Peptide anchoring spore coat assembly to the outer forespore membrane in Bacillus subtilis

Authors


*E-mail losick@mcb.harvard.edu; Tel. (+1) 617 495 4905; Fax (+1) 617 496 4642.

Summary

Spore formation in Bacillus subtilis involves the formation of a thick, proteinaceous shell or coat that is assembled around a specialized membrane known as the outer forespore membrane. Here we present evidence that the assembling coat is tethered to the outer forespore membrane by a 26-amino-acid peptide called SpoVM, which is believed to form an amphipathic helix. We show that proper localization of SpoVM is dependent on SpoIVA, a morphogenetic protein that forms the basement layer of the spore coat, and conversely, that proper localization of SpoIVA is dependent on SpoVM. Genetic, biochemical and cytological evidence indicates that this mutual dependence is mediated in part by contact between an amino acid side-chain located near the extreme C-terminus of SpoIVA and an amino acid side-chain on the hydrophilic face of the SpoVM helix. Evidence is also presented that SpoVM adheres to the outer forespore membrane via hydrophobic, amino acid side-chains on the hydrophobic face of the helix. The results suggest that the SpoVM helix is oriented parallel to the membrane with the hydrophobic face buried in the lipid bilayer.

Introduction

The sorting of proteins to their correct subcellular destination represents a terminal step in cellular morphogenesis. Typically, localization signals encoded within the primary amino acid sequence direct newly synthesized proteins to various chemically distinct destinations within a cell (Blobel, 1980). Of particular interest, however, is the mechanism by which proteins are sorted to subcellular regions that harbour no obvious physical uniqueness. In bacteria, these regions are often patches of membrane either at the cell poles or at recent sites of cell division. To date, it is largely unknown how cells are able to distinguish these subcellular regions from others in a cell. We have sought to elucidate the mechanism by which this occurs by examining the localization of model substrates during sporulation in the Gram-positive bacterium Bacillus subtilis.

When exposed to conditions of nutrient limitation, B. subtilis initiates an unusual cell division event that divides the cell into two unequal-sized compartments that initially lie adjacent to one another: the mother cell, and the forespore (the smaller compartment) (Losick et al., 1986; Stragier and Losick, 1996). The asymmetrically positioned division septum (called the polar septum) then curves as the mother cell engulfs the forespore in a phagocytic-like process. Ultimately, the leading edges of the engulfing membrane fuse, which pinches off the forespore as a double membrane-enclosed organelle in the mother cell cytosol. A thick, unique layer of peptidoglycan called the cortex is then synthesized between the two membranes enveloping the forespore, and the mother cell deposits a proteinaceous shell or coat on the surface of the outer forespore membrane. The mechanism by which the proteins that make up the coat are guided to the outer forespore membrane and subsequently anchored to it is largely unknown (Driks, 1999). Ultimately, the mature (ripened) spore is released into the milieu by a programmed lysis of the mother cell, at which time it is resistant to a wide variety of environmental extremes.

Although the mother cell is initially delineated by a contiguous membrane, a large number of sporulation-specific proteins synthesized in the mother cell are targeted specifically to the mother cell face of the engulfing membrane. Among these proteins is a 26-amino-acid-long peptide called SpoVM, hereafter referred to simply as VM. VM is synthesized under control of the mother cell-specific transcription factor σE and almost quantitatively localizes to the engulfing membrane (Levin et al., 1993; van Ooij and Losick, 2003). As engulfment proceeds, VM remains associated with the curving membrane until it ultimately decorates the outer forespore membrane. Cells harbouring a spoVM deletion produce an ill-defined cortex and a thin, loosely attached coat, resulting in a dysfunctional spore that is not heat resistant (Levin et al., 1993). Furthermore, substitution of three N-terminal proximal amino acids that mediate localization of VM also produces heat-sensitive spores, suggesting that spore morphogenesis is dependent on proper localization of VM (van Ooij and Losick, 2003). Despite the severe phenotype caused by its absence, the precise function of this peptide has been unclear.

Evidence indicates that VM is an amphipathic alpha helix (Prajapati et al., 2000). Previous reports had hypothesized that positively charged residues along the polar face of the helix mediate an unspecific ionic interaction with negatively charged phospholipid head groups, providing a mechanism by which VM adheres to the membrane (Prajapati et al., 2000). Curiously, the three N-terminal residues implicated in the proper localization of VM lie along this charged face, suggesting that these residues recognize a specific defining feature of the polar septum.

Here we report the isolation of an allele-specific suppressor mutation that reverses the localization and sporulation defect of one of the VM mislocalization mutants. The suppressor mutation causes an amino acid substitution near the extreme C-terminus of the coat protein SpoIVA (hereafter simply IVA), which forms the basement layer of the spore coat (Piggot and Coote, 1976; Roels et al., 1992; Driks et al., 1994). Surprisingly, VM mislocalization mutants also cause the mislocalization of IVA. This reciprocal dependence appears mediated by direct contact between VM and the C-terminal region of IVA. Further, we demonstrate that the positively charged residues of VM are not primarily responsible for the association of VM with the membrane. Rather, our results suggest that the positively charged residues of the VM helix face the cytosol and that the hydrophobic residues are embedded within the phospholipid bilayer. We propose a model in which the basement layer of the spore coat is anchored to the outer forespore membrane via a direct association between IVA and VM.

