Dual localization pathways for the engulfment proteins during Bacillus subtilis sporulation


*E-mail kpogliano@ucsd.edu; Tel. (+1) 858 822 1314; Fax (+1) 858 822 1431.


Engulfment in Bacillus subtilis is mediated by two complementary systems, SpoIID, SpoIIM and SpoIIP (DMP), which are essential for engulfment, and the SpoIIQ-SpoIIIAGH (Q-AH) zipper, which provides a secondary engulfment mechanism and recruits other proteins to the septum. We here identify two mechanisms by which DMP localizes to the septum. The first depends on SpoIIB, which is recruited to the septum during division and provides a septal landmark for efficient DMP localization. However, sporangia lacking SpoIIB ultimately localize DMP and complete engulfment, suggesting a second mechanism for DMP localization. This secondary targeting pathway depends on SpoIVFA and SpoIVFB, which are recruited to the septum by the Q-AH zipper. The absence of a detectable localization phenotype in mutants lacking only SpoIVFAB (or Q-AH) suggests that SpoIIB provides the primary DMP localization pathway while SpoIVFAB provides a secondary pathway. In keeping with this hypothesis, the spoIIB spoIVFAB mutant strain has a synergistic engulfment defect at septal thinning (which requires DMP) and is completely defective in DMP localization. Thus, the Q-AH zipper both provides a compensatory mechanism for engulfment when DMP activity is reduced, and indirectly provides a compensatory mechanism for septal localization of DMP when its primary targeting pathway is disrupted.


The phagocytosis like process of engulfment (Fig. 1) is a key step in the sporulation pathway of Bacillus subtilis, B. anthracis and various Clostridia species (reviewed by Errington, 2003; Hilbert and Piggot, 2004). Shortly after the asymmetrically positioned cell division event that generates the smaller forespore and larger mother cell, the mother cell membrane migrates around the forespore, until the leading edge of the engulfing membrane meets and fuses to release the forespore into the mother cell cytoplasm. Engulfment therefore mediates a striking reorganization of the sporangium, from two cells that lie side by side, to a unique structure in which one bacterial cell lies within the cytoplasm of another. Engulfment provides an ideal system to study how bacterial cells are able to move macromolecules, localize proteins and catalyse membrane fusion, the final step of both engulfment and cell division.

Figure 1.

The sporulation pathway of B. subtilis.
A. Early in sporulation, two FtsZ rings assemble at each pole of the cell.
B. The sporulation septum separates the smaller forespore and larger mother cell, and traps the forespore chromosome, which is pumped into the forespore by SpoIIIE (arrows).
C. The first step of engulfment, septal thinning, commences as the peptidoglycan within the septum (grey) is thinned, a step requiring the mother cell-expressed proteins SpoIID, SpoIIM, SpoIIP, and facilitated by SpoIIB.
D. Next, the mother cell membrane migrates up and around the forespore, which also requires DMP and, under certain conditions, the Q-AH zipper.
E and F. (E) Ultimately, the migrating membrane meets and (F) fuses at the forespore pole, a step requiring the SpoIIIE DNA translocase (Sharp and Pogliano, 1999). Proteins shown above the arrow are always essential for the step, while those shown below slow but do not completely block engulfment under some conditions.
G. Mutants completely defective in septal thinning (i.e. spoIID, spoIIM or spoIIP null mutations) produce a bulge phenotype in which the growing forespore pushes through the unthinned septum.
H. Mutants that slow septal thinning (i.e. spoIIB null mutations) also produce a bulge, but the engulfing membranes are ultimately able to migrate around septal peptidoglycan.

The first step of engulfment is septal thinning, during which septal peptidoglycan is degraded, starting in the middle of the septum and proceeding towards the edges (Fig. 1C). This step requires three essential engulfment proteins, SpoIID and SpoIIM, which are expressed in the mother cell (Lopez-Diaz et al., 1986; Driks and Losick, 1991; Smith and Youngman, 1993), and SpoIIP, which is expressed in both cells but only required in the mother cell (Frandsen and Stragier, 1995; Dworkin and Losick, 2005). Both SpoIID (Abanes-De Mello et al., 2002) and SpoIIP (Chastanet and Losick, 2007) have peptidoglycan hydrolase activity, demonstrating that septal thinning involves peptidoglycan degradation. The second step of engulfment is the migration of the engulfing membrane around the forespore (Fig. 1D–E). Thus far, no proteins specifically essential for membrane migration under all conditions have been identified (although SpoIIQ is necessary under certain conditions, as described below). However, certain missense mutations in spoIID and spoIIP slow both septal thinning and membrane migration, suggesting that these proteins are required throughout engulfment (Abanes-De Mello et al., 2002). This conclusion is supported by the localization of SpoIID, SpoIIM and SpoIIP to the leading edge of the engulfing membrane (Abanes-De Mello et al., 2002). A secondary mechanism for membrane migration depends on the interaction between the forespore protein SpoIIQ and its mother cell ligand SpoIIIAH (Blaylock et al., 2004; Broder and Pogliano, 2006). This interaction provides a DMP-independent mechanism for membrane migration that can compensate for reduced DMP activity and mediate engulfment in cells in which the peptidoglycan has been enzymatically removed (Broder and Pogliano, 2006). Thus, membrane migration appears to be mediated by two separate engulfment modules, the DMP module and the Q-AH module.

