|Home | About | Journals | Submit | Contact Us | Français|
Endospores formed by Bacillus subtilis are encased in a tough protein shell known as the coat, which consists of at least 70 different proteins. We investigated the process of spore coat morphogenesis using a library of 40 coat proteins fused to GFP and demonstrate that two successive steps can be distinguished in coat assembly. The first step, initial localization of proteins to the spore surface, is dependent on the coat morphogenetic proteins SpoIVA and SpoVM. The second step, spore encasement, requires a third protein, SpoVID. We show that in spoVID mutant cells, most coat proteins assembled into a cap at one side of the developing spore but failed to migrate around and encase it. We also found that SpoIVA directly interacts with SpoVID. A domain analysis revealed that the N-terminus of SpoVID is required for encasement and is a structural homolog of a virion protein, whereas the C-terminus is necessary for the interaction with SpoIVA. Thus, SpoVM, SpoIVA and SpoVID are recruited to the spore surface in a concerted manner and form a tripartite machine that drives coat formation and spore encasement.
Elucidating the molecular mechanisms governing the formation of supramolecular structures is a fundamental objective in developmental biology. The challenge of controlling the assembly of dozens of proteins into large organized functional structures is shared by all organisms. Examples include cellular machines as diverse as the bacterial flagellum (McCarter, 2006), the divisome (Goehring and Beckwith, 2005), eukaryotic nuclear pore complexes (Alber et al., 2007), clathrin-coated vesicles (Kaksonen et al., 2005) and various envelopes, including exoskeletons (Frand et al., 2005) and viral capsids (Dokland et al., 1997). Though diverse in structure and function, all of the machines above must be assembled in a defined order and their assembly must be co-ordinated with various gene expression programs, including development, cell cycle and stress responses. Here, we are interested in the morphogenesis of the B. subtilis spore coat, where more than 70 different proteins are assembled during spore development into a complex multi-layered structure that encases the spore (Driks, 1999; Driks, 2002a; Henriques et al., 2004; Henriques and Moran, 2007; Kim et al., 2006; Kuwana et al., 2002; Lai et al., 2003).
Endospores (hereafter spores) formed by Bacillus and Clostridium species are metabolically dormant cells characterized by their extreme resistance to heat, radiation and chemicals (Nicholson et al., 2000). In Bacillus species, spore formation (sporulation) is initiated in response to nutrient limitation. However, since nutrients are required for sporulation of Clostridium species, the mechanisms of sporulation initiation differ between the two genera (Paredes et al., 2005; Stephenson and Hoch, 2002; Woods and Jones, 1986). Although spores can remain dormant for extended periods of time (Nicholson, 2003), spore germination in Bacillus species occurs as soon as a sufficient amount of nutrients is available to sustain cell growth (Setlow, 2003). An impaired coat structure undermines several of the spore resistance and germination properties (Henriques et al., 2004). By creating a physical barrier, the coat protects the spore against digestion by eukaryotic predators (Klobutcher et al., 2006; Laaberki and Dworkin, 2008) and large molecules, such as lysozyme. A second function of the coat is to mediate proper recognition and access of nutrients to specific receptors found on the inner spore membrane (Setlow, 2003). The coat further facilitates germination by housing cell wall hydrolases that are required during germination for degradation of the spore peptidoglycan (Lambert and Popham, 2008; Ragkousi et al., 2003).
Sporulation has been studied in a great amount of detail in the model organism B. subtilis (Eichenberger, 2007; Errington, 2003; Piggot and Coote, 1976; Piggot and Losick, 2002; Stragier and Losick, 1996). Shortly after initiation of sporulation, an asymmetric division separates the sporulating cell, generating two cellular compartments: a large mother cell and a smaller forespore. Asymmetric division is followed by a phagocytosis-like process, where the forespore is engulfed by the mother cell, creating a protoplast surrounded by a double membrane. The spore peptidoglycan (cortex) is synthesized between the two membranes. Spore coat assembly begins shortly after asymmetric division and continues until the coat encases the entire spore. This event marks the completion of the maturation process and the mother cell lyses, releasing the spore in the environment.
Spore coat morphogenesis is a complex process in B. subtilis (Driks, 1999; Henriques and Moran, 2007). Coat proteins are all produced in the mother cell under the control of the σE and σK factors and their identification was first achieved by reverse genetics (Bourne et al., 1991; Cutting et al., 1991; Donovan et al., 1987; Naclerio et al., 1996; Sacco et al., 1995; Zhang et al., 1993; Zheng et al., 1988). A second group of coat proteins was identified more recently based on gene expression profiles and proteomics-based approaches (Eichenberger et al., 2003; Henriques et al., 1995; Henriques et al., 1997; Kuwana et al., 2002; Lai et al., 2003; McPherson et al., 2005). Since the determination of the σE and σK regulons by transcriptional profiling (Eichenberger et al., 2004; Steil et al., 2005), it has become possible to systematically characterize mother cell genes of unknown function and identify spore coat proteins fused to the green fluorescent protein (GFP) based on their characteristic subcellular localization pattern (Eichenberger et al., 2003; Kim et al., 2006).
