Cell wall assembly in Bacillus subtilis: how spirals and spaces challenge paradigms

Authors


*E-mail ebrown@mcmaster.ca; Tel. (+1) 905 525 9140 ×22392; Fax (+1) 905 522 9033.

Summary

Although the bacterial cell wall has been the subject of decades of investigation, recent studies continue to generate novel and controversial models of its synthesis and assembly. Here we compare and contrast the transcompartmental biosyntheses of peptidoglycan and teichoic acid in Bacillus subtilis. In addition, the current paradigms of B. subtilis wall assembly and structure are distinguished from emerging models of murein insertion and organization. We discuss evidence for the directed, cytoskeleton-dependent insertion of nascent peptidoglycan and the existence of a periplasmic compartment. Furthermore, we summarize the challenges these findings represent to the existing paradigm of murein insertion. Finally, motivated by these new developments, we discuss outstanding issues that remain to be addressed and suggest research directions that may contribute to a better understanding of cell wall assembly in B. subtilis.

Introduction

The bacterial cell wall is a fascinating structure that is simple in design and yet sophisticated and intricate in assembly. This structure is so invaluable to bacteria that historically it has been the key target of therapeutic antimicrobials. It is also considered by some to be one of the defining features of the entire domain Bacteria (see discussion in Koch, 2003). Designed to withstand turgor pressure exerted upon the bacterial cell, the cell wall must be rigid to provide physical integrity to the cell. Seemingly in contrast, the cell wall must also be porous to allow diffusion of large macromolecules and degradable to allow cell expansion. These properties are provided by several cell wall designs in the domain Bacteria, most notably those of the Gram-negative and the Gram-positive bacteria.

The Gram-positive bacterial cell wall is mainly comprised of a thick murein layer with covalently bound anionic polymers that provide the cell with structural integrity and defines its shape. Though it has been the subject of lengthy study, renewed interest in the bacterial cell wall has arisen because highly complex processes involved in the assembly of the Gram-positive cell wall are now coming to light. In fact, as outlined below, new research into cell wall ultrastructure may challenge our definition of the Gram-positive bacterium.

Although there is clearly great variety in the cell wall structures of Gram-positive bacteria, this review focuses on the cell wall assembly of vegetative Bacillus subtilis. More specifically, it covers the synthesis and assembly of peptidoglycan and anionic polymers, the major components of the Gram-positive cell wall. The distinct cellular localizations of precursor synthesis and polymer assembly are highlighted along with recent models of how information is conveyed between these cellular compartments. Emerging concepts in murein insertion and wall organization that challenge our suppositions of the Gram-positive cell wall are also discussed.

Cell wall components

The cell wall in B. subtilis for all its complexity is comprised of mostly three components, peptidoglycan, anionic polymers and selected proteins. The latter component has been estimated to be ∼10% of total cellular protein (Merchante et al., 1995), although proteomic analysis of the cell wall binding proteins identified only 11 proteins under specified conditions (Antelmann et al., 2002). In contrast, peptidoglycan and anionic polymers constitute the bulk of the cell wall and are present in approximately equal proportions. The latter fact may be surprising to some given that peptidoglycan has long been regarded as the predominant component of the cell wall. This misconception may have been strengthened by the plethora of antibacterial compounds targeting peptidoglycan synthesis, e.g. penicillin, bacitracin, cycloserine and fosfomycin, etc., and the paucity of compounds thought to target anionic polymer biosynthesis, e.g. only tunicamycin (Pooley and Karamata, 2000) and daptomycin (Canepari et al., 1990), although the mechanisms of the latter are not without controversy (Laganas et al., 2003; Bouhss et al., 2004).

Peptidoglycan

Peptidoglycan forms a three-dimensional meshwork that surrounds the cell and provides the structural integrity required for cell survival. The rigidity of the mesh is imparted by the simple alternation of N-acetylmuramic acid (MurNAc) and N-acetylglucosamine (GlcNAc) residues joined by β1,4 glycosidic linkages into glycan strands. Short peptide chains, usually comprised of four alternating l- and d-amino acids and referred to as stem peptides, are covalently bound to the carboxyl group of MurNAc. As the glycan strands adopt a helix-like conformation, with four disaccharide units per turn, the stem peptides extend radially from the glycan strand with each stem peptide oriented perpendicular to the preceding stem peptide (Burge et al., 1977). Stem peptides on adjacent glycan strands can form peptide bonds, or cross-bridges, thereby creating a three-dimensional meshwork of strands. In this manner glycan chains from many layers become covalently linked to form a supramolecular structure.

The cross-bridges also impart a degree of flexibility to the cell wall. Lapidot and Irving (1979a,b) conducted 15N NMR studies of intact and isolated Gram-positive cell walls and showed that while the glycan constituents and stem peptides were relatively rigid, the cross-bridging peptide linkages (especially the pentaglycine bridge of staphylococcal peptidoglycan) exhibited relatively large movements. It is hypothesized that this flexibility allows changes in the packing of the glycan strands in response to cell growth (Lapidot and Irving, 1979a,b). A similar interpretation of peptide side-chain flexibility was made on the basis of X-ray diffraction studies of bacterial peptidoglycan (Burge et al., 1977).

The MurNAc-GlcNAc disaccharide backbone is highly conserved among bacterial species. In contrast, the individual components of the stem peptide and cross-bridge can vary, although a di-amino acid, e.g. lysine or diaminopimelic acid, and a terminal dipeptide, e.g. d-ala-d-ala, are required for cross-bridge formation. The amino acid residues of peptidoglycan, owing to their stereo configuration, require specialized enzymes for degradation, perhaps a mechanism to avoid accidental cleavage of the peptidoglycan by extracellular proteases.

