PqsE Is Essential for RhlR-Dependent Quorum Sensing Regulation in Pseudomonas aeruginosa

Pseudomonas aeruginosa is a versatile bacterium found in various environments. It can cause severe infections in immunocompromised patients and naturally resists many antibiotics. The World Health Organization listed it among the top priority pathogens for research and development of new antimicrobial compounds. Quorum sensing (QS) is a cell-cell communication mechanism, which is important for P. aeruginosa adaptation and pathogenesis. Here, we validate the central role of the PqsE protein in QS particularly by its impact on the regulator RhlR. This study challenges the traditional dogmas of QS regulation in P. aeruginosa and ties loose ends in our understanding of the traditional QS circuit by confirming RhlR to be the main QS regulator in P. aeruginosa. PqsE could represent an ideal target for the development of new control methods against the virulence of P. aeruginosa. This is especially important when considering that LasR-defective mutants frequently arise, e.g., in chronic infections.

LasR-defective mutants frequently arise in various environments (18)(19)(20)(21)(22). It could be expected that these mutants would be unable to regulate QS-dependent genes; however, we have shown that RhlR is also able to activate the transcription of LasR target genes when the latter is nonfunctional (23). Indeed, LasR-defective strains expressing RhlR-regulated functions are found (22,24,25), implying that QS is not abolished in the absence of LasR. In recent work, a lasR mutant isolated from the lungs of an individual with cystic fibrosis expressed a rhl system that acted independently of the las system (26). It allowed this strain to produce factors essential for its growth under a specific condition that would normally require a functional LasR. When evolved under controlled conditions, this strain gained a mutation in MvfR (PqsR) making it unable to produce PQS and to activate the RhlR-dependent genes, highlighting the link between the pqs operon and RhlR.
Although a thioesterase activity of PqsE could participate in the biosynthesis of HAQs (27), the protein encoded by the last gene of the pqs operon is not required, since a pqsE mutant shows no defect in HAQ production (14). On the other hand, PqsE is implicated in the regulation of genes that include many of the RhlR-dependent targets, such as the phz and hcn operons and the lecA gene, through an unknown mechanism (28)(29)(30)(31)(32)(33). An impact of PqsE on the RhlR-dependent regulon was proposed; for instance, PqsE could enhance the affinity of RhlR for C 4 -HSL (28) or even synthesize an alternative ligand for RhlR (34). Importantly, such function is independent of its thioesterase function, as inhibitors of this activity had no impact on the regulatory functions of PqsE (27,28).
In this study, we validate that activation of RhlR-dependent QS strongly relies on the presence of a functional PqsE and reveal that this is especially important for activation of the rhl system in cases where LasR is not functional. This makes RhlR the key QS regulator and points to PqsE as an essential effector for full activation of this regulation. These findings thus strengthen the position of RhlR as the master regulator of QS and place PqsE at the center of QS regulatory circuitry in P. aeruginosa.

RESULTS AND DISCUSSION
RhlR is not the main activator of C 4 -HSL production. Quorum sensing regulation is typically described as a partnership between a LuxI-type AHL synthase and a LuxR-type transcriptional regulator. The LuxR-type regulator is activated by a cognate AHL and then regulates the transcription of target genes as well as the gene encoding the synthase, which upregulates AHL production, resulting in an autoinducing loop. In P. aeruginosa, the 3-oxo-C 12 -HSL synthase LasI is associated with the LasR regulator and the C 4 -HSL synthase RhlI with the RhlR regulator. Interestingly, LasR regulates the transcription of both rhlI and rhlR genes (2,5,7,35); actually, it has been argued that LasR, and not RhlR, is the primary regulator of rhlI (35). Accordingly, we previously reported that C 4 -HSL production is decreased in a lasR mutant (23,26). Indeed, a study in strain 148 showed that LasR binds the lux box found in the promoter region of rhlI but that RhlR does not (36), while other studies showing a direct regulation of rhlI by RhlR were actually performed in a heterologous host, in the absence of LasR (7,35). Together, these reports would suggest that RhlR mostly activates the transcription of rhlI when LasR is unable to.
