Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
The Escherichia coli actin homologue MreB is part of a helical cytoskeletal structure that winds around the cell between the two poles. It has been shown that MreB redistributes during the cell cycle to form circumferential ring structures that flank the cytokinetic FtsZ ring and appear to be associated with division and segregation of the helical cytoskeleton. We show here that the MreB cytoskeletal ring also contains the MreC, MreD, Pbp2 and RodA proteins. Assembly of MreB, MreC, MreD and Pbp2 into the ring structure required the FtsZ ring but no other known components of the cell division machinery, whereas assembly of RodA into the cytoskeletal ring required one or more additional septasomal components. Strikingly, MreB, MreC, MreD and RodA were each able to independently assemble into the cytoskeletal ring and coiled cytoskeletal structures in the absence of any of the other ring components. This excludes the possibility that one or more of these proteins acts as a scaffold for incorporation of the other proteins into these structures. In contrast, incorporation of Pbp2 required the presence of MreC, which may provide a docking site for Pbp2 entry.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Escherichia coli and other rod-shaped bacteria contain a helical, membrane-associated cytoskeletal structure that coils around the cell cylinder between the two cell poles (Jones et al., 2001; Shih et al., 2003). In the species that have been examined the helical cytoskeletal structure contains the actin homologue MreB, one of a group of morphogenetic proteins that are required to maintain the shape of rod-shaped cells. The MreB-associated helical cytoskeleton has been implicated in several cellular processes, including maintenance of cell shape, chromosome segregation and establishment of cell polarity (reviewed in Shih and Rothfield, 2006).
In addition to MreB, several other proteins required for maintenance of the cylindrical shape of the cell are also present as helical structures that extend along the length of the cell cylinder. These include MreC [shown in Bacillus subtilis (Leaver and Errington, 2005) and Caulobacter crescentus (Divakaruni et al., 2005; Dye et al., 2005)], MreD [shown in B. subtilis (Leaver and Errington, 2005)] and Pbp2 [shown in C. crescentus (Figge et al., 2004) and E. coli (den Blaauwen et al., 2003)]. MreC and MreD are bitopic and polytopic integral membrane proteins, respectively, whereas Pbp2 (also referred to as MrdA or PbpA) is a bitopic membrane protein. Pbp2 acts cooperatively with RodA (MrdB), a polytopic membrane protein, in catalysing cylindrical murein synthesis during the longitudinal growth of rod-shaped cells (Ishino et al., 1986). The localization pattern of RodA has not previously been established.
It has been suggested that one or more of the Mre proteins may act as a scaffold to establish the helical organization of murein-synthesizing proteins in rod-shaped cells (Kruse et al., 2005). However, the relationships of the proteins within the helical structures or whether some or all are actually located within the same structures is unclear. Similarly, it has not been clearly established whether the assembly of specific proteins into cytoskeletal structures requires the presence of one or more of the other proteins. In the case of C. crescentus at least, disruption of MreB helices by limited treatment with the MreB-depolymerizing drug A22 did not prevent appearance of MreC or Pbp2 cables (Divakaruni et al., 2005; Dye et al., 2005) although long-term treatment was reported to cause disappearance of the MreC coiled structures (Divakaruni et al., 2005).
Prior to septation, MreB is incorporated into annular structures (‘cytoskeletal rings’) that are located near midcell (Shih et al., 2003; Figge et al., 2004; Gitai et al., 2004; Leaver and Errington, 2005; Vats and Rothfield, 2007). The annular structures are present as single rings that extended circumferentially around the cell cylinder and as pairs of rings (doublets) that flank the cytokinetic FtsZ ring. Similar results were obtained by immunofluorescence of chromosomally encoded MreB and by fluorescence microscopy of Yfp-MreB, showing that the ring structures were not artefacts of MreB overexpression or due to the Yfp-moiety in Yfp-MreB (Vats and Rothfield, 2007). The MreB-associated rings have been suggested to be intermediates in the division and segregation of the helical MreB cytoskeleton during the cell cycle, thereby ensuring its equal partition into the two daughter cells (Vats and Rothfield, 2007).
