The Intersection of the Staphylococcus aureus Rex and SrrAB Regulons: an Example of Metabolic Evolution That Maximizes Resistance to Immune Radicals

ABSTRACT Staphylococcus aureus is the most pathogenic member of the Staphylococcaceae. While it acquired an arsenal of canonical virulence determinants that mediate pathogenicity, it has also metabolically adapted to thrive at sites of inflammation. Notably, it has evolved to grow in the presence of nitric oxide (NO·). To this end, we note that the Rex regulon, composed of genes encoding dehydrogenases, metabolite transporters, and regulators, is much larger in S. aureus than other Staphylococcus species. Here, we demonstrate that this expanded Rex regulon is necessary and sufficient for NO· resistance. Preventing its expression results in NO· sensitivity, and the closely related species, Staphylococcus simiae, also possesses an expanded Rex regulon and exhibits NO· resistance. We hypothesize that the expanded Rex regulon initially evolved to provide efficient anaerobic metabolism but that S. aureus has co-opted this feature to thrive at sites of inflammation where respiration is limited. One distinguishing feature of the Rex regulon in S. aureus is that it contains the srrAB two-component system. Here, we show that Rex blocks the ability of SrrA to auto-induce the operon, thereby preventing maximal SrrAB expression. This results in NO·-responsive srrAB expression in S. aureus but not in other staphylococci. Consequently, higher expression of cytochromes and NO· detoxification are also observed in S. aureus alone, allowing for continued respiration at NO· concentrations beyond that of S. simiae. We therefore contend that the intersection of the Rex and SrrAB regulons represents an evolutionary event that allowed S. aureus to metabolically adapt to host inflammatory radicals during infection.

respiratory flux and/or the oxidation state of the menaquinone pool must exist for the majority of the SrrAB orthologs to function. When stimulated, SrrA drives the expression of both S. aureus cytochromes (cytochromes aa 3 and bd), the anaerobic ribonucleotide reductase, pyruvate-formate lyase, NOÁ-detoxifying flavohemoprotein, as well as heme synthesis and iron-sulfur cluster repair proteins (10). Essentially, when respiratory flux wanes, SrrA increases the capacity of the electron transport chain to optimize the energy state of the cell. This is particularly important for NOÁ resistance since NOÁ detoxification, iron-sulfur (Fe-S) cluster repair, and maximization of cytochrome content all enable S. aureus to maintain positive energy balance in the presence of this immune radical (19).
Here, we show that the Rex regulon is significantly expanded in S. aureus compared with most other CoNS, save S. simiae and other members of the SAC. We show that this expansion is necessary and sufficient for NOÁ resistance and that this trait is not exclusively associated with S. aureus. We further show that SrrAB is autoregulated and Rex repressed, and therefore, NOÁ responsive, only in S. aureus. Thus, the merging of two metabolic regulons may represent an evolutionary event aimed at allowing S. aureus to achieve a metabolic state compatible with host inflammation.

RESULTS
The expanded Rex regulon is necessary and sufficient for NOÁ resistance. Inhibition of respiration in S. aureus, either by oxygen depletion or NOÁ exposure, is known to induce the expression of genes normally repressed by Rex. Given that S. aureus is highly resistant to NOÁ while other staphylococci generally are not, we sought to investigate the relationship between the Rex regulon and S. aureus NOÁ resistance. We conducted full-genome searches for Rex binding sites (TTGTGAW 6 TCACAA) located within 400 bp upstream of an annotated start codon and allowing a maximum of two mismatches in the following genomes: S. aureus COL, S. simiae CCM 7213, S. epidermidis RP62A, S. haemolyticus JCSC1435, S. saprophyticus ATCC 15305, Staphylococcus carnosus TM300, Staphylococcus pseudintermedius HKU10-03, S. lugdunensis HKU09-01, Staphylococcus warneri SG1, Staphylococcus pasteuri SP1, and Macrococcus caseolyticus JCSC5402 (Table S1 in the supplemental material). S. aureus possessed, by far, the most (38 putative Rex-regulated genes), followed by S. simiae with 29 putative Rex-regulated genes (Fig. 1A). NOÁ-sensitive S. epidermidis only encodes 16 putative Rex-regulated genes, and S. haemolyticus and S. saprophyticus encode even fewer (Fig. 1A).
