Structural basis for the O-acetyltransferase function of the extracytoplasmic domain of OatA from Staphylococcus aureus

Many bacteria possess enzymes that modify the essential cell-wall polymer peptidoglycan by O-acetylation. This modification occurs in numerous Gram-positive pathogens, including methicillin-resistant Staphylococcus aureus, a common cause of human infections. O-Acetylation of peptidoglycan protects bacteria from the lytic activity of lysozyme, a mammalian innate immune enzyme, and as such is important for bacterial virulence. The O-acetylating enzyme in Gram-positive bacteria, O-acetyltransferase A (OatA), is a two-domain protein consisting of an N-terminal integral membrane domain and a C-terminal extracytoplasmic domain. Here, we present the X-ray crystal structure at 1.71 Å resolution and the biochemical characterization of the C-terminal domain of S. aureus OatA. The structure revealed that this OatA domain adopts an SGNH-hydrolase fold and possesses a canonical catalytic triad. Site-specific replacement of active-site amino acids revealed the presence of a water-coordinating aspartate residue that limits esterase activity. This residue, although conserved in staphyloccocal OatA and most other homologs, is not present in the previously characterized streptococcal OatA. These results provide insights into the mechanism of acetyl transfer in the SGNH/GDSL hydrolase family and highlight important evolutionary differences between homologous OatA enzymes. Furthermore, this study enhances our understanding of PG O-acetyltransferases, which could guide the development of novel antibacterial drugs to combat infections with multidrug-resistant bacterial pathogens.

