The Requirement for Pneumococcal MreC and MreD Is Relieved by Inactivation of the Gene Encoding PBP1a ^,

Abstract

MreC and MreD, along with the actin homologue MreB, are required to maintain the shape of rod-shaped bacteria. The depletion of MreCD in rod-shaped bacteria leads to the formation of spherical cells and the accumulation of suppressor mutations. Ovococcus bacteria, such as Streptococcus pneumoniae, lack MreB homologues, and the functions of the S. pneumoniae MreCD (MreCD_Spn) proteins are unknown. mreCD are located upstream from the pcsB cell division gene in most Streptococcus species, but we found that mreCD and pcsB are transcribed independently. Similarly to rod-shaped bacteria, we show that mreCD are essential in the virulent serotype 2 D39 strain of S. pneumoniae, and the depletion of MreCD results in cell rounding and lysis. In contrast, laboratory strain R6 contains suppressors that allow the growth of ΔmreCD mutants, and bypass suppressors accumulate in D39 ΔmreCD mutants. One class of suppressors eliminates the function of class A penicillin binding protein 1a (PBP1a). Unencapsulated Δpbp1a D39 mutants have smaller diameters than their pbp1a⁺ parent or Δpbp2a and Δpbp1b mutants, which lack other class A PBPs and do not show the suppression of ΔmreCD mutations. Suppressed ΔmreCD Δpbp1a double mutants form aberrantly shaped cells, some with misplaced peptidoglycan (PG) biosynthesis compared to that of single Δpbp1a mutants. Quantitative Western blotting showed that MreC_Spn is abundant (≈8,500 dimers per cell), and immunofluorescent microscopy (IFM) located MreCD_Spn to the equators and septa of dividing cells, similarly to the PBPs and PG pentapeptides indicative of PG synthesis. These combined results are consistent with a model in which MreCD_Spn direct peripheral PG synthesis and control PBP1a localization or activity.

INTRODUCTION

Three basic patterns of peptidoglycan (PG) synthesis have emerged from the study of rod-shaped (bacillus), spherical (coccus), and ellipsoid (ovococcus) bacteria (reviewed in references 12, 71, and 72). In rod-shaped bacteria, like Escherichia coli, Bacillus subtilis, and Caulobacter crescentus, PG synthesis occurs at two locations catalyzed by distinct sets of proteins (12, 35, 36, 71). Lateral PG synthesis elongates the side walls of these bacteria and results in their rod shape (12, 35, 71). Lateral PG synthesis is mediated by actin-like MreB paralogue proteins, which form filamentous helical structures in spirals over the length of cells (12, 35, 66). MreB is thought to organize the MreC and MreD proteins, which in turn recruit specific penicillin-binding proteins (PBPs) and other division proteins that catalyze lateral PG formation (26, 54, 55, 70). Septal PG synthesis is organized by FtsZ ring formation, followed by the orderly assembly of the divisome complex, which includes a set of PBPs different from those involved in lateral PG synthesis (16, 22, 23). The mutational interruption of lateral or septal PG synthesis of rod-shaped bacteria generally results in the formation of spherical or elongated cells, respectively (3, 4, 9, 31). Although useful, this two-state model is obviously an oversimplification (35), and recent results reveal that MreB, MreC, MreD, and specific PBPs also form annular rings adjacent to FtsZ rings at the septa of dividing E. coli cells (67). Moreover, some PBPs and division regulators, such as PBP1 and GpsB of B. subtilis, shuttle between the lateral and septal PG synthesis machineries (9).

Much less is known about the mechanisms of PG synthesis in coccus and ovococcus species. In spherical species, such as Staphylococcus aureus, only a septal PG synthesis machinery seems to be present (20, 47, 72), although S. aureus still encodes homologues of proteins like MreC and MreD that mediate lateral PG synthesis in rod-shaped bacteria (35). In contrast, ovococcus species, such as Streptococcus pneumoniae and Lactococcus lactis, form American football-shaped cells by a combination of peripheral sidewall and septal PG synthesis that occurs in the midcell regions of dividing cells (Fig. 1) (11, 42). Ovococcus species lack an MreB homologue, and peripheral and septal PG synthesis are likely coordinated with and organized by FtsZ ring formation (64, 65, 72). Because two modes of PG biosynthesis are required to form ellipsoid-shaped bacteria, models have been proposed for two different PG synthesis machineries (18, 21, 33, 72). Based on the composition of the complexes in rod-shaped cells, it has been hypothesized that specific sets of homologous cell division proteins mediate peripheral (MreC, MreD, RodA, and PBP2b) and septal (FtsZ, EzrA, PBP2x, FtsW, and DivIB/FtsL/DivIC) PG synthesis in ovococcus bacteria (72). A recent study correlating the effects of mutations or antibiotic treatments to changes in cell shapes supports the idea that the PBP2x and PBP2b class B transpeptidases are associated with septal and peripheral PG synthesis, respectively, in L. lactis (46) and provides some of the first direct evidence for a two-state model in ovococci. However, most other aspects of this model are largely untested, and it remains to be determined whether the two ovococcus PG synthesis machineries exist as two distinct complexes (Fig. 1) or form one large complex at midcell (72). In addition, the roles of class A (dual-function transglycosylase-transpeptidase) PBP1a, PBP1b, and PBP2a are unknown.

Fig. 1.

Two-state model of PG biosynthesis in ovococcus bacteria, such as S. pneumoniae. To attain an ellipsoid shape, two machineries have been proposed that carry out peripheral and septal PG synthesis (12, 71, 72). At the start of a division cycle, components of both machines locate to the equators of cells (bottom). One machine (orange dots) carries out peripheral PG synthesis (light blue; top) between the future equator and septum of dividing cells. Peripheral PG synthesis corresponds to a form of lateral (sidewall) PG synthesis in rod-shaped bacteria. At some point, septal PG synthesis (medium blue; green dots) commences to divide the cell in two. For clarity, the peripheral and septal machines are shown at separate locations at the middle of dividing cells, although the spatial locations and compositions of the two machines have not been established (12, 72). The red dot corresponds to PG hydrolases that remodel the PG and allow septal separation. See the text for additional details.

In this paper, we report the first characterization and localization of the MreC and MreD proteins in a coccus or ovococcus species. In rod-shaped bacteria, MreC and MreD are essential proteins that play crucial roles in lateral PG synthesis (3, 5, 15, 27, 31). The genes encoding MreC and MreD are cotranscribed with that encoding MreB (17), and all three Mre proteins are involved in organizing the helical insertion of new PG during sidewall elongation (13, 31, 70). The exact functions of MreC and MreD remain unknown. MreC consists of three domains: a cytoplasmic amino-terminal tail, a membrane-spanning domain, and an extracytoplasmic coiled-coil domain required for self dimerization (62). MreD is an integral membrane protein with five membrane-spanning domains (27, 31). The depletion of MreC or MreD in rod-shaped bacteria leads to the formation of spherical cells and eventual lysis (3, 27, 31). Fractionation and two-hybrid analyses demonstrated that E. coli and B. subtilis MreC are in the cell membrane, where they interact with MreD (28, 62). MreC has been shown to interact with multiple PBPs in C. crescentus and B. subtilis (13, 62). In E. coli, C. crescentus, and B. subtilis, MreC localizes helically along the cylindrical walls of cells (15, 31, 67). In E. coli and B. subtilis, the helical sidewall localization of MreC is organized by MreB family proteins (4, 25, 26, 58, 66), whereas the migration of MreC to the paraseptal rings adjacent to the FtsZ ring of dividing E. coli cells is independent of MreB or MreD (67). In contrast to E. coli and B. subtilis, the helical filament formation of MreC is independent of MreB in C. crescentus, where MreC is located detached from the inner membrane in the periplasm (15).

