The sporulation transcription factor Spo0A is required for biofilm development in Bacillus subtilis

Abstract

Biofilms are structured communities of cells encased in a polymeric matrix and adherent to a surface, interface or each other. We report here that the soil bacterium Bacillus subtilis forms biofilms. By confocal scanning laser microscopy, we observed that B. subtilis adhered to abiotic surfaces and formed a three-dimensional structure ≥ 30 µm in depth. These biofilms appeared to be at least partly encased in an extracellular polysaccharide matrix, as they could be stained with Calcofluor, a polysaccharide-binding dye. To understand the molecular mechanism of biofilm formation, we screened previously characterized mutants for a defect in biofilm formation. We found that mutations in spo0A, which encodes the major early sporulation transcription factor, caused a defect in biofilm formation. spo0A mutant cells adhered to a surface in a monolayer of cells rather than a three-dimensional biofilm. The requirement of Spo0A for biofilm development appears to result from its role in negatively regulating AbrB. Mutations in abrB suppressed the biofilm defect of a spo0A mutant, indicating that AbrB negatively regulates at least one gene that is required for the transition from a monolayer of attached cells to a mature biofilm. Implications of biofilm development for the ecology of B. subtilis are discussed.

Introduction

Many bacterial species exhibit two modes of growth, the free-floating planktonic mode and the sessile biofilm mode. It is the biofilm mode of growth that has been proposed to be the major bacterial life style in nature (Costerton et al., 1995). Biofilms are structured communities of microbial cells living adherent to a surface, interface or each other and encased in a self-produced polymeric matrix (Costerton et al., 1995; Davey and O'Toole, 2000). Biofilm formation has been proposed to be a developmental process that involves large changes in gene expression (Prigent-Combaret et al., 1999;O'Toole et al., 2000). Isolation of mutants defective in biofilm development in several genera of bacteria have begun to reveal some of the gene products that are involved in biofilm formation; these include motility, cell surface structures and exopolysaccharides (EPS) (Pratt and Kolter, 1999; Davey and O'Toole, 2000). Furthermore, studies from at least one bacterium, Escherichia coli, indicate that ≈ 38% of the genes in the genome are differentially expressed in biofilm-grown cells, indicating that there are many functions involved in biofilm formation that remain unknown (Prigent-Combaret et al., 1999;Kuchma and O'Toole, 2000). We have chosen to study the molecular mechanism of biofilm formation in the Gram-positive soil bacterium, Bacillus subtilis.

Biofilm formation has largely been studied in bacteria that live in aquatic environments, where biofilm formation provides several benefits to these bacteria (Costerton et al., 1995; Davey and O'Toole, 2000). Organic nutrients will concentrate on surfaces (Costerton et al., 1995) and, thus, biofilm formation on these surfaces is a mechanism for obtaining nutrients. The extracellular polymeric matrix of biofilms, which is often composed of EPS, may also serve to concentrate nutrients through absorption (Wolfaardt et al., 1998). In addition, the extracellular polymeric matrix may protect biofilm cells from environmental stresses, such as osmotic shock and pH shifts (Flemming, 1993;Davey and O'Toole, 2000).

Although biofilms appear to be the major form of bacterial growth in aquatic environments (Costerton et al., 1995), it is not known what role biofilms play for bacteria growing in terrestrial environments. Bacteria in terrestrial environments face widely fluctuating conditions, including changes in temperature, pH and moisture, and biofilm growth would appear to be an excellent strategy to protect soil bacteria from some of these stress factors (Davey and O'Toole, 2000). However, the biofilm-forming properties of very few soil bacteria have been studied. One soil bacterium that has been shown to form biofilms is Pseudomonas fluorescens, which can form biofilms on plant roots (O'Toole and Kolter, 1998; Biancotto et al., 2001). It has been suggested that the structured community of microbial cells that colonizes the rhizosphere may be considered a biofilm (Anellal et al., 1998). It is important to understand the physiology of bacteria growing in the soil and the rhizosphere, as bacteria in this environment carry out important activities, such as bioremediation, stimulating plant growth and protection of plants from pathogens (van Veen et al., 1997).

We are studying the molecular mechanism of biofilm development in the ubiquitous soil bacterium, B. subtilis. B. subtilis can promote plant growth and protect against pathogens (Utkhede and Smith, 1992; Aska and Shoda, 1996; Podile and Prakash, 1996; Emmert and Handelsman, 1999). B. subtilis is also likely to have an important role in the degradation of organic polymers in the soil. Sequencing of the B. subtilis genome revealed a number of genes involved in the degradation of plant products, including cellulose, hemicellulose and pectin (Kunst et al., 1997). Studying the signals that trigger the transition of B. subtilis from a planktonic to a biofilm mode of growth may enhance our understanding of the role that biofilms play in the ecology of this important soil bacterium. B. subtilis has been a model organism for the study of Gram-positive bacterial physiology and, thus, the study of biofilm development in B. subtilis should advance our knowledge of the mechanisms of biofilm formation in Gram-positive bacteria.

In this report, we provide evidence that B. subtilis forms biofilms. We also report on the identification of two transcription factors, Spo0A and AbrB, that regulate the formation of biofilms. One of these transcription factors, Spo0A, is also required for sporulation; however, sporulation itself is not required for biofilm development. Our results suggest that some of the signals that regulate sporulation may also regulate biofilm development in B. subtilis.

