Bacterial SPOR domains are recruited to septal peptidoglycan by binding to glycan strands that lack stem peptides

Significance

Cell division in bacteria is mediated by a large collection of proteins that assemble into a contractile ring at the division site. Understanding how these proteins are targeted to that site is important for understanding how division is spatially and temporally regulated. Here we demonstrate that a septal-targeting domain found in many bacterial cell-division proteins works by binding to a cell-wall structure that is present only transiently during cell division. These findings support a model for how different cell-wall hydrolases work together during separation of daughter cells and suggest a mechanism for coordinating synthesis and degradation of the cell wall during division.

Abstract

Bacterial SPOR domains bind peptidoglycan (PG) and are thought to target proteins to the cell division site by binding to “denuded” glycan strands that lack stem peptides, but uncertainties remain, in part because septal-specific binding has yet to be studied in a purified system. Here we show that fusions of GFP to SPOR domains from the Escherichia coli cell-division proteins DamX, DedD, FtsN, and RlpA all localize to septal regions of purified PG sacculi obtained from E. coli and Bacillus subtilis. Treatment of sacculi with an amidase that removes stem peptides enhanced SPOR domain binding, whereas treatment with a lytic transglycosylase that removes denuded glycans reduced SPOR domain binding. These findings demonstrate unequivocally that SPOR domains localize by binding to septal PG, that the physiologically relevant binding site is indeed a denuded glycan, and that denuded glycans are enriched in septal PG rather than distributed uniformly around the sacculus. Accumulation of denuded glycans in the septal PG of both E. coli and B. subtilis, organisms separated by 1 billion years of evolution, suggests that sequential removal of stem peptides followed by degradation of the glycan backbone is an ancient feature of PG turnover during bacterial cell division. Linking SPOR domain localization to the abundance of a structure (denuded glycans) present only transiently during biogenesis of septal PG provides a mechanism for coordinating the function of SPOR domain proteins with the progress of cell division.

A defining feature of bacteria is the peptidoglycan (PG) cell wall or “sacculus” that confers cell shape, protects the cell against lysis caused by turgor pressure, and is the ultimate target of many clinically relevant antibiotics, including β-lactams and vancomycin (1–4). The structure of PG is characterized by a network of glycan strands connected at regular intervals by oligopeptide cross-links (Fig. 1). The glycans are built from a repeating disaccharide of β-1,4–linked N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc), from which branch the oligopeptides that mediate cross-linking. During cell division, a new PG wall is assembled between the incipient daughter cells (5, 6), and subsequent cell separation depends on the activity of a collection of PG hydrolases that cleave various bonds in the PG mesh (7).

Fig. 1.

Model for SPOR domain localization to septal PG in E. coli. (Upper) The PG sacculus is shown with glycan strands running perpendicular to the long axis of the cell and joined by peptide cross-links depicted with arrows. Septal PG is indicated as a blue ring. (Lower) Recruitment of a prototypical SPOR domain protein (yellow; domains not to scale) to the midcell by binding to denuded glycans. PG is composed of GlcNAc (G) and MurNAc (M) amino sugars cross-linked by stem peptides (colored balls) that extend from the MurNAc residues. Processing of septal PG by cell-wall amidases during daughter-cell separation creates regions of denuded glycan backbone that serve as binding sites for SPOR domains. For simplicity a single SPOR domain is shown binding to a tetrasaccharide, but these details are not yet established. Denuded glycans are ultimately degraded by LTs, resulting in delocalization of SPOR domain proteins (not shown).

In the model Gram-negative bacterium Escherichia coli, daughter-cell separation is driven primarily by three cell-wall amidases (8–10). These enzymes cleave the amide bond that joins the lactyl group of the MurNAc to the α-amino group of the first amino acid (l-Ala) of the stem peptide, releasing as products free oligopeptides and “denuded” glycan strands (Fig. 1). Lytic transglycosylases (LTs) that cleave the MurNAc–GlcNAc glycosidic bond and endopeptidases that cleave stem peptides make minor but still significant contributions to daughter-cell separation (9, 11). In Bacillus subtilis, a model Gram-positive species, the enzymology of daughter-cell separation is not as well characterized. The available evidence indicates the process is driven primarily by an endopeptidase that cleaves within the stem peptide, with secondary contributors being an amidase that removes stem peptides and a glucosaminidase that degrades the glycan chain (12–14). Thus, the current view is that cell-wall amidases are more important for daughter-cell separation in E. coli than in B. subtilis. It is not known whether this difference reflects evolutionary happenstance or differences in the structure of the cell wall. Although the composition of the PG sacculus is virtually identical in these two organisms, there are profound differences in architecture (2). For example, the PG of E. coli is, for the most part, only one molecular layer thick, whereas B. subtilis has a multilayered PG sacculus that contains copious amounts of teichoic acids (1, 2).

Several years ago it became apparent that a widespread subset of bacterial cell-division proteins is targeted to the division site by a small PG-binding domain known as a “SPOR” domain (15–18). E. coli has four SPOR domain proteins, all of which are involved in cell division: DamX, DedD, FtsN, and RlpA. Of these, only FtsN is essential (19). SPOR domains are thought to direct proteins to the division site by binding to denuded glycans created by the cell-wall amidases that mediate daughter-cell separation. In support of this hypothesis, in a cosedimentation assay the periplasmic domain of FtsN (which includes the SPOR domain) binds much better to wild-type E. coli sacculi than to sacculi from a triple-amidase mutant (20). Moreover, the purified periplasmic domain of FtsN binds to long glycan strands (≥25 disaccharides) released by amidase digestion of PG (20). The length requirement suggests binding is cooperative or involves the formation of a higher-order structure of the SPOR domain with PG. In vivo, septal localization of all four E. coli SPOR domains requires both septal PG synthesis and amidase activity (16). Finally, amino acid substitutions in the SPOR domains from DamX and FtsN that impair septal localization in vivo also impair PG binding in vitro, indicating that PG binding is necessary for septal localization (21, 22).

Nevertheless, it is not known whether PG binding is sufficient for septal localization or whether denuded glycans are enriched in septal PG. Even the biochemical specificity of FtsN’s SPOR domain for long, denuded glycans is uncertain, because fragments of FtsN that contain the SPOR domain also bind muropeptides released by the digestion of PG sacculi with an LT (23); this observation suggests that the SPOR domain binds to short glycans that contain stem peptides. Here we use purified components and a microscopy-based assay to demonstrate that SPOR domains are targeted to the division site by binding to denuded glycans, which are indeed enriched in septal PG. Besides resolving central questions about SPOR domain function, these findings have implications for the regulation of bacterial cell division and for models of how different types of PG hydrolases work together to process septal PG.

Results

SPOR Domains Bind Septal Regions of Purified E. coli PG Sacculi.