Results

Amino acid substitutions causing mislocalization of SpoVM

To study the effect of amino acid substitutions on the subcellular localization of VM, we used a previously described fusion of the peptide to the green fluorescent protein (VM–GFP). Because VM tagged with GFP was non-functional with respect to sporulation (van Ooij and Losick, 2003), we also created a FLAG-tagged derivative of VM that was capable of supporting sporulation. Cells producing VM-FLAG were only modestly impaired in sporulation, producing heat-resistant spores at an efficiency of 18% that of the wild type. We took advantage of the relative sporulation proficiency of cells producing VM-FLAG to confirm key results obtained with VM–GFP by carrying out immunofluorescence microscopy using anti-FLAG antibodies. Investigations into the effects of the amino acid substitutions on spore formation were carried out using wild-type (untagged) VM.

As seen in Fig. 1A, VM–GFP localizes to the patch of mother cell membrane that engulfs the forespore, which is known as the engulfment membrane or the outer forespore membrane. This can be observed both in sporangia that are undergoing engulfment (in which VM–GFP can be seen as an arc) and in sporangia that have completed engulfment (in which the fusion protein can be seen as a ring that completely surrounds the forespore). Note that VM–GFP localized almost exclusively to the engulfment membrane, with little or no signal being detected in the mother cell (cytoplasmic) membrane that surrounds the sporangium. Likewise, VM-FLAG localized exclusively to the forespore when viewed by immunostaining using anti-FLAG antibodies (Fig. S1A–C). In contrast, fusion proteins harbouring alanine substitutions at positions 3, 6 or 9 of VM exhibit a significant level of mislocalization to the cytoplasmic membrane that surrounds the sporangium (van Ooij and Losick, 2003; compare, for example, the localization of VMI6A–GFP in Fig. 2C with that of VM–GFP in Figs 1A and 2A). Interestingly, and as noted previously (van Ooij and Losick, 2003), strains producing such mislocalization mutants of untagged VM are severely impaired in sporulation, producing spores six orders of magnitude less efficiently than the wild type (Table 1, strains A–D, genotypes listed in Table 2). Evidently, and as inferred from the behaviour of the GFP-tagged proteins, even moderate levels of mislocalization profoundly impede morphogenesis.

Figure 1.

Proper localization of VM depends on IVA.
A–C. VM–GFP localization (A) in the wild type (strain CVO1195), (B) in an IVA deletion mutant (KR128) and (C) in an IVA deletion mutant harbouring a complementing copy of IVA at the thr locus (KR209).
D–F. Membranes of cells in (A)–(C) visualized with the dye FM4-64. The fraction of the total fluorescent signal present in the area corresponding to the forespore is reported below each image as ‘% forespore localization’ (see Experimental procedures).

Figure 2.

A–D. Mislocalization of VMI6A is corrected by IVAG486V VM–GFP localization (A) in the presence of wild-type IVA (KR131) and (B) in cells harbouring IVAG486V (KR139). VMI6A–GFP localization (C) in the presence of wild-type IVA (KR133) and (D) in cells harbouring IVAG486V (KR141).
E–H. Membranes of cells in (A)–(D) visualized with the dye FM4-64. Quantification of the fluorescence signal at the forespore is as described for Fig. 1.

Table 1.  Sporulation efficiencies of strains harbouring various VM and IVA alleles.
StrainaVMIVASporulation efficiency
  • a. 

    Strain A: CVO1422; B: CVO1402; C: CVO1399; D: CVO1405; E: KR21; F: KR37; G: KR44; H: KR46; I: KR47. Genotypes are listed in Table 2.

