The peptidoglycan walls of bacterial cells prevent cell lysis and determine cell shape. For decades, research has addressed basic biological questions concerning not only wall structure and architecture, but also the mechanism of wall assembly. How are existing wall bonds broken in order to allow insertion of new material, and how are synthetic and lytic processes co-ordinated to accommodate cell growth, maintain cell shape, avoid lysis and drive cell division? Peptidoglycan synthesis and structure also has significant applied considerations. For some pathogenic species the wall is the outer layer presented to a host organism, and peptidoglycan can be both a stimulator and a target of the host immune system. Peptidoglycan synthesis is the target of some of our most successful classes of antibiotics, including the β-lactams such as penicillin. Extensive study has revealed reaction details involved in the fine scale polymerization of peptidoglycan, chemical studies have revealed gross variations in structural modifications, and morphological studies have allowed the visualization of the presence, absence, amount and turnover of peptidoglycan in particular cellular locations. However, determination of the three-dimensional arrangement of peptidoglycan strands in vivo has proven extremely challenging. The structure is not regularly repetitive enough to visualize using biophysical methods. Fixation steps required for electron microscopy can alter the structure and may not be fast enough to prevent some degradation. Several recent advances are allowing better visualization of in vivo fine structure and morphological changes.
The basic chemical structure of peptidoglycan is conserved across all bacteria (Vollmer et al., 2008). Glycan strands are polymerized from N-acetylglucosamine-N-acetylmuramic acid disaccharide precursors on the outer surface of the cytoplasmic membrane. Each N-acetylmuramic acid residue carries a short peptide side-chain, and these side-chains are then cross-linked by a transpeptidase to produce a mesh network surrounding the entire cell. While this two-dimensional structure is conserved, the three-dimensional structure can fall into one of two categories corresponding to the Gram-negative and Gram-positive classification of bacteria.
Gram-negative bacteria possess a very thin wall, essentially a single layer of cross-linked peptidoglycan strands, within the periplasmic space between the cytoplasmic and outer membranes (Matias et al., 2003; Gan et al., 2008). The shallowness of this layer indicates that the peptidoglycan strands must lie parallel to one another and that they must lie parallel to the plane of the membrane. Several pieces of evidence indicate that the glycan strands run approximately circumferentially around the long axis of a rod-shaped cell, rather than parallel to the long axis (Fig. 1). First, these cells grow via elongation of the rod, and approximately 50% of the PG is degraded and recycled during each generation (Goodell, 1985). It is easy to model an elongation mechanism of cleaving peptide cross-links between PG strands with insertion and cross-linking of new glycan strands around the circumference of the cell (a process for which many biochemical data exist) (Holtje, 1998). A mechanism that would separate longitudinally oriented glycan strands with equal insertion of new strands around the cell is much harder to imagine. Second, fluorescence microscopy allowing visualization of the movement of proteins involved in PG insertion indicates a spiral movement around the circumference of the wall (van Teeffelen et al., 2011), and probes for newly inserted PG reveal a similar pattern (Kuru et al., 2012). Finally, direct imaging of the surfaces of Gram-negative sacculi reveals apparent PG strands running in a circumferential path (Gan et al., 2008).
Gram-positive bacteria possess PG walls that are 5- to 10-fold thicker than those of the Gram-negatives. Gram-positives shed large amount of PG from their surfaces and generally carry out little or no PG recycling. This has been modelled as an inside-to-outside growth mechanism where nascent PG is incorporated in an unstressed configuration on the inside and moves outward into a stressed layer as the outer layers are degraded and released (Koch and Doyle, 1985). The thickness of the Gram-positive wall can accommodate additional models for the arrangement of the PG strands (Fig. 1). Alternative models, all with some supporting evidence, have included: (i) strands parallel to the membrane (Ghuysen, 1968), either circumferential like that for the Gram-negative wall or potentially longitudinal, (ii) 50 nm coiled cables of PG strands arranged circumferentially (Hayhurst et al., 2008), or (iii) a ‘Scaffold’ model in which short PG strands are arranged radially from the cell, perpendicular to the membrane plane (Vollmer and Holtje, 2001; Dmitriev et al., 2005).
