The Capsular Polysaccharide of Staphylococcus aureus Is Attached to Peptidoglycan by the LytR-CpsA-Psr (LCP) Family of Enzymes*

Background: Staphylococcus aureus LcpABC attach wall teichoic acids (WTA) to peptidoglycan. Results: S. aureus capsular polysaccharide (CP5) is linked to peptidoglycan in a manner requiring lcpABC genes. Conclusion: Unlike WTA, CP5 attachment is mediated preferentially by LcpC. Significance: LCP proteins display substrate preferences for the transfer of undecaprenyl-bound polymers to peptidoglycan. Envelope biogenesis in bacteria involves synthesis of intermediates that are tethered to the lipid carrier undecaprenol-phosphate. LytR-CpsA-Psr (LCP) enzymes have been proposed to catalyze the transfer of undecaprenol-linked intermediates onto the C6-hydroxyl of MurNAc in peptidoglycan, thereby promoting attachment of wall teichoic acid (WTA) in bacilli and staphylococci and capsular polysaccharides (CPS) in streptococci. S. aureus encodes three lcp enzymes, and a variant lacking all three genes (Δlcp) releases WTA from the bacterial envelope and displays a growth defect. Here, we report that the type 5 capsular polysaccharide (CP5) of Staphylococcus aureus Newman is covalently attached to the glycan strands of peptidoglycan. Cell wall attachment of CP5 is abrogated in the Δlcp variant, a defect that is best complemented via expression of lcpC in trans. CP5 synthesis and peptidoglycan attachment are not impaired in the tagO mutant, suggesting that CP5 synthesis does not involve the GlcNAc-ManNAc linkage unit of WTA and may instead utilize another Wzy-type ligase to assemble undecaprenyl-phosphate intermediates. Thus, LCP enzymes of S. aureus are promiscuous enzymes that attach secondary cell wall polymers with discrete linkage units to peptidoglycan.

Bacteria elaborate capsules as a defense mechanism and protective layer to withstand changes in the environment (1,2). During infection, encapsulation provides for escape from opsonophagocytosis and complement-mediated killing (3)(4)(5). Capsules are typically composed of high-molecular weight polysaccharides (CPS) 3 that are firmly attached to the cell sur-face (2,6,7). CPS consists of repeat units that vary greatly between species as well as within bacterial species, yielding distinct serotypes. Such structural diversity favors escape from adaptive immune responses. Typically, genes are required for CPS synthesis cluster in defined loci carrying conserved and variable genes, the latter of which account for serotype specificity. For example, Streptococcus agalactiae may synthesize up to 10 different serotypes (8,9), and no fewer than 80 and 93 capsular serotypes have been reported for Escherichia coli and Streptococcus pneumoniae, respectively (1,10). Despite their extensive structural diversity, many capsular polysaccharides are synthesized via the Wzy-dependent pathway: repeat units are assembled onto undecaprenyl-P in the cytoplasm, flipped across the membrane, and polymerized in a non-processive manner whereby a dedicated glycosyltransferase transfers a nascent capsule polymer from its undecaprenyl-PP carrier to the non-reducing end of an incoming undecaprenyl-PP-linked repeat unit (reviewed in Refs. 1, 2, and 11). This polymerization mechanism was first elucidated as one of three pathways (Wzy-, synthase-, and ABC transporter-dependent pathways) responsible for the synthesis of O-polysaccharide that defines the O-antigen serological specificity in Gram-negative bacteria (11)(12)(13). O-Polysaccharide is transferred from its undecaprenyl-PP carrier onto the lipid A-core at the periplasmic face of the plasma membrane (14). Wzy-dependent polymerization is ubiquitous in bacteria and based on the genetic conservation of CPS loci, and appears to account for capsule synthesis in most Gram-positive bacteria (1). Only two synthase-dependent loci have been described including type III CPS of S. pneumoniae (15) and the hyaluronic acid capsule of Streptococcus pyogenes (16). ABC transporter-dependent pathways have not been associated with capsular polysaccharide synthesis in Gram-positive bacteria.
The genetic requirement for CP8 synthesis has been described for 11 of the 16 genes (25), but only 5 CPS gene products have been characterized biochemically (19). Nonetheless, the genetic composition of the cluster clearly suggests that CPS synthesis in S. aureus occurs in a Wzy-dependent manner. In both S. agalactiae and S. pneumoniae, the biosynthetic enzymes are encoded by an operon beginning with a conserved gene, cpsA/cps2A (1, 26 -28). cpsA deletion mutants of S. pneumoniae and S. agalactiae produce less capsule, a phenotype originally attributed to a possible role in transcription (29 -31). CpsA encodes a protein that belongs to a family of highly homologous proteins, LytR-CpsA-Psr (LCP), present often as numerous paralogues in most Gram-positive organisms (32). In streptococcal species, CpsA/Cps2A has recently been implicated in mediating the transfer/attachment of capsule from its undecaprenyl-phosphate lipid carrier onto peptidoglycan, where loss of cpsA, lytR, and psr result in reduced capsule production, and release of the polymer into culture supernatants (31,33). A cpsA-like homologue is conspicuously missing in the CPS gene cluster of S. aureus, and tethering of the capsule to the envelope of staphylococci has not been revealed. In this work, we explore the nature of capsule attachment in staphylococci. We show that the capsule can be extracted along with cell wall fragments and can be released by chemical treatment or enzymatic degradation of peptidoglycan. Unlike wall teichoic acid (WTA), attachment of CP5 is not dependent on TagO, the enzyme that synthesizes the linker unit tethering WTA to peptidoglycan. However, similar to WTA, CPS is not attached to murein sacculi in a mutant lacking all three lcp genes, ⌬lcp. We conclude that CP5 is attached to peptidoglycan in a manner requiring LCP enzymes.

