Staphylococcus aureus Glucose-Induced Biofilm Accessory Protein A (GbaA) Is a Monothiol-Dependent Electrophile Sensor

Abstract

Staphylococcus aureus is a commensal pathogen that has evolved to protect itself from unfavorable conditions by forming complex community structures termed biofilms. The regulation of the formation of these structures is multifactorial and in S. aureus involves a number of transcriptional regulators. GbaA (glucose-induced biofilm accessory protein A) is a tetracycline repressor (TetR) family regulator that harbors two conserved Cys residues (C55 and C104) and impacts the regulation of formation of poly-N-acetylglucosamine-based biofilms in many methicillin-resistant S. aureus (MRSA) strains. Here, we show that GbaA-regulated transcription of a divergently transcribed operon in a MRSA strain can be induced by potent electrophiles, N-ethylmaleimide and methylglyoxal. Strikingly, induction of transcription in cells requires C55 or C104, but not both. These findings are consistent with in vitro small-angle X-ray scattering, chemical modification, and DNA operator binding experiments, which reveal that both reduced and intraprotomer (C55–C104) disulfide forms of GbaA have very similar overall structures and each exhibits a high affinity for the DNA operator, while DNA binding is strongly inhibited by derivatization of one or the other Cys residues via formation of a mixed disulfide with bacillithiol disulfide or a monothiol derivatization adduct with NEM. While both Cys residues are reactive toward electrophiles, C104 in the regulatory domain is the more reactive thiolate. These characteristics enhance the inducer specificity of GbaA and would preclude sensing of generalized cellular oxidative stress via disulfide bond formation. The implications of the findings for GbaA function in MRSA strains are discussed.

Graphical Abstract

Staphylococcus aureus is a Gram-positive commensal pathogen of considerable socioeconomic impact due to its ability to cause a wide range of infections and its extensive multidrug resistance. Clinically acquired methicillin-resistant S. aureus (MRSA) was first discovered in 1961, immediately following the introduction of the potent β-lactam antibiotic methicillin.¹ Various Staphylococcus spp. are known to be members of the human microbiota and constitute upward of 50% of the nasal microbiota in humans;² Staphylococcus spp. are therefore opportunistic pathogens that gain entry into host tissues through wounds, syringes, or implanted medical devices.

USA300 MRSA strains are virulent and readily transmissible.³ The most common impact of USA300 strains is skin and soft tissue infections (SSTI), but they also cause toxic shock syndrome, necrotizing pneumonia, and endocarditis, largely in immunocompromised patients.³ Whole genome sequencing of USA300 versus the Newman strain,⁴ a methicillin-sensitive S. aureus (MSSA) strain, reveals Panton-Valentine leucocidin (PVL) and arginine catabolic mobile element (ACME) to be common genetic elements in community-associated (CA) MRSA that is proposed to contribute to the enhanced virulence of these strains.^5,6 The success of USA300 strains in causing disease is often attributed to the increased level of expression of global transcriptional regulators agr and sae, which are responsible for quorum sensing and expression of virulence factors in Staphylococcus spp., respectively.⁷ The deletion of agr and sae independently affected the expression of virulence factors RNAIII, α-hemolysin, and PVL among other virulence genes.^7,8 Another advantageous factor for CA-MRSA infection is the ability to produce biofilms that are known to be negatively regulated by the global regulators agr and sae.^9,10

Biofilms are polymeric conglomerates of extracellular DNA, polysaccharides, and proteins. Biofilm production represents a strategy deployed by a vast number of microorganisms to attach to surfaces, to cope with nutrient scarcity, to tolerate antibiotic insult, or to avoid opsonization by host macrophages.^11–13 Biofilms are generally comprised of a community of various microbial species and promote horizontal gene transfer; as a result, biofilms not only confer resistance against antibiotics but also contribute to the rapid spread of antibiotic resistance within the community.¹⁴ USA300 MRSA strains are known to form robust biofilms. Biofilms in Staphylococcus ssp. are broadly classified as polysaccharide-based, ica-dependent biofilms, biosynthesized by the icaADBC gene cluster, or as protein-based, ica-independent biofilms.¹⁵ The ica-dependent biofilms, also known as polysaccharide intercellular adhesin (PIA), are composed of poly-N-acetylglucosamine (PNAG). In contrast, a major feature of ica-independent biofilms is the biofilm accessory protein (Bap), consisting of 2000 amino acids.¹⁶ The ica gene cluster is regulated by IcaR, a tetracycline repressor (TetR) family transcriptional repressor,¹⁷ whose expression is known to be regulated by genes encoding global regulators, including agr, sae, sar, and the alternative sigma factor, sigB; however, the IcaR-sensing inducer molecule, recently proposed to be autoinducing peptide II (AIP-II), has not been clearly established.^16,18,19

More recently, the presence of glucose in the growth medium was shown to be an important factor for the elaboration of PNAG/PIA biofilms in clinically isolated superbiofilm-elaborating S. aureus strains, under the control of the glucose-induced biofilm accessory gene A or GbaA (also known as Rob, regulator of biofilms).^20,21 Like IcaR,¹⁷ GbaA is a member of the TetR family transcriptional repressor family,^22,23 which collectively comprises a very large protein superfamily characterized by a highly conserved N-terminal three-helix motif (α1–α3) that recognizes a palindromic DNA operator sequence and a C-terminal all-helical domain (α4–α8 or α9) that has evolved to sense a specific small molecule “inducer” thus controlling the expression of specific genes in response to cellular concentrations of that inducer (Figure S1).²³ Specific inducers range from inorganic ions such as Zn(II)^24,25 to a variety of small organic molecules,²³ including the namesake of the superfamily, tetracycline.²² GbaA is predicted to negatively regulate biofilm formation in S. aureus by repressing GbaA-regulated genes, consistent with the finding that a clinical isolate, S. aureus TF2758, harbors a translationally terminated allele of gbaA and is a robust biofilm former.^20,21 A model for GbaA function hypothesizes that GbaA binds to a TATTT motif in the icaR–icaA intergenic region and represses the transcription of ica genes. The physiological inducer of GbaA remains an open question that was not previously addressed.^20,21

GbaA (SAUSA300_2515) in S. aureus USA300 FPR3757 is reported to regulate the expression of a divergently transcribed operon that contains seven genes (SAUSA300_2512–SAUSA300_2518), including GbaA, by binding to a consensus TetR operator upstream of the SAUSA300_2515 gene that physically overlaps the promoter in a 131 bp intergenic region (Figure 1A,B). The biological activities of the proteins encoded by these genes are largely unknown, with the possible exception of GbaB (SAUSA300_2516), predicted to encode a short chain oxidoreductase that oxidizes a secondary alcohol to an aldehyde in a carbohydrate substrate and thus could be involved in PNAG biosynthesis.²⁰ Investigation of the primary structure of GbaA reveals that it harbors two cysteine (Cys) residues that are conserved in all Staphylococcus spp., C55 and C104 (Figure 1C). This was of interest to us because previous work showed that GbaA-regulated genes were upregulated in an early study of nitrite-induced dispersal of the S. aureus N315 biofilms mediated by downstream reactive nitrogen species (RNS) as determined by a microarray analysis (Figure S2A), results corroborated here for the USA300 FPR3757 strain (Figure S2B–D).²⁶ This same study revealed strong induction of the cst operon via the thiol persulfide [reactive sulfur species (RSS)]-sensing repressor CstR characterized later,²⁷ which we show here also impacts biofilm dynamics (Figure S2D), as well as genes encoding various transition metal (Zn and Cu) efflux and iron storage proteins (Figure S2A).²⁶ Given the emerging physiological crosstalk between RSS and reactive nitrogen species (RNS) thought to involve the one-electron-reduced form of nitric oxide, nitroxyl,^28,29 which is directly sensed by CstR through dithiol chemistry,^30,31 this suggested to us that GbaA might employ one or both of its conserved cysteines to impact expression of GbaA-regulated genes. This hypothesis was further motivated by the finding that GbaA is persulfidated in S. aureus strain Newman under conditions of sodium sulfide/RSS misregulation.³²

Figure 1.

