Salt-Induced Stress Stimulates a Lipoteichoic Acid-Specific Three-Component Glycosylation System in Staphylococcus aureus

ABSTRACT

Lipoteichoic acid (LTA) in Staphylococcus aureus is a poly-glycerophosphate polymer anchored to the outer surface of the cell membrane. LTA has numerous roles in cell envelope physiology, including regulating cell autolysis, coordinating cell division, and adapting to environmental growth conditions. LTA is often further modified with substituents, including d-alanine and glycosyl groups, to alter cellular function. While the genetic determinants of d-alanylation have been largely defined, the route of LTA glycosylation and its role in cell envelope physiology have remained unknown, in part due to the low levels of basal LTA glycosylation in S. aureus. We demonstrate here that S. aureus utilizes a membrane-associated three-component glycosylation system composed of an undecaprenol (Und) N-acetylglucosamine (GlcNAc) charging enzyme (CsbB; SAOUHSC_00713), a putative flippase to transport loaded substrate to the outside surface of the cell (GtcA; SAOUHSC_02722), and finally an LTA-specific glycosyltransferase that adds α-GlcNAc moieties to LTA (YfhO; SAOUHSC_01213). We demonstrate that this system is specific for LTA with no cross recognition of the structurally similar polyribitol phosphate containing wall teichoic acids. We show that while wild-type S. aureus LTA has only a trace of GlcNAcylated LTA under normal growth conditions, amounts are raised upon either overexpressing CsbB, reducing endogenous d-alanylation activity, expressing the cell envelope stress responsive alternative sigma factor SigB, or by exposure to environmental stress-inducing culture conditions, including growth media containing high levels of sodium chloride.

IMPORTANCE The role of glycosylation in the structure and function of Staphylococcus aureus lipoteichoic acid (LTA) is largely unknown. By defining key components of the LTA three-component glycosylation pathway and uncovering stress-induced regulation by the alternative sigma factor SigB, the role of N-acetylglucosamine tailoring during adaptation to environmental stresses can now be elucidated. As the dlt and glycosylation pathways compete for the same sites on LTA and induction of glycosylation results in decreased d-alanylation, the interplay between the two modification systems holds implications for resistance to antibiotics and antimicrobial peptides.

INTRODUCTION

Staphylococcus aureus is an opportunistic pathogen that commonly inhabits the pharynx and skin (1, 2). In the past two decades, this bacterium has made headlines due to the clinical rise of methicillin-resistant S. aureus (MRSA). Despite its namesake, MRSA is capable of resisting various beta-lactam antibiotics and not just methicillin (3). Beta-lactams and other commonly used antibiotics, such as vancomycin, target the biosynthetic machineries that assemble the Gram-positive bacteria cell wall. The biosynthesis of these cell wall components is essential for MRSA's fitness and survival and thus presents an attractive target for antibiotic development (4–7).

The cell wall of S. aureus is comprised of a thick peptidoglycan matrix embedded with various surface-associated proteins and unique phosphodiester linked alditol polymers called teichoic acids (TA) (8). There are two types of TAs, namely, wall teichoic acid (WTA) and lipoteichoic acid (LTA). In S. aureus, WTA is comprised of polyribitol phosphate (poly-RboP) repeats anchored to the peptidoglycan via an N-acetylglucosamine (GlcNAc)-N-acetylmannosamine linkage unit (9), while LTA is comprised of poly-glycerophosphate (poly-GroP) repeats anchored directly to the cell membrane via a diacylglycerol-gentiobiose moiety (10–12). TAs are essential for maintaining the integrity of the cell envelope where they play numerous roles, including osmoprotection, regulating peptidoglycan autolysis, and coordinating cell division (9, 10, 13). In fact, deletion of both WTA and LTA is synthetically lethal (14, 15). The biosynthetic pathways for both forms of TAs have been established in S. aureus (9, 10). However, the biosynthesis of more complex forms of TAs in other species, as well as modifications in model TAs such as those in S. aureus, is still being elucidated. Modification of both WTA and LTA parent polymers can have vast impact on function. For instance, d-alanine (d-Ala) ester addition by the dltABCD operon has been shown to regulate autolysis during cell wall remodeling (16), localize penicillin-binding proteins (17), and confer resistance to antibiotics such as daptomycin (18) and antimicrobial peptides (19–21). The d-alanylation pathway is currently the most well defined (20, 22, 23), whereby DltA loads the carrier protein DltC with d-Ala. The functions of DltB and DltD are not as clear but are thought to transport the DltC–d-Ala intermediate across the membrane and subsequently d-alanylate TA acceptor substrates. Glycosylation of TAs has remained relatively underexamined by comparison, but recent studies have made progress in demonstrating the roles of TarM and TarS in transferring α- and β-GlcNAc, respectively, to WTA in S. aureus (24–27). WTA modification with β-GlcNAc imparts beta-lactam resistance, is critical to colonization, and plays a role alongside d-Ala modifications in cell division and morphogenesis (28, 29). Both TarS and TarM utilize cytoplasmic UDP-GlcNAc donors with glycosylation occurring before the WTA polymer is transported to the outer surface of the membrane. The TarM/TarS family of WTA glycosylases is widely distributed among TA-harboring Gram-positive bacteria and has been characterized among other Firmicutes, including in Bacillus subtilis (30).

