VraT/YvqF Is Required for Methicillin Resistance and Activation of the VraSR Regulon in Staphylococcus aureus

Abstract

Staphylococcus aureus infections caused by strains that are resistant to all forms of penicillin, so-called methicillin-resistant S. aureus (MRSA) strains, have become common. One strategy to counter MRSA infections is to use compounds that resensitize MRSA to methicillin. S. aureus responds to diverse classes of cell wall-inhibitory antibiotics, like methicillin, using the two-component regulatory system VraSR (vra) to up- or downregulate a set of genes (the cell wall stimulon) that presumably facilitates resistance to these antibiotics. Accordingly, VraS and VraR mutations decrease resistance to methicillin, vancomycin, and daptomycin cell wall antimicrobials. vraS and vraR are encoded together on a transcript downstream of two other genes, which we call vraU and vraT (previously called yvqF). By producing nonpolar deletions in vraU and vraT in a USA300 MRSA clinical isolate, we demonstrate that vraT is essential for optimal expression of methicillin resistance in vitro, whereas vraU is not required for this phenotype. The deletion of vraT also improved the outcomes of oxacillin therapy in mouse models of lung and skin infection. Since vraT expressed in trans did not complement a vra operon deletion, we conclude that VraT does not inactivate the antimicrobial. Genome-wide transcriptional microarray experiments reveal that VraT facilitates resistance by playing a necessary regulatory role in the VraSR-mediated cell wall stimulon. Our data prove that VraTSR comprise a novel three-component regulatory system required to facilitate resistance to cell wall agents in S. aureus. We also provide the first in vivo proof of principle for using VraT as a sole target to resensitize MRSA to β-lactams.

INTRODUCTION

Infection with methicillin-resistant Staphylococcus aureus (MRSA) has become a major public health concern (1, 2). The recent emergence and dissemination of virulent community-associated (CA) MRSA (3–6) strains and vancomycin-resistant strains (7) underscore the need to develop new antimicrobials. One possible approach in the therapy of MRSA infections is to resensitize a resistant strain to β-lactams.

Analysis of a vancomycin-intermediate-resistant Staphylococcus aureus (VISA) strain revealed that a two-component regulatory system (TCRS) called VraSR was upregulated compared with an isogenic vancomycin-susceptible strain (8) and inactivation of vraSR in a vancomycin-susceptible strain increased vancomycin susceptibility (9). Although vraSR is characterized as vancomycin resistance associated, it is also induced by β-lactam antimicrobials (9–11) and antimicrobials from other classes that target the cell wall, such as daptomycin (12), mersacidin (13), and certain cationic peptides (14). Accordingly, inactivation of vraSR attenuates resistance to β-lactams (9, 11, 15) as well as vancomycin (9, 11) and daptomycin (16). It is therefore possible that an inhibitor of VraS or VraR could broadly restore susceptibility to strains that are resistant to these antimicrobials. In support of this, we have shown that deletion of the vra operon in MRSA improves the outcome of β-lactam therapy in vivo in mouse models of infection (17).

Methicillin, oxacillin (Ox), and related congeners are β-lactams that are resistant to the β-lactamase that is produced by most S. aureus strains. Methicillin resistance emerged shortly after the introduction of this compound in clinical use (18). The methicillin resistance mechanism involves the acquisition of a gene (mecA) encoding a penicillin-binding protein (PBP 2a) with a low affinity for β-lactams, the class of compounds to which penicillin and methicillin belong (19). mecA expression is regulated by a dedicated regulatory system (MecI/MecR1) encoded upstream of mecA and by BlaR1/BlaI β-lactamase-regulatory proteins (20). Inactivation of vraSR decreases methicillin resistance without decreasing mecA expression (15), supporting the idea that the methicillin resistance phenotype is influenced by other chromosomal factors (20).

vraS and vraR together encode a histidine kinase (HK) and response regulator (RR) on a 2.8-kb transcript downstream of two other genes of unknown function, which we call vraU and vraT (previously called yvqF) (10), to reflect their location in the vra operon upstream of vraS (Fig. 1A). When induced by a cell wall synthesis inhibitor, VraS and VraR autoactivate the expression of the vra operon and about 46 other unlinked genes in the vra regulon, several of which encode known or putative cell wall biosynthesis enzymes (9, 10) that can presumably repair cell wall damage.

Fig 1.

Expression of vra operon genes in vra operon deletion strains. (A) Depiction of the single gene deletions in the vra operon of USA300 strain 923. Gray arrows, gene replaced in each construct. Strain M23 is a derivative of an MRSA USA300 isolate (strain 923) containing a vra operon deletion (17). ΔvraU, ΔvraT, ΔvraS, and ΔvraR, single nonpolar deletions in the respective gene. (B) Northern blot of strains with the indicated single gene deletions in the vra operon grown in the absence (Uninduced) or presence (Induced) of oxacillin and hybridized with a vraS-specific probe. Lanes are labeled with the corresponding strains: wild-type USA300 strain (strain 923), the vra operon deletion mutant (M23), the vraR deletion mutant (ΔR), the vraS deletion mutant (ΔS), the vraU deletion mutant (ΔU), and the ΔvraT deletion mutant (ΔT). (C) Western blot of lysates probed with rabbit antiserum specific for VraR (αVraR). A similar abundance of VraR in the presence (lanes +) or absence (lanes −) of oxacillin in strain 923ΔvraT (ΔvraT) is demonstrated. The absence of a VraR-specific product in M23 confirms the specificity of the antiserum for VraR.

VraS is a putative membrane protein belonging to the subfamily of intramembrane-sensing histidine kinases (IMHKs), so called because there is not an obvious extracellular sensing domain characteristic of most HKs (22). VraS responds to antimicrobial-elicited cell wall stress by autophosphorylation and subsequent transphosphorylation of VraR, its cognate RR (23). Phosphorylated VraR binds to its own promoter and presumably to genes in the vra stimulon to activate or repress their transcription (24). It is known that cell wall stress is required, but the specific molecular signal responsible for induction of activation of VraS and VraR is still unknown. The many unrelated classes of antibiotics that affect cell wall integrity and induce this system suggest that a by-product of cell wall damage elicits the response.

VraT (previously called YvqF) is a putative membrane protein that is a homologue of the cell wall response protein LiaF encoded in the cell wall-responsive locus liaHGFSR of Bacillus subtilis (22, 25). VraT/LiaF homologues are conserved in loci containing cell wall-responsive TCRS homologues in Firmicutes, Gram-positive bacteria with low G+C contents (25). Although the role of LiaF in the cell wall stress response has been well characterized in Bacillus species, its role in S. aureus has only recently been studied (26, 27).

The mechanism by which vra effects methicillin resistance is unclear. Since mecA is not involved, one or more of the genes in the vra regulon might be responsible for loss of resistance in the mutant. Alternately, VraU or VraT could be directly responsible for the attenuation of methicillin resistance seen in a vraSR mutant since the operon is autoactivated and downregulated in the vraSR mutant (15). McCallum et al. recently demonstrated that inactivation of vraT (referred to by them as yvqF) attenuated the methicillin resistance phenotype of a laboratory-constructed MRSA strain, BB270 (28). However, the MIC of Ox of the mutant was well within the range of clinical resistance. Since BB270 is a recombinant MRSA strain created in the background of laboratory-adapted strain NCTC 8325, the effect on methicillin resistance might not be observed in clinical MRSA strains. The purpose of this study was to define the roles of VraU and VraT in regulation of the cell wall stimulon and in the methicillin resistance phenotype in a virulent clinical USA300 MRSA strain. We also investigated the outcomes of Ox therapy of a vraT mutant in murine models of lung and skin infection.

