Distal and proximal promoters co-regulate pqsR expression in Pseudomonas aeruginosa

Summary

The ubiquitous bacterium Pseudomonas aeruginosa is an opportunistic pathogen that can cause serious infections in immunocompromised individuals. P. aeruginosa virulence is controlled partly by intercellular communication, and the transcription factor PqsR is a necessary component in the P. aeruginosa cell-to-cell signaling network. PqsR acts as the receptor for the Pseudomonas quinolone signal, and it controls the production of 2-alkyl-4-quinolone molecules which are important for pathogenicity. Previous studies showed that the expression of pqsR is positively controlled by the quorum-sensing regulator LasR, but it was unclear how LasR is able to induce pqsR transcription. In this report we further investigated the control of pqsR, and discovered two separate promoter sites that contribute to pqsR expression. LasR-mediated activation occurs at the distal promoter site, but this activation can be antagonized by the regulator CysB. The proximal promoter site also contributes to pqsR transcription, but initiation at this site is inhibited by a negative regulatory sequence element, and potentially by the H-NS family members MvaT and MvaU. We propose a model where positive and negative regulatory influences at each promoter site are integrated to modify pqsR expression. This arrangement could allow for information from both environmental signals and cell-to-cell communication to influence PqsR levels.

Graphical Abstract

Introduction

Pseudomonas aeruginosa is a gram-negative bacterium that inhabits soil and aquatic environments, and it can act as an opportunistic pathogen with the ability to infect plants, insects, and mammals (Pirnay et al., 2005, Rahme et al., 1995, Chugani et al., 2001, Lau et al., 2004). This organism can also infect humans, causing approximately 8% of healthcare-associated infections in the United States and serious chronic infections in individuals with cystic fibrosis (Sievert et al., 2013, Döring et al., 2012). These infections are generally difficult to treat due to the frequency of antibiotic-resistant phenotypes (Döring & Pier, 2008, Kallen & Srinivasan, 2010). P. aeruginosa utilizes a wide range of virulence determinants when causing disease, and the development of treatments to inhibit virulence factor production is thought to be a promising strategy to fight these infections (Van Delden, 2004, Fothergill et al., 2012). However, the control of virulence factor production is complex, with the synthesis of many virulence determinants being directed both by environmental cues and by cell-to-cell signaling. P. aeruginosa possesses three main signaling systems that are important for pathogenicity. Two of these, the las and rhl signaling systems, function through the signaling molecules N-(3-oxododecanoyl) homoserine lactone (3-oxo-C₁₂-HSL) and N-butyryl homoserine lactone (C₄-HSL), respectively, and are similar to homoserine lactone based signals found in other gram-negative bacteria (Pearson et al., 1994, Pearson et al., 1995). These molecules act as respective ligands for the transcription factors LasR and RhlR in a process of intercellular communication known as quorum sensing, which is thought to be a mechanism to coordinate gene expression with cell density (Pearson et al., 1994, Ochsner & Reiser, 1995). The third P. aeruginosa cell-to-cell signaling system functions primarily through the quinolone compound 2-heptyl-3-hydroxy-4-quinolone, referred to as the Pseudomonas quinolone signal (PQS) (Pesci et al., 1999). This signal serves as a coinducer for the transcriptional regulator PqsR (also known as MvfR), which positively regulates quinolone production to create an auto-regulatory loop (McGrath et al., 2004, Wade et al., 2005, Xiao et al., 2006a). Together these signaling systems form a global regulatory network and are estimated to control the expression of up to 12% of the genes present in the P. aeruginosa genome (Déziel et al., 2005, Wagner et al., 2003, Schuster et al., 2003).

The primary role for PqsR appears to be to induce the expression of the pqsABCDE operon, which encodes proteins necessary for the biosynthesis of PQS and for signal transduction (Wade et al., 2005, Xiao et al., 2006b). The majority of the genes controlled by PQS signaling are regulated through the activity of PqsE, which plays a non-essential role in quinolone synthesis, but appears to enhance the activity of the rhl signaling system (Farrow et al., 2008, Rampioni et al., 2010, Hazan et al., 2010, Drees & Fetzner, 2015). In addition to PQS, the enzymes encoded by pqsABCDE are involved in the production of at least 55 other quinolone compounds (Déziel et al., 2004, Lépine et al., 2004, Drees & Fetzner, 2015). The function of many of these compounds is not known, but studies have suggested that PQS and other quinolones participate in a wide variety of physiological functions, including iron metabolism, membrane vesicle formation, inter-species interactions, biofilm formation, and host immune modulation (Tashiro et al., 2013, Kim et al., 2010, Bredenbruch et al., 2006, Diggle et al., 2007, Mashburn & Whiteley, 2005, Mashburn-Warren et al., 2008, Recinos et al., 2012). Along with these functions, it has been shown that PqsR and quinolone production are crucial for virulence in animal infection models, and PQS has been isolated from the sputum of cystic fibrosis patients infected with P. aeruginosa (Déziel et al., 2005, Collier et al., 2002). These findings demonstrate the importance of quinolones for P. aeruginosa physiology and pathogenicity. The production of quinolones is also directly influenced by the las and rhl signaling systems. LasR and 3-oxo-C₁₂-HSL positively regulate the expression of both pqsR and pqsH, the latter of which encodes a protein that performs the final step of PQS synthesis, whereas RhlR and C₄-HSL negatively regulate PQS by influencing expression from the pqsA promoter (Pesci et al., 1999, Gilbert et al., 2009, Schuster et al., 2003, McGrath et al., 2004, Brouwer et al., 2014). In addition, there is evidence that other signals modulate quinolone production. PQS is still generated in a lasR mutant strain, although its production is delayed, showing that quinolone biosynthesis can be induced in the absence of LasR (Diggle et al., 2003). Furthermore, unlike the other P. aeruginosa signaling molecules, PQS production is delayed until cells reach the stationary phase of growth (McKnight et al., 2000), suggesting that specific conditions are needed to induce quinolone synthesis.

