Essential PcsB putative peptidoglycan hydrolase interacts with the essential FtsX_Spn cell division protein in Streptococcus pneumoniae D39

Abstract

The connection between peptidoglycan remodeling and cell division is poorly understood in ellipsoid-shaped ovococcus bacteria, such as the human respiratory pathogen Streptococcus pneumoniae. In S. pneumoniae, peptidoglycan homeostasis and stress are regulated by the WalRK (VicRK) two-component regulatory system, which positively regulates expression of the essential PcsB cysteine- and histidine-dependent aminohydrolases/peptidases (CHAP)-domain protein. CHAP-domain proteins usually act as peptidoglycan hydrolases, but purified PcsB lacks detectable enzymatic activity. To explore the functions of PcsB, its subcellular localization was determined. Fractionation experiments showed that cell-bound PcsB was located through hydrophobic interactions on the external membrane surface of pneumococcal cells. Immunofluorescent microscopy localized PcsB mainly to the septa and equators of dividing cells. Chemical cross-linking combined with immunoprecipitation showed that PcsB interacts with the cell division complex formed by membrane-bound FtsX_Spn and cytoplasmic FtsE_Spn ATPase, which structurally resemble an ABC transporter. Far Western blotting showed that this interaction was likely through the large extracellular loop of FtsX_Spn and the amino terminal coiled-coil domain of PcsB. Unlike in Bacillus subtilis and Escherichia coli, we show that FtsX_Spn and FtsE_Spn are essential in S. pneumoniae. Consistent with an interaction between PcsB and FtsX_Spn, cells depleted of PcsB or FtsX_Spn had strikingly similar defects in cell division, and depletion of FtsX_Spn caused mislocalization of PcsB but not the FtsZ_Spn early-division protein. A model is presented in which the interaction of the FtsEX_Spn complex with PcsB activates its peptidoglycan hydrolysis activity and couples peptidoglycan remodeling to pneumococcal cell division.

Ovococcus bacteria are ellipsoid-shaped like American footballs and include Streptococcus and Lactococcus species that are important as mammalian pathogens and in industry (1–3). Of this group, S. pneumoniae is a serious human pathogen that causes pneumonia, otitis media, meningitis, and septicemia, and it annually kills more than 1.6 million people worldwide (4–6). The compositions of the supramolecular machineries that mediate peripheral and septal peptidoglycan (PG) synthesis in ovococci are only starting to emerge (7–11). However, relatively little is known about the roles of PG hydrolases in cell division as opposed to autolysis or competence-dependent fratricide (12–15). The PcsB protein in S. pneumoniae has been a leading candidate for a PG hydrolase involved in cell division (12, 13, 16–19).

PcsB emerged as a candidate PG hydrolase in cell division, because it is essential in serotype 2 strain D39 and possibly other serotypes of S. pneumoniae (Discussion) (16, 18, 19). Consistent with its essential role, the amino acid sequence of PcsB is highly conserved in all serotypes of S. pneumoniae, much more so than the amino acid sequences of surface virulence factors (17). The pcsB gene is the only essential member of the regulon controlled by the WalRK_Spn (VicRK) two-component system (Fig. 1). The WalRK_Spn regulon also includes genes encoding other nonessential PG hydrolases (12, 19, 20). Depletion of PcsB or WalRK_Spn leads to an arrest of cell growth and the development of misshapen cells with severe division and PG biosynthesis defects (16, 18, 19). Consistent with this regulatory linkage, constitutive expression of pcsB from a synthetic promoter uncouples pcsB expression from the WalRK_Spn two-component system and renders the WalR_Spn (VicR) response regulator nonessential (19).

Fig. 1.

Schematic diagrams of the pneumococcal pcsB operon and the PcsB protein. pcsB is a single gene operon between mreCD and rpsB that is regulated positively by the WalR_Spn response regulator, which binds upstream of the P_pcsB promoter (8). Putative transcription terminators (lollipops) and the insertion point that creates the PcsB-l-FLAG³ fusion protein and adds the selectable P_c-erm marker are indicated. The PcsB polypeptide contains the following segments: a signal peptide that is cleaved off during export (16), a coiled-coil domain containing two putative leucine zipper motifs, an alanine-rich linker, and a CHAP domain containing a characteristic catalytic triad of 3 aa (Cys292-His343-N366). Simultaneous L78S and L219P amino acid changes in the putative leucine zipper motifs of the coiled-coil domain cause a temperature-sensitive phenotype. Additional details in the text.

The domain structure of PcsB_Spn is also suggestive of a PG hydrolase. The PcsB_Spn polypeptide consists of four segments: a conventional signal peptide domain that is cleaved off during secretion (16), an extended coiled-coil domain containing putative leucine zipper motifs, a variable alanine-rich linker region, and a putative cysteine- and histidine-dependent aminohydrolases/peptidases (CHAP) domain (Fig. 1). CHAP domains contain a cysteine–histidine–asparagine triad in their active sites that resembles the fold of papain proteases (21–23). In bacteriophage lysins, CHAP domains function as PG amidases and endopeptidases (24–26). By comparison, the functions of only a couple bacterial CHAP domain proteins have been characterized (16, 27–29). Like PcsB homologs in Streptococcus species, the Sle1 CHAP-domain protein of Staphylococcus aureus seems to play fundamental roles in cell division (28). KO mutants of sle1 have been reported that are severely impaired for cell division and growth (28). Purified Sle1 protein lacking its signal peptide and the Sle1 CHAP domain alone both exhibited robust hydrolase activity on S. aureus PG in zymograms (28). In addition, purified Sle1 protein reduced the turbidity of S. aureus PG preparations and released dimer PG peptides, consistent with an activity as an N-acetylmuramyl-l-alanine amidase (i.e., Sle1 cleaves the amide bond between N-acetyl-muramic acid in the glycan chain and l-Ala in PG peptides that are cross-linked as dimers) (28).

