The YycG/YycF two-component system, originally identified in Bacillus subtilis, is very highly conserved and appears to be specific to low G + C Gram-positive bacteria. This system is required for cell viability, although the basis for this and the nature of the YycF regulon remained elusive. Using a combined hybrid regulator/transcriptome approach involving the inducible expression of a PhoP′-′YycF chimerical protein in B. subtilis, we have shown that expression of yocH, which encodes a potential autolysin, is specifically activated by YycF. Gel mobility shift and DNase I footprinting assays were used to show direct binding in vitro of purified YycF to the regulatory regions of yocH as well as ftsAZ, previously reported to be controlled by YycF. Nucleotide sequence analysis and site-directed mutagenesis allowed us to define a potential consensus recognition sequence for the YycF response regulator, composed of two direct repeats: 5′-TGT A/T A A/T/C-N5-TGT A/T A A/T/C-3′. A DNA-motif analysis indicates that there are potentially up to 10 genes within the B. subtilis YycG/YycF regulon, mainly involved in cell wall metabolism and membrane protein synthesis. Among these, YycF was shown to bind directly to the region upstream from the ykvT gene, encoding a potential cell wall hydrolase, and the intergenic region of the tagAB/tagDEF divergon, encoding essential components of teichoic acid biosynthesis. Definition of a potential YycF recognition sequence allowed us to identify likely members of the YycF regulon in other low G + C Gram-positive bacteria, including several pathogens such as Listeria monocytogenes, Staphylococcus aureus and Streptococcus pneumoniae.
Bacteria are characterized by their ability to respond and adapt rapidly to changes in environmental and nutritional conditions. This signal transduction capacity, allowing bacteria to integrate diverse signals and trigger an appropriate response, often involves two-component systems (TCSs), composed of a sensor kinase and a response regulator (Hoch and Silhavy, 1995). The sensor kinase is composed of two domains. The sensing (input) domain often has multiple membrane spanning regions and is widely divergent among TCSs, reflecting the variety of signals to which these systems can respond. The kinase (transmitter) domain is highly conserved, and phosphorylates the cognate response regulator upon activation by a specific stimulus. The response regulator is also composed of two domains: a highly conserved receiver domain implicated in phosphotransfer, and an output domain, involved in DNA-binding and transcription activation. Phosphorylation of the response regulator therefore effects global changes in gene expression that are appropriate to the detected stimulus.
The high degree of conservation among TCSs and the fact that they are ubiquitous among bacteria has made them an attractive target for novel classes of antimicrobial compounds (Barrett and Hoch, 1998). Although TCSs constitute one of the largest known families of transcriptional regulators in bacteria, few to date have been shown to be essential for cell viability. Among these, DivJ/DivK (Ohta et al., 1992; Hecht et al., 1995) and CckA/CtrA (Quon et al., 1996; Jacobs et al., 1999) interact in a multicomponent regulatory pathway to control cell cycle regulation, cell division and DNA replication in Caulobacter crescentus (Wu et al., 1998). The MtrB/MtrA TCS has been shown to be essential in Mycobacterium tuberculosis, but its function remains unknown (Zahrt and Deretic, 2000). Another system, YycG/YycF, is a member of the EnvZ/OmpR TCS family, but appears to be specific to low G + C Gram-positive bacteria. It is the most highly conserved TCS in these bacteria, with YycF response regulators sharing 70% amino acid sequence identity on average. The YycG/YycF system is present in all the low G + C Gram-positive genomes whose sequence is available, and has been reported to be essential for cell growth in Bacillus subtilis, Staphylococcus aureus, Streptococcus pyogenes, Streptococcus pneumoniae and Listeria monocytogenes (Fabret and Hoch, 1998; Lange et al., 1999; Martin et al., 1999; Federle et al., 1999; Throup et al., 2000; Fukuchi et al., 2000; Kallipolitis and Ingmer, 2001). However, the basis for this, the signal to which YycG/YycF responds and the nature of the YycF regulon remain unknown. Although conventional genetic approaches have so far been hampered by the essential nature of the YycG/YycF system, this has made it a prime target for the development of novel antibacterial compounds (Yamamoto et al., 2001). YycF was shown to be involved in competence development under microaerobic conditions in S. pneumoniae (Echenique and Trombe, 2001), is thought to be involved in resistance to MLS antibiotics in S. aureus via regulation of ssa expression (Martin et al., 2002), and plays a role in activating transcription from the nonessential P1 promoter of the B. subtilis ftsAZ operon (Fukuchi et al., 2000), although neither of these functions explain its essential nature.
More recently, a systematic approach using high-density microarrays and transcriptome analysis to identify regulons controlled by the entire complement of TCSs in B. subtilis was undertaken (Kobayashi et al., 2001; Ogura et al., 2001). This involved the use of strains in which the response regulator is overproduced in the absence of its cognate kinase, which could not be applied to study the YycG/YycF system due to the essential nature of the yycG gene (Kobayashi et al., 2001).
