Correspondence: Waldemar Vollmer, Institute for Cell and Molecular Biosciences, Medical School, University of Newcastle upon Tyne, Catherine Cookson Building, Framlington Place, Newcastle upon Tyne NE2 4HH, UK. Tel.: +44 0 191 222 6295; fax: +44 0 191 222 7424; e-mail: firstname.lastname@example.org
The normal, unmodified glycan strands of bacterial peptidoglycan consist of alternating residues of β-1,4-linked N-acetylmuramic acid and N-acetylglucosamine. In many species the glycan strands become modified after their insertion into the cell wall. This review describes the structure of secondary modifications and of attachment sites of surface polymers in the glycan strands of peptidoglycan. It also provides an overview of the occurrence of these modifications in various bacterial species. Recently, enzymes responsible for the N-deacetylation, N-glycolylation and O-acetylation of the glycan strands were identified. The presence of these modifications affects the hydrolysis of peptidoglycan and its enlargement during cell growth. Glycan strands are frequently deacetylated and/or O-acetylated in pathogenic species. These alterations affect the recognition of bacteria by host factors, and contribute to the resistance of bacteria to host defence factors such as lysozyme.
The glycan strands of the bacterial peptidoglycan (murein) are synthesized by polymerization of lipid II, which is, in most cases, GlcNAc (β1→4) MurNAc-peptide linked to an undecaprenyl pyrophosphoryl lipid anchor (GlcNAc, N-acetylglucosamine; MurNAc, N-acetylmuramic acid) (van Heijenoort, 1998, 2001; Vollmer et al., 2008). Therefore, the unmodified, polymerized glycan strands consist of alternating, β-1,4-linked GlcNAc and MurNAc residues. However, there are no bacterial species known to have such unmodified, poly(GlcNAc-MurNAc) glycan strands in their mature peptidoglycan, as glycan strands invariably become modified or linked to other cell-wall polymers shortly after their synthesis (Fig. 1). For example, the glycan strands of Gram-negative species have a 1,6-anhydro ring at the terminal MurNAc residue, whereas Gram-positive species attach other cell-wall polymers (e.g. teichoic acids, capsular polysaccharides) via phosphodiester bonds to their GlcNAc or MurNAc residues. In addition, many pathogenic species contain secondary glycan strand modifications affecting their interaction with elements of the immune system. This review describes the currently known modifications of glycan strands and the occurrence of these modifications in different bacterial species. In recent years, several of the enzymes responsible for these modifications have been identified and characterized. The focus of this review will be the N-deacetylation, N-glycolylation and O-acetylation of glycan strands.
N-Deacetylation of GlcNAc and MurNAc residues
Occurrence of N-deacetylated peptidoglycan
The muramidase lysozyme hydrolyses the glycan strands of peptidoglycan between C1 of MurNAc and C4 of GlcNAc. As early as 1971, while studying the glycan strand composition of the lysozyme-resistant Bacillus cereus, Araki et al. (1971b) identified the presence of a high proportion of nonacetylated glucosamine (GlcN) residues in the peptidoglycan of this species. A later study also identified nonacetylated muramic acid residues (MurN) residues in the peptidoglycan of Bacillus anthracis (Zipperle et al., 1984). The presence of these nonacetylated amino sugars accounted for the observed lysozyme-resistance, as isolated cell walls could not be digested by lysozyme unless they were chemically N-acetylated by acetic anhydride. Other Bacillus species, including B. anthracis, Bacillus thuringensis and Bacillus subtilis strain 168, contain deacetylated GlcN and MurN residues in their peptidoglycan (Table 1). However, B. subtilis strain W23 contains only GlcN but not MurN residues, indicating that the MurNAc of its peptidoglycan is fully acetylated (Zipperle et al., 1984). Nonacetylated GlcN was detected in the peptidoglycan of several other bacteria, including S. pneumoniae (Ohno et al., 1982; Vollmer & Tomasz, 2000), Lactobacillus fermentum (Logardt & Neujahr, 1975) and Listeria monocytogenes (Boneca et al., 2007), whereas nonacetylated MurN was detected in small quantities in S. pneumoniae (Vollmer & Tomasz, 2000) and Micrococcus lysodeiktikus (Hoshino et al., 1972). Deacetylated amino sugars appear to be widespread in peptidoglycan, and the application of modern analytical techniques is likely to reveal their presence in other bacterial species.
Table 1. Occurrence of N-deacetylation, muramic-δ-lactam and O-acetylation in bacterial petidoglycans
Deacetylated amino sugars are formed from the GlcNAc and MurNAc residues by peptidoglycan deacetylases. The supernatant of B. cereus cells contains an enzymatic activity capable of deacetylating GlcNAc residues in peptidoglycan. It was noted that the peptidoglycan deacetylase present in the crude fraction has a different biochemical property to that of the known GlcNAc-6-phosphate deacetylase (Araki et al., 1971a). The deacetylation reaction is most likely to occur on polymerized peptidoglycan because deacetylated precursors have not been detected in species with deacetylated peptidoglycan, and known deacetylases have a predicted extracytoplasmic localization (Vollmer & Tomasz, 2000). Furthermore, inactivation of deacetylase genes is tolerated in different species with none or at most minor effects on cell growth. Therefore, the absence of deacetylated amino sugars in the peptidoglycan appears not to affect the peptidoglycan biosynthetic enzymes, the transglycosylases and transpeptidases in these mutants.
