The YvcK protein is required for morphogenesis via localization of PBP1 under gluconeogenic growth conditions in Bacillus subtilis

Authors


E-mail galinier@ifr88.cnrs-mrs.fr; Tel. (+33) 4 91 16 45 71; Fax (+33) 4 91 71 89 14.

Summary

The YvcK protein was previously shown to be dispensable when B. subtilis cells are grown on glycolytic carbon sources but essential for growth and normal shape on gluconeogenic carbon sources. Here, we report that YvcK is localized as a helical-like pattern in the cell. This localization seems independent of the actin-like protein, MreB. A YvcK overproduction restores a normal morphology in an mreB mutant strain when bacteria are grown on PAB medium. Reciprocally, an additional copy of mreB restores a normal growth and morphology in a yvcK mutant strain when bacteria are grown on a gluconeogenic carbon source like gluconate. Furthermore, as already observed for the mreB mutant, the deletion of the gene encoding the penicillin-binding protein PBP1 restores growth and normal shape of a yvcK mutant on gluconeogenic carbon sources. The PBP1 is delocalized in an mreB mutant grown in the absence of magnesium and in a yvcK mutant grown on gluconate medium. Interestingly, its proper localization can be rescued by YvcK overproduction. Therefore, in gluconeogenic growth conditions, YvcK is required for the correct localization of PBP1 and hence for displaying a normal rod shape.

Introduction

The YvcK protein was previously shown to be dispensable when B. subtilis cells are grown on glycolytic carbon sources but essential for growth on Krebs cycle intermediates and on carbon sources metabolized via the pentose phosphate pathway (Görke et al., 2005). Indeed, a yvcK mutant exhibited normal cell morphology and normal growth on minimal medium containing glycolytic substrates like glucose, sucrose, fructose, glycerol or inositol. However, it did not grow either on DSM solid medium or on minimal medium plates containing arabinose, ribose, gluconate, citrate or succinate as carbon source; its growth was significantly impaired on fumarate and malate. In liquid medium, the yvcK mutant grew during 4–6 h, harbouring abnormal cell shapes followed either by lysis or by growth arrest. For example, during the first hours of growth, yvcK mutant cells exhibited aberrant morphologies; either as very long filaments in DSM medium or they swelled in minimal medium containing gluconate or malate (Görke et al., 2005). As already observed in B. subtilis mutants affected in cell wall formation and/or in morphogenesis (Formstone and Errington, 2005; Lazarevic et al., 2005; Carballido-López et al., 2006), high magnesium concentration restored normal growth and cell morphology of the yvcK mutant (Görke et al., 2005).

The YvcK protein belongs to the UPF0052 uncharacterized protein family related to CofD, a transferase involved in coenzyme F420 biosynthesis in archaea and high G+C content Gram-positive bacteria (Graupner et al., 2002). YvcK is present in a wide variety of bacteria and its role in all the bacteria that do not possess coenzyme F420 is still unknown. Bio-informatic analysis and data from 3D structures (Forouhar et al., 2008) suggest that YvcK could be a transferase with an NAD(P)-binding Rossmann-fold domain. However, despite the conservation of a glycine-rich loop, the YvcK homologues may use different substrates. In particular, no binding of nucleotides to YvcK has been detected whereas GTP and GDP were found to interact with CofD (Forouhar et al., 2008). YvcK may catalyse a different biochemical reaction from that of the transferase CofD.

The Escherichia coli homologue was shown to partially rescue YvcK in B. subtilis, because the E. coli ybhK gene restores to some extent the phenotype of the B. subtilis yvcK mutant. This observation suggests that the two proteins have been functionally conserved in these Gram-positive and Gram-negative bacteria (Görke et al., 2005). In Staphylococcus aureus, the yvcK gene, like many genes that play a role in metabolism, is essential (Chaudhuri et al., 2009). It has been proposed that, in a B. subtilis yvcK mutant, the synthesis of cell wall precursors from carbon sources that do not enter directly in the upper part of glycolysis would be slowed down (Görke et al., 2005).

In this paper, we observed that YvcK is organized as a helical structure within the cell. This organization has already been observed for the cytoskeletal proteins belonging to the MreB family (Carballido-López et al., 2006). These proteins form helical cables in the cell that encircle and run along the length of the rod-shaped bacterial cell (van den Ent et al., 2001; Jones et al., 2001; Carballido-López and Errington, 2003). Interestingly, an overproduction of YvcK is able to restore a normal rod-shaped morphology in an mreB mutant grown on PAB whereas an additional copy of mreB is able to restore a normal rod-shaped morphology in a yvcK mutant strain grown on gluconeogenic carbon sources. Furthermore, overproduction of YvcK in an mreB mutant strain on one hand, and reciprocally, overproduction of MreB in a yvcK mutant strain both rescue the restoration of PBP1 localization.

