Roles for MreC and MreD proteins in helical growth of the cylindrical cell wall in Bacillus subtilis



Actin homologues of the MreB family have an important role in specifying the morphology of many non-spherical eubacteria. The mreC and mreD genes have been implicated in control of cell morphology but their precise functions are unknown. In Bacillus subtilis the MreB homologue Mbl directs helical insertion of new cell wall material in the cylindrical part of the rod-shaped cell. Depletion of either MreC or MreD abolishes the control of cell shape. In the presence of high concentrations of magnesium cells depleted of MreC or MreD can be propagated indefinitely, although they have a spheroidal shape. We show that growth of the spheroidal mutants is based on insertion of new wall material at cell division sites and that this localized growth is dependent on cell division. Under some conditions the MreC and MreD proteins localize in a helical configuration. This localization pattern resembles that of the helical cables of Mbl protein. These results suggest that MreC and MreD act in a morphogenic pathway that couples the helical cytosolic Mbl cables to the extracellular cell wall synthetic machinery, which is critical for cylindrical elongation of the rod-shaped cells.


Bacteria have a wide range of cell morphologies. These morphologies are generally maintained across a wide range of growth conditions and throughout the growth and cell division cycles. Bacillus subtilis has a characteristic rod shape, with a cylindrical side wall and hemispherical poles. During division a cross-wall or septa is synthesized that forms the new poles of the resulting daughter cells. Most work on the control of this cell shape has been performed in B. subtilis and Escherichia coli through the isolation and characterization of rod mutants (Rogers and Thurman, 1978; Wachi et al., 1987; Levin et al., 1992; Varley and Stewart, 1992). Many of these mutations lie in genes encoding proteins that catalyse the synthesis of the cell wall, for example tagF, pbpA and pbpH (Pooley et al., 1992; Wei et al., 2003). This is consistent with an important role for the wall in determination or maintenance of cell shape.

Moreover, purified cell walls retain the shape of the cells from which they were obtained, implying that information about the shape of the cell is built into the wall (see Höltje, 1998). Other rod mutants have mutations in genes that do not directly catalyse the cell wall but still control cell shape, and include the mreBCD operon, and rodA (Henriques et al., 1998; Jones et al., 2001; Formstone and Errington, 2005). We recently discovered that mreB and its homologues mbl and mreBH in B. subtilis, encode actin homologues that polymerize to form cytoskeletal filaments that follow helical paths around the cell periphery just under the cytoplasmic membrane (Jones et al., 2001; Soufo and Graumann, 2004; Carballido-Lopez R., unpubl.). Similar results have subsequently been reported for mreB homologues from other diverse bacteria (Kruse et al., 2003; 2005; Figge et al., 2004; Gitai et al., 2004; Slovak et al., 2005). A novel staining method for nascent cell wall synthesis, based on use of a fluorescent derivative of vancomycin (a cell wall synthesis inhibitor), suggested that in B. subtilis at least, the insertion of new wall material also occurs in a helical manner (Daniel and Errington, 2003). This helical insertion appeared to be dependent on the Mbl protein but not on MreB. Although these experiments established a prominent role for Mbl, the detailed molecular mechanisms that couple Mbl to cell wall synthesis in rod-shaped bacteria remain to be worked out. The above results suggest that Mbl and MreB have divergent functions. However, the precise function of MreB is currently unclear (Formstone and Errington, 2005).

The final catalytic steps in synthesis of the major cell wall polymer, peptidoglycan (PG) in eubacteria, are carried out by penicillin-binding proteins (PBPs). PBP's have two distinct activities, transglycosylation and transpeptidation (Goffin and Ghuysen, 1998; Höltje, 1998). Transglycosylation is used to create linear glycan strands that are cross-linked by transpeptidation to produce the huge mesh work of PG that covers the whole surface of the cell. E. coli and B. subtilis both appear to have distinct wall synthetic systems specialized for cross-wall formation during division and side wall elongation during growth. Thus, specific inhibition or depletion of PBP3 of E. coli and Pbp2B of B. subtilis gives rise to a specific defect in cell division (Pogliano et al., 1997; Daniel et al., 2000). Similarly PBP2 of E. coli and either Pbp2A or PbpH of B. subtilis (the latter two seem to be functionally redundant (Wei et al., 2003) appear specialized for elongation. Absence of the division system leads to the formation of long, aseptate cell filaments, whereas absence of the elongation system leads to the formation of rounded cells. Proteins of the RodA and FtsW family also have specialized functions and are localized in side wall and division septum formation, although their biochemical function remains unclear (Henriques et al., 1998).

mreC and mreD– the other widely conserved genes thought to be involved in cell shape control – almost invariably lie in an operon immediately downstream of mreB (Wachi et al., 1989; Levin et al., 1992; Varley and Stewart, 1992). MreC is predicted to have a single transmembrane span with its major C-terminal domain outside the cytoplasmic membrane. The MreD protein is predominantly hydrophobic with four or six transmembrane spans and both N- and C-termini inside the cell. Several groups have investigated the genetics of the B. subtilis mreBCD region and concluded that these genes are involved in cell shape (Levin et al., 1992; Varley and Stewart, 1992; Lee and Stewart, 2003; Soufo and Graumann, 2003). We previously reported difficulties in complementing an mreB null-mutation that suggested polar effects on transcription of mreC and mreD (Jones et al., 2001). We also reported preliminary indications that mreC and mreD were not essential, based on ability to insert inactivation plasmids into each of those genes. It now appears that the plasmids may have integrated aberrantly, suggesting that correct insertion was lethal. Similar problems were reported by Varley and Stewart (1992) in an early characterization of this region. Lee and Stewart (2003) have recently reported that mreC is essential, although the cells could be protected from lysis by providing an osmotically balanced medium. The classical rodB1 mutation conferred a temperature-sensitive Rod phenotype (Rogers and Thurman, 1978) and was shown to lie in the mreD gene (Levin et al., 1992; Varley and Stewart, 1992), suggesting that it too is essential.

