O-Acetylation of Peptidoglycan Is Required for Proper Cell Separation and S-layer Anchoring in Bacillus anthracis*

O-Acetylation of the MurNAc moiety of peptidoglycan is typically associated with bacterial resistance to lysozyme, a muramidase that serves as a central component of innate immunity. Here, we report that the peptidoglycan of Bacillus anthracis, the etiological agent of anthrax, is O-acetylated and that, unusually, this modification is produced by two unrelated families of O-acetyltransferases. Also, in contrast to other bacteria, O-acetylation of B. anthracis peptidoglycan is combined with N-deacetylation to confer resistance of cells to lysozyme. Activity of the Pat O-acetyltransferases is required for the separation of the daughter cells following bacterial division and for anchoring of one of the major S-layer proteins. Our results indicate that peptidoglycan O-acetylation modulates endogenous muramidase activity affecting the cell-surface properties and morphology of this important pathogen.

Peptidoglycan, the major component of bacterial cell walls, consists of a polymer of the disaccharide N-acetylmuramic acid (MurNAc) 2 -(␤-1,4)-N-acetylglucosamine (GlcNAc) and associated stem peptides. Neighboring glycan strands are cross-linked via their peptide moieties, thereby creating a thick three-dimensional molecular meshwork that surrounds the cell. The sacculus formed by this polymer comprises a rigid exoskeleton responsible for bacterial shape and resistance to cytoplasmic turgor pressure. Given the importance of this structure, its synthesis and maintenance are targets of numerous antimicrobial compounds. In particular, lysozyme, which is found at ϳ5 mg/ml in mucosal secretions and serves as the major effector of innate immunity (1), hydrolyzes the ␤-1,4 glycosidic bond between MurNAc and GlcNAc.
Not surprisingly, bacteria have evolved strategies to inhibit the action of such hydrolytic enzymes (2). One mechanism involves additional structures such as S-layers or capsules around the peptidoglycan layer to restrict access of externally acting enzymes. Another relies on postsynthetic modification of the peptidoglycan macromolecule. In Bacillus anthracis, the etiological agent of anthrax, N-deacetylation of the MurNAc and GlcNAc residues has been demonstrated to confer resistance to lysozyme (3). A variety of other bacteria, both Gram-positive and Gram-negative, add acetate to the C-6 hydroxyl group of MurNAc residues (reviewed in Ref. 4), which inhibits lysozyme through steric hinderance. The extent of this modification varies with bacterial strain and culture age and ranges from 15% to over 70%. A direct correlation between the level of peptidoglycan O-acetylation and lysozyme resistance has been reported for several bacteria, including the important human pathogens Staphylococcus aureus (5) and Neisseria gonorrhoeae (6). Peptidoglycan Oacetylation is known to be a maturation event, occurring within the peptidoglycan sacculus following its synthesis and assembly (4).
Two distinct families of membrane bound O-acetyltransferases are responsible for peptidoglycan O-acetylation (7). With S. aureus, mutation of the oatA gene, encoding a putative O-acetyltransferase, drastically decreases the O-acetyl peptidoglycan levels and renders the bacteria sensitive to lysozyme (8). The production of OatA for peptidoglycan Oacetylation appears to be confined to Gram-positive bacteria and homologs of S. aureus OatA have been found in Streptococcus pneumoniae, Enterococcus faecalis, and Lactococcus lactis (9 -12). Although OatA has not been characterized biochemically, it is classified as a member of the membranebound O-acetyltransferase (MBOAT) family of acyltransferases (13), and it is thus predicted to be an integral membrane protein possessing twelve membrane-spanning helices. Its mode of action has been postulated to involve the transfer of acetate from cytoplasmic pools of acetyl-CoA to peptidoglycan followings its translocation across the cytoplasmic membrane (Fig. 1).
Rather than using OatA, Gram-negative bacteria produce a second family of enzymes, called Pat, for the O-acetylation of their peptidoglycan. Genes encoding these enzymes have been identified in a number of species, including the human pathogens N. gonorrhoeae, N. menningitidis, Helicobacter pylori, and species of both Campylobacter and Proteus (7). These proteins are homologs of Alg, proteins involved in the O-acetylation of the exopolysaccharide secreted by Pseudomonas aeruginosa, alginate (7,14). In their respective genetic loci, the pat genes are presumably co-transcribed with one or two genes encoding putative type II secreted proteins, bearing  In addition to protecting bacteria from the innate immune response of hosts, the O-acetylation of peptidoglycan is postulated to control autolysin activity (19). Autolysins are a collection of enzymes endogenous to bacteria that cleave peptidoglycan. They are required for peptidoglycan biosynthesis, as well as for cell division and septation and their name is derived from the fact that their nonspecific activity results in cell lysis (20). Given this lethal potential, autolysin activity must be tightly regulated during the cell cycle. As with lysozyme, O-acetylation sterically hinders the function of many autolysins. With B. anthracis, the main autolysins required for cell separation remain unknown, but it has been suggested that the Sap protein, one of the two main S-layer proteins, is a cell wall hydrolase (21).
