|Home | About | Journals | Submit | Contact Us | Français|
The response regulatory protein Spo0A of Bacillus subtilis is activated by phosphorylation by multiple histidine kinases via a multicomponent phosphorelay. Here we present evidence that the activity of one of the kinases, KinD, depends on the lipoprotein Med, a mutant of which has been known to cause a cannibalism phenotype. We show that the absence of Med impaired and the overproduction of Med stimulated the transcription of two operons (sdp and skf) involved in cannibalism whose transcription is known to depend on Spo0A in its phosphorylated state (Spo0A~P). Further, these effects of Med were dependent on KinD but not on kinases KinA, KinB, and KinC. Additionally, we show that deletion or overproduction of Med impaired or enhanced, respectively, biofilm formation and that these effects, too, depended specifically on KinD. Finally, we report that overproduction of Med bypassed the dominant negative effect on transcription of sdp of a truncated KinD retaining the transmembrane segments but lacking the kinase domain. We propose that Med directly or indirectly interacts with KinD in the cytoplasmic membrane and that this interaction is required for KinD-dependent phosphorylation of Spo0A.
The soil-dwelling, Gram-positive bacterium Bacillus subtilis is able to draw on a wide repertoire of adaptive responses to cope with adverse environmental circumstances. Several of these responses, such as spore formation, cannibalism, and biofilm formation, require the same master regulatory protein, Spo0A (11, 13, 15, 18, 35, 36). Spo0A is a member of the response regulator family of transcription factors and is active in its phosphorylated state (Spo0A~P) (14, 22). When phosphorylated, Spo0A can act as both an activator and repressor and is known to directly regulate about 120 genes (13, 25) and indirectly over 500 genes (12, 25).
Phosphorylation of Spo0A is governed by a multicomponent phosphorelay (5). At least four autophosphorylating sensor kinases, KinA, KinB, KinC, and KinD (the contribution of a fifth, KinE, is less certain), transfer phosphoryl groups to the relay protein Spo0F (21). Spo0F, in turn, transfers the phosphoryl group to Spo0B, which then phosphorylates Spo0A (5). It is thought that the activity of the kinases is regulated by specific signals whose nature is largely mysterious. KinA and KinB are particularly important for entry into sporulation (29, 37), whereas KinC and KinD contribute to triggering biofilm formation (23). Three of the kinases are integral membrane proteins (KinB, KinC, and KinD), whereas the fourth, KinA, is cytosolic. Flux through the relay is also governed by phosphatases, which dephosphorylate Spo0F~P or Spo0A~P (30, 31).
The activity of KinA is subject to inhibition by two other proteins: Sda and KipI (6, 32, 39). Both proteins bind to KinA, thereby blocking autophosphorylation. The synthesis of Sda is induced in response to DNA damage, preventing sporulation if chromosome replication is impaired (6, 16, 19, 33). The production of KipI appears to be linked to the utilization of nitrogen, but the specific signal that results in its inhibition of KinA is not known (39).
The activity of KinB is also influenced by interaction with other proteins. In addition to inhibiting KinA, Sda also inhibits the autophosphorylation reaction of KinB (3, 6, 20). To be functional, KinB also requires the presence of a lipoprotein, KapB. The kapB gene is in an operon with the kinB gene, and the two proteins are thought to form a complex in the cytoplasmic membrane (10).
Here we report that the activity of KinD also depends on a partner, the lipoprotein Med. Med was originally identified as a positive regulator of the competence gene comK, but its function remained obscure (27, 28). Our attention was drawn to Med through our studies of the phenomenon of cannibalism, in which cells that have activated Spo0A in response to nutrient limitation produce a toxin and a killing factor that kills sibling cells that have not activated Spo0A (11, 15). Colonies of cells that exhibit cannibalism are delayed in sporulation. It is thought that nutrients released by the dead cells delay sporulation by reversing or slowing the activation of Spo0A in the toxin- and killing factor-producing cells. The toxin and the killing factor are produced by the operons sdpABC (here designated sdp) and skfABCDEFGH (here designated skf), respectively. Both operons are indirectly under the control of Spo0A via AbbA and AbrB (2, 13). Additionally, Spo0A~P directly binds to and activates the skf promoter at low levels and acts as a direct repressor of skf transcription at high levels. Colonies of cells that are mutant for sdp or skf are mutant for cannibalism and exhibit accelerated sporulation. A previous survey of members of the Spo0A regulon for genes involved in cannibalism revealed that a mutation in the gene med caused accelerated sporulation (25). Here we show that Med acts by stimulating phosphorylation of Spo0A and that the target of Med is KinD.
