SEARCH

SEARCH BY CITATION

Keywords:

  • regulation;
  • RNA duplex;
  • RNA/protein interaction;
  • RNA/RNA interaction;
  • small non-coding RNAs

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Co-ordination of regulatory effects mediated by ncRNAs
  5. Adaptation processes controlled by ncRNAs
  6. Aspects and consequences of ncRNA/mRNA interactions
  7. Concluding remarks
  8. Acknowledgements
  9. Funding
  10. References

Regulatory ncRNAs (non-coding RNAs) adjust bacterial physiology in response to environmental cues. ncRNAs can base-pair to mRNAs and change their translation efficiency and/or their stability, or they can bind to proteins and modulate their activity. ncRNAs have been discovered in several species throughout the bacterial kingdom. This review illustrates the diversity of physiological processes and molecular mechanisms where ncRNAs are key regulators.


Abbreviations used:
aa

amino acid

Csr

carbon storage regulator

Eno

enolase

Fur

ferric uptake regulator

G6P

glucose 6-phosphate

ncRNA

non-coding RNA

nt

nucleotide

OMP

outer membrane protein

PTS

phosphoenolpyruvate phosphotransferase systems

TnaA

tryptophanase

QS

quorum sensing

RNAP

RNA polymerase

SD

Shine—Dalgarno sequence

TIR

translation initiation region

5′UTR

5′untranslated region

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Co-ordination of regulatory effects mediated by ncRNAs
  5. Adaptation processes controlled by ncRNAs
  6. Aspects and consequences of ncRNA/mRNA interactions
  7. Concluding remarks
  8. Acknowledgements
  9. Funding
  10. References

Regulatory ncRNAs (non-coding RNAs) are crucial regulators that enable the cell to adjust its physiology to environmental changes. NcRNAs exert their effects following the same principles in all organisms, and may impact on all steps of gene expression (Storz et al., 2005). In bacteria, functional ncRNAs can be generated via processing (e.g. RnpB, SsrA or SsrS) or as primary transcripts, the latter representing the most abundant class of known ncRNAs. Regulatory ncRNAs may also form chimaeras with protein-encoding sequences that they control, e.g. riboswitches or 5′UTRs (5′ untranslated regions) of messenger RNAs. This review is focused on the diversity and the importance of physiological processes controlled by ncRNAs and the diversity of their mechanisms of action (Figure 1). Excellent reviews refer to other regulatory RNA elements such as riboswitches, 5′UTRs or ncRNAs encoded by extrachromosomal genetic elements (transposons, phages and plasmids) (Brantl, 2007; Coppins et al., 2007; Weaver, 2007; Winkler and Breaker, 2005).

image

Figure 1. Effects of chromosomally encoded ncRNAs on their mRNA or protein targets

As well as examples provided in the text, additional ones are ncRNAs with a mode of action that is known or speculated. Notations are: ncRNA/target; +, RNA duplex degradation; −, no degradation of the RNA duplex; (d), degradation not demonstrated; (h), involvement of Hfq not demonstrated; (p), pairing not demonstrated; (dh), neither Hfq involvement nor degradation demonstrated; (he), Hfq involvement not demonstrated and effect predicted. An, Anamoeba sp.; Bb, Borrelia burgdorferi; Bs, B. subtilis; Ca, Clostridium acetobutylicum; Col, ncRNA from the plasmid ColE1, target from E. coli host (tryptophanase, TnaA); Ct, Chlamydia trachomatis; Ec, E. coli; Ent, enterobacteriaceae; Lm, L. monocytogenes; Pa, Ps. aeruginosa; Sa, S. aureus; St, Salmonella Typhimurium; Sy, Synechocystis Sp.; Ub, ubiquitous in all bacteria; Vc, V. cholerae. MICs/OMPs*, ncRNAs acting on envelope integrity; these are: MicF, MicC, MicA, MicX, RybB, RydC, RseX, OmrA, OmrB, InvR and IpeX (note that IpeX does not require Hfq to act on ompC mRNA), (see Table* on the right side of the panel; references available in Guillier et al., 2006; Vogel and Papenfort, 2006; Bossi and Figueroa-Bossi, 2007; Davis and Waldor, 2007). References are: DsrA- and RprA/rpoS (Majdalani et al., 1998, 2001); GlmY- and GlmZ/glmS (Kalamorz et al., 2007; Urban et al., 2007); GcvB/gltI/argT/ilvJ/dppA/oppA (Sharma et al., 2007); OxyS/rpoS (Zhang et al., 1998); OxyS/fhlA (Altuvia et al., 1998); DsrA/hns* (Lease and Belfort, 2000), *DsrA is supposed to pair with both the TIR and the 3′ end of hns mRNA; DicF/ftsZ (Faubladier et al., 1990); p3RNA/glnA (Fierro-Monti et al., 1992); Spot42/galK (Møller et al., 2002a, 2002a); GadY/gadX (Opdyke et al., 2004); RldD/ldrD (Kawano et al., 2002); Rli ncRNAs (Mandin et al., 2007); IthA/hctA (Grieshaber et al., 2006); RnpB/RnpA, tmRNA/SmpB and 4.5S/Ffh (Pichon and Felden, 2007); for other ncRNAs, references are mentioned in the text.

Download figure to PowerPoint

Co-ordination of regulatory effects mediated by ncRNAs

  1. Top of page
  2. Abstract
  3. Introduction
  4. Co-ordination of regulatory effects mediated by ncRNAs
  5. Adaptation processes controlled by ncRNAs
  6. Aspects and consequences of ncRNA/mRNA interactions
  7. Concluding remarks
  8. Acknowledgements
  9. Funding
  10. References

Regulatory ncRNAs directly affect the expression of effector and regulatory proteins. They exert their effects by two modes of action, depending on the nature of the target molecule (Figure 1). ncRNAs can base-pair to mRNAs to form RNA duplexes that modify translation efficiency and/or stability of mRNAs. In Escherichia coli, the Hfq protein is generally involved in facilitating the formation of duplexes (Aiba, 2007; Brennan and Link, 2007). This class of ncRNA comprises numerous characterized examples (antisense RNAs). ncRNA of the second class bind to proteins and modify their activities.