Spore coat assembly is initiated by genes under the control of σE, shortly after asymmetric division and continues after engulfment of the forespore by the mother cell, with late gene expression under the control of σK. Assembly of spore coat proteins into a supramolecular structure is controlled by a small group of proteins, the coat morphogenetic proteins (Driks, 1999; Henriques and Moran, 2007). Two morphogenetic proteins sit at the top of the assembly pathway: SpoVM (hereafter VM) and SpoIVA (hereafter IVA). VM, a short amphipatic α helix, has the ability to bind to positively curved membranes, including those that are formed at the beginning of the engulfment process (Ramamurthi et al., 2009), and is essential for spore coat assembly (Levin et al., 1993). IVA is required for attachment of the spore coat (Piggot and Coote, 1976; Roels et al., 1992) and restriction of VM to the outer forespore membrane (Ramamurthi et al., 2006). Conversely, the pattern of subcellular localization of a GFP fusion to IVA is impaired in VM cells (Price and Losick, 1999). IVA can self-assemble into higher order structures in an ATP-dependent manner (Ramamurthi and Losick, 2008). In addition to IVA and VM, the main coat morphogenetic proteins in B. subtilis also include SpoVID (Beall et al., 1993; Driks et al., 1994), SafA (Ozin et al., 2000; Takamatsu et al., 1999) and CotE (Zheng et al., 1988).
In this report, we are primarily concerned with SpoVID (hereafter VID), its interaction with IVA and VM and their concerted roles in directing the deposition of every coat protein to the spore surface. Some aspects of the VID phenotype are reminiscent of IVA mutants, which led the original studies to conclude that VID played a role similar to IVA in anchoring the coat to the spore surface (Beall et al., 1993). However, IVA mutants differ from VID mutants in the sense that IVA spores are devoid of cortex. We show here that the roles of IVA and VID in coat assembly are also distinct. In contrast to IVA, VID is dispensable for initial localization of coat proteins to the spore surface, but essential for a late stage in coat morphogenesis that we define as spore encasement. Our conclusion is based on fluorescence microscopy experiments, where we recorded the patterns of subcellular localization for 39 coat protein-GFP fusions in VID cells. We also engineered a series of targeted VID in-frame deletion mutants and demonstrated that spore encasement is dependent on the N-terminal domain of VID that, according to a consensus fold-recognition structure prediction method, could be a structural homolog of a phage coat protein. Our deletion analysis also identified two adjacent regions at the C-terminus of VID, including a short conserved region that directly interacts with IVA, that are involved in the recruitment of VID itself to the forespore surface. Finally, we propose a two-step model of spore coat assembly. The first step, the initial localization of coat proteins to the spore suface is primarily controlled by IVA. The second step, which we name encasement, is the transition from an asymmetric cap of fluorescence on one side of the developing spore to a symmetric distribution of individual coat proteins around the circumference of the spore. Encasement is driven by the concerted action of VM and VID. Hence, the main stages in coat assembly are specified by the assembly of a tripartite molecular machine, formed by VM, IVA and VID at the surface of the developing spore.
To examine genetic interactions in the subcellular localization of IVA, VM and VID to the surface of the forespore, we imaged IVA-GFP, VM-GFP and VID-GFP fusions in sporulating cells in all combinations of fusions and deletions (Fig. 1). A table listing all strains is supplied in the Supporting Information (Table S1), as well as a table providing a detailed quantification of the localization patterns (Table S2). Our IVA-GFP and VID-GFP fusions were functional in spore germination assays that were used as an indirect measurement of spore coat integrity (Fig. S1). We did not assay the functionality of the VM-GFP fusion, as it had been previously characterized (van Ooij and Losick, 2003).
In otherwise wild type sporulating cells, green fluorescent signals corresponding to VID-GFP (Fig. 1A, first row), VM-GFP (Fig. 1B, first row), IVA-GFP (Fig. 1C, first row) appeared shortly after asymmetric division of the sporangium, 2 hours after initiation of sporulation by suspension in Sterlini-Mandelstam (SM) medium (Sterlini and Mandelstam, 1969). During the engulfment process, the VID-GFP and VM-GFP signals overlapped with the curved septum revealed by membrane staining (FM4-64, red). By contrast, IVA-GFP did not immediately track along the engulfing membranes. After engulfment was completed (hour 4), the red signal around the forespore was lost because the impermeability of the plasma membrane prevented staining of the forespore membranes (Pogliano et al., 1999). By that time, VID-GFP and VM-GFP had formed continuous rings of uniform intensity at the surface of the forespore (shells in three dimensions). However, the IVA-GFP signal consisted of two strongly staining polar caps that did not appear to be connected. In electron micrographs of thin-sectioned spores (Driks, 1999), the coat often appears to be thicker at the poles. It is possible that these differences in thickness may be a consequence of a non-uniform distribution of IVA around the spore. Nevertheless, we also note that in merodiploid strains (i.e cells expressing the native copy of spoIVA in addition to the GFP fusion), a more uniform ring is obtained (Price and Losick, 1999; Ramamurthi and Losick, 2008) indicating that small differences in protein levels and/or functionality of GFP fusions can directly impact the completeness of the fluorescent rings.
