A Conserved Cysteine Residue of Bacillus subtilis SpoIIIJ Is Important for Endospore Development

During sporulation in Bacillus subtilis, the onset of activity of the late forespore-specific sigma factor σG coincides with completion of forespore engulfment by the mother cell. At this stage, the forespore becomes a free protoplast, surrounded by the mother cell cytoplasm and separated from it by two membranes that derive from the asymmetric division septum. Continued gene expression in the forespore, isolated from the surrounding medium, relies on the SpoIIIA-SpoIIQ secretion system assembled from proteins synthesised both in the mother cell and in the forespore. The membrane protein insertase SpoIIIJ, of the YidC/Oxa1/Alb3 family, is involved in the assembly of the SpoIIIA-SpoIIQ complex. Here we show that SpoIIIJ exists as a mixture of monomers and dimers stabilised by a disulphide bond. We show that residue Cys134 within transmembrane segment 2 (TM2) of SpoIIIJ is important to stabilise the protein in the dimeric form. Labelling of Cys134 with a Cys-reactive reagent could only be achieved under stringent conditions, suggesting a tight association at least in part through TM2, between monomers in the membrane. Substitution of Cys134 by an Ala results in accumulation of the monomer, and reduces SpoIIIJ function in vivo. Therefore, SpoIIIJ activity in vivo appears to require dimer formation.


Introduction
Protein secretion and membrane protein insertion are fundamental processes in all living organisms, and several pathways are known to serve those purposes, namely the Sec, Tat, and YidC/ Oxa1/Alb3 pathways [1,2,3]. These pathways have been extensively studied in bacteria, specifically in the model organisms Escherichia coli and Bacillus subtilis. The most prominent one is the Sec pathway. Its core is the SecYEG channel, which is utilised both for secretion and membrane protein insertion, using distinct piloting factors that guide each class of proteins to SecYEG. The SecB chaperone in E. coli (possibly CsaA in B. subtilis) guides proteins to be secreted. In contrast, nascent membrane proteins are delivered to SecYEG by the signal recognition particle (SRP) [reviewed in 1,4]. The twin-arginine translocation (Tat) pathway has the peculiarity of permitting the transport of previously folded proteins [reviewed in 2]. YidC/Oxa1/Alb3 family members, found in archaea, bacteria and in eukaryotic organelles, participate in the biogenesis of membrane proteins. The first protein from this family to be identified was Oxa1 in mitochondria of Saccharomyces cerevisiae. Members of this family are polytopic membrane proteins that share a core architecture of five transmembrane (TM) segments. They perform critical roles in membrane insertion and folding and also in the assembly of energy-transducing multimeric membrane complexes [reviewed in 3]. The best-studied proteins from this family in bacteria are YidC (E. coli) and SpoIIIJ (B. subtilis), respectively, which function with the Sec channel or independently of it. The Sec-independent function of these proteins is conserved and similar to that of Oxa1 in mitochondria, which lack the Sec system [3,5,6]. The large periplasmic region of YidC (absent in SpoIIIJ and from the conserved core that characterises the family) was found to be monomeric [7,8]. However, full-length YidC was observed as a dimer through cryo-electron microscopy [9], suggesting that the determinants for dimerisation lie within the conserved, membrane-associated, regions of the protein. SpoIIIJ is required for spore formation in B. subtilis [10]. The conversion of vegetative cells into highly resistant spores is one of the responses that this organism uses to cope with highly stressful situations [11,12,13]. Besides SpoIIIJ, another member of this family exists in B. subtilis, YqjG. Although they can substitute for each other during growth, only SpoIIIJ supports efficient sporulation [10,14,15,16,17].
Spore differentiation takes place in a cell divided into two unequally-sized compartments, between which successive waves of gene expression in one compartment or the other are activated in a coordinated manner and in register with the course of morphogenesis [11,12,13]. Transcription is controlled by the successive activities of cell type-specific RNA polymerase sigma factors. From the forespore, s F signals the activation of s E in the larger mother cell, which in turn is required for activity of s G in the forespore following engulfment completion. s G is in turn responsible for the activation of s K in the mother cell [18,19]. A key process of endosporulation is the engulfment of the forespore by the mother cell. The onset of s G activity is normally coupled to engulfment completion, when the forespore becomes isolated from the extracellular milieu and surrounded by two membranes [13,19]. At this stage of development, assembly of a specialised secretion system that spans the two forespore membranes is thought to allow the mother cell to feed the forming spore, allowing the continuation of macromolecular synthesis [10,20,21,22,23,24]. Eight mother cell-specific proteins encoded by the spoIIIA operon, and the forespore-specific SpoIIQ protein form the complex. SpoIIIJ is needed specifically during this step of sporulation possibly because it participates in the biogenesis of the secretion complex [10,20,23,24,25,26]. Evidence suggests that SpoIIIJ interacts with one of the spoIIIA proteins, SpoIIIAE, in the context of the Sec translocon to promote its correct insertion into the forespore membranes [20,24].
