APPLIED AND ENVIRONMENTAL MICROBIOLOGY, May 2002, p. 2495–2502 Vol. 68, No. 50099-2240/02/$04.00�0 DOI: 10.1128/AEM.68.5.2495–2502.2002Copyright © 2002, American Society for Microbiology. All Rights Reserved.

Metabolic Commensalism and Competition in a Two-SpeciesMicrobial Consortium

Bjarke B. Christensen,1,2 Janus A. J. Haagensen,1 Arne Heydorn,1 and Søren Molin1*BioCentrum-DTU, Molecular Microbial Ecology Group, Technical University of Denmark, DK-2800 Lyngby,1 and Division of

Microbiological Safety, Danish Veterinary and Food Administration, DK-2860 Søborg,2 Denmark

Received 23 October 2001/Accepted 6 February 2002

We analyzed metabolic interactions and the importance of specific structural relationships in a benzylalcohol-degrading microbial consortium comprising two species, Pseudomonas putida strain R1 and Acineto-bacter strain C6, both of which are able to utilize benzyl alcohol as their sole carbon and energy source. Theorganisms were grown either as surface-attached organisms (biofilms) in flow chambers or as suspendedcultures in chemostats. The numbers of CFU of P. putida R1 and Acinetobacter strain C6 were determined inchemostats and from the effluents of the flow chambers. When the two species were grown together inchemostats with limiting concentrations of benzyl alcohol, Acinetobacter strain C6 outnumbered P. putida R1(500:1), whereas under similar growth conditions in biofilms, P. putida R1 was present in higher numbers thanAcinetobacter strain C6 (5:1). In order to explain this difference, investigations of microbial activities andstructural relationships were carried out in the biofilms. Insertion into P. putida R1 of a fusion between thegrowth rate-regulated rRNA promoter rrnBP1 and a gfp gene encoding an unstable variant of the greenfluorescent protein made it possible to monitor the physiological activity of P. putida R1 cells at differentpositions in the biofilms. Combining this with fluorescent in situ hybridization and scanning confocal lasermicroscopy showed that the two organisms compete or display commensal interactions depending on theirrelative physical positioning in the biofilm. In the initial phase of biofilm development, the growth activity ofP. putida R1 was shown to be higher near microcolonies of Acinetobacter strain C6. High-pressure liquidchromatography analysis showed that in the effluent of the Acinetobacter strain C6 monoculture biofilm themetabolic intermediate benzoate accumulated, whereas in the biculture biofilms this was not the case, sug-gesting that in these biofilms the excess benzoate produced by Acinetobacter strain C6 leaks into the surround-ing environment, from where it is metabolized by P. putida R1. After a few days, Acinetobacter strain C6 colonieswere overgrown by P. putida R1 cells and new structures developed, in which microcolonies of Acinetobacterstrain C6 cells were established in the upper layer of the biofilm. In this way the two organisms developedstructural relationships allowing Acinetobacter strain C6 to be close to the bulk liquid with high concentrationsof benzyl alcohol and allowing P. putida R1 to benefit from the benzoate leaking from Acinetobacter strain C6.We conclude that in chemostats, where the organisms cannot establish in fixed positions, the two strains willcompete for the primary carbon source, benzyl alcohol, which apparently gives Acinetobacter strain C6 a growthadvantage, probably because it converts benzyl alcohol to benzoate with a higher yield per time unit than P.putida R1. In biofilms, however, the organisms establish structured, surface-attached consortia, in whichheterogeneous ecological niches develop, and under these conditions competition for the primary carbonsource is not the only determinant of biomass and population structure.

Bacteria often live in consortia bound to surfaces, such as inbiofilms, flocs, or granules (5). Under these conditions thebacteria are positioned in a heterogeneous environment withgradients of nutrients and waste products as a consequence ofdiffusion and mass transport processes, and it is therefore to beexpected that this heterogeneity is reflected in the physiologyof the individual cells. In agreement with this, consortia likebiofilms often appear as rather complex and heterogeneousassemblies consisting of clusters of bacteria embedded in poly-meric substances, which are separated by void regions (cell-free channels) (12, 13, 27, 28). The development of the cell-free regions may support transport of nutrients and wasteproducts to and from the deeper layers of the biofilms (7). Forexample, DeBeer et al. (8) showed by using microelectrodes

that at the same depths in a biofilm, oxygen concentrations inthe void regions were much higher than those in adjacentclusters of biomass.

The development of what seem to be structurally organizedcommunities may argue for the presence of overall regulatoryelements, which control the formation of the community struc-tures (6). However, changes in structural organization havebeen shown to be significantly affected by the nutrients sup-plied to the community (13, 27). Furthermore, mathematicalmodeling of bacterial growth in biofilms has indicated thatsimple rules based on nutrient gradients, diffusion rates, andbiomass production may determine basic features of biofilmstructures (18, 26). Thus, even though there may be regulatoryfactors that are actively involved in control of biofilm forma-tion, parameters like mass transport, substrate concentrations,diffusion gradients, detachment-attachment mechanisms, andflow rates probably all have significant influence on biofilmstructures.

Syntrophic relationships between different organisms have

* Corresponding author. Mailing address: BioCentrum-DTU, Mo-lecular Microbial Ecology Group, Building 301, The Technical Uni-versity of Denmark, DK-2800 Lyngby, Denmark. Phone: 45 45 25 2513. Fax: 45 45 88 73 28. E-mail: imsm@pop.dtu.dk.

2495

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 04

Jan

uary

202

2 by

164

.70.

153.

