Bacillus subtilis is a ubiquitous soil bacterium that forms biofilms in a process that is negatively controlled by the transcription factor AbrB. To identify the AbrB-regulated genes required for biofilm formation by B. subtilis, genome-wide expression profiling studies of biofilms formed by spo0A abrB and sigH abrB mutant strains were performed. These data, in concert with previously published DNA microarray analysis of spo0A and sigH mutant strains, led to the identification of 39 operons that appear to be repressed by AbrB. Eight of these operons had previously been shown to be repressed by AbrB, and we confirmed AbrB repression for a further six operons by reverse transcription-PCR. The AbrB-repressed genes identified in this study are involved in processes known to be regulated by AbrB, such as extracellular degradative enzyme production and amino acid metabolism, and processes not previously known to be regulated by AbrB, such as membrane bioenergetics and cell wall functions. To determine whether any of these AbrB-regulated genes had a role in biofilm formation, we tested 23 mutants, each with a disruption in a different AbrB-regulated operon, for the ability to form biofilms. Two mutants had a greater than twofold defect in biofilm formation. A yoaW mutant exhibited a biofilm structure with reduced depth, and a sipW mutant exhibited only surface-attached cells and did not form a mature biofilm. YoaW is a putative secreted protein, and SipW is a signal peptidase. This is the first evidence that secreted proteins have a role in biofilm formation by Bacillus subtilis.
Many bacteria exhibit an attached biofilm mode of growth, in which cells grow as an organized three-dimensional community adherent to a surface and encased in a self-produced, polymeric matrix (Costerton et al., 1995; Davey and O’Toole, 2000). It has been proposed that most bacteria in nature are found in a biofilm mode of growth (Costerton et al., 1995). Biofilms are also found in medical and industrial settings, where they can be problematic as the result of the increased resistance of biofilm cells to antimicrobial agents (Kuchma and O’Toole, 2000). Relatively few genes have been identified that are required for bacteria to form biofilms, and little is known about how bacteria coordinate biofilm formation with other phenotypic states that bacteria exhibit.
Bacillus subtilis, a Gram-positive soil bacterium, provides a model system to study the molecular mechanisms controlling biofilm formation. At least three global regulatory proteins, AbrB, Spo0A and Sigma-H, affect biofilm formation by B. subtilis (Branda et al., 2001; Hamon and Lazazzera, 2001). Spo0A is required for surface-attached cells to transition to a three-dimensional biofilm structure (Branda et al., 2001; Hamon and Lazazzera, 2001). The role of Spo0A in this process is to repress expression of AbrB (Fig. 1) (Hamon and Lazazzera, 2001). Spo0A binds and directly represses the abrB promoter (Strauch et al., 1990), and an abrB mutation restores biofilm formation to a spo0A mutant strain (Hamon and Lazazzera, 2001). Thus, AbrB is a negative regulator of the initiation of biofilm formation. Sigma-H may indirectly repress AbrB expression and stimulate the initiation of biofilm formation, as Sigma-H is known to activate expression of spo0A (Fig. 1) (Predich et al., 1992). However, Sigma-H is not essential for the initiation of biofilm formation, as mutants defective for Sigma-H do not have an obvious defect in the initiation of biofilm formation but, rather, have a defect in the formation of fruiting bodies on the surface of biofilms formed by wild isolates of B. subtilis (Branda et al., 2001). Consistent with Spo0A and Sigma-H stimulating the initiation pathway for biofilm formation, many genes that were shown to be regulated by Spo0A and Sigma-H in genome-wide expression profiling studies were also found to be differentially expressed as cells transitioned from a planktonic state to a biofilm state (Stanley et al., 2003). Which of the Spo0A and Sigma-H regulated genes are regulated by AbrB is not fully known, and the genes repressed by AbrB that are required for biofilm formation are unknown.
AbrB, Spo0A and Sigma-H are global regulatory proteins that control many processes that occur during the transition of B. subtilis from exponential growth to stationary phase. In addition to biofilm formation, these transcription factors regulate differentiation of a subpopulation of cells into genetically competent cells capable of taking up exogenous DNA, formation of environmentally resistant spores, and acquisition of new food sources through the production of degradative enzymes and antibiotics (Phillips and Strauch, 2002). This raises the interesting question of how B. subtilis coordinates the decision to enter these different phenotypic states. AbrB has a significant role in this decision making process, as the role of Spo0A and Sigma-H in degradative enzyme production, antibiotic production, and genetic competence is due to their role in repressing AbrB expression (Phillips and Strauch, 2002). Unlike for the biofilm formation pathway, at least some of the genes, required for the formation of these other phenotypic states that are repressed by AbrB, have been identified. However, there have been no studies to generate a complete picture of the genes and, thus, the physiological processes regulated by AbrB.
Here, we present the identification of AbrB-regulated genes that are induced under biofilm formation condition. We identified 57 genes encoded in 39 operons that appear to be repressed by AbrB. More than half of these genes are of unknown function, and many of the genes of known function are involved in metabolism and energy generation. To assess the role of some of these AbrB-regulated genes in biofilm formation, we disrupted 23 of the 39 operons and tested these mutants for their ability to support biofilm formation. Two genes were identified, sipW, which encodes a signal peptidase, and yoaW, which encodes a secreted protein, that are required for normal biofilm formation.
