The Eukaryotic-Like Ser/Thr Kinase PrkC Regulates the Essential WalRK Two-Component System in

Summary

A central question in bacterial physiology is how bacteria sense and respond to their environment. The archetype of bacterial signaling systems is the two-component signaling system composed of a sensor protein histidine kinase that activates a transcription factor response regulator in response to a specific signal. In addition, bacteria also have signaling systems composed of eukaryotic-like Ser/Thr kinases and phosphatases. Even though these systems do not have dedicated transcription factors, they are capable of affecting gene expression. Here we show that a eukaryotic-like Ser/Thr kinase conserved in all sequenced Gram-positive bacteria converges with an essential two-component signaling system to regulate gene expression in the model organism Bacillus subtilis. We show that this eukaryotic-like Ser/Thr kinase phosphorylates the response regulator of a highly conserved and essential two-component signaling system, thereby increasing its activity. This phosphorylation results in the regulation of genes involved in the essential process of cell wall metabolism. Given that bacterial cell wall metabolism is the target of many known antibiotics, and mutations in both of these signaling systems change the antibiotic sensitivity of a number of important Gram-positive pathogens, we expect that our analysis will suggest novel insight into the emergence of antibiotic resistance.

Introduction

Regulatory Ser/Thr phosphorylation has been assumed to be largely absent from prokaryotes. However, recently, phosphoproteomic analysis has identified numerous (~50) proteins phosphorylated on Ser or Thr residues in both Gram-positive and Gram-negative bacteria [1]. Likely candidates for the enzymes responsible for these modifications are bacterial proteins with structural and mechanistic similarities to eukaryotic Ser/Thr kinases (eSTKs) and phosphatases (eSTPs) [2,3]. These enzymes have been identified in phylogenetically diverse bacteria, and eSTKs may be directly responsible for a number of the observed Ser/Thr phosphorylations. Thus, how these kinases regulate cellular physiology remains unclear, but their conservation across the majority of sequenced bacterial species suggests that they play an important role.

Regulatory phosphorylation in bacteria has largely been studied in the context of two-component systems (TCS) that govern the expression of a regulon in response to a physiological signal. These signaling systems include a sensor histidine kinase, which when activated transphosphorylates a cognate response regulator, typically a DNA-binding protein, on a conserved aspartate residue. Once phosphorylated, the response regulator dimerizes and binds specific sequences in the promoters of the genes in its regulon. Hints that eSTKs might similarly be involved in transcriptional regulation come from studies demonstrating that mutations in eSTKs cause large-scale changes in gene expression [4–6]. However, in contrast to TCS systems which typically use response regulator phosphorylation to regulate gene expression, eSTKs and the eSTPs generally lack identified DNA binding motifs, suggesting they likely mediate changes in gene expression indirectly. Although eSTKs and eSTPs are also not generally associated with a specific partner transcription factor, at least some of the observed transcriptional changes may be due to direct phosphorylation of a transcription factor. As the regulons governed by some TCSs have been extensively studied, overlap between transcriptional changes caused by mutations in eSTKs and eSTPs and certain TCS systems has been observed [7,8]. A plausible mechanism for these effects is that an eSTK phosphorylates components of a TCS system and thereby affecting gene expression. Consistently, eSTKs from a number of organisms have been shown to phosphorylate TCS proteins, and these modifications affect expression of genes in the respective TCS regulon [7,8].

All sequenced Gram-positive bacteria contain a single membrane-bound eSTK comprised of an extracellular domain composed of three or four PASTA repeats and a cytoplasmic domain with extensive homology to eukaryotic Ser/Thr kinases [3,9]. The PASTA (Peptidoglycan binding and Ser/Thr kinase associated) motif was originally identified in penicillin-binding proteins that are responsible for peptidoglycan polymerization, suggesting that it mediates peptidoglycan recognition [10]. Consistent with this hypothesis, isolated PASTA domains interact in vitro with a peptide corresponding to the peptidoglycan stem peptide [11–13]. In the Gram-positive bacterium B. subtilis, the PASTA-containing eSTK is PrkC, which is required for spore germination [14] and for regulation of the production of a secreted muramidase [15] in response to muropeptides. PrkC is transcribed during all phases of growth, and is co-transcribed with the eSTP PrpC [16]. Thus, PrkC could play a role in the regulation of peptidoglycan metabolism in B. subtilis.

