Bacillus subtilis σ^V Confers Lysozyme Resistance by Activation of Two Cell Wall Modification Pathways, Peptidoglycan O-Acetylation and d-Alanylation of Teichoic Acids

Abstract

The seven extracytoplasmic function (ECF) sigma (σ) factors of Bacillus subtilis are broadly implicated in resistance to antibiotics and other cell envelope stressors mediated, in part, by regulation of cell envelope synthesis and modification enzymes. We here define the regulon of σ^V as including at least 20 operons, many of which are also regulated by σ^M, σ^X, or σ^W. The σ^V regulon is strongly and specifically induced by lysozyme, and this induction is key to the intrinsic resistance of B. subtilis to lysozyme. Strains with null mutations in either sigV or all seven ECF σ factor genes (Δ7ECF) have essentially equal increases in sensitivity to lysozyme. Induction of σ^V in the Δ7ECF background restores lysozyme resistance, whereas induction of σ^M, σ^X, or σ^W does not. Lysozyme resistance results from the ability of σ^V to activate the transcription of two operons: the autoregulated sigV-rsiV-oatA-yrhK operon and dltABCDE. Genetic analyses reveal that oatA and dlt are largely redundant with respect to lysozyme sensitivity: single mutants are not affected in lysozyme sensitivity, whereas an oatA dltA double mutant is as sensitive as a sigV null strain. Moreover, the sigV oatA dltA triple mutant is no more sensitive than the oatA dltA double mutant, indicating that there are no other σ^V-dependent genes necessary for lysozyme resistance. Thus, we suggest that σ^V confers lysozyme resistance by the activation of two cell wall modification pathways: O-acetylation of peptidoglycan catalyzed by OatA and d-alanylation of teichoic acids by DltABCDE.

INTRODUCTION

Bacillus subtilis provides an important model system for the investigation of antibiotic resistance mechanisms in Gram-positive bacteria. As a soil-dwelling bacterium, B. subtilis inhabits a highly variable and competitive environment and, as a consequence, has evolved an arsenal of protective stress responses. Soil bacteria include many of the most prolific producers of antibiotics, including members of both the phylum Firmicutes (including Bacillus spp.) and, most notably, the class Actinobacteria. Antibiotics frequently target the bacterial cell envelope, including both the peptidoglycan cell wall and the cell membrane. In response to low levels of antibiotics and other cell envelope-active compounds, B. subtilis induces complex and multifaceted cell envelope stress responses (38).

Regulation of cell envelope stress responses in B. subtilis frequently involves one or more of seven extracytoplasmic function (ECF) sigma (σ) factors (σ^M, σ^W, σ^V, σ^X, σ^Y, σ^Z, σ^YlaC). The three most active in nonstressed cells, and the best characterized, are σ^M, σ^W, and σ^X (33). σ^M regulates a large set of genes that encode functions essential for cell division and envelope synthesis, and its expression is induced by cell envelope-active antibiotics, acid, heat, ethanol, and superoxide stresses (25, 37). The σ^W regulon includes at least 60 genes that inactivate, sequester, or eliminate toxic compounds from the cell, and its expression is induced by a variety of cell envelope-active compounds, detergents, and alkali stress (14, 17, 32, 39, 65). The σ^X regulon includes genes which serve to alter cell surface properties to provide protection against antimicrobial peptides (15) and is also induced by antibiotics that inhibit cell wall synthesis (20).

The ECF σ factors of B. subtilis, like those of other bacteria, are regulated at multiple levels (59). In general, each σ factor is cotranscribed with an adjacent gene encoding an anti-σ factor, which is usually a membrane protein that sequesters its cognate σ factor to the cytoplasmic membrane (66). In response to an inducing signal, the anti-σ factor is inactivated, often by proteolytic degradation (11, 31). The released σ factor then binds core RNA polymerase (RNAP) and directs the activation of specific promoter sites. In most cases, but not all, expression of ECF σ factors is positively autoregulated. To date, studies of B. subtilis suggest that each ECF σ factor (with the exception of σ^Z) activates its own expression (2) but does not activate the expression of other ECF σ factors (33). In some cases, expression is also directed by an additional σ^A-dependent promoter. In contrast, in Mycobacterium tuberculosis, activation of one ECF σ factor can induce the expression of another, leading to a transcriptional cascade (55). The potential for transcriptional cascades, in which activation of an ECF σ factor induces the expression of another transcription factor (or even another ECF σ factor), complicates efforts to define those targets that are transcribed directly as a result of σ factor reprogramming of RNAP.