A new approach, potentially allowing systematic investigation of any TCS of unknown function, is presented in this paper. This ‘domain-swapping’ method involves the generation of hybrid response regulators, based on the modular nature of these proteins (Parkinson, 1995), by fusing the receiver domain from a response regulator of known function with an output (DNA-binding) domain from a regulator of unknown function. In response to the appropriate stimulus, the sensor kinase of the known system should phosphorylate its cognate receiver domain in the hybrid response regulator protein, which can then bind to the promoters of genes that constitute the regulon of the TCS of unknown function. Here we report the successful use of this approach to investigate the essential YycG/YycF two-component system of unknown function in B. subtilis, using PhoR/PhoP as the system of known function. Through this analysis we have established the recognition sequence for the YycF response regulator and identified potential members of the YycF regulon in B. subtilis and other Gram-positive bacteria, including genes involved in cell division, cell wall metabolism and membrane-bound transport systems.
A combined hybrid regulator/transcriptome approach to identify members of the YycG/YycF regulon
In order to identify members of the YycG/YycF regulon in B. subtilis, a novel approach was developed using a hybrid response regulator as presented in Fig. 1. The principle is to generate a chimerical response regulator comprising the receiver domain from a TCS of known function (i.e. one whose stimulus and regulon are well characterized) fused to the output domain of a TCS of unknown function. When the appropriate signal is applied, the sensor kinase of the known TCS should recognize and phosphorylate the receiver domain of the hybrid response regulator protein. The activated hybrid regulator can then bind and activate the expression of genes controlled by the TCS of unknown function. The choice of TCSs to be used in this approach was mandated by the following considerations: (i) the activating stimulus of the characterized TCS should be easily manipulated; (ii) the response should be rapidly turned on when the activating stimulus is applied and (iii) the TCSs of known and unknown function should be closely related to maximize the chance of generating a functional hybrid response regulator. Based on these criteria, the well-characterized PhoR/PhoP system (Hulett, 2002) was chosen as the TCS of known function as phylogenetic analysis indicates it is the one most closely related to YycG/YycF in B. subtilis.
The amino-terminal domains of PhoP and YycF are very similar, with nine identical residues out of 13 in α helix 5 of PhoP, known to play an important role in interaction with the carboxy-terminal domain (Allen et al., 2001; Birck et al., 2003; Chen et al., 2003). Gene fusion by strand overlap extension (SOE-PCR; Horton et al., 1989) was used to create a hybrid phoP′-′yycF gene, where codons M1 to R120 are from phoP, encoding the entire receiver domain, and codons Q121 to D236 are from yycF, encoding the linker region and the output domain (see Experimental procedures). The hybrid gene was then cloned under the control of the PxylA xylose-inducible promoter using plasmid pXT (Derréet al., 2000), allowing integration as a single copy at the thrC locus of B. subtilis (see Experimental procedures).
Bacillus subtilis strain AH014 carries the phoP′-′yycF gene at the thrC locus and a deletion in the phoP gene to prevent PhoR/PhoP regulon induction upon phosphate limitation (see Table 1). This simplifies the pattern of changes in gene expression that should result from activation of the hybrid PhoP′-′YycF response regulator by the PhoR sensor kinase in response to phosphate limitation. Strain AH014 also contains the wild-type yycFG locus as it is essential. Bacteria were grown in low phosphate defined medium (LPDM) with or without 1% xylose to induce expression of the phoP′-′yycF hybrid regulator gene. Cells were harvested 1, 4 or 5 h after the onset of phosphate limitation and RNA was extracted. RNA preparations were reverse transcribed and used to probe GenoSys Panorama macroarray filters carrying the full gene complement of the B. subtilis genome. A single gene, yocH, was consistently found to be induced more than fourfold in the presence of the hybrid PhoP′-′YycF (data not shown). The yocH gene product has not been characterized but is thought to be a likely autolysin (Smith et al., 2000).
pXT derivative containing the hybrid phoP′-′yycFresponse regulator gene
pXT derivative containing the hybrid yycF′-′phoPresponse regulator gene
pMUTIN4 derivative for inducible expression of the yycF operon
pDG646 derivative for deletion of the phoPR locus
Northern analysis of yocH expression in a strain producing the PhoP′-′YycF hybrid regulator
To verify that the differential induction of yocH is specifically mediated by the hybrid PhoP′-′YycF response regulator, Northern analysis was performed on RNA samples prepared from strains AH014 and AH029 (phoR– derivative of AH014) grown in LPDM in the presence or absence of xylose (Fig. 2). Because the endogenous YycG/YycF system is still present in this strain, activation by the PhoP′-′YycF hybrid regulator will only be observed as increased or persistent gene expression. Indeed, the levels of yocH transcript are significantly higher after the onset of phosphate starvation in cells from the culture containing xylose (expressing the hybrid PhoP′-′YycF response regulator) when compared with the culture lacking xylose (Fig. 2, upper panel). Transcription of yocH continues to persist, following phosphate starvation, in cells expressing the PhoP′-′YycF hybrid response regulator (T135, T165), whereas it falls to a low and barely detectable level at the corresponding time points in cells in which phoP′-′yycF is not induced.
To verify that reduced yocH transcript levels observed at T135 and T165 in cells growing without xylose were not due to RNA degradation, the filter was stripped of the yocH probe and rehybridized with a yxiE probe. Expression of yxiE is known to be induced by phosphate starvation in a SigB-dependent manner (Antelmann et al., 2000) and would therefore be expected to be present at similar levels in both cultures. As shown in Fig. 2 (upper panel), yxiE is induced approximately 75 min after phosphate limitation and the transcript levels are approximately the same in both cultures at T135 and T165 indicating that the RNA has not been degraded in these samples. These combined macroarray and Northern analyses show that yocH expression is positively regulated by the PhoP′-′YycF hybrid regulator. This increased and persistent expression of yocH in the presence of the PhoP′-′YycF regulator was not observed in the AH029 strain lacking the phoR gene, indicating that phosphorylation of the PhoP′-′YycF hybrid regulator by the PhoR kinase in response to phosphate limitation is required (Fig. 2, lower panel).