While considerable knowledge has been amassed over decades of research regarding the general role of peptidoglycan as a stress bearing structure important in cellular integrity, remarkably little is known about how this structure is assembled and integrated among wall components. Herein we explore exciting and emerging work aimed at an understanding of these poorly characterized aspects of cell wall biogenesis.

Anionic polymers

Anionic polymers are comprised of the wall teichoic acids (WTA) and the structurally related lipoteichoic acid (LTA) that is anchored in the plasma membrane (Archibald et al., 1993). WTA, the major class of anionic polymers in B. subtilis, is a phosphate-rich polymer that is covalently bound to peptidoglycan. WTA polymer structure can vary in different strains of B. subtilis: strain 168 has a glycerol phosphate (GroP) polymer, whereas strain W23 has a ribitol phosphate (RboP) polymer (Burger and Glaser, 1964; Glaser, 1964).

A so-called ‘minor teichoic acid’ polymer (Duckworth et al., 1972; Shibaev et al., 1973), comprised of a glucose-N-acetyl-galactosamine-phosphate repeating unit has also been reported. Assembly of this minor teichoic acid polymer is not well characterized but is known to be catalysed by enzymes encoded in the gga locus, presumably using nucleotide-activated sugar precursors (Estrela et al., 1991). The synthesis of this polymer is impaired at elevated temperature although the cause of this phenomenon has not been identified (Boylan et al., 1972).

Under limiting phosphate, WTA gives way to the incorporation of an anionic polymer, teichuronic acid, that is devoid of phosphate residues (Ellwood and Tempest, 1972). Little is known about the molecular details of assembly of this polymer; however, the tua gene cluster has been identified in B. subtilis 168 and implicated in its synthesis (Ward, 1981; Soldo et al., 1999). Convention had held that WTA is entirely replaced by teichuronic acid under phosphate-limiting conditions. Recently, we demonstrated that not only was teichoic acid synthesis essential for growth in B. subtilis under phosphate-limiting conditions but that significant amounts of WTA persisted under these conditions (Bhavsar et al., 2004). The latter point remains a controversial subject in the phosphate-limitation literature (see the discussion in Bhavsar et al., 2004). Nevertheless, this work suggested that WTA synthesis was an essential function that could not be replaced by teichuronic acid biogenesis.

Our biochemical understanding of the synthesis of the second major class of anionic polymers, LTA, is still limited as much of the LTA research has focused on the immunogenic properties of this compound (reviewed in Fournier and Philpott, 2005). However, it is known that this glycerol phosphate polymer is anchored to the membrane via a gentibiosyl-diacylglycerol moiety with the substituted polymeric region extending into the cell wall matrix (Price et al., 1997; Jorasch et al., 1998). Steps involved in the polymerization of LTA have yet to be identified. However, it is interesting to note that the stereochemistry of the glycerol phosphate precursors of the LTA polymer are opposite those of the WTA polymer, i.e. the monomeric units of LTA and WTA are l-glycerol-1-phosphate (S-glycerol phosphate) and l-glycerol-3-phosphate (R-glycerol phosphate) respectively. Pulse chase analysis of Streptococcus sanguis and Staphylococcus aureus cells implicated phosphatidylglycerol as the precursor for LTA, in contrast to the nucleotide-activated precursor, CDP-glycerol, for WTA (Emdur and Chiu, 1974; Glaser and Lindsay, 1974). To date the best-studied aspect of LTA synthesis is its substitution with d-alanine. This modification has been shown to be important for the regulation of autolytic activity, protection from antibiotic action and even elicitation of immune response in pathogenic Gram-positive species (reviewed in Neuhaus and Baddiley, 2003).

By far the most puzzling aspect of anionic polymer biosynthesis is the function of these abundant structures. In this review, we focus on WTA in particular. We outline paradoxes around its dispensability and highlight deficiencies in the current understanding of the synthesis and assembly of teichoic acid into cell wall.

Precursor biosynthesis

The initial steps of WTA and peptidoglycan polymer biosynthesis show paradigmatic similarity (see Fig. 1). Due to a requirement for nucleotide-activated sugars and soluble amino acids, both polymers are initially synthesized in the cytoplasmic compartment but culminate in the membrane. Both utilize isoprenoid-derived undecaprenyl moieties in their terminal products. However, one major difference in the biosynthesis pathways of these polymers is their site of polymerization. In the case of WTA, polymerization of the glycerol phosphate chain occurs intracellularly, whereas peptidoglycan polymerization occurs on the extracytoplasmic face of the membrane.

Figure 1.

Transcompartmental biosynthesis of teichoic acid and peptidoglycan. The biosynthesis of peptidoglycan and teichoic acid precursors is depicted on the left and right side of the figure respectively. Solid arrows denote catalysis by the indicated enzyme. Unknown enzymatic reactions are denoted by question marks. Block arrows denote where activated substrate is utilized in the biosynthesis pathway. The convergence of both pathways on the common substrate undecaprenyl-phosphate is highlighted in the black rectangle. The distinction between the cytosolic, membrane and cell wall compartments of the cell are indicated by the dashed grey line. Note that polymerization of teichoic acid occurs intracellularly, whereas polymerization of peptidoglycan occurs extracellularly.

