JOURNAL OF BACTERIOLOGY, JUlY 1976, p. 237-248Copyright © 1976 American Society for Microbiology

Vol. 127, No. 1Printed in U.SA.

Evidence Against the Involvement of Chemotaxis inSwarming ofProteus mirabilis

FRED D. WILLIAMS,* DAVID M. ANDERSON,' PAUL S. HOFFMAN,2 ROBERT H.SCHWARZHOFF, AND SHARLEEN LEONARD3

Department ofBacteriology, Iowa State University, Ames, Iowa 50011

Received for publication 19 April 1976

Nonswarming and nonchemotactic mutants ofProteus mirabilis were isolatedafter mutagenesis with N-methyl-N'-nitro-N-nitrosoguanidine or ultravioletlight. These mutants were used in experiments to determine if chemotaxis isinvolved in the swarming of P. mirabilis. Nonchemotactic mutants failed toform chemotactic bands in a semisolid casein hydrolysate medium, yet theyswarmed on the same medium containing 1.5% agar. Nonswarming mutantswere attracted towards individual amino acids and components of tryptose. Incross-feeding experiments, no evidence was obtained to indicate the productionof a diffusable chemical repellent. In studies with the wild-type P. mirabilis, noclear-cut negative chemotaxis was seen even though three different assays wereused and numerous chemicals were tested. Additional evidence against theinvolvement of chemotaxis in swarming comes from finding that dialysis doesnot interfere with swarming; swarm cells will swarm immediately when trans-ferred to fresh media, and swarm cells will swarm on an agar-water mediumsupplemented with a surfactant. These data indicate that chemotaxis is notinvolved in the swarming ofP. mirabilis.

Two members of the genus Proteus, Proteusmirabilis and Proteus vulgaris, will readilyswarm on common laboratory media. This phe-nomenon results in the bacterium completelycovering the surface of a solid medium, usuallywithin 24 h. Swarming of Proteus is alwaysaccompanied by morphological changes in thecells. Nonswarming cells are about 2 to 4 um inlength, whereas swarm cells reach lengths ofup to 80 ,tm and possess many more flagella perunit of surface area than do the short forms (9).Groups of swarm cells actually move over theagar surface and produce the swarming phe-nomenon. Frequently, periodic swarming oc-curs in that swarming proceeds for a period oftime and then stops. The cessation of swarmingis accompanied by the swarm cells undergoingdivision into short cells, a process called consol-idation. The resulting short cells undergo aperiod of growth, and then new swarm cells areformed at the periphery of the growth zone andswarming starts again. This produces a zona-tion effect, with concentric rings of swarminggiving a "bulls eye" appearance on the agarsurface.

' Present address: 998 Church St., Ventura, Calif. 93001.2 Present address: Department of Biology, Virginia Po-

lytechnic Institute and State University, Blacksburg, Va.24060.

3Present address: R. R. #2, Holstein, Iowa 31025.

Since the original description ofswarming byHauser (8), investigators have attempted to ex-plain the swarming phenomenon. The mostwidely accepted theory is the negative chemo-taxis hypothesis of Lominski and Lendrum(12). Although based upon weak experimentalevidence, this hypothesis accounts for the mor-phological changes in the cells and the periodic-ity frequently associated with swarming.Briefly, they propose that during growth ofPro-teus on a solid surface the cells excrete one ormore toxic waste products that inhibit cell divi-sion, thus leading to the formation of swarmcells. The toxic agent diffuses through the agarmedium, out from the region of growth, estab-lishing a gradient of the toxic substance. Theswarn cells detect the gradient and move overthe agar surface down the gradient towards alower concentration of the toxic agent (negativechemotaxis). When the cells have moved to asubtoxic region, the inhibition of division isrelieved and the swarm cells stop swarmingand start dividing. The growth of short formsresults once again in the production of the toxicsubstance, swarm cells are again produced, andswarming starts again.The Lominski and Lendrum hypothesis can

be easily modified into one that proposes posi-tive chemotaxis for swarming. This theorywould resemble the depletion of nutrients hy-

237

on March 20, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

238 WILLIAMS ET AL.

pothesis of Moltke (13). Growth of a centralcolony depletes the available nutrients in theimmediate area, resulting in the formation of agradient of nutrients. Swarm cells are formeddue to the depletion of nutrients, and, onceformed, they move up the nutrient gradient(positive chemotaxis) to richer regions.Because of the attractiveness of the Lominski

and Lendrum hypothesis and the recent ad-vances in our understanding of bacterial chem-otaxis (3), we felt that we were in a position totest whether chemotaxis is involved in swarm-ing by using several different experimental ap-proaches. Through the use of nonchemotacticand nonswarming mutants we felt that anyrequirement for chemotaxis could be detected,and if chemotaxis is required for swarming achemical gradient of either an attractant orrepellent could be implicated. This report de-scribes the results of studies leading to theconclusion that chemotaxis is not involved inthe swarming of Proteus.

MATERIALS AND METHODSOrganisms. The wild-type P. mirabilis, strain

IM47, used in these studies was obtained from theculture collection of the Bacteriology Department,Iowa State University. All of the mutants of P.mirabilis produced in this study were derived fromIM47. Cultures of Proteus morganii, Pseudomonasaeruginosa, Serratia marcescens, and Escherichiacoli K-12 were obtained from the same culture col-lection. Additional nonswarming mutants ofP. mir-abilis were obtained from W. M. Bain (University ofMaine, Orono) and C. D. Jeffries (Wayne State Uni-versity, Detroit, Mich.). All cultures were main-tained on slants of either nutrient agar or Trypti-case soy agar and stored at 4 C.

Culture media. Nutrient broth with NaCl (NB)consisted of nutrient broth (GIBCO) supplementedwith 0.5% NaCl. Tryptose broth contained 2.0%tryptose (Difco) and 0.5% NaCl. Trypticase soy brothwith yeast extract (TSYE) consisted of 0.5% yeastextract added to Trypticase soy broth (BBL). Casi-tone phosphate broth (CP) was composed of 1% Casi-tone (Difco); 10-2 M potassium phosphate, pH 7.0;10-3 M MgSO4 7H20; 0.001% nicotinic acid; and0.5% NaCl. Casein hydrolysate (CH) medium wasprepared with a basal salts solution modified fromthat of Jones and Park (10). It contained, per liter:K2HPO4, 11.2 g; KH2PO4, 4.8 g; (NH4)2S04, 0.54 g;NH4Cl, 0.54 g; Fe(NH4)2(SO4)2*6 H20, 20 mg;MgSO4 * 7 H20, 40 mg; nicotinic acid, 2.4 mg; NaCl,4.8 g; vitamin-free Casamino Acids (Difco), 3.0 g.The MgSO4, nicotinic acid, NaCl, and CasaminoAcids were autoclaved as separate solutions andadded aseptically. The Fe(NH4)2(SO4)2 was preparedin 0.02 N HCI, membrane filtered (0.45-,um poresize, Millipore Corp.), and added aseptically.

