P. aeruginosa and S. aureus strains used in this study are shown in Table 1. CF isolates of P. aeruginosa were originally obtained from Dr. David Speert and previously described by our lab (Nguyen et al., 2014). Brain heart infusion (BHI) was used for routine culture of S. aureus and P. aeruginosa. Dialyzed trypticase soy broth (DTSB) for iron studies was prepared as previously described for iron-depleted medium (Oglesby-Sherrouse et al., 2014). Cultures were supplemented with 100 μM FeCl₃ as indicated.

To prepare extracts for antimicrobial assays, P. aeruginosa strains were grown in DTSB medium supplemented with or without 100 μM FeCl₃ for 18 h at 37°C. OD₆₀₀ of cultures was measured and cultures were centrifuged for 5 min at 14,000 RPM (~16,000 × g) in a tabletop centrifuge. Supernatant volume collected for extraction was normalized to the lowest culture OD in each biological replicate. Supernatants were extracted as previously described (Collier et al., 2002) and 1 ml was evaporated to dryness. Dried extracts were resuspended in 20 μL of 100% ethyl alcohol. For antimicrobial assays, the methicillin-resistant S. aureus strain M2 was grown for 18 h in BHI medium at 37°C. Growth was measured by OD₆₀₀, then diluted to an OD 0.05 in DTSB medium supplemented with or without 100 μM FeCl₃. 1 μL of P. aeruginosa extracts was added to 200 μL of diluted S. aureus culture. 1 μL 100% ethyl alcohol was used as a solvent control. Cultures were incubated for 18 h with shaking at 37°C in a 96-well plate, and S. aureus cell density (OD₆₃₀) was measured spectroscopically in a BioTek® Synergy™ HT plate reader.

Strains were grown for 18 h in DTSB supplemented with or without 100 μM FeCl₃ as indicated. Cells were spun down and supernatants harvested for AQ extraction as previously described by Collier et al. (2002). Nalidixic acid and deuterated N-dodecanoyl-L-homoserine lactone (C₁₂-HSL) were used as internal standards. LC-MS/MS analysis was performed as previously described (Ortori et al., 2011; Nguyen et al., 2015).

To quantify antimicrobial activity against S. aureus, a liquid co-culture system using transwell cell culture inserts (Corning Costar®, NY, USA) was performed as previously described (Nguyen et al., 2015). Briefly, P. aeruginosa and S. aureus strains were grown overnight in DTSB for 18 h at 37°C. S. aureus cultures were diluted to an OD₆₀₀ of 0.05 in DTSB supplemented with or without 100 μM FeCl₃, and 600 μL of the resulting cell suspension was inoculated into the bottom of the transwell plate. A transwell insert with a 0.4 μm membrane was then placed onto the plate, and 100 μL of P. aeruginosa cultures, diluted to an OD₆₀₀ of 0.05, were inoculated on top of the membrane. The transwell plates were incubated at 37°C for 18 h under static growth conditions, and S. aureus cell density (OD₆₃₀) was measured spectroscopically in a BioTek® Synergy™ HT plate reader.

Real time PCR (qPCR) analysis of P. aeruginosa and S. aureus gene expression in broth cultures was carried out as previously described (Oglesby et al., 2008; Nguyen et al., 2014; Reinhart et al., 2015), using the Applied Biosystems StepOne Plus Real Time PCR System (Life Technologies). Primers and probes used in this study are listed in Table S1. For S. aureus cell lysis: cell pellets were resuspended in 100 μL of 2.5 μg/μL lysozyme and 0.25 μg/μL lysostaphin in TE and incubated for 45 min at 37°C. Relative amounts of cDNA were determined by the ΔΔC_T method, and expression was normalized to oprF cDNA for P. aeruginosa or rpoB cDNA for S. aureus detected in each sample.

Bacteria were grown in DTSB for 18 h at 37°C, with or without 100 μM FeCl₃ supplementation as indicated. Each culture was harvested and extracted with acidified ethyl acetate as described by Collier et al. (2002). One half of the resulting organic extract was transferred to a clean tube and evaporated to dryness. Samples were resuspended in 1:1 acidified ethyl acetate:acetonitrile and analyzed by thin-layer chromatography (TLC) with a synthetic PQS standard (Pesci et al., 1999).