Divergently transcribed GbaA-regulated operon as defined in previous work^20,21 and a structural model of S. aureus FPR3757 GbaA. (A) Divergently transcribed operon encoding a glyoxalase-like protein (SAUSA300_2512), NmrA (SAUSA300_2513), duf2316 (SAUSA300_2514), GbaA (SAUSA300_2515), a short chain oxidoreductase GbaB (SAUSA300_2516), an amidohydrolase-like protein (SAUSA300_2517), and a protein that belongs to the α,β-hydrolase superfamily (SAUSA300_2518), all regulated by GbaA. (B) Putative −10 and −35 promoter elements for the two σ^A promoters within the intergenic region, with the palindromic operator bound by GbaA represented by the two facing arrows. The initiation codons of the duf 2316 and gbaA genes are colored red. (C) Both C55 and C104 are conserved within many Staphylococcus spp. (D) Structural model of the GbaA homodimer (top) obtained by homology modeling using Modeler⁷⁶ and 2REK, which possesses 28% similarity over 97% query coverage, as the search template. The S^γ atoms of the two conserved Cys residues in each protomer are quite close to one other in this model (≈6.5 Å), potentially sufficiently close to form a disulfide bond.

In this work, we show that S. aureus GbaA is a stable homodimer that senses electrophiles N-ethylmaleimide (NEM) and methylglyoxal (MG) in bacterial cells and in vitro reacts with NEM to form thioether adducts on each Cys. Although GbaA thiolates also react with other oxidants in vitro, including bacillithiol and glutathione disulfides (BSSB and GSSG), S-nitroso glutathione (GSNO), and glutathione persulfide (GSSH), the latter of which accumulates in cells under conditions of sulfide toxicity,^30,31 only NEM and MG are significant inducers of GbaA-regulated gene expression in vivo. We further show that these and other common oxidants result in a disulfide bond linking C55 and C104 of the same GbaA protomer, revealing that C55 and C104 are physically close to one another. Strikingly, formation of a disulfide bond is not regulatory (inhibitory) for DNA binding, nor does it significantly change the global conformation of the homodimer. In contrast, characterization of the component single Cys mutants of GbaA (C55A and C104A) incubated with the oxidants BSSB and GSSG/GSSH gives rise to the expected mixed di- and polysulfide species, which are inhibitory to DNA binding. These findings reveal that monothiol chemistry rather than dithiol chemistry governs GbaA DNA binding activity and regulatory function in cells. These results are fully consistent with expectations of a monothiol electrophile sensor^33,34 and the finding that single Cys mutants of GbaA are active as NEM sensors in cells, while the double-Cys substitution mutant is not. The functional implications of these findings and of S. aureus GbaA in the context of the bona fide NEM sensor NemR from Escherichia coli and Acinetobacter nosocomialis^35–37 are discussed.

MATERIALS AND METHODS

S. aureus USA300 FPR3757 Bacterial Strains.

The FPR3757 SAUSA300_2515∷Tn(Erm) (gbaA) strain was constructed using insertions in the Nebraska Transposon Mutant Library (see Acknowledgments) (Table S1). The gbaA∷Tn(Erm) insertion was transduced into S. aureus FPR3757 with selection on erythromycin. Other S. aureus FPR3757 strains harboring transposon insertion in other genes, including the SAUSA300_2516∷Tn(Erm) (gbaB), SAUSA300_2513∷Tn(Erm), SAUSA300_0083∷Tn(Erm) (t a u E), S A U S A 3 0 0 _ 0 0 8 4 ∷ T n (E r m) (c s t R), SAUSA300_0085∷Tn(Erm) (cstA), and SAUSA300_0086∷Tn(Erm) (cstB), were prepared in the same way as the gbaA strain (Table S1).

S. aureus FPR3757 Growth Curves.

Ten milliliters of TSB broth prewarmed at 37 °C was inoculated with the appropriate S. aureus strain from a frozen glycerol cell stock and grown overnight (~16 h). The cells were diluted to an OD₆₀₀ of 0.01 in freshly prepared, chemically defined Hussain–Hastings–White modified medium (HHWm) with 0.2 mM cystine used as the sole sulfur source.²⁷ Cells were grown to an OD₆₀₀ of 0.2 and diluted again into a fresh aliquot of HHWm and cystine with the appropriate stressor added at mid log phase when the cell density reaches an OD₆₀₀ of ~0.2–0.3. The cell density was monitored every hour by measuring the OD₆₀₀. All cultures were grown at 37 °C while being shaken in loosely capped 50 mL Falcon tube unless otherwise specified. A Spectronic 20 Genesys spectrophotometer (ThermoScientific) was used to record OD₆₀₀.

RNA Extraction and Quantitative Real Time Polymer-ase Chain Reaction (qRT-PCR).

Fifteen milliliters of recovered cultures of S. aureus strains was grown in HHWm +TS as described previously to an OD₆₀₀ of ≈0.2, at which time 5 mL of cell culture was washed with phosphate saline buffer (PBS), centrifuged, and stored at −80 °C until further use. Pellets were thawed and resuspended in 1 mL of TriReagent (catalog no. TR-118, Molecular Research Center) and then transferred to tubes containing 0.1 mm silica beads (Lysing matrix B, catalog no. 6911–100, MP Biomedicals) and lysed on a bead ruptor homogenizer; 200 μL of chloroform was added followed by mixing adequately by pipetting and centrifuging at 16000g for 10–15 min. The aqueous layer was extracted using 70% ethanol. RNA purification was then completed using a Qiagen RNeasy minikit. A commercial DNase (Invitrogen) was used to digest all of the DNA according to the manufacturer’s instructions. cDNA was synthesized from the mRNA using random hexamers and commercially available Flex cDNA synthesis kit (Quanta Biosciences). Reactions without the reverse transcriptase were performed as control experiments to check for possible DNA contamination.

qRT-PCRs designed to quantify levels of the duf 2316 (SAUSA300_2514) and gbaB (SAUSA300_2516) transcripts included 6 μL of diluted cDNA, 2 μL each of 2 μM QPCR primers (see Table S1 for sequences), 10 μL of 2× Brilliant III Ultra-Fast SYBR Green QPCR Master Mix (catalog no. 600882, Agilent), and 0.3 μL of 2 μM ROX reference dye. The relative levels of the mRNA transcript were measured using the MX3000P thermocycler (Stratagene) running SYBR Green with the dissociation curve program and normalized to the amount of 16S rRNA. The thermal profile included one cycle at 95 °C for 3 min and 40 cycles at 95 °C for 20 s and 59 °C for 20 s. Subsequently, a dissociation curve starting at 55 °C going to 95 °C in 0.5 °C increments with a dwell time of 30 s was performed to assess the specificity of the reactions. Three biologically independent samples were measured for each treatment, and the mean [±standard deviation (SD)] values are reported. Transcript amounts were compared by a ΔΔCt method, by normalizing the threshold Ct against the Ct of 16s rRNA followed by the normalization against the WT, no stress. Fold change is defined as 2^−ΔΔCt. Errors in Ct were propagated by multiplying the power (ΔΔCt) to the error percent, and statistical significance was established using the Student’s unpaired t-test for with statistics reported as n.s. (not significant), p > 0.05, *p ≤ 0.05, **p ≤ 0.01, and *****p ≤ 0.00001.

