PcsB is a protein of unknown function(s) that influences the cell morphology of several pathogenic species of streptococcus. PcsB contains a CHAP (cysteine, histidine-dependent amidohydrolase/peptidase) domain found in bacterial murein hydrolases; however, direct links between steps in cell wall biosynthesis and PcsB function(s) have not been demonstrated. We show here that pcsB is essential in the human respiratory pathogen, Streptococcus pneumoniae, that depletion of PcsB is bacteriostatic and that alanine substitutions in the conserved cysteine and histidine residues of the CHAP domain appear to be lethal. We stained wild-type parent and mutant bacteria deficient in expression of PcsB with fluorescent vancomycin and DAPI to determine patterns of cell wall synthesis and nucleoid segregation respectively. The wild-type parent strain exhibited ordered, simultaneous septal and equatorial cell wall synthesis. In contrast, reduced expression of PcsB resulted in formation of long chains of cells in which peptidoglycan synthesis occurred at nearly every division septum and cell equator. Severe depletion of PcsB led to abnormal, uncontrolled cell wall synthesis at misplaced septa and around large cells. Together, these physiological properties are consistent with a role for PcsB as a murein hydrolase that balances the extent of cell wall synthesis in S. pneumoniae. Finally, we show that the defects in morphology and cell wall synthesis that result from depletion of PcsB strongly resemble those caused by depletion of the essential VicRK two component regulatory system (TCS). This result and the essentiality of pcsB support the hypothesis that the essentiality of the VicRK TCS results from its positive regulation of PcsB expression.
Two component signal transduction systems (TCSs) probably play important roles in regulating genes required for successful colonization and infection by human opportunistic pathogens, such as Streptococcus pneumoniae (Lau et al., 2001; Hava and Camilli, 2002; Hubbard et al., 2003; Kadioglu et al., 2003). Recently, we reported an unusual regulatory link between the essential VicRK TCS of S. pneumoniae and cell wall biosynthesis (Ng et al., 2003). Microarray analysis of S. pneumoniae depleted for expression of VicRK showed large decreases in the amounts of transcripts of four genes [pcsB (spr2021), spr0096, spr1875 and lytB] that are potentially involved in cell wall hydrolysis and the regulation of cell wall biosynthesis (Ng et al., 2003). Of these four VicRK-regulated genes, we initially focused on pcsB(Fig. 1A), because reduced expression of the VicRK TCS resulted in defects in cell shape, size and morphology that resembled those reported previously for mutants of homologues of pcsB in other species of streptococcus (Chia et al., 2001; Reinscheid et al., 2001; 2003). Indeed, constitutive expression of pneumococcal pcsB+ suppressed the essential requirement for the VicRK TCS and allowed the isolation of vicR null mutants (Ng et al., 2003). This finding is consistent with the hypothesis that the requirement for the VicRK TCS is partly or wholly caused by its positive regulation of pcsB expression (Ng et al., 2003).
The function(s) of PcsB is/are not known in any species of streptococcus. The S. pneumoniae PcsB protein and its homologues from S. mutans and S. agalactiae (group B streptococcus) can be divided into four domains (Fig. 2). The amino-terminus contains a secretory signal peptide. A large amino-terminal domain of about 200 amino acids follows that contains a leucine zipper motif (Mattos-Graner et al., 2001), but lacks other known functional motifs. Then there is a spacer of variable length, with that of S. agalactiae and S. pneumoniae the largest and smallest respectively. Finally, the carboxyl-terminal domain of about 120 amino acids contains the CHAP (cysteine, histidine-dependent amidohydrolase/peptidase) motif found in peptidoglycan hydrolases (Anantharaman and Aravind, 2003; Bateman and Rawlings, 2003; Rigden et al., 2003). PcsB lacks known attachment motifs found in Gram-positive bacteria (Navarre and Schneewind, 1999), but is associated with the cell walls of S. mutans and S. agalactiae (Chia et al., 2001; Reinscheid et al., 2003). A significant portion of PcsB is secreted into the growth media of cultures of S. mutans and S. agalactiae (Chia et al., 2001; Mattos-Graner et al., 2001; Reinscheid et al., 2001; 2003), and S. mutans PcsB is an immunodominant antigen in humans (Chia et al., 2001). pcsB mutants show three prominent shared phenotypes in S. mutans and S. agalactiae (Chia et al., 2001; Reinscheid et al., 2001; 2003). Although pcsB is not essential, S. mutans and S. agalactiae pcsB mutants grow slowly (Chia et al., 2001; Reinscheid et al., 2001), and the S. agalactiae mutant requires the osmoprotectant sorbitol (Reinscheid et al., 2001). pcsB mutants in both species of streptococcus are hypersensitive to several stress conditions (Chia et al., 2001; Reinscheid et al., 2001; 2003). Of most note, S. mutans pcsB mutants form pleiomorphic cell shapes with multiple in-growths of the cell wall, whereas S. agalactiae pcsB mutants grow in clumps instead of chains with cell separation in several planes and multiple division septa within single cells.
Despite these suggestive morphological defects, no experimental evidence currently links PcsB function to steps in peptidoglycan biosynthesis. Cell wall hydrolase activity was not detected for purified S. agalactiae PcsB (Reinscheid et al., 2001), and the peptidoglycan composition is not changed in an S. agalactiae mutant compared with its wild-type parent (Reinscheid et al., 2003). A purified homologue of PcsB from Enterococcus faecium also lacked detectable cell wall hydrolytic activity (Teng et al., 2003). In S. mutans pcsB mutants, the cell walls seemed to be fragile and in-growths of cell wall material seemed to be at locations other than normal septa, leading to the suggestion that a stress response occurred instead of a defect in cell wall biosynthesis per se (Chia et al., 2001). Moreover, PcsB homologues seem to play other roles in S. mutans and related bacteria. For example, the PcsB homologue of S. mutans acts as a glucan-binding protein (Mattos-Graner et al., 2001) and that of E. faecium avidly binds to eukaryotic extracellular matrix proteins (Teng et al., 2003).
