Targets of the master regulator of biofilm formation in Bacillus subtilis

Authors

  • Frances Chu,

    1. Department of Molecular and Cellular Biology, Harvard University, 16 Divinity Avenue, Cambridge, MA 02138, USA.
    Search for more papers by this author
  • Daniel B. Kearns,

    1. Department of Molecular and Cellular Biology, Harvard University, 16 Divinity Avenue, Cambridge, MA 02138, USA.
    Search for more papers by this author
    • Department of Biology, Indiana University, Bloomington, IN 47408, USA.

  • Steven S. Branda,

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

  • Roberto Kolter,

    1. Department of Microbiology and Molecular Genetics, Harvard Medical School, Boston, MA 02115, USA.
    Search for more papers by this author
  • Richard Losick

    Corresponding author
    1. Department of Molecular and Cellular Biology, Harvard University, 16 Divinity Avenue, Cambridge, MA 02138, USA.
      *E-mail losick@mcb.harvard.edu; Tel. (+1) 617 495 4905; Fax (+1) 617 496 4642.
    Search for more papers by this author

*E-mail losick@mcb.harvard.edu; Tel. (+1) 617 495 4905; Fax (+1) 617 496 4642.

Summary

Wild strains of the spore-forming bacterium Bacillus subtilis are capable of forming architecturally complex communities of cells. The formation of these biofilms is mediated in part by the 15-gene exopolysaccharide operon, epsA-O, which is under the direct negative control of the SinR repressor. We report the identification of an additional operon, yqxM-sipW-tasA, that is required for biofilm formation and is under the direct negative control of SinR. We now show that all three members of the operon are required for the formation of robust biofilms and that SinR is a potent repressor of the operon that acts by binding to multiple sites in the promoter region. Genome-wide analysis of SinR-controlled transcription indicates that the epsA-O and yqxM-sipW-tasA operons constitute many of the most strongly controlled genes in the SinR regulon. These findings reinforce the view that SinR is a master regulator for biofilm formation and further suggest that a principal biological function of SinR is to govern the assembly of complex multicellular communities.

Introduction

Bacteria of many kinds are known to be capable of forming multicellular communities with complex morphological features. Three examples in well-studied bacteria are fruiting body formation by Myxococcus xanthus, aerial mycelium formation by Streptomyces coelicolor, and the formation of elaborate multicellular communities (biofilms) by Bacillus subtilis, the subject of the present communication (Kelemen and Buttner, 1998; Branda et al., 2001; Jelsbak and Sogaard-Andersen, 2003). Wild strains of B. subtilis, a Gram positive soil bacterium, are known to form robust pellicles at liquid–air interfaces in standing culture and architecturally complex colonies on solid medium (Branda et al., 2001). A striking feature of these colonies is fruiting body-like aerial structures in which spore formation takes place preferentially at the apical tips. Laboratory strains of B. subtilis, in contrast, form thin, fragile pellicles and flat, featureless colonies.

Previous studies have identified several transcriptional control proteins that help to govern the formation of these multicellular communities, including the sporulation regulatory proteins Spo0A, σH, and the transition state regulator AbrB (Branda et al., 2001; Hamon and Lazazzera, 2001; Hamon et al., 2004). Of special significance, however, is the DNA binding protein SinR, which appears to be a master regulator for the transition from a state in which cells are capable of swimming or swarming to a sessile state characteristic of biofilms in which long chains of cells are bound to each other in tight bundles (Gaur et al., 1991; Kearns et al., 2005). SinR is a direct negative regulator of the 15-gene operon epsA-O, which is believed to be responsible for the production of an exopolysaccharide that holds the chains of cells together in bundles (Kearns et al., 2005). Genetic experiments indicate that SinR is the downstream member of a regulatory pathway that consists of SinI, a known antagonist of SinR, and two proteins of unknown function, YlbF and YmcA (Bai et al., 1993; Branda et al., 2004; Kearns et al., 2005).

In confirmation and extension of the view of SinR as a master regulator for multicellularity, we report that SinR also controls the transcription of a three-gene operon, yqxM-sipW-tasA, that is required for the formation of robust biofilms. We report that SinR binds to multiple sites within the promoter region for the operon, thereby repressing its transcription. Thus, SinR directly controls the transcription of at least 18 genes that are directly involved in the formation of an extracellular matrix that allows cells to adhere to each other in biofilms. Finally, we present evidence from transcriptional profiling that indicates that genes in the yqxM-sipW-tasA and epsA-O operons constitute the majority of the most strongly regulated members of the SinR regulon.

Results

tasA is required for robust biofilm formation

To discover previously unrecognized genes involved in the assembly of multicellular communities, we performed insertional mutagenesis in the wild strain 3610 using the transposon mini-Tn10 and screened for mutants displaying altered colony morphologies and defects in pellicle formation (see Experimental procedures). One of the insertions recovered from our screen was located in tasA, the terminal member of the three-gene operon yqxM-sipW-tasA (Serrano et al., 1999; Stover and Driks, 1999). A transduction experiment using phage SPP1 indicated that the Tn10 insertion and the mutation causing the defect in biofilm formation were apparently inseparable, a finding that reinforced the view that the tasA::Tn10 insertion was indeed responsible for blocking biofilm formation (data not shown).

To investigate further the apparent requirement for tasA in biofilm formation, we created an insertion/deletion (null) mutation by replacing the tasA coding sequence with a drug-resistance gene. The resulting null mutant formed featureless colonies in contrast to the architecturally complex colonies formed by the wild type (compare wild type and tasA in Fig. 1). Moreover, the tasA deletion/insertion mutation could be complemented with either a copy of the entire operon inserted into the chromosome at the amyE locus (Fig. 2A, first complementation construct) or a construct containing the promoter for the operon (hereafter PyqxM) fused to the tasA open reading frame (and similarly inserted in the chromosome at amyE) (Fig. 2A, second complementation construct). We conclude that tasA is required for development of complex colony architecture.

