Post-translational control of Bacillus subtilis biofilm formation mediated by tyrosine phosphorylation


E-mail; Tel. (+44) (0) 1382 386335; Fax (+44) (0) 1382 388216.


A biofilm is a complex community of cells enveloped in a self-produced polymeric matrix. Entry into a biofilm is exquisitely controlled at the level of transcription and in the Gram-positive organism Bacillus subtilis it requires the concerted efforts of three major transcription factors. Here, we demonstrate that in addition to transcriptional control, B. subtilis utilizes post-translational modifications to control biofilm formation; specifically through phosphorylation of tyrosine residues. Through our work we have assigned novel roles during biofilm formation to two proteins; the protein tyrosine kinase PtkA and the protein tyrosine phosphatase PtpZ. Furthermore by introducing amino acid point mutations within the catalytic domains of PtkA and PtpZ we have identified that the kinase and phosphatase activities, respectively, are essential for function. PtkA contains a conserved C-terminal tyrosine cluster that is the site of autophosphorylation; however, our in vivo analysis demonstrates that this domain is not required during biofilm formation. With the aim of identifying the target(s) of PtkA controlled during biofilm formation we used a systematic mutagenesis approach but, despite extensive efforts, it remained elusive. Our findings highlight the complexity of biofilm development by revealing an additional level of regulation in the form of protein tyrosine phosphorylation.


Bacteria are capable of ‘multicellular’ behaviours that benefit the bacterial community as a whole. These processes are diverse in both nature and utility. This can be seen using Bacillus subtilis as an example where the processes encompass the production of exoproteases (Veening et al., 2008), swarming motility (Kearns and Losick, 2003), genetic competence (Dubnau, 1991; Bai et al., 1993; Hamoen et al., 2000), sporulation (Piggot and Hilbert, 2004) and biofilm formation (Branda et al., 2001; Hamon and Lazazzera, 2001). A biofilm is a complex population of cells, which adhere to a solid surface or interface that are encased in a protective self-secreted polymeric matrix (Branda et al., 2005). In the natural environment bacteria exist predominantly in biofilms, and in such communities, contribute towards the majority of chronic infections (Costerton et al., 1995). As well as being important in clinical and industrial settings, biofilm formation presents an ideal system for investigating the complex network of transcription events used by bacteria while co-ordinating cell fate decisions (Lopez and Kolter, 2010).

Biofilm formation by B. subtilis can be manifested in the laboratory as the formation of a structured pellicle at the air–liquid interface or as a colony with complex architecture on nutrient agar (Branda et al., 2001). The initiation of biofilm formation is dependent on three key transcription factors, namely ComA (Lopez et al., 2009a), Spo0A (Branda et al., 2001; Hamon and Lazazzera, 2001) and DegU (Stanley and Lazazzera, 2005). ComA is activated by phosphorylation at high cell density and triggers the production of the lipopeptide surfactin (Nakano and Zuber, 1991). The release of surfactin into the environment triggers the activation of Spo0A by phosphorylation (Lopez et al., 2009b). Phosphorylation of Spo0A is mediated by a phosphorelay system that is, in addition, stimulated by starvation (Piggot and Hilbert, 2004). Spo0A∼P regulates the expression of approximately 120 genes according to the specific affinity of the promoter element for Spo0A∼P (Fawcett et al., 2000; Molle et al., 2003; Fujita et al., 2005). This allows the activation of different subgroups of genes at different levels of Spo0A∼P (Fujita et al., 2005). The main role of Spo0A∼P during the activation of biofilm formation is to facilitate the activation of transcription from the epsA–O and yqxM–sipW–tasA operons that are required for the biosynthesis of the extracellular matrix (Lopez et al., 2009c). Spo0A∼P achieves this by removing the directly mediated inhibition of transcription exerted by two repressor proteins called SinR and AbrB (Kearns et al., 2005; Chu et al., 2006; 2008; Murray et al., 2009a).

DegU is a response regulator that is activated by phosphorylation through DegS, its cognate histidine sensor kinase, in response to various environmental signals (Mukai et al., 1990; Steil et al., 2003; Murray et al., 2009b). Changes to the phosphorylation state of DegU alter its ability to bind to different promoter regions and thus it can regulate incompatible processes such as protease production and motility (Kobayashi, 2007; Verhamme et al., 2007; Tsukahara and Ogura, 2008). DegU∼P activates the transcription of yuaB and yvcA that encode two proteins that are needed for biofilm formation (Kobayashi, 2007; Verhamme et al., 2007; 2009).

In addition to controlling biofilm formation, SinR and DegU are involved in the regulation of other multicellular behaviours in B. subtilis (Gaur et al., 1986; 1991; Kodgire et al., 2006) (Mukai et al., 1990; Amati et al., 2004; Shimane and Ogura, 2004). This knowledge suggested to us that there may be an intimate link between the two regulators. As a consequence of initial experiments designed to investigate this, we have identified that biofilm formation by B. subtilis is controlled at the post-translational level by protein tyrosine phosphorylation. Modulation of protein activity using phosphorylation of serine/threonine/tyrosine (Ser/Thr/Tyr) residues was initially believed to be confined to eukaryotes (Deutscher and Saier, 2005). Prokaryotes were thought to be limited to phosphorylation of histidine and aspartic acid residues; a key signalling feature of two-component signal transduction systems (Hoch, 2000). However, this paradigm was revoked in the late 1970's by the identification of Ser/Thr phosphorylation in bacteria (Wang and Koshland, 1978; Manai and Cozzone, 1979) and again when tyrosine phosphorylation was detected in the mid 1990's (Duclos et al., 1996). Serine, threonine and tyrosine phosphorylation has subsequently emerged as an important mechanism for mediating the dynamic control of diverse cellular processes in bacteria, including polysaccharide biosynthesis (Morona et al., 2000; Grangeasse et al., 2003; Soulat et al., 2007), DNA metabolism (Petranovic et al., 2007), cell division (Wu et al., 1999) and resistance to antimicrobial compounds (Lacour et al., 2008).

Bacterial tyrosine kinases are structurally different from their eukaryotic counterparts and thus have been proposed as novel targets for antimicrobial compounds (Lee et al., 2008; Olivares-Illana et al., 2008; Grangeasse et al., 2009). Therefore, understanding the mechanism(s) and role(s) of modification of tyrosine residues by phosphorylation in bacteria is important. Like eukaryotic tyrosine kinases, bacterial tyrosine kinases (hereafter BY-kinases) are ATP-dependent; but the active site of ATP binding and hydrolysis is not conserved between the eukaryotic and prokaryotic proteins (Lee and Jia, 2009). Furthermore, BY-kinases possess unique conserved motifs that are needed for catalytic activity, which are absent from the eukaryotic tyrosine kinases and, as such, could represent potential targets for small molecule inhibitors (Grangeasse et al., 2009; Lee and Jia, 2009). In Gram-negative bacteria BY-kinases are a single polypeptide where the kinase catalytic activity and modulator domains are fused (Mijakovic et al., 2005a). In contrast, in Gram-positive bacteria the BY-kinase is split into two proteins where the genes encoding the modulator and kinase domains are colocated on the chromosome and cotranscribed (Grangeasse et al., 2007). In Gram-positive bacteria the C-terminal domain of the membrane bound modulator protein interacts with the N-terminal region of the cytoplasmic kinase to allow activation (Mijakovic et al., 2003; Grangeasse et al., 2009).

Bacillus subtilis contains one confirmed and one predicted BY-kinases known as PtkA and EpsB respectively. EpsB (also known as PtkB) is a predicted BY-kinase and is encoded in the 15 gene operon epsA–O that is required for the production of a secreted exopolysaccharide (EPS) needed for biofilm formation (Branda et al., 2001; Mijakovic et al., 2003). PtkA is a confirmed and promiscuous BY-kinase (Mijakovic et al., 2003; Jers et al., 2010). B. subtilis also possesses three bacterial protein tyrosine phosphatases. First, there is PtpZ that belongs to the polymerase and histidinol phosphatase family and functions to dephosphorylate the targets of PtkA (Mijakovic et al., 2003; 2005b). Second, B. subtilis possesses two tyrosine phosphatases belonging to the low molecular weight protein tyrosine phosphatase family, YwlE and YfkJ, that modulate a distinct set of targets from PtpZ (Mijakovic et al., 2005a). Our study began with an investigation into the link between SinR and DegU and the impact that simultaneous deletion of these regulators had on biofilm formation. These experiments led to the discovery of a novel role for tyrosine phosphorylation mediated by PtkA during biofilm formation by B. subtilis. We go on to demonstrate that the known and/or predicted substrates of PtkA are not targeted during post-translational regulation of biofilm formation.

