Capillary Gas Chromatographic Analysis of Carbohydrates ofLegionella pneumophila and Other Members of the Family
LegionellaceaeALVIN FOX,'* PAULINE Y. LAU,2 ARNOLD BROWN,1'3'4 STEPHEN L. MORGAN,2 Z.-T. ZHU,2t MICHAEL
LEMA,4 AND MICHAEL D. WALLA2Department of Microbiology and Immunology, School of Medicine,' Department of Chemistry,2 and Department of
Medicine,3 University of South Carolina, Columbia, South Carolina 29208, and Research Service, William Jennings BryanDorn Veterans Administration Hospital, Columbia, South Carolina 292014
Received 25 August 1983/Accepted 28 November 1983
Legionella pneumophila, the causative agent of Legionnaires disease, and related organisms havepreviously been characterized primarily by conventional bacteriological methods, DNA-DNA hybridiza-tion, antigenic analysis, and fatty acid analysis. By capillary gas chromatographic analysis for carbohy-drates, we have shown that muramic acid and glucosamine, characteristic markers of bacterial cell walls,were present in samples of L. pneumophila and a group of legionella-like organisms. Some bacterial samplescontained two unusual isomeric aminodideoxyhexoses (Xl and X2). L. pneumophila was characterized bythe absence of fucose and the presence of the peak Xl. Tatlockia micdadei (Legionella micdadei) wasdistinguishable by the presence of large amounts of rhamnose and fucose and by the absence of Xl and X2.Fluoribacter strains were much more variable in their carbohydrate composition. These data suggest that, inaddition to other reported techniques, carbohydrate profiling by capillary gas chromatography can be avaluable diagnostic method in reference microbiology laboratories for differentiating members of the familyLegionellaceae.
Legionella pneumophila (3) and other members of thefamily Legionellaceae are similar in rather general pheno-typic characteristics (2, 3, 7, 14, 15, 20-22). However, thereare simple tests, including colony color on dye-containingmedia, colony fluorescence, pigment production, gelatinstarch and hippurate hydrolysis, and serology, which candistinguish between various members of this family (12, 29),and important differences have been noted by studies ofDNA homology (2, 5, 7, 12, 15, 21, 22), DNA base composi-tion (5, 12), fatty acid analysis (14, 20, 23-25), ubiquinoneprofiles (17), analysis of soluble peptide patterns by poly-acrylamide gel electrophoresis (19), and antigenic analysis(8).
Analysis of the carbohydrates of microorganisms by gaschromatography (GC) and GC-mass spectrometry (MS) hasbeen used for the characterization and classification ofseveral groups of bacteria such as the streptococci (27) andthe bacteria often associated with meningitis (4). We previ-ously reported improvements in carbohydrate analysis byusing a modified alditol acetate derivatization method fol-lowed by capillary GC or GC-MS (10, 11, 16). This tech-nique, based on high-resolution capillary chromatography, isa simple and reproducible procedure which we have em-ployed routinely for the analysis of bacterial components. Inthis paper we describe the carbohydrate profiles of L.pneumophila and a group of legionella-like organisms. Theprocedure is sufficiently easy to perform that it could beadopted as a routine confirmatory technique in referencemicrobiology laboratories for the identification of theseorganisms.
* Corresponding author.t Present address: Department of Chemistry, Shanxi University,
Taiyuan, People's Republic of China.
MATERIALS AND METHODS
Bacterial strains, media, and preparation of bacterial cells.The bacterial strains used in this study and their sources arelisted in Table 1. After incubation for 3 days on bufferedcharcoal-yeast extract agar (26) in air at 37°C, the confluentbacterial growth was scraped from four to five agar plateswith a bent Pasteur pipette and was suspended in distilledwater. The bacterial cells were killed by heating them inflowing steam at 100°C for 1 h and then were centrifuged atroom temperature for 10 min at 10,000 rpm with a JA-20rotor in a Beckman J21B centrifuge (Beckman Instruments,Inc., Palo Alto, Calif.). The heat-killed cells were resus-pended in distilled water and then recentrifuged. This wash-ing step was performed three times. The pellet was thenlyophilized and stored in a desiccator at 4°C.
