Dynamic localization of penicillin-binding proteins during spore development in Bacillus subtilis

Bacteria display a high degree of subcellular organization, with proteins localizing to distinct sites in the cell at distinct times in the bacterial cell cycle. Examples include the assembly of cell-division proteins in a ring at the middle of the bacterial cell (Errington et al., 2003), the formation of dynamic helical filaments by the prokaryotic actin homologues MreB and ParM (Errington, 2003b; Gerdes et al., 2004), and the polar localization of, for example, chemoreceptors (Maddock & Shapiro, 1993). A process that involves both redistribution of cellular proteins and time- and compartment-dependent gene expression in bacteria is the formation of endospores. Endospores, which are very tough survival structures formed upon nutrient starvation, can be formed by several ancient lineages of bacteria. The development of spores has been best studied in the Gram-positive model organism Bacillus subtilis (for recent reviews see Errington, 2003a; Hilbert & Piggot, 2004; Piggot & Losick, 2001).

Recent work from various groups, focusing on the (re)distribution of several proteins during sporulation, has revealed intriguing examples of protein localization and distribution. At the start of sporulation, an asymmetric division septum is formed (a schematic overview of division and engulfment is shown in Fig. 1A). The switch from medial to asymmetric division is accomplished by an upshift in the expression of the key cell-division gene ftsZ, and a concomitant redistribution of FtsZ from a medial ring to a spiral that extends to both poles of the cell (Ben-Yehuda & Losick, 2002). The spiral pattern is also observed for FtsA and EzrA (Ben-Yehuda & Losick, 2002), which are other components of the division machinery that localize early in the division process (for a recent overview of B. subtilis cell division see Errington et al., 2003). One of the two possible asymmetric division sites is then committed to division by the action of SpoIIE (Barák & Youngman, 1996; Feucht et al., 1996), resulting in the localization of other cell-division proteins to the asymmetric division site, and subsequent division (see Errington, 2003a; Hilbert & Piggot, 2004).

(23K): Fig. 1. Schematic overview of sporulation and observed redistribution of GFPPBP2c and GFPPBP2d. The cell wall is shown in grey, membranes in black, and GFPPBP molecules as white ellipsoids. (A) 1, Vegetative cell with random distribution of GFPPBP. 2, 3, Asymmetric division. The number of PBPs ending up in what is going to be the prespore inner membrane is lower than the number of PBPs ending up in the mother cell membrane. 4, Septal thinning and recruitment of GFPPBP to the prespore septum. 5, Engulfment. GFPPBPs follow the engulfing mother cell membrane. 6, After membrane fusion, the outer prespore membrane has a high concentration of GFPPBP2c or 2d. (B) An alternative model. 2, GFPPBPs are recruited to the asymmetric division septum before division is complete. 3, This leads to an increased concentration of GFPPBPs on both sides of the asymmetric septum. 5 and 6, GFPPBPs follow the membrane during engulfment, and end up evenly distributed over the outer and inner prespore membranes.

Following asymmetric division, the asymmetric septum is degraded, and the mother cell membrane migrates around the prespore membrane in a phagocytosis-like process called engulfment. Abanes-De Mello et al. (2002) showed that membrane migration during engulfment depends on the activity of three proteins expressed in the mother cell, SpoIID, SpoIIM and SpoIIP, and that these three proteins are associated with the leading edge of the engulfing membrane. SpoIID is a cell wall hydrolase which is required not only for septal thinning at the onset of engulfment, but also for the completion of engulfment (Abanes-De Mello et al., 2002). The authors suggested an attractive model in which the cell wall acts as a scaffold along which the hydrolase tracks, dragging the mother cell membrane and other engulfment proteins along (Abanes-De Mello et al., 2002). The DNA translocase SpoIIIE also migrates at the leading edge of the engulfing membrane, and is involved in membrane fusion when the engulfing membranes meet (Sharp & Pogliano, 1999), so that the prespore is now surrounded by an inner membrane and a mother-cell-derived outer membrane. Work from the same group has shown that prespore-expressed membrane proteins also localize to the septum, track along with the engulfing mother cell membrane, and then spread out along the inner prespore membrane, sometimes to assemble into regular structures (Rubio & Pogliano, 2004). The tracking of these prespore-expressed proteins is dependent on an unidentified mother-cell-expressed protein, suggesting that the proteins interact with a localized partner protein made in the mother cell or with parts of the septal peptidoglycan (PG) (Rubio & Pogliano, 2004).

