and PFBO derivatives (8, 14), we decided to investigate whether two-dimensional GC with ECD would be useful for detecting trace amounts of mycobacteria in a complex environment by using the corresponding TSA and 2-eicosanol derivatives as analytes. Negative-ion-SIM was used for unequivocal identification of TSA in parallel studies; its use is highly recommended as a reference method since reten-

LARSSON ET AL.

2232

J. CLIN. MICROBIOL. 16:0

a

18:O

TSA

0(

1

1

1

n

qb

d

e.

4

c-

e

C.

4

cb

pz

c-

4

«b

e

q. Ci. ...

q» ..

«.

!i



!i

z :; ..

..

18:0 c

FIG. 2. Detection of tuberculostearic acid (TSA) in a suspension of M. leprae prepared from a naturally infected armadillo using FID (a and b) and ECD with column switching (c). Peak designations refer to the numbers of carbon atoms and double bounds in the fatty acid derivatives.

tion time comparisons, even when using capillary columns with stationary phases of different polarity, permit only tentative identification. TSA and 2-eicosanol are released from mycobacterial lipids through hydrolysis. Clearly, hydrolysates of clinical

and environmental samples will contain several common fatty acids of even-numbered carbon chain lengths in concentrations exceeding those of TSA by several orders of magnitude (cf. the studied M. leprae preparation, Fig. 2a and b). Attempts to analyze such preparations with GC-ECD without column switching will therefore result in either overloading of the detector (injecting concentrated extracts) or lowered sensitivity (injecting diluted extracts), even though excess amounts of the halogenated reagents may have been carefully removed after derivatization, e.g., by using acid and base washing, disposable columns, or other purification means (1). On the basis of our experiences, in practice it is virtually impossible to utilize the full inherent sensitivity of ECD without column switching for detecting picogram amounts of TSA and 2-eicosanol in hydrolysates of biological samples containing only minor proportions of mycobacteria. The advantage of two-dimensional GC is that only relevant, narrow fractions of the sample are selectively injected onto the analytical column connected to the ECD, leading to considerably improved selectivity. Thus, not only excess amounts of the reagents but also other disturbing components, including the common fatty acids (palmitic acid, stearic acid, etc.) are prevented from entering the analytical column. The extent to which the latter compounds are removed will vary, depending on their GC retention time characteristics. Both TSA and 2-eicosanol were easily detected in a few nanograms of cultivated mycobacteria (Fig. 1), and picogram amounts of TSA could be detected in a preparation of M. leprae (Fig. 2c). Obviously, the latter preparation consisted mainly of nonmycobacterial material (probably tissue components). When MUSIC-ECD is used, the retention times of the studied compounds on the pre-column must be determined first. In the present study, the maximum amounts of the TSA and 2-eicosanol derivatives injected onto the pre-column (connected to the FID) were approximately 300 ng. Initially, when we used splitless injection, adsorption of trace amounts of these compounds (particularly the TSA derivative) occurred in the injector, resulting in ghost peaks observed by ECD at subsequent injections. However, the problem disappeared once we had changed to direct oncolumn injection onto the wide-bore capillary pre-column. The parameters used for trapping the relevant fractions) should be selected in such a way that the compound(s) of interest is introduced onto the analytical column quantitatively while the coinjection of nonrelevant and disturbing compounds is minimized. In practice, however, a certain degree of overlapping must be tolerated; for example, the coinjection of a minute proportion of stearic acid derivative with TSA is almost unavoidable because these derivatives have similar retention times. The MUSIC system requires relatively little sample preparation and considerably improves the performance of GCECD analysis of halogenated derivatives of bacterial constituents. Further studies are in progress at our laboratories aimed at determining whether MUSIC-ECD can, with a minimum of sample preparation, be used to directly detect mycobacterial components in clinical specimens. The technique would be useful for analyzing specimens which, although negative in Ziehl-Neelsen staining, are suspected to contain mycobacteria. ACKNOWLEDGMENTS

This study was supported by grants from the Swedish Heart and Lung Foundation.

VOL. 27, 1989

DETERMINATION OF LIPIDS BY TWO-DIMENSIONAL GC

2233

A. Valeur is gratefully acknowledged for the mass spectrometric analyses.

marine microfouling community by capillary gas chromatography-chemical ionization mass spectrometry with negative ion detection. J. Microbiol. Methods 3:331-344.

