Prokaryotic regulatory small RNAs act by a conserved mechanism and yet display a stunning structural variability. In the present study, we used mutational analysis to dissect the functional anatomy of RybB, a σE-dependent sRNA that regulates the synthesis of major porins in Escherichia coli and Salmonella. Mutations in the chromosomal rybB locus that altered the expression of an ompC–lac fusion were identified. Some of the mutations cluster within a seven-nucleotide segment at the 5′ end of the sRNA and affect its ability to pair with a sequence 40 nucleotides upstream from ompC translation start site. Other mutations map near the 3′ end of RybB, destabilizing the sRNA or altering its binding to Hfq. The 5′ end of RybB is also involved in ompD regulation. In this case, the sRNA can choose between two mutually exclusive pairing sites within the translated portion of the mRNA. Some of the RybB 5′ end mutations affect the choice between the two sites, resulting in regulatory responses that diverge from those observed in ompC. Further analysis of RybB target specificity identified chiP (ybfM), a gene encoding an inducible chitoporin, as an additional member of the RybB regulon. Altogether, our results indicate that an heptameric ‘seed’ sequence is sufficient to confer susceptibility to RybB regulation.
Non-coding small RNAs have emerged as relevant components of the regulatory repertoires of all living cells. Micro RNAs in eukaryotes and small regulatory RNAs in prokaryotes modulate gene expression at the post-transcriptional level through the establishment of base pair interactions with short complementary sequences in mRNA targets. In bacteria, sRNAs often pair with sequences immediately upstream of the translation start site or within the first few codons of the mRNA (Gottesman, 2005; Aiba, 2007; Vogel, 2009; Waters and Storz, 2009). Formation of the RNA duplex interferes with ribosome binding and inhibits translation initiation. Exposure of endoribonuclease cleavage sites in the untranslated mRNA stimulates its decay. The latter effect, although not essential for regulation, is thought to confer robustness to the process and to make it irreversible (Morita et al., 2006). A notable variation of this theme is encountered in the inhibition of OmpD protein synthesis by MicC sRNA in Salmonella. MicC pairs with the sequence within the internal portion of ompD mRNA (codons 23–26) where formation of the RNA duplex does not affect ribosome binding nor progression; rather it acts by eliciting mRNA cleavage by RNAse E (Pfeiffer et al., 2009).
Occasionally, sRNA pairing within the 5′ untranslated region of an mRNA can prevent the target sequence from participating in a secondary structure that limits ribosomal access. In such cases, the interaction stimulates translation initiation and the sRNA thus acts as positive regulator (Majdalani et al., 2005; Frohlich and Vogel, 2009). Interestingly, some sRNAs can downregulate certain genes while activating others. For example, RyhB, an sRNA synthesized under iron-limiting conditions, inhibits translation of several mRNAs-encoding iron-storage proteins but activates translation of shiA mRNA, which encodes a permease for shikimate, a siderophore precursor (Prevost et al., 2007).
Unlike transfer RNAs, the other major class of ‘small RNAs’ in living cells, which all conform to a strictly conserved construction plan, bacterial regulatory sRNAs stand out for their seemingly erratic domain organization. Some sRNAs pair with mRNA targets through their 5′ ends (Bouvier et al., 2008; Guillier and Gottesman, 2008) while others have their pairing domain internally (Geissmann and Touati, 2004; Prevost et al., 2007) or even adjacent to the transcription termination hairpin (Figueroa-Bossi et al., 2009; Rasmussen et al., 2009). Lack of a common architecture, combined with the elusive nature of the Hfq-binding determinant, makes it difficult to predict the sequence elements required for sRNA activity. Identification of such elements can profit from the analysis of regulatory mutants in which sRNA activity is impaired (Bossi and Figueroa-Bossi, 2007; Figueroa-Bossi et al., 2009). In the present study, this mutational approach was used to dissect the functional anatomy of RybB, a small RNA that downregulates the synthesis of several outer membrane proteins (OMPs) in E. coli and Salmonella (Vogel, 2009). Initially identified in E. coli cells from stationary cultures (Wassarman et al., 2001), RybB was later found to be under the control of alternative sigma factor σE and to accumulate under conditions causing or mimicking envelope stress (Johansen et al., 2006; Papenfort et al., 2006; Thompson et al., 2007). RybB targets include the mRNAs for major outer porins, such as OmpC, OmpD, OmpN and OmpW (Papenfort et al., 2006; Bouvier et al., 2008). By preventing excessive accumulation of these proteins, RybB plays an important role in the maintenance of envelope homeostasis. An in vitro study of the RybB interaction with ompN mRNA showed that the initial 16 nucleotides of the sRNA form an RNA duplex with the portion of the mRNA spanning from the second to the seventh codon (Bouvier et al., 2008). Although neither the Shine–Dalgarno sequence nor the initiation codon were included in the duplex, the structure effectively prevented binding of the ribosome to the mRNA (Bouvier et al., 2008).