Previously, it was not known if B. anthracis O-acetylated its peptidoglycan. However, as noted above, this bacterium does produce an S-layer, an additional cell wall component composed of self-assembled proteins that encapsulates the cell. The major components of the S-layer, the Sap and EA1 proteins, are anchored to the cell wall using a secondary polysaccharide and are required for cell separation (22,23).
Herein, we demonstrate that B. anthracis does produce O-acetylated peptidoglycan and that, uncommonly, both families of peptidoglycan O-acetyltransferases, Oat and Pat, are used by this bacterium for full O-acetylation. Moreover, the O-acetylation of peptidoglycan is required for separation of B. anthracis cells, as well as for proper assembly and attachment of its S-layer.

EXPERIMENTAL PROCEDURES
Bacterial Strain Construction-The bacterial strains and plasmids used in this study are listed in Table 1, while the  oligonucleotides used are listed in supplemental Table S1. Standard procedures were used to prepare and handle recom-FIGURE 1. Pathways for the O-acetylation of peptidoglyan. A single integral membrane protein, OatA, in Gram-positive bacteria translocates acetate from cytoplasmic pools of acetyl-CoA across the cytoplasmic membrane and then transfers it directly to peptidoglycan. In Gram-negative bacteria, an integral membrane protein, PatA, functions like OatA to translocate acetate across the cytoplasmic membrane, but a second protein, PatB, localized to the periplasm receives the acetate for its transfer to peptidoglycan.  (24) and transformants were selected using either MLS (1 g/ml erythromycin and 250 g/ml lincomycin), 5 g/ml chloramphenicol, 100 g/ml spectinomycin, or 10 g/ml tetracycline. All reagents were from Sigma unless noted. Mutagenesis of B. anthracis Sterne used the plasmid pKS1 that is temperature sensitive for replication (25). Plasmid minipreps of pKS1 derivative plasmids transformed into E. coli JM110 cells were used to electroporate B. anthracis competent cells. The B. anthracis transformants were grown in LB at the non-permissive temperature (37°C) without antibiotic and selected for resistance to the desired antibiotic (tetracycline, kanamycin, or spectinomycin). For each transformation, 100 CFUs were screened for resistance to the selected antibiotic and erythromycin sensitivity, indicating that insertion of the plasmid and its subsequent excision had occurred. Deletion of the desired locus and the excision of the plasmid were confirmed by PCR using flanking oligonucleotide pairs. Details for the preparation of B. anthracis and B. subtilis strains are presented in supplemental materials.
Peptidoglycan Purification-A single colony of B. anthracis or B. subtilis was grown for 3 h at 37°C in 3 ml of either BHI ϩ 0.1% glycerol (BHIG) or Lennox LB, respectively. This culture was diluted 1:1000 into 300 ml of the same media prewarmed at 37°C. For B. anthracis, cells were harvested at A 600 ϭ 1.0 and for B. subtilis, cells were harvested at A 600 ϭ 0.7 by centrifugation (8,000 ϫ g, 10 min, room temperature), resuspended and washed once with 50 ml of 25 mM sodium phosphate buffer (pH 6.5). Peptidoglycan was isolated from these cells and purified using the boiling SDS procedure as previously described, taking care to maintain the pH at 6.5 (12).
Measurement of Peptidoglycan O-Acetylation-Purified insoluble peptidoglycan was resuspended in 25 mM sodium phosphate buffer (pH 6.5) and fragmented by continuous sonication for 2 min using a VCX 130 Sonics sonicator. Samples were treated with 500 mM NaOH (final concentration) for 30 min at room temperature to release any ester-linked acetate, neutralized using an appropriate volume of 500 mM H 2 SO 4 and subjected to ultracentrifugation (100,000 ϫ g, room temperature) using a Beckman Airfuge (Beckman Coulter, Mississauga, ON). Quantification of released acetate was performed using the Megazyme Acetic Acid Assay kit (Megazyme International Ireland, Ltd., Wicklow, Ireland). The extent of Oacetylation is presented as a percentage of muramic acid content, which was determined by quantitative aminosugar analysis of corresponding insoluble peptidoglycan pellets after their hydrolysis in 4 M HCl at 96°C, in vacuo, for 18 h (26).
The O-acetylation levels in both B. subtilis and B. anthracis strains were identical both prior to and following hydrogen fluoride treatment, thus excluding the possibility that the Oacetate levels measured originated from secondary cell wall polysaccharides (27,28).
Peptidoglycan Hydrolysis by Muramidases-The peptidoglycan samples were diluted in 25 mM sodium phosphate buffer (pH 6.5) to an A 600 ϭ 0.5. When appropriate, lysozyme was added to a final concentration of 200 g/ml and treated and untreated samples were incubated at 37°C for 1 h. The A 600 was measured, and the results expressed as a percentage of the untreated samples. The mutanolysin and PlyL digestions used the same experimental conditions, but with 5 g/ml of protein.
