Dynamics of spore coat morphogenesis in Bacillus subtilis

Authors

  • Peter T. McKenney,

    1. New York University, Center for Genomics and Systems Biology, Department of Biology, 12 Waverly Place, 8th floor, New York, NY 10003, USA.
    Search for more papers by this author
  • Patrick Eichenberger

    Corresponding author
    1. New York University, Center for Genomics and Systems Biology, Department of Biology, 12 Waverly Place, 8th floor, New York, NY 10003, USA.
    Search for more papers by this author

E-mail pe19@nyu.edu; Tel. (+1) 212 998 8247; Fax (+1) 212 995 4266.

Summary

Spores of Bacillus subtilis are encased in a protective coat made up of at least 70 proteins. The structure of the spore coat has been examined using a variety of genetic, imaging and biochemical techniques; however, the majority of these studies have focused on mature spores. In this study we use a library of 41 spore coat proteins fused to the green fluorescent protein to examine spore coat morphogenesis over the time-course of sporulation. We found considerable diversity in the localization dynamics of coat proteins and were able to establish six classes based on localization kinetics. Localization dynamics correlate well with the known transcriptional regulators of coat gene expression. Previously, we described the existence of multiple layers in the mature spore coat. Here, we find that the spore coat initially assembles a scaffold that is organized into multiple layers on one pole of the spore. The coat then encases the spore in multiple co-ordinated waves. Encasement is driven, at least partially, by transcription of coat genes and deletion of sporulation transcription factors arrests encasement. We also identify the trans-compartment SpoIIIAH–SpoIIQ channel as necessary for encasement. This is the first demonstration of a forespore contribution to spore coat morphogenesis.

Introduction

Upon entering sporulation, Bacillus subtilis cells divide asymmetrically. The smaller compartment ultimately becomes a dehydrated spore that encases the genome within two protective layers, the cortex, made up of peptidoglycan and the spore coat made up of at least 70 individual proteins (Henriques and Moran, 2007). The larger of the two compartments, the mother cell, synthesizes the proteins of the spore coat, which are deposited on the surface of the developing forespore (Driks et al., 1994). B. subtilis spore coats are necessary for resistance to digestion by the nematode Caenorhabditis elegans and the protozoan Tetrahymena thermophila (Klobutcher et al., 2006; Laaberki and Dworkin, 2008). The presence of enzymes in the spore coat, such as the alanine racemase YncD, suggests that the spore coat may also have a function in buffering the environment within the developing sporangium and around the released spore (Steichen et al., 2003; Todd et al., 2003; Chesnokova et al., 2009; McKenney et al., 2010).

While there may be over 70 spore coat proteins in total, only a small number of them, the morphogenetic proteins, have been shown to be necessary for assembly of the coat. The spore coat is organized into four distinct layers: a basement layer closely apposed to the membrane, the inner coat, outer coat and crust (McKenney et al., 2010). Assembly of each coat layer is dependent on one major morphogenetic protein (Roels et al., 1992; Driks et al., 1994; Takamatsu et al., 1999; McKenney et al., 2010). The roles of these proteins were inferred from the phenotypes of deletion strains observed by electron microscopy. In cells deleted of spoIVA the coat does not localize to the spore surface, but self-assembles into aggregates in the mother cell cytoplasm (Piggot and Coote, 1976; Roels et al., 1992). These data suggest that SpoIVA anchors the coat to the spore surface. Deletion of safA leads to the mis-assembly of the inner coat (Takamatsu et al., 1999; Ozin et al., 2000), while deletions of cotE and cotZ cause mis-assembly of the outer coat and crust respectively (Driks et al., 1994; McKenney et al., 2010; Imamura et al., 2011).

Fusion of coat proteins to the green fluorescent protein (GFP) has allowed the easy observation of the localization of spore coat proteins within live developing sporangia (Webb et al., 1995). Observation of the localization of individual spore coat protein fusions in cells deleted of the coat morphogenetic proteins established a spatially scaled genetic interaction map that was able to accurately predict the localization of individual proteins within the spore coat (Kim et al., 2006; McKenney et al., 2010). This map, however, is static and only represents fully mature spore coats. In this work we examine the morphogenesis of the spore coat in time-course and attempt to determine how spore coat morphogenesis is organized in time.

Sporulation is driven by a complex programme of transcriptional regulation that includes five sporulation-specific sigma factors and five other sporulation-specific transcription factors (de Hoon et al., 2010). In the mother cell, gene activation is driven by the sequential activation of σE and σK, resulting in early and late waves of gene expression (Kroos et al., 1989; Cutting et al., 1990). The transcription factor SpoIIID regulates subsets of the σE regulon as a repressor and an activator, creating distinct pulses of gene expression (Halberg and Kroos, 1994; Eichenberger et al., 2004). The σK regulon is regulated by GerE in a similar manner (Zheng et al., 1992; Eichenberger et al., 2004).

Coat genes are expressed throughout sporulation under the control of the early mother cell-specific sigma factor σE and the late mother cell-specific sigma factor σK (Eichenberger et al., 2003; 2004). The localization of spore coat proteins begins shortly after asymmetric division (Driks et al., 1994; Pogliano et al., 1995; Webb et al., 1995), but new coat proteins are added to the spore coat throughout sporulation (Costa et al., 2004; van Ooij et al., 2004; Masayama et al., 2007). In order to determine the dynamics of spore coat morphogenesis, we used a library of 41 spore coat proteins fused to GFP and observed protein localization relative to a series of morphological markers in time-course. The combination of a fluorescent membrane stain and two distinct events observable by phase-contrast microscopy have allowed us to establish six kinetic classes of spore coat protein localization. These kinetic classes correlate well with previous data describing the regulation of spore coat gene promoters during sporulation (Eichenberger et al., 2003; 2004).

