A matter of life and death: cell wall homeostasis and the WalKR (YycGF) essential signal transduction pathway

Authors

  • Sarah Dubrac,

    1. Biology of Gram-Positive Pathogens, CNRS URA 2172, Department of Microbiology, Institut Pasteur, 25, Rue du Dr Roux, 75724 Paris Cedex 15, France.
    Search for more papers by this author
    • The first two authors contributed equally to this work.

  • Paola Bisicchia,

    1. Smurfit Institute of Genetics, Trinity College Dublin, Dublin 2, Ireland.
    Search for more papers by this author
    • The first two authors contributed equally to this work.

  • Kevin M. Devine,

    Corresponding author
    1. Smurfit Institute of Genetics, Trinity College Dublin, Dublin 2, Ireland.
    Search for more papers by this author
  • Tarek Msadek

    Corresponding author
    1. Biology of Gram-Positive Pathogens, CNRS URA 2172, Department of Microbiology, Institut Pasteur, 25, Rue du Dr Roux, 75724 Paris Cedex 15, France.
    Search for more papers by this author

*E-mail tmsadek@pasteur.fr; Tel. (33)-1-45-68-88-09; Fax (33)-1-45-68-89-38.

**E-mail kdevine@tcd.ie; Tel. (353)-1-896 1872; Fax (353)-1-671 4968.

Summary

The WalK/WalR (aka YycG/YycF) two-component system (TCS), originally identified in Bacillus subtilis, is very highly conserved and specific to low G+C Gram-positive bacteria, including a number of important pathogens. An unusual feature is that this system is essential for viability in most of these bacteria. Recent studies have revealed conserved functions for this system, defining this signal transduction pathway as a crucial regulatory system for cell wall metabolism, that we have accordingly renamed WalK/WalR. Here we review the cellular role of the WalK/WalR TCS in different bacterial species, focusing on the function of genes in its regulon, as well as variations in walRK operon structure and the composition of its regulon. We also discuss the nature of its essentiality and the potential type of signal being sensed. The WalK histidine kinase of B. subtilis has been shown to localize to the divisome and we suggest that the WalKR system acts as an information conduit between extracytoplasmic cellular structures and intracellular processes required for their synthesis, playing a vital role in effectively co-ordinating peptidoglycan plasticity with the cell division process.

Introduction

Bacteria can survive a multiplicity of physical, chemical and biological insults, displaying an impressive capacity to adapt to changing environmental and nutritional conditions. This often involves significant modifications in gene expression, leading to alterations in cellular structure, physiology and metabolism. Two-component systems (TCSs) are among the mechanisms most commonly used by bacteria to sense and respond appropriately to a wide range of signals, through autophosphorylation of a sensor histidine kinase, usually membrane-bound with an extracellular sensing loop, and phosphorylation of its cognate response regulator protein, a transcriptional regulator which in turn appropriately alters gene expression. However, as these signals are generally transitory, it is unusual for a TCS to be essential for viability. Nevertheless, several essential systems have been described, including the Caulobacter crescentus CenK/CenR TCS involved in cell envelope biogenesis (Skerker et al., 2005) and MtrB/MtrA of Mycobacterium tuberculosis, controlling expression of the essential replication gene dnaA (Fol et al., 2006). The most widely distributed essential TCS, practically ubiquitous among Firmicutes and specific to that phylum, is the WalK/WalR system, the only essential TCS in this group. Originally described in Bacillus subtilis (Fabret and Hoch, 1998; Fukuchi et al., 2000), this system has since been reported as essential for cell viability in several closely related pathogens (Staphylococcus aureus, Streptococcus pneumoniae, S. mutans, S. pyogenes), and referred to under various designations (YycG/YycF, VicK/VicR, MicA/MicB) (Martin et al., 1999; Lange et al., 1999; Throup et al., 2000; Echenique and Trombe, 2001; Wagner et al., 2002; Dubrac and Msadek, 2004; Senadheera et al., 2005; Liu et al., 2006). Attempts to inactivate the walRK locus in Listeria monocytogenes, Enterococcus faecalis and Staphylococcus epidermidis were unsuccessful, indicating that it is likely to be essential in these bacteria as well (Kallipolitis and Ingmer, 2001; Hancock and Perego, 2004; Qin et al., 2006).

Assigning a physiological role to WalKR was at first hampered by its essentiality, precluding conventional genetic analysis. Indeed, both its function and the nature of its regulon had remained mostly uncharacterized for the past 10 years. However, a combination of approaches, including the use of hybrid response regulators, conditional mutants, or WalKR depletion or overproduction using inducible expression systems, coupled with genetic, phenotypic and transcriptomic analyses, has recently shown that WalKR controls cell wall metabolism in all of the bacteria studied. This TCS also plays a role in cell membrane synthesis/fluidity, cell division, exopolysaccharide synthesis and oxidative stress in some cases. As this essential TCS de facto constitutes a highly attractive target for antimicrobial therapy, this aspect has also been the subject of much investigation for the design of specific WalK inhibitors as novel antimicrobial compounds, with several reports in the literature (reviewed in Dubrac and Msadek, 2008). To simplify the diverse nomenclature for WalKR orthologues generated from sundry phenotypic studies (see above), we have renamed the system WalK/WalR to reflect its major role as a regulator of cell wall metabolism (Dubrac et al., 2007).

The WalKR system has attracted considerable interest recently, yielding insights into its physiological function, the nature of its essentiality and its potential activating signal. In this review, we outline these studies, summarizing the current knowledge on the system as well as the function of the WalK/WalR TCS in different bacterial species, and present a model suggesting that the WalKR system plays a unique and key role in synchronizing cell wall metabolism with the division process.

