CLINICAL MICROBIOLOGY REVIEWS, Oct. 1991, p. 422-438 Vol. 4, No. 40893-8512/91/040422-17$02.00/0Copyright © 1991, American Society for Microbiology

Applications of Cellular Fatty Acid AnalysisDAVID F. WELCH

Department of Pediatrics, University of Oklahoma Health Sciences Center, and Clinical Microbiology Laboratories,Oklahoma Medical Center, P.O. Box 26307, Oklahoma City, Oklahoma 73126

INTRODUCTION ................................................................... 422MOLECULAR BASIS OF CFA ANALYSIS ................................................................... 423Nomenclature of Fatty Acids .................................................................. 423Lipid Content of Microbes ................................................................... 423

INSTRUMENTATION AND DATA PROCESSING ................................................................... 424METHODS ................................................................... 424FEASIBILITY OF WIDESPREAD USE OF GLC IN CLINICAL LABORATORIES ..........................425

Practical Considerations ................................................................... 425Comparison with DNA Probes ................................................................... 426

CHARACTERIZATION OF BACTERIA ................................................................... 427Identification to Genus or Species ................................................................... 427

Staphylococcus spp....................................................................427Streptococcus and Enterococcus spp.................................................................. 427Gram-positive bacilli ................................................................... 427Mycobacterium and Nocardia spp................................................................... 428Enterobacteriaceae ................................................................... 428Vibrionaceae............................................................................... 428Neisseria and Moraxella spp...................................................................428Campylobacter and Helicobacter spp..................................................................429Pseudomonas spp. and nonfermenters ................................................................... 429Legionella spp....................................................................430Miscellaneous bacteria................................................................... 430Spirochetes ................................................................... 430Anaerobes .................................................................. 430

Identification to Subspecies and Epidemiologic Typing ...............................................................431Campylobacterjejuni ................................................................... 431Pseudomonas cepacia and Staphylococcus epidermidis ..............................................................431

CHARACTERIZATION OF OTHER MICROBIAL SPECIES .......................................................431Rickettsia spp.................................................................431Chlamydia spp.................................................................432Fungi................................................................ 432Blood Parasites ................................................................ 432Mycoplasmas and Viruses ................................................................ 432Unclassified Organisms ................................................................ 433

DETECTING MARKERS OF INFECTION IN CLINICAL MATERIAL..........................................433MEASURING ANTIMICROBIAL RESISTANCE ................................................................ 433CONCLUSION ................................................................ 434ACKNOWLEDGMENTS ................................................................ 434REFERENCES ................................................................ 434

INTRODUCTION a probe. The main requirement of CFA analysis is properinstrumentation. Technological development by industry

Medically important microorganisms can be identified in involved in the manufacture of chromatographs and capillarymany ways. Conventional methods rely on the expression of columns has made it possible for microbiologists to usecertain properties that are usually mediated directly by gas-liquid chromatography (GLC) more easily. Improvedenzyme activity. Extension of this approach to include microelectronics and computer-aided interpretation of datanumerical identification (120) or automated systems to ana- have also facilitated numerous applications of GLC in clin-lyze results often strengthens conclusions. Immunodiagnos- ical microbiology.tic and nucleotide hybridization techniques have improved In practice, whole cells of bacteria or yeasts are treated tosensitivity, specificity, precision, and ease of testing. Chemo- cause release of their fatty acids, which are converted to ataxonomy is also precise and can result in the definition of methyl ester derivative and then analyzed by GLC. Earlyhighly discriminatory properties. Cellular fatty acid (CFA) attempts to apply CFA analysis to bacterial identificationanalysis falls into this category (146). In contrast to antigen were made in the 1950s (62). In 1963, Abel et al. (1) were thedetection and hybridization, this technique does not require first to present evidence suggesting that CFA analysis by

422

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

APPLICATIONS OF CELLULAR FATTY ACID ANALYSIS 423

GLC could successfully identify bacteria. Other early stud-ies were aimed at CFA analysis from the standpoint ofbacterial virulence factors. It was recognized that rough(avirulent) strains of Vibrio cholerae lacked branched-chainCFAs but that such chains were present in smooth (virulent)strains (14). An alteration of the lipid A composition ofSalmonella spp. was reported to correlate with a change invirulence (97). Abel et al. (1) identified different CFA pat-terns within various members of the family Enterobac-teriaceae and in some gram-positive bacteria, but no attemptwas made to identify specifically the CFAs that character-ized each genus or species that they tested. Moreover, theanalyses required numerous steps, and the apparatus usedfor extraction and esterification was cumbersome and tech-nically complicated by today's standards. Nevertheless,their report established the potential usefulness of CFAanalysis and established the foundation of further investiga-tion.

This review will discuss the development of CFA analysispertaining to microbial identification from the mid-1960s tothe present. For the purpose of this review, CFAs will bedefined as the components of any cellular lipid that have acarbon chain length of 9 to 20 atoms. This includes themajority of fatty acids located in the cell membrane asglycolipid and phospholipid. It includes the fatty acid con-stituents of lipopolysaccharides (LPS) but does not includethe long-chain (24 to 90 carbon atoms) mycolic acids or theisoprenoid quinones. Understanding the molecular basis anddevelopment of the analytic instrumentation is central to anappreciation of the potential use and limitations of thistechnology. The CFA composition of most microbes hasbeen studied, but data to predict the success of applicationsin diagnostic microbiology are incomplete in many casesbecause of the limited scope of the studies or use of olderinstrumentation. Thus, a fair degree of caution must beexercised in examining the earlier literature and comparingresults of different studies. In addition to changes in tech-nique with respect to chromatographic equipment, the CFAcomposition of cells varies, at least quantitatively, accordingto culture conditions (46, 72, 90). Nonetheless, the existinginformation from these older studies interpreted in light ofwhat has been reported recently can be used to formreasonable conclusions about the role of CFA analysis inclinical microbiology.

MOLECULAR BASIS OF CFA ANALYSIS

The source of fatty acids in microbial cells is lipid,primarily that of the cell membranes (e.g., phospholipid) orthe lipid A component of LPS in gram-negative bacteria andlipoteichoic acid in gram-positive bacteria. In addition tophospholipid, fungi synthesize sterols (e.g., ergosterol) as amajor lipid. The fatty acid content of all lipids is determinedby the particular type of biosynthetic pathway of a givenspecies. The process is initiated by synthesis of the coen-zyme A ester of a fatty acid, with a molecule of acetylcoenzyme A used as primer. Most bacteria synthesize fattyacids with chain lengths of 10 to 19 carbon atoms, and themost prevalent fatty acids are those with 16 or 18 carbonatoms. The 16-carbon saturated CFA hexadecanoic acid, inparticular, is highly conserved among prokaryotes. Thevariable properties that make an organism's CFA composi-tion distinctive include quantitative differences in CFA con-tent and the presence of other CFAs, of which more than 100have been identified (132). The usual profile features 5 to 15CFAs in significant amounts. Organisms in genera with

smaller genomes, such as Rochalimaea spp., tend to havefew fatty acids, while other eubacteria, such as Xanthomo-nas spp., have more than 20 fatty acids. Bacteria containsome CFAs that are unique, i.e., not generally found ineukaryotic cells. Branched-chain and cyclopropane-contain-ing CFAs characterize many gram-positive and gram-nega-tive bacteria, respectively, but are not found in fungi (or inhumans). Conversely, the polyunsaturated fatty acids foundin higher organisms tend to be absent in aerobic bacteria.Gram-negative bacteria generally have a higher proportionof saturated and monounsaturated CFAs with an even-numbered chain of carbon atoms than gram-positive bacte-ria. The latter, represented by Bacillus spp. and staphylo-cocci, tend to have saturated, branched-chain CFAs with anodd-numbered chain of carbon atoms and lower amounts ofstraight-chain, saturated CFAs. Coryneforms and strepto-cocci have straight-chain and unsaturated CFAs.

Nomenclature of Fatty Acids

Fatty acids are properly named according to the number ofcarbon atoms, the type of functional groups, and the double-bond location(s). The systematic name can be simplified bywriting C followed by the number of carbon atoms to the leftof a colon and the number of double bonds on the right(Table 1). Different conventions are encountered for locatingthe double-bond positions and cis or trans isomers. w, thelowercase letter omega of the Greek alphabet, indicates thedouble-bond position from the hydrocarbon end of the chain,and c and t indicate the cis and trans configurations of thehydrogen atoms. It is important to note that, as with otherconventions, numbering of branched-chain, cyclopropane-containing, and hydroxy fatty acids proceeds from thecarboxyl end of the molecule. To illustrate these principles,Fig. 1 shows the long form of two fatty acids commonlyfound in bacteria. Some fatty acids were at one time given acommon name to reflect the source from which they wereoriginally identified, e.g., "lactobacillic" acid from Lacto-bacillus spp. Since not all fatty acids have been givencommon names, the use of such terminology can be confus-ing, and the systematic or simplified form is usually pre-ferred.

Lipid Content of Microbes

Figure 2 illustrates typical cell membrane structures andhypothetical examples of commonly found fatty acid moi-eties in two phospholipids. Figure 3 illustrates the lipid Amolecule of Escherichia coli. The fatty acids constituting E.coli lipid A are qualitatively and quantitatively reflected inthe results of whole-cell CFA analysis with respect to therelative proportions of C14:0, C12:0, and 3-OH-C140. While3-hydroxy fatty acids in general are a marker for gram-negative bacteria, 3-OH-C14:0is indicative of the type of lipidA found in E. coli (53). The fatty acid composition of lipid Ain other gram-negative bacteria may be different from that ofE. coli lipid A and similarly is reflected in the CFA analysisof whole cells. For example, the lipid A of Bacteroidesfragilis contains five fatty acids instead of the six foundin E. coli (161). The CFAs of B. fragilis found in LPSare iso-C14:0, 3-OH-C15:0s iso-3-OH-C160, 30HC16:0, and3-OH-C17:0. Besides the lipids and fatty acids typical of otherbacteria, many anaerobes also contain unique lipids calledplasmalogens (56). These are phospholipid analogs with anether linkage instead of the usual ester linkage of a fatty acid

VOL. 4, 1991

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

CLIN. MICROBIOL. REV.

TABLE 1. Nomenclature of some fatty acids commonly found in bacteria

Type of Namefatty acid Systematic Simplified Common

Saturated Dodecanoic C12:0 Lauric acidTetradecanoic C14:0 Myristic acidHexadecanoic C16:0 Palmitic acidOctadecanoic C18:0 Stearic acidEicosanoic C20:0 Arachidic acid

Unsaturated cis-9-Hexadecenoic C16:1w7c Palmitoleic acidcis-9-Octadecenoic C18:1w9c Oleic acidcis-11-Octadecenoic C18 lw7c Vaccenic acid

Branched chain 13-Methyltetradecanoic Iso-C15:012-Methyltetradecanoic Anteiso-C,5.010-Methyloctadecanoic 10-Me-C19.0 Tuberculostearic acid

Hydroxy 3-Hydroxytetradecanoic 3-OH-C14:0 P-Hydroxymyristic acid

Cyclopropane cis-11,12-Methylene-octadecanoic C19:Oycll,12 Lactobacillic acid

a Adapted from reference 134.

to a glycerol carbon. The derivative formed is a dimethylac-etal instead of the typical methyl ester derivative.

INSTRUMENTATION AND DATA PROCESSING

The only commercially available GLC system dedicated tothe identification of bacteria and yeasts by CFA analysis isthe Microbial Identification System. It was codeveloped bythe Hewlett-Packard Co. and Microbial ID, Inc., Newark,Del., and is now marketed by Microbial ID. The systemconsists of a gas chromatograph equipped with a flameionization detector, 5% methylphenyl silicone fused-silicacapillary column (25 m by 0.2 mm), automatic sampler,integrator, computer, and printer (163). The original database for identification of aerobic bacteria was developed bySasser (132). Software libraries for the identification of alarge number of aerobes, anaerobes, mycobacteria, andyeasts have subsequently been developed and updated.Equipment designed and dedicated for the purpose of micro-biologic identification is not a necessary requirement, but theMicrobial Identification System greatly facilitates the controlof conditions during analysis and the interpretation of re-sults. An element of automation is added by the automaticsampler, which lets the operator run up to 100 sampleswithout intervention. Upon injection of a sample into thecolumn containing a specified flow of hydrogen (carrier) gas,the fatty acids are separated because of different retentiontimes under conditions of increasing temperature. A com-puter-controlled temperature program begins at 170°C and is

H H H H H H H H H H H H H H H O

H-C-C-C-C-C-C-C_-C-C-C-C-C-C'-C-C -C-C'-C-OHH H H H H H H H H H 4 H H H H H H

H HH H H H H H C H H H H H H H H H O

/ \ I IIH-C-C-C-C-C-C-C-C-C-C-C-C-C-C-C-C-C-C-OH

H H H H H H H H H H H H H H H H H

FIG. 1. Molecular formulas of cis-li-octadecenoic acid or C18.1'a7c or vaccenic acid (top) and cis-11,12-methylene-octadecanoicacid or C19:cyc11,12 or lactobacillic acid (bottom).

gradually increased to 270°C at 5°C/min. When the methylester derivatives reach the end of the column, a signal fromthe flame ionization detector is recorded as a peak by theintegrator. The area under the peak reflects the relativeamount of individual fatty acids. The retention time of amixture of known fatty acids is used by the computer, or bythe individual if done manually, to calculate an equivalentchain length for the molecule. The equivalent chain length isequal to the number of carbon atoms of a straight-chainsaturated fatty acid or to a number that can be calculated byinterpolation with a mathematical formula for other fattyacids. The accuracy of naming fatty acid peaks by comparingretention times with those of a known mixture is high whenthe computer is used, but definitive identification can bemade only by mass spectrometry. For most applications,mass spectrometry is unnecessary. The amounts of CFAsdetected are calculated as a percentage of the total amount,and a summary can be printed at the end of each run to showthe names and amounts of the CFAs and, optionally, themost likely identification according to similarity to entries inthe data base (132, 163). The multivariate statistical methodof principal-component analysis is used by the computer asthe basis for interpreting data and matching an unknownwith data base entries. Each CFA quantitatively representsthe objects of the principal-component analysis, resulting inpattern recognition as the basic means of identifying anisolate (89). Numerical analysis of CFA data can also beperformed, resulting in a computer-generated dendrogram.An unweighted pair-matching method may be used to showsimilarities at the genus, species, and subspecies levels(128).

