Intraspecies Signaling between Common Variants of Pseudomonas aeruginosa Increases Production of Quorum-Sensing-Controlled Virulence Factors

Coculture interactions between lasR loss-of-function and LasR+ Pseudomonas aeruginosa strains may explain the worse outcomes associated with the presence of LasR− strains. More broadly, this report illustrates how interactions within a genotypically diverse population, similar to those that frequently develop in natural settings, can promote unpredictably high virulence factor production.

I n chronic infections and healthy microbiomes, genetic diversity frequently arises and persists within clonally derived microbial populations, and recent data highlight that heterogeneity within a population can pose challenges to clearance and treatment (1)(2)(3)(4). Genotypic and phenotypic complexity is particularly well documented in the chronic lung infections associated with the genetic disease cystic fibrosis (CF), and studies have convincingly demonstrated that within a species, a common set of genes is under selection across strains and hosts (5)(6)(7)(8)(9)(10)(11).
That the levels of pyocyanin in WT/ΔlasR cocultures were higher than the level in each strain alone was not dependent on the initial ratios of WT to ΔlasR cells (Fig. 1D). We saw increased coculture colony pigmentation when the initial proportions of WT cells were at 0.2, 0.3, 0.5, 0.7, and 0.8 of the initial inoculums, with the balance comprised of ΔlasR cells (Fig. 1D).
No increase in pyocyanin was observed at any ratio when the WT was cocultured with the ΔlasR complemented derivative strain (ΔlasR ϩ lasR), indicating that the phenomenon was dependent on the lasR mutation (Fig. 1D). To assess the relative abundances of WT and ΔlasR cells in coculture, we competed each strain against a neutrally tagged WT strain (PA14 att::lacZ). We found that ΔlasR cells increased in proportion after 16 h of growth in colony biofilms regardless of the starting proportion whereas the proportions of untagged WT cells remained stable (Fig. 1E). We have previously shown that Anr activity is higher in ΔlasR strains and contributes to the competitive fitness of the ΔlasR strain against WT P. aeruginosa in colony biofilms (27,45), but Anr was not required for coculture pyocyanin production (Fig. S2).
Pyocyanin is a product regulated by quorum sensing (QS) through the transcription factors LasR, RhlR, and PqsR (46)(47)(48), and because QS regulation is cell density dependent, it was important to assess the coculture population size relative to that of the monoculture. Total CFU counts did not increase in WT/ΔlasR mixed cultures relative to the level for either strain alone (Fig. S1D). Instead, we found that WT/ΔlasR cocultures had fewer CFU than WT monocultures on lysogeny broth (LB) medium (Fig. S1D). Taken together, these data suggested that altered behavior, rather than cell number, contributed to the increased phenazine profile of LasRϪ strains.
Independent of its ability to produce autoinducers, the WT promotes RhlR/Idependent signaling in a ⌬lasR strain. In the canonical QS pathway, LasR regulates both PqsR and RhlR, and mutants lacking either regulator in a WT background have impaired pyocyanin production (49,50). Both pqsR and rhlR were required in ΔlasR cells for pyocyanin production in coculture with the WT (Fig. 2A). To determine if coculture increased RhlR-or PqsR-dependent signaling in ΔlasR strains, we fused lacZ to the promoters of rhlI and pqsA (PrhlI and PpqsA, respectively) which provide activity readouts of each respective regulator (17). We examined the interactions between WT and the ΔlasR strain in single-cell-derived colonies by spreading suspensions containing ϳ50 cells of WT with ϳ50 cells of either a ΔlasR PrhlI-lacZ or ΔlasR PpqsA-lacZ strain on LB agar containing the colorimetric ␤-galactosidase substrate 5-bromo-4-chloro-3indolyl-D-galactopyranoside (X-Gal). Intercolony distances and ␤-galactosidase activity in ΔlasR strains were measured. We found that the rise in PrhlI-lacZ activity was inversely correlated with the distance to a WT colony (Fig. 2B). Pearson correlation analyses showed that 54% of the variability in ΔlasR PrhlI-lacZ strain activity could be explained by changes in the distance to a WT colony (P value of Յ 0.0001). The increased PrhlI-lacZ activity in the ΔlasR strain was not observed in the ΔlasR ΔrhlR strain, and close proximity to another ΔlasR PrhlI-lacZ colony did not affect promoter activity (Fig. 2B, inset). Because C4HSL (which is synthesized by RhlI) activates RhlR and because proximity to the WT stimulated ΔlasR PrhlI-lacZ strain activity, we examined the role of RhlI in the ΔlasR strain response. We observed that a ΔlasR ΔrhlI strain was greatly impaired in the induction of pyocyanin upon coculture with the WT (Fig. 2A), which suggests that WT production of C4HSL was insufficient to complement the ΔlasR ΔrhlI strain and further posits activation of RhlR and C4HSL synthesis in ΔlasR strains. Although PqsR was required in ΔlasR cells for coculture pyocyanin production, there was no significant correlation with proximity to the WT for ΔlasR PpqsA-lacZ strain activity (Fig. 2B). Collectively, these data indicated that a diffusible factor produced by the WT stimulated RhlR-dependent signaling in the ΔlasR strain to induce downstream production of RhlR-and PqsR-dependent factors.