AWTWT0.94
BI6AWT2.4 × 10−6
CF3AWT1.9 × 10−6
DP9AWT2.6 × 10−6
EI6AG486V0.44
FΔG486V4.4 × 10−6
GF3AG486V2.6 × 10−3
HP9AG486V6.1 × 10−4
IWTG486V0.59
Table 2.  Strains used in this study.
StrainGenotypeReference/source
PY79Prototrophic derivative of B. subtilis 168Youngman et al. (1984)
EL200ΔspoVM::specCutting et al. (1997)
CVO1195amyE::spoVM–gfp catvan Ooij and Losick (2003)
CVO1399ΔspoVM::spec amyE::spoVMF3Acatvan Ooij and Losick (2003)
CVO1402ΔspoVM::spec amyE::spoVMI6Acatvan Ooij and Losick (2003)
CVO1405ΔspoVM::spec amyE::spoVMP9Acatvan Ooij and Losick (2003)
CVO1422ΔspoVM::spec amyE::spoVM catvan Ooij and Losick (2003)
KR21ΔspoVM::spec amyE::spoVMI6AspoIVAG486V 
KR37ΔspoVM::spec spoIVAG486V 
KR44ΔspoVM::spec amyE::spoVMF3AspoIVAG486V 
KR46ΔspoVM::spec amyE::spoVMP9AspoIVAG486V 
KR47ΔspoVM::spec amyE::spoVM spoIVAG486V 
KR139ΔspoVM::tetΔspoIVA::neoR amyE::spoVM–gfp cat thrC::spoIVAG486Vspec 
KR141ΔspoVM::tetΔspoIVA::neoR amyE::spoVMI6A–gfp cat thrC::spoIVAG486Vspec 
KR160thrC::gfp–spoIVA spec 
KR163ΔspoVM::tet amyE::spoVMI6Acat thrC::gfp–spoIVA spec 
KR165ΔspoVM::tet amyE::spoVM cat thrC::gfp–spoIVA spec 
KR171ΔspoVM::tet amyE::spoVMI6Acat thrC::gfp–spoIVAG486Vspec 
KR173ΔspoVM::tet amyE::spoVM cat thrC::gfp–spoIVAG486Vspec 
KR178ΔspoVM::tet thrC::gfp–spoIVA spec 
KR318amyE::Phyperspank -spoVM–gfp spec 
KR323amyE::Phyperspank -spoVMK2A,K7A,K10A,R24A,K25A–gfp spec 
KR333amyE::Phyperspank -spoVML12E–gfp spec 
KR335amyE::Phyperspank -spoVMI15E–gfp spec 
KR328amyE::Phyperspank -spoVML12E,I15E–gfp spec 
KR329amyE::Phyperspank -spoVML12W,I15W–gfp spec 
KR359ΔspoVM::tet amyE::spoVM-FLAG cat 
KR375ΔspoVM::tetΔspoIVA::neo amyE::spoVM-FLAG cat 

Allele-specific suppression of the mislocalization mutant VMI6A

We took advantage of the severe phenotype of VMI6A to select for a suppressor mutation that would restore the capacity of cells producing the mislocalization mutant to sporulate. We reasoned that such a suppressor mutation might identify a partner protein that helps VM localize to the engulfment membrane. Accordingly, cells harbouring VMI6A were subjected to repeated cycles of sporulation, heat-treatment to kill cells that had not successfully formed spores, germination and growth. After three such cycles, one such suppressor mutant emerged, which upon further characterization was found to harbour a point mutation in the 492-codon-long sporulation gene spoIVA (hereafter referred to as IVA): a guanidyl to thymidyl transversion that converted codon 486 from glycine to valine. The G486V substitution was highly effective in suppressing the deleterious effect of the I6A substitution in VM as cells that were doubly mutant for both proteins sporulated almost as efficiently as did the wild type (Table 1, strain E). Interestingly, the G486V substitution mutant of IVA caused little impairment of sporulation on its own (i.e. in cells that were wild type for VM; strain I).

Next, we asked whether the IVAG486V mutation was simply bypassing the requirement for VM in sporulation or was an allele-specific suppressor of VMI6A. The results of Table 1 show that spoIVAG486V did not obviate the need for VM altogether nor did it restore the capacity of cells harbouring VMF3A or VMP9A to sporulate with comparable efficiency to that of IVAG486VVMI6A double mutant. [The IVAG486V allele did increase the sporulation efficiencies of the VMF3A and VMP9A mutants by two to three orders of magnitude, but these cells were still largely unable to sporulate (Table 1, strains G–H).] We conclude that IVAG486V is an allele-specific suppressor and that the substitution of valine for glycine at position 486 of IVA somehow restores the proper function of VM harbouring a substitution of alanine for isoleucine at position 6.

Proper subcellular localization of VM–GFP depends on IVA

The simplest interpretation of the results so far is that IVA, which is itself localized to the region around the outer forespore membrane (Roels et al., 1992; Driks et al., 1994; Webb et al., 1995; Price and Losick, 1999), contacts VM and thereby facilitates its localization to the engulfment membrane. Consistent with this idea, little or no VM–GFP was detected at the cytoplasmic membrane in wild-type sporangia whereas a substantial level of mislocalization was observed in mutant sporangia lacking IVA (Fig. 1A and B). Experiments in which we attempted to quantify these results by scanning across the fluorescent images indicated that about 95% of the fluorescent signal was associated with the outer forespore membrane in wild-type sporangia, but only about 70% was properly positioned in sporangia lacking IVA (Fig. 1 and Experimental procedures). Conversely, little or no signal above the background could be detected at the cytoplasmic membrane in the wild type whereas about 30% of the total signal was mislocalized in the mutant.

We confirmed the requirement of IVA for the proper localization of VM-FLAG using immunostaining (Fig.  S1D–F). In contrast to the GFP fusion, which although mislocalized appeared principally associated with the cytoplasmic membrane, VM-FLAG was seen to be distributed in the cytosol in the absence of IVA. Perhaps the displacement of VM-FLAG from the membrane was a consequence of the fixation process required for immunofluorescence microscopy. Nonetheless, it is clear that proper localization of VM required IVA.

Finally, that mislocalization was due to the absence of IVA rather than a polar effect of the IVA deletion mutation on the expression of a downstream gene was confirmed by the complementation experiment of Fig. 1C, which shows that proper localization was restored to the mutant when a copy of IVA was introduced into the chromosome at an ectopic locus (thr).

Earlier work had shown that the localization of VM, which is produced in the mother cell under the control of σE, is dependent on the product of an unknown gene or genes under the control of this mother cell-specific transcription factor (van Ooij and Losick, 2003). The present results indicate that one such gene is IVA.