To verify that RhlR is not the main regulator of C 4 -HSL production in a LasR-positive background, we measured concentrations of this AHL in cultures using liquid chromatography-tandem mass spectrometry (LC-MS/MS). The production of C 4 -HSL is only detectable at the stationary phase in a lasR mutant, while in a rhlR mutant, the production is only slightly delayed compared to that of wild-type (WT) P. aeruginosa strain PA14 (Fig. 2). This concurs with the often-overlooked idea (e.g. see reference 37) that it is LasR, rather than RhlR, that is primarily responsible for activating the transcription of rhlI and thus the production of C 4 -HSL, the ligand of RhlR. Interestingly, production is even more diminished in a double lasR pqsE mutant, while it is not affected at all in the ΔpqsE mutant, indicating PqsE has a role in LasR-independent activation of C 4 -HSL production (Fig. 2).
PqsE is important for LasR-independent quorum sensing. A plausible explanation for the results presented in Fig. 2 is that RhlR is a secondary regulator of rhlI, mostly important in the absence of LasR only, and that the absence of PqsE negatively affects the activity of RhlR only when LasR is not functional. To verify this hypothesis, we needed to investigate the activity of RhlR through one of its primary targets. Phenazines are redox-active metabolites produced by P. aeruginosa and are synthetized via two redundant operons: phzA1-G1 (phz1) and phzA2-G2 (phz2). These operons are almost identical and encode proteins that catalyze the synthesis of phenazine-1-carboxylic acid (PCA). PCA converts into derivatives such as pyocyanin, the blue pigment characteristic of P. aeruginosa cultures (38). The phz operons are differentially regulated depending on conditions, but the phz1 operon shows higher expression than phz2 in planktonic cultures of strain PA14 (39). The promoter of the phz1 operon contains a las box which can be recognized by both LasR and RhlR (40). We measured the activity of a chromosomal phzA1-lux reporter in both lasR and rhlR mutants to verify their involvement in the regulation of the transcription of the phz1 operon (Fig. 3). The transcription of phz1 is completely abolished in a rhlR mutant but it is still observed in a lasR mutant, although it starts much later than for the WT (after an optical density at 600 nm [OD 600 ] of 4.0). This is consistent with the delayed production of pyocyanin (23,41) and C 4 -HSL ( Fig. 2) observed in cultures of a lasR mutant. Since we know that transcription of phz1 and production of pyocyanin are abrogated in a double lasR rhlR mutant (23,41), these results indicate that RhlR, but not LasR, regulates the transcription of phzA1 and that RhlR is responsible for the late activation of phzA1 expression in a lasR-negative background. We used transcription of the phz1 operon to further study the influence of PqsE on RhlR-dependent regulation. Even if cultures of a pqsE mutant do not show any visible pyocyanin, we still observe clear expression of phz1 (Fig. 3). Since there is no pyocyanin produced in the WT until an OD 600 of around 2.5 even if there is expression from the phzA1 promoter, there seems to be a minimal level of expression of phz genes for detectable pyocyanin. Also, pyocyanin is not a direct product of the phz operons and it is possible that other enzymes (e.g., PhzM or PhzS) implicated in the conversion of PCA to pyocyanin do not follow the same pattern of expression in this background (29). The transcription of phzA1 is completely abolished in a double lasR pqsE mutant. Many studies report an impact of PQS-dependent QS on the regulation of the phz operons or pyocyanin production (28,31,39,41,42). More specifically, this effect necessitates a functional PqsE (28,42).
Because LasR regulates the expression of rhlI (5, 7, 23), we performed a ␤-galactosidase assay using a rhlI-lacZ reporter to verify the impact of PqsE on the transcription of rhlI. As expected, transcription of rhlI is much delayed in a lasR mutant (Fig. 4). This is compatible with the late activation of phz1 we observed (Fig. 3) and is apparently occurring because RhlR takes the relay in activating the transcription of rhlI following the initial activation by LasR. When the pqsE gene is inactivated in a lasR background, very low transcription of rhlI is observed (Fig. 4) which concurs with the production of C 4 -HSL in this background (Fig. 2) and which agrees with a PqsEdependent activity of RhlR. Again, since RhlR takes over regulating the production of C 4 -HSL following the initial activation by LasR, the transcription of rhlI slows down in rhlR and rhlI mutants after an OD 600 of 2.0, when LasR main activity is decreasing (the levels of 3-oxo-C 12 -HSL are rapidly declining) (23,31). Together, these data point to a role for PqsE in LasR-independent regulation of the rhl system. PqsE/RhlR/C 4 -HSL collude to activate LasR-independent quorum sensing. Since C 4 -HSL has an effect on RhlR activity (2,7,28), we needed to better understand the functional complementary of C 4 -HSL with PqsE in modulating the activity of RhlR. We measured the activity of the phzA1-lux reporter in a rhlI mutant as well as in a double rhlI pqsE mutant. Transcription of phzA1 in the rhlI mutant was delayed, but not abolished, suggesting that RhlR utilizes its AHL ligand to activate the phz1 operon but that its presence is not essential (Fig. 5A). However, when both C 4 -HSL and PqsE are absent (rhlI pqsE double-negative background), there is no residual transcription of phz1 (Fig. 5A), like in the rhlR-negative background (Fig. 3). The profile of expression of phz1 significantly differs between pqsE and rhlI mutants (P values of Ͻ0.05 from OD 600 s of 3.0 to 3.6). In the pqsE mutant, the expression starts at an OD 600 of around 2.0, while in the rhlI mutant, it starts later (OD 600 of around 3.5) and keeps augmenting through the rest of the growth curve. This suggests that both elements increase the activity of RhlR through different mechanisms.