Interestingly, a new class of integral membrane proteins (YfhO) was recently implicated in ribitol phosphate WTA glycosylation with GlcNAc moieties in Listeria monocytogenes (31). In this proposed glycosylation pathway, YfhO adds GlcNAc residues to WTA outside the cytoplasmic membrane using extracytoplasmic undecaprenol monophosphate GlcNAc (Und-P-GlcNAc) donor produced by the CsbB protein family. An analogous Listeria sp. serovar system for WTA galactosylation has also now been described (32). In the model S. aureus strain NCTC8325, there are both CsbB (SAOUHSC_00713) and YfhO (SAOUHSC_01213) orthologues, suggesting the genetic potential to glycosylate WTA as in L. monocytogenes. Instead, however, we present evidence here that the S. aureus YfhO orthologue glycosylates LTA with GlcNAc and is dependent on CsbB for donor substrate. We show that wild-type S. aureus LTA has low GlcNAc levels under normal growth conditions but that these levels are markedly elevated by either overexpressing CsbB, reducing d-alanylation, or inducing cell envelope stress using growth media containing high levels of NaCl.

RESULTS

CsbB and YfhO do not glycosylate WTA.

To examine whether CsbB-YfhO is indeed a WTA-specific glycosylation system, we first overexpressed the putative components of the membrane-bound extracellular glycosylation system in S. aureus. Typically, these systems consist of three proteins: the membrane associated Und-P-sugar glycosylase (i.e., CsbB), the integral membrane glycosyl transferase (i.e., YfhO), and an integral membrane protein involved in flipping the sugar loaded Und carrier from the cytoplasmic to the extracellular face of the membrane (40, 41). Since no flippase genes were located in proximity to either CsbB (SAOUHSC_00713) or YhfO (SAOUHSC_01213), a protein similarity search using the annotated Und-P-sugar transporter from the Shigella flexneri O-antigen glycosylation system encoded by the bacteriophage Sfx (41) was performed. A lone transporter candidate in the S. aureus genome with the same conserved domain was identified (GtcA; SAOUHSC_02722). The S. aureus GtcA protein shares 24% identity over 83% of its length (128 amino acids total) with the putative S. flexneri transporter and is predicted to have the four-transmembrane pass topology common to this transporter class. Hence, both CsbB and the putative transporter GtcA were overexpressed on a single multicopy plasmid to ensure transport would not be limiting. We transformed this plasmid into strains that were null for either of the known cytoplasmic WTA glycosylases tarM, tarS, or both, as we speculated modification could be masked by the resident GlcNAc glycosylases. Crude WTA was extracted from these strains and analyzed by polyacrylamide gel electrophoresis (PAGE) (see Fig. S1A in the supplemental material). Because TarM and TarS have redundant GlcNAcylating activities differing only in anomeric configurations, the single deletion profiles were not discernible from the wild type. In the ΔtarM ΔtarS double-knockout strain from which WTA glycosylation was eliminated, nonglycosylated WTA migrated at lower molecular weights. Contrary to our expectation, simultaneous overexpression of CsbB and GtcA (P_cadcsbB-gtcA) did not reestablish higher-molecular-weight WTA in the ΔtarM ΔtarS double knockout and deletion of ΔyfhO had no effect on WTA regardless of the presence of P_cadcsbB-gtcA. To confirm this, WTA was purified from overexpression strains for nuclear magnetic resonance (NMR) analysis (see Fig. S1B in the supplemental material). The structure of WTA isolated from the different constructs was indifferent to the expression levels of CsbB-GtcA-YfhO.

CsbB and YfhO induces LTA structural heterogeneity.

Since LTA is structurally similar to WTA, we next examined whether instead LTA is the actual CsbB-GtcA-YfhO target substrate. Gründling and coworkers previously identified a galactose LTA modification system in L. monocytogenes that uses an extracytoplasmic three-component glycosylation system (42). Furthermore, GlcNAc-modified LTA has been isolated from certain S. aureus strains, although no genes have been annotated, and there also is a lack of consensus on whether LTA glycosylation occurs in all S. aureus lineages (10, 36, 43). In order to rapidly analyze LTA for glycosylation, we sought to develop a quick PAGE method to monitor LTA structure analogous to that used for WTA analysis (see Fig. S1 in the supplemental material). Initial attempts at separating purified LTA directly on PAGE were poorly resolved since no discrete bands were observed (see Fig. S2, lane 1, in the supplemental material). We posited that LTA, unlike hydrolyzed WTA, aggregates due to the covalently attached diacylglycerol moiety. In order to deacylate LTA while preserving glycosidic linkages, we screened a panel of lipases (see Fig. S2 in the supplemental material). In contrast to acid or base deacylation (44), pretreatment with Resinase HT (Strem Chemicals) consistently produced single-band resolution with no concurrent LTA polymer degradation upon PAGE analysis.

With a facile assay in hand, LTA PAGE profiles were compared to assess the effect of CsbB-GtcA-YfhO overexpression (Fig. 1). Overexpression of CsbB-GtcA caused the discrete single-band resolution to disappear and produce a “smear” in the wild type, regardless of whether YfhO was cooverexpressed or expressed at endogenous chromosomal levels. Smearing is consistent with induced LTA structural heterogeneity, as would be expected from glycosylation. Moreover, this is a YfhO-dependent phenotype since there was no effect in the ΔyfhO background even when P_cadcsbB-gtcA was present. The diagnostic smearing could be restored with plasmid borne YfhO only when P_cadcsbB-gtcA was also present. Hence, basal expression of YfhO from the chromosomal copy appears sufficient to modify LTA, and significant LTA modification depends on constitutive CsbB-GtcA expression.

FIG 1.

PAGE analysis of purified LTA from RN4220 derived strains. LTA was first treated with Resinase HT at pH 8.5 for 16 h to hydrolyze the lipid chains and d-Ala esters. This enabled single-band resolution of the LTA poly-GroP backbone, producing a profile consistent with minimal glycosylation in the wild type. Modification of LTA is yfhO dependent and, provided P_cadcsbB-gtcA is expressed, induces a loss in band resolution. Each band indicates a poly-GroP polymer of n+1 repeat length.

CsbB and YfhO glycosylates LTA with GlcNAc.