Previous studies have revealed significant overlap in the regulons controlled by σ^M, σ^W, and σ^X, and as a result, the stimulons induced by various cell envelope stresses often overlap extensively (38, 46). Deciphering of the stimulons induced by cell envelope-active compounds is complex due to both the induction of multiple stress-responsive regulators by a single stimulus and substantial overlap between the target genes activated by each ECF σ factor. Regulon overlap in B. subtilis results largely from the fact that ECF σ factors recognize similar promoter sequences that share a highly conserved AAC motif in the −35 region and a CGT motif in the −10 region but may differ in other discriminatory positions (33). In some cases, promoters are exclusively activated by only one ECF σ factor, whereas in other cases two or more ECF σ factors can activate a single target promoter (16, 44). As a result of this regulon overlap, some phenotypes are evident only when two or more of the ECF σ factors are deleted (43, 45).

In contrast to the roles of σ^M, σ^W, and σ^X, those of the other four ECF σ factors (σ^V, σ^Y, σ^Z, and σ^YlaC) are still poorly understood. An initial study of σ^Y showed that this σ factor controls a small regulon and likely controls the expression of a toxic bacteriocin and its cognate immunity gene (19). The regulons and functions of σ^Z and σ^YlaC have not been well defined. Two previous studies have sought to define the set of genes regulated by σ^V (2, 67). However, the prolonged incubation after the induction of σ^V, the potential for cross-regulation as noted above, and the lack of a specific natural inducing signal have prevented clear insights into the unique physiological role(s) of σ^V.

Here we show that the ECF σ factor σ^V plays a major role in resistance to lysozyme. The σ^V regulon is strongly and specifically induced by lysozyme and includes ∼20 operons. Two of the σ^V-regulated operons play major roles in lysozyme resistance: the dlt operon and oatA, which is transcribed as part as the sigV operon. We conclude that lysozyme resistance in B. subtilis is largely mediated by the activation of two cell wall modification pathways: OatA-dependent peptidoglycan O-acetylation and d-alanylation of teichoic acids by DltABCDE.

MATERIALS AND METHODS

Strain construction and growth conditions.

All of the B. subtilis strains used in this study were constructed in the 168 background (Table 1). Unless otherwise stated, bacteria were grown in liquid Luria-Bertani (LB) medium at 37°C with vigorous shaking or on solid LB medium containing 1.5% Bacto agar (Difco). All cloning was done in Escherichia coli DH5α using ampicillin (100 μg/ml) for selection. Chromosomal DNA and plasmid DNA transformations were performed as previously reported (30). The antibiotics used for selection were spectinomycin (100 μg/ml) and macrolide-lincosamide-streptogramin B (contains 1 μg/ml erythromycin and 25 μg/ml lincomycin).

Table 1.

B. subtilis strains and plasmids used in this study

Strain or plasmid	Genotype or description	Construction or reference
Strains
168	trpC2	Lab strain
BSU2007	168 ΔsigMWXYVZ ΔylaC (Δ7ECF)	1
HB12010	BSU2007 amyE::PxylA-sigV	pVG001 → BSU2007
HB12020	BSU2007 amyE::PxylA-sigW	pVG003 → BSU2007
HB12035	BSU2007 amyE::PxylA-sigM	pVG0013 → BSU2007
HB12036	BSU2007 amyE::PxylA-sigX	pVG004 → BSU2007
HB12027	168 sigV::kan	LFH-PCR → 168
HB12082	168 amyE::PxylA-sigV	pVG001 → 168
HB0048	CU1065 dltA::spc	19
HB12093	168 dltA::spc	HB0048 ChrDNAa → 168
HB12083	168 oatA::tet	LFH-PCR → 168
HB12092	168 oatA::tet dltA::spc	HB12083 ChrDNA → HB-12093
HB12129	168 sigV::kan oatA::tet dltA::spc	HB12027 ChrDNA → HB-12092
HB10016	168 sigM::tet	44
HB10102	168 sigW::mls	44
HB10103	168 sigX::kan	44
HB10107	168 sigM::tet sigX::kan sigW::mls (Δ3)	44
Plasmids
pVG001	pSWEET-PxylA-sigV (Cmr)	This work
pVG003	pSWEET-PxylA-sigW (Cmr)	This work
pVG004	pSWEET-PxylA-sigX (Cmr)	This work
pVG0013	pSWEET-PxylA-sigM (Cmr)	This work

^aChrDNA, chromosomal DNA.

Inducible expression of ECF σ factors in B. subtilis.

The pSWEET plasmid, which is integrated into the amyE locus, was used to construct xylose-dependent expression strains (10). sigV, sigM, sigW, and sigX were amplified from strain 168 chromosomal DNA using primers 4556/4557, 4970/4590, 4558/4559, and 4560/4561 (see Table S1 in the supplemental material), respectively, and cloned into pSWEET using PacI and BamHI sites to create pVG001, pVG003, pVG004, and pVG013, respectively (Table 1). Inducible expression from each construct was checked using reporter strains: the induction of each σ factor specifically activated expression from its cognate autoregulatory promoter site. The plasmids were transformed into a B. subtilis strain carrying in-frame deletions of all seven ECF σ factor genes (Δ7ECF) (1) with chloramphenicol (5 μg/ml) selection to create strains HB12010, HB12020, HB12036, and HB12035, respectively (Table 1). The same strategy was used to integrate an ectopic copy of P_xylA-sigV into wild-type strain 168 to generate strain HB12082.

