By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
The Bacillus subtilis extracytoplasmic function (ECF) σ factor σM is inducible by, and confers resistance to, several cell envelope-acting antibiotics. Here, we demonstrate that σM is responsible for intrinsic β-lactam resistance, with σX playing a secondary role. Activation of σM upregulates several cell wall biosynthetic enzymes including one, PBP1, shown here to be a target for the beta-lactam cefuroxime. However, σM still plays a major role in cefuroxime resistance even in cells lacking PBP1. To better define the role of σM in β-lactam resistance, we characterized suppressor mutations that restore cefuroxime resistance to a sigM null mutant. The most frequent suppressors inactivated gdpP (yybT) which encodes a cyclic-di-AMP phosphodiesterase (PDE). Intriguingly, σM is a known activator of disA encoding one of three paralogous diadenylate cyclases (DAC). Overproduction of the GdpP PDE greatly sensitized cells to β-lactam antibiotics. Conversely, genetic studies indicate that at least one DAC is required for growth with depletion leading to cell lysis. These findings support a model in which c-di-AMP is an essential signal molecule required for cell wall homeostasis. Other suppressors highlight the roles of ECF σ factors in counteracting the deleterious effects of autolysins and reactive oxygen species in β-lactam-treated cells.
The bacterial cell envelope is crucial for maintaining cell shape and counteracting turgor pressure and is an important target for many antimicrobial compounds (Walsh, 2003). The cell envelope of Bacillus subtilis contains a cytoplasmic membrane surrounded by layers of cross-linked peptidoglycan (PG), membrane-associated lipoteichoic acids (LTA), and wall-associated teichoic acids (WTA) (Foster and Popham, 2002; Scheffers and Pinho, 2005). PG is a polymer of alternating N-acetylglucosamine (NAG) and N-acetylmuramic acid (NAM) glycan chains cross-linked by peptide sidechains. The newly synthesized lipid-linked NAG-NAM units are polymerized to glycan strands by the action of transglycosylase (TG). Concurrent with, or soon after this polymerization, the peptide side chains on the NAM residue are cross-linked by transpeptidase (TP). Both TG and TP are activities of high molecular weight penicillin-binding proteins (HMW PBPs), and they are the targets of moenomycin and the β-lactam antibiotics respectively (Waxman and Strominger, 1983; Foster and Popham, 2002; Macheboeuf et al., 2006).
β-lactam antibiotics are characterized by the presence of a β-lactam ring which mimics the D-Ala-D-Ala dipeptide substrate of HMW PBP and inhibits the transpeptidation reaction by covalent modification of the TG active site (Macheboeuf et al., 2006). This inhibition disrupts cell wall biosynthesis, triggers the formation of reactive oxygen species (ROS), and results in cell lysis and death (Kohanski et al., 2007; 2010; Gusarov et al., 2009). Synthesis and incorporation of new PG glycan strands into the existing cell wall require close co-ordination between the biosynthetic machinery (including HMW PBPs) and autolytic enzymes that allow the separation of already cross-linked glycan strands. When properly co-ordinated, the cell grows normally and maintains proper cell shape. Conversely, agents that prevent this co-ordination by inhibiting TG or TP activities of PBPs, or by activating autolysins, lead to lysis and cell death (Fig. 1). Several models have been advanced to explain how this co-ordination occurs, but the existence and precise architecture of the proposed biosynthetic holoenzyme is still unclear (Carballido-Lopez and Formstone, 2007; Vollmer and Bertsche, 2008).
There are three major mechanisms that confer high-level β-lactam resistance as described for the Gram-positive genera Staphylococcus and Streptococcus and the Gram-negative species Escherichia coli and Pseudomonas spp. These are: (i) expression of β-lactamase(s) that inactivate the antibiotics; (ii) expression of mutated or mosaic PBP alleles that have low affinity for β-lactams; and (iii) the expression of a β-lactam specific efflux pump (Poole, 2004; Wilke et al., 2005). B. subtilis displays a significant level of intrinsic resistance against a variety of β-lactam antibiotics, but the underlying mechanisms are poorly understood. Although there are three putative β-lactamase genes (penP, ybbE and yblX) in the genome, no β-lactamase activity is detected in the growing cells or supernatant (Colombo et al., 2004). No penicillin-insensitive PBP alleles have been identified nor does an efflux pump-based mechanism appear to be applicable to B. subtilis and other Gram-positive bacteria. Therefore, the molecular basis of this intrinsic, moderate level β-lactamase resistance is unclear. Recent results suggest that extracytoplasmic function (ECF) σ factors play a role in resistance to β-lactam antibiotics: a triple mutant (strain sigMWX) as well as a mutant lacking all seven ECF σ factors (strain Δ7ECF) is sensitive to β-lactam antibiotics including ampicillin, penicillin G, aztreonam and cefuroxime (CEF) (Mascher et al., 2007; Luo et al., 2010).
In this study, we investigated the roles of ECF σ factors in providing intrinsic resistance to β-lactam antibiotics and, in particular, to CEF. We found that σM plays a primary role in β-lactam resistance, with σX as a secondary determinant. We identified Tn7 insertions mutations that restored CEF resistance to a sigM mutant. Genetic analysis reveals a central role for the recently identified signal molecule cyclic-di-AMP (c-di-AMP), synthesized in part by a σM-activated diadenylate cyclase (DAC), in cell wall homeostasis. In addition, our results highlight the key role of previously defined pathways by which ECF σ factors regulate autolysin activity and resistance to ROS.