In contrast, there has not been a convincing demonstration of PG hydrolase activity by any PcsB homolog from a Streptococcus species. In S. pneumoniae, Cys292Ala and His343Ala mutations in PcsB are not tolerated (18), suggesting that the CHAP-domain catalytic amino acids are required for PcsB function in cells. However, several different PG hydrolysis assay formats have failed to show activity of purified PcsB (16, 30). Recently, weak PG hydrolytic activity was reported in zymograms for the PcsB homolog, called CdhA, of S. pyogenes (31). However, the critical control was not included, showing loss of activity of mutant proteins containing amino acid changes of the catalytic cysteine or histidine. Some other properties suggest that PcsB may also play a scaffolding role in cell division complexes. Curiously, less than one-half of PcsB typically remains bound to cells grown in culture, and the rest of the PcsB is secreted into the medium (16, 30, 32). Secretion of substantial amounts of PcsB homologs into the culture medium was reported for other bacterial species in a strain- and medium-dependent manner (32–34). Cell-bound PcsB is an abundant protein of ∼5,000 monomers per cell (16), consistent with a role in complex formation. Finally, although the Cys292 and His343 catalytic amino acids are essential in PcsB, unexpectedly, an Asn366Ala change is tolerated. This result suggests that if the CHAP domain of PcsB is a PG amidase, then another amino acid can substitute for the conserved Asn in the catalytic triad.

A previous report showed that PcsB is likely membrane-bound (32). During secretion, the signal peptide is fully removed from cell-bound and secreted PcsB (16). Processed PcsB is not a hydrophobic protein and lacks predicted amphipathic helices. Therefore, we hypothesized that PcsB must interact with a membrane-bound protein on the outside cell surface. If PcsB PG hydrolase activity strictly depended on such an interaction, it would account for the lack of PG hydrolase activity of purified PcsB. Likewise, protein interactions are implicit to the hypothesis that PcsB acts as a scaffold or regulatory protein. A corollary of these hypotheses is that the PcsB interactor protein would itself be essential. In this paper, we used unbiased chemical cross-linking and immunoprecipitation approaches to show that PcsB interacts on the pneumococcal cell surface with the FtsX_Spn integral membrane protein, which in turn, interacts with the cytoplasmic FtsE_Spn protein. We also show that both FtsX_Spn and FtsE_Spn are essential and that this interaction directs PcsB to equatorial and septal sites of dividing cells. A model is proposed in which the PG hydrolytic activity of PcsB is dependent on its interaction with the FtsEX_Spn complex, which interacts with the FtsZ_Spn-dependent divisome.

Results

PcsB Is Bound on the Outer Surface of the Cell Membrane by Hydrophobic Interactions.

PcsB was epitope-tagged with two versions of the FLAG tag (DYKDDDDK) to aid in detection and pull-down experiments. One construct (PcsB-FLAG) has the 8-aa FLAG tag (16), and the other (PcsB-l-FLAG³) has a 10-aa linker (l) followed by three tandem copies of FLAG (35) fused to the carboxyl terminus of PcsB (Table S1). Each construct was expressed from the native chromosomal pcsB locus of laboratory strain R6 or an unencapsulated derivative of strain D39 (Table S1). Western blotting with anti-PcsB antibody showed that both PcsB-FLAG and PcsB-l-FLAG³ are stable, and no loss of the epitope tag was detected in pneumococcal extracts (Fig. S1) (16). Neither fusion caused any defect in cell growth, cell division, or morphology or the relative amount of PcsB secreted into the culture medium compared with parent strains (see below) (16). Because a modest decrease of less than fourfold of the cellular PcsB amount results in observable division and morphology defects (16), we conclude that the PcsB-FLAG and PcsB-l-FLAG³ fusions are functional.

To better understand PcsB binding to pneumococcal cells, we first performed a series of subcellular fractionation and wash experiments. For strain IU1845 (R6 PcsB-FLAG) growing exponentially in brain–heart infusion (BHI) broth (Materials and Methods), cell-bound PcsB-FLAG was recovered only from the membrane fraction (Fig. 2A), consistent with a previous report (32). Membrane binding of PcsB-FLAG rather than aggregation and precipitation during fractionation was also confirmed by cosedimentation of PcsB-FLAG and membranes in sucrose step gradients (35). We did not detect any PcsB_Spn in the cell wall fraction. This result contrasts with earlier reports that the PcsB homologs of S. agalactiae and S. mutans fractionate in both the cell wall and membrane fractions (33, 36). In our fractionation procedure, we digested away cell walls for a longer period with a combination of mutanolysin and lysozyme than in these earlier studies, and we confirmed microscopically that all digested cells formed spherical protoplasts that lysed on hypotonic treatment (Materials and Methods).

Fig. 2.

PcsB binds to the outer surface of the cell membrane by hydrophobic interactions. (A) Fractionation of S. pneumoniae cells after PcsB-FLAG by Western blotting using anti-FLAG antibody. Supernates (S) or pellets (P) are marked for centrifugation steps, and cell fractions are indicated below the blots. Strain IU1845 was grown exponentially and fractionated as described in Materials and Methods. Secreted PcsB-FLAG was recovered from the culture supernatant by TCA precipitation. Bound PcsB-FLAG was from cell pellets, which were washed briefly with 5 M LiCl or digestion buffer. Washed cells were digested with mutanolysin and lysozyme (PG digestion lanes), and protoplasts were lysed by adding hypotonic buffer (membrane lanes). (B) Release of PcsB-FLAG from the cell surface of IU1845 was performed by washing with the chemicals indicated. Exponentially grown strain IU1845 was washed, and cell-bound (P) or released (S) PcsB-FLAG was detected by Western blotting as described in Materials and Methods. Released PcsB-FLAG was recovered from supernates by trichloroacetic acid (TCA) precipitation. Disruption of cell membrane was monitored by release of intracellular Era protein with anti-Era antibody (38). Arrows indicate large releases of PcsB-FLAG from cell surfaces by low concentrations of SDS that did not disrupt cell membranes. The experiment was performed three times with similar results.