The flexibility of the cell response to environmental changes necessitates a rigorous co-ordination between ncRNAs and protein regulators, and interactions between regulatory cascades to provide the cell with the ability to adapt to various signals simultaneously. The co-ordination of ncRNA-mediated effects occurs mainly by regulating the abundance of ncRNAs via synthesis (transcription) and/or stability (Repoila et al., 2003; Romby et al., 2006). Several examples of ncRNAs described in this review are known to be at the crossroads between regulons.

Adaptation processes controlled by ncRNAs

  1. Top of page
  2. Abstract
  3. Introduction
  4. Co-ordination of regulatory effects mediated by ncRNAs
  5. Adaptation processes controlled by ncRNAs
  6. Aspects and consequences of ncRNA/mRNA interactions
  7. Concluding remarks
  8. Acknowledgements
  9. Funding
  10. References

Transcription reprogramming

Bacterial RNAP (RNA polymerase) exists as two forms, the core enzyme (consisting of two ‘α’, one ‘β’ and one ‘β′’ subunit) and the holoenzyme (the core plus the σ subunit). The σ subunit confers promoter selectivity to RNAP (RNA polymerase). The different σ factors encoded by bacterial genomes each favour transcription of particular sets of genes. In E. coli, the core RNAP bound to the Eσ70 (housekeeping sigma factor σ70) ensures the main transcriptional activity in the cell (Mooney et al., 2005; Jin and Cabrera, 2006).

In E. coli, SsrS (or 6S) is a highly structured 184 nt long ncRNA that resembles a melted DNA promoter. SsrS forms a specific complex with Eσ70 via interactions with σ70, β and β′ subunits, and stabilizes the association between σ70 and core RNAP. SsrS was shown to selectively modulate σ70 promoter usage by Eσ70, thus defining a subset of genes transcriptionally controlled by SsrS (Wassarman, 2007; Cavanagh et al., 2008). Biogenesis of SsrS relies on co-ordinated transcription and maturation processes (Kim and Lee, 2004). SsrS accumulates throughout growth and reaches elevated levels when bacteria enter stationary phase, at which time 80–90% of Eσ70 is sequestered in an SsrS—Eσ70 complex. An elegant series of in vitro and in vivo experiments demonstrated that at high intracellular nucleotide concentrations, i.e. conditions mimicking outgrowth from stationary phase, SsrS is used as a template by Eσ70 to transcribe a 14–20 nt long RNA (pRNA). Upon pRNA synthesis, the Eσ70–SsrS—pRNA complex dissociates. Eσ70 is released and the SsrS—pRNA complex is subsequently degraded (Gildehaus et al., 2007; Wassarman, and Seacker, 2006). Figure 2 describes the current understanding of the Eσ70-transcription reprogramming mediated by SsrS.

image

Figure 2. Transcription reprogramming by SsrS

During exponential growth, the RNAP is able to transcribe all the Eσ70-dependent promoters. In parallel, SsrS accumulates and binds tightly to Eσ70. At the end of the exponential growth, more than 80% of Eσ70 is bound to SsrS. In stationary phase, the complex Eσ70-SsrS is able to use only a certain type of Eσ70-dependent promoter (‘SsrS-non sensitive’), seperate to the SsrS-sensitive ones. Thus, SsrS provokes a reprogramming of the main transcriptional activity. During stationary phase, levels of NTPs are low. At the outgrowth from stationary phase, NTP levels increase to a threshold compatible with the use of the ncRNA SsrS as a template by Eσ70; pRNA is synthesized, provoking the dissociation of the Eσ70–SsrS—pRNA complex. SsrS—pRNA is degraded and Eσ70 re-uses all the Eσ70-dependent promoters expressed during the exponential growth.

Download figure to PowerPoint

Thus the ncRNA SsrS encodes its own regulatory ncRNA, pRNA, and adjusts the bulk of Eσ70-dependent transcription to appropriate metabolic resources by sensing nucleotide concentration. By doing so, SsrS indirectly affects expression of genes transcribed by other σ factors that compete with σ70 for core RNAP (Wassarman, 2007, and references cited within). SsrS is widespread in bacteria and such a mechanism is certainly ubiquitous.

Carbon metabolism

Storage and flux of carbon

In nature, bacteria encounter alternate periods of abundance and famine, imposing major switches in gene expression and co-ordination of global regulatory networks. Among those, the Csr (carbon storage regulator) system ensures connections between carbon metabolism and diverse traits in many Gram-negative bacteria, including motility, QS (quorum sensing), biofilm development and virulence. CsrA, an RNA binding protein, is the central component of the Csr system. CsrA is called RsmA and RsmE in Pseudomonas and Erwinia species respectively, where it plays a crucial role in virulence. In E. coli, CsrA represses metabolic processes related to stationary phase (such as glycogen synthesis and glucogenesis), and stimulates pathways during the exponential growth phase (e.g. glycolysis, motility, chemotaxis and use of particular carbon sources). CsrA binds to GGA motifs contained in, or in the vicinity of, TIRs (translation initiation regions) of mRNA targets, and stabilizes or destabilizes transcripts (Babitzke and Romeo, 2007; Baker et al., 2007; Bejerano-Sagie and Xavier, 2007; Toledo-Arana et al., 2007).

In E. coli, CsrB (369 nt) and CsrC (242 nt) ncRNAs bind to and antagonize the effects of the CsrA protein. These ncRNAs carry repeated GGA motifs that mimic sequences bound by CsrA. The sequence, the length and the number of repeats contained in Csr ncRNAs vary from one species to another. For instance, in pathogenic species only one Csr-like ncRNA is present in Erwinia carotovora (RsmB), two in Pseudomonas aeruginosa (RsmY and Z), and three in Vibrio cholerae (CsrB, C and D) and in Pseudomonas fluorescens (RsmX, Y and Z), (Lapouge et al., 2008).

