Streptococcus pneumoniae is an oval-shaped Gram-positive coccus that lives in intimate association with its human host, both as a commensal and pathogen. The seriousness of pneumococcal infections and the spread of multi-drug resistant strains call for new lines of intervention. Bacterial cell division is an attractive target to develop antimicrobial drugs. This review discusses the recent advances in understanding S. pneumoniae growth and division, in comparison with the best studied rod-shaped models, Escherichia coli and Bacillus subtilis. To maintain their shape, these bacteria propagate by peripheral and septal peptidoglycan synthesis, involving proteins that assemble into distinct complexes called the elongasome and the divisome, respectively. Many of these proteins are conserved in S. pneumoniae, supporting the notion that the ovococcal shape is also achieved by rounds of elongation and division. Importantly, S. pneumoniae and close relatives with similar morphology differ in several aspects from the model rods. Overall, the data support a model in which a single large machinery, containing both the peripheral and septal peptidoglycan synthesis complexes, assembles at midcell and governs growth and division. The mechanisms generating the ovococcal or coccal shape in lactic-acid bacteria have likely evolved by gene reduction from a rod-shaped ancestor of the same group.
Research on bacterial cell division is not only driven by our curiosity to understand a fundamental biological process but also by the possibility to exploit cell division proteins as primary targets to develop novel broad-spectrum antibacterials (Vicente et al., 2006a; Vollmer, 2006; Lock and Harry, 2008). Over the last years, substantial progress has been made in elucidating several aspects of cell division in the rod-shaped model organisms Escherichia coli and Bacillus subtilis (for recent reviews de Boer, 2010; Lutkenhaus et al., 2012; Egan and Vollmer, 2013). Cell division proteins have been shown to be targeted to the division site, where they engage in a complex pattern of interactions (Di Lallo et al., 2001; 2003; Karimova et al., 2005) to assemble the functional division machinery called the divisome (Goehring and Beckwith, 2005; Vicente et al., 2006b). Significant advances have been made in the molecular and functional characterization of the main cell division protein, the tubulin-like FtsZ (Erickson et al., 2010), allowing the identification of compounds that inhibit its activity (reviewed in Lock and Harry, 2008; Adams and Errington, 2009; Foss et al., 2011). However, the lack of knowledge on the precise roles of other cell division proteins has so far limited their use in screens to identify cell division inhibitors. Moreover, it is important to characterize and validate the function of cell division proteins not only in model organisms but also in pathogenic species, for which the resistance to conventional antibacterials is an increasingly severe problem and new therapeutic agents are urgently required.
Streptococcus pneumoniae (the pneumococcus) is an important opportunistic Gram-positive pathogen that, according to the World Health Organization, annually causes more than a million deaths worldwide (http://www.who.int/ith/diseases/pneumococcal/en/index.html). S. pneumoniae finds its ecologic niche in colonizing the nasopharynx of healthy children and adults, without apparent negative effects, although carriage is a prerequisite for its spread among its human hosts (Shak et al., 2012; Simell et al., 2012). S. pneumoniae causes a variety of diseases that range from middle ear infections and sinusitis to pneumonia, bacteraemia and meningitis (Shak et al., 2012). Invasive pneumococcal diseases are particularly insidious: they are still associated with an overall high mortality and are prevalent among individuals with underdeveloped or compromised immune systems, such as infants, the elderly and HIV patients, and often occur after respiratory infections, such as influenza (Klugman, 2011). The severe toll of S. pneumoniae on human health is exacerbated by the spreading of strains resistant or tolerant to many of the commonly used antibiotics, calling for increased efforts to develop new vaccines and antimicrobials (Sham et al., 2012).
The interest in cell division of S. pneumoniae started mainly with the intention to identify and characterize targets to develop novel antimicrobials. Initially, the division and cell wall (dcw) (see Table 1 for glossary) gene cluster of S. pneumoniae was identified and shown to differ in its organization from the dcw clusters of other bacteria (Massidda et al., 1998). Genomic analysis of the dcw cluster of S. pneumoniae also led to the identification of an extended syntenic region downstream of the ftsA and ftsZ genes, present in both low and high GC% Gram-positives and cyanobacteria, but absent in Gram-negatives. This region contains yet uncharacterized genes as well as genes encoding homologues of YlmG, YlmF/SepF and DivIVA that are involved in chromosome segregation, cell morphology and/or cell division in various species (Massidda et al., 1998; Fadda et al., 2003; 2007; Flärdh, 2003; Ramos et al., 2003; Miyagishima et al., 2005; Ramirez-Arcos et al., 2005; Hamoen et al., 2006; Ishikawa et al., 2006; Kabeya et al., 2010). Further studies focused on the localization of the cell division protein FtsZ (Morlot et al., 2003) and on the localization and properties of FtsA, including its ability to polymerize in vitro in presence of ATP (Lara et al., 2005).
Table 1. Glossary for bacterial cell wall and cell division
Eukaryotic cytoskeletal protein with ATP-binding and ATPase activity that forms helical polymers
Peptidoglycan (PG) hydrolase (muramidase, glucosaminidase, endopeptidase or amidase) capable of lysing the cell by degrading the PG sacculus
Complex carbohydrate bound to the bacterial cell surface. Major virulence factor present in many pathogens, including pneumococci
Division and cell wall gene cluster; a conserved chromosomal region that contains many genes involved in cell wall biosynthesis and cell division
Division; derived from the defective phenotype of the rod-shaped B. subtilis cell division mutants, in genes other than fts or independently discovered from fts genes
The multi-protein assembly that synthesize the cross-wall septum and carry out cell division
The multi-protein assembly that synthesize the lateral wall during cell elongation
Filamenting temperature sensitive; derived from the filamentous phenotype of rod-shaped E. coli cell division mutants unable to divide at the non-permissive temperature
PG precursor; structure in pneumococci: undecaprenolpyrophosphoryl-(N-acetylglucosaminyl)N-acetylmuramic acid-L-Ala-D-iGln-L-Lys-D-Ala-D-Ala
Minicell; derived from the minicell phenotype, i.e. the formation of small, DNA-less spherical cells resulting from an incorrect division near the cell pole
Murein region ‘e’; E. coli chromosomal region containing genes encoding proteins responsible for the determination of the rod-shape
Enzymes for the synthesis of PG precursors
Synonym for PG, the major component of the bacteria cell wall required for osmotic stability of the cell and for maintaining cell shape
Bag-shaped macromolecule made of PG and surrounding the cytoplasmic membrane
Penicillin-binding protein; PG synthase (class A and class B PBP) or hydrolase (class C PBP); main target for β-lactam antibiotics
Single filament made of actin or tubulin molecules
Teichoic acid (TA)
Anionic cell wall polymers in Gram-positive bacteria, containing ribitol-phosphate or glycerol-phosphate subunits; pneumococci contain unusually complex TAs
Eukaryotic cytoskeletal protein with GTP-binding and GTPase activity that forms tubular polymers
The dynamic, ring-like structure made of FtsZ polymers and membrane-bound proteins that affect FtsZ polymer dynamics
Other relevant work in pneumococcus focused on the localization of the peptidoglycan (PG) synthases, the penicillin-binding proteins (PBPs) and the late cell division proteins FtsQ (DivIB), FtsB (DivIC), FtsL and FtsW. These studies provided insights into divisome assembly and supported the notion of two interdependently operating cell wall synthesis machineries responsible for peripheral growth and cell division, which is in line with the hypothesis of a two-site model for streptococcal growth and division (Lleo et al., 1990; Massidda et al., 1998; Morlot et al., 2003; 2004; Noirclerc-Savoye et al., 2005; Le Gouëllec et al., 2008; Zapun et al., 2008).
Further evidence for this model came from several divergent lines of research. The first concerns the characterization of the mreC and mreD genes that are required to maintain the ‘ovococcal’ cell shape, supporting their role in peripheral PG synthesis (Land and Winkler, 2011). The second line of evidence concerns the role of the S. pneumoniae Ser/Thr protein kinase StkP that coordinates cell growth and division and thus is also involved in cell shape maintenance during the cell cycle (Beilharz et al., 2012; Fleurie et al., 2012). Also, the two-site model in ovococci, including S. pneumoniae, was supported by super-resolution fluorescence microscopy of cells labelled at PG synthesis sites in combination with measurements of cell length, width and cross-wall diameter showing that newborn cells grow with little constriction prior to a longer phase in which cells both elongate and constrict (Wheeler et al., 2011).
Streptococci divide in parallel planes perpendicular to their long axis (Higgins and Shockman, 1970; 1976). Their cell separation, albeit closely linked with division, is not strictly synchronized, resulting in diplococci and/or short chain formation (Tomasz, 2000). In S. pneumoniae, the putative endo-beta-N-acetylglucosaminidase LytB acts at the very end of the division process and is responsible for chain dispersion (Garcia et al., 1999; De Las Rivas et al., 2002). Another putative PG hydrolase, PcsB, has been recently implicated with PG remodelling during cell division (Ng et al., 2003; 2004; Barendt et al., 2009; 2011; Giefing-Kröll et al., 2011) and has been shown to require FtsEX for proper function (Sham et al., 2011).