Chemicals. L-Rhamnose, L-fucose, D-xylose, D-mannose,D-galactose, and D-glucose standards were obtained fromSupelco, Bellefonte, Pa. 2-Deoxy-D-ribose, D-ribose, mu-ramic acid, D-glucosamine hydrochloride, D-galactosaminehydrochloride, and methylglucamine were obtained fromSigma Chemical Co., St. Louis, Mo. D-Glycero-L-manno-heptose and Pasteurella multocida lipopolysaccharide (LPS)were gifts from Paul Rebers, National Animal DiseaseCenter, Ames, Iowa. Salmonella typhimurium LPS waspurchased from Difco Laboratories, Detroit, Mich.The following derivatization reagents and chromatograph-
ic solvents were glass distilled: acetic anhydride (AlltechAssociates Inc., Applied Sciences Div., State College, Pa.),ultrex-grade glacial acetic acid (J. T. Baker Chemical Co.,Phillipsburg, N.J.), chloroform and methanol (Burdick &Jackson Laboratories, Muskegon, Mich.). N,N-dioctylme-thylamine was purchased from ICN Pharmaceuticals, Inc.,Plainview, N.Y. Reagent-grade sulfuric acid and sodium
F. gormanii (L. gormanii) LS-13 PUH-CDCa CDC, Centers for Disease Control, Atlanta, Ga.; PVA, Veter-
ans Administration Medical Center, Pittsburgh, Pa.; GSPH, R. Yee,Graduate School of Public Health, University of Pittsburgh, Pitts-burgh; PUH, A. W. Pasculle, Presbyterian-University Hospital,Pittsburgh, Pa.; WVA, P. Edelstein, Wadsworth Veterans Adminis-tration Medical Center, Los Angeles, Calif.
borohydride were obtained from Fisher Scientific Co.,(Pittsburgh, Pa.). Clin Elut hydrophilic extraction columnsand Bond Elut hydrophobic extraction columns were pur-chased from Analytichem International, Lawndale, Calif.All glassware was washed separately with acid and chloro-form before use.
TABLE 2. Relative peak areas obtained for equal amounts ofneutral and amino sugar standards derivatized and
Carbohydrate analysis. The neutral and amino sugar con-tents of bacteria and bacterial components were determinedby a previously described modification of the alditol acetateprocedure (10) which permits the simultaneous preparationof multiple samples and decreases background peaks derivedfrom the sample, side reactions, and impure reagents. Onemilligram of each sample was hydrolyzed in 0.5 ml of 2 Nsulfuric acid under vacuum for 3 h, the acid was neutralizedwith N,N-dioctylmethylamine, and the sample was thenreduced with sodium borohydride and acetylated with aceticanhydride. Hydrophobic and hydrophilic columns wereused, repectively, in prederivatization and postderivatiza-tion cleanup steps. The samples were analyzed with aPerkin-Elmer Sigma 3B gas chromatograph (The Perkin-Elmer Corp., Norwalk, Conn.) fitted with a capillary inletsystem, an SP-2330 capillary column prepared in our labora-tory (16), and a flame ionization detector. A sample splitratio of ca. 18:1 was used. Other chromatographic conditionsincluded the following: injector temperature, 240°C; detectortemperature, 295°C; initial oven temperature, 100°C; initialhold, 0.5 min; and temperature program, 30°C/min to a finaltemperature of 245°C. The column helium flow rate was ca. 2ml/min. Peak areas were determined with a Hewlett-Packardmodel 3390A integrator (Hewlett-Packard, Avondale, Pa.).A sample containing external standards was analyzed
TABLE 4. Carbohydrate composition of LPS isolates from S.typhimurium and P. multocida
b Percent sample dry weight of abequose was an estimate calcu-lated by assuming that detector response factors for abequose anddeoxyribose were identical since an abequose standard was notavailable.