A comparison of walRK operon structures and gene function

Although the surrounding genetic environments vary somewhat, three types of conserved walRK operon structures are observed in low G+C percentage Gram-positive bacteria (Fig. 1A). Bacilli and Listeria species have a 6-cistron operon (walR, walK, yycH, yycI, yycJ, yycK). Staphylococci, lactobacilli and Enterococcus faecalis lack the last cistron, while only three are present in streptococci (walR, walK, yycJ; Fig. 1A).

Figure 1.

Genetic map of the walRK operon.
A. The three types of structural organization found in low G+C% Gram-positive bacteria. Major bacterial species for each type of organization are indicated on the right.
B. Detailed genetic map of the B. subtilis walRK operon. The walRK region is bracketed by two potential transcription terminator-like stem-loop structures. Two independent promoters have been proposed, a Sigma A-dependent promoter upstream from walR, and a likely Sigma G-dependent promoter upstream from yycK. Three transcripts were detected: one covering the entire operon and one encompassing walR and part of walK, both initiated from the Sigma A-dependent promoter, and a third corresponding to yycK initiated from the likely Sigma G-dependent promoter (Fukuchi et al., 2000).

The players

The first two genes of the operon, respectively, encode the WalR response regulator and the WalK histidine kinase (Fig. 1A), whose structural organization is detailed below. Although WalK orthologues are generally typical of the sensor histidine kinase protein family, with two transmembrane domains and an extracellular loop, differences in structure, operon organization and essentiality suggest that the signal being sensed might vary among closely related bacteria. WalR is a member of the largest subfamily of response regulators, the OmpR family. The remaining genes encode between one and three accessory proteins, whose function has been extensively studied in a few cases.

YycH and YycI.  The YycH and YycI proteins appear to be present only in bacteria where the WalK kinase has a two-transmembrane domain/extracellular loop-type configuration, except for L. lactis where both proteins are absent. However, the segment separating the transmembrane regions in WalK of L. lactis has only four amino acids and, thus, might not be functional as an extracellular loop.

The presence of YycH and YycI is often linked to WalK essentiality: the kinase is essential in B. subtilis, S. aureus, E. faecalis and L. monocytogenes but not in streptococci or lactococci, which lack the two proteins (Fabret and Hoch, 1998; Martin et al., 1999; O'Connell-Motherway et al., 2000; Throup et al., 2000; Kallipolitis and Ingmer, 2001; Wagner et al., 2002; Hancock and Perego, 2004; Senadheera et al., 2005). In B. subtilis, the walK gene can only be inactivated when the strain contains a point mutation in walR, leading to a constitutively active form of the response regulator (WalRD54H) (Fukuchi et al., 2000).

YycH and YycI do not share any significant amino acid sequence similarities, yet their tertiary structures are highly similar and both are extracellular proteins anchored by an amino-terminal transmembrane domain (Szurmant et al., 2006; Santelli et al., 2007). They both play a role in negatively controlling WalK activity in B. subtilis, and their absence leads to aberrant regulation of the WalKR regulon, with associated growth and cell wall defects and a 10-fold upregulation of the WalKR-dependent yocH gene (Howell et al., 2003; Szurmant et al., 2005; 2007). YycH and YycI form a ternary complex with WalK and the three proteins have been shown to interact through their transmembrane domains (Szurmant et al., 2005; 2007; 2008). The fact that YycH and YycI are absent in streptococci and lactococci suggests that the signal being sensed and the regulation of WalK activity in these bacteria may differ from that in Bacillus, Listeria, Enterococcus and Staphylococcus.

YycJ.  All identified walRK operons include a yycJ gene (Fig. 1A). Although its function is unknown, Hidden Markov Model predictions suggest YycJ is a member of the metallo-β-lactamase superfamily, possibly indicating a role in cell wall metabolism. Deletion of yycJ in B. subtilis affects colony morphology and mutants accumulate secondary mutations that prevent sporulation (Szurmant et al., 2007). In S. pneumoniae, YycJ is required for optimal growth only when WalR levels are reduced, and constitutive expression of yycJ was shown to attenuate the competence defect of a walK mutant (Throup et al., 2000; Wagner et al., 2002; Ng et al., 2003). YycJ also seems to be involved in pneumococcal virulence, as overexpression of yycJ decreased virulence in a murine intraperitoneal infection model (Wagner et al., 2002) and a yycJ null mutation strongly attenuated bacterial growth in a murine respiratory tract infection model (Throup et al., 2000). Inactivation of yycJ in S. mutans has a more pronounced phenotype, affecting growth, natural competence, oxidative stress tolerance and biofilm formation (Senadheera et al., 2007). Inactivation of walK in S. mutans also affects genetic competence development, oxidative stress resistance and biofilm formation; in addition, inactivation of yycJ led to increased expression of the gtfB/C genes, encoding glucosyltransferases involved in adhesion, which are known to be positively regulated by WalR (Senadheera et al., 2005; 2007; Deng et al., 2007). Together, these data suggest that YycJ also participates in the WalKR signal transduction pathway and may exert a negative effect on WalKR activation. In contrast, S. aureus cells lacking yycJ did not display any distinguishing phenotypes and cell morphology and oxidative stress resistance were not affected (O. Poupel and S. Dubrac, unpubl. results).

YycK.  The last gene of the operon, yycK (aka yyxA), is present only in Listeria and Bacillus, and encodes a likely HtrA-type serine protease. YycK plays a role in heat, acid and penicillin stress response as well as pathogenesis in L. monocytogenes (Stack et al., 2005), but does not appear to influence WalKR activity in B. subtilis (Szurmant et al., 2005).