To confirm the PAGE observations and quantify the levels of glycosylation, purified LTA samples extracted from early-stationary-phase cultures of S. aureus RN4220 were analyzed by ¹H NMR (Fig. 2). In wild-type RN4220, very low intensity signals with chemical shifts consistent with both the α-anomeric and methyl group protons of GlcNAc were observed (Fig. 2A). These peaks are consistent with the α-GlcNAc assignments previously made by Hartung and coworkers using LTA extracted from S. aureus subsp. aureus Rosenbach 1884 (strain H, DSM 20233) (36). However, the amount of GroP subunits modified with GlcNAc is at most 2% in comparison to the ∼15% levels previously reported in strain H LTA (see Table S2 in the supplemental material). LTA analysis from the S. aureus subsp. aureus NCTC8325 HG003 derived strain TM226 (the RN4220 strain parent lineage) likewise revealed an extremely low basal GlcNAc content. Introduction of P_cadcsbB-gtcA into the RN4220 wild type increased the GlcNAc levels to 5%, whereas introduction of both P_cadcsbB-gtcA and P_cadyfhO further raised the GlcNAc levels to 6% (see Fig. S3 and Table S2 in the supplemental material). Deletion of chromosomal yfhO completely removed any GlcNAc related peaks, regardless of whether P_cadcsbB-gtcA was present (Fig. 2C). Overexpression of YfhO alone only slightly increased the GlcNAc levels beyond the trace amount observed in the wild type (Fig. 2A, bottom panel), indicating that (lack of) expression of CsbB-GtcA is the main factor controlling LTA glycosylation levels in wild-type S. aureus.

FIG 2.

Expression of CsbB-GtcA and YfhO induces GlcNAcylation of LTA in S. aureus. (A) NMR spectra of purified LTA extracted from various strains. Shading of peaks corresponds to LTA structural components highlighted in panel B and whose integration was used in calculating d-Ala and GlcNAc percent compositions (summarized in Fig. S3; see also Table S2 in the supplemental material). Strains were as follows: RN4220, wild type; TM684, P_cadcsbB-gtcA; KK791, P_cadcsbB-gtcA + P_cadyfhO; KK737, ΔyfhO; KK753, ΔyfhO + P_cadcsbB-gtcA; and KK860, P_cadyfhO. (C) Signal from the α-anomeric proton (∼5.1 ppm) and the acetate proton (∼2.1 ppm) of GlcNAc (in boldface text in panel B). Spectra are color coded to match the labels in panel A.

Since S. aureus elaborates many GlcNAc containing cell envelope polymers including peptidoglycan and poly-N-acetylglucosamine that could potentially contaminate LTA-purified preparations (45–47), it was necessary to unambiguously confirm covalent attachment of GlcNAc to the poly-GroP backbone. We thus adapted a hydrofluoric acid (HF)-mediated WTA depolymerization method for LTA monomer analysis by electron spray ionization-mass spectrometry (ESI-MS) (48–50). LTA extracted from two different strains harboring P_cadcsbB-gtcA were compared; one control sample from a ΔyfhO background (KK753) and the other from a strain cooverexpressing YfhO (Fig. 3). From both strains, the dominant mass profile observed was a family of peaks consistent with protonated (dehydrated) and sodiated glycerol-d-Ala LTA monomer (m/z = 164.1, 146.1, and 186.09, respectively). Thus, the majority of notoriously labile d-Ala esters (23) remained largely intact during HF treatment. In addition, a number of ions could be assigned to the acylglycerol moiety and gentiobiose core from both samples (see Table S3 in the supplemental material). Three prominent strain-specific ion signals corresponding to GlcNAc were observed at m/z 204.10 (free GlcNAc, oxonium), 296.15 (glycerol-GlcNAc, H⁺), and 318.14 (glycerol-GlcNAc, Na⁺) when CsbB-GtcA-YfhO was overexpressed (Fig. 3A). In contrast, no GlcNAc signals were detected in the ΔyfhO background (Fig. 3B), confirming both covalent GlcNAc attachment to LTA and a complete dependence on YfhO activity for glycosylation.

FIG 3.

Electron spray ionization-mass spectrometry of KK791 (RN4220 + P_cadcsbB-gtcA + P_cadyfhO) (A) and KK753 (ΔyfhO + P_cadcsbB-gtcA) (B). The GlcNAc-related ions are observed at m/z 204.10 (GlcNAc, oxonium), m/z 296.15 (glycerol-GlcNAc, H⁺), and m/z 318.14 (glycerol-GlcNAc, Na⁺) in KK791, whereas these signals are absent in KK753. A complete list of assignable signals is provided in Table S3 in the supplemental material.

Depletion of d-Ala promotes high level LTA glycosylation.

Based on the NMR data (Fig. 2; see also Table S2 in the supplemental material), C2 GroP hydroxyls from wild-type RN4220 LTA are ∼85% substituted by d-Ala, a finding which is in agreement with previous reports (20, 36). The level of d-alanylation among the purified LTA extracts remained in the 80 to 90% range, regardless of how much GlcNAc modification (up to 6% maximum) was achieved through the overexpression of CsbB-GtcA-YfhO. We hypothesized that this glycosylation ceiling may not solely be due to limited gene expression but rather is capped by competition with the d-alanylation machinery. Simply put, if LTA GroP approaches 90% occupation by d-Ala, the maximum amount of GlcNAc substitution can only be 10% since both moieties are attached to the C2 hydroxyl of the poly-GroP backbone.

