A summary of the bacterial strains and oligos used in this study is listed in Supplementary Tables 1, 2, respectively.

B-medium is a modified LB medium suitable for staphylococci cultivation by adding 1 g/L potassium phosphate (Brückner, 2006). In cultivation steps, the ratio of the bacterial suspension to the total volume of the flask was less than or equal to 1:3 to ensure sufficient aeration.

RPMI medium (catalog number 72400021) was purchased from LIFE Technologies. Formaldehyde (FA), diamide, and NaOCl was bought from Fisher Scientific, MP Biomedicals, and Alfa Aesar, respectively. 4-Methylumbelliferyl-β-D-glucuronide hydrate (MUG), methylhydroquinone (MHQ), methylglyoxal (MG), H₂O₂, cumene hydroperoxide (CHP), and mevalonate were obtained from Sigma-Aldrich. Thiourea, N-acetylcysteine (NAC), and catalase was purchased from Carl Roth, Hölzel Diagnostika, and MP Biomedicals, respectively. Stressors were dissolved in sterilized Milli-Q water for β-galactosidase and survival assays.

Completely assembled chromosomal sequences of S. aureus strains were retrieved from the NCBI website¹ in May 2021. TBLASTN program from the standalone BLAST ncbi-blast-2.11.0+ used WP_000838037.1 as a query to search for possible proteins. Then, we used BLASTP program to identify the homology between the proteins retrieved via TBLASTN to WP_000838037.1. We consider 35% amino acids identity and protein length 375 ± 38 amino acids (10% deviation from WP_000838037.1) as a cutoff to limit OfrA-like proteins (Shi et al., 2020). Identical proteins were filtered out using SDDC program (Ibrahim et al., 2017).

Multiple sequence alignments were done using Clustal Omega² with the default parameters. Phylogenetic trees were constructed using RAxML 8.0.0 software with the following setup (−f a −# autoMRE −m PROTGAMMAAUTO) (Stamatakis, 2014). Tree visualization and annotation were done using ggtree v2.0.1 (Yu, 2020).

To construct EI011, we exchanged P_hla in pKO10 with a 1-kb fragment upstream of ofrA (Ohlsen et al., 1997). The reporter plasmid was transformed into E. coli IM08B and then electroporated into S. aureus JE2 strain (Monk et al., 2015). We confirmed the single crossover event by sequencing of the amplified fragment using primer in the plasmid and another one upstream of the 1-kb fragment. To construct EI046 (JE2ΔofrA), the allelic exchange in S. aureus JE2 strain was mediated by cloning the upstream and downstream fragments into pBASE6 shuttle vector (Geiger et al., 2012). After the double crossover with counter-selection, polymerase chain reaction (PCR) and sequencing were used to identify the mutant (Geiger et al., 2012).

Overnight cultures in RPMI were diluted into final OD₆₀₀ = 0.05 and incubated with serial dilutions of each compound. Minimum inhibitory concentration (MIC) is the minimum concentration that results in not more than (OD₆₀₀ = 0.1) after 24-h incubation at 37°C with shaking at 200 rpm. OD₆₀₀ were measured using Synergy H1 plate reader.

The reporter strain was conditioned in RPMI for 24 h. We diluted the overnight culture 1:100 into fresh medium. In the overnight and the diluted culture, 10 mg/ml chloramphenicol was added as a final concentration. The resulting culture was grown in 37°C until transition from exponential phase to stationary phase (OD₆₀₀ = 1.25 ± 0.05). 500 μl of the bacterial culture was supplemented with the stressor at the specified concentrations. After 2 h in 37°C with shaking at 200 rpm, samples were taken for analysis as indicated in the study of Vidal-Aroca et al. (2006).

Overnight cultures grown in RPMI were diluted 1:100 to reach OD₆₀₀ = 0.5. Samples were taken to represent the control before stress. Substances were added to the indicated concentrations and incubated for 15 min. After the incubation, the cultures were immediately put on ice to transfer to −80°C. RNA was isolated using RNAeasy Mini Kit following the manufacturer’s instructions. DNase I treatment was done using RapidOut DNA removal kit followed by cDNA synthesis via SuperScript IV Reverse Transcriptase utilizing random hexamer primers and ofrA-specific primer with non-staphylococcal tag (Supplementary Table 2). Quantitative PCR was performed with Biozym Blue S’Green qPCR Kit and rho and rpoB as the internal controls (Sihto et al., 2014).

