Methicillin resistance in Staphylococcus aureusrequires glycosylated wall teichoic acidsStephanie Browna, Guoqing Xiab,1, Lyly G. Luhachackc, Jennifer Campbella, Timothy C. Mereditha, Calvin Chena,Volker Winstelb, Cordula Gekelerb, Javier E. Irazoquic, Andreas Peschelb, and Suzanne Walkera,1

aDepartment of Microbiology and Immunobiology, Harvard Medical School, Boston, MA 02115; bCellular and Molecular Microbiology Section, InterfacultyInstitute of Microbiology and Infection Medicine, University of Tübingen, Elfriede-Aulhorn-Strasse 6, D-72076 Tübingen, Germany; and cProgram ofDevelopmental Immunology, Department of Pediatrics, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02115

Edited by Laura L. Kiessling, University of Wisconsin, Madison, WI, and approved September 6, 2012 (received for review May 31, 2012)

Staphylococcus aureus peptidoglycan (PG) is densely functionalizedwith anionic polymers called wall teichoic acids (WTAs). Thesepolymers contain three tailoring modifications: D-alanylation, α-O-GlcNAcylation, and β-O-GlcNAcylation. Here we describe the dis-covery and biochemical characterization of a unique glycosyltrans-ferase, TarS, that attaches β-O-GlcNAc (β-O-N-acetyl-D-glucosamine)residues to S. aureus WTAs. We report that methicillin resistantS. aureus (MRSA) is sensitized to β-lactams upon tarS deletion. Un-like strains completely lacking WTAs, which are also sensitive toβ-lactams, ΔtarS strains have no growth or cell division defects. Be-cause neither α-O-GlcNAc nor β-O-Glucose modifications can conferresistance, the resistance phenotype requires a highly specific chem-ical modification of the WTA backbone, β-O-GlcNAc residues. Thesedata suggest β-O-GlcNAcylated WTAs scaffold factors required forMRSA resistance. The β-O-GlcNAc transferase identifiedhere, TarS, isa unique target for antimicrobials that sensitize MRSA to β-lactams.

PBP2A | antibiotic resistance | beta lactam potentiation | murein |WTA glycosylation

Peptidoglycan (PG), a cross-linked polymeric matrix that sur-rounds bacterial cells, is essential for survival (1). It is a major

target for clinically used antibiotics, including the β-lactams (2).Methicillin resistant Staphylococcus aureus (MRSA) is responsiblefor a large percentage of β-lactam–resistant infections and there is anongoing need for new strategies to treat these infections (3). Re-sistance in MRSA involves the acquired gene,mecA, which encodesPBP2a, a transpeptidase that functions to cross-link peptidoglycanstrands when native transpeptidases are inactivated by β-lactams (4–6). Many other genes native to S. aureus must be present for mecA-mediated resistance to operate (7–10). Understanding how differ-ent factors participate in β-lactam resistance is of paramountimportance for developing new approaches to combat MRSA.Preventing wall teichoic acid (WTA) synthesis by blocking

TarO, which catalyzes the first step in theWTA pathway, sensitizesMRSA to β-lactams (11, 12). S. aureus WTAs are covalentlyattached to PG and consist of a poly(ribitol phosphate) [poly(RboP)] backbone containing three tailoring modifications: D-alanylation, α-O-GlcNAcylation, and β-O-GlcNAcylation (Fig. 1)(13, 14). WTAs have been implicated in cell shape, cell division,biofilm formation, phage infectivity, and pathogenesis, and evidencesuggests that the tailoring modifications contribute to these func-tions (13, 14). For example, D-alanylation promotes biofilm for-mation, increases resistance to cationic antibiotics, and plays arole in virulence, whereasO-GlcNAcylation is required for infectionby serogroups A, B, and F phages of S. aureus (13–18).We sought to determine whether the WTA polymer itself or

a backbone modification regulates various WTA functions inS. aureus. Our studies required knowing the genes that encodethe tailoring enzymes. The genes involved in D-alanylation andα-O-GlcNAcylation were previously identified, but the gene en-coding the transferase that attaches β-O-GlcNAc residues toWTAs was not (19–23). Here we report the identification of thisunique gene, tarS, and we profile the substrate preferences of the

encoded β-O-GlcNAc transferase.We show that theWTAbackboneis required for proper cell division, whereas β-O-GlcNAcylation isrequired to maintain β-lactam resistance in MRSA. Unlike ΔtarOMRSA strains, which are also β-lactam sensitive, ΔtarS strainshave nomorphological or cell division defects. Hence, the β-lactamsusceptibility is not coupled to the global defects in cell divisionthat occur in the absence of WTAs. Through genetic manipula-tions of WTA tailoring enzymes, we demonstrate that the properstereochemical linkage of the sugar to the WTA backbone andthe C2 N-acetyl substituent are required for the resistant phe-notype. Because the structural features required are specific, wespeculate that β-O-GlcNAcylatedWTAs interact with cell surfacefactors necessary for β-lactam resistance. We conclude that TarSis a unique possible target for inhibitors to be used in combinationwith β-lactams to overcome MRSA infections.

