INFECTION AND IMMUNITY, Sept. 1974, p. 481-488Copyright © 1974 American Society for Microbiology

Vol. 10, No. 3Printed in U.S.A.

Growth Inhibition Among Strains of Neisseria gonorrhoeaedue to Production of Inhibitory Free Fatty Acids and

Lysophosphatidylethanolamine: Absence of BacteriocinsDIANA L. WALSTAD, RONALD C. REITZ, AND P. FREDERICK SPARLING

Departments of Medicine, Bacteriology and Immunology, and Biochemistry, University of North CarolinaSchool of Medicine, Chapel Hill, North Carolina 27514

Received for publication 15 April 1974

Each of 50 tested strains of Neisseria gonorrhoeae produced growth-inhibitorysubstances that were active against most strains of gonococci or meningococci,but not against species other than the Neisseria. There were quantitativedifferences among different strains in production of the inhibitor and sensitivityto it, but not of sufficient magnitude to permit routine strain typing. Theinhibitor was associated with the cell pellet (crude cell envelope) and was notinducible with mitomycin C. Inhibitory activity was thermostable and resistedalkali and proteolytic enzymes. The inhibitor was quantitatively recovered fromwhole cells by chloroform-methanol extraction. Separation of total gonococcallipids by silica gel chromatography revealed inhibitory activity in both the freefatty acid and the phospholipid fractions. The major phospholipid, phos-phatidylethanolamine, had no inhibitory activity, but monoacyl phosphatidyl-ethanolamine, a minor phospholipid, was quite inhibitory. It is likely that the"bacteriocin" of N. gonorrhoeae strains results from the degradation of phos-phatidylethanolamine to inhibitory long-chain free fatty acids and monoacylphosphatidylethanolamine.

There is no satisfactory means for classifyingclinical isolates of Neisseria gonorrhoeae. Bac-teriocin typing has been used to classify severalspecies of bacteria including Escherichia coliand Pseudomonas aeruginosa. (6, 24). Thismethod is dependent upon the production of aproteinaceous antibacterial substance that ex-hibits strain-specific growth inhibition of mem-bers of the same, or closely related, species. In1966 Kingsbury (17) isolated a bacteriocin froma strain of N. meningitidis, and Counts et al. (3)subsequently employed bacteriocin typing ofmeningococci as an epidemiological tool. Flynnand McEntegart recently reported the apparentsuccess of a similar "bacteriocin" typingscheme for gonococci (7). In this report weexplore the potential efficacy of strain typinggonococci by these methods and the mech-anisms of growth inhibition between gonococci.

MATERIALS AND METHODSMicroorganisms and media. Strains of N. gonor-

rhoeae included both recent clinical isolates fromDurham, N.C., and older laboratory strains obtainedfrom widely varying geographical regions of theUnited States, Europe, and Asia. (The latter werekindly supplied by R. Martin, Atlanta; A. Reyn,Copenhagen; and G. Brooks, Indianapolis, who also

sent six strains used by Flynn and McEntegart [7].)All strains were characterized as N. gonorrhoeae byGram stain, oxidase reaction, and sugar fermenta-tions. Strains of N. meningitidis were obtained locallyfrom clinical sources, and their identity was con-firmed by fermentation of glucose and maltose but notsucrose, lactose, or fructose. Colicin-producing andindicator strains of E. coli kindly supplied by J. K.Davies were: K53 (colicin El), CA42 (colicin E2),CA38 (colicin E3), K235 (colicin K), and AB1133(indicator strain sensitive to colicins El, E2, E3, andK). Pyocin-producing (P1 and P10) and indicatorstrains (Ti and T5) of Pseudomonas aeruginosa wereobtained from J. J. Farmer; the pyocin-producingstrains were induced with mitomycin C as describedby Farmer and Herman (6). Strains of other specieswere from the local culture collection.

Cultures were stored at -70 C in Trypticase soybroth (BBL) containing 20% glycerol. Plate mediaused throughout was GC base medium (Difco) plus1% defined supplements 1 and 2 (GCBDS medium)(30). Soft agar for overlays consisted of Mueller-Hin-ton broth (Difco), 0.6% agar, and 2% supplements 1and 2. Diphasic liquid cultures consisted of a lowerphase containing 10 ml of Mueller-Hinton or GC basemedium with 2% agar and an upper phase containing10 ml of Mueller-Hinton or peptone broth plus 2%supplements in 125-ml Erlenmeyer flasks. Peptonebroth consisted of 1.5% proteose peptone no. 3 (Difco),0.4% K2HPO4, 0.1% KH2PO4, and 0.5% NaCl. Mini-

481

on October 16, 2017 by guest

http://iai.asm.org/

Dow

nloaded from

WALSTAD, REITZ, AND SPARLING

mal medium A of Davis and Mingioli (4) (minimalbroth Davis, Difco) was prepared without glucose andwas used for suspending and diluting cells. All cul-tures were incubated at 37 C in a 5% CO2 incubator.

