Escherichia coli Hfq, RNase E catalytic domain (residues 1–529) and RNA degradosome were prepared using previously described procedures (9,30,31). The degradosome preparation includes RNase E, RhlB, enolase and polynucleotide phosphorylase (PNPase). Wild-type 30S subunits were prepared from E. coli MRE600 using zonal centrifugation (32,33). Initiator tRNA (fMet-tRNA^fMet) was purified using HPLC and E. coli initiation factors IF1, IF2 and IF3 were purified following published protocols (34).

RNAs were synthesized by in vitro transcription (IVT) as described (27). In brief, IVT templates were amplified by PCR from plasmid pVP042 that carries the Salmonella Typhimurium ompD gene (28). The forward primer introduces a T7 promoter sequence upstream of the transcription initiation site. IVT reactions containing 3 μg of template DNA, 5 mM of each of ATP, UTP, GTP and CTP, 10 mM DTT (dithiothreitol) and 0.5 U/μl RNaseOUT™ (Invitrogen) were incubated with purified recombinant T7 RNA polymerase in 40 mM Tris, pH 8.0, 25 mM MgCl₂, 2 mM spermidine at 37°C for 5 h. IVT products were DNase I-treated, resolved by 4% denaturing PAGE and excised RNA bands were electroeluted using an Elutrap™ Electroelution System Kit (Whatman), followed by separation using a PureLink™ RNA Micro Kit (Invitrogen).

Isothermal titration calorimetry (ITC), used to quantify RNA annealing, was performed using a MicroCal ITC200 microcalorimeter (Malvern Panalytical). ompD and MicC 12mers (Dharmacon) were buffer exchanged in 25 mM Tris pH 7.5, 50 mM NaCl, 50 mM KCl, 10 mM MgCl₂, 1 mM DTT using Micro Biospin 6 Chromatography Columns (Bio-Rad). ITC measurements were performed at 30°C with stirring at 1000 rpm. Following an initial injection of 0.4 μl, injections of 2 μl MicC 12mer variants into 12mer ompD were separated by 100 s to allow the system to equilibrate between injections. Data were analyzed using the ORIGIN software (OriginLab).

30S PIC was prepared following a previously described strategy (35). In brief, 30S subunits were incubated in buffer TAKM₂₀ (50 mM Tris–HCl pH 7.5, 70 mM NH₄Cl, 30 mM KCl, 20 mM MgCl₂) for 30 min at 37°C for reactivation. Reactivated 30S subunits were incubated with a 2.5-fold excess of mRNA, 2-fold excess of initiation factors IF1, IF2 and IF3, and a 2.5-fold excess of initiator fMet-tRNA^fMet (hereafter tRNA) in the presence of 250 μM GTP or GTPγS (Jena Bioscience) in TAKM₇ buffer (35). The complete complex was purified from the free components by size exclusion chromatography using a BioSEC5-1000 A matrix (Agilent), and the assembly was corroborated by denaturing protein and RNA electrophoresis gels. A fraction containing all components was used for cleavage assays, directly following purification.

Degradation assays were carried out in degradation buffer (25 mM Tris–HCl, pH 7.6, 50 mM NaCl, 50 mM KCl, 10 mM MgCl₂, 2 mM DTT, 0.5 U/μl RNaseOUT™) as described previously (27). Substrates (free ompD, ompD-PIC) were prepared as described above and pre-incubated with indicated MicC sRNA or 12-mer variants and Hfq at 37°C, at final concentrations of 0.2 μM if not otherwise indicated. Reactions were started by addition of the enzyme at indicated final concentrations. Reaction aliquots were withdrawn at indicated time points and quenched with equal volume stop buffer at 50°C for 30–45 min. Samples were resolved on 8–10% denaturing polyacrylamide gels, stained in SYBR Gold (Invitrogen) and visualized under UV light. Bands were quantified with GeneTools or ImageJ and plotted with GraphPad PRISM. Degradation rates of full-length sRNAs and mRNA segments were determined by fitting the data to the one phase decay fit equation in GraphPad PRISM, [I = (I₀ – P) × e^-kt) + P] with relative abundance, I; plateau, P; degradation rate, k; reaction time, t. Turnover rates of intermediates were determined by fitting the relative abundance to two-exponential equation [I = e^-kt – e^-jt] with relative abundance, I; degradation rate, k; formation rate, j; reaction time, t. Cleavage site specificities were calculated by dividing initial intermediate formation rates by the respective RNA degradation rate. Preference multiples were determined by dividing specificities for sites +83 and +72 of individual reactions. Reactions with free ompD were carried out in at least triplicates, reactions with ompD-PIC were done with triplicates for the degradosome (all MicC variants) and RNase E (1–529) (MicC WT, W1 and S1).