We found that both IVA and VM were essential for normal VID-GFP subcellular localization. However, their respective contributions to the VID-GFP localization process were distinct. In IVA cells (Fig. 1A, middle row), recruitment of VID-GFP to the forespore surface was severely impaired. These cells exhibited a diffuse fluorescent signal in the mother cell cytoplasm with some enrichment of the signal near the septum (hour 2). At the later time point (hour 4), additional regions of signal enrichment were sometimes observed at various locations in the mother cell cytoplasm. These regions may correspond to aggregated coat proteins. Our observations are consistent with the previously characterized role of IVA in anchoring coat proteins to the spore surface (Piggot and Coote, 1976; Roels et al., 1992). Furthermore, the dependency on IVA for VID subcellular localization had already been recognized in immunofluorescence microscopy experiments carried out by Ozin et al. (2001).
The VID-GFP localization pattern was also modified in VM cells (Fig. 1A, last row), but in a largely different manner than in IVA cells. In comparison to IVA cells, a brighter, crescent-shaped, VID-GFP signal was noted at the surface of the forespore in VM cells, with minimal diffusion into the mother cell cytoplasm. However, no complete VID-GFP fluorescent rings were ever observed. These observations imply that the initial recruitment of VID-GFP to the forespore surface is more directly correlated with the presence of IVA than it is with that of VM. Nevertheless, VM is necessary for the completion of encirclement of the forespore by VID-GFP.
Next, we determined if VID was required for VM-GFP and IVA-GFP subcellular localization. VM-GFP localization was unaffected in VID cells (Fig. 1B, last row), a result that was anticipated considering that VM can localize to curved membranes in vitro (Ramamurthi et al., 2009). However, IVA-GFP localization was incomplete in VID cells (Fig. 1C, last row). Similarly to the localization of VID-GFP in VM cells reported above, a crescent-shaped fluorescent signal was observed on the mother cell proximal side of the forespore, but no fluorescence was detected on the mother cell distal side. An equivalent pattern has been described for GFP-IVA in VM cells (Ramamurthi et al., 2006) and was observed under our experimental conditions as well (Fig. 1C, middle row). Thus, VM and VID are both necessary for the transition of IVA-GFP from a single cap to a full ring of fluorescence. Considering that VM-GFP subcellular localization is independent of VID, we infer that VM is located upstream of VID in the genetic hierarchy controlling ring formation. However, since IVA is necessary for the proper localization of both VID-GFP (see above) and VM-GFP (Fig. 1B, middle row), we placed it at the top of the genetic hierarchy controlling the targeting of coat proteins to the forespore surface. As previously demonstrated, IVA is involved in restricting VM-GFP to the outer forespore membrane (Ramamurthi et al., 2006).
Taken together, we interpret our data and previously published experiments to indicate that two successive steps can be distinguished in the coat morphogenesis process. The first step is the targeting of coat proteins, including VID and VM, to the forespore surface, a process primarily controlled by IVA (Fig. 1D, first row). The second step, that we have named spore encasement, is the formation of a full shell of spore coat proteins around the circumference of the spore (see below). This later morphogenetic transition is controlled by VM and VID, which act in part by promoting assembly of IVA around the spore (Fig. 1D, second row). Our results indicate that these two processes are distinct phenomena that can be genetically uncoupled.
In order to investigate further the process of spore encasement, we analyzed the subcellular localization of individual spore coat proteins fused to GFP in VID, VM and IVA cells. Our library of coat protein-GFP fusions consists, in addition to VID-GFP, IVA-GFP and VM-GFP, of 37 constructs for a total of 40 fusions. These experiments allowed us to identify which proteins required VID, VM or both genes for spore encasement. The complete results of our analysis are displayed in Figs. S2 (for VID dependency) and S3 (for VM dependency). Selected examples are presented in Fig. 2 (which also includes data for IVA dependency), with a detailed quantification provided in Table S3.
As observed above for IVA-GFP, we noticed that most coat protein-GFP fusions were still able to anchor at the forespore surface in VID and VM cells. However, in most cases, the final morphogenetic transition was not completed, in the sense that no complete fluorescent rings were ever obtained. For a majority of fusions, a single fluorescent polar cap was visible on the mother cell proximal side of the forespore, whereas a fluorescent signal rarely (if ever) appeared on the distal side of the forespore. In VID cells, this characteristic pattern of a spore encasement defect was observed for IVA-GFP (Fig. 1C), CotE-GFP (Fig. 2A), CotO-GFP (Fig. 2B) and YaaH-GFP (Fig. 2C), as well as CotA-GFP, CotB-GFP, CotD-GFP, CotG-GFP, CotM-GFP, CotS-GFP, CotT-GFP, CotU-GFP, CotW-GFP, CotZ-GFP, Tgl-GFP, YheD-GFP, YhjR-GFP, YisY-GFP, YknT-GFP, YmaG-GFP, YncD-GFP, YtxO-GFP, YutH-GFP, YuzC-GFP and YybI-GFP (Fig. S2). An identical pattern was observed in VM cells (Figs. 2A-C and S3). By contrast, in IVA cells (Fig. 2A-C), morphogenesis was blocked at an earlier stage, as exemplified by the presence of single dots of fluorescence at various locations in the mother cell cytoplasm. Thus, these data suggest that, for this large group of coat proteins, VID and VM are both necessary for mediating spore encasement, whereas IVA is required for initial localization to the surface of the forespore.
