Vectorial signalling mechanism required for cell–cell communication during sporulation in Bacillus subtilis
Instituto de Biología Molecular y Celular de Rosario (IBR) and Departamento de Microbiología, Facultad de Ciencias Bioquímicas y Farmacéuticas, Universidad Nacional de Rosario, 2000 Rosario, Argentina.
Instituto de Biología Molecular y Celular de Rosario (IBR) and Departamento de Microbiología, Facultad de Ciencias Bioquímicas y Farmacéuticas, Universidad Nacional de Rosario, 2000 Rosario, Argentina.
Instituto de Biología Molecular y Celular de Rosario (IBR) and Departamento de Microbiología, Facultad de Ciencias Bioquímicas y Farmacéuticas, Universidad Nacional de Rosario, 2000 Rosario, Argentina.
Spore formation in Bacillus subtilis takes place in a sporangium consisting of two chambers, the forespore and the mother cell, which are linked by pathways of cell–cell communication. One pathway, which couples the proteolytic activation of the mother cell transcription factor σE to the action of a forespore synthesized signal molecule, SpoIIR, has remained enigmatic. Signalling by SpoIIR requires the protein to be exported to the intermembrane space between forespore and mother cell, where it will interact with and activate the integral membrane protease SpoIIGA. Here we show that SpoIIR signal activity as well as the cleavage of its N-terminal extension is strictly dependent on the prespore fatty acid biosynthetic machinery. We also report that a conserved threonine residue (T27) in SpoIIR is required for processing, suggesting that signalling of SpoIIR is dependent on fatty acid synthesis probably because of acylation of T27. In addition, SpoIIR localization in the forespore septal membrane depends on the presence of SpoIIGA. The orchestration of σE activation in the intercellular space by an acylated signal protein provides a new paradigm to ensure local transmission of a weak signal across the bilayer to control cell–cell communication during development.
How cells communicate with each other is a fundamental biological question relevant to the development of multicellular organisms, microbial pathogenesis and communities of microorganisms. In the bacterium Bacillus subtilis, cell–cell signalling pathways play a role in co-ordinating gene expression during the process of spore formation. Upon the initiation of sporulation, the developing cell divides asymmetrically, generating two cells of unequal size and dissimilar fate: a small cell (the prospective spore), referred to as the forespore (also called the prespore) and a large cell called the mother cell (Stragier and Losick, 1996; Hilbert and Piggot, 2004). Initially, these two cells lie side by side, but shortly after polar division the mother cell membranes migrate around the forespore in a phagocytic-like process known as engulfment, generating a cell within a cell. The mother cell is necessary for spore formation, but ultimately lyses (Grossman, 1995). During this developmental process, the mother cell and the forespore follow separate programmes of gene expression directed by alternative sigma factors. However, cell–cell signalling pathways ensure that gene expression in one cell is co-ordinated with gene expression in the other (Losick and Stragier, 1992; Stragier and Losick, 1996; Rudner and Losick, 2001). After polar division, the first cell type-specific transcription factor, σF, is activated in the forespore (Stragier and Losick, 1996; Kroos and Yu, 2000; Hilbert and Piggot, 2004). This in turn activates expression of spoIIR, initiating an intercompartmental signalling cascade that activates the mother cell-specific regulator σE, causing mother cell differentiation (Karow et al., 1995; Hilbert and Piggot, 2004). A striking feature of SpoIIR is that it is an intercellular signalling protein (Londoño-Vallejo and Stragier, 1995) that induces the conversion of pro-σE, the product of the spoIIGB gene, to its active form by proteolytic removal of its pro-amino sequence, an N-terminal extension of 27 residues (LaBell et al., 1987; Stragier et al., 1988). SpoIIR is synthesized in the prespore (Karow et al., 1995), and is believed to be exported to the inter-membrane septal space where it is thought to activate the membrane-associated pro-σE processing enzyme, SpoIIGA. Consistent with SpoIIR acting as a cell–cell signal, characterization of the spoIIR gene indicated that it encoded a protein with a secretion signal (Karow et al., 1995) and it was reported that conditioned medium from bacteria engineered to express spoIIR could trigger processing of pro-σE to σE when added to protoplasts expressing the spoIIG operon products, SpoIIGA and pro-σE (Hofmeister et al., 1995). Thus, the main function of SpoIIR would be to tie the transcription programme in the mother cell to the transcription programme in the forespore by allowing σE to become active only after σF has itself been activated (Stragier and Losick, 1996).
The mechanism by which information is transduced across the lipid bilayer by SpoIIR and SpoIIGA is not yet known. One intriguing property of this cell–cell signalling pathway is that it is stringently dependent on concomitant lipid synthesis. In an earlier work, we showed that de novo lipid synthesis at the onset of sporulation is required for spore formation and that reduced levels of lipid synthesis allow formation of the asymmetric septum but prevent activation of σE (Schujman et al., 1998). We speculated that mother cell septal membrane phospholipids play a key role on the activity, assembly or localization of the SpoIIGA protein or in the cell–cell signalling mediated by SpoIIR.