Dynamic localization of proteins from the mother cell to the prespore has also been shown for the morphogenetic protein SpoIVA, which is involved in assembly of the spore cortex and coat (Lewis & Errington, 1996), and for SpoIVFB, a polytopic membrane protein involved in the activation of a transcription factor (Rudner et al., 2002). In the latter case, Rudner et al. (2002) showed that SpoIVFB is inserted into the cytoplasmic membrane in a dispersed fashion, after which the protein diffuses to, and is captured in, the outer prespore membrane (the diffusion-and-capture model).

This work is concerned with the localization of penicillin-binding proteins (PBPs), which are proteins involved in the synthesis of the cell wall PG during vegetative growth, cell division and sporulation (for recent reviews on cell wall synthesis during growth and sporulation, see Foster & Popham, 2001; Popham, 2002) (Table 1). Early work on expression profiles of PBPs during vegetative growth and sporulation indicated roles for PBPs 2B, 3, 4* and 5* in sporulation (Sowell & Buchanan, 1983; Todd et al., 1983). Studies on mutant strains, transcriptional profiling and localization have now identified a number of PBPs as playing (putative) roles during sporulation. The class A bifunctional transglycosylase/transpeptidase PBP1 is part of the division machinery that operates during asymmetric division, and is required for efficient sporulation (Scheffers & Errington, 2004). Two other class A PBPs, 2c and 2d, play a redundant role in spore PG synthesis. A strain in which the genes for these PBPs 2c and 2d (pbpF and pbpG, respectively) are inactivated is incapable of completing sporulation, and shows defects in spore PG synthesis (McPherson et al., 2001). The class B transpeptidase PBP2b, the only essential PBP in B. subtilis, is required for the asymmetric cell division, and localizes to the asymmetric septum (Daniel et al., 2000). Another class B PBP, SpoVD, is essential for spore formation, and is required for the synthesis of cortical PG (Daniel et al., 1994), whereas PBP4b does not seem to have an effect on spore PG, but is expressed under the control of the mother-cell-specific σ^E factor (Eichenberger et al., 2003; Wei et al., 2004). Finally, two low-molecular-mass PBPs are involved in spore PG synthesis. These are the carboxypeptidases 5* and DacF, which have partially redundant roles in regulating the degree of cross-linking of the spore PG, and a double mutant for both proteins has decreased spore heat resistance (Popham et al., 1995, 1999).

Table 1. Summary of B. subtilis PBPs implicated in sporulation

In this study, I made use of a recently constructed collection (Scheffers et al., 2004) of fusions of green fluorescent protein (GFP) to PBPs of B. subtilis to study the localization of these proteins during sporulation. I found that PBP2c and 2d localize to the prespore division septum, and follow the leading edge of the mother cell membrane during engulfment. Also, a localization pattern was identified for the as yet uncharacterized PbpX that resembles the redistribution of FtsZ upon the start of sporulation. A pbpX knockout strain has no obvious phenotype. These results show that for some PBPs, sporulation induces a dynamic redistribution of their localization patterns. General methods.
The strains and plasmids used in this study are listed in Table 2. B. subtilis cells were made competent for transformation with DNA either by the method of Kunst & Rapoport (1995), or by the method of Anagnostopoulos & Spizizen (1961) as modified by Jenkinson (1983). DNA manipulations and Escherichia coli DH5α transformations were carried out using standard methods (Sambrook et al., 1989). Solid medium used for growing B. subtilis was nutrient agar (Oxoid), and liquid medium was either casein hydrolysate (CH) medium (Sterlini & Mandelstam, 1969) or S medium (Sharpe et al., 1998) supplemented with 1 % (v/v) CH (S+), with antibiotics added as required. Chloramphenicol was used at 5 µg ml¹, spectinomycin at 50 µg ml¹, erythromycin at 0·5 µg ml¹, lincomycin at 12·5 µg ml¹, and kanamycin at 5 µg ml¹. Media used for growing E. coli were 2x TY (tryptone yeast extract medium; Sambrook et al., 1989) and nutrient agar supplemented with ampicillin (100 µg ml¹) as required.