Identification of genes regulated by AbrB
A profile of the genes differentially expressed during biofilm formation by B. subtilis has recently been published (Stanley et al., 2003). Seventy of the genes that were differentially expressed after 24 h incubation under biofilm formation conditions had previously been shown to be activated by Spo0A and/or Sigma-H (Fawcett et al., 2000; Britton et al., 2002). These 70 genes may be regulated by Spo0A or Sigma-H directly or indirectly through Sigma-H and Spo0A repression of abrB (Fig. 1). To determine which of these 70 genes are regulated by AbrB, we compared the gene expression profile of spo0A abrB or sigH abrB mutant strains to a wild-type strain under biofilm formation conditions. AbrB-repressed genes should be those of the 70 genes that are not differentially expressed or have increased expression in the spo0A abrB or sigH abrB mutant, as AbrB is not present in the mutant strains and is similarly depleted in the wild-type strain due to repression of abrB by Spo0A. In contrast, genes that are positively regulated by Spo0A or Sigma-H independently of AbrB should be those of the 70 genes that have decreased expression in the spo0A abrB or sigH abrB mutant strains.
DNA microarrays comprised of 4074 of the 4100 open reading frames of the B. subtilis genome were used to monitor the differences in mRNA levels between by the wild-type strain and either a spo0A abrB or sigH abrB mutant strain after growth of the cells under biofilm formation conditions for 24 h (see Experimental procedures). The RNA from the wild-type and the mutant strains was fluorescently labelled with Cy5 and Cy3, respectively, through the generation of cDNA. The DNA microarrays were simultaneously hybridised with the wild-type cDNA and one of the mutant cDNA samples to determine the ratio of gene expression. Those genes that had highly variable expression ratios were eliminated from further analysis as previously described (Stanley et al., 2003).
From the wild-type versus spo0A abrB DNA microarray experiments, approximately 60% of the genes had reproducible expression ratios. As much as a 5.1-fold difference in the expression level for a gene was observed. Iterative outlier analysis was applied to those genes that had reproducible ratios to determine which genes had significantly different expression between the two strains. 100 genes were identified through this approach as differentially expressed between the wild-type and spo0A abrB strains (see Supplementary material, Table 1).
. Shown is the ratio of gene expression for the first gene in the operon that is not in parenthesis.
*Indicates genes for which previously published data indicate that these genes are regulated by AbrB. Genes in parenthesis are those that appear to be regulated in a similar manner to other genes in the same operon but whose expression value were too variable to be identified as differentially expressed.
. Indicated is the ratio reported in Stanley et al. (2003) for expression of the indicated gene under biofilm formation conditions (B) divided by the expression of the gene under planktonic conditions (P). A ratio of >1 indicates a higher level of expression in the biofilm formation condition sample.
. Indicated is the ratio reported by Fawcett et al. (2000) for expression of the gene in the wild-type (WT) strain divided by the expression in the spo0A mutant strain (0 A). A ratio of >1 represents a higher level of expression in the wild-type strain. ND represents those genes for which no data was available.
. Indicated is the ratio of expression of the gene in the wild-type strain (WT) divided by the expression of the gene in the spo0A abrB strain (0AabrB). Both stains were grown under biofilm formation conditions for 24 h. A ratio of >1 represents a higher expression level in the wild-type strain. ND represents those genes for which no reproducible ratio was obtained.
. Indicated is the ratio reported by Britton et al. (2002) for expression of the gene in the wild-type strain (WT) divided by the expression in the sigH mutant strain (sigH). A ratio of >1 represents a higher level of expression in the wild-type strain. ND represents those genes that were not identified as differentially expressed in the study.
. Indicated is the ratio of expression of the gene in the wild-type strain (WT) divided by the expression of the gene in the sigH abrB strain (sigHabrB). Both stains were grown under biofilm formation conditions for 24 h. A ratio of >1 represents a higher expression level in the wild-type strain. ND represents those genes for which no reproducible ratio was obtained.
unknown (yqxM), signal peptidase (sipW), spore component (tasA)
yvd(F, G, H, I) J (K, malL, pgcM)
similar to maltodextrin utilization system
yvq (I), H,
ywq (H, I), J, K, L
yxbB, A, yxnB (asnH, yxaM)
unknown (yxbB, A, yxnB), asparagine synthetase (asnH), similar to tetracycline efflux pump (yxaM)
From the wild-type versus sigH abrB DNA microarray experiments, approximately 54% of the genes gave reproducible expression ratios. As much as a 7.1-fold difference in the expression level for a gene was observed. Iterative outlier analysis identified 140 genes as differentially expressed between the wild-type and sigH abrB strains (see Supplementary material, Table 2).
Table 2. . RT-PCR analysis of AbrB-regulated genes.
. Strains used are as follows; wild type, BAL 218; spo0A, BAL679; spo0A abrB, BAL734. The numbers represent the ratio of gene expression as a function of veg expression and are the average of three independent experiments. The errors represent the standard error of the mean.
