Decreased Vancomycin Susceptibility in Staphylococcus aureus Caused by IS256 Tempering of WalKR Expression

Abstract

Vancomycin-intermediate Staphylococcus aureus (VISA) strains often arise by mutations in the essential two-component regulator walKR; however their impact on walKR function has not been definitively established. Here, we investigated 10 MRSA strains recovered serially after exposure of vancomycin-susceptible S. aureus (VSSA) JKD6009 to simulated human vancomycin dosing regimens (500 mg to 4,000 mg every 12 h) using a 10-day hollow fiber infection model. After continued exposure to the vancomycin regimens, two isolates displayed reduced susceptibility to both vancomycin and daptomycin, developing independent IS256 insertions in the walKR 5′ untranslated region (5′ UTR). Quantitative reverse transcription-PCR (RT-PCR) revealed a 50% reduction in walKR gene expression in the IS256 mutants compared to the VSSA parent. Green fluorescent protein (GFP) reporter analysis, promoter mapping, and site-directed mutagenesis confirmed these findings and showed that the IS256 insertions had replaced two SigA-like walKR promoters with weaker, hybrid promoters. Removal of IS256 reverted the phenotype to VSSA, showing that reduced expression of WalKR did induce the VISA phenotype. Analysis of selected WalKR-regulated autolysins revealed upregulation of ssaA but no change in expression of sak and sceD in both IS256 mutants. Whole-genome sequencing of the two mutants revealed an additional IS256 insertion within agrC for one mutant, and we confirmed that this mutation abolished agr function. These data provide the first substantial analysis of walKR promoter function and show that prolonged vancomycin exposure can result in VISA through an IS256-mediated reduction in walKR expression; however, the mechanisms by which this occurs remain to be determined.

INTRODUCTION

Methicillin-resistant Staphylococcus aureus (MRSA) is an opportunistic pathogen, and the glycopeptide antibiotic vancomycin is generally the first choice of treatment for serious MRSA infections (1). Although vancomycin has been in clinical use for over 5 decades, reduced vancomycin susceptibility in clinical MRSA isolates was first reported in 1997 (2). Subsequent studies have divided the vancomycin resistance phenotype into two major categories (3). The first comprises the rarely observed high-level resistance (MIC, ≥16 μg/ml) which results from the horizontal acquisition of the enterococcal vanA operon (4). The second resistance phenotype comprises both vancomycin-intermediate (VISA) and -heteroresistant (hVISA) S. aureus strains. MRSA isolates in the VISA/hVISA category have lower MICs (2 to 8 μg/ml) but remain clinically relevant and are linked to a higher incidence of persistent bloodstream infection and treatment failure (5–8).

Comparative genomic studies using isogenic vancomycin-susceptible S. aureus (VSSA) pairs isolated before and following vancomycin treatment failure or following vancomycin exposure in vitro have demonstrated the existence of multiple potential pathways in the generation of the VISA phenotype (9–14). Several genes have been implicated, but certain regulatory genes comprising members of two-component regulatory systems (TCS) are overrepresented. Associations between VISA and mutations in the graRS and vraRS TCS have been demonstrated in numerous studies (11, 12, 15–19) and recently mutations in genes of the walKR TCS have been shown to be responsible for a large proportion of clinical VISA isolates (9, 13, 20). walKR is the only essential two-component regulator in S. aureus, and this is thought due to its role in the regulation of cell wall metabolism-associated genes and specifically as a regulator of peptidoglycan biosynthesis during cross bridge hydrolysis (21–23). Bidirectional allelic exchange experiments have shown that apart from increasing the vancomycin MIC, walKR mutations can also lead to decreased susceptibility to the lipopeptide antibiotic daptomycin (9). These studies have also demonstrated that walKR mutations can lead to other phenotypic alterations commonly associated with VISA strains, including reduced autolytic activity, increased cell wall thickening, decreased virulence, and a reduction in biofilm formation and agr activity (9, 13). A further notable finding is that VISA-associated walKR mutations can occur throughout the various domains of both walK and walR. Although their functional influence on WalKR activity has not been measured, it has been assumed that these mutations inhibit or reduce normal protein activity (9, 13, 20). However, it has also been reported that a spontaneous IS256 insertion into the walKR 5′ untranslated region (5′ UTR) results in walKR upregulation, as determined by microarray and quantitative reverse transcription-PCR (qRT-PCR) analysis, and reduced vancomycin susceptibility (24). In the latter case it was hypothesized that the integration of IS256 created a strong hybrid promoter upstream of walR (24).

Using comparative genomics and site-directed mutagenesis, we have previously shown that single nucleotide polymorphisms (SNPs) within both the graS and walK genes contribute to a decrease in vancomycin susceptibility in the VISA strain of the isogenic VSSA clinical isolate pair JKD6009/JKD6008 (9, 18). We have subsequently sought to determine the types of mutation induced in the parental VSSA strain JKD6009 when placed under vancomycin selection pressure in vitro. Here we report our observations relating to walKR mutations in 10 isolates recovered after in vitro exposure to vancomycin simulating human pharmacokinetics from an in vitro hollow fiber infection model (HFIM) over a 10-day period. Our findings have led us to a detailed analysis of the walKR promoter and an investigation into the impact of IS256 integration on the expression of both walKR and several previously described walKR-controlled cell wall metabolism genes.

MATERIALS AND METHODS

HFIM.