We used high-resolution image analysis to measure the position of early-localizing spore coat proteins and found that major components of all four layers of the spore coat are present at the earliest stages of spore coat morphogenesis. Our data suggest that the spore coat initially forms an organized scaffold on the mother cell proximal (MCP) pole. In a previous study we described the phenomenon of encasement, the transition of individual spore coat protein fusions from a single cap of fluorescence on the MCP pole of the spore to a full shell of protein around the circumference of the spore (Wang et al., 2009). Here, analysis of our time-course localization data revealed that encasement occurs by the co-ordinated movement of groups of fusions in a series of successive waves. Encasement appears to proceed in order from basement layer and inner coat to outer coat to crust. Thus, the timing of encasement may be the major organizing phenomenon of spore coat morphogenesis. We also identify two novel factors necessary for encasement, SpoIIIAH and SpoIIQ, the protein components of the transmembrane channel linking the mother cell and the forespore. This is the first demonstration of a contribution to spore coat morphogenesis from within the forespore.

Results

Time-course analyses of spore coat assembly

We began this work by gathering time-course microscopy data during sporulation for our library of 41 spore coat protein–GFP fusions (Kim et al., 2006; Wang et al., 2009; McKenney et al., 2010). Because sporulation is regulated at the population level (Veening et al., 2005; 2008; Eswaramoorthy et al., 2010), synchronization is not as efficient as cell cycle synchronization in Caulobacter crescentus (Evinger and Agabian, 1977). Nonetheless, each of the morphological states predominates in the cell population at time points separated by approximately 30 min under our standard conditions. Sporulation was induced by resuspension in Sterlini-Mandelstam medium (Sterlini and Mandelstam, 1969). Two hours after resuspension, a septum is built near one pole of the developing sporangium (Ryter, 1965). We systematically mapped the localization of each spore coat protein from hour 2 to 4 after resuspension to a series of morphological markers provided by the fluorescent membrane stain FM4-64 (Pogliano et al., 1999), allowing us to distinguish between early engulfment (stage 1 in Fig. 1), late engulfment (stage 2) and post engulfment (stage 3) cells. With a phase-contrast objective we can also distinguish between phase dark (stage 4) and phase bright (stage 5) forespores during late sporulation. Phase dark spores began to predominate at hour 5 while phase bright spores appear at hour 6. The result is a time-series of morphological markers from the beginning to the end of sporulation that allowed us to map the population average for the localization of each protein throughout morphogenesis (Figs 1A and 2A). We collected time-course fluorescence microscopy images for all 41 spore coat protein–GFP fusions and present these data in Figs S2–S42, and we present data for six representative fusions in Figs 1, 2 and S1.

Figure 1.

Three kinetic classes of early-localizing spore coat proteins. A. Left column: A diagram of the morphological stages of sporulation revealed by the membrane stain FM4-64 (red) and phase-contrast optics (grey). 1 = early engulfment, 2 = late engulfment, 3 = post engulfment, 4 = phase dark, 5 = phase bright. In the remaining columns, typical GFP fusion localization patterns are drawn in green for one representative of each kinetic class over the time-course of sporulation. Class I = SpoVID-GFP, Class II = CotE-GFP, Class III = CotZ-GFP. B–F. Strains containing fusions of GFP to a spore coat protein in an otherwise wild-type strain, (B) PE654 (spoVID-gfp), (C and D) PE632 (cotE-gfp), (E and F) PE680 (cotZ-gfp) were incubated at 37°C to the indicated time point after resuspension. Samples were stained with FM4-64. A merged image of FM4-64 membrane stain (red) and a coat protein–GFP fusion (green) is shown in the left column, the coat protein–GFP fusion alone is shown in the centre column, FM4-64 alone (B,C,E) or a brightfield image taken with phase-contrast (D,F) is shown in the right column. Number labels of carets correspond to the stages of sporulation on the left of the diagram above (A). Quantification of the staging of individual cells at the relevant time points is shown in Table S1. Complete time-course microscopy for SpoVID-GFP, CotE-GFP and CotZ-GFP is shown in Fig. S1. Complete time-course microscopy for 19 class I, II and III fusions can be found in Figs S2–S20.

Figure 2.

Three kinetic classes of late-localizing spore coat proteins. A. Left column: A diagram of the morphological stages of sporulation revealed by the membrane stain FM4-64 (red) and phase-contrast optics (grey). 1 = early engulfment, 2 = late engulfment, 3 = post engulfment, 4 = phase dark, 5 = phase bright. In the remaining columns, typical GFP fusion localization patterns are drawn in green for one representative of each kinetic class over the time-course of sporulation. Class IV = YsnD-GFP, Class V = CotA-GFP, Class VI = CotD-GFP. B–D. Strains containing fusions of GFP to a spore coat protein in an otherwise wild-type strain, (B) PE490 (ysnD-gfp), (C) PE676 (cotA-gfp), (D) PE659 (cotD-gfp) were incubated at 37°C to the indicated time point after resuspension salts. Samples were stained with FM4-64. A merged image of FM4-64 membrane stain (red) and a coat protein–GFP fusion (green) is shown in the left column, the coat protein–GFP fusion alone is shown in the centre column, a brightfield image taken with phase-contrast is shown in the right column. Number labels of carets correspond to the stages of sporulation on the left of the diagram above (A). Quantification of the staging of individual cells at the relevant time points are shown in Table S1. Complete time-course microscopy for YsnD-GFP, CotA-GFP and CotD-GFP is shown in Fig. S1. Complete time-course microscopy for 22 class IV, V and VI fusions can be found in Figs S21–S42.

We failed to observe the localization of any spore coat protein fusion on the flat septum immediately after asymmetric division. As the septum began to exhibit curvature (stage 1 cells in Fig. 1), we observed the localization of 19 spore coat protein fusions adjacent to the septum (Figs S43–S45). Additional experiments will be necessary to determine if sequential assembly exists at the temporal resolution of seconds; however, we feel justified in reporting that these 19 fusions localize simultaneously to the forespore surface at the level of resolution of our time-course experiments.

While the initial localization of these coat protein fusions appeared nearly simultaneous, the rate at which the individual coat proteins encased the spore differed dramatically. Among the early-localizing fusions we defined three classes of fusion encasement kinetics (Fig. 1). A first class of fusions (defined here as class I) appeared to follow the mother cell membrane closely as it engulfs the forespore and a nearly complete ring of GFP fluorescence was often observed while a nearly full ring of FM4-64 fluorescence was also present. This is exemplified by SpoVID-GFP in Fig. 1B (stage 2 cell, see also: Figs S2–S7, S43).