Chemicals. Mitomycin C was from Mann Re-search. Fatty acids were purchased from Nu-Chek-Prep Inc. (Elysian, Minn.). Phosphatidylethanola-mine (PE) and monoacyl PE (lyso PE) were obtainedfrom Applied Sciences Laboratories (State College,Pa.), and phosphatidylglycerol (PG) was from Supel-co, Inc. (Bellefonte, Pa.). Monoacyl phosphatidycho-line (PC) was prepared by treating PC from egg yolkwith phospholipase A2 (26), and the monoacyl PC wasseparated from the free fatty acids (FFA) before useby preparative thin-layer chromatography on silicagel by using a petroleum ether-diethyl ether-aceticacid developing solvent (70:30: 1). Pronase (grade B)was from Calbiochem. Bovine serum albumin waspurchased from Sigma Chemical Co. (St. Louis) intwo forms, purified fraction V and fatty acid free. Allother chemicals were of highest available purity.

Inhibition of gonococci by whole cells or cellextracts. A replica-inoculating device was used toinoculate 11 to 21 "producer" strains onto agar plates(usually GCBDS). After incubation for 24 to 48 h, theplates were exposed to chloroform vapor for 1 min,and then overlaid with 0.2 ml of an indicator strainsuspension (adjusted to 100 Klett units) in 3.0 ml ofsoft agar ("double-layer" technique [8]). After over-night incubation, the size of the clear zone of inhibi-tion of the indicator strain around each underlyingproducer strain was recorded. Most strains were alsotested by the "cross-streaking" method described byFredericq (9) and modified by Flynn (7). Inhibition byextracts of whole cells or by purified reagents wasdetermined by spotting 0.01 ml of the materials to betested onto GCBDS plates, which were allowed to drybefore being overlaid by the indicator strain in softagar. Quantitation of inhibitory activity of suspen-sions of cells or of cell extracts was expressed as thehighest dilution resulting in unequivocal growth inhi-bition. All tests were performed in duplicate.

Mitomycin C induction of growth inhibitor.Eleven gonococcal strains exhibiting strong inhibitionof other strains by the double-overlay or cross-streakmethods were tested for mitomycin C inducibility oftheir putative bacteriocin (7). Early log-phase cellswere incubated at 33 C to 38 C for 5 h in brothcontaining 0.01 to 0.32 Ag of mitomycin C per ml.The concentrations of mitomycin C were slightly lessthan the minimal growth-inhibiting concentration foreach strain. After mitomycin induction, the cells werewashed by centrifugation and incubated for another17 h in fresh broth. The cells were then pelleted bycentrifugation for 10 min at 20,000 x g, and the cellpellet and supernatant fluid were tested for inhibitoryactivity against sensitive indicator strains. Brothmedia included Trypticase soy broth, peptone broth,and Mueller-Hinton broth, with addition of 2% de-fined supplements (30) or 2% supplement C (Difco).Incubations were performed with varying degrees ofaeration or environmental CO2. In some instances,samples were taken hourly during mitomycin C in-

duction to detect possible presence of a labile growthinhibitor.

Attempts to solubilize the inhibitor. Strains re-sulting in definite growth inhibition of other gonococciwere grown for 24 h on GCBDS plates, suspended inminimal medium A, and washed twice by centrifuga-tion. Whole cells were ruptured at 0 C in a Frenchpressure cell at 20,000 psi. Deoxyribonuclease (2,ug/ml) was added, and the cell pellet (crude envelopefraction) and supernatant were collected by centrifu-gation for 30 min at 30,000 x g (4 C). The cell pelletcontaining most of the inhibitor was washed twice,suspended, and stirred for 15 min to several hours atroom temperature in one of several reagents describedas being effective in removing proteins from membra-nous material (23): 1 and 5 M NaCl; 0.85% KCl, 2 and20 mM ethylenediaminetetraacetic acid; n-butanol;and a series of buffers of increasing alkalinity [0.025M tris(hydroxymethyl)aminomethane-hydrochloride,pH 7.0 to 9.0; Sorenson glycine-sodium hydroxidebuffer, pH 10.0 to 12.0 (5)]. The treated cell pelletsuspensions were centrifuged for 30 min (30,000 x g,4 C), and the centrifugate was resuspended to itsoriginal volume in minimal medium A. The superna-tant and extracted cell pellet, in addition to theextracting reagent alone (as control), were tested forinhibitory activity.Pronase digestion. Whole gonococcal cells or the

alkali-extracted inhibitor, soluble pyocins, and coli-cins were exposed to 2 mg of Pronase per ml in 0.02M tris(hydroxymethyl)aminomethane-hydrochloride(pH 7.4) for 2 h at 37 C. Sensitivity of gonococcalgrowth inhibitor to Pronase produced during growthon GCBDS plates was determined in the doubleoverlay technique by incorporating 2 mg of Pro-nase per ml in the soft agar overlay containing theindicator strain.

