OUP user menu

Dynamics of development and dispersal in sessile microbial communities: examples from Pseudomonas aeruginosa and Pseudomonas putida model biofilms

Mikkel Klausen, Morten Gjermansen, Jan-Ulrich Kreft, Tim Tolker-Nielsen
DOI: http://dx.doi.org/10.1111/j.1574-6968.2006.00280.x 1-11 First published online: 1 August 2006


Surface-associated microbial communities in many cases display dynamic developmental patterns. Model biofilms formed by Pseudomonas aeruginosa and Pseudomonas putida in laboratory flow-chamber setups represent examples of such behaviour. Dependent on the experimental conditions the bacteria in these model biofilms develop characteristic multicellular structures through a series of distinct steps where cellular migration plays an important role. Despite the appearance of these characteristic developmental patterns in the model biofilms the available evidence suggest that the biofilm forming organisms do not possess comprehensive genetic programs for biofilm development. Instead the bacteria appear to have evolved a number of different mechanisms to optimize surface colonization, of which they express a subset in response to the prevailing environmental conditions. These mechanisms include the ability to regulate cellular adhesiveness and migration in response to micro-environmental signals including those secreted by the bacteria themselves.

  • biofilm
  • development
  • dynamics
  • evolution
  • Pseudomonas aeruginosa
  • Pseudomonas putida


Bacteria in nature often exist in surface-associated sessile communities referred to as biofilms (Costertonet al,1995). Microbial biofilms have been subject to intense study during the last decade mainly for two reasons. First, it is of basic scientific interest to understand how bacteria form and live in multicellular communities. Second, because bacteria in biofilms in many cases are difficult to eliminate via biocide or antibiotic treatment or host immune responses, biofilm formation causes considerable problems in industrial and medical settings (Costertonet al,1999), and knowledge about the environmental cues, genetic elements, and molecular mechanism which are involved when bacteria form biofilms is the basis for rational design of strategies to control biofilm development.

A substantial part of the studies of microbial biofilms carried out during the last decade has involved laboratory setups such as microtitre trays and flow chambers. The model biofilms grown in these setups most likely differ a lot from the biofilms which are formed in industrial, clinical, and environmental settings. Nevertheless experiments with model biofilms have identified and provided information about a number of factors which are important during biofilm development and should be relevant beyond the model systems. These factors include bacterial attachment mechanisms, surface-associated spreading mechanisms, cell-to-cell interconnecting components, dispersion mechanisms, and the environmental cues and underlying genetic elements which are involved. In the present review we present examples of work carried out on these topics with model biofilms formed by Pseudomonas putida and Pseudomonas aeruginosa. These two pseudomonads within the controlled conditions of laboratory setups represent different pathways of biofilm development. Our current knowledge is presented in accordance with the biofilm developmental cycle, i.e. attachment, microcolony formation, maturation, and dispersion. Finally, we discuss whether the presented types of biofilm development should be regarded as a form of microbial development, and we touch upon issues of sociality and cooperation in biofilms.

Development and dynamics in P. aeruginosa biofilms

Studies of P. aeruginosa biofilm development and dynamics have provided illustrative examples of the diversity of ways surface-associated microbial multicellular structures can develop. Pseudomonas aeruginosa is capable of flagellum-driven swimming motility and type IV pili-mediated surface-associated twitching motility. Both swimming and twitching motility can contribute to the dynamic structural changes that occur during P. aeruginosa biofilm development.

Transport of P. aeruginosa cells to a surface before attachment is a dynamic equilibrium between diffusive, convective and active flagellum-driven transport (van Loosdrechtet al,1990). Mutations in genes encoding flagella (O'Toole & Kolter, 1998; Saueret al,2002), type IV pili (Dezielet al,2001; Chiang & Burrows, 2003) and cup fimbria (Valletet al,2001) were shown to affect the ability of P. aeruginosa to attach to surfaces in microtiter tray and flow-chamber setups. However, growth conditions under which lack of flagella and/or type IV pili did not affect surface attachment in a flow chamber have been reported as well (Klausenet al,2003b). After P. aeruginosa has attached to a surface it either detaches, stay attached at the position of attachment, or move along the surface by twitching motility (Singhet al,2002; Klausenet al,2003a,b). The gene sadB (PA5346) has been shown to be involved in regulation of the attachment–detachment frequency in a flow-chamber setup (Caiazza & O'Toole, 2004).