Here we describe a detailed analysis of the function of the MreC and MreD of B. subtilis. The results suggest that both proteins have critical roles in growth of the lateral wall of B. subtilis and that they probably have a direct role in the process whereby Mbl cables inside the cell govern helical synthesis of the cylindrical external cell wall.


mreC and mreD are essential under normal growth conditions

Genetic analysis of the mreBCD minCD locus of B. subtilis has been complicated by several factors: the possibility of more than one essential gene being present in the cluster; potential translational coupling between mreC and mreD; and expression levels in merozygotes with duplications of parts of the operon. Lee and Stewart (2003) reported that mreC is essential, but mreD has not been conclusively characterized. We previously reported that mreC and mreD are not essential (unpubl. results cited by Jones et al., 2001) but the single crossover integrations on which this conclusion was based may have occurred aberrantly. To determine definitively whether mreC and mreD are essential, we designed a strategy for making in-frame deletions of these genes. A selectable kanamycin resistance determinant (neo) was inserted just upstream of the mreB promoter. The site of insertion proved to be silent, as there was no detectable effect on growth or morphology (data not shown). Segments of DNA were amplified from this strain and ligated to generate in-frame deletions of mreC and mreD (annotated ΔmreC and ΔmreD). The ΔmreC and ΔmreD mutations were introduced into B. subtilis by linkage to the neo marker in a strain carrying a xylose-inducible copy of the mreBCD cluster (Pxyl-mreBCD) placed at an ectopic locus (amyE). Deletion derivatives of both mreC and mreD were readily generated and both turned out to be xylose dependent, suggesting that both genes were essential. mreD is located immediately upstream of, and probably coregulated with, the minC and minD genes (Levin et al., 1992; Varley and Stewart, 1992). The ΔmreD mutation was not polar on minC or minD because there was no significant increase in minicell formation (1000 cells counted). Moreover, the ΔmreD mutation was readily complemented by a Pxyl-mreBCD construction (strain 3452) under inducing conditions. To check that the ΔmreC mutation was not polar on mreD it was transformed into a recipient strain bearing Pxyl-mreC at amyE (strain 3437). The resultant strain (3461) grew well with wild-type morphology in the presence of xylose, demonstrating that the ΔmreC mutation was not polar.

To confirm the essential nature of mreC and mreD, the xylose inducible mreC and mreD strains (strains 3461 and 3452, respectively) were grown in the presence and absence of xylose (Fig. 1A and B). In the presence of xylose the growth of both these strains was comparable to that of the wild-type strain (data not shown). However, in the absence of xylose, both strains began to grow more slowly and, after the cultures were diluted back, there was no further growth (Fig. 1A and B). Hence, mreC and mreD are both essential for cell viability under these conditions. The essential nature of mreC was reported previously by Lee and Stewart (2003) using a construction in which two copies of mreD, minC and minD were present.

Figure 1.

Effect of the depletion of MreC or MreD on growth rate and cell morphology.
A and B. Growth curves of the MreC depletion strain (3461) (A) and the MreD depletion strain (3452) (B) in PAB supplemented with 0.5% xylose (▪), 20 mM MgCl2 (◆), 0.5 M Sucrose (▴) or with no addition (•).
C–F. Phase contrast images of strains 3461 and 3452 grown with (C, E) and without (D, F) 0.5% xylose.
G. Phase contrast images of the mreC null strain (3481) grown in PAB supplemented with MSM. Scale bar 5 µm.

Rescue of the lethal effects of the ΔmreC and ΔmreD mutations by magnesium chloride

Detailed analysis of the lethal phenotypes of the ΔmreC and ΔmreD deletions was hampered by cell lysis and low growth rates. Lee and Stewart (2003) previously showed that an mreC mutant could be rescued by growth in medium osmotically balanced by sucrose. For our experiments we initially chose a protoplast transformation medium containing 20 mM MgCl2, 0.5 M Sucrose and 20 mM Maleic acid (MSM) (Chang and Cohen, 1979). Penicillin assay broth (PAB) medium supplemented with MSM was able to prevent the lysis of cells depleted of MreC or MreD (strains 3461 and 3452) and restore an exponential growth rate (data not shown). In fact, depleted cells could be propagated indefinitely under these conditions. Control experiments excluded the possibility that sucrose induces expression from the Pxyl promoter (data not shown). Further experiments in media supplemented with 0.5 M sucrose or 20 mM MgCl2 alone showed that MgCl2 was the component of MSM responsible for restoring the growth of cells depleted of MreC or MreD (Fig. 1A and B). Sucrose reduced cell lysis in cells depleted of MreC or MreD but only improved the growth rate to a limited degree (Fig. 1A and B) and cells could not be indefinitely propagated in this medium.

Cell shape defects in the MreC and MreD depletion strains

Cells from the growth curves shown in Fig. 1 were examined microscopically by phase contrast microscopy. This revealed that both the inducible MreC and MreD strains (3461 and 3452) had normal rod-shaped cells when grown in the presence of xylose (Fig. 1C and E). After growth in the absence of inducer the cells became deformed and lytic; empty sacculi were frequent, accounting for the growth defect noted above (Fig. 1D and F). Cells were stained with a membrane dye (nile red; Greenspan et al., 1985) to visualize their septa and allow accurate cell length and width measurements. Depletion of MreC or MreD resulted in short, fat cells (Fig. 2A and D) (Table 1). After prolonged depletion in the presence of MgCl2 the cells became very short and fat and eventually spheroidal, but no significant lysis was observed (Fig. 2C and F). Addition of sucrose resulted in rounded cells (Fig. 2B and E), but some cell lysis was still observed (data not shown). Although cell shape was clearly affected in the MreC- or MreD-depleted cells they still appeared to be capable of making division septa (arrows in Fig. 2). The presence of xylose, sucrose or MgCl2 had no visible effect on the morphology of a wild-type strain (Fig. 2G–I).

Figure 2.