Since the absence of LasR seems to impose the requirement for PqsE to achieve efficient RhlR activity, we overexpressed pqsE in a lasR-null background. As previously shown (43), the constitutive expression of PqsE augments and advances the transcription of phzA1 (Fig. 5B). When we added exogenous C 4 -HSL in the lasR mutant bearing a plasmid-borne pqsE, the transcription of phz1 started even earlier and reached higher Luminescence of the phzA1-lux chromosomal reporter was measured in a lasR::Gm background with either empty vector pUCP20 or pUCP20-pqsE with or without the addition of C 4 -HSL. The values are means Ϯ standard deviations (error bars) from three replicates. levels than with either one separately (P values of 0.046 and 0.002, respectively). Farrow et al. (28) proposed that PqsE acts by enhancing the affinity of RhlR for C 4 -HSL. However, we see that PqsE increases the activity of RhlR even in the absence of RhlI ( Fig. 4 and 5A), thus not supporting this hypothesis; our data suggest that RhlR full activity depends on both C 4 -HSL and PqsE and that their impact is cumulative.
The induction of RhlR activity by PqsE in the absence of rhlI could be explained by the proposed PqsE-dependent production of a putative alternative RhlR ligand. Indeed, Mujurkhee and colleagues (13) observed activation of rhlA transcription by adding culture-free fluids from a ΔrhlI mutant to a QS mutant expressing rhlR under the control of an arabinose-inducible promoter. They proposed in a subsequent study that this activity was PqsE dependent (34). We thus tested the effect of pqsE, rhlI, and rhlI pqsE mutants cell-free culture fluids on the activation of phzA1-lux in the rhlI pqsE doublenegative background. As expected, the activity of the reporter is strongly induced by culture supernatants from PA14 or a pqsE mutant (which both contain C 4 -HSL). On the other hand, there is no activation by supernatants from rhlI and rhlI pqsE mutants (see Fig. S1 in the supplemental material), even when combined with an overexpression of rhlR (data not shown). This argues against an unknown RhlR inducer whose production would require PqsE. The same results were obtained when using an hcnA-lacZ reporter (data not shown).
To validate our model, we looked at the regulation of the hcnABC operon, a dual target of both LasR and RhlR (12,41), and obtained results similar to what we observed for the phz1 operon and the rhlI gene (see Fig. S2). Taken altogether, our data highlight a possible homeostatic loop between RhlR-RhlI-PqsE and demonstrate that PqsE is essential for maintaining control of RhlR-dependent QS functions in a LasR-independent way.
Excess RhlR, but not C 4 -HSL, can overcome a PqsE deficiency. We then sought to better understand how C 4 -HSL and PqsE both contribute to RhlR activity. First, we verified if overproduction of C 4 -HSL could counterbalance a lack of PqsE. It was already shown that adding C 4 -HSL alone could not restore pyocyanin production in a triple ΔlasR ΔrhlI ΔpqsA mutant, but that adding PQS and C 4 -HSL together could (41). We thus used a plasmid-borne plac-rhlI for constitutive C 4 -HSL production and measured its effects on the transcription of phz1 and on pyocyanin production in various backgrounds. Overexpression of rhlI complements the transcription of phz1 in a lasR mutant enough to show pyocyanin production at the stationary phase ( Fig. 6A; see also Fig. S3). As expected, this complementation was not as efficient when a pqsE mutation was added to the lasR-negative background, as there was even less transcription of phz1 (P values of Ͻ0.05 at all growth phases) (Fig. 6A). Taken together, these results confirm that C 4 -HSL cannot counterbalance the absence of PqsE and highlight an important role for PqsE in regulating RhlR-dependent genes; this is especially striking in the absence of LasR.