Results and discussion
σM is the major ECF σ factor involved in the intrinsic resistance to cefuroxime
Previously, we showed that a null mutant lacking all seven ECF σ factors (strain Δ7ECF) has higher sensitivity to numerous antibiotics (including several β-lactams) compared to the wild-type (WT) strain (Luo et al., 2010). To clarify the role of ECF σ factors in mediating the intrinsic resistance to β-lactam antibiotics, we here sought to identify both the ECF σ factor(s) and the relevant pathways responsible for resistance using CEF as a model β-lactam.
Isogenic strains carrying single or multiple mutations in genes encoding ECF σ factors were tested for CEF susceptibility using disk diffusion and minimal inhibitory concentration (MIC) assays. A sigM null mutant showed elevated sensitivity to CEF, whereas other single mutants showed little or no change (Fig. 2). The double sigM sigX mutant displayed high sensitivity equivalent to the Δ7ECF strain. A sigW mutant showed no effect, although effects on β-lactam resistance have been seen in other B. subtilis strain backgrounds (Lee et al., 2012). None of the other four ECF σ factors played a role in CEF resistance, even when a multiple mutant strain was tested (Fig. 2). We conclude that σM is the major ECF σ involved in the intrinsic resistance to CEF, with σX playing a secondary role apparent in strains lacking σM. These results suggest that the major resistance pathway(s) depend exclusively on σM for their expression, with one or more additional pathways that can be activated by either σM or σX (as revealed in the double sigM sigX mutant). As described previously, several ECF σ factor promoters can be recognized by more than one ECF σ factor (Huang et al., 1998; Qiu and Helmann, 2001; Mascher et al., 2007). As we will show later in this study, genes encoding the transcription factors Abh and Spx are recognized by both σM and σX and are involved in CEF resistance.
Antibiotic resistance pathways are often transcriptionally activated in the presence of the cognate antibiotic. ECF σ factors typically autoregulate their own expressions and we and others have previously characterized the relevant autoregulatory promoters (Helmann, 2002; Asai et al., 2003; Thackray and Moir, 2003). We therefore monitored the effect of CEF on expression from the autoregulatory promoters for sigM, sigW and sigX. In each case, a two- to threefold induction was observed (Table 1). In contrast, low (basal) activity and no induction were detected for the other four ECF σ factors (sigY, sigV, sigZ, ylaC) (data not shown). This induction profile is consistent with prior results demonstrating that σM, σX and σW are responsive to cell envelope stress and are activated by an overlapping set of inducers (Minnig et al., 2003; Mascher et al., 2007; Eiamphungporn and Helmann, 2008; Hachmann et al., 2009).
Table 1. ECF σ promoter activities induced after treatment with 8 µg ml−1 CEF for 30 min.
Activities (Miller Units) were measured using β-galactosidase assays and the means and SE are reported.
3.7 ± 0.5
10.1 ± 0.5
38.6 ± 1.3
99.7 ± 3.5
35.6 ± 1.9
71.3 ± 2.3
CEF targets PBP1, 2a, 2b and 4
The σM regulon is known to include several enzymes involved in various aspects of cell wall synthesis including one HMW PBP (PBP1, encoded by ponA) (Eiamphungporn and Helmann, 2008). In most cases, σM-dependent promoters serve to upregulate gene expression in response to stress, but are not solely responsible for expression due to the presence of other promoters. In the case of ponA, this gene can be transcribed from two promoters: one is σM-dependent, and the other is σA-dependent. We here hypothesized that one mechanism of resistance might be the σM-dependent upregulation of PBP1 or other factors involved in assembly or function of cell wall biosynthetic complexes.
To identify the targets of CEF, we performed bocillin-FL competitive labelling assays (Zhao et al., 1999; Kawai et al., 2009). Five HMW PBPs (PBP1, 2a, 2b, 2c, 4) and one low molecular weight penicillin-binding protein (LMW PBP) (PBP5) were detected by bocillin-FL labelling and CEF competed with bocillin-FL for binding to PBP1, 2b, 2c and 4 (Fig. 3). Since only six PBPs can be detected in this assay, it is possible that other PBPs are also targets for CEF. No differences in either PBP profile or relative affinity for CEF binding were apparent in a comparison of the CEF-sensitive sigMWX mutant and the WT strain using the bocillin-FL labelling assay (data not shown). This suggests that mutants lacking ECF σ factors are not altered in their CEF susceptibility due to a gross change in the levels of PBPs.
Since PBP1 is a target for CEF, we hypothesized that σM-mediated upregulation of PBP1 might contribute to β-lactam resistance. However, deletion of ponA did not alter CEF susceptibility (a ponA null mutant and WT have an identical zone of inhibition). Thus, upregulation of PBP1 by σM does not appear to be a major mechanism of CEF resistance. We next tested whether B. subtilis expresses β-lactamase using the chromogenic substrate nitrocefin (Ross et al., 2009). No β-lactamase activity could be detected (either prior to or after CEF treatment) in the WT, sigMWX or Δ7ECF strains (data not shown). Thus, the role of ECF σ factors in CEF resistance does not appear to be due to alterations in CEF targets or due to degradation by β-lactamases.
Tn7 mutagenesis reveals multiple pathways involved in CEF resistance
To gain insights into the pathways contributing to CEF resistance, we performed Tn7 transposon mutagenesis and selected for mutations that restored CEF resistance to a sigM mutant. The Tn7 transposon derivative we used harbours an outward-facing, xylose-inducible promoter which thereby allows recovery of both loss of function (gene disruption) and gain of function (xylose-dependent upregulation) mutations (Bordi et al., 2008). Insertion libraries were generated in vitro using WT genomic DNA as a target and then transformed into competent B. subtilis cells with selection for both the transposon (spcR) and CEF resistance. In an initial study, we recovered numerous insertions linked to sigM. In these strains, a functional copy of sigM had been co-transformed into the recipient cells. Although this result confirms the importance of σM in CEF resistance, it was otherwise uninformative. Therefore, all subsequent experiments used a Tn7 mutant library generated in a sigM mutant (HB10216) background.
