The PqsR and RhlR Transcriptional Regulators Determine the Level of Pseudomonas Quinolone Signal Synthesis in Pseudomonas aeruginosa by Producing Two Different pqsABCDE mRNA Isoforms

Abstract

Regulation of gene expression plays a key role in bacterial adaptability to changes in the environment. An integral part of this gene regulatory network is achieved via quorum sensing (QS) systems that coordinate bacterial responses under high cellular densities. In the nosocomial pathogen Pseudomonas aeruginosa, the 2-alkyl-4-quinolone (pqs) signaling pathway is crucial for bacterial survival under stressful conditions. Biosynthesis of the Pseudomonas quinolone signal (PQS) is dependent on the pqsABCDE operon, which is positively regulated by the LysR family regulator PqsR and repressed by the transcriptional regulator protein RhlR. However, the molecular mechanisms underlying this inhibition have remained elusive. Here, we demonstrate that not only PqsR but also RhlR activates transcription of pqsA. The latter uses an alternative transcriptional start site and induces expression of a longer transcript that forms a secondary structure in the 5′ untranslated leader region. As a consequence, access of the ribosome to the Shine-Dalgarno sequence is restricted and translation efficiency reduced. We propose a model of a novel posttranscriptional regulation mechanism that fine-tunes PQS biosynthesis, thus highlighting the complexity of quorum sensing in P. aeruginosa.

INTRODUCTION

The opportunistic pathogen Pseudomonas aeruginosa is a ubiquitous bacterium which is able to thrive in a wide variety of environments. It is frequently associated with severe nosocomial infections and is found to be the leading cause of morbidity and mortality among people with cystic fibrosis (1). The remarkable ecological success of P. aeruginosa can be attributed to its large metabolic versatility and its sophisticated quorum sensing (QS) network. This cell-to-cell communication enables P. aeruginosa to control expression of numerous virulence factors and is involved in biofilm formation, thus facilitating establishment of acute and chronic infections.

QS in P. aeruginosa is tightly regulated by at least three different systems organized in a hierarchical manner. The acylhomoserine lactone (AHL)-dependent las system is considered to stand at the top of the hierarchy. It is composed of the LuxRI homologues LasR and LasI. The signal synthase LasI directs the synthesis of N-(3-oxododecanoyl)-l-homoserine lactone (3-oxo-C₁₂-HSL), which is the ligand of the LasR receptor (2, 3). The las system is interconnected with the rhl system, as LasR bound to its autoinducer 3-oxo-C₁₂-HSL induces production of RhlRI (4). LasR-3-oxo-C₁₂-HSL and RhlR-C₄-HSL direct transcription of several QS-regulated virulence factors, including pyocyanin, hydrogen cyanide, and exotoxin A (5, 6). In addition, P. aeruginosa possesses a third QS system, pqs, based on 2-alkyl-4-quinolone (AQ) signal molecules (7). The Pseudomonas quinolone signal (PQS) and its direct precursor 2-heptyl-4-quinolone (HHQ) are active members of the over 50 different AQs produced by P. aeruginosa (8). Like AHLs, they play an important role in the expression of several virulence factors as well as in inducing a protective stress response toward deteriorating environmental conditions (9–11). Investigations of PQS biosynthesis revealed that the major synthase genes are arranged in a polycistronic operon (pqsABCDE) and that transcription of this operon is under the control of PqsR, the LysR-type transcriptional regulator of the pqs system (12, 13). PqsR, which is activated by HHQ and PQS binding (14), plays a critical role in the pathogenicity of P. aeruginosa and is regulated by both the las and rhl system (13, 14) and the small RNA PhrS (15). While LasR activates pqsR transcription and subsequently enhances pqsABCDE expression, RhlR was found to be a repressor of pqsR transcription (14). Interestingly, RhlR was also found to inhibit pqsABCDE expression by binding to an las-rhl box centered at bp −311 upstream of the pqsA transcriptional initiation site (16). We have recently shown that there is an alternative transcriptional start site (pqsA −339) just downstream of the RhlR-binding site (17), indicating that repression of PqsA production via RhlR might be posttranscriptional.