MATERIALS AND METHODS

Strains and constructs.

The strains and constructs used in this study are listed in Table 1. The primers used in the cloning constructs are listed in Table S1 in the supplemental material. A nonpolar deletion of each gene in the vra operon (vraU, vraT, vraS, and vraR) was produced in strain 923, a USA300 MRSA strain, employing the allelic replacement vector pMAD, as described previously (29) and detailed in the supplemental material. To prevent the inserted gene from creating a negative polar effect on expression of downstream genes in the operon, the promoter of the antibiotic resistance gene was included and the transcriptional terminator was excluded. To delete vraU or vraT, the tetracycline resistance marker (tet) from the S. aureus-Escherichia coli shuttle vector pAW8 was used. To replace vraS or vraR, the cat gene was obtained from pLI50. To select the deletion mutants, strain 923 harboring a pMAD vra deletion construct underwent serial incubations in tryptic soy broth (TSB) at 30°C and 42°C as described previously (17, 29). For each construct, one white colony growing at 42°C was selected for study. The chromosomal deletion was confirmed by PCR, Southern blotting, and DNA sequencing. Strains 923ΔvraU, 923ΔvraT, 923ΔvraS, and 923ΔvraR have confirmed single nonpolar deletions in vraU, vraT, vraS, and vraR, respectively (Fig. 1; Table 1). For complementation, each gene was amplified by PCR and inserted into pCN48 (30), provided by Richard Novick (30). To construct pVraT, the vraT open reading frame (ORF) was first inserted in the BamHI and KpnI sites in pCN48 using primers VraT-f-BamHI and VraT-r-KpnI (see Table S1 in the supplemental material). To construct plasmid pP_vraT-VraT, a 264-bp vra-specific promoter fragment was isolated by PCR using primers Vra promoter f3-SalI and Vra promoter f3-r-BamHI (see Table S1 in the supplemental material) and inserting it into pVraT.

Table 1.

Plasmids and strains used in this study^a

Plasmid or strain	Description	Source or reference	Selection in S. aureus
Plasmids
pMAD	Allelic replacement vector	29	Em
pCN48	Vector for constructing complementation plasmids	30	Em
pVraT	VraT ORF in pCN48	This study	Em
pPvra-VraT	VraT ORF expressed from the vra promoter (Pvra) in pCN48	This study	Em
Strains
923	CA-MRSA USA300	15	Ox
923vraS	923 vraS insertion mutant M2	15	Em
923ΔvraT	vraT replaced by tetL (Q3009)	This study	Tet
923ΔvraU	vraU replaced by tetL (Q3005)	This study	Tet
923ΔvraS	vraS replaced by cat (Q2920)	This study	Cm
923ΔvraR	vraR replaced by cat (Q2721)	This study	Cm
M23	Strain 923 with a vra operon deletion	17	Em
Q3009-1	923ΔvraT harboring pVraT	This study	Em
Q3009-7	923ΔvraT harboring pCN48 vector	This study	Em
Q3009-3	923ΔvraT harboring pPvraT-vraT in pCN48	This study	Em
Q3078	M23 harboring pPvraT-vraT	This study	Em

^aErythromycin (Em) was used at 5 mg/liter, chloramphenicol (Cm) was used at 5 mg/liter, tetracycline (Tet) was used at 10 mg/liter, and oxacillin (Ox) was used at 2 mg/liter.

Broth MIC.

MICs were determined using Clinical and Laboratory Standards Institute (CLSI) guidelines and 24-well plates (Corning). Ox (a β-lactam that has replaced methicillin in clinical use) was used at final concentrations of 0 to 256 mg/liter. Erythromycin (5 mg/liter) was added to all cultures that contained a pCN48 derivative plasmid. The plates were incubated at 37°C and read at 24 h.

Isolation of RNA and Northern blot analysis.

Induction of the vra regulon was accomplished as follows: S. aureus-saturated cultures were diluted 1:100 in TSB and incubated in a 1:10 culture-to-air volume ratio with shaking (250 rpm) in 50-ml Falcon tubes (30 by 115 mm). After incubation at 37°C for 1 h (optical density at 600 nm [OD₆₀₀], 0.2), the cultures were treated with Ox (1 mg/liter) or water (uninduced) and incubated for an additional hour or as indicated. To release RNA, cells were lysed using recombinant lysostaphin (300 μg/ml; AMBI Products, Lawrence, NY) at room temperature for 10 min. The possible effect of lysostaphin on the cell wall stimulon (31) was avoided by incubating at nonphysiological temperature in a TE (Tris-EDTA) buffer for a short period. Moreover, any possible effects caused by lysostaphin were minimized since the oxacillin-treated and -untreated cells were lysed using identical conditions. The general procedures used for isolation of RNA and quality assurance and the conditions used for Northern blotting were as described previously (10). The PCR primers used to construct probes for the detection of mecA, pbp2, vraS-vraR, murZ, and sgtB are given in Table S2 in the supplemental material. All probes were labeled with [α-³²P]dATP (PerkinElmer) using a Prime-a-Gene labeling system (Promega) as described previously (10).

Real time RT-PCR.

Overnight cell cultures were diluted 1:100 in TSB, incubated at 37°C with shaking for 1 h (OD₆₀₀, ∼0.2), and treated with water or Ox (1 μg/ml) for 1 h.

The Ox concentration used for induction of gene expression and the 1-h duration of induction were chosen on the basis of data showing that the culture does not begin to die until after 1 h using up to 2 mg/liter of Ox (data not shown). Thus, the effect of growth inhibition by Ox was minimized and the effects on gene expression are largely due to cell wall stress prior to cell death. The cell pellets were lysed using lysostaphin (200 μg/ml) in TE at room temperature for 10 min. RNA was extracted using a Qiagen RNeasy kit with DNase treatment, and the RNA integrity and concentration were determined using a NanoDrop 1000 spectrophotometer (Thermo Scientific) and an Agilent 2100 bioanalyzer (Agilent Technologies). cDNA was synthesized using 500 ng of RNA and a high-capacity cDNA reverse transcription (RT) kit (Applied Biosystems). Quantitative RT-PCR (qRT-PCR) was performed in triplicate using the TaqMan universal master mix (Applied Biosystems) and a 7500 Fast cycler (ABI). Differences were analyzed using the ΔΔCT method (32). Each reaction mixture contained 2 μl of cDNA in a 25-μl reaction volume. Primer and probe sets were designed using the Prime Time RT-PCR design software available on the Integrated DNA Technologies (IDT) website. The gyrB probe labeled with the fluorophore Cy5 was used as the endogenous control. To assess vraT expression, the primers used were VRAT-F (TGCGTTATTTAATACACAAGTTTAAACC) and VRAT-R (5′-GTGTACGTTGCTCACCAAAC-3′) and the 6-carboxyfluorescein (FAM)-labeled probe (5′-FAM/AGT TGC GAC/ZEN/GGA TGA GGT TAT GAC TTC-3′) modified at the 3′ end with Iowa Black FQ to quench nonspecific fluorescence.