Antimicrobial resistance is one of the leading healthcare burdens of the century and is only predicted to worsen. Current projections predict that antimicrobial resistant infections will overtake cancer as a leading cause of death worldwide by 2050 (1). Among the biggest threats is methicillin-resistant Staphylococcus aureus. Individuals infected with methicillin-resistant S. aureus are estimated to be 64% more likely to die than those infected with non-drug-resistant strains of S. aureus. Both the Centers for Disease Control and Prevention and the World Health Organization have highlighted the desperate need for the research and development of novel antimicrobials to combat multidrug-resistant infections (2,3). One approach researchers are taking is to search for ways to disarm bacteria with nontraditional therapeutic agents (4). By targeting virulence factors that significantly contribute to the ability of a bacterium to colonize a host or cause infection, it will be possible to prevent infection without otherwise affecting survivability of the bacterium. It is thought that drugs targeting virulence factors may suffer less from the development of resistance (4).
The peptidoglycan (PG) layer of Gram-negative and Grampositive bacteria is an essential component of the cell envelope involved in shape determination and resisting turgor pressure. PG is composed of a glycan backbone consisting of alternating N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc) residues. Glycan chains are cross-linked by short peptides attached to MurNAc residues to form a mesh-like sacculus surrounding the cytoplasmic membrane. The importance of this macromolecule is highlighted by the number of antimicrobials that target PG and steps in its biosynthesis pathway.
Lysozyme is a muramidase of the innate immune system that hydrolyzes the ␤-1,4-glycosidic bond between MurNAc and GlcNAc residues, causing bacterial cell lysis. Because of the unique nature of PG, released fragments serve as important recognition motifs for immune receptors, activating the immune response in the early stages of an infection (5). Many pathogens have therefore evolved a strategy to defend against the host immune system through modification to their PG.
One such modification is O-acetylation of the C6-hydroxyl of MurNAc residues of PG, which sterically hinders binding of lysozyme (6,7). This modification is widespread among Gramnegative and Gram-positive bacteria, but it is most predominant in pathogens (8 -11). Bera et al. (12) discovered that only pathogenic species of Staphylococcus produce O-acetylated PG, and they are resistant to lysozyme. The levels of PG O-acetylation can range from 20 to 80% depending on the organism, environmental conditions, and growth phase of the culture (13)(14)(15). For example, the levels of PG O-acetylation increase by 10 -40% as Enterococcus faecalis cells enter stationary the phase and a further 10 -16% as the cells enter the viable but nonculturable state (16). In addition to providing resistance to lysozyme, PG O-acetylation has important implications in virulence, including increasing disease severity and downstream complications (17,18), conferring resistance to bacteriocins (19) and ␤-lactam antibiotics (10), and influencing the immune response (19,20). PG O-acetylation is considered important for virulence in numerous pathogens such as S. aureus (8,17,20), Streptococcus pneumoniae (10), Listeria monocytogenes (19), Neisseria meningitidis (21), Neisseria gonorrhoeae (9,18), Helicobacter pylori (22), and E. faecalis (23).
In Gram-positive bacteria, the enzyme responsible for PG O-acetylation is O-acetyltransferase A (OatA), first identified in S. aureus (8). Homologs of OatA have since been identified in S. pneumoniae (10), L. monocytogenes (19), E. faecalis (23), Lactobacillus plantarum (24), Lactococcus lactis (25), and several other Staphylococcus species (12). OatA is a bimodular protein consisting of an N-terminal acyltransferase 3 integral membrane domain and an extracellular C-terminal SGNH/GDSLhydrolase domain. SGNH hydrolases are a large family of esterases and lipases that possess four consensus residues, Ser, Gly, Asn, and His, that comprise their active sites and are involved in their mechanism of action (26). The catalytic Ser of these enzymes is found in a GDSL sequence motif. The N-terminal domain of OatA is predicted to contain 11 transmembrane helices and is thought to shuttle acetyl groups across the cytoplasmic membrane to the C-terminal domain for their subsequent transfer onto PG (27). It is still unknown whether the two domains remain attached after translation; S. aureus OatA possesses a noncanonical signal peptidase site between the two domains, and the C-terminal domain alone has been detected in spent culture media (28).
We recently described the crystal structure of the C-terminal domain of OatA from S. pneumoniae and experimentally confirmed the function of this domain as an O-acetyltransferase with a reaction mechanism involving a Ser-His-Asp catalytic triad (29,30). Preliminary characterization of the C-terminal domain of S. aureus OatA was also performed, including identification of the putative catalytic triad residues. Furthermore, the substrate specificity of the C-terminal domains of S. pneumoniae and S. aureus OatA was investigated with regard to the stem peptide, and it was found that the enzymes had distinct preferences for muroglycans with tetra-and pentapeptide stems, respectively (29). Given that PG O-acetylation is a postbiosynthetic modification, occurring after incorporation of lipid II precursors into the pre-existing sacculus (31-34), OatA must work intimately with the PG biosynthetic machinery. Here, we present the crystal structure of the C-terminal domain of S. aureus OatA and characterization of its mechanism of action as an O-acetyltransferase. Examination of the active center suggests that the staphylococcal enzyme, as well as those produced by most other Gram-positive pathogens, use a novel process for preventing simple hydrolysis of the acetyl-enzyme intermediate compared with the previously characterized streptococcal OatA.

Crystallization and structure determination of SaOatA C
Efforts to crystallize the engineered extracytoplasmic domain of S. aureus OatA, encompassing residues 445-601 (SaOatA C ), were unsuccessful, despite the removal of potentially disordered regions that could hinder crystallization. This included significant truncation of the N-terminal interdomainlinker regions and two C-terminal lysine residues. To find additional areas of disorder, we analyzed the amino acid sequence of SaOatA C and found that it was enriched with residues of high conformational freedom (9.6% Lys and 5.1% Glu). Therefore, we used surface entropy reduction (35) in an effort to reduce surface disorder and promote crystallization of the domain. We identified three clusters of predicted high entropy surface residues: Lys 464 and Lys 465 , Lys 495 and Lys 496 , and Glu 551 and Lys 552 , using the SERp server. Each residue in these cluster pairs was replaced with alanine residues to produce three new constructs for crystallization. The specific activities of the resultant SaOatA C variants were similar to WT enzyme (Table 1), and all three variants crystallized.
(E551A/K552A)-SaOatA C crystallized only in the presence of zinc salts, so we suspected that Zn 2ϩ ions were bound to the protein and well-ordered. Indeed, we detected an anomalous signal from protein-bound zinc, and this signal was sufficient for phase determination using single-wavelength anomalous dispersion (Zn-SAD) ( Table 2). (E551A/K552A)-SaOatA C crystallized with a dimer in the asymmetric unit and contained three Zn 2ϩ ions, two of which are coordinated by the putative catalytic His and Asp residues at the dimer interface. We presume that these protein-bound Zn 2ϩ ions are not biochemi-