Results

Adherence of B. subtilis to abiotic surfaces

Biofilms are structured communities of cells adherent to a surface and encased in an extracellular polymeric matrix, and the first step in biofilm formation is generally adherence to a surface (Watnick and Kolter, 2000). To determine whether B. subtilis is capable of forming biofilms, we determined whether B. subtilis would adhere to the wells of a microtitre plate, using a modified version of the procedure described by O'Toole et al. (1999). B. subtilis was incubated in the wells of a polyvinylchloride microtitre plate containing biofilm growth medium (described in Experimental procedures). After 24 h, the growth medium and unattached cells were removed, and the cells that remained attached were visualized by staining with crystal violet (CV). A ring of stained cells was attached to the wells at what had been the air to medium interface (Fig. 1A). This strain of B. subtilis adhered to borosilicate glass and polypropylene surfaces, but not to polystyrene (data not shown).

Figure 1. Microtitre plate assay of biofilm formation by wild-type and mutant B. subtilis strains.

A. Photograph of the wells of a microtitre plate after staining adherent BAL218 (wild-type) biofilms with crystal violet.

B. OD₅₇₀ of solubilized crystal violet from e plate assay over time for different mutant strains. (Closed squares) BAL218 (WT); (closed triangles) BAL359 (spo0A); (closed diamonds) BAL373 (abrB); (closed circles) BAL663 (spo0A abrB). Each data point is an average of 16 wells, and error bars indicate the standard error. Representative data from one of at least five independent experiments are shown.

To characterize the growth conditions under which B. subtilis forms biofilms, a variety of growth media was tested for the ability to support adherence to microtitre plates. Growth in an LB medium or minimal medium with glucose, succinate or glycerol as a carbon source did not support reproducible adherence to the microtitre plates (data not shown). However, if LB medium was supplemented with 0.1% glucose, 1 mM MgSO₄, 0.15 M ammonium sulphate and 34 mM citrate and the medium buffered to pH 7 (i.e. biofilm growth medium), reproducible adherence of B. subtilis to the microtitre plates was observed (Fig. 1A). Elimination of glucose, MgSO₄ or the buffer from the biofilm growth medium resulted in reduced and non-reproducible biofilm formation (data not shown). These data suggest that this strain of B. subtilis has rather strict growth medium requirements for biofilm formation.

Biofilms of B. subtilis were also only observed if the cells were incubated under relatively static conditions (i.e. mixing once every 12 h). If cells were grown in biofilm growth medium in flasks with shaking at 200 r.p.m., the cells grew with a faster doubling time (2.4 doublings h⁻¹ versus 0.5 doublings h⁻¹ under e plate conditions) and reached a higher OD₆₀₀ (5 versus 0.5 under microtitre plate conditions), but did not form biofilms over the period of 48 h (data not shown). These data suggest that, under microtitre plate conditions, the cells may be oxygen limited and that this could be a signal for biofilm formation.

The time course of adherence of B. subtilis cells to e plates was determined. CV associated with the adhered cells was solubilized and quantified by measuring absorbance at 570 nm (see Experimental procedures). Measurable levels of adhered cells were observed 24 h after inoculation (Fig. 1B). This rate of adherence was significantly different from the rate of growth of the cells. Measurement of OD₆₀₀ of the microtitre plate wells over time indicated that growth ceased after 4 h of incubation (data not shown). The long delay between the time of entry into stationary phase and adherence to a surface suggests that the cells undergo adaptation in order to adhere.

The levels of adhered cells increased eightfold between the period of 24–48 h of incubation (Fig. 1B). This increase in CV staining appeared to result from an increase in the number of adhered cells, as this corresponded to an increase in the number of colony-forming units (cfu) of adhered cells (data not shown). This increase in the level of adhered cells also appeared to result from division of these adhered cells rather than recruitment of planktonic cells, as this increase was observed when planktonic cells were removed and fresh medium added. For reasons that are unknown, no additional increase in the level of biofilm formation was observed at later time points.

Confocal scanning laser microscopy of B. subtilis cells adhered to an abiotic surface

To determine whether the cells that were adhering to the microtitre plates were actually forming a three-dimensional biofilm, adherent B. subtilis cells were analysed by confocal scanning laser microscopy (CSLM). CSLM scans in both the X–Y and the X–Z planes and, therefore, can be used to view a three-dimensional structure. In order to view adhered B. subtilis cells by CSLM, a B. subtilis strain that had a multicopy plasmid carrying the gene for the green fluorescent protein (GFP) was grown adhered to glass coverslides. These coverslides were placed on microscope slides with concave depressions filled with buffer, such that the structure of the adhered cells would not be disrupted.

Adhered B. subtilis cells appeared to form a three-dimensional, multicellular structure typical of a biofilm (Fig. 2A and B). Viewing the cells from the X–Y plane (i.e. from the top looking perpendicular to the surface), it appeared that the adhered cells formed a mat of cells. From the X–Z plane (i.e. from the side looking horizontal to the surface), it was observed that the adhered cells formed a structure with significant depth. This depth ranged from ≈ 30 µm after 48 h of incubation (Fig. 2A) to ≈ 60 µm after 72 h of incubation (data not shown).

