Early activation of quorum sensing in Pseudomonas aeruginosa reveals the architecture of a complex regulon
© Schuster and Greenberg. 2007
Received: 22 March 2007
Accepted: 22 August 2007
Published: 22 August 2007
Skip to main content
© Schuster and Greenberg. 2007
Received: 22 March 2007
Accepted: 22 August 2007
Published: 22 August 2007
Quorum-sensing regulation of gene expression in Pseudomonas aeruginosa is complex. Two interconnected acyl-homoserine lactone (acyl-HSL) signal-receptor pairs, 3-oxo-dodecanoyl-HSL-LasR and butanoyl-HSL-RhlR, regulate more than 300 genes. The induction of most of the genes is delayed during growth of P. aeruginosa in complex medium, cannot be advanced by addition of exogenous signal, and requires additional regulatory components. Many of these late genes can be induced by addition of signals early by using specific media conditions. While several factors super-regulate the quorum receptors, others may co-regulate target promoters or may affect expression posttranscriptionally.
To better understand the contributions of super-regulation and co-regulation to quorum-sensing gene expression, and to better understand the general structure of the quorum sensing network, we ectopically expressed the two receptors (in the presence of their cognate signals) and another component that affects quorum sensing, the stationary phase sigma factor RpoS, early in growth. We determined the effect on target gene expression by microarray and real-time PCR analysis. Our results show that many target genes (e.g. lasB and hcnABC) are directly responsive to receptor protein levels. Most genes (e.g. lasA, lecA, and phnAB), however, are not significantly affected, although at least some of these genes are directly regulated by quorum sensing. The majority of promoters advanced by RhlR appeared to be regulated directly, which allowed us to build a RhlR consensus sequence.
The direct responsiveness of many quorum sensing target genes to receptor protein levels early in growth confirms the role of super-regulation in quorum sensing gene expression. The observation that the induction of most target genes is not affected by signal or receptor protein levels indicates that either target promoters are co-regulated by other transcription factors, or that expression is controlled posttranscriptionally. This architecture permits the integration of multiple signaling pathways resulting in quorum responses that require a "quorum" but are otherwise highly adaptable and receptive to environmental conditions.
In Pseudomonas aeruginosa, two acyl-homoserine lactone quorum sensing (acyl-HSL QS) systems, LasR-LasI and RhlR-RhlI, control the expression of partially overlapping sets of genes. Many of the regulated genes are implicated in virulence and biofilm formation of this opportunistic pathogen. LasI and RhlI are enzymes that synthesize the acyl-HSL signals 3-oxo-dodecanoyl (3OC12)-HSL and butanoyl (C4)-HSL, respectively. LasR and RhlR are receptors that specifically bind the signal generated by their cognate synthases to regulate transcription of target genes [1–5]. Three independent transcriptome analyses identified several hundred such genes [6–8]. The numbers of identified genes and their expression patterns varied among studies depending on experimental conditions and statistical criteria used. In our study  we noted that the expression of most quorum-controlled genes increased in the stationary phase of growth and could not be advanced to logarithmic phase by the addition of exogenous signal, confirming earlier observations with individual quorum-controlled genes [9–11]. This suggested that the activation of most quorum-controlled genes is not triggered by the accumulation of signal, and seems to require additional factors.
If LasR, RhlR or RpoS levels were limiting the expression of QS-controlled genes during logarithmic growth of wildtype P. aeruginosa, early expression of these factors should advance quorum-controlled gene expression from stationary to logarithmic phase. To test this hypothesis, we constructed strains for regulatable expression of LasR, RhlR, and RpoS. Each allele, under the control of an arabinose-inducible araBAD promoter, was inserted in single copy into a neutral site of the chromosome of a lasR, rhlR mutant or an rpoS mutant, respectively. This arrangement allowed the careful titration of expression levels in logarithmic phase such that they matched those of the wildtype in stationary phase. For LasR and RhlR, it also allowed us to assess the impact of each regulator on gene expression independently and uncoupled from the QS hierarchy, where LasR-LasI is required for RhlR-RhlI expression [20, 23, 24]. We have thoroughly validated this approach by monitoring target gene expression and regulator protein levels in the engineered strains and in the parent strain throughout growth (see below).