EXPERIMENTAL PROCEDURES
Growth Media and Reagents-S. aureus strains were propagated in tryptic soy broth or agar at 37°C with antibiotic selection when necessary. Erythromycin and chloramphenicol were used at 10 g ml Ϫ1 , spectinomycin at 200 g ml Ϫ1 , and tunicamycin at 1 g ml Ϫ1 , unless specified otherwise. For capsule production, S. aureus strains were cultured on Columbia agar plates supplemented with 2% NaCl and the appropriate antibiotic. E. coli strains were grown in lysogeny broth supplemented with ampicillin at 10 g ml Ϫ1 when needed. To assess growth of staphylococcal strains, stationary phase cultures were normalized to A 600 ϭ 3.0, diluted 1:100 into 3 ml of tryptic soy broth, and incubated with shaking at 37°C. Sample aliquots were taken at 3.5 h and serially diluted to determine colony-forming units (cfu).
Subcellular Fractionation of CPS Producing Staphylococci-Overnight cultures of S. aureus were spread onto Columbia agar plates and incubated at 37°C for 16 h. Cells harvested from one plate were suspended in 5 ml of 50 mM Tris-HCl (pH 7.5) and normalized by optical density at 600 nm (A 600 ϭ 50). One ml of suspension was vortexed vigorously and split into two 0.5-ml aliquots. One aliquot was treated with 10 g ml Ϫ1 lyso-staphin at 37°C for 2 h (total fraction). The second aliquot was centrifuged for 5 min at 21,000 ϫ g and the sediment containing intact cells was separated from the supernatant. Molecules in the supernatant fractions were subsequently digested with DNase (50 g ml Ϫ1 ) and RNase (50 g ml Ϫ1 ) supplemented with 2 mM MgCl 2 for 2 h at 37°C, and then treated with Pronase E (1 g ml Ϫ1 ) at 37°C for 16 h. Extracts were centrifuged at 21,000 ϫ g for 10 min, and soluble material was analyzed for the presence of CP5.
Preparation of Intact Murein Sacculi-Murein sacculi were prepared as described (41). Briefly, cells were scraped off Columbia agar, suspended in 4% SDS and boiled for 30 min. Cells were subsequently washed five times in water to remove detergent, and broken in a bead-beating instrument (MP Biomedicals). Murein sacculi were sedimented by centrifugation (10,000 ϫ g, 10 min), washed twice with water, and suspended in 50 mM Tris-HCl (pH 7.5), 10 mM CaCl 2 , 20 mM MgCl 2 , for digestion with DNase (10 g/ml) and RNase (50 g/ml) for 2 h at 37°C, and then treated with trypsin (100 g/ml) for 16 h at 37°C. Murein sacculi were again sedimented by centrifugation, suspended in 1% SDS, boiled for 15 min to inactivate enzymes, washed twice with water, once with 8 M LiCl, once with 100 mM EDTA, twice with water, once with acetone, and twice more with water. Purified murein sacculi were suspended in water, normalized to optical density (A 600 ϭ 10), and stored at Ϫ20°C until further use.
Extraction of Macromolecules Bound to Peptidoglycan-CP5 that associated with murein sacculi was liberated through chemical or enzymatic treatment as follows. Briefly, samples containing CP5 were incubated for 16 h with 40% hydrofluoric acid at 4°C, or 0.1 M NaOH at room temperature. The remaining insoluble peptidoglycan material was sedimented and washed three times with 1 M Tris-HCl (pH 7.5), and incubated with lysostaphin as described above. For enzymatic release of CP5 associated with intact murein sacculi, sample aliquots (50 l of A 600 ϭ 10) were solubilized by digestion with either lysostaphin (10 g) in 0.5 M Tris-HCl (pH 7.5), autolysin amidase (20 g), or glucosaminidase (20 g) in 0.1 M Tris-HCl (pH 7.0), at 37°C for 16 h. Insoluble material was removed by centrifugation (21,000 ϫ g, 10 min), and soluble material was transferred to a new tube for analyses.
Immunoblotting-To examine the protein contents of extracts from murein sacculi preparation or subcellular fractionations, sample aliquots were mixed (v/v) with sample buffer (0.1 M Tris-HCl, pH 8.0, 4% SDS, 20% glycerol, 2 mM ␤-mercaptoethanol, 0.04% bromphenol blue), heated at 90°C for 10 min, and subjected to 12% SDS-PAGE prior to electrotransfer to poly(vinylidene difluoride) (PVDF) membranes. In some instances when total bacterial loads were monitored, staphylococci from washed cultures were lysed with lysostaphin and precipitated with 7% trichloroacetic acid. The precipitates were washed once with acetone, dried, and solubilized in 100 l of 0.5 M Tris-HCl (pH 8.0), 4% SDS, separated on 12% SDS-PAGE prior to transfer to PVDF membranes. Membranes were blocked by incubation in 2% bovine serum albumin for 1 h at room temperature. To prevent binding of primary antibodies to protein A, 0.8 mg of human IgG was added to 10 ml of blocking buffer. After incubation (1 h, room temperature), immune serum (primary polyclonal antibodies at a dilution of 1:5,000 for anti-50 S ribosomal subunit L6, anti-SrtA, and anti-SpA, and at a dilution of 1:15,000 for anti-LcpA, anti-LcpB, and anti-LcpC) was added, and the membranes were incubated for an additional hour. Membranes were washed 3 times for 10 min in TBS-T (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 0.1% Tween 20), and incubated 1:20,000 in secondary antibody (goat antirabbit coupled to IRDye 680) for 1 h. Membranes were washed again in TBS-T, and fluorescence at 700 nm was measured with an infrared scanner (Li-Cor Biosciences Odyssey imager). Immunoblotting for CP5 was performed in a similar manner except that samples were spotted directly onto nitrocellulose membrane (primary polyclonal antibodies were used at a dilution of 1:5,000).
