A number of P. aeruginosa virulence genes are regulated by temperature
We used RNA-seq to examine the global gene expression of P. aeruginosa
PA14 at 37°C and 28°C, with an objective to assess whether virulence genes respond to alterations in temperatures encountered in different environments (Table S1
). Using replicate experiments, we identified a total of 144 genes that showed statistically significant over-expression of transcripts at 37°C, and 234 genes whose mRNA levels were reduced at 37°C compared to 28°C (Table S2
). These represent transcripts of genes that may be activated or repressed during infection of warm-blooded mammals and possibly respond to temperature as one of several input signals for their coordinated expression. Therefore, the levels of transcripts for 6.4% of the genes of the P. aeruginosa
PA14 genome were affected by temperature. Gene ontology (GO) enrichment analysis showed significant enrichment for genes related to protein secretion, phenazine biosynthesis and regulation of alginate biosynthesis among the genes upregulated at 37°C (p
, and 6×10−3
respectively; Materials and Methods
). Elevated expression of the regulatory gene cluster (mucABCD
) leads to repression of alginate production. However, this effect can be overcome in isolates from cystic fibrosis patients in which common mutations in mucA
lead to overproduction of the alginate exopolysaccharide. At 28°C, we observed upregulation of ribosomal protein genes, histidine catabolic genes, and aromatic and catechol-related genes (p
, and 4×10−3
, respectively). We found a strong enrichment for virulence-related genes among those upregulated at 37°C (hypergeometric test p
). Indeed, the expression of 43 out of 238 annotated virulence genes was altered in response to the temperature shift, with most of these (32, 74%) being upregulated at 37°C (; Materials and Methods
). The vast majority of the genes, 94%, were transcribed only from the strand containing the ORF, thus allowing quantification and comparison of the gene expression with strand-insensitive sequencing.
Expression of virulence-related genes at different temperatures.
At the two temperatures used in our transcriptome analysis, we observed few significant changes in the expression of cold-response (capB and PA14_05960) or heat-shock genes (htpG), and did not find evidence for extensive cold- or heat- shock response that is indicative of thermal stress. This suggests that P. aeruginosa has evolved to thrive under a range of conditions that may include a variety of environmental temperatures without the need to activate protective survival responses.
The type III secretion system is preferentially expressed at 37°C
The mRNA levels for the majority of the T3SS components (regulators, secretion machinery and effectors) were higher at 37°C () suggesting that this important P. aeruginosa virulence mechanism is subject to temperature regulation. Since the RNA-seq was carried out in calcium replete media, which is not optimal for the expression of the T3SS genes, we analyzed the production of the secretin component PscC by Western blots in two different strains, PA14 and PAO1, grown on calcium-depleted media, which promotes T3SS expression (). In both strains we observed immunoreactive PscC at 37°C but not at 28°C. Similarly, we observed temperature-regulated production of species-specific effectors of the T3SS, ExoS and ExoU in PAO1 and PA14, respectively, where they were detected only at 37°C (), confirming that the temperature regulation is not a PA14-specific trait.
Temperature-regulated expression of selected virulence factors.