Expression of csrA, csrB and csrC is regulated by multiple networks that integrate various signals and an autoregulatory loop (Figure 3). CsrA activates the two-component system BarA/UvrY, where BarA is the sensor protein and UvrY the cognate response regulator. This activation stimulates CsrB and CsrC synthesis, which subsequently antagonizes CsrA activity and provides negative feedback (Babitzke and Romeo, 2007; Lapouge et al., 2008). Other signal transmission pathways are possibly involved in this regulatory loop. For instance, BarA activity depends on pH and carbon sources, and UvrY might be activated by other kinases (Babitzke and Romeo, 2007). Another level of input is provided by the stability of CsrB and CsrC ncRNAs. In E. coli, the CsrD protein specifically targets CsrB and CsrC for degradation by RNase E, the essential endoribonuclease of the degradosome (Suzuki et al., 2006; Carpousis, 2007). The control of CsrB and CsrC degradation via CsrD is still not known but appears as an important issue in understanding gene expression reprogramming orchestrated by CsrA.

image

Figure 3. CsrA/CsrB and CsrC regulatory loop in E. coli

CsrA activates the expression of the two-component system BarA/UvrY that subsequently activates the expression of CsrB and CsrC. CsrB and CsrC ncRNAs bind to CsrA and counteract its translational regulatory effects. Degradation of CsrB and CsrC is controlled by CsrD. mRNAs bound by CsrA are marked in blue. Examples of biological processes controlled by the regulatory loop CsrA/CsrB/CsrC are shown in the grey boxes.

Download figure to PowerPoint

Sugar utilization and metabolic intermediates

Bacterial carbohydrate PTS (phosphoenolpyruvate phosphotransferase systems) mediate the uptake of sugars and their concomitant phosphorylation (Deutscher et al., 2006). The main PTS importing glucose contains the transporter PtsG (or EIICBGlc), encoded by the gene ptsG. When glucose is present in the medium, the activity of PtsG generates G6P (glucose 6-phosphate), which is toxic if it accumulates. In E. coli, the ncRNA SgrS (227 nt) negatively regulates ptsG and restricts the influx of G6P. SgrS pairs with the TIR of ptsG mRNA in an Hfq-dependent manner, resulting in translational inhibition of ptsG and certainly a subsequent degradation of the RNA duplex by RNase E (Kawamoto et al., 2006; Morita et al., 2005; Morita et al., 2006; Vanderpool and Gottesman, 2004). The base-pairing region SgrS/ptsG is necessary but not sufficient for the translational control of ptsG and the degradation of the RNA duplex. In addition, SgrS-mediated control requires the localization of the ptsG mRNA at the inner membrane coupled with membrane insertion of the nascent peptide (Kawamoto et al., 2005). Since (i) SgrS/ptsG base-pairing inhibits translation and promotes degradation, and (ii) translation of at least the signal peptide is necessary for the SgrS effect, this implies that ptsG translation starts, and then SgrS and free ribosomes compete for binding to the mRNA (Kawamoto et al., 2005; Vanderpool and Gottesman, 2005). Rapid degradation of the repressed mRNA requires a functional interaction between RNase E and Eno (enolase), two partners of the degradosome (Carpousis, 2007). RNase E-dependent degradation of the SgrS/ptsG duplex is modulated by Eno in response to G6P levels. The signal sensed by Eno remains unknown, but this enzyme is a crucial ‘sensor/regulator’ for the stability of ptsG mRNA (Morita et al., 2004).

Transcriptional control of sgrS is another factor affecting ptsG mRNA stability via the abundance of SgrS. Two transcriptional regulators of sgrS have been characterized, SgrR and Mlc. SgrR, encoded by the gene divergent to sgrS, activates sgrS in response to G6P. Mlc, a glucose-specific repressor protein, represses sgrS independently of SgrR (Vanderpool and Gottesman, 2004, 2007).

Arginine catabolism, nitrogen and carbon sources

In Bacillus subtilis, the use of arginine (ornithine, isoleucine and valine) as carbon and/or nitrogen source requires the products of rocABC and rocDEF operons encoding the catabolic and transport enzymes. Transcription of roc operons is strictly dependent on σL, a transcription initiation factor homologous to σ54 of E. coli. RocR and AhrC are transcriptional activators of rocABC and rocDEF, and are both required for σL-dependent transcription initiation. Levels of RocR and AhrC are tightly controlled to ensure appropriate synthesis of enzymes involved in the metabolic response: RocR negatively regulates its own synthesis and the expression of AhrC expression is controlled by the ncRNA SR1 (205 nt) (Gardan et al., 1997; Heidrich et al., 2006). In an elegant study, SR1 was shown to base-pair to ahrC mRNA by patches of complementary regions within the coding sequence of the mRNA, far downstream of the TIR (80 nt). The interaction between SR1 and ahrC promotes conformational changes in the TIR of ahrC and prevents the assembly of a translation initiation complex. The formation of the SR1/ahrC duplex does not require Hfq and is not a target for degradation (Heidrich et al., 2006, 2007).

SR1 expression is induced during entry into stationary phase and is repressed by the use of sugars as carbon source. Both these controls operate at the transcriptional level and depend on the two carbon catabolite repressors CcpA and CcpN (Licht et al., 2005). In addition, L-arginine stimulates sr1 transcription, indicating a retrocontrol of the expression of rocABC and rocDEF operons via the negative effect of SR1 on the positive transcriptional regulator AhrC (Heidrich et al., 2006). The actors involved in sr1 L-arginine-dependent transcription are not known.