Identification of peptidoglycan deacetylase genes
The first gene encoding a peptidoglycan GlcNAc deacetylase, pgdA, was identified in S. pneumoniae (Vollmer & Tomasz, 2000). A pgdA mutant strain lacked deacetylated amino sugars in its peptidoglycan. The amino acid sequences of the pneumococcal peptidoglycan deacetylase PgdA, rhizobial NodB chitooligosaccharide (nodulation factor) deacetylases and fungal chitin deacetylases are similar, which is not surprising because their substrates share a common structural feature. All three substrates (peptidoglycan, nodulation factor and chitin) have a backbone of β-1,4-linked GlcNAc residues (MurNAc in peptidoglycan is the 3-lactyl ether of GlcNAc). According to the classification of Henrissat (Coutinho & Henrissat, 1999, 2002), these enzymes belong to carbohydrate esterase family 4 (CE4, PFAM01522), which includes peptidoglycan GlcNAc deacetylases (EC 3.1.1.-), rhizobial NodB chitooligosaccharide deacetylases (EC 3.5.1.-), chitin deacetylases (EC 22.214.171.124), acetyl xylan esterases (EC 126.96.36.199), and xylanases A, C, D and E (EC 188.8.131.52). Members of this family hydrolyse either N-linked acetyl groups from GlcNAc residues (peptidoglycan deacetylases, rhizobial NodB chitooligosaccharide deacetylases, chitin deacetylases) or O-linked acetyl groups from O-acetylxylose residues (acetyl xylan esterases and xylanases A, C, D and E) of their substrates. Interestingly, some members of this family have low substrate specificity: the chitin deacetylase CDA from Mucor rouxii deacetylated both, chitin and xylan, and the xylan esterase AxeA from Streptomyces lividans is also active on soluble chitin substrates. By contrast, neither of these enzymes is able to deacetylate peptidoglycan from B. subtilis 168 (Caufrier et al., 2003), perhaps because of structural constraints due to the bulky MurNAc-peptide moiety in the latter substrate.
Figure 2 shows the sequence alignment of the five currently recognized peptidoglycan deacetylases: PgdA from S. pneumoniae, PdaA from B. subtilis, BC1960 from B. cereus, BC3618 from B. cereus and PgdA from L. monocytogenes, together with the NodB chitooligosaccharide deacetylase from Mesorhizobium loti and the chitin deacetylase CDA1 from Saccharomyces cerevisiae. The carbohydrate esterase family CE4 contains over 800 ORFs (Bornscheuer, 2002), many of which encode for hypothetical polysaccharide deacetylases in bacterial genomes. The amino acid sequence alone is not sufficient to determine the substrate (peptidoglycan, chitin, chitooligosaccharide or xylan) of a hypothetical deacetylase of family CE4. Therefore, structural analysis of the peptidoglycan of the mutant strain lacking a hypothetical gene as well as biochemical characterization of the gene product are required to establish its functional activity. For example, both B. cereus and B. anthracis encode 10 putative polysaccharide deacetylases (Psylinakis et al., 2005). A combination of sequence comparison, pattern-based analysis, phylogenetic clusters, networked cellular pathways and chromosomal neighbourhoods of functionally related genes (Ivanova et al., 2003; Overbeek et al., 2003) suggested that six of these ORFs encode for peptidoglycan deacetylases, and three encode for chitin deacetylases (there was no assignment for one deacetylase). Subsequent biochemical characterization of two of the deacetylase gene products from B. cereus, BC1960 and BC3618, confirmed their peptidoglycan GlcNAc deacetylase activity (Psylinakis et al., 2005).
Structure and biochemical property of peptidoglycan deacetylases
PdaA from B. subtilis was the first bacterial peptidoglycan deacetylase for which biochemical activity and structure was determined (Fig. 3) (Blair & van Aalten, 2004; Fukushima et al., 2005). The structure revealed a modified (β/α)8 fold, with an c. 3-nm-long and 0.9-nm-wide groove on one side of the protein. Residues conserved in family CE4 deacetylases are clustered within the groove which forms the catalytic site. The structure was also solved with a bound GlcNAc residue or with a Cd2+ ion. Many members of the carbohydrate esterase family CE4 have been reported to be metal ion-dependent. PdaA is expressed with a predicted cleavable signal peptide. Biochemical studies with a soluble form of PdaA revealed a particular substrate specificity (Fukushima et al., 2005); PdaA was inactive against peptidoglycan, peptidoglycan fragments carrying short peptides (obtained by treatment with the l,d-endopeptidase LytF) and hexa-N-acetylchitohexaose, but showed activity against oligo-(GlcNAcMurNAc) glycan strands (obtained from peptidoglycan by treatment with the amidase CwlH). In these strands, MurNAc residues, but not GlcNAc residues, were deacetylated by PdaA. Remarkably, PdaA is required for the formation of muramic δ-lactam during sporulation of B. subtilis (Fukushima et al., 2002), and it could artificially produce this structure in Escherichia coli although its in vitro activity was limited to the deacetylation of MurNAc residues.
The peptidoglycan GlcNAc deacetylase A (PgdA) from S. pneumoniae is about twice the size of typical family CE4 proteins. PgdA has a predicted N-terminal membrane anchor followed by a large region of unknown function, and a C-terminal deacetylase domain (Vollmer & Tomasz, 2000). The crystal structure of a soluble form of PgdA was solved and shows a three-domain architecture, with the N-terminal domain from amino acids 46 to 160, a smaller domain from amino acids 161 to 268 and the deacetylase domain from amino acids 269 to 463 (Fig. 3) (Blair et al., 2005). Although there are significant topological differences between PgdA and PdaA (from B. subtilis), the overall fold of the deacetylase domain and the catalytic core are similar in both structures. Structural analysis did not reveal possible in vivo functions for the N-terminal and middle domains of PgdA, both of which are absent in PdaA. PgdA contained a Zn2+ ion, complexed with two histidines and an aspartic acid, and this metal binding triad is conserved in members of the CE4 family. The crystal structures of PgdA and of two acetylxylan esterases (Taylor et al., 2006) contain an acetate residue which interacts with the Zn2+ ion. In PgdA the amino acids Asp-275, His-417, Tyr-367, Trp-385, Trp-392 and Leu-415 bind the acetate residue, whereas the amino acids Asp-276, His-326 and His-330 form a Zn2+-binding triad. These amino acids for acetate and Zn2+-binding are conserved in other peptidoglycan deacetylases (Fig. 2). As expected from the crystal structure, the activity of PgdA is metal-dependent and the addition of EDTA inactivates the enzyme (Blair et al., 2005). The enzyme activity was highest in the presence of Co2+, but PgdA was also active in the presence of Zn2+, Ni2+ or Fe2+, whereas it showed little or no activity with Cu2+, Cd2+, Mg2+ or Ca2+. PgdA deacetylated the middle GlcNAc residue of an artificial GlcNAc3 substrate. Insights from the crystal structure and from activity assays using several PgdA variants with amino acid exchanges of the conserved amino acids lead to a proposed reaction mechanism. Accordingly, the catalytic base Asp-275 abstracts a proton from a water molecule, creating a nucleophilic attack on the carbonyl carbon of the acetate to create a tetrahedral oxyanion intermediate. His-417 then protonates the intermediate on the nitrogen atom, generating a free amine and releasing the acetate product. Two acetylxylan esterases from Streptomyces lividans and Clostridium thermocellum belonging to the CE4 family are very similar to the deacetylase domain of PgdA with respect to the overall three-dimensional structure and the metal ion coordination. These acetylxylan esterases contained an acetate in their crystal structure, and, as both were most active in the presence of a Co2+ ion, a reaction mechanism similar to that of PgdA has been proposed (Taylor et al., 2006).