neutral and amino sugars, respectively. All bacterial sampleswere analyzed in triplicate, with the internal standardsomitted from the third sample so that the absence of thesesugars as natural components could be confirmed. Theamount of each sugar component (as a percentage of thesample dry weight), A, was calculated by the followingformula: A = [(H1 x S2 x Al)/(H2 x S1 x A2)] x 100, whereH1 is the area of the component peak in the sample chro-matogram, H2 is the area of the component peak in theexternal standard chromatogram, S1 is the area of the
176
B
6
C
BI
17
I I J18Ill lSi:ii II
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14TIME (minutes)
FIG. 1. Capillary column chromatograms of the alditol acetatederivatives of (A) a mixture of neutral and amino sugars, (B) S.typhimurium LPS hydrolysate and (C) P. multocida LPS hydroly-sate. See Table 2 for peak identification. When the chromatogramsare compared, it should be noted that there was 25 times the amountof internal standards (arabinose and methylglucamine) in sample Cas compared with sample B.
concurrently with each batch of bacterial samples. Table 2presents the relative response factors for equal amounts ofneutral and amino sugars carried through the entire derivati-zation and GC procedure. Although the specific capillarycolumn used was chosen to minimize selective degradationof amino sugars compared with neutral sugars (16), thecolumn produced considerably smaller peak areas for aminosugars compared with neutral sugars. Visual comparison ofrelative peak areas in chromatograms without considerationof these response factors can lead to misinterpretation of therelative amounts of carbohydrates present in the sample.
Arabinose and methylglucamine were added to bacterialand external standard samples as internal standards for
17
118
17
I 18
I I I I I I I I I I I I I I I0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
TIME (minutes)
FIG. 2. Capillary column chromatograms of whole bacterial cellhydrolysates. (A) L. pneumophila; (B) T. micdadei; (C) F. boze-manae; and (D) F. dumoffii. See Table 2 for peak identification.When chromatograms are compared, it should be noted there wasthree times the amount of internal standards in samples A and B ascompared with samples C and D. Glucose was also present in traceamounts in these organisms as confirmed by MS.
internal standard peak in the sample chromatogram, S2 is thearea of the internal standard peak in the external standardchromatogram, Al is the amount of the component in theexternal standard (in micrograms), and A2 is the total amountof the sample (in micrograms). This single-level calibrationproved to be satisfactory based upon previous experimentswhich demonstrated a linear calibration curve and zerointercept (10). The use of internal standards to correct forcomponent loss during sample processing and GC analysis isbased on the assumption that any loss of sample componentsor external standards is identical to the loss of the internalstandards in the analytical sample or external standardsample, respectively.The identity of carbohydrate components was confirmed
by capillary GC-MS in which a Finnigan model 4021 system(Finnigan Corp., San Jose, Calif.) was used. The chromato-graphic column used for GC-MS was a flexible, fused silicacolumn which was static coated with SE-52 and cross-linkedin our laboratory. The SE-52 column was installed in the gaschromatograph and threaded through the GC-MS transferline so that the column effluent passed directly into the ionsource of the mass spectrometer. The GC-MS chromato-graphic conditions were otherwise similar to those used forGC.
RESULTS
Table 3 lists the retention times relative to arabinose of theneutral and amino sugars present in standards or in samplesemployed in this study. Figure 1 compares the chromato-grams of a standard mixture of sugars (Fig. 1A) with those ofhydrolysates of LPS isolated from S. typhimurium (Fig. 1B)and from P. multocida (Fig. 1C). These LPS specimens wereanalyzed to determine whether the method could detectheptose and ketodeoxyotonic acid (KDO) (6, 28) in bacterialsamples since neither of these sugars was detected in strainsof L. pneumophila or the legionella-like organisms analyzed.L-Glycero-D-mannoheptose was identified in S. typhimuriumLPS and both L-glycero-D-mannoheptose and D-glycero-D-mannoheptose were identified in P. multocida LPS bycapillary GC-MS. However, KDO was not detected in eitherLPS preparation, probably due to its destruction by the highconcentration of mineral acid used for hydrolysis (28). Table4 summarizes the composition of LPS isolated from S.typhimurium and P. multocida.