Cells were washed briefly with several chemicals to try to release PcsB-FLAG from the membrane surface (37). Strong ionic washes (5.0 M LiCl or 1.0 M MgCl₂) or chemical reagents that disrupt covalent disulfide (DTT) or thioester linkages (100 mM Na₂CO₃) did not remove substantial amounts of PcsB-FLAG from the cell surface (Fig. 2 A and B). Control experiments in which Western blots were stripped and reprobed with anti-Era antibody showed that none of these conditions released Era protein from its locations in the cytoplasm and inner membrane (Fig. 2B) (38). Even brief treatment with a strong chemical denaturant (8.0 M urea) only released ∼25% of PcsB-FLAG from pneumococcal cell surfaces (Fig. 2B). In contrast, very low concentrations of detergent [0.05% or 0.1% (wt/vol) SDS] released almost all of the PcsB-FLAG from cell surfaces but none of the cytoplasmic Era (Fig. 2B). Similar results were obtained when purified cell membranes were washed with the same chemicals (Fig. S2), confirming that the pneumococcal cell wall did not block PcsB release from the whole cells. We conclude that hydrophobic interactions make a major contribution to the binding of PcsB-FLAG to the outer membrane surface of pneumococcal cells.

PcsB Interacts with the FtsEX_Spn Complex in Cell Membranes.

To identify PcsB interactors, we used reversible chemical cross-linking to stabilize complexes followed by pull down of FLAG-tagged proteins. Unencapsulated strain IU3126 (R6 PcsB-l-FLAG³) was grown exponentially and treated briefly with 0.1% (wt/vol) formaldehyde, and PcsB-l-FLAG³ was pulled down with anti-FLAG agarose from purified membranes as described in Materials and Methods. In Western blots with anti-FLAG antibody, a single PcsB-l-FLAG³ band was detected in cross-linked samples that had been heated (Fig. 3A). In samples that were not heated, a prominent complex that included PcsB-l-FLAG³ was detected with a mass of ∼100 kDa. A distinct, less prominent complex was also detected with a mass of ∼120 kDa as well as some complexes with higher molecular masses. We cut out the 100- and 120-kDa bands, removed the cross-links by heating, and determined the compositions of the bands by MALDI MS as described in Materials and Methods. As expected, we identified several proteins with molecular masses corresponding to the size of these bands. In addition, we detected prominent tryptic fragments corresponding to the smaller proteins: PcsB (41.0 kDa) and FtsX_Spn (34.3 kDa) or PcsB, FtsX_Spn, and FtsE_Spn (25.8 kDa) in the 100- or 120-kDa bands, respectively (Fig. 2A). No other prominent tryptic fragments corresponding to smaller proteins were detected. We conclude that the major 100-kDa band contains a complex of PcsB and FtsX_Spn, likely in a stoichiometry of 1:2, respectively, and the 120-kDa band contains a complex of PcsB, FtsX_Spn, and FtsE_Spn, likely in a stoichiometry of 1:2:1, respectively.

Fig. 3.

PcsB forms a complex with division proteins FtsEX_Spn in cell membranes. (A) Pull down of PcsB-l-FLAG³ complexes in membranes of exponentially growing cells that were treated with formaldehyde are shown, which was described in Materials and Methods. Bands on Western blots were detected with anti-FLAG antibody. Left lanes, samples that were not cross-linked. FT, flow-through fractions; W, washed fractions; EL, eluted fractions. Right lanes, cross-linked samples that were heated to break cross-links or not heated to maintain complexes. The composition of the bands was determined by MALDI MS as described in Materials and Methods. The experiment was done two times with similar results. (B) Pull down of cross-linked PcsB-l-FLAG³ complexes in membranes detected by Coomassie blue staining. Samples were prepared, heated to break cross-links, and resolved by SDS/PAGE as described in the text and Materials and Methods. Left lane, extract from R6 control strain that does not express a FLAG-tagged protein. Right lane, extract from strain IU3126 expressing PcsB-l-FLAG³. The unique pull-down bands are indicated, and their identities were confirmed by MALDI MS. The experiment was performed three times with similar results.

To confirm this conclusion, we pulled down PcsB-l-FLAG³ from membrane fractions of cross-linked cells, removed cross-links by heating, and located possible interactors on SDS/PAGE gels stained with Coomassie dye (Fig. 3B). To reduce background, we filtered eluents of anti-FLAG affinity columns through 100-kDa cutoff filters before removing cross-links. Even with this treatment, numerous faint background bands were detected by general Coomassie staining in pull-down samples of the R6 control strain not expressing PcsB-l-FLAG³ (Fig. 3B, left lane). Nevertheless, the pull-down samples from the strain expressing PcsB-l-FLAG³ contained three prominent bands not present in the control extracts (Fig. 3B, right lane). MALDI MS identified these three bands as PcsB-l-FLAG³, FtsX_Spn, and FtsE_Spn. Together, these results support the conclusion that PcsB is in a complex with membrane division protein FtsX_Spn, which in turn, binds to cytoplasmic FtsE_Spn (Discussion) (39–41).