LITERATURE CITED 1. Brooks, J. B., M. I. Daneshvar, D. M. Fast, and R. C. Good. 1987. Selective procedures for detecting femtomole quantities of tuberculostearic acid in serum and cerebrospinal fluid by frequency-pulsed electron capture gas-liquid chromatography. J. Clin. Microbiol. 25:1201-1206. 2. Daneshvar, M. I., and J. B. Brooks. 1988. Improved procedure for preparation of pentafluorobenzyl derivatives of carboxylic acids for analysis by gas chromatography with electron capturedetection. J. Chromatogr. Biomed. Apple. 433:248-256. 3. French, G. L., C. Y. Chan, S. W. Cheung, R. Teoh, M. J. Humphries, and G. O'Mahoney. 1987. Diagnosis of tuberculous

9. Portaels, F., C. Asselineau, I. Baess, M. Daffé, G. Dobson, P. Draper, D. Gregory, R. M. Hall, T. Imaeda, P. A. Jenkins, M. A. Lanéelle, L. Larsson, M. Magnusson, D. E. Minnikin, S. R. Pattyn, G. Wieten, and P. R. Wheeler. 1986. A cooperative

4.

5.

6. 7. 8.

meningitis by detection of tuberculostearic acid in cerebrospinal fluid. Lancet i:118-120. Larsson, L., P. Draper, and F. Portaels. 1985. Use of gas chromatography to differentiate Mycobacterium leprae from cultivable armadillo-derived mycobacteria, M. aviumlintracellulare and M. Iepraemurium by analysis of secondary alcohols. Int. J. Lepr. 53:441-446. Larsson, L., G. Odham, G. Westerdahl, and B. Olsson. 1987. Diagnosis of pulmonary tuberculosis by selected ion monitoring: improved analysis of tuberculostearate in sputum using negative-ion mass spectrometry. J. Clin. Microbiol. 25:893-896. Larsson, L., A. Sonesson, and J. Jimenez. 1987. Ultrasensitive analysis of microbial fatty acids using gas chromatography with electron capture detection. Eur. J. Clin. Microbiol. 6:729-731. Mardh, P. A., L. Larsson, N. Hoiby, H. C. Engbaek, and G. Odham. 1985. Tuberculostearic acid as a diagnostic marker in tuberculous meningitis. Lancet i:367. Odham, G., A. Tunlid, G. Westerdahl, L. Larsson, J. Guckert, and D. White. 1985. Determination of microbial fatty acid profiles at femtomolar levels in human urine and the initial

taxonomic study of mycobacteria isolated from armadillos infected with Mycobacterium leprae. J. Gen. Microbiol. 132: 2693-2707. 10. Prabhakaran, K., E. B. Harris, and W. F. Kirchheimer. 1976. Hypopigmentation of skin lesions in leprosy and occurrence of o-diphenoloxidase in Mycobacterium leprae. Pigm. Cell 3:152164. 11. Rosenfeld, J. M., O. Hammerberg, and M. C. Orvidas. 1986.

Simplified methods for preparation of microbial fatty acids for analysis by gas chromatography with electron-capture detection. J. Chromatogr. Biomed. Appl. 378:9-16. 12. Sonesson, A., L. Larsson, and J. Jimenez. 1987. Use of pentafluorobenzyl and pentafluoropropionyl/pentafluorobenzyl esters of bacterial fatty acids for gas chromatographic analysis with electron capture detection. J. Chromatogr. Biomed. Apple. 417: 366-370. 13. Sonesson, A., L. Larsson, and J. Jimenez. 1989. Two-dimensional gas chromatography with electron-capture detection used in the determination of specific peptidoglycan and lipopolysaccharide constituents of gram-negative bacteria in infected human urine. J. Chromatogr. Biomed. Appl. 490:71-79. 14. Sonesson, A., L. Larsson, G. Westerdahl, and G. Odham. 1987. Determination of endotoxins by gas chromatography: evaluation of electron capture and negative ion chemical ionization mass spectrometric detection of halogenated derivatives of P-hydroxy myristic acid. J. Chromatogr. Biomed. Appl. 417: 11-25.