Semisolid media contained 0.2% agar (Difco) andsolid media contained 1.5% agar (Difco). Petri platescontaining a solid medium were always dried over-

night at 35 C before use to remove droplets from theagar surface.The chemotaxis medium described by Adler (2)

was used in the quantitative chemotaxis assay forwashing the cells and preparing the attractant solu-tions. It contained 10-2 M potassium phosphatebuffer (pH 7.0) and 10-4 M ethylenediaminetetraace-tic acid. The ethylenediaminetetraacetic acid wasadded as the free acid and was neutralized withpotassium hydroxide. The medium was preparedwith deionized water, filter sterilized, and stored atroom temperature.

Isolation of nonswarming mutants. Cells of P.mirabilis (IM47) from 10 ml of a 14-h, agitated cul-ture in TSYE were collected by centrifugation (900x g for 10 min) at room temperature. The pellet wassuspended in 10 ml of a 0.85% sodium citrate solu-tion, containing 500 Ag of N-methyl-N'-nitro-N-ni-trosoguanidine per ml (Aldrich Chemical Co.), ad-justed to pH 5.0 with HCI. This suspension wasincubated at 35 C for 1 h and was then centrifuged,and the cells were suspended in cold NB. This proce-dure normally resulted in a 99 to 99.9% reduction inviable cell numbers. While the suspension was re-frigerated, viable cell counts were determined onthe mutagenized suspension by using spread platesof nutrient agar. Nutrient agar without NaCl failsto support swarming, so that discrete colonies areproduced. On the basis of the viable cell count, thesuspension was diluted to a concentration of 500 to750 viable cells/ml.Two methods were used for the selection of non-

swarming mutants from the diluted suspension. Thefirst method involved plating 0.1 ml of the suspen-sion on TSYE or nutrient agar (GIBCO) with 0.5%NaCl (NA); both media support normal swarming ofIM47. The plates were incubated at 35 C until theindividual colonies began to show evidence ofswarming. Nonswarming colonies were picked andreinoculated onto the same medium to allow confir-mation of their inability to swarm. The second selec-tion method involved a type of "blind selection" inwhich 0.1 ml of the diluted suspension was plated onnutrient agar without NaCl and incubated for 16 hat 35 C. Normal swarming does not occur underthese conditions. Individual colonies were thenpicked and reinoculated onto individual sectors ofquadrant plates (Quad-Petri, Lab-Tec Products)containing TSYE. These plates were then incubatedat 35 C for 16 h and examined for the presence ofnonswarmers.

Isolation of nonchemotactic mutants. A 200-mlNB culture of P. mirabilis IM47 was grown to mid-log phase at 35 C with shaking and then was har-vested by centrifugation at 10,000 x g. The cellswere resuspended in cold 0.5% sterile NaCl and keptat 0 C before use. A 25-ml sample was dispensed intothe bottom of half of a glass petri plate (100 mm) andexposed to a 15-W germicidal lamp (ChampionG15T8) for 20 to 55 s with a dose rate of 12 ergs/mm2.During ultraviolet irradiation the culture was con-stantly agitated on a lateral shaking machine. Sam-ples were removed at 10-s intervals for the determi-nation of viable cell numbers and inoculation ofenrichment plates. Care was taken to avoid photo-

J. BACTERIOL.

on March 20, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

CHEMOTAXIS AND SWARMING 239

reactivation. To enrich for nonmotile and nonchem-otactic mutants, each irradiated sample was inocu-lated onto the center of five CH soft-agar plates,which were incubated at 35 C until the outermostchemotactic band of cells reached the edge of theplate. At that time a transfer was made from thecenter of the plate to a fresh plate of the samemedium. This procedure was continued for up to 16serial transfers. To isolate mutants, 0.1 ml was re-moved from the center of a soft-agar plate, diluted,and pour plated in CH soft agar. After incubation for14 to 16 h at 35 C, each plate was examined for thepresence of nonchemotactic and nonmotile colonies.These were detected as colonies without chemotacticbands of cells or colonies with sharp distinct edges inthe soft agar. Those that were found were pickedand transferred to individual plates of soft agar forthe verification of defects in chemotaxis or motility.Plates with between 10 and 20 colonies proved opti-mal for detecting mutants.

Confirmed mutants were cloned and then exam-ined for their ability to give typical reactions of P.mirabilis in a number of biochemical tests. Eachmutant was further characterized for motility,swarming, and chemotaxis. Motility tests weremade in NB and soft-agar media at 25 and 35 C bythe direct microscopic observation of wet mounts orhanging-drop preparations. Swarming was tested onNA, tryptose agar, TSYE agar, and CH agar at 25and 35 C. After 24 h each plate was examined forswarming, and if no swarming was observed thecentral colony was examined microscopically for thepresence of motility and swarm cells at the colonyedge. Qualitative tests for chemotaxis towardsamino acids were made by looking for the formationof chemotactic bands in tryptose soft agar or CH softagar at 25 and 35 C.

Several different techniques were used to test fornegative chemotaxis. The method of Doetsch andSeymour (7) was used to test the wild-type P. mira-bilis IM47 and control cultures of S. marcescens, P.aeruginosa, and P. morganii. In this test a suspen-sion ofbacteria fills the inside of a flat capillary tubeplugged with agar at each end. One agar plug con-tains the test agent; the other plug serves as acontrol. The capillary is observed microscopically forthe presence of a dense band of cells moving awayfrom the test agent. All cultures were grown at roomtemperature (20 to 22 C) in either CP broth or Trypti-case soy broth (BBL) and washed before use. Eachculture was examined for motility before use, andthe suspension was used only if at least 60% of thecells displayed normal motility. The following com-pounds were tested in the chemotaxis assay: hydro-chloric acid, pH 1.0 to 5.0; FeCl3, 10-1 and 10-2 M;FeSO4, 10-1 and 10-2 M; ZnC12, 10-2 M; 2-methylbu-tylamine, 10-2 M; isoamylamine, 10-2 M; isobutyla-mine, 10-2 M; NaOH, 1 N; indole, 10-2 M; succinicacid, 10-' M; sodium acetate, 10-1 M; and hydrogenperoxide, 20%.Two methods of Tso and Adler (15) were also used

to test for negative chemotaxis. In the first method(agent in the plug), bacteria were suspended in soft-agar chemotaxis medium in a petri plate. Hard-agarplugs containing the test compound were then

placed into the soft agar in the petri plate. Chemi-cals tested by this method included: sodium acetate,indole, L-leucine, old overswarmed CP agar, andFeCl1. The concentration range tested for all com-pounds was between 10-6 and 1.0 M. During theincubation, samples were removed from areas nearthe test plugs and away from the test plugs for use inmicroscopic determinations of relative bacterial mo-tility. A dense band of motile cells migrating awayfrom the test plug constituted negative chemotaxis.