We distinguished a second kinetic class of fusion proteins (class II) as those that localize simultaneously with class I, but begin to encase the spore only after engulfment is complete. FM4-64 is impermeable to membranes. Until membrane fission occurs on the mother cell distal (hereafter MCD) pole of the forespore, the forespore membranes are continuous with the plasma membrane of the sporangium and are accessible to the surrounding medium. After membrane fission, the forespore is contained as a free-floating protoplast within the sporangium, disconnected from the plasma membrane. Therefore, the membranes surrounding the forespore are no longer stained. Sporangia stained at this stage (stage 3 cells in Fig. 1) appear as empty rods (Pogliano et al., 1999). In the case of all class II fusions, we did not observe encasement of the forespore before the disappearance of FM4-64 staining of the forespore membranes. Class II fusions (illustrated by CotE-GFP, Fig. 1C and D, Figs S8–S15) appeared to begin encasing the forespore from initial foci near the site on the MCD pole where fission of the engulfing membranes occurred (stage 3 cell, Figs 1C and S44). These foci then expanded from the cell midline to form a cap of fluorescence covering the MCD hemisphere (stage 4 cell, Fig. 1D).

The third kinetic class of spore coat proteins (class III, represented by CotZ-GFP, Fig. 1E and F, Figs S16–S21) localized simultaneously with the class I and class II fusions (Fig. S45); however, they did not begin to encase the spore until phase dark spores had appeared around hour 4.5 of sporulation (stage 5 cell, Figs 1F and S46). Taken together, these data suggest that spore encasement by early-localizing spore coat proteins takes place in a cascade of three temporally distinct waves.

Similarly, we can distinguish between three kinetic classes among fusions that localize only after the completion of engulfment. The largest class, Class V (illustrated by CotA-GFP), includes 17 spore coat protein fusions that localize exclusively on phase dark spores (Figs 2C and S24–S40). Consistent with the accessibility of both poles of the spore to the mother cell cytoplasm after the completion of engulfment, these fusions tend to localize simultaneously to both poles of the spore (stage 4 cell, Fig. 2C). We observed localization dynamics that were intermediate between class III and class V for three fusions (class IV, illustrated by YsnD-GFP, Figs S21–S23). These fusions did not exhibit convincing localization before the completion of engulfment (stage 2 cell, Fig. 2B) and yet were often present in sporangia that did not contain phase dark spores (stage 3 cell, Fig. 2B). Finally, the two fusions of Class VI (illustrated by CotD-GFP) have the most delayed initial localization of all the 41 spore coat protein fusions and localize exclusively to phase bright spores (stage 5 cell, Figs 2D, S41 and S42).

Evidence for a spore coat scaffold

Considering our localization time-course data, we favour a model of coat morphogenesis in which a spatially organized scaffold of spore coat proteins assembles on the MCP pole, as originally suggested by Driks et al. (1994). When we combine our spatially weighted genetic interaction map (McKenney et al., 2010) with the localization kinetics data discussed above, we find proteins of all four coat layers within the group of 19 early-localizing fusions (Table 1). Such a model requires that all four layers are present and spatially organized upon localization to the MCP pole. Our previous study only measured coat protein localization in mature phase bright spores. To directly test our model, we measured, at sub-pixel resolution, the localization of a morphogenetic protein of each spore coat layer at hour 3 in sporangia that were near the completion of engulfment (with the exception of the inner coat where we used YaaH-GFP as a proxy for SafA-GFP, which is not functional). When comparing the measurements from hour 3 and hour 6 we found that at hour 3 each of the four fusions are present with the same ordering from basement layer to crust: GFP-SpoIVA, YaaH-GFP, CotE-GFP and CotZ-GFP. Also, they are each separated with a statistical significance that is similar to our previous data from phase bright spores [P < 10−3 (Wilcoxon test) for each adjacent distribution, Fig. 3]. Our data suggest that spore coat layers can be distinguished very early in spore coat morphogenesis (Fig. 3). Thus, the spore coat initially assembles as a spatially organized scaffold containing the principal organizers of all four coat layers.

Table 1.  Kinetics, genetics and transcription. Thumbnail image of
Figure 3.

A spore coat scaffold. Strains PE1560 (gfp-spoIVA), PE655 (yaaH-gfp), PE632 (cotE-gfp) and PE680 (cotZ-gfp) were incubated to hours 3 and 6 after resuspension in sporulation salts containing MitotrackerRed (Invitrogen) at 37°C, Cells were imaged by collecting Z-series of five planes separated by 0.3 µm imaged using phase contrast and TexasRed (MitotrackerRed) and FITC (GFP) filters. Image stacks were deconvolved using Autoquant (MediaCybernetics). Images were analysed using PSICIC, a collection of MATLAB (The Mathworks) functions, as described in the Experimental procedures. Data were plotted using the boxplot function in R (http://www.r-project.org). Each box of the boxplot is bounded by the upper and lower quartile with the median in bold. Outliers are data points outside of ± 1.5 of the inter-quartile range indicated by circles. n = 100 for each sample. Colour code: Cyan = basement layer, Gold = inner coat, Blue = outer coat, Red = crust.

Waves of transcription – waves of encasement

We propose that the waves of spore coat protein encasement are driven, at least in part, by waves of transcriptional regulation of coat protein gene promoters. All 19 genes encoding fusions of the first three kinetic classes were shown to be regulated by the early mother cell-specific sporulation sigma factor σE by transcriptional profiling, and complementary analyses such as primer extension and binding motif searches [(Eichenberger et al., 2003), Table 1]. All six transcriptional regulatory regions of class I fusions are under the positive control of σE, while three of the six have been described as negatively regulated by SpoIIID (Eichenberger et al., 2004). These data are consistent with the genes being expressed in a short strong pulse that correlates well with the early encasement dynamics of these fusions.