Here we have purified SpoIIIJ from E. coli cells and shown that it forms a dimer. We also provide evidence that a cysteine residue predicted to be in the second transmembrane segment of SpoIIIJ facilitates dimer formation by forming a disulphide bond at the dimer interface. Replacement of the cysteine by an alanine residue results in accumulation of the monomer and under certain conditions impairs the activity of s G and reduces sporulation. Our results suggest a model in which a disulphide-bond contributes, with additional non-covalent interactions, to the formation of a SpoIIIJ dimer, and that this dimer is important for SpoIIIJ activity during sporulation.

Media, bacterial strains and general techniques
The B. subtilis strains used in this work are congenic derivatives of the Spo + strain MB24 (trpC2 metC3) and are listed in Table S1. Luria-Bertani (LB) medium was used for growth or maintenance of E. coli and B. subtilis, and sporulation was induced in Difco sporulation medium (DSM) [27,28].

Strains and plasmids
A spoIIIJ C134A allele for expression in E. coli was constructed in two steps: an initial amplification from chromosomal DNA of the wild-type B. subtilis strain MB24 using primers pairs J112D with JC134A_D, and JC134A_R with Jhis (Table S2); next, the PCR fragments were joined using the external primers through splicing by overlap extension (SOE) [29]. The final PCR product was cleaved with NcoI and EcoRI, ligated to pETDuet-1 (Novagen, Darmstadt, Germany) similarly digested to obtain pFiV1 (plasmids are listed in Table S3), resulting in a fusion to a His 6 tag at the 39 of spoIIIJ C134A .
For controlled expression of His 6 fusions to spoIIIJ and spoIIIJ C134A in B. subtilis from a promoter that can be induced with isopropyl-b-D-thiogalactopyranoside (IPTG), chromosomal DNA of MB24 was amplified with primers J174D and JhisR (for wild-type spoIIIJ) and primer pairs J174D with JC134A_R, and JC134A_D with JhisR, being the last two products joined by PCR (SOE) with the external primers (for the spoIIIJ C134A mutant). The final PCR products were digested with BglII and SphI and ligated to pDH88 [30] similarly digested, generating pFiV2 and pFiV3, respectively. Both pFiV2 and pFiV3 were digested with EcoRI and BamHI, the inserts were recovered and ligated to pDG1664 [31] similarly digested to produce pFiV4 and pFiV5, respectively; these plasmids were inserted at the non-essential locus thrC of JOB44 [16] through a double recombination event, producing AH5425 (DspoIIIJ DthrC::P spac -spoIIIJ-his 6 ) and AH5426 (DspoIIIJ DthrC::P spac -spoIIIJ C134A -his 6 , respectively. The two strains were transformed with DyycR::P sspE -cfp from AH6566 [24] (constructed with DNA from BTD2633 (kindly provided by D. Rudner) whilst selecting for Cm r , resulting in AH5431 and AH5432, respectively (Table S1). JOB44 was transformed with DNA of BTD2633 resulting in AH5433 (DspoIIIJ DyycR::P sspE -cfp).
To construct pLC138, two PCR products were obtained with primer pairs PYqjG-460D with YA50C_R, and YA50C_D with YqjG-His-R, which were joined by PCR (SOE) using the external primers. Both the final PCR product and pLC111 (Côrte and Henriques, unpublished data) were digested with BamHI and ligated. To construct the yqjG A50C/C142A -his 6 derivative, primers PYqjG-460D with YC142A_R were used to amplify the first half of the gene from pLC138, and primers YC142A_D with YqjG-His-R for the second half, using pLC115 (Côrte and Henriques, unpublished data) as a template. The products were joined by PCR (SOE) with the external primers, digested with BamHI and cloned into pLC111 similarly digested, resulting in pLC155. Transformation of JOB44 with pLC155 resulted in AH5382 (DspoIIIJ DamyE::P yqjG -yqjG A50C/C142A -his 6 ).
The whole-cell extract was separated in a soluble and a membrane fraction by ultracentrifugation (100,0006g, 45 min, at 4uC). Membrane proteins were extracted from the pellet with 2% dodecyl maltoside (DDM), 20 mM Tris-HCl pH 7.6, 20% glycerol, 0.5 M NaCl and 10 mM imidazole, for 1 h on ice, and again ultracentrifuged (as above). Purification proceeded using a Ni 2+ -NTA affinity column (Qiagen, Hilden, Germany) previously equilibrated with a buffer composed of 0.5 M NaCl, 0.1% DDM 20 mM Tris-HCl pH 8 with 20 mM imidazole. The column was washed five times with the same buffer. Bound proteins were was eluted with a similar buffer containing 25 mM, 50 mM or 100 mM imidazole.