Iron homoeostasis

Iron is a vital metal for most organisms, since it is a cofactor of a large number of enzymes involved in essential biological processes (e.g. photosynthesis, respiration, DNA biosynthesis). However, in the presence of oxygen, iron is detrimental since it catalyses the formation of hydroxyl radicals and ion superoxides that cause oxidative damage (Touati, 2000; Imlay, 2003). Therefore, intracellular iron levels must be rigorously controlled to satisfy both physiological needs and to avoid toxic effects. Bacteria respond to fluctuations in iron concentration by co-ordinating the expression of genes encoding proteins involved in iron transport and storage, and iron-dependent enzymes. Usually the expression of these genes is connected to other metal metabolism pathways and to the oxidative stress response (Andrews et al., 2003). In bacteria, Fur (ferric uptake regulator) is a key transcriptional repressor that controls iron metabolism and the use of iron as cofactor (Fe2+) (Hantke, 2001; Moore and Helmann, 2005). At low intracellular iron concentrations, Fur is inactive and iron-uptake genes are expressed. Above a threshold, Fur represses the expression of these genes and activates the expression of genes encoding iron-containing and iron-storage proteins. In E. coli and related species, this activation relies on two consecutive repression steps: (i) Fur represses the transcription initiation of ryhB, encoding the ncRNA RyhB (90 nt); and (ii) via an Hfq-dependent process, RyhB base-pairs to mRNAs of iron-containing proteins, thereby blocking their translation and promoting their degradation (Massé et al., 2007, and references therein). More than 56 genes might be directly repressed by RyhB, including genes involved in the TCA cycle, glycolysis, oxidative stress, iron—sulfur cluster formation, and aerobic and anaerobic respiration. Thus, RyhB increases intracellular iron by limiting expression of iron-using proteins, and thus ensuring its availability for essential iron-requiring proteins (Jacques et al., 2006).

The role of RyhB is not limited to its involvement in the global control exerted by Fur. RyhB also provides feedback by acting on fur mRNA translation (Vecerek et al., 2007). The fur-encoding sequence is cotranscribed and cotranslated with an upstream small open reading frame (28 aa), uof. RyhB pairs to the uof-fur mRNA in an Hfq-dependent manner within the TIR of uof, blocking translation and reducing the abundance of the bicistronic mRNA. Thus RyhB exerts a post-transcriptional control on fur and diminishes the abundance of Fur when iron is scarce, i.e. when Fur is not needed (Vecerek et al., 2007). The Fur/RyhB system relies on a negative feedback regulatory loop, since Fur and RyhB mutually repress each other. The negative feedback loop ‘Fur/RyhB’ certainly exists in other Gram-negative bacteria, since ryhB orthologues have been found, e.g. in Ps. aeruginosa (two copies of ryhB named prrF1 and prrF2), Vibrio, Shigella and Neisseria species (in the latter called nrrF), (Massé et al., 2007; Vecerek et al., 2007). In two species, Synechocystis and Anamoeba, a cis-antisense ncRNA directly controls translation of fur mRNA, IsiR and α-fur, respectively (Duhring et al., 2006; Hernandez et al., 2006).

RyhB also activates the expression of other genes involved in iron homoeostasis. For instance, RyhB interacts with the mRNA of shiA, a gene encoding a transporter of shikimate, a compound required for siderophore synthesis. RyhB pairing stabilizes shiA mRNA and activates its translation by releasing the SD (Shine—Dalgarno) sequence that otherwise is embedded in a secondary structure (Massé et al., 2007; Prévost et al., 2007).

Envelope homoeostasis

The bacterial envelope provides structural integrity to the cell, protects it from internal turgor pressure (mechanical barrier), and is responsible for vital exchanges between the cytoplasm and the outside environment (selective barrier). To survive in a changing environment, bacteria must constantly adjust the nature and abundance of envelope components (Scott and Barnett, 2006; Bos et al., 2007). In Gram-negative bacteria, the cell envelope is composed of two phospholipid membranes separated by a periplasmic space containing peptidoglycan. OMPs (outer membrane proteins) span the external membrane and function as a selective barrier for hydrophilic solutes. The nature and the abundance of OMPs vary in response to environmental changes and result from the interplay of regulatory networks that ensure envelope homoeostasis. Among these networks, two-component systems, such as EnvZ-OmpR or CpxA-CpxR, and the extracytoplasmic transcription initiation factor σE, are essential players, which modulate the expression of many genes encoding envelope components (e.g. OMPs, phospholipid and lipopolysaccharide synthesis, and proteins involved in the assembly of OMPs and lipopolysaccharides) (Rowley et al., 2006; Rhodius et al., 2006; Ruiz et al., 2006; Bos et al., 2007). Within this complex picture, at least twelve ncRNAs have been shown to downregulate the expression of OMPs in E. coli and related species, thus preventing a build-up of unassembled or misfolded OMPs in the envelope. In a few cases, the mode of action is not entirely clear, but all these ncRNAs probably act in the same fashion: they inhibit protein synthesis by base-pairing to the TIR of their mRNA target(s) in an Hfq-dependent manner followed by subsequent degradation of both RNAs (see Figure 1). Such a negative control would relieve internal or external envelope stresses via activation of the σE-pathway, which is connected to other envelope stress response networks (EnvZ-OmpR; σH, CpxA-CpxR), (Figure 4). Several excellent reviews and original papers have been published in the last two years on the control of envelope integrity by ncRNAs (such as Udekwu et al., 2005; Douchin et al., 2006; Guillier et al., 2006; Papamichail and Delihas, 2006; Vogel and Papenfort, 2006; Davis and Waldor, 2007; Valentin-Hansen et al., 2007).

image

Figure 4. Current model of the envelope integrity control by ncRNAs in Enterobacteriaceae

Environmental signals and stresses sensed directly by σE or by other envelope regulatory systems (e.g. EnvZ-OmpR; σH; CpxA-CpxR) lead to the activation of the σE regulon and synthesis of OMPs. ncRNAs repress the expression of OMPs. So far, only ncRNAs under the control of σE or the CpxA-CpxR system are known.

Download figure to PowerPoint

Toxin and antitoxin ncRNAs

A TA (toxin—antitoxin) module consists of a gene encoding a toxin (a small protein with bacteriostatic or bacteriocidal properties), and a gene encoding an antitoxin (a protein or a ncRNA) counteracting the effect of the toxin. TA modules are widespread in bacterial genomes (Pandey and Gerdes, 2005). The biological significance of TA systems is debated but there is evidence for their role in stress physiology and quality control of gene expression (Gerdes et al., 2005; Magnuson, 2007). Bacterial RNA antitoxins have been reviewed in Gerdes and Wagner (2007). Here we present four recently described systems.