We diluted overnight cultures 1:100 in fresh RPMI followed by incubation at 37°C with shaking at 200 rpm until mid-logarithmic phase. OD₆₀₀ were adjusted to be 0.4 after collecting the bacteria by centrifugation for 10 min at 4°C and 4,000 rpm. After adding the indicated concentration and incubation at 37°C for the indicated time interval, serial dilutions of the bacterial suspension were made followed by plating of 80 μl on LB agar using single plate-serial dilution spotting (SP-SDS) method (Thomas et al., 2015).

The exposure time to 1.5 mM NaOCl was 30 min, while we challenged the bacteria against 40 or 30 mM H₂O₂ for 1 h. MHQ and methylglyoxal were exposed for 3 h.

In H₂O₂ survival assay and after 1-h exposure, samples were centrifuged at 4,000 rpm for 10 min and then resuspended in sterile PBS supplemented with 10 mg/ml catalase. In NaOCl survival assay, the serial dilutions of the bacteria were made in sterile LB to quench the remaining NaOCl.

Overnight cultures of S. aureus JE2 and EI046 (JE2ΔofrA) strains were grown at 37°C with shaking at 200 rpm in B-medium. Genomic DNA was extracted using DNeasy Blood and Tissue Kit (from Qiagen) according to the manufacturer’s protocol modification for Gram-positive bacteria. Whole genome sequencing and variant calling were done by MicrobesNG. The raw sequenced reads are deposited in SRA database (BioProject ID: PRJNA812552).

Similar protocol was applied as done in the study of Flannagan et al. (2018). RAW 264.1 macrophage cell line was prepared by passaging in RPMI medium supplemented with 10% FCS and Pen/Strep. Passages 12–15 were used in the bacterial survival assays. Bacteria grown in BHI to logarithmic phase were washed in sterile PBS, resuspended in RPMI, and added (∼5 × 10⁷ CFU) to ∼5 × 10⁶ RAW 264.1 cells in 24-well plates (MOI = 1:10).

Extracellular bacteria were killed by treatment with 150 μg/ml gentamicin in 1 h. We considered zero-time by adding fresh RPMI + 10% FCS medium. Samples (n = 5) were taken at time = 4, 24, and 48 h while the 4-h samples were set as the normalization factor as indicated (Fritsch et al., 2019). Bacterial counts were achieved as indicated above using SP-SDS method on LB agar.

Venous blood specimens were collected from four healthy human blood donors (age: 21–32 years, gender: two women and two men) in tubes supplemented with anti-coagulant (1.6 mg/ml EDTA). Blood was kept at room temperature (RT) until use. We used a similar protocol to van der Maten et al. (2017) with some modifications.

Bacteria grown in BHI to logarithmic phase were washed two times with sterile PBS. Afterward, 30 μl of the bacterial suspension (2.2 × 10⁷ CFU/ml) was mixed with 100 μl of human blood (final concentration = 5 × 10⁶ CFU/ml). Saponin (final concentration = 1%), immediately or after 60-min incubation at 37°C with shaking, was added to blood–bacteria mixture for cell lysis. After incubation for 20 min at 4°C, viable bacterial cells were determined using SP-SDS method on LB agar in two technical replicates.

Dilutions (1:100) of three independent overnight cultures of S. aureus JE2 and EI046 (JE2ΔofrA) strains were grown at 37°C with shaking at 200 rpm in RPMI medium until OD₆₀₀ = 0.5. We extracted total RNA using RNeasy Mini Kit (from Qiagen) as in the manufacturer’s manual. DNA digestion was done using RapidOut DNA Removal Kit. Negative amplification in PCR using 16S rDNA primers was taken as an evidence of successful DNase treatment. Evaluation of RNA quality, rRNA depletion, cDNA library generation, and sequencing were done by the Core Unit Systems Medicine Facility at the University Hospital Würzburg. Adaptor trimming was done using Cutadapt software. Trimmed reads were aligned to the reference genome (NC_007793). We used READemption pipeline for reads mapping, coverage calculations, gene quantification, and differential gene expression analysis (Förstner et al., 2014). We developed scripts for gene set enrichment analysis (GSEA) using clusterProfiler (Wu et al., 2021). Regulon analysis was done using self-written R scripts. RNA-seq data are available in NCBI’s Gene Expression Omnibus (GSE196683).

