Switching off small RNA regulation with trap-mRNA

Authors


*E-mail valentin@bmb.sdu.dk; Tel. (+45) 6550 2375; Fax (+45) 6550 2467.

Summary

Small non-coding regulatory RNAs in bacteria have been shown predominantly to be tightly regulated at the level of transcription initiation, and sRNAs that function by an antisense mechanism on trans-encoded target mRNAs have been shown or predicted to act stoichiometrically. Here we show that MicM, which silences the expression of an outer membrane protein, YbfM under most growth conditions, does not become destabilized by target mRNA overexpression, indicating that the small RNA regulator acts catalytically. Furthermore, our regulatory studies suggested that control of micM expression is unlikely to operate at the level of transcription initiation. By employing a highly sensitive genetic screen we uncovered a novel RNA-based regulatory principle in which induction of a trap-mRNA leads to selective degradation of a small regulatory RNA molecule, thereby abolishing the sRNA-based silencing of its cognate target mRNA. In the present case, antisense regulation by chb mRNA of the antisense regulator MicM by an extended complementary sequence element, results in induction of ybfM mRNA translation. This type of regulation is reminiscent of the regulation of microRNA activity through target mimicry that occurs in plants.

Introduction

Chromosomally encoded small non-coding RNAs (sRNAs) regulate important aspects of physiology in enteric bacteria, and it has become increasingly clear that a major class of these riboregulators act by an antisense mechanism to modify the translation and/or the stability of target mRNAs (Waters and Storz, 2009). The cellular role of a significant portion of these base pairing RNAs is to control stress responses and expression of outer membrane proteins (Gottesman et al., 2006; Guillier et al., 2006; Vogel and Papenfort, 2006; Valentin-Hansen et al., 2007; Vanderpool, 2007; Vogel, 2009). In particular, studies of the stress-induced sRNAs have uncovered common regulatory features of sRNA-mediated post-transcriptional regulation. First, the sRNA genes are generally highly regulated at the transcriptional level, and frequently are expressed as part of well-known stress response regulatory systems. Second, they require the Sm-like protein Hfq which in vitro strongly enhances the on-rate of duplex formation between sRNA and target RNA (Møller et al., 2002; Zhang et al., 2002; Kawamoto et al., 2006; Arluison et al., 2007; Soper and Woodson, 2008; Rasmussen et al., 2009). Third, pairing typically occurs at or near the ribosome-binding site of target mRNA and leads to translational inhibition and concomitant rapid degradation of the message. Intriguingly, studies on the RyhB and SgrS RNAs (involved in iron and glucose-phosphate stress respectively) led to the discovery of rapid co-degradation of sRNA with the mRNA target in an RNase E-dependent manner, and preliminary data suggest that several other well-studied regulatory RNAs are co-degraded with their mRNA targets (Masséet al., 2003; Kawamoto et al., 2006; Aiba, 2007). All of these are induced under stress conditions and regulate metabolic pathways specifically related to that temporary condition. Because sRNA-coupled degradation enables a rapid shutdown of the regulatory RNA as soon as its transcription ceases, it has been argued that all chromosomally encoded base paring RNAs are likely to act stoichiometrically, i.e. being destroyed in a 1:1 manner with the target RNA (Masséet al., 2003).

It is clear, however, that a number of base paring RNAs are expressed not only in response to certain stress conditions but also under standard growth conditions. One such RNA is the recently studied MicM (formerly RybC/SroB) which downregulates the expression of the ybfMN operon encoding a conserved porin (YbfM) and a putative lipoprotein (Rasmussen et al., 2009). MicM regulation of ybfMN is strictly dependent on the Sm-like protein Hfq and genetic and biochemical studies provided evidence that MicM acts by an antisense mechanism to silence ybfM and ybfMN mRNAs in vivo. In wild-type cells these messages are barely detectable and basal expression of MicM from a tightly controlled promoter was sufficient for a near-complete elimination of the two transcripts.

Here we have extended our analysis of this regulatory case and carried out investigations aimed at elucidating MicM regulation and action as well as the biological role of YbfM and the significance of its sRNA-mediated silencing. The studies uncovered a novel gene-regulatory system with a twist as well as its physiological relevance. The negative regulator of MicM activity is not a regulatory protein but instead a polycistronic mRNA that carries a special element with complementarity to MicM RNA. Induction of this mRNA causes destruction of the sRNA regulator, thereby abolishing the sRNA-based silencing of its cognate target mRNA. We propose to use the term ‘trap-mRNA regulation’ for defining this mechanism of post-transcriptional inhibition of small regulatory RNA activity.

