The δ Subunit of RNA Polymerase Guides Promoter Selectivity and Virulence in Staphylococcus aureus

Abstract

In Gram-positive bacteria, and particularly the Firmicutes, the DNA-dependent RNA polymerase (RNAP) complex contains an additional subunit, termed the δ factor, or RpoE. This enigmatic protein has been studied for more than 30 years for various organisms, but its function is still not well understood. In this study, we investigated its role in the major human pathogen Staphylococcus aureus. We showed conservation of important structural regions of RpoE in S. aureus and other species and demonstrated binding to core RNAP that is mediated by the β and/or β′ subunits. To identify the impact of the δ subunit on transcription, we performed transcriptome sequencing (RNA-seq) analysis and observed 191 differentially expressed genes in the rpoE mutant. Ontological analysis revealed, quite strikingly, that many of the downregulated genes were known virulence factors, while several mobile genetic elements (SaPI5 and prophage ϕSA3usa) were strongly upregulated. Phenotypically, the rpoE mutant had decreased accumulation and/or activity of a number of key virulence factors, including alpha toxin, secreted proteases, and Panton-Valentine leukocidin (PVL). We further observed significantly decreased survival of the mutant in whole human blood, increased phagocytosis by human leukocytes, and impaired virulence in a murine model of infection. Collectively, our results demonstrate that the δ subunit of RNAP is a critical component of the S. aureus transcription machinery and plays an important role during infection.

INTRODUCTION

Bacterial gene transcription is a complex, multifactorial process that involves several key enzymes and regulatory elements. It is driven by the activity of DNA-dependent RNA polymerase (RNAP) and its associated proteins, which form a multisubunit enzyme consisting of one β subunit, one β′ subunit, two identical α subunits, and one ω subunit (reviewed in reference 1). Together these form the RNAP apoenzyme, which is able to perform RNA elongation and termination; however, initiation requires the involvement of a σ factor. Typically, most bacterial species harbor several different σ factors, a primary one (σ^A or σ⁷⁰) that mediates housekeeping gene transcription and a variety of alternative sigma factors, which aid in the response to unfavorable environmental conditions and stress.

In certain Gram-positive species and particularly the Firmicutes (see Fig. S1 in the supplemental material), an additional RNAP subunit is present, termed the δ factor, or RpoE (2). In Bacillus subtilis, the 173-amino-acid delta subunit has been shown to reduce nonspecific binding of RNAP to DNA and lead to an elevated preference for DNA regions that include promoter sequences (3, 4). In early experiments, it was shown that RpoE has a role specifically confined to promoter selection and influences the ability of RNAP to form open promoter complexes (5, 6). Full-length RpoE has also been described as displacing nucleic acids from RNAP-DNA or RNAP-RNA complexes in vitro, which may explain the decreased affinity of RNAP-RpoE for nonpromoter regions. This feature of RpoE activity has additionally been attributed to its ability to stimulate RNAP recycling by enhancing the release of the complex from terminator sequences (6).

The effects of RpoE are far from arbitrary, since it has been shown that the δ subunit of RNAP not only distinguishes between promoter and nonpromoter sequences but also is important in the recognition of individual promoter features. A comparison of two ϕ29 promoters by transcription runoff revealed selective, promoter-dependent effects of RpoE (7). While the stronger of these two promoters was largely unaffected by RpoE, the weaker promoter showed significantly decreased expression in the presence of this factor. In order to explain the effect of RpoE on promoter selection, Rabatinova et al. (8) investigated the role of initiating nucleotide triphosphate (iNTP) recognition by RNAP and RNAP-RpoE complexes. iNTP is required to stabilize open complex formation and therefore to initiate transcription. It is thought that by destabilizing DNA-RNAP complexes, RpoE increases the amount of iNTP required to transcribe from certain promoters. Finally, in an approach aiming to identify functional regions of RpoE, Lopez de Saro et al. (9) created several truncated variants of the protein. In doing so, they showed that the N-terminal portion is required for binding to RNAP and also guides the correct orientation of the disordered and polyanionic C-terminal region. A limitation of these studies regarding RpoE function, however, is that they were performed in vitro, using only single promoters. Therefore, the true extent of δ-subunit influence on genome-wide transcriptional effects is still somewhat restricted at this point.

Phenotypically, rpoE mutants of B. subtilis have an extended lag phase upon subculturing of stationary-phase cells in fresh medium, as well as a moderately changed cell morphology (10). Furthermore, competition experiments revealed that when cocultured with the wild type, δ-factor-lacking strains show decreased fitness and are outcompeted during growth over several days (8). Finally, a role for RpoE in sporulation has been suggested, since rpoE loss was shown to suppress a mutation (pdhC) that negatively influences sporulation (11).

The role of RpoE has also been addressed for a small number of pathogenic organisms, most notably in streptococcal species. RpoE in Streptococcus mutans is most abundant during exponential and early stationary phases of growth, and rpoE deletion causes an extended lag phase. S. mutans rpoE mutants are also more sensitive to environmental stresses (H₂O₂ stress and acid stress) and show alterations in biofilm formation and virulence (12). Interestingly, rpoE deletion was also reported to increase self-aggregation, coaggregation with other oral microorganisms (13), and an elevated ability to bind human extracellular matrix proteins. For Streptococcus agalactiae, there have been a limited number of studies which show greatest rpoE expression during exponential growth (14) and attenuated virulence of rpoE mutants in a rat sepsis model (15, 16) and human whole-blood survival assays.