The parental methicillin-resistant VSSA strain JKD6009 (Australasian ST239) has been described previously (25) and was utilized for all experiments. An in vitro hollow fiber infection model (HFIM) was utilized, as previously described (26), to evaluate the effect of selected vancomycin dosing regimens on the change in bacterial burden and amplification of resistance in MRSA over 240 h (unpublished data). The HFIM employed a C3008 cellulosic cartridge (FiberCell Systems, Frederick, MD). Each vancomycin dosing regimen was administered in a manner to achieve the same area under the curve (AUC) over 10 days as would be expected in humans. Overnight cultures of MRSA isolates were taken to generate a high starting inoculum of over 10⁸ CFU/ml. Samples were taken from the HFIM at multiple time points throughout the 10-day experiment at days 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10 to assess growth of viable bacteria, incubated for 24 h, and quantified by colony count. To simulate human antibiotic dosing, the following vancomycin regimens were administered in the HFIM, assuming a free (f) fraction of 50% and a 6-h half-life with the indicated pharmacokinetic parameters in plasma for maximal concentration (C_max), minimum concentration (C_min), and AUC: 500 mg every 12 h (fC_max of 10 mg/liter, fC_min of 2.5 mg/liter, and fAUC of 112.5), 1,000 mg every 12 h (fC_max of 20 mg/liter, fC_min of 5 mg/liter, and fAUC of 225), 2,000 mg every 12 h (fC_max of 40 mg/liter, fC_min of 10 mg/liter, and fAUC of 450), and 4,000 mg every 12 h (fC_max of 80 mg/liter, fC_min of 5 mg/liter, and fAUC of 900). Isolates recovered after exposure to different vancomycin dosing regimens were tested for vancomycin and daptomycin susceptibility using the standard Etest according to the manufacturer's instructions (BioMérieux, Marcy l'Etoile, France). An isolate was defined as vancomycin-intermediate S. aureus (VISA) if the vancomycin MIC was >2 to 8 mg/liter (1). The macro-Etest method, as a measure of glycopeptide resistance in all strains, was also performed according to the manufacturer's instructions (BioMérieux, Marcy l'Etoile, France).

DNA methods and molecular techniques.

Standard methods were used for DNA manipulation, plasmid extraction, and Sanger sequencing. Bacterial genomic DNA was extracted using a GenElute kit according to the manufacturer's instructions (Sigma-Aldrich). The walKR region was amplified using PCR with primers 1901 and 1908 (see Table S1 in the supplemental material). DNA sequencing of the resulting amplicons was performed using standard Sanger sequencing with the primers 1901, 1903, 1906, and 1908 (see Table S1). Where IS256 integration into the walKR 5′ UTR was detected, the sequencing primer IS256rev was also used (see Table S1). Taq DNA polymerase (Roche) in reactions of 32 cycles was used for PCR. All primers used in this study are listed in Table S1 in the supplemental material.

Preparation of total RNA.

Total RNA was extracted with an RNeasy microkit (Qiagen) as previously described (27) using bacterial cultures that had entered exponential growth phase (optical density at 600 nm [OD₆₀₀] = 0.4 to 0.6). Two on-column DNA digestions were performed using RNase-free DNase (Qiagen) to remove genomic DNA. RNA concentration was determined spectrophotometrically, and RNA quality was determined by analysis of the A_260/280 ratio and by electrophoresing and visualizing a 5-μl aliquot through a 1% agarose gel.

qRT-PCR experiments.

Quantitative RT-PCR (qRT-PCR) experiments were performed using the oligonucleotides listed in Table S1 in the supplemental material. RNA was prepared from exponential-phase cultures (OD₆₀₀ = 0.5) as described above and previously (25). cDNA synthesis using SuperScript II RNase H reverse transcriptase (Invitrogen) was performed as described above and included a SuperScript II negative control. Relative expression levels were determined using the ΔΔC_T method as previously described (25) and were normalized against gyrB levels (28). For chromosomal gene expression analysis, results for the IS256 mutants and JKD6009 were compared, while for GFP expression analysis, results were compared to those for the control strain TPS3313, which contained only the T7 terminator sequence upstream of the GFP gene. Results were obtained from three biological repeats, each performed in technical triplicates, except for GFP expression experiments, which used 2 biological repeats performed in technical triplicates. Water-only and reverse transcriptase-minus DNA contamination controls were used in each experiment. Results were expressed as mean fold change (log₂) of mRNA compared to that for JKD6009 or empty vector and analyzed using GraphPad Prism (v 5.0d) by a one-sample t test, testing if the means were significantly different from 1.0 (i.e., significantly different from expression in JKD6009 or empty vector). P values of <0.01 were considered significant.

PE experiments.

The primer extension (PE) protocol used was modified from that described by Lloyd and colleagues (29) and used 10 to 20 μg of total bacterial RNA. Three (JKD6009) or four (BPH0391 and BPH0397) biological replicates were performed. To the RNA-primer mix, 6 μl of 5× first-strand buffer (Invitrogen), 15 mM dithiothreitol (DTT) (Invitrogen), 1 mM deoxynucleoside triphosphate (dNTPs) (Promega), 1 U RNasin (Promega), and 100 U of Superscript II RNase H⁻ reverse transcriptase (Invitrogen) were added. After 1 h of incubation at 42°C, 2 μl of 5× first-strand buffer, 1.5 mM dNTPs, 1 U RNasin, 15 mM DTT and 100 U of Superscript II were added and incubated for a further hour at 42°C. Ten nanograms of RNase A (Sigma) was then added, and the mixture was incubated at 37°C for 30 min. The resultant cDNA was precipitated and washed once with 70% ethanol, dried, and stored at −20°C until analysis. Capillary electrophoresis was performed on an Applied Biosystems 3730 DNA analyzer using Liz 500 size standards to generate a standard curve (Applied Biosystems). Genemapper version 3.7 (Applied Biosystems) was used to analyze the sample files, with automated allele calling verified by manual inspection. The sized cDNA fragments were then mapped to their respective first-strand synthesis primer binding sites to identify the putative transcription start sites (TSS). Consensus S. aureus SigA −10 and −35 sequences were taken from 28 published S. aureus promoters and displayed using WebLogo3 (30) (see Table S2 in the supplemental material).

