Adaptation of bacteria to the prevailing environmental and nutritional conditions is often mediated by two-component signal transduction systems (TCS). The Bacillus subtilis YycFG TCS has attracted special attention as it is essential for viability and its regulon is poorly defined. Here we show that YycFG is a regulator of cell wall metabolism. We have identified five new members of the YycFG regulon: YycF activates expression of yvcE, lytE and ydjM and represses expression of yoeB and yjeA. YvcE(CwlO) and LytE encode endopeptidase-type autolysins that participate in peptidoglycan synthesis and turnover respectively. We show that a yvcE lytE double mutant strain is not viable and that cells lacking LytE and depleted for YvcE exhibit defects in lateral cell wall synthesis and cell elongation. YjeA encodes a peptidoglycan deacetylase that modifies peptidoglycan thereby altering its susceptibility to lysozyme digestion and YdjM is also predicted to have a role in cell wall metabolism. A genetic analysis shows that YycFG essentiality is polygenic in nature, being a manifestation of disrupted cell wall metabolism caused by aberrant expression of a number of YycFG regulon genes.
Bacteria use two-component systems (TCS) to monitor and respond appropriately to the prevailing conditions. A canonical TCS has four domains usually organized into two proteins: a sensor kinase and a response regulator (for reviews see Hoch and Silhavy, 1995; Stock et al., 2000). Sensor kinases detect specific stimuli that activate their autokinase activity. The activated kinase phosphorylates the response regulator that is usually a transcription factor thereby altering its activity and effecting changes in gene expression. By functioning as a cognate pair, the elicited cellular response is appropriate to the initial stimulus.
There are 34 TCS encoded in the Bacillus subtilis genome (Kunst et al., 1997; Fabret et al., 1999). While the cellular roles of many TCS are now known in some detail, the YycFG system remains something of an enigma. Uniquely among the complement of TCS in B. subtilis, YycFG is essential for survival (Fabret and Hoch, 1998). An orthologous TCS is found in the low G+C group of bacteria where two genomic organizations are observed: the yycFGHIJ-type operon found in Bacillus, Staphylococcus and Listeria species and the yycFGJ-type operon found among Streptococcus and Lactococcus species (Fabret and Hoch, 1998; Lange et al., 1999; Martin et al., 1999; O'Connell-Motherway et al., 2000; Throup et al., 2000; Kallipolitis and Ingmer, 2001; Ng et al., 2004; Liu et al., 2006). YycF orthologues are very similar and are essential in most members of the low G+C group of bacteria with the exception of Lactococcus lactis (O'Connell-Motherway et al., 2000). More heterogeneity is found among YycG orthologues: YycG is essential in B. subtilis and Staphylococcus aureus and has two transmembrane domains flanking a large extracellular loop (154 amino acids in B. subtilis). The YycG orthologue in L. lactis, llkinC, is non-essential and has only four amino acids in the loop region, while in Streptococcus, YycG has a single transmembrane domain and is dispensable under some conditions (Ng et al., 2003; 2004). Such diversity within the YycG-sensing domains suggests variability in the signals being detected in different bacteria. Although the signal(s) to which YycG responds is unknown, studies by Hoch and colleagues show that YycH and YycI are located extracytoplasmically but tethered to the membrane and that they interact with YycG to form a ternary complex that regulates its activity (Szurmant et al., 2005; 2006; 2007). This insight is consistent with yycH and yycI being found only in bacteria that have a YycG protein containing a large extracytoplasmic loop (Szurmant et al., 2007). Control of cell wall and cell membrane homeostasis is the predominant theme of YycFG function that has emerged from the many studies already cited. The phenotypic effects of YycFG depletion are dependent on the bacterium under investigation: YycFG-depleted B. subtilis are aberrant in size and shape with some cells lacking contents (Fabret and Hoch, 1998); a strain of S. aureus with a temperature-sensitive mutation in YycF resulted in hypersusceptibility to MLS-type antibiotics and hypersensitivity to unsaturated fatty acids (Martin et al., 1999; 2002) while YycFG participates in competence development in Streptococcus pneumoniae (Echenique and Trombe, 2001; Wagner et al., 2002; Ng et al., 2003). YycFG-controlled genes that participate in cell wall homeostasis include those encoding the autolysins YkvT and YocH in B. subtilis (Howell et al., 2003), LytM in S. aureus (Dubrac and Msadek, 2004) and LytN, LytB and PcsB in S. pneumoniae (Wagner et al., 2002; Ng et al., 2003). Howell et al. (2006) report that the teichoic acid biosynthetic operons tagAB and tagDEF are regulated by YycFG in B. subtilis while fatty acid biosynthesis is reported to be controlled by YycFG in S. pneumoniae (Mohedano et al., 2005; Ng et al., 2005). Expression of the cell division genes ftsAZ are YycFG-controlled in B. subtilis (Fukuchi et al., 2000).
There are intriguing links between YycFG, the phosphate limitation-inducible PhoPR TCS and phosphate metabolism in B. subtilis. YycFG clusters with PhoPR and ResDE (which controls the cellular response to oxygen availability) in a phylogenetic tree of B. subtilis TCS. The similarity between YycF and PhoP was demonstrated experimentally by showing that the PhoP′–′YycF and YycF′–′PhoP hybrid response regulators are functional. The relationship between YycFG, PhoPR and phosphate limitation is manifest in the findings that the PhoR kinase can phosphorylate YycF in addition to its cognate PhoP response regulator, while YycG can only activate its cognate YycF response regulator; that YycFG is required for normal induction of a PhoPR-mediated phosphate limitation response; and that expression of some genes (e.g. phoPR, tagAB, tagDEF) is controlled by both YyFG and PhoPR (Howell et al., 2003; 2006). These observations imply a complex relationship between the YycFG and PhoPR TCS and their respective regulons.
There is only a small number of essential TCS in bacteria, among them YycFG and orthologous TCS that are found in the low G+C group of bacteria. However, the reason for YycFG essentiality has been established unambiguously only in S. pneumoniae.Ng et al. (2003) have shown that the essential nature of YycFG (VicRK) in S. pneumoniae derives from its activation of pcsB expression, a gene that encodes a putative murein hydrolyase – constitutive expression of pcsB suppresses the requirement for the YycFG orthologues VicRK. The essential nature of YycFG in B. subtilis has not yet been established. With the exception of the tagAB and tagDEF operons, none of the identified YycFG-controlled genes is essential (Kobayashi et al., 2003). However, YycFG-controlled expression of teichoic acid biosynthesis is not likely to be the cause of essentiality as YycFG depletion results in continued expression of the tag operons (Howell et al., 2006).
In this study we sought to establish the function of the YycFG TCS in B. subtilis, to identify constituent genes of the YycFG regulon and to determine the reason for its essentiality. We show that YycFG regulates cell wall metabolism. Among the five newly identified genes of the YycFG regulon are yvcE and lytE encoding endopeptidase-type autolysins, one or other of which is essential for growth and that perform an important function in lateral cell wall synthesis and cell elongation. We show that YycFG essentiality is polygenic in nature being a manifestation of disrupted cell wall metabolism caused by aberrant expression of a number of YycFG regulon genes.