Murein precursor (Lipid II) biosynthesis

Genes involved in the intracellular synthesis of the peptidoglycan precursor, annotated mur(AA)BCDEFG and mraY, have been identified and biochemical roles for each gene product have been demonstrated in vitro. The peptidoglycan precursor, Lipid II, is formed stepwise, beginning with the conversion of UDP-N-acetylglucosamine (UDP-GlcNAc) to UDP-N-acetylmuramic acid (UDP-MurNAc)-pentapeptide (see Fig. 1). The latter molecule is formed by the initial synthesis of UDP-MurNAc from UDP-GlcNAc and phosphoenolpyruvate in reactions catalysed by MurAA and MurB. The pentapeptide moiety is added in successive steps by MurC, MurD, MurE, and MurF that add l-ala, d-glu, diaminopimelic acid (DAP), and the dipeptide, d-ala-d-ala respectively. MurNAc-pentapeptide-1-phosphate is then translocated to the membrane compartment via covalent bond formation with a membrane-embedded undecaprenyl-phosphate molecule to form Lipid I (see Fig. 1). This reaction is catalysed by MraY. The synthesis of Lipid II is completed by the addition of GlcNAc from UDP-GlcNAc to the 4-hydroxyl of MurNAc, catalysed by MurG (see Fig. 1). The substantial literature on the initial steps of peptidoglycan synthesis has been summarized elsewhere (Archibald et al., 1993; Foster and Popham, 2002) and a comprehensive review of Lipid II biosynthesis has been published (van Heijenoort, 2001a).

Anionic polymer biosynthesis

Our understanding of anionic polymer biosynthesis lags behind that of peptidoglycan biosynthesis. The major WTA biosynthetic genes have been identified in B. subtilis 168, tagABDEFGHO, and B. subtilis W23, tarABIJKLDFO (Mauel et al., 1991; Lazarevic et al., 2002; Soldo et al., 2002a; Minnig et al., 2005). Of the seven genes identified in strain 168, one is dispensable (tagE) and one unstudied (tagA). Precise deletion mutants conditionally complemented by an ectopically cloned inducible gene copy revealed that tagD, tagB and tagF are indispensable for cell viability (Bhavsar et al., 2001; 2004). In addition, transcriptional repression of either tagO or tagGH also revealed these genes to be indispensable for growth (Lazarevic and Karamata, 1995; Soldo et al., 2002a).

Discovery of the tag and tar gene clusters, in addition to early biochemical studies of crude enzymatic preparations derived from temperature-sensitive mutant strains, led to functional predictions for most of the tag/tar gene products (Lazarevic et al., 2002). It is noteworthy that in vitro biochemical investigations of recombinant enzymes have only recently begun to probe the biosynthesis of teichoic acid and have been restricted to TagD, the CDP-glycerol pyrophosphorylase (Park et al., 1993), MnaA, the UDP-GlcNAc 2-epimerase (Soldo et al., 2002b), TagB, the ‘Tag primase’ (Bhavsar et al., 2005), and TagF, the ‘Tag polymerase’ (Schertzer and Brown, 2003). The demonstration of in vitro biochemical activity using recombinant enzymes notwithstanding, much work remains with respect to the latter two enzymes. TagB and TagF are remarkable in that they catalyse unique chemical reactions and share no homology to characterized enzymes. Detailed in vitro biochemical characterization of this potentially novel class of enzymes, e.g. glycerol-3-phosphate primase and polymerase, is fraught with significant technical hurdles owing to their membrane-bound substrates and will be further discussed below.

Models of teichoic acid biogenesis are rooted in analyses of the teichoic acid gene clusters with the benefit of many years of physiological, analytical and genetic work on teichoic acid. Nevertheless, some significant gaps remain in our understanding of the synthesis of this polymer. According to proposed models of teichoic acid biosynthesis (Pooley and Karamata, 1994; Neuhaus and Baddiley, 2003), the teichoic acid polymer is synthesized on a membrane-embedded undecaprenyl-pyrophosphate-linked GlcNAc-N-acetylmannosamine (ManNAc) disaccharide (see Fig. 1). The disaccharide moiety is thought to be produced by the successive action of TagO, proposed to transfer GlcNAc-1-P from UDP-GlcNAc to undecaprenyl-phosphate, and TagA, believed to transfer ManNAc from UDP-ManNAc to the 4-hydroxyl of GlcNAc. UDP-ManNAc was shown to be the MnaA-catalysed epimerization product of UDP-GlcNAc (Soldo et al., 2002b). A single glycerol phosphate (GroP) is added to the 4-hydroxyl of ManNAc by TagB and is subsequently used by TagF as a primer for the synthesis of the GroP polymer. Glucosylation of the polymer by TagE and polymer export through the ABC-type transporter, TagGH, complete intracellular teichoic acid biosynthesis.

Outstanding issues

Lipid II biosynthesis and export

The enzymatic reactions involved in Lipid II biosynthesis have been studied biochemically and, in fact, can be reconstituted in cell-free systems to synthesize peptidoglycan (Wong et al., 1998; Barbosa et al., 2002). It is important to note that the detailed characterization of Lipid II biosynthesis applies, for the most part, to the cytoplasmic steps of this process. Recent successes with recombinant expression and purification of B. subtilis MraY (Bouhss et al., 2004) and Escherichia coli MraY (Lloyd et al., 2004) are likely to shed greater light on the mechanistic details of Lipid I formation. In contrast, our understanding of the mechanism of Lipid II movement across the cell membrane remains unclear (see Fig. 1). The lag in biochemical investigation of the membrane-localized enzymes relative to those localized in the cytoplasm highlights the difficulty of these types of studies – a problem that arguably plagues the field of teichoic acid biogenesis to an even greater extent (see below). Advances in the chemical synthesis of soluble analogues of Lipid I (Men et al., 1998; Cudic et al., 2001) hold particular promise for the careful study of MurG and MraY. Nevertheless, a full understanding of the final membrane-linked steps of peptidoglycan biosynthesis will surely require in vitro systems that faithfully reconstitute the spatial and interfacial properties of the native environment. Indeed, recapitulation of the system may include multienzyme complexes that are increasingly thought to be important to cell wall synthesis in bacteria (see below).

