A major protein component of the Bacillus subtilis biofilm matrix


  • Steven S. Branda,

    1. Department of Microbiology and Molecular Genetics, Harvard Medical School, Boston, MA 02115, USA.
    Search for more papers by this author
      Biosystems Research, Sandia National Laboratories, Livermore, CA 94551 USA.
  • Frances Chu,

    1. Department of Molecular and Cellular Biology, Harvard University, Cambridge, MA 02138, USA.
    Search for more papers by this author
  • Daniel B. Kearns,

    1. Department of Molecular and Cellular Biology, Harvard University, Cambridge, MA 02138, USA.
    Search for more papers by this author
      ‡Department of Biology, Indiana University, Bloomington, IN 47408, USA.
  • Richard Losick,

    1. Department of Molecular and Cellular Biology, Harvard University, Cambridge, MA 02138, USA.
    Search for more papers by this author
  • Roberto Kolter

    Corresponding author
    1. Department of Microbiology and Molecular Genetics, Harvard Medical School, Boston, MA 02115, USA.
      *E-mail rkolter@hms.harvard.edu; Tel. (+1) 617 432 1776; Fax (+1) 617 738 7664.
    Search for more papers by this author

  • Present addresses: Biosystems Research, Sandia National Laboratories, Livermore, CA 94551 USA. Department of Biology, Indiana University, Bloomington, IN 47408, USA.

*E-mail rkolter@hms.harvard.edu; Tel. (+1) 617 432 1776; Fax (+1) 617 738 7664.


Microbes construct structurally complex multicellular communities (biofilms) through production of an extracellular matrix. Here we present evidence from scanning electron microscopy showing that a wild strain of the Gram positive bacterium Bacillus subtilis builds such a matrix. Genetic, biochemical and cytological evidence indicates that the matrix is composed predominantly of a protein component, TasA, and an exopolysaccharide component. The absence of TasA or the exopolysaccharide resulted in a residual matrix, while the absence of both components led to complete failure to form complex multicellular communities. Extracellular complementation experiments revealed that a functional matrix can be assembled even when TasA and the exopolysaccharide are produced by different cells, reinforcing the view that the components contribute to matrix formation in an extracellular manner. Having defined the major components of the biofilm matrix and the control of their synthesis by the global regulator SinR, we present a working model for how B. subtilis switches between nomadic and sedentary lifestyles.


Biofilms share an important structural feature: their constituent cells are bound together by an extracellular matrix that mainly consists of macromolecules, including polysaccharides, proteins, and nucleic acids, that are produced by the cells themselves (Branda et al., 2005). While extracellular matrices have been shown to play essential roles in the establishment and maintenance of biofilm structure, our current understanding of the biofilm matrix is rudimentary at best. The molecular composition of the matrix is likely to be complex, yet few components have been identified in even the most thoroughly characterized biofilms. Still less is known about the regulation of matrix components; e.g. how is production of an individual component co-ordinated with that of the others? To address these issues we have initiated a comprehensive analysis of the extracellular matrix of biofilms produced by the spore-forming Gram positive soil bacterium Bacillus subtilis.

In standing liquid culture B. subtilis constructs a floating biofilm, or pellicle, through a dramatic switch in its growth pattern: single, swimming cells migrate to the air–liquid interface and begin to proliferate as long chains of non-motile cells that are bound together in parallel arrangement (Branda et al., 2001). These bundled filamentous cells give rise to larger features of the biofilm architecture. In the case of biofilms formed by undomesticated strains of B. subtilis (e.g. NCIB 3610, hereafter referred to as our ‘wild’ strain), such features include aerial projections that serve as preferential sites of sporulation within the biofilm, reminiscent of microbial fruiting bodies. The wild B. subtilis strain also forms biofilms on agar plates (i.e. colonies) that show a similar organization of cells and analogous architectural features.

Our previous work strongly suggested that bundling of the cell chains is a key step in B. subtilis biofilm development. Specifically, we identified several genes required for formation of a mature biofilm, and found that many of them play a role in cell chain bundling in nascent biofilms (Branda et al., 2004). Some of the ‘bundling’ genes appear to mediate the biosynthesis of an exopolysaccharide (EPS); for instance, pgcA (formerly known as yhxB) encodes an enzyme that catalyses the production of nucleotide sugars (Lazarevic et al., 2005), while the 15-gene epsA-O operon is predicted to encode biosynthetic machinery that uses nucleotide sugar substrates to produce EPS (Branda et al., 2001; 2004; Kearns et al., 2005). This suggests that, as in many other microbial biofilms (Branda et al., 2005), the cells of the B. subtilis biofilm are bound together by an extracellular matrix that is, at least in part, composed of EPS. The importance of EPS in B. subtilis biofilm formation is further evident in the fact that the eps operon is directly repressed by SinR, a transcription factor that acts as master regulator of biofilm formation and, more specifically, controls the transition from single motile cells to bundled cell chains (Kearns et al., 2005).

