JOURNAL OF BACTERIOLOGY, Nov. 1977, p. 628-640Copyright © 1977 American Society for Microbiology

Vol. 132, No. 2Printed in U.S.A.

Chemotaxis by Bdellovibrio bacteriovorus Toward PreySUSAN C. STRALEY AND S. F. CONTI*

Thomas Hunt Morgan School of Biological Sciences, University ofKentucky, Lexington, Kentucky 40506

Received for publication 1 August 1977

A chemotaxis assay system that uses a modified Boyden chamber wascharacterized and used for measurements of chemotaxis by Bdellovibrio bacter-iovorus strain UKi2 toward several bacterial species. Bacteria tested includedboth susceptible and nonsusceptible cells (Escherichia coli, Pseudomonasfluorescens, Bacillus megaterium, and B. bacteriovorus strains UKi2 and D).None was attractive to bdellovibrios when present at densities below 107 cellsper ml. Chemotaxis toward E. coli was studied most extensively; underconditions that minimized effects of osmotic shock to the cells, E. coli andexudates from E. coli at densities as high as 10 cells per ml failed to elicit achemotactic response. Cell-free filtrates from mixed cultures of bdellovibriosand E. coli neither attracted nor repelled bdellovibrios. The data indicate thatbdellovibrios do not use chemotaxis to locate prey cells.

Bdellovibrio bacteriouorus is a predatorybacterium that grows at the expense of othergram-negative bacteria (30, 32). To survive, abdellovibrio must locate and successfully pene-trate a prey cell before losing its viability dueto rapid starvation (20). Hespell et al. (20)have estimated from their data on viability ofthree strains of B. bacteriovorus under starva-tion conditions that about 1.5 x 105 prey cellsper ml would have to be present for a bdello-vibrio to have a 50% chance of survival if itencountered its prey solely by random collision.This would limit the distribution of bdello-vibrios in nature to environments containingrelatively dense populations of prey. The re-quired density of prey would be lower if suitablesources of carbon and energy were present torelieve the starvation of the bdellovibrios or ifthe predator could use a "search" mechanismsuch as positive chemotaxis (motion toward achemical) for the location of prey cells (20).

Speculation that bdellovibrios might locatetheir prey by means of chemotaxis has ap-peared in the literature (e.g., references 30, 32)without any reported attempt to make directexperimental tests. We present direct tests ofthe chemotactic attraction of B. bacteriovorusto a variety of bacteria, including both suscep-tible and nonsusceptible prey cells. The experi-ments show that B. bacteriouorus strain UKi2probably does not locate individual prey cellsby means of chemotaxis.

MATERIALS AND METHODS

Bacteria. Strains of B. bacteriovorus used werethe facultatively predacious strain UKi2 (9); obli-

gately predacious strains D and 114 obtained, re-spectively, from S. C. Rittenberg, University ofCalifornia at Los Angeles, and D. Abram, PurdueUniversity; and a facultatively predacious strain,UKi2Smr, a streptomycin-resistant derivative ofstrain UKi2, isolated in our laboratory by selectionwith streptomycin.

E. coli B, obtained from D. Abram, Purdue Uni-versity, is a derivative of the ATCC strain 15144. E.coli BSmr, a streptomycin-resistant derivative of E.coli B, was isolated in our laboratory by J. J. Tudor.Bacillus megaterium was obtained from A. D. Hitch-ins, University of Kentucky; it is a derivative ofATCC strain 19213. Pseudomonas fluorescens wasobtained from ATCC strain 13525.

Media. Prey cells used for propagation of bdello-vibrios were suspended in dilute nutrient broth(DNB, organic components from Difco) supple-mented with 0.002 M CaCl2 and 0.003 M MgCl2 (28).E. coli and B. bacteriovorus strain UKi2 grownnonpredaciously were propagated in peptone-yeastextract medium (PYE, components from Difco) (2).For some experiments, E. coli was grown in aglycerol-salts medium (P medium of Kaiser andHogness + 0.5% [wt/vol] glycerol [1, 21]). B. mega-terium and P. fluorescens were grown on Trypticasesoy broth (TSB; Becton, Dickinson, and Co.). Forone experiment, B. megaterium was grown on theglucose-salts medium of Tuominen and Bernlohr(36).Suspending medium (SM) in which cells were

washed and suspended for chemotaxis experiments,was glass-distilled water containing 10- 3M tricine[N-tris(hydroxymethyl)methylglycine NaOH] (pH7.35); 0.002 M CaCl2; and 0.003 M MgCl,. For someexperiments, glycerol or glucose was added from afilter-sterilized stock solution to SM, giving thesame concentration as had been present in glycerol-or glucose-salts medium (0.5% or 0.03% [wt/vol],respectively).

628

on March 25, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

CHEMOTAXIS TOWARD PREY BY BDELLOVIBRIO 629

Chemicals and membrane filters. All chemicalswere obtained from commercial sources and were

reagent grade; tricine was from the Sigma ChemicalCo.

Unless otherwise specified, all membrane (micro-pore) filters were obtained from the Millipore Corp.

Cultivation. Incubation of all cultures was at30°C; liquid cultures were grown with shaking (gy-ratory shaker, model G-25 for flasks and Psychroth-erm incubator for tubes; New Brunswick ScientificCo.).

E. coli B, B. megaterium, and P. fluorescens were

maintained on PYE agar (1.0% [wt/vol] agar, Difco)in plates and were transferred monthly. Stock cul-tures of E. coli used to grow B. bacteriovorus con-

tained about 5 x 109 cells per ml and were preparedby inoculating 250 ml of PYE in a 500-ml Erlen-meyer flask with cells from a plate and incubatingthe culture for 18 to 24 h. The cells were centrifuged,washed once with 250 ml of DNB, and suspended in250 ml of DNB. Such stock cultures were stored at4°C and used within 2 weeks.

E. coli BSmr was used for propagation of B.bacteriovorus strain UKi2Sm' and was handled as

was E. coli B, except that the growth mediumcontained 125 Ag of streptomycin base per ml, andthe density of the final stock culture (in DNB +streptomycin) was about 2 x 109 cells per ml.

For use as a source of attractants in chemotaxisexperiments, E. coli was grown in PYE or in glyc-erol-salts medium from a 1% (vol/vol) inoculum of astationary-phase culture into 250 ml of medium in a

500-ml nephelometer (or Erlenmeyer) flask. Cul-tures were grown in PYE to early exponential phase(13 Klett units [Klett-Summerson colorimeter,green filter no. 54] representing 9 x 107 cells per

ml, with an average of 1.2 cells per chain), lateexponential phase (64 Klett units; 6 x 108 cells per

ml and an average of 1.5 cells per chain), andstationary phase (340 Klett units; 5 x 109 cells per

ml and an average of 1.0 cell per chain). Culturesgrown in glycerol-salts medium were incubated for3.5 h.

For use in chemotaxis experiments, B. megate-rium or P. fluorescens cells from a plate were addedto 10 ml of TSB in a culture tube and grown for 12h. Usually, one or two additional 12-h transfers(10% [vol/vol]; 10-ml cultures) were made; for thefinal transfer, a 1% (vol/vol) inoculum from a sta-tionary-phase (12-h) culture was made into 250 mlof TSB in a 500-ml nephelometer flask. P. fluores-cens was grown for 12 h; the B. megaterium cellswere grown to early exponential phase (9 Klettunits; 107 cells per ml; average of 4 cells per chain)or stationary phase (290 Klett units; 6 x 108 cellsper ml; average of 2 cells per chain). The bacilli didnot sporulate under these conditions. For one exper-

iment, B. megaterium, which had been growing on

glucose-salts agar (1.0% [wt/vol] agar) was smearedheavily onto glucose-salts agar and grown for 12 h.Cells were inoculated into 250 ml of glucose-saltsmedium lacking MnCl2 in a 500-ml Erlenmeyerflask and grown for 72 min. About 12% of the cellsin the resulting culture contained spores or hadinitiated visible spore formation.