Numerous similarities between the Gram-negative and Gram-positive wall synthetic machinery suggest that the process of peptidoglycan incorporation will be similar. The complement of penicillin-binding proteins (PBPs), the enzymes that polymerize and cross-link the strands, are similar among most species (Goffin and Ghuysen, 1998), and circumferential spiral patterns of ‘Mre’ shape-determining proteins, PBPs, and nascent peptidoglycan have been observed in both Gram-positive and Gram-negative species (Daniel and Errington, 2003; Scheffers and Pinho, 2005; Tiyanont et al., 2006; Dominguez-Escobar et al., 2011; Garner et al., 2011; van Teeffelen et al., 2011). However, the possibility remains that the spiral PG incorporation in Gram-positives is the insertion of new strands into a perpendicular-type structure.
The study of PG synthesis in the Gram-positive species Bacillus subtilis opens the possibility of examining a unique process that takes place during endospore formation (Fig. 2). During this developmental process, two cells cooperate to produce a single spore. The step in sporulation called ‘engulfment’ involves migration of the larger cell's membrane around the smaller cell, resulting in the developing spore being surrounded by two membranes and within the larger cell's cytoplasm. The spore ‘cortex’ peptidoglycan is produced between these two membranes, maintains the high turgor of the dehydrated core in the dormant spore, and is rapidly degraded during spore germination. Previous studies have implicated both PG synthesis and PG-cleaving proteins in the engulfment process (Abanes-De Mello et al., 2002; Broder and Pogliano, 2006; Meyer et al., 2010; Morlot et al., 2010). B. subtilis actually possesses two complementary engulfment-promoting systems, one that requires PG metabolism and one that does not (Broder and Pogliano, 2006). Data are consistent with the presence of a PG layer between the engulfing membrane and the forespore membrane (Fig. 2), but technical challenges have prevented the visual observation or chemical detection of PG in this location.
Three advances are allowing new insights into PG wall structure and metabolism. Electron cryotomography (ECT) (Jensen and Briegel, 2007) allows the visualization in three dimensions of native cell structures, without alterations potentially introduced by the fixation, embedding, sectioning and staining methods associated with traditional electron microscopy. ECT has allowed the examination of PG structures at < 10 nm detail, even revealing PG strand orientation in Gram-negative walls (Gan et al., 2008). Reagents for fluorescent labelling of PG structures, including both total and nascent PG (Kuru et al., 2012), allow the visualization of time-courses of cell structure alterations. Computational molecular modelling of PG structures can allow prediction of the morphological outcomes of different strand arrangements and insertion patterns. These methods have been applied in two articles in this issue of Molecular Microbiology to address outstanding questions about the insertion pattern of PG in B. subtilis cells (and presumably other Gram-positive rods) and the morphology of PG structures produced during endospore formation.
Beeby et al. (2013) examine the arrangement of PG strands in B. subtilis vegetative cells using a combination of ECT and computational modelling of potential PG structures. ECT images of whole cells and purified sacculi revealed a uniform electron density throughout the wall and no evidence of PG ‘cables’ that had been previously postulated based upon atomic force microscopy of B. subtilis walls (Hayhurst et al., 2008). While no orientation of glycan strands could be observed directly using ECT, some novel physical properties of the sacculi were detected. When sacculi were fragmented, they predominantly split in a circumferential pattern and liberated sheets of peptidoglycan that coiled back on themselves, always with the outer face of the PG rolled into the interior of the coil.