METHODS

The usual preparation of samples for CFA analysis con-sists of hydrolysis of the whole-cell fatty acids to formsodium salts and then methylation of the CFA esters to makethem volatile in the gas chromatograph. Various proceduresinvolving acid or base hydrolysis followed by esterificationwith methanol have been described before (78, 95, 99, 111,134). Recent improvements in methods have optimized therecovery of CFAs that formerly were difficult to identify

424 WELCH

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

APPLICATIONS OF CELLULAR FATTY ACID ANALYSIS

LIPID BILAYER.....CELL MEMBRANE

FIG. 2. Diagram of cell membrane and example of fatty acid composition. Adapted from reference 129.

reliably. Miller (96) described a simple washing procedurewith NaOH that removes free acids and secondarily pre-vents the tailing of hydroxy acid peaks in fused-silica capil-lary columns. This and other refinements in the proceduresdeveloped by Lambert and Moss (78, 99) have led to arelatively simple four-step process for preparation of sam-ples (132, 163).

First, after cells (approximately 50 mg, wet weight) areharvested from culture plates incubated for 24 to 48 h,saponification is conducted in a sodium hydroxide-methanolsolution for 30 min at 100'C. This liberates the fatty acidfrom cellular lipid. The second step is methylation with HCOin methanol at 80'C for 10 min. Third, the fatty acid methylesters are extracted into a solution of hexane and methyltertiary butyl ether (10 min), and finally, the extract iswashed in aqueous NaOH for 5 min and then transferred toa GLC vial which is capped for injection into the GLC. Thus,the entire process of sample preparation takes roughly 1 h.Multiple samples can be processed simultaneously, with theincremental increase in the time required generally propor-tional to the number of samples.

Derivatives of CFAs other than methyl esters can beanalyzed. For direct analysis of clinical material in whichincreased sensitivity may be required, i.e., requiring elec-tron capture rather than flame ionization detection, haloge-nated CFAs are prepared. Many of the methods pertaining tothe preparation of trichloroethanol esters have been devel-oped by Brooks and co-workers (20, 21, 33) and have beenmost successfully applied to cerebrospinal fluid (CSF) (21)and diarrheal stool specimens (19). A drawback of theseapplications is that the sample preparation steps are moreinvolved and time-consuming than those for the preparationof fatty acid methyl esters. In addition, a reversed-phasecolumn chromatography step is recommended to partitionCFAs of interest from interfering constituents (34).

FEASIBILITY OF WIDESPREAD USE OF GLC INCLINICAL LABORATORIES

The prospects for using GLC of bacterial CFAs in diag-nostic microbiology have been appreciated for several years.

The idea was ahead of its time, however, because thetechnology for making it practical and dependable in theclinical laboratory was lacking. The developmental status ofmost applications reached a plateau in the 1970s and did notjustify widespread use of the method until recently. Thesituation was aptly summed up by O'Leary in a 1975 review(123): "We are still convinced that fast analyses and properinterpretations of cellular lipid contents can be used toidentify pathogens, but so far it seems that this is an ideawhose time has not yet come, but is close." Interestingforesight was also contained in the title of a contribution byMoss to the proceedings of the Third International Sympo-sium on Rapid Methods and Automation in Microbiology:"Chromatographic Analysis: a New Future for ClinicalMicrobiology" (98).Two major advances that have brought the technology

forward in terms of making it appropriate for use in theclinical laboratory can be cited. One is the development andimplementation of fused-silica capillary columns (99, 104). Incontrast to packed columns and those of greater width, thesecolumns allow reproducible recovery of hydroxy fatty acidsand the ability to distinguish several isomers of fatty acidswith the same carbon chain length (99). The second advanceis the efficient data processing afforded by modern micro-computer systems (5, 48). Of considerable importance alsoare the contributions of Miller (96) and of Moss and others(78, 95, 99, 111), who defined practical steps for samplepreparation in the clinical laboratory. A natural extension ofcurrent technology will be further automation and thusreduced technologist involvement in the saponification,methylation, and extraction procedures. It also seems pos-sible to concomitantly scale down the process, since theactual sample requirement is only a few microliters. Thus,fewer cells or even single colonies from primary culturescould be analyzed instead of large inocula from pure subcul-tures.

Practical Considerations

From a practical standpoint, the chief drawback of CFAanalysis mainly involves the preparation of samples. Al-

VOL. 4, 1991 425

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

CLIN. MICROBIOL. REV.

HO v7

HN

.-O H-G¶X

FAITVAGVS

u

P=O

OH 0-

FIG. 3. Diagram of fatty acids constituting E. coli lipid A. Adapted from reference 53.

though the complexity and number of steps have beenminimized by recent methodologic improvements (78, 99,132), sample preparation is relatively laborious comparedwith preparation for other identification methods used inclinical microbiology. While the process is amenable tohandling one to many samples at a time, it is impractical tohandle very few and laborious to handle very many (.40 to50) at once. Sufficient technical expertise is necessary toensure proper preparation of samples. Complete saponifica-tion by proper exposure of cells in a boiling water bath andcareful control of the time and temperature during methyla-tion are critical steps. The quantity of cells that must beharvested may be achieved easily with a loopful for somestrains but with great difficulty for certain strains that havedifferent growth characteristics or colonial properties.

Properly functioning instruments require attention tomaintenance. Except for the regular replacement of columnliners and injection port septa, the demands of GLC main-tenance are not generally very great. High-purity gases are

required, but their acquisition and storage should not be a

problem for laboratories that maintain CO2 incubators or

anaerobic chambers. The use of gases is economical, as isthe use of other reagents, which can be prepared in largequantities and stored at room temperature. Before analysis,samples can be stored in a freezer if desired. The risk ofchemical contamination of samples during preparation mustbe controlled by using clean glassware and avoiding theintroduction of interfering substances, especially those suchas particulate rubber or oil matter. In contrast to samplepreparation, the performance of GLC, including data entry,storage, and retrieval, is economical in terms of manuallabor.

Comparison with DNA Probes

There are similarities as well as differences between thetwo approaches of CFA analysis and use of DNA probes (4,138) in clinical microbiology. Both are widely applicable,since most pathogens contain fatty acids and all containDNA (or RNA). In this context, CFA analysis requiressophisticated instrumentation, whereas hybridization tech-nology requires a battery of probes. Both methods poten-tially offer a high level of sensitivity and specificity, althoughthe ultimate sensitivity and specificity are theoreticallyachievable with probes combined with DNA amplification.The sensitivity of GLC, however, can be made to exceedthat of DNA probes without amplification. By using anelectron capture detector, femtomole (10-15) quantities of aMycobacterium tuberculosis CFA can be detected (20). Bothmethods can provide results the same day that tests arebegun, both have similar requirements of sample prepara-tion, and both potentially can be adapted to direct detectionof pathogens in clinical material.Both techniques also share limitations to their widespread

use in clinical laboratories. They are generally unwarrantedfor common pathogens such as Staphylococcus aureus thatare easily identified by one or two simple tests. Moreover,although CFA and DNA probe analyses are relatively rapid,many simpler tests are more rapid and less expensive.Though not as significant a problem as that associated withthe use of radiolabeled probes, the disposal of waste in theform of organic solvents is a minor drawback of GLC. Theuse of either approach in smaller laboratories may not becost-effective because of a relatively high cost for low-volume testing. Conversely, laboratories that provide refer-

426 WELCH

I

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

APPLICATIONS OF CELLULAR FATTY ACID ANALYSIS 427

ence services, mycobacteriology, or high-volume testingmay find either technology equally applicable. Characteriza-tion of unusual isolates that may be frequently encounteredin a reference laboratory can be more readily achieved byGLC than by DNA hybridization because of the limitedarray of specific probes available. Many laboratories arebeginning to address and evaluate the judicious use of DNAprobe technology and should consider doing the same forGLC to advance capabilities in the laboratory diagnosis ofinfectious diseases.

CHARACTERIZATION OF BACTERIA

Analysis of CFA composition can be done in conjunctionwith other tests or as the main determinant in identification.In general, the greater the dependence on CFA analysis, thegreater the need to control variables such as culture mediaand growth conditions. These must be controlled as carefullyas possible to arrive at valid conclusions from CFA analysisas the basis for differentiation at the subspecies level. On theother hand, medium, length and atmosphere of incubation,and temperature generally have little impact on the qualita-tive detection of CFAs. The influence of temperature, forexample, is seen mainly in the relative proportions of CFAsrather than in loss or gain of major components. In E. coli,an increase in temperature results in decreased unsaturatedrelative to saturated fatty acids (90). For bacteria that formcyclopropane CFAs from monoenoic acids, cells harvestedduring the stationary phase of growth contain larger amountsof these CFAs than cells harvested during exponentialgrowth (72, 90).

Identification to Genus or Species

Staphylococcus spp. S. aureus and coagulase-negative spe-cies have qualitatively similar CFA compositions (47, 76,122). They consist mainly of the branched-chain iso-C15:0,anteiso-C15:0, and anteiso-C170 CFAs and C18:0 and C20:0CFAs. From the results of early investigation (65), it wasconcluded that S. aureus could not be distinguished from S.epidermidis by CFA patterns but that there was a differencebetween Micrococcus and Staphylococcus spp. The latterspecies contain iso-C19:0. anteiso-C19:0 and C20:0, whereasC15:0 is present in Micrococcus spp. but is either absent (65,139) or present in only very low amounts (47) in Staphylo-coccus species. With the exception of S. saprophyticus andS. warneri, other species of coagulase-negative staphylo-cocci are not readily distinguished by CFA analysis. S.warneri is characterized by a C22 CFA, and S. saprophyticuscontains larger amounts of C16:0 and C20:0 than of otherchains (47). It is interesting that the species of Micrococcusstudied by Jantzen et al. (65) included Micrococcus muci-laginosus (Stomatococcus mucilaginosus), now recognizedwith increasing frequency in vascular-catheter-related infec-tions. Their study of five strains revealed the absence ofC16:1, which characterized other micrococcal species, andthe presence of relatively high concentrations of iso-C14.0and C16:0 CFAs.

Notwithstanding the reported similarity between theCFAs of S. aureus and other species, which have probablybeen studied in some cases as a heterogeneous group, itseems that well-characterized species of the coagulase-negative staphylococci deserve further study to determinethe usefulness of CFA analysis in identifying them. Theresults of one such study have been reported recently.Kotilainen et al. (76) found that the CFA compositions of

blood isolates could be used to distinguish S. epidermidis, S.haemolyticus, S. warneri, S. capitis, S. lugdunensis, S.simulans, and S. hominis, provided that results were inter-preted by cluster analysis. Quantitative differences in theamounts of individual CFAs were insufficient without com-puter analysis to distinguish species. These results are notentirely consistent with those of O'Donnell et al. (122), whoshowed that numerical analysis of CFA data separated only50% of the strains studied into homogeneous clusters. Inview of increased rates of nosocomial infections caused bycoagulase-negative staphylococci, this technique may behelpful in readily characterizing (i) isolates from a series ofblood cultures in a patient or (ii) paired isolates fromindwelling line and venipuncture sites, thus guiding interpre-tation as to the clinical significance of positive blood cultureisolates.

Streptococcus and Enterococcus spp. CFA analysis has beenused as an aid in classification of the streptococci. Bosley etal. (12) compared the CFAs of streptococci (175 cultures)with those of related genera. Streptococcus spp. containC16:1.5, which the genera Aerococcus, Enterococcus, Pedi-ococcus, and Lactococcus do not. Those investigators alsoreported the presence of significant quantities of two rela-tively unusual unsaturated CFAs, C16:1i9c and C16:1.9g, inAerococcus spp. Clinically important enterococci containthe cyclopropane-containing CFA C19:0cyc11,12' which wasnoted in an earlier study to be present in reproduciblydifferent amounts among four species of enterococci (2). Thelowest amount, an average of 1%, occurs in Enterococcuscasseliflavus; intermediate amounts are found in E. faecalisand E.faecium; and the greatest amount, an average of 17%,occurs in E. durans (2). Difficulties involving phenotypicvariability at the species level may be overcome to a certainextent, but CFA profiles also indicated heterogeneity amongcertain species. The profile of "S. milleri" (S. anginosus)differs qualitatively by source of isolates. For example,vaginal strains reportedly contain C17:0, which is absent inoral strains (45).Attempts to identify Lancefield groups by CFA were made

by Drucker (44), who determined that representative strainsof different groups had profiles that could be distinguishedonly by computer analysis. Determining the CFA composi-tion may be more helpful among streptococcal species notidentifiable by group carbohydrate antigen. For example,Streptococcus mutans and S. salivarius are distinguishedfrom related species by the presence of C20:0 and C20:1 (80,130). A relatively large amount of C20:1 can be used todistinguish S. mutans or S. salivarius from a physiologicallysimilar species such as S. bovis (130). In summary, theapplicability of CFA analysis for Streptococcus spp. isminimal for those species that can be identified by immuno-diagnostic reagents and seems to be greatest for otherspecies and for differentiation among related genera.