Strains of B. bacteriovorus were stored as lyophi-lized stocks from which cultures were startedmonthly. In intervals between experiments (1 to 2days), Bdellouibrio lysates were stored at 4°C. Pre-daciously grown B. bacteriovorus cultures weregiven at least two sequential 12-h transfers on E.coli by using a 10% (vol/vol) inoculum from a lysatein a 5- or 10-ml total volume in a culture tube. Forall bdellovibrios except strain D, a multiplicity ofinfection of 0.2 was used for the final transfer; thelysate to be used as inoculum was first passedthrough a membrane filter (1.2-,m-diameter pores)to remove E. coli cells and aggregated bdellovibrios(38).

For nonpredacious growth of B. bacteriovorusstrain UKi2, a 12-h, predaciously grown culturewas passed through a membrane filter (1.2-,mpores), diluted 1:9 in PYE in a culture tube, andgrown for 14 to 17 h. The cells were washed once bycentrifugation and suspended in DNB.Enumeration of cells. Densities of suspensions of

B. bacteriovorus were determined with a Petroff-Hausser counting chamber (9). For E. coli, B. meg-aterium, and P. fluorescens, a Coulter Counter wasused (model ZB equipped with a 19-,m-diameterorifice). Cell densities determined with the CoulterCounter were less than actual cell densities, sincethe instrument counted pairs or chains of cells assingle particles. This difference was appreciable forB. megaterium; therefore, in chemotaxis experi-ments involving this organism, both Coulter countsand Petroff-Hausser counts were made.

Preparation of chemotactically responsive bdel-lovibrios. A 10-ml portion of lysate from preda-ciously grown bdellovibrios was passed through amembrane filter (1.2-gm pores), washed with 10 mlof SM, and removed by backwashing with 2.5 ml ofSM. After enumeration, the suspension was dilutedto contain 108 cells per ml. Motility of the cells waschecked by microscopic observation.

Preparation of attractants for chemotaxis exper-iments. When E. coli B or P. fluorescens was usedas a source of attractants, cells were collected ontoa membrane filter (0.45-,um pores) and washed twicewith SM (or SM containing glycerol). The cells wereremoved from the filter by Vortex mixing (30 s) in 3ml of SM. When B. megaterium was used as asource of attractants, cells were washed twice bycentrifugation (at 5 to 10°C) in SM. Filtration wasnot used to collect B. megaterium cells due to rapidplugging of the filter.When B. bacteriouorus strain D was used as a

source of attractants, 10 ml of lysate was passedthrough a membrane filter (1.2-,um pores), and thefilter was washed with 20 ml of SM. Without thiswash, most of the needle-like cells were retained bythe filter (the purpose of which was to pass bdello-vibrios while retaining E. coli cells and aggregatesof bdellovibrios); however, the wash also flushedthrough some E. coli cells (as many as 108 cells perml) which were not subsequently removed from thesuspension. Cells in the filtrate were collected ontoa membrane filter (0.22-gm pores), washed with 10ml of SM, and harvested by backwashing with 3 mlof SM. Smaller bdellovibrios, which might pass

VOL. 132, 1977

on March 25, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

630 STRALEY AND CONTI

through a filter having 0.45-gum pores (used inchemotaxis chambers; see below) were removed bycollecting the 3 ml of harvested bdellovibrios ontosuch a filter. The larger cells remaining on thefilter were collected by backwashing with 2 ml ofSM. The ratio of bdellovibrios to E. coli cells in thisfinal suspension was 25:1.

Nonpredaciously grown B. bacteriovorus strainUKi2 used as a source of attractants was harvestedby passing liquid cultures through a membranefilter (1.2-gm pores) and washing twice with SM.The elongated vibrios and spiral forms retained bythe filter were collected by Vortex mixing in SM.

Filtered mixtures of E. coli and B. bacterioLorusstrain UKi2, having progressively advanced devel-opment of the predator-prey interaction, were pre-pared as follows. Washed cells of E. coli, which hadbeen grown to exponential phase in glycerol-saltsmedium, were mixed with washed, predaciouslygrown cells of B. bacteriovoruts strain UKi2 so as tomake cultures containing 101 E. coli cells per mland having MOIs ranging from 0.0 to 5.0. Thesewere incubated for 12.5 h and passed through amembrane filter (0.20-/im pores; Nalge) to removeintact cells. The resulting filtrates, subsequentlyplated out to verify the absence of bdellovibrios,were used as sources of attractants in a chemotaxisexperiment.

"Exudates" from E. coli were prepared from ex-ponential-phase cells. The cells were washed twiceby centrifugation at room temperature and sus-pended in 3 ml of SM (or SM containing glycerol).This final suspension, containing about 2 x 10''cells per ml, was kept at room temperature for 25min and then passed through a membrane filter(0.22-,um pores). The resulting filtrate ("exudate")was serially diluted in SM for use as a source ofattractants in chemotaxis experiments. In addition,a 2-ml portion of SM was similarly filtered anddiluted for use in a control test. The dilution wasthe same as that used in preparing the exudaterepresenting 109 E. coli cells per ml.

Yeast extract solution was prepared as describedpreviously (33).

Protocol of chemotaxis experiments. Chemotaxisto intact cells was measured using a modified Boy-den chamber similar in design to that of Denningand Davies (8) and machined from Delrin (Fig. 1).In the chamber, two identical compartments wereseparated by a membrane filter (0.45-gm pores[Millipore] or 0.4-,um pores [Nuclepore]). B. bacter-iouorus was small enough to pass through the mem-brane, but the various attractant bacterial cellswere not. Prey cells were injected into the "testcompartment" of the assembled chamber, bdellovib-rios were injected into the "Bdelloibrio compart-ment," and the chamber was incubated at 30°C for30 min. During this time, diffusion created a concen-tration gradient of the attractant, to which thebdellovibrios could respond by swimming up thegradient into the test compartment. After incuba-tion, the contents of the test compartment wereremoved and serially diluted in DNB; bdellovibriosthat had entered the test compartment were countedby plating samples (double-layer technique [91 with

Bde//ovIbr,o M,cropore TestCornportment Membrnoe Comportment

VAod,fied E3oyoer -nomabe,4,tte, TT v Denn,ng 8 P R -.oyies

FIG. 1. Modified Boyden chamber used for meas-uring chemotaxis by B. bacteriouorus toward itsprey. Two identical compartments are separated bya micropore membrane through which the bdellouib-rios, but not the prey, can pass. Prey cells wereinjected into the test compartment of the assembledchamber; bdellovibrios were injected into the Bdel-louibrio compartment, and the chamber was incu-bated for 30 min at 30°C. The number of bdellouib-rios that entered the test compartment in responseto attractants released by the prey cells was mea-sured.

DNB agar; E. coli B or E. coli BSmr for lawnformation; two plates per serial dilution).

Chemotaxis experiments were usually beguneither 25 or 90 min after the harvesting of theattractant bacteria. During the 25-min "preincuba-tion," the attractant cells were allowed to stand(without shaking) in SM at room temperature (23 to27°C); during the 90-min preincubations, they wereheld at 30°C for as long as possible (ca. 72 min) andat room temperature for the rest of the time. Allbdellovibrios, SM, and yeast extract to be used inan experiment were similarly preincubated. Imme-diately before injecting any sample into the testcompartment, it was aerated in the syringe bydrawing and then expelling air through it.

Every experiment included two types of controltests: chambers that contained 0.1% (wt/vol) yeastextract and ones with only SM in the test compart-ment. The former served as a "positive control,"indicating the chemotactic responsiveness of thecells on a given day; the latter provided a measureof the number of bdellovibrios entering the testcompartment due to random swimming.The terms used to describe the results of a chem-

otaxis experiment are response, background, con-centration-response curve, peak concentration, peakresponse, and threshold, as defined by Mesibov andAdler (26).