Two other peptidoglycan GlcNAc deacetylases have been characterized biochemically, BC1960 and BC3618 from B. cereus (Psylinakis et al., 2005). Both were inactive against GlcNAc but were active against the oligosaccharides GlcNAc2–8 and showed highest activity against GlcNAc4. In these substrates, all GlcNAc residues were deacetylated except the one at the reducing terminus. Both enzymes deacetylated GlcNAc residues in a peptidoglycan fragment (GlcNAcMurNAc-l-Ala-d-Glu) and in peptidoglycan from Helicobacter pylori and B. cereus.
Biological role of peptidoglycan deacetylation
The presence of deacetylated sugars in the peptidoglycan strands strongly reduces the activity of the muramidase lysozyme. Indeed, interactions between the acetyl groups in a hexasaccharide glycan strand and amino acids in a long groove in the lysozyme molecule are important for substrate binding (Blake et al., 1965; Vocadlo et al., 2001). Several studies have shown that deacetylated peptidoglycan is a poor substrate for lysozyme, and that the activity of lysozyme can be restored by chemical acetylation of the substrate (Amano et al., 1977, 1980; Westmacott & Perkins, 1979; Vollmer & Tomasz, 2000). Lysozyme is ubiquitous in phages, bacteria, fungi and mammals. It is an important factor of the innate immune system in humans and is present in many tissues and body liquids. In addition, lysozyme is secreted in large amounts by cells of the immune system at sites of bacterial infection. Consequently, pathogenic bacteria such as S. pneumoniae, L. monocytogenes or B. anthracis appear to deacetylate their peptidoglycan as a means of resisting the activity of lysozyme. Deacetylation is not the only modification with an effect on the activity of lysozyme. For example, O-acetylation of MurNAc and the covalent linkage of other cell-wall polymers such as teichoic acid also increase resistance to lysozyme (Bera et al., 2007).
An S. pneumoniae mutant lacking a functional pgdA gene became more sensitive towards lysozyme in the stationary phase of growth (Vollmer & Tomasz, 2000). A pdgA mutant expressing a type 3 capsule exhibited significantly reduced virulence in a intraperitoneal mouse model as compared with the parental strain, indicating that PgdA is a putative virulence factor (Vollmer & Tomasz, 2002). It is likely that the effect of peptidoglycan deacetylation on the interaction with the host might be multifactorial and not restricted to the effect on lysozyme activity. Deacetylation introduces additional positive charge into the cell wall, potentially affecting the binding of specific cell-wall proteins and other compounds such as capsular polysaccharide, which is the major virulence determinant of S. pneumoniae. In addition, increasing the positive charge of the cell wall is likely to increase the resistance of the bacterium to cationic antimicrobial peptides, which are important elements of the innate immune system (Peschel, 2002). Thus, next to d-alanylation of teichoic acids (Peschel et al., 1999) and lysinylation of phospholipids (Staubitz et al., 2004), the deacetylation of peptidoglycan could be a third way to introduce positive charges into the cell wall to protect pathogens against antimicrobial peptides of the host organism. The role of peptidoglycan deacetylation in virulence has been recently confirmed in L. monocytogenes (Boneca et al., 2007). A pgdA mutant strain was very sensitive and rapidly killed within macrophages. In contrast to the wild-type, the mutant strain induced a strong IFN-β response in a TLR2- and Nod1-dependent manner. Thus, deacetylation of peptidoglycan has an effect on pathogen recognition via different host receptors.
It is possible that some of the bacterial peptidoglycan hydrolases (autolysins), in particular those cleaving in the glycan strands (glucosaminidases and muramidases), have different activities against fully acetylated or deacetylated peptidoglycan. Hydrolases are required for cleavage of the septum prior to separation of the daughter cells, and for peptidoglycan turnover (the release of cell wall into the surrounding medium) during growth (Vollmer et al., 2008). In S. pneumoniae, cell separation is performed by the LytB glucosaminidase (Garcia et al., 1999), and this enzyme appears not to be affected by the state of acetylation of the peptidoglycan because cell separation is normal in pgdA mutant strains (Vollmer & Tomasz, 2000). It is not known whether deacetylase mutants alter the amount of peptidoglycan turnover products and their structural composition, as compared with the wild-type. An interesting variation of this theme appears to be the mechanism by which Bdellovibrio bacteriovorus infects an E. coli cell. Members of the genus Bdellovibrio are Gram-negative bacteria that prey obligately on other Gram-negative bacteria. Bdellovibrio bacteriovorus efficiently exploits the nutrients of its prey by forming a stable bdelloplast, multiplying in the periplasm of its prey. Upon attachment to a prey cell, Bdellovibrio bacteriovorus penetrates the outer membrane and the peptidoglycan layer but remains in the periplasm. To enable it to penetrate the peptidoglycan layer, Bdellovibrio bacteriovorus hijacks the autolytic enzymes of E. coli by an as yet unknown mechanism. After penetration, Bdellovibrio bacteriovorus deacetylates the prey's peptidoglycan to prevent further cleavage by the E. coli autolysins (Thomashow & Rittenberg, 1978; Tudor et al., 1990). This presumably prevents premature lysis of the E. coli cell by its own autolysins. The deacetylases involved and the mechanisms of their regulation are not known.