Shape phenotype of cells depleted of MreC and MreD in the presence of sucrose and magnesium visualized by fluorescence microscopy and TEM.
A–F. Cells of the MreC depletion strain (3461) (A, B and C) and cells of the MreD depletion strain (3452) (D, E and F), grown in PAB (A, D), PAB supplemented with 0.5 M sucrose (B, E) or 20 mM MgCl2 (C, F), and stained with nile red.
G–I. Cells of B. subtilis 168 grown in PAB supplemented with either 0.5% xylose (G), 0.5 M Sucrose (H) or 20 mM MgCl2 (I), stained with nile red. Scale bar 2 µm.
J–L. Cells of the MreD depletion Strain (3452) grown in PAB supplemented with 0.5% xylose (J), 0.5 M Sucrose (K) or 20 mM MgCl2 (L) and subjected to electron microscopy. Insets show enlargements of cell wall and arrows point to septum. Scale bar 500 nm. Magnification, approximately ×20 800, and ×104 000 for insets.

Table 1. Cell length and width measurements for wild type (168) and MreC and MreD inducible strains (3461 and 3452) grown with and without inducer in the presence of 20 mM MgCl2.
StrainInducerAverage cell length (µm)aAverage cell width (µm)a
  • a

    . Mean ± standard deviation; 500 cells counted.

1682.92 ± 0.700.81 ± 0.08
3461+2.76 ± 0.660.89 ± 0.07
1.67 ± 0.511.47 ± 0.28
3452+2.91 ± 0.770.89 ± 0.07
1.94 ± 0.561.42 ± 0.28

The shape defect of the MreC or MreD depleted cells might reflect a defect in cell wall synthesis. To investigate this hypothesis, the MreC and MreD inducible strains (3461 and 3452) were grown with and without inducer then subjected to transmission electron microscopy (TEM). Examples of typical MreD-depleted cells are shown in Fig. 2. Cells grown with inducer were indistinguishable from wild type in terms of morphology and particularly in the appearance of their cell walls (Fig. 2J). As expected, the depleted cells grown with sucrose or MgCl2 were generally spheroidal, but were still capable of forming septa, though these septa were often asymmetric and resulting in the formation of multiple compartments (Fig. 2K and L). This could be an indirect effect due to incorrect positioning of the division machinery in abnormally shaped cells. Such aberrant Z-rings are often observed in fat cells (see Fig. 1H of Jones et al., 2001; for example). Interestingly, for depleted cells recovered with MgCl2 the cell wall material was essentially indistinguishable from that of the controls, in terms of thickness and general appearance (Fig. 2J and L), whereas depleted cells rescued with sucrose had a thicker and more uneven cell wall (Fig. 2K). Similar results were obtained for the MreC depletion strain (data not shown, see also Lee and Stewart, 2003).

Isolation of an mreC null mutant

The results of the depletion experiments showed that mreC and mreD are essential, though viability could be rescued by high concentrations of MgCl2. Depletion experiments always carry the caveat that small quantities of wild-type protein are probably made even under repressing conditions. To bypass this problem and to determine whether mreC null strain could be made, chromosomal DNA from the MreC inducible strain (3461) was used to transform wild-type B. subtilis to neomycin resistance on NA plates. One hundred transformants were screened and all were found to have either failed to inherit the mreC deletion (22%) or to have acquired the deletion and the unlinked copy of Pxyl-mreC at the amyE locus (78%). When these experiments were repeated with selective plates containing 20 mM MgCl2 and 0.5 M sucrose mreC null mutants were readily obtained. These mutants were dependent on MgCl2, and had a severe shape defect. These results confirmed that mreC is essential for growth in normal levels of MgCl2 and that cells lacking mreC are viable only in the presence of high concentrations of MgCl2. Cells depleted of MreC are spheroidal but tend to maintain some polarity as evident in Fig. 2A and C. In comparison, cells of the mreC null strain (3481) showed little or no sign of a defined long axis (Fig. 1G). This suggests that the residual ability of the MreC depleted cells to elongate was due to background expression from the Pxyl promoter in repressing conditions, providing some MreC function. Unfortunately, this strain was unstable and difficult to propagate. Attempts to isolate an mreD null mutant by a similar approach have so far failed, suggesting that the requirement for MreD may be more stringent than for MreC.

Fluorescent vancomycin staining

The above results confirmed that MreC and MreD are required for correct cell shape and probably control the spatial organization of cell wall synthesis. To determine the localization of nascent PG synthesis in these mutants we used a recently developed method based on staining with a fluorescent derivative of the antibiotic vancomycin (VAN-FL) (Daniel and Errington, 2003). Vancomycin binds to the terminal d-ala–d-ala of PG precursors and probably to precursors recently incorporated into PG; however, after the subsequent processing of the d-ala–d-ala in mature PG, vancomycin no longer binds (Reynolds, 1989). Hence, it has been hypothesized that VAN-FL acts as a marker for precursor transport and nascent PG synthesis (Daniel and Errington, 2003). In wild-type B. subtilis VAN-FL gives a strong band of staining at the mid cell corresponding to division sites, and a fainter pattern of spots and transverse bands in the cylindrical side wall corresponding to helical insertion (Daniel and Errington, 2003). MreC and MreD inducible strains (3461 and 3452) grown with inducer had a wild-type VAN-FL staining pattern with staining at division sites and in the cylindrical part of the cell (Fig. 3A and B). In the absence of MreC the spherical cells had distinct, approximately medial, transverse bands of VAN-FL, probably corresponding to division septa, but no pattern of spots and transverse bands and very little staining, generally, away from the prominent bands and arcs (Fig. 3C). This suggests that cell wall synthesis mainly occurs at division sites and that helical incorporation is lost or greatly reduced in the depleted mutant. Similar results were obtained with the MreD-depleted cells (Fig. 3D). Occasionally, cells with a ring-shaped area of VAN-FL staining were evident, presumably corresponding to partially completed division septa viewed from the side (inset Fig. 3C).