Production of VraR antiserum.

The ∼20-kDa VraR protein antigen was produced by expressing the vraR ORF from strain 923 in pETt28a. The PCR primers used to amplify the vraR ORF from strain 923 are given in Table S1 in the supplemental material. The protein was purified from E. coli using a His-bind kit (Novagen) as recommended by the manufacturer. Antiserum was collected from rabbits after 4 injections with VraR (Pacific Immunology). The specificity of the antiserum for VraR is demonstrated in Results. Procedures for Western blotting are provided in the supplemental material.

Analysis of membrane topography.

A hidden Markov model (HMMTOP) was used to compare the predicted transmembrane (TM) topology of the VraT and LiaF homologues using HMMTOP (version 2.1) public-domain software (http://www.sacs.ucsf.edu/cgi-bin/hmmtop.py).

Microarray comparison of strain 923 and isogenic vraT and vraS mutants.

Samples from three biological replicates from each strain grown in the presence or absence of oxacillin were examined. Cell growth, Ox induction, and RNA isolation were performed as described above for RT-PCR. Ten micrograms of total RNA was processed for biotin-labeled target preparation for hybridization to Affymetrix S. aureus expression arrays (part number 900154; according to the GeneChip Expression Analysis Technical Manual; Affymetrix, Inc.). The array contains sets of probes to over 3,300 S. aureus open reading frames from strains COL, Mu50, N315, and NCTC 8325. Four micrograms of fragmented and biotin-labeled targets was hybridized to the array for 16 h at 45°C and 60 rpm in an Affymetrix 640 hybridization oven. Arrays were washed and stained with streptavidin-phycoerythrin. The fluorescent signals were amplified using a biotinylated antistreptavidin antibody in an Affymetrix GeneChip 450 fluidics station (Affymetrix, Inc.), and the signals were detected in a GeneChip 3000 7G scanner.

The expression data were extracted using the GeneChip operating software (MicroArray Suite, version 5.0, software; Affymetrix, Inc.), and the CEL files were imported into the GeneSpring program (version 11.5.1; Agilent) to identify differentially expressed genes. Normalization of GeneChip signals was performed using the quantile algorithm robust multiple-array analysis. Analysis of variance was used to identify significantly different gene expression between strains or conditions.

Mouse lung and skin infection model.

The models of Ox treatment of MRSA lung and skin infection in mice (6-week-old male C57BL/6 mice [Harlan Laboratories, Indianapolis, IN]) were previously described (17, 33) and approved by the University of Chicago Institutional Animal Care and Use Committee. In each study, 16 mice were infected with either strain 923 or strain 923ΔvraT; 8 mice infected with each strain were treated with either Ox (400 mg/kg of body weight) or saline. All statistical data were analyzed by using the Prism (version 5) program (GraphPad Software, Inc., San Diego, CA). Differences were considered significant if the P value was <0.05. See the figures for details.

Microarray data accession number.

The microarray data have been deposited in the Gene Expression Omnibus (GEO) database under accession number GSE39627 (www.ncbi.nlm.nih.gov/geo).

RESULTS

Characterization of single gene deletions in the vra operon.

Each gene in the vra operon was deleted by inserting an antibiotic resistance cassette (Fig. 1A) as described in Materials and Methods. To confirm that other genes located downstream of the inserted cassette in the vra operon were not downregulated as a result, Northern blotting was performed on RNA obtained from cultures grown in the presence of Ox (induced) or absence (uninduced) (Fig. 1B). The vraS probe detects the vra transcript, which contains all 4 vra genes. As expected, in the wild-type (wt) control strain 923, the vraS transcript was barely detected in the absence of Ox and its level dramatically increased in the Ox-induced culture. In the vra operon deletion strain M23 (17), the vraS transcript was absent under both conditions, as expected. In strain 923ΔvraR, induction of vraS by Ox was dramatically attenuated, consistent with our previous findings and supporting the notion that the vra operon is positively autoregulated (10).

In 923ΔvraU, the vraS transcript was expressed in the absence of Ox using the resistance cassette's promoter but increased in the presence of Ox, demonstrating that VraU is not required for vra operon induction.

In strain 923ΔvraT, the vraS transcript was expressed constitutively using the resistance cassette's exogenous promoter. In contrast to 923ΔvraU, the vra transcript was not inducible by Ox in strain 923ΔvraT, as evidenced by a similar abundance under inducing and noninducing conditions.

Western blotting using VraR antiserum confirmed that in wt strain 923, VraR production increased in the presence of Ox compared with the level of production in its absence (Fig. 1C). In contrast, in strain 923ΔvraT, a similar abundance of VraR was produced in the absence and presence of Ox. The specificity of the antiserum for VraR was validated by a lack of reactivity in the operon deletion mutant M23 (Fig. 2D).

Fig 2.

Effect of vra operon deletions on methicillin resistance. (A) Broth MICs of Ox for wild-type MRSA strain 923 (black bars) and isogenic strains with deletions in the indicated vra operon genes (vraU, vraT, vraS, vraR) (gray bars). (B) qRT-PCR showing the relative amount of vraT expressed from strain 923, 923 grown with oxacillin (923+Ox), and 923ΔvraT complemented with pCN48 expressing vraT from an endogenous plasmid promoter (ΔvraT pVraT+Ox) or the vra promoter (ΔvraT pP_vra-VraT+Ox). The error bars indicate the minimum and maximum relative quantities (RQ) obtained from 3 biological replicate experiments. (C) MIC of Ox for strain 923ΔvraT after complementation with the vector pCN48 or pCN48 expressing vraT from an endogenous plasmid promoter (pVraT) or from the vra promoter (pP_vra-VraT). Error bars indicate SDs of the means determined from at least 3 independent experiments. (D) MICs of Ox for strain 923, the isogenic vra operon deletion strain M23, and M23 complemented with pP_vra-VraT (M23+vraT). The data show that vraT is not sufficient for Ox resistance. Error bars indicate the SDs of the means determined from at least 3 experiments.

Characterization of strains 923ΔvraU and 923ΔvraT.

The MIC of Ox for strain 923 was significantly higher than that for strain 923ΔvraT, 923ΔvraS, or 923ΔvraR (Fig. 2A). In contrast, the MIC of Ox for strain 923ΔvraU was similar to that for strain 923. The decreased MIC of Ox for 923ΔvraT could be documented, despite the constitutive expression of vraS or vraR (Fig. 1B and C) from the chromosome. We also confirmed that the DNA sequences of vraS and vraR in the ΔvraT mutant were the same as those in the wild type, ruling out the possibility that spontaneous mutations arose during construction of the mutant.