Results and discussion

DegU and SinR function in parallel to control biofilm formation

Recently we hypothesized that DegU and SinR were part of two separate but converging pathways that are required for biofilm formation by B. subtilis (Verhamme et al., 2009). Epistasis analysis presented here supports this conclusion. Mutations in sinR and degSU were introduced separately, and in combination, into the B. subtilis isolate NCIB3610. Colony architecture and pellicle formation were used as two independent measures of the capability to form a biofilm (Branda et al., 2001). The phenotype exhibited by the following strains was assessed with respect to the wild-type parental strain (NCIB3610); degSU (NRS1499), sinR (NRS1859) and degSU sinR (NRS1860). The degSU sinR (NRS1860) double mutant exhibited a morphology that was distinct from the wild-type strain and was dissimilar from either of the previously characterized single mutant strains, degSU (NRS1499) and sinR (NRS1859) (Fig. 1A). Consistent with previous findings, the degSU mutant strain was unable to form a robust stable pellicle (Stanley and Lazazzera, 2005; Kobayashi, 2007; Verhamme et al., 2007) and the sinR mutant formed a very robust rugose pellicle (Fig. 1B) (Kearns et al., 2005). The degSU sinR double mutant formed a pellicle that was distinct from either single mutant strain or wild-type strain (Fig. 1B).

Figure 1.

Complex colony architecture assay of B. subtilis.
A. Strains 3610, degSU (NRS1499), sinR (NRS1859) and degSUsinR (NRS1860) were grown on MSgg agar plates for 40 h at 37°C prior to photographing. Scale bar represents 5 mm.
B. Pellicles were allowed to grow in MSgg media overnight at 37°C prior to photography.
C. Strains as detailed above were patched onto an LB agar plate and grown for 16 h at 37°C prior to scraping and photography.

DNA microarray analysis identifies subset of genes co-regulated by DegU and SinR

It was noted that by comparison with both the wild-type and single mutant strains, the double degSU sinR mutant (NRS1860) tightly adhered to the surface of an agar plate (Fig. 1C). This led to the hypothesis that a novel set of DegU and SinR co-regulated genes existed and that one (or more) was responsible for the enhanced adhesion phenotype. It also provided a possible mechanism to identify novel loci involved in the control of biofilm formation. To identify these potential gene(s), DNA microarray analysis was used. As the number of genes controlled by DegU is significantly greater than those controlled by SinR, the expression profile of the degSU strain was compared with that of the degSU sinR strain (see Experimental procedures) (Dartois et al., 1998; Ogura et al., 2001; Tsukahara and Ogura, 2008) (Kearns et al., 2005; Chu et al., 2006; Kodgire et al., 2006). By identifying loci that were upregulated more than twofold in the degSUsinR strain compared with the degSU strain, a combination of putative DegU and SinR co-regulated genes, and known SinR-regulated genes, was generated (Table S1). A finalized list of 11 DegU and SinR co-regulated genes was compiled by removing known SinR-repressed genes (Table S1). For a subset of the genes (ptpZ, yomI, ywbC and yvfO) semi-quantitative reverse transcription polymerase chain reaction (RT-PCR) was used to verify the DNA microarray results. An increase in transcription that was at least threefold higher in the degSU sinR mutant versus the wild-type strain (P < 0.05) was observed confirming that transcription was elevated in the absence of DegU and SinR (Fig. S1).

Screening of genes for a role in biofilm formation

We next tested if one of the genes identified through the DNA microarray analysis was responsible for the enhanced surface adhesion exhibited by the degSU sinR strain. To do this, a mutation in each gene was constructed and introduced into the wild-type and degSU sinR mutant strains (Table S3). None of the mutations screened were successful in relieving the enhanced cell-surface adhesion phenotype (data not shown). It is likely that a combination of mutations would be required to relieve this phenotype but this was not tested. In parallel, taking into account the role that DegU and SinR have in controlling biofilm formation (Kearns et al., 2005; Kobayashi, 2007; Verhamme et al., 2007) the impact of deleting each gene on biofilm formation was assessed. The biofilm morphology after 24 and 48 h incubation was monitored. Mutation of 10 of the 11 genes identified by DNA microarray analysis did not affect biofilm formation but deletion of ptpZ (NRS1827) altered the architecture of the biofilm by comparison with the wild-type strain (Fig. S2). For the remainder of this study, focus was devoted to the characterization of the tkmA operon and its members to which ptpZ was shown to belong (Fig. 2).

Figure 2.

Transcriptional analysis of the tkmA operon.
A. 5′RACE analysis was used to identify the transcriptional start site of the tkmA operon. Genes are represented by hollow arrows containing the gene name in italics. A black, stalked arrow schematically indicates the start position of transcription. A stalked, hollow circle indicates the translational stop site. A section of the tkmA promoter sequence can be found directly below the operon structure schematic. Positive and negative numbers refer to the position of the nucleotide with reference to the translational start site, designated as +1. The translational start codon ATG is highlighted by a grey box. Conserved sequences, including the ribosome binding site (RBS), −10 TATAA box and −35 box, are outlined in black. The proposed transcriptional start site (−33 bp downstream of the translational start site) is highlighted by an asterisk. The capital letters M, G, E indicate the translated sequence of TkmA. A DNA sequencing readout showing the predicted transcriptional start site preceded by an engineered poly-G tail can be seen directly below the DNA sequence of tkmA.
B. RT-PCR analysis of the tkmA operon using cDNA synthesized from degSUsinR (NRS1860) RNA. The level of transcription shown was obtained by comparison with wild-type levels of transcription and normalized to the transcription of the control gene veg. The dashed line represents the transcriptional level of 1 in the wild-type strain, 3610. The asterisks indicate genes whose transcription is significantly different from wild-type transcription levels (P-value < 0.05) where n ≥ 3. The error bars represent standard error of the mean.
C. Amplification of regions of DNA internal to tkmA, ptpZ and ugd was amplified from cDNA generated using an ugd-specific reverse primer. Lane 1 contains 100 bp ladder (NEB Biolabs™). Lanes 2 and 3 contain tkmA; lanes 4 and 5 contain ptpZ; and lanes 6 and 7 contain ugd PCR product. Lanes 2, 4 and 6 show PCR products amplified from cDNA template, while lanes 3, 5 and 7 show PCR products amplified from gDNA template. Primer combinations used for amplification of tkmA, ptpZ and ugd can be found in Table S4.

Organization of the tkmA locus

The gene encoding PtpZ is found within a region of the chromosome including tkmA, ptkA and ugd. Initially we wanted to establish if these four genes formed an operon. This was necessary to establish as: (i) the DNA microarray analysis only identified ptpZ as differentially expressed, (ii) ugd is not colocated in the same region of the genome in other Gram-positive bacteria (data not shown) and (iii) there is a 198 bp gap between the stop codon of ptpZ and the start codon of ugd that could contain a promoter element (Fig. 2A). We tested whether, as seen for ptpZ (Fig. S2), transcription of tkmA and ugd was upregulated in the absence of DegU and SinR. We used RT-PCR analysis with cDNA synthesized from RNA isolated from the wild-type (NCIB3610) and the degSUsinR mutant strain (see Experimental procedures). For tkmA, ptpZ and ugd, respectively, a 10-fold, sevenfold and threefold increase in transcription was observed in the absence of SinR and DegU by comparison with the wild-type strain (Fig. 2B). These findings are consistent with the genes being co-regulated and suggest that the DNA microarray analysis was the limiting factor in the identification of tkmA, ptkA and ugd as differentially expressed in the absence of degU and sinR.

tkmA, ptkA, ptpZ and ugd are cotranscribed

To determine whether tkmA, ptkA, ptpZ and ugd were encoded on a single transcript, mRNA was isolated from wild-type B. subtilis and cDNA was synthesized using a reverse ugd gene-specific primer (Fig. 2C). The cDNA generated corresponded to the ugd transcript and was used as a template for PCR with three primer pairs that were specific to the internal coding regions of tkmA, ptpZ and ugd. As a negative control, a sample that had been subjected to the same treatment but which lacked the reverse transcriptase was used to ensure that the PCR products were not the result of any contaminating genomic DNA (data not shown). As a positive control, genomic DNA was used to ensure the product amplified was the correct size (Fig. 2C). All three gene-specific PCR products were amplified from the cDNA indicating that tkmA, ptkA, ptpZ and ugd are cotranscribed (Fig. 2C).