Results

Turnover of MicM and ybfM RNA

The recent finding that MicM RNA is capable of downregulating its target with unprecedented efficiency (Rasmussen et al., 2009), prompted us to analyse more closely the turnover of MicM. To investigate the effect of ybfM mRNA on MicM (and vice versa) we employed a ΔmicM, ΔybfM strain carrying the micM gene under control of the arabinose-inducible araBAD promoter and a truncated form of the ybfM gene (ybfM′) under control of a synthetic IPTG-inducible lac promoter. Thus, in this strain the expression of regulatory RNA and target RNA can be switched on and off at will, simply by adding or removing inducers. In a first series of experiments we expressed MicM and the ybfM′ transcript for 20 min, either alone [Fig. 1, pBAD-micM/pNDM220 (closed circles); pBAD33/pNDM-ybfM (closed triangles)] or in combination [Fig. 1, pBAD-micM/pNDM-ybfM (open circles/triangles)], and monitored their disappearance when inducers were removed. The results show that the lifetime of MicM is unaffected by target expression (half-life ∼27 min) whereas the target RNA is cleared from the cell much more rapidly in the presence of MicM RNA than in its absence (half-lives being ∼4 min and ∼27 min respectively) (Fig. 1). Thus the experiment did not result in any sign of target-coupled sRNA degradation. Also, it should be noted that the estimated half-life of MicM is very similar to that determined in a rifampicin run-out experiment (Vogel et al., 2003). This is in striking contrast to the stress-induced regulatory RNAs which are cleared from the cell much more rapidly in the absence of rifampicin than in its presence, i.e. the sRNAs become destabilized with ongoing transcription where their target mRNAs are synthesized.

Figure 1.

The half-life of ybfM′ mRNA is affected by MicM, but not vice versa.
A. Northern blot analysis of MicM and ybfM′ RNA levels. SØ928 ΔmicMΔybfM cells containing either pBAD-micM/pNDM220, pBAD33/pNDM-ybfM or pBAD-micM/pNDM-ybfM were grown to an OD450 of 0.7 in LB medium and then induced for 20 min with 1 mM arabinose and 2 mM IPTG. Following induction, the cultures were centrifuged and the cells were resuspended in pre-warmed LB medium containing 0.2% glucose but without inducers. Samples were taken 0, 2.5, 5, 10, 20, 40 and 80 min after termination of micM/ybfM expression.
B. Quantification of RNA levels from Northern blots. MicM values in the presence and absence of ybfM mRNA are indicated by open and closed circles respectively. ybfM mRNA values in the presence and absence of MicM are indicated by open and closed triangles respectively. Amounts of MicM and ybfM′ mRNA in analysed samples are indicated on the left ordinate. When expressed alone the ybfM/MicM RNA ratio was ∼1:2. Estimated half-lives (T½) are indicated on the right ordinate.

In a next series of experiments cells were grown in the presence of arabinose – thus only expressing the micM gene. At time zero arabinose was removed to terminate micM expression and IPTG was added to induce ybfM′ expression. Subsequently, accumulation of ybfM′ mRNA and disappearance of MicM RNA were monitored by Northern blot analysis. The results are presented in Fig. 2, along with data from parallel cultures in which only expression of sRNA or target RNA had been switched on. Whereas MicM decay was unaffected by ybfM′ mRNA expression (Fig. 2A, lanes 8–14 versus lanes 15–21), the accumulation of the target RNA was strongly inhibited by MicM until the cellular level of this RNA species became almost undetectable (Fig. 2A, lane 20). At this time point (∼32 min), the ybfM′-transcript had accumulated to ∼700 nmol in the RNA sample from cells without MicM expression, while just ∼100 nmol was present in samples from cells with MicM expression up until time zero (Fig. 2B). As cell samples contained approximately 50 nmol of MicM when ybfM′ expression was induced (Fig. 2B; zero time point), it follows that one molecule of the regulatory RNA can participate in the degradation of several molecules of target RNA. Based on these findings we infer that MicM decay is uncoupled from that of its target and that MicM RNA acts catalytically rather than stoichiometrically.

Figure 2.