Purification of His 6 -PlyL-A DNA fragment containing ba4073 that encodes PlyL was obtained by digestion of the pET-22a derivative (29) with BamHI and NdeI and was ligated to pET15b digested with BamHI and NdeI. The resulting plasmid was transformed into E. coli BL21 (DE3) strain (strain JDE758). The His 6 -PlyL protein was overproduced and purified using Ni 2ϩ affinity chromatography (Qiagen). The specificity of the purification was verified using SDS-PAGE followed by Coomassie Brilliant Blue staining and immunoblotting against the His tag. The activity of the purified His 6 -PlyL was verified by zymogram analysis (30) using peptidoglycan of wild-type B. anthracis (JDB1575).
In Vitro Modification of Peptidoglycan-To remove the Oacetyl group from muropeptides, the peptidoglycan was incubated in 80 mM NaOH at room temperature for 3 h followed by washing with 25 mM sodium phosphate buffer (pH 6.5) (26). Peptidoglycan samples were N-acetylated according to the method of Vollmer and Tomasz (31) with the following modifications. A suspension of 2 mg/ml peptidoglycan in ddH 2 O (pH 6.5) was cooled on ice, to which 0.25 volumes of freshly prepared 5% acetic anhydride and 0.25 volumes of saturated NaHCO 3 were added to a final pH of 6.8. The mixture was first incubated for 30 min on ice with shaking and then incubated for 1 h at room temperature with shaking. The peptidoglycan was recovered by centrifugation at 20,000 ϫ g for 20 min, washed 3ϫ with sodium phosphate buffer (pH 6.5) and resuspended in buffer.
Fluorescent Microscopy of Vegetative Cells-B. anthracis cells were prepared for microscopy by growing strains for 16 h at 30°C on LB agar. A single colony was used to inoculate 1 ml of BHIG and further diluted 500-fold into 3 ml of BHIG. For the strains carrying pAD123 or its derivatives, chloramphenicol was added to the liquid medium. The cultures were incubated until A 600 ϭ 0.5, at which point, samples were taken. For visualization of membranes, the membrane dye FM4 -64 (100 mg/ml stock; Invitrogen) was added to a final concentration of 10 g/ml. The samples were centrifuged (3000 ϫ g, 1 min), and the cell pellet resuspended in mounting medium (90% glycerol, 1ϫ PBS) prior to mounting the cells on poly-L-lysine-coated cover slips. All samples were observed with a Nikon Eclipse 90i with a 100ϫ objective. Phase contrast or fluorescence images were captured by using a Hamamatsu Orca-ER and recorded and processed using the imaging software NIS elements BR2.30 (Nikon).
Assay of Autolysis-The autolysis rates of the wild type and the various mutant B. anthracis strains were analyzed according to Mesnage et al. (22). Briefly, the bacteria were grown in BHIG at 37°C until an A 600 ϭ 0.5. 10 mM of sodium azide was added to the cultures, and the A 600 was measured every 30 min. The results are expressed as a percentage of the initial value.
Analysis of Cell Surface and Supernatant Proteins-B. anthracis strains were grown on LB agar for 16 h at 30°C. For each strain, single CFUs were resuspended in 1 ml of LB and diluted 1000-fold into 50 ml of LB and incubated in 300 ml of baffled flasks shaken at 37°C until A 600 ϳ0.5. The bacteria were centrifuged (3000 ϫ g, 5 min, room temperature) using a swinging bucket centrifuge. Filtered supernatant (0.45 m) was supplemented with 200 M phenylmethylsulfonyl fluoride (PMSF) and was concentrated by trichloroacetic acid (TCA) precipitation (10% [w/v], 4°C, 20 min), centrifugation (20,000 ϫ g, 15°C, 5 min), and three washes in 96% ethanol. The dried pellet was prepared for SDS-PAGE by resuspending in 1/200 of the initial volume in SDS-PAGE sample buffer (1% (w/v) SDS, 1 mM EDTA, 10% (v/v) glycerol, 5% (v/v), ␤-mercaptoethanol, 0.0025% (w/v) bromphenol blue, and 50 mM Tris-HCl, pH 7.5); the suspension was then boiled for 3 min at 100°C and loaded on a 10% SDS-PAGE. LiCl extracts were made by suspending the cell pellet in 1/100 of the initial culture volume of 4 M LiCl, 200 M PMSF, and 50 mM Tris-HCl (pH 7.5), followed by incubation with constant agitation at 4°C for 1 h. The cells were removed by centrifugation (20,000 ϫ g, 4°C, 5 min), and the supernatant was concentrated by TCA precipitation and prepared for SDS-PAGE as described above. The protein samples were run on 10% acrylamide gels, then fixed and stained using Coomassie Brilliant Blue.