Figure 3.

Cells depleted of MreC or MreD grow exponentially, but growth is dependent upon the division machinery.
A–D. Cells of the MreC and MreD depletion strains (3461 and 3452) grown with (A, B) and without (C, D) 0.5% xylose, stained with VAN-FL and subjected to fluorescence microscopy.
E–H. Cells of the MreC, FtsZ and MreD, FtsZ double depletion strains (3464 and 3465), depleted of either MreC (E) or MreD (F) alone, or depleted of both MreC and FtsZ (G) or MreD and FtsZ (H), strained with VAN-FL and subjected to fluorescence microscopy.
I and J. Growth curves of strains 3464 and 3465 depleted of MreC, diluted back then grown exponentially for 1.5 h then depleted of FtsZ (□) or with continued FtsZ synthesis (▪). Strain BB11 Grown with (•) or without (○) FtsZ.
K–S. Cells from time points 0 (K, N, Q) and 2.5 (L, M, O, P, R, S) of growth curves I and J, stained with nile red and subjected to fluorescence microscopy. Cells depleted of MreC (N, O), MreD (Q, R), or FtsZ (M) only and cells depleted of both MreC and FtsZ (P) or MreD and FtsZ (S).

MreC and MreD deficient cells require the division machinery for cell growth

If cells depleted of MreC or MreD were growing by cell wall synthesized only at septa, their growth should be dependent on the division machinery. To test this, derivatives of the mreC and mreD conditional strains were constructed that also carried a conditional, isopropyl β- d-thiogalactoside (IPTG)-dependent allele of the cell division gene ftsZ (from strain BB11, Beall and Lutkenhaus, 1991). Thus, these strains could be grown in the presence of xylose and IPTG and then depleted for either or both proteins by removing one or both inducers (strains 3464 and 3465). Cultures of these strains were grown with MgCl2, to prevent lysis in the absence of MreC or MreD. As before, cells depleted of MreC or MreD grew exponentially as spheroids. Once fully depleted of MreC or MreD, the cultures were diluted back to an optical density (OD) of 0.01. When the cultures had reached an OD of 0.1, cells were washed in fresh medium. The culture was split into two portions and IPTG was added to only one portion, then the growth curve was continued. After about 1 h the culture depleted of both MreC and FtsZ stopped growing, whereas the culture depleted of MreC alone continued to grow exponentially (Fig. 3I). Similar results were obtained for the MreD, FtsZ inducible strain, although the latter cultures grew to a slightly higher OD (Fig. 3J). As previously reported (Beall and Lutkenhaus, 1991), the FtsZ inducible strain (BB11) grew as filaments in the absence of IPTG (Fig. 3M) and as rods with IPTG (Fig. 3K and L). Thus, cells depleted for MreC or MreD require cell division to maintain exponential growth, and cells lacking FtsZ require MreC and MreD to elongate. Samples of cells from these cultures were stained with membrane dye or VAN-FL and observed by microscopy. Cells depleted only of MreC or MreD (Fig. 3N, O, Q and R) had a similar shape defect to the single depletion strains (Fig. 3E and F). The double-depleted cells had severely aberrant cell shapes, and were generally elongated and fat, or fat and rounded (Fig. 3P and S), suggesting that, if wall synthesis was occurring, it was severely defective in its spatial organization. VAN-FL staining of the cells was consistent with this notion. Small dots and patches of fluorescence were evident, scattered all over the surface of the cells (Fig. 4G and H). We suggest that this pattern is due to delocalized PG synthesis that results in cell death, although we cannot rule out the possibility that the VAN-FL staining pattern is disrupted in dying cells.

Figure 4.

GFP-MreC and GFP-MreD localize in a banded pattern that resembles the GFP-Mbl localization. Cells expressing GFP-MreC (A, C; strain 3416) and GFP-MreD (B, D; strain 3417) were subjected to fluorescence microscopy with the focus of the microscope through the medial plane of the cell (A, B) and with the focus adjusted by 250 nm (C, D). Insets show no-neighbours deconvolution and arrows point to bands and spots in the membrane.

MreC and MreD localize in helical patterns in the membrane

The above results suggested that MreC and MreD are required for the spatial organization of wall synthesis in the cylindrical part of the cell. Work from this laboratory has shown that this wall synthesis occurs in a helical pattern dependent on the actin homologue Mbl (Daniel and Errington, 2003). To test whether MreC and MreD were associated with these helical sites of synthesis, green fluorescence protein (GFP) fusions were made to both proteins. MreC is predicted to have a single transmembrane helix with the N-terminus of the protein inside the cell. MreD is predicted to have four to six transmembrane domains, again with the N-terminus likely to be inside. A strain was made by a Campbell-type integration vector that truncated the wild-type copy of the mreC gene and fused gfp to a full-length copy of the mreC gene. In this strain expression of the gfp fusion gene and the downstream genes was controlled by the Pxyl promoter (strain 3417). A similar strain was constructed fusing the gfp gene to mreD (3416). Both strains were viable but dependent on xylose. The cell morphology in the presence of xylose was essentially normal, showing that the fusion genes were functional. The functionality of both constructs supports the prediction that the N-termini of both proteins are intracellular because GFP is not normally functional when fused to a protein that is translocated to the extracellular side of the membrane (Feilmeier et al., 2000).