ResultsIdentification of a β-O-GlcNAc Transferase in S. aureus. We recentlyidentified a phage-resistant S. aureus strain, K6, which producesnonglycosylated WTAs (21). This strain carried a transposonin the tarM gene, and the encoded TarM protein was identifiedas a WTA α-O-GlcNAc transferase. We subsequently found thata targeted tarM deletion strain, RN4220ΔtarM, is still susceptibleto S. aureus serogroup B phages ϕ11, ϕ52A, and ϕ80 (Fig. 2A).Because phage infection depends on WTA glycoepitopes (18),these data suggested that WTAs are glycosylated in theRN4220ΔtarM strain. NMR analysis of WTAs from the ΔtarMstrain showed a WTA structure consistent with β-O-GlcNAcy-lated ribitol phosphate (Fig. 2 B and C) (24). Thus, S. aureus strainRN4220 contains a β-O-GlcNAc transferase in addition to thepreviously identified α-O-GlcNAc transferase.Two putative glycosyltransferase-encoding genes, designated

SAOUHSC_00644 and SAOUHSC_00228, were identified inWTA operons of S. aureus NCTC8325 (Fig. S1). We used an invitro reconstitution approach to test their function as WTA β-O-GlcNAc transferases. Each gene was overexpressed as an N-terminal 10-His fusion protein in Escherichia coli and purified byNi2+-affinity chromatography. The poly(RboP)-WTA substratewas synthesized in vitro using a combined chemical and enzymaticapproach (Fig. 3A), and the purified proteins were incubated withthis polymer substrate and UDP-[14C]-GlcNAc. Reaction mixtureswere separated by PAGE and analyzed using phosphorimaging.No reaction was detectable under any conditions in incubation

Author contributions: S.B., G.X., T.C.M., J.E.I., A.P., and S.W. designed research; S.B., G.X.,L.G.L., J.C., T.C.M., C.C., V.W., and C.G. performed research; S.B., G.X., L.G.L., J.C., C.C.,V.W., and C.G. analyzed data; and S.B., G.X., J.E.I., A.P., and S.W. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

See Commentary on page 18637.1To whom correspondence may be addressed. E-mail: suzanne_walker@hms.harvard.eduor guoqing.xia@med.uni-tuebingen.de.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1209126109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1209126109 PNAS | November 13, 2012 | vol. 109 | no. 46 | 18909–18914

MICRO

BIOLO

GY

SEECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Aug

ust 1

7, 2

021

mixtures containing protein encoded by SAOUHSC_00644, andthe function of this protein remains unknown. However, in reactionscontaining protein encoded by gene SAOUHSC_00228, theradiolabeled sugar starting material disappeared and radio-labeled products having similar mobility to [14C]-WTA polymersappeared (Fig. 3B). We conclude that SAOUHSC_00228 enc-odes a WTA glycosyltransferase, which we designated TarS.

TarS Prefers UDP-GlcNAc as a Donor Substrate and Is Specific for Poly(Ribitol Phosphate) WTAs.Most S. aureus strains attach O-GlcNActo their WTAs, whereas other bacterial strains attach glucose,GalNAc, or other carbohydrates (15). To examine the donorspecificity of TarS, we tested its ability to use four other UDPsugars as donors for glycosylation of poly(RboP)-WTA (Fig. 3B).The results showed that TarS can use UDP-Glc and UDP-GalNAcas alternative donor substrates, but not UDP-galactose or UDP-glucuronic acid. Thus, structural changes at either the C2 or C4positions of the sugar donor are tolerated, althoughUDP-GalNAc isa far less efficient substrate thanUDP-GlcNAcorUDP-Glc. InTarSreactions incubated with poly(RboP)-WTA and equimolar amountsof UDP-GlcNAc and UDP-Glc, the UDP-GlcNAc peak disap-peared, whereas the UDP-Glc peak area was largely unchanged,indicating that UDP-GlcNAc is the preferred donor (Fig. 3C).We tested the ability of TarS to transfer GlcNAc to possible

alternate acceptors. Using different combinations of previouslycharacterized WTA biosynthetic enzymes (25–27) we prepared aWTA pathway intermediate for acceptor testing. CDP ribitol, ribitolphosphate, lipoteichoic acid (LTA), and poly(GroP)-WTAwere alsotested as acceptor substrates. No glycosylation of any of thesesubstrates was observed (Fig. 3D and Fig. S2), showing that TarS isspecific for substrates containing multiple RboP units. These datasupport the prediction that WTA glycosylation in S. aureus, as inBacillus subtilis (28), occurs after polymer synthesis is complete.