In Staphylococcus aureus, the only information thus far regarding RpoE is that strains lacking this factor display defects in starvation-induced stationary-phase survival/recovery and slightly increased acid sensitivity (17). Notably, however, no investigation of its role in cellular physiology and virulence has been performed. Therefore, in this study, we investigated the role of RpoE in promoter selectivity using transcriptome sequencing (RNA-seq) technology and determined that in S. aureus, the δ subunit specifically guides RNAP toward strongly expressed promoters. Deletion of this factor results in a normalizing of transcriptional activity across genes, guiding RNAP away from key virulence-affecting loci. Such effects strongly impair the pathogenic potential of S. aureus, leading to diminished infection using a systemic model of sepsis and increased phagocytosis by human leukocytes. Collectively, our results suggest that rpoE is involved in orchestrating the ability of S. aureus to react and adapt to environmental changes and thus plays a critical role during virulence.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Strains used for experimental procedures are listed in Table 1. An rpoE transposon mutant in S. aureus USA300 strain JE2 was acquired from the Nebraska transposon library (NARSA). This mutation was transferred via ϕ11 transduction to our USA300 Houston wild-type strain (18). Successful transduction of this mutation was confirmed by PCR, using gene-specific (OL1709 and OL1710) and transposon-specific (OL14721 and OL1472) primers. Cultivation of bacteria was performed in tryptic soy broth (TSB) at 37°C. Where required, erythromycin (5 mg/ml), lincomycin (10 mg/ml), or chloramphenicol (10 mg/ml) was added to the medium. Synchronous cultures were obtained as described by us previously (19).

TABLE 1.

Strains, plasmids, and primers used in this study

Strain, plasmid, or primer	Genotype and/or description or sequencea	Reference or source
E. coli
DH5α	Cloning strain	47
S. aureus
RN4220	Restriction-deficient strain	Lab stocks
USA300 HOU	USA300-HOU MRSA isolate cured of pUSA300-HOU-MRSA	19
NE 646	USA300 JE2 rpoE::Bursa, rpoE mutant	NARSA
JAI1287	USA300 HOU rpoE::Bursa, rpoE mutant	This study
JAI1607	USA300 HOU rpoE::Bursa, pMK4::rpoE-His6, rpoE+	This study
JAI1569	USA300 HOU rpoE::Bursa, pOS1sGFP-PsarA-sod, rpoE mutant	This study
JAI1570	USA300 HOU pOS1sGFP-PsarA-sod	This study
Plasmids
pMK4	Gram-positive shuttle vector	48
pJAI101	pMK4::rpoE-His6	This study
pAW102	pMK4::rpoE-His6trunc1, truncated RpoE, N terminus and half-C terminus	This study
pAW103	pMK4::rpoE-His6trunc2, truncated RpoE, N terminus	This study
pAW104	pMK4::rpoE-His6trunc3, truncated RpoE, half-N terminus	This study
pOS1sGFP-PsarA-sod	sarA promoter controlling expression of gfp	25
Primers
OL1471	TTTATGGTACCATTTCATTTTCCTGCTTTTTC	49
OL1472	AAACTGATTTTTAGTAAACAGTTGACGATATTC	49
OL1709	ATGAAAATTCAAGATTATACAAAAC	This study
OL1710	AATAGTTGGTGCGATTTTCTCTTC	This study
OL1184	AGCCGACCTGAGAGGGTGA	50
OL1185	TCTGGACCGTGTCTCAGTTCC	50
OL1973	CGCGGATCCAAGGACCAATTGGCAAAGAACGAC	This study
OL1975	ACCGGAGTCGACTTAATGGTGATGGTGATGGTGATCGTTGAAGTCTTCTTCGTCTTC	This study
OL1976	GAGCGTTAGGTGATTATGAGTACG	This study
OL1978	CACTTCTTCGTCTTCGTCTAGTTC	This study
OL2018	TTAATCGTTGAAGTC	This study
OL2019	GTTTTGTATAATCTTGAATTTTC	This study
OL2020	CAATAAATGATTTTTCATCAACC	This study
OL2021	CAATCATTCGGATATAGTCATTG	This study
OL2022	GATGAAGATGAACTAGACGAAG	This study
OL2931	ATGGTCGACTTAGTGGTGATGGTGATGATGATTTTCAATTTCTTCGTACTC	This study
OL2932	ATGGTCGACTTAGTGGTGATGGTGATGATTTCAATATCATCTACCGAATACC	This study
OL2933	ATGGTCGACTTAGTGGTGATGGTGATGATGTTGATCATCTGTTTGAGCTGG	This study
OL2981	TCGTACTCATAATCACCTAAC	This study
OL2982	GGCGAAACAATGAACTTAT	This study

^aUnderline denotes restriction enzyme site.

Construction of an rpoE complemented strain.

The rpoE gene and its promoter were amplified via PCR, using primer pair OL1973 and OL1975, which are located 200 bp upstream from the rpoE translational start codon and at the 3′ end of the rpoE coding region, respectively. In addition to the native rpoE sequence, a hexahistidine (His₆) tag was included in the reverse primer (OL1975) to create a fusion protein that could be used for downstream purification. For the cloning of truncated rpoE-His₆ fragments, primers 2931 (half of the N terminus only, amino acids 1 to 159), 2932 (the N-terminal half of the protein, amino acids 1 to 264), and 2933 (the N-terminal half of the protein and half of the C terminus, amino acids 1 to 396) were used instead of 1975. The PCR products were cloned into shuttle vector pMK4 and transformed into chemically competent Escherichia coli DH5α. Clones were confirmed via PCR using the same set of oligonucleotides used for cloning. Additionally, Sanger sequencing using primers for the pMK4 multiple cloning site (M13Fw and M13Rv) was performed to confirm fidelity of the construct. The plasmids were transformed into S. aureus RN4220 by electroporation and confirmed by PCR. Correct clones were used to generate a ϕ11 lysate for transduction into the S. aureus USA300 Houston rpoE transposon mutant, which was again confirmed as described above.