Purification of Proteins-E. coli XL1 Blue was used for the purification of glutathione S-transferase fusions to the autolysin amidase (AM) (42) or glucosaminidase (GL) domains (43). E. coli BL21(DE3) was used for the purification of soluble Lcp proteins carrying an N-terminal His 6 tag. Bacterial cultures were grown to A 600 ϭ 0.6 in LB with 100 g ml Ϫ1 of ampicillin at 37°C and lcp expression was induced with 1 mM isopropyl 1-thio-␤-D-galactopyranoside overnight at room temperature. Bacterial cells were sedimented by centrifugation (10,000 ϫ g, 10 min) and suspended in lysis buffer A (50 mM Tris-HCl (pH 7.5), 150 mM NaCl) for the purification of GST fusions and in PBS (pH 7.4), containing 20 mM imidazole for His 6 -tagged proteins. Cells were lysed by passage through the French Pressure cell (twice at 15,000 p.s.i.). Crude lysates were cleared by centrifugation (100,000 ϫ g, 30 min), and the supernatant was loaded by gravity flow onto glutathione-Sepharose beads (GE Healthcare) pre-equililibrated or nickel-nitrilotriacetic acid beads (Qiagen) pre-equilibrated in their respective lysis buffer. Columns were washed with 20 volumes of lysis buffer, and bound proteins were eluted with either 20 mM reduced glutathione or 500 mM imidazole. Proteins in the eluates were dialyzed into the lysis buffer, quantified by bicinchoninic acid assay (Pierce), and kept at 4°C for immediate use or stored frozen at Ϫ80°C.
Purification and Conjugation of CP5 for Antibody Production-S. aureus CP5 was purified according to a previously published method (44) with some modifications. S. aureus Reynolds (capsular polysaccharide type 5) was grown on Columbia agar plates supplemented with 2% sodium chloride to induce capsular polysaccharide synthesis. After overnight incubation at 37°C, bacteria were scraped off agar plates and suspended in 25 ml of buffer A (50 mM Tris-HCl, 2 mM MgSO 4 , pH 7.5). The staphylococcal suspension was incubated with lysostaphin (150 g/ml) at 37°C with slow rotation for 2 h. Samples were then treated with DNase and RNase (50 g/ml) for an additional 2 h at 37°C, followed by overnight treatment with Pronase E (5 units/ml) at 37°C. Samples were centrifuged to sediment bacterial debris and the supernatant was sterilized by filtration (0.2-m Millipore nitrocellulose filter). The filtrate was dialyzed against buffer B (50 mM NaOAc, 10 mM NaCl) and loaded onto a MonoQ anion exchange column equilibrated with buffer B, the recorded absorbance of collected fractions at 206, 215, 260, and 280 nm was measured. CPS was eluted with 400 ml of linearly increasing gradient of salt (0.01-1 M NaCl). Collected fractions were analyzed to identify elution of capsular polysaccharide (immunoblotting), peptidoglycan (Morgan Elson reaction), and teichoic acid (phosphate content). Fractions containing capsular polysaccharide were pooled and treated with 10 mM NaIO 4 to oxidize contaminating teichoic acid polymers for 30 min at room temperature. Samples were dried in a rotatory lyophilizer. The resulting powder was dissolved in 1 ml of sterile water and chromatographed on a Sephacryl S-300 column equilibrated with buffer B. Fractions containing capsular polysaccharide were pooled and lyophilized. A full wavelength scan (200 -800 nm, Varian) was performed to determine the purity of isolated CP5. Nuclear magnetic resonance (NMR) spectroscopy was used to verify the structure of the capsular polysaccharide by comparison with previously published spectra (22). For NMR analysis, dried polysaccharide was solubilized into deuterated water and analyzed using a 500 MHz NMR spectrometer (Varian Inova 500) at an indicated probe temperature of 50°C.
Generation of Rabbit Polyclonal Antibodies-For antibody generation, rabbits (Charles River Laboratories, 6-month-old female New Zealand White rabbits) were immunized with 500 g of CP5-CRM197 conjugate or purified Lcp proteins emulsified in Complete Freund's adjuvant (Difco). For booster immunizations, samples were emulsified in incomplete Freund's adjuvant and injected 24 or 48 days following the initial immunization. Rabbits were bled on day 60 and serum stored for future use.
Assessment of the in Vitro Activity of Murein Hydrolases-The enzymatic activity of purified AM and GL derived from autolysin was assessed by using osmotically stable bacterial cells as substrates. Briefly, S. aureus Newman (25 ml) grown to A 600 ϭ 1 and bacteria were sedimented by centrifugation at 8000 ϫ g for 10 min. Bacteria in the pellet were suspended in 5 ml of water, 5 ml of ethanol/acetone (1:1) was added, and samples were incubated on ice for 30 min. This treatment rendered the cells osmotically stable. Cells were washed with 20 ml of ice-cold water and suspended in 100 l of 0.1 M Tris-HCl (pH 7.5) for incubation with lysostaphin, or in 0.1 M Tris-HCl (pH 7.0) for incubation with AM or GL. Cell wall-anchored proteins released by these treatments were recovered in the supernatant of acetone-treated protoplasts after centrifugation at 21,000 ϫ g for 10 min. Protein A in the soluble extracts was identified by immunoblot using a rabbit polyclonal anti-SpA serum (␣-SpA) from our laboratory collection.