Generation of mutant strains.

Long flanking homology (LFH) PCR was used to generate deletion mutations in which the designated coding region was largely replaced with an antibiotic resistance cassette as previously described (45, 64). Strain 168 chromosomal DNA was used for PCR amplification of flanking fragments of each gene using primers 5148/5501 and 5502/5151 for sigV and 5156/5157 and 5158/5159 for oatA (see Table S1 in the supplemental material). The PCR products were joined to an antibiotic resistance cassette using joining PCR with outside primers. The final LFH product was used to transform strain 168 with selection for kanamycin (10 μg/ml) for sigV::kan and tetracycline (5 μg/ml) for oatA::tet.

Lysozyme sensitivity measurements.

Lysozyme sensitivity was determined using a disk diffusion assay performed as described previously (7, 45). Briefly, wild-type strain 168 and various mutant strains were grown to mid-logarithmic phase (optical density at 600 nm [OD₆₀₀] of 0.4) in LB medium at 37°C with aeration. A 100-μl aliquot of each culture was mixed with 4 ml of 0.75% Müller-Hinton (MH) soft agar (kept at 50°C) and directly poured onto MH plates (containing 15 ml of 1.5% MH agar). The plates were then dried for 20 min in a laminar airflow hood. Filter paper disks containing 5 μl of 100 mg/ml lysozyme were then placed on top of the agar, and the plates were incubated at 37°C overnight. The diameters of the inhibition zones (clear zones) were measured.

RNA extraction for transcriptome analyses.

A culture of HB12010 (Δ7ECF P_xylA-sigV) was grown in LB at 37°C with shaking to an OD₆₀₀ of 0.4 and then incubated for 20 min either with or without 2% xylose. A culture of 168 was grown similarly and either left untreated or treated with 0.5 μg/ml lysozyme. Total RNA was isolated from three different biological replicates for each experiment with the RNeasy Mini Kit following the manufacturer's instructions (Qiagen Sciences, Germantown, MD). After DNase treatment with Turbo DNA-free (Ambion), RNA concentrations were quantified using a NanoDrop spectrophotometer (NanoDrop Technologies Inc., Wilmington, DE) and kept at −20°C.

Microarray analyses.

Twenty micrograms of total RNA was used to make cDNA using the SuperScript Plus Indirect cDNA Labeling System (L1014-04; Invitrogen). cDNA was labeled using Alexa Fluor labeling, and microarray analysis were performed as described previously (27). Six microarrays (biological triplicates with a dye swap) were analyzed for both the σ^V regulon and lysozyme stimulon determinations. Images were processed and normalized using the GenePix Pro 4.0 software package, which produces red (R) and green (G) fluorescence intensity pairs for each gene. Each expression value is represented by up to 12 separate measurements (duplicate spots on each of six arrays). Mean values and standard deviations were calculated with Microsoft Excel. The normalized microarray data sets were filtered to remove those genes that were not expressed at levels significantly above the background under either condition (sum of mean fluorescence intensities, <20). In addition, the mean and standard deviation of the fluorescence intensities were computed for each gene, and those for which the standard deviation was greater than the mean value were ignored. The n-fold induction values were calculated using the average signal intensities from the three arrays under the different conditions.

qRT-PCR.

For quantitative real-time PCR (qRT-PCR), specific primers were designed using the B. subtilis genome sequence to amplify 100-bp products (see Table S1 in the supplemental material). Two micrograms of total RNA (isolated as described above for transcriptome analysis) was used to make cDNA using TaqMan reverse transcription reagents following the manufacturer's instructions (Applied Biosystems). The cDNA was used for qRT-PCR using iQ SYBR green Supermix in an Applied Biosystems 7300 Real Time PCR System. Quantification of 23S RNA levels was used as an internal control. The n-fold change was calculated using the difference between the cycle thresholds under the two conditions.

Determination of consensus promoter sequences.

Promoter consensus sequence alignment was performed using the Weblogo software (http://weblogo.berkeley.edu/). The σ^X regulon (11 promoters), σ^W regulon (30 promoters), and σ^M regulon (30 promoters) are based on previously published results (14, 15, 17, 25).

Microarray data accession number.

The microarray data sets are available in the NCBI GEO database under accession number GSE31563.

RESULTS AND DISCUSSION

Induction of σ^V in a Δ7ECF strain identifies direct targets of σ^V RNAP.