BN-PAGE
Blue-native polyacrylamide gel electrophoresis (BN-PAGE) was performed using gradient gels (5-15%) as described previously [33]. An elution fraction of 9 mg of His-tagged purified SpoIIIJ containing 100 mM imidazole and 500 mM NaCl was used. HMW-Native (Amersham Biosciences) was used as a molecular weight marker.

Fluorescence microscopy
Samples (0.6 ml) of DSM cultures were collected at 4 and 6 h after the onset of sporulation, from cultures with (0.5 mM) or without IPTG, and resuspended in 0.2 ml of phosphate-buffered saline (8 mM sodium phosphate [pH 7.5], 150 mM NaCl) supplemented with 10 mg ml 21 of the lipophilic membrane dye FM4-64 (Molecular Probes) and with 0.2 mg ml 21 of the DNA dye 4-,6-diaminodino-2-phenylindole (DAPI). The cells were then observed by fluorescence microscopy [34,35]. For the quantitative analysis of P sspE -cfp expression at least 200 cells were scored for the fluorescence patterns designated by low (class a) or high (class b). ImageJ (http://imagej.nih.gov/ij/) was used for quantification of the fluorescence signal. Low and high fluorescence refers to the intensity of the fluorescent signal as determined by a threshold defined with ImageJ, using s G -inactive cells and s G -active cells as references.
MalPEG labelling of B. subtilis SpoIIIJ-His 6 Cells grown in LB were collected by centrifugation, resuspended in 10 mM Tris-HCl pH 7.4, 1 mM EDTA, 1 mM PMSF, 1 mM Tris(2-carboxyethyl)phosphine (TCEP) and lysed with a French pressure cell (19 000 lb/in 2 ). Membranes were isolated by a 60min centrifugation at 100,0006 g, resuspended in 50 mM Tris-HCl pH 6.8, 1 mM TCEP and further incubated either at 37uC or 80uC, with or without methoxypolyethylene glycol 5000 maleimide (malPEG), in the presence of 6 M urea and SDS (1 or 2%) or with no SDS nor urea. Other conditions for labelling of SpoIIIJ-His 6 were as follows: 10% SDS; 2% SDS and 8 M Urea; 1% Triton and 4 M GdnHCl. Note that GdnHCl and SDS form a precipitate and could not be used in conjunction for this analysis.

Whole-cell lysates and immunoblot analysis
B. subtilis strains were grown in DSM and samples collected one hour before the end of the growth phase (defined as the onset of sporulation, or T 0 ), and two and four hours thereafter. Cells were resuspended in 10 mM Tris-HCl pH 7.4, 1 mM EDTA, 1 mM PMSF, 0.1 mM dithiothreitol (DTT) and lysed with a French pressure cell as above. Proteins (30 mg) were electrophoretically resolved on 12.5% polyacrylamide gels (SDS-PAGE) [36], transferred to nitrocellulose membranes and subjected to immunoblot analysis with a mouse anti-His 6 antibody (Novagen, Darmstadt, Germany) for the detection of SpoIIIJ-His 6 and YqjG-His 6 . The proteins were visualised with the ECL detection system (Amersham Biosciences) as described by the manufacturer.

A role for Cys134 in the oligomerisation of SpoIIIJ
We purified SpoIIIJ with a C-terminal His 6 tag, SpoIIIJ-His 6 , from cells of E. coli C43(DE3). SpoIIIJ-His 6 was solubilised with DDM from the membrane fraction obtained by ultracentrifugation and further purified over a Ni 2+ -column (Fig. S1A). About 0.8 mg/L of essentially pure SpoIIIJ-His 6 , as judged by SDS-PAGE analysis, was obtained (Fig. S1). SpoIIIJ is a polytopic membrane protein with 5 predicted transmembrane (TM) ahelices and a calculated molecular mass of 27 kDa (Fig. 1A). In agreement, analysis by far-UV circular dichroism indicates that the purified and detergent-solubilised SpoIIIJ-His 6 corresponds to a folded a-helical rich protein (Fig. S2). Under reducing SDS-PAGE conditions, the most abundant form of the protein migrates with an apparent mass of 23 kDa (Fig. 1B) (see also [16]). However, higher molecular weight forms of SpoIIIJ-His 6 were detected following Coomassie staining of SDS-PAGE gels, in the absence of the reducing agent DTT (Fig. 1B). Multimeric forms of SpoIIIJ-His 6 have been observed before in the membrane fraction of E. coli cells over-producing this protein [24]. The high molecular weight forms of SpoIIIJ-His 6 in Fig. 1B, have an apparent size of 30 and 45 kDa (Fig. 1B, bands ''a'' and ''b'', respectively; these are indicated by red arrows). Immunoblotting with an anti-His 6 antibody did not reveal additional forms ( Fig. 1B; shorter forms of the protein are discussed below). This suggested that disulphide bonds between two cysteine residues of SpoIIIJ could be responsible for the formation of the 30 and 45 kDa forms.