A signaling system that is known to be important for peptidoglycan metabolism is the WalRK TCS, a well conserved, essential TCS found in low-GC-Gram-positive bacteria including B. subtilis [17–19]. Mutations in the response regulator WalR that alter its function significantly result in the generation of B. subtilis L-forms, variants that lack cell wall [20]. The essentiality of WalR likely derives from its regulation of at least one WalRK regulon member, although the identity of that gene, or combination of genes, is not known for most WalRK-containing species (see S. pneumoniae for an exception [21]). This essentiality is unusual; WalRK is the only essential TCS (out of 34) in B. subtilis [17]. The core WalRK regulon conserved in different species comprises ~10 genes involved in cell wall metabolism, including several peptidoglycan hydrolases [18]. In B. subtilis, ~20 WalRK regulon genes have been identified through several approaches including the use of a hybrid regulator [22], an IPTG-inducible WalRK operon [23,24], and ChIP [25]. Differential regulation of WalRK regulated genes is observed upon treatment with various cell wall-targeting antibiotics, suggesting that the WalK activating ligand is a molecule derived from the cell wall [18].

As the eSTK PrkC and the TCS WalRK have functional overlap, it is plausible that these two types of signaling systems converge in Gram-positive bacteria. Consistent with this hypothesis, deletion of the PrkC homologs in S. pneumoniae (StkP; [5]), S. mutans (PknB; [6]), S. pyogenes (Stk; [26]) and S. aureus (PknB; [4]) resulted in the down-regulation of genes known to be under control of either WalR or its homolog VicR. In the model Gram-positive B. subtilis, yocH, a well-characterized member of the WalRK regulon encoding a cell wall hydrolase, is regulated by both WalR and PrkC. WalR directly binds the yocH promoter in vitro [22], and yocH transcription is activated by WalR expression in vivo [27]. Intriguingly, yocH also demonstrates PrkC-dependent activation in response to cell wall-derived muropeptides [15]. Thus, these data suggest a functional interaction between PrkC and the WalRK TCS. Here we use in vitro and in vivo approaches to demonstrate that, consistent with this hypothesis, PrkC regulates the WalRK TCS through phosphorylation of WalR, resulting in the regulation of genes involved in cell wall metabolism by an eSTK.

Results

PrkC regulates yocH transcription in stationary phase

Prior work demonstrated that expression of the WalR regulon gene yocH is controlled both by WalR binding to P_yocH [22,27] and by PrkC through an unknown mechanism [15]. These observations suggest two possible mechanisms of PrkC dependent yocH activation: 1) PrkC acts on the WalRK TCS to regulate yocH and consequently other members of the WalR regulon, or 2) PrkC acts through another (unknown) mechanism on yocH (Fig 1A). The first hypothesis predicts that changes in PrkC activity resulting in the activation of yocH would also result in transcriptional changes across the WalR regulon, namely increased activation of genes activated by WalR, as well as increased repression of genes repressed by WalR. The second hypothesis predicts that changes in PrkC activity that activate yocH may have little, or no consistent, effect on the expression of other members of the WalR regulon.

To test these alternate hypotheses, we first determined when PrkC activity has the strongest effect on yocH expression. We constructed a transcriptional fusion of the yocH promoter region to the luxABCDE operon, originally from P. luminescens, optimized for expression in Gram-positive bacteria [28] and integrated it into an ectopic locus in the chromosome. As the luminescence signal from the luxABCDE operon is relatively unstable, it can be used to measure both increases and decreases in gene expression [29]. Therefore, we used the P_yocH-luxABCDE transcriptional reporter (hereafter referred to as P_yocH-lux) to continuously monitor changes in yocH expression throughout growth in rich media (LB) in a plate reader. Under these conditions, the doubling time for the WT strain harboring the P_yocH-lux reporter is roughly 22 minutes in exponential phase, comparable to the optimal growth rate of B. subtilis in LB at 37°C. We measured yocH expression as well as the corresponding growth curves (OD₆₀₀) in the plate reader every 5 minutes in four genetic backgrounds: wild type (‘WT’), ΔprpC (no phosphatase), ΔprkC (no kinase), and Δ(prpC-prkC) (neither phosphatase nor kinase). Under these conditions, the growth of these four strains is consistent with the previously reported stationary phase survival defect of ΔprkC strains and increased longevity of ΔprpC strains (Fig 1B) [30]. To highlight the consistent similarity between the ΔprkC and Δ(prpC-prkC) strains, indicating that the observed ΔprpC phenotypes are epistatic to prkC, ΔprkC mutants are grouped by color (Fig 1, shades of blue). Furthermore, these experiments also demonstrated that while there is no growth phenotype for ΔprkC and ΔprpC mutant strains during log phase growth, mild differences begin to appear during the transition to stationary phase and become more pronounced later in stationary phase prior to the onset of lysis (Fig 1B). Since transcriptional changes near the onset of lysis may be complicated to interpret, we chose to examine PrkC-dependent changes in gene expression at four characteristic growth points, as determined by position on the growth curve and corresponding measured OD₆₀₀ (arrows indicate points for WT, Fig 1B): 1) exponential phase (OD₆₀₀~0.1), 2) late log (OD₆₀₀~0.2), 3) transition phase (OD₆₀₀~0.3), and 4) stationary phase (OD₆₀₀~0.5). We picked OD₆₀₀~0.5 as a representative point for stationary phase measurements, as we observe comparable growth rates for all genetic backgrounds near this OD₆₀₀, and it is prior to the onset of lysis. Note that OD₆₀₀ values measured by the plate reader differ from those measured by a standard laboratory spectrophotometer. See materials and methods for more details.