In the present study, we show that the MreB-associated cytoskeletal rings are multiprotein structures that contain MreB, MreC, MreD, Pbp2 and RodA. Assembly of MreB, MreC, MreD and Pbp2 into the cytoskeletal rings was dependent on the presence of FtsZ ring, an essential component of the cell division machinery, but required no other known components of the division apparatus. On the other hand, entry of RodA into the cytoskeletal ring required one or more additional components of the division machinery.
The MreBCD and RodA proteins were capable of assembling into cytoskeletal rings and coiled structures independently of each other or Pbp2. This shows that none of these proteins provides a scaffold for entry of the other components and argues against a hierarchical mechanism of assembly of the cytoskeletal structures. In contrast, entry of Pbp2 into the rings and coiled elements did not occur in the absence of MreC, suggesting that Pbp2 may assemble by using MreC as a docking site.
Protein components of the E. coli cytoskeletal ring
Previous work has shown that MreB can organize into two types of cytoskeletal structures: coiled helical elements that extend along the long axis of the cell and transverse singlet and doublet ring structures (Shih et al., 2003; Vats and Rothfield, 2007). The singlet and doublet MreB ring structures were visualized as transverse bands by fluorescence microscopy of Yfp-MreB in wild-type strain MC1000/pLE7 (Fig. 1A and B). The Yfp-MreB singlet bands were clustered around midcell (Fig. 1E–G) and the doublet bands were predominantly clustered on either side of midcell (Fig. 1H). This distribution pattern is consistent with previous evidence that the single bands are formed adjacent to the Z-ring and that the doublet structures flank the FtsZ ring and septation site (Vats and Rothfield, 2007).
The rings were present predominantly in the longest cells of the population (Fig. 1G), suggesting that they are formed in the latter part of the division cycle. Similar results were obtained by analysis of immunofluorescence patterns from cells expressing chromosomal MreB under control of the normal mreB promoter elements, and by analysis of the distribution patterns of Yfp-Pbp2, RodA, MreC and MreD (shown for Yfp-Pbp2 in Fig. S1). The progressive increase in numbers of cells with MreB rings in the longest size classes was associated with a progressive decrease in cells that contained only the coiled helical structure (Fig. 1D). This is consistent with previous evidence that appearance of the cytoskeletal rings may reflect redistribution of molecules from the extended helical elements (Vats and Rothfield, 2007).
To determine whether other proteins were associated with the MreB rings, we studied the fluorescence distribution patterns of Yfp-labelled MreC, MreD, Pbp2 and RodA. These four proteins, together with MreB, are required to establish the cylindrical shape of rod-shaped cells.
In exponentially growing cultures, Yfp-MreC, Yfp-MreD, Yfp-Pbp2 and Yfp-RodA were present as transverse bands, usually located near midcell. The transverse structures were present both as single bands (singlets, Fig. 2A-iii, B-iii, C-iii and D-iii) and as paired bands (doublets, Fig. 2A-iv, B-iv, C-iv and D-iv). The ratio of singlets to doublets was ∼3:1 in 2320 cells examined and no cells contained more than two bands. The MreC, MreD, Pbp2 and RodA transverse bands were indistinguishable from the MreB ring structures that have been described previously (Vats and Rothfield, 2007).
Complementation studies showed that Yfp-Pbp2 and Yfp-RodA were functional based on their ability to restore rod shape to temperature-sensitive spherical Pbp2ts and RodAts cells. Although Yfp-MreC and Yfp-MreD were unable to complement the corresponding deletion strains PV1 [ΔmreC] and YLS6 [ΔmreD::kan], Yfp-MreC and Yfp-MreD colocalized with chromosomally encoded MreB in MreB immunofluorescence experiments (see below), implying that the Yfp moieties did not interfere with the localization characteristics of the proteins.
The Yfp-labelled MreC, MreD, Pbp2 and RodA proteins were also present in pole-to-pole double-stranded helical structures (Fig. 2A-i, ii, B-i, ii, C-i, ii and D-i, ii). The helical structures were present in ∼93% of the wild-type cells and were often also visible in cells containing the cytoskeletal bands (Fig. 2B-iv and C-iv). The intensity of the helices was usually diminished in these cells as compared with the helical structures in cells that did not contain transverse bands. This suggests that ring formation occurred by redistribution of molecules from the pre-existing helical structures or possibly by a shift in the topology of new cytoskeletal assembly.
