Evidence for an autonomous 5′ target recognition domain in an Hfq-associated small RNA

Abstract

The abundant class of bacterial Hfq-associated small regulatory RNAs (sRNAs) parallels animal microRNAs in their ability to control multiple genes at the posttranscriptional level by short and imperfect base pairing. In contrast to the universal length and seed pairing mechanism of microRNAs, the sRNAs are heterogeneous in size and structure, and how they regulate multiple targets is not well understood. This paper provides evidence that a 5′ located sRNA domain is a critical element for the control of a large posttranscriptional regulon. We show that the conserved 5′ end of RybB sRNA recognizes multiple mRNAs of Salmonella outer membrane proteins by ≥7-bp Watson–Crick pairing. When fused to an unrelated sRNA, the 5′ domain is sufficient to guide target mRNA degradation and maintain σ^E-dependent envelope homeostasis. RybB sites in mRNAs are often conserved and flanked by 3′ adenosine. They are found in a wide sequence window ranging from the upstream untranslated region to the deep coding sequence, indicating that some targets might be repressed at the level of translation, whereas others are repressed primarily by mRNA destabilization. Autonomous 5′ domains seem more common in sRNAs than appreciated and might improve the design of synthetic RNA regulators.

Small RNAs that act on trans-encoded target mRNAs by short base pairing are important posttranscriptional regulators in many organisms. The two most abundant classes to date are the microRNAs of eukaryotes and the Hfq-associated small regulatory RNAs (sRNAs) of gram-negative bacteria such as Escherichia coli and Salmonella (1). The ∼22-nt microRNAs use well-established mechanisms and machinery to repress mRNAs by short seed pairing within the 3′ UTR, and a single microRNA might regulate hundreds of genes in parallel (2).

The bacterial sRNAs are also increasingly found to control multiple targets, although by binding the 5′ region of bacterial mRNAs (3). In fact, some sRNAs have an impact on dozens of genes under stress or altered growth conditions (4–7), with target sites being known for a subset of the regulated mRNAs. The Sm-like protein, Hfq, is required for intracellular stability and target annealing of the sRNAs (1, 8). Global analyses of Hfq-bound transcripts suggest an excess of potential targets over regulators (9, 10), further arguing that multiple targeting might be the general mode of sRNA action.

Unlike the universal length and seed pairing of microRNAs, there are few common denominators for Hfq-dependent regulators. The sRNAs dramatically vary in size (50–250 nt) and secondary structure (11), and sRNA-mRNA interactions range from >30-bp duplexes in MicF-ompF or Spot42-galK (12, 13) to only 6 bps that are critical in SgrS-ptsG (14). Most of the sRNAs analyzed to date inhibit translational initiation of targets by sequestering the Shine–Dalgarno (SD) or start codon (AUG) sequences of the ribosome binding site (RBS); how the recognition of these conserved RBS elements would ensure highly specific target selection is little understood (15). Interestingly, there are a few recent examples of sRNA binding upstream or downstream of SD/AUG sequences (16–21), suggesting that the mRNA window for target repression could be broader than the RBS. Emerging evidence suggests that those sRNA nucleotides interacting with multiple targets might be maintained by selection (18) and often cluster in the 5′ sRNA region (6, 16, 20, 22–24).

The ∼80-nt RybB sRNA was recently shown to accelerate mRNA decay of many major and minor outer membrane proteins (OMPs) in Salmonella and E. coli (25, 26). RybB is activated by the alternative σ^E factor when excessive OMP synthesis or envelope damage causes periplasmic folding stress (25–28). As such, RybB is a major facilitator of σ^E-directed global OMP repression (25, 29) and is required for envelope homeostasis and feedback regulation of σ^E (25, 27).

Pulse expression of Salmonella RybB reduces the half-lives of stable omp mRNAs from of ≥10 min to ∼1 min (25), and both RybB and its putative targets interact with Hfq (10), all of which predicts direct regulation by RNA interactions. Biocomputational algorithms failed to predict statistically significant RybB pairing with the RBS of targets (25), however, and the only known RybB site thus far was discovered in the coding sequence of Salmonella ompN mRNA (17).

