Hydrogen Peroxide Affects Growth of S. aureus Through Downregulation of Genes Involved in Pyrimidine Biosynthesis

Reactive oxygen species (ROS) play a crucial role in the cellular defense against S. aureus, as evidenced by the importance of this pathogen in patients lacking the ROS-generating phagocyte NADPH oxidase NOX2. ROS concentrations required to kill S. aureus in vitro are much higher than those found in the phagosome. We therefore hypothesized that sublethal ROS concentrations may play a role in S. aureus gene dysregulation and investigated the in vitro transcriptomic response of S. aureus to sublethal concentrations of hydrogen peroxide (H2O2). A striking observation of these experiments was a coordinated and massive downregulation of genes involved in pyrimidine metabolism. Using transposon insertion mutants, we demonstrated that deletion of carA, a gene involved in pyrimidine synthesis, led to a significant growth defect and to an increased sensitivity of S. aureus to added H2O2. The phenotype of the carA mutant could be reversed through supplementation with the pyrimidine precursor uracil, or with a multicopy vector encoding carA. As opposed to the impact of ROS on extracellular survival, carA deletion did not affect the intracellular survival in neutrophils. Our results raise the possibility that ROS-dependent downregulation of pyrimidine metabolism might be a survival strategy of S. aureus, allowing colonization through intracellular survival, while decreasing the risk of killing the host through dampened extracellular growth.


INTRODUCTION
Staphylococcus aureus is a Gram-positive, round-shaped facultative anaerobic bacterium, discovered in 1881 by Alexander Ogston, a Scottish surgeon. S. aureus colonizes approximately 30% of the human population (1) and its main site of colonization is the nasal cavity (2,3). While S. aureus is generally commensal, it causes a broad spectrum of severe infections (4). S. aureus is able to adapt to the environment (5) and, in the era of antibiotics, it has rapidly developed or acquired antimicrobial resistances. Development of new anti-staphylococcal treatments is now considered a high priority by the WHO (6).
Reactive oxygen species (ROS) generated by phagocytes are key players in the defense against S. aureus. This concept is largely based on the phenotype of chronic granulomatous disease (CGD), a genetic disorder caused by loss of function mutations in the ROS-generating phagocyte NADPH oxidase NOX2. CGD patients often suffer from severe, recurrent and/or persistent infections with S. aureus (7). Thus, in a simplistic model, NOX2derived ROS are killing S. aureus by targeting DNA, proteins and lipids (8). However, S. aureus possesses a complex antioxidant defense system, including two superoxide dismutases and one catalase (9) and is quite resistant to microbicidal concentrations of ROS. The relationship between ROS and S. aureus might be more complex than a direct lethal effect on S. aureus and we hypothesized a more subtle mechanism that could affect the bacterial fitness.
Among the different ROS generated by NOX2, hydrogen peroxide (H 2 O 2 ) may have a predominant role. High concentration of this non-radical oxidant can damage cells and tissues. However, based on available research, it is unlikely that these very high concentrations (>50 mM) can be reached within the phagosome as phagosomal H 2 O 2 concentrations are in the micromolar range (10). However, there is increasing evidence that there are H 2 O 2 gradients and therefore the local H 2 O 2 concentration in close proximity to the NADPH oxidase might be higher and reach the low millimolar range (11). H 2 O 2 is now increasingly recognized as an intra-and intercellular signaling messenger (12). H 2 O 2 can impact cell phenotype through a variety of mechanisms, including regulation of gene expression (13). H 2 O 2 reversibly oxidizes specific cysteine residues of key protein targets to regulate their function in eukaryotic cells (14). A similar effect of H 2 O 2 on bacterial signaling was observed (15). For instance, several redox-sensitive transcriptional regulators exist in bacteria (16,17) and these regulators usually control the expression of genes involved in defense against oxidative stress (18,19). However, alteration of bacterial signaling by H 2 O 2 can also be used by host cells as a defense mechanism. Intestinal cells produce low H 2 O 2 concentrations that interferes with bacterial signaling and weakens the fitness of potential intestinal pathogens (20).
In this study, we addressed a significant open question in the field: concentrations of ROS that can be achieved in vivo are sublethal for S. aureus, yet NOX2-derived ROS play a crucial role in the host defense against this microorganism. We found a major impact of sublethal ROS concentrations on gene expression in S. aureus. The importance of understanding the interaction of ROS with this important pathogen is severalfold: first, it should provide at least a part of the answer to the oldest question in CGD research, namely why do patients that cannot generate ROS have such difficulties to defend themselves against S. aureus; second, a better knowledge of the effect of H 2 O 2 on S. aureus transcriptome might provide new therapeutic targets within S. aureus, which is now considered as a high priority pathogen for the development of new treatments by the WHO.