IVAG486V corrects the mislocalization defect of VMI6A–GFP

We next tested whether IVAG486V corrects the mislocalization defect of VMI6A–GFP. VM–GFP localizes almost exclusively to the engulfment membrane in the presence of either wild-type IVA or IVAG486V (Fig. 2A and B). The tolerance of IVAG486V by wild-type VM–GFP is consistent with the observation that otherwise wild-type cells harbouring IVAG486V sporulate efficiently (Table 1). However, in the presence of wild-type IVA, a significant level of VMI6A–GFP was associated with the cytoplasmic membrane surrounding the sporangium (Fig. 2C). In contrast, when wild-type IVA was replaced with the IVAG486V mutant protein, VMI6A–GFP was found to localize almost exclusively to the outer forespore membrane in sporangia that were undergoing engulfment or had completed engulfment, a pattern that was indistinguishable from that observed for wild-type VM–GFP (Fig. 2D). Thus, IVAG486V not only suppresses the sporulation defect of VMI6A but also corrects its tendency to mislocalize to the cytoplasmic membrane as judged by use of a GFP fusion.

IVA binds VM but not VMI6A

A simple explanation for the discovery that the sporulation and localization defects of VMI6A were suppressed by a single-amino-acid change in IVA is that the two proteins interact directly and that isoleucine6 of VM is a contact site with glycine486 of IVA. Conceivably, the long side-chain of the isoleucine mediates a hydrophobic interaction with the short side-chain of the glycine. If so, then the replacement of isoleucine6 with alanine would eliminate the hydrophobic interaction, while the further replacement of glycine486 with valine would restore the interaction.

To investigate whether VM and IVA do indeed interact, we tagged VM at its C-terminus with six histidine residues and purified the his-tagged protein by affinity chromatography. Next, we prepared an extract from wild-type sporulating cells and incubated it with the purified VM-His6. The mixture was then applied to a Ni2+-agarose affinity column and eluted with imidazole. The results of the immunoblot analysis of Fig. 3 (left side) show that IVA was retained on the VM-His6-containing resin and was eluted with imidazole. As a control, neither a protein (labelled with an asterisk) that cross-reacted with the anti-IVA antibodies nor the non-sporulation protein σA was found in the imidazole eluate. Furthermore, and importantly, when affinity chromatography was carried out with a his-tagged derivative of VMI6A (VMI6A-His6) little or no IVA could be detected in the imidazole eluate (right-side of Fig. 3). [As a control, immunoblot analysis with antibodies against the histidine tag showed that VMI6A-His6 had been efficiently retained on the column (bottom of Fig. 3)] Strictly speaking, these results do not establish that IVA and VM are in direct contact; conceivably, IVA and VM interact indirectly via an unknown bridging protein. Nonetheless, the demonstration of allele-specific suppression in combination with the results of affinity chromatography strongly supports the view that VM and IVA interact directly and do so in part through an I6–G486 contact site.

Figure 3.

Affinity chromatography of IVA using immobilized VM-His6. Extracts of sporulating cells (PY79) were incubated with purified VM-His6 (left) or VMI6A-His6 (right), applied to a Ni2+-agarose column, and eluted. The presence of IVA and, as controls, σA and VM-His6 or VMI6A-His6 was monitored in the load (L), the flow-through (FT), the wash (W) and the elution fractions (E1–E3) by immunoblotting. When blotting for IVA or σA, elution fractions were concentrated 10-fold. The asterisk (*) indicates an unknown protein that reacts non-specifically with anti-IVA antiserum. The slower migration of IVA in the eluates as compared with the load was due to differences in the buffer components as demonstrated by showing that the mobility was indistinguishable when mixtures of the fractions were subjected to electrophoresis (data not shown).

Proper subcellular localization of GFP–IVA depends on VM

The pattern of subcellular localization of IVA is similar to that of VM–GFP, as fluorescence from a GFP–IVA fusion is known to colocalize with the outer forespore membrane (Price and Losick, 1999; Fig. 4A). This observation is consistent with the idea that IVA acts as a receptor for VM at the engulfment membrane. Conversely, however, the proper subcellular localization of GFP–IVA is known to be dependent on VM (Price and Losick, 1999). Evidently then, VM and IVA are dependent on each other for their proper deployment within the sporangium.

Figure 4.

Proper localization of IVA requires VM.
A–F. Localization of GFP–IVA (A) in the wild type (KR160), (B) in a VM deletion mutant (KR178), and in a VM deletion mutant harbouring at thr a wild-type copy of VM (KR165) (C) or VMI6A (KR163) (D). GFP–IVAG486V localization in the presence of either wild-type VM (KR173) (E) or VMI6A (KR171) (F).
G–L. Membranes of cells in (A)–(F) visualized with the dye FM4-64.