To test this hypothesis, we replaced the d-Ala carrier protein dltC with a kanamycin resistance cassette and analyzed LTA by PAGE and ¹H NMR (Fig. 4). While the overall amount of LTA extracted from the dltC::Kan^r strain was significantly lower than in the wild type, it was apparent that the extracted LTA migrated more slowly than that of the wild-type parent (Fig. 4A). We then overexpressed CsbB-GtcA-YfhO in the dltC::Kan^r background, which restored the levels of extractable LTA to near wild-type levels. In addition, LTA from this strain was shifted to a higher molecular mass and pronouncedly smeared, as would be expected for high-level GlcNAc modification. To quantify this, LTA extracted from both strains was analyzed by ¹H NMR (Fig. 4B). The levels of d-Ala were significantly lower in the dltC::Kan^r background, as would be expected. More importantly, the ΔdltC LTA was now 8% substituted with GlcNAc (∼4.5-fold higher than in the dltC⁺ parent strains). Overexpression of CsbB-GtcA-YfhO in the dltC::Kan^r mutant further elevated GlcNAc levels to 28% (Fig. 4C), indicating substitution in one of every four GroP repeating units. This degree of substitution was by far the highest observed in any of the tested constructs (see Fig. S3 and Table S2 in the supplemental material), suggesting that a combination of overexpression of the three-component glycosylation system in tandem with lowered d-alanylation activity is the most effective route to highly glycosylated LTA.

FIG 4.

Depletion of d-Ala promotes GlcNAc modification of LTA. (A) PAGE of column-purified LTA extracted from wild-type RN4220, TM582 (dltC::Kan^r), and KK868 (dltC::Kan^r + P_cadcsbB-gtcA + P_cadyfhO) strains. When CsbB-GtcA-YfhO are overexpressed, a slowly migrating smear diagnostic of substantial LTA modification is observed. (B and C) NMR spectra of LTA isolated from dltC::Kan^r (B) and dltC::Kan^r (C) strains overexpressing CsbB-GtcA-YfhO.

Sodium chloride-induced cell stress increases GlcNAc modification of LTA.

Our results indicated that expression of CsbB-GtcA (or, conversely, a lack of chromosomal expression in the wild type) is the key genetic determinant controlling whether LTA is modified with GlcNAc residues in S. aureus. These two genes putatively load Und-P with GlcNAc inside the cytoplasm (CsbB) and flip the sugar-loaded carrier to the outer surface of the cytoplasmic membrane (GtcA) in a three-component glycosylation system completed by the LTA α-O-GlcNAc transferase (YfhO) whose chromosomal expression appeared to be sufficient to realize GlcNAcylation. To narrow which of the two genes (CsbB or GtcA) may be the dominant factor in regulating LTA glycosylation, we overexpressed them singly on a plasmid and monitored the resulting LTA structure by PAGE (see Fig. S4 in the supplemental material). Since the expression of CsbB was clearly the most important determinant in inducing LTA modification, we next focused on potential regulation of the CsbB promoter to identify environmental conditions that may induce LTA GlcNAcylation in the wild type. Bioinformatics analysis using the comprehensive transcript data set of Mäder et al. suggests that SAOUHSC_00713 is under the control of the cell wall stress-induced alternative sigma factor SigB (51). However, RN4220 is known to be deficient in SigB expression due to a small deletion in rsbU, a component of the SigB activation cascade (52). Regulation experiments were thus conducted in the rsbU⁺ strain TM226 that, like RN4220, also does not appreciably modify LTA with GlcNAc under standard growth conditions. To further corroborate intact csbB regulation in TM226, sigB was cloned into a multicopy overexpression vector under the control of the protein A (spa) gene promoter (53). Gene expression analysis by reverse transcription-quantitative PCR (RT-qPCR) confirmed 5-fold overexpression of sigB, resulting in overexpression of both csbB (2.4-fold) and the sigB expression control marker asp23 gene (18-fold) (54). Once more, LTA was now highly modified with GlcNAc at 10% when sigB was overexpressed.

To determine whether LTA GlcNAcylation is indeed controlled by SigB under environmentally induced stress cues, S. aureus was subjected to a panel of known cell envelope stress-inducing conditions previously reported to induce SigB (55, 56). While we did not observe altered LTA PAGE profiles for all conditions tested, LTA extracted from cells grown in tryptic soy broth (TSB) supplemented with 10% NaCl or 30 mM KOH produced PAGE profiles diagnostic of modified LTA (Fig. 5). NaCl was thus chosen for further study. To assess the effects of high-salt stress induction over time on CsbB-GtcA-YfhO gene expression and the corresponding changes to LTA structure, a 4-h time course experiment was conducted after shock exposure to 10% NaCl (Fig. 6A). Strikingly, the loss of LTA band resolution intensified with increased time of NaCl exposure with discrete single banding disappearing as early as the 1-h time point. During this initial 1-h period, there was at most a single round of cell division. Based on these results, a large-scale S. aureus culture was grown in 10% NaCl to mimic the 4-h growth point and the purified LTA analyzed by ¹H NMR (Fig. 6B). LTA was appreciably glycosylated with GlcNAc at 5.5% when extracted from cells cultured under NaCl cell envelope stress.

FIG 5.

PAGE of LTA samples extracted from wild-type TM226 grown under cell envelope stress-inducing conditions. LTA samples were extracted from cultures grown for approximately five generations under the indicated stress-inducing conditions. When subjected to growth in the presence of 10% NaCl or 30 mM KOH, single-band resolution was eliminated in comparison to the well-resolved bands present in the wild type.

FIG 6.

LTA modification and gene expression analysis of S. aureus TM226 grown in 10% NaCl-containing TSB medium. (A) LTA was harvested for PAGE at various time points. Single-band resolution of LTA is gradually lost as the high-salt exposure time is prolonged. (B) NMR spectra of LTA from wild-type TM226 grown in TSB supplemented with 10% NaCl. The GlcNAc composition of LTA increased to approximately 5.5%, while d-Ala decreased to 72.8%. (C) High-salt stress induces the expression of SigB-regulated genes, including the CsbB-GtcA genes, as measured by RT-qPCR.