Expression profile of the walRK operon

Three transcripts are detected for the walRK operon in B. subtilis: one extends through all six cistrons of the operon (Fig. 1B) and a second shorter transcript encodes only WalR, with both transcripts present during exponential growth. The third transcript encodes only YycK and is detected during the late sporulation phase. Expression analysis is consistent with a SigA-type promoter upstream from the walR gene and a second likely SigG-type promoter located upstream of yycK (Fabret and Hoch, 1998; Fukuchi et al., 2000). However, the decreased walRK operon expression observed as cells enter transition phase does not result in lowered levels of the WalK and WalR proteins in B. subtilis stationary phase cells (Fabret and Hoch, 1998; Szurmant et al., 2005; Howell et al., 2006).

In B. subtilis, the walRK operon does not appear to be autoregulated (Fabret and Hoch, 1998). While one transcriptome study indicated that the walRK operon is autoregulated in S. pneumoniae (Mohedano et al., 2005), a second study did not reveal any autoregulation and in vitro gel mobility shift assays failed to demonstrate binding of WalR to the operon promoter region (Ng et al., 2005).

Structural features of the WalK kinase

WalK of B. subtilis belongs to Class IIIA, a histidine kinase subgroup also including the closely related ResE and PhoR proteins (Fabret et al., 1999). Many histidine kinases also act as phosphoprotein phosphatases, catalysing the dephosphorylation of the cognate response regulator. Structure-based and mathematical modelling-based analyses indicate that WalK acts as a so-called ‘bifunctional sensor’ i.e. with both kinase and phosphatase activities, although this was not experimentally demonstrated (Alves and Savageau, 2003). The WalK crystal structure has yet to be resolved, however, a three-dimensional model of the conserved ATP-binding subdomain of S. epidermidis WalK was constructed in silico and used in a structure-based virtual screen for specific inhibitory compounds (Qin et al., 2006).

The domain structure of WalK orthologues is shown in Fig. 2: they have one or two transmembrane sections, a HAMP domain, as well as the canonical HisKA and HATPase_c motifs that define this protein family.

Figure 2.

Domain architecture of the WalK histidine kinase and its cognate WalR regulator. The graphical representation is derived from data in the SMART database at http://smart.embl-heidelberg.de. Domains and their co-ordinates are indicated based on the WalK and WalR proteins of Bacillus subtilis. HAMP: Histidine kinases, Adenylyl cyclases, Methyl binding proteins, Phosphatases domain; PAS: PER-ARNT-SIM sensor domain; HisKA: dimerization and phosphoacceptor domain of histidine kinases; HATPase_c: Histidine kinase-like ATPases; REC: CheY-homologous receiver domain; Trans_reg_c: DNA binding domain. The extracellular loop of WalK extends from co-ordinates 34–183. The star symbol indicates that the indicated extracellular PAS domain of WalK was not predicted by a HMM but by a computational-modelling approach (Szurmant et al., 2008).

Most WalK orthologues contain a poorly conserved extracytoplasmic loop of 142–154 amino acids, flanked by two transmembrane domains, with the exception of Lactococcus lactis, where the ‘loop’ has only four amino acids. WalK orthologues in streptococcal species lack an extracytoplasmic loop, with a single transmembrane domain and 4–12 extracellular amino acids (Ng and Winkler, 2004). Extracytoplasmic loops are known to play a crucial role in detecting external stimuli in several histidine kinases, while the transmembrane domains function as anchors to the cytoplasmic membrane, but can also be involved in signal perception (Mascher et al., 2006; Szurmant et al., 2008).

Although the amino-terminal sensing domains of WalK are not well conserved between bacterial species, the presence of an intracellular PAS (PER-ARNT-SIM) domain is a distinguishing feature of WalK orthologues. These domains adopt typical α/β folds and are involved in sensing signals such as oxygen, light, redox potential or the presence of specific ligands (Taylor and Zhulin, 1999), although their role in WalK kinases remains unclear. Substitution of a conserved residue in the N-terminal cap of the WalK PAS domain of S. pneumoniae resulted in an in vitro auto-phosphorylation defect, and the PAS domain is required for competence repression under microaerobiosis in this bacterium (Echenique and Trombe, 2001), but not for in vitro autophosphorylation of the isolated cytoplasmic histidine kinase domain (Clausen et al., 2003). The WalKR system in B. subtilis is one of the rare examples where physiologically relevant cross-talk has been demonstrated, as the PhoR kinase, which also has a PAS domain, is able to phosphorylate WalR upon phosphate limitation (Howell et al., 2006). Although in addition to WalK and PhoR there are seven other histidine kinases with PAS domains in B. subtilis, it is interesting to note that in S.  aureus the only two proteins with PAS domains are PhoR and WalK, whereas in S. pneumoniae and S. mutans WalK is the only PAS-domain protein (Ulrich and Zhulin, 2007). Interestingly, the B. subtilis WalK extracellular loop is also predicted to adopt a PAS-like fold (Szurmant et al., 2008; Fig. 2), which may play a role in signal detection.

The WalR response regulator and its recognition sequence

WalR is a typical member of the OmpR/PhoB family of response regulators, with the signature winged helix–turn–helix DNA-binding Trans_reg_c domain (Fig. 2; Martinez-Hackert and Stock, 1997; Mizuno and Tanaka, 1997). The structures of the isolated S. pneumoniae WalR receiver domain and E. faecalis WalR carboxy-terminal domain have both been determined (Bent et al., 2004; Trinh et al., 2007). WalR orthologues typically display a high level of amino acid sequence identity (generally > 70%) particularly in the DNA-binding domain. This conservation reflects that observed for the WalR DNA recognition motifs in different bacteria: WalR binds to two hexanucleotide direct repeats separated by five nucleotides, and a consensus binding sequence of 5′-TGTWAH-N5-TGTWAH-3′ was defined through protein/DNA interaction studies in B. subtilis and S. aureus (Howell et al., 2003; Dubrac and Msadek, 2004), with a slightly different consensus (5′-TGTNAN-N5-NGTNANA-3′) described in S. pneumoniae (Ng et al., 2005).