Here we present evidence for a new model in which σE activation is tightly coupled to fatty acid synthesis, rather than to phospholipid synthesis. First we show that the proteolytic processing and secretion of SpoIIR in the interseptal space is strictly dependent on the activity of the type II fatty acid synthase. Second, we demonstrate that a conserved threonine (T27) is important for cleavage of the SpoIIR signal sequence. Finally, we show that septal localization of SpoIIR is dependent on SpoIIGA. All these results are consistent with the idea that SpoIIGA is essential to spatially restrict SpoIIR signalling to the septum and that the SpoIIR pre-protein requires a particular fatty acid modification in T27 to be processed by a specific signal peptidase. This model provides a new paradigm to ensure local transmission of a weak signal across the bilayer to control cell–cell communication during development.
Processing of pro-σE to σE requires fatty acid synthesis but is independent of phospholipid synthesis
Previous experiments indicated that the fungal toxin cerulenin is a potent inhibitor of processing of pro-σE to its active form (Schujman et al., 1998). This inhibition is accomplished through the ability of cerulenin to bind to and irreversibly inactivate B. subtilis FabF, the enzyme that catalyses a key step in fatty acid biosynthesis. Importantly, cessation of fatty acid synthesis also inhibits the production of phosphatidic acid (PtdOH), the precursor of membrane phospholipids (Fig. 1). The fact that cerulenin treatment blocks these two essential biosynthetic pathways (Fig. 1) does not allow distinguishing if processing of pro-σE is affected by inhibition of fatty acid or phospholipid synthesis. In the latter pathway (Fig. 1) PlsX initially catalyses the synthesis of the intermediate fatty acyl-phosphate (acyl-PO4). Next, PlsY transfers the fatty acid from the activated acyl intermediate to the 1 position of glycerol-3-PO4. Finally, acyl-glycerol-3-PO4 (acyl-G3P) is acylated to PtdOH by PlsC. PlsX is a key co-ordinator of membrane lipid synthesis and its inactivation leads to cessation of both fatty acid and phospholipid synthesis (Paoletti et al., 2007). On the other hand, depletion of PlsY and PlsC results in a profound inhibition of phospholipid synthesis, although the de novo fatty acid synthesis continues, leading to their accumulation as a free species (Paoletti et al., 2007). To know whether the action of cerulenin on pro-σE processing was before or after the synthesis of PtdOH, we analysed sporulation efficiencies of strains containing either the plsX, plsY or plsC genes under the control of inducible promoters. As expected, a plsX conditional mutant displayed a severe sporulation deficiency in the absence of inducer (Table 1). In agreement with this observation, the β-galactosidase activity of a σE-dependent PspoIID–lacZ fusion contained in a plsX conditional strain was totally curtailed when PlsX was depleted (Table 1). In contrast, depletion of PlsY or PlsC did not affect sporulation efficiency (Table 1) and the transcription of spoIID–lacZ in plsY or plsC strains was similar to that found in the wild type (Table 1). Therefore, these results suggest that σE activation is coupled to fatty acid synthesis and does not require that newly synthesized fatty acyl groups are incorporated into membrane phospholipids.
Table 1. Sporulation efficiency of strains deficient in phospholipid synthesis.
Conditionally expressed gene
The sporulation frequency was measured by calculating the ratio of the number of heat stable spores to viable cells at t24.
β-Galactosidase activity of the σE-dependent spoIID promoter fused to lacZ was assayed in samples taken at t3. MU, Miller Units.
Fatty acid synthesis is required for secretion of the SpoIIR signal
The SpoIIR protein contains a putative Sec-type signal sequence (Karow et al., 1995), suggesting that SpoIIR may be secreted from the prespore into the intermembrane septal space where it is thought to activate SpoIIGA. Since SpoIIR is an exported protein it is conceivable and even likely that this exoprotein can also be translocated directly into the extracellular environment serving as a means of communication between cells. If this were the case, a cell synthesizing SpoIIR could export this peptide to the extracellular medium and then it could be transferred to the inter-membrane septal space of a cell lacking SpoIIR to promote pro-σE processing. To test this possibility, a mixed culture of two Spo- strains was devised: a ΔspoIIR spoIIGA+ acceptor strain (Fig. 2A) was induced to sporulate in the presence of a spoIIR+ΔspoIIGA donor strain (Fig. 2A). Under these conditions, the sporulation efficiency of the mixed culture was ∼ 500-fold higher than in simple cultures (Fig. 2B, black bars). Thus, SpoIIR produced under the control of σF in a donor prespore was indeed transferred to the intermembrane space of a SpoIIR-deficient cell, stimulating the processing of σE and enhancing its sporulation frequency. This complementation test provides us with a tool to test if σE processing in mixed cultures requires fatty acid synthesis in the SpoIIR-donor cell, in the SpoIIR-acceptor cell or in both partners. For this purpose, we introduced the fabF1 mutation, which codes for a FabF enzyme resistant to inhibition by cerulenin (Schujman et al., 1998; 2001), in a SpoIIR donor or in a SpoIIR acceptor strain. Then, mixed culture experiments were devised in which combinations of SpoIIR donors or SpoIIR acceptors which were sensitive or resistant (CerR) to cerulenin were subjected to sporulation in the presence of the antibiotic (Fig. 2A). The results of these experiments showed that the sporulation efficiency in mixed cultures was only decreased by the antibiotic when the SpoIIR-donor cell was sensitive to cerulenin (Fig. 2B, grey bars).
