The bacterial strains and plasmids used in this study are listed in Table 1. The primers used are listed in Table 2. When necessary, the corresponding antibiotics (100 μg/ml ampicillin and 50 μg/ml kanamycin) were added to the medium. S. aureus was usually cultured in Tryptone Soy Broth medium at 37°C and 220 rpm, to which 15 μg/ml chloramphenicol and 10 μg/ml erythromycin were added when necessary.

To construct arcR knockout plasmid, the upstream and downstream homologous arm fragments were amplified from wild-type strain NCTC8325 and then ligated by overlap PCR. The obtained fragment was cloned into the pBTs plasmid digested with EcoR I and Hind III. Ampicillin was used to screen the correct transformants. The pBTs derivative pBTs-arcR was electroporated into RN4220 for modification and transformed into NCTC8325 for allelic replacement as described (Hu et al., 2015). PCR and sequencing were used to verify the mutation. For the construction of complementary plasmids, arcR and katA genes were amplified from NCTC8325, and the obtained fragments were ligated into pCN51 plasmid digested with Pst I. The derivates pCN51-arcR and pCN51-katA were transformed into RN4220 and subsequently electroporated into ΔarcR mutant. Expression of arcR and katA was induced by 10 μM CdCl₂ if necessary.

The MICs of S. aureus to a variety of antibiotics were determined in Mueller-Hinton Broth medium as described previously (Borrero et al., 2014). For the determination of time-dependent bactericidal curves, overnight culture was diluted into fresh TSB medium at a ratio of 1:100 and incubated at 37°C and 220 rpm until OD₆₀₀ reached 1.0. The MICs of ciprofloxacin, ofloxacin, norfloxacin, levofloxacin, moxifloxacin, and garenoxacin were 0.32, 0.25, 0.48, 0.19, 0.125, and 0.03 μg/ml, respectively. Thiourea (150 mM) was added when the strains grew to OD₆₀₀ 0.6. Samples were taken out at certain time points, diluted 10-fold, and colonies counted by dropping plate.

The overnight cultures were diluted with fresh TSB medium to OD₆₀₀ 0.05 and then, H₂O₂ at a final concentration of 0.4 mM was added to the cultures as described previously (Oogai et al., 2016). They were incubated at 37°C with 150 rpm and 1 ml aliquots were removed at the indicated times. The supernatant was removed by centrifugation and pellet was washed twice with sterile 1 × PBS. The precipitates were diluted 10-fold and then dropped onto TSB plates. The plates were cultured overnight at 37°C, and colony counts were performed on the next day.

The whole blood of mice was collected with an anticoagulant tube, and neutrophils were extracted using a mouse peripheral blood neutrophil isolation kit (Solarbio, Beijing, China). The concentration of extracted cells was counted by cell counter. Measurement of ROS levels was determined as previously described with minor modifications (Wu et al., 2009). Neutrophil cells were diluted to 10⁵/ml using warm Hanks’ balanced salt solution (HBSS) containing 100 mM luminol and 1 U/ml horseradish peroxidase. The diluted cell suspension was added to a 96-well plate at 200 μl/well, and the reaction was performed at 37°C for 10 min. The bacterial strains were added at a multiplicity of infection (MOI) of 5. At the same time, 10⁶ cells were removed and infected with an MOI of 10. The reaction was carried out in a cell incubator at 37°C, and 20 μl reactant was removed every hour for dilution and dropping plate counting, to which 100 mM NAC was added if necessary.

When the bacteria were cultured in TSB medium to OD₆₀₀ 1.0, they were treated in PBS containing 0.4 mM H₂O₂ for 10 min and collected by centrifugation. Total RNA was extracted using RNAprep Pure Cell/Bacteria Kit (Tiangen Biotech, Beijing, China). Then, cDNA was synthesized with 1 μg RNA and PrimeScript™ IV 1st strand cDNA Synthesis Mix (Takara, Dalian, China). For the qRT-PCR, cDNA was mixed with specific primers (Table 2) and chamQ universal SYBR quantitative RT-PCR master mix (Vazyme, Nanjing, China). 16S rRNA gene was used as internal controls.