An alignment of 21 direct WalR binding sites identified in B. subtilis, S. aureus, S. pneumoniae and S. mutans through sequence analysis and gel mobility shift or DNase I footprinting assays was performed and is shown in Fig. 3A. Using this alignment, a revised WalR consensus binding site sequence was generated, 5′-TGTNDH-N5-BKBWRN-3′, as shown in Fig. 3B, which may prove useful in predicting further members of the WalKR regulon through genomic analysis.

Figure 3.

Alignment of WalR binding sites.
A. Alignment of upstream nucleotide sequences for genes where binding of purified WalR has been shown. The indicated WalR binding sites were either mapped by DNaseI footprinting experiments, or predicted following gel mobility shift assays (Howell et al., 2003; Dubrac and Msadek, 2004; Ng et al., 2005; Senadheera et al., 2005; Bisicchia et al., 2007). Sequences indicated in bold correspond to the previously established WalR recognition sequence (Howell et al., 2003; Dubrac and Msadek, 2004).
B. WebLogo graphic of the revised WalR consensus binding site generated by sequence alignments (http://weblogo.berkeley.edu/).

WalR appears to both activate and repress gene expression in B. subtilis, depending on the position of the WalR binding site relative to the regulated promoter (Bisicchia et al., 2007). A model for a complex between the WalR DNA-binding domain and DNA was constructed in silico based on the known structure of the E. coli PhoB/DNA complex (Blanco et al., 2002; Trinh et al., 2007). It was suggested that in WalR the α-loop, which connects the two helices of the helix–turn–helix domain, might interact with the Sigma A subunit of RNA polymerase, while the loop linking the DNA recognition helix to the C-terminal β-hairpin might also contact DNA (Trinh et al., 2007), although this latter loop has also been suggested to be involved in WalR dimerization (Watanabe et al., 2003).

The WalKR system plays a major role in controlling cell wall metabolism

The first identified member of the WalKR regulon was ftsAZ in B. subtilis, signalling a potential role in cell division (Fukuchi et al., 2000). Expression of this operon is complex, driven by three distinct promoters, with only one nonessential promoter controlled by WalKR. The link with cell division was recently confirmed, as WalK and FtsZ co-immunoprecipitate, and WalK localization at the incipient septum requires FtsZ (Fukushima et al., 2008).

Recent studies indicate that cell wall metabolism genes predominate in the WalKR regulon, as summarized in Table 1 for the three bacteria where this system has been extensively studied. In B. subtilis, WalKR activate expression of yocH, yvcE (cwlO) and lytE, all encoding autolysins, and of ydjM whose product is predicted to be cell wall-associated (Howell et al., 2003; Bisicchia et al., 2007). In addition, WalKR negatively regulate expression of yoeB (iseA) and yjeA, encoding proteins that modulate autolysin activity: IseA inhibits autolysin activity while YjeA is a peptidoglycan deacetylase that modulates peptidoglycan degradation by lysozyme (Bisicchia et al., 2007; Salzberg and Helmann, 2007). Depletion for both YvcE and LytE results in reduced cell wall synthesis and a defect in lateral cell wall synthesis and elongation (Bisicchia et al., 2007). Thus, when activated, WalKR controls expression of autolysins that mediate lateral cell wall synthesis and cell elongation, while proteins that modulate or inhibit autolysin activity are produced when WalKR is not active. Although cell viability of B. subtilis requires the activity of either LytE or YvcE, control of lytE and yvcE expression does not account for WalKR essentiality in B. subtilis as uncoupling of their expression from WalKR regulation did not allow walRK inactivation (Bisicchia et al., 2007). The essential nature of the system is thought to be polygenic, likely arising from an imbalance in cell wall synthesis and turnover caused by misregulation of a number of genes encoding autolysins as well as autolysin-regulating proteins.

Table 1.  WalKR-dependent genes involved in cell wall metabolism, fatty acid metabolism and virulence in B. subtilis, S. aureus and S. pneumoniae.Thumbnail image of

In S. aureus, bioinformatics and transcriptional studies showed WalKR-dependent regulation of nine cell wall metabolism genes (Dubrac et al., 2007). These encode the two major S. aureus autolysins, AtlA and LytM, and two recently characterized lytic transglycosylases, IsaA and SceD (Dubrac and Msadek, 2004; Dubrac et al., 2007; Stapleton et al., 2007), as well as five proteins with a CHAP amidase domain (SsaA, SA0620, SA2097, SA2353, SA0710) (see Table 1; Dubrac et al., 2007). Direct binding of WalR to the promoter regions of ssaA, isaA and lytM was shown, indicating that genes involved in cell wall metabolism are controlled directly by the WalKR system (Dubrac and Msadek, 2004; Dubrac et al., 2007). Consistent with this role in cell wall metabolism is the finding that cell wall biosynthesis and turnover are strongly reduced in WalKR-depleted S. aureus cells, as is sensitivity to Triton X-100 and lysostaphin (Dubrac et al., 2007). However, individually, none of these cell wall metabolism genes appear to account for the essentiality of the system in S. aureus.

Streptococcus pneumoniae transcriptome studies revealed WalKR-mediated activation of the cell wall hydrolase-encoding genes pcsB, lytB and lytN, and several additional genes encoding proteins with LysM-type cell wall binding domains (Ng et al., 2003). WalKR depletion causes defects in cell morphology and murein synthesis that are similar to those observed when cells have reduced levels of the PcsB autolysin (Ng et al., 2004). Although there are conflicting reports in the literature regarding the essentiality of pcsB in S. pneumoniae, control of its expression by WalKR accounts for the essentiality of this TCS, the only instance where it can be ascribed to expression of a single gene (Ng et al., 2003; 2004; Giefing et al., 2008). Several genes of the S. pneumoniae WalR regulon are also controlled by StkP, a serine/threonine-type kinase, suggesting that StkP could influence WalKR activity (Saskova et al., 2007).

