The Involvement of Cell-to-Cell Signals in the Development of a Bacterial Biofilm

See allHide authors and affiliations

Science  10 Apr 1998:
Vol. 280, Issue 5361, pp. 295-298
DOI: 10.1126/science.280.5361.295


Bacteria in nature often exist as sessile communities called biofilms. These communities develop structures that are morphologically and physiologically differentiated from free-living bacteria. A cell-to-cell signal is involved in the development of Pseudomonas aeruginosa biofilms. A specific signaling mutant, alasI mutant, forms flat, undifferentiated biofilms that unlike wild-type biofilms are sensitive to the biocide sodium dodecyl sulfate. Mutant biofilms appeared normal when grown in the presence of a synthetic signal molecule. The involvement of an intercellular signal molecule in the development of P. aeruginosa biofilms suggests possible targets to control biofilm growth on catheters, in cystic fibrosis, and in other environments where P. aeruginosa biofilms are a persistent problem.

Certain bacteria, such as the fruiting bacteria, communicate with each other to form structured macroscopic groups (1, 2). Recently, it has become apparent that in appropriate environments, common bacteria exhibit similar social behavior. Microscope observations of living bacterial biofilms attached to a glass surface have revealed that these sessile microbial biofilm populations have a complicated structural architecture (3, 4). Biofilms of mixed bacterial communities and of individual species such as Pseudomonas aeruginosa that develop on solid surfaces exposed to a continuous flow of nutrients form thick layers consisting of differentiated mushroom- and pillar-like structures separated by water-filled spaces. The structures consist primarily of an extracellular polysaccharide (EPS) matrix or glycocalyx in which the bacterial cells are embedded (5). The finding that P. aeruginosa produces at least two extracellular signals involved in cell-to-cell communication and cell density-dependent expression of many secreted virulence factors suggests cell-to-cell signaling could be involved in the differentiation of P. aeruginosa biofilms, much as cell-to-cell signaling is involved in the development of specialized structures of fruiting bacteria like Myxococcus(1, 2). Thus, we initiated this study of the role of intercellular signals in P. aeruginosa biofilm development.

The two cell-to-cell signaling systems identified in P. aeruginosa are the lasR-lasI andrhlR-rhlI (also called vsmR-vsmI) systems (6-10). The lasI gene product directs the synthesis of a diffusible extracellular signal,N-(3-oxododecanoyl)-l-homoserine lactone (3OC12-HSL). The lasR product is a transcriptional regulator that requires sufficient levels of 3OC12-HSL to activate a number of virulence genes, including lasI, and the rhlR-rhlI system (11-14). The rhlI product directs the synthesis of the extracellular signal,N-buytryl-l-homoserine lactone, which is required for activation of virulence genes and expression of the stationary-phase σ factor, RpoS, by the rhlR gene product (13-16). At sufficient population densities these self-produced signals reach the concentrations required for gene activation. Thus, this type of gene regulation has been termed quorum sensing and response (17). Recently, acylhomoserine lactones have been detected in naturally occurring biofilms (18).

Because quorum sensing requires a sufficient density of bacteria, neither of the P. aeruginosa signals would be expected to participate in the initial stages of biofilm formation, attachment, and proliferation. However, these signals may be involved in biofilm differentiation. To test this hypothesis, we monitored biofilm formation of wild-type (WT) P. aeruginosa PAO1 and alasI-rhlI double mutant that makes neither of the quorum-sensing signals (19). Both strains adhered to and proliferated on the glass surface of the reaction chamber and reached a steady state within 2 weeks. However, the mutant biofilm was thin, about 20% of the WT thickness, and the cells were more densely packed (Fig. 1A). Furthermore, the WT formed characteristic microcolonies composed of groups of cells separated by water channels, whereas the mutant appeared to grow rather as continuous sheets on the glass surface. These results are consistent with the hypothesis that although not involved in the initial attachment and growth stages of biofilm formation, one or both of theP. aeruginosa quorum-sensing systems participates in the subsequent biofilm differentiation process.

Figure 1

Characteristics of P. aeruginosaWT and quorum-sensing mutant biofilms in flow-through continuous-culture reaction vessels (31). (A) Depth of biofilms (mean ± SD of 20 measurements) (open bars) and cell-packing as determined by a nearest-neighbor analysis of cells at the glass bioreactor surface (filled bars) (32,33). Strains: WT, PAO1; lasI, rhlImutant, PAO-JP2; rhlI mutant, PDO100; and lasImutant, PAO-JP1. (B) Epifluorescence and scanning confocal photomicrographs of the WT and the lasI mutantP. aeruginosa biofilms containing the GFP expression vector pMRP9-1. (Top) Epifluorescence photomicrographs of the WT (PAO1) and the lasI mutant (PAO-JP1) grown with or without the autoinducer, 3OC12-HSL added to the medium. (Bottom) Saggital views of Z series of wild-type andlasI mutant biofilms (with or without 3OC12-HSL) acquired by scanning confocal laser microscopy. Because the bacterial cells contain GFP, the color correlates with cell density.

To determine whether lasI, rhlI, or both are required for the normal development of the P. aeruginosa biofilm, we tested mutants defective in one or the other of these genes (19, 20). As indicated by measuring average thickness of biofilms and cell packing (Fig. 1A), therhlI mutant formed biofilms similar to that of the WT, and the lasI mutant formed biofilms similar to that of the double mutant.