0.47 ± 0.1
0.23 ± 0.04
0.75 ± 0.1
1.1 ± 0.3
0.75 ± 0.1
1.4 ± 0.2
1.0 ± 0.1
0.84 ± 0.1
2.3 ± 0.3
1.4 ± 0.4
1.1 ± 0.2
2.2 ± 0.2
0.77 ± 0.2
0.47 ± 0.06
0.86 ± 0.1
0.56 ± 0.1
0.38 ± 0.03
1.3 ± 0.07
0.68 ± 0.2
0.22 ± 0.05
1.5 ± 0.3
Of the 70 genes previously shown to be activated by Spo0A or Sigma-H and induced under biofilm formation conditions (Stanley et al., 2003), 13 had lower expression in either the spo0A abrB or sigH abrB mutant strain (see Supplementary material, Table 3). This suggests that Spo0A or Sigma-H, independently of AbrB, activated these 13 genes. These data further suggest that the remaining 57 genes may be regulated by Spo0A or Sigma-H through AbrB (Table 1). These 57 genes are part of 39 known or putative operons. Many genes that are part of these operons were not identified as differentially expressed. However, these genes have similar expression profiles to the genes that were identified as differentially expressed, except their variances were high in the Stanley et al. (2003) study. These genes are listed in Table 1 in parentheses as they are likely to be AbrB regulated. Consistent with the operons in Table 1 being regulated by AbrB, seven of the operons have previously been shown to be regulated by AbrB under planktonic conditions: aprE (Jan et al., 2000; Ferrari et al., 1988), dppABCDE (Slack et al., 1991), hutPHUIGM (Fisher et al., 1994), phrE (McQuade et al., 2001), qcrABC (Yu et al., 1995), spoVG (Robertson et al., 1989), and yqxM sipW tasA (Stover and Driks, 1999a).
Table 3. . Biofilm formation by mutants defective for AbrB-repressed genes.
To confirm via a second method that the genes identified as regulated by AbrB under planktonic conditions were indeed regulated by AbrB under biofilm formation conditions, we monitored the expression of the yqxM sipW tasA operon. This was achieved by measuring the levels of β-galactosidase activity from wild-type, spo0A and spo0A abrB strains containing a sipW-lacZ transcriptional fusion. sipW is the second gene in the three gene operon (Stover and Driks, 1999a). The levels of β-galactosidase activity in these strains were measured after growth under planktonic conditions and after growth for 16 h under biofilm formation conditions.
The sipW-lacZ fusion exhibited a 21-fold higher expression under biofilm formation conditions versus planktonic conditions in the wild-type strain (data not shown). The expression of sipW-lacZ in the spo0A mutant strain was near the detection threshold under biofilm formation conditions (0.11 units of activity for the spo0A mutant strain versus 28 units for the wild-type strain), indicating that sipW is under the positive transcriptional control of Spo0A under biofilm formation conditions. In contrast, for the spo0A abrB mutant strain, 98 units of activity were observed under biofilm formation conditions, indicating that AbrB inhibits sipW transcription in a spo0A mutant strain. These data confirm that the expression of the sipW operon is regulated in a similar manner under biofilm formation conditions as under planktonic conditions (Stover and Driks, 1999a).
To determine whether any of the genes listed in Table 1, which were not previously determined to be regulated by AbrB, were indeed AbrB-regulated, reverse transcription-PCR (RT-PCR) was utilised. This allowed the expression of a gene to be monitored and expressed as a ratio with respect to a control gene, veg, that was known not to be differentially expressed (Ollington et al., 1981; Ollington and Losick, 1981; Gilman and Chamberlin, 1983) and see Experimental procedures). We examined the expression profile of seven genes from seven different operons: yqxM (the first gene in the operon containing sipW), ycnK, yoaW, yocH, yvqH, yxbB, and yxbC in wild-type, spo0A and spo0A abrB mutant strains grown for 24 h under biofilm formation conditions. The gene yqxM was used as a positive control. It was expressed at levels twofold higher in the wild-type strain than in the spo0A mutant strain (Table 2), indicating that yqxM is under the positive transcriptional control of Spo0A. Expression of yqxM in the spo0A abrB mutant strain was 3.5-fold higher than in spo0A mutant strain (Table 2), indicating that AbrB mediated the repression of yqxM seen in the spo0A strain. Similar expression profiles were seen for ycnK, yoaW, yocH, yvqH, yxbB and yxbC (Table 2). These data support the microarray analysis that identified these genes as repressed by AbrB under biofilm formation conditions.
Microtitre plate analysis of biofilm formation by mutants defective for AbrB-regulated genes
To determine whether any of the 39 operons identified as AbrB repressed are involved in biofilm formation, we disrupted 23 of the operons by disrupting the first gene of the operon (Table 3). These 23 operons represent 17 of the 21 operons of unknown function and six of the 18 operons of known function. These 23 mutants were tested for their ability to form biofilms in the microtitre plate assay (Table 3). Nine of the 23 mutants exhibited significantly (P < 0.01) altered levels of biofilm formation. Six mutants, tagE, ymfJ, ywqH, yqxI, yqxM and yoaW, formed reduced levels of biofilms, and three mutants, yhjM, ydeH and era, formed increased levels compared with the wild-type strain. A growth defect did not appear to be responsible for the altered levels of biofilm formation by all but one of these mutants; only the tagE mutant grew slower than wild type in microtitre plates (data not shown). We chose to further study the effect of the yqxM and the yoaW operons on biofilm formation as only these mutant strains exhibited a greater than twofold difference in biofilm formation.
The biofilm defect of yqxM and yoaW mutant strains could be due to the loss of YqxM or YoaW proteins, respectively, or loss of another protein encoded in the same operon as yqxM or yoaW. Analysis of the DNA sequence surrounding yoaW indicated that this gene is unlikely to be co-transcribed with another gene. The sequence of a transcription terminator can be found both up and downstream of yoaW, and downstream of yoaW there is a gene transcribed in the opposite orientation (Fig. 2). Thus, the phenotype of yoaW mutant cells is likely to be due to loss of YoaW.
