SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

The characteristic shape of bacterial cells is mainly determined by the cell wall, the synthesis of which is orchestrated by penicillin-binding proteins (PBPs). Rod-shaped bacteria have two distinct modes of cell wall synthesis, involved in cell elongation and cell division, which are believed to employ different sets of PBPs. A long-held question has been how these different modes of growth are co-ordinated in space and time. We have now identified the cell division protein, EzrA, and a newly discovered protein, GpsB, as key players in the elongation–division cycle of Bacillus subtilis. Mutations in these genes have a synthetic phenotype with defects in both cell division and cell elongation. They also have an unusual bulging phenotype apparently due to a failure in properly completing cell pole maturation. We show that these phenotypes are tightly associated with disturbed localization of the major transglycosylase/transpeptidase of the cell, PBP1. EzrA and GpsB have partially differentiated roles in the localization cycle of PBP1, with EzrA mainly promoting the recruitment of PBP1 to division sites, and GpsB facilitating its removal from the cell pole, after the completion of pole maturation.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

A fundamental question in cell biology is how cells determine their shape. In eukaryotic cells, shape is determined mainly by the cytoskeleton, a highly dynamic network of protein fibres that resides in the cytoplasm. Notably, prokaryotes also have cytoskeletal elements that control cell shape by orchestrating the synthesis of the bacterial cell wall (Wachi et al., 1987; Bi and Lutkenhaus, 1991; Jones et al., 2001; Ausmees et al., 2003; Daniel and Errington, 2003; Figge et al., 2004; Carballido-López, 2006). The cell wall forms a rigid, protective macromolecular structure composed predominantly of polymers of peptidoglycan (PG) that are interconnected via covalent cross-links thereby creating, in essence, a single gigantic molecule called the sacculus, which envelops the entire cell (Höltje, 1998). Importantly, growth and division of bacterial cells have to be tightly co-ordinated with synthesis of the sacculus to maintain its structural integrity and to prevent lysis. Therefore, the cell wall must be dynamic during the cell cycle with continuous synthesis, remodelling and turnover.

The discovery and development of penicillin several decades ago highlighted the bacterial cell wall and its metabolism as a crucial target for antimicrobial compounds. Penicillin and related β-lactam antibiotics inhibit the action of enzymes (penicillin-binding proteins; PBPs) that catalyse the formation of PG cross-links in the cell wall. Early research on β-lactams demonstrated the existence of two modes of cell wall synthesis in rod-shaped bacteria: one responsible for formation of the sidewall (elongation), and one responsible for formation of the cross-wall (cell division) (Lederberg, 1957; Schwarz et al., 1969). Indeed, the existence of septum-specific and elongation-specific PBPs was demonstrated in both Escherichia coli (Spratt, 1975) and Bacillus subtilis (Yanouri et al., 1993; Wei et al., 2003). Although several aspects of the division and elongation modes have been addressed, a decades-old question remains as to how the two distinct growth modes are co-ordinated in space and time.

Recent advances have established that rod-shaped bacteria use two large macromolecular complexes to direct synthesis of the cell wall during either cell elongation or cell division (Fig. 1). These complexes, which differ in composition, are positioned by two distinct cytoskeletal systems involving homologues of either actin (MreB) (van den Ent et al., 2001) or tubulin (FtsZ) (Löwe and Amos, 1998) respectively (Fig. 1). The MreB proteins form dynamic helical structures underneath the membrane in the cylindrical part of the cell of non-spherical bacteria (Jones et al., 2001; Carballido-López and Errington, 2003; Figge et al., 2004; Gitai et al., 2004; Slovak et al., 2005), and they are usually absent in bacteria with a coccoid morphology (Jones et al., 2001; Daniel and Errington, 2003). Furthermore, the introduction of mutations into anyone of the genes of the mreBCD operon causes cells to become spherical (Wachi et al., 1987; Levin et al., 1992; Jones et al., 2001; Lee and Stewart, 2003; Figge et al., 2004; Formstone and Errington, 2005; Kruse et al., 2005; Hu et al., 2007). These discoveries suggest that bacterial actins control cell wall synthesis in the cylindrical part of rod-shaped cells. This view is supported by the observation that antibiotics binding to nascent PG form an underlying helical pattern along the lateral wall that is reminiscent of that formed by MreB-like proteins (Daniel and Errington, 2003; Tiyanont et al., 2006). Further evidence for such a link is provided by MreC, which localizes helically and is tightly associated with the bacterial actin cytoskeleton, but has also been shown to interact directly with the enzymes catalysing synthesis of the cell wall, i.e. the PBPs (Divakaruni et al., 2005; Kruse et al., 2005; Leaver and Errington, 2005; van den Ent et al., 2006). Taken together, these results suggest that the bacterial actin cytoskeleton ultimately organizes cell wall synthesis in the cylindrical part of the cell by providing a scaffold for the recruitment and assembly of a macromolecular elongation complex to which it could also provide topological information from inside the cell (Scheffers and Pinho, 2005; Carballido-López, 2006; van den Ent et al., 2006).

image

Figure 1. Schematic diagram representing the classical view of the macromolecular complexes involved in cell wall synthesis during elongation (left) and division (right) of Bacillus subtilis. These complexes use different cytoskeletal proteins (MreB and FtsZ) that are not only involved in the recruitment and assembly of the complexes, but also provide topological information from the interior of the cell that is used by the penicillin-binding proteins (PBPs) synthesizing cell wall material on the outside of the cell.

Download figure to PowerPoint

The idea for the existence of a second macromolecular complex for the formation of the cell division septum (Fig. 1) originally came from depleting proteins of the division machinery and observing the formation of elongated or filamentous cells (as opposed to the bloated, spherical cells in the absence of proteins involved in cell elongation) (Ricard and Hirota, 1973; Beall and Lutkenhaus, 1991; Dai et al., 1993; Harry et al., 1993; Levin and Losick, 1994; Daniel et al., 2000). Cell division is initiated by the formation of a ring-like structure (the Z ring) by the FtsZ tubulin homologue at the future division site (Bi and Lutkenhaus, 1991). This ring acts as a scaffold for more than 10 other proteins that together govern and co-ordinate the constriction of the cell membrane and the synthesis of new cross-wall material to make up the septum (Errington et al., 2003; Vicente et al., 2006). In Gram-positive bacteria, formation of this septum is followed by a process of septum splitting and sister cell separation during which the mature, hemispherical cell poles are formed. Completed poles are inert in the sense that no further cell wall synthesis or turnover takes place (Cole and Hahn, 1962; Kirchner et al., 1988; de Pedro et al., 1997). Most of the proteins that were initially recruited to the division site are no longer present in the newly formed cell poles, but how this disassembly process occurs is poorly understood.