Handling of membrane filters and Boyden cham-bers. Filters were soaked in three changes of dis-tilled water, autoclaved in distilled water, anddrained free of excess water before use. Boydenchambers were washed in ethanol, rinsed in tapwater, and then extensively rinsed in distilled wa-ter.

In tests for leakage of chemotaxis chambers, thechambers contained membrane filters having 0.22-gm pores (Millipore) or 0.2- am pores (Nuclepore).Bdellovibrios could not pass through these filtersby random swimming, and any bdellovibrios de-tected in the test compartment (which containedSM) would be due to leakage around the filter. Agasket was required to prevent leakage with thethin Nuclepore filters; for this purpose, a circle of

J. BACTERIOL.

on March 25, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

CHEMOTAXIS TOWARD PREY BY BDELLOVIBRIO 631

nylon tulle, well washed in distilled water, wasplaced next to the membrane filter.

Experimental conditions. The temperature atwhich the chambers were incubated (30°C) and thepH of the suspending medium (7.35) lie within theranges of these conditions that are optimal forchemotaxis by B. bacteriovorus (24); they also aresimilar to the temperature (30°C) and pH (7.2) atwhich the cells were grown. The salts contained inSM are not required for chemotaxis by B. bacterio-vorus, but were included, since the role of thesesalts in the predator-prey interaction is not com-pletely understood at this time.

Preliminary tests with the Boyden chambers useda strong attractant, yeast extract (33). Comparisonsof chemotactic and background responses of B. bac-teriovorus strain UKi2 were made with chamberscontaining membrane filters (Millipore Corp.) hav-ing pores 0.45 and 0.8 Am in diameter. The magni-tudes of the chemotactic responses were similar forthe two filters, but the background response for the0.8-,um pore size was significantly higher than thatfor the 0.45-Am-diameter pores. The smaller poresize was chosen for all other experiments usingmembrane filters (Millipore Corp.), since it provideda more sensitive measure of chemotaxis and wouldbetter retain small prey cells in the test compart-ment.Nuclepore and Millipore membrane filters gave

comparable results, but the Nuclepore filters provedmore difficult to handle because of problems withleakage.

Figure 2 shows the chemotactic and backgroundresponses by B. bacteriovorus strain UKi2 as afunction of the time of incubation of the chambersbefore removal of the contents of the test compart-ments. The accumulation of cells in the test com-partment was rapid for the first 20 min and ap-proached a plateau by 30 min. An incubation periodof 30 min was used in the remainder of the experi-ments, since this period was sufflcient time for thedevelopment of a near-maximal chemotactic re-sponse and of a background response free of theapparent limitation seen after longer incubations.

Figure 3 shows the chemotactic and backgroundresponses by B. bacteriovorus strain UKi2 as afunction of the density of the suspension of bdello-vibrios in the Bdellovibrio compartment. There wasa linear increase in the chemotactic accumulationof cells in the test compartment as the number ofcells in the Bdellovibrio compartment was increased(the power law from the log-log plot is unity). Thebackground responses increased roughly in parallelto the chemotactic responses up to a density of 10'cells per ml; at higher cell densities, the backgroundresponses showed a plateau or a decrease in size,whereas the chemotactic responses continued to in-crease in linear fashion. The unknown factors caus-ing plateaus for the chemotactic responses in similarexperiments with the capillary assay system (1, 24)apparently were not limiting for chemotaxis in theBoyden chambers, but might have been affectingrandom swimming through the membrane at highcell densities. For subsequent experiments (exclud-ing that of Fig. 4), the Bdellovibrio compartment

0 104

O4

103

I0

0 10 20 30 40 50 60

INCUBATION, MINUTES

FIG. 2. Effect of time of incubation of the cham-bers on chemotaxis and random swimming by B.bacteriovorus strain UKi2 . (0O) Chemotaxic re-sponse: chambers contained 0.5%o (wtlvol) yeast ex-tract. (O) Random swimming: chambers containedsuspending medium. The points represent the aver-ages of responses obtained with two Boyden cham-bers; the error bars represent the individual re-sponses. Bdellovibrios were present at a density of 5X 108 cells per ml. Other conditions for the experi-ment were as described in the text.

108

107 [

06 F

zLu2

cr

ITt0u

v)uj

Z

v)0xm

0_jLIj0m

I05 -

64-107 108 109

CELLS/ML IN BDELLOV/BR/O COMPARTMENT

FIG. 3. Effect of the density of the suspension ofbdellovibrios on chemotaxis and random swimmingby B. bacteriovorus strain UKi2. (U) Chemotaxis;(0) random swimming. Notation and experimentalconditions as in the legend to Fig. 2. Nucleporemembrane filters were used in the chambers.

contained 108 cells per ml, a density within thelinear portion of the curve for chemotaxis and atthe peak or plateau for random swimming.Oxygen consumption was measured polarograph-

ically at 29.5°C (Yellow Springs Instruments model53 oxygen monitor).

RESULTSBoyden chamber assay system. Figure 4 is

a concentration-response curve for yeast ex-

I I

VOL. 132, 1977

on March 25, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

632 STRALEY AND CONTI

8

010E

C)E 7

.0

cn

0

2'amr

151)0 10 10 10

Percent Yeast ExtractFIG. 4. Chemotactic response of B. bacteriovorus

strain UKi2 toward various concentrations (percent,wtlvol) of yeast extract. Notation and experimentalconditions as in the legend to Fig. 2 except that thebackground response (random swimming) is shownby the point for 0% yeast extract.

tract; it shows a threshold concentration atabout 10-3% yeast extract and a peak concen-

tration at about 3 x 10-2S%. These data showthat the modified Boyden chamber yields infor-mation qualitatively similar to that obtainedwith the capillary system of Adler (1, 24, 33).The peak response occurs at a lower concen-

tration of yeast extract in the Boyden chamberwhen compared with that found with the capil-lary assay system (0.03% yeast extract as com-

pared with 0.25 to 0.5%).This may result from a limitation by the

membrane filter on the rate at which bdello-vibrios can enter the test compartment. If themaximal rate were reached for a concentrationof attractant lower than that eliciting the max-

imal response in the capillary assay, then stillhigher concentrations of attractant in the testcompartment could not elicit greater responsesin the Boyden chamber within the fixed periodof incubation. This difference in peak concen-

trations seen between the capillary and Boydenchamber assay systems means that differencesin attractiveness of high concentrations ofstrong attractants cannot be distinguished withthe Boyden chamber whereas they can withthe capillary system.Threshold concentrations for attraction by

yeast extract are the same when measuredwith the two assay systems, indicating that

the Boyden chamber system is as sensitive asthe capillary system for weak attractants andfor low concentrations of strong attractants.The number of cells entering the test com-

partment in response to 0.1% (wt/vol) yeastextract amounted to ca. 10% of the cells in theBdellovibrio compartment for strain UKi2 (33determinations). The number was ca. 4% forstrain UKi2Smr (11 determinations). The per-centage for strain UKi2 was twice that ob-tained with the capillary system (24).The average ratio of peak response (toward

yeast extract) to the background response (SMin the test compartment) measured with theBoyden chamber was at least twice that of thecapillary system (24). With the Boyden cham-ber, the ratio was ca. 74 for B. bacteriovorusstrain UKi2 (33 determinations) and ca. 94 forstrain UKi2Sm' (11 determinations).

There was considerable day-to-day variationin both the background and chemotactic re-sponses (for both assay systems [24, 331). Forthe Boyden chamber system, there was no ap-parent correlation between the background re-sponse and the chemotactic response to yeastextract. However, there was a correlation be-tween the responses to yeast extract and thechemotactic responses to attractant bacteria inreplicate experiments: strong responses toyeast extract were associated with greater re-sponses toward attractant bacteria. However,this relationship varied among different pairsof replicate experiments; for this reason, datafrom replicate experiments were not normal-ized: background responses and responses to0.1% (wt/vol) yeast extract are reported forevery experiment. Data were discarded if theratio of the response toward yeast extract tothe background response was less than 10:1.The comparisons show that the Boyden

chamber assay system, not previously appliedto studies of bacterial chemotaxis, is well suitedfor use in this study of chemotaxis by B. bacter-iovorus toward its prey. The prey are physicallyretained in the test compartment, whereas thebdellovibrios are free to enter in response toattractants excreted by the prey. The Boydenchamber also has the property of high sensitiv-ity at low concentrations of attractants.Chemotaxis toward susceptible bacteria.