Given differences in colony pigmentation between WT/ΔlasR ΔrhlR and WT/ΔlasR ΔrhlI ( Fig. 2A) cocultures, C4HSL cross-feeding between the WT and ΔlasR strain likely occurred. Because C4HSL is diffusible and produced by WT cells, we tested the hypothesis that C4HSL or other acyl-homoserine lactones (AHLs) produced by the WT were necessary to induce RhlR-dependent activity in ΔlasR cells cocultured with the WT. To test this hypothesis, we cocultured the ΔlasR strain with ΔrhlI cells or ΔlasI ΔrhlI cells which lack both acyl-homoserine lactone synthases. Surprisingly, we found that like WT/ΔlasR cocultures, ΔrhlI/ΔlasR cocultures had higher levels of pyocyanin than monocultures (Fig. 2C). Similarly, ΔlasI ΔrhlI/ΔlasR cocultures had higher levels of pyocyanin production than monocultures though the interaction was delayed by ϳ24 h relative to the interaction of the WT/ΔlasR cocultures (Fig. 2C). Consistent with the activity of the ΔlasR PpqsA-lacZ strain, which was not induced in coculture with WT, the PQSdeficient ΔpqsA strain supported high pyocyanin colony pigmentation in coculture with ΔlasR cells after 24 h of extended incubation (Fig. 2C). The AHL-independent activation observed in ΔlasI ΔrhlI/ΔlasR cocultures and the striking differences in pyocyanin production observed between the strongly stimulating ΔrhlI/ΔlasR cocultures and the weakly stimulating WT/ΔlasR ΔrhlI cocultures suggested that the ΔlasR strain may rely more heavily on production of its own autoinducer for activation in coculture. Consistent with this model, we found that the ΔlasR strain produces RhlR/RhlI (RhlR/I)dependent AHLs in coculture with an AHL-sensing reporter strain (i.e., ΔlasI ΔrhlI strain with a lacZ promoter fusion to an AHL-responsive gene) ( Fig. S3A and B). The dispensable contribution of WT-produced autoinducers implicated a novel signaling interaction in coculture-dependent activation of RhlR/I activity in the ΔlasR strain (Fig. 2D).
To assess whether WT-induced RhlR activity in the ΔlasR strain was sufficient to elicit other RhlR/I-controlled phenotypes in addition to pyocyanin production, we tested whether coculture with LasRϩ strains enhanced swarming, a surface-associated motility which requires the production of RhlR-regulated rhamnolipid surfactants (51). While Mixed Genotype Coculture Promotes Virulence Factors ® the rhamnolipid-defective mutant ΔrhlA, ΔlasR, and ΔlasR ΔrhlR strains were unable to swarm, cocultures of the ΔlasR strain with the ΔrhlA strain swarmed considerably. The phenomenon was dependent on RhlR as the ΔrhlA/ΔlasR ΔrhlR cocultures did not swarm (Fig. S4). Altogether, these data implicated broad activation of RhlR-mediated QS in LasRϪ strains cocultured with LasRϩ P. aeruginosa.