A total of 520 CEF-resistant colonies were obtained in 10 separate experiments. DNA sequence analysis identified 25 unique insertions localized to 10 different genes (Table 2). All of the insertions increased CEF resistance in a sigM mutant, although none restored resistance to WT levels (Table 2). The most frequently observed insertion occurred in yybT, an orthologue of a gene recently renamed gdpP (see below). We therefore performed an additional round of selection, transforming the sigM Tn7 library into a sigM gdpP double mutant strain (HB10257). This selection led to the recovery of insertions in two genes (lytE and clpP). Both triple mutants (sigM gdpP lytE::Tn7 and sigM gdpP clpP::Tn7) were at least as CEF-resistant as WT (Table 2). These results indicate that gdpP likely affects a different resistance pathway than lytE and clpP. Although our selection plates contained xylose, in no case was CEF resistance dependent on xylose suggesting that in each case we have recovered gene disruption mutations that lead to increased CEF resistance.
Table 2. Tn7 insertions that can restore CEF resistance in a sigM or a sigM gdpP mutant.
The resistance to CEF was tested using disk diffusion assay with biological triplicates, and repeated twice. The zone of inhibition (mean ± SE) was used for the score. The resistance level of WT is defined as ‘+++’, and ΔsigM is ‘−’.
The sigM strain has the same growth rate as WT (100%). Strains with noticeably reduced growth rates are labelled with an asterisk.
Insertions in a sigM background
Penicillin-binding endopeptidase X
Wall teichoic acid biosynthesis
Regulate expression of SlrR
Negative regulator of Spo0A–P
Initiation of sporulation
Cytochrome aa3-600 quinol oxidase
Transfer-messenger RNA (tmRNA)
Insertions in a sigM gdpP background
ATP-dependent Clp protease proteolytic subunit
Genes identified in this suppressor mutation are involved in a variety of pathways and functions (Table 2). We categorized them into three groups using two criteria: (i) direct or indirect involvement in cell wall metabolism, and (ii) mild or strong effect on CEF resistance. The first group included several insertions that inactivated genes directly involved in cell wall metabolism including lytE, pbpX, tagA and ymdB. LytE is a major autolytic endopeptidase in vegetative cells (Margot et al., 1998; Smith et al., 2000). LytE interacts with the actin-like protein MreBH along the cylindrical part of cell wall and with FtsZ and PBP2b at the division septum. It is, therefore, closely related to cell wall synthesis (Carballido-Lopez et al., 2006). The inactivation of lytE presumably increases β-lactam resistance by delaying cell lysis. PbpX is a LMW PBP that is located at the septum during vegetative growth (Scheffers et al., 2004). Its function is unknown, although it was shown previously to be activated by σX (Cao and Helmann, 2004). YmdB was recently reported to regulate the expression and/or activity of a transcriptional regulator SlrR, which in turn affects the activity of both σD and the regulator of biofilm formation, SinR, and likely indirectly modulates autolysin activity (Diethmaier et al., 2011). Finally, TagA is a key enzyme in the synthesis of teichoic acids, a major component of the cell wall (Mauel et al., 1991; D'Elia et al., 2009). The second and third groups of insertions are not directly linked to cell wall homeostasis. The second group, including kinD, spo0A, qoxAB and ssrA insertions, had relatively mild effects on CEF resistance. Further studies are needed to define the mechanisms of these effects, but in several cases the mutant strains grew more slowly than WT strain under our experimental conditions and this may contribute to their increased β-lactam resistance (Table 2).
Here, we focus on the third group of mutations (gdpP, rsiX and clpP) for further analysis since they resulted in strong CEF resistance and have been linked to σM and its regulon members. We recovered 11 independent insertions within the 1980 bp coding sequence of gdpP (formerly yybT). GdpP is a transmembrane protein containing three functional domains: a haem-binding PAS domain, a degenerate GGDEF domain and a DHH/DHHA1 phosphodiesterase (PDE) domain (Rao et al., 2010; 2011). The Staphylococcus aureus orthologue has recently been renamed GdpP to indicate that it is a GGDEF domain protein containing PDE (Corrigan et al., 2011) and we therefore adopt this same designation for B. subtilis. RsiX is the anti-σ factor cognate for σX. We hypothesized that the rsiX::Tn7 insertion increased β-lactam resistance by upregulation of σX. Tn7 insertions in clpP led to the highest level of CEF resistance observed in this study (Table 2). ClpP is a component of the Clp protease. In B. subtilis, the ClpP proteolytic core can pair with any of the three Clp ATPases (ClpX, ClpC and ClpE) and form a large hetero-oligomeric Clp protease. Clp protease recognizes and degrades a wide range of proteins, including non-native proteins and stress response regulators, and it is therefore involved in multiple cell development and stress response pathways (Frees et al., 2007). Here, we present evidence that these three insertion mutations affect three interrelated pathways for CEF resistance (Fig. 1).
The role of σX in CEF resistance is in part through regulation of abh and spx
We hypothesized that the Tn7 insertion in rsiX restored CEF resistance by upregulation of σX which, as noted above, plays a secondary role in CEF resistance that becomes important in the absence of σM (Fig. 1). As predicted, epistasis experiments indicated that σX is downstream of RsiX: a sigM sigX rsiX strain was as sensitive to CEF as the sigM sigX strain (Fig. 4A).