Figure 2. CSLM analysis of wild-type and spo0A mutant strains of B. subtilis expressing GFP. Biofilms of cells expressing GFP from a multicopy plasmid were grown on the surface of glass coverslides and then analysed by CSLM.

A, C and E. Standard projections of the biofilms; each image is a compilation of 25 sections through the X–Y plane. Scale bars represent 20 µm.

B, D and F. A single section through the X–Z plane. Scale bars represent 10 µm. A representative image of those obtained on at least three independent occasions is shown.

A and B. BAL682 (wild-type) biofilms grown for 48 h.

C and D. BAL685 (spo0A) biofilms grown for 72 h.

E and F. BAL690 (spo0A abrB) biofilms grown for 48 h.

Staining of adhered B. subtilis cells with a polysaccharide-binding dye

An additional definition of a biofilm is that the cells are encased in an extracellular polymeric matrix (Davey and O'Toole, 2000). This matrix is often composed, at least in part, of exopolysaccharides (Davey and O'Toole, 2000). Calcofluor, a polysaccharide-binding dye (Wood, 1980), has been used to stain the extracellular matrix of biofilms formed by other bacteria (Stewart et al., 1995). Therefore, to determine whether the adhered structures of B. subtilis were encased in a polysaccharide matrix, we stained adherent B. subtilis cells that lack GFP with Calcofluor and then observed these cells by CSLM.

CSLM of the X–Y plane of these multicellular structures of B. subtilis revealed Calcofluor staining (Fig. 3A). An analysis of the X–Z plane revealed that not only the surface of the structures was stained, but that there was staining throughout the depth of the structure (Fig. 3B). To determine whether the ability of these cells to bind Calcofluor was a property of growing in this adherent structure, planktonic cells were stained and observed by CSLM. Actively growing planktonic cells do not stain with Calcofluor (data not shown), indicating that production of the polysaccharide is a unique property of B. subtilis cells in an adherent, three-dimensional structure. These Calcofluor staining data, along with the CSLM data from the GFP-expressing cells, strongly indicate that B. subtilis forms biofilms that are at least partially encased in an extracellular polysaccharide matrix.

Figure 3. Staining of B. subtilis biofilms with a polysaccharide-binding dye. BAL218 (wild-type, i.e. lacking GFP) biofilms were grown adherent to glass coverslides for 48 h. Calcofluor-stained cells were viewed by CSLM. Each image is of a single section though the biofilm.

A. X–Y plane.

B. X–Z plane.

Biofilm formation by a spo0A mutant of B. subtilis

To determine the molecular mechanism of biofilm development in B. subtilis, we sought to identify genes that are involved in biofilm formation. Mutants of previously characterized genes that have a role in cell–cell signalling were tested for a defect in biofilm formation in the microtitre plate assay. We postulated that, because these structures are multicellular, cell–cell signalling may play a role in biofilm development in B. subtilis, as has been shown for biofilm development in Pseudomonas aeruginosa (Davies et al., 1998). All the mutants tested, comA, comP, comQ, opp, phrA, phrC, phrE, rapC, rapB and spo0A (Magnuson et al., 1994; Perego and Hoch, 1996;Solomon et al., 1996), are defective in the production of, or response to, an extracellular signalling peptide. Only one mutant had a significant effect on biofilm formation, spo0A, which is defective in the major early sporulation transcription factor (data not shown).

The level of spo0A mutant cells that adhered to microtitre plates was quantified (Fig. 1B). At all time points tested, the spo0A mutant exhibited a level of CV staining that did not exceed an OD₅₇₀ of 1. This is significantly below the level of CV staining observed with wild-type cells, which typically ranged from an OD₅₇₀ of 0.8–1.5 at the 24 h time point to 3–4.5 at the 60 h time point. This apparent defect in adherence of the spo0A mutant was not caused by lack of growth of the spo0A mutant strain. The spo0A mutant and the wild-type strain reached similar OD₆₀₀ levels ≈ 4 h after inoculation (data not shown). The defect in adherence of the spo0A mutant also does not appear to result from lack of survival of the spo0A mutant cells. The total number of cells present in the wells of the microtitre plates were counted as cfu, and similar numbers were observed for the spo0A and wild-type strains after 24, 48 and 72 h of incubation (data not shown). These data indicate that spo0A mutant cells are defective in biofilm formation.

To determine the step in the biofilm development pathway at which the spo0A mutant was blocked, GFP-expressing, spo0A mutant cells that adhered to a surface were examined by CSLM. Sections through the X–Y plane revealed that spo0A mutant cells were able to adhere to a glass surface (Fig. 2C). However, these cells did not appear to be forming the three-dimensional structure of the wild-type cells. Sections of spo0A mutant cells through the X–Z plane revealed that only a monolayer of cells ≈ 2.5 µm in depth had adhered to the surface (Fig. 2D). Smaller but similar monolayers of wild-type cells were also observed at very early time points (i.e. < 24 h; data not shown). These data indicate that the transcription factor Spo0A must regulate the expression of at least one gene that is required for cells to transition from a monolayer to a mature biofilm.