Structure of S. aureus OatA
cally relevant because they appear to distort the active site. This was supported by the observation that zinc, among other firstrow transition metal cations, inhibits SaOatA catalysis in vitro (36). Diffraction data were therefore collected for a crystal of (K495A/K496A)-SaOatA C that grew in the absence of zinc, and the structure was solved by molecular replacement using the (E551A/K552A)-SaOatA C structure as a search model. This structure was determined to 1.71 Å resolution and refined to R work /R free values of 16.6%/19.6% ( Table 2). The overall structure of (K495A/K496A)-SaOatA C adopts an ␣/␤-hydrolase fold with five parallel ␤-strands sandwiched between seven ␣-helices (Fig. 1). The asymmetric unit contains two molecules of the protein; however, size-exclusion chromatography revealed that SaOatA C is monomeric, and thus the dimeric conformation seen in the crystal is likely not physiologically relevant. A sodium ion was seen in the bend of ␣-helix 4 and ␤-strand 5, coordinated by three water molecules and the backbone carbonyl oxygens of Ala 550 , Arg 553 , and Val 556 . The putative catalytic triad Ser-His-Asp residues align within a shallow active site on the surface of the protein. In chain B, the catalytic His 578 is found in two conformations, facing both toward and away from the active site (occupancies of 0.54 and 0.46, respectively), suggesting that His 578 is inherently flexible (Fig. S1).
Searching for structural homologs with DALI revealed that SaOatA C most closely resembles the C-terminal domain of OatA from S. pneumoniae (SpOatA C ) (PDB code 5UFY) ( Fig.  2A). The SaOatA C structure is also homologous to numerous uncharacterized proteins proposed to belong to the SGNH/ GDSL hydrolase family. Of the characterized proteins in the PDB database, aside from SpOatA C , SaOatA C most closely resembles Axe2, an acetylxylan esterase from Geobacillus stearothermophilus (PDB code 4JHL), a family 3 carbohydrate esterase from Clostridium thermocellum (PDB code 2VPT), and a family 3 carbohydrate esterase from Talaromyces cellulolyticus (PDB code 5B5S) ( Fig. 2A). The most prominent difference between SaOatA C and its structural homologs is the geometry of the putative catalytic triad. The distances between N␦1 of His 578 and O␦2 of Asp 575 and between the ⑀2 of His 578 and O␥ of Ser 453 are 3.2 and 4.3 Å, respectively, which are significantly longer than in homologs (Fig. 2B). However, in the absence of a bound ligand, the crystal structure of SaOatA C could represent the inactive resting state of the enzyme and not the active conformation. Indeed, the dual conformations of His 578 seen in chain B of the crystal structure suggests that His 578 is flexible (Fig. S1), and the active site may undergo conformational changes during catalysis. Such conformational changes have been reported previously for SpOatA C and the PG esterase Ape1 (29,37).