In light of these findings, we decided to investigate further the dependence of GFP–IVA localization on VM. Indeed, and as previously reported, GFP–IVA was drastically mislocalized in the absence of VM (Fig. 4B). However, the pattern of GFP–IVA mislocalization was different from that of mislocalized VM–GFP. First, a significant portion of the fluorescence from GFP–IVA was seen in the mother cell cytosol, as opposed to the cytoplasmic membrane. Second, the remaining fluorescence was concentrated in a spot. Such a spot was noticed previously (Price and Losick, 1999), but our present analysis reveals that this spot was characteristically located in close proximity to the polar septum. It therefore appears that GFP–IVA is able to localize to the vicinity of the polar septum independently of VM but requires VM to associate with, and become uniformly distributed along, the outer forespore membrane during engulfment.

The results of Fig. 3 show that this mislocalization was dependent on the I6–G486 contact site between the proteins. Thus, GFP–IVA was severely mislocalized in the presence of VMI6A (Fig. 4D), but a normal pattern of localization was seen in the presence of VMI6A in sporangia producing GFP fused to the IVAG486V mutant (GFP–IVAG486V). Interestingly, the extreme C-terminal residues of IVA had previously been implicated as being required for the proper localization of IVA (Price and Losick, 1999). The results presented here suggest that these residues accomplish this by mediating an interaction with VM. We therefore conclude that GFP–IVA depends on VM to track with the outer forespore membrane during engulfment and that this dependence involves direct interaction between the proteins.

VM associates with the membrane via hydrophobic interactions

Our results so far suggest a model in which VM anchors IVA to the membrane thereby enabling it to track with the outer forespore membrane during engulfment. To investigate this model we sought to identify residues involved in the association of VM with the membrane. We reasoned that at 26 amino acids in length, VM is unlikely to be an integral membrane protein. Rather, it has been suggested that VM is an amphipathic alpha helix wherein all six positively charged residues lie along one face, and all but three hydrophobic residues lie along the opposite face (Prajapati et al., 2000; Fig. 5A). (The three exceptions include I6, which we have shown is a contact site with IVA). Prajapati et al. have hypothesized that VM associates with the negatively charged head groups of the membrane via the positively charged face of the proposed amphipathic helix.

Figure 5.

VM associates with the membrane via hydrophobic residues.
A. Alpha helical model of VM. Panel on the right is rotated 180° along the long axis relative to the panel on the left. Hydrophobic residues are shaded in grey, and positively charged residues are circled in green.
B. Wild-type and mutant forms of VM–GFP in cells engineered to produce the fusion proteins during growth. From the left: wild type (KR318), substitution of residues 2, 7, 10, 24 and 25 with alanine (KR323), substitution with glutamic acid at position 12 (KR333) or 15 (KR335), double substitution with glutamic acid at positions 12 and 15 (KR328), and double substitution with tryptophan at positions 12 and 15 (KR329). Below are the corresponding images of cells stained with the membrane dye FM4-64.
C. Extracts of strains in (B) were separated into soluble and insoluble fractions by centrifugation. VM–GFP and σA were detected by immunoblotting. T, total extract; S and P, supernatant and pellet, respectively, after 100 000 g centrifugation.

We sought to test this hypothesis by examining the effect of amino acid substitutions of the positively charged residues on the localization of VM–GFP. To simplify the analysis and to avoid the potential influence of other sporulation proteins on the membrane association of VM–GFP, we examined the subcellular localization of the fusion protein in cells engineered to produce VM–GFP during growth. Consistent with the idea that VM–GFP associates with the membrane, the results of Fig. 5B (first panel) show that fluorescence was enriched at the periphery of vegetative cells engineered to produce the fusion protein (van Ooij and Losick, 2003) (Fig. 5B, first panel). However, a quintuple mutant in which five positively charged residues were replaced with alanine was not appreciably impaired in its ability to associate with the cytoplasmic membrane (Fig. 5B, second panel).

Many proteins that associate with membranes via amphipathic helices reportedly do so via hydrophobic residues (Johnson and Cornell, 1999; Bechinger, 2000). In such instances, the long axis of the helix is believed to be oriented parallel to the membrane bilayer, with the hydrophobic face of the helix buried within the membrane and the polar face exposed to the cytosol. Such associations are typically disrupted by introduction of charged residues in the hydrophobic face. We wondered whether VM associates with the membrane in this manner. According to the helical projection of VM, residues 12 and 15 (leucine and isoleucine respectively) are predicted to lie well within the hydrophobic face and are furthest away from the polar face (Fig. 5A). The third and fourth panels in Fig. 5B show the localization of VM–GFP, produced in vegetative cells, where either position 12 or 15 was changed to glutamic acid. Evidently, the introduction of a single negative charge in the hydrophobic face was insufficient to entirely disrupt the association of VM–GFP with the membrane. Next, we created a double mutant with glutamic acid at both positions, as transmembrane regions can often tolerate the introduction of a single charged residue (Bechinger, 2000). The fifth panel of Fig. 5B shows that the VML12E,I15E–GFP double mutant was almost entirely cytosolic, exhibiting no measurable enrichment at the cell periphery. To confirm that the inability of VML12E,I15E–GFP to associate with the membrane was due to the introduction of negative charges in the hydrophobic face per se and not due simply to the specific elimination of the leucine and isoleucine residues, we introduced a neutral amino acid (tryptophan) at positions 12 and 15 rather than negatively charged residues. The final panel in Fig. 5B shows that VML12W,I15W–GFP was able to associate with the membrane, indicating that hydrophobic face tolerates neutral, but not polar amino acid substitutions.