The second pulse of gene expression in the mother cell is driven by positive regulation of promoters by both σE and SpoIIID, as SpoIIID accumulates to levels sufficient to promote gene expression. All positive regulation by SpoIIID occurs in promoters of classes II–V. While only two of the seven promoters of class II were categorized by transcriptional profiling as positively regulated by SpoIIID, we believe more may have been missed due to the experimental design of the microarray experiments, where spoIIID dependency was analysed at a single time point during sporulation (Eichenberger et al., 2004, see data for PcotO-lacZ below). The next pulse of mother cell gene expression is driven primarily by σK. For class III fusions, four of six promoters were characterized as both σE- and σK-dependent, suggesting that this late transition to encasement may be controlled by σK-dependent expression of class III promoters.

To examine the dynamics of promoter activity we fused one representative promoter of each of the first three kinetic classes to lacZ and examined the level of β-galactosidase activity in deletions of mother cell transcription factors (Fig. S47). The class I transcriptional fusion of lacZ to spoIVA (which contains 2 promoters, P1 and P2) was unaffected by deletion of spoIIID, while the activity of the lacZ fusion to the class II promoter PcotO was reduced in the absence of SpoIIID. For a class III promoter, full activity of the lacZ fusion to PcotY, the promoter of cotZ (Zhang et al., 1994), required SpoIIID, σK and GerE. Thus, it is possible that waves of transcription may underlie our observed waves of encasement.

Mother cell transcription factors regulate encasement

To test the importance of transcriptional regulation to encasement, we observed the localization of representatives of each kinetic class of coat protein fusion in strains containing deletions of the genes encoding the mother cell-specific transcription factors. Fusions from class I were unaffected by deletion of spoIIID (data not shown). We observed an arrested morphogenesis phenotype for CotE-GFP (Fig. 4A and B) and six of the seven class II fusions (Table 1, Figs S48–S54). The six fusions formed a small second cap of fluorescence on the MCD pole, but appeared arrested at this state.

Figure 4.

Effect of SpoIIID and σK on class II fusions. Strains PE632 (cotE-gfp), PE1698 (cotE-gfp, spoIIID::tet) and PE836 (cotE-gfp spoIVCB::erm) were incubated to hours 3, 4 and 5 after resuspension in sporulation salts at 37°C. Cells were stained with FM4-64 and imaged using fluorescence microscopy. A–C. representative fields are shown for each strain at hour 4. Merged membrane stain (red) and CotE-GFP (green) are shown in the left column. CotE-GFP alone is shown in the right column. D. Table of data from the quantification of CotE-GFP in the indicated mutant backgrounds. Images were analysed using ImageJ (http://rsbweb.nih.gov/ij/) as described in the Experimental procedures. The first four columns of data represent the percentage of the cell population with CotE-GFP morphogenesis at the indicated stage. Single Cap = MCP pole cap only, 2 Caps = caps on both the MCP and MCD poles. Unconnected = 2 distinct ROIs (regions of interest) after thresholding (threshold = 4x background) on both poles of the spore. Connected = a single ROI that protrudes towards the MCD pole. The normal progression of CotE-GFP localization is Single Cap to Unconnected to Connected. Unconnected MCD pole caps were measured and the area in pixels is recorded in the MCD cap area column.

We quantified the spoIIID phenotype for CotE-GFP by measuring the area of the MCD cap of fluorescence in cells that have completed engulfment (Fig. 4D) and made two observations. First, at the population level we observed a delay in the progression of cells through the three phases of CotE-GFP assembly: single MCP cap, two unconnected polar caps and two connected caps. The proportion of cells with two measureable caps in spoIIID deletion cells is reduced beginning at hour 3 of sporulation (wild type: 55%, spoIIID: 22%). These spoIIID cells rarely transition from two unconnected polar caps to connected caps. At hour 5, 42% of wild-type cells have connected CotE-GFP caps compared with 0% of spoIIID deletion cells (Fig. 4D). Second, we observe a significant reduction in size of the MCD cap beginning shortly after the completion of engulfment at hour 3 (wild type: 26 ± 20 pixels, spoIIID: 15 ± 10 pixels) that persists in spoIIID deletions until hour 5 (wild type: 35 ± 23 pixels, spoIIID: 19 ± 14 pixels) when spoIIID deletion cells begin to lyse. Cells deleted of sigK (spoIVCB) are unaffected (Fig. 4C), arguing against the possibility that the spoIIID effect on encasement of class II proteins is an indirect effect of the lack of σK activity. We also measured CotO-GFP and CotM-GFP in the same manner with very similar results (data not shown).

While six of seven class II fusions require SpoIIID for 2nd cap morphogenesis, all six class III fusions have strongly altered morphogenesis in the absence of SpoIIID. CotZ-GFP (Fig. 5) and five other class III fusions (Figs S55–S60) all appear to be arrested at the single-cap stage of morphogenesis in the absence of SpoIIID. We also tested the effects of deletion of sigK and gerE on the morphogenesis of the class III fusions. CotZ-GFP shows dramatic requirements for SpoIIID, σK and GerE (Fig. 5A–E). CotZ-GFP is arrested as a single dot in both the spoIIID and sigK deletion strains (Fig. 5A–D). In cells deleted of gerE the fusion expands out to a structure resembling a cap on the MCP pole, but it arrests there and does not transition to encasement (Fig. 5E). The effects of sigK and gerE deletion on the other five class III fusions are variable. In general, the fusions initiate encasement forming a small cap on the MCD pole, but they fail to form complete rings of protein. Taken together, the defects in encasement by coat protein fusions in transcription factor mutants coupled with the altered activity of coat protein promoters suggests that the waves of encasement we observed are driven, at least in part, by the level of protein expression.

Figure 5.