Plasmids expressing MicC variants were constructed by cleavage of the plasmid pKP21-13 encoding wild type MicC with BsaXI (Weaker 1, 2 and 3) or BamHI/HpaI (Stronger 1, 2 and 3), and replacement of the removed fragment with hybridized oligonucleotides introducing the desired mutation: W1for/W1rev (pBAD-MicC-W1), W2for/W2rev (pBAD-MicC-W2), W3for/W3rev (pBAD-MicCW3), S1for/S1rev/MicCStrongerFor/MicCStrongerRev (pBAD-MicC-S1), S2for/S2rev/MicCStrongerFor/MicCStrongerRev (pBAD-MicC-S2) and S3for/S3rev/MicCStrongerFor/MicCStrongerRev (pBAD-MicC-S3) (Table 1).

All strains unsed in this study are listed in Table 2. Phage P22 transduction was employed to transfer each single chromosomal modification to a fresh Salmonella wild-type background and to obtain strains carrying multiple mutations. To eliminate the Kan^R cassette of λRed-derived mutants, cells were transformed with the FLP recombinase expression plasmid pCP20 (36). Mutant susceptibility to kanamycin and loss of the temperature-sensitive FLP expression plasmid were tested.

Salmonella strains ΔmicC, ΔmicC rne701 and ΔmicC Δrnc carrying either a control plasmid (pBAD-ctrl.) or MicC expression vectors (pBAD-MicC, pBAD-MicC-S1, pBAD-MicC-S2, pBAD-MicC-S3, pBAD-MicC-W1, pBAD-MicC-W2, pBAD-MicC-W3) were grown aerobically at 37°C in LB broth supplemented with ampicillin (100 μg/ml) to mid-exponential phase (OD₆₀₀ of 0.6) when expression from the pBAD promoter was induced by addition of arabinose (0.1% final concentration). Total bacteria samples were collected, mixed with 0.2 volumes of stop-mix (95% ethanol and 5% phenol, v/v) and snap-frozen in liquid nitrogen. Total RNA was isolated using the Hot Phenol method as described previously (37). For Northern blot analysis, 5 μg of total RNA were separated on 5% polyacrylamide (7 M urea) gels in 1xTBE buffer and electroblotted. Membranes were hybridized with gene-specific P³² 5′ end-labelled DNA-oligonucleotides (MicC: JVO-0718; ompD mRNA: JVO-4314; 5S rRNA: JVO-0322) at 42°C in Roti-Hybri-Quick hybridization solution (Roth) and washed in three subsequent steps with SSC wash buffers (5×/1×/0.5 × SSC) supplemented with 0.1% SDS and imaged with a Typhoon FLA 7000 (GE Healthcare) phosphor-imager.

Earlier studies of RNase E activity on the ompD transcript in vivo identified position +83 in the coding region as a preferred cleavage site when the sRNA MicC is expressed (27,28). In vitro, RNase E cleaves ompD at additional sites downstream of the +83 site, probably due to changed accessibility of this fragment in the in vitro system; however, the +83 cleavage is present only in the presence of MicC and is substantially boosted when the sRNA carries a 5′ monophosphate (27). Moreover, in the absence of MicC a cleavage site at position +72 is observed, which is within the MicC-ompD pairing region, and is protected from RNase E activity when the sRNA-mRNA pairing occurs. For the following experiments, we are concentrating on the efficiency of the +83 cleavage as the reflection of the in vivo degradation pattern, and on the absence of cleavage at position +72, which is a marker of efficient sRNA binding.