B. anthracis Peptidoglycan is O-Acetylated-
The presence and levels of O-linked acetate associated with highly purified peptidoglycan from either vegetative or spore forms of B. anthracis Sterne 34F2 were determined. Relative to muramic acid content, ϳ40% (38.1 Ϯ 3.6%) of the muropeptides obtained from growing cells of B. anthracis (A 600 ϭ 0.8) were O-acetylated. The extent of this O-acetylation underwent a modest, but significant increase 12 h after exiting exponential phase to 58.1% (Ϯ 3.5%). However, the level of muropeptide O-acetylation was found to be reduced to 33% (Ϯ 2.4%) in spore peptidoglycan.
Identification of Both Oat and Pat Enzymes for the O-Acetylation of B. anthracis Peptidoglycan-We searched the B. anthracis genome for genes encoding hypothetical analogs of proteins known to be involved in peptidoglycan O-acetylation in other Gram-positive species. Thus, using S. aureus OatA (ZP_06848822) as the querry, a tblastn search led to the identification of bas1490 (YP_027759), and bas5308 (YP_031545) as potentially encoding OatA homologs. These two ORFs encode hypothetical proteins of 397 and 343 amino acids with theoretical molecular masses of 46,017 Da and 39,447 Da, respectively. They are thus considerably smaller than S. aureus OatA, which is composed of 604 amino acids and has a theoretical mass of 69,455 Da. Nonetheless, like OatA, both are predicted to be integral membrane proteins with ten hypothetical membrane spanning helices (Fig. 2). However, sequence analysis revealed that different regions of the two bas ORFs align with OatA, suggesting that only one, if not both, may function as an O-acetyltransferase.
A tblastn search of all bacterial genomes using bas5308 as the query identified a relatively small number of homologs. A phylogenic analysis generated from an alignment of these hypothetical protein sequences with other OatA homologs indicated a distant relationship. As seen in Fig. 3, the OatA homologs are confined almost exclusively to the Firmicutes, with most from either the Bacillales or the Lactobacillales, followed by the Clostridia; the lone remaining homolog is from the Actinobacteria family. No OatA homolog has been found in any Gram-negative bacteria. Whereas the majority of the bas5308 homologs are also from the Firmicute bacteria (Bacillales, Lactobacillales, and Clostridia), others are from the Bacteroidetes and two would appear to be from two different families of Gram-negative bacteria. Again, the difference between bas5308 and bas1490 is reflected in the phylogram, as the latter was the most distantly related to each of the other sequences (Fig. 3).
To confirm the function of Bas1490 and Bas5308 as OatA paralogs, we generated insertional mutations in these genes and analyzed the level of peptidoglycan O-acetylation of the resultant strains. Whereas there was no substantial effect of the bas1490 mutation on peptidoglycan O-acetylation, mutation of bas5308 resulted in a significant reduction relative to wild-type levels ( Table 2). Unexpectedly however, this loss was not complete, as cells were found to retain over 70% of their original acetylation levels (33% O-acetylation compared with 46% for the wild type) ( Table 2). Nonetheless, these data suggested that Bas5308 is a functional peptidoglycan O-acetyltransferase and given that both it and its homologs are smaller than OatA and that they appear to be distantly related (Fig. 3), we have renamed Bas5308 as OatB.
As it was clear that B. anthracis must possess an additional system for the O-acetylation of its peptidoglycan, a tblastn search of its genome was conducted using P. aeruginosa algI as the query. Surprisingly, not one, but two genes, bas0844 and bas0845, were identified that have predicted amino acid sequences with 41 and 59% similarity to AlgI, respectively. More importantly, these two hypothetical proteins are 27 and 29% identical and both are 62% similar to N. gonorrhoeae PatA, a putative peptidoglycan O-acetyltransferase (16, 18). The bas0844 and bas0845 genes would appear to be transcribed divergently and they exist immediately upstream of bas0843 (YP_027119) and bas0846 (YP_027122) (Fig. 4), genes that were identified previously as encoding potential acetylesterases (7). However, it has been demonstrated recently that the N. gonorrhoeae homolog, while resembling an  FEBRUARY 18, 2011 • VOLUME 286 • NUMBER 7 acetylesterase, actually functions as an acetyltransferase, and it was subsequently renamed PatB (16).

