Mechanism of Pseudomonas aeruginosa RhlR Transcriptional Regulation of the rhlAB Promoter

Abstract

Pseudomonas aeruginosa contains two transcription regulators (LasR and RhlR) that, when complexed with their specific autoinducers (3-oxo-dodecanoyl-homoserine lactone and butanoyl-homoserine lactone, respectively) activate transcription of different virulence-associated traits. We studied the RhlR-dependent transcriptional regulation of the rhlAB operon encoding rhamnosyltransferase 1, an enzyme involved in the synthesis of the surfactant monorhamnolipid, and showed that RhlR binds to a specific sequence in the rhlAB regulatory region, both in the presence and in the absence of its autoinducer. Our data suggest that in the former case it activates transcription, whereas in the latter it acts as a transcriptional repressor of this promoter. RhlR seems to repress the transcription of other quorum-sensing-regulated genes; thus, RhlR repressor activity might be of importance in the finely regulated expression of P. aeruginosa virulence-associated traits.

Pseudomonas aeruginosa is an opportunistic pathogen that causes serious nosocomial infections. The secretion of numerous toxic compounds and hydrolytic enzymes involved in bacterial pathogenesis is coordinately expressed at high bacterial densities by a mechanism called quorum sensing (4, 38). The quorum-sensing response is triggered by the accumulation in the extracellular medium of certain compounds called autoinducers (AIs). In several proteobacteria the AIs are N-acyl-homoserine lactones (4). At high bacterial densities AIs reach a threshold concentration and interact with specific transcriptional regulators, usually belonging to the LuxR family. These protein-AI complexes are responsible for the regulation of transcription (4, 38).

P. aeruginosa quorum-sensing regulation of gene expression depends on the production mainly of two AIs, butanoyl-homoserine lactone (C₄-HSL) and 3-oxo-dodecanoyl-homoserine lactone (3-O-C₁₂-HSL), that, coupled with RhlR and LasR proteins, respectively, activate gene expression (4, 38). The LasR(3-O-C₁₂-HSL) complex promotes transcription of several virulence-associated traits including that of the gene coding for the transcriptional regulator RhlR (10, 22).

The RhlR(C₄-HSL) complex promotes the expression, among others, of the genes coding for the enzymes involved in rhamnolipid synthesis: the rhlAB operon, encoding rhamnosyltransferase 1 (19), and rhlC, encoding rhamnosyltransferase 2 (23). Besides the well-documented ability of RhlR to activate transcription, it has been suggested that this protein acts as a repressor in at least two different promoters (1, 14). It has been reported recently that RhlR can act as a repressor of its own gene, rhlR (14), and it has been suggested that this protein, in the absence of C₄-HSL, has a negative effect on the LasR(3-O-C₁₂-HSL)-dependent expression of lasB, encoding elastase B (1). It has recently been reported that RhlR dimerizes both in the presence and in the absence of C₄-HSL and that 3-O-C₁₂-HSL causes the monomerization of the protein (32), while LasR forms multimers and binds DNA only when coupled with 3-O-C₁₂-HSL (9).

The transcriptome of the P. aeruginosa quorum-sensing response has recently been analyzed (28, 33). This analysis showed that hundreds of genes, representing approximately 6% of the P. aeruginosa genome, are either activated or repressed by LasR(3-O-C₁₂-HSL) and/or RhlR(C₄-HSL) (28, 33). The existence of a large group of genes repressed by the quorum-sensing response was reported by the two research groups that made this analysis (28, 33), but one of the groups found a larger number of repressed genes (33). This discrepancy might arise from differences in culture conditions or perhaps from differences in the genetic background of the P. aeruginosa PAO1 derivatives used, as discussed recently (31).

P. aeruginosa contains a third member of the LuxR family of transcriptional regulators called QscR (2) that exerts a negative effect in the expression of several quorum-sensing-regulated traits (2, 11). It was recently reported that QscR forms multimers in the absence of C₄-HSL and 3-O-C₁₂-HSL and lower-order oligomers in the presence of these AIs (11). It was also shown that QscR forms heterodimers, both with RhlR and with LasR (11), revealing an additional level of regulation of the quorum-sensing response at the level of protein interaction among members of the LuxR family of transcriptional regulators.