We then looked at the overexpression of RhlR, since it partially restores pyocyanin production in a ΔpqsE background (30). We observed an augmentation in both the transcription of phzA1 and pyocyanin production ( Fig. 6B and S3). Figure S3 shows that when RhlR is overexpressed, both lasR and lasR pqsE mutants produce higher levels of pyocyanin, coupled with strong activation of phzA1-lux expression in both backgrounds. This is the first ever report of restoration of phz1 transcription and pyocyanin production in the absence of PqsE. Surprisingly, we observed a discrepancy between the transcription from the phzA1 promoter and pyocyanin production, which indicates that the transcription of the target genes shows a more realistic portrait of the activity of RhlR than only looking at pyocyanin production. Further supporting our model, the transcription of phzA1 and the production of pyocyanin when rhlR was overexpressed were higher in the lasR mutant than in the lasR pqsE mutant (P value of Ͻ0.05 at OD 600 s of 2.0 to 4.0), and these results again confirm an effect of PqsE on RhlR activity.
PqsE affects RhlR regulatory activity on its targets, including itself, in the absence of LasR. The very late activity of phz1 in lasR-negative backgrounds can be explained by low levels of RhlR, whose initial transcription also requires LasR (2,(5)(6)(7)35). When measuring the activity of an rhlR-lacZ reporter, there was indeed a lower transcription of rhlR in a lasR mutant (Fig. 7). Since overexpression of rhlI did not lead to full activation of the phz genes in a double lasR pqsE mutant background (Fig. 6A), we hypothesized that this was instead caused by low transcription of the rhlR gene. Interestingly, the level of rhlR transcription was even lower in the double lasR pqsE mutant background than in the single lasR mutant. This result is unexpected since the transcription of rhlR is weakly affected in a pqsE-null background (30). Because RhlR can activate the target genes of LasR when the latter is absent (23), we hypothesized that RhlR could therefore regulate itself, explaining the impact of PqsE only in the absence of LasR. Transcription of rhlR-lacZ was accordingly lower in a double lasR rhlR mutant, to levels similar to those in the lasR pqsE mutant (nonsignificant, P Ͼ 0.05 at all growth phases) (Fig. 7). This indicates that RhlR directs its own transcription only in the absence of LasR and that PqsE is important for this activity. These data confirm that PqsE is an essential element in RhlR activity when LasR is not functional.
Conclusion. The complex quorum sensing circuitry of P. aeruginosa has been extensively studied, and we know all three systems are intimately intertwined (44,45). Although RhlR is often believed to form a traditional autoinducing pair with rhlI, we confirm here that LasR really is the main activator of C 4 -HSL production and that RhlR activation of rhlI is mainly observed in the absence of a functional LasR. LasR is also an activator of the pqs operon and thus of PqsE. However, production of C 4 -HSL and PQS are not completely abolished in a lasR mutant, only delayed. In a lasR-null background, the importance of RhlR and PqsE on the activation of phzA1, rhlI, or hcnA is higher than in the WT, since LasR is at the top of the regulation cascade. This allowed us to observe that RhlR is able to fully activate target genes only if PqsE is present. The function of PqsE has been a subject of many studies but is still enigmatic (32). In this work, we show that PqsE most likely promotes the function of RhlR and that this effect seems independent of the presence of C 4 -HSL or another putative ligand, as previously proposed.
Under laboratory conditions, P. aeruginosa can afford a late activation of QS or even no activation of QS at all. In a more competitive environment, it is likely there is pressure to control these genes and to activate their transcription independently of LasR when necessary. PqsE could thus be important as a trigger for stronger and/or earlier RhlR activity. A growing number of studies report on the presence of LasRdeficient variants in chronic infections settings (18,19,22). With the absence of a functional LasR in these strains, the traditional QS hierarchy is altered and independent expression of RhlR becomes necessary for the bacteria to activate functions important for survival in hosts, such as virulence factors (like exoproteases and HCN) or biofilm formation (rhamnolipids and lectins).