Levels of yocH transcript are increased when the yycFG genes are overexpressed
Because the PhoP′-′YycF hybrid response regulator is an artificial construct, we sought to establish whether yocH expression was indeed regulated by the natural YycG/YycF TCS. In strain AH9912, the yycFG operon was placed under the control of the IPTG-inducible Pspac promoter. Strain AH9912 was grown in LB medium containing no IPTG, 100 µM or 1 mM IPTG. In the absence of IPTG, cells grew to an OD600 of approximately 0.1 and subsequently lysed, but grew normally and at similar rates in the presence of 75 µM, 100 µM or 1 mM IPTG (Fig. 3A).
Cells were harvested from the growing cultures at different times, RNA was extracted and the abundance of yocH transcript established by Northern analysis. As shown in Fig. 3B, the level of yocH transcript for the culture containing 100 µM IPTG is low at T-60, increases at T-40 and T-20 and decreases to a low level at T0, the transition phase. In cultures grown with 1 mM IPTG, the level of yocH transcript is dramatically increased at T-60 and T-40. These data indicate that yocH is regulated by YycG/YycF. Interestingly, despite the continued high level of IPTG in this culture, the yocH transcript level at T-20 and T0 is greatly reduced compared to the earlier time-points. This stationary phase shut-off of yocH transcription suggests that despite increased levels of YycG/YycF, the TCS is no longer activated as cells reach the transition phase and that the specific signal to which it responds may be restricted to the exponential phase.
The yocH transcription start site was determined by primer extension analysis (data not shown) and is indicated in Fig. 6C. The best SigA-type promoter corresponding to this initiation point of transcription has a − 35 region (TTGATT) and a − 10 region (TACGAT) with an unusual 16 nucleotide spacing between them.
Purification and phosphorylation of ‘YycG and YycF
As shown above, yocH expression is controlled by the YycG/YycF two-component system. In order to determine whether this regulation is direct, the YycG and YycF proteins were overproduced and purified. The yycF coding sequence was cloned in plasmid pET28/16 creating a translational fusion adding six histidine residues to the carboxy-terminus of YycF and placing the gene under the control of an inducible T7 bacteriophage promoter (see Experimental procedures). In order to overproduce ‘YycG, a DNA fragment encoding the isolated cytoplasmic histidine kinase domain was cloned in pET28/16, adding six histidine residues to the carboxy-terminus of the protein (see Experimental procedures). SDS-PAGE analysis showed overproduced bands in the soluble fraction of crude extracts from cells carrying pETYycF or pETYycG, which were absent in the control extract from cells carrying the pET28/16 vector alone (Fig. 4A, lanes 3, 5 and 2 respectively). The overproduced band corresponding to ‘YycG displayed the expected apparent molecular mass (approximately 46 kDa, Fig. 4A, lane 6), whereas the recombinant YycF protein migrated slightly higher than the expected size (approximately 35 kDa instead of the deduced molecular mass of 28 kDa) (Fig. 4A, lane 4). Such anomalous electrophoretic migration has previously been reported for other response regulators such as DegU (Kunst et al., 1988; Dahl et al., 1991). His-tagged YycF and ‘YycG were purified in a single step using a Ni-NTA agarose column (see Experimental procedures) and SDS-PAGE analysis revealed a purity greater than 95% (Fig. 4A, lanes 4 and 6 respectively).
Autophosphorylation of the ‘YycG histidine kinase was demonstrated by incubating the protein with [γ-32P]-ATP, followed by SDS-PAGE and autoradiography, as shown in Fig. 4B, lane 1. When the YycF and ‘YycG purified proteins were incubated together, autophosphorylation of ‘YycG and phosphotransfer to YycF was observed upon addition of [γ-32P]-ATP (Fig. 4B, lanes 2–5). Phosphorylation of YycF by ‘YycG was rapid and efficient, occurring in less than one minute (Fig. 4B, lane 2). This contrasts a recent report where the authors were unable to demonstrate in vitro phosphorylation of the orthologous proteins from S. pneumoniae (Wagner et al., 2002).
Identification of the YycF recognition sequence
In order to determine whether YycF binds directly to the yocH promoter region, an in vitro approach based on gel mobility shift DNA-binding assays and DNase I footprinting was performed. The purified YycF protein was used in gel mobility shift DNA-binding assays with a 314 bp radiolabelled DNA fragment, generated by PCR with oligonucleotides OSA7/OSA8 and corresponding to the yocH promoter region (positions − 234 to + 79 with respect to the transcription initiation site), in the presence of an excess of non-specific competitor DNA [1 µg of poly (dI-dC)]. YycF bound specifically, forming one major protein/DNA complex with the yocH DNA fragment (Fig. 5A, lanes 1 and 2). The specificity of this DNA binding was checked by competition assays. Two different unlabelled fragments were tested for their capacity to compete with the labelled yocH promoter region DNA fragment for YycF binding. The specific DNA fragment was identical to the radiolabelled probe described above. As shown in Fig. 5A an 80-fold excess of unlabelled specific competitor DNA fully prevents the formation of the YycF/yocH complex. A non-specific DNA competitor fragment synthesized by PCR with primers OSA52/OSA53 (see Table 2), corresponding to a 310 bp region located just downstream from the yycF promoter region (positions − 20 to + 290 with respect to the transcription initiation site), was not able to titrate the YycF protein, even when added at a 400-fold excess (Fig. 5B), thus demonstrating that binding of YycF to the yocH promoter is specific.