It is now clear that a previously unsuspected level of sophistication ensures that cell division occurs at the correct place and time in the cell cycle (Errington et al., 2003; Barak and Wilkinson, 2007). Several factors contribute to the formation and correct placement of the Z ring, one of which is EzrA. Previous work has shown that the integral membrane protein EzrA seems to fulfil two contradictory functions. In its absence, cells frequently form multiple Z rings, not only at mid-cell but also at polar sites, indicating that EzrA normally prevents their formation and thus acts as a negative regulator of FtsZ assembly (Levin et al., 1999). Paradoxically, deletion of the ezrA gene also causes a delay in cell division which results in longer cells (Levin et al., 1999; Chung et al., 2004). This suggests a positive role for EzrA during cell division. Although EzrA has always been thought of as a cell division protein, we noticed that cells lacking EzrA are not only longer, but also significantly thinner than wild-type cells. This came as a surprise, as known mutants with a changed cell diameter are normally affected in processes related to lateral wall synthesis or turnover, such as those in the mreBCD operon (Formstone and Errington, 2005; Leaver and Errington, 2005), mreBH[affecting the cell wall hydrolase LytE (Carballido-López et al., 2006)] or ponA, encoding PBP1 (Murray et al., 1998; Pedersen et al., 1999). The thinning of ezrA mutant cells therefore hinted that EzrA might also have a role in cell elongation. To explore this uncharacterized role, we performed a synthetic lethal screen with ezrA searching for defects related to cell wall synthesis. We identified a conserved protein that, together with EzrA, helps to co-ordinate the switch between cylindrical and septal cell wall synthesis by re-localization of PBP1, which we now show is a key component of both the elongation and division machines. The results also highlight the existence of ongoing PG synthesis during cell pole maturation, and the likely importance of removing PBP1 from completed cell poles.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Identification of gpsB using a synthetic lethal screen

To characterize the potential role of EzrA in the elongation system, we set up a genetic screen to identify synthetic lethal mutations. This screen was modelled after the classical yeast method (Bender and Pringle, 1991), and which was recently applied successfully in E. coli (Bernhardt and de Boer, 2004; 2005). Therefore, we created the unstable vector called pLOSS* (for Lethal Or Synthetic Sick; Fig. 2A). To screen for mutations that were synthetically lethal or sick in the absence of EzrA, strain 3362 (ezrA::tet) containing pLOSS*-ezrA+ was transformed with the mariner transposon plasmid pMarB and subsequently mutagenized (Le Breton et al., 2006), while maintaining selection for pLOSS*-ezrA+. The library of individual transposon mutants (n ≈ 30 000) was scraped from selective plates, and screened for mutants forming solid blue colonies on plates containing X-gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) only (pLOSS* carries a constitutively expressed lacZ gene). After the elimination of false-positives (see Experimental procedures), 49 mutants were obtained in which pLOSS*-ezrA+ was apparently stabilized. To confirm that the stabilization of the plasmid was due to the transposon insertion and not due to a second site mutation, or the integration of pLOSS*-ezrA into the genome, genomic DNA was isolated from these mutants and backcrossed into the parent strain (3362 containing pLOSS*-ezrA+). In five cases all the transformants obtained were blue, suggesting that maintenance of the unstable plasmid was essential for cell survival. PCR analysis showed that the transposon mapped to the ylm (sepF) region, deletion of which had previously been shown to generate a lethal cell division block in combination with ezrA (Hamoen et al., 2006). Sequencing showed that in these mutants the transposon had inserted either in the coding sequence of sepF (ylmF), or near the end of the upstream ylmE gene (data not shown).

image

Figure 2. A. Schematic representation of the synthetic lethal screening plasmid pLOSS*. The lacZ gene is constitutively expressed from the Bacillus subtilis divIVA promoter. pLOSS* has a large multiple cloning site (MCS) downstream of the IPTG-inducible Pspac promoter. Indicated sites in the MCS are unique. Resistance gene abbreviations as follows: spcR, spectinomycin; ampR, ampicillin. B. Backcrossing of the gpsB::TnYLB-1 allele into strain 3362 (ezrA::tet) carrying pLOSS*-ezrA+ results in the formation of large (blue) colonies (appear black in Fig. 2B) that maintain the unstable plasmid as well as small (white) colonies that have lost pLOSS*-ezrA+. The white colonies grow poorly and quickly acquire suppressor mutations (arrows). C. Alignment and structural predictions for GpsB proteins from several low-GC Gram-positive bacteria. Protein sequences similar to that of B. subtilis GpsB were identified by blast searches and aligned using clustalw. Each protein was tested for the presence of a coiled-coil region and the letters ‘a’ and ‘d’ were placed above the alignment at positions directed by the COILS output (Lupas et al., 1991). A helix-breaking domain that contains at least one proline residue (highlighted in bold face) follows the coiled-coil region. Conserved amino acids, present in all GpsB sequences, are dark shaded. The consensus sequence shows residues that are present in 80% of the sequences. D. Schematic representation of the likely domain structure of GpsB.

Download figure to PowerPoint

The other 44 backcrosses gave mixtures of blue and white colonies, suggesting that the combination of mutations caused cells to be synthetically sick rather than synthetically lethal. Indeed, in some cases the white colonies that no longer possessed the plasmid were substantially smaller than the blue colonies (Fig. 2B). In addition, they frequently seemed to acquire suppressor mutations allowing them to grow better (see arrows in Fig. 2B). Mapping and sequencing of five such mutants revealed that the transposon had inserted into a previously uncharacterized gene called ypsB, hereinafter called gpsB (for guiding PBP1 shuttling; see below).

GpsB is related to the N-terminus of DivIVA and conserved in a range of Gram-positive bacteria

The B. subtilis gpsB gene encodes a protein of 98 amino acids. Sequence similarity searches against the B. subtilis genome indicated that the N-terminus of GpsB is related to the coiled-coil region of the cell division protein DivIVA. During vegetative growth, DivIVA recruits the FtsZ-assembly inhibitor MinCD to division sites and cell poles (Marston et al., 1998), which, together with the process termed nucleoid occlusion, ensures that cells divide at mid-cell (Wu and Errington, 2004). Bioinformatic analysis showed that GpsB homologues are present in the genomes of a range of other Gram-positive organisms, including Streptococcus pneumoniae, Staphylococcus aureus and Listeria monocytogenes (Fig. 2C). Alignment of the amino acid sequences enabled three putative domains to be assigned to the GpsB proteins (Fig. 2D). The N-terminal domain (residues 1–65) aligns with DivIVA and contains five pseudo-heptad repeats that are predicted to form parallel double-stranded coiled-coils (Lupas et al., 1991). The coiled-coil is followed by a small helix-breaking domain (residues 66–75), which diverges from the DivIVA sequence and characteristically contains at least one proline. The third, C-terminal domain (residues 76–98) consists of a highly conserved run of 23 amino acids that is unique to GpsB.