Posttranscriptional repression events in prokaryotes occur mainly either by binding of small regulatory RNAs to target mRNA molecules or by formation of secondary structures in the mRNA, which play an important role in posttranscriptional regulation of gene expression in bacteria (18, 19). One common form of an RNA regulatory element is the so-called riboswitch. These regulatory elements usually reside in the noncoding region of the mRNA and regulate gene expression by forming alternative structures in response to binding of a specific metabolite (20). Bacteria commonly mask the Shine-Dalgarno (SD) sequence to block access of the 30S ribosomal subunit. With this, translation initiation becomes highly dependent on the folding structure of the initiation region of the mRNA (21). A well-studied example of such a regulatory mechanism includes RNA thermosensors. They form a zipper-like structure which unwinds with increasing temperature, allowing successful binding of the ribosome (22). For example, in Yersinia the virulence factor LcrF is expressed at 37°C, but access to the SD sequence is abolished at 26°C (23, 24). Thermosensors are also known to occur in Pseudomonas species. Recently, the bacterial small heat shock protein IbpA was found to be under the control of two temperature-sensitive hairpin structures in the 5′ untranslated region (UTR) of ibpA in P. aeruginosa (25).

In the present study, we show that repression of PQS biosynthesis by RhlR in P. aeruginosa is due to conformational masking of the translation initiation site of the pqsABCDE transcript. We demonstrate that RhlR promotes transcription of pqsABCDE from an alternative transcriptional start site. The resulting long 5′ UTR folds into a structure, which hinders association of the 30S ribosomal subunit with the mRNA and impedes translation initiation.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Unless otherwise noted, bacterial strains listed in Table 1 were grown in Luria broth (LB) medium at 37°C with shaking at 180 rpm. Escherichia coli DH5α was routinely used for subcloning and propagation. For plasmid selection and maintenance, antibiotics were added at the following final concentrations: for E. coli, 100 μg/ml ampicillin, 12.5 μg/ml tetracycline, and 15 μg/ml gentamicin; for P. aeruginosa, 400 μg/ml carbenicillin, 100 μg/ml tetracycline, and 30 μg/ml gentamicin.

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Relevant feature(s)	Source or reference
Strains
E. coli DH5α	F− endA1 glnV44 thi-1 recA1 relA1 gyrA96 deoR nupG ϕ80dlacZΔM15 Δ(lacZYA-argF)U169 hsdR17 (rK− mK+) λ−	43
E. coli BL21(DE3)	F− ompT hsdSB(rB− mB−) gal dcm	Stratagene
E. coli S17-1	Mobilizing strain for RP4 Mob-containing plasmids
PA14	Wild type	44
PA14 ΔpqsR mutant	pqsR knockout mutant	This study
PA14 ΔrhlR mutant	rhlR knockout mutant	This study
Plasmids
pEX18Ap	Gene replacement vector with MCS from pUC18, oriT+ sacB+, Ampr	26
pEX18Ap2	pEX18TAp derivative, 845-bp fragment containing 5S rRNA and lacZ-alpha genes and multiple cloning site (MCS) removed by inverse PCR, novel MCS generated with unique restriction sites for XhoI, PstI, SmaI/XmaI, XbaI, SacI, HindIII, NheI, NotI, MluI, KpnI, BamHI, EcoRI, Ampr	27
pEX18Ap-ΔpqsR::FRT-Gm	Gene replacement vector for PA14 pqsR containing an FRT-Gm cassette, Ampr	This study
pEX18Ap2-ΔrhlR::FRT-Gm	Gene replacement vector for PA14 rhlR containing an FRT-Gm cassette, Ampr	This study
pFLP3	FLP expression vector, sacB+ oriT+, Ampr Tcr	29
pBBR1-MCS5-Terminator-RBS-Lux (pMTRL)	Broad-host-range low-copy-number vector pBBR1-MCS5 harboring luxCDABE and terminators lambda T0 rrnB1 T1 for plasmid-based transcriptional fusions, Gmr	30
p339	−501 to −338 fragment upstream of pqsA cloned into pMTRL using SpeI and PstI sites, Gmr	This study
p71	−256 to −70 fragment upstream of pqsA cloned into pMTRL using SpeI and PstI sites, Gmr	This study
plong	−501 to +1 fragment upstream of pqsA cloned into pMTRL using SpeI and PstI sites, Gmr	This study
pUCP20	Escherichia-Pseudomonas shuttle vector with beta-lactamase (bla) and LacZ alpha peptide (lacZ alpha) genes; Apr Cbr	45
pUCP20-TOEpqsA_(1)	−72 to +60 fragment of pqsA containing a T7 promoter (5′-GAAATTAATACGACTCACTATAGG-3′) and an EcoRI restriction site at the 3′ end cloned into pUCP20 using the SmaI site, Ampr	This study
pUCP20-TOEpqsA_(2)	−140 to +60 fragment of pqsA containing a T7 promoter (5′-GAAATTAATACGACTCACTATAGG −3′) and an EcoRI restriction site at the 3′ end cloned into pUCP20 using the SmaI site, Ampr	This study
pUCP20-TOEpqsA_(3)	pUCP20-TOEpqsA_long containing a mutation in the 5′ UTR of psqA, where CGTTC was replaced by AAGAA, Ampr	This study
pUCP20_ΔRBS	Plasmid-borne ribosomal binding site AGGAAA of pUCP20 was replaced by CCTCGC; Ampr	This study
pUCP20_ΔRBS-pqsA-His6	Fragment containing −80 bp of the 5′ UTR and the entire coding sequence of pqsA was cloned into pUCP20_ΔRBS using KpnI and HindIII sites, Ampr	This study
pUCP20_ΔRBS-ΔpqsA-His6	pUCP20_ΔRBS-pqsA-His6 containing a mutation in the 5′-UTR of psqA, where CGTTC was replaced by AAGAA, Ampr	This study