We here investigate the role of SpoIIB in engulfment. Mutants lacking SpoIIB show uneven dissolution of septal peptidoglycan, although most sporangia ultimately complete engulfment, with the engulfing membranes migrating around this residual peptidoglycan (Margolis et al., 1993; Perez et al., 2000). These results suggest either that SpoIIB spatially regulates septal thinning, that it recruits peptidolgyan hydrolases to the septum, or that it is a peptidoglycan hydrolase itself. The studies in this article, and those of Chastanet and Losick (2007), which were published while our work was in preparation, support a role for SpoIIB in localization of DMP. Specifically, we here show that SpoIIB is recruited to the sporulation septum during division where it is necessary for efficient DMP localization. However, mutants lacking SpoIIB ultimately complete engulfment and recruit DMP to the sporulation septum in a manner that depends on SpoIVFA and SpoIVFB, which are recruited to the septum by the Q-AH zipper (Doan et al., 2005; Jiang et al., 2005). Thus, DMP localization is normally mediated by SpoIIB, which appears to serve (directly or indirectly) as a septal landmark protein, but the Q-AH zipper can also mediate DMP localization in the absence of SpoIIB. The secondary engulfment mechanism (Q-AH), therefore also provides a secondary mechanism to localize the DMP proteins to the septum.


Localization of SpoIIB

Previous studies demonstrated that SpoIIB, a protein required for efficient septal thinning, localizes to the sporulation septum in an FtsZ-dependent manner (Perez et al., 2000). However, these studies were based on immunofluorescence experiments with a SpoIIB–myc fusion protein and we were unable to visualize the protein in living cells. We revisited the subcellular distribution of SpoIIB by creating a fusion protein in which the C-terminus of SpoIIB is fused to the red fluorescent protein mCherry (Shaner et al., 2004), which can be visualized in living cells and is fluorescent when fused to a secreted protein (Chen et al., 2005). The spoIIB–mCherry gene fusion replaced the wild-type spoIIB gene and was able to fully support wild-type levels of spore formation (Table S1). Sporulating cells were harvested and stained with mitotracker green, which stains cell membranes, and allows the various stages of engulfment to be visualized. The sporulation septum is initially flat (Fig. 2A, arrow) and then curves around the forespore (Fig. 2C, arrow) until it is fully engulfed (Fig. 2C, double arrowhead). SpoIIB–mCherry initially localized to the middle of the sporulation septum, where septal thinning likely starts (Fig. 2A), similar to the SpoIIB–myc fusion (Perez et al., 2000). However, we found the SpoIIB–mCherry fusion localized to the leading edges of the engulfing mother cell membrane (Fig. 2B) until the completion of membrane migration (Fig. 2C, double arrowhead). This localization pattern was in contrast to previous studies, in which SpoIIB quickly disappeared upon the start of membrane migration (Perez et al., 2000). This is likely due to instability of the c-myc epitope previously used to localize SpoIIB, as our preliminary results indicate that SpoIIB–mCherry and native SpoIIB can be detected at later times of sporulation than SpoIIB–myc (J. Fredlund-Gutierrez, unpubl. data). SpoIIB–mCherry also moved around the forespore in a manner that depended on the DMP proteins but not on SpoIIQ (Fig. 2E–H). Thus, our new results suggest that SpoIIB localizes in a manner more similar to the DMP proteins than previously appreciated.

Figure 2.

Localization of SpoIIB–mCherry. Live cells expressing SpoIIB–mCherry (red) and stained with the membrane stain Mitotracker Green (green). Sporulation was induced by re-suspension at 30°C and samples were harvested at the indicated time.
A–C. SpoIIB–mCherry in wild type (KP1026) (A) at t2 (2 h after the initiation of sporulation), (B) at t2.5 and (C) at t3. SpoIIB localizes to the septum (arrow), and to the potential second division site (arrowhead) at the opposite pole. In engulfing sporangia, SpoIIB localizes to the leading edge of the engulfing membrane (C, arrow), where it remains until the completion of engulfment (C, double arrowhead).
D. Cartoon representation of SpoIIB–mCherry during engulfment.
E–H. SpoIIB–mCherry in (E) spoIID (KP1027), (F) spoIIP (KP1028), (G) spoIIM (KP1029) and (H) spoIIQ (KP1030) strains. SpoIIB–mCherry is restricted to the septal midpoint in spoIID and spoIIP mutants (arrows, E and F), but moves to the edge of the septum in the spoIIM mutant (arrow, G), and shows wild-type localization in the spoIIQ mutant (H).
I. Co-immunoprecipitation of SpoIID by SpoIIP-FLAG. KP4156 (spoIIP-FLAG; lanes 4–6) and the negative control strain PY79 (wild type; lanes 1–3) were solubilized by 0.5% digitonin after 2.5 h of sporulation at 37°C and then subjected to immunoprecipitation using anti-FLAG antibodies. Proteins were visualized by Western blotting using anti-SpoIIP (top), anti-SpoIID (middle) and anti-SpoIIIE (negative control; bottom). W, U and B represent whole cell, unbound and bound fractions respectively.
Scale bar in (A), 1 μm.

Interactions between the DMP proteins

The similarity in localization patterns of SpoIIB and the DMP proteins throughout engulfment (Abanes-De Mello et al., 2002) suggested that these proteins might exist in a complex with each other. Indeed, it has recently been reported that SpoIID and SpoIIP interact by affinity chromatography (Chastanet and Losick, 2007). In keeping with this finding, we noted that SpoIID and SpoIIP-FLAG were co-immunoprecipitated from sporulating B. subtilis cells with high efficiency (Fig. 2I), suggesting that these proteins are in a complex. However, similar experiments failed to detect an interaction between SpoIIB-FLAG and SpoIIP. Thus, if these proteins directly interact, the interactions are not sufficiently robust to allow their co-immunoprecipitation under our conditions. Our studies and those of Chastanet and Losick (2007) clearly indicate that SpoIID and SpoIIP interact.