We conclude that de novo fatty acid synthesis is required for either the synthesis of SpoIIR in the prespore or its transport across the bilayer, and that fatty acid synthesis is dispensable for SpoIIR-mediated proteolytic activation of pro-σE processing in the mother cell.
Fatty acid synthesis is required for SpoIIR processing
If fatty acid synthesis is required for SpoIIR activation, then we would expect that inhibition of the type II fatty acid synthase system would block the synthesis or processing of this signalling protein. We therefore tested SpoIIR synthesis and processing using a functional SpoIIR C-terminally tagged with His6, which is expressed in B. subtilis under the control of the IPTG-inducible hyper spac promoter (Pspac-hy). To examine whether the inhibition of lipid synthesis blocks synthesis or processing of SpoIIR, we performed Western blot analysis, using hexa-His antibodies, of proteins extracted from cells expressing spoIIR-His6 treated with cerulenin at various times following the initiation of sporulation. As shown in Fig. 3A, Western blot analysis of the different cell extracts revealed two species migrating in SDS-PAGE, with apparent molecular masses of 26 and 17 kDa. The slow migrating band displays the expected size (25.1 kDa) and should correspond to unprocessed SpoIIR, whereas the faster migrating species should correspond to mature SpoIIR since it was the predominant band detected in cerulenin-free cultures (Fig. 3A). It is worth mentioning that the expected molecular weight of the processed protein is 23.7 kDa, suggesting an abnormal migration of the mature protein. The overall levels of SpoIIR were almost similar in cells treated or untreated with cerulenin, at least within 36 or 72 min after the addition of the antibiotic; at earlier times (12 min), the level of SpoIIR was substantially decreased (86%) after the addition of cerulenin (Fig. 3A). The reduced quantities of SpoIIR observed at 12 min after the addition of cerulenin (Fig. 3A) could be due to a higher instability of the unprocessed protein. In fact, processing of SpoIIR was totally suppressed in cells that were treated with cerulenin at 12 or 36 min after the onset of sporulation (Fig. 3A) producing only the slow migrating species. Processing of SpoIIR was restored, however, if cerulenin treatment happened 72 min after the onset of sporulation. It should be noted that in the experiments depicted in Fig. 3A we investigated the effect of cerulenin on SpoIIR processing using a construct where SpoIIR is under the control of a heterologous inducible promoter. However, cerulenin also inhibited SpoIIR processing when spoIIR was expressed under its native promoter (Fig. 3B, left panel). Altogether, these results indicate that cells that do not synthesize lipids at the onset of sporulation produced SpoIIR but did not process it, indicating that de novo lipid synthesis is required for SpoIIR processing. Moreover, by comparing the pattern of SpoIIR production and its processing with the sporulation efficiency of each culture (Fig. 3A, bottom, and Fig. 3B, left panel bottom), there is a clear correlation between inhibition of SpoIIR processing and inhibition of sporulation by cerulenin. We conclude that SpoIIR secretion from the prespore into the intermembrane space separating it from the mother cell is tightly coupled to fatty acid synthesis.
Inhibition of fatty acid biosynthesis specifically blocks SpoIIR processing
To provide a solid proof that de novo fatty acid synthesis is specifically required for SpoIIR processing, we engineered and tested a fusion of SpoIIR to the signal peptide of the extracellular α-amylase (AmyE) from B. subtilis (Ohmura et al., 1983). The goal was to determine whether this construct was able to trigger the activation of σE and if its processing was linked to fatty acid synthesis. Thus, we constructed a chimera, named SPAmyE-SpoIIR, where the first 23 residues of the N-terminal domain of SpoIIR were replaced by the 33 residues corresponding to the signal peptide of AmyE (SPAmyE). To examine whether the inhibition of lipid synthesis affects the processing of SPAmyE-SpoIIR, we expressed this chimeric protein under the control of the spoIIR promoter in cells treated or untreated with cerulenin. As shown in Fig. 3B (right), processing of SPAmyE-SpoIIR was unaffected by the addition of cerulenin 36 min after the onset of sporulation. Furthermore, the frequency of sporulation of cells expressing SPAmyE-SpoIIR was essentially the same in the presence or absence of cerulenin added at t0.6 (Fig. 3B, right panel, bottom). Similar results were obtained if the signal peptide of the levansucrase SacC of B. subtilis (Martin et al., 1987) was peptidically linked to SpoIIR lacking the first 23 residues of its N-terminal domain (data not shown).
We conclude that the effect of cerulenin on SpoIIR processing is highly specific for this signalling protein, supporting the notion that the requirement of fatty acid synthesis for SpoIIR processing is a mechanism for integrating the metabolic state of the cell with the developmental commitment to sporulate.