Results

YvcK is organized as a helical structure in the cell

In order to obtain insights in the role of YvcK in B. subtilis, we determined its subcellular localization by immunofluorescence microscopy using antibodies directed against YvcK (Fig. 1A). After verifying that no signal was detected in the yvcK mutant strain by Western blot and by microscopy, we analysed the signal in a wild-type strain. For most of the cells observed, we detected a regular pattern of fluorescence. Three-dimensional deconvolution of Z stacks taken from various cells indicates that the YvcK protein is organized into coiled helical structures extending from one pole to the other (Fig. 1A4 and A5).

Figure 1.

YvcK is organized as a helical structure in vivo. A. Localization of YvcK by immunofluorescence. A1 – Western blot analysis: detection of YvcK in 20 µl of crude extract using an anti-YvcK antibody: in cells from strain SG63 deleted of YvcK (lane 1) and in the wild-type strain (lane 2). A2 – DIC and A3 – immunofluorescence microscopy of the WT strain. A4 and A5 – The brightest image of the deconvoluted stack of eight cells (A4) and of two cells with a higher magnification (A5) suggesting a helical structure of the YvcK protein. The scale bars represent 1 µm. B. Localization of YvcK-GFP by live-cell fluorescence microscopy. B1 – Western blot analysis: detection of YvcKGFP in 20 µl of crude extract using an anti-GFP antibody: in the wild-type strain containing either GFP (strain SG187) (lane 1) or YvcKGFP fusion in a wild-type background (strain SG188) (lane 2) and in ΔyvcK background (strain SG116) (lane 3). B2 – DIC and B3 – Fluorescence microscopy of the strain SG116 expressing the YvcKGFP fusion. B4 – Fluorescence pattern of an isolated cell of strain SG116 cropped prior deconvolution. B5 – The brightest image of the deconvoluted stack showing a helical structure of the YvcKGFP protein. B6 – Bicubic extrapolation for x,y and z dimensions of the deconvoluted stack in order to obtain a smoothed image. B7 – Three-dimensional reconstruction of the YvcK cytoskeletal structure, top viewpoint (left panel) and bottom viewpoint (right panel). B8, B9 – Transverse (x,z planes) and longitudinal (y,z planes) sections of the stack deconvoluted and extrapolated. B10, B11 – The two images represent the longitudinal and transverse sections of the deconvoluted stack obtained from a cell from SG116 strain, that expressed the fluorescent protein YvcKGFP (in green), stained by FM4-64 (in red). The scale bars represent 1 µm.

To confirm the localization pattern of YvcK observed by immunofluorescence, we constructed a strain encoding the YvcK protein fused at its C-terminus with the green fluorescent protein (GFP) (Fig. 1B1); the expression of yvcKgfp is under the control of the native promoter of yvcK, located in front of the yvcI gene (Galinier et al., 1997; Görke et al., 2005). To test the functionality of the YvcKGFP fusion, bacteria grown on gluconate as sole carbon source were observed by microscopy. They present a normal rod shape (Figs 1 and 2), thus indicating that YvcKGFP is functional. To obtain a better fluorescence signal, we also constructed strain SG116 to overproduce YvcKGFP. We further analysed YvcKGFP localization in both strains by live-cell fluorescence microscopy (Figs 1B3 and 2B). In many exponential phase cells (n = 100), we observed a regular pattern of fluorescence as previously visualized by immunofluorescence. The deconvoluted images elucidated the precise 3D cellular organization of YvcKGFP (Fig. 1B4 to B9 and Movie S1). The 3D reconstruction clearly shows the two strands of a closed coiled circle (Fig. 1B7). The geometrical parameters seem less regular than in 2D view; for the five coils visible from the top viewpoint (Fig. 1B7 left panel), only the second and third ones were parallel. The transverse and longitudinal sections (Fig. 1B8 and B9) of the YvcK structure show an inner non-fluorescent region surrounded by ring-like fluorescent structures, probably composed of helix coils. The membrane staining of a cell from SG116 strain by FM4-64 showed that the helical structure of the YvcKGFP protein runs up and down the cell adjacent to the inner surface of the membrane (Fig. 1B10 and B11).

Figure 2.

Subcellular localization of YvcKGFP in various genetic backgrounds. All the bacteria were grown on PAB medium then analysed by DIC and fluorescence microscopy. The YvcKGFP protein was overexpressed from the promoter Pxyl in the presence of 0.5% xylose or expressed from the promoter PyvcI. A. The YvcKGFP protein was overexpressed in the mutant strain SG129 (ΔmreB, ΔyvcK, amyE::Pxyl yvcKgfp). B. The YvcKGFP protein was expressed in the wild-type strain SG163 (yvcKgfp), in the strain SG172 (ΔmreB, yvcKgfp) and in the double mutant strain SG156 (yvcKgfpΔmblΔmreBH) (B). The fluorescence microscopy was illustrated, for each strain, by the brightest image of the deconvoluted stack. The scale bars represent 1 µm.