We tested whether the apparent expansion of the Rex regulon in S. aureus contributes to NOÁ resistance. We noticed that S. simiae encodes almost as many Rex-regulated genes as S. aureus, including ldh1, a gene not found in S. epidermidis or other CoNS, and one that is known to contribute to NOÁ resistance (11). We therefore compared the growth of S. aureus, S. simiae, and S. epidermidis while enduring NOÁ stress. Following the addition of NOÁ, S. aureus and S. simiae did not exhibit a growth defect, while S. epidermidis lagged in growth until the high concentration of NOÁ dissipated after 5 h of exposure (Fig. 1B). Since Rex is a repressor, we hypothesized that overexpressing it might prevent the production of dehydrogenases that are important for maintaining redox balance in the absence of respiration. Indeed, overexpression of Rex from the constitutive lgt promoter prevented growth of S. aureus in the presence of NOÁ but did not affect untreated cells (Fig. 1C). Taken together, these data suggest that the apparent expansion of the Rex regulon is necessary and sufficient for NOÁ resistance. Additionally, overexpression of Rex inhibited anaerobic growth, suggesting that any time respiration is hindered, derepression of the Rex regulon is essential for growth (Fig. 1D). Furthermore, it appears that this expansion occurred sometime after the last common ancestor shared by S. aureus and S. simiae diverged from the S. epidermidis lineage (Fig. S1) since both species are NOÁ resistant, while S. epidermidis is not.
SrrAB expression is responsive to NOÁ exposures in S. aureus only. NOÁ exposure is known to induce the expression of SrrAB, which, in turn, drives expression of the SrrA regulon. Rex and SrrA both bind directly to the srrAB promoter, so we hypothesized that Rex and/or SrrA are responsible for the NOÁ responsiveness of S. aureus srrAB (9,20). Since the putative binding sites for Rex and SrrA are not well conserved in CoNS ( Fig. 2A), we tested whether srrAB promoters from any other species responded to the presence of NOÁ. Cloning the promoters for srrAB from S. aureus, S. simiae, S. epidermidis, S. saprophyticus, and S. haemolyticus so that each drove green fluorescent protein (GFP) expression showed that only the S. aureus srrAB promoter is NOÁ responsive ( Fig. 2B and C). This did not correlate with basal SrrAB expression levels in the absence of NOÁ exposure (Fig. S2).
The putative Rex binding sites are ;20 bp upstream of the 235 sequence, which is not consistent with preventing RNA polymerase from accessing the srrAB promoter ( Fig. 2A). However, deletion of rex resulted in a modest 5-fold induction of srrAB even in the absence of NOÁ (Fig. 3A). This, in turn, led to elevated levels of SrrA-activated cytochrome expression in some instances as well ( Fig. S3A and B). Furthermore, the Drex mutant had no effect on srrAB expression in the presence of NOÁ (Fig. S3B). These observations are consistent with Rex-mediated repression of SrrAB expression as the source of NOÁ responsiveness in S. aureus. However, the DsrrB mutant demonstrated virtually no expression of SrrAB and exhibited severe reduction in the expression of SrrAB-regulated genes both in the presence and the absence of NOÁ ( Fig. 3A; Fig. S3A and B). Moreover, the double Drex DsrrB mutant phenocopied the DsrrB mutant ( Fig. 3A; Fig. S3A and B). The epistatic relationship between Rex and SrrB on SrrAB expression is more consistent with Rex preventing the auto-induction of SrrAB expression by SrrA. Since the known Rex-repressed ldh1 was NOÁ inducible in S. simiae (Fig. S3C), the lack of induction of srrAB by NOÁ in S. simiae cannot be due to a defect in Rex derepression. Rather, the SrrA binding site is significantly divergent between S. aureus and CoNS, explaining the unique NOÁ responsiveness of SrrAB expression in S. aureus ( Fig. 2A).