Wall teichoic acid biosynthesis and export

Arguably the most pressing question in the teichoic acid research field is its physiological function. To date no definitive function has been attributed to teichoic acid and this fact, in conjunction with its indispensable phenotype; frame an interesting problem in Gram-positive physiology. In B. subtilis 168 at least 10 biosynthetic enzymes, the majority of which are genetically organized in divergent transcriptional units and subject to complex regulation (Mauel et al., 1994; Liu et al., 1998; Howell et al., 2003), catalyse the formation of a polymer that comprises up to half of the cell wall material. Despite all of the investigation into the chemical nature and biosynthesis of teichoic acid, its indispensability cannot be attributed to anything beyond its localization to the cell wall. However, the literature contains speculation of specific function(s) for teichoic acid in many diverse physiological processes, e.g. cation chelation, autolysin regulation, cell elongation and wall protonation (reviewed in Archibald et al., 1993). Recently, Matias and Beveridge (2005) showed that teichoic acid plays a crucial role in conveying rigidity to the cell wall because isolated teichoic acid-extracted cell walls had increased flexibility. This finding corroborates those made after examination of teichoic acid-extracted bacterial cell walls by X-ray diffraction (Burge et al., 1977).

The problem outlined in the preceding paragraph has become a legitimate paradox in Staphylococcus aureus with the recent report of a viable deletion mutant of the tagO orthologue in this organism (Weidenmaier et al., 2004). The authors carried out a preliminary analysis of this strain and found that it was devoid of teichoic acid. This surprising finding is in stark contrast to the indispensability of tagO in B. subtilis (Soldo et al., 2002a) and the indispensability of the teichoic acid biosynthetic gene tagF in another staphylococcal strain, Staphylococcus epidermidis (Fitzgerald and Foster, 2000). The current tenet of teichoic acid indispensability in Gram-positive organisms that stems from years of research describing WTA essentiality in the model Gram-positive organism B. subtilis (Boylan et al., 1972; Briehl et al., 1989; Mauel et al., 1989; Bhavsar et al., 2001; 2004) may now be called into question. Greater detailed examination of teichoic acid gene dispensability and the related cell physiology are required to reconcile this disparity.

In comparison with their soluble counterparts, our understanding of the membrane-localized synthesis reactions of cell wall polymers has progressed at a slower pace. The ideal characterization of an enzymatic reaction utilizes purified recombinant enzyme, pure substrate(s), a robust assay and spectral characterization of the product. In the case of WTA biogenesis the cytosolic enzymes MnaA and TagD have met this standard. However, applying the same level of scrutiny to the WTA polymer assembly enzymes has been especially troublesome. The lipophilic nature of the membrane-embedded substrates has made it difficult to carry out in vitro studies on the recombinant enzymes involved in undecaprenyl-pyrophosphoryl disaccharide synthesis and glycerol phosphate priming/polymerization. The latter steps are of particular interest in that they represent novel reactions, catalysed by prototypical enzymes and are largely uncharacterized.

Until very recently, biochemical studies of the teichoic acid biosynthetic enzymes that catalyse additions to the undecaprenyl group had been performed with crude or partially purified enzyme fractions. A major advancement in the field occurred with the demonstration of glycerol phosphate incorporation onto a membrane-bound acceptor using recombinantly purified TagB and TagF enzymes (Schertzer and Brown, 2003; Bhavsar et al., 2005). These studies have built upon the vintage work of teichoic acid priming and polymerization by utilizing pure recombinant enzymatic preparations and also by analytically characterizing the reaction products. These studies have paved the way for new work directed towards a mechanistic understanding of this prototypical family of enzymes that includes the domain dissection and site-directed mutagenesis studies aimed at understanding teichoic acid priming and polymerization steps (Schertzer et al., 2005).

Just as studies of the final steps of peptidoglycan synthesis have been complicated by technical difficulties associated with structurally complex, membrane-bound substrate, so too have studies of teichoic acid biogenesis been limited by the fact that they rely on membrane or a purified, but uncharacterized, component therefrom (Murazumi et al., 1981) as the source of undecaprenyl substrates. Perhaps the biggest advance for the study of teichoic acid assembly enzymes will be the production and utilization of soluble undecaprenyl analogues to allow in vitro study of reactions that normally act upon membrane-embedded substrates. Such analogues have been synthesized for use in the study of both peptidoglycan (Men et al., 1998; Cudic et al., 2001) and lipopolysaccharide (Montoya-Peleaz et al., 2005) biosynthetic enzymes that utilize undecaprenyl-pyrophosphoryl-linked substrates. Indeed, very recently Walker’s group has demonstrated the utility of this approach, chemi-enzymatically synthesizing a soluble analogue of GlcNAc-diphospho-undecaprenol that replaces the C55 undecaprenyl group with a C13 moiety (Ginsberg et al., 2006). This substrate was reacted with recombinantly purified B. subtilis TagA – the first reported in vitro activity for this enzyme – and the product was subjected to spectral characterization. Furthermore, the TagA reaction product was used as a substrate for recombinantly purified TagB, verifying the glycerophosphotransferase activity previously reported (Bhavsar et al., 2005). The use of a soluble acceptor in high concentration translated into a reaction turnover number that was vastly higher to that reported with endogenous substrates (Bhavsar et al., 2005; Ginsberg et al., 2006) and holds tremendous potential for the detailed characterization of enzymes that can utilize this undecaprenol analogue, e.g. TagO, TagA, TagB and TagF.