The mechanism responsible for the intracellular organization of YvcK is unknown. YvcK could have the intrinsic ability to self-assemble or to be associated with an already assembled template filament present in the cell. However, despite our efforts, we did not obtain any clue that it forms a filamentous structure on its own. For example, we checked whether YvcK was able to form a filamentous structure in yeast cells as previously shown for MreB (Srinivasan et al., 2007). We did not see any filament formation of the YvcKGFP protein in the yeast Saccharomyces cerevisiae in contrast with the filament formation observed with MreBGFP, which served as positive control (data not shown).

YvcK localization does not require the presence of MreB

YvcK subcellular organization is similar to the cytoskeletal filament formed by the three actin-like proteins. Indeed, in B. subtilis, MreBH, MreB and Mbl colocalize and appear to form a triplex helical organization closely associated with the inner surface of the membrane (Carballido-López et al., 2006). In order to determine whether this cytoskeleton is required for the cellular organization of YvcK, the localization of YvcKGFP was firstly monitored by fluorescence microscopy in mreB mutant cells. The chromosomal DNA of SG116 strain was used to transform 3725 strain to obtain SG129 strain. Surprisingly, while the mreB mutant (3725 strain) had an altered morphology (Formstone and Errington, 2005), the SG129 strain deleted for mreB but expressing YvcKGFP from Pxyl, has a normal morphology (Fig. 2A) when bacteria were grown on PAB medium in the absence of additional Mg2+ and sucrose. Furthermore, the YvcK coiled structure is present. In a study carried out on EF-Tu that was shown to interact and to colocalize with MreB in B. subtilis, the disruption of mreB inhibited the formation of the EF-Tu structures (Defeu Soufo et al., 2010). We also constructed a strain deleted for mreB and containing YvcKGFP expressed from the yvcK native promoter PyvcI. As expected, this strain was found to present the same morphology alterations as an mreB mutant, on PAB medium in the absence of Mg2+ and sucrose (see Fig. 2B). Furthermore, despite an aberrant cell shape in this strain, the YvcKGFP seems to present a regular pattern of fluorescence but its patchy localization is not enough to warrant the YvcK helical localization.

We also tested if YvcK requires at all or to some extent the presence of the two other MreB isoforms, MreBH or Mbl, for its cellular organization as a partial functional redundancy has been observed among the three MreB isoforms (Kawai et al., 2009a). We first attempted to obtain an mreB mbl mreBH triple mutant strain that expressed a YvcKGFP fusion. However, despite our efforts, we could not transform an mreB mbl mreBH triple mutant (Schirner and Errington, 2009) that is severely affected in its shape and in its cell wall and thus seems ‘untransformable’. We then carried out immunolocalization of YvcK in this triple mutant. Nevertheless, as a result of the aberrant cell shape and the abnormal cell wall, the YvcK localization was ambiguous, like that obtained in a spheroplast from wild-type cell, and we could not determine if YvcK required the three MreB isoforms for its correct localization (data not shown). Even though we do not have any conclusive evidence with the triple mutant, we were able to determine the localization pattern of YvcKGFP in mreBH mbl double mutant cells. As shown in Fig. 2B, the YvcKGFP seems to be still organized in the mreBH mbl double mutant cells, even if the helical structure seems altered because of the abnormal size and shape of the cell.

Conversely, we also investigated whether the YvcK structure influenced the cytoskeletal organization of MreB. We therefore analysed the localization pattern of CFPMreB by fluorescence (Fig. S1). We showed that, in the yvcK mutant cells, the MreB coiled structures were unaffected when bacteria were grown on glycolytic media (xylose and glucose used as carbon sources). Indeed, the PAB medium, which was used to study the mreB mutant strains (Formstone and Errington, 2005), is a complex medium comprised of enzymatic digests of proteins combined with yeast and beef extracts that contains glucose as carbon source. Therefore, when bacteria were grown in the presence of low concentration of inducer and on gluconate as sole carbon source, they displayed an abnormal shape (Fig. S1). In these conditions, the CFPMreB seems to present a regular pattern of fluorescence but, similarly to the results obtained with YvcK, its patchy localization is not enough to warrant the MreB helical localization.

All these observations suggest that YvcK localization does not require the presence of MreB and reciprocally, MreB localization does not require the presence of YvcK. In contrast, a normal rod-shape is necessary to observe a normal helical localization of both proteins.

