The bacterial endospore is a dormant and heat-resistant form of life. StoA (SpoIVH) in Bacillus subtilis is a membrane-bound thioredoxin-like protein involved in endospore cortex synthesis. It is proposed to reduce disulphide bonds in hitherto unknown proteins in the intermembrane compartment of developing forespores. Starting with a bioinformatic analysis combined with mutant studies we identified the sporulation-specific, high-molecular-weight, class B penicillin-binding protein SpoVD as a putative target for StoA. We then demonstrate that SpoVD is a membrane-bound protein with two exposed redox-active cysteine residues. Structural modelling of SpoVD, based on the well characterized orthologue PBP2x of Streptococcus pneumoniae, confirmed that a disulphide bond can form close to the active site of the penicillin-binding domain restricting access of enzyme substrate or functional association with other cortex biogenic proteins. Finally, by exploiting combinations of mutations in the spoVD, stoA and ccdA genes in B. subtilis cells, we present strong in vivo evidence that supports the conclusion that StoA functions to specifically break the disulphide bond in the SpoVD protein in the forespore envelope. The findings contribute to our understanding of endospore biogenesis and open a new angle to regulation of cell wall synthesis and penicillin-binding protein activity.
Bacteria of the genera Bacillus and Clostridium can form endospores (Piggot and Hilbert, 2004; Paredes et al., 2005). In the form of an endospore a bacterium is dormant and can endure chemical and physical conditions that would otherwise kill the cell. The spore core, which corresponds to the cytoplasm of the vegetative cell and harbours the chromosome, is surrounded by several protective layers. The cortex layer, made of mainly peptidoglycan, helps maintaining the dehydrated state of the core and is important for heat resistance. The inner and outer coat layers, localized on the outer side of the cortex layer, consist mainly of protein and protect the endospore against chemicals and hydrolytic enzymes.
The endospore forms within the cell cytoplasm in a process termed sporulation. This developmental process, induced by nutrient starvation and high cell density, takes many hours for the bacterium to complete. Initially, asymmetrical cell septation takes place resulting in two differently sized cells; a larger mother cell and a smaller forespore. The forespore is subsequently engulfed by the mother cell. As a result, the forespore cytoplasm is surrounded by a double membrane and the forespore is at this stage like an organelle inside the mother cell cytoplasm (Hilbert and Piggot, 2004).
The cortex peptidoglycan layer is assembled in the compartment between the two forespore membranes from precursors primarily synthesized in the mother cell (Popham, 2002). Cortex peptidoglycan differs from peptidoglycan of vegetative cells by lack of associated teichoic acids, the presence of muramic-δ-lactam and an overall lower extent of cross-linking between glycan strands (Popham, 2002). Despite a wealth of information available on penicillin-binding proteins (PBPs) and peptidoglycan composition, structure and synthesis, the details of cell wall biogenesis, including endospore cortex morphogenesis, are largely not understood at the molecular level (Popham, 2002; Silvaggi et al., 2004; Sauvage et al., 2008). Some PBPs in Bacillus subtilis function in both vegetative and spore peptidoglycan synthesis, while others are specific for cortex synthesis (Buchanan and Neyman, 1986; Daniel et al., 1994). Late in sporulation the coat layers are deposited on the surface of the outer forespore membrane and the mature endospore is finally released by lysis of the mother cell. If nutrients again become available the spore can in minutes germinate and form a vegetative cell.
Sporulation is a tightly controlled cell-differentiation process involving a phosphor-relay and a sigma-factor cascade with associated intricate patterns of activation and repression of gene expression (Errington, 2003; Piggot and Hilbert, 2004) The mother cell-specific transcription factor σE is activated after cell septation and governs the transcription of many genes encoding proteins for cortex and coat maturation (Steil et al., 2005).
The StoA protein comprises a thioredoxin-like domain (with a –CysProProCys– motif), as shown by the recently determined structure of the protein (Crow et al., 2009a). The thioredoxin-like domain is membrane-anchored by an N-terminal peptide segment. The low redox midpoint potential (−248 mV at neutral pH) and other biophysical properties, its close similarities to ResA, and findings with BdbD-deficient mutants, indicate that StoA functions to break disulphide bonds in proteins on the outer side of the cytoplasmic membrane (Erlendsson et al., 2004; Crow et al., 2009a). The CcdA protein transfers reducing equivalents from thioredoxin (TrxA) in the cytoplasm across the membrane to reduce StoA and ResA (Carlsson Möller and Hederstedt, 2008). Absence of either StoA or CcdA in B. subtilis results in heat-sensitive spores that have no visible cortex layer as determined by electron microscopy (Erlendsson et al., 2004; Imamura et al., 2004). The observed effect of StoA deficiency is more dramatic than that of CcdA deficiency and cultures of a StoA CcdA double-deficient mutant form less heat-resistant spores than either single mutant (Schiött and Hederstedt, 2000; Erlendsson et al., 2004).
The finding that a protein thiol-disulphide oxidoreductase is involved in peptidoglycan synthesis was unexpected because the functions of known enzymes central to this process, i.e. transglycosylases, transpeptidases, carboxypeptidases and endopeptidases, are not dependent on thiol/disulphide chemistry, as discussed before (Möller and Hederstedt, 2006). Furthermore, B. subtilis CcdA and StoA are required for endospore synthesis only when the oxidizing, extra-cytoplasmic, thiol-disulphide oxidoreductase BdbD is active; lack of BdbD effectively suppresses the sporulation-defective phenotype caused by StoA and CcdA deficiency (Erlendsson and Hederstedt, 2002; Erlendsson et al., 2004).