A conserved threonine is required for SpoIIR processing
Figure 3 showed that there is a strong correlation between inhibition of SpoIIR processing and inhibition of sporulation by cerulenin. However, according to previous work by Londoño-Vallejo (1997), processing should not be essential for SpoIIR signalling. This was based on the finding that point mutations at positions −3 (G21V) and −1 (A23M) from the presumed cleavage site of the SpoIIR leader peptide blocked its processing, although fully complemented the sporulation deficiency of spoIIR mutants. In face of these conflicting findings, we substituted glycine 21 for aspartate and alanine 23 by methionine in SpoIIR-His6 expressed under the control of its own promoter. The SpoIIRG21D,A23M mutant strain was induced to sporulate and processing of SpoIIR was assessed by immunoblot analysis with hexa-His antibodies. Processing of this mutant protein was unaffected (Fig. 4) and the mutant protein was able to support wild-type levels of heat-resistant spores (data not shown). Based on these observations, we conclude that G21 and A23 are not required for cleavage of the N-terminal extension. The discrepancy between our results and those of Londoño-Vallejo (1997) could be due to the fact that this author tested the processing of SpoIIR mutant proteins using an indirect approach, based on the secretion of a SpoIIR-levansucrase hybrid protein.
When SpoIIR sequences from nine different Bacillus species were aligned (Fig. S1A) a conserved threonine or tyrosine was found at position 27. In eight of the SpoIIR analysed sequences the most probable cleavage site is located downstream from a conserved threonine or tyrosine. However, in B. subtilis the predicted primary cleavage site (Fig. S1B) is residue A23 that is located upstream of T27 and that is dispensable for SpoIIR processing (Fig. 4). Strikingly, two residues, A28 and K30, lying downstream of the conserved T27 are predicted as secondary cleavage sites (Fig. S1B) by the algorithm SignalP version 3.0 (http://www.cbs.dtu.dk/services/SignalP/) (Bendtsen et al., 2004). To investigate the significance of this, we switched residue T27 to an alanine or to a tyrosine. The mutant protein with amino acid substitution T27A was completely impaired in processing (Fig. 4). However, SpoIIR processing was unaffected by substitution of T27 with Y. As an independent measure of SpoIIR processing, we analysed σE activity in the SpoIIR mutants using a σE-responsive PspoIID fused to lacZ. This assay allowed us to monitor the effect of the mutations throughout sporulation rather than at a particular time point. Consistent with the immunoblot analysis, the amino acid substitutions in the putative primary cleavage site of SpoIIRG21D,A23M expressed under PspoIIR did not interfere with σE activity (Fig. S2). However, a change of tyrosine 27 for alanine completely blocked σE activation, while substitution of the conserved threonine by tyrosine did not alter σE activity.
Altogether, these data indicate that T27 is crucial for the proteolytic conversion of the precursor of SpoIIR to mature SpoIIR, and that A23 is not the cleavage site of the signal peptide. Instead, we propose that, similarly to SpoIIR from other Bacillus species, the cleavage motif is located downstream of T27. We also infer that release of SpoIIR into the interseptal space is indeed important to activate SpoIIGA in the mother cell.
Use of a GFP–SpoIIR fusion to monitor SpoIIR translocation and processing
Given that SpoIIR processing seems to be crucial for signalling we sought to directly visualize SpoIIR processing, by fluorescence microscopy, during B. subtilis spore formation. To this end, we used a B. subtilis strain harbouring a fusion of the green fluorescent protein (GFP) to the N-terminus of full-length SpoIIR (Rubio and Pogliano, 2004). In contrast to the findings of Rubio and Pogliano (2004), who reported that the GFP–SpoIIR fusion was fully functional, we found that when this gfp–spoIIR fusion was expressed as the sole source the SpoIIR, it substantially reduced the sporulation efficiency of B. subtilis and this was because of a delay in the activation of σE (see below). However, when gfp–spoIIR was expressed in a strain that also expressed native spoIIR (VD221) the kinetics of sporulation was normal (data not shown). Thus, we used strain VD221 to monitor GFP–SpoIIR localization during development. Strain VD221 was induced to sporulate by resuspension and visualized by fluorescence microscopy at different stages of differentiation. Consistent with a previous report (Rubio and Pogliano, 2004), we found that GFP–SpoIIR first localized in early sporangia as a focus at the septal midpoint (Fig. 5, arrow 1), and then moved along with the mother cell membrane (Fig. 5, arrow 2). During the early engulfment processes, the fusion protein closely followed the engulfing mother cell membrane as it moved around the forespore (Fig. 5, arrow 3) but at later times the fluorescence signal accumulated in the forespore cytoplasm (Fig. 5, arrow 4). We interpret our results to indicate that at later times the N-terminal signal sequence of the pre-protein was cleaved by the leader peptidase releasing GFP which freely diffused in the cytosol. Western blot analysis with GFP antibodies (Fig. S3A) showed two bands of apparent masses of 65 and 27 kDa, corresponding to GFP–SpoIIR and GFP, respectively, thereby confirming that GFP–SpoIIR was processed at the forespore septal membrane. Detection of faster migrating bands suggests that the GFP fused to the signal peptide is being degraded after SpoIIR maturation. These data strongly suggest that GFP–SpoIIR is first localized to the sporulation septum and then translocated by the Sec protein translocation pathway. Supporting that the recognition and cleavage of the internal signal peptide contained in GFP–SpoIIR is carried out by the Sec system, several examples of processing of internal signal sequences present in secreted proteins have been recently reported in Gram-positive bacteria (Antelmann et al., 2001; Lu et al., 2009; Powers et al., 2011).