YvcK structures localize separately from MreB filaments

Because we did not manage to characterize the cellular organization of YvcK in an mreB mbl mreBH triple mutant strain, we analysed whether the YvcK and MreB structures colocalize. Indeed, if the YvcK helical structure needed the cytoskeletal template formed by the MreB proteins, we expected the two proteins to colocalize, as it was already shown for EF-Tu (Defeu Soufo et al., 2010). Therefore, we analysed the SG224 strain expressing both YvcKGFP and mRFPMreB by fluorescence microscopy (Fig. 3). We observed that most of the cells expressed the two fluorescent proteins (Fig. 3A). The images (Fig. 3B) and the resulting profile plots (Fig. 3C) indicate that mRFPMreB filament seems not to colocalize with YvcKGFP structures, although overlapping signals were also observed. Indeed, a detailed analysis of 17 cells expressing both fluorescent protein fusions shows that, of 107 fluorescent foci analysed, 55% were coincident (overlay of peaks corresponding to MreB and to YvcK from the profile plots), whereas 45% of the foci were distinct i.e. with at least a difference of ±0.2 µm for the top of the peak (Fig. 3C). Furthermore, in a single cell, the two structures do not harbour exactly the same pitch (about 0.9 µm for MreB and about 0.8 µm for YvcK) and thus can not colocalize all along the cell axis. However, we cannot rule out the possibility that YvcK and MreB colocalize partially and that the MreB helix provides several anchor points that are sufficient to help stabilize the helical structure of YvcK. To detect a putative interaction between these two proteins, we carried out several assays; namely, bacterial two hybrid, pull down and Tap-tag assays but we did not detect any specific interactions between YvcK and MreB. Altogether, our results suggest that the helical structure of YvcK in the cell does not use the cytoskeletal filament formed by the MreB proteins as template.

Figure 3.

Subcellular localization of YvcKGFP and mRFP MreB. The strain SG224 was grown on LB medium in the presence of 1% xylose during 2 h, then cells were analysed by fluorescence microscopy. The YvcKGFP protein was expressed from the yvcK promoter PyvcI whereas mRFPMreB protein was overexpressed from the promoter Pxyl. A. A merged image where the two proteins YvcKGFP and mRFPMreB are shown in green and in red, respectively; the yellow and orange regions represent regions of colocalization. The scale bars represent 1 µm. B. Three individual cells. YvcK (green) and MreB (red) are shown in the left and middle columns, respectively; overlays are shown in right panels. C. Profile plots through reoriented images of corresponding cells displayed in B.

YvcK overproduction compensates the lack of MreB

We observed that in an mreB mutant strain grown on PAB medium in the absence of Mg2+ and sucrose, the yvcKgfp fusion expressed from PyvcI did not restore a normal shape whereas the same recombinant gene expressed from Pxyl promoter restored it. We therefore decided to measure the expression level of yvcKgfp from PyvcI and from Pxyl by q-RT-PCR (Table 1). We showed that the expression of yvcKgfp driven from the Pxyl is at least sixfold higher than that obtained from the native promoter PyvcI (Table 1). We carried out quantitative Western blot analysis to compare the protein level in these strains (Fig. 4). We showed that the crude extracts obtained from strains that overexpressed yvcKgfp from Pxyl contained about 6- to 8.5-fold more of YvcKGFP protein in comparison with those obtained from strains that expressed yvcKgfp from PyvcI native promoter.

Table 1.  Comparison of the expression of yvcKgfp driven by the native promoter of yvcK or by the strong Pxyl promoter.
StrainGenotypeExpression of yvcKgfp
  1. All the strains were grown on LB medium supplemented with 0.3 M sucrose, 25 mM MgSO4 and 0.5% (w/v) xylose and the cells were collected at OD600 of 0.4. After RNA extraction, the relative expression of the gene yvcKgfp was determined by q-RT-PCR using a couple of specific primers yvckRT5′ and yvckRT3′.

SG163yvcKgfp1.00 ± 0.05
SG172ΔmreB yvcKgfp0.89 ± 0.06
SG116ΔyvcK Pxyl yvcKgfp6.78 ± 0.12
SG129ΔmreB ΔyvcK Pxyl yvcKgfp9.22 ± 0.32
Figure 4.

Comparison of quantities of YvcKGFP expressed from PyvcI or from Pxyl by quantitative Western blot. The YvcKGFP protein was expressed in the wild-type strain SG163 and in the ΔmreB, yvcKgfp strain SG172 (A), and overexpressed in the strain SG116 (ΔyvcK, amyE::Pxyl yvcKgfp) and in ΔmreB mutant strain SG129 (ΔmreB, ΔyvcK, amyE::Pxyl yvcKgfp) (B). The four strains were grown on LB medium supplemented with 0.3 M sucrose, 25 mM MgSO4 and xylose 0.5% (w/v) until DO600 = 0.4. After centrifugation, the four pellets were resuspended by 80 µl lysis buffer. For each one, 2, 4, 8 and 16 µl of crude extract were separated by SDS-PAGE. After transfer, YvcK was detected using antibodies directed against YvcK. To estimate the relative quantity of YvcK in crude extract and to compare the different lanes, we used an internal standard, the PrkC protein, which was detected using specific antibodies (Madec et al. 2002). After quantification, we showed that the overproducer SG116 synthesized sixfold more of YvcKGFP than SG163 whereas the overproducer SG129 synthesized 8.4-fold more of YvcKGFP than SG172.