The objective of the work reported here was to identify the protein reduced by StoA in B. subtilis and which functions directly or indirectly in endospore cortex synthesis in the intermembrane compartment of the forespore. To find such a protein, we first made a bioinformatics-based search and then experimentally analysed the role of each candidate protein for cortex synthesis. The results obtained pointed to the SpoVD protein as a probable StoA substrate. From a combination of in silico, in vitro and in vivo experiments we present evidence that a disulphide bond in SpoVD is a substrate for StoA reductase activity. We conclude that the activity of SpoVD in the forespore intermembrane compartment of sporulating Bacillus cells is regulated by a thiol-disulphide switch.
Search for candidate StoA substrate proteins
Cortex synthesis starts after engulfment of the forespore and occurs primarily during stage IV of the sporulation process (Hilbert and Piggot, 2004). Since many genes required for cortex synthesis in B. subtilis are under control of σE, we reasoned that a StoA substrate protein with a direct function in cortex synthesis should meet the following three criteria. It should: (i) be encoded by a gene transcribed from a σE-dependent promoter, (ii) be a protein secreted to the intermembrane compartment of the forespore envelope or be a membrane protein with functional domains exposed to this compartment, and (iii) contain at least one cysteine residue that by the action of BdbD can be oxidized to form a disulphide bond with a second cysteine residue in the same protein or in another protein (such that the presence of the disulphide bond affects the activity of the substrate protein).
Bacillus subtilis strain 168 has about 4100 protein-encoding genes. The σE regulon contains at least 253 genes (Eichenberger et al., 2004). Additional 276 genes are obtained if the results from three studies on σE-dependent gene expression (Eichenberger et al., 2003; 2004; Feucht et al., 2003) are combined and not filtered to remove inconsistent data. In a computer-based search we made use of all these 529 genes to find those that encode proteins containing cysteine residues and predicted to have an N-terminal hydrophobic segment (which may function as a signal sequence in secretion of the protein or as a membrane anchor) or several transmembrane segments. The program tmhmm (Krogh et al., 2001) was used to predict the membrane protein topology. The position of Cys residues were then mapped on the topology models. In parallel we searched the genome sequences of four isolates of Bacillus anthracis, two isolates of Bacillus cereus, and one strain each of Bacillus halodurans, Bacillus thuringensis, Clostridium acetobutylicum, Clostridium perfringens and Clostridium tetani for orthologues to the 529 B. subtilis proteins. We also searched three genomes of Listeria species, two genomes of Lactobacillus species and one genome of Lactococcus (these Gram-positive bacteria are related to B. subtilis but do not form endospores) to examine which proteins are unique for endospore-forming bacteria. The genome sequences used are specified in Table S1. Finally, we searched all predicted B. subtilis membrane proteins for Cys–n–Cys motifs (where n is 1, 2, 3, 4 or 5 residues other than cysteine) and mapped these motifs on topology models.
The computer-based search combined with manual inspection and elimination of some of the suggested proteins based on rational arguments (e.g. SpoIIE is not likely to be a StoA substrate since it is required for σF activation, which occurs prior to cortex synthesis) brought forward the following candidate proteins: SpmB, YloB, YyaD and YjcA. Screening the literature for proteins known to be involved in sporulation and having the features predicted for a StoA substrate also yielded SpoVB, SpoVD, SpoVE and YabQ.
Predicted membrane topologies and cysteine residue contents of the candidate proteins are shown in Fig. 1. The genes for the eight proteins are distributed over seven loci on the B. subtilis chromosome (Fig. S1). Table S2 presents a summary of the size and cysteine content of the proteins and presents paralogous proteins. Also indicated in the table is whether an orthologue of the respective B. subtilis protein is present in phylogenetically related bacteria.
Phenotype of strains deficient in the respective candidate StoA target protein
To determine the relative importance of the candidate proteins for cortex synthesis we inactivated the yabP, yabQ, yloB, spmB, spoVD, spoVB, spoVE, yyaD and yjcA genes in B. subtilis strain 1A1.
Four of the nine constructed mutants showed a clear defect in forming heat-resistant spores (Table 2). LMD5 (spoVB) showed a clear defect whereas LMD1p (yabP) and LMD6 (spoVE) were severely affected forming few heat-resistant spores. Notable is that YabQ is expected to be sporulation defective (Asai et al., 2001) but we found only a modest defect with strain LMD1q (Table 2). No heat-resistant spores could be detected in cultures of strain LMD4 (spoVD). The result with LMD4 was confirmed by analysis of a different SpoVD-deficient strain, LMD19 (Table 2), with a spectinomycin resistance gene cassette inserted in the middle of spoVD (Table 1).
Table 2. Sporulation efficiency of B. subtilis mutant strains derived from strain 1A1.
The genotypes of the strains are presented in Table 1. The strains carry in the chromosome either a disruption or a point mutation in the gene for the indicated protein. On minimal glucose medium and complex medium (nutrient sporulation medium with phosphate; NSMP) the strains grew similarly to the parental strain 1A1 and did not depend on the presence of isopropyl-β-d-thiogalactopyranosid (IPTG) in the medium. IPTG was added to strains with plasmid pMUTIN2mcs inserted into the gene of interest in order to avoid polar effects on transcription of genes located in the same operon downstream of the insertion (see Fig. S1).