Our cell biological studies provide support for additional interactions between the engulfment proteins. For example, in the absence of either spoIIP or spoIID, GFP–SpoIIM accumulated at the middle of the septum (Fig. 3C and D, arrowheads), as did SpoIIB–mCherry (Fig. 2E and F). This suggested that either SpoIIP and SpoIID or the septal thinning they mediate are required to move the engulfment complex across the septum. SpoIIM was also required for localization of SpoIIP to the septum, as the spoIIM mutant resulted in the almost complete delocalization of SpoIIP (Fig. 3H). We did not investigate the localization of GFP–SpoIID, because it interacts with SpoIIP and because the GFP fusion protein is degraded to release the transmembrane segment fused to GFP, resulting in high background of non-localized GFP even in wild-type strains (Abanes-De Mello et al., 2002). These results suggest that SpoIIM is necessary for septal localization of SpoIIP (and we presume also SpoIID), while SpoIID and SpoIIP or septal thinning is necessary for the movement of the DMP proteins and SpoIIB to the edges of the septum.

Figure 3.

Localization of GFP–SpoIIM and GFP–SpoIIP in mutant backgrounds. Localization of GFP–SpoIIM (A–D, green) and GFP–SpoIIP (E–H, green) in sporangia sampled 1 h and 45 min after the initiation of sporulation by re-suspension (t1.75). Membranes are stained with FM 4-64 (red).
A–D. GFP–SpoIIM localization in (A) wild type (KP721), (B) spoIIB (KP762), (C) spoIID (KP763) and (D) spoIIP (KP722) strains. GFP–SpoIIM localizes at the leading edge of the engulfing mother cell membrane in wild type (arrow), but is randomly localized in the spoIIB strain (arrowheads) and restricted to the septal midpoint in spoIID and spoIIP strains (double arrowheads).
E–H. Localization of GFP–SpoIIP in (E) wild type (KP722), (F) spoIIB (KP774), (G) spoIID (KP775) and (H) spoIIM (KP776) strains. GFP–SpoIIP localizes to the leading edge and assembles distinct foci at the leading edge and within the septum of wild type (arrow), but shows mostly random distribution in spoIIB and spoIIM strains (arrowheads) and is trapped at the middle of the septum in the spoIID strain (double arrowhead).
Scale bar in (A), 1 μm.

Localization of SpoIIM and SpoIIP to the septum partially depends on SpoIIB

An attractive candidate for a protein mediating septal localization of DMP is SpoIIB, which is synthesized before polar septation (Margolis et al., 1993). We therefore investigated the genetic requirements for DMP localization. We first localized GFP–SpoIIM and GFP–SpoIIP in strains lacking SpoIIB. The spoIIB mutation showed reduced septal targeting of both GFP–SpoIIM (Fig. 3A–B) and GFP–SpoIIP (Fig. 3E–F) compared with wild-type strains, with the proteins appearing randomly distributed throughout the mother cell membrane. These results suggest that SpoIIB is necessary for efficient localization of SpoIIM and SpoIIP, as has also been concluded by Chastanet and Losick (2007). However, one caveat to these studies is that although the N-terminal GFP fusions to SpoIIM and SpoIIP support sporulation and engulfment at wild-type efficiencies (Abanes-De Mello et al., 2002), they are only partially functional, and produce synergistic engulfment defects when introduced into strains with mutations that normally just slow engulfment (such as spoIIQ, Broder and Pogliano, 2006). Indeed, GFP–SpoIIM and GFP–SpoIIP failed to support engulfment when introduced into spoIIB strains (Fig. 3B and F), which normally complete engulfment (Margolis et al., 1993; Perez et al., 2000) and thus should show some localization of the DMP proteins.

Rescue of the synergistic engulfment defect of SpoIIP–GFP, spoIIB strains

Our concern that GFP fusions might not accurately reflect localization of the engulfment proteins led us to seek alternative localization methods. First, we expressed GFP–SpoIIP in the presence of wild-type SpoIIP, which rescued the synergistic engulfment defect and sporulation defect in spoIIB null strains (Fig. 4A–B, Table S1). In these strains, sporangia that were engulfing showed localization of GFP–SpoIIP to the leading edges of the engulfing membrane (Fig. 4B, arrows), while sporangia that were not engulfing showed only partial localization (Fig. 4B, arrowheads). Quantitative analysis of GFP–SpoIIP in engulfing spoIIB sporangia showed that SpoIIP was enriched approximately eightfold at the leading edge of the engulfing membrane relative to the mother cell membrane, a similar enrichment as in wild-type sporangia (Fig. 4; Fig. S1). This suggests that the absence of SpoIIB slows, but does not completely block localization of SpoIIP to the leading edge of the engulfing membrane. We also used immunofluorescence microscopy to localize untagged SpoIIP and found that localization of SpoIIP was clearly reduced in the spoIIB strain (Fig. S2). Together our results suggest that SpoIIB is directly or indirectly required for the efficient localization of SpoIIP and SpoIIM to the engulfing membrane and that there is likely a secondary and less efficient mechanism to localize these proteins in the absence of SpoIIB.

Figure 4.

Localization of GFP–SpoIIP in the presence of untagged SpoIIP to rescue synergistic engulfment defects. Localization of GFP–SpoIIP (green) in sporangia also expressing untagged SpoIIP sampled at t1.75. Membranes are stained with FM 4-64 (red). Pixel intensities (fourth column) of GFP–SpoIIP signals are represented as surface plots from the highest intensity (red) to the lowest (blue), as described in Experimental procedures.
A. In wild type (KP1035), GFP–SpoIIP localized to the leading edge of the engulfing mother cell membrane with some foci forming at the septal midpoint and around the forespore (arrows).
B. In the spoIIB mutant (KP1036), GFP–SpoIIP localizes to the leading edge in engulfing sporangia and to the edges of septal bulges (arrows). Non-engulfing sporangia show additional delocalized GFP–SpoIIP fluorescence (arrowhead).
C–E. GFP–SpoIIP in (C) spoIIIAGH (KP1037), (D) spoIIQ (KP1038) and (E) spoIVFAB (KP1039) strains. Nearly wild-type localization is seen in these strains with a slight increase in mother cell fluorescence (arrows) and fewer clear foci at the septal midpoint (double arrowheads).
F–H. GFP–SpoIIP in (F) spoIIB spoIIIAGH (KP1040), (G) spoIIB spoIIQ (KP1041) and (H) spoIIB spoIVFAB (KP1042) strains. GFP–SpoIIP shows a more random distribution (arrowheads) with some accumulation at regions with increased membrane fluorescence (*) and septa.
Scale bar in (A), 1 μm.