The cleavage site in ompD is downstream of the base-pairing region between the mRNA and the sRNA, which suggests that the RNA-RNA interaction helps to align RNase E to attack the substrate (Figure 1A). Consistently, when an ompD mRNA fragment that encompasses the 69 nts long 5′ UTR and the first 118 nts of the coding region (ompD₁₁₈) is incubated with the highly purified RNase E catalytic domain (RNase E 1–529) in vitro, 5′P-MicC induces cleavage at the +83 site (9,27) (Figure 1B; Supplementary Figure S1). The level of intact MicC decreases over the course of the reaction, giving rise to deactivated MicC species that are cleaved at positions +9 and/or +24 and therefore lack the seed region (27) (Figure 1B). The full RNA degradosome assembly also produces this expected cleavage pattern with the +83-cleavage intermediate in the presence of 5′P-MicC (Figure 1C; Supplementary Figure S1). The +72-cleavage intermediate corresponds to degradosome cleavage within the region of complementarity to the MicC seed sequence and, in the absence of MicC, is the dominant intermediate species (Figure 1C, lanes 2–4). The in vitro reaction conditions do not support cleavage by the exoribonuclease PNPase as they do not contain inorganic phosphate (30,38); therefore, all degradation events are expected to be the result of RNase E activity.

To explore if the ompD transcript remains accessible to RNase E activity in the early stages of translation, we examined how the 30S PIC might impact cleavage of the mRNA. The ompD₁₁₈ fragment was assembled with purified 30S ribosomal subunit, initiator fMet-tRNA^fMet, initiation factors IF1, IF2, IF3 and GTP to form the ompD-PIC (Figure 1D) which was subsequently separated from the free components by size exclusion chromatography using a BioSEC-5 1000 Å matrix and used for cleavage assays directly following purification.

When associated with PIC components, the ompD mRNA was inefficiently processed by the isolated catalytic domain of RNase E, which is unable to access the +83-cleavage site in the presence of MicC (Figure 1E; Supplementary Figure S1). In contrast, the full RNA degradosome assembly can process ompD when the mRNA is engaged with the 30S PIC (Figure 1F; Supplementary Figure S1), producing +83-cleavage intermediates in the presence of MicC. This result indicates that the interaction of the C-terminal domain of RNase E and/or the other degradosome components with the 30S subunit modulates the nuclease's access to the sRNA-induced cleavage site in the transcript and therefore could play a role in gene expression regulation. Notably, in the presence of the 30S PIC, intact MicC levels decrease in parallel with that of the ompD transcript, whereas in the absence of the PIC, the sRNA is degraded more slowly than ompD. This suggests that the PIC might expose cleavage sites on both the sRNA and mRNA and may facilitate synchronized degradation of the two RNAs by the RNA degradosome. As stable complexes can form between the 30S subunit and the RNA recognition core of the degradosome, comprising the non-catalytic segment 603–850 from RNase E, RhlB helicase and enolase (31,39,40), PIC assembled on ompD might be also interacting with the C-terminal domain of RNase E and/or other degradosome components directly. This interaction could play a role in regulating access to cleavage sites on an mRNA that is engaged in the 30S PIC.

It is anticipated that an sRNA can be an effective trans-activator of RNase E if the molecule can provide an accessible monophosphate group that can bind to the 5′-end sensing pocket of the enzyme and boost the catalytic activity. Many trans-encoded sRNAs bear a seed region at the very 5′-end (41–43) and can therefore engage in duplex formation with a target such that it might be available to activate RNase E allosterically. Accordingly, if the duplex formed by sRNA and mRNA is recognized by RNase E, it must melt at least partially for the 5′-end of the sRNA to be accommodated in the enzyme's 5′-end sensing pocket. To test this hypothesis, and to explore the impact on activity within the PIC, the seed region of MicC was altered to change its interaction free energy with ompD. Wild type MicC forms an imperfect 12 bp duplex with the target sequence in ompD (14). Six different MicC seeds were designed with different predicted energies for binding ompD: three ‘weaker’ and three ‘stronger’ compared to wild type MicC (Figure 2A). To confirm the ranking for duplex stability, the free energies for pairing between MicC variants and ompD were quantified experimentally using isothermal titration calorimetry (Supplementary Table S1).