131.

been demonstrated in several microbial ecosystems, such as theinterspecies electron transfer from H2 or formate in anaerobicdigesters (1, 25) and the relationship between ammonia-oxi-dizing and nitrite-oxidizing species in nitrifying communities(22). Communities involving xenobiotic degradation (for a re-view, see reference 21) and oral communities (2) are otherexamples of tight metabolic associations between communityspecies.

Applications of scanning confocal laser microscopy (SCLM),fluorescence in situ hybridization (FISH), and microelectrodeshave led to a rapidly increasing understanding of structure-function relationships in microbial communities. FISH and theuse of microelectrodes have shown that the nitrite-oxidizingbacteria are clustered around ammonia-oxidizing bacteria innitrite-oxidizing zones (inner part of the biofilm) in a nitrifyingwastewater treatment biofilm (17). In addition, Ramsing et al.(20) demonstrated a negative correlation between sulfate-re-ducing bacteria and the oxygen profile in a photosyntheticbiofilm, and in anaerobic granular sludge digesters the struc-tural relationship between different species was shown to behighly organized (9, 11, 23).

To obtain a better understanding of the function of channelstructures and of structural relationships between species inrelation to the overall functionality of the microbial commu-nities, we also need to be able to determine the physiologicalstate of the individual cells in the microbial consortium. Re-cently, a reporter system based on a fusion between the rRNApromoter from Escherichia coli and a reporter gene encodingan unstable derivative of the green fluorescent protein (GFP)was developed (24). This reporter system has been used tomonitor growth activity in the present study of species rela-tionships in benzyl alcohol-degrading communities comprisingtwo species, Acinetobacter sp. strain C6 and Pseudomonasputida strain R1.

MATERIALS AND METHODS

Strain and growth conditions. Throughout this investigation, derivatives of P.putida R1 and Acinetobacter strain C6 described in Table 1 were used. P. putidaR1 (SM1700) with the mini Tn5-Km-rrnBP1-gfp[AGA]-T0-T1 cassette inserted inthe chromosome encodes an unstable variant of the GFP (AGA) with a slowerdegradation half-life, of approximately 6 to 7 h (19), than those previouslydescribed (24). The use of a slowly degrading GFP protein reduced the ability tomonitor rapid changes in growth activity. However, cells with a much reducedgrowth rate could be detected when this GFP protein variant was used.

A streptomycin-resistant mutant of the natural isolate of Acinetobacter strainC6 was isolated as a spontaneous mutant on Luria-Bertani (LB) broth platescontaining 200 �g of streptomycin per ml.

Strains were grown in FAB medium [1 mM MgCl2, 0.1 mM CaCl2, 0.01 mMFe-EDTA (Sigma Chemical Co., St. Louis, Mo.), 0.15 mM (NH4)SO4, 0.33 mMNa2HPO4, 0.2 mM KH2PO4, and 0.5 mM NaCl] supplied, unless otherwise

mentioned, with benzyl alcohol (Merck, Darmstadt, Germany) as the sole carbonsource. When required, antibiotics were added at final concentrations of 100�g/ml for streptomycin, 50 �g/ml for nalidixic acid, and 10 �g/ml for kanamycin.

Chemostat experiments. The chemostats were made from a 50-ml plasticsyringe (Terumo Europe N.V., Leuven, Belgium) with a rubber stopper contain-ing a glass tube for outlet of the effluent culture and another for intake of air,which was passed through a 0.2-�m-pore-size filter. In addition, a thin hypoder-mic needle was inserted for injection of medium and another was inserted forwithdrawal of samples from the chemostat. All chemostats were coated with adimethyl-dichlorosilane solution (Sigma) to avoid attachment of cells to thechemostat walls. In addition, occurrence of bacterial growth on the chemostatwall was tested after each experiment by scraping of samples from the wallfollowed by plating. In all experiments the total CFU of suspended cells were atleast 100 times higher than the number of surface-attached cells, indicating thata potential bias of the chemostat data by development of biofilms on the che-mostat walls was insignificant in the present experiments. Six chemostats wererun in parallel at the same time. The chemostats were supplied with FABmedium containing 5 mM benzyl alcohol (Merck), and the dilution rate was keptat 6 ml/h with a 205S peristaltic pump (Watson Marlow Inc., Wilmington, Mass.).The chemostats were inoculated with 25 ml of an overnight culture of either P.putida R1 or Acinetobacter strain C6 grown in FAB minimal medium supple-mented with 2 mM benzoate and 5 mM benzyl alcohol. In chemostats where P.putida R1 and Acinetobacter strain C6 were mixed, 25 ml of overnight cultures ofboth strains were mixed before inoculation. The chemostats were sampled at1-day intervals. For each sample the optical density at 450 nm was measured andthe CFU per milliliter for each strain was enumerated by plating on LB brothplates containing antibiotics for selection of P. putida R1 or Acinetobacter strainC6. The chemostats were checked for contaminating organisms by plating on LBbroth plates without addition of antibiotics (data not shown).