Thus, we investigated the effect of an overproduction of YvcK, without the GFP tag, in the mreB mutant background and vice versa (Fig. 5). We observed that an overproduction of YvcK rescued a normal rod shape of an mreB mutant grown on PAB medium in the absence of additional Mg2+ and sucrose (Fig. 5A). This rescue is surprising because: (i) YvcK and MreB do not harbour any similarity of sequence and of structure and (ii) the carbon source provided by PAB is glucose and no phenotype was detected for yvcK mutant when bacteria were grown on glycolytic carbon source (Görke et al., 2005). This observation suggests that, in particular conditions like those generated by the disruption of mreB, an overproduction of YvcK has an observable cellular role when bacteria were grown on glycolytic carbon sources, like glucose. We also observed that an overproduction of YvcK rescued a normal rod shape of an mreB mutant grown on a gluconeogenic carbon source as sole carbon source like gluconate in the absence of additional Mg2+ and sucrose (Fig. S2). Conversely, when the yvcK mutant is transformed by the pDG148-MreB plasmid, normal growth and normal physiology are rescued when bacteria were grown on the gluconate medium (Fig. 5B). We also tested if this bypass was specific of MreB or could be obtained with another MreB-like protein. After verifying that the constructs were functional in 2505 (Carballido-López et al., 2006) and 2535 strains (Jones et al., 2001), we observed that neither the insertion of pDG148-Mbl plasmid nor of the pDG148-MreBH plasmid rescued a normal growth and a normal shape of a yvcK mutant when bacteria were grown on gluconeogenic carbon sources (Fig. 5B).

Figure 5.

Observation of the cell shape of mreB mutants (A) and yvcK mutants (B) in various genetic backgrounds by DIC microscopy. A. Observation of strains 168 (wild-type), 3725 (ΔmreB), SG103 (ΔmreB-amy::Pspac mreB) and SG102 (ΔmreB-amy::Pspac yvcK) during growth in PAB medium in the presence of 1 mM IPTG. B. Observation of strains SG63 (ΔyvcK), SG100 (ΔyvcK, pDG148), SG101 (ΔyvcK, pDG148-YvcK), SG111 (ΔyvcK, pDG148-MreB), SG113 (ΔyvcK, pDG148-Mbl) and SG115 (ΔyvcK, pDG148-MreBH) during growth on CE-gluconate medium in the presence of 1 mM IPTG. The scale bars represent 2 µm.

The YvcK overproduction compensates the absence of MreB by relocalizing PBP1

To understand how the two proteins YvcK and MreB can rescue each other, we investigated the effect of a ponA deletion. Indeed, J. Errington and coworkers (Kawai et al., 2009b) have carried out a transposon mutagenesis to obtain suppressor mutations that rescue the growth of an mreB mutant. Transposon insertions were found in the ponA gene that encodes the penicillin-binding protein PBP1. This kind of approach has already been carried out for yvcK (Görke et al., 2005) but no transposon insertion was found in ponA. Furthermore, the localization of PBP1 was shown to be dependent on MreB (Kawai et al., 2009b). In addition, the Mg2+ requirement of an mreB mutant strain was correlated to a PBP1 delocalization; a deletion of ponA, the gene encoding PBP1, restored the viability of this mutant. We also observed previously an Mg2+ dependency of the yvcK mutants grown in CE-gluconate liquid medium (Görke et al., 2005). We thus tested the effect of a ponA deletion in a yvcK mutant and analysed the growth of bacteria in gluconeogenic conditions in the absence of Mg2+. As described previously (Görke et al., 2005), we observed that the strain deleted for yvcK can not grow either on solid (Fig. 6A), or in liquid CE-gluconate medium (Fig. 6B). As already observed for the mreB mutant (Kawai et al., 2009b), the deletion of ponA prevented the bulging and the lysis phenotypes and thus restored the viability of the yvcK mutant in these gluconeogenic growth conditions (Fig. 6B).

Figure 6.

The abnormal morphology of the yvcK mutant cells is associated with a delocalization of PBP1. A and B. Effect of a ponA deletion on the growth of a strain deleted for yvcK. Strains 168, SG63 (ΔyvcK), SG189 (ΔyvcK, ΔponA), PS2062 (ΔponA) were grown overnight in LB. After centrifugation, they were plated on CE-glucose or CE-gluconate then incubated overnight (A) or grown in liquid CE-gluconate medium (B) at 37°C. The cell shape was observed by DIC microscopy during the growth in liquid medium and the images were taken 4 h after inoculation as indicated by an asterisk in the corresponding growth curves (168, diamonds, SG63: squares, PS2062: triangles and SG189: circles). The scale bars represent 2 µm. C and D. Localization of GFP-PBP1 in various genetic backgrounds. Strains YK706 (WT) and SG199 (ΔyvcK) and SG218 (ΔyvcK, pDG148-YvcK) were grown on CE-gluconate medium and, for SG199, on CE-glucose medium(C). Strains YK706 (WT), SG204 (ΔmreB) and SG217 (ΔmreB, pDG148-YvcK) were grown in LB and, for SG204, in LB in the presence of 25 mM MgSO4 (D) then analysed by fluorescence microscopy. As expected, strain SG199 (ΔyvcK) grown on CE-glucose medium and strain SG204 (ΔmreB) grown in LB in the presence of 25 mM MgSO4 show a normal morphology and a normal localization of GFP-PBP1. The YvcK protein was expressed from the promoter Pspac in the presence of 100 µM IPTG whereas the GFP-PBP1 was expressed from the promoter Pxyl in the presence of 0.25% xylose; in these conditions, the yvcK mutant has an observable phenotype. The scale bars represent 2 µm.