GFP reporter plasmid construction.

Plasmid pALC2084 is an Escherichia coli-S. aureus shuttle vector that contains a tetracycline-inducible GFP_UVR gene (31). There is a unique KpnI site between the tetracycline-inducible promoter and the GFP gene, and this site was used to clone walKR 5′-UTR regions from JKD6009, BPH0397, and BPH0391 that were likely to harbor putative promoter elements as defined by primer extension analysis. A common reverse primer (RHSall [see Table S1 in the supplemental material]) containing a 5′ KpnI site was used to amplify each walKR 5′ UTR in conjunction with various unique forward primers that contained a 5′ KpnI site followed by a 47-bp T7 termination sequence followed by a 3′ walKR 5′-UTR-specific binding region (see Table S1 in the supplemental material). 5′-UTR amplicons and pALC2084 were digested with KpnI (New England BioLabs). The plasmid was additionally treated with shrimp alkaline phosphatase (Affymetrix/USB). The 5′-UTR amplicons were ligated into pALC2084 using T4 DNA ligase (New England BioLabs), and the ligated construct was used to transform NEB-5α competent E. coli cells (New England BioLabs). Transformants were selected on LB-ampicillin (100 μg/ml) plates, and single colonies were screened by PCR using the pALCF/RHSall primer pairs to select for colonies containing pALC2084 with the correctly orientated walKR 5′-UTR insert. Plasmids were extracted from NEB-5α cells using a QIAprep Spin Miniprep kit (Qiagen), transformed into S. aureus strain RN4220, and selected on heart infusion (HI)-chloramphenicol (12.5 μg/ml) plates. A control vector containing only the T7 terminator sequence was constructed by hybridizing oligonucleotides corresponding to the T7 terminator sequence flanked by KpnI sites. Following KpnI digestion, the double-stranded T7 terminator sequence was cloned into the KpnI site of pALC2084 as described above. Splice overlap extension PCR (32) was used to alter the sequence of the putative JKD6009 promoter motif at TSS bp −145 with oligonucleotides 6009.SO.F and 6009.SO.R (see Table S1 in the supplemental material). Each PCR consisted of 20 cycles of 94°C for 1 min, 50°C for 1 min, and 72°C for 3 min and then 94°C for 1 min, 72°C for 10 min, and holding at 4°C. Two overlapping PCR products were obtained, and 2 μl (∼50 ng DNA) from each was used in a subsequent reaction using the outermost primers for each product to yield a full-length fragment incorporating the desired mutation. Each product was then ligated into pALC2084 as described above. To alter the predicted −10 element of the JKD6009 bp −23 TSS, an oligonucleotide was designed to be used in place of the RHSall primer (83SDM [see Table S1 in the supplemental material]). This oligonucleotide was used in a PCR in conjunction with either of the forward primers. All constructs were confirmed correct by DNA sequencing using the pALCF primer.

GFP reporter assays.

GFP expression in pALC2084 derivatives was measured using a FLUOstar Optima plate scanner (BMG Lab Technologies). One hundred microliters of overnight cultures of plasmid-transformed S. aureus RN4220 strains grown in heart infusion (HI) broth supplemented with 12.5 μg/ml chloramphenicol was used to inoculate 10 ml fresh HI-chloramphenicol medium. Two-hundred-microliter aliquots of this mixture were added in triplicate to wells of a 96-well flat-bottomed clear plate. Plates were incubated at 37°C for 10 h. Each well was scanned using an excitation filter of 485 nm and an emission filter of 520 nm. Fluorescence readings were taken every 10 min, and the average of 20 flashes per well was taken to be the measure of fluorescence. Prior to each reading, the plates were shaken for 5 min in an orbital motion. Replicates were averaged for each experiment, and the average value for the strain containing the T7 terminator control plasmid (empty vector control, TPS3313) was taken as background. This value was used to normalize all experimental samples. For each fluorescence analysis experiment, an identical parallel experiment that assayed the OD₆₀₀ of each well was performed. This was done in order to correlate fluorescence emission data with bacterial growth phase. Differences in GFP fluorescence between the strains were not evident until early stationary phase and were still increasing at the completion of the 10-h experiments, presumably because of the lag time between GFP gene transcription and the production of functional fluorescent protein (33). Fluorescence readings were therefore taken at the 10-h time point. Results were expressed as the mean and range for the biological replicates.

Whole-genome sequencing.

Whole-genome sequencing of S. aureus BPH0391, BPH0391R, and BPH0397 isolates was achieved using Ion Torrent semiconductor sequencing on the PGM platform. Each genome was sequenced following the manufacturer's instructions (Life Technologies) with a single 316 chip with 100-bp or 200-bp chemistry, yielding 147.65 to 699.56 Mbp of Phred score quality of >Q20. The resulting sequence reads for these isolates and also JKD6009 (previously sequenced [9, 34]) were aligned using Nesoni (Victorian Bioinformatics Consortium; http://vicbioinformatics.com/) to the S. aureus JKD6008 reference genome (accession no. CP002120.1) (34). JKD6008 is a clinical VISA derivative of JKD6009 (25). A genome-wide variant analysis was performed using Nesoni, and allelic variability at any nucleotide position in BPH0391, BPH0391R, and BPH0397 relative to JKD6009 was tallied to generate a list of differences for each genome. The position of IS256 within JKD6009 was identified by mapping all paired-end Illumina reads to a reference IS256 sequence and then remapping any unmapped reads from a read pair back to the JKD6008 reference genome. This was done twice, once for the 5′ end of IS256 and again for the 3′ end. In a mapping plot (viewed in Artemis), the reads that mapped immediately adjacent to the location of the known IS256 copies presented as a coverage spike at both ends of the genomic region in which they had inserted and as a single coverage spike for novel IS256 insertions relative to the reference. The same procedure was applied for the single-end Ion Torrent reads for the isolates BPH0391, BPH0391R, and BHP0397 except that a custom utility was written to split any read that straddled both the end of the IS256 and some flanking sequence. This segment of the read from the IS256 flanking sequence was then mapped back to the reference to locate the IS256 positions.