DNase I footprinting assays were then performed in order to precisely determine the location and sequence of the YycF binding site. As shown in Fig. 6A and C, YycF protects a region on the non-template strand extending from positions − 73 to − 51 relative to the transcription initiation site. On the template strand, the protected region was more extended (positions − 80 to − 22). This type of asymmetrical protection pattern was also noted in the case of PhoP (Birkey et al., 1998). No obvious difference in affinity or in the extent of the protected region was noted when DNase I experiments were carried out with increasing quantities of unphosphorylated or phosphorylated YycF (data not shown). However, we cannot exclude that this may be due to inefficient phosphorylation by ‘YycG, or that a significant amount of YycF may have been phosphorylated during overproduction in E. coli as recently reported for the Spo0A response regulator (Lewis et al., 1999).
YycF was previously shown to bind to the ftsAZ promoter region as determined by gel shift experiments, but no DNase I footprinting analysis had been performed (Fukuchi et al., 2000). We confirmed that YycF binds specifically to the ftsAZ promoter region using a gel mobility shift DNA-binding assay (data not shown). In order to define a consensus recognition sequence for YycF, DNase I footprinting was performed on a 246 bp DNA fragment carrying the ftsAZ promoter region (positions − 224 to + 21 with respect to the transcription initiation site), generated by PCR with oligonucleotides OSA10/OSA11. As shown in Fig. 6B and C, YycF protects a region extending from positions − 70 to − 46 on the non-template strand. For both yycF and ftsAZ, binding of YocH occurs directly upstream of the − 35 promoter sequences (Fig. 6C), in agreement with its role as a positive regulator of transcription.
A comparison of the DNA regions protected by YycF from DNase I cleavage allowed us to identify a conserved motif constituted by two direct repeats separated by 5 bp. As shown in Fig. 6C, this direct repeat is 5′-TGT A/T A A/T/C N5 TGT A/T A A/T/C-3′. In order to test whether this direct repeat is important for binding of DNA by YycF, site-directed mutagenesis on the yocH promoter region was carried out by SOE-PCR (see Experimental procedures), resulting in two promoter fragments called PyocHA and PyocHB (Fig. 7A). The PyocHA fragment has a mutation changing three nucleotides in one of the two repeats, whereas PyocHB has an increased spacer region of six nucleotides instead of five between the two repeats (see Fig. 7A). As shown in Fig. 7B, DNase I footprinting experiments indicate that mutations in one of the two repeats or increased spacing between the direct repeats prevents binding by YycF, whereas under the same conditions YycF binds to the unmodified DNA fragment. Thus, the 5′-TGT A/T A A/T/C N5 TGT A/T A A/T/C-3′ direct repeat is essential for binding by YycF. The same type of mutagenesis was carried out on the ftsAZ promoter, confirming the importance of the direct repeat for YycF binding (data not shown).
Identification of potential members of the YycF regulon
An extensive DNA-motif analysis of the complete B. subtilis genome was performed to identify likely target genes belonging to the YycF regulon. For this purpose, we used the ‘search pattern’ function in the SubtiList database (http://www.genolist.pasteur.fr/SubtiList; Moszer et al., 2002). Seventeen genes were found to have this sequence on either strand within the 400 base pair region upstream from the translation initiation codon, many of which appear to be involved in cell wall metabolism (yocH, ykvT, tuaAB, tagA/tagD, cotT; see Fig. 8).
In order to confirm binding of YycF on promoter regions containing a potential binding site, DNase I footprinting experiments were performed and we were able to show specific binding of YycF to the ykvT promoter as well as to the promoter region of the tagA/tagD divergon. For ykvT, DNase I footprinting experiments were performed with a 348 bp DNA fragment, generated by PCR using oligonucleotides OSA74/OSA75, and extending from positions − 342 to + 5 with respect to the translation initiation codon. As shown in Fig. 9A and C, YycF protects a region containing the identified direct repeat and extending from − 250 to − 203 on the template strand, with respect to the translation initiation codon. For the tagA/tagD divergon, DNase I footprinting experiments were performed with a 410 bp DNA fragment, generated by PCR using oligonucleotides OSA84/OSA85, and extending from positions − 411 to − 1 with respect to the translation initiation codon. As shown in Fig. 9B and C, YycF protects a region containing the identified direct repeat and extending from − 289 to − 229 on the template strand, with respect to the tagD translation initiation codon.
PhoR may phosphorylate YycF through in vivo crosstalk
The hybrid response regulator approach was adapted in an effort to establish the signal to which the YycG/YycF system responds. In this instance, the reverse hybrid response regulator was constructed, fusing the receiver domain of YycF (amino acids 1–119) and the output domain of PhoP (amino acids 121–239) using the same strategy. The hybrid yycF′-′phoP gene was then placed under the control of the PxylA promoter and integrated at the thrC locus of strain MH5117 to generate strain AH019. Strain AH019 also has a deletion in the phoP gene and therefore the PhoR/PhoP regulon is not induced upon phosphate starvation via the Pho system in this strain. Activation of the hybrid YycF′-′PhoP response regulator by the YycG kinase in response to a cognate stimulus should activate expression of the PhoP regulon and this can be assessed by measuring the level of phoA transcript and/or alkaline phosphatase specific activity.