As a biochemical approach to investigating the membrane association of VM–GFP, we carried out a fractionation experiment in which soluble proteins were separated from insoluble material in cell extracts by ultracentrifugation. Figure 5C is an immunoblot of extracts prepared from the cells in Fig. 5B in which ‘T’ represents total extract, ‘S’ the supernatant fluid from centrifugation and ‘P’ the pellet. The use of specific antibodies revealed that most of the VM–GFP was in the pellet, whereas, as a control, the cytosolic protein σA was almost entirely found in the supernatant fluid. Similar results were obtained with the quintuple mutant, confirming that membrane association was not mediated by the five positively charged residues on the charged face of the helix (Fig. 5C, second set). In striking contrast, the VML12E,I15E–GFP double mutant protein was found almost entirely in the supernatant fluid. When, however, residues 12 and 15 were replaced with tryptophan, the resulting VML12W,I15W–GFP double mutant was found in the pellet fraction. Taken together, we conclude that positively charged residues do not participate in the anchoring of VM to the plasma membrane. Indeed, in other experiments (data not shown), we were unable to extract VM–GFP from the insoluble fraction with high ionic strength (1 M NaCl or 100 mM Na2CO3; data not shown), further suggesting that the positively charged face was largely dispensable for membrane association. Instead, VM principally associates with the membrane via hydrophobic interactions. Accordingly, VM (but not a control integral membrane protein) was readily extractable from the insoluble fraction with 5 M urea, a chaotropic agent, a finding consistent with the idea that the protein is peripherally associated with the membrane (data not shown). In toto, these results support a model in which VM is an amphipathic helix whose hydrophobic face is buried within the phospholipid bilayer and whose polar face (which, pleasingly, includes the IVA contact site) is exposed to the cytosol.

Discussion

The coat is a complex macromolecular structure consisting of over 25 different gene products that are packaged in a tight shell around the outside of the spore (Driks, 1999). This shell both protects the spore from environmental insults and senses the restoration of favourable environmental conditions under which circumstances the spore can germinate and resume vegetative growth. Despite the identification of many components of the coat, the mechanism by which coat proteins are targeted to the surface of the developing spore is unknown. Here, we have presented evidence that VM acts as a bridge between IVA, which creates the basement layer of the coat, and the outer membrane that surrounds the developing forespore.

The conclusion that VM acts as a bridge rests on two principal findings. First, we obtained genetic and biochemical evidence for an amino acid contact site between residue 6 of VM, which is located on the largely hydrophilic face of the amphipathic helix, and residue 486, which is near the extreme C-terminus of IVA. Both residues are in fact conserved among VM and IVA orthologues in various Bacillus and Clostridium species. We have further shown the proper subcellular localization of VM is dependent on IVA and that, conversely, the localization of IVA is dependent on VM.

The second finding is that VM adheres to the membrane and does so via hydrophobic amino acids on the hydrophobic face of the helix. Cytological and biochemical evidence indicates that leucine 12 and isoleucine 15, but not charged residues on the polar face of the helix, are required for membrane association of VM. The simplest interpretation of these results is that the non-polar face of the helix is oriented parallel to, and buried within, the phospholipid bilayer. This is opposite to a model in which the positively charged polar face of the helix is parallel to and interacts with the negatively charged head groups of the phospholipid bilayer (Prajapati et al., 2000). Our proposal for how VM inserts into the membrane is reminiscent of that for several other proteins that associate with phospholipid bilayers via amphipathic helices (Craig et al., 1994; Andersen et al., 1997; Garner et al., 1998; Johnson and Cornell, 1999; Ford et al., 2002). For example, at least two members of the Arf family of proteins (Arf1 and Arf6) that regulate vesicle biogenesis during intracellular traffic harbour an N-terminal amphipathic alpha helix (Pasqualato et al., 2002). When Arf1 and Arf6 are bound to GTP, this alpha helix is exposed and its hydrophobic face is embedded in the membrane. In its inactive GDP-bound state, Arf1 and Arf6 sequester this N-terminal alpha helix so that it is unavailable to interact with the membrane. Similarly, the bacterial cell division protein MinD, in its ATP-bound state, exposes a C-terminal amphipathic helix whose hydrophobic residues along one face penetrate into the hydrophobic interior of the phospholipid bilayer (Zhou and Lutkenhaus, 2003).

Amphipathic alpha helical membrane association domains typically exhibit two traits. First, they are unordered structures until they interact with phospholipids (Johnson and Cornell, 1999). Accordingly, circular dichroism spectroscopy has revealed that VM exhibits alpha helical characteristics only in the presence of phospholipids or organic solvents (Prajapati et al., 2000). Second, this motif is typically employed by proteins that shuttle on and off the membrane. To date, it is unclear whether VM exhibits this shuttling feature. Our results suggest that during sporulation, VM is exclusively and stably associated with the outer forespore membrane. It is conceivable, though, that a population of VM molecules does in fact shuttle between the cytosol and the membrane during sporulation to recruit additional IVA molecules to the polar septum and that our methods simply do not detect this population. Alternatively, perhaps the ready potential of VM to be displaced from the plasma membrane allows for the rapid removal of the basement layer of the spore coat during germination. In any case, it appears that the basement layer of the spore coat, and by extension, the rest of the spore coat, is anchored, at least initially, to the surface of the outer forespore membrane by a small amphipathic peptide.