GpsB is important for growth at high salt concentrations

To investigate the function of gpsB, a null mutant was constructed that had the majority of the gpsB coding sequence replaced with a kanamycin cassette. The resulting strain, 4221, was viable and grew similarly to the wild-type strain under normal growth conditions (see below). However, a previous microarray study had shown that gpsB (ypsB) was one of the genes whose expression is significantly induced under high-salt conditions (Steil et al., 2003). Therefore, wild-type and single-mutant cells were inoculated onto plates containing increasing amounts of NaCl. Strikingly, growth of the single-mutant strain was strongly reduced above a concentration of 0.6 M salt (Fig. 3A). Microscopic examination revealed that the mutant cells (Fig. 3C) were considerably longer than wild-type cells (Fig. 3B) and lysed frequently. In addition, mutant cell poles were often swollen (arrows). At even higher salt concentrations the wild-type strain also formed polar bulges on solid medium, and eventually lysed (data not shown; see also Murray et al., 1998). We conclude that GpsB has an important role in growth under salt stress, during which it promotes cell division and helps to reduce polar bulging and cell lysis.

image

Figure 3. GpsB is important for growth under high-salt conditions. A. Growth of the wild-type (GpsB+) and gpsB (GpsB-) mutant strain on PAB plates containing increasing amounts of NaCl. B and C. Phase-contrast images of typical fields of cells of wild type (B) and the gpsB mutant (C). Samples were taken from PAB plates containing 0.7 M NaCl.

Download figure to PowerPoint

Simultaneous deletion of gpsB and ezrA induces defects in both cell division and cell wall synthesis

The identification of gpsB as being synthetically sick with ezrA indicated that GpsB is important under normal conditions as well as during salt stress, although its function is normally masked by an activity provided by EzrA. Examination of the double mutant lacking EzrA and GpsB confirmed that the cells grew very poorly in liquid PAB medium with only a very slow increase in optical density (Fig. 4A, open triangles). Microscopic examination of the strain showed extensive lysis and cells that were seriously impaired in division, resulting in a broad range of cell lengths (Fig. 4B, Fig. S1). Moreover, cells of the double-deletion strain (4222) were significantly thinner than each of the single mutants (Table 1, Fig. S1) and about 10% of the cells displayed the formation of bulges at division sites or cell poles (Fig. 4B–D), similar to those observed for the gpsB single mutant grown under salt stress (Fig. 3C).

image

Figure 4. Characterization of the ezrA::tet gpsB::kan double mutant (strain 4222). A. Growth curves of strains 168 (wild type), 3362 (ezrA::tet), 4221 (gpsB::kan) and 4222 in PAB medium. 10 mM Mg2+ (10 mM) was added to one of the 4222 cultures (solid squares). B. Phenotype of the ezrA gpsB double mutant grown in PAB medium. Arrows highlight polar bulges. C. Time-lapse microscopy showing bulge formation (i) and subsequent lysis (ii) in a filamentous cell. D. Nascent PG synthesis in a bulging, filamentous cell of the double mutant (i), as visualized with fluorescently labelled vancomycin (Van-FL; ii). E. Van-FL-stained cells of the wild-type, single- and double-mutant strains grown in CH medium in the presence of 10 mM Mg2+. F. Van-FL-stained filamentous cells of the double-mutant strain grown in the presence of 10 mM Mg2+. Arrows in (Dii)–(F) highlight bulging cell poles. Scale bars represent 5 μm.

Download figure to PowerPoint

Table 1.  Average lengths and widths of exponentially growing wild-type and mutant cells in PAB supplemented with or without 10 mM Mg2+.
StrainWidth ± SD (μm)Length ± SD (µm)
  1. The number of cells counted for each strain is in parentheses.

PAB
 1680.91 ± 0.058 (237)4.56 ± 1.15 (327)
 3362 (ezrA::tet)0.82 ± 0.058 (249)8.51 ± 4.97 (305)
 4221 (gpsB::kan)0.91 ± 0.061 (283)4.70 ± 1.07 (249)
 4222 (ezrA::tet gpsB::kan)0.77 ± 0.110 (245)9.21 ± 6.20 (185)
 4223 (ponA:spc)0.65 ± 0.067 (242)7.04 ± 4.31 (197)
PAB + 10 mM Mg2+
 1680.88 ± 0.062 (310)3.95 ± 0.94 (292)
 3362 (ezrA::tet)0.82 ± 0.049 (271)5.98 ± 2.03 (333)
 4221 (gpsB::kan)0.85 ± 0.054 (264)4.20 ± 0.94 (264)
 4222 (ezrA::tet gpsB::kan)0.76 ± 0.058 (274)6.12 ± 2.82 (270)
 4223 (ponA:spc)0.71 ± 0.054 (254)3.85 ± 1.67 (274)

The extensive lysis and the bulging cell pole phenotype suggested that cell wall synthesis might be affected in the double-mutant strain. Previous reports have shown that the phenotypes of several mutants affected in cell wall formation can be rescued by the addition of mM concentrations of magnesium (Mg2+) to the growth medium (Rogers et al., 1976; Formstone and Errington, 2005; Leaver and Errington, 2005; Carballido-López et al., 2006), although the exact mechanism by which this works is unclear (perhaps Mg2+ cross-links anionic polymers in the cell, e.g. teichoic acid, thereby stiffening the wall or affecting autolysis activity). Similarly, polar bulging and the growth of wild-type cells under high-salt conditions were shown to be rescued by Mg2+ (Murray et al., 1998). In keeping with this idea, growth of the double-mutant strain was substantially improved in the presence of high levels of Mg2+ (Fig. 4A, compare filled squares with open triangles, and Fig. S2). The enhanced amount of Mg2+ promoted cell division and greatly reduced cell lysis and bulging, although the cells remained thinner than those of either single mutant (Table 1, Figs S1 and S2), and the overall growth rate was still not normal (Fig. 4A).

Delocalized cell wall synthesis in the absence of EzrA and GpsB

A characteristic feature of double-mutant cells (and salt-stressed gpsB cells) was the formation of bulges, which were mostly located at cell poles but also occurred at newly formed division sites (arrow in Fig. 4C). To establish if cell wall synthesis was active during bulge formation, cells of the double-mutant strain were grown at normal (low) magnesium levels and stained with fluorescently labelled vancomycin (Van-FL), which is thought to label sites where nascent PG synthesis occurs (Daniel and Errington, 2003). Strikingly, bright fluorescence was observed around the entire surface of the bulge (see arrow in Fig. 4D). In addition, staining was observed at division sites, albeit irregularly, and in the cylindrical part of the cell. To characterize this effect in more detail, we examined the topology of cell wall synthesis in single- and double-mutant cells grown in the presence of high Mg2+ (Fig. 4E and F). As reported previously (Daniel and Errington, 2003; Tiyanont et al., 2006), wild-type cells showed strong helical staining of the lateral cell wall. Ongoing or recent division sites also stained strongly but mature cell poles (outer edges of chains) were never stained, presumably because cell wall synthesis and turnover are shut down at these sites (Clarke-Sturman et al., 1989; de Pedro et al., 1997). Both single mutants showed a similar staining pattern (Fig. 4E). However, double-mutant cells were quite different with intense staining at both internal division sites and, importantly, old cell poles (arrows). This polar staining was even more evident in chaining cells, in which about 40% of the poles stained brightly (Fig. 4F). Moreover, the pattern of staining of division sites in such filaments appeared irregular. Some division sites stained brightly, whereas others hardly stained at all (Fig. 4F). Taken together, these results show that the localization of cell wall synthesis at division sites is perturbed in the absence of EzrA and GpsB, and in particular, that the double mutant is defective in completing the cell pole maturation process during which cell wall synthesis and turnover are shut down.