Another method referred to as the "cells in thewell assay" (15) was also used. Wells (8 mm) werecut in the solidified test agar plates, and these wellswere then filled with a solution of bacteria sus-pended in 0.3% soft chemotaxis agar. The plateswere allowed to stand at room temperature for 30min before being observed. Chemotactic cells in thisexperiment would be those that formed a ring in thecenter of the well.The quantitative chemotaxis assay described by

Adler (2) was used to characterize the chemotacticresponse towards amino acids of IM47 and thosenonswarming mutants derived from it that weremotile. This method quantifies the number of bacte-ria that swim into a capillary tube filled with anattractant. Cells from a hard-agar surface were sus-pended in chemotaxis medium and then harvestedby centrifugation at 900 x g for 10 min at roomtemperature. The pellet was gently resuspended inthe small amount of liquid that remained after de-canting, and 5 ml of chemotaxis medium was added.This process was repeated a second time to wash thecells before use. The final cell suspension was ad-justed to a density of 5 x 10' viable cells/ml and thenexamined microscopically by using a hanging-droptechnique to judge the degree of motility in thesuspension. Cells harvested in this manner wouldnormally retain their motility throughout the washprocedure and for over 2 h after the last wash.Chemotactic assays were conducted at 30C for 50min.

Cross-feeding. Three types of cross-feeding exper-iments were performed to test for the presence of adiffusable substance that would promote swarmingof a nonswarming mutant or bring about prematureswarming of a wild-type swarmer.The first method involved mutant-mutant crosses

in which the ability of one mutant to stimulateswarming of a second mutant was examined. Platesof TSYE agar were inoculated with the two culturesto be tested as two divergent streaks that ap-proached each other at one side of the plate but didnot touch. The plates were incubated at 35 C andexamined over a period of 48 h for evidence ofswarming.The second method involved crosses between

wild-type swarmers and nonswarming mutants, inwhich the ability of a wild type to stimulate swarm-ing of a mutant was examined. Plates ofTSYE agarwere prepared with a strip of membrane filter, 0.8cm wide and 8 cm long (cut from 142-mm filter disks;0.45-Am pore size, Millipore Corp.), embedded in theagar to divide each plate in half. One side of theplate was inoculated with a wild-type strain, andthe plate was incubated at 35 C for 3 h. The other

VOL. 127, 1976

on March 20, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

240 WILLIAMS ET AL.

side of the plate was then inoculated with a mutant,and the plate was reincubated at 35 C and examinedover a period of 48 h for evidence of swarming by themutant. In both of these methods, mutants thatshowed any sign of swarming were checked for re-version by inoculating cells from the swarm bandonto fresh TSYE agar.The third method involved crosses between mu-

tants and wild types in which the ability of themutant to stimulate premature swarming of thewild types was examined. At the beginning of theexperiment, three sectors of individual quadrantplates containing NA were inoculated with a differ-ent mutant as a single streak. The plates were incu-bated at 35 C for 2 h and were then inoculated with asingle wild type on each sector. In the three sectorswith the mutants, the wild type was inoculated asthe second leg of a "V" so that the mutant and wildtype were in direct contact at the vertex. In thefourth quadrant, the wild type was simply inocu-lated as a single streak, and this served as a control.The plates were reincubated at 35 C and examinedmicroscopically to determine exactly when the wildtype began to swarm. Mutants that showed anyability to induce premature swarming of the wildtype, as compared with the control, were examinedin a duplicate experiment.

Isolation of swarm cells. Swarm cells were iso-lated comparatively free of short forms by washingthe cells from the outermost swarm band from eitherCP agar or tryptose agar using the wash medium.The cells were washed twice before use, and micro-scopic examination showed 80 to 90% swarm cells.

Dialysis experiments. A plastic chamber (2.5 cmin diameter and 5 mm in depth), open at the top andhaving two ports entering the chamber at oppositesides, was used for dialysis. The chambers weremade from tissue culture chambers (Research PlusLaboratories Inc.) in which the central pillar hadbeen flattened out, leaving a slightly rounded bot-tom in the well of the chamber. CP agar was asepti-cally pipetted onto an 8- by 5-cm piece of steriledialysis tubing and allowed to dry as a thin film atroom temperature overnight in a sterile petri dish.The CP agar-covered dialysis tubing was laid on topof the dialysis chamber so that it came into contactwith the dialyzing medium. A uniform flow of di-alysis medium into and out of the chamber wasregulated by a variable-speed polystaltic pump(Beckman). The flow rate was between 3 and 4 ml/min. P. mirabilis IM47 was grown overnight in CPbroth and washed, and an inoculum was placed ontothe center of the agar-covered dialysis tubing thathad already been placed onto the dialysis chamberto equilibrate. CP agar-covered dialysis tubing wasalso placed onto CP agar plates and inoculated toserve as a control. Dialysis was allowed to continuefor 30 h at room temperature. If no swarming oc-curred on the dialyzed agar, the agar-covered di-alysis tubing was removed from the dialysis cham-ber and placed onto a plate of CP agar and observed.

RESULTSNonswarming mutants. A total of 23 non-

swarming mutants were obtained. In addition,

J. BACTERIOL.

10 other nonswarming mutants ofP. mirabiliswere obtained from other investigators (thesewere derived from other wild-type strains). Themutants isolated in this laboratory were as-signed the designation Nsw and a 3-digit num-ber. Those mutants obtained from other labora-tories have the strain designation used in thatlaboratory.When the mutants were examined under dif-

ferent culture conditions, a number of themdemonstrated the ability to swarm. Table 1summarizes the responses of the 33 mutants onthree different media at two temperatures. Theability to form swarm cells and the motility ofthe cells under nonswarming conditions arealso given. Of the Nsw mutants, those with anumber between 100 and 200 were selected onnutrient agar with salt, and 5 of the 12 demon-strated the ability to swarm on a richer me-dium. Those mutants with a strain designationbetween 200 and 300 were selected on TSYEagar. Five of the Nsw mutants (Nsw2O2, -203,-206, -211, and -214) were obtained by using theblind selection technique by which they wereselected on the basis of their response in anisolated environment. No information wasavailable on the selection methods of the mu-tants from other laboratories.