We performed the experiments reported herein to address several key issues about the function and regulation of PcsB in S. pneumoniae. We report that pcsB is strictly essential in S. pneumoniae and the conserved cysteine and histidine residues in the CHAP domain of PcsB are required for its function. This result and staining of mutants expressing varying amounts of PcsB with fluorescent vancomycin (Fl-Van), which detects regions of active cell wall synthesis, support the hypothesis that PcsB acts as a hydrolase in S. pneumoniae. Finally, we demonstrate that the defects in cell morphology caused by depletion of the VicRK TCS are similar to those caused by depletion of PcsB, thereby supporting the idea that the essentiality of the VicRK TCS results from its strong positive regulation of PcsB.
pcsB is essential for growth of S. pneumoniae
Homologues of pneumococcal pcsB are essential in E. faecium (sagA; Teng et al., 2003), conditionally essential in S. agalactiae (pcsB; Reinscheid et al., 2001), but not essential in S. mutans (gsp-781; Chia et al., 2001) and Staphylococcus aureus (ssa; Martin et al., 2002). In S. pneumoniae, we tried several times to delete pcsB and replace it with Pc -ermAM (erythromycin resistance gene driven by the constitutive synthetic Pc promoter; Lee and Morrison, 1999) by transformation with linear polymerase chain reaction (PCR) amplicons (Fig. 1C; Experimental procedures). Transformation reactions were spread onto TSA II blood agar plates in the absence or presence of 0.5 M of the osmoprotectant sorbitol, which is required for growth of an S. agalactiaeΔpcsB mutant (Reinscheid et al., 2001; 2003). We were unable to recover any deletion-replacement mutants of pcsB in S. pneumoniae R6 (data not shown; Ng et al., 2003). We next replaced the native promoter and ribosome-binding site (RBS) of the chromosomal pcsB gene with the fucose-inducible promoter PfcsK (Chan et al., 2003) and the RBS from fcsK (fuculose kinase) (strain IU1545; Table 1; Fig. 1D). There was a critical concentration of PcsB protein required for growth of this strain. Strain IU1545 failed to grow in BHI medium containing 0.02% (wt/vol) fucose, whereas it grew with nearly the same doubling times in BHI containing concentrations of fucose greater than 0.05% (wt/vol) (Fig. 3). Concentrations of sorbitol ranging from 0.1 to 0.5 M could not bypass the requirement for fucose (data not shown). Finally, we found that we could delete the essential vicR response regulator gene in strain IU1545 (PfcsK-pcsB+) in the presence of 0.2% (wt/vol) fucose (data not shown; strain IU1602; Table 1). This result confirms our previous conclusion that constitutive expression of PcsB suppresses the essential requirement for the VicRK TCS (Ng et al., 2003).
Strains were constructed by transformation of indicated recipient with linear double-stranded synthetic PCR amplicon DNA. After single-colony isolation, all strain constructions were confirmed by PCR analysis of chromosomal DNA using flanking primers (see ‘Experimental procedures’ and ‘Supplemental data’ at http://sunflower.bio.indiana.edu/~mwinkler/index.htm).
<> indicates total replacement of the reading frame of a gene by the indicated antibiotic resistance marker.
SpcR, resistant to spectinomycin; ErmR, resistant to erythromycin; KanR, resistant to kanamycin; TetR, resistant to tetracycline.
EL1404 transformed with linear bgaA::[kan-t1t2-Pc]- sag0017aa1−25/pcsBaa28−392 amplicon
EL59 transformed with linear PCR amplicon bgaA::[kan-t1t2-PfcsK]-pcsB+
IU1599 transformed with linear PCR amplicon ΔpcsB < > Pc-ermAM
IU1545 transformed with linear PCR amplicon ΔvicR::Pc-ermAM
EL59 transformed with linear bgaA::[kan-t1t2-Pc]-smu.22 + amplicon
EL776 transformed with linear bgaA::[kan-t1t2-Pc]-smu.22 + amplicon
KanR ErmR SpcR
EL1221 transformed with linear bgaA::[kan-t1t2-Pc]-smu.22 + amplicon
KanR ErmR TetR
EL1404 transformed with linear bgaA::[kan-t1t2-Pc]-smu.22 + amplicon
KanR ErmR SpcR
EL1221 transformed with linear bgaA::[kan-t1t2-Pc]-pcsB + amplicon
KanR ErmR TetR
EL1404 transformed with linear bgaA::[kan-t1t2-Pc]-pcsB + amplicon
EL59 transformed with linear PCR amplicon pcsB+-Pc-ermAM-rpsB +
EL776 pcsB+bgaA::Pc-sag0017 +
EL776 transformed with linear bgaA::[kan-t1t2-Pc]- sag0017 + amplicon
Putative transcription termination structures surround pcsB (Fig. 1A), and our previous microarray analyses did not indicate co-transcription of pcsB with mreCD or rpsB under the conditions tested (Ng et al., 2003). To further rule out possible polar effects of expression from the PfcsK promoter on downstream expression of rpsB, which encodes a ribosomal protein that is probably essential, we constructed merodiploid strains expressing ectopic copies of pcsB+ inserted into the dispensable bgaA chromosomal gene (Fig. 1E and F). This extra copy of pcsB+ was under control of either the constitutive Pc promoter (Fig. 1E; strain IU1533; Table 1) or the fucose-inducible PfcsK promoter (Fig. 1F; strain IU1599; Table 1). Now we were able to delete and replace the native pcsB+ gene in strains IU1533 and IU1599 by transformation with a linear PCR amplicon containing Pc-ermAM (Fig. 1C) (strains IU1547 and IU1600 respectively; Table 1), where 0.2% (wt/vol) fucose was included in all transformation and selection steps for strain IU1599. Like strain IU1545 (Fig. 3), strain IU1600 showed an absolute requirement for concentrations of at least 0.05% (wt/vol) fucose for growth (data not shown). This result confirms that pneumococcal pcsB is indispensable for growth, independent of the downstream rpsB gene. Finally, the dependence of strain IU1600 on fucose refutes the possible explanation that transcription in the opposite direction from the Pkan-kan marker into the mreCD operon was responsible for the essentiality of pcsB in strains EL1454 and IU1545 (Fig. 1B and D respectively; Table 1).