Colocalization of SpoIIB with FtsZ and SpoIIM and SpoIIP

If SpoIIB acts as a septal landmark for DMP targeting, then it should localize to the septum with FtsZ and before DMP. To test this hypothesis, we colocalized SpoIIB–mCherry with FtsZ–GFP, GFP–SpoIIM and GFP–SpoIIP. These studies showed that SpoIIB initially localizes with FtsZ, but remains localized after FtsZ is lost from the completed septum (Fig. 5A–B). Furthermore, SpoIIB localized to the septum earlier in sporulation than GFP–SpoIIM and GFP–SpoIIP (Fig. 5C–F). To more precisely measure the relative times at which these proteins reach the septum, we took advantage of observations that the forespore chromosome is translocated into the chromosome ∼15 min after polar septation and ∼30 min before the completion of engulfment (Fig. 1A–C; Pogliano et al., 1999; Errington, 2001), and that incompletely and completely translocated forespore chromosomes are readily distinguished by DAPI staining (Fig. 5C, arrow and arrowhead respectively). Sporangia with completely translocated forespore chromosomes are therefore more advanced in sporulation than those with incompletely translocated chromosomes. Quantification of the localization data revealed that 90% of early sporangia with incompletely translocated forespore chromosomes showed SpoIIB–mCherry but not GFP–SpoIIP, while 94% of later sporangia (with a completely translocated chromosome) showed both proteins. Similarly, 92% of early sporangia showed SpoIIB–mCherry but not GFP–SpoIIM, while 87% of later sporangia showed both proteins (Table S2). Thus, the temporal order in which FtsZ, SpoIIB, SpoIIM and SpoIIP reach the septum is similar to their genetic dependency.

Figure 5.

Colocalization of SpoIIB–mCherry with FtsZ–GFP, GFP–SpoIIM and GFP–SpoIIP. Sporulation of SpoIIB–mCherry strains was conducted at 30°C, and the sporangia were counterstained with the DNA stain DAPI (blue).
A and B. Colocalization of SpoIIB–mCherry (red) and FtsZ–GFP (green) in strain KP1043 induced to sporulate by re-suspension at 30°C. Samples were collected at t2(A) or t3 (B). FtsZ–GFP localizes to polar division sites first (arrow) followed by SpoIIB–mCherry (arrowhead). SpoIIB–mCherry remains localized after FtsZ–GFP is lost from the completed septum (double arrowhead).
C and D. Colocalization of SpoIIB–mCherry (red) and GFP–SpoIIM (green) in strain KP1032 at t2.5(C) or t3 (D). SpoIIB–mCherry (KP1032) localizes to the septum first (arrow), followed by GFP–SpoIIM (arrowheads).
E and F. Colocalization of SpoIIB–mCherry and GFP–SpoIIP in strain KP1031 at t2.5 (E) or t3 (F). SpoIIB–mCherry localizes to the septum first (arrow), followed by GFP–SpoIIP (arrowhead).
Scale bar in (A), 1 μm.

Evidence for a SpoIIB-independent targeting pathway

Our experiments with cells expressing GFP–SpoIIP in the presence of wild-type SpoIIP suggested the possibility of a second, SpoIIB-independent mechanism for DMP localization, as sporangia lacking spoIIB ultimately localize DMP (Fig. 4B) and complete engulfment (Margolis et al., 1993; Perez et al., 2000). One obvious candidate for a secondary DMP targeting pathway is the SpoIIQ-SpoIIIAH zipper, which recruits proteins required for engulfment-dependent gene expression to the septum (Blaylock et al., 2004; Doan et al., 2005; Jiang et al., 2005). We therefore tested GFP–SpoIIP localization in spoIIQ and spoIIIAGH single mutants, which complete engulfment slightly more slowly than wild type (Broder and Pogliano, 2006). These strains also expressed native SpoIIP to avoid synergistic engulfment defects. GFP–SpoIIP localized to the septum in both single mutants (Fig. 4C–D), although two subtle phenotypes were noted. First, the pixel intensity data sometimes revealed a slightly increased fluorescence within forespore-distal regions of the mother cell membrane (Fig. 4C–D, arrows), suggesting that some GFP–SpoIIP was released from the septum in the absence of Q or AH (this is most evident in Fig. 4D). Second, while GFP–SpoIIP normally forms foci surrounding the forespore, as well as at the leading edge of the engulfing membrane (Fig. 4A), in the absence of either Q or AH, foci were only formed at the leading edge of the engulfing membrane (Fig. 4C–D, double arrowheads).