To characterize PrkC-dependent yocH expression at each of these characteristic growth phases, we compared the normalized luminescence (relative luminescence, RLU, normalized by OD₆₀₀) of the P_yocH-lux reporter in each genetic background at each characteristic growth phase (Fig 1C). Interestingly, although prkC and prpC are expressed throughout all growth phases [16], mutants in prkC and prpC did not exhibit strong changes in yocH expression until stationary phase (Fig 1C). To determine more precisely when in the growth cycle changes in yocH expression begin to appear, we plotted normalized luminescence as a function of OD₆₀₀ for data acquired at 5 minute intervals for each genetic background (Fig 1D), yielding a synchronized picture of gene expression as a function of growth state. For clarity, these graphs show only the pre-lysis data (to maximum OD₆₀₀) for each strain. In WT cells, yocH expression is high in exponential phase and falls in stationary phase (Fig 1C and 1D, gray), consistent with prior work [31]. Interestingly, this trend was reversed in a ΔprpC background beginning at the exit from log phase (OD₆₀₀~0.3, Fig 1D, ii). This effect is PrkC dependent as both ΔprkC and Δ(prpC-prkC) strains show very similar expression profiles (compare shades of blue, Fig 1D). The WT expression profile (gray) lies between ΔprpC and ΔprkC, and is strongly biased towards ΔprkC. This suggests that under these conditions, PrkC exerts a relatively small average effect on yocH expression in WT cells, ~3.5 fold (S1 Fig), when compared to the total possible PrkC-dependent effect observed in a ΔprpC background. The difference in expression between ΔprpC and ΔprkC backgrounds suggests that PrkC is capable of exerting a >20 fold change on yocH expression in stationary phase.

PrkC regulates the WalR regulon genes iseA and pdaC

If PrkC regulates yocH expression through a WalR-dependent mechanism (Fig 1A), we would expect it to control expression of other genes in the WalR regulon. Two genes repressed by WalR were found to have the highest average WalR occupancy of their promoters in a ChIP study [25]: iseA (yoeB), encoding an inhibitor of autolysin activity [32], and pdaC (yjeA), a peptidoglycan deacetylase [33]. To determine if PrkC regulates iseA and pdaC under the same conditions where we observed PrkC-dependent regulation of yocH, we constructed P_iseA-luxABCDE and P_pdaC-luxABCDE transcriptional reporters. Using these reporters and the same growth conditions and techniques used to monitor yocH expression, we measured iseA and pdaC expression using normalized luminescence as a function of OD₆₀₀ (Fig 2A). Strikingly, we observed that both iseA and pdaC exhibit strong PrkC-dependent repression at the same OD₆₀₀ and growth phases where yocH exhibited PrkC-dependent activation. That is, beginning at the exit from log phase (OD₆₀₀~0.3) and becoming most significant in stationary phase, both iseA and pdaC show increased repression in a ΔprpC background that is relieved in both ΔprkC and Δ(prpC-prkC) backgrounds (Fig 2A). We examined this growth phase dependence by extracting the normalized luminescence at four specific optical densities from the continuous data (Fig 2B). Similarly to what we observed for yocH, the WT strains have expression profiles in stationary phase between the ΔprpC and ΔprkC backgrounds, consistent with opposing regulation. In strains carrying a ΔprpC mutation, P_iseA-lux in stationary phase is expressed ~4x less and P_pdaC-lux is expressed >10x less than in strains carrying Δ(prpC-prkC) mutations (Fig 2B). As yocH is activated by WalR and both iseA and pdaC are repressed by WalR, these data strongly support a model (Fig 1A) where PrkC acts through the WalRK system to transcriptionally regulate yocH, iseA, pdaC, and other members of the WalR regulon.