LasR and RhlR activate transcription through binding to a DNA sequence called the las box, which has been defined by the presence of a conserved sequence and identified in all genes known to be activated by these quorum-sensing transcription regulators (36, 37). The las box has a very similar sequence as that recognized by other members of the LuxR family (3, 36). Furthermore, the LasR(3-O-C₁₂-HSL) complex has been shown to bind and activate the expression of the Vibrio fischeri lux operon, which is normally activated by the LuxR(3-oxo-hexanoyl-homoserine lactone) complex (6). The LuxR protein binds to the lux box mainly when forming a complex with its corresponding AI (3).

Several quorum-sensing transcriptional regulators belonging to the same family as RhlR and LasR have been purified (12, 16, 18, 24, 34, 41, 42). Recently, the three-dimensional structure of the first member of this family was obtained (39). The TraR activator from Agrobacterium tumefaciens was crystallized in complex with its corresponding AI and its target DNA sequence (39). It was found to consist of an asymmetric dimer that interacts with its target DNA sequence through a carboxy-terminal helix-turn-helix motif and with the AI through a hydrophobic pouch present in the amino-terminal half of the protein (39).

It is interesting that some members of the LuxR family of regulatory proteins have different mechanisms for transcriptional activation. In the case of P. aeruginosa LasR (9), A. tumefaciens TraR (12), V. fischeri LuxR (3), and Erwinia carotovora CarR (34), the proteins bind to the DNA target sequence only when complexed with their corresponding AI. The binding of A. tumefaciens TraR to its AI has three effects: it increases the binding affinity for its target DNA sequences, it is necessary for this protein to attain an active conformation, and it prevents degradation by endogenous proteases (41, 42). In the case of E. carotovora CarR (34) and P. aeruginosa LasR (9), it has been shown that the binding of the AI causes its multimerization and binding to the target DNA sequence. On the other hand, the Erwinia chrysanthemi ExpR regulator can bind to its DNA target sequence even in the absence of its corresponding AI, but its conformation is modified by its binding (18). ExpR functions as a transcriptional activator of some promoters, but it is also a repressor of its own promoter. When ExpR acts as a repressor, the presence of AI causes dissociation from its DNA target sequence (24). Pantoea stewartii EsaR (16) and Serratia marcescens SpnR (8) bind to their target sequences in the absence of AI acting as a repressor; this interaction is reversed by AI binding. None of the studied transcriptional regulators of the LuxR family is able to bind to the same target sequence both in complex with AI and in its absence.

The aim of this work was to further characterize the mechanism of transcriptional regulation of P. aeruginosa RhlR. To achieve this, we studied the heterologous expression of the P. aeruginosa rhlAB promoter in Escherichia coli. We determined that the RhlR binding DNA sequence corresponds to the predicted las box in the rhlAB upstream region and also that RhlR binds to this DNA sequence, both in vivo and in vitro, in the presence and in the absence of C₄-HSL. The binding of RhlR(C₄-HSL) to the las box is a necessary condition for rhlAB expression while the binding of RhlR alone to the same DNA sequence seems to repress rhlAB transcription.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

E. coli strains DH5α (27) and ET8000 (13) were cultured in Luria-Bertani (LB) broth (15) and used as stated. P. aeruginosa PAO1 (7) was grown in phosphate-limited peptone-glucose-ammonium salts medium (PPGAS; pH 7.2) containing NH₄Cl₂ (0.02 M), KCl (0.02 M), Tris-HCl (0.12 M), MgSO₄ (0.0016 M), glucose (0.5%), and peptone (1%) (40). Plasmids used are listed in Table 1. When appropriate, isopropyl-β-d-thiogalactoside (IPTG) was added to the medium at the concentration stated in the text. Antibiotics (Sigma Chemical Co.) were used to supplement selection media at the following concentrations in micrograms per milliliter for E. coli and P. aeruginosa, respectively: ampicillin, 150 and not used; carbenicillin, not used and 250; tetracycline, 15 and 100. N-Acyl-homoserine lactones (C₄-HSL and 3-O-C₁₂-HSL) were purchased from Quorum Sciences Inc. (Coralville, Iowa). These AIs were used at a concentration of 10 μM, unless otherwise stated. β-Galactosidase activity was determined as reported previously (15). One Miller unit corresponds to 1 nM o-nitrophenyl-β-d-galactopyranoside hydrolyzed per min and per absorbance unit at 600 nm; the data shown in this work represent the averages of triplicate measurements from two independent experiments.

TABLE 1.