We became interested in investigating the control of pqsR expression, since a pqsR mutant strain does not produce detectable levels of PQS (Wade et al., 2005, Déziel et al., 2004), and because pqsR appeared to have a complex promoter region, making it a likely point for the integration of multiple signals that influence quinolone production (see Fig. 1C for a model). The pqsR coding sequence is separated from the divergently transcribed nadA gene by 747 bp, and two transcriptional start sites were identified at −190 and −278 bp relative to the pqsR translational start site (Wade et al., 2005). LasR positively regulates pqsR transcription from a binding site that is distant from these start sites (Farrow et al., 2015, Xiao et al., 2006b). Additionally, upstream from pqsR is a small open reading frame (uof), and the translation of uof appears to be linked to the translation of pqsR, providing a point for post-transcriptional control of PqsR expression (Sonnleitner et al., 2011). While these elements provided some clues about pqsR regulation, it was uncertain how these components functioned together to control pqsR. In these studies we focused on investigating the process of pqsR transcriptional initiation. We found that pqsR transcription is also controlled from a distal promoter site, and show that specific interactions at both the distal and proximal promoters are important for regulating pqsR expression.

Fig. 1. LasR induces transcription at a promoter that is distant from the pqsR translational start site.

A. Primer extension analyses of transcripts from the pqsR-nadA intergenic region. Lanes labeled A, C, G, and T contain DNA sequencing ladders that terminate at the corresponding nucleotide positions. The result in lane 1 was generated using RNA isolated from P. aeruginosa strain PAO1, and the result in lane 2 was generated using RNA from strain PAO1 carrying a pqsR’-lacZ fusion on pMWC1003.

B. DNA sequence of the pqsR promoter region in the vicinity of the transcriptional start sites identified in panel (A) (transcriptional start sites shown in bold). The LasR binding site is boxed, and sequences with similarity to σ⁷⁰-type promoter −10 elements are underlined. Numbering indicates nucleotide positions relative to the pqsR translational start site.

C. A diagram of pqsR’-lacZ fusions and results from β-gal assays performed using P. aeruginosa strain PAO1 harboring the indicated lacZ fusion integrated on the bacterial chromosome. Bent arrows on the diagram indicate the relative positions of transcriptional start sites, and numbering is in base pairs relative to the pqsR translational start site. As a control, P. aeruginosa strain PAO1 carrying a promoter-less lacZ construct integrated on the chromosome produced 4 ± 2 Miller units of activity. Data are presented in Miller units as the mean ± SD of results from duplicate assays from at least three separate experiments.

Results

LasR activates pqsR transcription at a distal promoter site

Previously we found that LasR only interacts with a single binding site in the pqsR promoter region centered at 278 bp upstream from the pqsR transcriptional start site 1 (TS1), and that this binding site was important for the expression of pqsR (Farrow et al., 2015). However, the mechanism used by LasR to activate transcription from this distal binding site was unclear. Another study that analyzed the P. aeruginosa transcriptome by high-throughput sequencing analysis identified a third transcriptional start site in the pqsR promoter region upstream from the pqsR TS1 (Dötsch et al., 2012). This potential start site was much closer to the LasR binding site, and we hypothesized that there was an additional promoter in the pqsR-nadA intergenic region that could affect pqsR transcription. To confirm the position of this putative start site we performed primer extension analyses on RNA from P. aeruginosa strain PAO1 using primers that would anneal to the pqsR coding strand upstream from the pqsR TS1. These experiments confirmed the presence of a third transcriptional start site (TS3) at −482 bp relative to the pqsR translational start site, and additionally revealed a fourth start site at −463 bp, 19 bp downstream from TS3 (Fig. 1A). As expected, these sites were more evident when we analyzed RNA from strain PAO1 carrying a pqsR’-lacZ fusion on plasmid pMWC1003 to enrich for RNA from the pqsR promoter (Fig. 1A). Inspection of the DNA sequences near these sites showed that there were appropriately spaced sequences upstream from each site with similarity to the −10 element of a σ⁷⁰-type promoter (Fig. 1B). We could not identify −35 elements upstream from either start site, but the LasR binding site overlapped with the −35 region from TS3. This arrangement suggested that transcriptional initiation at these sites was directly controlled by LasR.

To further examine the regulation of this promoter site and its contribution to pqsR transcription, we constructed a series of lacZ fusions that linked various sections of the pqsR promoter region to lacZ (Fig. 1C). These pqsR’-lacZ fusions were integrated onto the P. aeruginosa strain PAO1 chromosome at the attTn7 site, and cultures of the resulting strains were assayed for β-gal activity. The strain carrying the entire pqsR-nadA intergenic region fused to lacZ produced 324 ± 36 units of β-gal activity (Fig. 1C). Strains carrying lacZ fusions that only contained either the pqsR TS2 or both the pqsR TS1 and TS2, but not the pqsR TS3 or TS4, only produced approximately 20% of the β-gal activity produced by the full-length pqsR’-lacZ fusion (Fig. 1C). These results indicated that sequences upstream from −446 bp relative to the pqsR translational start site are important for the full induction of pqsR transcription. Surprisingly, a lacZ fusion that contained only the pqsR TS3 and TS4 sites and the distal promoter (from −458 to −776 bp relative to the pqsR translational start site) produced approximately 10-fold more β-gal activity than the full-length pqsR’-lacZ fusion (Fig. 1C). This finding suggested that transcriptional initiation at the TS3/4 promoter site was very efficient, but it was unclear why we observed such a large difference in activity when lacZ was fused within the pqsR coding region. Finally, we examined the role of LasR in the activation of both the full-length pqsR’-lacZ fusion and the pqsR TS3/4-lacZ fusion by altering base pairs in the LasR binding site on each to prevent LasR from interacting with the pqsR promoter region (Farrow et al., 2015). These mutations caused the β-gal activity produced by each of the lacZ fusions to be greatly reduced (Fig. 1C). Taken together these experiments showed that LasR activates transcription at the TS3/4 promoter site, and that this promoter is important for the full level of pqsR expression.