Gram-positive bacilli. The coryneforms are probably dis-tinguishable on the basis of CFAs, but many have beendifficult to identify to the species level by classical means.

Therefore, few studies have applied CFA analysis to thisgroup. In general, these organisms contain large amounts ofC18:0, C16:0, and C18:1, and some species contain tubercu-lostearic acid (TBSA), the 10-methyl octadecanoic acidtypical of mycobacteria (6). Corynebacterium diphtheriaedoes not contain TBSA, but other species of Corynebacte-rium contain small amounts and Rhodococcus species con-

tain larger amounts of it. GLC with well-characterizedstrains may aid in overcoming the problems of classificationthat persist with members of this group. Athalye et al. (6)

VOL. 4, 1991

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

CLIN. MICROBIOL. REV.

compared profiles of unidentified pathogenic coryneformswith those of several reference strains and found similaritiesthat may provide aids to identification.

Listeria monocytogenes is characterized by branched-chain CFAs 15 and 17 carbons long (both iso and anteiso),C16:0, and C14:0 (126). The composition of Bacillus species isgenerally the most complex in terms of branched-chainCFAs, with typically seven or more iso or anteiso forms (70).In addition to large amounts of C16:0, Bacillus cereus ischaracterized by iso-C15:0, iso-C13:0, iso-C17:0, iso-C14:0,anteiso-C15:0,. and anteiso-C13:0 (119). A large study recentlyexamined 561 isolates encompassing the genera Corynebac-terium, Arcanobacterium, Actinomyces, Brevibacterium,Erysipelothrix, Oerskovia, Propionibacterium, Rothia, List-eria, Kurthia, and Jonesia (7). Strains were assigned to oneof two broad groups on the basis of overall amounts ofbranched-chain versus straight-chain CFA. Several but notall species within the two groups could be precisely identi-fied by further quantitative analysis of the CFA profiles,resulting in the investigators' conclusion that conventionaltests in conjunction with GLC should be performed foridentification of coryneforms and similar bacteria.Mycobacterium and Nocardia spp. Mycobacteria have been

chosen frequently for study by CFA analysis because oftheir high content of cellular lipid. Because of the abundanceof mycolic acids in mycobacterial species, several studieshave addressed the separation of compounds of both types,i.e., those with chains of >20 carbons (mycolic acid cleavageproducts) and those with chains of <20 carbons (constitutivefatty acids) (58, 87, 150). Quantitative analysis of fatty acidsbelonging to these two groups results in accurate identifica-tion to the species level (58, 69, 82, 150, 151). Among isolatesencountered over a 2-year period, Jantzen et al. (69) deter-mined that M. tuberculosis could be identified withoutexception, largely on the basis of C26:0 at concentrations of1 to 13%. Other studies suggest that differentiation of speciesmay be achieved without establishing special chromato-graphic parameters to detect mycolic acid cleavage productsas well as constitutive CFAs (82, 93, 124, 143). Tisdall et al.(144) concluded that GLC was equally accurate as and morerapid than conventional identification. All isolates of M.gordonae and M. kansasii and 85% of isolates of M. tuber-culosis were correctly identified in a prospective clinicalstudy. In a subsequent follow-up study, there were nodiscrepancies in the identification of 42 isolates of M. tuber-culosis. Of 325 strains overall, 320 matching identificationswere obtained by GLC and biochemical profiles (143). Themost frequent discrepancies occurred with M. avium, M.scrofulaceum, and M. gastri.

All species of mycobacteria are characterized by an abun-dance of C16:0, Cl8i,:,9, and, with the exception of M.gordonae, the single-methyl branched constituent TBSA(69, 82, 144). M. gordonae may also be recognized by thepresence of iso-C14:0 (82, 144). However, the closely relatedspecies M. avium, M. intracellulare, and M. scrofulaceumcannot be easily differentiated, even on the basis of mycolicacid cleavage products (69). Differentiation of species be-longing to Runyon group I (M. kansasii and M. marinum) ispossible (69), as is differentiation of the less commonlyencountered species M. terrae, M. xenopi, M. flavescens(93), and M. malmoense (69, 150). Unusual branched-chainCFAs, i.e., a 14-carbon chain in M. kansasii and a 15-carbonchain in M. marinum, distinguish these species (141, 144).

Analysis of CFAs may also be useful in differentiating thegroup IV rapid growers M. fortuitum and M. chelonae fromNocardia spp. (49). A C20:1 CFA absent in mycobacteria was

identified in nine strains of Nocardia asteroides and a CFAidentified by mass spectra as 2-methyl-2-octadecenoic acidwas found in all strains of M. fortuitum, but neither of theseCFAs was detected in M. chelonae or in Nocardia spp. Inaddition, M. fortuitum and M. chelonae, but not N. asteroi-des, reportedly contain C14:0 (49).

Enterobacteriaceae. Typical CFAs are primarily saturatedC17:0cyc, C19:ocyc, and 3-OH-C14:0 Most also contain C16:1 invariable amounts (9, 88). Clear separation of members on thebasis of CFA composition has not consistently been estab-lished, particularly by earlier investigations. Differencesbetween strains of the same species have been noted, andsimilarities may be observed from one genus to the next (88).As with other groups, the explanation is largely the uncer-tainty of taxonomic relationships. More recent studies,however, suggest that the CFA composition of members ofthe family Enterobacteriaceae is at least genus specific andin some cases specific enough for identification to species (9,154, 163). For example, Morganella spp. can be distin-guished from Proteus and Providencia spp. by the presenceof a significant amount of C12:0 (154). Other minor quantita-tive differences distinguish Proteus vulgaris and Providenciaalcalifaciens from the other species of the Proteus-Providen-cia group.

Algorithms based on cluster analysis of CFA data wereestablished by Boe and Gjerde (9) to separate members ofthe families Vibrionaceae and Enterobacteriaceae. Withinthe latter family, Boe and Gjeide were able to identify twospecies of Salmonella, E. coli, Enterobacter cloacae, Mor-ganella spp., and Klebsiella pneumoniae on the basis ofquantitative differences in 17 CFAs. As taxonomic relation-ship among members of this family continue to becomevalidated, it will be possible to apply CFA analysis forcomparison with phenotypic characterization. It is not ex-pected, however, that certain species or even genera such asEscherichia (E. coli) and Shigella that have high DNArelatedness could ever be wholly distinguished by CFAs,given the presently accepted classification.

Vibrionaceae. The Vibrionaceae may be distinguishedfrom the Enterobacteriaceae by the presence of significantamounts of branched-chain CFAs, especially iso-C16.0 andiso-C18:0, in addition to greater amounts of C16:1 (9). Variousspecies of Vibrio and Aeromonas can be distinguished byCFAs. Urdaci et al. (149) studied Vibrio isolates belonging to22 species and found that satisfactory separation of most ofthem could be achieved on the basis of principal-componentanalysis of the CFA data. Isolates identified to species levelincluded V. vulnificus, V. fluvialis, and V. damsela. How-ever, V. cholerae could not be distinguished from V. mim-icus, and V. parahaemolyticus clustered with V. alginolyti-cus and V. natriegens. Among Aeromonas species, the ratioof C16:0 to C17:0 was used to separate clinical isolates into thespecies designated Aeromonas hydrophila (mean ratio, 8.6),A. sobria (mean ratio, 22.7), and A. caviae (mean ratio, 42.3)(26). Low amounts of iso-C15:1 and iso-C17.1 may serve todistinguish Aeromonas from Vibrio species (149), but theseCFAs were not reported in Aeromonas spp. by other inves-tigators (26).

Neisseria and Moraxella spp. CFA analysis was first usedwith members of the genus Neisseria to address problemspertaining to classification. Consistent with findings fromDNA studies, results of GLC showed a marked differencebetween Neisseria (Moraxella) catarrhalis and other species(86). M. catarrhalis contains significant amounts of C10:0 andC18:0 but less C14:0 than is found in the other species (66, 86).The "true" Neisseria spp. are qualitatively similar, with the

428 WELCH

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

APPLICATIONS OF CELLULAR FATLY ACID ANALYSIS 429

exceptions of N. flavescens and N. elongate, which contain

30OHHC16:0. The pathogenic Neisseria spp. contain largeamounts of C16:0, C16:1, C12:0, 30HC12:0, and C14:0 (107). Incontrast to Moss et al. (107), who reported essentiallyidentical patterns of CFAs in N. gonorrhoeae and N. men-

ingitidis, Jantzen et al. (66) presented a numerical analysis ofdata that suggests that minor quantitative differences may besufficient to reliably distinguish N. gonorrhoeae and N.meningitidis.

Campylobacter and Helicobacter spp. With the recognitionof new species of Campylobacter-like organisms, CFA com-

position has become regarded as an important parameter in

characterizing true members of this genus. Human isolatesof various other species such as "Campylobacter cinaedi"and "C. fenelliae" have CFA profiles that distinguish themfrom other campylobacters. The CFAs individually uniqueto these organisms suggest that these species may not bemembers of the genus Campylobacter. Their assignment to

the genus Helicobacter has been proposed (152). Blaser et

al. (10) first demonstrated that C. jejuni could be differenti-ated from C. fetus or C. intestinalis by the presence of

C19:Ocyc. Three isolates of another species, C. landis, inwhich Cl9:Ocyc is absent, were recognized by CFA analysisamong 23 isolates that were from patients with gastroenteri-tis and were initially identified as C. jejuni (29). Majordifferences in the CFA composition of species were furtherdocumented by Lambert et al. (83). In addition to theassociation ofCl9:ocyc withC. jejuni, they found that a group

including C. fetus was characterized by the presence of

3-0H-C14:0 and3-0H-C16:0. This observation has been con-

firmed recently by Brondz and Olsen (17). Unlike Campylo-bacter spp., Helicobacter pylori (formerly C. pyloridis) ischaracterized by unusually low amountsof C16:0 (57). Themajor CFAs of Helicobacter spp. are C14:0, C19:ocyc, C18:1,and C18:0.

Pseudomonas spp. and nonfermenters. The genus Pseudo-monas contains a larger number of species that can becharacterized by CFA composition, which greatly facilitatesidentification. This is in part due to the fact that typical CFAprofiles of pseudomonads have several CFAs which, when

quantitatively analyzed, are distinctive. Examples of nearlyall major CFA types, i.e., saturated, hydroxy, branchedchain, cyclopropane, and unsaturated, are found in thisgroup (37, 39, 40, 42, 101, 113, 131). The division ofPseudomonas species into eight GLC groups (101) parallelsRNA homology grouping. The GLC groups I and II de-scribed by Moss and Dees (101) in 1976 still correspond tothe respective RNA groups consisting of Pseudomonasaeruginosa and others (group I) and P. cepacia and others(group II). CFA analysis is rarely needed to identify P.

aeruginosa, of course, because much simpler and more

reliable techniques exist. However, Veys et al. (156) pro-

posed the use of GLC to identify nonfermenters routinely,including other pseudomonads belonging to a total of 35

species. They identified 19 distinct GLC groups. Membersbelonging to these groups differed primarily in the types ofhydroxy CFAs (Table 2).Nonfermenters associated with rarely encountered but

important infections can be readily identified by GLC. Forexample, Eikenella corrodens and two other pathogens, a

Kingella sp. and Cardiobacterium hominis, with which itshares certain phenotypic characteristics, are distinguish-able by CFAs (157). Eikenella corrodens contains less C14:0than the other two and has C18:1.,7c and C16:0 as the majorCFAs and C16:1, C12:0, C14:0, and C18:0 in smaller amounts(125, 157). Cardiobactenium, hominis is the only one of the

CD

D.

la

0

Co

0%

110 00 (O t .P- w" 0'Cm-j(\~ P

C, C)- B~~~~~~~Ca 0-¢

X

x Ax

X Xe

Ox X

VOL. 4, 1991

00o

*0.

_.0C

U,o,o,

0

O0

8

0

o

C,

0

8

t'

..

o

_

cn

O

Q2

0

o

-3t\)rm

C,

n0

N.n

SD

C.

0

C,U,

=r

0

=-CDl

It

(ACz

.1119

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

CLIN. MICROBIOL. REV.

three characterized by the absence of 3-OH-C12:0. Anotherexample is in cystic fibrosis patients, from whom the impor-tant pathogen P. cepacia can usually be isolated and identi-fied from respiratory specimens with little difficulty by usingselective media. However, when a closely related speciessuch as P. gladioli is present, differentiation of P. cepaciamay be problematic. Furthermore, the distinction is likely tobe important, since P. cepacia is a known prognostic factorin pulmonary disease of cystic fibrosis patients but P.gladioli is not. The CFA profile of P. gladioli differs fromthat of P. cepacia by the presence of 3-OH-C10:0 andsubstantially smaller amounts of C16:1 and C17:0cyc (27).