These distribution patterns were similar to those previously described for the E. coli MreB protein, showing that all five rod shape-determining proteins of the mre and mrd operons assemble into morphologically similar helical and ring structures in actively growing E. coli cells.
To estimate the frequency of the ring structures, the number of MreB, MreC, MreD, Pbp2 and RodA bands was expressed as a fraction of the total number of cells that showed clear fluorescent coils and/or bands. This showed that approximately 10–25% of cells in exponentially growing cultures contained the cytoskeletal bands. The results were similar in fixed and unfixed cells and in cells in which MreB from the normal chromosomal mreB locus was localized by immunofluorescence. This compares with the approximately 5% incidence previously reported (Vats and Rothfield, 2007), where numbers of bands were expressed relative to all fluorescent cells, whether or not any clear internal structures could be detected.
Double-label fluorescence studies provided further evidence that the five proteins (MreB, MreC, MreD, Pbp2, RodA) are located within the same cytoskeletal structures. Immunofluorescence using anti-MreB primary antibody and Alexa-labelled secondary antibody was used to identify MreB while Yfp fluorescence was used to characterize the distribution of Yfp-labelled MreC, MreD, Pbp2 and RodA in the same cells.
The double-label studies showed that Yfp-labelled MreC, MreD, Pbp2 and RodA colocalized with MreB in the singlet (Fig. 3A, C, E and G) and doublet (Fig. 3B, D, F and H) ring structures. These results suggest that these five proteins are likely to be present within the same cytoskeletal ring structures in E. coli.
Assembly of the cytoskeletal rings
To determine whether the assembly of cell shape maintenance proteins into the cytoskeletal ring is a hierarchical process in which one or more of the components is required for the subsequent entry of the other proteins, we studied the localization of each Yfp-labelled protein in the absence of one or more of the other proteins.
The localization experiments were carried out in ΔmreB, ΔmreC and ΔmreD deletion strains, and in temperature-sensitive pbpAts and rodAts strains grown at 42°C. In all cases, the rod shape of the cells was lost and the cells grew as spheres (Figs 4–8). The spheroidal cells underwent division, but at a slower rate than wild-type organisms. The ability to divide may have been associated with an increase in cellular FtsZ levels, previously reported for ΔmreB cells (Kruse et al., 2005; Shih et al., 2005). We have observed that there was also a moderate (< 2-fold) increase in cellular FtsZ concentration in the ΔmreC and ΔmreD cells used in the present study (data not shown).
The Yfp-Pbp2ts and Yfp-RodAts mutant proteins showed characteristic helical and ring distribution patterns at 30°C that were similar to those of the wild-type proteins. However, when the cells were grown at 42°C, the labelled Pbp2ts and RodAts proteins were diffusely distributed within the spherical cell, with no evidence of coiled or ring structures (data not shown). This confirmed that the temperature-sensitive defects affected the assembly characteristics of the proteins.
Fluorescence localization studies showed that Yfp-MreB was organized as bright singlet and doublet rings in ΔmreC (Fig. 4A-ii, iii), ΔmreD (Fig. 4B-ii, iii), pbpAts (Fig. 4C-ii, iii) and rodAts (Fig. 4D-ii, iii) cells. The ratio of singlets to doublets in the spherical cells was 8:1 in 780 cells examined. The bands were present in ∼7% of the cells and were generally located near the equator of the cell. In cells that were undergoing cytokinesis, one MreB band of the doublet was present in each half of the dividing cell (Fig. 4A-iii, C-iii and D-iii). Based on previous work (Vats and Rothfield, 2007), these bands presumably arose from doublet bands that were present in the parent cell prior to the division event.