Construction of knockout mutants.

To create the single-knockout mutants in the wild-type PA14 parental strain, we used an adapted version of the gene replacement method in reference 26 by using plasmid pEX18Ap for pqsR and pEX18Ap2 (27) for rhlR and a gentamicin resistance cassette flanked by flippase recombination target (FRT-Gm) originating from plasmid pPS856. The mutant fragments were constructed by PCR extension overlap (28) with the following primers: for pqsR, upstream region pqsRup2FEcoRI/pqsRup2RBamHI and downstream region pqsRdwFcompup/pqsRdwRHindIII; and for rhlR, upstream region uprhlRNotI-fw/uprhlRNheI-rv and downstream region dorhlRNheI-fw/dorhlRHindIII-rv (Table 2). The BamHI site was introduced between the upstream and downstream regions of pqsR and at the NheI site for rhlR. These restriction sites were used to insert the FRT-Gm cassette. In the case of pqsR, the primers were designed to target for deletion of the 529-bp upstream region of pqsR, as well as 487 bp part of the coding sequence, while for rhlR the primers were designed to delete the entire coding sequence of the gene. The resulting plasmids, pEX18Ap-ΔpqsR::FRT-Gm and pEX18Ap2-ΔrhlR::FRT-Gm, were transferred into P. aeruginosa PA14 by two-parental mating using the donor strain E. coli S17-1. P. aeruginosa cells were selected on nalidixic acid (20 μg/ml) and gentamicin (50 μg/ml). The occurrence of the double crossover was checked by plating at least 30 colonies from the mating result on gentamicin- and carbenicillin (400 μg/ml)-containing agar plates. Gentamicin-resistant and carbenicillin-sensitive bacteria were isolated, and the insertion of the FRT-Gm cassette was ensured by PCR. Finally the FRT-Gm cassette was removed from the chromosomal DNA with help of flippase encoded on pFLP3 plasmid (29). The knockout mutation was confirmed by PCR using primers annealing outside any pEX18Ap/pEX18Ap2-mediated deletion regions.

TABLE 2.