Septal PG synthesis is affected in the absence of EzrA and GpsB

The extensive lysis as well as the cell division defects in the double mutant prompted us to examine the synthesis of the septum in more detail. Electron micrographs of wild-type cells (Fig. S3A) confirmed many previous observations that synthesis of the complete septum of membrane and a thick plate of cell wall is usually completed well before hydrolysis of the wall material begins, which in turn gives rise to swelling of the new hemispherical cell poles. EzrA mutants had a similar septal morphology, though as expected, division was less frequent and the cells were longer (not shown). However, in the double mutant, cells were frequently observed in which wall hydrolysis and cell separation were apparently occurring before completion of synthesis of the cross-wall, as indicated by the deep indents in the cell wall at putative division sites (Fig. S3B). These results confirm the defects of the double-mutant strain in division and point to an important role for EzrA and/or GpsB in the recruitment or activation of one or more components of the septal synthesis machinery.

Complex interactions of GpsB with both division and cell elongation proteins

The complex phenotype of double-mutant cells, with defects related to both cell division and cell wall synthesis, prompted us to test for interactions with various division and wall morphogenesis proteins using a bacterial two-hybrid screen. For this, the coding sequence of gpsB was cloned into each of the four two-hybrid vectors (Karimova et al., 1998). The frequently described role of coiled-coils in establishing dimerization (Lupas et al., 1991), and the predicted presence of such a domain in GpsB (Fig. 2D), suggested that this protein might form multimeric structures. Indeed, reciprocal crosses with the constructs consistently showed a strong self-interaction of GpsB (Fig. S4A). Moreover, this also showed that all four constructs resulted in expressed and functional fusion proteins. Next, these constructs were used to screen for interactions with a range of proteins involved in cell wall synthesis or cell division. Surprisingly, reproducible strong interactions were found with only three other proteins: PBP1, a high-molecular-weight bi-functional PBP reported to catalyse both the transglycosylation and transpeptidation of the PG precursors during cell division (Murray et al., 1998; Pedersen et al., 1999), MreC, which is involved in growth of the cylindrical cell wall (elongation) and is tightly associated with the bacterial actin (MreB) cytoskeleton (Leaver and Errington, 2005), and EzrA (Fig. 5A). These interactions appeared GpsB-specific, as they were not observed when the homologous DivIVA protein was used as bait (only a DivIVA–DivIVA self-interaction was seen; Fig. S4B).

image

Figure 5. Bacterial two-hybrid analyses of GpsB. A. Interactions of GpsB with various proteins involved in cell division and elongation. The T18 and T25 fragments of the adenyl cyclase protein were fused to the N- or C-terminus of GpsB as indicated. Strong positive interactions are indicated with the red font. B. Schematic representation of the domain organization of PBP1. C. Bacterial two-hybrid interactions between truncated forms of PBP1 with GpsB, MreC and EzrA.

Download figure to PowerPoint

The specificity and strength of these interactions were surprising because most of the proteins associated with division and cell wall assembly tend to have multiple relatively weak interactions (Fig. S4B; Daniel et al., 2006; R. Emmins and R.A. Daniel, unpubl. results). Indeed, all of these constructs used are known to be at least partially functional because they show interactions in certain pair-wise combinations (Fig. S4B; van den Ent et al., 2006; and not shown). That GpsB interacted strongly with PBP1, but not with any of the other PBPs, was especially intriguing. PBP1 is a large protein of 914 amino acids consisting of a 37-amino-acid cytoplasmic N-terminus, a single, 23-amino-acid transmembrane region and an 854-amino-acid extracellular PBP domain (Fig. 5B). As GpsB is predicted to be a cytosolic protein, interaction with PBP1 was expected to occur inside the cell. Indeed, tests on various truncated forms of PBP1 showed that GpsB interacts strongly with a PBP1 truncation comprising only the cytoplasmic tail and the transmembrane region (Fig. 5C). GpsB did not interact with any of the other truncations, so GpsB presumably interacts with PBP1 at or close to the membrane surface.

Previous work has shown that PBP1, along with other PBPs, interacts with MreC (van den Ent et al., 2006). As a control experiment, the truncated forms of PBP1 were tested for interaction with full-length MreC (Fig. 5C). Strong interactions with MreC were observed with the PBP1 constructs that expressed the transmembrane domain on its own or in any combination with the other domains. Therefore, we conclude that PBP1 interacts with MreC via its transmembrane domain. Moreover, this confirmed the functionality of the constructs expressing the transmembrane domain-containing fusions of PBP1 that did not interact with GpsB.

The identification of PBP1 as an interaction partner for GpsB was interesting because some aspects of the ponA deletion phenotype (the ponA gene codes for PBP1) resemble that of the double mutant lacking the ezrA and gpsB genes. Cells of a ponA mutant strain are longer and thinner than wild-type cells and require increased amounts of Mg2+ to grow in liquid (Murray et al., 1998; Table 1, Figs S1 and S2). However, deletion of ponA does not cause polar bulging (Fig. S2). If the synthetic defects of the double-mutant strain were related to PBP1, one might expect EzrA also to interact with PBP1. To analyse this, EzrA was tested for interaction with the full-length and truncated forms of PBP1 (Fig. 5C). These experiments showed a reproducible strong interaction between EzrA and PBP1, for which the transmembrane domain of PBP1 was sufficient. To summarize, these results showed that GpsB and PBP1 interact strongly, and that both proteins also interact with both EzrA and MreC.

GpsB shuttles between the lateral wall and sites of cell division during the cell cycle

To test the physiological relevance of these putative interactions of GpsB with both elongation and division proteins, we analysed the localization of GpsB in living cells, using a C-terminal GFP fusion. Remarkably, two distinct patterns of GpsB–GFP fluorescence were observed in populations of cells: in some cells GpsB was present as discrete spots along the periphery of the cylindrical part of the cell which sometimes appeared to form a banded, potentially helical pattern; in other cells fluorescence occurred as a bright mid-cell band that appeared to constrict to form a dot at the junction between the two separating cells (Fig. 6A and B, arrows).

image

Figure 6. Shuttling of GpsB and PBP1 between division and elongation sites. A and B. Phase-contrast (A) and GFP images (B) of a typical field of exponentially growing cells expressing GpsB–GFP (strain 4224). Arrows point to cells at a late stage of cell division. C. Plot of cell length and frequency for cells with different types of GFP localization pattern (see insert for colour coding of different classes of pattern) for cells from the experiment illustrated in (A) and (B). D–F. Data from a similar experiment with a strain (2083) expressing GFP–PBP1. Scale bars represent 5 μm.