In ovococcus species that produce MreC and MreD, such as Streptococcus pneumoniae, Streptococcus mitis, Streptococcus mutans, Streptococcus thermophilus, Lactococcus lactis, and Enterococcus faecalis, the mreCD genes are always located immediately upstream of the pcsB gene, which plays roles in cell division and PG biosynthesis (L.-T. Sham, unpublished data) (1, 19, 42, 43). pcsB is essential in serotype 2 strains of S. pneumoniae but essentiality can be masked by the accumulation of different kinds of bypass suppressor mutations (S. M. Barendt, unpublished data) (1). Curiously, several important ovococcus species, including Streptococcus pyogenes (group A Streptococcus) and Streptococcus agalactiae (group B Streptococcus), do not encode identifiable homologues of MreC and MreD (72). Previously, we reported that MreC and MreD were not essential in the R6 laboratory strain of S. pneumoniae (1). Further experiments reported here for the prototypical virulent serotype 2 strain D39 show that MreC and MreD are indeed essential and likely play roles in peripheral PG synthesis in pneumococcus. We show that the ability to knock out the mreCD genes in the R6 and D39 backgrounds is due to the accumulation of suppressor mutations, including inactivation or knockout mutations in the pbp1a gene, which encodes a major class A PBP that influences the diameter of pneumococcal cells. Finally, we show that MreC and MreD localize to the equators and septa of dividing pneumococcal cells. We were unable to detect strong filament formation of the type observed in rod-shaped bacteria.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Bacterial strains used in this study were derived from S. pneumoniae serotype 2 strain D39 and its unencapsulated laboratory strain derivative R6 (30) (see Table S1 and Fig. S1 in the supplemental material). Bacteria were grown on plates containing Trypticase soy agar II (modified; Becton-Dickinson) and 5% (vol/vol) defibrinated sheep blood (TSAII-BA) and incubated at 37°C in an atmosphere of 5% CO₂. Strains were cultured statically in Becton-Dickinson brain heart infusion (BHI) broth at 37°C in an atmosphere of 5% CO₂, and growth was monitored by optical density at 620 nm (OD₆₂₀) using a Spectronic 20 spectrophotometer fitted for the measurement of capped tubes (outer diameter, 16 mm). Bacteria were inoculated into BHI broth from frozen cultures or colonies, serially diluted into the same medium, and propagated overnight. Overnight cultures that were still in exponential phase (OD₆₂₀ of 0.1 to 0.6) were diluted back to an OD₆₂₀ of 0.005 to start final experimental cultures, which did not contain antibiotics. Overnight cultures of strains containing genes fused to the inducible P_fcsK promoter (6) contained 0.2 to 0.8% (wt/vol) l-fucose. Cells from these overnight cultures were collected by centrifugation (10 min at 3,200 × g at 25°C), washed twice in BHI broth lacking l-fucose, and suspended in fresh BHI broth containing (0.2 to 0.8% [wt/vol]) or lacking l-fucose. Sensitivity to penicillin G was determined as described previously (1).

Construction and verification of mutants.

Strains containing antibiotic markers were constructed by transforming linear DNA amplicons synthesized by overlapping fusion PCR into competent pneumococcal cells as described previously (43, 49, 51, 53). Primers synthesized for this study are listed in Table S2 in the supplemental material. Transformations were carried out as described before (2). TSAII-BA plates were supplemented, as appropriate, with the following final concentrations of antibiotics: 250 μg kanamycin per ml, 150 μg spectinomycin per ml, 0.3 μg erythromycin per ml, or 200 μg streptomycin per ml. All constructs were confirmed by PCR, where reaction mixtures contained genomic DNA prepared from cell lysates (2), and by DNA sequencing. Amplicons containing these mutations were purified using a PCR cleanup kit (Qiagen) and sequenced in reaction mixtures containing 1 μl of BigDye Terminator reagent (Applied Biosystems) as described previously (30). Sequences were aligned and analyzed using Vector NTI software (Invitrogen).

Mapping of ΔmreCD suppressor mutations by whole-genome sequencing.

D39 strains IU3897 and IU3898 containing suppressor mutations that allowed the growth of a ΔmreCD deletion mutant were isolated as described in Results. Overnight cultures still in exponential phase (see above) were diluted into 30 ml of BHI broth in 50-ml conical tubes to an OD₆₂₀ of ≈0.001 and grown to an OD₆₂₀ of ≈0.050. Cells were collected by centrifugation (10,000 × g for 10 min at 4°C). Genomic DNA was purified from collected cells using a MasterPure Gram-positive DNA purification kit (Epicenter Biotechnologies) according to the manufacturer's protocol. Five μg of purified genomic DNA from each strain was used to construct a genomic library for single-end-read Illumina DNA sequencing as described by Lazinski and Camilli (http://www.tucf.org/htseq_protocol_for_illumina_paired.pdf). DNA from IU3897 and IU3898 was tagged with bar-coded adapters Illumina bar 1A and Illumina bar 1B, respectively (see Table S2 in the supplemental material). Illumina DNA sequencing in a single-lane run was performed by the Tufts University Core Facility, which assembled the sequences based on the published D39 genome sequence (30) and performed bioinformatic analyses. Full coverage was obtained for the genomes of each mutant with >200 reads of most base pairs and only one small gap in one mutant.

Reverse transcription-PCR (RT-PCR).

Total pneumococcal RNA was purified as described previously (44). To detect mreCD-pcsB cotranscripts, cDNA was synthesized from total RNA using primer WN212 (see Table S2 and Fig. S2 in the supplemental material) and the Stratascript first-strand synthesis kit according to the manufacturer's instructions. Cotranscripts were detected by PCR amplification in reaction mixtures containing rTth DNA polymerase (Applied Biosystems), synthesized cDNA, and primer pairs AL005/WN212, AL006/WN212, and WN259/WN212 (see Table S2 and Fig. S2). PCR amplification reaction mixtures containing genomic DNA as the template or that omitted reverse transcriptase in the first step of the procedure were used as positive or negative controls, respectively.

Northern blotting.

Total RNA was purified from pneumococcal strains as described previously (44). Northern blots were performed using a Northern Max kit (Ambion). RNA Millennium size markers (Ambion) and 20 to 30 μg of total RNA from the indicated strains were separated in individual lanes on denaturing 1% (wt/vol) agarose gels and transferred to Bright Star Plus positively charged nylon membranes (Ambion) according to the manufacturer's instructions. Primers MreC and PcsB (see Table S2 in the supplemental material) were end labeled using T4 polynucleotide kinase (New England BioLabs) and 1.67 pmol of [γ-³²P]ATP (6,000 Ci per mmol; Perkin Elmer) and purified on Sephadex G-25 quick-spin columns (Roche). Radiolabeled probes were hybridized to blots in Ultrahyb-Oligo buffer (Ambion) according to the manufacturer's instructions. Labeled Northern blots were exposed to X-ray film to obtain images. Blots were stripped and hybridized with a radiolabeled probe specific for 5S RNA (see Table S2) to confirm the equal loading and transfer of RNA samples (61).

Microscopy.

Five hundred-μl samples were taken from exponentially growing BHI broth cultures at an OD₆₂₀ of 0.1 to 0.2. Cells were collected by centrifugation in a microcentrifuge (21,000 × g for 10 min at 25°C) and resuspended in 50 μl of room-temperature BHI broth. Concentrated cells were stained without fixation by adding 4,6-diamidino-2-phenylindole (DAPI), a 1:1 mixture of vancomycin and Bodipy-FL-conjugated vancomycin (FL-V) (Molecular Probes), or FM-4-64 (Molecular Probes) to a final concentration of 0.2, 2, or 10 μg per ml, respectively, as described previously (1). Cells were viewed using an epifluorescent microscope, and images were recorded as described before (1). In control experiments, cells in exponential phase (OD₆₂₀ of 0.10 to 0.20) were examined without staining to ensure that cell morphology defects observed were not caused by the concentrations of vancomycin used for staining (data not shown). Length-to-width aspect ratios (ARs) were determined from phase-contrast micrographs by using Nikon NIS-Elements AR software as described previously (1).