A heat map shows that the early expression of LasR, RhlR, and RpoS induced many genes in logarithmic phase to levels close to those of the wildtype in stationary phase (Fig. 3B). Nineteen genes were also activated by the addition of 3OC12-HSL and C4-HSL signals alone, but their induction levels were comparatively low and all of them could be induced further by early expression of RhlR (Fig. 3). Early expression of RpoS often only partially activated those genes that were more highly activated by LasR or RhlR. Because RpoS also has a small effect on LasR and RhlR expression , it appears that RpoS is capable of advancing many genes only indirectly through LasR and RhlR. Overall, there was a good correlation between the genes induced by early expression of LasR or RhlR, and their signal specificity determined by addition of 3OC12-HSL or both 3OC12-HSL and C4-HSL to a signal generation mutant . For example, a cluster of rhl-specific genes was advanced by C4-HSL-RhlR but not by 3OC12-HSL-LasR (Fig. 3C), whereas several las-specific genes were advanced by 3OC12-HSL-LasR but not C4-HSL-RhlR (Fig. 3D). However, there were exceptions. PA5481 and PA5482, for example, can be advanced by 3OC12-HSL-LasR but not C4-HSL-RhlR (Fig. 3D) although in a signal synthesis mutant they responded better to both signals than to 3OC12-HSL alone . Such genes may require the binding of both regulators to their promoters for activation, or they may require additional factors that are themselves under the control of 3OC12-HSL-LasR.
Among the genes significantly induced by early expression of LasR, RhlR, or RpoS were 47% of all genes activated early by addition of signals to cells in conditioned medium . The large majority of this subset was induced by early expression of RhlR. RhlR was one of the genes triggered by media conditioning  and it also induced in minimal medium . Thus, the advancement of QS in conditioned medium appears to be partially mediated through super-regulation of RhlR.
Most quorum-controlled genes showed little to no activation by LasR, RhlR or RpoS expression in logarithmic phase. One-hundred and twenty of these genes are late genes. Their quorum-dependent (and in most cases RpoS-independent) induction in the wildtype strain is delayed until stationary phase (Additional file 2). These late genes could not be advanced although several are predicted to be directly activated by LasR or RhlR as they possess conserved sequence elements, so-called las-rhl boxes, in their promoter regions [7, 8, 11]. Thus it appears that 3OC12-HSL-LasR, C4-HSL-RhlR, and RpoS levels are not limiting for the activation of many quorum-controlled genes.
As indicated above, there was generally a good correlation between regulator specificity and acyl-HSL signal specificity (as determined by early expression of LasR or RhlR in a signal receptor mutant and by addition of 3OC12-HSL or both signals to a signal synthesis mutant, respectively). An exception is PA0179, which was 3OC12-HSL specific  but responded to both LasR and RhlR expression in logarithmic phase (Fig. 4). This suggests that RhlR-C4-HSL is capable of activating PA0179, but this effect is masked in the signal generation mutant because 3OC12-HSL-LasR is required for RhlR expression and 3OC12-HSL-LasR already saturates the PA0179 promoter. Hence, our expression strategy provides a different assessment of QS promoter specificity because it functions independently of the QS cascade.
We hypothesized that most genes advanced by QS are directly activated by QS. If this hypothesis is true, then we should be able to identify conserved sequence elements, so-called las-rhl boxes, upstream of these genes. In our previous study, we identified las-rhl box like sequences in 40 out of the 168 predicted quorum-controlled promoters . Many of these promoters are associated with QS advanced genes. Of the 53 predicted promoters advanced by LasR, 16 contain a las-rhl box-like sequence, and of the 26 predicted promoters advanced by RhlR, 17 contain a las-rhl box-like sequence. Thus, it appears that genes advanced by RhlR are for the most part also directly regulated by this transcription factor, whereas genes advanced by LasR are mostly indirectly regulated (i.e. genes directly regulated by LasR are more evenly distributed among the genes that can and cannot be advanced). However, our data do not provide evidence for indirect regulation by LasR via a regulatory cascade. Candidate regulatory genes mvfR, PA2591, and vqsR, which are under direct transcriptional control by LasR [7, 17, 26, 27], are not among the genes activated early (Additional file 1).