O-Acetylation of B. anthracis Peptidoglycan
To investigate the possible role of these gene products in the O-acetylation of B. anthracis peptidoglycan, we constructed strains carrying insertional mutations in either bas0844 or bas0845, aware that such insertions might have a polar effect on the expression of bas0843 and bas0846, respectively. Both of these mutant strains exhibited a similar reduction in O-acetyl peptidoglycan content to ϳ30% (Table  2). Based on both these data and the sequence alignments, bas0844 and bas0845 were renamed patA1 and patA2, respectively. A strain carrying the double mutations of both ⌬patA1 and ⌬patA2 exhibited a lower level of peptidoglycan O-acetylation than the respective single mutant strains, sug-  gesting a partial redundancy in the functions of the two gene products. The triple mutant strain JBD2583 (⌬oatB, ⌬patA1, and ⌬patA2) was almost devoid of the modification ( Table 2). As no in vitro assay has been developed to monitor peptidoglycan O-acetylation by either the Oat or Pat proteins (16), we further studied these putative O-acetyltransferases phenotypically. Thus, each of B. anthracis oatB, patA1, and patA2 were expressed individually in wild-type B. subtilis, a bacterium that was not known previously to be O-acetylated, and then measuring the subsequent changes in O-acetylpeptidoglycan levels. Initially, we measured the peptidoglycan Oacetylation of B. subtilis (PY79), an organism that is highly sensitive to lysozyme (32). Surprisingly, this peptidoglycan was ϳ35% O-acetylated (Table 3). A potential source of this activity in B. subtilis is yrhL, which encodes a hypothetical peptidoglycan O-acetyltransferase homologous to S. aureus OatA (Fig. 3.). Consistent with this assignment, a deletion of yrhL, which we have renamed oatA, reduced the level of Oacetyl-muropeptide content by ϳ3-fold as compared with wild type (Table 3). When this gene was expressed from an ectopic locus in the ⌬oatA strain, peptidoglycan O-acetylation was restored to wild-type levels. In contrast, expression of either the B. anthracis oatB, patA1 or patA2 genes individually in the B. subtilis ⌬oatA mutant strain led to only a modest restoration of peptidoglycan O-acetylation ( Table 3). The Oacetylation of B. subtilis ⌬oatA peptidoglycan increased significantly only upon co-expression of patA1 with bas0843, further suggesting that both genes of the putative OAP cluster are required for peptidoglycan O-acetylation and that the insertion into the patA1 gene results in the inactivation of the entire OAP1 cluster.

Resistance of B. anthracis Peptidoglycan to Lysozyme-By
analogy with other organisms, the resistance of B. anthracis to lysozyme could be due to O-acetylation of MurNAc within peptidoglycan. Consistent with the presence of this modification, incubation of a suspension of isolated wild type B. anthracis peptidoglycan with lysozyme did not result in a decrease in turbidity, whereas incubation with mutanolysin, a muramidase able to digest O-acetylated and N-deacetylated peptidoglycan, or with PlyL, a N-acetylmuramoyl-L-alanine amidase, led to a reduction in turbidity (Fig. 5). However, lysozyme did not hydrolyze peptidoglycan isolated from the single mutant strains ⌬oatB, ⌬patA1, ⌬patA2, the double mu-  bas0843 and bas0844 (designated patA1) ORFs of OAP1 and the bas0845 (patA2) and bas0846 ORFs of OAP2 are separated by 25 and 10 nucleotides, respectively. The intergenic region between patA1 and patA2 is 227 nucleotides. The bas0841/sap and bas0842/eag genes encode for the S-layer proteins Sap and EA1, respectively. bas0847 is a small, non-conserved ORF of unknown function. The bas0848 gene encodes a putative membrane protein that is not homologous to any known protein, but it is conserved in members of the B. cereus group. The relative sizes of the hypothetical proteins encoded by the OAP clusters are depicted below and the consensus motifs that define the PatA and PatB proteins (7) are shown.  tant strain ⌬patA1⌬patA2 or the triple mutant strain ⌬patA1⌬patA2⌬oatB any more than peptidoglycan derived from the wild type strain (Fig. 5). Similarly, chemical de-Oacetylation of the B. anthracis native peptidoglycan did not facilitate its hydrolysis by lysozyme, demonstrating that removal of O-acetyl groups alone is not sufficient to change peptidoglycan susceptibility to this enzyme. Previous in vitro studies with B. anthracis peptidoglycan have shown that N-deacetylation of its MurNAc and GlcNAc moieties leads to an inhibition of its hydrolysis by hen egg lysozyme (3). The removal of the N-acetyl group of the GlcNAc residues is performed in vivo by N-deacetylases (33). However, the B. anthracis genome encodes at least ten hypothetical N-deacetylases, of which at least six are likely to be specific for peptidoglycan, making a genetic approach to their study not feasible. Hence, we adopted a biochemical approach to investigate the effect of N-deacetylation on peptidoglycan susceptibility to lysozyme digestion. Under neutral conditions of pH, treatment of peptidoglycan with acetic anhydride results in the esterification and amidation of any free hydroxyl and amino groups, respectively. Incubation of a peptidoglycan suspension treated in this manner with lysozyme led to a 30% reduction in turbidity, confirming that N-deacetylation partially blocks the activity of lysozyme (Fig. 5). When this fully acetylated peptidoglycan was subjected to mild alkaline treatment, leading to de-O-acetylation, it became an even better substrate for lysozyme, confirming that O-acetylation, in addition to N-deacetylation, protects B. anthracis peptidoglycan from lysozyme.