Identification of additional members of the YycFG regulon
To identify the constituent genes of the YycFG regulon, the transcriptome of normally growing cells was compared with the transcriptome of cells harvested at the precise point of growth cessation due to YycFG depletion. It was reasoned that such an experimental approach would increase the likelihood of identifying the genes whose aberrant expression leads to growth cessation and cell lysis upon YycFG depletion. Three experimental replicates were performed as described in Experimental procedures and the growth profiles and analysis of the RNA preparations are presented in Fig. S1. The complete results of the transcriptomic analysis can be assessed at http://www.tcd.ie/genetics/devine/bisicchia2007.php. To assess whether YycFG depletion resulted in global changes in the gene expression profile, a Pearson correlation coefficient was calculated for YycFG-replete and YycFG-depleted cultures using (i) expression of all genes and (ii) expression of 268 of the 271 genes reported to be essential in B. subtilis (Kobayashi et al., 2003). The correlation coefficients were 0.9864 and 0.9969, respectively, showing that YycFG depletion had not resulted in a global change in RNA levels at the point of harvesting. Analysis of the transcriptome data also confirmed that expression of the yycFGHIJyyxA operon was reduced as expected in the cultures grown without IPTG: interestingly yycG, yycH, yycI, yycJ and yyxA transcripts were reduced to a greater extent than was the yycF transcript suggesting processing and/or differential stability of the operon transcript (data not shown).
Eight genes that show significant changes in transcript level between YycFG-replete and YycFG-depleted cells are listed in Table 1. Three genes are activated by YycFG: ydjM encodes a protein of unknown function with no orthologue in current databases; yvcE (renamed cwlO) encodes an endopeptidase-type autolysin previously characterized by Yamaguchi et al. (2004) and yocH, previously shown to be YycFG-regulated by Howell et al. (2003). Five genes are repressed by YycFG: yjeA is homologous to peptidoglycan deacetylases; yocE (des) encodes a fatty acid desaturase (Aguilar et al., 1998); dhbF encodes a protein that participates in siderophore biosynthesis (May et al., 2001); yoeB encodes a protein that modulates autolysin activity (Salzberg and Helmann, 2007) while ykuP encodes a protein of unknown function.
Table 1. Genes that show differential expression upon YycFG depletion.
To confirm the results of the array analysis, the expression profiles of seven genes listed in Table 1 (dhbF was excluded because of transcript size) were established by Northern analysis (using new RNA preparations). The results (Fig. 1) show that yocH, yvcE and ydjM expression profiles differ somewhat during the growth cycle: yocH and ydjM expression is high during exponential growth and falls to a lower level post-exponentially; yvcE expression is high during early exponential growth, falls to a lower level between T−20 and T0 with a further decrease evident during stationary phase. However, YycFG depletion (AH9912 –IPTG) results in dramatically reduced transcript levels of all three genes, confirming positive regulation by this TCS. Of the four genes negatively regulated by YycFG, expression is low or undetectable during exponential growth in wild-type strain 168, but increases significantly in YycFG-depleted cells (AH9912 -IPTG), again confirming the transcriptome results. Thus, we have identified six genes whose expression is responsive to the cellular YycFG level.
YycF directly regulates expression of ydjM, yvcE, yjeA and yoeB
Gel shift analysis was performed to establish whether YycF directly regulates expression of these six genes (direct regulation of yocH expression was previously shown, Howell et al., 2003). A biotin-labelled DNA fragment spanning the promoter region of each gene was prepared by PCR, while a fragment containing the htrA promoter was used as a negative control. Each reaction also contained the cognate ′YycG kinase. The results are shown in Fig. 2. It is evident that YycF∼P binds to the fragments containing the promoters of yvcE, yjdM, yjeA and yoeB (Fig. 2, lanes 2–4). In each case, YycF∼P binding to the labelled DNA fragment can be competed by addition of unlabelled DNA fragment, confirming the specificity of interaction (data not shown). No retardation of the fragment containing the htrA promoter was observed as expected even upon addition of 20 μM YycF∼P (htrA, lane 2). However, YycF∼P did not bind to the des and ykuP promoters (data not shown). We therefore conclude that YycFG directly regulates expression of ydjM, yvcE, yoeB and yjeA.
Transcriptional analysis of the ydjM, yvcE, yoeB and yjeA genes
To characterize the ydjM, yvcE, yoeB and yjeA gene promoters and to establish the effect of YycFG on their expression, we performed primer extension and transcriptional fusion analysis. We established a unique initiation point of transcription for ydjM, yvcE, yoeB and yjeA, shown underlined and in bold as +1 in Table 2. The promoter sequences of genes positively and negatively regulated by YycFG were then aligned relative to their initiation points of transcription (Table 2): the YycF-regulated promoters of yocH (Howell et al., 2003) and ftsA (Fukuchi et al., 2000) are also included. In a previous study, we established a potential consensus YycF recognition sequence consisting of two hexanucleotide direct repeats separated by five nucleotides: 5′-TGT A/T A A/T/C.N5.TGT A/T A A/T/C-3′ (Howell et al., 2003). This YycF consensus sequence is present in the promoter regions of yvcE and ydjM (putative YycF binding motifs are in red capitals, Table 2). However, while these promoters have consensus SigA-type −10 motifs, the sequence at the −35 position has little resemblance to a SigA-type −35 motif. The promoters of yvcE and ydjM differ therefore from those of yocH and ftsA as the latter have canonical SigA-type −35 and −10 motifs in addition to YycF recognition sequences. It is noteworthy that despite lacking a canonical −35 motif, the positioning of the YycF binding motif relative to the −10 motif in the yvcE and ydjM promoters is the same as that seen in the ftsA promoter. Interestingly, there are putative SigA-type −35 motifs within the promoter region of yvcE and ydjM (dotted underlined in Table 2), although their positioning is unusual relative to the −10 motifs (an 11-base-pair spacing between the putative −10 and −35 regions of the yvcE promoter and an extended 23-base-pair spacing in the ydjM promoter) and their significance is uncertain. The positioning of YycF binding sites within the yoeB and yjeA promoters is characteristic of negative regulation: the putative YycF binding sequences are positioned either overlapping (yoeB) or downstream (yjeA) of the putative SigA-type promoters (Table 2). Our mutational analysis indicates that yoeB has two potential YycF binding motifs overlapping the putative SigA-type promoter and initiation point of transcription with the promoter distal motif having an unusual 3 bp spacing (Table 2). The yjeA gene has one putative YycF binding motif that overlaps the initiation point of transcription. This promoter analysis of the newly identified genes of the YycFG regulon consolidates the YycF binding consensus proposed by Howell et al. (2003).
Table 2. Alignment of promoters regulated by the YycFG two-component system.
Transcriptional bgaB fusions were constructed with the wild-type and mutated promoter fragments of each gene (mutations are shown in green type in Table 2) to confirm that the putative binding motifs participate in YycFG-mediated control of gene expression. The results of the expression analysis are presented in Fig. 3. Expression of ydjM (squares, Fig. 3) and yvcE (squares, Fig. 3) is low during exponential growth and in both cases increases somewhat during stationary phase. The increased accumulation of β-galactosidase post-exponentially, despite decreased transcript levels (see Fig. 1), can be explained by continued yvcE and ydjM expression albeit at a low level in the absence of cell division, coupled with the known stability of β-galactosidase in B. subtilis (Schrogel and Allmansberger, 1997). However, mutation of the putative YycF binding sites (constructs ydjM1–bgaB and yvcE1–bgaB) significantly reduces expression of both fusions with less than 5 units of activity being observed at all stages of the growth cycle in each case (triangles, Fig. 3). The converse result was obtained with the yjeA and yoeB fusions. The yjeA and yoeB fusions accumulate approximately 5 U and 50 U of BgaB activity during exponential growth with levels increasing during the stationary phase of the growth cycle (squares, Fig. 3). However, mutation of the putative YycF binding sequences in the promoters of both genes (fusions yoeB1–bgaB, yoeB2–bgaB and yjeA1–bgaB, Table 2) results in significantly higher expression at all stages of the growth cycle (triangles and circles, Fig. 3). In summary these data confirm that the YycF binding sequences identified within the promoter regions of the ydjM, yvcE, yoeB and yjeA do mediate YycFG-controlled expression of these genes.