Here, we report genetic evidence that the conserved 5′ end of RybB constitutes an autonomous multiple-target binding domain used to select many omp mRNAs by short (≥7 bp) Watson–Crick pairing. This domain is essential and sufficient for target repression and adjusting σ^E activity. RybB sites in omp mRNAs are associated with a unique 3′ adenosine signal and are located in a broad window ranging from 5′ located stability elements into the deep coding sequence, suggesting varying contributions of translational inhibition and mRNA decay to target repression. An autonomous 5′ domain might be more common in sRNAs than appreciated and also governs the prototypical MicF-ompF regulation.

Results

OMP Depletion by Constitutive RybB Expression.

RybB effects have been studied at the omp mRNA level following short expression of the sRNA (25, 26). To assay regulation at the protein level, we used the constitutive expression plasmid pP_L-RybB (17), which confers equivalent RybB levels throughout growth (Fig. 1) and approximately fivefold more than the chromosomal rybB gene after σ^E induction (SI Appendix, Fig. S1). We observed full depletion of three 30- to 35-kDa proteins—OmpA, OmpC, and OmpD—known as the most abundant Salmonella porins during regular growth (Fig. 1). Depletion of the fourth major porin, OmpF, was confirmed using strains with individual omp gene disruptions (SI Appendix, Fig. S2). In contrast, there was no effect on the OmpX porin (Fig. 1) whose mRNA is refractory to RybB in Salmonella (25), suggesting that repression was specific and not caused by a general block of OMP biogenesis. Because constitutive RybB expression did not affect bacterial growth rate or viability, it lent itself to genetic dissection of RybB target pairing.

Fig. 1.

Constitutive RybB expression causes porin depletion in Salmonella. RNA and protein samples were prepared from WT Salmonella carrying plasmids pJV300 (empty vector; lanes 1–4) or pP_L-RybB (lanes 5–8) after growth to exponential phase (OD₆₀₀ of 0.5) or stationary phase (OD₆₀₀ of 1.5, 2, or 3.4). (Top) Autoradiograms of a Northern blot probed for RybB or 5S rRNA (loading control). (Middle) Coomassie-stained SDS gel of corresponding total protein. The major OmpA/C/D porin bands are labeled. (Left) Comigrating marker proteins; sizes are expressed in kilodaltons. (Bottom) Western blot using antisera against major porins or OmpX (control) as indicated.

5′ Domain of RybB Regulates Multiple omp mRNAs.

Alignment of RybB homologs of diverse γ-proteobacterial species revealed that this sRNA possesses an exceptionally well-conserved 5′ end (17). Moreover, we previously obtained evidence that RybB nucleotides 1–16, dubbed R16 (Fig. 2A and SI Appendix, Fig. S3), are sufficient for base pairing and repression of ompN mRNA (17). To this end, we grafted R16 onto either of two control sRNAs, TMA or TOM; the resulting chimeric sRNAs, R16TMA or R16TOM (Fig. 2B), repressed ompN invariably from wild-type (WT) RybB (17).

Fig. 2.

First 16 nucleotides of RybB are sufficient for porin repression. (A) Structure of Salmonella RybB RNA based on alignment in SI Appendix, Fig. S3 and probing (17). A horizontal bar denotes the R16 sequence. (B) Schematic drawing and color code explaining the origin of chimera RNAs (R16TMA and R16TOM) and control RNAs (TMA and TOM) from Salmonella RybB, MicA, and OmrB. (C) Comparison of target regulation in a Salmonella ΔrybB strain transformed with pJV300 (lane 1) or pP_L-RybB (lane 2) or in plasmids expressing the following RybB mutants from a P_L promoter: Δ1–9 (lane 3); R16TMA or R16TOM, which carry the R16 sequence (lane 4 or 5); and their parental control RNAs, TMA (5′ truncated MicA; lane 6) or TOM (5′ truncated OmrB; lane 7). (Upper) Bacteria were cultured to an OD₆₀₀ of 2.0, and major porin regulation was determined on a stained SDS gel (first panel) or by Western blot analysis (second panel). (Lower) Western blot analysis of regulation of translational gfp fusions to the 5′ regions of RybB target mRNAs (fadL: −100/+63 relative to AUG, ompA: −133/+45, ompC: −79/+36, ompD: −66/+36, ompF: −112/+36, ompS: −66/+36, ompW: −29/+36, tsx: −76/+36) or to ompX::gfp or gfp control vectors (SI Appendix, Fig. S5). (Right) Abundance of GFP fusion proteins was assayed after overnight growth for 12 h. The ompA::gfp fusion was analyzed in an E. coli ΔrybB strain.