Purified SPOR domains bind to purified PG sacculi in a cosedimentation assay, but whether this binding reflects the specific association of the SPOR domains with septal PG is not known (15, 17, 20, 21). To address this question, we purified hexahistidine (His₆)-tagged fusions of GFP to the SPOR domains from four E. coli cell-division proteins: DamX, DedD, FtsN, and RlpA. These constructs are referred to hereafter as “GFP-DamX^SPOR,” “GFP-DedD^SPOR,” “GFP-FtsN^SPOR,” and “GFP-RlpA^SPOR.” These various SPOR domains exhibit only ∼18% amino acid identity in pairwise alignments (17). Purified PG sacculi isolated from wild-type E. coli cells were immobilized by adhering them to glass slides coated with poly-l-lysine. After blocking with BSA, GFP-tagged SPOR domains were added and given 30 min to bind, at which time unbound protein was washed away and the sacculi were visualized by fluorescence microscopy (Fig. 2A). In these images, the sacculi appeared dark against a lighter background. Septal localization of GFP-SPOR constructs was readily apparent as a bright band of fluorescence across the middle of the sacculi. The fraction of sacculi exhibiting septal localization ranged from 36% for GFP-FtsN^SPOR to 55% for GFP-DedD^SPOR. Septal localization also was observed using an alternative assay format in which GFP-DamX^SPOR was added to a suspension of purified E. coli PG sacculi, and the assay mixtures were visualized directly (without any washing steps) by fluorescence and phase-contrast microscopy (Fig. S1A). When immobilized sacculi were labeled with anti-PG antibody, they appeared more uniformly fluorescent (Fig. S1B), as reported previously (10). We conclude that septal enrichment of GFP-SPOR constructs reflects binding to a septal PG structure rather than simply a greater abundance of PG in division septa than in the cell cylinder.

Fig. 2.

GFP-SPOR fusion proteins localize to septal PG in purified E. coli sacculi. (A) Representative fluorescence micrographs of the indicated GFP-tagged SPOR domains binding to immobilized sacculi from wild-type or mutant E. coli strains. All GFP-SPOR proteins were added at 250 nM, and sacculi were washed before imaging. Numbers in the top row show the percentage of wild-type sacculi exhibiting septal localization (mean ± SD from three experiments using independent sacculi preparations, with ≥200 sacculi scored in each case). (Scale bars, 5 μm.) Solid arrows indicate strong septal localization of GFP signal; dashed arrows indicate weak septal GFP signal; arrowheads indicate constrictions with little or no GFP signal. Sacculi were isolated from EC251 (WT), EC3708 (ΔLTs), EC3486 (ΔamiAC), and RP77 (ΔamiABC). (B) Quantification of GFP-SPOR binding to septal regions of purified PG sacculi from wild-type E. coli. Graphs pertain to the GFP-SPOR construct noted at top of the respective column of images and present composite data based on averaging the fluorescence intensity across 10 septa per strain. Data are representative of two experiments.

Fig. S1.

Characterization of GFP-SPOR and anti-PG antibody binding to purified PG sacculi. (A) GFP-DamX^SPOR exhibits septal localization when assayed with E. coli sacculi in suspension. GFP-DamX^SPOR was used at 50 nM and photographed under fluorescence (GFP filter) and phase-contrast microscopy without any washing steps to remove unbound or loosely bound proteins. Arrows point to an example of septal localization, which was observed in 42% of the sacculi (n = 201). (Scale bar, 5 μm.) (B) Anti-PG antibodies label sacculi uniformly. Immobilized PG sacculi from E. coli or B. subtilis were labeled with polyclonal anti-PG antiserum followed by goat anti-rabbit antibody conjugated to Alexa Fluor 488. Note that in this experiment we were unable to visualize the E. coli sacculi under phase-contrast microscopy, as is often the case. (C) Proteolytic removal of Lpp is not necessary for SPOR domain localization. E. coli sacculi purified by our standard procedure but without the protease digestion step were immobilized on poly-l-lysine–treated slides and used as binding substrates for the various GFP-SPOR constructs (250 nM in each case). The images shown are fluorescence micrographs (GFP filter). All experiments were performed at least twice.

Binding of SPOR Domains to Sacculi from PG Hydrolase Mutants.

The sacculi used in these experiments were treated with protease to remove Lpp (also called “Braun’s lipoprotein”) that is attached covalently to an amino acid in the stem peptides (1). However, we observed clear septal localization of GFP-SPOR constructs even when protease digestion was omitted (Fig. S1C). Thus, Lpp is not a SPOR domain target, nor does it block SPOR domain-binding sites.

To test the importance of denuded glycans as binding sites for SPOR domains, we isolated PG sacculi from E. coli mutants lacking either two (ΔamiAC) or all three (ΔamiABC) cell-wall amidases. As noted above, these enzymes remove stem peptides from glycan strands, so denuded glycans presumably are reduced in the ΔamiAC mutant and are absent in the ΔamiABC mutant. Notably, septal accumulation of the four SPOR domains was greatly reduced in sacculi from the double-amidase mutant and was not detected in sacculi from the triple-amidase mutant (Fig. 2). Septal localization of a Tat-targeted GFP-DamX^SPOR fusion protein also was amidase dependent in vivo (Fig. S2), as previously reported (16).

Fig. S2.

Effect of PG hydrolase mutations on localization of GFP-DamX^SPOR in vivo. The upper panel (“Sacculi”) is for comparison purposes and shows localization of GFP-DamX^SPOR (250 nM) on purified E. coli sacculi immobilized on poly-l-lysine–coated slides. Sacculi were washed before imaging. The lower panels (“Cells”) show live E. coli cells producing either Tat-targeted GFP-DamX^SPOR or Tat-targeted GFP. The latter serve as a negative control and indicate that the weak accumulation of GFP fluorescence in septal regions of the amidase mutants is likely the result of an enlarged periplasmic space rather than a specific association of GFP-DamX^SPOR with septal PG. By the same logic, the high-level accumulation of GFP-DamX^SPOR at septa of the ΔLTs mutant probably is caused by the accumulation of denuded glycans rather than by enlargement of the periplasmic space, because Tat-targeted GFP does not accumulate at these sites.

In many bacteria, including E. coli, the glycan backbone of PG is degraded by LTs that cleave the β-1,4 glycosidic bond via a mechanism that creates a 1,6-anhydroMurNAc end (7, 24, 25). Thus, denuded glycans generated by cell-wall amidases are expected to persist longer and accumulate to higher than normal levels in an LT mutant. We therefore constructed a mutant that lacks five of the seven known LTs (ΔmltADEC Δslt) (7, 9, 26), hereafter referred to as “ΔLTs.” It should be noted that our ΔLTs mutant also lacks rlpA, which encodes an LT in Pseudomonas aeruginosa (27), but so far, despite considerable effort, we have been unable to obtain any evidence indicating that RlpA from E. coli has LT activity. As reported previously for a similar mutant (9), our ΔLTs mutant has a moderate chaining phenotype, indicating that these enzymes contribute to the processing of septal PG in E. coli. Remarkably, sacculi from the ΔLTs mutant were exceptionally good binding substrates for GFP-tagged SPOR domains (Fig. 2). Quantification of septal fluorescence intensity indicated a 20- to 40-fold increase in SPOR domain accumulation compared with wild-type sacculi (based on the areas under the curves in Fig. 2B). Septal localization of Tat-targeted GFP-DamX^SPOR also was enhanced in the ΔLTs mutant in vivo (Fig. S2).