Legionella spp. Legionella species contain large amountsof branched-chain CFAs (81, 100, 103, 115). The majorcomponent, iso-C16.0, typically accounts for 25 to 50% of thetotal CFA and has been used as the basis for dividing isolatesinto major CFA groups (81). Quantitative differences inconjunction with ubiquinone content allow the differentia-tion of all Legionella species and may permit more rapididentification than any other means. Certain species may beidentified by CFA composition alone. Legionella long-beachae, for example, is quantitatively different from otherspecies in the relative amounts of iso-C16:0 and C16:0 itcontains (106). In addition to the branched-chain CFAs,Legionella spp. also contain small amounts of hydroxy andcyclopropane CFAs, although the amounts are considerablysmaller than in other gram-negative bacteria (81, 103).Through generation of the data for entry in a computer database such as the Microbial Identification System (81), per-formance of GLC for identification of Legionella spp. be-comes a practical approach for laboratories handling theseisolates. Unrelated pathogens that may be confused withLegionella spp. because of growth on charcoal-yeast extractagar (e.g., Francisella tularensis) could also be readilydifferentiated by virtue of CFA content.

Miscellaneous bacteria. Members of the family Pasteurel-laceae possess similar CFAs (15, 64) and Haemophilusinfluenzae (64) or H. aphrophilus (13) closely resembleActinobacillus actinomycetemcomitans in CFA composi-tion, but Pasteurella multocida can be separated by itsgreater amount of saturated and unsaturated C18 CFAs.Quantitative differences may also permit differentiation ofH. ducreyi, H. parahaemolyticus, H. paraphrophilus, andH. suis from other species (64). Gardnerella (formerly Hae-mophilus) vaginalis does not resemble Haemophilus speciesin CFA content. It lacks 3-OH-C14.0 and other hydroxyCFAs commonly found in gram-negative bacteria (31, 64).Major CFA constituents of G. vaginalis are C18:1 and C18:0,with smaller amounts of C14:0, C16:0, and C16:1. Jantzen et al.(64) reported the absence of C18:2 in G. vaginalis, whereasthis CFA was identified in amounts from 4 to 20% in thestudy by Csango et al. (31). This discrepancy illustrates thelimitation in drawing conclusions from separate studies inwhich preparation and analysis of cells for CFA may havebeen performed by different methods.

Differentiation of the unusual from the more frequentlyencountered organisms at certain sites of infection can befacilitated by CFA analysis. Four organisms that occur alongwith Pasteurella multocida in wound infections resultingfrom dog bites were studied by Dees et al. (43). Pasteurellamultocida was the only one that contained 3-OH-C14:0.Flavobacterium spp. (group IIj) and Capnocytophaga can-imorsus (DF-2) were similar in possessing large amounts ofiso-C15:0 but could be differentiated from each other by thepresence of iso-C17:1 in Flavobacterium spp. Group DF-3organisms contain substantially more anteiso-C15:0 than iso-

C15:0 (158). Group EF-4 and M-5 organisms contain 3-OH-C12:0, and they can be separated on the basis of 2-OH-C16:0,Canteiso-17:0, and C170cyc' which are all present in group EF-4organisms. Capnocytophaga spp. have been further charac-terized and found to contain the uncommon CFAs 3-OH-iso-C14:0 and 3-OH-iso-C16:0 (38). These serve to distinguishCapnocytophaga spp. from other gliding bacteria, but theyalso are found in Flavobacterium spp. (102).The application of CFA analysis in studying another

formerly unclassified group (IVe) proved useful in establish-ing proper classification of the urinary tract pathogen Ohi-gella urealytica (35). The phenotypic characteristics of thisorganism resemble those of Alcaligenes, Brucella, and Bor-detella spp., but the CFA compositions of these organismsare sharply different. 0. urealytica (IVe) contains C18.1 inlarge amounts; Brucella spp., except Brucella canis (36), arecharacterized by large amounts of Cl9:ocyc, and Alcaligenesspp. and Bordetella bronchiseptica contain C17:0cyc (35).Cyclo CFAs were not detected by Jantzen et al. (68) inBordetella pertussis, whereas significant amounts werefound in Bordetella parapertussis and, along with 2-OH-C12:0, in Bordetella bronchiseptica. Additional investiga-tions of the CFA compositions of Brucella spp. have shownspecific patterns for Brucella abortus, Brucella melitensis,Brucella canis, and Brucella ovis (28, 43, 140).

Phenotypically similar members of the Achromobactergroup can be separated by CFA composition. Achromobac-ter xylosoxidans primarily consists of C16:0, C17:Ocyc' andhydroxy CFAs, whereas other Achromobacter species con-tain large amounts of C18j1 and C19 0cyc CFAs (40).

F. tularensis has one of the more distinctive patterns,being characterized by long-chain CFAs (63, 118) and thehydroxy CFAs 2-OH-C10:0, 3-0H-C16:0, and 3-OH-C18.0(63). Because there is little similarity between the CFAs ofF. tularensis and those of other gram-negative bacteria,GLC is highly specific for the identification of F. tularensis(61). Yersinia pestis and Y. pseudotuberculosis have beenreported to contain 16- and 18-carbon hydroxy CFAs, butYersinia spp. are different in containing C17:0cyc (145).Spirochetes. Treponema pallidum purified from rabbit tes-

ticular tissue has been inconclusively analyzed in regard toCFA content. The CFAs found (mainly C16:0, C18:1, C18:0,and C18:2) were present in the bacteria but were also found insmaller amounts in uninfected tissue (92). Up to 50% of thetotal CFAs in Leptospira spp. are in the form C16:0 (74). Therest are mainly unsaturated C,16:1 C18:1' C18:2, and C14:2CFAs.

Anaerobes. The most common application of GLC inidentification of anaerobes has been the detection of meta-bolic end products of glucose fermentation. Analysis ofwhole cells also is potentially useful and has been proposedas offering a practical approach, at least to identification ofBacteroides spp. (153). The key in a proposed scheme is therelative amount of the 15-carbon CFAs that predominate inclinically encountered species (94, 95). The species withinthree GLC groups delineated by Mayberry et al. (95) wereidentified on the basis of ratios of one 15-carbon CFA toanother.The clinically important Fusobacterium species are

clearly separated from Bacteroides spp. by CFAs. Fusobac-teria are generally characterized by the presence of 3-OH-C14:0, C14:0, C16:0, C16:1, 3-OH-C16:0, C18:0, and C18:1 (25, 60,67). In contrast to Bacteroides spp., Fusobacterium spp.contain no methyl branched CFAs. The amounts of 3-OH-C14:0 vary according to species, and these differences andthe presence of 3-OH-C16:0 aid in the identification of Fuso-

430 WELCH

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

APPLICATIONS OF CELLULAR FATTY ACID ANALYSIS 431

bacterium nucleatum (67). Calhoon et al. (25) used the ratioof C16:0 to 3-OH-C16:0 to differentiate Fusobacterium nucle-atum from other species isolated from the oral cavity.Propionibacterium species contain anteiso-C15:0 or iso-

C15:0 as the most abundant CFA (105). However, Propioni-bacterium acnes is characterized by relatively minoramounts of anteiso-C15:0 compared with the amounts inother propionibacteria. Significant amounts of branched-chain C17:0 and C16:0 CFAs are also found in these species.The anaerobic gram-positive cocci have not been suffi-

ciently studied in the context of a clear understanding oftheir taxonomic status (79). Variable amounts of 14- and15-carbon branched-chain CFAs serve to differentiate sev-eral of the species (79, 110). They range from none inPeptostreptococcus micros to significant quantities of iso-C14:0 in Peptostreptococcus anaerobius and iso-C15:0 inPeptostreptococcus saccharolyticus (79). Similarly, with theclostridia, studies demonstrating the utility of CFA analysisin diagnostic microbiology are lacking. Of 41 isolates studiedby Moss and Lewis (112), all were categorized in fourgroups. A large amount of C12:0, C14:0, C16:0, and C16.1characterized group I, which was represented by Clostrid-ium perfringens. Group II lacked C20:0. Groups II, III, andIV, represented respectively by Clostridium sporogenes, C.bifermentans, and C. histolyticum, possessed other qualita-tive and quantitative differences in CFAs, but 10 additionalspecies were assigned to group IV. Cundy et al. (32) reportedrecently that 51 of 52 isolates of C. difficile were accuratelyidentified by CFA analysis with the Microbial IdentificationSystem. However, in comparison with headspace GLC, theMicrobial Identification System method was thought to betoo complicated. Toxigenic types A, B, and E of C. botuli-num were studied by Fugate et al. (52). In contrast to theresults of CFA analysis with toxigenic types of C. perfrin-gens (112), those investigators suggested that quantitativeanalysis of the CFAs could serve to distinguish types of C.botulinum. Additional information is needed to validate theusefulness of CFA analysis for Clostridium spp. and theanaerobes in general. Beyond providing an aid to routineidentification, there may be useful applications in specialcircumstances, such as the isolation of a Clostridium sp.suspected to be C. botulinum from a wound infection.Because it is not be possible to distinguish C. botulinumfrom C. sporogenes phenotypically by commonly used tests,GLC could potentially provide a rapid means of distinguish-ing the two.

Identification to Subspecies and Epidemiologic Typing

In general, the discriminatory power of CFA analysis isnot great enough to provide useful typing systems. Signifi-cant advantages of this analysis, however, include ease,reproducibility, storage of information when computerized,and universal typeability. The applicability of GLC forsubgrouping (typing) also depends on the species in ques-tion. Those with more complex profiles, i.e., the gram-negative bacilli and especially Pseudomonas species, aremore amenable to GLC typing than are those with simplerCFA patterns. At least three studies (29, 76, 116) haveattempted to use GLC as a tool for identification to thesubspecies level, and they suggest its use in epidemiologictyping.

Campylobacterjejuni. Coloe et al. (29) proposed GLC as aneasy method of epidemiologic typing based on CFA analysisof several human and animal isolates of C. jejuni. The ratioof C18:1 to C19:Ocyc was used to place isolates into one of

three groups, resulting in 85% of the isolates of human originand 56% of the isolates of other animal origin being assignedto separate groups. It is uncertain whether variations in theamounts of cyclopropane-containing CFAs and the corre-sponding precursor (C16: 1.w9c or C18:1) can be relied on as trueindicators of difference in CFA composition. If the amountsof cyclopropane-containing CFA and its unsaturated precur-sor are combined, there is actually very little difference inthe groups identified by Coloe et al. (29). For example,instead of a distribution of 14, 21, and 37% for the amountsof C18:1 in groups I, II, and III, respectively, of humanisolates, those amounts when added to the amounts ofC19:Ocyc become 41, 43, and 48%, representing questionabledifferences. A stronger case could be made for distinction oftypes by analyzing differences, if they existed, among otherfeatures.Pseudomonas cepacia and Staphylococcus epidermidis. Not

only use of the overall CFA composition but also computer-assisted numerical analysis greatly enhance our ability torecognize valid subgroups (116). In a study of P. cepaciaisolates collected from five cystic fibrosis centers, 42 strainswere divided into five subgroups on the basis of overall CFAcompositions. The distribution in subgroups varied fromcenter to center, but most centers had a predominant sub-group, which suggests that this subgroup was acquired in allcases from a common source. Furthermore, the subgroupingwas corroborated in part by biotype distribution in that arelatively unusual biotype of lysine decarboxylase-negativestrains all came from the same center, and six of the sevenstrains also fell within the same GLC subgroup.

Correlation analysis has been applied recently to the CFAdata on blood culture isolates from 60 patients (76). OfStaphylococcus strains serially isolated from the same pa-tient, 97.6% gave a correlation value of >95. The correlationvalues of nonidentical strains (isolates from different pa-tients) of S. epidermidis varied from 34.2 to 94.9, and amongdifferent species, the correlation value ranged from 59 to 77.Consideration of the use of GLC with numerical or correla-tion analysis for subgrouping strains collected for epidemio-logic studies is warranted when an efficient but not highlydiscriminating typing method is desired.

CHARACTERIZATION OF OTHERMICROBIAL SPECIES

A limited number of studies have addressed identificationof microbial pathogens other than bacteria by GLC of CFAs.Purification of a sufficient quantity of the pathogen in ques-tion is the primary challenge for those organisms that are notcultivable on artificial media. GLC can potentially be used asan aid to identification of those organisms that can beprepared in sufficient quantity and purity. Rickettsia andChlamydia spp. tend to have CFAs similar to those of otherbacteria, whereas of the various yeasts, several are charac-terized by large amounts of di- and triunsaturated 18-carbonCFAs.

Rickettsia spp.

The CFAs of five Rickettsia spp. and Rochalimaea quin-tana were determined by Tzianabos et al. (148). Rickettsiarickettsia, R. akari, R. typhi, R. prowazekii, and R. canadawere purified by a standard method from yolk sacs, and allhad similar CFA profiles. Large amounts of C18:1, C16:1, andC16:0 characterize these species. Quantitative but not quali-tative differences in CFA composition were found. The

VOL. 4, 1991

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

CLIN. MICROBIOL. REV.