To determine whether R16 also regulated other targets, we assayed effects of R16TMA or R16TOM and of a previously undescribed 5′ deleted variant of RybB (Δ1–9; Fig. 2B) on the accumulation of abundant OMPs (Fig. 2C, Top). Both R16TMA and R16TOM down-regulated the major OMPs (lanes 4–5), whereas their parental TMA and TOM RNAs did not (lanes 6–7), suggesting a broader role of R16 in OMP repression. The Δ1–9 truncation rendered RybB unable to deplete major OMPs (lane 3), corroborating that an intact 5′ end is pivotal to regulation. Importantly, all tested sRNAs were transcribed from the same constitutive promoter as in pP_L-RybB and accumulated to comparable levels in vivo (SI Appendix, Fig. S4). Thus, RybB-mediated OMP repression is stringently correlated with the presence of the R16 sequence.

Bacterial sRNAs typically target the SD or AUG sequence, which can be assayed in vivo using a well-established translational gfp reporter system (30). We constructed reporter plasmids for eight RybB targets (Fig. 2C and SI Appendix, Fig. S5), fusing the 5′ UTR and up to 21 codons to the NH₂ terminus of GFP. All fusions are constitutively transcribed to score specifically for posttranscriptional regulation. Reporters were individually combined in Salmonella with the above WT or mutant RybB plasmids, and regulation of the fusion proteins was determined by Western blot analysis (Fig. 2C, Bottom).

Wild-type RybB repressed all target fusions (Fig. 2C, lane 2), predicting that the cloned mRNA regions contained a RybB target site; these repressions were specific because RybB did not regulate gfp alone or an ompX::gfp control fusion. As with the major OMPs, the RybBΔ1–9 variant did not regulate any of the fusions (Fig. 2C, lane 3). In contrast, R16TMA and R16TOM down-regulated all fusions (Fig. 2C, lanes 4 and 5) to the same degree as WT RybB; we attributed these regulations to R16 because the parental control RNAs (TMA and TOM) alone had no impact on reporter activity (Fig. 2C, lanes 6 and 7). Together, these results narrowed down the sequence space for the prediction of RybB binding.

RybB Target Sites in omp mRNAs.

We used the RNAhybrid algorithm (31) to predict RNA duplexes of R16 with the above cloned 5′ target mRNA fragments. Fig. 3A shows the highest scoring interaction for each target, all of which have a change in free energy (ΔG°) in a −24.2 to −18.8 kcal/mol⁻¹ range, similar to the known RybB-ompN pairing (included for comparison) yet far more favorable than the −13 to −8 kcal/mol⁻¹ range expected for random RNA pairing (15). Intriguingly, each interaction involves 5′ terminal pairing of RybB, commonly of seven or more consecutive Watson–Crick pairs.

Fig. 3.

RybB target sites on omp mRNAs. (A) Antisense complementarity of R16 (red) with the above tested omp mRNA fragments (black) predicted by RNAhybrid, sorted by relative position to the AUG start codon. The RybB-ompN pairing (17) is shown for comparison. Numbers in parentheses denote mRNA nucleotides relative to AUG (A is +1); the predicted change in free energy of an interaction is given below. Vertical arrows denote the C₂→G mutation in RybB-M2 RNA and the compensatory G→C change in M2’ mRNA alleles. The 3′ adenosine flanking most RybB sites is marked in yellow. (B) Validation of selected interactions. Stationary phase Salmonella ΔrybB cells carrying gfp-target fusions in combination with RybB expression or control plasmids were analyzed by FACS. Results are plotted in fluorescence histograms derived from 30,000 events per sample. Cellular fluorescence is given in arbitrary units (GFP intensity). Histograms on the same line represent the WT or M2’ mutant fusions of the same target. Each fluorescence histogram represents triplicate results from pPl-RybB (red), pPl-RybB-M2 (blue), or pJV300 (black) samples.