Biofilm Assays.

Strains were streaked to TSB agar plates, and individual colonies were inoculated into 5 mL of TSB. Cultures were incubated overnight at 37 °C while being shaken in biofilm medium containing TSB, 1% glucose, and 3% NaCl (TSBg). Then 4 μL samples were used to inoculate 196 μL of biofilm medium in 96-well flat-bottom plates pretreated with bovine plasma (biofilms) or untreated (planktonic). The cultures were then incubated at 37 °C without being shaken. Biofilms were then washed twice with PBS by gentle pipetting and aspiration, permeabilized, fixed with 200 μL of ethanol, dried, and stained with crystal violet. These assays were carried out in biological quadruplicate and technical triplicate.

GbaA Overexpression Plasmid Construction.

Genomic DNA from S. aureus strain Newman⁴ and appropriate primers (Table S1) were used to PCR amplify the coding sequence of NWMN_2477, which is identical to the SAUSA300_2515 gene. Following PCR amplification, the gbaA gene was cloned into the pHis parallel E. coli overexpression plasmid³⁸ harboring an N-terminal His₆ tag, followed by a TEV protease recognition site using Gibson isothermal assembly.³⁹ The C55A and C104A mutants were constructed using PCR-based site-directed mutagenesis.

Expression and Purification of GbaA and Mutant GbaA Proteins.

Plasmid DNA was transformed by a 45 s heat shock at 42 °C in E. coli Rosetta (DE3)/pLysS (Novagen) and selected on LB agar plates with 100 μg/mL ampicillin and 24 μg/mL chloramphenicol. A few colonies were scraped using sterilized inoculation tips, and each was used to inoculate 15 mL of LB broth containing only 100 μg/mL ampicillin and grown at 37 °C while being vigorously shaken. When the OD₆₀₀ reached ≈0.6, cells were diluted into 1 L of LB broth containing 100 μg/mL ampicillin and grown at 37 °C while being shaken until the OD₆₀₀ again reached ≈0.6. Isopropyl β-thiogalactopyranoside (IPTG) was then added to a final concentration of 0.5 mM to induce the expression of GbaA, and cells were grown at 16 °C while being shaken overnight. Cells were then collected the next day by centrifugation (8000g for 15 min) and suspended in lysis buffer [25 mM Tris-HCl (pH 8.0), 500 mM NaCl, 0.2–0.5 mM tris(2-carboxyethyl)-phosphine hydrochloride (TCEP), and 1 mM phenylmethanesulfonyl fluoride (PMSF)]. The cell wall was disrupted by sonication on ice (1/2 in. probe at 60% power, 300 cycles, with a 1 s pulse every 6 s), and the cell debris removed by centrifugation (10000g for 15–20 min). GbaA was precipitated by saturating the solution with 70% (w/v) ammonium sulfate followed by centrifugation at 4 °C. GbaA was then resuspended with buffer A [25 mM Tris (pH 8.0), 500 mM NaCl, and 0.2–0.5 mM TCEP] for the initial purification steps. The solution was centrifuged at 4 °C, passed through a 0.45 μm Millex GV Durapore PVDF membrane, and then purified using two 5 mL HisTrap HP Ni-NTA columns connected serially on an ÄKTA pure protein purification system (GE Healthcare Life Sciences). Using a gradient elution protocol with buffer B [25 mM Tris (pH 8.0), 500 mM NaCl, 500 mM imidazole, and 0.2–0.5 mM TCEP], the Ni-bound protein was eluted and dialyzed with 0.5 mM EDTA and TEV protease over buffer A overnight. Cleaved GbaA lacking the His₆ tag was then isolated by being passed through the Ni-NTA column as described, followed by another round of purification using a HiLoad 16/60 Superdex 200 prep grade gel-filtration column using buffer A. Fractions (2 mL) were collected and pooled upon inspection of the ultraviolet–visible chromatogram and sodium dodecyl sulfate–polyacrylamide gel electrophoresis gel analysis. GbaA purified in this way was aliquoted into 1 mL stocks with 5% or 10% glycerol and stored in −80 °C until use, with the protein (protomer) concentration determined using an ε₂₈₀ of 20860 M⁻¹ cm⁻¹. The number of reduced Cys residues was quantified using Ellman’s reagent, as previously described,⁴⁰ with the value ranging from 1.7 to 1.8 mol of Cys/mol of protomer (2.0 expected) for individual GbaA preparations. The C55A and C104A GbaA mutants were purified and characterized in exactly the same way where the thiol concentration was 0.9 mol per monomer (expected value of 1.0).

Reactions of GbaA with Glutathione Disulfide (GSSG), S-Nitrosylated Glutathione (GSNO), and Bacillithiol Disulfide (BSSB).

A stock aliquot of reduced GbaA was buffer exchanged with buffer A without TCEP inside a Vacuum Atmospheres glovebox (<10 ppm O₂) using a 10 kDa cutoff Amicon ultra 0.5 mL centrifuge unit. End-point reactions were carried out inside the glovebox at ambient temperature by adding a 10–20-fold excess of either oxidant (GSSG or BSSB) or GSNO (prepared as previously described⁴¹) to 100 μM GbaA (protomer) and allowing the reaction to proceed overnight (15–17 h). The reaction products were analyzed by LC-ESI-MS or MALDI-MS as described below.

In Situ Glutathione Persulfide (GSSH) Generation and GbaA Reactions with GSSH.

A 50 mM stock solution of Na₂S (Sigma-Aldrich, CAS Registry No. 1313-82-2) was freshly prepared in 300 mM phosphate (pH 7.4) under an argon atmosphere (<10 ppm O₂) in a glovebox just prior to use. Fifty microliters of this Na₂S solution was added to 50 μL of a freshly prepared 10 mM GSSG solution (5:1 S²⁻:GSSG) in an Eppendorf tube. The tube was tightly capped and removed from the argon chamber, and the contents were allowed to react for 30 min in a 30 °C bath. After the reaction, the concentration of GSSH was determined by a cold cyanolysis assay as previously described.^27,30,42 This solution was used directly for the GbaA derivatization reactions at the indicated concentration of GSSH.

LC-ESI-MS Analysis of Intact GbaA Species.

LC-ESI-MS analysis of intact GbaA preparations was performed in the Indiana University Mass Spectrometry Facility using a Waters/Micromass LCT Classic time-of-flight (TOF) mass spectrometer with a CapLC inlet. GbaA was loaded onto a 50 mm Agilent BioBasic C8 reverse phase column (5 mm particle size, 300 Å pore size) in solvent A (5% acetonitrile, 95% water, and 0.1% formic acid) and eluted with a 20 min linear gradient from 10% solvent A to 90% solvent B (95% acetonitrile, 5% water, and 0.1% formic acid). Data were collected and analyzed using MassLynx Software (Waters, Milford, MA).