Flow chamber experiments. Biofilms were grown at room temperature inthree-channel flow chambers with individual channel dimensions of 1 by 4 by 40mm. The flow system was assembled and prepared as described previously (3).The substratum was a microscope glass coverslip (st1; Knittel Gläser, Braun-schweig, Germany). Each channel was supplied with a flow of 3 ml/h (flow rateof 0.2 mm/s) of FAB medium containing 0.5 mM benzyl alcohol. Flow cells wereinoculated with mixtures of overnight cultures of P. putida R1 and Acinetobacterstrain C6 grown in LB medium. P. putida R1 was diluted 20 times, and Acineto-bacter strain C6 was diluted 4 times, in 0.9% NaCl. After the medium flow wasstopped, the flow channels were turned upside down and 250 �l of the dilutedmixture was carefully injected into each flow channel with a small syringe. After1 h, the flow channels were turned around and the flow was resumed using a 205Speristaltic pump (Watson Marlow). Enumeration of detaching biofilm cells wasperformed by collecting 0.5 to 1 ml of cells from the flow chamber effluent in anEppendorf tube kept on ice. The cells were vortexed for at least 20 s, which wasshown by microscopic inspection to be enough to ensure dispersion of cellclumps. The effluent cells were enumerated for each strain by plating on LBbroth plates containing antibiotics for selection of P. putida R1 or Acinetobacterstrain C6. The flow chambers were checked for contaminating organisms byplating on LB broth plates without antibiotics (data not shown).

Oligonucleotide probes. For in situ 16S rRNA hybridizations two differentoligonucleotide probes were used. A probe specific for P. putida subgroup A,PP986 (14), was labeled with the indocarbocyanine dye CY5, and the probespecific for Acinetobacter sp. strain C6, ACN449 (15), was labeled with theindocarbocyanine dye CY3. Hybridization was performed in 30% formamide at37°C, as this stringency proved to be sufficient for distinguishing the species in themodel consortium. The probes were purchased from Hobolth DNA Syntese(Hillerød, Denmark).

TABLE 1. Strains

Strain Relevant characteristics Source or reference

P. putida R1 (SM1700) P. putida R1 (Nalr)a � HB101 (RK600) � CC118�pir(pSM1695); P. putida R1 with mini Tn5-Km-rrnBP1-gfp[AGA]-T0-T1 cassette inserted into the chromosome; Nalr

Kmr; � subclass of class Proteobacteriab

19

Acinetobacter sp. strain C6 (CKL01) Natural isolate made Strr; � subclass of class Proteobacteriac This work

a P. putida R1 (JB156) from reference 3.b From reference 14.c From reference 15.

2496 CHRISTENSEN ET AL. APPL. ENVIRON. MICROBIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 04

Jan

uary

202

2 by

164

.70.

153.

131.

Embedding and 16S rRNA hybridization of hydrated biofilm samples. Inorder to avoid further degradation of the unstable GFP, the embedding andhybridization procedure, as previously described (3, 4, 15), was slightly modified.The biofilms were fixed by carefully inoculating 500 �l of ice-cold 4% parafor-maldehyde solution directly into the flow channel, which was then kept on ice forat least 1 h to ensure complete fixation of all cells in the biofilm. The fixedbiofilms were washed with 1� phosphate-buffered saline by pumping the solutionthrough the channels for 20 min (flow rate, 0.25 mm s�1), and finally the biofilmwas embedded in 1 ml of 20% acrylamide solution containing 200:1 acrylamide-bisacrylamide (Sequagel; National Diagnostics, Atlanta, Ga.), 8 �l of N,N,N�,N�-tetramethylethylendiamine (Kodak International Biotechnologies Inc., New Ha-ven, Conn.), and 20 �l of 1% ammonium persulfate (InternationalBiotechnologies Inc.). After the polyacrylamide was allowed to solidify for atleast 1 h, the glass coverslip was carefully loosened from the flow cell, and thebiofilm containing the polyacrylamide block was lifted out of the flow channel.The block was subsequently cut into slices of approximately 5 mm, placed on asix-well hybridization slide (Novakemi ab, Enskede, Sweden), and prehybridizedat 37°C in 45 �l of the hybridization buffer (washing solution I [0.9 M NaCl, 100mM Tris, pH 7.2] containing 30% formamide at 37°C). After 30 min, the pre-hybridization buffer was removed and 30 �l of the hybridization buffer containing75 ng of each probe was added to the hybridization well. The slide was incubatedat 37°C for at least 3 h in a moisturized chamber. The polyacrylamide blocks werewashed in 45 �l of washing solution I for 30 min at 37°C, followed by washing in45 �l of washing solution II (0.9 M NaCl, 100 mM Tris, pH 7.2) for another 30min at 37°C. Finally, the acrylamide blocks were rinsed in 45 �l of Milli-Q waterand then immediately mounted on an object glass with a drop of SlowFadephosphate-buffered saline-based antifade solution (Molecular Probes) and acoverslip on top.

Microscopy and image analysis. All microscopic observations and image ac-quisitions were performed with a TCD4D SCLM (Leica Lasertechnik GmbH,Heidelberg, Germany) equipped with an argon-krypton laser and three detectorsand filter sets for simultaneous monitoring of fluorescein isothiocyanate-GFPand the indocarbocyanine dyes CY3 and CY5.

The x-y images were presented as extended-focus images, which are producedby taking the confocal images from the different depths of the biofilm andprojecting them into a single image. The extended-focus images and verticalcross sections through the biofilm were generated by using the IMARIS softwarepackage (Bitplane AG, Zurich, Switzerland) running on an Indigo2 workstation(Silicon Graphics, Mountain View, Calif.). Images were further processed fordisplay by using Photoshop software (Adobe, Mountain View, Calif.)

HPLC analysis. Samples subjected to high-performance liquid chromatogra-phy (HPLC) analysis were taken from the effluent of the flow channels. Thesamples were filtered through a 0.2-�m-pore-size filter, and the contents ofbenzyl alcohol and benzoate were measured with a Shimadzu (Tokyo, Japan)HPLC equipped with a Supelcosil C18 reverse-phase column (Supelco Park,Bellefonte, Pa.) and a UV-visible detector set at 206 nm. The mobile phase wasa solution of 40% acetonitrile (HPLC grade; Sigma Aldrich) and 60% NaH2PO4

(50 mM, pH 3,0) supplied at a flow rate of 1 ml min�1.