The RybB target sites are strikingly diverse in relative mRNA position (Fig. 3A). Canonical binding sites overlapping SD/AUG must fall within a −16 to +3 window on mRNA (+1 is A of the start codon AUG); only tsx would be targeted so. Unexpectedly, RybB would recognize the majority of targets in the coding sequence (fadL, ompA, ompD, ompN, ompS, and ompW) or far upstream of the RBS (ompC and ompF). We used a previously published RybB-M2 mutant (17) carrying a C₂→G change to validate the six most representative interactions (Fig. 3B). Quantitative single-cell measurements of gfp reporter activity using FACS showed that, as expected, the M2 point mutation disrupting the 5′ helix of RybB-target duplexes (Fig. 3A) consistently abrogated target regulation (Fig. 3B, Left). Reciprocally, we constructed compensatory M2’ alleles of all six target fusions, seeking restoration of base pairing (Fig. 3A; G→C change at positions −42, −39, −8, +19, +31, and +49 in ompC, ompF, tsx, ompW, ompA, and fadL, respectively). In line with our predictions, the M2’ reporters were insensitive to WT RybB but regulated by RybB-M2 sRNA (Fig. 3B, Right). Including ompN pairing (17), seven RybB-omp interactions are now firmly validated in vivo.

5′ End of RybB Is an Autonomous Regulatory Domain.

Next, we asked whether the short R16-target duplexes were sufficient to repress native omp mRNAs in the σ^E response. To this end, we replaced the chromosomal RybB sequence with the R16TMA chimera while keeping the σ^E-dependent promoter of the rybB locus. Northern blots showed that following σ^E induction, WT RybB and R16TMA accumulated to comparable levels. Importantly, the chromosomally encoded R16TMA RNA down-regulated the native ompC/D/F/N/S target mRNAs with almost identical kinetics as did WT RybB, demonstrating that R16-mediated pairing is sufficient for target repression under physiological conditions (Fig. 4A).

Fig. 4.

Chromosomal R16 is sufficient to repress omp mRNAs in the σ^E response. The transcribed region of Salmonella rybB was chromosomally replaced with the R16TMA sequence. (A) WT, ΔrybB, and R16TMA strains were transformed with a pBAD-RpoE plasmid. Levels of omp mRNAs before or following arabinose induction of σ^E at exponential phase for 5 and 10 min were analyzed on Northern blots. RybB/R16TMA expression was detected with an LNA probe for the R16 sequence. (B) WT, ΔrybB, and R16TMA strains were grown to late stationary phase (OD₆₀₀ of 3), and the relative expression of two σ^E-responsive genes (rpoE and degP) was determined by quantitative RT-PCR. RNA levels in WT are set to 1. (C) Reporter activity (arbitrary fluorescence values) of the above strains but carrying a gfp fusion to the σ^E-dependent rybB promoter was determined at indicated stages of growth.

We and others have reported that ΔrybB strains suffer chronic envelope stress as judged by elevated σ^E activity even under standard growth conditions, likely attributable to excessive OMP synthesis in the absence of RybB (25–27). To test whether R16 could counteract envelope stress on its own, we compared the levels of rpoE and degP mRNA, two sensitive markers of σ^E induction (29), among Salmonella WT, ΔrybB, and R16TMA strains at stationary phase. Fig. 4B shows that R16TMA fully reversed the twofold (degP) or 7.5-fold (rpoE) increase of these σ^E-dependent transcripts to WT levels. Next, we used a gfp reporter fusion to the highly σ^E-dependent rybB promoter (25) to assay envelope stress over growth (Fig. 4C). Samples taken at five points from late exponential through stationary phase consistently showed severalfold increased rybB-gfp reporter activity, hence chronic σ^E induction, in ΔrybB vs. WT Salmonella. In contrast, the σ^E activity in the R16TMA strain was indistinguishable from WT, suggesting full complementation. These results establish that the conserved R16 sequence is sufficient to control a global regulatory circuit under the conditions tested.