We diluted overnight cultures 1:100 in the respective medium (with supplementation if necessary) and allowed the bacterial growth in 37°C and shaking at 200 rpm for 16 h (stationary phase) (Sullivan and Rice, 2021). 2 ml of the bacteria was centrifuged at 16,000 rpm for 2 min and then washed in sterilized water. OD₆₀₀ were recorded for normalization. 400 μl of methanol was added to the washed bacterial pellets and incubated at 55°C for 3 min. After centrifugation at 16,000 rpm for 2 min, 300 μl of the methanolic extract was added to 700 μl methanol. 200 μl of the solution was measured in three technical replicates at A₄₆₅ with infinite 200Pro machine and methanol as a blank.

Overnight cultures in RPMI were 1:100 diluted into fresh RPMI medium and incubated in 37°C with shaking at 200 rpm until OD₆₀₀ = 0.5. Then, we diluted the bacteria down to ∼5 × 10⁵ cells and mixed with different concentration of streptonigrin (0–2 μg/ml). OD₆₀₀ were measured using Synergy H1 plate reader.

Statistical analysis was done under R version 3.6.1 using rstatix R package version 0.6.0 and ggpubr R package version 0.4.0. Statistical tests were indicated in the corresponding figure legends. We considered statistical significance if p < 0.05.

Utilizing TBLASTN of WP_000838037.1 (SAUSA300_0859 gene product) against S. aureus USA300_FPR3757 genome, we found that OYEs are encoded from two paralogous genes (SAUSA300_0859 and SAUSA300_0322). We propose to name SAUSA300_0859 as old yellow enzyme flavin oxidoreductase A (ofrA) and SAUSA300_0322 as ofrB. Upon NCBI’s CDD search, OfrA and OfrB contain “OYE_like_4_FMN” domain. OfrA and OfrB orthologs are conserved in B. subtilis as YqiG and YqjM, respectively (Figure 1A). Multiple sequence alignment shows that E. coli_NemA and P. fluorescens_XenB do not belong to the same class of Gram-positive OYEs (Figure 1A); rather, NemA and XenB belong to Class-I of classical OYEs whereas YqjM belongs to Class-III OYEs (Scholtissek et al., 2017). OfrA does not belong to any of the studied OYEs classes and represent a novel class of OYEs (Scholtissek et al., 2017). Therefore, we hypothesized that OfrA could play different roles in S. aureus than NemA in E. coli. Since S. aureus has a wide spectrum of genomic lineages, we were interested to study ofrA conservation in S. aureus strains and Firmicutes.

OfrA is encoded in all publicly available 749 chromosomes of S. aureus strains with 98–100% amino acid identities (Supplementary Table 3 and Supplementary Figure 1A). On genus level, OfrA is encoded from the 28 staphylococcal representative chromosomes. Staphylococcal OfrA orthologs cluster in three distinct clades (Figure 1B, Supplementary Figure 1B, and Supplementary Table 4). The phylogenetic tree illustrates that OYEs are not subjected to horizontal transfer; rather, they evolved within the encoding organism to adapt to certain function.

Although different OYEs are encoded in a number of representative Firmicutes’ chromosomes, OfrA-like orthologs are limited to only a few species (Figure 1C and Supplementary Figure 2). However, OfrB is not conserved across the different staphylococci (Figure 1D). In fact, BLASTP retrieved OfrA orthologs from the ofrB-minus genomes (such as S. saprophyticus, S. hominis, and S. epidermidis). From our analysis, we learned that some of the ofrB-minus genomes encode variants of the OYEs, other than OfrB, with varying lengths and/or other fused protein domains. Apparently, an earlier speciation event in a common ancestor resulted in OfrA and OfrB differences. Since OfrA is associated with MT02 resistance and is found conserved in staphylococci, we intended to understand the function of ofrA in the human pathogen S. aureus as an example of OYEs.