Oligonucleotides

Primer	Sequence (5′→3′ direction)a
Mutagenesis
pqsRup2FEcoRI	GAGAATTCATCCACCGGGCAGCCCAG
pqsRup2RBamHI	CGGGATCCGTTAGCGTAGCCACCGGCCAGGC
pqsRdwFcompup	TGGCTACGCTAACGGATCCCGTCGGCTACACCAAGGCGTTC
pqsRdwRHindIII	CCCAAGCTTGAGAACGCTCTACTCTGGTGCG
uprhlRNotI-fw	TATGCGGCCGCTGCAGCGCGCCTACGCG
uprhlRNheI-rv	TCAGTCAGTCAGCTAGCTGCAGTAAGCCCTGATCGATAAAATGCA
dorhlRNheI-fw	GCTAGCTGACTGACTGAAGCGCAGGGCGCGCCG
dorhlRHindIII-rv	TATAAGCTTGGCGGCGTAGCGCGAAAGC
lux reporter
pqsA−339-SpeI-fw	TCAGACTAGTGGAGGCTGCAAATGGCA
pqsA−339-PstI-rv	ATACTGCAGGCAAACCAATGACAACCCACTTTGC
pqsA−71-SpeI-fw	TCAGACTAGTTGCCGCCCTTCTTGCTTG
pqsA−71-PstI-rv	ATACTGCAGTCAGGAACGGGAAACTAGCGG
pqsA-ATG-PstI-rv	ATACTGCAGGACAGAACGTTCCCTCTTCAGC
CD spectroscopy
Short	TGACAAAGCAAGCGCTCTGGCTCAGGTATCTCCTGATCCGGATGCATATCGCTGAAGAGGGAACGTTCTGT
Long	CGTTCCTGACAAAGCAAGCGCTCTGGCTCAGGTATCTCCTGATCCGGATGCATATCGCTGAAGAGGGAACG
Derep	AAGAACTGACAAAGCAAGCGCTCTGGCTCAGGTATCTCCTGATCCGGATGCATATCGCTGAAGAGGGAACG
His6-tagged pqsA
pUCP20_RBSmut-fw	GGAATTGTGAGCGGATAACAATTTCACACCCTCTGCAGCTATGACCATGATTACGAATTCCCCG
pUCP20_RBSmut-rv	CGGGGAATTCGTAATCATGGTCATAGCTGCAGAGGGTGTGAAATTGTTATCCGCTCACAATTCC
pqsA_KpnI-fw	TATCGGTACCCCCGTTCCTGACAAAGCAAGCG
pqsA_mut_KpnI-fw	TATCGGTACCCCAAGAACTGACAAAGCAAGCGCTCTGGC
pqsA_6His-rv	TATCAAGCTTTCATCATCAGTGGTGGTGGTGGTGGTGACATGCCCGTTCCTCCGGAAGGTT
Toeprinting
T7-pqsA200nt-fw	GAAATTAATACGACTCACTATAGGCTCCCCGAAACTTTTTCGTTCGGACTC
T7-pqsA132nt-fw	GAAATTAATACGACTCACTATAGGTGACAAAGCAAGCGCTCTGGC
Toe_EcoRV-rv	TAAGATATCGGTATCGGGATCGAAATCGAGGCG
Toe_short-rv	TCGAAATCGAGGCGGAACAGAACC
5′UTRpqsA_mut-f	CCAGAGCGCTTGCTTTGTCAGTTCTTGGAAACTAGCGGCGCTGGGC
5′UTRpqsA_mut-r	GCCCAGCGCCGCTAGTTTCCAAGAACTGACAAAGCAAGCGCTCTGG

^aRestriction sites are underlined.

Bioluminescence assays.

To generate luxCDABE (lux) reporter fusion plasmids, pqsA 5′ leader sequence fragments were PCR amplified from PA14 chromosomal DNA, digested with SpeI and PstI, and subcloned into pMTRL (30). p339 was generated from primers pqsA−339-SpeI-fw and pqsA−339-PstI-rv, p71 from primers pqsA−71-SpeI-fw and pqsA−71-PstI-rv, and plong from primers pqsA−339-SpeI-fw and pqsA-ATG-PstI-rv (Table 2). The resulting plasmids were transferred into the P. aeruginosa PA14 wild-type, PA14 ΔpqsR, and PA14 ΔrhlR strains. The reporter strains were grown overnight with the appropriate antibiotic and then subcultured from an optical density at 600 nm (OD₆₀₀) of 0.05 in BM2 medium [7 mM (NH₄)₂SO₄, 40 mM K₂HPO₄, 22 mM KH₂PO₄, 0.4% glucose, and 2 mM MgSO₄] containing 0.01% Casamino Acids and 10 μM FeSO₄ and grown to mid-exponential growth phase at 37°C (OD₆₀₀ ∼ 2.0). Bioluminescence was monitored using an EnSpire Multimode plate reader (PerkinElmer). Promoter activities are given as the relative luminescence of 200 μl of the cultures measured in a 96-well plate divided by the OD₆₀₀ (relative light units [RLU] OD₆₀₀⁻¹). All results represent the means from at least three independent replicates.

Thermal unfolding of DNA oligonucleotides.

UV absorption spectra of 71-bp-long oligonucleotides (Table 2) were recorded on a JASCO J-815 circular dichroism (CD) spectrometer at a concentration of 10 μM in 50 mM potassium phosphate buffer (pH 7.2). Thermal unfolding of the secondary structure was monitored as an increase in absorption at 255 nm as a function of temperature in intervals of 1°C and a ramp rate of 4°C per minute. To determine the melting temperatures, raw data were fitted with an equation for a dual-step unfolding of a monomer with corrections for linear changes of the CD signal before and after the unfolding transition (31).

Generation of His₆-tagged fusion pqsA and immunoblotting.