Previous studies have reported that artificial induction of σ^V induces dozens of genes (2, 67), suggesting that this ECF σ factor is likely important under some undefined conditions. However, only a small subset of genes were consistently detected in these studies, which involved long incubations after σ^V induction (at least 2 h). Therefore, we sought to reinvestigate the σ^V regulon under conditions that reduce indirect effects and preclude transcriptional cascades due to the activation of other ECF σ factors.

To define the σ^V regulon, we induced the expression of σ^V in a strain devoid of all other ECF σ factors (Δ7ECF) (1, 43). We used DNA microarray hybridization to monitor transcriptional changes 20 min after the induction of σ^V to selectively detect direct effects and thereby define promoters activated by σ^V RNAP. Analysis of the resulting transcriptome revealed the upregulation of ∼30 operons, including many known from previous work to be also regulated by σ^M, σ^W, and/or σ^X (Fig. 1A and Table 2; see Table S2 in the supplemental material). There was also weak upregulation of some members of the σ^B-dependent general stress response (see Table S2). These results indicate that σ^V can directly activate numerous promoter sites independently of any influence it may also have on the expression of other ECF σ factors.

Fig. 1.

The σ^V regulon. (A) Scatterplot representing the average expression levels of genes in induced (+ Xylose) versus noninduced (− Xylose) cultures of B. subtilis Δ7ECF P_xylA-sigV. The known regulons of σ^M, σ^X, or σ^W (MXW); σ^B; and the genes belonging to the sigV operon are labeled. (B) Promoter consensus sequence alignment performed using the Weblogo software (http://weblogo.berkeley.edu/) and representing the promoters that are regulated by σ^M, σ^X, or σ^W and also regulated by σ^V (σ^MXWV; top) and those promoters that are regulated by σ^M, σ^X, or σ^W that are not regulated by σ^V (σ^MXW; bottom).

Table 2.

The σ^V-regulated genes and their functional roles

Category, operon	n-Fold changea		Regulatorb	Function	Reference(s)c
	± Xylose	± Lysozyme
Regulation
sigV-rsiV	6	73	V	ECF σ and anti-σ factor	58
abh	26	2	MXW	Transition state regulator (AbrB paralogue)	60
ywaC	8	2	MWX	ppGpp synthase	48
Cell division and shape, maf-radC-mreBC-minCD	3	2	MW	Cell division and shape determination	22
Cell envelope
bcrC	42	4	MX	Undecaprenyl pyrophosphate phosphatase	9,16
ddl-murF	3	2	MWX	Peptidoglycan biosynthesis
dltABCDE	14	2	MX	d-Alanylation of teichoic acids	15
pbpX	21	3	XW	Penicillin-binding protein	15
oatA-yrhK	46	63	V	oatA, O-acetylation of peptidoglycan; yrhK, unknown	40
Detoxification
yrhHIJ	26	3	MXW	yrhH, putative methyltransferase; yrhIJ, cytochrome P450	41
yqjL	8	2	MW	Hydrolase, paraquat resistance	18
Miscellaneous
mmgD	5	2	E	2-Methylcitrate synthase	13
scoB	3	2	E	Succinyl CoA:3-oxoacid CoA-transferase (subunit B)d
spoIIB	2	2		Stage II sporulation	51
yutH	2	2		Spore coat-associated protein	61
Unknown
yebC	7	3	M	Putative integral inner membrane protein	25
yocL	5	2	E	Hypothetical protein	26
ycgR	3	2	M	Putative permease	25
ytvB	3	4		Putative conserved membrane protein
ydgA	2	2	K	Conserved hypothetical protein	56
yvaFE	2	2		Putative transcriptional regulator and transporter
ytwF	2	2		Putative sulfur transferase
ycgQ	2	2	M	Conserved hypothetical protein	25

^aThe n-fold change shown for operons represents the average of the n-fold changes for each gene in the operon (except for for sigV-rsiV, where the value for rsiV is reported).

^bV refers to σ^V, M refers to σ^M, X refers to σ^X, W refers to σ^W, E refers to σ^E, and K refers to σ^K.

^cWhere no reference is listed, the function annotation is based on GenoList (http://genodb.pasteur.fr/cgi-bin/WebObjects/GenoList).

^dCoA, coenzyme A.

The most dramatic effect of inducing σ^V was the strong upregulation of the sigV operon itself, consistent with prior reports of positive autoregulation (2). The induction of sigV itself is not informative, since this gene was induced by xylose. However, we also observed very strong induction (>40-fold) of genes downstream of sigV, including rsiV (encoding anti-σ^V), oatA (encoding a peptidoglycan O-acetyltransferase) (40), and yrhK (unknown function). Since the strain background used for this study (Δ7ECF) carries an in-frame deletion of sigV, this induction is likely indicative of the autoregulation that would occur in response to natural inducers.