Expression of gfp–spoIIR as the only source of SpoIIR led to the appearance of an abortively disporic phenotype (data not shown), characteristic of cells in which spoIIR expression is delayed (Khvorova et al., 2000; Hilbert and Piggot, 2004). In concert with this finding, immunoblot analysis of the accumulation of mature SpoIIR and σE showed that the processing of both pre-proteins was significantly delayed in cells expressing gfp–spoIIR (Fig. S3A and B). It should be noted that σE is normally active within about 4 min of spoIIR transcription (Eldar et al., 2009). Thus, although the delayed processing of GFP–SpoIIR is a non-physiological process it clearly supports a causal relationship between GFP–SpoIIR processing and σE activation.
Inhibition of fatty acid synthesis does not affect localization and translocation of SpoIIR
The inhibition of SpoIIR processing by cerulenin could be due to the drug interfering in several different steps of the secretion process. For example, cerulenin might inhibit the septal targeting or the delivery of SpoIIR to the Sec apparatus. Thus, we next investigated the effect of the inhibition of fatty acid synthesis on GFP–SpoIIR localization, translocation and processing.
To determine if inhibition of fatty acid synthesis perturbed SpoIIR localization, we treated strain VD221 with cerulenin at different time points after the onset of sporulation. Analysis of GFP–SpoIIR localization at hour three of sporulation (Fig. 6 and Table 2) showed that treatment of these cultures with cerulenin gave rise to different patterns of GFP–SpoIIR localization, depending on the time in which the antibiotic was added. Previous morphological analysis revealed that most B. subtilis cells treated with cerulenin at the onset of sporulation are unable to form septa, or synthesize primarily thick or abnormal septa (Schujman et al., 1998). In agreement with these early observations, when cerulenin was added after 12 min of resuspension (t0.2) the asymmetric septation was inhibited in more than 65% of the cells thus preventing gfp–spoIIR expression (Table 2 and Fig. 6, arrow 1). However, at least 35% of the scored cells initiated septa and in these cells SpoIIR was localized both in the prespore and in the mother cell (Table 2 and Fig. 6, arrow 2). This is probably because in cells developing an aberrant septum (Schujman et al., 1998), σF is abnormally activated in both sporangium compartments. In cells treated with cerulenin 36 min after resuspension (t0.6) it was observed that 63% of the cells formed an asymmetrical polar septum, and in 48% of these cells GFP–SpoIIR was localized in the polar septum (Table 2 and Fig. 6, arrows 3 and 4). This septal localization of GFP–SpoIIR in cerulenin-treated cells is similar to the one observed at early sporulation stages of untreated cultures (Fig. 5, arrows 1 and 2). However, as these lipid-starved cells were unable to initiate engulfment, the movement of GFP–SpoIIR along with the mother cell membranes was blocked (Fig. 6, arrows 3 and 4). Strikingly, GFP–SpoIIR remained associated to the asymmetric septum even after hour four of sporulation (data not shown), thus indicating that SpoIIR achieves proper septal localization in the absence of de novo synthesis of membrane lipids, which instead is necessary for its processing and/or translocation. Consistent with the fluorescence microscopy results, Western blot analysis with anti-GFP antibodies showed that cells deprived of lipid synthesis at t0.6 of sporulation produced GFP–SpoIIR normally but did not process it (Fig. S3A). Finally, when cerulenin was added 54 min after resuspension (t0.9), 40% of the cells processed GFP–SpoIIR (as judged by the GFP fluorescence visualized in the cytosol) once forespore engulfment was initiated (Table 2 and Fig. 6, arrows 5 and 6). These results agree with a previous report (Schujman et al., 1998) showing that the inhibitory effect of cerulenin on σE processing and sporulation disappears when the antibiotic was added at more advanced stages of sporulation.
Table 2. Quantification of septal membrane morphologies and GFP localization in cerulenin-treated cells.
Signal fluorescence not confined to the forespore.
Cells of strain VD221 (amyE::PspoIIR–gfp–spoIIR) were treated as described in Fig. 6 and harvested a t3 for fluorescence microscopy examination.
The fluorescence microscopy results described above indicate that de novo fatty acid synthesis is dispensable for recruitment of SpoIIR to the septal membrane. However, the microscopy data alone does not allow us to conclude if GFP–SpoIIR is being translocated across the membrane when fatty acid synthesis is inhibited. Namely, the septally located GFP–SpoIIR could be fully translocated although not processed or alternatively, the fusion protein could become localized to the asymmetric septum even if it does not get translocated across the membrane. In this case, the protein could be targeted to the asymmetric septum by interactions with the intracellular portion of forespore proteins that localize to the septal membrane before passing to the secretory translocase. Thus, to distinguish if fatty acid synthesis is required for SpoIIR translocation, we used a ΔspoIIGA strain containing a construct that replaces the promoter of the spoIIG operon with the σF-directed spoIIQ strong promoter (PspoIIQ-spoIIGAB) so as to enhance and confine spoIIGAB transcription to the prespore (Chary et al., 2010). To monitor processing of σE in the prespore, we assayed the levels of β-galactosidase activity coded by a lacZ reporter gene fused to the σE-controlled PspoIID. If fatty acid synthesis is required for translocation of SpoIIR by the secretion apparatus, then the addition of cerulenin should block the activation of the spoIID promoter since SpoIIR would not be translocated outside of the prespore to activate the extracellular domain of SpoIIGA inserted in the prespore membrane. On the other hand, if fatty acid synthesis is required for processing of SpoIIR, the addition of cerulenin would not inhibit its translocation to the interseptal space, although SpoIIR would remain tethered to the forespore membrane via the un-cleaved signal peptide. In this scenario SpoIIR could in principle interact with extracellular domains of SpoIIGA and activate σE. As shown in Fig. 7, the activation of the spoIID promoter was not inhibited by cerulenin. However, the activation of σE was completely blocked by deletion of spoIIR (Fig. 7), thus confirming that in this genetic background (ΔspoIIGA, PspoIIQ-spoIIGAB) SpoIIR acts as an autocrine signal to activate processing of σE (Chary et al., 2010).