PrkC expression complements ΔprkC phenotypes

We tested whether inducible heterologous PrkC expression complements a ΔprkC mutation specifically in stationary phase by comparing the expression of the WalR reporters in strains carrying an inducible copy of prkC (P_spac-prkC) to their expression in WT and ΔprkC backgrounds (Fig 3). To confirm that these results represent systematic changes in expression from transition phase through early stationary phase (OD₆₀₀~0.3–0.5), we also compared the normalized luminescence as a function of OD₆₀₀ for each strain and IPTG concentration (S2 Fig). Without the addition of the inducer IPTG, the strain to test complementation, ΔprkC P_spac-prkC, appears similar to ΔprkC for all three promoters (compare two lightest shades of blue, Figs 3 and S2). Under these conditions, the small amount of “leaky” expression without inducer is sufficient to complement expression of P_yocH, and nearly complement P_iseA (compare gray and lightest shade of blue, Figs 3 and S2). At 10 μM IPTG, P_spac-prkC complements the ΔprkC deletion for the expression of P_pdaC, and P_yocH and P_iseA also display similar expression to the wild type (compare gray and blue, Figs 3 and S2). For each promoter, the effect of increasing inducer concentration was consistent with increasing WalR activity specifically in stationary phase: increased repression of pdaC and iseA, and increased activation of yocH. Although ΔprkC P_spac-prkC was grown in the continuous presence of inducer, changes in gene expression for pdaC, iseA, and yocH did not appear until early stationary phase, consistent with the previous observation that the PrkC-dependent effect on the WalR regulon is growth phase specific (Figs 1 and 2). Thus, a low level of prkC expression in trans is sufficient to complement expression of pdaC, iseA, and yocH in stationary phase.

PrkC overexpression resembles ΔprpC phenotypes

The eSTK prkC is co-transcribed with its cognate eSTP prpC. Previous work suggested that the kinase activity of PrkC and the phosphatase activity of PrpC have opposing modes of action on B. subtilis physiology. Specifically, mutants in prkC and prpC have opposing stationary phase lysis phenotypes ([30] and Fig 1B) and overexpression of PrpC was shown to prevent PrkC mediated induction of yocH [15]. Consistent with these observations, prkC and prpC mutations have opposing effects on the expression of the WalR regulon genes yocH, iseA, and pdaC in stationary phase (Figs 1 and 2). We therefore sought to determine if the expression phenotypes for WalR regulon genes in a ΔprpC mutant background reflect high net PrkC kinase activity. To test this, we overexpressed prkC from P_spac-prkC (in a strain lacking endogenous prkC) using higher levels of IPTG induction than were necessary to complement the ΔprkC mutation (Fig 3). Higher levels of induction (100 μM or 1 mM IPTG) of P_spac-prkC caused expression of P_yocH to increase and expression of P_iseA and P_pdaC to decrease (Fig 4), consistent with further increases in WalR activity. These effects are similar to those observed in the presence of a ΔprpC mutation (green) suggesting that kinase overexpression overwhelms the phosphatase activity under these conditions (Fig 4). This further suggests that the ΔprpC expression phenotypes observed for yocH, iseA, and pdaC in stationary phase (Figs 1 and 2) are caused by high levels of PrkC-dependent WalR phosphorylation, generated by the absence of phosphatase activity. With 100 μM or 1 mM IPTG induction (shades of dark blue), the reporters in the ΔprpC mutant (green) background all appear similar (Fig 4), suggesting that the reporters in the ΔprpC background reflect the maximal kinase activity for that condition.

Although the prpC-prkC transcript is detected from exponential through stationary phase in LB [16], the heterologous overexpression of PrkC (Fig 4) produces only mild expression phenotypes prior to stationary phase. This observation, combined with the data indicating that PrkC-dependent transcriptional changes in the WalR regulon occur primarily in stationary phase (Figs 1 and 2), suggests that these effects are not simply caused by changes in prkC expression.

The eSTK PrkC phosphorylates WalR in vitro

As PrkC is membrane-bound and does not contain any predicted DNA binding domains, the observed co-regulation of three separate genes in the WalR regulon is likely to be indirect. One potential mechanism is a direct interaction of PrkC with WalR resulting in the phosphorylation of WalR on Ser or Thr residue(s). We examined this possibility by taking advantage of the previous observation that the cytoplasmic domain of PrkC autophosphorylates in vitro [34] and observed that PrkC robustly transphosphorylated WalR (Fig 5A). Mass spectrometry analysis of a PrkC-WalR kinase reaction performed with cold ATP suggested that WalR Thr101 was the only phosphorylated residue. To verify this interpretation, we generated mutant WalR proteins containing either a T101S or a T101A mutation and repeated the in vitro kinase assays. Strikingly, PrkC did not transphosphorylate either WalR T101S or WalR T101A mutant proteins, suggesting PrkC has a high degree of specificity for Thr101 (Fig 5A). WalR Thr101 is located at the dimerization interface of the receiver domain (Fig 5B, red).