Transmission Electron Microscopy-Bacteria were washed twice with 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, bathed in fixative (2% glutaraldehyde, 4% paraformaldehyde, 0.1 M sodium cacodylate buffer) overnight at 4°C, and post-fixed with 1% OsO 4 in 0.1 M sodium cacodylate buffer for 60 min. Fixed samples were stained in 1% uranyl acetate in maleate buffer for 60 min, serially dehydrated with increasing concentrations of ethanol, embedded in spurr resin for 48 h at 60°C, thin sectioned (90 nm) using a Reichert-Jung Ultracut device, and post-stained in uranyl acetate and lead citrate. The samples were imaged on the FEI Tecnai F30 with a Gatan charge-coupled device (CCD) digital micrograph.
Statistical Analyses-The abundance and release of macromolecules from the cell wall of wild type, mutant, and complemented strains was analyzed for statistical analysis by analysis of variance with Dunnett's multiple comparison test. All data were analyzed using Prism (GraphPad Software, Inc.); p values Ͻ0.05 were deemed significant.

RESULTS
Purification of CP5 for the Generation of Antibodies-To examine the capsule of S. aureus, we first purified CP5 from the well characterized strain Reynolds and raised a polyclonal antibody. S. aureus Reynolds was grown on Columbia agar supplemented with 2% NaCl to favor capsule production (45) and capsular materials were extracted following sequential incubations of bacterial cells with lysostaphin, DNase and RNase, and Pronase E. Soluble materials obtained following these treatments were subjected to anion-exchange chromatography, and eluted fractions with the highest absorbances at 206 and 215 nm were pooled (Fig. 1A). The presence of CP5 in these fractions was later verified by dot blot with specific rabbit CP5 antiserum (Fig. 1A, lower panel). The pooled fractions were chromatographed on a Sephacryl S-300 size exclusion column. Fractions with the highest absorbances at 206 and 215 nm and lowest protein content were pooled (Fig. 1B). Enrichment for capsule was documented retroactively by dot blot (Fig. 1B,  lower panel). The purity of CP5 in the pooled fractions was verified by an absorbance scan (200 -800 nm) revealing that protein contaminants were indeed separated following size exclusion chromatography and no longer present in the pooled fractions (Fig. 1C) and its structure was confirmed by NMR spectroscopy (Fig. 1D). The chemical shifts of protons associated with the N-or O-acetyl groups were identified at 2.0 -2.1 ppm, and chemical shifts of protons (H) associated with the sugar molecules were matched to 3.5-5.0 (H1-H5) and 1.2-1.3 ppm (H6), as previously reported (22). This purified CP5 was used to generate polyclonal antibodies in a rabbit following chemical conjugation to the conventional carrier protein, CRM197.
The Capsular Polysaccharide of S. aureus Is Associated with Murein Sacculi-Previous work suggested that extraction of CPS could be enhanced under acidic conditions or following incubation of staphylococci with lysostaphin (7,46,47). Nonetheless, routine preparations of CPS involved the recovery of capsular materials from the supernatant of previously autoclaved bacterial suspensions (47). Here, we asked whether CP5 is covalently attached to the murein sacculi of S. aureus New-man and, if so, by what mechanism. Staphylococci were harvested, boiled in SDS, washed extensively, and broken in a bead beater. Cellular molecules (nucleic acids and proteins) were enzymatically degraded, and murein sacculi were purified. Lysostaphin was used to hydrolyze peptidoglycan and thus solubilize macromolecules otherwise associated with murein sacculi. When spotted on a nitrocellulose membrane and stained with a rabbit polyclonal anti-CP5 serum (␣-CP5), immunoreactive species were observed in murein sacculi of wild type strain S. aureus Newman (WT), but not from the murein sacculi of a variant lacking capO, a gene that is known to be essential for capsule synthesis (48) (Fig. 2A, left panel). To examine the chemical nature of the capsule linkage to peptidoglycan, we first asked whether it was sensitive to hydrolysis with weak acid, a treatment known to release phosphodiester and O-acetyl bonds (41,49,50). Incubation of murein sacculi with hydrofluoric acid removed all CP5 immune reactive signals, indicating that capsule is linked to peptidoglycan via an acid labile bond ( Fig. 2A, left panel). To account for the possibility that hydrofluoric acid treatment removed O-acetyl moieties, thereby abrogating antibody binding to CP5, we subjected murein sacculi to alkaline lysis with 0.1 M NaOH for 16 h at room temperature, a treatment that removes O-acetyl groups (21,49). Alkaline lysis did not abolish antibody detection of CP5 ( Fig. 2A,  right panel), indicating the CP5 linkage to peptidoglycan is indeed susceptible to hydrofluoric acid hydrolysis.
CP5 Association with Murein Sacculi Does Not Require TagO-Attachment of WTA polymers to peptidoglycan requires a linker unit that is synthesized by TagO a member of the WecA family of proteins that links UDP-GlcNAc and undecaprenyl phosphate (C 55 -P) to generate C 55 -PP-GlcNAc (51). We wondered whether S. aureus CP5 synthesis requires tagO expression as well. Murein sacculi were extracted from the S. aureus Newman tagO variant, treated with lysostaphin, and spotted on nitrocellulose membrane. In a complementary approach, wild type S. aureus Newman was grown in the presence of low concentrations of tunicamycin, an antibiotic from Streptomyces clavuligerus that inhibits the glycosyl transferase activity of TagO at 1 g ml Ϫ1 . At higher concentrations (10 -20 g ml Ϫ1 ), tunicamycin inhibits MraY, an enzyme essential for peptidoglycan biosynthesis. Dot blot analysis of murein sacculi digested with lysostaphin revealed continued synthesis and association of CP5 with peptidoglycan regardless of whether the tagO gene and its product were genetically or chemically inactivated (Fig.  2, A and B). These data indicate that tagO expression and TagO activity are dispensable for CP5 synthesis and assembly in the cell wall envelope. Of note, bioinformatics analyses (BLAST searches) suggest that TagO and MraY are the only members of the WecA family present in S. aureus. Interestingly, depletion of WTA, elicited upon tagO deletion or tunicamycin inhibition, resulted in an increase in CP5 as compared with wild type (Fig.  2B).
