18.5 N&V Feature Greenbe#822F58 12/5/06 6:34 PM Page 300

Vol 441|18 May 2006 NEWS & VIEWS FEATURE

MICROBIAL SCIENCES The superficial life of microbes Roberto Kolter and E. Peter Greenberg The social activities and organization of are crucial to their ecological success. But it is only in recent years that we have begun to study these secret societies.

Most surfaces on this planet teem with micro- and establish a sedentary yet remarkably liquid cultures, which produce homogeneous bial life, creating ecosystems of diverse organ- diverse community (Fig. 1a). These are com- populations of genetically identical cells, isms that flourish in slimy beds of their own munities in the sense that we humans organize growth in generates a large amount of making. The plaque encrusting our teeth, the ourselves into communities with division of genetic diversity 2. How can a single cell, with a slippery coating on river stones, the gunge labour — as the surface-associated population single genetic complement, give rise to a clogging up water pipes or infected wounds: grows, the becomes increasingly biofilm population in which the individual cells these are just a few examples of the microbial sophisticated in its activities, with individual are genetically different from one another? The ‘biofilms’ that form anywhere there is a surface cells taking on specific tasks. As a result, simplest explanation may be that in any biofilm, with a little moisture and some nutrients. biofilms can develop intricate architectures; individual cells are stuck in the same place, Although microbes by and large live in such striking mushroom-like structures can bloom attached to their neighbours and the slime that biofilm communities, most of our understand- on submerged surfaces, and aerial projections surrounds them, so their access to nutrients will ing of their stems from experiments sprout from surfaces exposed to the air vary as gradients form within the biofilms using liquid cultures of dispersed, free-swim- (Fig. 1b,c). The diversity of biofilm architec- through metabolic activity. As a consequence, ming ‘planktonic’ cells. In the past decade, how- tures is akin to a miniature coral reef, with many microniches are likely to arise, as are ran- ever, the number of studies performed on their structures differing enormously depend- dom spontaneous mutants that can exploit surface-associated microbes has increased dra- ing on the species present and the environ- those microniches. As various mutants grow in matically. Today, we recognize that most, if not mental conditions. their location, the result will be a biofilm con- all, microbial species can form biofilms. The Natural biofilms nearly always harbour a taining a multitude of genotypes. This genetic physiological differences between free-living multitude of microbial species — the region diversity within a strain may provide a sort of individuals and communal biofilm-associated around a single tooth, for example, is often insurance, as the population can adapt better cells are becoming apparent, as are the regula- sheathed in a community of several hundred to sudden environmental changes than can a tory mechanisms that underlie the switch species1. Within such microbial ecosystems, genetically homogeneous population. between these two lifestyles. species richness may ensure stability of the To understand these minute ecosystems, it ecosystem in the face of changing environ- is necessary to be able to grow, observe and The diverse universe of biofilms mental conditions, very much as species rich- manipulate biofilms in the laboratory. This When individual bacterial cells encounter a ness is thought to aid the stability of macro- is accomplished using transparent flow cells, surface under conditions propitious for growth, scale ecosystems such as a tropical rainforest. where bacteria attach to a glass surface and they almost invariably undergo dramatic Even in the context of artificial, single- are continuously fed fresh nutrients3. The ensu- lifestyle changes. The nomadic single cells species biofilms, strain diversification seems to ing biofilm growth can be followed using settle down on the surface, where they divide be the rule. Unlike the growth of bacteria in confocal laser microscopy. In such flow cells,

ab c

Figure 1 | Biofilm formation and architecture. a, Individual bacteria b, These mushroom-like structures are characteristic of many submerged undergo a reversible lifestyle switch between a nomadic and a sedentary biofilms. In this Pseudomonas aeruginosa biofilm, which was grown existence. In the process, they lose motility and become enclosed in a gooey in a flow cell, they are about 150 m high. c, A biofilm of P. aeruginosa extracellular matrix. As the community grows, different cell types appear grown on a semi-solid agar surface (about 10 mm in diameter). The as the original strain diversifies to allow cells to take on different tasks colony displays aerial projections or wrinkling characteristic of an or to exist in different ‘microniches’, shown by differently coloured cells. air-exposed biofilm.