The observed regulatory effect of temperature on the T3SS suggests that this virulence mechanism should play an important role in P. aeruginosa
infections of warm-blooded mammals. The importance of the T3SS for acute human nosocomial infections has been demonstrated and T3SS-deficient P. aeruginosa
strains are attenuated in murine models of pneumonia, bacteremia, keratitis and burn wound infections 
. However, the T3SS has been shown to contribute to virulence in models where the hosts were infected at their optimal growth temperatures, such as Drosophila melanogaster
, Acanthamoeba castellani
, Galleria mellonella
and more recently, Danio rerio
(zebrafish) embryos 
although this virulence mechanism appears non-essential for the infection of adult gnotobiotic fish 
. P. aeruginosa
can also infect Caenorhabditis elegans
, although its T3SS is expressed but not required for virulence 
. Since the T3SS contributes to the virulence of P. aeruginosa
even at reduced temperatures, it is conceivable that the environment of specific hosts provides additional signals that overcome those provided by the temperature or other environmental factors. Previous work has shown that contact between P. aeruginosa
and mammalian cells can trigger the expression of T3SS even in the presence of calcium concentrations that in vitro
inhibit the expression of these genes 
Effect of growth temperature on the expression of genes encoding phenazine biosynthesis enzymes
The second group of genes showing a significant increase in mRNA levels encodes enzymes responsible for the biosynthesis of phenazines, which are well characterized virulence factors of P. aeruginosa
. Phenazines are nitrogen-containing heterocyclic secondary metabolites that serve as signaling molecules influencing gene expression during environmental adaptations including biofilm formation 
. Phenazines are capable of producing reactive oxygen species toxic to eukaryotic cells and other bacteria 
. These molecules are also involved in electron shuttling to alternate terminal acceptors particularly during anaerobic growth 
. Moreover, the production of phenazines in other Pseudomonas
species is temperature regulated 
. In P. aeruginosa
, two unlinked gene clusters, phzA1-G1
encode enzymes catalyzing the synthesis of the core molecule phenazine-1-carboxylic acid (PCA) 
. Although levels of transcripts from both of the phz
clusters were elevated at 37°C, the temperature effect was more pronounced for phzA1-phzG1
(). Two additional genes located adjacent to the phzA1-G1
operon encode enzymes involved in the modification of PCA: PhzM (a methyltransferase) and PhzS (a monooxygenase). These two enzymes modify PCA to give pyocyanin, while PhzS alone can convert PCA to 1-hydroxy-phenazine 
. The levels of phzM
mRNAs were also increased at 37°C. Finally, the P. aeruginosa
chromosome contains the phzH
gene at an unlinked site; it encodes the enzyme for the conversion of PCA to phenazine-1-carboxamide. However, we did not detect significant expression of this gene at either temperature, and, therefore, it is unlikely that this modification step takes place in P. aeruginosa
under the conditions tested.
Phenazine is synthesized by a pathway utilizing metabolites that are also precursors for aromatic amino acids, TCA cycle intermediates and other bioactive molecules 
. We therefore examined the levels of transcripts for enzymes of the entire pathway and its branches, starting from condensation of erythrose-4-phosphate and phosphoenol pyruvate to various end products (Figure S1
). Genes found in several operons encoding enzymes of these pathways were significantly more expressed (ranging from 6.6- to 35-fold) at 28°C than 37°C. Specifically, antABC
, the genes for enzymes catalyzing the synthesis of catechol from anthranilic acid, and genes encoding CatABC and PcaDIJF, responsible for the conversion of catechol to the TCA cycle intermediates succinate and acetyl-CoA, were preferentially expressed at 28°C when compared to 37°C (). We also detected temperature regulation of the antR
gene encoding the transcriptional activator of the ant
operon, which was not the case for mRNA levels of catR
, the regulatory gene of the cat
operon. Although numerous regulatory inputs are involved in regulating these genes at the transcriptional and post-transcriptional levels 
, appears that the conversion of anthranilate to TCA cycle intermediates may occur less effectively at 37°C than at 28°C, making more precursors available for other biosynthetic pathways.
Our results therefore show that at 37°C there should be a significant increase in the levels of two metabolic intermediates affecting the synthesis of important signaling and virulence-enhancing molecules (Figure S1
). An increase in the concentration of chorismic acid would be the consequence of higher levels of PhzC at 37°C and constitutive, temperature-independent transcription of genes encoding enzymes of the shikimate pathway (AroF1BQ1EKAC). Chorismic acid can be also converted to anthranilic acid, which would accumulate at 37°C, in part because of a decrease in its flow through the breakdown pathway to the TCA cycle intermediates succinate and acetyl-CoA, resulting from lower levels the enzymes (CatABC, PcaDIJF). An important consequence of this redistribution of various metabolic intermediates at 37°C is the diversion of chorismic acid towards the biosynthesis of phenazine. Anthranilic acid is also a precursor for the biosynthesis of the quorum sensing molecules alkyl-4 quinolones (AQs). Therefore, the reduction of the levels of the ant
transcripts at 37°C (and by inference, levels of corresponding catabolic enzymes) could lead to an increase of the production of 2-heptyl-4-quinolone (HHQ) and 2-heptyl-3-hydroxy-4-quinolone (PQS), two important regulators of P. aeruginosa
cell-to-cell communication and virulence gene expression 
. We tested these predictions directly by comparing the levels of pyocyanin and PQS in P. aeruginosa
supernatants of cultures grown at 28°C and 37°C (). We found increased concentrations of both of these molecules at 37°C, although the effect on PQS production was much more pronounced when cultures reached the stationary phase of growth. Moreover, the pyochelin siderophore pathway also benefits from the altered levels of the various precursors of the aromatic amino acid pathway, since increased levels of chorismic acid as the consequence of a decrease in its flow towards the TCA cycle makes more of it available for the synthesis of pyochelin, although mRNA concentrations encoding the enzymes of this pathway were not affected by temperature.