Location of the transcription start site

To determine the transcription start site of the tkmA operon 5′RACE was used and the start site was localized to either −33 or −32 bp upstream of the tkmA coding sequence initiation codon (Fig. 2A). It was not possible to distinguish this more specifically as the anchor primer that is needed for the 5′RACE (NSW367) contains a series of guanine residues, which cannot be distinguished from guanine residues in the amplified promoter region (Fig. 2A). We, however, favour the guanine located at −33 bp upstream from the initiation codon of tkmA for the transcriptional start site. Our reasoning being that there is a perfect −10 binding site (TATAAT) for the housekeeping RNA polymerase Sigma-A centred 10 bp upstream from this nucleotide (Fig. 2A). Additionally a sequence with a four out of six base pair match (TGGAGA) to the canonical −35 binding site (TTGACA) for the housekeeping RNA polymerase Sigma-A is centred at 35 bp upstream from the nucleotide (Fig. 2A) (Moran et al., 1982). When 5′RACE was applied to the region upstream from the ugd translation start site, no transcription initiation site could be identified (data not shown). Taken together the simplest explanation is that the tkmA–ptkA–ptpZ–ugd operon is a single transcript that is upregulated in the absence of DegU and SinR and driven from a single transcriptional start site at 33 bp upstream from the translation initiation codon of tkmA (Fig. 2A).

Electrophoretic mobility shift analysis

We wanted to establish whether DegU and/or SinR directly regulated transcription from the tkmA promoter. To test this we used in vitro electrophoretic mobility shift assays with purified and phosphorylated DegU-His6 and purified SinR-His6. Prior to EMSA analysis we confirmed that purified DegS-His6 was capable of phosphorylating DegU-His6in vitro using [γ-32P]-ATP (data not shown). A 542 bp region of DNA upstream from the translation start site of tkmA was amplified by PCR. The results showed that the mobility of the promoter region of tkmA was impeded by the addition of 1 µM DegU∼P during gel electrophoresis (Fig. 3A). Greater retention was seen with increasing amounts of DegU∼P (3 µM). Unlabelled DNA corresponding to the tkmA promoter was able to out-compete access to the binding domain of DegU∼P on the radiolabelled DNA indicating that the interaction was specific (Fig. 3A). Consistent with this a region of DNA corresponding to the internal coding region of tkmA was not able to compete for the binding domain of DegU∼P (Fig. 3A). As transcription from the tkmA promoter is upregulated in the absence of degU (Fig. S1) we conclude that DegU∼P functions as an inhibitor, and not an activator, of transcription. It is important to note that while DegU activates biofilm formation (Stanley and Lazazzera, 2005; Kobayashi, 2007; Verhamme et al., 2007) DegU also functions to inhibit biofilm formation (Verhamme et al., 2007). These findings add to the loci regulated by DegU∼P during the formation of a biofilm (Kobayashi, 2007; Verhamme et al., 2007; 2009).

Figure 3.

Electrophoretic mobility shift analysis of the tkmA operon promoter region. DNA binding reactions were conducted with γ-32P labelled DNA corresponding to the region −500 to +42 bp. Labelled DNA (1 ng) was loaded into each lane, with or without purified (A) DegU∼P-His6 or (B) SinR-His6. Further binding reactions include the addition of 100 ng of unlabelled tkmA promoter region PCR product (PtkmA) (−500 to +42 bp fragment) or unlabelled internal tkmA (i-tkmA) (+3 to +379 bp) to the EMSA reaction.

In contrast, SinR-His6 did not interact with the tkmA promoter DNA (Fig. 3B). The activity of purified SinR-His6 protein was established by the presence of a positive interaction between the known SinR-regulated promoter, PyqxM and the purified SinR-His6 (Chu et al., 2006) (data not shown). These results were as anticipated as in silico analysis of the tkmA promoter region did not identify any DNA sequences corresponding to the SinR consensus binding (Chu et al., 2006). The inhibitory action exerted by SinR on transcription from the tkmA promoter is therefore unlikely to be directly mediated (Figs S1 and 3B). However, we cannot exclude the possibility that the increase in transcription may in fact be due to pleiotropic effects on gene expression that occur when sinR and degU are deleted in combination. In addition, we cannot rule out the possibility that SinR may require an additional transcriptional regulator in order to exert its regulatory effect on the transcription of the tkmA operon. Such a situation exists for the lytABC operon where SlrR and SinR are required for a protein–DNA interaction to be established (Chai et al., 2010). It is also interesting to note that transcription of the tkmA operon is also directly regulated by Spo0A∼P (Fawcett et al., 2000; Molle et al., 2003). This provides the second example of a locus involved in biofilm formation where the Spo0A and DegU regulatory cascades converge; the first being the promoter element of yuaB (Verhamme et al., 2009).

Construction and analysis of in frame deletion strains

Having defined the structure of the tkmA operon we wanted to establish the role of tyrosine phosphorylation during the formation of a biofilm. The initial ptpZ mutant that was constructed as part of the screening process was generated using single cross-over disruption (Fig. S2). As tkmA–ptkA–ptpZ–ugd form an operon (Fig. 2), in frame deletions in ptkA and ptpZ were made to circumvent downstream effects on transcription and/or the stoichiometry of the remaining proteins (see Experimental procedures). It has been reported that PtkA and PtpZ play a role in controlling DNA replication (Petranovic et al., 2007) and that a ptpZ mutant exhibits a severe growth defect (Petranovic et al., 2007). A growth defect would render interpretation of biofilm formation ambiguous; therefore, we tested whether either the ΔptkA strain or the ΔptpZ strain had a reduced growth rate. In both Luria–Bertani (LB) (data not shown) and MSgg liquid medium the ΔptkA and ΔptpZ mutants had a doubling time equal to that of the parental strain (Fig. 4A). Absence of PtkA has been shown to have deleterious effects on cell length and nucleoid positioning (Petranovic et al., 2007). Therefore, we analysed cell size and nucleoid distribution at the single cell level using microscopy. Deletion of neither ptkA nor ptpZ influenced cell size or nucleoid positioning (Fig. 4B and C). In toto, in NCIB3610, deletion of either ptkA or ptpZ does not result in a gross growth or DNA replication defect. It is possible that differences in either the method of strain construction or the genetic background may explain this discrepancy.

Figure 4.

Growth and nucleoid localization of ΔptkA and ΔptpZ mutants compared with 3610. (A) Growth of 3610, ΔptkA (NRS2544) and ΔptpZ (NRS2222) in MSgg media. Results are representative of > 3 independent rounds. Phase-contrast and fluorescent images of NCIB3610, ΔptkA (B) and ΔptpZ (C) collected using Zeiss Axio Imager M1. Scale bars represent 5 µm.

Tyrosine phosphorylation mediated by PtkA is required for biofilm maturation

Having established that the ΔptkA and ΔptpZ mutants did not exhibit a growth defect, the impact of deleting ptkA and ptpZ on biofilm formation was tested. Three independent indicators of biofilm formation were assessed; namely complex colony architecture, pellicle formation and the ability to form environmentally resistant spores (Branda et al., 2001; Vlamakis et al., 2008). All of these processes have been linked with the biosynthesis and assembly of the extracellular matrix. Starting with the ΔptkA strain (NRS2544); we determined that it was unable to develop the complex radial structures typical of the mature B. subtilis biofilm on semi-solid agar (Fig. 5B). In contrast to the reduction in colony architecture complexity (Fig. 5B), the pellicle formed by the ΔptkA strain showed extensive three dimensional structural complexity but it entirely lacked the ‘fruiting body structures’ that are present on the wild-type pellicle (compare Fig. 5E with F). The altered colony morphology and lack of ‘fruiting body’ formation on the pellicle correlated with the decreased ability of the ΔptkA strain to sporulate. After 24 h incubation under biofilm formation conditions 13 ± 2.5% of the ΔptkA mutant population had sporulated by comparison with the wild type that exhibited 25 ± 4% sporulation (n = 3; P < 0.06 Student's two-way t-test). After 72 h the difference in the sporulation efficiency widened such that only 37 ± 10% of the ΔptkA cells in the colony had sporulated by comparison with 103 ± 9% of the wild-type strain (n = 3; P < 0.01 Student's two-way t-test). In contrast after 72 h planktonic growth there was no difference between the sporulation levels of the wild-type strain and that of ΔptkA mutant indicating that the decrease in sporulation did not reflect a reduction in the sporulation efficiency per se (data not shown) (Vlamakis et al., 2008). To confirm that the biofilm morphology of the ΔptkA mutant was specific to the absence of PtkA, a wild-type copy of ptkA was introduced into the amyE locus under the control of a heterologous promoter; Pspank-hyptkA–lacI (NRS2804). The biofilm architecture was fully restored compared with that shown by the wild-type strain both in the absence of IPTG and in the presence of 10 µM IPTG (compare Fig. 5B with G). These findings confirm that the altered biofilm phenotype was specific to the disruption of ptkA.