In addition to the ring structures, the spherical cells contained MreB coils that extended around the spheres (Fig. 4A-i, B-i, C-i, D-I and E-ii). The coiled elements appeared as intertwined filamentous structures that wound around the spherical cell, frequently crossing each other one or more times. Further work will be needed to determine whether the coiled filaments in the spherical cells show any ordered long-range organization. The Yfp-MreB rings (Fig. 4E-i) and coils (Fig. 4E-ii) were also present in ΔmreBCD cells, excluding the possibility that MreC and MreD might be able to substitute for each other in providing a scaffold needed for MreB assembly into the organized structures. This finding is in contrast to reports that the localization of MreB required MreC and MreD in B. subtilis (Leaver and Errington, 2005) and in E. coli (Kruse et al., 2005). The reason for this discrepancy remains to be determined.
Similar results were obtained in studies of Yfp-labelled MreC (Fig. 5), MreD (Fig. 6) and RodA (Fig. 8). The proteins were organized in coiled as well as singlet and doublet ring structures in ΔmreB (Figs 5A, 6A and 8A), ΔmreC (Figs 6B and 8B) and ΔmreD (Figs 5B and 8C) cells, and in temperature-sensitive pbpAts (Figs 5C, 6C and 8D) and rodAts (Figs 5D and 6D) cells grown at 42°C. In exponentially growing cultures, coils and rings were present in ∼15% and ∼5% of the spherical cells respectively. The ratio of singlets to doublets for each of the Yfp-labelled proteins was approximately 8:1 in 1030 cells. The Yfp-MreC and Yfp-MreD rings and coils were also present in ΔmreBCD cells (Fig. 4F and G).
Double-label fluorescence studies confirmed that MreB and Yfp-MreC colocalized within the same ring structure in spherical ΔmreD cells (Fig. 3I), as had been suggested by the similarity of their localization patterns in the single-label experiments. MreB and Yfp-Pbp2 also colocalized in similar experiments (data not shown).
We conclude that the assembly of MreB, MreC, MreD and RodA into the cytoskeletal rings and coiled structures occurs independently of any of the other known components of MreB-associated annular or helical elements of the E. coli cytoskeleton.
In contrast, assembly of Pbp2 into the cytoskeletal structures required the presence of MreC. Yfp-Pbp2 was organized as rings and coiled structures in ΔmreB, ΔmreD and RodA-inactivated cells (Fig. 7A, C and D), but was not present in rings or coiled structures in ΔmreC cells (Fig. 7B). Instead of forming rings and coiled structures, Yfp-Pbp2 was irregularly distributed in ΔmreC cells, sometimes as irregular clumps in the periphery of the cell. Quantitative immunoblot analysis confirmed that the cellular concentration of Pbp2 was not diminished in the ΔmreC cells (Fig. S2).
These results show that Pbp2 entry into the cytoskeletal rings and coiled structures is dependent on MreC but is relatively independent of MreB, MreD and RodA. In cells that contained MreC (i.e. in ΔmreB, ΔmreD and rodAts cells), the Yfp-Pbp2 bands and coils often appeared somewhat clumpy and less sharply defined (Fig. 7A, C and D) than the structures formed by the other proteins in the same mutant backgrounds (Figs 5, 6 and 8). This suggests that, unlike the other proteins, the normal integration of Pbp2 can be perturbed by changes in the overall organization of the cytoskeletal structures.
Relation of the cytoskeletal ring to the division apparatus
Bacterial cell division is preceded by assembly of a multiprotein complex in which the essential cell division protein FtsZ first assembles into a ring structure at the future division site. In E. coli this is followed by the sequential entry of at least 11 proteins (Fig. 9A) to form the mature septasome (Aarsman et al., 2005; and reviewed in Buddelmeijer and Beckwith, 2002). The FtsZ ring is thought to act as a scaffold for the subsequent entry of the other septasomal components (reviewed in Shih and Rothfield, 2006). The membrane-anchoring proteins FtsA or ZipA are each individually capable of supporting Z-ring formation whereas the subsequent entry of the other septasomal proteins requires the presence of both FtsA and ZipA (Fig. 9A) (Pichoff and Lutkenhaus, 2002). We previously showed that the assembly of MreB into the cytoskeletal rings requires the presence of the Z-ring (Vats and Rothfield, 2007).
