As has been the case for other bacteria, the PhrS RNA and other sRNAs in the opportunistic human pathogen Pseudomonas aeruginosa have been identified by a combination of approaches. PhrS was first annotated (as P20) in a computational screen of intergenic regions for genetic features associated with sRNAs; a search that led to the identification of 17 sRNAs (Livny et al., 2006). PhrS also was detected (as 1887) among the 14 sRNAs identified in another computational screen of intergenic regions for sequences indicative of RNA structure conservation (González et al., 2008). In addition, the sRNA was one of the 11 sRNAs present after co-immunoprecipitation with the RNA-binding protein Hfq (Sonnleitner et al., 2008).
Pseudomonas aeruginosa strains lacking the RNA chaperone Hfq show reduced virulence, which is due to decreased fitness and ability to cope with stresses in the host environment, as well as reduced motility or biofilm formation (Sonnleitner et al., 2003) similar to what has been found for other pathogenic bacteria (reviewed in Chao and Vogel, 2010). Some of the pleiotropic effects of an hfq deletion mutant can, as in E. coli, be attributed to decreased Hfq-dependent synthesis of the sigma factor RpoS. However, RpoS-independent effects, such as altered expression of pyocyanin (PYO), have also been observed (Sonnleitner et al., 2006) and can now, at least in part, be explained by PhrS-dependent regulation (Sonnleitner et al., 2011).
Insights into PhrS function have come from several different lines of experimentation including studies of PhrS expression. Previous work showed that although the stationary-phase induced PhrS RNA binds Hfq in vitro, the 50% reduction of its steady-state levels in an hfq deletion strain could not be attributed to the reduced RNA stability usually associated with the absence of Hfq (Sonnleitner et al., 2008). In the present study, Sonnleiter et al. show that the reduction in PhrS levels is due to an indirect effect of Hfq on PhrS transcription via the oxygen-responsive regulator ANR, which is downregulated in the absence of Hfq. The phrS promoter contains a conserved ANR binding site, and assays of a transcriptional phrS–lacZ reporter gene fusion as well as Northern blot analysis of PhrS levels showed ANR-dependent induction of phrS when cells enter stationary phase and face oxygen limitation. Anaerobic regulation was first observed for an E. coli sRNA, FnrS, whose expression is activated by FNR, the functional orthologue of ANR, and which downregulates aerobic functions and enzymes associated with oxidative stress (Boysen et al., 2010; Durand and Storz, 2010).
The finding that PhrS co-immunoprecipitates with and binds to the Sm-like protein Hfq, which is required for base pairing between sRNAs and mRNAs of limited complementarity in E. coli and Salmonella (reviewed in Waters and Storz, 2009), suggested that the sRNA acts by base pairing with target mRNAs. Proteome analysis upon PhrS overexpression uncovered upregulation of the GroEL chaperone, the OprD outer membrane protein, and a putative periplasmic binding protein (Sonnleitner et al., 2008), while whole genome expression analysis after short-term induction of PhrS revealed downregulation of the terminal cytochrome oxidases CyoAB (Sonnleitner et al., 2011). The most striking result of the transcriptome study, however, was that PhrS induction led to increased levels of genes required for the synthesis of the virulence factor PYO. In fact, an increase in the blue PYO pigment, which can react with molecular oxygen to generate oxidative stress, was evident from a change in the colour of the liquid culture upon 3 h of PhrS overproduction. The genes for PYO synthesis, as well as 25 out of 42 genes induced by PhrS overexpression, are regulated by the common transcription regulator PqsR, indicating the PhrS RNA might directly base pair with and regulate the pqsR mRNA. In P. aeruginosa, PqsR is activated by the co-inducer 2-heptyl-3-hydroxy-4-quinolone (PQS), a cell density signal, whose synthesis in turn is controlled by PqsR-regulated genes. Thus, the PhrS RNA connects an environmental condition (oxygen limitation when cell density increases in batch culture) to a quorum sensing regulatory loop that controls many virulence factors, analogous to how the Vibro Qrr RNAs impact the HapR/LuxR quorum sensing regulatory loops (Lenz et al., 2005). It will be interesting to see whether a phrS deletion strain shows altered virulence and how many other genes are directly activated or repressed by PhrS.