Elevated SrrAB activity in S. aureus allows for optimum respiratory capacity during NOÁ stress. Since the SrrA regulon includes genes involved in cellular respiration and NOÁ detoxification, we reasoned that these genes may be expressed to a higher degree in S. aureus than S. simiae upon stimulation with NOÁ. As expected, srrA, qoxB, and hmp transcripts were more abundant in S. aureus than in S. simiae 15 min after NOÁ treatment (6-fold, 2-fold, and 43-fold, respectively) ( Fig. 3B and Fig. S4A). Furthermore, 60 min after NOÁ exposure, cydA and hmp transcripts were more abundant in S. aureus by 8-fold and 15-fold, respectively ( Fig. 3C and Fig. S4B). Therefore, since SrrAB is NOÁ responsive in S. aureus alone, this species overproduces downstream effectors such as cytochrome production and NOÁ-detoxifying enzymes compared to closely related S. simiae.
A consequence of a relatively overactive SrrAB regulon is the optimization of respiratory activity in the presence of NOÁ. NOÁ will temporarily halt respiration through competitive binding of cytochrome heme cofactors. Once NOÁ levels have been reduced via enzymatic detoxification, however, cellular respiration can resume. We measured this in vitro by using amperometric probes to measure oxygen and NOÁ concentrations in cell suspensions of S. aureus COL and S. simiae in real time. Representative traces show both the spike and clearance of NOÁ and the halt and resumption of oxygen consumption via respiration (Fig. S5). Since Hmp is the primary means of NOÁ detoxification in these species and since it is induced much more in S. aureus due to overexpression of SrrAB, the NOÁ consumption rate was significantly higher in S. aureus than S. simiae upon stimulation (Fig. 4A). Interestingly, while it is known that S. aureus exhibits little or noÁconsumption without stimulation, S. simiae seems to express Hmp constitutively, as the NOÁ consumption rate was not affected by prior exposure to this immune radical (Fig. 4A). Similarly, since both QoxABCD and CydAB were induced by NOÁ more robustly in S. aureus, this species exhibited NOÁenhanced respiratory capacity, while S. simiae did not (Fig. 4B). Given that NOÁ-exposed S. aureus exhibits enhanced NOÁ detoxification and expresses relatively higher levels of cytochromes upon NOÁ exposure than S. simiae, we tested whether S. aureus could resume respiration in the presence of higher levels of NOÁ than its closely related species. Indeed, we found that S. aureus is able to resume respiration at extracellular NOÁ concentrations more than five times that of S. simiae (Fig. 4C), a trait likely to serve the pathogen at sites of inflammation.

DISCUSSION
Compared to most coagulase-negative staphylococci, S. aureus is able to grow much better in the absence of respiration, whether being cultivated anaerobically or in the presence of respiratory inhibitors such as NOÁ (21). Here, we demonstrate that the expanded Rex regulon is necessary and sufficient for this trait as follows. In the absence of respiration, overexpressing the Rex repressor prevents derepression of the regulon. Consequently, these strains cannot grow anaerobically or in the presence of following NOÁ exposures (administered as 10 mM DETA-NO; n = 3) relative to untreated expression levels. Expression levels were normalized to that of rpoD, and induction levels were compared between species for a given gene/time point using Student's t test using the Holm-Sidak method (***, P # 0.0001; **, P # 0.01; *, P # 0.05).