The impact of folding structures in the 5′ UTR on the in vivo translation efficiency of pqsA in E. coli and P. aeruginosa was analyzed by cloning C-terminal His-tagged pqsA from nucleotide −80 to the stop codon into the KpnI/HindIII sites of pUCP20_ΔRBS, a pUCP20 derivate lacking the plasmid-borne ribosomal binding site generated by site-specific mutagenesis (QuikChange II site-directed mutagenesis kit; Agilent Technologies) by using the primer set pUCP20_RBSmut-fw/pUCP20_RBSmut-rv (Table 2), according to the manufacturer's instructions. pUCP20_ΔRBS-pqsA-His₆ was generated using the primer set pqsA_KpnI-fw/pqsA_6His-rv, and primers pqsA_mut_KpnI-fw/pqsA_6His-rv were used for generation of pUCP20_ΔRBS-ΔpqsA-His₆ (Table 2). To prepare samples for Western blot analysis, whole-cell lysates of cultures grown to exponential growth phase in LB medium were normalized for protein content, and 10 μl of an OD₆₀₀ of 10.0 was separated by SDS-PAGE (10% acrylamide) after 15 min of incubation at 95°C. We used a His-tagged mouse IgG1 monoclonal antibody (Novagen) at a dilution of 1:1,000 as the primary antibody. A4a goat anti-mouse IgG and IgM (Dianova) were used as a secondary antibody at a dilution of 1:2,000. Blots were developed using Lumi-Light Western blotting substrate (Roche), and chemiluminescence was detected using a Las-1000 Luminescent image analyzer (Fujifilm).

Toeprinting analysis.

Primer extension inhibition (toeprinting) assays were performed as described previously (24). Plasmid templates were generated by cloning PCR-amplified fragments comprising a T7 promoter sequence (5′-GAAATTAATACGACTCACTATAGG-3′) at the 5′ end, an EcoRV site at the 3′ end, and nucleotides −140 and −72 to +60 of pqsA, with usage of primers T7-pqsA200nt-fw/Toe_EcoRV-rv and T7-pqsA132nt-fw/Toe_EcoRV-rv (Table 2), into SmaI cut pUCP20 (blunt end treated), resulting in pUCP20-TOEpqsA_(2) and pUCP20-TOEpqsA_(1), respectively. To generate pUCP20-TOEpqsA_(3), site-specific mutagenesis of pUCP20-TOEpqsA_(2) was carried out using the QuikChange II site-directed mutagenesis kit (Agilent Technologies) with the primers 5′UTRpqsA_mut-f/5′UTRpqsA_mut-r (Table 2), according to the manufacturer's instructions. For in vitro transcription with T7 RNA polymerase, plasmids were linearized by digestion with EcoRV, and the primer Toe_short-rv was used for reverse transcription. The experiments were carried out at 37°C in the presence and absence of E. coli 30S ribosomal subunits.

RESULTS

RhlR induces transcription of the PQS biosynthetic operon via an alternative transcriptional start site.

In an early attempt to investigate regulation of PQS synthesis, RhlR-C₄-HSL was found to negatively regulate pqsR transcription and thus to inhibit PQS synthesis (14). Later, Xiao and colleagues showed binding of RhlR to an las-rhl box in the 5′ leader sequence of pqsA (Fig. 1A) (16). Deletion of the entire las-rhl box significantly increased pqsA transcription in P. aeruginosa wild-type cells but not in ΔrhlR mutant cells. It was therefore suggested that binding of RhlR to the las-rhl box lowers pqsA transcription (16).

FIG 1.

Regulatory elements within the 5′ UTR of the pqsA gene and constructs used in this study. (A) Promoter region of pqsA. The pqsA transcriptional start sites are indicated by bent arrows at pqsA −339 and pqsA −71, and the pqsA ATG start codon is underlined. The binding sites of RhlR, PqsR, and the ribosome are boxed and labeled. (B) Model of the pqsA promoter region. (C) pqsA promoter fusion (p339 and p71) and translational (plong) constructs used in the present study. − and + indicate positions to the start codon of pqsA. Lines are not drawn to scale.

To address this further, we generated luxCDABE promoter fusions that contained the recently predicted pqsA −339 (17) (p339) and pqsA −71 (p71) transcription initiation sites of the pqsABCDE operon (Fig. 1B and C). Expression of lux in the P. aeruginosa PA14 parental strain and ΔrhlR and ΔpqsR mutant strains was monitored during exponential growth phase (17). Consistent with Dötsch and colleagues (17), we found expression from the region containing pqsA −339 as well as pqsA −71 (Fig. 2). While pqsA −339-induced luminescence was detected mostly in the PA14 ΔpqsR strain and silent in the absence of RhlR, pqsA −71 was active in the PA14 ΔrhlR strain but silent in the absence of PqsR. This demonstrates that induction of pqsA −339 is strictly dependent on RhlR, whereas induction of pqsA −71 is dependent on PqsR. Interestingly, activity of the pqsA −71 fusion doubled in the ΔrhlR mutant strain compared to that in the wild-type strain. Since it has previously been shown that RhlR is a transcriptional repressor of pqsR (14), the increased activity of pqsA −71 could be a consequence of an elevated level of PqsR in the PA14 ΔrhlR strain.