In contrast, yqxM is known to be the first gene in a three gene operon containing sipW and tasA (Fig. 2) (Stover and Driks, 1999a; Serrano et al., 1999). Thus, we sought to determine whether the defect in biofilm formation observed in the microtiter plate assay for the yqxM mutant could be the result of a defect in either tasA or sipW expression. Mutants with non-polar mutations in yqxM or tasA, or a mutation in sipW, which is polar on tasA, were tested for biofilm formation using the microtitre plate assay. As shown in Fig. 3, the yqxM and the tasA deletion strains formed biofilms at levels that were statistically indistinguishable from the wild-type strain. In contrast, the sipW deletion strain showed a 7.5-fold decrease in biofilm formation (P < 0.001), which is comparable to the level observed for the yqxM operon mutant strain. These data indicate that neither yqxM nor tasA individually are required for biofilm formation and that the lack of SipW is largely responsible for the biofilm defect of a yqxM operon mutant.
SipW is a signal peptidase that is only known to process TasA and YqxM (Stover and Driks, 1999a,b; Serrano et al., 1999). Thus, the biofilm defect of a sipW mutant could have been due to the inability of a sipW mutant strain to process both TasA and YqxM. To determine whether tasA and yqxM have redundant functions in biofilm formation, a double tasA yqxM mutant, which should still produce SipW (Stover and Driks, 1999c), was tested for biofilm formation using the microtitre plate assay. As shown in Fig. 3, the double mutant showed a 2.6-fold increase (P < 0.001) in the level of biofilm formation compared to the wild-type strain. The reason high levels of biofilm formation were observed for the tasA yqxM mutant is not known. These data support the conclusion that neither TasA nor YqxM are required for biofilm formation and indicate that SipW processes at least one protein other than TasA or YqxM that is required for biofilm formation.
Microscopic analysis of surface-adhered sipW and yoaW mutant cells
To determine the step in biofilm formation at which the yoaW and the sipW mutants were arrested, surface adhered yoaW and sipW mutant cells were analyzed by confocal scanning laser microscopy (CSLM). A view of the surface in the X–Y plane (i.e. perpendicular to the surface) showed that, similar to the wild-type cells, yoaW cells were able to adhere to the surface of the glass slide (compare Figs 4D and 4A). Cross sections through the structure formed by the yoaW mutant cells, in the X–Z plane, showed a structure with reduced depth compared to the structure formed by the wild-type cells (compare Figs 4D and 4A). The structure formed by the yoaW mutant strain showed an average depth of 8.0 ± 0.5 µm. This is in contrast to the wild-type strain, which formed three-dimensional structures with average depth of 14 ± 0.7 µm. The fold difference in the depth of the biofilm formed by the wild-type and the yoaW mutant strains as seen by CSLM is similar to the fold difference in the level of biofilm formation observed in the microtiter plate assay.
A view of surface adhered sipW mutant cells by CSLM in the X–Y plane showed that sipW mutant cells were able to adhere to the surface of the glass slide (Fig. 4C). However, these cells appear unable to cluster together to form microcolonies. Cross-sections through the field, in the X–Z plane, indicate that the sipW mutant cells adhere in a monolayer, as CSLM images showed a structure with a depth ≥ 4 µm (Fig. 4C).
To further confirm that sipW was the only gene in the yqxM operon with a role in biofilm formation, the tasA yqxM mutant strain that expresses SipW was analyzed by CSLM. As shown in Fig. 4B, the structure formed by the yqxM tasA double mutant strain showed a similar biofilm morphology to the wild-type strain, both in the X–Y and the X–Z sections. The average depth the biofilm formed by the mutant strain was 10 ± 1.0 µm, which was not considered significantly different from the wild-type structure as determined by a standard t-test. These data indicate that tasA and yqxM have no apparent role in biofilm formation.
In this study, we have generated a list of genes that appear to be activated during biofilm formation caused by the relief of AbrB repression. This was done by comparing the data from DNA microarray experiments performed in this study, which determined the gene expression profile of spo0A abrB mutant and sigH abrB mutant cells under biofilm formation conditions, to three previously published DNA microarray experiments (Fawcett et al., 2000; Britton et al., 2002; Stanley et al., 2003). Genes regulated by AbrB are of interest as they are candidate genes for having a role in biofilm formation as well as other processes that occur during the transition to stationary phase. We have tested mutants defective for 23 of the 39 operons identified as AbrB-regulated for their ability to form biofilms. From this analysis, we have identified two genes, sipW and yoaW, which are necessary for B. subtilis to form a biofilm with a three-dimensional structure similar to a wild-type strain.
Comparative analysis of independent DNA microarray experiments led to the identification of AbrB-regulated genes
In this study, we aimed to identify AbrB-regulated genes differentially expressed under biofilm formation conditions by comparing the results from DNA microarray experiments performed in different laboratories under different growth conditions. Stanley et al. (2003) previously identified Spo0A- or Sigma-H-regulated genes that were differentially expressed under biofilm formation conditions, by comparing the transcription profile of wild-type cells grown under biofilm formation conditions to the transcription profile of spo0A or sigH mutant cells under planktonic conditions (Fawcett et al., 2000; Britton et al., 2002; Stanley et al., 2003). As a subset of the Spo0A and Sigma-H-regulated genes should have been controlled by AbrB, we compared the results of the analysis performed by Stanley et al. (2003) to the transcription profile of spo0A abrB and sigH abrB mutant cells under biofilm formation conditions. Using this method, we have identified 39 operons as AbrB repressed.