SOS-induced toxins in E. coli

The IstR-TisB module

istR and tisB are two non-overlapping, divergently transcribed genes. istR encodes two ncRNAs, IstR-1 (75 nt) and IstR-2 (140 nt), and tisB encodes a 29 aa SOS-induced toxin (TisB) (Vogel et al., 2004). IstR-1 and IstR-2 originate from different promoters and share a transcriptional terminator. IstR-1 is constitutively expressed and is the antitoxin of TisB. In normal growth conditions, the level of tisB mRNA is very low and IstR-1 is sufficient to titrate tisB. ItsR-1 pairs to the 5′UTR of tisB mRNA more than 80 nt upstream of the SD and competes with ribosomes for an upstream ribosome entry site (‘standby’ model) (Darfeuille et al., 2007, Unoson and Wagner, 2007). The RNA duplex ItsR-1/tisB forms independently of Hfq and is degraded by RNase III. Under SOS-inducing conditions, the level of tisB increases, the pool of existing ItsR-1 is consumed and tisB is translated, then bacterial growth stops due to the synthesis of the toxin. Then the constitutive expression of IstR-1 permits growth to restart when the SOS stress signal stops (Vogel et al., 2004). The reason why TisB is toxic is not entirely clear, but it has been shown that TisB inserts into the inner membrane, which results in a decreased pool of ATP (Unoson and Wagner, 2008). The role of IstR-2 is currently unknown.

The SymR-SymE module

symR and symE are overlapping and divergently transcribed genes, encoding an ncRNA, SymR (77 nt), and an SOS-induced toxin, SymE (113 aa), respectively. SymR is highly stable and present in 10-fold excess compared to symE mRNA. It represses symE translation by base-pairing (SymR is entirely complementary to symE). It has been suggested that the SymR/symE duplex would form in an Hfq-independent fashion and would be degraded (Kawano et al., 2007). Ectopic expression of SymE decreases protein synthesis and promotes the cleavage of several mRNAs and ncRNAs. SymE co-purifies with ribosomes, as observed for other mRNA-cleaving toxins (e.g. RelE), suggesting that SymE may act as an endoribonuclase. However, further biochemical evidences will be needed to establish such an activity for SymE.

Intriguingly in E. coli, other toxin-encoding genes are induced during the SOS response, e.g. hokE or the above example, tisB (Fernandez De Henestrosa et al., 2000; Vogel et al., 2004). In the latter case, induction of TisB, as for SymE, blocks translation and leads to RNA cleavage, suggesting a role of SOS-induced toxin in surveillance of RNA quality after SOS damage or other stresses (Kawano et al., 2007; G.H.R. Wagner, personal communication).

The lytic-toxin TxpA of B. subtilis

ratA and txpA are two convergent genes, expressing a 222 nt ncRNA, RatA, and a 59 aa toxic protein mediating cell lysis respectively. The 3′ ends of RatA and txpA mRNA overlap by 75 nt. The perfect complementarity between the two transcripts allows for duplex formation, inhibiting TxpA production. The duplex forms independently of Hfq and is a substrate for degradation. The txpA-ratA module is located in the skin region of the B. subtilis chromosome. skin is excised during spore formation to generate the functional coding sequence of σK, a specific regulatory protein of the mother-cell. The mother-cell is an ultimate stage of cell differentiation during spore formation. It has been suggested that the presence of the txpA-ratA module in the skin region would avoid premature cell differentiation (Silvaggi et al., 2005).

Rcd, a plasmid-encoded ncRNA toxin targeting a host enzyme

Rcd is a peculiar example of ncRNA in E. coli: (i) it is encoded by a plasmid and not the chromosome; (ii) it does not belong to any indexed TA module family; and (iii) it behaves as a toxin and not as an antitoxin.

The stable maintenance of a multicopy plasmid can be compromised by natural dimerization (or multimerization) due to homologous recombination. Dimer resolution of the E. coli plasmid ColE1 is ensured by a complex system involving two recombination cer sites of the multimer and proteins encoded by the host chromosome (XerC, XerD, ArgR and PepA), (Summers, 1998). During the recombination process resolving the dimer, the Rcd ncRNA (70 nt) is transcribed from a promoter located at the cer site (Sharpe et al., 1999). Rcd is a crucial player in ColE1 stability since it delays cell division until completion of multimer resolution (Balding et al., 2006). To do so, the ncRNA Rcd binds to the host enzyme TnaA (tryptophanase) and increases its affinity for tryptophane, leading to an increased production of intracellular indole (a product of tryptophane conversion by TnaA). The burst of indole slows down cell division and enables plasmid resolution (Chant and Summers, 2007).

Co-ordinators of pathogenesis

During infection, pathogenic bacteria adapt to the changing environment they encounter in the host by appropriate temporal expression of their virulence genes. The co-ordination of stress response programmes and virulence gene expression relies on shared regulators. Several ncRNAs co-ordinating stress response and virulence have been described (reviewed in Geissmann et al., 2006; Romby et al., 2006; Toledo-Arana et al., 2007). We will describe two well-documented examples of pathogens using ncRNAs as crucial co-ordinators of their virulence traits.

The RNAIII paradigm in Staphylococcus

Staphylococcus aureus is a major cause of nosocomial infections and toxin-mediated diseases. Virulence traits of S. aureus are under a temporal control orchestrated by the agr system consisting of two divergent transcription units, encoding RNAII and RNAIII. The RNAII operon encodes an autoinducer QS system that activates the transcription of RNAIII (514 nt), the effector molecule of the agr system, essential for the virulence of numerous S. aureus isolates (Novick, 2003). RNAIII encodes δ-haemolysin (hld), but is also an ncRNA that targets at least five mRNAs encoding virulence factors: hla (α-haemolysin), spa (cell-surface protein A), sa1000 (a fibrinogen-binding protein), sa2353 (secretory antigen precursor) and rot (a pleiotropic transcription factor of many genes involved in virulence and general metabolism) (Morfeldt et al., 1995; Saïd-Salim et al., 2003; Huntzinger et al., 2005; Geisinger et al., 2006; Boisset et al., 2007). The 5′end of RNAIII promotes the translation of hla mRNA (α-haemolysin) by pairing to the 5′UTR and competing directly with an intramolecular RNA secondary structure that sequesters the SD (Novick et al., 1993; Morfeldt et al., 1995). In contrast, certain stem-loops within the 3′ domain of RNAIII and/or the internal hairpin-loop 7 (depending on the mRNA target) base-pair to TIRs of spa, sa1000, sa2353 and rot mRNAs and blocks their translation. These RNA duplexes form independently of Hfq and induce degradation initiated by RNase III (Huntzinger et al., 2005; Geisinger et al., 2006; Boisset et al., 2007). By acting on these different targets, the expression of RNAIII is a key player in the temporal control of virulence genes, reflecting the infection process of S. aureus (Figure 5). At low cell density, RNAIII expression is low, bacteria undergo a stealth colonization of host tissues by expressing adhesins (SA1000) and proteins that hide the pathogen from host defenses (Protein A, SA2353). Then, when cells reach a certain density, the QS control operating on the agr system triggers a collective bacterial response: RNAIII is induced, turning off adhesins and switching on haemolysins and degradative enzymes (e.g. α- and δ-haemolysins) responsible for tissue destruction in the host and further progression of the bacterial colonization. In parallel, RNAIII co-ordinates metabolic switches by acting on rot mRNA (Saïd-Salim et al., 2003).