Binding of GFP-DamX^SPOR to Sacculi from B. subtilis.

Although cell-wall amidases do not drive daughter-cell separation in B. subtilis, we nevertheless assayed sacculi from this organism as binding substrates for purified GFP-tagged E. coli SPOR domains. Interestingly, we observed weak fluorescence throughout the sacculus and, in most cells, a more prominent band of fluorescence at putative division sites. This result is shown in Fig. 3 for GFP-DamX^SPOR in both assay formats in Fig. S3A and for all four SPOR domains using immobilized sacculi. Regardless of the SPOR domain, septal localization was generally cleaner using immobilized sacculi, presumably because that assay format includes washing steps. Of the various GFP-SPOR constructs tested, GFP-DamX^SPOR typically exhibited the best combination of bright septal signal with low general background binding throughout the cell cylinder. Some preparations of B. subtilis sacculi exhibited considerable polar localization (Fig. S4), especially when assayed in suspension. This phenomenon might reflect persistence of denuded glycans after cell separation.

Fig. 3.

Septal localization of GFP-DamX^SPOR binding to purified B. subtilis sacculi. (A) Representative micrograph of GFP-DamX^SPOR binding to immobilized PG sacculi isolated from wild-type B. subtilis strain PY79 grown at 30 °C. GFP-DamX^SPOR was tested at 100 nM, and sacculi were washed before imaging. The fraction of sacculi scoring positive for septal localization of GFP-DamX^SPOR (white arrow) was 71% (n = 136 sacculi). (B) Representative micrograph of GFP-DamX^SPOR binding to B. subtilis PG sacculi in suspension. GFP-DamX^SPOR was added at 10 nM, and sacculi were imaged directly, without any washing steps to remove loosely bound protein. Sacculi appear elongated because they were isolated from strain PY79 grown at 37 °C, conditions under which there is a mild chaining phenotype. White and orange arrows denote septa with and without GFP-DamX^SPOR, respectively. Septa that lack GFP signal presumably are older and no longer contain denuded glycans. (Scale bars, 5 μm in A and B.) (C) Representative micrograph from an assay in which GFP-DamX^SPOR(Q351K) mutant protein was tested at 1 μM for binding to sacculi isolated from strain PY79 grown at 37 °C. No examples of septal localization or even much fluorescence were observed (n = 100 sacculi). All experiments were performed three or more times. Note that purified B. subtilis sacculi contained phase-dark inclusions of unknown composition. These inclusions were visible under phase-contrast but not fluorescence microscopy and moved around inside the sacculi. They did not correspond to sites where GFP-DamX^SPOR domains localized.

Fig. S3.

GFP-SPOR fusion proteins localize to septal PG in purified B. subtilis sacculi. (A) Fluorescence micrographs of immobilized B. subtilis sacculi isolated from wild-type strain PY79 grown at 30 °C and assayed for binding with the indicated GFP-SPOR domain fusion proteins at 100 nM. Arrows indicate examples of septal localization. (Scale bar, 5 μm.) (B) Representative images from a competition assay using sacculi in suspension and 10 nM GFP-DamX^SPOR in the absence (−) or presence of a 100-fold molar excess of wild-type or mutant DamX^SPOR domain lacking a GFP tag, as indicated. Sacculi were imaged without washing to remove loosely bound protein. Images are representative of at least three experiments.

Fig. S4.

Septal-specific localization of GFP-DamX^SPOR does not require the inclusion of a blocking agent such as BSA to suppress nonspecific binding elsewhere on the sacculus. B. subtilis sacculi were assayed in suspension (A) or after immobilization (B) with the indicated concentrations of BSA or β-lactoglobulin. GFP-DamX^SPOR was used at 100 nM in this experiment. Note that considerable polar localization was observed using this preparation of sacculi, perhaps reflecting the persistence of residual denuded glycan after cell separation is complete. Images are representative of three experiments.

Several controls were performed to validate the use of B. subtilis sacculi. Because assay mixtures routinely include high levels of BSA, one might worry that BSA directs SPOR domains to septa by blocking alternative binding sites in the cell cylinder. However, this is not the case because similar results were obtained when BSA was used at lower concentrations, omitted entirely, or replaced with β-lactoglobulin (Fig. S4). As a further control, we ascertained that anti-PG antibody decorated B. subtilis sacculi relatively uniformly (Fig. S1B). This finding indicates that local differences in cell-wall thickness do not account for the strong accumulation of GFP-SPOR signal in septal regions. Finally, genetic evidence argues that GFP-DamX^SPOR binds similar sites on both E. coli and B. subtilis sacculi. This inference is based on the finding that an amino acid substitution (Q351K) known to impair septal localization and PG-binding in E. coli (21) also impairs binding to B. subtilis sacculi in our microscopy-based assay (Fig. 3C). Moreover, binding of GFP-DamX^SPOR was out-competed by excess wild-type DamX^SPOR protein lacking a GFP tag but not by a DamX^SPOR(Q351K) mutant derivative (Fig. S3B).

Binding of SPOR Domains to Sacculi After Treatment with PG Hydrolases in Vitro.

One potential complication with interpreting results obtained using sacculi isolated from mutants (Fig. 2) is that during the course of cell growth those PGs might have been modified in ways that cannot be predicted simply by knowing what genes were deleted. We therefore studied the binding of GFP-SPOR constructs to purified B. subtilis PG sacculi that had been treated in vitro with purified LT or amidase. For these assays we started with GFP-DamX^SPOR because it is comparatively well-behaved, and we used B. subtilis sacculi because they are much easier to observe under phase-contrast microscopy than the thinner sacculi from E. coli.

RlpA of P. aeruginosa (PaRlpA) is an LT with the unusual (but for our purposes advantageous) property of specifically degrading glycan chains that lack stem peptides (27). Exposure of sacculi to PaRlpA essentially abrogated binding of GFP-DamX^SPOR, even though there was no change in the appearance of these sacculi under phase-contrast microscopy (Fig. 4A). Because PaRlpA has a SPOR domain and localizes to division sites in vivo (16, 17, 27), it was necessary to verify that lack of GFP-DamX^SPOR binding did not result from competition for denuded glycans. We did so in two ways. (i) Treatment of sacculi with a catalytically defective PaRlpA variant that has a D157N amino acid substitution (27) did not impair binding by GFP-DamX^SPOR (Fig. 4A), even though the inactivated enzyme contained the SPOR domain. (ii) Conversely, treatment of sacculi with a catalytically active PaRlpA deletion derivative lacking the C-terminal SPOR domain (27) did prevent subsequent binding of GFP-DamX^SPOR (Fig. 4B). PaRlpA treatment prevented binding of GFP-DamX^SPOR not only at division sites but also throughout the cell cylinder, indicating that essentially all DamX^SPOR binding sites in B. subtilis PG sacculi, even those in the cell cylinder, involve denuded glycans. This finding implies that cell-wall amidases are active throughout the cell cylinder in B. subtilis.

Fig. 4.