Analysis of biofilm thickness. The thicknesses of the biofilms were measuredusing a specific function on the digitally controlled microscope (DMRXA mi-croscope; Leica Mikroskopie und Systeme GmbH, Wetzlar, Germany) whichmakes it possible to measure the distance between two focused planes.

RESULTS

Quantitative analysis of mixed-species and monospeciesconsortia of P. putida R1 and Acinetobacter strain C6 in che-mostats and in flow chambers. P. putida R1 and Acinetobacterstrain C6 were grown as suspended cultures in chemostats, inwhich heterogeneous structural relationships between the or-ganisms are not established. Acinetobacter strain C6 and P.putida R1 were established either as monocultures (Fig. 1A) ortogether in a mixed culture (Fig. 1B). As shown in Fig. 1,constant cell densities were reached in the monoculture andmixed-culture chemostats after approximately 3 and 5 days,respectively. With respect to Acinetobacter strain C6, the celldensity remained constant in both chemostats over the courseof the next 5 days until the experiment was terminated,whereas a slight increase in P. putida R1 cell numbers wasobserved during the last 2 days. In both the mixed and mono-species cultures Acinetobacter strain C6 reached the same celldensity of approximately 109 CFU/ml. In contrast, the celldensities of P. putida R1 were lower than the correspondinglevels of Acinetobacter strain C6 in both the mixed-species andthe monospecies chemostats. When growing alone in the che-mostat, the cell density of P. putida R1 was between 10 and 100times lower than that of Acinetobacter strain C6, whereas in thechemostats in which Acinetobacter strain C6 and P. putida R1were mixed, the density of P. putida R1 was reduced to a 500-to 1,000-fold-lower level relative to that of Acinetobacter strainC6. These results show that Acinetobacter strain C6 producesmore biomass than P. putida R1.

To investigate how the observed differences in the metabolicefficiency affected the composition in surface-associated cul-tures over time, nondestructive in situ methods were em-ployed. Two parameters were measured: (i) biofilm thicknessand (ii) CFU of cells collected from the flow chamber effluentsper milliliter. Although the ratios between biomass, voids, and

FIG. 1. Time course analysis of the numbers of P. putida R1 (■ ) and Acinetobacter strain C6 (F) cells collected from chemostats where thestrains were established either as monospecies cultures (A) or as mixed cultures (B). CFU were enumerated on LB plates containing theappropriate antibiotics for selection of P. putida R1 and Acinetobacter strain C6, respectively. Error bars indicate standard deviations.

VOL. 68, 2002 METABOLIC INTERACTIONS IN MICROBIAL CONSORTIA 2497

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 04

Jan

uary

202

2 by

164

.70.

153.

131.

channels may vary in internal parts of the biofilm (which wasnot taken into account when measuring biofilm thickness), thethickness of the biofilm may, as shown for other biofilm con-sortia (10), be used as a relative measure of the total biomassaccumulated in the flow channels. As shown in Fig. 2, bothAcinetobacter strain C6 and P. putida R1 monocultures reacheda thickness of about 7 to 10 �m, whereas the binary biofilmreached a thickness of approximately 12 to 14 �m.

The cells collected from the effluent are those cells thatdetach from the flow chamber biofilm. Only in cases where thebiofilm cell density has reached a steady state for each specieswill the effluent cells become an exact representation of thebiofilm population. On the other hand, previous studies haveshown (3) that in control experiments in which Acinetobacterstrain C6 and P. putida R1 were the only species or the pre-dominant species, the effluent data do represent reliable esti-

mates of the entire biofilm population. Determinations of thebiofilm effluent cell numbers showed that a ratio of approxi-mately 1:5 of Acinetobacter strain C6 to P. putida R1 in themixed biofilm was reached after a few days (Fig. 3B). Varyingthe ratio of Acinetobacter strain C6 and P. putida R1 in theinoculum did not change this ratio significantly (data notshown). After 2 to 3 days the effluent cell numbers of Acineto-bacter strain C6 in the monoculture (Fig. 3A) and mixed-culture (Fig. 3B) biofilms had reached a level which remainednearly constant throughout the rest of the experiment. In themonoculture biofilm of P. putida R1, the cell number increasedat a lower rate but reached higher numbers than Acinetobacterstrain C6 (Fig. 3A), whereas a constant high level of P. putidaR1 in the mixed biofilm was reached within a few days (Fig.3B). Thus, the presence of Acinetobacter strain C6 apparentlyresults in a faster establishment of P. putida R1 in the mixedconsortium.

It was previously observed that in mixed-species biofilmsgrowing with benzyl alcohol as the sole carbon and energysource, the expression of the benzoate-inducible promoter,Pm, from the TOL pathway inserted into P. putida R1 wasinduced in regions near microcolonies of Acinetobacter strainC6 (15), and it was speculated that benzoate leaking fromAcinetobacter strain C6 caused the induction of Pm in P. putidaR1. In order to investigate this explanation, we performedHPLC analysis of benzyl alcohol and benzoate in flow channeleffluents, and at the same time this analysis was used to furtherassess the metabolic efficiencies of the two organisms.