To confirm that the susceptible phenotype of strain 923ΔvraT was solely due to the absence of vraT, we complemented strain 923ΔvraT using two constructs. In plasmid pP_vra-vraT, vraT was expressed from the vra promoter (P_vra). In plasmid pVraT, vraT was expressed from an upstream promoter present on pCN48. We confirmed that the expression of vraT in the complemented strains was similar to or greater than that in strain 923 in the presence of Ox (Fig. 2B). Regardless of the construct used, the MIC of Ox was restored to that of strain 923 when vraT was expressed in trans in 923ΔvraT (Fig. 2C). Thus, vraT expression was sufficient to restore resistance in the vraT deletion mutant. In contrast, methicillin resistance in vra operon deletion strain M23 (17) could not be complemented with vraT (M23 with vraT) (Fig. 2D). This shows that VraT requires VraS and VraR to facilitate the methicillin resistance phenotype.

Since the vraT mutant had increased susceptibility to methicillin, it was possible that expression of mecA was decreased. However, mecA transcript abundance was similar in the wt and ΔvraR, ΔvraS, ΔvraU, and ΔvraT mutant strains, as shown by Northern blotting (Fig. 3). The impact of the deletions on the vra regulon was examined by evaluating expression of the murZ, sgtB, and pbp2 genes, which was previously shown to be controlled by VraSR, by Northern blotting (Fig. 3). Induction of all three genes by Ox was attenuated in strains 923ΔvraT, 923ΔvraS, and 923ΔvraR compared with that in wt strain 923. In contrast, induction of these genes in strain 923ΔvraU was maintained at levels similar to those documented in strain 923.

Fig 3.

Effect of vra operon deletions on expression of mecA and genes in the vra regulon. Northern blots of USA300 strain 923 and isogenic strains with single gene deletions in the vra operon. The strains were cultured in the presence of Ox or were untreated prior to isolation of RNA. Membranes were probed for mecA and the vra regulon genes murZ, sgtB, and pbp2. Lane 1, strain 923; lane 2, ΔvraSR operon deletion strain M23; lane 3, ΔvraR strain; 4, lane ΔvraS strain; 5, ΔvraU strain; 6, ΔvraT strain.

Microarray analysis.

The results shown in the Northern blot suggested that VraT has a regulatory role in the vra regulon. To determine if this finding extends to the entire vra regulon, we compared the genome-wide transcriptional stress response to Ox of strains 923, 923ΔvraT, and 923vraS (10) by microarray analysis.

Ox-induced genes in wt strain 923.

Among the 79 ORFs that were induced ≥2-fold by Ox in strain 923 (see Table S3 in the supplemental material), 42 (53.0%) had significantly decreased expression in strain 923vraS and 43 (54.4%) had significantly decreased expression in strain 923ΔvraT (Table 2). This is similar to the number of VraSR-dependent genes induced by vancomycin identified by Kuroda et al. (9).

Table 2.

ORFs downregulated in vraS and vraT mutants^a

Fold induction by Ox in wt strain 923	Fold decrease in mutantsa		ORF description	Function	S. aureus locus no.		First to identify ORF to be vraSR regulated
	923ΔvraT	923vraS			USA300	N315
2.7	1.4	13	vraR	vra operon	SAUSA300_1865	SA1700	Kuroda et al. (9)
2.9	1.3	2.6	vraS	vra operon	SAUSA300_1866	SA1701	Kuroda et al. (9)
4.6	232.4	3	vraT	vra operon	SAUSA300_1867	SA1702	Kuroda et al. (9)
5.6	2.5	4.4	vraU	vra operon	SAUSA300_1868	SA1703	Kuroda et al. (9)
4.9	2.7	3.7	Hypothetical mechanosensitive channel MscS	Membrane transport	SAUSA300_0362	SA0349	This study
10	13.8	16.6	vraX	Cell wall related	USA300HOU_0572	SAS016	Kuroda et al. (9)
7.6	6.8	7.4	Hypothetical protein	Hypothetical	SAUSA300_0622	SA0591	This studyb
3.1	2.6	3.5	Probable membrane protein, SNARE associated	Transport and protein processing	SAUSA300_0866	SA0824	This study
2.1	5.3	6.4	Chitinase B	Hypothetical	SAUSA300_0964	SA0914	Kuroda et al. (9)
3.7	2.4	2.4	guaC, GMP reductase	Purine metabolism	SAUSA300_1235	SA1172	This studyb
3.4	2.2	2.4	MsrR, methionine sulfoxide reductase regulator	Regulation of protein folding	SAUSA300_1257	SA1195	This studyb
4.2	2.3	2.8	Hypothetical protein	Hypothetical	SAUSA300_1314	SA1254	Kuroda et al. (9)
12	7.9	15.1	Hypothetical protein	Hypothetical	SAUSA300_1606	SA1476	Kuroda et al. (9)
2.5	2.1	2.1	prsA, peptidyl-prolyl cis-isomerase	Protein translocation and folding	SAUSA300_1790	SA1659	Kuroda et al. (9)
3.3	4.1	4	sgtB, monofunctional glycosyltransferase	Peptidoglycan synthesis	SAUSA300_1855	SA1691	Kuroda et al. (9)
9.9	11.1	11.8	RNA methyltransferase	RNA modification	SAUSA300_1877	SA1712	Kuroda et al. (9)
2.1	3.6	3.6	Hypothetical phage protein/ϕPVL, ORF41-like	Phage-related processes	SAUSA300_1961	SA1795	This study
2.7	4.8	2.1	Hypothetical protein	Hypothetical	SAUSA300_1964	SA1799	This study
2.7	4.7	2.1	Antirepressor phage ϕNM	Phage-related processes	SAUSA300_1966	SA1801	This study
3.4	5.7	2.9	Putative phage transcriptional regulator	Phage-related processes	SAUSA300_1968	SA1804
3.4	2.4	2.7	murZ	Peptidoglycan synthesis	SAUSA300_2078	SA1926	Kuroda et al. (9)
3.1	2.5	2.9	lytR	Peptidoglycan degradation	SAUSA300_2259	SA2103	Kuroda et al. (9)
8.3	11.7	17.4	Hypothetical protein	Hypothetical	SAUSA300_2269	SA2113	Kuroda et al. (9)
4.2	3.5	4	tcaA, teicoplanin resistance protein	Antimicrobial resistance	SAUSA300_2302	SA2146	Kuroda et al. (9)
3.5	2.8	3	Hypothetical protein	Hypothetical	SAUSA300_2378	SA2221	Kuroda et al. (9)
4.3	2.1	3.3	fmtA	Peptidoglycan synthesis	SAUSA300_2386	SA2230	Kuroda et al. (9)
3.4	4.2	4.5	Transcriptional regulator, MerR family	Regulation	SAUSA300_2445	SA2296	Kuroda et al. (9)
3.4	4.1	4.2	GTP pyrophosphokinase, relA homologue	Metabolism	SAUSA300_2446	SA2297	Kuroda et al. (9)
3.1	3.6	3.2	Drp35 lactonase	β-Lactam resistance	SAUSA300_2621	SA2480	Kuroda et al. (9)
2.8	2.4	2.7	Hypothetical protein	Hypothetical	SAUSA300_2622	SA2481	This studyb
2.1	9.8	4.9	Nitrite extrusion protein narUK	Nitrate/nitrite metabolism	SAUSA300_2333	SA2176	This study
2.5	5.5	4	Nitrate reductase gamma chain, narI	Nitrate/nitrite metabolism	SAUSA300_2340	SA2182	This study
2.8	9.6	5.2	Respiratory nitrate reductase, delta subunit	Nitrate/nitrite metabolism	SAUSA300_2341	SA2183	This study
2.3	7.5	4	Nitrate reductase beta chain, narH	Nitrate/nitrite metabolism	SAUSA300_2342	SA2184	This study
2.7	6.5	4.6	Uroporphyrin-III C-methyltransferase, putative	Nitrate/nitrite metabolism	SAUSA300_2344	SA2186	This study
3.6	6.8	5.5	Assimilatory nitrite reductase, nirD, nasE	Nitrate/nitrite metabolism	SAUSA300_2345	SA2187	This study
2.2	6.3	4	Nitrite reductase protein nasD, nirB	Nitrate/nitrite metabolism	SAUSA300_2346	SA2188	This study
18.7	32.8	37.2	Hypothetical, related to SCIN	Pathogenesis	SAUSA300_2493	SA2343	Kuroda et al. (9)
ORFs with ≥2-fold lower expression in either 923ΔvraT or 923vraSa
2.4	2.3	1.9	psmβ1 and -β2	Antibacterial and hemolytic	SAUSA300_1068	NF	This study
2.1	3.3	1.7	RecT	Phage-related processes	SAUSA300_1960	SA1794	This study
2.5	2	1.7	Phage ϕN315, ϕPVL ORF39-like	Phage-related processes	SAUSA300_1962	SA1797	This study
2.7	4.5	1.9	Phage protein, ϕN315	Phage-related processes	SAUSA300_1967–68	SA1802	This study
2.2	1.8	2.1	Hypothetical protein	Hypothetical	SAUSA300_0639	SA0608	Kuroda et al. (9)
2.1	1.9	2.4	Hypothetical protein	Hypothetical	SAUSA300_0779	SA0749	This study
2.8	1.9	2.6	Signal peptidase, spsA	Folding, sorting, and degradation	SAUSA300_0867	SA0825	Kuroda et al. (9)
2.4	1.7	2	Undecaprenyl pyrophosphatase synthetase, uppS	Peptidoglycan synthesis	SAUSA300_1153	SA1103	This studyb
2.7	1.8	2.2	PTS system glucose-specific enzyme II A	Glycolysis and starch metabolism	SAUSA300_1315	SA1255	Kuroda et al. (9)