All strains used in this study are listed in Table 1. Bacillus subtilis PY79 and 3610 and derivatives were grown in Luria-Bertani broth (LB) at 37°C for propagation. For measurements of gene activity with lacZ reporter genes, cells were induced to sporulate by nutrient exhaustion at 37°C in Difco sporulation medium (DSM) (26). For colony formation, cells were grown on solid MSgg medium (5 mM potassium phosphate, 100 mM morpholinepropanesulfonic acid [MOPS; pH 7], 2 mM MgCl2, 50 μM MnCl2, 50 μM FeCl3, 700 μM CaCl2, 1 μM ZnCl2, 2 μM thiamine, 0.5% glycerol, 0.5% glutamate, and 50 μg/ml [each] of threonine, tryptophan, and phenylalanine) containing 1.5% Bacto agar. When appropriate, isopropyl β-d-1-thiogalactopyranoside (IPTG; Sigma) was added to a final concentration of 1 mM, and xylose was added to a final concentration of 20 mM. When needed, antibiotics were added at the following concentrations for growth of B. subtilis: 10 μg per ml of tetracycline, 100 μg per ml of spectinomycin, 10 μg per ml of kanamycin, 5 μg per ml of chloramphenicol, 0.4 μg per ml of phleomycin, and 1 μg per ml of erythromycin.
General methods for molecular cloning and strain construction were performed according to published protocols (34). Long-flanking PCR mutagenesis was applied to make insertional knockout mutations in the med and kinD genes (38). Primers used in strain construction are listed in Table 2. Introduction of DNA into PY79 derivatives was conducted by transformation (17). SPP1 phage-mediated general transduction was used to introduce antibiotic resistance marker-linked mutations or overexpression constructs into derivatives of 3610 (40).
All oligonucleotide primers used in this study for strain construction are listed in Table 2. Construction of the Psdp-lacZ and Pskf-lacZ reporters was described previously (11, 15). To construct the IPTG-inducible Phyperspank-med construct, a PCR product containing an optimized ribosome binding site and the med open reading frame was generated using primers AVB007/AVB008 and cloned into the NheI and SalI sites of either pDR111, an amyE locus integration plasmid, or pDP150, a thrC integration plasmid (modified pDR111; gift of D. Rudner). The resulting Phyperspank-med-plus-lacI fragment was integrated into the chromosome at either amyE or thrC by double recombination to create strains AB392 and AB340, respectively. To construct the xylose-inducible PxylA-kinD* strain, a PCR product containing an optimized ribosome binding site and the first 882 bp (plus a stop codon) of the kinD open reading frame was generated using primers AVB176/AVB177 and cloned into the HindIII and SphI sites of pEH211 (modified pDR150; gift of D. Rudner). The resulting PxylA-kinD*-plus-xylR fragment was integrated into the chromosome at thrC by double recombination.
To construct the strain (AB434) that contains Psdp-lacZ, Phyperspank-med, and PxylA-kinD*, the amyE::Phyperspank-med construct was first introduced by transformation into AHB282 (8), a derivative of PY79, which contains an additional amyE site (ywrk::Tn917::amyE::cm) at position 317° on the circular chromosome marked by a chloramphenicol resistance gene. Transformants were selected for spectinomycin (Spec) resistance and chloramphenicol (Cm) sensitivity (Cmr Specs) for integration of Phyperspank-med (Specr) into the amyE sequence at the 317° position, but not into the native amyE locus.