Parallel time point culture aliquots were withdrawn and analyzed in tandem for gene expression by RT-qPCR (Fig. 6C). To gauge the extent of SigB activation, expression of both sigB and the known SigB regulated marker asp23 (54) were monitored. There was a sharp increase in both sigB and asp23 transcript levels at 5 min. Likewise, both csbB and gtcA were upregulated 8-fold and 2-fold at the same time point. Over time, the levels of all of these upregulated targets waned, although csbB expression levels remained somewhat elevated out to 4 h. In contrast, yfhO and dltB expression did not change throughout the time course of the experiment.

DISCUSSION

There is growing appreciation for the role and dynamic nature of LTA modifications in altering cellular function, and here we report the identification and preliminary characterization of genes from S. aureus involved in LTA glycosylation with GlcNAc residues. We demonstrate that the expression of both CsbB and YfhO is absolutely required to modify LTA with GlcNAc residues in a pathway that likely resembles other characterized three-component glycosylation systems (57). These glycoconjugate systems are widespread in bacteria, modifying O-antigens, lipid A, mycolylarabinogalactan, and other cell surface polymers in addition to TAs. Glycosylation involves a three component pathway similar to the prototype O-antigen glycosylation outlined for the Shigella flexneri serotype converting bacteriophage Sfx (41). First, a DPM1 (dolichol-phosphate mannose) family glycosyltransferase charges an undecaprenol lipid carrier with a sugar from an NDP-sugar donor to form a membrane-associated Und-P-sugar. The CsbB protein belongs to this family of enzymes and has a predicted topology consistent with the characteristic two C-terminal transmembrane α-helices. Next, the Und-P-sugar is transported to the outer leaflet of the cell membrane by a small integral membrane “flippase” typically possessing four transmembrane α-helices. In S. aureus, GtcA was identified as the most likely candidate for this role based on the predicted topology. Finally, a second glycosyltransferase belonging to the GT-C family then utilizes Und-P-sugar donor to glycosylate an acceptor molecule to form the product glycoconjugate. GT-C family enzymes share limited overall primary sequence homology but are generally multipass integral membrane proteins (8 to 13 transmembrane helices) with an active-site aspartate-containing motif in the first extracytoplasmic loop (58, 59). The candidate LTA α-O-GlcNAc transferase YfhO identified here possesses all of these characteristics.

Unlike most three component glycosylation systems, however, the genes for CsbB, GtcA, and YfhO are located at noncontiguous positions spread throughout the S. aureus chromosome. The lack of clustering with any TA-associated genes in S. aureus makes YfhO family glycosyltransferase substrate prediction challenging, particularly given the similarity in TA structure between type I poly-GroP LTA and poly-RboP WTA. In certain serovars of L. monocytogenes (31), YfhO is a WTA GlcNAcylating enzyme, while in S. aureus it is now clear that LTA is the preferred ligand. In the absence of any known TA discriminating amino acid sequence motifs, experimental characterization remains necessary. While both TA substrates are structurally similar phosphodiester linked alditol polymers, WTA and LTA have mechanistically distinct biosynthetic pathways (9, 10, 13). LTA is polymerized on the outer cell surface by the enzyme LtaS, which polymerizes GroP from phosphatidylglycerol donors (37, 60), while WTA is built from polymerization of RboP entirely within the cytoplasm (9). In this respect, LTA is more similar to Wzy-dependent O-antigen pathways whereby the polysaccharide is polymerized from building blocks on the extracellular surface, and WTA is more similar to ATP-binding cassette (ABC) transporter pathways that extrude a polymer completed entirely within the cytoplasm. Mann et al. have demonstrated that the same three-component glycosylation system is capable of modifying both types of glycopolymer-forming systems (61), providing conceptual evidence for the mechanistic feasibility of a common GT-C family glycosyltransferase to modify preassembled (WTA) or surface-polymerized (LTA) glycopolymers using common machinery.

In contrast to S. aureus, where GlcNAcylated LTA is difficult to detect in most strains (10), many Bacillus spp. elaborate LTA richly substituted with GlcNAc modifications under standard culture conditions (62). An LTA glycosylation pathway using Und-P-sugar donor was originally proposed (62), which has been supported by recent work in B. subtilis proposing a three-component LTA GlcNAcylation system similar to the one described here in S. aureus (63). In B. subtilis, yfhO and csbB are cotranscribed in the same transcriptional unit that can be further induced by the extracytoplasmic sigma factor σ^X, as well as the alternative sigma factor σ^B (64, 65). These genes were thus suggested to be involved in modifying the cell envelope in response to stress possibly through glycosyl transferase activity, though their exact functions remained unclear at that time. Subsequent work in B. subtilis discovered that csbB expression in the absence of yfhO induced marked cell stress, likely due to Und carrier lipid sequestration as Und-P-sugar dead-end intermediates (66). While these studies are largely consistent with a common role for LTA glycosylation in both B. subtilis and S. aureus, there are clearly unique aspects to how the CsbB-GtcA-YfhO system functions in S. aureus. The most apparent difference is in the tightly repressed basal level of LTA GlcNAcylation, since less than 2% of the GroP units are modified with GlcNAc in the wild type (Fig. 2). Low expression of CsbB appears to dictate LTA GlcNAc levels in S. aureus. Regulatory control at the Und-P sugar loading step is a logical checkpoint, given the known toxicity of accumulating Und dead-end intermediates (6, 66, 67). By regulating flux into the three-component glycosylation pathway, no Und intermediates can accumulate and peptidoglycan/WTA assembly will not have to compete with a depleted pool of carrier. In contrast to CsbB, we observed basal YfhO expression under standard culture conditions in both SigB-positive (TM226) and SigB-defective (RN4220) genetic backgrounds and did not observe appreciable YfhO upregulation upon sigB overexpression or salt-induced cell envelope stress. Consistent with this result, other transcriptional profiling studies in S. aureus have not identified YfhO as a member of the SigB regulon. Both CsbB and the putative transporter GtcA, however, have SigB-driven promoters (50, 51, 68), and we were able to observe induction upon high-salt exposure (Fig. 6C). The tight basal regulation and transient induction under certain cell culture conditions is a satisfying explanation for variations in previous reports regarding LTA-GlcNAc levels in S. aureus. Since the genetic determinants CsbB-GtcA-YfhO identified here are distributed ubiquitously throughout the species, strain-specific LTA glycotypes are a less likely explanation.