^aStrains were compared with the wild type in the presence of Ox. ORFs are listed in the ascending order of the USA300 locus number. The SA prefix indicates that the annotation is from the genome sequence of strain N315; the USA300 prefix is from the sequence of USA300 strain FPR3757. USA300HOU_0572 is from the genome sequence of USA300_TCH1516 (CA methicillin-susceptible S. aureus). Functions were ascertained from annotations in the KEGG database. NF, not found; SCIN, staphylococcal complement inhibitor; PTS, phosphoenol pyruvate phosphotransferase system.

^bORFs that are newly identified members of the vraSR stimulon that were induced by Ox by Utaida et al. (36).

Of most importance was the finding that the majority of ORFs that were downregulated in the presence of Ox in strain 923ΔvraT were also downregulated in strain 923vraS (Table 2); 36 ORFs (excluding the ORFs in the vra operon) were downregulated at least 2-fold in both mutants compared with their regulation in strain 923. Interestingly, 5 ORFs with the greatest induction in strain 923 also had the greatest decrease in expression in the vraS and vraT mutants, including SAS016 (identical to vraX [8]), SAUSA300_1606, SAUSA300_2269, SAUSA300_1877, and SAUSA300_0622 (Table 2).

When comparing the vraS- and vraT-dependent genes identified in our analysis to those previously published (9), 22 additional members of the vraSR regulon were identified (Table 2). These included uppS (undecaprenyl pyrophosphatase), involved in the synthesis of the lipid carrier of lipid II peptidoglycan precursors; a hypothetical mechanosensitive channel (SAUSA300_0362); a ϕPVL ORF41-like hypothetical phage protein (SAUSA300_1961); an antirepressor phage protein; a GMP reductase (SAUSA300_1235); six genes from the assimilatory nitrogen gene cluster; and phenol-soluble modulins (PSMs; psmβ1 and -β2). As indicated in Table 2, although some of these genes are previously recognized members of the cell wall stimulon (36), they have not previously been classified as members of the vraSR regulon.

The results for sgtB, murZ, and pbp2 are concordant with the Northern blotting data (Fig. 3). Since PSMs were not previously associated with the vraSR regulon, their expression was validated in the vraS mutant using qRT-PCR. It was found that expression of psmβ1 and -β2 decreased in strain 923vraS to an average of 30 to 34% of that for Ox-induced strain 923 (data not shown).

Ox-repressed genes in wt strain 923.

In strain 923, 102 ORFs were downregulated by Ox at least 2-fold (see Table S3 in the supplemental material). Compared with strain 923 grown in the presence of Ox, 28 ORFs were derepressed ≥2-fold in 923ΔvraT and 42 were derepressed in 923vraS (Table 3). Importantly, 21 ORFs were derepressed ≥2-fold in both mutant strains (Table 3). Although none of these were previously identified by Kuroda et al. (9), the fact that they were derepressed in both mutant strains strengthens the validity of the data and supports the conclusion that they are members of the vra regulon that are normally repressed by VraSR in the presence of Ox. These data also support the conclusion that VraT, VraS, and VraR corepress the same genes in the cell wall stimulon. Interestingly, many of the genes repressed by Ox and regulated by vraSR are virulence factors, including protein A (spa), repressor of toxins (rot), and the genes essA and esxA, involved in type 6 secretion (37).

Table 3.

ORFs derepressed in vraS and vraT mutants^a

Fold repression by Ox in wt	Fold derepression in mutantsa		ORF description	Function	S. aureus USA300 locus	Locus
	923ΔvraT	923vraS
6.4	5.4	2.5	Protein A, Spa	Pathogenesis	SAUSA300_0113	SA0107
2.1	2.5	2.2	SirA, iron ABC transporter, lipoprotein	Transporter	SAUSA300_0117	SA0111
2	2.4	1.9	Branched-chain amino acid transport protein	Amino acid transport	SAUSA300_0188	SA0180
3.2	2.3	2.9	AcpD, FMN-dependent NADH azoreductase	synthesis of membrane-derived oligosaccharides from UDP-glucose	SAUSA300_0206	SA0204
3.7	3.4	2	Antiholin-like protein LrgA	Peptidoglycan degradation	SAUSA300_0256	SA0252
2.9	2.6	1.8	Antiholin-like protein LrgB	Peptidoglycan degradation	SAUSA300_0257	SA0253
4.4	3.7	2.6	Hypothetical protein	Hypothetical	SAUSA300_0274	SA0269
2.3	2.6	2.1	Hypothetical protein	Hypothetical	SAUSA300_0275	NF
3.9	2.4	2.8	Staphyloxanthin biosynthesis protein	Transport and pathogenesis	SAUSA300_0277	SA0270
2.4	2.3	2.5	EsxA	Transport and pathogenesis	SAUSA300_0278	SA0271
2.5	2.5	3.4	EsaA	Transport and pathogenesis	SAUSA300_0279	SA0272
2.8	2.3	2.2	Acid phosphatase 5′-nucleotidase, lipoprotein	Degradation, transport	SAUSA300_0307	SA0295
2.5	4.1	4.4	Superoxide dismutase (Mn/Fe family)	Stress response	SAUSA300_0314	SA0303
2.3	2.8	2.5	Hypothetical	Hypothetical	SAUSA300_0938	NF
2.2	2.4	2.2	Glycosyltransferase, group 1 family protein	Starch/sucrose metabolism	SAUSA300_0939	SA0522
2.5	2.9	2.6	Putative lipoprotein	Lipoprotein	SAUSA300_1106	SA1056
2.3	2.2	1.8	Hypothetical	Hypothetical	SAUSA300_1210	NF
6.8	3.6	5.1	Thermonuclease	Endoribonuclease, pathogenesis	SAUSA300_1222	SA1160
3.2	2	2.3	Probable nitroreductase family protein	Nitrate/nitrite metabolism	SAUSA300_1986	SA1840
2.8	2.1	1.9	ABC transporter NatA	Transporter	SAUSA300_2288	SA2132
2.6	2.2	2.2	Hypothetical	Hypothetical	SAUSA300_2289	SA2133
9.6	3.1	2.6	ABC transporter	Transporter	SAUSA300_2453	SA2302
6.3	2.3	1.9	Membrane-spanning protein SmpB	Membrane spanning	SAUSA300_2454	SA2303
2.2	2.4	2.3	Hypothetical	Hypothetical	SAUSA300_2528	SA2377
4.3	3.5	3.6	Repressor of toxins Rot	Regulation, pathogenesis	SAUSA300_2543	SA2403
2.2	2.7	2.5	Citrate transporter	Transporter, metabolism	SAUSA300_2552	SA2411