To further characterize RybB and to study its action at additional targets, we designed a genetic screen for randomly induced mutations that altered the ability of RybB to downregulate ompC and ompD mRNAs. Characterization of these mutants confirmed the role of the 5′ sRNA portion in target recognition, as well as the contribution of specific residues to the selectivity and strength of the regulation. Mutations in the central portion of RybB identified residues involved in Hfq binding. A change near the 3′ end of the sRNA revealed a role of the transcription termination hairpin in RybB function.
The genetic screen
Mutants of RybB were generated through a procedure that combines mutagenic PCR with lambda red recombination (Bossi and Figueroa-Bossi, 2007). Chromosomal recombinants were screened in a Salmonella strain carrying a copy of the rpoE (σE) gene under the control of the chromosomal PBAD promoter, and an ompC–lacZ fusion whose regulatory range lies within the sensitivity window of MacConkey indicator medium (see Experimental procedures). Mutant candidates, identified by their altered colony colour on MacConkey-lactose plates supplemented with arabinose, were purified and analysed by DNA sequencing. This analysis identified isolates carrying mutations in the chromosomal rybB locus. These mutants fell into two general classes. One class carried changes in the rybB promoter region that cause the −35 or the −10 motifs to depart from the consensus sequence of σE promoters (Fig. S1). These mutations presumably relieve ompC repression by lowering rybB transcription and were not further analysed. A second class of mutants, the majority in our study, contained changes within the structural portion of the rybB gene. These mutations cluster at three separate locations, corresponding to the 5′ end, the centre and the 3′ end of the sRNA (Fig. 1A). The 5′ end mutations mapped within the segment previously shown to base pair with ompN mRNA (Bouvier et al., 2008).
In addition, the mutagenic PCR approach was applied to a DNA region extending about 500 bp on either side of the ompC AUG start codon. Mutants were screened for increased ompC–lac expression as above. Most of the mutations obtained map upstream from the main ompC promoter. These changes increase ompC transcription from either the main promoter or from a second weaker promoter further upstream (data not shown). Two mutations, however, fall within the 5′ untranslated region (5′UTR) of ompC mRNA, within a 10 nucleotide sequence complementary to the 5′ end of RybB (Fig. 1A). One of the two alleles, ompC U36A, is predicted to restore base pairing in the presence of the rybB allele A4U. Strains carrying either one or both of these mutations were constructed, and ompC expression was quantified. The results in Fig. 1B show that, separately, ompC U36A and rybB A4U cause increased expression of the ompC–lacZ fusion as well as the accumulation of ompC mRNA. However, both these effects are suppressed when the two alleles are combined (Fig. 1B). This suggests that the repression mechanism involves a base pair interaction and that the resulting decline in ompC mRNA levels reflects the destabilization of this mRNA. The observation that ompC U36A affects regulation to a greater extent than rybB A4U might be explained by a greater difficulty to accommodate an A : A mismatch, as opposed to a U : U in the duplex structure (Du et al., 2005). The second ompC mutation, A31G, causes an A : U base pair to be replaced by a G : U base pair. This mutant is more difficult to interpret given the conservative nature of the change and its lateral position in the duplex. Perhaps the A31G change antagonizes RybB pairing indirectly, by affecting the size of a secondary structure on the 5′ side of the duplex region (see Fig. S3). The portion of the RybB sequence that pairs with ompC mRNA is the same previously shown to interact with ompN mRNA (Bouvier et al., 2008). However, while in the latter and in other RybB-regulated mRNAs (see below), the target sequence lies within the coding region, the ompC target site is located in the 5′UTR, relatively distant from the start codon.