Some ‘bundling’ genes lack an obvious link to EPS production, however, and therefore might contribute to the extracellular matrix by different means. Among the most intriguing of such genes are yqxM, sipW and tasA (Branda et al., 2004; Hamon et al., 2004), which together comprise a three-gene operon that is directly repressed by SinR (Chu et al., 2006). The sipW gene encodes a type I signal peptidase that is specifically required for the processing and secretion of the proteins encoded by yqxM and tasA (Serrano et al., 1999; Stover and Driks, 1999a). YqxM is a protein of unknown function that has been detected in cell culture mainly under conditions of high salt (Stover and Driks, 1999b). In contrast, TasA has been detected in stationary phase and sporulating cultures, both in the medium and in association with the cells and spores (Serrano et al., 1999; Stover and Driks, 1999a; Hirose et al., 2000; Antelmann et al., 2001). Several different functions have been attributed to TasA: it appears to have broad-spectrum antibacterial activity (Stover and Driks, 1999a), and seems to play subtle roles in spore coat assembly and germination (Serrano et al., 1999).

Chu et al. (2006) demonstrated that all three members of the yqxM-sipW-tasA operon are needed for the formation of robust biofilms. Here we demonstrate that the tasA gene product is the major protein component of the biofilm extracellular matrix. We further demonstrate that YqxM is important for the proper localization of TasA to the matrix. In light of these and previous results, we propose a model in which SinR regulates the lifestyle switch from swimming single cells to a structured multicellular community by co-ordinating the synthesis of the major components of the extracellular matrix.


The role of tasA in formation of the biofilm matrix

In the wild strain of B. subtilis 3610, biofilm development critically depends upon a switch in growth pattern: from single, motile cells to non-motile cell chains that are bundled together in parallel arrangement (Branda et al., 2001). ‘Bundling’ mutants, which transit to growth as chains of cells that are not bundled together, typically form biofilms of unusual morphology; for instance, eps mutants produce flat pellicles and colonies. In a previous study we demonstrated that mutation of the yqxM-sipW-tasA operon has dramatic effects on cell chain bundling and biofilm architecture (Branda et al., 2004). Thus, it was not surprising to find that targeted disruption of either yqxM or tasA generated mutants that failed to bundle when forming pellicles (data not shown).

We have hypothesized that cell bundling during B. subtilis biofilm formation is accomplished through production of an extracellular matrix, and that failure to bundle is indicative of a defect in that matrix (Branda et al., 2001). In this study we used scanning electron microscopy (SEM) to visualize the extracellular material of pellicles formed by the wild strain as well as by tasA and eps mutants (Fig. 1). We found that in wild pellicles the cells were almost completely enveloped in thick sheets of extracellular material. Pellicles formed by tasA featured less extracellular material, such that many cells were left nearly bare, and the material itself had a fibrous appearance typical of EPS-rich biofilm matrices. Still less extracellular material was evident in eps pellicles, appearing as occasional thread-like strands. It is important to note that sample preparation for SEM included several dehydration steps (see Experimental procedures), and therefore the electron micrographs of Fig. 1 cannot be considered faithful representations of fully hydrated biofilms. Nevertheless, it is clear from the results of these experiments that an extracellular matrix is present in B. subtilis biofilms, and that at least some of its physical properties depend upon the tasA and eps genes.

Figure 1.

Scanning electron microscopy of wild and mutant pellicles, at 10 000× magnification. Measure bar is 200 µm.