Pyochelin production by ⌬lasR cells is required for coculture interactions. With evidence indicating that induction of RhlR activity in ΔlasR cells can occur in both mixed-strain spot colonies and adjacent colonies independent of autoinducer crossfeeding, we sought to gain further insight into the mechanisms that underlie the WT-ΔlasR cell interactions. We investigated the transcriptomes of the lasR mutant in coculture with either the WT or itself via RNA sequencing (RNA-seq). We grew ΔlasR colony biofilms on LB medium physically separated from a lawn of either the ΔlasR or WT strain by two filters with 0.22-m pores to prevent mixing of genotypes while allowing for the passage of small molecules. In order to examine ΔlasR strain transcriptional profiles, RNA was extracted from cells within the ΔlasR colony biofilms grown on the topmost filter for 16 h (Fig. 3A). As expected, no lasR reads were detected in our sequencing data to suggest that the wild type was sufficiently excluded by filter separation. Expression levels of a total of 199 genes in the ΔlasR strain were higher, and those of 198 genes were lower by a |log 2 (fold change)| of Ն1 with a P value of Ͻ0.05, in coculture with the WT than levels in the ΔlasR strain alone (Table S1). Gene Ontology (GO) term analyses through PantherDB (52) indicated that the upregulated gene set was significantly overrepresented in two pathways related to siderophore biosynthesis: the pyoverdine biosynthetic process and salicylic acid biosynthetic process (an upstream precursor of pyochelin) with ϳ44and ϳ77-fold enrichment, respectively (P values of Ͻ0.005). Twenty-eight out of the 33 genes in the pyochelin and pyoverdine siderophore biosynthesis-and acquisition-related GO families were significantly upregulated in ΔlasR cells upon coculture with the WT (Fig. 3B) (i.e.,|log 2 (fold change)| of Ն0 with a P value of Ͻ0.05). Other low-iron-responsive genes were differentially expressed, including the has genes involved in heme uptake and antABC genes (Table S1). While we observed stimulation of rhlI promoter activity and increased production of RhlR-regulated products, we did not see a broad transcriptional pattern indicative of RhlR activation at this early time point (Table S1), and this point is discussed below.
Given that siderophore biosynthesis genes were upregulated in ΔlasR cells cocultured with WT cells, we qualitatively examined production of fluorescent pyoverdine and pyochelin siderophores in monocultures and cocultures. To determine the contribution of both pyoverdine and pyochelin by the ΔlasR strain to fluorescence, genes required for pyoverdine biosynthesis (ΔlasR ΔpvdA strain), pyochelin biosynthesis (ΔlasR ΔpchE strain), or both pathways (ΔlasR ΔpvdA ΔpchE strain) were disrupted (Fig. 3C). Increased fluorescence attributable to both pyoverdine and pyochelin in coculture was due to siderophore production by ΔlasR strains, consistent with the RNA-seq data, as the increased fluorescence in WT/ΔlasR cocultures was lost in coculture when the ΔlasR strain was replaced with a ΔlasR ΔpvdA, ΔlasR ΔpchE, or ΔlasR ΔpvdA ΔpchE mutant. While cocultures of the WT and the pyoverdine-deficient derivative ΔlasR ΔpvdA strain (i.e., WT/ΔlasR ΔpvdA coculture) showed increased pyocyanin production relative to that of either monoculture, the ΔlasR ΔpchE and ΔlasR ΔpvdA ΔpchE strains did not promote pyocyanin production in coculture with the WT, as observed by colony pigmentation (Fig. 3D). The decrease in ΔlasR strain-derived pyocyanin was not due to decreased fitness as disruption of pvdA and pchE individually in ΔlasR cells had no effect on the final proportions; in contrast, the ΔlasR ΔpvdA ΔpchE strain had a significant defect in competitive fitness compared to the fitness of the ΔlasR parental strain (Fig. S5). These data suggested that pyochelin played a role in the coculture interaction. To test this model, we complemented the pyocyanin defect in the siderophoredeficient