Download figure to PowerPoint

Quantification (Fig. 6C) and time-lapse microscopy (Movie S1) confirmed that localization of this protein changes dynamically during progression of the cell cycle. In young, newborn cells (2–3 μm) the GFP signal was associated with the new cell pole, from which it was rapidly redistributed into the lateral wall when cells elongated. When cells had reached a length of about 5 μm, the GFP signal gradually moved from the lateral wall towards the new division sites, until all detectable fluorescence was localized as a band at mid-cell, which in turn constricted as cell separation progressed. Just before the two daughter cells had separated completely, a bright fluorescent dot remained between them, at the site at which they seemed to be connected. Such cells have probably completed formation of the division septum and are undergoing maturation of the hemispherical cell poles. Finally, the polar fluorescence in the newborn cells disappeared, coinciding with the re-appearance of fluorescence in the lateral wall. These data suggest that GpsB shuttles between the lateral wall and the division site in a cell cycle-dependent manner. In addition, it infers that the likely interaction sites of GpsB with EzrA (division) and MreC (elongation) may be separated in time and space.

GFP–PBP1 localizes to sites of both cell division and elongation

PBP1 was previously shown to be mostly associated with cell division sites (Pedersen et al., 1999; Scheffers and Errington 2004). However, as noted above, in the absence of ponA cells are longer and thinner than wild-type cells. This contrasts with the effects of depletion of cell division-specific proteins such as PBP2B, for which cells form long filaments, but cell diameter is not detectably affected (Daniel et al., 2000; R.A. Daniel, unpubl. results). This suggested that PBP1 might have a role not only in cell division but also in elongation, which would be in agreement with the interactions detected between PBP1 and both EzrA (division) and MreC (elongation) respectively (Fig. 5C). To test this hypothesis, we re-analysed the localization of a previously described GFP fusion to PBP1 (Scheffers et al., 2004), to see if we could detect not only septal, but also lateral cell wall localization [with the difference that due to the improved sensitivity of the microscope, the amount of inducer (xylose) added to the cultures could be reduced, while still visualizing fluorescence, thereby preventing the masking of putative specific localization sites by spreading of the fusion protein]. Cells expressing the GFP–PBP1 fusion as the only copy of PBP1 in the cell were indistinguishable from those of a wild-type strain, indicating that the fusion is functional (Scheffers et al., 2004; this work). As expected, GFP–PBP1 fluorescence was the brightest in those cells that had started to divide, in which it was localized as a single band of fluorescence at mid-cell (Fig. 6D and E; arrows). However, in pre-divisional cells and in cells that had completed cell division, GFP–PBP1 localized to the lateral wall in a discrete pattern consisting of dots that sometimes resembled the tilted bands of helical proteins such as Mbl (Jones et al., 2001) as well as GpsB. These differences in the localization pattern of PBP1 between cells in the population, as for GpsB, seemed to correlate with the state of the cell cycle. Because of the relatively long exposure times (> 2 s) required to detect the fluorescence of the GFP–PBP1 fusion, time-lapse microscopy was not feasible. Instead, the localization pattern of GFP–PBP1 in the cell was related to the length of the cell (Fig. 6F). In young, newborn cells, GFP–PBP1 localized mostly to the new cell pole. When cells started to elongate, the fluorescent signal was rapidly removed from the cell pole and redistributed into the lateral wall. This helical-like localization pattern was observed until about the time of cell division, at which point increasing amounts of GFP–PBP1 formed a mid-cell band, and this coincided with the weakening or disappearance of the fluorescent signal in the lateral wall. When the onset of cell separation became visible (phase contrast), most of the GFP–PBP1 appeared to be at mid-cell. As cell separation neared completion, a bright dot of GFP–PBP1 was often observed at the connection between the two new cell poles of the separating daughter cells. Finally, after completion of cell separation, fluorescence was gone from those new cell poles, and redistributed into the lateral wall again.

The localization patterns of GpsB–GFP and GFP–PBP1 were strikingly similar throughout the cell cycle. To exclude that the helical-like localization of GpsB was due to (passive) interaction with PBP1, we analysed the fluorescence of GpsB–GFP in the absence of PBP1. No significant difference could be discerned (data not shown).

Localization of GFP–PBP1 is perturbed in the absence of EzrA and GpsB

The phenotype of cells lacking EzrA and GpsB implied defects in cell division as well as elongation, and, apart from the bulging cell poles, resembled a ponA deletion. Also, PBP1 and GpsB interact strongly with each other and with proteins involved in division and elongation. These results and the pattern of redistribution of both proteins throughout the cell cycle suggested a model in which EzrA and GpsB co-ordinate the distribution and movement of PBP1 between elongation and division sites. To test this hypothesis, the localization of GFP–PBP1 was studied in the absence of EzrA and GpsB. Notably, in the absence of GpsB (Fig. 7C; Fig. S5) the localization pattern of GFP–PBP1 was comparable to that of the wild-type strain (Fig. 7A). In contrast, when EzrA was absent (Fig. 7B), the clear separation between the localization patterns of elongating versus dividing cells was disrupted: cells failed to properly recruit GFP–PBP1 to division sites (or keep it there), as judged by a significant decrease in the number of cells in which GFP–PBP1 only localized at mid-cell; instead in most longer cells (> 8 μm) GFP–PBP1 localized mainly to the lateral wall (Fig. S5). Comparable results were obtained with the localization of GpsB–GFP in the absence of EzrA (data not shown). In the absence of both EzrA and GpsB, localization of GFP–PBP1 was further perturbed and appeared even more dispersed (Fig. 7D). Most cells failed to orchestrate the shift in localization of GFP–PBP1 between the elongation and division sites properly, with GFP–PBP1 being localized simultaneously at elongation and division sites, as well as at old cell poles (see particularly the red and purple bars in Fig. S5). Whereas some cells showed GFP–PBP1 simultaneously at elongation and division sites, regardless of the progression of cell division, others completely failed to recruit any GFP–PBP1 to division sites. The most characteristic difference observed, though, was the distinct fluorescence associated with old cell poles regardless of cell cycle progression (arrows in Fig. 7D), which was never seen for the other strains. Taken together, these results indicate that EzrA and GpsB are crucial, but partially redundant for the proper movement of PBP1 between the lateral wall and division sites, with a possible distinction of function in which EzrA is mostly important for recruiting PBP1 to the cell division site, and GpsB facilitating its removal after completion of cell division.

image

Figure 7. Localization of GFP–PBP1 in wild-type cells (A; strain 2083) or in the absence of EzrA (B; strain 4225), GpsB (C; strain 4226), or both (D; strain 4227). Arrows point to the unusual occurrence of GFP–PBP1 localization at cell poles. On the right side phase-contrast images are shown corresponding to the fluorescence images on the left. Scale bars represent 5 μm.