CP5 Is Linked to the Glycan Strands of Peptidoglycan-The peptidoglycan of S. aureus is comprised of the repeating disaccharide, N-acetylmuramic acid (NAM)-(␤1-4)-N-acetylglucosamine. The D-lactyl moiety of N-acetylmuramic acid is amidelinked to tetrapeptides that are cross-linked via pentaglycyl cross-bridges to wall peptides of adjacent glycan strands (Fig.  3A) (52). Where pentaglycine cross-linking does not occur, the terminal pentaglycyl cross-bridge is linked to surface proteins, a reaction that is catalyzed by the transpeptidase sortase A (SrtA) (Fig. 3A) (53,54). Several murein hydrolases have been characterized for the substrate specificities in cleaving the staphylococcal peptidoglycan. AM cut the amide bond that links the wall peptide to NAM, thereby liberating glycan strands. GL cleave the N-acetylglucosamine-(␤1-4)-NAM bond, and endopeptidases, such as lysostaphin, cut the pentaglycine cross-bridge (Fig. 3A). Murein sacculi prepared from wild type strain Newman or the isogenic capO variant were incubated with lysostaphin, AM, or GL (43). If CP5 were attached to NAM A, extract derived from S. aureus Reynolds grown on Columbia salt agar were subjected to anion-exchange chromatography on a MonoQ column equilibrated with buffer B (50 mM NaOAc, 10 mM NaCl), and absorbance of eluted fractions was recorded at 206 (blue) and 215 nm (pink). Type 5 capsular polysaccharide (CP5) elution was identified by immunoblotting with specific rabbit antiserum. Green bar indicates sample fractions that were pooled for further purification. B, capsular polysaccharide sample (A) was chromatographed on a Sephacryl S-300 size exclusion column equilibrated with buffer B and absorbance at 206 and 215 nm was recorded. CP5 elution was identified by immunoblotting with specific antiserum. The bracket indicates fractions pooled for lyophilization. C, absorbance scan (200 -800 nm) of purified CP5. D, 500 MHz one-dimensional 1 H NMR spectroscopy of purified CP5 at 50°C. FIGURE 2. S. aureus CP5 is attached to peptidoglycan in a TagO-independent manner. A, murein sacculi extracted from Newman wild-type (WT), capO, and tagO strains grown on Columbia agar supplemented with 2% NaCl were incubated for 16 h with (ϩ) or without (Ϫ) 40% hydrofluoric acid (HF) at 4°C, or 0.1 M NaOH at room temperature. Acid-or base-treated sacculi were subsequently neutralized and digested with lysostaphin. Solubilized materials were spotted onto a nitrocellulose membrane and probed for CP5 immune reactivity using the ␣-CP5 antiserum. B, WT, capO, and tagO strains were grown on Columbia agar supplemented with 2% NaCl in the presence of 0, 1, or 5 g/ml of tunicamycin (Tun), which inhibits TagO-glycosyl transferase activity. Murein sacculi were extracted, solubilized by lysostaphin treatment, and analyzed by immunoblotting using the ␣-CP5 antiserum.
or N-acetylglucosamine residues of peptidoglycan, incubation with lysostaphin or AM, which eliminate peptidoglycan crosslinking, would be expected to release capsule from murein sacculi. In contrast, GL cleavage would not be expected to solubilize the capsule, as cut glycan strands would remain associated through a network of cross-linked wall peptides. These predictions were tested. As expected, treatment of murein sacculi with lysostaphin, AM, or GL caused a decrease in turbidity, or a decline in A 600 (Fig. 3B). Large amounts of CP5 were released from murein sacculi treated with lysostaphin or AM (Fig. 3, C and D). In contrast, incubation with GL did not promote release of CP5 from murein sacculi (Fig. 3, C and D).
To ensure that the purified AM and GL displayed their expected activities, staphylococci were incubated with either lysostaphin for 2 h, or AM or GL for 16 h at 37°C. Soluble material was separated from cells by centrifugation, and analyzed by immunoblotting for the presence of protein A (SpA), a SrtA-anchored surface protein (Fig. 3E). Equal amounts of SpA were released upon treatment with either AM or GL, whereas lysostaphin, as expected (the enzyme cuts the anchoring point of surface proteins), was more effective (Fig. 3, C and D). Taken together, these data indicate that CP5 is attached to the glycan strands of the staphylococcal peptidoglycan.
LCP Enzymes Are Required for CP5 Attachment to Peptidoglycan-We wondered whether S. aureus lcp genes are also involved in the attachment of CP5 to the staphylococcal envelope albeit that capsule anchoring requires neither murein linkage units nor the expression of tagO. To test this possibility, lcp mutant alleles were generated in S. aureus Newman, as MSSA1112 and its variants (37) do not elaborate CP5. Compared with the wild type parent strain Newman, the ⌬lcp variant displayed a 2-log reduction in viability, as shown by enumerating colony forming units (cfu) in broth cultures grown to the same A 600 (Fig. 4, A and B); this phenotype was also observed with the MSSA1112 parent (37). In comparison, the tagO mutant exhibited only a 1-log reduction in cfu as compared with Newman (Fig. 4B). Transmission electron micrographs of wild type bacteria revealed regular-shaped cells with septa placed at the mid-cell (Fig. 4C, top row). Cells of the tagO variant often failed to separate fully and displayed aberrant septal placement (Fig. 4C), as reported before (55). Micrographs of ⌬lcp cells revealed defects similar and more exacerbated than those of the tagO mutant (Fig. 4C). Consistent with previous reports that neither tagO nor ⌬lcp elaborate cell wall-bound teichoic acid, the outer electron dense layer was missing in the envelope of ⌬lcp variant Newman (Fig. 4C) (37,55).
