Journal of Applied Bacteriology 1992, 73,426-432

Analysis of polar lipids from some representative enterobacteria,Plesiomonas and Acinetobacter by fast atom bombardment-mass spectrometry H.S. Aluyi, Valerie Bootel, D.B. Drucker, J.M. Wilson’ and Y.H. Ling2 Department of Cell and Structural Biology, University of Manchester Dental School, Manchester and ’Department of Chemistry, and ‘Medical Computation Unit, University of Manchester, Manchester, UK 4097/01/92: accepted 27 May 1992 H. S . ALUYI. V. B O O T E , D.B. DRUCKER, J . M . WILSON AND Y.H. LING. 19w.Fast atom bombardment-mass spectrometry (FAB-MS) was used to analyse lipid extracts of bacteria to assess its usefulness for analysing anionic phospholipids of potential chemotaxonomic value. The following micro-organisms were tested : Acinetobacter calcoaceticus, Acinetobacter sp., Citrobacter freundii, Enterobacter cloacae (2 strains), Escherichia coli (3 strains), Hafnia alvei, Klebsiella oxytoca, Klebsiella pneumoniae, Morganella morganii, Plesiomonas shigelloides, Proteus mirabilis (3 strains), Serratia liquefaciens and Serratia marcescens. Negative-ion spectra provide data for twenty-seven major carboxylate anions (m/z 209-325) and for thirty-seven major phospholipid anions (m/z 645-774). Generally, the largest carboxylate peaks were due to 16 : 1, 16 : 0, cycl7 and 18 : 1 while the largest phospholipid anion peaks were due to PE(32 : I), PE(33 : l), PE(34 : l), PE(34 : 2), PG(30 : 2), PG(31 : 2), PG(32 : 2), PG(34 : 1) and PS (33 : 0). However, quantitative differences were observed. For example, Acinetobacter lacked PE ( 3 3 : 1) but had exceptionally high peaks at m/z 748, PS(33 : 0), and m/z 281, octadecanoate. Unknown ‘carboxylate’ peaks were detected at m/z 254, 256, 261, 268, 282 and 301. In some cases, unknown peaks appeared to constitute possible homologous series being separated by A m/z of 14( = methylene). For chemotaxonomic purposes, the complexity of the data required numerical analysis. Using the Pearson coefficient of linear correlation, as a measure of association, it was possible to compare all strains analysed. Typical results for strain comparisons were as follows : Ent. cloacae vs Ent. cloacae, r = 0-90 (Ent. cloacae vs Ac. calcoaceticus, r = 0-46). Thus FAB-MS represents an excellent means of obtaining large quantities of data on polar lipids of a range of bacterial isolates, which may be suitable for chemotaxonomic purposes.

INTRODUCTION

The concept of using chemical analyses for microbial taxonomic purposes, or chemotaxonomy, has gained widespread acceptance in microbiology (Goodfellow & Minnikin 1985). Suitable data can be collected by various chemical analytical techniques such as gas chromatography (GC) (Drucker 198I), high performance liquid chromatography (HPLC) (Drucker 1987), reverse phase thin layer chromatography (RP-TLC) (Collins 1985) or sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) (Khwaja et al. 1990). These are primarily separative, rather than chemical identification techniques. An alternative Currespondenre t o : Dr D.B. Drucker, Department of Cell and Structural B i o l u ~ University , of Manchester Dental School, Higher Cambridge Street, Manchester M I 5 6FH. U K .

approach, which can be combined with a separative technique, is mass spectrometry (MS) (Drucker & Jenkins 1989). M S has historically been an identification technique for chemically pure substances although, by using combined GC-MS, it has been possible to separate mixtures of solutes using gas chromatography followed by M S analysis of eluting peaks. This approach has proved successful for many groups of compounds, e.g. sugars (Aluyi & Drucker 1979), carboxylic methyl esters, enol ethers and acetals (Drucker & Jenkins 1989) and amines (Tavakkol et al. 1985). In these MS studies molecules were ionized either by electron impact, EI-MS, or by chemical ionization with a reagent gas, CI-MS. EI-MS produces a fingerprint of molecular fragments without necessarily revealing a molecular ion. CI-MS is a gentler technique; the molecular ion is usually seen and there is less collateral molecular fragmen-