Download figure to PowerPoint

GpsB is sufficient to reposition PBP1 in the cell

The above results showed that GpsB is an important factor needed for proper re-localization of PBP1 during the cell cycle. If GpsB was making a direct and significant contribution to the movement of PBP1 (as indicated by the bacterial two-hybrid data), artificial re-localization of GpsB might generate a corresponding re-localization of PBP1. To test this, we fused GpsB to the DivIVA protein that targets to division sites and remains at mature cell poles in B. subtilis (Edwards and Errington, 1997). The construct expressing the fusion protein was introduced into strain 4226 carrying the xylose-inducible copy of GFP– PBP1. Strikingly, when expression of the fusion protein was induced, GFP–PBP1 not only localized to the lateral wall and cell division sites, but was also associated with (mature) cell poles (Fig. 8A). Expression of DivIVA alone had no such effect on the localization of GFP–PBP1 (Fig. 8B). Therefore GpsB alone is capable of directing the localization of GFP–PBP1 in support of the bacterial two-hybrid data.

image

Figure 8. GpsB is sufficient to reposition GFP–PBP1 in the cell. A. GFP–PBP1 localization in a strain expressing a DivIVA–GpsB fusion. B. Equivalent experiment for cells expressing DivIVA with no GpsB fusion. Arrows point to examples of cell poles with prominent fluorescence. Scale bar represents 5 μm.

Download figure to PowerPoint

PBP1 is required for polar bulge formation

The frequently observed localization of GFP–PBP1 and cell wall synthetic activity to cell poles in the absence of EzrA and GpsB, as well as the formation of polar bulges, suggested that PBP1 might be required for their formation. To test this, we deleted ponA in the double mutant lacking ezrA and gpsB. In the absence of all three proteins, no bulging at polar or division sites was detected, showing that PBP1 is indeed required for bulge formation (Fig. S2).

To see if the polar bulges formed under high-salt conditions were also dependent on PBP1, we grew the ponA mutant strain on solid medium containing increasing amounts of NaCl. The ponA mutant grew, albeit slowly, on PAB plates supplemented with up to 1.0 M NaCl. In the presence of high salt, no polar bulges were detected, showing that also under these conditions PBP1 is required for their formation (not shown).

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

The synthesis of the bacterial cell wall remains the principal target of clinically prescribed antibiotics. Therefore understanding its regulation and control not only provides insights into a fundamental process of development, but also allows the identification of new targets for novel antimicrobial compounds. Ever since the observation that different penicillin derivatives cause different morphologies upon cell death (Schwarz et al., 1969) bacteriologists have been fascinated by the question of how cell wall synthesis is organized. The identification of distinct PBPs involved in either cell division or elongation established that different complexes were responsible for the two modes of cell wall synthesis (Spratt, 1975). Despite ever more sophisticated research into the pattern of cell wall synthesis (Begg et al., 1990; de Pedro et al., 1997; Daniel and Errington, 2003; Bertsche et al., 2006) novel insights into the molecular basis for the switch between septal cell wall synthesis and lateral wall synthesis in rod-shaped bacteria have been elusive. The work described here has revealed that PBP1 of B. subtilis, which is involved in both elongation and division, dynamically relocates from the septum to the lateral wall during the cell cycle. Furthermore this switch requires the redundant activity of EzrA and a new cell division associated protein GpsB (Fig. 9). It has also revealed a new role for PG synthesis, involving PBP1, during pole maturation. Regulation of GpsB could therefore be a key control point in spatial organization of cell wall synthesis during the cell cycle.

image

Figure 9. Schematic representation of the shuttling of PBP1 protein between the elongation and division complexes, indicating the partially distinct roles of EzrA and GpsB.

Download figure to PowerPoint

Multiple roles for PBP1 in elongation, division and pole maturation

Previous localization studies have emphasized the prominent localization of PBP1 at division sites (Pedersen et al., 1999; Scheffers et al., 2004). However, our results show that in non-dividing cells the protein localizes to the cylindrical part of the cell, likely comprising a key component of the helical MreB-dependent cell wall elongation machine. The constellation of phenotypic effects of ponA mutants, including increased length, decreased width and sensitivity to Mg2+ concentration [in confirmation and extension of previous results (Murray et al., 1998; Pedersen et al., 1999)], shows that the two localization patterns probably reflect the dual role of this protein in elongation and division.

Detailed examination of the PBP1 localization pattern revealed a striking dynamic redistribution of the protein from division sites to the cell cylinder and back again, during cell cycle progression. An unexpected feature of this redistribution pattern was the invariable presence of a bright division site fluorescence in cells apparently very close to cell separation, contrasting with the striking absence of such a signal in cells that had completed separation. This suggests that PBP1 remains at the new cell poles until cell separation is complete but is then rapidly redistributed to the cylinder, and away from the pole. In Gram-positive bacteria cell division proceeds via two distinct steps. First, constriction results in the invagination of the cell membrane and wall to produce a septal plate that physiologically separates the cells. Dissolution of the wall material that holds the nascent sister cells is then required to allow them to separate, and the time taken for this step can be variable and rather long (Holmes et al., 1980). During cell separation the poles of the newly formed cells expand, maturing from their initially flat configuration to the inflated hemispherical state characteristic of mature poles. In principle, expansion of the pole could occur by limited cell wall hydrolysis (Koch and Burdett, 1986). However, the presence of PBP1 at the division site during cell separation suggests that PBP1 actively contributes to maturation of the pole. Failure to complete the maturation process correctly, as occurs in the absence of EzrA and GpsB, generates bulging cell poles, presumably resulting from continued or aberrant PG synthesis.

An elongation–division switch controlled by EzrA and GpsB

Our results suggest that EzrA and GpsB are key components of the system that regulates PBP1 localization during cell cycle progression. The effects on division, cell width and Mg2+ dependence of growth are common to both ezrA gpsB double mutants and ponA single mutants. Thus, impaired PBP1 localization or activation could be responsible for the major elements of the double mutant phenotype. Bulging is specific to the double mutant (or gpsB single mutants under high-salt conditions) but we propose that this is due in part to deranged PBP1 activity, as the effect was eliminated when ponA was deleted. In support of this model, PBP1 was found to interact strongly with both EzrA and GpsB. We therefore propose that EzrA and GpsB contribute crucially to the regulation of PBP1 action.

EzrA and GpsB are probably differentiated in the way they contribute to this task. Previous work has demonstrated a clear requirement for EzrA in cell division, via correctly regulating the timing and positioning of FtsZ rings (Levin et al., 1999), and in vitro experiments have established that EzrA can directly bind to and regulate the assembly of FtsZ polymers (Chung et al., 2007). It therefore seems likely that EzrA provides a key target for recruitment of PBP1 to the division site, through a direct protein–protein interaction (Fig. 9). In the course of this work, we noticed that ezrA mutants, unlike any of the other division mutants we have examined, are also thinner than wild-type cells and sensitive to Mg2+ levels, both indicative of an additional defect in lateral wall synthesis. As we have now shown that localization of PBP1 is perturbed in ezrA mutants it seems likely that the elongation defect is at least partly mediated by mislocalization of PBP1.