Studies of microcolony formation of twitching and nontwitching cells by time-lapse confocal laser scanning microscopy (CLSM) have shown that cells attach to a surface and form microcolonies by keeping a balance between cell division, surface migration and detachment (Singhet al,2002; Klausenet al,2003a,b). Experiments with mixtures of cyan fluorescent protein-tagged and yellow fluorescent protein-tagged cells provided evidence that microcolonies developed by clonal growth in a flow-chamber system (Fig. 1) (Klausenet al,2003a). On the contrary, microcolony formation by motility driven cell-to-cell aggregation has been observed in a microtitre tray setup (O'Toole & Kolter, 1998). Conditions that promote extensive twitching motility of P. aeruginosa have been shown to prevent microcolony formation in a flow-chamber setup. Interestingly, this could be a mechanism which prevents biofilm formation in humans as lactoferrin, a component of the innate immune system, induce extensive twitching motility in P. aeruginosa by chelating iron (Singhet al,2002).

Figure 1

Time-lapse CLSM in a colour-coded Pseudomonas aeruginosa PAO1 wild-type biofilm. The biofilm was initiated with a 1:1 mixture of yellow fluorescent protein-tagged and cyan fluorescent protein-tagged P. aeruginosa wild-type bacteria, and was grown in a flow-chamber irrigated with citrate minimal medium. The structural development in the biofilm was followed by time-lapse CLSM. The shown CLSM side view projections were acquired after 3, 6, 8, 10, 13, 15, 18, 20 and 23h of biofilm development. The side of the boxes corresponds to 230μm and the height of the boxes corresponds to 44μm. Reproduced from Molecular Microbiology48: 1511–1524.

Extracellular polymeric substances (EPS), which consist of polysaccharide, extracellular DNA, and protein, have important cell-to-cell interconnecting functions in biofilms. Five gene clusters in P. aeruginosa potentially encode functions for exopolysaccharide biosynthesis. The alg biosynthesis operon (PA3540–PA3551) appears not to be expressed in P. aeruginosa flow-chamber biofilms (Wozniaket al,2003), but evidently play a role in biofilm formation by P. aeruginosa in particular in the lungs of cystic fibrosis patients (Govan & Deretic, 1996). The psl cluster (PA2231–PA2245) is involved in the production of a mannose-rich exopolysaccharide, which was shown to play a role in P. aeruginosa biofilm formation (Friedman & Kolter, 2004b; Jacksonet al,2004; Matsukawa & Greenberg, 2004). Bacteria with a mutation in pslA displayed delayed microcolony formation in a flow chamber setup (Matsukawa & Greenberg, 2004). The psl locus was also found to be important for early biofilm formation in a microtitre tray assay (Friedman & Kolter, 2004b). The pel (PA3058–PA3064) cluster encodes production of a glucose rich matrix component which was shown to facilitate biofilm formation of some P. aeruginosa strains at the liquid air interface (Friedman & Kolter, 2004a). Mutations in the pel locus also made P. aeruginosa biofilms susceptible to shear force in a microtitre tray setup (Friedman & Kolter, 2004a). In addition, two other gene clusters (PA1381–PA1393 and PA3552–3558) potentially code for synthesis of exopolysaccharides which may have roles in biofilm development under some conditions (Matsukawa & Greenberg, 2004). Addition of DNase to young (24-h-old) flow chamber grown P. aeruginosa biofilms was shown to remove the biofilm cells, indicating an importance of extracellular DNA as a biofilm matrix component under some conditions (Whitchurchet al,2002). Evidence has been presented that the extracellular DNA in P. aeruginosa biofilms may be generated via at least two different pathways. A basal level of extracellular DNA appears to be generated via a pathway which is not linked to quorum sensing, while a larger amount of extracellular DNA appears to be generated via a quorum-sensing regulated pathway (Allesen-Holmet al,2006). Evidence has been presented that the extracellular DNA in P. aeruginosa biofilms originates from lysis of a small subpopulation of the cells (Allesen-Holmet al,2006). The mutants lasIrhlI (PA1432–PA3476), pqsA (PA0996) and fliMpilA (PA1443–PA4525) formed biofilms with low extracellular DNA levels and showed increased susceptibility to treatment with SDS, indicating a stabilizing effect of the extracellular DNA in biofilms (Allesen-Holmet al,2006). In addition, mature wild-type biofilms that were treated with DNase for a short time before SDS treatment showed elevated susceptibility to the SDS treatment (Allesen-Holmet al,2006). In agreement with a role of quorum sensing in biofilm development Davies (1998) found that, unlike the wild-type, a P. aeruginosa lasI mutant was unable to structurally mature into mushroom shaped SDS-resistant microcolonies, but developed a flat SDS-susceptible biofilm in a flow-chamber setup. Evidence is accruing that cup fimbria in addition to their role in initial biofilm formation also play a role as cell-to-cell interconnecting compounds in mature biofilms. The sadARS genes (PA3946–3948) codes for a putative sensor histidine kinase and two response regulators, and mutations in any of these genes were shown to result in biofilms with an altered mature structure (Kuchmaet al,2005). In another study, the sadARS genes (PA3946-3948, termed rocARS) were shown to regulate biosynthesis of cup fimbria (Kulasekaraet al,2005).