Mapping the rpoE promoter.

Rapid amplification of cDNA ends (RACE) was used to identify the rpoE transcriptional start site. For this approach, we used the 5′RACE core set (TaKaRa) and primers OL2018, OL2019, OL2020, OL2021, and OL2022 (Table 1). PCR products were TA cloned using the StrataClone PCR cloning kit (Agilent). The transcriptional start site was identified by sequencing (MWG operon) 10 plasmids that originated with the TA cloning using the universal primers M13Fw and M13Rv.

qPCR.

Quantitative real-time PCR (qPCR) analysis was conducted as described previously (20) using the rpoE gene-specific primers OL1976 and OL1978 and OL2981 and OL2982 and 16S rRNA primers OL1184 and OL1185. Three independent replicates were used to calculate final values.

Zymograms and Western blot analysis.

S. aureus strains were grown as described above, with samples taken at the time points specified. Zymograms were performed with culture supernatants as described by us previously (21). Intracellular and secreted protein fractions were harvested and subjected to SDS-PAGE and Western blot analysis as described by us previously (21). Immunoblotting was performed with a mouse monoclonal anti-His₆ (Covance) or anti-LukS (IBT) antibody at a 1:1,000 or 1:20.000 dilution, respectively, overnight at 4°C. Secondary antibodies were affinity-purified horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG or goat anti-rabbit IgG (Cell Signaling Technology).

RpoE pulldown assay.

A coprecipitation was performed to identify proteins that interact with RpoE. Briefly, rpoE-negative strains carrying pMK4::rpoE-His₆ or pMK4::rpoEtrunc1-His₆ were harvested after 3 h of growth, and intracellular fractions were isolated as described previously (22). Cellular extracts were mixed with nickel-nitrilotriacetic acid (Ni-NTA) agarose beads (Qiagen), and samples were incubated with agitation at 37°C for 1 h. Beads were then washed 5 times (50 mM NaH₂PO₄, 300 mM NaCl, and 20 mM imidazole, pH 8.0) while rotating for 10 min at 37°C. After each wash step, beads were allowed to settle for 10 min, the supernatant was removed, and fresh washing buffer was added. RpoE and any interacting proteins were recovered by the addition of elution buffer (50 mM NaH₂PO₄, 300 mM NaCl, and 250 mM imidazole, pH 8.0). Isolated proteins were then used either to perform SDS-PAGE with subsequent silver staining or for mass-spectrometric analysis, as described by us previously (23).

Transcriptomic analysis via RNA-seq.

For both the wild-type and mutant strains, three independent cultures were grown as described above. Synchronized cultures were harvested after 3 h, and RNA was isolated using a Qiagen RNeasy kit (Qiagen). To ensure complete removal of genomic DNA, RNA was treated with DNase I (Turbo DNA free; Ambion). The concentration and purity of RNA were determined using an Agilent 2100 bioanalyzer. Following quantification, equimolar amounts of RNA, from each of the three replicate preparations, were pooled, and rRNA was removed using the MICROBExpress (Invitrogen) and Ribo-Zero (Epicentre) kits. Removal of rRNA was confirmed, again employing the Agilent 2100 bioanalyzer. rRNA-depleted samples were then used for RNA-seq analysis using the IonTorrent Total RNAseq kit v2 according to the manufacturer's instructions (Life Technologies). Templated ion sphere particles were generated using the Ion OneTouch 200 template kit, v2. Sequencing was performed on an IonTorrent 318 chip using an Ion PGM 200 sequencing kit. Data generated were exported to the CLC Genomics Workbench software package for analysis. Reads were aligned to the USA300 FPR genome, and expression values for each gene were determined as RPKM (reads per kilobase per million reads) values. An RPKM threshold of detection value of 10 was imposed as a lower-level cutoff, and data were normalized using the quantile normalization approach (24). Genes demonstrating changes in expression lower than 2-fold were excluded from further analysis.

Assay to detect alpha-hemolysin activity.

The activity of alpha-hemolysin was determined as described by us previously (25) using pooled, whole human blood (Bioreclamation). Hemolysin activity (Ha) was calculated using the following equation: Ha = (OD₅₄₃ × 1,000)/(vol. of sample × 15 min × OD₆₀₀ × 0.5).

Coinfection of whole human blood.

The ability of strains to survive in human blood was determined by seeding equal amounts of exponentially growing wild-type and rpoE mutant bacteria into pooled, whole human blood (Bioreclamation). Pooled blood was used to minimize effects that result from single-donor samples, such as being preconditioned by S. aureus infection or variation due to the immune status of the donor. Samples were incubated with shaking at 37°C, with aliquots withdrawn hourly to determine the bacterial burden by serial dilution and plating on TSA and TSA supplemented with erythromycin. Bacteria growing on TSA reflect the total number of cells in the sample, while TSA containing erythromycin allows only rpoE mutants to grow.

Analysis of differential phagocytosis by human leukocytes using flow cytometry.