TasA is a component of the biofilm matrix

Two lines of evidence suggested that the TasA protein itself might be a component of the B. subtilis extracellular matrix. First, TasA has been shown to be a relatively abundant secreted protein; indeed, laboratory strains produce enough TasA that it is readily detected in the conditioned medium of vegetative cell cultures, as well as in spores (Serrano et al., 1999; Stover and Driks, 1999a; Hirose et al., 2000; Antelmann et al., 2001). Second, our characterization of tasA mutants implicated TasA as being important for the structure of the biofilm matrix (see Fig. 1). To test the hypothesis that TasA is indeed a protein component of the biofilm matrix, we attempted to detect TasA in extracellular material extracted from pellicles formed by the wild strain (Fig. 2). In these experiments the wild strain was grown in MSgg medium at 22°C without agitation for 72 h. The pellicles produced under these conditions were mature enough that they could be manipulated easily, yet young enough that they contained few spores (data not shown); the latter detail was important because in laboratory strains TasA is associated with spores (Serrano et al., 1999; Stover and Driks, 1999a) and we sought to avoid this potential complication in our analyses. Pellicles were harvested simply by lifting them from the surface of the culture medium; in effect, this manipulation yielded two biochemical fractions: the pellicle and the conditioned medium. Residual cells in the conditioned medium were removed by passage through a 0.2 µm filter. Proteins present in the conditioned medium were precipitated, quantified, separated by SDS-PAGE and subjected to Western blot analyses.

Figure 2.

Biochemical fractionation of wild-type and yqxM pellicles and localization of TasA using immunodetection. Localization of σA exclusively in the cell fraction indicates there was no cell lysis during the fractionation.

As evident from the results shown in Fig. 2, the conditioned medium (‘Medium’ fraction) contained little if any TasA, as detected using highly specific anti-TasA polyclonal antibodies and chemiluminescence. The anti-TasA antibodies used were those generated by Stover and Driks (1999a); as a control a tasA mutant strain did not give any detectable signal in the Western blot analyses (not shown). As a control for release of cytoplasmic proteins, the Western blot membrane was reprobed using highly specific polyclonal antibodies raised against σA, the dominant sigma-factor in vegetative cells. Only trace amounts of σA were detected in the ‘Medium’ fraction, indicating that the harvested pellicles were not rife with cell lysis and concomitant release of cytoplasmic proteins into the medium. The fact that we failed to detect TasA in the medium is interesting, because a number of groups have reported that laboratory strains of B. subtilis, when grown in shaken cultures, secrete relatively large quantities of TasA into the medium (Hirose et al., 2000; Antelmann et al., 2001). It is possible that our wild strain does not share this property of laboratory strains. Alternatively, our standing culture conditions may strongly favour incorporation of TasA into the extracellular matrix, whereas shaking of the culture may disrupt such a process.

The pellicle fraction was itself further fractionated to separate the cells from the extracellular material. This was accomplished through mild sonication followed by differential centrifugation (see Experimental procedures). Two new fractions were generated: a ‘Cell’ fraction, which mainly consisted of cells derived from the pellicle; and a ‘Matrix’ fraction, which mainly consisted of extracellular material derived from the pellicle. To ensure its being cell-free, the latter fraction was passaged through a 0.2 µm filter. Proteins in the fractions were precipitated and subjected to Western blot analysis as described above. We detected TasA signal in the ‘Cell’ fraction, as well as in the ‘Matrix’ fraction (Fig. 2). In contrast, σA was detected exclusively in the ‘Cell’ fraction. This result strongly suggested that the pellicle-fractionation process left most cells intact, a conclusion supported by observations made during phase-contrast microscopy of pellicle fractions (data not shown). Taken together, these results indicate that in the wild-type pellicle TasA is primarily associated with the extracellular matrix, and that it is not released into the medium.

Interestingly, TasA localization to the matrix appears to depend on the presence of a functional yqxM gene. This is evident in the results shown in the lower six panels of Fig. 2. In contrast to TasA localization in wild pellicles, no matrix-associated TasA was detected in a yqxM mutant. Furthermore, the insertion of a yqxM complementation construct into the yqxM mutant restores TasA localization to the matrix (data not shown). In the yqxM fractionation the anti-σA antibodies once again served to ascertain that there was no cell lysis during pellicle growth and fractionation. The fact that a double band is detected with the anti-TasA antibody in the yqxM pellicle suggests that in the absence of YqxM some pre-TasA may accumulate; this observation was not pursued further.

Previous studies have shown that some TasA can be found associated with spores (Serrano et al., 1999; Stover and Driks, 1999a). For this reason we sought to determine whether localization of TasA in the extracellular matrix somehow depended upon sporulation. This seemed an unlikely scenario, as we had detected TasA in extracellular material derived from relatively young (72 h), nearly sporeless pellicles (Fig. 2 and data not shown). Nevertheless, we addressed this possibility by determining the localization of TasA in pellicles formed by a mutant that cannot sporulate. The mutant we chose to study, sigE, lacks the ability to synthesize σE, a sigma-factor that activates the expression of genes essential for sporulation (Piggot and Losick, 2002). The σE protein is required for key developmental events that follow the committed step of the sporulation process; thus, the sigE mutant is able to initiate sporulation but becomes trapped at an intermediate stage, unable to return to vegetative growth (Dworkin and Losick, 2005). Despite this defect, sigE forms pellicles that are indistinguishable, in terms of their gross morphology, from those formed by the wild strain (Branda et al., 2001). Fractionation and Western analysis of sigE pellicle cultures revealed a familiar pattern: TasA mainly segregated with the extracellular material but was associated with the cells as well, whereas it was not detected in the conditioned medium (data not shown). To summarize, in pellicles TasA is primarily localized to the extracellular matrix in a YqxM-dependent fashion, even when the pellicle does not contain spores.