Lipoteichoic acid biosynthesis

In contrast to the emerging studies with recombinant WTA synthesis enzymes, virtually nothing is known about the enzymes that catalyse LTA synthesis. A paucity of genetic mutants affected in LTA synthesis has precluded even the most rudimentary of speculations as to how this polymer is assembled. The only characterized genes involved in the LTA biogenesis pathway are ypfP, a nonessential gene that is responsible for the incorporation of the gentiobiosyl portion of the membrane anchor (Price et al., 1997; Jorasch et al., 1998); and the dlt operon, whose dispensable gene products are responsible for the d-alanylation of LTA (Perego et al., 1995). Conversely though, the same paucity of genetic mutants that confound our understanding of LTA assembly also lends credence to the hypothesis that this polymer has an essential role in Gram-positive physiology.

Extracellular assembly of the cell wall

Following their intracellular synthesis the Lipid II and anionic polymer cell wall precursors must be organized and functionally assembled (Fig. 1). Undoubtedly, cell wall assembly will be the subject of renewed interest as a result of several controversial proposals in the areas of cell wall organization and architecture. In the following sections we review the conventional understanding of cell wall assembly, novel research that questions this understanding, and outstanding issues that require further investigation.

Assembly of WTAs

Following polymerization of teichoic acid in the cytosol on a lipid anchor, the entire polymer is proposed to move to the outer face of the cell membrane via an ABC transporter-mediated process. The hypothesis is based on the similarity of TagGH to well-characterized bacterial polysaccharide transporters and the phenotype of a conditional tagGH mutant that is consistent with lesions in teichoic acid synthesis (Lazarevic and Karamata, 1995). The subsequent step of covalent attachment to the 6-hydroxyl of muramate remains poorly described, both biochemically and genetically. Venerable work suggests that this covalent attachment occurs concomitantly with glycan strand synthesis (Mauck and Glaser, 1972).

Assembly of murein

Following extrusion through the membrane, Lipid II is assembled into glycan strands by transglycosylation to the reducing end of the growing strand (Ward and Perkins, 1973), resulting in the liberation of an undecaprenyl-pyrophosphate molecule (see Fig. 1). This glycosyltransfer reaction can be catalysed by the Class A high molecular weight penicillin binding proteins (HMW PBP). These are bifunctional proteins also possessing a carboxy-terminal transpeptidase domain. There are no reported monofunctional glycosyltransferases in B. subtilis (McPherson and Popham, 2003). The reader is referred to a more detailed review of the glycosyltransferase reaction (van Heijenoort, 2001b). Transpeptidation of the pentapeptide moieties are catalysed by Class A or Class B HMW PBPs. The Class B HMW PBPs possess an uncharacterized amino terminal domain, in addition to a carboxy-terminal transpeptidase domain (Ghuysen, 1991).

There are four Class A HMW PBPs (PBP1a, PBP1b, PBP2c and PBP4), five putative Class B HMW PBPs (PBP2a, PBP2b, PBP3, PbpH and YrrR), and three low molecular weight (LMW) PBPs (PBP5, PBP4a and PbpX) in vegetative B. subtilis. Interestingly, McPherson and Popham reported the creation of a viable B. subtilis strain that contained simultaneous deletions in all four of the Class A HMW PBPs theoretically rendering this strain unable to polymerize peptidoglycan strands (McPherson and Popham, 2003). However, this surprising achievement was explained when the authors demonstrated that an unknown glycosyltransfer activity remained detectable in this quadruple mutant strain.

Murein insertion, i.e. Lipid II polymerization and incorporation into the cell wall matrix in B. subtilis, occurs along the cell cylinder and at the cell septum. Ironically, the cell pole, originally synthesized as a septum, is rather metabolically inert. Experiments that utilized a variety of techniques to visualize cell wall insertion and turnover (Birdsell et al., 1975; Pooley, 1976a,b; Anderson et al., 1978; Mobley et al., 1984; Merad et al., 1989) led in part to the ‘inside-to-outside’ model of cell growth (Koch and Doyle, 1985). Briefly, this model states that murein insertion occurs in a relaxed state, randomly along the cell cylinder and adjacent to the cell membrane. Older strands of stressed peptidoglycan are turned over at the cell surface and the newer peptidoglycan strands migrate towards the periphery of the cell wall, thereby undertaking more stress due to turgor pressure. In the following sections we discuss new concepts of murein assembly that challenge aspects of the ‘inside-to-outside’ growth paradigm.

Rethinking the ‘inside-to-outside’ growth paradigm

Cytoskeleton-directed murein insertion

One of the most significant advances in microbial biology in recent years has been the discovery of broadly conserved cytoskeletal elements in bacteria. These include the bacterial counterparts to actin, tubulin and intermediate filaments identified as MreB, FtsZ and crescentin respectively (Moller-Jensen and Lowe, 2005). Interestingly, in B. subtilis MreB is one of the three proteins with homology to actin, the other two being Mbl and MreBH. The MreB and Mbl proteins that are thought to exert control over cell width and length, respectively, are surprisingly dispensable in B. subtilis (Jones et al., 2001; Formstone and Errington, 2005). Both MreB and Mbl proteins have been shown by fluorescence microscopy to localize in helical filaments proximal to the cytoplasmic leaflet of the cell membrane of B. subtilis (Jones et al., 2001) and similar results have been reported in other organisms (Kruse et al., 2003; Shih et al., 2003; Figge et al., 2004). Indeed, recently the MreB orthologue from Thermotoga maritima was shown to self-assemble and possess ATPase activity in a manner similar to actin (Esue et al., 2005).

A connection between the putative microbial cytoskeleton and cell wall assembly was forged when it was observed that nascent peptidoglycan, monitored using a fluorescent vancomycin derivative, was incorporated into the cell wall in a helical pattern similar to that formed by Mbl (Daniel and Errington, 2003). Further investigation revealed that nascent peptidoglycan incorporation into the cell cylinder was in fact dependent on the presence of helical Mbl filaments, but not MreB (Daniel and Errington, 2003; Formstone and Errington, 2005). In the absence of Mbl, however, wall incorporation continued at septal sites. Remarkably, a model for the helical assembly of cell surface components was originally proposed nearly 30 years ago by Mendelson (1976). This hypothesis followed the isolation of a germination mutant in B. subtilis, impaired for release from the spore coat that grew with unusual double helix morphology.