This review will take a critical look at some questions that emerged from all these studies, considering similarities and differences with what has been reported for the extensively studied rod-shaped cell division models E. coli and B. subtilis. We hope to stimulate fruitful discussions and novel lines of research, with the advantage of the new and remarkable tools that are now available for S. pneumoniae.
Growth and cell division in the model rods
To maintain their rod shape, E. coli and B. subtilis grow by alternating septal PG synthesis that occurs during cell division, with peripheral PG synthesis, that occurs during cell elongation. These processes are guaranteed by specialized proteins that assemble in two distinct separate machineries called, respectively, the divisome (Goehring and Beckwith, 2005; Vicente et al., 2006b; Lock and Harry, 2008; de Boer, 2010; Lutkenhaus et al., 2012; Egan and Vollmer, 2013) and the elongasome (den Blaauwen et al., 2008; Typas et al., 2012). In the two-competing sites model, these machineries compete with each other, so that no elongation occurs during septum formation and vice versa (Lleo et al., 1990), although the two processes are now known to be interdependent and some elongation also takes place during septation (van der Ploeg et al., 2013).
In E. coli, more than two dozen different proteins participate in the division process. Of these, FtsZ, FtsA, ZipA, FtsK, FtsE, FtsX, FtsQ, FtsB, FtsL, FtsW, PBP3 (FtsI) and FtsN are essential or conditionally essential for cell division and constitute the core of the division machinery. The assembly of these proteins into the divisome follows an almost linear hierarchical order, such that the recruitment of a particular protein or sub-complex depends on the prior localization of a specific set of earlier recruited proteins (de Boer, 2010; Lutkenhaus et al., 2012; Egan and Vollmer, 2013; Huang et al., 2013). This is only partially true for the Gram-positive counterpart B. subtilis (Errington et al., 2003). Nevertheless, both model organisms assemble the complete divisome in at least two temporally distinct steps (Aarsman et al., 2005; Gamba et al., 2009). The tubulin-like FtsZ protein is the first to localize to midcell to form the so called Z-ring well before the beginning of cell division (Bi and Lutkenhaus, 1991). FtsA and ZipA are recruited to the division site together with FtsZ or just after its localization, and are required to stabilize the Z-ring by tethering the FtsZ filaments to the membrane and to recruit the downstream division proteins (de Boer, 2010; Lutkenhaus et al., 2012; Egan and Vollmer, 2013; Huang et al., 2013). FtsA is an actin-like protein fairly conserved among eubacteria. However, differently from actin and other bacterial actin-like proteins which contain four structural subdomains (1A, 2B, 2A and 2B), FtsA is characterized by the absence of the subdomain 1B and the presence of an additional domain 1C on the opposite site of the molecule (van den Ent et al., 2001), which is important for its specific function. In E. coli, FtsA is essential and interacts with itself, FtsZ and other cell division proteins (Goehring and Beckwith, 2005; Vicente et al., 2006b). ZipA (from Z-interacting protein) is also essential to the division process (Hale and de Boer, 2002), but it is significantly less conserved than FtsA. In E. coli, the requirement for ZipA can be bypassed by the mutant FtsA(R286W) allele, named FtsA*, suggesting that the two proteins have a somewhat redundant function (Geissler et al., 2007). FtsA-polymerization is important for in vivo function (Pichoff and Lutkenhaus, 2005; Pichoff et al., 2012) and is supported in a recent study based on its crystal structure, showing that FtsA can polymerize in the presence of a lipid monolayer surface and forms actin-like protofilaments (Szwedziak et al., 2012). Interestingly, the analysis of E. coli ftsA mutants suggests that the ability of FtsA to recruit the late cell division proteins is inversely proportional to its propensity for self-interactions, which is modulated by ZipA (Pichoff et al., 2012). However, a specific role for ZipA in stabilizing the Z-ring by protecting FtsZ from ClpXP-directed degradation, supporting its essential role in cell division, has been recently reported (Pazos et al., 2013). FtsA is not essential In B. subtilis, although cells lacking ftsA form thermosensitive filaments with few Z-rings assembled (Beall and Lutkenhaus, 1992). ZipA is absent in Gram-positive bacteria, including B. subtilis, which possesses the negative Z-ring regulator EzrA (Levin et al., 1999) which is thought to share some features with ZipA (Erickson et al., 2010) and SepF (YlmF) that stabilizes the Z-ring and has a somewhat redundant function with FtsA (Ishikawa et al., 2006). SepF is recruited early to the division site in a FtsZ-dependent manner and shows a synthetic lethal division effect with EzrA (Hamoen et al., 2006). The Z-ring-associated proteins ZapA and ZapB assemble early at the division site and are necessary to stabilize the Z-ring, although they are not essential for the division process (Gueiros-Filho and Losick, 2002; Ebersbach et al., 2008; Mohammadi et al., 2009). ZapC and ZapD are two recently found non-essential components of the E. coli divisome that also stabilize the Z-ring (Hale et al., 2011; Durand-Heredia et al., 2012), but they are absent in B. subtilis. The Z-ring still requires some time before the late cell division proteins, recruited by FtsZ and FtsA, are subsequently assembled to form the mature divisome, although the signals for further maturation are still unknown. The ABC transporter encoded by ftsE and ftsX, whose inactivation in E. coli is conditionally lethal depending on the osmolarity of the growth medium, is targeted to the midcell via FtsZ-FtsE interactions (Schmidt et al., 2004; Corbin et al., 2007; Reddy, 2007). The membrane-bound FtsEX complex is required to recruit EnvC, an activator of the septal PG hydrolases, AmiA and AmiB, required for septum splitting (Uehara et al., 2010; Yang et al., 2011). In B. subtilis, FtsEX is not involved in cell division during vegetative growth, but is required for the proper localization of the septum near the pole during the initiation of sporulation (Garti-Levi et al., 2008). The divisome matures by incorporating the late cell division proteins. This stage includes the assembly of FtsK, a bifunctional protein that connects cell division to DNA segregation (Liu et al., 1998; Massey et al., 2006). The B. subtilis FtsK homologue SpoIIIE is required for proper distribution of chromosomal DNA only during the development of an endospore (Wu and Errington, 1994). In E. coli, FtsK localization is needed for septal recruitment of FtsQ (DivIC), FtsB (DivIC) and FtsL that are incorporated as a trimeric complex and are thought to interact with the PG biosynthetic machinery (Buddelmeijer and Beckwith, 2004). FtsW and PBP3 (FtsI) are required for septal PG synthesis while their orthologues, RodA and PBP2, respectively, are required for PG synthesis during cell elongation (den Blaauwen et al., 2008; Typas et al., 2012). FtsW is a conserved essential integral transmembrane protein that, together with RodA and the B. subtilis homologue SpoVE, belongs to the SEDS (shape, elongation, division and sporulation) family of integral membrane proteins (Ikeda et al., 1989). FtsW has been recently shown to function as the long-sought flippase for the transport of the PG precursor lipid II across the membrane (Mohammadi et al., 2011). In E. coli, FtsW has been shown to interact with PBP3 (FtsI), FtsQ, FtsL and FtsN in a bacterial two-hybrid assay (Di Lallo et al., 2003; Karimova et al., 2005). FtsW recruits the essential putative transpeptidase PBP3 (FtsI) that interacts in vitro and in vivo with the bifunctional glycosyltransferase-transpeptidase PBP1B for synthesizing the septal PG (Bertsche et al., 2006; Egan and Vollmer, 2013). The interaction between FtsW and PBP3 was confirmed in vivo and the FtsW–PBP3 complex was isolated in vitro (Fraipont et al., 2011). Finally, FtsN is one of the last E. coli essential cell division protein to reach the midcell (Addinall et al., 1997). Although its precise function remains unknown, it interacts with both PBP3 and PBP1B (Müller et al., 2007) and has been suggested to initiate constriction by signalling the completion of the late divisome assembly to its cytoplasmic components (Gerding et al., 2009; Rico et al., 2010). FtsN is absent from B. subtilis which, as expected, also lacks homologues of LpoA/B outer membrane lipoproteins required in E. coli to activate PG synthases and of members of the Tol-Pal complex that function to invaginate the outer membrane during constriction (Typas et al., 2012).