SaOatA C uses a Ser-His-Asp catalytic triad
SGNH hydrolases are characterized by four consensus sequence blocks (I, II, III, and V) containing conserved residues, Ser, Gly, Asn, and His that, as noted above, give this family of enzymes their name (Ser 454 , Gly 475 , Asn 507 , and His 578 in SaOatA) (26). Most SGNH hydrolases possess a catalytic triad consisting of a Ser from block I and an Asp and His from block V. The putative catalytic triad residues of SaOatA C , Ser 453 , His 578 , and Asp 575 were previously replaced with Ala in a longer construct of the enzyme (SaOatA 435-603 ) and assayed for activity using p-nitrophenyl-acetate (pNP-Ac) (29). We repeated these experiments using the new construct with 4-methylumbelliferyl-acetate (4MU-Ac) as substrate. The stability of all SaOatA C variants was verified by a thermal shift assay after purification. The (S453A)-SaOatA C variant had no detectable activity, whereas the (H578A)-SaOatA C and (D575A)-SaOatA C variants had 2.08% and 1.02% residual esterase activity, respectively ( Table 1). As seen previously, WT SaOatA C displays very limited transferase activity toward chito-oligosaccharides as acceptors (29), and as such, it was not possible to determine the rates of transfer for all SaOatA C variants. A qualitative analysis of any reaction products by LC-MS demonstrated that the truncated WT SaOatA C retained transferase activity; however, the sensitivity of the assay is not sufficient to discriminate between no transfer and limited transfer that may be seen with weakly active SaOatA C variants (Fig. S2).
Most SGNH hydrolases employ a double-displacement reaction mechanism involving a covalent acyl-enzyme intermediate at the catalytic serine residue. Indeed, this reaction mechanism was recently confirmed for SpOatA C (30). We employed a sim- where I is the intensity of the reflection hkl, ⌺ hkl is the sum over all reflections, and ⌺ i is the sum over i measurements of reflection hkl. ϽI(hkl)Ͼ is the mean value of I(hkl). b CC1 ⁄ 2 is the Pearson correlation coefficient calculated between two random half data sets.
where F obs and F calc are the observed and calculated structure factors, respectively. R free was calculated identically except that all reflections belonged to a test set consisting of only 5% of the data, chosen at random.  (Fig. 3B). Based on the MS/MS fragmentation pattern of the m/z 1138.59 parent ion, we were able to map the acetyl modification to Ser 453 . This m/z 1138.59 ion was not detected among the digestion products of the nosubstrate control reaction.

SaOatA C possesses a typical three-component oxyanion hole
In most SGNH hydrolases, the oxyanion hole is typically formed by three conserved hydrogen-bond donors: the back-bone amide of the catalytic Ser of the block I consensus sequence, the backbone amide of Gly from block II, and the side-chain amide of Asn from block III. In SpOatA C , the block II Gly is replaced by a Ser, and the loop adopts a type I ␤-turn. In contrast, SaOatA C retains the Gly, and the loop adopts the typical type II ␤-turn seen in homologous SGNH esterases (Fig. 2B). The backbone amide of Gly 476 in SaOatA C faces the active site and thus likely participates in stabilizing the transition state. Interestingly, the N␦ 2 of block III Asn 507 is only 3.1 Å away from the O␥ of Ser 453 , closer than typically seen in other SGNH hydrolases. We tested the importance of Asn 507 in catalysis by its replacement with Ala. The (N507A)-SaOatA C variant had no detectable esterase activity toward 4MU-Ac, suggesting an important role in the catalytic mechanism (Table 1).
We also previously proposed the importance of a (V/I)(G/ S)(R/V) motif in the block II loop (29). In the resting-state structure of SpOatA C , a water molecule was observed coordinated by the backbone carbonyl of Val 460 and the backbone carbonyl of Val 462 (equivalent residues Val 475 and Arg 477 in SaOatA C ) (29). In contrast, because of the opposite turn of the block II loop in SaOatA C , the carbonyl of Val 475 does not face the active site, and no water molecule is seen coordinated at the active site in this position ( Fig. 4 and Fig. S3). Nonetheless, Val 475 is highly

Structure of S. aureus OatA
conserved among OatA homologs, and in accordance, we replaced Val 475 with Gly and saw complete abolishment of esterase activity (Table 1).