Since the effect of sigX on CEF resistance is greatly enhanced in a sigM mutant background (Figs 2 and 4A), we hypothesized that the relevant genes involved in CEF resistance can be activated by either σM or σX. The regulons of σM and σX have been characterized, and several target promoters have been defined that are activated by both ECF σ factors (Cao and Helmann, 2004; Eiamphungporn and Helmann, 2008). We chose six such target operons (abh, spx, dltABCDE, lytR, yceCDEF and bcrC) for further analysis. In a WT background, only the spx null mutant showed increased CEF susceptibility. When introduced into the sigM null mutant, the abh and spx mutations both increased CEF sensitivity (Fig. 4B). The abh and spx CEF-sensitive phenotypes in both sigM and sigMX background can be complemented using IPTG-inducible abh or spx alleles respectively (Figs S1 and S2). These results suggest that spx and abh can account for at least part of the role of σX in CEF resistance. We also defined the MIC of single and multiple mutant strains of sigM, abh and spx. Although their differences in CEF susceptibility are readily detected in the disk diffusion assay (Fig. 4A and B), mutant strains of sigMX, sigM abh, sigM abh spx have the same MIC of 0.03 µg ml−1 when measured in liquid medium (Table S3). We therefore focus here on the differences observed on solid medium.
Abh is a paralogue of AbrB and together these two transition state regulators regulate biofilm formation, autolysin activity and antibiotic production and resistance (Strauch et al., 2007; Luo and Helmann, 2009; Murray et al., 2009; Murray and Stanley-Wall, 2010). The transcription of abh is dependent on σX and σM, with σX being the major regulator (Huang and Helmann, 1998; Luo and Helmann, 2009). Recently, an abh mutant was shown to be sensitive to β-lactam antibiotics ampicillin, carbenicillin and cephalexin (Murray and Stanley-Wall, 2010). Resistance to ampicillin was restored by inducing the expression of the transcriptional regulator slrR, or by inactivating genes encoding major autolysins (lytC and lytF encoding amidase and DL-endopeptidase respectively) (Murray and Stanley-Wall, 2010). These results support a model (Fig. 1) in which Abh indirectly activates the expression of SlrR (Murray et al., 2009). SlrR forms a heteromeric complex with SinR which represses both the lytABC and lytF operons (Chai et al., 2010a). Thus, σX and σM play partially redundant roles in β-lactam resistance by activating Abh, which in turn activates SlrR to enable repression of autolytic enzymes.
We used a genetic approach to explore the role of ClpP and Spx in β-lactam resistance. As noted above (Table 2), a clpP::Tn7 mutation greatly increased CEF resistance in the sigM gdpP mutant strain (HB10264). The clpP null mutation also increased CEF resistance in WT and null mutant strains of sigM and both sigM and sigX (Fig. 4C). Spx is a ClpXP substrate (Nakano et al., 2002). The spx mutation masked the effect of clpP in the WT, sigM and sigM sigX strain backgrounds (Fig. 4C). These epistasis results imply that spx is downstream of clpP in the CEF resistance pathway and is the major ClpP substrate that plays a role in β-lactam resistance. Thus, we predict that the major impact of the clpP mutation is to enhance accumulation of Spx in the cell. To test this idea, an IPTG-inducible copy of spx or spxDD (a Clp protease insensitive variant; Nakano et al., 2003) was introduced in the sigM and sigM sigX mutant strains. An increase in CEF resistance was observed when either spx or spxDD was induced (although the effect was much more dramatic with the protease-insensitive allele), suggesting that the accumulation of Spx can increase resistance to CEF in B. subtilis (Fig. S2). In addition, we performed disk diffusion assays with strains lacking either clpX or clpC (Fig. S3). Deletion of clpX can strongly increase CEF resistance in both strain backgrounds of WT and sigM mutant, while deletion of clpC only showed minor effect. This result is consistent with the major role of ClpP in CEF resistance being the ClpXP-dependent degradation of Spx.
We also note that the effect of the clpP mutation may not be limited to enhancing accumulation of Spx, since mutation of clpP also led to a small increase in CEF resistance in an spx mutant background. This effect was most notable in strains mutant for sigM or sigM and sigX (Fig. 4C). A small increase in CEF resistance was also found with a clpC mutant (Fig. S3). Therefore, we suggest that there are other ClpP protease substrates that also contribute, albeit modestly, to CEF resistance. One candidate is SlrR which, as noted above, has been implicated in the downregulation of autolysins and is subjected to degradation by ClpCP (Chai et al., 2010b) (Fig. 1). A second candidate and a ClpCP-degraded substrate is MurAA. MurAA is a UDP-N-acetylglucosamine 1-carboxyvinyltransferase, which catalyses the first committed step in PG biosynthesis (Kock et al., 2004).
c-di-AMP as an emerging second messenger found in Bacteria
The most frequent insertions recovered in our selection (Table 2) were in gdpP and inactivate a PDE known to degrade c-di-AMP, an emerging second messenger found in Bacteria and likely in Archaea (Romling, 2008). c-di-AMP was discovered as a metabolite bound in the crystal structure of DisA which catalyses its synthesis from ATP (Witte et al., 2008). DisA was initially characterized as a DNA integrity scanning protein that signals the integrity of the DNA and thereby enables sporulation to proceed (Bejerano-Sagie et al., 2006). This led to a model in which the DisA diadenylate cyclase (DAC; DUF147 domain) signals chromosome integrity: DAC activity can be strongly inhibited by binding of DisA to branched chain nucleic acid structures that might form as recombination intermediates.
DisA is the only confirmed diadenylate cyclase (DAC) in B. subtilis (Witte et al., 2008; Oppenheimer-Shaanan et al., 2011). However, B. subtilis encodes two additional candidate DAC proteins (containing DUF147 domains): YbbP and YojJ (Romling, 2008). The DisA DAC domain is linked to a helix–hairpin–helix non-specific DNA-binding domain which allows DAC activity to be regulated by DNA integrity. In contrast, YbbP is predicted to be membrane-localized and YojJ cytosolic, but little is known of how their activities might be regulated. Of relevance to the present study, transcription of disA is regulated by both σA and σM (Eiamphungporn and Helmann, 2008).