While the major role of the WalKR TCS is clearly linked to regulation of cell wall metabolism, transcriptome studies in B. subtilis and S. pneumoniae also suggest a role in regulating fatty acid metabolism. In S. pneumoniae, expression of genes involved in fatty acid biosynthesis increases concomitantly with WalKR levels in the cell, resulting in increased fatty acid chain lengths and lowered cell membrane fluidity (Table 1) (Mohedano et al., 2005). However, a second transcriptome study did not support direct regulation of fatty acid biosynthesis by WalR, although a walR null mutant (constitutively expressing the essential gene pcsB) was shown to require fatty acids for optimal growth in some conditions, indicating that the pneumococcal WalKR might be involved in mediating membrane integrity (Ng et al., 2005). WalKR depletion in B. subtilis results in increased expression of the membrane phospholipid desaturase gene des, but this effect appears to be indirect (Bisicchia et al., 2007). The apparent involvement of WalKR in membrane homeostasis might therefore be an indirect effect of its role in cell wall metabolism.

Studies in S. mutans have shown that WalKR directly activate expression of gtfB, gtfC, gtfD and ftf, encoding glucosyl- and fructosyltransferases, respectively, which play a role in the formation of dental caries (Senadheera et al., 2005). In addition, a recent study showed that WalKR positively control synthesis of the major autolysin Atl, and of Smu0629, a putative thiol-disulphide oxidoreductase involved in Atl production and maturation (Ahn and Burne, 2007). WalR of S. mutans was shown to be essential, whereas WalK was dispensable but required for optimal growth, similar to the situation in S. pneumoniae (Bhagwat et al., 2001, Senadheera et al., 2005).

The WalKR system of S. pyogenes also controls cell wall metabolism (regulation of the pcsB orthologue), nutrient uptake and resistance to osmotic stress (Liu et al., 2006). Although a knockout walR mutant was described in S. pyogenes, it showed impaired growth in non-immune human blood and serum as well as attenuated virulence (Liu et al., 2006). The only other bacterium in which WalR has been reported not to be essential is Lactococcus lactis (O'Connell-Motherway et al., 2000). As outlined above, variations in WalKR essentiality might reflect intrinsic differences in cell wall metabolism between Gram-positive rods and cocci.

Additional WalKR-dependent phenotypes

Oxidative stress and anaerobiosis

Several studies suggest a role for WalKR in sensing and responding to oxygen levels and/or redox potential. In S. aureus, the hypersensitivity of a walR thermosensitive mutant to MLSB antibiotics is reduced in microaerobic conditions, while in S. pneumoniae WalKR mediates competence repression under microaerobiosis (Martin et al., 1999; Echenique and Trombe, 2001). A study of virulent S. pneumoniae strains showed that inactivation of walK resulted in a shorter lag phase during adaptation to growth under anaerobic conditions, and in attenuated virulence after intranasal challenge (Kadioglu et al., 2003). In S. mutans, deletion of walK resulted in increased susceptibility to oxidative stress, and higher WalKR levels were observed upon treatment with hydrogen peroxide, suggesting a role for WalKR in sensing and responding to oxidative stress, while disruption of walK causes increased biofilm formation under aerobic conditions, indicating the involvement of WalKR in the adaptive response to redox conditions (Ahn and Burne, 2007; Deng et al., 2007). While these studies suggest a role for WalKR in sensing and responding to redox potential and/or oxygen levels, consistent with the presence of a PAS domain in WalK, definition of the precise role played by WalKR under these conditions awaits further investigation.

Antibiotic resistance

A number of reports link the S. aureus WalKR system with vancomycin-intermediate resistance (VISA strains). In one VISA clinical isolate, the WalKR TCS was found to be drastically upregulated, due to an insertional mutation in the walRK promoter that enhanced its efficiency (Jansen et al., 2007). In agreement with this finding, whole-genome sequencing of isolates of the same S. aureus strain recovered periodically from a patient undergoing vancomycin treatment established a link between increased vancomycin resistance and a mutation in the yycH gene, whose product downregulates activity of the WalKR system in B. subtilis (Szurmant et al., 2005; Mwangi et al., 2007). Thus, both studies link vancomycin resistance to increased WalKR levels or activity. WalKR overproduction in a susceptible S. aureus strain was also shown to result in increased resistance to vancomycin, clearly indicating a role of WalKR in developing resistance to this antibiotic, probably through regulation of cell wall metabolism (Jansen et al., 2007).

Daptomycin exerts its effect at the level of the cell membrane, by penetrating the membrane lipid layers and dissipating the proton motive force (Silverman et al., 2003). Laboratory-derived strains and clinical isolates of S. aureus with decreased susceptibility to daptomycin were found to have missense mutations in walK (Friedman et al., 2006). However, it is not clear how these mutations affect WalK activity or whether the WalKR TCS plays a positive or negative role in daptomycin resistance, particularly as the strains carried additional mutations affecting mprF (encoding a lysylphosphatidylglycerol synthetase) and the rpoB and rpoC genes (Friedman et al., 2006). Interestingly, one mutation found in a daptomycin resistant clinical isolate was a single nucleotide insertion early in the walK open reading frame, predicted to result in a truncated inactive WalK protein, even though walK was previously shown to be essential in S. aureus. To explain this conundrum, the authors suggest that the truncated WalK protein might retain some functionality, or that WalR might be phosphorylated by a non-cognate histidine kinase (Friedman et al., 2006). However, this strain also contained a mutation in mprF, which might affect the WalK requirement for cell viability. Nevertheless, the possibility that changes in WalK activity might lead to higher daptomycin resistance cannot be excluded, and the link between WalKR and resistance to daptomycin is in agreement with the observed role of this TCS in modulating cell membrane metabolism.