MicM facilitates degradation of excess ybfM mRNA.
A. Northern blot analysis of MicM and ybfM mRNA levels. Cells of strain SØ928 ΔmicMΔybfM containing either plasmids pBAD33/pNDM-ybfM, pBAD-micM/pNDM220 or pBAD-micM/pNDM-ybfM were grown in LB medium to an OD450 of 0.7 and induced with 1 mM arabinose for 20 min. The culture was centrifuged briefly to remove the inducer, and cells were resuspended in fresh pre-warmed LB medium containing 200 µM IPTG. Samples were withdrawn at times 0, 2, 4, 8, 16, 32 and 64 min after IPTG induction of MicM expression.
B. Quantification of RNA levels from Northern blots. MicM levels in the presence and absence of ybfM mRNA are indicated by open and closed circles respectively, and RNA amounts (nmol) in the samples are shown on the left ordinate. ybfM mRNA levels in the presence and absence of MicM are indicated by open and closed triangles respectively, and mRNA amounts (nmol) in the samples are shown on the right ordinate.

A genetic screen for factors affecting MicM-mediated ybfM silencing

To screen for factors that compromise MicM downregulation of ybfM expression we constructed a translational fusion with the lacZ reporter gene and inserted this construct in single copy at the λ attachment site attB of a Δlac strain. The resulting reporter strain formed white colonies on medium containing the β-galactosidase indicator 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (Xgal), while a Δlac, ΔmicM strain with this fusion formed deep blue colonies. The reporter strain was transformed with a pBR322-based Escherichia coli genomic library with the rationale that increased level of a putative DNA element or overproduction of a putative RNA transcript/protein interfering with silencing of ybfM expression should result in blue-coloured (Lac+) colonies on Xgal indicator plates. We screened approximately 14 000 transformants. Plasmids from Lac+ colonies were purified, their activity confirmed by retransformation, and inserts were subsequently determined by DNA sequencing. This resulted in inserts that could be grouped in two classes (Table 1).

Table 1.  Plasmids that affect ybfM-lacZ expression.
ClassNumberGenes insertedPhenotype
  1. Number of isolates refers to non-identical plasmid isolates.

  2. Genes inserted refer to shared genes of plasmid isolates with overlapping sequences. A prime of one side of a gene indicates that it is interrupted on this side. The symbol (+) refers to intensity of blue colouring of colonies on Xgal containing LB agar plates.

14rppH, ′ptsP+
14yadB, pcnB, folK++/+++
211ybfM-UTR, ybfM++/+++
210chbB′, ′chbC++/+++

One class encodes protein factors, PAPI or RppH, that promote RNA degradation (Xu and Cohen, 1995; Mohanty et al., 2004; Deana et al., 2008). Poly(A) polymerase (PAPI), encoded by the pcnB gene, is the main enzyme responsible for RNA polyadenylation, which can destabilize bacterial transcripts by facilitating 3′- to 5′-end exonuclease-mediated decay of RNAs with a structured 3′ end, e.g. transcripts that ends with Rho-independent transcription terminators, including the GlmY RNA and plasmid encoded antisense RNAs (Xu et al., 1993; Dam and Gerdes, 1997; Söderbom et al., 1997; Görke and Vogel, 2008; Urban and Vogel, 2008). RppH is a newly identified pyrophosphohydrolase which triggers RNA degradation by 5′-end pyrophosphate removal. Specifically, the presence of a monophosphate at a 5′-end enhances the catalytic efficiency of the endonuclease RNase E, and further increases the specificity of cleavage site selection within an RNA substrate (Deana et al., 2008).

The second class carries two types of inserts with gene fragments possessing complementarity to MicM. One contains the well-investigated target sequence in the translational initiation region of ybfM (Rasmussen et al., 2009), and the 11 non-identical inserts share a common sequence of 70 nucleotides that has 5′- and 3′-ends located 30 and 28 nucleotides respectively, from the MicM complementary region of 12 nucleotides (Fig. 3B). The other contains the 84-nucleotide intergenic region (IGR) between the first and second gene of the chitobiose, chbBCARFG, operon (Keyhani and Roseman, 1997; Plumbridge and Pellegrini, 2004). This IGR possesses complementarity to the same region of MicM that is used for targeting ybfM; however, the complementarity is extended to 21 nucleotides, including two central mismatches (Fig. 3B). The 10 non-identical inserts share a sequence that has a 5′ end located at −20 from the stop codon of chbB and a 3′ end located at +162 from the start codon of chbC. In all cases the fragments are inserted in pBR322 such that the coding strand of ybfM and chbBC will be transcribed from the tet-promoter. The chitobiose operon (Fig. 3A) encodes genes for the transport and degradation of the N-acetylglucosamine disaccharide, chitobiose (GlcNAc)2, the main degradation product of chitin, which constitutes the second-most abundant organic polymer in nature after cellulose (Keyhani and Roseman, 1997; Keyhani et al., 2000a,b,c; Plumbridge and Pellegrini, 2004). The chb operon encodes its own specific regulator, ChbR, an AraC-type dual repressor–activator, which is necessary for transcription activation in the presence of chitobiose in vivo, and expression of the operon allows E. coli and relatives to utilize chitobiose and chitotriose as sole carbon, nitrogen and energy source (Keyhani and Roseman, 1997; Plumbridge and Pellegrini, 2004).