In addition to the higher molecular weight forms of SpoIIIJ-His 6 visualised in Fig. 1B (bands ''a'' and ''b''), species migrating closely together at around 15 kDa were also seen, particularly by immunoblot analysis (Fig. 1B). The apparent molecular weight of band ''a'' is compatible with a dimer of full-length mature SpoIIIJ (that is, cleaved by signal peptidase II). Band ''b'' could, in turn, represent a heterodimer formed by full-length mature SpoIIIJ-His 6 and one of the species found around the 15 kDa region of the gel. To test this hypothesis, bands ''a'' and ''b'' were excised from an SDS-PAGE gel and re-run separately. Starting with band ''a'', both ''a'' and monomeric SpoIIIJ-His 6 were detected by immunoblot analysis (Fig. 1C, red and black arrows, respectively). Thus, band ''a'' seems to be composed of two full-length mature SpoIIIJ monomers. Starting with band ''b'', both ''b'' and monomeric SpoIIIJ-His 6 were detected, along with this band, ''c'', that seems to correspond to the slower migrating species of the doublet presumably resulting from processing of SpoIIIJ-His 6 migrating just above the 15 kDa region of the gel ( Fig. 1B and C). N-terminal sequencing of these two species of about 15 kDa, indicates that both start at residue L122 (Fig. 1B). These species are possibly generated by proteolytic cleavage of SpoIIIJ-His 6 at L122, located in the cytoplasmic loop connecting TM1 and TM2, and removing TM1 (Fig. 1A). Cleavage at this position would generate fragments of 11 kDa (from Cys23 to part of the first cytoplasmic loop (including TM1), fragment ''d'' in Fig. 1A) and 16 kDa (the remainder of the first cytoplasmic loop to the end of the protein (including TM2), fragment ''c'' in Fig. 1A). The faster migrating species is not generated by a second cleavage event close to C-terminal of the larger species, as both are detected with an antibody against the C-terminally located His 6 tag (Fig. 1B). Possibly, the difference in mobility of the doublet is caused by some other modification of the proteolytic product of SpoIIIJ-His 6 . In any event, these observations support the idea that SpoIIIJ-His 6 purified from E. coli forms dimeric species. Moreover, they suggest that SpoIIIJ-His 6 can form a heterodimer with a form of the protein lacking TM1.
SpoIIIJ-His 6 has two Cys residues. One is located in the lipoprotein-type signal peptide and is modified during insertion of the protein into the membrane (Côrte and Henriques, unpublished data). A second, unmodified in mature SpoIIIJ-His 6 , is Cys134, located in TM2 (Fig. 1A). Cys134 in SpoIIIJ is conserved among orthologues of the protein, as shown in the alignment of Fig. S3A. A cysteine residue is found at the homologous position of YqjG (Cys142), the other member of the YidC/Oxa1/Alb3 family found in B. subtilis, and also in the E. coli YidC protein. As shown with several membrane topology prediction programs (see Materials and Methods), this residue seems to be always located within a TM segment (Fig. S3A). A helical wheel projection of the second TM segment of SpoIIIJ (Fig. S3B) showed it is a typical hydrophobic helix [37,38]. This residue was the likely candidate for formation of disulphide bonds between SpoIIIJ monomers. To test this, we overproduced and purified a C-terminal His-tagged variant of SpoIIIJ in which Cys134 was substituted by an alanine (Fig. S1B). In contrast to the wild-type, no higher molecular weight bands of SpoIIIJ C134A -His 6 were detected either by Coomassie-staining or immunoblotting of SDS-PAGE gels in the presence or absence of DTT (Fig. 1B). Far-UV CD analysis showed that the C134A substitution is not deleterious for the SpoIIIJ fold, as the spectrum obtained for this purified and detergent solubilised variant is essentially identical to that of the wild-type protein ( Fig. S2A and B). Nevertheless, whereas reduction of SpoIIIJ-His 6 with DTT yields minor secondary structure interconversions compatible with structural rearrangements upon reduction of the presumed C134-C134 disulphide, the same is not observed in the case of SpoIIIJ C134A -His 6 (Fig. S2C).
As an independent test of oligomer formation by SpoIIIJ-His 6 , we used blue-native (BN) PAGE and size exclusion chromatography (SEC). By BN-PAGE, three species were observed that exhibited molecular weights compatible with a monomer, a dimer (the most represented species) and traces of a possible hexamer ( Fig. 2A). SEC of purified SpoIIIJ-His 6 revealed two main peaks (Fig. 2B) both of which contained the protein (as show by the immunoblot of the peak fractions; insert in Fig. 2B). The larger peak corresponds to a species with a calculated size (75 kDa) closer to that of a SpoIIIJ-His 6 dimer (Fig. 2B). The smaller peak is caused by a species with a calculated size (47 kDa) closer to that of monomeric SpoIIIJ-His 6 (Fig. 2B). The differences in these sizes from that estimated from BN-PAGE analysis probably results from the presence of dodecyl maltoside (DDM), the amount of which we were unable to estimate by SEC-MALLS (data not shown). Peaks with the same estimated sizes were also observed for SpoIIIJ C134A -His 6 but in this case, the size of peak B, corresponding to the smaller species, increased relative to peak A (Fig. 2B). Quantification of the peak areas shows ratios of B to A of 1:9 for SpoIIIJ-His 6 and 1:2.5 for SpoIIIJ C134A -His 6 .