F A B - M S 427

tation. A new form of ionization is by fast atom bombardment (FAB) (Heller et al. 1987, 1988; Platt et al. 1988) with xenon. This can ionize and desorb neutral solutes and desorb ionic solutes from the surface of a solvent of low volatility such as meta-nitrobenzyl alcohol. FAB-MS tends not necessarily to fragment molecules so that each solute can yield a peak of a specific mass to charge, m/z, value. In the case of anions, this is equivalent to mol wt - 1 (loss of a proton). FAB-MS offers a means of separating mixtures of solutes and simultaneously providing the anions’ m/z values. Thus it separates mixtures and provides far more standard peak descriptors than retention times or indices used in GC. Because FAB-MS selectively desorbs surface active polar lipids in lipid extracts, carboxylate and phospholipid anions are particularly readily detectable in negative ion FAB-MS (Heller et al. 1988). Carboxylic esters have proved especially valuable in chemotaxonomy (Drucker 198l), in spite of variability induced by variation in experimental parameters (Drucker & Owen 1973;Drucker 1976). Phospholipids have proved less useful for chemotaxonomic purposes. This has been due to the immense amount of effort required to obtain a good analysis-even for a single strain. One excellent study has used a reversed phase HPLC separation of phospholipid p-methoxybenzoates derived from a single strain, Escherichia coli K12 (Batley et al. 1982). The possibility of comprehensively screening several bacterial isolates for phospholipid composition in a single study was first realized by Heller et al. (1987) who profiled bacteria by FAB-MS of polar lipids. This theme was developed (Platt et al. 1988) by using linear regression analysis of resorption mass spectra for microbial chemotaxonomic purposes. Recently, Pramanik et al. (1990) have suggested a possible role for FAB-MS as a rapid means of bacterial detection and identification following examination of several actinomycetes and representatives of Escherichia coli, Proteus mirabilis and Pseudomonas aeruginosa. A limited amount of FAB-MS data on polar lipids is also available for E. coli, Pr. mirabilis and Enterobacter cloacae which were examined during an examination of the effects of experimental parameters (Aluyi et al. 1992). The aim of this study is to take a first step toward developing FAB-MS of polar lipids as a routine chemotaxonomic tool by examining a larger number of strains than has previously been studied. The intention is: (a) to discover which polar lipids can be detected; and (b) to determine their possible chemotaxonomic significance. MATERIALS AND METHODS Micro-organisms

The bacteria studied were : Acinetobacter calcoaceticus, Acinetobacter sp., Citrobacter freundii, Enterobacter cloacae

(2 strains), Escherichia coli (3 strains), Hafnia alvei, Klebsiella oxytoca, Klebsiella pneumoniae, Morganella morganii, Plesiomonas shigelloides, Proteus mirabilis (3 strains), Serratia liquefaciens and Serratia marcescens. All were kindly donated by the Hope Hospital Microbiology Department, Salford. T h e isolates were chosen to represent different genera within Enterobacteriaceae and for some species several strains were tested to obtain initial data on strain variability. The two Acinetobacter strains and Plesiomonas were included for comparative purposes. The identities of strains were confirmed by simultaneously culturing them on 5% horse blood Columbia agar and in 3 x 75 ml nutrient broth (Lab-m, Bury) at 37°C for 24 h. The agar growth was used for Rapid 20E (Biomerieux UK, Basingstoke) testing. Identities of all isolates were confirmed with 99% probability of current identification. Culture and harvestlng

The triplicate broth cultures were sampled after shaking at 37°C for 24 h for purity plates, then harvested by centrifugation. Pellets were twice washed with 2 ml distilled water before being lyophilized using an Edwards modulyo freezedrier (Edwards, Crawley).