Figure 3.

Analysis of multicopy suppressor plasmids that enhance ybfM-lacZ expression.
A. Genetic organization of the chb chromosomal region represented by Class 2 plasmid inserts.
B. Detailed view of the intergenic region common to all chb inserts in Class 2 plasmids. Possible base paring between MicM and chbBC IGR is indicated; 15 and 48 nucleotides separate the paired bases closest to the chbB stop and chbC start codons respectively. The complementarity is extended to 21 nucleotides, including two central mismatches compared with the 12-nucleotide complementarity between MicM and the ybfM target (shown below). Boxed nucleotides indicate base positions that are changed in the mutant variant MicM 55+56.
C. Northern blot showing the effect of Class 1 and 2 plasmids on MicM and ybfM/ybfMN RNA levels. Wild-type SØ928 was transformed with either empty vector, Class 1 or Class 2 plasmids each containing the shortest chromosomal fragments isolated of the loci indicated in the figure. RNA was isolated from exponentially grown cultures LB medium at an OD450 of 0.7.

Transcription of the chbB-chbC IGR causes destabilization of MicM RNA

The activities of both classes of plasmids were further analysed by Northern blot analysis. Wild-type strain SØ928 containing either empty vector or plasmids with the smallest genomic fragments of ybfM, chbBC, pcnB and rppH was grown exponentially in Luria–Bertani (LB) and total RNA was prepared for detection of MicM and ybfMN transcripts. In agreement with previous findings, the presence of MicM RNA reduces ybfM/ybfMN transcripts to low levels barely detectable using Northern analysis (Fig. 3C, lane 1; empty pBR322 vector). The presence of a pBR-ybfM plasmid obtained in the genetic screen does not significantly affect the level of MicM RNA (Fig. 3C, lane 2). This is in agreement with the stability studies presented in Figs 1 and 2. However, ybfM/ybfMN mRNAs are induced and multiple smaller transcripts are detected, presumably due to transcription of the genomic ybfM fragment from the pBR322 plasmid. This result confirms the Lac+ phenotype obtained in the genetic screen and suggests that plasmids with ybfM′ inserts inhibit MicM activity by producing competing RNA substrates, thereby resulting in increased ybfM/ybfMN mRNA accumulation and activity. Intriguingly, the presence of the second type of plasmid carrying the chbBC-IGR resulted in a markedly reduced MicM level (∼10-fold), indicating that expression of this region interferes with MicM stability (Fig. 3C, lane 3). In line with this, the ybfM/ybfMN mRNAs accumulated. Finally, the presence of plasmids pBR-pcnB or pBR-rppH containing genomic fragments of pcnB or rppH, respectively, resulted in slightly reduced MicM RNA levels and the ybfM/ybfMN mRNAs were barely detectable (Fig. 3C, compare lane 1 with lanes 4–5).

Taken together, the chbB′C′ fragment of the chbBCARFG operon contains an extended MicM target sequence and promotes a significant reduction in MicM RNA levels and importantly, accumulation of ybfM/ybfMN mRNA.

Growth on chitobiose leads to degradation of MicM and concomitant induction of ybfM/ybfMN by an anti-antisense mechanism