Discussion

Bacterial antisense regulation was traditionally associated with the extended base-pairing potential of cis-antisense RNAs transcribed opposite to a single target gene in plasmids and mobile genetic elements (32). Although structural analyses have revised the concept of long RNA duplexes (33, 34), few cis-antisense RNAs are found in sufficiently diverse bacteria to predict important RNA nucleotides by hint of sequence conservation. In contrast, Hfq-dependent sRNAs often act on multiple targets and carry short stretches of sequence conserved in a wider range of species. These characteristics have allowed us to identify R16 as a 5′ located functional domain that comprises ∼20% of RybB (Fig. 2A) and can act autonomously to control a functionally related yet structurally diverse family of mRNAs as well as the overall state of the cell (Fig. 4).

A survey of conservation patterns and target pairing of 18 well-characterized E. coli/Salmonella sRNAs shows that more than a third of them possess a conserved 5′ end and act by 5′ terminal pairing (SI Appendix, Figs. S6 and S7 and Table S1). Intriguingly, enrichment of conserved nucleotides toward the 5′ end was most prominent in MicF, the founding member of the class of trans-acting antisense RNAs (35). Similar to RybB, the conserved 5′ end of MicF is sufficient for target repression when fused to TMA RNA, suggesting that MicF also carries an autonomous 5′ target-binding domain (SI Appendix, Fig. S6). The principles of target recognition by 5′ terminal pairing sRNAs such as RybB and of the modular structure of sRNAs in general could facilitate the design of artificial regulators for the control of synthetic regulatory circuits.

Structure of RybB Target Sites.

Recent work has implicated conserved 5′ nucleotides in target pairing of several Hfq-dependent sRNAs (6, 16, 20, 22–24). For example, mutation in the 5′ region of the highly similar OmrA/B sRNAs impaired repression of several E. coli mRNAs under conditions of sRNA overexpression, prompting complementarity searches with this region to predict an additional OmrA/B target (22). However, whether the involved 5′ nucleotides sufficed for regulation under physiological conditions or needed to partner with other sRNA regions for productive target recognition remained unknown.

Our data suggests that target selection by RybB is fully determined by short 5′ terminal Watson–Crick pairing (Fig. 3A). First, sRNA chimeras carrying the R16 sequence (R16TMA and R16TOM) regulate targets almost indistinguishably from WT RybB. Second, σ^E-induced chromosomal RybB and R16TMA RNAs promote the same omp mRNA decay (Fig. 4); thus, R16 suffices for target recognition under native expression conditions. Third, the M2 point mutation at the second RybB position abrogates repression (Fig. 3) even if the typically short 5′ helix is followed by strong downstream pairing as in RybB-ompN (17) (Fig. 3A).

Conservation patterns in mRNAs further argue that target recognition is restricted to short 5′ terminal pairing of RybB. For example, alignment of enterobacterial ompC sequences reveals conservation of 5′ UTR nucleotides −50 to −41 complementary to RybB nucleotides 1–9 (Fig. 5A). Although the adjacent ompC nucleotides are conserved as well, they do not extend the RybB-ompC helix (Figs. 3A and 5A), suggesting counterselection of long pairing. Similar patterns mark RybB sites in other Salmonella omp mRNAs (SI Appendix, Fig. S8).

Fig. 5.

Conservation and distribution of RybB target sites. (A) Alignment of ompC 5′ UTRs of various enterobacteria (abbreviations given in SI Appendix, Fig. S3). Gray boxes indicate the conserved SD sequence and RybB target site. Inverted arrows indicate the 5′ hairpin. Red, blue, or black coloring indicates strong, medium, or weak conservation, respectively. (B) Approximate positions of RybB sites in omp target mRNAs. Most RybB targets are recognized in the coding sequence, and core binding generally occurs outside the canonical SD/AUG region. Local alignment of RybB target mRNAs (C) or targets regulated by other 5′ end-pairing sRNAs (D) (SI Appendix, Fig. S8). Graphs were generated using Weblogo (http://weblogo.berkeley.edu/logo.cgi).

