A Combination of Glycerol and Manganese Promotes Biofilm Formation in Bacillus subtilis via Histidine Kinase KinD Signaling

Abstract

The spore-forming bacterium Bacillus subtilis forms matrix-enclosed biofilms in response to environmental cues that to date remain poorly defined. Biofilm formation depends on the synthesis of an extracellular matrix, which is indirectly regulated by the transcriptional regulator Spo0A. The activity of Spo0A depends on its phosphorylation state. The level of phosphorylated Spo0A (Spo0A∼P) is controlled by a network of kinases and phosphatases, which respond to environmental and physiological signals. In spite of significant progress in understanding biofilm development, the fundamental question of how cells sense the environmental cues that trigger biofilm formation has largely remained unaddressed. Here, we report that biofilm formation of B. subtilis in LB medium is triggered by a combination of glycerol and manganese (GM). Moreover, LB medium with GM significantly stimulates biofilm-associated sporulation and production of an undefined brown pigment. We further show that transcription of the major operons responsible for matrix production and biofilm formation is dramatically enhanced in response to GM. We also establish that KinD is a principal histidine kinase responsible for sensing the presence of GM exclusively by its extracellular CACHE domain. Finally, we show that GM has a similar biofilm-promoting effect in two related Bacillus species, B. licheniformis and B. cereus, indicating that the biofilm-promoting effect of GM is conserved in Bacillus species.

INTRODUCTION

The vast majority of bacteria often grow as elaborate multicellular communities, referred to as biofilms (1, 2). Biofilm formation is an ancient prokaryotic adaptation which allows bacteria to survive in hostile environments (2, 3). The spore-forming bacterium Bacillus subtilis can form architecturally complex colonies on solid medium and pellicles at the interface of liquid medium and air (4). In spite of significant progress in understanding biofilm development in B. subtilis, the fundamental question concerning the specific signals that trigger biofilm formation has largely remained unaddressed. Biofilm formation depends on the synthesis of an extracellular matrix that holds the constituent cells together. The matrix has two main components, an exopolysaccharide synthesized by the products of the epsA-epsO operon and amyloid fibers encoded by tasA, located in the tapA (formerly yqxM) operon (5–7). Both operons are under the negative control of two repressors, SinR and AbrB, which act synergistically to repress the matrix genes (7, 8). Loss of either one results in formation of an extremely robust biofilm (7, 8). Derepression is triggered in part by the action of SinI (7), whose expression is in turn activated by Spo0A (8, 9). Spo0A also directly represses the gene for AbrB, a transition-state regulator, in post-exponential-phase cells (10).

The activity of Spo0A depends on its phosphorylation state (11). Low and intermediate levels of phosphorylated Spo0A (Spo0A∼P) lead to induction of the epsA-epsO and tapA operons, which results in production of the extracellular matrix and, thus, biofilm formation (12). At high levels of Spo0A∼P, the matrix genes are repressed (12). Simultaneously, sporulation genes are induced, and as a result, the matrix-producing cells transform to become spores. The level of Spo0A∼P is controlled by a network of histidine kinases that directly or indirectly phosphorylate Spo0A (11). Five distinct sensor kinases (KinA, KinB, KinC, KinD, and KinE) have the capability of transferring a phosphoryl group into the phosphorelay to control the level of Spo0A∼P present at any moment in the cell (13–17). The current thinking in the field is that these kinases respond to various environmental and physiological cues, but the nature of these cues and how the kinases respond to them are not known in most cases.

It was recently proposed that the kinase KinD contains an extracellular domain, the so-called CACHE domain (18), for sensing small chemical molecules released from the plant host during colonization (19). In another recent study (20), the authors suggested that the transmembrane domain of KinD is involved in osmosensing. On the basis of those recent studies, it seems that KinD is likely bifunctional (or multifunctional) in signal sensing. Here we show that a combination of glycerol and manganese (GM) strongly promotes biofilm formation as well as biofilm-associated sporulation and pigment production in B. subtilis grown in lysogeny broth (LB) medium. We further report that the histidine kinase KinD is a major kinase responsible for sensing the biofilm-promoting effect of GM. We also provide evidence indicating that the biofilm-promoting effect of GM is conserved among the Bacillus species.

MATERIALS AND METHODS

Strains and growth media.