Figure 5.

Complex colony architecture and pellicles of ΔptkA. Strains 3610 (A) and ΔptkA (B) (NRS2544) were assessed as described in Fig. 1. Scale bar represents 5 mm (A–D) and 1 mm (E and F). (G) ΔptkA + amyE–Pspank-hy–ptkA–lacI (NRS2804) was grown on MSgg agar with and without the addition of 10 µM IPTG. The scale bar represents 5 mm.

The kinase activity of PtkA is required for biofilm formation

Bacterial tyrosine kinases (BY-kinase) share sequence conservation and recent structural analysis of a PtkA homologue showed that the Walker A, Walker B and ‘DxD’ motifs are required to co-ordinate and stabilize four water molecules and an Mg2+ ion into the active site (Fig. 6A) (Olivares-Illana et al., 2008). Therefore, the activity of PtkA is dependent on binding and hydrolysis of ATP coupled to the transfer of the released phosphate moiety onto a target substrate protein(s) (Mijakovic et al., 2003). We aimed to confirm whether or not the catalytic kinase activity of PtkA was needed during biofilm formation. To achieve this, two strains were made that resulted in substitutions on the chromosome leading to alterations in the PtkA amino acid sequence: (i) an amino acid point mutation in the second of the aspartic acid residues in the conserved ‘DxD’ motif required for ATP hydrolysis catalysis; (PtkA D83A; NRS2795) and (ii) a dual ‘DxD’ aspartic acid to alanine substitution; (PtkA D81A and D83A; NRS2796). The nucleotide substitutions were introduced into the chromosome using pMAD (Arnaud et al., 2004) and in each case an isolate that had been subjected to the mutagenesis but had retained the wild-type sequence was retained as a control (see Experimental procedures). Each of the negative control strains exhibited wild-type morphology (data not shown). Consistent with the ‘DxD’ motif being required for the activity of PtkA, mutation to ‘AxA’ resulted in a biofilm morphology that was indistinguishable from the complete in frame ΔptkA strain (NRS2544) (compare Figs 5B with 6D). Additionally consistent with the altered colony morphology 32 ± 9% of the mutant cell population sporulated after 72 h incubation by comparison with 103 ± 9% of the wild-type cells after 72 h incubation (n = 3; P < 0.006 Student's two-way t-test). The specificity of the ‘DxD’ point mutations could be demonstrated as wild-type morphology was restored upon introduction of the Pspank-hyptkA–lacI allele at the amyE locus (NRS2807) (compare Fig. 6D with E). A strain carrying the single ‘DxD’ alanine substitution had a phenotype that was intermediate between the parental strain and the ΔptkA (compare Fig. 6C and D with Fig. 5B); thus underscoring the importance of both aspartic acid residues in the BY-kinase (Mijakovic et al., 2003).

Figure 6.

Complex colony architecture of ΔptkA point mutation strains. (A) Schematic representation of PtkA protein sequence. Conserved motifs are highlighted by coloured boxes containing the specific motif residues. Amino and carboxyl termini are indicated by N and C respectively. Strains (B) 3610, (C) ptkA D83A (NRS2795), (D) ptkA D81A–D83A (NRS2796), (E) ptkA D81A–D83A + Pspank-hyptkA (NRS2807) and (F) ΔptkA Y225A–Y227A–Y228A (NRS2799) were grown on MSgg agar plates with and without the addition of 10 µM IPTG as required for 40 h at 37°C prior to photographing. Scale bar represents 5 mm.

It has been previously noted that in addition to its role as a bacterial tyrosine kinase, PtkA also exhibits ATPase activity (Mijakovic et al., 2003). To (indirectly) distinguish whether the ATPase or kinase activity of PtkA was responsible for its role in biofilm formation, an in frame deletion of tkmA was constructed (NRS3541) (Table 1). TkmA is the cognate BY-kinase modulator that interacts with PtkA allowing it to phosphorylate its target proteins (Mijakovic et al., 2003; Jers et al., 2010). Thus in the absence of TkmA, PtkA is unable to exhibit kinase activity and can no longer phosphorylate substrate proteins but its ATPase activity remains intact (Mijakovic et al., 2003). Upon deletion of tkmA we observed a reduction in colony complexity and a loss of complex aerial structures and fruiting bodies (Fig. 7C); consistent with TkmA (and thus PtkA) being required for biofilm formation. The heterologous expression of tkmA under the control of the Pspank-hy promoter was able to complement the tkmA mutation (in the absence of IPTG) (Fig. 7D). Taken together this demonstrates that TkmA is required for biofilm formation and supports the hypothesis that the kinase activity of PtkA is essential for its role during biofilm formation.

Table 1.  Strains used in this study.
StrainRelevant genotypeaSource/constructionb
  • a. 

    Antibiotic resistance cassettes are indicated as follows: spc, spectinomycin resistance; kan, kanamycin resistance.

  • b. 

    Arrow indicates direction of strain construction. Plasmids (DNA) and SPP1 (phage) were used to transform or transduce recipient strain noted above. Full plasmid and primer lists can be found in Tables S4 and S3 respectively. BSGC represents the Bacillus genetic stock centre.

NRS14993610 ΔdegSU (spc)SPP1 NRS1183 → NCIB3610
NRS18593610 ΔsinR (kan)SPP1 NRS1858 → NCIB3610
NRS18603610 ΔsinR (kan)ΔdegSU (spc)SPP1 NRS1858 → NRS1499
NRS35413610 ΔtkmApNW395 → NCIB3610
NRS25443610 ΔptkApNW341 → NCIB3610
NRS35283610 ΔtkmAΔptkApNW387 → NCIB3610
NRS22223610 ΔptpZpNW330 → NCIB3610
NRS24713610 ΔugdpNW331 → NCIB3610
NRS32323610 tuaD::pNW377 (cml)SPP1 NRS3227→ 3610
NRS22273610 ΔytcA (kan)SPP1 NRS2059 → 3610
NRS32313610 Δugd + ΔytcA (kan) + tuaD::pNW377 (cml)SPP1 NRS3227→ NRS2821
NRS27953610 ptkA D83ApNW348 → NCIB3610
NRS27963610 ptkA D81A–D83ApNW348 → NCIB3610
NRS27993610 ptkA Y225A–Y227A–Y228ApNW349 → NCIB3610
NRS35443610 ΔtkmA + amyE::Pspank-hytkmA–lacI (spc)SPP1 NRS3255 → NRS3541
NRS35363610 ΔtkmAΔptkA + amyE::Pspank-hytkmA ptkA–lacI (spc)SPP1 NRS3265 → NRS3528
NRS35353610 ΔtkmAΔptkA + amyE::Pspank-hytkmA–lacI (spc)SPP1 NRS3255 → NRS3528
NRS35373610 ΔtkmAΔptkA + amyE::Pspank-hyptkA–lacI (spc)SPP1 NRS3255 → NRS2801
NRS28043610 ΔptkA + amyE::Pspank-hy–ptkA–lacI (spc)SPP1 NRS2801 → NRS2544
NRS28073610 ptkA D81A–D83A + amyE::Pspank-hy–ptkA–lacI (spc)SPP1 NRS2801 → NRS2796
NRS24673610 ΔptpZ + amyE::Pspank-hyptpZ–lacI (spc)SPP1 NRS2173 → NRS2222
NRS24683610 ΔptpZ + amyE::Pspank-hy–ptpZ (H196A)–lacI (spc)SPP1 NRS2193 → NRS2222
Figure 7.