To extend the mutational analysis, RNA from ompC 5'UTR was synthesized in vitro and its interaction with in vitro-made RybB RNA or with purified Hfq protein was probed by footprinting techniques. The results in Fig. 1C show that RybB protects the segment between U29 and C39 of ompC mRNA from digestion by RNases A and T1, thus confirming the participation of this sequence in a RNA duplex. Moreover, the analysis indicates the presence of a secondary structure adjacent to the 5′ side of the pairing segment in ompC mRNA. This structure does not appear to be disrupted by the formation of the RNA duplex, suggesting that pairing is confined to the initial 10 nucleotides of RybB. Significantly, the portion of ompC mRNA protected by Hfq binding begins within the pairing segment and extends as many as 40 bases downstream (Fig. 1A and C). The entire region is A/U-rich and includes two AAUAA motifs proposed to bind Hfq with high affinity (Soper and Woodson, 2008; Figueroa-Bossi et al., 2009). The extended Hfq binding might reflect the participation of an additional sRNA in ompC regulation, MicC, which pairs with a 16 base sequence a few residues downstream from to the RybB pairing region (grey bar in Fig. 1A) (Chen et al., 2004).
The effects of rybB mutations on ompC and ompD regulation were assessed measuring β-galactosidase activity in strains carrying ompC::lacZ and ompD::lacZ translational fusions or monitoring ompC and ompD mRNAs by Northern blot hybridization. The results from the group of alleles at the 5′ end of the sRNA (the pairing mutants) are shown in Fig. 2A. Again, one can see that the β-galactosidase measurements closely parallel the Northern signals. Some of the mutations cause as little as a twofold effect on ompC regulation, underlining the sensitivity of the genetic screen. The largest defect is observed in a mutant with a single base deletion at position 2 of the sRNA. The ompC and ompD expression levels in this strain (fivefold and fourfold higher than the repressed levels, respectively) approach those of a strain deleted for the entire rybB gene, indicating that the ΔC2 allele completely inactivates the sRNA. Interestingly, certain alleles affect ompC and ompD expression differentially. The G7A change in particular, rather than relieving ompD repression, actually makes it tighter than in the wild type. Such hyper-repression is reflected both in the β-galactosidase activities and in the Northern blot analysis. Examination of the sequence on the 5′ side of the ompD gene reveals two potential RybB pairing sites resulting from the tandem duplication of the six base sequence AGUGGC in the region between the sixth and ninth codon of the mRNA (Fig. 2B). The G7A change is expected to increase the affinity of RybB for the site proximal to the AUG, as it replaces a G : U base pair with an A : U base pair. Thus, these findings tentatively indicate that the proximal site is recognized by RybB. The question is then whether the distal site can also function as a RybB pairing sequence. To answer it, identical mutations were introduced at corresponding positions of the two putative pairing sequences, as well as a compensatory change in the sRNA (Fig. 3A). The ompD G16C mutation allele does not have a significant effect on ompD expression while ompD G22C lowers expression below the wild-type level. Hyper-repression is also observed when either of these alleles is combined with compensatory change C5G in the sRNA (Fig. 3B). Thus, these results confirm that both sites are functional targets for RybB. As predicted, combining ompD G16C and ompD G22C leads to the complete loss of RybB-mediated regulation (Fig. 3B). Downregulation is restored by rybB C5G, supporting the involvement of base pairing in the regulatory mechanism. It should be noticed that the two ompD mutations result in the replacement of a valine codon (GUG) for a leucine codon (CUG). The fact that these two triplets are the most frequently used codons for the respective amino acids in E. coli (Hénaut and Danchin, 1996) rules out the possibility that changes in codon usage could have influenced the results.
Overall, the data in Figs 2 and 3 suggest that the two pairing interactions are mutually exclusive. The hyper-repressive patterns of some allele combinations are difficult to interpret and suggest that still unknown factors contribute to the effectiveness of the repression mechanism. For example, in strains carrying rybB mutation G7A, hyper-repression could be in part attributable to an increase in the steady state levels of RybB, which remains unexplained (see Fig. 2A).
The Hfq-binding mutants
All rybB mutations outside the pairing segment have similar effects on ompC and ompD regulation (Fig. 4; data not shown). With the exception of allele U70C (discussed below), these mutations cause RybB levels to decrease significantly, suggesting that they affect the stability of the sRNA rather than its activity (Fig. 4). Three alleles, U30C, G33A and C37U, affect nucleotides in the central portion of the sRNA near the bottom of the CG-rich stem loop structure that likely constitutes the Rho-independent transcription terminator (Fig. 1A). Because binding to chaperon protein Hfq is essential for RybB stability (our unpublished data), we speculated that the three mutations might lower the affinity of the sRNA for Hfq. To test this hypothesis, we compared the binding of wild-type and mutant sRNAs to purified Hfq in a gel mobility shift assay. Both U30C and G33A cause a small but reproducible reduction in the RybB affinity for Hfq (Fig. 5). The U30C allele shortens a run of four U residues, an alteration consistent with the known affinity of Hfq for poly U RNAs and A/U-rich tracks. In contrast, the defect of the G33A mutant is surprising given that the change increases the A/U richness in the region. The G33A position is at the hinge between a single-stranded region and the stem-loop terminator structure. Perhaps the change affects the relative orientations of these two domains in a way less favourable to the Hfq interaction. Footprinting experiments in Fig. 6 show that both the single-stranded region near the basis of the terminator stem and the sequence of the 5′ arm of the stem are weakly protected by Hfq, suggesting that both domains come in contact with the protein.