We conclude that de novo fatty acid synthesis is dispensable for recognition of SpoIIR by the translocation machinery and its export across the septal membrane. On the other hand, fatty acid synthesis is essential for removal of the SpoIIR signal peptide and, presumably, for the release of the mature protein into the interseptal space. This role of fatty acid synthesis is crucial since ordinarily SpoIIGA in the mother cell predominates (Chary et al., 2010) and so σE activation is confined in the mother cell. However, as shown here, if expression of SpoIIGA in the prespore is artificially enhanced it can be activated by SpoIIR, even if processing of its signal sequence is inhibited in cells exposed to cerulenin.
SpoIIR is captured in the septal membrane by SpoIIGA
The discovery that GFP–SpoIIR initially localized to the sporulation septum and then was released into the engulfment intermembrane space, led us to test the fate of GFP–SpoIIR in a B. subtilis mutant strain that fails to initiate engulfment. To this end, we introduced gfp–spoIIR in a strain in which the parental spoIIGA was inactivated, thus preventing the expression of the SpoIIGA processing enzyme, so as to block any σE-directed gene required for engulfment. Sporulation was induced by resuspension and GFP–SpoIIR was localized at different times of sporulation (Fig. 8). Unexpectedly, from t1.5 of sporulation we found that GFP–SpoIIR was localized all around the forespore membrane (Fig. 8, arrows 1 and 2), contrasting with the septal localization of the fusion protein observed in spoIIGA+ strains at a similar sporulation time (Fig. 5, arrows 1 and 2). At a later stage of development (t4), the GFP fluorescence was localized in the prespore cytoplasm (Fig. 8, arrow 3) indicating that SpoIIR was cleaved by the signal peptidase, as confirmed by Western Blot analysis using anti-GFP antibodies (Fig. S3A). These results strongly suggest that SpoIIGA is involved in the capture of its own activator, SpoIIR, in the septum.
To determine whether this recruitment pathway also operates to restrict SpoIIR secretion to the sporulation septum, rather than secrete it randomly throughout the cell membrane of the forespore, we took advantage of the complementation test depicted in Fig. S4. Using this tool, we tested if expression of the SpoIIGA protein in the donor cell suppressed the extracellular export of SpoIIR. To this end we used as donors isogenic strains expressing SpoIIR in either a spoIIGA+ or a ΔspoIIGA background (Fig. S4A). Two of the three donor strains also contained the spoIIIG mutation to block sporulation in cells expressing SpoIIR and SpoIIGA. Each of the donor strains was mixed with a ΔspoIIR spoIIGA+ acceptor strain and subjected to sporulation, observing that sporulation frequency is significantly enhanced only in mixed cultures in which SpoIIGA is missing in donor cells (Fig. S4B). These results suggest that in sporulating B. subtilis, SpoIIGA is necessary to anchor SpoIIR at the forespore septal membrane. This mechanism could explain the vectorial export of SpoIIR from the forespore to the intermembrane septal space and enhance the effective interaction with the mother cell-located SpoIIGA. In summary, the interaction between SpoIIGA and SpoIIR in the forespore and in the mother cell septal membranes is critical for localization of SpoIIR as well as for the activation of σE.
In this study we demonstrated that processing (and subsequent release) of SpoIIR to the intermembrane space to trigger the SpoIIGA-mediated activation of pro-σE, in the mother cell, is dependent on fatty acid synthesis in the prespore. We analysed a number of mutants of SpoIIR to demonstrate that T27 is required for proper proteolytic processing and intercellular transport of SpoIIR during the developmental programme. In addition, we have demonstrated that SpoIIR signalling activity, contrary to what is generally assumed (Londoño-Vallejo, 1997; Imamura et al., 2008), requires its processing and release from the membrane to activate pro-σE. This finding is significant because it seems to indicate that SpoIIR located in the intercellular space but still connected to the membrane, is unable to trigger processing of pro-σE in the mother cell. Finally we report that septal localization of GFP–SpoIIR requires SpoIIGA.
A model for the vectorial cell–cell signalling by SpoIIR
Altogether, our results are most consistent with a novel model for SpoIIR transport and activation of σE that is based on two premises: (i) SpoIIR cannot activate SpoIIGA present in the mother cell membrane unless it has been released into the interseptal space. This is supported by our observations that processing of SpoIIR signal sequence is essential for σE activation, and (ii) SpoIIR localization to the forespore septum is mediated by its interaction with SpoIIGA protein. This second premise is supported by direct microscopic observations that showed that septal localization of SpoIIR was dependent on SpoIIGA.