Dmitriev et al. ® NOÁ ( Fig. 1C and D). In addition, S. simiae, which also possesses an expanded Rex regulon, is also highly resistant to NOÁ compared to other coagulase-negative staphylococci (Fig. 1B). Various dehydrogenases and metabolite transporters comprise the Rex regulon, and while the substrates for these enzymes/transporters are largely unknown, they are predicted to be small organic acids and/or amino acids. The expanded Rex regulon would solve a problem with the metabolic strategy of S. aureus during NOÁ stress as we know it today: homolactic fermentation would not allow for incorporation of carbon into biomass. Indeed, host immune cells employ homolactic fermentation and convert one mole of glucose to two moles of lactate, resulting in redox-balanced energy production, but these cells are not replicating. For S. aureus to divide and generate a gram of biomass, it consumes 12 g of glucose, 11 for energy and 1 for biomass (21). If all the glucose is converted to lactate, all carbon would be excreted as waste. Rather, the ability of S. aureus to reduce exogenous substrates to regenerate NAD 1 allows the organism to use some of the glucose carbon for the production of biomass. Metabolic Evolution of the Rex and SrrAB Regulons ® S. simiae may have evolved to use this metabolic strategy to thrive in the anaerobic primate gut, while S. aureus adopted it to thrive at sites of inflammation. Both environments would require efficient respiration-independent growth.
While the last common ancestor shared by S. aureus and S. simiae may have evolved an expanded Rex regulon to thrive anaerobically, the fact that S. aureus adapted to inflammatory radicals would require additional evolutionary changes. One change is the autoregulatory feedback loop of SrrAB (Fig. 3A). Rex prevents the auto-induction of srrAB, but when the Rex regulon is derepressed, SrrA maximizes srrAB transcription. Higher levels of phosphorylated SrrA leads to higher levels of cytochromes and NOÁdetoxifying flavohemoprotein (Hmp) (Fig. 3B and C). This would allow S. aureus to "outcompete" host immune radical production and continue respiring despite their presence. Indeed, when exposed to NOÁ, S. aureus resumed respiration and oxygen consumption at NOÁ levels $5-fold higher than S. simiae (Fig. 4C). When S. simiae senses a buildup of NADH, it is most likely due to it entering the anaerobic environment of the primate gut. Therefore, it would not be necessary to induce cytochromes or Hmp. In contrast, a common reason for S. aureus to sense high NADH is because of host immune radicals, which inhibit respiration. In response, overproducing cytochromes, NOÁ detoxification, and Fe-S cluster repair systems provide a metabolic advantage aimed at overcoming the respiratory hinderances of host inflammation. This may be especially true in tissues where glucose is less abundant since respiration is key for metabolizing gluconeogenic substrates in S. aureus (22,23).
Both Rex and SrrA have been shown to directly bind the srrAB promoter, and there are two potential Rex binding sites upstream of the 235 and one for SrrA ( Fig. 2A) (9,20). However, only one Rex site is active since there was only one shift when incubating recombinant Rex with the srrAB promoter (9). While we do not know definitively which site is bound, either could potentially interfere with SrrA auto-activation. One overlaps entirely with the predicted SrrA binding site, and the other is downstream where binding by Rex could interfere with the SrrA-RNA polymerase interactions. Furthermore, neither Rex binding sites are completely conserved among coagulase-negative staphylococci, including S. simiae. Moreover, the SrrA binding site is completely degenerate in all species other than members of the SAC (Fig. 2A). This implies that the SrrA auto-activation and the Rex repression of this operon evolved relatively recently in S. aureus. The SrrA requirement for the srrAB promoter likely stems from mutations that accumulated in the 235 region. Indeed, while the 210 is completely conserved, the 235 is highly variable, which is consistent with the requirement of SrrA for srrAB transcription in S. aureus, but with relatively constitutive expression in other species.
Another indicator that S. simiae has evolved to hypoxic or anaerobic environments is the constitutive NOÁ-consuming activity exhibited by this species. While the clonal complex 30 (CC30) lineage of S. aureus encodes both a NOÁ reductase and Hmp, most clones only harbor the gene for the flavohemoprotein (hmp). Similarly, S. simiae only encodes an Hmp for NOÁ detoxification. In S. aureus, Hmp is relatively scarce until the cell encounters NOÁ stress (Fig. 4A). In contrast, in S. simiae, Hmp is constitutively expressed and is not induced by exogenous NOÁ in the environment. It is known that Hmp expression in the absence of NOÁ can lead to ROS production, and therefore, the enzyme could be toxic in the presence of oxygen (24). The fact that hmp is constitutively expressed in S. simiae could indicate this organism is generally found in low-oxygen environments. Alternatively, like S. aureus, S. simiae also encodes an NOÁ synthase. Low-level NOÁ production by this nitrous oxide system (NOS) might be enough to prevent Hmp from spontaneously reducing molecular oxygen.