A second group of fusions also presented distinct alterations of their localization pattern in VID cells. In contrast to the previous category, a fluorescent signal was usually observed on both sides of the forespore. However, similarly to the first group, ring morphogenesis was incomplete, as exemplified by the presence of several unconnected small dots of fluorescence encircling the maturing forespore. No strong decrease of the fluorescence signals was noted in VID cells, thus arguing against degradation of unassembled fusion proteins. Furthermore, the absence of diffuse fluorescence in the mother cell of VID cells suggests that the initial targeting of these fusions to the forespore surface was independent of VID. This group comprises SafA-GFP (Fig. 2D), CotP-GFP, CotQ-GFP, OxdD-GFP, YeeK-GFP, YjqC-GFP, YjzB-GFP, YlbD-GFP, YppG-GFP and YxeE-GFP (Fig. S2). By contrast, in VM cells (Figs. (Figs.2D2D and S3), assembly was, once again, blocked at the single cap stage, whereas in IVA cells, morphogenesis was blocked at an earlier stage (Fig. 2D). These results suggest that for these fusions VM is the main factor driving encasement, even though the contribution from VID remains necessary.
Finally, a third group is composed of fusions to VM (Fig. 1B), YhaX (Fig. 2E), LipC (Masayama et al., 2007) and YsnD (Kim et al., 2006) that form complete fluorescent rings in VID cells (Fig. S2). However, in VM cells (Figs. (Figs.2E2E and S3), ring formation of YhaX-GFP, LipC-GFP and YsnD-GFP was impaired, suggesting that, for these fusions, VM is driving encasement independently of VID (and possibly with the help of another coat protein that remains to be identified). In IVA cells, YhaX-GFP produced a diffuse fluorescent signal that contrasts with the punctuate pattern that was observed for most coat protein fusions in that background (Fig. 2E). Taken together, these data suggest that VM is at the top of a hierarchy regulating encasement, while for 35 out of 39 coat proteins VID is also contributing to encasement.
In order to examine in more detail the role of VID in the encasement step, we carried out a time course analysis of the subcellular localization of CotE-GFP in wild type, VID and IVA cells (Fig. 3 and Table S4). CotE is a morphogenetic protein located at the interface between the inner and outer coat layers (Driks et al., 1994) and required for assembly of the outer spore coat layer (Zheng et al., 1988). Previous studies using immunoelectron microscopy have suggested that VID was required for the maintenance of the CotE ring around the forespore (Driks et al., 1994). We observed that recruitment of CotE-GFP to the surface of the forespore begins as soon as the protein is produced. The CotE-GFP fluorescent signal can be detected initially as a single dot of fluorescence, at or near the center of the curved septum (hour 1.5). Importantly, and in contrast to VID-GFP and VM-GFP, the CotE-GFP protein does not seem to track along the engulfing membranes. Even in otherwise wild type cells, there is a post-engulfment stage when the assembly of CotE-GFP stalls at the single mother cell proximal polar cap stage (hour 2). Up to that stage, the spatio-temporal pattern of subcellular localization of CotE-GFP was identical in wild type (Fig. 3A) and VID cells (Fig. 3B). In wild type cells, a fluorescent signal appeared at the distal pole at hour 2.5. This second polar cap was fully extended at the next time point (hour 3) and the ring began to close at hour 4. However, encasement did not occur in VID cells and we never observed a complete ring of CotE-GFP fluorescence, even after prolonged incubation of the corresponding culture. By contrast, in IVA cells (Fig. 3C), the fluorescence pattern never progressed beyond the single dot stage, again illustrating the distinct roles that IVA and VID play in coat assembly. Thus, we propose that the main function of VID in CotE-GFP localization, as well as the localization of every other coat protein with a similar dependence on VID, is to drive the transition from initial localization of the protein to encasement, the formation of a complete ring. In the experiments presented below, we will use this property as a morphogenetic marker to assess the functionality of VID in-frame deletion mutants.
Next, we sought to determine which parts of VID were required for its morphogenetic properties. Amino acid (a.a.) sequence comparisons of VID orthologs in endospore-forming bacteria revealed regions of conserved sequence at the N- and C-termini of the protein, whereas the central region of the protein is highly variable in length and sequence composition (Fig. S4).
Using FFAS-03 (Jaroszewski et al., 2005), a consensus fold-recognition structure prediction method, we obtained two strong hits (FFAS > 200) for the N-terminal sequence of VID (hereafter N-domain, a.a. 1 to 137): to the coat protein for the bacteriophage PP7 (Fig. S5A) (Tars et al., 2000) and to the lipocalin-like fold (Fig. S5B) (Grzyb et al., 2006). The main difference between these fold predictions is the last secondary structure element of the N-domain, which would be a helix in the phage protein prediction and a strand in the lipocalin prediction.
The C-terminal region of VID includes a LysM domain between a.a. 525 and 575 (Costa et al., 2006). LysM domains have the ability to bind peptidoglycan and have been identified in several other bacterial proteins, including coat proteins SafA and YaaH (Buist et al., 2008; Ozin et al., 2000; Takamatsu et al., 1999). It has been suggested that the LysM domain facilitates targeting of VID by binding to cell wall material located in the space between the two spore membranes or to cell wall precursors, even before the cortex is formed (Costa et al., 2006; Henriques and Moran, 2007; Ozin et al., 2000; Ozin et al., 2001; but see the Discussion). However, the region of sequence conservation at the C-terminus extends beyond the LysM domain and includes residues 500 to 524. Hereafter, we refer to this second region of conservation as region A (because we show below that it is essential to the interaction with IVA).