Some transcripts from the distal pqsR promoter terminate prematurely

These findings provided a more clear explanation of how LasR activates transcription at the pqsR promoter, but did not explain the high levels of β-gal activity we observed from the pqsR TS3/4-lacZ fusion. Studies that attempted to identify small RNAs (sRNAs) in P. aeruginosa suggested that additional small transcripts may originate from the pqsR promoter region (Gómez-Lozano et al., 2012, Wurtzel et al., 2012), providing a potential alternative function for the TS3 promoter. To identify any additional transcripts generated from the pqsR promoter region we performed Northern blot analyses using single-stranded DNA oligonucleotides as probes (for the relative positions of each probe see Fig. 2A). As expected, blots using a probe that anneals to the nadA coding strand downstream from the nadA transcriptional start site (probe 4, blot not shown) and a probe that anneals to the pqsR coding strand downstream from the pqsR TS2 (probe 1) each detected a transcript that was larger than the largest molecular weight standard at 766 bp (Fig. 2B), which should correspond to the coding mRNA for each gene. We also identified two sRNAs of approximately 270 and 290 nucleotides in length using probes 3, 7, 2, and 6 that annealed to the pqsR coding strand (Fig. 2B, blots for probes 7 and 2 not shown). Based on the estimated sizes of the RNAs and the region covered by the probes, it appeared that these transcripts originated at the pqsR TS3 and TS4, and terminated between the pqsR TS1 and TS2. The location and orientation of these sRNAs is similar to what was predicted by the prior studies (Wurtzel et al., 2012, Gómez-Lozano et al., 2012), although the sizes and positions of these transcripts do not exactly match the previous predictions which were generated through RNA sequencing-based transcriptomic analyses. These results confirmed that sRNAs are generated from the pqsR promoter region and clearly defined their locations.

Fig. 2. Mapping of transcripts that originate from the pqsR TS3/4 transcriptional start sites.

A. A diagram of the pqsR-nadA intergenic region and the relative positions of oligonucleotide probes and primers used for Northern blot and RT-PCR analyses. Probes used for Northern blots are shown above the model, and primers for RT-PCR are shown below. Northern blot probes that hybridized to small transcripts in the pqsR promoter region are marked with an asterisk.

B. Representative results from Northern blot analyses performed using RNA isolated from P. aeruginosa strain PAO1 after 3 h of growth in LB medium. The number of the probe used for each analysis is indicated above the image of each blot. The blots shown were performed using single-stranded oligonucleotide probes that hybridized to the pqsR coding strand, and asterisks indicate the position of the small transcripts identified by these analyses. Unneeded lanes were removed from the image of each blot to show the position of transcripts directly next to the corresponding molecular weight markers.

C. Results of RT-PCR analyses performed using RNA isolated from P. aeruginosa strain PAO1 and oligonucleotide primers shown in (A). Equivalent amounts of RT-PCR reactions were analyzed by agarose gel electrophoresis. The primers used for each set of reactions are indicated below each image. Lane 1 corresponds to control reactions performed without reverse-transcriptase present, lane 2 corresponds to experimental reactions with reverse-transcriptase, and lane 3 corresponds to control reactions performed with chromosomal DNA as the template instead of RNA. The data presented are representative of two separate experiments performed using RNA isolated from two different cultures.

Since our data indicated that some of the transcripts which initiated at the TS3/4 promoter terminated prior to reaching pqsR, we wondered if there were also transcripts from this promoter site that carried the pqsR coding sequence. To test this we performed RT-PCR using the primers shown in Fig. 2A. We were able to amplify fragments from cDNA corresponding to either the pqsR coding sequence or the upstream small RNAs (primer pairs F3/R1 and F2/R2, respectively; Fig. 2C). We also detected a transcript using primers that annealed upstream from the pqsR TS1 and to the pqsR coding sequence (primers F2/R1, Fig. 2C), showing that pqsR can be encoded on a mRNA that initiates upstream from TS1. We additionally performed reactions with a primer that corresponds to the nadA coding sequence and a primer downstream from the pqsR TS1 as a negative control, which did not produce a DNA fragment (primer pair F4/R2, Fig. 2C). Together with our previous experiments, these results implied that LasR induces transcription at the pqsR TS3/4 promoter, generating small transcripts that terminate before pqsR and longer mRNAs that carry the pqsR coding region.

Confirmation of the pqsR TS1 as a transcriptional start site

The discovery of the distal promoter site resolved many of our questions regarding LasR-mediated control of pqsR expression. With this in mind, we now became curious about the role of the promoter at the pqsR TS1, since the TS3/4 promoter site appeared to have the most significant role in controlling pqsR transcription. We previously identified the TS1 by primer extension analysis, and other studies that analyzed the P. aeruginosa transcriptome identified this site as well (Wade et al., 2005, Wurtzel et al., 2012, Dötsch et al., 2012). However, since we now had established that some pqsR-coding RNAs began upstream from TS1, it was possible that the TS1 represented the cleavage site of a longer RNA instead of a transcriptional start site. To examine this possibility we treated P. aeruginosa RNA with 5’-phosphate-dependent exonuclease (TEX), which allowed us to discriminate between between RNA processing sites versus initiation sites. This is due to the fact that RNAs with a 5’-monophosphate, such as those that are produced by cleavage of the RNA molecule, are degraded by this enzyme, whereas those with a 5’-triphosphate, which is a feature of the initiating nucleotide, are not subject to degradation. After this treatment we mapped the pqsR TS1 by S1 nuclease protection assays as an alternative to techniques that rely on reverse transcriptase. As a control for this experiment we also examined the 5’-end of the mature tryptophanyl-tRNA (tRNA^Trp), which has a defined cleavage site (Pettersson & Kirsebom, 2008). The +1 site of the tRNA^Trp was clearly evident in S1 protection assays using untreated RNA, but we did not detect any of this RNA after treatment with TEX (Fig. 3B), confirming that this treatment effectively degraded processed RNAs. When we examined the pqsR TS1 using untreated RNA, we observed the 5’ end of a transcript at the expected site, and another potential start site 11 bp downstream from TS1 at −267 relative to the pqsR translational start site (Fig. 3A). In contrast to tRNA^Trp, when we mapped the pqsR TS1 using TEX-treated RNA, both of the potential start sites were detected at a similar level as when we used untreated RNA (Fig. 3A). These results confirmed the position of the pqsR TS1 and showed that it is a genuine transcriptional start site.