The SPOR domain from DamX binds denuded glycan strands. (A) Immobilized PG sacculi from B. subtilis were treated with 2 mg/mL purified PaRlpA to remove denuded glycan strands or with a catalytically inactive variant [PaRlpA(D157N)]. Sacculi were washed after 10 or 60 min, as indicated, to terminate digestion of PG and then were tested as binding substrates for 100 nM GFP-DamX^SPOR. (B and C) Results from a similar assay in which sacculi were treated for 2 h with increasing concentrations of an RlpA derivative that lacks the SPOR domain [PaRlpA(ΔSPOR)] (B) or E. coli amidase D (EcAmiD) (C) to remove stem peptides. All assays included a no-enzyme control (−). Images are representative of three experiments.

AmiD of E. coli cleaves the amide bond between MurNAc and the stem peptide (28). Limited digestion of purified B. subtilis sacculi with purified AmiD had no obvious effect on the appearance of the sacculi under phase-contrast microscopy (Fig. 4C). Nevertheless, binding of GFP-DamX^SPOR increased dramatically and was no longer restricted to septal regions (Fig. 4C). Similar results were obtained using sacculi from E. coli (Fig. S5).

Fig. S5.

Amidase treatment enhances binding of GFP-DamX^SPOR to purified PG sacculi from E. coli. PG sacculi immobilized on a poly-l-lysine–coated glass slide were exposed to the indicated concentration of AmiD for 2 h and then were washed to terminate PG digestion. Subsequently, 100 nM GFP-DamX^SPOR was added and allowed to bind for 30 min. Sacculi were washed to remove loosely bound GFP-DamX^SPOR and photographed under fluorescence (GFP filter set). The exposure time was 4 s in all cases. Unadjusted images allow a direct comparison of the amount of GFP bound to the sacculi. Where indicated, brightness and contrast were adjusted uniquely for each image, better to reveal the distribution of the GFP tag on the sacculi. The experiment was repeated twice with similar results.

We next investigated the effect of treating sacculi with PaRlpA and AmiD on the subsequent binding of the other three E. coli GFP-SPOR constructs (Fig. S6). Overall, the results were similar to those for GFP-DamX^SPOR, although some modest differences were noted. Before enzyme treatment, both GFP-RlpA^SPOR and GFP-DedD^SPOR exhibited more binding to the cell cylinder than did GFP-DamX^SPOR and GFP-FtsN^SPOR. Moreover, although treatment of sacculi with PaRlpA essentially abrogated binding by GFP-DamX^SPOR and GFP-FtsN^SPOR, the binding by GFP-RlpA^SPOR and GFP-DedD^SPOR was reduced but not eliminated. Importantly, treating sacculi with AmiD to remove stem peptides resulted in a dramatic enhancement of GFP-SPOR binding in all cases. Collectively, these findings indicate that all the SPOR domains included in this study bind denuded glycans, but they may do so with different affinities or have somewhat different binding-site requirements.

Fig. S6.

The SPOR domains from FtsN, RlpA, and DedD bind denuded glycan strands. Immobilized PG sacculi from B. subtilis were incubated with buffer (−), PaRlpA(ΔSPOR), or AmiD and then were tested as binding substrates for GFP-FtsN^SPOR (500 nM), GFP-RlpA^SPOR (250 nM), or GFP-DedD^SPOR (250 nM). Images are representative of two experiments.

Discussion

Cell division in bacteria is mediated by a large assortment of proteins that assemble into a contractile ring structure at the division site (5, 6, 29, 30). Understanding how these proteins are targeted to that site is important for understanding how division is spatially and temporally regulated. The division apparatus is generally thought of as a multiprotein complex (or a collection of smaller complexes) whose assembly is driven, for the most part, by protein–protein interactions (23, 31–36). However proteins that contain a small PG-binding domain known as a “SPOR domain” are thought to be recruited by binding to septal PG (15–17). In particular, at least in E. coli, SPOR domains are hypothesized to bind to denuded glycan strands that accumulate in septal PG because of the intense activity of cell-wall amidases that drive the separation of daughter cells (16, 20).

Here we used a microscopy-based assay to demonstrate the septal localization of four E. coli SPOR domains in a purified system. This finding should dispel any lingering doubt about whether SPOR domain localization is driven by binding to something other than a PG structure, especially because any hypothetical contaminant in our purified sacculi would have to be enriched in septal regions from both E. coli and B. subtilis. We also provide convincing support for the proposal that the physiologically relevant binding site is a region of glycan strand that lacks stem peptides (16, 20). For example, binding of GFP-SPOR constructs was reduced when denuded glycans were removed by treating sacculi with PaRlpA. Conversely, elevating the content of denuded glycans by subjecting sacculi to limited digestion with AmiD increased GFP-SPOR binding.

Although it is clear that SPOR domains bind denuded glycans, many questions concerning the SPOR–PG interaction remain to be answered. What is the binding affinity (K_d) for this interaction? What chemical features of the denuded glycan are important for binding? Does binding require a long stretch of denuded glycan, as suggested for FtsN^SPOR (20)? Do SPOR domains bind as monomers or multimers? Is binding cooperative? The answers to these questions could be different for different SPOR domains, in part because SPOR domains are not well conserved at the level of amino acid sequence (17, 21). Further evidence for functional diversity comes from an in vivo study showing treatments that perturb the biogenesis of septal PG reduce septal localization of different SPOR domains to different degrees (16). The in vitro findings presented here also point in this direction. We observed small but reproducible differences in the fractional occupancy of septa and in the shapes of the fluorescence intensity profiles across septa (Fig. 2). We also observed that the SPOR domains from RlpA and DedD exhibited more binding to the cell cylinder and less sensitivity to removal of denuded glycans by PaRlpA than did the SPOR domains from DamX and FtsN (Fig. 4 and Figs. S3 and S6). One interesting idea is that PaRlpA might leave behind some short, denuded glycans that can serve as binding sites for the SPOR domains from RlpA and DedD but not from DamX or FtsN. However, less interesting possibilities cannot be ruled out, such as that the fraction of active molecules is different in different protein preparations. In any case, it will be interesting to learn the molecular basis for these differences and whether they are important for targeting different SPOR domain proteins to different sites.

GFP-DamX^SPOR might be considered a “stain” for denuded glycans. Our findings reveal this form of PG is not distributed uniformly in E. coli or B. subtilis but instead is enriched in division septa. The accumulation of denuded glycans in B. subtilis septal PG was somewhat unexpected, given the lack of evidence that amidases drive daughter-cell separation in that organism. Overall, our findings support a model for PG processing in which amidases remove stem peptides before LTs (or glucosaminidases, in the case of B. subtilis) degrade the glycan backbone (8, 9, 27, 37, 38). The reasons for this sequence of events are not immediately obvious because the great majority of these enzymes can degrade PG that contains stem peptides (24, 26). In E. coli, amidases and/or their activators localize sharply to the septal ring (39–41). It is not known whether E. coli LTs also accumulate at the midcell, but most are anchored to the outer membrane, and this anchoring presumably causes them to lag behind the amidases during cell division (7, 24). Whatever the underlying mechanism, sequential activity of amidases and enzymes that cleave the glycan backbone is likely to be an ancient and widespread feature of bacterial cell division because it pertains in both E. coli and B. subtilis, organisms that diverged at least 1 billion years ago.