In the present study, evidence that assembly of MreC, MreD, Pbp2 and RodA into the cytoskeletal ring required one or more components of the septasomal division apparatus first came from studies of ftsA12ts zipA1ts double mutant cells. When FtsA and ZipA were simultaneously inactivated in these cells by growth at 42°C, Z-rings failed to assemble (Fig. S3), as reported previously (Pichoff and Lutkenhaus, 2002). Under these conditions, MreB (Vats and Rothfield, 2007), MreC, MreD, Pbp2 and RodA failed to form ring structures (Figs 9B-i, C-i and D-i and 10A). On the other hand, assembly of proteins into the helical cytoskeletal structures along the length of the cell appeared unaffected in the ftsAts zipAts double mutant cells. This indicated that the Z-ring and/or downstream septasomal proteins are required for the assembly of MreB, MreC, MreD, Pbp2 and RodA into the cytoskeletal rings but are not needed for assembly of the helical cytoskeletal structures.
Evidence that the Z-ring alone was sufficient to support assembly of MreC, MreD and Pbp2 into the cytoskeletal rings was provided by localization studies in ftsAts and zipAts single mutants in which Z-rings are formed (Fig. S4) but other septasomal components fail to enter the septasomal apparatus (Pichoff and Lutkenhaus, 2002). Inactivation of FtsA or ZipA by growth of WC1006 (ftsA12ts) or PS223 (zipA1ts) at 42°C did not prevent the formation of MreC (Fig. 9B-ii, iii), MreD (Fig. 9C-ii, iii) or Pbp2 (Fig. 9D-ii, iii) ring structures. Similar results were obtained when ZipA was depleted by the growth of CH5/pCH32 (zipA::aph/repAts zipA+ftsZ+) at 42°C (Fig. S5A–C). These findings were similar to previous results for Yfp-MreB localization (Vats and Rothfield, 2007). Thus, the FtsZ ring appears to be the only component of the division apparatus required for the assembly of MreB, MreC, MreD and Pbp2 into cytoskeletal rings.
In contrast, the Z-ring alone was not capable of supporting assembly of RodA into the cytoskeletal ring structure. Fluorescence localization studies showed that Yfp-RodA failed to assemble into ring structures in ftsAts (Fig. 10B) and zipAts (Fig. 10C) single mutant cells grown at 42°C, where FtsZ rings were present (Fig. S4) but other septasomal proteins were absent. This showed that the presence of the FtsZ ring was not sufficient to support RodA assembly into the cytoskeletal rings, and suggested that additional components of the septasomal apparatus were required for this process.
To further investigate this, the distribution pattern of Yfp-RodA was determined in cells in which entry of septasomal proteins was blocked at different stages of septasome assembly. The septasome assembly sequence (Fig. 9A) was interrupted by growth of ftsKts and ftsQts cells under non-permissive conditions, or by treatment with aztreonam, a specific inhibitor of Pbp3 (FtsI) (Georgopapadakou et al., 1982). Under these conditions proteins downstream of FtsK, FtsQ or FtsI in the assembly pathway fail to enter the septasomal ring (Pichoff and Lutkenhaus, 2002; Geissler and William, 2005). This revealed that Yfp-RodA was present in cytoskeletal ring structures in ftsQts and FtsI-inactivated cells (Fig. 10E and F) but not in ftsKts cells (Fig. 10D). This implies that assembly of RodA into the cytoskeletal ring requires the presence of FtsK or some FtsK-dependent cellular component. In contrast Yfp-MreB rings were present in the FtsK-inactivated cells (Fig. 10G), confirming the specificity of the RodA assembly defect.
Although ZapA and ZapB proteins are not essential components of the division machinery, both localize to the midcell in an FtsZ-dependent manner (Gueiros-Filho and Losick, 2002; Goehring and Beckwith, 2005; Ebersbach et al., 2008). Neither ZapA nor ZapB was required for the formation of MreB or RodA rings (Fig. S6). However, the number of rings formed was lower than in the ftsAts, zipAts or aztreonam-induced FtsI– filamentous cells.
Cell division in rod-shaped bacteria produces daughter cells of identical cell shape. As part of this process, the helical cytoskeleton that is required for maintenance of rod shape is faithfully divided and segregated into the two progeny cells (Vats and Rothfield, 2007). The division and segregation of the E. coli helical cytoskeleton is associated with the formation and dynamic behaviour of the MreB-associated cytoskeletal ring structures that flank the potential division site and are formed prior to and independently of cytokinesis, based on studies of septation-blocked cells (Vats and Rothfield, 2007).