Cross-feeding experiments were performedbetween nonswarming mutants whereby indi-vidual mutants were examined for their abilityto stimulate swarming of other mutants. Ini-tially, crosses were made between potential"producers" (organisms that form swarm cells)and "detectors" (organisms that do not formswarm cells), but eventually nearly every mu-tant was crossed with every other mutant. Noevidence for a diffusable agent that would stim-ulate a nonswarming mutant to swarm wasobtained. Several cases of swarming by the mu-tants were observed, but, in nearly all cases,cells removed from the swarm band retainedthe ability to swarm when placed on fresh me-dium. The orientation of these apparent rever-sions seemed random, and the movement of theswarm bands did not appear to be affected bythe other mutant. Occasional stimulation ofnutritional and temperature-sensitive mutantswas observed, but these results were not repro-ducible.Using the second type of cross-feeding

method involving the plates with membranefilters, 26 of the nonswarming mutants werecrossed with 23 different wild-type swarmers.The membrane filter embedded in the agar pre-vented the wild type from swarming into themutant colonies for 24 h. A significant numberof reversions were noted, and the number ap-peared to increase as the mutant cultures were

on March 20, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

CHEMOTAXIS AND SWARMING 241

TABLE 1. Characteristics of nonswarming mutants

Ability to swarm

Mutants TSYE TAb NA Swarm cell Motility.formatio

25C 35C 25C 35C 25C 35C

Nsw104, Nsw107, - - - - - - - -Nsw2O2, Nsw2O6,Nsw211, Nsw215,Nsw221, Nsw235,75

Nsw115, Nsw231 - - - - - - + -Nsw109, Nsw214, - - - - - - - +Nsw228

Nsw2O3, Nsw122, - - - - - - + +2NS, 4NS, 60,91b, DA, Ml

Nsw1lO + - + - + - + +Nsw227, NswlOl + - + - + - - +Nsw112, Nsw114, + + + - - - + +Nsw106, Nsw108

77NS, 73NS + + + + - - + +Nsw127 + - + - - - + +91NR + - - - - + +

a Under nonswarming conditions.b TA, Tryptose agar.

serially transferred in broth through the courseof the experiments. When fresh cultures wereprepared from the stock slants, however, thenumber of reversions decreased significantly.No evidence was obtained for the presence of adiffusable agent, produced by a wild type, thatwould stimulate swarming of a mutant.

In the third type of cross-feeding experiment,7 different wild-type swarmers were crossedwith 32 nonswarming mutants. Again, no evi-dence was obtained for the presence ofa diffusa-ble agent, produced by a mutant, that wouldstimulate premature swarming of a wild type.The incubation period, required before swarm-ing of the wild type occurred, ranged between3.5 and 6 h for the different strains, but itremained essentially constant for a single wildtype crossed with the various mutants. Cases ofslight stimulation were re-examined using thesame procedure, but no reproducible stimula-tion could be detected.Chemotaxis of nonswarming mutants in

soft agar. The positive chemotactic responses ofthe wild-type swarmers and the nonswarmingmutants were determined by using tryptosesoft agar; Adler (1) used a similar medium todemonstrate the response of E. coli. He foundthat when a soft-agar plate was centrally inocu-lated, well-defined bands of cells emerged fromthe colony edge and moved out through the softagar as concentric rings. The cells in each ringutilized a particular nutrient in the mediumand responded chemotactically in the gradientthat was established. The cells in each subse-

quent band utilized a different component ofthe medium.The chemotactic response of the different

swarming strains ofP. mirabilis were variable,as indicated by the pattern of bands thatformed. The chemotactic rings normally be-came evident after 2 to 3 h of incubation at 35 Cand were well formed over most of the plateafter 8 to 10 h.When the nonswarming mutants were exam-

ined in this way and compared with their re-spective wild type, three different responseswere noted. A large number were nonmotileand, therefore, failed to respond chemotacti-cally. Mutants of this type demonstrated a verydistinct colony edge with no sign of spreading.Mutants of the second type were motile anddemonstrated a spreading colony with diffuseedges, but no chemotactic bands formed. Micro-scopic examination of cells from this type ofcolony showed that they were uncoordinated,with few cells showing any translational move-ment. Mutants of the third type were motileand demonstrated chemotactic bands, but inthe cases where the wild-type parents wereavailable for comparison the patterns of bandswere different. Table 2 summarizes the re-sponses of the absolute nonswarming mutantstested.The quantitative chemotactic responses of

IM47, two mutants that were chemotactic insoft agar (NswlO9 and Nsw228), and the twomutants that were motile but nonchemotacticin soft agar (Nsw2O3 and Nsw 214) were deter-

VOL. 127, 1976

on March 20, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

242 WILLIAMS ET AL.

TABLE 2. Chemotactic responses of nonswarmingmutants on tryptose soft agar

Mutant Motilitya Chemotaxisb

NswlO4NswlO7Nswll5Nsw2O2Nsw2O6Nsw2llNsw215Nsw221Nsw231Nsw23575Nsw2O3Nsw214NswlO9Nsw122Nsw2282NS4NS60DAMl

+

+

+

a As indicated by movement of the colony edgeand microscopic examination.bAs indicated by the formation of bands of cells

moving away from the central colony.c Positive chemotaxis but pattern of bands was

altered when compared to the wild type.d Wild types were not available for comparison.