image

Figure 5. Temporal control of virulence traits by RNAIII

At low cell density (e.g. exponential phase), the agr system is not induced: RNAIII levels are low and mRNA targets enabling tissue colonization are expressed (sa1000, spa, sa2353 and rot). In these conditions bacteria are silent for host defences. At high cell density (e.g. stationary phase), the agr system is autoinduced via QS, RNAIII levels increase and inhibit translation of mRNA targets (i.e. sa1000, sa2352, spa and rot). Meanwhile, RNAIII that already encodes δ-haemolysin also activates translation of α-haemolysin. The induction of the agr system via QS allows the response of the entire S. aureus community. Such response leads to a local tissue destruction of the host and enables further colonization by the pathogen.

Download figure to PowerPoint

An arsenal of ncRNAs in Vibrio

Vibrio cholerae is a human pathogen responsible for food- and water-associated infections. In Vibrio species, QS systems and ncRNAs form an intricate network that regulates biofilm formation, bioluminescence and virulence traits. This network transduces multiple signals to a response regulator, LuxO, via a phosphorylation cascade that uses a relay protein, LuxU. LuxU is directly connected by phosphorylation to at least two sensor proteins, CqsS and LuxP-Q. In addition, LuxO is indirectly connected to the Csr system (orthologous to the Csr system of E. coli), consisting of three ncRNAs (CsrB-C-D) and the VarS-VarA system (equivalent to BarA-UvrY; see above). Phosphorylated LuxO is required for the σ54-dependent transcription of four ncRNAs, Qrr1–4 (96 to 108 nt). Qrr ncRNAs base-pair to the TIR of hapR mRNA in an Hfq-dependent manner and destabilize the transcript. hapR mRNA encodes HapR, a master transcription regulator of many genes, including virulence genes that it represses. In addition, Qrr1–4 activate translation of vca0939 mRNA by an Hfq-dependent base-pairing mechanism, most likely by destabilizing a secondary structure that prevents ribosome access to the SD. vca0939 encodes a GGDEF domain-containing protein characteristic of those that synthesize the intracellular messenger molecule cyclic di-GMP (Lenz et al., 2004, 2005; Waters and Bassler, 2006; Hammer and Bassler, 2007; Tu and Bassler, 2007).

Aspects and consequences of ncRNA/mRNA interactions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Co-ordination of regulatory effects mediated by ncRNAs
  5. Adaptation processes controlled by ncRNAs
  6. Aspects and consequences of ncRNA/mRNA interactions
  7. Concluding remarks
  8. Acknowledgements
  9. Funding
  10. References

The majority of ncRNAs thus far characterized exert their control of gene expression through antisense mechanisms, i.e. by base-pairing to mRNAs (see Figures 1 and 6). In most cases, they are trans-encoded, i.e., the genes encoding the ncRNA and the target mRNA do not overlap. Typically, ncRNA/mRNA hybrids involve short stretches of sequence complementarity between partners, including non-canonical base-pairs (e.g. G:U pairs) and bulges (Figure 6A). Interactions between ncRNAs and their cognate mRNAs are remarkably efficient, since a limited number of nucleotides on the mRNA are targeted by the ncRNA to form the duplex. For instance, 9 nt are involved in RyhB targeting of sodB mRNA, and only 6 nt on ptsG appear to be critical for the effect of SrgS (Geissmann and Touati, 2004; Kawamoto et al., 2006). Such interactions are facilitated by secondary structures of RNAs and originate in single stranded regions, as demonstrated for the formation of RNAIII/sa1000, spa and rot mRNAs duplexes. In at least two cases, RNAIII/rot and OxyS/fhlA, interactions do not involve an extended duplex but remain as two short loop—loop base-pairing complexes (Argaman and Altuvia, 2000; Boisset et al., 2007). In theory, these features enable an ncRNA to regulate multiple mRNA targets (e.g. in the case of RNAIII, MicA, RyhB, RybB and GcvB); however, this feature complicates mRNA target predictions directly from ncRNA sequence (Tjaden et al., 2006; Wagner and Darfeuille, 2006; Mandin et al., 2007; Vogel and Wagner, 2007).

image

Figure 6. General features and consequences of the ncRNA/mRNA interaction

(A) Different regions of the mRNA that can be targeted by an ncRNA. The portion of the ncRNA sequence base-pairing to the mRNA is symbolized in red and named the ‘base-pairing region’. Note that the base-pairing region can be formed by blocks and contain parts that do not interact with the target mRNA (indicated by * on parts A and B of the Figure). The same situation can be found on the mRNA target, leading to an RNA duplex carrying bulges and mismatches. The ncRNA can pair to the TIR of the mRNA, which is normally recognized and bound by the 30S subunit of the ribosome to initiate translation. The TIR spans from −30 to +16 relative to the translation initiation codon (AUG). The ncRNA can also pair to an mRNA sequence located upstream of the TIR or within the encoding sequence. (B) Most well-known mechanisms of translation regulation mediated by ncRNAs. Translation inhibition (left part) occurs by pairing the ncRNA to the TIR of the mRNA, thereby blocking the binding of ribosomes. Translation activation (right part) occurs by pairing the ncRNA to an mRNA region that otherwise pairs to the TIR and prevents binding of ribosomes. The duplex formation releases TIR and becomes accessible to ribosomes. (C) RNase E- and RNase III-dependent mRNA degradation mediated by ncRNAs. The RNase E-dependent degradation (left) of the ncRNA/mRNA duplex appears to be associated with a Hfq-dependent duplex formation (see text). RNase E cleaves single-stranded RNA sequences either flanking or contained in the duplex. RNase E is the endoribonuclease of the degradosome, the machinery ensuring the RNA turnover in enterobacteriae. Note that RNase E and Hfq are not ubiquitous in the bacterial kingdom. The RNase III-dependent degradation (right part) is initiated by the recognition of the double-stranded RNA duplex and the cleavage of this region.