In the rod-shaped models, cell elongation precedes cell division and requires a distinct set of proteins, namely MreB, MreC, MreD, PBP2, PBP1A, RodA and RodZ, that assemble to form the elongasome (reviewed in den Blaauwen et al., 2008; Typas et al., 2012). MreB, together with the associated membrane proteins MreC and MreD, controls or localizes the PG synthesis complex responsible for cell elongation along the longitudinal axis of the cell (Jones et al., 2001; Kruse et al., 2005), while FtsZ and other early cell division proteins control complexes responsible for ‘preseptal’ cell elongation and for cell division (Varma et al., 2007). MreB is an actin-like protein (van den Ent et al., 2001) that assembles to form filamentous polymers or patches on the periphery of the cell membrane and orchestrates PG synthesis forming a complex with the membrane proteins MreC and MreD (Kruse et al., 2005; Leaver and Errington, 2005) required to localize the proteins involved in lateral elongation, RodA and PBP2. Consistent with a role in rod-shape maintenance, inactivation of the genes encoding MreB, MreC, MreD, RodA and PBP2 in E. coli resulted in non-dividing giant spherical cells prone to lysis, and their essentiality could be suppressed by overexpression of FtsZ, among other conditions (Vinella et al., 1993; De Pedro et al., 2001; Bendezú and de Boer, 2008). RodZ is the most recently identified component of the elongasome (Shiomi et al., 2008). It is a bitopic membrane protein that in E. coli interacts with MreB and is required for proper cell shape (Bendezú et al., 2009; Gerdes, 2009). Successive studies have clarified the MreB requirement of RodZ at a structural level by showing that it binds to monomeric MreB and this interaction is crucial for tethering MreB filaments to the membrane (van den Ent et al., 2010). In E. coli, RodZ is conditionally essential, and its shape and growth defects at non-permissive conditions can also be suppressed by elevated levels of FtsZ (Bendezú et al., 2009) or by mutations in the genes encoding MreB, RodA and PBP2 (Shiomi et al., 2013). B. subtilis requires a similar set of cell elongation proteins, but has two additional MreB homologues, MbI and MreBH (Carballido-López and Formstone, 2007) and a protein not present in E. coli, GpsB, that are also required for proper rod-shape (Claessen et al., 2008; Tavares et al., 2008).
Growth and division S. pneumoniae
Like the model rods, some of the more ovoid cocci, named ovococci (Zapun et al., 2008), were predicted to have two systems for PG synthesis, the peripheral and the septal, respectively, while the more spherical cocci have only the septal one (Lleo et al., 1990). Accordingly, inhibition of septum formation would cause cell elongation in those cocci with two systems, but not in those that have only one (Lleo et al., 1990).
The current model for growth and division of oval-shaped cocci is based on time-lapse ultrastructural reconstruction (Higgins and Shockman, 1970; 1976), visualization of the central growth zone by cell wall degradation in ethanolamine/choline pulse-labelled cells (Laitinen and Tomasz, 1990) and recently confirmed by super-resolution microscopy (Wheeler et al., 2011). This model is consistent with the characteristic shape of ovococci, like S. pneumoniae, that is achieved by both peripheral and septal PG synthesis, although with the latter prevailing over the former. Accordingly, cell division starts with an initial inward growth of the cross wall at the cell equator, marked by a microscopically visible ‘wall band’ or ‘equatorial ring’ (Higgins and Shockman, 1970). Soon after, on each side of the original ring, two new equatorial rings are formed and the initial centripetal growth remains constant while new PG is inserted in between the newly generated rings. The resulting peripheral growth, estimated about 300 nm for S. pneumoniae (Wheeler et al., 2011), progresses until the newly forming internal hemispheres have reached approximately the size of old ones, that is when the two rings approach the equators of future daughter cells, that will mark the next division sites. At this time, septal PG synthesis rapidly resumes and the complete septum is split by PG hydrolases for separation of the daughter cells. Thus, oval-shaped streptococci, in addition to peripheral and septal PG synthesis, require additional PG synthesis to produce their characteristic prolate poles (Higgins and Shockman, 1970; 1976).
In agreement with this model, the S. pneumoniae chromosome encodes proteins that are proven or likely to be part of the cell division (divisome) and/or the elongation (elongasome) machineries (Table 2). Additionally, other proteins, like StkP, PhpP and PBP3 (DacA), that have a regulatory role in pneumococcal growth and division are also listed in Table 2.
Table 2. List and features of established or putative S. pneumoniae proteins involved in cell growth and division
aGene locus in strains R6 (spr numbers) and TIGR4 (SP numbers).
bProteins with different lengths in R6 and TIGR4 are indicated by the respective numbers.
cSee also main text for function and references.
dThe symbol means self-interactions. Interactions conserved in S. pneumoniae and E. coli are shown in bold.
eReferences for protein essentiality, localization and interactions are only given for S. pneumoniae. For systematic studies on gene essentiality in S. pneumoniae, see Thanassi and colleagues (2002) and Song and colleagues (2005). See main text for further references.
fReferred to gene inactivation in Rx1 strain.
gPolar localization when purified GFP-LytB was added to cells.
CE, conditionally essential; E, essential; nd, not determined; NE, not essential; PG, peptidoglycan.
Tubulin structural homologue; GTPase; First protein in cell division; Z-ring formation at midcell
The relative ease of genetic manipulation of S. pneumoniae has been exploited in a number of studies on gene essentiality, including two global studies (Thanassi et al., 2002; Song et al., 2005). Several aspects complicate the interpretation of gene essentiality in S. pneumoniae, which may depend on the inactivation method (e.g. insertion/duplication, insertion/deletion or allelic replacement), the growth condition (e.g. media composition and incubation atmosphere) and the genetic background used. As shown in Table 2, a number of genes were found conditionally essential depending on these variables (Giefing et al., 2008; Le Gouëllec et al., 2008; Barendt et al., 2009; Osaki et al., 2009; Giefing-Kröll et al., 2011; Sham et al., 2011; Agarwal et al., 2012; Goldová et al., 2012), even in the case of closely related strains (Gasc et al., 1997). Moreover, for some gene deletions the observed phenotypic variability has been attributed to the presence or absence of a polysaccharide capsule (Barendt et al., 2009). Whether these variations are due to the very dynamic pneumococcal genome or to those more trivial factors mentioned previously, the clear message is that information concerning gene essentiality or mutant phenotypes in S. pneumoniae must be taken with caution.
Several studies have examined the role of S. pneumoniae proteins in cell growth and division by cellular localization (Table 2) which often provided a clearer result than gene inactivation studies. Proteins known or thought to be part of the division machinery are targeted to the septum and the equators in dividing S. pneumoniae cells, with the exception of DivIVA and the putative PG hydrolase PcsB, which are present simultaneously at the division septa and the cell poles. In addition, a GFP-fusion of the putative endo-beta-N-acetylglucosaminidase LytB, localizes preferentially at the poles when added to the cells. Unlike the model rods, however, the proteins also believed to be part of the elongation machinery, like PBP2b, MreC and MreD, localize to midcell, raising the question if in S. pneumoniae, proteins involved in the peripheral and septal PG synthesis are assembled into two distinct and separate complexes or into a single large one (Zapun et al., 2008; Sham et al., 2012).
S. pneumoniae contains the cell division proteins present in B. subtilis and other Gram-positives, and lacks some present in E. coli, like ZipA, ZapC, ZapD and FtsN (Table 2). Almost all of the proteins mentioned in Table 2 have been shown to localize to the septum in dividing S. pneumoniae cells. While a precise order of recruitment to midcell has not yet been established, the timing of localization based on fluorescence studies suggests that also in pneumococci, divisome formation occurs in at least two steps. Consistent with this, although no mechanism for targeting the division machinery to the nascent septum has been identified in S. pneumoniae, the essential cell division initiator proteins FtsZ and FtsA localize to midcell at the earliest stages of the process (Morlot et al., 2003; Lara et al., 2005) while the septal markers DivIB (FtsQ), DivIC (FtsB), FtsL, FtsW, PBP2x, PBP1a (Morlot et al., 2003; 2004; Noirclerc-Savoye et al., 2005), and the cell division protein DivIVA (Fadda et al., 2007; Beilharz et al., 2012), localize only after the FtsZ-ring has assembled. The Z-ring formation requires about half of the cell cycle before septation can occur (Fadda et al., 2007), in agreement with the observation that 46–50% of the cell cycle is devoted to cell elongation (Wheeler et al., 2011). As the equivalent E. coli and B. subtilis counterparts, pneumococcal FtsZ and FtsA interact with themselves and with each other (Lara et al., 2005; Maggi et al., 2008) and with other cell division proteins, including ZapA (Table 2). The ability of FtsA to form large polymers in the presence of ATP in vitro was first described for S. pneumoniae (Lara et al., 2005); however, it still awaits in vivo validation. Recently, however, two variants of the S. pneumoniae FtsA protein truncated in domain 1C or 2B (β-strand S12 and S13), previously shown to impair E. coli FtsA functionality in vivo (Rico et al., 2004), were found to inhibit the ATP-dependent polymerization of the pneumococcal wild-type protein and their ability to polymerize in vitro, confirming the importance of these regions for FtsA function (Krupka et al., 2012).