The contribution of TasA and EPS to biofilm structure

The results presented thus far, when combined with those published previously, have led to the identification of two critical components of the extracellular matrix. TasA is located in the extracellular matrix of pellicles (Fig. 2), and is required for structural integrity of the matrix (Fig. 1) as well as for the development of biofilm architecture (see Supplementary material and Fig. S1). The presumed EPS product of the eps operon is also required for matrix structure (Fig. 1) as well as for biofilm architecture (Branda et al., 2001). Moreover, carbohydrate analyses indicate a reduction in carbohydrate content and complexity in pellicles formed by an eps mutant (data not shown), consistent with the idea that EPS is also a component of the extracellular matrix.

Having identified two potential components of the matrix, it was important to determine whether their individual contributions to the matrix are made through a common molecular mechanism, or through complementary mechanisms. To distinguish between these possibilities, we constructed a tasA eps double mutant and analysed its capacity to form pellicles. Strikingly, the tasA eps double mutant failed to form a pellicle of any sort – a cumulative phenotype, in which it differed from the phenotypes of the single mutants (flat pellicles in the case of tasA; flat, fragile pellicles in the case of eps) (Fig. 3, top row). These results indicate TasA makes a contribution to the matrix that is different from and complementary to that made by EPS. Furthermore, the observation that the tasA eps double mutant make no pellicle whatsoever emphasizes the particular importance of TasA and EPS to biofilm formation and, more specifically, to the extracellular matrix.

Figure 3.

Double mutant phenotypes with respect to pellicle formation (top row) and cell filamentation and bundling (bottom row). Cells in the bottom row were viewed using phase-contrast microscopy, at 600× magnification. Measure bar is 10 µm.

Transcription of the genes encoding the components of the extracellular matrix, eps and yqxM-sipW-tasA, is directly repressed by SinR, and these 18 genes appear to constitute many of the most strongly regulated members of the SinR regulon (Chu et al., 2006). When conditions are propitious for biofilm formation, the SinR antagonist SinI is synthesized, causing derepression of eps and yqxM-sipW-tasA and initiation of matrix synthesis. Thus, it is not surprising that a sinI mutant phenocopies the tasA eps double mutants in terms of pellicle formation (Fig. 3, top row). However, there is a clear difference between the double mutant and the sinI mutant that is evident from phase contrast microscopy of shaken MSgg cultures (Fig. 3, bottom row). The tasA eps double mutant, like its parental single mutants, forms long chains of cells that are not bundled together (Fig. 3, bottom row). In contrast, the sinI mutant produced only individual cells (Fig. 3, bottom row). The simplest explanation for this difference may be found in the fact that SinR is thought to promote the expression of autolysins (Sekiguchi et al., 1988; Kuroda and Sekiguchi, 1993; Rashid and Sekiguchi, 1996). In the absence of SinI, SinR might in some way aid in cell separation after division. Another conclusion that can be drawn from the results shown in Fig. 3 is that the formation of cell chains is not in itself sufficient for pellicle formation; rather, matrix components are required as well. This is evident in the fact that the tasA eps double mutant makes chains but utterly fails to make a pellicle.

Extracellular complementation of matrix component mutants

Given the results indicating that TasA, and EPS are the major components of the matrix, it was of interest to determine whether corresponding mutants might exhibit extracellular complementation when cocultured. To carry out such experiments we grew cultures of the mutants alone and in coculture. The pellicle phenotypes of the wild strain and the two mutants used are recapitulated in the top row of Fig. 4, along with the pellicle resulting from mixed culture. When the tasA mutant was cocultured with the eps mutant the resulting pellicle was similar to that of the wild type. The observed extracellular complementation is a particularly dramatic result because it is completely opposite to the phenotype of the tasA eps double mutant which is completely unable to form a pellicle (see Fig. 3, top row). As can be seen in the lower row of Fig. 4, the characteristic aerial projections that are evident in wild colonies are missing in colonies formed by tasA and eps mutants, yet when the mutants are cocultured the aerial projections reappear. Thus, TasA and EPS can ‘assemble’ extracellularly into an apparently functional matrix even when produced by different cells.