In the end, here, we present evidence that the expanded Rex regulon in certain species of staphylococci is necessary and sufficient for NOÁ resistance. We also suggest that this expansion originally served as an adaptation to low-oxygen environments but was co-opted by S. aureus to thrive at sites of inflammation. This required additional evolutionary adaptations, namely, the Rex-repressed and -autoregulated SrrAB system, which controls cytochrome production and NOÁ detoxification. This adaptation likely allows S. aureus specifically to avoid the cytotoxic effects of host NOÁ.

MATERIALS AND METHODS
Bacterial strains and growth conditions. Strains used in this study are described in Table 1.All strains were grown in either brain heart infusion medium (BHI; Difco, Sparks, MD) or chemically defined PN medium supplemented with 0.5% glucose (25). Cultures were shaken at 250 rpm unless otherwise specified. Antibiotic selection in S. aureus (E. coli) was performed using the following concentrations: 25 mgÁml 21 kanamycin, 5 mgÁml 21 erythromycin, 20 mgÁml 21 chloramphenicol, and 100 mgÁml 21 ampicillin. All restriction enzymes were purchased from New England Biolabs (Ipswich, MA).
AR0352 was generated via allelic replacement using the E. coli-S. aureus shuttle vector pBTK as previously described (Cooke, PLoS One). AR1626 and AR1630 were created via U11 phage transduction of NE588 into S. aureus COL or AR0352, respectively. GFP reporter strains driven by srrAB promoters were constructed as follows. Homologous srrAB promoters were PCR amplified from S. aureus COL, S. simiae, S. epidermidis, S. haemolyticus, and S. saprophyticus genomic DNA, resulting in amplicons with 59 BamHI and EcoRI restriction sites for directional ligation into the GFP reporter transcriptional fusion vector pJF119. Plasmids were then propagated through E. coli via electroporation (with ampicillin selection), harvested using a QIAprep Spin miniprep kit (Qiagen, Hilden, Germany), and then transformed into S. aureus restriction-deficient strain RN4220 (with chloramphenicol selection) (26). Plasmids were finally transduced into S. aureus COL using U11 phage lysates made from the transformed RN4220 strains.  rex allele cloned into NdeI of pOS1-P lgt This study was supplemented with chloramphenicol for antibiotic selection when appropriate. We added 10 mM NOC-12 (EMD Millipore Sigma, Temecula, CA) and 1 mM diethylamine NONOate (DEA-NO) (Sigma-Aldrich, St. Louis, MO) when cultures concurrently reached an OD 660 of 0.15, and then growth was allowed to resume. GFP reporter experiments. Cells were grown at 37°C in 200 ml BHI medium supplemented with chloramphenicol and shaken aerobically (1 mm orbital) on a Synergy HTX plate reader (Biotek, Winooski, VT) for 24 h. When cultures concurrently reached an OD 660 of 0.2, DETA-NO (Acros Organics, Fair Lawn, NJ) was added to a final concentration of 10 mM, and then growth was allowed to resume.