Conserved Asp 457 limits esterase activity
We observed a water molecule in the active site of SaOatA C coordinated by the O␦1 of Asp 457 , the backbone carbonyl of

Structure of S. aureus OatA
Ile 577 , and the O␥ of the catalytic Ser 453 that was not present in any of the structural esterase homologs or SpOatA C (Fig. 4). We investigated the role of the water molecule by substituting Asp 457 with Ala and Asn. Surprisingly, the D457A and D457N variants of SaOatA C displayed a 4-fold increase in esterase activity and still maintained the ability to transfer to chitooligosaccharides (Table 1). To investigate the prevalence of an Asp at this position in SGNH hydrolases/transferases, we analyzed sequences from 200 known and hypothetical OatA homologs from Gram-positive bacteria and characterized esterases from the SGNH hydrolase family. We found that an Asp is highly conserved in OatA homologs from the Staphyloccocus, Bacillus, Listeria, and Lactobacillus genera, among others (Fig.  S4). In contrast, an Asp residue was not found at this position (catalytic Ser ϩ 4) in any of the characterized SGNH hydrolases that naturally act as esterases (Fig. 2C). In most OatA homologs from Streptococcus species, an Arg residue is found in this position (Fig. S4). We propose that Asp 457 is a conserved feature in many SGNH family transferases and serves as a sentry to limit esterase activity through the coordination of a water molecule that could otherwise approach an acetyl-enzyme intermediate. It was observed that transferase activity also increased with the replacement of Asp 457 with Ala or Asn. The reason for this enhanced activity is not known, but it is possible that these replacements increase the accessibility of the pseudo-substrate chitooligosaccharides, in addition to water, used as acceptors for the in vitro assays of transferase activity.

Discussion
OatA belongs to the SGNH hydrolase family of enzymes along with numerous esterases with a wide range of substrate specificities. The mechanism by which OatA acts as a transferase was widely unknown until recently. The structure of SpOatA C elucidated structural features that distinguished it from SGNH hydrolase family esterases, including an inverted turn of the block II loop, a conserved valine in block II, a hydrophobic active site wall, and an atypical two-residue oxyanion hole (29). Unexpectedly, the active site of SaOatA C more closely resembles that of structurally homologous esterases than that of SpOatA C . SaOatA C has the conserved Gly in block II, and the loop adopts the typical type II ␤-turn seen in the homologous esterases. As a consequence, the water molecule that is coordinated by the backbone of Val 460 in the block II sequence of SpOatA C in its resting state is not seen in the structure of SaOatA C . Furthermore, replacement of Val 460 of SpOatA C with Gly or Ala increases esterase activity while resulting in loss of transferase activity (29). To explain this loss, it was proposed that Val 460 may contribute to the effective binding of the carbohydrate acceptor. In contrast, a comparative replacement of the homologous Val in SaOatA C , Val 475 resulted in a total loss of esterase activity. It is possible that replacement of Val 475 with Gly in SaOatA C disrupts correct positioning of the block II loop, which may impact the ability of the backbone amide of Gly 476 to stabilize the transition state; such stabilization of the transition state formed by SpOatA C does not appear to involve its block II loop (29).
A thorough kinetic analysis of SpOatA C confirmed that the enzyme employs a double-displacement reaction mechanism (30). Accordingly, we propose that SaOatA C follows a similar reaction mechanism (Fig. S5), wherein the carboxyl group of Asp 575 forms a salt bridge with a nitrogen atom in the imidazole ring of His 578 , enabling His 578 to deprotonate Ser 453 . The nucleophilic Ser 453 attacks the carbonyl carbon of the acetyl donor, generating a tetrahedral transition state. Residues at the active center of the enzyme form an oxyanion hole that stabilizes the transition state, which then collapses into a covalently bound acetyl-enzyme intermediate. We were able to unequivocally identify Ser 453 of SaOatA C as the site of acetylation (Fig.  3). The acetyl donor is released upon acquisition of a proton from His 578 . The glycan accepter, a MurNAc residue of the PG backbone, can then bind the active site cleft. His 578 , now acting as a base, abstracts a proton from the C6-OH of MurNAc, rendering the carbon atom nucleophilic and resulting in its attack on the carbonyl center of the acetyl-Ser 453 intermediate. This leads to the formation of a second tetrahedral transition state, collapse of which results in the release of the O-acetylated product and free enzyme.
We propose that the transition state is stabilized by the backbone amide of Ser 453 in block I, the side-chain amide of Asn 507 in block III, and the backbone amide of Gly 476 in block II. The three-residue oxyanion hole is typical of SGNH hydrolases, but distinguishes SaOatA C from SpOatA C , which appears to employ an oxyanion hole formed of two residues (29). Replacement of the SpOatA C block III Asn 491 with Ala gave 58% residual esterase activity (29). Further kinetic analysis suggested that Asn 491 may play a larger role in substrate binding than in stabilization of the first transition state (30). In contrast, replacement of Asn 507 of SaOatA C with Ala resulted in a complete loss of activity, suggesting that this residue may play a more critical role in SaOatA C . Absolute identification of oxyanion hole H-donors would require analysis of a ligand-bound structure, ideally with a covalently bound transition-state mimic. Such a structure was achieved for SpOatA C using the mechanistic inhibitor methanesulfonyl fluoride, forming a methylsulfonyladduct structure (PDB code 5UG1). Unfortunately, methanesulfonyl fluoride and related analogs do not significantly inhibit SaOatA C ; thus a different transition state mimic will need to be found.
Bioinformatic analysis shows that OatA homologs form two distinct clades, wherein the Streptococcus genus forms a phylo-