Growth of E. coli using chitobiose as a sole energy and carbon source requires the induction of the chb operon (Keyhani and Roseman, 1997; Plumbridge and Pellegrini, 2004). To further investigate the physiological condition under which MicM-mediated riboregulation is switched off, we next compared the RNA levels of MicM and ybfM/ybfMN in early and late exponential cultures of wild-type cells cultivated in glucose, glycerol and chitobiose minimal media respectively. Consistent with the results obtained in the genetic screen, MicM levels became strongly reduced (∼10-fold) upon induction of chb-operon expression by growth in chitobiose medium (Fig. 4A, lanes 5–6) compared with growth in glucose or glycerol medium (Fig. 4A, lanes 1–4). The expected concomitant strong accumulation of ybfM/ybfMN mRNA (Fig. 4A, lanes 5–6) clearly indicates a physiological role of the YbfM porin in chitobiose utilization. To test directly, whether a complementary interaction between MicM and the chb-IGR is critical for the stability of the sRNA, we employed a mutant form of MicM (MicM55G-56C, see Fig. 3B) that is specifically defective in base paring with wild-type ybfM but capable of binding to and regulating a mutant form of ybfM carrying compensatory mutations (Rasmussen et al., 2009). In addition we showed that (i) under standard growth conditions, short-term expression of wild-type micM and micM 55+56 from the arabinose-inducible araBAD promoter resulted in similar cellular levels of the sRNAs; and (ii) induced levels of MicM and MicM 55+56 were not significantly affected by YbfM expression (Rasmussen et al., 2009). The experiment presented in Fig. 4B revealed that the level of wild-type MicM in the ΔmicM deletion strain upon induction by arabinose in chitobiose minimal medium was significantly reduced compared with the level of mutated MicM (Fig. 4B, lanes 2 and 4). Collectively, these findings indicate that MicM destabilization during chb operon induction requires base paring to its alternative target in the chbBC-IGR, which thereby acts as a trap-element.

Figure 4.

Effects of carbon and energy source on MicM and ybfM/ybfMN RNA expression.
A. Northern blot analysis of wild-type SØ928 grown in minimal media containing either 0.2% glucose, 0.2% glycerol or 0.1% chitobiose as sole carbon and energy source. Samples were taken in early exponential growth phase (EE) OD450∼0.3 and in late exponential growth phase (LE) OD450∼0.9 as indicated.
B. Chitobiose-induced degradation of MicM is compromised by MicM base mutations in the ybfM/chb complementary region. Cultures of SØ928 ΔmicM/pBAD-MicM and SØ928 ΔmicM/pBAD-MicM 55+56 were grown in minimal medium containing 0.1% chitobiose to an OD450 of 0.8. Samples were taken before (−) and after (+) 20 min of arabinose induction (0.2%). The levels of MicM were determined using Northern blot analysis.

Biological role of YbfM

In an attempt to obtain information about a putative role for YbfM we carried out a combined literature search for chitobiose and YbfM porin. Intriguingly, the search revealed that a comparative genomic analysis for reconstruction of the chitin and N-acetyl-glucosamine utilization pathways in proteobacteria had let to the view that the YbfM protein in enterobacteria constitutes a porin for smaller chitooligosaccharides such as chitobiose (Yang et al., 2006). This information, together with our findings, supports the notion that chb-operon mRNA is a bona fide riboregulator, rather than mere target of MicM and it is easy to envisage a gene regulatory circuit in which the ybfM messages are eliminated from the cell by MicM until chitooligosaccharides accumulate in the environment and the YbfM porin is needed. At that point the expression of the chitobiose operon is switched on and MicM binds to a special trap element in the mRNA. In turn the MicM sRNA is subjected to degradation thus resulting in YbfM porin expression (Fig. 5).

Figure 5.

mRNA-mediated antisense regulation of antisense RNA makes sense. In the absence of chitobiose, the main product of chitin degradation, the expression of the YbfM porin in silenced very efficiently in a catalytic manner at the post-transcriptional level by MicM RNA (OFF). At the same time the chb operon is transcriptionally repressed (OFF) due to binding by ChbR (encoded by the chb operon) and NagC (not shown). Under conditions where E. coli or related species encounter degradation products such as chitobiose made by chitinolytic organisms, the ChbR regulator is converted from a repressor to an activator via an unknown signal and transcription of the chb operon is induced (ON). Under these conditions MicM RNA is trapped and targeted for degradation by an extended target sequence in the chbBC mRNA intergenic region (coloured blue). This extended target sequence, thus, mimics the cognate MicM target (ybfM) (coloured red). Consequently, the ybfM/ybfMN mRNAs accumulate rapidly and the YbfM porin is synthesized (ON). YbfM is likely to function as an outer membrane porin for chitooligosaccharides (see Discussion) and act in concert with the specific PTS transporter (EIIChb permease) in mediating uptake of chitooligosaccharides.