Plasmids used in this study

Plasmid	Relevant characteristic(s)	Reference
pECP61.5	Contains an rhlA::lacZ translational fusion and rhlR under ptac; Apr	21
pMP220	Vector to construct transcriptional lacZ fusions; Tcr	29
pMPCG	pMP220 derivative with an rhlA::lacZ fusion	This study
pUCP20	Cloning vector with plac; Apr	35
pGMYC	pUCP20 expressing rhlR under plac	This study
pMT1	pUCP20 expressing lasR under plac	This study
pThio-R	Plasmid expressing a Thio-RhlR fusion protein; Apr	This study

Strain ET8000rpoN::Tn5 was constructed by transducing the rpoN::Tn5 mutation from strain MX848 (20) to strain ET8000 by using phage P1cm1,clr100 as described elsewhere (15).

Enzymes and reagents.

Plasmids were purified using the Wizard Plus Minipreps DNA purification system (Promega Corporation, Madison, Wis.). DNA bands were cut from agarose gels and purified with the Geneclean III kit (Bio 101, La Jolla, Calif.). DNA restriction enzymes were purchased from New England Biolabs (Beverly, Mass.) or Roche Diagnostics Corporation (Indianapolis, Ind.), and used according to the manufacturers' instructions.

Nucleic acid techniques.

DNA manipulation was performed as reported previously (26). DNA sequencing reactions were done using the Thermosequenase kit (Amersham Life Science Corp., Cleveland, Ohio). Oligonucleotides were radioactively labeled with the T4 polynucleotide kinase (Amersham Life Science Corp.), by using [γ-³²P]ATP (Amersham International) as substrate. Blunt-ended DNA fragments were obtained with the Klenow fragment of DNA polymerase I (Amersham Life Science Corp.).

Plasmid pMPCG was constructed by ligating the 0.8-kb HindIII-BamHI fragment from plasmid pUO58 (19) containing the rhlAB promoter into the same restriction sites of plasmid pMP220 (29) to create an rhlA::lacZ transcriptional fusion. Plasmid pGMYC (14) was constructed by ligating a 1.3-kb blunt-ended SmaI-KpnI fragment from pUO58 (19) unto pMOS-Blue and then subcloning rhlR into the SmaI site of pUCP20 (35) to place rhlR under plac control.

Primer extension analysis.

E. coli strains containing plasmid pECP61.5 (21) were grown to an optical density of 1.5 at 600 nm, and their total RNA was extracted using a GlassMAX RNA Microisolation reagent assembly (Life Technologies Inc., Gaithersburg, Md.). Primer extension reactions were performed using RAV2 reverse transcriptase (Amersham Life Science Corp.) according to the manufacturer's instructions. All reactions were carried out at 42°C. The experiments were done using a primer (oligo1) that corresponds to the sequence between −159 and −179 with respect to the rhlA ATG codon: 5′-GGGGCTTGTGTGGGTCTTGC-3′.

Gel retardation assays.

A 216-bp radiolabeled DNA fragment containing the rhlAB regulatory region was synthesized by PCR, with plasmid pECP61.5 (21) as template and oligo1 and oligo2 (5′-CATGCCTTTTCCGCCAACCCCTCGC-3′) as primers. One of the primers used to amplify this fragment was radiolabeled with [γ-³²P]ATP (Amersham International) prior to the PCR, using T4 polynucleotide kinase. Binding reactions were carried out in buffer containing 10 mM Tris-HCl (pH 7.9), 1 mM EDTA, 1 mM dithiothreitol, 60 mM KCl (Fig. 2A and B) or 100 mM KCl (Fig. 2C), and 10% glycerol. Calf thymus DNA (300 μg/ml) was included in all reaction mixtures. After 20 min of incubation at room temperature (Fig. 2A and B) or 37°C (Fig. 2C), samples were size fractionated on 6% polyacrylamide gels in 0.5× TBE buffer (27).

FIG. 2.

RhlR DNA binding with and without C₄-HSL. The figure shows gel retardation assays of a 216-bp end-labeled PCR product of the rhlA 5′ region. The assays were done in the presence of E. coli DH5α/pECP61.5 cell extract induced with 0.5 mM IPTG. (A) Increasing amounts of cell extract expressing RhlR without AI (lanes 2 to 6) or without protein (lane 1). The protein concentration of the cell extract was 28 μg/ml. (B) Competition assay using increasing amounts of a nonlabeled PCR product containing the rhlAB regulatory region. (C) Gel retardation assay with no protein (lane 1), a cell extract of E. coli DH5α (lane 2), or a cell extract of E. coli DH5α/pECP61.5 grown in the absence of C₄-HSL (lanes 3 to 6) or in the presence of 10 mM C₄-HSL (lanes 7 to 10). (D) Immunoblot of the same cell extracts used in panel C. Shown are results for a cell extract of E. coli DH5α (lane 1) and a cell extract of E. coli DH5α/pECP61.5 grown with 0.5 mM IPTG in the absence of C₄-HSL (lane 2) or in the presence of 20 μM C₄-HSL (lane 3).