300 © 2006 Publishing Group 18.5 N&V Feature Greenbe#822F58 12/5/06 6:34 PM Page 301

NATURE|Vol 441|18 May 2006 NEWS & VIEWS FEATURE

experimental reproducibility is best obtained are carried out respectively by proteins that using single-species systems. As a consequence, Sedentary have structural domains possessing diguany- (biofilm) many microbial species have been analysed late cyclase and phosphodiesterase enzymatic as monocultures. Yet, in natural settings, these activities (referred to as GGDEF and EAL microbes nearly always grow in the context of Sensor domains, respectively, because of the highly multi-species communities. To approximate the GGDEF conserved amino-acid sequences they con- natural setting more closely, investigators have 2 GTP tain). Genomic analyses of dozens of microbial begun to foray into systems containing two or 2 GMP c-di-GMP genomes show that most bacteria harbour a more microbial species. Some potentially inter- pGpG multitude of these proteins — Vibrio vulnifi- esting multi-species model systems are poised cus, an extreme example, has 66 GGDEF for molecular analyses. Among these are EAL proteins and 33 EAL proteins18. Often, the mixed-species biofilms of the opportunistic Sensor GGDEF or EAL segment is fused to one pathogens Pseudomonas aeruginosa (a bac- of several types of environment-sensing terium) and Candida albicans (a fungus)4; two domains, implying that the production and Nomadic bacterial species that colonize teeth, Strepto- (planktonic) breakdown of c-di-GMP are controlled by the coccus gordonii and Veillonella atypica 5; and two bacterium’s surroundings. species that occur in the soil environment Figure 2 | The lifestyle switch. The second- The GGDEF and EAL proteins, and by around plant roots, P. aeruginosa and Agro- messenger molecule c-di-GMP mediates the extension c-di-GMP, are involved in many 6 bacterium tumefaciens . switch between nomadic and sedentary lifestyles. diverse cellular activities, but the mechanism Many species of bacteria have a multitude of by which c-di-GMP acts remains largely Sticking together enzymes capable of making c-di-GMP from GTP unknown. It may act to integrate the many Given the diversity of species that form (GGDEF proteins), and of breaking it down by external inputs sensed by GGDEF and EAL biofilms, it is not surprising that the molecular hydrolysing it into GMP (EAL proteins). Overall, proteins, allowing this single messenger to mechanisms involved in biofilm development it seems that increased intracellular levels of instigate multiple indirect effects. Or perhaps reflect the remarkable variety of the microbial c-di-GMP favour a sedentary existence, whereas there are small cellular ‘compartments’ where reduced levels of the second messenger favour world. Different species build biofilms differ- a nomadic existence. local fluctuations in the concentration of ently, and many strains have multiple biofilm- c-di-GMP cause independent outcomes. formation pathways. For instance, some that production of flagella and of extracellular Initially, c-di-GMP was shown to function by mutants of the soil bacterium Pseudomonas matrix are mutually exclusive processes: bac- activating the enzyme that synthesizes extra- fluorescens that cannot make a biofilm when teria give up the ability to move in order to set- cellular cellulose in Gluconoacetobacter xyli- grown with glucose as the sole carbon source tle down. The regulatory circuitry involved in nus. This activity occurs at what is known as do form biofilms when the sole carbon source this lifestyle switch is beginning to be mapped the post-translational level, once the target is glutamate7. out. In , for instance, the switch protein, in this case the enzyme, has been So, do any general principles emerge? One between flagellar synthesis and matrix pro- synthesized19. By extension, investigators feature that does seem to be common to all duction is controlled by the master gene regu- believed c-di-GMP acted exclusively at the biofilms is that the cells secrete a ‘matrix’ to lator SinR (refs 