Phagocytosis of wild-type and mutant cells was determined using pooled, whole human blood (Bioreclamation) and flow cytometry, as described by us previously (25). Experiments comparing the wild type and mutant were conducted using the same batch of blood in order to exclude variations resulting from different blood samples. These studies were facilitated by the generation of green fluorescent protein (GFP)-expressing variants of the USA300 Houston wild type and its rpoE mutant using plasmid pOS1sGFP-P_sarA-sod RBS (26). Cells were analyzed using forward and side scattering in a fluorescence-activated cell sorting (FACS) Excalibur cytometer (BD Biosciences). FACSDiva version 6.1.3 software (BD Biosciences) was used to analyze the data. Results represent the averages for three independent experiments and are presented as the percentages of GFP-positive cells plus or minus standard errors of the means (SEM).

Coinfection model of murine sepsis and dissemination.

In order to determine the virulence of strains, a murine coinfection model of sepsis was employed. These experiments were conducted as described previously (27), with the following modifications. Briefly, 6-week-old, female CD-1 Swiss mice were purchased from Charles River Laboratories and housed in the vivarium at the College of Medicine, University of South Florida. Ten mice were inoculated by tail vein injection with 100 μl bacterial suspension (1 × 10⁸ CFU/ml), containing the wild type and the rpoE mutant in a 1:1 ratio. The infection was then allowed to proceed for 7 days before mice were euthanized and the kidneys were collected. Each organ was homogenized in 3 ml sterile PBS, and the numbers of CFU/kidneys for the wild type and its rpoE mutant were determined via serial dilution, as described above. This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Institutional Animal Care and Use Committee of the University of South Florida (permit number A-4100-01).

RESULTS

S. aureus RpoE shows homology to delta subunit proteins of other Gram-positive bacteria.

The majority of information regarding the δ subunit of RNAP comes from studies in B. subtilis and streptococcal species. We therefore compared the primary structure of the S. aureus counterpart to those of its homologs in other organisms to determine if differences in sequence were apparent (Fig. 1A). Interestingly, the proteins analyzed show significant variation across their sequences yet retain homology in the N terminus, with a conserved region between amino acid residues 14 and 91 (S. aureus amino acid numbering). A sequence search (pfam.sanger.ac.uk) revealed this region as containing a predicted helix-turn-helix (HARE-HTH) DNA-binding motif, which is also found in the B. subtilis version of the protein. Notably, the N-terminal region of the HTH motif (residues 14 to 57) shows decreased similarity between proteins from various species, while the C-terminal portion of the motif (residues 58 to 91) displays a high degree of homology. The presence of such an α-helix-rich region has previously been discussed for B. subtilis (28); however, its biological function is currently unknown. We also analyzed the similarity of RpoE between various staphylococci (Fig. 1B). As expected, we observed a much larger degree of similarity for protein sequences within the genus, although again, the C terminus displayed less conservation than the N terminus.

FIG 1.

Protein alignment of the δ subunit of RNAP. Alignments of representative RpoE proteins from different species were prepared using CLC Main Workbench software. (A) Comparison of RpoE sequences between S. aureus and other Gram-positive organisms. Colors represent the degree of similarity, with red being the most identical and blue/black the most divergent. Broken black lines next to the species names represent alignment of proteins to the consensus sequence. The green rectangle denotes the region of highest conservation among different species; the red arrow marks the HARE-HTH motif identified by Pfam search using the S. aureus protein; blue and yellow arrows represent α-helices and β-sheets, respectively, identified previously in the B. subtilis protein (28). (B) Comparison of RpoE between staphylococci.

rpoE is transcribed from a single σ^A-dependent promoter during exponential growth.

To investigate the role of rpoE in S. aureus, we first began by defining its promoter. Using 5′ RACE, we detected a single transcriptional start site 68 bp upstream of the translation initiation codon, which is preceded by a strong σ^A-type promoter, comprised of a −35 sequence of TTGcgA (lowercase letters represent divergence from the consensus), followed by a perfect 17-nucleotide (nt) spacer and a consensus −10 sequence (TATAAT). Additionally, a strong ribosome binding site (AGGAaG) 9 nt upstream of the ATG initiation codon was observed. To track the growth phases and situations in which the S. aureus cell deploys RpoE, we next analyzed its transcription using quantitative real-time PCR (Fig. 2). We detected strongest expression of rpoE during exponential growth, with a peak between 1 and 2 h and a subsequent decline at later time points. To correlate these findings with protein stability, we assessed the abundance of RpoE using a His₆-tagged variant expressed from its own promoter on the shuttle vector pMK4. Consistent with our transcriptional findings, we identified high protein levels during exponential growth, which again declined over time.

FIG 2.

rpoE expression in S. aureus is maximal during exponential growth. (A) Quantitative real-time PCR analysis was used to measure rpoE expression from 3 independent cultures. Error bars ± SEM are shown. (B) Western blot analysis using an anti-His antibody. Samples were standardized using the Bradford assay, and an equal amount of protein for each time point was loaded onto gels. Shown is a representative image of 3 independent experiments.

RpoE interacts directly with RNA polymerase.

To determine whether the δ subunit binds to core RNAP in S. aureus, we performed a pulldown assay using a plasmid-encoded, His-tagged variant of the RpoE protein. Intracellular lysates were generated from exponentially growing rpoE mutants bearing either the pMK4::rpoE-His₆ construct or pMK4 alone (control). RpoE-His₆ and its interacting partners were then purified by pulldown assay and analyzed by SDS-PAGE and mass spectrometry. We identified 3 major protein bands on SDS-PAGE gels compared to control samples with molecular weights that correlate to those expected for RNA polymerase subunit β (RpoB), β′ (RpoC), and δ itself (Fig. 3). Using mass spectrometry, we confirmed that the identities of these bands did indeed correspond to the suggested components of RNAP. A fourth protein was also identified in test samples and was found to be elongation factor Tu (Tuf). Collectively, our results demonstrate that RpoE is part of the S. aureus transcription machinery and most likely interacts directly with the β and/or β′ subunit of RNAP, although we cannot rule out that weaker interactions with other subunits of this complex also occur.