^aStrains were compared with the wild type in the presence of Ox. ORFs are listed in the ascending order of the USA300 locus number. The SA prefix indicates that the annotation is from the genome sequence of strain N315; the USA300 prefix is from the sequence of USA300 strain FPR3757. NF, not found.

Membrane topology predictions of VraT.

VraTSR is orthologous to the cell wall-active antibiotic response two-component system in Bacillus (LiaSR) and other Firmicutes (25). Accordingly, VraT is similar to LiaF/YvqF encoded in the LiaSR operon that is present in diverse Firmicutes bacterial species (Table 4; Fig. 4). However, whereas VraT positively influences expression of the cognate vraSR operon, LiaF negatively modulates LiaSR expression in Bacillus (25). To gain insight into the functional differences between the LiaF and VraT homologues, we analyzed the protein sequences using topological prediction tools. All members of this family contain a conserved C-terminal domain (DU2154) and 2 to 5 N-terminal TM-spanning helices (Fig. 4; Table 4). Using a hidden Markov model (HMMTOP), it is predicted that the S. aureus VraT orthologue has 5 TM-spanning helices, the most among this family of proteins. (An alternate prediction algorithm, HMMTMM, predicts there are 4 transmembrane helices for VraT). In contrast, HMMTOP predicts that LiaF (from B. subtilis) has just 2 TM helices. The lia and vra operons also differ with respect to the number of genes present; several genes in the lia operon in Bacillus spp. are absent in staphylococci (4 are present in staphylococci and 5 are present in bacilli). The different membrane architectures and operon structures and the disparate roles of LiaF and VraT between bacilli and staphylococci might reflect the strategies that they have evolved to respond to the distinct challenges that they encounter in distinct environmental niches. Interestingly, the VraT homologues from other staphylococcal species also have a predicted TM architecture that differs from that of S. aureus (Table 4), suggesting that there might be different response repertoires even among staphylococci. The actual orientation of the protein in the membrane remains to be determined for VraT.

Table 4.

Transmembrane topologies of VraT/LiaF homologues^a

Organism (strain)	GenBank accession no.	Locus tag	aa positions of TM helices	psi BLAST e value (% identity)
Staphylococcus aureus (Mu50)	NP_372410.1	SAV1886	10–27, 32–49, 56–73, 78–95, 102–119	Query
Staphylococcus haemolyticus JCSC1435	YP_252984.1	SH1069	10–28, 33–51, 66–90	8e−139 (81)
Staphylococcus epidermidis (M23864:W1)	ZP_04817379	HMPREF0793_0247	10–28, 33–51, 66–90	6e−135 (86)
Staphylococcus simiae CCM 7213	ZP_09147023	SS7213T_08747	10–28, 33–51, 64–88	3e−135 (91)
Staphylococcus lugdunensis HKU09-01	YP_003471302.1	SLGD_01043	10–28, 33–51, 66–90	2e−132 (82)
Staphylococcus warneri L37603	ZP_04679015.1	STAWA0001_0601	10–28, 33–51, 66–90	2e−128 (86)
Staphylococcus pseudintermedius	YP_004149802	SPSINT_1638	10–28, 33–51, 64–88	7e−121 (69)
Listeria monocytogenes serotype 4b strain F2365	YP_013641	LMOf2365_104	10–34, 55–79	6e−17 (30)
Enterococcus faecium TX1330	ZP_03980568.1	HMPREF0352_0461	21–45, 66–90	3e−12 (26)
Streptococcus pneumoniae	NP_357936	spr0342	6–23, 30–46, 53–77	4e−10 (27)
Bacillus anthracis (Ames)	NP_843911.1	BA_1455	22–46, 61–85	5e−09 (25)
Bacillus subtilis subsp. spizizenii TU-B-10	YP_004878824	GYO_3616	6–30, 57–81	2e−06 (25)

^aTopologies were determined using the HMMTOP server (http://www.enzim.hu/hmmtop/) for prediction of transmembrane helices and membrane topology using hidden Markov models. The proteins listed here were a subset of those identified by psi BLAST analysis using the S. aureus VraT protein sequence as the query. All proteins with significant alignment belong to the cell wall-active response family, including LiaF from Bacillus spp., and contain a conserved domain in the carboxy terminus (DUF154). S. aureus has a unique membrane topology compared with other Staphylococcus species and all other members from this family (DUF154).

Fig 4.

Predicted membrane topology of VraT from S. aureus and the orthologous cell wall-responsive protein LiaF from Bacillus subtilis. A hidden Markov model was used to compare the predicted transmembrane topology of the VraT and LiaF homologues using the HMMTOP (version 2.1) public-domain software (http://www.sacs.ucsf.edu/cgi-bin/hmmtop.py). See Table 4 for further details. Whereas LiaF is predicted to contain 2 TM helices, VraT contains as many as 5. The HMMTMM algorithm predicts 4 TM helices in VraT. The VraT protein from strain Mu50 (GenBank accession number NP_372410.1) was analyzed. The LiaF protein analyzed was from Bacillus subtilis subsp. spizizenii TU-B-10 (GenBank accession number YP_004878824).

In vivo effect of vraT deletion on methicillin resistance in murine pneumonia.

Although the MIC of Ox for 923ΔvraT dramatically decreased in vitro compared with that for the wt parent strain, the presence of mecA and the fact that the MICs are above the susceptibility breakpoint (4 to 8 mg/liter) suggested that the strains could still exhibit resistance in vivo during therapy. To evaluate this, we compared Ox therapy in mice infected with strains 923 and 923ΔvraT in our in vivo models of skin and lung infection (Fig. 5).

Fig 5.