After formation of the initial microcolonies structural biofilm development by P. aeruginosa appears to be conditional. For example, a flat biofilm was formed in flow chambers irrigated with citrate minimal medium (Klausenet al,2003b), while a heterogeneous biofilm with mushroom-shaped multicellular structures was formed in flow chambers irrigated with glucose minimal medium (Klausenet al,2003a). Pseudomonas aeruginosa biofilm development in flow chambers with citrate as carbon source was shown to occur via formation of initial microcolonies by clonal growth of sessile cells at the substratum, followed by expansive migration of the bacteria out on the substratum, and the eventual formation of a flat biofilm (Fig. 1) (Klausenet al,2003b). As biofilm formation by a P. aeruginosa pilA mutant (deficient in biogenesis of type IV pili) occurred without the expansive phase, and resulted in discrete protruding microcolonies, it was suggested that the expansive migration of the bacteria out on the substratum was type IV pili-driven. CLSM time-lapse microscopy indicated that the shift from nonmotile to migrating cells occurred when the initial microcolonies reached a certain size (Fig. 1), suggesting that the shift may by induced by some sort of limitation arising in the initial microcolonies. The formation of mushroom-shaped structures in glucose-grown P. aeruginosa biofilms was shown to occur in a sequential process involving a nonmotile bacterial subpopulation which formed the initial microcolonies by growth in certain foci of the biofilm, and a migrating bacterial subpopulation which initially formed a monolayer on the substratum, and subsequently formed the mushroom caps by aggregating on top of the microcolonies (mushroom stalks) via a type IV pili-dependent process (Klausenet al,2003a). In colour-coded biofilms this type of dynamic development gave rise to mushroom-shaped structures with stalks composed of clonal subpopulations of one colour, and caps composed of aggregated subpopulations of mixed colour (Fig. 2a). Growth of the initial microcolonies in the glucose-grown biofilms continued past the point where spreading by twitching motility prevented further microcolony formation in the citrate-grown biofilms. In pilA/wild-type mixed biofilms formation of mushroom-shaped structures could occur also with citrate as the carbon source (Fig. 2b). In these citrate-grown mixed biofilms the pilA bacteria formed the mushroom stalks, and the wild-type bacteria accumulated on top of these stalks and formed the mushroom caps. The fact that a twitching deficient hyperpiliated P. aeruginosa pilT mutant formed biofilms with large mushroom-shaped structures (Chiang & Burrows, 2003), and our unpublished experiments with mixtures of pilA mutants and twitching deficient but piliated mutants, indicate that type IV piliation of the cells is required for mushroom cap formation, while type IV pili-driven motility, although not absolutely required, can facilitate the process substantially. The extracellular DNA in P. aeruginosa biofilms was shown to be present in high concentrations on the microcolonies in young biofilms and between the stalk-forming and the cap-forming subpopulations in mature biofilms (Fig. 3) (Allesen-Holmet al,2006). Because type IV pili bind to DNA (Aaset al,2002; Van Schaiket al,2005), it is tempting to speculate that the high concentration of extracellular DNA on the mushroom stalks might cause accumulation of migrating (either twitching or swimming) piliated bacteria resulting in the formation of the mushroom caps.