The target sites in mRNAs share little sequence except for the four nucleotides recognized by the extreme 5′ end of RybB. However, almost all RybB sites are followed by an unpaired adenosine (Fig. 5C), and this 3′ adenosine signal is also observed with other 5′ terminal pairing sRNAs (Fig. 5D). Thus, despite radical differences in underlying machinery and mechanisms of target regulation, some features of bacterial 5′ pairing sRNAs are reminiscent of animal microRNAs, which select multiple targets by short Watson–Crick pairing of a 5′ located conserved “seed” (microRNA nucleotides 2–7), flanked by 3′ adenosine in the target (2, 36). Because these features are instrumental in the discovery of microRNA target sites, extra scores for conserved sRNA domains and a 3′ adenosine might help to improve target predictions in bacteria.

Mechanisms of mRNA Regulation by RybB.

Translational interference on the basis of sRNA complementarity to SD or AUG was proposed to serve as the primary mechanism of target repression (37), irrespective of concomitant mRNA degradation by recruitment of RNase E (38), which should relax the RBS dependence. A stringent requirement to target the conserved RBS elements is difficult to reconcile with sRNAs having to discriminate genuine targets from thousands of cellular messengers (39). Intriguingly, the RybB sites determined here reveal almost no tight contacts with SD or AUG (Figs. 3A and 5B). Instead, RybB operates within a >100-nt window around AUG, which is much larger than the −20/+15 window for general antisense inhibition of initiating 30S ribosomes previously suggested by structural and in vitro interference data (17, 40–42). Thus, during its evolution toward a global regulator, RybB might have selected the most suitable pairing region in the diverse omp targets, likely at the expense of stringent translational control. In other words, current models predict efficient competition of RybB with ribosomes for the tsx (SD sequestration) and ompD/N/S/W targets [“five-codon window” in the 5′ proximal coding sequence (17)].

In contrast, four RybB sites lie well outside the 30S contacts, either in the 5′ UTR (ompC and ompF) or in the coding sequence downstream of the fifth codon (ompA and fadL). Pairing in the deep coding sequence is unlikely to abrogate translation because elongating 70S ribosomes unwind RNA duplexes much stronger than RybB-fadL or RybB-ompA (43). Of note, MicC sRNA was recently shown to repress Salmonella ompD mRNA at codons 23–26 without 30S or 70S inhibition but induced RNase E-dependent decay (16). Similarly, we find that RybB requires RNase E for repression in the deep coding sequence (SI Appendix, Fig. S9), suggesting that RybB acts on fadL and ompA by alternative gene silencing downstream of translational initiation similar to MicC-ompD (16).

RybB pairing upstream of the ompC and ompF RBS could involve translational repression, induced mRNA decay, or both. 30S inhibition at such distal sites was recently reported for Hfq-dependent regulators: GcvB targets a translational enhancer element of gltI mRNA (18), RyhB targets a leader peptide of fur (19), and OmrAB targets a conserved RNA structure element of csgD (20). None of these models readily applies here, and possible translational control of ompC and ompF is currently being investigated. However, we note that RybB binds within a conserved 5′ hairpin of ompC mRNA (Fig. 5A). Such hairpins commonly stabilize bacterial messengers by impeding 5′ pyrophosphorylation and RNase E-mediated degradation (44, 45). Thus, RybB pairing to the 3′ flank of the ompC hairpin (−64 to −41) might impair full formation of this stability element to promote the accelerated ompC mRNA decay that follows σ^E induction (25, 26, 29). Altogether, the non-RBS sites support the emerging importance of induced mRNA decay in Hfq-dependent regulations (38, 46). Moreover, recent work on Hfq-dependent regulations involving RNase E showed that diverse sRNAs undergo coupled degradation with their targets (47, 48). In the case of RybB, such concomitant consumption of the sRNA with omp mRNAs would permit fast recovery of OMP synthesis once the σ^E-inducing envelope stress is alleviated.

Modular sRNA Structure and Implications for Synthetic Design.

Our key finding is that the 5′ end of RybB constitutes an autonomous target recognition domain that, on transfer to another sRNA, suffices to both control many OMPs and suppress activation of the σ^E response over growth (Fig. 4). The σ^E regulon is complex, comprising ∼100 genes (28, 29), and R16 can be predicted biocomputationally (31) to target additional mRNAs with σ^E-related functions. Interestingly, the other σ^E –dependent sRNA, MicA, also uses conserved 5′ proximal nucleotides to select multiple mRNAs (26, 49–52).