Figure 1.

Mutations in yqxM, sipW and tasA impair colony surface architecture. Shown are 25× images of colonies grown on MSgg agar for 4 days at 22°C. Scale bar is 1 mm. The strains were as follows: wild type (3610), ΔtasA::spec (SSB505), ΔsipW::tet (FC171), ΔyqxM(13234) (FC268).

Figure 2.

Complementation experiments confirm that yqxM, sipW and tasA are each required for robust biofilm formation. Coding regions of the genes are represented by horizontal arrows, PyqxM is shown as a bent arrow, and transcriptional terminators are shown as stem-loop symbols. Coding regions that were replaced with antibiotic-resistance cassettes are depicted as black, horizontal bars, and the in frame deletion of ΔyqxM(13234) is shown as two parallel, slanted lines with ‘IF’ in between. The top portions of panels A, B and C indicate the mutations in each of the genes that were being complemented. The lower portion of each panel shows the complementation constructs that were inserted into the chromosome at the amyE or thrC locus. The + symbol indicates robust complementation, and the – symbol indicates the absence of complementation.
A. The ΔtasA::specR mutant strain was SSB505 and the complementation strains harboured ΔtasA::specR and the inserts amyE:: PyqxM-yqxM-sipW-tasA (FC202) or amyE:: PyqxM-tasA (FC210).
B. The ΔsipW::tetR mutant strain was FC171 and complementation strains harboured ΔsipW::tetR and the inserts: amyE:: PyqxM-yqxM-sipW-tasA (FC224), amyE:: PyqxM-yqxM-sipW (FC223), amyE:: PyqxM-sipW (FC188) or amyE:: PyqxM-ΔyqxM(13234)-sipW (FC255).
C. The mutant ΔyqxM(13234) strain was FC268 and the complementation strain contained ΔyqxM(13234) and thrC:: PyqxM-yqxM (FC270).

Confirmation that sipW is required for biofilm formation

In light of these findings, it was conceivable that the previously observed effects of mutations in sipW on the development of biofilms and complex colony architecture were the consequence of a polar effect on the expression of tasA (Branda et al., 2004; Hamon et al., 2004). As shown in Fig. 1, cells harbouring a newly constructed deletion/insertion mutation in sipW were conspicuously defective in forming architecturally complex communities of cells. Next, we asked whether the mutation could be complemented with constructs lacking tasA. We initially found that the sipW mutation could be complemented by a construct that contained the first two genes of the operon (Fig. 2B, second complementation construct). But a construct that contained sipW (alone) fused to PyqxM failed to restore pellicle formation and colony architecture (Fig. 2B, third complementation construct). However, we noticed that the yqxM and sipW open reading frames overlap by five codons. This observation raised the possibility that the two genes are translationally coupled and hence that translation of sipW mRNA might depend on ribosomes that load at the yqxM ribosome binding site (Adhin and van Duin, 1990). We therefore created a construct corresponding to the first two genes in the operon but containing an in frame deletion within the yqxM open reading frame. That is, the construct contained the PyqxM promoter and the normal translation initiation sequences for yqxM but not the remainder of the yqxM gene except for the 3′ terminal region of overlap with sipW. This construct was able to complement the sipW null mutation (Fig. 2B, fourth complementation construct). In toto, our findings are consistent with the idea that yqxM and sipW are translationally coupled, but, more importantly, the results demonstrate that sipW is indeed required for proper multicellular behaviour.

yqxM is also required for robust biofilm formation

Finally, we created an in frame deletion of the remaining member of the operon yqxM. As seen in Fig. 1, the resulting mutant formed colonies that were smoother, much flatter, and less architecturally complex than the wild type. In addition, the colonies’ aerial structures were noticeable only on the outer perimeter, rather than throughout, as in wild type. To ensure that the mutation was not exerting a polar effect on downstream gene expression, we demonstrated that the in frame deletion of yqxM was complemented by a PyqxM-yqxM construct (Fig. 2C). Thus, each and every member of the yqxM-sipW-tasA operon appears to play a critical role in proper biofilm and fruiting body formation.

SinR inhibits yqxM-sipW-tasA transcription

In previous work, we identified SinR as a transcriptional control protein for the exopolysaccharide operon epsA-O (Kearns et al., 2005). This 15-gene operon is believed to be responsible for the synthesis of an extracellular matrix that bundles chains of cells together during biofilm formation, and its transcription was shown to be under the direct negative control of the DNA binding protein (Branda et al., 2001; Kearns et al., 2005). The finding that all members of the yqxM-sipW-tasA operon were involved in the development of biofilm architecture raised the question of whether, like epsA-O, the operon is under the control of SinR.

As a test of the idea that SinR controls the transcription of the yqxM-sipW-tasA operon, we fused a promoterless lacZ gene to the promoter region of yqxM to create PyqxM–lacZ and introduced the transcriptional fusion into strains mutant for SinR and SinI. We also introduced a mutation in the epsH gene of the exopolysaccharide operon into these strains to alleviate the cell-bundling phenotype caused by a sinR mutation, which prevents accurate measurements of cell number and gene expression (Kearns et al., 2005). The epsH mutation prevented cell clumping while having little or no effect on PyqxM–lacZ expression (Table 1). In a medium that promoted biofilm formation (MSgg), expression of the PyqxM–lacZ fusion was increased almost fivefold in the absence of SinR (Table 1). In contrast, the absence of SinI, an antagonist of SinR, decreased PyqxM–lacZ expression by approximately 100-fold, consistent with prior reports that SinI acts as a negative modulator of SinR (Bai et al., 1993). We conclude that, like epsA-O, the yqxM-sipW-tasA operon is under the transcriptional control of the SinI/SinR regulatory system.

Table 1.  SinR represses expression of the yqxM-sipW-tasA operon.
GenotypeaActivity (MU)b
  • a

    . The following PyqxM–lacZ-containing strains were used FC129, FC134, FC135, FC136, FC205 and FC206.