Previous analysis of a panel of pqs mutants demonstrated that antimicrobial activity of P. aeruginosa against S. aureus is dependent upon multiple AQs, including HQNO (Filkins et al., 2015; Nguyen et al., 2015). HQNO possesses innate antimicrobial activity against S. aureus due to its ability to bind to and inhibit the activity of cytochrome b (Lightbown and Jackson, 1956; Machan et al., 1992). Recent studies also demonstrate that both siderophores and AQNOs are required for antimicrobial activity against S. aureus during mixed biofilm growth (Filkins et al., 2015), indicating that iron may play a role in AQNO-mediated antimicrobial activity. We therefore sought to determine how iron supplementation affected the antimicrobial activity of AQNOs against S. aureus. Since P. aeruginosa produces several congeners of AQNOs, with varying alkyl chain lengths and levels of saturation, we chose to first analyze extracts of P. aeruginosa culture supernatants instead of assaying the effects of individual AQNO metabolites. AQs were extracted from wild type PAO1, the ΔpqsA mutant, which lacks production of all AQs, and the ΔpqsL mutant, which is specifically defective for AQNO production (Figure 1). S. aureus was then cultured in high or low iron media in the presence of each of these extracts. As expected, wild type PAO1 extracts substantially reduced S. aureus growth (by 50%) as compared to the solvent control (Figure 2A). This effect was eliminated when S. aureus cultures were subjected to either the ΔpqsA or ΔpqsL extracts (Figure 2A), demonstrating the specific role of AQNOs in inhibiting S. aureus growth in this assay. Notably, iron supplementation of P. aeruginosa cultures had no impact on the ability of the resulting extracts to inhibit S. aureus growth (Figure 2A). In contrast, iron depletion of S. aureus cultures significantly enhanced the antimicrobial activity of both the high and low iron PAO1 extracts (Figure 2A). This was not due to effects of the Pseudomonas growth medium, as extracted high or low iron DTSB media did not significantly affect S. aureus growth (Supplementary Figure S1). These data indicate that while iron does not affect on the production of AQNOs by P. aeruginosa, it does impact the susceptibility of S. aureus cultures to these metabolites.

To directly determine if AQNO production is regulated by iron, we quantified AQNO levels in supernatants of P. aeruginosa cultures grown in low or high iron media using liquid chromatography tandem mass spectrometry (LC-MS/MS). For this analysis we selected HQNO, an AQNO congener with a C7 (heptyl) alkyl chain, and NQNO, an AQNO congener with a C9 (nonyl) alkyl chain. Strikingly, our results show that the impact of iron on AQNO production is dependent upon alkyl chain length: production of HQNO is repressed by iron depletion, while that of NQNO is induced by iron depletion (Figures 2B,C). These data further support the idea that iron depletion does not induce overall AQNO production by P. aeruginosa, and that antimicrobial activity of AQNOs is instead dependent upon iron depletion of S. aureus cultures. To directly test this idea, we subjected S. aureus low and high iron cultures to HQNO and NQNO synthesized as described previously (Ortori et al., 2011). While HQNO and NQNO suppressed growth of S. aureus when cultured in either high iron medium, iron depletion significantly enhanced the antimicrobial activity of these metabolites (Figure 3A). As expected, decreases in cell density correlated closely with changes in cellular respiration as determined by staining cultures with triphenyltetrazolium chloride (TTC), indicating that the effects of AQNOs are due to inhibition of cytochrome activity vs. loss of cell viability (Table S2). Combined, these results show that iron starvation sensitizes S. aureus cultures to the antimicrobial activity of AQNOs.

We next sought to determine how PQS and HHQ contribute to iron-regulated antimicrobial activity against S. aureus. PQS and HHQ both function as quorum sensing molecules, activating the expression of genes that contribute to S. aureus growth suppression (Deziel et al., 2004; Diggle et al., 2007). PQS is additionally able to chelate iron (Bredenbruch et al., 2006; Diggle et al., 2007), which could also contribute to growth suppression of S. aureus. Our previous work showed that pqsH, which is responsible for the conversion of HHQ to PQS (Figure 1), is not required for S. aureus growth suppression during co-culture (Nguyen et al., 2015). Deletion of pqsH also had no effect on the antimicrobial activity of culture supernatant extracts (Figure 4A). Combined, these results indicate the iron-chelation activity of PQS is dispensable for S. aureus growth suppression. Alternatively, it is possible that deletion of pqsH alters the production of AQ metabolites, resulting in enhanced antimicrobial activity by other mechanisms. We therefore used LC-MS/MS to quantify production of HHQ and HQNO, and related congeners, in the ΔpqsH mutant. While no significant changes in either HQNO or NQNO were observed (Supplementary Figures S2C,D), these results showed more than a 10-fold increase in HHQ and NHQ production in the ΔpqsH mutant as compared to the wild type strain in both high and low iron conditions (Figures 4B,C). It is therefore possible that increased production of HHQ suppresses defects in antimicrobial activity that are specific to loss of PQS.