ΔpvdA ⌬pchE/ΔlasR ⌬pchE coculture with pyochelin-containing extracts from cultures of ΔpvdA cells which cannot produce pyoverdine or control extracts from siderophore-deficient ΔpvdA ΔpchE cultures (Fig. S6A gives supernatant absorption spectra). The two extracts were analyzed using a chrome azurol S (CAS) assay (53) to confirm that chelator activity was present in the ΔpvdA cell supernatant extracts but not in extracts from ΔpvdA ΔpchE cultures (Fig. S6B). Medium supplemented with pyochelin-containing extracts, but not siderophore-free extracts, restored pyocyanin production in ΔpvdA ΔpchE/ΔlasR ΔpchE cocultures (Fig. 3E), lending further support to the idea that pyochelin was required for coculture interactions. Consistent with this requirement, iron supplementation suppressed siderophore production, as expected, and diminished coculture pyocyanin in WT/ΔlasR cocultures (Fig. 3F) alongside a FIG 3 Biosynthesis of the coculture-induced iron scavenging siderophore pyochelin is required in the ΔlasR strain for pyocyanin production when it was cultured with the wild type (WT). (A) Scheme for the collection of RNA from ΔlasR colony biofilms grown above a lawn of ΔlasR or WT cells. DE, differential expression. (B) Volcano plot showing differential expression (log 2 ) for ΔlasR cells grown over WT relative to ΔlasR cells grown over ΔlasR on the x axis; the y axis shows the Ϫlog 10 P value for the difference between sample types. Genes involved in pyoverdine (blue) and pyochelin (green) iron acquisition systems are indicated. ccmC and ccmF (indicated with arrows) of the pyoverdine GO term are involved in c-type cytochrome biosynthesis, and strains with knockouts of these genes are reported to produce more pyochelin. Mixed Genotype Coculture Promotes Virulence Factors ® decrease in ΔlasR strain RhlR/I-dependent AHL activity in coculture with the AHLsensing reporter strain ( Fig. S3C and D). Collectively, these data support a model in which pyochelin production by the ΔlasR strain is induced and required for pyocyaninpromoting interactions with the WT through initiation of a low-iron response (Fig. 3G).
In coculture with the WT, the ⌬lasR strain responds to citrate, a pyochelininducible metabolite. Many of the upregulated genes in the ΔlasR strain upon coculture with the WT have annotations related to organic acids, such as anthranilate and citrate (Table S1 and Fig. S7A). Several lines of evidence suggest that anthranilate was not the factor that stimulated RhlR activity and pyocyanin production in coculture. First, anthranilate supplementation (up to ϳ15 mM) did not alter ΔlasR strain phenazine production (Fig. S7B). Further, cocultures of the ΔlasR mutant with the anthranilate synthase mutant ΔphnAB strain, with reduced extracellular anthranilate (Fig. S7C), or with the ΔpqsA mutant (Fig. 2C), which accumulates it (54), did not alter coculture phenazine production. Anthranilate is also generated from tryptophan catabolism through the kynurenine pathway (55). Given that coculture pyocyanin production could occur in the absence of exogenous amino acids (Fig. S1A), we concluded that the kynurenine pathway was likely not involved. Together, these data suggested that anthranilate was not a stimulating metabolite.
In light of the observation that 20% of the most strongly differentially expressed genes [|log 2 (fold change)| of Ն2 with a P value of Ͻ0.05] were implicated in citrate sensing, transport, catabolism, and anabolism, as annotated by UniProt (56) and www .pseudomonas.com (57), we looked at all genes with annotations related to citrate to identify broad expression patterns (Fig. 4A). Among the genes induced in ΔlasR/WT cocultures [|log 2 (fold change)| of Ն0 with a P value of Ͻ0.05] were genes annotated as citrate responsive or playing roles in citrate sensing and transport or metabolism, with the most strongly upregulated citrate genes involved in sensing, transport, and catabolism specifically (Fig. 4B).