Endospore formation was scored as heat-resistant cells after growth in NSMP medium for 2 days at 37°C. Sporulation efficiency was calculated as colony-forming units (cfu) after heating the culture at 80°C for 15 min divided by cfu of not heated culture. Sporulation assays were performed at least four times with each strain including sister clones. Representative results are shown.
Electron microscopy analysis confirmed (Piggot and Coote, 1976; Daniel et al., 1994) that spores of SpoVD-deficient strains lack detectable cortex and show morphological features that are very similar to those of heat-sensitive spores formed by StoA- and CcdA-defective strains (Erlendsson et al., 2004). Spores of strains lacking YabP, YabQ, SpoVB and SpoVE showed defective cortex layer but coat layers were also affected. The very similar morphology of spores of SpoVD- and StoA-deficient strains and the presence of two invariant appropriately positioned cysteine residues in the protein made us postulate that SpoVD is a substrate to StoA in B. subtilis.
SpoVD protein is a membrane-anchored protein appearing at an early stage during sporulation
The SpoVD protein (molecular mass 71 kDa) is a predicted high-molecular-weight, class B, PBP with five paralogues in B. subtilis (Table S2). Two cysteine residues are conserved among orthologues in endospore-forming bacteria (Fig. 1) but not in paralogous proteins, e.g. PBP2B (Fig. 2). The major part of SpoVD comprises a membrane-extrinsic domain of some 600 amino acid residues which is predicted, using the signal sequence analysis program SignalP (Bendtsen et al., 2004), to be anchored to the membrane by one single trans-membrane segment, as illustrated in Fig. 1.
To determine the localization and presence of SpoVD in sporulating cells we grew strains LMD18 (wild-type reference), LMD14 (SpoVD-deficient) and LUA14 (blocked in stage III of sporulation) for 30 h in NSMP medium and collected samples of culture during growth. Heat-resistant endospores appeared in the wild-type culture at about 5 h (T5) after the end of exponential growth phase and spore production was completed at 30 h. Strains LMD14 and LUA14 produced no heat-resistant spores, as expected. SpoVD protein of expected size was found in the membrane fraction of LMD18 and LUA14 (Fig. S2). No SpoVD antigen was detectable in the soluble fractions (immunoblots not shown). In the culture of LMD18 (wild type), SpoVD antigen was first seen at about T1, with a peak in the amount at about T4. The protein was not detectable in extracts harvested at about T8 or later (except for strain LUA14), suggesting that SpoVD is degraded or trapped by the coat layers in spores and therefore not detected by the analytical method used here. The latter explanation is supported by the observation that also BdbD protein, which is constitutively produced (Crow et al., 2009b) and was used as an internal control, was found in low amounts in samples taken after T4 (except for LUA14) (Fig. S2). StoA protein in membranes of sporulating cells shows a similar pattern to that observed here for SpoVD (Crow et al., 2009a).
Biochemical properties of SpoVD
SpoVD without the N-terminal membrane anchor part (residues 1–32) was produced as a water-soluble GST fusion protein in the cytoplasm of Escherichia coli. The soluble SpoVD domain, termed sSpoVD, was isolated from cell-free lysates by affinity chromatography followed by cleavage of the fusion protein, as described in Experimental procedures.
The identity of the purified sSpoVD protein of 615 residues (65 kDa) was assessed by Edman sequence analysis and by mass-spectrometry of tryptic peptides (18 different peptides; 42% sequence coverage). The analysis confirmed the N- (GSVQF–) and C-terminal (–DEKEAAD) sequence of the polypeptide.
To probe the redox state of the two cysteine residues in SpoVD we incubated the protein with monomethyl polyethylene glycol 5000 2-maleimidoethyl ether (MAL-PEG), which reacts with thiol groups forming a covalent adduct. Bound MAL-PEG adds a mass of about 5 kDa to the protein and this can be monitored by SDS-PAGE as a protein mobility shift. Reduced, Tris(2-carboxyethyl)phosphine (TCEP)-treated, sSpoVD reacted with MAL-PEG and this resulted in two products corresponding to the reaction with one or two cysteine residues (Fig. 3). At low concentrations of MAL-PEG (< 0.01 mM) predominantly one cysteine residue was modified (Fig. S3). Oxidized and untreated (as isolated) sSpoVD did not react with MAL-PEG indicating that the two cysteine residues form a disulphide bridge. Gel permeability chromatography of both untreated and diamide-oxidized sSpoVD indicated that the protein is a monomer (apparent mass 58 kDa) showing that the disulphide bond in SpoVD is intramolecular.
SpoVD is known to bind penicillin (Daniel et al., 1994) but the properties of isolated protein and role of redox state for binding have not been analysed. After incubation of isolated sSpoVD with a large molar excess of 3H-benzylpenicillin we found that up to 0.3 mol of penicillin could be covalently bound per mol of sSpoVD (Fig. S4). The apparent binding association constant was found to be in the nM range (Fig. S4). The off rate is low since incubation of 3H-benzylpenicillin-labelled sSpoVD in the presence of a large excess of cold benzylpenicillin did not decrease the amount of covalently bound radioactivity. Both oxidized and reduced sSpoVD reacted with penicillin but the reduced protein bound 1.4-fold more penicillin (Table 3 and Fig. S4). The results show that covalent binding of penicillin to SpoVD is affected by the redox state and presumably by the presence or absence of the disulphide bond. We conclude that SpoVD can covalently bind one penicillin molecule per polypeptide as predicted from the amino acid sequence. The obtained apparent submolar stoichiometry in penicillin-binding capacity can be explained by the presence of a fraction of protein being in a not active conformation, by an over-estimation of sSpoVD protein (as determined using the BCA method) despite the apparent purity of the preparations (see lane 1 in Fig. 3 and Fig. S3) and by impurities in the commercial preparation of 3H-benzylpenicillin (see Experimental procedures).