In order to investigate the contribution of the N-domain to coat morphogenesis, we designed an experimental system where encasement by a CotE-YFP fusion was used as a reporter for VID morphogenetic activity. We generated a series of in-frame deletions in VID where every construct (which were all driven by the native VID promoter) was fused to the coding sequence of cfp and integrated at the non-essential amyE locus in a VID null strain. To control for proper expression and localization of the resulting VID-CFP fusions, we carried out fluorescence microscopy and Western blotting experiments at hourly intervals after initiation of sporulation. For each construct, we determined the localization of the CotE-YFP fusion, which was expressed from the native cotE locus. On Western blots using anti-GFP antibodies, bands of the expected sizes were obtained for all of the constructs with minimal differences in signal intensities (Fig. S6A).
All of the deletions that were generated in the N-domain (Δ24-136, Δ86-136 and Δ125-136) recapitulated the phenotype of the null mutant strain, i.e. CotE-YFP localization was blocked at the single cap stage (Fig. 4). Importantly, the initial localization to the forespore surface of the corresponding VID-CFP fusions was not affected, either temporally or spatially, implying that the recruitment of VID to the forespore surface is independent of residues 24 to 136. We also determined the localization of a few other YFP fusions from our library (to IVA, YtxO, CotP and SafA) in the strain that expresses the SpoVIDΔ86-136-CFP fusion (Fig. S7) and observed disrupted localization patterns that are consistent with those reported in Figs. Figs.22 and S2, confirming that the morphogenetic function of the N-domain was not limited to CotE.
In addition, we analyzed the function of the central region of VID (hereafter region M, where M stands for middle region). We observed that the morphogenetic activity of the protein is largely maintained when extended fragments of region M (Δ140-424 and Δ140-484) are missing (Figs. S6B and S8). Some minor changes in activity were noted for the largest deletion construct (VIDΔ140-484-CFP). The shape of both the VID-CFP and CotE-YFP rings was perturbed and the time necessary to complete ring formation extended by approximately an hour. Although further studies will be necessary to clarify the role of the central region in spore coat assembly, our data show that most of region M is dispensable for spore encasement.
Next, we attempted to determine the respective roles of the two conserved regions at the C-terminus of the protein: the LysM domain (from a.a. 525 to 575) and region A (from a.a. 500 to 524). Three C-terminally truncated VID-CFP constructs were generated: a VID1-525-CFP construct, which only removes the LysM domain; and two constructs (VID1-500-CFP and VID1-484-CFP) that remove both the LysM domain and region A. In contrast to the N-terminus, we find that the C-terminus appears to be necessary for the initial localization of VID-CFP itself to the spore surface.
The localization of each construct to the forespore surface was impaired, as indicated by the presence of a strong and diffuse fluorescence signal in the mother cell cytoplasm (Fig. 5). Western blot analyses revealed fusions of the expected sizes (Fig. S6C), confirming that the observed fluorescence pattern was due to the deletion within the fusion proteins and not to the inadvertent cleavage of the CFP moiety. When the deletion was limited to the LysM domain, the mislocalization was less pronounced and a significant enrichment of the signal was still noted at the forespore surface (on both sides of the forespore, although the mother cell distal cap is less extended than the mother cell proximal cap). By contrast, no enrichment at the forespore surface was noted for either VID1-500-CFP or VID1-484-CFP, suggesting that all of the necessary residues involved in localization to the forespore surface had been lost.
In the case of the VID1-525-CFP fusion, it seems that a sufficient amount of the morphogenetic protein localized properly to permit some signal from CotE-YFP to appear at the mother cell distal forespore pole, suggesting a residual amount of encasement activity (Fig. 5). By contrast, in the cases of VID1-500-CFP and VID1-484-CFP, encasement was prevented to a level comparable to what was observed in the VID null strain. Thus, similarly to the N-domain, the C-terminal region of VID is necessary for spore encasement. However, unlike the N-domain, this may be an indirect effect caused by the failure of the morphogenetic protein to reach its proper subcellular location. Altogether, results from Fig. 5 imply that the C-terminal region is essential for localizing VID to the forespore surface with a more significant contribution of region A, than from the LysM domain.
In order to investigate the contribution of the 484-525 region independently of the LysM domain, we generated two additional in-frame deletion constructs. The first construct (VIDΔ500-525-CFP) removes region A, whereas the second construct (VIDΔ484-500-CFP) leaves region A intact. Western blots for each construct are displayed in Fig. S6D. The VIDΔ500-525-CFP construct showed a diffuse fluorescent signal in the mother cell cytoplasm and some preferential localization at the surface of the forespore (Fig. 5). However, contrary to the truncation of the LysM domain, no localization of VIDΔ500-525-CFP was observed on the mother cell distal side of the forespore. As a consequence of this incomplete localization pattern, CotE-YFP ring formation was stalled at the initial localization stage. In contrast, the deletion of a.a. 484 to 500 had no observable effect, demonstrating that this region was dispensable for VID function, at least under our experimental conditions.
Taken together, our data indicate that the C-terminal region of VID constitutes a targeting signal, which comprises two components: the LysM domain and region A. Both elements contribute to the targeting of VID.