To our knowledge, this is the first report of a comparative analysis of DNA microarray experiments to study a bacterial system. This type of analysis is powerful at generating new information quickly, as it relies on previously existing data. However, as it compares DNA microarray analyses with cells grown under different growth conditions, it can only provide a partial list of the genes controlled by a transcription factor. Such a partial list may be sufficient in cases such as that presented here, where a partial list of genes regulated by AbrB was sufficient to identify genes required for biofilm formation.
Large-scale analyses such as DNA microarray analyses will inherently generate some level of false positive results (Murphy, 2002; Hatfield et al., 2003). Any comparative DNA microarray analyses will compound the false positive error rate inherit in each DNA microarray experiment. Thus, it is important to consider where false positives might arise. One place is from our assumption that genes, known to be activated by Spo0A or Sigma-H under planktonic conditions and known to be induced under biofilm formation conditions, were indeed induced under biofilm formation conditions due to the activity of Spo0A or Sigma-H, and not due to the activity of another transcription factor. If these genes were differentially expressed under biofilm formation conditions due to the activity of another transcription factor, they should not be differentially expressed in the spo0A abrB or sigH abrB mutant cells and would be falsely identified as AbrB-regulated genes. At least six genes, ctaC, D, F, sspB, tagE, and bcd, identified as AbrB-regulated in this study are known to be regulated by other transcription factors (Liu and Taber, 1998; Liu et al., 1998; Nicholson et al., 1989; Debarbouille et al., 1999). Further work will be necessary to determine whether these genes are induced under biofilm formation conditions due to relief of AbrB repression or another transcription factor.
False positive results may also have been generated by the inclusion of genes for which the expression profile data was incomplete. For example, expression data was not obtained for ctaO from the spo0A DNA microarray experiment and for the sigH abrB DNA microarray experiment. In this instance, we used data from the sigH microarray analysis and data from the spo0A abrB microarray analysis to deduce AbrB regulation. Although this is an indirect method of identifying ctaO as AbrB-regulated, we choose not to exclude genes with similarly incomplete expression data, as of the five genes, spoVG, sspB, yfhD, yhjM, and yvdJ, with similarly missing data one, spoVG, is known to be AbrB-regulated (Robertson et al., 1989).
Despite the limitations of this study, the level of false positives appears to be low, as AbrB regulation for 13 of the 39 operons (33%) identified in this study has been confirmed. Seven of these operons have been shown in previous studies to be regulated by AbrB. An additional six operons, chosen at random, were shown via RT-PCR to be regulated by AbrB (Table 2). The finding that 100% of those operons chosen for further analysis were AbrB regulated indicates that the incidence of false positives is likely to be low.
Two AbrB regulated genes, sipW and yoaW, are essential for forming a mature biofilm
Previous data had indicated that AbrB repressed at least one gene required for biofilm formation (Hamon and Lazazzera, 2001). In this study, we showed that, of the 23 operons identified as AbrB-regulated and tested for a role in biofilm formation, nine had a significant effect on biofilm formation under the conditions tested in this study. The effect of seven of these operons was small and therefore not studied further. The effect of two operons, the yqxM and the yoaW operons, was larger and was further studied by CSLM. Both yqxM and yoaW operon mutants showed defects in forming a mature biofilm structure.
The YoaW protein is required at a late stage of biofilm formation, as a yoaW mutant formed a structure with reduced depth compared to a wild-type strain. A mechanism that has been proposed to explain why an Escherichia coli wcaF mutant strain exhibits biofilms with reduced depth is that this mutant forms fragile biofilm structures that loose depth upon rinsing for microscopic examination (Danese et al., 2000). WcaF is required for extracellular polysaccharide production, which is presumably part of the extracellular matrix of E. coli biofilms (Danese et al., 2000). As the phenotype of an E. coli wcaF mutant resembles that of a B. subtilis yoaW mutant, YoaW could have a similar role to WcaF in forming the extracellular matrix required for a mature biofilm.
In contrast to the yoaW operon that appears to be required at a late stage in the biofilm formation, one gene in the yqxM operon is required at an early stage in biofilm formation. Microscopic images of surface attached sipW mutant cells indicated that these cells were able to attach to the surface of a glass slide, but were unable to organize into microcolonies. The stage in the biofilm formation pathway that is blocked in a sipW mutant is the same stage in biofilm formation that was previously shown to be repressed by AbrB (Hamon and Lazazzera, 2001). Thus, one of the major functions of AbrB in repressing biofilm formation is to repress sipW.
As SipW is a signal peptidase, its role in biofilm formation is most likely to process a secreted protein that has a role in allowing surface-adhered cells to form a mature biofilm. Different functions have been shown to enable surface-attached cells to transition to a mature biofilm in different bacterial systems. Synthesis of a polysaccharide intercellular adhesin is required by Staphylococcus epidermidis, twitching motility is required under some conditions by Pseudomonas aeruginosa, and flagellar motility is required by E. coli (O’Toole and Kolter, 1998; Pratt and Kolter, 1998; Cucarella et al., 2001). These observations suggest that the SipW-processed protein may have a role in producing an intercellular adhesin or a motility structure in B. subtilis. Identifying the protein(s) processed by SipW that is required for biofilm formation would distinguish these possibilities.