In a second step, the matrix generated with CONSENSUS was used to search the upstream regions of all previously identified quorum-controlled genes. Fourty-two predicted QS promoters showed a significant match. Not surprisingly, most sequences had also been identified in our previous study . In addition, we identified sequences with similarity to the RhlR consensus upstream of the following genes: PA2081, PA2146, PA3361, PA4141, PA4217 (qsc132), and PA5356 (glcC). Of these, the latter three were found to be advanced by RhlR.
When P. aeruginosa is grown in complex medium, LasR, RhlR and RpoS protein levels increase in stationary phase (Fig. 2), correlating with the activation of most quorum-controlled genes. Our approach to investigating the advancement of quorum sensing gene expression was to adjust the expression levels of regulators in log-phase cells to match the levels in stationary phase. This is an overexpression strategy in which proteins are maintained within physiological levels known to exist in P. aeruginosa. Many QS genes could be activated in logarithmic phase by the early expression of LasR, RhlR, and RpoS, but not by signal addition. Thus, the levels of these three proteins can be critical in modulating the quorum-response in P. aeruginosa. Several of the genes were induced by more than one regulator, confirming overlapping specificities and co-regulation. Genes that directly and exclusively respond to QS are likely among the subset of genes whose maximal expression can be advanced by increased expression of LasR or RhlR during logarithmic phase. In fact, our results suggested that most genes advanced by RhlR are directly regulated by this transcription factor, allowing us to build a RhlR consensus sequence.
Many other QS genes fail to be advanced by 3OC12-HSL-LasR and C4-HSL-RhlR, although it has been shown that some of these genes are directly regulated by these signal and response systems. This confirms and extends a recent observation for an individual gene, rhlA, which is not significantly activated in logarithmic phase even when C4-HSL-RhlR is present . These results suggest that the corresponding promoters are co-regulated by other transcription factors, likely constituting a network motif known as a multi-input dense overlapping regulon . Some of the regulatory inputs may affect translation rather than transcription, as is the case for small regulatory RNAs that modulate quorum sensing gene expression [33, 34]. The overall topology allows for specific responses to a multitude of signals, and may provide the basis for the exceptional environmental adaptability of P. aeruginosa. It also provides a simple explanation for the seemingly discordant sets of quorum-controlled genes identified by two separate groups under different culture conditions [7, 8, 35].
Taken together, our results indicate that co-regulation of target genes is an important, and perhaps predominant, feature of the P. aeruginosa QS network in addition to the well established super-regulation of the central components, LasR-LasI and RhlR-RhlI. Thus, a thorough understanding of the QS network will necessitate a comprehensive analysis of target promoter architecture, which should include the global identification of transcription factor binding sites. Technologies such as ChIP-chip, chromatin immunoprecipitation and microarray analysis , make this approach feasible.
Bacterial strains and plasmids
Strain or plasmid
Reference or origin
F-, φ80dlacZΔM15 Δ(lacZYA-argF)U169
deoR recA1 endA1 hsdR17(rk-, mk+)
phoA supE44 λ-thi-1 gyrA96 relA1
recA pro hsdR RP4-2-Tc::Mu-Km::Tn7
PAO1 derivative; ΔlasR::TcR
PAO1 derivative; ΔrhlR::GmR
PAO1 derivative; rpoS::GmR
PAO1 derivative; ΔrhlRΔlasR, pBADlasR
PAO1 derivative; ΔrhlRΔlasR, pBADrhlR
PAO1 derivative; rpoS::GmR, pBADrpoS
Source of Flp recombinase; ApR
araC-pBAD cloned in pBBR1MCS-5; GmR
lasR in pJN105
rhlR in pJN105
rpoS in pJN105
mini-CTX1 based plasmid containing
araC-pBAD from pBAD30; TcR
lasR in pSW196
rhlR in pSW196
rpoS in pSW196
Strains for regulatable expression of LasR, RhlR, and RpoS were constructed as follows: Alleles of lasR, rhlR, and rpoS were placed under control of an arabinose-inducible promoter and inserted in single copy into the chromosome of P. aeruginosa using a specialized integration-proficient plasmid system [37, 38]. The strains contained mutations in the respective chromosomal loci; i.e. lasR or rhlR were expressed in PAO lasR rhlR, and rpoS was expressed in PAO rpoS. Because of constraints with antibiotic resistance markers, lasR and rhlR expression constructs were first introduced into an isogenic PAO rhlR single mutant. To construct a double mutant background, a chromosomal lasR mutation was introduced into these strains by transformation with chromosomal DNA isolated from a PAO lasR mutant.