Bacteria were cultured to OD₆₀₀ 1.0 using TSB medium, and 1 ml of culture was collected by centrifugation and washed twice with PBS. The cultures were incubated in PBS with or without 0.4 mM H₂O₂ for 30 min. The reaction samples were collected by centrifugation, washed twice with PBS, and broken by Ultrasonic Cell Disruptor (Xinzhi, Ningbo, China). Catalase activity was detected using a catalase assay kit (Beyotime, Shanghai, China).

For the construction of ArcR expression plasmid, the arcR gene fragment was amplified from NCTC8325 genome and the PCR product was ligated with pET28a digested with EcoRI. Kanamycin was used to screen the correct transformants. PCR and sequencing were used for verification to obtain the correct expression of pET28A-his6-ArcR. The correctly constructed pET28a-His6-ArcR plasmid was transferred into E. coli BL21(DE3). The transformers were grown in LB medium containing 50 μg/ml kanamycin at 37°C until OD₆₀₀ reached 0.6 and were induced with 0.5 mM IPTG at 37°C for 4 h. The cells were harvested and lysed by sonication with lysis buffer (25 mM HEPES, 5 mM β-mercaptoethanol, 500 mM NaCl, and 10% glycerol, pH 7.8). The His-tagged fusion proteins were purified in nickel nitrogen triacetate agarose solution (Qiagen, Beijing, China). The bounding protein was eluted by elution buffer (25 mM HEPES, 5 mM β-mercaptoethanol, 500 mM NaCl, 10% glycerol, and 500 mM imidazole, pH 7.8). Target protein was stored at −80°C until use. The concentration of purified proteins was determined by SDS-PAGE and bicinchoninic acid as the standard protein.

EMSA was performed as previously described with minor modifications (Hellman and Fried, 2007). We used NCTC8325 as the template to amplify katA promoter region DNA fragment. The 30 ng DNA fragment was incubated with 0, 100, 200, and 400 ng purified ArcR in binding buffer (25 mM HEPES, 1 mM dithiothreitol, 200 mM NaCl, and 10% glycerol, pH 7.8) at 37°C for 30 min. The 8% polyacrylamide gel was pre-electrophoresed in 1× Tris-borate-EDTA buffer (0.044 M Tris, 0.044 M boric acid, and 0.001 M EDTA, pH 8.0) for 1 h to remove impurities. After adding the sample, electrophoresis was performed for 1 h 40 min on ice. At the end of the electrophoresis, the glue was stained with 0.5 μg/ml ethidium bromide. Imaging was performed using a gel imager (Bio-Rad, Hercules, CA, USA).

To understand whether ArcR affected antibiotic resistance of S. aureus, the MICs of ΔarcR mutants to ciprofloxacin, ofloxacin, linezolid, tetracycline, vancomycin, erythromycin, chloramphenicol, norfloxacin, levofloxacin, moxifloxacin, garenoxacin, and daptomycin were determined. Wild-type NCTC8325 and ΔarcR mutant displayed similar levels of resistance (MICs) to the above antibiotics (Table 3). However, following treatment with 0.32 μg/ml ciprofloxacin, 0.5 μg/ml ofloxacin, 0.5 μg/ml norfloxacin, 0.5 μg/ml levofloxacin, 0.2 μg/ml moxifloxacin, or 0.06 μg/ml garenoxacin, the survival rate of ΔarcR mutant was approximately 342-, 158-, 74-, 20-, 93-, or 87-fold lower than that of wild-type NCTC8325 strain after 10 h sterilization (Figures 1A–F).

The tolerance of S. aureus to fluoroquinolones is usually associated with mutations of topoisomerase II genes, including gyrA and gyrB, and high expression of NorA, NorB, NorC, and SdrM belonging to MFS superfamily efflux pump (Lowy, 2003; Ding et al., 2008; Xu et al., 2011). Expression of norA, norB, norC, and sdrM was detected in ΔarcR mutant, and there was no significant difference compared with the wild-type strain (Supplementary Figures 1A–D). These results suggest that ArcR influenced the resistance of S. aureus against fluoroquinolone antibiotics through other mechanisms.