The spacing between the YycF binding site and the −10 region of the ydjM promoter is crucial for YycFG-mediated regulation of expression
The yvcE and ydjM promoters have three characteristic features: (i) a 33 bp spacing between a consensus SigA-type −10 motif and the YycF binding motif; (ii) the absence of a SigA-type −35 motif at the expected position; and (iii) a SigA-type −35 motif at a non-canonical position within the promoter. To evaluate the contributions of these features to YycF-mediated control of gene expression, we made a 6 bp deletion (bases marked in blue, Table 2) in the promoter regions of the ydjM–bgaB and ydjM1–bgaB fusions yielding ydjM2–bgaB and ydjM3–bgaB respectively. The effects of these deletions are twofold: the spacing between the YycF binding motif and the −10 motif is altered while the putative −35 motif (TTGAAA, dotted underline) is concomitantly placed at the normal SigA-type position. The expression profiles of these promoters are shown in Fig. 4. Expression of the wild-type promoter (squares, ydjM–bgaB) is low throughout the growth cycle and is further reduced when the YycF binding site is mutated (triangles, ydjM1–bgaB). However, when the putative −35 motif is placed at the normal position to give a canonical 17 bp spacing (diamonds, ydjM2–bgaB), expression increases during the exponential growth phase (to 20 U) and significantly, continues throughout the stationary phase (to ∼60 U). This indicates that this −35 motif is functional when placed in the context of a canonical SigA-type promoter. To assess whether YycF contributes to the expression of the ydjM2–bgaB construct, one of the direct repeats was mutated (shown in green, Table 2) to give fusion ydjM3–bgaB. Results show that mutation of this motif had no effect on expression (circles, Fig. 4). As this mutation was shown to abolish YycF-mediated control of the wild-type promoter (triangles, Fig. 4), it is concluded that the spacing between the YycF binding motif and the SigA-type −10 motif is critical for YycF-mediated control of expression. While the putative SigA-type −35 motif can function when normally placed, whether it plays a role in normal promoter function requires further investigation.
Expression of the LytE autolysin is controlled by YycFG
The preceding sections established three characteristics of YycFG regulon genes: (i) promoters have a copy of the YycF binding motif, (ii) YycF-activated promoters often have a canonical SigA-type −10 motif but not a −35 motif at the expected position, and (iii) the YycFG regulon has a high proportion of genes involved in cell wall metabolism. We used these criteria to identify additional YycF-controlled genes that might have been missed in the array analysis perhaps because their transcript level was not sufficiently depleted at the point of growth cessation. We focused on the lytE gene as it fulfilled all three criteria: (i) it is expressed during exponential growth from a SigA-type promoter that has a canonical −10 motif but the sequence at the −35 position (TATTTC) does not conform to a SigA-type motif, (ii) there is a potential YycF binding site positioned upstream of the −35 region, and (iii) it encodes an dl-endopeptidase-type autolysin like the YycFG-regulated yvcE (Table 2; Ishikawa et al., 1998). To determine whether lytE expression is dependent on YycFG, we performed Northern analysis on RNA harvested from cultures of wild-type strain 168 and from strain AH9912 grown under YycFG-replete (+IPTG) and YycFG-depleted (–IPTG) conditions. The growth profiles of each culture are shown in Fig. 5A. Northern analysis (Fig. 5B) shows that in wild-type strain 168 lytE is expressed at a low level during exponential growth (T60 and T80) in Luria–Bertani (LB) medium and that the level of transcript increases in the stationary phase (T120 and T140). This differs from the profile in SM medium where expression is turned off when cells enter stationary phase (Ishikawa et al., 1998). Cells grown in the presence of 1 mM IPTG show an increased level of lytE transcript during exponential growth (T60 and T80) when compared with the level seen in wild-type cells. It is interesting that YycF control of yvcE expression is also most obviously manifest early during exponential growth at T−60 and T−40 (Fig. 1). In cultures grown without IPTG where YycFG becomes limiting, the level of lytE transcript is initially similar to that seen in wild-type cultures [compare T60–T100 between wild-type and AH9912 (–IPTG)] but subsequently decreases. A duplicate gel using these RNA preparations showed a large increase in the level of yoeB transcript in the AH9912 (–IPTG) culture confirming that YycFG depletion had occurred (data not shown). This lytE expression profile is consistent with the array analysis in that the level of lytE transcript is not decreased at the point of growth cessation upon YycFG depletion (the point of RNA harvesting for the array analysis) but does decrease at later time points. Experiments using lytE–bgaB transcriptional fusions showed that the post-exponential induction of lytE expression is YycFG-independent. To ascertain whether YycF directly controlled expression of lytE, gel shift analysis was performed on a fragment containing the lytE promoter. Results show that the mobility of the lytE promoter fragment is retarded upon addition of YycF∼P protein (Fig. 5C, lanes 2–4). In summary, it is concluded that lytE expression is directly regulated by the YycFG TCS.
Phenotypic analysis of strains mutated in genes of the YycFG regulon
A genetic analysis was undertaken to determine which gene or combination of genes contributes to YycFG essentiality. Strains carrying null mutations in any one of the genes yocH, ydjM, yvcE, lytE, yoeB and yjeA are viable indicating that no individual gene is essential, confirming previous results (Kobayashi et al., 2003). Our attention then focused on the four genes positively regulated by YycFG: yocH, yvcE, lytE and ydjM. It was possible to construct strains with the following combination of double mutants: BP087 (ΔyocH::kanrΔyvcE::spcr), BP071 (ΔyocH::kanrΔydjM::tetr), BP083 (ΔyvcE::spcrΔydjM::tetr), BP122 (ΔlytE::cmrΔyocH::kanr) and BP123 (ΔlytE::cmrΔydjM::tetr). However, we were unable to construct a strain with null mutations in both yvcE and lytE that encode endopeptidase-type autolysins. Attempts to generate the double mutant by transformation of strain BP079 (ΔyvcE) with chromosomal DNA from strain L16638 (ΔlytE) resulted in only three colonies, which were each shown to be mutated in one or other gene only and presumably arose through congression. This supports the view that cellular growth requires either YvcE(CwlO) or LytE. To verify that the yvcE and lytE genes cannot both be inactivated, strain BP115 (PxylyvcEΔlytE::cmr) was constructed where yvcE expression is inducible by xylose in a lytE null mutant background. This strain was grown in the presence and absence of xylose and the growth profiles are shown in Fig. 6A, together with those of control strains wild-type 168 (diamonds), L16638 (ΔlytE::cmr, triangles) and BP079 (ΔyvcE::spcr, circles). It is evident that BP079 (ΔyvcE::spcr) grows similarly to wild-type strain 168 (compare diamonds and circles) whereas L16638 (ΔlytE::cmr) displays a growth defect (triangles). In the presence of 1% xylose, strain BP115 grows similarly to strain L16638 (ΔlytE::cmr), consistent with both being mutant for the lytE gene (compare closed squares and triangles). However, when BP115 is grown in the absence of xylose inducer (open squares), growth rapidly ceases but the cells do not lyse. This is consistent with the transformation data reported above and shows that both yvcE and lytE cannot be inactivated in B. subtilis. We then examined the phenotype of the cells from all cultures harvested at the point of the growth curve indicated by an arrow (Fig. 6A). The informative data comes from a comparison of the morphologies of strains L16638 (ΔlytE::cmr) and strain BP115 grown with and without xylose inducer. The cellular morphology of strains L16638 and BP115 (grown with 1% xylose) were very similar with regular rod-shaped cells of normal length (Fig. 6B, panels a and c). There was an even distribution of nucleoids as viewed by DAPI staining (Fig. 6B, panels b and d). The morphology of cells from strain BP115 grown without xylose inducer was very different: (i) the cells were very short (1.5 μm ± 0.33 in length compared with 3.5 μm ± 0.5 for L16638 and BP115 grown with xylose inducer), and (ii) the nucleoids almost completely fill the cylinder with some cells being devoid of nucleoids (Fig. 6B, panels e and f). Figure 6B (panel g) shows a cluster of cell images at higher magnification: cells are very short, irregularly shaped with bulging and occasional bent cylinders and some are empty (arrows). This phenotype suggested these cells were defective in elongation of the cell cylinder during growth, implying that an endopeptidase-type autolysin activity is required for an essential function in this process that can be performed either by LytE or by YvcE(CwlO). To test this hypothesis, we examined cell wall synthesis in these strains using fluorescein-labelled vancomycin, a compound that binds to the d-Ala-d-Ala residues of Lipid II thereby blocking the transpeptidation reaction of peptidoglycan synthesis (Daniel and Errington, 2003; Kahne et al., 2005). Cells of strains L16638 (ΔlytE::cmr) and strain BP115 grown with xylose inducer (Fig. 6B, panels h and i) show patches of peptidoglycan synthesis perhaps arranged in a helical pattern along the length of the cell. However, when strain BP115 is grown without xylose [i.e. depleted for both LytE and YvcE(CwlO); Fig. 6B, panel j] the intensity of labelling is greatly reduced in the truncated cell cylinders and the patches of labelling are weaker in intensity and of reduced dimension. It is concluded that one or other of the endopeptidase-type autolysins LytE or YvcE(cwlO) is essential in B. subtilis and that they play an important role in lateral cell wall synthesis and cell elongation during cell growth.