Muramic acid δ-lactam residues in spore peptidoglycan
Bacterial endospores have unique features with respect to heat and chemical resistance. Spores have a thick peptidoglycan known as the spore cortex that is responsible for dormancy (Popham, 2002). Spores of Bacillus species and Clostridium sporogenes contain a high abundance of a spore-specific muramic acid δ-lactam, which is generated by intramolecular amide bond formation between the carboxyl group of the lactyl group at position 3 of MurNAc, and the amino group at position 2. In B. subtilis, about 50% of the MurNAc residues in the spore peptidoglycan are modified to the δ-lactam, and these residues appear to be distributed regularly at every second muramic acid position along the glycan strands (Warth & Strominger, 1972; Atrih et al., 1996; Popham et al., 1996b).
There are two preconditions for δ-lactam formation: (1) the lactyl group must be free (i.e. without attached peptide), and (2) the MurNAc residue needs to be deacetylated. Interestingly, only two enzymes are required to produce muramic acid δ-lactam in B. subtilis, the amidase CwlD and the peptidoglycan MurNAc deacetylase PdaA. CwlD removes the peptide from the acetylmuramic acid residue in peptidoglycan (Sekiguchi et al., 1995). Mutants lacking this enzyme produced no peptide-free MurNAc residues during sporulation and lacked the muramic acid δ-lactam modification (Sekiguchi et al., 1995; Atrih et al., 1996; Popham et al., 1996a). Although an amidase activity of CwlD could not be determined in vitro, overproduction of the enzyme in E. coli resulted in the formation of typical amidase products (Gilmore et al., 2004). PdaA was active in vitro as a peptidoglycan MurNAc deacetylase. Bacillus subtilis mutants without PdaA produced spore peptidoglycan with fully acetylated MurNAc residues lacking muramic acid δ-lactam (Fukushima et al., 2002). Interestingly, heterologous production of CwlD and PdaA resulted in the formation of muramic acid δ-lactam in E. coli, showing that both proteins are necessary and sufficient for muramic acid δ-lactam production (Gilmore et al., 2004).
Mutants lacking CwlD or PdaA are able to produce intact endospores, indicating that the muramic acid δ-lactam modification is not required for sporulation, spore dehydration or heat resistance. By contrast, spores lacking muramic acid δ-lactam residues were unable to complete the germination (outgrowth) process to produce viable cells (Popham et al., 1996a). In particular, the mutant spores were unable to degrade the spore cortex during germination. Thus, the muramic acid δ-lactam modification appears to serve as a ‘marker’ for spore peptidoglycan recognized by germination-specific hydrolases. The specificity of the germination-specific hydrolases for spore peptidoglycan (with muramic acid δ-lactam) ensures that the new peptidoglycan of the germ (without muramic acid δ-lactam) remains intact (Popham et al., 1996a; Atrih & Foster, 2001; Chirakkal et al., 2002).
N-Glycolylation of muramic acid
The presence of a glycolyl residue (instead of acetate) at the two-amino group of muramic acid was first described in Mycobacterium smegmatis (Adam et al., 1969). Subsequent studies revealed that this modification is the hallmark of closely related genera within the Actinomycetales (Uchida & Aida, 1979). Therefore, the identification of glycolate in the peptidoglycan is used in bacterial taxonomy for the classification of Actinomycetales (Uchida et al., 1999). A glycolylated peptidoglycan is present in most genera with mycolic acids (the mycolata) including Mycobacterium, Rhodococcus, Tsukamurella, Gordonia, Nocardia, Skermania and Dietzia (Azuma et al., 1970; Sutcliffe, 1998). The only exception is the genus Corynebacterium, members of which contain mycolic acid and N-acetylated rather than N-glycolylated muramic acid (Azuma et al., 1970; Uchida & Aida, 1979). In addition, there are several closely related genera within the Actinomycetales which contain N-glycolyl muramic acid in the peptidoglycan but not mycolic acids, including Actinoplanes, Asanoa, Catellatospora, Catenuloplanes, Couchioplanes, Dactylosporangium, Glycomyces, Longispora, Microbacterium, Micromonospora, Okibacterium, Pilimelia, Spirilliplanes, Verrucosispora and Virgisporangium (Evtushenko et al., 2002; Matsumoto et al., 2003; Li et al., 2005; and references therein).
In contrast to other modifications in the glycan strands (such as N-deacetylation or O-acetylation), which appear to occur at the polymerized peptidoglycan, the N-glycolyl modification is introduced during synthesis of UDP-linked precursors. Initial analysis showed that the precursors were quantitatively N-glycolylated in Mycobacterium phlei and M. smegmatis (Petit et al., 1970; Takayama et al., 1970; Lederer, 1971). However, the quantitative N-glycolylation of muramic acid observed in these studies appears to be a consequence of the enrichment procedure for the precursors using cycloserine. When cycloserine was omitted, the precursor pool contained both an N-glycolylated and N-acetylated precursor (Mahapatra et al., 2005). This is consistent with the analysis of mycobacterial peptidoglycan, which also contains both substituents at muramic acid residues (Adam et al., 1969; Lederer, 1971; Mahapatra et al., 2005). Thus, glycolylation affects only a fraction of the muramic acid residues in mycobacteria, and unmodified acetylated muramic acid is also present both in the precursors and in the peptidoglycan. In Rhodococcus rhodochrous the extent of N-glycolylation of peptidoglycan is dependent on growth conditions. When cells were grown in the presence of glycerol, N-glycolylation was suppressed until the later stages of growth. However, this pattern of glycolylation might be specific to this species, as it was not observed in Rhodococcus erythropolis or Rhodococcus globerulus (Sutcliffe, 1998).