Complex colony architecture of ΔtkmA and ΔtkmAΔptkA deletion strains. Strains (A) 3610, (B) ΔptkA (NRS2544), (C) ΔtkmA (NRS3541), (D) ΔtkmA + Pspank-hytkmA (NRS3544), (E) ΔtkmAΔptkA (NRS3528), (F) ΔtkmAΔptkA + Pspank-hytkmA–ptkA (NRS3536), (G) ΔtkmAΔptkA + Pspank-hytkmA (NRS3535) and (H) ΔtkmAΔptkA + Pspank-hyptkA (NRS3537) were grown on MSgg agar plates without the addition IPTG for 40 h at 30°C prior to photographing. Scale bar represents 5 mm.

Interestingly, deletion of tkmA did not phenocopy the ΔptkA strain (NRS2544) (compare Fig. 7B with C). Therefore, we hypothesized that TkmA retained activity in the absence of PtkA and that TkmA was capable of interacting with other proteins. To test this we constructed a strain lacking both tkmA and ptkA (NRS3528) (Table 1). We predicted that if our hypothesis was correct restoration of tkmA expression in the strain mutant for both tkmA and ptkA would return the colony phenotype to that of a ptkA mutant. When grown on MSgg media ΔtkmAΔptkA (NRS3528) displayed a complex colony architecture that was indistinguishable to the ΔtkmA mutant (NRS3541) (compare Fig. 7C with E). We confirmed that addition of both the tkmA and ptkA coding regions (NRS3536) under the control of an inducible promoter was necessary and sufficient to return the mutant phenotype back to that of wild-type B. subtilis (Fig. 7F). From this we conclude that both TkmA and PtkA are required for biofilm formation and that TkmA is epistatic to PtkA. Next, only the tkmA coding region was introduced into the ΔtkmAΔptkA mutant strain (NRS3535). The complex colony architecture of this strain was indistinguishable from the ΔptkA mutant (NRS2544) (compare Fig. 7G with B). As a change to colony architecture was observed upon expression of tkmA, this indicated that, as hypothesized above, TkmA can function (at least in part) in the absence of PtkA. In contrast, expression of only the ptkA coding region in the ΔtkmAΔptkA mutant strain (NRS3537) had no impact on complex colony architecture (Fig. 7H). This further supports the conclusion that substrate phosphorylation by PtkA during biofilm formation relies on activation of its kinase activity through interaction with TkmA. When taken together we conclude that the kinase activity of PtkA is required for its role during biofilm formation, whereas the ATPase activity of PtkA is not.

The terminal tyrosine residues of PtkA are not required for biofilm formation

The conserved C-terminal tyrosine domain serves as the site of autophosphorylation on PtkA (Fig. 6A); a process that is dependent on the catalytic activity of PtkA. In vitro assays indicate that the C-terminal tyrosine residues are not needed for phosphorylation of substrate proteins (Mijakovic et al., 2003) but structural analysis of a PtkA homologue demonstrates that the C-terminal tyrosine domain interacts with the neighbouring kinase active site (Olivares-Illana et al., 2008). Taken together, these findings render the role of the conserved tyrosine residues ambiguous. The influence that PtkA has on colony morphology presents a robust and simple route to clarify if the terminal tyrosine residues play a role in vivo during biofilm formation. To investigate the role of the terminal tyrosine cluster of PtkA during biofilm formation we constructed a strain in which the three terminal tyrosine residues were mutated to alanine (PtkA Y225A; Y227A; Y228A; NRS2799) (Fig. 6A).

In sharp contrast to mutating the ‘DxD’ motif of PtkA, mutation of the terminal tyrosine residues to alanine (NRS2799) did not alter the morphology the colony formed (Fig. 6F). These findings clearly demonstrate that the C-terminal tyrosine residues are not required in vivo by PtkA during biofilm formation. While the C-terminal tyrosine cluster is a conserved feature of BY-kinases' (Mijakovic et al., 2005a), the effects of phosphorylation of these residues are not conserved. For example, phosphorylation of the terminal tyrosine cluster of CpsD from Streptococcus pneumonia activates polysaccharide production (Bender et al., 2003; Morona et al., 2003); whereas in Acinetobacter lwoffii dephosphorylation is thought to activate the kinase Wzc (Nakar and Gutnick, 2003). It is unusual to conserve features of a protein, which serve no obvious purpose; therefore, further investigation into the role of the terminal tyrosine cluster is required as it remains to be identified if it is required for other processes controlled by PtkA, for example, protein localization (Jers et al., 2010).

Tyrosine dephosphorylation controlled by PtpZ influences early stages of biofilm formation

An approach similar to that used for PtkA was taken to assess the impact of deleting ptpZ on biofilm formation. As determined for the ΔptkA mutant, the ΔptpZ mutant was unable to develop the complex radial structures typical of the maturing B. subtilis biofilm at 24 h. Once more, in contrast to the reduction in colony architecture complexity (Fig. 8A), the pellicle formed by the ΔptpZ strain showed extensive three dimensional structural complexity but entirely lacked the ‘fruiting bodies’ that are present on the wild-type pellicle (compare Figs. 5E with 8B). These alterations in colony morphology and pellicle morphology were accompanied by a reduction in the level of sporulation from ∼25 ± 4% in the wild-type strain to ∼8.0 ± 3.6% in the ΔptpZ mutant (n = 3; P < 0.05 two-way Student's t-test). The decrease in sporulation was found to be specific to cells growing in the biofilm colony as after 24 h growth in liquid medium the wild-type strain and the ΔptpZ mutant had sporulated at a frequency of ∼33 ± 1.5% and ∼40 ± 9% respectively. However, in contrast to the ΔptkA mutant, the impact of deleting ΔptpZ became less apparent over time; such that by 48 h the lack of PtpZ had only a minor impact on the ability to form a biofilm (data not shown). The sporulation analysis confirmed this conclusion. After 72 h incubation ∼65 ± 3% of the ptpZ mutant population had sporulated, a value that was not significantly different from the wild-type strain (n = 3; P = 0.09 two-way Student's t-test). As dephosphorylation is less energetically demanding by comparison with the energy required to phosphorylate a tyrosine residue; perhaps this may explain the decrease in impact of the ΔptpZ mutation over time?

Figure 8.

Complex colony architecture of ΔptpZ point mutation strains.
A. Strains 3610, ΔptpZ (NRS2222) were assessed in the 3610 background as described in Fig. 1. Scale bar represents 5 mm.
B. Pellicles of ΔptpZ (NRS2222) were assessed in the 3610 background as described in Fig. 1. Scale bar represents 5 mm (left) and 1 mm (right).
C and D. Strains (C) ΔptpZ + Pspank-hy–ptpZ (NRS2467) and (D) ΔptpZ + Pspank-hy–ptpZ (H196A) (NRS2468) were grown on MSgg media with or without the addition of 100 µM IPTG.

To ensure that the ΔptpZ mutant phenotype was specific to the absence of PtpZ, a wild-type copy of ptpZ was introduced to the amyE locus in the ΔptpZ mutant under the control of a heterologous promoter; Pspank-hyptpZ–lacI (NRS2467). In the presence of 100 µM IPTG biofilm architecture was comparable to that of the wild-type strain (Fig. 8C). These findings confirm that the altered biofilm phenotype was specific to the disruption of ptpZ. The phosphatase activity of PtpZ has been shown to be reduced by ∼95% when histidine at position 196 is mutated to alanine (Mijakovic et al., 2005b). Therefore, to establish whether the phosphatase activity of PtpZ was required to control biofilm formation a mutant allele of ptpZ (PtpZ–H196A) was introduced into the amyE locus under the control of a heterologous promoter; PspankhyptpZ–H196A–lacI (NRS2468). In contrast to the wild-type allele of PtpZ, the mutant PtpZ–H196A protein was not capable of restoring wild-type colony morphology (Fig. 8D). These findings support the conclusion that the phosphatase activity of PtpZ is required.

It is interesting to note that deletion of both ptkA and ptpZ results in a loss of colony complexity and a reduction of sporulation because of the lack of formation of fruiting body like structures. The a priori is that kinase and phosphatase pairs act to counteract the effects of one another in order to finely tune the activity of their target protein and therefore, the mutant strains would have opposite phenotypes. Our results contradict this assumption. However, consistent with our findings a decrease in polysaccharide production and biofilm formation was observed in Burkolderia cepacia upon mutation of both the tyrosine autokinase BceF and its low molecular tyrosine phosphatase partner, BceD (Ferreira et al., 2007). We therefore favour the hypothesis that a cycle of phosphorylation and dephosphorylation facilitated by TkmA/PtkA and PtpZ, respectively, is required to form a mature B. subtilis biofilm.