Depletion of PcsB is bacteriostatic and reversible
We further investigated the physiological consequences of depletion of PcsB in pneumococcus. Strain IU1545, which has PcsB expression under control of the fucose-inducible PfcsK promoter (Fig. 1D; Table 1), was grown in BHI containing 0.2% (wt/vol) fucose to mid-exponential phase [OD620 (13 mm) ≈0.1; ≈5 × 107 colony-forming units (CFU) per ml], at which point the culture was harvested by centrifugation, washed in BHI and resuspended in BHI lacking or containing 0.2% (wt/vol) fucose (see Experimental procedures). The culture containing fucose continued to grow as indicated by increases in optical density and CFU until a stationary phase plateau was reached (Fig. 4). In contrast, the culture lacking fucose rapidly stopped growing, and the optical density and CFU, which can reflect the presence of diplococci and chains of cells (see Discussion), remained relatively constant for 6 h after removal of fucose (Fig. 4). Addition of 0.2% (wt/vol) fucose to cultures resulted in a small, but reproducible decrease in optical density lasting about 1 h, whereupon cultures resumed growth (Fig. 4). These data indicate that depletion of PcsB is bacteriostatic and reversible in pneumococcus within the times used in these experiments.
Point mutations that change the conserved residues in the CHAP domain of pneumococcal PcsB are lethal
We tested whether the conserved cysteine (C292) or histidine (H343) residue in the CHAP domain of PcsB (Fig. 2) is required for viability of S. pneumoniae. Mutagenic PCR amplicons that changed C292 or H343 to alanine (A) in PcsB were fused to a downstream Pc-ermAM marker (Fig. 1G) and used to transform pcsB+ parent strain R6 or the pcsB+/bgaA::PfcsK-pcsB+ merodiploid strain IU1599 in the presence of 0.2% (wt/vol) fucose (see Experimental procedures). Single colonies of erythromycin resistant transformants were isolated on TSAII blood agar plates containing 0.2% (wt/vol) fucose and 0.3 µg erythromycin per ml (see Experimental procedures), and DNA prepared from lysed single colonies was sequenced to test for the presence of the C292A or H343A point mutation (see Experimental procedures).
We did not recover the C292A or H343A point mutation in 28 individual transformants of the parent R6 strain from three independent experiments. However, we did recover both CHAP domain point mutations at the native pcsB locus in the IU1599 merodiploid strain. In these transformations, small and large colonies were recovered. Sequence analysis showed that the large colonies contained wild-type pcsB+, whereas the small colonies contained the C292A or H343A pcsB mutation, possibly suggesting a dominant-negative phenotype. These CHAP mutants could be single colony isolated at least twice on TSAII blood agar plates. However, the pcsB(C292A)/bgaA::PfcsK-pcsB+ or pcsB(H343A)/bgaA::PfcsK-pcsB+ mutants could not be propagated in liquid BHI containing 0.2% (wt/vol) fucose or as confluent regions of growth on TSAII blood agar plates containing 0.2% (wt/vol) fucose. During propagation, the C292A or H343A mutations reverted back to the wild-type sequences encoding cysteine or histidine residues, respectively, possibly by recombination with the PfcsK-pcsB+ gene in the merodiploid. Thus, there is strong selective pressure against maintaining the CHAP domain mutations in the merodiploid strains. Although the merodiploid strains containing the CHAP mutations could not be characterized further, our combined data support the hypothesis that the conserved cysteine and histidine residues in the PcsB CHAP domain are required for PcsB function, stability, or both.
Ordered, simultaneous equatorial and septal cell wall synthesis in wild-type parent strain R6
If PcsB is an essential murein hydrolase, then patterns of cell wall synthesis might be altered in bacteria expressing different amounts of PcsB. To test this hypothesis, we stained cells with Fl-Van, which Errington and co-workers have developed as a tool to visualize regions of newly synthesized peptidoglycan in Gram-positive bacteria (Jones et al., 2001; Pinho and Errington, 2003). Fl-Van binds to the d-alanine-d-alanine residues at the carboxyl-termini of peptidoglycan precursors (Jones et al., 2001; Pinho and Errington, 2003), which are concentrated in regions of newly synthesized cell wall (Jones et al., 2001; Pinho and Errington, 2003). Fl-Van can also bind to muropeptides containing d-alanine-d-alanine residues incorporated into mature cell walls, such as those of S. aureus (Pinho and Errington, 2003).
To use Fl-Van to localize regions of cell wall synthesis in S. pneumoniae, we first needed to establish staining patterns of the wild-type parent strain R6. Unlike S. aureus, the mature peptidoglycan of S. pneumoniae R6 contains a very low content of d-alanine-d-alanine muropeptides (Garcia-Bustos et al., 1987). Therefore, to detect specific staining of regions of newly synthesized cell wall (Fig. 5), we did not need to pregrow S. pneumoniae R6 cells in medium containing d-serine, which replaces terminal d-alanine-D-alanine with d-alanine-d-serine residues in the muropeptides of the mature peptidoglycan (see Pinho and Errington, 2003). The d-alanine-d-alanine residues present in the mature peptidoglycan seemed to be evenly distributed along the surface of cells and caused very light background staining (Fig. 5). Moreover, pregrowth in medium containing d-serine did not change equatorial and septal staining patterns of the wild-type (Fig. 5) or mutant bacteria (see below; data not shown).