Sonnleitner et al. showed that while pqsR transcript levels were only slightly affected by PhrS, the expression of translational reporter gene fusions was clearly increased upon sRNA induction. This adds PhrS to the growing list of sRNAs that activate translation. Most of the translation activators (S. aureus RNAIII, E. coli DsrA, RprA, RyhB, GlmZ and ArcZ, as well as V. cholerae Qrr1) act by an anti-antisense mechanism (reviewed in Fröhlich and Vogel, 2009). In these examples, as first illustrated for hla-RNAIII (Morfeldt et al., 1995) and rpoS-DsrA (Lease et al., 1998; Majdalani et al., 1998), the ribosome binding site (RBS) of the target mRNA is sequestered in a stable secondary structure in the absence of the sRNA. The activating sRNA functions as a structural competitor and increases translation by opening the inhibitory stem-loop structure, thereby facilitating ribosome access. Other positive acting sRNAs control transcript stability by the processing of 3′ or 5′ ends (E. coli GadY RNA or Clostridium VR-RNA) or the generation of 5′ proximal stabilizing structures to prevent degradation by RNases (Streptococcus FasX RNA) (reviewed in Podkaminski and Vogel, 2010).
In the case of the pqsR mRNA, Sonnleitner et al. identified an uof of 40 amino acids in the 5′ UTR as the primary target of translational activation by PhrS anti-antisense regulation (Fig. 1). Translational reporter gene fusions to uof and pqsR as well as stop mutations within uof showed that pqsR translation is coupled to translation of the uof, which ends immediately upstream of the RBS of pqsR. During aerobic growth, translation of both uof and pqsR is limited by an inhibitory structure that sequesters the RBS of both of the genes. When cells enter stationary phase and face oxygen limitation, ANR induces PhrS transcription, which in turn binds to and opens the inhibitory structure at the uof RBS. The resulting increase in uof translation is coupled to pqsR translation, ultimately leading to increased transcription of the PqsR target genes and elevated PYO and PQS levels. Mutational analysis of both PhrS and the uof–pqsR mRNA confirmed the inhibitory structure, which represses uof and pqsR translation, and revealed that the highly conserved region of PhrS interacts with sequences upstream of the RBS of uof. Previously, E. coli RyhB RNA was the only example of a sRNA targeting an uof. However, in contrast to PhrS activation of pqsR translation, RyhB RNA indirectly inhibits translation of fur by directly inhibiting translational initiation of an uof whose translation is positively coupled to that of fur (Vecerek et al., 2007).
In addition to acting as a base pairing sRNA, PhrS itself encodes a 37-amino-acid protein in the region 5′ to the sequences involved in base pairing. This is reminiscent to what has been found for the E. coli base pairing SgrS RNA, which encodes the 43-amino-acid SgrT protein (Wadler and Vanderpool, 2007), and the S. aureus base pairing RNAIII, which encodes the 26-amino-acid δ-haemolysin protein (Janzon et al., 1989). As for SgrT and δ-haemolysin, translation of the PhrS ORF has been confirmed. However, since the uof product is dispensable for activation of uof–pqsR, the function of the peptide, which is highly conserved in Pseudomonads as well as other species, remains unknown.
Other sRNAs characterized in P. aeruginosa also provide interesting twists on sRNA themes (Table 1). For example, a bioinformatic screen of intergenic regions for properties of the E. coli ryhB gene, namely a Fur binding site followed by a Rho-independent terminator, uncovered duplicated Fur-regulated sRNAs PrrF1 and PrrF2 (Wilderman et al., 2004). Although PrrF1 and PrrF2 are regulated in the same way as E. coli RyhB and target many of the same mRNAs, the sequences of these two sRNAs differ significantly from RyhB. It is intriguing that while the sRNA controlling iron homeostasis is encoded by a single gene in E. coli, it is present as two copies in P. aeruginosa as well as in Salmonella and Yersinia pestis. Moreover, the two genes are located in tandem in P. aeruginosa but are found in separate locations in other Pseudomonads. Recently, the tandem prrF1 and prrF2 genes have been reported to be transcribed as a single PrrH transcript (Oglesby-Sherrouse and Vasil, 2010), but the advantage of a single PrrH transcript compared with the individual PrrF1 and PrrF2 RNAs is not clear.