TarS β-O-GlcNAcylates WTAs in S. aureus Cells. To investigate thefunction of TarS in cells, we deleted tarS fromwild-type RN4220 andthe RN4220ΔtarM strain and checked susceptibility of the mutantsto phagesϕ11, ϕ52A, andϕ80. TheΔtarS strain was as susceptible tothe test phage as the parental andΔtarM strains, but theΔtarMΔtarSdouble mutant was phage resistant (Fig. 2A). To establish theanomeric stereochemistry of the TarS product, we extracted WTAs(25) from the deletion strains and analyzed NMR spectra of theirmonomer units. β-O-GlcNAcylated ribitol phosphates have ananomeric β-O-GlcNAc 1H resonance at 4.73 ppm and a 13C reso-nance at 101.6 ppm (24). This peak is absent inWTAs extracted fromthe ΔtarMΔtarS strain, but is present inWTAs from theΔtarM strainand the ΔtarMΔtarS strain expressing TarS from a plasmid (Fig. 2 BandC andFig. S3). Thus,TarS β-O-GlcNAcylatesWTAs inS. aureus.

The WTA Polymer Backbone, but Not Its Tailoring Modifications, IsEssential for Properly Regulated Cell Division. WTA null (ΔtarO)strains have several phenotypes. They are temperature sensitive,

reach stationary phase at a lower density than wild-type strains,undergo Triton-X–induced autolysis at an increased rate, andshow increased susceptibility to lysostaphin (13, 14). Perhapsmore striking, the absence of WTAs results in massively dysre-gulated cell division and sensitizes MRSA to β-lactams (11). Tocorrelate phenotypes observed in S. aureusWTA null strains withparticular WTA structural features, we deleted dltA, tarM, andtarS singly and in combination from S. aureus MW2, a well-char-acterized community-acquired MRSA strain (29) that producessimilar amounts of α- and β-O-GlcNAcylated WTAs. DeletingdltA prevents attachment of D-alanyl esters to both lipo- and wallteichoic acids (15). Deleting tarM results in the expression of onlyβ-O-GlcNAcylated WTAs, deleting tarS leads to the productionof only α-O-GlcNAcylated WTAs, and deleting both genes resultsin unglycosylated WTAs (Fig. S4A).Several phenotypes of the deletion strains were examined and

it was found that deletion of the O-GlcNAc transferase genesalone or in combination had a negligible effect on cell growth rates,in vitro fitness, autolysin activation, lysostaphin susceptibility, andbiofilm formation (Fig. S5). In contrast, preventing D-alanylationled to several phenotypes that are also observed in the ΔtarOstrain, either herein or in previous publications, including a de-creased growth rate, increased Triton-X–induced autolysis, anddecreased biofilm formation, consistent with previous studies onother ΔdltA strains (17, 30).To determine whether any of the WTA substituents play a role

in regulating cell division, we compared transmission electronmicroscopy (TEM) images of the mutant strains to those of thewild-type and ΔtarO strains. Whereas the MW2ΔtarO strainshowed defects in septal placement and daughter cell separation,none of the other mutant strains, including the triple mutantlacking all WTA decorations, displayed any septal abnormalities(Fig. 4A and Fig. S6). Hence, the cell division defects observed inΔtarO strains are not due to the absence of a tailoring modifi-cation, but to the absence of the anionic poly(RboP) backbone.

WTA β-O-GlcNAc Modifications Are Required to Maintain β-LactamResistance in MRSA. Because preventing WTA expression sensi-tizes MRSA strains to β-lactams by an unknown mechanism (11),we measured β-lactam minimum inhibitory concentrations (MICs)for the MW2 deletion strains to determine whether any tailoringmodifications were required for resistance. Other antibiotics wereincluded as controls (Fig. 4B and Fig. S7A). As observed pre-viously, the ΔdltA strain was more sensitive to cationic antibioticsthan wild-type S. aureus, perhaps because the increased negativesurface charge density attracts cationic molecules. This strainalso showed reduced MICs for methicillin and imipenem, butunlike the ΔtarO strain, it was not more susceptible to several otherβ-lactams. The ΔtarM strain showed no change in sensitivity to anyantibiotics. In striking contrast however, the ΔtarS strain showedincreased susceptibility to the same set of β-lactams as the ΔtarOstrain, but not to any other antibiotics, including other cell-

Fig. 1. Wall teichoic acids in S. aureus contain three different types of tailoring modifications. S. aureus WTAs are composed of ribitol-phosphate repeats(two repeats are shown in red) tailored with D-alanines, α-O-GlcNAcs, and β-O-GlcNAcs. Enzymes responsible for α-O-GlcNAcylation and D-alanylation werepreviously identified. The enzyme (blue arrow) that attaches β-O-GlcNAcs to poly(RboP)-WTA was unidentified before this work.