Both Z-ring positive (ZapA and ZapB) and negative (EzrA) regulatory proteins have been found by homology in S. pneumoniae, but they have not been characterized. Unlike in B. subtilis (Levin et al., 1999), EzrA seems to be essential in S. pneumoniae, whereas ZapA and ZapB could be inactivated in different genetic backgrounds (Thanassi et al., 2002; Song et al., 2005; Table 2). SepF (YlmF) was identified in S. pneumoniae as part of the extended dcw cluster and its inactivation resulted in severe division defects, including multi-septated, sausage-like cells with thinner septa at an early stage of constriction and occasionally elongated cells and minicells (Massidda et al., 1998; Fadda et al., 2003). Interestingly, SepF is part of the B. subtilis divisome where it plays a crucial role in Z-ring stability (Hamoen et al., 2006; Ishikawa et al., 2006), thus it will be worthwhile to verify this in S. pneumoniae. In contrast to E. coli, FtsE and FtsX are essential in S. pneumoniae, although they share a conserved function in regulating the recruitment and the activity of the PG hydrolase PcsB involved in cell wall remodelling (Sham et al., 2011; see next). Little is known about the properties of pneumcoccal FtsK, except that it is not essential for viability and that, as its E. coli counterpart (Di Lallo et al., 2003; Karimova et al., 2005), it interacts with itself, FtsZ, ZapA, FtsQ and FtsL (Maggi et al., 2008). Better characterized are FtsQ (DivIB), FtsB (DivIC) and FtsL that, similar to what has been reported for E. coli (Buddelmeijer and Beckwith, 2004), interact with each other and are incorporated as a trimeric complex into the S. pneumoniae divisome (Noirclerc-Savoye et al., 2005; Masson et al., 2009). In S. pneumoniae, FtsQ (DivIB) does not seem to be required for growth in rich medium, although the null mutant forms chains of diplococci and a small fraction of enlarged cells with defective septa (Le Gouëllec et al., 2008). In the absence of FtsQ (DivIB), the amount of FtsL drops dramatically, indicating that FtsQ, similarly to its proposed role in B. subtilis (Daniel et al., 2006), is crucial to stabilize the essential FtsL protein against proteolytic cleavage (Le Gouëllec et al., 2008). However, overexpressed FtsL only partly complements the ftsQ null mutant phenotype, suggesting that in pneumococci FtsQ may have an additional function unrelated to FtsL. Both FtsB (DivIC) and FtsL are essential in S. pneumoniae (Le Gouëllec et al., 2008). FtsW is also essential in S. pneumoniae according to both global gene inactivation studies (Thanassi et al., 2002; Song et al., 2005). It has 10 hypothetical membrane-spanning segments, a large extracellular loop, and both N- and C-termini located in the cytoplasm (Gérard et al., 2002). In agreement with the fact that it is a late cell division protein, FtsW is recruited to midcell after FtsZ where it remains longer and is occasionally observed at the cell poles (Morlot et al., 2004). Similarly to what has been shown in E. coli, S. pneumoniae FtsW interacts with FtsQ (DivIB) and FtsL, although an interaction between FtsW and PBP2x, the putative homologue of PBP3 (FtsI), or with any of the other PBPs, was not detected (Maggi et al., 2008). This is consistent with the suggestion that the PBPs involved in septum formation in S. pneumoniae localize to the septum by their substrate (Morlot et al., 2004) rather than by protein–protein interactions. S. pneumoniae has a set of six PBPs (Hakenbeck et al., 1999) including three bifunctional glycosyltransferase-transpeptidases (PBP1a, PBP1b and PBP2a), two monofunctional transpeptidases (PBP2b and PBP2x) and the DD-carboxypeptidase PBP3 (DacA) (Table 2; see next). Both PBP2b and PBP2x have been found essential in pneumococci (Kell et al., 1993), but a more recent global gene inactivation study, considering polar effect, found PBP2x as the only essential PBP (Song et al., 2005). The general assumption that PBP2x is involved in PG septal synthesis during cell division (Massidda et al., 1998; Morlot et al., 2003; Zapun et al., 2008; Sham et al., 2012) still requires experimental confirmation. The first localization data suggesting that PBP2x and PBP2b are part of different PG synthesis machineries at the septum and periphery, respectively (Morlot et al., 2003), were later proven to be incorrect (Zapun et al., 2008). In the revised model, both PBP2b and PBP2x localize to midcell during cell division (Zapun et al., 2008) where they interact with different cell division proteins (Maggi et al., 2008; Table 2). After cell division is completed, the daughter cells separate with the help of the PG hydrolase LytB (Garcia et al., 1999; De Las Rivas et al., 2002; see below).
DivIVA, whose name derives from its homologue in B. subtilis (Cha and Stewart, 1997; Edwards and Errington, 1997), is another protein crucial for proper pneumococcal cell division. It is highly conserved among Gram-positive bacteria but absent in Gram-negatives, with the possible exception of the δ-proteobacteria (Akiyama et al., 2003). This protein consists of a significantly conserved N-terminal domain and a less conserved C-terminal domain, varied in amino acids composition and length and rich in coiled-coil regions (Oliva et al., 2010). In agreement with this feature, DivIVA has been shown to interact with itself and with a number of other proteins in a species-specific manner (see below). Other relevant features of DivIVA are (i) its simultaneous localization at the cell septa and at the poles, (ii) its ability to accumulate at membrane regions with increased negative curvature (Lenarcic et al., 2009) and (iii) its ability to bind the cell membrane through its conserved N-terminal domain (Oliva et al., 2010). However, despite the common aspects shared by the DivIVA homologues, studies on the physiological role of this protein in different bacterial species have revealed distinct, albeit morphogenesis-related, functions.
In B. subtilis, DivIVA is part of the division-site selection system, functionally replacing the missing E. coli MinE (Cha and Stewart, 1997; Edwards and Errington, 1997). Consistently, during B. subtilis vegetative growth, DivIVA works together with the cell division inhibitors MinCD and MinJ to ensure correct FtsZ positioning at the midcell (Bramkamp et al., 2008; Patrick and Kearns, 2008). In addition, B. subtilis DivIVA has a second, quite distinct role during sporulation, where it is involved in chromosome segregation, attracting the chromosome origin and DNA-binding proteins RacA, Spo0J and Soj to the cell poles (Thomaides et al., 2001; Ben-Yehuda et al., 2003; Wu and Errington, 2003). Inactivation or depletion of divIVA in other Gram-positives that lack the MinCDJ homologues has revealed a variety of other phenotypes. In actinobacteria, divIVA has been shown to be essential and its function is related to polar growth, morphogenesis and, more recently, to chromosome segregation (Flärdh, 2003; Hempel et al., 2008; Kang et al., 2008; Letek et al., 2008; Ginda et al., 2013). In E. faecalis, divIVA depletion results in enlarged cells of altered shape with chromosome segregation defects (Ramirez-Arcos et al., 2005). In S. pneumoniae, its inactivation results in the formation of chains of unseparated, larger and rounder cells with incomplete septa and often devoid of nucleoids (Fadda et al., 2003). Further characterization of the pneumococcal DivIVA, based on its localization and protein–protein interactions, led to a model proposing a common MinCD-independent function that evolved differently in the various species depending on specific protein–protein interactions (Fadda et al., 2007). According to this model, S. pneumoniae DivIVA is targeted to midcell after FtsZ localization and assembles with the rest of the cell division proteins; once there, it ensures correct division through positioning the PG hydrolytic enzymes, likely PcsB and LytB, and possibly the late cell division proteins involved in the PG synthesis, required for septum formation and splitting and for maturation of the distinctive pointed streptococcal poles (Fadda et al., 2007; Vicente and García-Ovalle, 2007). Indeed, in ovococci, their prolate poles cannot be formed merely by stretching of the cell wall after septum splitting, but require insertion of additional PG (Higgins and Shockman, 1976). This function would be dispensable in those species, like B. subtilis, in which formation of the oblate poles requires only stretching of the cell wall after septum splitting (Koch and Burdett, 1986; Koch, 1992; Koch, 2002). Consistent with the function proposed for S. pneumoniae DivIVA, this protein is completely dispensable in S. aureus (Pinho and Errington, 2004) and only associated with cell separation defects in Listeria monocytogenes (Halbedel et al., 2012). In fact, despite the presence of MinCDJ in Listeria, DivIVA is not required for division-site selection but to recruit the PG hydrolases p60 and MurA to midcell, from where they are secreted in a SecA2-dependent manner (Halbedel et al., 2012). Interestingly, the Sec system has been shown to localize to the midcell at some point of pneumococcal cell division (Tsui et al., 2011), but no connection with export of PG hydrolases has been reported. Instead, PcsB and LytB, were shown to interact with DivIVA in a two-hybrid assay and this interaction has been thought to account, at least in part, for the divIVA null mutant phenotype (Fadda et al., 2007; Giefing-Kröll et al., 2011). However, PcsB has been recently shown to require the cell division proteins FtsEX for proper localization and activation (Sham et al., 2011; and below). Considering these data, a direct protein–protein interaction between DivIVA and PcsB appears unlikely, although a direct or indirect role of DivIVA in positioning PcsB cannot be completely excluded. Another important aspect of DivIVA is that this protein is phoshorylated by a conserved eukaryotic-type Ser/Thr protein kinase in several different Gram-positive bacteria. As phosphorylation signalling is related to regulation of cell morphology, growth and division, this aspect is considered below.