Samples were taken at different time points after initialcolonization of mono- and mixed-culture flow chamber bio-films of P. putida R1 and Acinetobacter strain C6. In the P.putida R1 monoculture biofilm (Fig. 4A), the amount of de-graded benzyl alcohol increased slowly over the entire sam-pling period, reaching 75% conversion after 10 days. In theAcinetobacter strain C6 monoculture biofilm, more than 90%of the benzyl alcohol was catabolized 4 days after the initialcolonization, but in contrast to the case for the P. putida R1biofilm, a large amount of benzoate was accumulated (Fig.4B). Thus, Acinetobacter strain C6 does in fact leak benzoatefrom cells when grown as a biofilm. The effluent from a mixed

FIG. 2. Time course analysis of biofilm thickness when P. putida R1(■ ) and Acinetobacter strain C6 (F) were grown as monoculture bio-films or when the two strains were grown as mixed biofilms (Œ). Eachpoint represents the mean of the mean thickness in at least threeindependent flow channels run in parallel. Error bars indicate standarddeviations. The mean thickness in each flow channel was taken as themean from four random microscope viewing fields in which the thick-ness was measured at five different positions (a total of 20 positions ineach flow channel).

FIG. 3. Time course analysis of the numbers of P. putida R1 (■ ) and Acinetobacter strain C6 (F) cells collected from flow channel effluentswhere the strains were established either as monoculture biofilms (A) or as mixed biofilms (B). CFU were enumerated on LB plates containingthe appropriate antibiotics for selection of P. putida R1 and Acinetobacter strain C6. Error bars indicate standard deviations.

2498 CHRISTENSEN ET AL. APPL. ENVIRON. MICROBIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 04

Jan

uary

202

2 by

164

.70.

153.

131.

culture of the two species showed removal of benzyl alcohol aswell as efficient degradation of benzoate (Fig. 4).

In situ analysis of the mixed biofilm consortium. To furtheranalyze the metabolic interactions between Acinetobacterstrain C6 and P. putida R1 cells and their consequences for thestructural development of the consortium, mixed-species bio-films were fixed and embedded 1, 2, 3, and 6 days after theinitial colonization. For specific identification of the two or-ganisms, we used FISH with the fluorescence-labeled probesPP986 and ACN449, targeting P. putida R1 and Acinetobacterstrain C6, respectively. By employing SCLM in combinationwith FISH, the exact positions of Acinetobacter strain C6 and P.putida R1 cells could be visualized. The changes and distribu-tions of microbial growth activity of P. putida R1 cells wereanalyzed by insertion of a transposon with the growth phase-regulated promoter rrnBP1 fused to a gfp gene encoding anunstable variant of GFP. Applications of this type of monitorsystem inserted into P. putida R1 for determinations of growthactivity in biofilms have previously been demonstrated (24).This reporter system can be used to distinguish between fast-growing (bright green) and slow-growing or nongrowing (weakor no signal) cells.

The images presented in Fig. 5 represent examples of themost frequently occurring structural relationships between thetwo species observed at days 1, 2, 3, and 6 after the initialcolonization. In order to ensure that the changing structuresobserved were a consequence of a systematic process and notjust some random process occurring in some cases and not inothers, the experiments was run in five independent roundseach time with three independent flow channels running inparallel. In all experiments the same developmental patternwas observed, indicating that the dynamic changes in the struc-tural relationships between the two species are caused by non-random processes.

One day after inoculation of the strains, the P. putida R1cells with the highest apparent growth activity (highest GFPsignal) were observed in the regions near small microcoloniesof Acinetobacter strain C6 cells (Fig. 5A). As the Acinetobacterstrain C6 microcolonies grew in size, the P. putida R1 cells withhighest growth activity were located in the center of the colo-

nies (Fig. 5B). At day 3, two distinct types of associations haddeveloped. In one of these, P. putida R1 was excluded from thecenter of the rather large Acinetobacter strain C6 colonies (Fig.5C), and the cells with the highest growth activity were nowobserved at the periphery of the colonies. After 6 days, manyof these Acinetobacter strain C6 microcolonies were overgrownby P. putida R1 cells and the fluorescence intensity of the P.putida R1 cells near the Acinetobacter strain C6 colonies wasreduced to low levels (Fig. 5D). In the other type of associa-tion, small Acinetobacter strain C6 microcolonies started tobuild up in the upper layers of the P. putida R1 biofilm (Fig.5E). The result was production of large structures (at least 100to 200 �m thick) in which Acinetobacter strain C6 microcolo-nies became integrated in the P. putida R1 cell structures (Fig.5F).

DISCUSSION

Competition for substrate is considered to be one of themajor evolutionary driving forces in the bacterial world, andnumerous experimental data obtained in the laboratory underwell-controlled conditions show how different organisms, orvariants of one organism, may effectively outcompete othersbecause of better utilization of a given energy source. Suchcompetition experiments may be performed very convincinglyin chemostats, where nutrient limitation allows for metaboliccompetition among the cells present in the reactor. One im-portant factor for optimal performance of a chemostat is thehomogeneous distribution of the suspended cells, and it isconsidered important to prevent surface attachment of thecells to the reactor walls. Under such optimal conditions, inwhich only one nutrient is limiting for growth, it may be ex-pected that an organism will totally outcompete all others if ithas an improved efficiency of substrate utilization relative tothe others. The concentration of the particular nutrient willalways be below the threshold level for the less efficient cells,and the rate of washout is therefore higher than the rate of cellproliferation for these cells.

In the present study the two organisms were grown as mono-species and as mixed-species cultures with benzyl alcohol as the

FIG. 4. HPLC analysis of the flow channel effluents. The contents of benzyl alcohol (A) and benzoate (B) were measured for monoculturebiofilms of P. putida R1 (■ ) and Acinetobacter strain C6 (F), and for the binary Acinetobacter strain C6-P. putida R1 biofilm (Œ). In all flow channelbiofilms the inlet concentration of benzyl alcohol was 0.5 mM. Error bars indicate standard deviations.