The SigB regulon is the dominant alternate sigma factor responsible for mounting cell envelope stress response in S. aureus, controlling a variety of cell wall-related processes conferring antibiotic resistance and adaption to environmental stress (50, 54, 56). The LTA GlcNAc tailoring modification observed here presumably enables the bacteria to alter its cell wall physiology to tolerate such environmental challenges. Curiously, we did not observe LTA GlcNAcylation under all of the classical SigB stress-inducing conditions tested (Fig. 5). High-NaCl conditions and 30 mM KOH treatments were by far the most potent effectors, inducing a remarkably rapid phenotypic shift in the LTA PAGE profile (Fig. 5). Conversion was apparent after at most a single round of cell division had occurred (Fig. 6A). Once more, while we did observe robust induction of csbB within minutes of high-salt exposure and a decrease in d-Ala LTA content (Fig. 6B and C), we did not observe a coordinated transcriptional repression of the dlt operon, as previously reported for S. aureus and Lactobacillus casei when cultured in high-salt media (69, 70). The downregulation of dlt would help rationalize how such a rapid change in the LTA GlcNAcylation state could be achieved, particularly if the sole mechanism for switching to a GlcNAc-rich LTA is through dilution of unmodified mature LTA with a highly GlcNAcylated nascent polymer via cell division. This raises the intriguing possibility of GlcNAcylation of mature LTA already within the cell envelope prior to challenge with NaCl or KOH. For this to occur, existing d-Ala linkages would likely have to be removed since the two LTA modification systems are in competition for a shared poly-GroP substrate. Transcription-independent mechanisms that lower d-Ala LTA content have been proposed. Kiriukhin and others have demonstrated that high NaCl concentrations inhibit transacylation from d-alanyl-DltC to the C2-OH of GroP and stimulate hydrolytic cleavage of the d-alanyl thioester (23, 71). Alternatively, a recently described TA d-Ala esterase FmtA could play a role in reducing the LTA d-alanylation content under certain conditions (72). The mechanism of dealanylation by KOH treatment is more straightforward, as d-Ala LTA ester linkages are extremely labile to base. Regardless of the mechanism, modification of mature LTA chains would require reorientation so as to place the acceptor 2-OH of the GroP repeating unit within the putative active site of YfhO. The active site of GT-C family enzymes is likely within extracytoplasmic loop 1 at the membrane surface (58, 59). LTA conformation is highly dynamic, and biophysical studies have demonstrated a reversible transition from an extended conformation to a random coil that packs against the membrane surface in a high-NaCl solution (23, 73). It remains unknown, however, how GlcNAcylation impacts LTA structure and function, but the induction by cell envelope stress demonstrated here would suggest a net stabilizing affect.

MATERIALS AND METHODS

Strains, plasmid construction, and growth conditions.

All S. aureus strains used in this study are listed in Table 1 and are derived from the reference strain NCTC8325. Gene deletions were made with the E. coli-S. aureus shuttle vector pKFC for allelic exchange using 1-kb flanking DNA homology arms amplified by PCR (33). TM852 was generated using pKFC with the dltC::Kan^r cassette from SHM084. Plasmid assemblies were performed using an In-Fusion cloning kit (Clontech). S. aureus was typically grown in TSB at 37°C with agitation. Where necessary, antibiotic markers were selected with chloramphenicol (10 μg/ml), erythromycin (10 μg/ml), or a combination of kanamycin and neomycin (25 μg/ml each). The strains and plasmids are listed in Table 1.

TABLE 1.

Strains and plasmids used in this study^a

Strain or plasmid	Genotype	Source or reference
Strains
RN4220		Kreiswirth et al. (74)
TM226	S. aureus HG003 ϕ11::FRT	Santiago et al. (75)
SHM084	S. aureus Newman dltC::Kanr	S. Walker, Harvard Medical School (unpublished data)
TM684	RN4220(PcadcsbB-gtcA)	This study
KK713	RN4220 tarO::Tetr	This study
KK717	RN4220 tarO::Tetr(PcadcsbB-gtcA)	This study
KK731	RN4220 ΔSA664 (tarM)	This study
KK737	RN4220 ΔyfhO	This study
KK739	RN4220 ΔSA228 (tarS)	This study
KK742	RN4220 ΔtarM ΔtarS	This study
KK753	RN4220 ΔyfhO(PcadcsbB-gtcA)	This study
KK776	RN4220(PcadcsbB)	This study
KK777	RN4220(PcadgtcA)	This study
KK791	RN4220(pLI50, PcadcsbB-gtcA, PcadyfhO)	This study
TM852	RN4220 dltC::Kanr	This study
KK860	RN4220(PcadyfhO)	This study
TM864	RN4220(PspAsigB)	This study
KK868	RN4220 dltC::Kanr(PcadcsbB-gtcA, PcadyfhO)	This study
Plasmids
pKFC	E. coli-S. aureus shuttle vector for cloning and allelic exchange; Carbr (E. coli), Cmr (S. aureus)	Kato and Sugai (33)
pLI50	E. coli-S. aureus shuttle vector; Carbr (E. coli), Cmr (S. aureus)	Lee et al. (76)
pLI50 PcadC	pLI50 with Pcad cadC for cadmium-inducible expression; Carbr (E. coli), Cmr (S. aureus)	This study
pCN59	E. coli-S. aureus shuttle vector, Pcad cadC blaZ transcriptional terminator; Carbr (E. coli), Apr (S. aureus), Emr (S. aureus)	Charpentier et al. (77)
PspAsigB	pLI50 with protein A (spa) promoter for constitutive gene expression; Carbr (E. coli), Cmr (S. aureus)	This study
PcadcsbB-gtcA	pLI50 PcadC cadC expressing both SAOUHSC_00713 and SAOUHSC_002722	This study
PcadyfhO	pCN59 expressing SAOUHSC_01213	This study
PcadcsbB	pLI50 PcadC expressing SAOUHSC_00713	This study
PcadgtcA	pLI50 PcadC expressing SAOUHSC_002722	This study