What might be the benefit of using a SPOR domain rather than a protein–protein interaction to target a protein to the division site? On a conceptual level, SPOR domains sense the abundance of a transient intermediate (denuded glycans) in the division process and thereby enable the cell to regulate the function of SPOR domain proteins in response to the status (or progress) of septum assembly. In contrast, recruitment mechanisms based on protein–protein interactions seem better suited to ensuring that all the necessary components of the division apparatus are in place before it begins to function (41, 42). On a more practical level, it is readily apparent that the SPOR domain of PaRlpA brings the protein directly to the substrate it degrades (27). However, rationalizing why a protein such as FtsN should have a SPOR domain is more difficult. The primary function of FtsN is to stimulate synthesis of septal PG, so by equipping FtsN with a SPOR domain, evolution has made the seemingly illogical choice of rendering PG synthesis somewhat dependent on cell wall amidases that act later in the process of cytokinesis. As proposed by de Boer and colleagues, one benefit of making FtsN dependent on amidase activity is that this dependency creates a positive feedback loop that drives a burst of septal PG synthesis (16). We suggest an additional benefit is to help coordinate constriction of the outer membrane with inward growth of the PG layer. Hydrolysis of septal PG is required for proper invagination of the outer membrane, which in turn is required for resistance to bile salts and other noxious compounds (9, 43, 44). Tethering FtsN to denuded glycans could help prevent septal PG synthesis from running too far ahead of the amidases that are needed for maintaining outer membrane integrity during cytokinesis. In this scenario, removal of denuded glycans by LTs facilitates outer membrane constriction and promotes inward movement of FtsN and any other SPOR domain proteins that stimulate septum assembly.

Materials and Methods

SI Materials and Methods contains descriptions of bacterial strains, plasmids, strain constructions and growth conditions, and methods for the isolation of PG sacculi from E. coli and B. subtilis. It also contains procedures for purification of His₆-tagged GFP-SPOR proteins, for labeling sacculi with GFP-SPOR constructs and anti-PG antibodies, for localization of Tat-targeted GFP-DamX^SPOR fusions in live E. coli cells, and for microscopy. Finally, it includes a discussion of critical parameters for the assay used to visualize the binding of purified GFP-SPOR constructs to isolated PG sacculi. Bacterial strains, plasmids, and oligonucleotide primers are listed in Tables S1–S3, respectively.

Table S1.

Strains used in this study

Strain	Genotype or relevant features	Source or reference
B. subtilis
PY79	Wild-type	C. Ellermeier, The University of Iowa
E. coli strains for experiments
EC251	MG1655 (wild-type)	Lab collection
EC1898	EC251/pDSW997	(17)
EC2368	EC251/pDSW962	(50)
EC2691	RP77/pDSW997	This study
EC2723	RP77/pDSW962	This study
EC3486	EC251 ΔamiA<>frt ΔamiC<>frt	This study
EC3708	ΔrlpA<>frt ΔmltA<>frt Δslt<>frt ΔmltD<>frt ∆mltE<>frt ∆mltC<>frt	This study
EC3906	EC3486/pDSW997	This study
EC3907	EC3486/pDSW962	This study
EC3908	EC3708/pDSW997	This study
EC3909	EC3708/pDSW962	This study
RP77	CS109 amiB yje::res amiA::cam amiC::kan	(10)
E. coli donors for strain construction by P1 transduction
JW2428-1	BW25113 ΔamiA764::kan	(51)
JW2542-6	BW25113 ΔmltF732::kan	(51)
JW2671-1	BW25113 ΔmltB776::kan	(51)
JW2784-1	BW25113 ΔmltA741::kan	(51)
JW5018-1	BW25113 ΔmltD730::kan	(51)
JW5449-1	BW25113 ΔamiC742::kan	(51)
JW5481-1	BW25113 ΔmltC738::kan	(51)
JW5821-1	BW25113 ΔemtA783::kan (emtA is also called mltE)	(51)
E. coli plasmid hosts for cloning or protein overproduction
BL21	fhuA2 [lon] ompT gal [dcm] ΔhsdS	New England Biolabs
BL21(DE3)	λ(DE3) fhuA2 [lon] ompT gal [dcm] ΔhsdS	(28)
Shuffle T7	F′ lac, pro, lacIQ/Δ(ara-leu)7697 araD139 fhuA2 lacZ::T7 gene1 Δ(phoA)PvuII phoR ahpC* galE (or U) galK λatt::pNEB3-r1-cDsbC (SpecR, lacIq) ΔtrxB rpsL150(StrR) Δgor Δ(malF)3	New England Biolabs
XL1-Blue	F′ proAB lacIq lacZΔM15 Tn10 Tetr/recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 lac	Stratagene
Protein overproduction strains
EC1903	BL21/pDSW1000 (for His6-DamXSPOR)	(17)
EC2068	BL21/pDSW1086 (for His6-DamXSPOR(Q351K))	(21)
EC2249	BL21(DE3)/pDSW1171 (for His6-GFP-DamXSPOR)	This study
EC2251	BL21(DE3)/pDSW1171 (for His6-GFP-DedDSPOR)	This study
EC2292	BL21(DE3)/pET-28a-amiD (for His6-AmiD)	(28)
EC3087	BL21/pDSW1557 (for His6-PaRlpA)	(27)
EC3204	BL21/pDSW1600 (for His6-PaRlpA(D157N))	(27)
EC3387	BL21/pDSW1650 (for His6-PaRlpA(ΔSPOR)-mCherry)	This study
EC3476	Shuffle T7/pDSW1666 (for His6-GFP-FtsNSPOR)	This study
EC3477	BL21/pDSW1667 (for His6-GFP-RlpASPOR)	This study
EC3479	BL21/pDSW1669 (for His6-GFP-DamXSPOR(Q351K))	This study

Table S3.

Oligonucleotide primers used in this study

Name	Purpose	Sequence (5′-3′)
P761	Cloning gfp-ftsNSPOR into pQE-80L	CCCAAGCTTCAACCCCCGGCGGCGAGCCG
P1126	Cloning gfp-dedDSPOR into pQE-80L	GCCAAGCTTTAATTCGGCGTATAGCCCATT
P1128	Cloning gfp-rlpASPOR into pQE-80L	GCCAAGCTTTACTGCGCGGTAGTAATAAAT
P1137	Cloning gfp-damXSPOR into pQE-80L	CTGAAGCTTCTTCAGATCGGCCTGTACCT
P1403	Cloning gfp-SPOR fusions into pQE-80L	CACAGATCTAACAATAACGATCTCTTTCAG
P1787	Cloning P. aeruginosa rlpA(ΔSPOR)-mCherry into pQE-80L	CATTGATCATCCAGCAAGGCGCCCCAGCAG
P1900	Cloning P. aeruginosa rlpA(ΔSPOR)-mCherry into pQE-80L	CTGAAGCTTTTACTTGTACAGCTCGTCCATG
P1926	Colony PCR of amiA locus	TGGCGTGAACGGTCGAATTAG
P1927	Colony PCR of amiA locus	CATTACGCAACACCCGACTAC
P1930	Colony PCR of amiC locus	CGCCGAGCATGATGACAAACG
P1931	Colony PCR of amiC locus	ATGCCAAATCCGTGATCGGGG
P1935	Colony PCR of mltA locus	TTGTGCAAAATTTGCGTCAGG
P1936	Colony PCR of mltA locus	GCGCACTTAGTCAATAATCAGG
P1941	Colony PCR of mltD locus	ATTGCAACAACCTGAAGAGCG
P1942	Colony PCR of mltD locus	TCACCAGTGATATCATCATGG
P1943	Colony PCR of mltE locus	CCTCGACGTTATTGACTTGATG
P1944	Colony PCR of mltE locus	TTCTTAACGTCAGCCCGACGG
P1947	Colony PCR of slt locus	AACGGCAATGACTGGTTAGC
P1948	Colony PCR of slt locus	GCCATAATATGTCGCCATTG
P1949	Colony PCR of mltC locus	TGTCGTCATTCACCGGCTGC
P1950	Colony PCR of mltC locus	TAACGGCGAGATTCCGCAGAG