VOL. 68, 2002 METABOLIC INTERACTIONS IN MICROBIAL CONSORTIA 2499

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 04

Jan

uary

202

2 by

164

.70.

153.

131.

FIG. 5. SCLM micrographs showing the structural relationships between Acinetobacter strain C6 and P. putida R1 cells with low and highactivity, respectively, in a mixed biofilm consortium which was supplied with 0.5 mM benzyl alcohol as the sole carbon source. At days 1 (A), 2 (B),3 (C and E), and 6 (D and F) after inoculation, biofilms were embedded and hybridized. P. putida R1 and Acinetobacter strain C6 were hybridizedwith PP986 labeled with CY5 (blue) and ACN449 labeled with CY3 (red), respectively. The active P. putida R1 cells were monitored as cellsemitting green fluorescence due to the rrnBP1-gfp[AGA] fusion inserted in the chromosome of P. putida R1. These cells appear as cyan due to

2500 CHRISTENSEN ET AL. APPL. ENVIRON. MICROBIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 04

Jan

uary

202

2 by

164

.70.

153.

131.

sole carbon source, both as surface-attached organisms (bio-films) in flow chambers and as suspended cultures in chemo-stats. In the chemostats, differences in the respective cell num-bers at steady state showed that Acinetobacter strain C6produced the highest yield on the limiting concentration ofbenzyl alcohol. When Acinetobacter strain C6 and P. putida R1were mixed in the chemostat, both organisms rapidly reachedrelatively stable cell densities in which the cell number ratiowas approximately 1 to 500 in favor of Acinetobacter strain C6.This again showed the competitive advantage of Acinetobacterstrain C6, but it also showed an unexpected stable presence ofP. putida R1 at a low but significant level. In the flow chamber-based biofilms, effluent cell counts showed that within a fewdays the two organisms reached nearly stable cell densities, inwhich P. putida R1 was present in higher numbers than Acin-etobacter strain C6 (approximately 5 to 1). Furthermore, mea-surements of the biofilm thickness suggested that the simulta-neous presence of both strains resulted in a significant increaseof the overall biomass in the biofilm. Thus, despite the betterutilization of benzyl alcohol by Acinetobacter strain C6, P.putida R1 was maintained in both growth systems. This findingmay be explained by the previous suggestion that growing withbenzyl alcohol as the carbon source makes Acinetobacter strainC6 cells excrete benzoate, a carbon source readily utilized by P.putida R1, into the reactor environment (15). This hypothesishas been confirmed in the present study. The HPLC analysis ofthe effluent from flow chambers shows that benzoate accumu-lates in Acinetobacter strain C6 monospecies biofilms, whereasin the presence of P. putida R1 almost all the excreted benzo-ate is degraded. In the P. putida R1 monospecies biofilm,degradation of benzyl alcohol was rather inefficient, and only alow concentration of benzoate was detected in the effluent.This suggests that Acinetobacter strain C6 relatively easily con-verts benzyl alcohol to benzoate, whereas the degradation ofbenzoate is slow, resulting in a metabolic bottleneck that leadsto accumulation of benzoate. In contrast, P. putida R1 encodesa better pathway for degradation of benzoate, which has thecapacity to mineralize both the benzoate derived from theconversion of benzyl alcohol to benzoate via its own degrada-tion pathway and the external supply from Acinetobacter strainC6.

To further explain the differences in ratios between P. putidaR1 and Acinetobacter strain C6 in the flow chambers and che-mostats, a detailed analysis of the distribution and activity ofthe P. putida R1 cells in the flow chamber biofilms was carriedout. In situ rRNA hybridization for identification of the indi-vidual organisms was used to analyze the spatial distribution ofthe two strains in the flow chamber biofilms. After 2 to 3 daysof growth, large numbers of P. putida R1 cells were found tocluster around large surface-associated microcolonies of Acin-

etobacter strain C6. By employing the rrnBP1::gfp[AGA] mon-itor cassette inserted into P. putida R1, it was shown that the P.putida R1 cells in these regions had a higher growth activitythan those further away from the Acinetobacter strain C6 mi-crocolonies. Similar distribution patterns have also been ob-served in nitrifying biofilms (17) and in a two-species biofilmgrowing with chlorobiphenyl as the sole carbon and energysource (16). However, in neither of these studies was it possiblefor the individual strains to be established as monospeciesbiofilms under the conditions described, and thus, in contrastto the present study, there was no competition for the primarycarbon source in the mixed biofilms.

For the Acinetobacter strain C6 microcolonies that were ini-tially established at the flow chamber glass surface, a stablestructural relationship between P. putida R1 and Acinetobacterstrain C6 cells was not maintained, because P. putida R1started to overgrow the Acinetobacter strain C6 microcolonies(Fig. 5C and D). The result was a significant reduction in thegrowth activity of P. putida R1 (reduced GFP signal) in theseregions. We suggest that the establishment of P. putida R1 cellson the outside of the surface-associated microcolonies reducedthe supply of benzyl alcohol to the Acinetobacter strain C6 cellsand thereby also the production of benzoate from these cells.Instead, small microcolonies of Acinetobacter strain C6 cellsbecame established in the upper layer of the P. putida R1biofilm where they were near the medium flow and therebyexposed to a constant supply of benzyl alcohol. This allowedthe Acinetobacter strain C6 cells to accumulate benzoate inthese regions, which was metabolized by the associated P.putida R1 cells, which obtained a growth advantage over other,nonassociated P. putida R1 cells (Fig. 5E). We suggest that dueto the metabolic interactions, these mixed Acinetobacter strainC6-P. putida R1 structures grew faster than the rest of thebiofilm and resulted in the production of large structures ofbiofilm. Thus, taking all these observations together, the twoorganisms in the biofilm competed as well as exhibited com-mensal interactions, depending on the physical positioning ofthe organism. Similar interacting mechanisms, i.e., differentspecies being able to degrade the primary substrate but havingdifferent capacities for degradation of the metabolic interme-diates, may be expected to prevail in natural environments.This may explain why some natural communities grown on asingle carbon source, such as in the studies presented by Wol-faardt et al. (27) and Møller et al. (13), often develop quitecomplex biofilm compositions and organizations.