Virulence

The role of WalKR in infection and virulence was primarily studied in streptococcal species, where the walK kinase gene is not essential. Considering the fundamental role of the WalKR TCS, it is surprising that null walK mutants of S. pneumoniae R6 and of clinical isolates of serotype 3 and 22 display wild-type virulence when tested by murine intraperitoneal challenge or in a murine respiratory tract infection model (Throup et al., 2000; Wagner et al., 2002; Kadioglu et al., 2003). However, the virulence of strains D39 and S6 was significantly attenuated by the walK mutation following intranasal challenge (Wagner et al., 2002). Several virulence genes, such as pspA and piaBCDA, encoding a surface virulence factor involved in the evasion of complement during infection and an ABC transporter involved in iron uptake, respectively, are members of the WalKR regulon and it would be surprising if this TCS does not emerge as a regulator of virulence in S. pneumoniae (Ng et al., 2003; 2005; Ren et al., 2004a,b; Mohedano et al., 2005).

As mentioned above, a walR mutant of group A Streptococcus was successfully constructed and shown to be impaired in its capacity to grow in human blood and to kill mice after subcutaneous infection (Liu et al., 2006). The WalKR system also promotes the capacity of S. mutans to attach to the tooth surface through positive regulation of gtfB, gtfC, gtfD and ftf expression (Senadheera et al., 2005).

WalKR are required in S. aureus for biofilm formation, a key step favouring propagation of infection (Dubrac et al., 2007), while the sdrD and ebpS genes that are involved in S. aureus virulence through interaction of their gene products with the extracellular host matrix are positively controlled by the WalKR system and preceded by WalR binding sites (S. Dubrac, unpubl. results). Clearly more research is required to elucidate the impact of the WalKR system in vivo, but it is highly likely that fine-tuning of WalKR activity during the course of infection is crucial for bacterial propagation and persistence within the host.

Investigating the WalK activation signal: a transcriptome approach

The nature of the signal(s) sensed by WalK is unknown. However, based on the observation that WalKR depletion or treatment with vancomycin lead to similar transcriptional changes in B. subtilis, the signal sensed by WalK was proposed to emanate from normal cell wall metabolism, as the antithesis of cell wall stress (Cao et al., 2002; Bisicchia et al., 2007). To explore this hypothesis, a survey of transcriptome studies reporting the response of B. subtilis to different cell wall active antibiotics was undertaken. Conditions that lead to WalK activation would be reflected in increased expression of yocH, yvcE and ydjM with concomitant decreased expression of yoeB and yjeA, while conditions that lead to the absence of WalK signal would have the opposite transcriptional profile. Changes in expression of des (reflecting changes in membrane fluidity) (Cybulski et al., 2002) and liaI (induced by cell envelope stress) (Mascher et al., 2004) were also monitored. The results for cell wall-active antibiotics are shown in Table 2; values consistent with WalKR deactivation/depletion are shown in red while those consistent with increased WalKR activity/levels are shown in blue. The full analysis is available as supplementary material, Table S1.

Table 2.  Transcriptional changes of WalKR regulon genes in Bacillus subtilis in response to cell wall acting antibiotics. Thumbnail image of

WalKR activity appears decreased when cells are treated with antibiotics that target the early stages of peptidoglycan precursor synthesis (Table 2, Fig. 4A), e.g. fosfomycin inhibits the first step of peptidoglycan synthesis (conversion of UDP-GlcNAc into UDP-MurNAc), while bacitracin inhibits recycling of the membrane lipid carrier and hence its availability (Jordan et al., 2008). In contrast, WalKR activity is increased by antibiotics targeting the late stages of peptidoglycan synthesis, e.g. β-lactams that inhibit transglycosylation and transpeptidation activities of the penicillin binding proteins (PBPs) (Table 2, Fig. 4A). These observations are consistent with the idea that WalK detects levels of available Lipid II in the extracytoplasmic compartment, where both Lipid II and the WalK sensing domain are located (Fig. 4A). Thus, antibiotics acting on early steps of peptidoglycan synthesis would lower Lipid II levels (causing deactivation of WalK) whereas β-lactams block utilization of Lipid II by PBPs, causing Lipid II levels to increase, thereby activating WalK. The effect of treating cells with vancomycin and ristocetin is interesting in light of this hypothesis: although both these antibiotics act in the later stages of peptidoglycan synthesis, by binding to the D-Ala-D-Ala moiety of Lipid II to block transglycosylation and transpeptidation reactions (Kahne et al., 2005), their effect on WalKR activity is similar to that of the early acting antibiotics (Table 2, Fig. 4A). We suggest therefore that binding of vancomycin or ristocetin mimics lowered Lipid II levels; i.e. Lipid II bound to vancomycin or ristocetin is unavailable for cell wall synthesis and hence WalK is deactivated (Table 2). These results support the hypothesis that the level and/or availability of Lipid II might be a signal for WalK activation (Fig. 4A). We further speculate that D-Ala-D-Ala might play a role in WalK signalling, as it is this moiety of Lipid II that is masked by vancomycin/ristocetin binding. Interestingly, β-lactams are structural homologues of D-Ala-D-Ala and so may be capable of activating WalK either directly, or indirectly by Lipid II accumulation through inhibition of its utilization by the PBPs. There are several interesting aspects to this hypothesis that are amenable to experimental verification: (i) Lipid II is the only cell wall synthetic intermediate that is present in the same cellular compartment as the sensing loop domain of WalK, and can thus function as a link between intracellular synthesis and extracellular utilization (Fig. 4A); (ii) Lipid II levels would reflect cell wall synthetic activity (consistent with the idea that the WalK signal emanates from normal cell wall metabolism) – high Lipid II levels signifying rapid cell wall synthesis and low levels signifying reduced cell wall synthesis; (iii) two autolysins encoded by members of the WalKR regulon are endopeptidases: YvcE and LytE – extracytoplasmic D-Ala-D-Ala levels could therefore signal the necessity for expression of these endopeptidases to cleave the cross-linking peptide bridges of the cell wall so that nascent Lipid II could be incorporated. Cleavage of the terminal D-Ala residue upon transpeptidation would lower the level of the D-Ala-D-Ala signal – in this manner WalKR would provide a homeostatic mechanism co-ordinating endopeptidase production with transpeptidation activity during cell wall synthesis; and (iv) there are D-Ala–D-Ala residues within the mature cell wall that are not cross-linked (Fig. 4A) – however, as they are physically removed from the WalK sensing domain it is conceivable that they do not participate in signalling.