This suggests that the interaction of GFP–SpoIIP with the Q-AH zipper [which also forms foci around the forespore (Blaylock et al., 2004; Rubio and Pogliano, 2004)] is necessary for the assembly of multiple SpoIIP foci surrounding the forespore and for the efficient retention of GFP–SpoIIP at the septum. In the spoIIB spoIIQ and spoIIB spoIIIAGH double mutants, GFP–SpoIIP localization was more severely affected, and many sporangia showed a uniform distribution of GFP–SpoIIP within the mother cell membrane (Fig. 4F–G, arrowheads). In the absence of SpoIIQ and SpoIIB, GFP–SpoIIP sometimes formed foci within the mother cell, mostly at sites that showed increased FM 4-64 membrane staining (Fig. 4G, asterisk), suggesting that it simply accumulated within the cell as locations that showed more mother cell membrane. These results suggest that the Q-AH zipper is involved in targeting DMP to the septum in the absence of SpoIIB, as in the spoIIB strain inactivation of either component of the zipper severely compromised DMP targeting to the sporulation septum.

It seemed possible that the engulfment proteins were not directly recruited by the Q-AH zipper, but rather were recruited by another protein that depended on the zipper. Two candidates for such proteins are SpoIVFA and SpoIVFB, which are required for activation of σK and are recruited to the septum by the Q-AH zipper (Doan et al., 2005; Jiang et al., 2005). To determine if these proteins were involved in DMP localization, we examined the effect of a spoIVFAB mutation that abolishes both proteins on GFP–SpoIIP localization. While the single spoIVFAB mutation had only subtle effects on GFP–SpoIIP localization (similar to those described above for spoIIQ spoIIIAGH mutants; Fig. 4E, double arrowhead), a spoIIB spoIVFAB double mutant showed a phenotype similar to spoIIB spoIIQ and spoIIB spoIIIAGH mutants, with GFP–IIP showing a uniform mother cell membrane distribution (Fig. 4H, arrowhead). Together, these results suggest that the DMP complex can be localized to the septum by a secondary pathway involving the Q-AH zipper and SpoIVFA or SpoIVFB, but that this pathway is only essential when the primary DMP localization pathway (SpoIIB) is compromised.

Elimination of both the primary (SpoIIB) and secondary (SpoIVFAB) localization pathways blocks engulfment at the stage of septal thinning

If SpoIVFA and SpoIVFB provided a compensatory pathway to localize the DMP proteins when SpoIIB was absent, then the spoIIB spoIVFAB double mutant should show a synergistic engulfment defect even in the absence of GFP fusions. To test this hypothesis, we used a membrane fusion assay that discriminates between engulfed and unengulfed sporangia (Sharp and Pogliano, 1999). This assay relies on the membrane impermeable stain FM 4-64 and the membrane permeable stain Mitotracker Green. When applied to sporangia that have completed engulfment, the forespore membranes fail to stain with FM 4-64 and stain only with Mitotracker Green (Fig. 6A, arrow), while the forespores of engulfing sporangia are accessible to both stains and therefore appear yellow in overlays (Fig. 6A, arrowhead). The spoIVFAB mutant completed engulfment at nearly wild-type efficiency (57% at t3.5 versus 77% for wild type). The spoIIB mutant showed a transient septal thinning defect, with some sporangia showing septal bulges (Fig. 6C, double arrowhead), some showing the onset of membrane migration (Fig. 6C, arrowhead) and others had completed engulfment (Fig. 6C, arrow), albeit more slowly than wild type (14% at t3.5). However, the spoIIB spoIVFAB double mutant showed a block at septal thinning, with sporangia showing no membrane migration and septal bulges (Fig. 6D, double arrowhead) that appeared nearly identical to those in the spoIIP null strain (Fig. 6E, double arrowhead). This finding is consistent with the hypothesis that SpoIVFA and SpoIVFB or another protein they recruit to the septum is able to mediate DMP localization to the septum via the Q-AH zipper when SpoIIB is absent .

Figure 6.

Synergistic engulfment defect in the absence of SpoIIB and SpoIVFAB. The completion of engulfment was followed using a membrane fusion assay using the membrane impermeable stain FM 4-64 (red) and the membrane permeable stain Mitotracker Green (MTG, green). During engulfment FM 4-64 stains all the membranes (arrowhead), while after engulfment, MTG but not FM 4-64 stains the forespore membranes (arrow). While wild type engulfs smoothly, mutants with septal thinning defects show the bulging of the forespore into the mother cell resulting in a bubble of membrane (double arrowheads). Strains were sporulated by resuspension at 37°C and scored for the percent of sporangia completing membrane fusion at t3.5. (A) wild type (PY79), (B) spoIVFAB (KP1013), (C) spoIIB (KP343) shows both bulges and complete engulfment, (D) spoIIB spoIVFAB (KP1014) shows no engulfment and mostly sporangia with bulges, (E) spoIIP (KP513) also shows a complete defect in septal thinning. Scale bar in (A), 1 μm.