Fluorescent vancomycin stained two discrete regions of cells of parent strain R6 growing exponentially in BHI medium (Experimental procedures). Ovoid cells that appeared about to divide contained long bands of equatorial staining and minimal separation of nucleoids, which were stained with DAPI (Fig. 5A). Cells further along in division were more elongated, showed a single band of septal staining on what had been the equator and contained completely separated nucleoids (Fig. 5B). As constriction at septa became more pronounced, simultaneous parallel septal and equatorial staining by Fl-Van was detected (Fig. 5C and D). As constriction at the septum progressed, the region of septal staining with Fl-Van continued to shrink (Fig. 5D) until the cells had divided but were still attached (Fig. 5E). These cells retained the equatorial Fl-Van staining (Fig. 5E).
Underexpression of PcsB leads to formation of long chains of cells and excessive septal and equatorial cell wall synthesis
We reported previously that constitutive expression of pcsB led to the formation of long chains of cells (Fig. 6A; Ng et al., 2003). Chain formation is a common phenotype of murein hydrolase mutants (see, e.g. De Las Rivas et al., 2002; Heidrich et al., 2002). In these experiments, we had replaced the chromosomal pcsB promoter and RBS with the synthetic Pc promoter and the RBS of the ermAM gene (Fig. 1B; Claverys et al., 1995; Lee and Morrison, 1999). Cells in these long chains were generally ovoid, but somewhat irregularly shaped (Fig. 6A). Staining of bacteria constitutively expressing PcsB from the Pc promoter (strain EL1454; Pc-pcsB+Table 1) with Fl-Van revealed parallel bands at most septal and equatorial regions (Fig. 6B) and co-staining with DAPI indicated discrete separated nucleoids (Fig. 6C). In fact, the most prominent staining suggested that the chains were composed of pairs of late divisional cells, such as those shown in Fig. 5C and D for the parent strain R6. Interestingly, pairs or short chains of two or three pairs of these late divisional cells, but not long chains, predominated in BHI culture medium supplemented with 0.062 M d-serine (data not shown; see Discussion). It is noteworthy that the Fl-Van staining remained confined to the equatorial and septal regions and did not appear generally along the cell walls of this mutant as it grew (Fig. 6B). This staining specificity suggests that Fl-Van continued to indicate regions of active cell wall synthesis, rather than general aberrant incorporation of d-alanine-d-alanine residues into the muropeptides of mature cell walls.
We hypothesized that this chain formation and ‘zebra’ staining pattern reflected decreased expression of PcsB from the Pc promoter compared to that from the native PpcsB promoter. Antibody is not yet available to quantify pneumococcal PcsB directly by Western blotting. To test this hypothesis, we examined the cell morphology of strains expressing low and high amounts of PcsB from the fucose-inducible promoter PfcsK. Strain IU1545 (PfcsK-pcsB+Fig. 1; Table 1) grown in medium containing the minimum concentration of fucose tested that supported growth [0.05% (wt/vol); Fig. 3] formed a mixture of short and long chains (data not shown). These chains exhibited a ‘zebra’ staining pattern with Fl-Van similar to that of strain EL1454 (Fig. 6). In contrast, strain IU1545 grown in excess [0.2% (wt/vol)] fucose was largely diplococcal (Fig. 7A), similar to parent strain R6 (Fig. 5). Additional merodiploid constructs containing Pc-pcsB+ or PfcsK-pcsB+ inserted ectopically into the dispensable bgaA locus (Zahner and Hakenbeck, 2000) supported the conclusion that underexpression of PcsB leads to the formation of chains of cells in which peptidoglycan synthesis occurs at most equatorial and septal regions (data not shown). Results with these ectopic constructs in the bgaA locus also supported the notion that the phenotypes observed did not result from disturbing expression of the upstream mreCD operon or downstream rpsB gene at the pcsB locus (Fig. 1A; data not shown).
Severe depletion of PcsB results in aberrant cell morphology and aberrant cell wall synthesis
The morphology and Fl-Van and DAPI staining patterns of cells severely depleted for PcsB also supported the idea that PcsB plays a direct role in modulating cell wall synthesis, possibly as a murine hydrolase. Expression of PcsB from the fucose-inducible PfcsK promoter was abruptly turned off by shifting strain IU1545 (PfcsK-pcsB+; Fig. 1; Table 1) to medium lacking fucose (Fig. 4). Severe depletion of PcsB caused an assortment of aberrant cell shapes, including large circular cells, clusters of tightly grouped cells and kinked chains of irregularly shaped cells (phase contrast; Fig. 7B). The large round cells, which were usually still connected to smaller cells, stained with Fl-Van over their entire surfaces (Fig. 7B), and the entire cytoplasm was stained with DAPI (Fig. 7B). In contrast, in the clumps of cells and irregular chains, Fl-Van staining remained localized and indicated irregular, aberrant regions of cell wall synthesis (Fig. 7B). In these cells, planes of cell wall synthesis were no longer ordered and parallel as in parent strain R6 (Fig. 5). Instead, regions of cell wall synthesis radiated in multiple planes from fixed foci and were present at irregular angles in adjoining cells (Fig. 7B).
Given the complex mechanisms that coordinate peptidoglycan biosynthesis and cell division (see, e.g. Errington, 2003; Morlot et al., 2003; Pinho and Errington, 2003), some level of aberrant nucleoid segregation would not be surprising in cells depleted for PcsB. However, DAPI staining showed that most of these irregularly shaped cells still retained discrete nucleoids, even in small, constricted cells (Fig. 7B). We have no way of knowing whether these nucleoids contained complete chromosomes, and in some cases, we did observe loss of nucleoids (Fig. 7B). Finally, the resumption of growth of strain IU1545 upon fucose addition after 2 h of PcsB depletion (Fig. 4) was accompanied by the reappearance of normal looking cells after 2 h, which was the first time point examined (Fig. 7C).