Effect of Ox treatment on pneumonia (A and B) and skin infection (C and D) among mice infected with wild-type CA-MRSA (strain 923) or its isogenic vraT deletion mutant (ΔvraT). Mice were infected with wild-type MRSA strain 923 or its isogenic vraT deletion mutant and treated with Ox. Eight animals were tested in each group. (A) Survival at 18, 24, 48, and 72 h after inoculation with 4.8 × 10⁸ CFU/mouse. Groups of mice were compared for survival by a log rank test. (B) Numbers of bacterial CFU in the lung at 24 h after infection with a sublethal inoculum (1.1 × 10⁸ CFU/mouse). Horizontal lines indicate the median numbers of CFU for each group. 923/Sa, mice infected with strain 923 and treated with normal saline; 923/Ox, mice infected with strain 923 and treated with Ox; ΔvraT/Sa mice infected with strain 923ΔvraT and treated with normal saline; ΔvraT/Ox, mice infected with strain 923ΔvraT and treated with Ox. (C) Area of dermonecrosis after subcutaneous infection with 1.4 × 10⁷ CFU plotted over 13 days. The areas of dermonecrosis in groups of mice were compared by an unpaired t test. (D) Numbers of bacterial CFU in the lesions 3 days after inoculation. Bacterial densities were compared by a Mann-Whitney test.

The mortality among mice infected with wt MRSA strain 923 treated with Ox or saline did not differ significantly (P = 0.85) (Fig. 5A). Similar mortality was observed among untreated mice infected with strain 923ΔvraT. In contrast to mice infected with 923ΔvraT and treated with saline, all mice infected with strain 923ΔvraT survived the 72-h observation period if they were treated with Ox (P = 0.025). Moreover, the rate of survival of Ox-treated mice infected with strain 923ΔvraT was significantly greater than that of Ox-treated mice infected with strain 923 (P = 0.009) (Fig. 5A).

The bacterial burden in the lungs at 24 h after inoculation was significantly lower in mice infected with strain 923ΔvraT and treated with Ox than in mice infected with strain 923ΔvraT and treated with saline (P < 0.001) or Ox-treated mice infected with strain 923 (P < 0.001) (Fig. 5B). In contrast, there was no significant difference in the lung bacterial burden in mice infected with wt strain 923 when those receiving Ox and saline treatments were compared (P = 0.28).

In vivo effect of vraT deletion on methicillin resistance in murine skin infection.

Dermonecrotic lesions developed in mice inoculated with strain 923ΔvraT when treated with saline, whereas they did not in Ox-treated mice inoculated with this strain (P < 0.0001) (Fig. 5C and D). Despite the absence of dermonecrosis, four of the eight 923ΔvraT-inoculated, Ox-treated mice developed a very small subcutaneous abscess (diameter, <2 mm) that resolved within 7 days postinoculation.

Saline-treated mice inoculated with strain 923 or 923ΔvraT developed dermonecrotic lesions of similar size (P = 0.75) that peaked at 2 days after inoculation and diminished thereafter (Fig. 5C). Dermonecrotic lesions uniformly appeared in Ox-treated mice inoculated with strain 923 and diminished with a time course similar to that observed among the saline-treated animals. However, the mean maximal area of dermonecrosis was significantly smaller than that observed among saline-treated controls (P = 0.042).

Among Ox-treated mice infected with strain 923ΔvraT, the bacterial burden in the skin lesions was significantly lower than that in the skin lesions of the saline-treated mice (P < 0.001) and Ox-treated mice infected with strain 923 (P < 0.001) (Fig. 5D). There was no significant difference in the bacterial burden in the skin lesions between the saline-treated mice inoculated with strain 923 and those inoculated with strain 923ΔvraT (P = 0.38). Similarly, the bacterial burden in the lesions of the Ox- or saline-treated mice infected with strain 923 did not differ significantly (P = 0.88).

DISCUSSION

Two-component regulatory systems provide a means for bacteria to respond to changes or stressors in their environment by inducing or repressing genes required for adaptation (34). VraS and VraR are tuned to respond and adapt to antimicrobials that interfere with cell wall synthesis or perturb the cell membrane (9, 10, 12, 14, 15). Utaida et al. described a large set of genes, collectively referred to as the cell wall stimulon, that is activated in response to cell wall-acting antibiotics from distinct chemical classes, Ox, bacitracin and d-cyclosporine (36). Kuroda et al. described the VraSR two-component regulatory system that was upregulated in VISA strains (8) and by using microarray analysis with a vraSR mutant of MRSA strain N315 identified vancomycin-induced genes that were dependent on VraSR (9). The role of VraSR in the Ox resistance phenotype has now been demonstrated in many different genetic backgrounds of S. aureus (9, 11, 15). Daptomycin also activates a VraSR-dependent cell wall stimulon (12), and accordingly, VraSR has been shown to be involved in daptomycin nonsusceptibility (16). The microarray analysis that we performed extends these studies. We have identified newly recognized members of the cell wall stimulon and the vraSR stimulon (Table 2). Some of the newly identified members of the vraSR regulon are previously recognized members of the cell wall stimulon, but the regulatory association with VraSR had not been made (36). Interestingly, a large majority of the genes that we showed were repressed by Ox and repressed by VraSR encode virulence factors, including spa (protein A), rot (repressor of toxins), and essA and esxA, which are involved in type 6 secretion (37).

This study in particular sheds new light on the role of VraT, the determinant encoded immediately upstream of vraS in the 4-gene vra operon. We show that VraT acts together with VraS and VraR to induce about half the genes in the Ox stimulon and that vraT is essential for optimal expression of the methicillin resistance phenotype in the USA300 epidemic MRSA background that is prevalent in the United States. Our data agree with and extend the findings of a similar study reported by McCallum et al., who also created a nonpolar mutation in vraT (called yvqF by them) (28). However, several differences exist between our study and theirs. First, McCallum et al. (28) studied a laboratory-adapted, recombinant strain of 8325 (strain BB270) which has a defective SigB global regulatory system (39, 40). This may be important, given the finding that although the MIC for BB270 vraT decreased 2.7-fold compared to that for the parent strain, the MIC of Ox was 96 mg/liter, well within clinical resistance, and much higher than that for strain 923ΔvraT (4 to 8 mg/liter). Thus, the SigB defect could unpredictably affect the vraSR regulon. We also extend the observations of McCallum et al. (28) by demonstrating the effect of the vraT mutation on the entire vra regulon.

The importance of VraT in the methicillin resistance phenotype has potential implications for the development of novel therapies against MRSA infections. Although we found a decreased MIC of Ox for the vraT mutant in vitro, the possibility existed that the MICs of Ox were not attenuated sufficiently to improve Ox therapy in vivo. Our results alleviate this concern, since Ox was indeed significantly more effective in treating mice infected with the vraT mutant strain than in mice infected with the wild-type strain. Thus, disabling a single target in the operon is sufficient to attenuate methicillin resistance during infection of a live host. This finding extends our previous observation that deletion of the entire vra operon improves the outcome of Ox therapy in mice (17).