Table 2. Bacterial strains and plasmids

Construction of GFP fusions.
This was performed as described previously (Scheffers et al., 2004). Approximately one-third of each of the promoter-proximal parts of the pbpG, spoVD, dacB, dacF and pbpE genes was amplified using PCR, and cloned into pSG4902 (Wu & Errington, 2003). All primers and restriction endonucleases used are listed in Table 3. Transformation of the resulting plasmids (Table 2) into B. subtilis, with selection for chloramphenicol resistance resulted in several strains, each of which carried a gfp fusion to a gene of interest at the chromosomal locus as the only copy of the gene of interest, and under the control of the P_xyl promoter. Correct integration at the chromosomal locus was confirmed by PCR.

Table 3. Primers Restriction sites are underlined.

The vectors pMDS13 and pMDS14, which contain gfp under the control of the prespore-specific P_spoIIQ promoter and the mother-cell-specific P_spoIID promoter, respectively (Sharp & Pogliano, 2002), were used for the construction of N-terminal GFP fusions to PBPs that are expressed during sporulation. Full-length pbpF, pbpG and pbpX were amplified by PCR with primer sets DJS163DJS164, DJS165DJS166 and DJS175DJS176, respectively. The PCR products were digested with EagI, and ligated into EagI-digested pMDS13 or pMDS14. The correct orientation of the inserts was confirmed by PCR and sequencing. The resulting plasmids (Table 2) were transformed into strain 168, with selection for chloramphenicol resistance, and integration at the amyE locus was confirmed by screening the transformants for failure to degrade starch.

Construction of pbpX mutant strains.
Using primer pairs DJS149DJS150 and DJS151DJS152 (Table 3), ∼1·6 kb PCR fragments were generated containing the first 292 bp of pbpX plus upstream sequences, and the last 257 bp of pbpX plus downstream sequences. These fragments were cut with BamHI and EcoRI, respectively, and ligated to a BamHIEcoRI-digested PCR product containing a neo cassette, which was generated using primers km3 and km4, and plasmid pKM1 as a template. The ligation product was subjected to another PCR reaction using primers DJS149 and DJS152, and the resulting PCR product was transformed into strain 168, with selection for kanamycin resistance, generating strain 3906, which contains a deletion of pbpX codons 98305 (out of 391 codons). Correct integration of the ligation product into the chromosome was confirmed by PCR and sequencing.

Using primer pair DJS128DJS129, a fragment of pbpX (bp 161441) was amplified by PCR. The fragment was cut with HindIII and BamHI, and ligated into HindIIIBamHI-digested pMUTIN4, a vector that allows inactivation of the target gene as well as the monitoring of its expression through a transcriptional lacZ fusion (Vagner et al., 1998), generating pSG5313. pSG5313 was transformed into strain 168 to give strain 3905. Correct integration of the plasmid into the chromosome was confirmed by PCR.

Sporulation methods.
Sporulation was induced by growth to OD₆₀₀ ∼0·8 in CH, followed by resuspension in a starvation medium (SM; Partridge & Errington, 1993; Sterlini & Mandelstam, 1969). Cell pellets were washed with SM prior to resuspension to remove xylose, unless stated otherwise. Time zero (T₀) was defined as the point at which the cells were resuspended in the starvation medium.

β-Galactosidase activity was assayed as described by Errington (1986). One unit of β-galactosidase catalyses the production of 1 nmol 4-methylumbelliferone min¹. Alkaline phosphatase activity was measured as described by Errington & Mandelstam (1983) and Glenn & Mandelstam (1971). Sporulation efficiency was tested by determining the number of heat-resistant spores formed in the cultures at 10 h (T₁₀) or 25 h (T₂₅).

Microscopy.
Microscopy was performed essentially as described previously (Scheffers et al., 2004). Image acquisition was done as described by Lewis & Errington (1997), using Metamorph version 6.0 software (Universal Imaging Corporation). DNA was stained with Hoechst 33342 (1 µg ml¹; Molecular Probes). Membranes were stained with FM95.5 (4 µg ml¹; Molecular Probes). Images from a single focal plane were deconvolved using the No Neighbours algorithm from the Metamorph software package. Overlays of micrographs were assembled using Metamorph, before exporting the images to Adobe Photoshop version 6.0.