CFAs of Rochalimaea quintana grown on blood agar me-dium resemble the other rickettsial CFAs except that theylack the 3-OH-C14:0 found in the others. Additional studieshave examined Rochalimaea quintana and revealed similarfindings (135, 162). That C18l: accounts for such a largepercentage of the total CFAs (40 to 60%) is a distinctivefeature of Rochalimaea quintana. In contrast to other re-searchers, Westfall et al. (162) analyzed Rochalimaea quin-tana and related species by electron capture GLC andreported that it contained C9:0. However, they noted thatcomplex media may have been the source of this unusualCFA. Neither of the studies by Tzianabos et al. (148) orSlater et al. (135) found C9.0 in Rochalimaea quintana.

Coxiella burnetii has a CFA profile markedly differentfrom that of other rickettsiae (148, 165). Significant amountsof branched-chain CFAs (anteiso-C15:0, anteiso-cl7:0, andanteiso-C14:0) are found, along with C16:0, C18:0, C20:0, andC16:1. The similarity between the CFA patterns of Coxiellaand Legionella spp. is interesting, although DNA hybridiza-tion has revealed no relatedness between the two (148).

Chlamydia spp.

The lipid elementary bodies of Chlamydia trachomatispurified from McCoy cells consist largely of C16:0, iso-C15:0,anteiso-C15:0. and C18:0 (8). Serotypes D and L3 can bedistinguished from serotypes C and G by quantitative differ-ences in the amounts of the branched-chain C15:0 CFAs. C.psittaci is identifiable and can be similarly differentiatedfrom the C. trachomatis serotypes (8). The CFAs of Chla-mydia spp. were absent in uninfected McCoy cells except fora minor amount of C18,:, which was common to uninfectedMcCoy cells as well as to the Chlamydia serotype.

Fungi

The CFA compositions of various yeasts and a fewfilamentous fungi have been investigated. Initial studies ofyeasts by Gangopadhyay et al. (54), Gunasekaran andHughes (59), and Moss et al. (114) resulted in somewhatconflicting data with respect to the CFAs detected. Theirresults and those from more recent studies indicate that atleast group-specific and, for some, species-specific CFAprofiles can be identified. Substantial quantitative differ-ences among Saccharomyces spp. (large amounts of C16:1)and Candida spp. (large amounts of C18.2 and C16:1) separatethese genera. Candida spp. are also characterized by thepresence of C16:0, C18:0, C18,:, and C18.3 (73). Torulopsisglabrata is characterized by a high ratio of C16:1 to C16:0, andCryptococcus species can be identified by the absence oronly trace amounts of C16:1 relative to the amounts in otheryeasts (91, 114). The results of a study recently reported byMarumo and Aoki (91) show a 95% overall agreement withstandard methods in the identification of Candida spp., T.glabrata, and Cryptococcus neoformans. These results,analyzed by discriminate analysis, yielded rates of correctidentification ranging from slightly less than 90% for Can-dida parapsilosis and Candida albicans to 100% for T.glabrata, Cryptococcus neoformans, and other Candidaspp.Brondz and Olsen (16) examined both cellular carbohy-

drates and CFAs and found that multivariate analyses dis-tinguished Candida albicans from T. glabrata and fromSaccharomyces cerevisiae. In contrast to other investiga-tors, Brondz et al. (18) reported in another study the findingof significant amounts of C12:0 and C14:0 in Candida, Toru-

lopsis, and Saccharomyces spp. A methodologic variable inthe form of alcoholysis with ethanol, propanol, or butanolwas reported to influence the recovery of these more volatileCFAs.Approaches to the identification of filamentous fungi on

the basis of CFA composition have not been described.Notwithstanding the greater technical difficulty in workingwith molds rather than yeasts, CFA analysis may in thefuture become a useful aid as a supplement to morphologicfeatures for identification. Fusarium spp. (137) and Tricho-phyton rubrum (75, 164) are examples of fungi that have beenanalyzed for CFA content. The mycelial mass of growthfrom fluid cultures was harvested and subjected to fatty acidmethyl esterification by these investigators. At variousstages of fungal growth, C16:0, C18:0, C18:1, and particularlylarge amounts of C18:2 accounted for most of the CFAs of T.rubrum. C16:1 and C20:0 were also found. The principal CFAs(C16:0, C18:0, C18:1, and C18:2) of Fusarium oxysporum weresimilar to those of T. rubrum, raising the question of whetherthere is less diversity of CFAs among molds than amongother microorganisms. Additional studies need to be per-formed with related species and genera to develop moreexperience before GLC can be applied to the identification offilamentous fungi.

Blood Parasites

Schistosomes, trypanosomes, Leishmania spp., and plas-modia have been analyzed for CFA composition. Schisto-soma mansoni purified from infected mice (adult stage of theparasite) has a complex fatty acid composition (136), includ-ing saturated, unsaturated, and branched-chain componentswith carbon-chain lengths of 12 to 24 atoms. The fatty acidsin the total lipid fractions in Trypanosoma cruzi consistmainly of C18:2, C18:1, C18:0, and C18:0 (142). Polyunsaturatedacids are also found.

Promastigotes of Leishmania spp. have been studied todetermine whether CFA composition can be used to differ-entiate various species. Means other than host specificity aregenerally unable to separate members of this genus. Vessalet al. (155) concluded that distinctive features of the CFAprofiles of Leishmania donovani, L. tropica, and L. enriettiiwere helpful in identification. A significant amount of C18:3 inL. donovani and its complete absence in the other twospecies, along with the absence of all unsaturated 18-carbonCFAs in L. enriettii, were the major differences.Using Plasmodium lophurae and P. berghei released from

erythrocytes, Wallace (160) determined the CFA composi-tion of lipid fractions. Both species have complex profilesconsisting mainly of C16:0, C18:0, and C18:1. P. berghei wasreported to contain greater amounts of saturated CFAs thanP. lophurae; it was noted that these parasites and the bloodof the hosts share similar compositions.

Mycoplasmas and Viruses

Because of their requirement for exogenous sources ofgrowth factor from either complex medium or host cells,mycoplasmas and viruses can have widely variable fatty acidcompositions. The incorporation of several or as few as onefatty acid into membranes will allow mycoplasmas to grow,while only Mycoplasma (Acholeplasma) laidlawii is knownto synthesize its own CFAs (127).

Lipid constitutes the envelope of viruses such as influenzavirus. Variable amounts of several fatty acids with chains of12 to >20 carbons have been shown to occur, suggesting that

432 WELCH

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

APPLICATIONS OF CELLULAR FATTY ACID ANALYSIS 433

there may be strain differences in the influenza virus (11).Other investigators have analyzed the fatty acid contents ofuninfected and virus-infected cells and found modification inthe fatty acid compositions. For example, cells persistentlyinfected by measles virus have more C16.0 and less C18.1compared with uninfected or lyrically infected monkey kid-ney cells (3).

Unclassified Organisms

CFA characterization is useful in initially placing unclas-sified species into groups. Establishing the similarity of aCFA profile to that of a recognized species, especiallyamong the nonfermenters, steers definitive taxonomic stud-ies based on genetic analysis. Examples include determina-tion of the CFA composition of "Flavobacterium-like" (41)and "Moraxella-like" (108) organisms. Many belonging tothese groups are unusual opportunistic pathogens that maygrow slowly and need to be distinguished from traditionalfastidious pathogens. In characterizing a collection of strainsthat resembled Methylobacterium extorquens, Wallace et al.(159) relied heavily on CFA composition to define groupswith similarities and groups with differences. CFA analysismay reveal differences not readily apparent by phenotypiccharacterization. An unclassified species responsible forserious infection in association with chronic granulomatousdisease (147) was confirmed to be distinct from Pseudomo-nas cepacia by virtue of containing C12:0, which did notcharacterize other members of the pseudomallei group. Theisolate was phenotypically similar to Pseudomonas cepaciaand Pseudomonas pickettii. Similarly, distinct CFA profileshave recently been established for the agent of cat scratchdisease (109) and for an unidentified fastidious organismcausing septicemia (135). Of seven CFAs found in the catscratch disease bacillus, an unusual component, 11-methyl-octadec-12-enoic acid, was detected by GLC, identified bymass spectrometry, and proposed to serve as a marker of thecat scratch disease agent (109).

DETECTING MARKERS OF INFECTION INCLINICAL MATERIAL

The possibility of diagnosing meningitis by GLC of CSFwas reported in 1976 (133). Initial studies focused on thedifferences in patterns of infected and uninfected CSF with-out associating particular derivatized compounds with anygiven pathogen. In most cases, derivatives of more than oneclass of compounds (acids, alcohols, and amines) wereanalyzed. Cryptococcal (30, 133), enteroviral (24, 30, 133),herpesvirus (30), measles virus, varicella-zoster virus,Rocky Mountain spotted fever, meningococcal (24), nocar-dial (22), and tuberculous (21, 30, 51) meningitis have beenstudied. Modification ofGLC resulting in increased sensitiv-ity is central to the direct analysis of clinical specimens (20,85). Preparation of samples appropriate for fatty acid analy-sis by frequency-pulsed electron capture (FPEC) GLC in-volves production of pentafluorobenzyl (20, 33) or trichloro-ethyl (21) derivatives. FPEC-GLC alone (20, 21) or massspectrometry with selected ion monitoring in addition toFPEC-GLC (50, 51, 117, 121) is then used to analyzesamples. The latter approach is thought by some investiga-tors to be necessary for optimal resolution of analytes. Thegreatly increased sensitivity of FPEC-GLC along with theheavy background of fatty acids of host origin are therebymanaged. Another approach has been to use reversed-phasechromatography columns (34), which simplify the removal of

certain components that cause interfering peaks. For com-puter-assisted interpretation of results, methods that ignorecertain peaks normally encountered in clinical specimenscan be installed.Another special consideration relative to direct CFA anal-

ysis is specificity. In contrast to analysis of cells grown incultures for which a profile of CFAs is interpreted qualita-tively and quantitatively to arrive at an identification, FPEC-GLC depends more on detection of selected CFAs asmarkers indicating a microbe's presence. Thus, any of theCFAs uniquely found among prokaryotes could serve as amarker of infection. In CSF, for example, TBSA has beenstudied most extensively in this regard (20, 50, 51, 84, 117).Such markers are useful for diagnosis when detected innormally sterile sites and in the setting of a clinical pictureconsistent with the agent represented by the CFA marker.TBSA is also a component of Nocardia spp. (22) and othercoryneforms, however, so detection in specimens such as

sputum must be interpreted cautiously. Despite these limi-tations, FPEC-GLC promises rapid diagnosis of tuberculosisbecause of its good sensitivity. Increasing promise of appli-cability to the rapid diagnosis of tuberculosis has beenshown by two recent studies (21, 117). Muranishi et al. (117)tested specimens from the respiratory tract of 223 patientswith active tuberculosis. All of 61 specimens with positivesmears and cultures were found to be positive for TBSA(100% sensitivity), and 84% of culture-positive, smear-neg-ative patients were detected by TBSA measurement. Of 160controls, 9.4% were TBSA positive. In 75 patients withmeningitis, Brooks et al. (21) reported 95% sensitivity and91%'specificity of FPEC-GLC for diagnosis of tuberculousmeningitis on the basis of either the presence of TBSA inCSF or the PFEC-GLC profile of CSF. FPEC-GLC has alsobeen applied directly to naturally infected armadillo tissuefor detection of Mycobacterium leprae (77), and attemptshave been made to analyze joint fluid for diagnosing septicarthritis (23). Results of the latter study, which comparedprofiles of gonococcal, streptococcal, and staphylococcalarthritis with those of traumatic synovitis, were probablybased on an analysis of metabolic products.

Perhaps it would be useful to reexamine FPEC-GLC forthe rapid diagnosis of opportunistic infections as well as ofthe rarely encountered, diagnostically difficult infectionssuch as tuberculous meningitis in the light of current epi-demiology. All of the published experience with FPEC-GLCand diagnosis of cryptococcal meningitis, for example, ap-peared before recognition of AIDS. Bacteremia due toMycobacterium avium-Mycobacterium intracellulare, com-

mon in patients with this syndrome, is often of highermagnitude than bacteremias in immunocompetent adults(166), suggesting that serum might serve as the source ofdetectable amounts of CFAs in FPEC-GLC. It would beinteresting to determine the applicability of the technique tothe diagnosis of other relatively common opportunistic in-fections such as histoplasmosis and to actual detection of thehuman immunodeficiency virus.

MEASURING ANTIMICROBIAL RESISTANCE

Among the many prospects for future development ofCFA analysis in clinical microbiology are methods of assess-

ing antimicrobial susceptibility or resistance. The basis ofthis possibility lies'either in the direct effects of antimicrobialagents on membrane lipids or on quantitative analysis of therate of synthesis and incorporation of fatty acids into lipid.The latter would be analogous to the adaptation of DNA

VOL. 4, 1991

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

CLIN. MICROBIOL. REV.

probes used for identification purposes to a system for rapidsusceptibility testing with mycobacteria. Kawa et al. (71)successfully used the Gen-Probe system for isoniazid sus-ceptibility testing of M. tuberculosis in 3 days on the basis ofa comparison of the total amount of nucleotide hybridized inthe presence and absence of various concentrations ofisoniazid. Similarly, one may be able to predict susceptibilityor resistance by comparing the total area of CFA peaks fromanalysis of a strain grown in the presence and the absence ofantimicrobial agents.