Discussion

Most small non-coding regulatory RNAs in bacteria have been shown to be tightly regulated at the transcriptional level and the general view held, is that they function by antisense mechanisms on trans-encoded mRNAs in a stoichiometric fashion. Prior to this work, we reported that the conserved MicM RNA downregulates ybfMN expression through complementary interaction with the translational initiation region of ybfM messages, causing them to be degraded. Furthermore, we observed that MicM was able to downregulate its target with unprecedented efficiency (Rasmussen et al., 2009). However, the physiological significance of the sRNA-mediated regulation was far from clear, because to the best of our knowledge no function had been ascribed to the YbfM porin in E. coli and its close relatives. Also, the regulation of MicM expression remained unclear. In this regard alignments of promoter regions of conserved sRNA genes have in several cases let to the identification of binding sites for transcription factors, e.g. the σE-dependent MicA and RybB RNAs and cAMP-CRP-dependet CyaR RNA (Figueroa-Bossi et al., 2006; Johansen et al., 2006; Papenfort et al., 2006; Thompson et al., 2007; Johansen et al., 2008; Papenfort et al., 2008; De and Gottesman, 2009). For MicM, however, such alignments only revealed conservation of two sequence motifs that contain good matches to the σ70−10 and −35 boxes. Moreover transcriptional analysis indicates that sequences upstream of the −35 motif are not essential for MicM promoter expression (Mandin and Gottesman, 2009).

The goal of the present study was to shed some light on this gene regulatory circuit. We first provided evidence that MicM is unusual in several aspects. First, in striking contrast to the stress-induced sRNAs, MicM does not become destabilized upon overexpression of its target mRNA. Rather it appears that MicM acts catalytically in silencing the ybfMN expression. Second, we have not been able to identify classical transcriptional regulators of micM transcription, despite employing highly sensitive genetic screens including transposon mutagenesis of strains carrying fusions that link either the micM or the ybfM gene to lacZ. Although these studies were non-productive, they indicated that micM might be under post-transcriptional control. The subsequent genetic analysis described in the present work uncovered what was first believed to be a novel target for MicM. However, the finding that expression of the chbBC-IGR of the chitobiose operon caused destruction of MicM; that the IGR possessed extended complementarity to the same region of MicM known to be involved in pairing with the ybfM mRNAs; and finally, the realization that the YbfM porin is functionally connected with chb operon products revealed a neatly straightforward mechanism of switching off sRNA-mediated gene silencing. Here, an mRNA molecule, the chb operon message, is proposed to act as an antisense RNA that destroys the antisense regulator MicM, thereby alleviating silencing of ybfMN expression. In support of this view, the potential base paring capacity is retained in related bacteria such as Citrobacter, Klebsiella, Proteus, Salmonella and Enterobacter sp. 638. Furthermore, in parallel experiments in Salmonella, Figueroa-Bossi and coworkers demonstrated that MicM (they have renamed the sRNA ChiX) binds directly to the complementary region of the chbBC IGR and that proper complementary interaction is a prerequisite for chb-RNA-mediated regulation of MicM activity (Figueroa-Bossi et al., 2009). Most importantly, the authors also provided experimental evidence that YbfM acts as a porin for chitotriose, further supporting the hypothesis that the chb-operon mRNA functions as a bona fide riboregulator. To substantiate this hypothesis it would be important to define the role of Hfq in chb-mediated degradation of MicM. Hfq binds efficiently to MicM and ybfM mRNA and is essential for in vivo stability of the sRNA as well as for MicM-mediated decay of the ybfMN messages, and in vitro, Hfq strongly promotes annealing between sRNA and its target (Mandin and Gottesman, 2009; Rasmussen et al., 2009; Sittka et al., 2008). We predict that the same is the case for trap-mRNA-mediated degradation of MicM. In this regard a deep sequencing analysis of mRNA targets for Hfq in Salmonella identified chbB and chbC transcripts (also known as celA and celB) as being Hfq bound in vivo (Sittka et al., 2008) and the chbBC-IGR sequence is comprised predominantly of A/U-rich tracks which are preferred sequence elements for tight binding by Hfq (Brennan and Link, 2007). Other important objectives for further work include determination of the degradation pathways of MicM RNA and ybfMN mRNA and determination of whether chb mRNA turnover is coupled to and dependent on pairing with MicM RNA. Furthermore, it will be important to understand the different outcomes of MicM pairing to the two target RNAs and define the minimal RNA element required for MicM destruction.