Construction of GFPPBP2d and GFPPBP4*, and localization during vegetative growth
I set out to study the localization of PBPs during sporulation in B. subtilis. Previously, GFP fusions were constructed to 11 PBPs expressed during vegetative growth (Scheffers et al., 2004). To include additional PBPs specific for sporulation, a similar strategy was attempted for the pbpG, spoVD, dacB, dacF and pbpE genes (encoding PBP2d, SpoVD, PBP5*, DacF and PBP4*, respectively, none of which are essential). This approach was successful for fusions to pbpG and pbpE. A strain containing a genetic fusion to dacB was obtained, but when it was studied under the microscope, the GFPPBP5* fluorescence signal was not detected, so this strain was not studied in more detail. Unfortunately, several attempts to obtain GFP fusions to spoVD and dacF using this strategy also failed.

When expressed during vegetative growth, GFPPBP2d localized in a dispersed fashion along the membrane, whereas GFPPBP4* localized in a punctate pattern (Fig. 2). These patterns were similar to the dispersed or punctate localization patterns observed with most PBPs expressed during vegetative growth (Scheffers et al., 2004).

(79K): Fig. 2. Localization of GFPPBP2d and GFPPBP4* during vegetative growth. Fluorescence micrographs of cells expressing GFPPBP2d (A) and GFPPBP4* (B). Illumination for fluorescence was 1 s. Bar, 5 µm.

PBP2c, PBP2d and PbpX localize to the prespore
During sporulation, the cell-division-specific PBPs PBP2b and PBP1 localize to the asymmetric septum formed during sporulation (Daniel et al., 2000; Scheffers & Errington, 2004). To determine the localization of other PBPs during sporulation, the GFPPBP fusion strains were grown in the presence of 0·5 % xylose to allow expression of the fusion protein. At OD₆₀₀ ∼0·8, cells were washed with sporulation salts to remove xylose, since the presence of xylose delays the onset of sporulation. Subsequently, cells were induced to sporulate using the resuspension method (see Methods), and the localization of the GFPPBP fusions was followed. Using this method, PBP2c and PbpX were seen to accumulate specifically at the prespore (Fig. 3A, F; Table 1). PBP2c arrives at the prespore septum, and then follows the mother cell membrane during engulfment. After completion of engulfment, GFPPBP2c is localized around the prespore. Whether GFPPBP2c is predominantly present in the inner or outer prespore membrane cannot be distinguished (see Discussion). PbpX seemed to localize earlier during sporulation, being detected at both poles before cells committed to form the asymmetric septum (Fig. 4C), and subsequently followed the mother cell membrane during engulfment (Fig. 3F). None of the other PBPs tested, PBP4, PBP2a, PBP3, PbpH, PBP4b, PBP5, PBP4a and PBP4*, showed a similar localization pattern (Fig. 3, and data not shown). Also, GFPPBP2d fluorescence was not detected above background levels at 2 h (T₂) and 3 h (T₃), even though PBP2d plays a role in sporulation (Fig. 3B; McPherson et al., 2001). However, when cells were resuspended in the presence of xylose, PBP2d was readily detected at the prespore in a pattern similar to that of PBP2c (Fig. 3C). The localization patterns of all of the other PBPs studied were not affected by the addition of xylose to the sporulation medium. Western blotting of cultures resuspended in the absence of xylose confirmed that the levels of GFPPBP2c, GFPPbpX, GFPPBP4b and GFPPBP4* were not altered significantly 2 h into sporulation, but that GFPPBP2d could no longer be detected, indicating that it is subject to proteolysis (data not shown). Also, when resuspended in the absence of xylose, strains expressing GFPPBP2c, GFPPBP2d, GFPPBP4b and GFPPbpX had similar spore counts compared to wild-type, showing that the GFPPBPs do not interfere with sporulation (data not shown).

(82K): Fig. 3. Localization of GFPPBP fusions during sporulation. Fluorescence micrographs of cells expressing GFPPBP2c (A), GFPPBP2d (B, C), GFPPBP4b (D), GFPPBP4* (E) and GFPPbpX (F) are shown as overlays of the GFP image (green) and DNA staining (red). Images were taken directly after resuspension of the cells in sporulation medium (T0), and 2 h (T2) and 3 h (T3) after resuspension. Cells were resuspended in sporulation medium without xylose, except for cells expressing GFPPBP2d, which were resuspended in the absence (B) and presence (C) of xylose. Bar, 5 µm. (A; T2), (B; T0 and T3) and (D; T0 and T2) are composite images.