As the PhoR and YycG histidine kinases are closely related, and since many cell wall metabolism genes whose expression is known to be dependent on phosphate levels are preceded by potential YycF binding sites (pstS, tagA/tagD, tuaAB) it was conceivable that the YycG/YycF system could be involved in sensing phosphate levels in the cell during the exponential growth phase. Strain AH019 was therefore grown in low phosphate medium (LPDM) in the presence or absence of xylose (i.e. in the presence or absence of the hybrid YycF′-′PhoP response regulator), cells were harvested throughout growth and transcription of phoA was detected by Northern analysis. The results show that the level of phoA transcript in cells grown under phosphate limitation conditions (LPDM) in the absence of xylose is not detectable (Fig. 10, upper panel). However, the phoA transcript was clearly detectable in cells producing the hybrid YycF′-′PhoP response regulator during phosphate starvation (Fig. 10, upper panel). In the absence of the PhoR kinase this effect disappeared and the phoA transcript was barely detectable even after phosphate starvation (Fig. 10, lower panel). These results were confirmed by measuring PhoA enzymatic activity under the same conditions (data not shown). These results show that the YycF′-′PhoP hybrid response regulator is functional and can activate the PhoP-regulated phoA promoter in response to phosphate limitation, in a PhoR-dependent manner. This also suggests that the activity of the YycG/YycF two-component system might be activated during phosphate starvation through phosphorylation by the PhoR kinase.
We report the novel application of a hybrid response regulator construct to identify the regulon of the essential YycG/YycF two-component system in B. subtilis. Although other domain-swapping approaches were recently reported using the E. coli CheY, OmpR and PhoB response regulators, these studies were entirely focused on the interactions between the receiver and output domains of the hybrid regulators and were not designed as a tool for studying TCSs of unknown function (Allen et al., 2001; Walthers et al., 2003). It is however, clear from these studies that the linker domain plays an important role in the interaction between the receiver and output domains, with some combinations leading to inactive hybrid proteins. Indeed, it was shown that a OmpR′-′PhoB hybrid response regulator is active, whereas a PhoB′-′OmpR chimera is not (Walthers et al., 2003). In our case, for both the PhoP′-′YycF and YycF′-′PhoP hybrid response regulators, the linker domain for each chimerical protein was systematically associated with its cognate output domain. The evidence presented here indicates that the PhoP′-′YycF and YycF′-′PhoP chimerical proteins are both functional in vivo and can regulate the expression of genes controlled by the native YycF or PhoP regulators respectively.
A comparative macroarray analysis of the B. subtilis transcriptome during growth under phosphate limiting conditions, in the presence or absence of the PhoP′-′YycF hybrid response regulator, identified yocH as a member of the YycG/YycF regulon. Northern analysis indicated that expression of yocH was indeed increased and persisted longer when expression of the hybrid PhoP′-′YycF regulator was induced by xylose and the system was activated by phosphate limitation. This occurred only when the PhoR kinase was present, indicating that PhoR/PhoP′-′YycF is a functional hybrid two-component system, capable of being activated by phosphate limitation. Expression of yocH was also increased during exponential growth when YycG/YycF are overproduced. In this case, yycG/yycF were expressed from an IPTG-inducible promoter. Under these conditions, yocH expression was activated only during the exponential growth phase, suggesting that the signal activating the YycG/YycF system is only present in the vegetative phase.
A direct interaction between YycF and the yocH promoter was demonstrated by gel shift and DNase I footprinting analysis and revealed that the YycF binding region was located just upstream from the − 35 box, which is typical for positive regulators that assist RNA polymerase recruitment (Hochschild and Dove, 1998). Comparison of the regions protected by YycF in the yocH and ftsAZ promoter regions allowed us to define a conserved tandemly repeated sequence with a spacer of five nucleotides: 5′-TGT A/T A A/T/C N5 TGT A/T A A/T/C-3′. As the direct repeat is six base pairs, the two sequences repeat with an 11 bp interval, which is consistent with the fact that YycF belongs to the OmpR family of response regulators. Indeed, these proteins have a winged helix–turn–helix DNA-binding domain (Mizuno and Tanaka, 1997; Martinez-Hackert and Stock, 1997) and are known to bind to tandem direct repeats with an approximately 11 bp interval, as shown for PhoB of E. coli (Blanco et al., 2002)
A detailed DNA motif analysis indicates that the YycF consensus recognition sequence occurs 17 times within 400 bases of a start codon in the chromosome of B. subtilis. The corresponding genes are indicated in Fig. 8. Most of these genes appear to play a role in cell-wall metabolism (yocH, ykvT, tuaAB, tagA/tagD, cotT) in addition to ftsAZ, encoding cellular division proteins. Data from the Japanese and European B. subtilis genome functional analysis research programs (Kobayashi et al., 2003) indicate that this list includes three essential operons: ftsAZ, previously shown to be regulated by YycF (Fukuchi et al., 2000), as well as the tagDEF and tagAB divergent operons which encode key components of teichoic acid biosynthesis and cell wall metabolism. However, the ftsAZ operon is transcribed from three independent promoters and the proximal one, controlled by the YycG/YycF system, is not essential (Gonzy-Treboul et al., 1992). The identification of the tag divergon as a member of the YycF regulon provides a likely candidate to explain the essential nature of the YycG/YycF system. Indeed, we have shown by DNase I footprinting that YycF binds specifically to the predicted recognition sequence in the promoter region of the tagD/tagA genes, as well as to that upstream from the ykvT gene, encoding a potential cell wall hydrolase, further reinforcing the notion that the YycG/YycF system may act to control cell wall metabolism. Although the direct repeat with an appropriate spacing is clearly required for YycF binding, it may not be sufficient as preliminary experiments indicate that YycF does not bind to the yycR, pstS and yvsH promoter regions which also contain potential binding sites (data not shown). An involvement of YycG/YycF in cell wall metabolism is consistent with results obtained from S. aureus where a strain containing a temperature-sensitive yycF allele displayed attenuated virulence, ilv auxotrophy, increased sensitivity to the MLS family of antibiotics and alterations in membrane integrity at the restrictive temperature (Martin et al., 1999). Recently, the hypersensitivity to the MLS family of antibiotics has been linked to the lack of synthesis of a staphylococcal antigen, Ssa, in the thermosensitive mutant (Martin et al., 2002). Experiments are in progress to test direct binding of the ssa promoter region by the purified YycF protein from S. aureus. This constitutes a likely hypothesis since the ssa promoter region also contains a potential YycF binding site (S. Dubrac, unpublished results).