FIG 3.

RpoE interacts with the S. aureus RNAP complex. A His-tagged version of RpoE was purified by pulldown assay from S. aureus cell lysates, alongside empty vector control samples, and run on an SDS-polyacrylamide gel (silver stained). Samples were subjected to mass-spectrometric analysis for protein identification. Shown are the identities of bands, alongside their molecular masses.

To explore which components of RpoE mediate this interaction, we generated 3 His₆-tagged, truncated forms of the protein on the shuttle vector pMK4 in the rpoE mutant background. These contained the following: (i) the N terminus and half of the C terminus (amino acids 1 to 369), (ii) only the N terminus (amino acids 1 to 264), or (iii) half of the N terminus only (amino acids 1 to 159). Abundances of the truncated proteins were first measured by Western blotting (see Fig. S2A in the supplemental material). Interestingly, only the longest of the truncated versions was detected in these studies. To identify if this lack of protein resulted from a lack of truncated protein expression, we performed qPCR analysis (see Fig. S2B in the supplemental material). Importantly, we observed robust expression for all variants of rpoE. Since the proper construction of all plasmids was confirmed by Sanger sequencing, we concluded that the shortest variants of RpoE are likely characterized by protein instability. We next repeated our pulldown experiments using the longest of the RpoE truncated proteins to examine RNAP binding. Upon doing so, we observed results identical to those for the full-length RpoE protein (data not shown). As such, we were able to show that the C terminus of RpoE, at least in part, is dispensable for interaction with RNAP, further supporting the hypothesis that δ subunit-RNAP interactions are mediated by the RpoE N terminus. Interestingly, we also showed that in S. aureus, deleting the entire RpoE C terminus results in inherent instability for the entire protein and a lack of functionality for the S. aureus δ subunit.

The S. aureus δ subunit of RNAP strongly influences expression of virulence determinants and genes encoded on mobile genetic elements.

Given that RpoE functions as a component of the transcriptional complex, we next set out to assess its role in influencing gene expression in S. aureus. This was achieved using RNA sequencing (RNA-seq) technologies and our rpoE mutant strain. Samples of the wild type and mutant were taken during exponential growth (3 h), and the changes in gene expression were compared. Upon analysis, we observed 191 differentially expressed genes in the rpoE mutant strain compared to expression in the parent (Fig. 4), with 83 positively regulated and a further 108 that are repressed (see Table S1 in the supplemental material). To confirm data derived from our RNA-seq experiments, we performed qPCR with a representative subset of genes, which confirmed the direction and fold change of expression observed (see Fig. S3 in the supplemental material).

FIG 4.

Genomic map of altered transcription profiles upon rpoE deletion. The outermost circle (yellow) represents a heat map comparing changes in gene expression for the rpoE mutant compared to results for the wild type. The middle circle (black) depicts RPKM values for the wild type, while the inner circle (red) represents RPKM values for the rpoE mutant (see Materials and Methods for an explanation). ϕSA3usa and ϕSA2usa denote both prophages found in S. aureus USA300.

When genes with decreased expression were reviewed by ontology (Fig. 5), one of the most striking observations was that the transcription of a very large number of known virulence factors is reduced in the rpoE mutant. These include 9 of the 10 major secreted proteases, components of 3 of the 4 bipartite toxins, nuclease, several phenol-soluble modulins (PSMs), alpha-hemolysin, beta-hemolysin, and a number of known immune evasion factors. To ensure that changes in gene expression for virulence factors was not due to unintended mutations in global regulatory systems, we sequenced the agr operon, RNAIII, and the sae operon. No alteration in gene sequence was found between the rpoE mutant and the wild-type strain, indicating that the observed changes were due to ablation of RpoE activity. Changes in the expression of key virulence regulation loci were noted (RNAII and RNAIII, ∼1.4-fold; saeR, 2.8-fold), although to a significantly lesser extent than that observed for most virulence factors (4- to 8-fold). This suggests that the changes in expression of agr and sae do not completely explain the major alterations observed in virulence factor expression upon rpoE deletion.

FIG 5.

Ontological grouping of genes influenced by RpoE activity in S. aureus. A total of 191 genes showed at least a 2-fold alteration of expression in the rpoE mutant compared to that in the parental strain. Genes are grouped according to known or predicted ontologies.

Analysis of genes with increased expression in the mutant revealed, quite remarkably, that almost the entire ϕSA3usa prophage is upregulated (39 genes; increases ranging from 2.9- to 29-fold). Additionally, for the SaPI5 pathogenicity island, another mobile genetic element, expression of 9 genes was increased. Further, we observed upregulation of the nar-nir operon, as well as a variety of membrane and lipoproteins.

rpoE disruption leads to a normalizing of gene expression in S. aureus.

When a heat map of the changes in gene expression is reviewed (Fig. 4), it is apparent that those genes with altered expression in the mutants are for the most part evenly distributed over the entire genome (with the exception of phage ϕSA3usa, SaPI5, and the nar-nir operon). Interestingly, however, upon rpoE disruption, a normalizing of expression appears to occur. Specifically, the majority of genes that had decreased expression in the mutant are typically highly transcribed in the wild-type strain. Furthermore, genes that are typically expressed at a low level in the parent are significantly upregulated in the rpoE-lacking strain (Fig. 6). Therefore, the difference in expression between strongly and weakly transcribed genes becomes dampened, and due to the normalizing effect of rpoE disruption, growth-phase-specific expression patterns are less pronounced. These data support the suggestions from other organisms that RpoE influences promoter selectivity of RNAP based on promoter strength (6), whereby RpoE enhances the transcription of strong promoters and limits expression from weaker elements. In S. aureus such a function appears to have evolved to include the favoring of virulence factor promoters, which are commonly among the most highly expressed genes.