Helical models for cell elongation are consistent with a controlled and spatially restricted mechanism of nascent peptidoglycan incorporation into the cell wall and are reminiscent of the non-random insertion model proposed by Schlaeppi et al. (1982). This model was based upon the segregation of nascent peptidoglycan units after supplementing B. subtilis with radiolabelled cell wall precursors. Models of non-random nascent peptidoglycan insertion are at odds with the paradigmatic view of random insertion described by Koch and Doyle (1985). However, it is interesting to note that Koch was prescient in his caveats of the ‘inside-to-outside’ nascent peptidoglycan insertion model. He suggested that zonal, i.e. non-random, cell cylinder extension would require cell shape control such as that afforded by a stabilizing cytoskeleton (Koch, 1988). In light of the discovery of the microbial cytoskeleton in Gram-positive and Gram-negative rod-shaped bacteria, the ‘inside-to-outside’ model of cylinder growth must be re-examined.

Experiments using Caulobacter crescentus have implicated the MreB homologue in this organism as a key determinant of peptidoglycan insertion along the cell cylinder or at the cell septum (Figge et al., 2004). These authors suggest that in C. crescentus MreB directs the site of peptidoglycan incorporation via a direct interaction with PBPs. A mechanism by which the intracellular actin-like cables (MreB in E. coli and Mbl in B. subtilis) convey information to the extracellular PBPs has recently been suggested with the demonstration that the MreC and MreD proteins, whose structural genes are usually found operonic with mreB, form a complex in the cell membrane that can interact with MreB or Mbl in E. coli or B. subtilis respectively (Kruse et al., 2005; Leaver and Errington, 2005; Stewart, 2005). According to the resulting model MreB/Mbl designates the site of nascent peptidoglycan insertion and transmits that information to the requisite PBP across the membrane via MreCD. Thus, it appears that the newly discovered cytoskeletal elements of microbes may direct a pivotal process in the assembly of cell wall.

New evidence for a Gram-positive periplasm

As outlined above, recent technological advances have challenged tenets surrounding cell wall insertion in B. subtilis by allowing the visualization of bacterial cytoskeletal elements and their influence on murein insertion. Similarly, an unconventional method of visualizing B. subtilis ultrastructure was recently reported by Matias and Beveridge (2005) and has provided data that challenge paradigms of Gram-positive cell wall organization. These authors employed a cellular vitrification and non-dehydrative sample preparation strategy for cryo-transmission electron microscopy (cryo-TEM) examinations of B. subtilis 168. The advantage of this ‘frozen-hydrated’ technique is that the image contrast does not rely on heavy metal stains that are known to have varying affinities for cellular structures. Instead, the image contrast in ‘frozen-hydrated’ cryo-TEM is derived from the differential density of two regions and thus, the contrast is an indication of mass distribution within a given area (Matias and Beveridge, 2005). The authors argue that this preparative method likely preserves native cell wall architecture because vitrified cells can resume metabolism upon thawing.

Cryo-TEM of ‘frozen-hydrated’B. subtilis cells has yielded some remarkable and novel observations regarding wall ultrastructure. The cell wall of ‘frozen-hydrated’B. subtilis appears biphasic, possessing an ‘inner wall zone’ and ‘outer wall zone’. Teichoic acid and peptidoglycan were inferred to be localized to the higher density ‘outer wall zone’, based in part upon the examination of isolated cell walls that retained cell shape in the absence of the ‘inner wall zone’. The ‘inner wall zone’ was characterized by lower density and ability to accommodate membranous blebs under plasmolysis conditions (Matias and Beveridge, 2005). These findings led the authors to make the provocative hypothesis that the ‘inner wall zone’ demarcates a 22 nm periplasmic space in B. subtilis (see Fig. 2). The notion of a periplasmic space in B. subtilis has been previously suggested, though not observed, and in fact a specific profile of proteins and enzyme activities has already been defined for the B. subtilis‘periplasm’ (Merchante et al., 1995). The specificity of its contents, as well as its discrete localization, suggests that the periplasmic space can be considered a unique cellular compartment. Interestingly, an inner wall zone with low mass density would seem to be at odds with the ‘inside-to-outside’ model of murein insertion that predicts the existence of a highly cross-linked layer of unstressed murein lying adjacent to the outer face of the cell membrane (Holtje, 1998). Such a layer should presumably have been a prominent feature under ‘frozen-hydrated’ conditions. Thus, these new ultrastructural findings seem not to support a model for peptidoglycan assembly that depends on random insertion of an unstressed murein along the cell cylinder. Indeed, the existence of a large periplasmic compartment is arguably consistent with a model for cytoskeletal-directed assembly and insertion of newly synthesized peptidoglycan. In the latter example, there surely would be a need for localized, extracytoplasmic, macromolecular complexes involved in peptidoglycan assembly. A large periplasmic region could afford the required space for this biosynthetic machinery. Very recently, cryo-TEM examination of S. aureus under ‘frozen-hydrated’ conditions revealed similar wall architecture to that found in B. subtilis suggesting that S. aureus also possesses a periplasmic space (Matias and Beveridge, 2006) strengthening the need for a re-examination of Gram-positive cellular architecture.

Figure 2.