18910 | www.pnas.org/cgi/doi/10.1073/pnas.1209126109 Brown et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 1

7, 2

021

wall–active antibiotics. Complementation of MW2ΔtarMΔtarS witha plasmid expressing catalytically active TarS restored wild-typeresistance levels, whereas complementation with a plasmid ex-pressing TarM or a catalytically inactive variant of TarS did not(Figs. S4B and S7 B and C). These results established that thesensitized phenotype of the tarS deletion strains is not due topolar effects, but to the lack of a functional WTA β-O-GlcNActransferase. Consistent with the importance of tarS in mediatingresistance, its expression was fourfold up-regulated in the pres-ence of β-lactams, whereas the expression of tarM was unchanged(Fig. S8A). QRT-PCR analysis showed that expression of the geneencoding the β-lactam resistant transpeptidase PBP2a, mecA,was unaffected by tarS deletion, indicating that sensitivity is notdue to down-regulation of the resistance determinant. Expres-sion levels of several other genes implicated in β-lactam resistance,including pbp4 and fmtA, were also unaffected (Fig. S8B). Takentogether, these results are consistent with a direct role for β-O-GlcNAcylated WTAs in maintaining β-lactam resistance.To obtain further information on the structural requirements

of the WTA sugar residue in resistance, we sought to replaceβ-O-GlcNAc with another β-linked sugar. B. subtilis W23 makespoly(RboP)-WTAs containing β-O-glucose WTA modifications(31), but the tailoring enzyme had not been identified. There arefour predicted glycosyltransferases, tarEMNQ, in the W23 WTAgene cluster. Since TarQ has the highest homology to TarS, weconstructed two B. subtilis W23 strains, one containing a markeddeletion of tarQ (W23ΔtarQ) and the other containing an addi-tional deletion of genes tarEMN (W23ΔQΔtarEMN). Quadrupole(Q)TOF-MS analysis of extracted WTAs showed that glucosyla-tion was absent in both mutant strains (Fig. S9A). Because de-letion of tarQ was sufficient to abolish glucosylation, we concludedthat TarQ is the primary WTA β-O-glucosyl transferase in B.subtilis W23. The tarQ gene was therefore expressed from aplasmid in MW2ΔtarMΔtarS, and this tarQ expression strain wasshown to produce WTAs modified with glucose (Fig. S9B). How-ever, unlike strains expressing tarS from a plasmid, this strain wasnot resistant to β-lactams (Fig. S7C). The resistance phenotypethus depends not only on the β-anomeric configuration of the WTAsugar modification, but also on the presence of a C2 N-acetylmoiety (Fig. 4C). Hence, the structural features required forresistance are specific, suggesting that β-O-GlcNAcylated WTAsestablish selective interactions with other cell surface factors re-quired for resistance.To determine whether β-O-GlcNAcylated WTAs play a role in

β-lactam resistance in other MRSA strains, we deleted tarS fromfour additional community- and hospital-acquired MRSA strainsand measured their β-lactam MICs. These tarS deletion strainswere also sensitized to β-lactams (Fig. S7D), implying that β-O-GlcNAcylated WTAs are a general resistance factor in MRSAstrains. Accordingly, TarS is a possible target for small moleculesthat sensitize MRSA to β-lactams.

Fig. 2. TarS is a β-O-GlcNAc-WTA transferase in vivo. (A) Soft agar assay forphage susceptibility of RN4220 wild type, ΔtarM, ΔtarS, ΔtarMΔtarS, andthe double mutant complemented with tarS (ΔtarMΔtarSPtarS). The doublemutant, ΔtarMΔtarS, leads to resistance to serogroup B phages ϕ52A, ϕ11,and ϕ80. Mutation 201A, leading to a truncated tarS gene, was identified inthe previously reported strain K6. This explains previous findings (21) thatthis strain, which contains a transposon insertion in tarM, is completely phage

resistant. (B) 1H NMR spectra of chemically extracted and degraded WTAmonomers from the RN4220 ΔtarM, ΔtarMΔtarS, and ΔtarMΔtarS PcadtarSstrains (D2O; Varian; 400 mHz). Arrow points to the H-1 proton of the β-O-GlcNAc residue on ribitol phosphate. Peak corresponding to the amide ofthe C2 N-acetyl group attached to the sugar is identified as NHAc. Both thesingle knockout and the tarS-complemented strain produce β-O-GlcNAcy-lated WTAs, but the double knockout does not. (Fig. S3) (C) Heteronuclear13C,1H single quantum correlation (HSQC) of the WTA monomer from theRN4220 ΔtarM strain. The H-1 proton of the β-O-GlcNAc residue is indicatedby the arrow and confirms the signal at δ4.73 is indicative of a β-O-GlcNAc-ribitol-phosphate WTA monomer. These NMR data correlate with previousNMR analyses of β-O-GlcNAc-ribitol-phosphate WTA monomers (21, 24).