The level of c-di-AMP in the cell is controlled by both its rate of synthesis by DAC and its degradation by a c-di-AMP-specific PDE (Fig. 1). B. subtilis GdpP (formerly YybT) is a c-di-AMP PDE in vitro (Rao et al., 2010; 2011) and in vivo (Oppenheimer-Shaanan et al., 2011). In vegetatively growing B. subtilis, 1.7 µM c-di-AMP was measured which increased, in a DisA-dependent manner, to near 5 µM early during sporulation. A gdpP deletion strain of B. subtilis was shown to have an over fourfold increase in c-di-AMP levels in early sporulating cells (Oppenheimer-Shaanan et al., 2011). Similarly, a ∼15-fold increase was observed with a gdpP mutation in S. aureus (from 2.8 µM to 42.9 µM). In S. aureus, elevated levels of c-di-AMP suppress the growth defects associated with an inability to synthesize LTA and alter both autolysin expression and the level of PG cross-linking (Corrigan et al., 2011).
In B. subtilis, the synthesis and degradation of c-di-AMP is correlated with β-lactam resistance
GdpP is a transmembrane protein with three functional domains: a haem-binding PAS domain, a degenerate GGDEF domain and a DHH/DHHA1 PDE domain (Rao et al., 2010; 2011). In accordance with the emerging model of c-di-AMP as a signal molecule, we hypothesized that it was the loss of GdpP PDE activity that conferred CEF resistance. We therefore complemented the sigM gdpP strain with an IPTG-inducible GdpP, a truncated GdpP lacking the DHH/DHHA1 domain (GdpP1–303), or a mutated GdpP (GdpPD420A) carrying a single amino acid substitution which abolishes PDE activity (Rao et al., 2010). Induction of WT GdpP conferred an extreme CEF sensitivity (Fig. 5). In contrast, neither of the mutant GdpP proteins increased sensitivity to CEF (Fig. 5), suggesting that it is the PDE activity that affects CEF sensitivity.
One consequence of antibiotic stress is the activation of σM which leads to elevated expression of the DisA DAC. We therefore hypothesized that a sigM null mutant might have decreased c-di-AMP levels that could be compensated by mutation of GdpP, the c-di-AMP degrading PDE. Indeed, a disA deletion mutant displayed a small but reproducible increase in sensitivity to CEF, with an MIC of 3 µg ml−1 compared to 4 µg ml−1 for WT (Table S3). This is consistent with the recent report that DisA accounts for perhaps 50% of the c-di-AMP present in cells as monitored early in sporulation (Oppenheimer-Shaanan et al., 2011). As expected, the induction of GdpP in the disA mutant led to a large increase in CEF susceptibility (Fig. 6), consistent with the notion that even disA cells contain substantial c-di-AMP that contributes to CEF resistance. This suggests that B. subtilis contains at least one additional DAC, presumably encoded by either or both the DAC domain-containing proteins YbbP and YojJ.
c-di-AMP is essential for cell growth
To gain insights into the relative contributions of disA, ybbP and yojJ to c-di-AMP synthesis, we mutated each of these loci individually and in combination. Deletion of ybbP resulted in the highest CEF sensitivity (as seen in the uninduced sample in Fig. 6, and MIC of 1 µg ml−1, Table S3). Deletion of yojJ, however, had no effect. Induction of GdpP increased CEF sensitivity in all three DAC mutant backgrounds (Fig. 6). We conclude that YbbP is the major DAC contributing to intrinsic β-lactam resistance in growing cells, and that both synthesis and degradation of c-di-AMP affects CEF resistance. This result is consistent with the recent suggestion that DisA functions primarily in early sporulation, with a comparatively minor contribution in (unstressed) vegetative phase cells (Oppenheimer-Shaanan et al., 2011). It is interesting to note that YbbP and GdpP are both membrane-localized, although the signals that might control their synthesis and activity are unknown.
The expression of YbbP is poorly characterized, but it is noteworthy that it is encoded immediately downstream of the sigW-rsiW operon and it may be, in part, activated by σW. However, σW has no effect in CEF resistance in our B. subtilis WT strain background (Fig. 2). We therefore asked whether σM or σX have a role in regulating ybbP. Multiple null mutants of sigM, sigX and ybbP were constructed and tested for their susceptibilities to CEF. The mutation in ybbP is clearly additive to both sigM and sigX mutations (Fig. S4). In addition, the transcriptional start site of ybbP was mapped to 72 bp upstream of its start codon using 5′RACE. A σA promoter is present upstream of the assigned start site (TTCACTtgctaaatcgaaatgtggTATAATgggctcG; upper case letters indicate the −35, −10, +1 regions respectively). Together, these results suggest that ybbP is not part of the σM or σX regulatory pathways.
We next sought to construct double and triple null mutants of disA, ybbP and yojJ. A disA ybbP double mutant strain could not be obtained, suggesting that this combination of mutations is lethal, whereas double mutants of disA yojJ and ybbP yojJ were viable. We conclude that c-di-AMP is essential for viability and that the basal level of expression of either DisA or YbbP is sufficient to support growth. An essential role for DAC proteins has also been suggested in Listeria monocytogenes since it was impossible to disrupt the single DAC-encoding gene in this organism (Woodward et al., 2010). Similarly, DAC genes were identified in screens for essential genes in Mycoplasma spp., Streptococcus pneumoniae and S. aureus (Song et al., 2005; Glass et al., 2006; French et al., 2008; Chaudhuri et al., 2009).