A relative amount of 1 correspond to an estimated amount of 0.3 mol of penicillin bound per mol of sSpoVD (Fig. S4).
Student's t-test shows P < 0.001 (n = 3) for these results as compared with those obtained with reduced sSpoVD.
The sample was incubated with 30 mM benzylpenicillin for 15 min before radioactive penicillin was added.
Isolated sSpoVD (0.33 mg ml−1) was oxidized using 2 mM diamide or reduced using 2 mM TCEP or left untreated and then, after removal of the redox agent, incubated in the presence of 15 μM phenyl-4(n)-[3H]-benzylpenicillin. Radioactivity bound to polypeptides after SDS-PAGE was determined by scintillation counting as described in Experimental procedures. Glutathione S-transferase (GST) was used as a control protein in the experiment.
There are no three-dimensional structural data available for SpoVD but X-ray crystal structure data for the membrane-extrinsic part of a number of high-molecular-weight PBPs are available (Contreras-Martel et al., 2009) and recently the structure of E. coli PBP1b including the membrane anchor part was published (Sung et al., 2009). The sequence of the SpoVD membrane-extrinsic part is similar (overall c. 33% identity) to that of Streptococcus pneumoniae PBP2x (Fig. 2 and Fig. S5) (Yanouri et al., 1993) for which several crystal structures have been determined (Gordon et al., 2000; Dessen et al., 2001). The membrane-extrinsic part of PBP2x consists of three structural domains; an N-terminal pedestal domain (about 210 residues), a middle penicillin-binding domain (about 340 residues) and a C-terminal domain (about 140 residues that shows little sequence similarity to SpoVD). Residues Asp-373 and Ser-396 in the centre of the penicillin-binding domain of PBP2x correspond in position to Cys-332 and Cys-351 in SpoVD (these residues are indicated by triangles in Fig. 2).
We used X-ray 2.4 Å crystal structure information for PBP2x (Fig. S5) to model the structure of the N-terminal and penicillin-binding domains of SpoVD comprising residues 54–556 (Fig. 4A). With the Modeler integrated within the Accelrys Discovery Studio package we produced a homology model with a Verify score (Lüthy et al., 1992) of 169.41 (high score: 187.49 and low score: 84.37) indicative of a proper fold. The Ramachandran plot showed 89.8% (307) residues in allowed region, 9.1% in marginally allowed and 1.2% (4) in disallowed regions. Residues of the penicillin-binding catalytic site consensus motifs SXXK (Ser-294 and Lys-297), SXN (Ser-350 and Asn-352) and KT/SG (Lys-496, Thr-497 and Gly-498) are positioned appropriately in the model. Five of these residues are indicated in the modelled structure shown in Fig. 4. Residue Ser-294 forms an acyl covalent adduct upon reaction with β-lactam or d-alanine in a peptide side-chain of un-cross-linked peptidoglycan. Cys-332 is found at the end of a loop and Cys-351 is residue X in the SXN active-site motif (Fig. 2). SpoVD was modelled both with and without a disulphide bond between Cys-332 and Cys-351 (Fig. 4B and C). The two structures are very similar and essentially only differ in the region of the mentioned loop adjacent to the substrate binding cleft of the penicillin-binding domain.
Normal mode analysis using the extracted pdb files and the program molmol (Koradi et al., 1996) provided information about the difference in polypeptide flexibility between the model of oxidized and reduced SpoVD (Fig. S6). In the presence of the disulphide bond the modelled protein was much more rigid in the region of the loop at the entrance to the penicillin binding active site. This difference in polypeptide dynamics depending on the disulphide bridge could have considerable effects on the function of SpoVD.
Functional importance of cysteine residues in SpoVD
To investigate the physiological role of the two cysteine residues and the disulphide bond in SpoVD we replaced these residues individually with an aspartate and serine, respectively, to mimic the situation in PBP2x (Fig. 2). A Cys-332Ser mutation was also made. Mutations in the codons for the cysteine residues were introduced into the spoVD gene in the B. subtilis chromosome, to assure normal expression (both in level and in time) of spoVD and the murE-mraY-murD downstream genes (Fig. S1).
SpoVD with the Cys-332Asp mutation (LMD21) was found to be functional in endospore biogenesis whereas a Cys-332Ser mutation (strain LMD22) resulted in abnormal sporulation efficiency (Table 2), suggesting that a negatively charged residue is required at position 332 in SpoVD. Strain LMD24 with the Cys-351Ser mutation produced heat-sensitive spores (Table 2) and contained membrane-bound SpoVD protein (Fig. S2), indicating that this mutation inactivates SpoVD.
We next exploited the Cys-332Asp mutation in SpoVD to prove in vivo that SpoVD is a substrate for StoA. Because Cys-332Asp cannot form an intramolecular disulphide bond, we reasoned that strains with this variant of SpoVD would form heat-resistant spores independent of the presence of StoA and CcdA.