^aAll S. aureus strains are derivatives of the reference strain NCTC8325. Phenotypes: Kan^r, kanamycin and neomycin resistance; Carb^r, carbenicillin resistance; Cm^r, chloramphenicol resistance; Em^r, erythromycin resistance; Ap^r, apramycin resistance.

WTA extraction.

WTA was isolated as previously described (34). Briefly, 20 ml of stationary-phase S. aureus was pelleted, washed with 30 ml of buffer 1 [50 mM 2-(N-morpholino)ethanesulfonic acid (MES), pH 6.5] and resuspended in 30 ml of buffer 2 (50 mM MES, 4% [wt/vol] sodium dodecyl sulfate [SDS]). Samples were placed in a boiling water bath for 1 h and pelleted. Cell wall sacculus was resuspended in buffer 2, transferred to a 2-ml microcentrifuge tube, and pelleted by centrifugation (16,000 × g, 1 min). The cells were washed consecutively with buffer 2, buffer 3 (50 mM MES [pH 6.5], 2% NaCl), and then finally with buffer 1. The pellet was then resuspended in 1 ml of 20 mM Tris-HCl (pH 8.0)–0.5% (wt/vol) SDS containing 20 μg of proteinase K, followed by incubation at 50°C for 4 h on a shaking heat block. Samples were then washed once with buffer 3 and three times with MilliQ distilled H₂O (dH₂O) to remove SDS before being thoroughly resuspended in 1 ml of 0.1 M NaOH. Hydrolysis was performed at room temperature for 16 h with constant agitation before neutralization with 250 μl of 1 M Tris-HCl (pH 7.8). Insoluble cell wall debris was removed by centrifugation (16,000 × g, 1 min), and the WTA-containing supernatant was extracted into a clean tube.

TA-PAGE.

For PAGE analysis of both WTA and LTA, a separating gel (20%T, 6%C) and stacking (3%T, 0.26%C) was cast using a Bio-Rad Protein II xi system (20 cm by 16 cm by 0.75 mm) as previously described (34). Crude extracts were diluted 1:3 in loading buffer (50% glycerol in running buffer with a trace of bromophenol blue) to a maximum sample volume of 15 μl. The gel was developed at 4°C with a constant current of 40 mA per gel for approximately 18 to 20 h in Tris-Tricine running buffer (0.1 M Tris, 0.1 M Tricine [pH 8.2]). Gels were then stained with alcian blue (1 mg/ml, 3% acetic acid), followed by silver staining, according to established protocols (35).

LTA purification.

LTA was purified using the 1-butanol extraction method as described elsewhere (36, 37). A 20-ml TSB starter culture was grown until mid-log phase and used to inoculate 4 liters of TSB. After overnight incubation, stationary-phase cells were pelleted and washed with 100 ml of resuspension buffer (50 mM sodium citrate [pH 4.7]). Washed cells were kept cool and disrupted by bead beating with zirconia-silica beads (0.1-mm-diameter beads; four 2.5-min pulses; BioSpec Products BeadBeater). Cell envelope components were collected by centrifugation (30,000 × g, 1 h) and resuspended in 70 ml of resuspension buffer, to which 50 ml of 1-butanol was added. Samples were incubated at 37°C for 45 min with shaking and then centrifuged (30,000 × g, 1 h) to remove insoluble material and induce phase separation. The lower aqueous phase was lyophilized and resuspended in 5 ml of equilibration buffer (15% 1-propanol in 50 mM citrate [pH 4.7]). The crude lysate was centrifuged (20,000 × g, 10 min) and filtered (0.22-μm pore size) to remove residual debris. The clarified supernatant was then subjected to hydrophobic interaction chromatography on a HiPrep Octyl 4FF (16/10) column (GE Healthcare) and LTA eluted with a linear gradient of 15 to 60% 1-propanol in 50 mM citrate (pH 4.7). Fractions were analyzed for phosphate content by colorimetric molybdate assay (38). Briefly, 100 μl of each fraction was combined with 200 μl of ashing solution (2 M H₂SO₄, 0.44 M HClO₄) and incubated uncapped at 120°C for 3 h in a fume hood. Then, 1 ml of reducing solution (3 mM ammonium molybdate, 0.25 M sodium acetate, 1% ascorbic acid) was added, followed by incubation in a 45°C water bath for 2 h with occasional mixing. Subsequently, 250 μl of each fraction was transferred to a 96-well plate and read at 700 nm. Phosphate-containing fractions were combined, dried on a rotary evaporator, and dissolved in 5 ml of MilliQ dH₂O for extensive dialysis using cellulose ester dialysis membranes (molecular mass cutoff, 0.5 to 1.0 kDa; Spectrum Labs). The dialysate was lyophilized in a preweighed vial to determine LTA yield (typically 5 to 15 mg per 4 liters of culture) and stored at −20°C.

¹H NMR.