Effect of temperature on production of pyocyanin and PQS.
The temperature-dependent redirection of transcripts encoding the components of the aromatic amino acid biosynthetic machinery, towards the synthesis of phenazines and signaling molecules such as AQs, provides new insight into the role of environmental modulation of global survival strategies of P. aeruginosa. Although our study focused on the influence of temperature on transcript levels, the precise molecular mechanisms that accomplish thermoregulation are unclear. However, the results of this RNA-seq study provide a basis for the design of rational genetic and biochemical experiments to probe the molecular details of signal transduction, gene expression and protein function at different temperatures that are undoubtedly coordinated with other physical or chemical environmental inputs.
The transcriptome structure of P. aeruginosa PA14
Understanding the transcriptome structure and operon organization in bacteria is essential for understanding bacterial RNA-based regulation 
. To produce a detailed transcriptome structure map of P. aeruginosa
, we used the mapping of the RNA-seq reads, which provides transcript coverage in a non-strand-specific manner. To complement this approach, we also used a strand-specific 5′-end sequencing method, which identifies active transcription start sites (TSS) at a single nucleotide resolution across the entire genome 
. The combination of the two sequencing methods provided a comprehensive view of the P. aeruginosa
PA14 transcriptome in an unbiased manner.
We were able to map transcription start sites (TSSs) for 2,117 transcriptional units (TU), spanning 3,325 protein-coding genes (55% of all protein coding genes, Table S3
). A total of 1,854 genes (56%) were found to be included in polycistronic TUs (Figure S2A
). Over 61% of the multi-gene operons are bicistronic, and only 19% contain 4 or more genes. This general operon organization is highly similar to that found in other bacteria. For example, in the Gram-positive Listeria monocytogenes
, in which the operon structure was determined by hybridization of RNA to tiling-arrays, it was found that 60% of the genes were transcribed in multi-gene operon structures, with about half of the multi-gene TUs transcribed as bicistronic mRNAs 
. In Geobacter sulfurreducens
, polycystronic mRNAs accounted for approximately 50% of all transcripts 
and a recent determination of Escherichia coli
K12 found that 35% of the TUs are polycistronic 
. Since these organisms vary greatly in genome size, growth conditions and physiology, there appears to be a general design principle for bacterial operon organization, which has no correlation with the genome coding capacity. We note that TU structures can be flexible and may change when conditions change 
; however, since in this study we tested only two growth conditions, a global analysis of alternative TU structures could not have been performed.
Definition of transcriptional units and the analysis of 5′ untranslated regions (5′ UTRs)
The capacity of 5′ UTRs to regulate transcriptional and post-transcriptional processes in cis
has been demonstrated in numerous organisms 
. Until now, the mapping of 5′ UTRs has not been performed systematically on P. aeruginosa
, thus limiting the discovery of new cis
-regulatory elements. Since our TSS mapping defines the 5′ ends of TUs, it also reveals the 5′ UTRs of the immediate downstream genes. Inspection of the 5′ UTRs for the 2,117 TUs we defined showed a median 5′ UTR length of 47 nt (Figure S2B
), similar to the 5′ UTR lengths reported in E. coli
(20–40 nt) 
and Synechococcus elongatus
(30 nt) 
. Most of the 5′ UTRs (77%) were shorter than 100 nt, with 60 genes completely lacking 5′ UTRs. Interestingly, 115 5′ UTRs were longer than 200 nt, which suggests that a significant fraction of these leader regions may function as cis-
regulatory RNA elements. We scanned the sequences of all 5′ UTRs longer than 100 nt using RFAM, and found that 13 of those contain known riboswitches and RNA-leaders (Table S3
). Since most of the well-characterized riboswitches models derive from highly divergent bacteria, such as Bacillus subtilis
, the presence of these long 5′ UTRs suggests that many post-transcriptional regulatory sequences in the P. aeruginosa
are yet to be discovered.