Formation of the dimer size species (peak A) does not require Cys134 but taken all together our results suggest that Cys134 stabilises a multimeric (possibly a dimeric) form of SpoIIIJ-His 6 , in line with the suggestion that this residue is involved in formation of a disulphide bond between two monomers of the protein.

Cys134 of SpoIIIJ is part of a protein-protein interface in B. subtilis
Cys134 is thought to be located in the plane of the membrane (Fig. 1A and S3), and involved in formation of a disulphide bond between SpoIIIJ-His 6 monomers (Fig. 1). We wanted to examine whether Cys134 is located within the membrane and if it forms part of a protein-protein interaction interface. The membrane fraction of whole cell lysates prepared from B. subtilis cultures was isolated by ultracentrifugation and the available Cys residues in the proteins present in the sample were labelled with the Cysmodifying reagent methoxypolyethylene glycol 5000 maleimide (malPEG for simplicity). This reagent binds free sulfhydryl groups in cysteine residues forming thioether bonds, whilst adding an extra 5 kDa per available cysteine residue. As malPEG induces distortion in gel migration, the minimum concentration required to obtain appreciable labelling was determined and found to be 1 mM (data not shown). As malPEG is membrane-impermeable, SDS was also used to solubilise the membranes. If the cysteine is in the membrane plane, labelling was expected only in the presence of SDS. To increase the availability of free sulfhydryl groups by reducing eventual disulphide bonds, Tris(2-carboxyethyl)phosphine (TCEP) was used. Unlike DTT, this reagent does not compete with the cysteine residues for malPEG. As a positive control for labelling we used a substituted form of YqjG in which a Cys residue was introduced at position 50. As for the highly similar SpoIIIJ (Fig. 1A), this position is predicted to be located outside the cytoplasmic membrane, and facing the extracellular space (Fig. 3A). This residue should be more readily labelled with malPEG than a residue located in the plane of the membrane. Cys142 of YqjG (Fig. 3A) is homologous to Cys134 of SpoIIIJ and was substituted by an alanine to facilitate interpretation of the labelling pattern. Labelling of SpoIIJ with malPEG was not detected at 37uC even in the presence of 2% SDS (Fig. 3B) or at higher concentrations of the detergent (not shown). The level of SpoIIIJ in mixtures incubated at 80uC in the absence of SDS is strongly reduced (Fig. 3C, lanes 6 and 8), because under these conditions the protein failed to enter the gel (not shown). The combined action of detergents and denaturants has been shown to disclose normally hidden residues [39,40]. However, none of the combinations herein tested (GdnHCl, urea, Triton and SDS), result in labelling of Cys134 of SpoIIIJ-His 6 (see Materials and Methods). However, SpoIIIJ-His 6 was labelled at 80uC in the presence of 2% SDS (Fig. 3C). Thus, high temperature and the presence of a detergent are required for labelling Cys134 of SpoIIIJ-His 6 with malPEG. In contrast, labelling of the positive control YqjG A50C/C142A -His 6 was obtained in the presence of 1% SDS and also in its absence ( Fig. 3B and C). We observed a shift in the apparent molecular weight for malPEG-derivatised YqjG A50C/ C142A -His 6 larger than the expected 5 kDa, which has also been reported previously [41].
Together, these results are compatible with the view that Cys134 is located within the membrane plane, and is part of a tight protein-protein interaction surface.