We show here that all the known proteins of the E. coli MreB-associated helical cytoskeleton are also components of the cytoskeletal ring structures, and the studies also add the RodA protein to the list of known components of the MreB-associated cytoskeleton. The cytoskeletal rings contain proteins thought to play a direct cytoskeletal function (MreB, MreC and MreD) and proteins that play a role in murein synthesis during cell elongation (Pbp2 and RodA). It remains to be determined whether assembly of these proteins into the cytoskeletal rings results from redistribution of molecules from the helical structures, as suggested by the accompanying decrease in intensity of the helices, or whether cytoskeletal ring assembly results from a change in the pattern of deposition of newly synthesized molecules, presumably accompanied by turnover of the components of the pre-existing helical elements.
The colocalization of proteins within the same annular structures, as shown by the double-label studies, implies that they act cooperatively in the structural organization and biological functions of the rings. The five proteins were also present in the coils of the helical structures along the long axis of the cell. Thus, it seems reasonable to view the cytoskeletal rings and helical structures as different forms of the same basic cytoskeletal elements that undergo a topologically specific reorganization event as part of the normal cell cycle.
What is the mechanism of assembly of the MreB-associated multiprotein helical cytoskeleton and cytoskeletal rings? It has been suggested that one or more of the Mre proteins may provide a scaffold for the entry of other proteins into the cytoskeletal helices (Daniel and Errington, 2003; Figge et al., 2004; Divakaruni et al., 2007). In E. coli, it was reported that MreB failed to form coils or rings in cells depleted of MreC, MreD or RodA (Kruse et al., 2005), implying that these three proteins might be required for the entry of MreB into the filamentous cytoskeleton. Similarly, in B. subtilis, the helical pattern of MreB appeared to depend on the presence of MreC and MreD (Leaver and Errington, 2005), and depletion of MreC or MreD was associated with a decrease in the number of cells with visible MreB-coiled structures (Defeu-Soufo and Graumann, 2005). It also has been reported that Pbp2 does not form midcell bands during the C. crescentus cell cycle and that the regular banding pattern of Pbp2 (probably indicating helical structures) is perturbed after prolonged growth in the absence of MreB synthesis (Figge et al., 2004). Colocalization studies of MreB and MreC in the helical structures of Caulobacter crescentus indicated that the two proteins were out of phase, implying they were not present within the same structure (Dye et al., 2005). On the basis of the existing evidence, it was not clear whether one or more of the cytoskeletal proteins provided a nucleation or docking site for entry of the other proteins.
The present study showed that in E. coli the MreB, MreC, MreD and RodA proteins assemble into cytoskeletal ring and coiled structures independently of each other in the spherical cells of mutants that lacked each of the other proteins, or that contained deletions of the chromosomal mreBCD operon (Fig. 11). This establishes that none of the proteins plays an obligatory role in entry of the other components and therefore these proteins do not serve a general scaffold function for cytoskeletal ring assembly. Similarly, each of the four proteins (MreB, MreC, MreD and RodA) assembled into filamentous coils that wound around the spherical cells of mutants that lacked one or more of the other cytoskeletal ring proteins. These results appear to exclude lattice-directed, hierarchical or concerted assembly mechanisms involving these proteins in formation of the cytoskeletal ring and coiled structures. Because of the spherical cell shape, it could not be determined whether the MreB, MreC, MreD and RodA coiled elements were organized as helical or quasi-helical structures or were present as more disorganized coiled elements. The independent assembly of the proteins into the cytoskeletal rings and coiled structures does not exclude the possibility that they interact with each other within the ring after their original capture.
Unlike the independent assembly patterns of MreB, MreC, MreD and RodA, the assembly of Pbp2 into rings and coiled structures was dependent on the presence of MreC (Fig. 11). This suggests that MreC is a docking site for Pbp2 entry or that Pbp2 enters the structures as a Pbp2–MreC complex. Either explanation would be compatible with cross-linking, co-immunoprecipitation and two-hybrid experiments that have shown interactions between Pbp2 and MreC (Figge et al., 2004; Divakaruni et al., 2005; van den Ent et al., 2006).