The strains used and generated in this study are listed in Table S2 in the supplemental material and are isogenic unless otherwise indicated. For routine growth, all strains were propagated in LB (10 g of tryptone, 5 g of yeast extract, and 5 g of NaCl per liter) or on solid LB medium supplemented with 1.5% agar. The B. subtilis wild strain NCIB3610 and its derivatives were regularly cultured in LB medium. Biofilms were generated at 30°C in the novel biofilm-promoting medium LBGM (LB plus 1% [vol/vol] glycerol and 0.1 mM MnSO₄) and were assayed by a method similar to that described previously (21). When appropriate, antibiotics were added at the following concentrations for growth of B. subtilis: 10 μg/ml of tetracycline, 100 μg/ml of spectinomycin, 10 μg/ml of kanamycin, 5 μg/ml of chloramphenicol, and 1 μg/ml of erythromycin. For growth of Escherichia coli, the concentrations of added antibiotics were as indicated previously (21).

Strain construction.

All insertion-deletion mutants were generated by using long-flanking-homology PCR (22). The generated constructs were inserted by double-crossover recombination into neutral integration sites (amyE and lacA) in the genome of B. subtilis by inducing natural competence (4). The KinD-DegS hybrid kinase was created with overlapping PCR using the primers listed in Table S3 in the supplemental material. In brief, the first DNA fragment covering the promoter region and the N-terminal sequence of kinD was amplified by PCR using primers kinD-degS-F1 and kinD-degS-MR1. A second DNA fragment covering the C-terminal sequence of degS was amplified by PCR using primers kinD-degS-MF1 and kinD-degS-R1. These two PCR fragments were joined together after another round of overlapping PCR amplification. The final PCR products were cloned into BamHI and EcoRI restriction sites of the vector pDG1662 (23), which is a vector for amyE locus integration and carries the chloramphenicol resistance marker. Expression of the chimeric kinase is therefore under the control of the kinD native promoter in the recombinant plasmid. The generated vector was transformed into E. coli DH5α cells. The insert was verified by PCR as well as by DNA sequencing. The purified plasmid DNA was transformed into naturally competent cells of PY79. The constructs were then transferred to NCIB3610 by SPP1 phage transduction as described previously (6, 21).

The insertion-deletion mutant of ΔglpK (FC96) was constructed using long-flanking PCR mutagenesis by Frances Chu and provided to us as a gift. Construction of the insertion-deletion mutant of ΔglpF (YC877) was performed similarly by applying long-flanking PCR mutagenesis (22). The four primers used for generating the ΔglpF deletion mutation were delta/glpF-P1, delta/glpF-P2, delta/glpF-P3, and delta/glpF-P4, listed in Table S3 in the supplemental material. KinC and KinD overexpression strains (strains MF1889 and MF2147) were originally constructed by Masaya Fujita in the PY79 strain background (24). The overexpression constructs were then introduced into the NCIB3610 strain background by SPP1 phage-mediated transduction, generating strains YC1014 and YC1015, respectively.

Assays of biofilm formation.

B. subtilis cells were grown in LB broth at 37°C to mid-log phase. For colony formation, 2 μl of the cells was spotted onto LBGM medium (or some other LB-based medium) solidified with 1.5% agar. Plates were incubated at 30°C for 72 h prior to analysis. For pellicle formation, 9 μl of the cells was mixed with 9 ml of LBGM broth (or some other LB-based broth) in 6-well plates (VWR). Plates were incubated at 30°C for 48 h. Images were taken using either a Nikon CoolPix 950 digital camera or a SPOT camera (Diagnostic Instruments).

Assays of β-galactosidase activities.

The entire colonies formed on solid medium were collected and resuspended in phosphate-buffered saline (PBS) buffer. Typical long bundled chains of cells in the biofilm colony were disrupted using mild sonication, as described previously (5). The optical density (OD) of the cell samples was normalized using the OD at 600 nm (OD₆₀₀). One milliliter of cell suspensions was collected and assayed as described previously (21).

Sporulation assay.

Biofilm colonies were formed on LB and LBGM solid media at 30°C as described above. The entire colonies were collected and suspended in PBS buffer. Chained and bundled cells in the biofilm colony were disrupted by mild sonication. Cells were serially diluted. Heat kill was performed at 80°C for 20 min in a water bath. Total cell numbers before and after heat kill were quantified by the plating method. Sporulation efficiency was calculated by dividing the total number of viable spores after heat kill by the total number of cells before heat kill.

Growth curve experiment.