Accumulation of wild-type and substituted SpoIIIJ during sporulation in B. subtilis spoIIIJ is required for sporulation [10]. To evaluate the ability of SpoIIIJ C134A to support efficient sporulation, both the wild-type and C134A form of SpoIIIJ were expressed from the IPTGinducible P spac promoter, placed at the non-essential thrC locus in a spoIIIJ deletion background (Fig. 4A). When measured 24 hours after the onset of sporulation in DSM, the non-induced C134A strain produced very low levels of heat-resistant spores when compared to its wild-type counterpart (Table 1). Importantly, the level of heat-resistant spores for the non-induced wild-type allele was close to that found for a wild-type strain (i.e., expressing spoIIIJ from its normal locus) ( Table 1). In contrast, upon addition of IPTG both the wild-type and the C134A mutant exhibited sporulation levels similar to a wild-type strain ( Table 1). The fact that a low spore titre was observed for the non-induced mutant but not for the non-induced wild-type protein, suggested that under these conditions SpoIIIJ C134A -His 6 accumulates to lower levels or is less functional than the wild-type form. To distinguish between these possibilities, we analysed the levels of either protein under inducing and non-inducing conditions. When induced, both the wild-type and C134A forms of SpoIIIJ accumulate to levels close to those of the protein produced from its native locus (Fig. S4). When non-induced, the levels of SpoIIIJ or SpoIIIJ C134A are both barely detectable, but similar to one another (Fig. 4B). Hence, it seems unlikely that different expression levels account for the difference in spore titre observed for the two forms of the protein ( Table 1). Note that a species with the expected size for a dimer is seen for both SpoIIIJ wt -His 6 and SpoIIIJ C134A -His 6 (Fig. 4B, red arrows). The representation of this species did not increase when the experiment was repeated in the absence of DTT, suggesting that Cys134 is not sensitive to the reducing agent (data not shown), in agreement with the idea that this residue is part of a tight protein-protein interaction surface ( Fig. 3; above). Because the level of heat-resistant spores for the non-induced wild-type allele was close to that found for a wild-type strain (Table 1, see also above), the data shows that low levels of wild-type SpoIIIJ are sufficient for efficient sporulation. Together, these results suggest that SpoIIIJ C134A is less functional than the wild-type protein, and that this reduced functionality emerges at low expression levels of the corresponding alleles.

Requirement of Cys134 of SpoIIIJ for s G activity
Since SpoIIIJ is required for late forespore-specific gene expression during sporulation [10], we wanted to assess the impact of the C134A substitution on the activity of s G . We used the s G -controlled transcriptional fusion P sspE -cfp [23] to monitor the activity of s G in single cells by fluorescence microscopy (Fig. 4C). DSM cultures of strains expressing either spoIIIJ or spoIIIJ C134A from P spac in a spoIIIJ mutant background were sampled at hours 4 and 6 after the onset of sporulation. The cells were observed by fluorescence microscopy after staining with the membrane dye FM4-64 and the nucleoid stain DAPI (see Materials and Methods), to identify and score the morphological stage of sporulation. Sporulating cells were assigned to two classes according to the intensity of the CFP signal: with a low (a) and with a high (b) fluorescence signal (Fig. 4C). High-fluorescent cells completed engulfment and activated s G , thus leading to high expression of the reporter fusion P sspE -cfp; on the other hand, low fluorescence mainly results from the weaker utilization of the sspE promoter by the preceding sigma factor, s F , in the forespore line of gene expression [24].
For the non-induced strain bearing the P spac -spoIIIJ allele 23.4% of the cells at hour 4 of sporulation, and 23.5% of the cells at hour 6 were scored as highly fluorescent (class b), an indication of active s G ; in contrast, the P spac -spoIIIJ C134A -expressing strain under non-inducing conditions showed a large proportion of lowfluorescence cells (class a; 38.1% at hour 4 and 58.8% at hour 6), whilst the high-fluorescence cells were scarce (class b; 6.8% at hour 4 and 11.6% at hour 6) (Fig. 4C). A distribution similar to that of the non-induced P spac -spoIIIJ C134A strain was obtained for a spoIIIJ null mutant (not shown), as expected, as in these cells only s F is active [10]. Upon addition of IPTG (to 0.5 mM), similar percentages of highly fluorescent, class b cells were obtained for the strains expressing either spoIIIJ C134A (26.8% at hour 4 and 34% at hour 6) or its wild-type counterpart (23.6% and 41.3% at hour 6, respectively) (Fig. 4C). Together, the results show that expression of spoIIIJ C134A does not cause a delay in sporulation. Rather, the results suggest that the C134A substituted form of SpoIIIJ is less active than the wild-type protein, but the lower activity of spoIIIJ C134A is evident in vivo only when lower than normal levels of the proteins is produced.

Discussion
B. subtilis SpoIIIJ, a member of the YidC/Oxa1/Alb3 family of insertases, forms a dimer. The Cys134 residue, predicted to reside within the TM2 segment of SpoIIIJ, has a role in dimer formation, presumably by establishing a disulphide bond. Several lines of evidence suggest that formation of an intramembrane disulphide bond, involving Cys134 promotes formation of, or stabilises a SpoIIIJ dimer. SpoIIIJ was detected as a monomer and a dimer by SDS-PAGE, but only a monomeric species was seen in the presence of the reducing agent DTT (Fig. 5A). In contrast, SpoIIIJ C134A , in which the only Cys residue in mature SpoIIIJ was changed to an Ala, migrated as a monomeric species, even in the absence of DTT (Fig. 5A). Also, the monomer:dimer ratio, as estimated by SEC, was higher for SpoIIIJ C134A in comparison to the wild-type protein. Circular dichroism spectroscopy analysis suggested that DDM-solubilised SpoIIIJ is a folded, mainly ahelical protein, and we found that the addition of DTT caused an alteration in the content of a-helices and anti-parallel b-sheets for SpoIIIJ but not for SpoIIIJ C134A . This change in secondary structure may be explained, at least in part, by a change in the monomer:dimer ratio. Clearly, despite its importance, Cys134 is not absolutely required for dimer formation. Rather, the dimer appears to be sustained by a disulphide bond along with other non-covalent interactions. Oligomerisation is a widespread mechanism for regulating protein function [42]. For SpoIIIJ, dimer formation is important, at least for sporulation (see also below), and an inter-chain disulphide bond may be advantageous as it might stabilise the functional form of the protein. More generally, inter-chain disulphide bonds may help coping with naturally occurring mutations that reduce the potential for dimerisation or oligomerisation [43]. The Cys134 residue is predicted to occur within the membrane plane, in TM2. Conservation of a Cys residue at a homologous position in the YqjG paralogue, and among SpoIIIJ orthologues, underscores the importance of this residue, and presumably of disulphide bond formation for optimal functionality.