L. pneumophila (nine strains) and Fluoribacter dumoffii
(Legionella dumoffli) isolate NY-23 contained a peak (X1)that differed slightly in retention time from L-glycero-D-mannoheptose and D-glycero-D-mannoheptose. Other le-gionella-like organisms contained both Xl and a second peak(X2) which also did not coelute with either of these heptoses.Peaks Xl and X2 were not identified in any Tatlockiamicdadei (Legionella micdadei) strain.By electron impact and methane chemical ionization MS,
we have identified the peaks labeled Xi and X2 as alditolacetates of aminodideoxyhexoses. The chemical ionizationmass spectrum of peak Xi indicates that the molecularweight of this compound is 375, based on the interpretationof an observed ion at mass 376 as the addition of a proton (M+ 1 peak) and an ion at mass 404 as the addition of C2H5 (M+ 29 peak) from the methane reagent gas. The losses ofacetic acid (mass, 60) and ketene (mass, 42) from the (M + 1)ion are also observed at masses 316 and 334, respectively.The electron impact and methane chemical ionization spec-tra of component X2 are very similar to their Xl counter-parts, indicating that X1 and X2 are isomers. In addition tothe identification of these two rather unusual sugars, match-ing retention times and mass spectra, compared with stan-dards, were obtained for the following neutral and aminosugars found in the hydrolysates of members of the Legion-ellaceae: rhamnose, fucose, ribose, mannose, glucose, mu-ramic acid, and glucosamine.
Figure 2 shows typical chromatograms obtained for L.pneumophila, T. micdadei, Fluoribacter bozemanae (Le-gionella bozemanii), and F. dumoffii. The chromatogramsprovide characteristic fingerprints. Tables 5 to 7 summarizeresults obtained with representative strains of each genus ofthe Legionellaceae. L. pneumophila Philadelphia-1 was ih-cluded as a control in every batch of samples analyzed toallow comparisons between experiments. L. pneumophila(Fig. 2A) contained rhamnose, ribose, mannose, and traceamounts of glucose and was characterized by the absence offucose and by the presence of Xl. T. micdadei (Fig. 2B)contained these sugars, but rhamnose was present in signifi-cantly greater amounts. Large amounts of fucose were alsopresent, whereas neither sugar Xl nor sugar X2 could bedetected. Fluoribacter strains were much more variable intheir sugar compositiot. Four of five strains containedmoderate amounts of rhamnose, and three of five strainscontained moderate amounts of fucose. The presence of Xland X2 was noted in four of the five strains tested. StrainNY-23 contained only Xl. Figures 2C and D, showing F.
TABLE 5. Carbohydrate composition of L. pneumophila% Sample dry wta
Mean ± SD 0.13 ± 0.03 0.18 ± 0.08 0.19 ± 0.07 0.54 ± 0.19 0.30 + 0.15 0.47 ± 0.18a Rha, Rhamnose; Rib, ribose; Man, mannose; Glu, glucose; Mur, muramic acid; Gin, glucosamine.b Percent sample dry weights of Xi were estimates based on relative areas of Xi and muramic acid in sample chromatograms.c tr, Levels near background noise (less than 0.1% sample dry weight).d ND, Not detectable.
bozemanae and F. dumoffii, respectively, illustrate the vari-ability noted among Fluoribacter strains. Muramic acid andglucosamine were present in moderate amounts in all iso-lates and were presumably derived primarily from peptido-glycan.
DISCUSSION
Genetic and biochemical data (12), ubiquinone patterns(17), fatty acid analysis (23-25), peptide profiles (19), andantigenic characterization (8) can be used to separate L.pneumophila and the legionella-like organisms into at leastthree groups. Table 8 provides a summary of the availablegenetic and biochemical data which allow one to distinguishamong these organisms.Most classical methods of bacterial characterization con-
sist of groups or series of individual tests each of whichmeasures one discriminating feature. Recent chemotaxo-nomic approaches have involved techniques which cananalyze the multiple structural components of bacterial cellssuch as peptides by polyacrylamide gel electrophoresis andcarbohydrate or fatty acid fingerprinting by GC. Theseinstrumental methods provide measurements of a number ofdiscriminating features simultaneously. Although fatty acidprofiling has been used for the characterization of membersof the Legionellaceae, detailed analysis of the sugar compo-sition of this group of agents has not previously beenreported.The alditol acetate procedure used in this work was
developed as a specific method for the analysis of carbohy-drates in bacterial samples (10). The combination of hydro-phobic and hydrophilic extraction columns, respectively,before and after the acetylation step acts as a selectivecleanup step. The chromatograms obtained by using thismethod are relatively clear of extraneous peaks. For this
reason, relatively low amounts of sugars in bacterial samplesmay be identified. However, the identification of sugarcomponents should be confirmed not only by matchingretention times of standards but also, if possible, by MS. Amore detailed description of the capillary GC-MS identifica-tion of carbohydrates from legionellae will be publishedelsewhere (M. D. Walla, P. Y. Lau, S. L. Morgan, A. Fox,and A. Brown, submitted for publication).Our analyses show that L. pneumophila, T. micdadei, and
Fluoribacter species can be distinguished from one anotherby their carbohydrate profiles. L. pneumophila was charac-terized by the absence of fucose and by the presence of anaminodideoxyhexose. Fluoribacter strains were much morevariable in their sugar composition. Four of five strainscontained moderate amounts of rhamnose, and three of fivestrains contained moderate amounts offucose. The presenceof two aminodideoxyhexose isomers was noted in four of thefive strains tested. Strain NY-23 contained only one isomer.T. midadei was distinguished by the presence of largeamounts of both fucose and rhamnose and by the absence ofeither aminodideoxyhexose.