Restriction sites are underlined.

SI Materials and Methods

General Procedures.

Bacterial strains, plasmids, and oligonucleotide primers are listed in Tables S1, S2, and S3, respectively. Bacteria were grown in LB medium containing antibiotics as appropriate: kanamycin at 40 µg/mL, ampicillin at 200 µg/mL, and chloramphenicol at 30 µg/mL. PCR used Phusion DNA polymerase (New England Biolabs). Oligonucleotide primers were purchased from Integrated DNA Technologies. α-Chymotrypsin (catalog no. C4129), BSA (BSA Fraction V, catalog no. A7906), β-lactoglobulin (catalog no. L7880), DNase I (catalog no. DN25), and RNase A (catalog no. R503) were from Sigma Aldrich. l-(tosylamido-2-phenyl) ethyl chloromethyl ketone (TPCK)-treated trypsin (catalog no. 3740) was purchased from Worthington Biochemical Corp. Hydrofluoric acid (48–51%; catalog no. 7664-39-3) was obtained from VWR International. Autoclaved deionized H₂O was used during PG purifications.

Table S2.

Plasmids used in this study

Plasmid	Relevant features	Source or reference
pCP20	λPR::FLP λcI857 bla cat RepTS (pSC101 derivative)	(52)
pDSW962	P204::sstorA-gfp lacIq bla pBR ori	(50)
pDSW992*	P204::TTgfp-ftsNSPOR(amino acids 240–319)	(17)
pDSW994*	P204::TTgfp-dedDSPOR(141–220)	(17)
pDSW995*	P204::TTgfp-rlpASPOR(283-362)	(17)
pDSW997*	P204::TTgfp-damXSPOR(338–428)	(17)
pDSW1000†	PT5::His6-damXSPOR(338–428)	(17)
pDSW1089*	P204::TTgfp-damXSPOR(338-428; Q351K)	(21)
pDSW1171†	PT5::His6-gfp-damXSPOR(338–428)	This work
pDSW1173†	PT5::His6-gfp-dedDSPOR(141–220)	This work
pDSW1504‡	pEXG2::‘ rlpA(ΔSPOR)Pa-mCherry dacC’	(27)
pDSW1557†	PT5::His6-rlpA (28-341)Pa	(27)
pDSW1600†	PT5::His6-rlpA(28-341; D157N)Pa	(27)
pDSW1650†	PT5::His6-rlpA(ΔSPOR)Pa-mCherry	This work
pDSW1666†	PT5::His6-gfp-ftsNSPOR(240–319)	This work
pDSW1667†	PT5::His6-gfp-rlpASPOR(283–362)	This work
pDSW1669†	PT5::His6-gfp-damXSPOR(338-428; Q351K)	This work
pET28a-amiD	PT7::His6-amiD lacIq kan pBR ori	(28)

^*These plasmids are derivatives of pDSW962.

^†These plasmids are derivatives of the Qiagen protein overproduction vector pQE-80L (Engineered P_T5 promoter with Lac repressor binding sites, N-terminal His₆ tag, lacI^q, bla, pBR ori).

^‡ΔSPOR refers to a deletion encompassing codons 269–341 of P. aeruginosa rlpA.

Strain Construction.

The double amidase mutant EC3486 and the ΔLTs strain EC3708 were constructed by a series of P1 transductions using various Keio collection donors listed in Table S1. The initial recipient was our laboratory isolate of MG1655, designated “EC251.” After transduction to kanamycin resistance, acquisition of the expected gene replacement was confirmed by colony PCR using primers that flank the gene in question (Table S3). Then the Kan^R element was evicted using FLP recombinase produced from pCP20. Excision of the Kan^R element was confirmed by colony PCR. This process was continued until all the desired mutations had been introduced.

Plasmid Construction.

Plasmids are listed in Table S2. Plasmids for overproduction of His₆-GFP–tagged SPOR domains were derivatives of pQE-80L and were constructed as follows. pDSW1171 [P_T5::His₆-gfp-damX(338–428)] was constructed by amplifying a portion of the torA signal sequence, gfp, and the coding region for the SPOR domain from damX (codons 338–428) from pDSW997 with primers P1403 and P1137. The resulting 1,153-bp product was digested with BglII and HindIII and then was ligated to pQE-80L that had been cut with BamHI and HindIII. pDSW1669 [P_T5::His₆-gfp-damX(338–428) carrying a Q351K substitution] was constructed similarly using pDSW1089 as the template. pDSW1173 [P_T5::His₆-gfp-dedD(141–220)] was constructed similarly using primers P1403 and P1126 with pDSW994 as the template. pDSW1666 [P_T5::His₆-gfp-ftsN(240–319)] was constructed similarly using primers P1403 and P761 with pDSW992 as the template. pDSW1667 [P_T5:: His₆-gfp-rlpA(283–362)] was constructed similarly using P1403 and P1128 with pDSW995 as the template. All constructs were verified by DNA sequencing. Plasmid pDSW1650 for overproduction of PaRlpA lacking a SPOR domain was constructed by amplifying rlpA codons 28–268 by PCR using primers P1787 and P1900 with pDSW1504 as template. The product was cut with BclI and HindIII and ligated to pQE-80L that had been cut with BamHI and HindIII.

Protein Purification.

All purified proteins used in this study carried His₆ tags and were overproduced from plasmids in E. coli hosts. BL21 was the overproduction host for GFP-DamX^SPOR, GFP-DamX^SPOR(Q351K), GFP-DedD^SPOR, GFP-FtsN^SPOR, GFP-RlpA^SPOR, DamX^SPOR, DamX^SPOR(Q351K), PaRlpA, PaRlpA(D157N), and PaRlpA(ΔSPOR). Shuffle T7 was the host for GFP-FtsN^SPOR. BL21(DE3) was the host for AmiD. For a typical purification, 1 L of cultures was grown in LB medium at 30 °C to OD₆₀₀ = 0.5 and induced by the addition of isopropyl-β-d-thiogalactopyranoside to 1 mM for 2–3 h. Cells were harvested and broken by sonication, and the His₆-tagged protein was purified by cobalt-affinity chromatography as described (21). Protein concentration was determined by A₂₈₀ and bicinchoninic acid assay with BSA as standard; these two methods produced values in close agreement. Purity was judged to be ∼95% by Coomassie staining after SDS-PAGE.