Fig. 3.

Confirmation of the pqsR TS1 as a transcriptional start site. Results of S1 nuclease protection assays performed using RNA isolated from strain PAO1 carrying a pqsR’-lacZ fusion on pMWC1003. RNA was either untreated or treated with Terminator 5’-phosphate-dependent exonuclease (TEX) before being used for nuclease protection assays. Lanes labeled A, C, G, and T contain DNA sequencing ladders that terminate at the corresponding nucleotide positions. Arrowheads indicate the positions of (A) the pqsR TS1 or (B) the 5’-end of the processed tryptophanyl-tRNA, which served as a positive control for TEX activity. The data presented are representative of two separate experiments, and similar results were obtained using RNA isolated from strain PAO1.

A negative regulatory element is encoded in the vicinity of the pqsR TS1

Once we verified the pqsR TS1 we wanted to assess the contribution of the TS1 promoter to pqsR expression. To do this we decided to alter sequence elements upstream from TS1 to inactivate this promoter while leaving the distal pqsR TS3/4 promoter intact. We previously noted putative −10 and −35 elements for a σ⁷⁰-type promoter upstream from the TS1 (Wade et al., 2005). However, a closer inspection of this region showed that the potential −10 element we identified is actually centered at −14 relative to TS1, and that more appropriately spaced sequences upstream from TS1 also have some similarity to the σ⁷⁰ consensus binding sequence (Fig. 4A). To resolve this issue we constructed pqsR’-lacZ fusions with base substitutions in either of the potential −10 elements (shown in Fig. 4A). The mutations in the originally identified −10 and −35 elements caused little change in the amount of β-gal activity produced by the pqsR’-lacZ fusion (Fig. 4B). In contrast, changing the alternative −10 site caused a 10-fold increase in β-gal activity (Fig. 4B). This was unexpected, since we assumed that the disruption of core promoter elements would cause a decrease in transcription. To further ensure that the pqsR TS1 promoter was disabled, we constructed another pqsR’-lacZ fusion with the pqsR TS1 core promoter region deleted (from −5 to −37 relative to TS1, Fig. 4A). The β-gal activity produced by this pqsR’-lacZ fusion (Δ core) was also approximately 10-fold higher than the activity produced by the wild-type pqsR’-lacZ fusion (Fig. 4B). These results indicated that the sequences immediately upstream from TS1 limit the expression of pqsR.

Fig. 4. Alteration of sequences close to the pqsR TS1 cause increased transcription of a pqsR’-lacZ fusion.

A. A diagram of DNA sequences in the vicinity of the pqsR TS1 showing mutations that were introduced on pqsR’-lacZ fusions used for assays shown in (B). The pqsR TS1 is shown in bold. Numbering indicates nucleotide positions relative to the pqsR translational start site. Sequences with similarity to σ⁷⁰-type promoter −10 and −35 elements are underlined. Nucleotide substitutions that were made to generate the indicated pqsR’-lacZ fusions are shown below each sequence. The DNA sequence indicated by the bracket was removed to generate the pqsR’-lacZ fusion with the TS1 core promoter region deleted (Δ core).

B. Results of β-gal assays using P. aeruginosa strain PAO1 carrying a pqsR’-lacZ fusion with the indicated mutation integrated on the bacterial chromosome. Data are presented in Miller units as the mean ± SD of results from duplicate assays from at least three separate experiments.

While trying to understand what these results implied about the regulation of pqsR, we noted that in our first set of experiments we observed small transcripts from the upstream TS3/4 promoter which terminate between TS1 and TS2 (Fig. 2). We considered the possibility that perhaps while trying to disrupt the TS1 promoter, we had instead altered sequences that were part of a transcriptional termination signal. If this were the case, then the increased β-gal activity produced by the altered pqsR’-lacZ fusions could be due to increased transcriptional read-through from the distal promoter, and this would also be consistent with the high level of β-gal activity produced by the pqsR’ TS3/4-lacZ fusion (Fig. 1). To test this hypothesis we decided to transfer the alternative −10 mutation at the TS1 promoter (alt −10 mut, Fig. 4A) onto the P. aeruginosa chromosome, and then examine the transcripts that are generated with this mutation present. The resulting pqsR promoter mutant strain produced approximately 2.5-fold more PQS than the wild-type strain (31.1 ± 3.1 ng µl⁻¹ PQS for strain PJF-pqsR-promut versus 12.7 ± 3.0 ng µl⁻¹ for strain PAO1), suggesting that the introduction of this mutation on the chromosome caused an increase in PqsR expression as we expected. To assess the production of transcripts from the pqsR promoter region we performed qRT-PCR using primers that would amplify probes from cDNAs corresponding to either transcripts upstream from TS1 (region 2, Fig. 5A), RNAs that begin upstream from TS1 and extend beyond TS2 (region 3, Fig. 5A), or transcripts from within the pqsR coding region (region 1, Fig. 5A). We additionally examined transcript levels in a lasR mutant strain as a control, since we had established that transcription from the pqsR TS3/4 promoter was controlled by LasR. In agreement with the data generated using pqsR’-lacZ fusions (Fig. 1 and Fig. 4), the relative expression of the pqsR coding region was decreased to only 0.58 ± 0.07-fold of the wild-type level in the lasR mutant, whereas in the promoter mutant strain there was a 10.8 ± 1.94-fold increase in pqsR expression (Fig. 5B). When we examined the relative expression of transcripts upstream from TS1 (region 2), which should include all transcripts that initiate at the distal promoter, we found that these were decreased 14-fold (0.07 ± 0.01-fold change) in the lasR mutant compared to the wild-type, but expression of these transcripts in the promoter mutant strain was similar to the wild-type strain (Fig. 5C). These results showed that the TS1 promoter mutation did not affect transcription from the TS3/4 promoter, and confirmed that activation of the distal pqsR promoter is strongly dependent on LasR. Similarly, we observed that the production of transcripts detected from region 3, which should only include RNAs that initiate at the distal promoter and extend beyond the early termination point, was greatly decreased in the lasR mutant strain (0.11 ± 0.08-fold compared to wild-type, Fig. 5D). However, in the promoter mutant strain we saw only a small increase in the expression of these RNAs (Fig. 5D). If the mutation near the TS1 had disrupted a termination signal for RNAs that originated from the TS3/4 promoter, then we would have expected to find a large increase in the number of transcripts that initiated upstream from TS1 and extended past TS2 in the promoter mutant strain, similar to the large increase in the expression of the pqsR coding region (region 1) observed in this strain (Fig. 5B). Instead, these results implied that enhanced pqsR expression in the promoter mutant strain was not the result of transcriptional read-through from the distal promoter. The alternative explanation for the increased pqsR expression in the promoter mutant strain was that the TS1 promoter mutation caused increased transcription from a proximal promoter site. To examine this possibility we directly analyzed transcripts that initiated at the TS1 promoter by S1 nuclease protection assays using an equal amount of RNA from the wild-type and pqsR promoter mutant strains. These transcripts appeared to be more abundant in the promoter mutant strain (Fig. 5E), implying that the alternative −10 mutation at TS1 caused increased transcriptional activation at this site. Taken together with the results from our mutational analysis of the sequences near the pqsR TS1, these data show that there is a negative regulatory element present in these sequences which inhibits pqsR expression by limiting transcriptional initiation at the TS1 promoter, and not by affecting the termination of transcripts from the TS3/4 promoter.