Most of the remaining genes that responded strongly to the induction of σ^V are known members of the σ^M, σ^W, and σ^X regulons. Since this experiment was done in a background carrying in-frame deletions of all three of these ECF σ factors, we conclude that this reflects an overlap of the promoter recognition properties of these ECF σ factors rather than a transcriptional cascade. The induced operons (Table 2) include abh, ywaC, bcrC, dltABCDE, pbpX, and yqjL. Abh is a paralog of AbrB and functions as a transition state regulator affecting antibiotic synthesis and resistance (21, 44), YwaC is a ppGpp synthase (48), BcrC functions as an inducible undecaprenyl pyrophosphate phosphatase and thereby contributes to bacitracin resistance (9, 16), the dlt operon encodes enzymes for teichoic acid d-alanylation (49), PbpX is a low-molecular-weight penicillin-binding protein with an unknown function, and YqjL contributes to resistance to paraquat by an unknown mechanism (18).

These transcriptional profiling results indicate that σ^V activates both its own operon and a well-defined subset of the σ^M, σ^X, and σ^W regulons. Previously, we demonstrated that a key feature distinguishing σ^X- and σ^W-specific promoters is the sequence of the −10 consensus element: promoters with the sequence CGTA are generally recognized by σ^W, those with the sequence CGAC are recognized by σ^X, and those with the sequence CGTC may be recognized by both (52). To begin to define possible promoter features that account for the ability of σ^V to activate its specific subset of target operons, we created a consensus alignment of those promoters that belong to the σ^M, σ^X, or σ^W regulon that were not activated by the induction of σ^V (MXW) and compared this with the consensus of those that were also activated by σ^V (MXWV). Interestingly, the consensus for the σ^MXWV-regulated genes contains a T-rich −30 to −26 region that is not conserved in the genes regulated only by σ^MXW (Fig. 1B). We are currently testing the hypothesis that this spacer region sequence is important for promoter recognition by σ^V.

σ^V is specifically induced by lysozyme.

One of the genes most strongly induced by σ^V is oatA (formerly yrhL), which is immediately downstream of sigV-rsiV. B. subtilis oatA encodes an ortholog of a Staphylococcus aureus peptidoglycan O-acetyltransferase and has been shown genetically to affect levels of peptidoglycan O-acetylation (40). OatA provides lysozyme resistance in pathogenic Staphylococcus species (5, 7) and Lactococcus lactis (62). Furthermore, possible orthologs of σ^V were found to be induced by lysozyme exposure in Enterococcus faecalis (42) and Clostridium difficile (35). Together, these results suggest that B. subtilis sigV, and therefore the σ^V regulon, might be induced by lysozyme and thereby provide lysozyme resistance.

To test for induction by lysozyme, we used reporter strains with lacZ fusions to the autoregulated promoters of sigV, sigM, sigW, and sigX. β-Galactosidase measurements with and without lysozyme treatment show that the sigV promoter is strongly (∼90-fold) and specifically induced by lysozyme (Fig. 2A). To verify and extend these results, we performed qRT-PCR with RNA isolated from wild-type cells treated with different concentrations of lysozyme (Fig. 2B). As little as 0.01 μg/ml lysozyme strongly induced sigV activity (∼10-fold), and even at 1 μg/ml lysozyme, there was little, if any, observable lysis of cells during the 20 min of treatment. This demonstrates that activation of σ^V is extremely sensitive to even mild digestion of the cell wall and is not correlated with cell lysis.

Fig. 2.

σ^V is strongly and specifically induced by lysozyme. (A) β-Galactosidase activity of P_sigV-lacZ, P_sigM-lacZ, P_sigW-lacZ, and P_sigX-lacZ with or without treatment with a subinhibitory concentration of lysozyme. The reporter strains were grown to an OD₆₀₀ of 0.4 and then incubated for 20 min either with (+ Lys) or without (− Lys) 1 μg/ml lysozyme. This experiment was performed in three biological replicas and repeated at least three times. Bars represent mean values, and error bars indicate standard deviations. Student t tests were performed, and a statistically significant difference (P value, <0.005) between the control (− Lys) and lysozyme-treated (+ Lys) cells is denoted by an asterisk. (B) qRT-PCR of sigV expression under lysozyme induction. Strain 168 was grown to an OD₆₀₀ of 0.4 and incubated for 20 min with the addition of different concentrations of lysozyme. qRT-PCR was performed with primers specific for sigV and for 23S rRNA as a control. The bars show the n-fold change in induction after treatment with lysozyme. The results shown are representative of experiments performed at least three times.

The lysozyme stress response is dominated by strong activation of the σ^V regulon.