Wild-type (NCIB3610), ΔglpK mutant (FC96), and ΔglpF mutant (YC877) cells were first grown in LB medium at 23°C overnight. On the next morning, cells were diluted 1:100 into M9 minimal medium (25) with either 0.5% glucose or 05% glycerol as the sole carbon source. Cells were grown under shaking culture conditions (150 rpm) at 37°C, and the cell optical density of the samples was measured periodically for a total of 12 h. Each condition had 3 replicates, and the growth curve experiment was repeated twice. Representative results are shown.

RNA extraction and real-time RT-PCR.

The details of the real-time reverse transcription-PCR (RT-PCR) experiments can be found in the Materials and Methods in the supplemental material.

RESULTS

A combination of glycerol and manganese promotes multicellular development by B. subtilis.

The starting point of this study was the finding that a combination of glycerol and manganese (GM) strongly promotes biofilm formation by B. subtilis strain NCIB3610 (4) in lysogeny broth (LB) medium (here referred to as LBGM medium) (Fig. 1A). This observation was quite surprising because LB medium is considered not optimal for biofilm formation for B. subtilis NCIB3610 (Fig. 1A) (21). Both components, glycerol (1%, vol/vol) and manganese (0.1 mM), were required for the robust biofilm phenotype, as the presence of glycerol alone (1%, vol/vol; LBG) or manganese alone (0.1 mM; LBM) stimulated biofilm formation little or relatively less profoundly (Fig. 1A). Noteworthy, it was previously shown that addition of glycerol or manganese alone could stimulate biofilm formation by a B. subtilis B1 strain, albeit at much higher concentrations: up to 80 g per liter for glycerol (8%, vol/vol) and up to 1,000 mM for manganese (26). Apparently, in our study, there was a synergistic activity between glycerol and manganese in stimulating the biofilm phenotype when added at much lower concentrations. We also attempted to replace glycerol in this combination with a variety of different carbon sources at similar concentrations, including sugars, such as glucose, galactose, mannose, and sucrose, and organic acids, such as succinic acid, glutamic acid, and lactic acid. None of the above-mentioned carbon sources tested, when combined with manganese, had a strong biofilm-promoting activity similar to that of glycerol (data not shown). This indicates that glycerol is unique in stimulating biofilm formation. We further wondered if simple carbohydrates such as glucose, which represses utilization of glycerol via catabolite repression (27), could impair glycerol-dependent biofilm stimulation. We therefore tested the effect of addition of glucose to the LBGM medium. Surprisingly, our results show that addition of glucose did not impair biofilm formation in LBGM medium (see Fig. S1 in the supplemental material).

Fig 1.

Effect of glycerol and manganese on B. subtilis NCIB3610 multicellular development. (A) A combination of glycerol (1%, vol/vol) and manganese (0.1 mM) added to LB medium (LBGM medium) promotes robust biofilm formation. Effects of addition of glycerol (LBG medium) or manganese (LBM medium) alone are also shown. An undomesticated wild strain of B. subtilis, NCIB3610, was used here. (B) Transcription of the operons responsible for the matrix production is upregulated in response to GM. (Left) Results from RL4582 cells that bear the P_tapA-lacZ transcriptional fusion; (right) results from RL4548 cells that bear the P_eps-lacZ transcriptional fusion. (C) Addition of GM to LB medium also greatly stimulates biofilm-associated sporulation, as the total spore accounts increased about 100-fold in a biofilm colony formed on an LBGM agar plate compared with the counts on an LB agar plate.

Biofilm formation depends on the synthesis of extracellular matrix, whose production is specified by two major operons: the epsA-epsO and tapA operons (5–7). The epsA-epsO operon is responsible for the production of the exopolysaccharide, whereas the tapA operon is responsible for the production and assembly of an amyloid-like fiber (21, 28). We hypothesized that the dramatic increase in biofilm formation in the presence of GM could be due to upregulation of those genes involved in matrix synthesis. To test this hypothesis, we analyzed the effect on matrix gene expression by addition of GM using transcriptional fusions of the promoters for epsA-epsO and tapA to the lacZ gene encoding β-galactosidase. The expression of the two transcriptional fusions was enhanced drastically in response to the addition of GM (Fig. 1B), increasing about 20- and 9-fold for tapA (left) and epsA-epsO (right). This result suggests that the strong biofilm-stimulating activity in response to the addition of GM was indeed due to upregulation of the matrix genes. To differentiate the contribution of glycerol alone and manganese alone in upregulation of the matrix genes, we similarly measured transcription of epsA-epsO (by using the transcriptional fusion described above) in cells from the biofilms formed on LB, LBG, and LBM media. Interestingly, the addition of glycerol alone to LB medium had a quite significant effect on epsA-epsO transcription, whereas the addition of manganese alone had a less profound effect (see Fig. S2A in the supplemental material). This result was rather unexpected since the biofilm phenotype in LBM medium seemed to be more significant than that in LBG medium (Fig. 1A).