Strains containing the Cys-332Asp mutation in SpoVD and a deletion of the stoA or ccdA gene were constructed. As one control we also constructed a bdbD-deficient strain. The strains, and the background isogenic reference strain, were grown for sporulation and the production of heat-resistant spores was monitored as a test for cortex synthesis. The results, presented in Fig. 5, demonstrate that this single amino acid substitution in SpoVD suppresses the negative effect on endospore synthesis otherwise observed with StoA- or CcdA-deficient mutants containing wild-type SpoVD. The SpoVD Cys-332Asp mutation also efficiently suppresses the effect of StoA CcdA double deficiency. These results are exactly as expected if the disulphide bond in SpoVD inactivates the protein and is a substrate reduced by StoA.
In the work reported in this article we first identified SpoVD as a putative target protein for the thioredoxin-like, membrane-bound, thiol-disulphide oxidoreductase StoA in sporulating B. subtilis cells. With purified protein we then showed that the two cysteine residues in SpoVD, Cys-332 and Cys-351, are surface-exposed and redox-active. As supported by structural modelling, the two residues in the oxidized protein are probably linked together by an intramolecular disulphide bond (Fig. 4). Finally we demonstrated that StoA and its electron donor protein, CcdA, are dispensable for the production of heat-resistant spores if residue Cys-332 in SpoVD is replaced by an aspartate (Fig. 5). From the available data we conclude that formation of the disulphide bond in SpoVD is catalysed by the activity of BdbD (a thioredoxin-like disulphide oxidizing enzyme which cooperates with BdbC) (Bolhuis et al., 1999; Möller and Hederstedt, 2006; Kouwen and van Dijl, 2008) whereas StoA (in cooperation with CcdA) functions to break that bond (Fig. 6).
At the stage in sporulation when the forespore becomes completely engulfed, the direct contact of the forespore with the external milieu is broken. A considerable change in the environment, including redox status, of what becomes the forespore intermembrane compartment can therefore be envisaged to accompany engulfment. The inner and outer membranes of the forespore envelope are both derived from the cytoplasmic membrane of the vegetative cell and most likely contain BdbD and CcdA, which have functions in vegetative cells (Kouwen and van Dijl, 2008). SpoVD and StoA (both encoded by σE-dependent genes) are translated in the cytoplasm of the mother cell and inserted into the forespore outer membrane with their catalytic domains facing the intermembrane compartment. At present we do not know if the two proteins are sublocalized to the forespore or inserted also into the cytoplasmic membrane of the mother cell. SpoVE, being a member of the SEDS family of proteins and proposed to interact with SpoVD, is synthesized in the mother cell and directed post-translationally to the forespore envelope (Real et al., 2008). At a later stage in sporulation, due to the σG dependence of the stoA gene, StoA is inserted into the inner membrane of the forespore (Imamura et al., 2004) and spoVD expression is downregulated due to repression by SpoIIID (Zhang et al., 1997).
Newly membrane-inserted SpoVD is presumably immediately oxidized by the action of BdbD leading to formation of the disulphide bond between residues Cys-332 and Cys-351. This apparently inactivates SpoVD by a direct effect of the bond on the activity of the penicillin-binding domain and/or on the interaction with other proteins involved in cortex biogenesis. The Cys-351 thiol group, positioned in the active-site region, appears very important for the activity of SpoVD as shown by the effect of the Cys-351Ser mutation (Table 2). Substitution of Cys-351 with another amino acid can have essentially the same effect on SpoVD activity as a disulphide bond between Cys-332 and Cys-351 (equivalent to a reversible chemical modification of Cys-351). The disulphide bond, however, would have more profound effects than an amino acid substitution in that it also restricts the dynamics of the protein including flexibility at the active site (Fig. S6).
A complete lack of SpoVD protein causes a complete block in the production of heat-resistant spores (Table 2) (Daniel et al., 1994). In contrast, cultures of StoA- and CcdA-deficient strains show a leaky phenotype with the production of low amounts (0.03–4% of wild-type) of heat-resistant spores (Schiött and Hederstedt, 2000; Erlendsson et al., 2004). This difference can be understood in the light of our findings. SpoVD is presumably inactive if the disulphide bond is present in the protein. Disulphide bonds can form and break also in the absence of enzymes with thiol-disulphide oxidoreductase activity (BdbD and StoA in this case) but at a slow rate. For kinetic and thermodynamic reasons, SpoVD in the cell would always be a mixture of oxidized and reduced proteins (Steen Jensen et al., 2009). The fraction of reduced, active, SpoVD and the amount of this protein is expected to vary stochastically between cells and also subcellularly in membranes. Such a variation in SpoVD activity combined with a trigger-like function of SpoVD (as discussed below) can explain why some heat-resistant spores form in cultures of, for example, StoA-deficient strains.
Bacillus subtilis PBP2B is very similar to SpoVD but does not contain any cysteine residues in the membrane-extrinsic domain (Yanouri et al., 1993). The genes for these two proteins (pbpB and spoVD) have probably originated as the result of a tandem gene duplication in the chromosome (Fig. S1). PBP2B is essential for growth, has a role in septation of the cell during vegetative growth, and is thought to interact with FtsW (Yanouri et al., 1993). SpoVD is sporulation-specific and believed to assemble with SpoVE and probably other not yet identified proteins in a dynamic complex operating in cortex biogenesis. The N-terminal and/or C-terminal domains flanking the penicillin-binding domain in SpoVD could be major determinants for the assembly and composition of this protein complex in the forespore outer membrane. Notably, SpoVE is an integral membrane protein with two invariant cysteine residues (Fig. 1) not found in homologous proteins that function in vegetative cells, e.g. RodA and FtsW (Real et al., 2008). We cannot rule out the possibility of redox communication between SpoVE and SpoVD.