Figures 5 and 6 are representative sets of datafor chemotaxis of B. bacteriovorus strain UKi2toward cells of E. coli grown in PYE and P./luorescens grown in TSB. P. fluorescens hadpreviously been determined to be susceptibleto predation by B. bacteriovorus strain UKi2.There was no significant attraction of bdello-vibrios by prey cells until densities of 10" and

J. BACTERIOL.

on March 25, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

CHEMOTAXIS TOWARD PREY BY BDELLOVIBRIO 633

10

Early Exponentiol Phase25 Min Preincubation

E 6

10

E~~~~~~~4

10500 10 10 10 10 0.1% yeast

E. co/i Cells/ml extract

FIG. 5. Chemotactic response of B. bacteriovorusstrain UKi2 toward exponential-phase cells of theprey E. coli grown in PYE and preincubated in SMfor 25 min. The circles and squares represent datataken in experiments run on different days. Eachpoint represents the average of responses obtainedwith from one to three chambers. The error barsrepresent the individual responses (where two cham-bers were used) or + one standard deviation of theresponses (where three chambers were used).

109 cells per ml. In all, seven tests were madewith E. coli grown in PYE (four tests for earlyexponential-phase cells, one for late exponen-tial phase, and two for stationary phase); andtwo tests were made with stationary-phase P.fluorescens, all employing 25-min preincuba-tions of the prey. In all tests these prey cellsdid not attract bdellovibrios when present atdensities as great as 107 cells per ml. For adensity of 10 P. fluorescens or E. coli cells perml, three of the nine tests yielded chemotacticresponses that were relatively weak but signif-icantly greater than the background responses.At a density of 109 cells per ml, both E. coliand P. fluorescens were chemotactically attrac-tive to B. bacteriovorus strain UKi2, and insome tests this attraction was strong.

Preincubation of the E. coli cells in bufferfor 90 min did not result in a buildup of attrac-tants relative to the levels present after 25 minof preincubation (four tests, two with earlyexponential- and two with stationary-phasecells grown in PYE).

In one experiment with E. coli cells grownto early exponential phase in PYE and prein-cubated in buffer for 90 min, some of thechambers contained Nuclepore membrane fil-

ters (which are 10 ,um thick as compared with150 gm for Millipore filters; data from themanufacturers). These gave data comparableto those from chambers containing Milliporemembrane filters. This indicated that prey cellsseparated from bdellovibrios by only about 10bdellovibrio-lengths were not more attractiveto bdellovibrios than ones separated by about150 cell-lengths.For the experiments described so far, the

prey cells were in a non-nutrient medium, andany attractants released would be due to endog-enous metabolism. Therefore, a test was madein which early exponential-phase E. coli cellsgrown in PYE were suspended in SM contain-ing 0.5% (wt/vol) glycerol, a compound readilymetabolized by the E. coli but not chemotacti-cally attractive to B. bacteriouorus (unpub-lished data). Glycerol also was present in thesuspending medium for the bdellovibrios andin the yeast extract solution. The metabolismof the glycerol by the E. coli cells during theexperiment did not increase the chemotacticattractiveness of these prey to B. bacteriovorus.A series of tests was made with E. coli cells

grown in a minimal medium, glycerol-saltsmedium, and suspended in SM alone or SMcontaining glycerol (Table 1). E. coli cells inSM without added glycerol showed no signifi-cant attractiveness toward bdellovibrios when

25 Min Preincubat,on I6 710

106c0 i6' 4+i

0 10 104 106 108 01% yeastextract

P f/uorescens Cells/ml

FIG. 6. Chemotactic response of B. bacteriovorusstrain UKi2 toward cells of the prey P. fluorescensgrown in TSB and preincubated in SM for 25 min.Notation as in the legend to Fig. 5. All points excepttwo represent averages of responses obtained withthree chambers (these two points are marked bycrosses).

VOL. 132, 1977

on March 25, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

634 STRALEY AND CONTI

the E. coli were present at densities lowerthan 109 cells per ml; E. coli present at 109cells per ml elicited a moderately strong re-

sponse from the bdellovibrios. However, whenglycerol was included in SM, the E. coli cellswere not attractive to bdellovibrios at any ofthe densities tested.The data of Table 1 coupled with those for

exponential-phase E. coli grown in PYE andpreincubated for 25 min show the followingpattern: the ratio of the chemotactic responsetoward 101 E. coli cells per ml to the back-ground response (toward SM) was greatest forcells grown in PYE and suspended in SM; nextgreatest were the ratios for cells grown in PYEand suspended in SM containing glycerol andfor cells grown in minimal medium and har-vested into SM; the lowest ratio (ca. unity)obtained for cells grown in minimal mediumand suspended in SM containing glycerol. Thispattern probably reflects the contributions ofseveral factors. (i) The prey cells probably ex-

perienced osmotic down-shocks when washedand suspended in SM. Osmotic down-shocksaccompanying the washing of cells grown in

rich organic medium can cause transient leak-age of a wide variety of compounds into themore dilute washing medium (4, 11, 12). Ifcompounds that were attractants to B. bacter-iovorus were released, then prey which re-

ceived the strongest osmotic shock (E. coligrown in PYE and suspended in SM) would

release more attractants into SM than thosethat received weaker shocks. (ii) Prey grown

in a richer medium might have more concen-

trated internal pools of attractive compounds(such as potassium, a strong attractant forbdellovibrios [unpublished data], and any or-

ganic compounds serving an osmoregulatoryrole for the cells [4, 11, 12]) than do cells grownin minimal medium. Upon osmotic shock, cellswith richer pools of compounds would be ex-

pected to release a greater amount of thesecompounds into SM than would cells with lessconcentrated pools. (iii) Glycerol in SM mightdecrease the osmotic shocks experienced byprey grown in minimal medium during wash-ing and suspension in SM. (iv) Oxygen con-

sumption by the prey is greater for cells har-vested into SM containing glycerol than forcells harvested into SM alone.The effect of oxygen depletion might be to

elicit a repulsion, or negative chemotaxis,which would counteract any attraction to prey.

This possibility was explored in separate exper-

iments (see later). A consideration of the effectsof the media in which the prey were grown andharvested (factors i through iii above) suggeststhat the smallest artifacts due to osmotic shockshould be in measurements made with cellsgrown in minimal medium and suspended in

SM containing glycerol.One test with E. coli cells was made with B.

bacteriovorus strain 114, which is obligately

TABLE 1. Chemotaxis of B. bacteriovorus toward susceptible and nonsusceptible bacteria grownin minimal media

Response,a bdellovibrios entering test compartment (x 104)

Attractant bacteria 0.1% (wt/vol)No cells 10' cells/ml 10' cells/ml 109 cells/ml yeast extract

E. coli in SMb 85 6 65 18 65 13 120 + 80 1400 ± 10020 10 26 8 27 7 110 ± 60 690 ± 32011 8 14 7 22 5 59 51 420 ± 100

E. coli in SM + glycerol" 1.6 + 0.8 1.8 + 0.6 1.9 + 0.6* 1.9 + 0.7* 150 ± 808.5 2.0 8.9 3.2 27 16 27 21 460 + 706.2 + 4.6 7.5 ± 6.3 2.8 + 1.5 1.0 + 0.0* 170 ± 604.4 1.8 7.8 3.0 13 + 4 7.8 0.7 260 ± 2013 6 8.1 4.2 7.0 ± 0.6 4.7 2.6 1300 200

B. megaterium in SM + glucosed 38 + 15 7.4 + 4.2 7.6 ± 2.1 9.7** 390 ± 9011 In all tables, unless otherwise noted, the numbers are averages of responses obtained with three

chambers. The limits of error are ± one standard deviation of the responses. * Two chambers were used;error bars give the individual responses. ** One chamber was used.

b B. bacteriovorus strain UKi2 and exponential-phase cells of E. coli preincubated for 30 min.C B. bacteriovorus strain UKi2 and exponential-phase cells of E. coli harvested into SM containing the

same concentration of glycerol as was in the growth medium. Top two sets of data: E. coli cells werepreincubated for 25 min. Bottom three sets of data: preincubation was for 30 min.

d B. bacteriovorus strain UKi2Smr and B. megaterium cells harvested into SM containing the sameconcentration of glucose as was in the growth medium. Preincubation was for 25 min.