FIG 6.

RpoE influences RNAP promoter selectivity based on promoter strength. The 191 genes identified as being differently expressed in the rpoE mutant were evenly separated into four groups based on strength of expression in the wild type. Specifically, the first quartile represents the 47 genes most highly expressed under standard conditions (3 h of growth in TSB) from the pool of 191. Conversely, the fourth quartile refers to the 47 genes expressed at the lowest levels from the pool of 191. Upon rpoE disruption, a normalizing affect is observed whereby highly expressed genes are overwhelmingly downregulated and low-expression genes are significantly upregulated.

RpoE disruption leads to the decreased accumulation and activity of key virulence factors.

Given that virulence factors were among the most profoundly affected upon rpoE disruption, we next sought to determine if these changes led to alterations in protein abundance and activity. As such, we first assessed Panton-Valentine leukocidin (PVL) protein levels via Western blotting (Fig. 7A). We determined that, much like the case with RNA-seq data, a significant decrease in the LukS signal was detected in culture supernatants from the rpoE mutant compared to results for the parental and complemented strains. Following this, we next performed activity profiling for alpha-hemolysin using whole human blood. Upon analysis of the hemolytic capacity of culture supernatants, we observed a 3-fold decrease in hemolytic activity upon rpoE disruption, which was largely restored by supplying rpoE in trans (Fig. 7B). Finally, given that 9 of the 10 secreted proteases had diminished expression in the mutant strain, we performed protease activity assays. When the mutant strain was cultured on casein nutrient agar, we observed a decrease in clearing around the rpoE mutant in comparison to results for the parent (Fig. 7C). Using zymography with gelatin as a substrate, we observed even more pronounced effects, with a significant reduction in proteolytic activity and with several key proteolytic bands proving to be absent in the mutant strain (Fig. 7D). Collectively, these findings confirm not only the RNA-seq observations that RpoE preferentially influences RNAP toward highly expressed virulence factor genes but that rpoE ablation leads to the diminished transcription, protein synthesis, and activity of central pathogenic determinants.

FIG 7.

RpoE ablation leads to decreased accumulation and activity of key virulence factors. (A) LukS (33 kDa) Western blot using supernatant (15 h) from an rpoE mutant and its parental and complemented strains. The amount of protein loaded was standardized using a Bradford assay. (B) Hemolytic activity assay using culture supernatants and pooled, whole human blood. Data presented are from at least three independent replicates, with error bars ± SEM shown; P values were determined using Student's t test *, P < 0.05; **, P < 0.005; ns, not significant). (C and D) Protease activity assay using casein nutrient agar (C) or gelatin zymography (D). W, wild type; M rpoE mutant; C, complemented rpoE mutant.

An rpoE-defective strain demonstrates decreased virulence in both human and murine models of infection.

To determine if these global alterations in virulence factor production led to measurable outcomes for pathogenesis, we next investigated the ability of an rpoE mutant to survive and compete against the wild type in whole human blood. Accordingly, exponentially growing wild-type and rpoE mutant cells were inoculated into whole human blood in a 1:1 ratio. Aliquots were then withdrawn every hour for 3 h, and the number of CFU/ml of each strain was determined via plating on TSA alone or TSA containing erythromycin (which selects for the antibiotic resistance cassette in the mutant strain). Over the infection period, the percentage of rpoE mutant cells continually decreased (Fig. 8A), with a 1:0.46 ratio observed at 2 h (31.6% recovery of the mutant inoculum) and a 1:0.2 ratio observed at 3 h (17% recovery of the mutant inoculum). By comparison, identical studies performed with TSB over a similar 3-h period revealed no changes in viability for the rpoE mutant strain (see Fig. S4 in the supplemental material). This suggests that a virulence defect, rather than decreased fitness, mediates the impaired survival in whole human blood.

FIG 8.

The δ subunit of RNAP is required for virulence in both human and murine models of infection. (A) Both the WT and mutant strains were inoculated into pooled (5 donors) whole human blood at a 1:1 ratio and incubated for 3 h at 37°C. Aliquots were withdrawn at the times specified, and the numbers of CFU/ml were determined by plating on either TSA or TSA containing erythromycin (to select for the mutant strain). Data are generated from three independent replicates; error bars ± SEM are shown. (B) Wild-type and rpoE mutant strains harboring a constitutively expressing GFP gene were separately incubated in whole human blood. Samples were withdrawn, and the percentage of GFP⁺ granulocytes was measured by FACs. Data are generated from three independent replicates; error bars ± SEM are shown. (C and D) Mice were infected with an equal ratio of rpoE mutant and wild-type cells. After 7 days, mice were euthanized and bacterial loads in the kidneys were assessed. Box-and-whisker plots represent the minimum and maximum values (whiskers), as well as the 25th to 75th percentiles (boxes). The median for each group is indicated as a solid black line. *, P < 0.05; **, P < 0.005; ***, P < 0.001; determined using Student's t test.