We then investigated the localization of a GFP-PBP1 in a yvcK mutant strain. Growth in CE-gluconate liquid medium showed that the bulging of the yvcK mutant cells was associated with a delocalization of GFP-PBP1. However, growth in the presence of 25 mM of Mg2+ (data not shown) or growth on CE-glucose liquid medium showed that, as expected, the shape of the yvcK mutant cells was normal and the localization of GFP-PBP1 was normal too (Fig. 6C). We also analysed the effect of an overproduction of YvcK on the localization of PBP1, in the ΔmreB and in the ΔyvcK backgrounds (Fig. 6C and D). For both strains, the overproduction of YvcK restores a normal pattern of localization of GFP-PBP1. This result shows that the recovery of growth and of cell morphology by the YvcK overproduction in a ΔyvcK or in a ΔmreB background is correlated with a restoration of PBP1 localization.

Discussion

In this paper, we showed that, in B. subtilis, the YvcK protein necessary for growth on gluconeogenic carbon sources is organized as a helical structure within the cell. Upon overproduction, it is able to restore a normal rod-shaped morphology in an mreB mutant by rescuing the PBP1 localization.

A helical organization of YvcK in the cell, autonomous of MreB

Numerous proteins that encircle and run along the length of the rod-shaped cell have recently been reported in the literature and this helical organization seems to be relatively widespread. However, the molecular mechanisms involved in these cellular localizations remain unknown or unclear and the reasons why proteins harbour this organization are enigmatic. Among them, are the components of the actin-like proteins of the MreB family (Carballido-López, 2006; Cabeen and Jacobs-Wagner, 2007). They form filamentous structures on their own. Others like LytE in B. subtilis, a cell wall hydrolase (Carballido-López et al., 2006) and PBP1 (Scheffers et al., 2004; Claessen et al., 2008) are involved in cell wall organization and hence in cell shape; their cellular localization being driven by the cytoskeleton. Then, some proteins like DnaA (Boeneman et al., 2009), EF-Tu (Defeu Soufo et al., 2010), SetB (Espeli et al., 2003) or RNaseE and some degradosome components in E. coli (Taghbalout and Rothfield, 2007) do not have any obvious link with cell shape. Furthermore, two of them, DnaA (Boeneman et al., 2009) and RNaseE (Taghbalout and Rothfield, 2007) from E. coli, do not colocalize with MreB filament and it was proposed that they adopt their helical structures without being directly associated with the cytoskeleton formed by the filament of MreB. The mechanism responsible for the intracellular organization of YvcK is unknown. We did not obtain any clue that it forms a filament on its own. We showed that YvcK organization is not driven by the MreB cytoskeleton, and we can wonder whether this particular localization plays a role in the rescue of mreB mutant, when YvcK is overexpressed, and in the PBP1 localization.

Functional similarities between YvcK and MreB

The MreB and YvcK proteins do not have any similarity of sequence and of structure. MreB is the main protein of bacterial cytoskeleton (Jones et al., 2001; Carballido-López et al., 2006) whereas YvcK is a protein of unknown function required for growth on gluconeogenic carbon sources and it belongs to the family of transferases (Görke et al., 2005; Forouhar et al., 2008). However, in addition to their similar cellular localization, we show here some functional similarities between MreB and YvcK. In some conditions, the absence of one of these proteins is correlated with an abnormal bulging of the cell that can be rescued by high concentrations of magnesium (Formstone and Errington, 2005; Görke et al., 2005). The effect of magnesium is enigmatic but it has been proposed that it might affect the structure of the cell wall (Formstone and Errington, 2005). If it is the case, we can hypothesize that, in a yvcK mutant cell grown on gluconeogenic carbon sources, the structure of cell wall may be affected. The cytoskeleton is implicated in the control of the synthesis of peptidoglycan, the major component of the cell wall. Indeed, the final stages of the synthesis of peptidoglycan require the PBP proteins on one hand and on the other it has been shown that MreB directly recruits the major PBP, PBP1, to the lateral cell wall (Kawai et al., 2009b). In this study, we observed that, when yvcK mutant cells were grown on gluconeogenic carbon sources, the abnormal shape of the cell is correlated to an abnormal localization of PBP1. In these conditions, the absence of YvcK has the same effect as MreB inactivation and seems to stop the dynamic localization of PBP1 in the cylindrical part of the cell and its accumulation at the cell poles.