Structure of S. aureus OatA
genetically separate clade to Staphylococcus, Bacillus, and other genera (Fig. S4). The differences that we have observed between the structures of S. pneumoniae and S. aureus OatA suggest that the enzymes from these clades may use different mechanisms to minimize, if not prevent, water from serving as the acetyl acceptor during their respective double-displacement reaction mechanisms. We previously noted the occurrence of a conserved Val/Ile adjacent to the oxyanion hole block III Asn in Streptococcus OatA homologs, proposing that this hydrophobic residue may stabilize carbohydrate acceptor substrates (29). In contrast, in Staphylococcus and species from the same clade, this position is most commonly occupied by a Thr or Ser residue, which would not engage in the same hydrophobic interactions. We identified a water molecule coordinated by Asp 457 in the structure of SaOatA C and determined that this residue played an important role in limiting the esterase activity of the enzyme, while maintaining transferase activity. This suggests that the two distinct clades of OatA homologs utilize different mechanisms to preclude water from their active site to catalyze efficient and nonwasteful transfer of acetyl groups to peptidoglycan only. Our data suggest that the coordination of a water molecule in the active site by Asp 457 may be the method by which SaOatA C and the majority of OatA homologs belonging to the same phylogenetic clade favor transferase activity. Unfortunately, the absence of other OatA structures prevents us from verifying whether or not these structural features are conserved among homologs within the same clade. The reason for these differences also remains unknown. Perhaps the selective pressure for divergence into two clades was substrate specificity recognizing that the staphylococcal OatA O-acetylates MurNAc residues with pentapeptide stems, whereas the streptococcal enzyme has specificity for residues with tetrapeptide stems (29).
PG O-acetylation is a common modification employed by pathogenic Gram-positive bacteria as a means to evade the host innate immune system. Despite knowledge of the modification for decades, OatA from S. pneumoniae was the only PG O-acetylating enzyme from a Gram-positive bacteria characterized before this study. Our data reinforce the mechanism of action proposed for both S. pneumoniae OatA C and N. gonorrhoeae peptidoglycan O-acetyltransferase B (PatB) (30,38). We have previously validated SaOatA C and N. gonorrhoeae PatB as antibiotic targets with a high-throughput small-molecule screen (39). The structure of SaOatA C will assist in the design of anti-virulence drugs against OatA. Furthermore, our discovery of the differences between the active sites of S. pneumoniae and S. aureus OatA C is an important consideration in developing narrow-or broad-spectrum OatA inhibitors for the treatment of important human pathogens for which current antibacterial therapies are being threatened by multidrug resistance.