Selection of walKR IS256 wild-type revertant.

To identify S. aureus that had lost the IS256 insertion in the walR 5′ UTR, a single colony of BPH0391 was inoculated into 5 ml of HI broth (Oxoid) within a 50-ml tube and incubated at 37°C for 18 h with agitation at 250 rpm. Fifty microliters of this broth was inoculated into another 5 ml of HI broth and incubated again. This passaging was repeated a further two times (four passages in total). After the final passage, 3 μl of the culture was streaked onto 6 horse blood agar plates (Oxoid) and incubated at 37°C for 18 h, and 240 of the resulting colonies were replica patched onto HI agar plates containing 2 μg/ml vancomycin (Van-2) and HBA plates without vancomycin. Colonies that grew on HBA but not on Van-2 were screened by PCR with primers 1901 and 1908 (see Table S1 in the supplemental material) for loss of IS256 in the walR 5′ UTR.

Sequence read accession number.

Sequence reads have been deposited in the NCBI Sequence Read Archive under submission number SRA059342.

RESULTS

In vitro evolution of VISA.

After exposure to simulated human exposures of vancomycin ranging from 500 mg to 4,000 mg administered every 12 h, five isolates were recovered on day 1 and five isolates were recovered on day 10 from the parental strain JKD6009 using HFIM (Tables 1 and 2). The details of bacterial killing, population analysis profiles, and vancomycin pharmacodynamics are to be reported elsewhere. At the completion of the experiments, two strains demonstrated stable colony heterogeneity, identified as differences in colony pigmentation (white [W] or gray [G]) (Tables 1 and 2). The white and gray colony types were stable on subculture and were investigated independently. Six strains were VISA. Two strains also demonstrated nonsusceptibility to daptomycin.

Table 1.

Plasmids and bacterial strains used in this study

Plasmid or strain	Description	Reference or source
Plasmids
pALC2084	Tet-inducible, GFP expression vector	31
pTPS3283	Full-length 330-bp walR promoter region from JKD6009 cloned into KpnI site of pALC2084	This study
pTPS3284	5′-truncated walR promoter region from JKD6009 (excludes potential TSS at −244) cloned into KpnI site of pALC2084	This study
pTPS3285	5′-truncated walR promoter region from JKD6009 (excludes potential TSS at −244 and −145) cloned into KpnI site of pALC2084	This study
pTPS3286	Full-length 330-bp walR promoter region from BPH0397 cloned into KpnI site of pALC2084	This study
pTPS3287	5′-truncated walR promoter region from BPH0397 (excludes potential TSS at −144) cloned into KpnI site of pALC2084	This study
pTPS3288	Full-length 330-bp walR promoter region from BPH0391 cloned into KpnI site of pALC2084	This study
pTPS3290	Empty vector control, pALC2084 containing only the T7 terminator sequence	This study
pTPS3306	Plasmid pTPS3284 with site-directed mutation of the putative −10 promoter, upstream of the −144 TSS	This study
pTPS3307	Plasmid pTPS3283 with site-directed mutation of the putative −10 promoter, upstream of the −144 TSS	This study
pTPS3311	Plasmid pTPS3285 with site-directed mutation of the putative −10 promoter, upstream of the −23 TSS	This study
pTPS3312	Plasmid pTPS3283 with site-directed mutation of the putative −10 promoters, upstream of the −23 and −144 TSS	This study
E. coli DH5α	Highly transformable strain	NEB
S. aureus strains
RN4220	Strain capable of maintaining shuttle plasmids	46
JKD6009	Parental MRSA, VSSA clinical isolate	10, 25
BPH0395	VSSA strain derived in vitro from JKD6009	This study
BPH0397	VISA strain derived in vitro from JKD6009	This study
BPH0398	VISA strain derived in vitro from JKD6009	This study
BPH0393(W)	VSSA strain derived in vitro from JKD6009	This study
BPH0393(G)	VSSA strain derived in vitro from JKD6009	This study
BPH0391	VISA strain derived in vitro from JKD6009	This study
BPH0391R	VSSA strain derived in vitro from BPH0391	This study
BPH0399	VISA strain derived in vitro from JKD6009	This study
BPH0396(W)	VISA strain derived in vitro from JKD6009	This study
BPH0396(G)	VISA strain derived in vitro from JKD6009	This study
BPH0394	VSSA strain derived in vitro from JKD6009	This study
TPS3283	RN4220 containing pTPS3283	This study
TPS3284	RN4220 containing pTPS3284	This study
TPS3285	RN4220 containing pTPS3285	This study
TPS3286	RN4220 containing pTPS3286	This study
TPS3287	RN4220 containing pTPS3287	This study
TPS3288	RN4220 containing pTPS3288	This study
TPS3290	RN4220 containing pTPS3290	This study
TPS3306	RN4220 containing pTPS3304	This study
TPS3307	RN4220 containing pTPS3305	This study
TPS3311	RN4220 containing pTPS3311	This study
TPS3312	RN4220 containing pTPS3312	This study
TPS3313	RN4220 containing pTPS3290	This study

Table 2.