Fig. 5. The mutation of a sequence element upstream from the pqsR TS1 causes increased transcriptional initiation at this site.

A. Diagram of the pqsR-nadA intergenic region showing the relative positons of DNA fragments amplified and quantified by qRT-PCR. The star shows the relative position of the alternative −10 mutation in the pqsR promoter.

B., C., and D. Relative expression was analyzed by qRT-PCR using total RNA isolated from either the wild-type strain (PAO1), a ΔlasR strain (PAO-R1), or a pqsR promoter mutant strain (PJF-pqsR-promut) carrying the alt −10 mutation (see Fig. 4A) on the bacterial chromosome. Results are shown as the mean fold change ± SD, relative to the expression in strain PAO1, from at least three separate experiments.

E. Equivalent amounts of total RNA from (Lane 1) the pqsR promoter mutant and (Lane 2) the wild-type strains was used for S1 nuclease protection assays. The results presented are representative of two separate experiments performed using RNA isolated from two different cultures.

Factors that control pqsR expression have distinct roles at each promoter site

All of these experiments provided new insight into the overall structure of the pqsR promoter region, but we still wished to understand more about the interactions that occur at each promoter site and how they function together to influence pqsR expression. We were also curious about the operation of the TS1 negative regulatory element, and we speculated that this region may act as a binding site for a repressor. Several factors have been identified that bind the pqsR promoter region and can act as repressors. One is the YebC-family protein PmpR, although its precise binding site within the pqsR promoter was not identified (Liang et al., 2008). Another is the transcriptional regulator CysB, which represses pqsR expression by competing with LasR for binding to the pqsR promoter at the las box (Farrow et al., 2015). The functionally redundant H-NS family members MvaT and MvaU can also interact with the pqsR-nadA intergenic region, and can act to silence transcription (Castang et al., 2008, Farrow et al., 2015). However their potential role in controlling pqsR was not fully investigated due to the fact that the presence of at least one of these proteins is required to maintain cell viability. To examine the role of these proteins we generated plasmids that allowed the expression of each to be controlled by the inducible P_BAD promoter. We then transferred these plasmids into P. aeruginosa strains which harbored on the chromosome either the full wild-type pqsR’-lacZ, only the distal pqsR promoter fused to lacZ, only the proximal pqsR’-lacZ, or the full pqsR’-lacZ fusion with the alternative −10 mutation at TS1. This approach allowed us to examine the regulatory effect of each protein at each of the pqsR promoter sites, and to determine if these effects required the TS1 negative regulatory element.

We over-expressed each protein at a level that did not affect cell growth, and then compared the level of β-gal activity produced to the activity generated by a strain carrying the empty expression vector. When we expressed PmpR, we saw only a small decrease in the β-gal activity produced by any of the pqsR’-lacZ fusions (Fig. 6). These results showed that either PmpR does not significantly affect pqsR transcription under the conditions we tested, or that this experimental approach was not able to detect regulation by PmpR. We should note that PmpR-mediated repression was previously shown using a pmpR mutant strain, and not through over-expression (Liang et al., 2008). However, over-expression of CysB did cause changes in pqsR’-lacZ activity. CysB caused the activity of the full-length pqsR’-lacZ to decrease by approximately 46% (Fig. 6A), and caused the activity of the TS3/4 pqsR’-lacZ to drop to only 5% of the activity produced without over-expression (Fig. 6B). These data concur with the established role of CysB in blocking LasR from activating pqsR. In contrast, we observed that expression of CysB caused an increase in the β-gal activity produced when only the proximal pqsR promoter was fused to lacZ (Fig. 6C). Interestingly, we have previously shown that purified CysB interacted with two sites in the pqsR promoter region: it bound to one fragment that contained the LasR binding site, and also with lower affinity to a fragment containing the TS1 and TS2 (Farrow et al., 2015). Together these findings suggest that CysB can positively affect pqsR transcription at a proximal promoter site, although it is uncertain whether this form of regulation occurs under normal physiological conditions. We additionally saw that CysB expression had a less pronounced negative impact on the β-gal activity produced by a pqsR’-lacZ fusion with the alternative −10 mutation at TS1 (Fig. 6D). This result was in keeping with the notion that inhibitory effects at the distal promoter site would have less impact when a proximal promoter is activated. The over-expression of MvaT and MvaU also altered pqsR’-lacZ activity, but with a different pattern of effects. Both MvaT and MvaU expression decreased the β-gal activity produced from the full pqsR’-lacZ by approximately 36% and 23%, respectively (Fig. 6A), but neither affected the activity of the TS3/4 pqsR’-lacZ fusion (Fig. 4B). Instead, MvaT caused an approximately 73% decrease, and MvaU a 47% decrease in the activity produced by the proximal pqsR’-lacZ fusion (Fig. 4C), which was greater than the effects seen on the full pqsR’-lacZ. Similarly, the activity of the pqsR’-lacZ with the TS1 alternative −10 mutation was lowered by 70% or 39% in response to either MvaT or MvaU expression (Fig. 6D). These experiments indicated that MvaT and MvaU can repress pqsR transcription from a proximal promoter site. They also showed that this inhibition did not depend on the negative regulatory element near the pqsR TS1. Overall, these findings demonstrated that each of the pqsR promoter sites is controlled differentially in response to specific regulatory factors.