In preliminary experiments with cells of strains 3417 and 3416 grown in media supplemented with 0.5% xylose, GFP-MreC and GFP-MreD were clearly associated with the membrane and division septa. The fluorescence was relatively dispersed around the membrane as reported by Lee et al. (2003) using imunofluorescence against MreC, but some cells showed spots in the membrane where GFP-fluorescence was more intense. To counter any possible effects of over-expression on localization, the inducer was titrated to find the lowest concentration that could maintain growth. In these conditions GFP-fluorescence was reduced but spots of relatively high fluorescence were still observed in the membrane. These were particularly prominent when images were deconvolved and inverted (arrows in inset Fig. 4A and B). Because the GFP signal was weak, long exposure times were needed and the resultant bleaching made the collection of a Z-series impossible. Instead, a single, long-exposure image was taken with the focus of the microscope adjusted by 250 nm from the medial plane of the cell. This allowed visualization of the surface of the cell and revealed that GFP-MreC and GFP-MreD localize in a banded pattern (arrows in inset Fig. 4C and D). This pattern was reproduced in three separate experiments and could be seen in a high proportion of cells visible in a field of view. The patterns obtained for helical structures of bright GFP fusions, such as Mbl-GFP, differ depending on whether the plane of focus is medial, apical or basal. For a medial plane, the helix passes through the image plane almost perpendicularly, giving a pattern of dots along the edges of the cell. However, in apical or basal planes the helix lies almost parallel to the image plane and should appear as a pattern of oblique bands. The oblique banding patterns seen for both MreC-GFP and MreD-GFP were therefore consistent with an underlying helical pattern. The application of similar methods to GFP fusions to several other membrane proteins (TagG, TagF and YneS) did not give rise to this kind of pattern when grown under similar conditions, so it appears that this is not an artefact of the imaging method (data not shown).

It seemed likely that the helical patterns were due to an interaction with Mbl or MreB, which have previously been shown to have a filamentous helical localization in B. subtilis and to be involved in cell shape determination (Jones et al., 2001; Daniel and Errington, 2003). To test for such an interaction, the ΔmreB mutation of strain 3725 (Formstone and Errington, 2005) and an IPTG inducible copy of the mbl gene (strain 3468) were introduced into the Pxyl-gfp-mreC and Pxyl-gfp-mreD strains (3417 and 3416) creating strains 3469, 3470, 3471 and 3472. Unfortunately, the presence of the gfp-mreC or gfp-mreD allele enhanced the shape defect of the mreB and mbl null mutations, resulting in severe shape defects and unstable strains. Under such conditions the aberrant localization patterns observed (data not shown) could be attributed to either the lack of MreB or Mbl proteins or to a secondary effect resulting from the shape defect. This phenotype also suggested that the GFP fusion proteins were not completely functional. To avoid this problem the gfp fusions were examined in strains that placed the gfp-fusions at the amyE locus leaving the wild-type copy of the genes intact (strains 3475 and 3476). These strains were then combined with the ΔmreB mutation and the IPTG inducible copy of the mbl gene as before, generating strains 3478, 3479, 3480 and 3482. However, helical or banded GFP fluorescence was not seen for either fusion in the presence of its wild-type allele, suggesting that the wild-type proteins are preferentially recruited to the helical sites (data not shown).


The mreC and mreD genes are required for cell shape

The mreC and mreD genes lie in an operon immediately downstream of mreB in a wide range of bacteria. Although little is known about their precise function, they were likely to be involved in cell shape determination, based on their genetic association with mreB and the cell-shape phenotype of mutants in E. coli and B. subtilis (Wachi et al., 1987; Levin et al., 1992; Varley and Stewart, 1992). Several publications have addressed the construction of mre mutants and analysis of their phenotype (Jones et al., 2001; Lee and Stewart, 2003; Kruse et al., 2005); however, in each case the strain constructions were complicated by potential polar effects and gene dosage effects. We decided to analyse this definitively by making in-frame deletions of mreC and mreD, complemented by inducible copies of the deleted genes placed at the amyE locus. Neither in-frame deletion had a significant polar effect on downstream gene expression. The results showed that mreC is essential under normal growth conditions, which confirmed the previous result of Lee and Stewart (2003) for mreC. Previous work showing that the temperature sensitive rodB1 mutation mapped to mreD suggested that mreD may be essential (Levin et al., 1992; Varley and Stewart, 1992). We have now shown that mreD is also essential for growth in normal conditions. Overall, these results concur with recent work on E. coli showing that mreC and mreD are essential (Kruse et al., 2005).

Depletion of either MreC or MreD resulted in the formation of spheroidal cells that were prone to cell lysis. Because the cell wall is the major determinant of cell shape, MreC and MreD might be involved in cell wall synthesis. Also the observed lysis would support this idea as it suggests that the cell wall is weakened in the absence of MreC or MreD. However, the lysis was suppressed by MgCl2, but the spheroidal shape remained, and the structure of the cell wall (as seen by TEM) was comparable to that of wild-type cells. When MreC- and MreD-depleted cells were grown in the presence of sucrose the growth rate was not restored to that of wild type and the cell wall appear thick and uneven. This suggests that the effect of MgCl2 is mediated by changing the structure of the cell wall rather than by osmotic protection. There are several reports in the literature in which divalent cations have had such an effect. The rodB mutants (allelic to mreD) were shown to be dependent upon Mg++ (Rogers et al., 1976) and loss of ponA (encoding PBP1) results in an increased requirement for Mg++ or Ca++ (Murray et al., 1998). Recently, a similar phenotype was reported for an mreB null mutant (Formstone and Errington, 2005) (although in that case, a virtually normal rod shape was restored). The phenotype of Mg++ dependence, associated with a shape defect appears to be characteristic of genes involved in shape determination and cell wall synthesis. The mechanism whereby divalent cations have these effects is not yet clear, but the phenomenon should be useful in that media supplemented with Mg++ allows the study of cells depleted of MreC and MreD.

MreC and MreD influence cell wall synthesis

Our results suggest that MreC and MreD are involved in cell wall synthesis. Fluorescent vancomycin (VAN-FL) has proved to be a useful tool in probing the synthesis of the cell wall in bacteria (Daniel and Errington, 2003). Vancomycin acts as a marker for precursor transport and nascent PG synthesis by binding to the terminal d-ala, d-ala of the PG precursor prior to the processing that occurs in mature PG. VAN-FL staining of wild-type B. subtilis cells shows that synthesis occurs in two places: an intense band at about mid cell, representing sites of division; and in bands and spots associated with the cylindrical part of the cell. This pattern has been interpreted as being helical, and its presence is dependent on the actin homologue Mbl, which forms helical structures (Daniel and Errington, 2003). The staining pattern is consistent with a model of cell growth consisting of two phases: division, in which the septum is synthesized; and elongation, during which the cylinder elongates, eventually doubling in length. MreC- or MreD-depleted cells grew as spheroids or very short rods in the presence of MgCl2, suggesting that they are unable to elongate. Because the mreC null mutant had an even more extreme phenotype, it seems likely that the depletion strain retains residual MreC activity, allowing limited elongation.