Figure 4.

Extracellular complementation during cocultivation of mutants defective in the synthesis of matrix components. Cultures in which the two strains were mixed are designated with a ‘+’ between the two genotypes utilized. Top row: pellicle formation. Bottom row: colony edges, as visualized using a stereomicroscope at 32× magnification. Measure bar is 200 µm.

The involvement of the yqxM-sipW-tasA operon in surface adhesion

In addition to our results implicating the yqxM-sipW-tasA operon in the formation of structured communities, others have reported effects of sipW on biofilm formation. Guided by results from transcriptional profiling experiments, Hamon et al. (2004) generated mutations in each of the operon's genes in the laboratory strain JH642 and tested their effects on surface adhesion. Their results led them to the conclusion that of the three genes of the operon only sipW, and not yqxM or tasA, was required for surface adhesion. Given that sipW encodes a signal peptidase whose only known substrates are encoded by yqxM and tasA (Stover and Driks, 1999a), Hamon et al. (2004) hypothesized that other, as-yet undiscovered SipW substrates are required for surface adhesion. At first glance this hypothesis might seem to be in conflict with our results.

It seemed plausible that this apparent discrepancy might arise from differences between the systems under investigation. To test this possibility, we analysed the lab strain PY79 and an otherwise isogenic tasA deletion mutant for their ability to adhere to a surface using the crystal violet staining assay (O’Toole and Kolter, 1998; O’Toole et al., 1999) that was used by Hamon et al. (2004). We found that the lab strain produced a robust ring of surface-associated cells on the tube wall at a position corresponding to the air–liquid interface of the standing culture (Fig. 5A). As a negative control, a spo0A mutant showed a strong defect in surface adherence, as evident in the less robust ring that it produced (Fig. 5A) and the comparatively small amount of crystal violet associated with that ring (Fig. 5B); these results are consistent with those of a previous study (Hamon and Lazazzera, 2001). We were most interested to find that a mutant lacking the entire yqxM-sipW-tasA operon showed a clear defect in surface adhesion, whereas one lacking only tasA adhered at least as well as the parental strain (Fig. 5A and B). These results are entirely in accordance with those reported earlier (Hamon et al., 2004). Taken together, the simplest interpretation of these results is that sipW is indeed the only gene of the operon needed for the adherence of cells of laboratory strains to a solid surface, whereas all three genes play important roles in the development of complex, structured communities by the wild strain.

Figure 5.

Crystal violet staining (A) and quantification (B) of biofilm formed on a solid surface by laboratory strain PY79 and several mutant derivatives.

Is γ-polyglutamic acid a component of the biofilm matrix?

Another extracellular polymer, γ-polyglutamic acid, has been implicated in cellular adherence to solid surfaces (Stanley and Lazazzera, 2005) and could be an additional matrix component. As the genes responsible for the synthesis of γ-polyglutamic acid, pgsBCA, have been identified in B. subtilis (Ashiuchi et al., 1999; 2001), we constructed derivatives of our wild strain bearing mutations in them. We detected no discernable effect on colony or pellicle phenotype, and thus conclude that γ-polyglutamic acid does not contribute significantly to the extracellular matrix in the wild strain (Fig. 6). Production of γ-polyglutamic acid has been linked to colony mucoidy (Stanley and Lazazzera, 2005). In this regard, it is interesting to recall that in the wild strain background a lack of Spo0A leads to a mucoid phenotype (Branda et al., 2001 and see Fig. 6). This mucoidy was lost when the pgs genes were disrupted (Fig. 6). These results indicate that in the wild strain mucoidy is the result of γ-polyglutamic acid production and that its synthesis is somehow repressed by Spo0A. In contrast, Spo0A is required positively for pellicle formation and the development of complex architecture (Branda et al., 2001) as well as for solid surface-associated biofilm formation (Hamon and Lazazzera, 2001). Thus it would appear that γ-polyglutamic acid synthesis is normally repressed during biofilm development. The observation that some wild strains are constitutively mucoid might indicate that they have altered regulation of γ-polyglutamic acid synthesis (Stanley and Lazazzera, 2005).

Figure 6.

The role of γ-polyglutamic acid in colony development. Measure bar is 5 mm.