PBP1, YvcK and MreB relationship

The dynamic localization of PBP1 seems to be a complex process as it is highly regulated to achieve cell wall synthesis at the good time and place during the cell cycle. A PBP1 abnormal localization is lethal for the bacterial cell. It has been proposed to disturb PBP1 activity, as the effect was eliminated when ponA was deleted [our results and (Claessen et al., 2008; Kawai et al., 2009b)]. PBP1 localization is tightly associated to the cell cycle and is dependent on various cell division proteins (Scheffers and Errington, 2004). Furthermore, two proteins involved in the elongation–division cycle EzrA, and GpsB, also have an unusual bulging phenotype associated with disturbed localization of PBP1 and both have been found to interact with PBP1 (Claessen et al., 2008). For the moment, we do not know if the rescue of PBP1 localization in an mreB mutant cell by the YvcK overproduction is a direct or indirect effect. No interaction has been detected between YvcK and MreB but we can speculate on a possible transient complex between MreB, YvcK and PBP1 or on a putative interaction between YvcK and PBP1 and we will investigate these potential dynamic interactions in the future. Using the database and web resource STRING [http://string-db.org/ (Jensen et al., 2009)], we showed that there is no relationship between the presence of YvcK and the bacterial shape and no co-occurrence among PBP1, MreB and YvcK. Indeed, MreB is found in nearly all rod-shaped bacteria (Dye and Shapiro, 2007) whereas YvcK is found in rod-shaped bacteria but also in spherical bacteria like streptococci or staphylococci in which it is essential (Chaudhuri et al., 2009).

In conclusion, the elucidation of the role of the YvcK protein, not only in rod-shaped bacteria but also in spherical bacteria lacking MreB, will probably give clues on the molecular mechanisms involved in two key processes: the carbon metabolism on one hand and the maintenance of bacterial morphology on the other.

Experimental procedures

Plasmid and strain constructions

Standard procedures for molecular cloning and cell transformation of B. subtilis or E. coli were used. All the strains, primers and plasmids used in this study are listed in Tables S1, S2 and S3 respectively.

For the generation of fluorescent fusion proteins, the yvcK gene was amplified by PCR using the primers 5′YvcKCterGFP and 3′YvcKCterGFP and inserted between KpnI and HindIII sites in pSG1154 (Lewis and Marston, 1999). The B. subtilis strains SG63 (yvcK mutant) and 3725 (mreB mutant) were then transformed with pSG1154-yvcKGFP and protein expression was induced for 4 h with 1% xylose (w/v). The yvcK gene was also amplified using the primers 5′YvcK-deb and 3′YvcK-fin and inserted between KpnI and XmaIII sites in the pMutin-GFP+ (Kaltwasser et al., 2002). The resulting plasmid was used to transform B. subtilis wild-type strain 168, 3725, 2535 (mreBH mutant) (Carballido-López et al., 2006) and 2505 (mbl mutant) (Jones et al., 2001). In all these resulting strains, the expression of the YvcK fused to GFP is driven by the PyvcI of the yvcIJKLcrhyvcN operon whereas the native yvcK gene and the downstream genes are controlled by the Pspac promoter.

For the complementation of the SG63 strain, DNA fragments encoding MreB, Mbl and MreBH were generated by PCR using specific primers and inserted into the pDG148-Stu plasmid (Joseph et al., 2001). Strain SG63 was transformed with pDG148, pDG148-MreB, pDG148-Mbl or pDG148-MreBH plasmid. For the rescue of the strain 3725, the two genes yvcK and mreB under the control of the Pspac promoter from pDG148-YvcK and pDG148-MreB, respectively, were digested by EcoRI and BamHI enzymes and inserted into these restriction sites of pAC5 plasmid. The two pAC5 resulting plasmids containing either yvcK or mreB gene were used to transform the 3725 strain by a double crossing-over event and selected for chloramphenicol resistance. The expression of all the genes under Pspac control (at the amyE locus or on the pDG plasmid) was induced with 100 µM IPTG.

Growth tests

Bacillus subtilis strains were grown in LB, DSM, CE-minimal medium and on PAB medium (Difco antibiotic Medium 3) as previously described (Formstone and Errington, 2005; Görke et al., 2005). For the overnight pre-culture of ΔmreB mutant strains, the PAB medium was supplemented with 0.3 M sucrose and 25 mM MgSO4.

q-RT-PCR

1 µg of RNA was reverse-transcribed using the standard protocol of Superscript II Reverse Transcriptase (Invitrogen) and 100 ng of random primers. The resulting cDNA was diluted 16-fold and 5 µl were used for the q-RT-PCR reaction. This step was performed on a Mastercycler ep Realplex (Eppendorf) using the SYBR Premix Ex Taq (Perfect Real Time) PCR Kit (Takara Bio Group, Japan) in a final volume of 20 µl according to the manufacturer's instructions. Melting curves were analysed to control for specificity of the PCR reactions. Data from three independent experiments were analysed and normalized with the software supplied with the mastercycler. The relative units were calculated from a standard curve plotting four different dilutions (1/80, 1/400, 1/2000 and 1/10 000) against the PCR cycle number (Ct) at which the measured fluorescence intensity reached the threshold, specified so that it is significantly above the noise band of the baseline (10-fold above the standard deviation value).