We here present genetic and cell biological evidence that SpoIIB is incorporated into the septum during division and serves directly or indirectly as a landmark for localization of SpoIIM and then SpoIIP and SpoIID to the septum. Thus, SpoIIB appears to play a role similar to Caulobacter crescentus TipN, which localizes to the septum and remains at the cell pole to serve as a landmark for polar localization (Huitema et al., 2006; Lam et al., 2006). We also show that a localization hierarchy exists within the engulfment machinery, with SpoIIM being required to recruit and retain SpoIIP at the septum, which is required together with SpoIID to move the machinery around the forespore. Our results are similar to those recently published by Chastanet and Losick (2007), although our studies have allowed the identification of a second, SpoIIB-independent, pathway for DMP localization. Specifically, when we eliminated the synergistic engulfment defect observed in spoIIB strains expressing GFP–SpoIIP (by also expressing untagged SpoIIP), we noted that localization was partially rescued, with engulfing sporangia showing almost wild-type localization of GFP–SpoIIP at the leading edge of the engulfing membrane. This localization depended on SpoIVFAB, which in turn depends on the Q-AH zipper for localization (Blaylock et al., 2004; Doan et al., 2005; Jiang et al., 2005).
The DMP engulfment proteins therefore appear to have two distinct mechanisms by which they can reach the sporulation septum, SpoIIB, which assembles a septal landmark for DMP during cytokinesis, and the Q-AH zipper (via SpoIVFAB), which assembles after cytokinesis (Fig. 7). Interestingly, while the Q-AH zipper proteins and SpoIVFAB are present in the genomes of all endospore-forming bacteria, SpoIIB is present only in the Bacilli, not in the Clostridia (Stragier, 2002). This suggests either that DMP localization in the Clostridia depends entirely on the Q-AH zipper via SpoIVFAB or that the Clostridia have another primary pathway for DMP localization that can substitute for SpoIIB.
We recently reported that the Q-AH zipper mediates engulfment in cells whose cell walls have been enzymatically removed (Broder and Pogliano, 2006). This protoplast engulfment does not require SpoIVFAB, SpoIIB or DMP, indicating that the requirement for Q-AH for protoplast engulfment is not mediated by their ability to localize DMP or SpoIVFAB to the septum. However, we have here shown that in intact cells, the Q-AH zipper also makes a second apparently distinct contribution to engulfment, by localizing SpoIVFAB, which can mediate DMP localization in the absence of SpoIIB. Three observations suggest that SpoIVFAB might also interact with DMP in cells containing SpoIIB. First, the absence of SpoIVFAB (or SpoIIQ or SpoIIIAGH) results in the release of a small amount of SpoIIP from the septum into the mother cell cytoplasmic membrane (Fig. 4), suggesting that SpoIVFAB is necessary for the efficient retention of SpoIIP at the septum. Second, the absence of SpoIVFAB (or SpoIIQ or SpoIIIAGH) results in the loss of the foci that SpoIIP normally assembles around the forespore, although localization to the leading edge of the engulfing membrane is maintained (likely by SpoIIB). Finally, Doan et al. (2005) reported that SpoIVFA is slightly delocalized in the absence of DMP, suggesting an interaction between SpoIVFA and DMP. Thus, although the interaction between SpoIVFAB and the DMP proteins has not yet been detected biochemically, evidence is building that this interaction occurs in wild-type cells. If this is indeed the case, then this interaction would bring proteins involved in engulfment together with proteins involved in activating engulfment-dependent gene expression. The localization studies presented here also suggest that there are two distinct populations of SpoIIP (and likely also SpoIID), one recruited to the leading edge of the engulfing membrane by SpoIIB and a second that interacts with the protein complex containing the Q-AH zipper and the σK-processing machinery (SpoIVFA, SpoIVFB and BofA).
We emphasize that it remains unclear if the SpoIIB-dependent or SpoIVFAB-dependent localization pathways for DMP are mediated by their direct protein–protein interaction with the DMP proteins (or a subset). While the physical interaction between SpoIID and SpoIIP has been demonstrated by both affinity chromatography (Chastanet and Losick, 2007) and co-immunoprecipitation (Fig. 2), we have thus far failed to detect any robust and repeatable interaction between SpoIIB or SpoIVFAB with SpoIID, SpoIIM or SpoIIP (although we have not tested all possible combinations, as described in the Experimental procedures). Indeed, it seems possible that certain steps in the localization hierarchy might be mediated indirectly, for example, by other unknown proteins that interact with the Q-AH zipper or SpoIIB or by one protein modifying the bacterial cell wall in a manner that allows a second protein to bind. Consistent with this latter proposal, the engulfment proteins include two cell wall hydrolases likely capable of binding the wall (SpoIID and SpoIIP), although the specificity or affinity of this interaction has not yet been determined. In addition, SpoIIB and SpoIIP show limited sequence similarity with the CwlC amidase (Errington et al., 2003; Soding et al., 2005; Chastanet and Losick, 2007) and both SpoIVFA and SpoIIQ show some sequence similarity to the M23 family of endopeptidases (Rudner and Losick, 2002), some of which cleave peptide cross-bridges in peptidoglycan. It is therefore possible that these proteins modify the peptidoglycan in a manner that allows subsequent proteins to bind in a modification-specific manner.
Peptidoglycan is the perfect molecule to serve as a landmark in bacterial cells, as unlike membrane proteins, it is immobile for long periods of time (until it is degraded or recycled) so that modifications (amidation, distinct peptide cross-linking or trimming reactions) could remain in the same cellular location. It is also tempting to speculate that such modifications are necessary to allow SpoIID and SpoIIP to cleave peptidoglycan, as these enzymes must be subject to strict spatial and temporal control, as their unregulated activity could lead to cell lysis.

Figure 7.

Model for DMP localization.
A. SpoIIB (red ring and circles) is targeted to the septum during cytokinesis (ring) and remains at the septum. After septation, DMP (green square) are synthesized by the mother cell transcription factor σE, and recruited to the septum by SpoIIB. The Q-AH zipper forms by the interaction of the forespore protein SpoIIQ (light blue) with the mother cell protein SpoIIIAH (royal blue), which recruits the SpoIVFA and SpoIVFB proteins (purple) that are required for late mother cell gene expression. SpoIIB is primarily responsible for localization of DMP to the leading edge of the engulfing membrane, while SpoIVFAB allow the DMP proteins to assemble foci behind the leading edge.
B. In the absence of SpoIIB, the Q-AH zipper can indirectly mediate DMP localization via SpoIVFAB.
C. Our results suggest that there are two localization pathways for DMP, a primary SpoIIB-dependent pathway and a compensatory SpoIVFAB-dependent pathway. SpoIIB is synthesized before polar septation (Margolis et al., 1993; Perez et al., 2000) and is therefore drawn here as though the protein is present in both daughter cells, although this has not been experimentally tested.