It was reported that ROS was involved in the bactericidal process of aminoglycoside, β-lactam, and fluoroquinolone antibiotics (Kohanski et al., 2007). To determine the role of ROS in the increased sensitivity of ΔarcR mutant to fluoroquinolones, thiourea that neutralize intracellular ROS was added into the medium when killing curves were detected. Thiourea supplementation increased the survival rate of the ΔarcR mutant strain to a level similar to that of the wild type when treated with the same concentration of antibiotics (Figures 2A,B). These results suggest that a defective oxidative stress response leads to increased sensitivity to fluoroquinolone antibiotics in ΔarcR mutant.

The bacterial response to oxidative stress is a manifestation of virulence, which can help bacteria to resist ROS produced by phagocytes and enhance their infectivity. Activated neutrophils produce O₂•⁻, H₂O₂, nitric oxide (NO), and hypochlorite (HOCl), which kill invading pathogenic bacteria through oxidative burst (Forman et al., 2002; van der Veen et al., 2009). To understand the role of ArcR in the bacterial oxidative stress response, bacteria were incubated with neutrophils. The wild type and ΔarcR mutant could introduce the same level ROS produced by neutrophils when they were incubated with neutrophils (Supplementary Figure 2). However, the survival rate of ΔarcR mutant was significantly lower than that of wild-type strain when they were treated with neutrophils (Figure 3A). The presence of ROS-neutralizing N-acetylcysteine (NAC) increased the survival rate of ΔarcR mutant after incubation with neutrophils (Figure 3B). We used H₂O₂ to verify the role of ArcR in bacterial tolerance to oxidative stress, and the ΔarcR mutant showed higher sensitivity to H₂O₂ than wild-type strain (Figure 3C). These results suggest that ArcR contributes to bacterial resistance to oxidative stress.

The increased sensitivity to ROS suggested that the expression of genes involved in ROS response might be reduced in ΔarcR mutant. PerR is a major global regulator of H₂O₂ response, which is a repressor of peroxide reaction (Dubbs and Mongkolsuk, 2012). The expression of PerR was not significantly altered in ΔarcR mutant (Figure 4A). We measured the expression of other ROS resistance genes. In the presence or absence of H₂O₂, mRNA levels of katA, which encodes primary catalase, were approximately fourfold lower in the ΔarcR mutant than in the wild-type strain (Figure 4B). Whereas, the mRNA levels of ahpC, ahpF, dps, and hmp were similar between the wild-type and ΔarcR mutant (Figures 4C–F). Total catalase activity was lower in ΔarcR mutant than that in wild-type strains when treated with or without H₂O₂, which consistent with the change of katA mRNA levels (Figure 4G). Complement with arcR or katA in ΔarcR mutant restored the catalase to wild-type level (Figure 4G).

Overexpression of katA in ΔarcR mutant strain restored bacterial tolerance to neutrophils, H₂O₂, ofloxacin, and ciprofloxacin (Figures 5A–D). These results confirmed that downregulation of katA expression was responsible for bacterial susceptibility to H₂O₂ which increased Fenton reaction and OH• production and finally resulted in decreased tolerance to fluoroquinolone antibiotics.

The ArcR regulator of S. aureus contains two domains that are characteristic of members of the Crp family of regulatory proteins. Structural prediction showed that the N terminus of ArcR had a circular nucleotide binding domain, and the C terminus contained a highly conserved helix-turn-helix DNA-binding domain. To explore how ArcR regulated katA transcription, we used pET 28a to express ArcR protein in E. coli BL21. Electrophoretic mobility shift assay (EMSA) was performed to detect the binding of ArcR to 244-bp DNA in the promoter region of katA. The amount of free DNA gradually decreased as the amount of ArcR protein increased (Figure 6A). The specificity of binding was reflected in the fact that DNA mobility was not affected when an unrelated His-tagged TetR protein was used. It also did not bind to 267-bp DNA of tetA promoter region (Figures 6B,C). These results suggested that the binding of ArcR to the katA promoter region was specific.