Immunofluorescent microscopy (IFM) was performed to localize MreC and MreD in exponentially growing cells. MreC and MreD were tagged at their carboxyl termini with a peptide linker (L) followed by three tandem copies of the FLAG epitope tag (L-FLAG³) as described previously for other proteins (2, 69). The MreC-L-FLAG³ and MreD-L-FLAG³ proteins were expressed from their native chromosomal locus in unencapsulated laboratory strain R6 and in an unencapsulated derivative of strain D39 (strains IU3238, IU4868, and IU4970; see Table S1 in the supplemental material). IFM was performed using commercially available purified polyclonal anti-FLAG antibody (F7425; Sigma) exactly as described previously (2, 69). Positive-control experiments showed the proper localization of FLAG-tagged FtsZ at equators and septa (69), and negative-control experiments showed no detectable staining of bacteria that did not contain a FLAG-tagged protein (data not shown). Only single bands of the expected molecular masses were detected for MreC-L-FLAG³ and MreD-L-FLAG³ in Western blots (see below) probed with anti-FLAG antibody, indicating no significant cleavage of the tag from the fusion proteins.

Western blotting.

The preparation of extracts for Western blotting was done in two ways, which gave comparable yields of the membrane-bound MreC and MreD proteins. For routine Western blotting, 1-ml samples of exponentially growing cultures at an OD₆₂₀ of ≈0.2 were microcentrifuged (21,000 × g for 10 min at 25°C). Cell pellets were lysed directly by resuspension in 100 μl of lysis buffer (1% [wt/vol] SDS, 0.1% [vol/vol] Triton X-100), and protein concentrations were determined using the DC protein assay (Bio-Rad). Eighty μl of cell lysates was mixed with 80 μl of 2× SDS-PAGE sample buffer (Bio-Rad) containing freshly added 5% (vol/vol) β-mercaptoethanol and heated for 10 min at 95°C. Cell lysate samples usually containing 15 μg of protein were resolved on 4% stacking-10% SDS-PAGE gels in Tris-glycine buffer as described before (69). Electrophoresis, membrane transfer, and immunodetection using ECL chemiluminescent reagents were performed as described in reference 69. Bands on blots were detected by autoradiography on XAR film and quantitated by photonic imaging using an IVIS Xenogen VivoVision Lumina system and software as described before (69).

Cellular amounts of MreC were determined by quantitative Western blotting in unencapsulated strain R6 as described previously (1, 69). Briefly, 1.0-ml samples of the parent R6 strain (EL59) and an R6 ΔmreCD mutant (IU1983) were taken from exponentially growing BHI broth cultures at an OD₆₂₀ of 0.2 to 0.3 and centrifuged (21,000 × g for 10 min at 25°C). Pellets were suspended in 100 μl of a lysis buffer containing 40 μg mutanolysin per ml (69) and incubated at 37°C for 10 min. One hundred μl of 2× SDS sample buffer (Bio-Rad) containing 5% β-mercaptoethanol was added, and samples were incubated at 90°C for 8 min. Lysates were microcentrifuged (3,330 × g for 3 min at 25°C), and 10 and 25 μl of the supernatants were resolved by SDS-PAGE (1, 69). His₆-(N)-MreC lacking a membrane domain was purified as a standard as described in reference 1. Standard curves for MreC quantification were generated by adding 0.0025 to 0.025 μg of purified His₆-(N)-MreC to 10 μl of lysate of the R6 ΔmreCD deletion mutant prior to loading onto SDS-PAGE gels. Proteins were transferred onto nitrocellulose membranes, and MreC protein was detected using polyclonal anti-MreC antibody as reported previously (1). Chemiluminescent detection and quantification were performed using an IVIS imaging system (see above) (1, 69). CFU determinations of the R6 cultures used for lysate preparations were determined by serial dilution and plating onto TSAII-BA plates. Molecules per cell were calculated from CFU, taking into account that the unencapsulated R6 strain formed primarily diplococci and chains of two diplococci at a 3:1 ratio (data not shown), which means on average that each CFU corresponded to ≈2.3 cells.

RESULTS

Pneumococcal mreCD and pcsB are transcribed independently.

To determine whether the conserved grouping of mreCD with pcsB implies cotranscription or regulation, we first performed RT-PCR on the mreCD-pcsB intercistronic region (see Materials and Methods and Fig. S2 in the supplemental material). Total RNA was prepared from laboratory strain R6 and virulent serotype 2 strain D39, including DNase treatment to remove contaminating genomic DNA. cDNA was generated by reverse transcription from reverse primer WN212, which anneals 198 nucleotides downstream from the pcsB start codon (see Fig. S2). This cDNA was used as the template in PCRs carried out using primer WN212 and forward primers annealing within mreC, mreD, or the mreD-pcsB intercistronic region (see Fig. S2). Reverse transcription-dependent amplification products were obtained with each primer pair, indicating the presence of a cotranscript extending at least from within mreC to pcsB (see Fig. S2).

However, Northern blots indicated considerable independent expression of pcsB. Northern analysis was performed using probes specific for mreC and pcsB (Fig. 2A). Total RNA was prepared from cultures of the parent R6 strain (EL59), an R6 ΔmreCD mutant (IU1983), and a ΔpcsB mutant containing a characterized bypass suppressor mutation (IU3363) (S. M. Barendt, unpublished). For the parent and ΔmreCD strains, hybridization with the pcsB-specific probe detected a ≈1.2-kb transcript (Fig. 2B) corresponding to the predicted size of a monocistronic pcsB transcript extending from the P_pcsB promoter to a predicted factor-independent terminator downstream of pcsB (Fig. 2A). This pcsB transcript was absent from RNA from the ΔpcsB bypass mutant (Fig. 2B). No mreCD-pcsB cotranscript was detected using the pcsB-specific probe (Fig. 2B). The mreC-specific probe failed to detect any mreC transcripts, despite the presence of 5S rRNA on blots (data not shown). Thus, a considerable amount of pcsB transcript originating from the WalR_Spn-dependent P_pcsB promoter, but no mreCD or mreCD-pcsB cotranscript, was detected in Northern blots.

Fig. 2.

Northern blots indicate that pcsB_Spn is a monocistronic operon. (A) Diagram of the mreCD-pcsB-rpsB region of the pneumococcal chromosome drawn to scale. rpsB encodes ribosomal subunit S2. The mapped P_pcsB promoter, which is regulated by the WalRK_Spn (VicRK) two-component regulatory system (44, 50) and putative factor-independent transcription terminators (lollipops), are indicated. The bars indicate positions of probes used in Northern blots. The stop codon of mreC overlaps the start codon of mreD by 1 bp; the intercistronic space between the stop codon of mreD and the start codon of pcsB is 93 bp and contains a putative transcription terminator ending 44 bp upstream from the pcsB start codon; the WalR_Spn (VicR) response regulator binding site is 59 to 79 bp upstream of the pcsB start codon (44); the transcription start of the P_pcsB promoter is a G residue 29 bp upstream of the pcsB start codon (L.-T. Sham, unpublished); the downstream putative transcription terminator starts 6 bp after the stop codon of pcsB. (B) Northern blots of total RNA extracted from the R6 parent strain (EL59), ΔmreCD mutant IU1983, and a D39 ΔpcsB mutant containing a defined suppressor mutation (S. M. Barendt, unpublished). RNA was extracted from exponentially growing cells, and Northern blotting was performed as described in Materials and Methods. The positions of RNA standards are indicated. The upper panel shows a blot hybridized with the pcsB-specific probe. A single band was detected in R6 and ΔmreCD lanes but not in the ΔpcsB sup lane, with a size corresponding to a transcript extending from the P_pcsB promoter to the terminator after pcsB. The lower panel shows the same blot stripped and reprobed for 5S rRNA, indicating approximately equal loading in all three lanes. Hybridization with the mreC-specific probe did not reveal an mreCD transcript (data not shown; see Discussion). The experiment was done three times independently with the same results.

MreC and MreD are not essential in laboratory strain R6.