Tryptic Digest and MALDI-MS Analysis.

One hundred microliters of a chemically derivatized GbaA sample was incubated with 100 μL of 8 M urea and 0.2 M iodoacetamide (IAM) in an Eppendorf tube inside the glovebox at specific time points for ~30 min. The Eppendorf tubes were removed from the glovebox and placed on ice; the contents were precipitated with 25% (v/v) trichloroacetic acid (TCA) and centrifuged (rotor speed, time), and the resulting pellet was washed twice with acetone. The protein was then dried using a SAVANT ISS110 Speed-Vac concentrator and redissolved in 20 μL of 100 mM ammonium bicarbonate and 2 M urea. Trypsin was added, and the digest was allowed to proceed at 37 °C overnight. The products were absorbed to a C18 zip-tip, eluted with an 100% acetonitrile/0.1% TFA wash with 1 μL of the digest spotted onto a CCA (α-cyano-4-hydroxycinnamic acid) matrix, and injected on a Bruker MALDI-MS instrument at the Indiana University Mass Spectrometry Facility. The resulting data were analyzed using Autoflex analysis software.

Ratiometric Pulsed Alkylation Mass Spectrometry (rPA-MS) of GbaA.

These reactions were carried out essentially as described previously.⁴³ Briefly, the protein was removed from −80 °C storage, thawed over ice, and introduced into the glovebox (<10 ppm O₂). The protein solution was buffer exchanged into reaction buffer [25 mM Tris-HCl and 500 mM NaCl (pH 7.5)] by using an Amicon Ultra 0.5 mL filter with a 10 kDa molecular weight (MW) cutoff (UFC5010), and the concentration determined by absorption at 280 nm (ε = 20860 M⁻¹ cm⁻¹). Reaction mixtures (600 μL) contained 50 μM GbaA (dimer) and 500 μM d₅-NEM (Cambridge Isotope Laboratories, catalog no. 36078-37-2) at time zero, and 50 μL was withdrawn at pulse times (t) of 30 s and 1, 2.5, 5, 10, 20, 40, 60, 80, and 100 min, added to 50 μL of a quenching solution [8 M urea and 10 mM H₅-NEM (Sigma-Aldrich)], mixed thoroughly, incubated for 30 min, then precipitated with 25% (v/v) TCA, and prepared for tryptic digestion and MALDI analysis as described previously.⁴³ To obtain the species fraction as a function of pulse time, t, the intensities for all isotopic peaks corresponding to the H₅-NEM adduct (m/z 1384.7, 1385.7, 1386.7, and 1387.7) and the d₅-NEM adduct (m/z 1389.7, 1390.7, 1391.7, and 1392.7) were summed to represent A(H₅-NEM) and A(d₅-NEM), respectively. The mole fraction (Φ) of each of the two singly derivatized peptides in a mixture is calculated as the ratio of the peak area integration (A) of the ith derivatized species to the sum of the integrated areas of all A(H₅+d₅) alkylated species.

$${\Phi }_{{\text{H}}_5-\text{NEM}}=\frac{\text{A}\left({\text{H}}_5-\text{NEM}\right)}{\text{A}\left({\text{H}}_5-\text{NEM}\right)+\text{A}\left(d_5-\text{NEM}\right)}$$

$${\Phi }_{d_5-\text{NEM}}=\frac{\text{A}\left(d_5-\text{NEM}\right)}{\text{A}\left({\text{H}}_5-\text{NEM}\right)+\text{A}\left(d_5-\text{NEM}\right)}$$

${\Phi }_{{\text{H}}_5}$ and ${\Phi }_{d_s}$ for each time point were then obtained from three independent experiments, and the mean value and standard error of the mean plotted as a function of reaction time (t). The data points were then fitted to a single-exponential rate equation to k as previously described⁴³ with ${\Phi }_{{\text{H}}_5}$ and ${\Phi }_{d_s}$ at time zero constrained to 1.0 and 0, respectively.

Fluorescence Anisotropy-Based DNA Binding Titrations.

The duplex DNA operator was obtained by heating 1 μM fluorescein-labeled and 1.05 μM (excess) unlabeled complementary ssDNAs (Table S1) in an opaque microcentrifuge tube at 95 °C on a heating block, followed by slow cooling at room temperature; this annealed dsDNA was used without further purification. Ten nanomolar labeled OP1 duplex DNA was added to 3 mL of buffer A [25 mM Tris (pH 8.0), 500 mM NaCl, and 0.2–0.5 mM TCEP]. Experiments was carried out at 25.0 °C in a thermostated sample holder using 495 nm as the excitation wavelength (slit width of 1 mm) and a 515 nm cutoff filter using an ISS (Champaign, IL) PC1 spectrofluorometer operating in the L-format. A known concentration of GbaA was loaded on a 100 μL HAMILTON gastight syringe and titrated into the cuvette using a kd Scientific automated syringe pump. The observed anisotropy (r_obs) versus time t was recorded and converted into r_obs versus [GbaA]. The r_obs values were normalized for the fractional saturation (Θ) of OP1 calculated from the equation Θ = (r_obs − r₀)/(r_complex − r₀) from 0 to 1, where r_complex represents the maximum anisotropy obtained for each titration and r₀ is the initial anisotropy of the OP1 DNA. The data were fit to a simple nondissociable 1:1 GbaA dimer binding model using Dynafit.⁴⁴ Electrophile-mediated dissociation of preformed GbaA–DNA complexes (10 nM DNA, 100–150 nM protein, and 1 mM electrophile) were performed by sequentially adding protein and then electrophile to free DNA while monitoring the anisotropy of the DNA over time in buffer A.

Small-Angle X-ray Scattering (SAXS) Data Collection and Envelope Reconstruction.

GbaA samples were thawed from −80 °C, and samples prepared for SAXS analysis. Reduced GbaA was used as is. For the preparation of disulfide cross-linked (oxidized) GbaA, 10 protomer molar equivalents of GSSG was incubated with 1 equiv of GbaA in 25 mM Tris-HCl and 500 mM NaCl and samples were diluted to final concentrations of 5, 10, and 15 mg/mL in the same buffer, with 0.2 mM TCEP present only for the reduced GbaA sample. Data were acquired for these samples by the Small-Angle X-ray Scattering Core Facility, Frederick National Laboratory for Cancer Research, and the resulting raw data subjected to buffer subtraction and data merging using EMBLATSAS. The intensities were normalized by their I₀ values, extrapolated from the Guinier plot. The radius of gyration (R_g) was also calculated from the Guinier plot and reported. The pairwise distance distribution function (PDDF) was constructed for each sample and used as the input for DAMMIF, which was run in the “slow mode” for 10 steps with P₂ symmetry for initial three-dimensional envelope structure reconstruction. The best envelope output from DAMMIF was collected and used as an input for DAMMIN for further refinement for 20 steps. The envelope that best fit the model is shown here.

RESULTS

GbaA Responds to Electrophilic Stressors in Cells.