Images of MreC- or MreD-depleted cells stained with a membrane dye showed that the cells form septa, and VAN-FL staining and TEM studies clearly showed that these septa contain wall material. This shows that loss of MreC or MreD does not prevent synthesis of cell wall material in cell division septa. Indeed, it seems that this wall synthesis is responsible for the continued growth of the mutant cells. In support of this idea growth of cells depleted of MreC or MreD was dependent on the cell division protein FtsZ. This dependence provides strong support for the notion that MreC and MreD are required for cell wall synthesis in the cylindrical part of the cell, leading to cell elongation.

MreC and MreD may connect the cytoskeleton to the cell wall synthetic machinery

Mbl and MreB form helical cables that underlie the membrane in B. subtilis (Jones et al., 2001; Carballido-Lopez and Errington, 2003; Soufo and Graumann, 2005; Formstone and Errington, 2005). Cell wall is made by helical incorporation of new PG in the side wall, and this helical incorporation is dependent upon Mbl (Daniel and Errington, 2003). These findings suggest that Mbl provides the positional information that directs cell wall synthesis in the cylinder, but until now, it was unclear how that positional information traversed the membrane. Cell wall synthesis is proposed to be synthesized by a multi-enzyme complex which possesses all the enzymatic activities required to inset a new strand into the existing cell wall (Höltje, 1998). We showed that at least under some conditions the transmembrane proteins MreC and MreD localize in a banded pattern in the membrane reminiscent of the Mbl or MreB localizations. This makes it likely that MreC and MreD interact with either Mbl or MreB and may provide the bridge between the intracellular cytoskeleton and the extracellular cell wall synthetic machinery. This may be a conserved function for these proteins because Kruse et al. (2005) have shown an interaction between MreC and MreB in E. coli, though they did not detect a helical localization pattern for either MreC or MreD. As the phenotypic consequences of mutating mreC and mreD are worse than mutating mbl, MreC and MreD proteins may have some biochemical role other than bridging the membrane.

Figure 5 shows a schematic view of the proteins that probably participate in the putative cell elongation machinery of B. subtilis, and their likely arrangement in the membrane. The similarity of the mreC and mreD phenotypes, and the proximity of their genes, suggests that they are a functional unit, and 2-hybrid analysis in E. coli and heterospecific expression experiments of the B. subtilis genes supports this idea (Lee et al., 2003; Kruse et al., 2005). These proteins could interact with Mbl on the cytosolic face of the membrane via the extreme N-terminus of MreC or either terminus of MreD. The major domain of MreC and/or surface-exposed loops of MreD could contact the extracellular cell wall synthetic machine. MreC proteins are predicted to have coiled-coil domains (Kruse et al., 2005), which are often involved in protein–protein interactions. Two likely candidates for the cell wall synthetic machine in B. subtilis are Pbp2A and PpbH because their mutant phenotypes identify them as being involved in elongation. Although neither of these two genes is essential, a Rod phenotype is observed if both are absent (Wei et al., 2003). Finally, the conserved RodA protein has been implicated in functioning of the cylindrical elongation machinery in both E. coli and B. subtilis (Henriques et al., 1998; de Pedro et al., 2001). It encodes a multiple transmembrane protein, probably involved in some aspect of PBP function. Elucidation of the biochemical functions of MreC and MreD in cell elongation represents an important challenge for future work in this area.

Figure 5.

Schematic model for the arrangement of MreC and MreD in a putative cell elongation complex. MreC and MreD lie in the membrane and act as a bridge between Mbl, on the cytosolic face of the membrane, and the cell wall synthetic machinery, including Pbp2A and PbpH, on the extracellular side of the membrane. MreC is shown dimerizing through its coiled coil domain, and RodA is included as part of the elongation machinery.

Experimental procedures

Bacterial strains

B. subtilis strains are described in Table 2.

Table 2. B. subtilis strains and plasmids used in this study.
NameRelevant GenotypeaConstruction
  • a

    . Resistance gene abbreviations are as follows: neo, neomycin; spc, spectinomycin; ble, phleomycin; cat, chloramphenicol: bla, ampicilin; erm, erythromycin and lincomycin.

  • Δ, deletion; Ω, insertion.