R6 is a commonly used laboratory strain that is diplococcal due to the absence of capsule (1). R6 contains more than 70 additional drift mutations compared to its serotype 2 D39 progenitor strain (30). We constructed deletion amplicons in which the mreC, mreD, or mreCD reading frames were replaced by the P_c synthetic promoter driving an erm (erythromycin resistance) gene or by the promoterless in-frame add9 gene (spectinomycin resistance) (Table 1, rows 1, 4, 5, and 6; also see Table S1 in the supplemental material). Both types of mreCD insertion amplicons readily transformed the R6 strain and yielded large numbers of colonies characteristic of transformations of control amplicons containing the resistance markers in other genes (Table 1, rows 2, 3, and 7). All transformants appeared after overnight incubations, and recovered ΔmreCD, ΔmreC, or ΔmreD mutants grew like the R6 parent strain in culture and had normal cell morphology (Fig. 3A and C and data not shown). The R6 parent and ΔmreCD mutants also responded similarly to stress responses, including high temperature (limit for growth on plates, 41.4°C) and sensitivity to 0.1 μg penicillin G per ml (data not shown) (1, 61).

Table 1.

Relative transformation of ΔmreCD amplicons into laboratory strain R6 and into the D39 genetic background^a

Amplicon	No. of colonies in recipient strain
	R6	R6 ΔbgaA′::Pfcsk-mreCD with fucose	D39 Δcps	D39 Δcps ΔbgaA′::Pfcsk-mreCD with fucose
1. ΔmreCD<>Pc-erm	>500	>500	<20	>500
2. pcsB+-Pc-erm	>500	>500	>500	>500
3. lctO::erm	>500	NTb	>500	NT
4. ΔmreCD<>aad9	>500	>500	<20	>500
5. ΔmreC<>aad9	>500	>500	<20	>500
6. ΔmreD<>aad9	>500	>500	<20	>500
7. purR::aad9	>500	>500	>500	>500

^aUnencapsulated pneumococcal strains R6 (EL59), R6 ΔbgaA′::P_fcsK-mreCD (IU2920), D39 Δcps (IU1945), and D39 Δcps ΔbgaA′::P_fcsK-mreCD (IU3890) were grown and transformed as described in Materials and Methods. An equal amount (100 ng) of linear double-stranded DNA amplicon was used for each transformation. Transformation reactions and the plating of IU2920 and IU3890 were performed in the presence of 0.8% (wt/vol) fucose. Numbers of transformants were scored after 14 to 18 h of incubation for strains other than D39 Δcps, which was incubated for at least 24 h. The experiments were performed at least three times independently with similar results.

^bNT, not tested.

Fig. 3.

MreC and MreD are not required for the growth or maintenance of cell shape in laboratory strain R6. (A and B) Growth curves of the R6 parent (EL59) and ΔmreCD mutant IU1983 and merodiploid strains IU2920 (mreCD⁺//P_fcsK-mreCD⁺) and IU2922 (ΔmreCD//P_fcsK-mreCD⁺) (see Table S1 in the supplemental material) grown in the presence or absence of fucose as described in Materials and Methods. IU2920 and IU2922 were maintained in the presence of fucose at all stages of the strain construction and storage and were deprived of fucose only in growth experiments. (C) Phase-contrast micrographs of strain R6, ΔmreCD mutant IU1983, and the merodiploid (IU2922 [ΔmreCD//P_fcsK-mreCD⁺]) with and without fucose sampled during exponential growth at an OD₆₂₀ of approximately 0.2. The experiment was performed five times independently with the same results.

We tested for suppressor accumulation during transformation with the ΔmreCD amplicons. mreCD⁺ transcripts were expressed under the control of the fucose-inducible P_fcsK promoter from the ectopic ΔbgaA′ locus (6, 52). Quantitative RT-PCR and Western blotting confirmed that mreCD⁺ transcript and MreC protein were overexpressed by ≈5-fold or underexpressed by ≈2-fold from the ectopic site in the presence or absence of fucose, respectively, compared to the expression of the R6 parent (see Tables S3 and S4 in the supplemental material). The transformation of the ΔmreC, ΔmreD, and ΔmreCD amplicons into the R6 strain ectopically expressing MreCD (with fucose) was indistinguishable from the transformation of the parent R6 strain (Table 1, rows 1, 4, 5, and 6). Recovered ΔmreCD mutants in the ectopic expression strain were maintained in the presence of fucose to prevent the selection of suppressor mutations in response to MreCD loss. The removal of fucose resulted in the same growth properties as those in the presence of fucose (Fig. 3B). We conclude that MreCD are not essential in laboratory strain R6 and that the lack of MreCD does not appreciably affect growth (Fig. 3A and B) or cell morphology (Fig. 3C) in the R6 background.

MreC and MreD are essential in the strain D39 genetic background and are required for normal cell shape.

In contrast to laboratory strain R6, the transformation of ΔmreC, ΔmreD, and ΔmreCD amplicons into an unencapsulated isogenic derivative (IU1945) of serotype 2 strain D39 yielded few colonies (<20) of diverse sizes only after longer (>24-h) incubation (Table 1, rows 1, 4, 5, and 6). Transformations with control amplicons again yielded large numbers of colonies overnight (Table 1, rows 2, 3, and 7). The same patterns of transformation were observed in encapsulated strain D39 and derivatives, indicating that capsule did not play a role in the low rate of recovery of ΔmreC, ΔmreD, and ΔmreCD mutants (see below and data not shown). The appearance of relatively low numbers of colonies of variable sizes after longer incubation is a hallmark of suppressor mutation accumulation (1, 49).

To test this idea and the essentiality of MreCD in the D39 background, we transformed the ΔmreC, ΔmreD, and ΔmreCD amplicons into the merodiploid strain expressing an extra copy of mreCD⁺ from an ectopic site (Table 1, rows 1, 4, 5, and 6). This time, the ΔmreCD mutant amplicons readily transformed the merodiploid strain, producing uniform-sized colonies in 16 to 18 h, similarly to those in the control transformations (Table 1, rows 2, 3, and 7). The D39 Δcps ΔmreCD (or ΔmreC or ΔmreD)//P_fcsK-mreCD⁺ merodiploid strains grew in the presence of 0.8% (vol/vol) fucose like the D39 Δcps mreCD⁺ parent, and the merodiploid cells appeared as typical diplococci with equatorial and septal staining by fluorescent vancomycin (FL-V) (Fig. 4; also see Fig. S3 and S4 in the supplemental material), which binds to PG pentapeptides in regions of active PG biosynthesis (1, 2, 42, 46, 47).

Fig. 4.

MreC and MreD are essential in the D39 genetic background. Merodiploid strain IU3931 (ΔmreCD//P_fcsK-mreCD⁺) was constructed in an unencapsulated derivative of strain D39 as described in Materials and Methods and Table S1 in the supplemental material. IU3931 was maintained in the presence of fucose at all stages of construction and storage prior to growth experiments. (A) Growth curves following the resuspension of starter cultures grown in the presence of fucose in fresh BHI broth containing 0.8% (wt/vol) fucose or lacking fucose (see Materials and Methods). In no-fucose cultures, there was some variation in the maximum OD₆₂₀ reached before the onset of rapid lysis. This variation may reflect slight differences in the amount of MreCD accumulated before depletion. (B) Phase-contrast and fluorescent micrographs of cells sampled at the indicated times and examined directly or stained with fluorescent vancomycin (FL-V) as described in Materials and Methods. Bars indicate 2 μm. The experiment was performed independently at least three times with the same results. Similar results were obtained for the individual depletion of MreC or MreD in strain IU3927 or IU3925, respectively (see Fig. S3 and S4 in the supplemental material).

In contrast, merodiploid mutants depleted of MreC, MreD, and MreCD by shifting to medium lacking fucose abruptly stopped growing after ≈4 h, after which cell lysis occurred (Fig. 4; also see Fig. S3 and S4 in the supplemental material). At 3 h after the shift, cells depleted for MreCD (or MreC or MreD) formed chains of nearly spherical cells that stained with FL-V primarily at equators but not at septa (Fig. 4; also see Fig. S3 and S4). Cell lysis and debris were already visible after 3 h of depletion. At 5 h of MreCD (or MreC or MreD) depletion, the few remaining intact cells were largely spherical and showed aberrant FL-V staining. We conclude from these experiments that MreC and MreD are indeed essential for the growth and normal cell division of strain D39 and that the depletion of MreC, MreD, or both results in the formation of unstable spherical cells that rapidly lyse.

Null mutations in pbp1a suppress the requirement for pneumococcal mreCD in strain D39.