Under conditions of ptac induction by IPTG, a considerable fraction of RhlR is present in inclusion bodies. This insoluble protein does not bind to the DNA fragment containing the las box (data not shown).

In vivo methylation protection assays.

In vivo methylation protection assays of the rhlAB operon by binding of the RhlR protein in E. coli were performed as described previously (5) with the following modifications. E. coli DH5α cells with the plasmids pMPCG and pGMYC were grown on LB medium until stationary phase. Dimethyl sulfate was added at a final concentration of 0.1%, and cells were incubated for 1 min. The cell pellet was washed twice with 50 ml of ice-cold saline phosphate solution. Plasmid DNA was purified and cleaved at the methylated positions by incubation with 10% piperidine at 90°C for 30 min. Piperidine was vacuum evaporated, and plasmid DNA was washed twice with 70% ethanol. Primer extension reactions were performed using the purified plasmid DNA as template and either end-labeled oligo1 or end-labeled oligo2 as primer for each DNA strand. The concentrated extension products were separated by electrophoresis in a 7% polyacrylamide gel containing 8 M urea. The results shown in Fig. 2 correspond to one of the rhlA DNA strands, but we found that the same DNA sequence was protected when we analyzed the other strand by using oligo2 (data not shown). Determination of DNA sequences was done with the same primers.

Overproduction of the thioredoxin (Thio)-RhlR protein fusion.

To construct a Thio-RhlR protein fusion, the rhlR gene was PCR amplified using oligo3 (5′-GAGACTGCAGGTCGACTCAGATGAGGCCCAG-3′) and oligo4 (5′-GATAGGTACCAGAATTCATGAGGAATGACGGA-3′). The product was digested with KpnI and SalI and cloned into plasmid pThioC (Invitrogen) digested with the same enzymes; the resulting plasmid pThio-R was transformed into E. coli DH5α. This strain was cultured at 30°C, and when induced with 0.5 mM IPTG, most of the expressed fusion protein formed inclusion bodies. Cells were harvested by centrifugation and ruptured by sonication. The pellet obtained after centrifugation of cell lysate was resuspended and size-fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The band corresponding to Thio-RhlR protein was cut, crushed, and used to immunize rabbits.

Immunoblotting techniques.

New Zealand rabbits were immunized with the protein fusion Thio-RhlR in order to produce polyclonal antibodies. E. coli cultures were grown until they reached the indicated optical density at 600 nm. The cells were collected and lysed by boiling for 5 min in loading buffer (26). Equal amounts of protein from each lysate were separated by sodium dodecyl sulfate-14% polyacrylamide gel electrophoresis. The proteins were transferred by electroblotting from the gel to Hybond-C nitrocellulose membranes (Amersham Life Science Corp.). RhlR was detected by using the rabbit polyclonal antiserum raised against ThioR-RhlR.

RESULTS

Transcription of rhlAB in E. coli is σ⁵⁴ independent.

The transcription start site of the P. aeruginosa rhlAB operon has been reported elsewhere (21). To further characterize this promoter, we evaluated its functionality in E. coli. We determined that the rhlAB transcription start site in E. coli coincided with the previously reported site for P. aeruginosa (Fig. 1). The rhlAB regulatory region contains canonical −10 and −35 sequences corresponding to a promoter transcribed by RNA polymerase (RNA-P) containing a σ⁷⁰ subunit (21). However, the transcription of this promoter has been proposed to be dependent on the RNA-P containing the σ⁵⁴ subunit, based on the reduced level of its expression on a P. aeruginosa rpoN mutant (21). We analyzed the role of σ⁵⁴ in rhlAB expression in E. coli by determining its level of transcription and start site in a mutant devoid of this sigma factor. Figure 1 shows that the expression of this promoter is independent of σ⁵⁴.

FIG. 1.