(66K): Fig. 4. GFPPbpX appears to form spirals, and localizes to both polar division sites early in sporulation. GFP fluorescence of GFPPbpX is dispersed along the membrane (A) and sometimes appears to spiral out (B) from the midcell division septum to the poles upon sporulation. Cells were photographed 80 min after resuspension in sporulation medium. From left to right, a fluorescence micrograph, a deconvolved image of the micrograph, and an overlay of the deconvolved GFP image (green) and DNA staining (red). (C) GFPPbpX is present at both polar division sites (cells indicated with arrows). Fluorescence micrographs and an overlay of the GFP image (green) and DNA staining (red) are shown for cells 60 min (i) and 80 min (ii) after resuspension in sporulation medium. Bars (notice size difference in B), 5 µm.

GFPPbpX appears to spiral out to both asymmetric sporulation division sites
GFPPbpX localizes to the division septum during vegetative growth (Scheffers et al., 2004). After resuspending GFPPbpX-expressing cells in sporulation medium, a quick change in its localization pattern was detected. Typically, between T₁ and T₂, before a complete asymmetric division septum was formed, the PbpX signal became dispersed along the membrane, with differences in signal intensity along the membrane (Fig. 4A). In some cases, these signals looked like spirals (Fig. 4B) similar to the spirals observed for FtsZ, FtsA and EzrA in sporulating cells (Ben-Yehuda & Losick, 2002). The spiral pattern was more apparent when the image was deconvolved. However, these spiral structures were only observed in rare cases. Interestingly, GFPPbpX was seen to appear at both poles, often with a slight difference in intensity on either pole (Fig. 4C). Polar localization of the cell-division-specific PBPs PBP1 and PBP2b was seen only in cells that appeared to already have committed one of the potential polar division sites to become the asymmetric septum (Daniel et al., 2000; Scheffers & Errington, 2004). Thus, GFPPbpX localizes to both poles, similar to cell-division proteins like FtsZ, and the sporulation-specific SpoIIE (Levin & Losick, 1996; Levin et al., 1997).

A pbpX knockout strain has no distinguishable phenotype
Given the interesting localization pattern observed with GFPPbpX, I decided to study pbpX in more detail. pbpX was identified as a gene encoding an endopeptidase based upon sequence similarity (Foster & Popham, 2001), and it has recently been described as part of the σ^X regulon (Cao & Helmann, 2004). pbpX was inactivated in two ways: by replacing 624 internal bases from the gene with a neo resistance marker, and by use of the pMUTIN-4 vector, which generates a lacZ transcriptional fusion to pbpX allowing the determination of the pbpX expression pattern (Methods). Both knockout strains grew at an identical rate, and with similar spore counts compared to wild-type B. subtilis (Table 4). Correct formation of the asymmetric sporulation septum was followed by expression of σ^E-dependent genes, since activation of this sigma factor is dependent on septation (Piggot & Losick, 2001). The σ^E-dependent synthesis of alkaline phosphatase was measured for both strains, and was found to be indistinguishable from wild-type (result not shown), showing that deletion of pbpX has no effect on septation during sporulation. The appearance of the ΔpbpX strain was indistinguishable from that of the wild-type (Table 4). The transcriptional activation of pbpX followed a pattern typical for weak expression during vegetative growth, with no induction upon sporulation (result not shown).

Table 4. Properties of pbpX knockout strains

pbpX is part of the σ^X regulon, which is thought to modulate aspects of cell envelope metabolism, possibly through regulation of cell-surface modification (Cao & Helmann, 2004; Huang & Helmann, 1998). A sigX-null mutant is impaired in its ability to survive at high temperature (Huang et al., 1997). To determine whether this effect is mediated by pbpX, the survival of the wild-type and the ΔpbpX strain was tested, after transferring exponentially growing cells to 54 °C for 30 min. Survival of the ΔpbpX strain was similar to that of the wild-type (Table 4), showing that pbpX is not responsible for the decrease of heat resistance observed in a sigX null strain.

It is concluded that pbpX is a non-essential gene in B. subtilis, with no obvious phenotype during vegetative growth or sporulation.