Fig. 6.

Differential regulation of the pqsR distal and proximal promoters in response to the over-expression of potential regulators. β-gal assays were performed using derivatives of strain PAO1 that harbored either (A) the wild-type pqsR’-lacZ fusion, (B) the pqsR TS3/4 promoter fused to lacZ, (C) the TS1 and TS2 promoters (without distal promoter sequences) fused to lacZ, (D) or the full-length pqsR’-lacZ fusion with the alternate −10 mutation in the TS1 promoter. Regulatory proteins PmpR, CysB, MvaT, and MvaU were expressed in each strain from a plasmid construct derived from pHERD20T. Each strain was grown in LB medium supplemented with the inducer arabinose for 6 h, and then cultures were assayed for β-gal activity. Data are presented in Miller units as the mean ± SD of results from duplicate assays from at least three separate experiments.

Discussion

PqsR is recognized as a key regulator in the P. aeruginosa cell-to-cell signaling network that governs the expression of multiple virulence factors, and this study focused on understanding how pqsR expression is regulated. Through these experiments we were able to map the pqsR promoter, showing that pqsR transcription is controlled at two separate sites. From these data we can propose a model of pqsR transcriptional activation (Fig. 7). We found that pqsR transcription can initiate at the proximal TS1 promoter site (Fig. 3), which we suggest acts as a basal promoter. The fact that deletion of lasR, or disruption of the las-rhl box, causes a large decrease in expression from the distal pqsR promoter, but a more modest decrease in expression of the pqsR coding region (Fig. 1 and Fig. 5) supports the notion that the proximal promoter sites contribute to pqsR transcription in the absence of LasR. However, expression from the proximal promoter is potentially limited in two ways. One of these is through the activity of MvaT and MvaU, which specifically inhibited transcription from the proximal, but not the distal promoter site (Fig. 6). Another limiting component is the negative regulatory sequence element that we discovered near the pqsR TS1 (Fig. 4 and Fig. 5). We found this element through experiments where we altered sequences which appeared to be part of a σ⁷⁰-dependent core promoter element. Instead of reducing transcription, these mutations caused a large increase in transcriptional initiation (Fig. 5). A recent report found that there was poor conservation of σ⁷⁰ (RpoD) binding motifs in P. aeruginosa (Schulz et al., 2015), and it remains unclear which sequences are important for promoter recognition at TS1. It is also uncertain why these mutations caused increased transcription; the changes may have simply caused a more favorable condition for initiation or promoter clearance, or this regulatory element may act as a binding site for a yet unidentified repressor.

Fig. 7.

A proposed model of pqsR transcriptional activation at two distinct promoter sites. A basal level of pqsR transcription occurs at proximal promoter sites, but this may be limited by the binding of repressors such as MvaT and MvaU, or another unidentified repressor (R?) to these sites. As cell populations increase, LasR + 3-oxo-C₁₂-HSL activates transcription from the distal start sites, but CysB can compete with LasR for binding at this promoter. Transcription from the distal promoter provides more copies of the pqsR coding region for translation, and could also aid in the displacement of repressors from downstream promoter sites.

As levels of 3-oxo-C₁₂-HSL increase along with population density, LasR binds this signaling molecule and becomes active. It can then induce pqsR transcription from the distal TS3/4 start sites, leading to the production of multiple RNAs (Fig. 2). We found that some of these transcripts terminate prior to reaching the pqsR coding sequence, at a point between the pqsR TS1 and TS2. No potential rho-independent transcriptional terminators were identified in this region using the prediction programs ARNold (Naville et al., 2011) or TransTermHP (Kingsford et al., 2007). Therefore it would be interesting to know what contributes to the premature termination of these transcripts, since this could have a significant impact on pqsR expression. Other transcripts from the distal promoter extend into pqsR (Fig. 2), and these are likely responsible for the increased pqsR expression when this promoter is active. In addition, LasR-directed transcription may positively influence pqsR expression in another way as well. We speculate that active transcription from the distal promoter site may displace repressors bound in the vicinity of the proximal promoter, allowing RNA polymerase to have better access to this promoter site. The regulators ToxT in Vibrio cholera and SsrB in Salmonella enterica can both counter H-NS-mediated silencing and directly activate transcription at some promoter sites, possibly by altering the promoter confirmation (Yu & DiRita, 2002, Walthers et al., 2011). Furthermore, some evidence suggests that RNA polymerase can displace H-NS from DNA when it is actively engaged in transcription (Dame et al., 2006). Either of these mechanisms would allow LasR to indirectly have a positive effect on the proximal promoter site by interfering with the binding of the H-NS family members MvaT and MvaU. Collectively our investigations imply that pqsR is transcribed from two separate promoter sites simultaneously, with positive and negative regulatory influences occurring at each site.