Quantitative lipid extraction of the inhibitorfrom whole cells. One to two grams of cells, removedby scraping from 24-h GCBDS plate cultures, wassuspended in minimal medium A to a final volume ofabout 4.5 ml. A 2.0-ml volume of the cell suspensionwas set aside, and another 2.0-ml sample was pipettedinto a Waring blender and extracted for lipids withchloroform-methanol (1: 2) essentially as described byGarbus (12). Great care was taken to insure maximalrecovery of the lipids; this included two additionalchloroform washes of the cell material after acidifica-tion with 0.08 ml of 50% H2S04. The combinedchloroform extracts were desiccated at room tempera-ture under vacuum, and the lipids were resuspendedin 2.0 ml of chloroform. Serial dilutions of the lipidsolution and the control, unextracted cell suspensionwere tested for inhibitory activity as described.

Extraction and separation of the lipids. A 100-mgamount of lyophilized cells was suspended in 2.5 ml ofdistilled water. Total lipids were extracted by themethod of Bligh and Dyer (1) and taken up in 5 ml ofbenzene. Separation of the lipids from 3 jg of eachextract was accomplished on Gelman ITLC (type SG)chromatography sheets. The solvent systems wereisooctane-ethyl acetate (100:3.4, vol/vol) for neutrallipids and chloroform-10% acetic acid in methanol

482 INFECT. IMMUNITY

on October 16, 2017 by guest

http://iai.asm.org/

Dow

nloaded from

VOL. 10, 1974 GROWTH INHIBITION BETWEEN N. GONORRHOEAE STRAINS

(100: 15, vol/vol) for phospholipids. Great care wastaken to insure that the ITLC papers and the solventwere free of water. Such caution was necessary toachieve the separation ofPG from PE. The chromato-grams were dried and either charred with H2SO4 ordeveloped with ninhydrin. Appropriate neutral andphospholipid standards were employed as controls.

Preparative thin layer chromatography was used inexperiments to determine which lipids were growthinhibitory. A 1.5-ml volume of the lipid extract wasconcentrated, spotted onto Silica Gel G plates (0.25mm), and developed with petroleum ether-ethylether-acetic acid (70:30: 1). The lipid bands were lo-cated by spraying only the part of the chromatogramcontaining the standards with 2, 7-dichlorofluoresceinand using ultraviolet light. The bands representingthe phospholipids,. the FFA, and the solvent frontwere scraped from the plate and eluted with chloro-form-methanol (2: 1, vol/vol). The eluates were evap-orated under vacuum or N2, suspended in chloro-form, and tested for inhibition of growth of indicatorstrain FA5.

The fatty acid concentrations were determined onthe total lipid extract of N. gonorrhoeae FA5 and N.meningitidis NM33 and on the FFA fraction fromFA5. The procedure of Metcalf et al. (21) was used toprepare the methyl esters which were then separatedon a column (6 foot by 0.08 inch, intemal diameter[ca. 1.8 m by 0.2 cm]) packed with 12% EGSS-Y asdescribed previously (25). Arachidic acid (20:0) wasused as an internal standard since only trace amountsof this acid were found.

RESULTSApparent bacteriocin production. Each of

50 tested strains of N. gonorrhoeae producedsubstances that inhibited the growth of othergonococci. The results were difficult to repro-duce by the cross-streak method of Flynn andMcEntegart (7), but were more reproducible bythe double-overlay method (Fig. 1). The size ofthe zone of inhibition of the indicator strainvaried according to several variables. The omis-sion of starch or the substitution of Mueller-Hinton medium for GCBDS resulted in largerzones of inhibition. Zone sizes and strain differ-ences in production of inhibitor were maximizedif the plates containing 24-h growth of producerstrains were refrigerated for 48 h prior to theoverlaying with the indicator strain. The clonaltype (15) of the producer or the indicator strainhad no significant effect on the amount ofinhibition observed.Some gonococci clearly showed a more notice-

able growth inhibitory effect than others (Fig.1), but all 50 strains inhibited each of 30indicator gonococci to some degree. (In somecases, this was observable only by visualizingthe area over the CHCl3-killed producer spotwith a dissecting microscope.) There was no

...

FIG. 1. Inhibition of N. gonorrhoeae strain FA18 by11 selected strains of N. gonorrhoeae. GCBDS plateswere incubated for 24 h and refrigerated for 48 hbefore being overlaid with the indicator strain. StrainFA18 is relatively sensitive to inhibition; "resistant"strains showed barely discemible zones of inhibition.

evidence for immunity of producer strains totheir own inhibitors. Sensitivity to inhibitionalso varied, but the differences were relativelysmall. The differences in apparent productionof inhibitor and in sensitivity to inhibition weresufficient for grouping strains into four arbitrarycategories (either good or poor producer, orsensitive or resistant indicator), but furthersubdivisions were impossible.