We tested the hypothesis that suppressor mutations were accumulating in ΔmreCD mutants by performing Illumina sequencing of the genomes of independent isolates that arose in transformations of strain D39 Δcps with the ΔmreCD<>P_c-erm amplicon (Table 1, row 1). We confirmed that the isolates contained the ΔmreCD<>P_c-erm mutation and lacked a second copy of mreCD⁺ that might have arisen by gene duplication (52). Suppressor mutations were mapped by cost-effective whole-genome sequencing (see Materials and Methods). Besides the ΔmreCD<>P_c-erm insertion, each isolate contained two point mutations of the D39 mreCD⁺ Δcps parent (Table 2). We focused on the pbp1a (Asp92Gly) mutation in isolate IU3897 (Table 2, row 1), because PBP1a is a prominent pneumococcal class A PG synthesis enzyme (see Introduction). The other suppressor isolate (IU3898) contained mutations in two genes of unknown function (Table 2, rows 3 and 4), which were not characterized further in this study.

Table 2.

Mutations in D39 Δcps ΔmreCD suppressor mutants determined by Illumina whole-genome sequencing^a

Gene containing indicated mutation	Function	Amino acid change
Isolate IU3897b (ΔmreCD)
1. spd_0336 (pbp1a) (GAC→GGC)	PBP1a; dual-function transglycosylase-transpeptidase	Asp92Gly
2. spd_1108 (GGU→AGU)	Hypothetical 5′-methylcytosine restriction system component protein	Gly274Ser
Isolate IU3898 (ΔmreCD)
3. spd_0177 (GGU→AGU)	Hypothetical M24 peptidase that hydrolyzes Xaa-Pro dipeptides	Gly290Ser
4. spd_1346 (UUA→UGA)	Hypothetical YceG-like family protein	Leu354Stop

^aIU3897 and IU3898 were isolated from the small number of colonies that arose following independent transformations of unencapsulated D39 Δcps strain IU1945 with the ΔmreCD<>P_c-erm amplicon as described for Table 1. Illumina whole-genome sequencing was performed as described in Materials and Methods. Besides the ΔmreCD<>P_c-erm insertion, the mutational changes shown are compared to the sequence of the IU1945 parent.

^bThe sequence determination of IU3897 contained a single, unresolved 43-bp gap (bp 1834486 to 1834528) in spd_1849 (putative SpoIIIJ-associated protein). Other results showed that the pbp1a (Asp92Gly) mutation was sufficient to allow growth of the ΔmreCD mutant (see Table 3), so the sequencing gap was not filled in.

We moved the pbp1a (Asp92Gly) allele into a fresh encapsulated D39 background as described in Materials and Methods (Table 3; also see Table S1 in the supplemental material). DNA sequencing confirmed that the resulting D39 rpsL1 pbp1a (Asp92Gly) mutant lacked the spd_1108 (Gly274Ser) mutation that also was present in the starting suppressed strain IU3897 (Table 2, row 2). Strikingly, ΔmreCD, ΔmreC, or ΔmreD amplicons more readily transformed the constructed pbp1a (Asp92Gly) mutant than its isogenic pbp1a⁺ parent (Table 3, rows 1, 4, 5, and 6). pbp1a (Asp92Gly) mutant cells lacked overt shape defects and were in somewhat shorter chains than those of the encapsulated D39 rpsL1 parent (Fig. 5 A). pbp1a (Asp92Gly) ΔmreCD (or ΔmreC or ΔmreD) mutant cells formed chains containing some aberrantly shaped cells compared to the shapes of pbp1a (Asp92Gly) mutant cells (Fig. 5A, arrows). These results indicate that the pbp1a (Asp92Gly) mutation is sufficient to suppress the lethality caused by the absence of MreC and MreD, although cell division is not entirely normal in the suppressed mutants.

Table 3.

Suppression of ΔmreC, ΔmreD, and ΔmreCD mutation lethality by the pbp1a (Asp92Gly) mutation in an encapsulated derivative of strain D39^a

Amplicon	No. of colonies in recipient strain
	pbp1a+	pbp1a (Asp92Gly) mutant
1. ΔmreCD<>Pc-erm	<20	>500
2. pcsB+-Pc-erm	>500	NTb
3. lctO::erm	>500	>500
4. ΔmreCD<>aad9	<20	>500
5. ΔmreC<>aad9	<20	>500
6. ΔmreD<>aad9	<20	>500
7. purR::aad9	>500	>500

^aThe pbp1a (Asp92Gly) allele was moved to encapsulated strain IU1781 (D39 rpsL1) to give strain IU4418 [D39 rpsL1 pbp1a (Asp92Gly)] (see Table S1 in the supplemental material). IU1781 and IU4418 were transformed as described in Materials and Methods and footnote a of Table 1.

^bNT, not tested.

Fig. 5.

Suppression of the requirement for MreC and MreD in the D39 genetic background by the pbp1a (Aps92Gly) point mutation (A) or by a Δpbp1a deletion (B). The pbp1a (Asp92Gly) mutation was selected as a spontaneous suppressor (Table 2) and then moved into a clean genetic background (Table 3; also see Table S1 in the supplemental material). Phase-contrast micrographs (A and B, top row) and fluorescent micrographs of cells stained with FL-V (bottom row of B) were taken as described in Materials and Methods for cells growing exponentially. Isogenic sets of strains containing the indicated mutations were constructed in the encapsulated IU1781 or unencapsulated IU1945 parent strain (see Table S1) and are shown. Representative micrographs are shown for each strain, and arrows mark some defective cells of the ΔmreC, ΔmreD, or ΔmreCD mutants. Bars indicate 2 μm. The experiment represented in panel A or B was performed twice or at least three times, respectively, with similar results.

The PBP1a (Asp92Gly) amino acid change is in the N-terminal transglycosylase (TG) domain, and some amino acid changes in the TG domain of PBP1a lead to the inactivation of both the transglycosylase and transpeptidase activities (59). We were unable to cross a Δpbp2a mutation into the pbp1a (Asp92Gly) mutant, confirming that the pbp1a (Asp92Gly) allele is a null mutation (data not shown). Consistent with this conclusion, a Δpbp1a deletion mutation also acted as a suppressor that abrogated the requirement for MreC and MreD in the D39 background (Table 4, rows 1, 4, and 5). We performed these studies in the unencapsulated D39 Δcps rpsL⁺ background, because the presence of capsule masks or alters some cell division defects in S. pneumoniae D39 strains (1, 2). D39 Δpbp1a Δcps mutants grew like the D39 pbp1a⁺ Δcps parent (data not shown) and formed diplococci that stained normally with FL-V (Fig. 5B), but unexpectedly they had significantly reduced widths and lengths compared to their parent, resulting in smaller cells with increased average aspect ratios (Fig. 6). The phenotypes of the Δpbp1a mutants were not caused by polarity, because pbp1a is transcribed oppositely to the next downstream gene (spd_0335), and the intercistronic region, including a putative factor-independent terminator, was maintained between pbp1a and spd_0335 in Δpbp1a mutations (see Table S1 in the supplemental material). In addition, a D39 pbp1a (Asp92Gly) Δcps mutant, which contains a nonpolar, single-base-pair change that inactivates PBP1a (Table 2), formed the same narrower and shorter cells as the D39 Δpbp1a Δcps mutant compared to the cells of the D39 pbp1a⁺ Δcps parent (data not shown).

Table 4.

Suppression of ΔmreC, ΔmreD, and ΔmreCD mutation lethality is specific for Δpbp1a mutations in the strain D39 background^a

Amplicon	No. of colonies in recipient strain
	D39 Δcps parent	Δpbp1a mutant	Δpbp2a mutant	Δpbp1b mutant
1. ΔmreCD<>Pc-erm	<20	>500	<20	<20
2. pcsB+-Pc-erm	>500	>500	NTb	NT
3. lctO::erm	>500	>500	>500	>500
4. ΔmreC<>aad9	<20	>300	None	None
5. ΔmreD<>aad9	<20	>300	None	None
6. purR::aad9	>500	>300	>200	>100

^aD39 Δcps (IU945), D39 Δcps Δpbp1a::kan-rpsL⁺ (K164), D39 Δcps Δpbp2a::kan-rpsL⁺ (K166), and D39 Δcps Δ pbp1b::kan-rpsL (K180) were transformed as described in Materials and Methods and footnote a of Table 1.