PrkC has specificity for WalR in vitro

Work with M. tuberculosis eSTKs revealed specificity for particular phosphoacceptors based on sequences surrounding the phosphosite [35]. A ClustalO alignment of the 34 B. subtilis response regulators identified eight response regulators with a Thr or Ser residue at the position homologous to WalR Thr101 (PhoP, YvrH, PsdR, YkoG, YclJ, YrkP, LytT, and CssR; Fig 5C). In all of these response regulators, the Ser or Thr residue is surrounded by well-conserved residues (Fig 5C, bold), including those located on the surface exposed loop from the +1 to +6 position (Fig 5B, yellow). However, despite this strong conservation, when we purified the eight response regulators and performed in vitro kinase assays with PrkC, we observed that PrkC much more robustly phosphorylated WalR than the other response regulators following 10’ incubation (Fig 5D). Increasing the incubation time of the reactions to 30’ did not increase the phosphorylation of any other substrates (S3A Fig). We verified that similar amounts of total protein were loaded for each response regulator (S3B Fig), and that none of the response regulators exhibited a background signal in the absence of PrkC (S3C Fig).

WalR is phosphorylated on Thr101 in vivo

The ability of PrkC to affect the expression of WalR-dependent genes and to phosphorylate WalR in vitro on Thr101 suggests that this modification occurs in vivo. To detect WalR Thr phosphorylation in vivo, we immunoprecipitated FLAG-tagged WalR from OD₆₀₀ matched stationary phase LB cultures of wild type (WT), ΔprpC, and ΔprkC strains. As a control for non-specific immunoprecipitation of proteins, strains lacking a FLAG-tagged protein were also used. Each sample was run on a gel cast with Phos-tag acrylamide, allowing for the separation of phosphorylated proteins by SDS-PAGE (Fig 6, top). To determine if any shifted bands observed were dependent on Phos-tag, and therefore indicative of phosphorylation, a control gel without Phos-tag was run in parallel (Fig 6, bottom). Since Thr phosphorylations are relatively heat stable compared to Asp phosphorylations, samples were boiled prior to gel loading. Visualized by subsequent Western blotting for the FLAG-tagged protein, a Phos-tag-dependent second band was observed in the WT sample, was significantly enhanced in the ΔprpC sample, but did not appear in the ΔprkC sample (Fig 6, top). This result indicated the presence of a heat stable PrkC-dependent WalR phosphorylation in vivo.

To verify that this PrkC-dependent WalR phosphorylation was on the same residue observed in vitro, mass spectrometry was performed on WalR-FLAG immunoprecipitated from ΔprpC cells. In two separate experiments, phospho-peptides containing Thr101 were found by mass spectrometry, providing further evidence that this modification occurs in vivo (S4 Fig). To confirm that the experimentally observed phospho-peptides and unmodified peptides had masses that are consistent with the WalR Thr101 region, the experimentally determined masses were compared to both the theoretically predicted masses and the masses observed from synthetic peptides (S4 Fig).

WalR Thr101 is required for the PrkC-dependent effect on yocH expression

PrkC-dependent WalR Thr101 phosphorylation was observed in vivo, suggesting that WalR Thr101 phosphorylation was responsible for the PrkC-dependent gene expression phenotypes observed for yocH (Fig 1). Therefore, we generated a strain containing a walR T101A mutation at the endogenous locus. Since the WalR T101A mutant may exhibit PrkC independent changes in expression compared to wild type WalR, direct comparison of yocH expression between waR^T101A and walR^WT backgrounds cannot be performed. Therefore, we generated P_yocH reporter strains in the walR^T101A background in otherwise WT, ΔprpC, and Δ(prpC-prkC) backgrounds to test for PrkC dependence. In parallel, we measured the expression of yocH using normalized luminescence as a function of OD₆₀₀ in both walR^T101A and walR^WT backgrounds (Fig 7A). Log phase yocH expression was comparable between walR^WT and walR^T101A, indicating that the walR^T101A mutant is functional at the yocH promoter in vivo (Fig 7A, compare OD₆₀₀<0.3, left and right). Strikingly, in the walR^T101A background, PrkC-dependent yocH activation in stationary phase was lost, with the WT, ΔprpC and Δ(prpC-prkC) backgrounds all showing similar expression profiles (Fig 7A, left). This is in contrast to the walR^WT background, particularly in stationary phase (OD₆₀₀>0.4), where the strongest PrkC-dependent activation is observed in a walR^WT background (Fig 7A, right inset). At the same stationary phase OD₆₀₀ used in Figs 1–4, we found that in the walR^WT background, a >20 fold difference is observed between the ΔprpC and Δ(prpC-prkC) backgrounds, whereas no significant difference is observed in the walR^T101A mutant background (Fig 7B). In the 40 min prior to OD₆₀₀~0.5 (‘Stationary Phase’) we observed comparable growth rates for WT, ΔprpC, ΔprkC, Δ(prpC-prkC) in both walR^WT (S5 Fig, top) and walR^T101A (S5 Fig, bottom) backgrounds.