We next sought to establish whether YycFG-mediated control of yvcE and lytE expression is the sole reason for YycFG essentiality. Our strategy involved attempting to delete the yycFG genes from a strain constitutively expressing both yvcE and lytE. Strain BP113 was constructed in which yvcE is expressed from the xylose-inducible Pxyl promoter (shown in the previous paragraph to sustain growth in the presence of inducer) and the lytE gene is expressed using the IPTG-inducible Pspac promoter (PspaclytE PxylyvcE). This strain was transformed with DNA from strain BP056 (yycFG::neorlacA::PspacyycG amyE::PxylyycF) selecting for neor. As a positive control, DNA from strain BP056 was transformed into strain BP053 (lacA::PspacyycG amyE::PxylyycF) selecting for neor and hence yycFG deletion. Strains BP113 and BP053 were shown to be equally competent (data not shown). When strain BP053 was transformed with chromosomal DNA from strain BP056, 200 transformants were obtained per ng of chromosomal DNA whereas no transformants were obtained upon transformation of strain BP113 with the same amount of DNA. Therefore, we conclude that YycFG-controlled expression of yvcE and lytE is not the only reason for the essentiality of this TCS. To assess whether YycFG-controlled expression of yvcE contributes to the growth cessation and lysis phenotype upon YycFG depletion, we examined the growth profile of strain BP125 (PspacyycFGIJyyxA PxylyvcE) with and without inducers as appropriate. Only a small amelioration of cell lysis was observed, showing yvcE to be only one of a number of contributors to the growth cessation and lysis upon YycFG depletion (data not shown). To assess what other YycFG-controlled genes might contribute to this phenotype, we constructed two triple mutant strains: BP090 (ΔyvcE::spcrΔydjM::tetrΔyocH::kanr) and BP124 (lytE::CmrΔydjM::tetrΔyocH::kanr). Both strains were obtained at very low transformation frequencies (30-fold reduced in both cases) and the cells of BP090 were fragile, very short, some were devoid of nucleoids and the strain could not be propagated in liquid culture. As a strain mutated in yvcE only grows like wild type, these data show that the lowered ydjM and yocH expression levels that occur upon YycFG depletion also contribute to the growth cessation and lysis phenotype.
We observed that any strain carrying a deleted ydjM gene [BP068 (ΔydjM::tetr), BP071 (ΔyocH::kanrΔydjM::tetr) and BP083 (ΔyvcE::spcrΔydjM::tetr)] had a shiny colony morphology, a reduced growth rate in LB broth, reached stationary phase at a lowered optical density, displayed a reduced sporulation frequency and the cells were shorter than wild type during exponential growth. However, we noted that the growth defect observed during culture of strain BP068 (ΔydjM::tetr) in LB was less evident when cultured in Schaeffer's and Sterlini–Mandelstam media. The latter media contain high concentrations of Mg++ (Schaeffer et al., 1965; Nicholson and Setlow, 1990), an ion capable of restoring almost wild-type growth and morphology to mutants impaired in cell wall homeostasis (Formstone and Errington, 2005; Lazarevic et al., 2005; Carballido-Lopez et al., 2006). We found that addition of 25 mM MgCl2 to LB medium rescued the growth defect suggesting that ydjM also participates in cell wall homeostasis (data not shown). These genetic analyses show that YycFG essentiality is probably polygenic in nature, being a manifestation of disrupted cell wall metabolism caused by aberrant expression of a number of genes upon YycFG depletion. Of the YycFG regulon genes, lytE and yvcE(cwlO) encode endopeptidase-type autolysins with one or other protein being essential for lateral cell wall synthesis and cell elongation in B. subtilis. However, while contributing to YycFG essentiality, they are not its sole cause.
The YvcE and LytE autolysins participate in peptidoglycan synthesis and turnover respectively
This and previous studies show that control of cell wall metabolism is a predominant theme of the YycFG function in B. subtilis (Howell et al., 2003; 2006). Therefore, we decided to assess the contributions of YocH, YvcE(CwlO), LytE and YdjM to peptidoglycan synthesis and turnover by measuring incorporation of [14C]-N-acetylglucosamine into, and its loss from cell walls, respectively, during exponential growth between OD600 0.1 and 1.0. The results are shown in Fig. 7A. The growth rate of all cultures was similar and is represented by a single curve (symbol X, Fig. 7A). Incorporation in wild-type strain 168 (circles) increases throughout the growth period, reaching a value of 18 × 104 cpm/OD600. Similar incorporation was observed in cultures of strain L16638 (ΔlytE::cmr) and AH023 (yocH::kanr) showing that neither mutation significantly affected peptidoglycan synthesis under these conditions (data not shown). A lowered incorporation reaching 12–14 × 104 cpm/OD600 was observed in strain BP079 (squares, yvcE::spcr), a 23–33% reduction in radioactivity incorporation during this period. A similar lowered incorporation was observed in strain BP087 (yvcE::spcryocH::kanr) (data not shown). These profiles of incorporation were found to be highly reproducible for all strains during this experiment. These data show that YvcE(CwlO) functions in peptidglycan synthesis in B. subtilis.
To assess the role of these autolysins in peptidoglycan turnover during exponential growth, cells grown for two generations in [14C]-N-acetylglucosamine were harvested, washed and re-suspended in pre-warmed media containing unlabelled N-acetylglucosamine. The amount of cell-associated radioactivity and radioactivity free in the medium was then determined for each sampled point. When added together at each point, these values always totalled 100% of the radioactivity that was present at the T0 point of each experiment. The results are shown in Fig. 7B. In wild-type strain 168 (circles) there was an 80% decrease in cell-associated radioactivity during the experiment that occurred with a characteristic sigmoidal kinetic profile, previously observed by de Boer et al. (1982). A virtually identical pattern of turnover was observed for strain BP087 (triangles, yvcE::spcryocH::kanr). However, the loss of cell-associated radioactivity in strain L16638 (diamonds ΔlytE::cmr) was significantly slower (50% loss in ∼51 min) than that observed for the wild-type or BP087 (yvcE::spcryocH::kanr) strains (50% loss in ∼36 min). Neither cell wall synthesis nor turnover was affected by mutations in ydjM (data now shown). We conclude that the YvcE(CwlO) autolysin participates in cell wall synthesis and the LytE autolysin in cell wall turnover but that YocH and YdjM do not detectably affect either process under the conditions of the experiment.