In the microbial world, existence within surface-associated structured multicellular communities may be the rule, rather than the exception. The organizing principle afforded by surfaces appears to have been universally exploited by microbes during the course of evolution; most microorganisms appear to be capable of forming biofilms of some sort or another. Molecular genetic approaches have begun to elucidate the mechanisms by which microbes build such communities. Comparison of biofilm formation by many different organisms has led to the appreciation that the extracellular matrix is absolutely essential for biofilm structure (Branda et al., 2005). Here we present molecular details as to how the model microbe B. subtilis produces its extracellular matrix during biofilm formation. Like several other microbes, B. subtilis builds a matrix that contains both EPS and protein. Mutations that eliminate EPS production have a severe effect on biofilm formation, while those that eliminate TasA tend to have less severe effects. Most importantly, elimination of both EPS and TasA leads to a particularly severe phenotype in which pellicle formation is entirely prevented. We interpret these results as an indication that EPS and TasA are the two most important matrix components in B. subtilis biofilms.

What might be the roles of the various components in assembling the extracellular matrix? Clearly, we have only begun to answer this question. The absence of TasA in the matrix in a yqxM mutant indicates that YqxM is somehow involved in the proper localization of TasA to the matrix. TasA is reported to have antibacterial activity against a wide variety of bacteria (Stover and Driks, 1999a). This leads us to speculate that TasA plays a dual role in biofilm formation. It contributes to the structure of the extracellular matrix while also warding off competing species of bacteria. Finally, the results from extracellular complementation studies also shed some light as to the functioning of the various matrix components; TasA and the EPS can be synthesized in different cells and still be assembled properly outside the cell to yield a functional matrix.

In closing, we summarize our current understanding regarding the regulation and synthesis of the extracelluar matrix by presenting a working model (Fig. 7). The master regulator SinR co-ordinately represses the expression of the eps and yqxM-sipW-tasA operons, thereby controlling the synthesis of the major components of the biofilm matrix (Kearns et al., 2005; Chu et al., 2006). The decision as to whether to adopt the nomadic or sedentary lifestyles thus appears to hinge on the state of SinR. When conditions favour the nomadic lifestyle, SinR is capable of fully repressing matrix synthesis because its activity is unhindered. There are numerous reports that link SinR as a positive effector of motility and cell separation (Pooley and Karamata, 1984; Sekiguchi et al., 1988; Kuroda and Sekiguchi, 1993; Barilla et al., 1994; Marquez-Magana et al., 1994; Fredrick and Helmann, 1996; Rashid and Sekiguchi, 1996). Under conditions propitious for biofilm formation, the repressive activity of SinR is antagonized by its binding to SinI, which, in turn, leads to matrix formation. In some unknown manner this also leads to repression of motility and cell separation. Prior work has indicated that SinR-mediated repression of genes involved in biofilm formation is somehow modulated by Spo0A and by YlbF/YmcA (Kearns et al., 2005). Exactly how this occurs and how environmental signals are sensed and transduced remains unknown and is the subject of current investigation.

Figure 7.

Model for the regulatory circuitry controlling the switch between the nomadic and sedentary lifestyles of B. subtilis. The leftward arrow, which represents a positive effect of SinR on motility and cell separation, is shown in light grey to indicate that the interaction is not well understood and is likely to be indirect.

Experimental procedures

Growth media

Luria–Bertani (LB) broth: 1% tryptone (Difco), 0.5% yeast extract (Difco), 0.5% NaCl. MSgg broth: 100 mM morpholinepropane sulphonic acid (MOPS) (pH 7), 0.5% glycerol, 0.5% glutamate, 5 mM potassium phosphate (pH 7), 50 µg ml−1 tryptophan, 50 µg ml−1 phenylalanine, 2 mM MgCl2, 700 µM CaCl2, 50 µM FeCl3, 50 µM MnCl2, 2 µM thiamine, 1 µM ZnCl2 (Branda et al., 2001). Media were solidified through addition of agar (Difco) to 1.5%; plates were allowed to dry at room temperature for 40 h before use. Antibiotic concentrations (final): MLS (1 µg ml−1 erythromycin, 25 µg ml−1 lincomycin); spectinomycin (100 µg ml−1); tetracycline (500 µg ml−1 for broth, 10 µg ml−1 for agar); chloramphenicol (5 µg ml−1).