The cytoskeletal rings assemble adjacent to the septasomal cell division ring (Vats and Rothfield, 2007). What initiates this site-specific cellular differentiation event? The spatial relationship of the cytoskeletal and septasomal rings is consistent with the observation that the FtsZ component of the septasome was required for assembly of all proteins into the cytoskeletal ring but was not required for assembly of the same proteins into the helical cytoskeletal structures along the length of the cell. In the case of MreB, MreC, MreD and Pbp2, the Z-ring was the only septasomal component required for assembly of the proteins into the cytoskeletal rings (Fig. 11) as shown by the observation that the four proteins were present in the cytoskeletal rings of cells that expressed FtsZ and either FtsA or ZipA. Under these conditions, the Z-ring is assembled at the future division site but additional septasomal proteins are not incorporated (Pichoff and Lutkenhaus, 2002). In contrast, assembly of RodA into the cytoskeletal ring structures required a functional FtsK protein in addition to the basic FtsZ ring. Because FtsK plays a role in chromosome resolution (Steiner et al., 1999; Aussel et al., 2002) as well as cell division (Wang and Lutkenhaus, 1998), further work will be needed to fully interpret these results.
On the basis of the present results, it seems clear that the pathway leading to cytoskeletal ring assembly begins with the induction of FtsZ ring formation at midcell, normally directed by the MinCDE division site-specificity system (Rothfield et al., 2005). The FtsZ ring, in turn, induces assembly of MreB, MreC and MreD into the cytoskeletal rings, possibly by redistribution of molecules from the pre-existing helical structures. It is not yet known how this is accomplished. As discussed above, Pbp2 and RodA appear to enter the structure subsequently.
FtsZ might induce cytoskeletal ring assembly by interacting directly with the cytoskeletal proteins. In this case, the fact that the flanking cytoskeletal rings move apart subsequent to their formation (Vats and Rothfield, 2007) indicates that any direct interactions with FtsZ would be transient, restricted to the stage of ring assembly. Because MreB, MreC and MreD enter the cytoskeletal ring independently of each other, the use of FtsZ as a nucleation site or template for ring assembly would require that each of the proteins interact directly with FtsZ. This would imply that the 40.3 kDa FtsZ protein contains binding sites for at least eight different proteins – each of the Mre cytoskeletal proteins plus MinC, FtsA, ZipA, SulA and DicB (Mukherjee et al., 1998; Johnson et al., 2002; Cordell et al., 2003). If an intermediate protein were interposed between FtsZ and the cytoskeletal ring components, the intermediary protein would require sites for interaction with at least four proteins: FtsZ, MreB, MreC and MreD. At this time it is not known whether FtsZ can interact directly with any of the cytoskeletal proteins, and no candidates for intermediary assembly proteins have been identified. Because of the relative simplicity of an FtsZ–cytoskeletal protein interaction model, it will be of great interest to determine whether such interactions can be demonstrated. However, the large number of interacting sites within a single protein that are required by the direct interaction model suggests consideration of models that do not require a specific topological target protein. For example, the Z-ring could provoke a local change in membrane lipid organization or a local modification of murein structure, either of which might provide a domain for the initial assembly of the cytoskeletal ring components.
Studies of septation-blocked cells previously showed that the two doublet cytoskeletal rings move in opposite directions after their formation (Vats and Rothfield, 2007). It was speculated that the movement apart of the doublet rings, and the accompanying separation of the two progeny helical cytoskeletal structures, might be driven by insertion of new murein between the doublet rings (Vats and Rothfield, 2007). This idea is given further credence by the demonstration in the present work that the cytoskeletal rings contain at least two proteins involved in murein biosynthesis, Pbp2 and RodA. It is easy to imagine that these, and perhaps other murein-synthesizing proteins that may be present in the ring, are oriented to catalyse the progressive intercalation of new murein subunits into the murein sacculus between the rings. This would displace the rings away from midcell in opposite directions towards the cell poles, as was observed in time-lapse experiments (Vats and Rothfield, 2007). If this were correct, the new murein could correspond to the zone of preseptal murein that is formed, also in an FtsZ-dependent process, prior to initiation of septation in E. coli (de Pedro et al., 1997; Rothfield, 2003).