Since subcellular localization of VID is strictly dependent on IVA in fluorescence microscopy experiments (Fig. 1A), we decided to determine whether VID and IVA interacted with one another in biochemical assays. In a first series of experiments, we used GST fusions to VID produced in E. coli (Figs. (Figs.6A6A and S9A) and measured their ability to pull down IVA. The results presented in Fig. 6B showed the retention of purified IVA by a fusion to GST to full length VID. In order to more precisely map the interacting region, we used a series of truncated forms of VID (Costa et al., 2006) and observed that IVA was pulled down by a L201-A575 fragment that is lacking the entire N-domain. By contrast, it was not recovered by the T499, R202, A302, N399 and L201-N399 fragments, suggesting that neither the N-domain, nor the central region of VID were able to interact with IVA in the absence of the C-terminal region. From these experiments, we conclude that VID interacts directly with IVA and that the region delimited by residues 499 to 575 is required for the interaction.
To confirm this interaction using a different assay, we performed affinity blotting experiments (Einarson and Orlinick, 2002). We generated a series of VID constructs containing a His6 tag (Fig. 6C). The protein fragments (including His6-mCherry used as negative control) were produced in E. coli, purified on Ni2+-NTA agarose columns, resolved by SDS-PAGE and stained with Coomassie (Fig. S9B). The proteins were transferred to a nitrocellulose membrane and hybridized with purified His6-IVA. Next, the binding of IVA to the membrane was detected by immunoblotting with an anti-IVA antiserum. We observed binding of IVA to full-length VID, but not to the purified LysM domain or to the first 120 residues of VID (Fig. 6D). As a control, we showed that the anti-IVA antibody did not cross-react with VID or mCherry. Next, we tested the binding of IVA to the protein fragments VIDΔLysM, VID121-575 and VID121-521, which all contain region A. Although VIDΔLysM and VID121-521 did not contain the LysM domain, every fragment was able to recruit IVA (Fig. 6E). Therefore, our data are consistent with the interpretation that region A is necessary for the interaction with IVA, even in the absence of the the LysM domain.
The assembly of 70 proteins into a functional spore coat constitutes a paradigm for the formation of a supramolecular structure in a cell undergoing development (Driks, 1999; Henriques and Moran, 2007). Coat assembly is a dynamic and inherently asymmetric process, as exemplified by the patterns of subcellular localization previously reported for the coat proteins IVA, SafA, CotE, YheD and YutH (Ozin et al., 2001; Price and Losick, 1999; van Ooij et al., 2004; Webb et al., 1995). We began this work with a characterization of the effects of the deletion of VID on coat assembly. We have presented evidence that VID is a coat protein whose morphogenetic activity is required during a later stage of coat assembly. Specifically, we propose that VID is governing a morphogenetic transition that we named spore encasement. Our conclusion is supported by the analysis of the patterns of subcellular localization for 39 coat protein-GFP fusions in VID cells. We have shown that all of these fusions initially localized to the forespore surface independently of VID and 90% (35 out of 39 fusions) failed to complete formation of the ring pattern typical of normal coat protein localization. In a majority of cases (70%, 25 out of 35 fusions), the assembly process was blocked at the stage just after initial localization as a single cap of fluorescence on the mother cell proximal side of the forespore. Given that the spore coat phenotype of the VID mutant strain is markedly different from that of IVA cells, we propose that encasement is indeed a distinct second step of spore coat morphogenesis.
Our findings confirm and expand the conclusions of (Ozin et al., 2001) who observed by immunofluorescence microscopy that VID was required for SafA to fully encircle the forespore. In addition, our fluorescence microscopy data are reminiscent of previously published transmission electron microscopy images of VID cells (Beall et al., 1993). Although the principal conclusion of that original study was that the main role of VID was to anchor the spore coat to the forespore surface, we note that in some of these images, the spore coat material found in the center of the mother cell had a horseshoe shape, which would be consistent with incomplete spore encasement, followed by detachment of the spore coat from the spore surface. This phenomenon can also be observed in the VID mutant strains analyzed by Driks et al. (1994). We rarely observed detached caps of fluorescence in the mother cell of VID cells. However, we attribute this difference to the fact that fluorescence microscopy is a less disruptive method than electron microscopy and does not require fixing, staining and sectioning of sporulating cells. Taken together, these observations and our data suggest that, in coat assembly, VID facilitates the morphogenetic transition from single polar cap to a complete shell.
We also draw a parallel between B. subtilis spore coat assembly and B. anthracis exosporium formation. The exosporium constitutes the outermost spore structure in B. anthracis, B. cereus and several other spore forming bacteria, although B. subtilis spores are devoid of it (Driks, 2002b; Henriques and Moran, 2007). In the absence of the exosporium protein ExsY, assembly of the B. anthracis exosporium is blocked at the single cap stage (Boydston et al., 2006; Steichen et al., 2007). Thus, although different morphogenetic proteins regulate spore encasement during exosporium and spore coat formation, both processes include at least two successive steps (Giorno et al., 2009).
Our second major finding is that the N-domain of VID is essential for the morphogenetic process of spore encasement. We predicted that the first 150 residues of VID fold into a single protein domain with two likely folds. The first fold corresponds to the coat protein of the PP7 phage (Tars et al., 2000). Although hypothetical, this previously unsuspected structural homology between two proteins that belong respectively to spore and phage envelopes suggests that similar mechanisms may be operating during the assembly of spore and phage external protective structures. Specifically, the PP7 phage capsid protein undergoes multimerization during phage morphogenesis, and our preliminary results also indicate that both the full-length VID protein and the N-domain alone can form multimeric structures (data not shown). The second prediction is a lipocalin-like fold. The β-barrel structure characteristic of lipocalins is usually a binding site for small hydrophobic molecules, such as lipids (Campanacci et al., 2006). It remains to be tested whether VID can bind lipids in the manner of some lipocalins. However, since lipocalins can also bind other proteins (Skerra, 2008), the similarity to the lipocalin fold suggests that the N-domain may be involved in protein-protein interactions.