The availability of genome-wide TSS maps can allow the correction of false computational gene annotations 
. Indeed, we were able to correct the annotation of 46 genes (Table S3
Since gene annotation in P. aeruginosa
is largely limited to the description of ORFs, a genome-wide characterization of promoter regions and binding sites of regulators of transcription is often difficult. The availability of a comprehensive list of TSSs in P. aeruginosa
now allows an accurate search for promoter and regulatory regions for specific genes or operons. To demonstrate the utility of this approach, we selected LasR, the transcriptional regulator of quorum sensing and virulence genes. LasR independently regulates the expression of a number of genes. It is positioned at the top of a regulatory network that includes additional quorum sensing systems such as Rhl, PQS and Qsc and, therefore, its regulon is extensive 
. The number of genes directly or indirectly controlled by LasR was shown to include over 300 genes or about 6% of the genome 
. A recent chromatin immunoprecipitation coupled to microarray hybridization study (ChIP-chip) 
has characterized 35 LasR binding sites in the P. aeruginosa
PAO1 genome, but the lack of TSS information prevented accurate localization of the LasR sites relative to the transcript. By analyzing the TSSs of genes marked as LasR-bound by the ChIP-chip study, we were able to establish that this transcription factor has a preferential positioning of 51–52 bases upstream to the start of transcription (). Based on this analysis, we computationally identified 17 new, previously unrecognized putative LasR binding sites on the P. aeruginosa
PA14 genome (Table S4
; Materials and Methods
). An additional 20 sites, previously shown to be bound by LasR in strain PAO1 
were also identified in the PA14 strain. The vast majority of the sites recognized by LasR were immediately adjacent to the −35 element of the promoter, making it likely that it contacts RNA polymerase during transcription initiation (). Two genes, PA14_21030 (ATP-dependent Clp protease subunit) and PA14_16250 (lasB
, previously identified by Gilbert et al 
) contained multiple LasR binding sites at their promoters ().
Identification of new LasR binding sites.
To confirm that the new LasR binding sites identified by our computational methods represent direct targets of regulation by this transcription factor, we analyzed these promoters for direct binding by LasR using electrophoretic mobility shift assays (EMSAs) and for LasR-dependent expression using lacZ transcriptional fusions. When purified LasR was used to analyze DNA-protein interactions, we detected binding of this protein to all but one fragment (). Based on the low concentrations of LasR (0.3 pM) that were required to give a complete or near-compete shift, relatively high affinity binding was detected for at least five DNA fragments corresponding to the promoter regions of PA14_09480, PA14_09490, PA14_33830, and PA14_48530. For the rest of the DNA fragments, the extent of binding of LasR varied and in a number of cases, only a weak shift was observed even at the highest concentrations (30 pM) tested. No binding of LasR to the promoter region of PA14_23220 was detected.
To determine whether the newly identified LasR-dependent binding sites indeed respond to LasR in vivo
, we have cloned the same fragments used in the EMSAs into a transcriptional lacZ
reporter plasmid and incorporated the gene fusion into the chromosome of wild-type P. aeruginosa
PA14 and its isogenic lasR
mutant. LasR-dependent lacZ
expression was assessed at two time points corresponding to early stationary phase (8 hr growth) and late stationary phase (12 hr growth). In general, the levels of expression for most of the fusions, and correspondingly LasR dependence, were higher at later stages of growth (Figure S3
). A number of fusions were poorly expressed in P. aeruginosa
, and because of this, no LasR-dependent regulation was observed for the fusions transcribed from the promoters of PA14_03490, PA14_23220, PA14_18800, and PA14_33890. This lack of expression is consistent with the transcriptome data (RPKM values of 30 or less for these genes) suggesting they are not expressed under these conditions. Interestingly, we observed LasR binding to three of these promoters (PA14_03490, PA14_18800, and somewhat weakly, PA14_33890) in our in vitro
assay. Two conclusions could be drawn from these results. First, the predicted LasR binding sites adjacent to promoter regions do not necessarily indicate that a particular gene is expressed under all conditions. For example, PA14_03490, PA14_18800, and PA14_33890 could be expressed in environments that are different from these used in our work (LB, aerobic and at 37°C). Moreover, full transcriptional activation of certain LasR-regulated genes could require the input of additional regulatory factors. This regulatory complexity of quorum sensing would permit a more fine-tuned response of groups of genes to specific environmental inputs.