The presence of GM also significantly stimulated biofilm-associated sporulation; there was a more than 100-fold increase in the sporulation efficiency in a biofilm colony grown on LBGM medium compared to that in a biofilm colony grown in LB medium (Fig. 1C). In addition, the presence of GM also promoted production of a brown pigment and secretion of the pigment around the biofilm colony (see Fig. S3A in the supplemental material). Interestingly, the secreted pigment seemed to be biofilm specific since a biofilm-defective strain of B. subtilis (ΔepsH) which bears a null mutation in one of the eps genes whose products are involved in making exopolysaccharides and is thus unable to form biofilms (7) did not show significant pigment secretion (see Fig. S3B in the supplemental material). Furthermore, the pigment production was not associated with sporulation, as the ΔsigF mutant, which is blocked in sporulation due to the mutation in the gene encoding the key sporulation sigma factor SigF (29), was still able to secret the brown pigment (see Fig. S3B in the supplemental material). It will be interesting to characterize the chemical composition and understand the function of the pigment in future studies.

The biofilm-promoting effect of GM is mediated mainly by the histidine kinase KinD.

We next wished to elucidate the signaling pathway that enables GM to trigger such robust biofilm formation. Our assumption was that Spo0A must be involved in this process, since it is known to have a pivotal role in regulating biofilm formation (9). The five sensory histidine kinases (from KinA to KinE) that are capable of donating the phosphoryl group to Spo0A are the putative candidates for sensing the presence of GM (14, 16). We wondered whether any one of these sensor kinases was involved in sensing the presence of GM and promoting biofilm formation. To answer this question we screened mutants for each of the five sensor kinases for their ability to respond to the addition of GM. As shown in Fig. 2, the ΔkinD mutant (and, less significantly, the ΔkinC mutant) had the most defective phenotype, while all other mutants were able to respond to the addition of GM by forming robust biofilm. Moreover, the elevated expression of matrix genes in response to GM discussed above was found to depend on KinD (see Fig. S2B in the supplemental material). These results indicate that KinD might be principally responsible for sensing the presence of GM. Interestingly, the biofilm phenotype of the ΔkinD mutant was almost as defective as that of the mutants deficient in phosphorelay, namely, Δspo0F, Δspo0B, and Δspo0A (Fig. 2), which further indicates that the presence of GM was sensed by KinD and then transduced to Spo0A via the phosphorelay. It was previously shown that a mutation in the abrB gene, whose protein product is a transition state regulator and controls both epsA-epsO and tapA, suppresses the biofilm-defective phenotype caused by Δspo0A (30). In agreement with these data, we found that a deletion in the abrB gene also suppressed the ΔkinD mutation, once again suggesting that KinD is located upstream of Spo0A and AbrB in the pathway (Fig. 2).

Fig 2.

Histidine kinase KinD is a major kinase in sensing the presence of GM. Colony development and pellicle formation on LBGM by the wild type (WT) and various mutant strains were compared. The strains used here were as follows: wild type (NCIB3610) and ΔkinA (RL4566), ΔkinB (RL4563), ΔkinC (RL4565), ΔkinD (RL4569), ΔkinE (RL4570), ΔkinAB (RL4573), ΔkinCD (RL4577), Δspo0F (RL4567), Δspo0B (RL4568), Δspo0A (RL4620), ΔkinD ΔabrB (YC863), and ΔabrB (YC668) mutants. The ΔkinD mutant showed the most defective phenotypes in biofilm formation, fairly similar to those of the mutants deficient in phosphorelay: Δspo0F, Δspo0B, and Δspo0A.