We next sought to obtain a global view of the lysozyme stress response by monitoring the changes in the transcriptome induced by brief treatment (20 min) with sublethal levels of lysozyme known to be sufficient for full induction of the σ^V regulon. This allows us to compare the response elicited in wild-type cells upon naturally inducing σ^V with that elicited by the more artificial situation of ectopically inducing σ^V in the Δ7ECF background. Remarkably, the lysozyme stimulon is dominated by the strong (>50-fold) induction of sigV and the immediately adjacent rsiV, oatA, and yrhK genes. Thus, not only does lysozyme selectively activate σ^V, there are no other cell envelope stress systems that appear to respond strongly to this level of lysozyme. Altogether, the lysozyme stimulon includes weak induction of as many as 76 operons (Fig. 3; see Table S3 in the supplemental material), including several members of the Fur regulon (3). Other inducible genes have functions in cell wall biosynthesis, cell division, and antibiotic resistance (Table 2; see Table S3 in the supplemental material). Overall, there is very good congruence between the σ^V regulon and the lysozyme stimulon (Table 2).

Fig. 3.

The lysozyme stimulon. The scatterplot represents the average expression levels of genes in induced (+ Lysozyme) versus noninduced (− Lysozyme) cultures of B. subtilis 168. Genes belonging to the regulons of σ^M, σ^X, or σ^W (MXW); σ^B; Fur; and AbrB and the genes belonging to the sigV operon are labeled.

One notable difference between the σ^V regulon (Fig. 1A) and the lysozyme stimulon (Fig. 3) is that in the former, the induction of genes also potentially regulated by other ECF σ factors was generally much stronger. This likely reflects the fact that the σ^V regulon was determined in a strain background devoid of other ECF σ factors that might have otherwise contributed to the background expression of these genes. In other words, in wild-type cells, the lysozyme-dependent induction of some genes is superimposed on their basal transcription. Of the seven ECF σ factors, at least two (σ^M and σ^X) are found associated with RNAP in nonstressed cells (23) and this likely contributes to basal gene expression.

oatA is cotranscribed with sigV and rsiV.

The transcriptional analyses above revealed a coordinate induction of oatA with the upstream sigV and rsiV genes, suggestive of a likely operon structure. However, a 110-bp gap separates rsiV from the downstream oatA gene and this region contains a predicted transcription terminator. To test the hypothesis that these genes can be expressed as a single transcript, we performed qRT-PCR of the intergenic junctions between sigV and rsiV, between rsiV and oatA, and between oatA and yrhK (Fig. 4). We detected an increase in the expression of all intergenic junctions correlated with the induction of sigV. These results suggest that the predicted terminator is, at best, only partially efficient and the downstream genes can be expressed as part of a readthrough transcript. Furthermore, there is no predicted σ^V-dependent promoter in the intergenic region between rsiV and oatA, and lacZ fusions made with different fragments upstream of oatA have no activity, even when σ^V is induced (data not shown). Inspection of published tiling array data (54) also supports a likely four-gene operon extending from sigV to yrhK. We therefore conclude that oatA is cotranscribed with sigV, rsiV, and yrhK and that the σ^V-dependent induction of oatA reflects the activity of the sigV autoregulatory promoter.

Fig. 4.

oatA is part of the sigV operon. qRT-PCR of intergenic junctions after induction of σ^V. (A) Strain 168 was grown to an OD₆₀₀ of 0.4 and incubated for 20 min either with (+Lys) or without (−Lys) 0.5 μg/ml lysozyme. (B) Strain 168 P_xylA-sigV was grown to an OD₆₀₀ of 0.4 and incubated for 20 min either with or without the addition of 2% xylose. qRT-PCR was used to quantify the n-fold change in each junction region after treatment. The results shown are representative of experiments repeated at least three times.

σ^V plays a central role in lysozyme resistance.

We assessed the role of σ^V in lysozyme resistance using a modified disk diffusion protocol (7). A sigV null mutant is nearly as sensitive as a strain missing all seven ECF σ factors, implying that this single ECF σ factor is the major lysozyme resistance determinant (Fig. 5A). In preliminary studies using a strain with the sigV gene disrupted by a codirectional kan cassette, the role of σ^V in lysozyme resistance was partially masked by readthrough transcription into oatA (data not shown). Therefore, we used an allelic replacement mutation in which the sigV gene was disrupted by a divergently oriented antibiotic resistance cassette (Fig. 5A). Identical results were also seen when an in-frame deletion of sigV (from Δ7ECF) was used instead of an allelic replacement mutant (data not shown). σ^V was not induced by, and had no apparent role in resistance to, mutanolysin, which is known to cleave peptidoglycan irrespective of N-acetylmuramic acid (MurNAc) O-acetylation (63). Moreover, σ^V does not appear to confer resistance to several cell wall-active antibiotics (bacitracin, nisin, moenomycin, d-cycloserine, polymyxin B, cefuroxime, fosfomycin, vancomycin, and ramoplanin) (data not shown).

Fig. 5.