Resistance levels and walKR mutation details of 10 S. aureus strains recovered after in vitro exposure to vancomycin

Straina	Exposure profile (time [h], vancomycin [g])	Phenotypeb	MIC (μg/ml)				walKR mutation
			Vancomycin (Etest)	Vancomycin (macro-Etest)	Teicoplanin (macro-Etest)	Daptomycin (Etest)
JKD6009	None	VSSA, DS	1.5	4	6	0.19
BPH0395	240, 0.5	VSSA, DS	1.5	4	6	0.125	None detected
BPH0397	240, 1	VISA, DNS	4	8	8	1.5	IS256 insertion in walKR 5′ UTR at position −38c
BPH0398	240, 2	VISA, DS	4	8	12	1	walK CAA deletion, position 1111–1113, ΔQ371
BPH0393(W)	240, 4	VSSA, DS	1.5	4	6	0.125	None detected
BPH0393(G)	240, 4	VSSA, DS	1	4	4	0.094	None detected
BPH0391	24, 0.5	VISA, DS	3	8	12	1	IS256 insertion in walKR 5′ UTR at position −50
BPH0399	24, 1	VISA, DS	3	6	8	0.38	None detected
BPH0396(W)	24, 2	VISA, DNS	3	8	12	2	walK CAA deletion, position 1111–1113, ΔQ371
BPH0396(G)	24, 2	VISA, DS	4	8	24	0.125	None detected
BPH0394	24, 4	VSSA, DS	1.5	4	6	0.19	None detected
BPH0391R		VSSA, DS	1.5	4	6	0.19

^aStrains BPH0393 and BPH0396 were heterogeneous populations and comprised both white (W) and gray (G) colonies.

^bVSSA, vancomycin-susceptible S. aureus; VISA, vancomycin-intermediate S. aureus; DS, daptomycin susceptible; DNS, daptomycin nonsusceptible.

^cIS256 in reverse orientation compared to that seen in BPH0391 and previously reported (24).

DNA sequence analysis of the walKR locus in VISA mutants.

Previous studies have found a high proportion of walKR mutations in VISA strains (9, 13, 20, 35). We therefore wished to determine the mutational status of this locus in our 10 in vitro-derived strains. PCR of each isolate using the 1901/1908 primer pair (see Table S1 in the supplemental material) and subsequent DNA sequence analysis of the PCR amplicon revealed that four of the 10 in vitro-derived isolates had walKR mutations, while six isolates had no walKR mutations. BPH0398 and BPH0396(W), which both tested as VISA, possessed identical 3-bp deletions in walK, which are predicted to result in the deletion of a single amino acid (ΔQ371) (Table 2). This mutation occurs in a short CAA repeat and has previously been described in another laboratory MRSA strain (13). VISA mutants BPH0391 and BPH0397 produced PCR amplicons that were approximately 1.3 kb larger than expected (Fig. 1). Sequence analysis identified the presence of IS256 in the walKR 5′ UTRs of these strains (Fig. 2A to C). S. aureus strain BPH0391 contained IS256 at position −50 from the walR ATG start codon, while strain BPH0397 contained IS256 in the reverse orientation at position −38 (Fig. 2B and C).

Fig 1.

PCR amplification of the walKR region in 10 S. aureus strains recovered after in vitro exposure to vancomycin. The entire walKR operon was sequenced for mutation detection. Two strains (BPH0391 and BPH0397) showed PCR amplicon sizes approximately 1.3 kb larger than the predicted size of 2.7 kb, caused by the insertion of IS256 within the walR 5′ UTR.

Fig 2.

Identification of TSS and promoters in the 5′ UTRs of JKD6009, BPH0397, and BPH0391. (A) walR 5′-UTR sequence of JKD6009. Predicted TSS are shown with red nucleotides, and potential −10 and −35 elements are boxed in red. Nucleotides altered by SDM are shown in bold above the −10 boxes. The walR sequence is highlighted in brown, with the ATG start codon in bold. Locations of IS256 integrations are indicated by vertical arrows: BPH0397, blue; BPH0391, red; and SA137/93A (24), green. The location of the primer used for PE analysis is shown under the walR sequence. (B and C) Similar analyses for the BPH0391 (B) and BPH0397 (C) walR 5′-UTR sequences. The IS256 sequence is indicated by orange highlighting. (D) Primer extension results for JKD6009 and BPH0397. JKD6009 consistently showed three clear peaks representing TSS at positions −23 (83-bp peak), −145 (204.5-bp peak), and −244 (304-bp peak) from the walR ATG start codon (the y axis shows arbitrary fluorescence units, and the x axis shows size [sz] in bp, height [ht], and area under the curve [ar]). Results for BPH0397 showed low reproducibility. This example shows peaks corresponding to the predicted TSS at positions −23 (83-bp peak) and −144 (204-bp peak). These were the only peaks consistently observed over 4 separate experiments. Results for BPH0391 also showed low reproducibility, with only a predicted TSS at position −23 consistently observed (data not shown). (E) Consensus S. aureus SigA −10 and −35 sequences taken from 28 published S. aureus promoters and displayed using WebLogo3 (30) (see Table S2 in the supplemental material).

Comparative genomics of BPH0391, BPH0397, and wild-type JKD6009.