The catalytic activities of SpoVD are not known but the protein most likely has transpeptidase activity as indicated by the penicillin-binding properties (Table 3 and Fig. S4). There is no transglycosylase-like domain in SpoVD. The available information thus suggests that SpoVD catalyses the formation of peptide cross-links between glycan strands in cortex peptidoglycan. Several other high-molecular-weight PBPs are also engaged in spore synthesis in B. subtilis, e.g. PBP2C, PBP2D and PBP3 (Scheffers and Pinho, 2005). The SpoVD orthologues B. subtilis PBP2B and S. pneumoniae PBP2x function in septal cell wall synthesis (Daniel et al., 2000; Morlot et al., 2003). This could suggest that SpoVD has a similar function and possibly plays a critical role only in an initial phase of cortex synthesis, i.e. has a primer or trigger-like function essential for subsequent action of enzymes in cortex peptidoglycan synthesis. This would be consistent with the fact that SpoVD-deficient strains produce spores without a trace of cortex layer, are completely blocked in production of heat-resistant spores, and accumulate peptidoglycan precursors in the mother cell (Daniel et al., 1994; Vasudevan et al., 2007). The proposed substrate of SpoVD, i.e. nascent peptidoglycan strands, is large and contains muramic-δ-lactam residues and therefore few pentapeptide side-chains that can be cross-linked. We found only a small effect in covalent binding of benzyl-penicillin depending on the redox state of SpoVD (Table 3). This small effect might be explained by the size and structural differences of native substrate and penicillin and does not rule out the possibility that the disulphide bond in SpoVD in the cell blocks binding of endogenous substrate.
Functional or structural importance of the two cysteine residues in SpoVD is indicated by their absolute conservation in Bacillus and Clostridium species and by the fact that they are only found in PBPs of endospore formers (Table S2 and Fig. 2). Some Clostridium species, such as C. acetobutylicum, entirely lack membrane-bound thiol-disulphide oxidoreductases as judged from the genome sequence. C. perfringens is unusual in that it contains CcdA and StoA orthologues of unknown function (Möller and Hederstedt, 2006). Disulphide bond formation in SpoVD might not occur to a significant extent under anoxic conditions, therefore excluding a need for a StoA-like protein in Clostridium species. Alternatively, reduction of the disulphide bond in SpoVD in the anaerobes is performed by novel thiol-disulphide reductases (Dutton et al., 2008).
It is intriguing that the proposed regulation of SpoVD activity through a thiol-disulphide redox switch seemingly is not very important to the cell, as shown by for example the production of normal amounts of heat-resistant spores by BdbD and StoA double-deficient strains (Fig. 5). Synthesis of StoA in the forespore, driven by the σG-dependent promoter in front of stoA, is sufficient for synthesis of heat-resistant spores and the membrane anchor is seemingly not required for function of StoA (Imamura et al., 2004). This may indicate that the role of StoA is to activate SpoVD only in the forespore envelope and not in the cytoplasmic membrane of the mother cell or vegetative cell. Such activation of SpoVD by StoA specifically in the forespore envelope could exist to assure correct temporal and spatial activity of SpoVD and avoid interference in cell wall synthesis from fortuitously synthesized or mis-located SpoVD protein. Remarkably, the role of StoA in sporulation is analogous to that of ResA in cytochrome c maturation in B. subtilis. ResA and StoA are structurally very similar proteins and both are reduced by CcdA, but they have distinct and non-overlapping substrate specificities (Crow et al., 2009a). ResA functions to break a disulphide bond in apo-cytochrome polypeptides formed by the action of BdbD (Erlendsson et al., 2003; Lewin et al., 2006). In the absence of BdbD, ResA is not required for cytochrome c maturation. StoA and ResA in conjunction with CcdA have thus evolved to break unwanted disulphide bonds resulting from the action of BdbD. The BdbD protein is required for maturation of two Com proteins that are essential for uptake of exogenous DNA (Meima et al., 2002).
In this article we present evidence for the presence of a regulatory disulphide bond in the SpoVD protein. We conclude that the formation and disruption of the disulphide is catalysed by BdbD and StoA respectively. The crystal structures of both these extra-cytoplasmic thiol-disulphide oxidoreductases have been determined (Crow et al., 2009a,b). Further studies on the catalytic activity of StoA and BdbD with SpoVD as substrate in vitro with purified components, structural analysis of co-crystals of these proteins, as well as determination of the redox state of SpoVD in vivo in BdbD- and StoA-deficient B. subtilis strains are experimental approaches that can be adopted to validate our findings. The results of that type of future investigations will deepen the knowledge about enzyme substrate discrimination and how PBPs operate in peptidoglycan synthesis.