Quantitative reverse transcriptase real-time PCR. (i) RNA extraction. Cells were grown at 37°C in 60 ml of BHI medium in 500-ml baffled flasks. At an OD 660 of 0.5, a 25-ml sample of cells was collected and mixed with 25 ml of ice-cold 1:1 ethanol/acetone in order to prevent RNA degradation before immediately being stored at 280°C until further use. After adjusting the remaining culture to a volume of 25 ml, DETA-NO was added to a final concentration of 10 mM, and cells were shaken for an additional 15 or 60 min under the same conditions. After 15 or 60 min, the 25-ml culture was collected and stored at 280°C in ethanol/acetone as previously described. Frozen cell suspensions were thawed at room temperature, pelleted via centrifugation, and resuspended in 250 ml of TE buffer, pH 8.0. They were then sequentially frozen in a dry ice/ethanol bath and thawed at 60°C a total of three times before being transferred to Lysing matrix B tubes (MP Biomedicals, Solon, OH). RNA extraction was further carried out with a PureLink RNA minikit (Invitrogen, Carlsbad, CA) per the manufacturer's instructions with additional modifications. Briefly, tubes were bead beat for 60 s in a standard cell disruptor and then placed on ice for 5 min before the addition of 650 ml lysis buffer containing 10 ml b-mercaptoethanol and 1 ml buffer and completion of a second identical bead beating step. Following centrifugation and the standard binding and wash steps with optional on-column PureLink DNase treatment, RNA eluted in 50 ml of RNase-free water was further treated with 1 ml of off-column DNase I (New England BioLabs, Ipswich, MA) at 37°C for 60 min to ensure complete removal of contaminating DNA. Reaction mixtures were deactivated at 75°C for 10 min and mixed with both 350 ml lysis buffer and 250 ml 100% ethanol before being transferred to spin cartridges and eluted as instructed by the manufacturer.
(ii) qRT-PCR. RNA was quantified and assessed for purity via spectrophotometry. Quantitative reverse transcriptase real-time PCR (qRT-PCR) was performed using the Power SYBR green RNA-to-Ct 1step kit (Applied Biosystems, Vilnius, Lithuania) as per the manufacturer's instructions with 50 ng of RNA per reaction. Utilized primers are listed in Table 2, and primer efficiencies were determined empirically by creating a standard curve of amplification cycle (C T ) values plotted against various concentrations of genomic DNA used for amplification. Primer efficiencies ranged from 1.76 to 2.02. For a given reaction, where TGOI TrpoD is the ratio of transcript abundance for any gene of interest to that of rpoD, E is the efficiency for the corresponding primer set, and C T is the amplification cycle at which the arbitrary threshold fluorescence was met. Fold induction was determined by dividing the calculated transcript ratio for a given gene expressed under NOÁ stress by its corresponding ratio for expression in the absence of NOÁ.
Determination of nitric oxide and oxygen consumption. Cells were grown in 200 ml of BHI in 2,000-ml flasks at 37°C and 200 rpm. At an OD 660 of 0.5, cells were harvested and immediately spun down in 250 ml Sorvall centrifuge tubes. Alternatively, at an OD 660 of 0.5, diethylene triamine NONOate (DETA-NO) was added to a final concentration of 10 mM, and cultures were shaken for an additional hour before being harvested in the same way. After being washed once with phosphate-buffered saline (PBS), cells were pelleted once more and resuspended to a final OD 660 of 1.0 in PBS bubbled with air for 2 h. For a typical experiment, 60 ml of culture was transferred to a 100-ml beaker containing a magnetic stir bar, and all steps were conducted at 37°C. The culture was stirred at high intensity for 15 min to ensure maximal aeration before being sealed with a 3-holed rubber stopper, leaving no headspace in the beaker. ISO-NOP and ISO-OXY-2 amperometric probes (World Precision Instruments, Sarasota, FL) were inserted through the stopper along with an air-tight pipette tip to seal the injection port when not in use. The entire apparatus was air-tight, and both probes were allowed to polarize to a final minimal current level before conducting an experiment. With the culture being stirred at moderate intensity, glucose was added to a final concentration of 0.1% in order to initiate respiration. After allowing oxygen to be consumed for 2 min, as indicated by the probe tracing, the rapidly releasing NOÁ donor proline NONOate (PROLI-NO) (Cayman Chemical, Ann Arbor, MI) was added to a final concentration of 100 mM (resulting in an immediate release of 200 mM NOÁ). Continuous measurements were taken until all dissolved oxygen was consumed. NOÁ concentration was determined via comparison to a standard curve of PROLI-NO injections at doubling concentrations, while % O 2 present was determined by setting the baseline current and the minimally detected current at the end of an experiment to 100% O 2 and 0% O 2 , respectively.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.