We next sequenced the genomes of BPH0391 and BPH0397 to detect any additional mutations (SNPs, indels, or additional IS256 insertions) that might have occurred during HFIM vancomycin exposure. As shown in Table 3, BPH0391 had one additional mutation, while BPH0397 had three. The BPH0391 mutation was am SNP predicted to cause an S163L amino acid change in the gene coding for pseudouridine-5-phosphate glycosidase (Table 3). We could find no obvious link between its role in catalyzing the conversion of pseudouridine 5-phosphate to d-ribose 5-phosphate and uracil and a VISA phenotype, so we assumed that the mutation was neutral with regard to vancomycin susceptibility. Of the three mutations found in BPH0397, the first was an intergenic SNP between two conserved hypothetical genes of unknown function. The second was a single-nucleotide deletion in treP (which is a pseudogene in JKD6009 due to a single-nucleotide insertion) that is predicted to return the gene to its correct reading frame. In order to rule out the possibility of this mutation being VISA associated, we sequenced this gene from 20 single colonies of JKD6009. Seven of these colonies had identical or similar mutations that were all predicted to return the gene to its correct reading frame, and we therefore concluded that the restoration of treP function in BPH0397 was not VISA associated. The final mutation identified in BPH0397 was a novel IS256 insertion within agrC (encoding the system's sensor kinase) that is likely to inactivate agr function. The limited genome sequence variation in BPH0391 and BPH0397 indicated that the IS256 walR 5′-UTR insertion in these mutants was likely the cause of the VISA phenotype.

Table 3.

Genome mutations found in strains BPH0391 and BPH0397 compared to parental strain JKD6009

Strain	Mutation (JKD6008 chromosome position)	Locus tag (JKD6008)	Protein product	Effect of mutation
BPH0391	IS256 insertion in walR 5′ UTR (24338)	Intergenic		Decrease in walKR transcription
	T to C (332823)	SAA6008_00288	Pseudouridine-5′-phosphate glycosidase	S163L amino acid change
BPH0391R	T to C (332823)	SAA6008_00288	Pseudouridine-5′-phosphate glycosidase	S163L amino acid change
BPH0397	IS256 insertion in walR 5′ UTR (24350)	Intergenic		Decrease in walKR transcription
	G to A (420152)	Intergenic
	G deletion (517326)	treP SAA6008_00478	Phosphotransferase system trehalose-specific IIBC component	Corrects frameshift mutation
	IS256 insertion within agrC (2178434)	agrC	Agr sensor kinase	Inactivation of agr system

Loss of IS256 in BPH0391 leads to reversion from VISA to VSSA.

An IS256 insertion upstream of walR has been previously associated with VISA (24). However, to prove that the IS256 insertion within the walR 5′ UTR observed here caused the VISA phenotype, we subjected S. aureus BPH0391 to serial passaging in the absence of vancomycin and, by replica plating, identified colonies that returned to vancomycin susceptibility (see Materials and Methods). PCR screening with the 1901/1908 primer pair (see Table S1 in the supplemental material) revealed one colony that had lost IS256. This isolate (BPH0391R) was subjected to MIC testing, and macro-Etest showed full reversion to the JKD6009 VSSA MIC (Table 2). Whole-genome sequencing of BPH0391R confirmed that there were no additional genome changes compared to BPH0391, with the BPH0391-specific SNP at nucleotide position 332823 retained in the revertant (Table 3). These sequence data confirmed that BPH0391R is an isogenic version of BPH0391, lacking only the walR 5′-UTR IS256 insertion, and prove that this insertion caused the VISA phenotype.

Transcriptional analysis of walKR and walKR-controlled genes.

We next sought to test the impact of IS256 on walKR expression in BPH0391 and BPH0397. We used qRT-PCR and measured walR and walK gene expression in each mutant, BPH0391R, and JKD6009. Expression of walR and walK in both BPH0391 and BPH0397 was 2-fold downregulated compared with that in JKD6009 but was restored to wild-type levels in BPH0391R (Fig. 3). We then tested the expression of three genes encoding autolysins that have been previously shown to be WalKR regulated (ssaA) or to have putative WalR upstream binding sites (sak and sceD) (36). Expression of ssaA was 3- to 4-fold upregulated in the mutants, suggesting that it is negatively regulated by walKR. There was no significant change in expression of sak or sceD for either mutant, indicating that the reduction in vancomycin susceptibility did involve these loci (Fig. 3).

Fig 3.

Quantitative RT-PCR results for expression of mRNAs from walR, walK, sak, ssaA, sceD, and RNAIII in VISA strains BPH0391 and BPH0397 and VSSA strain BPH0391R compared to the VSSA parent JKD6009. Results are expressed as fold changes in mRNA expression relative to that in the parental strain JKD6009 and depicted are the mean and standard error of the mean obtained from at least three biological repeats. Expression levels marked with an asterisk were significantly different from that for JKD6009 (P < 0.01).

The inactivation of AgrC by the second IS256 insertion in BPH0397 prompted us to also measure levels of the effector molecule RNAIII in all strains. As predicted, RNAIII levels were dramatically reduced in BPH0397 (Fig. 3). Interestingly, BPH0391, which has no agr mutation, also exhibited reduced RNAIII expression (approximately 4-fold compared to JKD6009). This expression difference was not restored in BPH0391R, indicating that the change in RNAIII expression was not caused by the IS256 walR insertion.

Characterization of walKR promoters and gene expression in the wild-type and IS256 mutants.