We next determined if PQS, due to its ability to chelate iron, exhibits innate antistaphylococcal activity by subjecting S. aureus low and high iron cultures to PQS and related congeners (C9-PQS and C1-PQS). Interestingly, only C9-PQS caused a small but significant decrease in S. aureus culture density as compared the ethanol control (p < 0.05), and this effect was only observed in low iron (Figure 3A). To determine if this activity was due to iron chelation, we also treated S. aureus cultures with NHQ and HHQ, as well as a PQS mimic, 3-NH₂-PQS. The replacement of an amino group for the hydroxyl group on the 3′ position of the quinolone allows for binding and activation of PqsR, but not iron chelation ability (Ilangovan et al., 2013). Surprisingly, 3-NH₂-PQS, HHQ, and NHQ were all able to suppress growth of S. aureus in low iron (Figure 3A), indicating that the 3′ hydroxyl group of PQS inhibits the growth suppressive activity of these AQs. As observed for HQNO, the reduction in S. aureus culture density upon HHQ supplementation correlated with decrease in TTC staining, demonstrating a similar mechanism of growth inhibition (Table S2). Our analysis further shows that addition of either PQS or HHQ has an additive effect on S. aureus growth when provided in combination with HQNO (Supplementary Figure S3). The combination of all three AQs did not significantly alter the growth suppression of S. aureus compared to HQNO in combination with either PQS or HHQ (Supplementary Figure S3). These data demonstrate a novel role for these signaling molecules in mediating antimicrobial activity against S. aureus.

A recent report by Filkins et al. demonstrated that co-culture with P. aeruginosa significantly altered S. aureus gene expression in an AQNO-dependent manner (Filkins et al., 2015). The most up-regulated genes were associated with fermentation pathways of S. aureus including formate acetyltransferase (pflB), L-lactate dehydrogenase (ldh), and alcohol dehydrogenase (adh). To determine whether synthesized AQNOs were able to induce these same genes, we performed real time PCR on S. aureus cultures grown with HQNO in DTSB supplemented with or without iron. HQNO induced pflB, ldh, and adh gene expression 100–1000 fold in low iron compared to the ethanol control, similar to that observed by Filkins et al. (Figure 3B). Strikingly, HQNO's ability to induce pflB, ldh, and adh expression in S. aureus was reduced in iron-replete conditions (Figure 3B). Thus, iron levels significantly impact on the ability of HQNO to induce fermentative metabolism genes in S. aureus.

The studies by Filkins et al. were conducted with co-cultures of S. aureus with P. aeruginosa, or in the presence of P. aeruginosa culture extracts (Filkins et al., 2015), leaving the possibility that other P. aeruginosa metabolites contribute to the AQNO-dependent shift in S. aureus metabolism. Since we showed that HHQ possesses innate antimicrobial activity against S. aureus, we assayed gene expression of S. aureus treated with HHQ or PQS to determine if either could drive S. aureus to fermentative metabolism. While the effects of HHQ were not as robust as that of HQNO, HHQ did cause a small but significant induction of pflB and ldh expression as compared to the ethanol control (Figure 3B). Moreover, induction of these genes by HHQ was only noted under iron-depleted conditions (Figure 3B—light gray bars). In contrast, PQS had no significant effect on pflB, ldh, or adh expression in either high or low iron condition compared to the ethanol control (Figure 3B). Notably, the ability of the individual AQ metabolites to induce fermentative gene expression correlated with their growth suppressive activity against S. aureus (Figure 3A). Thus, while HQNO appears to be the major metabolite responsible for increasing expression of S. aureus fermentative genes and growth suppressive activity, HHQ is also capable of altering S. aureus gene expression related to fermentative metabolism and inhibiting growth of S. aureus. Moreover, the sensitivity of S. aureus cultures to each of these metabolites is enhanced by iron depletion, presenting a novel means by which iron affects antimicrobial activity of P. aeruginosa against S. aureus.