The present investigation of the structure-function relation-ships in a binary biofilm growing with benzyl alcohol as the onlyadded carbon and energy source thus offers an explanation ofthe contradictory population data from suspended cultures inchemostats and in flow chamber biofilms, respectively. In the

the combination of green (GFP) and blue (hybridization). For each panel similar images were collected from at least five independent biofilmexperiments. Panels A and B are representative of the biofilm structures observed at day 1 and 2. Panels C and D are examples of the largeAcinetobacter strain C6 microcolonies that developed after 3 days (C) and which later were overgrown by P. putida R1 (D). Panels E and F areexamples of Acinetobacter strain C6 microcolonies (arrow) that were established in the upper part of P. putida R1 cell clusters after 2 to 3 days(E), which resulted in production of large macrostructures of associated P. putida R1 and Acinetobacter strain C6 cells (F). All x-y plots arepresented as extended-focus images. Shown to the right and above the x-y plots are vertical sections through the biofilm collected at the positionsindicated by the white triangles. The arrow indicates the direction of flow. Bars, 20 �m. Note that the green fluorescent intensities on the imageshave been amplified to give the best differentiation between P. putida R1 cells with low and high activity in each image. Thus, the intensity levelsin the different images cannot be directly compared.

VOL. 68, 2002 METABOLIC INTERACTIONS IN MICROBIAL CONSORTIA 2501

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 04

Jan

uary

202

2 by

164

.70.

153.

131.

chemostat the individual cells are all surrounded by the sameenvironmental conditions, and they cannot stay in the reactorif their growth rate is lower than the dilution rate (no adher-ence). In the biofilm there is a range of conditions surroundingthe cells due to the heterogeneity of the consortium structure;i.e., microniches develop in which the supplies of primary andsecondary nutrients differ significantly. There will therefore belocations in the biofilm where the conditions favor one or theother or both of the organisms, depending on the local struc-ture. In addition, the cells may adhere to the surface or to eachother, thus reducing washout. One important consequence ofthe biofilm configuration, therefore, is that substrates may bemore optimally utilized by the consortium (increased mineral-ization of metabolic intermediates), resulting in faster degra-dation of the primary nutrient and a faster buildup of biomass.The reorganization of mixed-species consortia in response tothe nutrient conditions in order to achieve optimal conditionsfor the present organisms may require active motility coupledwith chemotaxis, cell-to-cell communication signals for coor-dinated organizational development, or a liquid flow movingcells around for detachment and reattachment. In the presentconsortium neither chemotaxis nor cell-to-cell communicationis the obvious mechanism behind the presented developmentof the consortium, because our unpublished investigationshave not so far shown evidence of any of these properties in P.putida R1. It is therefore concluded that passive transport ofcells by the flow through the biofilm may be solely responsiblefor the continuous structural development taking place in theconsortium.

ACKNOWLEDGMENTS

This work was supported by grants from the Danish BiotechnologyProgram.

We thank Anne Nielsen and Tove Johansen for expert technicalassistance.

REFERENCES

1. Boone, D. R., R. L. Johnson, and Y. Liu. 1989. Diffusion of the interspecieselectron carriers H2 and formate in methanogenic ecosystems and its impli-cations in the measurement of Km for H2 or formate uptake. Appl. Environ.Microbiol. 55:1735–1741.

2. Bradshaw, D. J., K. A. Homer, P. D. Marsh, and D. Beighton. 1994. Meta-bolic cooperation in oral microbial communities during growth on mucin.Microbiology 140:3407–3412.

3. Christensen, B., C. Sternberg, J. B. Andersen, L. Eberl, S. Møller, M.Givskov, and S. Molin. 1998. Establishment of new genetic traits in a mi-crobial biofilm community. Appl. Environ. Microbiol. 64:2247–2255.

4. Christensen, B. B., C. Sternberg, J. B. Andersen, R. J. Palmer, A. T. Nielsen,M. Givskov, and S. Molin. 1999. Molecular tools for study of biofilm phys-iology. Methods Enzymol. 310:20–42.

5. Costerton, J. W., Z. Lewandowski, D. E. Caldwell, D. R. Korber, and H. M.Lappin-Scott. 1995. Microbial biofilms. Annu. Rev. Microbiol. 49:711–745.

6. Davies, D. G., M. R. Parsek, J. P. Pearson, B. H. Iglewski, J. W. Costerton,and E. P. Greenberg. 1998. The involvement of cell-to-cell signals in thedevelopment of a bacterial biofilm. Science 280:295–298.

7. DeBeer, D., P. Stoodley, and Z. Lewandowski. 1994. Liquid flow in hetero-geneous biofilms. Biotechnol. Bioeng. 44:636–641.

8. DeBeer, D., P. Stoodley, F. Roe, and Z. Lewandowski. 1994. Effects of biofilmstructures on oxygen distribution and mass-transport. Biotechnol. Bioeng.43:1131–1138.