To measure β-galactosidase activity, we used a kinetic assay described previously (7). Briefly, cells were cultured in DSM at 37°C in a water bath with shaking. A total of 1 ml of culture was collected every 30 min. Cells were spun down, and pellets were resuspended in 1 ml Z-buffer (60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, 1 mM MgSO4, 50 mM β-mercaptoethanol at pH 7.0). Ten microliters of these cells was added to individual wells of a clear 96-well plate containing 90 μl of 0.2 mg/ml lysozyme in Z-buffer. Cell lysis was allowed to proceed for 20 to 30 min at 37°C. Twenty microliters of 4 mg/ml 2-nitrophenyl β-d-galactopyranoside (ONPG; Sigma-Aldrich) in Z-buffer were then added to each well and mixed thoroughly. Absorbance at 420 nm for each reaction was read once per minute for 1 h at 37°C in a Synergy 2 plate reader (BioTek). β-Galactosidase activity (in arbitrary units [AU]) is reported as the rate of ONPG conversion (i.e., maximum rate of metabolism [Vmax], with units of the change in the optical density at 420 nm [ΔOD420] per minute) divided by the OD600 value of the sample at the time of collection. Standard curve analysis revealed that 1 AU was ~4 Miller units. β-Galactosidase activity for each strain was measured a minimum of three times in separate experiments. A representative experiment is shown.
For colony formation, cells were first grown to exponential-growth phase in LB broth, and 1 μl of these cultures was spotted onto solid MSgg medium containing 1.5% Bacto agar. The plates were incubated at 30°C for 72 h. Images of the colonies were taken using a SPOT camera (Diagnostic Instruments).
Cells were grown as biofilms on MSgg plates and incubated at 30°C for 72 h prior to harvesting for quantification. Samples were kept on ice during the following procedure. Entire colonies were harvested, resuspended in 1 ml phosphate-buffered saline (PBS), and subjected to mild sonication to disrupt the matrix and obtain single cells. For quantification of spores, each preparation was incubated at 80°C for 20 min to kill vegetative cells. To determine viable cell counts, serial dilutions were plated from each preparation before and after the 80°C incubation.
We asked if the accelerated sporulation phenotype of a med mutant was due to impaired expression of the sdp and skf operons. To do this, we examined the effect of a med deletion mutation (Δmed) on the expression of lacZ fused to the promoters for sdp (Psdp-lacZ) and skf (Pskf-lacZ) during sporulation in DSM. We observed that deletion of med led to decreased transcription from both promoters (Fig. 1), providing an explanation for the cannibalism defect of a med mutant. To ensure that the effect of the deletion was not due to a polar effect on the expression of the downstream gene comZ, we deleted comZ and observed no effect on the expression of sdp and skf (data not shown). Also the med deletion mutation was complemented by a copy of med that had been inserted at the thrC locus (data not shown).
Next, we determined the effect of overproducing Med on the expression of sdp and skf. To do this, we constructed a fusion of the med gene with the IPTG-inducible promoter Phyperspank and examined the effect of inducing this construct on the expression of Psdp-lacZ and Pskf-lacZ during sporulation. We observed that expression of both genes was markedly elevated when Med was overexpressed (Fig. 1).
Because transcription of both sdp and skf is dependent on the activity of Spo0A, we wondered if Med might act by enhancing Spo0A~P levels. We asked if Med acted through either KinC or KinD, as both of these kinases, like Med, are dispensable for sporulation but are thought to contribute to lowering AbrB levels (via Spo0A~P-mediated repression) during the transition to stationary phase (21). We decided to focus on sdp expression, as its regulation by Spo0A is less complex than that of skf.