In toto, our results suggest the following ‘reciprocal localization’ model for the subcellular positioning of VM and IVA to the outer forespore membrane (Fig. 6). We propose that VM serves as membrane anchor that tethers IVA (and hence the entire spore coat for which IVA creates a basement layer) to the outer membrane that surrounds the forespore. Our evidence suggests that the hydrophobic face of VM becomes buried in the lipid bilayer of the membrane. Meanwhile, at least one hydrophobic residue on the opposite, polar face of the helix, isoleucine six, directly contacts a residue near the C-terminus of IVA, thereby indirectly tethering IVA to the membrane. This is a reciprocal localization model in that proper localization of VM depends on IVA. Evidently, IVA contributes positional information to the IVA–VM complex, but the nature of this positional information is unknown. An important challenge for the future will be to elucidate the nature of the positional signals that cause the proper localization of the IVA–VM complex.

Figure 6.

Reciprocal interaction model for the localization of VM and IVA. A sporangium is depicted in which the mother cell is engulfing the forespore. Above, an expansion of the engulfing membrane is shown. The amphipathic alpha helical VM is depicted such that it lies parallel to the patch of membrane that is contiguous with the membrane surrounding the mother cell. The hydrophobic face of VM is buried within the phospholipid bilayer. Residue six of VM lies on the exposed polar face of the alpha helix and contacts residue 486 near the C-terminus of IVA, thereby tethering the complex to the engulfing membrane.

Experimental procedures

Strain construction

Strains are otherwise isogenic derivatives of B. subtilis PY79 (Youngman et al., 1984). B. subtilis competent cells were prepared as described previously (Wilson and Bott, 1968). The spoVM–gfp fusion at amyE was created by cloning spoVM open reading frame and 400 bases of upstream sequence into pKL147 as described previously (van Ooij and Losick, 2003) to create pKC1. To place spoVM–gfp at amyE, the liberated EcoRI–HindIII fragment from pKC1 was cloned into the integration vector pDG1662 (Guerout-Fleury et al., 1996) to create pKC2. spoVM-FLAG at amyE was created by PCR amplifying the spoVM region using primers VMProm5′Eco and VMFLAG3′Bam (primer sequences are listed in Table S1), digesting the PCR product with EcoRI and BamHI, and cloning into the integration vector pDG1662 to create pKR95. The gfp–spoIVA fusion at thrC was created by first amplifying sequences upstream of the spoIVA reading frame using primers IVAprom5′Eco and IVAprom3′Hind, digesting with EcoRI and HindIII, and cloning into the integration vector pDG1731 (Guerout-Fleury et al., 1996), to create pIVAprom. Next, spoIVA was PCR amplified using primers IVA5′Xho and IVA3′Bam, and digested with XhoI and BamHI. gfp was PCR amplified using primers gfp5′Hind and gfp3′Xho, and digested with XhoI and HindIII. The two digested products were then cloned into pIVAprom to create plasmid pKR5. spoIVA was placed at thrC by PCR amplifying the spoIVA open reading frame and upstream sequences using primers (IVAprom5′Eco and IVA3′Bam), digested with EcoRI and BamHI, and cloned into pDG1731. All plasmids were introduced into B. subtilis by double recombination.

Site-directed mutations in spoVM–gfp were created using the Quikchange site-directed mutagenesis kit (Stratagene) using complementary primers that harboured the desired mutation flanked by approximately 15 bases on either side and using pKC2 as the template. Primers were evaluated on http://tim.saraogtim.com/molbio/qctemp.php. Mutagenesis was confirmed by DNA sequencing. Mutant alleles of spoVM–gfp were placed under the control of the IPTG-inducible promoter Phyperspank by PCR amplification using primers VMRBS5′Hind and gfp3′Nhe, digesting with HindIII and NheI, and cloning into pDR111 (gift of David Rudner) and introduced at amyE by double recombination.

Plasmids pKR3 and pKR23, harbouring spoVM-His6 and spoVMI6A-His6, respectively, under the control of the T7 promoter, were created by PCR amplifying the spoVM open reading frame with primers VM5′Nde or VMI6A5′Nde and VM3′His6Bam, and cloning into pET29a (Novagen). pKR3 and pKR23 were transformed into Escherichia coli BL21(DE3).

Isolation of spontaneous suppressors

Strain CVO1402, harbouring spoVMI6A as the only copy of spoVM, was grown in 100 ml of Difco sporulation medium (DSM) supplemented with 5 μg ml−1 chloramphenicol for 24 h at 37°C in order to accumulate spontaneous mutations and sporulate. Forty millilitres of this culture was removed and incubated at 80°C for 20 min in order to kill any cells that had not completed sporulation. Thirty-three millilitres of the heat-killed culture was then diluted into 300 ml of fresh DSM/Cm and allowed to sporulate for 24 h as above. The procedure was repeated two more times, and sporulation efficiency was calculated as the fraction of heat-resistant (80°C for 20 min) colony-forming units.