13–15). This protein directly post-translational level on enzymes in other hold themselves in place and to provide a represses genes encoding matrix components bacteria. But it now seems that c-di-GMP can buffer against the environment. The make-up and activates those encoding flagellar compo- affect many processes, not only by acting of each matrix is different, and, depending on nents (seemingly indirectly). SinR activity post-translationally, but also by regulating the contributing species and environmental itself is antagonized by another protein, SinI, gene expression. For example, it functions at conditions, there can be various mixes of poly- whose abundance and activity are modulated the level of gene-transcription control in saccharides, proteins and even nucleic acids8. by changing environmental conditions. The P. aeruginosa20. No gene-regulatory proteins The rich diversity among the matrix compo- inverse regulation of genes involved in motil- that bind to c-di-GMP have yet been identi- nents can be glimpsed by mentioning what is ity and matrix components is apparent in fied conclusively, so how this messenger known about the polysaccharides secreted by several other organisms, including E. coli, passes on the external signal to the master just one species, P. ae r ug inos a , one of the most- P. aeruginosa, Salmonella enterica and Vibrio regulators is unclear. studied organisms in terms of biofilm forma- cholerae, although the specific regulatory mol- Even though the structure of c-di-GMP and tion. Many strains of this bacterium have the ecules used have not been identified16,17. its involvement in regulating cellulose synthe- capacity to extrude at least three types of extra- But how is the lifestyle switch thrown? An sis in G. xylinus were reported21 in 1987, it was cellular polysaccharide, called alginate, Pel and individual cell must have some way of recog- only in the past few years, when it became Psl. Alginate was initially considered the major nizing that it is near a suitable surface, and of apparent that c-di-GMP occurs in bacterial biofilm polysaccharide, but it now seems that passing that information on to the master pathogens, that interest in this molecule really it is a major matrix component only of the gene regulators and other effectors so that intensified. In this sense, the history of biofilms formed by unusual variants9. Pel and it can cease roaming and settle down. In sev- c-di-GMP research resembles the history of Psl are apparently more ubiquitous: the genes eral species, the messenger seems to be a quorum sensing — the means by which bacte- involved in the synthesis of Pel occur in all P. small cytoplasmic molecule called bis- ria detect the presence of others of their kind. aeruginosa strains studied, whereas the genes (3–5)-cyclic dimeric guanosine monophos- As early as 1968, there were reports of what we associated with Psl are present in only some10–12. phate (c-di-GMP)18 (Fig. 2). Low intracellu- now know as quorum sensing in the marine But even the Pel genes have diverse regulatory lar c-di-GMP concentrations are found in bacterium Vibrio fischeri22. But it was only in regions, suggesting that the various strains planktonic cells, but these levels rise once the the mid-1990s that quorum sensing was rec- express this matrix component differently. cells have given up their nomadic lifestyle. ognized as a feature of many pathogens23, lead- Furthermore, it looks as though c-di-GMP ing to a flood of interest in this process. One Lifestyle choices can control the synthesis of diverse cellular has to wonder how many interesting molecu- Motile bacteria generally have flagella — components involved in both motility and lar mechanisms have already been found in corkscrew-like appendages that rotate to pro- matrix production in response to changing microbes but have yet to receive much atten- pel them. These structures occur only in the environments. tion because they are not being studied in microbes’ planktonic form, and it now seems The synthesis and degradation of c-di-GMP pathogens.