The biofilm-forming phenotype for many of the mutants mentioned above was previously examined in another biofilm-inducing medium, MSgg (17, 31). In comparison to LBGM, MSgg is a minimal medium in which the mutants deficient in phosphorelay (Δspo0F, Δspo0B, and Δspo0A) were also found to be severely defective in biofilm formation (9, 31). On the contrary however, the mutants for each of the individual kinases, especially KinC and KinD, showed little or a very mild difference in phenotype from that of the wild-type strain in MSgg (see Fig. S4 in the supplemental material and reference 17). Therefore, KinD (and, to a lesser degree, KinC) does seem to be more important in promoting robust biofilm formation in LBGM medium than in MSgg. Since MSgg also contains glycerol (0.5%, vol/vol) and manganese (0.1 mM), one possible explanation for the difference in the two media is that in MSgg, the roles of the sensory kinases (especially KinC and KinD) in stimulating biofilm formation are redundant, whereas in LBGM medium, their roles in that are much more complementary. As further evidence, the ΔkinC ΔkinD double mutant is severely defective in both LBGM medium and MSgg (Fig. 2) (17).

If KinD and/or KinC is indeed that important in stimulating biofilm formation in LBGM medium, overexpression of KinD or KinC may lead to increased levels of biofilm production. Moreover, the presence of GM may further enhance the robustness of the biofilms. To test this, we constructed two strains in which the native kinD (or kinC) gene was under the control of the hyperspank promoter (see Materials and Methods). These two strains overproduce either KinD (YC1078) or KinC (YC1077) upon addition of the inducer IPTG (isopropyl-β-d-thiogalactopyranoside). As shown in Fig. 3A, overexpression of KinD alone (in the presence of IPTG but in the absence of GM) in LB medium resulted in slightly elevated levels of biofilm formation; in comparison, very thin pellicles were formed in the absence of IPTG. In LBGM medium, overexpression of KinD with IPTG resulted in the formation of thick and structured floating pellicles (Fig. 3A, +IPTG), unlike the pellicles of the same strain not treated with the inducer (Fig. 3A, −IPTG). The results from KinC overexpression (Fig. 3B) were unexpected because overexpression of KinC in the presence of GM had a strong negative impact on growth, especially under static culture conditions, and no pellicle formation was observed (Fig. 3B). Our results further strengthened the hypothesis that the sensory kinase KinD mediates the biofilm-stimulating effect of GM and that both the protein levels of KinD and the presence of GM are important for biofilm stimulation.

Fig 3.

Overexpression of KinD stimulates biofilm formation. (A) Overexpression of KinD in strain YC1078 stimulated floating pellicle formation in both the absence and presence of GM. (B) The effect of KinC overexpression on biofilm formation was similarly assessed in strain YC1077; to overexpress KinC or KinD, IPTG was added to the medium at a final concentration of 100 μM. Cells were incubated at 30°C. Images were taken after 48 h of incubation. Note that overexpression of KinC in the presence of GM had a strong negative impact on cell growth.

To further support the hypothesis presented above that KinD mediates the biofilm-inducing effect of GM, we compared the transcription of epsA-epsO (by using the same transcriptional fusion described above) in the ΔkinD mutant in LB, LBG, and LBM media. As seen in Fig. S2B in the supplemental material, there was no significant change in epsA-epsO transcription when glycerol or manganese alone was added to LB medium, indicating that KinD mediates the biofilm-stimulating effect of GM.

The presence of GM is sensed by the extracellular sensor domain of KinD.

We further asked, what is the unique characteristic of KinD distinguishing it from other sensor kinases? According to bioinformatics analysis (see the Materials and Methods in the supplemental material), KinD bears an extracellular CACHE domain (19). CACHE is present in many bacterial sensory histidine kinases and is capable of sensing small molecules, often in the presence of cofactors, such as metal ions (see Table S1 in the supplemental material) (18). This finding is consistent with the hypothesis that KinD can directly sense glycerol or its derivative through the CACHE domain, while the Mn²⁺ could be a possible cofactor of the interaction.

A KinD CACHE mutant was previously constructed in order to test the role of this domain in KinD in sensing plant-released chemical signals during colonization and root-associated biofilm formation of B. subtilis on the plant host (19). The CACHE mutant of KinD contains amino acid substitutions in the putative signal recognition motif (¹³¹RSFF¹³⁴ >¹³¹LLDS¹³⁴) and may have lost the ability to bind to its cognate signal molecule (19). We took advantage of that and tested whether the KinD CACHE mutant is also defective in stimulating biofilm formation in response to the addition of GM. Indeed, and as shown in Fig. 4, the strain expressing the CACHE mutant of KinD (ΔkinD amyE::kinD^mut) was unable to stimulate biofilm formation even in the presence of GM, similar to what was observed for the ΔkinD mutant, whereas the mutant strain with the wild-type kinD complementation (ΔkinD amyE::kinD^WT) formed robust biofilms comparable to those formed by wild-type cells (Fig. 4). This result strongly indicates that the biofilm-promoting activity of GM is mediated by the CACHE domain of KinD.