mined on cells removed from hard-agar me-

dium that would support swarming of the wildtype. Figure 1 illustrates the responses of IM47,NswlO9, and Nsw228 towards 14 amino acids.The response of IM47 to six other amino acids(isoleucine, leucine, proline, threonine, tyro-sine, and valine) was determined to be negligi-ble, and the mutants were, therefore, not testedtowards those amino acids. The two mutantsthat failed to respond chemotactically in softagar also failed to respond in the quantitativeassay, so the values for these two organisms are

omitted from the figure.Chemotaxis of swarm cells. When swarm

cells were harvested by the procedure describedabove, a suspension judged to be at least 90%swarm cells was obtained. Motility of theswarm cells was sluggish, but a high proportionof them showed translational movement. Thissuspension was assayed using the quantitativemethod towards alanine, asparagine, gluta-mate, histidine, methionine, phenylalanine,serine, and a 2% solution of tryptose. No signifi-cant response was obtained for alanine, aspara-

gine, or glutamate, and the net response for theremaining five attractants ranged between2,800 for serine and 7,200 for histidine. Whereasmicroscopic examination of the suspension indi-

cated that the swarm cells were motile, thesecells failed to undergo the rapid directionchanges demonstrated by the short forms.Nonchemotactic mutants. Numerous non-

chemotactic mutants were isolated by using theenrichment procedure. Because it is very likelythat all the mutants isolated from a particularenrichment series are identical, we havegrouped the mutants to indicate that they wereobtained from identical enrichment plates (seeTable 3). The 17 groups of mutants isolatedwere examined for their positive chemotacticabilities on CH soft agar at two different tem-peratures. Table 3 lists the results of this studyas well as motility determinations performedduring the progress of the experiment; nonmo-tile mutants, temperature-sensitive mutants,and positive chemotaxis mutants were isolatedwith the mutagenesis procedure used.The 17 different groups ofmutants were char-

acterized as to their swarming ability on fourdifferent media at two different temperatures.Table 4 summarizes the results of this experi-ment and also lists information on motility andswarm cell formation. The majority of thosemutants capable of motility showed an abilityto swarm at both temperatures. To test for thepossibility that swarming was caused by revert-ants, cells were isolated from a swarm bandand rechecked for chemotaxis in CH soft agar.All isolates showed impaired chemotaxis iden-tical to the mutant phenotype.

Motility studies of mutants. In overnightNB cultures (12-18 h), those mutants that weredeficient in positive chemotaxis demonstrated adistinctive type of motility. Whereas the wild-type P. mirabilis IM47 showed a pattern oftumbles and short runs similar to that de-scribed by Berg and Brown (5), positive chemo-taxis mutants manifested a greater tendency totumble. An extreme example of this trait wasshown by mutant Nch9 after incubation atroom temperature. When viewed microscopi-cally, the majority of the motile cells from thisculture tumbled almost unceasingly and be-haved rather like tiny "pinwheels."Those mutants that were earlier found to be

nonmotile in CH soft agar (Nchl8 and Nch39)were likewise nonmotile when cultured in NB.Negative chemotaxis. With the capillary

tube method of Doetsch and Seymour no nega-tive chemotactic responses were observed for P.mirabilis, whereas all the control organismsexhibited negative chemotactic responses tomost chemical agents tested (Table 5). The con-trol organisms responded to the chemical gra-dients by forming dense bands of motile bacte-ria, which migrated along with the diffusingchemicals down the capillary tubes away from

J . BACTERIOL.

on March 20, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

CHEMOTAXIS AND SWARMING

IL In.I

I

m

ASP ASN CYS GLU SLN

GLY HIS LYS MET PHE SER TYR

FIG. 1. Quantitative chemotactic responses of IM47 (open bars), NswI09 (closed bars), and Nsw228(crosshatched bars) to seven amino acids. The cells were harvested from the surface of a tryptose hard-agarplate. All values represent the averages of three determinations. Cysteine was used at a concentration of10-2M; the remaining amino acids were used at a concentration of10-1 M.

the source. Bacteria within a band were highlymotile and appeared to be trapped within theband. Rapid directional changes were exhibitedby the bacteria within the band, whereas bacte-ria outside the band, on the side away from thediffusing chemical, traveled longer distancesbetween directional changes. Only nonmotilebacteria were observed in the region betweenthe test agar plug and the band of cells. Thosebacteria that moved out of the band into themore concentrated region of the gradient (to-wards the test agent) were observed to quicklychange directions, eventually returning to theband. Likewise, bacteria that moved out of theband into the less concentrated region of thegradient (towards the control end) also rapidlychanged directions and returned to the band.The result of the activities ofmany motile bac-teria appeared to confine the bacteria within a

narrow region of the gradient, and this confine-ment of motile bacteria was visualized and in-

terpreted as a distinct band of motile chemotac-tic bacteria. The numbers of motile bacteriawithin the band increased because of the en-trance ofnew bacteria as the band moved downthe capillary tube.P. mirabilis IM47 never formed a band in

response to any of the chemicals tested. As atoxic chemical diffused down the capillary tube,more and more of the bacteria became immobi-lized. Under microscopic observation (x400),motile cells of P. mirabilis were seen enteringthe toxic region and becoming immobilized be-fore any avoidance response was initiated. Aline of demarcation was seen moving down thecapillary tube. On the proximal side (testagent), nonmotile bacteria were seen, whereason the other side (control) normal motility wasobserved. Motile cells were not observed to en-

ter the lethal region of the diffusing gradientand reverse direction, as was seen with thecontrol organisms. Swarm cells ofP. mirabilis

6 F

5

4

3

ALA ARO

ALA ARS

DUD

6

5

4

3

2

l m&A - JLML I -J-LILb -iA 1, I243VOL. 127, 1976

ItI0

x

U)-i-iw0

0w

9-jm2MC)L)-9

10

6x

cn-j-jwu

awI.-4-jn2DC-)u4

on March 20, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

244 WILLIAMS ET AL.

IM47 also failed to demonstrate a chemotacticresponse to the compounds tested. These cellsalso swam into the lethal region of the gradientand quickly became immobilized. Swarm cells

TABLE 3. Characteristics ofpositive chemotaxismutants on CH soft agar

Chemotaxis MotilityMutant'

24 C 35 C 24 C 35 C

Nchl8- - - -

Nch39 - - - -Nch2-7 + - + +Nch22-26 + - + +Nch9-11 - - + +Nchl2 - - + +Nchl5-17 - - + +Nchl9-21 - - + +Nch27-28 - - + +Nch29-32 - - + +Nch33-34, 36-37 - - + +Nch38 - - + +Nch4O-42 - - + +Nchl3 - +b + +Nch8 +b +b + +Nchl4 +b +b + +Nch35 +b +b + +

aMutants listed on the same horizontal line wereisolated from the same enrichment plate.bShowed positive chemotactic bands of smaller

diameter than those produced by the wild type(IM47).

did not tumble or initiate any rapid directionalchanges in response to the diffusing gradient.When the chemical in -the plug technique of