No significant alteration of Hfq binding could be detected in the C37U mutant (data not shown). This allele replaces a C : G base pair at the bottom of the terminator stem with an U : G pair. Although a conservative change, the position of the change might be critical for the overall stability of the structure. Destabilization of the terminator could in turn result in increased RybB turnover (see below).
The terminator mutants
A last group of rybB mutations are clustered within a short segment corresponding to the 3′ arm of the terminator stem. All these mutations disrupt base pairing in the lower portion of the stem and, with one exception, cause the partial or complete destabilization of the sRNA (Fig. 4). These findings support the notion that the terminator structure is a stabilizing element in prokaryotic RNAs (Mott et al., 1985; Aiba et al., 1991; Guarneros and Portier, 1991). Intriguingly, the U70C allele does not cause any detectable decrease in RybB levels (Fig. 4). To characterize this mutant, in vitro synthesized sRNA was subjected to ribonuclease probing. From the cleavage pattern in Fig. 7, it appears that U70C causes the terminator region to rearrange in an alternative configuration. In particular, the increase in the reactivity of G36, G57 and G58 to RNAase T1 and of U54 to RNAse V1 is consistent with a model in which the entire 5′ arm of the stem slides in a 5′ direction to allow a sequence in the apical portion of the arm (CCCGCC) to pair with the complementary segment generated by the U70C change (GGCGGG) (see Fig. 7B). As a result, the terminator stem is shortened while the single-stranded region on its 5′ side is extended. Somehow surprisingly, given the magnitude of the conformational change, gel mobility assays failed to reveal any alteration in Hfq binding (data not shown). Nonetheless, the latter findings are consistent with the in vivo stability of the mutant sRNA (Fig. 5). From the regulatory phenotype of the mutant, one would infer that the conformational rearrangement somehow hampers RybB ability to interact with its target sequences.
A new member of the RybB regulon
ChiP (YbfM) is an inducible porin that allows passage of chitin-derived oligosaccharides (chitobiose and chitotriose) across the outer membrane (Figueroa-Bossi et al., 2009). ChiP synthesis is normally inhibited by constitutively made ChiX (MicM) sRNA, which pairs with chiP mRNA-preventing translation (Figueroa-Bossi et al., 2009; Rasmussen et al., 2009). ChiX inhibition is relieved in the presence of chitosugars as a result of the accumulation of an RNA species that pairs with ChiX sRNA and promotes its degradation (Figueroa-Bossi et al., 2009; Overgaard et al., 2009). The presence of a sequence complementary to the 5′ end of RybB within the initial portion of the chiP coding sequence led us to examine whether chiP might be regulated by RybB. Initial tests, in a strain deleted for the chiX gene and carrying a chiP–lacZ fusion, showed the levels of β-galactosidase activity to nearly double upon deleting the rybB gene (data not shown). To confirm these data and assess the role of base pairing in the regulation, chiP mutations predicted to act as compensatory changes for RybB alleles A4U and C5G were isolated. The effects of the various changes on chiP expression were quantified measuring β-galactosidase activity in the chiP–lacZΔchiX background (Fig. 8). The results in Fig. 8B show that chiP and rybB mutations derepress chiP–lacZ when present separately but they lose their effects when combined in compensatory pairs. Thus, these findings conclusively identify chiP is a target for RybB regulation. The relatively narrow range of the regulatory response (less than a 50% variation) is likely ascribable to the shortness of the chiP:RybB RNA duplex (limited to seven base pairs).