The glycolyl modification is introduced into the last soluble, cytoplasmic precursor (UDP-MurNAc-pentapeptide) by a monooxygenase (hydroxylase) in the presence of molecular oxygen and NADPH (Fig. 4). Such activity has been found in M. phlei and in Nocardia asteroides (Gateau et al., 1976; Essers & Schoop, 1978). Recently, the namH gene encoding a UDP-MurNAc-pentapeptide monooxygenase was identified in M. smegmatis (Raymond et al., 2005). The namH gene product has sequence similarity to eukaryotic enzymes responsible for N-glycolylation of sialic acids. Homologues of namH are present in species producing N-glycolylated peptidoglycan but not in the related actinobacteria Corynebacterium glutamicum and Streptomyces albus, which do not have this modification. An M. smegmatis mutant lacking a functional namH gene was viable but lacked N-glycolate modifications to its precursors and peptidoglycan.
The role of the glycolate residue is not known. As compared with acetate, glycolate has an extra hydroxyl group. It has been speculated that the extra hydroxyl group participates in hydrogen bonding within the cell envelope and consequently the stability of the envelope (Brennan & Nikaido, 1995). Interestingly, an M. smegmatis namH mutant lacking the N-glycolate modification was hypersensitive to lysozyme and β-lactam antibiotics. Therefore, the N-glycolylation of mycobacterial peptidoglycan could have a similar role in protection against lysozyme as has N-deacetylation and O-acetylation in other species.
O-Acetylation of peptidoglycan
Occurrence of O-acetylated peptidoglycan
O-Acetylated peptidoglycan was first detected in Micrococcus luteus (Brumfitt et al., 1958) and Streptococcus faecalis ATCC 9790 (now Enterococcus hirae ATCC 9790) (Abrams, 1958). The peptidoglycan of these species contains a proportion of MurNAc with an extra acetyl group linked to C6-OH to form a 2,6-N,O-diacetyl muramic acid residue. Thus far, O-acetylation of GlcNAc residues has not been observed. O-Acetylation of peptidoglycan is more prevalent than N-deacetylation and occurs in both Gram-positive and Gram-negative species, including many important pathogens (Table 1). For example, O-acetylated peptidoglycan is present in the Gram-positive B. cereus, Staphylococcus aureus, Enterococcus hirae and S. pneumoniae. Gram-negative species with this modification include Neisseria gonorrhoeae, Neisseria meningitidis, H. pylori and Proteus mirabilis. The extent of O-acetylation varies between <20% and 70% in different species and strains. Peptidoglycan O-acetylation has been studied extensively by the group of Anthony Clarke, who has published excellent overviews on this topic (Clarke & Dupont, 1992; Clarke et al., 2000; Weadge et al., 2005).
Detection of O-acetyl groups in peptidoglycan
The ester bond of O-linked acetate is significantly weaker than the amide bond of N-linked acetate. Therefore, acetate is lost from O-acetylated MurNAc even at mild alkaline or acidic conditions. Treatment with an acid is frequently used in purification of peptidoglycan to remove other cell-wall polymers such as teichoic acids. In addition, HPLC analysis of peptidoglycan fragments is often performed at acidic pH and elevated temperature, and these conditions can cause the loss of O-linked acetyl groups (W. Vollmer, unpublished observations). For example, O-acetylated fragments were not observed in peptidoglycan from S. pneumoniae unless the protocols for the isolation and analysis of peptidoglycan were designed to preserve the labile O-acetyl groups (Bera et al., 2005). Similarly, O-acetylated peptidoglycan fragments were not observed in H. pylori with the standard muropeptide analysis protocol although new data demonstrate the occurrence of O-acetyl residues in the peptidoglycan of this species (Weadge et al., 2005). For the detection of O-acetyl groups the peptidoglycan sample has to be purified at neutral pH. Various methods have been developed to analyse peptidoglycan for the presence of O-linked acetate. For example, acetate can be released by a mild alkaline treatment, followed by analysis on an anion exchange column (Dupont & Clarke, 1991a; Bera et al., 2005). This method is used in combination with the quantification of peptidoglycan by high-performance anion-exchange chromatography-using pulsed-amperometric detection (PAD) to determine the extent of O-acetylation in peptidoglycan (Clarke, 1993a). Alternatively, O-acetylated peptidoglycan fragments (muropeptides) can be released by treatment with a muramidase (Cellosyl), followed by separation by a modified HPLC method preserving O-acetyl groups and MS (Bera et al., 2005).
O-Acetylation and de-O-acetylation of MurNAc residues
Because O-acetyl groups are not present on the peptidoglycan precursor lipid II, it is assumed that the acetylation reaction occurs on polymerized peptidoglycan (Gmeiner & Kroll, 1981; Dougherty, 1983a, b; Lear & Perkins, 1983, 1986, 1987; Gmeiner & Sarnow, 1987; Snowden et al., 1989). This is consistent with pulse-chase experiments performed in P. mirabilis and N. gonorrhoeae, showing that only non-O-acetylated peptidoglycan subunits are incorporated into the peptidoglycan. During maturation of the incorporated material, the degree of peptide cross-linkage increases and the new material becomes O-acetylated (Gmeiner & Kroll, 1981; Dougherty, 1983a, b; Lear & Perkins, 1983, 1986, 1987; Gmeiner & Sarnow, 1987). Although the O-acetylation reaction occurs without a significant time lag after the incorporation of the new material, the rate of O-acetylation was lower than the rate of peptide cross-linking. These results and the absence of O-acetyl groups on the precursors led to the suggestion that O-acetylation must occur outside the cytoplasm at the newly polymerized peptidoglycan, presumably by a membrane-bound enzyme. Indeed, such membrane-associated O-acetyltransferases were identified recently.