An exciting question is how widespread mRNA-mediated antisense control is in biology. It appears obvious that this strategy is simple and versatile and would allow post-transcriptional gene regulation under diverse physiological conditions. Thus, induction of an mRNA carrying a short, properly placed trap sequence element should suffice for this mechanism to take place. Our findings reveal yet another modular and transplantable RNA-based regulator switch that may have functioned in primitive organisms, and further suggests that present cells retain some of these ancient genetic control systems. Other examples of RNA-based regulatory circuits include the on-off switchable self-cleavage regulatory elements present within the mRNA itself, as first found in the glmS mRNA of Gram-positive bacteria, encoding an enzyme involved in the synthesis of glucosamine 6-phosphate (Winkler et al., 2004). Furthermore, the mRNA-based regulation in bacteria is reminiscent of target mimicry for regulation of microRNA activity in plants (Franco-Zorrilla et al., 2007). Here a family of non-coding RNAs bears a 23-nucleotide sequence that resembles a cleavable miRNA target site, but contrary to mRNA target sites that show almost perfect complementarity, the pairing to non-coding RNA targets is interrupted by a mismatched loop. In consequence this site is not cleaved and instead negatively regulates miRNA activity through mimicry. In addition, translation of the glmS mRNA in enterobacteria is positively controlled by two homologue sRNAs, GlmY and GlmZ. Whereas the latter sRNA directly targets glmS mRNA, the GlmY RNA acts indirectly by protecting GlmZ by a yet unknown mechanism that might involve RNA mimicry to compete with factors that tag the highly similar sRNA for cleavage (Urban and Vogel, 2008). Thus, in bacteria there is accumulating evidence that various RNA-based control systems are key players in sugar metabolism (for a recent review, see Görke and Vogel, 2008).

Experimental procedures

Bacterial strains and plasmids

All strains used are derivatives of E. coli K-12 strain SØ928 (Δdeo,Δlac) (Valentin-Hansen et al., 1978). Construction of ΔmicM and ΔybfM deletion strain derivatives of SØ928 is described in Rasmussen et al. (2009). The chromosomal ybfM′-lacZ′YA translational fusion was constructed as follows. A DNA fragment corresponding to nucleotides −124 to +241 respective to the transcription start site of ybfM was amplified using primers ‘ybfM-lacZ forw’ (5′-GCGAATTCCTTTCAACCGCAAACTTAAGCC) and ‘ybfM-lacZ rev’ (5′-CCGGATCCCAGTAATAGATACCGCCGG) and chromosomal DNA of strain SØ928 as template. The resulting PCR product was subsequently cloned in-frame with the lacZ′ gene between the EcoRI and SmaI sites of the R1 low-copy number translation fusion vector pOU246. A EcoRI-SalI fragment from the resulting plasmid carrying the ybfM′-lacZ′YA gene fusion was ligated with a EcoRI-SalI fragment of plasmid pTAC3590 containing the aphA (KanR) gene and λattP, thereby generating circular DNA substrates for integration into chromosomally attB site (Atlung et al., 1991). To do this, strains SØ928 and SØ928ΔmicM carrying plasmid pBK491 were transformed with the ligated fragments. Plasmid pBK491 (AmpR, repts) encodes the int gene under control of the λPR promoter as well as the λ repressor CI857 gene; thus, cells carrying this plasmid expresses the Int protein at 42°C, but the plasmid cannot be replicated at this temperature. Recombinants were selected at 42°C using LA plates containing 40 µg ml−1 kanamycin, and recombination was verified by PCR using primers ‘attB forw’ (5′-CTCTGCCAGATGGCGC) and ‘attB rev’ (5′-CACAGGTTGCTCCGGGCT).

Plasmids pBAD-micM, pBAD-micM 55+56 and pNDM-ybfM′ are described in Rasmussen et al. (2009). The pBAD-micM and pBAD-micM 55+56 plasmids are derivatives of the low-copy vector pBAD33 (Guzman et al., 1995) and pNDM-ybfM′ is derivatives of the R1 low-copy-vector pNDM220 (Gotfredsen and Gerdes, 1998). The pNDM-ybfM′ plasmid carries the leader region and the first 39 codons of the ybfM gene followed by an in-frame stop codon and the strong rrnB terminator.

