Cytoskeletal structures in bacteria play an important role in organizing the enzymatic complexes that remodel the cell wall, a robust peptidoglycan (PG) exoskeleton that protects cells from osmotic pressure, preventing the rupture of the cytoplasmic membrane (CM), and that endows cells with their characteristic shape. It is now established that all three types of cytoskeletal elements, originally defined in eukaryotes, regulate growth of the bacterial PG layer. For instance, the intermediate filament-like crescentin is used as a pressure-generating molecular spring to cause the cell curvature in Caulobacter crescentus by inducing anisotropic growth of the PG (Ausmees et al., 2003; Cabeen et al., 2009). Moreover, the widely conserved tubulin-like FtsZ protein serves as the nucleation point for the PG biosynthetic machinery at the division septum to create two new cell poles (Margolin, 2005). Finally, helical ribbons of the actin-like MreB line the inner face of the CM to position PG biosynthetic enzymes along the cylinder long axis (Shaevitz and Gitai, 2010; White et al., 2010).
The PG layer is composed of densely repeated disaccharide N-acetyl glucosamine-N-acetyl muramic acid strands cross-linked to one another by pentapeptide side-chains. Gram-positive bacteria feature a multilayered PG linked to the CM via embedded teichoic acid moieties. By contrast, the single PG layer of Gram-negative bacteria is covalently riveted to the outer membrane by Braun's (Lpp) and other PG-binding lipoproteins (Silhavy et al., 2010). Because gaps in the PG layer can result in lysis, cells face the challenge of extending and splitting the PG layer during growth and division, respectively, without compromising its integrity. Therefore, the enzymes responsible for remodelling of the PG layer must be under tight spatio-temporal control. Penicillin binding proteins (PBPs), the famous cellular targets of many β-lactam antibiotics that are in clinical use, play an important role in PG remodelling and synthesis (Kohanski et al., 2010). Class A (also known as high molecular weight) PBPs are the primary cellular PG synthetases, possessing both transglycosylation and transpeptidation activity for the polymerization and cross-linking of new PG strands, respectively. By contrast, the class B PBPs are solely monofunctional transpeptidases. Aberrant localization and/or unrestrained activity of these bifunctional PG enzymes can have a detrimental effect on cell morphology and viability. For example, when the localization of the bifunctional PBP1 enzyme along the Bacillus subtilis cell cylinder and at the septum is perturbed by inactivation of MreB, cells swell and eventually rupture. These phenotypes are mitigated by compensatory mutations in PBP1, indicating that proper topological control of PBP1 is critical for the integrity of the PG layer (Kawai et al., 2009).
Enter Foulquier et al. (2011) who uncover an analogous role for YvcK, a B. subtilis homologue of the CofD family of enzymes that are involved in the biosynthesis of the coenzyme F420 in archaea and high G+C Gram-positive bacteria. B. subtilis does not possess coenzyme F420, yet YvcK is essential for growth on certain carbon sources. While YvcK- cells grow normally on media with glycolytic carbon sources, they die in medium favouring gluconeogenesis. Switching YvcK- cells from glycolytic to gluconeogenetic growth is accompanied with severe shape defects, causing the cells first to bulge, swell and finally burst. These phenotypes resemble the defect in PG homeostasis exhibited by MreB- cells. Indeed, akin to the MreB- mutant, the shape and growth defect of the YvcK- mutant is suppressed by supplementing the medium with Mg2+ (Gorke et al., 2005). This déjà vu prompted the authors to test if other parallels exist between YvcK and MreB. In situ and in vivo they observed YvcK within two interwoven helical assemblies that stretch from pole to pole beneath the CM in a pattern strikingly similar to the known subcellular disposition of MreB (Fig. 1). However, careful deconvolution fluorescence microscopic analysis resolved the MreB and YvcK helices to separate structures. In support of this, genetic analyses showed that the integrity of each structure does not depend on the presence of the other. Strikingly, however, under conditions of overexpression one protein can functionally compensate for the other's absence.