An important and as yet unsolved question is the source of the acetyl group required for the O-acetylation reaction. The source of the acetyl group may vary from species to species. It has been suggested that the acetate is transferred from the N-2 position of either GlcNAc or MurNAc to the O-6 position of MurNAc in P. mirabilis (Dupont & Clarke, 1991b). The indirect evidence is based on two observations. First, radioactive, O-linked acetate can be recovered from the peptidoglycan of cells fed with [acetyl-3H]-GlcNAc. Second, the addition of other potential sources for acetate, pyruvate, acetyl phosphate or acetate, does not interfere with the amount of radioactive O-acetate. The formation of radioactive, O-linked acetate was also observed in (ether- or toluene-) permeabilized cells of P. mirabilis upon addition of [acetyl-3H]-UDP-MurNAc, a soluble precursor for peptidoglycan biosynthesis, thus confirming a possible mechanism of N→O acetyl transfer (Dupont & Clarke, 1991b, c). By contrast, the peptidoglycan of P. mirabilis does not contain N-deacetylated sugars in its peptidoglycan, excluding a transfer mechanism in the high-molecular-weight polymer. Instead, it was proposed that the acetate comes from soluble peptidoglycan turnover products, which are known to be released in high amounts from the peptidoglycan sacculus during growth of Gram-negative species (Clarke & Dupont, 1992). Thus far, there is no experimental evidence for the existence of N-deacetylated turnover products, and the enzyme responsible for the acetyl transfer reaction has not been identified in P. mirabilis.
Other species without detectable N→O acetyl transfer reaction are likely to utilize cytoplasmic acetyl-coenzyme A (CoA) or acetyl phosphate as source of acetate. This would require the transport of the acetate moiety across the cytoplasmic membrane followed by its transfer to MurNAc residues of nascent peptidoglycan. This model is supported by the recent identification of peptidoglycan O-acetyltransferase genes that encode integral membrane proteins presumably capable of transporting acetate across the membrane (Bera et al., 2005). Two types of O-acetyltransferases have been described corresponding to different mechanisms of peptidoglycan O-acetylation (Fig. 5) (Clarke et al., 2000). The first mechanism involves a single protein which performs both the transport of acetate across the membrane and its transfer onto the peptidoglycan. Presumably, O-acetyltransferases of the OatA-type are such enzymes. The second mechanism involves two proteins, one for acetate transport across the membrane and the other for catalysing its transfer to MurNAc. The acetate transport genes of this system are unknown. Although experimental proof is lacking, there are several candidate genes for these O-acetyltransferases (designated Pat) which are unrelated to OatA. Interestingly, irrespective of the species, the pat gene is always clustered on the chromosome with one or two ape genes, the products of which have O-acetylpeptidoglycan esterase activity capable of removing O-acetyl groups from peptidoglycan. Therefore, it was suggested that pat/ape systems have a function in spatial and temporal regulation of the state of peptidoglycan O-acetylation, which is required for controlling the activities of autolytic enzymes (Weadge et al., 2005).
Peptidoglycan O-acetyltransferases of the OatA-type
OatA was the first known peptidoglycan O-acetyltransferase gene and was identified in Staphylococcus aureus (Bera et al., 2005). Inactivation of oatA results in a loss of peptidoglycan O-acetylation. Similarly, inactivation of the homologous adr gene results in the loss of O-acetylation in S. pneumoniae (Crisostomo et al., 2006). A peptidoglycan O-acetyltransferase activity has not yet been shown for the corresponding proteins. OatA and Adr belong to a large protein family (PFAM01757) with more than 1000 confirmed or hypothetical acyltransferases active on a variety of substrates (http://www.sanger.ac.uk//cgi-bin/Pfam/getacc?PF01757). Members of this protein family are present in Gram-positive and Gram-negative bacteria with or without known peptidoglycan O-acetyl modification, as well as in eukaryotic organisms such as Caenorhabditis elegans and Drosophila melanogaster. Some of these proteins might function in the acylation of O-antigen (lipopolysaccharide, LPS) in Gram-negative species such as Salmonella enterica. By contrast, many of the yet uncharacterized OatA-like proteins might O-acetylate peptidoglycan in Gram-positive species, for example in L. monocytogenes, Streptococcus pyogenes or Enterococcus faecalis. However, this has to be confirmed for each of these species by analysis of the peptidoglycan of mutant strains lacking the putative O-acetyltransferase gene.
Figure 6 shows a sequence alignment of OatA from Staphylococcus aureus (Q7A3D6), Adr from S. pneumoniae (NP_359459) and the homologous proteins from Enterococcus faecalis (NP_814528), L. monocytogenes (NP_464816) and N. gonorrhoeae (YP_207241). There are regions of high amino acid sequence conservation close to the N-termini of these proteins. They have 11 (10 in the case of the N. gonorrhoeae protein) predicted transmembrane regions, which are, in most cases, at similar positions within the sequences. The predicted integral membrane region is located at the N-terminal part and covers approximately two-thirds of the protein sequence. The C-terminal regions of these proteins are predicted to have an extracytoplasmic localization which is consistent with an extracytoplasmic substrate for O-acetylation.
Peptidoglycan O-acetylation and de-O-acetylation by Pat/Ape proteins
A hypothetical peptidoglycan O-acetyltransferase gene patA has been identified in the genome of N. gonorrhoeae by its sequence similarity to the algI, a Pseudomonas aeruginosa exoenzyme for the O-acetylation of alginate (Weadge et al., 2005). Both PatA and AlgI are members of the O-acyltransferase family 1 (PFAM03062) and have the sequence motifs characteristic of these proteins (Clarke et al., 2000; Weadge et al., 2005). This protein family is also named MBOAT for ‘membrane-bound O-acyltransferases’ (Hofmann, 2000). It is a very large family and includes, for example, Porcupine, a modulator protein of developmental proteins in D. melanogaster, and the bacterial DltB enzymes, which presumably attach d-alanine to teichoic acids. Thus far, none of the proteins of this family has a confirmed peptidoglycan O-acetyltransferase activity, and there is no example of a mutant strain lacking the corresponding gene and having a defect in peptidoglycan O-acetylation. The following two points argue for a role of PatA in peptidoglycan O-acetylation (Weadge et al., 2005): first, N. gonorrhoeae does not produce alginate, making it likely that PatA has another exocytoplasmic substrate such as peptidoglycan; secondly, the patA gene is located on the chromosome of gonococci and other species immediately adjacent to one or two ape genes, the products of which remove O-acetyl groups from peptidoglycan (Weadge et al., 2005). At least 17 species contain a pat/ape gene cluster including N. meningitidis, N. gonorrhoeae, H. pylori, B. anthracis and B. cereus. Several of these species were not known to have O-acetylated peptidoglycan, and the presence of the pat/ape gene cluster in their genome prompted a reinvestigation of their peptidoglycan structure. Interestingly, of nine species tested, previously unknown O-acetylation of peptidoglycan was detected in eight species, including B. cereus, Bacillus fragilis, Campylobacter jejuni and H. pylori.