It is interesting to note that neither of the known SipW-processed proteins, YqxM or TasA, were required for biofilm formation. It is tempting to speculate that the production of the SipW-processed protein that is required for biofilm formation is also AbrB regulated. Seven genes identified as AbrB regulated, aprE, qcrA, sdpA, xynD, yocH, yoaW, yqxI, yraI, encode proteins that have a putative signal sequence and therefore could be processed by SipW. Five of these genes were tested for a role in biofilm formation, including yoaW, but mutations in these genes did not result in a phenotype similar to sipW. Preliminary experiments to detect processing of YoaW by SipW were unsuccessful due to our inability to detect secreted YoaW (data not shown). It is possible that genes encoding multiple SipW-processed proteins will need to be deleted before a phenotype similar to SipW is observed.
Potential role for AbrB-regulated genes under biofilm formation conditions
Many of the AbrB-regulated genes identified in this study may contribute to the metabolic capacity of biofilm cells. AbrB has previously been shown to regulate the expression of extracellular degradative enzymes, nitrogen utilization enzymes, amino acid metabolism enzymes and transporters, all of which would expand the range of potential nutrient sources used by B. subtilis (Phillips and Strauch, 2002). In this study, we identified genes that encode enzymes involved in all of these functions. In addition, we identified genes involved in carbon source utilization, nucleotide metabolism, membrane bioenergetics, antibiotic resistance, and delay of sporulation. The altered metabolic capacity conferred by increased expression of these genes may reflect the type of nutrients found in the environment where B. subtilis typically forms biofilms as well as the general physiology of cells in a biofilm.
Two extracellular degradative enzymes were identified in this study, XynD and AprE, which are involved in xylan and protein degradation, respectively (Stahl and Ferrari, 1984; Wong et al., 1984; Gosalbes et al., 1991). Production of these extracellular enzymes may be advantageous to B. subtilis during biofilm formation, as these enzymes and consequently the nutrients released may be less likely to diffuse away from the cells in a biofilm. Xylan is a plant material, and the induction of a xylan-degrading enzyme during biofilm formation may indicate that decaying plant material is a common surface on which B. subtilis forms biofilms. Similarly, the induction of a maltodextran utilizing operon, the yvdF operon (Cho et al., 2000), may indicate that starch, a component of plants, is a common nutrient encountered in the environment where B. subtilis forms biofilms.
Regulation of genes involved in utilizing amino acids as nitrogen sources, the bkdR and hutP operons (Kimhi and Magasanik, 1970; Debarbouille et al., 1999), and in transport of dipeptides, the dppA operon (Cheggour et al., 2000), may reflect the utilization of substrates released by the extracellular protease AprE. Conversely, the increased expression of genes involved in the synthesis of the amino acids asparagine [asnH – part of the yxbB operon (Yoshida et al., 1999)), leucine (the ilvB operon (Grandoni et al., 1992)], and threonine [the hom operon (Parsot and Cohen, 1988)], may reflect the lack of these amino acids in the environment in which B. subtilis forms biofilms.
Three operons involved in membrane bioenergetics, ctaC, ctaO, and qcrA, were identified as regulated by AbrB under biofilm formation conditions. The ctaC and qcrA operons encode enzymes that form the cytochrome oxidase pathway for shuttling electrons from menaquinone to O2, one of four pathways that B. subtilis possesses for reducing O2 ( Saraste et al., 1991; Yu et al., 1995; Winstedt and von Wachenfeldt, 2000). The differential regulation of these pathways suggests that they have different functions (Winstedt and von Wachenfeldt, 2000). The terminal oxidases of E. coli also exhibit differential regulation, such that the oxidase with a higher affinity for O2 is expressed maximally at low O2 concentrations (Govantes et al., 2000). It is possible that the O2-reducing pathway that has increased expression under biofilm formation conditions has a higher affinity for oxygen, as it is well documented that O2 concentrations are severely limited in the depths of biofilms (Costerton et al., 1995).
The increased expression of CtaO may be important for the activity of the cytochrome oxidase pathway. Heme A is found in the terminal oxidase of this pathway. CtaO has heme O synthesis activity, which is proposed to be a precursor for Heme A synthesis (Mogi et al., 1994). Similarly, the increased expression of a putative copper import protein, YcnJ, may be important for the cytochrome oxidase pathway. The terminal oxidase of this pathway is a heme-copper oxidase (Winstedt and von Wachenfeldt, 2000).
Several genes involved in cell wall functions exhibited increased expression under biofilm formation conditions. Alterations in the cell wall are interesting as they may change how the cells interact with surfaces. TagE, which is involved in teichoic acid biosynthesis in B. subtilis, was identified in our studies (Mauel et al., 1991). Teichoic acids have been shown to play a role in adherence of Staphylococcus aureus cells to surfaces (Gross et al., 2001). Mutants lacking tagE exhibited decreased levels of biofilm formation. However, they also exhibited a slow growth rate, making it difficult to discern the role of teichoic acids in biofilm formation by B. subtilis.