Bacterial Strains, Culture Medium and Growth Analysis
All bacterial strains used in this study are summarized in Table 1. All strains were cultured at 37°C in tryptic soy broth (TSB) or on tryptic soy agar (TSA). When antimicrobial agents were indicated, they were added to the following concentrations: erythromycin (Sigma-Aldrich) 5 µg ml -1 and chloramphenicol (AppliChem) 10 µg ml -1 . For growth restoration of the carA mutant strain, the medium was supplemented with uracil 2.5 mM (Sigma-Aldrich).
Growth dynamics was followed on 384-wells plate (Corning #3640) using a plate reader Infinite 200 Pro (Tecan) at 37°C with an orbital shaking of 5 mm. The absorbance was measured every 6 minutes at optical density 595 nm (OD 595nm ).

In Vitro Exposure to H 2 O 2
After an overnight culture, S. aureus was inoculated in TSB to an OD 595nm of 0.01 and 50 µL of the bacterial culture were dispensed on a 384-wells plate (Corning #3640) and incubated RNA Isolation, RNA-Seq and qRT-PCR Total RNA was extracted as described previously by Fischer et al. (25). Briefly, overnight cultures were diluted to an OD 595nm of 0.01 and grown in fresh TSB for 3h at 37°C with constant shaking (180 rpm). Bacteria were washed and lysed with lysostaphin 25 µg/ml (Sigma) and RNA was isolated with RNeasy mini kit plus (Qiagen). gDNA was removed with DNase I digestion as previously described by Schuster et al. (26). Total RNA was quantified with Cubit or Nanodrop. RNA-Seq analysis was made as described by Mikheyeva et al. (27). Briefly, the RNA integrity was determined using Agilent 2100 BioAnalyzer (Agilent Technologies) to verify the quality of extracted RNA and one microgram of total RNA was ribo-depleted with the Ribo-Zero kit (Illumina). For the library preparation, the truseq total RNA stranded was used. Using Illumina HiSeq 4000 sequencer, oriented 50 bases single-read sequencing was performed. Finally, RNA-Seq and data analysis were carried as described in Cherkaoui et al. (28). Low quality reads and reads containing adapter and poly-N were removed and remaining reads were aligned on USA300 genome (accession number: CP000255). For qRT-PCR, total RNA was reverse transcribed using PrimerScript Reverse Transcriptase (TaKaRa) and genes were quantified using a Brilliant SYBR green master mix (Agilent). The primers used in this study are described in Table 2. For RNA isolated after phagocytosis, a preamplification with the TaqMan ™ PreAMp Master Mix Kit (Applied Biosystems) was performed. Quantitative PCR (qPCR) reactions were performed in a Bio-Rad CFX96 and normalized using intensity levels recorded for the hu gene as previously described in Garzoni et al. (29).

Isolation of Human Neutrophils
Human neutrophils were purified from 10 ml citrated blood samples of healthy donors after obtaining their informed consent. Total blood cells were separated by sequential Ficoll-Hypaque differential density centrifugation as described in Genestet et al. (32). Neutrophils in the pellet were resuspended in phosphate-buffered saline (PBS) solution. Erythrocytes were lysed twice with a frozen hypotonic lysis buffer as described by Dri et al. (33). Briefly, the lysis buffer contained 155 mM NH 4 Cl, 10 mM KHCO 3 , 0.1 mM EDTA for a pH of 7.4. After erythrocytic lysis, cells were centrifugated for 10 minutes at 460 g at 4°C and neutrophils were resuspended in Ca +2 and Mg +2 free HEPES-buffered saline (HBS) solution containing 140 mM NaCl, 5 mM glucose, 5 mM KCl, 5 mM HEPES and 0.2% of bovine serum albumin (BSA). Cells were kept on ice until use. Directly before use, cell suspensions were supplemented with 1 mM CaCl 2 and 1 mM of MgCl 2 .

Microbicidal Activity Assay With Human Neutrophils
The microbicidal activity of neutrophils was assessed according to the method described in Decleva et al. (34). In brief, 4x10 6 neutrophils/ml were incubated at 37°C with shaking 160 rpm with serum-opsonized S. aureus at a bacteria/neutrophil ratio of 3:1. After one hour of incubation, aliquots were diluted 50 times in water with NaOH (pH 11) for 5 minutes in order to lyse neutrophils. Then, tubes were vortex and diluted into 0.9% NaCl solution and plated on Petri dishes. The next day, CFU were counted and the percentage of killing was calculated according to the number of CFU at T0.

Hydrogen Peroxide Significantly Affects the Expression of Genes Involved in Pyrimidine Metabolism
We first investigated growth of S. aureus JE2 strain at different H 2 O 2 concentrations ( Figure 1). H 2 O 2 was added at the early exponential phase (OD 595nm 0.25) and the effect on growth  Figure 1B). The impact of sublethal H 2 O 2 concentration on gene expression in S. aureus was addressed by RNA-Seq. H 2 O 2 was added at the early exponential phase and after one hour of exposure to 20 mM (or 0 mM for control conditions), total bacterial RNA was extracted and their respective transcriptome was determined. This time point was chosen in order to study bacterial response after the immediate stress response such as catalase expression. Figure 2 illustrates the differential expression analysis for S. aureus treated with H 2 O 2 versus untreated. We observed 190 differentially expressed genes with a fold change ≤ -2 and ≥ 2 with statistical significance (False Discovery Rate (FDR)). Among these genes, 98 were upregulated and 92 downregulated ( Table 1 and Supplementary Data). Genes with unknown function (n=85, 44.7%) were removed from Table 1. In order to identify gene categories affected by the presence of H 2 O 2 , we analyzed RNA-seq data based on TIGRFAM gene categories. Figure 2A illustrates the number of differentially expressed genes on the total number of genes in the corresponding TIGRFAM main role. TIGRFAM is a protein families database designed to support genome annotation (35,36). Only genes with expression that were statistically (FDR<0.05) and significantly changed (fold change ≤-2 or ≥2) were analyzed. Pathways including DNA replication, recombination and repair and DNA interactions had the highest number of genes upregulated, followed by riboflavin, FMN and FAD pathways. The upregulated genes belonged to various families, including genes involved in riboflavin metabolism [typically involved in redox response (37)], phage genes, genes involved in DNA repair, as well as several gene encoding non-identified hypothetical proteins. The downregulated genes belonged to different gene families, including degradation of proteins, peptides, and glycopeptides, ribosomal proteins or toxin production and resistance. Interestingly, the category of genes involved in pyrimidine ribonucleotide biosynthesis was the pathway with the most important downregulation. There was an almost complete downregulation of the different genes involved in the pyrimidine ribonucleotide biosynthesis (fold changes ranging from -7 to -2) ( Figure 2B). We confirmed a dose-dependent downregulation observed in the RNA-Seq by qRT-PCR ( Figure 2C). We decided to focus our research on this gene family.
Inactivation of carA and pyrP Genes Affects S. aureus Growth Next, we addressed the impact of inactivation of pyr genes on the fitness of S. aureus. To study this question, we used a library of mutant strains, the Nebraska Transposon Mutant Library (NTML) (38,39), to address the question of inhibition of the pyrimidine pathway. We used six mutants of the pyrimidine biosynthesis pathway: carA, pyrB, pyrD, pyrE, pyrF, pyrP. The biosynthesis of pyrimidine is divided into two different pathways: i) the de novo pathway, which uses glutamine, ATP and bicarbonate for uridine monophosphate (UMP) synthesis and, ii) the salvage pathway, which uses extracellular uracil to synthetize UMP ( Figure 3A). Several pyr genes such as pyrP, pyrB, pyrC, carA, pyrF and pyrE are located on the same operon and transcribed from a single promoter ( Figure 3B).
We first investigated the growth of pyr mutants and we observed two mutants displaying a growth defect. Growth was strongly affected for carA and pyrP mutants ( Figures 4A, B). The growth of pyrB, pyrD, pyrE, pyrF mutants was not affected ( Figures 1A-D and Supplementary Data).
To analyze the effect of H 2 O 2 on each pyr mutant, we exposed each mutant to eight different H 2 O 2 concentrations, ranging from 2.5 mM to 160 mM of H 2 O 2 for one hour and observed the effect on their growth rate. The growth rate was normalized according to the growth rate in absence of H 2 O 2 and compared with the parental strain. The estimated IC 50 for each mutant is represented on each graph. pyrB, pyrD, pyrE and pyrF mutants have a H 2 O 2 sensitivity close to the parental strain with an IC 50 respectively at 80.5 mM, 73.5 mM, 72.7 mM and 46.7 mM respectively ( Figure 2 and Supplementary Data). In contrast, pyrP and carA mutants were more sensitive to H 2 O 2 than the parental strain with IC 50 estimated to 14.5 mM and 11.4 mM ( Figures 5A, B).
Among all pyr mutants found in the NTML, we observed that loss of function of two pyr genes, carA and pyrP, induced a drastic alteration of bacterial fitness. The carA gene encodes for the small subunit of the carbamoyl phosphate synthase, a subunit of a large enzymatic complex associated with CarB, initiating the first reaction in pyrimidine and arginine metabolism (40) ( Figure 3A). The pyrP gene encodes for an uracil permease located in the bacterial membrane, which transports uracil into the cell (41) ( Figure 3A) and methods). Bacterial growth and sensitivity to H 2 O 2 was further analyzed. CarA complementation restores carA mutant growth, similar to wild-type JE2 strain ( Figure 4B). Moreover, the complemented strain showed a strong decreased sensitivity to H 2 O 2 with an IC 50 of 107.6 mM, compared to the IC 50 of the carA mutant (IC 50 = 11.42 mM) ( Figure 5B). These results show that loss of function of carA gene not only affects bacterial growth but also plays a role in the sensitivity to H 2 O 2 .

Uracil Restores Growth Defect in carA Mutant
As uracil is a common and natural pyrimidine derivative, we supplemented the culture medium with uracil. As depicted in Figure 4C, 2.5 mM uracil led to a substantial recovery of growth of the carA mutant. We tested higher concentrations of uracil, but they did not further enhance bacterial growth and even had a deleterious effect on growth above 5 mM (data not shown). Note that the recovery was delayed and incomplete suggesting that this compensatory mechanism might be saturated and does not fully compensate for the loss of the carA gene. carA is also involved in arginine metabolism. We therefore tested different concentrations of arginine and did not observe any growth restoration ( Figure 4D).

H 2 O 2 Induces a Decreased DNA Replication in S. aureus
We then wanted to study if the downregulation in pyrimidine metabolism observed after addition of sublethal concentration of H 2 O 2 was due to a decreasedbacterial DNA replication. We performed an EdU-Click labeling assay and compared the incorporation of the nucleotide analog 5-ethynyl-2'deoxyuridine (EdU) in S. aureus JE2 with and without 20 mM of H 2 O 2 by flow-cytometry ( Figure 6).  (Figures 6B, C). These results suggest that the effect H 2 O 2 on genes involved in pyrimidine metabolism leads to a decreased DNA replication.  carA Gene Is Not Involved in Survival After Neutrophil Phagocytosis Due to the increased sensitivity of carA mutant to H 2 O 2 , we investigated the impact of the lack of carA expression in a more complex model than in vitro H 2 O 2 exposure. We first analyzed the expression of carA gene after one hour of phagocytosis ( Figure 7A) and observed downregulation similarly to what we observed in vitro. Then, we assessed the survival of JE2, carA mutant and carA complemented strains after phagocytosis by human neutrophils. Opsonized S. aureus strains were incubated with freshly isolated human neutrophils for 1 hour and their survival was measured by counting the colonies on agar plate after 24 hours, in presence or absence of an irreversible NOX2inhibitor (DPI). The number of CFUs was normalized according the CFUs number at T0. We observed that in presence of neutrophils, the survival rate dropped around 40% for all the three strains ( Figure 7B). The role of the ROS generating NOX2 was confirmed by the fact that the killing activity of neutrophils was inhibited by 5 µM of DPI. Therefore, the absence of carA expression did not affect the bacterial survival after phagocytosis in presence and in absence of ROS.

DISCUSSION
The aim of this study was to understand why ROS are crucial in the host defense against S. aureus. We observed important changes in gene expression at sublethal H 2 O 2 concentrations, in particular a concerted downregulation of enzymes of pyrimidine metabolism. This downregulation was caused by a decreased in DNA replication and was strongly associated with decreased extracellular, but not intracellular survival of the bacteria. We hypothesize that this is a long-term survival mechanism of S. aureus: avoiding growth of extracellular S. aureus which can cause severe infection and lead to death of its host, while assuring its intracellular survival. After one hour of exposure to a sublethal concentration of H 2 O 2 , the number of upregulated and downregulated genes was approximately equal, suggesting that under our experimental conditions, H 2 O 2 does not lead to unspecific toxicity, but rather induces a cellular response. Interestingly among upregulated genes, the expression of genes involved in antioxidant response was unchanged, most likely because they are expressed within minutes after the onset of the oxidative stress and already decreased to normal levels one hour after addition of H 2 O 2 . Our hypothesis was that H 2 O 2 affects S. aureus general fitness and we decided to focus on downregulated genes. Among downregulated genes, we observed that most of the genes involved in pyrimidine biosynthesis were strongly affected in S. aureus following one hour of exposure to a sublethal concentration of H 2 O 2 . We identified two genes in this family, carA and pyrP, that were important for bacterial growth in rich medium and H 2 O 2 sensitivity. We observed impaired DNA replication after H 2 O 2 exposure. This decrease in DNA replication and transcription was dependent on the downregulation of pyrimidine biosynthesis. This concerted response represents a compensation mechanism due to the presence of H 2 O 2 ensuring bacterial survival. Indeed, similar  amount of H 2 O 2 resulted in bacterial death of a carA mutant, a noncompensable mutation. We then confirmed that the expression of carA was also decreased after one hour of phagocytosis by human neutrophils. Furthermore, the fitness of the carA mutant was particularly affected and could be reversed by genetic rescue. Yet, survival of carA mutant after phagocytosis by human neutrophils was not different from the parental strain. Pyrimidine metabolism leads to the formation of pyrimidine nucleotides (namely uracil, cytosine and thymine), which are used for nucleic acid synthesis (44), energy production (UTP or CTP) and other key cellular functions (45). Both carA and pyrP are involved in the first step of two pathways, de novo and salvage pathways, and, unlike the other genes in the pathway, they are crucial, and no alternative pathway can substitute their function. We think that in nutrient rich media, the salvage pathway is working, as bacteria are able to grow, however it is less efficient than the de novo synthesis pathway, explaining why the growth of the carA mutant is delayed (but not inhibited as can be seen in minimal media).
It has been observed that carA plays an important role in virulence in other bacterial species, such as Pseudomonas syringae (46), Escherichia coli (47), Xanthomonas citri (48) and Francisella tularensis (49). Similar to the genomic organization characterized in B. subtilis, several pyr genes (pyR, pyrP, pyrB, pyrC, carA, carB, pyrF and pyrE) of S. aureus are located on an operon and transcribed from a single promoter ( Figure 3B). The transcription of the operon is negatively regulated by the binding of PyrR to specific anti-termination sites [described in detail in Turnbough and Switzer (50)]. PyrR peptide sequence does not contain cysteines redox-sensitive residues (44). However, two known redox-sensitive transcription factors, MgrA and SarZ, are known to negatively regulate expression of pyrimidine genes under oxidative stress (9,51). Both regulators have a direct effect on pyrR but MgrA also controls the expression of genes involved in biosynthesis of another pyr gene that are not located on pyr operon (51), such a pyrG, which was also downregulated in our transcriptomic analysis. This finding supports a potential role of MgrA through an additional transcription factor. The specific redox regulation of MgrA and SarZ effect on the downregulation of genes involved in pyrimidine biosynthesis pathway could be specifically addressed using redox biochemical approach, such as the biotin switch assay (52).