In order to explore this survival defect in human blood more closely, we next assessed interaction of the rpoE mutant and wild-type strains with human leukocytes. Accordingly, GFP-expressing variants of each strain were used to separately infect whole human blood. Samples were withdrawn at 2 h and 4 h, and the percentage of each strain present in granulocyte cells was determined using FACs analysis, as described by us previously (25). We determined that at both time points, there were significantly more rpoE mutant cells within human granulocytes than cells of the parental strain (Fig. 8B). These increased rates of phagocytosis are likely mediated by decreased virulence factor expression seen in the δ subunit mutant and explain the survival defect observed in whole human blood.

To determine if these ex vivo human findings were recapitulated in vivo, we assessed the infectious capacity of an rpoE mutant using a murine model of sepsis. Accordingly, we coinfected mice with 1 × 10⁷ CFU/ml of the wild-type and rpoE mutant strains in a 1:1 ratio. After 7 days, mice were euthanized, the kidneys were harvested, and the bacterial burdens of the wild-type and mutant were determined. Three mice died before the end of the infection period and were excluded from analysis. When determining bacterial loads within the kidneys of surviving mice (Fig. 8C), we found a 10-fold decrease in rpoE mutant cells from the inoculum, with only 10% of the total bacterial burden resulting from the δ subunit mutant (Fig. 8D). These findings demonstrate that RpoE-mediated guidance of RNAP toward highly expressed promoters, including virulence factors, is required for successful S. aureus infection.

DISCUSSION

In this study, we sought to explore the role of the delta subunit of RNAP in the major human pathogen S. aureus. Despite being first identified as a unique, Gram-positive specific component of the RNAP apoenzyme more than 30 years ago (2), very little is understood about its role and function. Herein we show that in S. aureus, it is highly expressed early in growth and seemingly interacts with the β and β′ subunits of RNAP, suggesting they might be its binding partners in vivo. Interestingly, other subunits of RNAP were not copurified, leading to the conclusion that a direct interaction with the α, ω, and σ subunits is perhaps unlikely. Importantly, these data are the first evidence from any organism that points to putative interaction partners for RpoE, providing much-needed understanding of its functional role.

As a component of the RNAP complex, the role of the δ subunit is clearly confined to activities concerning gene expression. Therefore, we sought to understand and analyze its target genes in S. aureus, using RNA-seq technologies, during exponential growth (a time when RpoE is highly abundant in the S. aureus cell). We identified almost 200 genes that were changed in expression upon rpoE disruption. Interestingly, these genes displayed a strong pattern in terms of RpoE influence, with typically highly expressed genes in the wild type being strongly downregulated in the mutant, while many low-expression genes were upregulated. As such, loss of the δ subunit in S. aureus appears to lead to a normalization of gene expression, abrogating the more variable, temporal expression patterns typically observed. Such a finding is in line with the hypothesis that RpoE weakens binding of RNAP and DNA and therefore influences weaker promoters more than stronger ones (7). In support of this, Juang and Helmann (6, 29) distinguished 3 different kinds of promoters affected by RpoE: (i) weak promoters that are repressed by RpoE, (ii) moderate-strength promoters that show decreased open complex formation in the presence of the delta subunit, and (iii) strong promoters that are resistant to RpoE inhibition.

Such a role for the delta subunit of RNAP also appears to be conserved in S. aureus based on our findings, and this opens up an interesting point of evolution. It is believed that the influence of RpoE in selectively guiding RNAP toward important promoters is to mediate rapid changes in gene expression, and thus survival, during stress. Such a notion is supported by observations from both B. subtilis (8) and S. mutans (30), which demonstrate that rpoE ablation leads to an impaired ability to recover from stresses. Specifically, in S. mutans (30), it has been shown that an rpoE mutant has decreased accumulation of oxidative-stress defense proteins upon hydrogen peroxide challenge and a concomitant reduction in survival. These lower levels of protective proteins are not caused by downregulation in the mutant per se but rather a stronger, RpoE-influenced stress response in the parent, leading to increased accumulation of key protective proteins. This demonstrates that in the absence of RpoE, cells are less primed to respond transcriptionally to unfavorable conditions and mount a sufficiently protective response.

In S. aureus a lone study that references the δ subunit of RNAP, where an rpoE mutant was identified as having a markedly decreased ability to recover following prolonged starvation, exists (17). Herein we show (see Fig. S5 in the supplemental material) that such an effect is true even after overnight growth, since we observe an extended lag phase of our rpoE mutant similar to those described for other low-G+C Gram-positive organisms. It seems that such a defect is confined to the initial adaptation to new environments, since we observed otherwise unimpaired growth after entrance into log phase, with equal numbers of viable wild-type and mutant bacteria (data not shown). As such, it appears that RpoE may have evolved in Gram-positive bacteria, and the Firmicutes specifically, to facilitate a rapid switching of gene expression profiles that is geared toward the natural expression level of genes within a cell. The reason for this could be that by centralizing such a response within a key protein within the transcription complex, specific and global alteration in gene expression can be rapidly achieved upon environmental change, leading to expedient and efficient adaption and survival. Given that S. aureus is a highly human-adapted pathogen, much of its gene expression is dedicated to survival within the human host. Consequently, we contend that the δ subunit of RNAP has evolved in S. aureus along these virulence-adapted lines, since a large number of the most highly expressed genes in this organism are virulence specific. Therefore, RpoE provides S. aureus with a survival advantage upon entering the human host, which results in a survival defect upon the abrogation of its activity.