9. Harmsen, H. J., H. M. Kengen, A. D. Akkermans, A. J. Stams, and W. M. deVos. 1996. Detection and localization of syntrophic propionate-oxidizingbacteria in granular sludge by in situ hybridization using 16S rRNA-basedoligonucleotide probes. Appl. Environ. Microbiol. 62:1656–1663.

10. Heydorn, A., A. T. Nielsen, M. Hentzer, C. Sternberg, M. Givskov, B. K.Ersbøll, and S. Molin. 2000. Quantification of biofilm structures by the novelcomputer program COMSTAT. Microbiology 146:2395–2407.

11. MacLeod, F. A., S. R. Guiot, and J. W. Costerton. 1990. Layered structure ofbacterial aggregates produced in an upflow anaerobic sludge bed and filterreactor. Appl. Environ. Microbiol. 56:1598–1607.

12. Massol-Deya, A. A., J. Whallon, R. F. Hickey, and J. M. Tiedje. 1995.Channel structures in aerobic biofilms of fixed-film reactors treating con-taminated groundwater. Appl. Environ. Microbiol. 61:769–777.

13. Møller, S., D. R. Korber, G. M. Wolfaardt, S. Molin, and D. E. Caldwell.1997. Impact of nutrient composition on a degradative biofilm community.Appl. Environ. Microbiol. 63:2432–2438.

14. Møller, S., A. R. Pedersen, L. K. Poulsen, E. Arvin, and S. Molin. 1996.Activity and three-dimensional distribution of toluene-degrading Pseudomo-nas putida in a multispecies biofilm assessed by quantitative in situ hybrid-ization and scanning confocal laser microscopy. Appl. Environ. Microbiol.62:4632–4640.

15. Møller, S., C. Sternberg, J. B. Andersen, B. B. Christensen, J. L. Ramos, M.Givskov, and S. Molin. 1998. In situ gene expression in mixed-culture bio-films: evidence of metabolic interactions between community members.Appl. Environ. Microbiol. 64:721–732.

16. Nielsen, A. T., T. Tolker-Nielsen, K. B. Barken, and S. Molin. 2000. Role ofcommensal relationships on the spatial structure of a surface-attached mi-crobial consortium. Environ. Microbiol. 2:59–68.

17. Okabe, S., H. Satoh, and Y. Watanabe. 1999. In situ analysis of nitrifyingbiofilms as determined by in situ hybridization and the use of microelec-trodes. Appl. Environ. Microbiol. 65:3182–3191.

18. Picioreanu, C., M. C. M. Van Loosdrecht, and J. J. Heijnen. 1998. Mathe-matical modeling of biofilm structure with a hybrid differential-discrete cel-lular automaton approach. Biotechnol. Bioeng. 58:101–116.

19. Ramos, C., L. Molbak, and S. Molin. 2000. Bacterial activity in the rhizo-sphere analyzed at the single-cell level by monitoring ribosome contents andsynthesis rates. Appl. Environ. Microbiol. 66:801–809.

20. Ramsing, N. B., M. Kuhl, and B. B. Jørgensen. 1993. Distribution of sulfate-reducing bacteria, O2, and H2S in photosynthetic biofilms determined byoligonucleotide probes and microelectrodes. Appl. Environ. Microbiol. 59:3840–3849.

21. Reanney, D., P. C. Gowland, and J. H. Slater. 1983. Genetic interactionsamong microbial communities. Symp. Soc. Gen. Microbiol. 34:379-422.

22. Schramm, A., L. H. Larsen, N. P. Revsbech, N. B. Ramsing, R. Amann, andK. H. Schleifer. 1996. Structure and function of a nitrifying biofilm as de-termined by in situ hybridization and the use of microelectrodes. Appl.Environ. Microbiol. 62:4641–4647.

23. Sekiguchi, Y., Y. Kamagata, K. Nakamura, A. Ohashi, and H. Harada. 1999.Fluorescence in situ hybridization using 16S rRNA-targeted oligonucleotidesreveals localization of methanogens and selected uncultured bacteria inmesophilic and thermophilic sludge granules. Appl. Environ. Microbiol. 65:1280–1288.

24. Sternberg, C., B. B. Christensen, T. Johansen, A. T. Nielsen, J. B. Andersen,M. Givskov, and S. Molin. 1999. Distribution of bacterial growth activity inflow-chamber biofilms. Appl. Environ. Microbiol. 65:4108–4117.

25. Thiele, J. H., M. Chartrain, and J. G. Zeikus. 1988. Control of interspecieselectron flow during anaerobic digestion: role of floc formation in syntrophicmethanogenesis. Appl. Environ. Microbiol. 54:10–19.

26. Wimpenny, J. W. T., and R. Colasanti. 1997. A unifying hypothesis for thestructure of microbial biofilms based on cellular automaton models. FEMSMicrobiol. Ecol. 22:1–16.

27. Wolfaardt, G. M., J. R. Lawrence, R. D. Robarts, S. J. Caldwell, and D. E.Caldwell. 1994. Multicellular organization in a degradative biofilm commu-nity. Appl. Environ. Microbiol. 60:434–446.

28. Zahid, W. M., and J. J. Ganzarczyk. 1994. Fractal properties of the RBCbiofilm structure. Wat. Sci. Technol. 29:271–279.

2502 CHRISTENSEN ET AL. APPL. ENVIRON. MICROBIOL.

Dow

nloa

ded

from

http

s://j

ourn

als.

asm

.org

/jour

nal/a

em o

n 04

Jan

uary

202

2 by

164

.70.

153.

131.