To determine whether all three DAC proteins (DisA, YbbP, YojJ) are active and could support growth, we integrated an IPTG-inducible copy of each gene into a ybbP null mutant and then attempted to introduce a disA null mutation by chromosomal transformation. Indeed, a disA ybbP double mutant could be obtained when any one of the three genes (disA, ybbP or yoj) was induced (Fig. 7A). This strategy also allowed construction of IPTG-dependent disA ybbP yojJ triple mutant strains in which growth could be supported by any one of three DAC-encoding genes. We note that the Pspac(hy) promoter used in this work is slightly leaky and, as a result, the disA ybbP Pspac(hy)–disA strain was able to grow even in the absence of IPTG. However, the disA ybbP Pspac(hy)–yojJ strain grew slowly and the disA ybbP Pspac(hy)–ybbP was unable to grow unless at least 50 µM IPTG was present (data not shown). These results suggest that all three of these putative DAC proteins are biologically active and able to support growth when expressed.
The essential role of c-di-AMP is linked to PG homeostasis
Since a reduced level of c-di-AMP is linked to high CEF sensitivity, we tested whether c-di-AMP is involved in cell wall homeostasis. Depletion of c-di-AMP in strain disA ybbP Pspac(hy)–ybbP by growth in the absence of inducer IPTG led to cell lysis as monitored both by following optical density (Fig. 7B) and by light microscopy (Fig. S5). The lysis phenotype can be suppressed either by the presence of IPTG (inducing the expression of ybbP) or by supplementation of the growth medium with SMM (sucrose, MgSO4 and maleic acid), sucrose or MgSO4. SMM has been used previously to stabilize protoplasts and support the growth of cell wall-free l-form cells (Chang and Cohen, 1979; Leaver et al., 2009). Similarly, sucrose likely functions as an osmotic protectant, and Mg2+ has been shown to restore growth and WT morphology of many PG defective mutants including single mutants of ponA, rodA, mreB, mreC, mreD, mbl and a double mutant of pbpAH (Murray et al., 1998; Formstone and Errington, 2005; Leaver and Errington, 2005; Kawai et al., 2009; 2011; Schirner and Errington, 2009). This is reminiscent of recent results from Corrigan et al. (2011) who showed that osmotic protectants support the growth of a LTA-deficient mutant of S. aureus and that this requirement can be bypassed by a gdpP mutation. The S. aureus gdpP mutant displayed an increase in both c-di-AMP and PG cross-linking. Collectively, these results suggest that c-di-AMP plays an essential role in PG homeostasis (Fig. 1).
σM and c-di-AMP are involved in resistance to other cell wall antibiotics
We next tested whether c-di-AMP is involved in resistance to other antibiotics. Induction of GdpP in strain sigM gdpP Pspac(hy)–gdpP leads to high sensitivity to aztreonam, cefixime and moenomycin in addition to CEF as monitored using disk diffusion assays (Fig. 8). Cefixime is a third-generation cephalosporin, aztreonam is a monobactam, and moenomycin is a glycolipid. As β-lactams, cefixime and aztreonam target PBP TPs. Moenomycin, on the other hand, targets the TG activity of HMW PBPs (Lovering et al., 2007). Although aztreonam is generally found to have poor activity against Gram-positive bacteria (Georgopapadakou et al., 1982; Guay and Koskoletos, 1985), we observed using bocillin-FL labelling that aztreonam can derivatize PBP 1, 2c and 4 in B. subtilis (Fig. 3). As also noted for CEF, mutation of sigM converts B. subtilis from an aztreonam non-susceptible to a susceptible strain, and this susceptibility is modulated by gdpP (Fig. 8). Thus, the function of c-di-AMP is not limited to CEF resistance, as would be expected if it functions to support balanced cell wall synthesis.
A model for the role of ECF σ factors in β-lactam resistance
The genetic analyses presented herein lead to an integrated model in which the ECF σ factors σM and σX contribute to β-lactam resistance by the antibiotic-inducible activation of regulatory proteins that affect three distinct pathways (Fig. 1). B. subtilis PG is a dynamic structure, which is continuously synthesized, modified and hydrolysed. It is notable that σM-activated promoters have been previously mapped preceding several genes involved in PG synthesis (including mreB, bcrC, divIB, divIC, ddl, murB, murF, rodA, pbpX and ponA), one of the four paralogous LTA synthases (yfnI), and cell wall modification enzymes (dltABCDE) (Eiamphungporn and Helmann, 2008). Thus, σM appears to function to positively regulate cell wall assembly and structure in response to antibiotic stress. β-lactam antibiotics inhibit the TP activity of PBPs and thereby inhibit glycan strand cross-linking. This inhibition disrupts the balance between PG synthesis and hydrolysis and endogenous autolysins trigger cell lysis. In addition, β-lactams trigger ROS formation and cell death. Both autolysin-dependent and independent mechanisms contribute to the bactericidal effect (Kohanski et al., 2007; Dubee et al., 2011).
Extracytoplasmic function σ factors counteract the effects of β-lactams by activating at least three distinct pathways (Fig. 1). First, σM contributes to the expression of one of three c-di-AMP synthases (DisA). The cellular level of c-di-AMP is regulated by both DAC synthases (DisA, YbbP and YojJ) and the cognate PDE (GdpP). At least one DAC is required for cell growth, indicating an essential role of c-di-AMP. The cell lysis phenotype of our DAC depletion strain together with the recent report from Corrigan et al. (2011) suggest a positive link between c-di-AMP and PG cross-linking. However, the role of c-di-AMP may be not limited to cross-linking, since c-di-AMP also modulates susceptibility to moenomycin, which targets the TG domain of PBP and thereby inhibits the polymerization of the PG glycan strands.