We measured citrate in the supernatants of WT and ΔlasR LB cultures based on the following observations: (i) ΔlasR strains induced low-iron-responsive genes when grown near the WT but not itself; (ii) ΔlasR strain pyochelin production was necessary for coculture interactions that led to increased pyocyanin; (iii) citrate sensing and catabolism genes were induced in ΔlasR cells by the presence of WT cells; and (iv) numerous microbes, including Pseudomonas putida, were shown to secrete citrate and other organic acids when iron limited (44,(58)(59)(60). Citrate was detected in both WT and ΔlasR strain LB culture supernatants (Fig. 4C), and amendment with extracts containing 50 M pyochelin increased extracellular citrate concentrations by ϳ2-fold in WT cultures compared to levels in cultures supplemented with extracts lacking pyoverdine and pyochelin, with a much smaller stimulation in ΔlasR cultures under the same conditions (Fig. 4C). This suggested that WT-produced citrate may be involved in WT/ΔlasR coculture interactions and that citrate release was enhanced by pyochelin produced by the ΔlasR strain.
Citrate induces RhlR-dependent activity and RhlI levels in a ClpX proteasedependent manner in ⌬lasR cells. To determine if citrate was sufficient to stimulate RhlR activity in ΔlasR cells, we analyzed its effects on rhlI promoter fusion activity, colony morphology, and RhlI protein levels. We found that citrate increased rhlI promoter activity (PrhlI) in ΔlasR cells and that its effects were dependent on the presence of RhlR (Fig. 5A). Citrate was sufficient to promote increases in colony pigmentation and colony smoothness, previously characterized to be RhlR-mediated in ΔlasR cells (29) (Fig. 5A, inset). In contrast, citrate caused a small but significant reduction in WT PrhlI activity compared to that the LB control (Fig. 5A).
To determine if RhlI protein levels were influenced by citrate, we utilized an arabinose-inducible rhlI-hemagglutinin (HA) construct to assess RhlI protein levels and stability in the absence and presence of citrate independent of RhlR transcriptional control. RhlI-HA was functional as swarming defects of the ΔrhlI mutant were complemented upon expression of RhlI-HA but not by the empty vector (Fig. 5B, inset). RhlI-HA protein levels were 3-fold higher in the ΔlasR strain upon citrate supplementation than in the controls (Fig. 5B). Consistent with the absence of an increase in rhlI promoter activity in WT strains (Fig. 5A), RhlI-HA protein levels were not higher with citrate in the ΔlasR complemented strain (ΔlasR ϩ lasR strain) (Fig. 5B). The differential responses to citrate were also observed in LasRϪ and LasRϩ pairs of clinical isolates (CIs). LasRϪ CIs from acute (strain 388D) and chronic (strain DH2415) infections had RhlI-HA levels 1.4and 1.7-fold higher, respectively, in the presence of citrate (Fig. 5C), whereas alterations in RhlI-HA protein levels in LasRϩ CIs from acute (strain 550A) or chronic (strain DH2417) infections were not observed (Fig. 5D). Through this work, we successfully identified citrate as a molecule in coculture that specifically promoted RhlI protein levels in LasRϪ strains, but not in LasRϩ strains, by posttranscriptional control. In an attempt to identify transporters involved in the ΔlasR strain response to citrate and/or other coculture metabolites, we deleted two organic acid transporters: dctA (61) and PA14_51300 (62) in the ΔlasR strain background. We found that the ΔlasR ΔdctA strain showed induction of pyocyanin when it was cocultured with the WT and that induction was dependent on RhlR (Fig. S7D). Similar results were obtained with the ΔlasR Mixed Genotype Coculture Promotes Virulence Factors ® ΔPA14_51300 strain (Fig. S7E), suggesting that these transporters were not required for the interaction, perhaps due to redundant functions of other proteins or the involvement of other import mechanisms.