Ribose, which was generally present in strains of Legion-ellaceae, presumably originated largely from RNA, althoughsome might have been a component of another bacterialsubstance such as LPS. Deoxyribose, a component of DNA,and KDO, a component of gram-negative LPS, were notdetected in any strains of Legionellaceae studied; however,these deoxysugars would not have survived the hydrolysisconditions employed (6, 27, 28), and KDO was, in fact, notdetected in the LPS of other gram-negative bacteria ana-lyzed. Heptose, usually present in bacterial LPS (6, 28), wasalso not detected in samples of legionellae, although it wasdetected in the LPS of other gram-negative bacteria. Thepresence of LPS in L. pneumophila has been reported (9,30); however, heptoses are either not present in the LPS of
TABLE 7. Carbohydrate composition of Fluoribacter species
Species % Samples dry wtaRha Fuc Rib Man Glu Xlb X2 Mur Gln
a For abbreviations, see Table 6, footnote a.b Percent sample dry weights of Xl and X2 were estimates based on relative heights of Xl and X2 and muramic acid in sample
chromatograms.c See Table 5, footnote c.d ND, Not detectable.
DNA guanine-plus-cytosine content 38.8 44.3 40.4(molM)
Colony color on dye-containing mediad White to slightly Blue grey Pastel greengreen
Colony fluorescence (blue-white) +
Brown pigment from tyrosine + +
Catalase + + +
Starch hydrolysis + +
Gelatinase + +
Hippurate hydrolysise +
Ubiquinonef Q12/11/13 Q10/11/12/13 Q9/10/11/12
Major fatty acids (%)g12-Methyltetradecanoic 11-20 39-40 24-3114-Methylpentadecanoic 32-42 10-11 14-2014-Methyl-cis-9-pentadecenoic 2-13 tre trcis-9-Hexadecenoic 13-15 9-10 11-1614-Methylhexadecanoic 5-11 22-25 12-24a Range of homology at 60 to 65°C between strains of a species (intraspecific) or between strains representing the different species or genera
the Legionellaceae or are present in quantities below thedetection limits of our methodology.Muramic acid is a unique amino sugar found in nature only
in peptidoglycan, the glycan backbone of which is a polymerof alternating units of muramic acid and glucosamine. Wehave shown the presence of muramic acid in species ofLegionella, Tatlockia, and Fluoribacter. Peptidoglycan hasbeen isolated from L. pneumophila and characterized byamino acid analysis (1, 18), but the presence of peptidoglycanin species of Tatlockia and Fluoribacter has not been previ-ously reported. Interestingly, the glucosamine/muramic acidratio was significantly higher than 1:1 in most isolates, thuspossibly indicating the presence of an additional glucosam-ine-containing macromolecule.The data presented in this article suggest that, in addition
to other reported techniques, carbohydrate profiling bycapillary GC can be a valuable method for distinguishingamong members of the Legionellaceae. The method de-scribed is sufficiently simple to be routinely used in refer-ence microbiology laboratories.