We first compared Psdp-lacZ expression in a kinC mutant, a med kinC double mutant and a kinC mutant that also contained Phyperspank-med. We observed that removing or overexpressing med diminished or stimulated, respectively, Psdp-lacZ expression in the absence of KinC (Fig. 2A). We conclude that because the contributions of Med and KinC were additive, these proteins evidently act in separate pathways to promote sdp expression. We next compared Psdp-lacZ expression in a kinD mutant, a med kinD double mutant, and a kinD mutant that also contained Phyperspank-med. We observed that removing or overexpressing med had little effect on Psdp-lacZ expression in the absence of KinD (Fig. 2B). In other words, the effect of the kinD mutation was epistatic compared to that of the med mutation. These results suggest that Med and KinD act in the same pathway to promote expression of sdp.
We also tested the dependence of KinA and KinB on Med. Mutations of kinA and kinB impaired sdp expression only to a small extent. Nonetheless, the effects of removing or overexpressing med on Psdp-lacZ expression in the absence of KinA (Fig. 3A) or KinB (Fig. 3B) were similar (that is, additive) to what we observed in the absence of KinC. We conclude that Med acts separately from KinA, KinB, and KinC and is instead specifically required for the function of KinD.
A survey of genes required for efficient biofilm formation in B. subtilis revealed that a med mutation confers a modest defect in biofilm colony morphology (9). Biofilms are composed of long chains of cells that are held together by an extracellular matrix consisting of protein and polysaccharide. The production of matrix is governed by a regulatory network that, like cannibalism, requires Spo0A activity. We wondered whether Med influenced biofilm formation through KinD. To test this we examined the colony morphology of derivatives of the wild-type strain (3610), which forms architecturally complex colonies on solid, biofilm-inducing (MSgg) medium. We first compared the colony morphologies of a med mutant, a kinD mutant, and a kinC mutant in the 3610 background. We observed that all three single mutations resulted in similar subtle defects in colony morphology in which the colony was flatter and wider than that of the wild type (Fig. 4, top). We then compared a kinC kinD double mutant (which is known to exhibit a severe block in biofilm formation ), a kinC med double mutant, and a kinD med double mutant. We observed that both the kinC kinD double mutant as well as the kinC med double mutant had far more severe defects in colony morphology than the single mutant alone (Fig. 4, middle). In contrast, a med kinD double mutant was indistinguishable from the med or kinD single mutants. Lastly, we compared a strain that contained Phyperspank-med, a strain that contained Phyperspank-med as well as a kinC mutation, and a strain that contained Phyperspank-med as well as a kinD mutation. We observed that overexpression of Med resulted in a hyper-wrinkly colony morphology. Deletion of kinC had little, if any, effect on the colony morphology of a Med overexpression strain. However, deletion of kinD resulted in a colony that was indistinguishable from that resulting from a kinD mutation alone (Fig. 4, bottom). Thus, it appears that Med acts through KinD and independently of KinC to promote biofilm development.
It has been reported previously that deletion of kinD results in a small increase in the number of spores formed under biofilm-inducing conditions. More importantly, deletion of kinD partially restores sporulation to an eps tasA double mutant, which is blocked in matrix production. An eps tasA double mutant is normally conspicuously delayed in sporulation under biofilm-inducing conditions (1). These observations have led to the idea that KinD is a checkpoint protein that links spore formation to matrix production, with KinD acting as a phosphatase in the absence of matrix and as a kinase as matrix accumulates (1). We wondered if Med had a similar effect on sporulation in a biofilm. To investigate this, we harvested biofilms after 72 h of growth at 30°C and calculated the number of heat-resistant spores as a fraction of total living cells. We then expressed this number as a ratio relative to the wild-type strain (3610). We first compared a med mutant, a kinD mutant, a med kinD double mutant, an eps tasA double mutant, an eps tasA kinD triple mutant, and an eps tasA med triple mutant. We found that both single mutants had a slight increase in the number of heat-resistant spores compared to the wild type and that the med kinD double mutant was indistinguishable from either of the single mutants. We also observed that deletion of either kinD or med resulted in a partial restoration of sporulation in an eps tasA mutant background (Table 3). We next tested the effect of overexpressing med on sporulation in a biofilm. We compared a strain that contained Phyperspank-med, a strain that contained Phyperspank-med as well as a kinD mutation, and a strain that contained Phyperspank-med as well as an eps and a tasA mutation. We observed a 10-fold reduction in the number of heat-resistant spores relative to that of the wild type when Med was overproduced. However, when we introduced a kinD mutation into a Phyperspank-med background, sporulation was restored to levels comparable to that resulting from a kinD mutation alone. Overexpression of med in an eps tasA double mutant background did not have a noticeable additional inhibitory effect on sporulation (Table 3). In sum, these results indicated that the effect of Med on sporulation during biofilm formation is similar to that of KinD and that the effect of Med is dependent on KinD.