The lasR ORF was amplified from P. aeruginosa PAO1 genomic DNA by polymerase chain reaction (PCR) using primers 5'-N6 GAATTCTGATTAACTTTATAAGGAGGAAAACATATG GCCTTGGTTGACGGTTTTC-3' and 5'-N6 GCGGCCGCGGCAAGATCAGAGAGTAATAA GAC-3'. The underlined sequences indicate EcoRI and NotI restriction sites, respectively. The sequences in italics indicate a T7 gene enhancer element and an optimized ribosomal binding site (RBS) . These sequences were included to enhance translation of LasR because expression levels from the native RBS (as assessed by Western blotting) were low even when induced fully. The rhlR and rpoS alleles were subcloned from pJN105.rhlR and pJN105.rpoS. These plasmids were constructed previously (see below). The lasR PCR product was cut with EcoRI and NotI, and pJN105.rhlR and pJN105.rpoS were cut with EcoRI-NotI and KpnI-SpeI, respectively. The digested fragments containing lasR, rhlR, or rpoS were ligated with appropriately cut pSW196 , which contains an arabinose-inducible araBAD promoter inserted into plasmid mini-CTX1 . The resulting plasmids pSW196.lasR and pSW196.rhlR were each mobilized into PAO rhlR, and pSW196.rpoS was mobilized into PAO rpoS via E. coli S17-1. Mini-CTX1 encodes a site-specific integrase which mediates insertion at the chromosomal attB site, and it also contains Flp recombinase target sites flanking the multiple cloning site thus allowing in vivo excision of unwanted plasmid backbone DNA. To accomplish excision, pFLP2, encoding Flp recombinase, was introduced into the P. aeruginosa strains containing the respective lasR, rhlR, or rpoS expression constructs. This plasmid was then cured via the pFLP2-encoded sacB counterselectable marker. To determine whether strains were constructed properly, the size of fragments encompassing the attB region was determined by PCR with the primers 5'-CGTACAACGTGCCGGATATCG-3' and 5'GCTTCGGGATAAGCCAATCCTG-3' and subsequent agarose gel electrophoresis. In contrast to lasR and rhlR constructs, the rpoS expression construct did not insert at the attB site, even after repeated attempts. It did insert, however, at another, unknown site. We were unable to determine the exact insertion site by arbitrary PCR. We verified that the transcript profile of the rpoS mutant carrying the rpoS construct under non-inducing conditions was very similar to that of the original rpoS mutant, indicating that the insertion did not result in significant unintended gene expression changes. Transcription of neighboring chromosomal DNA from the araBAD promoter is minimized by strong T4 transcriptional terminators . To generate PAO lasR rhlR double mutant backgrounds, a lasR::TcR mutation was introduced by transformation of chemically competent cells with 80 μg of genomic DNA isolated from PAO lasR::TcR as described .
The vectors pJN105.rhlR and pJN105.rpoS were constructed as follows: The rhlR and rpoS ORFs including native ribosomal binding sites were amplified from PAO1 genomic DNA by PCR. The primers for rhlR were 5'-N6 GAATTCATCGATCAGGGCTTACTGCAATG-3' and 5'-N6 TCTAGAGCGCTTCAGATGAGACCCAGC-3' with the underlined sequences indicating EcoRI and XbaI restriction sites, respectively. The primers for rpoS were 5'-N6 GCTAGCAAGGGATAACGACATGGCAC-3' and 5'-N6 GAATTCTCACTGGAACAGCGC GTCACT-3' with the underlined sequences indicating NheI and EcoRI restriction sites, respectively. The resulting PCR fragments were cut with the indicated restriction enzymes and inserted into the multiple cloning site of appropriately digested plasmid pJN105  to give pJN105.rhlR and pJN105.rpoS.