The temporal pattern of activation and the stimulation of RhlI-HA in the absence of RhlR control suggested that the RhlI protein may precede signal amplification via the  (63); however, we found no apparent difference in RhlS expression levels in our RNA-seq reads in coculture. To begin to unravel the mechanisms by which citrate promoted RhlR/Idependent signaling and RhlI stability in ΔlasR cells, we analyzed the role of two proteases previously found to target and degrade RhlI (i.e., Lon and ClpXP) (64). Given that knockouts of Lon protease have a less substantial rise in RhlR/I expression in ΔlasR strains than in the WT (65), we focused on the role of ClpXP in ΔlasR cells. We found that citrate induction of RhlI-HA protein levels in the ΔlasR strain relative to that in the LB control was dependent on functional ClpX protease (Fig. 5E). More specifically, when ClpX, a protease shown to degrade RhlI, is nonfunctional (i.e., ΔlasR clpX::TnM strain), RhlI-HA levels did not increase on LB plus citrate relative to that of the LB control, unlike the level in the ΔlasR strain comparator (Fig. 5E). Under LB culture conditions, RhlI-HA levels were 3.20-Ϯ 2.1-fold higher in the ΔlasR clpX::TnM strain than in the ΔlasR strain, which mirrors the 3-fold induction observed for the ΔlasR strain on LB medium with or without citrate. Under citrate-supplemented conditions, no significant difference in RhlI-HA levels was observed for the ΔlasR clpX::TnM relative to that in the ΔlasR strain (fold change of 1.01 Ϯ 0.53). In other words, as previously noted for WT strains, ClpX may degrade RhlI in ΔlasR cells and play a role in the ΔlasR cell response to citrate. The distinct responses and mechanisms identified between LasRϩ and LasRϪ strains under iron limitation and exposure to the low-iron-associated molecules, citrate and pyochelin, enabled increases in antagonistic factors beyond monoculture levels as an emergent property of P. aeruginosa intraspecies interactions.

DISCUSSION
In this study, we described an emergent outcome of coculturing LasRϪ and LasRϩ strains of P. aeruginosa in which their interactions promoted the toxic exoproducts pyocyanin and rhamnolipids ( Fig. 6 provides a model). We determined that, in coculture, the ΔlasR strain produces the siderophore pyochelin and that exogenous pyochelin induced citrate release more strongly in the WT than in ΔlasR strain. Citrate increased RhlI protein levels and induced RhlR-dependent activity only in ΔlasR cells and not WT cells (Fig. 6). Western blot analyses of RhlI-HA expressed from a regulated promoter led us to propose that the increase in RhlR-dependent signaling is due to decreased degradation of RhlI by ClpXP, a known negative regulator of RhlI (65,66), or through other mechanisms of posttranscriptional regulation. The differences in siderophore production, citrate release, and RhlR/I-dependent activation between P. aeruginosa LasRϩ and LasRϪ strains in coculture reflect the pronounced differences between  (1). Citrate (and diffusible autoinducer) released by the wild type in coculture stimulates RhlR/I-dependent activity in a ΔlasR strain (2). Citrate stabilizes RhlI protein in ΔlasR cells potentially through a ClpXP protease-dependent mechanism (3), which ultimately promotes the production of antagonistic factors like pyocyanin toxin and rhamnolipid surfactant above monoculture levels (4).
Mixed Genotype Coculture Promotes Virulence Factors ® strains that drive QS reactivation. Previous studies have shown that LasRϪ strain colony morphology and phenazine production change in the presence of other species such as Candida albicans (29) and Staphylococcus aureus (see Fig. 3B in reference 67), and future work will determine if pyochelin and citrate also participate in these interspecies interactions as many microbial interactions have been shown to be influenced by iron availability (68)(69)(70)(71). Furthermore, the induction of RhlR activity that can occur in late-stationary-phase ΔlasR cultures (30,72) may relate to changes in iron or TCA cycle intermediates. While we found that WT-produced autoinducers, including 3OC12HSL, C4HSL, and PQS, were not required for coculture stimulation, they clearly contributed to the enhanced RhlR-dependent activity, which is consistent with intercolony QS interactions demonstrated previously (73).