Isolation of PG Sacculi from E. coli.

The following procedure was derived from published protocols (45, 46). An overnight culture of the desired strain was diluted 1:500 into 100 mL of fresh LB medium in a 500-mL flask and grown in a New Brunswick C24 incubator-shaker at 30 °C and 210 rpm to OD₆₀₀ = 0.5 (∼10¹⁰ cells). Flasks were transferred to an ice-water bath to chill the cells, which then were collected by centrifugation for 10 min at 8,000 × g and 4 °C. The cell pellet was suspended in 3 mL of 1% NaCl and added dropwise to 6 mL of 8% SDS in a 25-mL Corex screw-cap centrifuge tube equipped with a stirring bar and sitting in a boiling water bath. The tube then was capped loosely, and samples were boiled for 5 h with gentle stirring. Water was added as needed to maintain volume. Samples then were incubated overnight at room temperature with continued stirring. The next morning, samples were boiled again for 2 h, allowed to cool to about 40 °C, and then divided among several thick-walled microfuge tubes (Beckman; catalog no. 357448). Sacculi were collected using a tabletop ultracentrifuge (Beckman Coulter OptimaTM Ultracentrifuge, TLA55 rotor; 50,000 rpm) by centrifugation at 100,000 × g for 10 min at 37 °C. Pellets were resuspended and pooled into a final 3-mL volume of 4% SDS in a 25-mL Corex tube. Samples were boiled again for 2 h with gentle stirring. Sacculi were collected as above and washed five times with autoclaved deionized H₂O to remove SDS. The pellets were resuspended in 1 mL PBS (137 mM NaCl, 3 mM KCl, 9 mM NaH₂PO₄, and 2 mM KH₂PO₄, pH 7.4) in a standard 1.5-mL microfuge tube. Then α-chymotrypsin was added to 200 µg/mL. Samples were incubated for 2 h at 37 °C; then a second dose of α-chymotrypsin (200 µg/mL) was added, and incubation was continued overnight. Next morning, one-tenth volume of 10% SDS solution was added, and the microfuge tubes were sealed using lid locks, and placed in a 100 °C heat block for 2 h. Samples were transferred into thick-walled microfuge tubes, and sacculi were collected using a tabletop ultracentrifuge as described above. Samples were washed with autoclaved deionized H₂O at least six times until they were free of SDS as determined using a methylene blue test (47). The final pellet was resuspended into 400 µL of water, stored at 4 °C, and used within 3 wk. For purposes of assaying binding, we assumed no loss of sacculi during the course of purification, so the final suspension is considered to contain 10¹⁰ sacculi in 400 μL.

Isolation of PG Sacculi from B. subtilis.

The purification procedure is adapted from David Popham’s laboratory protocol and has been described (48). An overnight culture of B. subtilis PY79 was diluted 1:200 into 100 mL of LB medium in a 500-mL flask and grown at 30 or 37 °C and 210 rpm to OD₆₀₀ = 0.5. The flask was placed into an ice bath for 5 min to chill cells, which then were collected by centrifugation for 10 min at 8,000 × g and 4 °C. The cell pellet was suspended in 2 mL of cold H₂O and was dripped over a 2-min period into 50 mL of boiling 4% SDS in a 250-mL beaker equipped with a stirring bar and placed on a magnetic stirrer/hot plate. Cells were boiled for 30 min with gentle stirring, and the volume was maintained at ∼50 mL by adding additional water as needed. The sample then was allowed to sit until it cooled to about 40 °C. The sample was divided into two 50-mL conical centrifuge tubes and centrifuged for 10 min at 8,000 × g to collect sacculi. The pellets were resuspended in 20 mL of warm water (∼60 °C), combined in one tube, and centrifuged as above. The pellet was resuspended and washed by centrifugation at least five more times until SDS was undetectable by a methylene blue test (47). The pellet was resuspended into 800 µL of H₂O and transferred into a 1.5-mL microfuge tube before stock solutions of the following reagents were added to achieve a final volume of 1 mL containing 100 mM Tris⋅HCl (pH 7.5), 20 mM MgSO₄, 10 µg/mL DNase I, and 50 µg/mL RNase A. The sample was incubated at 37 °C for 2 h. Then 100 µg of TPCK-treated trypsin and 10 µL of 1 M CaCl₂ were added, and the sample was incubated at 37 °C overnight. The next morning, sacculi were collected using a microcentrifuge (10 min at 13,000 × g), and the pellet was washed twice with 1 mL of H₂O. The sample was resuspended into 1 mL of H₂O and transferred to a 1-mL plastic microfuge tube equipped with a screw cap, and sacculi were collected by pelleting in a microfuge as above. To cleave off teichoic acids, the pellet was resuspended in 1 mL of 49% hydrofluoric acid and incubated at 4 °C for 48 h with gentle mixing on a rocker. The sample was centrifuged at 13,000 × g for 10 min. The pellet was resuspended in 1 mL of H₂O and washed by centrifugation five more times to remove residual acid. The final pellet was suspended in 400 µL of water and stored at 4 °C and used within 3 wk.

Localization of GFP-SPOR Proteins in Purified PG Sacculi.

PG sacculi in suspension.

Assay mixtures (40 μL) contained 25 mM sodium phosphate (pH 7.5), 200 mM NaCl, 1 mg/mL BSA, ∼10⁷ sacculi, and variable amounts of GFP-SPOR domain protein as indicated in the figures and figure legends. The reaction mixtures were incubated on ice for 30 min; then 3 μL were spotted onto a glass slide and examined immediately by phase-contrast and fluorescence microscopy.

Immobilized PG sacculi.

The assay for the binding of GFP-tagged SPOR domains to immobilized PG sacculi was adapted from a published procedure for localization of proteins by immunofluorescence microscopy in fixed cells adhered to poly-l-lysine–treated multiwell slides (49). To each well we added 10 µL containing ∼10⁷ purified sacculi in water. Sacculi were allowed to adhere for 10 min; then the suspension was removed by aspiration, and wells were washed three times with 10 μL of PBS to remove any nonadherent sacculi. Wells were allowed to air dry for 5 min. Then 10 μL of 20 mg/mL BSA in PBS (BSA/PBS) was added to rehydrate the samples and block any free poly-l-lysine–coated surfaces. After 20 min, wells were washed once with PBS. Then we added 10 μL of GFP-SPOR protein in BSA/PBS, and incubation was continued for 30 min. Unbound protein was removed by washing the wells five times with 10 μL of PBS. Finally, 3 μL of PBS was added per well, a coverslip was mounted, and samples were examined immediately under the microscope.

Treatment of Sacculi with PaRlpA or AmiD.

The effect of treating sacculi with PG hydrolases was determined using immobilized B. subtilis sacculi. Briefly, 10 μL of enzyme in BSA/PBS were added after the blocking step and incubated at 37 °C in a covered Petri dish with droplets of water to minimize evaporation. Enzymes were removed by washing the wells five times with 10 μL of PBS, at which point 10 μL of GFP-DamX^SPOR protein in BSA/PBS was added. Subsequent procedures were as described above.