We have shown that yocH is expressed in B. subtilis primarily during exponential growth, with levels dropping rapidly as cells enter stationary phase (Figs 2 and 3 and data not shown). This is in agreement with previous data indicating that the yycFG operon is expressed essentially during exponential growth and rapidly shut off as cells enter stationary phase (Fabret and Hoch, 1998). Furthermore, we show here that the YycG/YycF system seems to be active only during the exponential phase (Fig. 3B). Interestingly, expression of ykvT and the tag divergon also occurs mainly during exponential growth and is rapidly shut off as cells enter stationary phase, consistent with the idea that their expression is controlled by the YycG/YycF TCS (Mauël et al., 1994; Micado database: http://www.locus.jouy.inra.fr/cgi-bin/genmic/madbase_home.pl).
There appear to be close links between the YycG/YycF system and the PhoR/PhoP system. The function of the PhoR/PhoP regulon is to scavenge and provide additional phosphate sources for the cell during phosphate starvation (Hulett, 2002). When phosphate conditions become limiting, cells stop dividing, and the PhoR/PhoP two-component system is activated. The highly related YycG/YycF and PhoR/PhoP systems may both function to sense phosphate levels, one signalling phosphate availability to exponentially growing cells and the other responding to phosphate limitation stress. Indeed, the PhoR and YycG histidine kinases share a similar structure, including a conserved PAS domain, known to be involved in ligand binding, protein–protein interaction and sensing signals such as redox potential, light, oxygen or energy levels (Taylor and Zhulin, 1999; Galperin et al., 2001). In PhoP, the PAS domain extends from amino acid residues 234–300, whereas in YycG it is located between residues 263 and 329 and followed by an associated PAC domain (residues 328–370) as determined using the SMART database (Schultz et al., 1998).
Studying the YycG/YycF two-component system is of particular interest as it is essential for bacterial survival and specific to low G + C Gram-positive bacteria, including many major pathogens. Consequently, it constitutes a prime target for novel antibacterial compounds. This study reveals the nature of the YycG/YycF regulon in the Gram-positive model bacterium B. subtilis. Work is now in progress to extend our results to other Gram-positive bacteria, including several pathogens. The amino acid sequence of the YycF DNA recognition helix is practically invariant in these bacteria, indicating that the DNA recognition sequence should also be the same. Detailed genome sequence analyses revealed 12 potential target genes for YycG/YycF regulation in S. aureus, 13 in L. monocytogenes, and five in S. pneumoniae. As mentioned above, experiments are in progress to study the Yyc regulon in S. aureus and some preliminary results suggest that the YycF box is very similar in B. subtilis and in S. aureus (S. Dubrac, unpubl. results).
Although bioinformatic analysis suggests that the YycG/YycF regulon in B. subtilis may contain up to 17 genes, only yocH was identified in the macroarray analysis using the PhoP′-′YycF hybrid regulator. Because YycG/YycF is an essential two-component system and cannot be deleted, genes controlled by the PhoP′-′YycF hybrid regulator will only be recognized if their expression deviates from the normal expression pattern. For example, if expression of a gene is fully activated or repressed by the endogenous YycF protein, the presence of the PhoP′-′YycF hybrid will probably make little difference to the expression profile and this gene will not be recognized as a member of the regulon in the transcriptome analysis. This limitation does not apply to non-essential two-component systems, as the analysis can be performed in strains with a mutation inactivating the response regulator gene. A second contributing factor is that genes identified in the transcriptome analysis were those whose expression persisted in cells during phosphate-limitation induced stationary phase. Genes involved in cell division, such as ftsAZ, might not be expressed in this non-growing state, and it has been shown that ykvT and the tag divergon are not expressed under these conditions. In general therefore, genes whose expression is limited to growing and dividing cells might not be identified, unless their expression can be reactivated by the hybrid regulator during the stationary phase. These considerations illustrate some of the limitations of the hybrid regulator approach and but also indicate areas for future research.