With that said, the manner by which RpoE mediates its function cannot be explained solely by mere promoter strength, since there are highly and low-level-expressed loci in S. aureus that are not influenced by its activity. When one compares RpoE protein sequences between various Gram-positive bacteria, there does appear to be a large variation in similarity other than in regions of the collected N termini. The N-terminal regions are characterized by the presence of several α-helices and β-sheets that were previously identified in the B. subtilis protein (28, 31) and are predicted to be present in the S. aureus counterpart (www.predictprotein.org). Such conservation is perhaps to be expected, since the N terminus is thought to be the component that mediates interaction with core RNA polymerase (9), which is itself a highly conserved protein complex. Unusually, this conserved region of α-helices and β-sheets is predicted to form a winged HTH domain (HARE-HTH). Although they have a wide range of functions in nature, including transcriptional regulation, DNA degradation, and restriction modification (32), all functions studied thus far for HARE-HTH domains involve some form of DNA binding. Thus, even though no direct interaction with nucleic acids has yet been shown for the δ subunit of RNAP, a function in recognizing transcriptionally relevant DNA features would appear to be possible, if not probable, via this HARE-HTH domain. Such a situation would thus explain the selectivity of RpoE that extends beyond mere promoter strength.

By comparison, the C terminus of the protein shows little conservation between different species and is in fact the part of the protein responsible for significant interspecies variation and alterations in protein length. Due to its highly acidic nature and strongly disordered structure, thus far the function of the C terminus has remained obscure. Herein we show that deletion of this C terminus in S. aureus seemingly results in decreased stability of the RpoE protein, suggesting its importance for proper folding and function. In addition, it has been suggested that the C terminus may mediate interaction with nucleic acids and have a role in RpoE-dependent displacement of RNA and DNA from the RNAP complex (9). As such, it too may influence promoter selection in a manner hitherto not demonstrated.

Another remarkable observation from our RNA-seq analysis was the simultaneous upregulation of mobile genetic elements within the rpoE mutant. Specifically, we observed that most genes on the S. aureus pathogenicity island 5 (SaPI5) and prophage ϕSA3usa displayed strong increases in expression. In the context of the phage element, it is possible that these effects are mediated by alterations in transcription of a common regulator encoded within the prophage region (33). The shift between lytic and lysogenic pathways is commonly mediated by an altered abundance of competing regulators, which seek to swing the balance from one lifestyle to the other (reviewed in references 34 and 35). While the specific molecular details of the lytic/lysogenic decision are not known for ϕSA3usa, we do observe alterations in regulatory elements on this prophage in the rpoE mutant. Specifically, the phage transcriptional regulator SAUSA300_1968 is strongly altered in expression (+12-fold). This finding, together with the upregulation of most of the adjacent phage genes, leads to the postulation that SAUSA300_1968 may be a positive and global regulator of ϕSA3usa gene transcription. In terms of SaPI5, there is limited understanding of regulatory events that modulate expression of the genes it encodes. However, it is possible that an event similar to that suggested for ϕSA3usa occurs, since these elements are believed to have evolved from phages themselves. In line with this, SaPI5 in USA300 contains at least 2 uncharacterized regulatory elements (33), one of which (SAUSA300_0804) displays elevated gene expression upon rpoE deletion. As such, it is possible that upregulation of this element could account for changes in the expression of the pathogenicity island, as described for ϕSA3usa.

Alternatively, it is possible that RpoE has a wider role in the S. aureus cell in providing cellular immunity against phage infection. Although quite different, immunity systems in bacteria that seek to abrogate phage infections, such as clustered, regularly interspaced short palindromic repeat (CRISPR) elements, do exist (36). In such a scenario, RpoE may be involved in silencing foreign elements and therefore protecting the core genome.

In the context of virulence, the various defects observed for the rpoE mutant are readily explained. Numerous elements that are known to play integral roles in S. aureus disease causation, including Panton-Valentine leukocidin (PVL) (37–40) (reviewed in reference 41), alpha-hemolysin (42–44), and different proteases (21, 25), were shown to be decreased in the rpoE mutant. To explore how these changes are mediated, we first ensured (by Sanger sequencing and analysis of mRNA transcript fidelity from the RNA-seq reads) that unintended mutations were not present in key global regulators. We next assessed expression of key virulence regulators and found the agr locus to be unaffected by ΔrpoE deletion. We did observe downregulation of saeR (−2.8-fold) in our RNA-seq experiment, which may explain some of the effects observed (45). Indeed, Nygaard et al. (46) show that deletion of saeRS results in a downregulation of various virulence factors, many of which were also affected in the rpoE mutant. However, the limited alteration in saeR expression in the rpoE mutant is unlikely to be the sole reason for the effects observed. Specifically, the changes observed by Nygaard et al. result from complete loss of the saeRS locus (∼120- and 178-fold downregulation). In our results, we see similar effects on virulence factor synthesis, with much smaller changes in sae system expression (e.g., for splE, ΔsaeRS led to a −1.16-fold change in expression; ΔrpoE led to −2.89-fold change). Furthermore, rpoE deletion causes changes in gene expression that are not seen in the saeRS mutant. For example, aureolysin and several other virulence factors (e.g., V8 protease and PSMs) are either positively affected or unaffected in the saeRS mutant, while being negatively affected in the rpoE mutant. Accordingly, the effects observed in the rpoE mutant appear to be direct and gene specific rather than being caused solely by the differential expression of major virulence regulators.

In summary, we herein demonstrate the importance of the RNAP δ subunit for gene regulation and virulence in S. aureus. We reveal that rpoE loss results in an uncoupling of well-ordered regulatory circuits, resulting in viable yet less-fit S. aureus cells that are unable to rapidly adapt to stress, such as that encountered during infection. We describe a new layer of transcriptional regulation in S. aureus that, as in other bacteria, functions in a promoter-specific manner to deploy its global effects and, in this highly human-adapted pathogenic organism, uniquely targets those regions of the genome that are associated with virulence.