Biofilm formation by mutants defective in the Spo0A phosphorelay

Spo0A is a response regulator protein that is required for the transcription of genes required early in sporulation (Grossman, 1995). Spo0A is activated by phosphorylation through a phosphorelay, which transfers phosphate from one of several histidine protein kinases (KinA, KinB, KinC, KinD or KinE) to a response regulator protein, Spo0F, and then to Spo0A through a phosphotransferase protein, Spo0B (Burbulys et al., 1991; Jiang et al., 2000). To determine whether, under biofilm growth conditions, Spo0A was being activated by phosphorylation through the same phosphorelay as under sporulation conditions, mutants that are defective in the phosphorelay proteins, Spo0F and Spo0B, were tested for a defect in biofilm formation using the microtitre plate assay (Table 1). At all time points tested, the spo0F and spo0B mutant cells exhibited a level of CV staining that did not exceed an OD₅₇₀ of 0.8. This is significantly below the level of CV staining observed with wild-type cells, but similar to the level of CV staining observed with the spo0A mutant cells. The biofilm defect of the spo0F and spo0B mutants was not caused by a defect in the growth rate of these cells or by a defect in the survival of these cells (data not shown). These data indicate that the Spo0A phosphorelay proteins, Spo0F and Spo0B, are required for biofilm formation.

Table 1.   Biofilm formation by sporulation mutants and Spo0A phosphorelay mutants.

Strain^a	Biofilm formation^b (normalized to wild type)
a . Relevant genotypes are indicated in parenthesis. b. Biofilm formation by the indicated strains was quantified using the microtitre plate assay after growth of the biofilms for 48 h. The OD570 of the solubilized CV for 24 microtitre plate wells was averaged and then normalized to the wild-type data. Standard errors for each data point were < 10% of the average. Representative data for at least four independent experiments are shown.
BAL 218 (WT)	1.0
BAL 359 (spo0A)	0.22
BAL 369 (spo0B)	0.34
BAL 370 (spo0F)	0.49
BAL 666 (spoIIGB)	0.91
BAL 667 (spoIIAC)	0.83
BAL 344 (kinA)	1.0
BAL 354 (kinB)	1.1
BAL 363 (kinC)	0.45
BAL 691 (kinD)	0.61
BAL 692 (kinE)	0.42

Five kinases are capable of donating phosphate to Spo0F in vitro (Jiang et al., 2000). However, only mutations in kinA and kinB have been reported to have a significant effect on sporulation under laboratory conditions (Trach and Hoch, 1993). To determine whether KinA and KinB were also the major kinases donating phosphate to Spo0A under biofilm growth conditions, the biofilm defect of kinA and kinB mutants was analysed using the microtitre plate assay (Table 1). At all time points analysed, kinA and kinB mutants exhibited similar levels of CV staining compared with the wild-type strain. These data suggest that the major kinases required for sporulation are not required for biofilm formation.

To determine whether the KinC, KinD or KinE kinase was responsible for donating phosphate to the Spo0A phosphorelay under biofilm growth conditions, mutants defective for each of these kinases were tested for a defect in biofilm formation using the microtitre plate assay (Table 1). At all time points, the kinC, kinD and kinE mutants exhibited an ≈ 2.5-fold defect in biofilm development compared with the wild-type strain. This decrease in the level of CV staining for the kinC, kinD and kinE mutants does not appear to result from differences in the growth of these mutants or from differences in the number of viable cells (data not shown). This indicates that KinC, KinD and KinE are the kinases that donate phosphate to the Spo0A phosphorelay during biofilm development.

Biofilm formation by mutants defective in stage II sporulation sigma factors

Spo0A is required for sporulation as a result of its role in activating stage II sporulation genes (Stragier and Losick, 1996). To determine whether the defect in biofilm formation of a spo0A mutant was caused by a defect in the ability to form spores, spoIIGB and spoIIAC mutants were analysed for the ability to form biofilms using the microtitre plate assay (Table 1). spoIIGB and spoIIAC encode the mother cell sigma factor, σ^E, and the forespore sigma factor, σ^F, respectively, and require Spo0A for their expression in vivo (Stragier and Losick, 1996). Both the spoIIGB and spoIIAC mutants exhibited similar levels of CV staining to the wild-type strain (Table 1). These data indicate that sporulation is not required for biofilm development.

To determine whether the levels of Spo0A activity were sufficient to support sporulation under biofilm growth conditions, we quantified the percentage of cells in the biofilm that were heat-resistant spores. Biofilms of the wild-type strain were grown in microtitre plates and, after removing unadhered cells, the biofilm cells were resuspended, and the percentage of cfu that were heat-resistant spores was determined (see Experimental procedures). Approximately 0.1% (standard deviation = 0.1) of the biofilm cells formed spores under these growth conditions. This may be an overestimate of the percentage of spores, as the number of total cfus is probably low because of some clumping of the cells. This degree of sporulation is similar to the percentage of planktonic cells that sporulate under these conditions. These data indicate that a small percentage of the cells in a biofilm sporulate and that sporulation is not required for biofilm formation.