Finally, our analysis showed that the C-terminus of VID is essential for targeting VID itself to the surface of the spore. This region has two main components, the previously characterized LysM domain (a.a. 525-575) and the directly adjacent region A (a.a. 500-524). Both are involved in recruitment to the forespore surface, although the contribution of region A seems more important. A parallel can be drawn to the morphogenetic protein SafA, which also has a LysM domain (at its N-terminal end) followed by a conserved stretch of amino acids. In that case as well, both components are necessary for the subcellular localization of the protein to the surface of the forespore (Costa et al., 2006). Given that LysM domains have been shown to interact with peptidoglycan, it has been suggested that the LysM domain of VID may facilitate the initial targeting of the protein to the outer forespore membrane by recognizing peptidoglycan or peptidoglycan precursors, even before synthesis of the spore cortex (Ozin et al., 2000; Ozin et al., 2001; Costa et al., 2006; Henriques and Moran, 2007). However, unlike other proteins with LysM domains, VID does not contain a signal sequence for secretion, suggesting that an atypical mechanism would have to be used to translocate the LysM domain across the outer forespore membrane. An alternative possibility is that the LysM domain contributes to the initial targeting of VID by facilitating its interaction with IVA, presumably through region A, and that it only binds to the cortex at a later stage of development, when the forespore outer membrane becomes more permeable, or even during spore germination, when the cell wall hydrolases that were stored in the spore coat are released to carry out cortex degradation.
Our model of spore coat morphogenesis distinguishes between two successive steps: initial localization of coat proteins to the spore surface and spore encasement (Fig. 7). In this model, we propose specific roles for the three principal coat morphogenetic proteins: IVA, VM and VID. IVA is essential to the initial localization step of every coat protein. This conclusion is supported by the observation that in electron micrographs of IVA cells, the coat appears to be completely detached from the spore, as swirls of aggregated proteins floating in the middle of the cell (Piggot and Coote, 1976; Roels et al., 1992). Similarly, IVA is required for recruitment of VID through a direct interaction, presumably involving region A. Although purified VM-GFP is able to localize to positively curved membranes in vitro and when heterologously expressed in Escherichia coli and Saccharomyces cerevisiae, it also physically interacts with IVA. This interaction appears to be necessary to restrict VM to the outer spore membrane in vivo (Ramamurthi et al., 2006).
We propose that the next step, encasement, is driven by the combined actions of VID, and VM. IVA has been shown to form polymer cables upon the addition of ATP to purified IVA in vitro (Ramamurthi and Losick, 2008). Subsequently, VM and VID may be necessary to physically bend cables of IVA around the spore surface. The structural homology of the N-domain of VID to a phage coat protein, as well as our preliminary observations (data not shown), suggests that VID may also be able to multimerize. Therefore, one could envision a mechanism whereby VID, IVA and VM form a tripartite machine that polymerizes on top of the positively curved outer spore membrane and drives encasement of the spore by the spore coat. Once the components are properly localized, VID multimerizes to form a phage envelope-like structure around the surface of the spore while simultaneously binding to IVA and bending IVA cables around the outer spore surface. VM, via its interaction with IVA and binding to the outer spore membrane, localizes the continually polymerizing machine to the outer spore membrane as engulfment proceeds. The molecular details of how the machine assembles remain to be discovered.
Our model does not currently provide a detailed mechanism for the formation of the second focus of VID-GFP on the mother cell distal side of the spore. This second focus would presumably act as a nucleation site to promote the formation of the second polar cap. Further studies will be necessary to distinguish between alternative mechanisms. For instance, continuous expansion in a single plane of the first single focus could lead to the formation of the second focus on the opposite side of the spore. This mechanism would imply an intermediate structure connecting the two foci in the form of a thin fluorescent ring around the spore. Similar rings have been observed before for the sporulation protein YabP-GFP (van Ooij et al., 2004), but we did not detect such an intermediate in time course experiments with VID-GFP. However, it remains possible that this structure is short-lived and difficult to capture with traditional approaches. Alternatively, the formation of the second focus might be conditioned by specific positional cues on the distal site of the forespore. Such a cue could be provided by the sporulation protein SpoIIIE (or a protein that interacts with it). It has indeed been shown that the single fluorescent focus formed by a SpoIIIE-GFP fusion migrates from the mother cell-proximal to the distal side of the forespore during engulfment (Sharp and Pogliano, 1999).
Considering that every coat protein requires IVA for localization to the spore, an explanation of the encasement of IVA itself may explain the mechanism of encasement for the remainder of the coat. However, we favor a model with contributions from additional morphogenetic proteins, such as SafA and CotE, to relay the effects of VID and IVA to downstream components of the coat protein network. This hierarchical organization of the assembly process is supported in part by the observation that the N-domain of VID interacts directly with SafA (Costa et al., 2006). The existence of the second and third classes of VID-dependent coat proteins, those that form an incomplete punctuate ring and those that are unaffected by a VID deletion respectively, also suggest the contribution of other coat proteins to promote optimal spore encasement. However, each class of coat proteins is fully dependent on VM for encasement implying a branched hierarcy of encasement regulation: class 1 includes 70% of the coat proteins (i.e. those that are strictly dependent on VM and VID for encasement), class 2 includes 20% of the coat proteins (strictly dependent on VM and partially dependent on VID) and class 3 includes 10% of the coat proteins (strictly dependent on VM and potentially dependent on yet unidentified coat proteins).