The rhlAB promoter is σ⁵⁴ independent. The figure shows identification of the rhlAB transcription start site by primer extension analysis (A) and analysis of the level of expression (B) in different E. coli strains. Shown are results with the following strains: DH5α/pECP61.5 (I), ET8000/pECP61.5 (II), and ET8000rpoN::Tn5/pECP61.5 (III). The presence (+) or absence (−) of 10 μM C₄-HSL is indicated. MU, Miller units.

In support of these results, the RhlR transcriptional regulator does not have sequence similarity to members of the EBP family of transcriptional activators like NtrC and NifA, which are specific for expression of σ⁵⁴-dependent promoters (5). Thus, RhlR cannot activate a σ⁵⁴-dependent promoter, and the data presented in this work are related to the transcriptional regulation of rhlAB at a σ⁷⁰-dependent promoter. It is very likely that the RhlR(C₄-HSL)-dependent expression of this promoter in P. aeruginosa is also independent of this sigma factor, since, as already mentioned, this protein does not belong to the EBP family (17). Thus, the observed dependence of rhlAB expression on a functional σ⁵⁴ protein (21) seems to be due to an indirect effect. It was reported elsewhere (14) that rhlR has four different transcriptional start sites and that one of these is σ⁵⁴ dependent. This could explain, at least in part, the apparent regulation of P. aeruginosa rhlAB expression by this alternative sigma factor.

RhlR specifically binds to the rhlAB las box with or without C₄-HSL.

The rhlAB upstream region contains a single las box which is centered at −42 nucleotides with respect to the transcriptional start site (21). To study the specific binding of the RhlR protein in vitro, gel retardation assays were performed using a 216-bp DNA fragment corresponding to the rhlAB upstream region including the las box, with an E. coli DH5α/pECP61.5 (ptac-rhlR and rhlA::lacZ) cell extract overexpressing RhlR (approximately 5% of total cell protein; data not shown). Retardation of the labeled DNA fragment depended on the presence of plasmid pECP61.5 and on the use of a DNA fragment which contains the las box sequence (Fig. 2). Interestingly, we found that RhlR binds to its target sequence both in the presence and in the absence of its AI, C₄-HSL (Fig. 2). This is in contrast to other transcriptional regulators of the LuxR family, which bind to DNA under only one of these conditions (3, 6, 9, 12, 18, 34). The addition of 3-O-C₁₂-HSL alone or with C₄-HSL has no effect on RhlR binding to its target DNA (data not shown).

Figure 2 shows results suggesting that the binding affinity of RhlR for its target DNA sequence is higher than that of RhlR(C₄-HSL). A cell extract of DH5α/pECP61.5 grown in the absence of C₄-HSL is able to retard a higher amount of the labeled DNA fragment containing the rhlA las box than is a cell extract of the same strain grown in the presence of 20 μM C₄-HSL (Fig. 2C). These two cell extracts contained the same RhlR concentration (Fig. 2D). Furthermore, a cell extract of DH5α/pECP61.5 grown in the absence of AI also shows a higher retardation ability than does the same extract incubated for 2 min or overnight with C₄-HSL, even though the AI concentration needed to modify the apparent affinity of RhlR for the las box is much higher (data not shown). These results show that RhlR has a higher affinity for the rhlA las box than does RhlR(C₄-HSL); however, to determine the dissociation constant for their target DNA, it will be necessary to use purified RhlR.

Gel retardation assays show only whether a protein interacts with a specific DNA fragment, but more precise information on what type of complex is formed or the DNA sequence involved in this interaction cannot be obtained with this type of experiment. To determine whether the conformation of the RhlR/las box complex was modified by the interaction of RhlR with its AI, we used an in vivo DNA methylation protection assay (Fig. 3). To do this experiment, we used two plasmids, pGMYC (plac-rhlR) and pMPCG (rhlA::lacZ). In agreement with the results obtained in the gel retardation assay, we found that RhlR binds in vivo, both in the presence and in the absence of C₄-HSL, to a DNA region that includes the sequence defined as the las box. Figure 3 shows that several guanine residues in this region were protected from methylation or were hypermethylated. Even more, the in vivo footprinting assay enabled us to detect several differences in the way that RhlR binds to its target DNA sequence depending on its coupling to C₄-HSL (Fig. 3). Changes in the methylation of guanine residues, either protection or hypermethylation, were apparent in the sequence spanning from the guanine at position −20 to that at −60 with respect to the rhlAB transcription start site, which includes the las box (Fig. 3). Guanines at positions −39, −40, −49, −58, and −64 are protected, and those at −20 and −65 are hypermethylated (Fig. 3).

FIG. 3.