^bNT, not tested.

Fig. 6.

Δpbp1a mutants are shorter and narrower than their unencapsulated D39 parent or deletion mutants lacking the other two class A PBPs (PBP1b or PBP2a). The indicated strains were grown exponentially, phase-contrast micrographs were taken, and cell dimensions were measured as described in Materials and Methods. Measurements were confined to single cells in diplococci in later stages of division, where separation points were clearly visible, and only one cell in each diplococcus was measured. Standard errors of the means are indicated, and P values were determined relative to the parent strain by unpaired t tests using GraphPad Prism software. Bars indicate 2 μm. The experiment was performed three times.

In contrast to the Δpbp1a mutant, Δpbp1a ΔmreCD (or ΔmreC or ΔmreD) mutants formed short chains of cells containing misshaped cells, some of which stained aberrantly with FL-V. Like the pbp1a (Asp92Gly) mutation, the Δpbp1a deletion was sufficient to allow the growth of cells lacking MreC and MreD, although cell division was not fully restored. The formation of chains by unencapsulated pneumococcal strains is a common phenotype that we have observed for several mutants deficient in PG biosynthesis, including pcsB, Δpmp23, and ΔdacA mutants (1, 2). Finally, mutants lacking the other two class A PBPs (PBP2a and PBP1B) did not suppress the requirement for MreC and MreD (Table 4) and formed cells of dimensions similar to those of the pbp2a⁺ pbp1b⁺ parent strain (Fig. 6). Thus, the suppression of lethality of ΔmreCD mutations was confined to the absence of PBP1A among the class A PBPs.

We tested whether an allele difference in pbp1a underlies the requirement for MreC and MreD in strain D39 but not in laboratory strain R6. Among the many mutations acquired by R6, R6 protein Pbpla (Pbp1a^R6) contains two amino acid differences (Ala124 instead of Thr and Gln388 instead of Asp) compared to its D39 progenitor (30) or other serotype strains, such as TIGR4 (60). We reconstructed pbp1a^R6 in the R6 background and swapped it into a D39 cps⁺ rpsL1 strain. Conversely, we reconstructed the pbp1a gene of strain D39 (pbp1a^D39) in D39 cps⁺ rpsL1 and swapped it into the R6 background (Table 5; also see Table S1 in the supplemental material). The ΔmreCD amplicon again readily transformed the reconstructed R6 strain pbp1a^R6 but showed low transformation efficiency of the R6 strain containing the pbp1a^D39 allele along with suppressor accumulation (Table 5, row 1). We conclude that the pbp1a^R6 allele was required for the suppression of the requirement for MreC and MreD in the R6 genetic background. However, the converse was less clear in the D39 background, where the pbp1a^R6 allele led to more transformants than the reconstructed pbp1a^D39 allele (Table 5, row 1) but not nearly as many as for the control amplicon (Table 5, row 2) or in the R6 background. Taken together, these results suggest that the pbp1a^R6 allele and other mutations in the R6 background are required for the strong suppression of mreCD essentiality.

Table 5.

Relative transformation of R6 and D39 strains containing swapped pbp1a alleles^a

Amplicon	No. of colonies in recipient strain with indicated pbp1a allele
	R6 pbp1aR6	R6 pbp1aD39	D39 pbp1aR6	D39 pbp1aD39
1. ΔmreCD<>Pc-erm	>300	<20	∼60	∼20
2. pcsB+-Pc-erm	>500	>500	>500	>500

^aR6 rpsL1 pbp1a^R6 (IU4151), R6 rpsL1 pbp1a^D39 (IU4153), D39 rpsL1 pbp1a^R6 (IU4155), and D39 rpsL1 pbp1a^D39 (IU4157) were transformed as described in Materials and Methods and footnote a of Table 1. All strains were constructed by allele exchange of corresponding Δpbp1a<>kan-rpsL⁺ deletion mutants (see Table S1 in the supplemental material).

MreC is an abundant protein in S. pneumoniae.

As anticipated from its homologues in other bacterial species (27, 32), pneumococcal MreC fractionated with cell membranes (L.-T. Sham, unpublished). We quantified the amount of cellular MreC in laboratory strain R6 growing exponentially in BHI broth as described in Materials and Methods. Purified, soluble His₆-(N Δ32 amino acids)-MreC′, lacking its N-terminal membrane anchor, was used as a standard (1). Known concentrations of protein were spiked into extracts of an R6 ΔmreCD strain to generate a linear standard curve (see Fig. S5 in the supplemental material) in Western blots using anti-MreC polyclonal antibody. Amounts of MreC in comparable amounts of extracts of the R6 mreC⁺ parent strain were determined from the standard curve and converted to monomers per cell as described in Materials and Methods. This analysis showed that strain R6 cells contained ≈17,000 monomers (8,500 dimers) of MreC, indicating that MreC is a relatively abundant protein is S. pneumoniae.

MreC localizes to the equators and septa of dividing S. pneumoniae cells independently of MreD.

In rod-shaped cells, MreC forms filamentous lateral spirals and localizes to paraseptal rings at division septa (see Introduction) (15, 31, 67). The localization of MreC and MreD in ovococcus bacteria has never been reported. To localize MreC and MreD, we performed IFM as described in Materials and Methods. Initial attempts to localize MreC with anti-MreC polyclonal antibody did not yield a signal. Therefore, we used the approach of epitope tagging MreC and MreD with a peptide linker followed by three tandem copies of FLAG (L-FLAG³), which has worked well in locating other essential pneumococcal proteins by IFM (2, 69). We constructed unencapsulated R6 and D39 Δcps strains expressing MreC-(C)-L-FLAG³ from the native mreCD locus in the chromosome (Fig. 7) or under the control of the P_fcsK promoter in the ectopic bgaA site (see Table S1 in the supplemental material). We used unencapsulated strains for these studies, because capsule often masks cell division defects of mutants (1, 2). Growth, cell division, and cell shape were normal in D39 strains expressing MreC-L-FLAG³ from its chromosomal locus, indicating that the fusion did not significantly impair MreC or MreD functions (data not shown). We binned images of diplococci at different stages of division before observing fluorescence (Fig. 7).

Fig. 7.

MreC_Spn-L-FLAG³ localizes to the equators and septa of dividing pneumococcal cells. The upper diagram shows the chromosomal location of the L-FLAG³ epitope tag fused to the carboxyl terminus of MreC. The indicated P_c-erm cassette was used to select for the fusion construction in unencapsulated R6 and D39 background strains (see Table S1 in the supplemental material). The P_c-erm insertion did not cause any defects in growth or cell division in the D39 background, where MreC and MreD are essential (Fig. 4; also see Fig. S3 and S4). The middle panels show phase-contrast and pseudocolored fluorescent images of cells of R6-derived strain IU3238 (MreC-L-FLAG³) subjected to IFM as detailed in Materials and Methods. Before observing fluorescence, more than 100 cells were binned according to the stage of division, and representative phase and IFM images are shown for cells at each stage. The bottom diagram summarizes the location of MreC-L-FLAG³ (green dots) during the progression of the cell cycle, where opposed equatorial dots likely correspond to rings. The similar specific localization of MreC-L-FLAG³ was observed in at least five independent experiments using native or ectopic constructs expressing MreC-L-FLAG³ (see Results). Bars indicate 2 μm.

In IFM, MreC-L-FLAG³ located specifically to regions of PG synthesis at midcells, equators, and septa (Fig. 7). The same patterns were observed in unencapsulated R6 and D39 expressing MreC-L-FLAG³ from its native locus (Fig. 7 and data not shown) or an ectopic site in the presence of 0.8% (wt/vol) fucose in R6 mreC⁺//P_fcsK-mreC-L-FLAG³ merodiploid strains (data not shown). In control experiments, no fluorescence was observed in the isogenic parent strains lacking the MreC-L-FLAG³ construct or in merodiploid strains in the absence of fucose (data not shown), and only a single band corresponding to the expected molecular mass was detected on Western blots probed with anti-FLAG antibody of extracts of strains expressing MreC-L-FLAG³ (data not shown). Notably, the tight localization of MreC-L-FLAG³ to equators and septa closely matched the staining pattern of FL-V at different stages of cell division (Fig. 5B) (2, 42).