Regulation of phenazine biosynthetic operons
The P. aeruginosa
PA14 genome contains two temperature regulated gene clusters, encoding the enzymes for the synthesis of phenazine-1 carboxylic acid (phzS
, and phzA2-G2
; ). Regulation of the phzA1-G1
operon and phzM
by acyl-homoserine-lactones (HCLs) has been shown previously using DNA microarrays 
. Using the precisely-mapped TSSs, we independently identified in the phzA1
intergenic region a conserved palindromic LasR consensus binding sequence CTACCAGATCTTGTAG
, with the C
positioned 52 nt from the site and on the opposite strand, the sequence CTACAAGATCTGGTAG
with the C
54 nt from the TSS of phzM
(, ). Since the consensus LasR binding site is palindromic, the binding of this transcription factor to a single site in an intergenic region could allow activation of the two divergent promoters, a rather uncommon arrangement of regulatory sequences bound by positive transcription regulators in bacteria. Moreover, since the LasR binding site resembles the predicted binding site for a second quorum sensing regulator RhlR 
, and LasR regulates the expression of rhlR
, it is conceivable that the same palindromic double stranded DNA sequence serves as a binding site for either LasR or RhlR, or both for these two transcriptional units.
Role of LasR and RhlR in regulating the biosynthesis of pyocyanin.
Promoter-lacZ fusion experiments in wild-type, lasR, and rhlR mutants of P. aeruginosa showed strong LasR- and RhlR-dependent expression of the lacZ reporter fused to the phzA1 promoter fragment (). Similarly, the phzM promoter directed transcription in a LasR- and RhlR-dependent manner, although the fold decrease of transcription of the fusion in lasR and rhlR mutants was not as large as seen for the phzA1-lacZ fusions, owing to its high level of expression in wild-type P. aeruginosa. There was only modest regulation by LasR or RhlR for the phzS and phzA2 fusions ().
Because LasR controls the expression of rhlR, it was impossible to differentiate which of the two quorum sensing transcription factors regulated the expression of phzA1 or phzM. We therefore engineered phzA1-lacZ and phzM-lacZ promoter fusion plasmids and the expression of the reporter gene was measured in response to N-(3-oxododecanoyl) homoserine lactone (3O-C12-HSL) or N-butyryl homoserine lactone (C4-HSL), in E. coli, co-expressing LasR or RhlR on an inducible plasmid. DNA fragments containing the promoter regions of rsaL and rhlA were used as controls for specificity of expression of activation of the reporter constructs by LasR and RhlR, in response to 3O-C12-HSL and C4-HSL, respectively ().
An approximately 3.7-fold and 6-fold induction of phzA1-lacZ was seen in E. coli expressing either LasR or RhlR, in response to their cognate autoinducers. The phzM-lacZ fusion showed a significant, albeit modest, effect of the homoserine lactone autoinducers and LasR, RhlR (1.6-and 1.3-fold, respectively). The results from the expression of the phzA1-lacZ and phzM-lacZ fusions in P. aeruginosa and heterologous expression in E. coli indicate that binding of LasR and RhlR to the same site, in the context of the respective promoters, leads to the enhanced expression of phzA1 over phzM in overall magnitude and dependence on these two quorum sensing regulators.