Like in the model rods, also in ovococci peripheral elongation precedes cell division. S. pneumoniae lacks the rod-shape determinant MreB, but has all the other components that are known or thought to constitute the elongasome, whose features are reported in Table 2. So far, only MreC and MreD have been characterized in S. pneumoniae (Land and Winkler, 2011). Interestingly, these proteins were found to be conditionally required depending on the specific genetic background. In particular, mreC and mreD are dispensable in the unencapsulated S. pneumoniae R6 strain, where their inactivation does not affect growth or morphology. In contrast, mreC and mreD are required in the encapsulated progenitor strain of R6, D39, and in its unencapsulated derivative D39Δcps (Land and Winkler, 2011). Consistently, D39 merodiploid strains carrying a second copy of the genes under the control of an inducible promoter rapidly stop growing and start lysing upon MreC, MreD or MreCD depletion. However, the requirement for mreCD in the D39 genetic background could be relieved by suppressor mutations, including some that eliminate the function of PBP1a. Inactivation of pbp1A resulted in thinner cells, particularly evident in the D39 unecapsulated derivative. Likewise, inactivation of the gene encoding PPB1a relieved the requirement for mreC, mreD or both, although cells of the resulting double mutants were misshaped and some had defective PG synthesis (Land and Winkler, 2011). Together, these results confirm that pneumococcal MreC and MreD are involved in peripheral elongation and support the previous hypothesis that, similarly to the rod-shaped models, two PG biosynthetic complexes also exist in S. pneumoniae (Massidda et al., 1998; Morlot et al., 2003; Zapun et al., 2008). Given the absence of MreB and the midcell localization of MreC and MreD, it remains to be clarified how this/these complexes are assembled (Land and Winkler, 2011).
Besides MreC, MreD and PBP1a, other likely components of the pneumococcal elongasome are GpsB, PBP2b, RodA and RodZ (Table 2). However these proteins are yet to be studied. S. pneumoniae GpsB (YpsB) was identified as the paralog of DivIVA, whose gene is located in the conserved chromosomal region that also contains pbp1A (Massidda et al., 1998) and is not essential according to both global gene inactivation studies (Thanassi et al., 2002; Song et al., 2005). GpsB (YpsB) is a late component of the B. subtilis divisome and is required for efficient cell division in thermosensitive ftsA null mutants (Tavares et al., 2008) and for the cell cycle-dependent localization of the B. subtilis PBP1 (Claessen et al., 2008), allowing regulation of elongation versus cell division. PBP2b is the monofunctional transpeptidase assumed to be involved in PG synthesis during peripheral growth (Massidda et al., 1998; Morlot et al., 2003; Zapun et al., 2008; Sham et al., 2012). While the essentiality of PBP2b remains to be confirmed in S. pneumoniae, the PBP2b homologues are not essential in the closely related ovococcal species S. sanguinis, S. thermophilus and L. lactis (Thibessard et al., 2002; Pérez-Núñez et al., 2011; Xu et al., 2011). Little is known about RodA that, like its septal counterpart FtsW, was found essential in both global gene inactivation studies in S. pneumoniae (Thanassi et al., 2002; Song et al., 2005), but not in S. thermophilus and S. sanguinis (Thibessard et al., 2002; Xu et al., 2011).
Little is known about the function of RodZ which, unlike the other elongasome proteins, does not show a coherent distribution with respect to cell shape (Margolin, 2009). Intriguingly, RodZ is present in all streptococci, including those species that lack MreB and even those, like S. pyogenes, that also lack MreC, MreD, PBP2b and RodA (Zapun et al., 2008). In this respect, it is interesting to underline that, although coccal-shaped streptococci would not be predicted to elongate if cell division is blocked, elongated cells were recently observed in S. pyogenes upon treatment with methicillin, suggesting that the septal PG synthesis machinery alone can be sufficient to determine the coccus-to-rod transition (Raz et al., 2012). In streptococci of both ovococcal and coccal shape, new cell wall is inserted only in a growth zone at midcell (the cell's equator), as determined by vancomycin labelling (Daniel and Errington, 2003). Therefore, it is possible that cell wall growth is facilitated by a single large machinery, directed by FtsZ and other early cell division proteins, that contains both peripheral and septal PG synthesis complexes, whose components may vary in different species. A representative model taking into account these considerations is shown in Fig. 1 for S. pneumoniae. Further support for this model comes from two independent studies in Lactococcus lactis and S. pneumoniae, where a block in septum formation by methicillin or deletion of the gene encoding the Ser/Thr kinase, respectively, resulted in elongated cells with multiple sites of PG insertion, at which both early and late cell division proteins were localized (Pérez-Núñez et al., 2011; Beilharz et al., 2012). To work properly, such a large midcell machinery would require a regulatory mechanism able to monitor the cell cycle progression and signal when it is time to stop elongating and start dividing. Finally, if the proposed model is true, it could be predicted that in both ovococcal- and coccal-shaped streptococci, an early block in cell division, such as the inhibition of FtsZ, would result in cell enlargement and eventually lysis (due to the simultaneous block in cell elongation and division), while a later block, as the one caused by methicillin, would result in the transition from coccus-to-rod (due to a block in cell division but not in elongation).
Assembly of wall teichoic acid and capsular polysaccharides
Apart from PG, the pneumococcal cell wall contains two other major polymers, an unusually complex, choline-containing wall teichoic acid (WTA) and the capsular polysaccharides (CPS). There are more than 90 chemically different versions of CPS, which define the specific serotype and play a crucial role in bacterial survival inside the host. The precursors for both WTA and CPS are synthesized in the cytoplasm and transported across the cytoplasmic membrane for insertion into the cell wall (Yother, 2011; Denapaite et al., 2012). Interestingly, both polymers appear to be inserted together with the new PG at the central growth zone, as elegantly shown for WTA using sequential ethanolamine/choline-labelling of cells followed by the specific removal of choline-containing cell wall segments with the pneumococcal autolysin LytA (Laitinen and Tomasz, 1990). Both WTA and CPS are covalently attached to PG by members of the LytR-Cps2A-Psr (LCP) family of phosphotransferases, which in B. subtilis associate with the MreB cytoskeleton (Kawai et al., 2011). The three pneumococcal LCP proteins localize to midcell and have redundant roles in the attachment of CPS and presumably WTA to PG (Eberhardt et al., 2012). Capsule polymer length is affected by a phosphoregulatory system composed of the autophosphorylating tryrosine kinase CpsD (also called Wze), the membrane protein CpsC (Wzd) and the phosphatase CpsB (Wzb) (reviewed in Yother, 2011). CpsC and CpsD interact with each other in the presence of ATP and localize as a complex to midcell (Henriques et al., 2011). Cells lacking CpsC or CpsD produce less capsule, which is still cell wall attached but depleted from division septa (Henriques et al., 2011). Together, these data indicate that secondary cell wall polymers like WTA and CPS are inserted into the cell wall at the site of new PG synthesis and coordinated with PG synthesis, although the precise mechanisms of cell wall assembly and its regulation remain to be deciphered.
Cell wall remodelling and separation
Bacterial growth and cell division requires the activities of PG hydrolases, presumably to allow for an increase in the cell wall surface area and to split the septal PG for daughter cell separation (Vollmer, 2012). The latter function has been demonstrated in many species, which grow in chains of unseparated cells in the absence of one or more septum-splitting PG hydrolases (Vollmer et al., 2008). An essential role of PG hydrolases in cell growth has been assumed for a long time but could be shown only recently: E. coli requires at least one of the DD-endopeptidases Spr, YebA and YdhO for cell elongation and incorporation of new PG into the sacculus (Singh et al., 2012), and B. subtilis requires one of two DL-endopeptidases, CwlO or LytE, for growth (Bisicchia et al., 2007). S. pneumoniae has 11 known or putative PG hydrolases. Single deletion of five of the corresponding genes leads to aberrant cell division and morphology (dacA, dacB, pmp23, pcsB) or severe cell chaining (lytB), indicating a role of the gene products in the placement, synthesis and/or cleavage of the division septum (Barendt et al., 2011).