  • b

    .β-Galactosidase activity, calculated in Miller Units (MU), was measured from late-exponential phase cells grown in MSgg and is the average of three replicas.

Wild type 90 ± 4
ΔepsH 92 ± 25
ΔsinRΔepsH428 ± 54
ΔsinIΔepsH0.6 ± 0.3
Δspo0HΔepsH9.3 ± 1.5
ΔsinRΔspo0HΔepsH284 ± 21

Characterization of the PyqxM promoter

Two observations suggested that the yqxM-sipW-tasA operon is likely to be under the direct negative control of SinR. First, upstream of the translation start site for the first member of the operon, yqxM, are sequences that strongly resemble the previously identified, consensus binding site for SinR (see below; Kearns et al., 2005). Second, the genes for both SinR and SinI are located immediately adjacent to, and in convergent orientation with, the yqxM-sipW-tasA operon (Fig. 2; Gaur et al., 1988; Bai et al., 1993).

Before turning to the question of whether the operon is controlled by SinR directly, we sought to localize the transcriptional start site for the yqxM-sipW-tasA operon. To do this, we mapped the 5′ end (putative transcription start site) of transcripts from the operon using primer extension analysis (Fig. 3B). The apparent transcription start site, designated as +1, was located 60 base pairs upstream of the initiation codon for the yqxM open reading frame. As expected, a higher level of yqxM-derived transcripts were observed in a sinR mutant than in the wild type (Fig. 3B). Sequences that resembled the canonical ‘−10’ (TATAAT) and ‘−35’ (TTGACA) regions for promoters recognized by RNA polymerase containing the house-keeping sigma factor σA were centred 10 and 33 base pairs upstream of the start site (Moran et al., 1982). The putative ‘−10’ region was a three out of six match (TACTCT) to the canonical −10 sequence and the putative ‘−35’ region a four out of six match (TTTAAA) to the canonical −35 sequence. These sequences were separated from each other by a spacing of 17 base pairs, which is optimal for promoters recognized by σA-RNA polymerase (Moran et al., 1982). As a functional test of the putative promoter, we created a deletion of the PyqxM–lacZ fusion that extended from the downstream direction to position −47, hence removing the transcription start site and the putative −10 and −35 sequences (Fig. 4). Consistent with our promoter assignment, the deletion-mutated fusion was incapable of directing lacZ expression.

Figure 3.

Mapping the bindings sites for SinR in the yqxM-sipW-tasA promoter region. Panel A shows the results of a DNase I footprinting experiment. A 231 bp fragment of PyqxM-containing DNA was 5′ end-labelled at the terminus downstream of the promoter, mixed with purified SinR at the indicated concentrations, and treated with DNase I as described in Experimental procedures (panel A). Vertical bars indicate regions of DNA that were protected from DNase I digestion. In a primer extension experiment, RNA was used as a template for extension of the 5′ radiolabelled primer ‘+59yqxM_R’ as described in Experimental procedures (panel B). RNA was purified during the mid-exponential phase of growth from cells of the wild-type strain 3610 (‘wt’ in the figure) and the ΔsinR mutant strain DS92 (‘sinR’ in the figure) grown in MSgg. The extension product corresponding to the putative transcriptional start site of the operon is labelled with an asterisk. The other lanes are dideoxy sequencing ladders produced with the same primer and terminated with ddTTP (T), ddGTP (G), ddCTP (C) and ddATP (A). The DNA sequence corresponding to the two panels is shown between the panels. The DNA sequence is the complement of the sequencing ladder in panel A. The arrow to the right of the sequence indicates the location of the putative transcriptional start site (+1). The numbers to the right indicate the centres of binding sites for SinR and the numbers to the left, along with the dashed boxes indicate the putative ‘−10’ and ‘−35’ sequences.

Figure 4.

Functional dissection of the PyqxM promoter. The figure shows the sequence of PyqxM and the downstream binding sites for SinR. The dashed lines indicate the putative −10 and −35 regions and the solid, black lines indicate the inverted-repeat, binding sites for SinR. A series of deletions were created that extended from the downstream direction to the positions indicated with black, vertical lines and numbering. Each truncated DNA was joined to a promoterless lacZ gene and inserted into the chromosome at the amyE locus. The activity of β-galactosidase (in Miller Units, MU) for each lacZ reporter construct is indicated at the right. Activities were measured at the late-exponential phase of growth and the values shown are the average of three replicas. In the case of the truncation to position +3, activity was measured both in a sinR+ strain (FC187) and in a sinR mutant strain (FC239). The strain harbouring the intact PyqxM–lacZ fusion construct was FC122 and the strain harbouring the lacZ fusion contract that had been truncated to position −47 is FC190.

It was previously reported that expression of the operon is dependent upon the alternative sigma factor σH, the product of the spo0H gene (Serrano et al., 1999; Stover and Driks, 1999). Because the promoter did not contain sequences that resembled those recognized by σH-containing RNA polymerase, we wondered whether the dependence on σH was an indirect effect due to the known (partial) dependence of sinI expression on σH (Gaur et al., 1988; Britton et al., 2002). SinI is an antagonist of SinR and hence a reduction in SinI levels would be expected to interfere with transcription from the PyqxM promoter (Bai et al., 1993). To test this hypothesis, we examined the effect of mutations in spo0H and sinR on expression of PyqxM–lacZ. Consistent with previous reports, the absence of σH resulted in a 10-fold decrease in PyqxM–lacZ expression as compared with the wild type (Table 1; Serrano et al., 1999; Stover and Driks, 1999). However, in the absence of both σH and SinR, expression of the PyqxM–lacZ fusion was approximately the same as that observed in the absence of SinR alone (Table 1). The simplest interpretation of these results is that the yqxM-sipW-tasA operon is not transcribed by σH-containing RNA polymerase and that σH exerts its effect indirectly (presumably through the level of SinI).