Co-purification of IVA with VM-His6

SpoVM-His6 and SpoVMI6A-His6 were purified to homogeneity from BL21(DE3) pKR3 and BL21(DE3) pKR23, respectively, as follows. Overnight cultures grown at 37°C in LB supplemented with 50 μg ml−1 kanamycin for plasmid maintenance were diluted 1:50 into 1 l of LB/Kan and grown at 37°C for 2 h. IPTG was added to 1 mM to induce expression of the his-tagged construct and the culture was grown at 37°C for 3 h. Harvested cells were resuspended in 30 ml of buffer B (8 M Urea, 0.1 M NaH2PO4 and 10 mM Tris at pH 8.0), sonicated, and incubated on ice for 30 min to solubilize inclusion bodies. The lysate was forced through a 21-gauge needle twice to reduce viscosity, and centrifuged at 15 000 g to pellet cell debris. Cleared lysate was then placed on 500 μl of (bed volume) Ni2+-NTA agarose (Qiagen), washed with 50 ml of wash buffer (50 mM Tris at pH 7.5, 150 mM NaCl, 0.1% Triton X-100 and 50 mM imidazole), and eluted with wash buffer containing 200 mM imidazole. Purity was assessed by separating load, flow-through, wash and elutions by 15% PAGE and detecting with Coomassie stain. Protein concentration was determined with Coomassie Plus Bradford kit (Pierce).

Overnight cultures of PY79 grown at 22°C were diluted into 300 ml of CH medium (Sterlini and Mandelstam, 1969) and induced for sporulation by resuspension of exponential-phase cells in SM medium. Harvested cells were resuspended in 20 ml of buffer A (50 mM Tris at pH 7.5, 150 mM NaCl, 0.2 mg ml−1 lysozyme) and incubated at 37°C for 10 min. Triton X-100 was added to 0.1%, and the lysate was incubated on ice for 10 min. Lysate was cleared of debris as described above.

Nine millilitres of PY79 cleared lysate was incubated with 1 ml of purified SpoVM-His6 or SpoVMI6A-His6 (5 μM final concentration of peptide) for 1 h at 4°C, gently tumbling. Each binding reaction was placed onto Ni2+-NTA agarose as above, washed with 50 ml of wash buffer, and eluted as above (3 × 1 ml fractions). IVA and σA were detected in the various fractions by immunoblotting using specific antisera. VM-His6 and VMI6A-His6 were detected by antisera directed against the His6 epitope (GeneTex). For the detection of IVA and VM, elution fractions were concentrated 10-fold by precipitating with trichloroacetic acid and resuspension in sample buffer.

Immunoblot analysis of VM–GFP solubility

Overnight cultures of KR318, KR323, KR328 and KR329 were diluted 1:20 into 20 ml of CH medium and grown for 90 min at 37°C. IPTG was added to 1 mM and the cultures were incubated for 1 h at 37°C. Harvested cells were lysed and fractionated as described previously (Ellermeier et al., 2006). Briefly, cells were suspended in 6 ml of SM buffer (0.5 M sucrose, 20 mM MgCl2, 10 mM potassium phosphate at pH 6.8, and 0.1 mg ml−1 lysozyme) and incubated for 10 min at 37°C to form protoplasts. The protoplasts were harvested at 15 000 g, and suspended in 6 ml of lysis buffer (250 mM EDTA, 10 mM potassium phosphate at pH 6.8, 50 mM NaCl) and vortexed to lyse protoplasts. Cell debris was pelleted as described above, and 50 μl was removed from the supernatant for analysis and labelled ‘T’ for total cell extract. Four millilitres of the supernatant was removed and centrifuged at 100 000 g to pellet membranes. Fifty microlitres was removed from the supernatant for analysis and labelled ‘S’ for 100 000 g supernatant. The pellet was resuspended in 4 ml of buffer B (above), and 50 μl was removed for analysis and labelled ‘P’ for 100 000 g pellet. VM–GFP was detected in the various fractions by immunoblotting using antisera raised against GFP.

Microscopy

Cells were harvested and resuspended in PBS containing 1 μg ml−1 of the membrane dye FM4-64 (Molecular Probes). Three microlitres of the suspension was placed on a glass microscope slide and cells were immobilized using a freshly prepared coverslip treated briefly with poly l-lysine. The equipment used and analysis of images were as previously described (Fujita and Losick, 2002). Fluorescence was quantified using Image J software that calculated the vertically averaged pixel intensity across the cross-section of single-cell images and plotted them as a function of distance along the long axis of the cell. Quantifying the area under the curve corresponding to the forespore or mother cell revealed the pixel intensity at those locations. The ratio of total pixel intensity of the forespore to the total intensity of the cell is reported as ‘per cent forespore localization’, along with standard deviation. Reported percentages are averages of 10 representative cells. Immunofluorescence microscopy was performed with strains KR359 and KR375 using rabbit anti-FLAG antibodies as previously described (Teleman et al., 1998).

Acknowledgements

We thank members of the laboratory for helpful discussions. This work is supported by Ruth L. Kirschstein National Research Service Award GM072408 to K.S.R. and NIH Grant GM18568 to R.L.

Ancillary