OAP-mediated O-Acetylation Is Required for Cell Separation and Activity of Autolysins-The colony morphologies of B. anthracis ⌬oatB and ⌬patA1⌬patA2 strains grown on agar plates were indistinguishable from the wild type and liquid cultures of B. anthracis ⌬oatB exhibited similar growth kinetics as the wild type (data not shown). However, under identical conditions of growth, liquid cultures of the ⌬patA1 and ⌬patA2 mutant strains contained longer chains of cells compared with the wild type (24.6 Ϯ 6.3 and 20.7 Ϯ 6.1 cells per chain, respectively, versus 10.7 Ϯ 2.9 cells) (Fig. 6). Moreover, both ⌬patA1⌬patA2 and ⌬patA1⌬patA2⌬oatB mutant strains displayed even longer chains than the single mutants, and these cells were often twisted and hence not countable using routine microscopic procedures. Expression of patA1 alone under the control of its own promoter in the ⌬patA1 and ⌬patA1⌬patA2 mutants did not restore the wild-type cell morphology even though the protein was produced at levels comparable to wild-type (data not shown). However, wildtype morphology was restored in both of these mutants when patA1 was co-expressed with bas0843 (viz. the complete OAP1 cluster) (Fig. 6). This requirement of proteins to restore wild-type morphology is consistent with the finding that the co-expression of their genes significantly increases the O-acetylation level in both the B. anthracis ⌬patA1⌬patA2 and B. subtilis ⌬oatA mutants (Tables 2 and 3). Similarly, while expression of patA2 alone did not complement either the ⌬patA2 or ⌬patA1⌬patA2 mutations, co-expression with bas0846, representing the entire OAP2 cluster, resulted in complementation of each (data not shown).
MurNAc O-acetylation has been suggested to control the activity of autolysins (12,19,34). We therefore tested the autolytic activity of B. anthracis and the putative peptidoglycan O-acetyltransferase mutants by addition of sodium azide, a known inducer of autolysis (35), to growing cultures. The ⌬patA1, ⌬patA2, ⌬patA1⌬patA2, and ⌬patA1⌬patA2⌬oatB mutant strains did not undergo lysis under these conditions, as the optical density of the cultures did not change after 140 min of incubation ( Fig. 5 and not shown). In contrast, cultures of the B. anthracis ⌬oatB strain underwent autolysis at a level comparable to that of the wild type (Fig. 7).
OAP-mediated O-acetylation Is Required for Anchoring of a S-layer Protein-B. anthracis is known to produce an S-layer mainly comprised of two proteins named Sap and EA1 (22,23). Under the growth conditions tested, Sap, with an apparent size of 94 kDa as determined by mass spectrometry, was found to be mostly cell-associated with only a minor fraction detected in the extracellular milieu (Fig. 8). This pattern of  localization was found to be similar in each mutant strain that was shown to have only partial loss of peptidoglycan O-acetylation (Fig. 8). Thus, the majority of Sap remained cell associated in each of the single ⌬patA1, ⌬patA2, and ⌬oatB mutants. However, the distribution was noticeably different in the ⌬patA1⌬patA2 double mutant. This strain exhibited a 4-fold reduction in the amount of cell wall associated Sap and a corresponding increase in the amount found in the extracellular milieu (Fig. 8). This difference was not seen in the single ⌬patA1 and ⌬patA2 strains, indicating a possible functional redundancy. Consistent with this interpretation, Sap was associated with the B. anthracis cell wall when either OAP1 or OAP2 was produced. Thus, a minimal amount of O-acetylation appears to be sufficient but necessary for the proper assembly of the S-layer in B. anthracis.

DISCUSSION
In this study, we have demonstrated for the first time that the peptidoglycan of both B. anthracis and B. subtilis is Oacetylated, and that B. anthracis is the first bacterium known to use both of the O-acetylation systems found in bacteria, viz., OatA of Gram-positive organisms and the two-component system of Gram-negative bacteria encoded by the OAP gene cluster. This is also the first report of bacterial cells requiring a minimal level of peptidoglycan O-acetylation for the proper assembly of an S-layer.
B. anthracis Peptidoglycan O-Acetylation-Peptidoglycan O-acetylation is observed in a wide variety of bacterial species including the human pathogens S. aureus, Campylobacter jejuni, H. pylori, N. gonorrhoeae, N. menningitidis, species of Enterococcus, and in all Proteus species (4,7,18). The reported extent of this modification ranges between 15 and 70% relative to the MurNAc content, although whether this variability reflects the particular growth phase of the cell cultures or actual differences in the extent of O-acetylation of these species is not known. Peptidoglycan derived from exponential phase growth of B. anthracis was ϳ40% O-acetylated, consistent with the presence of this modification in the closely related bacterium B. cereus (7). This modification increases along the growth curve and peptidoglycan derived from B. anthracis cells ϳ12 h following exit from exponential phase was more O-acetylated (ϳ60%). This difference could reflect differential activity during these growth phases of the O-acetyltransferases and/or the possible existence of O-acetyl esterases. Alternatively, the increased level of O-acetylation could simply be a result of reduced peptidoglycan turnover in postexponential growth (36), resulting in an accumulation of the modification.