PBP2c and PBP2d localize independently of each other
PBP2c and PBP2d play redundant roles during sporulation, and the presence of at least one of these PBPs is required for the synthesis of the spore germ cell wall (McPherson et al., 2001). The finding that both GFPPBP2c and GFPPBP2d localize to the prespore was in line with this observation. The localization of both GFP fusion proteins in the absence of the other PBP was studied. As shown in Fig. 5, each of the proteins GFPPBP2c and GFPPBP2d was able to localize in the absence of the other protein. Again, for GFPPBP2d, xylose had to be present in the sporulation medium for continued synthesis. This shows that these proteins are not dependent on each other for correct localization, as might be expected from the fact that their functions appear to be redundant during sporulation (McPherson et al., 2001).

(86K): Fig. 5. PBP2c and PBP2d localize to the prespore independently of each other. Fluorescence micrographs of cells from strain 3912 containing an insertional inactivation of pbpG, and expressing GFPPBP2c (A), and from strain 3915 containing a pbpF knockout, and expressing GFPPBP2d (B). Images were taken 2 h after resuspending the cells for sporulation. Xylose was added to the sporulation medium for strain 3915. Bar, 5 µm. (A) is a composite image.

The sporulation efficiency of a ΔpbpF strain with gfppbpG under the control of P_xyl (strain 3915), when grown and sporulated without xylose, was less than 0·3 % of that for the wild-type and the ΔpbpF strain (Table 5). A similar result was obtained with a ΔpbpG strain with gfppbpF under control of P_xyl (strain 3912). This was expected, since either PBP2c or PBP2d is required for efficient sporulation (McPherson et al., 2001). Xylose fully restored sporulation for strains 3912 and 3915, showing that the presence of GFPPBP2c or GFPPBP2d in the absence of PBP2d or PBP2c, respectively, is sufficient for sporulation (Table 5). This shows that GFPPBP2c and GFPPBP2d are fully functional. The presence of xylose in both the growth and the sporulation medium increased the spore count in the single-knockout strains (PS1869 and BFA1208). This was because the extra carbon source in the sporulation medium allowed the cells to continue growing, causing both a delay of the initiation of sporulation and an increase in the number of cells at the initiation of sporulation.

Table 5. GFPPBP2c and GFPPBP2d are functional

Localization of GFPPBPs under control of sporulation-specific promoters
Localization of GFPPBP2c, GFPPBP2d and GFPPbpX to the prespore was observed with protein expressed in the predivisional cell from the P_xyl promoter. I wanted to test whether similar patterns could be observed when the proteins were expressed only during sporulation, with expression switched on in either the mother cell or the prespore compartment. Compartment-specific expression makes it possible to distinguish targeting to the outer prespore membrane from targeting to the inner prespore membrane. Vectors were used that allow the expression of N-terminal GFP fusion proteins under control of the prespore-specific P_spoIIQ promoter or the mother-cell-specific P_spoIID promoter (kindly provided by Dr K. Pogliano; see Sharp & Pogliano, 2002). Expression from both promoters resulted in very strong fluorescence signals that required two-dimensional deconvolution to provide sufficient resolution. As shown in Fig. 6, expression of GFPPBP2c, GFPPBP2d and GFPPbpX under control of the prespore-specific promoter P_spoIIQ resulted in rather uniform labelling of the inner prespore membrane, although for GFPPBP2d, signal throughout the prespore was also observed, which is attributed to the instability of GFPPBP2d (Fig. 6B, T₄). This instability would also explain the uniform fluorescence in the cytoplasm when GFPPBP2d was expressed under control of the mother-cell-specific promoter P_spoIID (Fig. 6B). In contrast to this, GFPPBP2c, when expressed in the mother cell, localized to the outer prespore membrane in a pattern that shows that GFPPBP2c follows the mother cell membrane during engulfment. GFPPbpX, when expressed in the mother cell, was found uniformly distributed along the cytoplasmic membrane and the outer prespore membrane. The early redistribution of GFPPbpX, and localization to the asymmetric septa, is precluded by the fact that expression from P_spoIID occurs only after septation has occurred (see Hilbert & Piggot, 2004). These results confirm that the GFPPBP2c localization to the outer prespore membrane is specific, since in this experiment GFPPBP2c is expressed after closure of the sporulation septum, which separates the inner prespore membrane from the mother cell membrane that will develop into the outer prespore membrane.