The conclusion that PcsB forms a complex with FtsEX_Spn was also confirmed by a reciprocal pull-down experiment of FtsE_Spn. Cytoplasmic FtsE family proteins are ATPase subunits that form a complex with cognate integral membrane FtsX proteins, and these complexes resemble an ABC transporter (39–41). We show below that FtsX_Spn and FtsE_Spn are essential proteins in S. pneumoniae. We were unable to construct an active tagged version of FtsX_Spn. However, we were able to construct an active FtsE_Spn-l-FLAG³ fusion under the control of the fucose-inducible P_fcsK promoter in the ectopic bgaA site (Fig. S4 and Table S1) (12, 42). We expressed FtsE_Spn-l-FLAG³ ectopically, because ftsE_Spn-ftsX_Spn forms an operon with overlapping translational signals, and it was unlikely that we could construct a carboxyl terminal fusion to FtsE_Spn without disrupting the translation of FtsX_Spn.

The ftsE⁺//P_fcsK-ftsE-l-FLAG³ merodiploid and a ΔftsE//P_fcsK-ftsE-l-FLAG³ mutant grew normally in the presence of fucose, indicating that FtsE_Spn-l-FLAG³ was functional and complemented the ΔftsE deletion. The ftsE⁺//P_fcsK-ftsE-l-FLAG³ merodiploid strain was grown exponentially, cells were treated with formaldehyde, and FtsE_Spn-l-FLAG³ was pulled down as described in Materials and Methods. Cross-linked FtsE_Spn-l-FLAG³ was detected in higher molecular mass complexes of ≥100 kDa on Western blots with anti-FLAG antibody (Fig. S4A, lane 4). Control experiments showed no complexes of FtsE_Spn-l-FLAG³ in extracts of cells that were not treated with formaldehyde (Fig. S4A, lane 2), and no signal was detected for the parent strain that does not express FtsE_Spn-l-FLAG³ (Fig. S4A, lanes 1 and 3). Samples containing cross-linked FtsE_Spn-l-FLAG³ were heated, and Western blots were performed with anti-PcsB antibody (16). A PcsB band was readily detected from the strain expressing FtsE_Spn-l-FLAG³ (Fig. S4B, lane 2) but not in the control experiment using a strain that does not express FtsE_Spn-l-FLAG³ (Fig. S4B, lane 1). These results strongly support the idea that FtsE_Spn and PcsB are in the same membrane-bound complex with FtsX_Spn.

Large Extracellular Loop of FtsX_Spn Interacts with the Coiled-Coil Domain of PcsB.

Sequence alignments and the TMHMM program (43) suggest that FtsX_Spn has a similar topology as the topology determined experimentally for FtsX_Eco (39). The 34.3-kDa FtsX_Spn protein likely contains four transmembrane domains interspersed with one large extracellular loop (ECL1; 15.9 kDa) and one smaller ECL (ECL2; 3.0 kDa) extracellular domains. We tested for interactions between purified PcsB lacking its signal peptide and ECL1_Spn by Far Western blotting (Fig. 4 and Materials and Methods). PcsB protein bound to ECL1 in this assay (Fig. 4, columns 1 and 2). Control experiments showed that ECL1 did not bind to the purified LytA amidase (Fig. 4, column 7) or DacA dd-carboxypeptidase, and no signal was detected on the blots when purified ECL1 or primary anti-FLAG antibody was omitted. We could not test binding between PcsB and ECL2 by this method, because ECL2-His₆-FLAG became insoluble after purification. We cloned and purified the separate coiled-coil (CC) domain and CHAP domains of PcsB (Fig. 1 and Table S1). In Far Western blots, we could detect binding of ECL1 to the CC domain (Fig. 4, columns 5 and 6) but not the CHAP domain (Fig. 4, columns 3 and 4) of PcsB. Together, these results support the hypothesis that the ECL1 domain of FtsX_Spn specifically interacts with the CC domain of PcsB.

Fig. 4.

Purified PcsB and the coiled-coil (CC) domain of PcsB, but not the CHAP domain of PcsB, interacts with purified ECL1 of FtsX_Spn. Proteins were purified, and Far Western blots were performed as described in Materials and Methods. The blot contains the amounts of purified proteins indicated at the top. The blot was probed with purified ECL1-FLAG protein and developed with anti-FLAG antibody. The expected position on the blot of PcsB-His₆, PcsB^CC-His₆, and the CHAP domain are indicated. LytA should run between PcsB^CC-His₆ and PcsB-His₆. No signal was detected on control blots when purified ECL1 or primary anti-FLAG antibody was omitted. The experiment was performed three times with similar results.

FtsX_Spn and FtsE_Spn Are Essential and Depletion of FtsX_Spn Phenocopies Depletion of PcsB.