301 © 2006 Nature Publishing Group 18.5 N&V Feature Greenbe#822F58 12/5/06 6:34 PM Page 302

NEWS & VIEWS FEATURE NATURE|Vol 441|18 May 2006

Box 1 Biofilms in human health and disease decade have the tools of molecular been brought to bear on biofilms — a form Environmental microbiologists and engineers that has immense biological significance. With have long recognized that biofilms form on recent technological developments that allow surfaces — clogging pipework, for instance, or analysis of multi-species biofilms, the next aiding water clean-up — but this fact has only decade should see further integration of mol- recently been appreciated fully by those in the ecular approaches into the study of biofilm medical sciences. The intimate relationship biology, and significant progress in our

between the human body and its resident CARR CDC/R. M. DONLAN, J. understanding of more complex problems microbes is now beginning to be elucidated. of bacterial social behaviour. At an applied Such biofilms are mostly beneficial. By level, we can hope to develop agents that colonizing our skin, teeth and the mucosal control the biology of biofilms — to allow surfaces lining our gut and airways, thousands of them to be analysed further, to treat biofilm different microbial species play an essential role infections, and to control biofilms for benefi- in our nutrition and serve as a primary line of cial uses (Box 1). Clearly, much remains to be defence against invading pathogens. But discovered about the diverse world of bac- although they are vital to our well-being, we know terial biofilms, and it seems the time is ripe for very little about the microbiota that thrive on and such discoveries. ■ in our bodies. We scarcely know how to cultivate Roberto Kolter is in the Department even a small percentage of these microbial of and , inhabitants so as to be able to study them closely. systemic infection. Unfortunately, for reasons Harvard Medical School, Boston, Some biofilms, however, are extremely harmful that remain poorly understood, biofilm- Massachusetts 02115, USA. to us. Notably, foreign objects such as catheters associated microbes are particularly e-mail: [email protected] and prostheses provide enticing surfaces for impervious to many antimicrobial agents, so E. Peter Greenberg is in the Department colonization, and microbes often quickly take biofilm-related infections are difficult to treat. of Microbiology, University of Washington, up residence (for example, the image shows A major goal in this field is to develop therapeutic Seattle, Washington 98195, USA. a Staphylococcus aureus biofilm found on an strategies that can control the undesirable e-mail: [email protected] in-dwelling catheter, magnified 1,180 times). growth of biofilms while leaving beneficial The resulting biofilms can become reservoirs for biofilms intact. R.K. & E.P.G. 1. Paster, B. J. et al. J. Bacteriol. 183, 3770–3783 (2001). 2. Boles, B. R., Thoendel, M. & Singh, P. K. Proc. Natl Acad. Sci. USA 101, 16630–16635 (2004). 3. Palmer, R. J. Jr Methods Enzymol. 310, 160–166 (1999). Community relations mounting evidence that bacteria do have sys- 4. Hogan, D. A., Vik, A. & Kolter, R. A. Mol. Microbiol. 54, 1212–1223 (2004). Once an advance party of bacteria has set up tems that monitor and respond to quorum- 5. Egland, P. G., Palmer, R. J. Jr & Kolenbrander, P. E. Proc. Natl camp on a pristine surface, the cells can let sensing signals from other species. In some Acad. Sci. USA 101, 16917–16922 (2004). each other know they are there by exuding cases these systems are very specific, and in 6. An, D., Danhorn, T., Fuqua, W. C. & Parsek, M. R. Proc. Natl 23 Acad. Sci. USA 103, 3828–3833 (2006). quorum-sensing signals . Planktonic bacteria others there may be methods for sensing sig- 7. O’Toole, G. A. & Kolter, R. Mol. Microbiol. 28, 449–461 use these chemical signals to regulate a variety nals or cues produced by almost any member (1998). of processes, depending on population density. of the local microflora27. 8. Branda, S. S., Vik, S., Friedman, L. & Kolter, R. Trends Microbiol. But in many bacterial species the same signals If different species can sense one another, 13, 20–26 (2005). 