ACKNOWLEDGMENTSThis work was supported in part by Public Health (NIH) Service
Biomedical Support grant 2 S07 RRO160-07 (University of South
Carolina) and by Public Health Service grants GM 27135 and EY04715 from the National Institutes of Health and by the VeteransAdministration Medical Research Service.The assistance of Matthew Przybyciel (Department of Chemistry,
University of South Carolina) with preparation of capillary columnsis greatly appreciated.
LITERATURE CITED1. Amano, K.-I., and J. C. Williams. 1983. Peptidoglycan of
Legionella pneumophila: apparent resistance to lysozyme hy-drolysis correlates with a high degree of peptide cross-linking. J.Bacteriol. 153:520-526.
2. Brenner, D. J., A. G. Steigerwalt, G. W. Gorman, R. E.Weaver, J. C. Feeley, L. G. Cordes, H. W. Wilkinson, C. Pat-ton, B. M. Thomason, and K. R. Lewallen-Sasseville. 1980.Legionella bozemanii sp. nov. and Legionella dumoffii sp. nov.:classification of two additional species of Legionella associatedwith human pneumonia. Curr. Microbiol. 4:111-116.
3. Brenner, D. J., A. G. Steigerwalt, and J. E. McDade. 1979.Classification of the Legionnaires' disease bacterium: Legion-ella pneumophila, genus novum, species nova, of the familyLegionellaceae familia nova. Ann. Intern. Med. 90:656-658.
4. Brice, J. L., T. G. Tornabene, and F. M. LaForce. 1979.Diagnosis of bacterial meningitis by gas liquid chromatography.I. Chemotyping studies of Streptococcus pneumoniae, Hemo-philus influenzae, Neisseria meningitidis, Staphylococcus aur-
eus and Escherichia coli. J. Infect. Dis. 140:443-452.5. Brown, A., G. M. Garrity, and R. M. Vickers. 1981. Fluori-
bacter dumoffii (Brenner et al.) comb. nov. and Fluoribactergormanii (Morris et al.) comb. nov. Int. J. Syst. Bacteriol.31:111-115.
6. Bryn, K.-I., and E. Jantzen. 1982. Analysis of lipopolysaccha-rides by methanolysis, trifluoroacetylation and gas chromatog-raphy on a fused silica column. J. Chromatogr. 240:405-413.
7. Cherry, W. B., G. W. Gorman, L. H. Orrison, C. W. Moss,A. G. Steigerwalt, H. W. Wilkinson, S. E. Johnson, R. M. Mc-Kinney, and D. J. Brenner. 1982. Legionella jordanis: a newspecies of Legionella isolated from water and sewage. J. Clin.Microbiol. 15:290-297.
8. Collins, M. T., F. Espersen, N. H4iby, S.-N. Cho, A. Friis-Mpller, and J. S. Reif. 1983. Cross-reactions between Legion-ella pneumophila (serogroup 1) and twenty-eight other bacterialspecies, including other members of the family Legionellaceae.Infect. Immun. 39:1441-1456.
9. Flesher, A. R., S. Ito, B. J. Mansheim, and D. L. Kasper. 1979.The cell envelope of the Legionnaires' disease bacterium.Morphologic and biochemical characteristics. Ann. Intern.Med. 90:628-630.
10. Fox, A., J. R. Hudson, S. L. Morgan, Z.-T Zhu, and P. Lau.1983. Capillary gas chromatographic analysis of alditol acetatesof neutral and amino sugars in bacterial cell walls. J. Chroma-togr. 256:429-438.
11. Fox, A., J. H. Schwab, and T. Cochran. 1980. Muramic aciddetection in mammalian tissues by gas-liquid chromatography-mass spectrometry. Infect. Immun. 29:526-531.
12. Garrity, G. M., A. Brown, and R. M. Vickers. 1980. Tatlockiaand Fluoribacter: two new genera of organisms resemblingLegionella pneumophila. Int. J. Syst. Bacteriol. 30:609-614.
13. Hebert, G. A. 1981. Hippurate hydrolysis by Legionella pneu-mophila. J. Clin. Microbiol. 13:240-242.
14. Hebert, G. A., C. W. Moss, L. K. McDougal, F. M. Bozeman,R. M. McKinney, and D. J. Brenner. 1980. The Rickettsia-likeorganisms TATLOCK (1943) and HEBA (1959): bacteria pheno-typically similar to but genetically distinct from Legionellapneumophila and the WIGA bacterium. Ann. Intern. Med.92:45-52.