Biofilm formation is independent of the activity of Ugd

We wanted to identify the downstream target(s) of PtkA that is phosphorylated during biofilm formation. In vitro analysis currently indicates that PtkA has 13 confirmed substrates (Mijakovic et al., 2003; 2006; Jers et al., 2010). Additionally, using phosphoproteome analysis a further eight proteins have been identified as tyrosine phosphorylated (Levine et al., 2006; Eymann et al., 2007; Macek et al., 2007) (Table S2). One obvious potential target of PtkA during biofilm formation was Ugd. Ugd is regulated by both PtkA and PtpZ (Mijakovic et al., 2003; 2005b) and is cotranscribed with ptkA and ptpZ (Fig. 2). Ugd is a uridine-5′-diphophoglucose (UDP-glucose) dehydrogenase (Mijakovic et al., 2003). When phosphorylated on tyrosine-70 by PtkA, Ugd is activated and can catalyse the oxidation of UDP-glucose to UDP-glucuronate (Petranovic et al., 2009). UDP-glucuronate is subsequently used as a precursor for the production of teichuronic acid, which is a component of the cell wall (Soldo et al., 1999). It has been postulated that Ugd may be required for biofilm formation (Mijakovic et al., 2005a; Hung et al., 2007). To test whether Ugd was required for biofilm formation we constructed an in frame deletion in ugd (NRS2471). The Δugd mutant exhibited a colony morphology that was indistinguishable from the parental strain biofilm (Fig. 9), indicating that Ugd is not needed. However, there is one additional proven and one putative, UDP-glucose dehydrogenase encoded on the B. subtilis chromosome by tuaD and ytcA respectively (Mijakovic et al., 2003; 2005a). Functional redundancy between these proteins is a possibility; therefore, a series of single, double and a triple deletion mutant strains were constructed and the ability to form a mature and robust colony was tested. None of the strains tested had an influence on colony architecture (Figs 9 and S3). Taken together these findings indicate that Ugd, YtcA and TuaD are not the downstream substrates of PtpZ and PtkA regulated during post-translational control of biofilm formation by B. subtilis.

Figure 9.

Complex colony architecture of Δugd, tuaD::pNW377 and ΔytcA, as single mutants and in combination. Strains Δugd (NRS2471), tuaD::pNW377 (NRS3232), ΔytcA (NRS2227) and Δugd + tuaD::pNW377 + ΔytcA (NRS3231) were assessed in the 3610 background as described in Fig. 1. Scale bar represents 5 mm.

Biofilm formation is independent of any of the known tyrosine phosphorylated proteins

Having eliminated Ugd, TuaD and YtcA as targets of PtkA during biofilm formation, we widened our search. Global phosphoproteome analyses have broadened our understanding of the processes that are regulated by tyrosine phosphorylation in bacteria (Levine et al., 2006; Eymann et al., 2007; Macek et al., 2007). We therefore chose to take a systematic mutagenesis approach to try and identify the target of PtkA during biofilm formation. After removing Ugd and TuaD (Fig. 9), 19 proteins have been identified as tyrosine phosphorylated and therefore potential targets of PtkA (Table S2). Seven genes are known to encode essential proteins (asd, aspS, yaaD, yurY, infA, ssbA and eno) and thus were excluded from our analysis. The gene ssbB was also excluded from the analysis as SsbB shares 63% identity and 81% similarity to SsbA, an essential protein. This left 11 proteins (AhpF, CitC, Ldh, OppA, RocA, YjoA, YvyG, YnfE, SdhA, YxxG and YorK) of which mutations could be constructed in nine of the corresponding genes (ahpF, citC, ldh, oppA, rocA, yjoA, yvyG, ynfE and yorK). Mutations of both sdhA and yxxG were found to result in a severe growth defect (data not shown). After construction and verification of the mutant strains the impact on colony morphology was tested. None was found to influence colony formation (Fig. S4). Therefore, despite extensive efforts to identify the substrate phosphorylated by PtkA it remained elusive. It is of course possible that one or more of the essential protein is (are) the target of PtkA; however, it is important to remember that B. subtilis strain NCIB3610 is the progenitor of the B. subtilis 168 strain that was used in the phosphoproteome analyses and they contain several single nucleotide polymorphisms (Srivatsan et al., 2008) (Levine et al., 2006; Eymann et al., 2007; Macek et al., 2007). Some of these genome level differences alter the ability of the two strains to form a biofilm (Branda et al., 2001; Kobayashi, 2008). The NCIB3610 strain also contains a plasmid of ∼85 kbp, which could conceivably encode proteins that are tyrosine phosphorylated and required for biofilm formation (Earl et al., 2007). Therefore, it will be necessary to conduct global proteome analyses using NCIB3610 to identify further possible targets of PtkA that are modulated during biofilm formation.

Concluding remarks

The principal finding of this work is that B. subtilis uses tyrosine phosphorylation as a mechanism of influencing biofilm formation. To achieve this it uses TkmA, a BY-kinase modulator; PtkA, a BY-kinase and PtpZ, a protein tyrosine phosphatase. In addition, we have shown for the first time that TkmA, the BY-kinase modulator is capable of interacting with an additional target(s) in the absence of PtkA. It will be interesting to investigate whether or not TkmA is able to interact with the predicted tyrosine kinase EpsB as has previously been speculated (Mijakovic et al., 2005a) or if it has an entirely unrelated function. EPS biosynthesis is essential for biofilm formation and in a wide range of bacterial species tyrosine phosphorylation is an important mechanism that regulates EPS biosynthesis. For example, in Escherichia coli, a tyrosine kinase (Wzc) and phosphatase (Wzb) pair regulate the biosynthesis and export of colonic acid (Lacour et al., 2008), a thick mucoid polymer that is required for biofilm structure and depth (Danese et al., 2000). Additionally, in Streptococcus thermophilus, a tyrosine kinase (EpsD) and phosphatase (EpsB) pair control the biosynthesis of EPS by regulating the phosphogalactosyltransferase activity of EpsE (Minic et al., 2007). It would seem reasonable to suggest that EpsB plays this more traditional role in EPS biosynthesis in B. subtilis given its location on the chromosome and the requirement for the epsA–O operon for biofilm formation (Branda et al., 2001; Mijakovic et al., 2005a). In contrast with these examples, PtkA and PtpZ are not located within an operon involved in the biosynthesis of an EPS that is required for biofilm formation as deletion of ugd did not impact biofilm formation (Fig. 9). Thus the mechanism by which PtkA and PtpZ influence biofilm formation using tyrosine phosphorylation is novel and warrants further investigation.

Experimental procedures

Growth conditions and strain construction

Escherichia coli and B. subtilis strains were routinely grown in LB medium (10 g NaCl, 5 g yeast extract and 10 g tryptone per litre) at 37°C unless otherwise stated. E. coli strain MC1061 [F′lacIQ lacZM15 Tn10 (tet)] was used for the routine construction and maintenance of plasmids. Where appropriate MSgg medium (5 mM potassium phosphate and 100 mM MOPS at pH 7.0 supplemented with 2 mM MgCl2, 700 µM CaCl2, 50 µM MnCl2, 50 µM FeCl3, 1 µM ZnCl2, 2 µM thiamine, 0.5% glycerol, 0.5% glutamate) (Branda et al., 2001) was used for analysis of biofilm formation. B. subtilis strains used in this study are listed in Tables 1 and S3. Strains were constructed using standard protocols (Harwood and Cutting, 1990). Phage transductions were conducted as described previously (Kearns and Losick, 2003). When required, antibiotics were used at the following concentrations: ampicillin 100 µg ml−1, chloramphenicol 5 µg ml−1, erythromycin 1 µg ml−1, lincomycin 25 µg ml−1, kanamycin 25 µg ml−1and spectinomycin 100 µg ml−1. When required, IPTG was added to the medium at the concentrations specified.

Surface adhesion assay

Cell-surface adhesion of strains constructed from the NCIB3610 parent strain was assessed by patching each strain on a 1.5% LB agar plate. Strains were grown for 16 h at 37°C prior to the cells being scraped from the surface of the agar plate using a disposable pipette tip ensuring equal pressure was applied to each patch.