There are several explanations for these observations. First, the two overlapping regulatory sequences found on the opposite DNA strands are not functionally identical although the differences are rather minor (). There are two differences in the regulatory sequences adjacent to each promoter (T/A at position 5 and C/G at position 12). These could account for strong preference for LasR and RhlR activation towards the transcription of phzA1. Moreover, the predicted promoters (−10 and −35 regions) are not identical and these could also provide additional context for the extent and magnitude of LasR and RhlR activation. Finally, the distances of the LasR element from the TSSs of phzA1 and phzM are different, although well within those seen for other LasR activated genes, and this could also contribute to the overall transcriptional level and extent of influence of quorum sensing regulators and autoinducers. Conceivably, all of these factors could incrementally contribute to regulated expression.
Detection of expressed non-coding RNAs
Non-coding RNAs (ncRNAs) are now appreciated as important regulators of diverse processes in bacteria. Although P. aeruginosa
has been studied extensively, only 38 intergenic ncRNAs (sRNAs) have been identified in its genome so far 
. By combining our TSS mapping and the whole-transcriptome data, we determined 223 events of intergenic transcription based on the current genome annotation (). Of these, 23 represent protein-coding genes that were annotated in other genomes but so far escaped detection in PA14 (Table S5
; Materials and Methods
), and 35 additional transcripts that might represent non-conserved protein coding genes (Table S6
). The sequences of the remaining 165 transcripts lack an ORF and therefore most likely represent non-coding transcripts (Table S6
). Most of the previously described sRNAs (31/38, 86%) were detected in our sRNA set, and we could define the exact TSS for 26 of them. Most of the sRNAs we identified (69.5%) are conserved only within other P. aeruginosa
strains, and only a few (10.5%) are conserved outside of the genus Pseudomonas
Distribution of temperature regulated genes, and segments transcribing antisense RNAs and intergenic small RNAs in the genome of P. aeruginosa PA14.
In addition, we identified 384 genes with overlapping transcription on the reverse strand, representing cis
-encoded antisense RNA (asRNA) transcription, Such asRNAs are known to be abundant regulators in many prokaryotes 
but until now they have not determined globally in P. aeruginosa
We next examined the distribution of the DNA segments that are transcribed into asRNAs and sRNAs in the genome of P. aeruginosa
PA14. Specifically, we assessed whether they are part of the flexible or core genome, as determined previously 
(). We noted an over 2-fold enrichment in the location of sequences specifying both sRNAs (Fisher's exact test p
0.003) and antisense RNAs (Fisher's exact test p
) in regions that have been previously designated as regions of genome plasticity (i.e. the flexible genome); sRNAs and asRNAs were found less frequently in the conserved core genome. For example, the annotated 115 kb pathogenicity island PAPI-1 carries only two genes with predicted function in transcriptional regulation, RL0037 (a RcsB ortholog) and a predicted transcription regulator RL0012 
. Strikingly, in addition to the 175 annotated genes in PAPI-1, we were able to detect additional 46 ncRNAs (10 sRNAs and 36 asRNAs) representing 21% of the genes in the island; this number of asRNAs in PAPI-1 is enriched by over 5-fold compared to the core genome. Since the flexible genome represents, in most instances, horizontally-acquired genes, it is likely that these segments specify regulatory elements controlling the expression of co-acquired target genes. However, it could be that genes expressed from other genomic islands or from the core genome could be subject to regulation by the trans
-acting sRNAs at a post-transcriptional level 
. These findings raise the possibility that the ability to acquire ncRNA regulators via horizontal gene transfer contributes significantly to bacterial evolution, where new phenotypes emerge by post-transcriptional activation or repression of genes already present in the recipient cells. An alternative explanation is that horizontally acquired sequences in the flexible genome were newly introduced to the genome, and due to lack of evolutionary time to ameliorate their coding- and non-coding sequences, the acquired sequences contain random promoter sequences that are recognized by Pseudomonas
Interestingly, two of the predicted sRNAs (Lrs1 and Lrs2) were identified as having a conserved binding site with similar binding affinity for LasR (). Monitoring the expression of lrs1
with transcriptional lacZ
fusions showed LasR dependence () throughout the entire growth phase of the cultures, although there was a slight increase in the expression of the lrs1-lacZ
fusion and lrs1
transcript levels (Figures S3
B) when cultures reached a late stationary phase of growth. A Northern blot with a probe designed for lrs1
with RNA extracted from wild-type PA14, a ΔlasR
mutant, and a ΔlasR
strain complemented with LasR on a plasmid indicated that lrs1
expression is indeed dependent on LasR (Figure S4B
). We also note that during the late stationary phase Lrs1 can be detected even in a lasR
mutant, suggesting that its regulation may be directed by another transcription factor expressed or activated at this late growth phase. We further demonstrated using EMSA that the concentrations of Lrs1 (but not Lrs2) in P. aeruginosa
were dependent on the RNA chaperone Hfq through a direct interaction (). We also observed that Lrs1 transcript levels, but not Lrs2 levels, are reduced in a hfq
mutant, possibly due to reduced transcript stability (). These results point toward an overlapping regulatory network that includes a transcriptional regulatory cascade and a sRNA-regulated post-transcriptional regulatory component. The quorum sensing regulatory system therefore very likely integrates, in addition to cell density, a number of regulatory inputs and directs these to the most optimal responses for survival in a particular environment.