Figure 4.

Figure 4.

Model for signal-dependent WalKR induction and its signal transduction.
A. Global representation of peptidoglycan biosynthesis as established in B. subtilis (modified from Bhavsar and Brown, 2006). The transmembrane WalK/YycI/YycH complex has been positioned in order to reflect the hypothesis concerning the signal sensed by this complex. Antibiotics whose action is thought to interfere with the WalK activation signal (also listed in Table 2) are indicated. The proposed model is based on data concerning the effect of cell wall targeting antibiotics. Briefly, antibiotics targeting the first steps of peptidoglycan biosynthesis, whose action results in a lower availability of Lipid II, decrease WalKR activity. A similar effect has also been characterized for glycopeptides that mask the D-Ala-D-Ala moiety of Lipid II. Conversely, antibiotics whose action results in higher availability of Lipid II (i.e. β-lactams) increase WalKR activity. These observations, combined with the topology of the WalK/YycI/YycH complex, lead us to postulate that Lipid II availability could be the signal sensed by the WalK kinase.
B. Summary of the up-to-date knowledge of the WalKR TCS, the cofactors involved in its signal transduction pathway, the regulated functions and the global effect of this TCS in B. subtilis (modified from Fukushima et al., 2008). According to this model, WalK and FtsZ colocalize at the septum during cell division, where cell wall metabolism is particularly active. This entails high activity of the WalKR system resulting in increased synthesis and activation of several cell wall hydrolases and cell-division initiation proteins, whereas synthesis of proteins having a protective effect against cell wall degradation is downregulated. Thus, during cell division, the WalKR system ensures the high cell wall plasticity required for division. By contrast, during stationary phase, the WalK sensor is located around the cell and the effect of the WalKR negative modulators, YycH and YycI, is higher. This results in lower activity of the WalKR system and a low cell wall turnover rate, adapted to the reduced cell wall plasticity required during stationary phase. Further studies are required in order to determine whether YycH and YycI also localize to the septum, whether interactions between the WalK kinase and its two negative modulators, YycH and YycI, can also occur at the septum and whether the interaction between WalK and FtsZ is direct.

Figure 4.

Figure 4.

Model for signal-dependent WalKR induction and its signal transduction.
A. Global representation of peptidoglycan biosynthesis as established in B. subtilis (modified from Bhavsar and Brown, 2006). The transmembrane WalK/YycI/YycH complex has been positioned in order to reflect the hypothesis concerning the signal sensed by this complex. Antibiotics whose action is thought to interfere with the WalK activation signal (also listed in Table 2) are indicated. The proposed model is based on data concerning the effect of cell wall targeting antibiotics. Briefly, antibiotics targeting the first steps of peptidoglycan biosynthesis, whose action results in a lower availability of Lipid II, decrease WalKR activity. A similar effect has also been characterized for glycopeptides that mask the D-Ala-D-Ala moiety of Lipid II. Conversely, antibiotics whose action results in higher availability of Lipid II (i.e. β-lactams) increase WalKR activity. These observations, combined with the topology of the WalK/YycI/YycH complex, lead us to postulate that Lipid II availability could be the signal sensed by the WalK kinase.
B. Summary of the up-to-date knowledge of the WalKR TCS, the cofactors involved in its signal transduction pathway, the regulated functions and the global effect of this TCS in B. subtilis (modified from Fukushima et al., 2008). According to this model, WalK and FtsZ colocalize at the septum during cell division, where cell wall metabolism is particularly active. This entails high activity of the WalKR system resulting in increased synthesis and activation of several cell wall hydrolases and cell-division initiation proteins, whereas synthesis of proteins having a protective effect against cell wall degradation is downregulated. Thus, during cell division, the WalKR system ensures the high cell wall plasticity required for division. By contrast, during stationary phase, the WalK sensor is located around the cell and the effect of the WalKR negative modulators, YycH and YycI, is higher. This results in lower activity of the WalKR system and a low cell wall turnover rate, adapted to the reduced cell wall plasticity required during stationary phase. Further studies are required in order to determine whether YycH and YycI also localize to the septum, whether interactions between the WalK kinase and its two negative modulators, YycH and YycI, can also occur at the septum and whether the interaction between WalK and FtsZ is direct.

The only piece of transcriptomic data apparently inconsistent with this hypothesis is the de-activation of WalKR regulon genes observed upon treatment of cells with oxacillin, a β-lactam antibiotic (Table 2). However, cumulative data suggests that oxacillin may exert multiple effects on the bacterial cell, including oxidative stress (Singh et al., 2001; Singh and Moskovitz, 2003; Pechous et al., 2004), which may affect WalK sensing and result in WalKR de-activation. It is interesting that with the exception of vancomycin and bacitracin, none of the other antibiotics cause differential expression of liaI, suggesting that the WalK signal is very specific and not merely a result of cell wall stress conditions that are sensed by the LiaSR TCS (Mascher et al., 2006). If WalK is sensing free levels of Lipid II outside the cytoplasmic membrane, it would be interesting to test whether genes involved in synthesis of Lipid II or PBPs are also under WalKR control, an area under active investigation.