We used primer extension (PE) analysis to identify transcription start sites (TSS) in JKD6009 and the two IS256 mutants (Fig. 2D). Three potential TSS at positions −23, −145, and −244 from the ATG initiation start codon were identified for JKD6009 (Fig. 2A and D). To identify likely promoter elements, we established consensus SigA −10 and −35 binding motifs (position relative to the TSS) from a collection of 28 published S. aureus SigA-dependent promoters (Fig. 2E; see Table S2 in the supplemental material). Inspection of the walKR 5′-UTR sequence revealed appropriately spaced potential promoter elements upstream of each TSS (Fig. 2A). PE was also performed using RNA extracted from the two IS256 mutants. However, the results were less reproducible, suggesting reduced walR expression compared to that in JKD6009 and in line with qRT-PCR analysis of walR (Fig. 3). Nevertheless, from repeated PE experiments (four biological repeats), BPH0397 consistently displayed a peak at position −144, while both BPH0397 and BPH0391 intermittently displayed a peak at position −23 (Fig. 2B and C). Based on these TSS predictions, potential −10 and −35 SigA promoter motifs were identified, suggesting that both IS256 mutants possessed hybrid promoters corresponding to the −23 TSS, where the wild-type −35 motif had been replaced by an IS256 sequence. This sequence was different for each mutant because of the differences in IS256 location and orientation (Fig. 2B and C).

To confirm the locations and strengths of the predicted promoters in JKD6009 and the IS256 mutants, we cloned 330-bp fragments of the walKR 5′ UTRs from JKD6009, BPH0397, and BPH0391 into pALC2084, immediately upstream of the GFP_UVR reporter gene (Fig. 4A). We then transformed S. aureus RN4220 with these expression constructs and compared GFP mRNA and protein expression levels with those for RN4220 containing empty vector (Fig. 4). The highest GFP transcriptional expression was observed for the construct harboring the wild-type walKR upstream sequence (TPS3283), with mRNA levels 12-fold above background. In contrast, GFP expression from the IS256 constructs (TPS3286 and TPS3288) was only 2- to 4-fold above background (Fig. 4B). Fluorescence measurements were not as sensitive as qRT-PCR, but they also showed the same significant pattern of difference between the wild type and the mutants (Fig. 4C). These data confirm that the IS256 insertions have reduced expression of walKR.

Fig 4.

GFP reporter assay construct details and GFP transcript and protein expression results. (A) Wild-type and IS256 mutant walR 5′ UTRs were cloned into pALC2084 upstream of the GFP reporter gene. 5′ UTRs spanned potential TSS as predicted from PE analysis. Stars indicate predicted TSS. Dotted lines represent IS256 sequence. Shaded regions represent T7 transcriptional terminator sequence. Crosses through the constructs represent positions within the predicted −10 boxes upstream of the −23 or −145 TSS that were altered by site-directed mutagenesis. (B and C) Analysis of GFP gene transcription (B) and GFP fluorescence (C) in S. aureus RN4220 strains containing the corresponding plasmid constructs from panel A. Results are presented relative to GFP expression from strain TPS3313, containing only the T7 terminator upstream of the GFP gene, which has been normalized to a value of one. All results are expressed as the mean and the range for two (GFP mRNA measurement) or three (GFP fluorescence) biological replicates.

To delineate the DNA regions harboring the observed promoter activity, we prepared two truncated versions of the full-length wild-type sequence, which excluded the −244 TSS (TPS3284) or both the −244 and −145 TSS (TPS3285) (Fig. 4A). Expression testing showed that full promoter activity was attributable to the region harboring the −145 and −23 TSS, while the −244 region conferred no additional activity (Fig. 4B and C). Site-directed mutations were constructed to confirm promoter predictions within these two regions exhibiting expression activity. Mutation of the −10 box for the TSS at −145 (TPS3307) reduced expression to half of wild-type levels, while mutation of the −10 boxes at both −145 and −23 (TPS3312) reduced expression to background, suggesting that active promoters were present (and had been correctly identified) in these regions (Fig. 2 and Fig. 4).

A truncated version of the IS256 mutant BPH0397 with the putative −144 TSS and promoter region removed (TPS3287) was also tested but showed no significant decrease in expression, although the full-length construct (TPS3286) already displayed very low expression levels that were only 2-fold higher than background (Fig. 4).

DISCUSSION

Recent studies have highlighted the significant impact that mutations within the essential TCS walKR can have on the vancomycin susceptibility of S. aureus, with 50 to 60% of clinical VISA strains possessing walKR coding-region mutations (9, 13, 20). In our analysis of 10 laboratory isolates derived from the same VSSA parent under extensive selective pressure after exposure to simulated human dosing regimens of vancomycin for up to 10 days, four of the six strains demonstrating the VISA phenotype were found to possess walKR mutations. Four of these strains showed an increase in daptomycin MIC, with two becoming daptomycin nonsusceptible, and all four contained a walKR nucleotide mutation or promoter sequence IS256 insertion. It appears (with the caveat that small sample numbers were involved) that the mutational spectrum seen in these in vitro-derived strains differs from that in clinical isolates. For example, we have previously described the clinical isogenic VSSA pair JKD6009 (used as the parental strain in this study) and JKD6008, which has a G223D mutation within walK (9). This mutation was not observed in any of the in vitro-derived VISA strains. Interestingly, the single coding sequence mutation that was observed (Δ371Q in walK) arose independently on two occasions and has previously been observed in another in vitro-derived VISA strain (13). However, this mutation has not been observed in over 50 clinical VISA strains found to harbor walKR mutations (9, 13, 20), suggesting that it may be specific to laboratory-derived strains, where the primary selection pressure is the antibiotic and other stressors such as host immune responses are absent.