According to our model (Fig. 9), SpoIIR is first translocated to the external side of the septal membrane by the SecYEG translocon and there it interacts with the extracellular domain of SpoIIGA present in the forespore membrane (Fig. 9A). We therefore suggest that a crucial function of the extracellular portions of forespore SpoIIGA would be to retain the SpoIIR pre-protein in the septal membrane to provide spatial and temporal co-ordination to SpoIIR processing. Nevertheless, our microscopy analysis does not allow distinguishing if GFP–SpoIIR localization occurs before or after its translocation across the septal membrane. Thus, our results are also consistent with the hypothesis that the extracellular domain of SpoIIGA associated to the mother cell membrane contributes, or even wins in the competition with forespore SpoIIGA, to retain unprocessed SpoIIR at the forespore septal membrane (Fig. 9B). However, our data strongly suggest that the interaction between SpoIIR and SpoIIGA in opposing membranes should be ineffective to activate σE, unless the signal peptide of SpoIIR is processed and the mature signalling protein is released into the interseptal space. Our findings fit with and extend the recent model of Chary et al. (2010) that posits that σE compartmentalization can be partially attributed to a competition between the mother-cell and forespore compartments for SpoIIR. Normally, SpoIIGA is predominantly located in the mother cell septum and this confines σE activation to the mother cell. Because of the preponderance of SpoIIGA in the mother cell membrane, it wins the competition for the limited amount of mature SpoIIR released in the interseptal space. One gap in the Chary et al. (2010) model is how the vectorial secretion of SpoIIR is achieved. Here we propose that SpoIIGA in the forespore (Fig. 9A) and/or in the mother cell (Fig. 9B) also acts as receptor for the SpoIIR pre-protein, favouring its septal translocation, processing and release.
One important question raised by our work is: why would cleavage of the SpoIIR signal peptide come to depend on concomitant prespore fatty acid synthesis? Fatty acid modification of cellular proteins plays diverse roles in the regulation of membrane interactions, intracellular sorting, stability and membrane micropatterning (Hannoush and Sun, 2010; Salaun et al., 2010). We ruled out the involvement of a diacylglycerol-modified lipoprotein, since PlsY and PlsC activities are both required for the synthesis of this lipid moiety (Paoletti et al., 2007). Instead, we speculate that the SpoIIR N-terminus undergoes internal acylation by directly incorporating a long chain fatty acid to the conserved T27 (Fig. 9C) to assist the recognition and/or cleavage of the SpoIIR N-terminal extension by a signal peptidase (Fig. 9D). In support of this hypothesis, we demonstrate that replacement of T27 with alanine, completely abolished SpoIIR processing. However, when T27 was substituted by a tyrosine, SpoIIR was normally processed during sporulation. We propose that a hydroxyester bond between SpoIIR's hydroxyl-bearing threonine or tyrosine residues and a fatty acid is important for cleavage of the SpoIIR signal peptide. SpoIIR is synthesized in very limited amounts (Karow et al., 1995). Thus, SpoIIR modification with a fatty acid could function to concentrate the acylated protein in a membrane microdomain harbouring both SpoIIGA and a signal peptidase. This protein association determined by a membrane microdomain could act in parallel with the recruitment of SpoIIR by an interaction with SpoIIGA and favour the enzymatic cleavage of the pre-protein signal peptide and release of the mature protein into the interseptal space. Regulation of signalling pathways by compartmentalization of proteins in microdomains is an unusual mode of developmental control in bacteria, but a precedent is known. The histidine kinase KinC which is involved in biofilm formation in B. subtilis, is activated when it is recruited in a membrane domain analogous to lipid rafts in eukaryotic cells (Lopez and Kolter, 2010). If our model is correct, then what is the nature of the compound that serves as a fatty acid donor during SpoIIR secretion and which enzyme catalyses SpoIIR acylation? In eukaryotes the presumed donor for post-translational acylation of internal threonines or serines is acyl-CoA (Yang et al., 2008). This acyl donor is also synthesized in B. subtilis by long chain fatty acyl-CoA ligases that are induced at the onset of sporulation (Matsuoka et al., 2007). Nevertheless, the best known example of fatty acid modification of secreted proteins in bacteria is the acylation of Escherichia coli haemolysin A (HlyA). HlyA requires the covalent amide linkage of fatty acids to internal lysine residues for activity. Acylation of HlyA is carried out by HlyC which is a homodimeric putative acyltransferase, using acyl-acyl carrier protein (ACP) as the fatty acid donor (Stanley et al., 1998). In contrast to E. coli, fatty acid modifications of proteins in Gram-positive bacteria are poorly understood. For example, the fatty acid acylation of lipopeptides in B. subtilis is well established although the enzymes that activate and provide the fatty acid necessary for initiation or synthesis are as yet unidentified [for a review see Stachelhaus et al. (2002)].
Why is such a complex sequence of events required for σE activation in B. subtilis? It has been extensively established that precise temporal and spatial regulation of gene expression is a critical prerequisite to control cell differentiation. Here we established a link between fatty acid synthesis and σE activation that seems designed to tightly co-ordinate metabolic conditions with progression of the developmental pathway, thus ensuring the survival and genome integrity during germination of the future spore.