Fungi offer an interesting problem in this context becauseall of the newer antifungal agents target membrane lipids.There is either a direct effect on membranes, causing releaseof fatty acids, or an inhibition of lipid biosynthesis. Amarked effect of imidazole antifungal agents on the CFAcomposition of Candida albicans is seen as a ratio ofunsaturated to saturated CFAs that decreases from 2.3 to 1.1in correlation with growth inhibition (55). A shift in therelative amounts of C18:0 and C16:0 is also seen, thus sug-gesting that alterations in the overall CFA composition maypredict antifungal susceptibility of Candida albicans. Itwould be important to study resistant strains to confirm thatdifferences exist. Whereas the targets of most antibacterialagents such as the cell wall, protein synthesis, or DNA arenot directly taken advantage of for measuring antibacterialresistance, fungal lipid may serve as a practical and usefulindicator to measure the effects of antifungal agents.

CONCLUSION

The technology of CFA analysis using GLC has gainedapplicability for several purposes. It offers considerablepower as a tool for microbial identification, because charac-teristic patterns of CFAs can be defined for several microbesto the species level and results are achievable rapidly. Whilemany applications remain highly specialized, laboratoriesengaged in mycobacteriology or frequent identification ofunusual isolates should consider GLC as an effective ap-

proach to identification of these isolates. Experience withCFA analysis of these and other microbial groups for pur-

poses of identification is rapidly developing, making theversatility of the method more fully realized. GLC is pres-

ently applicable on a broad scale in conjunction with con-

ventional tests for identification of bacterial pathogens.Other applications, including characterization at the subspe-cies level, identification of nonbacterial pathogens, anddirect analysis of clinical material, are enticing as a new

technology for achieving epidemiologically and diagnosti-cally useful information.

ACKNOWLEDGMENTS

Excellent secretarial support was provided by Rose Stursa.Geoffrey Mukwaya and Denise Pickett provided helpful discussionsand review of the manuscript.

REFERENCES1. Abel, K., H. deSchmertzing, and J. I. Peterson. 1963. Classifi-

cation of microorganisms by analysis of chemical composition.I. Feasibility of utilizing gas chromatography. J. Bacteriol.85:1039-1044.

2. Amstein, C. F., and P. A. Hartman. 1973. Differentiation ofsome enterococci by gas chromatography. J. Bacteriol. 113:38-41.

3. Anderton, P., T. F. Wild, and G. Zwingelstein. 1981. Modifi-cation of the fatty acid composition of phospholipid in measlesvirus-persistently infected cells. Biochem. Biophys. Res. Com-mun. 103:285-291.

4. Anonymous. 1989. DNA technology and rapid diagnosis ofinfect. Lancet ii:897-898.

5. Aston, J. W. 1977. Computer processing of fatty acid analysisdata. J. Chromatogr. 131:121-130.

6. Athalye, M., W. C. Noble, and D. E. Minnikin. 1985. Analysisof cellular fatty acids by gas chromatography as a tool in theidentification of medically important coryneform bacteria. J.Appl. Bacteriol. 58:507-512.

7. Bernard, K. A., M. Bellefeuille, and E. P. Ewan. 1991. Cellularfatty acid composition as an adjunct to the identification ofasporogenous, aerobic gram-positive rods. J. Clin. Microbiol.29:83-89.

8. Bidawid, S., S. Chow, C. W. Ng, E. Perry, and S. Kasatiya.1989. Fatty acid profiles of Chlamydia using capillary gaschromatography. Antonie van Leeuwenhoek. J. Microbiol.55:123-31.

9. Boe, B., and J. Gjerde. 1980. Fatty acid patterns in theclassification of some representatives of the families Entero-bacteriaceae and Vibrionaceae. J. Gen. Microbiol. 116:41-49.

10. Blaser, M. J., C. W. Moss, and R. E. Weaver. 1980. Cellularfatty acid composition of Campylobacter fetus. J. Clin. Micro-biol. 11:448-451.

11. Blough, H. A., J. P. Merlie, and J. M. Tiffany. 1969. The fattyacid composition of incomplete influenza virus. Biochem.Biophys. Res. Commun. 34:831-834.

12. Bosley, G. S., P. L. Wallace, C. W. Moss, A. G. Steigerwalt,D. J. Brenner, J. M. Swenson, G. A. Hebert, and R. R.Facklam. 1990. Phenotypic characterization, cellular fatty acidcomposition, and DNA relatedness of aerococci and compar-ison to related genera. J. Clin. Microbiol. 28:416-421.

13. Braunthal, S. D., S. C. Holt, A. C. R. Tanner, and S. S.Socransky. 1980. Cellular fatty acid composition of Actinoba-cillus actinomycetemcomitans and Haemophilus aphrophilus.J. Clin. Microbiol. 11:625-630.

14. Brian, B. L., and E. W. Gardner. 1968. Fatty acids from Vibriocholerae lipids. J. Infect. Dis. 118:47-53.

15. Brondz, I., and I. Olsen. 1984. Determination of board cellularfatty acids in Actinobacillus actinomycetemcomitans and Hae-mophilus aphrophilus by gas chromatography and gas chroma-tography-mass spectrometry. J. Chromatogr. 308:292-288.

16. Brondz, I., and I. Olsen. 1990. Multivariate analyses of cellularcarbohydrates and fatty acids of Candida albicans, Torulopsisglabrata, and Saccharomyces cerevisiae. J. Clin. Microbiol.28:1854-1857.

17. Brondz, I., and I. Olsen. 1991. Multivariate analyses of cellularfatty acid in Bacteroides, Prevotella, Porphyromonas, Wo-linella, and Campylobacter spp. J. Clin. Microbiol. 29:183-189.

18. Brondz, I., I. Olsen, and M. Sjostrom. 1989. Gas chromato-graphic assessment of alcoholyzed fatty acids from yeasts: anew chemotaxonomic method. J. Clin. Microbiol. 27:2815-2819.

19. Brooks, J. B. 1986. Review of frequency-pulsed electron-capture gas-liquid chromatography studies of diarrheal dis-eases caused by members of the family Enterobacteriaceae,Clostridium difficile, and rotavirus. J. Clin. Microbiol. 24:687-691.

20. Brooks, J. B., M. I. Daneshvar, D. M. Fast, and R. C. Good.1987. Selective procedures for detecting femtomole quantitiesof tuberculostearic acid in serum and cerebrospinal fluid byfrequency-pulsed electron-capture gas-liquid chromatography.J. Clin. Microbiol. 25:1201-1206.

21. Brooks, J. B., M. I. Daneshvar, R. L. Haberberger, and I. A.Mikhail. 1990. Rapid diagnosis of tuberculous meningitis byfrequency-pulsed electron-capture gas-liquid chromatographydetection of carboxylic acids in cerebrospinal fluid. J. Clin.Microbiol. 28:989-997.

22. Brooks, J. B., J. V. Kasin, D. M. Fast, and M. I. Daneshvar.1987. Detection of metabolites by frequency-pulsed electron-capture gas-liquid chromatography in serum and cerebrospinalfluid of a patient with Nocardia infection. J. Clin. Microbiol.25:445-448.

23. Brooks, J. B., D. S. Kellogg, C. C. Alley, H. B. Short, H. H.

434 WELCH

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

APPLICATIONS OF CELLULAR FATTY ACID ANALYSIS 435

Handsfield, and B. Huff. 1974. Gas chromatography as apotential means of diagnosing arthritis. I. Differentiation be-tween straphylococcal, streptococcal, gonococcal, and trau-matic arthritis. J. Infect. Dis. 129:660-668.

24. Brooks, J. B., J. E. McDade, and C. C. Alley. 1981. Rapiddifferentiation of Rocky Mountain spotted fever from chicken-pox, measles, and enterovirus infections and bacterial menin-gitis by frequency-pulsed electron capture gas-liquid chro-matographic analysis of sera. J. Clin. Microbiol. 14:165-172.

25. Calhoon, D. A., W. R. Mayberry, and J. Slots. 1983. Cellularfatty acid and soluble protein profiles of oral fusobacteria. J.Dent. Res. 62:1181-1185.

26. Canonica, F. P., and M. A. Pisano. 1988. Gas-liquid chromato-graphic analysis of fatty acid methyl esters of Aeromonashydrophila, Aeromonas sobria, and Aeromonas caviae. J.Clin. Microbiol. 26:681-685.

27. Christenson, J. C., D. F. Welch, G. Mukwaya, M. J. Muszyn-ski, R. E. Weaver, and D. J. Brenner. 1989. Recovery ofPseudomonas gladioli from respiratory tract specimens ofpatients with cystic fibrosis. J. Clin. Microbiol. 27:270-273.

28. Cole, P. J., A. J. Sinclair, J. F. Slattery, and D. Burke. 1984.Differentiation of Brucella ovis from Brucella abortus bygas-liquid chromatographic analysis of cellular fatty acids. J.Clin. Microbiol. 19:896-898.

29. Coloe, P. J., P. Cavanaugh, and J. Vaughan. 1986. The cellularfatty acid composition of Campylobacter species isolated fromcases of enteritis in man and animals. J. Hyg. 96:225-229.

30. Craven, R. B., J. B. Brooks, D. C. Edman, J. D. Converse, J.Greenlee, D. Schlossberg, T. Furlow, J. M. Gwaltney, Jr., andW. F. Miner. 1977. Rapid diagnosis of lymphocytic meningitisby frequency-pulsed electron capture gas-liquid chromatogra-phy: differentiation of tuberculous, cryptococcal, and viralmeningitis. J. Clin. Microbiol. 6:27-32.

31. Csango, P. A., N. Hagen, and G. Jagars. 1982. Method forisolation of Gardnerella vaginalis (Haemophilus vaginalis).Characterization of isolates by gas chromatography. ActaPathol. Microbiol. Immunol. Scand. 90:89-93.

32. Cundy, K. V., K. E. Willard, L. J. Valeri, C. J. Shanholtzer, J.Singh, and L. R. Peterson. 1991. Comparison of traditional gaschromatography (GC), headspace GC, and the microbial iden-tification library GC system for the identification of Clostrid-ium difficile. J. Clin. Microbiol. 29:260-263.

33. Daneshvar, M., and J. B. Brooks. 1988. Improved procedurefor preparation of pentafluorobenzyl derivatives of carboxylicacids for analysis by gas chromatography with electron-cap-ture detection. J. Chromatogr. 433:248-256.

34. Daneshvar, M. I., J. B. Brooks, and R. M. Winstead. 1987.Disposable reversed-phase chromatography columns for im-proved detection of carboxylic acids in body fluids by electron-capture gas-liquid chromatography. J. Clin. Microbiol. 25:1216-1220.

35. Dees, S., S. Thanabalasundrum, C. W. Moss, D. G. Hollis, andR. E. Weaver. 1980. Cellular fatty acid composition of groupIVe, a nonsaccharolytic organism from clinical sources. J.Clin. Microbiol. 11:664-668.

36. Dees, S. B., D. G. Hollis, R. E. Weaver, and C. W. Moss. 1981.Cellular fatty acids of Brucella canis and Brucella suis. J. Clin.Microbiol. 14:111-112.

37. Dees, S. B., D. G. Hollis, R. E. Weaver, and C. W. Moss. 1983.Cellular fatty acid composition of Pseudomonas marginateand closely associated bacteria. J. Clin. Microbiol. 18:1073-1078.

38. Dees, S. B., D. E. Karr, D. Hollis, and C. W. Moss. 1982.Cellular fatty acids of Capnocytophaga species. J. Clin. Mi-crobiol. 16:779-783.

39. Dees, S. B., and C. W. Moss. 1975. Cellular fatty acids ofAlcaligenes and Pseudomonas species isolated from clinicalspecimens. J. Clin. Microbiol. 1:414-419.

40. Dees, S. B., and C. W. Moss. 1978. Identification of Achromo-bacter species by cellular fatty acids and by production of ketoacids. J. Clin. Microbiol. 8:61-66.

41. Dees, S. B., C. W. Moss, D. G. Hollis, and R. E. Weaver. 1986.Chemical characterization of Flavobacterium odoratum, Fla-

vobacterium breve, and Flavobacterium-like groups Ie, IIh,and If. J. Clin. Microbiol. 23:267-273.

42. Dees, S. B., C. W. Moss, R. E. Weaver, and D. Hollis. 1979.Cellular fatty acid composition of Pseudomonas paucimobilisand groups IIk-2, Ve-1, and Ve-2. J. Clin. Microbiol. 10:206-209.

43. Dees, S. B., J. Powell, C. W. Moss, D. G. Hollis, and R. E.Weaver. 1981. Cellular fatty acid composition of organismsfrequently associated with human infections resulting from dogbites: Pasteurella multocida and groups EF-4, IIj, M-5, andDF-2. J. Clin. Microbiol. 14:612-616.

44. Drucker, D. B. 1974. Chemotaxonomic fatty-acid fingerprintsof some streptococci with subsequent statistical analysis. Can.J. Microbiol. 20:1723-1728.

45. Drucker, D. B., and S. M. Lee. 1981. Fatty acid fingerprints of"Streptococcus milleri," Streptococcus mitis, and related spe-cies. Int. J. Syst. Bacteriol. 31:219-225.

46. Drucker, D. B., and F. J. Veazey. 1977. Fatty acid fingerprintsof Streptococcus mutans NCTC 10832 grown at various tem-peratures. Apple. Environ. Microbiol. 33:221-226.