We have also shown that two in vitro-derived VISA strains (BPH0391 and BPH0397) acquired independent IS256 insertions that led to reduced walKR expression. Furthermore, the identification and genomic characterization of a VSSA revertant of BPH0391 confirmed that the IS256 insertion within the walKR 5′ UTR was also the cause of the VISA phenotype. A recent report of a S. aureus InsTet^G+2^Cm transposon mutant library also described hVISA/VISA mutants with independent walKR 5′-UTR insertions (37). These findings suggest that altered expression of wild-type WalKR proteins is another relatively common mechanism for VISA formation, as it has previously also been reported in the VISA strain SA137/93A (24). However, in this previous study, walKR expression was reported to be upregulated in the IS256 walKR 5′-UTR mutant. The contrasting findings of our studies may be due to the different hybrid promoters formed from the different sites of insertion, i.e., position −58 for SA137/93A (relative to the WalR start codon) compared with positions −50 and −38 for BPH0391 and BPH0397, respectively (Fig. 2A). Although it might be possible for walKR upregulation to induce VISA, with over 40 distinct VISA-associated walKR mutations now described (9, 13, 20), it seems unlikely that many of these could represent gain-of-function changes. Further evidence against walKR gain-of-function mutations causing VISA has emerged from a recent study which found that constant WalR activation failed to produce a VISA phenotype (38). We were unsuccessful in our attempts to obtain S. aureus strain SA137/93A to test in our reporter assays, but more comparative investigation of these IS256 walKR 5′-UTR mutants could help our understanding of the role of walKR in VISA.

The primer extension and GFP reporter experiments that we describe here have provided the first detailed analysis of walKR promoter structure. Two TSS, located −23 and −145 bp from the walR ATG start codon, were identified, and mutation of the putative −10 boxes (relative to the TSS) for these SigA-like promoter sequences resulted in GFP gene downregulation and reduced cellular fluorescence. A third potential TSS, located 244 bp upstream from the walR ATG start codon, was also investigated. However, the sequence in this region had no impact on gene expression in our reporter experiments (Fig. 4). Neither of the SigA promoters identified in this study match that predicted in earlier work using consensus −10 and −35 sequences recognized by SigA in Bacillus subtilis (39), and we note that the sequences identified in that study are polymorphic across various S. aureus strains (data not shown). Overall, we present evidence for a walKR 5′ UTR that contains at least two SigA promoters that would allow for complex regulation.

Primer extension analysis of both the IS256 mutants also showed a TSS located 23 bp upstream of the walR start codon (Fig. 2D). In both of these mutants, the −10 sequences for this TSS remain unchanged but the predicted −35 elements are located within IS256 (Fig. 2B and C). These hybrid promoters presumably are able to produce sufficient quantities of WalKR for cell viability, but both our plasmid-based and native chromosome qRT-PCR experiments show that these elements are significantly less active than the wild-type promoter located at position −23 (Fig. 2 and 4). An additional TSS located at position −144 was detected in BPH0397, and while its presence did increase GFP gene transcription and protein production, the effect was modest. The most significant cause of reduced walKR transcription appears to be the loss of the second wild-type TSS located at position −145, which has been replaced by IS256 in both mutants.

WalR has been shown to bind to the promoter regions and directly regulate genes encoding certain S. aureus autolysins, including ssaA, lytM, and isaA. Other genes that are also important for peptidoglycan turnover and cell wall metabolism, such as sceD and sak, are positively regulated by WalKR and are predicted to harbor WalR recognition sequences in their 5′ UTRs (36, 40). We therefore compared by qRT-PCR the expression of sak, ssaA, and sceD in BPH0391 and BPH0397 against that in JKD6009. There was a significant increase in ssaA transcription in both mutants that was restored to the wild-type level in the revertant (BPH0391R), implying that in JKD6009, ssaA is negatively regulated by walKR. In a recent study that involved uncoupling the expression of several cell wall metabolism-associated genes from their walKR-dependent regulation, ssaA was one of only two genes (the other being the glycyl-glycyl endopeptidase gene lytM) that when overproduced was capable of compensating for reduced walKR expression with respect to cell viability (21). The finding of increased ssaA transcription in our own walKR-depleted strains is therefore not surprising and suggests that it may be necessary for normal bacterial growth. However, there was no change in expression of sak and sceD, which is in contrast to previous reports. The upregulation of sceD is a common phenomenon in VISA strains (10, 41, 42). Indeed, it has been proposed as a potential biomarker for VISA due to its consistent statistically significant level of induction (43). These observations, and other studies that have looked at the transcriptome in WalKR mutants, point to a broader role for this essential regulator beyond peptidoglycan metabolism (9, 21). Further research is required to better define the WalR regulon and thus understand the molecular basis for a VISA phenotype that is induced through WalKR perturbation.

In an effort to further clarify the genetic basis for the observed phenotypes of both IS256 mutants, we performed whole-genome sequencing on BPH0391 and BPH0397 and compared them to the previously sequenced parental strain JKD6009 (Table 3). Mutational differences were limited, and while we cannot rule out the impact of any of them on VISA development, the only non-walKR mutation of obvious interest was the integration of IS256 into the agrC gene of BPH0397. The impact of this mutation was assessed using qRT-PCR targeting the agr effector molecule RNAIII. As predicted, both mutants demonstrated a reduction in agr activity, with BPH0391 RNAIII expression levels shown to be approximately 4-fold less than those of the parent strain. The impact of the agr mutation in BPH0397 was predictably substantial, with RNAIII expression levels reduced >12-fold (Fig. 3). Downregulation of agr is commonly reported in VISA strains lacking agr mutations (10, 24, 41, 44), a finding that presumably accounts for the lower virulence reported for these strains (9, 45). Interestingly, RNAIII expression was not restored to wild-type levels in BPH0391R, suggesting that other differences in BPH0391 compared to JKD6009, such as the predicted S168L amino acid substitution in SAA6008_00288, might have caused this change.

Our analysis of BPH0391 and BPH0397 indicates the important role played by IS256 in shaping S. aureus and shows that even among nearly isogenic strains, a VISA phenotype can emerge through different combination of mutations that arise over very short time scales. These mutants will also be useful tools to help address the more fundamental question of the role of WalKR in the physiology of S. aureus.