Newly made PG contains abundant pentapeptides that in many bacteria are trimmed by carboxypeptidases to tetra- and/or tripeptides during PG maturation. In S. pneumoniae there are two carboxypeptidases that provide the main activities for the removal of pentapeptides: the class C penicillin-binding protein DD-carboxypeptidase PBP3 (encoded by dacA) cleaves the pentapeptides to generate tetrapeptides (Severin et al., 1992; Morlot et al., 2005; Nemmara et al., 2011), which are the substrate for the LD-carboxypeptidase DacB that forms the tripeptides (Barendt et al., 2011). PBP3 and DacB localize over the entire cell surface and at midcell in some dividing cells (Barendt et al., 2011), consistent with their capability of almost quantitatively removing the pentapeptides from the cell wall. The E. coli DD-carboxypeptidase PBP5 localizes to sites of active PG synthesis including the septation site, presumably by recognizing its pentapeptide substrates (Potluri et al., 2010). In S. pneumoniae, inactivation of PBP3 by clavulanic acid results in increased autolysis (Severin et al., 1997), and the truncation of dacA or the deletion of dacA, dacB or both genes causes similar morphological defects, including heterogeneity in cell size and shape and misplaced division septa (Schuster et al., 1990; Barendt et al., 2011). The reasons for these defects are not known, but may in part be due to improper localization of cell division proteins. In cells lacking PBP3, the PG synthase PBP2x and the lipid II flippase FtsW have altered patterns of co-localization with FtsZ during the cell cycle (Morlot et al., 2004). Whether the mis-localization of PBP2x and FtsW is caused by altered protein interactions within the divisome in the absence of PBP3 or by the abnormal, pentapeptide-rich cell wall is not known. The high proportion of penta- or tetrapeptides in the cell wall of the dacA or dacB mutant could potentially affect the activities of enzymes utilizing these peptides as acceptors in transpeptidation reactions, including PBPs that attach new PG to the cell wall and sortases that anchor surface proteins to the cell wall.
Pmp23 has been identified as putative PG hydrolase based on the presence of a PG carbohydrate cleavage enzyme (PECACE) domain (Pagliero et al., 2005). While Pmp23 does not have any other predicted enzymatic domain, the protein databases contain many sequences containing a PECACE domain linked to various PG hydrolase domains. The PECACE domain of Pmp23 shares conserved amino acid residues with the lytic transglycosylase Slt70 of E. coli and degraded Micrococcus PG in a zymogram assay (Pagliero et al., 2005), although its activity and specificity against pneumococcal PG remains to be established. Interestingly, pmp23 mutant strains have distorted division septa and grow with irregular, more rounded cell shape and larger size (Pagliero et al., 2005; Barendt et al., 2011), but the reasons for these morphological defects are unknown.
The predicted PG hydrolase PcsB has been identified in group B streptococcus in which the deletion of the pcsB gene causes reduced growth rate, cell clumping and misplaced division septa (Reinscheid et al., 2001). Similar phenotypes were observed in mutants lacking the pcsB homologues cdhA in group A streptococcus (GAS) (Pancholi et al., 2010), gbpB in Streptococcus mutans (Mattos-Graner et al., 2001) and sagA in Enterococcus faecalis (Breton et al., 2002). In S. pneumoniae, the pcsB gene is located adjacent to the mreC and mreD genes on the chromosome. PcsB appears to be essential in the capsule type 2 strain D39 and the non-encapsulated strain R6 (Barendt et al., 2009), but the gene could be deleted in four pneumococcal strains with other genetic backgrounds (Giefing-Kröll et al., 2011). The deletion or depletion of pcsB results in slow growth and strongly altered cell morphology with irregular septa and clumps of unseparated cells of different sizes (Barendt et al., 2009; Giefing-Kröll et al., 2011). The WalRK (also called VicRK) two-component system is required for the sufficient expression of pcsB (Ng et al., 2003; 2004; 2005). Consequently, the depletion of walRK results in similarly severe growth phenotypes as the depletion of pcsB. However, walRK can be deleted in a strain constitutively expressing pcsB, indicating that pcsB is the only essential member of the walRK regulon (Ng et al., 2003). In GAS, the expression of the PcsB homologue CdhA is regulated via phosphorylation of the response regulator WalR by the eukaryotic-type Ser/Thr kinase SP-STK, indicating convergence of WalK and SP-STK signalling cascades (Pancholi et al., 2010; Agarwal et al., 2011). Interestingly, the expression of pneumococcal pcsB is positively regulated by the SP-STK homologue StkP (Saskova et al., 2007), suggesting that a similar mode of regulation also exists in S. pneumoniae.
PcsB has a signal peptide followed by a predicted coiled-coil region with two putative leucine zipper motifs, and a C-terminal cysteine, histidine-dependent amidohydrolase/peptidase (CHAP) domain which is present in many PG endopeptidases, like the E. coli DD-endopeptidases Spr and YdhO, the B. subtilis DL-endopeptidases CwlO and LytE, and the large family of Tse/Tae DL- and DD-endopeptidases that function as effectors of type VI secretion systems in many Gram-negative bacteria (Russell et al., 2012). Purified PcsB lacked detectable PG hydrolyzing activity (Barendt et al., 2009; Giefing-Kröll et al., 2011), although two of the three predicted active-site residues, Cys292 and His343, were essential for PcsB function in the cell (Ng et al., 2004), suggesting that PcsB is an active enzyme in vivo. Interestingly, recent work from the Winkler laboratory suggested that PcsB might be activated by the membrane-bound ABC transporter FtsEX (Sham et al., 2011). PcsB interacts via its coiled-coil region with the extracellular ECL1 domain of FtsX, which recruits PcsB to the division septum. Depletion of FtsE or FtsX results in severe morphological defects similar to those seen in PcsB-depleted cells, indicating that FtsEX is required for PcsB function (Sham et al., 2011). It remains to be shown whether FtsEX activates the PG hydrolase activity of PcsB. Interestingly, E. coli appears to use a similar system for activating PG hydrolases during septation (Yang et al., 2011). In this species, FtsEX recruits EnvC to the division site. EnvC, although having a LytM peptidase domain, does not itself cleave PG but interacts with and activates two septum-splitting amidases, AmiA and AmiB (Uehara et al., 2010; Yang et al., 2011). The activation of AmiB involves a conformational change to remove an α-helix away from the active site cleft (Yang et al., 2012). Hence, it is possible that both, E. coli and S. pneumoniae couple the hydrolysis of ATP by FtsEX with the activation of PG hydrolases required for PG cleavage during septation (Sham et al., 2011; Yang et al., 2011). Other mechanisms may contribute to the regulation of PG hydrolase activity. For example, B. subtilis has a specific protein inhibitor (IseA) of DL-endopeptidases of the CHAP family (LytE, LytF, CwlS and CwlO) that has essential functions in growth and cell division (Arai et al., 2012).
The N-acetylglucosaminidase (glucosaminidase) LytB contains a signal peptide followed by 15–18 imperfect repeats forming a teichoic acid choline-binding domain and the C-terminal catalytic domain (Garcia et al., 1999; De Las Rivas et al., 2002). Purified LytB degraded pneumococcal PG and the purified GFP-LytB fusion protein localized to the poles of cells grown either in the presence of choline or ethanolamine, indicating that LytB-targeting to the cell pole does not depend on the nature of the amino alcohol present in the teichoic acids (De Las Rivas et al., 2002). Cells lacking LytB grow with normal morphology but in long chains with deeply constricted septa, suggesting a role of LytB in separating the daughter cells after cell division (Garcia et al., 1999; De Las Rivas et al., 2002). Consistent with this function, purified LytB is capable of dispersing the cell chains of a lytB mutant strain. This activity of LytB depends on the presence of choline residues in the teichoic acids, as LytB cannot disperse the chains of ethanolamine-grown lytB cells (De Las Rivas et al., 2002). It remains to be established how the cell restricts the activity of LytB to the division site and how LytB cleavage in PG leads to daughter cell separation.
Regulation of cell growth and cell division
Prokaryotes often use reversible protein phosphorylation, catalyzed by protein kinases and phosphatases, to transmit cell cycle signals and to respond to environmental changes. Two component systems, consisting of a histidine kinase with a cognate response regulator, are the most abundant type of bacterial signalling systems (Stock et al., 2000). However, eukaryotic-type Ser/Thr protein kinases (ESTKs) as well as Ser/Thr phosphatases (ESTPs) operate in numerous bacterial species and have been shown to regulate various cellular functions, including cell cycle and cell division, in many Gram-positive bacteria (reviewed in Pereira et al., 2011).
Lactic acid bacteria (LAB) encode a conserved eukaryotic-type signaling system consisting of a Ser/Thr protein kinase and a cognate, PP2C type, phosphatase. The protein kinases belong to conserved subfamily of ESTKs found in Gram-positive bacteria and consist of a cytoplasmic kinase domain, a transmembrane region and a C-terminal part outside the cell. The extracellular part is made up of three or four PASTA (for ‘penicillin-binding protein and serine/threonine kinase associated’) domains (Yeats et al., 2002) which were first identified in the high-molecular-weight class B PBP2x of S. pneumoniae. It was suggested that PASTA domains can bind PG fragments that might act as a signalling molecule (Yeats et al., 2002). This hypothesis was supported by the finding that PASTA domains of protein kinase PrkC from B. subtilis bind PG in vitro and activate spore germination in response to cell wall-derived muropeptides (Shah et al., 2008).