PcsB homologues from S. mutans and S. agalactiae do not replace pneumococcal PcsB functions
Blast analysis of 20 complete and unfinished genome sequences of different species of streptococcus showed that each genome contains one gene whose product displays a high degree of sequence similarity to pneumococcal PcsB over its whole length (unpubl. obs.). We tested whether the closest homologues of S. pneumoniae PcsB from S. mutans (SMU.22; Gsp-781) or S. agalactiae (SAG0017) could replace pcsB function in S. pneumoniae. We constructed a set of merodiploid strains in which the native pcsB+ gene remained intact and ectopic copies of S. pneumoniae pcsB+, S. agalactiae pcsB (sag0017 ) or S. mutans gsp-781 (smu.22) were placed under control of the Pc promoter in the dispensable bgaA locus of S. pneumoniae R6 (strains IU1533, IU1554 and IU1606 respectively; Table 1). As expected from our previous results (Ng et al., 2003), constitutive expression of S. pneumoniae pcsB+ from the Pc promoter in the merodiploid strain allowed significantly better growth of bacteria depleted for vicRKX expression (strain IU1549) than bacteria containing only the native pcsB+ gene (strain EL776; data not shown; Ng et al., 2003). The severe growth defects caused by depletion of vicR expression in ΔvicK or ΔvicX mutants (Ng et al., 2003) were also suppressed by constitutive expression of the ectopic copy of pneumococcal pcsB+ (strains IU1616 and IU1617; Table 1; data not shown). Finally, the native pcsB+ gene could be deleted in the merodiploid strains expressing pneumococcal pcsB+ from the bgaA locus [stains IU1547 (Pc) and IU1600 (PfcsK), Table 1; see above].
In contrast, constitutively expressed S. agalactiae pcsB+ (sag0017) or S. mutans gsp-781+ (smu.22) did not function well in S. pneumoniae (data not shown). S. agalactiae pcsB + (sag0017) or S. mutans gsp-781+ (smu.22) was cloned under the control of the constitutive Pc promoter in the bgaA locus of merodiploid strains that also contained the native pneumococcal pcsB + gene (strains IU1628 and IU1608 respectively; Table 1). Strain IU1628 (pcsB +/sag0017 +) or IU1608 (pcsB +/gsp-781+) grew marginally better than or as poorly as, respectively, bacteria containing only the native pcsB + gene (strain EL776) when expression of the vicRKX operon was depleted (data not shown; see Fig. 8). The marginally better growth of the S. agalactiae merodiploids (pcsB +/sag0017 +) was consistent with microscopic examination, which indicated less severe defects in cell shapes and Fl-Van staining patterns for strain IU1628 than for strain EL776 (data not shown; see Fig. 8 and Ng et al., 2003). Backcrosses confirmed that the marginal growth phenotype of the S. agalactiae merodiploid strain was probably not caused by the accumulation of suppressor mutations (data not shown).
Nevertheless, constitutive expression of the ectopic S. agalactiae pcsB + (sag0017 ) or S. mutans gsp-781+ (smu.22) gene failed to suppress the severe growth defects caused by depletion of vicRKX operon expression in ΔvicK or ΔvicX mutants (strains IU1558, IU1561, IU1611 and IU1612, Table 1; data not shown). In addition, the native pneumococcal pcsB+ gene could not be knocked out in merodiploids constitutively expressing S. agalactiae pcsB + (sag0017) or S. mutans gsp-781+ (smu.22) from the bgaA locus (data not shown). Finally, S. pneumoniae mutants deleted for the secretory signal peptide encoded at the beginning of pcsB (Fig. 2) were not viable (data not shown). However, control experiments using chimeric proteins indicated that the secretory signal peptide sequence of S. agalactiae SAG0017 could replace that of S. pneumoniae PcsB (data not shown). Thus, neither the S. agalactiae nor S. mutans homologue of PcsB could simply replace the function(s) of pneumococcal PcsB, although S. agalactiae PcsB may have a low level of activity in S. pneumoniae (see Discussion).
Depletion of the VicRK TCS causes defects in cell morphology and murein synthesis indistinguishable from those caused by severe depletion of PcsB
Previously we showed that bacteria depleted for expression of the essential VicRK TCS formed abnormally shaped cells compared to those of the R6 parent (phase contrast, Fig. 8; see Ng et al., 2003). On the basis of this and other results, we concluded that the essentiality of the VicRK TCS is partly or wholly caused by downregulation of pcsB+ expression when the expression of the VicRK TCS is depleted (Ng et al., 2003). If this hypothesis is correct, then depletion of the VicRK TCS should lead to the same defects in cell wall synthesis as those caused by severe depletion of PcsB (Fig. 7). This was indeed the case for strain EL776 (PfcsK-vicRKX+Table 1), in which expression of the VicRK TCS was greatly reduced by growth in medium containing a very low concentration of fucose [0.001% (wt/vol)] (Fig. 8; Ng et al., 2003). The defects in cell morphology, cell wall synthesis (Fl-Van staining) and nucleoid content (DAPI staining) of cells depleted for VicRK TCS expression (Fig. 8) were extremely similar to those of cells severely depleted for PcsB (Fig. 7). Moreover, the localized staining of strain EL776 (Fig. 8) again supports the contention that Fl-Van indicates regions of new peptidoglycan synthesis and not aberrant incorporation of d-alanine-d-alanine residues into the muropeptides of mature cell walls. Under these culture conditions, strain EL776 continues to grow, although slowly, and aberrant general incorporation of d-alanine-d-alanine residues would lead to uniform staining around cells, which was not observed (Fig. 8).
We report here several new findings about the function and regulation of pcsB in S. pneumoniae. We show that the pcsB gene product strongly influences patterns of cell wall synthesis in S. pneumoniae R6 (Figs 6 and 7B), and several lines of evidence are consistent with PcsB acting as a cell wall hydrolase. First, PcsB is essential in S. pneumoniae, and both the conserved histidine and cysteine residues in the CHAP domain of PcsB are required for viability (Bateman and Rawlings, 2003; Rigden et al., 2003). In other peptidoglycan hydrolases, these two residues act together to allow cell wall cleavage by a nucleophilic attack mechanism (see Bateman and Rawlings, 2003; Rigden et al., 2003).