SinR binds to the promoter region for the yqxM-sipW-tasA operon

To determine if SinR binds to the promoter region for the yqxM-sipW-tasA operon, we performed electrophoretic mobility shift assays (EMSA) using purified SinR. The results of Fig. 5 show that SinR retarded the electrophoretic mobility of a 231 bp DNA fragment containing the yqxM promoter region with a similar binding affinity to that of DNA containing the epsA-O (Peps) promoter region (Kearns et al., 2005). The presence of multiple species of electrophoretically retarded PyqxM fragments suggests that the promoter region contains at least two binding sites for SinR. As a negative control, SinR did not cause a shift in the mobility of DNA containing the promoter region for a gene (yvbA) that is not under the control of SinR (Fig. 5; Kearns et al., 2005). These results are consistent with the idea that the yqxM-sipW-tasA operon is directly controlled by SinR.

Figure 5.

SinR binds to DNA containing the promoter region for the yqxM-sipW-tasA operon. EMSA in which 5′ radiolabelled DNAs (indicated on the upper left corner of each panel) were mixed with the indicated concentrations of purified SinR. The sizes of the PyvbA-, PyqxM- and PepsA-containing DNAs were 396, 231 and 225 base pairs respectively (see the Experimental procedures). The horizontal arrow indicates the positions of probe not bound to SinR.

SinR binds to sites located upstream and downstream of the transcription start site

DNase I footprinting experiments were carried out to determine the location of SinR binding sites in the PyqxM promoter region. The binding assays were performed using a PyqxM-containing DNA fragment that was radiolabelled at the 5′ terminus downstream of the promoter and incubated with increasing concentrations of purified SinR. DNase I was added to the SinR-bound DNA and the radiolabelled fragments were then subjected to electrophoresis to identify regions that were protected from DNase I cleavage by SinR. In accordance with the EMSA data, the footprinting experiment revealed two regions that were protected by SinR (Fig. 3A): a short and distinct region upstream of the promoter (labelled −70 in Fig. 3A) and a larger slightly less distinct region located downstream of the start site (labelled +7 +16 in Fig. 3A). Both regions contained sequences that were strikingly similar to the sequences found in front of epsA and which were shown to be involved in SinR binding, GTTCTYT (in which Y is C or T) (Kearns et al., 2005; see below). The upstream region, which was centred at position −70, contained the sequence ATTCTTT, which differs from the consensus binding site at one position. The downstream region was a near-perfect inverted repeat containing the sequences GTTCTTT and GTTCTCT, which were centred at positions +7 and +16 respectively. Because these sequences are located immediately downstream of the transcription start site, the binding of SinR to the inverted repeat sequence would be appropriately positioned to occlude RNA polymerase and hence repress transcription.

To investigate the functional significance of the inverted repeat, we constructed a deletion of PyqxM-containing DNA that extended from the downstream direction to position +3. The deletion-mutated DNA was fused to lacZ and inserted into the chromosome at the amyE locus. The resulting deletion-mutated fusion retained the putative ‘−10’ and ‘−35’ sequence elements for the PyqxM promoter but lacked the SinR binding sites located downstream of the start site. As shown in Fig. 4, the level of expression of the deletion-mutated fusion was approximately fivefold higher than that of the full-length PyqxM–lacZ fusion. This high level of expression was largely or completely independent of the presence or absence of SinR (Fig. 4). These results are consistent with the idea that the inverted repeat is an operator site that plays a critical role in SinR-mediated repression of yqxM-sipW-tasA expression. Conceivably, the upstream site contributes to the stability of SinR binding to the inverted repeat through the formation of a DNA loop, but we have not investigated this possibility.

The members of the yqxM-sipW-tasA and epsA-O operons constitute the majority of strongly controlled genes in the SinR regulon

To identify additional genes under SinR control, we performed gene microarray analysis with RNA from cells of a sinI mutant (in which SinR-controlled genes should be repressed) and a sinR mutant (in which SinR-controlled genes should be derepressed) (Bai et al., 1993). Again, and for reasons explained above, an epsH mutation was introduced into the mutants to prevent cell clumping. Strikingly, members of the yqxM-sipW-tasA and epsA-O operons represented a high proportion of the genes that were most strongly controlled by SinR (Table 2). The expression ratios between the derepressed and repressed states ranged from 14 to 50 for members of the yqxM-sipW-tasA operon and between 3.6 and 50 for members of the epsA-O operon. A small number of additional genes also exhibited strong regulation by SinR. These were rapG, spoVG, yvfV, yvfW, yvgN and ywbD, and, as we argue below, they too are likely to be under the direct negative control of SinR (Table 2). Our microarray analysis was carried out with RNA from cells in the exponential phase of growth in Luria–Bertani (LB) medium. Therefore, it is possible that SinR contributes to the regulation of other genes, which due to additional levels of control, were not expressed in LB medium even in the absence of SinR. For example, transcription of the known SinR-controlled gene aprE was not detected under our growth conditions (Gaur et al., 1991). Nonetheless, we believe that our microarray analysis has probably detected many of the direct targets of SinR regulation, at least in cases in which SinR is the principal or exclusive regulator.

Table 2.  Genome-wide identification of genes repressed by SinR.
GeneRatioaFunction
  • a

    . Ratios of relative RNA levels in ΔsinR::specΔepsH::tet (DS207) versus ΔsinI::specΔepsH::tet (DS383) mutants. RNA was harvested from exponential phase cells grown to 1 OD600 in LB broth at 37°C.