Fig 4.

Cells expressing the CACHE mutant of KinD fail to form robust biofilms in LBGM medium. Colony development and floating pellicle formation in LBGM medium by strains of the wild type (NCIB3610), a ΔkinD mutant complemented with either the wild type (CY189) or the CACHE mutant of kinD (CY190), and a ΔkinD mutant (RL4569) are compared.

To further prove the involvement of the CACHE domain in sensing the presence of GM, we constructed a hybrid histidine kinase which harbors the CACHE domain from KinD and the kinase and the ATPase domains from DegS (Fig. 5A). DegS is a sensory histidine kinase that is able to phosphorylate DegU, a DNA-binding response regulator (32). The DegS-DegU system controls many genes (both positively and negatively), including the fla-che operon, which encodes dozens of genes involved in motility and chemotaxis (33). Following the introduction of the hybrid kinase construct to a ΔdegS ΔkinD mutant, the expression of fla-che operon was controlled by the chimeric kinase. We applied real-time RT-PCR to examine the expression of the fla-che operon in the hybrid kinase background in the presence or absence of glycerol (see Materials and Methods). We specifically measured the expression of the flgB gene, the first gene in the fla-che operon. It was previously shown that highly phosphorylated DegU represses expression of the fla-che operon (33). We therefore hypothesized that if glycerol or its derivative activates DegU phosphorylation via the hybrid kinase, the flgB gene should be downregulated. The results from the real-time RT-PCR analysis showed that the addition of glycerol downregulated transcription of flgB very strongly. There was about a 10-fold reduction in the abundance of mRNA encoding flgB in response to glycerol (Fig. 5B). The similar experiment in the ΔkinD ΔdegS double mutant strain background without the hybrid kinase showed no significant changes in flgB expression in response to glycerol (see Fig. S5 in the supplemental material). This result provides further evidence that the CACHE domain of KinD can independently sense glycerol. Interestingly, the hybrid kinase was not activated in response to Mn²⁺ (data not shown), which probably indicates that manganese does not act directly on KinD in the native situation.

Fig 5.

Sensing of glycerol is mediated by CACHE, the extracellular sensor domain of KinD. (A) Schematic demonstration of the construction of the hybrid kinase (HK) harboring the extracellular sensor domain from KinD and the kinase and the ATPase domains from DegS. (B) Analysis by real-time RT-PCR of the expression of flgB, the first gene in the fla-che operon, which becomes under the control of the KinD-DegS chimeric kinase (MS178) in response to glycerol. The results represent the means ± standard deviations (SD) of two independent experiments.

Glycerol uptake or metabolism mutants are partially capable of promoting biofilm formation in response to GM.

We wanted to further test whether glycerol itself or metabolic derivatives of glycerol exerted the biofilm-promoting effect. To do so, we constructed two deletion mutants (the ΔglpF and ΔglpK mutants) that were defective in either glycerol uptake or metabolism. Among the deleted genes, glpF encodes a glycerol uptake facilitator and the mutation blocks the uptake of glycerol into the cells (34). The glpK gene encodes a glycerol kinase that is required in the first step of glycerol metabolism (34). We then assayed the mutants for their ability to respond to the addition of glycerol and to promote biofilm formation. As shown in Fig. 6A, both the ΔglpK and the ΔglpF mutants were only partially capable of responding to the addition of GM by promoting relatively less profound biofilm formation (Fig. 6A). This result indicates that the biofilm-promoting effect of glycerol may not be attributed solely to glycerol itself; rather, an intermediate metabolite of glycerol (catabolized by GlpK) may also be important in biofilm stimulation. We picked the ΔglpK and ΔglpF mutants for further assays of growth in a minimal medium with glycerol as the sole carbon source. As shown in Fig. 6B, both mutants failed to grow in the minimal medium with glycerol as the sole carbon source (although the ΔglpF mutant grew slightly at late time points) but grew well in the minimal medium with glucose as the sole carbon source (Fig. 6B). The result confirms that the glpK-encoded glycerol kinase and glpF-encoded glycerol transport facilitator are critical in utilization of glycerol as a carbon source.