Download figure to PowerPoint

Regulation of protein synthesis

Translation inhibition

Translation inhibition by ncRNAs mainly occurs by base-pairing to sequences that overlap or are adjacent to the SD and/or translation initiation codon within the TIR of the mRNA target (Figure 6A). The formation of the RNA duplex is generally sufficient to prevent binding of the ribosome (Figure 6B). Translation inhibition is supposed to occur on nascent mRNA targets: as mRNAs are being synthesized, initiating ribosomes and ncRNAs compete for access to the same binding region. However, the mechanism may be more complex. For instance, translation inhibition of ptsG mRNA by SrgS requires a nascent peptide and imposes a membrane location of the translation complex before free ribosomes and SrgS compete (Aiba, 2007). Whether SrgS/ptsG depicts an atypical case or a more general mechanism is not known, but it is tempting to speculate that such a process may specifically apply to mRNAs competent for translation and encoding envelope proteins. It remains to be seen whether the mechanisms employed by ncRNAs controlling envelope homoeostasis are in keeping with this hypothesis (MICs/OMPs in Figure 1).

Protein encoding sequences on mRNAs are rarely targets for ncRNAs (Figures 1 and 6A). However, the example of SR1 in B. subtilis indicates another possible mode of action of ncRNAs. SR1 prevents the recruitment of ribosomes on ahrC by pairing with its coding region and inducing structural changes within the TIR (Heidrich et al., 2007) (Figure 1 and Figure 6A). This mode of action differs from most described ncRNA/mRNA interactions, where the pairing region is nested in the TIR (see Figure 1). ‘SR1/ahrC’ may be an exception to the rule, but it may indicate that target searches should focus more on such unexpected sites. However, some features of the ‘SR1/ahrC’ system have been reported; for instance, ItsR-1 with tisB mRNA pairing occurs outside the TIR. In this case, translation inhibition results from ncRNA binding to a standby site (which prevents ribosome binding), rather than from conformational changes of the mRNA. Also, in Salmonella enterica serotype Typhimurium it was demonstrated that the GcvB ncRNA inhibits translation of ABC transporters by targeting C/A rich regions far upstream of the TIRs (e.g. concerning gltI, argT and ilvJ) (Sharma et al., 2007). Similarly to SR1/ahrC, in Listeria monocytogene, the ncRNA RliB interacts with feoA mRNA outside TIR within the encoding sequence. Effects of RliB on feoA translation are not yet known (Mandin et al., 2007).

Translation activation

In contrast to inhibition, translation activation by ncRNAs has been reported in only a few cases (see Figure 1, first column). The only direct mechanism described so far involves pairing of an ncRNA with the 5′UTR of an mRNA, upstream of the TIR. Pairing induces structural changes, rendering the SD accessible to ribosomes (e.g. DsrA- and RprA/rpoS, RyhB/shiA, RNAIII/α-hla) (see Figures 6A and 6B).

In addition, GlmY and GlmZ ncRNAs in E. coli activate glmS translation by an indirect mechanism that has a stabilizing effect. glmS mRNA encodes the glucosamine-6-phosphate synthase and is derived from RNase E processing of the bicistronic mRNA glmUS. GlmY and GlmZ promote this processing at the end of the glmU coding sequence, releasing a stable glmS mRNA (Kalamorz et al., 2007; Urban et al., 2007).

mRNA target degradation

Actors in duplex degradation

Translation inhibition is generally associated with rapid degradation of the ncRNA/mRNA duplex by RNase E or RNase III (Figure 6C), indicating that ncRNAs might act stoichiometrically and might be consumed as they exert their effect (Massé et al., 2003; Morita et al., 2005; Dühring et al., 2006). However, elegant work has demonstrated that the formation of RyhB/sodB and SrgS/ptsG duplexes is sufficient for translation inhibition, and that RNase E-dependent degradation is a subsequent step that renders inhibition irreversible (Morita et al., 2005).

RNase III is known to cleave perfect RNA duplexes of about two helical turns, and acts occasionally on different, specific RNA substrates (Pertzev and Nicholson, 2006; MacRae and Doudna, 2007). In cases of perfect complementarity between RNA partners, duplexes formed should be substrates for RNase III, such as those generated by cis-antisense ncRNAs (e.g. RldD/ldrD, Rat/txpA and IsiR/isiA). Duplexes formed by trans-antisense ncRNA can also be substrates for RNase III. For example, duplexes formed between IstR-1 and tisB, and RNAIII and spa mRNA, interfere with translation initiation, but in vivo inhibition is enhanced by RNase III (Vogel et al., 2004; Huntzinger et al., 2005).