YjeA deacetylates peptidoglycan, altering its susceptibility to lysozyme digestion
The finding that expression of yjeA, encoding a putative peptidoglycan deacetylase, is repressed by YycFG during exponential growth extends the complement of genes in its regulon that participate in cell wall metabolism. This was an especially interesting datum in that it addressed an observation we had repeatedly made but were unable to explain. In establishing expression of yocH–bgaB, cell lysates were always clear from cells grown under conditions of YycFG sufficiency (+IPTG) but remained turbid when prepared from cells grown under conditions of YycFG depletion (–IPTG; Howell et al., 2003). Studies have shown that deacetylation of the N-acetylglucosamine moiety of the peptidoglycan carbohydrate chain makes it more resistant to lysozyme degradation (De Las Rivas et al., 2002). We therefore hypothesized that YjeA is a peptidoglycan deacetylase and that increased yjeA expression in YycFG-depleted cells (see Fig. 1) results in the cell wall peptidoglycan becoming less acetylated and hence resistant to lysozyme digestion, the first step in the preparation of cell lysate for β-galactosidase assays. To test this hypothesis we examined the sensitivity to lysozyme digestion of cell walls prepared from wild-type strain 168, and from strain AH9912 grown with YycFG sufficiency (+IPTG) and YycFG depletion (–IPTG). Cells were harvested from the YycFG-depleted culture of strain AH9912 at the point of growth cessation (similar to that shown in Fig. S1). Lysozyme was added to purified cell wall suspensions (OD595 = 0.3 of each strain) and digestion was monitored spectrophotometrically. The results (Fig. 8A) show that cell walls prepared from a YycFG-depleted culture (open squares) are more resistant to lysozyme digestion than are cell walls from the wild-type strain 168 (open triangles) and from strain AH9912 overexpressing YycFG (open circles). To confirm that reduced acetylation of peptidoglycan in cell walls from YycFG-depleted cells was the cause of this resistance to degradation, the lysozyme digestion profile was established for each cell wall preparation after chemical acetylation using acetic anhydride (see Experimental procedures). The rate of digestion by lysozyme was increased for all three preparations (Fig. 8A, compare closed and open symbols). Of particular importance is the observation that chemical acetylation increases the sensitivity to lysozyme digestion of cell walls prepared from YycFG-depleted (–IPTG) cells, to a level that is virtually indistinguishable from the rate of digestion of acetylated cell walls prepared from wild-type strain 168 (compare closed squares and triangles, Fig. 8A).
We then sought to establish if the decreased susceptibility to lysozyme digestion of cell walls prepared from YycFG-depleted cells was due to derepression of yjeA. Cell walls were prepared from strains BP070 (ΔyjeA::kmr) and strain BP088 (ΔyjeA::kmr PspacyycFGHIJyyxA) grown under YycFG-depleted conditions (cells were harvested at the point of growth cessation) and treated with lysozyme as described. The results (Fig. 8B) show that cell walls from strain BP070 (ΔyjeA::kmr, symbol X) are degraded at approximately the same rate as that observed for wild-type cell walls (triangles). However, it is evident that cell walls prepared from a culture of strain BP088 that has a null mutation in yjeA and is depleted for YycFG (–IPTG) are no longer resistant to lysozyme but show a digestion rate that is similar to wild-type strain 168 [compare closed rectangles BP088 (–IPTG) and closed triangles (strain 168) with closed squares AH9912 (–IPTG)]. These results show that derepression of the putative peptidoglycan deacetylase YjeA in cells depleted for YycFG leads to increased resistance of the cell walls to lysozyme digestion.
We then evaluated the contribution of YycFG-controlled autolysins YvcE, LytE and YocH to cell wall autodigestion and the effect of the YjeA peptidoglycan deacetylase on endogenous cell wall-degrading activity. The results show that cell walls prepared from wild-type strain 168 show approximately 35% digestion during the 90 min incubation while cell walls prepared from YycFG-depleted cells show virtually no autodigestion during the same incubation period (data not shown). However, LytE, YvcE, YocH and YjeA do not appear to play a prominent role in autolysis (data not shown). In summary these data show that cell walls prepared from YycFG-depleted cells are more resistant to lysozyme digestion due to reduced acetylation of peptidoglycan caused by derepression of the yjeA gene. While native cell walls have considerable autodigestive capacity, cell walls from YycFG-depleted cells are highly deficient in autodigestion activity but the LytE, YvcE, YocH and YjeA proteins do not play a role in this activity.
The aim of this study was to establish the function of YycFG and to identify the constituent genes of the YycFG regulon with a view to understanding why this TCS is essential in B. subtilis. We identify five new members of the YycFG regulon that participate in cell wall metabolism: yvcE(cwlO) and lytE encode murein hydrolases, yjeA encodes a peptidoglycan deacetylase, yoeB encodes a protein that modulates autolysin activity (Salzberg and Helmann, 2007) and ydjM has a cell wall-associated function. We show that the two endopeptidase-type autolysins LytE and YvcE(CwlO) function in cell wall turnover and synthesis, respectively, that one or other of these enzymes is essential for cell growth and that they play an important role in lateral cell wall synthesis and cell elongation. Our genetic analysis reveals YycFG essentiality to be polygenic in nature, arising from aberrant expression of multiple genes that participate in cell wall synthesis and turnover in growing cells. We propose that YycG senses a cell wall- or membrane-associated signal(s) indicating that conditions are conducive to growth and that YycF∼P then activates expression of proteins that are required for the correct timing and location of cell wall synthesis and turnover.