FtsX_Spn and FtsE_Spn were categorized as likely essential in a global transposon mutagenesis study of serotype 4 strain TIGR4 (44). Consistent with this interpretation, we could not construct deletion/insertion mutations in ftsE_Spn or ftsX_Spn in an unencapsulated derivative of serotype 2 strain D39, unless ftsE⁺_Spn or ftsX⁺_Spn, respectively, was also expressed from an ectopic locus (Fig. 5). Additional confirmation of the essentiality of ftsX_Spn was obtained by titrating the expression of FtsX_Spn in a D39 Δcps ΔftsX//P_fcsK-ftsX⁺ merodiploid strain (Fig. 5). At concentrations of fucose >0.1% (wt/vol), the phenotypes of the unencapsulated ftsX merodiploid strain were indistinguishable from those phenotypes of its D39 Δcps ftsX⁺ parent; both strains grew with similar doubling times and formed diplococci, whose septa and equators stained with fluorescent vancomycin (FL-V), which binds to PG pentapeptides ending in d-Ala-d-Ala in regions of active PG synthesis in growing WT cells (Fig. 5A, Left) (16, 18, 19). At 0.05% (wt/vol) fucose, FtsX_Spn underexpression caused chaining, rounding of cells, and thick equatorial staining with FL-V, although the growth rate and yield of the ftsX merodiploid were only slightly reduced compared with the growth rate and yield of the parent strain (Fig. 5A, Center). Continued depletion of FtsX_Spn halted growth and led to the formation of nearly spherical cells that contained aberrant division septa and peripheral staining with FL-V (Fig. 5A, Right). The growth and cell morphology defects caused by depletion of FtsX_Spn are similar to those defects caused by depletion of PcsB (16, 18). We attempted to perform parallel titration experiments using a D39 Δcps ΔftsE//P_fcsK-ftsE⁺ merodiploid strain. The ectopic copy of ftsE⁺ complemented the ΔftsE mutation and allowed growth, but we were unable to reduce ectopic expression of FtsE_Spn sufficiently to interfere with growth. This topic was not pursued more in this study, because we conclusively showed that FtsX_Spn and FtsE_Spn are essential for normal growth and division of S. pneumoniae. Moreover, the similar defects in cell morphology caused by depletion of FtsX_Spn or PcsB support the hypothesis that both proteins function in the same pathway.

Fig. 5.

Defective cell morphology and FL-V staining and impaired growth caused by depletion of FtsX_Spn. (A) Morphology of cells depleted of FtsX_Spn. Depletion of FtsX_Spn was brought about in ftsX merodiploid strain IU4723 (D39 Δcps ΔftsX// P_fcsK-ftsX⁺) by decreasing the concentration of fucose in the culture medium as described in Materials and Methods. Samples were taken for phase-contrast microscopy or staining with FL-V and observation by epifluorescent microscopy at OD₆₂₀ ∼ 0.2 for cultures containing 0.05% or 0.1% (wt/vol) fucose or 300 min after fucose was removed completely from the culture medium. (B) Growth curve (linear scale) of strain IU4723 depleted for FtsX_Spn. The experiment was performed three times with similar results. (Scale bar, 2 μm.)

PcsB Localizes to Equators and Division Septa and Becomes Mislocalized on FtsX_Spn Depletion.

The FtsEX_Eco complex of Escherichia coli interacts with the FtsZ_Eco divisome apparatus (39, 45, 46). Therefore, if pneumococcal PcsB and FtsX_Spn interact, then we would expect preferential localization of PcsB to division sites, and FtsX_Spn depletion should lead to mislocalization of PcsB. We performed immunofluorescent microscopy (IFM) to localize PcsB. IFM with anti-PcsB did not produce strong signals or consistent results. Consequently, we localized PcsB-l-FLAG³ expressed from the native pcsB locus (Fig. 1) with anti-FLAG antibody (Materials and Methods). As mentioned above, cells expressing PcsB-l-FLAG³ from its native locus were indistinguishable from their parent strains. PcsB-l-FLAG³ consistently localized to septa, equators, or both places in ∼77% of 87 dividing cells that were scored (Fig. 6 and Fig. S3A). In S. pneumoniae, these sites of PG synthesis are indicated by FL-V staining (Fig. 5A, Left) (16, 18, 19) and FtsZ_Spn localization (Fig. S5, Left) (47–49). Lower amounts of PcsB-l-FLAG³ were sometimes detected at other locations around cells but not as prominently as at the septa and equators. So far, we have not discerned a pattern for this weaker localization of PcsB-l-FLAG³. In ∼23% of scored cells, PcsB-l-FLAG³ was present faintly in no discernable pattern at all. We conclude that PcsB preferentially localizes to sites of active PG synthesis and cell division at the septa and equators; however, this localization may be dynamic, and therefore, PcsB is occasionally observed in lower amounts at other cellular locations.

Fig. 6.

Localization of PcsB to equators and septa of dividing pneumococcal cells depends on FtsX_Spn. Strains IU4959 (D39 Δcps pcsB-l-FLAG³) and IU5016 (D39 Δcps pcsB-l-FLAG³ ΔftsX::P-aad9// P_fcsK-ftsX⁺) were grown exponentially, FtsX_Spn was depleted in strain IU5016, and IFM with anti-FLAG antibody and staining with DAPI were performed as described in Materials and Methods. Representative images are shown from three independent experiments. Column 1, phase-contrast images showing cell outlines; column 2, location of PcsB-l-FLAG³ (pseudocolored green); column 3, DAPI staining of DNA nucleoids (pseudocolored red); column 4, overlaid images of columns 1 (phase) and 2 (PcsB-l-FLAG³). (Scale bar, 2 μm.) Additional details in the text.

We found that PcsB-l-FLAG³ mislocalized from the septa and equators on depletion of FtsX_Spn. In BHI broth containing 0.8% (wt/vol) fucose, PcsB-l-FLAG³ localized to the septa and equators in ftsX merodiploid strain IU5016 (D39 Δcps pcsB-l-FLAG³ ΔftsX::P_c-aad9//P_fcsK-ftsX⁺) (Fig. S3B), similar to isogenic control strain IU4959 (D39 Δcps pcsB-l-FLAG³ ftsX⁺) (Fig. 6 and Fig. S3A). In 0.05% (wt/vol) fucose, the ftsX merodiploid strain grew only slightly worse than in fucose concentrations >0.1% (wt/vol) (Fig. 5). However, cells started to chain, and PcsB-l-FLAG³ was not localized or even visible in many cells (Fig. 6 and Fig. S3C). In BHI broth lacking fucose, the ftsX merodiploid strain stopped growing, and relatively little PcsB-l-FLAG³ was detected in any of the cells (Fig. 6 and Fig. S3D). As a control, we followed the localization of an FtsZ_Spn-mCherry fusion expressed from its native chromosome locus in ftsX merodiploid strain IU5150 (D39 Δcps ftsZ-mCherry ΔftsX::P_c-aad9//ΔbgaA′ P_fcsK-ftsX⁺). The FtsZ_Spn-mCherry fusion protein did not cause cell morphology defects or temperature sensitivity (Fig. S5, Left) and localized to the midcell regions of dividing cells, which was reported previously for FtsZ_Spn in S. pneumoniae (47–49). Depletion of FtsX_Spn caused division defects, but FtsZ_Spn-mCherry continued to localize to midcells on moderate depletion of FtsX_Spn (Fig. S5, Center), which caused PcsB-l-FLAG³ mislocalization (Fig. 6 and Fig. S3C). Even on severe depletion of FtsX_Spn, FtsZ_Spn-mCherry could occasionally be seen at midcells, although it also appeared mislocalized in many cells (Fig. S5, Right).