PrkC-dependent repression of iseA and pdaC is reduced in a WalR T101A mutant

To confirm that WalR Thr101 is also required for the PrkC-dependent repression of iseA and pdaC, we performed experiments similar to those described above for yocH. Log phase expression of iseA and pdaC in walR^T101A and walR^WT backgrounds was comparable, suggesting that WalR T101A is also functional at these promoters as a repressor (S6 Fig). Whereas both P_iseA and P_pdaC show strong PrkC-dependent repression in the walR^WT background in stationary phase, in the walR^T101A background, the repression is at least ~10 fold relieved between ΔprpC and ΔprkC (Fig 8). Consistent with a loss of PrkC-dependent activity at these promoters in stationary phase, the loss of repression in the walR^T101A mutant is primarily due to the reduction of repression in the ΔprpC background, not to an increase in expression in a ΔprkC background. The residual-PrkC dependent effect on these promoters may be due to additional PrkC activity not dependent on WalR Thr101 phosphorylation that is sensed by WalK (S6 Fig). As iseA and pdaC were identified as the two sites in the B. subtilis chromosome with the highest WalR occupancy [25], it is possible that they are more sensitive than yocH to WalRK activity.

Discussion

Here, we provide the first known mechanism for PrkC-dependent gene regulation in B. subtilis by demonstrating that PrkC phosphorylates the response regulator WalR on Thr101 both in vitro and in vivo. Overall, this work supports a model (Fig 9) where PrkC, a PASTA-domain-containing eSTK, is activated and phosphorylates WalR at Thr101. This is supported by the in vitro data demonstrating direct phosphorylation of WalR Thr101 by PrkC (Fig 5), as well as the presence of PrkC-dependent phosphorylation observed at Thr101 in vivo (Figs 6 and S4). Transcriptional data from the WalR-dependent genes yocH, iseA, and pdaC (Figs 1,2, 7 and 8) demonstrates that this secondary phosphorylation increases both the activation and repression of WalR regulon genes (Fig 9, bottom). In ΔprpC strains, increases in WalR Thr101 phosphorylation are observed in vivo, and changes in expression of WalR-dependent genes is consistent with high levels of PrkC activity (Figs 4 and 6), demonstrating that PrpC and PrkC have opposing modes of action.

The effect of PrkC on the expression of the WalR regulon is most significant as growth slows in stationary phase (Figs 1 and 2). As continuous expression of prkC in trans does not cause changes in gene expression during rapid growth, the mild effect of PrkC on the WalR regulon in log phase is likely not due to lack of expression. Rather, it suggests that PrkC regulation of the WalRK system becomes significant as the activity of the WalRK system begins to decrease in stationary phase [31]. Thus, by acting as a second sensor for the system, PrkC can provide an input to increase WalR activity even as WalK activity decreases. Since both PrkC and WalK are hypothesized to sense cell wall-related processes, and both are highly conserved signaling systems in Gram-positive bacteria, this may suggest that they have separate sensory inputs in vivo, such as sensing different aspects or different steps in cell wall growth and metabolism. The PASTA domain of PrkC interacts in vitro with muropeptides [13], suggesting that it may directly sense some aspect of the cell wall. WalK localizes to septal regions, and it has been hypothesized that some aspect of cell wall synthesis during active division is required to keep WalK activity high [18,36–39]. Taken together, separate sensory inputs working through PrkC and WalK may regulate the expression of genes responsible for cell wall metabolism by phosphorylation of WalR on two different sites.