Specificity of growth inhibition. Thegrowth inhibitors produced by 12 tested strainsof gonococci were active against each of the sixstrains of N. meningitidis tested, but notagainst one strain of N. lactamicus or againstany of five strains each of E. coli, Enterobacter,Klebsiella, Proteus, group A streptococci, andStaphylococcus aureus. Two of five strains of P.aeruginosa were slightly inhibited.

Isolation of the inhibitor. All attempts toinduce production of increased amounts of thegrowth inhibitor with mitomycin C were unsuc-cessful. No inhibitory activity was found in thesupernatant of mitomycin-induced cultures,and the activity associated with the cell pelletwas not increased.When cells were disrupted by passage

through a French press and then centrifuged at30,000 x g for 30 min, nearly all of the inhibi-tory activity remained in the cell wall-enrichedfraction (cell pellet) rather than in the superna-

483

on October 16, 2017 by guest

http://iai.asm.org/

Dow

nloaded from

WALSTAD, REITZ, AND SPARLING

tant. When the cell pellet was exposed toneutral solutions (KCl or NaCl) of varying ionicstrength (up to 5 M), or to butanol, no release ofinhibitor was achieved. Partial release of theinhibitor from the cell pellet was achieved with20 mM ethylenediaminetetraacetic acid or bymild alkali treatment (pH 8.0 to 10.0). Progres-sively greater release of inhibitor was achievedwith more strongly alkaline buffers, and at pH12 all cellular material was saponified, withcomplete recovery of inhibitory activity (alkaliinhibitor).Most of the inhibitor remained in the super-

natant after a 60-min 100,000 x g centrifuga-tion, and it was not dialyzable. Like gonococcalwhole-cell suspensions, the alkali inhibitor wasnot affected by autoclaving (121 C) for 20 min orby digestion with deoxyribonuclease, Pronase,or trypsin. The alkali inhibitor was completelyblocked by addition of 2% (wt/vol) bovineserum albumin and partially blocked by addi-tion of 1% soluble starch (Difco) to the mediaused for testing inhibitory activity. In contrast,soluble colicins El, E2, E3, and K and twopyocins were completely inactivated by Pronaseor autoclaving and were not affected by 2%bovine serum albumin or 1.0% starch (Table 1).Similar results were obtained when inhibition

between whole cells was tested in agar mediumcontaining Pronase or bovine serum albumin.The gonococcal inhibitors were unaffected byPronase but inhibited by 2% bovine serum

TABLE 1. Activity of the gonococcal inhibitor andknown bacteriocins under various conditions

Inhibitory titer

GonococcalbConditions Pyo- Coli- Alkali-

ci i"Whole solubi-cells lizedinhibitor

Control (untreated) .. 100 625 8 8Pronase (0.2%, 2 h,37C) .............. 0 0 8 8

Heat (121 C,20 min) ........... 0 0 8 8

BSAc (2.0%) ......... 100 625 0 0Starchc (1.0%) ....... 100 625 4 4

a Soluble pyocin PI or colicin E2 were tested againstappropriate indicator strains.

bGonococcal strain FA5 was tested for inhibitoryactivity against FA5.

c Bovine serum albumin (BSA) or starch was addedto the agar medium used for measuring inhibitorytiters.

albumin, whereas colicins and pyocins werePronase sensitive and bovine serum albuminresistant. Thus, the gonococcal inhibitor re-leased into agar or prepared by alkaline diges-tion of whole cells or cell pellets differed inseveral important respects from classical bac-teriocins and seemed unlikely to be a protein.

Strain typing with alkali inhibitor. Alka-line extraction of approximately 0.5 g (wetweight) of cells from 20 strains into a volume of2.0 ml resulted in inhibitory titers of 1: 2 to 1 :8when tested on GCBDS plates. The same inhib-itors were bactericidal at dilutions of 1: 256 to1:1,012 to cells growing exponentially in pep-tone broth. Differences in sensitivity to inhibi-tion among the 20 strains were no greater thanfourfold.

Lipid nature of gonococcal inhibitor.Chloroform-methanol extraction of washed,whole cells resulted in quantitative recovery ofinhibitory activity (Table 2). Similar extractionof uninoculated agar plates or of N. meningiti-dis NM33 grown under identical conditions didnot result in recovery of any inhibitory material.Therefore, the chloroform-extractable gonococ-cal inhibitor must have been the result of theendogenous metabolism of the strain and not anartifact resulting from presence of inhibitorylipids in the agar medium.Chromatographic separation of the chloro-

form-methanol extracts revealed that the majorlipid components of N. gonorrhoeae FA5 wereFFA and phospholipids, whereas N. meningiti-dis NM33 showed only phospholipids (Fig. 2).Similar results were obtained with each of sevenother strains of N. gonorrhoeae tested in thismanner; two of three additional strains of N.