This type of complex regulatory scheme should allow for fine-tuning of pqsR expression, which is in keeping with the lifestyle of P. aeruginosa. Quinolone signaling and production can have major effects on P. aeruginosa physiology, and modulation of this system is likely important during transitions to the many different environments which these bacteria inhabit. As surroundings change, the control of receptor availability can act as a counterpoint to signal production in cell-to-cell signaling systems. Indeed, the other P. aeruginosa signal receptors, LasR and RhlR, are each subject to mechanisms working at both transcriptional and post-transcriptional levels to alter their expression (Balasubramanian et al., 2013, Siehnel et al., 2010, Gupta et al., 2009). Furthermore, the translation of pqsR can be positively affected by the small RNA PhrS, which appears to be up-regulated in response to reduced oxygen levels (Sonnleitner et al., 2011). In this type of environment PQS production is limited by oxygen availability since molecular oxygen is required for the final step in PQS synthesis (Toyofuku et al., 2008, Schertzer et al., 2010). In this situation, increased PqsR levels could help to maintain this signaling circuit by making more receptor available for the less potent coinducer 2-heptyl-4-quinolone, allowing for continued pqsABCDE expression and quinolone production. This helps to explain the quinolone production that is observed under anaerobic conditions (Schertzer et al., 2010), and provides an example of how modulating PqsR levels may be beneficial under different conditions. The control of the components in each P. aeruginosa cell-to-cell signaling system is an opportunity for these bacteria to integrate information from environmental cues with intercellular communication, and understanding the regulation of these signaling components should provide new insight into the stimuli that drive virulence factor production in this pathogen.

Experimental Procedures

Bacterial strains and growth conditions

Bacterial strains used in this study are listed in Table S1. Bacteria were cultured in lysogeny broth (LB, Lennox formulation), and were freshly plated from frozen stocks to begin each experiment. Unless indicated otherwise, cultures were incubated at 37°C with shaking at 260–280 rpm. When necessary for selection or to maintain plasmids, cultures were supplemented with 200 µg ml⁻¹ carbenicillin and/or 30 µg ml⁻¹ gentamicin.

Primer extension analysis

To isolate RNA for primer extension analyses, P. aeruginosa strains were sub-cultured in LB medium at an optical density at 660 nm (OD₆₆₀) of 0.05, and incubated at 37°C for 6 h. Cells were harvested by centrifugation, and treated with RNAprotect Bacteria Reagent (Qiagen). Cells were then incubated in lysis buffer (30 mM Tris pH 8, 1 mM EDTA, 10 mg ml⁻¹ lysozyme, 2 mg ml⁻¹ proteinase K) for 5 min at room temperature, and RNA was isolated using TRIzol Reagent (Life Technologies). Primer extension analyses were performed as described elsewhere (Farrow et al., 2015), using either primer PQSR PEXT4 or PQSR PEXT 5 and 80 µg of RNA per reaction. Extension products and sequencing ladders were separated on denaturing 8% polyacrylamide gels and visualized by autoradiography.

Chromosomal lacZ strains and β-galactosidase (β-gal) assays

To generate lacZ fusions that contained defined segments of the pqsR promoter, DNA fragments corresponding to either +75 to −276, +75 to −446, or −458 to −776 relative to the pqsR translational start site were amplified by PCR. The oligonucleotide primers used for these amplifications contained either PstI or HindIII restriction sites (Table S2). The amplified fragments were digested with the appropriate enzymes and ligated into plasmid pUC18T-mini-Tn7T-Gm-lacZ, which had been digested with the same enzymes, to generate plasmids pJF-Tn7T-PpqsRtc-TS2, pJF-Tn7T-PpqsRtc-prox, or pJF-Tn7T-PpqsRtc-dist (Table S1). To generate pqsR’-lacZ fusions that contained nucleotide substitutions or deletions in the pqsR promoter region, inverse PCR was performed using Pfu Turbo DNA polymerase (Agilent technologies) and oligonucleotide primers with a 5’ phosphate group (Table S2). The template DNA for these reactions was either pJF-Tn7T-PpqsR’Tc or pJF-Tn7T-PpqsRtc-dist. The resulting PCR products were re-circularized by ligation to produce the desired constructs. All plasmids were analyzed by DNA sequencing to ensure that the desired mutations were present. The lacZ fusion from each plasmid was integrated onto the P. aeruginosa chromosome at the attTn7 site as described elsewhere (Choi & Schweizer, 2006). The plasmid-encoded gentamicin resistance gene was also removed from the chromosome as described elsewhere (Choi & Schweizer, 2006). For β-gal assays, washed cells from overnight cultures of chromosomal lacZ fusion strains were used to inoculate 10 ml of LB medium to an OD₆₆₀ of 0.05. These sub-cultures were incubated at 37°C for 6 h, at which time pqsR’-lacZ expression was near its maximal level (Xiao et al., 2006b and data not shown), and then assayed for β-gal activity as described elsewhere (Miller, 1972). For β-gal assays performed using strains carrying protein expression constructs, washed cells from overnight cultures were used to inoculate 10 ml of LB medium supplemented with 200 ug ml⁻¹ carbenicillin to an OD₆₆₀ of 0.05. Then 1 ml aliquots of inoculated media were transferred into 13 ml round bottom tubes and additionally supplemented with either water, 0.025% arabinose (inducer for CysB, MvaT, MvaU expression), or 0.1% arabinose (inducer for PmpR expression). These concentrations of arabinose were chosen to produce similar levels of over-expression for each protein as judged by SDS-PAGE. Cultures in tubes were incubated with shaking at 37°C for 6 h, and then assayed for β-gal activity.

Northern blot analysis

To isolate RNA for Northern blot analyses, P. aeruginosa strains were sub-cultured in LB medium to an OD₆₆₀ of 0.05, and incubated at 37°C for either 3 h or 6 h. Then RNA was isolated using the same procedure as described for primer extension analysis. For each Northern blot analysis, 15 µg of RNA per lane was separated by electrophoresis on a denaturing 6% polyacrylamide gel. Next, RNA was transferred from the gel onto a BrightStar-Plus nylon membrane (Life Technologies) by semi-dry transfer, and RNA was fixed to the membrane by UV crosslinking. Probes for Northern blot analyses were generated by labeling single-stranded DNA oligonucleotides on the 5’ end using [γ-³²P]ATP and T4 polynucleotide kinase. Hybridizations were carried out in ULTRAhyb-Oligo buffer (Life Technologies) at 42°C using the procedures recommended by the manufacturer. Northern blots were visualized by autoradiography.