Lipid extraction

Detergent-washed glassware was specially rinsed in methanol-chloroform (2 : 1, v/v) to ensure removal of hydrophobic contaminants. Gloves were worn to reduce the risk of contamination of samples. Pipetting was preferred to decanting. Lyophilized cells (10 mg amounts) were suspended in 2 ml methanokhloroform and dispersed using a vibratory mixer. They were occasionally subjected to remixing over a period of 4 h. After centrifugation at 3000 g for 20 min, lipid extract was removed; cells were reextracted as before then re-centrifuged and the second lipid extract was united with the first. The pooled extract was dried with a rotary evaporator then redissolved in 1 ml chloroform and washed using 1 ml water. T h e lower, chloroform, phase was retrieved, dried in a rotary evaporator and stored over silica in a vacuum desiccator. Mass spectrometry

The triplicate growth extracts were analysed separately after solution in m-nitrobenzyl alcohol (Aubignac 1990) and negative-ion FAB-MS spectra were obtained using a Kratos Concept 1s mass spectrometer. This provided a wealth of information; typically, over 500 peaks were obtained per sample with m/z values between 200 and 1000.

428 H . S . ALUYI E T A L .

18: I

Data analysls

The initial analyses of the chemical data were performed manually in order to detect possible problems. One problem encountered was that the usual output of the mass spectrometer data system shows m/z values as integers which may be produced by rounding up or down the measured values, e.g. the peak from PG(34 : l ) phosphatidyl glycerol has m/z 747.5 (Heller et al. 1987). T o confirm that the correct non-integral masses were being produced (based on integer atomic weights), non-integral masses were printed out and truncated. Accurate mass print out would also detect any acyl PG, e.g. acyl PG(30 : 0) which has a molecular weight of 962.0. Once tables of normalized data had been obtained for major peaks, it was possible to compare data for different strains, using the Pearson coefficient of linear correlation as a measure of association (Drucker et al. 1982). Pearson's coefficient specifically gives increased weight to larger peaks which are the ones most accurately measured by FAB-MS. Calculations were performed on the University of Manchester's Amdahl 5980 mainframe computer and the SPSS package of programs.

16:O

loor

1

0 200

250

300

350

400

450

550

500

600

PS (33: 0) PE(34: I )

I

PEW:^)/

800

700

9 00

Fig. 1 Negative-ion fast atom bombardment-mass spectra for polar lipids of Acinetobacter calcoaceticus

748(PS/33 : 0), moreover the peak at m/z 702(PE/33 : 1) was not detected and the peak at m/z 688(PE/32 : 1) was much less significant than for the enterobacteria examined. T o assess the quantitative differences evident between isolates, it was necessary to compute measures of association between strains. Results of representative comparisons are shown in Table 3.

RESULTS

Representative negative-ion FAB-MS spectra for polar lipids of Ac. calcoaceticus and Ser. marcescens are depicted in Figs 1 and 2, respectively. Major negative-ion FAB-MS peaks are tabulated for carboxylate anions in Table 1 and for phospholipid anions in Table 2. Data in the tables are expessed as normalized peak intensities with nominal m/z values. Peak identities were calculated using accurate m/z values from the M S output. Twenty-seven 'carboxylate anion' peaks were used together with thirty-seven 'phospholipid anions'. In the case of 'carboxylate anions', an even m/z value would indicate presence of nitrogen in certain 'unknown' peaks. It is of interest that the latter may form a homologous series separated by A m/z of 14 which is equivalent to a methylene group, i.e. peaks at 254, 268 and 282. The major carboxylate anions have the following m/z values (identities): 253(16 : l), 255(16 : 0), 267(cyc 17), 281(18 : 1). Uniquely, in the case of the two Acinetobacter strains, m/z 281 peak is the most intense; for the other organisms tested, the most intense peak is usually m/z 255 but occasionally m/z 253. Turning to phospholipid anions, these can be ascribed to homologous series of phosphatidylethanolamines, phosphatidylglycerols or phosphatidylserines. T h e major anionic phospholipids within the enterobacteria tested are m/z 688(PE/32 : I), 702(PE/33 : l), 714(PE/34 : 2), 716(PE/34 : l), 689(PG/30 : 2) and 703(PG/ 31 : 2). Acinetobacter differs from the enterobacteria examined having major peaks at m/z 717(PG/32 : 2) and m/z