It is often assumed that the membrane environment does not favour formation of disulphide bonds, but an increasing number of reports argue in favour of the existence of such intramembrane disulphide bonds [e.g., 44,45,46,47]. The experiments using malPEG were designed to test whether a free cysteine residue was accessible for modification in SpoIIIJ. We have found that labelling with malPEG only took place by exposing the protein to high temperature in the presence of high concentrations of SDS in (Fig. 3). The harsh conditions required for labelling of Cys134 suggest that this residue is present at a tight interface, as the bulky malPEG molecule may have only limited access to the cysteine residue [39]. This may contribute to the incomplete labelling observed (Fig. 3). Since the labelling experiments were implemented with extracts prepared from B. subtilis cultures, another factor contributing for incomplete labelling could be the existence of a mixed population of SpoIIIJ in several oligomeric states and/ Strains producing SpoIIIJ-His 6 or YqjG A50C/C142A -His 6 in a spoIIIJ mutant background were grown in liquid LB, and samples withdrawn. Cells were resuspended in a buffer containing 1 mM TCEP, lysed and membranes isolated. The membranes were resuspended in the presence of 1 mM TCEP and further incubated with or without malPEG in the presence of 2% SDS (lanes 1-2), 1 or 0% SDS (lanes 3 and 4, respectively). Proteins (30 mg) were electrophoretically resolved by SDS-PAGE and immunoblotted with an anti-His 6 antibody for the detection of SpoIIIJ-His 6 and YqjG A50C/C142A -His 6 . (C) Samples from cultures of strains producing SpoIIIJ-His 6 or YqjG A50C/C142A -His 6 were withdrawn at the onset of the stationary growth phase in LB medium. Samples were treated as in (B), except that no TCEP was added to the French press buffer. Incubation proceeded at the indicated temperatures in the presence of 1 mM TCEP and with or without 1 mM malPEG and SDS, as indicated. In both B and C, grey and black arrows indicate SpoIIIJ-His 6 or YqjG A50C/C142A -His 6 or labeled (shifted) reaction products, respectively. The position of molecular weight markers (in kDa) is shown. doi:10.1371/journal.pone.0099811.g003 or belonging to distinct protein complexes allowing different degrees of malPEG accessibility.
Distinct oligomeric states have been reported for members of the YidC/Oxa1/Alb3 family of membrane protein insertases. Oxa1 from Neurospora crassa was suggested to form tetramers [48] and Oxa1 from S. cerevisiae was observed to form only dimers [9], only tetramers [49], or dimers and tetramers [50]. Alb3 from Arabidopsis thaliana is able to form dimers [51]. Finally, YidC from E. coli was reported to form dimers [9,52,53,54] and was also purified as a mixture of monomers and dimers [55,56]. We cannot exclude that forms of SpoIIIJ of order higher than the dimer are formed and have physiological relevance. The BN-PAGE, for examples, suggests that a SpoIIIJ hexamer can be formed, and at least part of the dimer population could serve as an intermediate for the formation of this species. However, we cannot presently exclude that the hexamer is only formed due to overproduction of the protein in E. coli. A role for TM1 of SpoIIIJ (corresponding to TM2 of YidC) in SpoIIIJ-SpoIIIJ interactions is hinted from the fact that a heterodimer composed of a full-length, mature, and a truncated SpoIIIJ monomer (lacking TM1) was observed but not a homodimer of two truncated SpoIIIJ monomers (Fig. 1). This role for TM1 may be direct or indirect. TM1 could itself be involved in protein-protein interactions, or it could maintain the regions directly interacting (possibly TM2, with C134) in the correct position (Fig. 5B). Interestingly, TM2 of YidC (corresponding to SpoIIIJ's TM1) was suggested to have a structural role as well, in helix-helix interactions [57]. In addition, an interaction between TM2 and 3 of YidC (corresponding to TM1 and 2 of SpoIIIJ, respectively) was also suggested [58]. Importantly, a recent study by Geng and co-authors in which a series of SpoIIIJ/ YqjG chimeras were analysed [59], also highlights the critical role of TM2 of SpoIIIJ (containing Cys134) for its role during sporulation.