Strains, plasmids and growth conditions
The plasmids and E. coli strains used in the present study are described in Supplemental information (Table S1). To generate N-terminal Yfp fusion plasmids, mreC, mreD, pbp2 and rodA genes were amplified by PCR from chromosomal DNA of E. coli PB103, in a reaction that introduced XbaI and HindIII sites upstream and downstream of the genes respectively. XbaI/HindIII-digested PCR products were then ligated into the XbaI/HindIII site of plasmid pLE7 [Plac-yfp::mreB], thereby replacing the mreB gene to generate pVP1 [Plac-yfp::mreC], pVP2 [Plac-yfp::mreD], pVP3 [Plac-yfp::rodA] and pVP6 [Plac-yfp::pbpA]. pVP4 [Plac-yfp::rodAts] and pVP7 [Plac-yfp::pbpAts] were constructed in the same manner by ligating XbaI/HindIII-digested PCR products containing pbp2ts and rodAts fragments amplified from LMC582 [pbpAts] and SP5211 [rodAts] cells. ΔmreC and ΔmreD deletion mutants created using the linear DNA chromosomal deletions were confirmed by PCR using specific primers for mreC and mreD.
Strains were grown in LB medium at 37°C unless otherwise indicated; strains PS234, PS233, TOE1, CH5/pCH32 and LMC582 were grown at 30°C and then shifted to 42°C for 2 h; strain TOE44 was grown at 30°C and then incubated for 3 h at 42°C. Growth of CH5/pCH32 [zipA::aph/repA(ts) zipA+ftsZ+] at 42°C for 2 h leads to loss of ZipA rings due to loss of the zipA+ plasmid (Vats and Rothfield, 2007).
Antibiotics (100 μg ml−1 ampicillin or 25 μg ml−1 kanamycin) were added when plasmids were present. Where indicated, 1 μg ml−1 aztreonam was added for 2 h. Strain WC1006 was grown in the presence of 0.4% NaCl; RG60 was grown in the presence of 1% NaCl. For localization studies of Yfp fusion proteins, the strains were first grown in the presence of 0.4% glucose (w/v) at 37°C except for RG60 (ftsE/X::kan) and temperature-sensitive mutants which were grown at 30°C. The cells were re-suspended in the same medium without glucose and incubated for 90 min. After pelleting by centrifugation, the cells were washed once with the same medium, re-suspended to A600 0.05 and grown in the presence of 20 μM IPTG for 90–120 min at the indicated temperatures prior to preparation for microscopy. To study localization of Yfp fusion proteins in temperature-sensitive mutants, 5 μM IPTG was present for 90–120 min during the period of growth at 42°C.
Fluorescence microscopy of cells containing Yfp-labelled proteins was performed as described earlier (Vats and Rothfield, 2007). For immunofluorescence microscopy, polyclonal blot-purified anti-MreB and affinity-purified anti-FtsA, anti-ZipA and anti-FtsZ were used as primary antibodies. Cells were fixed in the culture medium using 1% formaldehyde and 0.01% glutaraldehyde for 1 h at room temperature and then permeablized with 0.5 μg ml−1 lysozyme in TGE buffer and stained with anti-MreB or anti-FtsZ antibodies as described earlier (Vats and Rothfield, 2007). For double-label experiments, strains containing Yfp fusion plasmids were grown in the presence of IPTG (see above). The cells were fixed, permeablized and stained with purified rabbit anti-FtsZ or rabbit anti-MreB as primary antibody and Alexafluor 594-conjugated goat anti-rabbit IgG as secondary antibody. Images were collected using the Openlab image acquisition program (Improvision). Alexa- and Yfp-fluorescence were detected using Texas Red (Chroma Set ID 41004) and YGFP (Chroma Set ID 41029) filters respectively (Vats and Rothfield, 2007).
We thank Mary Osborn and Aziz Taghbalout for useful discussions. This work was supported by Grant GM R37-06032 from the US National Institutes of Health.