The MicC series was first tested as 12mers encompassing the seed region for their capacity to guide RNase E catalytic domain (RNase E (1–529)) to cleave ompD. All 12mers harbored a 5′-monophosphate group to activate RNase E cleavage in trans, and while each can induce +83-cleavage of ompD by RNase E (1–529), the cleavage efficiency differs for the various MicC seeds (Figure 2B). Remarkably, when the duplex between sRNA and mRNA was perturbed by substitutions in the first or second position from the 5′-end of the seed region, the cleavage at position +83 becomes more efficient compared to the wild type MicC seed (Figure 2B, Weaker 1 and 2). When the mutation in the seed region was in the third position from the 5′-end, the sRNA-12mer behaved closer to wild type (Figure 2B, Weaker 3), which suggests that the first two nucleotides of the seed are the most important in influencing the speed and efficiency of RNase E action in vitro. This agrees with the crystal structure of RNase E in complex with a fragment of MicC sRNA (9), in which only the first two terminal nucleotides of the RNA are engaged with the enzyme: the first nucleotide is buried in the 5′ binding pocket, whereas the second one interacts with R141. To form these interactions, the 5′-terminal nucleotides must be free and readily peeled from the mRNA.

The influence of the MicC seed variants in mediating RNase E (1–529) cleavage was investigated further in time course reactions to quantify the effect of the different MicC seeds on ompD degradation (Figure 2C, D). The weaker MicC 12mer mutants caused between 1.3- and 2-fold faster ompD degradation. Formation rates of +83-cleavage intermediates showed marked increase in the presence of Weaker 1 (13-fold) and Weaker 2 (2.5-fold), supporting the above-described observations. Comparing formation rates for +83 intermediates in proportion to respective ompD degradation rates shows that especially the Weaker 1 seed drastically increased RNase E preference for the +83 site (Figure 2E). The efficiency of the +83-cleavage in the presence of the stronger seeds is comparable to that of the wild type MicC 12mer; however, the overall ompD degradation is slower when the stronger substitution is introduced in the 8th position from the beginning of the seed region or when combined substitutions are made at both the 8th and 12th positions (Figure 2C, D, Stronger 1 and 3). These substitutions seem to trap the enzyme in a non-active conformation, which is consistent with the hypothesis that the sRNA-mRNA duplex must melt to facilitate RNase E action. Surprisingly, the fastest degradation rate for ompD was observed in the presence of the Stronger 2 seed, where a cleavage product larger than the +83-intermediate accumulated (Figure 2C). The differences in ompD degradation rates suggests that fold recognition might be another aspect of RNase E action on the sRNA-mRNA duplex, in addition to duplex strength.

Consistent with the observations made for full-length MicC, the +72 site of ompD that is accessible to RNase E attack in the absence of the sRNA also becomes protected in the presence of the various MicC 12mer variants. The abundance of the cleavage product from the +72 site increases as the sRNA-mRNA interaction becomes weaker, while the stronger MicC seed mutants confer greatest protection (Figure 2C).

The observation that seeds with readily peeled 5′ nucleotides boost the efficiency of ompD deactivation supports a model in which the two 5′-terminal nucleotides of the MicC seed must be unwound to activate RNase E in trans. To assess possible cooperative involvement of components within the RNA degradosome, we compared the activity of recombinant degradosome and RNase E (1–529) towards ompD in the presence of full-length monophosphorylated MicC with the different seed variants (Figure 3A; Supplementary Figure S1). The degradosome reactions are very similar to those with RNase E (1–529), with Weaker 1 and 2 boosting RNase E activity leading to accelerated +83 cleavage and increased MicC co-deactivation, whereas stronger seeds impede MicC co-degradation (Figure 3A; Supplementary Figure S1).

The series of weaker/stronger MicC constructs was also tested for the cleavage activity of RNase E catalytic domain (1–529) and the degradosome on the purified ompD₁₁₈-PIC complex (Figure 3B; Supplementary Figure S1). As described above, isolated RNase E catalytic domain, in contrast to the full degradosome, was unable to cleave the ompD-PIC at position +83 in the presence of wild type MicC (Figure 1E). While this is also the case in the presence of stronger MicC variants (Figure 3B, upper panel, S1, S2, S3), RNase E catalytic domain can cleave the ompD-PIC at position +83 when the duplex of sRNA and mRNA is weakened at the 5′-end of the sRNA seed (Figure 3B, upper panel, W1, W2, W3). In contrast, the RNA degradosome cleaves the ompD-PIC at the position +83, also in the presence of the stronger MicC seeds, which indicates that the degradosome assembly supports RNase E activity on PIC-engaged mRNA. Moreover, as the isolated catalytic domain can access the MicC-ompD duplex when not engaged on PIC, but is unable to cleave ompD in the presence of WT and stronger MicC variants within PIC, 30S-PIC assembly might restrict access of RNase E to the sRNA–mRNA duplex, but this restriction can be overcome in the presence of the other degradosome components.