RybB is a small non-coding RNA that is synthesized in response to σE activation and downregulates the expression of several major outer membrane proteins (Johansen et al., 2006; Papenfort et al., 2006; Thompson et al., 2007). Because overaccumulation of OMPs is one of the signals leading to σE activation, RybB activity contributes to the maintenance of OMP homeostasis and serves as basis for autogenous control. Here we used a mutational approach to define the RNA sequence elements that participate in RybB-mediated regulation. The study was carried out entirely in the Salmonella chromosome and relied on a genetic screen sensitive enough to detect mutations causing as little as a twofold effect in RybB-mediated regulation. The results identified the sequence of the 5′ end of RybB as the main determinant of recognition of ompC and ompD target mRNAs. The same region had been previously shown to interact with ompN mRNA (Bouvier et al., 2008), leading to the prediction that the 5′ end of RybB would be involved in the regulation of additional targets (Vogel, 2009). A similar pattern was recently demonstrated for two redundant sRNAs with a wide target range, OmrA and OmrB, which also pair with their targets through their 5′ ends (Guillier and Gottesman, 2008). Our genetic analyses indicate that the first seven bases of the RybB sequence play the predominant role in the interaction with either ompC or ompD mRNAs. This finding is strongly reminiscent of post-transcriptional regulation by eukaryotic miRNAs, where a seven base ‘seed’ region is the minimal requirement for regulation (Lewis et al., 2003; Brennecke et al., 2005). In ompC mRNA, a secondary structure near the 5′ side of the pairing region limits the total length of the duplex to 10 base pairs or less. The pairing segment, 40 nucleotides upstream from the start codon, lies just outside the portion of the mRNA covered by the 30S subunit of ribosome during initiation (Huttenhofer and Noller, 1994). Still, formation of the RybB:ompC RNA duplex might sequester residues needed for efficient docking of the ribosome onto the mRNA and thus slow down translation initiation (Sharma et al., 2007). Alternatively, formation of the duplex could stimulate cleavage of ompC mRNA by a ribonuclease (Masséet al., 2003; Morita et al., 2005). Footprinting experiments showed that Hfq binds the region between the RybB pairing sequence and the ompC AUG in vitro, consistent with the presence of two putative high-affinity binding sites (AAUAA) within the region. Intriguingly, the upstream motif is part of the sequence recognized by MicC sRNA (Chen et al., 2004). This raises the possibility that conditions that stimulate MicC activity could be incompatible with optimal (Hfq-assisted) RybB recognition of its target sequence. The second AAUAA motif lies in the interval between the Shine–Dalgarno sequence and the initiation codon. Hfq binding to this site might exert a direct effect on ompC mRNA translation, either contributing to obstruction of the ribosome entry site and/or causing ompC mRNA destabilization through the recruitment of RNAse activity (Morita et al., 2005). This possibility is in agreement with genetic evidence indicating that the effects of hfq mutations on OMP homeostasis are partially independent of MicA and RybB sRNAs (Bossi et al., 2008).
The study of ompD regulation showed the presence of two seed regions for RybB, one spanning codons 5 to 7, the other spanning codons 7 to 9. Surprisingly, mutations that affect pairing at either site increase repression, suggesting that the redundancy is somehow detrimental to RybB activity. Perhaps collisions can occur between RybB molecules trying to pair simultaneously at the two positions.
In the course of this study, we found that the chitoporin gene, chiP, is an additional RybB-downregulated gene in Salmonella. Synthesis of ChiP protein was recently shown to be specifically induced in the presence of chitosugars in the growth medium (Figueroa-Bossi et al., 2009). The induction mechanism involves inactivation of a different sRNA, ChiX (MicM), which normally inhibits translation of chiP mRNA (Figueroa-Bossi et al., 2009; Overgaard et al., 2009). ChiP accumulates massively following induction (Figueroa-Bossi et al., 2009). Therefore, it is not surprising that chiP may be also subject to the RybB-dependent homeostatic control as other major OMPs.
Strains and growth conditions
Strains used in this study were all derived from Salmonella enterica serovar Typhimurium strain MA3409 (Figueroa-Bossi et al., 1997). The genotype of relevant strains used are shown in Table 1. Bacteria were cultured at 37°C in liquid media or in media solidified by the addition of 1.5% Difco agar. LB broth (Bertani, 2004) was used as complex medium. Carbon-free medium (NCE) (Maloy and Roth, 1983), supplemented with the appropriate carbon source, was used as minimal medium. Antibiotics (Sigma) were included at the following final concentrations (in LB): chloramphenicol, 10 µg ml−1; kanamycin monosulphate, 50 µg ml−1; sodium ampicillin 75 µg ml−1; spectinomycin dihydrochloride, 80 µg ml−1; tetracycline hydrochloride, 25 µg ml−1. LB plates containing 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (Xgal; from Sigma), 40 µg ml−1, were used to monitor lacZ expression in bacterial colonies. Liquid cultures were grown in New Brunswick gyrotory shakers and growth was monitored by measuring the optical density at 600 nm with a Milton-Roy Spectronic 301 spectrophotometer.