Fig 6.

Mutants defective in glycerol uptake or metabolism are partially capable of promoting biofilm formation in response to GM. (A) Mutants with the glycerol uptake or metabolism mutations ΔglpF (YC877) and ΔglpK (FC96) are only partially capable of promoting biofilm formation in response to GM compared to the ability of the wild type. (B) Growth curves of the ΔglpF and ΔglpK strains in minimal medium supplemented with either 0.5% glucose (left) or 0.5% glycerol (right) as the sole carbon source.

Next we wanted to address how products of glycerol metabolism would be involved in biofilm formation if mutants with the glpK and glpF deletions were still capable of responding to GM. One possibility was that the phenotypes of the mutants with the glpK and glpF deletions were actually due to manganese. Indeed, our results indicate that mutants with the glpK or glpF mutation may still respond to manganese (see Fig. S6A in the supplemental material), as the phenotypes of the ΔglpK, ΔglpF, and wild-type strains in LBM medium were fairly similar. To quantitatively examine the effect of the glpK or glpF deletion on the response of the cells to the addition of manganese, we constructed two reporter strains in which the P_epsA-lacZ reporter fusion was introduced into the glpK and glpF deletion mutants (generating YC1250 and YC1251, respectively). The activities of the P_epsA-lacZ reporter from the two deletion mutants described above were compared to the activity of the reporter from the wild-type strain (YC110) by using cells from biofilm colonies on LBM medium (see Fig. S6A in the supplemental material). No significant difference was observed among the three tested strains (see Fig. S6B in the supplemental material). Taken together, these results suggest that the biofilm-promoting effect of glycerol is likely associated with its metabolism, while manganese has an independent mechanism for activation of Spo0A which is not related to glycerol metabolism.

The biofilm-promoting effect of GM is conserved in Bacillus species.

Finally, we wondered whether the biofilm-promoting effect of GM is conserved in other Bacillus species. We tested this hypothesis in two related Bacillus species: Bacillus licheniformis and Bacillus cereus. B. licheniformis is mainly involved in food spoilage, while some B. cereus strains are considered highly pathogenic to humans (35, 36). Interestingly, the similar biofilm-promoting effect caused by addition of GM was very dramatic in both bacteria, as judged by pigment production as well as formation of wrinkled colonies (Fig. 7A). Addition of GM also strongly stimulated formation of floating pellicles in B. licheniformis, very similar to what was seen in B. subtilis, whereas in B. cereus, only weak floating pellicles were seen in the presence of GM (Fig. 7B). A possible explanation for the observed phenotypes might be the high similarity of the CACHE domain of the B. subtilis KinD to the CACHE domain of the B. licheniformis KinD (59% sequence identity) but the relatively lower similarity to the CACHE domain of the B. cereus KinD (39% sequence identity). Overall, our finding indicates that the biofilm-stimulating effect in the presence of GM could be highly conserved in Bacillus species (Fig. 7C).

Fig 7.

The biofilm-promoting effect of GM is conserved in Bacillus species. (A) GM has a strong effect on colony morphology and pigment production in B. licheniformis ATCC 8480 and B. cereus ATCC 14579. (B) GM promotes formation of floating pellicles in B. licheniformis but not B. cereus. (C) Schematic representation of the signal transduction pathway responsible for the activation of a master regulator of matrix production and biofilm formation in B. subtilis.

DISCUSSION

A key question that remains unclear in the field of biofilm development is how the cells sense the environmental cues that trigger biofilm formation. B. subtilis provides a facile system to address this question, as the molecular pathways involved in regulation of cellular differentiation are well characterized. B. subtilis cells use the levels of Spo0A∼P, controlled by a network of five histidine kinases, as a sign to initiate biofilm formation. Recent studies showed that four histidine kinases, KinA, KinB, KinC, and KinD, are predominantly involved in controlling the Spo0A∼P level for biofilm formation (17, 37–39). In the present study, we establish that KinD is the most important sensor kinase for biofilm development in LBGM medium. Along with this finding, we note that KinC also has some contribution to the biofilm phenotype in LBGM medium, as a ΔkinC ΔkinD double mutant showed a very severe null phenotype which was very similar to that of the mutants deficient in phosphorelay (Fig. 2).