YycFG controls expression of genes involved in cell wall metabolism in growing cells
The evidence presented here shows that the YycFG TCS controls expression of genes involved in cell wall metabolism. Three constituent genes of the B. subtilis YycFG regulon, lytE, yvcE and yocH, encode autolysins: LytE and YvcE are members of the dl-endopeptidease autolysin family II. Both enzymes have a C-terminal NlpC/P60-type peptidase domain but their N-terminal domains differ: YvcE(CwlO) has two coiled-coil domains while LytE has three LysM domains. YocH is a putative amidase with two LysM domains (Smith et al., 2000). YycF-dependent expression of all three autolysins is highest during exponential growth, but they are also expressed at lower levels post-exponentially. Our evidence indicates that YvcE(CwlO) and LytE function in lateral cell wall metabolism and cell elongation: (i) yvcE mutants have reduced cell wall synthesis during exponential growth, while cell wall turnover is unaffected. Conversely, lytE mutants have normal cell wall synthesis but have reduced cell wall turnover. (ii) There is an essential requirement for an endopeptidase-type autolysin activity in cell elongation in B. subtilis that is fulfilled by either YvcE(CwlO) or LytE. (iii) Depletion for both YvcE(CwlO) and LytE results in short cells with a bent/bumpy-type cell morphology with occasional empty cells and lateral cell wall synthesis is very low in such cells. This phenotype is very similar to that reported for cells mutated in genes that participate in lateral cell wall metabolism, e.g. mreC, mreD, mbl and lytE mreBH at low Mg2+ concentrations (Jones et al., 2001; Carballido-Lopez and Errington, 2003; Daniel and Errington, 2003; Soufo and Graumann, 2003; Leaver and Errington, 2005; Carballido-Lopez et al., 2006). Previous work localized LytE to cell poles and septa, participating in cell separation (Yamomoto et al., 2003) while the work of Carballido-Lopez et al. (2006) extended its role, showing that LytE is also positioned in a helical pattern along the cylindrical cell wall of growing cells, a distribution that requires interaction between MreBH and LytE to target the autolysin to this location. The latter study proposed that LytE is required for controlled cell wall maturation during growth, being inserted into newly synthesized peptidoglycan and participating in cell wall turnover. Our results, showing that mutation of lytE decreases cell wall turnover during growth, support their prediction. However, we can extend these results, showing that an endopeptidase-type murein hydrolase activity is an essential requirement for cell growth and that such an activity plays an important role in lateral cell wall synthesis and cell elongation that is executed either by LytE or by YvcE(CwlO). This study and the studies of Carballido-Lopez et al. (2006) and others provide insight into the requirements and mechanism of lateral cell wall synthesis and cell elongation during growth. Current models propose a dispersed pattern of peptidoglycan insertion into the elongating cell wall (for reviews see Scheffers and Pinho, 2005; Carballido-Lopez, 2006). LytE is helically located along the cell cylinder and interacts with MreBH (Carballido-Lopez et al., 2006). We hypothesize that YvcE(CwlO) has a similar location and interacts perhaps with one of the other actin isoforms MreB or Mbl in B. subtilis (a model under investigation). Thus, one or other of these autolysins provides an essential mureinolytic endopeptidase activity that is required for insertion of new peptidoglycan units along the cell cylinder thereby allowing cell expansion. Further insight into these processes may be provided by our observation that native cell walls prepared from growing cells have significant autodigestive capability, while those prepared from YycFG-depleted cells harvested at the point of growth cessation are highly deficient in such activities. Surprisingly, mutation of YocH, YvcE(CwlO), LytE or YjeA did not decrease the autodigestion capability, prompting us to investigate whether there are other autolysin-encoding genes with YycF-activated promoters. None were identified among the 45 autolysins listed in the review of B. subtilis autolysins by Smith et al. (2000), consistent with our array analysis. There are three possible explanations for this paradox: (i) it is possible that additional YycFG-controlled autolysin or peptidoglycan modifying encoding genes will be identified in B. subtilis, (ii) the autolytic activity of YycFG-depleted cell walls may be inhibited by the elevated levels of YoeB that occur under these conditions, as Helmann and colleagues have shown that YoeB modulates autolysin activity and protects cells from autolysis (Salzberg and Helmann, 2007), and (iii) depletion of YycFG-controlled cell wall metabolic activities might generate a form of peptidoglycan that is refractory to digestion by other autolysins. This latter model predicts hierarchies of autolytic activity where a particular autolysin would generate appropriate substrates for other autolysins. Spatial restriction of autolytic activity can also be achieved as autolysins would only be active at locations where appropriate substrates were generated.
Two other members of the YycFG regulon, ydjM and yjeA, are noteworthy in the context of cell wall metabolism. YdjM is a protein of unknown function whose expression is activated by YycF and is highest during the exponential period of the growth cycle. We propose that it participates in cell wall metabolism as the ydjM mutant phenotype can be partially suppressed by addition of 10 mM Mg2+, a feature noted for other cell wall mutants (Carballido-Lopez et al., 2006 and references therein). We are unable to detect tagged proteins by Western or fluorescence microscopy so either YdjM is produced in low amounts or the tagged protein is unstable (D. Noone and K.M. Devine, unpubl. results). YjeA encodes a peptidoglycan deacetylase. Expression of yjeA is low during exponential growth of the wild-type strain, but increases in cells depleted for YycFG and is therefore negatively regulated by this TCS. Here we show that cell walls prepared from YycFG-depleted cells display an increased resistance to lysozyme degradation that is dependent on YjeA activity, consistent with the yjeA expression profile, the location of the protein and the known properties of peptidoglycan deacetylases (Eymann et al., 2004; Tjalsma and van Dijl, 2005). We did not observe an effect of YjeA on autodigestion by endogenous autolysins and its precise role in controlling autolysin activity in B. subtilis remains to be established. However, in the context of YycFG function, it further exemplifies the role of this TCS in controlling cell wall metabolism, in this case controlling autolysin activity indirectly by modifying peptidoglycan. In summary YycFG is a regulator of cell wall metabolism, with a complete listing of the constituent genes of its regulon presented in Table 3. It controls expression of autolysins [YocH, YkvT, YvcE(CwlO) and LytE], proteins that modify autolysin activity (YjeA and YoeB), teichoic acid biosynthesis (TagAD) and other cell wall-associated proteins (YdjM).
We sought to establish the reason for YycFG essentiality in B. subtilis. As all YycFG-regulated genes can be individually mutated, it was evident that essentiality must be polygenic (this study; Kobayashi et al., 2003). Here we establish that YycFG controls expression of LytE and YvcE(CwlO), the functions of which partially overlap as one or other protein is essential for cell growth. The level of lytE transcript at the point of growth cessation upon YycFG depletion is not lower than that seen in wild-type cells (Fig. 5B; although it does decrease subsequently), nor can YycFG be deleted from cells constitutively expressing LytE and YvcE(CwlO) showing that YycFG-controlled expression of yvcE and lytE is not the sole reason for its essentiality. However, the similarity in phenotype of YycFG-depleted and YvcE LytE-depleted cells suggests that lowered levels of these autolysins contributes significantly to the essentiality of YycFG. The complexity of YycFG essentiality is underscored by additional observations: (i) cells depleted for YvcE and LytE cease growth but do not lyse, as do YycFG-depleted cells, (ii) ydjM mutants also have a cell wall-associated defect, and (iii) when a lytE or yvcE mutation was combined with both yocH and ydjM mutations, both triple mutant combinations were obtained at extremely low frequency and had severe growth and morphological defects. Therefore, we conclude that YycFG essentiality is polygenic in nature arising from abnormal cell wall metabolism due to aberrant expression of multiple genes under YycFG control.
YycF-dependent expression of ydjM derives from an unusual promoter structure
Many of the SigA-type promoters that are activated by YycF have unusual features. In addition to the YycF binding sequence, there is a canonical SigA-type −10 motif present in the yvcE, ydjM and lytE promoters but the sequence at the −35 position has little resemblance to a canonical SigA-type −35 motif. However, there are putative −35 motifs positioned within the promoters of ydjM and yvcE (23 and 11 bp from the −10 motif respectively). Our results show that ydjM expression is almost completely abolished when the YycF binding motifs are mutated. A new expression profile, with higher expression levels, is obtained upon deletion of 6 bp that generates a canonical 17 bp spacing between the SigA-type −35 and −10 motifs of the mutated promoter. However, ydjM expression is now constitutive throughout the growth cycle and importantly is independent of YycF. We conclude that YycF-dependent expression of this promoter requires specific positioning of the YycF binding motif within the promoter structure. Although the putative SigA-type −35 motif is functional when normally placed within this promoter, it is not clear whether it plays a role in expression under normal conditions. A precedent exists in B. subtilis for a promoter with such a structure. The SigA-type promoters of spoIIE and spoIIG are activated by Spo0A∼P, have a SigA-type −10 motif but no recognizable −35 motif at the expected position while having putative −35 motifs at unusual positions (21 and 22 nucleotides, respectively, from the −10 motif). Expression of these promoters is dependent on phosphorylation of Spo0A that increases its affinity for weak ‘OA’ boxes located within the promoter (Kenney et al., 1989; Satola et al., 1991; York et al., 1992; Baldus et al., 1995). This mechanism restricts expression of these promoters to periods when Spo0A∼P levels are high in the cell. As activation of gene expression by YycF is most evident during the exponential period of the growth cycle, it will be interesting to establish whether this occurs by a similar mechanism.