9. Wozniak, D. J. et al. Proc. Natl Acad. Sci. USA 100, are now known to be cues for biofilm develop- how do they interact? Can biofilm communi- 7907–7912 (2003). ment. This was first shown in P. aeruginosa, ties encourage or discourage individuals in the 10. Matsukawa, M. & Greenberg, E. P. J. Bacteriol. 186, in which quorum-sensing signals control the planktonic community over entry to the 4449–4456 (2004). 11. Friedman, L. & Kolter, R. J. Bacteriol. 186, 4457–4465 expression of hundreds of genes throughout biofilm? In P. aeruginosa biofilms, the quo- (2004). 24,25 the genome . In certain surroundings, rum-sensing response results in the activation 12. Jackson, K. D., Starkey, M., Kremer, S., Parsek, M. R. & mutants defective in quorum signalling of a battery of potential defence systems, Wozniak, D. J. J. Bacteriol. 186, 4466–4475 (2004). 13. Kearns, D. B., Chu, F., Branda, S. S., Kolter, R. & Losick, R. are also defective in biofilm formation; for including a system for producing cyanide and Mol. Microbiol. 55, 739–749 (2005). instance, they do not grow mushroom-like one that releases to attack other 14. Chu, F., Kearns, D. B., Branda, S. S., Kolter, R. & Losick, R. structures under conditions where these struc- bacteria25,28. Whether this response does Mol. Microbiol. 59, 1216–1228 (2006). 15. Branda, S. S., Chu, F., Kearns, D. B., Losick, R. & Kolter, R. tures would normally form. The mutants still indeed constitute a defence against interlopers Mol. Microbiol. 59, 1229–1238 (2006). make biofilms, however, and in certain condi- is being studied experimentally by tagging the 16. Simm, R., Morr, M., Kader, A., Nimtz, M. & Romling, U. tions where mushroom structures would not biofilm bacteria and the planktonic bacteria Mol. Microbiol. 53, 1123–1134 (2004). usually form, these biofilms are indistinguish- with differently coloured fluorescent proteins 17. Watnick, P. I., Lauriano, C. M., Klose, K. E., Croal, L. & Kolter, R. Mol. Microbiol. 39, 223–235 (2001). able from those formed by unmutated strains. (G. J. Balzer, M. H. Hentzer and M. R. Parsek, 18. Romling, U., Gomelsky, M. & Galperin, M. Y. Mol. Microbiol. This observation may be due to the fact that personal communication). 57, 629–639 (2005). the effects of the quorum-sensing mutations 19. Weinhouse, H. et al. FEBS Lett. 416, 207–211 (1997). 20. Hickman, J. W., Tifrea, D. F. & Harwood, C. S. Proc. Natl on biofilm formation are not apparent visually, Outlook for the future Acad. Sci. USA 102, 14422–14427 (2005). or that quorum sensing is involved in specific Surfaces afford a greater capacity for organiza- 21. Ross, P. et al. Nature 325, 279–281 (1987). steps in biofilm formation such as mushroom tion than do liquids, and so microbes encoun- 22. Kempner, E. S. & Hanson, F. E. J. Bacteriol. 95, 975–979 (1968). building. So it seems that biofilm formation is tering surfaces can form aggregates that begin 23. Fuqua, W. C., Winans, S. C. & Greenberg, E. P. J. Bacteriol. a social activity that is also governed by both to display some attributes of multicellularity. 176, 269–275 (1994). genetic and environmental factors (reviewed These groups of cells are held in place by an 24. Davies, D. G. et al. Science 280, 295–298 (1998). 25. Whiteley, M., Lee, K. M. & Greenberg, E. P. Proc. Natl Acad. in ref. 26). extracellular matrix and can use intercellular Sci. USA 96, 13904–13909 (1999). An area that will see increasing attention signalling for communication. In doing so, 26. Parsek, M. R. & Greenberg, E. P. Trends Microbiol. 13, 27–33 involves the question of whether signalling in they develop intriguing features that we are (2005). biofilm biology occurs between species. Multi- only just beginning to understand. 27. Waters, C. M. & Bassler, B. L. Annu. Rev. Cell Dev. Biol. 21, 319–346 (2005). species biofilm model systems are now avail- Traditionally, microbes have been studied 28. Mashburn, L. M. & Whiteley, M. Nature 437, 422–425 able to address such questions, and there is in their planktonic forms. Only in the past (2005).

302 © 2006 Nature Publishing Group