47. Durham, D. R., and W. E. Kloos. 1978. Comparative study ofthe total cellular fatty acids of Staphylococcus species ofhuman origin. Int. J. Syst. Bacteriol. 28:223-228.

48. Eerola, E., and O.-P. Lehtonen. 1988. Optimal data processingprocedure for automatic bacterial identification by gas-liquidchromatography of cellular fatty acids. J. Clin. Microbiol.26:1745-1753.

49. Fourche, J., M. Capdepuy, and J. Texier-Maugein. 1989. Gaschromatographic fatty acid determination to differentiate No-cardia asteroides, Mycobacterium fortuitum and Mycobacte-rium chelonei. J. Chromatogr. 478:142-146.

50. French, G. L., C. Y. Chan, S. W. Cheung, and K. T. Oo. 1987.Diagnosis of pulmonary tuberculosis by detection of tubercu-lostearic acid in sputum by using gas chromatography-massspectrometry with selected ion monitoring. J. Infect. Dis.156:356-362.

51. French, G. L., C. Y. Chan, S. W. Cheung, R. Teoh, M. J.Humphries, and G. 0. Mahony. 1987. Diagnosis of tuberculousmeningitis by detection of tuberculostearic acid in cerebrospi-nal fluid. Lancet ii:117-119.

52. Fugate, K. J., L. B. Hansen, and O. White. 1971. Analysis ofClostridium botulinum toxigenic types A, B, and E for fattyand carbohydrate content. Appl. Microbiol. 21:470-475.

53. Galanos, C., 0. Luderitz, E. T. Rietschel, 0. Westphal, H.Brade, L. Brade, M. Freudenberg, U. Schade, M. Imoto, H.Yoshimura, S. Kusumoto, and T. Shiba. 1985. Synthetic andnatural Escherichia coli free lipid A express identical endo-toxic activities. Eur. J. Biochem. 148:1-5.

54. Gangopadhyay, P. K., H. Thadepalli, I. Roy, and A. Ansari.1979. Identification of species of Candida, Cryptococcus, andTorulopsis by gas-liquid chromatography. J. Infect. Dis. 140:952-958.

55. Georgopapadakou, N. H., B. A. Dix, S. A. Smith, J. Freuden-berger, and P. T. Funke. 1987. Effect of antifungal agents on

lipid biosynthesis and membrane integrity in Candida albicans.Antimicrob. Agents Chemother. 31:46-51.

56. Goldfine, H. 1964. Composition of the aldehydes of Clostrid-ium butyricum plasmalogens. J. Biol. Chem. 239:2130-2134.

57. Goodwin, C. S., R. K. McCulloch, J. A. Armstrong, and S. H.Wee. 1985. Unusual cellular fatty acids and distinctive ultra-structure in a new spiral bacterium (Campylobacter pyloridis)from the human gastric mucosa. J. Med. Microbiol. 19:257-267.

58. Guerrant, G. O., M. A. Lambert, and C. W. Moss. 1981. Gaschromatographic analysis of mycolic acid cleavage products inmycobacteria. J. Clin. Microbiol. 13:899-907.

59. Gunasekaran, M., and W. T. Hughes. 1980. Gas-liquid chro-matography: a rapid method for identification of differentspecies of Candida. Mycologia 72:505-511.

60. Hofstad, T., and N. Skaug. 1980. Fatty acids and neutral sugarspresent in lipopolysaccharides isolated from Fusobacteriumspecies. Acta Pathol. Microbiol. Scand. 88:115-120.

61. Hollis, D. G., R. E. Weaver, A. G. Steigerwalt, J. D. Wenger,

VOL. 4, 1991

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

CLIN. MICROBIOL. REV.

C. W. Moss, and D. J. Brenner. 1989. Francisella philomiragiacomb. nov. (formerly Yersinia philomiragia) and Francisellatularensis biogroup novicida (formerly Francisella novicida)associated with human disease. J. Clin. Microbiol. 27:1601-1608.

62. James, A. T., and A. J. P. Martin. 1952. Gas-liquid partitionchromatography: the separation and micro-estimation of vola-tile fatty acids from formic acid to dodecanoic acid. Biochem.J. 50:679-690.

63. Jantzen, E., B. P. Berdal, and T. Omland. 1979. Cellular fattyacid composition of Francisella tularensis. J. Clin. Microbiol.10:928-930.

64. Jantzen, E., B. P. Berdal, and T. Omland. 1980. Cellular fattyacid composition of Haemophilus species, Pasteurella multo-cida, Actinobacillus actinomycetemcomitans and Haemophi-lus vaginalis (Corynebacterium vaginale). Acta Pathol. Micro-biol. Scand. 88:89-93.

65. Jantzen, E., T. Bergan, and K. Bovre. 1974. Gas chromatogra-phy of bacterial whole cell methanolysates. VI. Fatty acidcomposition of strains within Micrococcaceae. Acta Pathol.Microbiol. Scand. Microbiol. Immunol. 82:785-798.

66. Jantzen, E., K. Bryn, T. Bergan, and K. Bovre. 1974. Gaschromatography of bacterial whole cell methanolysates. V.Fatty acid composition of Neisseriae and Moraxellae. ActaPathol. Microbiol. Scand. Microbiol. Immunol. 82:767-779.

67. Jantzen, E., and T. Hofstad. 1981. Fatty acids of Fusobacter-ium species: taxonomic implications. J. Gen. Microbiol. 123:163-171.

68. Jantzen, E., E. Knudsen, and R. Winsnes. 1982. Fatty acidanalysis for differentiation of Bordetella and Brucella species.Acta Pathol. Microbiol. Immunol. Scand. 90:353-359.

69. Jantzen, E., T. Tangen, and J. Eng. 1989. Gas chromatographyof mycobacterial fatty acids and alcohols: diagnostic applica-tions. Acta Pathol. Microbiol. Immunol. Scand. 97:1037-1045.

70. Kaneda, T. 1967. Fatty acids in the genus Bacillus. I. Iso- andanteiso-fatty acids as characteristic constituents of lipids in 10species. J. Bacteriol. 93:894-903.

71. Kawa, D. E., D. R. Pennell, L. N. Kubista, and R. F. Schell.1989. Development of a rapid method for determining thesusceptibility of Mycobacterium tuberculosis to isoniazid usingthe Gen-Probe DNA hybridization system. Antimicrob.Agents Chemother. 33:1000-1005.

72. Knivett, V. A., and J. Cullen. 1965. Some factors affectingcyclopropane acid formation in Escherichia coli. Biochem. J.96:771-776.

73. Kobayashi, K., H. Suzinaka, and I. Yano. 1987. Analysis offatty acid compositions of various yeasts by gas-liquid chro-matography using a polar column. Microbios 51:37-42.

74. Kondo, E., and N. Ueta. 1972. Composition of fatty acids andcarbohydrates in Leptospira. J. Bacteriol. 110:459-467.

75. Kostiw, L. L., E. E. Vicher, and I. Lyon. 1966. The fatty acidsof Trichophyton rubrum. Mycopathol. Mycol. Apple. 29:145-150.

76. Kotilainen, P., P. Huovinen, and E. Eerola. 1991. Applicationof gas-liquid chromatographic analysis of cellular fatty acidsfor species identification and typing of coagulase-negativestaphylococci. J. Clin. Microbiol. 29:315-322.

77. Kusaka, T., and S. Izumi. 1983. Gas chromatography ofconstitutive fatty acids in Mycobacterium leprae. Microbiol.Immunol. 27:409-414.

78. Lambert, M., and C. W. Moss. 1983. Comparison of the effectsof acid and base hydrolysis on hydroxy and cyclopropane fattyacids in bacteria. J. Clin. Microbiol. 18:1370-1377.

79. Lambert, M. A., and A. Y. Armfield. 1979. Differentiation ofPeptococcus and Peptostreptococcus by gas-liquid chromatog-raphy of cellular fatty acids and metabolic products. J. Clin.Microbiol. 10:464-476.

80. Lambert, M. A., and C. W. Moss. 1976. Cellular fatty acidcomposition of Streptococcus mutans and related strepto-cocci. J. Dent. Res. 55:A96-102.

81. Lambert, M. A., and C. W. Moss. 1989. Cellular fatty acidcompositions and isoprenoid quinone contents of 23 Legionellaspecies. J. Clin. Microbiol. 27:465-573.

82. Lambert, M. A., C. W. Moss, V. A. Silcox, and R. C. Good.1986. Analysis of mycolic acid cleavage products and cellularfatty acids of Mycobacterium species by capillary gas chroma-tography. J. Clin. Microbiol. 23:731-736.

83. Lambert, M. A., C. M. Patton, T. J. Barrett, and C. W. Moss.1987. Differentiation of Campylobacter and Campylobacter-like organisms by cellular fatty acid composition. J. Clin.Microbiol. 25:706-13.

84. Larsson, L., J. Jimenez, A. Sonesson, and F. Portaels. 1989.Two-dimensional gas chromatography with electron capturedetection for the sensitive determination of specific mycobac-terial lipid constituents. J. Clin. Microbiol. 27:2230-2233.

85. Larsson, L., A. Sonesson, and J. Jimenez. 1987. Ultrasensitiveanalysis of microbial fatty acids using gas chromatographywith electron capture detection. Eur. J. Clin. Microbiol. 6:729-731.

86. Lewis, V. J., R. E. Weaver, and D. G. Hollis. 1968. Fatty acidcomposition of Neisseria species as determined by gas chro-matography. J. Bacteriol. 96:1-5.

87. Luquin, M., V. Ausina, F. L. Calahorra, F. Belda, M. G.Barcelo, C. Celma, and G. Prats. 1991. Evaluation of practicalchromatographic procedures for identification of clinical iso-lates of mycobacteria. J. Clin. Microbiol. 29:120-130.

88. Machtiger, N. A., and W. M. O'Leary. 1973. Fatty acidcomposition of paracolons: Arizona, Citrobacter, and Provi-dencia. J. Bacteriol. 114:80-85.

89. Maliwan, N., R. W. Reid, S. R. Plisha, T. J. Bird, and J. R.Zvetina. 1988. Identifying Mycobacterium tuberculosis cul-tures by gas-liquid chromatography and a computer-aidedpattern recognition model. J. Clin. Microbiol. 26:182-187.

90. Marr, A. G., and J. L. Ingraham. 1962. Effect of temperatureon the composition of fatty acids in Escherichia coli. J.Bacteriol. 84:1260-1267.

91. Marumo, K., and Y. Aoki. 1990. Discriminant analysis ofcellular fatty acids of Candida species, Torulopsis glabrata,and Crytococcus neoformans determined by gas-liquid chro-matography. J. Clin. Microbiol. 28:1509-1513.

92. Matthews, H. M., T. K. Yang, and H. M. Jenkin. 1979. Uniquelipid composition of Treponema pallidum (Nichols virulentstrain). Infect. Immun. 24:713-719.

93. Mayall, B. C. 1985. Rapid identification of mycobacteria usinggas liquid chromatography. Pathology 17:24-28.

94. Mayberry, W. R. 1980. Hydroxy fatty acids in Bacteroidesspecies: D-(-)-3-hydroxy-15-methylhexadecanoate and its ho-mologs. J. Bacteriol. 143:582-587.

95. Mayberry, W. R., D. W. Lamb, Jr., and K. P. Ferguson. 1982.Identification of Bacteroides species by cellular fatty acidprofiles. Int. J. Syst. Bacteriol. 32:21-27.

96. Miller, L. T. 1982. Single derivatization method for routineanalysis of bacterial esters, including hydroxy acids. J. Clin.Microbiol. 16:584-586.

97. Modak, M. J., S. Nair, and A. Venkataraman. 1970. Studies onthe fatty acid composition of some salmonellas. J. Gen. Micro-biol. 60:151-157.

98. Moss, C. W. 1981. Chromatographic analysis: a new future forclinical microbiology, p. 147-150. In R. C. Tilton (ed.), Rapidmethods and automation in microbiology. Proceedings of theThird International Symposium on Rapid Methods and Auto-mation in Microbiology. American Society for Microbiology,Washington, D.C.

99. Moss, C. W. 1981. Gas-liquid chromatography as an analyticaltool in microbiology. J. Chromatogr. 203:337-347.

100. Moss, C. W., W. F. Bibb, D. E. Karr, G. 0. Guerrant, andM. A. Lambert. 1983. Cellular fatty acid composition andubiquinone content of Legionella feeleii sp. nov. J. Clin.Microbiol. 18:917-919.

101. Moss, C. W., and S. B. Dees. 1976. Cellular fatty acids andmetabolic products of Pseudomonas species obtained fromclinical specimens. J. Clin. Microbiol. 4:492-502.

102. Moss, C. W., and S. B. Dees. 1978. Cellular fatty acids ofFlavobacterium meningosepticum and Flavobacterium speciesgroup lib. J. Clin. Microbiol. 8:772-774.

103. Moss, C. W., and S. B. Dees. 1979. Further studies of the

436 WELCH

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

APPLICATIONS OF CELLULAR FATTY ACID ANALYSIS 437

cellular fatty acid composition of Legionnaires disease bacte-ria. J. Clin. Microbiol. 9:648-649.

104. Moss, C. W., S. B. Dees, and G. 0. Guerrant. 1980. Gas-liquidchromatography of bacterial fatty acids with fused-silica cap-illary column. J. Clin. Microbiol. 12:127-130.