The stimulatory relationship between LasRϩ and LasRϪ strains was remarkably stable as it was observed when strains were mixed within single spot colonies (Fig. 1A) and when strains were separated by either filters (Fig. 3) or millimeter distances on an agar plate (Fig. 2B). The LasRϪ/LasRϩ interactions occurred across distinct media (see Fig. S1A in the supplemental material), among genetically diverse LasRϩ and LasRϪ clinical isolates (Fig. S1B), and over a wide range of relative proportions of each type (Fig. 1C). Of note, colonies inoculated at a 80:20 WT-to-ΔlasR cell ratio had more zones with the lasR mutant-associated phenotypes described as sheen and lysis than colonies with a 20:80 WT-to-ΔlasR cell ratio (Fig. 1C). At both ratios, ΔlasR cell numbers increased slightly relative to level of the wild type (Fig. 1D). We propose that the reduced appearance of sheen and lysis in mixed colonies inoculated with more ΔlasR cells reflects a requirement for a sufficient proportion of ΔlasR cells to initiate the WT-ΔlasR cell interactions that activate RhlR and restore a more WT-like phenotype to LasRϪ cells. Furthermore, if RhlR signaling is not fully activated in ΔlasR cells, there may be regions of increased ΔlasR cell killing via WT-produced factors such as cyanide (74).
The consequences of this intraspecies interaction may explain the worse outcomes exhibited by patients in which LasRϪ strains are detected (13), but future studies that include genotypes, monoculture and coculture phenotypes, and longitudinal outcome data will be required. RhlR plays other important roles in host interactions (75) which may benefit P. aeruginosa LasRϪ strains. The observation that rhlR mutants are rare in natural isolates and that LasRϪ strains with active RhlR are virulent (25,76) underscores the relevance of this mechanism and highlights the importance of understanding how microbial interactions influence RhlR activity.
As studies of inter-and intraspecies interactions progress, it is becoming increasingly clear that the environment can dictate the outcome of microbial interactions (77). In fact, even the importance of QS regulation for fitness depends on nutrient sources and conditions (78,79). As ΔlasR cell-produced pyochelin was a key component of the interaction and as pyochelin production is repressed under conditions of excess iron, it was not surprising that iron supplementation suppressed the interaction (Fig. 3F and Fig. S3C and D) without significantly altering the final colony CFU count or strain ratios relative to those of the LB control (Fig. S4). Siderophore-mediated iron uptake is often required in vivo (34,80,81) due to iron sequestration by host proteins (82)(83)(84)(85); thus, in vivo settings may support these interactions.
Interestingly, pyoverdine, the higher-affinity siderophore, was not required for the coculture response, mirroring findings that genes for biosynthesis of pyoverdine, but not pyochelin, are commonly disrupted in isolates from chronic CF patients (39)(40)(41). In the absence of pyoverdine (i.e., the ΔlasR ⌬pvdA strain), we observed more pyocyanin in coculture with the WT than with the ΔlasR strain (Fig. 3D), and we speculate that this is due to increased pyochelin production by ΔpvdA cells, but future studies will be required to test this model. It was interesting to find that in WT/ΔlasR coculture, heme-related proteins, hasAP, hasS, and hasD, were among the top eight mostupregulated genes by the ΔlasR strain because the presence of lasR mutants and heme content are both reported biomarkers of disease progression in CF patients (13,86). Coculture-induced lasR mutant phenotypes may link these two correlative observations.
Citrate, a TCA intermediate released under iron limitation as a result of overflow metabolism (43,44,60,87), can be used by P. aeruginosa and other microbes for iron acquisition due to its iron chelating properties (88). The increased siderophore production by ΔlasR cells in coculture likely reflects different metabolic strategies between genotypes. Ongoing work will investigate the mechanisms that drive differences in metabolism and iron requirements in order to determine how these differences shape microbial and host interactions. It is likely that Crc-mediated catabolite repression is involved in the response to citrate and the control of RhlI levels (64,66). That a mechanism exists for the induction of RhlR-mediated QS in response to citrate secreted when iron is limiting dovetails with reports of increased expression of the P. aeruginosa QS regulon in low iron in LasRϩ cells (89)(90)(91). This coordinate regulation may aid in iron acquisition as QS-controlled phenazines, such as pyocyanin, reduce poorly soluble Fe 3ϩ to Fe 2ϩ and facilitate its uptake via the Feo system (92). Additionally, rhamnolipids have been employed for iron remediation (93,94), which suggests that their surfactant activity may increase P. aeruginosa substrate iron uptake in part through hydroxy-alkylquinolone-dependent mechanisms (95).