Previously, we showed that the bisquaternary bisnaphthalimide MT02 induces ofrA (El-Hossary et al., 2018). Thus, we hypothesized that compounds with an electron-deficient center (such as electrophilic stress generators) could similarly induce ofrA. To avoid the quenching activity of standard laboratory media (TSB and LB), we used RPMI as a well-defined medium, which, in addition, mimics the host environment (Meerwein et al., 2020).

To screen for important induction conditions, we constructed a reporter strain (EI011) that harbors a chromosomally encoded β-galactosidase from a promoter-less lacZ gene under the control of P_ofrA (Supplementary Figure 3). Since β-galactosidase assay is protein-based, we chose 2-h exposure time to report for ofrA induction. We tested a range of RES conditions such as diamide, fosfomycin (Fosfo), formaldehyde (FA), methylglyoxal (MG), and MHQ at the MIC against EI011 to avoid false negative results from stressors’ toxicities in higher concentrations. The MIC concentration did not significantly affect the bacterial growth in the 2-h experimental time (Supplementary Figure 4).

β-Galactosidase assays suggest that diamide, formaldehyde, methylglyoxal, and MHQ induce ofrA (Figure 2A). MHQ results in the highest upregulation (21-folds), while formaldehyde, methylglyoxal, and diamide result in approximately fourfold upregulation. However, there is no upregulation upon exposure to fosfomycin (Figure 2A). Furthermore, the β-galactosidase assays show a dose-dependent induction by diamide, formaldehyde, methylglyoxal, and MHQ (Figure 2B).

Since diamide is a non-specific disulfide-stress inducer, we were interested in the induction with toxic aldehydes (formaldehyde and methylglyoxal) and quinone-stress (MHQ). Reverse transcription quantitative polymerase chain reaction (RT-qPCR), comparing ofrA mRNA levels after 15-min exposure in S. aureus JE2 strain background, confirms the results obtained by the reporter strain (Figure 2C). After formaldehyde, methylglyoxal, and MHQ exposure for 15 min, there were 38-, 4-, and 10-fold upregulation in ofrA, respectively.

Staphylococcus aureus faces electrophilic stress in many natural niches including host–pathogen interface. Therefore, we were interested in elucidating the role of OfrA in S. aureus survival in electrophilic stress conditions in more detail. To address this, we used a marker-less deletion mutant in S. aureus JE2 (EI046 = ΔofrA) as well as a complemented strain (EI047 = pofrA) which harbors a plasmid-based expression of ofrA from its natural promoter. To assure the absence of secondary mutations that could affect any phenotype, we sequenced the whole genome of JE2 and ΔofrA. The results show that there are no discriminating mutations in ΔofrA compared to JE2 except for ofrA mutation.

In bacterial survival assays, we compared the survival of ΔofrA vs. JE2 strain after 3-h exposure of 0.5 mM MHQ and 2 mM methylglyoxal compared to an untreated control. In quinone stress, JE2 strain survived 90%; however, ΔofrA survived only 61%. pofrA with restored ofrA expression complemented the survival defect phenotype (Figure 3A).

ΔofrA exhibits a similar survival defect in methylglyoxal compared to the parental strain which we could restore in the complemented strain pofrA (Figure 3B). We concluded that OfrA is important to mediate quinone-stress and toxic aldehydes and that OfrA is an important factor in S. aureus defense against electrophilic stress conditions.

Since NemA was reported to be important in hypochlorite stress in E. coli, we also analyzed the role of OfrA in ROS and hypochlorite stress conditions. β-Galactosidase assays suggest that oxidative stress [H₂O₂ and cumene hydroperoxide (CHP)] and hypochlorite (NaOCl) stress induce ofrA (Figure 2A). Moreover, ofrA induction follows a dose response at different concentrations of H₂O₂ and NaOCl (Figure 2B).

Next, we analyzed the survival of the deletion mutant ΔofrA vs. JE2 wild-type (WT) after 1-h exposure of 40 mM H₂O₂ compared to untreated control. The ofrA mutant strain had decreased survival in 40 mM H₂O₂ compared to WT (Figure 3C). Moreover, the complementation in pofrA restored the WT phenotype indicating that OfrA enhances S. aureus survival in oxidative stress.