Second, ECF σ factors affect the expression and regulation of autolysins. Both σM and σX activate the transcription of abh, whose product indirectly activates the expression of SlrR, which directly represses expression of LytC and LytF (Luo and Helmann, 2009; Chai et al., 2010a; Murray and Stanley-Wall, 2010). Another autolytic endopeptidase (LytE) was identified by Tn7 mutagenesis as a contributor to β-lactam susceptibility. These findings support the notion that preventing autolysis can increase β-lactam resistance.
Third, our analysis of the β-lactam resistance phenotype of a clpP null mutant identified Spx, a regulator of pathways that protect the cell against ROS (Zuber, 2009), as a contributor to β-lactam resistance. The clpP mutant strain may also have elevated levels of SlrR, a known inhibitor of autolysin expression (Chai et al., 2010b). Although the model we have developed here (Fig. 1) is already quite complex, it certainly underestimates the true complexity of the adaptive responses mediated by ECF σ factors and other regulators that conspire to protect cells against antibiotics and other chemical insults.
Bacterial strains and growth conditions
Bacillus subtilis strains used are derivatives of strain168 (trpC2) and are shown in Tables 3 and S2. E. coli strain DH5α was used for standard cloning procedures. Bacteria were grown in Luria–Bertani (LB) (10 g tryptone, 5 g yeast extract and 5 g NaCl per litre) broth at 37°C with vigorous shaking. Antibiotics were added to the growth medium when appropriate: 100 µg ml−1 ampicillin for E. coli, and 1 µg ml−1 erythromycin plus 25 µg ml−1 of lincomycin (MLS, macrolide-lincomycin-streptogramin B resistance), 10 µg ml−1 chloramphenicol, 100 µg ml−1 spectinomycin (Spc), 5 µg ml−1 tetracycline and 10 µg ml−1 kanamycin for B. subtilis. OD600 readings were taken on a Spectronic 21 spectrophotometer.
Gene deletions were generated by replacing the coding region with an antibiotic resistance cassette using long flanking homology PCR (LFH-PCR) followed by DNA transformation as previously described (Mascher et al., 2003). Chromosomal DNA transformations were performed as described previously (Harwood and Cutting, 1990).
The IPTG-inducible constructs were generated using vector pPL82 (Quisel et al., 2001). PCR products were amplified from B. subtilis 168 chromosomal DNA, digested with endonucleases, and cloned into pPL82. pPL82 contains a chloramphenicol resistance cassette, a multiple cloning site downstream of the Pspac(hy) promoter, and the lacI gene between the two arms of the amyE gene. Primer pairs used for PCR amplification are 5249/5250 for disA, 5252/5253 for ybbP, 5255/5256 for yojJ, 5244/5245 for gdpP and 5244/5258 for gdpP1–303. All oligonucleotide sequences are listed in Table S1. The sequences of the inserts were verified by DNA sequencing (Cornell DNA sequencing facility). pPL82–gdpPD420A was generated using overlap joining PCR with pPL82–gdpP as DNA template. Primer pairs 5244/5293 and 5294/5245 were first used to amplify the up and down fragments of gdp respectively. The gdpPD420A mutation was generated using primers 5293 and 5294. A joining PCR was then performed with the up and down fragments as template and primer pairs 5244/5245. The PCR product was cloned into pPL82 as above, and the insert was verified by DNA sequencing. Plasmids were linearized by ScaI and used to transform B. subtilis, where they integrated into the amyE locus.
Antibiotic susceptibility tests
Susceptibility tests for antibiotics were conducted using disk diffusion assay and MIC test. Mueller Hinton (MH, Sigma-Aldrich) medium was used for both assays. Disk diffusion assays were performed as previously described (Luo et al., 2010). The bottom agar is 15 ml MH broth supplemented with1.5% agar, and the top agar is 4 ml MH broth supplemented with 0.75% agar. We used BBL™ Sensi-Disc™ Susceptibility Test Discs (BD; cefixime 5 µg, cefoxitin 30 µg, ceftriaxone 30 µg, ceftazidime 30 µg, cefoperazone 75 µg, amoxicillin 30 µg, ampicillin 10 µg, piperacillin 100 µg, oxacillin 1 µg, piperacillin 100 µg, imipenem 10 µg, meropenem 10 µg and Isoniazid 1 µg) and also prepared disks using Whatman filter paper disks (7 mm in diameter) and freshly made stocks of antibiotics (aztreonam 30 µg, CEF 6 µg, penicillin G 10 U, nalidixic acid 30 µg, novobiocin 250 µg, vancomycin 30 µg, polymycin B 250 µg and moenomycin 50 µg). The zone of growth inhibition was measured after overnight growth at 37°C. For MIC test, fresh single colonies were first grown in MH broth to an OD600 of 0.4, and diluted 1:100 in MH broth, and 200 µl of the diluted culture was dispensed in Bioscreen 100-well microtitre plate. Growth was measured spectrophotometrically (OD600) using a Bioscreen incubator (Growth Curves USA, Piscataway, NJ) at 37°C with vigorous shaking. The absorbance was recorded every 30 min for 24 h. Inhibition was defined as a final OD600 < 0.2 at the 12 h time point (after 12 h, suppressor mutants started to grow up). All antibiotics susceptibility tests were performed with biological triplicates and repeated at least twice.