Given that anthranilate did not alter ΔlasR colony morphology or phenazine production, we did not further investigate anthranilate as a cross-fed metabolite involved in WT-ΔlasR cell interactions. We speculate that the increased expression of anthranilate catabolism genes in coculture may be more reflective of increases in RhlR activity than increased exposure to anthranilate given reports highlighting RhlR activation of antABC and catABC anthranilate catabolism genes (59).
As the presence of heterogeneous genotypes within single-species populations becomes increasingly appreciated, it is important to understand how commonly encountered genotypes interact to influence population-level behavior. Other work shows that cocultures can influence the survival of other genotypes (96,97). Here, we show that intergenotype interactions lead to increased RhlR-dependent signaling in LasRϪ strains. It is likely that a wide array of such interactions has yet to be uncovered.

MATERIALS AND METHODS
Strains and growth conditions. Strains used in this study are listed in Table S2 in the supplemental material. Bacteria were maintained on LB (lysogeny broth) medium with 1.5% agar. Saccharomyces cerevisiae strains for cloning were maintained on yeast-peptone-dextrose (YPD) medium with 2% agar. With the exception of pyochelin complementation experiments, which were performed in 12-well dishes with a 2-ml total volume containing 50 M pyochelin (or an equal-volume negative-control extract), colony biofilm assays were performed in 100-mm petri dishes with a 25-ml total volume. Where stated, a 20 mM concentration of the indicated metabolite was added to the medium (liquid or molten agar). Planktonic cultures were grown on roller drums at 37°C. Artificial sputum medium (ASM) was made as described previously (27).
Competition assays. Competition assays were performed to determine the relative fitness of P. aeruginosa mutants. Strains to be competed were grown overnight and adjusted to an optical density at 600 nm (OD 600 ) of 1. Competing strains were combined with the PA14 att::lacZ strain in a 1:1 ratio unless otherwise stated. Following a 15-s vortex, 5 l of the combined suspension was spotted on LB agar. After 16 h, colony biofilms (and agar) were cored, placed in 1.5-ml tubes with 500 l of LB, and agitated vigorously for 5 min using a Genie Disruptor (Zymo). This suspension was diluted, spread on LB plates supplemented with 150 g/ml 5-bromo-4-chloro-3-indolyl-D-galactopyranoside (X-Gal) using glass beads, and incubated at 37°C until blue colonies were visible (ϳ24 h). The numbers of blue and white colonies per plate were counted, and the final proportions were recorded. Each competition was run in triplicate on 3 separate days.
Additional methods. See Text S1 in the supplemental material for methods describing plasmid construction, pyocyanin quantification, colony proximity image analysis, swarming, RNA collection and processing, pyochelin extraction and quantification, citrate quantification, ␤-galactosidase quantification, Western blotting, and acyl-homoserine lactone activity assays.
Data availability. Data for RNA-seq analysis of P. aeruginosa ΔlasR grown on the ΔlasR or WT strain in coculture has been uploaded to the Gene Expression Omnibus (GEO) repository (https://www.ncbi .nlm.nih.gov/geo/) under accession number GSE149385.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TEXT S1, DOCX file, 0.03 MB.

ACKNOWLEDGMENTS
Research reported in this publication was supported by grants from the Cystic Fibrosis Foundation HOGAN19G0 and NIH/NIAID T32AI007519 (D.L.M.). Additional support came from NIGMS P20GM113132 through the Molecular Interactions and Imaging Core (MIIC), STANTO19R0 from the Cystic Fibrosis Foundation, and NIDDK P30-DK117469 (Dartmouth Cystic Fibrosis Research Center). RNA-seq was carried out at Dartmouth Medical School in the Genomics Shared Resource, which was established by equipment grants from the NIH and NSF and is supported in part by a Cancer Center Core Grant (P30CA023108) from the National Cancer Institute.
We also thank Georgia Doing for preliminary RNA-seq analysis, Pat Occhipinti for swapping the antibiotic marker on the rhlI-HA expression vector, and Carla Cugini for the ΔlasRclpX::TnM mutant.