Brown et al. PNAS | November 13, 2012 | vol. 109 | no. 46 | 18911

MICRO

BIOLO

GY

SEECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Aug

ust 1

7, 2

021

TarS Is Required for Oxacillin Resistance in a Caenorhabditis elegansModel.The bacterivorous nematode C. elegans dies upon ingestionof pathogenic S. aureus, and is commonly used as a simple modelhost to evaluate the virulence of different bacterial strains in thepresence and absence of antibiotics (32, 33). Using this host, weevaluated the virulence of MW2ΔtarS compared with wild type,and determined bacterial susceptibility to treatment with oxacillinin vivo by evaluating nematode survival over time (Fig. S10). Inthe absence of antibiotic, the MW2ΔtarS strain was slightly at-tenuated compared with wild type, suggesting a modest role forthe β-O-GlcNAc WTA moieties in pathogenesis in this system.At the highest tested dose, oxacillin partially rescued nematodesinfected with wild-type MW2. In contrast, at the lowest dose tested,the antibiotic completely rescued animals infected with MW2ΔtarS.We conclude that β-O-GlcNAcylation of WTAs plays an impor-tant role in β-lactam resistance in vivo as it does in vitro.

DiscussionWTAs play an important role in regulating cell division in S. aureusand are also involved in maintaining resistance to β-lactams inMRSA strains (11). In this paper, we used biochemical and ge-netic approaches to examine the roles of the WTA polymer

backbone and three tailoring modifications in these importantphenotypes. We showed that the polymer backbone alone isrequired for properly regulated cell division because deletion ofall three tailoring genes did not affect septal placement or cellseparation. We also showed that a single WTA tailoring modi-fication, β-O-GlcNAcylation, mediated by a newly discoveredglycosyltransferase named TarS, is required to maintain wild-typelevels of β-lactam resistance.The mechanism by which β-O-GlcNAcylated WTAs confer

resistance to β-lactams in MRSA is not yet known. TarS deletiondoes not cause apparent growth or cell division defects and itsdeletion does not down-regulate expression of mecA or othertested genes implicated in β-lactam resistance. MRSA β-lactamresistance cannot be suppressed by expressing glycosyltransferasesthat incorporate either α-O-GlcNAcs or β-O-glucose residues intoWTAs. Because the susceptibility phenotype seems directly at-tributable to the absence of the β-O-GlcNAc modifications onWTAs, we propose a scaffolding role for β-O-GlcNAcylatedWTAsin methicillin resistance. Consistent with this hypothesis, it has beenpreviously speculated that WTAs can scaffold PG synthesis anddegradation proteins (11, 34–36). Moreover, a recent paper hasshown that both PBP2a and FmtA directly bind WTAs in vitro

Fig. 3. Characterization of the O-GlcNAc transferase, TarS, that attaches β-O-GlcNAc to poly(RboP)-WTAs. (A) Scheme for in vitro chemoenzymatic synthesisof poly(RboP)-WTA and its use to assay candidate β-O-GlcNAc WTA transferases for their ability to transfer a radiolabeled sugar to the polymer. (B) Auto-radiographs of polyacrylamide gels of heat-treated (−) or active (+) TarS enzyme incubated with poly(RboP)-WTA and various [14C]-UDP-donor substrates.UDP-glucose and UDP-GalNAc are used as donor substrates, but TarS has a preference for UDP-GlcNAc (C). (C) Overlay of HPLC chromatograms (260 nmdetection) of TarS incubated with equimolar amounts of UDP-glucose and UDP-GlcNAc at 0 min and 120 min. TarS reacts preferentially with UDP-GlcNAc.Injections of authentic standards were used to determine the identity of the observed peaks. (D) Structures of acceptors tested are tabulated and results aresummarized (Fig. S2). The only tolerated acceptor substrate is poly(RboP)-WTA.

18912 | www.pnas.org/cgi/doi/10.1073/pnas.1209126109 Brown et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 1

7, 2

021

(37). Polyvalent scaffolding by glycopolymers (glycolipids andglycoproteins) is a known mechanism for organizing proteins in avariety of biological processes (38, 39).Whereas there are no genesof high homology to tarS in Staphylococcus sciuri, the organismfrom which PBP2a was speculated to derive (40, 41), all depositedS. aureus strain sequences in theNational Center for BiotechnologyInformation contain tarS. It is possible that PBP2amay exploit β-O-GlcNAcylated WTAs by binding to them directly or by interactingwith other cellular factors that are scaffolded by these polymers.There are many genes native to S. aureus that have been shown toplay a role in maintaining β-lactam resistance (7–10), includinggenes encoding proteins with extracellular domains that couldinteract with β-O-GlcNAc WTA tailoring modifications.In closing, we note that there is considerable interest in com-

pounds that restore β-lactam sensitivity to resistant microorganisms(42). Because β-lactam resistance in MRSA involves a resistanttranspeptidase that depends on the function of other cellular factorsfor activity (4–6), compound combinations that target the activity ofthese other factors could be useful for overcomingMRSA infections.β-lactamase inhibitors have been highly successful as components ofcompound combinations to treat many β-lactam–resistant infections(2, 3). Because deletion of the tarS gene sensitizes MRSA to

β-lactams, we propose that TarS is a unique target for compoundsused in combination with β-lactams to treat MRSA infections.