Biofilm formation by mutants defective for the AbrB transcription factor

One function of Spo0A is to repress the transcription of AbrB (Strauch et al., 1990), a transcriptional repressor of several genes, many of which are not required for sporulation (Strauch and Hoch, 1993). To determine whether the biofilm defect of the spo0A mutant resulted from lack of repression of abrB, a spo0A abrB double mutant was analysed for biofilm formation using the microtitre plate assay (Fig. 1). When compared with the spo0A mutant, the spo0A abrB mutant exhibited significantly higher levels of CV staining. The spo0A abrB mutant exhibited only a 1.5-fold defect in biofilm formation compared with the wild-type strain. This indicates that AbrB has a negative role in regulating biofilm formation, and the defect in biofilm formation of a spo0A mutant can, in large part, be explained by its role in repressing th transcription of abrB.

To determine whether the biofilm formed by the spo0A abrB mutant was a three-dimensional structure similar to the wild-type strain, a GFP-expressing, spo0A abrB mutant biofilm was analysed by CSLM. Observation of the biofilm in both the X–Y plane (Fig. 2E) and in the X–Z plane (Fig. 2F) indicated that the spo0A abrB mutant formed a three-dimensional biofilm. The spo0A abrB mutant formed biofilms that were ≈ 30 µm in depth after 24 h of growth, similar to the wild-type strain. This is contrast to the spo0A mutant, which only formed a monolayer of attached cells.

Biofilm formation by the abrB single mutant was also analysed. In the microtitre plate assay, the abrB mutant had a similar level of CV staining to the wild-type strain, which was 1.5-fold higher than that observed with the spo0A abrB mutant strain (Fig. 1). The abrB mutant biofilm was also analysed by CSLM, and the abrB mutant biofilm appeared to be indistinguishable from the wild-type strain (data not shown). That the abrB mutant shows a higher level of biofilm formation than the spo0A abrB mutant in the microtitre plate assay indicates that spo0A may have a role in regulating biofilm formation in addition to regulating abrB.

Discussion

Bacillus subtilis biofilm formation

Biofilm formation is accepted as the major bacterial lifestyle in nature among microorganisms that live in aquatic environments. The lifestyle of microorganisms living in soil is less well understood. The experiments presented in this study provide some of the first evidence that B. subtilis, a ubiquitous soil bacterium, forms biofilms. CLSM of B. subtilis cells adhered to a surface revealed a multicellular, three-dimensional structure, characteristic of a biofilm. These biofilms appeared to be at least partially encased in a polysaccharide matrix, as these biofilms stained with a polysaccharide-binding dye. Exponentially growing, planktonic cells did not stain with this polysaccharide-binding dye, indicating that this matrix is synthesized during the transition from individual planktonic cells to a mature biofilm. We have found that B. subtilis will adhere and form a biofilm on a variety of surfaces in the laboratory. We have not yet tested whether B. subtilis adheres to the surfaces that are normally encountered in the soil. However, the evidence that B. subtilis does form biofilms suggests that biofilms may play a role under some conditions in the growth and survival of B. subtilis in the soil.

Role of the sporulation transcription factor Spo0A in biofilm formation

Biofilm formation has been proposed to be a developmental process involving large changes in gene expression (O'Toole et al., 2000). Consistent with this, we identified two transcription factors that regulate biofilm formation in B. subtilis. One of these is the early sporulation transcription factor, Spo0A. spo0A mutant cells were able to adhere to a surface, but CSLM analysis revealed that these mutant cells attached only as monolayers. This suggests that the spo0A mutant may be defective in the cell–cell interactions that are necessary to form a multicellular biofilm. The genes regulated by Spo0A that are required for these intercellular interactions are not known. Gene products have been identified in other bacterial species that are involved in intercellular adhesion; these include the Ica proteins of staphylococci that synthesize β-1,6-linked glucosaminoglycan (Cramton et al., 1999) and the outer membrane protein, Antigen 43, of Escherichia coli (Danese et al., 2000). Further study of B. subtilis biofilm formation is necessary to determine the type of gene products involved in intercellular adhesion in this bacterium.

The other transcription factor identified in these studies as regulating biofilm formation in B. subtilis is AbrB. AbrB negatively regulates biofilm formation, and the biofilm defect of a spo0A mutant was largely suppressed by deleting abrB. This suggests that AbrB may repress the expression of genes whose products are involved in intracellular adhesion. However, the rate of biofilm formation by an abrB mutant was not faster than that of a wild-type strain, indicating that there may be another signalling pathway that prevents inappropriate expression of these AbrB-repressed genes. That the biofilm defect of spo0A could be suppressed by a mutation in abrB is consistent with the observation that Spo0A binds to the promoter region and represses the transcription of abrB (Strauch et al., 1990). Thus, it appears that the major role of spo0A in biofilm formation is to regulate abrB negatively.

Spo0A may have an additional role in biofilm formation beyond repressing abrB expression. This is suggested by the different phenotype of an abrB mutant versus a spo0A abrB mutant in the microtitre plate assay. The abrB mutant consistently showed a higher level of biofilm formation than the spo0A abrB mutant. The reason for this difference is not known. It may be due to an increase in the percentage of the surface area colonized by an abrB mutant. Development of assays to quantify surface area coverage will be necessary to address this model. That the spo0A abrB mutant does not appear to have the same biofilm properties as the abrB mutant suggests that Spo0A may regulate the expression of at least a gene, other than abrB, that contributes to biofilm development in B. subtilis.