Characterization of the LasR-regulated small RNAs Lrs1 and Lrs2.
To further investigate the possible roles of Lrs1, we constructed an isogenic deletion mutant lacking the lrs1
sequence. We first determined the precise 5′ and 3′ ends of the transcript following intramolecular ligation and synthesis of cDNA across the junction. The sequence of the cDNA was used to design PCR primers for deletion of the entire lrs1
in P. aeruginosa
PA14. Surprisingly, the culture of the mutant strain was devoid of the characteristic blue-green pigment of P. aeruginosa
, a color that stems from production of pyocyanin (Figure S4C
). Measurements of levels of pyoverdine, the fluorescent pigment produced by P. aeruginosa
, showed no detectable differences between culture supernatants of wild-type and the lrs1
mutant (data not shown). To detect the transcriptional changes that might contribute to this phenotype we constructed additional RNA-seq libraries from wild type and Δlrs1
isolates. Interestingly, the gene expression in both libraries was almost identical (Pearson r>0.99), except for a single operon, the anthranilate dioxygenase operon (antABC
), which was upregulated by more than 2-fold in the mutant strain. This might explain a highly efficient metabolic flux to this pathway, which in turn depleted the production of pyocyanin, due to lack of chorismic acid (Figure S1
); in turn this may provide an explanation for the lack of the color in the Δlrs1
cultures. Strikingly, Δlrs1
also resulted in a significant (over 4-fold) upregulation in the expression of the two PrrF sRNAs, which are regulators of a number of iron-related genes 
The Pseudomonas transcriptome browser
As part of this study, we created a comprehensive graphical transcriptome map of P. aeruginosa
. This map provides a high-resolution view of the Pseudomonas
transcriptome structure and its regulatory sequences. To make this data an accessible resource for further research we generated an online repository and a data-viewer, The Pseudomonas
transcriptome browser, which is available at http://www.weizmann.ac.il/molgen/Sorek/pseudomonas_browser/
. The transcriptome browser allows navigating the transcriptome and genome data, and provides direct links to additional Pseudomonas
In conclusion, the high-resolution analysis of the transcriptome of the opportunistic pathogen P. aeruginosa presented here provides a framework for studying the regulatory mechanisms that allow this common soil organism to become a successful human pathogen in various compromised hosts. In this work, we provide a detailed snapshot of the operon organization, the abundance of coding and non-coding transcripts and their sites of transcription initiation, taken at temperatures encountered most likely in the natural environment and in the human body. Taking advantage of this data, we could significantly expand the number of genes that are under the direct control of LasR, the master quorum-sensing regulator, including two novel sRNAs. In addition to creating a valuable tool for the P. aeruginosa research community, The Pseudomonas transcriptome browser, our work demonstrates an important function for temperature in controlling the expression of many of P. aeruginosa virulence factors. Considering the wide range of reservoirs, including the human body, this detailed transcriptome analysis should provide insights into the interactions between the various mechanisms that direct the assimilation of various signals and directing selective and coordinated gene expression to favor survival of this organism in often challenging environments.