To identify a potential physiological inducer of GbaA-regulated genes in cells, we stressed S. aureus MRSA strain USA300 FPR3757 with various redox and electrophilic stressors at early mid log phase of growth [OD₆₀₀ ≈ 0.2 (Figure S3)]. The addition of 5 mM hydrogen peroxide (H₂O₂) and 1.5 mM chloramine-T (HOCl) was strongly growth inhibitory, as was the addition of 0.1 mM N-ethylmaleimide (NEM) (vide infra). In contrast, the addition of 0.1 mM methylglyoxal (MG), 0.2 mM disodium sulfide (Na₂S), 5 mM sodium nitrite (NaNO₂), 5 μg/mL Ag nanoparticles (Ag Nano), and 0.2 mM linoleic acid (LA) gave rise to little or no growth phenotype under these conditions (Figure S3).

To investigate the effect of these stressors on the transcription of GbaA-regulated genes, we collected cells at 15 min following the addition of the stressor and extracted mRNA for reverse transcription-based qRT-PCR analysis of duf 2316 (SAUSA300_2514) and gbaB (SAUSA300_2516) that are known to be regulated by GbaA (Figure 2A,B).⁴⁵ Our qRT-PCR analysis reveals that while NEM and MG strongly induce the expression of both GbaA-regulated genes, no other stressor results in a statistically significant effect on the transcription, with the possible exception of sodium nitrite, which only slightly induces the transcription of duf 2316 (Figure 2A).²⁶ N-Ethylmaleimide and methylglyoxal are both potent electrophiles and are capable of inducing GbaA-regulated genes, in striking contrast to oxidative stresses such as chloramine-T and hydrogen peroxide. NEM is a significant inhibitor of growth at concentrations of >50 μM (Figure 2C), while MG gives rise to a significant growth phenotype at concentrations of >250 μM (Figure S4). As GbaA harbors two nucleophilic Cys residues (Figure 1C), we hypothesized that one or both might react with the electrophile to form a thioether adduct that ultimately drives dissociation of GbaA from the DNA operator–promoter region and transcriptional derepression.

Figure 2.

Transcriptional response of GbaA in S. aureus FPR3757 toward oxidative and electrophilic stressors. Quantitative real time PCR (qRT-PCR) data showing the relative fold change of the mRNA transcripts for genes regulated by GbaA, namely, (A) duf 2316 and (B) gbaB, with respect to WT unstressed cells, under the following conditions: 0.05 mM N-ethylmaleimide (NEM), 0.1 mM methylglyoxal (MG), 0.2 mM disodium sulfide (Na₂S), 5 mM sodium nitrite (NO₂⁻), 5 μg/mL Ag nanoparticles (Ag NP), 0.2 mM Angeli’s salt, 5 mM hydrogen peroxide (H₂O₂), and 0.2 mM linoleic acid (LA). The unstressed gbaA strain is shown for reference. (C) Dose-dependent growth curves of WT S. aureus USA300 FPR3757 exposed to the indicated concentrations of NEM, added to mid log cultures (OD₆₀₀ ≈ 0.2). (D) Relative expression of duf 2316 with 0.05 mM NEM stress added to the gbaA strain complemented with the indicated gbaA allele constitutively expressed on a pOS1 plasmid vs the WT strain (n.s., not significant; *p ≤ 0.05; **p ≤ 0.01; *****p ≤ 0.00001).

To test this and to determine if one Cys or both Cys residues in GbaA were required to sense NEM and MG in cells, we complemented a gbaA mutant with constitutively expressed gbaA alleles corresponding to the wild type (WT), an allele containing only C55 in the DNA binding domain (C104A), one containing C104 only (C55A), or one containing neither Cys (C55A/C104A) and stressed these cells with 0.05 mM NEM (Figure 2D). The WT GbaA-complemented gbaA strain gave rise to wild-type-like induction of duf 2316, revealing the pOS1 plasmid-expressed GbaA is functional in this assay. Strikingly, both single-Cys GbaA mutants (C55A and C104A) are functional, as well, to an extent similar to that of the wild-type-complemented strain, while the corresponding double mutant (C55A/C104A GbaA) could not be induced by NEM. These data taken collectively reveal that the presence of one Cys or the other endows GbaA with NEM-sensing activity in vivo. We also tested if these gbaA strains complemented with distinct gbaA alleles result in a growth phenotype when stressed with either 0.05 mM NEM or 0.25 mM MG (Figure S5). We find the growth of these strains is indistinguishable from that of the wild type, even a strain expressing the double GbaA mutant (C55A/C104A GbaA) that cannot be induced by NEM (Figure 2D). This suggests that GbaA-regulated genes are not required to resist the effects of NEM- or MG-mediated growth inhibition, suggesting that neither is a physiologically relevant electrophile sensed by GbaA; this is in contrast to NemR-regulated genes in E. coli and A. nosocomialis.^35–37

GbaA Forms Intraprotomer Disulfide Bonds upon Incubation with Various Oxidants.

We were next interested in understanding the origin of the apparent specificity of GbaA for reactive electrophile stress (RES) sensing in cells, using purified GbaA (Figure S6). This is important because although the expression experiments reveal that one Cys or the other is required for GbaA function, they provide no insight into the breadth of oxidative thiol chemistry that is available to GbaA. Indeed, many bacterial regulatory proteins use reversible dithiol–disulfide or dithiol–tetrasulfide bond formation to regulate DNA binding,^27,46 which is of particular interest here because the two cysteines in our GbaA model appear close enough to form a disulfide bond (Figure 1D). The prediction from our characterization of the single-Cys gbaA mutant strains, however, is that even if a disulfide is capable of forming in GbaA, it is unlikely to be regulatory.

To test this, we carried out anaerobic, end-point reactions of GbaA with a 20-fold molar protomer excess of diamide, glutathione disulfide, S-nitrosoglutathione, or bacillithiol disulfide, and with CuSO₄ in air, and compared the reaction products to that of reduced (unreacted) GbaA. These reactions were terminated by an excess of iodoacetamide (IAM) to cap any free remaining thiol and subjected to trypsin digestion and MALDI-TOF mass spectrometry to identify the peptide fragments (Figure 3 and Table 1). We find that all oxidants result in the disappearance of two peptide fragments containing the IAM-capped C55 at mass values of 1316.6 and 1559.7 Da, while two corresponding C55–C104 disulfide-cross-linked peptides at mass values of 2531.2 and 2774.6 Da were observed in all oxidant samples, albeit to widely varying degrees (Figure 3 and Table 1). We note that the highest relative abundance of mixed disulfide was obtained with bacillithiol disulfide (BSSB), which is the physiological low-molecular weight thiol in S. aureus³⁰ and other Gram-positive bacteria.⁴⁷ Although the reduced and capped peptide containing C104 was not detected in these mass spectra, likely due to low peptide ion counts, the formation of a C55–C104 disulfide bond in the diamide-cross-linked sample as representative of the others was confirmed by ESI-MS/MS (Figure S7). Although these data cannot distinguish an intraprotomer (C55–C104) from an interprotomer (C55–C104′) cross-link, the near absence of a dimer peak in ESI mass spectra of the intact, disulfide cross-linked GbaA strongly argues for an intraprotomer linkage (Figure S8; vide infra).

Figure 3.