As a further test of the idea that med and kinD interact genetically, we constructed a fusion of the first 882 bp of the kinD open reading frame with the xylose-inducible promoter PxylA. This construct (here designated kinD*) contained both transmembrane segments and the extracellular double PAS-like sensor domain of KinD but did not contain the histidine kinase or ATPase domains. We noticed that when we induced expression of this construct, Psdp-lacZ expression was impaired. We wondered if the impaired level of Psdp-lacZ expression was dependent on Med. We tested the effect of inducing PxylA-kinD* in a med mutant as well as in a strain that contained Phyperspank-med. We observed that inducing kinD* had no effect in the absence of med and, strikingly, that overproduction of Med restored Psdp-lacZ expression in a KinD*-producing strain (Fig. 5). This result suggests that expression of kinD* has a dominant-negative effect on expression of Psdp-lacZ but that this effect is dependent on Med. These results are consistent with the following ideas: (i) Med and KinD interact, (ii) nonfunctional, truncated KinD, KinD*, and wild-type KinD compete for a limited number of Med molecules, thereby titrating Med, and (iii) when Med is overproduced, titration is overcome, restoring KinD activity.
The principal contribution of this work is the elucidation of a function for the previously mysterious lipoprotein Med. Med was originally identified as a positive regulator of the competence gene comK by Ogura et al. in 1997 (28). At the time, the authors rejected the idea that Med might be acting through Spo0A to induce comK expression because a med mutation did not affect sporulation. Our results show that Med does indeed act through Spo0A, and this is most likely the basis for the modest effect of a med mutation on comK expression that the authors observed. The authors also later characterized Med as a cell surface-localized lipoprotein (27, 28).
How does Med act to promote the activity of KinD? The simplest interpretation of our results is that Med interacts directly with KinD. However, efforts to demonstrate such an interaction by chemical cross-linking and by means of “pulldown” experiments with His-tagged Med have so far been unsuccessful. Therefore, we cannot exclude the possibility that the interaction is indirect. Whether the interaction is direct or indirect, the use of a truncated KinD shows that the transmembrane and extracellular sensor domains of KinD are all that is required to interact with Med, whose N-terminal region is anchored in the cytoplasmic membrane and whose C-terminal region projects from the outside surface of the membrane. Assuming that Med and KinD are in contact, we propose that the functional form of KinD is as a complex with Med. If so, KinD is not alone in its requirement for a lipoprotein partner. KinB, another kinase capable of activating Spo0A, requires a lipoprotein, KapB, for its activity. Although there is no evidence of a direct interaction between KinB and KapB, the fact that their genes are in the same operon has led to the idea that the two proteins physically interact (10).
Med acts by stimulating phosphorylation (and under some conditions, dephosphorylation) of Spo0A indirectly through KinD. Previous work has shown that Spo0A~P directly represses transcription of med (13, 25). The negative regulation of med by Spo0A thus constitutes a negative feedback loop wherein the activity of KinD is curtailed as Spo0A~P levels rise. This may allow other kinases, such as KinA and KinB, to take over phosphorylation of Spo0F and drive Spo0A~P levels high enough for sporulation to occur.
This work was supported by NIH grants GM18568 to R.L. and MH090948 to J. Liu and R.L.
Published ahead of print on 27 May 2011.