|sRNA||Length (nucleotides)||Protein bound||Expression||Process regulated||Reference|
|PhrS||212||Hfq||Induced by low oxygen via ANR||Quorum sensing||Sonnleitner et al. (2011)|
|PrrF1||∼110||Hfq||Induced by low iron via Fur derepression||Iron homeostasis||Wilderman et al. (2004)|
|PrrF2||∼110||Hfq||Induced by low iron via Fur derepression||Iron homeostasis||Wilderman et al. (2004)|
|PrrH||∼325||Repressed by iron and haem||Iron and haem homeostasis||Oglesby-Sherrouse and Vasil (2010)|
|RsmY||124||RsmA, Hfq||Induced by multiple signals via the GacS–GacA two-component system||Biofilm formation||Reviewed in Lapouge et al. (2008)|
|RsmZ||∼120||RsmA||Induced by multiple signals via the GacS–GacA two-component system||Biofilm formation||Reviewed in Lapouge et al. (2008)|
|CrcZ||407||Crc||Induced in the presence of poor carbon sources via RpoN and the CbrA–CbrB two-component system||Catabolite repression||Sonnleitner et al. (2009)|
|RgsA||120||Directly regulated by RpoS and indirectly regulated by the GacS–GacA two-component system||Oxidative stress response?||González et al. (2008)|
|PhrD||72||Hfq||Sonnleitner et al. (2008)|
A few sRNA regulators that act by binding to proteins to modify their activity rather than by base pairing have also been found in P. aeruginosa. Two that have been fairly extensively studied in Pseudomonas species are RsmY and RsmZ, members of the CsrB/C, RsmY/Z family of sRNAs, which sequester the CsrA/RsmA family of RNA binding proteins by carrying multiple repeats of the GGA recognition sequence (reviewed in Lapouge et al., 2008). As in other organisms, P. aeruginosa RsmY and RsmZ are induced by a two-component system (GacS/GacA) in response to multiple environmental stimuli to modulate a wide range of physiological responses including virulence, quorum sensing, biofilm formation and secretion. Interestingly, Hfq also impacts quorum sensing by binding and stabilizing the RsmY RNA (Sonnleitner et al., 2006). Another protein binding sRNA characterized in P. aeruginosa is CrcZ, whose expression is induced by RpoN and the CbrA–CbrB two-component system (Sonnleitner et al., 2009). This RNA binds to Crc, which brings about catabolite repression by binding to CA-rich sequences in mRNAs such as amiE. CrcZ contains five CA motifs and is able to titrate the Crc protein away from the amiE mRNA in vitro, a mechanism reminiscent of CsrB/C and RsmY/Z RNAs.
The P. aeruginosa sRNAs exhibiting distinct combinations of features of previously characterized sRNAs are opening up interesting mechanistic questions. For example, in the case of PhrS, does sRNA targeting of the uof rather than sequences proximal to pqsR have consequences? It will be interesting to see how many additional RNAs target uofs to activate or repress gene expression. In addition, do both the 37-amino-acid protein encoded by PhrS and the 40-amino-acid protein encoded by the uof upstream of pqsR have functions, and if so, are the functions connected to or distinct from the regulatory function of PhrS? Moreover, how many other mRNAs encoding small proteins also act as regulatory sRNAs? In the case of CrcZ, how is the sRNA bound by the Crc protein? In addition, how many other sRNAs carrying sequence repeats to titrate RNA binding proteins will be found? For the 50% of bacteria that lack Hfq, such as the human pathogens Helicobacter pylori or Chlamydia trachomatis, do sRNAs base pair in the absence of RNA chaperone proteins or do new RNA binding proteins remain to be identified? It is striking that similar physiological themes continue to be connected to sRNA regulation in a wide range of bacteria including iron, carbon and amino acid metabolism, membrane remodelling, anaerobic regulation and quorum sensing. Is this because these physiological responses most benefit from regulation by sRNAs or because only a limited number of sRNAs have been characterized? All of these questions emphasize the importance of continued study of sRNAs in a wide range of bacteria. The most exciting aspect of the many ongoing deep sequencing projects is that a treasure trove of bacterial sRNAs of unknown functions is coming available. Undoubtedly, still other twists on known mechanisms, as shown for P. aeruginosa PhrS, as well as completely new mechanisms of regulation will become evident as more sRNAs are studied in all bacteria.