TABLE 2. Quantitative recovery of gonococcal growthinhibitor from whole cells by lipid extraction

Dilution showing inhibitory

Sample activitya

0 1:2 1:4 1:8 1:16 1:32

N. gonorrhoeae FA5Whole cells + + + + - -

Chloroform-methanolextract of FA5 + + + + - -

N. meningitidis NM33Whole cells + + + + + -

Chloroform-methanolextract of NM33 _ _

a Inhibitory activity tested against indicator strainFA5 by the double-layer method. +, Inhibition; -, noinhibition.

484 INFECT. IMMUNITY

on October 16, 2017 by guest

http://iai.asm.org/

Dow

nloaded from

VOL. 10, 1974 GROWTH INHIBITION BETWEEN N. GONORRHOEAE STRAINS 485

M sprayed with ninhydrin, these two compounds,A as well as PE, gave the characteristic pink color

of primary or secondary amines. Based on theninhydrin reaction and R, values, one of theunknown compounds of FA5 was clearly identi-fied as monoacyl PE, and the other was identi-fied tentatively as glyceryl phosphorylethanola-

B mine.When the lipid components of the two strains

were separated by preparative thin-layer chro-matography and tested for inhibition,, both theFFAs and the phospholipids of FA5 were inhibi-tory, but the phospholipids of NM33 were not.Since the FFAs of FA5 were quite inhibitory, itbecame important to determine their concen-tration and composition. Analysis by gas-liquidchromatography showed that the FFA fraction

D B

c

E

I 2 3 4 5

FIG. 2. Neutral lipids of N. gonorrhoeae FA5 andN. meningitidis NM33. Solvent system is isooctane-ethyl acetate (100:3.4). 1, Mono-, di-, triglyceridemixture; 2, oleic acid; 3, FA5; 4, NM33; 5, cholesterol.Only the two spots for FA5 (at positions D and E),representing free fatty acids and phospholipids, wereinhibitory when eluted from preparative TLC; none ofthe lipid material from NM33 was inhibitory.

meningitidis also displayed plentiful FFA onsilica gel chromatograms. Thus, FFA were uni-formly present in seven strains of N. gonor-rhoeae, but were an inconstant finding amongN. meningitidis strains.When the phospholipids were separated in a

more polar solvent, gonococcal strain FA5 andmeningococcal strain NM33 again appeared todiffer (Fig. 3). Both strains contained PG andPE, but FA5 appeared to have small quantitiesof two other compounds. When another chro-matogram of the separated phospholipids was

.;:

i E

F

2 3 4 5 6 7 8 9 10 11FIG. 3. Phospholipids of N. gonorrhoeae FA5 and

N. meningitidis NM33. Solvent system is chloro-form-10% acetic acid in methanol (100:15). 1, NM33;2, FA5; 3, monoacyl PE (lyso PE); 4, PG; 5, PE; 6, PEand PG, 7, NM33 and PG; 8, FA5 and PG; 9, PC; 10,PC, monoacyl PC (lyso PC) and cardiolipin; 11, PCand monoacyl PE. FA5 and NM33 both reveal PG andPE at positions B and C, respectively. FA5 and NM33both reveal PG and PE at positions B and C,respectively. FA5 shows monacyl PE (position D) andphosphorylethanolamine at the origin (position G). Ad-ditional phospholipid standards of cardiolipin, PC, andmonoacyl PC were located at positions A, E, and F,respectively.

%:. ..: AA.1 -

ih

on October 16, 2017 by guest

http://iai.asm.org/

Dow

nloaded from

WALSTAD, REITZ, AND SPARLING

TABLE 3. Total cellular fatty acids of N. gonorrhoeaeFA5 and N. meningitidis NM33 compared to the FFA

fraction of FA5

Total fatty acids of: FFA fraction of

Fatty NM33 FA5acidacd

AMoll ;IMol/( d wt )

Mol %

g (dry Mol% g (dry Mol% (drywt)wt) wt)

12: lb 1.2 0.9 4.5 3.3 0.2 0.814:0 10.8 7.9 8.6 6.3 1.0 3.815:0 9.8 7.2 0.2 0.2 1.3 5.016:0 58.6 42.8 54.3 40.0 5.2 19.516:1 43.7 31.9 56.0 41.3 9.5 36.217:0 0.5 0.4 TRc17:1 0.5 0.4 TR18:0 0.8 0.6 0.8 0.6 2.0 7.618:1 10.3 7.5 10.7 7.6 6.1 23.319:Ob 0.3 0.2 0.5 0.4 0,9 3.4Total 136.5 135.6 26.2

a The free fatty acid (FFA) fraction of N.gonorrhoeae FA5 represented 19.3% of the total fattyacids. N. meningitidis NM33 had no FFA fraction.

bTentative identification.c TR, Trace.