J. BACTERIOL.

on March 25, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

CHEMOTAXIS TOWARD PREY BY BDELLOVIBRIO 635

predacious. The E. coli cells (early exponentialphase; grown in PYE) were not chemotacticallyattractive to this strain of B. bacteriovorus,thus making it unlikely that the facultativenature of strain UKi2 was associated with itsweak chemotactic response to its prey.One test was made to see if any substances

released from prey cells during the predaciousgrowth of strain UKi2 would be attractive tobdellovibrios of the same strain. Cultures wereinitiated, having E. coli cells present at 10'cells/ml and bdellovibrios added so as to pro-duce MOIs ranging from 0 to 5. These weregrown for 12.5 h, the cells were removed byfiltration, and the filtrates were tested forchemotactic attractiveness. Microscopic obser-vations of the suspensions prior to filtrationrevealed that predation by bdellovibrios in thecultures ranged from essentially none (no E.coli cells seen containing bdellovibrios) to anadvanced stage (unattacked E. coli rarelyseen). The filtrates from the cultures were notattractive to the bdellovibrios, nor were thechemotactic responses less than the back-ground response, which would indicate repul-sion, or negative chemotaxis (35).Chemotaxis toward nonsusceptible bacte-

ria. B. megaterium will not serve as prey forB. bacteriovorus strain UKi2 (unpublisheddata). This nonsusceptible species, which issensitive to streptomycin, was tested for chem-otactic attractiveness to a streptomycin-resist-ant derivative of B. bacteriovorus strain UKi2.The media used in the subsequent plating anal-ysis contained streptomycin, which preventedovergrowth by B. megaterium. In all, threetests were made, one with exponential- andtwo with stationary-phase cells grown in TSBand preincubated for 25 min (Fig. 7). Therewas no detectable chemotactic response untilthe prey density reached 107 cells per ml. Athigher densities, there was an increasinglystrong response, which reached a maximallevel almost as high as the response to yeastextract and which declined to the backgroundlevel, presumably due to saturation of thechemotactic machinery of the bdellovibrios (1).

Preincubation of the B. megaterium cells for90 min did not make the cells more attractiveto B. bacteriovorus than they were after 25min of preincubation (data of Fig. 7 and onetest for exponential-phase cells).A control experiment was run to test the

adequacy of washing of the B. megateriumcells after their growth in the rich TSB me-dium. Stationary-phase B. megaterium cellswere washed 2, 4, and 6 times with SM beforetesting for chemotactic attractiveness to B.

90 Min Preincubot,ol

O 106E

c):

5910

410 A

2 4 6 80 1100 10 10 0 %

yeas8 meqaferlum Cells/ml exraoct

FIG. 7. Chemotaxis of B. bacteriovorus strainUKi2Smr toward stationary-phase cells of the non-susceptible bacterium B. megaterium preincubatedin SM for 91 min. Notation as in the legend to Fig.5; all points represent averages of responses obtainedwith three chambers.

bacteriovorus. After each pair of washes, thedensity of the suspension was determined, anda sample was diluted to 10W cells per ml andpreincubated for 25 min in SM prior to testingfor chemotactic attractiveness. There was nodifference in the attractiveness of the B. mega-terium washed twice, as usual, or four to sixtimes (and the background responses and re-sponses to 0.1% yeast extract remained con-stant throughout the experiment). The strongchemotactic responses by B. bacteriovorus tohigh densities of B. megaterium therefore werenot due to carryover of attractants from thegrowth medium of the cells.The main conclusions from these experi-

ments with B. megaterium are that, as withE. coli, densities of bacilli below 107 cells perml failed to attract bdellovibrios. High densi-ties of B. megaterium (101 cells per ml) at-tracted B. bacteriovorus more strongly thansuspensions of E. coli having similar cell dens-ities. This might have been due in part to thelarger volume of B. megaterium cells as com-pared with that of E. coli cells; larger cellswould be expected to release more material percell than smaller cells.

One test was made in which the B. megate-rium cells were grown in a glucose-salts mini-mal medium and harvested into SM containingglucose (Table 1); as with E. coli (Table 1),these cells failed to attract bdellovibrios fordensities as high as 109 cells per ml.

VOL. 132, 1977

on March 25, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

636 STRALEY AND CONTI

Other nonsusceptible bacteria tested forchemotactic attractiveness to B. bacteriovorusstrain UKi2Smr were B. bacteriovorus strainsD and UKi2 (Table 2; tests have been madeshowing the inability ofB. bacteriovorus strainUKi2Smr to grow on B. bacteriocorus strainsUKi2 and D). Both of these strains are strepto-mycin sensitive; therefore their interferencewith the plating analysis was prevented by thepresence of streptomycin. These prey also failedto elicit detectable chemotactic responses whenpresent at densities lower than 10O cells perml. For strain UKi2, the response elicited by10 cells per ml was significantly greater thanthe background response, whereas for strain Dit was not. For the nonpredaciously grown

UKi2, the cells ranged from elongated vibriosto long spirals many times the length of preda-ciously grown bdellovibrios; this may accountfor the stronger chemotactic response by B.bacteriovorus strain UKi2Smr to strain UKi2than to strain D.

Effect of oxygen consumption by prey. It is

unlikely that the failure of bdellovibrios to beattracted to the lower densities of susceptiblebacteria was due to depletion of oxygen by theprey, causing a repulsion of bdellovibrios thatbalanced attraction. The endogenous respira-tion of B. bacteriouorus strain UKi2 is about

15 times lower than that of an equal number ofearly exponential-phase E. coli cells. The oxy-gen consumption by bdellovibrios would begreater than that by the prey for densities ofprey of less than about 7 x 1Of cells per ml.However, during a 30-min experiment, E.

coli present at 10O cells per ml would use moreoxygen than would the bdellovibrios. For 10OE. coli cells per ml, all of the oxygen originallypresent in the suspension would be consumedduring the experiment; any oxygen present atthe end of the experiment would have enteredby diffusion. Thus, at high densities of prey(10O and 10' cells per ml), a strong gradient ofoxygen concentration would be established inthe Boyden chamber. The bdellovibrios mightbe able to respond to such a gradient by repul-sion, which would cause them to remain in theBdellovibrio compartment.

Table 3 shows the chemotactic response ofbdellovibrios to exudates prepared by filtrationof a dense suspension of E. coli cells that hadbeen grown either in PYE or in glycerol-saltsmedium. The resulting exudate was seriallydiluted and tested for chemotactic attractive-ness. For these experiments, the prey cellswere absent, but any stable substances ex-

creted by them during the 25-min preincuba-tion (see above) would be present. The concen-

TABLE 2. Chemotaxis of B. bacteriouorus strain UKi2Smr toward B. bacteriouorusResponse, bdellovibrios entering test compartment ( x 104)

Attractant bacteria 0.1% Wt/volNo cells 10 cells/ml 10" cells/ml 109 cells/ml yeast extract

B. bacteriouorus strain DI 4.1 + 1.4 3.8 + 2.3 13 +10 25 + 10- 120 + 40*B. bacteriouorus strain UKi2b 1.1 + 0.5 2.2 + 0.3 7.8 + 3.1 15 + 6 54 + 12*a Predaciously grown B. bacteriovorus strain D was harvested into SM and preincubated for 48 min. The

suspension contained some E. coll cells (see text).Nonpredaciously grown B. bacterioLorus strain UKi2 was harvested into SM and preincubated for 135

min. For definition of asterisk, see footnote a, Table 1.