Cloning of C-terminal S. aureus oatA variants
The generation of SaOatA C(445-601) possessing single sitespecific amino acid replacements was achieved by site-directed mutagenesis. PCR products incorporating the desired mutations were obtained using KAPA HiFi polymerase with pDSAC71 (harboring oatA C encoding residues 445-601 of fulllength S. aureus OatA) as template and the appropriate primers listed in Table S1. Following PCR amplification, the reaction was incubated with Dpn1 (Thermo Fisher Scientific, Mississauga, Canada) for 1 h at 37°C, followed by transformation into Escherichia coli DH5␣. The sequences of all resultant plasmids were verified before use.

Overproduction and purification of SaOatA C
The genes encoding SaOatA C(445-601) and variants were expressed in E. coli BL21 (DE3) transformed with pDSAC71, and the overproduced recombinant proteins were purified by a combination of affinity chromatography and gel filtration as previously described (39). Gel filtration buffer consisting of 50 mM Tris, pH 7.5, 150 mM NaCl was used when the protein was being purified for the purpose of crystallography. In all other instances, the gel filtration buffer consisted of 50 mM sodium phosphate, pH 6.5, 150 mM NaCl. Fresh immobilized metal affinity chromatography resin was used for each SaOatA C variant to prevent cross-contamination.
The stability of each purified protein was assessed by SYPRO Orange thermal shift assay as a means to infer proper folding (40). Briefly, SaOatA C and variants were diluted to 5 M in 50 mM sodium phosphate, pH 7.0, and mixed with 2ϫ SYPRO Orange (Thermo Fisher Scientific) in 50-l reactions. The melting temperature (T m ) of each protein was determined using a StepOnePlus real-time PCR machine using a temperature gradient of 4 to 95°C over 60 min. The data were analyzed using the StepOnePlus software.

Crystallization
SaOatA C surface entropy variants were concentrated to 30 mg/ml by ultrafiltration using an Amicon Ultra-15 centrifugal filter (10-kDa molecular mass cutoff; Millipore) (4,000 ϫ g, 4°C). Commercial Midwest Center for Structural Genomics crystallization suite sparse matrix crystallization screens 1-4 (Microlytic North America Inc., Burlington, MA, USA) were prepared at room temperature with the E551A/K552A and K495A/K496A forms of SaOatA C . Crystallization screening by sitting-drop vapor diffusion was set up using a Gryphon robot (Art Robbins Instruments, Sunnyvale, CA, USA) with 1-l protein drops and a protein-to-reservoir ratio of 1:1 for a final drop volume of 2 l. Crystal trays were stored at 22°C. Optimization of crystal conditions was performed to produce crystals of (E551A/K552A)-SaOatA C in 0.008 M zinc acetate, 20% PEG 3350, and crystals of (K495A/K496A)-SaOatA C in 27% PEG 6000, 0.015 M sodium citrate.

X-ray diffraction data collection and structure determination
Crystals were cryoprotected for 30 s in reservoir solution supplemented with 60% (v/v) ethylene glycol prior to vitrification in liquid nitrogen. Zn-SAD data for (E551A/K552A)-SaOatA C were collected on Beamline 08B1-1 at the Canadian Synchrotron Light Source (Saskatoon, Canada). Native data for (K495A/K496A)-SaOatA C were collected on Beamline 17-ID2 at the National Synchrotron Light Source II (Upton, NY). The

Structure of S. aureus OatA
data were indexed and scaled using HKL2000 (41). Three zinc sites were located in the (E551A/K552A)-SaOatA C Zn-SAD data using HKL2MAP (42), and density modified phases were calculated using SOLVE/RESOLVE (43). The resulting electron density map was of good quality and allowed for PHENIX AutoBuild (44) to build 100% of the protein. Manual model building was done in COOT (45) alternated with refinement using PHENIX.REFINE (46). The structure of native (K495A/ K496A)-SaOatA C was determined by molecular replacement using PHENIX AutoMR (46) with the zinc-incorporated derivative as the search model. Manual model building and refinement was carried out as described previously. All molecular models were generated using PyMOL.