105. Moss, C. W., V. R. Dowell, Jr., D. Farshtchi, L. J. Raines, andW. B. Cherry. 1969. Cultural characteristics and fatty acidcomposition of propionibacteria. J. Bacteriol. 97:561-570.

106. 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.

107. Moss, C. W., D. S. Kellog, Jr., D. C. Farshy, M. A. Lambert,and J. D. Thayer. 1970. Cellular fatty acids of pathogenicNeisseria. J. Bacteriol. 104:63-68.

108. Moss, C. W., D. G. Hollis, and R. E. Weaver. 1988. Culturaland chemical characterization of CDC groups EO-2, M-5, andM-6, Moraxella (Moraxella) species, Oligella urethralis, Aci-netobacter species, and Psychrobacter immobilis. J. Clin.Microbiol. 26:484-492.

109. Moss, C. W., G. Holzer, P. L. Wallace, and D. G. Hollis. 1990.Cellular fatty acid compositions of an unidentified organismand a bacterium associated with cat scratch disease. J. Clin.Microbiol. 28:1071-1074.

110. Moss, C. W., M. A. Lambert, and G. L. Lombard. 1977.Cellular fatty acids of Peptococcus variabilis and Peptostrep-tococcus anaerobius. J. Clin. Microbiol. 5:665-667.

111. Moss, C. W., M. A. Lambert, and W. H. Merwin. 1974.Comparison of rapid methods for analysis of bacterial fattyacids. Apple. Microbiol. 28:80-85.

112. Moss, C. W., and V. J. Lewis. 1967. Characterization ofclostridia by gas chromatography. Appl. Microbiol. 15:390-397.

113. Moss, C. W., and S. B. Samuels. 1972. Cellular fatty acidcomposition of selected Pseudomonas species. Apple. Micro-biol. 24:596-598.

114. Moss, C. W., T. Shinoda, and J. W. Samuels. 1982. Determi-nation of cellular fatty acid compositions of various yeasts bygas-liquid chromatography. J. Clin. Microbiol. 16:1073-1079.

115. Moss, C. W., R. E. Weaver, S. B. Dees, and W. B. Cherry.1977. Cellular fatty acid composition of isolates from Legion-naires' disease. J. Clin. Microbiol. 6:140-143.

116. Mukwaya, G. M., and D. F. Welch. 1989. Subgrouping ofPseudomonas cepacia by cellular fatty acid composition. J.Clin. Microbiol. 27:2640-2646.

117. Muranishi, H., M. Nakashima, R. Isobe, T. Ando, and N.Shigematsu. 1990. Measurement of tuberculostearic acid insputa, pleural effusions, and bronchial washings. Diagn. Mi-crobiol. Infect. Dis. 13:235-240.

118. Nichols, P. D., W. R. Mayberry, C. P. Antworth, and D. C.White. 1985. Determination of monounsaturated double-bondposition and geometry in the cellular fatty acids of the patho-genic bacterium Francisella tularensis. J. Clin. Microbiol.21:738-740.

119. Nishanen, A., T. Kiutamo, S. Raisanen, and M. Raevuori. 1978.Determination of fatty acid compositions of Bacillus cereusand related bacteria: a rapid gas chromatographic methodusing a glass capillary column. Apple. Environ. Microbiol.35:453-455.

120. O'Brien, M., and R. Colwell. 1987. Characterization tests fornumerical taxonomy studies. Methods Microbiol. 19:69-104.

121. Odham, G., L. Larsson, and P. A. Mardh. 1987. Demonstrationof tuberculostearic acid in sputum from patients with pulmo-nary tuberculosis by selected ion monitoring. J. Clin. Invest.63:813-819.

122. O'Donnell, A. G., M. R. Nahaie, M. Goodfellow, D. E. Minni-kin, and V. Hajek. 1985. Numerical analysis of fatty acidprofiles in the identification of staphylococci. J. Gen. Micro-biol. 131:2023-2033.

123. O'Leary, W. M. 1975. The chemistry of microbial lipids. Crit.Rev. Microbiol. 4:41-63.

124. Olson, W. P., M. J. Groves, and M. E. Klegerman. 1989. Fattyacids of Mycobacterium bovis BCG. Microbios 57:151-155.

125. Prefontaine, G., and F. L. Jackson. 1972. Cellular fatty acid

profiles as an aid to the classification of "corroding bacilli" andcertain other bacteria. Int. J. Syst. Bacteriol. 22:210-217.

126. Raines, L. J., C. W. Moss, D. Farshtchi, and B. Pittman. 1968.Fatty acids of Listeria monocytogenes. J. Bacteriol. 96:2175-2177.

127. Rodwall, A. 1968. Fatty-acid composition of Mycoplasmalipids: biomembrane with only 1 fatty acid. Science 160:1350-1351.

128. Romesburg, H. C. 1990. Cluster analysis for researchers.Robert E. Krieger Publishing Co., Malbar, Fla.

129. Rose, A. H. 1968. Chemical microbiology, p. 4-56. Butter-worths, London.

130. Ruoff, K. L., M. J. Ferraro, J. Holden, and L. J. Kunz. 1984.Identification of Streptococcus bovis and Streptococcus sali-varius in clinical laboratories. J. Clin. Microbiol. 20:223-226.

131. Samuels, S. B., C. W. Moss, and R. E. Weaver. 1973. The fattyacids of Pseudomans multivorans (Pseudomans cepacia) andPseudomans kingii. J. Gen. Microbiol. 74:275-279.

132. Sasser, M. 1990. Identification of bacteria through fatty acidanalysis, p. 199-204. In Z. Klement, K. Rudolph, and D. Sands(ed.), Methods in phytobacteriology. Akademiai Kiado,Budapest.

133. Schlossberg, D., J. B. Brooks, and J. A. Shulman. 1976. Thepossibility of diagnosing meningitis by gas chromatography:cryptococcal meningitis. J. Clin. Microbiol. 3:239-245.

134. Shaw, N. 1974. Lipid composition as a guide to the classifica-tion of bacteria. Adv. Appl. Microbiol. 17:63-108.

135. Slater, L. N., D. F. Welch, D. Hensel, and D. W. Coody. 1990.A newly recognized fastidious gram-negative pathogen as acause of fever and bacteremia. N. Engl. J. Med. 323:1587-1593.

136. Smith, T. M., T. J. Brooks, Jr., and H. B. White, Jr. 1969.Fatty acid composition of adult Schistosoma mansoni. Lipids4:31-36.

137. Starratt, A. N., and C. Madhosingh. 1967. Sterol and fatty acidcomponents of mycelium of Fusarium oxysporum. Can. J.Microbiol. 13:1351-1355.

138. Tenover, F. C. 1988. Diagnostic deoxyribonucleic acid probesfor infectious diseases. Clin. Microbiol. Rev. 1:82-101.

139. Theodore, T. S., and C. Panos. 1973. Protein and fatty acidcomposition of mesosomal vesicles and plasma membranes ofStaphylococcus aureus. J. Bacteriol. 116:571-576.

140. Thiele, 0. W., J. Asselineau, and C. Lacave. 1969. On the fattyacids of Brucella abortus and Brucella melitensis. Eur. J.Biochem. 7:393-396.

141. Thoen, C. O., A. G. Karlson, and R. D. Ellefson. 1972.Differentiation between Mycobacterium kansasii and Myco-bacterium marinum by gas-liquid chromatographic analysis ofcellular fatty acids. Apple. Microbiol. 24:1009-1010.

142. Timm, S. L., A. D. Pereira-Netto, and M. M. Oliveira. 1982.Fatty acids of Trypanosoma cruzi. Comp. Biochem. Physiol.71:397-402.

143. Tisdall, P. A., D. R. DeYoung, G. D. Roberts, and J. P. Anhalt.1982. Identification of clinical isolates of mycobacteria withgas-liquid chromatography: a 10-month follow-up study. J.Clin. Microbiol. 16:400-402.

144. Tisdall, P. A., G. D. Roberts, and J. P. Anhalt. 1979. Identifi-cation of clinical isolates of mycobacteria with gas-liquidchromatography alone. J. Clin. Microbiol. 10:506-514.

145. Tornabene, T. G. 1973. Lipid composition of selected strains ofYersinia pestis and Yersinia pseudotuberculosis. Biochim.Biophys. Acta 306:173-185.

146. Tornabene, T. G. 1985. Lipid analysis and the relationship tochemotaxonomy. Methods Microbiol. 18:209-234.

147. Trotter, J. E., T. L. Kuhls, D. A. Pickett, S. Reyes de la Rocha,and D. F. Welch. 1990. Pneumonia caused by a newly recog-nized pseudomonad in a child with chronic granulomatousdisease. J. Clin. Microbiol. 28:1120-1124.

148. Tzianabos, T., C. W. Moss, and J. E. McDade. 1981. Fatty acidcomposition of Rickettsiae. J. Clin. Microbiol. 13:603-605.

149. Urdaci, M. C., M. Marchand, and P. A. Grimont. 1990.Characterization of 22 Vibrio species by gas chromatography

VOL. 4, 1991

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from

CLIN. MICROBIOL. REV.

analysis of their cellular fatty acids. Res. Microbiol. 141:437-452.

150. Valero-Guillien, P., F. Martin-Luengo, L. Larsson, J. Jimenez,I. Juhlin, and F. Portaels. 1988. Fatty and mycolic acids ofMycobacterium malmoense. J. Clin. Microbiol. 26:153-154.

151. Valero-Guillen, P. L., and F. Martin-Luengo. 1983. A gas lipidand thin-layer chromatographic study of Mycobacterium for-tuitum. Tubercle 64:283-290.

152. Vandamme, P., E. Falsen, R. Rossau, B. Hoste, P. Segers, R.Tytgat, and J. De Ley. 1991. Revision of Campylobacter,Helicobacter, and Wolinella taxonomy: emendation of genericdescriptions and proposal of Arcobacter gen. nov. Int. J. Syst.Bacteriol. 41:88-103.

153. Van der Auwera, P., M. Labbe, W. R. Mayberry, K. P.Ferguson, and D. W. Lambe, Jr. 1986. Identification of Bac-teroides by cellular fatty acid profiles: application to theroutine microbiological laboratory. J. Microbiol. Methods4:267-275.

154. Vasyurenko, Z. P., and Y. N. Chernyavskaya. 1990. Confirma-tion of Morganella distinction from Proteus and Providenciaamong Enterobacteriaceae on the basis of cellular and lipo-polysaccharide fatty acid composition. J. Hyg. Epidemiol.Microbiol. Immunol. 34:81-90.

155. Vessal, M., H. R. Rezai, and P. Pakzad. 1974. Leishmaniaspecies: fatty acid composition of promastigotes. Exp. Parasi-tol. 36:455-461.

156. Veys, A., W. Callewaert, E. Waelkens, and K. Van den Abbeele.1989. Application of gas-liquid chromatography to the routineidentification of nonfermenting gram-negative bacteria in clin-ical specimens. J. Clin. Microbiol. 27:1538-1542.

157. Wallace, P. L., D. G. Hollis, R. E. Weaver, and C. W. Moss.1988. Cellular fatty acid composition of Kingella species,

Cardiobacterium hominis, and Eikenella corrodens. J. Clin.Microbiol. 26:1592-1594.

158. Wallace, P. L., D. G. Hollis, R. E. Weaver, and C. W. Moss.1989. Characterization of CDC group DF-3 by cellular fattyacid analysis. J. Clin. Microbiol. 27:735-737.

159. Wallace, P. L., D. G. Hollis, R. E. Weaver, and C. W. Moss.1990. Biochemical and chemical characterization of pink-pig-mented oxidative bacteria. J. Clin. Microbiol. 28:680-693.

160. Wallace, W. R. 1966. Fatty acid composition of lipid classes inPlasmodium lophurae and Plasmodium berghei. Am. J. Trop.Med. Hyg. 15:811-813.

161. Weintraub, A., U. Zahringer, H.-W. Wollenweber, U. Seydel,and E. T. Rietschel. 1989. Structural characterization of thelipid A component of Bacteroides fragilis strain NCTC 9343lipopolysaccharide. Eur. J. Biochem. 183:425-431.

162. Westfall, H. N., D. C. Edman, and E. Weiss. 1984. Analysis offatty acids of the genus Rochalimaea by electron capture gaschromatography: detection of nonanoic acid. J. Clin. Micro-biol. 19:305-310.

163. White, M. A., M. D. Simmons, A. Bishop, and H. A. Chandler.1988. Microbial identification by gas chromatography. J. R.Nav. Med. Serv. 74:141-146.

164. Wirth, J. C., and S. R. Anand. 1964. The fatty acids ofTrichophyton rubrum. Can. J. Microbiol. 10:23-27.

165. Wollenweber, H.-W., S. Schramek, H. Moll, and E. T. Riet-schel. 1985. Nature and linkage type of fatty acids present inlipopolysaccharides of phase I and II Coxiella burnetii. Arch.Microbiol. 142:6-11.

166. Wong, B., F. F. Edwards, and T. E. Kiehn. 1985. Continuoushigh-grade Mycobacterium avium-intracellulare bacteremia inpatients with the acquired immunodeficiency syndrome. Am.J. Med. 78:35-40.

438 WELCH

on May 16, 2020 by guest

http://cmr.asm

.org/D

ownloaded from