Three molecular models of PG were assembled to examine if they behaved like the fragmented sacculi upon relaxation of the stress holding them in a sheet conformation. One model contained multiple layers of circumferentially oriented PG strands, cross-linked both within and between layers. The other two models used PG strands oriented perpendicular to the membrane, in one case with all strands uniformly aligned and spanning the width of the wall, in another case with some strands offset to the inner and outer wall regions to simulate the observed inside-to-outside growth pattern. Each model was stretched to twice its original size to simulate cell turgor and growth. The molecular models were then ‘relaxed’ to simulate the experimental PG fragmentation, and deformation of the modelled PG was observed. The uniform perpendicular model did not exhibit curling and maintained a constant thickness. The circumferential model and the offset perpendicular model both curled towards the outer surface, the curling of the circumferential model more closely matching the experimental observation. The circumferential model also exhibited a significant wall thickening upon relaxation.
The researchers then examined ECT images and observed a similar degree of PG thickening upon fragmentation (relaxation), suggesting that the circumferential model most closely approximates the in vivo situation. To further compare experimental observations with the potential PG strand orientations, the effect of denaturing conditions on sacculus size was observed using ECT. While the cross-linking peptides can be relaxed by denaturing conditions, the linear glycan strands should be much less flexible. Denaturing conditions resulted in an increase in sacculus length but not diameter (or circumference), suggesting that the peptide cross-links are predominantly oriented along the long axis of the cell. This orientation is consistent with circumferential PG strand orientation but not with longitudinal or perpendicular models.
The article by Tocheva et al. (2013) in this issue examines the production of PG structures during the formation of B. subtilis endospores. ECT of sporulating cells allowed the visualization of a thin electron-dense layer resembling PG between the mother cell and forespore membranes early during the process of engulfment. ECT of purified sacculi revealed that this presumptive PG layer became thinner as sporulation progressed, consistent with previously observed septal thinning, and that the edges were connected to the mother cell wall even as the membranes began movement around the forespore (Fig. 2). The use of a method for fluorescent staining of nascent PG indicated that the observed structure was certainly PG and that this PG progressed around the forespore with the engulfing membranes throughout the process. These observations refine a model for the roles of PG synthesis and degradation in the engulfment process (Fig. 2). The authors propose that PG synthesis takes place at the leading edge of the engulfing membrane, with addition of a thin PG layer templated on the mother cell wall. Simultaneously, the back side of this thin layer is cleaved from the mother cell wall by engulfment-promoting PG lytic enzymes, creating a space for the engulfing membrane to move forward. The finding that the PG layer surrounding the forespore is only about 2 nm thick means that this must be a single-layer structure similar to that of Gram-negative cells. This again negates the 50 nm coil model for Gram-positive PG insertion, unless a completely different PG synthesis mode is involved in engulfment than in vegetative wall synthesis, and makes a perpendicular PG insertion model very difficult to imagine.
The thin layer of PG surrounding the forespore may represent all or part of the ‘germ cell wall’ that has been shown to be the first synthesized around the spore (Meador-Parton and Popham, 2000) and that remains after spore germination and cortex degradation (Atrih et al., 1998; Dowd et al., 2008). This initial spore PG layer could also serve as a template for subsequent production of the thicker cortex PG layer. It was previously observed that a strain lacking two PG-synthetic enzymes that are specifically expressed within the developing forespore, PBPs 2c and 2d, produce disorganized cortex PG (McPherson et al., 2001). A possible explanation is that these proteins produce the first layer of PG during engulfment, and that in their absence there is no template for the cortex. In this situation, engulfment would have to proceed via an alternative pathway that does not require PG synthesis (Broder and Pogliano, 2006).
The results of this recent work stimulate new questions and increase the odds of gaining useful answers. An interesting experiment would be the ECT visualization of engulfment in a strain lacking PBPs 2c and 2d, to observe the presence or absence of PG between the engulfing membranes and the production of cortex PG in the absence of a template. While it is increasingly clear that PG is incorporated in a circumferential spiral pattern in both Gram-positives and Gram-negatives, details of this insertion are lacking. Is this incorporation in the form of long continuous strands that are subsequently broken, which would result in straight growth of the rod? Or are PG strands inserted in short stretches, which will require an additional method of spatial regulation to insure straight growth of the wall? The combined use of ECT, fluorescent probes of PG structure, molecular modelling and clever genetics may well advance our understanding of these processes and others in the near future.