Steady-state kinetics of SaOatA C
The specific activity of SaOatA C acting as an esterase and transferase was determined as previously described (39). Briefly, SaOatA C (5 M) was incubated in 50 mM sodium phosphate, pH 6.5, at room temperature with 0.1 mM 4MU-Ac as substrate. For transferase assays, SaOatA C (5 M) was incubated in 50 mM sodium phosphate, pH 6.5, at 37°C with 0.1 mM 4MU-Ac as acetyl donor and 2 mM pentaacetyl-chitopentaose (Megazyme) as acetyl acceptor. Product release was monitored fluorometrically using a Synergy plate reader with excitation and emission wavelengths of 325 and 450 nm, respectively. Control reactions were performed using gel-filtration buffer in place of SaOatA C to account for spontaneous substrate hydrolysis. The rate of background hydrolysis was subtracted from the rate of reactions with enzyme to determine a rate of esterase activity. The rate of transfer was determined as the difference between the reaction rates with and without acceptor. Each reaction was performed in triplicate. Analyses and graphs were generated in GraphPad Prism 5.

Qualitative analysis of SaOatA C transferase activity
The ability of SaOatA C to transfer acetyl groups to chitooligosaccharides was determined by qualitative end-point analysis by LC-MS/MS. SaOatA C (5 M) was incubated with 0.1 mM 4MU-Ac and 1 mM pentaacetyl-chitopentaose (Megazyme) at 37°C for 18 h. The reaction products were separated by LC-MS/MS using an Agilent 1200 HPLC system interfaced with an Agilent UHD 6520 Q-TOF mass spectrometer (Agilent Technologies Inc., Santa Clara, CA, USA) housed in the Mass Spectrometry Facility of the Advanced Analysis Centre of the University of Guelph. Data analysis was performed using Mass-Hunter qualitative analysis, version B.06.00 (Agilent).

Trapping acetyl-SaOatA C intermediate
Direct observation of the covalent acetyl-SaOatA C intermediate was achieved by real-time analysis of a reaction by LC-MS as previously described (30) with minor modifications. The reaction mixture consisted of 5 M SaOatA C in 50 mM sodium phosphate, pH 6.5, with 1 mM pNP-Ac as acetyl donor and was incubated for 15 min in an MS-grade sample vial. A control reaction without pNP-Ac was also performed. The site of O-acetylation was identified by a tryptic digest of an acetonequenched reaction mixture of SaOatA C with pNP-Ac, as previ-ously described (30). In this case, the reaction mixture consisted of 5 M SaOatA C in 50 mM sodium phosphate, pH 6.5, with 1 mM pNP-Ac, incubated for 30 min in an MS-grade sample vial. A control reaction without pNP-Ac was also performed. All samples and experiments were run as previously described (30). Data analyses were performed using MassHunter qualitative analysis, version B.06.00 (Agilent).

Other analytical procedures
Nucleotide sequencing of PCR products and plasmids was performed by the Genomics Facility of the Advanced Analysis Center (University of Guelph). Protein concentrations were determined using the Pierce BCA protein assay kit with BSA serving as the standard. SDS-PAGE on 15% acrylamide gels was conducted by the method of Laemmli (47) with Coomassie Brilliant Blue staining. Surface entropy reduction analyses were conducted using the SERp server (35).

Data availability
The atomic coordinates and structure factors of the reported crystal structures have been deposited in the Protein Data Bank under codes 6VJP and 6WN9. The authors declare that all other data supporting the findings of this study are available within the paper and its supporting information.