(70K): Fig. 6. GFPPBP2c is targeted to the prespore outer membrane when expressed under control of PspoIID. Fluorescence micrographs of cells are shown as overlays of the deconvolved GFP image (green) and DNA staining (red). Cells were expressing GFPPBP2c (A), GFPPBP2d (B) and GFPPbpX (C) under the control of the prespore-specific promoter PspoIIQ, and under the control of the mother-cell-specific promoter PspoIID. Cells were processed for microscopy 3 h (T3) and 4 h (T4) after resuspension in sporulation medium. Bar, 5 µm.

The developmental switch from vegetative growth to sporulation in bacteria causes major changes in the subcellular organization of proteins, and gene expression. Work from several groups has shown that proteins involved in asymmetric cell division (Ben-Yehuda & Losick, 2002; Daniel et al., 2000; Scheffers & Errington, 2004), engulfment (Abanes-De Mello et al., 2002; Sharp & Pogliano, 1999), spore coat and cortex assembly (Price & Losick, 1999; Van Ooij et al., 2004), and mother-cell- or prespore-specific gene expression (Rubio & Pogliano, 2004; Rudner et al., 2002) localize to the prespore in a distinct and dynamic fashion. A striking feature of this localization is that, at least for membrane proteins that are targeted to the outer prespore membrane, a shift from a random membrane distribution to a highly organized localization occurs in accordance with a model in which these proteins diffuse laterally through the membrane to their site of action, where they are captured by (an) as yet unknown factor(s) (Rudner et al., 2002).

This paper is concerned with the localization of PG-synthesizing proteins during sporulation (summarized in Table 1). A collection of GFPPBP fusion proteins constructed earlier (Scheffers et al., 2004) was used, in addition to new fusions to include PBPs implicated in sporulation. Expression of the GFPPBP fusion proteins is driven by the P_xyl promoter, which is switched on during vegetative growth, but switched off during sporulation by removal of xylose from the sporulation medium. This procedure reveals the localization of membrane proteins, and their redistribution upon sporulation, with no newly synthesized GFPPBPs in the prespore. Out of 11 GFPPBP fusions tested, only three showed a change in localization patterns during sporulation, and two of these fusions are to PBPs known to be involved in sporulation. This strongly suggests that the pattern changes are not caused by artefacts.

A striking change in protein localization was observed with GFPPbpX, which started by localizing to the division septum at midcell, and then appeared to spiral out in a pattern resembling FtsZ (Ben-Yehuda & Losick, 2002), and was then found at both asymmetric potential division sites. Although GFPPbpX spirals were rare, it should be noted that spiralling seems to be less obvious for membrane proteins than for cytosolic proteins (compare EzrAGFP to FtsZGFP and FtsAGFP in Ben-Yehuda & Losick, 2002). GFPPbpX then appeared at both asymmetric septa, with unequal distribution of fluorescence intensity, as observed for SpoIIE by Wu et al. (1998). This observation adds credibility to the redistribution observed for GFPPbpX, since other cell-division proteins, notably PBPs (Daniel et al., 2000; Scheffers & Errington, 2004), only localize to the asymmetric septum when one of the potential division sites has been committed to division (see Hilbert & Piggot, 2004). The unequal distribution possibly reflects which asymmetric division site is chosen for septum formation. This question could be resolved by following GFPPbpX distribution in individual sporulating cells with time. Despite this striking fluorescence pattern, pbpX does not play a critical role in B. subtilis. pbpX was inactivated in two ways, but effects on cell growth, cell shape or sporulation efficiency were not detected. Thus, PbpX cannot be an essential component of the cell-division machinery, but it could be associated with (a) component(s) from the division machinery, which it follows from the midcell division site to both asymmetric cell-division sites. A possible role for the endopeptidase PbpX is the quick removal of PG that connects two cells after vegetative division, or thinning of the sporulation septum prior to engulfment.

GFPPBP2c and GFPPBP2d, which are randomly distributed along the membrane in vegetative cells (Scheffers et al., 2004), are redistributed during sporulation: both proteins localized to the sporulation septum, followed the engulfing membrane, and were finally concentrated in the prespore membrane. There are two ways in which this redistribution can be achieved (see Fig. 1). First, GFPPBP2c/d in the mother cell is recruited to the septum after septum closure, and then follows the mother cell membrane during engulfment. As a result, the majority of GFPPBP2c/d will be located in the outer prespore membrane (Fig. 1A). Alternatively, GFPPBP2c/d can be recruited to the sporulation septum during septum formation, after which GFPPBP2c/d is found on both sides of the asymmetric septum. Following the membrane during engulfment results in GFPPBP2c/d being distributed throughout both inner and outer prespore membranes (Fig. 1B). The fact that GFPPBP2c localizes to the engulfing membrane when expressed from the mother-cell-specific P_spoIID, which is switched on only after closure of the sporulation septum, argues in favour of the first model.