Cultures were harvested at early logarithmic phase (OD600 = 0.2) and early stationary phase (OD600 = 2). RNA isolation as well as subsequent cDNA synthesis, labeling, and P. aeruginosa GeneChip genome array (Affymetrix, Santa Clara, CA) processing were performed as described previously [8, 15]. Each experiment was done in duplicate. A control experiment confirmed that the addition of arabinose had no significant effect on gene expression. The GeneChip profiles of P. aeruginosa in logarithmic phase with and without arabinose showed the same correlation as those of two replicate cultures grown in the absence of arabinose (not shown). Data were processed with the Affymetrix GeneChip Operating Software 1.1. The web-based program Cyber-T  was used for statistical analysis as described . The p-value threshold was 0.001, and the corresponding confidence estimate termed posterior probability of differential expression, PPDE (<p), was ≤ 0.97. This p-value was chosen based on graphical analysis of the distribution of p-values and their association with the non-uniform distribution (indicating differential expression) and the uniform distribution (indicating no differential expression) . To determine which quorum-controlled genes qualify as late genes, we demanded that transcript levels in stationary phase were significantly higher in wildtype P. aeruginosa compared to a lasR, rhlR mutant, and that this difference was larger in stationary phase than in logarithmic phase. Global gene expression patterns ("heat-maps") were constructed with GeneSpring 7.2 (Agilent Technologies, Palo Alto, CA). Transcript profiles of individual genes were normalized to the median expression level. Microarray data were deposited in the EMBL ArrayExpress repository (Accession number E-MEXP-1183).
P. aeruginosa culture conditions, as well as RNA isolation and cDNA synthesis procedures were identical to those used for microarray analysis. Real-time PCR was performed with an ABI Prism 7900 Sequence Detection System (Applied Biosystems, Foster City, CA). Cycling parameters were 10 min at 95°C followed by 40 cycles of 15 s at 95°C and 1 min at 60°C. Dissociation profiles of the amplified products were run to evaluate non-specific amplification. Each PCR reaction contained 1× SYBR Green Master Mix (Applied Biosystems), 1 ng cDNA template, and 7.5 pmol of each primer in a 25-μl volume. Gene-specific primers were designed and data were analyzed using Primer Express and SDS 2.1 software, respectively (Applied Biosystems). Relative transcript levels were determined by using the standard curve method and by using the nadB transcript as a calibrator. Standard curves were constructed with 10-4 to 10 pg of RNA-free genomic DNA purified from P. aeruginosa PAO1 (Genomic-tip kit, Qiagen). Experiments were performed in duplicate.
For protein analysis, samples were withdrawn from the same P. aeruginosa cultures that were used for transcript profiling. Cells were harvested by centrifugation. Pellets were suspended in lysis buffer containing 25 mM Tris-HCl, pH 7.8, 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 0.5% Tween-20, 10% glycerol, 2 μM 3OC12-HSL and 10 μM C4-HSL. The suspensions were sonicated, and the resulting lysates subjected to ultracentrifugation at 100,000 × g for 15 min. Protein concentrations were determined by using the Bradford assay. Approximately 2 μg of each supernatant fraction was separated by 12.5% SDS-PAGE. The separated proteins were blotted onto a nylon membrane. The membrane was treated with polyclonal antibodies against LasR, RhlR, or RpoS. Proteins were detected by using a secondary anti-rabbit horseradish peroxidase-conjugated IgG and chemiluminescent substrate (Pierce, Rockford, IL). Antibodies against P. aeruginosa LasR and RhlR were generated in rabbits immunized with His10-LasR or His10-RhlR affinity-purified under denaturing conditions. Polyclonal rabbit antibody against P. aeruginosa RpoS was obtained from V. Venturi .
Ten predicted promoter regions (-400 to -1 relative to the translational start) associated with the genes most highly induced by either LasR alone or by RhlR alone (Additional file 1) were each subjected to the pattern search algorithm CONSENSUS . Overlap of sequences with upstream open reading frames was disallowed. A matrix length of 16 bp was chosen for the final alignment, because it yielded the most consistent results. The highest scoring matrix from the final cycle was used to depict a sequence alignment in WEBLOGO . The same matrix was also used in PATSER  to identify sequences in upstream regulatory regions of other quorum-controlled genes. The default weight score of 7 was used as the lower threshold.
We thank Esther Volper and the University of Iowa DNA Facility for microarray processing. We thank Milica Sevo and Vittorio Venturi for providing anti-RpoS antibody. We also thank Dan Wozniak and Herbert Schweitzer for providing the integration-proficient plasmids. This work was supported by USPHS grant GM-59026.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.