To examine the contribution of lcp genes for display of CP5 in S. aureus Newman, the wild-type, tagO, capO, LCP single (lcpA, lcpB, lcpC), double (lcpAB, lcpAC, lcpBC), and triple mutant (⌬lcp) strains, and the ⌬lcp variant complemented with single LCP genes (⌬lcp/plcpA, ⌬lcp/plcpB, ⌬lcp/plcpC) were cultured on Columbia agar with NaCl. Bacteria were scraped from plates, suspended in 50 mM Tris-HCl (pH 7.5) buffer, and vortexed vigorously. Samples were sedimented and non-cellassociated macromolecules were recovered in the supernatant. The presence of CP5 and cytosolic ribosomal protein L6 in supernatant fractions was examined by dot blot and immunoblot, respectively (Fig. 5, A and B, S fractions). The data indicated that staphylococci grown on Columbia agar were not prone to lysis, as ribosomal protein L6 was not detected in the supernatant fractions. The variant lacking lcpC accumulated CP5 in the supernatant fraction. The lcpAB mutant did not display a defect in CP5 retention. The lcpAC, lcpBC, and triple (⌬lcp) mutants displayed levels of CP5 release commensurate to the lcpC mutant (the quantification for three independent experiments is shown in Fig. 5C). Of note, the increased release of capsule in the supernatant fraction could be rectified by expression of lcpC on a plasmid (plcpC) in the ⌬lcp mutant (Fig.   FIGURE 3. CP5 association with glycan strands of peptidoglycan. A, schematic of peptidoglycan structure and site of cleavage by muralytic enzymes. WTA is attached to the 6Ј-hydroxyl of N-acetylmuramic acid (NAM), whereas sortase A substrates such as protein A (SpA) are anchored to the pentaglycine cross-bridge. B, murein sacculi from S. aureus Newman WT and capO strains were incubated at 37°C with muralytic enzymes lysostaphin (L), AM, GL, or buffer (B) alone. Turbidity (A 600 ), as a measure of peptidoglycan cleavage, was monitored for 12 h. C, murein sacculi solubilized as in B were spotted onto nitrocellulose membrane and analyzed for CP5 release by immunoblot using the ␣-CP5 antiserum. A representative blot of seven replicates is shown. D, CP5 release from C was quantified by densitometry measurements. The value of 1 was attributed to samples treated with lysostaphin and used for normalization. Statistical differences compared with buffer treatment were determined by analysis of variance with Dunnett's multiple comparison test (***, p Ͻ 0.001). E, muralytic enzyme activity of lysostaphin, AM, and GL was assessed by release of the SpA protein (black arrow) from S. aureus Newman. Staphylococci, treated with ethanol/acetone to create osmotically stable cells, were incubated with lysostaphin, AM, GL, or buffer (B) at 37°C. Solubilized material was separated from intact cells by centrifugation, resolved by SDS-PAGE, and analyzed by immunoblotting for protein A release using ␣-SpA antibody and for cell lysis using ␣-L6 (ribosomal protein L6) antiserum. Numbers to the left of the blots indicate molecular mass markers (kDa). A and C). Plasmid-encoded lcpA and lcpB failed to reduce capsule release in the supernatant fractions of ⌬lcp mutant bacteria to statistically significant levels (Fig. 5, A and C). To document that similar numbers of cells were processed for the dot blot analyses, bacterial cultures were lysed with lysostaphin and the protein content in these total extracts was examined for the presence of ribosomal protein L6 by immunoblot (Fig. 5B). Next, murein sacculi were extracted from washed bacterial cells previously scrapped off Columbia agar plates. Dot blot analyses using ␣-CP5 antibodies showed that all strains except for the capO mutant produced CP5 (data not shown). However, capsular polysaccharide was greatly diminished in murein sacculi from the lcpC mutant and absent from sacculi of the ⌬lcp variant (Fig. 5, A and D). Mutants lacking lcpA or lcpB had no effect on CP5 attachment whether deleted individually or in combination. Interestingly, the ⌬lcp defect was restored by overproducing anyone of the three lcp genes on a plasmid (Fig. 5, A and  D). Together, these data show that lcp genes are essential for the attachment of CP5 to peptidoglycan. Moreover, lcpC encodes the key enzyme for immobilization of capsular polysaccharide to the staphylococcal cell wall.

5,
LcpC Requirement for Capsule Attachment Is Not Due to Preferential Expression over LcpA and LcpB-The data described above suggest a key contribution of LcpC toward CP5 attachment in the envelope of staphylococci. Because bacteria are grown to stationary phase on Columbia agar, a condition that is more favorable for capsule production, we wondered whether LcpC production might be specifically enhanced. Staphylococcal strain Newman and the isogenic ⌬lcp mutant were grown either in tryptic soy broth or Columbia agar. Bacteria were normalized to similar optical densities and lysed with lysostaphin. Proteins in extracts were separated by SDS-PAGE and transferred to PVDF membranes for incubation with polyclonal antibodies against LcpA, LcpB, and LcpC, as well as SrtA, which served as a loading control (Fig. 6). This analysis did not reveal any noticeable changes in LCP protein levels, indicating that lcpC is not differentially expressed compared with lcpA and lcpB when cells are induced for CP5 expression. Thus, LcpC appears to have evolved to preferentially recognize capsular polysaccharide synthesis intermediates for attachment to the cell wall envelope.