In experiments with the full degradosome, the presence of the weaker MicC variants in reactions with ompD-PIC resulted mostly in +72-cleavage intermediates, whereas these same seeds provided more protection from +72-cleavage in the absence of the PIC (Figure 3A, B, compare lower panels). The +72 cleavage might be also a result of further degradation of the +83 intermediate, and therefore might be arising from faster progression of the reaction. With wild type or stronger pairing, separation of the MicC seed and complementary ompD fragment requires more energy than when weaker MicC variants are used. This separation might be facilitated by the PIC, but only when the whole degradosome is present. Moreover, all MicC variants were cleaved faster when ompD was engaged in the PIC (compare MicC bands between Figure 3A and B lower panels), and the trend is more pronounced for the weaker series, indicating that the sRNA pairing with the mRNA must be disrupted for RNase E cleavage to continue. These findings support a model in which the PIC assists the degradosome in remodeling the sRNA–mRNA duplex during or following +83-cleavage causing faster and/or complete sRNA-mRNA duplex melting, which results in the access to the +72-cleavage site on ompD and presenting MicC cleavage sites for sRNA-deactivation.

The impact of varying MicC seed strength on ompD decay was also investigated in vivo.

MicC wild type and weaker/stronger variants were individually expressed from a pBAD vector in strains of Salmonella Typhimurium lacking chromosomal micC (Figure 4A). Following induction, cell extracts were prepared in a time series and MicC and mRNA fragments of ompD were detected by northern blot analysis (Figure 4). Upon expression of the wild-type sRNA (Figure 4B), a bi-phasic response was observed, with an initial rapid decrease in ompD transcript levels to less than 10% of the starting levels within 4 min, followed by a slower decay (Figure 4B, D). All MicC variants were expressed within the first minutes of induction and capable of mediating ompD degradation in vivo. Unexpectedly, Weaker 1 and 2 were the least effective in triggering ompD degradation, while Stronger 2 and Stronger 3 were the most effective MicC variants. These observations stand in striking contrast to their behavior in vitro and suggest that additional processes and factors might be involved in defining the in vivo degradation rate that are not recapitulated with the purified in vitro conditions. For instance, the RNA-RNA interactions mediated by the weaker series might be disfavored by processes that prevent off-targeting due to mismatching, and that only a subset of sRNA-mRNA duplexes would induce degradation. For the stronger series, the sRNAs might trigger an alternative degradation pathway mediated for example by RNase III, which cleaves double-stranded RNA duplexes (44–46). In fact, MicC/ompD complexes with stronger base-pairing are better substrates for RNase III in in vitro assays (Supplementary Figure S2). In an RNase III-deficient strain of Salmonella, the Stronger 3 MicC variant has a similar impact on ompD as wild type MicC and accumulates to higher levels (Supplementary Figure S2).

To test if the degradosome organization contributes to the process of MicC-mediated degradation of ompD, the assays were repeated with cells expressing RNase E (1–701), which is a truncated version of RNase E that lacks the portion of the C-terminal domain encompassing the second RNA binding motif (AR2) and the interaction sites for enolase, helicase and PNPase, but retains the membrane-association signal and the first RNA binding microdomain. All MicC constructs, including the wild-type, exhibited a pronounced lag phase before the onset of ompD degradation (Figure 4D), suggesting that the C-terminal domain of RNase E supports enzyme activity in sRNA-mediated mRNA degradation, most likely through substrate capture. This is consistent with results of single molecule studies of the action of the sRNA SgrS on the ptsG transcript in vivo, where a truncation of RNase E that disrupts of the degradosome assembly impacts on the kinetics of substrate cleavage (21).

We noted that the +83-cleavage product accumulated transiently with the stronger seeds in vivo, while this intermediate was not detected for the wild type seed (Figure 4C). This suggests that the subsequent degradation steps in vivo are inefficient, which might arise because of product inhibition of RNase E, in which the enzyme is not able to separate the strong seed from its target to proceed with degradation. Given that the rate of decay of MicC on the PIC is dependent on the strength of the seed pairing (in vitro), the in vivo results may also reflect a process where sRNA lifetime is modulated upon meeting a target.