Second, underexpression of PcsB led to the formation of long chains of cells (Fig. 6), which is a phenotype often found for strains deficient in cell wall hydrolytic activity (De Las Rivas et al., 2002; Heidrich et al., 2002). These chains of cells exhibited a unique staining pattern with Fl-Van and DAPI. The chains seemed to be composed of pairs of late divisional cells (Fig. 6) that had the most cell wall synthesis occurring at the equator of every cell and at every other septum between cells (Fig. 6). DAPI staining showed that the nucleoids had not separated in cells containing equatorial cell wall synthesis (Fig. 6). Thus, underexpression of PcsB led to a cell separation defect and excess, but ordered cell wall synthesis. This pattern is consistent with PcsB acting as a cell wall hydrolase that counterbalances cell wall synthesis and participates in cell separation. Pairs or short chains of pairs of late divisional cells, but not long chains, were predominantly observed when BHI broth was supplemented with 0.062 M d-serine, which slowed down cell growth considerably (data not shown). d-serine incorporation has been reported to reduce the interstrand cross-linking in the peptidoglycan of Gram-positive bacteria (see De Jonge et al., 2002). Perhaps this change in cell wall composition allows more efficient relative hydrolysis in cells underexpressing PcsB so that long chains of cells are separated.
Third, severe depletion of PcsB led to rapid cessation of growth (Fig. 4) accompanied by the appearance of aberrantly shaped cells with irregular patterns of cell wall synthesis (Fig. 7B). This synthesis sometimes occurred over the entire surface of enlarged cells (Fig. 7B). Other irregularly shaped chains of cells had lost the parallel planes of equatorial and septal cell wall synthesis observed in the wild-type R6 parent strain (Figs 5 and 7B). These chains exhibited unusual regions of cell wall synthesis that radiated from single points or sometimes appeared to lay at right angles (Fig. 7B). Although pcsB function is essential, short-term severe depletion of PcsB for 6 h was not catastrophically lethal (Fig. 4); however, only one cell in each chain needed to survive to be counted as a CFU. Moreover, severe depletion for 2 h could be reversed (Fig. 4). The small, but reproducible, decrease in culture optical density observed upon the resumption of PcsB expression might reflect lysis of some of the large, round and more fragile looking cells formed upon severe depletion of PcsB (Figs 4 and 7B).
We also carried out several experiments with antibiotics to determine whether changes in PcsB expression caused phenotypes consistent with those reported previously for other murein hydrolases. Depletion or overexpression of murein hydrolases often leads to decreased or increased susceptibility, respectively, to antibiotics that inhibit cell wall synthesis, such as penicillin (see, e.g. Heidrich et al., 2002). We observed that S. pneumoniae strains constitutively expressing PcsB from the constitutive Pc or PfcsK promoter in the presence of fucose [EL1454 (Pc-pcsB+bgaA+) and IU1545 (PfcsK-pcsB+bgaA+) respectively; Table 1], were much more sensitive to autolysis induced by addition of 0.1 µg penicillin G per ml to cultures growing exponentially in BHI medium at 37°C (data not shown). This autolysis was dependent on the function of the major pneumococcal autolysin, LytA (data not shown). However, this increased sensitivity to penicillin G was not correlated with fucose concentration for strain IU1545 and was lost when Pc-pcsB+ or PfcsK-pcsB+ was expressed ectopically from the bgaA+ locus, which was shown in control experiments not to influence autolysis (data not shown). Therefore, on the one hand, constitutive expression of PcsB did increase antibiotic sensitivity, similar to other murein hydrolases. On the other hand, increased sensitivity did not seem to be simply correlated with PcsB expression levels.
Together, these data support a model in which PcsB acts as a critical murein hydrolase that modulates cell wall biosynthesis, which is a dynamic process involving synthetic and hydrolytic enzymes (reviewed in Höltje, 1998; Höltje and Heidrich, 2001). According to this model, when PcsB amount is underexpressed or depleted, a balance between peptidoglycan synthesis and turnover or trimming is disturbed. This imbalance leads to aberrant peptidoglycan synthesis at places where cell wall synthesis normally does not occur at that time in the cell cycle. PcsB may also play roles in directing murein synthetic and hydrolytic enzymes to specific regions of the cell wall. The facts that PcsB contains a leucine zipper motif in its amino-terminal domain (Fig. 2) and that mutations in the CHAP domain are possibly dominant negative (see Results) are suggestive of dimerization or some other protein interactions. Specific interactions in cell wall biosynthesis may explain why the closest homologues of PcsB from S. mutans (SMU.22) and S. agalactiae (SAG0017) (Fig. 2) failed to complement mutations in pneumococcal pcsB (see Results), although other explanations, such as protein instability, are certainly possible. Further experiments are needed to determine the biochemical functions and interactions of pneumococcal PcsB.
The essentiality of pneumococcal PcsB is noteworthy, especially if it turns out to function as a murein hydrolase. In E. coli, no single gene encoding a cell wall hydrolase is essential, and as many as seven or eight hydrolases can be knocked out without seriously affecting cell growth, although chains of cells begin to appear (Heidrich et al., 2002). There has been speculation as to why cell wall hydrolases seem to be so redundant and apparently dispensable in E. coli, especially because hydrolysis must be integrated with synthesis to maintain an intact sacculus (Höltje, 1998; Höltje and Heidrich, 2001). In pneumococcus, no other murein hydrolytic enzyme studied so far is essential, including the LytA and LytC autolysins (Sanchez-Puelles et al., 1986; Garcia et al., 1999), the low-molecular-weight penicillin-binding d,d-carboxypeptidase encoded by dacA(pbp3) (Schuster et al., 1990; Krauss and Hakenbeck, 1997), and the endo-β-N-acetylglucosaminidase encoded by lytB (De Las Rivas et al., 2002). dacA(pbp3) mutants exhibit several morphological defects, including thickened, irregular cell walls, clumping of cells and misplaced septa (Schuster et al., 1990; Krauss and Hakenbeck, 1997), some of which are also elicited by PcsB depletion (Fig. 7B). Intriguingly, it was recently reported that this d,d-carboxypeptidase plays a much greater and more complex role than previously appreciated in organizing the cell division process in S. pneumoniae (Morlot et al., 2004). Unlike the irregularly shaped cells of pcsB and dacA(pbp3) mutants (Fig. 6; Schuster et al., 1990; Krauss and Hakenbeck, 1997), lytB mutants form long chains of normally shaped cells, suggesting that the LytB hydrolase catalyses a final step in cell division that disperses intact cells (De Las Rivas et al., 2002).