Bacterial strains and growth media
Strains and plasmids used in this work are presented in Table 1 and Table S3 respectively. E. coli MM294 and TOP10 were used in cloning experiments and strain MC1061 was used for protein production. E. coli strains were grown at 37°C in LB medium or on LB agar plates (Sambrook and Russel, 2001). B. subtilis strains were cultivated at 30°C or 37°C in LB medium, nutrient sporulation medium with phosphate (NSMP) (Fortnagel and Freese, 1968) or on tryptose blood agar base (TBAB) plates (Difco). Antibiotics were used when appropriate at the following concentrations: kanamycin 50 μg ml−1; ampicillin 100 μg ml−1 for E. coli, and chloramphenicol 4 μg ml−1; erythromycin 0.3 μg ml−1 or 1 μg ml−1; kanamycin 10 μg ml−1, lincomycin 12.5–20 μg ml−1; spectinomycin 100 μg ml−1, tetracycline 15 μg ml−1 for B. subtilis.
Plasmid DNA was isolated using Quantum miniprep (Bio-Rad) or EZNA plasmid miniprep kit I (Omega Bio-tek) or by CsCl density gradient centrifugation. Chromosomal DNA from B. subtilis was isolated according to Marmur (1961). E. coli was transformed by electroporation (Hanahan et al., 1991) and B. subtilis was grown to natural competence as described by Hoch (1991). PCR was carried out using Taq DNA polymerase (New England Biolabs) or Phusion polymerase (Finnzymes) and B. subtilis chromosomal DNA or plasmid DNA as template. Oligonucleotides used are listed in Table S4. DNA ligation was performed using T4 DNA ligase (New England Biolabs).
All DNA fragments cloned in plasmids were verified by sequence analysis using specific and universal vector primers. Three errors were revealed in the B. subtilis 168 spoVD sequence (Daniel et al., 1994) and they affect residues 540–549 in the predicted polypeptide sequence. The DNA sequence errors in spoVD were recently identified also by the re-sequencing of the strain 168 genome (Barbe et al., 2009).
Construction of the plasmids for inactivation of genes in B. subtilis is described in Supporting information. Plasmid for the production of a thrombin-cleavable GST–sSpoVD fusion protein in E. coli was obtained as follows. A fragment of the spoVD gene encoding residues 33–645 was first amplified by PCR using B. subtilis 1A1 chromosomal DNA and primers: LY009 and LY011 (Table S4). The PCR product was cloned into pCR®-Blunt-II-TOPO® resulting in plasmid pLYM018. The insert was cut out from the plasmid pLYM018 using BamHI and EcoRI and ligated into pGEX2T cut with the same enzymes, resulting in plasmid pLYM020.
Construction of B. subtilis strains with a mutated chromosomal spoVD gene
Site-specific mutations in spoVD were generated using the QuikChange II kit and protocol (Stratagene) and plasmid pLYM018 (Table S3) as the template and primers LY012–LY017 (Table S4). The resulting plasmids were denoted pLYM021 (C332D), pLYM022 (C332S) and pLYM024 (C351S) respectively.
Bacillus subtilis strain LMD12, containing a spectinomycin gene inserted into the unique PstI site in spoVD (the site is located between the codons for the two cysteine residues of SpoVD), was obtained by transformation of strain 1A1 with linearized pLJM2. Plasmid pLJM2 was obtained by insertion of the spectinomycin resistance gene of pSPC(+) on a PstI fragment into pLJM1. pLJM1 was in turn generated by amplification of a 5′-truncated spoVD fragment using PCR with primers JM1 and JM2 (Table S4). The PCR product was cloned in pCR-Blunt-II-Topo and then moved as a SacI–EcoRI fragment to pBAD/gIIIA giving pLJM1.
Site-specific point mutations in the chromosomal spoVD gene in B. subtilis was introduced by transformation of strain LMD12 with a mixture of one of the plasmids pLYM021, pLYM022 or pLYM024 that had been linearized by cleavage with ScaI and chromosomal DNA of strain LMD15 that is identical to LMD12 except that it contains an erythromycin resistance gene inserted into amyE. Erythromycin-resistant transformants were selected on TBAB plates and then tested for spectinomycin sensitivity. Chromosomal DNA was isolated from transformants that were erythromycin-resistant and spectinomycin-sensitive. The spoVD gene was amplified from the chromosomal DNA by PCR and the presence of the desired mutation was confirmed by DNA sequence analysis. Strains containing multiple mutations were obtained by transformation as indicated in Table 1.
Preparation of B. subtilis cells for analysis by electron microscopy was performed essentially as described before (Erlendsson et al., 2004). Cultures were grown in NSMP medium at 37°C and harvested by centrifugation 24 h after entry into stationary phase. After fixation in 3% glutharaldehyde in 50 mM phosphate buffer (pH 6.5) overnight, the cells were post-fixed in 1% osmium tetroxide for 1 h, dehydrated and embedded in Epon. Sections were stained in 2% uranyl acetate and in lead citrate according to the method of Reynolds (1963). Sections were examined using a JEOL 1230 transmission electron microscope.