FIG 2.

Promoter activity of the two alternative transcriptional start sites of pqsA. The P. aeruginosa PA14 wild-type strain and ΔpqsR and ΔrhlR mutant strains containing plasmids p339 and p71 were cultured as described in Materials and Methods and assayed for lux expression. Data are presented in relative luminescence, and error bars represent one standard deviation of the mean value from three biological replicates (*, P ≤ 0.05; **, P ≤ 0.01).

The RhlR-dependent long 5′ UTR of pqsA blocks the translation initiation site.

We could demonstrate that RhlR is an activator of pqsA transcription but simultaneously represses PQS production. We hypothesized that base pairing in the translation initiation region of the longer transcript might result in gene silencing due to inaccessibility of the SD sequence to the 30S ribosomal subunit. To test this hypothesis, we predicted the secondary structures of the pqsA −71-induced 5′ leader sequence (Fig. 3A, short) and of a longer construct containing six additional nucleotides present in the pqsA −339 transcript (Fig. 3A, long). In contrast to the short mRNA, the long RhlR-induced transcript is predicted to sequester the SD domain.

FIG 3.

Folding dynamics and stability of secondary structures formed by the 5′ UTR of pqsA. (A) Predicted secondary structures in the 5′ leader sequence of the mRNA induced by PqsR (short; −14.10 kcal/mol), RhlR (long; −29.80 kcal/mol) and a derepressed RNA variant (derep; −14.10 kcal/mol). Secondary structures were generated using the CONTRAfold method (46). The color code represents base pairing probabilities in the structure ensemble, whereby high values (red) close to 1 are the most probable. Numbers indicate nucleotide positions relative to A of the AUG start codon. The AGGGAA Shine-Dalgarno sequence is encircled, and mutated nucleotides are emphasized by an asterisk. (B) Thermal unfolding behavior of 71-mer single-stranded DNA oligonucleotides equivalent to the RNA sequences shown in panel A monitored by UV absorption at 255 nm. Markers represent the raw data recalculated to the fraction of unfolded DNA. Continuous lines represent the data fits used to calculate the melting temperatures.

We next analyzed the thermal stability of 71-mer single-stranded DNA oligonucleotides equivalent to the sequence of the mRNAs (Fig. 3B). To equate the expansion at the 5′ end of the “long” RhlR-induced primer (which comprises the entire sequence predicted to participate in hairpin formation), six nucleotides (TTCTGT) were added to the 3′ end of the “short” primer. Both unfolding curves can be best explained by two separate unfolding events. Strikingly, the long primer underwent thermal unfolding at significantly higher temperatures (44°C and 46°C) and with greater cooperativity than the short oligonucleotide (26°C and 44°C). This strongly supports formation of a stable secondary structure in the RhlR-mediated transcript of pqsABCDE. This effect could be reversed by site-specific mutagenesis of five nucleotides upstream of pqsA −71, responsible for base pairing with the SD region, with random noncomplementary nucleotides (CGTTC replaced by AAGAA).

To examine a potential inhibitory role of the secondary structure in the 5′ UTR of pqsA, we generated an additional lux fusion construct (Fig. 1C). The plong construct is a translational fusion and harbors both promoter regions and the additional 70 nucleotides downstream of pqsA −71 containing the pqsA ribosomal binding site (RBS). As with the transcriptional reporters, exponentially grown cells were monitored for luminescence expression levels. We detected a significant decrease in luminescence (up to 3-fold) in the wild-type strain compared to that in the ΔrhlR mutant strain (Fig. 4). Although activity of pqsA −339 has been verified in the PA14 ΔpqsR strain (Fig. 2), we were unable to measure any luminescence signals in the translational fusion construct. Hence, the present data support our hypothesis that RhlR mediates posttranscriptional control on the pqs operon by producing an alternative transcript which leads to alterations in the folding pattern of the 5′ UTR in a way that inhibits efficient translation.

FIG 4.

Translational control of pqsA expression. P. aeruginosa strain PA14 WT and ΔpqsR and ΔrhlR mutant strains containing the plasmid plong were cultured as described in Materials and Methods and assayed for lux expression. Data are presented in relative luminescence, and error bars represent one standard deviation of the mean value from three biological replicates.

RhlR-induced transcription of the pqs operon abolishes efficient translation of the mRNA.