Other genes involved in cell wall functions are part of the ykfA operon, which has been proposed to play a role in recycling cell wall peptides. YkfA has amino acid similarity to LdcA of E. coli, a l,d-carboxypeptidase that was shown to be required to recycle cell wall peptides for new cell wall synthesis (Templin et al., 1999). Interestingly, LdcA is essential for survival of E. coli in stationary phase, presumably due to the decreased ability to synthesize cell wall peptides de novo under stationary phase conditions (Templin et al., 1999). As biofilm formation in B. subtilis occurs in stationary phase, YkfA may be important for survival of B. subtilis cells under these conditions. YkfC is an endopeptidase that can cleave cell wall peptides, and YkfB has l-Ala-d/l-Glu epimerase activity (Schmidt et al., 2001). These proteins are proposed to have a role in degrading cell wall peptides, which could be useful to cells in biofilms as a source of nutrients.
Biofilm cells exhibit increased resistance to antimicrobial agents. Thus, it is of particular interest that yxaM has increased expression in B. subtilis under biofilm formation conditions. YxaM is 23% identical and 45% similar to the tetracycline efflux pump, TetA(P), of Clostridium septicum and perfringens (Sloan et al., 1994). At this time it is not known whether B. subtilis cells are more resistant to tetracycline when growing in a biofilm.
As AbrB regulated processes occur during the transition to stationary phase, the question arises as to how B. subtilis coordinates these processes. It is therefore interesting to note one operon responsible for delaying sporulation, sdpA, has been identified as differentially expressed under biofilm formation conditions. Gonzalez-Pastor et al. (2003) postulate that the sdpA operon allows cells to wait as long as possible before committing to sporulation, which is a dormant state. Expression of the sdpA operon under biofilm formation conditions may be a mechanism to allow biofilm formation to occur under conditions where nutrient concentrations are low, but sufficient to allow metabolic activity.
In summary, we have performed a comparative analysis of DNA microarrays conducted in different laboratories and under different conditions. To our knowledge, this is the first time such an approach has been applied to study gene regulation in a bacterial system. From this analysis, we have successfully identified AbrB-regulated genes that are activated under biofilm formation conditions. These studies have furthered our knowledge of the process regulated by AbrB. We have further shown that 9 of these AbrB-regulated genes have a significant effect on biofilm formation. We analyzed in greater detail the phenotype of two mutants that showed the greatest defect in biofilm formation, sipW and yoaW. The gene sipW encodes a signal peptidase, which is the first protein that is not a transcription factor that has been shown to have a role in the transition of surface attached B. subtilis cells to a biofilm. The gene yoaW encodes a secreted protein, which is also the first protein that is not a transcription factor that has been shown to affect the depth of B. subtilis biofilms. Understanding how SipW and YoaW contribute to biofilm formation will further our understanding of the mechanisms used by bacteria to form biofilms.
LB was used for growth of both E. coli and B. subtilis for routine strain construction and maintenance. For growth of B. subtilis under biofilm growth conditions, biofilm growth medium was used (Hamon and Lazazzera, 2001). Antibiotics were used at the following concentrations as appropriate: 100 µg ml−1 ampicillin, 5 µg ml−1 chloramphenicol, 5 µg ml−1 neomycin, 100 µg ml−1 spectinomycin.
Strain and plasmid construction
The B. subtilis strains used in this study are described in Table 4 and were constructed by transformation with chromosomal DNA or plasmids using standard protocols (Harwood and Cutting, 1990). All strains are derivatives of the parental strain BAL218 (JH642) and contain trpC2 and pheA1 mutations (Perego et al., 1988).
Table 4. .Bacillus subtilis strains used in this study.
Disruption of 23 operons identified as AbrB-regulated was accomplished by integrating a plasmid containing an internal fragment of the gene into the chromosome of strain BAL218. The internal fragment of the genes were obtained by PCR amplification using primers that amplify from the positions indicated in Supplementary materialTable 4. The PCR product was blunt cloned into the SmaI site of pBL132 (Stanley et al., 2003). The newly constructed plasmid was then integrated into the chromosome at the gene locus using selection for chloramphenicol resistance associated with the plasmid.
DNA microarray experiments
RNA for use in DNA microarray experiments was isolated from biofilm cells of BAL218 (wild-type), BAL678 (sigH::cat::spec, abrB::Tn917), and BAL734 (spo0AD56N:: spec, abrB::Tn917) cells after incubation for 24 h under biofilm formation conditions. The cells were grown under planktonic conditions in biofilm growth medium at 37°C with shaking at 200 r.p.m. to late-exponential phase (OD600 = 2.5). The cells were then diluted to an OD600 of 0.1 in 20 ml of fresh biofilm growth medium and placed in 250 ml beakers. The beakers were incubated at 37°C without shaking for 24 h. At this point, the cells had formed an air-medium interface biofilm, and the growth medium below the biofilm was removed using a pipet with minimal disruption to the biofilm. The biofilm cells were washed in 5 ml of wash buffer (15 mM (NH4)2S04, 80 mM K2HPO4, 44 mM KH2PO4, 3.4 mM sodium citrate, 1 mM MgSO4) and the RNA was extracted from these cells using a Qiagen RNAeasy Maxiprep kit according to the manufacture's instructions. The isolated RNA was treated with DNaseI (Qiagen) to remove any contaminating genomic DNA.