S. pneumoniae encodes a single Ser/Thr protein kinase, StkP, and a co-transcribed cognate phosphatase, PhpP (Echenique et al., 2004; Nováková et al., 2005). StkP acts as a dimer and forms a signalling pair with PhpP, a predicted cytoplasmic protein (Pallová et al., 2007; Osaki et al., 2009). Several StkP substrates playing a role in cell wall metabolism and cell division were identified, including the phosphoglucosamine mutase GlmM, the cell division proteins DivIVA, FtsZ and FtsA, and the PG precursor biosynthesis enzyme MurC (Nováková et al., 2005; Giefing et al., 2010; Nováková et al., 2010; Beilharz et al., 2012; Fleurie et al., 2012; Falk and Weisblum, 2013). Moreover, the putative endo-beta-N-acetylglucosaminidase LytB and the putative PG hydrolase PcsB are expressed at lower levels in an stkP mutant (Saskova et al., 2007). Interestingly, LytA and LytB were identified to be phosphorylated in vivo in a global study of the pneumococcal phosphoproteome, although the kinase(s) responsible for their modification were not identified (Sun et al., 2010).
The phenotypes of null mutant strains suggest that StkP could be involved in regulation of cell division (Giefing et al., 2008; Nováková et al., 2010), virulence, competence (Echenique et al., 2004) and stress resistance (Saskova et al., 2007). Consistent with a role in cell division, StkP and PhpP localize to the division site where active PG synthesis occurs (Beilharz et al., 2012). The PASTA domains of StkP bind synthetic and native PG subunits and β-lactam antibiotics in vitro (Maestro et al., 2011) and are required for septal localization of StkP, suggesting that they bind to newly made and still uncross-linked PG chains in vivo (Beilharz et al., 2012). Moreover, StkP and PhpP delocalize in the presence of antibiotics that target the latest stages of cell-wall biosynthesis and in cells that have stopped dividing. Cells of the unencapsulated strain Rx1 and the encapsulated strain D39 mutated for stkP or overexpressing PhpP are perturbed in cell wall synthesis and display elongated morphologies with multiple, often unconstricted division rings. On the contrary, S. pneumoniae cells overexpressing StkP or depleted for PhpP are significantly smaller and rounder (Beilharz et al., 2012; Goldová et al., 2012). These data indicate that StkP/PhpP play an important role in coordinating cell wall synthesis during growth and division, which in S. pneumoniae is required to achieve its characteristic oval shape. In this sense, StkP may act as a molecular switch that, through phosphorylation of key division substrates, signals the shift from peripheral to septal cell wall synthesis (Beilharz et al., 2012). A parallel study on the role of StkP in pneumococcal cell division partially confirmed these results and proposed diverse functions for the different StkP domains in the R6 derived strain R800 (Fleurie et al., 2012). In R800, the deletion of the stkP gene or the expression of a truncated protein lacking its kinase domain (StkP-PASTA-TMH) resulted in round and chaining cells, while an elongated morphology was observed in mutants expressing a truncated protein lacking the PASTA domains (StkP-KD-TMH) or the catalytically inactivated StkP(K42M) (Fleurie et al., 2012). The elongated phenotype was previously observed in the Rx1 mutant StkP(K42R), the stkP mutant lacking the PASTA domains (Nováková et al., 2010), and in the null stkP mutants of the TIGR 4, Rx1, R6 and D39 genetic backgrounds (Giefing et al., 2008; Beilharz et al., 2012). Indeed, the round and chaining phenotype, connected with the stkP deletion or truncation (Fleurie et al., 2012), was not observed in other studies (Giefing et al., 2008; Nováková et al., 2010) and may be due to the different genetic background, growth condition used or to suppressor mutations in stkP null strains, which show an elevated mutation rate (Osaki et al., 2009; Beilharz et al., 2012). To date, the molecular mechanisms of StkP's regulatory function on its substrates remain poorly understood. Fleurie and co-workers reported that the StkP-dependent phosphorylation of the major StkP substrate DivIVA at a single residue, Thr201, is crucial for its activity. Accordingly, R800 cells expressing the unphosphorylatable DivIVA(T201A) show an elongated shape with a polar bulge and display aberrant spatial organization of nascent PG synthesis (Fleurie et al., 2012). In contrast, in the Rx1 and D39 strains, the conditional expression of DivIVA(T201A) results in the complementation of the divIVA null phenotype and no obvious phenotypic defects (Nováková et al., 2012). Due to these conflicting results, the role and the importance of DivIVA phosphorylation by StkP in pneumococcal cell division remains yet unclarified.
Genes encoding homologues of the StkP/PhpP signalling pair are present in the genomes of all LAB, but they were characterized only in S. agalactiae (Rajagopal et al., 2003), S. pyogenes (Jin and Pancholi, 2006), S. mutans (Hussain et al., 2006) and E. faecalis (Kristich et al., 2007). Protein kinase Stk1 and its cognate phosphatase Stp1 regulate growth, virulence, stress adaptive response and cell separation of S. agalactiae (Rajagopal et al., 2003). Both stk1 and stp1 null mutants form extensive chains in comparison with the diplococcal wild type; however, the cell division phenotype of these mutants was not characterized further (Rajagopal et al., 2003). Analysis of the phosphoproteome derived from the corresponding mutant strains identified several substrates of Stk1 and Stp1, including the cell division proteins DivIVA (SAK_0586), FtsZ (SAK_0581), the GpsB homologue (SAK_0373) (Silvestroni et al., 2009) and the Z-ring negative regulator EzrA (SAK_0709) (Burnside et al., 2011), among other proteins of unknown functions unrelated to morphogenesis. The protein kinase STK in S. pyogenes is important for colony morphology, cell growth, cell division and virulence (Jin and Pancholi, 2006). Mutant cells lacking STK aggregate have an increased volume, a loosely attached fibrillar electron-dense outer layer and show incomplete septation (Jin and Pancholi, 2006). Moreover, strains with an inactivated gene for the cognate phosphatase STP have significant cell division defects containing multiple asymmetric and parallel division septa with thicker cell walls (Agarwal et al., 2011). The defects of STK-depleted cells can be partially explained by the reduced expression of the cdhA gene, which encodes the S. pneumoniae PcsB homologue (Pancholi et al., 2010; Agarwal et al., 2011) and of mur1 and mur2, encoding the LytA and LytB homologues, respectively (Bugrysheva et al., 2011). The S. mutans protein kinase PknB and its cognate phosphatase PppL regulate various processes including cell division, genetic competence, stress resistance, production of bacteriocins, biofilm formation and cariogenesis (Hussain et al., 2006; Banu et al., 2010). Cells of the pknB null mutant display an abnormal cell shape, larger volume and higher tendency to lyse. Cells with inactivated pppL have normal size but divide irregularly with division plane often placed perpendicular to the previous one. Whole-genome transcriptome analysis revealed that PknB regulates the expression of the mreC and mreD genes required for peripheral PG synthesis and the putative PG hydrolase gene SMU_984, which may account for the changes in cell shape observed in the pknB and pppL mutants (Banu et al., 2010). The protein kinase IreK (PrkC) and its cognate phosphatase IreP reciprocally regulate the intrinsic cephalosporin and bile detergent resistance in E. faecalis (Kristich et al., 2007). IreK-depleted cells show morphological defects and massive cell wall lesions, and they often lyse forming cell ghosts (Kristich et al., 2007).
Overall, in streptococci and enterococci, ESTKs and their cognate ESTPs are required for correct septum formation and placement and cell separation, most probably through phosphorylation of important cell division proteins. Indeed, FtsZ, DivIVA and several cell wall metabolism enzymes are phosphorylated by ESTKs in other bacteria including mycobacteria, corynebacteria and streptomyces (Pereira et al., 2011), suggesting that cell division regulation via reversible phosphorylation of proteins is widespread in Gram-positive bacteria. However, such regulatory mechanisms appear to be absent in the two most studied model organisms: in B. subtillis the StkP homologue PrkC has no obvious role in regulating cell division or cell shape, but instead controls sporulation and germination (Madec et al., 2002; Shah et al., 2008), and E. coli lacks homologues of ESTKs genes in its genome (Galperin et al., 2010). Nevertheless, in a global study of protein arginine phosphorylation, B. subtilis DivIVA was found to be phosphorylated at Arg-102 (Elsholz et al., 2012), suggesting that alternative types of phosphorylation signalling systems can cover the role of ESTKs.