A survey of stress conditions or treatment with other antimicrobial compounds shows that many display the signature changes in gene expression expected from modulation of WalKR expression: in particular triclosan, an inhibitor of fatty acid synthesis, and ethanol and heat stresses display a profile indicative of WalKR deactivation, while high osmolarity and metal ion stresses show increased activation of WalKR (see Table S1).

A model for WalKR-mediated signal transduction in sensing and co-ordinating cell growth and cell division

We propose that the WalKR TCS senses conditions of active cell wall synthesis, perhaps detected as the accumulation of the peptidoglycan biosynthetic precursor Lipid II. We propose that WalK senses the D-Ala-D-Ala moiety of Lipid II, signalling active cell wall synthesis and the necessity for autolysin production to allow incorporation of the Lipid II-linked cell wall precursor into the nascent cell wall. This model would explain the essential nature of the system: WalKR would be required for viability because it would sense the ‘healthy’ status of actively replicating cells and perform the crucial task of adjusting cell wall metabolism to the requirements of cell growth.

Interestingly, in B. subtilis the WalK kinase localizes at the septum through association with FtsZ, and its activity requires the formation of a proper septum (Fukushima et al., 2008). A role for the B. subtilis WalKR system in septal function is supported by transcriptional regulation of the essential cell division genes ftsAZ (Fukuchi et al., 2000) and by the filamentous growth phenotype of a strain carrying a temperature-sensitive walR allele (Fabret and Hoch, 1998). On the basis of these findings, it was recently proposed that WalKR might function in co-ordinating cell wall remodelling and cell division in B. subtilis (Fukushima et al., 2008).

In this report, we extend this model by integrating it with our proposal of the possible signal being detected by the WalK kinase during growth. A schematic representation of WalKR function in B. subtilis is reported in Fig. 4B, showing the two scenarios of growing and quiescent cells.

During the exponential phase of growth, when cells are actively dividing, WalK is located at the septum and is part of the divisome (Fukushima et al., 2008). The YycH and YycI auxiliary proteins are known to assemble with WalK independently of FtsZ cellular levels, indicating that the formation of this complex precedes participation of WalK in the divisome (Fukushima et al., 2008). However, whether WalK activation requires maintenance of this complex at the septum or physical separation of WalK from the auxiliary proteins YycH and YycI is currently unknown.

In growing cells, cell wall synthesis occurs rapidly and at a high rate, in order to provide cell wall material for cell elongation and septum formation. Cell wall precursors are produced in the cytoplasm and then attached to a lipid moiety, undecaprenol phosphate or Lipid II. This is flipped outside the cell and the subunits are then incorporated into the peptidoglycan. In growing cells the Lipid II precursor is turned over rapidly and is presumably present at high levels. Crucially, both Lipid II and the WalK sensing loop (Fig. 2 and 4A) are extracytoplasmic and can therefore physically interact, perhaps through the Lipid II D-Ala-D-Ala moiety. This would result in WalK activation and subsequent phosphorylation of its cognate response regulator WalR, leading to continued synthesis of autolysins required for cell wall remodelling and cell elongation (LytE, YvcE and YocH), and of the cell division proteins FtsA and FtsZ, while simultaneously shutting down production of YoeB and YjeA, which inhibit and modulate autolysin activity.

In non-growing cells, i.e. during the stationary phase of growth, cell wall synthesis levels diminish, and the Lipid II precursor does not accumulate to the same extent, resulting in reduced WalK activation. In addition, septa are no longer formed in these cells, and WalK might be displaced from the septa and hence inactive. This would result in lowered autolysin synthesis and de-repression of the autolysin-regulating genes yoeB and yjeA. In addition, expression of the cell division genes ftsAZ would no longer be activated by WalKR.

As a result, the cell adjusts its metabolism to the new, non-growing conditions, by reducing both synthesis and activity of autolysins in order to maintain a balanced cell wall synthesis/turnover ratio and avoid cell lysis.

The basic mechanism of WalKR function outlined for B. subtilis is likely to be generally applicable to the low G+C Gram-positive bacteria where the system is conserved, although differences in cell shape and wall architecture might result in deviations from this model.

For instance, S. aureus does not have orthologues of YdjM, LytE and YvcE, although WalKR does control production of numerous autolysins, while in streptococci, which lack YycH and YycI, and where WalK does not have an extracellular loop, significant variations from the proposed model must exist. Such differences will be fruitful to explore, as they have the potential to yield major insights into the control of cell division, cell shape and cell wall metabolism in different bacteria.

Acknowledgements

The authors would like to thank the anonymous reviewers for their constructive comments and useful suggestions. S.D. and T.M. would like to thank Ivo Gomperts-Boneca and James A. Hoch for helpful discussion. Work in the Biology of Gram-positive Pathogens Unit (Institut Pasteur) was supported by research funds to T.M. from the European Commission (Grants BACELL Health, LSHG-CT-2004-503468, StaphDynamics, LHSM-CT-2006–019064, and BaSysBio, LSHG-CT-2006-037469), as well as research funds from the Centre National de la Recherche Scientifique (CNRS URA 2172) and the Institut Pasteur (Grand Programme Horizontal N°9). Work at the Smurfit Institute of Genetics was supported by grants to KMD from Science Foundation Ireland (Principal Investigator Award, 03/IN3/B409), from the EU (BACELL Health, LSHG-CT-2004-503468) and from Enterprise Ireland (SC/02/109).

Ancillary