Figure 2

(a) CLSM images were acquired in a 4-day-old biofilm which was initiated with a 1:1 mixture of cyan fluorescent protein-tagged and yellow fluorescent protein-tagged Pseudomonas aeruginosa PAO1 wild-type bacteria, and grown on glucose minimal medium. (b) CLSM images were acquired in a 4-day-old biofilm which was initiated with a 1:1 mixture of yellow fluorescent P. aeruginosa PAO1 wild-type and cyan fluorescent P. aeruginosa pilA mutant, and grown on citrate minimal medium. The central pictures in the left panels show horizontal CLSM sections, while the two flanking pictures in the left panels show vertical CLSM sections. The thin white lines in the central pictures indicate the positions of the vertical sections, while the thin white lines in the flanking pictures indicate the position of the horizontal section. The right panels show the corresponding three-dimensional CLSM images. The bars represent 20μm. Reproduced from Molecular Microbiology50: 61–68.

Figure 3

Biofilms of green fluorescent protein-tagged Pseudomonas aeruginosa PAO1 wild-type bacteria were grown in flow-chambers irrigated with glucose minimal medium, and the extracellular DNA was stained red with propidium iodide or DDAO. CLSM images were acquired in a 2-day-old DDAO-stained biofilm (a) and in a 5-day-old propidium iodide-stained biofilm (b), and are represented as horizontal optical sections located close to the substratum flanked by vertical optical sections. Because the DDAO stain intrinsically gives rise to more intense fluorescence than the propidium iodide stain, the observed difference in fluorescence should not be taken as evidence that the young biofilm contains more extracellular DNA than the established biofilm. The bars represent 20μm. Reproduced from Molecular Microbiology59: 1114–1128.

Dispersion of biofilms in response to perturbation in the carbon availability has been reported for P. aeruginosa biofilms. Sauer et al. (2004) reported that P. aeruginosa biofilms grown in flow chambers on glutamate medium responded to an abrupt upshift in carbon availability by initiating a dispersion process that lead to the majority of the biomass being released from the biofilm. The extent of dispersion was dependent on the carbon source used and was associated with increased expression of flagella and down regulation of twitching motility. The production of rhamnolipid biosurfactant has also been associated with dispersal of cells from P. aeruginosa biofilms (Boleset al,2005). During biofilm development of P. aeruginosa in rich medium local dispersion was observed as a hollowing out of some microcolonies (Purevdorj-Gageet al,2005). Through careful microscopic inspection it was observed that two subpopulations existed in the microcolonies. On the outside of the microcolonies a wall forming subpopulation of nonmotile cells was present, and on the inside of the microcolonies a motile rapidly moving subpopulation was present. The motile subpopulation eventually found its way out of the microcolony which resulted in microcolonies with a central void. This phenomenon has been termed ‘seeding dispersal’ and was shown to be dependant on the microcolonies reaching a critical size.