The importance of PcsB is emphasized by its wide distribution in Gram-positive bacteria. The genomes of all sequenced species of streptococcus and related bacteria, such as Lactococcus lactis, each contain a single homologue with strong amino acid similarity to S. pneumoniae PcsB over its full length, excluding a variable linker region (Fig. 2; http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi). S. pneumoniae contains only one other CHAP domain protein, choline-binding protein D (CbpD), which is not essential but seems to play a role in nasopharyngeal colonization and adherence to eukaryotic cells (Gosink et al., 2000). The genomes of some other species of streptococcus encode several additional proteins containing CHAP domains. Some of these CHAP domain-containing proteins may have partially redundant functions, which could account for why pcsB is essential in some species but not others (Chia et al., 2001; Reinscheid et al., 2001). Finally, homologues of pneumococcal PcsB are found outside of S. pneumoniae and its immediate evolutionary relatives. For example, B. subtilis YvcE and E. faecium SagA contain amino- and carboxyl-terminal domains with amino acid sequence similarity to those of pneumococcal PcsB, again joined by a variable linker region (see Introduction; Kunst et al., 1997; Rigden et al., 2003; Teng et al., 2003).
The chromosomal location of pcsB and its homologues is also of potential interest and underscores a possible critical link to cell division. In S. pneumoniae (Fig. 1A), S. mitis, S. mutans, E. faecium, E. faecalis and L. lactis, the pcsB homologue is located adjacent to and downstream from the mreCD cluster (see also Mattos-Graner et al., 2001), which encodes proteins of unknown function that help to determine cell shape in B. subtilis and possibly other bacteria (Levin et al., 1992; Errington, 2003; Lee and Stewart, 2003). Our previous transcription data suggested that pcsB was strongly positively regulated by the VicRK TCS, whereas mreCD was not (Ng et al., 2003). Control experiments reported above (see Results) also did not reveal an obvious transcriptional link between the mreCD cluster and pcsB in S. pneumoniae. However, the conditions tested so far have been limited. Curiously, mreCD is not adjacent to the sag0017 pcsB homologue in S. agalactiae or S. pyogenes (see http://www.tigr.org/tigr-scripts/CMR2/CMRHomePage.spl); in fact, strong homologues of pneumococcal MreCD were not found in S. agalactiae or S. pyogenes by standard homology searching methods.
Besides providing controls for our mutant studies, the staining patterns of the wild-type R6 parent strain with Fl-Van and DAPI are noteworthy for two reasons (Fig. 5). First, they are fundamentally different from those reported recently for S. aureus, where cell wall synthesis is confined mainly to septal regions (Pinho and Errington, 2003). Second, they lend support to a recent model of cell division of S. pneumoniae proposed by Vernet and co-workers (Morlot et al., 2003). Immunofluorescence localization demonstrated that two different sets of high-molecular-weight penicillin-binding proteins (PBPs) are located at the equators and division septa of S. pneumoniae cells (Morlot et al., 2003). PBP2a and PBP2b, which are equatorially localized, were proposed to carry out peripheral peptidoglycan synthesis between separating equatorial ring structures away from and preceding synthesis at division septa. PBP2x and PBP1a were proposed to carry out septal cell wall synthesis by a mechanism that is not strictly coupled to FtsZ ring constriction (Morlot et al., 2003). These protein localization results are consistent with findings from earlier pulse-chase labelling studies of cell wall synthesis in S. pneumoniae (Briles and Tomasz, 1970). The pattern of Fl-Van staining reported here (Fig. 5) closely matches that predicted by this model of cell wall biosynthesis (see fig. 6 in Morlot et al., 2003). In particular, this model predicts simultaneous equatorial and septal peptidoglycan synthesis in dividing cells, such as that shown in Fig. 5C and D. Moreover, the equatorial synthesis observed was not diffuse and was likely restricted to equatorial ring structures, again as predicted by this model.
Finally, the results presented here provide crucial support for the hypothesis that the VicRK TCS is essential in S. pneumoniae primarily because of its strong positive regulation of pcsB (Ng et al., 2003). One tenet of this hypothesis is that pcsB itself is essential, which is established here. The second tenet of this hypothesis is that depletion of the VicRK TCS or PcsB should result in the same defects in cell morphology and cell wall synthesis. Comparison of Figs 7B and 8 shows that this is indeed the case, and as might be expected, cells deficient in regulation continued to grow slowly (Fig. 8), whereas those deprived of PcsB abruptly stopped growing (Figs 4 and 7B). Many questions remain to be answered as to why a TCS controls the transcription level of a gene that seems to be essential for normal cell wall biosynthesis and division of S. pneumoniae.
Bacterial strains and growth conditions
Bacterial strains used in this study were derived from S. pneumoniae R6, which is an unencapsulated mutant of the serotype 2 virulent strain D39 (Table 1). An isolate of strain R6 was assigned the unique strain designation EL59 to track its isogenic derivatives. S. pneumoniae, which is a facultative anaerobe, was cultivated statically in brain heart infusion broth (BHI; Difco Laboratories) at 37°C in an atmosphere of 5% CO2. Growth was monitored by change in OD620 in a Spectronic 20 spectrophotometer fitted for direct measurement of tubes with a 13 mm diameter. For CFU determinations, bacteria were serially diluted into sterile 0.9% (wt/vol) saline solution and spread onto trypticase soy agar II (modified) containing 5% (v/v) defibrinated sheep blood (TSAII BA; Becton Dickinson BBL). For antibiotic selection, plates were supplemented with 200 µg of kanamycin per ml, 100 µg of spectinomycin per ml, 0.25 µg of tetracycline per ml or 0.3 µg of erythromycin per ml. l-fucose (Sigma) was added to media as indicated in the text and figures and was routinely present at a final concentration of 0.2% (wt/vol) in BHI for overnight cultures. TSAII BA selection plates used for transformations and single-colony isolations of bacteria routinely contained 0.2% (wt/vol) fucose.