The phenotype of gpsB single mutants is quite different from that of ezrA. Under standard conditions the cells appear normal but under salt stress the cells are sick, and undergo bulging and lysis in a manner similar to that of the double mutant under non-stress conditions. Again, the bulging and lysis are dependent on PBP1 and rescuable by high Mg2+. The phenotypic effects of gpsB mutants, as well as the localization of GpsB in the cell pole throughout its maturation (but not in older cell poles), suggest that its main role could lie in removal of PBP1 from the cell pole after pole maturation is completed. This would presumably be followed by re-engagement of PBP1 with the elongation complex in the lateral cell wall, probably involving interactions with MreC (Fig. 9). However, the division defects of ezrA gpsB double-mutant cells, as well as salt-stressed gpsB single-mutant cells, suggest that like EzrA, GpsB can contribute to the recruitment of PBP1 to the division complex, perhaps by providing an additional bridge between PBP1 and EzrA (Fig. 9). It therefore appears that EzrA and GpsB have distinct, though partially redundant roles. Taken together, our results suggest that EzrA and GpsB play important roles in controlling the dynamic subcellular localization of PBP1 during cell cycle progression. This regulatory mechanism helps to ensure that the major bi-functional PBP is in the right place at the right time as the cell switches back and forth between elongation and division, thus maintaining proper cell shape.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Bacterial strains, plasmids and oligonucleotides

The strains and plasmids used in this study are listed in Table S1. Table S2 shows the oligonucleotide sequences.

Growth conditions and media

Bacillus strains were grown at 37°C on solid nutrient agar (Oxoid) plates or in liquid Difco antibiotic medium 3 (PAB) or casein hydrolysate (CH) medium, supplemented, where required, with salt, 1.5% agar or the appropriate antibiotics. For growth of strain 4222 (ezrA::tet gpsB::kan), 10 mM MgSO4 was added to the growth medium. Bacillus transformants were selected on nutrient agar (Oxoid) supplemented with 5 μg ml−1 chloramphenicol, 50 μg ml−1 spectinomycin, 10 μg ml−1 tetracyclin, 5 μg ml−1 kanamycin or 1 μg ml−1 erythromycin with 25 μg ml−1 lincomycin for selection of cat, spc, tet, kan and erm markers respectively. DNA manipulation and E. coli transformations were performed by standard methods (Sambrook et al., 1989).

Construction of pLOSS*

Primers pLS20 FW and pLS20 REV were used to amplify a 1125 bp fragment, flanked by SapI restriction sites, from the large (±55 kb) theta-replicating plasmid pLS20 (Koehler and Thorne, 1987). This fragment contains all the information required for autonomous replication in B. subtilis (Meijer et al., 1995). The PCR product was digested with SapI and ligated into the unique SapI site of pMutin4 (Vagner et al., 1998). The resulting vector, pLOSS, was able to replicate autonomously in B. subtilis.

To further decrease the stability of pLOSS, two point mutations were introduced in the DnaA box contained on the cloned pLS20 fragment by site-directed mutagenesis (primers GA[RIGHTWARDS ARROW]CC FW and GA[RIGHTWARDS ARROW]CC REV) thereby changing the consensus 5′-TGTGGATAA sequence into 5′-TGTGCCTA. Three further modifications were made to improve the vector: first, the B. subtilis divIVA promoter was cloned into pLOSS to constitutively express the lacZ gene. Second, the Erm resistance marker was replaced by the Spc resistance marker, and third, a multiple cloning site was introduced in between the unique NotI and BamHI sites of pLOSS, downstream of the Pspac promoter. All the primers used are listed in Table S2. The resulting end vector, pLOSS* (Fig. 2A), is available from the Bacillus Genetic Stock Centre (http://www.bgsc.org/).

The ezrA synthetic lethal screen

To screen for mutants that were synthetically lethal in combination with an ezrA chromosomal deletion, the promoter and coding sequence of the ezrA gene were amplified with primers SL-ezrA FW and SL-ezrA REV. These primers introduce NotI and BamHI restriction sites respectively (see Table S2). The PCR product was digested and inserted into NotI–BamHI-restricted pLOSS*. After introduction of pLOSS*-ezrA+ into strain 3362 (ezrA::tet), cells were transformed with the transposon plasmid pMarB at 30°C (Le Breton et al., 2006), while maintaining selection for the unstable plasmid. Several transformants were picked and individually grown for 8 h in LB medium supplemented with spectinomycin at 37°C. The cells from each culture were then plated and incubated overnight at 50°C on plates containing kanamycin and spectinomycin, or erythromycin and spectinomycin. The following day we selected the culture that gave the highest ratio of kanamycin-resistant colonies versus erythromycin-resistant colonies and the widest range of colony sizes and morphologies. It was assumed that these criteria were indicative of cells that had undergone successful and random transposition events. The selected culture was then used to generate a library of about 30 000 colonies, which were scraped from selective plates containing kanamycin and spectinomycin. The harvested cells were washed with LB medium and plated on nutrient agar plates containing X-gal only. Plates were incubated for 8–12 h at 50°C (to promote loss of the unstable plasmid), and then transferred to 37°C. Cells forming solid blue colonies were streaked on X-gal plates to eliminate false-positives. Mutants that remained uniformly blue, and those that formed mixtures of blue and white colonies in which the white colonies were retarded in growth, were screened for the correct antibiotic resistance pattern by patching them on plates containing erythromycin (pMarB), spectinomycin (pLOSS*) or kanamycin (TnYLB-1). Genomic DNA isolated from ErmS SpcR KanR mutants was isolated and backcrossed into strain 3362 containing pLOSS*-ezrA+ to confirm that plasmid stabilization was due to the transposon insertion and not due to a second-site mutation, or integration of the unstable plasmid into the genome. Mutants that fulfilled all of these criteria were subjected to inverse PCR amplification and sequencing of the transposon insertion site as described previously (Le Breton et al., 2006).

Construction of the GpsB–GFP fusion plasmid

A PCR fragment including the entire gpsB gene, from 353 bp upstream of its −35 consensus to 9 bp beyond its stop codon, was amplified with primers Y100 and Y101. Primer Y100 introduced an XhoI site into the sequence and primer Y101 a ClaI site that changed the stop codon to ATC. The PCR product was digested and inserted into XhoI–ClaI-restricted pSG1151 (Lewis and Marston, 1999). The resultant plasmid pYG1 encoded a GpsB–GFP fusion protein with a short linker (IDKLDIEFLQ) derived from the pSG1151 polylinker, which was integrated via a single-cross-over event at the chromosomal gpsB locus. The resulting strain 4224 contains intact copies of both gpsB and gpsB–gfp, with both genes under the control of the native σA promoter.