B. subtilis
 168 trpC2 Laboratory stock
 2056 trpC2ΩamyE::(spc Pxyl-mreBCD) Jones et al. (2001).
 3416 trpC2ΩmreD::pSG5271 (cat Pxyl-gfp-mreD)This work
 3417 trpC2ΩmreC::pSG5276 (cat Pxyl-gfp-mreC)This work
 3427 trpC2Ωneo3427This work
 3437 trpC2ΩamyE::(spc Pxyl-mreC)This work
 3452 trpC2Ωneo3427 ΔmreDΩamyE::(spc Pxyl-mreBCD)This work
 3460 trpC2Ωneo3427 ΔmreCΩamyE::(spc Pxyl-mreBCD)This work
 3461 trpC2Ωneo3427 ΔmreCΩamyE::(spc Pxyl-mreC)This work
 3464 trpC2Ωneo3427 ΔmreCΩamyE::(spc Pxyl-mreC) Ωchr::pJSIZΔpble (Pspac-ftsZ ble)This work.
 3465 trpC2Ωneo3427 ΔmreDΩamyE::(spc Pxyl-mreBCD) Ωchr::pJSIZΔpble (Pspac-ftsZ ble)This work.
 3468 trpC2Ωmbl::pSG5262 (erm Pspac-mbl)This work
 3469 trpC2ΩmreD::pSG5271 (cat Pxyl-gfp-mreC) Ωmbl::pSG5268 (erm Pspac-mbl)This work
 3470 trpC2ΩmreC::pSG5276 (cat Pxyl-gfp-mreD) Ωmbl::pSG5268 (erm Pspac-mbl)This work
 3471 trpC2Ωneo3427 ΔmreBΩmreD::pSG5271 (cat Pxyl-gfp-mreC)This work
 3472 trpC2Ωneo3427 ΔmreBΩmreC::pSG5276 (cat Pxyl-gfp-mreD)This work
 3475 trpC2ΩamyE::(spc Pxyl-gfp-mreC)This work
 3476 trpC2ΩamyE::(spc Pxyl-gfp-mreD)This work
 3478 trpC2ΩamyE::(spc Pxyl-gfp-mreC) Ωmbl::pSG5268 (erm Pspac-mbl)This work
 3479 trpC2ΩamyE::(spc Pxyl-gfp-mreD) Ωmbl::pSG5268 (erm Pspac-mbl)This work
 3480 trpC2Ωneo3427 ΔmreBΩamyE::(spc Pxyl-gfp-mreC)This work
 3481 trpC2Ωneo3427 ΔmreCThis work
 3482 trpC2Ωneo3427 ΔmreBΩamyE::(spc Pxyl-gfp-mreD)This work
 3725 trpC2Ωneo3427 ΔmreB Formstone and Errington (2005)
 BB11 chr::pJSIZ?pble (Pspac-ftsZ ble) Beall and Lutkenhaus (1991)
 pBEST501 bla neo Itaya et al. (1989)
 pMUTIN4 bla erm P spac -lacZ lacI Vagner et al. (1998)
 pSG1729 bla amyE3′ P xyl -gfp amyE5′ Lewis and Marston (1999)
 pSG4902 bla cat P xyl -gfp Wu and Errington (2003)
 pSG5256 bla amyE3′ P xyl -mreC amyE5′ This work
 pSG5262 bla erm P spac -mbl′-lacZ lacI This work
 pSG5264 bla amyE3′ P xyl -gfp-mreC amyE5′ This work
 pSG5265 bla amyE3′ P xyl -gfp-mreD amyE5′ This work
 pSG5271 bla cat P xyl -gfp-mreD′ This work
 pSG5276 bla cat P xyl -gfp-mreC′ This work

Construction of mreC and mreD inducible strains

A neomycin resistance cassette was inserted upstream of the mreBCD operon. Primer pairs LENm1, LENm2 and LENm3, LENm4 were designed to polymerase chain reaction (PCR) amplified two 3 kb fragments, which stretched upstream and downstream from a point between the radC STOP codon and the mreB promoter sequence (oligo sequences are shown in supplementary material (Table S1). These two fragments were digested with XbaI and ligated to the neo cassette from pBEST501, then used to transform into B. subtilis strain 168 to kanamycin resistance, giving rise to strain 3427. An in-frame deletion of mreC was constructed by PCR amplifying two 3 kb fragments from strain 3427: the first starting 3 kb upstream of the neo cassette and finishing 30 bp downstream of the mreC start codon (primers mreDSF3 and LE2CUS); and the second starting 30 bp upstream of the mreC stop codon and finishing 3 kb downstream (primers LE2CDS and mreUSF5). These two fragments were digested with SalI, ligated together, then used to transform strain 2056 (Jones et al., 2001) to Kanamycin resistance, thus creating strain 3460, which carried a deletion of mreC at the chromosomal locus and a xylose-inducible copy of mreBCD at the amyE locus. A similar strategy was used for mreD with primer pairs mreDSF3, LE2DUS and mreUSF5, LE2DDS creating strain 3452. To complement the in-frame deletion of mreC with an inducible copy of mreC alone a strain was constructed that carried a xylose-inducible copy of mreC at the amyE locus. The mreC gene was PCR amplified using primers mreCAE1 and mreCAE2 and inserted between the AvrII and XhoI sites of pSG1729, generating pSG5256, which was then used to transform wild-type B. subtilis to spectinomycin, generating strain 3437. Chromosomal DNA from strain 3460 was used to transform strain 3437 to kanamycin resistance, and transformants were screened by PCR for the ΔmreC allele, giving rise to strain 3461. For the construction of a null mutant of mreC B. subtilis 168 was transformed to kanamycin resistance with chromosomal DNA of strains 3461 and plated onto selective plates containing MSM and 15 µg ml−1 kanamycin. Transfomants were screened for strains that had the in-frame deletion but not the amyE::Pxyl copy of the gene, giving rise to strain 3481.

Construction of an IPTG-inducible mbl mutant

The first 300 bp of the mbl gene were PCR amplified with primers pMUT4:mblF and pMUT4:mblR, then cloned between the EcoRI and BamHI sites of pMUTIN4, creating pSG5262. pSG5262 was then used to transform B. subtilis 168 to erythromycin resistance, resulting in strain 3468 in which the full length mbl gene is expressed form the IPTG inducible promoter Pspac.