In conclusion, we have shown that VID and VM play critical roles in spore coat assembly by mediating the process of spore encasement and that encasement is a process of spore coat morphogenesis distinct from initial localization under the control of IVA. Both the N-terminal and C-terminal conserved regions of VID are required for full morphogenetic activity. However, the role of the C-terminal region may be limited to the targeting and anchoring of the protein to the surface of the spore via either a direct interaction with IVA or by facilitating this interaction, whereas the N-domain is likely to constitute the active morphogenetic component necessary for encasement. These three proteins appear to be the minimal requirements to form a shell of IVA protein around the spore and thus, to recruit the rest of the coat. The predicted structural homology to a phage coat protein essential for capsid formation raises the intriguing possibility that parts of the molecular mechanism of formation of protective envelopes are conserved from bacteriophages to bacterial spores.
All strains used here were derivatives of the wild-type strain PY79 (Youngman et al., 1984) and are listed in Table S1. Strains construction (including VID mutagenesis, production of GST fusions to VID, production and purification of full-length and truncated forms of His6-VID and His6-IVA) is detailed in the Supporting Information. Every plasmid and oligonuleotide primer used is listed in Tables S5 and S6, respectively.
Cells were incubated at 37°C in hydrolyzed casein medium to A600nm of 0.6, pelleted by centrifugation at 4000 g for 5 min and resuspended in SM medium (Sterlini and Mandelstam, 1969). After resuspension, cells were returned to 37°C and imaged at the indicated times by fluorescence microscopy.
Fluorescence microscopy was performed as described before (Kim et al., 2006). Briefly, 1ml aliquots of the sporulating cultures were transferred into microcentrifuge tubes and spun down at 8000RPM for 2 minutes in a microcentrifuge. Pellets were resuspended in 100μL PBS supplemented with the membrane dye FM4-64 (Invitrogen) at 1.5μg ml-1 final concentration. Two μl of the concentrated sample were placed on a microscope slide and covered by a poly-L-lysine-treated coverslip for analysis. Control experiments were performed (Figures S10 and S11, Table S4) to confirm that neither the centrifugation conditions nor the poly-L-lysine treatment influenced the localization patterns of the GFP fusions. Images were taken using a Nikon 90i motorized fluorescent microscope equipped with a Roper 1K monochrome digital camera and driven by the NIS-Elements AR 3.0 software. Images were processed with Adobe Photoshop 8.0 for minor adjustments of brightness and contrast.
Pull down assays using GST fused to the full-length VID protein or to various deletion forms of the protein were conducted essentially as described before (Costa et al., 2006), except that purified IVA (10 μg) was used as the prey protein. IVA was detected by immunoblot analysis using an anti-His6 antibody (Novagen) at a 1:10000 dilution.
The protocol described by Einarson and Orlinick (2002) was followed with minor modifications. The partially purified full-length and truncated forms of His6-VID and His6-mCherry (5 μM final concentration) were resolved by SDS-PAGE (10% gels for the VID1-120 fragment and VID-LysM domain, and 15% gels for all the other forms) and transferred to a nitrocellulose membrane. The membranes were incubated for 10 min in basic buffer (20mM HEPES pH 7.9; 50mM KCl; 10mM MgCl2; 1mM DTT; 0,1 Nonidet P-40) and then for 1 hour with blocking buffer (5% non-fat dry milk in basic buffer). Next, the membranes were incubated overnight with gentle agitation at 4°C in interaction buffer (1% nonfat dry milk; 5% glycerol in basic buffer) containing 5 nM of purified His6-IVA. The membrane was then washed three times with PBS 0.2% Triton X-100 for 10 minutes and twice with PBS containing 0.2% Triton X-100 and 100 mM KCl. Finally, bound IVA was detected by immunoblotting (Costa et al., 2006) with an anti-IVA antibody at a 1:500 dilution.
We are grateful to Adam Driks for discussion, comments on the manuscript, sharing of unpublished data and strains, Richard Losick for comments on the manuscript and strains, Jonathan Dworkin, Alan Grossman and Patrick Stragier for strains. We thank Teresa Costa and Ana M. Almeida for participating at an earlier stage of the project, Uschi Auguste, Alex LoPinto, Aparna Modi, Lara Winterkorn and Kevin Wu for technical assistance. We acknowledge the financial support of NIH grant GM081571 to PE and training grant in Developmental Genetics 5T32HD007520 to PTM, Department of the Army award number W81XWH-04-1-0307 to PE and RB, NSF DBI-0820757 to RB, grant POCTI/BIA-BCM/60855/2004 and ERA-Net Pathogenomics CDIFFGEN from Fundação para a Ciência e a Tecnologia (F.C.T) to AOH and a Collegiate Research Scholar fellowship from New York University to SB. The content of this material does not necessarily reflect the position or the policy of the Government and no official endorsement should be inferred.