Insertional inactivation of gpsB

Primers Y401 and Y402 were designed to amplify a 2079 bp fragment of the chromosome that included the entire cotD ypsA gpsB and rnpB coding regions. The primers introduced Asp718 and BamHI restriction sites respectively (see Table S2). After digestion with the appropriate enzymes, the PCR product was cloned into the integration plasmid pSG1301 to produce pYPB1.

Plasmid pYPB1 was digested with HindIII to remove a 264 bp product that encodes the majority of GpsB from codons 7–95. The plasmid backbone was filled in with Klenow and gel purified. A kanamycin cassette was isolated from pBEST502 by Asp718–HindIII digestion, the 5′ overhangs were filled in with Klenow and the blunt ended cassette was gel purified. A blunt end ligation of the kanamycin cassette to the gel purified filled in HindIII-digested pYPB1 resulted in pYPB2.

In pYPB2 the kanamycin cassette disrupts gpsB after the seventh codon, and is flanked by two segments of B. subtilis DNA of approximately 1100 bp and 600 bp respectively. The plasmid can integrate into the chromosome by either single- or double-cross-over events. Kanamycin-resistant, chloramphenicol-sensitive colonies resulting from a double-cross-over event result in a chromosomal replacement of gpsB with the gpsB::kan disruption. Disruptions were verified by PCR analysis and sequencing.

Construction of the DivIVA–GpsB fusion

To construct a DivIVA–GpsB fusion protein, the coding sequences of the divIVA (without its stop codon) and gpsB genes were amplified by PCR, digested with the appropriate enzymes and ligated together into pDR111 that had been cut with HindIII and SphI. In the resulting construct the fusion protein could be expressed from the IPTG-inducible Pspank-hy promoter. Typically, 0.5 mM IPTG was used for induction. The control plasmid only contained the coding sequence of divIVA, including its stop codon.

Construction of p25-N

The bacterial two-hybrid vector p25-N was constructed in two steps. First, a fragment encoding the T25 fragment of the Bordetella pertussis adenylate cyclase protein was amplified from pKT25 (Karimova et al., 1998) using primers p25-N FW1 and p25-N REV1, generating a 747 bp product. These primers introduced a multiple cloning site, equivalent to that between the BamHI–EcoRI sites in pUT18 (Karimova et al., 1998), upstream of the T25 coding sequence, as well as an XhoI site downstream of it. The vector backbone was generated by PCR amplification of pKT25 with primers p25-N FW2 and p25-N REV2, resulting in a 2652 bp product. These primers incorporated a multiple cloning site downstream of Plac, equivalent to that between the PstI and BamHI sites in pKT25 (Karimova et al., 1998), and an XhoI site at the opposing end of the PCR product. Following digestion with XhoI and BamHI, both PCR products were ligated together, resulting in the 3369 bp p25-N plasmid, which enables C-terminal fusions of the T25 adenylate cyclase fragment to any gene of interest.

Bacterial two-hybrid analyses

To screen for interactions of GpsB with various proteins involved in cell wall synthesis or cell division, the gpsB coding sequence was amplified by PCR and cloned into each of the four bacterial two-hybrid vectors (pUT18, pUT18C, PKT25 and p25-N; Table S1), resulting in an N- (pUT18C) or a C-terminal (pUT18) fusion with the T18 domain of the adenylate cyclase enzyme, or an N- (pKT25) or a C-terminal (p25-N) fusion with the T25 domain (Karimova et al., 1998). In addition, the coding sequences of pbpB, ponA, rodA, divIVA, minC, minD, mreB, ftsW, ftsL, divIB, mreC, mreD, mbl, noc, ftsZ, ftsA, zapB, divIC, zapA, pbpA, pbpC, pbpD and pbpE were amplified by PCR and cloned into pKT25 and pUT18C to make N-terminal fusions, whereas the coding sequence of ezrA was cloned into p25-N and pUT18 resulting in C-terminal fusions. All genes were cloned into the two-hybrid vectors using XbaI and KpnI, except for pbpA, which was cloned as an XbaI–BamHI fragment.

To study the interactions between GpsB and truncated forms of PBP1, the coding sequences corresponding to amino acids 2–37 (N-terminus), 2–59 (N-terminus and transmembrane domain), 38–59 (transmembrane domain), 38–914 (transmembrane domain and extracellular PBP domain) and 60–914 (extracellular PBP domain) of PBP1 were cloned into pUT18C. All primer sequences that were used for the bacterial two-hybrid analyses are available upon request.

To assay for interactions using the bacterial two-hybrid system, co-transformations of each pair of plasmids were spotted onto minimal medium plates (Daniel et al., 2006) containing ampicillin (100 μg ml−1), kanamycin (25 μg ml−1), 0.1 mM IPTG and 0.004% X-gal. Pictures were taken after 40 h of growth at 30°C. Under these conditions, control transformations with empty vectors remained white for up to 72 h of incubation.

Microscopic imaging

For fluorescence microscopy, cells from an overnight liquid or solid culture were diluted into fresh CH medium containing 0.25% xylose and 10 mM MgSO4, and grown at 30°C to mid-exponential phase. Cells were then mounted directly onto poly lysine slides or on microscope slides covered with a thin film of 1.2% agarose in water, as described previously (Glaser et al., 1997). For time-lapse microscopy, cells were applied on microscope slides covered with a thin layer of 1.5% agarose made from CH medium and photographed every 15–20 min. Images were acquired with a Sony CoolSnap HQ cooled CCD camera (Roper Scientific) attached to a Zeiss Axiovert 200 M microscope. Images were analysed with ImageJ (http://rsb.info.nih.gov/ij/). All images in a given figure were taken with the same excitation and camera gain, on a single day using the same growth conditions. Image manipulation was limited to altering brightness and contrast to obtain optimal prints.

Fluorescent vancomycin staining

To label sites of nascent PG synthesis, cells were grown at 30°C in CH or PAB medium, and treated with fluorescently labelled vancomycin (BODIPY® FL Conjugate, Invitrogen) as described previously with the exception that cells were fixed in 3.6% formaldehyde (Daniel and Errington, 2003).

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank various current and former members of the Errington lab for stimulating discussions, as well as Susan Perry, Uli Schwarz and Jochen Höltje for their help and support. Furthermore, we are indebted to David Rudner for providing plasmids. D.C. was supported by EMBO and Marie Curie Long-term Fellowships. Work in the Errington lab was supported by a grant from the Biotechnology and Biological Sciences Research Council (BB/C506613/1), and in the Edwards lab by a Wellcome Trust project grant (064258/Z/01/Z).

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
MMI_6210_sm_Movie_S1.mov325KSupporting info item
MMI_6210_sm_Figures_S1-S5_and_Tables_S1-S2_and_Movie_S1.pdf4536KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.