Purified LTA was deuterium exchanged by two rounds of lyophilization using 500 μl of 99.8% D₂O. LTA was then resuspended in 550 μl of 99.9% D₂O for NMR analysis using a Bruker Avance III 600 at 298K with water presaturation, a 10-ppm spectral width, a 5.45-s acquisition time, a 5-s recycle delay, a 0.3-Hz line broadening, and 512 scans. The collected spectra were analyzed and quantified using Bruker TopSpin 3.5. The chemical shifts assigned to each LTA component were integrated and adjusted for the number of protons, as described previously (36). Briefly, the GlcNAc anomeric and acetate integrals were averaged for four protons (GlcNAc_av.). The integral between 3.7 and 4.2 ppm was added to the Gro-3-CH integral, and the underlying GlcNAc resonances (5H) were deducted to obtain the total amount of glycerol (5H⁺). Average LTA chain lengths were estimated by determining the ratio of glycerol (5H⁺) to CH₂+CH₃ (average, 58H⁺) of the lipid chains, the levels of d-Ala modification by the ratio of d-Ala-αH to glycerol (5H⁺) and, similarly, the level of GlcNAc modification by the ratio of GlcNAc_av. to glycerol (5H⁺). Standard deviations were calculated from at least two experimental repetitions.

Small-scale LTA extraction for PAGE.

Either 20 ml of stationary-phase cells or 10 optical density units/ml of exponentially growing cells were pelleted and washed once with 2 ml of resuspension buffer (50 mM sodium citrate [pH 4.7]). The cells were disrupted on a MagNA Lyser (Roche) with four 30-s pulses (7,000 rpm), and the cell envelope fragments were pelleted (20,000 × g, 4°C, 1 h). Pellets were resuspended in 350 μl of resuspension buffer, and an equal volume of 1-butanol was added. The extraction was carried out on a shaking heat block at 37°C for 45 min. The crude extract was centrifuged (20,000 × g, 4°C, 1 h) to pellet insoluble materials, and the lower aqueous phase was then extracted into a clean tube.

For PAGE analysis, LTA was first enzymatically deacylated with lipase. For the butanol aqueous-phase crude extract, 50 μl of the aqueous layer was combined with 50 μl of 50 mM Tris (free base) and then adjusted to pH 8.5 with 2 μl of 1 M NaOH. For column-purified LTA, samples were dissolved to 5 μg/μl in 100 μl of 10 mM Tris (pH 8.5). For both crude and purified LTA, deacylation was carried out by adding 1 μl of Resinase HT, and the reaction mixture was incubated at 50°C for 16 h. Lipase-treated samples were then analyzed by PAGE and stained as described above.

Dephosphorylation of LTA.

LTA was monomerized by dephosphorylation with hydrofluoric acid. Approximately 1 mg of purified LTA was resuspended in 100 μl of 47% HF and incubated at room temperature for 20 h. HF was then evaporated in a closed vessel with a stream of filtered air until there was no residual liquid using a KOH pellet trap with the outlet tubing inserted into a saturated base solution. Dried samples were resuspended in 200 μl of 5 mM ammonium bicarbonate, and the pH was adjusted to 7 with dilute ammonium hydroxide before lyophilization.

ESI-MS.

MS analysis was performed on a Waters Q-TOF Premier quadrupole/time of flight mass spectrometer (Waters Corporation [Micromass, Ltd.], Manchester, UK). Operation of the mass spectrometer was performed using MassLynx software version 4.1 (Waters). Samples were introduced into the mass spectrometer using a Waters 2695 high-performance liquid chromatograph. The samples were analyzed using flow injection analysis. The mobile phase consisted of 50% acetonitrile (LC-MS grade) and 50% aqueous 10 mM ammonium acetate with a flow rate of 0.15 ml/min. The nitrogen drying gas temperature was set to 300°C at a flow of 7 liters/min with a capillary voltage of 2.8 kV. The mass spectrometer was set to scan from 100 to 1,000 m/z in positive- and negative-ion modes, using ESI.

Cell envelope stress response induction.

For stress-inducing conditions, overnight cultures were diluted 1:100 into prewarmed TSB at 37°C. Upon reaching an optical density at 600 nm (OD₆₀₀) of ∼0.2, cultures were diluted 10-fold into prewarmed TSB with final concentrations of 7% ethanol (EtOH), 10% NaCl, 30 mM KOH, 1 mM MnCl₂, or 0.006% Tween 20. Cultures were grown to an OD₆₀₀ of 0.5, and 20 ml was harvested for small-scale LTA extraction as detailed above. A second 1-OD unit/ml equivalent of the cell aliquot was rapidly pelleted and resuspended in 1 ml of RNAlater (Applied Biosystems). Pelleted cells were stored at −20°C before RNA extraction for RT-qPCR.

RT-qPCR.

RNA was extracted using the SurePrep TrueTotal RNA purification kit (Fisher Scientific) according to the manufacturer's protocol with modifications. To facilitate S. aureus lysis, the lysozyme digestion solution was supplemented with 13 μg of lysostaphin. The on-column DNase I digestion was performed using 2 U of DNase I (RNase-free; NEB). Upon elution, RNA was treated again with 2 U of DNase I in a 100-μl reaction at 37°C for 15 min. Total RNA was cleaned by a second column purification. RNA integrity was assessed by morpholinepropanesulfonic acid-formaldehyde-agarose gel electrophoresis and quantified by determining the absorbance at 260 nm. Synthesis of cDNA was carried out using a Maxima first-strand cDNA synthesis kit (Thermo), while real-time qPCRs were performed using PowerUP SYBR green master mix (Applied Biosystems) on an AriaMX real-time PCR system (Agilent). Data analysis was performed using the accompanying software. Primers (see Table S1 in the supplemental material) were designed and validated according to MIQE guidelines (39).