Construction of plasmid pNW312

Plasmid pNW312 used to introduce the ptpZ gene under the control of the IPTG inducible promoter, Pspankhy at the non-essential amyE locus is a derivative of plasmid pDR111 (Britton et al., 2002). The ptpZ coding region, including ribosome binding site, was amplified from genomic DNA isolated for NCIB 3610 using primers NSW172 (5′-GGATGTCGACCATCGTGCCCGTTATTTATTT-3′) and NSW173 (5′-GGATGCATGCGAATCGGCTGTTAAAAAAAACC-3′). The PCR was cloned into the SalI and SphI sites of pDR111 using the restriction sites engineered into the primers (underlined in the primer sequence). The insert was sequenced to ensure that PCR-generated mistakes were not introduced into ptpZ. Additional plasmids were constructed in an identical manner for tkmA and ptkA, separately and in combination, using the primers and plasmids detailed in Tables S4 and S5.

Construction of plasmid pNW316

Plasmid pNW316 was used to introduce the ptpZ gene containing a point mutation encoding for a substitution of histidine at position 196 to an alanine, under the control of the IPTG inducible promoter, Pspankhy at the non-essential amyE locus and is a derivative of plasmid pNW312. The point mutation was introduced using primers NSW205 (5′-GTAGCCTCAGATGCCGCTAATGTGAAAACGAGA-3′) and NSW206 (5′-TCTCGTTTTCACATTAGCGGCATCTGAGGCTAC-3′). Following PCR of methylated template DNA the reaction was digested using DpnI (NEB). The remaining DNA was transformed into E. coli. Plasmids containing the insert were sequenced. Additional point mutations were inserted in an identical manner for ptkA using the primers and plasmids detailed in Tables S4 and S5.

Construction of in frame deletion strains

Plasmid pNW330 used for the construction of an in frame, markerless deletion of ptpZ was constructed as follows. Primers NSW225 (5′-GCATGGATCCGGAGTGGACATTTTGGCGCT-3′) and NSW218 (5′-GCATGTCGACATCGATCATGTCCTAGCCCCC-3′) were used to amplify the region located upstream from the coding sequence of ptpZ. The PCR product was cloned into pMAD (Arnaud et al., 2004) using the restriction sites engineered into the primers (underlined in the primer sequence) resulting in pNW328. Primers NSW219 (5′-GCATGTCGACGGTTTTTTTTAACAGCCGATTCTC-3′) and NSW220 (5′-GCATCCCGGGCACTGTGCTTTTCGTTACCAC-3′) were used to amplify the region located downstream from the coding sequence of ptpZ, which was cloned into pNW328 using the restriction sites engineered into the primers (underlined in the primer sequence) resulting in pNW330. To introduce the mutation into NCIB3610, pNW330 was transformed into 168. Phage was harvested and was used to transduce NCIB3610. NRS2222 was constructed by integration and curing of pNW330 in NCIB3610. Further plasmids were constructed for in frame mutations in an identical manner for tkmA, ptkA, ugd and for tkmA and ptkA simultaneously using the primers and plasmids detailed in Tables S4 and S5.

Construction of chromosomal directed substitutions

Plasmid pNW348 was used for the construction of an in frame, markerless point mutation of ptkA at the ‘DxD’ motif. Primers NSW246 (5′-GGGCGATATCGGATCCGCATACTTTTGATTATGATTGTAACCGCG-3′) and NSW249 (5′-CCGGGAGCTCGAATTCTTTCTCTAGAACATAAAGAGCCTCTTGA-3′) were used to amplify ptkA, which was cloned into pUC19 using the restriction sites engineered into the primers (underlined in the primer sequence). Primers NSW250 (5′-AAAGTGCTCCTGATTGCTGCTGCTTTGCGAAAACCAACA-3′) and NSW251 (5′-TGTTGGTTTTCGCAAAGCAGCAGCAATCAGGAGCACTTT-3′) were used to introduce point mutations as previously described. The resulting product was sequenced to ensure that the desired mutations were introduced. The region of DNA carrying ptkA (D81A–D83A) was cloned into pMAD using the restriction sites engineered into the primers. The point mutations were introduced into NCIB3610 as described above for the in frame deletions. Further plasmids were constructed for in frame mutations in an identical manner for ptkA (Y225A–Y227A–Y228A) using the primers and plasmids detailed in Tables S4 and S5.

Biofilm analysis

Analysis of biofilm formation was performed as described (Branda et al., 2001; Verhamme et al., 2007). B. subtilis strains were inoculated from a fresh LB plate and grown to mid-late exponential phase in LB. The undiluted culture (10 µl) was spotted onto an MSgg plate containing 1.5% agar (containing IPTG as required) and incubated at 37°C for the time period indicated. For pellicle formation the culture was diluted 1000-fold into 1.5 ml MSgg media in a 24-well plate and was incubated for 16 h at 37°C. Images of the pellicles and bacterial colonies were captured using a Leica MZ16 FA stereoscope using LAS software version 2.7.1.

Sporulation analysis

Sporulation analysis was conducted as described previously (Vlamakis et al., 2008). Colonies were grown on solid MSgg media for 24 or 72 h at 37°C prior to cell collection. Liquid cultures were prepared by growing cells in 10 ml MSgg media inoculated with cells from an OD600 of 0.01. Cultures were grown for 24 or 72 h with shaking at 37°C.

Single cell image analysis

Strains for analysis were grown in LB at 37°C with shaking to an OD600 of 0.3. Cells were collected by centrifugation and were concentrated fivefold by resuspension in wash buffer [15 mM (NH4)2SO4, 80 mM K2HPO4, 44 mM KH2PO4, 3.4 mM sodium citrate, 1 mM Mg2SO4]. The concentrated cell suspensions were stained using Hoechst 33342 (10 ng ml−1) (Invitrogen™) for 5 min and were washed twice in wash buffer. Stained cell suspension (2 µl) was spotted onto thin agarose slides and viewed using an Axio Imager M1 microscope (Zeiss). Images were analysed using AxioVision Rel. version 4.6 image capture software (Zeiss).

RNA isolation

Cells for RNA isolation were inoculated into 25 ml of LB at an OD600 of 0.01 from an overnight lawn plate and grown at 37°C with shaking at 200 r.p.m. in 250 ml flasks. When an OD600 of 1 was reached, 3 ml of cells were harvested by centrifugation at 13 000 g for 1 min and immediately processed for RNA extraction using the RiboPure™-Bacteria RNA extraction kit (Ambion®) according to the manufacturer's instructions. RNA was checked for quality by running on an agarose gel and quantified using a Nanodrop™.

DNA microarray analysis

Bacillus subtilis microarrays consisting of 4105 gene-specific oligonucleotides (Compugen) were printed at the Research Technology Support Facility at Michigan State University. Five oligonucleotides that have no significant similarity to the B. subtilis genome were used as negative controls. To identify the genes that were upregulated in the absence of both SinR and DegU, RNA was extracted from a degSU (NRS1499) and a degSUsinR (NRS1686) mutant strain. For the DNA array analysis two independent RNA samples were used for each strain and two technical DNA array replicates were conducted. cDNA was synthesized using either 3 or 6 µg of total RNA by reverse transcribed with Superscript III (Invitrogen) for 2 h at 50°C in the presence of amino-allyl-dUTP (Sigma). RNA was degraded by hydrolysis using 15 µl of 0.1 M NaOH and incubation at 70°C for 10 min. The sample was neutralized by the addition of 15 µl of 0.1 M HCl and the cDNA purified using a MinElute PCR purification kit (Qiagen) and eluted in 10 µl H2O. To fluorescently label the cDNA, 0.5 µl of fresh 1 M NaHCO3 was added to the cDNA and was used to hydrate an aliquot of either GE Healthcare Fluorolink Cy3 dye (degSU cDNA) or GE Healthcare Fluorolink Cy5 dye (degSUsinR cDNA). After 1 h incubation at room temperature unincorporated dye was removed using the MinElute PCR purification kit. Both labelled cDNA populations were applied onto a microarray and hybridized overnight at 42°C for 16–18 h. After washing, hybridized microarray slides were analysed using a GenePix™ 4100A array scanner (Axon Instruments). Images were processed using the GenePix Pro 5.0 software, which generates red-green fluorescent intensity values for each spot. The fluorescent signal intensities were imported into Microsoft Excel (available from The microarray datasets were filtered to remove those genes that were expressed at levels less than two standard deviations above the average background values in both channels. The intensity values were normalized relative to the ratio generated by dividing the total value of Cy3 fluorescence with that of the total value of Cy5 fluorescence measure per array. The average value from the four spots per gene was calculated and those genes with an expression level of greater than 2.0 after normalization were selected. At this point the genes that had previously been identified as directly repressed by SinR were eliminated (Chu et al., 2006), for example, genes in the epsA–O operon and the yqxM–sipW–tasA operon (Table S1). This resulted in a group of 11 genes that had an average expression value of > 2.0 and that had previously not been demonstrated to be SinR repressed.