Last, we tested whether severe depletion of FtsX_Spn led to release of PcsB-l-FLAG³ from cells (Fig. 7). To maximize depletion of FtsX_Spn in this experiment, we first grew the ftsX_Spn merodiploid strain in 0.05% (wt/vol) fucose and then shifted it to medium lacking fucose (Materials and Methods). PcsB-l-FLAG³ was largely released into the medium when FtsX_Spn was severely depleted under these conditions (Fig. 7, Right), which still did not cause overt lysis of cells. Thus, depletion of FtsX_Spn led to mislocalization of PcsB, even when the FtsZ_Spn division protein, which likely interacts with the FtsEX_Spn complex, continued to localize to midcells. Together, these results support the contention that PcsB and FtsX_Spn interact.

Fig. 7.

PcsB-l-FLAG³ is released to the culture medium when FtsX_Spn is severely depleted. Strain IU5016 was grown in BHI broth supplemented with 0.05% (wt/vol) fucose at 37 °C and diluted into fresh BHI broth containing 0.8% (wt/vol) fucose, 0.05% (wt/vol) fucose, or no fucose as described in Materials and Methods. When cultures reached early exponential phase (OD₆₂₀ ∼ 0.2) or after ∼300 min, whichever came first, cells were collected by centrifugation, and PcsB-l-FLAG³ released to culture supernates (S) or in cell pellets (P) was detected by Western blotting using anti-PcsB antibody as described in the Materials and Methods. Percentages of total PcsB-l-FLAG³ in each sample in the supernatant (S) or pellet (P) are indicated. Phase-contrast microscopic examination of cells at the time samples taken for Western blotting did not reveal obvious lysis. Comparable results were obtained in three independent experiments.

CC Domain of PcsB Is Essential for Function.

Because of its interaction with FtsX_Spn, the CC domain of PcsB should be essential for function and cell growth. To test this hypothesis, we used error-prone PCR to introduce random point mutations into pcsB in vitro (Materials and Methods). After transformation of these mutated pcsB amplicons, we screened for temperature-sensitive (Ts) mutations that allowed colony growth at 32 °C but not 41.4 °C. Isolated Ts mutants contain amino acid changes in either the CC or CHAP domain of PcsB. In particular, one Ts mutant contains two amino acid changes (L78S L219P) in each of the predicted leucine zipper motifs in the CC domain of PcsB (Fig. 1). Constructed single mutants containing either the L78S or L219P change grew normally at 41.4 °C, indicating that the Ts phenotype requires both amino acid changes. The Ts phenotype caused by the pcsB^{L78S L219P} mutations was fully complemented by expression of WT pcsB⁺ from an ectopic site, showing that the pcsB^{L78S L219P} mutations are recessive. Importantly, the L78S L219P amino acid changes did not decrease the relative cellular amount of PcsB in pneumococcal cells grown at 32 °C or 41.4 °C (Fig. S6), arguing against a decrease in the stability of the mutant PcsB^{L78S L219P} protein. Consistent with a loss of function, cells expressing PcsB^{L78S L219P} showed mild or severe defects in cell morphology and division at 32 °C or 41.4 °C, respectively. (Fig. S7). At the nonpermissive temperature, the pcsB^{L78S L219P} double mutant formed spherical cells that phenocopied the defect, causing severe depletion of PcsB or FtsX (Fig. 5). From these combined results, we conclude that the CC domain of PcsB is required for the function of this essential protein.

Discussion

Relatively little is known about the regulation of PG hydrolases that remodel the PG during cell division. Unlike autolysins produced by bacteriophages or during some stages of bacterial growth, the activity and location of remodeling PG hydrolases are likely to be tightly regulated and coordinated with the divisome. In the pathogenic ovococcus bacterium, S. pneumoniae, only four (PcsB, Pmp23 putative lytic transglycosylase, DacA dd-carboxypeptidase, or DacB ld-carboxypeptidase) of 11 proteins with predicted domains of PG hydrolases are required individually for normal cell division (12). Of these four proteins, only PcsB is essential for growth in serotype 2 strains (16, 18). PcsB has a two-domain structure characteristic of many extracellular Gram-positive PG hydrolases (Fig. 1) (13, 50). Besides a putative catalytic CHAP domain that could function as a PG amidase or endopeptidase (21–23, 27), PcsB contains an extended CC domain that could bind to another protein and attach PcsB to the cell surface. Its highly conserved amino acid sequence in different pneumococcal serotype strains has made PcsB_Spn a leading candidate as a new vaccine target. Nevertheless, several purified constructs of PcsB have not exhibited PG hydrolase activity in different biochemical assays (12, 30).