Development and dynamics in P. putida biofilms

Initial attachment and colonization by P. putida on the surface of plant roots and fungal hyphae have been proposed to involve flagella (Yanget al,1994; Turnbullet al,2001). Flagella, however, do not seem to be required for efficient attachment and biofilm formation by P. putida under all conditions as a nonflagellated fliM (PP4358) mutant formed biofilms in a flow-chamber system within the same timeframe as the wild-type (Gjermansenet al,2005). In addition, Choy (2004) reported that for a P. putida morA (similar to PP0672) mutant, which was shown to be hyper-flagellated and hyper-motile, initial biofilm formation was impaired. In a transposon screen for P. putida mutants defective in adhesion to seeds Espinosa-Urgel (2000) identified a mutant, mus-24 (PP0168), which had defects in both seed adhesion and adhesion to abiotic materials like glass and plastic. The mus-24 locus of P. putida is highly homologues to the lapA gene of Pseudomonas fluorescens (Pfl_0133 in strain PfO-1), and a detailed study of a P. fluorescens lapA mutant documented that LapA was necessary for nonreversible binding to surfaces (Hinsaet al,2003). LapA belongs to a group of large surface proteins which contain highly repetitive sequences and share functional features such as adhesion to surfaces (recently reviewed by Lasa & Penades, 2006). Our unpublished results suggest that LapA also functions as cell-to-surface adhesin in flow-chamber-grown P. putida biofilms, and that it in addition is important for later stages in biofilm development as a part of the cell-to-cell interconnecting matrix.

Cellulose is an important constituent of the biofilm matrix material of many bacteria and has also been suggested to have a role in P. putida biofilms (Camesano & Abu-Lail, 2002; Chang & Halverson, 2003). In P. fluorescens an acetylated form of cellulose plays an important role in maintaining the structural rigidity of air–liquid interface biofilms (Spierset al,2003; Spiers & Rainey, 2005), and the production of this cellulose-like polymer confers an ecological advantage during the colonization of plant surfaces (Galet al,2003). Pseudomonas putida KT2440 has a complete cellulose biosynthetic operon (PP2634–2638), and the production of calcofluor white-stainable material in flow-chamber-grown P. putida biofilms suggests that cellulose is expressed in this system (Nelsonet al,2002; our unpublished results). In addition to cellulose and LapA protein P. putida has been reported to produce substantial amounts of extracellular DNA in the sessile mode of growth (Steinberger & Holden, 2005), which may function as biofilm stabilizing matrix material.

Structural biofilm development by P. putida in flow chambers irrigated with citrate minimal medium has been investigated using dual fluorescent labelling and CLSM (Tolker-Nielsenet al,2000). Biofilm development occurs initially by the formation of compact microcolonies via clonal growth of nonmotile bacteria (Fig. 4a). When the compact microcolonies reach a critical size, the bacteria inside the microcolonies are no longer firmly interconnected and they move rapidly by means of flagellum-driven motility. The combination of local dissolution of the cell-to-cell interconnecting bonds and flagellum-driven motility, results in release of cells from inside of the microcolonies, and leads to a shift in the biofilm architecture from compact microcolonies to loose protruding structures (Fig. 5). The loose protruding structures in mature P. putida biofilms contain a mixture of bacteria from separate microcolonies (Fig. 4b) indicating that a substantial dynamic reorganization occurs during development (Tolker-Nielsenet al,2000). Time-lapse microscopy revealed that the structural rearrangement was preceded by rapid movement of bacteria inside the microcolonies, and the lack of this movement in microcolonies formed by a fliM mutant suggested that it was flagellum-driven. Consistent with the described developmental steps in P. putida biofilms, Sauer and colleagues found that the expression of flagella in P. putida populations was down regulated quickly after surface attachment, but upregulated again at later stages of biofilm development (Sauer & Camper, 2001). Evidence has been provided that the local biofilm dissolution and release of cells from the interior of the microcolonies is induced by local carbon starvation (Gjermansenet al,2005). It was also found that global carbon starvation led to rapid dispersal of the entire biofilm within few minutes after the onset of starvation (Fig. 6). Microscopic analysis indicated that the cells were swimming rapidly during the global dispersion process, but experiments with a nonmotile fliM mutant showed that dispersion did not occur because of the motility per se, but through modulation of the cell-to-cell interconnecting compounds (Gjermansenet al,2005). Experiments with a biofilm dispersion deficient mutant indicated that the local and global dispersion process utilizes the same molecular mechanisms.