Role of the Spo0A phosphorelay in biofilm formation

Spo0A is a response regulator that is activated by phosphorylation through a multicomponent phosphorelay (Burbulys et al., 1991). This phosphorelay appears to be required to activate Spo0A for biofilm formation. Mutants defective for Spo0F, the response regulator that receives phosphate from the kinases, or Spo0B, the phosphotransfer protein that transfers phosphate from Spo0F to Spo0A, had a similar defect in biofilm formation to a mutant defective for Spo0A. This phosphorelay serves to integrate multiple signals for the initiation of sporulation (Ireton and Grossman, 1992; 1994;Ireton et al., 1993). The major signal that appears to activate this phosphorelay for sporulation is starvation, but this phosphorelay can be modulated by signals arising from quorum sensing, DNA damage, chromosome replication and partitioning and the TCA cycle (Grossman, 1995). It will be interesting to determine whether these same signals that affect Spo0A activity for sporulation affect Spo0A for biofilm development.

Evidence that different signals may activate Spo0A for biofilm development comes from the observation that different kinases were required for biofilm development compared with sporulation. Five kinases have been identified that are capable of donating phosphate to Spo0F (Jiang et al., 2000). However, in vivo, only KinA and KinB have been demonstrated to have a large effect on sporulation, and the physiological significance of KinC, KinD and KinE donating phosphate to Spo0A has been unclear (Trach and Hoch, 1993). The studies presented here suggest that the major function of the KinC, KinD and KinE kinases is to activate Spo0A for biofilm formation. Mutants defective for KinC, KinD or KinE are not as defective as a spo0A mutant, suggesting that these kinases are partially redundant and that deletion of the genes for all three kinases would be necessary to observe a phenotype similar to a spo0A mutant. The reason for the lack of an affect of KinA and KinB on Spo0A activity under biofilm formation conditions is not understood. Each kinase appears to be subject to a different environmental and/or metabolic signal (Jiang et al., 2000), but the molecular natures of these signals are unknown. It may be that, under our biofilm growth conditions, the signals that activate KinA and KinB are not present.

That KinC, KinD and KinE have a more significant role in biofilm development than KinA or KinB is consistent with the previous observation that KinC and KinD also had a more significant role in regulating abrB expression (Jiang et al., 2000). In planktonic cells, abrB expression is highest during exponential growth and then decreases during entry into stationary phase (Strauch et al., 1989; Furbass et al., 1991;O'Reilly and Devine, 1997). The major kinases required for this proper temporal control of abrB are KinC and KinD (Jiang et al., 2000). kinA and kinB mutants do not have a large effect on abrB expression (Jiang et al., 2000), presumably because KinA and KinB only become active at a later time when abrB expression has already been maximally repressed through the action of KinC and KinD.

To be a biofilm or to be a spore?

Although both sporulation and biofilm development require Spo0A, these two pathways are genetically separable. Mutants defective in stage II sporulation genes, which are activated by Spo0A and are absolutely required for sporulation, were not defective in biofilm development. Conversely, biofilm development does not appear to be required for sporulation as, under typical laboratory conditions that promote efficient sporulation (i.e. > 90% of the cells form spores), the cells remain free floating and unattached to other cells or to the sides of the culture vessel. However, these pathways are not mutually exclusive; a small percentage of biofilm cells formed spores in this study. It appears that, upon activation of Spo0A, the cells are faced with a decision to proceed directly down the sporulation development pathway or transition to the biofilm development pathway before the decision to form a spore.

How the cell decides to form a spore or a biofilm upon activation of Spo0A is not simply explained by the role of Spo0A in regulating the transcription of abrB. Under both sporulation conditions and biofilm formation conditions, Spo0A represses transcription of abrB. One model for how the cells decide to form biofilms versus spores upon activation of Spo0A proposes that there are additional activator or repressor proteins involved in this process. In this model, there is an activator or repressor that allows the expression of sporulation genes only in the presence of the physiological signal for sporulation, and there is also an activator or repressor that only allows the expression of biofilm genes in the presence of the physiological signal for biofilm formation.

An alternative model for how the cells decide to form biofilms versus spores upon activation of Spo0A proposes that the decision is determined by the level of phosphorylated Spo0A (Spo0A∼P). In this model, the physiological signal for biofilm formation leads to low levels of Spo0A∼P, which are sufficient to activate biofilm genes, and the physiological signal for sporulation leads to high levels of Spo0A∼P, which are necessary to activate stage II sporulation genes. This model is similar to a model proposed previously to explain the temporal regulation of abrB. abrB appears to be one of the first genes that is affected by Spo0A as it becomes phosphorylated and, thus, it has been suggested that lower levels of Spo0A∼P are needed for repression of abrB versus activation of stage II sporulation genes (Trach and Hoch, 1993;Jiang et al., 2000).

This second model suggests that, in going from little or no Spo0A∼P to the high levels that would promote sporulation, the cells would pass through a stage of having low levels of Spo0A∼P. Despite this, there are conditions in which sporulation is efficient and biofilm development does not occur. This apparent incongruence could be explained by proposing that the decision to form a biofilm is determined by the length of time that the cells only receive the physiological signal sufficient for low levels of Spo0A∼P and biofilm development. This is supported by the observation that we do not observe significant levels of biofilm formation until ≈ 20 h after the entry of cells into stationary phase, the time at which we would expect Spo0A to become phosphorylated to at least a low level. Future experiments will be necessary to distinguish these models of how cells decide to form a biofilm or a spore.