Materials and MethodsThe bacterial strains, plasmids, and primers for PCR amplification are listed inTable S1. More detailed methods can be found in SI Materials and Methods.

Cloning, Overexpression, and Purification of SAOUHSC_00228 and SAOUHSC_00644.Genes were PCR amplified from RN4220 DNA ligated into a pET-47b(+) vector(Novagen), which had been modified to a cleavable N-terminal His-tag with10-His residues transformed into Rosetta2(DE3)pLysS cells for overexpressionby isopropyl-β-D-1-thiogalactopyranoside (IPTG) at 16 °C. The cell pellet wasresuspended in 50 mM Tris·HCl pH 8, 200 mM NaCl, 0.1% Tween-20, and25% (vol/vol) glycerol, benzonase, and rLysozyme (Novagen) was added andthe cells sonicated. The clarified lysate was purified using nickel resin. Purifiedprotein was stored as 50% (vol/vol) glycerol stocks at either −20 °C or −80 °C.The yield is ∼5 mg/L for SAOUHSC_00228 and 33 mg/L for SAOUHSC_00644.

SAOUHSC_00228 and SAOUHSC_00644 Substrate Testing Reactions. Three-mi-croliter reactions containing 15 μM 228 or 130 μM 644, 20 mM Tris pH 8,5 mM MgCl2, 63 μM UDP sugar, 7 μM UDP-[14C]-sugar, and 7 μM purified neryl-polyRboP-polymer or 5 μM UDP-[14C]-GlcNAc and 15 μM ribitol-5-phosphate,CDP-ribitol, or farnesyl-PP-GlcNAc-ManNAc-GroP-RboP were constructed.Three-microliter reactions contained 7 μM neryl-PP-GlcNAc-ManNac-GroP,

Fig. 4. The anionic WTA polymer backbone, but not any tailoring modifications, is required for proper cell division. β-O-GlcNAc WTA modifications arerequired for β-lactam resistance in MRSA. (A) TEM images of MW2 wild type, a WTA null strain (ΔtarO), and two strains containing deletions in genes re-sponsible for WTA tailoring modifications show that deleting WTAs leads to dysregulated cell division. No obvious morphological abnormalities are observedin cells lacking WTA decorations. (Scale bars, 500 nm.) (Fig. S6) (B) MIC values (μg/mL) of tested antibiotics against S. aureus strains. Data representative of atleast six experiments. Highlighted values differ from wild type by at least fourfold. β, β-lactams; C, cationic antibiotics. ΔtarS, like ΔtarO, is susceptible to alltested β-lactams. Consistent with previous results, the ΔdltA strain is more susceptible to cationic antibiotics than wild type. (Fig. S7 A and D) (C) Graphic tableof results showing only MRSA strains that contain β-O-GlcNAc WTA modifications are able to maintain β-lactam resistance. MRSA strains containing β-O-glucose or α-O-GlcNAc WTA substituents are sensitive to β-lactams. MICs are tabulated in Fig. S7C.

Brown et al. PNAS | November 13, 2012 | vol. 109 | no. 46 | 18913

MICRO

BIOLO

GY

SEECO

MMEN

TARY

Dow

nloa

ded

by g

uest

on

Aug

ust 1

7, 2

021

700 μM CDP-glycerol, 20 mM Tris pH 8, 5 mM MgCl2, 500 nM TagF, 10 μM228 or 130 μM 644, 7 μM [14C]-UDP-GlcNAc, and 63μM UDP sugar. As acontrol to ensure TagF activity, the reaction was also set up in the samemanner except with the addition of 21 μM CDP-[14C]-glycerol and cold UDP-GlcNAc. Reactions testing LTA (Sigma) were set up in a similar manner exceptTagF and CDP-glycerol were omitted. As a negative control, identical reac-tions were set up using heat-treated enzyme. After 2 h at room temperaturethe reactions were quenched with an equal volume of dimethylformamideand analyzed by gel electrophoresis and phosphorimaging.