PBP2c is expressed during both vegetative growth and sporulation, under control of σ^G (Popham & Setlow, 1993a). The observed redistribution presumably reflects the behaviour of the vegetatively expressed PBP2c during wild-type sporulation. Interestingly, since a pbpFpbpG double mutant has no detectable defects in its spore germ cell wall, which is synthesized from the surface of the inner prespore membrane, but is severely affected in its cortical PG, which is synthesized from the outer prespore membrane (McPherson et al., 2001), it has been suggested that the more important site of PBP2c action is in the outer prespore membrane (Popham, 2002). The observed pattern reflects this mode of action of PBP2c.

PBP2d expression is specific to the prespore (Pedersen et al., 2000), but when expressed as a GFP fusion protein during vegetative growth, GFPPBP2d localized in a dispersed fashion along the membrane, similar to various other PBPs described earlier (Scheffers et al., 2004). To follow GFPPBP2d during sporulation, it was necessary to keep xylose present in the sporulation medium. Interestingly, even though GFPPBP2d was not expressed in the compartment in which it is expressed naturally, it did seem to recognize a targeting signal that guides it to the prespore septum and the engulfing membrane. Unfortunately, when expressed from sporulation-specific promoters, either in the prespore or in the mother cell, GFPPBP2d was degraded rapidly, making it impossible to confirm the localization of GFPPBP2d when expressed in the mother cell, or to study the localization of GFPPBP2d in the prespore in detail. PBP2c and PBP2d play redundant roles in sporulation (McPherson et al., 2001), and in agreement with this, do not depend on each other for their localization to the prespore.

PBP2c and 2d show localization patterns that are similar to patterns observed for SpoIVFB, a protein that localizes to the prespore outer membrane by diffusion-and-capture (Rudner et al., 2002). We see two possibilities for the diffusion-and-capture of PBP2c and PBP2d. First, it is possible that PBP2c and PBP2d are actively targeted to the prespore, or captured at the prespore membrane, via an unidentified protein pathway. This active targeting could make sense for PBP2c, which is expressed in both mother cell and prespore, but not for PBP2d, which is expressed in the prespore alone in the wild-type situation. So, if this model were true, under our experimental conditions, GFPPBP2d redistribution should not be observed, unless the protein factor that is recognized is present in the space between inner and outer prespore membranes, and accessible to PBPs present in either membrane. Secondly, PBP2c and PBP2d may be recruited to the prespore by the presence of substrate or substrate analogues. GFPPBP2c and GFPPBP2d follow the engulfing membrane, which contains SpoIID, SpoIIM and SPoIIQ at its leading edge (Abanes-De Mello et al., 2002). SpoIID is a cell wall hydrolase, which is suggested to use the cell wall as a track to drag along the membrane during engulfment (Abanes-De Mello et al., 2002). The hydrolase activity of SpoIID would release PG building blocks that PBP2c and PBP2d could recognize as substrates, and maybe recycle by using them for synthesis of the spore germ wall or cortex, even before engulfment is complete. Recent work in Staphylococcus aureus (Pinho & Errington, 2005) has shown that some high-molecular-mass PBPs depend on the presence of substrate for their correct localization. This has also been suggested for PBP localization in Streptococcus pneumoniae (Morlot et al., 2004). Targeting of PBP2c and PBP2d to the prespore by the availability of substrate is an attractive model, although the question would remain of why there is a difference in substrate binding between PBPs 2c and 2d and the other high-molecular-mass PBPs.

I thank Jeff Errington for his support, stimulating discussions and critical reading of the manuscript; Mariana Pinho for valuable comments on the manuscript; other members of the laboratory for helpful discussions and advice; Kit Pogliano and Aileen Rubio (University of California, San Diego) for plasmids pMDS13 and pMDS14. This work was supported by a Marie Curie Postdoctoral Fellowship (HPMF-CT-2001-01421), and a grant from the Biotechnology and Biological Sciences Research Council.

Footnotes

†Present address: Molecular Microbiology, Faculty of Earth and Life Sciences, Vrije Universiteit, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands.