Experimental evolution selects for PYO tolerance in S. aureus

We first determined the bactericidal effect of PYO on the S. aureus strain JE2 by quantifying survival of exponential phase cells upon treatment with a range of PYO concentrations. PYO production by P. aeruginosa can range from 0 to ~30 µg/mL (~150 µM) depending on medium composition and infection site [31,44] although higher concentrations have been observed [30,45] and are likely to exist in local microenvironments. We tested PYO concentrations that covered this range and observed a concentration-dependent effect of PYO on S. aureus cell density, including moderate growth reduction at 12.5 and 25 µM, growth inhibition at 50 and 100 µM, and killing at 200 and 400 µM (Fig 1A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Experimental evolution of S. aureus selects for enhanced survival in PYO.

(A) Early exponential phase S. aureus cells were treated with the indicated concentration of PYO. Values indicate S. aureus viable cell counts after 20 hours of PYO exposure. Data shown are the geometric mean ± geometric standard deviation of four biological replicates. (B) Schematic of experimental evolution to select for PYO-tolerant S. aureus. Overnight cultures were grown to early exponential phase prior to addition of 200 µM PYO. After incubation for 20 hours, viable cell counts were enumerated, and the remaining culture was grown overnight in the absence of PYO. This process was repeated six times for two independent populations. (C) Viable cell counts of two independently evolved populations, after each treatment with 200 µM PYO. Yellow fill indicates the passage at which isolates were selected. (D, E) Selected isolates (indicated on the x-axis) from populations (D) ‘A’ or (E) ‘B’ were assayed for tolerance to 200 µM PYO. Data shown are the geometric mean ± geometric standard deviation of three biological replicates. (A, C, D, E) The dashed line indicates the mean initial cell density (CFU/mL) for all strains at the time of addition of PYO or DMSO as a control (0 µM PYO). Significance is indicated for comparison to the DMSO control (0 µM PYO) (A), or WT (D, E) as determined by a one-way ANOVA using Dunnett’s correction for multiple comparisons (* P < 0.05, ** P < 0.01, *** P < 0.001, # P < 0.0001).

https://doi.org/10.1371/journal.pgen.1011610.g001

To identify potential adaptations that increase survival of S. aureus upon PYO exposure, we decided to use experimental evolution of cells via repeated exposures to 200 µM PYO – the lowest bactericidal concentration identified. We evolved S. aureus by treating early exponential phase cells with PYO for 20 hours, recovering the surviving cells overnight in medium without PYO, and repeating this process over several iterations (Fig 1B). During the evolution, the populations did not grow in PYO, and we estimate that each recovery phase consisted of 6-13 generations. Because we observed loss of cell viability with this concentration of PYO and recovered surviving cells, our expectation was that we would identify mutants exhibiting increased survival during treatment with PYO. We evolved two independent populations (‘A’ and ‘B’) to identify mutations specific to PYO tolerance, as these mutations should be common to both evolved populations. As the number of treatments increased, we observed enhanced survival of both independent populations when treated with PYO (Fig 1C). Individual isolates from the evolved populations (designated A1, A2, B1, etc.) also exhibited higher survival following PYO exposure, indicating that these evolved strains had acquired increased tolerance to P. aeruginosa-derived PYO (Figs 1D, 1E and S1 Fig). The selection of strains exhibiting reduced cell death, rather than a continuation of growth, during treatment with PYO suggested that we were selecting for PYO tolerance, instead of resistance.

Loss-of-function mutations in the CodY global regulator confer tolerance to PYO

Next, we sought to identify common mutations that characterize PYO-tolerant isolates from each evolved population, and thus sequenced and analyzed genomes from 14 terminal isolates (7 isolates each from populations A and B). Due to the diversity of mutations observed in the first 7 isolates of population A, we selected an additional 4 for sequencing, bringing the total to 18 sequenced isolates. While we observed diverse mutations among different isolates (S1 Data), each of the 18 isolates had at least one mutation associated with the codY gene (S1 Table), which encodes a well-characterized pleiotropic transcriptional regulator conserved across gram-positive bacteria [43]. In S. aureus, CodY regulates the expression of virulence and metabolic genes in response to nutritional cues [46,47]; however, a role in modulating tolerance to interspecies antimicrobials has not, to our knowledge, been described. Coding sequence mutations observed in our evolved isolates were present in both the substrate-sensing and DNA-binding domains, and we also observed mutations in the promoter region and the start codon (Fig 2A and S1 Table). Considering the diversity of mutations observed in our evolved isolates, we expect that these mutations arose during the recovery phase of the evolution and were selected for during PYO treatment. Isolation of the R61K mutation, which has previously been described to substantially reduce CodY activity [46], and an ablated start-codon, as well as the diversity of mutations across both functional domains suggested that the mutations we observed in our evolved isolates likely resulted in loss of CodY function.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Loss of CodY function confers PYO tolerance.

(A) Schematic of CodY-associated mutations observed in the evolved isolates and where they reside within the protein and upstream sequence. The CodY^R222C allele is recapitulated in the codY* mutant. (B) Viable cell counts of the WT and codY* mutant after 20 hours of treatment with either DMSO as a control or 200 µM PYO. Data shown are the geometric mean ± geometric standard deviation of three biological replicates. (C) Viable cell counts of the WT, codY* mutant, and a transposon mutant of codY (codY::Tn) after 20 hours of treatment with 200 µM PYO. Data shown are the geometric mean ± geometric standard deviation of four biological replicates. (B, C) Dashed lines indicate the mean initial cell density (CFU/mL) for all strains at the time of PYO or DMSO addition. Significance is shown for comparison to the respective WT condition as tested by a (B) two-way ANOVA using Tukey’s correction or (C) one-way ANOVA using Dunnett’s correction for multiple comparisons (# P < 0.0001).

https://doi.org/10.1371/journal.pgen.1011610.g002

To determine if a CodY mutation is sufficient to recapitulate the PYO tolerance phenotype of our evolved isolates, we reconstructed one of our evolved alleles CodY^R222C (hereafter referred to as codY*), in the parental strain. When the codY* mutant was treated with 200 μM PYO, we observed significantly greater survival (~100-fold) compared to the WT (Fig 2B), while no difference was seen upon exposure to the DMSO control, suggesting that the codY mutations were selected for due to the PYO exposure during the evolution, and likely did not cause adaptation to just the media conditions. In addition, a mutant with a transposon insertion in codY knocking out CodY activity (from the Nebraska Transposon Mutant Library [NTML] [48]) phenocopied the codY* mutant when treated with PYO (Fig 2C), providing further evidence that loss of CodY function mediates PYO tolerance. The survival of the codY* mutant was similar to that of the evolved isolates (Figs. 1D and 1E), indicating that codY mutations in these isolates likely cause the PYO tolerance phenotype.

We also tested whether loss of CodY function confers resistance to PYO, allowing for growth at higher PYO concentrations. We found that the codY* mutant exhibited greater growth than the WT at relatively low concentrations of PYO (12.5 and 25 µM), while a PYO concentration of 50 µM almost completely inhibited growth for both the WT and the codY* mutant (S2 Fig). These data suggest that while a codY* mutation confers some resistance to low concentrations of PYO, selection of codY mutations in the experimental evolution was largely due to increased survival and tolerance to 200 μM PYO. In addition, because we observed this moderate enhancement of PYO resistance in the codY* mutant (S2B and S2C Figs, Fig 2C), we tested whether a previously identified PYO resistance mutation would also engender increased PYO tolerance. It has been shown that loss of QsrR, a quinone-sensing repressor of quinone detoxification genes [49], results in PYO resistance in S. aureus [40], allowing for enhanced growth in up to 32 µM PYO compared to the parental strain. While our experimental conditions and growth medium are different, we observed greater sensitivity of the ΔqsrR mutant to 50 – 200 µM PYO compared to the WT (S3 Fig), indicating that although loss of QsrR allows for increased growth in the presence of lower concentrations of PYO, it does not confer increased survival or tolerance to higher concentrations of PYO in our assay conditions. Together, these data indicate that CodY loss-of-function mutations enhance S. aureus survival in the presence of high concentrations of PYO via a mechanism distinct from previously identified adaptive mutations.

The codY* mutation confers protection against PYO-dependent killing in co-culture with P. aeruginosa

P. aeruginosa is known to produce a variety of additional molecules that may have antimicrobial effects, including against S. aureus [15]. We tested whether the codY* mutation conferred increased tolerance to several such P. aeruginosa antimicrobials. We saw either no difference or a minor increase (1.5- to 2-fold) in survival or growth between WT and the codY* mutant in the presence of the phenazines 1-hydroxyphenazine, or phenazine-1-carboxylic acid, the alkyl quinolones PQS, or HHQ, the biosurfactant rhamnolipids, or the siderophores pyoverdine, or pyochelin (S4A-S4C, S4E Fig). However, we did observe an approximately 10-fold increase in survival of the codY* mutant when treated with HQNO (S4D Fig), an alkylquinolone that also inhibits respiration in S. aureus [50,51], indicating some overlap in tolerance to another P. aeruginosa toxin.

Further, to determine the effect of the codY* mutation in a co-culture of S. aureus-P. aeruginosa, we first tested whether PYO and other phenazines can contribute to competitive killing of S. aureus. We performed co-culture experiments of WT S. aureus with the P. aeruginosa strain PA14 and isogenic mutants deficient for either both phenazine biosynthesis operons phzA1-G1 and phzA2-G2 (Δphz1/2) or just PYO (ΔphzM) in media modified to enhance PYO production [44]. When co-cultured with the parental P. aeruginosa PA14 strain, we observed an approximately 10-fold decrease in S. aureus survival (Fig 3A), that was not seen in co-culture with either the Δphz1/2 or ΔphzM mutants (Fig 3A), suggesting that PYO may be a major contributor to S. aureus killing in these conditions. Finally, we observed that the codY* mutation was protective in co-culture with the parental P. aeruginosa strain compared to WT S. aureus (Fig 3B). Together, these data suggest that codY* contributes to S. aureus survival against PYO-dependent killing in co-culture with P. aeruginosa.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. The codY* mutant is tolerant to PYO-dependent killing by P. aeruginosa in co-culture.

(A) Viable cell counts of S. aureus JE2 after 18 hours of co-culture with P. aeruginosa WT, a phenazine-null mutant (Δphz1/2), or a PYO-null mutant (ΔphzM). (B) Viable cell counts of S. aureus WT or the codY* mutant after 18 hours of co-culture with P. aeruginosa PA14. (A, B) Data shown are the geometric mean ± geometric standard deviation of three biological replicates and the dashed lines indicate the mean initial cell density of all S. aureus strains. Significance is shown for comparison to the (A) PA14 condition or (B) WT condition as tested by (A) one-way ANOVA using Dunnett’s correction for multiple comparisons or (B) unpaired, two-tailed t-test (# P < 0.0001).

https://doi.org/10.1371/journal.pgen.1011610.g003

A CodY^R222C mutant exhibits expected transcriptional changes in metabolism and virulence gene expression

CodY regulates a substantial proportion of the S. aureus genome based on branched-chain amino acid (isoleucine, leucine, and valine) and nucleotide (GTP) availability [47,52]. In the presence of sufficient intracellular concentrations of these nutrients CodY is active and functions primarily to repress its target genes [46]. As nutrients become scarce, CodY activity decreases, thereby facilitating the expression of amino acid transport and biosynthesis genes and host-targeting virulence factors.

To identify differentially expressed genes between the two strains that may explain the increased PYO tolerance of the codY* mutant, we determined the transcriptional response of the WT and codY* mutant to PYO at early (30 minutes) and late (120 minutes) time points (S2 Data). The gene expression changes in the codY* mutant compared to WT under control conditions at 30 minutes were consistent with previous reports of the CodY regulon [46,47] (S5 Fig and S2 Data) indicating that CodY^R222C disrupts CodY-dependent regulatory activity, similar to a codY null mutant. We observed no difference in gene expression at 120 minutes, likely due to the natural de-repression of CodY at this time point in the WT strain.

In addition, in the codY* mutant we observed overexpression of the agr quorum sensing system (S5B Fig) that regulates virulence gene expression [53,54], and can impart long-lived protection from oxidative stress [55]. However, we did not observe enhanced PYO sensitivity of an agrA::Tn mutant and a codY* agrA::Tn double mutant exhibited protection against PYO comparable to the agr competent strains (S6 Fig), suggesting that overexpression of agr in the codY* mutant is not responsible for increased PYO tolerance.

S. aureus responds to PYO by inducing stress response and iron acquisition genes and suppressing metabolism

Considering that S. aureus and P. aeruginosa are frequently found to co-infect patients, and S. aureus may thus be exposed to PYO, we determined the response of the WT S. aureus JE2 strain to PYO. Enriched pathways among the PYO-induced genes included iron acquisition (siderophore and heme metabolism), energy generation (menaquinone biosynthesis and the TCA cycle), and stress response to oxidative stress, metals, and DNA damage (Fig 4), indicating that PYO likely generates reactive oxygen species that are known to affect metal homeostasis [56–58], causes DNA damage, and leads to metabolic alterations in S. aureus. While the codY* mutant overall showed a similar response to PYO (S7 and S8 Figs), at the early time point we observed broad repression of amino acid and nucleic acid metabolism, energy generation, and translation-associated pathways (S7A and S7B Figs), suggesting that PYO rapidly elicits a more metabolically dormant state in the codY* mutant, a phenomenon that has been associated with antibiotic tolerance and persistence [36,59].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. S. aureus increases expression of genes associated with iron acquisition and stress responses while suppressing metabolism- and translation-related genes in response to PYO.

Differential gene expression of WT S. aureus in response to 200 µM PYO after 30 (A, B) and 120 (C, D) minutes of incubation in PYO. (A, C) Enriched GO pathways at 30 and 120 minutes, respectively, among upregulated and downregulated genes and their fold enrichment relative to the expected number of observed genes. (B, D) Volcano plots of log₂(fold change gene expression) and -log₁₀(false discovery rate) at 30 and 120 minutes, respectively. Upregulated genes are shown in light red and downregulated genes are shown in light blue. Further highlighted genes indicate the over- and under-expressed genes comprising the associated functional pathways in the legend (see S3 Data for a list of the included genes). Enriched GO pathways (C) and differentially expressed genes (D) at 120 minutes include only those not also observed in the codY* mutant compared to WT from the 30-minute DMSO comparison (S5 Fig), but a full list can be found in S8 Fig and as part of S2 Data.

https://doi.org/10.1371/journal.pgen.1011610.g004

We also noted that cidA, part of the cidABCR operon that plays a central role in controlling carbon metabolism and programmed cell death [60,61], was the most overexpressed gene in the WT JE2 strain at both early (~111-fold) and late (~29-fold) time points, while cidB and cidC were also highly upregulated (Fig 4B and 4D). We therefore tested whether any of the cid genes were involved in PYO-dependent cell death or tolerance. We observed no effect for mutations in any of the cidABCR genes in either the WT or codY* backgrounds (S9 Fig), suggesting that, despite its strong overexpression, the cid operon does not contribute to cell death or survival in these conditions.

Since we also observed enrichment of DNA repair pathways among upregulated genes, we determined if PYO treatment led to DNA damage using a TUNEL assay. We observed significant DNA damage after 2-, 4-, and 20-hour treatment with PYO in the WT strain (S10A Fig). Further, while WT exhibited a moderate, but not statistically significant, increase in TUNEL staining compared to the codY* mutant in PYO at 2 hours, there was no significant difference at other time points (S10B and S10D-F Figs). Since we do not observe killing following 2- and 4-hour treatment with PYO (S10C Fig), these data suggest that while PYO does induce DNA breaks, these likely do not explain the differential PYO tolerance between WT and the codY* mutant.

The codY* mutant exhibits greater suppression of translation and expression of stress response genes than the WT

We next directly compared gene expression between the codY* mutant and WT in the PYO condition to identify transcriptional differences that could potentially explain the PYO tolerance of the codY* mutant. Apart from CodY-dependent amino acid metabolism pathways, at the early time-point, translation was enriched among the downregulated genes (Fig 5A and S11A Fig), consistent with the greater metabolic suppression seen in the codY* mutant at this time-point. Further, several genes and pathways potentially involved in stress responses were upregulated in the codY* mutant. These included pxpBCA encoding 5-oxoprolinase that converts 5-oxoproline to glutamate [62], and gltBD encoding glutamate synthase, both of which generate glutamate that has been implicated in the response to oxidative and other stresses [63–65], and adhC encoding alcohol dehydrogenase, that has been implicated in resistance to oxidative and nitrosative stress [66,67], which were consistently upregulated at both time points (Fig 5A and 5B). Additionally, the carotenoid biosynthesis pathway, and catalase-encoding katA, which are both important for resistance to oxidative stress [56,68] were upregulated at the later time-point (Figs 5B and S11 Fig) suggesting that the codY* mutant has a more robust stress response to PYO which may underlie its increased survival.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. The codY* mutant exhibits enhanced expression of amino acid metabolism and stress response genes and downregulates translation compared to the WT in response to PYO.

Volcano plots showing log₂(fold change gene expression) and -log₁₀(false discovery rate) in response to 200 µM PYO in the codY* mutant compared to WT after 30 (A) and 120 (B) minutes. Highlighted genes indicate the over- and under-expressed genes comprising the associated functional pathways (see S3 Data for a list of the included genes) or selected genes associated with stress responses. Enriched GO pathways for the respective volcano plots can be found in. S11 Fig.

https://doi.org/10.1371/journal.pgen.1011610.g005

ATP depletion protects S. aureus from the bactericidal effects of PYO

Given the increased metabolic suppression in the codY* mutant upon PYO exposure, and that ATP depletion in S. aureus and E. coli is associated with the formation of persister cells and reduced susceptibility to antibiotics [69,70] we hypothesized that metabolic quiescence may promote protection from PYO. We therefore tested the effect of proton motive force decoupling agents carbonyl cyanide m-chlorophenyl hydrazone (CCCP), a protonophore that reduces ATP synthase activity [71,72], or sodium arsenate, which reduces ATP by forming unproductive ADP-arsenate [69,73], on PYO tolerance. Treatment with either 10 μM CCCP or 30 mM arsenate resulted in WT PYO tolerance comparable to that of the codY* mutant (Fig 6A). We observed no additional protective effect of ATP depletion via addition of 10 μM CCCP on PYO tolerance of the codY* mutant (S12 Fig). While CCCP did not interfere with growth in the absence of PYO, arsenate arrested growth at the point of addition.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Metabolic suppression and hydrogen peroxide stress response mediate PYO tolerance.

Viable cell counts of the indicated S. aureus strains after 20-hour treatment with the stated reagents. (A) Tolerance to 200 µM PYO of the S. aureus codY* mutant, and WT in the presence of the ATP depleting agents CCCP (10 µM) and sodium arsenate (30 mM). (B) Tolerance to 200 µM PYO of the WT and the codY* mutant in the presence of glutathione or sterile water as a control. (C) Effect of glutathione on WT survival during treatment with 200 µM PYO, 60 mM hydrogen peroxide, or 0.1 mM paraquat. (D) Susceptibility of WT and the codY* mutant to hydrogen peroxide. (E) PYO sensitivity of transposon mutants disrupting overexpressed genes from RNA-seq in WT and the codY* mutant backgrounds. (F) Overexpression of selected genes with their native promoters from a multi-copy plasmid. pKM16 expresses the fluorescent protein dsRed3.T3 from the sarA1 promoter of S. aureus and is used as a control. Supplemental reagents were added at the indicated concentrations immediately prior to addition of 200 µM PYO, DMSO, or sterile water. Dashed lines indicate the mean initial cell density (CFU/mL) for all strains prior to addition of PYO, DMSO, or other reagents, while the lower dotted lines in (C) and (E) indicate the limit of detection. Data shown are the geometric mean ± geometric standard deviation of the following numbers of biological replicates: (A) five, (B) four, (C) three, (D) four, (E) four, (F) five. Significance is shown relative to (A) the respective WT condition, (B) WT without glutathione, or the indicated strains, (C) the respective no glutathione condition, (D) the respective WT condition, (E) the respective WT mutant, or (F) pKM16, and was determined by (A-E) a two-way ANOVA using (A) Dunnett’s, (B, C, D) Tukey’s, or (E) Šídák’s correction, or (F) one-way ANOVA using Dunnett’s correction for multiple comparisons (* P < 0.05, ** P < 0.01, *** P < 0.001, # P < 0.0001).

https://doi.org/10.1371/journal.pgen.1011610.g006

Despite their metabolic dormancy, persister cells maintain some active metabolic processes [74,75] which can be necessary for the protective effects of persistence [76–78]. To determine whether CCCP-mediated protection from PYO still required active metabolism, we tested the effect of higher CCCP concentrations on both the WT and codY* mutant strains. While 40 μM CCCP did not adversely affect growth in the DMSO control, this concentration resulted in an increased susceptibility to PYO in both the WT and codY* mutant strains (S12 Fig), suggesting that active metabolism is likely required for basal PYO tolerance in the WT and increased PYO tolerance in the codY* mutant. Together, these data indicate that partial metabolic suppression in the codY* mutant may contribute to PYO tolerance.

Mitigation of hydrogen peroxide toxicity is protective against PYO

PYO toxicity is in part mediated by the generation of diverse ROS [33,79], and the oxidative stress response pathway is induced in both the WT and codY* mutant upon PYO exposure (Figs 4 and S7 Fig). To determine whether the PYO-induced cell death observed in the WT was mediated by ROS, we screened S. aureus mutants deficient in different ROS responses for sensitivity to PYO. These included mutants in the catalase (katA) and alkyl hydroperoxide reductase (ahpC, ahpF) hydrogen peroxide detoxification systems [80,81], the peroxiredoxins thiol peroxidase (tpx) and thiol-dependent peroxidases (bsaA, gpxA2) predicted to detoxify peroxides [82], both superoxide dismutases (sodA, sodM) [83], and a biosynthesis gene for bacillithiol (bshA) predicted to function in redox homeostasis [82]. Mutations in katA, ahpC, and ahpF sensitized WT to PYO-mediated killing (S13 Fig), indicating that detoxification of hydrogen peroxide is critical for PYO tolerance. Interestingly, a mutant of perR, a master repressor of the oxidative stress response [56,57], phenocopied the PYO tolerance of the codY* mutant (S13 Fig). A perR mutant constitutively expresses several stress response genes, including katA, ahpCF, dps, trxB, and ftnA [56], corroborating that an enhanced ROS response is protective against PYO.

To further test whether ROS were responsible for the killing effect of PYO, we supplemented WT with the antioxidant glutathione [84–86], and observed substantially increased survival of the WT, comparable to the codY* mutant (Fig 6B). Similar to the CCCP treatment, we observed no additional protective effect of glutathione on the codY* mutant. Glutathione supplementation also provided partial protection against hydrogen peroxide, but not paraquat-induced superoxide (Fig 6C), indicating that hydrogen peroxide is a primary mediator of cell death under these conditions, consistent with the enhanced PYO sensitivity of the katA and ahpCF mutants (S13 Fig).

In the codY* mutant, katA was overexpressed 2.2- to 2.5-fold compared to WT in DMSO (S2 Data) and in the later response to PYO (Fig 5B), suggesting that the codY* mutant exhibits an elevated basal tolerance to hydrogen peroxide and a greater response to ROS stress compared to WT. We found that the codY* mutant exhibited increased resistance to hydrogen peroxide (Fig 6D) but not superoxide (S14 Fig). Together, these data indicate that PYO-mediated toxicity is primarily mediated by hydrogen peroxide and that codY* confers greater tolerance to this ROS.

To identify the specific genes that mediate the increased tolerance of the codY* mutant to ROS, and specifically PYO, we assessed several stress response related genes whose transcripts were elevated in the codY* mutant either upon PYO exposure or compared to the WT in response to PYO. In addition to katA, ahpC, bsaA, and gpxA2, we selected a gene in the biosynthesis pathway of carotenoids (crtO) which function to alleviate hydrogen peroxide stress [68,87]; alcohol dehydrogenase (adhC) which is part of the Rex regulon responsive to redox stress and oxygen limitation [88,89]; MucB (umuC), an error-prone DNA polymerase which functions in DNA repair [90]; and glutamate synthase (gltB) since glutamate utilization can mediate protection against oxidative stress [63]. We transduced transposon mutations in each of these genes to the codY* mutant background. When challenged with PYO, we found that loss of katA (codY* katA::Tn) substantially reduced PYO tolerance and abolished the protective effects of codY*, suggesting that katA is required for PYO tolerance in the codY* mutant (Fig 6E). In contrast, while the codY* ahpC::Tn strain showed lower survival compared to the codY* mutant, it still showed higher tolerance compared to the ahpC::Tn strain.

Finally, we asked whether overexpression of these or other stress-related genes would be sufficient to confer PYO tolerance. When overexpressed from a multi-copy plasmid, we observed that katA protected WT from PYO-mediated killing at levels similar to codY* and expression of ahpCF exhibited a moderate, though not statistically significant, protective effect (Fig 6F). In contrast, overexpression of umuC, pxpBCA, or adhC was not sufficient to protect WT from PYO-mediated killing (Fig 6F).

Subsequently, we tested whether overexpression of katA would confer protection against P. aeruginosa in co-culture and similarly observed an approximately 10-fold increase in survival (S15 Fig). Taken together, these results indicate that an enhanced response to hydrogen peroxide stress is sufficient to mediate PYO tolerance and that the overexpression of katA in the codY* mutant contributes to its increased survival in PYO.

Experimentally evolved mutations in CodY are present in genomes of S. aureus clinical isolates

In our experimental evolution, we identified codY mutations in each of 18 sequenced isolates from two independently evolved populations. In total, we observed nine different mutations: seven in the coding sequence including an ablation of the start site, and two promoter mutations. Given that we did not isolate the same coding sequence mutation from both independently evolved populations, and that these mutations probably led to loss of CodY function, there is likely a large mutational space of CodY-inactivating alleles that may be selected for upon exposure to redox stress. Further, it has been shown that a codY deletion leads to increased in vivo virulence in a USA300 strain [91], suggesting that this could be an additional selective pressure for codY inactivating alleles.

To test whether codY mutations are seen in publicly available genomes of S. aureus strains, we queried the JE2 CodY protein sequence against a set of 63,983 S. aureus genomes from the NCBI Pathogen Detection Database (S4 Data). Interestingly, we identified multiple isolates which had CodY mutations that we observed in our study, including T125I, S178L, and R222C, as well as a variety of other mutations throughout the protein, suggesting that these mutations can and do arise in natural isolates (S4 Data). The mutations throughout the protein likely also reflect natural variation, and further work is required to determine the effect of such variation on CodY function, as well as adaptation to PYO and P. aeruginosa.

Bacterial strains and growth conditions

All strains and plasmids used in this study are described in S2 Table. Bacteria were cultured at 37 °C with shaking at 300 rpm in modified M63 medium [16] (13.6 g/L KH₂PO₄, 2 g/L (NH₄)₂SO₄, 0.4 µM ferric citrate, 1 mM MgSO₄; pH adjusted to 7.0 with KOH) supplemented with 0.3% glucose, 1x ACGU solution (Teknova), 1x Supplement EZ (Teknova), 0.1 ng/L biotin, and 2 ng/L nicotinamide for all experiments. For the co-culture experiments (shown in Fig 3 and S15 Fig), and experiments testing the effect of other P. aeruginosa antimicrobials (shown in S4 Fig), this medium was supplemented with 10 mg/mL tryptone, 2 µM ferric citrate, and 0.6% glycerol (M63T). Luria-Bertani (LB) broth (KD Medical) or LB agar (Difco) was used for cloning and enumerating colony forming units (CFU) and supplemented with 10 µg/mL erythromycin, 10 µg/mL chloramphenicol, 100 µg/mL spectinomycin, 50 µg/mL gentamicin, 25 µg/mL irgasan, or 100-200 µg/mL carbenicillin as required. Salt-free LB + sucrose plates were made with 10 g/L Bacto-tryptone, 5 g/L yeast extract,10% v/v sucrose, and 15 g/L agar. Pyocyanin (from Pseudomonas aeruginosa, Sigma-Aldrich [P0046]) (PYO) was resuspended to a concentration of 10 mM in dimethyl sulfoxide (DMSO), stored at -30°C, and used at the concentration(s) indicated for each experiment.

Pyocyanin and antimicrobial tolerance assays

Overnight cells were washed once in sterile PBS and normalized by their OD₆₀₀. Washed cells were then inoculated to an OD₆₀₀ of 0.1 in 1 mL of fresh, pre-warmed M63 (for PYO) or M63T (for the experiments in S4 Fig) and incubated for 2 hours in 14 mL polystyrene round-bottom test tubes. For the PYO tolerance assays shown in Figs 1D, 1E, and S1 Fig, cells were inoculated at an OD₆₀₀ of 0.05 into 125 mL Erlenmeyer flasks containing 5 mL of fresh, pre-warmed M63 and grown to an OD₆₀₀ of ~0.25. After incubation, 1 mL of culture was transferred to 14 mL polystyrene round-bottom test tubes. PYO or the indicated antimicrobial was added at the indicated concentration (or the same volume of the respective solvent was added as a control: DMSO for PYO, HHQ (Sigma-Aldrich [SML0747-10MG]), 1-hydroxyphenazine (Tokyo Chemical Industry [H0289]), and phenazine-1-carboxylic acid (MolPort [001-738-598]), ethanol for rhamnolipids (AGAE Technologies [R90-10G]), HQNO (Cayman Chemicals [15159]), and PQS (Sigma-Aldrich [94398-10MG]), or methanol for pyochelin (LGC Standards [TRC-P840365-1MG]), and pyoverdine (Sigma-Aldrich [P8124-1MG])) and the cells were incubated for an additional 20 hours. CFUs were enumerated by spot plating in triplicate. For experiments using hydrogen peroxide (Sigma-Aldrich [H1009]), paraquat (Fisher Scientific [US-PST-740]), carbonyl cyanide m-chlorophenyl hydrazone (CCCP; Sigma-Aldrich [215911-250MG]), or sodium arsenate (Sigma-Aldrich [S9663-50G]), the indicated supplement was added immediately prior to the addition of PYO.

Laboratory evolution of PYO tolerance

Two independent populations of S. aureus JE2 were grown overnight in M63, washed in PBS, and inoculated to an OD₆₀₀ of 0.05 into 125 mL Erlenmeyer flasks containing 5 mL of fresh, pre-warmed M63 and grown to an OD₆₀₀ of ~0.25. 1 mL of culture was transferred to a sterile 14 mL test tube containing 20 μL of 10 mM PYO – yielding a final concentration of 200 μM PYO – and incubated for a further 20 hours. Following incubation, 20 μL of culture was removed to enumerate CFU and the remaining culture pelleted by centrifugation. Pelleted cells were resuspended in fresh M63 and allowed to recover overnight in the absence of PYO. Overnight recovered cells were then diluted as above, and the procedure repeated up to six times. Populations after treatment 5 for Population A and treatment 7 for Population B were streaked out on LB plates, and individual colonies were selected from these to test for PYO tolerance, and for whole-genome sequencing. We evolved two independent populations to identify common mutations that likely led to PYO adaptation. Given that we identified only one gene that was mutated in common in all sequenced isolates from both populations (codY), and codY mutations led to specific PYO tolerance, we did not perform additional PYO or media-only evolutions.

Construction of S. aureus mutant strains

Primers used for cloning and verification are described in S3 Table. The locations of transposon insertions in strains acquired from the Nebraska Transposon Mutant Library (NTML) were verified by PCR and Sanger sequencing. Transductions were performed using φ11 or φ80 based on previously described procedures [120]. Briefly, overnight cultures bearing the transposon of interest were diluted 1:100 in a 125 mL Erlenmeyer flask containing 10 mL of BHI and grown for 2 hours (OD₆₀₀ ~1.0) at 37°C with shaking at 300 rpm. Then, cultures were supplemented with 150 µL of 1M CaCl₂ followed by the addition of 1-10 µL of empty phage, and incubation was continued overnight. Culture lysates were centrifuged for 10 minutes at ~4,000 rcf and the supernatant filtered with a 0.45 µm syringe filter. Subsequently, 100 to 1000 µL of phage lysate was used to transduce 1 mL of overnight culture of the recipient strain supplemented with 15 µL of 1M CaCl₂ in 15 mL conical tubes. Cultures were incubated as above for 20 minutes, at which point 200 µL of 200 µM sodium citrate was added, mixed by inversion, and centrifuged as above. Cells were then resuspended in BHI containing 1.7 mM sodium citrate, incubated as above for 1 hour, and then centrifuged as above. Transduced cells were concentrated 2-fold and 100-200 µL plated on at least three BHA plates containing 10 µg/mL erythromycin and 1.7 mM sodium citrate. Cells were allowed to grow at 37°C for up to 48 hours.

Cloning mutagenesis was performed using pIMAY* largely as previously described [121]. Briefly, the mutant codY* allele was amplified with ~900+ bp of homology on each side and cloned into pIMAY*. ~1 ug of purified plasmid isolated from E. coli DC10B was transformed into electrocompetent S. aureus JE2 and directly selected for integration by incubation at 37°C as previously described [122]. Integration was confirmed following additional overnight culture under antibiotic selection at 37°C using primers specific for plasmid integration. Integrated strains were then cultured at 28°C overnight for multiple passages without selection and plated on LB agar containing 20 mM para-chlorophenylalanine (PCPA) grown at 28°C. Chloramphenicol-sensitive colonies were screened by PCR and Sanger sequencing to identify the mutant allele.

Construction of the P. aeruginosa ΔphzM mutant

Mutants were constructed as described previously [38]. Briefly, we amplified upstream and downstream fragments flanking the phzM gene, and an FRT-site flanked Gentamycin resistance cassette (from a pPS856 plasmid template [123]). All primer sequences are listed in S3 Table. We performed overlap extension PCR of these fragments to generate the deletion construct, which was cloned into the PCR8/GW/TOPO vector (Invitrogen) by TA cloning. The deletion construct was transferred from the PCR8/GW/TOPO-ΔphzM plasmid to the pEX18ApGW vector [123] using the LR Clonase II Enzyme mix (Invitrogen) for an LR reaction. The resulting pEX18ApGW-ΔphzM plasmid was transformed into the conjugative S17-1 λ-pir E. coli strain, and then introduced into the P. aeruginosa PA14 strain using bi-parental conjugation.

For conjugations, the donor and recipient strains were grown overnight in LB (the donor strain with the appropriate antibiotic), and 500 µL of each overnight culture was used for the conjugation. The cultures were washed twice with PBS and resuspended together in 100 µL of 100mM MgSO₄. This mixture was spotted on to LB plates, incubated at 37ºC for 3 hours, scraped off with a plate spreader into PBS, and streaked on LB plates with the appropriate antibiotic to select for successful conjugants.

For the pEX18ApGW-ΔphzM plasmid, conjugants were selected on LB + irgasan + gentamicin plates, irgasan and gentamicin resistant colonies were selected on salt-free LB + 10% sucrose plates, and gentamicin-resistant carbenicillin-sensitive clones were tested for gene knockout by PCR and sequencing. A verified clone was conjugated with a S17-1 λ-pir E. coli strain carrying the pFLP2 plasmid that expresses the FLP recombinase [124], to excise out the gentamicin resistance cassette. Successful conjugants were selected on LB + irgasan + carbenicillin plates, irgasan and carbenicillin resistant colonies were selected on salt-free LB + 10% sucrose plates, and gentamicin and carbenicillin sensitive clones were tested for removal of the gentamicin cassette by PCR followed by sequencing.

S. aureus – P. aeruginosa co-culture assays

Overnight grown S. aureus cells were washed once in sterile PBS, OD normalized and inoculated to an OD₆₀₀ of ~0.05 in 1 mL of fresh, pre-warmed M63 supplemented with 10 mg/mL tryptone, 4 mM MgSO₄, and 0.6% glycerol (M63T) in 14 mL polystyrene test tubes and grown for 4 hours at 37ºC with shaking at 300 rpm. P. aeruginosa cultures were back-diluted in fresh M63T and grown as above for approximately 6 hours before dilution to an OD₆₀₀ of 0.1 in sterile PBS. P. aeruginosa was inoculated into 1 mL of S. aureus cultures at a final OD₆₀₀ of ~0.002 and grown as above for 18 hours. Viable S. aureus cells were enumerated on LB agar supplemented with 65 g/L sodium chloride to provide selection against P. aeruginosa.

Pyocyanin growth curves

Overnight cells were washed once in sterile PBS, OD normalized and inoculated to an OD₆₀₀ of ~0.05 in fresh, pre-warmed M63. 100 μL of culture was aliquoted in duplicate to the wells of a 96-well plate. Plates were parafilmed to reduce evaporation and cells were grown for 20 hours with shaking (807 cpm) at 37°C in a BioTek Synergy H1 microplate reader (Agilent). Optical density values are adjusted by the background value at T0.

Quantifying DNA damage using a TUNEL assay

Overnight cultures were washed and normalized as described above for the PYO tolerance assays. Washed cells were inoculated to a calculated OD₆₀₀ of 0.1 into 125 mL Erlenmeyer flasks containing 12 mL of fresh, pre-warmed M63 and incubated at 37°C for 2 hours with shaking at 300 rpm. After incubation, PYO at a final concentration of 200 µM or an equal volume of DMSO as a control was added and flasks were returned to the incubator. At the indicated time points, cells were pelleted by centrifugation at 14,400 rcf for 5 minutes, washed once with ice-cold PBS, and fixed on ice for 30 minutes in 1.5 mL of a 2.66% solution of PBS-paraformaldehyde (PFA). After initial fixation, cells were pelleted by centrifugation at 20,000 rcf for 3 minutes and washed with PBS to remove residual PFA. Washed cells were then resuspended in 1.25 mL of ice-cold 56% ethanol and stored for at least 24 hours at -30°C prior to TUNEL staining. TUNEL staining was performed using the APO-DIRECT Kits (BD Biosciences [51-6536AK, 51-6536BK]) according to the manufacturer’s instructions. Stained cells were analyzed using an Apogee MicroPLUS flow cytometer (ApogeeFlow Systems Inc). S. aureus cells were gated using medium and large angle light scatter. Fluorescently labeled DNA was excited using a 488-nm laser and collected using 515 and 610 emission filters for FITC and propidium iodide, respectively, and analyzed using FlowJo (v10.1). Comparisons were made using the Overton method to identify the proportion of the population that exhibits fluorescence compared to the control condition.

Whole-genome sequencing

Genomic DNA from S. aureus was isolated using a DNeasy Blood & Tissue Kit (Qiagen) with the addition of 5 μg/mL lysostaphin (Sigma) to the pretreatment regimen described for gram-positive bacteria. Sequencing libraries were prepared using the Illumina Nextera XT DNA Library Preparation Kit according to the manufacturer’s instructions and sequenced by the CCR Genomics Core using a NextSeq 550 75 Cycle High Output kit for single-end sequencing, or the 150 Cycle Mid Output kit or High Output kit for paired-end sequencing. Genomes were assembled and mutations identified with breseq 0.33.1 [125] using the JE2 reference genome on NCBI (NZ_CP020619.1) as a reference.

RNA-sequencing and analysis

Cells were cultivated in 125 mL Erlenmeyer flasks containing 10 mL of fresh, pre-warmed M63 medium at 37°C with shaking at 300 rpm. After the indicated treatment time bacterial RNA was stabilized using RNAprotect Bacteria Reagent (Qiagen) according to the manufacturer’s instructions and stored at -80°C prior to RNA isolation. RNA was isolated using a Total RNA Plus Purification Kit (Norgen) with some modifications for S. aureus. Briefly, cryo-preserved bacterial pellets were resuspended in 100 μL of TE buffer containing 3 mg/mL lysozyme and 50 μg/mL lysostaphin and incubated for 30 minutes at 37°C. Volumes of Buffer RL and 95% ethanol used in the protocol were increased to 350 μL and 220 μL, respectively. Following elution of total RNA, any remaining genomic DNA was removed by TURBO DNase (Thermo Fisher) treatment using the two-step incubation method as detailed in the manufacturer’s instructions. Removal of genomic DNA was confirmed by PCR.

Ribosomal RNA was then removed using Ribo-Zero rRNA Removal Kit for gram-positive Bacteria (Illumina) according to the manufacturer’s instructions. Removal of rRNA was confirmed by electrophoresis using an Agilent TapeStation. RNA libraries were prepared using NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (New England BioLabs) and sequenced using a NextSeq 550 75 Cycle High Output kit for single-end sequencing.

Sequence files were pre-processed using fastp [126]. Alignment was performed using Kallisto [127] and analyzed using EdgeR [128] and RStudio. Two independent RNA-seq experiments were performed and the replicate alignments were combined for analysis. Differential gene expression between conditions was performed with the glmQLFit function in EdgeR using an FDR significance of < 0.1 and a log₂-fold cutoff of ≥ 1 or ≤ -1. Enriched pathways were identified using Gene Ontology Resource (www.geneontology.org) by the PANTHER Overrepresentation Test (released 20240226) (Annotation Version and Release Date: GO Ontology database https://doi.org/10.5281/zenodo.10536401Released2024-01-17) after conversion of differentially expressed JE2 locus tags to NCTC8325-4 using AureoWiki [129]. Pathway enrichment was tested using Fisher’s exact test corrected for the false discovery rate. All scripts used to generate results and run the above programs are provided in S5 Data. Processed data files for each replicate are available in S6 Data.

Statistics

Statistical analysis of data was performed using GraphPad Prism 9 (GraphPad Software, San Diego, CA, United States). Significance was determined by one-way or two-way analysis of variance (ANOVA), or a one-sample t-test as indicated in the figure legends. Log-scale values were log-transformed prior to statistical analysis.