In contrast, spore cortex peptidoglycan is produced de novo during a late phase of sporulation, suggesting that the observed O-acetylation must be the result of an active enzymatic process. Consistent with this hypothesis, the genes of both OAP clusters are also expressed during sporulation (37), although we do not know which specific OAP cluster is responsible for the modification, if not both. The observation of cortex O-acetylation is, to our knowledge, the first report of this modification in spores, and it could play a role in the known high resistance of these cells (38). Interestingly, the peptidoglycan of VBNC (viable but non culturable) E. faecalis cells is highly O-acetylated and this modification has been suggested to play a role in the resilience of this important pathogen (16).
As noted above, we are reporting for the first time the presence of O-acetyl groups on the peptidoglycan from both B. anthracis and B. subtilis. Peptidoglycan of vegetative cells and spores of B. subtilis and of B. anthracis has been analyzed in previous studies but this modification was not mentioned (39 -41). However, O-acetyl groups are labile when subjected to basic or acidic pH (34). To analyze the peptidoglycan of both species, we kept the pH around 6.8 throughout our experiments. The protocols used for isolating peptidoglycan in the earlier studies were performed under basic conditions or without any buffering system, thus likely stripping any O-linked acetate that would have been present and thereby explaining the apparent discrepancy with our results.
B. anthracis O-acetyltransferases-Proteins homologous to either S. aureus OatA or to N. gonorrhoeae PatA/B (i.e. products of the OAP gene cluster) mediate the O-acetylation of peptidoglycan. We describe here the unusual coexistence of these two unrelated types of peptidoglycan O-acetyltransferases in B. anthracis. Not only are both present, but also they are both necessary for full peptidoglycan O-acetylation during growth. The lack of a detectable phenotype associated with the second hypothetical Oat homolog, Bas1490, suggests that perhaps this protein is not synthesized under the growth conditions examined here. Alternatively, it is possible that, despite its significant, but distant sequence similarity to all other OatA/B homologs as depicted in Fig. 3, Bas1490 does not function as a peptidoglycan O-acetyltransferase.
The B. anthracis Pat proteins may have evolved by lateral gene transfer as suggested for their N. gonorrhoeae counterparts (16). Interestingly, PatA homologs are also found in the genome of B. cereus, while absent in the genomes of other Bacillaceae, like S. aureus, B. subtilis, or L. monocytogenes (7). The O-acetylation of peptidoglycan in B. anthracis involving the PatA homologs also required functional bas0843 or bas0846 genes of the putative OAP clusters. Although the proteins encoded by bas0843 and bas0846 do not share enough identity with PatB to identify them as homologs, they nonetheless possess the conserved motifs of SNGH acyltransferases, importantly including the predicted catalytic triad residues Ser, His, and Asp (Fig. 4). Our findings thus suggest that these two proteins function as O-acetyltransferase paralogs. Accordingly, we propose that they function like PatB as part of a two component system (16) to receive translocated acetate from PatA and subsequently transfer it to peptidoglycan (Fig. 1). Unfortunately, the complexity of manipulating both an integral membrane protein with an associated peripheral membrane protein for the acetylation of a completely insoluble substrate has, to date, precluded the development of an in vitro protocol to assay peptidoglycan O-acetylation for any bacterium and thus, our postulate cannot be tested biochemically.
Whereas a functional PatA1 or PatA2 appear to be required for the proper separation of dividing cells (discussed further below), this was not the case for OatB. Why the different phenotypes? One possibility is that the two enzyme systems associate with different peptidoglycan synthesizing complexes. While little is known about peptidoglycan biosynthesis in B. anthracis, two enzymatic systems are thought to act in E. coli, one for lateral cell wall growth and a second for division/septation (reviewed in Ref. 42). If an analogous organization exists in B. anthracis, our data suggest that PatA1 and/or PatA2 would interact predominantly with complexes at the septum. This would leave OatB to be more associated with lateral peptidoglycan synthesis. However, if true, one would expect that the ⌬oatB mutation would have had a greater effect on overall levels of O-acetylation compared with the ⌬patA1⌬patA2 double mutation. This hypothesis was found to not be true. Moreover, a fusion of PatA1 to a fluorescent protein (mCherry), expressed within the OA1 cluster, exhibits homogeneous membrane localization of the protein and complements the double patA1 patA2 deletion (data not shown). Alternatively, the different phenotypes could be explained by a difference in substrate specificity of the Pat and Oat enzymes. The method of detection of the O-linked acetate on the peptidoglycan does not differentiate between O-acetylation of MurNAc and GlcNAc residues. Thus, it is possible that one of the two enzymatic systems catalyzes the O-acetylation of GlcNAc residues. Although GlcNAc O-acetylation has not been detected in any peptidoglycan to date, the O-acetylation of this aminosugar has been found on other bacterial polysaccharides. For example, the GlcNAc associated with the oligosaccharide produced by the plant symbiont Rhizobium leguminosarum as its nodulation factor is O-acetylated (43). We are currently conducting a careful analysis of the GlcNAc residues of B. anthracis peptidoglycan to determine if this modification does indeed occur in this pathogen.