  • b

    . Some genes from the eps operon do not appear to be differentially regulated in this analysis, specifically: epsH, epsI, epsJ, epsK, epsL and epsM. Control arrays comparing wild type versus ΔepsH::tet (DS76) mutant reveal that the tetracycline insertion cassette disrupting epsH results in high-level expression of genes immediately downstream (data not shown). We speculate that this high level of expression masks differential gene expression of the genes within the operon but downstream of epsH.

epsA51EPS biosynthesis genes
epsB10EPS biosynthesis genes
epsC21EPS biosynthesis genes
epsD27EPS biosynthesis genes
epsE28EPS biosynthesis genes
epsF 5.1EPS biosynthesis genes
epsGb16EPS biosynthesis genes
epsN 3.6EPS biosynthesis genes
epsO 5.4EPS biosynthesis genes
yqxM14Present work
sipW18Present work
tasA50Present work
rapG 5.9Rap phosphatase
spoVG10Sporulation
yvfV 3.9Similar to glycolate oxidase
yvfW 4.1Unknown
yvgN 4.2Similar to dehydrogenase
ywbD 3.9Unknown

An updated consensus sequence for DNA binding by SinR

The results presented above and those of our previous study make it possible to update the consensus DNA binding sequence for SinR (Kearns et al., 2005). Figure 6A lists all of the biochemically verified binding sequences for SinR on which basis the consensus sequence GTTCTYT is proposed (representing a slight modification of what was proposed previously; Kearns et al., 2005). The results are summarized quantitatively in the sequence logo of Fig. 6B. Importantly, all of the newly identified members of the regulon (rapG, spoVG, yvfV, yvfW, yvgN and ywbD) contain a sequence that closely resembles the consensus binding site in their promoter regions (Fig. 6A), and hence are likely to be under the direct negative control of SinR. The yvfV, yvgN and ywbD genes additionally contain putative SinR binding sites within their open reading frames, which could conceivably contribute to repression through DNA loop formation with SinR bound at the promoter. It will be interesting to see whether some or all of these newly identified regulon members play a role in biofilm formation.

Figure 6.

Consensus and putative SinR binding sites. Panel A lists SinR-regulated genes and their confirmed or putative SinR binding sites. The upper portion of the panel lists genes known from genetic and biochemical experiments to be under the direct control of SinR and their binding sites, which are based on the results of DNase I footprinting analyses in all cases except for aprE. The lower portion of the panel lists putative targets of SinR from the transcriptional profiling data of Table 2 and their putative SinR binding sites. The asterisks (*) indicate sequences that were found in reverse orientation on the coding strand. The double cross (‡) indicates sequences that were found in the open reading frames of their respective genes. The middle of panel A shows the consensus binding sequence, which was compiled from the biochemically confirmed targets in the upper portion of the panel. Panel B depicts the same as a sequence logo. The sequence logo was created using the website http://weblogo.berkeley.edu/.

Discussion

The present work makes two contributions to our understanding of biofilm formation in B. subtilis. First, we have discovered additional genes that are involved in the formation of architecturally complex communities of cells, namely tasA and yqxM. Second, we have discovered that that transcription of the yqxM-sipW-tasA operon is under the direct negative control of the protein SinR. We have shown that SinR is a potent repressor of the operon that works by binding to an operator located downstream of the transcription start site. In previous work we showed that the 15-gene epsA-O operon is also under the direct negative control of SinR (Kearns et al., 2005). The identification of the yqxM-sipW-tasA operon as a second major target of the repressor reinforces the view that SinR is a master regulator that governs biofilm formation. SinR acts downstream of the inhibitory protein SinI and two proteins of unknown function, YlbF and YmcA, in the pathway that governs the switch to multicellularity (Bai et al., 1993; Branda et al., 2004; Kearns et al., 2005). An important challenge in future work will be to elucidate the molecular chain of events that leads to relief from SinR-mediated repression of the epsA-0 and yqxM-sipW-tasA operons.

An unexpected outcome of this investigation is the discovery that the yqxM-sipW-tasA and epsA-O operons constitute many of the most strongly controlled genes in the SinR regulon. We did identify six additional genes that are subject to and are likely to be under its direct control. Conceivably, these newly identified members of the regulon also play a role in biofilm formation. But what is most striking about the results of transcriptional profiling was how few genes were identified other than members of the yqxM-sipW-tasA and epsA-O operons. In particular, we failed to identify any genes that could explain the results of previous reports implicating SinR in the regulation of motility (Sekiguchi et al., 1990; Barilla et al., 1994; Fredrick and Helmann, 1996), sporulation (Gaur et al., 1986; Mandic-Mulec et al., 1992; 1995), competence (Guillen et al., 1989; Liu et al., 1996), extracellular protease (Gaur et al., 1986; 1991; Olmos et al., 1997) and autolysin production (Kuroda and Sekiguchi, 1993; Rashid and Sekiguchi, 1996). Therefore, the contribution of SinR to these processes may in some cases represent secondary effects of the protein. In light of our present findings, we propose not only that SinR is a master regulator for biofilm formation but that this is likely to be one of its principal biological functions.

Voigt et al. (2005) recently proposed a model for the SinI/SinR regulatory system in which SinI-mediated inhibition of SinR and SinR-mediated repression of the gene encoding SinI creates a bistable switch. This model rests on a citation of unpublished results suggesting that SinR binds to sinI, thereby blocking its transcription (Smith et al., 1991). However, the reported binding site shares little similarity to the canonical binding sequence for SinR, and we cannot detect another sequence in or near the sinI promoter that resembles a SinR binding site. The authors also invoke SinR-mediated repression of spo0A in their model, but recent results indicate that SinR has little or no selective affinity for spo0A and spo0A does not exhibit a recognition sequence for SinR either (Kearns et al., 2005). Finally, a newly constructed sinI-lacZ fusion is not expressed at a higher level in the absence of SinR than in its presence (F. Chu, unpubl. results). Hence the model for a bistable switch is unlikely to be valid.

Finally, our investigation underscores the importance of studying biofilm formation in wild strains. The contributions of TasA and YqxM to multicellularity have gone unappreciated in laboratory strains, which form featureless colonies (Branda et al., 2001; Hamon et al., 2004). Likewise, the function of SinR as a master regulator for multicellularity had not been recognized in previous studies, which were carried out with laboratory strains. It may be that some, perhaps many, genes of unknown function are similarly involved in biological processes that have been lost or attenuated during the domestication of this bacterium.