Experimental procedures

Bacterial strains, genetic manipulations and growth conditions

Bacillus subtilis strains used in this study are derivatives of wild-type strain PY79 (Youngman et al., 1984) and are shown in Table 1. Mutations were introduced into PY79 by transformation (Dubnau and Davidoff-Abelson, 1971). B. subtilis strains were grown and sporulated at 37°C for GFP localization experiments, and at 30°C for mCherry experiments. Sporulation was induced by the resuspension method (Sterlini and Mandelstam, 1969). Cloning was performed in strain KJ801 (TG1 pcnB24-1), because the pcnB mutation lowers the copy number and partially alleviates toxicity of Bacillus membrane proteins. The ΔspoIVFAB::cat::tet allele in strains KP1013 and KP1014 was constructed from ΔspoIVFAB::cat (Lu and Kroos, 1994) using pCm::Tc to integrate the tet gene at cat (as described in Steinmetz and Richter, 1994).

Table 1.  Strains used in this study.
PY79Wild typeYoungman et al. (1984)
KP7spoIID298Lopez-Diaz et al. (1986)
KP343spoIIB::ermMargolis et al. (1993)
KP519spoIIP::tetFrandsen and Stragier (1995)
KP719spoIIM::Tn917Sandman et al. (1987)
KP721amyE::PspoIID–gfp–spoIIMΩcat, spoIIM::Tn917Abanes-De Mello et al. (2002)
KP722amyE::PspoIID–gfp–spoIIPΩcat, spoIIP::tetAbanes-De Mello et al. (2002)
KP762amyE::PspoIID–gfp–spoIIMΩcat, spoIIM::Tn917, spoIIB::ermThis study
KP763amyE::PspoIID–gfp–spoIIMΩcat, spoIIM::Tn917, spoIID298This study
KP764amyE::PspoIID–gfp–spoIIMΩcat, spoIIM::Tn917, spoIIP::tetThis study
KP765amyE::PspoIID–gfp–spoIIMΩcat, spoIIM::Tn917,ΔspoIIQ::specThis study
KP774amyE::PspoIID–gfp–spoIIPΩcat, spoIIP::tet, spoIIB::ermThis study
KP775amyE::PspoIID–gfp–spoIIPΩcat, spoIIP::tet, spoIID298This study
KP776amyE::PspoIID–gfp–spoIIPΩcat, spoIIP::tet, spoIIM::Tn917This study
KP777amyE::PspoIID–gfp–spoIIPΩcat, spoIIP::tet,ΔspoIIQ::specThis study
KP1026spoIIB–mCherryΩkanThis study
KP1027spoIIB–mCherryΩkan, spoIID298This study
KP1028spoIIB–mCherryΩkan, spoIIP::tetThis study
KP1029spoIIB–mCherryΩkan, spoIIM::Tn917This study
KP1030spoIIB–mCherryΩkan,ΔspoIIQ::specThis study
KP1031spoIIB–mCherryΩkan, amyE::PspoIID–gfp–spoIIPΩcat, spoIIP::tetThis study
KP1032spoIIB–mCherryΩkan, amyE::PspoIID–gfp–spoIIMΩcat, spoIIM::Tn917This study
KP1033amyE:::PspoIID–gfp–spoIIPΩcat, spoIIP::tet, spoIIB::erm,ΔspoIIQ::specThis study
KP1034amyE::PspoIID–gfp–spoIIPΩcat, spoIIP::tet, spoIIB::erm,ΔspoIIIAG-H::kanThis study
KP1035amyE::PspoIID–gfp–spoIIPΩcatThis study
KP1036amyE::PspoIID–gfp–spoIIPΩcat, spoIIB::ermThis study
KP1037amyE::PspoIID–gfp–spoIIPΩcat,ΔspoIIIAG-H::kanThis study
KP1038amyE::PspoIID–gfp–spoIIPΩcat,ΔspoIIQ::specThis study
KP1039amyE::PspoIID–gfp–spoIIPΩcat,ΔspoIVFAB::cat::tetThis study
KP1040amyE::PspoIID–gfp–spoIIPΩcat, spoIIB::erm,ΔspoIIIAG-H::kanThis study
KP1041amyE::PspoIID–gfp–spoIIPΩcat, spoIIB::erm,ΔspoIIQ::specThis study
KP1042amyE::PspoIID–gfp–spoIIPΩcat, spoIIB::erm,ΔspoIVFAB::cat::tetThis study
KP1043spoIIB–mCherryΩkan, ftsZ–gfpΩcatThis study
KP4156spoIIP-flagΩermThis study
KP1013ΔspoIVFAB::cat::tetThis study
KP1014ΔspoIVFAB::cat::tet, spoIIB::ermThis study

Construction of m-Cherry fusions

mCherry was PCR amplified from pRSETb–mCherry using the following primers: 5′-GGCGGCGGTACCGGATCTGGCGTGAGCAAGGGCGAGGAG-3′, KpnI site underlined, 5′-GGCGGCGGATCCTTACTTGTACAGCTCGTC-3′, BamHI site underlined. Fragments were digested with KpnI and BamHI, and ligated into KpnI- and BamHI-digested pEB71 (pUC derivative with loxP sites flanking a kanR cassette at the MCS) to form pJS7. The C-terminal spoIIB–mCherry fusion was constructed by PCR amplification of the last 500 base pairs of spoIIB using the following primers: 5′GGCGGCCATATGGATAAACAAACCTCAGG-3′, NdeI sited underlined, 5′-GGCGGCGGATCCTTACTTGTACAGCTCGTC-3′, BamHI site underlined. PCR fragments and pJS7 were digested with NdeI and BamHI and ligated together to create pJS8. pJS8 was sequenced, and then transformed into PY79, creating KP1026. The plasmid was able to integrate into the chromosome via a single recombination event at the native spoIIB locus. Derivatives of KP1026 were constructed by transforming KP1026 chromosomal DNA into various mutant backgrounds and selecting for kanamycin resistance.