TABLE 3. Chemotaxis ofB. bacteriovorus strain UKi2 toward exudates from E. coliTreatment ofE. coli by: Response, bdellovibrios entering test compartment ( x 105)

No cells, fil- 10' cells/ 3 x 10' cells/ 109 cells/ 0.1% (wt/Growth Harvesting No cells teredb ml" ml vol) yeastextract

PYE SM 2.1 1.1 1.3 + 0.1 1.5 + 0.1 1.1 - 0.2 3.7 - 1.3 51 + 41.0 0.4 2.7 ± 0.3 1.0 ± 0.3 2.7 ± 0.3 9.4 ± 2.2 71 ± 20

Glycerol-salts SM 2.4 + 0.6* 1.8 + 0.5 2.2 + 0.2 3.3 + 1.8 4.5 - 0.7 57 - 191.8 + 0.7* 4.1 + 0.9 3.2 + 1.2 12 + 5 11 + 8 38 + 15

Glycerol-salts SM + glycerol 2.3 + 1.1 1.5 + 0.4 1.4 +0.5 6.8 1.7 70 +261.4 ± 0.5 1.3 0.6 2.6 1.0 4.8 1.1 58 3

a E. coli cells were grown to exponential phase, harvested, and then preincubated for 25 min before being removed fromSM by filtration. The resulting filtrate was diluted and tested for chemotactic attractiveness. When glycerol was presentin SM, its concentration was that used in glycerol-salts medium.

" A volume of SM was filtered and diluted into unfiltered SM, mimicking the handling of the exudate from 109 E. collcells per ml.

"Cells per ml" denotes the density of the E. coli suspension corresponding to the diluted exudate. For definition ofasterisk, see footnote a, Table 1.

J. BACTERIOL.

on March 25, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

CHEMOTAXIS TOWARD PREY BY BDELLOVIBRIO 637

tration of oxygen in all samples would be equaland would remain constant during the experi-ment. This, in effect, represented a measure-ment of chemotaxis toward the stable attrac-tants produced by E. coli cells in the first 25min of an experiment without preincubation.

For technical reasons, the starting of incu-bations of the three replicate chambers for agiven data point were separated by 10-minintervals; therefore, 20 min elapsed betweenthe filling of the first and third chambers foreach test. There was no systematic decrease inchemotactic response obtained with the thirdset of chambers in any experiment. This indi-cated that any attractants present probablywere stable during the experiments (whichbegan 8 min after filtration to make the concen-trated exudate).The main conclusion to be drawn from Table

3 is that all of the exudates from densities ofE. coli as high as 108 cells per ml failed toelicit a chemotactic response greater than thebackground response. In four out of five of thetests, exudates from 3 x 108 cells per ml didnot elicit a significant chemotactic response.These results indicate, but do not prove, thatoxygen gradients did not cause appreciablerepulsion during tests for chemotaxis towarddensities of prey as high as 108 cells per ml.A second observation from the data in Table

3 is that exudates from 109 E. coli cells per mlin the presence of glycerol elicited a chemotac-tic response, whereas the corresponding sus-pension of intact cells (Table 1) did not. Perhapsthe strong oxygen gradient present under theseconditions did elicit repulsion which cancelledan attraction to intact cells at the highestconcentration tested.

DISCUSSION

The goal of this work was to see if bdellovib-rios locate their prey by means of chemotaxis.All but one of the experiments employed thefacultatively predacious B. bacteriovorus strainUKi2 (or its streptomycin-resistant derivative),originally isolated from sewage. One test wasmade with an obligately predacious strain 114originally isolated from soil. Susceptible bacte-ria tested for attractiveness were E. coli, whichis abundant in sewage, and P. fluorescens,commonly found in soil and water. The nonsus-ceptible cells tested were B. megaterium, whichis widely distributed in soil and water; Bbacteriovorus strain UKi2 itself; and B. bacter-iovorus strain D, a Bdellovibrio originally iso-lated from soil. This selection of attractantbacteria thus includes both susceptible andnonsusceptible species prevalent in various en-

vironments known also to contain bdellovib-rios.To evaluate the data shown in the Tables

and in Fig. 5 to 7 in terms of the usefulness ofchemotaxis for the location of prey by B. bacter-iovorus, we need to translate them into dis-tances through which prey can be detected bybdellovibrios. As pointed out by Hespell et al.(20), the probability that a randomly swimmingbdellovibrio will collide with a prey cell in-creases in proportion to the square of the radiusof the prey cell. If the prey is chemotacticallyattractive to bdellovibrios, its "radius" includesthe radius of the sphere through which theattraction is effective. Thus, a chemotactic at-traction that could substantially increase theeffective radius of the prey cell would greatlyreduce the concentration of susceptible bacteriathat must be present in an envrionment forbdellovibrios to be able to survive (20).We can make a rough translation of our

chemotaxis data by referring to a simplifiedmodel of the Boyden chamber (Fig. 8). If bdel-lovibrios are able to locate individual prey cellsby means of chemotaxis, then each prey cell inthe suspension is surrounded by a sphericalvolume, the radius of which is the distanceattractants diffuse before becoming too dilute

FIG. 8. Representation of weak attraction by preycells in the modified Boyden chamber used for thiswork. Upper panel: simplified diagram of a longitu-dinal section through a modified Boyden chamber(see text). Lower panels: distribution of attractiveareas (intersections with the micropore membrane ofattractiue volumes surrounding prey cells) over thesurface of the micropore membrane as viewed fromthe Bdellovibrio compartment (see text). (A) 107 preycells per ml in test compartment. (B) 108 prey cellsper ml in test compartment. These lower panels weregenerated by a computer, which distributed the preycells randomly through space and plotted the loca-tions of prey, whose attractive spheres intersectedwith the micropore membrane. For convenience indrawing, the radius of each resulting attractive areawas rounded off to the nearest of5 radii (representingthe nearest 0.25 cm in the original drawing).

Bdello,,Ibro eyPr TestComportment Cell Comportm.ent

P,ot,.d,,ngAft,.octive *-M,cropo,e Mem..brone

(-'B

A

VOL. 132, 1977

A

I

/---, <7r -1

1,

on March 25, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

638 STRALEY AND CONTI

to be detected by the chemotactic machinery ofthe bdellovibrios. A prey cell in the test com-partment of the Boyden chamber will be ableto exert an attractive force on bdellovibriosonly if it is near enough to the membrane(represented by a line in Fig. 8) that some ofits "attractive volume" protrudes into the Bdel-lovibrio compartment. A threshold chemotacticresponse toward individual cells should occurin the Boyden chamber when a large fractionof the surface of the micropore membrane iscovered with nonoverlapping attractive vol-umes (so that bdellovibrios swimming just nextto the membrane have a large probability ofencountering an attractive volume due to asingle prey cell).The lower two panels in Fig. 8 represent the

surface of the membrane as seen from insidethe Bdellovibrio compartment. The circles ofvarious diameters represent the areas of theintersections between the micropore membraneand the attractive volumes of prey cells in thetest compartment.

Figure 8A shows the distribution of attrac-tive areas for a density of prey eliciting a small(threshold) chemotactic response from the bdel-lovibrios; it contains the assumption that theradius of the attractive volume is approxi-mately equal to the average distance separat-ing prey cells at the threshold density. A gen-erous estimate for this density is 1 x 107 to 2 x107 cells per ml. Chemotaxis was detected forthis density only in two of three experimentswith stationary-phase B. megaterium grown inTSB. At a density of 107 cells per ml, theaverage distance separating prey cells is 29,um; therefore, in Fig. 8A the radius of thelargest attractive area represents 29 ,um. Thesurface in Fig. 8A is only half covered becauseattractive volumes can arise from only one sideof the membrane, the side facing the test com-partment. A density of 2 x 107 cells per mlwould give approximately complete coverageof the membrane with attractive areas. Someof these (the smaller circles in the figure) wouldrepresent volumes that would be too small tohave an effect eliciting chemotaxis. However,the larger volumes should be sufficient to en-compass several of the intervals of smoothswimming and abrupt changes of direction thatmake up the three-dimensional swimming pat-tern of a bacterium (reference 3 and our roughmeasurements from "motility tracks" [371 of B.bacteriovorus). Bdellovibrios entering theselarger protruding attractive volumes would ex-perience an attraction, and could respond, pre-sumably by biasing their three-dimensionalrandom walk (3). A small chemotactic responseshould be measured.