Nucleotide clumping of the bacterial genome has been described as another type of bacterial response to oxidative A B FIGURE 7 | carA expression and S. aureus survival 60 minutes after phagocytosis by human neutrophils. (A) Expression analysis of carA with qPCR one hour after phagocytosis by human neutrophils. Gene expression was normalized with the gene expression in non-phagocytosed bacteria. These results represent the mean ± SD of two independent experiments. (B) Bacterial survival was estimated after 60 minutes of co-culture with human neutrophils with a multiplicity of infection (MOI) of 3. ROS production inside the phagosome was either conserved (blue) or inhibited (green) by 5 µM of diphenyleneiodonium (DPI). CFU counts following the experiment were normalized according CFU count from the initial inoculum. These results represent the mean ± SD of three independent experiments. ns, not significant, p>0.05 by Kruskal-Wallis test. stress (53). This response is regulated by MgrA (53) and participates to bacterial stress tolerance by protecting the genome against ROS. Interestingly, the effect of H 2 O 2 on genes involved in pyrimidine biosynthesis pathway is not specific to S. aureus and a study in E. coli demonstrated that H 2 O 2 also induces a downregulation in genes involved in nucleotides and ribonucleotides production process (54). Intriguingly, we observed a downregulation in carA expression following intraphagosomal oxidative stress, but we did not observe any difference in the survival of carA mutant and the parental strain, in the presence or absence of intraphagosomal ROS. Two aspects of this difference are relevant for this discussion: i) the underlying biochemical mechanism, and ii) the impact on the S. aureus reservoir for long-term survival as observed in numerous chronic infections.
A possible biochemical explanation for the decreased expression of carA in intracellular survival would be a high level of pyrimidine or pyrimidine precursors, in particular uracil, within the phagosome. To the best of our knowledge, phagosomal concentrations of the respective metabolites have not been measured. Another explanation would be changes in metabolism of intracellular bacteria leading to a decreased catabolism of pyrimidine or pyrimidine precursors. A third aspect that might contribute to the decreased importance of carA for phagosomal survival might be the decreased growth rate of intracellular S. aureus leading a decreased requirement of nucleotide precursors. This type of behavior has been observed during prolonged residency of S. aureus in non-phagocytic cells (55).
In summary, we demonstrated a novel and selective mechanism allowing S. aureus to survive in the presence of ROS, namely a coordinated downregulation of the pyrimidine biosynthesis pathway and a decreased in DNA replication. This downregulation has the interesting feature to selectively interfere with the growth of extracellular, but not intracellular S. aureus. Given this dichotomy, our results suggest that S. aureus is not a naive victim of host defenses in this situation but has developed an evolutionary survival strategy by modulating growth rate. Future studies should investigate the mechanisms described here in clinical strains. In our study where RNA-Seq data needs to be interpreted in the context of the entire genome, the choice of the JE2 strain was pertinent and allowed to uncover this novel S. aureus survival strategy. The importance of this "go unnoticed" strategy may contribute to the high mortality of CGD patients following S. aureus infection [at least if untreated (7)] and contribute to their long-term epidemiological success in the numerous people carrying S. aureus.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ebi.ac. uk/ena, PRJEB43496.

ETHICS STATEMENT
Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. The patients/participants provided their written informed consent to participate in this study.