Microscopy and image analysis

To visualize GFP, samples from sporulating bacteria were harvested at the appropriate time, and stained with a final concentration of 5 μg ml−1 FM 4-64 and 0.2 μg ml−1 4′, 6-diamidino-2-phenylindole (DAPI; Invitrogen) and applied to poly Lysine-treated coverslips (Pogliano et al., 1999). To visualize mCherry, samples were stained with Mitotracker Green (5 μg ml−1, Invitrogen) and DAPI (0.2 μg ml−1; Invitrogen). The membrane fusion assay was performed as described (Sharp and Pogliano, 1999), using 5 μg ml−1 FM 4-64, 5 μg ml−1 Mitotracker Green and 0.2 μg ml−1 DAPI. Images were collected using an Applied Precision Spectris optical sectioning microscope (described in Liu et al., 2006). DeltaVision software was used to collect and deconvolve the images. Following deconvolution, images from the medial focal planes were saved as TIFF files and imported into Adobe Photoshop. Pixel intensity plots were made using pixel data from deconvolved but unadjusted GFP images. Data were imported into Microsoft Excel and pixel intensity plots created essentially as described (Ramirez-Arcos et al., 2001; Jiang et al., 2005; Broder and Pogliano, 2006), but with the following modifications. To adjust the background levels in a uniform manner based on image properties, we calculated the average GFP fluorescence resulting from background and from cellular autofluorescence from a region of the image containing a vegetative cell that did not express GFP. This value, plus two standard deviations, was taken as the background level as this results in a 95% confidence that pixel intensity values above this level are true GFP signal. The calculated background fluorescence was subtracted from the pixel values, and all negative numbers set at zero. The remaining pixel intensity was broken down into five equal bins, with the maximum shown as red and the minimum as blue.

Antibody production

Polyclonal antibodies specific for SpoIID, SpoIIM and SpoIIIE were produced in rabbits by Antibodies Inc. SpoIID-specific antibodies were produced from the purified His6-tagged extracellular domain of SpoIID (Abanes-De Mello et al., 2002). SpoIIP-specific antibodies were raised against the His6-tagged, full-length protein purified under denaturing conditions (Abanes-De Mello et al., 2002). SpoIIIE-specific antibodies were produced against purified untagged cytoplasmic domain of SpoIIIE (amino acids 186–787) produced using the IMPACT-CM system of NEB. This domain was cloned into the plasmid pTYB1 at the NdeI and SapI sites, following PCR amplification of the corresponding region of spoIIIE with the primers EBP162 (AAACGAAGCTCTTCCGCAGCCTTTGTATAGTTCATCCATGCCATGTGTAAT) and EBP166 (ATAATCCATATGTCGCTGCAAGAAACGCTAAAAAAGTG). The restriction sites are underlined.

Co-immunoprecipitation and Western blotting

Sporulating cells were harvested, suspended in HMS buffer (20 mM HEPES-NaOH, 20 mM MgCl2 and 20% sucrose, pH 7.6) and treated with 1 mg ml−1 lysozyme at 37°C for 10 min. The spheroplasts were harvested by centrifugation and re-suspended in ice-cold buffer A (20 mM HEPES-NaOH, 150 mM NaCl and 1 mM EDTA, pH 7.6) with 1 mM PMSF, 1 μg ml−1 leupeptin and 1 μg ml−1 pepstatin and then treated with 0.5% digitonin (Calbiochem) on ice for 30 min. The insoluble fraction was removed by ultracentrifugation by Beckman TLA120-2d rotor (40 000 r.p.m., 30 min) and the supernatant was incubated with anti-flag M2 affinity gel (Sigma) overnight at 4°C with gentle rolling. The affinity gel was washed three times with buffer A containing 0.1% digitonin and the bound proteins eluted by SDS-loading buffer (without reducing agent) at 42°C. Proteins were analysed by Western blotting as described previously (Blaylock et al., 2004; Jiang et al., 2005). Dilutions of 1:5000 of anti-SpoIIP, anti-SpoIID and anti-SpoIIIE antisera were used to probe proteins. This method failed to detect an interaction between SpoIIB-FLAG and SpoIIP, SpoIID or SpoIVFA, or between SpoIVFA-FLAG and SpoIID or SpoIIP, or between SpoIIP-FLAG and SpoIVFA.

Immunofluorescence microscopy

Immunofluorescence microscopy was performed as in Perez et al. (2000) with the following modifications. Cells were grown in CH medium at 30°C to an OD600 of 0.5–0.6 and resuspended in A+B sporulation media. Culture samples of 0.5 ml were taken 2.0, 2.5 and 3.0 h after resuspension and fixed for 20 min in 32 mM NaPO4 plus 2.6% paraformaldehyde and 0.004% glutaraldehyde. Cells were washed in PBS, concentrated into 100 μl of GTE and placed on ice for no more than 1.5 h. Cells were permeabilized using a final concentration of 1 mg ml−1 lysozyme for 2 min and then washed with PBS. Anti-SpoIIP polyclonal antibody was applied at a dilution of 1:10 000 and left on overnight at 4°C. Cells were washed and incubated with anti-rabbit IgG conjugated to Oregon Green 488 (Molecular Probes) at a 1:200 dilution for 2 h in the dark. Cells were again washed and then stained with 20 μg ml−1 DAPI and 10 μg ml−1 FM 4-64 in SlowFade equilibration buffer (Molecular Probes) for 5 min.


We thank Eric Becker for preparing the SpoIIIE antibodies. We also thank Nathan Shaner and Roger Tsien for providing the pRSETb-mCherry vector. This research was supported by the National Institute of Health (GM57045).