Construction of Deletion Strains. B. subtilis W23 was transformed with linearDNA harboring a resistance cassette flanked by 1,000 bp of DNA upstreamand downstream of the targeted genes (25). The RN4220 tarS gene wasdeleted using the pKOR1 vector, whereas, all other S. aureus gene deletions

were made using the pMAD plasmid. The ΔdltA strain is identical to thatwhich was published previously (16). For MW2 deletions, the plasmid wasfirst passaged through RN4220. The tarS/tarM deletion strains were backcomplemented with TarS by cloning the gene into the pLI50-PcadC shuttleexpression vector under the control of a cadmium-inducible promoter. TheD92A, D94A inactive TarS point mutant was made using the QuikChangeLightning (Agilent) kit according to the manufacturer’s instructions.

ACKNOWLEDGMENTS. We thank Charles Sheahan, Patricia Sanchez-Carballo,and Otto Holst for NMR assistance and Christiane Goerke and Petra Kühnerfor RT-PCR support. This work was supported by National Institutes of HealthGrants 1R01AI099144 and P01AI083214 (to S.W.) and T32AI007061 (to S.B.and T.C.M.), as well as by German Research Council Grants TRR34 (to A.P.)and SFB766 (to G.X.).

1. Vollmer W, Blanot D, de Pedro MA (2008) Peptidoglycan structure and architecture.FEMS Microbiol Rev 32:149–167.

2. Zapun A, Contreras-Martel C, Vernet T (2008) Penicillin-binding proteins and beta-lactam resistance. FEMS Microbiol Rev 32:361–385.

3. Boucher HW, et al. (2009) Bad bugs, no drugs: No ESKAPE! An update from the In-fectious Diseases Society of America. Clin Infect Dis 48:1–12.

4. Hartman BJ, Tomasz A (1984) Low-affinity penicillin-binding protein associated withbeta-lactam resistance in Staphylococcus aureus. J Bacteriol 158:513–516.

5. Matsuhashi M, et al. (1986) Molecular cloning of the gene of a penicillin-bindingprotein supposed to cause high resistance to beta-lactam antibiotics in Staphylococcusaureus. J Bacteriol 167:975–980.

6. Pinho MG, de Lencastre H, Tomasz A (2001) An acquired and a native penicillin-binding protein cooperate in building the cell wall of drug-resistant staphylococci.Proc Natl Acad Sci USA 98:10886–10891.

7. Berger-Bächi B, Rohrer S (2002) Factors influencing methicillin resistance in staphy-lococci. Arch Microbiol 178:165–171.

8. Lee SH, et al. (2011) Antagonism of chemical genetic interaction networks resensitizeMRSA to β-lactam antibiotics. Chem Biol 18:1379–1389.

9. Huber J, et al. (2009) Chemical genetic identification of peptidoglycan inhibitorspotentiating carbapenem activity against methicillin-resistant Staphylococcus aureus.Chem Biol 16:837–848.

10. De Lencastre H, et al. (1999) Antibiotic resistance as a stress response: Complete se-quencing of a large number of chromosomal loci in Staphylococcus aureus strain COLthat impact on the expression of resistance to methicillin. Microb Drug Resist 5:163–175.

11. Campbell J, et al. (2011) Synthetic lethal compound combinations reveal a funda-mental connection between wall teichoic acid and peptidoglycan biosyntheses inStaphylococcus aureus. ACS Chem Biol 6:106–116.

12. Maki H, Yamaguchi T, Murakami K (1994) Cloning and characterization of a geneaffecting the methicillin resistance level and the autolysis rate in Staphylococcusaureus. J Bacteriol 176:4993–5000.

13. Swoboda JG, Campbell J, Meredith TC, Walker S (2010) Wall teichoic acid function,biosynthesis, and inhibition. ChemBioChem 11:35–45.

14. Xia G, Kohler T, Peschel A (2010) The wall teichoic acid and lipoteichoic acid polymersof Staphylococcus aureus. Int J Med Microbiol 300:148–154.

15. Neuhaus FC, Baddiley J (2003) A continuum of anionic charge: Structures and func-tions of D-alanyl-teichoic acids in gram-positive bacteria. Microbiol Mol Biol Rev 67:686–723.

16. Peschel A, et al. (1999) Inactivation of the dlt operon in Staphylococcus aureus conferssensitivity to defensins, protegrins, and other antimicrobial peptides. J Biol Chem 274:8405–8410.

17. Peschel A, Vuong C, Otto M, Götz F (2000) The D-alanine residues of Staphylococcusaureus teichoic acids alter the susceptibility to vancomycin and the activity of auto-lytic enzymes. Antimicrob Agents Chemother 44:2845–2847.

18. Xia G, et al. (2011) Wall teichoic Acid-dependent adsorption of staphylococcal si-phovirus and myovirus. J Bacteriol 193:4006–4009.

19. May JJ, et al. (2005) Inhibition of the D-alanine:D-alanyl carrier protein ligase fromBacillus subtilis increases the bacterium’s susceptibility to antibiotics that target thecell wall. FEBS J 272:2993–3003.