Importantly, among LasR-deficient P. aeruginosa strains isolated from clinical settings, some still express a functional quorum sensing response through the activity of RhlR, independently of LasR (22,26). Since this study was limited to the prototypical strain PA14, it will be important to extend our findings and investigate the implication of PqsE in the activation of the RhlR regulon in diverse clinical and environmental isolates in order to better understand its role in QS gene regulation in P. aeruginosa.

MATERIALS AND METHODS
Strains, plasmids, and growth conditions. Bacterial strains are listed in Table 1. Plasmids used in this study are listed in Table 2. Unless otherwise stated, bacteria were routinely grown in tryptic soy broth (TSB; BD Difco, Canada) at 37°C in a TC-7 roller drum (NB, Canada) at 240 rpm or on lysogeny broth (LB) agar plates. When antibiotics were needed, the following concentrations were used: for Escherichia coli, 15 g/ml tetracycline and 100 g/ml carbenicillin, for P. aeruginosa, 100 g/ml gentamicin, tetracycline at 125 g/ml (solid) or 75 g/ml (liquid), and 250 g/ml carbenicillin. Diaminopimelic acid (DAP) was added to cultures of the auxotroph E. coli 7213 at 62.5 g/ml. All plasmids were transformed in bacteria by electroporation (46).
All experiments presented in this work were performed with three biological replicates and repeated at least twice.
Construction of the double ⌬pqsE mutants. A knockout in both rhlI and pqsE was constructed by transfer between chromosomes (46). The genomic DNA (gDNA) of strain ED297 rhlI::MrT7 was extracted using the EasyPure bacteria genomic kit (Trans Gen Biotech, China). Three milliliters of an overnight culture of ΔpqsE was centrifuged (16,000 ϫ g, 2 min) in separate microtubes. Pellets were washed twice with 300 mM sucrose. The pellets were combined in a final volume of 100 l 300 mM sucrose. Five hundred nanograms of gDNA was added to the bacterial suspension, and the mixture was transferred to a 0.2-mm electroporation cuvette. The cells were electroporated at 2,500 V, immediately transferred to 1 ml LB, and incubated at 37°C for 1 h. Selection was performed on LB agar containing gentamicin. Clones were selected and verified by PCR. The lasR::Gm mutation was introduced in the ΔpqsE background by allelic exchange using pSB219.9A as described (14,47).
Construction of phz1-lux chromosomal reporter strains. The mini-CTX-phz1-lux construct was integrated into the chromosomes of PA14 WT and mutants by conjugation on LB agar plates containing DAP with E. coli 7213 containing the pCDS101 plasmid. Selection was performed on LB agar plates containing tetracycline.
␤-Galactosidase activity assays and luminescence reporter measurements. Strains containing the reporter fusions were grown overnight in TSB with appropriate antibiotics and diluted at an OD 600 of 0.05 in TSB. For lacZ reporter assays, culture samples were regularly taken for determination of growth (OD 600 ) and ␤-galactosidase activity (48). For lux reporter assays, luminescence was measured using a Cytation 3 multimode microplate reader (BioTek Instruments, USA). When mentioned, C 4 -HSL was added at a final concentration of 20 M from a stock solution prepared in high-performance liquid chromatography (HPLC)-grade acetonitrile. Acetonitrile only was added in controls. All OD 600 measurements were performed with a NanoDrop ND100 spectrophotometer (Thermo Fisher Scientific, Canada).  rhlR gene under control of the lac promoter in pUCPSK John Mattick pUCPrhlI rhlI gene under control of the lac promoter in pUCPSK 47 pUCP20 Pseudomonas and Escherichia shuttle vector, Carb r 56 pUCP20-pqsE pqsE gene under control of the lac promoter in pUCP20, Carb r 57 Pyocyanin quantification. Overnight cultures of PA14 and mutants were diluted to an OD 600 of 0.05 in TSB and grown until an OD 600 of 4 to 5 was reached. Cells were removed by centrifugation at 13,000 ϫ g for 5 min, and the cleared supernatant was transferred to 96-well microplates. The absorbance at 695 nm was measured using a Cytation 3 multimode microplate reader. Pyocyanin production was determined by dividing the OD 695 by the OD 600 .
Data analysis. Statistical analyses were performed using R software version 3.6.3 (http://www.R -project.org) using one-way analysis of variance (ANOVA) with Tukey post hoc tests at different stages of growth. All conclusions discussed in this paper were based on significant differences. Probability (P) values of less than 0.05 were considered significant.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.