In 1.5 mM NaOCl, JE2 survived (70%) after 30 min compared to 33% survival of ΔofrA denoting OfrA importance in S. aureus survival against NaOCl (Figure 3D). However, pofrA failed to complement the mutant phenotype. We hypothesized that complementation with a high-copy number plasmid could result in overconsumption of cellular resources and therefore a decreased resistance against NaOCl taking in consideration the devastating non-specific effects of HOCl (da Cruz Nizer et al., 2020). From the whole genome sequencing results, there are no secondary mutations that could affect NaOCl survival phenotype. In addition, NaOCl results in ofrA induction (Figure 2A). Hence, we concluded that OfrA is also an important factor in S. aureus defense against ROS and hypochlorite stress conditions. To assure that there are no effects of strain growth behavior on the survival phenotypes, we observed the growth kinetics of the logarithmic phase cells of the three strains in RPMI medium (Supplementary Figure 5). Similar growth kinetics suggested that the growth behavior did not contribute to any of the measured survival phenotypes.

Macrophages produce reactive oxygen, chlorine, and electrophilic species as the killing factors against internalized S. aureus (Moldovan and Fraunholz, 2019). Since OfrA is important in survival in these stress conditions, we wondered whether ofrA mutation results in defective macrophage survival.

After 24 h, ΔofrA survival was reduced in RAW 264.7 macrophages compared to JE2 but this was not statistically significant. However, the difference became significant after 48 h. After 48 h, ΔofrA survived significantly (∼50%) less than JE2 in RAW 264.7 cell line (Figure 4A). In the complemented strain, the difference between pofrA and ΔofrA was statistically significant even after 24 h. The bigger difference could be explained by ofrA dosage effect from the high-copy number plasmid utilized in the complementation. We concluded that OfrA affects the bacterial fitness by enhancing S. aureus JE2 survival in macrophages.

The bacteria–immune response interaction in human blood determines the fatality of S. aureus-mediated bacteremia. We wondered whether OfrA contributes to S. aureus JE2 virulence via promoting survival in human blood. After 1 h of incubation with whole human blood, ΔofrA survives (∼23%) compared to the WT (∼65%) (Figure 4B). The complementation in pofrA restores the survivability of the mutant back to ∼46% (Figure 4B). In conclusion, ofrA contributes to S. aureus survival in whole human blood.

To understand ofrA function in S. aureus, we compared the transcriptome of ΔofrA vs. JE2 in mid-logarithmic phase in RPMI. Through RNA-seq experiment, we found that the ofrA mutant had decreased RNA abundances corresponding to 93 genes and increased RNA abundances corresponding to 95 genes (Supplementary Table 5). Several redox-related (SAUSA300_0339, SAUSA300_0340, SAUSA300_0212, SAUSA300_0213, ypdA, and cymR) and stress-related genes (csbD, clpB, sigB, and rsbW) are deregulated. Using regulon analysis and GSEA, we observed the following: (1) one-carbon metabolism is inhibited in ΔofrA indicating an unbalanced redox status (Shetty and Varshney, 2021), and (2) the carotenoid biosynthesis (crtOPQMN) is suppressed in the mutant (Supplementary Table 6).

To validate the results of RNA-seq analysis, we performed RT-qPCR to quantify the mRNA abundances of crtM, acuA, and rocD genes. RT-qPCR confirmed the results obtained by the RNA-seq analysis. In RT-qPCR, log₂ (fold change) of crtM expression is −0.7 ± 0.1 in ΔofrA compared to JE2 (Supplementary Figure 6). Moreover, log₂ (fold change) of acuA and rocD expression is −0.6 ± 0.2 and 2.9 ± 0.2, respectively (Supplementary Figure 6).

The carotenoid pigment (staphyloxanthin) production is mediated via the crtOPQMN operon (Götz, 2005). Staphyloxanthin (STX) is a virulence factor that affects the survival of S. aureus against oxidative stress and human neutrophils, so we were interested in quantifying STX levels in the ofrA mutant (Clauditz et al., 2006). Indeed, STX is decreased in ΔofrA compared to JE2 and pofrA (Figure 5A).

Acetyl-CoA is the key input of mevalonate pathway to produce farnesyl pyrophosphate (FPP), which enters the crtOPQMN pathway (Pelz et al., 2005). STX was previously shown to be decreased with glucose due to intracellular acetyl-CoA loss (Tiwari et al., 2018). To test the acetyl-CoA-dependency of STX phenotype in the ofrA mutant, we measured STX level in B-medium (contains no glycolytic substrates) ± 0.5% glucose.