This paradox of PcsB essentiality and lack of enzymatic activity led us to search for protein interactors in the pneumococcal cell membrane. We show here that cell-bound PcsB forms a complex with FtsEX_Spn in the cell membrane (Figs. 2, 3, and 7 and Fig. S4). Binding experiments provided additional evidence for an interaction between the CC domain of PcsB and the large ECL1 domain of FtsX_Spn (Fig. 4). Notably, the CC domain of PcsB is rich in hydrophobic leucine residues (Fig. 1), and PcsB is bound to the cell surface, partly by hydrophobic interactions (Fig. 2). A PcsB–FtsX_Spn interaction was also supported by mislocalization of PcsB from the equators and septa of dividing cells on depletion of essential FtsX_Spn (Figs. 5 and 6 and Fig. S3). Previously, bacterial two-hybrid assays suggested that PcsB might interact with DivIVA_Spn (49). An interaction was also reported between the PcsB homolog (GbpB_Smu) of S. mutans and ribosomal proteins L7/L12_Smu (33). However, it is difficult to reconcile direct binding between extracytoplasmic PcsB (Fig. 2) and cytoplasmic DivIVA_Spn or L7/L12_Spn, unless these abundant cytoplasmic proteins are released into the culture medium from lysed cells. However, then, it is hard to see how release and rebinding of these cytoplasmic proteins would lead to an essential mechanism involving PcsB. We did not detect either of these interactions in this study, where the short formaldehyde cross-linker was used to stabilize protein complexes containing PcsB in membranes.

The interaction between PcsB and the FtsEX_Spn complex suggests a possible explanation for the inactivity of purified PcsB. FtsE and FtsX homologs structurally resemble the cytoplasmic ATPase subunit and the membrane channel, respectively, of an ABC transporter (Fig. 8) (39, 40, 51). However, no extracellular transport receptor subunit has been associated with FtsEX, and mutations in the Walker box of FtsE_Eco impair constriction in E. coli cells (39). Consistent with a role in cell division, FtsE_Eco was shown to interact with FtsZ_Eco (46), and FtsX_Eco bound to FtsQ_Eco and FtsA_Eco (45). The binding of the CC domain of PcsB to the ECL1 of FtsX_Spn may be required for PcsB PG hydrolase activity. In this model, the interaction of PcsB and FtsEX_Spn would couple PG remodeling catalyzed by PcsB with Z-ring formation and cell division (Fig. 8). Consistent with this model, Ts mutations can be isolated in both the CC and CHAP domains of PcsB (Results and Materials and Methods). An initial attempt to stimulate a PG hydrolytic activity of purified PcsB by adding purified ECL1_Spn was not successful. However, activation of a PcsB PG hydrolase activity may depend on additional interactions besides the binding of ECL1_Spn to the PcsB CC domain, such as conformational changes induced by ATP hydrolysis by FtsE_Spn. Ongoing experiments will test these hypotheses.

Fig. 8.

Model for PcsB and FtsEX_Spn interactions and functions in S. pneumoniae. In this model, a direct PG hydrolytic activity of PcsB is regulated by interactions that include binding of the CC domain of PcsB and the ECL1 of FtsX_Spn. Membrane-bound FtsX_Spn interacts with the cytoplasmic FtsE_Spn ATPase subunit. This model postulates that PG remodeling activity by PcsB is coordinated with cell division through its interaction with the FtsEX_Spn complex, which interacts with FtsZ_Spn and other division proteins. Interaction of PcsB with the FtsEX_Spn complex could also be required for PcsB to activate other PG hydrolases involved in cell division. Additional details in the text.

Support for a general role of FtsEX in controlling PG hydrolysis emerged concurrently in E. coli from the Bernhardt laboratory and is described in the accompanying paper by Yang et al. (52). In E. coli, the FtsEX_Eco complex interacts with the EnvC_Eco adapter protein, which in turn, is required for the enzymatic activity of the AmiA and AmiB amidases (53–55). Thus, the fundamental function of FtsEX in coupling PG remodeling to cell division may be conserved among widely disparate bacterial species, such as E. coli and S. pneumoniae, although the output enzymes controlled by this interaction (in this case, EnvC_Eco-AmiAB_Eco or PcsB_Spn) are different. Consistent with these parallel functions, bioinformatic analysis places FtsX from ovococcus species, such as S. pneumoniae, and enteric bacteria, such as E. coli, into distinct phylogenetic groups (Fig. 9A). Cooccurrence analysis also shows that strong homologs of EnvC_Eco and FtsX_Eco occur together in the β-, γ-, and Δ/ε-proteobacteria (Fig. 9B, dark blue lines), whereas the combination of FtsX_Spn and PcsB_Spn homologs occurs almost exclusively in Streptococcaceae and Enterococcus species (Fig. 9B, red lines).

Fig. 9.

Phylogenetic and cooccurrence analysis of FtsX_Spn and FtsX_Eco. (A) Neighbor-joining phylogenetic trees of FtsX in different bacteria were generated as described in Materials and Methods. The percentages of replicate trees in bootstrap tests are shown next to branches in the tree. Results presented here show an interaction between PcsB_Spn and FtsX_Spn, and parallel findings show an interaction between FtsX_Eco and EnvC_Eco (52). Consistent with these different interactions, FtsX_Spn and FtsX_Eco are in distinctly different branches of the FtsX phylogenetic tree. (B) Distribution and cooccurrence of FtsX_Spn, FtsX_Eco, EnvC_Eco, and PcsB_Spn were determined by string analysis as described in Materials and Methods and are diagrammatically represented, where branches indicate groups only and not evolutionary relatedness. FtsX_Spn, FtsX_Eco, EnvC_Eco, and PcsB_Spn indicate homologs in organisms with high (>100 bits) similarity to the corresponding pneumococcal or E. coli proteins. FtsX_Low indicates FtsX homologs in organisms with low similarity scores (<100 bits) to either pneumococcal or E. coli FtsX. FtsX_Spn/Eco indicates homologs in organisms with high similarity scores (>100 bits) to both FtsX_Spn and FtsX_Eco. Cooccurrence of strong homologs to FtsX_Eco and EnvC_Eco occur throughout β-, γ-, and Δ/ε-proteobacteria, whereas cooccurrence of strong homologs of FtsX_Spn and PcsB_Spn are confined to the Streptococcaceae and Enterococcus faecalis. The Actinobacteria, Clamydiales, Bacillales, and Lactobacillaceae lack strong homologs of EnvC_Eco and PcsB_Spn and presumably have other classes of proteins that interact with their FtsX homologs. Additional details are in the text.