The important role of KinD in biofilm formation and biofilm-associated sporulation has been discussed in several recent studies (19, 20, 40). For example, Aguilar et al. (40) described KinD to be a bifunctional kinase/phosphatase likely sensing the presence of exopolysaccharides and proposed that its chief role is to act as a checkpoint protein linking spore formation to extracellular matrix production during B. subtilis biofilm formation. In another study (20), KinD was proposed to be capable of functioning as an osmosensor, possibly mediated by the transmembrane domain of KinD. KinD also seems to be important in B. subtilis colonization on the plant host by sensing plant-released signals and promoting biofilm formation on the root surface of the plant (19). It was also recently suggested that the activity of KinD depends on its partner, a lipoprotein termed Med (41). Based on the above-described studies as well as the present study, we suppose that the signal-sensing ability of KinD is multifaceted. Until now, we have not fully understood why a single kinase such as KinD possesses the ability to sense multiple distinct signals. It is possible that multiple domains of KinD may play different roles in signal sensing and signal transduction, especially under different inducing conditions.

It has been well established that wild-type cells of B. subtilis form robust biofilms mainly in specialized media, such as MSgg, E medium, and 2× SGG (4, 7, 42). It is not known why those media favor biofilm formation of B. subtilis. LB medium, on the other hand, is least favorable for biofilm formation by B. subtilis. In this study, we found that addition of glycerol and manganese can transform LB medium to a robust biofilm-promoting medium, LBGM medium. Importantly, both components (glycerol and manganese) were required for a robust biofilm phenotype (Fig. 1A), whereas the phenotypic effect of addition of glycerol or manganese alone to LB medium was relatively minor. This result is different from findings of a recent study which suggested that raising either the Mn²⁺or the glycerol concentration alone in specialized E medium (minimal medium) by the B1 strain of B. subtilis results in an increase in biofilm formation (26). In addition to the biofilm phenotype, LBGM medium also promoted robust sporulation and production of a brown pigment associated with biofilms. It was previously shown that manganese-dependent brown pigmentation of B. subtilis strain 168 was independent of sporulation (43). Our data indicate that the secreted pigment in LBGM medium is biofilm specific (see Fig. S3 in the supplemental material). Although we do not know the function of the secreted pigment, we presume that the pigment production is strongly associated with biofilm formation. The challenge for our future studies will be to elucidate the exact nature and functionality of the pigment and its connection to biofilm formation.

Cells of B. subtilis use a complex phosphoryl transfer system to control the activity of Spo0A, a transcriptional regulator of many cellular processes, including extracellular matrix production and sporulation. Our findings show that the phosphoryl transfer system and biofilm formation are activated by a combination of glycerol and manganese in LB medium. Our analysis indicated that the histidine kinase KinD, bearing the CACHE extracellular sensing domain, is capable of sensing glycerol or its derivative. Aside from being a signal molecule for the sensor kinase, we speculate that glycerol may play other roles in promoting biofilm formation by B. subtilis. This assumption is in part supported by the observation that the ΔglpK and ΔglpF mutants, which are defective in glycerol uptake or metabolism, lost a portion of their ability to promote robust biofilm formation in response to the addition of GM (Fig. 6). This observation indicates that possibly yet unknown molecules derived from glycerol metabolism may also contribute to the biofilm activation. Moreover, as we discussed earlier, when KinD senses glycerol, Mn²⁺ might act as a cofactor. Alternatively, it is also possible that Mn²⁺ acts as a cofactor further downstream, by promoting the efficiency of phosphoryl transfer in relay. For example, Mn²⁺ was previously shown to be complexed with Spo0F in a crystal structure (44). Spo0F is a critical component of the phosphorelay. Therefore, Mn²⁺ might be required for the activity of Spo0F and promote the efficiency of the relay; another possibility could be the regulation of Sda activity by Mn²⁺ (45), which in turn would influence the efficiency of phosphoryl transfer in the relay. Furthermore, as discussed above, the hybrid kinase was not activated in response to the Mn²⁺, which also supports the notion that Mn²⁺ acts downstream of KinD (in the phosphorelay) in the native condition.

The ability to form robust biofilms is conserved in the members of the Bacillus genus, which are highly ubiquitous in natural environments. In keeping with the idea that the KinD protein and proteins in the phosphorelay are also conserved in different Bacillus species, we found that the biofilm-promoting effect of GM is conserved in both B. licheniformis and B. cereus. These findings led us to propose that the signaling pathway for biofilm formation (Fig. 7C) is highly conserved in Bacillus species.