DISCUSSION
Using India ink negative staining, Gilbert first reported capsule production in S. aureus, studying a strain that had been isolated from a patient with "acute ulcerative gonococcal endocarditis" (56). When kept for 1 month on infusion agar slants, Gilbert observed two forms of colonies, smooth (S) forms with overall morphology and staining pattern similar to the original isolate and rough (R) forms, variants lacking capsule production. Gilbert (56) also reported that the supernatant forms were more virulent when inoculated into the peritoneal cavity of guinea pigs. A typing scheme was eventually developed based upon preparation of absorbed rabbit antiserum to prototype S. aureus strains and led to the identification of eight capsular serotypes (46). This method revealed the prevalence of CPS types 5 and 8 in hospital isolates of S. aureus and assigned the heavily encapsulated strains M and Smith diffuse to serotypes 1 and 2 (17). It is now appreciated that CP5 and CP8 are 2 of 13 different S. aureus serotypes and account for ϳ80% of human isolates (reviewed by Ref. 18).
Genes involved in the synthesis of bacterial capsular polysaccharides are organized in large genomic clusters where the presence of conserved genes, i.e. homologues shared among many different capsule clusters, provide insights on the mechanism of capsule assembly (1,57). The genetic content of S. aureus CPS clusters suggests that staphylococci employ the Wzy-dependent mechanism that involves the formation of repeat units and non-processive polymerization for the synthesis of capsule (1). A model for this pathway in S. aureus is shown in Fig. 7. The final step of the synthesis pathway involves transfer of translocated capsule polymer from its bactoprenol carrier onto peptidoglycan. This step is poorly characterized and the molecular principles for capsule attachment in Gram-positive bacteria have rarely been addressed. For example, NMR analysis of mutanolysin cleaved sacculi revealed that CPS type III of GBS is linked via a glycosidic linker to N-acetylglucosamine residues of peptidoglycan (6), whereas the Lancefield group B carbohydrate of GBS appears to be linked to N-acetylmuramic residues of peptidoglycan with an intervening linkage unit (6,58). In Bacillus anthracis, the capsule is composed of a poly-D-␥-glutamate polymer, and is also attached to peptidoglycan. However, precursor polyglutamate filaments are not synthesized on undecaprenyl-phosphate and attachment of filaments occurs directly on the free amino group of meso-diaminopicolate in a step catalyzed by the CapD transpeptidase (5). Here, we show that CP5 of S. aureus Newman is released by mild acid treatment from the cell wall. Furthermore, incubation of murein sacculi with endopeptidase or amidase, but not with glucosaminidase, released the capsular polysaccharide. These data suggest that the staphylococcal capsule is attached to the glycan strands of peptidoglycan.
LCP proteins have been proposed to catalyze the transfer of undecaprenyl phosphate-linked secondary cell wall polymer intermediates onto the amino sugar moieties of peptidoglycan (59). The typical structure of LCP enzymes is comprised of a short N-terminal cytoplasmic tail, followed by 1 to 5 transmembrane helices, and a C-terminal conserved domain (ϳ150 residues) exposed on the bacterial surface and shared among all members of the LCP family (32). Based on crystallization studies of S. pneumoniae Cps2A and B. subtilis TagT, the LCP domain contains conserved residues important for coordinating interactions with polyprenyl lipids, as well as Mg 2ϩ and pyrophosphate, and is endowed with pyrophosphatase activity in vitro (33,59).
LCP-encoding genes often lie in close proximity to gene clusters encoding the biosynthesis of secondary polymers for which they exert their catalytic effects. For instance, in B. subtilis, the lcp genes, tagTUV, all fall within a 50-kb region on the chromosome, directly adjacent to biosynthetic genes of teichuronic acid and minor teichoic acid, and within six open reading frames of the WTA gene cluster (59). Deletion of tagTUV is not tolerated and genetic depletion leads to loss of WTA (59). It is not known whether LCPs attach teichuronic acids and minor teichoic acids to the envelope of B. subtilis. S. pneumoniae encodes three lcp genes: cps2A, lytR, and psr, where cps2A is the first gene of the CPS operon. Disruption of cps2A alone or in combination with lytR results in reduced capsule synthesis, and release of CPS into the culture supernatants (33). S. agalactiae also encodes three LCP genes, and cpsA is the first gene of the type III capsule gene cluster. Deletion of cpsA results in reduced cell wall capsule (31). The contribution of LCPs to the group B carbohydrate of S. agalactiae has not been examined; assembly of this carbohydrate, however, requires the enzyme GbcO, the TagO orthologue (58). FIGURE 5. CP5 attachment to murein sacculi requires the LCP gene products. S. aureus Newman wild-type (WT), tagO (WTA-deficient), capO (capsuledeficient), and LCP single (lcpA, lcpB, and lcpC), double (lcpAB, lcpAC, and lcpBC), and triple (⌬lcp) mutant without or with complementing plasmids plcpA, plcpB, or plcpC were grown on Columbia agar supplemented with 2% NaCl. A, cells were harvested in 50 mM Tris-HCl (pH 7.5) buffer, normalized to the same optical density, and suspension was fractionated into supernatant (S) and cells from which murein sacculi digested with lysostaphin (MS) were further extracted. The presence of CP5 in the supernatant and MS fractions was examined by spotting samples on nitrocellulose and immunoblotting with the ␣-CP5 antiserum. B, cell suspensions harvested and normalized as in A were fractionated into supernatant (S) or incubated with lysostaphin (T, total). Supernatant and total fractions were separated by SDS-PAGE and transferred to PVDF membrane for Western blot analysis using the ␣-L6 antiserum to assess for cell lysis during growth and document that roughly the same quantity of cellular material were used for each strain examined. Numbers to the left of the blots indicate molecular mass markers (kDa). C and D, CP5 immunoreactive species in panel A were quantified by densitometry from at least three independent experiments. Amounts of CP5 in WT samples were arbitrarily set as 1 and all other samples were normalized against wild type CP5 levels. The white line marks the background level of CP5 immune reactive signals. Statistical differences relative to WT (denoted by asterisks directly above the bars) or lcpC were determined by analysis of variance with a Dunnett's multiple comparison test (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001). FIGURE 6. LcpC is not preferentially produced when cells are grown in capsule-inducing medium. S. aureus Newman wild-type and ⌬lcp strains were cultured under standard laboratory growth conditions in tryptic soy broth (TSB) to mid-log phase (A 600 ϭ 0.5), or capsule-inducing conditions on Columbia agar supplemented with 2% NaCl (C/NaCl). Cells were harvested, normalized to A 600 ϭ 2, and lysed with lysostaphin. Proteins in these lysates were separated by SDS-PAGE, transferred on PVDF membrane for Western blot analysis using ␣-LcpA, ␣-LcpB, ␣-LcpC, and ␣-Sortase A (SrtA) antisera. Levels of SrtA were monitored to document that roughly the same number of cells was used among strains examined. Numbers to the left of the blots indicate molecular mass markers (kDa). The data depict a representative of two independent experiments.