20. Perego M, et al. (1995) Incorporation of D-alanine into lipoteichoic acid and wallteichoic acid in Bacillus subtilis. Identification of genes and regulation. J Biol Chem270:15598–15606.

21. Xia G, et al. (2010) Glycosylation of wall teichoic acid in Staphylococcus aureus byTarM. J Biol Chem 285:13405–13415.

22. Nathenson SG, Ishimoto N, Anderson JS, Strominger JL (1966) Enzymatic synthesis andimmunochemistry of alpha- and beta-N-acetylglucosaminylribitol linkages in teichoicacids from several strains of Staphylococcus aureus. J Biol Chem 241:651–658.

23. Nathenson SG, Strominger JL (1962) Enzymatic synthesis and immunochemistry of N-acetylglucosaminylribitol linkages in the teichoic acids of Staphylococcus aureusstains. J Biol Chem 237:3839–3841.

24. Vinogradov E, Sadovskaya I, Li J, Jabbouri S (2006) Structural elucidation of the ex-tracellular and cell-wall teichoic acids of Staphylococcus aureus MN8m, a biofilmforming strain. Carbohydr Res 341:738–743.

25. Brown S, Meredith T, Swoboda J, Walker S (2010) Staphylococcus aureus and Bacillussubtilis W23 make polyribitol wall teichoic acids using different enzymatic pathways.Chem Biol 17:1101–1110.

26. Brown S, Zhang Y-H, Walker S (2008) A revised pathway proposed for Staphylococcusaureus wall teichoic acid biosynthesis based on in vitro reconstitution of the in-tracellular steps. Chem Biol 15:12–21.

27. Pereira MP, et al. (2008) The wall teichoic acid polymerase TagF efficiently synthesizespoly(glycerol phosphate) on the TagB product lipid III. ChemBioChem 9:1385–1390.

28. Allison SE, D’Elia MA, Arar S, Monteiro MA, Brown ED (2011) Studies of the genetics,function and kinetic mechanism of TagE: the wall teichoic acid glycosyltransferase inbacillus subtilis 168. J Biol Chem 286:23708–23716.

29. Baba T, et al. (2002) Genome and virulence determinants of high virulence commu-nity-acquired MRSA. Lancet 359:1819–1827.

30. Gross M, Cramton SE, Götz F, Peschel A (2001) Key role of teichoic acid net charge inStaphylococcus aureus colonization of artificial surfaces. Infect Immun 69:3423–3426.

31. Chin T, Burger MM, Glaser L (1966) Synthesis of teichoic acids. VI. The formation ofmultiple wall polymers in Bacillus subtilis W-23. Arch Biochem Biophys 116:358–367.

32. Irazoqui JE, et al. (2010) Distinct pathogenesis and host responses during infection ofC. elegans by P. aeruginosa and S. aureus. PLoS Pathog 6:e1000982.

33. Sifri CD, Begun J, Ausubel FM, Calderwood SB (2003) Caenorhabditis elegans asa model host for Staphylococcus aureus pathogenesis. Infect Immun 71:2208–2217.

34. Atilano ML, et al. (2010) Teichoic acids are temporal and spatial regulators of pep-tidoglycan cross-linking in Staphylococcus aureus. Proc Natl Acad Sci USA 107:18991–18996.

35. Schlag M, et al. (2010) Role of staphylococcal wall teichoic acid in targeting the majorautolysin Atl. Mol Microbiol 75:864–873.

36. Frankel MB, Schneewind O (2012) Determinants of murein hydrolase targeting tocross-wall of Staphylococcus aureus peptidoglycan. J Biol Chem 287:10460–10471.

37. Qamar A, Golemi-Kotra D (2012) Dual roles of FmtA in Staphylococcus aureus cell wallbiosynthesis and autolysis. Antimicrob Agents Chemother 56:3797–3805.

38. Bertozzi CR, Kiessling LL (2001) Chemical glycobiology. Science 291:2357–2364.39. Avci FY, Kasper DL (2010) How bacterial carbohydrates influence the adaptive im-

mune system. Annu Rev Immunol 28:107–130.40. Couto I, et al. (1996) Ubiquitous presence of a mecA homologue in natural isolates of

Staphylococcus sciuri. Microb Drug Resist 2:377–391.41. Wu SW, de Lencastre H, Tomasz A (2001) Recruitment of the mecA gene homologue

of Staphylococcus sciuri into a resistance determinant and expression of the resistantphenotype in Staphylococcus aureus. J Bacteriol 183:2417–2424.

42. Kalan L, Wright GD (2011) Antibiotic adjuvants: Multicomponent anti-infectivestrategies. Expert Rev Mol Med 13:e5.

18914 | www.pnas.org/cgi/doi/10.1073/pnas.1209126109 Brown et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 1

7, 2

021