15. Hebert, G. A., A. G. Steigerwalt, and D. J. Brenner. 1980.Legionella micdadei species nova: classification of a thirdspecies of Legionella associated with human pneumonia. Curr.Microbiol. 3:255-257.
16. Hudson, J. R., S. L. Morgan, and A. Fox. 1982. High-resolutionglass capillary columns for the separation of alditol acetates ofneutral and amino sugars. J. High Resolut. Chromatogr. Chro-matogr. Commun. 5:285-290.
17. Karr, D. E., W. F. Bibb, and C. W. Moss. 1982. Isoprenoidquinones of the genus Legionella. J. Clin. Microbiol. 15:1044-
1048.18. Keel, J. A., W. R. Finnerty, and J. C. Feeley. 1979. Fine
structure of the Legionnaires' bacterium. In vitro and in vivostudies of four isolates. Ann. Intern. Med. 90:652-655.
19. Lema, M., and A. Brown. 1983. Electrophoretic characteriza-tion of soluble protein extracts of Legionella pneumophila andother members of the family Legionellaceae. J. Clin. Microbiol.17:1132-1140.
20. Lewallen, K. R., R. M. McKinney, D. J. Brenner, C. W. Moss,D. H. Dail, B. M. Thomason, and R. A. Bright. 1979. A newlyidentified bacterium phenotypically resembling, but geneticallydistinct from Legionella pneumophila: an isolate in a case ofpneumonia. Ann. Intern. Med. 91:831-834.
21. McKinney, R. M., R. K. Porschen, P. H. Edelstein, M. L.Bissett, P. P. Harris, S. P. Bondell, A. G. Steigerwalt, R. E.Weaver, M. E. Ein, D. S. Lindguist, R. S. Kops, and D. J.Brenner. 1981. Legionella longbeachae species nova. Anotheretiologic agent of human pneumonia. Ann. Intern. Med. 94:739-743.
22. Morris, G. K., A. G. Steigerwalt, J. C. Feeley, E. S. Wong,W. T. Martin, C. M. Patton, and D. J. Brenner. 1980. Legion-ella gormanii sp. nov. J. Clin. Microbiol. 12:718-721.
23. Moss, C. W., and S. B. Dees. 1979. Cellular fatty acid composi-tion of WIGA, a rickettsia-like agent similar to the Legionnairesdisease bacterium. J. Clin. Microbiol. 10:390-391.
24. Moss, C. W., D. E. Karr, and S. B. Dees. 1981. Cellular fattyacid composition of Legionella longbeachae sp. nov. J. Clin.Microbiol. 14:692-694.
25. Moss, C. W., R. E. Weaver, S. B. Dees, and W. B. Cherry. 1977.Cellular fatty acid composition of isolates from Legionnairesdisease. J. Clin. Microbiol. 6:140-143.
26. Pasculle, A. W., J. C. Feeley, R. J. Gibson, L. G. Cordes, R. L.Myerowitz, C. M. Patton, G. W. Gorman, C. L. Carmack,J. W. Ezzeli, and J. N. Dowling. 1980. Pittsburgh pneumoniaagent: direct isolation from human lung tissue. J. Infect. Dis.141:727-732.
27. Pritchard, D. G., J. E. Coligan, S. E. Speed, and B. M. Gray.1981. Carbohydrate fingerprints of streptococcal cells. J. Clin.Microbiol. 13:89-92.
28. Seid, R. C., H. Schneider, S. Bondarew, and R. A. Boykins.1982. Quantitation of L-glycero D-mannoheptose and 3-deoxy-D-mannooctulosonic acid in rough core lipopolysaccharides bypartition chromatography. Anal. Biochem. 124:320-326.
29. Vickers, R. M., A. Brown, and G. M. Garrity. 1981. Dye-containing buffered charcoal-yeast extract medium for the dif-ferentiation of members of the family Legionellaceae. J. Clin.Microbiol. 13:380-382.
30. Wong, K. H., C. W. Moss, D. H. Hochstein, R. J. Arko, andW. 0. Schalla. 1979. Endotoxicity of the Legionnaires' diseasebacterium. Ann. Intern. Med. 90:624-627.