Induced expression of CotZ-GFP rescues encasement. A. A diagram representing localization of CotZ-GFP. Left column: CotZ-GFP expressed from the endogenous promoter in the indicated mutant backgrounds at hour 6 after resuspension. Right column: CotZ-GFP expressed from the IPTG-inducible promoter Phs at hour 4. Expression was induced at hour 3. B–E. Strains were incubated to the indicated time points at 37°C in sporulation salts, stained with FM4-64 and imaged using fluorescence microscopy. B: PE680 (cotZ-gfp). C: PE1698 (cotZ-gfp, spoIIID::tet). D: PE2022 (cotZ-gfp, spoIVCB::erm). E: PE1242 (cotZ-gfp, gerE::cat). F–I. Strains were incubated to hour 3 after resuspension in sporulation salts at 37°C. At hour 3, IPTG was added to a final concentration of 1 µM to induce gene expression. Samples were stained with FM4-64 and imaged using fluorescence microscopy 1 h after induction. F: PE2511 (amyE::Phs-cotZ-gfp, cotXYZ::neo). G: PE2493 (amyE::Phs-cotZ-gfp, cotXYZ::neo, spoIIID::tet). H: PE2065 (amyE::Phs-cotZ-gfp, cotXYZ::neo, spoIVCB::erm). I: PE2513 (amyE::Phs-cotZ-gfp, cotXYZ::neo, gerE::cat). J. Cell population stages and MCD pole cap areas were quantified using ImageJ as described in Fig. 4 and in Experimental procedures.

Coat protein expression is sufficient to partially rescue encasement

The experiments above suggested that proper transcriptional regulation is necessary for the completion of encasement. It is possible that the effects we are observing are manifestations of direct regulation of the promoters of individual fusions and that increased expression levels of these fusions may be sufficient to rescue encasement. To test this hypothesis, we cloned cotZ-gfp, which encodes the morphogenetic protein of the spore crust, in front of the IPTG-inducible promoter Phyperspank (Phs) and induced expression at hour 3, after the majority of cells had finished engulfment, but before the beginning of encasement of CotZ-GFP in wild-type cells. We performed these experiments in cells deleted of cotXcotYcotZ that were otherwise wild type or also deleted of spoIIID, sigK or gerE (Fig. 5F–J).

Within 30 min of the addition of inducer we began to observe encasement in a large fraction of cells in the wild-type, ΔsigK and ΔgerE strains. By 1 h after the addition of inducer (hour 4 of sporulation) greater than 70% of the post engulfment sporangia in the wild-type (Fig. 5F), ΔsigK (Fig. 5H) and ΔgerE (Fig. 5I) strains contained measureable 2nd caps; however, only 36% of sporangia in the ΔspoIIID strain contained a measureable 2nd cap (Fig. 5G). The 2nd caps formed in the ΔspoIIID strain were also 1/3 the size of the 2nd caps of the other strains (Fig. 5J), suggesting that induction of expression alone is sufficient to begin encasement in this background, but it is not sufficient to proceed to completion. The size and frequency of these 2nd caps of CotZ-GFP (14 ± 14 pixels, 35% of cells, Fig. 5J) are remarkably similar to the data obtained for CotE-GFP at hour 4 of sporulation in ΔspoIIID mutant cells (19 ± 13 pixels, 40% of cells, Fig. 4D). Because both strains are arrested at a similar stage of morphogenesis, and considering that CotZ-GFP is dependent on CotE for localization (Kim et al., 2006), this suggests that induction of expression may only restore encasement by CotZ-GFP up to the stage where encasement by CotE is blocked in the spoIIID background. Thus, encasement by CotZ-GFP likely requires a sub-stratum of CotE.

These data suggest four distinct genetically uncoupled steps in CotZ-GFP morphogenesis. First, at endogenous expression levels, σK is necessary to extend the initial dot of CotZ-GFP localization out to a crescent on the MCP pole of the forespore. Second, GerE is required for the subsequent transitions from a single cap to a second cap on the MCD pole and, eventually, a full ring. Both of these effects are presumably transcriptional on the cotZ promoter (PcotY) (Zhang et al., 1993). Any defects in the CotZ-GFP overexpression experiment should represent effects in ‘trans’ that are independent of PcotY and the abundance of CotZ-GFP. Third, as shown by the induced expression experiment, SpoIIID is necessary for expansion of the 2nd cap on the MCD pole, from a dot to a crescent. The simplest explanation for this effect would be that it is a consequence of the positive regulation of cotE by SpoIIID. In the absence of SpoIIID, encasement by CotE is incomplete (Fig. 4A), and as a result, encasement by CotE-dependent fusions, such as CotZ-GFP, may also be impaired. Further experiments would be necessary to determine whether the effect of SpoIIID on cotE expression is direct or if an unknown factor under the control of SpoIIID is involved. Fourth, an unknown factor under the control of σK is necessary for the final transition from two caps to a full ring. This factor is likely to be specific to CotZ-GFP as CotE-GFP is able to form a full ring in cells deleted of sigK.

A forespore requirement for encasement

Previous work from our laboratory characterized the roles of SpoVM and SpoVID in encasement (Wang et al., 2009). We initially examined spoVM as part of a microscopy-based candidate gene screen for deletions that cause defective encasement of spore coat protein fusions. We examined deletions involved in processes co-ordinated at the sporulation septum such as engulfment, cortex synthesis and inter-compartment signalling. We identified a very strong block of encasement in cells lacking SpoIIQ, a forespore protein that contributes to engulfment in certain contexts (Broder and Pogliano, 2006). SpoIIQ also interacts with the mother cell protein SpoIIIAH via a transmembrane domain in the inter-membrane space and forms a channel between the two compartments that resembles a type-III secretion apparatus (Blaylock et al., 2004; Doan et al., 2005; Camp and Losick, 2008; Meisner et al., 2008). The effect of spoIIQ on encasement was observed for every kinetic class of fusions, except class I. In particular we note that localization of SpoVM-GFP and SpoVID-GFP are not altered in spoIIQ cells (data not shown).

We performed a quantitative examination of the localization of CotE-GFP in cells deleted of spoVM, spoVID, spoIIQ or spoIIIAH (Fig. 6A and C–F). In cells lacking SpoVID, we rarely observed a 2nd cap on the MCD pole of the forespore (Fig. 6A, B and G, ΔspoVID: 4% of cells, wild type: 93% of cells), while in the population lacking SpoVM (Fig. 6C) we observe a very small cap of CotE-GFP fluorescence in 31% of cells. The 2nd cap in cells lacking SpoVM is small compared both with wild type and with cells lacking SpoIIID (Fig. 6D and H). In cells deleted of spoIIQ, we found a very strong block of encasement (Fig. 6E and G, ΔspoIIQ: 3% of cells). This is the first description of an effect of a forespore protein on spore coat morphogenesis. The phenotype in cells deleted of spoIIIAH was less strong with 25% of cells exhibiting a 2nd cap of CotE-GFP fluorescence (Fig. 6F). In this case the 2nd cap signal was not spatially separated from the MCP pole cap of florescence, perhaps reflecting structural instability of forespores in mutants of the spoIIIA operon (Doan et al., 2009).