S. aureus encodes three LCP genes, lcpABC. These genes are not part of any biosynthetic cluster but deletion of any one affects attachment of WTA to the bacterial peptidoglycan (37). Staphylococcal WTA is a polymer of 30 -50 ribitol-phosphate (Rbo-P) subunits connected via 1,5-phosphodiester bonds (60). Rbo-P n is tethered to peptidoglycan via the murein linkage unit, GlcNAc-ManNAc-(Gro-P) 2-3 (61), sequentially synthesized by TagO (GlcNAc), TagA (ManNAc), and TagBDF (Gro-P) onto undecaprenyl-phosphate (51,62). The product of this pathway is presumably flipped across the plasma membrane by the TagGH transporter (63). Attachment of the WTA polymer to the C6-hydroxyl of N-acetylmuramic acid within peptidoglycan occurs during cell wall assembly (64) and requires at least one LCP enzyme, as mutants with deletion of all three lcp genes no longer retain WTA in the envelope (37,65). The lcpA and lcpB mutants display the largest reduction in WTA anchoring, whereas deletion of lcpC causes only a small reduction (37). Here, we show that lcpC encodes the key LCP enzyme for attachment of CP5 to the peptidoglycan of S. aureus. Although the defect of the lcpC mutant is not further exacerbated in lcpAC or lcpBC double mutants, capsule attachment is entirely abolished in the triple mutant (⌬lcp).
Wzy-dependent synthesis has been best described for assembly of O-polysaccharides in Gram-negative bacteria (11)(12)(13). This polymer is directly attached to lipid A and does not bear a linkage unit. The function of S. aureus capsular genes fits a similar synthesis pathway with the exception of the last step that requires LcpC (Fig. 7). Considering the involvement of LCP proteins in transferring polymers for which synthesis is initiated by TagO (31,37,50,51,58,59,66,67), we examined whether capsule assembly requires tagO in S. aureus. Furthermore, recent work has shown that heterologous expression of Pseudomonas aeruginosa O11 lipopolysaccharide synthesis genes, which include wbpL, a UDP-FucNAc-1-phosphate transferase, together with CP5 genes capHIJK was sufficient to synthesize lipid-linked CP5 polymer repeats in E. coli (68). WbpL is a protein that harbors the WecA/WbcO/MraY glycosyl transferase homology domain, a TagO orthologue. Other than mraY and tagO, the S. aureus genome does not encode any other genes bearing this conserved domain that could fulfill this enzymatic role. Nevertheless, our data show that tagO is dispensable for CP5 synthesis, and thus we can exclude murein linkage units from the capsule assembly pathway. Interestingly, S. pneumoniae species do not encode tagO orthologues, other than mraY. The sugar transferase that initiates capsule synthesis by transferring glucose 1-phosphate onto the polyprenyl phosphate lipid carrier is believed to be encoded by cspE/wchA, the fifth gene in the 17-gene cps/cap cluster (1,69). When expressed in E. coli, this gene is sufficient to synthesize a glucose-P-lipid intermediate (70). Deletion of this gene in S. pneumoniae strains resulted in failure to elaborate a capsule (71). S. aureus capM encodes a protein with homology to WchA, and may serve as the membrane-bound sugar transferase for the synthesis of the UDP-FucNAc linkage (Fig. 7) (19). Although this biosynthetic pathway seems plausible to us, experimental proof for this model is not yet available.
Increased amounts of CP5 were observed when synthesis of WTA was inhibited or abrogated. This result suggests that both wall polymers may occupy the same C6 hydroxyl of N-acetylmuramic acid or adjacent sites on peptidoglycan such that occupancy by one polymer precludes attachment of another. This phenomenon has been described for S. pneumoniae (1), where the absence/reduction of one polymer (WTA or CPS) leads to increased abundance of the other (72). . Model for S. aureus capsular polysaccharide synthesis. Enzymes unique to CP5 or CP8 synthesis are denoted. Enzyme names in italic refer to those whose roles are predicted based on homology and bioinformatics, and whose functions have not been experimentally validated. CapK is hypothesized to serve as a flippase, transporting the lipid-linked capsule intermediate from the cytoplasm to the extracellular milieu, whereas the function of CapC (not depicted) is unknown.
In summary, we report here that CP5 is linked to the glycan strands of S. aureus peptidoglycan, presumably to the same (C6-hydroxyl of MurNAc) or proximal sites as used for the anchoring of WTA. Capsule attachment is catalyzed by LcpC, albeit that LcpA and LcpB can functionally substitute at least in part. Unlike WTA, capsule anchoring does not require TagOmediated synthesis of murein linkage units (GlcNAc-ManNac) and likely utilizes undecaprenyl-PP-FucNAc, thought to be formed by a reaction that is catalyzed by CapM.