σ^V confers resistance to lysozyme through regulation of oatA and the dlt operon. Zone-of-inhibition experiments were used to quantify the lysozyme sensitivity of B. subtilis strains. Strains were grown to an OD₆₀₀ of 0.4, and an inoculum of each culture was used to make a lawn of cells on 0.75% MH agar. Disks containing lysozyme were placed on top of the lawn, and the inhibition of growth was measured after incubation at 37°C for 16 h. Each bar represents the average zone of inhibition of a least three assays performed with three biological replicas of each strain. The zone of inhibition is expressed as the average total diameter (± the standard error) of the clear zone. (A) Lysozyme sensitivity of wild-type and ECF σ factor mutants. (B) Lysozyme sensitivities of wild-type, Δ7ECF, and ECF σ factor-inducible strains. The data shown are for cultures grown under inducing conditions (2% xylose). (C) Comparison of lysozyme sensitivities of the wild-type and oatA and dltA mutant strains. A statistically significant difference (P value, <0.005) between the wild-type and mutant strains (panels A and C) or between the Δ7ECF strain and the other strains tested (panel B), as determined by Student t tests, is denoted by an asterisk.

In contrast to sigV, single mutations of the most active ECF σ factors (σ^M, σ^X, and σ^W) did not affect lysozyme sensitivity. Even a sigM sigX sigW triple null mutant (Δ3) had only a modest increase in lysozyme sensitivity (Fig. 5A). Thus, these σ factors may play a small role in lysozyme resistance, but this is negligible in cells expressing σ^V. We next tested lysozyme resistance in the Δ7ECF strain upon the induction of various ECF σ factors (Fig. 5B). Only the induction of σ^V restored lysozyme resistance, whereas the induction of σ^M, σ^X, and σ^W had little, if any, effect. These results demonstrate that σ^V is both necessary and sufficient for the induction of lysozyme resistance determinants.

Lysozyme resistance is due to σ^V-dependent activation of OatA and Dlt.

Since the induction of σ^V can activate the expression of 20 or more operons (Fig. 1 and Table 2), we next sought to identify which σ^V-regulated genes are important for lysozyme resistance. We focused our attention on two σ^V-activated functions previously implicated in lysozyme resistance in other organisms: oatA and the dlt operon (5, 7, 42). Although a single mutation of either oatA or dltA did not affect lysozyme sensitivity, an oatA dltA double mutant was fully as sensitive as a sigV null mutant (Fig. 5C). We therefore suggest that upregulation of either or both of these operons can account for the role of σ^V in lysozyme resistance. Support for this notion is provided by the finding that the mutation of sigV in an oatA dltA double mutant does not further increase sensitivity (Fig. 5C). These results suggest that in B. subtilis, lysozyme resistance is provided by σ^V largely through the upregulation of oatA and the dlt operon.

The regulation of the dlt operon has been studied in detail. The dlt operon is potentially activated by σ^D (50), σ^X (15), and σ^M (25). As noted above, oatA is cotranscribed with sigV. To corroborate and extend our microarray results, we performed qRT-PCR studies with cells where σ^V was induced either ectopically with xylose or by lysozyme treatment (Table 3). As expected, lysozyme treatment strongly induced sigV and oatA in wild-type and sigX null mutant cells but not in the sigV null mutant. Conversely, ectopic induction of sigV also strongly induced oatA and dltA expression (Table 3). Ectopic induction of sigX induced dltA, consistent with the reported regulation of this operon by σ^X (15), but did not induce oatA.

Table 3.

qRT-PCR quantitation of sigV, oatA, and dltA expression

Condition and strain	n-Fold transcript induction
	sigV	oatA	dltA
± Lysozyme
Wild-type 168	50.5	25.7	2.6
ΔsigV mutant	NAa	1.3	2.0
ΔsigX mutant	82.4	19.6	5.8
± Xylose
Δ7 PxylA-sigV mutant	NA	5.6	5.8
Δ7 PxylA-sigX mutant	1.0	0.7	37.4

^aNA, not applicable. This induction value is not meaningful since the gene is deleted (row 2) or artificially induced from P_xylA (row 4).

As noted above, induction of σ^X is unable to restore lysozyme resistance to the Δ7ECF strain (Fig. 5B). However, induction of σ^X is clearly sufficient for strong activation of the dlt operon (Table 3). This suggests that induction of the dlt operon is not sufficient to provide lysozyme resistance in a strain lacking the other six ECF σ factors. In apparent contrast to this result, OatA and Dlt are redundant in providing lysozyme resistance to wild-type cells (only an oatA dltA double mutant was as sensitive to lysozyme as a sigV mutant; Fig. 5C). Therefore, in an oatA mutant, dlt appears to be the only σ^V-dependent operon required to provide resistance. However, in the Δ7 strain, which is significantly altered in its physiology (43), the artificial induction of Dlt (upon activation of σ^X) is not sufficient for lysozyme resistance. These results can be reconciled if other ECF σ factor-dependent genes also make contributions to lysozyme resistance, including perhaps pbpX (36).