Figure 6.

A forespore requirement for encasement. Strains containing cotE-gfp expressed from the endogenous promoter in the indicated genetic backgrounds were incubated to hour 5 after resuspension in sporulation salts at 37°C. Samples were stained with FM4-64 and were imaged using fluorescence microscopy. (A) PE632 (cotE-gfp). (B) PE1224 (cotE-gfp, spoVID::kan). (C) PE1471 (cotE-gfp, spoVM::kan). (D) PE1698 (cotE-gfp, spoIIID::tet). (E) PE1037 (cotE-gfp, spoIIQ::spc::cat). (F) PE2534 (cotE-gfp, spoIIIAH::tet). (G and H) Cell staging and MCD pole cap areas were quantified using ImageJ as described in Fig. 4 and in Experimental procedures.

SpoIIQ does not affect the activity of the canonical σE-dependent promoter PspoIID; however, it is required for activation of σG in the forespore (Londoño-Vallejo et al., 1997; Sun et al., 2000). SpoIIQ is also necessary for the proper localization of proteins on both sides of the septum (Campo et al., 2008). Thus, it is possible that alteration of σG-dependent gene expression or SpoIIQ-dependent protein localization in the forespore accounts for the encasement phenotype of CotE-GFP in cells deleted of spoIIQ. Deletion of sigG did not affect the formation of a full second cap of CotE-GFP localization (Fig. S61). Thus, it is likely that the encasement phenotype caused by deletion of spoIIQ is mediated post-transcriptionally, by interactions across the forespore membranes between SpoIIQ in the forespore and proteins in the mother cell.

Discussion

Encasement organizes spore coat morphogenesis

Our previous work established the existence of four distinct spore coat layers in mature spores; however, we did not address the organization of the spore coat at early time points during morphogenesis. The early localization of 19 coat protein fusions (kinetic classes I–III) from all four coat layers suggested that the initial localization of spore coat proteins to the MCP pole may be organized into distinct layers and we found that this is the case (Fig. 3). These data support an earlier model introduced by Driks et al. that proposed the deposition of a scaffold composed of SpoIVA and CotE and the morphogenesis of the coat within and over this skeleton structure (Driks et al., 1994). We add to the Driks et al. model by showing that components of all four spore coat layers, including the principal morphogenetic proteins, are present in the scaffold. Recent studies have examined the critical roles of SpoVM and SpoIVA to the earliest stages of coat morphogenesis (Ramamurthi et al., 2006; 2009; Ramamurthi and Losick, 2008). Here we find that SpoIVA and SpoVM are two members of a group of at least 19 spore coat proteins that contribute to formation of the spore coat scaffold.

We described in detail the temporal dynamics of encasement by the spore coat. For the early-localizing coat proteins, the transition from polar scaffold to concentric layers occurs in three successive waves of encasement. First, proteins of the innermost spore coat layer including the morphogenetic proteins SpoVM, SpoIVA, SpoVID encase the forespore concurrent with engulfment. We do not yet know the localization kinetics of the inner coat morphogenetic protein SafA as we have been unable to produce a functional fusion. It is likely that SafA would belong to kinetic class I as it directly interacts with the class I protein SpoVID (Ozin et al., 2000; Costa et al., 2006), and the SafA-dependent fusions YaaH-GFP and YuzC-GFP are members of kinetic class I. Second, proteins of class II including the morphogenetic protein of the outer coat, CotE, encase the forespore beginning soon after the completion of engulfment. This wave of encasement begins with the nucleation of the outer coat layer on the MCD pole. Finally, after spores turn phase dark, we begin to see a nucleation of the crust morphogenetic protein, CotZ, and other proteins of class III on the MCD pole.

Importantly, proteins are still being added to the more internal layers of the coat after the morphogenetic proteins of that layer have completed encasement. Kinetic classes IV–VI also include proteins from all four layers. This implies that the maturing coat remains permeable in order to allow access to late-expressed coat proteins. For instance, the σK-dependent coat protein fusions OxdD–GFP (inner coat) and LipC–GFP (basement layer) have predicted molecular weights of 70 and 53 kD respectively, suggesting that the spore coat is capable of allowing access to relatively large fusion proteins late in coat morphogenesis. Permeability is a requirement of spore coat function as germinants must be able to access germination receptors localized at the inner forespore membrane underneath both the cortex and the coat (Griffiths et al., 2011). The level of permeability may change as the spore coat matures. In a minority of cases, coat proteins appear to localize as patterns of puncta covering the spore surface (Costa et al., 2004; van Ooij et al., 2004; Imamura et al., 2010). These localization patterns may be maintained by a complex web of protein interactions within the spore coat; however, they do not appear to correlate with any class of genetic interactions, transcriptional regulation or localization kinetics.

The sequential nature of encasement by the morphogenetic proteins may be a mechanism that organizes the spore coat. SpoIVA has been shown to multimerize in an ATP-dependent fashion (Ramamurthi and Losick, 2008), SafA has been shown to self-interact and contains a predicted ATP-binding domain (Ozin et al., 2000; Costa et al., 2006), and CotE and CotZ have been shown to self-interact biochemically (Little and Driks, 2001; Krajcíkováet al., 2009). The polymerization and accumulation of each morphogenetic protein would create binding sites for each of its dependent fusions. This type of self-assembly based mechanism is analogous to the assembly of clathrin onto clathrin-coated vesicles during endocytosis (Schmid and McMahon, 2007).

Spore coat morphogenesis

In this work we have expanded the number of proteins involved in encasement from two, SpoVM and SpoVID (Wang et al., 2009), to four with the addition of SpoIIQ and SpoIIIAH. Although SpoIIQ and SpoIIIAH are not required for encasement by Class I proteins, they affect all the other classes. In addition, the mother cell-specific transcription factors, SpoIIID, σK and GerE also contribute to various degrees to spore encasement. We have not considered the roles of the mother cell transcription factor GerR or the post-translational regulator SpoVIF, which regulates GerE. A recent study characterized an effect of GerR on late-expressed spore coat gene promoters (Cangiano et al. 2010) and SpoVIF was also shown to affect the abundance of late-expressed spore coat gene transcripts (Kuwana et al. 2004). Thus, it is possible that GerR and SpoVIF may also affect the localization kinetics of late-expressed spore coat protein fusions.