To determine the relative expression level of a gene between wild-type and mutant cells, the wild-type and mutant RNA samples were labelled with Cy3 and Cy5, respectively, as previously described (Britton et al., 2002). The DNA microarray slides were simultaneously hybridized with the wild-type cDNA and a mutant cDNA to determine the ratio of gene expression between the wild-type and mutant strain as described in Britton et al. (2002). For each gene, the ratios from three independent experiments (i.e. independently grown and prepared samples) was averaged. The genes with a significantly different expression level were identified using iterative outlier analysis as previously described (Stanley et al., 2003).
Microtitre plate assay of B. subtilis biofilm formation
The microtitre plate assay measures the level of cells adhered to the surface of the microtiter plate wells. These assays were performed as described in Hamon and Lazazzera (2001). For one assay, the OD570 of between 16 and 24 wells were averaged. The standard error of the mean of these wells was <10%. Background levels of staining (i.e. wells incubated with growth medium lacking cells) were subtracted from the average. This assay was repeated on at least four separate occasions, and the average values from each independent assay was averaged to determine the level of biofilm formation for a strain.
Confocal scanning laser microscopy
Confocal scanning laser microscopy images were obtained as described (Hamon and Lazazzera, 2001). Briefly, 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. Sections through the X-Y plane and the X-Z plane (Z section) were obtained using the TCS-NT computer program. Each strain was examined on at least three separate occasions, and the average depth of the biofilm was determined for each strain using 6–10 independent Z section measurements.
Measurement of β-galactosidase specific activity under both planktonic and biofilm formation conditions was as described previously (Stanley et al., 2003), except that strains were grown under biofilm formation conditions for 16 h. The β-galactosidase specific activity of strains BAL881 [sipW–lacZ (cat)], BAL1293 [spo0AD56N-cat::spec, abrB::Tn917, sipW–lacZ (cat)], and BAL1300 [spo0AD56N-cat::spec, sipW–lacZ (cat)] was measured.
Reverse transcription-PCR analysis
The cDNA templates used for RT-PCR analysis were generated through reverse transcription of total RNA isolated from cells grown under biofilm formation conditions for 24 h in microtitre plates. The cells were grown in an identical manner to that for the microtiter plate assay. The complete contents of 10 microtitre plate wells were collected by pipetting and combined. The cells were vortexed to disrupt any cell aggregates, and the OD600 measured. RNA was then isolated from ∼107 cells using an RNAeasy miniprep kit according to the manufacture's instructions (Qiagen). This RNA was digested with DNaseI (Qiagen) until complete digestion of any contaminating genomic DNA was achieved. This was determined by the absence of a PCR product when using the total RNA as a template in a PCR reaction (data not shown). 100 ng of the total RNA was used in a reverse transcription reaction using random hexamers in accordance with the manufacture's instructions (Invitrogen First Strand cDNA synthesis Kit). 1 µl of the resulting cDNA was used in a 20 µl RT-PCR reaction.
RT-PCR analysis (Wong et al., 1994; Wall and Edwards, 2002) is a process where the relative expression of a gene is compared to that of an internal control, in this case veg, whose expression is known not to vary under different growth conditions (Ollington et al., 1981; Ollington and Losick, 1981; Gilman and Chamberlin, 1983). The number of PCR amplification cycles required for a linear increase in the PCR product for both the internal control gene veg and the gene of interest was determined for the cDNA generated from the wild-type sample. This was determined by following the PCR reaction over a number of cycles and assessing the amount of resulting product by agarose gel electrophoreses and ethidium bromide staining. The fluorescence level from the PCR product was measured using a Chemimager™ 4400 (Alpha Innotech Corporation). Once the number of cycles required for a linear amplification was calculated for veg and for the gene of interest with cDNA generated from the wild-type strain, a PCR reaction was performed with cDNA generated from BAL679 (spo0AD56N-cat::spec) and BAL734 (abrB::Tn917, spo0AD56N-cat::spec) to monitor the relative level of gene expression between the three strains. The PCR reactions conditions were 1× PCR reaction buffer, 1× Q-solution (Qiagen), 3.5 mM MgCl2, 0.5 µM each primer, 0.2 µM each dNTP, 0.8 units Taq DNA polymerase per 20 µl reaction (Qiagen). The sequence of the primers used is listed in Table 5. Between 24 and 32 cycles of amplification were used with 30 s at 95°C, 30 s at 48°C (except for yocH where 54°C was used) and 30 s at 72°C.
. The position of the primers is given, with +1 being the ‘A’ of the ATG start codon.
. Represented is the number of cycles used in the RT-PCR analysis. The number in parenthesis represents the number of cycles used at an annealing temperature of 54°C for analysis with the yocH primers.
+400 –+ 377
We are grateful for strains AGS157, AGS175 and AGS207, and plasmid pLGW201 kindly provided by Dr Adam Driks. We would like to thank Dr Matthew Schibler for advice and help with the confocal scanning laser microscopy performed at the UCLA Brain Research Institute confocal microscope facility. Thank you to Cynthia Lee for her help with strain construction. This work was supported in part by the University of California Academic Senate Council on Research of the Los Angeles Division, the UCLA foundation and a Frontiers of Science-Seed grant form the Howard Hughes Medical Institute (B.A.L.), and by the Public Health Services Grant AI48616 to B.A.L. and GM50895 to A.D.G. N.R.S. was supported by a long-term postdoctoral fellowship awarded by the EMBO. M.A.H. was supported in part by the Microbial Pathogenesis Training Grant (T32-AI07323) from the National Institute of Health and in part by the Warsaw fellowship.