In the R6 background where MreD is not required (Table 1), we found that the tight localization of MreC-L-FLAG³ was not affected by a ΔmreD mutation (data not shown). This result suggests that MreC does not require MreD for localization in the R6 genetic background. We also localized MreD-L-FLAG³ in an unencapsulated D39 derivative (Fig. 8). The MreD-L-FLAG³ fusion expressed from its chromosomal locus again did not affect growth or cell division in the D39 background (compare Fig. 5B to 8) and gave a single band on Western blots probed with anti-FLAG antibody (data not shown). Similarly to MreC-L-FLAG³, MreD-L-FLAG³ localized primarily to sites of PG synthesis at the equators and septa of dividing cells (Fig. 8). However, MreD-L-FLAG³ was more dispersed than MreC-L-FLAG³, and about half of the cells with labeling at equators, septa, or both places had MreD-L-FLAG³ located elsewhere in the cells (Fig. 8, second and third rows). These results suggest that MreD localizes to the same division sites as MreC in S. pneumoniae, but unlike MreC, MreD may cycle to other sites during different stages of cell division.

Fig. 8.

MreD-L-FLAG³ localizes to the equators and septa of dividing cells and to other regions in some cells. The construct used to express mreD-L-FLAG³ from the chromosome of D39 unencapsulated derivative IU4868 is shown in the upper panel. The L-FLAG³-tag fusion and downstream P_c-erm marker did not affect growth or cell morphology. Phase-contrast and IFM images were taken as described for Fig. 7, and representative images are shown (bars, 2 μm). More than 100 single cells and diplococci were binned by overall length, and MreD-L-FLAG³ localization was determined for cells in each bin, where E, S, or E+S indicates equatorial, septal, or both equatorial and septal localization, respectively. For each length grouping, the number of cells with MreD-L-FLAG³ at locations other than equators and septa are indicated. The predominant division stage in each grouping is diagrammed. Micrographs in rows 2 and 3 show typical examples of some cells with MreD-L-FLAG³ at other locations. The experiment was performed independently three times with similar results.

ΔmreCD mutations exacerbate pcsB phenotypes in laboratory strain R6.

Given the conserved proximity of mreCD and pcsB in the pneumococcal chromosome and their roles in cell division (1, 2, 19, 42), we examined whether there was a functional relationship between mreCD and pcsB in the R6 genetic background. The mreCD⁺ pcsB⁺//P_c-pcsB⁺ merodiploid strain, which has a constitutively expressed ectopic copy of pcsB⁺, expresses about 30% more PcsB than the R6 parent strain (1). Both the R6 mreCD⁺ pcsB⁺//P_c-pcsB⁺ and R6 ΔmreCD pcsB⁺//P_c-pcsB⁺ strains had similar cell morphology and staining patterns with FL-V (Fig. 9 A) and showed similar sensitivity to penicillin G (Fig. 9B). Previous experiments showed that the absence of MreCD did not alter the cellular amount of PcsB protein (1). The underexpression of PcsB in the R6 mreCD⁺ ΔpcsB//P_c-pcsB⁺ strain to about 30% of its wild-type level resulted in slower growth (data not shown), chaining, cell compression to a more spherical shape, and prominent staining of equators with FL-V, but the increase in sensitivity to penicillin was marginal at most (Fig. 9) (1). Strikingly, the lack of MreCD combined with the underexpression of PcsB in R6 ΔmreCD ΔpcsB//P_c-pcsB⁺ led to slow growth and cell yield (data not shown), extended chaining, severe cell compression, aberrant staining patterns with FL-V, and extreme sensitivity to penicillin (Fig. 9). Control complementation experiments showed that the severe phenotypes of R6 ΔmreCD ΔpcsB//P_c-pcsB⁺ were reversed by expressing MreCD or PcsB from controlled promoters or another ectopic site (data not shown). Thus, although not essential in the R6 genetic background, the lack of MreC and MreD exacerbates cell division defects caused by the depletion of certain other essential proteins.

Fig. 9.

ΔmreCD deletion exacerbates pcsB underexpression phenotypes in laboratory strain R6. (A) Cultures of the indicated strains were sampled during exponential growth, and phase-contrast and fluorescent micrographs of cells stained with FL-V were taken as described in Materials and Methods. Strains IU1979 and IU2926 underexpress PcsB, and strains IU3140 and IU2926 are deleted for mreCD. Bars indicate 2 μm. (B) Antibiotic sensitivity tests for the addition of a final concentration of 0.01 μg penicillin G per ml were performed on the strains shown in panel A as described in Materials and Methods. The experiment was performed at least five times independently. Additional images with membrane stain FM-4-64 and the DNA label DAPI are shown in Fig. S6 in the supplemental material. See the text for additional details.

DISCUSSION

We report here the first characterization of the MreC and MreD cell division proteins in an ovococcus species. MreC and MreD from S. pneumoniae have several remarkable features compared to their homologues in rod-shaped bacterial species (3, 15, 27, 31, 32). Like MreC in E. coli and B. subtilis (27, 62), pneumococcal S. pneumoniae MreC (MreC_Spn) is an abundant membrane-bound protein of about 8,500 dimers per cell (see Fig. S5 in the supplemental material). Like MreC and MreD from rod-shaped bacteria, MreC_Spn and MreD_Spn are localized to areas of active PG synthesis that correlate to staining with FL-V (Fig. 7 and 8). However, unlike rod-shaped bacteria that organize lateral wall PG synthesis into spiral filaments, MreC_Spn and MreD_Spn localize to pneumococcal equators and septa. MreC_Spn localization was confined to these division sites at all stages of division (Fig. 7), and MreC_Spn still localized normally in the absence of MreD_Spn in a ΔmreD mutant of laboratory strain R6 (see Results). In contrast, MreD_Spn localized to equators and septa but also was detected at other sites in about half of the cells (Fig. 8). Taken together, these results suggest that MreC_Spn and MreD_Spn have different dynamics and independent assembly during pneumococcal cell division, analogous to the formation of the paraseptal rings in E. coli (67, 68). We were unable to detect any filament formation of MreC_Spn and MreD_Spn by this method, in contrast to the tight localization detected for MreC_Spn at the equators and septa of dividing cells (Fig. 7). Although we cannot rule out weak filament formation by MreC_Spn, these results suggest that filament formation is not an intrinsic property of MreC_Spn, as was assumed in interpreting the structure of the carboxyl-terminal coiled-coil domain of MreC_Spn (34).

Like rod-shaped bacteria, MreC_Spn and MreD_Spn are essential for pneumococcal growth in prototypical virulent serotype 2 strain D39 and its unencapsulated derivatives (Tables 1 and 3 and Fig. 4; also see Fig. S3 and S4 in the supplemental material). Consistent with this conclusion, mreC and mreD could not be knocked out in serotype 4 strain TIGR4 by transposon mutagenesis (63). The depletion of MreCD, MreC, or MreD caused cells to stop growing, become spherical, form chains, and lyse (Fig. 4; also see Fig. S3 and S4). As cells became spherical, aberrant patterns of staining with FL-V appeared. In contrast, MreC_Spn and MreD_Spn were not essential for the growth of laboratory strain R6 (Table 1 and Fig. 3), which contains numerous mutational changes from its D39 progenitor strain (30). This difference in essentiality for mreCD between D39 and TIGR4 versus R6 strains likely is due to the accumulation of suppressor mutations, which can be readily selected in D39 ΔmreCD mutants (Tables 1 and 2). Whole-cell genome sequencing of two independent ΔmreCD suppressor mutants uncovered a new genetic relationship between PBP1a and MreCD in S. pneumoniae, which was analyzed further in this study (Tables 3 to 5). In addition, sequencing revealed that the suppression of mreCD essentiality can occur by at least one mechanism other than mutations in pbp1a (Table 2, rows 3 and 4). This second mechanism involves mutations in one or two genes (spd_0177 and spd_1346) encoding hypothetical proteins of unknown functions. Based on precedents from E. coli (3), the identification and characterization of these and additional ΔmreCD bypass suppressor mutations likely will increase the understanding of PG synthesis in S. pneumoniae and other species.