MALDI-MS analysis of a tryptic digest of reduced GbaA subjected to an anaerobic end-point reaction with various oxidative stressors, including diamide, glutathione disulfide, bacillithiol disulfide, and S-nitrosoglutathione, as well as CuSO₄ in air, as indicated (see Table 1 for observed and expected masses). Two spectral windows (mass ranges) are shown on the left and right, with the peptides of interest labeled.

Table 1.

MALDI-MS Masses for Cys-Containing Peptides Derived from GbaA Trypsin Digests

peptide fragment	expected MW (Da)	observed MW (Da)
A, (K)SDLC(CAM)YYVIQR(D)	1316.63	1316.60
B, (K)DKSDLC(CAM) YYVIQR(D)	1559.75	1559.71
C, (K)ALLQC(CAM) IEAGNNKLR(F)	1599.75	1600.21
D, (K)ALLQC(CAM) IEAGNNK(L)	1330.67	not detected
peptide A and peptide D less 2 × CAM (58.06 Da)	2531.2	2531.18
peptide B and peptide D less 2 × CAM (58.06 Da)	2774.3	2773.88, 2774.48, 2774.61, 2774.73

GbaA Adopts a Homodimeric Assembly State in Solution.

Regulatory disulfide bond formation in proteins is often accompanied by a change in tertiary or quaternary structure that impacts its hydrodynamic properties.⁴⁶ To investigate this, we carried out a global conformational analysis of the reduced and disulfide-oxidized forms of GbaA using small-angle X-ray scattering (SAXS) (Figure 4). The raw scattering profiles and pairwise distance distribution function histograms are very similar in both states (Figure 4A,B and Table S2). These data are consistent with the Guinier plots and corresponding Kratky plots (Figure 4C,D), which reveal well-folded conformations without significant aggregation in each case, and statistically indistinguishable R_g values [25.7 Å (Table S2)]. These data show that the native oligomeric state of GbaA is a homodimer (Table S2), consistent with most other members of the TetR family of repressors,²³ but with minimal to no change in structure upon intraprotomer disulfide bond formation (Figure 4 and Table S2). The reconstructed envelope for reduced GbaA obtained from the scattering data fits the GbaA homodimer model (see Figure 1D) quite well (Figure 4E).

Figure 4.

Small-angle X-ray scattering (SAXS) experiments carried out with the reduced and diamide-oxidized GbaA homodimer. (A) Raw scattering profiles of the reduced (red) and oxidized (blue) GbaA samples. (B) Unnormalized pairwise distance distribution function histograms (PDDF) of reduced (red) and oxidized (blue) GbaAs. (C) Guinier plots of reduced GbaA with its corresponding Kratky plot (inset). (D) Guinier plot of oxidized GbaA with its corresponding Kratky plot (inset). (E) SAXS envelope reconstructed from the experimentally observed scattering profile aligned with the predictive model (shown as a ribbon) for GbaA as shown from the “front” and two perpendicular rotation axes. The hydrodynamic parameters derived from these experiments for reduced and oxidized GbaA are listed in Table S2.

The Cysteines in GbaA Are Also Capable of Forming Adducts with GSSG, GSSH, BSSB, and NEM.

To further explore the oxidative chemistry of the cysteines in GbaA, we carried out anaerobic, end-point assays of wild-type GbaA and single-cysteine mutants C55A GbaA and C104A GbaA, with a 20-fold molar protomer excess of a model oxidant GSSG and its corresponding hydrodisulfide (persulfide) donor, GSSH,²⁷ along with BSSB, the physiological oxidant in S. aureus, and analyzed the products by ESI-MS. With GSSG, we observe only the anticipated −2 Da mass shift consistent with the loss of two hydrogen atoms in WT GbaA as a result of C55–C104 disulfide bond formation (Figure 5A). A similar result was obtained upon incubation with BSSB and GSSH (data not shown). In contrast, incubation of GSSG (Figure 5A), BSSB (Figure 5B), or GSSH (Figure 5C) with C55A or C104A GbaA results in the formation of the mixed disulfide with glutathione at 22460 Da corresponding to [M + 306] (Figure 5A) and mixed disulfide with bacillithiol at 22551 Da corresponding to [M + 397] (Figure 5B). For both mutants, we observe an additional peak at 22492 Da [M + ³²S + 306] obtained upon incubation with GSSH, which we assign as corresponding to a mixed trisulfide linkage with glutathione (Figure 5C). Because there is no resolving cysteine in C55A or C104A GbaA, a disulfide bond does not form, and any adduct formed with GSSG or GSSH becomes the final product of the reaction. In addition, we observe what appears to be full conversion of C55A GbaA to mixed di- and trisulfide products with both oxidants GSSG and BSSB, in striking contrast to C104A GbaA, which shows a far lower yield of S-thiolated products. This reveals that C104 in the regulatory domain of GbaA is more reactive than C55 in the DNA binding domain (Figure S1) toward these weakly reactive electrophiles. This finding is in contrast to reactions carried out with the more reactive electrophile and cellular inducer, NEM, which show that both cysteines in wild-type GbaA form RS-NEM adducts in a 10 min reaction (Figure S9 and Table S3). From these experiments, we conclude that both C55 and C104 in free GbaA are likely exposed to solvent and react with a number of small molecules, and that C104 is intrinsically more reactive toward weak electrophiles; they also provide evidence for the mixed thiol disulfide on C104 as an on-pathway intermediate for C55–C104 disulfide bond formation, but the formation of the disulfide appears quite slow.

Figure 5.

ESI-MS spectra that result from an anaerobic, end-point reaction of GbaA and single-cysteine mutants C55A GbaA and C104A GbaA with a 20-fold molar excess of (A) GSSG, (B) BSSB, and (C) GSSH. Conditions: 25 mM Tris-HCl, pH 7.5, 0.5 M NaCl, ambient temperature.

Ratiometric Pulsed Chase Alkylation Mass Spectrometry with NEM.

Because each cysteine in GbaA rapidly forms an NEM adduct (Figure S9 and Table S3), we wanted to measure the kinetics of NEM alkylation of these cysteines in wild-type GbaA. Unfortunately, the peptide containing C104 could not be observed due to the low ionization efficiency, so we focused on C55. We used a ratiometric pulsed chase alkylation mass spectrometry (rPA-MS) experiment previously developed, which allows quantitation of the rates of chemical reactivity of individual Cys residues in a protein.^43,48 Here, wild-type GbaA is first pulsed with a 10-fold molar excess of “heavy” deuterium-labeled NEM (d₅-NEM) for variable pulse times, t, and chased with a 1000-fold molar excess of unlabeled “light” NEM (H₅-NEM), with the reaction mixtures subjected to trypsin digestion and peak areas quantified by MALDI-MS (Figure 6A). A 5 Da mass shift between the two NEM variants allows us to capture the extent of reaction at different pulse time points (Figure 6A). We observe a very fast reaction, which gives rise to a rate constant k of 0.98 ± 0.09 min⁻¹ (Figure 6B). We conclude that C55 in the DNA binding domain is solvent-exposed and undergoes electrophilic addition to completion within 10–20 min of reaction.

Figure 6.