Semi-quantitative RT-PCR

RNA was isolated from cultures of 3610, degSU, sinR and degSUsinR as described above. The harvested RNA was DNase treated and cDNA was synthesized as previously published (Stanley and Lazazzera, 2005). To ensure the RNA samples were free from contaminant DNA, samples of RNA lacking Superscript III (Invitrogen) were treated in parallel with samples intended for cDNA synthesis. Synthesized cDNA (1 µl) was used in a standard 20 µl PCR reaction using Qiagen Taq DNA polymerase with the addition of 3.5 mM MgCl2 and 8.3 nl ml−1 SyberGreen (Sigma). PCR samples were held at 95°C for 10 min prior to a 25-cycle PCR reaction involving a melting step at 95°C, followed by an annealing step at 50°C and then an elongation step at 72°C. Each PCR step was held for the duration of 25 s. The genes ptpZ, yomI, ywbC and yvfO were amplified using gene-specific primers detailed in Table S4. To determine the melting temperatures of the PCR products the set point temperature was increased in 40 cycles (10 s each) by 1°C per cycle, starting from 50°C. Expression of ptpZ, yomI, ywbC and yvfO, was calculated as fold changes relative to the wild-type strain using the formula: Fold change = 2−ΔΔCt; with −ΔΔCt = [Ct(gene ×) − Ct(constitutive gene)]condition I − [Ct(gene x) − Ct(constitutive gene)]condition II (Talaat et al., 2002). The level of transcription of each gene was measured using Corbett qPCR reader, run using Rotorgene 8 software (Corbett) with expression of veg monitored as a reference gene (Hamon et al., 2004).

tkmA operon analysis

RNA was isolated from NCIB3610 grown in LB to an OD600 of 0.01, as described above. cDNA was synthesized using the reverse ugd-specific primer NSW220 and the resulting mixture was treated with RNase H (Invitrogen) for 20 min at 37°C. The following primer pairs were used for amplification of an internal region of ptkA (NSW225 and NSW254), ptpZ (NSW195 and NSW249) or ugd (NSW165 and NSW220). See Table S4 for primer sequences.


The transcription start site of tkmA was determined using 5′RACE conducted using DNase treated RNA harvested from NCIB3610 as described previously. To synthesize cDNA corresponding to the promoter region of the tkmA gene, 5 µg of DNase treated RNA was incubated with 2.5 pmol NSW227 (5′-GGGCTCTTAATGATGACA-3′) at 70°C for 10 min. The reaction mixture was made up to a total volume of 25 µl with the addition of 1× first strand buffer, 11.2 µM DTT and 0.4 mM dNTPs. The mixture was incubated at 42°C for 1 min prior to the addition of 200 U of Superscript II (Invitrogen™) and then incubated for a further 50 min. The reverse transcriptase was heat inactivated by incubation at 70°C for 15 min. To remove any remaining RNA template, RNase H. 500 ng of synthesized cDNA was 5′ dC-tailed using 20 units of terminal transferase (Tdt) (New England Biolabs™), supplemented with 1× Terminal Transferase Reaction Buffer (20 mM Tris-acetate, 50 mM potassium acetate and 10 mM Magnesium Acetate), 0.25 mM CoCl2 and 200 µM dCTP. The reaction was incubated at 37°C for 10 min and then 10 min at 65°C for to heat inactivate the Tdt. Touchdown PCR was used to amplify the tkmA promoter region using primers NSW367 (5′-GGCCACGCGTCGACTAGTACGGGUUGGGUUGGGUUG-3′) and NSW228 (5′-GGGTGTAAGTGCGAAGAAA-3′). The standard PCR reaction mixture included 1.5 mM MgCl2, 5 µl of dC-tailed cDNA, 0.4 mM each primer, 1 HotStart Taq bead and 1× PCR buffer (Promega). Annealing temperature was gradually reduced from 58°C to 48°C in 0.5 increments, with a final 20 cycles at 48°C. The resulting PCR product was visualized by gel electrophoresis and staining prior to recombination into the TOPO-TA vector pCR2.1 (Invitrogen™). The inserts cloned were sequenced and the transcription start site identified.

Purification of DegU, DegS and SinR

DegU-His6 was purified as described previously (Verhamme et al., 2007) with the following alteration. Following IMAC purification DegU-His6 was purified from contaminating DNA by application to a DEAE-FF column (GE Healthcare™). Purified DegU-His6 was dialysed before use [20 mM Tris-HCl (pH 8.0) 50 mM NaCl]. SinR-His6 was purified as previously described (Verhamme et al., 2009). The histidine tag used for purification was removed using Novagen's thrombin cleavage capture kit. For the overproduction of DegS-His6, plasmid pET28b–degS-His6 (Kobayashi, 2007), was induced by the addition of 500 µM IPTG at OD600 0.4. Growth was continued for 3 h prior to collection by centrifugation at 4°C. Cells were washed with ice cold 10 mM Tris-HCl (pH 7.6). DegS-His6 was purified from inclusion bodies according to the BugBuster protocol (Novagen). DegS-His6 was incubated on ice for 2 h in denaturing buffer [50 mM Tris-HCl (pH 7.6), 150 mM NaCl, 8 M urea and 10 mM imidazole] before the protein solution was applied to a HisTrap FF column (GE healthcare). DegS-His6 was slowly renatured by gradual removal of the urea over the course of 90 min at room temperature. DegS-His6 was eluted from the column by the addition of 500 mM imidazole and was dialysed in 50 mM Tris-HCl (pH 7.6), 200 mM KCl, 10 mM MgCl2, 0.1 mM EDTA, 1 mM dithiothreitol (DTT) and 50% glycerol.

Electrophoretic mobility shift assay

A PCR product corresponding to the promoter region of tkmA (PtkmA) was amplified using primers NSW146 and NSW147 and purified by gel extraction. The promoter DNA was labelled using 50 µCi [γ-32P]-ATP (Perkin-Elmer) and T4 polynucleotide kinase (New England Biolabs). Unincorporated ATP was removed from the labelled DNA using an Illustra Microspin G-25 column (GE healthcare). Phosphorylated purified DegU was produced essentially as described previously (Gueriri et al., 2008), with the exception that a final concentration of 10 µM purified DegU and 0.78 µM purified DegS was added to the phosphorylation reaction. The reaction was incubated for 30 min at 25°C. EMSA binding mixtures were prepared containing 15 mM Tris-HCl (pH 7.6), 0.2% Tween-20, 1 mM MgCl2, 60 mM NaCl, 4% glycerol, 15 mM DTT, 0.5 mg ml−1 bovine serum albumin, 1 µg poly (dI-dC) and a range of concentrations of DegU∼P, made up to a total volume of 30 µl. The reactions were initiated by the addition of 2 ng of 32P-labelled DNA probe. The reaction mixtures (15 µl) were loaded onto a 5% polyacrylamide (AccuFLOWGel; Acrylamide : Bis-Acrylamide solution 29:1) Tris Glycine gel (50 mM Tris-HCl, 400 mM glycine, 1.75 mM EDTA), which was run at 100 V for ∼1–2 h. Dried gels were exposed to X-ray film overnight at −70°C prior to development.


This work was funded by a BBSRC David Phillips Fellowship grant [BB/C520404/1]. T.B.K. is a recipient of a BBSRC doctoral training grant [BB/D526161/1]. We would like to thank the following people: the Sequencing Service (College of Life Sciences, University of Dundee, Scotland, for DNA sequencing, Professor Lazazzera, Professor Burkholder, Professor Britton and Professor Kroos for access to the B. subtilis oligonucleotides used to synthesis the DNA microarray slides, and Professor Norman Pratt and Dr Katie Robertson (Ninewells Hospital, University of Dundee) for use of the GenePix™ 4100A array scanner. We thank Dr Kazou Kobayashi for kindly providing plasmid pETdegS.