Experimental procedures

Strains and growth conditions

Strains 3610 and PY79 were grown at 37°C in LB broth (10 g tryptone, 5 g yeast extract and 5 g NaCl per litre) or on solid medium containing LB supplemented with 1.5% agar. For formation of architecturally complex colonies, cells were toothpick-inoculated onto MSgg medium (Branda et al., 2001) containing 1.5% Bacto agar and incubated at 22°C for 96 h. When appropriate, antibiotics were included at the following concentrations: 10 µg ml−1 tetracycline, 100 µg ml−1 spectinomycin, 5 µg ml−1 chloramphenicol, 5 µg ml−1 kanamycin and 1 µg ml−1 erythromycin plus 25 µg ml−1 linomycin.

Strain construction

All strains used in this study are listed in Table S1. All insertion/deletion mutants were generated using long flanking homology polymerase chain reaction (PCR) (using primers indicated in Table S2) and transformed into competent PY79 cells. DNA containing a spectinomycin (pDG1726) or tetracycline (pDG1514) drug resistance gene was used as a template for marker replacement (Guérout-Fleury et al., 1995). The marker replacements were then transferred to the 3610 background using SPP1-mediated generalized transduction (Yabsin and Young, 1974; Kearns et al., 2004).

All primers used in the construction of plasmids are listed in Table S2. An in frame deletion of yqxM was created by generating a PCR product containing PyqxM and the first 12 codons of the yqxM open reading frame using primers −500yqxM_EcoRI_F and +59yqxM_SphI_R and 3610 chromosomal DNA as a template. The second PCR product contained the last 19 codons of yqxM and the sipW and tasA genes and was generated using primers +678sipW_SphI_F and tasAcomp_BamHI_R. Using three-way ligation, the two PCR products were cloned into the EcoRI and BamHI sites of the plasmid pDG364, which carries a chloramphenicol-resistance marker and a polylinker between two arms of the amyE gene (Karmazyn-Campelli et al., 1992). The resulting construct was moved into a strain containing an insertion/deletion mutation that replaced the yqxM-sipW-tasA operon with a spectinomycin-resistance gene.

Complementation and reporter gene constructs

All primers used in the construction of plasmids are listed in Table S2. In all PCR reactions, 3610 genomic DNA was used as a template. To generate the PyqxM–lacZ transcriptional fusion, a PCR product containing the PyqxM promoter was amplified using primers −302yqxM_EcoRI_F and −26yqxM_BamHI_R. The PCR product was cloned into the EcoRI and BamHI sites of the plasmid pDG268, which carries a chloramphenicol-resistance marker and a polylinker upstream of the lacZ gene between two arms of the amyE gene (Antoniewski et al., 1990). To generate PyqxM–lacZ truncated to position +3, a PCR product was generated using primers −500yqxM_EcoRI_F and −60yqxM_BamHI_R. The PCR product was cloned into the EcoRI and BamHI sites of plasmid pDG268. To generate PyqxM–lacZ truncated to position −47, a PCR product was generated using primers −500yqxM_EcoRI_F and −106yqxM_BamHI_R. The PCR product was cloned into the EcoRI and BamHI sites of plasmid pDG268. To generate the yqxM complementation construct, a PCR product containing the yqxM gene and promoter region was amplified using primers −500yqxM_EcoRI_F and +768yqxM_BamHI_R. The PCR product was cloned into EcoRI and BamHI sites of plasmid pDG1664, which carries an erythromycin-resistance marker and a polylinker between two arms of the thrC gene (Guérout-Fleury et al., 1996). To generate sipW and tasA complementation constructs, a PCR product containing PyqxM was amplified from 3610 chromosomal DNA using primers −500yqxM_EcoRI_F and −12yqxM_XhoI_R. The sipW and tasA genes were PCR-amplified using primers −18sipW_XhoI_F and +599sipW_BamHI_R for the sipW gene and −17tasA_XhoI_F and tasAcomp_BamHI_R for the tasA gene. Through three-way ligation, the PCR products were cloned into the EcoRI and BamHI sites of plasmid pDG364, which carries a chloramphenicol-resistance marker and a polylinker between two arms of the amyE gene (Karmazyn-Campelli et al., 1992). To create the yqxM-sipW complementation construct, a PCR product was generated using primers −500yqxM_EcoRI_F and +599sipW_BamHI_R. The PCR product was cloned into EcoRI and BamHI sites of plasmid pDG364. A PCR product was amplified using primers −298yqxM_EcoRV_F and tasAcomp_BamHI_R to create a complementation construct containing the whole operon. The PCR product was cloned into plasmid pDG364 at the BamHI site and a blunt-ended site generated with Klenow DNA polymerase I. To create a complementation construct containing an in frame deletion of yqxM upstream of sipW, a PCR product containing PyqxM and the first 12 codons of the yqxM open reading frame was generated using primers −500yqxM_EcoRI_F and +59yqxM_SphI_R. A second PCR product containing the last 19 codons of yqxM and the entire sipW open reading frame was generated using primers +678yqxM_SphI_F and +599sipW_BamHI_R. Through three-way ligation, the two PCR products were cloned into the EcoRI and BamHI sites of plasmid pDG364. All constructs were transformed into competent PY79 cells and then moved into a 3610 background using SPP1-mediated transduction (Yabsin and Young, 1974; Kearns et al., 2004).

SPP1 phage transduction

Serial dilutions of SPP1 phage stock were added to 0.2 ml of dense cultures grown in TY broth (LB supplemented with 10 mM MgSO4 and 100 µM MnSO4) and incubated for 15 min at 37°C. Three millilitres of TY soft agar (TY supplemented with 0.5% agar) was added to the mixture and poured on top of freshly poured TY plates (TY supplemented with 1.5% agar). The plates were incubated at 37°C overnight and the plate containing near confluent plaques was harvested by adding 5 ml of TY to the plate and scraping the agar into a 50 ml conical tube. The tube was vortexed and centrifuged at 5000 g for 10 min. The supernatant fluid was treated with 25 µg ml−1 DNase I for 30 min at room temperature and passed through a 0.45 µm syringe filter before being stored at 4°C.