Ratiometric pulsed chase alkylation mass spectrometry (rPA-MS) of wild-type GbaA (see Materials and Methods for details). (A) Waterfall representation of the normalized peak intensities for H₅-NEM and the d₅-NEM adducts of the C55-containing peptide (K)SDLC⁵⁵YYVIQR(D). (B) Kinetic plot of the species fractions obtained as a function of pulse time, t, from triplicate experiments. The solid line represents a global fit to both data sets revealing a NEM alkylation rate k of 0.98 ± 0.09 min⁻¹. Conditions: 25 mM Tris-HCl, pH 7.5, 0.5 M NaCl.

Regulation of DNA Operator Binding by Thiol Modifications.

To elucidate the degree to which these various oxidative modifications of GbaA regulate DNA binding, we measured the DNA binding affinities of the wild-type and single-cysteine variants of GbaA in their reduced, disulfide-cross-linked, mixed disulfide, or NEM-alkylated states using a DNA fragment harboring the GbaA operator (GbaO) previously reported (Figure 1B).⁴⁵ We find that the reduced wild-type, C55A, and C104A GbaAs bind to the GbaO with nearly identical affinities, within a factor of 2, and on the order of 2.1 × 10⁷ M⁻¹, observed for wild-type GbaA under these high-salt solution conditions (0.5 M NaCl, pH 8.0) (Figure 7A–C and Table 2). The binding is specific, because GbaA exhibits no interaction with a control DNA operator, CstO (Figure 7A). Remarkably, installation of the C55–C104 disulfide bond in oxidized wild-type GbaA has no impact on DNA binding affinity (Figure 7A and Figure S10), a finding consistent with the in vivo regulation experiments (Figure 2) and the hydrodynamic studies (Figure 4). In striking contrast, the mixed glutathione disulfides in C55A and C104A GbaAs give rise to ≈10- and ≈4-fold lower affinities, respectively (Figure 7B,C), relative to those of their reduced proteins (Table 2). The BSSB-derivatized mixed disulfides of C55A and C104A GbaA bind ≈3-fold weaker than the S-glutathionylated proteins (Figure 7D,E), to a level comparable to that of the NEM-derivatized wild-type GbaA (Figure 7A and Table 2). Both NEM-derivatized Cys substitution mutants bind to the DNA operator even more weakly (Figure 7A–C and Table 2). These findings are fully consistent with the qRT-PCR experiments (Figure 2), which show that wild-type GbaA and component single Cys substitution mutants of GbaA are functional repressors, and that electrophilic addition to one or both Cys allosterically inhibits DNA binding, which drives transcriptional derepression in cells. In striking contrast, the disulfide-cross-linked GbaA is indistinguishable from reduced GbaA in this assay. We also preliminarily investigated the kinetics of NEM- and GSSG-mediated dissociation of the GbaA–DNA operator complex using this same fluorescence anisotropy-based assay (Figure S11) and found that wild-type and C55A GbaAs are rapidly dissociated from the DNA by NEM, while C104A GbaA releases more slowly. In contrast, the kinetics of GbaA dissociation upon addition of the weaker electrophile GSSG is very slow, further evidence that mixed disulfides are unlikely to be regulatory in cells.

Figure 7.

Specific modifications of C55 and C104 impact the DNA binding affinity of GbaA for the DNA operator GbaO to varying degrees. (A) Normalized fluorescence anisotropy-based DNA binding curves for wild-type reduced GbaA (blue squares), GSSG-oxidized GbaA (red circles), and NEM-adducted GbaA (green diamonds). A control cst operator (CstO) titrated with reduced GbaA shows no detectable binding (blue diamonds). (B) Normalized fluorescence anisotropy-based DNA binding curves for reduced C104A GbaA (blue squares) compared to a mixed disulfide glutathione adduct (red squares) and NEM-adducted C104A GbaA (green diamonds). (C) Normalized fluorescence anisotropy-based DNA binding curves for reduced C55A GbaA (blue squares) compared to a mixed disulfide glutathione C55A adduct (red squares) and NEM-adducted C55A GbaA (green diamonds). (D) Normalized fluorescence anisotropy-based DNA binding curves for reduced GbaA C104A (blue squares) compared to a mixed disulfide bacillithiol C104A adduct (red squares). (E) Normalized fluorescence anisotropy-based DNA binding curves for reduced C55A GbaA (blue squares) compared to a mixed disulfide bacillithiol C55A adduct (red squares). Conditions: 25 mM Tris-HCl, pH 7.5, 0.5 M NaCl with 0.5 mM TCEP added only for the reduced protein titrations. The binding parameters defined by the solid line through each data set are listed in Table 2.

Table 2.

Overall Dimer Binding Constants (K_a) and Changes in Anisotropy (Δr) Obtained for Various GbaA Preparations on the GbaA Operator DNA^a

DNA	GbaAb	treatment	Ka (×107 M−1)	Δr
GbaA operator	WT	reduced	2.1 ± 0.2	0.041 ± 0.002
GbaA operator	WT	diamide	2.9 ± 0.1	0.041c
GbaA operator	WT	GSSG	3.2 ± 0.4	0.042 ± 0.004
GbaA operator	WT	GSNO	2.2 ± 0.2	0.045 ± 0.007
CstR operatore	WT	none	NDd	0.001
GbaA operator	C55A	none	4.1 ± 0.3	0.035 ± 0.014
GbaA operator	C55A	GSSG	0.44 ± 0.04	0.035c
GbaA operator	C55A	BSSB	0.15 ± 0.05	0.033c
GbaA operator	C104A	none	2.0 ± 0.1	0.035 ± 0.008
GbaA operator	C104A	GSSG	0.49 ± 0.11	0.046c
GbaA operator	C104A	BSSB	0.14 ± 0.05	0.036c
GbaA operator	WT	NEM	0.11 ± 0.01	0.07c
GbaA operator	C55A	NEM	ND	NDc
GbaA operator	C104A	NEM	ND	NDc

^aConditions: 10 nM OP1 DNA, 25 mM Tris, pH 8.0, 500 mM NaCl, 0.2–0.5 mM TCEP for reduced (no reactions) proteins only. Representative titration data are shown in Figure 7 and Figure S10.

^bWild-type (WT) GbaA and component single-cysteine mutants (C55A and C104A) are as indicated.

^cOne technical replicate only; otherwise, the average results from two replicates are shown.

^dND, none detected (K_a ≤ 10⁵ M⁻¹).

^eFrom ref 27 as a nonspecific DNA binding control.

DISCUSSION

In this study, we define the functional relationship between the conserved cysteines and the regulatory activity of GbaA, a TetR family repressor known to impact the regulation of biofilm dynamics in certain clinical strains of methicillin-resistant S. aureus (MRSA) (Figure S2).^21,45 Cysteine is a relatively rare amino acid that, as a result of its unique ability to access a wide range of oxidative modifications, often plays a regulatory role, while pairs of Cys residues are almost always associated with reversible chemistry converting between dithiol and di- or tetrasulfide.⁴⁹ We make the surprising finding that electrophilic modification of either Cys or both conserved Cys residues in GbaA (C55 and C104) is regulatory in cells, while formation of the C55–C104 disulfide is not, although it can occur in vitro. Consistent with this, the global conformations of the reduced and oxidized GbaAs are virtually indistinguishable; thus, if this modification were to occur in the pathogen in the infected host in the presence of a physiological oxidant, e.g., HOCl,³⁶ it would not be regulatory. What drives GbaA off the DNA is, in contrast, monothiol derivatization by a cellular electrophile of the presumptive protein–DNA interface (C55) or at the base of the GbaA dimer (C104) (Figures 1D and 4E). These two Cys residues are not identical, as C104 is both more reactive toward weak electrophiles (Figure 5) and clearly accessible to derivatization when bound to the DNA (Figure S11). We conclude that GbaA is a bona fide electrophile sensor whose chemical specificity profile is not yet fully defined.