In vivo RhlR DNA binding. The figure shows dimethyl sulfate footprinting analysis of the RhlR binding to the 5′ rhlAB region. Lanes: 1, E. coli DH5α/pMPCG,pUCP20 supplemented with C₄-HSL; 2, E. coli DH5α/pMPCG,pGMYC supplemented with C₄-HSL; 3, E. coli DH5α/pMPCG,pUCP20 supplemented with 1 mM IPTG; 4, E. coli DH5α/pMPCG,pGMYC supplemented with 1 mM IPTG. oligo1 was used as the primer for the reverse transcriptase reaction and also to determine the DNA sequence (the ladder used to determine this sequence is shown). The sequence and position of the predicted las box and of the −35 promoter sequences are shown at the left of the figure. The positions of the protected (open circles) or hypermethylated (closed circles) guanine residues are shown; those marked with an asterisk are protected or hypermethylated only in the absence of C₄-HSL and in the presence of IPTG (lane 4).

The different conformation of the DNA-bound RhlR protein with or without C₄-HSL is apparent from the differential protection of the guanine residues at −20 and −39. The protection of guanine at −39 and hypermethylation of that at −20 by RhlR binding are apparent only in the presence of RhlR without C₄-HSL (Fig. 3).

RhlR uncoupled from C₄-HSL is a repressor of rhlAB transcription.

As was previously described (14, 21), we found that RhlR(C₄-HSL) is necessary for rhlAB transcriptional activation (Table 2). No expression of an rhlA::lacZ fusion was seen in the absence of C₄-HSL (Table 2). The LasR(3-O-C₁₂-HSL) complex can activate rhlAB promoter expression but at a considerably lower level than the RhlR(C₄-HSL) complex (21) (Table 2). Unexpectedly, however, we found that the expression of the rhlAB promoter is slightly, but reproducibly, lower when RhlR expression is induced by the addition of 1 mM IPTG, even in the presence of C₄-HSL (Table 2). This result suggests that RhlR unbound from its AI could act as an rhlAB transcriptional repressor. The overexpression of LasR as a result of IPTG addition had no repressor activity on rhlAB promoter expression (Table 2).

TABLE 2.

Expression of an rhlA::lacZ fusion in E. coli DH5α

Plasmid	β-Galactosidase activity in Miller units (% of maximum activity)a
	NA	AI	IPTG	AI + IPTG
pMPCG	2.3 ± 0.7 (1.4)	4.1 ± 0.3 (2.6)	1.9 ± 0.4 (1.2)	1.9 ± 0.4 (1.2)
pMPCG/pGMYC	1.5 ± 0.2 (0.9)	157 ± 14 (100)	1.2 ± 0.1 (0.7)	130 ± 12 (83)
pMPCG/pMT1	0.6 ± 0.2 (0.3)	12.3 ± 1.3 (7.8)	0.6 ± 0.3 (0.3)	12.1 ± 0.4 (7.6)

^aPercentage of the maximum activity detected in all experiments presented in the table. NA, no addition. AI refers to C₄-HSL (experimental results shown in the first two rows) or 3-O-C₁₂-HSL (experimental results shown in the third row) added at 10 μM. IPTG was added at 1 mM. All measurements were done after 8 h of growth on LB medium.

To further investigate whether the results presented in Table 2 had any significance, and whether indeed RhlR could act as a repressor of the rhlAB promoter, we used plasmid pECP61.5, which contains rhlR under the control of the tac promoter and an rhlA::lacZ fusion in the same plasmid. The use of a plasmid containing both the gene coding for the transcriptional regulator and the gene that is regulated minimizes the possibility of artifacts due to different copy numbers of the plasmids used. Using this system, we measured rhlA::lacZ expression and RhlR concentration in the cell soluble fraction by Western immunoblotting on DH5α/pECP61.5 cells grown in the presence of 10 μM C₄-HSL with two concentrations of and without 1 mM IPTG (Fig. 4). Under these conditions it was apparent that rhlA::lacZ is considerably repressed (up to approximately 33%), correlating with an accumulation of RhlR (Fig. 4).

FIG. 4.

Repression of the rhlAB promoter by RhlR in the E. coli background. The figure shows β-galactosidase activities (A) and RhlR concentrations detected by immunoblotting in the cell extract supernatant (B), after 8 h of growth of E. coli DH5α/pECP61.5 in the presence of 10 μM C₄-HSL and the following IPTG concentrations: 0, white bar and lane 1; 0.1 mM, black bar and lane 2; 0.5 mM, hatched bar and lane 3. The molecular size markers are shown with an arrow indicating the 30-kDa standard. MU, Miller units.