Below we describe, step-by-step, the process of spore coat morphogenesis and the contributions of this study (Fig. 7).

Figure 7.

A model of spore coat morphogenesis. In the left column we list the steps of spore coat morphogenesis that are distinguishable by fluorescence microscopy. The centre column contains diagrams of spore coat morphogenesis. Red lines indicate membranes of the sporangium, grey lines indicate the outline of cells at late time points. Layers of the spore coat are colour-coded as in McKenney et al. (2010) and Fig. 3: Cyan = Basement layer, Gold = Inner coat, Blue = Outer coat, Maroon = Crust. Genetic requirements of coat morphogenesis are indicated between the appropriate stages as identified by Roels et al. (1992) (spoIVA), Wang et al. (2009) (spoVID and spoVM) and our study. The right column indicates the transcription factors necessary for each step of spore coat morphogenesis.

Basement layer encasement

Spore coat morphogenesis begins shortly after asymmetric division of the sporangium on the MCP pole of the forespore surface. The initial localization of the coat to the forespore surface has been described to be under the genetic control of spoVM and spoIVA (Roels et al., 1992; Ramamurthi et al., 2006). According to our work, at least 19 coat proteins localize to this area simultaneously (Table 1, Figs 1, S1–S20 and S43–S45) to form an organized multilayered scaffold (Fig. 3) at a nucleation point near the middle of the septum that then spreads to cover the entire MCP hemisphere. As engulfment progresses, a set of at least six proteins including the morphogenetic proteins SpoVM, SpoIVA, SpoVID (and possibly SafA), track with the membrane and coat the spore surface with a basement layer of proteins (Figs S2–S7 and S43). All of the events described thus far require the transcriptional contributions of σE only, as none of these phenomena are disrupted in mutants of spoIIID, sigK or gerE (data not shown).

Establishment of spore polarity

After the completion of engulfment a second site of nucleation is detected on the far pole of the forespore, near the site of mother cell membrane fission (Fig. 1C stage 3 cell, Figs S8–S14 and S44). As we described above, the forespore protein SpoIIQ and the mother cell protein SpoIIIAH, appear to be required for the formation of a second cap of the outer coat morphogenetic protein CotE-GFP (Fig. 6E–G) and the rest of the fusions in classes II–VI (data not shown). The class I fusions GFP–SpoIVA, SpoVM–GFP, SpoVID–GFP and YaaH–GFP appear to be able to complete encasement in the absence of SpoIIQ (data not shown). SpoIIQ is an inner forespore membrane protein that has been shown to be capable of interacting directly with SpoIIIAH via a domain that is localized to the inter-membrane space between the mother cell and forespore compartments (Blaylock et al., 2004; Doan et al., 2005; Jiang et al., 2005). Deletion of spoIIQ has been shown to affect the localization of many mother cell expressed protein fusions including SpoIVFA and SpoIVFB of the σK processing complex, the SpoIID/SpoIIM/SpoIIP engulfment proteins and SpoIIIAG (Doan et al., 2005; 2009; Jiang et al., 2005; Campo and Rudner, 2006; Aung et al., 2007). Because the effect on coat morphogenesis is less severe in cells deleted of spoIIIAH than in cells deleted of spoIIQ, it is possible that the effect on coat morphogenesis is synthetic between multiple complexes. Nonetheless, this is the first demonstration of a forespore contribution to spore coat morphogenesis.

A recent paper described the spoIIQ mutant phenotype in high resolution (Doan et al., 2009). Cells deleted of spoIIQ complete engulfment; however, they are structurally unsound and frequently collapse. Under normal circumstances CotE-GFP first recognizes and localizes to the MCP pole, then recognizes and localizes to the MCD pole after the completion of engulfment. Thus, it is possible that the absence of spoIIQ and the failure of CotE-GFP to form a 2nd cap reflect a loss of spore polarity. Importantly, the establishment of spore polarity appears to be independent of membrane fission, as cells deleted of spoIIQ complete engulfment during sporulation by resuspension (Sun et al., 2000). Proteins such as SpoIVFA, SpoIVFB, BofA, SpoIIIAH and SpoIIIE and the engulfment machinery are prime candidate factors that should be present at the MCD pole upon the completion of engulfment. These proteins may act synergistically in defining spore polarity and in nucleating the MCD cap of the spore coat.

MCD cap expansion

In cells deleted of spoIIID, 12 of 13 class II and III fusions were arrested as a single cap and were unable to complete encasement (Figs S48–S60). These data suggest that SpoIIID (or a SpoIIID target represented by X in Fig. 7) is necessary for encasement waves II and III. Our quantitative analysis of CotE-GFP (and CotM-GFP and CotO-GFP, data not shown) suggests that SpoIIID affects expansion of the MCD pole cap, as most class II and all class III proteins appear to form a small MCD pole cap in a fraction of cells deleted of spoIIID (Figs 4, S48–S60).

Comparing the morphogenesis phenotypes of endogenously expressed and overexpressed CotZ-GFP (Class III) in the transcription factor deletion strains allowed us to distinguish between effects on promoter activity ‘in cis’ and the effects of other unknown factors that may affect CotZ-GFP morphogenesis. Overexpression was able to partially rescue the CotZ-GFP localization phenotype in spoIIID deletion cells (Fig. 5), driving encasement by CotZ-GFP to a stage similar to that seen with CotE-GFP in spoIIID deletion cells (Fig. 4). Previous experiments described a 75% reduction in promoter activity from a transcriptional fusion of lacZ to cotE in spoIIID deletion cells (Zheng and Losick, 1990). If the polymerization of morphogenetic proteins is necessary to drive encasement, it is possible that encasement by the overexpressed CotZ-GFP is limited by low levels of CotE in cells deleted of spoIIID. Thus, full encasement by the crust may require full encasement by the underlying outer coat.