Despite its small size and simple structure, RybB now appears to carry three distinct modules: (i) a specificity domain (R16) determining the target range, and thus physiological activity; (ii) an A/U-rich Hfq site (53) in the 30-nt region for general target annealing; and (iii) a ρ-independent terminator required for the biogenesis and stability of this global regulator. We believe that knowledge of modular sRNA structure will permit better design of synthetic antisense regulators, a task pioneered for single genes more than 3 decades ago (54). Indeed, substitution of R16 with a complementary sequence of a generic mRNA successfully generated a unique regulator (17). The very short 5′ terminal pairing and relaxed position dependence of target sites revealed in this study might now be exploited for reprogramming of regulatory circuits or metabolism through multiple targeting with a synthetic 5′ end of RybB.

Experimental Procedures

Oligonucleotides, Plasmids, Bacterial Strains, and Growth.

Bacterial strains and their construction details are listed in SI Appendix, Table S2. Strains were grown at 37 °C in LB or on LB plates. Ampicillin (100 μg/mL), kanamycin (50 μg/mL), chloramphenicol (20 μg/mL), and L-arabinose (0.2%) were added where appropriate. Salmonella WT (SL1344) or mutant strains were transformed with plasmids by electroporation as described (55). DNA, LNA (locked nucleic acid), and RNA oligonucleotides as well as plasmids are listed in SI Appendix, Tables S3 and S4. RybB expression and control plasmids have been described (17, 25, 55). Details of plasmids pFM58-1 (expressing RybBΔ1–9), pKP24-2 (MicF), pMB14 (pBAD-F14TMA), and pMB15 (pBAD-TMA) are found in SI Appendix.

SDS/PAGE and Western Blot Analysis of Salmonella Proteins.

Culture samples were taken according to 1 OD₆₀₀ and centrifuged for 4 min at 16,100 g at 4 °C, and pellets were resuspended in sample loading buffer to a final concentration of 0.01 OD₆₀₀/μL. Following denaturation for 5 min at 95 °C, 0.1/0.2 OD₆₀₀ equivalents of sample were separated in small/large SDS gels, respectively. Western blot analysis and quantification followed previously published protocols (30). OmpX was detected as in (56). OmpA/C/D/F were detected with major porin antisera (1:50,000 dilution; provided by R. Misra, Tempe, AZ) and anti–rabbit-HRP (1:5,000 dilution; Sigma–Aldrich) as the secondary antibody.

RNA Isolation and Northern Blot.

Total RNA was prepared and separated in 5 or 6% (vol/vol) polyacrylamide 8.3-M urea gels (5–15 μg of RNA per lane) and blotted as described (25). Membranes were hybridized at 42 °C with gene-specific [³²P] end-labeled DNA oligonucleotides, at 60 °C with an ompC-specific [³²P] PCR product (25) or with LNA probe 07150 at 37 °C in Rapid-hyb buffer (GE Healthcare). Detection of other omp mRNAs was done as described (25). Hybridization signals were quantified on a FLA-3000 PhosphorImager (Fuji) with AIDA software (Raytest).

gfp Reporter Assays.

Target fusions to gfp were constructed as described by Urban and Vogel (30) and in SI Appendix, Tables S5. sRNA-dependent fusion regulation on Western blots was as described (30). For reporter activities in single cells (FACS), we used a Canto Flow Cytometer (no. 337175; BD Biosciences) equipped with a blue excitation source (air-cooled, 20-mW, solid-state 488-nm laser) to measure forward angle light scatter (FSC), side scatter (SSC) and cell fluorescence (FITC). Instrument settings were in logarithmic mode (FSC-H: 516, SSC-A: 626; and FITC-A: 962). GFP fluorescence intensity was calculated from 30,000 events (maximum threshold of 10,000 events per second) of triplicate samples, using FCS Express (De Novo Software). Strains were grown for 12 h after fresh inoculation, and 1 mL of culture was centrifuged for 2 min at 7,500 × g at room temperature. Pellets were resuspended in PBS (pH 7.4) containing 2% (wt/vol) paraformaldehyde, kept in the dark at 4 °C (≤5 d), and diluted 250-fold in 1× PBS immediately before FACS analysis.