Figure 4

Biofilms of a 1:1 mixture of green fluorescent protein-tagged and red fluorescent protein-tagged Pseudomonas putida OUS82 wild-type bacteria were grown in flow-chambers irrigated with citrate minimal medium, and the spatial structures in the developing colour-coded biofilm were studied by the use of CLSM. A horizontal optical CLSM section was acquired close to the substratum in a 1-day-old biofilm (a), and 40μm from the substratum in a 4-day-old biofilm (b). The bars represent 20μm. Reproduced from Journal of Bacteriology182: 6482–6489 with permission from ASM.

Figure 5

Biofilms of green fluorescent protein-tagged Pseudomonas putida OUS82 wild-type bacteria were grown in flow chambers irrigated with citrate minimal medium, and the spatial structures in the developing biofilm were studied by the use of CLSM. Shadow projection CLSM micrographs recorded in a 1 (a), 3 (b), and 5 (c)-day-old biofilm are shown. The bars represent 20μm. Reproduced from Journal of Bacteriology182: 6482–6489 with permission from ASM.

Figure 6

Biofilms of green fluorescent protein-tagged Pseudomonas putida OUS82 wild-type bacteria were grown in flow chambers irrigated with citrate minimal medium, and the dispersion of the biofilm in response to carbon starvation was studied by the use of CLSM. Shadow projection CLSM micrographs were recorded 0 (a), 5 (b), and 15min (c) after stoppage of the medium flow. The bars represent 20μm. Reproduced from Environmental Microbiology7: 894–906.

Genetic analysis of the dispersion process in P. putida OUS82 lead to the identification of an operon, consisting of the two genes PP0164 and PP0165, which regulates the adhesiveness of the cells (Gjermansenet al,2005). Biofilms formed by a PP0164 mutant did not undergo the transition from compact microcolonies to loose protruding structures but instead the microcolonies continued to expand reaching almost complete substratum coverage. In addition, biofilms formed by the PP0164 mutant did not undergo dispersion in response to global carbon starvation. The PP0164 protein contains a putative signal protein for transport to the periplasm and a domain of unknown function, DUF920, which based on a bioinformatics approach has been proposed to function as a papain-like cysteine proteinase (Ginalskiet al,2004). Contrary to the PP0164 mutants, bacteria with mutations in the PP0165 gene were unable to initiate biofilm formation. The PP0165 gene encodes a putative inner membrane protein which contains an N-terminal periplasmic domain and C-terminal GGDEF and EAL domains. The GGDEF and EAL domains have been implicated in the turnover of the bacterial second messenger c-di-GMP. The GGDEF domain contains the diguanylate cyclase activity for synthesis of c-di-GMP, and the EAL domain contains a phosphodiesterase activity for degradation of c-di-GMP (Paulet al,2004; Schmidtet al,2005). Evidence that GGDEF and EAL domain proteins are involved in the regulation of cellular adhesiveness in diverse bacteria through modulation of the intracellular c-di-GMP levels is growing rapidly (D'Argenio & Miller, 2004; Römlinget al,2005). Comparative sequence analysis shows that PP0164/PP0165-homologues in a number of environmental species including P. putida, P. fluorescens and Shewanella oneidensis cluster together with genes encoding synthesis and export of proteins homologous to LapA. Our unpublished data suggest that PP0164 and PP0165 in conjunction regulate the expression or surface localization of the LapA protein and that this modulation play an important role in the formation/dispersal processes observed. Our experiments also suggest that LapA is not only a surface adhesin but also act as an important component of the matrix that interconnects the cells in the biofilm at later stages. One possible mechanism for the dispersion process could be dependant on the release of LapA from the bacterial surface in response to starvation by the action of the putative protease activity of the PP0164 protein. This hypothesis would be consistent with the very fast initiation of the dispersion process in response to starvation.