Our data imply that formation of a secondary structure within the RhlR-induced long pqsA transcript may inhibit pqsA translation. This effect should be reversed by site-specific mutagenesis liberating the SD sequence, as depicted in Fig. 3A. To test this, we analyzed PqsA production in vivo using His-tagged pqsA, whose transcription was under the control of the lac promoter but dependent on accessibility of the native SD sequence. Levels of PqsA-His were monitored during exponential growth in both P. aeruginosa PA14 and E. coli BL21 (Fig. 5A). Strikingly, we were unable to detect PqsA-His in cells harboring the native RhlR-induced sequence, while secondary structure destabilization in the derepressed construct drastically restored PqsA production in both bacterial strains, indicating that it is a general phenomenon not dependent on a Pseudomonas-specific factor. We also performed Western blot analysis by the use of an anti-PqsE antibody (32) to monitor PqsE production. No PqsE could be detected in the PA14 ΔpqsR strain, indicating that translation of not only pqsA but also the whole pqsABCDE operon is hindered in the RhlR-induced long pqsA transcript (data not shown).

FIG 5.

Implication of a secondary structure at the translation initiation site on PqsA production. (A) Analysis of His6-tagged PqsA produced under lac promoter control but dependent on the native translation initiation site in E. coli BL21 and PA14 WT. In vivo levels of PqsA-His were compared between the wild-type 5′ UTR comprising 80 nucleotides upstream of the start codon and a derepressed 5′ UTR causing liberation of the SD sequence. (B) Predicted folding patterns of the in vitro-transcribed pqsA constructs used for the toeprint assay: 1, PqsR-induced pqsA transcript (−22.70 kcal/mol; 2, RhlR-induced pqsA transcript (−46.40 kcal/mol); 3, a mutated RhlR-induced mRNA exhibiting a destabilized secondary structure (−29.80 kcal/mol). Folding dynamics of the mRNA constructs were predicted using the CONTRAfold method (46). Secondary structures are color coded according to base pairing probabilities in the structure ensemble, whereby high values (red) close to 1 are the most probable. The AGGGAA Shine-Dalgarno sequence is encircled, and the AUG start codon is indicated by a black line. Site-specific mutated nucleotides are emphasized by an asterisk. (C) Primer extension inhibition assay of the pqsA mRNAs drawn in panel B, including the 5′ UTR and the first 60 nucleotides of the pqsA coding sequence. Addition (+) or absence (−) of E. coli 30S ribosomal subunits is indicated. The AUG start codon and the SD sequence are marked on the right, and full-length products, structure 1, structure 2, and the toeprint are indicated on the left.

To further assess the exact molecular role of the hairpin loop, we performed primer extension inhibition (toeprinting) experiments to examine binding of the 30S ribosomal subunit to an mRNA molecule. In this method, ribosome-mRNA complex formation inhibits the primer extension reaction, resulting in a terminated product (toeprint) around position +17 with respect to the translational start site. We used the assay to investigate the ability of the ribosome to recognize the SD sequence and to form a translation initiation complex upstream of pqsA. Efficiency of ribosome binding was compared between the following in vitro transcribed mRNAs: PqsR-induced pqsA transcript (1), RhlR-induced pqsA transcript (2), and a mutated RhlR-induced mRNA exhibiting a destabilized secondary structure (3) (Fig. 5B). Consistent with impaired translation from the long transcript, a toeprint at position +17 was detectable in the PqsR-induced pqsA transcript but absent in the RhlR-induced pqsA transcript (Fig. 5C). Binding of the ribosome was partially restored upon destabilization of the secondary structure. This clearly demonstrates that the secondary structure of the RhlR-induced pqsA transcript, which comprises the SD sequence, prevents formation of the preinitiation complex. Additional reverse transcription products are indicative of additional double-stranded regions able to terminate reverse transcription (structures 1 and 2 in Fig. 5C). Together, these data demonstrate that RhlR mediates posttranscriptional repression of PQS synthesis by initiating a long pqsA transcript, in which the SD sequence is sequestered in a secondary structure.

DISCUSSION

Pathogens have developed mechanisms to persist and survive in various environments, including the human host. In P. aeruginosa, the production of the interbacterial signal molecule PQS is critical for survival under deteriorating conditions. PQS itself is a multifunctional molecule acting as a QS signal molecule (33), it has an iron-chelating activity, and it is essential for biofilm formation (6, 10). Furthermore, PQS plays a pivotal role in tuning cellular physiology and has been implicated in cell death under stressful conditions (6, 34). Recently, PQS was suggested to act as both a prooxidant and an inducer of an anti-oxidative stress response (9), emphasizing the importance of this molecule in environmental adaptation of P. aeruginosa. Therefore, production of PQS needs to be strictly controlled.