Media and growth conditions

Strains were grown at 37°C in liquid LB medium or AB minimal medium (Clark and Maaløe, 1967) containing 1 µg ml−1 thiamine and carbon and energy source (0.2% glucose, 0.2% glycerol or 0.1% chitobiose respectively). Chitobiose (N,N′-diacetylchitobiose) was purchased from Sigma-Aldrich. Inoculation from an overnight culture was done to an OD450 of ∼0.03. Samples for Northern blot analysis were taken at OD450∼0.3 for early or 0.9 for late exponential growth. In time-course experiments, the time zero point was withdrawn at OD450∼0.8. Final concentrations of arabinose and IPTG inducers, induction times and time points for samples taken, are indicated in the figure legends. Strains carrying plasmids pBAD33 and derivatives where grown in the presence of 30 µg ml−1 chloramphenicol for plasmid preservation. Strains carrying plasmids pNDM220 and derivatives where grown in the presence of 30 µg ml−1 ampicillin for plasmid preservation. Indicator plates (LB agar) for genetic screening contained 40 µg ml−1 Xgal, 100 µg ml−1 ampicillin and 50 µg ml−1 kanamycin.

Genetic screen

A multicopy plasmid DNA library screen was performed in strain SØ928 attB::ybfM′-lacZ′YA using a pBR322-based E. coli SØ928 genomic library. Two libraries were constructed using either EcoRI-HindIII restricted genomic DNA fragments of approximately 1–5 kb or using Sau3AI restricted DNA fragments of similar size. The libraries were ligated into EcoRI-HindIII and BamHI restricted pBR322 vector respectively. The screening strain SØ928 attB::ybfM′-lacZ′YA appeared completely pale on Xgal containing indicator plates due to MicM-mediated repression of the ybfM′-lacZ fusion construct (see text). Calcium chloride competent cells were subsequently transformed with each of the two plasmid libraries and ampicillin-resistant colonies were screened for Lac+ phenotype on indicator plates. Approximately 14 000 colonies were screened and positive clones, approximately 1 for each 200 transformants, were colony purified and plasmid DNA was isolated using a plasmid miniprep kit (Bio-Rad). Each plasmid that conferred the Lac+ phenotype when it was retransformed back into the screening strain, was selected for DNA sequencing using the forward and reverse primers ‘pBR CW’ (5′-CCATTATTATCATGACATTAACC) and ‘pBR CCW’ (5′-TAGGCGCCAGCAACCGCAC) for the EcoRI-HindIII plasmid library and CCW2′ (5′-GCATTGTTAGATTTCATACACG and ‘pBR CW2’ (5′-GCTACTTGGAGCCACTATCG) for the Sau3AI/BamHI plasmid library.

Northern analysis

Total RNA was isolated using the hot phenol method (Sambrook et al., 1989) and quantified on a NanoDrop 1000 (Thermo Scientific). For Northern analysis, 5 µg total RNA was fractionated by PAGE 4.5% (ybfM/ybfMN) or 8% (MicM) low-bis acrylamide and semidry blotted to a nylon membrane (Zeta-probe, Bio-Rad) using an electroblotting apparatus (Hoefer SemiPhor, Amersham Pharmacia) 1–2 h at 2 mA cm−2. Following UV-cross-linking, the membrane was hybridized with a 32P-end-labelled DNA oligonucleotide probe, complementary to ybfM (5′-CGACATTGCTGTAACACCGGCGATAGCCAGCGCCA) or MicM probe (5′-CGTCAAAGAGGAATTTCATTTTTTTATTATTATG) in PerfectHyb Plus Hybridization Buffer (Sigma-Aldrich) according to the manufacturer's instructions. Northern membranes were finally subjected to phosphoimaging using storage phosphor screens and a Typhoon Trio scanner (GE Healthcare). Band intensities were quantified using ImageQuant 5.0 (Molecular Dynamics). For quantitative estimations, the radiation level of each band was converted to nmol of the specific RNA species by comparison with in vitro transcribed RNA running on the same gel.

RNA half-life determination

In order to determine the half-lives of MicM and ybfM′, RNA synthesis from the induced PA1/O4/O3 or PBAD promoters was terminated. In brief, cells were pelleted by centrifugation to wash out the inducer and cell pellets were resuspended in an equal volume of pre-warmed medium supplemented with 0.2% glucose. Cell culture samples were snap frozen in liquid nitrogen immediately prior to centrifugation. Extracted RNA was analysed by Northern blotting as described. The intensities were converted to nmol and plotted and RNA half-life was plotted using the slope from each plot.

Acknowledgements

We thank Katja Christensen for enthusiastic and expert technical assistance. We would like to thank N. Figueroa-Bossi and L. Bossi for sharing results prior to publication. This work was funded by the Danish National Research Foundation (Danmarks Grundforskningsfond) Centre for mRNP biogenesis and metabolism, and The Danish Natural Science Research Council.

Ancillary