The B. subtilis strains were grown in Luria–Bertani (LB) broth (Difco), Spizizen minimal medium (SMM: Spizizen, 1958) or Difco sporulation medium (DSM: Schaeffer et al., 1965; Nicholson and Chambliss, 1985; Nicholson and Setlow, 1990). E. coli strains were propagated in LB broth. Antibiotics were supplied by Sigma, and used at the following concentrations: chloramphenicol (5 µg ml−1 for B. subtilis; 10 µg ml−1 for E. coli), macrolides (erythromycin 0.5 µg ml−1 plus lincomycin 12.5 µg ml−1), spectinomycin (100 µg ml−1), kanamycin (5 µg ml−1 for B. subtilis; 50 µg ml−1 for E. coli), cerulenin (3.3 or 10 µg ml−1) and ampicillin (100 µg ml−1). To quantify spore formation, cells were induced to sporulate in DS medium (Schaeffer et al., 1965). After 24 h growth in DS medium at 37°C, the number of colony-forming units that survived heat treatment (20 min at 80°C) was calculated. For all other experiments (including measurement of lacZ reporter gene activity, localization of GFP fusion proteins and immunoblot analysis), sporulation was induced by the resuspension method (Sterlini and Mandelstam, 1969; Nicholson and Setlow, 1990) with t representing the hours after onset of sporulation. When indicated, IPTG (isopropyl-β-d-thiogalactosidase) 0.2 mM or xylose 0.1% was added to the medium.
Strains and plasmids construction
All strains were derived from the laboratory strain JH642 (Dean et al., 1977). Details of strain and plasmid sources and construction are given in Supporting information. Additionally, see Table S1 for the full genotypes of strains and a description of plasmids, and Table S2 for a list of primers used in this study.
β-Galactosidase-specific activity (ΔA420 per min per ml of culture × 1000/A525) was determined at indicated times as previously described (Schujman et al., 1998) after pelleting cell debris and specific activity was expressed in Miller units (Sambrook et al., 1989).
Growth of mixed cultures
Mixed cultures were performed by growing each strain individually in DS medium until reaching an OD600 of 1. Then, two cultures were mixed in equal parts and the mixed cultures were incubated for 24 h. Spore formation was quantified as described in General methods.
Western blot analysis
For detection of His-tagged proteins, strains were induced to sporulate by resuspension method. A 10 ml aliquot of each culture was harvested at the indicated times, centrifuged and frozen. The pellets were resuspended in lysis buffer [50 mM Tris-HCl (pH 8.0) and 1 mM phenylmethylsulphonyl fluoride], adding 80 µl of buffer per OD600 unit and sonicated. After protein quantification by Lowry method (Lowry et al., 1951), 40 µg of proteins from each sample were boiled in the presence of loading buffer, loaded on a 15% SDS-polyacrylamide gel and transferred to nitrocellulose. For protein detection a 1:200 dilution of mouse monoclonal anti-His antibodies (R&D System) was applied, followed by a 1:3000 dilution of horseradish peroxidase-labelled goat anti-mouse antibodies, and enhanced chemiluminescence (ECL, Amersham). Similar sample preparation was applied for GFP fusion detection, loading 10 µg of proteins from each sample on a 12% SDS-polyacrylamide gel and after transferring to nitrocellulose membrane. A 1:1500 dilution of anti-GFP antibodies from rabbit (kindly provided by Richard Losick) was used, followed by 1:3000 dilution of alkaline phosphatase-labelled goat anti-rabbit antibodies. BCIP/NBT (Bio-Rad) was used as the alkaline phosphatase substrate. Detection of σE processing was performed using 1 ml aliquot from each culture harvested at the indicated times. The pellets were resuspended in 100 µl/OD600 lysis buffer and disrupted by incubating with lysozyme (500 µg ml−1) for 15 min at 37°C. Twenty microlitres from each protein sample was boiled for 5 min in the presence of loading buffer. Each sample was subjected to SDS-PAGE in a 12% acrylamide gel and transferred to nitrocellulose membrane. Proteins were revealed using anti-σE monoclonal antibody and a secondary antibody conjugated to alkaline phosphatase.
Microscopy was performed on a Nikon TE300 inverted microscope equipped with filters for green fluorescent protein (GFP) (Endow GFP set, 41018; Chroma Technology) and FM5-95 (Texas Red Brightline set, TXRED4040-B; Semrock). Images were captured with a Roper CoolSnap HQ camera. Exposure times varied from 0.2 to 2 s. Images were processed and analysed with NIS-Elements version 4.20 (Nikon Instruments) and ImageJ (http://rsb.info.nih.gov/ij/) software. To perform microscopy, cells from resuspended cultures were collected at the indicated times and mounted on slides covered with a layer of agarose-solidified phosphate-buffered saline (PBS). Membranes were stained with FM5-95 (Molecular Probes) at a final concentration of 1 µg ml−1. Dye was usually added directly to cells in medium 5 min before mounting and imaging.
We thank P. Piggot for sending strain SL14653 and helpful suggestions. This work was supported by grants from the Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Agencia Nacional de Promoción Científica y Tecnológica (FONCYT), Fundación Josefina Prats, FAPESP (08/58821-1) and CNPq (478019/2009-2). V.D. is a fellow from CONICET and G.E.S. and D.deM. are career investigators from CONICET. D.deM. is an international Research Scholar from Howard Hughes Medical Institute.