The repression of the rhlAB promoter by RhlR can also be visualized by adding IPTG to, and thus increasing the RhlR concentration in, an E. coli DH5α/pMPCG-pGMYC culture after 4 h of growth in the presence of AI (data not shown). Repression of rhlAB transcription by RhlR is a reversible phenomenon since addition of C₄-HSL to an E. coli DH5α/pMPCG-pGMYC culture after 4 h of growth in the presence or absence of 1 mM IPTG results in further rhlA::lacZ induction. We found that the final level of rhlA::lacZ expression is 22% higher in this experiment in the absence of IPTG, i.e., when RhlR is not overexpressed.

To obtain a clearer picture of the relationship between the level of RhlR unbound to C₄-HSL and the repression of the rhlAB promoter, we performed an experiment where both RhlR and C₄-HSL concentrations were modified (Fig. 5). We varied the level of rhlR induction and thus RhlR concentration by adding or not adding 0.1 mM IPTG and modified the AI concentration by adding 0, 10, or 50 μM C₄-HSL to E. coli DH5α/pECP61.5 cultures, at the beginning of the culture or at 3 h of growth. It is clear from these data (Fig. 5) that, under the condition where the RhlR concentration is higher (addition of 0.1 mM IPTG at the beginning of the culture) and C₄-HSL is low (10 μM), rhlAB expression is lower (Fig. 5, bar 2). However, with the same C₄-HSL concentration (10 μM), rhlAB expression is highest when RhlR expression is lowest (without IPTG addition) (Fig. 5, bar 3). If RhlR concentration is in between these conditions (addition of 0.1 mM IPTG at 3 h of growth), the level of rhlAB induction is also in between the previous values obtained (Fig. 5, bar 4) and is increased by increasing the C₄-HSL concentration to 50 μM (Fig. 5, bar 5). These results further support the hypothesis that the ratio of RhlR/RhlR(C₄-HSL) determines the level of induction of the rhlAB promoter. We conclude that, when this ratio is high, i.e., when a significant amount of unbound RhlR is present, due to either overexpression of RhlR or limitation of available C₄-HSL (Fig. 5), the expression of the rhlAB promoter is repressed by RhlR binding to the las box.

FIG. 5.

Effect of RhlR/RhlR(C₄-HSL) on rhlAB expression in the E. coli background. The figure shows β-galactosidase activities (A) and RhlR concentrations detected by immunoblotting (B) after 8 h of growth of E. coli DH5α/pECP61.5 under the following culture conditions: LB medium without supplement (black bar and lane 1), LB medium supplemented with 0.1 mM IPTG and 10 μM C₄-HSL (striped bar and lane 2), LB medium supplemented with 10 μM C₄-HSL (gray bar and lane 3), nonsupplemented LB medium with growth for 3 h and addition of 0.1 mM IPTG and 10 μM C₄-HSL (gridwork bar and lane 4), and nonnsupplemented LB medium with growth for 3 h and addition of 0.1 mM IPTG and 50 μM C₄-HSL (white bar and lane 5).

RhlR repressor activity is not specific in its expression in the E. coli background, since it is also detected in P. aeruginosa PAO1. Figure 6 shows that increasing RhlR concentration by adding 1 mM IPTG to a P. aeruginosa PAO1/pECP61.5 culture resulted in a reduced level of rhlAB expression, attaining almost 50% repression. Although we have not determined the RhlR and C₄-HSL concentrations, we found that addition of this AI to a P. aeruginosa PAO1 culture has no significant effect on rhlAB expression (Fig. 6), presumably because RhlR is expressed at a low level, as most regulatory proteins are, and thus there is not a significant amount of unbound RhlR in the cells. However, overexpression of rhlR by IPTG addition, causing an increased expression of RhlR (Fig. 6B) which in turn presumably results in a higher RhlR/RhlR(C₄-HSL) ratio, leads to rhlAB repression (Fig. 6A).

FIG. 6.

Repression of rhlAB promoter by RhlR in the P. aeruginosa background. The figure shows β-galactosidase activities (A) and RhlR concentrations detected by immunoblotting (B), after 12 h of growth of P. aeruginosa PAO1/pECP61.5. Lanes and bars: 1, nonsupplemented culture medium; 2, addition of 10 μM C₄-HSL; 3, addition of 10 μM C₄-HSL and 1 mM IPTG.