Construction of GFP fusions

The first 300 bp of the mreC gene was amplified by PCR with primers GFPLmreC1 and GFPLmreC2, then cloned between the ClaI and EcoRI sites of plasmid pSG4902, creating pSG5276. mreD was amplified with primers D902F2 and D902R and cloned between the BamHI and HindIII sites of plasmid pSG4902 (Wu and Errington, 2003), creating pSG5271. The parent plasmid pSG4902 contains a chloramphenicol resistance cassette and the Pxyl promoter upstream of the gfp gene followed by a multiple cloning site. Transformation of pSG5271 and pSG5276 into B. subtilis, with selection for chloramphenicol resistance, resulted in strains which carried a gfp fusion to either mreC or mreD at the mreBCD locus (3417 and 3416). The fusion gene in these strains was the only intact coding copy of the gene and expression of mreC or mreD– as well as the expression of the downstream genes – was controlled by the Pxyl promoter. To combine these gfp-fusions with the Pspac-mbl and ΔmreB alleles chromosomal DNA from strain 3468 was used to transform strains 3417 and 3416, with selection for resistance to erythromycin and chloramphenicol, generating strains 3469 and 3470 respectively. To combine the gfp-fusions with the ΔmreB allele, chromosomal DNA from strain 3725 was used to transform strains 3417 and 3416, with selection for resistance to kanamycin and chloramphenicol generating strains 3471 and 3472 respectively. gfp-fusions with the Pspac-mbl and ΔmreB alleles, Pxyl inducible gfp-mreC and mreD fusions were placed at the amyE locus. Primers PxylF and mreC1729R were used to PCR amplify gfp-mreC from strain 3417, this fragment was then cloned into pSG1729 at the AvrII and EcoRI sites of pSG1729 resulting in pSG5264. For mreD, primers D902F2 and mreD1729R were used to amplify full length mreD from B. subtilis 168, this fragment was then cloned into pSG1729 at the EcoRI and BamHI sites, resulting in pSG5265. pSG5264 and pSG5265 were then used to transform B. subtilis 168 to spectinomycin resistance, generating strains 3475 and 3476 respectively. To combine the gfp-fusions with the Pspac-mbl allele, chromosomal DNA from strain 3468 was used to transform strains 3475 and 3476 to erythromycin resistance, generating strains 3478 and 3479 respectively. To combine the gfp-fusions with the ΔmreB allele, chromosomal DNA from strain 3725 was used to transform strains 3475 and 3476 to kanamycin resistance generating strains 3480 and 3482 respectively.

Growth conditions

For growth experiments, B. subtilis strains were grown at 37°C in Difco Antibiotic Medium 3 (PAB) supplemented with 0.5 M sucrose, 20 mM maleic acid and 20 mM MgCl2 (MSM) (Chang and Cohen, 1979) or with 0.5 M sucrose or 20 mM MgCl2 alone and 0.5% xylose was added as necessary. For imaging of GFP fluorescence strains were grown at 30°C in S medium (Sharpe et al., 1998) with 0.3% xylose or CH medium with 0.3% xylose and 20 mM MgCl2. For VAN-FL staining strains were grown at 30°C PAB supplemented as before. For strains expressing genes under the control of the Pspac promoter, 0.2 mM IPTG was added to the media. For transformations cells were plated on nutrient agar (Oxoid) supplemented with 5 µg ml−1 chloramphenicol, 5 µg ml−1 kanamycin, 50 µg ml−1 Spectinomycin, 1 µg ml−1 phleomycin or 0.3 µg ml−1 erythromycin with 25 µg ml−1 lincomycin for selection of cat, neo, spc, ble and erm markers respectively. Nutrient agar was supplemented with 0.5% xylose, 1 mM IPTG or MSM, as appropriate. The concentration of kanamycin was increased to 15 µg ml−1 for transformations plated on nutrient agar containing MSM.

VAN-FL staining

Fluorescent vancomycin (Vancomycin BODIPY® FL conjugate) was purchased from Molecular Probes. VAN-FL staining was carried out as described by Daniel and Errington (2003) (Daniel and Errington, 2003). Briefly, VAN-FL was mixed one to one with non-labelled vancomycin and added to cultures giving a final concentration of 1 µg ml−1. The culture was then incubated for 15 min to allow absorption of the antibiotic. Cells were then fixed in 1.6% formaldehyde in PBS and incubated at 4°C for at least 1 h. Cells were then viewed by fluorescence microscopy with a 484 nm filter set.

Microscopic imaging

For fluorescence microscopy a 0.5 µl of sample of the culture was placed on a microscope slide coated with a thin layer of 1% agarose in Spizizen Minimal Medium (Anagnostopoulos and Spizizen, 1961) (w/v), and a coverslip was placed on top. Images were taken using a Sony CoolSnap HQ cooled CCD camera (Roper Scientific Ltd) attached to a Zeiss Axiovert 200 M microscope. The digital images were analysed with Metamorph version 4.6.9 (Universal Imaging), and the final Figs were generated by Photoshop version 6 (Adobe Systems Incorporated). Image manipulation was kept to a minimum. For general purposes images were scaled then saved as 8-bit images. For GFP imaging out of focus light was removed using the No Neighbours Deconvolution utility of the Metamorph program. The filter size, scale factor and result scale parameters were set to: 9, 0.97 and 2; 9, 0.90 and 9; 9, 0.97 and 2; and 9, 0.60 and 9 for Fig. 4A, B, C and D respectively. Images were then inverted.

Electron microscopy

Strains 3461 and 3452 were propagated with or without xylose in PAB supplemented with or without 0.5 M sucrose or 20 mM MgCl2. Cell samples were fixed in the culture medium by the addition of gluteraldehyde to a final concentration of 2.5%. Cells were then pelleted and fixation was continued overnight at 4°C. Cell pellets were washed with 200 mM phosphate buffer then post fixed with 1% osmium tetroxide in 100 mM phosphate buffer and incubated for 1 h at 4°C. Pellets were then washed in distilled water and en bloc stained with 2% aqueous uranyl acetate for 1 h at 4°C in the dark. The cells were then washed twice in water, dehydrated in acetone, and washed in propylene dioxide. Pellets were imbedded in epon-araldite resin that was allowed to polymerize over night at 65°C. Sections of 70 nm were cut, counter stained with 0.2% lead citrate, then examined and photographed on a Tecnai 12 electron microscope (FEI, Eindhoven, Holland).


This work was supported by grants from the Biotechnology and Biological Sciences Research Council and the Human Frontier Science Programme. M.L. was supported by a Graduate Studentship from the Medical Research Council. We thank Richard Daniel and Alex Formstone for helpful discussions, Heath Murray for critical reading of the manuscript and Mike Shaw for assistance with electron microscopy.