of FA5 constituted about 20% of the total fattyacids and that the FFA composition differedconsiderably from that of the total fatty acids(Table 3). The major fatty acids present, 16:0,16: 1 and 18: 1, represented 19.5, 36.2, and23.3%, respectively, of the FFA fraction and40.0, 41.3, and 7.6% of the total fatty acids. Theamount and composition of the fatty acids ofthe tested strains of N. gonorrhoeae and N.meningitidis were similar to those reported byothers (22, 31). The chromatography methodsused in our experiments would retain all hy-droxyacids; therefore, the ,B-hydroxy lauric acidreported by Moss et al. (22) would not havebeen observed.The sensitivity of N. gonorrhoeae and N.

meningitidis to purified FFAs of differing struc-ture is shown in Table 4. Representative strainsof both species were more sensitive to medium-and long-chain fatty acids than to short-chainfatty acids. Methyl esters of the major FFAswere not inhibitory, nor was the major phospho-lipid of both strains, PE. However, the monoa-cyl derivative of PE was quite inhibitory and,therefore, likely to be responsible for the inhibi-tion produced by FA5's phospholipid fraction.A representative mixture of the major fatty

acids found in the FFA fraction of N. gonor-rhoeae strain FA5 was prepared with the follow-ing final concentrations: 18:0, 1.3 mM; 18:1,

3.8 mM; 16:0, 3.1 mM; and 16:1, 6.0 mM. Thissolution was inhibitory to FA5 at a dilution of1: 16 and to NM33 at a dilution of 1: 64, but hadno effect on a strain of N. lactamicus which wasresistant to growth inhibition by gonococci.

DISCUSSIONWe sought to determine whether gonococci

produced bacteriocins that could be used as thebasis of a strain-typing system. The resultsshowed that each of 50 strains of N. gonorrhoeaeproduced inhibitory substances effectiveagainst gonococci and meningococci, but notagainst other species of bacteria. However, thedifferences among strains in production of andsensitivity to the inhibitor were too small topermit reliable strain typing. Our results do notsubstantiate, therefore, the report by Flynn andMcEntegart that nearly 75 of 100 strains ofgonococci could be separated into 13 types onthe basis of differences in patterns of inhibitionobserved by the cross-streak method (7). Sincethe strains used in this study included several ofthe indicator strains used by Flynn and McEn-tegart (7), our failure to confirm their findingsis not likely to be solely the result of straindifferences, nor are the differences likely to bedue to conditions of culture since they wereessentially the same (7).The results presented here suggest strongly

that production of true bacteriocins (24) bygonococci is quite uncommon. The inhibitorsproduced by gonococci were not mitomycin Cinducible, and they resisted Pronase digestionand autoclaving. Their activity was blocked bystarch or albumin, two compounds capable ofabsorbing or forming complexes with lipids (13,18). These properties of the gonococcal inhibitorwere unlike those demonstrated for known coli-cins and pyocins (Table 1) and suggested thatthe inhibitor might be lipid or lipopolysaccha-ride (13, 18, 19). This was confirmed by quanti-tative recovery of inhibitor by lipid extractionprocedures and by demonstration that the lipidextracts contained inhibitory long-chain FFAand phospholipids.

In 1947 Ley and Mueller described the pres-ence in agar media of fatty acids inhibitory togonococci and showed that this inhibition wasovercome by addition of starch to the medium(18). They also described differences in strainsusceptibility to inhibition by fatty acids (18).Brooks et al. found that "spent" GCBDS media(on which gonococci had been grown) containedincreased amounts of medium- and long-chainfatty acids (C10 to C18) compared to uninocu-lated media (2). We have extended these obser-

486 INFECT. IMMUNITY

on October 16, 2017 by guest

http://iai.asm.org/

Dow

nloaded from

VOL. 10, 1974 GROWTH INHIBITION BETWEEN N. GONORRHOEAE STRAINS

vations by showing that up to 20% of gonococcalfatty acids are in the free (unesterified) formand that the diffusion of lipids into the mediumaccounts for growth inhibition when gonococciare tested for bacteriocin production by stan-dard methods. Our observations also suggest a

similar mechanism may explain a part of thegrowth inhibition between certain meningo-cocci.

It is well known that anionic detergents,including fatty acids, are toxic to bacteria (13,29). The fact that surfactant activity increaseswith chain length of fatty acids to a maximumat C,4 or C,6 and then decreases above C,6 (13)corresponds with the inhibition pattern we

obtained for both FA5 and NM33 (Table 4).Also, realizing the detergent nature of themonoacyl derivative of PE, but not of PE itself,one can hypothesize that gonococcal self-inhibi-tion is surfactant in nature. However, it is possi-

ble that factors other than physiochemical ac-

tivity (surface tension) may be involved. Freeseet al. attribute the mechanism of lipophilicantimicrobial activity to the inhibition of sub-strate transport into the cell, thereby prevent-ing oxidative phosphorylation (10).We have not shown the mechanism by which