Table 1. Relevant Salmonella strains used in this work.
Restriction enzymes, T4 polynucleotide kinase and Taq DNA polymerase were from New England Biolabs, Pfu-Turbo DNA polymerase was from Stratagene, T4 DNA ligase was from USB. DNA oligonucleotides were obtained from Sigma Aldrich. The sequences of the oligonucleotides used in this work are shown in Table 2. Acrylamide-bisacrylamide (30%, 29:1) and other electrophoresis reagents were from Bio-Rad. Agarose was from Invitrogen. Hybond-N+ membranes and hybridization buffer used for Northern blot analysis were from GE Healthcare and from Applied Biosystems-Ambion respectively. MegashortScript T7 kit, Kinase Max kit, RNAses A, V1 and T1, yeast tRNA were all from Applied Biosystems-Ambion.
Changes generating mutations are shown in bold and underlined. Bold lettering in primers pp952 and ppC46 denotes the sequence of the T7 promoter. Sequence extensions providing homology for recombination in gene swapping experiments are in italics.
Generalized transduction was performed using the high-frequency transducing mutant of phage P22, HT 105/1 int-201 (Schmieger, 1972) as described (Lemire et al., 2008). Chromosomal engineering was carried out by the λred recombination method (Datsenko and Wanner, 2000; Yu et al., 2000) as previously described (Uzzau et al., 2001). Donor DNA fragments were generated by PCR using plasmid or chromosomal DNA templates. Amplified fragments were introduced into strains expressing phage λred operon from plasmid pKD46 (Datsenko and Wanner, 2000). When required, the antibiotic-resistance cassette introduced by recombination was excised upon transforming strains with plasmid pCP20, which expresses the Flp recombinase (Cherepanov and Wackernagel, 1995). Preparation of recipient bacteria, DNA electroporation and isolation and processing of recombinant clones were carried out as described (Lemire et al., 2008).
Construction of relevant strains
The constructs described below were obtained by the λred recombination method. DNA oligonucleotides used as primers for the PCR reactions are listed in Table 2. Template plasmids were pKD3, pKD4 and pKD13 (Datsenko and Wanner, 2000).
Construction of lac fusions
To readily identify chromosomal mutations affecting RybB-mediated regulation of ompC, at the beginning of this study, we sought to develop a reporter system suitable to reveal such mutants directly on lactose indicator plates. The strategy used involved: (i) inserting a promoterless lac operon (from plasmid pCE36; Ellermeier et al., 2002) immediately downstream from the transcription terminator of the ompC gene in a strain with a chromosomal PBAD–rpoE fusion; (ii) selecting Lac+ derivatives resulting from terminator deletions fusing ompC to lacZ and (iii) identifying Lac+ colonies whose colour on MacConkey agar-lactose (1%) plates turned from red to pink when the medium was supplemented with 0.2% arabinose. Insertion of the lac operon downstream from ompC was achieved integrating an aph cassette [conferring kanamycin resistance (KnR), amplified from plasmid pKD4 with DNA oligonucleotide primers pp893 and pp894] followed by its Flp recombinase-mediated replacement with suicide plasmid pCE36 as described (Ellermeier et al., 2002). Lac+ mutants were selected on NCE plates containing 0.2% lactose as the sole carbon source. For the colorimetric screen, Lac+ mutants were replica plated on MacConkey-lactose plates with or without arabinose. One of the isolates showing the colour change was analysed further. Sequence analysis revealed that the deletion, ompC6–lac, removes approximately two thirds of the ompC gene but does not generate an in frame fusion. Rather, the lacZ initiator AUG codon falls right at the junction with ompC DNA (overlapping, in − 1 frame, a UGA stop codon at the join point), suggesting that lacZ expression results from translational reinitiation at the ompC–lacZ boundary (Fig. S4). Finding that ompC6::lac responds to RybB regulation suggests that reinitiating ribosomes originate from the translation of the upstream ompC segment. The fusion served as basis for the screening of rybB mutants (see below).