DISCUSSION

Negative-ion FAB-MS of polar lipids yields data on possible fatty acyl substituents of anionic lipids and on free carboxylate which are less readily obtained by other means. Comparable data are published only for single strains of Ent. cloacae, E. coli and Pr. mirabilis (Aluyi et al. 1992), for E . coli and Pr. vulgaris (Heller et al. 1987) and for Pr. mirabills and two strains of E. coli (Pramanik et al. 1990). Dif-

I00

16:O

r

16:1

200

250

300

350

400

450

500

550

I00 -

PE/33: I P u 3 2 :I,

0

600

650

700

750

800

850

900

950

Fig. 2 Negative-ion fast atom bombardment-mass spectra for polar lipids of Serratia marcescens

FAB-MS 429

Table 1 Relative peak intensities of major carboxylate anions of some enterobacteria, detected in negative-ion FAB-MS spectra

5

*

5 -209 211 223 225 227 239 241 251 253 254 255 256 261 265 267 268 269 277 279 281 282 283 291 295 297 301 325

13 : 2 13:l 14 : 2 14 : 1 14:O 15:l 15:o

16:2 16 : I unknown 16 : 0 unknown unknown

1.9 2.6 2.2 2.5 4.4 0.0

3.5 2.0 9.5 2.4 15.1

3.5 2.3 17 : 2 2.1 (17: I/cycl7) 12.0 unknown 3.1 17 : 0 0.0 0.0 18:3 2.7 18:2 18 : I 10.7 3.0 unknown 3.0 18:O 19 : 3 3.3 (19: 1 cycl9) 3.7 2.6 19:O unknown 2.2 21 : o 0.0

1.4 1.0 0.0 1.7 1.9 1.1 2.5 2.7 0.0 1.4 2.2 2.5 3.8 4.5 6.0 2.9 4.5 3.2 3.2 4.5 4.9 2.2 1.6 2.5 16.7 8.6 15.5 2.7 1.5 2.2 23.2 16.5 19.3 4.0 3.2 3.5 0.0 0.9 0.0 1.2 1.9 2.5 6.8 11.5 16.2 2.7 3.2 1.1 2.2 0.0 0.2 0.9 2.0 1.2 1.6 1.8 1.3 11.6 14.6 4.6 2.5 3.3 0.9 2.2 4.6 1.4 1.2 1.9 1.1 1.9 3.7 0.0 0.0 0.0 0.0

0.0 0.0 0.0

1.6

1.7 1.1

1.2 2.3 1.o

0.7 2.1 14.1 2.7 18.0 3.3 0.0 2.0 14.0 2.6 1.7 1.2 2.6 11.3 2.3 2.1 1.4 1.4 1.6 0.0 3.8 0.4 2.3 5.6

1.8 1.4 0.8 1.3 1.5 1.4 1.6 1.8 1.4 I .o 0.9 1.9 2.0 1.1 0.9 1.2 I .7 2.4 1.3 1.6 I .7 4.7 2.8 2.6 2.8 1.0 1.2 2.4 2.7 1.1 2.9 4.1 1.7 1.4 1.2 2.3 1.5 2.3 2.3 3.3 18.4 7.8 20.0 18.2 23.3 1.0 4.0 3.4 4.1 3.1 17.2 21.2 17.7 21.8 20.6 8.0 4.1 3.2 4.1 3.8 2.1 0.5 0.9 0.0 0.0 1.3 2.8 0.7 1.3 1.3 8.4 19.0 5.0 6.2 0-4 1-2 2.0 4.3 1.2 1.5 2.4 2.4 1.7 2.1 2-1 0.8 1.4 1.3 0.9 1.0 2.0 3.0 1.4 2.0 2-6 14.2 4.1 7.4 13.8 14.1 2.9 0.9 1.5 2.9 3.1 1.6 1.3 2.3 4.0 1.5 0.0 1.3 2.0 2.7 1.1 1.1 1.1 1.2 1.1 1.1 0.9 1.4 0.9 1.0 0.5 2.1 4.1 0.5 1.4 1.5 1.3 2.9 3.3 2.0 0.5

ferences between studies tend to be quantitative except that, in the present study, Pr. mirabilis lacked any large peaks at m/z 885,899 and 913 due to an unknown lipid. Although peaks of intensity