Both SpoIIIJ and YidC function together with or independently of the Sec system [reviewed in 3,60]. For instance, SpoIIIJ participates in the biogenesis of the SpoIIIAE polytopic membrane protein, one of the components of the secretion system required for normal s G activity after engulfment completion, possibly in conjunction with the Sec pathway [20,24]. Depending on either mode of action, in conjunction with or independently of the Sec pathway, SpoIIIJ may function mostly as a monomer or mostly as a dimer. The precise role fulfilled by SpoIIIJ during membrane protein biogenesis could also vary regarding different substrates, either in guiding membrane protein insertion or later, in the folding or the assembly of membrane protein complexes [6]. Our observation that at low expression levels, SpoIIIJ C134A but not the wild-type protein, shows a reduced ability to support s G activity and sporulation ( Fig. 4 and Table 1), may suggest that a higher concentration of the mutant protein is needed to promote dimer formation, as the disulphide bond cannot be formed. Dimeric (or oligomeric) SpoIIIJ may be the form of the protein that best functions in the assembly of the SpoIIIA-SpoIIQ-based secretion system.
The basis for the observation that SpoIIIJ but not its paralogue YqjG supports efficient sporulation is not fully understood. Since a positional homologue of Cys134 is found in YqjG, it seems possible that this residue is also involved in disulphide bond formation in this protein. If so, formation of a disulphide bond between proximal TM2 segments does not distinguish SpoIIIJ and YqjG with respect to sporulation. Other regions of SpoIIIJ may be important, as a class of suppressor mutations that permit s G activation and sporulation in the absence of spoIIIJ map to TM5 and to the loop between TM4 and TM5 of yqjG [20]. Figure S1 Overproduction and purification of SpoIIIJ-His 6 and SpoIIIJ C134A -His 6 . Overproduction and purification of SpoIIIJ-His 6 (A) or SpoIIIJ C134A -His 6 (B) from E. coli strain C43(DE3) carrying pMS266 or pFV1, respectively. The cells were  Addition of DTT affects the spectrum of SpoIIIJ-His 6 , but has no effect on the spectrum of SpoIIIJ C134A -His 6 . (C) Relative estimates of the secondary structure of SpoIIIJ-His 6 (with or without 1 mM DTT, dark grey and black bars, respectively) and SpoIIIJ C134A -His 6 (with or without 1 mM DTT, white and light grey bars, respectively). Addition of DTT decreases the a-helix content by 6% and increases the content of antiparallel b-sheets (more 8%) of SpoIIIJ-His 6 , but does not significantly alter SpoIIIJ C134A -His 6 .

Supporting Information
(TIF) Figure S3 Conservation of Cys134 among SpoIIIJ orthologues. (A) Alignment of the transmembrane (TM) segment 2 of SpoIIIJ (red) and YqjG (green) proteins from several Bacillus species and of TM3 of YidC (blue) from E. coli. Conserved residues are shaded in grey except for the cysteine (yellow). The sequences were aligned with ClustalW [61]. (B) Helical wheel projection of TM2 of SpoIIIJ from B. subtilis (http://rzlab.ucr. edu/scripts/wheel/wheel.cgi). Circles denote hydrophilic residues and diamonds hydrophobic ones. The color code is as follows: green, hydrophobic residues, with the amount of green decreasing proportionally to the hydrophobicity; yellow, zero hydrophobicity; red, the most hydrophilic (uncharged) residue, the amount of red decreasing proportionally to the hydrophilicity. (TIF) Figure S4 Expression of SpoIIIJ-His 6 and SpoIIIJ C134A -His 6 in B. subtilis. The figure compares the levels of SpoIIIJ-His 6 and SpoIIIJ C134A -His 6 expressed from the thrC locus under the control of P spac , with the level of SpoIIIJ-His 6 expressed from the spoIIIJ locus under the control of its normal promoter (identified by the ''*'' symbol). The cultures were grown in liquid DSM in the presence (for the fusions at thrC) or in the absence of IPTG (for the fusion at the spoIIIJ locus), and samples withdrawn at hours 4 and 6 of sporulation. Proteins in whole-cell extracts were subject to immunoblot analysis with an anti-His 6 (upper panel) or an anti-s A antibody (lower panel). The gels were run under reducing conditions. The positions of SpoIIIJ-His 6 (wt or C134A) and s A are indicated by black arrows; red arrows show the position of presumed SpoIIIJ dimers. The position of molecular weight markers (in kDa) is shown. Note that the SpoIIIJ-His 6 * fusion, expressed from the spoIIIJ locus under the control of its normal promoter, has a linker longer than the fusions expressed under the control of P spac at thrC locus, and shows a slightly higher apparent mass [24]. (TIF)