Role of biofilm development in the lifestyle of B. subtilis

Why might B. subtilis, a soil bacterium, form a biofilm? Growth in biofilms would appear to be an ideal strategy for bacteria to survive the harsh conditions often encountered in soil. A major challenge that bacteria face in the soil is desiccation. One strategy to deal with this is to form a dormant spore that can survive periods of desiccation. However, sporulation is not induced by desiccation and, therefore, metabolically active bacilli and non-spore-forming genera of bacteria need another strategy to survive desiccation. The extracellular polymeric matrix of bacteria growing in biofilms has been suggested to slow down water loss during desiccation (Goerke et al., 2000). Consistent with this, exopolysaccharide-producing bacteria have been shown to be more resistant to desiccation than their non-exopolysaccharide-producing variants (Ophir and Gutnick, 1994). Thus, we propose that biofilm formation may allow B. subtilis to survive desiccation or other environmental stress while there are sufficient nutrients present to support metabolic activity. Knowing whether B. subtilis in the soil are in a biofilm mode of growth will require the identification of genes specifically induced in biofilms and probing B. subtilis in its natural environment for the expression of these genes.

Bacteria in the soil play many important and useful roles. Therefore, it is important to understand the physiology of bacteria growing in soil environments. B. subtilis may provide an ideal model system to study the physiology of bacteria growing in the soil.

Experimental procedures

Strain and plasmid construction

Bacillus subtilis strains used in this study are listed in Table 2. These strains were constructed by transformation with chromosomal DNA or plasmids using standard protocols (Harwood and Cutting, 1990). All strains are derivatives of the parent strain JH642 and, unless otherwise indicated, contain trpC and pheA mutations (Perego et al., 1988).

Table 2.   B. subtilis strains used in this study.

Strain	Relevant genotype	Reference or source
BAL218 (JH642)	trpC2 pheA1	Perego et al. (1988)
BAL216	abrB::Tn917	Perego et al. (1988)
BAL344	kinA::Tn917?HU19	Sandman et al. (1987)
BAL354	ΔkinB,kapB::phleo	A. D. Grossman
BAL359	Δspo0A475::cat	Grossman et al. (1992)
BAL363	ΔkinC::spec	LeDeaux and Grossman (1995)
BAL369	spo0BΔPst pheA^+	Weir et al. (1991)
BAL370	spo0FΔS	Kawamura and Saito (1983)
BAL373	ΔabrB::cat	A.D. Grossman
BAL663	abrB::Tn917Δspo0A475::cat	This study
BAL666	ΔspoIIGB::erm	Kenney and Moran (1987)
BAL667	spoIIAC1	Piggot (1973)
BAL679	spo0A(D56N)-cat::spec	Lazazzera et al. (1999)
BAL682	pBL95 (Pspachy-gfpmut2)	This study
BAL685	spo0A(D56N)-spec, pBL95	This study
BAL 690	spo0A(D56N)-spec, ΔabrB::cat, pBL95	This study
BAL691	kinD::pBL96	This study
BAL692	kinE::pBL97	This study

To view B. subtilis cells by laser microscopy, GFPmut2 (a GFP variant that fluoresces more intensely; Cormack et al., 1996) was introduced into B. subtilis and expressed from a Pspachy promoter on a multicopy plasmid. Pspachy is a strong, IPTG-inducible promoter (Levin et al., 2001). To place gfpmut2 under the control of Pspachy, gfpmut2 was subcloned from pKL147 (Lemon and Grossman, 1998) into pPL82 (Levin et al., 2001). pKL147 was digested with XbaI and SphI, and the gfpmut2 fragment was ligated into the multiple cloning site of pPL82 digested with the same enzymes to generate pBL93. Pspachy-gfpmut2 was then subcloned into a multicopy vector, pHP13, that replicates in B. subtilis (Haima et al., 1987;Bron, 1990). This was accomplished by digesting pBL93 with BamHI and EcoRI and cloning the Pspachy-gfpmut2 fragment into the corresponding sites of the multiple cloning site of pHP13 to create pBL95. pBL95 was transformed into B. subtilis strains by selecting for erythromycin resistance associated with the plasmid. Strains containing pBL95 express Pspachy-gfpmut2 constitutively, as these strains do not contain Lac repressor, which represses the transcription of Pspachy.

kinD and kinE were disrupted by integrating a plasmid containing an internal portion of the corresponding gene via single cross-over. To disrupt kinD, a 477 bp fragment, starting at base 699 relative to the A of the ATG start codon, was polymerase chain reaction (PCR) amplified and cloned into pGEM-cat (Youngman et al., 1989) to generate pBL96. To disrupt kinE, a 534 bp fragment, starting at base 690 relative to the A of the ATG start codon, was also PCR amplified and cloned into pGEM-cat (Youngman et al., 1989) to generate pBL97. kinD or kinE mutants that had pBL96 or pBL97 integrated at their respective genes were selected based on the chloramphenicol resistance associated with the plasmid.