Figure 8B illustrates the distribution of at-tractive areas when prey cells are present at adensity of l0X cells per ml. The radius of thelargest circle still represents 29 ,um; and forconvenience, only these largest circles weredrawn. For this density of prey, the membraneis heavily covered with attractive areas, manyof which are overlapping; a larger chemotacticresponse should be measured at this prey den-sity than at 107 prey cells per ml.

For the susceptible cells tested after growthin PYE, chemotactic responses were obtainedin only 5 out of 13 experiments when the preydensity was l0X cells per ml. When E. coli orB. megateriurn was grown in minimal medium,a density of 10' cells per ml failed to elicit achemotactic response in all experiments (Table1). In these latter experiments, the prey shouldhave received smaller osmotic shocks uponwashing and suspension than they would haveafter growth in PYE or TSB. Those experi-ments would therefore be expected to containless of an artifactual attraction due to sub-stances released upon osmotic shock. This prob-ably accounts for much of the difference seenbetween the results of Fig. 5 through 7 andTable 1. The observation that 90-min preincu-bations of E. coll or B. nmegaterium in suspend-ing medium did not increase its attractivenessto bdellovibrios is consistent with the view thatmost of the attractants detected for cells grownin rich media and present at a density of l08cells per ml were released upon osmotic shock(a transient effect). E. coll cells grown in PYEare larger than those grown in glycerol-saltsmedium, but not by the factor of 10 differenceseen in their apparent attractiveness to bdel-lovibrios. Oxygen gradients due to respirationby E. coli apparently did not complicate meas-urements of chemotaxis at a density of 10' E.coli cells per ml. Therefore, we feel that thethreshold density of prey for the attraction ofbdellovibrios actually was greater than 107 cellsper ml and probably exceeded 10 cells per ml.This means that B. bacteriovorus strain UKi2probably does not hunt down individual preycells from a long distance (greater than 29 ,um).

It may also mean that chemotaxis plays norole in attracting the predators to prey atshorter distances. At the subthreshold densityof 10X cells per ml, the average distance sepa-rating prey cells is 13.4 ,um; this suggests thatattractive volumes of radius 13.4 ,um are noteffective in eliciting net chemotactic responsesfrom bdellovibrios. If the average straight-lineswimming distance for bdellovibrios is at leastas great as that of E. coll (ca. 12 ,tm [31), thenattractive volumes having radii of r 12 ,malso are not likely to be effective.

J. BACTERIOL.

on March 25, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

CHEMOTAXIS TOWARD PREY BY BDELLOVIBRIO 639

The situation in the Boyden chambers inactual experiments differs from the descriptiondeveloped above in two main respects. First,the membrane is not infinitely thin; it is acellulose mesh 150 gm thick (for Milliporemembranes) or a 10-g.m-thick plastic barriercontaining cylindrical pores (for Nucleporemembranes). In either case, the main effectswould be to decrease the numbers of bdellovib-rios per unit volume next to the test compart-ment and to limit the freedom of movement ofbdellovibrios within the membrane. Since sim-ilar results were obtained with the two typesof filters and since the filters clearly are readilypenetrated by bdellovibrios, we assume thatthese barriers do not appreciably affect ouranalysis.

Second, the attractant bacteria were prein-cubated 25 or more min before being tested forattractiveness. Some preincubation was una-voidable, occurring while the suspension wasenumerated, serially diluted, and put into sy-ringes. The shortest and most convenient timefor such handling was 25 min. At the time oftesting, the suspensions contained a uniformbackground concentration of any stable attrac-tants released by the cells during the preincu-bation. During the experiment, these "back-ground" attractants would diffuse in a one-dimensional pattern, and on top of this patternwould be superimposed the radial diffusionsurrounding each cell. At very high densitiesof attractant bacteria, a101 cells per ml, theattractive volumes from the cells would overlapextensively, and one-dimensional diffusionwould be approximated. The net effect of thepreincubation is to emphasize the weakness ofany chemotaxis shown by bdellovibrios towardprey since high densities of prey, present alongwith a background of any compounds releasedduring the preincubation, failed to attract bdel-lovibrios.We obtained no evidence that susceptible

bacteria are more attractive than nonsuscepti-ble bacteria; B. bacteriovorus strain UKi2 prob-ably does not use chemotaxis to distinguishsusceptible from nonsusceptible cells.

Tlae experiment in which filtrates frommixed cultures were tested for attractivenessindicated that B. bacteriovorus strain UKi2neither searched for nor avoided infected preyby means of chemotaxis.We conclude that B. bacteriovorus strain

UKi2 probably locates its prey by means ofrandom swimming. The test made with B.bacteriovorus strain 114 further suggests thatit also does not use chemotaxis to locate preycells. These conclusions are consistent with therapidly accumulating evidence from field work

(13-15, 23, 25) and laboratory model studies(16, 17, 22) that bdellovibrios are able to in-crease in numbers only when the density ofprey is relatively high: 5 x 10 to 1 x 10 cellsper ml. These numbers are consistent with therough estimate that 1.5 x 105 prey cells per mlare required for a randomly swimming bdello-vibrio to have a 50% probability of striking aprey cell (20).

Bdellovibrios probably do play a role in theself-purification of polluted waters and soils inwhich the prey density is high. However, addi-tional biological factors, including other bacte-rial predators (15, 18, 27), predacious protozoa(6, 7, 10, 27, 29), invertebrate predators (27),algal toxins (27), competition for nutrients, andperhaps lysis by viruses (5); and physical proc-esses such as dilution, adsorption, and settling(13, 27) interact to further decrease the bacte-rial population.The question remains as to what role, if any,

chemotaxis plays in the predacious life cycle ofB. bacteriovorus. It is possible that bdellovib-rios might be attracted to large microcoloniesof prey cells. It is estimated that, in soils, mostof the bacteria are located on humus particlesand occur as microcolonies (19). The averagesize of such colonies is small (ca. 4 cells); butsome are as large as 100 cells (19). This numberof cells concentrated in a small area couldcorrespond to a denser aqueous suspensionthan was tested for prey cells in our studies.Such dense populations might release metabo-lites in large enough local concentration toattract nearby bdellovibrios to their vicinity.However, we speculate that any such chemo-tactic attraction would be relatively weak; oth-erwise, microcolonies of nonsusceptible preywould entrap bdellovibrios in a situation inwhich they could not grow.A second and less direct possible role of

chemotaxis in predation by B. bacteriovorus isattraction to sources of compounds (e.g., humusparticles in soil) that can serve as nutrients forpotential prey cells or that are needed forcellular maintenance by the predators them-selves. An investigation of the range of com-pounds that attract bdellovibrios is currentlyin progress.

Finally, negative chemotaxis and chemotaxistoward oxygen (aerotaxis) could function forBdellovibrio as they apparently do for otherbacteria. Negative chemotaxis could allow thepredators to avoid harmful compounds; andaerotaxis could allow the aerobic bdellovibriosto locate regions having an oxygen concentra-tion that is optimal for motility and growth.These possible roles of chemotaxis have not yetbeen investigated.

VOL. 132, 1977

on March 25, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

640 STRALEY AND CONTI

ACKNOWLEDGMENTS

This work was supported by National Science Founda-tion grant PCM75-19771.We express gratitude to Joseph P. Straley, Department

of Physics, University of Kentucky, who wrote the com-puter program used to generate the distributions of attrac-tive areas, shown in Fig. 8.