Hfq, the RNase E and the control of the RNA duplex destiny

The bacterial Sm-like protein Hfq is a key player in many ncRNA-mediated effects, as it binds ncRNAs with high affinity in vitro and helps interactions between ncRNAs and their mRNA targets in vivo (Valentin-Hansen et al. 2004). This conclusion comes from numerous examples of ncRNAs in E. coli and relatives, but it cannot be generalized (see Figure 1). In E. coli, examples are now known where Hfq is not required for formation of RNA duplexes. For instance, it seems quite general that cis-encoded ncRNAs do not require Hfq to form hybrids with their mRNA targets (Brantl, 2007). The trans-encoded ncRNA IstR-1 interacts with tisB mRNA independently of Hfq (Figure 1). In Gram-positive bacteria B. subtilis, S. aureus and L. monocytogenes, none of the ncRNA/mRNA duplexes characterized so far in vivo form in an Hfq-dependent manner (see Figure 1; Heidrich et al., 2006; Boisset et al., 2007; Mandin et al., 2007). In addition, although Hfq is well conserved in many bacterial species (Valentin-Hansen et al. 2004), it is absent in several Gram-negative and -positive species where ncRNA/mRNA interactions are currently studied, such as Helicobacter pylori and Enterococcus faecalis (F. Darfeuille and F. Repoila, unpublished work). A functional homologue to Hfq is likely to exist in these latter species, and maybe co-exists with Hfq in some bacteria where the hfq gene is present and where no Hfq-dependent ncRNA/mRNA interaction has been observed (e.g. B. subtilis, S. aureus and L. monocytogenes). Alternatively, it is conceivable that Hfq possesses a particular function in Enterobacteriacea and close relatives, and such a ‘helper’ protein might not be needed, since in many cases RNA structures are sufficient to promote efficient pairings.

In E. coli, the involvement of Hfq in RNA duplex formation seems to correlate with an effect of RNase E. This is particularly true for ncRNAs that mediate translation inhibition by pairing to TIRs of their target mRNAs (Figure 1). A direct interaction between Hfq and RNase E was revealed by the discovery of the ribonucleoprotein complex Hfq/RNase E/ncRNA (SrgS or RyhB). The Hfq/RNase E complex may be a general system to specifically degrade mRNAs targeted by ncRNAs when Hfq is involved in pairing (Morita et al., 2005, 2006). However, examples of Hfq-dependent translation activation where ncRNA/mRNA duplexes are not degraded (e.g. DsrA/rpoS, RprA/rpoS, RyhB/shiA, Qrr1-4/vca0939) indicate that association between Hfq and RNase E is controlled.

Concluding remarks

  1. Top of page
  2. Abstract
  3. Introduction
  4. Co-ordination of regulatory effects mediated by ncRNAs
  5. Adaptation processes controlled by ncRNAs
  6. Aspects and consequences of ncRNA/mRNA interactions
  7. Concluding remarks
  8. Acknowledgements
  9. Funding
  10. References

To date, ∼140 ncRNAs have been identified throughout the bacterial kingdom (Altuvia, 2007). In this review, we highlighted the impressive diversity of biological functions controlled by around 20 ncRNAs, and summarized the various modes of action currently known for fewer than 50 of them (Figure 1). The gap between the number of identified ncRNAs and their mechanisms suggests that many other processes under ncRNA-mediated regulation are yet to be discovered. Several genome-wide searches to identify ncRNAs have been performed in E. coli, and the vast majority of ncRNAs known are from this bacterium (∼80 out of 140) (Altuvia, 2007). In contrast, very few ncRNAs have been identified in other species, and very little is known about their mechanisms of action (Vogel and Sharma, 2005; Altuvia, 2007). The discovery of ncRNAs in bacteria other than E. coli will bring new insights into mechanisms and control of biological functions, since many bacterial traits, such as sporulation, biofilm development, antibiotic resistance, commensalism or virulence, are species-specific.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Co-ordination of regulatory effects mediated by ncRNAs
  5. Adaptation processes controlled by ncRNAs
  6. Aspects and consequences of ncRNA/mRNA interactions
  7. Concluding remarks
  8. Acknowledgements
  9. Funding
  10. References

We thank Gerhart Wagner and Jörg Vogel for communicating results prior to publication. We are grateful to Pascale Romby and Gerhart Wagner for their warm and constant support, and Alexandra Gruss and Aymeric Fouquier d'Herouel for comments and discussion on the manuscript. We apologize to colleagues whose original work has not been cited or replaced by reviews, due to space limitation.

Funding

  1. Top of page
  2. Abstract
  3. Introduction
  4. Co-ordination of regulatory effects mediated by ncRNAs
  5. Adaptation processes controlled by ncRNAs
  6. Aspects and consequences of ncRNA/mRNA interactions
  7. Concluding remarks
  8. Acknowledgements
  9. Funding
  10. References

F.D. is supported by INSERM (Institut National de la Santé et de la Recherche Médicale); F.R. is supported by INRA (Institut National de la Recherche Agronomique).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Co-ordination of regulatory effects mediated by ncRNAs
  5. Adaptation processes controlled by ncRNAs
  6. Aspects and consequences of ncRNA/mRNA interactions
  7. Concluding remarks
  8. Acknowledgements
  9. Funding
  10. References
Footnotes
  1. Regulatory ncRNA: RNA molecule, other than a tRNA or an rRNA, controlling biological processes by acting directly on proteins or mRNAs via an intermolecular reaction.

  2. Antisense RNA: Regulatory ncRNA pairing with an mRNA target.

  3. Cis-antisense ncRNA: Regulatory ncRNA transcribed from the complementary DNA strand to the DNA template of its mRNA target.

  4. Trans-antisense ncRNA: Regulatory ncRNA that is not transcribed from the complementary DNA strand to the DNA template of its mRNA target.

  5. Regulon: Collection of genes expressed under control of a common regulator (RNA or protein).

  6. Untranslated region (UTR): mRNA regions located 5′ or 3′ of a coding sequence within an mRNA.

  7. Shine—Dalgarno (SD) sequence: Sequence recognized and bound by the 16S ribosomal RNA. It is located upstream of the translation initiation codon (usually AUG) within the translation initiation region of the mRNA.

  8. Translation Initiation Region (TIR): Segment of an mRNA that is recognized and bound by the ribosome to initiate translation. It generally extends from about −30 nt to around +16 nt relative to the translation initiation codon.

  9. ncRNA/mRNA duplex (or RNA hybrid): Perfect or partial double stranded RNA segment formed by the interaction between an ncRNA and its mRNA target.

  10. Standby site: RNA region outside or near a TIR, which is bound by the ribosome, usually without sequence specificity. It facilitates ribosome access to structurally sequestered TIRs.