The O-acetylpeptidoglycan esterase activity of Ape1 from N. gonorrhoeae has been confirmed with the purified enzyme (Weadge & Clarke, 2006). The enzyme has a low level of activity against peracetylated xylan but is more active against O-acetylated peptidoglycan. Because inactivation of the ape1 gene was not possible, it has been suggested that the gene is essential (Weadge & Clarke, 2006). The Ape proteins identified in the databases have been divided into three families, Ape1, 2 and 3, based on sequence comparisons and on the organization of characteristic sequence motifs (Weadge et al., 2005). The Ape1 family was further subdivided into families Ape1a, Ape1b and Ape1c. All these proteins have an N-terminal signal sequence for the sec-dependent transport across the cytoplasmic membrane, which is consistent with the exocytoplasmic localization of their substrate, the O-acetylated peptidoglycan. Some Ape proteins are predicted to stay associated with the cytoplasmic membranes whereas others have a predicted cleavage site for removal of the signal peptide.
Biological roles of peptidoglycan O-acetylation
Soon after the discovery of peptidoglycan O-acetylation it was shown that this modification renders peptidoglycan resistant to the hydrolytic activity of hen egg white lysozyme (Brumfitt et al., 1958). In fact, most of the known muramidases have decreased or no activity against O-acetylated peptidoglycan, with only a few exceptions [e.g. N,O-diacetylmuramidase from Chalaropsis (Hash & Rothlauf, 1967) and mutanolysin from Streptomyces globisporus (Hamada et al., 1978)]. In addition, there was a direct correlation between the degree of O-acetylation of peptidoglycan samples obtained from 14 strains of P. mirabilis and the decreased activity of both human and egg white lysozyme (Dupont & Clarke, 1991a). O-Acetylation of peptidoglycan contributes to lysozyme resistance of pathogenic Gram-positive bacteria such as Staphylococcus aureus and S. pneumoniae and mutants of both species lacking peptidoglycan O-acetylation became more sensitive to exogenous lysozyme (Bera et al., 2005; Crisostomo et al., 2006). Interestingly, there is good correlation between pathogenicity, lysozyme resistance and the occurrence of O-linked acetate in the peptidoglycan of staphylococcal species (Bera et al., 2006). All 17 pathogenic staphylococcal species tested were resistant to lysozyme. A subset of seven species was further tested for the presence of O-linked acetate in the peptidoglycan, and all of those species contained O-linked acetate (Table 1). In contrast, 18 nonpathogenic species of staphylococci were all sensitive to lysozyme, and none of the four species tested (Staphylococcus arlettae, Staphylococcus carnosus, Staphylococcus condimenti and Staphylococcus lentus) contained O-linked acetyl groups in its peptidoglycan. Therefore, the O-acetylation of peptidoglycan is an important factor contributing to lysozyme resistance of pathogenic staphylococci. Next to O-acetylation, the loading of the cell wall with teichoic acid and the high degree of peptide cross-linkage in the peptidoglycan contribute to lysozyme-resistance in Staphylococcus aureus (Bera et al., 2007).
Strikingly, peptidoglycan O-acetylation is linked by (an) as yet unknown mechanism(s) to the cross-linking reaction in the synthesis of the peptidoglycan. This transpeptidation reaction is catalysed by the penicillin-binding proteins (PBPs), which are the targets of β-lactam antibiotics such as penicillin (Sauvage et al., 2008). The degree of O-acetylation is significantly reduced upon treatment with penicillin, and this is true for different species, including N. gonorrhoeae (Blundell & Perkins, 1981; Dougherty, 1983a, 1985), Staphylococcus aureus (Burghaus et al., 1983; Sidow et al., 1990) and P. mirabilis (Martin & Gmeiner, 1979). In N. gonorrhoeae, the decreased level of O-acetylation was indirectly linked to the inactivation of the carboxypeptidase PBP2 by penicillin (Dougherty, 1983a, 1985). PBP2 is involved in maturation of the newly synthesized peptidoglycan by trimming the pentapeptides to tetrapeptides. Therefore, it was proposed that peptidoglycan containing pentapeptides might be a poor substrate for O-acetylation. In S. pneumoniae, there was a different but related effect of O-acetylation on the expression of penicillin resistance. Inactivation of the adr gene encoding for the O-acetyltransferase resulted in the disappearance of O-acetyl groups in the penicillin-resistant mutant strain Pen6. Inactivation of adr resulted also in a significant reduction in the minimal inhibitory concentration of penicillin, indicating that a functional adr gene is required for the expression of high-level penicillin resistance in S. pneumoniae (Crisostomo et al., 2006). It was proposed that the O-acetyl modification in the peptidoglycan is required for the functioning of the penicillin-insensitive PBPs in the presence of antibiotic.