For quantitative measurement of ompC or ompD gene expression, in frame lacZ fusions were constructed. The procedure involved the insertion of the aph module of plasmid pKD4 in each of the genes, and Flp recombinase-mediated conversion of the insert to a lac fusion with plasmid pCE40 as described (Ellermeier et al., 2002). The primers used for the amplification of the aph cassette were ppB22 and ppB23 for the ompC construct, and pp998 and ppA86 for the ompD construct. Primer extension sequences were chosen so as to place the fusion boundary within ompC's or ompD's signal peptide-encoding segments. Disruption of the signal sequence was intended to prevent deleterious effects that might result from translocation of the β-galactosidase moiety of the hybrid proteins across cell membranes. In the final constructs, ompC94::lac and ompD96::lac, the lacZ coding sequence is fused to the 7th codon of ompC or ompD (Fig. S4).
Chloramphenicol-resistance (CmR) markers linked to rybB or ompC genes were obtained amplifying the cat module of plasmid pKD3 with the primers below, and using the resulting fragment for λred mediated recombination. Insertion Δ[STM0869]::cat (primers ppA11 and ppA12) places the cat gene 55 base pairs (bp) downstream from rybB gene in parallel orientation. Allele ΔyojN::cat (primers ppB31 and ppB32) places the cat gene 507 bp upstream from the coding sequence of the ompC gene in opposite orientation. A chromosomal PBAD promoter fusion to the σE (rpoE) gene (Δ[araBAD]76::Tn5TPOP45 PBAD-rpoE), obtained replacing the araBAD portion of the ara operon with the rpoE coding sequence was described in a previous study (Bossi and Figueroa-Bossi, 2007).
Random PCR mutagenesis
Chromosomal DNAs encompassing the region to be mutagenized and the linked cat marker (see above) were amplified by PCR under error-prone conditions (Bossi and Figueroa-Bossi, 2007). For rybB mutagenesis, the template was chromosomal DNA from strain MA9147 (Table 1) and the primers ppA40 and ppA41 (Table 2). For mutagenesis of ompC 5′UTR, template was chromosomal DNA from strain MA9242 (Table 1) and the primers were ppA62 and ppA63 (Table 2). In both amplification experiments, the priming sites were chosen so as to have approximately 0.5 Kb at both ends of the DNA products providing homology for recombination. The fragments obtained (2035 bp and 2043 bp, respectively) were used to transform strain MA9133 (ompC6::lac PBAD-rpoE) and CmR recombinants selected on MacConkey-lactose plate supplemented with 0.2% arabinose. Colonies of darker red colour than the background colonies were picked and purified by streaking on selective plates. The mutagenized region was PCR-amplified (ppA40/ppA8, RybB; pp832/ppA68 ompC 5′UTR) and subjected to DNA sequence analysis.
To obtain rybB allele C5G, a DNA fragment amplified from chromosomal DNA (strain MA9147) with primers ppD50 (which contains the desired base change) and ppA41 was introduced into strain MA9133 (Table 1) and CmR recombinants selected on MacConkey plates supplemented with 1% lactose and 0.2% arabinose. DNA from colonies of darker red colour was confirmed to contain the desired mutation. To introduce changes in the RybB pairing segment of ompD (alleles G16C, G22C and the double change), the procedure used was the same as in the isolation of the aph insertion in the signal sequence (above), except that forward primer was ppD47 in the making of G16C, ppD48 for G22C, and ppG24 to produce the double mutant. Reverse primer was ppA86 in all cases. Template was pKD13 plasmid DNA. Presence of the appropriate mutations in selected recombinant clones was confirmed by sequence analysis. Flp-mediated excision of the aph insert restored the ompD reading frame downstream from the mutated sites. These constructs were used directly for ompD mRNA analysis or they were converted to lacZ fusions for β-galactosidase determinations (see above). For the isolation of chiP mutations G14C and U15A, the aph cassette in strain MA9130 (chiP::aph; Table 1) was amplified by PCR using oligonucleotides ppG31 (G14C change) or ppG32 (U15A change) as forward primers, and ppA85 as reverse primer. The fragments so obtained were used to transform a λred-expressing recipient and KnR recombinants inheriting the desired changes identified by DNA sequencing. Constructs were then converted to lac fusions as described above.
RNA extraction and Northern analysis
RNA was prepared by the acid-hot-phenol method from exponentially growing cells (OD600 of 0.35) as previously described (Bossi and Figueroa-Bossi, 2007). RNA was separated on a 1.2% agarose-formaldehyde gel (ompC and ompD mRNAs) or on an 8% polyacrylamide-8 M urea gel (RybB) and blotted onto a nylon membrane. Blots were hybridized to 5′ end-labelled DNA oligonucleotide probes specific for the RNAs under study or for RNAs serving as internal standards (ssrA, 5S RNA). Oligonucleotides used were pp891 (ompC), pp931 (ompD, Figs 2 and 4) ppG25 (ompD, Fig. 3), pp929 (RybB), pp813 (ssrA) and ppB10 (5S RNA). The sequences of these probes are shown in Table 2. In the experiments in Figs 2 and 4, blots hybridized with pp891 were stripped of the label (2 × 15 min in hot 0.1% SDS) and re-hybridized with probe pp931.