Amino acid changes that inactivated PBP1a, such as Asp92Gly, and the absence of PBP1a suppressed the requirement for MreC and MreD in the D39 background (Tables 3 and 4 and Fig. 5). Thus, MreC and MreD are required for the growth of D39 strains when PBP1a is functional but are not required when PBP1a is inactivated or absent. However, the amino acid changes in PBP1a^R6 were not sufficient to account fully for the nonessentiality of mreCD in strain R6 compared to its importance in D39 (Table 5), suggesting that other mutations in the R6 background contribute to the suppression of the ΔmreCD requirement. On the other hand, swapping the pbp1a^D39 and pbp1a^R6 alleles in strain R6 restored the requirement for MreC and MreD in the R6 background (Table 5), which is consistent with the conclusion that there is a reinforcing genetic link between pbp1a and mreCD in S. pneumoniae.

PBP1a is a major class A bifunctional enzyme and the only pneumococcal class A enzyme implicated in the development of high levels of β-lactam resistance (10, 39, 56). Unlike PBP1a, the absence of the other pneumococcal class A enzymes, PBP2a and PBP1b, did not relieve the requirement for MreC and MreD (Table 4). PBP1a and PBP2a show a synthetic lethal relationship, in that pbp1a pbp2a double mutants are not viable (24, 45). However, the different suppression properties of Δpbp1a and Δpbp2a mutants (Table 4) indicate that this synthetic lethality is not due to simple functional redundancy. Consistent with this interpretation, we observed that the Δpbp1a deletion markedly reduced cell size and increased aspect ratios by reducing cell diameters of an unencapsulated derivative of D39 (Fig. 6). In contrast, Δpbp2a or Δpbp1b mutations did not significantly change cell dimensions (Fig. 6). The reduction in cell size caused by Δpbp1a was more apparent in the unencapsulated than in encapsulated D39 strains (compare Fig. 5A to 6), probably because of the tendency of the pneumococcal capsule to mask or alter cell division defects (1, 2). The smaller size of Δpbp1a mutants compared to that of the parent was not observed previously in the R6 background (24, 45), possibly due to the numerous differences in PG biosynthesis between the R6 and D39 genetic backgrounds, including the suppression of ΔmreCD deletions reported here and the significantly different PG peptide cross-bridge composition reported previously (1).

The phenotypes discussed above support a role for MreC, MreD, and PBP1a in peripheral PG synthesis (Fig. 1). Our current working model is that the pneumococcal peripheral PG synthesis complex contains MreC, MreD, and PBP1a, along with several other proteins suggested by previous work in pneumococcus and other bacteria, including class B PBP2b_Spn, RodA_Spn, shuttle protein GpsB_Spn, and another pneumococcal class A enzyme (see Introduction) (9, 46, 72). According to this model, the depletion of MreC, MreD, or both would decrease peripheral PG synthesis and lead to spherical cell formation (Fig. 4; also see Fig. S3 and S4 in the supplemental material) (31, 46). The marked reduction in width of Δpbp1a mutant cells (Fig. 6) is also consistent with reduced peripheral PG synthesis, which has been implicated in setting bacterial cell diameters by a mechanism that is largely unknown (9, 41, 71). The unusual genetic relationship reported here between MreCD and PBP1a (Tables 1, 3, and 4) would result if MreC and MreD are required for correct PBP1a localization, activity, or both in peripheral PG synthesis. The depletion of MreCD would not only diminish peripheral PG synthesis but also lead to aberrant PBP1a localization and activity that causes rapid cell lysis (Fig. 4). For example, PBP1a may shuttle between peripheral and septal PG synthesis, as has been proposed for its PBP1 homologue in B. subtilis (9, 26). The absence of S. pneumoniae MreCD (MreCD_Spn) may disrupt this process such that the absence of PBP1a obviates the requirement for MreC and MreD. Thus, mreCD_Spn are genetically essential, whereas pbp1a_Spn is not, but aberrant Pbp1a localization, activity, or both may underlie the lethality of mreCD_Spn mutations.

A somewhat parallel genetic relationship was observed in B. subtilis, where the inactivation of PBP1 suppressed the requirement for the MreB homologue (26). The deleterious mislocalization of PBP1 in the absence of MreB was invoked to explain this genetic relationship, and it was suspected that MreC, which interacts with MreB and PBP1 in B. subtilis, also is involved in the proper localization of PBP1 (26). In S. pneumoniae, the viable Δpbp1a ΔmreCD mutants form chains containing some cells with division defects (Fig. 5), implying that the remaining class A PBPs can carry out division sufficiently to allow remarkably normal growth. Aspects of this working model currently are being tested. Preliminary experiments indicate that other proteins that may mediate pneumococcal peripheral PG synthesis are essential for growth, including RodA_Spn and GpsB_Spn, in the presence or absence of PBP1a_Spn (data not shown). The abundance, essentiality, and surface location of conserved PG synthesis complex proteins like MreC suggest that they could be potential targets fot the development of pneumococcal vaccines.

Curiously, MreC and MreD are present in coccus species, such as S. aureus, that encode a single class A PBP and are not thought to carry out peripheral PG synthesis (57, 72). Notably, the mreCD genes are not essential in S. aureus (7), unlike in ovococcus S. pneumoniae (Table 1 and Fig. 4; also see Fig. S3 and S4 in the supplemental material). Conversely, MreC and MreD are absent from some ovococcus species, especially in the pyogenic group that includes S. pyogenes and S. agalactiae (35, 72). This absence raises the question of whether proteins other than MreC and MreD homologues organize peripheral PG synthesis in these species. There are several differences in the composition of the PG biosynthetic apparatus between these species and S. pneumoniae. For example, S. pneumoniae encodes a single low-molecular-weight d,d-carboxypeptidase PBP (DacA), whereas S. pyogenes and S. agalactiae encode several low-molecular-weight PBPs (35, 72). The modification of PG precursor substrates by low-molecular-weight PBPs has been implicated in localizing the PG synthetic apparatus (2, 8, 38, 40, 48).

Although MreC and MreD also are not required in the pneumococcus R6 genetic background (Tables 1 and 5 and Fig. 3), their absence still greatly exaggerated the cell division defects caused by the underexpression of the PcsB protein (Fig. 9). When present in ovococcus species, mreCD are always located upstream from pcsB, which is regulated by the WalRK_Spn (VicRK) two-component system from the P_pcsB promoter between mreCD and pcsB (Fig. 2) (1, 42, 43). We could detect a mreCD-pcsB cotranscript by sensitive RT-PCR methods (see Fig. S2 in the supplemental material), but the relative amount and significance, if any, of this cotranscript remain unknown for two reasons. First, there is considerable independent transcription of pcsB from the WalR_Spn-dependent P_pcsB promoter (Fig. 2), which is consistent with previous results from expression experiments (43). Second, despite the abundance of MreC, we were unable to detect an mreC transcript on Northern blots (see Results). This negative result could mean that the mreCD transcript is unstable or that mreCD is contained in a long transcript that did not transfer well. The latter interpretation is suggested by recent tiling array analyses indicating that mreCD are the last two genes in a nine-gene cotranscript that does not extend into pcsB (29).

Monocistronic expression was previously reported for gbpB_Smu, which is the pcsB homologue in S. mutans (37). Recently, it was proposed that GbpB_Smu is not a cell division protein at all but rather coordinates exopolysaccharide biofilm formation with cell growth and surface biogenesis (14). It is difficult to reconcile the exaggeration of division defects caused by PcsB underexpression in ΔmreCD mutants with this model, especially since the R6 strain does not produce the capsule exopolysaccharide, which is the only one known to be produced by D39-derived pneumococcal strains. New results to be reported elsewhere show that PcsB interacts with an essential membrane-bound cell division protein that may control the PG hydrolytic activity of PcsB (L.-T. Sham, unpublished).