How might such suppression occur? By analogy to the mislocalization of PBP1 by MreB (Kawai et al., 2009), the YvcK- phenotype seems to be caused by ectopic localization of PBP1. First, as shown for MreB, Foulquier et al. (2011) found that inactivation of PBP1 also corrects the YvcK- phenotype. Second, in misshapen YvcK- cells, PBP1 clusters at aberrant cellular positions. Third, overexpression of MreB (but not overexpression of the MreB paralogs MreBH or Mbl) or growth on glycolytic substrates restores PBP1 localization in YvcK- cells. Finally, YvcK overexpression also corrects the aberrant localization of PBP1 in MreB- cells. How YvcK controls the localization of PBP1 is unknown, but given that MreB can directly associate with PBP1 presumably via the N-terminal cytoplasmic tail (Kawai et al., 2009), it is conceivable that YvcK targets the same determinant. By analogy, the Gram-negative bacterium C. crescentus positions the bifunctional PBP PbpC specifically at the old cell pole by way of a direct interaction of the bactofilin cytoskeletal element with the cytoplasmic tail of PbpC (Kuhn et al., 2010). Based on these findings it appears that the organization of bifunctional PBP's is generally mediated through cytoskeletal elements. Interestingly, two landmark papers recently reported that the activities of bifunctional PBP1A and PBP1B are regulated by cognate outer membrane lipoproteins in Escherichia coli (LpoA and LpoB, respectively) (Paradis-Bleau et al., 2010; Typas et al., 2010). Thus, the regulation of bifunctional PBP's can occur from both flanking envelope layers.
Returning to the mechanism of suppression of the MreB- phenotype by extra YvcK, it is noteworthy that suppression can occur even on medium containing glycolytic carbon sources. As these conditions do not result in a growth and shape defect of YvcK- cells, might thus an imbalance(s) in the intracellular levels of a glycolytic intermediate underlie the PBP1-mediated shape defect and/or its suppression of YvcK- or MreB- cells? At least two regulatory genes of sugar metabolic pathways, that encoding the phosphotransferase system enzyme 1 (Pts1) or the carbon catabolite control protein A (CcpA), have been implicated in suppression of the MreB- growth defect (Kawai et al., 2009). Moreover, B. subtilis cells lacking the ManA mannose phosphate isomerase, an enzyme converting fructose-6-phosphate to mannose-6-phosphate, exhibit pronounced cell wall and shape defects presumably because of an underlying imbalance in (modified) hexose sugars (Elbaz and Ben-Yehuda, 2010).
That a putative metabolic enzyme can regulate the morphology of the cell from a cytoskeletal structure is clearly unusual, but (in hindsight) not entirely surprising. The biogenesis of the cell's envelope, organelles and other macromolecules is precisely tuned with nutrient availability in the environment and the resulting intracellular metabolic fluxes. YvcK joins a growing list of proteins with enzymatic signatures that are seemingly capable of functioning as metabolic sensors/modulators in cell shape homeostasis, potentially sampling internal or external metabolic fluctuations and translating this information into an appropriate morphogenetic output. For example in the realm of cytoskeletal regulators, the bifunctional CTP synthase enzyme (CtpS) of C. crescentus assembles into a filamentous structure that aligns, interacts with and modulates the crescentin cytoskeleton (Ingerson-Mahar et al., 2010). In B. subtilis, the glycosyltranferase homologue UgtP inhibits the assembly of the FtsZ tubulin, presumably in response to changes in uridine-5′-diphospho-glucose levels (Weart et al., 2007). In the Gram-negative intracellular pathogen Brucella abortus, the FumC fumarase is localized to the old cell pole where it interacts with PdhS (Mignolet et al., 2010), an essential histidine kinase that controls the activity of the master cell cycle regulator CtrA (Hallez et al., 2007). In C. crescentus, the NAD(H)-binding oxidoreductase KidO influences FtsZ assembly and the activity of the master regulator CtrA at specific times in the cell cycle. Finally, in a classic example, the C-signal (CsgA) protein that triggers fruiting body formation in Myxcoccus xanthus, is an NAD+-binding alcohol reductase homologue (Baker, 1994; Lee and Shimkets, 1994). Thus, homologues of metabolic enzymes are implicated in regulating cell polarity, cell shape, single or multicellular development in bacteria. Perhaps their ability to bind metabolites/cofactors or to modulate the levels of metabolites predisposes non-essential metabolic enzymes to evolve regulatory or sensory roles, thereby ensuring that morphogenetic events can be effectively tuned to the cell's current metabolic state.