WalR Thr101, the site of PrkC phosphorylation both in vitro and in vivo, is in the receiver domain at the α4-β5-α5 dimerization interface (Fig 5B). This region is important for both activity and regulation of protein-protein interactions in OmpR family response regulators such as WalR [40,41]. In response regulators of this type, this region has been shown to undergo important conformational changes when Asp phosphorylation occurs [42]. Therefore WalR Thr101 phosphorylation may impact the structural conformation of WalR in complex ways, including by altering the mechanism by which WalK-mediated Asp phosphorylation activates WalR. However, the gene expression data presented here suggests that the net effect of WalR Thr101 phosphorylation is consistent with an increase in WalR activity.

The high degree of sequence conservation between B. subtilis response regulators with conserved Ser or Thr residues in the receiver domain (Fig 5C) suggests that the specificity of PrkC for WalR is mediated by a region which is not conserved. Previous work dissecting the in vitro specificity of six different eSTKs in M. tuberculosis suggested that phosphorylation is enhanced with highly hydrophobic residues at the +3 and +5 positions surrounding the phosphosite [35]. Given that this region is surface exposed and the +3 site is conserved among the B. subtilis response regulators, the residue at the +5 position may govern specificity. This possibility is consistent with data that residues not directly adjacent to the phosphorylation site can affect the specific phosphorylation of response regulators by histidine kinases [43,44].

Other PrkC substrates including EF-Tu, EF-G, CpgA, YezB, YkwC, YvcK, and several metabolic enzymes [14,45–49] have been reported. However, evidence suggesting that PrkC-dependent phosphorylation of a transcription factor is important for gene regulation is relatively scarce. Recently, the transcription factor AbrB was found to be phosphorylated in vitro on S86 by PrkC and the two other B. subtilis eSTKs YabT and PrkD, resulting in a decrease in AbrB DNA binding activity [50]. However, AbrB phosphorylation in vivo was dependent on the presence of all three eSTKs, making the precise signal sensed and the dominant kinase responsible for physiologically relevant phosphorylation unclear.

Numerous examples of eSTK-dependent phosphorylation of transcription factors, including response regulators, have been reported in other Gram-positive bacteria (for recent reviews, see [2] and [8]). Although most of these examples are derived from in vitro observations, in vivo phosphorylation has been observed for both S. pneumoniae RR06 [51] and S. pyogenes CovR [52]. Examples in which specific in vitro phosphosites have been identified include S. aureus GraR [53] and VraR [54], M. tuberculosis DosR and Rv2175c [55–57], S. pneumoniae RitR [58] and RR06 [51], and Group A and Group B Streptococci CovR [52,59]. In some of these examples (e.g., RitR [58], GraR [53] and VraR [54]), the eSTK phosphorylates the response regulator in the DNA binding domain, suggesting that the effects on gene expression are caused by changing the affinity of the response regulator for its DNA binding sites. Alternatively, response regulators can be phosphorylated in the receiver domain (e.g., CovR [52,59], DosR [55], and VraR [54]). Of these, the most similar example to the WalR Thr101 phosphorylation we observed here is S. aureus VraR, which is phosphorylated in vitro on four residues including Thr106 located at the dimerization interface.

The essentiality of WalRK and its homologues in many low G+C Gram-positive bacteria, combined with its effect on virulence [60] and antibiotic resistance [61,62], has made mutations in the WalRK system in clinical pathogens of particular interest. Of note, single nucleotide substitutions within walR cause an increase in vancomycin resistance [63]. Whole genome sequencing of clinical isolates of VISA (vancomycin intermediate S. aureus) has identified WalR mutations (walR T101S and walR T101R) at the homologous Thr to the PrkC-WalR Thr phosphosite reported here [64]. This observation suggests that WalR Thr101 plays an important role in antibiotic resistance in S. aureus.

In conclusion, our characterization of PrkC-dependent phosphorylation of WalR helps to clarify the role of PrkC in stationary phase B. subtilis physiology and gene regulation. The intersection of the eSTK-eSTP pair PrkC-PrpC with the essential WalRK TCS system provides two sensory inputs to regulate the essential process of cell wall metabolism in B. subtilis. Parallels with closely related systems in Gram-positive pathogens suggest that this mechanism of regulation may be conserved and relevant to further understanding the role of eSTK phosphorylation of response regulators in Gram-positive physiology in the future.

Materials and Methods

Strains and strain construction

Bacillus subtilis strains were constructed by transformation into B. subtilis 168 trpC2 (PB2) obtained from Chet Price [30] and its derivatives unless otherwise noted. Strains used in each figure are listed in S1 Table, and details of strain and plasmid construction are in S2, S3 and S4 Tables. B. subtilis strains were transformed with 5–10 μl of plasmid DNA or 1–2 μl of genomic DNA using the two-step method [65]. Genomic DNA was prepared using the Wizard Genomic DNA Purification Kit (Promega) according to the manufacturer’s instructions. Cultures were grown in LB (Lennox).