Microtitre plate assay of B. subtilis biofilm formation

Bacillus subtilis biofilm formation was monitored using a modified version of the microtitre plate assay as described by O'Toole et al. (1999). B. subtilis cells were grown in 96-well polyvinylchloride (PVC) microtitre plates (Fisher Scientific) at 37°C in biofilm growth medium. Biofilm growth medium is Luria–Bertani (LB) medium plus 0.15 M ammonium sulphate, 100 mM potassium phosphate, pH 7, 34 mM sodium citrate, 1 mM MgSO₄ and 0.1% glucose. The inoculum for the microtitre plates was obtained by growing the cells in biofilm growth medium with shaking to mid-exponential growth and then diluting the cells to an OD₆₀₀ of 0.01 in fresh biofilm growth medium. Samples of 100 µl of the diluted cells were then aliquoted to each well of a 96-well PVC microtitre plate. The microtitre plates were incubated under stationary conditions, except that, at 12 h after inoculation, the cultures were mixed by pipetting up and down twice to oxygenate the medium. In addition, the spent growth medium was exchanged for fresh biofilm growth medium 12 h after mixing (i.e. 24 h after inoculation). Exchanging the medium resulted in less well-to-well variability in the level of biofilm formation (data not shown). The reasons for this are unknown, although it may be that the biofilms are less likely to detach in the presence of fresh medium. This cycle of mixing followed by exchanging the medium every 12 h was repeated for the time course of the experiment.

At various time points during the incubation of B. subtilis cells in the microtitre plates, the presence of adhered cells was monitored by staining with crystal violet (CV). Growth medium and non-adherent cells were removed from the microtitre plate wells, which were then rinsed with wash buffer (0.15 M ammonium sulphate, 100 mM potassium phosphate, pH 7, 34 mM sodium citrate, 1 mM MgSO₄). Cells that had adhered to the wells were stained with 1% CV in wash buffer at room temperature for 20 min. Excess CV was then removed, and the wells were rinsed with water. The CV that had stained the cells was then solubilized in 200 µl of 80% ethanol, 20% acetone. Biofilm formation was quantified by measuring the OD₅₇₀ for each well using a Bio-Rad model 550 plate reader.

Confocal scanning laser microscopy

For observation of B. subtilis biofilms by confocal scanning laser microscopy (CSLM), B. subtilis biofilms were grown using a modified version of the method described by Watnick and Kolter (1999). Biofilms of B. subtilis strains containing pBL95 were grown on glass coverslides (Fisher Scientific) in 6 ml of biofilm growth medium in 50 ml polypropylene conical tubes. As described for growth in the microtitre plates, the growth medium was mixed 12 h after inoculation, and then the medium was exchanged 12 h later. This cycle of mixing and medium exchange was continued for 48 h. By placing two slides, stuck together, into the conical tubes, biofilms grew on only one side of each glass slide. These slides were separated, rinsed with wash buffer and then fixed to slides with 0.5 mm depressions (Fisher Scientific) filled with wash buffer, such that the biofilm was submerged in the buffer.

Samples were observed using a Leica TCS-SP confocal laser microscope equipped with an argon ion laser. Samples were viewed using 488 nm as the excitation wavelength. Images were analysed using the tcs-nt computer program. Both projections through the X–Y plane and individual scans through the Z-section were obtained by this method.

Calcofluor staining of biofilms

To determine whether B. subtilis biofilms were encased in a polysaccharide matrix, we stained the biofilms with Calcofluor, a polysaccharide-binding dye. Biofilms (lacking GFP) were grown on glass coverslides as described above for CSLM. After rinsing, the slides were stained for 20 min with 10 ml of 75 µg ml⁻¹ Calcofluor (Sigma; fluorescent brighter 28) in wash buffer. The stained biofilms were then analysed by CSLM as described above.

Spore assays of biofilms

To determine whether B. subtilis biofilms contained spores, we harvested biofilms grown in microtitre plates to determine the percentage of viable cells that were heat-resistant spores. After growing wild-type B. subtilis biofilms in microtitre plates, as described above, the wells were rinsed with wash buffer, and the biofilms were manually scraped from the sides of the microtitre plate wells using a pipette tip and resuspended in 100 µl of wash buffer. The percentage of biofilm cells that formed heat-resistant spores was determined by comparing the total number of cfus with the number of cfus remaining after incubation of the resuspended biofilm samples at 80°C for 20 min. Spore assays were performed on biofilm cells at intervals that were 24 h after exchanging the spent medium for fresh medium (i.e. 48 and 72 h from the time of inoculation). Spore assays were performed at these times as fresh nutrient should inhibit spore formation, and 24 h after the addition of fresh nutrients should be sufficient time for cells to reach stationary phase (4 h) and develop into spores (8 h).

Acknowledgements

We thank Mathew Schribler of the UCLA Brain Research Institute Confocal Microscope Facility for assistance with the microscopy. We also thank Antje Hofmeister and Alan Grossman for the kind gift of strains. We thank April Jung for help in determining the conditions for growth of B. subtilis biofilms. We thank Nicola Stanley and Seema Mattoo for helpful suggestions on the manuscript. This work was supported by a Frontiers of Science grant from the Howard Hughes Medical Institute to B.A.L. M.A.H. was supported in part by the Microbial Pathogenesis Training Grant (T32-AI07323) from the National Institutes of Health.