Diverse mechanisms of lysozyme resistance.

Lysozyme hydrolyzes the β-1,4-glycosidic bond between MurNAc and N-acetylglucosamine (63). In addition to its muramidase activity, lysozyme also has a cationic antimicrobial peptide activity (34, 47).

There have been several different mechanisms reported for lysozyme resistance in Gram-positive bacteria. Most commonly, resistance is achieved by either modification of the peptidoglycan substrate by MurNAc O-acetylation (7, 12) or changes in the overall net charge of the cell envelope by d-alanylation (34). In Staphylococcus aureus, OatA-dependent O-acetylation and d-alanylation of teichoic acids function synergistically to provide full lysozyme resistance (5). In this organism, the GraRS two-component system plays a key role in lysozyme resistance by activating the expression of the dlt operon (34). Resistance to lysozyme in Enterococcus faecalis is provided by both oatA and dltA, and σ^V plays an additional, undefined role: a sigV dltA oatA triple mutant is more sensitive to lysozyme than either a sigV or a dltA oatA mutant (42). In this organism, the ECF σ^V factor acts as a general stress response σ factor (4) and is not required for the expression of oatA or dltA (42). Lactococcus lactis also protects its peptidoglycan by O-acetylation, and in this case, oatA transcription is regulated by SpxB (62). In L. monocytogenes, lysozyme resistance is achieved by both O-acetylation and N-deacetylation (53). In B. anthracis, lysozyme resistance is mediated by both O-acetylation, catalyzed by two distinct enzymes, and N-deacetylation (40) but the regulation of these resistance determinants is not yet characterized.

Here we define both the genetic determinants and regulatory pathways that control intrinsic lysozyme resistance in B. subtilis. In comparison with a highly lysozyme-resistant pathogen such as S. aureus, B. subtilis is relatively sensitive to lysozyme. Nevertheless, this organism clearly maintains an inducible resistance system controlled by σ^V. B. subtilis was shown previously to contain a functional OatA homolog: ∼35% of MurNAc residues were O-acetylated in wild-type cells, and this was reduced 2.5-fold in an oatA null mutant (40). Based on the results here, it seems likely that the level of MurNAc modification increases significantly in cells exposed to even low levels of lysozyme, thereby providing an adaptive mechanism for lysozyme resistance. B. subtilis also extensively modifies teichoic acids by d-alanylation (49), which is mediated by the products of the dlt operon (50). The dlt operon has been shown to be regulated by σ^D (50), σ^X (15), σ^M (25), and as shown here, σ^V. Our results indicate that in B. subtilis, σ^V is the major ECF σ factor responsible for lysozyme resistance: a sigV mutant is nearly as sensitive as the Δ7ECF mutant (Fig. 5A). When induced, σ^V activates the expression of both oatA, which is part of the autoregulated sigV operon, and the dltA operon (Tables 2 and 3). These two cell wall modification pathways provide redundant mechanisms of lysozyme resistance, as also noted in S. aureus (34).

While this report was in preparation, very similar results were obtained by Ellermeier and coworkers (36). They also note that sigV is cotranscribed with oatA and that induction of this operon by lysozyme contributes to lysozyme resistance. They further show that the dlt operon and pbpX also contribute to lysozyme resistance. While the results of these two studies are generally in good agreement, we find that a sigV null mutant is nearly as sensitive as the Δ7ECF strain whereas Ho et al. report that a sigV null mutation has only a modest effect (2-fold) on lysozyme resistance which is greatly enhanced in strains additionally defective in sigX and/or sigM. Our results also differ with respect to the importance of the dlt operon for lysozyme resistance. Whereas we see no significant effect of a dlt null mutation on lysozyme resistance (Fig. 5C), Ho et al. report that a dlt null mutation has a greater effect than a sigV null mutation. The reasons for these differences are unclear.

Although hen egg white lysozyme is commonly used for testing of lysozyme resistance, it is a surrogate for the physiological stresses likely to be encountered in the environment. In human mucosal secretions, lysozyme can be present at levels of up to 5 mg/ml and thereby provides an important component of innate immunity (24). Indeed, peptidoglycan O-acetylation and lysozyme resistance correlate with pathogenicity in S. aureus (6), in E. faecalis (42), and likely in other human pathogens. The role of lysozymes and lysozyme resistance mechanisms has not been as well studied in soil bacteria. However, soil bacteria are known to produce and in some cases secrete peptidoglycan-degrading enzymes. Myxococcus xanthus, which feeds on other soil bacteria, secretes several lytic enzymes (8), including at least one with lysozyme-like activity (29). Streptomyces coelicolor also secretes several muramidases (28). Bacillus spp. also produce a number of autolytic and peptidoglycan-degrading enzymes (57). We are currently studying the possibility that σ^V provides resistance to peptidoglycan-degrading enzymes produced by other soil bacteria.