Our previous studies showed that two clonal, longitudinally-isolated strains of P. aeruginosa from an individual CF patient exhibited iron-regulated antimicrobial activity against S. aureus (Nguyen et al., 2015). To determine if other CF isolates are capable of this activity, we examined clonal, longitudinally-isolated CF strains of P. aeruginosa from two additional patients: DSAM (DSAM-1, DSAM-2, and DSAM-3) and LNAP (LNAP-1, LNAP-2, and LNAP-3). These strains were isolated from CF patients ranging from age 2 to 22, providing a broad view of P. aeruginosa isolates from multiple stages of CF lung infection (Table 1). S. aureus was grown in co-culture with each CF isolate of P. aeruginosa, separated by a transwell membrane as previously described (Nguyen et al., 2015), in either high or low iron. With the exception of LNAP-3, each of these isolates exhibited iron-regulated antimicrobial activity against S. aureus (Figure 5A). As previously shown, the clonal JSRI-1 and JSRI-2 CF isolates also both exhibited iron-regulated antimicrobial activity against S. aureus (Figure 5A). These data indicate that iron-regulated antimicrobial activity against S. aureus is largely conserved by P. aeruginosa throughout CF lung infection, although variations in this activity between individual isolates does exist.

We next determined the potential role of individual AQs produced by CF isolates in iron-regulated antimicrobial activity against S. aureus. We previously found that PQS production of the JSRI-2 isolate is reduced as compared to the JSRI-1 isolate (Nguyen et al., 2014). However, co-culture with S. aureus restores PQS production to JSRI-2, potentially allowing for antimicrobial activity observed by this strain (Nguyen et al., 2015). To determine if PQS production similarly correlates with antimicrobial activity in the LNAP and DSAM CF isolates (Figure 5A), we performed thin layer chromatography (TLC) on culture supernatants from each isolate, grown either in mono-culture or co-culture with S. aureus. Similar to JSRI-2, DSAM-2 showed weak PQS production in monoculture (Figure 5B), yet was still able to exert antimicrobial activity against S. aureus (Figure 5A). Also similar to JSRI-2, we found that PQS production of DSAM-2 was restored to this isolate when co-cultured with S. aureus (Figure 5B). We also noted weak PQS production by LNAP-3 in mono-culture, which was not restored by co-culture with S. aureus (Figure 5B). Thus, co-culture with S. aureus restores PQS production to certain P. aeruginosa CF isolates, potentially allowing for iron-regulated antimicrobial activity against S. aureus.

Since the data above and our previous work shows that multiple AQs contribute to antimicrobial activity by P. aeruginosa laboratory strains (Nguyen et al., 2014), we next examined production of HQNO and HHQ by the JSRI, DSAM, and LNAP isolates. Similar to what was observed for PQS, production of both HQNO and NQNO was reduced in the JSRI-2, DSAM-2, and LNAP-3 isolates as compared to clonal isolates from the same patient (Table 2). HHQ and NHQ production were also reduced in JSRI-2 as compared to JSRI-1 (Table 2). In contrast, DSAM-2 and LNAP-3 showed increased production of HHQ and NHQ as compared to clonal isolates from the same patient (Table 2), similar to that observed in the PA14 ΔpqsH mutant (Figures 4B,C). Combined, these results suggest that JSRI-2 has an overall defect in AQ metabolism, while DSAM-2 and LNAP-3 are specifically defective for PQS and AQNO production.

To determine the molecular basis for altered AQ metabolite production by the JSRI-2, DSAM-2, and LNAP-3 isolates, we performed real time PCR of the pqsA and pqsH genes. These analyses show that pqsA gene expression is significantly reduced in the JSRI-2 isolate as compared to JSRI-1, while expression of pqsH is increased in the JSRI-2 isolate (Table 3), supporting the hypothesis that JSRI-2 exhibits an overall defect in AQ metabolism. In contrast, pqsA expression is maintained in DSAM-2 and increased in LNAP-3 as compared to their clonal isolates (Table 3), while pqsH expression is substantially reduced in these strains (Table 3). Thus, defects in the terminal step of PQS synthesis in these isolates correlates with an accumulation of HHQ intermediates. Surprisingly, DSAM-3, which showed no defects in PQS, HHQ, or HQNO production, exhibited significantly reduced expression of both pqsA and pqsH (Table 3). The rationale for this apparent paradox is not clear. Overall, these studies suggest that the multifactorial nature of AQ-dependent antimicrobial activity allows P. aeruginosa to maintain this activity amid changes in AQ metabolites during CF lung infection.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.