Reverse transcription-polymerase chain reaction (RT-PCR) analysis

P. aeruginosa strain PAO1 was cultured in LB medium for 6 h at 37°C and RNA was isolated as described for primer extension analysis. RNA samples were treated with TURBO DNase (Life Technologies) prior to RT-PCR analysis. RT-PCR was performed using the Access RT-PCR system (Promega) with 120 ng of RNA as a template and the oligonucleotide primers listed in Table S2. Only the reverse primers (complementary to the pqsR coding strand) were included in each reaction during the reverse transcription step. As controls, reactions that did not include reverse transcriptase and reactions that used P. aeruginosa strain PAO1 chromosomal DNA as a template were also performed. Aliquots of each completed reaction were analyzed by electrophoresis on an agarose gel.

S1 nuclease protection assays

RNA for S1 nuclease protection assays was isolated as done for primer extension analysis. For some experiments RNA was additionally treated with Terminator 5’-phosphate-dependent exonuclease (TEX) (Epicentre). 160 µg of total RNA was incubated in 1 × TEX reaction buffer B with 1.33 U µl⁻¹ RNase OUT (Life Technologies) and either with or without 0.067 U µl⁻¹ TEX for 60–90 min at 42°C. Reactions were quenched by extraction with 1:1 phenol:chloroform, and RNA was collected by ethanol precipitation. Probes for protection assays were prepared by labeling single-stranded DNA oligonucleotides with ³²P at the 5’ end using [γ-³²P]ATP and T4 polynucleotide kinase. For protection assays, 80 µg of RNA was suspended in 20 µl of a buffer containing 80% formamide, 40 mM PIPES (ph 6.4), 400 mM NaCl, and 1 mM EDTA, and then mixed with 1 × 10⁵ cpm (approximately 8–12 fmol) radiolabeled oligonucleotide. Mixtures were incubated at 75°C for 15 min, then at 37°C overnight. Next, 300 µl of a reaction mixture containing 1 × S1 reaction buffer, 20 µg ml⁻¹ sheared, denatured salmon sperm DNA, and 700 U S1 nuclease (Promega) was added to each sample, and reactions were incubated at 30°C for 1 h. Reactions that did not include S1 nuclease and reactions that contained only S1 nuclease and the oligonucleotide probe were also performed as controls. Reactions were quenched by adding 80 µl of a buffer containing 4 M ammonium acetate, 20 mM EDTA, and 40 ug ml⁻¹ of yeast tRNA (Life Technologies), and nucleic acids were collected by ethanol precipitation. Reaction products were analyzed by electrophoresis on denaturing 8% polyacrylamide gels along with DNA sequencing reactions that were generated using oligonucleotide primers that corresponded to the 5’ end of the probes used for the protection assays. Gels were visualized by autoradiography.

Construction of a pqsR promoter mutant strain

To create a plasmid for generating a pqsR promoter mutant strain, an approximately 2.85 kb DNA fragment that included the pqsR promoter and coding sequence was amplified by PCR. The oligonucleotide primers used for this amplification contained EcoRI sites (Table S2). The DNA fragment was digested with EcoRI and then ligated into pEX18Ap, which had also been digested with EcoRI, to produce pPqsR-suc. Plasmid pPqsR-suc was then used as the template for a PCR reaction with the oligonucleotide primers PQSRP mut-10alt and PQSRP DELCORE REV, which each had a 5’ phosphate group. The resulting DNA fragment was purified from an agarose gel and circularized by ligation to produce pPqsR-suc-alt-10mut. This plasmid carried a seven base pair substitution in the alternative −10 sequence upstream from the pqsR transcriptional start site 1 (Fig. 4). The pqsR promoter mutation was transferred from pPqsR-suc-alt-10mut onto the chromosome of P. aeruginosa strain PAO1 as described elsewhere (Farrow & Pesci, 2007). To confirm that the mutation was present, the pqsR promoter region was amplified by PCR using chromosomal DNA from the potential mutant strains as a template, and then this DNA fragment was analyzed by DNA sequencing.

Analysis of PQS production

To examine PQS production, 10 ml cultures of LB media were inoculated with several colonies of freshly grown P. aeruginosa cells and incubated at 37°C for approximately 6 h. Then washed cells from starter cultures were used to inoculate 10 ml sub-cultures to an OD₆₆₀ of 0.05, which were incubated at 37°C for 18 h. Next, samples of each culture were extracted with acidified ethyl acetate as described elsewhere (Carty et al., 2006). Samples of each extract were analyzed by thin-layer chromatography, and the amount of PQS in each extract was determined by comparison to known amounts of synthetic PQS using densitometry (Carty et al., 2006).

qRT-PCR analysis

For quantitative real-time PCR (qRT-PCR) analysis, RNA was isolated as described for primer extension analysis. Total RNA samples were treated with Turbo DNase (Life Technologies), and then used in cDNA synthesis reactions with SuperScript III reverse transcriptase (Life Technologies) as described elsewhere (Farrow et al., 2015). Real-time PCR was performed on a Bio-Rad CFX96 system using FastStart SYBR green master mix (Roche) and the following program: 95°C for 10 min; followed by 40 cycles of 95°C for 15 s, 58°C for 20 s, 72°C for 20 s, and 68°C for 5 s followed by a plate read; with a final step of 95°C for 10 s followed by a melt curve analysis. Oligonucleotide primers used for these reactions are listed in Table S2. At least two technical replicates were performed for each reaction. Relative expression was calculated using the Pfaffl method (Pfaffl, 2001), with rplU used as a reference gene.

Construction of protein expression plasmids

To generate plasmids for controlled protein expression in P. aeruginosa, the coding region for each protein (relative to the translational start site, from +1 to +763 for pmpR, from +1 to +451 for mvaT, and from +1 to +460 for mvaU) was amplified by PCR. The oligonucleotide primers for these reactions were designed to contain either an NcoI or a PciI restriction site that overlapped with the start codon (ATG) for each gene, and an XbaI restriction site downstream from the coding sequence. Each of these DNA fragments was digested with the appropriate enzymes, and then was ligated with vector plasmid pHERD20T, which had been digested with NcoI and XbaI, to produce plasmids pJF70 (PmpR expression), pJF72 (MvaT expression), and pJF73 (MvaU expression). These constructs placed the gene for each protein under the control of the inducible P_BAD promoter.