LITERATURE CITED

1. Adler, J. 1973. A method for measuring chemotaxisand use of the method to determine optimum condi-tions for chemotaxis by Escherichia coli. J. Gen.Microbiol. 74:77-91.

2. Althauser, M., W. A. Samsonoff, C. Anderson, and S.F. Conti. 1972. Isolation and preliminary characteri-zation of bacteriophages for Bdellovibrio bacteriovo-rus. J. Virol. 10:516-523.

3. Berg, H. C., and D. A. Brown. 1972. Chemotaxis inEscherichia coli analyzed by three-dimensional track-ing. Nature (London) 239:500-504.

4. Britten, R. J., and F. T. McClure. 1962. The aminoacid pool in Escherichia coli. Bacteriol. Rev. 26:292-235.

5. Carlucci, A. F., and D. Pramer. 1960. Evaluation offactors affecting the survival of Escherichia coli inseawater. IV. Bacteriophages. Appl. Microbiol.8:254-256.

6. Curds, C. R. 1973. The role of protozoa in the activatedsludge process. Am. Zool. 13:161-169.

7. Curds, C. R., and G. J. Fey. 1969. The effect of ciliatedprotozoa on the fate of Escherichia coli in the acti-vated-sludge process. Water Res. 3:853-867.

8. Denning, J. J. V., and R. R. Davies. 1973. Candidaalbicans and the chemotaxis of polymorphonuclearneutrophils. Sabouraudia 11:210-221.

9. Diedrich, D. L., C. F. Denny, T. Hashimoto, and S. F.Conti. 1970. Facultatively parasitic strain of Bdello-uibrio bacteriovorus. J. Bacteriol. 101:989-996.

10. Enzinger, R. M., and R. C. Cooper. 1976. Role ofbacteria and protozoa in the removal of Escherichiacoli from estuarine waters. Appl. Environ. Microbiol.31:758-763.

11. Epstein, W., and S. G. Schultz. 1965. Cation transportin Escherichia coli. V. Regulation of cation content.J. Gen. Physiol. 49:221-234.

12. Epstein, W., and S. G. Schultz. 1968. Ion transportand osmoregulation in bacteria, p. 186-193. In L. B.Guze (ed.), Microbial protoplasts, spheroplasts, andL-forms. Williams and Wilkins, Baltimore.

13. Fry, J. C., and D. G. Staples. 1974. The occurrence androle of Bdellovibrio bacteriovorus in a polluted river.Water Res. 8:1029-1035.

14. Guelin, A., and L. Cabioch. 1970. Sur l'utilisation dephenomene de bacteriolyse spontanee pour la con-naissance de l'6tat sanitaire des eaux douces et ma-rines. C. R. Acad. Sci. Paris 271:137-140.

15. Guelin, A., and L. Cabioch. 1972. Bacteriolyse sponta-nee et pouvoir bactericide des eaux douces et marines;isolement d'un nouveau micropredateur. C. R. Acad.Sci. Paris 274:3317-3319.

16. Guelin, A., and L. Cabioch. 1974. Caracteres dyna-miques de l'interaction entre le micropredateur Bdel-louibrio bacteriovorus et la bacterie-hote en fonctionde leurs densites initiales respectives. C. R. Acad.Sci. Paris 278:1293-1296.

17. Guelin, A., and P. Lepine. 1973. Multiplication del'agent bacteriolytique Bdellovibrio bacteriwvorus etintensite de la bact6riolyse en fonction de la densiteinitiale en bacteries-hotes. C. R. Acad. Sci. Paris276:1075-1076.

18. Guelin, A., P. Lepine, D. Lamblin, and J. Sisman.

J. BACTERIOL.

1968. Isolement d'un parasite bacterien sur lesgermes Gram-positifs a partir d'echantillons d'eaupolluee. C. R. Acad. Sci. Paris 266:2508-2509.

19. Gray, T. R. G., P. Baxby, I. R. Hill, and M. Goodfel-low. 1968. Direct observation of bacteria in soil, p.171-197. In T. R. G. Gray and D. Parkinson (ed.),The ecology of soil bacteria. University of TorontoPress, Toronto.

20. Hespell, R. B., M. F. Thomashow, and S. C. Ritten-berg. 1974. Changes in cell composition and viabilityofBdellovibrio bacteriouorus during starvation. Arch.Microbiol. 97:313-327.

21. Kaiser, A. D., and D. S. Hogness. 1960. The transfor-mation of Escherichia coli with deoxyribonucleic acidisolated from bacteriophage Xdg. J. Mol. Biol. 2:392-415.

22. Keya, S. O., and M. Alexander. 1975. Factors affectinggrowth of Bdellouibrio on Rhizobium. Arch. Micro-biol. 103:37-43.

23. Klein, D. A., and L. E. Casida, Jr. 1967. Occurrenceand enumeration of Bdellouibrio bacteriovorus in soilcapable of parasitizing Escherichia coli and indige-nous soil bacteria. Can. J. Microbiol. 13:1235-1241.

24. LaMarre, A. G., S. C. Straley, and S. F. Conti. 1977.Chemotaxis toward amino acids by Bdellovibrio bac-teriovorus. J. Bacteriol. 131:201-207.

25. Lambina, V. A., N. A. Chuvilskaya, L. A. Ledova, A.V. Afinogenova, and l. V. Averberg. 1974. Quantita-tive characteristics of distribution of Bdellouibriobacteriouorus in river water. Microbiologia 43:715-720.

26. Mesibov, R., and J. Adler. 1972. Chemotaxis towardamino acids in Escherichia coli. J. Bacteriol. 112:315-326.

27. Mitchell, R. 1968. Factors affecting the decline of non-marine microorganisms in seawater. Water Res.2:535-543.

28. Seidler, R. J., and M. P. Starr. 1969. Factors affectingthe intracellular parasitic growth ofBdellouibrio bac-teriovorus developing within Escherichia coli. J. Bac-teriol. 97:912-923.

29. Small, E. B. 1973. A study of ciliate protozoa from asmall polluted stream in east-central Illinois. Am.Zool. 13:225-230.

30. Starr, M. P., and R. J. Seidler. 1971. The bdellovibrios.Annu. Rev. Microbiol. 25:649-678.

31. Stolp, H. 1968. Bdellovibrio bacteriovorus, ein rauber-ischer Bakterienparasit. Naturwissenschaften 55:57-63.

32. Stolp, H. 1973. The bdellovibrios: bacterial parasites ofbacteria. Annu. Rev. Phytopathol. 11:53-76.

33. Straley, S. C., and S. F. Conti. 1974. Chemotaxis inBdellovibrio bacteriovorus. J. Bacteriol. 120:549-551.

34. Taylor, B. L., and D. E. Koshland, Jr. 1974. Reversalof flagellar rotation in monotrichous and peritrichousbacteria: generation of changes in direction. J. Bac-teriol. 119:640-642.

35. Tso, W.-W., and J. Adler. 1974. Negative chemotaxisin Escherichia coli. J. Bacteriol. 118:560-576.

36. Tuominen, F. W., and R. W. Bernlohr. 1971. Pyruvatekinase of the spore-forming bacterium, Bacillus li-cheniformis. I. Purification, stability, regulation ofsynthesis, and evidence for multiple molecularstates. J. Biol. Chem. 246:1732-1745.

37. Vaituzis, Z., and R. N. Doetsch. 1969. Motility tracks:technique for quantitative study of bacterial move-ment. Appl. Microbiol. 17:584-588.

38. Varon, M., and M. Shilo. 1968. Interaction ofBdellovib-rio bacteriovorus and host bacteria. I. Kinetic stud-ies of attachment and invasion of Escherichia coli Bby Bdellovibrio bacteriovorus. J. Bacteriol. 95:744-753.

on March 25, 2020 by guest

http://jb.asm.org/

Dow

nloaded from