Conclusions

In this study we have characterized the dynamics of spore coat morphogenesis by examining the localization kinetics of 41 spore coat protein fusions in time-course experiments. We have categorized the fusions into six distinct kinetic classes and have identified several genetic factors that are necessary to produce this diversity in spore coat protein localization. With our description of the role of SpoIIQ and SpoIIIAH in directing encasement, we have integrated the morphogenesis of the spore coat with the establishment of polarity of the forespore.

In addition to coat morphogenesis, the septum and the MCP pole are the locations of important phenomena during sporulation: translocation of the bisected chromosome into the forespore, degradation of the septal peptidoglycan, engulfment and inter-compartmental signalling (Errington, 2003). Proteins involved in these processes have potential roles in the establishment of the MCD pole and in spore coat morphogenesis.

The functional significance of spore polarity is unknown; however, polarity has been observed in the structure of the exosporium of Bacillus anthracis. Time-lapse microscopy of germinating B. anthracis spores showed that vegetative cells grow exclusively out of the equivalent of the MCD pole (Steichen et al., 2007). Perhaps, in addition to co-ordinating the morphogenesis of the MCD cap of the spore coat, SpoIIQ and SpoIIIAH also co-ordinate the assembly of a site that is easily broken down and cracked open during germination. These data suggest that coat assembly is fully integrated into the morphogenesis of the forespore itself. It should not come as a surprise that the important task of encasing the spore genome in its protective layers is co-ordinated from within.

Experimental procedures

Strains, plasmids and primers

All B. subtilis strains used here were derivatives of the wild-type strain PY79 (Youngman et al., 1984) and are listed in Table S2. The library of spore coat proteins fused to GFP has been described previously (Wang et al., 2009; McKenney et al., 2010) with one additional fusion (CwlJ-YFP), whose generation is described in the Supplemental methods in Supporting information.

Growth and sporulation conditions

Cells were incubated at 37°C in hydrolysed casein medium to A600 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

Fluorescence microscopy was performed as described before (Kim et al., 2006). Briefly, 1 ml aliquots of the sporulating cultures were transferred into microcentrifuge tubes and spun down at 8000 r.p.m. for 2 min in a microcentrifuge. Pellets were resuspended in 100 µl PBS supplemented with the membrane dye FM4-64 (Invitrogen) at 1.5 µg ml−1 final concentration. Samples intended for deconvolution were resuspended in SM medium containing 0.5 µg ml−1 Mitotracker Red (Invitrogen). Pellets were resuspended in PBS containing 1.0 µg ml−1 Mitotracker Red. Two microlitres of the concentrated sample were placed on a microscope slide and covered by a poly-l-lysine-treated coverslip for analysis. Images were taken using a Nikon 90i motorized fluorescent microscope equipped with a Roper 1 K monochrome digital camera and driven by the NIS-Elements AR 3.0 software. A 100× Plan Fluor 1.3NA objective (Nikon) was used for all image collection. Images were processed with ImageJ (http://rsbweb.nih.gov/ij/) and GIMP (http://www.gimp.org) for minor adjustments of brightness, contrast and colour balance.

Image analysis

PSICIC.  Phase-contrast and fluorescence microscopy images were analysed using PSICIC (Guberman et al., 2008), and was performed as described previously (McKenney et al., 2010). PSICIC establishes an internal co-ordinate system that facilitates comparison of cells within a field despite natural variations in cell size and shape. Each peak of fluorescence was called by identifying the first moment of intensity around a local maximum. The intensity value of each width-line was recorded as the sum of the intensities of a number of points spaced evenly on either side of the mid-line. The average width of a B. subtilis cell is 9.4 ± 0.9 pixels (n = 57), where 1 pixel equals 80 nm. We chose to use six evenly spaced points on both sides of the mid-line, for a total of 13 points, to ensure sampling of each pixel. For each width-line intensity value, we summed the intensities of the seven middle points of each width-line. The area of these seven points is roughly equal to the area of the smallest coat protein to be sampled, CotZ-GFP at hour 3.0 (data not shown). All samples were collected as 5-slice z-series with 0.3 µm spacing with phase contrast, TexasRed and FITC filters at each z-step. 3D deconvolution was performed on all z-series using a blind theoretical PSF for 30 iterations with Autoquant 1.4.1 (Media Cybernetics).

MCD cap quantification.  All measurements were performed using ImageJ. Background fluorescence was quantified by averaging the fluorescence intensity of 10 random cell-free regions of equal area using the Measure function. An ROI (region of interest) threshold (Threshold function) of the measured background multiplied by a constant of 4 was chosen empirically as it allows maximum separation of MCP and MCD caps in wild-type cells expressing CotE-GFP at hour 5.0. ROIs were drawn using Analyze particles. Cells were then examined and scored individually as ‘Single Cap’ (a sporangium that contains only a single ROI of the MCP cap) or ‘2 Caps’ (a sporangium that contains an MCP and MCD cap). Sporangia containing 2 Caps are subdivided into ‘Unconnected’ (2 separated ROIs of the MCP and MCD cap) and ‘Connected’ (a single continuous ROI that contains a protrusion of above-threshold pixels towards the MCD pole). The area of MCD caps in pixels was recorded for each sporangium containing ‘Unconnected’ CotE-GFP caps. The normal progression of CotE-GFP and CotZ-GFP localization in wild-type cells is from ‘Single Cap’ to ‘Unconnected’ to ‘Connected’.

Acknowledgements

We are grateful to Adam Driks, Kumaran Ramamurthi and Melissa DeFrancesco for critical reading of the manuscript and Jonathan Dworkin, Lee Kroos, Richard Losick and David Rudner for strains. We thank Kwang-Hwa Wu for the construction of the spoIIIAH deletion and Jenny Yeh for determining genetic interactions of the CwlJ–YFP fusion. We acknowledge the financial support of NIH grants GM081571 and GM092616 to P.E. P.T.M. was supported by an NIH training grant in Developmental Genetics 5T32HD007520 and the Horizon Fellowship in the Natural and Physical Sciences from the NYU Graduate School of Arts and Sciences.

Ancillary