After being grown to high density, 0.9 ml of recipient cell culture was mixed with 10 µl of SPP1 donor phage stock and 9 ml of TY. The mixture was then statically incubated at 37°C for 30 min and centrifuged at 5000 g for 10 min. The supernatant was discarded and the cell pellet was resuspended in the remaining volume of TY. One hundred microlitres of cells were then plated onto LB fortified with 1.5% agar, 10 mM sodium citrate and the appropriate antibiotic.

Transposon mutagenesis

Transposon mutagenesis was carried out using the 3610 strain DS1010, which contains the temperature-sensitive vector carrying mini-Tn10, pIC333 (Petit et al., 1990; Steinmetz and Richter, 1994; Kearns et al., 2004). To create each transposon library, DS1010 was inoculated into 3 ml of LB broth with spectinomycin and incubated at 22°C in a roller drum for 16 h. The culture was then diluted 20-fold into 2 ml of LB supplemented with spectinomycin and rolled for 3 h. The culture was transferred to 37°C for an additional 4 h. The culture was then diluted and plated onto LB plates containing 1.5% agar and spectinomycin and incubated overnight at 37°C. The resulting colonies were then arrayed onto MSgg plates supplemented with 1.5% agar and incubated at 22°C for 72 h to assay for fruiting body formation. Those colonies unable to form complex architecture characteristic of wild-type colonies were tested for MLS resistance (Kearns et al., 2004). Mutant phenotypes of MLS-sensitive colonies were verified on MSgg plates.

Identifying transposon insertions

Because the mini-Tn10 transposon possesses an Escherichia coli origin of replication, the transposon and flanking DNA was cloned into E. coli strain DH5α for sequencing (Steinmetz and Richter, 1994). Chromosomal DNA was isolated from mutants and 4 µl of DNA was digested using HindIII or EcoRI for 4 h at 37°C. Digests were extracted with phenol-chloroform and ligations were performed in a 200 µl volume at 15°C overnight. After conducting another phenol-chloroform extraction on the ligations, the ligations were transformed into E. coli strain DH5α and the transformants were selected for using spectinomycin resistance. The regions of DNA flanking the transposons were then sequenced using primers ‘my050’ or ‘my051’ that are complementary to the ends of the transposon.

β-Galactosidase assay

Cells at the late-exponential phase of growth (OD600∼2.0) cells were harvested from shaking cultures growing at 37°C in MSgg broth. One millilitre of cells were collected and suspended in an equal volume of Z buffer (40 mM NaH2PO4, 60 mM Na2HPO4, 1 mM MgSO4, 10 mM KCl and 38 mM 2-mercaptoethanol). Lysozyme was added to each sample to a final concentration of 0.2 mg ml−1 and incubated at 30°C for 15 min. Each sample was diluted in Z buffer to a final volume of 0.5 ml and the reaction was started with 100 µl of 4 mg ml−1 2-nitrophenyl β-D-galactopyranoside in Z buffer and stopped with 250 µl 1 M Na2CO3. After centrifuging the samples for 10 min at 15 000 g, the OD420 of the reaction mixture was measured. The β-galactosidase specific activity was calculated according to the equation: [OD420/(time × OD600)]× dilution factor × 1000.

Electrophoretic mobility shift assay (EMSA)

Radiolabelled probes were generated by PCR using 3610 chromosomal DNA and the following primer combinations: PepsAF2/PepsAR2 (Peps), ECH245/ECH246 (PyvbA), −214yqxMF/+17yqxMR (PyqxM). Each probe was PCR purified and 5′ end-labelled with 10 µCi of [γ-32P]-ATP (NEG002A, New England Nuclear) using polynucleotide kinase (New England Biolabs). Various concentrations of purified SinR were added to approximately 100 nM of radiolabelled probe (Kearns et al., 2004). DNA binding reactions were carried out in 30 µl volumes, including binding buffer (10 mM Tris HCl, 50 mM NaCl, 1 mM EDTA, 5% glycerol, 1 mM DTT, 10 µg ml−1 bovine serum albumin) and 25 µg ml−1 polydeoxyinosinic-deoxycytidlyic acid (poly dI-dC), at room temperature for 20 min. A 6% polyacrylamide 0.5× TBE gel was loaded with 10 µl of each binding reaction and resolved for 1 h at 200 mV.

Primer extension assay

Total RNA was isolated from mid-exponential phase cultures grown in MSgg broth. RNA was isolated using the hot acid/phenol method (protocol available at http://mcb.harvard.edu/losick/fawcettpaper/RNAprep.htm). The primer +59yqxM_R was 5′ end-labelled with 40 µCi [γ-32P]-ATP and polynucleotide kinase. The radiolabelled primer (0.2 pmol) was annealed to 20 µg total RNA in a 10 µl reaction volume of 1× first strand buffer (Invitrogen SuperScript First Strand Synthesis System for RT PCR). The annealing reaction was heated at 95°C for 1 min, transferred to 70°C for 10 min and placed on ice for 2 min. The extension reaction was carried out using 5 µl of the annealing reaction, 0.01 M DTT, 1 mM dNTP’s, 1× first strand buffer and 10 units of SuperScript II RNase H- Reverse Transcriptase (Invitrogen). The reaction was incubated for 45 min at 44°C and stopped with 5 µl of formamide loading buffer (80% deionized formamide, 10 mM EDTA, 1 mg ml−1 xylene cyanol FF, 1 mg ml−1 bromophenol blue).