Bacterial strains and growth media
Bacterial strains used in this study are listed in Table 1. Escherichia coli K12 strain TG1 (Gibson, 1984) was used for cloning experiments, and E. coli strain BL21 λDE3 (Studier and Moffatt, 1986) for protein overexpression and purification. E. coli strains were grown in Luria–Bertani (LB) medium and transformed by electroporation (Sambrook et al., 1989). Bacillus subtilis 168 trpC2 derivatives were grown in LB medium, high phosphate defined medium (HPDM) or low phosphate defined medium (LPDM) (Muller et al., 1997).
Plasmids and plasmid constructions
Plasmids and oligonucleotide primers used in this study are listed in Table 1 and Table 2 respectively. Standard procedures were used for DNA manipulation (Sambrook et al., 1989). Plasmids pXT (Derréet al., 2000) or pMUTIN4 (Vagner et al., 1998) were used for xylose or IPTG-inducible gene expression respectively. Plasmid pET28/16 (Chastanet et al., 2003), a derivative of plasmid pET28a (Novagen), was used for protein overexpression and purification. The input domain of PhoP was amplified by PCR using primers PhoP3 and AH212 generating a DNA fragment of 407 bp (codons 1–120) and fused to a DNA fragment encoding the output domain of YycF (482 base pair, oligonucleotides YycF1/AH221, codons 121–236), using the PCR-based method of Strand Overlap Extension (Horton et al., 1989), to construct the hybrid regulator gene phoP′-′yycF. The same method was used to construct the yycF′-′phoP hybrid regulator gene, but primers YP3 and YP4 were used to amplify the fragment encoding the PhoP output domain (428 base pair product, codons 121–239) and YP1 and YP2 to amplify that encoding the YycF input domain (422 base pair product, codons 1–119). The resulting fragments were digested with HindIII and BamHI and cloned into pXT, generating plasmids pAH20 and pAH21 which contain the phoP′-′yycF and yycF′-′phoP hybrid response regulator genes under control of the xylose-inducible PxylA promoter. The plasmids were then linearized using ScaI and transformed into strain MH5117, in which the phoP gene is deleted (Hulett et al., 1994), to give strains AH014 (PxylA phoP′-′yycF) and AH019 (PxylA yycF′-′phoP) respectively.
To generate a strain in which the yycFG operon was placed under the control of the inducible Pspac promoter, a 257 bp fragment generated by PCR using primers YycF4 and YycF5, carrying the ribosome-binding site and 5′ portion of yycF, was digested with EcoRI and BamHI and cloned into pMUTIN4 to generate pAH22. The circular plasmid was then transformed into B. subtilis strain 168, and integration of the plasmid through a single crossover event generated strain AH9912 in which the yycFG operon is placed under Pspac control.
Histidine kinases of the EnvZ/OmpR family of two-component systems have two amino-terminal transmembrane domains. In YycG, these are located from residues 14–34 and 183–203, as determined using von Heijne's algorithm (Claros and von Heijne, 1994). YycF and ‘YycG (isolated cytoplasmic histidine kinase domain, residues 218–611) were overproduced using plasmids pETyycF and pETyycG, respectively, constructed by cloning PCR-generated NcoI/XhoI DNA fragments corresponding to the yycF (721 bp, PCR TM295/TM296) or ‘yycG (1204 bp, PCR TM297/TM298) coding sequences between the NcoI and XhoI sites of plasmid pET28/16, replacing the stop codons with a XhoI restriction site. This allows the creation of translational fusions adding six histidine residues to the carboxy-terminus of the corresponding protein, placing expression of the genes under the control of a T7 bacteriophage promoter.
DNA macroarray analysis
Strain AH014 (PxylA phoP′-′yycF) was grown in LPDM with and without 1% xylose. 12 ml of culture were harvested by centrifugation (1 min at 9000 r.p.m.) after 1, 4 or 5 h after cessation of exponential growth. RNA isolation, radioactive cDNA synthesis and hybridization to B. subtilis genome Panorama macroarrays (Sigma Genosys) were carried out as previously described (Eymann et al., 2002). Data analysis was performed using ArrayVision software (Version 5.1; Imaging Research, St Catherines, Ontario, Canada) as described (Petersohn et al., 2001).
Strain AH014 was grown in HPDM overnight, and inoculated into 150 ml LPDM in the presence or absence of xylose. Cultures were grown at 37°C with shaking at 220 r.p.m. Cells were harvested at appropriate intervals, rapidly centrifuged at 8000 r.p.m. for 1 minute, the pellets snap frozen in a dry ice/ethanol bath and stored at − 80°C. RNA was extracted as previously described (Noone et al., 2000). Twenty-five microgram aliquots of total RNA were electrophoresed through an agarose gel (1.2% weight/volume) containing 2.2 M formaldehyde (Sambrook et al., 1989), and transferred to a positively charged nylon membrane (Biodyne B, Pall Gelman, Ann Arbor, Michigan) by capillary blotting. Membranes were hybridized with probes corresponding to phoA and yocH, synthesized by PCR using primer pairs PhoA2/PhoA3 and YocH1/YocH2, respectively, which were labelled using a PCR Dig-labelling mix (Roche) and detected using a digoxygenin detection kit (Roche). Membrane stripping was carried out by washing the membrane in stripping solution (50% formamide, 50 mM Tris-HCl pH 7.5, 5% SDS) at 80°C for 60 min.