Cell wall metabolism, cell wall stress and the nature of the signal sensed by the YycG sensor kinase
Prompted by our finding that YycFG controls expression of genes involved in cell wall metabolism, we sought to investigate a possible relationship between cell wall stress and YycFG depletion. Helmann and colleagues report the transcriptional changes that occur when B. subtilis cells are treated with bacitracin and vancomycin, antibiotics that induce cell wall stress by interfering with peptidoglycan synthesis (Cao et al., 2002; Mascher et al., 2003; for review see Bhavsar and Brown, 2006). Upon vancomycin treatment, yoeB and yjeA are among the most highly induced genes (9th and 12th respectively) while yvcE and ydjM are the genes most highly repressed with yocH expression being reduced to a lesser extent (see complete array results for the two studies at http://www.micro.cornell.edu/faculty.JHelmann.html). These profiles mirror those we observe upon YycFG depletion. The correlation between the transcriptional responses to vancomycin treatment and YycFG depletion supports our central finding that YycFG regulates cell wall metabolism during exponential growth and shows that aberrant expression of its regulon leads to cell wall stress. Thus, we propose that the signal being sensed by YycG emanates from normal cell wall metabolism and that it is the antithesis of cell wall stress. This is consistent with the fact that YycF-dependent activation of gene expression occurs primarily during the exponential growth period (this study; Howell et al., 2003; 2006). What is the nature of the signal detected by YycG? A clue might come from the different transcriptional responses to vancomycin and bacitracin: the transcriptional profile upon vancomycin but not bacitracin treatment mirrors that of YycFG depletion (Cao et al., 2002; Mascher et al., 2003). While vancomycin inhibits transpeptidation, perhaps leading to an accumulation of Lipid II intermediates extracytopasmically, bacitracin inhibits recycling of Lipid I thereby depleting the cell of Lipid II intermediate. As Lipid II is the only intermediate of the Lipid I/Lipid II cycle that is located in the same cellular compartment as the YycG sensing domain (the 154-amino-acid loop), we hypothesize that YycG is sensing some aspect (i.e. accumulation or incorporation into peptidoglycan) of Lipid II under normal growth conditions. The studies of Hoch and colleagues would support this model, showing that YycH and YycI are located extracytoplasmically, functioning to regulate the activation of the YycG kinase (Szurmant et al., 2005; 2006; 2007). In view of our finding that DesKR is activated upon YycG depletion (shown by increased expression of des, see Fig. 1), it is interesting that vancomycin treatment also leads to activation of cell membrane-associated regulators such as the TCS (LiaRS, YvqPQ and BceRS) and ECF-type sigma factors (SigW, SigX and SigM) (Cao et al., 2002; Mascher et al., 2003; 2004; Cao and Helmann, 2004; Jordan et al., 2006). Thus, there may also be a membrane-associated aspect to the signal being sensed by YycG. In this regard, it is noteworthy that a linkage between YycFG and the cytoplasmic membrane has been observed in other bacteria, although the mechanism is not yet established (Martin et al., 1999; Mohedano et al., 2005; Ng et al., 2005).
Bacterial strains, plasmids and growth conditions
The bacterial strains and plasmids used in this study are listed in Table 4. Details of strain construction are presented in Supplementary material. Strain TG1 was used for routine cloning in Escherichia coli (Gibson, 1984). B. subtilis strain 168 trpC2 was used throughout to generate mutant strains and to establish expression patterns using transcriptional fusions. Strains were grown in LB medium (Miller, 1972). IPTG was added at concentrations specified in the text while antibiotics were added to cultures at the following concentrations per ml: ampicillin 100 μg; tetracycline 13 μg; kanamycin 10 μg; spectinomycin 100 μg; erythromycin 3 μg; chloramphenicol 3 μg.
pDL derivative containing the yjeA promoter region (Apr Cmr)
pDL derivative containing the yjeA1 promoter region (Apr Cmr)
pDL derivative containing the yvcE promoter region (Apr Cmr)
pDL derivative containing the yvcE1 promoter region (Apr Cmr)
pDL derivative containing the ydjM promoter region (Apr Cmr)
pDG1515 derivative for deletion of the ydjM1 locus
pDG1727 derivative for deletion of the yvcE locus
pDG780 derivative for deletion of the yjeA locus
pDL derivative containing the ydjM1 promoter region (Apr Cmr)
pXT derivative for inducible expression of the yvcE gene
pMUTIN4 derivative for inducible expression of the lytE gene
pHTxyl derivative for inducible expression of the yvcE gene
pDL derivative containing the yoeB1 promoter region (Apr Cmr)
pDL derivative containing the yoeB2 promoter region (Apr Cmr)
pDL derivative containing the ydjM2 promoter region (Apr Cmr)
pDL derivative containing the ydjM2 promoter region (Apr Cmr)
DNA manipulations were carried out by standard procedures as described by Sambrook et al. (1989). Strand overlap extension PCR was carried out as described by Horton et al. (1989). The oligonucleotides used in this study are listed in Table S2.
Northern and primer extension analysis
Northern and primer extension analysis was carried out as described in Noone et al. (2000). Twenty-five micrograms of total RNA was used in each reaction. Probes for northern analysis were synthesized by PCR using primer pairs yoeB PROBE-1/yoeB PROBE-2 (for yoeB), ykuP PROBE-1/ykuP PROBE-2 (for ykuP), des PM−1/des PROBE-2 (for des), ydjM PROBE-1/ydjM PROBE-2 (for ydjM), yvcE PROBE-1/yvcE PROBE-2 (for yvcE) and LYTEF1/LYTER1 (for lytE). The yocH probe was as described in Howell et al. (2003). Primers yvcEPE1, yvcEPE2, ydjMPE1 and ydjMPE2 were used for primer extension from 168 wild-type total RNA and primers yoeBGSR, yoeBPE1, yjeAGSR and oBP55 were used for primer extension analysis of total RNA isolated from strain AH9912 grown in the absence of IPTG.
The Affymetrix (Affymetrix, USA) B. subtilis Genome Array (antisense) was used for transcriptome analysis of strain AH9912 grown under conditions of YycFG repletion (1 mM IPTG) and depletion (no IPTG). A culture of strain AH9912 grown in 100 μM IPTG was used to inoculate two LB cultures, one containing 1 mM IPTG and the other without IPTG. A culture of wild-type strain 168 B. subtilis was similarly prepared as a control. Cells were harvested from the three cultures throughout the following 2.5 h growth cycle and frozen in liquid nitrogen as described by Eymann et al. (2002). The point of growth cessation of the AH9912 culture grown without IPTG was determined from the growth curve. Then total RNA was prepared from B. subtilis strain 168 and from strain AH9912 grown in the presence of 1 mM IPTG and in the absence of IPTG at this point of the growth cycle of all three cultures and processed according to the method of Eymann et al. (2002). Four separate experiments were performed yielding four total RNA preparations for AH9912 cells grown without IPTG and with IPTG and for wild-type strain 168. The quality of each RNA preparation and the extent to which the cells were YycFG-depleted were estimated using gel electrophoresis and Northern analysis, probing for yocH transcript levels respectively. Three RNA preparations from cells grown under each condition (i.e. + and –IPTG) were chosen for use as probes in the microarray analysis. Biotinylated antisense RNA probe was prepared using the MessageAmpTM II-Bacteria Kit (Ambion) incorporating biotin-UTP and biotin-CTP. This was used to hybridize to the B. subtilis Affymetrix microarrays using standard procedures described for prokaryotic arrays. The arrays were washed in the Affymetrix GeneChip fluidics station 450 and scanned using the GeneChip 3000 scanner. The raw probe intensities were normalized using the non-linear normalization method, Qspline (Workman et al., 2002), and the logit-t method was applied to determine differentially expressed genes between the two categories (Lemon et al., 2003). The gene expression index values used to determine the fold changes were calculated by the use of the method developed by Li and Wong (2001), after applying the Robust Multi-arrays Average (RMA) method for background correction to the raw Perfect Match (PM) probe intensities (Irizarry et al., 2003), and normalizing these using Qspline (Workman et al., 2002).