Bocillin-FL competitive labelling assay
The bocillin-FL labelling assay was performed as previously described (Zhao et al., 1999; Kawai et al., 2009) with modifications. Overnight cultures of B. subtilis cells in LB were diluted 1:100 into 5 ml fresh LB broth, and incubated at 37°C with vigorous shaking. When cell cultures reached mid-log phase (OD600 0.4), the cultures were treated with either 0.05 µg ml−1 (final conc.) of bocillin-FL, or with additional challenge of 0.00625 µg ml−1 (final conc.) of CEF, or an additional 5 µg ml−1 aztreonam (final conc.) for 10 min. The cells were pelleted by centrifugation and kept at −20°C overnight. The pellet was thawed on ice and resuspended in 50 µl of 0.85% NaCl. The cell resuspension was boiled for 5 min with SDS loading buffer, and proteins were separated by 4–12% SDS-PAGE. To visualize the labelled PBPs, the gels were scanned with a Molecular Dynamics Typhoon PhosporImager (excitation at 488 nm and emission at 530 nm), and the images were analysed using ImageQuant TL (Amersham Biosciences).
The Tn7 mutagenesis libraries were generated with chromosomal DNA using in vitro transposition as described (Bordi et al., 2008). The library DNA was transformed into WT B. subtilis or a sigM mutant strain (HB10216), and the resulting transposants were grown in the presence of 100 µg ml−1 spectinomycin (Spc) with and without xylose (final concentration of 1%). Chromosomal DNA was prepared from these cultures using phenol-chloroform extraction (Sambrook and Russell, 2001) and considered an amplified Tn7 library. The amplified Tn7 library DNA was transformed into the sigM mutant strains (HB10016 or HB10216), and cells were plated on LB agar supplemented with 100 µg ml−1 Spc, 1% xylose and 2 µg ml−1 CEF (32× MIC of the sigM strain). Resulting transformants were streaked onto the same selection plate twice. In order to confirm that the increased CEF resistance was due to the presence of the transposon, we performed linkage tests by transforming the chromosomal DNA of the Tn7 mutants into the sigM mutant again and selecting with 100 µg ml−1 Spc. The resulting transformants (20 colonies for each strain) were then streaked on LB agar supplemented either with 100 µg ml−1 Spc or with 100 µg ml−1 Spc plus 2 µg ml−1 CEF. The transformants that can grow on both plates were counted as linked mutants, and strains with 100% linkage were subjected to Tn7 insertion position mapping using arbitrary PCR as previously described (Bordi et al., 2008). The dependence on xylose was tested by streaking cells on LB agar supplemented with 2 µg ml−1 CEF or with 2 µg ml−1 CEF plus 1% xylose. Tn7 mutagenesis with strain sigM gdpP (HB10257) was performed as described above, except that 4 µg ml−1 of CEF (MIC of the WT strain, and 64× MIC of the sigM strain) was used for selection.
β-Galactosidase activity test
Strains harbouring ECF σ promoter–lacZ fusions were grown overnight in LB broth containing appropriate antibiotics and diluted 1:100 into 5 ml LB medium. The culture was grown at 37°C with vigorous shaking to OD600∼ 0.4 (mid-log growth phase), and then split into two aliquots. One was challenged with 8 µg ml−1 of CEF and the other was untreated. The cultures were returned to 37°C, and samples were collected after 30 min. β-Galactosidase assays were performed as described by Miller (26), and each strain was tested in biological triplicates and repeated three times. Data were reported as the mean and SE.
The transcriptional start site of ybbP was determined using 5′ rapid amplification of cDNA ends (5′-RACE). Five micrograms of total RNA from a mid-log-phase LB culture was reversed transcribed to cDNA using TaqMan reverse transcription reagents (Roche) and oligo ybbP-rev-GSP3 (5584) as primer. The 3′ end of cDNA was tailed with poly-dCTP using terminal deoxynucleotidyl transferase (New England Biolabs). The tailed cDNAs were then amplified by PCR with primers AAP (3314) and ybbP-rev-GSP4 (5585). The PCR products were subjected to DNA sequencing (Cornell DNA sequencing facility).
Growth rate test
Fresh single colonies were first grown in MH broth to OD600 of 0.4, and diluted 1:100 in MH broth, and inoculated in Bioscreen microtitre plates with a total inoculum of 200 µl. Growth was measured spectrophotometrically (OD600) using a Bioscreen incubator (Growth Curves USA, Piscataway, NJ) at 37°C with vigorous shaking. The specific growth rate of each strain was calculated from the exponential growth phase. Each test was performed with biological triplicates and repeated twice.
Depletion of c-di-AMP and microscopic imaging
Strain HB10359 was grown in MH broth supplemented with 1 mM IPTG to mid-exponential phase, and collected by centrifugation. The cells were washed twice with MH medium, and resuspended to OD600 of 0.2 in fresh MH broth, or MH broth supplemented with 1 mM IPTG, SMM (20 mM MgCl2, 10% sucrose, 20 mM maleic acid, pH 7.0), 10% sucrose or 10 mM MgSO4. Two hundred microlitres of each cell resuspension was added a Bioscreen microtitre plate, and incubated at 37°C with vigorous shaking. For phase contrast and fluorescence microscopy, 1 µg ml−1 (final concentration) of cell membrane stain FM 4–64 (Invitrogen) was added to the cell culture, and incubated at 37°C for 30 min with shaking. Five microlitres of cells were then mounted on microscope slide coated with a thin film of 1% agarose as previously described in Glaser et al. (1997). Microscopy was performed using an Olympus BX61 epifluorescence microscope. Images were acquired using Cooke SensiCam and Slidebook software (Intelligent Imaging).
We thank Dr Peter Zuber (Oregon Health & Science University) for strains ORB4342 and ORB4342, and Dr David Popham (Virginia Tech) for stain PS2062, Dr Win Chai (Harvard University) for strains RL2173 and RL2774, Dr Esther Angert (Cornell University) for assistance with microscopy, and Marilyn Wang (Cornell University) for performing antibiotic screening. This work was supported by grant GM-047446.