Western blot

The cells were grown at 37°C in 20 ml of LB medium to OD600 = 1 then centrifuged for 10 min at 8000 r.p.m. at 4°C. Cell pellets were resuspended in 1/10 volume of lysis buffer containing 10 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% NP40, 1 mM PMSF, 25 U ml−1 benzonase and 10 mg ml−1 lyzozyme. Extracts were incubated for 10 min on ice then heated at 100°C for 10 min. Samples were run on a 12.5% SDS-PAGE and transferred to nitrocellulose membrane by electroblotting. The membrane was blocked with PBS solution containing 5% milk powder (w/v), for 1 h at room temperature with shaking. Then, the membrane was incubated either with anti-GFP antibody (Roche) used at 1/1000 dilution and the secondary antibody, a fluorescent Alexa Fluor 680 goat anti-mouse IgG (Molecular Probes) used at 1/5000 or with anti-YvcK antibody used at 1/2500 dilution and the secondary antibody, a fluorescent Alexa Fluor 647 goat anti-rabbit IgG (Molecular Probes) used at 1/5000. After three washes, the Blot membrane was scanned at 700 nm by the Odyssey Infrared imaging System (LiCor Biosciences).

Immunofluorescence

3 ml of LB was inoculated with one colony and incubated 2 h at 37°C. 1 ml of cells was then added to 10 ml of ice-cold 80% methanol and incubated 1 h. All the incubations were carried out at room temperature except when stated otherwise. 200 µl of freshly prepared 16% formaldehyde was added and incubated 5 min. Cells were centrifuged and washed once with 1 ml of ice-cold 80% methanol then resuspended in PBS-0.05% Tween to an OD600 of 0.5. 10 µl of cells was dropped on Poly-l-lysine coated slides and incubated for 5 min. Cells were permeabilized for 10 min at 37°C in 20 mM Sodium Phosphate buffer pH 6.2, 50 mM Sucrose, 500 µg ml−1 lysozyme. Cells were then washed with PBST and saturated with 40 µl of PBST- 5% BSA during 5 min. Primary anti-YvcK antibodies (dilution 1/100) were incubated 1 h. Cells were washed, saturated with 40 µl of PBST- 5% BSA during 5 min then incubated with the secondary antibody (1/400, anti-Rabbit Alexa Fluor) for 1 h in the dark. Cells were observed in differential interference contrast or Phase Contrast with an upright Nikon Eclipse E800 microscope at 100× NA 1.4 objective (498/516 nm). Images were recorded with Nikon digital camera DXM1200.

FM4-64 staining

1 µl of cells was placed on slides and mixed with 0.4 µl of FM4-64 (Molecular Probe) (10 mg ml−1 in DMSO) before microscopic observation. As previously mentioned, the cells were observed in Phase Contrast using an upright Nikon Eclipse E800 microscope with a 100× NA 1.4 objective (515/640 nm).

Image acquisition and processing

Images and Z-stack of 21 images were captured with step distance ranging from 0.15 µm using a Zeiss Axiovert 200 M microscope connected to a Hamamatsu ORCA ER camera. Image restoration was obtained by deconvolution using the Huygens Essentiel software (SVI). Three-dimensional visualization was performed with Imaris software package (Bitplane). Measures and analyses were performed with ImageJ (Rasband, W.S., ImageJ, U. S. National Institutes of Health, Bethesda, Maryland, USA, http://rsb.info.nih.gov/ij/, 1997–2008). 3D reconstruction was performed with ImageJ 3D Viewer (Schmid et al., 2010).

Colocalization

Cells were grown in LB medium in the presence of 1% xylose for 2 h (until OD600 between 0.3 and 0.6). The YvcKGFP and mRFPMreB fluorescent fusion proteins were visualized by fluorescence microscopy. Absence of crossed signal was checked using the strain SG163 expressing only YvcKGFP and with the strain SG223 expressing only mRFPMreB. In order to evaluate the pixel shift between different fluorescence filters, we have imaged multi 0.5 µm spectral fluorescent beads (TetraSpeck Fluorescent Microspheres Size Kit Molecular Probes, Invitrogen). Pixel shift between the green set filter and red set filter was measured at 2 pixels (0.2 µm) and images were corrected. This value was used as error margin for the coincidence fluorescent peaks scores. The profile plots obtained with ImageJ were processed with Origin (OriginLab, Northampton, MA 01060 USA) to subtract the baseline of the fluorescent signal.

Acknowledgements

This research was supported by the CNRS, the ANR and the Universities of Aix-Marseille. We thank T. Mignot, E. Cascales, T. Doan and B Ezraty, for valuable discussions, J. Domínguez-Escobar and R. Carballido-López for kind gift of unpublished plasmid RWB4, J. Errington, P. Graumann, P and B. Setlow for the kind gift of strains, S. Séror and V. Géli for supplying antibodies against PrkC and Alexafluor secondary antibodies respectively. We also particularly thank B. Khadaroo for critical reading of the manuscript and for the attenuation of the French touch.

Ancillary