Strains and mutant construction

All strains used in this study are listed in Table S1. Deletion of genes from the 168 chromosome, and their replacement by antibiotic resistance markers, was achieved through a long-flanking-homology PCR strategy (Wach, 1996). Primers used to create gene-replacement constructs are as follows (5′ to 3′): tasA::spc-1 (gtgagaaacttaagaaatcaaagtgg); tasA::spc-2 (acatgtattcacgaacgaaaatcgactgcagaagcaactc ctaaactc); tasA::spc-3 (attttagaaaacaataaacccttgcacatagc gaggataaaaattaataacagc); sipWop::spc-4 (ccgtccgactgggcgg aac); yve-yvf::tet-1 (cattgacatacaagcaatcctcgg); yve-yvf::tet-2 (gaacaacctgcaccattgcaagactcatattctcattcatgtattcatag); yve-yvf::tet-3 (ttgatcctttttttataacaggaattcccttcctgctcacatgtgagc); yve-yvf::tet-4 (ctgcggatgcagatcgatctc). Gene-replacement mutations were transferred from the 168 genetic background to PY79 by transformation (Harwood and Cutting, 1990), and to NCIB 3610 by SPP1-mediated generalized transduction (Yasbin and Young, 1974).

Assay for formation of solid surface-associated biofilms

Bacteria were grown on LB agar at 37°C overnight, resuspended in LB broth to an OD600 of ∼0.02, and grown at 37°C with agitation for ∼5 h, to an OD600 of ∼1.0. An aliquot of each culture was diluted 500-fold in MSgg broth + 200 µM NaCl, and 300 µl of the dilution were added to each of 12 sterile borosilicate glass tubes (10 × 75 mm; from VWR Scientific). Tube openings were covered with sterile tinfoil, and the cultures grown at 37°C without agitation for 60 h. At this point 500 µl of 1% crystal violet were added to each tube. After 10–15 min at 25°C without agitation, the tubes were drained of liquid via pipet, gently rinsed several times with water, and allowed to dry at 25°C overnight. For quantification, biofilm-associated crystal violet was solubilized through addition of 1 ml of 33% acetic acid and 10 min of intermittant vortexing at 25°C, and a sample diluted fivefold in 33% acetic acid for measurement of OD563 (readings of 0.1–1.6 were within the linear range). The 12 readings were averaged; one standard deviation typically represented < 15% of the mean. The assay was performed on three separate occasions. The means from each independent assay were averaged, and the standard deviation from this average value was calculated. The statistical significance of differences in average values was determined through analysis of variance (anova), using WINKS software (TexaSoft). Note that in contrast to common laboratory strains of B. subtilis, our wild strain does not appear to produce solid surface-associated biofilms (our unpublished observations), and therefore we were unable to determine whether mutation of tasA in that genetic background has any effect on solid surface-associated biofilm formation.

Analysis of colony and pellicle formation, and extracellular complementation

Bacteria were grown on LB agar and then in LB broth to an OD600 of ∼1.0, as described above. For colony architecture analysis, 3 µl of starting culture were spotted onto MSgg agar and the plates incubated at 30°C for 42–84 h, as indicated. For pellicle formation analysis, 12 µl of starting culture were added to 12 ml of MSgg broth contained within a well of a 6 well microtiter dish, and the dish incubated without agitation at 25°C for 60 h.

To test extracellular complementation, starting cultures of appropriate mutants were mixed at fixed ratios (1:1000–1000:1 in 10-fold increments), and the mixtures were used to form colonies and pellicles, as described above. For all mutant combinations tested we found that mixture ratios of 1:10–10:1 generated qualitatively similar results; thus, representative results from the 1:1 mixtures are presented. The final ratio of strains in the mature colony or pellicle was determined at the conclusion of each experiment, through measurement of colony-forming units on selective media; we found that in all cases the final ratio roughly corresponded to the starting ratio, and that the recovered strains displayed their original mutant phenotypes.

Light microscopy and photography

Whole colonies were photographed at low magnification (5×) using a Stemi SV6 stereomicroscope (Zeiss) equipped with a 0.63× ApoPlan objective lens, an AxioCam charge-coupled device (CCD) video camera system (Zeiss), and a computer interface. Fruiting bodies along the peripheral edges of the colonies were photographed at higher magnification (32×) using the same stereomicroscope and camera. Whole pellicles were photographed using a digital camera equipped with a close-up lens (Nikon).

To analyse the behaviour of individual cells growing in agitated MSgg broth, starting cultures of OD600 of ∼1.0 (see above) were diluted 1000-fold with MSgg, 3 ml aliquots were transferred to 18 ml glass tubes, and the tubes were incubated with agitation at 30°C for 12–18 h (final OD600 of ∼1.5). Samples of these cultures were examined at high magnification (600×) using an Optiphot-2 phase-contrast microscope (Nikon) equipped with a CCD video camera system (Optronics Engineering, Goleta CA) and a computer interface.