Schematic representation of a revised model of B. subtilis cell wall assembly. This conceptual figure depicts a magnified view of B. subtilis in cross-section. The cross-sectional plane is indicated by the grey box traversing the B. subtilis cell at the bottom of the figure. This model of cell wall assembly in B. subtilis incorporates the concepts of cytoskeletal-directed murein insertion and the existence of a periplasmic space (denoted in pale yellow). A multiprotein cytoskeletal complex (shown in red) comprised of Mbl, MreC and MreD relays positional information from the cytosol to the murein biosynthetic complex (shown in green) comprised of the PBPs, murein hydrolases (MltA) and scaffolding proteins (MipA) situated in the periplasmic space. Evidence for the protein–protein interactions depicted in this figure is given in the text. LTA (shown in brown) is anchored in the cell membrane (denoted by the lipid bilayer), traverses the periplasmic space and extends into the murein layer (glycan strands and stem peptides are represented by thick black bars and thin black lines respectively). WTA and the PBP transmembrane domain have been omitted for clarity.

Outstanding issues

The aggregate of research findings in support of localized incorporation of nascent peptidoglycan and its dependence on helical cytoskeletal components are compelling. The associated model is not necessarily exclusive of one that would propose the production of unstressed murein and its (localized) incorporation into peptidoglycan in an ‘inside-to-outside’ manner. Nevertheless, the most parsimonious explanation is that nascent peptidoglycan is inserted directly into cell wall in a cytoskeleton-controlled manner. This model invokes extracellular enzyme complexes that direct nascent peptidoglycan incorporation and regulate wall synthesis functions. Below we address the implications and outstanding issues surrounding the murein biosynthetic complex, including its composition, address and regulation.

Murein biosynthetic complex

The concept of localized murein insertion at sites controlled by the cytoskeleton predicts that enzymes critical to peptidoglycan assembly, e.g. PBPs, murein hydrolases and other yet to be identified components, will be in tight association with cell scaffolding proteins, possibly in a large higher order multienzyme complex. Still, it must be emphasized that the formation of a multicompartmental murein biosynthetic supercomplex has yet to be conclusively demonstrated (Fig. 2). Nevertheless, some recent studies are evidence of momentum in this research direction.

With respect to extracytoplasmic interactions among murein assembly proteins, Figge et al. (2004) used co-immunoprecipitation methods in conjunction with fluorescently labelled PBPs (Bocillin) in C. crescentus to demonstrate that three different PBPs co-precipitated with PBP2. Sepharose-immobilized MltA was used by Vollmer et al. (1999) to identify a potential new scaffolding factor from E. coli periplasm that potentiated the interactions of pure recombinant MltA with PBP1B. Similarly, immobilized PBP1C retained two other PBPs and MltA from solubilized E. coli membrane fractions (Vollmer et al., 1999). Thus reports to date suggest that PBPs may well form extracytoplasmic multienzyme complexes.

Regarding the components of a likely helical cytoskeleton, two hybrid experiments in E. coli showed pairwise interactions between MreB-MreC and MreC-MreD (Kruse et al., 2005; Leaver and Errington, 2005). These studies are consistent with fluorescence microscopy experiments using GFP-fused MreC and MreD that showed helical localization patterns reminiscent of Mbl and MreB in B. subtilis (Leaver and Errington, 2005). It should be noted, however, that a very recent report suggested that the periplasmic helical structure comprised of MreC in C. crescentus was, in fact, out of phase with that of the cytoplasmic MreB (Mbl orthologue) helical structure (Dye et al., 2005), although the authors believed that this anti-localization required some communication between MreB and MreC. Furthermore, MreB from C. crescentus cell extracts was not retained on a column containing immobilized C. crescentus MreC (Divakaruni et al., 2005). Nevertheless, in sum, the recent works on cytoplasmic and extracytoplasmic cytoskeleton represent important milestones in understanding the components of what is surely a highly complex scaffold in rod-shaped bacteria.

It is tempting to conclude that studies suggesting the existence of higher order complexes of peptidoglycan biosynthetic machinery and reports of a complex of membrane-localized cytoskeleton elements predict an interaction between these complexes. Clearly, the helical patterns evident in studies with fluorescein-labelled vancomycin (Daniel and Errington, 2003) are suggestive of such an interaction as are recent immunofluorescence experiments that showed MreB-dependent helical localization of PBP2 in C. crescentus (Figge et al., 2004). Although the connection between cytoskeleton and peptidoglycan biosynthesis has been further advanced with recent work in C. crescentus indicating that immobilized MreC retained a variety of PBPs in a detergent solubilized cell lysate (Divakaruni et al., 2005), the connection remains equivocal. Indeed, purified recombinant PBP 1B from E. coli can synthesize muropeptides whose biochemical properties are virtually identical to those isolated from whole cells (Bertsche et al., 2005).

The B. subtilis periplasm

Beveridge’s finding of a bona fide periplasm in B. subtilis could not be timelier. Put into context the 22 nm periplasmic space discovered by Matias and Beveridge (2005) is large enough to accommodate a protein complex the size of the bacterial ribosome. Indeed, it is difficult to imagine how a large murein biosynthetic supercomplex composed of PBPs, hydrolases, all localized on a helical scaffold, could function without a spacious extracellular compartment. Thus the emerging view of cell wall biosynthesis in B. subtilis as the work of a large cytoskeleton-localized extracytoplasmic multienzyme complex is entirely consistent with the concept of a large periplasmic space as an address for wall synthesis.

Inevitably the presence of a periplasmic space raises the question of how it is maintained. Would the turgor pressure inherent in bacteria not compress the cytoplasmic membrane against the cell wall? The cell cytoskeleton could presumably have a structural role in maintaining this space. Another factor could well be the presence of LTA. The periplasmic space of B. subtilis and other Gram-positive bacteria must contain LTA because this polymer is anchored in the cell membrane and extends into the cell wall. Perhaps LTA molecules and other protein components contribute significant osmolarity to the periplasmic space such that the most significant drop in osmolarity occurs across the cell wall. This would put the primary impact of turgor on cell wall and not on the cell membrane.