FFA are produced, but the chromatographicevidence (Fig. 3) suggests possible step-wisedegradation of PE to monoacyl PE and glycerylphosphorylethanolamine, with correspondingrelease of one or both FFAs from PE. This inturn, suggests the presence of phospholipases ingonococci. These enzymes have been reportedfor E. coli (11). In other experiments, we haveobserved essentially constant amounts of inhi-bition from cells of FA5 harvested during earlyexponential, late exponential, and stationaryphases of growth in peptone broth. When FA5,FA4, and four other strains of N. gonorrhoeae

TABLE 4. Inhibition of gonococci and meningococcia by purified fatty acids and lipids

Inhibition of N. gonorrhoeae FA51 growth at

Compound Chain length concentrations of:

1.0M 0.1 M 0.01 M 0.001M 0.0001 M

Saturated free fatty acidButyric acid 3 _C - - -

Valericacid 5- - - -Caproic acid 65 - - |Capryllic acid 8 + _ _ _Capric acid 10 + + - _Lauric acid 12 + + + _Myristic acid 14 + + + +Palmitic acid 16 + + + +Stearic acid 18 + + + _

Unsaturated free fatty acidsOleic acid 18:1 + + + +cis-Vaccinic 18:1 + + + +Palmitoleic 16:1 + + + +10-Undecenoic 11:1 + + _ _

Hydroxy free fatty acidRicinoleic 18:1 + + + +

Methyl esters of free fatty acidsMethyl palmitate 16 _ _ _ _Methyl stearate 18 _ _ _ _Methyl oleate 18:1 _ _ _ _

PhospholipidPhosphatidylethanolamine (PE) NTdd _ _Monoacyl PE NT NT + _

a Although NM33 was somewhat more sensitive than FA5 when similarly tested, the inhibition pattern wasalmost identical.

Tested by the double-overlay method.c No inhibition; +, inhibition.d NT, not tested.

487

on October 16, 2017 by guest

http://iai.asm.org/

Dow

nloaded from

488 WALSTAD, REITZ, AND SPARLING

were harvested during exponential growth phasein broth and their lipids were chromatographed,FFAs were readily apparent (data not shown).The Neisseria more closely resemble gram-

positive than typical gram-negative organismsin their sensitivity to medium- and to long-chain fatty acids (29). Since mutations or physi-cal methods that disrupt the integrity of theouter membranes of E. coli and Salmonellatyphimurium render them sensitive to medium-and long-chain fatty acids, the resistance ofthese organisms to C10 to C18 fatty acids hasbeen attributed to the penetration barrier com-posed by their outer envelope structure (10, 29).The susceptibility of Neisseria to antibioticsalso approximates the levels found in mostgram-positive species. These observations sug-gest a unique structure of the outer membranesof the Neisseria, which otherwise appear to betypically gram negative in structure by electronmicroscopy (16) and chemical analysis (20).Routine cultures of gonococci grown on choco-

late agar or GCBDS media rarely survive morethan 48 h of incubation. Although we cannotexclude the presence of other toxic metabolitesor autolytic enzymes (27), our demonstration ofthe production of self-inhibitory lipids duringgrowth may partially explain the short survivalof most gonococcal cultures.

ACKNOWLEDGMENTSWe thank Y. Ikeya and L. Guymon for fruitful suggestions

and M. Maness for the preliminary experiments.This work was supported by a grant from the John A.

Hartford Foundation, Inc., by Public Health Service Re-search Career Development Award A133032, from the Na-tional Institute of Allergy and Infectious Diseases to P. F. S.,and by Public Health Service Research grant AA00204 toR. C. R.

LITERATURE CITED

1. Bligh, E. G., and W. J. Dyer. 1959. A rapid method oftotal lipid extraction and purification. Can. J. Bio-chem. Physiol. 37:911-917.

2. Brooks, J. B., D. S. Kellogg, L. Thacker, and E. M.Turner. 1971. Analysis by gas chromatography of fattyacids found in whole cultural extracts of Neisseriaspecies. Can. J. Microbiol. 17:531-543.

3. Counts, G. W., L. Seeley, and H. N. Beaty. Identificationof an epidemic strain of Neisseria meningitidis bybacteriocin typing. J. Infect. Dis. 124:26-31.

4. Davis, B. D., and E. S. Mingioli. 1950. Mutants ofEscherichia coli requiring methionine or vitamin B,2. J.Bacteriol. 60:17-28.

5. Diem, K. (ed.). 1962. Buffer solutions, p. 314-315. InDocumenta Giegy. Geigy Pharmaceuticals/GeigyChemical Corporation, New York.

6. Farmer, J. J., and L. G. Herman. 1969. Epidemiologicalfingerprinting of Pseudomonas aeruginosa by the pro-duction of and sensitivity to pyocin and bacteriophage.Appl. Microbiol. 18:760-765.