Scanning electron microscopy (SEM) of pellicles

Pellicles were grown as described above, except at an incubation temperature of 30°C. Pellicle sections were carefully transferred to ∼10 mm diameter pieces of polydimethylsiloxane (PDMS) membrane, and then fixed in 4% paraformaldehyde, 2.5% glutaraldehyde, 87 mM sodium cacodylate (pH 7.4), at 25°C for 2 h. After three 15 min washes with 174 mM sodium cacodylate (pH 7.4) and dehydration through a graded series of ethanol, the samples were infiltrated with hexamethyldisilazane (HMDS), through one incubation in 50% HMDS (in 100% ethanol) at 25°C for 1 h and then two in 100% HMDS at 25°C for 30 min. At this point the PDMS-bound samples were mounted on pins, dried under vacuum overnight, sputter-coated with gold–palladium alloy, and examined by SEM.

Pellicle fractionation and Western analysis

Bacteria were grown in LB broth at 37°C to an OD600 of ∼1.0, as described above. A 10 µl of sample from this starting culture was added to 10 ml of MSgg broth contained within a 6 well plate. All MSgg cultures were incubated at 22°C without agitation for 72 h.

Culture pellicles were gently lifted from the surface of the medium and placed in 15 ml Falcon tubes; the conditioned medium (‘Medium’ fraction) was filter-sterilized (0.2 µm pore size) and stored at 4°C. In the case of tasA and yqxM samples, the entire culture was centrifuged at 9000 g for 20 min, and the supernatant was (‘Medium’ fraction) filter-sterilized and stored at 4°C. MS (MSgg broth minus the glycerol and glutamate, which were replaced with water) was added to the pellet tubes (10 ml per tube) to resuspend the cells.

The cells samples were subjected to mild sonication (two rounds of 12 one-second pulses at 20% power, using an UltraSonics Sonicator cell disrupter equipped with a small-tip probe). Phase-contrast microscopy of the samples revealed that this treatment thoroughly disrupted pellicles and cell clusters, mainly freeing single cells; only ∼5% of the vegetative cells were lysed (‘ghost cells’), while spores accounted for 10–20% of the total cell population (except in sigE samples).

Sonicated samples were then centrifuged at 9000 g at 4°C for 20 min. In each case the pellet (‘Cell’ fraction) was resuspended in 10 ml MS while the supernatant (‘Matrix’ fraction) was filter-sterilized, and both sets of fractions were stored at 4°C.

Once all of the fractions had been collected and cooled to 4°C, 2 ml of each ‘Cell’ fraction was treated with 100 µg ml−1 lysozyme for 15 min at 37°C. After the cells had been treated with lysozyme, 2 ml of the ‘Medium’, ‘Matrix’ and ‘Cell’ fractions were treated with trichloroacetic acid to a final concentration of 8%. Precipitated proteins were harvested by centrifugation at ∼13 000 g for 20 min, washed twice with cold acetone, dried under vacuum, and solubilized in 1× SDS Gel-loading Buffer (100 mM Tris-HCl pH 6.8, 4% SDS, 0.2% bromophenol blue, 20% glycerol, 200 mM 2-mercaptomethanol). Samples were run on a 12% polyacrylamide gel and electrotransferred to polyvinylidene fluoride membrane (Millipore). The proteins were immunoblotted with anti-TasA, anti-σA and anti-rabbit horseradish peroxidase conjugate. Anti-TasA was used at a dilution of 1:20 000 and anti-σA was used at a dilution of 1:10 000. The antibodies utilized were obtained from: anti-TasA, A. Driks (Stover and Driks, 1999a); anti-σA, M. Fujita and D. Rudner. The chemiluminescent signal was visualized on Kodak BioMax Film.


We thank R. Stearns and M. Ericsson for help with electron microscopy; C. Ellermeier and K. Ramamurthi for help with Western blots; A. Driks, D. Rudner and M. Fujita for generous gifts of antibodies; A. Driks, H. Misono and J. Dworkin for generous gifts of strains and plasmids; and A. Earl, P. Straight and H. Vlamakis for helpful discussions and critical reading of the manuscript. This work was supported by grants from the NIH GM58213 to R.K and GM18568 to R.L., and postdoctoral fellowships from the Charles A. King Trust (Bank of America, Boston MA) to S.S.B and the NIH (GM66612) to D.B.K.