To construct a strain where the phoPR locus was deleted, two DNA fragments (phoP′ and phoR′) were synthesized using primer pairs PhoPR ko1/PhoPR ko2 and PhoPR ko3/PhoPR ko4, and were cloned into pDG646 on either side of an erythromycin resistance gene. The phoP′ and phoR′ fragments consisted of nucleotides 83–333 of the phoP ORF and nucleotides 132–1713 of the phoR ORF respectively. The phoP′ fragment was digested with PstI and XbaI and cloned into pDG646. The resulting plasmid was cut with ClaI and XmaI and the phoR′ fragment was cloned generating plasmid pAH23. This plasmid was linearized and transformed into B. subtilis strain 168 to construct strain AH024. The phoPR deletion was confirmed by Southern blotting. Strains AH027 and AH028 were constructed by transformation of plasmids pAH020 and pAH021 into strain 168. Deletion of the phoPR locus in strains AH027 and AH028 was carried out by transformation with chromosomal DNA from strain AH024, thereby generating strains AH029 and AH030.
Overproduction and purification of ‘YycG and YycF
Plasmids pETyycF and pETyycG were introduced into a BL21 λDE3 strain, in which the T7 RNA polymerase gene is under the control of the inducible lacUV5 promoter, which also carries the pREP4 plasmid allowing co-production of the GroESL chaperonin, in order to optimize recombinant protein solubility (Amrein et al., 1995). The resulting strains were grown in 2 L of LB medium at room temperature, expression was induced during the mid-exponential growth phase by addition of 1 mM IPTG (isopropyl-β-d-thiogalactoside) and incubation pursued for 4 h. Cells were centrifuged at 10 800 g for 30 min, and resuspended in 1/50 of the culture volume of buffer I (50 mM NaH2PO4 pH 8, 300 mM NaCl, 20 mM imidazole). Cells were disrupted by sonication and cell debris removed by two consecutive 30 min centrifugation steps at 17 200 g. Escherichia coli crude protein extracts were loaded on a 0.2 ml Ni-NTA agarose (Qiagen) column equilibrated with buffer I. The column was washed with 10 volumes of buffer II (50 mM NaH2PO4 pH 6, 300 mM NaCl, 30 mM imidazole) and the proteins were eluted with an imidazole gradient (30 mM-500 mM). Fractions were pooled and dialysed against buffer III (50 mM NaH2PO4 pH 8, 300 mM NaCl, 50% glycerol) to remove imidazole and concentrate the protein solution approximately fourfold.
Protein phosphorylation assays
For ‘YycG autophosphorylation, 2 µg of protein was incubated 3 min at room temperature in 10 µl of phosphorylation buffer (100 mM Tris HCl pH 8, 200 mM KCl, 4 mM MgCl2, 0.5 mM DTT, 0.1 mM EDTA, 3.5% glycerol, 2.5 µM ATP, 4 µCi [γ-32P]-ATP). For phosphotransfer from ‘YycG to YycF, equal amounts of each protein were incubated together and the reaction was initiated by the addition of a radiolabelled ATP mixture as above. Reactions were stopped by adding 2 µl of SDS loading buffer and analysed by SDS-PAGE on 12% acrylamide gels, followed by autoradiography.
Gel mobility shift DNA-binding assays
DNA fragments corresponding to the promoter regions of yocH (314 bp, OSA7/OSA8), ftsA (246 bp, OSA10/OSA11), ykvT (350 bp, OSA74/OSA75) or tagA/tagD (410 bp, OSA84/OSA85) were generated by PCR using Pwo polymerase (Roche) and the indicated oligonucleotide pairs. Labelling, DNA-binding and electrophoresis were performed as described previously (Derréet al., 1999). For competition assays, non-labelled DNA fragments were synthesized by PCR with oligonucleotides OSA7/OSA8 and OSA52/OSA53 for specific and non-specific competition, respectively.These DNA fragments were then added to the binding mixtures.
DNase I footprinting
Radiolabelled DNA fragments were prepared as described above. YycF binding to DNA (5 × 104 cpm per reaction) was performed at room temperature in a buffer containing 25 mM NaH2PO4 pH 8, 50 mM NaCl, 2 mM MgCl2, 1 mM DTT and 10% glycerol. DNase I treatment was then performed as described previously (Derréet al., 1999).
Synthesis of mutagenised DNA fragments by SOE-PCR
Mutagenised DNA fragments PyocHA and PyocHB were synthesized by SOE-PCR (Ho et al., 1989). Briefly, two DNA fragments overlapping at their ends were synthesized by PCR using oligonucleotide pairs OSA7/OSA58 and OSA8/OSA59 for PyocHA and OSA7/OSA60 and OSA8/OSA61 for PyocHB. The second step consists of a fusion PCR using the two neo-synthesized PCR fragments and the OSA7 and OSA8 primers hybridizing at each end of the DNA fragment. Each mutagenised DNA fragment was then purified and sequenced.
We are very grateful to Georges Rapoport for many helpful discussions and critical reading of the manuscript. We thank Marion Hulett for the generous gift of strain MH5117, Etienne Dervyn for helpful discussion and Georg Homuth for assistance with array analysis. Research in the Unité de Biochimie Microbienne was supported by research funds from the European Commission (Grant QLG2-CT-1999–01455), the Centre National de la Recherche Scientifique, Institut Pasteur, and Université Paris 7. Research in the Smurfit Institute of Genetics was supported by research funds from the European Commission (Grant QLG2-CT-1999–01455), Enterprise Ireland SC/02/109 and AH was partly supported by HEA PRTLI.