Distribution of morphogenetic and cell division genes in S. pneumoniae and close relatives
The morphology of a bacterial cell depends on a complex interplay of many factors that drive the cell wall biosynthetic apparatus and is genetically determined by the presence or absence of shape determining genes. Indeed, many studies have considered the genetic content of these genes to evaluate phylogenetic relationships even among distantly related species. The wealth of complete bacterial genome sequences now allows a more detailed comparison of S. pneumoniae morphogenetic traits not only with the rod-shaped models but also with closely related species belonging to the same group of LAB. Besides having several similar phylogenetic traits, the LAB group contains species of rod (e.g. Lactobacillus spp.), ovococcal (e.g. Enterococcus spp., Lactococcus spp. and most of the Streptococcus spp.) and coccal (e.g. S. pyogenes) shape. Thus, this represents a good opportunity to follow dynamic evolutionary events in a restricted group of bacteria that have been somewhat overlooked in basic research.
The dcw cluster contains many essential genes involved in cell wall biosynthesis and cell division, and provides one of the best examples of a highly conserved region that survived the dynamic rearrangements during evolution often observed in eubacterial genomes (Ayala et al., 1994). The conservation of the dcw cluster is not just restricted to the sequence of the respective genes, but extends to their order. Moreover, the content and order of the genes in the dcw cluster correlates well with the morphology of the bacterial clades, with the elongated rods and spirochetes grouped together and separated from the cocci (Tamames et al., 2001; Mingorance et al., 2004). The correlation of the dcw conservation with cell shape has been mainly attributed to the needs of the cell to coordinate the expression of genes involved in cell wall synthesis and cell division, which differs in cells with or without an elongation phase. Co-expressing genes for PG precursor synthesis with cell division genes might ensure sufficient levels of precursors available for division (genomic channelling), adding a selective advantage to cells with two competing PG synthesizing systems, one for elongation and one for septation (Vicente et al., 2006b). On the other hand, if during evolution a species loses one or more components of the cell wall elongation machinery and, hence, its rod-shape, the coordinated expression of cell division and PG synthesis genes in the dcw cluster would no longer offer an advantage, resulting in its eventual dispersal. Hence, by comparing the dcw cluster and the degree of its dispersal in different species, we may not only learn about the coordination of cell wall synthesis and cell division, but we may obtain insights into bacterial evolution. Indeed, the currently available data indicate that cocci have likely evolved from a rod-shaped ancestor (Siefert and Fox, 1998; Gupta, 2000; Koch, 2003). Figure 1 shows a schematic representation of the dcw cluster of the LAB for which the complete genome is currently available, in comparison with the dcw clusters of E. coli and B. subtilis. As expected, the dcw cluster of LAB shows an organization more similar to that of the Gram-positive representative B. subtilis, marked by the absence of murE, murF, ftsW and ddl. These genes are located in other regions of their respective chromosomes, although murF and ddl are adjacent in Pediococcus, Enterococcus and most of the streptococci (Fig. 2). All LAB possess a region from ftsZ to ileS, containing from four to six genes, including the divIVA homologue, highly conserved in Gram-positive bacteria. Other differences in the dcw cluster, like the absence of mraZ and ylmE, seem to reflect more specific phylogenetic distinctions inside the LAB clade. Interestingly, compared with S. pneumoniae, the unconventional organization of the dcw cluster is only partly conserved in other streptococcal species, with the exception of the closely related S. pseudopneumoniae, S. mitis and S. oralis, which inhabit the same ecological niche and undergo heterologous genetic exchange, and the more distantly related L. lactis. Finally, the gene content and order of the dcw cluster in LAB is irrelevant with respect to cell shape. As shown in Fig. 2, based on the organization of the dcw cluster and irrespective of morphology, the LAB falls into two groups, with the Streptococcus/Lactococcus group having the most dispersed cluster. Curiously, the dcw cluster of some LAB contains the cell division gene encoding FtsK, which may suggest the existence of a common ancestor in which the dcw cluster was more extended and contained genes not only required for cell wall biosynthesis and division but also for other closely related processes like chromosome segregation.
The presence of the rod-shaped determinants was also evaluated in LAB. In E. coli, the MreB, MreC and MreD proteins are encoded by the mreBCD genes located in the mre operon (Wachi et al., 1987; 1989; Doi et al., 1988). In B. subtilis, the mreBCD genes are located in the divIVB operon, clustered together with the minCD genes (Levin et al., 1992; Varley and Stewart, 1992). In E. coli, the minCDE, whose products MinCDE are required for proper septum placement ensuring a correct positioning of the Z-ring at midcell, are located in a separate region than mreBCD (de Boer et al., 1988; 1989). As shown in Fig. 3, similarly to B. subtilis (Lee and Price, 1993), minE is absent in all the LAB, but its presence in the genome of some clostridia (Stragier, 2001) indicates that this gene may have been lost early in evolution. On the other hand, the genes encoding minC and minD are present only in some of the rod-shaped members of the LAB (Fig. 3), suggesting that, independently of their shape, some of the latter species must have evolved another mechanism to guarantee correct positioning of the Z-ring at the midcell. Finally, as expected from their morphogenetic character, the presence of the mreBCD genes shows a strong correlation with cell shape. As shown in Fig. 3, the LAB bacteria fall within four groups: (i) all Lactobacillus spp., Leuconostoc spp. and Oenococcus oeni have mreBCD and are rod shaped; (ii) the heterogeneous group of Pediococcus spp., Enterococcus spp., Lactococcus lactis and some of the streptococci have mreCD, but not mreB, and have an ovococcal shape; (iii) Streptococcus gordonii has only mreC and yet is ovococcal; and (iv) the remaining Streptococcus spp. lack all mre genes and have a coccal shape. The presence of mreB correlates with rod shape, with the special cases of Leuconostoc spp. and Oenococcus that can display both rod and ovoid morphologies (Dicks et al., 1995; Dicks and Holzapfel, 2009; Holzapfel et al., 2009). These data support the notion that cocci have evolved from a rod-shaped ancestor (Siefert and Fox, 1998; Gupta, 2000; Koch, 2003) and provide new insights on how this event may have happened inside a phylogenetically closely related group of bacteria. In the case of LAB, an early event in the transition from rod shape to ovococcal shape may have been the loss of mreB and, if present, its homologues mbl and mreBH that contribute to rod shape in B. subtilis (Carballido-López and Formstone, 2007). In fact, in addition to mreB, the LAB of group 1 possess at least mbI or mreBH that is completely absent from the genome of the LAB of groups 2, 3 and 4. The transition from ovococcal to coccal shape is accompanied by the loss of mreD and then mreC, and in most cases the concomitant loss of other genes, like pbp2b, whose product is thought to form a complex with MreCD (Zapun et al., 2008; Land and Winkler, 2011; Fig. 2). We can then speculate that besides having an immediate effect on cell morphology, the observed gene reduction could have relieved the need to maintain two competing PG synthesizing systems for elongation and division in favour of a single one, directed by FtsZ and perhaps by RodZ, which is present in all LAB, in agreement with the model shown in Fig. 1.
The initial studies on the morphogenesis of streptococcci were at their peak in the 1970s. This remarkable work culminated in a model for streptococcal cell growth and cell division that is still accepted. Later on, research on streptococci focused more on understanding the mechanisms of antibiotic resistance and virulence potential, as well as the possibility to prevent the infections caused by these important human pathogens. Nevertheless, the interest in cell division on streptococci was recently renewed by the possibility to exploit the cell division proteins as targets to develop novel broad-spectrum antibacterials, which are urgently required to compensate for the resistance to conventional antibiotics. In this review we have presented and discussed the major advances in understanding streptococcal cell shape, growth and division. These studies found that the major events in growth and division in S. pneumoniae are similar to what has been observed in the model rods, but they have also highlighted important aspects specific to pneumococci and their close relatives. Moreover, other processes closely connected to cell division, like chromosome segregation and nucleoid occlusion, still need to be elucidated in ovococci. In this light, it is clear that the molecular mechanisms that govern cell growth and division in oval-shaped cocci cannot be extrapolated solely from the information available from the rod-shaped models. The list of experimental tools available for S. pneumoniae, like the new plasmids for cellular localization of fluorescent proteins (Eberhardt et al., 2009; Henriques et al., 2013) and the in vitro systems for PG synthases and cell division proteins (Masson et al., 2009; Helassa et al., 2012), is continuously expanding, making this already excellent model even more appealing and useful for comparison to other bacteria. Besides providing new insights to the basic knowledge, these studies will validate cell division targets for the development of new antibacterial inhibitors active on septation.
OM was supported by the EU Commission through the SANITAS programme and by the RAS through the LR7/2007 program (Project CRP2-401). LN was supported by Czech Science Foundation (Project P302/12/0256 and P207/12/1568) and by Institutional Research Concept Grant RVO 61388971. WV was supported by the EU Commission through the DIVINOCELL programme. We thank Michael B. Whalen and Pavel Branny for critically reading the manuscript and making useful suggestions. We thank Miguel Vicente for lively and stimulating discussion on bacterial phylogeny. We thank Rocío Arranz and Cristina Patiño, Centro Nacional de Biotecnología (CSIC), Madrid, for the immunogold electron microscopy.