Is biofilm structure the outcome of a developmental program or the result of ecological adaptation?

Although P. putida and P. aeruginosa biofilm development can result in elaborate multicellular structures and appears to involve complex regulation, there is at present no evidence suggesting that these organisms possess a comprehensive genetic biofilm developmental program comparable to, for example, the fruiting body developmental program of myxobacteria (see Kaiser (2003) for a review about myxobacterial development). The elaborate biofilm structure may be nothing more than a spatial pattern emerging because of the actions of the individuals constituting the biofilm. Comparative analysis of the transcriptome or proteome of P. aeruginosa cells from biofilms and planktonic cultures (e.g. Whiteleyet al,2001; Saueret al,2002; Hentzeret al,2004) has not identified a comprehensive genetic program for biofilm development. The fact that biofilm development is conditionally dependent on, among other things, the type and amount of carbon source, implies that the organisms are able to respond to environmental factors while they are forming the biofilm. It is therefore more likely that the bacteria are not committed to a special developmental program during biofilm formation, but that they have evolved mechanisms to optimize surface colonization in response to the environmental conditions. These mechanisms include the ability to regulate cellular adhesiveness and motility in response to micro-environmental signals including those secreted by the bacteria themselves.

From this perspective of ecological adaptation of individual cells driving the evolution of attachment, biofilm formation, and dispersal mechanisms, the degree of cooperative behaviour of the members of the multicellular biofilm is called into question. Perhaps the evolution of some of the apparently cooperative traits can be explained more simply by the fitness advantages of this behaviour for the individual bacterium? In the following, we will examine the evidence for and against the cooperative nature of the above reviewed behaviour of P. aeruginosa and P. putida in biofilms.

To what degree does biofilm formation involve multicellular cooperation?

Most aspects of biofilm formation reviewed in this article can be viewed either as social behaviour or as single-cell behaviour (for examples of such a dual interpretation, see Redfield, 2002). Social behaviour may be positive (cooperation and altruism) or negative (spite and cannibalism). The key problem for the evolution of cooperation is the emergence of cheaters. They benefit from the cooperating individuals' investment into public goods without investing into cooperation themselves. This creates a conflict of interest between the fitness of the individual and the fitness of the group (Velicer, 2003; Kreft, 2004b; Travisano & Velicer, 2004). The evolution of cooperation in microorganisms can be explained either by kin selection theory (Hamilton, 1964; Griffinet al,2004) or by multilevel selection theory (Sober & Wilson, 1998). These two explanations complement rather than contradict each other (Queller, 2004). The predominantly clonal structure of the microcolonies, combined with refounding of them from single cells, acts as a purification mechanism to get rid of cheaters, and promotes the evolution of cooperation in biofilms (Kreft, 2004a; Kreft & Bonhoeffer, 2005). However, the mushroom caps in P. aeruginosa biofilms, and other multicellular structures which are formed via pathways that involve aggregation, are not clonal.

The production of the cell-to-cell interconnecting components in biofilms may be a cost each bacterium pays in order to contribute to the social activity of creating a protective biofilm domicile, or it may simply increase the adhesiveness of single cells allowing them to persist in an environment. Production of matrix components appears to be cooperative because cheaters who benefit from the existence of the matrix, but do not invest into its construction, may exist. Further, the matrix as a public good could be exploited by selfish mutants using it as a food source but damaging its integrity in turn.

The local biofilm dissolution and structural changes occurring during P. putida biofilm development may be a social process which creates optimal nutritional conditions to the biofilm population, or it may simply be a response of the individual cells to the local nutritional conditions. The latter view is supported by the experiments showing that almost all cells leave the biofilm if the flow, and therefore substrate supply, is switched off. The formation of regular mushroom-shaped structures in P. aeruginosa biofilms via a pathway that involves type IV pili-dependent cellular migration may be a coordinated social process that creates biofilms with architectures which allow optimal circulation and nutrient supply to the population, or it may, because nutrient gradients develop in biofilms, be the result of chemotaxis of individuals moving to a favourable nutrient-containing microenvironment.