The complex regulatory circuit of PQS synthesis involves the LysR-type transcriptional regulator protein PqsR, which recognizes and binds the signal molecule PQS and subsequently enhances transcription of the 4-quinolone biosynthetic operon pqsABCD, thus forming a positive autoregulatory loop (14). Expression of pqsR in turn is controlled by the las and rhl QS systems, interconnecting all three QS systems of P. aeruginosa. However, the transcriptional regulator of the rhl system, RhlR, was also shown to directly exert control on PQS biosynthesis by binding to the pqsABCDE promoter region (16). The present study provides molecular insights into this mechanism.

The discovery of an alternative transcriptional start site (pqsA −339) of the pqsABCDE operon suggested an additional level of direct transcriptional regulation of the PQS system in P. aeruginosa (17). The fact that an RhlR-binding box is located just upstream of this transcriptional start site led us to hypothesize that RhlR might be the transcriptional regulator of this promoter. Indeed, in the present study, we demonstrate that RhlR is actually a transcriptional activator of pqsABCDE and initiates transcription from pqsA −339. Further analyses revealed that RhlR binding to the pqsA promoter region induces the transcription of a pqsABCDE mRNA with an extended 5′ UTR that exhibits a stable secondary structure, which sequesters the SD sequence and inhibits translation initiation of pqsA. As a consequence, a PA14 ΔpqsR mutant strain displayed transcriptional activation of pqsA −339 but failed to efficiently induce production of PqsA (Fig. 6). By inducing a longer pqsABCDE transcript, RhlR prevents translation of the pqsA gene, whose product is responsible for priming anthranilate for entry into the PQS biosynthetic pathway and whose deletion is known to impede PQS production (35, 36). Translation of not only pqsA but also of the whole operon was abolished. We could not detect PqsE, encoded by the last gene on the operon, in a pqsR mutant, suggesting that the pqsA −339-induced transcript might be unstable as a consequence of ribosomes failing to load on its 5′ end. Hence, depending on the presence of sufficient intra- and extracellular concentrations of the signal molecules C₄-HSL and/or PQS, the PQS signaling pathway can either be induced by PqsR or inhibited by RhlR. The control of PQS production by two transcriptional regulators of different QS systems thus allows P. aeruginosa to fine-tune PQS signaling in response to cell density and environmental stimuli.

FIG 6.

Model of transcriptional and translational control of the pqs operon. P. aeruginosa releases the signal molecules N-butanoyl-l-homoserine lactone (C₄-HSL) and the Pseudomonas quinolone signal (PQS) in a cell-dependent manner. Upon binding to their cognate ligand, the transcriptional regulators RhlR-C₄-HSL and PqsR-PQS induce transcription of the pqs operon from the pqsA −339 and pqsA −71 transcription start sites, respectively. In the pqsA −339-induced mRNA, the formation of a hairpin at the translation initiation site of pqsA blocks access of the ribosome to the SD sequence. As a consequence, production of the anthranilate-coenzyme A ligase PqsA is hindered, rendering P. aeruginosa unable to generate HHQ/PQS.

Many bacterial mRNAs have been described to harbor structured elements in their 5′ leader sequence that control translation (37). Here, to our knowledge, we report for the first time on the modulation of bacterial protein levels via the production of two mRNA isoforms which form variable secondary structures at the translation initiation site and thus are translated at variable efficiency. We found that two transcriptional factors induce alternative transcripts that have profound effects on cell signaling-dependent phenotypes. Recently, a similar regulatory mechanism has been described for the crl gene in Escherichia coli. Overlapping promoters produce similarly sized mRNAs, but only one harbors a ribosomal binding site (38).

Although not well described in bacteria, different mRNA isoforms are known to play an important role in the regulation of translation in eukaryotes. In mammalians, up to 10 to 18% of all genes use multiple promoters (39). For instance, the axin2 gene, which is involved in early postnatal development and tumor suppression, has three promoters whose expression is strictly tissue specific (40, 41). The 5′ UTR of each mRNA isoform affects translation efficiency and mRNA stability due to the formation of different secondary structures. Another example in eukaryotes is the brca1 tumor suppressor gene, where transcription is induced from two separate promoters. Here, as observed for pqsA mRNA translation, a longer 5′ UTR of the brca1 mRNA is translated at lower efficiency due to formation of a stable secondary structure (42). Taken together, this study sheds light on the posttranscriptional regulation of PQS synthesis and illustrates that the promoter region of pqsA represents a major site of transcriptional and translational control. Differential secondary structures in mRNA isoforms that directly impact on translational efficiency of genes—a common mechanism of regulation in eukaryotes—might represent an underestimated mechanism of posttranscriptional control in bacteria and might play a more important role in bacterial adaptation than previously anticipated.