To further compare the WT and lasI mutant biofilms, we constructed a plasmid containing a gene encoding an enhanced green fluorescent protein (GFP) and introduced it into the two strains (21). This enabled us to image P. aeruginosacells in the biofilms by epifluorescence and scanning confocal microscopy (Fig. 1B). Scanning confocal microscopy was used to produce a side view of the WT and mutant biofilms. Only a few cells of an adherent cluster of the WT were apparent at the interface with the solid surface, and the cells appeared to be in a loose confederation with considerable intervening space between bacteria. Staining with alcian blue, which binds polysaccharides (22), showed that at least some of the intervening space consisted of an EPS matrix. The mutant biofilm was thin and much more uniform. A top view generated by epifluorescence microscopy showed the clusters of WT cells compared with the more uniform distribution of the lasI mutant (Fig.1B).

To confirm that abnormal biofilm formation in the lasImutant was due to the absence of 3OC12-HSL, we added this compound to the medium flowing through a reaction chamber with a mutant biofilm (23). In the presence of 3OC12-HSL, thelasI mutant formed biofilms of an average thickness and cell density similar to that of the WT biofilms (Fig. 1A), and as shown by epifluorescence and confocal microscopy, the addition of the quorum-sensing signal allowed the development of clusters of relatively loosely packed cells (Fig. 1B). From this experiment we conclude that the quorum-sensing signal 3OC12-HSL is required for normal biofilm differentiation, and that gradients of the signal do not appear to be necessary for this differentiation.

The EPS matrix is generally considered to be important in cementing bacterial cells together in the biofilm structure (24). The WT P. aeruginosa cells appeared to be embedded in an EPS matrix. Thus, we examined EPS levels in biofilms by measuring uronic acids (25), a constituent of the alginate EPS of P. aeruginosa (26), and by measuring total carbohydrates in biofilm samples (27). We detected no significant differences between the WT and the lasI mutant (Fig. 2) despite their markedly different appearance (Fig. 1). As shown previously, P. aeruginosabiofilm and free-floating (planktonic) cells also produce similar amounts of EPS (28). However, the distribution of the glycocalyx is different, with biofilm cells cemented to one another by the EPS matrix and planktonic cells having a compressed, incomplete glycocalyx (28). The mutant biofilms in our study may have a glycocalyx matrix similar to that of planktonic cells. This could result in the tight packing of mutant biofilm bacteria. The results suggest that in the lasI mutant, the initial stages of biofilm formation proceed as normal, but differentiation from attached planktonic bacteria into biofilm bacteria does not proceed. Our hypothesis is that in the WT, this differentiation is triggered when the cell mass produces a sufficient amount of the quorum-sensing signal, 3OC12-HSL. Although the signal generated by RhlI does not appear to participate in biofilm differentiation, there may well be other as yet unidentified signals implicated in this process.

Figure 2

Analysis of total carbohydrates (milligrams per milligram of total biofilm protein) and total uronic acids (nanograms per nanograms of total biofilm protein) in biofilm samples of the WT P. aeruginosa, PAO1, and the lasImutant, PAO-JP1 (34). The filled bars show the average value for total carbohydrates, and the open bars show the average values for total uronic acids. The averages of two separate experiments are shown; bars correspond to the range.

Because we have hypothesized that the abnormal, undifferentiated biofilm formed by the lasI mutant contains cells similar in physiology to planktonic cells, we examined whether the abnormal mutant biofilm might be sensitive to biocides that do not disrupt WT biofilms. Thus, we exposed biofilms of the WT and the lasI mutant to the detergent sodium dodecyl sulfate (SDS, 0.2%). This treatment had no detectable effect on the WT, but within 5 min of SDS addition, most or all of the bacteria in the lasI mutant biofilm detached from the surface and dispersed (Fig. 3). Exposure of a lasI mutant biofilm grown in the presence of synthetic 3OC12-HSL to 0.2% SDS for up to 24 hours had no detectable effect. As with WT biofilms, detachment and dispersal of the 3OC12-HSL–rescued lasI mutant biofilm were not evident (Fig. 3), and indeed the average thickness of this biofilm was not changed by SDS treatment (93 ± 21 μm before and 24 hours after SDS treatment versus 102 ± 21 μm before and 24 hours after SDS treatment of the WT biofilm).

Figure 3

SDS-induced detachment of mutant biofilm cells from a glass surface. Phase contrast photomicrographs of the P. aeruginosa WT strain, PAO1; the lasI mutant, PAO-JP1; and the lasI mutant grown in the presence of the autoinducer, 3OC12-HSL (AI), immediately before addition of SDS, and at the times indicated after SDS addition (35).

Our studies demonstrate that a cell-to-cell signal is required for the differentiation of individual cells of the common bacterium P. aeruginosa into complex multicellular structures. A mutation that blocks generation of the signal molecule hinders differentiation, and the resulting abnormal biofilm appears to be sensitive to the detergent biocide SDS. The control of biofilm differentiation and integrity by quorum sensing has important implications in medicine.Pseudomonas aeruginosa can colonize devices such as catheters (29), and it colonizes the lungs of most cystic fibrosis patients (30). Because of their innate resistance to antibiotics and other biocides, biofilms in these environments are difficult, if not impossible, to eradicate. Bacterial biofilms also present other problems of significant economic importance in both industry and medicine. Our finding of a connection between biofilm differentiation into clusters of bacteria resistant to the detergent biocide SDS and a quorum-sensing signal suggests that inhibition of these cell-to-cell signals could aid in the treatment of biofilms.

  • * To whom correspondence should be addressed. E-mail: epgreen{at}


Stay Connected to Science

Navigate This Article