Despite the increase of biofilm surface area owing to the formation of the loose protruding structures or the mushroom-shaped structures, the effective mass transfer area may actually decrease because only the top of the structures receives most of the substrate, unless convection in the valleys is dominant (Picioreanuet al,2000). If we do assume that the formation of the mushroom-shaped structures or loose protruding structures increase substrate flux to the biofilms as a whole, which would imply a group level benefit of the structural changes, why does the stalk-forming subpopulation not evolve to become a cap-forming subpopulation or the wall-forming subpopulation not evolve to be part of the loose protruding structures, thereby increasing their substrate supply? The fitness of the subpopulations would differ under most circumstances, and natural selection would be expected to shift the balance in favour of the fitter subpopulation. The split into the phenotypically different subpopulations appears to be epigenetic, however, because isolated cap subpopulations are able to form biofilms with mushroom-shaped structures (our unpublished data), and evolution, therefore, cannot select the fitter subpopulation. Circumventing natural selection between the subpopulations subjects the combination of the subpopulations to natural selection. Therefore, a population may increase its fitness by evolving a mechanism for epigenetic formation of subpopulations whose interactions give rise to synergistic benefits. The relative frequency of the subpopulations might have evolved to depend on the environmental conditions, which could be underlying causes resulting in the structural differences seen in citrate-grown and glucose-grown P. aeruginosa biofilms.

While the observed diversification of the biofilm populations into subpopulations may result in population-level fitness benefits because of substrate transfer optimization, it could also be an insurance mechanism (Boleset al,2004). For example, the bacteria in the caps or loose protruding structures could be able to benefit from better growth in good times, while the slower-growing or dormant bacteria in the stalks or walls could have better chances of survival in times of adversity (Wolfet al,2005).

The release of DNA via lysis of a small subpopulation of P. aeruginosa cells may serve the purpose of biofilm stabilization, procurement of food, elimination of damaged cells, or provide a common gene pool for diversification of the population. Whatever the purpose, cell lysis may be altruism if it is autolysis (Lewis, 2000) or cannibalism/fratricide if the lysed cells are killed by their kin/siblings (Gonzalez-Pastoret al,2003; Gilmore & Haas, 2005; Havarsteinet al,2006). Both altruism and cannibalism can give rise to group-level benefits such as diversification. The observed involvement of quorum sensing in DNA release from the lysing subpopulation, resulting in biofilm stabilization, is difficult not to interpret as social behaviour because the lysing cells that provide the DNA obviously do not benefit themselves directly.


Model biofilms formed by Pseudomonas aeruginosa and Pseudomonas putida in laboratory flow-chamber setups represent examples of surface-associated microbial communities which display dynamic developmental patterns. Dependent on the experimental conditions these model biofilms develop flat, mushroom-shaped, or loose protruding structures through a series of distinct steps where regulation of cellular migration and adhesiveness play important roles. Despite the appearance of these characteristic developmental patterns in the model biofilms, the available evidence suggests that these organisms do not possess a comprehensive genetic program for biofilm development. Instead, the bacteria appear to have evolved a number of different mechanisms to optimize surface attachment, microcolony formation, and dispersal, of which they express a subset in response to the prevailing environmental conditions. It appears that the increased adhesiveness that distinguishes biofilm cells from planktonic cells can be achieved by the expression of many different genes, and that the environmental conditions determine the actual cell-to-cell interconnecting components which are being expressed. While certain characteristics of biofilm formation, such as diversification of the population and DNA-release by cell lysis, presumably constitute social behaviour, the group benefits of the apparent biofilm structural engineering may not only lie in improved substrate supply but also in the establishment of subpopulations with different survival strategies.


This work was supported by grants from the Lundbeck foundation and the Danish Technical Research Council (grant no. 26-03-02) to T.T.N.


View Abstract