Physiological Role of B. anthracis O-Acetylation-As demonstrated with other bacteria, O-acetylation contributes to the resistance of B. anthracis peptidoglycan to hydrolysis by lysozyme (Fig. 5). The other contributing factor to this resistance was found to be the N-deacetylation of GlcNAc residues. To our knowledge, this is the first report of peptidoglycan protected from lysozyme hydrolysis by both forms of modification. With all other bacteria examined, either O-acetylation or N-deacetylation, but not both, is used to protect this essential cell wall heteropolymer from degradation.
The O-acetylation of spore cortex peptidoglycan was found to inhibit its lysis by lysozyme in vitro, but it is unlikely that this would be its physiological role. Intact spores are not vulnerable to the action of lysozyme because a spore coat surrounds the cortex, thus providing a permeability barrier to molecules Ͼ5 kDa and thereby preventing access of lysozyme to spore peptidoglycan in vivo (44). Therefore, the presence of this modification in spores suggests that it could play a role in controlling their autolysins in the same manner proposed for vegetative cells (12,19,34). Indeed, it was shown in this study that endogenous autolysins, at least those stimulated by the presence of azide, are influenced by the presence and/or absence of peptidoglycan O-acetylation. Nothing is known in detail about the complement of autolysins produced by B. anthracis. However, it has been shown in other bacteria that some autolysins have a substrate specificity for the presence of O-acetylation, while others are inhibited by it (34). Therefore, the defect in cell separation observed in the patA mutants could be the result of a failure of one or more of the B. anthracis autolysins to properly target and/or interact with peptidoglycan at the site of cell separation in the absence of O-acetylation. In addition, a similar mechanism of impaired substrate recognition could underlie the reduced autolysis observed in strains lacking functional PatA1 and PatA2.
Vegetative B. anthracis cells are surrounded by several structures including the paracrystalline S-layer located beneath the capsule. S-layer proteins are anchored to the cell wall through the binding of secondary cell wall polymers or directly to peptidoglycan (reviewed in 45). The finding that a decrease in peptidoglycan O-acetylation caused by the double ⌬patA1⌬patA2 mutation resulted in a defect in the anchoring of Sap, one of two major S-layer proteins of B. anthracis, suggests that this modification is necessary for proper S-layer attachment. Interestingly, the gene encoding Sap (bas0841) is located very near to the OAP1 and OAP2 clusters, although it is not in an operon-like arrangement (Fig.  4). As the addition of a pyruvyl group is necessary for Sap attachment (22), it is conceivable that peptidoglycan Oacetylation may be required for proper localization of secondary wall polysaccharides.
Role of Peptidoglycan Modifications in B. anthracis Pathogenesis-Alteration in peptidoglycan substitutions influences the survival in the host and the immunostimulatory properties of bacterial pathogens. For example, N-deacetylation of the GlcNAc residues in the peptidoglycan of L. monocytogenes, another important human pathogen and closely related to B. anthracis, is responsible for its resistance to lysozyme and survival following macrophage phagocytosis (46). A direct correlation has been observed between the extent of peptidoglycan O-acetylation in both N. gonorrhoeae (6) and P. mirabilis (4) and their susceptibility to hydrolysis by lysozyme. In fact, muramic acid from pathogenic staphylococci (e.g. S. aureus) is O-acetylated, whereas that from closely related, non-pathogenic species is not (5), suggesting that lysozyme resistance is a critical determinant of pathogenesis. It would appear that B. anthracis has evolved several mechanisms for peptidoglycan modification that may contribute to its success as a pathogen. Given the lability of peptidoglycan O-acetylation, the N-deacetylation of peptidoglycan is most likely the key mechanism of resistance to lysozyme in vivo (3).
The importance of interfering with peptidoglycan recognition during B. anthracis pathogenesis may be crucial for its survival. In the host, B. anthracis remodels its peptidoglycan as suggested by the abundant production of peptidoglycan hydrolases during infection (47). Because purified B. anthracis peptidoglycan stimulates an inflammatory response in vitro (48), controlling the composition of muropeptides shed during multiplication in the host may enable better survival of the pathogen. Indeed, the detection of MurNAc by Nod2, the eukaryotic intracellular muropeptide receptor, is altered by the conversion of MurNAc into anhydro-MurNAc (53). This conversion is catalyzed by bacterial lytic transglycosylases (19), which require a free C-6 hydroxyl group on muramyl residues for activity. Therefore, MurNAc O-acetylation may facilitate pathogenesis by interfering with innate immunity at the level of recognition.