Competition Assay.

The competition assay was conducted using B. subtilis sacculi in suspension as described above, except the unlabeled competitor [either DamX^SPOR or DamX^SPOR(Q351K)] was added to a final concentration of 1 µM and allowed to bind for 30 min on ice. Then GFP-DamX^SPOR was added to a final concentration of 10 µM, and incubation was continued on ice for an additional 30 min. Finally, 3 µL of the assay mixture was spotted onto a glass slide and examined immediately by fluorescence and phase-contrast microscopy.

Critical Parameters of the GFP-SPOR Localization Assay.

Initially we attempted to visualize GFP-DamX^SPOR binding to purified sacculi from E. coli in assay mixtures essentially identical to those used previously in cosedimentation assays (17, 20). We immediately encountered several difficulties. Although we observed rare instances of what appeared to be isolated rod-shaped sacculi with a green fluorescent band across the middle, for the most part the sacculi appeared as amorphous conglomerates that exhibited high levels of apparently nonspecific green fluorescence. Inclusion of BSA did not suppress the apparent nonspecific binding, nor did washing the sacculi by centrifugation and resuspension in fresh buffer. However, by examining sacculi under the microscope at each step of the purification procedure, we discovered they tended to lose rod morphology and aggregate after treatment with amylase. Thus, omitting amylase digestion yielded well-dispersed, rod-shaped sacculi to which GFP-DamX^SPOR bound predominately at the septal regions. In retrospect, we noted that several publications that have depicted fluorescently labeled E. coli sacculi also omitted the amylase-digestion step (10, 45). We do not know whether this similarity in methods is a coincidence or whether the issues related to amylase treatment have been encountered previously.

The fact that the sacculi used in our studies were not treated with amylase raises the question of whether GFP-SPOR domains are binding to some carbohydrate other than PG. We considered this possibility unlikely for several reasons. (i) That carbohydrate would have to be enriched in division septa of both E. coli and B. subtilis. (ii) That carbohydrate would have to be removed when sacculi are treated with PaRlpA and somehow created or exposed when sacculi are treated with AmiD. Indeed, in the case of sacculi from an E. coli amiABC mutant, that carbohydrate would have to be essentially nonexistent until the sacculi are treated in vitro with AmiD, at which point it would have to be generated in copious amounts. (iii) Amylase digestion appeared to increase binding of GFP-DamX^SPOR rather than remove binding sites, a result that is the opposite of what one would expect if SPOR domains bind to an amylase-sensitive target.

At this point we could convince ourselves that GFP-DamX^SPOR localized to septal PG in purified E. coli sacculi, but there still were issues with the assay. Because we often could not observe the sacculi under phase-contrast microscopy, we had to focus on them under fluorescence, during which time the GFP tended to bleach. Moreover, because the sacculi were in suspension, they tended to drift out of the focal plane, further complicating efforts to get both phase-contrast and fluorescence images of the same field of view. The issue with drift was solved by immobilizing the sacculi on poly-l-lysine–treated slides. The issue with transparency under phase-contrast microscopy was solved by switching to using sacculi from B. subtilis. One parameter of the assay that is not so important is the high levels of BSA that are included in the standard assay mixtures. Similar results are obtained if BSA is omitted or replaced with a low concentration of β-lactoglobulin (Fig. S4). Nevertheless, we recommend the inclusion of BSA to stabilize purified GFP-SPOR proteins against unfolding upon dilution.

In retrospect, it is fortunate that we started with GFP-DamX^SPOR, which turned out to exhibit consistent, bright, and specific localization to septal PG. When we started working with the other E. coli SPOR domains, we found that GFP-FtsN^SPOR also is highly specific for septal PG, but its fluorescence signals are not as bright. This reduced fluorescence could mean that binding sites for GFP-FtsN^SPOR are less abundant than the binding sites for GFP-DamX^SPOR. Alternatively, the SPOR domain from FtsN has an essential disulfide bond (22), so a significant fraction of the purified protein in our preparations might be inactive. Both GFP-RlpA^SPOR and GFP-DedD^SPOR are bright enough, but they exhibit higher levels of binding throughout the cell cylinder than does GFP-DamX^SPOR, especially when assayed with B. subtilis sacculi. We suspect that these differences reflect different binding-site preferences, but issues related to protein folding cannot be ruled out. In vivo localization studies indicate that RlpA is found not only at the division site but also in foci scattered throughout the cell cylinder in E. coli and P. aeruginosa (16, 27).

Labeling of PG Sacculi with Anti-PG Antibodies.

To label PG sacculi with anti-PG antibodies, we adapted a published procedure (10, 46). Immobilization of sacculi on poly-l-lysine–treated multiwell slides and blocking with BSA/PBS were performed as described in the procedure for localization of GFP-SPOR domain constructs on purified sacculi. Polyclonal rabbit anti-PG antiserum was diluted 1:200 in BSA/PBS, and 10 µL were added per well. After 1 h was allowed for antibodies to bind, samples were washed 10 times with 10 μL of PBS. Then 10 μL of a 1:250 dilution of Alexa Fluor 488 goat anti-rabbit IgG (in BSA/PBS) were added per well, and incubation was continued for 1 h. After wells were washed 10 times with 10 μL of PBS, samples were mounted in Vectashield (Vector Laboratories) and visualized by fluorescence (GFP filter set) and phase-contrast microscopy. Both the anti-PG antiserum and the Alexa Fluor 488 secondary antibody were a gift from K. Young, University of Arkansas for Medical Sciences, Little Rock, AR, but the former originated with M. de Pedro, University of Madrid, Madrid, and the latter was purchased from Invitrogen-Molecular Probes.

Localization of ^TTGFP-DamX^SPOR Fusions in Live Cells.

Transformants of EC251, EC3486, RP77, or EC3708 harboring pDSW962 or pDSW997 were grown overnight at 30 °C in LB medium with 200 μg ampicillin/mL. The next morning cultures were diluted 1:200 and grown under the same conditions to OD₆₀₀ = 0.5. Live cells were immobilized on agarose pads and imaged under phase-contrast and fluorescence (GFP filter set) microscopy as described (17).

Microscopy, Image Analysis, and Preparation of Figures.

Phase-contrast and fluorescence micrographs were recorded on an Olympus BX60 microscope equipped with a 100 UPlanApo objective. Images were captured using a Spot 2-cooled charge-coupled device camera (Diagnostic Instruments), a Uniblitz shutter, and a personal computer with Image-Pro software version 4.1 (Media Cybernetics). The GFP filter set was from Chroma Technology Corp (catalog no. 41017) and comprised a 450- to 490-nm excitation filter, a 495-nm dichroic mirror (long pass), and a 500- to 550-nm emission filter. Photograph exposure times were always 0.08 s for phase contrast and 4 s for fluorescence. Fluorescence intensity was quantified using the Line Profile measurement tool in Image-Pro. Levels, brightness, and contrast were used to adjust images for legibility. Figures were cropped and assembled in Adobe Photoshop and Adobe Illustrator.