Luminescence plate reader assays for growth curves and gene expression

Early log phase cultures were grown in LB inoculated from single colonies obtained from plates grown overnight at 37°C. The cultures were diluted 1:30 (150 μl final volume) into fresh LB in 96-well plates and grown at 37°C with continuous shaking for >20 h in a Tecan Infinite 200 plate reader. Measurements of luminescence and OD₆₀₀ were taken at 5 min intervals. Media only and lux- controls were used for background subtraction for OD₆₀₀ and luminescence respectively. All cultures were grown in triplicate. Note that OD₆₀₀ values measured in the Tecan plate reader differ from the spectrophotometer by both a scale factor and an offset, making OD₆₀₀~0.1 measurement in the plate reader correspond to OD₆₀₀~0.4 in a standard laboratory spectrophotometer.

Immunoprecipitation of WalR-FLAG

WalR-FLAG was immunoprecipitated from 25 ml (Phos-tag experiment) or 600 ml (mass spectrometry experiment) cultures at OD₆₀₀~0.4 using anti-FLAG M2 magnetic beads (Sigma M8823) according to the manufacturer’s instructions with the following modifications. Initial lysis to break down the cell wall was performed in 50 mM Tris pH 7.5, 150 mM NaCl, 0.1% NP-40, 5 μg/ml DNaseI, 3 mM MgCl₂, 1x Halt phosphatase inhibitor cocktail (Pierce), 1 mM PMSF, and 1 mg/ml lysozyme. After 20 min of lysis on ice with periodic vortexing, cell pellets were spun down and resuspended in lysis buffer without lysozyme and applied to 0.1 mm silica beads (BioSpec, 11079101z) in pre-chilled screw cap tubes. Cells were lysed using a FastPrep-24 5G instrument (MP Biomedicals) using 4 runs of 6.5 m/s for 40 seconds with samples placed on ice for 3 min between each run. Lysates were cleared by centrifugation at 19,000 x g for 20 min at 4°C. Elutions from the anti-FLAG M2 magnetic beads were performed competitively by addition of the 3x-FLAG peptide (Sigma F4799).

Separation and visualization of phosphorylated WalR by Phos-tag acrylamide

10% resolving gels for SDS-PAGE were cast with ±37.5 μM Phos-tag acrylamide (Wako, AAL-107, 304–93521), and 5% stacking gels were prepared as per the manufacturer’s instructions. Prior to gel loading, 35 μL of each sample was boiled at 100°C for 15 seconds to remove any residual heat labile Asp phosphorylation. Approximately 30 ng of WalR-FLAG was loaded per lane. Gels were run and transferred as recommended by the manufacturer with the following modifications: SDS-PAGE gels were loaded and run in parallel on a Bio-Rad mini-protean gel apparatus at constant voltage (150 V) for approximately 100 min (an additional ~20 minutes after the dye front ran off the gel) at 4°C. Gels were fixed twice for 15 min in transfer buffer supplemented with 5 mM EDTA to remove Mn²⁺ from the gel, followed by a 10 min incubation in transfer buffer. Transfer to PVDF membranes was performed using a wet tank method (Bio-Rad Mini Trans-Blot) at a constant 350 mA for 80 min. Subsequent Western blotting was performed using standard protocols with an anti-FLAG-Peroxidase antibody (Sigma-Aldrich s#SAB4200119).

In vitro kinase assays with PrkC

PrkC kinase reactions were performed in kinase buffer (50 mM Tris-HCl pH 7.5, 50 mM KCl, 10mM MgCl₂, and 0.5 mM DTT). All reactions were performed in the presence of 400 μM cold ATP and 2 μCi (~ 20 nM) of γ-³²P-ATP. The eSTK was used at a final concentration of 1 μM and substrates were added at a final concentration of 4 μM. Reactions were incubated at 37°C for 30 min unless otherwise indicated. 5X SDS-PAGE buffer was added to the reactions and subsequently boiled at 95°C for 5 min. Samples were run on a 12% SDS-PAGE gel and dried for 30 min at 80°C in a gel dryer. Dried gels were exposed to a phosphoscreen and visualized using a Typhoon Scanner (GE Healthcare).

Additional Materials and Methods, including protein purification and mass spectrometry for the detection of WalR phosphorylation both in vitro and in vivo, can be found in S1 Text.

Supporting Information

Attachment 1

Attachment 2

Attachment 3

Attachment 4

Attachment 5

Attachment 6

Attachment 7

Attachment 8

Attachment 9

Attachment 10

Attachment 11