VraSR responds to a wide range of cell wall antimicrobials and affects susceptibility to them as well. The role that VraT plays in conferring resistance to vancomycin was recently shown in a VISA clinical isolate with a point mutation in vraT; complementation of that isolate with a wild-type yvqF (vraT) gene restored vancomycin susceptibility in vitro (26). Thus, an inhibitor of VraT could be sufficient to potentiate glycopeptide as well as β-lactam therapy. This idea could extend to daptomycin therapy also since the role of VraSR in daptomycin resistance has been demonstrated (16).

It is not known exactly how VraT affects the methicillin resistance phenotype. However, we explored three possibilities. First, we considered but ruled out the possibility that the decreased resistance of the vraT mutant was due to decreased mecA gene expression. This is consistent with our findings in a vraSR mutant (15). The other possibilities that we considered were that (i) VraT plays a direct role in methicillin resistance by, for instance, detoxification or inactivation of Ox or (ii) VraT plays an integral role in the signal transduction and regulatory function of the VraSR system. The former possibility was ruled out, since VraT alone does not restore resistance to a vra operon deletion strain.

To address the second of the latter two possibilities, we undertook differential expression analysis to compare the stress response to Ox between the wt and the vraT and vraS mutants. The remarkable similarity of the Ox induction profiles of the vraT and vraS mutants demonstrates that VraT plays a crucial regulatory role in the VraSR-dependent cell wall stress response to β-lactams like Ox. This provides clear insight into the role of VraT that has not been rigorously evaluated previously and supports a model in which VraS, VraR, and VraT comprise a three-component regulatory system that is required to facilitate the methicillin resistance phenotype (Fig. 6). Briefly explained, given our knowledge of the HK/RR paradigm (41), our data, and data from other published studies, cell wall damage first imparts a modification of VraT which in turn influences autophosphorylation of VraS, ultimately leading to the phosphorylation of VraR. VraR then activates or represses genes through binding to target promoters (23, 24, 42). Recent data demonstrating that portions of VraS and VraT interact in vivo support this idea (28). Interactions between VraS and VraT might, for instance, modulate the degree of VraS phosphorylation or phosphatase activity upon VraR. Although it is possible that Ox interacts directly with VraS or VraT, this is unlikely, given that agents from classes as diverse as glycopeptides, a cyclic lipopeptide, cationic peptides, and β-lactams induce the regulon, suggesting that a by-product of inhibited cell wall synthesis rather than the drugs themselves provides the signal.

Fig 6.

VraTSR three-component regulatory system. Proposed model for the role of VraT in methicillin resistance in signaling to the VraTSR regulatory system. (A) The proposed membrane topology of VraT based on the HMMTOP algorithm predicts 5 TM helices connected by very short intervening segments (Fig. 4 and Table 4). The molecular sequence of events is not known with certainty, but we can speculate that cell wall damage first imparts a modification of VraT which in turn influences autophosphorylation of VraS, ultimately leading to the phosphorylation of VraR. VraR can then activate or repress its own operon and other genes in the cell wall stimulon. Brown arrows in vra operon, genes shown to be required for methicillin resistance; white arrow, nonessentiality for vraU in methicillin resistance; black double-headed arrows in membrane and cytoplasm, possible interaction between VraT and VraS following a stimulus. The signal for the stimulus is unknown but is likely created by a by-product of cell wall stress (lightning bolt). Dotted red and pink arrows, hypothetical interactions between the antimicrobial and VraS and VraT and between VraT and VraR; blue wavy line beneath vra operon, mRNA expression leading to expression of VraR, which in turn autoactivates the vra operon and members of the cell wall stimulon.

VraS belongs to the IMHK family of HKs, which is characterized by two TM helices with a small interconnecting loop that has been postulated to be buried in the membrane and therefore not available for sensing an extracellular environmental cue (22). In S. aureus, there are 3 additional two-component systems belonging to the IMHK family: GraSR (43), BraSR (44), and SaeSR (45). Recent discoveries about the way in which graSR and braSR operons self-regulate to respond to antimicrobial threats may provide insight into the regulation of vraTSR.

For instance, the notion that a third protein could play a regulatory role in the IMHK/RR pair has recently been elucidated for GraSR and BraSR, which, like VraSR, are involved in sensing and responding to antimicrobials. Like vraSR, the graSR operon responds to and regulates genes in response to cationic peptides, daptomycin, and vancomycin (43, 46). Like vraT in the vraSR operon, a third gene, graX, plays a role in signaling through GraSR and is encoded immediately upstream of graSR (46–49). However, VraT encodes a putative membrane protein, whereas GraX is likely localized in the cytoplasm (48). GraSR is also modulated by the VraFG ABC transporter encoded immediately downstream of graSR. Although it is similar to ABC transporters, VraFG plays a role in signaling through GraSR (48, 49) rather than by acting as a drug pump. Thus, GraXSR-VraFG is a 5-component regulatory system (48).

In response to bacitracin, BraSR activates an unlinked dedicated ABC transporter (VraDE) involved in detoxification of bacitracin (44). Like GraX and VraT, a protein of unknown function is encoded upstream of BraSR; however, its role is not yet known. Interestingly, expression of an ABC transporter, BraDE, encoded immediately downstream of braSR is activated by BraSR, but BraDE has no apparent detoxification role; instead, it acts only as a sensor in partnership with BraSR. Thus, VraFG and BraDE are members of a new family of ABC transporters in the Pep7E family that effect regulation of gene expression and antimicrobial resistance (50) but do not export the antimicrobials out of the cell (44, 47–51). VraE and BraE have 10 TM helices with a long putative extracellular loop (180 to 300 amino acids) between the 7th and 8th helices (47). It is therefore perhaps no accident that they partner with IMHKs since the extracellular loop senses the antimicrobial, thereby substituting for the missing loop from the partner IMHK (48, 50). Although it is a putative membrane protein and a partner with an IMHK, VraT (and LiaF) is not a member of the Pep7E signaling transporter family (50). As shown in our analysis, VraT possibly has 5 or 4 TM helices (according to the HMMTOP or HMMTMM algorithms, respectively), with each helix connected by a small interhelix bridge (3 to 5 amino acids) that would not be available to interact with antimicrobials external to the cell (48). However, the extracellular carboxy terminal tail predicted in the model for VraT in Figure 4 could conceivably interact with extracellular antibiotics. Therefore, VraT is not similar to either GraX or VraFG and it acts in a unique unknown way to relay a signal to VraSR.

Although methicillin resistance requires the presence of PBP 2a encoded by mecA, it is not sufficient; accessory factors termed factors essential for methicillin resistance (FEMs) are also important to the resistance phenotype (20). In contrast to other FEMs, such as FemA, FemB, and FemC, which are enzymes involved in peptidoglycan precursor synthesis (20), VraTSR plays a regulatory role. Indeed, two FEMs, fmtA and pbp2, are members of the vra regulon. Thus, the VraTSR three-component system regulates some of the previously identified FEMs.

In conclusion, this is the first study to demonstrate the role of VraT in methicillin resistance in an in vivo model and to conclusively provide evidence that VraT is involved in the global stress response of the VraSR system. Also, we provide the first in vivo proof of principle for the use of VraT as a sole target for antimicrobial development to resensitize MRSA to β-lactams.

That this was studied in the USA300 strain background is especially important, considering its widespread prevalence in the United States and its ability to rapidly disseminate, even among healthy individuals with no underlying predisposing risk factors for MRSA infection.