Production of GST–sSpoVD fusion protein
For the production of GST–sSpoVD fusion protein, E. coli MC1061/pLYM020 was grown at 37°C in 1 l portions in 5 l baffled E-flasks on a rotary shaker (200 r.p.m.). At OD600 = 0.6–0.8 gene expression was induced by addition of 1 mM IPTG (final concentration). After incubation for 5 h, cells were collected by centrifugation, washed in PBS (140 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4, pH 7.3) and stored as pellets at −20°C. The cells from 1 l of culture were suspended in 20 ml of ice-cold PBS and disrupted by passage (three times) through a French pressure cell at 18 000 Psi. The lysate was centrifuged at 48 000 g for 40 min and the obtained supernatant was centrifuged at 100 000 g for 1 h at 4°C. The final supernatant was mixed with 2 ml of 50% slurry of Glutathione Sepharose 4B (GE Healthcare) and the GST–sSpoVD fusion protein was purified according to manufacturer's instructions. Affinity-purified GST–sSpoVD was incubated with 50 units of thrombin (GE Healthcare) at room temperature for 5 h to cleave the fusion protein and then loaded onto a Sephacryl S-100 HR gel filtration column. Protein was eluted using 50 mM Tris-HCl, pH 8.0, containing 100 mM NaCl. Fractions containing sSpoVD were identified using SDS-PAGE and immunoblot with SpoVD antiserum, and then pooled and concentrated. The purity of the obtained preparations was assessed by SDS-PAGE and staining for protein (Fig. 3 and Fig. S3).
Modification of sSpoVD with MAL-PEG
sSpoVD protein in 50 mM Tris-HCl, 100 mM NaCl, pH 8.0 was reduced with 1 mM TCEP or oxidized with 1 mM diamide at room temperature for 30 min. Excess TCEP or diamide was removed using a YM10 column (Millipore). Untreated, reduced or oxidized sSpoVD (5 μg) was incubated with 1 μM to 1 mM mono-methyl polyethylene glycol 5000 2-maleimidoethyl ether (MAL-PEG) (≥ 90%, Fluka) at room temperature for 30 min. The samples were then applied directly onto a SDS-polyacrylamide gel.
Penicillin binding assay
Purified sSpoVD protein, 3.3 μg in 50 mM Tris-HCl, 100 mM NaCl, pH 8.0, was incubated at different concentrations of phenyl-4(n)-[3H]-benzylpenicillin (18 Ci mmol−1) (GE Healthcare) for 15 min at room temperature. The reaction was terminated by the addition of an excess (1 mg ml−1) of unlabelled benzylpenicillin (Sigma Chem). In competition assays, the protein was pre-incubated with ‘cold’ benzylpenicillin (30 mM) at room temperature for 15 min prior to labelling with [3H]-benzylpenicillin. sSpoVD was after incubation with penicillin subjected to electrophoresis using NuPAGE precast gels (Invitrogen). The gels were stained for protein with Coomassie brilliant blue G250 (Sigma) and dried. sSpoVD protein bands were cut out and incubated overnight in 200 μl of 3% H2O2 to dissolve the gel slices. To each sample was added 3 ml of Ultima Gold scintillation liquid (Perkin Elmer) and 3H was determined using a Tri-carb 2800TR liquid scintillation analyser (Perkin Elmer).
The concentration of active penicillin in the commercial preparation of phenyl-4(n)-[3H]-benzylpenicillin was determined as follows. B. subtilis 1A1 was cultured overnight in LB and diluted to OD600 = 0.2. Two millilitres of the diluted cultures were poured on a TBAB plate and then removed. The plate was dried at 37°C for 30 min. Different amounts of benzylpenicillin stock solution of known concentration and 5 μl of phenyl-4(n)-[3H]-benzylpenicillin were applied onto paper filter discs on the TBAB plate. After incubation at 37°C for 24 h the diameter of growth inhibition zones was measured and the concentration of penicillin in the radioactive sample calculated. The preparation of [3H]-benzylpenicillin was found to be about 60% pure as compared with the preparation of non-radioactive benzylpenicillin.
Structural modelling and protein dynamics analysis
Using Modeler within the Accelrys Discovery Studio package we created five models with, and five models without, a disulphide bond between residues Cys-332 and Cys-351. The models with the lowest probability density function (PDF) total energy (thiol-containing model, 2210.7 before and −9586.9 after loop refinement; disulphide-containing model, 2203.3 before and −6912.0 after loop refinement) were chosen as the best ones to continue with in minimization using the CHARMm force field. After minimization a Normal Mode Analysis protocol was run on the reduced and the oxidized model. A total of 20 modes were generated from each model. From each mode 50 pdb files were extracted to represent the dynamics shown in Fig. S6.
Protein concentrations were determined using the BCA reagent (Bio-Rad Chem) with bovine serum albumin as reference. SpoVD antiserum was obtained by immunizing rabbits with isolated sSpoVD produced in E. coli. Preparation of cell-free extracts from B. subtilis strains and immunoblot analysis were performed as described before (Crow et al., 2009a). The molecular size of sSpoVD was determined using high-performance liquid chromatography with an Ultraspherogel SEC 3000 column (Beckman). Approximately 1.2 nmol of sSpoVD in 10 μl was applied to the column equilibrated with 20 mM Tris-HCl, pH 8.0, 0.15 M NaCl, 1 mM DTT and eluted in the same buffer at a flow rate of 1 ml min−1. The molecular size of sSpoVD was calculated based on a calibration curve obtained using catalase (230 k), bovine serum albumin (67 k), ovalbumin (46 k), carboanhydrase (26 k), myoglobin (17.2 k) and horse heart cytochrome c (12.3 k).
We thank Jeff Errington for B. subtilis strain 785 and Richard Henriksson, Dan Li, Jennie Malm, Ingrid Stål, Hadija Trojer and Senad Zjajo for assistance in construction and analysis of strains, Rita Wallén for expert assistance with electron microscopy, Marit Lenman for help with mass-spectrometry and Nick Le Brun for critical reading of the manuscript. This work was supported by Grant 621-2007-6094 from the Swedish Research Council and by the Swedish National Research School in Genomics and Bioinformatics.