Tso and Adler (15) was used to assay negativechemotaxis with P. mirabilis, some apparentnegative chemotactic responses were observed.Zones of clearing surrounded by dense bands ofbacteria were observed around agar plugs con-

taining sodium acetate, indole, and FeCli. Thethreshold concentration for each was: sodiumacetate, 101- M; indole, 10-2 M; and FeCli, 10-2M. A 1-mm or greater zone of clearing, mea-

sured from the edge of the agar plug to theinner edge of the band ofbacteria encircling thezone of clearing, was considered a positive test.Upon microscopic examination of the zones ofclearing around agar plugs of sodium acetateand indole, motile bacteria were found,whereas no motile bacteria were observed inthe clear zones around FeCl:, agar plugs, whichabolished motility. It was impossible to deter-mine whether or not band formation wascaused by the chemotactic response of motilebacteria, because both motile and nonmotilebacteria were found in the band. The number ofmotile bacteria within the band was not signifi-cantly greater than the number of motile bacte-ria observed within the zones of clearingaround the test agar plugs. The presence of asmany motile bacteria in the clear zones as inthe band suggested that many of the motile

TABLE 4. Swarming responses ofpositive chemotaxis mutants

Ability to swarm

Mutants" TSYE TAc CHd NA Swarm cell Motilitybformationr

24C 35C 24C 35C 24C 35C 24C 35C

Nchl8 - - - - - - - - -

Nch39 - - - - - - - - -

Nch2-7 + - + - + - + - + +Nch8 + + + + + + + + + +Nch9-11 + + + + + + + + + +Nchl3 + + + + + + + + + +Nchl4 + + + + + + + + + +Nchl5-17 + + + + + + + + + +Nchl9-21 + + + + + + + + + +Nch22-26 + + + + + + + + + +Nch29-32 + + + + + + + + + +Nch35 + + + + + + + + + +Nch38 + + + + + + + + + +Nch4O-42 + + + + + + + + + +Nchl2 + + + - + - + - + +Nch27-28 + + + + + + + - + +Nch33-34, + + + + + + + - + +

36-37a Mutants listed on the same horizontal line were isolated from the same enrichment plate.b Determined microscopically with samples taken from the edge of colonial growth.c TA, Tryptose agar.d CH, Casamino Acids agar.

J. BACTERIOL.

on March 20, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

CHEMOTAXIS AND SWARMING 245

TABLE 5. Results of the Doetsch and Seymour assay for negative chemotaxis

Organism Test agent eliciting a negative chemotactic responseaP. aeruginosa ........... HCl, FeCl3, FeSO4, ZnCl2, MBA, IAA, NaOH, H202 (not tested: IBA,

succinic acid, acetate)P. morganii ........... HCl, FeC13, FeSO4, MBA, IAA, IBA, NaOH, Ind, succinic acid, H202

(not tested: acetate)S. marcescens ........... HCI, FeCl3, FeSO4, MBA, IAA, NaOH, H202 (not tested: IBA, AMC,

Ind, succinic acid, acetate)P. mirabilis IM47 ........... None (all agents tested)

a Abbreviations: MBA, 2-Methylbutylamine; IAA, isoamylamine; IBS, isobutylamine; Ind, indole; AMC,acetylmethylcarbinol.

bacteria were nonchemotactic in the assay. OldCP agar, on which swarming had been allowedto completely cover the agar surface, wastested, and no bands or zones of clearing wereformed in response to this agar medium. Anonmotile mutant (Nsw204) was tested in theassay system as a control. Zones of clearingwere observed around the 1 M sodium acetateagar plugs and around the 10-' M FeCl3 agarplugs, whereas no zones were observed around10-1 M indole agar plugs or around old, over-swarmed agar plugs. The diffusion of sodiumacetate caused a 1-mm (radius) zone of clearingto appear around the agar plug, whereas FeClicaused a 3- to 4-mm zone to appear. The re-sponses of motile P. mirabilis IM47 to sodiumacetate and FeCl3 at the same concentrationwere 2 mm and 4 to 5 mm, respectively.No negative chemotactic responses were ob-

served using the cells in the well assay system.If a response had occurred, a ring of bacteriawould have formed in the center of a well inresponse to a gradient of the diffusing com-pound. No rings were observed with indole,sodium acetate, or L-leucine. OverswarmedCP agar and FeCl3 were not tested.

Dialysis experiments. P. mirabilis was har-vested from CP broth (overnight growth) andeither inoculated directly or washed first andthen inoculated onto the dialysis chamber andthe control CP agar plates. Swarm cells wereseen at 7 h on the dialysis chamber and plates,and swarming commenced at 8 h and continuedfor 2.5 h. The distance covered by the firstswarm band for both the dialyzed colony andthe control was 10 mm (width measured fromedge of primary colony to edge of first swarmband). During the dialysis experiments the pHof the medium was monitorea with pH paperand found not to change, in contrast to nondi-alyzed control plates with which the pH rose to8.5 to 9.0.Experiments with swarm cells. Swarm cells

were collected from the second or third swarmband ofP. mirabilis IM47 grown on CP agar (18h at 30 C), by either washing the swarm cells off

the surface of the agar with wash medium andthen washing the swarm cells by the standardwash procedure or by scraping the swarm cellsdirectly off the agar with an inoculating loop.When the collected swarm cells (either proce-dure) were inoculated onto fresh CP agarplates, immediate swarming was observed. Themigration of rafts of swarm cells and the mo-tion of the swarming was similar to the swarm-ing originating from a primary colony. The dis-tance swarmed (width of swarm band) ap-peared to be related to the number of swarmcells inoculated onto the agar; however, noquantitative studies were made. Also, the dis-tance swarmed (width of swarm band) varieddepending on when the swarm cells were har-vested. Swarm cells harvested after 1 h ofswarming tended to swarm further when trans-ferred to fresh medium than did swarm cellsharvested near the end of the swarming period.The maximum distance swarmed for trans-ferred IM47 swarm cells was a width of 15 mm.Swarm cells harvested near the end of theswarm period swarmed an average of 5 mm.Swarm cells taken from other sources, such asthe defined agar medium, swarmed within 2min, but not immediately, when inoculatedonto CP agar. Immediate swarming was ob-served only between similar media, for exam-ple, CP agar to CP agar or TSYE to TSYE agar,but between different media swarming was de-layed about 2 min.When CP agar-grown swarm cells were inoc-

ulated onto a CP agar film on the dialysis tub-ing and dialyzed, immediate swarming was ob-served on both the dialyzed agar and the nondi-alyzed control plates.When washed swarm cells of IM47 were

placed onto a medium consisting of 1.5% agar inwater, the cells were motile but did not swarm.When the agar-water medium was supple-mented with 200 to 500 ,ug of Triton X-100,Tween 80, or Tergitol per ml, swarming oc-curred out to a radius of 3 to 5 mm. The surfac-tants appeared to be toxic to the swarm cells,because the cells stopped swarming sooner

VOL. 127, 1976

on March 20, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

246 WILLIAMS ET AL.

when exposed to higher concentrations and mo-tility was lost when swarming ceased.