While the concept of a periplasm in Gram-positive bacteria has been previously raised, its potential importance in Gram-positive physiology has been widely under-appreciated. With the recognition of its existence several new avenues deserve attention. Methods to isolate the components of this compartment and reconstitute their activities are generally lacking. Ongoing high resolution ultrastructural analyses of this space will need to address the localization and structure of individual components thought to reside in the periplasm. This will represent an interesting challenge because the cell wall and periplasmic space are too disordered and large to be studied by the high resolution yielding techniques of X-ray crystallography and nuclear magnetic resonance. However, X-ray crystallography has yielded atomic resolution information on a number of proteins involved in cell wall assembly including murein hydrolytic enzymes and PBPs (Macheboeuf et al., 2005; van Straaten et al., 2005). Given that the murein biosynthetic enzymes are likely organized in multiprotein complexes it will be of great value to study the structural arrangement of the individual protein components within the complex. Cryo-electron microscopy (cryo-EM), which offers resolution limited to approximately 10 angstroms, has been an ideal technique for the study of protein complexes such as the proteasome (Ortega et al., 2005) and could be used to study the structural architecture of a murein biosynthetic supercomplex. The structural information gleaned from crystallographic studies can be modelled into cryo-EM structures, thereby extending information to the organization of individual protein subunits within a larger supramolecular complex. This modelling approach has been successfully performed in the imaging of viral particles (Tang and Johnson, 2002). Nevertheless, the goal will be to determine how the multiprotein biosynthetic complexes are organized within the context of the cell wall and periplasmic compartments and this will require the overarching structural determination of these compartments. Cryo-tomography is a candidate technique for obtaining these structural determinations and has been successfully used to visualize the bacterial cytoskeleton and cytoplasm at a resolution of several nanometres (Grunewald et al., 2003; Kurner et al., 2005). Fortunately, the vitreous freezing and sample preparation techniques developed for the study of ‘frozen-hydrated’ cells by cryo-TEM (Matias and Beveridge, 2005) hold tremendous promise for visualization of the cell wall by tomography (Subramaniam, 2005). In this way it may be feasible to position into the compartmental context of the cell wall the atomic details of individual proteins and their organization within supramolecular complexes. The approach of correlating structural information of varying resolution may be a practical approach to bring to bear on the problem of cell wall structure and may yield important insights into the synthesis, organization and architecture of the cell wall.

Regulation of cell elongation and septation

Well before the suggestion of a cytoskeleton-directed extracellular enzyme complex responsible for cell elongation, a cell scaffold-controlled murein biosynthetic complex was proposed for cell septum synthesis in E. coli and termed the divisome (Nanninga, 1991). The divisome is conceptually very similar to its counterpart in elongation as both are thought to be composed of PBPs and accessory proteins, and associated with a cytoskeleton (Goehring and Beckwith, 2005). One of the most striking differences is the nature of the scaffold where FtsZ is the key cytoskeletal support for septum formation and analogous to the actin-like components Mbl/MreB of the cell cylinder elongation complex. Perhaps most remarkable is the long-held understanding that cell division and cell elongation are competing and exclusive processes in rod-shaped bacteria (Satta et al., 1994).

The mechanism for the switch from cell elongation to division has long been an elusive one. It is tempting to speculate that the co-ordination of cell division and elongation is rooted in compositional differences of the respective multiprotein complexes. This possibility is nicely illustrated in a provocative study by Daniel and Errington (2003) where depletion of Mbl led to a defect in incorporation of peptidoglycan into the cell cylinder but not the cell septum. Thus a co-ordinated switch from elongation to division could presumably occur by decoupling Mbl from the elongation complex and by engaging FtsZ with the divisome. Along these lines, it seems likely that a clearer understanding of the composition and role of the components of the divisome and cell elongation machinery will open the door to a detailed understanding of the switch from cell elongation to division.

Conclusions and future directions

Recent years have produced several exciting insights in the fields of cell wall precursor synthesis, cell wall insertion and cell wall assembly in B. subtilis. As outlined above these include the mechanistic study of teichoic acid synthesizing enzymes, the elucidation of an intercompartmental link between the cell wall and bacterial cytoskeleton and the physical delineation of the B. subtilis cell wall into the wall proper and a periplasmic space. However, these insights have in turn raised more questions about the process of cell wall assembly in B. subtilis. Indeed, we have endeavoured in this review to pose some of these questions and to speculate on their answers.

In this review we have highlighted the progress made in our understanding of the syntheses of peptidoglycan and anionic polymer precursors, due in large part to advances in techniques for the study of reactions localized to the membrane compartment. Furthermore, we have defined the paradox of teichoic acid essentiality in the absence of a clear cellular function for WTA in Gram-positive physiology. We have argued that the discovery of the bacterial cytoskeleton, its interaction with murein biosynthetic complexes and the visualization of a periplasmic space in B. subtilis present a fundamental challenge to the existing paradigm of cell wall insertion and cell elongation in this organism. Finally, we have illuminated research areas of emerging prominence arising from a revised model of cell wall assembly.

Acknowledgements

We would like to thank Dr Terry Beveridge, Dr Murray Junop and Dr Joaquin Ortega for useful discussions and to the anonymous referees for their thoughtful suggestions. We are also grateful for helpful discussions with other members of the laboratory. We apologize to any authors whose work was not cited due to size constraints. Work in our laboratory has been supported by an operating grant (MOP-15496) from the Canadian Institutes of Health Research. E.D.B. holds a Canada Research Chair in Microbial Biochemistry.

Ancillary