Transformation of S. pneumoniae
For transformation, S. pneumoniae cells were grown to OD620 (13 mm) ≈0.1. 100 µl of cells was added to 900 µl of BHI containing 10% (v/v) heat-inactivated horse serum (Sigma), 10 mM glucose, and 100 ng per ml of synthetic competence stimulatory peptide-1 (CSP-1) (generously provided by D. Morrison), and incubated for 13 min at 37°C in an atmosphere of 5% CO2. Following the addition of 0.2–1.0 µg of DNA per ml, the cells were incubated for an additional 1.5 h to allow phenotypic expression of antibiotic resistance markers. Transformants were recovered on TSAII BA containing appropriate antibiotics in nutrient broth soft agar overlays [0.8% (wt/vol) Bacto Nutrient Broth, 0.4% (wt/vol) Bacto Agar (Difco)].
Construction of S. pneumoniae mutants
Streptococcus pneumoniae strains (Table 1) containing defined mutations or ectopic copies of genes (Fig. 1) were constructed by transformation of competent pneumococcal cells with synthetic linear DNA synthesized by overlapping fusion PCR as previously described (Ng et al., 2003). Briefly, individual regions containing desired DNA sequences were amplified by PCR using Pfu (Stratagene) or rTth polymerases (Applied Biosystems) from source DNA using primers with 5′ extensions that allowed overlap and fusion of adjacent fragments. The primers synthesized for this study are listed in supplemental data at http://sunflower.bio.indiana.edu/~mwinkler/index.htm. PcsB homologues from S. agalactiae 2603 V/R and S. mutans UA159 were amplified from chromosomal DNA obtained from strains ATCC BAA-611 and ATCC 700610, respectively, purchased from the American Type Culture Collection. PCR amplicons corresponding to segments of final constructs were purified by gel electrophoresis and amplified together in a single PCR reaction containing only the outermost (flanking) set of primers. Final fusion PCR amplicons were purified by gel electrophoresis and used to transform competent pneumococcal cells as described above. Single colonies of bacteria were isolated, and the presence of desired mutations was verified by screening for changes in the sizes or restriction patterns of fragments amplified from chromosomal DNA using the flanking primers (Ng et al., 2003).
Generation of pcsB point mutations
The codons encoding the conserved cysteine (C292) or histidine (H343) residue in the CHAP domain were individually changed to an alanine codon (GCT) by a two-step PCR approach. Genomic DNA from strain IU1627 was used as the template. The 5′ portion of the mutagenic amplicon was amplified using the non-mutagenic primer WN0127 and mutagenic primer WN0131 (CTCCCCATGTAGCTTCTC CAATT for C292A) or WN0133 (GTAACAACCGCTACAG CACCATATCC for H343A). The 3′ portion of the mutagenic amplicon was amplified using mutagenic primer WN0130 (AATTGGAGAAGCTACATGGGGAG for C292A) or WN0132 (GGATATGGTGCTGTAGCGGTTGTTAC for H343A) and non-mutagenic primer WN0098. The 5′ and 3′ fragments for each construct were joined by PCR using primers WN0127 and WN0098. The presence of the desired mutations was verified by direct sequencing of each mutagenic amplicon. The final mutagenic amplicon carries the pcsB reading frame containing the desired mutation, linked to a selectable erythromycin marker, which is followed immediately by part of the rpsB gene. Transformation and selection of transformants (see Fig. 1G) were performed as described above. It was necessary to use large amounts of DNA during transformation to recover the desired point mutation in the transformants, possibly because strain R6 has a functional mismatch repair system.
Primers WN0098 and WN0127 were used to amplify the pcsB region from the chromosomal DNA of each transformant. BigDye Terminator (Applied Biosystems) sequencing reactions using primers WN0135 and WN0127 were performed to determine the presence of the desired point mutations.
Aliquots (50 µl) of bacterial cultures grown to OD620≈ 0.1 were stained directly without fixation by mixing with DAPI (final concentration: 0.2 µg ml−1) and Fl-Van (a 1:1 mixture of vancomycin and Bodipy FL-conjugated vancomycin, final concentration: 2 µg ml−1) for 5 min in the dark at room temperature. Stained bacterial cells were examined using a Nikon E-400 epifluorescent phase-contrast microscope equipped with a mercury lamp and filter blocks for fluorescence (DAPI: EX330-380, DM400, BA435-485; Fl-Van: EX460-500, DM505, BA510-560). Images from a 100× Nikon Plan Apo oil-immersion objective (NA 1.40) were captured using a cooled digital camera and processed with SPOT Advanced software.
We did not find it necessary to pregrow S. pneumoniae R6 in media containing d-serine to detect specific labelling patterns with Fl-Van, as was necessary for Staphylococcus aureus (Pinho and Errington, 2003). In fact, 0.125 M d-serine, which was the concentration used by Pinho and Errington (2003), inhibited the growth of S. pneumoniae R6 in BHI (data not shown). Pretreatment with lower concentrations of d-serine from 0.062 M, which inhibited growth rate and yield by about half, to 0.008 M, which did not inhibit growth, did not affect the Fl-Van staining patterns of parent strain R6 (data not shown). An additional control experiment showed that fucose addition to growth media did not affect the staining patterns of the parent R6 strain with Fl-Van and DAPI (data not shown).
We thank Barry Stein and Yves Brun for information about epifluorescent microscopy and Joel Lanie and Andrew Houppert for technical assistance. We thank Lilly Research Laboratories for providing some of the bacterial strains used in this work (‘EL’ designation in Table 1). This work was support by research funds from Indiana University Bloomington. Wai-Leung Ng is a predoctoral trainee on Grant NIGM-T32Gm0775 from the National Institutes of Health.