As expected, glucose decreased STX levels in WT (Figure 5B). However, the intracellular acetyl-CoA loss did not affect the ratio between ΔofrA and JE2 in STX production (Figure 5B). This result suggests that ΔofrA does not have decreased STX production via change in acetyl-CoA concentration. Conversely, the loss of crtM in JE2 and ΔofrA, transduced from strain Newman (Clauditz et al., 2006), resulted in the disappearance of the ofrA-dependent phenotype, and both strains become white (Figure 5C; Reichert et al., 2018). Therefore, we concluded that ofrA mutation could affect the mevalonate pathway.

The mevalonate pathway is classified into upper (mvaS, mvaA) and lower (mvaK1, mvaK2, and mvaD) mevalonate pathways (Reichert et al., 2018). The output of the upper mevalonate pathway is the mevalonate itself. So, we were interested to understand the dependency of ofrA-mediated STX phenotype on the presence of mevalonate.

We compared the STX production ±1 mM mevalonate in RPMI medium. The presence of mevalonate results in the disappearance of ofrA-mediated phenotype in ΔofrA compared to JE2 (Figure 5D). Therefore, we deduced that the ofrA mutant has decreased STX production via the upper mevalonate pathway in S. aureus.

To understand whether ROS hypersensitivity is linked to decreased STX, we challenged JE2ΔcrtM and JE2ΔcrtMΔofrA strains against H₂O₂ in the survival assay. As expected, crtM and ofrA mutations in JE2 resulted in decreased survival in ROS (Figure 5E). If low STX production is responsible for ROS-mediated killing, the double deletion mutants shall behave as ΔcrtM and ΔofrA. Contrary to this hypothesis, the double mutation in both genes, JE2ΔcrtMΔofrA, causes H₂O₂ hypersensitivity and more killing in 30 mM H₂O₂ (Figure 5E). Thus, crtM and ofrA are important in ROS survival but independent of each other.

From RNA-seq analysis, we know that ofrA mutation does not result in upregulation of sodA, sodM, katA, peroxidases, and hmp, which indicates that intracellular levels of O₂^– and H₂O₂ are within the WT levels (Supplementary Table 5). Therefore, ofrA-dependent ROS hypersensitivity is downstream to H₂O₂ production.

The only plausible explanation of ROS hypersensitivity we had is that ofrA contributes to the repair mechanism of thiol-oxidation caused by H₂O₂. This notion is supported by the fact that ofrA is generally induced with electrophilic, hypochlorite, and oxidative stress. Thiourea scavenges the hydroxyl radical that should decrease the H₂O₂-mediated killing (Wasil et al., 1987). As expected, the survival of ΔofrA was lower than JE2 strain in H₂O₂ (Figure 6A). Addition of 120 mM thiourea resulted in increased S. aureus JE2 survival and ΔofrA up to a similar level (Figure 6A). Therefore, we concluded that ofrA contributes to oxidative stress tolerance via a repair mechanism downstream to H₂O₂ but upstream to hydroxyl radical-mediated lethality.

Since MHQ is the highest induction condition (Figure 2A) and the hydroxyl radical is the main killing mechanism after H₂O₂ challenge (Figure 6A), we hypothesized that the survival defect of the mutant in ROS is secondary to disruption of thiol-dependent homeostasis upon ROS challenging.

In S. aureus, MHQ imposes oxidative and electrophilic stress (Fritsch et al., 2019). To test our hypothesis, we conducted MHQ survival assay ± N-acetyl cysteine (NAC). NAC supports the thiol-dependent redox homeostasis that acts as both reactive oxygen and electrophilic species scavenger (Pedre et al., 2021), and thiourea as ROS scavenger via thiol-independent mechanism. If our hypothesis was correct, thiourea would not be able to quench the electrophilic stress. 120 mM thiourea does not abolish MHQ toxicity in ofrA mutation; however, 1.25 mM NAC does (Figures 6B,C). We, therefore, concluded that ofrA plays a role in the thiol-dependent redox homeostasis, which affects the survival in oxidative, electrophilic, and hypochlorite stress, and that is an essential function during infection inside macrophages, and in human blood.