DISCUSSION

These studies were originated to examine ex-perimentally the negative chemotaxis hypothe-sis proposed by Lominski and Lendrum (12) toexplain the swarming ofProteus. The isolationof a number of nonswarming mutants seemedto present a direct approach for testing theproduction of the waste metabolite and for thedetection of the hypothetical gradient. Ourcross-feeding experiments attempted to detectthose mutants defective in the production ofthewaste product or defective in the detection ofthe waste products. We assumed that somenonswarming mutants would be defective be-cause they no longer produced the negativechemotactic agent; we presumed, however, thatthey still had the ability to respond to the agentif it were produced by another organism. Simi-larly, some mutants were assumed to be defec-tive in the detection of the waste product, butthey might still produce the chemical repellent.One might also assume that most nonmotilemutants still produced the repellent and wouldrepresent an excellent source of the negativechemotactic agent. The experimental results ofthe cross-feeding studies with nonswarmingmutants strongly contradict the presence of areadily diffu4sible metabolite that stimulatesswarming by nonswarming mutants or thatpromotes premature swarming by the wildtype. If the production of this metabolite alsoresults in a toxic environment that leads to theformation of swarm cells, some of these mu-tants certainly should be producers becausethey still form swarm cells, although swarmingdoes not follow. It therefore seems that thepossibility that all of the mutants were nonpro-ducers is very unlikely. First, the wild-typeculture would certainily be a producer and wefound no evidence that the wild type wouldpromote swarming by nonswarmers. Secondly,the blind selection technique used to isolate fiveof the nonswarming mutants was employed toavoid the criticism that all of the mutants werenonproducers, because the original selectiontechnique had these nonproducing mutantsbeing fed by neighboring colonies, and theyappeared as swarming colonies and weremissed. Although we only isolated five mutantsusing the blind selection approach, these mu-tants were not phenotypically different fromthe mutants isolated using the conventionalselection technique. Thus, we feel that thecross-feeding studies strongly argue against theLominski and Lendrum hypothesis by finding

J. BACTERIOL.

no evidence for a readily diffusible metabolitethat can be detected by the organism. Accord-ing to the negative chemotaxis hypothesis, it isthe swarm cells that move down the gradient ofrepellent, and perhaps the short cells do notrespond to the hypothetical metabolite. Severalof the nonswarming mutants and the wild typeform swarm cells on agar. If only these types ofcells respond to a repellent, we should havedetected swarming by those mutants that formswarm cells and premature swarming by thewild type in the cross-feeding experiments.This was not observed.The phenotypes of the nonswarming mutants

suggest, at the very least, that the theory ofnegative chemotaxis must be modified. Thelarge number of mutants that are both nonmo-tile and non-swarm cell producers (Table 1)indicates that the formation of elongated cellsis not simply the result of a toxic product. If itwas, one would expect that a mutant with de-fective motility would still form swarm cellsunless it also had a block in the production ofthe toxic agent. The relatively large number ofmutants of this type obtained casts doubt uponthe possibility that they represent double muta-tions.Further evidence against the negative chem-

otaxis hypothesis comes from the negativechemotaxis assays using the wild-type P. mi-rabilis IM47. In our studies three different as-say systems were used. With the Doetsch andSeymour system we used control organisms (P.aeruginosa, S. marsescens, and P. morganii),and these organisms responded to a number ofagents as reported by others (7), but we ob-served no response by P. mirabilis IM47 to anytest agent. Close microscopic examination ofboth the short forms and the swarm cells re-vealed that the cells became immobilized beforea negative chemotactic response could be elic-ited. A negative chemotactic response would becharacterized by a series of tumblings followedby net movement of bacteria away from thetoxic region (14, 15). The short forms tumblebriefly when they enter the lethal region of thegradient, whereas the swarm cells do not. Nei-ther move out of the lethal region, as was ob-served for the control organisms. According tothe negative chemotaxis theory for swarming,it is the swarm cells that recognize the subtledifferences in a gradient and, since this studyindicates that these cells are insensitive to po-tentially lethal compounds, it is unlikely thatswarm cells swarm because ofa recognition ofagradient of toxic metabolites.*In the Tso and Adler chemotaxis assay sys-tem, P. mirabilis appears to respond in a nega-tive chemotactic manner. As described by Tso

on March 20, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

CHEMOTAXIS AND SWARMING 247

and Adler using E. coli (15), zones of clearingappeared around test agar plugs and a denseband of cells encircled the zone. The thresholdresponses for P. mirabilis are 10 to 100 timeshigher than those reported for E. coli, and alsothe zones of clearing around the agar disks aresmaller for P. mirabilis. Tso and Adler testednonchemotactic motile mutants and found thatthe diffusion artifact described by Clayton (6)could not account for the negative chemotacticresults from their chemotactic strains ofE. coli.With the higher concentration of compounds(10 to lOOx) required for eliciting chemotacticresponses for P. mirabilis, the diffusion artifactmust be considered. The study with nonmotilemutants of P. mirabilis indicates that a zone ofclearing, which would ordinarily be labeled anegative chemotactic response, is a diffusionartifact for many of the compounds tested. Thisis an important point for P. mirabilis, since thenumbers of motile bacteria composing the bandare as numerous as the motile bacteria foundwithin the zone of clearing. If negative chemo-taxis exists for P. mirabilis, the response is 10to 100 times less than the negative chemotacticresponses reported for E. coli (15) and Salmo-nella (14), and this weak response is certainlynot supportive of the Lominski and Lendrumnegative chemotaxis hypothesis. We were una-ble to obtain sufficient numbers of swarm cellsto test them in the Tso and Adler assay sys-tems.With a minor change, the Lominski and Len-

drum negative chemotaxis hypothesis can bechanged into a positive chemotaxis hypothesis.The presence ofa waste product gradient can bereplaced by a gradient of nutrients produced bythe main colony during growth. This hypothe-sis is similar to that proposed by Moltke (13).The results of the studies with nonchemotac-