In E. coli, the FtsEX_Eco interactor is the EnvC_Eco regulatory protein that lacks enzymatic activity, despite amino acid similarities to LytM PG hydrolases (54, 55). At this stage, we cannot rule out that PcsB acts as a regulatory protein for some other PG hydrolases, and this possibility is included in our current working model (Fig. 8). A role as an adaptor or scaffolding protein is not mutually exclusive with function as a PG hydrolase. Besides FtsX_Spn (Fig. 3), we have not yet detected another interactor with PcsB, although these studies continue. If PcsB were acting as an adaptor, then deletion of genes encoding the PG hydrolases that interact with PcsB would result in lethal phenotypes for growth. To date, we have knocked out the spd_1874 gene, which encodes a putative N-acetyl muramidase (lysozyme), combined with nine other genes encoding established or putative PG hydrolases (Table S1) (12). None of these double mutations was lethal. Likewise, we constructed the triple mutant knocked out for all three of the genes (lytA (amidase) (56), cbpD (putative N-acetylmuramoyl-l-alanine amidase or endopeptidase) (57), and lytC [N-acetyl muramidase (lysozyme)] (58)) that mediate growth-dependent autolysis and competence-induced fratricide. This set includes the only two other predicted amidases in S. pneumoniae. The triple mutant was not impaired for growth (59). Additional combinations of PG hydrolase mutations are being constructed and tested, but currently, there is no evidence for PcsB acting as an adapter, analogous to EnvC_Eco.

Unlike their homologs in E. coli and B. subtilis, FtsX_Spn and FtsE_Spn are essential for growth, even in BHI broth containing 0.5% (wt/vol) NaCl, which rescues ftsEX_Eco mutants (40, 60). Strikingly, depletion of FtsX_Spn or PcsB results in similar defects in chaining and cell division (Fig. 5) (16, 18). This phenocopying is consistent with a direct interaction between PcsB and FtsX_Spn as well as mislocalization of PcsB on depletion of FtsX_Spn (Figs. S3 and S5). The essentiality of FtsEX_Spn matches the expectation that the membrane interactor of essential PcsB would itself be essential (shown in the Introduction). IFM indicates that PcsB localizes to the equators and septa of most dividing cells of an unencapsulated derivative of strain D39 (Fig. 6 and Fig. S3). We also observed lower amounts of PcsB at other sites in some cells, suggesting that PcsB localization may be dynamic and depend on stage of cell division. Septal localization was recently reported for the PcsB homolog (CdhA_Spy) of S. pyogenes (31). However, our results do not agree with the recent conclusion that PcsB is located at cell poles and is excluded from the septa of dividing cells (30). We clearly detect PcsB-l-FLAG³ at equators and septa in a pattern that matches regions of PG synthesis indicated by FL-V staining and FtsZ_Spn localization (Fig. S5, Left) (47–49), and we did not observe PcsB-l-FLAG³ at poles (Results, Fig. 6, and Fig. S3).

This discrepancy is not likely caused by using tagged PcsB. We did not detect any division defects in cells expressing carboxyl-terminal PcsB-l-FLAG³ from the native pcsB chromosomal locus (Figs. 1 and 6 and Fig. S3). In addition, moderate depletion of FtsX_Spn mislocalized PcsB-l-FLAG³ but not FtsZ_Spn from midcells (Fig. S3). Given its essentiality in pneumococcus and its localization in E. coli (39, 40), it is highly likely that FtsX_Spn interacts with the FtsZ_Spn divisome (Fig. 8). Moreover, IFM images in reference (30) do not include cell outlines, making it difficult to tell where staining occurred in cells and the stage of division. In fact, in some cells, PcsB seems to be located at the septa, and PcsB and FtsZ_Spn staining partially overlaps in some of the co-IFM images (30). We purposefully performed our IFM studies in unencapsulated derivatives of the prototypic serotype 2 strain D39, because capsule leads to chaining and masks or alters some morphological defects (12, 16). Although this other study was performed in an encapsulated serotype 4 strain (30), it is unclear whether the difference in genetic backgrounds or the presence of capsule would lead to a discrepancy in PcsB localization, if there is a discrepancy. IFM studies in progress will also examine whether PcsB localization is dynamic as a function of the cell cycle and will determine the localization of FtsX_Spn and FtsE_Spn expressed from their native chromosomal locus.

Materials and Methods

SI Materials and Methods contains descriptions of the bacterial strains used, culture growth conditions, strain constructions, procedure used to deplete essential FtsX_Spn, and fractionation of S. pneumoniae cells. It contains the protocols for formaldehyde cross-linking and pull down of PcsB complexes, purification of PcsB and its domains, ECL1_Spn, and LytA protein used in Far Western blotting, and the reciprocal pull down of FtsE_Spn complexes. IFM protocols and bioinformatic analyses of phylogenetic and cooccurrence of FtsX_Spn and PcsB and FtsX_Eco and EnvC_Eco are also described. Primers synthesized for this study are listed in Table S2.