7. Flynn, J., and M. G. McEntegart. 1972. Bacteriocinsfrom Neisseria gonorrhoaea and their possible role inepidemiological studies. J. Clin. Pathol. 25:60-61.

8. Fredericq, P. 1957. Colicins. Annu. Rev. Microbiol.

INFECT. IMMUNITY

11:7-22.9. Fredericq, P., J. Thibault, and A. Gratia. 1946. Re-

cherches sur les proprietes antibiotiques des germes dugenre Bacillus. C. R. Soc. Biol. (Paris) 140:1035:-1037.

10. Freese, E., C. W. Sheu, and E. Galliers. 1973. Function oflipophilic acids as antimicrobial food additives. Nature(London) 241:321-325.

11. Fung, C. K., and P. Proulx. 1969. Metabolism of phos-phoglycerides in E. coli. III. The presence of phospholi-pase A. Can. J. Biochem. 47:371-373.

12. Garbus, J., H. F. DeLuca, M. E. Loomans, and F. M.Strong. 1963. The rapid incorporation of phosphateinto mitochondrial lipids. J. Biol. Chem. 238:59-63.

13. Glassman, H. N. 1948. Surface active agents and theirapplication in bacteriology. Bacteriol. Rev. 12:105-148.

14. Hutchinson, R. I. 1970. Typing the gonococcus. Brit.Med. J. 3:107.

15. Kellogg, D. S., Jr., W. L. Peacock, Jr., W. E. Deacon, L.Brown, and C. I. Pirkle. 1963. Neisseria gonorrhoeae. I.Virulence genetically linked to clonal variation. J.Bacteriol. 85:1274-1279.

16. Kellogg, D. S., Jr., E. M. Turner, C. Callaway, L. Lee,and J. E. Martin Jr. 1971. Particulate fractions ofNeisseria gonorrhoeae. Infect. Immunity 3:624-632.

17. Kingsbury, D. T. 1966. Bacteriocin production by strainsof Neisseria meningitidis. J. Bacteriol. 91:1696-1699.

18. Ley, H. L., Jr., and J. H. Mueller. 1946. On the isolationfrom agar of an inhibitor for Neisseria gonorrhoeae. J.Bacteriol. 52:453-460.

19. Maeland, J. A. 1966. Antibodies in human sera againstantigens in gonococci demonstrated by a passive hae-molysis test. Acta Pathol. Microbiol. Scand. 67:102-110.

20. Maeland, J. A. 1971. Immunochemical investigations onNeisseria gonorrhoeae endotoxin. 2. Serological multi-specificity and other properties of phenol-water prepa-rations. Acta Pathol. Microbiol. Scand. 79:233-238.

21. Metcalf, L. C., A. A. Schmitz, and J. R. Pelka. 1966.Rapid preparation of fatty acid esters for gas chromato-graphic analysis. Anal. Chem. 38:514-515.

22. Moss, C. W., D. S. Kellogg, Jr.. D. C. Farshy, M. A.Lambert, and J. D. Thayer. 1970. Cellular fatty acids ofpathogenic Neisseria. J. Bacteriol. 104:63-68.

23. Penefskv, H. S., and A. Tzagoloff. 1971. Extraction ofwater-soluble enzymes and proteins from membranes,p. 204-219. In S. P. Colowick and N. 0. Kaplan (ed.),Methods in enzymology, vol. 22. Academic Press Inc.,New York.

24. Reeves, P. 1965. The bacteriocins. Bacteriol. Rev.29:24-45.

25. Reitz, R. C. 1972. Phosphatidyl choline species inEuglena gracilis. Biochem. Biophys. Acta 260:654-665.

26. Robertson, A. F., and W. E. M. Lands. 1962. Positionalspecificities in phospholipid hydrolyses. Biochemistry1:804-810.

27. Rogers, H. J. 1970. Bacterial growth and the cell enve-lope. Bacteriol. Rev. 34:194-214.

28. Schnaitnam, C. A. 1971. Effect of ethylenediaminetetra-acetic acid, Triton X-100, and lysozyme on the mor-phology and chemical composition of isolated cellwalls of Escherichia coli. J. Bacteriol. 108:553-563.

29. Sheu, C. W., and E. Freese. 1973. Lipopolysaccharidelayer protection of gram-negative bacteria against in-hibition by long-chain fatty acids. J. Bacteriol. 115:869-875.

30. White, I. A., and D. S. Kellogg, Jr. 1965. Neisseriagonorrhoeae identif-ication in direct smears by afluorescent antibody-counterstain method. Appl. Mi-crobiol. 13:171-174.

31. Yamakawa, T., and N. Ueta. 1964. Gas chromatographicstudies of microbial components. I. Carbohydrate andfatty acid constitution of Neisseria. Jap. J. Exp. Med.34:361-374.

on October 16, 2017 by guest

http://iai.asm.org/

Dow

nloaded from