DISCUSSION

We have shown in this work that RhlR can bind to the rhlAB las box, in vitro and in vivo, in the presence and absence of C₄-HSL (Fig. 2 and 3). These results are in agreement with recently reported observations showing that RhlR forms dimers both in the presence and in the absence of C₄-HSL (11). In addition, we have provided evidence suggesting that rhlAB expression is repressed by RhlR unbound to this AI (Fig. 4, 5, and 6). We propose that RhlR(C₄-HSL) activates transcription when bound to the las box by its direct interaction with RNA-P containing a σ⁷⁰ subunit, as has been reported for many bacterial transcriptional activators (25), including LuxR (30).

When RhlR is uncomplexed with C₄-HSL, it seems to bind its target DNA sequence with a different conformation (Fig. 3). We found that the protection of guanine at −39 and hypermethylation of that at −20 are apparent only when free RhlR binds to this DNA sequence (Fig. 3). This conformation might interfere with RNA-P binding to the promoter or else prevent it from proceeding with gene transcription, at the level of either initiation or elongation, as has been shown for repressors binding to the −35 promoter region (26).

In the absence of C₄-HSL, RhlR seems to bind the rhlAB las box with a slightly higher affinity than does RhlR(C₄-HSL) (Fig. 2); thus, it is possible that the unbound regulatory protein acts as an antiactivator of this promoter by competing with RhlR(C₄-HSL) for las box binding. If confirmed, this dual RhlR transcriptional regulatory mechanism might permit a very sensitive and rapid control of gene expression depending on the available C₄-HSL concentration.

The experimental conditions used in this work that enable us to propose that the rhlAB promoter is activated by RhlR(C₄-HSL) and repressed by RhlR implied the overexpression of this transcriptional regulator, so that physiological conditions were not maintained. However, we detected repression of rhlAB expression of more than 30% in E. coli and 50% in P. aeruginosa, and we suppose that such high levels of repression are not physiological either. As mentioned above, we propose that the dual function of RhlR as an activator and as a repressor of rhlAB expression depending on C₄-HSL binding represents a rapid and subtle genetic regulatory mechanism.

We reported here, as has already been shown (21), that transcription of the rhlAB operon is fully activated by RhlR(C₄-HSL) and that LasR(3-O-C₁₂-HSL) can activate its expression at a lower level (Table 2). We found that increasing LasR expression did not result in repression of rhlAB expression (Table 2), as expected from the inability of this transcriptional regulator to bind to las boxes in the absence of 3-O-C₁₂-HSL (9). There are other genes in P. aeruginosa, like lasB, that are regulated by these proteins in an opposite manner: they are fully activated by LasR(3-O-C₁₂-HSL) and to a lesser extent by RhlR(C₄-HSL) (21). It has been proposed, based on the regulation of the lasB promoter in different P. aeruginosa mutants, that RhlR when unbound to C₄-HSL acts as a repressor of the lasB promoter (1). This previous report thus provides experimental evidence with a different promoter (lasB) that agrees with the data presented in this work regarding RhlR repression of the rhlAB promoter. On the other hand, there is evidence suggesting that RhlR represses the expression of its own gene (14). These two reported signs of RhlR repressor activities were observed under conditions that did not imply the overexpression of this transcriptional regulator (1, 14). Thus, it is possible that RhlR repressor activity is a regulatory mechanism that controls several quorum-sensing-regulated genes in P. aeruginosa.

The transcriptome analysis of the P. aeruginosa quorum-sensing response has revealed that either LasR(3-O-C₁₂-HSL) or RhlR(C₄-HSL) or both negatively regulate gene expression (28, 33). However, one of these studies (32) detected a considerably larger number of repressed genes (31), with the use of a PAO1 lasI rhlI double mutant, whereas the other study used a lasR rhlR double mutant (28). It is possible that in the latter case the number of repressed genes is underestimated because at least one of these regulatory proteins, RhlR, is able to both activate and repress specific genes depending on its interaction with C₄-HSL, as shown in this work. Thus, the absence of RhlR from the cell will have a different effect than just the absence of AIs.

We conclude that the dual function of RhlR as an activator and repressor of gene transcription represents a novel element in the P. aeruqinosa quorum-sensing response that needs to be taken into account to understand the complex and fine-tuned genetic regulatory network controlling the expression of a significant fraction of this bacterial genome, including those genes encoding virulence-associated traits.