The four core proteins encoded by the other GbaA-regulated genes have not yet been characterized. They include a domain of unknown function (duf 2316; SAUSA300_2514), a putative NmrA (SAUSA300_2513), a vicinal oxygen chelate (VOC) enzyme of known structure that adopts a glyoxalase superfamily fold (PDB entry 4PAV; SAUSA300_2512), and GbaB, annotated as a short chain dehydrogenase (SAUSA300_2516).⁴⁵ Both NmrA and GbaB harbor Rossman folds that bind NADPH or NADP⁺ cofactors and thus may function as oxidoreductases; however, NmrA appears to lack a functional catalytic triad and is therefore projected to simply sense the redox [NAD(P)H/NAD(P)⁺ ratio] status of cells under GbaA regulon-induced conditions. Interestingly, NmrA has been linked to the regulation of nitrogen metabolism in the fungus Aspergillus via nicotinamide dinucleotide-modulated protein–protein interactions.⁵⁰ Although the substrate of GbaB is also not known, in previous work²¹ it was suggested that GbaB catalyzes the formation of a precursor to UDP-N-acetylglucosamine (UDP-GlcNAc), a key constituent of the PIA biofilm and bacterial capsule in other organisms,⁵¹ which might have higher activity as a result of increased glucose and glycolytic flux, which stimulates PIA formation.⁵² Clearly, unregulated expression of GbaA-regulated genes in a gbaA strain enhances biofilm formation in a gbaB-requiring fashion, in both engineered (Figure S2D) and clinically isolated MRSA strains.²⁰ An untargeted metabolomics screen carried out with the WT versus the gbaB strains under biofilm-forming conditions could be used to identify the substrate for GbaB and to quantify both UDP-GlcNAc and NAD(P)H levels in cells, while also uncovering what metabolic pathways are impacted by the loss of GbaB. Such studies may also narrow the search for the bona fide physiological electrophilic inducer sensed by GbaA.

Alternatively, it is known that GbaA-regulated genes are also relatively highly expressed in MRSA strains grown in TSB in the presence of the antibiotics erythromycin, fluconazole, colistin, and vancomycin.⁵³ This might imply that GbaA senses an electrophilic stressor that becomes cell abundant under conditions of general antibiotic stress. Indeed, direct GbaA induction by an electrophilic antibiotic(s) is also possible, given that another VOC family enzyme in S. aureus, FosB, a Mn(II)-dependent thiol S-transferase, detoxifies the antibiotic fosfomycin by catalyzing carbon adduction of l-cysteine or bacillithiol (BSH) to the ring-strained epoxide in fosfomycin, inactivating the molecule.⁵⁴ Interestingly. this chemistry is loosely present in reverse in the S-lactoylglutathione lyase GloA that catalyzes the cleavage of S-lactoylglutathione, the product of glutathionylation of methylglyoxal (MG), a toxic ketoaldehyde^49,55–58 that induces the gba operon (Figure 2) and becomes cell abundant under conditions of unbalanced glycolytic flux.⁵⁹ Increased glycolytic flux is important for biofilm formation in a number of microorganisms,^60–64 which also drives a switch to fermentative energy production and away from cellular respiration.⁶⁵ How these two phenomena are connected remains unclear but may involve the global regulator SrrAB. SrrAB senses oxygen tension and regulates autolysis and biofilm production in S. aureus, thus providing a potential link to hypoxia-induced biofilm production.⁶⁶ Although cellular levels of MG may increase under these conditions, MG may not be the primary or direct physiological inducer of GbaA because we were unable to detect an MG–GbaA covalent adduct by mass spectrometry, and efforts to dissociate preformed GbaA–DNA complexes by MG (see Figure S11) resulted only in protein aggregation (data not shown). In addition, the expression of GbaA-regulated genes does not appear to play a role in protecting S. aureus against MG toxicity (Figure S5).

We note that although TetR family transcriptional regulators are composed of a strongly conserved N-terminal helical DNA domain coupled to a C-terminal regulatory domain (Figure 1 and Figure S1), only a few TetRs harbor regulatory cysteines. These include the N-ethylmaleimide-sensing repressor NemR from E. coli and Acinetobactor spp., CosR from Campylobacter jejuni, and CprB from Streptomyces coelicolor.^35,67–69 In addition, all MRSA strains contain an as-yet uncharacterized TetR encoded by locus tag SAUSA300_2509 that, like GbaA, harbors two cysteines and is reported to sense a wide variety of oxidative and other thiol-modifying stresses.^70,71 Interestingly, SAUSA300_2509-regulated genes SAUSA300_2510 and SAUSA300_2511 are also strongly induced by nitrite-mediated biofilm dispersal (Figure S2A).

NemR was first characterized in E. coli and responds to a number of oxidative and electrophilic stressors, including bleach (HOCl), NEM, and the antibiotic showdomycin, and more weakly to iodoacetamide.^36,72 In A. nosocomialis, NemR is regulated by methylglyoxal, the proposed physiological electro-phile.³⁷ The crystal structure of EcNemR reveals that the one of the cysteines (C106) is proximate (≈4.9 Å) to K175 and forms a sulfenamide linkage upon reaction with N-chlorotaurine.^35,36 Although neither E. coli nor A. nosocomialis NemR is strongly related to GbaA (17% identity over just 30% sequence coverage) and GbaA does not significantly respond to chloramine T (Figure 2), A. nosocomialis NemR negatively regulates biofilm and pellicle formation, like S. aureus GbaA. Both Gram-negative NemRs regulate the expression of nemRA and gloA, which encode an NADPH-dependent oxidoreductase (NemA), long known to reduce NEM while also conferring resistance to toxic electrophiles, e.g., quinones, with the physiological substrate still unclear,⁷³ and GloA, the lactoylglutathione lyase (vide supra).³⁷ S. aureus USA300 FPR3757 does not encode nemRA and gloA, nor is EcNemA homologous to S. aureus NmrA; as a result, the physiological functions of GbaA-regulated genes and their potential roles in biofilm formation and antibiotic resistance await further investigation.

In conclusion, our studies establish that GbaA functions as a monothiol electrophile sensor in cells, and identification of what is likely a small molecule inducer(s) of DNA dissociation and transcriptional derepression of GbaA-regulated genes is a key future objective. This will likely be challenging, however, because 18 years have elapsed since the discovery of IcaR as the transcriptional repressor of icaADBC, and its physiological inducer has not yet been identified with a high degree of confidence.^74,75 The quorum-sensing factor autoinducer-2 (AI-2) may regulate biofilm formation via IcaR, but both positive and negative functions of AI-2 in biofilm formation have been reported in different strains of Staphylococcus epidermidis and S. aureus.^18,19 One possible model for GbaA holds that the product of the GbaB-catalyzed reaction is a physiological inducer of the gba operon because deletion of gbaB has a negative effect on biofilm formation.⁴⁵ What seems clear is that one or both cysteines in GbaA will function as chemical sensors in this process.