In penicillin-treated cells of Staphylococcus aureus, a decrease in O-acetylation occurred before the onset of lysis, suggesting that the decreased level of O-acetylation might be a prerequisite for autolysis (Sidow et al., 1990). In contrast, treatment of Staphylococcus aureus with bacteriostatic antibiotics (e.g. chloramphenicol or erythromycin) increased the level of peptidoglycan O-acetylation. It was speculated that treatment with these antibiotics might mimic the stationary phase of growth, during which the level of peptidoglycan O-acetylation is known to be increased (Burghaus et al., 1983). An increased level of O-acetylation was also observed in stationary cells of different Enterococcus species, and the degree of O-acetylation was particularly high in cells of Enterococcus faecalis in the viable but nonculturable state (Pfeffer et al., 2006). In this species, there were several autolysins present, which were affected in their activity by the presence of O-acetyl groups, whereas other autolysins were unaffected. Lytic transglycosylases are peptidoglycan hydrolases that are particularly abundant in Gram-negative species. For example, E. coli has six known lytic transglycosylases (Vollmer, 2006). They cleave the β-1,4-glycosidic linkage between MurNAc and GlcNAc, with the concomitant formation of a 1,6-anhydro ring at the MurNAc residue of the released product (Höltje et al., 1975; Höltje, 1996). Lytic transglycosylases cannot cleave the glycan strands at O-acetylated MurNAc residues because they require a free C6-OH group to form the 1,6-anhydro ring. Therefore, it has been proposed that O-acetylation and de-O-acetylation could be a means for spatial and temporal regulation of the activities of the lytic transglycosylases in certain Gram-negative species such as Neisseria sp. or H. pylori (Fig. 5) (Weadge et al., 2005; Weadge & Clarke, 2006, 2007).
The glycan strands in E. coli and other Gram-negative bacteria do not terminate with a reducing MurNAc residue but with 1,6-anhydroMurNAc (Höltje et al., 1975; Harz et al., 1990; Quintela et al., 1995). Also, the Gram-positive B. subtilis contains a low proportion (0.4%) of 1,6-anhydromuropeptides (Atrih et al., 1999), whereas another Gram-positive species, Staphylococcus aureus, did not contain this modification at all (Boneca et al., 2000). Muropeptides with a 1,6-anhydroMurNAc residue are released from peptidoglycan sacculi by the activity of lytic transglycosylases in a process termed peptidoglycan turnover (Vollmer & Höltje, 2001). It is not yet clear whether the 1,6-anhydroMurNAc residues present in the sacculus were formed by cleavage by lytic transglycosylases, or if they were formed by the synthetic transglycosylases when the glycan strand polymerization reaction terminated (Höltje, 1998). Interestingly, the 1,6-anhydroMurNAc-containing turnover products are used as signalling molecules for the induction of chromosomally encoded β-lactamase in some Gram-negative bacteria, e.g. Citrobacter freundii and Enterobacter cloacae (Höltje et al., 1994; Jacobs et al., 1994). The inhibition of peptidoglycan synthesis by β-lactams causes an uncontrolled activity of autolytic enzymes, which is characterized by a sudden increase in murein turnover products that are taken up into the cytoplasm. A transcriptional activator, AmpR, is inactivated by the binding of murein precursor UDP-MurNAc pentapeptide, but becomes activated by binding a turnover product 1,6-anhydroMurNAc-tripeptide, leading to the expression of AmpC β-lactamase.
Attachment of surface polymers to the glycan strands
Muramic acid-6-phosphate was identified by Liu & Gotschlich (1967) in cell-wall hydrolysates from eight Gram-positive species including different streptococci, S. pneumoniae, Staphylococcus aureus, Micrococcus lysodeikticus and Mycobacterium butyricum, and it was absent in peptidoglycan hydrolysates from Gram-negative E. coli and Neisseria catarrhalis (now Moraxella catarrhalis). In the cell wall of Gram-positive species, the phosphate residue is part of a phosphodiester bond, which links other cell-wall polymers to peptidoglycan. These polymers include the anionic teichoic acids and teichuronic acids, as well as capsular polysaccharides and the arabinogalactans. The latter are present only in mycobacteria, corynebacteria and nocardia. In most species, these cell-wall polymers are attached via a linkage unit to the glycan strands of the peptidoglycan. The structures of different linkage units are shown in Table 2. In most of the cases examined, the linkage is via a phosphodiester bond to the C-6 OH group of MurNAc residues. The type III capsular polysaccharide of Streptococcus agalactiae is the only known surface polymer that is linked via a phosphodiester bond and an oligosaccharide linker molecule (of unknown structure) to GlcNAc residues in the peptidoglycan (Deng et al., 2000). Very little is known about how these cell-wall polymers are attached to the peptidoglycan. In a few instances, the polymerization and attachment of cell-wall polymers could be observed in vitro (Elliott et al., 1975; Hancock & Baddiley, 1976; Kaya et al., 1983; Hancock et al., 2002; Yagi et al., 2003; Mills et al., 2004; Freymond et al., 2006; Seidel et al., 2007). However, the genetics and enzymology of these reactions are still unknown.
Table 2. Attachment of surface polymers to glycan strands in peptidoglycan in Gram-positive species
As summarized in this review, the oligo-(GlcNAcMurNAc) strands in the peptidoglycan are structurally modified during or soon after their synthesis in many bacteria. It will not be surprising if the known modifications are detected in additional species or if novel modifications are identified when modern analytical techniques are applied to peptidoglycans of other species. Only recently, the first genes for N-deacetylation, N-glycolylation and O-acetylation were identified and some of the corresponding enzymes were characterized. For example, biochemical data and crystal structures exist for few peptidoglycan deacetylases. However, the different pathways of O-acetylation of peptidoglycan are far less clearly understood. Future research in this area is required to identify the source of the acetate, to characterize the enzymatic activities of OatA- and Pat-like O-acetyltransferases, and to determine their structures. Linkage of secondary polymers such as teichoic acids, capsular polysaccharides or arabinogalactans to peptidoglycan strands is likely to be crucial for proper cell-wall architecture and integrity in Gram-positive bacteria. The structure of the linkage between these cell-wall polymers is known only for a limited number of bacterial species, and we have a complete lack of knowledge of the enzymes responsible for attachment of secondary cell-wall polymers to the glycan strands in peptidoglycan. Finally, further research is required to decipher the role of the different cell-wall modifications in bacterial physiology and in the interplay of commensals and pathogens with host factors during colonization and infection.
I thank Colin Harwood for critical reading of the manuscript. This work was supported by the European Commission through the EUR-INTAFAR project (LSHM-CT-2004-512138) and by the ‘Deutsche Forschungsgemeinschaft (DFG)’ within the ‘Forschergruppe Bakterielle Zellhülle (FOR 449)’.