tic mutants and the quantitative chemotaxisstudies with the nonswarming mutants contra-dict a theory involving positive chemotaxis. Ofthe nine groups of motile, nonchemotactic mu-tants, six showed swarming on all media testedat either of two temperatures. Of the threegroups of mutants that did not show normalswarming, they only showed this abnormalityon NA at 35 C. Mutant Nchl2 was somewhatmore hampered in that at 35 C it failed toswarm on any of the media tested except TSYEagar. Even with the exceptions, it can be statedthat the majority of motile, chemotaxis mu-tants swarmed, indicating that positive chemo-taxis is not required for swarming. The non-chemotactic mutants isolated were probablythe generally nonchemotactic type described byArmstrong et al. for E. coli (4), because theywere selected on a complex medium. Although

they were not tested for their chemotactic re-sponse towards other classes of chemicals, theydid lose the ability to form chemotactic bandson the semisolid CH medium (a medium thatsupports swarming when prepared with 1.5%agar). In all probability these mutants werealso defective in negative chemotaxis as well;however, we were unable to demonstrate aclear-cut negative chemotactic response withthe wild-type culture and consequently wereunable to test these mutants for negative chem-otaxis. The E. coli mutants described by Tsoand Adler (15) that lacked all positive chemo-taxis had also lost the negative chemotacticresponse; this might be the case with our mu-tants. If this assumption is correct, the non-chemotactic mutants provide strong evidencethat neither positive nor negative chemotaxis isrequired for swarming.The quantitative examination of chemotaxis

towards amino acids was examined in two non-swarming mutants that showed modified chem-otaxis in tryptose soft agar. The results indi-cated that, although NswlO9 does show im-paired chemotaxis, Nsw228 shows near normalchemotaxis; both mutants show the ability tomove towards attractants, yet neither swarms.When these results are combined with thosefrom the study of nonchemotactic mutants, itwould seem that chemotaxis is not involved inswarming. Furthermore, swarm cells of thewild-type culture failed to show a significantchemotactic response towards amino acids, andsince the swarm cell suspension also containedshort cells the weak response that was observedcould be due to those short cells in the suspen-sion.A number ofexperiments attempted to elimi-

nate the formation of a chemical gradient (at-tractant or repellent). When P. mirabilis IM47was growing on the surface of a solid mediumthat was being dialyzed from beneath, the orga-nism still swarmed. The results ofpH measure-ments indicated that the dialysis was efficientand that the formation of a gradient of dialyza-ble substances was unlikely. Even better evi-dence against the requirement of a gradient forswarming comes from the observation thatswarm cells removed from one agar surface willswarm almost immediately when placed onto afresh agar surface. A similar observation wasmade by Kvittingen in 1949 (11). We have sub-sequently found that no nutrients need be pres-ent in the second medium; swarm cellsswarmed on agar-water when any of a numberof detergents was present. Should either nega-tive chemotaxis or positive chemotaxis be in-volved in swarming, the establishment of agradient of the attractant or repellent would be

VOL. 127, 1976

on March 20, 2021 by guest

http://jb.asm.org/

Dow

nloaded from

248 WILLIAMS ET AL.

essential for any chemotactic response. It is dif-ficult to explain the immediate swarming ofswarm cells when transferred to fresh mediumif a gradient is required. Time is just too short.That they will swarm on the agar-water-deter-gent medium rules out the formation of anattractant gradient and a positive chemotacticresponse. Furthermore, a negative chemotacticresponse is unlikely because washed swarmcells move almost immediately after beingplaced onto fresh medium.The results presented here essentially rule

out the involvement of chemotaxis in theswarming of P. mirabilis. It is now clear thatswarming can be subdivided into at least threestages: the formation of swarm cells; the move-ment of swarm cells across the agar surface;and the division of swarm cells back to theshort forms (consolidation). Our current workis directed at determining the conditions thatlead to swarm cell production.

ACKNOWLEDGMENTSThe material reported in this paper includes work per-

formed for individual M.S. degrees by D.M.A., P.S.H., andR.H.S. at Iowa State University.

This work was partially supported by a small grant fromthe Office of the Vice President for Research, Iowa StateUniversity. S.L. was a National Science Foundation Under-graduate Research Participant.

LITERATURE CITED1. Adler, J. 1966. Chemotaxis in bacteria. Science 153:708-

716.

2. Adler, J. 1973. A method for measuring chemotaxis anduse of the method to determine optimum conditionsfor chemotaxis by Escherichia coli. J. Gen. Microbiol.74:77-91.

3. Adler, J. 1975. Chemotaxis in bacteria. Annu. Rev.Biochem. 44:341-356.

4. Armstrong, J. B., J. Adler, and M. M. Dahl. 1967.Nonchemotactic mutants of Escherichia coli. J. Bac-teriol. 93:390-398.

5. Berg, H., and D. A. Brown. 1972. Chemotaxis in Esche-richia coli analysed by three-dimentional tracking.Nature (London) 239:500-504.

6. Clayton, R. K. 1958. On the interplay of environmentalfactors affecting taxis and motility in Rhodospirillumrubrum. Arch. Mikrobiol. 29:189-212.

7. Doetach, Rt. N., and F. W. K. Seymour. 1970. Negativechemotaxis in bacteria. Life Sci. 9:1029-1037.

8. Hauser, G. 1885. Uber Faulnissbacterien und derenBeziemungen zur Septicamie. F. G. W. Vogel,Leipzig.

9. Hoeniger, J. F. M. 1965. Development of flagella byProteus mirabilis. J. Gen. Microbiol. 40:29-42.

10. Jones, H. E., and R. W. A. Park. 1967. The influence ofmedium composition on the growth and swarming ofProteus. J. Gen. Microbiol. 47:369-378.

11. Kvittingen, J. 1949. Studies of the life cycle of Proteushauser. Part 1. Acta Pathol. Microbiol. Scand. 26:24-50.

12. Lominski, I., and A. C. Lendrum. 1947. The mecha-nism of swarming of Proteus. J. Pathol. Bacteriol.59:688-691.

13. Moltke, 0. 1927. Contributions to the characterizationand systematic classification ofBac. proteus vulgaris(Hauser). Levin and Munksgaard, Copenhagen.

14. Tsang, N., R. M. Macnab, and D. E. Koshland, Jr.1973. Common mechanism for repellents and at-tractants in bacterial chemotaxis. Science 181:60-63.

15. Tso, W., and J. Adler. 1974. Negative chemotaxis inEscherichia coli. J. Bacteriol. 118:560-576.

J. BACTERIOL.

on March 20, 2021 by guest

http://jb.asm.org/

Dow

nloaded from