A sequencing ladder was generated to a DNA fragment by PCR using primers −598yqxM_F/+59yqxM_R. The PCR product (2 µl) was denatured using 0.2 N NaOH and 100 mM EDTA for 5 min at room temperature. Ammonium acetate was then added to a final concentration of 0.8 M and then DNA was ethanol precipitated and resuspended in 7 µl water. Sequenase (USB) reaction buffer and 0.4 pmol of radiolabelled primer were added to the denatured DNA. The mixture was heated to 65°C for 2 min and cooled to room temperature for 30 min. To the annealing mixture, 2 µl of 1.5 µM dNTP’s, 1 µl of 0.1 M DTT and 3.2 units of sequenase (USB) were added. After incubating the reaction mixtures at room temperature for 2 min, 3.5 µl of the reaction mixture was added to separate tubes containing 200 pmol dNTP's with 20 pmol of the appropriate ddNTP. The termination reactions were incubated at 37°C for 5 min and stopped with 4 µl of formamide loading buffer.

The primer extension and sequencing reactions were loaded onto an 8% sequencing gel after being heated to 95°C for 5 min (SequaGel Sequencing System, National Diagnostics). A total of 8 µl of the primer extension and 6 µl of the sequencing reaction were resolved on the gel and visualized using phosphoimaging.

DNase I footprinting assay

Primer +17yqxM_R was 5′ end-labelled with [γ-32P]-ATP and polynucleotide kinase. A PCR product of the PyqxM region was generated using 3610 chromosomal DNA as a template and 50 pmol of labelled primer +17yqxM_R and unlabelled primer −214yqxM_F. The PCR product was then PCR purified and the cpm was measured using a scintillation counter. Approximately 30 000 cpm of radiolabelled PCR product was added to varying concentrations of SinR protein in 100 µl footprinting buffer (20 mM Tris pH 8.0, 5 mM MgCl2, 5 mM CaCl2, 0.1 mM DTT, 0.1 mM EDTA, 50 µg ml−1 BSA) with 5 mg ml−1 poly dI-dC and incubated at room temperature for 15 min. Each sample was then incubated for 30 s with 2 µl DNase I (1:50 dilution of 1 U µl−1 stock, Invitrogen) before digestion was stopped by adding 25 µl of stop solution (1.5 M sodium acetate pH 5.3, 20 mM EDTA and 400 µg ml−1 glycogen). The reactions were then ethanol precipitated and resuspended in 8 µl of formamide loading dye (80% deionized formamide, 10 mM EDTA, 1 mg ml−1 xylene cyanol and 1 mg ml−1 bromophenol blue). A total of 6 µl of each reaction was loaded on an 8% sequencing gel and resolved before being visualized using phosphoimaging.

Transcriptional profiling assay

Total RNA was isolated from mid-exponential phase cultures grown in LB broth. RNA was isolated using the hot acid/phenol method (protocol available at http://mcb.harvard.edu/losick/fawcettpaper/RNAprep.htm). The strains used for comparison were DS207 (ΔsinR::spec; ΔepsH::tet) and DS383 (ΔsinI::spec; ΔepsH::tet). A total of 20 µg of RNA from each strain was used to apply to the array. The volume of RNA was brought up to 14 µl using DEPC-treated H2O and 1 µg µl−1 of random primer was added to each sample of RNA. The samples were incubated at 70°C for 10 min and incubated on ice for 2 min. To the appropriate tube, 2 µl of [Cy5]dUTP and Cy3[dUTP] were added along with 12 µl of reaction mix (2.5× Invitrogen First Strand buffer, 25 mM DTT, 2 mM ATP, 2 mM CTP, 2 mM GTP and 0.8 mM TTP, 100 units Roche Protector RNase inhibitor). After addition of [Cy5]dUTP and [Cy3]dUTP, the reactions were shielded from light for the rest of the procedure. To the reaction mixture, 200 units of SuperScript II RNase H-free reverse transcriptase (Invitrogen) were added. The tubes were incubated at room temperature for 10 min, then placed at 42°C for 2 h and 70°C for 15 min. After incubating the reactions, 3 units of RNase H (Invitrogen) was added and the mixtures were placed at 37°C for 20 min. The RNA samples to be compared were combined and a Qiagen PCR purification kit was used to purify the cDNA probes. A Speedvac was then applied to the samples until 5 µl of probe remained. To the remaining probe, 20 µg yeast tRNA, 5× SSC, 0.5% SDS and 33% formamide was added. The reactions were heated to 100°C for 1 min, spun briefly, and applied to an oligoarray containing >  99% of the annotated protein coding genes of the B. subtilis genome (Britton et al., 2002). Prior to hybridization, the slides were washed in rinsing solution (0.2% SDS) and blocked using aldehyde blocking solution (1.0 g NaBH4, 0.75× PBS and 25% ethanol). The slides were incubated in rinsing solution for 4 min, transferred for two rinses of 2 min each in H2O, 15 min in aldehyde blocking solution, two washes of 2 min each in rinsing solution, and lastly two rinses of 2 min each in H2O. The slides were they dried before applying probe. Hybridizations were performed overnight in CMT-hybridization chambers (Corning) plunged in a water bath at 42°C. After hybridization, unbound probe was washed off the slides by a 1 min incubation in wash buffer 1 (2× SSC and 0.2% SDS), 1 min in wash buffer 2 (2× SSC), and 1 min in wash buffer 3 (0.2× SSC). Finally, the slides were dried and scanned on a GenePix 4000B scanner (Axon Instruments). Images were processed and analysed with GenePix 4.0 software (Axon Instruments). Spots containing low signals were excluded from further analysis. In our final dataset, we included only genes in which there was a threefold change in expression, as seen using data from four replicates of the transcriptional profiling experiment.

Acknowledgements

We thank members of the Losick and Kolter laboratories, and, in particular, B. Gorbatyuk, J. Silvaggi and C. Ellermeier for helpful advice and discussions. This work was supported by an NSF Graduate Research Fellowship to F.C., NIH National Research Service Award GM66612 to D.K., a Postdoctoral Fellowship from the Charles A. King Trust (Bank of America, Boston MA) to S.S.B., NIH grant GM58213 to R.K, and NIH grant GM18568 to R.L.

Ancillary