Measurement of incorporation of 14C-labelled N-acetylglucosamine into B. subtilis cell walls
A B. subtilis culture was grown in LB medium at 37°C and 120 r.p.m. until the OD600 had reached a value of 0.5. A 2 ml aliquot of this culture was then added to 8 ml of pre-warmed LB medium containing 20 μl of 14C-labelled N-acetylglucosamine (7.4 MBq ml−1; Amersham) and 12.8 μl of unlabelled N-acetylglucosamine (5 ng ml−1; Sigma Chemical) to give a final concentration of 6.4 mg l−1. Growth was monitored turbidimetrically at OD600 and 400 μl of samples of the cell suspension were harvested in duplicate every 10 min, transferred to microcentrifuge tubes containing 400 μl of 0.1% sodium dodecylsulphate (SDS), pelleted by centrifugation at 15 000 r.p.m. for 10 min and the supernatant was discarded. The cell pellets were re-suspended in 400 μl of distilled water and mixed with 5 ml of scintillation liquid (Ecoscint A, National Diagnostic) and the radioactivity was measured using a liquid scintillation counter (Packard Tri-Carb). To calculate incorporation of 14C-labelled N-acetylglucosamine, the amount of radioactivity present in the T0 samples was subtracted from the value obtained for each sample.
Measurement of cell wall turnover of B. subtilis cell walls
In order to obtain radiolabelled cell walls, 1 ml of an exponentially growing culture (OD600 0.25) was added to 9 ml of pre-warmed LB medium containing 20 μl of 14C-labelled N-acetylglucosamine (7.4 MBq ml−1; Amersham) and incubated at 37°C, 120 r.p.m. for a time corresponding to two generations (44 min for all strains with the exception of strain BP068, for which cells were incubated for 60 min). Cells were pelleted by centrifugation at 6000 r.p.m. for 5 min at 25°C, and washed twice with pre-warmed LB medium. The cells were then re-suspended in 10 ml of pre-warmed LB containing 100 mg l−1 cold N-acetylglucosamine and incubated at 37°C, shaking at 120 r.p.m. Growth was monitored turbidimetrically at OD600 and 400 μl of samples of the cell suspension were taken in duplicate every 10 min and transferred into microcentrifuge tubes containing 400 μl of 0.1% SDS. Cells were pelleted by centrifugation at 15 000 r.p.m. for 10 min and the radioactivity of both the supernatant and the pellet was measured as described above.
The YycF and ′YycG proteins was purified according to the procedure described in Howell et al. (2006).
Gel mobility shift DNA binding assays
DNA fragments spanning the promoter regions of yjeA (485 bp, YjeAbELF1/YjeAELR1), ydjM (419 bp, YdjMbiotin/YdjMGSR), yvcE (385 bp, YvcEbiotin/YvcEGSR), lytE (268 bp, oBP177, oBP165biotin), yoeB (414 bp, YoeBbiotin/YoeBGSR) and htrA (300 bp, htrA biotin/htrAGSR) were generated by PCR using Phusion polymerase (NEB) and the indicated biotinylated oligonucleotide pairs. Biotinylated DNA fragments were gel purified and binding reactions, electrophoresis and detection were performed essentially as described (Howell et al., 2006), with the following modifications: 6% native polyacrylamide gels with a 80:1 acrylamide : bisacrylamide ratio were used with a Tris-acetate buffer system. All reactions contained ′YycG and were performed in the presence of poly [d(I-C)]. An aliquot of 2 ng of biotinylated promoter fragment was used and where appropriate 100× excess of unlabelled competitor DNA.
Purification, acetylation and lysozyme digestion of B. subtilis cell walls
Four hundred millilitres of cultures of B. subtilis strainsAH9912 and BP088 (both have the yycFGHIJyyxA operon under the control of the IPTG-inducible Pspac promoter) were grown in the presence and the absence of IPTG. In YycFG-depleted cultures (–IPTG), cells were harvested at the point of growth cessation while in YycFG-replete cultures (1 mM IPTG) cells were harvested at the same OD595 as the YycFG-depleted ones (in both cases OD595 ∼ 0.8). Cells were pelleted by centrifugation at 16 100 g for 5 min at 4°C and washed twice in 10 volumes of 10 mM sodium phosphate buffer (pH 7). Cell suspensions were transferred to 2 ml screw cap tubes containing 0.5 g of glass beads (Biospec; Bartlesville, OK, USA) and broken by vigorous shaking in a Fastprep bead beater (Bio101, Biospec; Bartlesville, OK, USA) for 45 s at speed 6.5. The process was repeated 10 times and the tubes were cooled on ice for 2 min in between. Unbroken cells and glass debris were removed by centrifugation at 5000 r.p.m. for 2 min at 4°C and the supernatant was transferred to new tubes. Cell wall fragments were pelleted by centrifugation at 16 100 g for 5 min at 4°C, re-suspended in 2% SDS and incubated twice for 5 min at 100°C in 2% SDS to remove proteins. Deproteinized cell walls were then washed twice in distilled water, twice in 10 mM sodium phosphate buffer (pH 7) and twice in distilled water and stored at −20°C. Cell walls were acetylated as described by Vollmer and Tomasz (2000). To determine lysozyme sensitivity, a suspension of cell wall (both non-acetylated and acetylated) was made in 10 mM Tris/HCl pH 8 and the turbidity was adjusted to OD595 = 0.3. Lysozyme was added at a final concentration of 20 μg ml−1 and digestion was estimated by measuring the decrease in OD595 at 25°C. Preparations without added lysozyme were used as negative controls. Native cell walls were prepared essentially as described above but the boiling step in 2% SDS was omitted. Native cell walls were suspended in 10 mM Tris/HCl to OD595 = 0.3, incubated at 37°C and the decrease in optical density was measured at 595 nm.
Cell samples for microscopic analysis were pelleted by centrifugation and fixed in 3% paraformaldehyde (in Spizizen's minimal salts, pH 7.5) for 10 min on ice. Cells were washed twice in Spizizen's minimal salts containing 15 mM sodium azide and mounted onto glass slides covered with 1.2% agarose in water. To visualize the nucleoids, 10 μl of DAPI (0.1 μg ml−1 in 50% v/v glycerol; Sigma) was added to the samples and incubated for 5 min in the dark before viewing. To visualize newly synthesized peptidoglycan, cells were grown in the presence of unlabelled and fluorescein-labelled vancomycin for 15 min (Molecular Probes) and prepared for microscopic examination as described by Daniel and Errington (2003). Images were acquired using an Olympus Fluoview 100 Confocal Microscope and analysed using Fluoview software version 1.3.
This work was supported by an SFI Investigator Programme grant (03/IN3/B409), by Health Research Board (Grant RP/101/2002) and by Enterprise Ireland (Grant SC/02/109) to Kevin M. Devine. The authors would like to thank H. Bjorn Nielsen, Simon Rasmussen and Rune Vejen Petersen for discussion of the array analysis, Orla Hanrahan, Petrina Delivani and Seamus Martin for assistance with confocal fluorescence microscopy and Inga Jende for purified proteins.