Activity of β-galactosidase was measured in toluene-permeabilized cells as described by (Miller, 1992) and is expressed in Miller units. Typically, the activity was measured in 10 µl of cultures grown to stationary phase. Reported values were the average of at least two independent determinations, each involving duplicate or triplicate samples.
In vitro RNA synthesis
Templates for in vitro transcription with T7 polymerase were generated by PCR from genomic DNA. To produce the ompC 5′UTR template, DNA was amplified with primers pp952 and pp953. To make the wild-type rybB template, primers were ppC46 and ppC47, while primers ppC46 and ppC57 were used to produce the rybB U70C allele (Table 2). In vitro transcription was performed with the MegashortscriptT7 kit (Ambion AM1354) according to the manufacturer's protocol. Transcription products were an 108 nucleotide RNA corresponding to the first 104 residues of ompC mRNA plus a four nucleotide extension from the T7 promoter (GGGU) at the 5′ end, and full-length RybB with an extra G at its 5′ end. After incubation for 2 h at 37°C, DNAse was added and incubation continued for further 15 min. The samples were treated with phenol and the RNA precipitated at − 20°C overnight with sodium acetate-ethanol-glycogen. RNA was recovered by centrifugation, resuspended in water and quantified by nanodrop reading. The RNA solution was adjusted to 5 pmol µl−1. RNA (10 pmol) was dephosphorylated and labelled at its 5′ end with [γ−32P]ATP (3000 Ci/mmol from Perkin Elmer) using the KinaseMax kit (Ambion AM1520). Labelled RNA was purified by 8% PAGE. The RNA band was eluted from the gel, phenol-extracted, ethanol precipitated at − 20°C and resuspended in water. Before use for Hfq binding or structural studies, RNA was heated in refolding buffer (50 mM Tris pH8, 0.1 M NaCl, 0.1 M KCl 1 mM MgCl2) at 85°C for 3 min, followed by 20 min at room temperature and then placed on ice.
Gel shift assays
7His-tagged Salmonella Hfq protein was purified as described (Figueroa-Bossi et al., 2009). Labelled ompC 5′UTR or rybB RNA (4 nM), was incubated with increasing amounts of protein Hfq as indicated, in 50 mM TrisHCl pH 7.8, 50 mM NaCl, 50 mM KCl, 10 mM MgCl and 6 µM yeast tRNA at 30°C for 30 min. Binding reactions were loaded on a 5% non-denaturing polyacrylamide gel. Electrophoresis was in 0.5 X TBE buffer at 4°C for 3 h at constant current of 15 mA. Results were analysed by phosphorimaging using ImageQuant software.
RNA Structural analyses
Enzymatic treatments were performed in 10 µl of reaction mix containing 0.2 pmol of RNA, 1 µg of yeast tRNA, 1 × Structure buffer (Ambion) and 2 pg of RNase A (Ambion AM2274, 1 µg ml−1) or 0.01 U of RNase T1 (Ambion AM2283, 1 U µl−1) or 0.01 U of RNAse V1 (Ambion AM2275, 0.1 U µl−1). Incubation was at 37°C for 12 min. Reactions were stopped by addition of 20 µl of ‘Precipitation/Inactivation’ buffer from the same manufacturer. Partial alkaline hydrolysis was performed according to Ambion's protocol as follows: 10 µl of reaction mix containing 0.2 pmol of RNA, 1 µg of yeast tRNA, 1 × Alkaline Hydrolysis buffer, were incubated at 95°C for 8 min, placed on ice and 20 µl of Precipitation/Inactivation buffer added. After recovery from precipitation, all samples were run on a 10% sequencing polyacrylamide gel in 0.5 × TBE. Results were analysed by phosphorimaging.
We thank Danièla Maloriol for technical assistance in the initial phase of this study. We are grateful to Modesto Carballo of the Servicio de Biología (CITIUS, Universidad de Sevilla) for assistance in experiments performed at the facility. This work was supported by Grant BLAN07-1_187785 from the French National Research Agency (to L.B.), and BIO2007-67457-CO2-02 and CSD2008-00013 from the Spanish Ministry of Science and Innovation (MCINN) and the European Regional Fund (to J.C.).