In vivo RNA interactome profiling reveals 3'UTR-processed small RNA targeting a central regulatory hub

Abstract

Small noncoding RNAs (sRNAs) are crucial regulators of gene expression in bacteria. Acting in concert with major RNA chaperones such as Hfq or ProQ, sRNAs base-pair with multiple target mRNAs and form large RNA-RNA interaction networks. To systematically investigate the RNA-RNA interactome in living cells, we have developed a streamlined in vivo approach iRIL-seq (intracellular RIL-seq). This generic approach is highly robust, illustrating the dynamic sRNA interactomes in Salmonella enterica across multiple stages of growth. We have identified the OmpD porin mRNA as a central regulatory hub that is targeted by a dozen sRNAs, including FadZ cleaved from the conserved 3'UTR of fadBA mRNA. Both ompD and FadZ are activated by CRP, constituting a type I incoherent feed-forward loop in the fatty acid metabolism pathway. Altogether, we have established an approach to profile RNA-RNA interactomes in live cells, highlighting the complexity of RNA regulatory hubs and RNA networks.

Fig. 1. iRIL-seq faithfully captures sRNA-target interactions.

a Schematic of iRIL-seq. Salmonella Hfq::FLAG strain carrying the plasmid pBAD-t4rnl1 (pYC582) was grown in LB. sRNA-target pairs were ligated to form chimeras in vivo by T4 RNA ligase induced for 30 min with L-arabinose. The ligation chimeras bound to 3xFLAG tagged Hfq were enriched using coIP from bacterial lysates. Chimeras were then purified, identified by deep sequencing, and used to determine the RNA interaction network by subsequent in silico analysis. Created with BioRender.com. b iRIL-seq captured known sRNA-target interactions. Genome browser screenshots showing the genomic locations of indicated RNAs covered by singleton or significant chimera reads (S-chimera, p < 0.05, one-sided Fisher’s exact test). Salmonella WT and Hfq::FLAG strains carrying empty vector or pYC582 were grown in LB. Bacteria were treated with L-arabinose for 30 min and grown to OD₆₀₀ of 2.0. Cells were collected and performed iRIL-seq. EV: empty vector. T4: pBAD-t4rnl1 (pYC582). ORFs and RNAs were indicated by gray boxes. c Distribution of each transcript type for singleton and S-chimeric fragments within one set of four iRIL-seq libraries. The total number of sequenced fragments was denoted in parentheses. hkRNA: four housekeeping RNA (RnpB, SsrS, Ffs and SsrA). IGR: intergenic region. EV and T4 are the same as in (b). d Number of S-chimeric fragments detected in each sample. EV and T4 are the same as in (b). e Number of S-chimeric fragments for different transcript types. RNA1, the 5’ terminal RNA in the chimera. RNA2, the 3’ terminal RNA in the chimera. f Distribution of RNA1 and RNA2 in S-chimeras for different transcript types.

Fig. 2. Global profiling of RNA-RNA interactome by iRIL-seq across growth stages.

a Western blot confirmed the pulldown of Hfq at three growth stages. Salmonella WT and Hfq::FLAG strains carrying pYC582 were grown in LB. T4 RNA ligase was induced with L-arabinose for 30 min, before reaching the indicated growth stages (EP: Exponential Phase at OD 0.5, ESP: Early Stationary Phase at OD 2.0, SP: Stationary Phase at OD 2.0 + 3 h). Cells were collected and subjected to iRIL-seq. Input, total proteins from bacterial lysates. IP, proteins after immunoprecipitation using an anti-FLAG antibody. Blot shown is representative of n = 2 biological replicates. b Relative fraction of individual sRNAs as singleton and S-chimeric fragments in iRIL samples. Percentage represents the reads of a given sRNA compared to all reads from top 100 sRNAs in a library. c Venn diagram analysis of RNA–RNA interactions at three growth stages. d–f Circos plots represent RNA interaction networks at three growth stages. RNA-RNA interactions from three growth stages were represented on the Salmonella chromosome. Circumference: Interacting sRNAs in order of genomic context. Labeled sRNAs interact with at least five putative targets, and their interactions are shown in color. Other interactions are black. Source data are provided as a Source Data file.

Fig. 3. Characterization of sRNA-target interactions in iRIL-seq datasets.

a Comparison of the sRNA-mRNA interactions found by iRIL-seq and RIL-seq under ESP condition. The bars indicate the numbers of predicted targets for 10 sRNAs that have most targets predicted. Venn diagram (inset) shows the overlap of all predicted targets for these 10 sRNAs between the two datasets. b Sequence motif identified in RNA2 in S-chimeras. M indicates the total number of RNA2 sequences. N indicates the number of RNA2 sequences containing the motif. c–f Motifs identified in the targets are complementary to the cognate sRNAs. m, the total number of target sequences. n, the number of target sequences containing the motif. g Known processed sRNAs (Supplementary Data 6) were enriched in S-chimeras compared to primary sRNAs. Similar trend was also observed by RIL-seq (Supplementary Fig. 3). h Genome browser screenshots showing the genomic locations of sRNAs as singleton and S-chimeric fragments in the iRIL-seq dataset at ESP. Dashed lines indicate the 5’ end of sRNA fragments found in S-chimeras in Hfq-associated iRIL-seq data.

Fig. 4. iRIL-seq determines mRNA regulatory hubs.

a Circos plot showing the mRNA regulatory hubs that interact with at least four sRNA candidates. The hub mRNAs were depicted on the left hemisphere, with their interacting sRNAs on the right. sRNAs interacting with ompD are in bold. b Genome browser screenshots showing chimeric reads mapped to the 5’ region of ompD. The location of Shine-Dalgarno sequence was indicated as an orange box. c Verification of OmpD regulation by different sRNAs. The Salmonella WT strain containing an empty vector or sRNA overexpression plasmids was grown overnight in LB. Total proteins were analyzed by 12% SDS-PAGE. The gel was stained with Coomassie brilliant blue, or subjected to Western blotting using a polyclonal anti-OMP antiserum. ΔompD served as a OmpD-null control. The OmpD bands are indicated by asterisks. The levels of OmpD from three replicates are quantified. Blot shown is representative of n = 3 biological replicates. Graph bar represents mean relative protein abundance. Error bars indicate standard deviations from n = 3 biological replicates. Source data are provided as a Source Data file.

Fig. 5. FadZ is a 3’UTR-processed sRNA repressor of the OmpD porin.

a Schematic of the genomic context of the fadBA operon and FadZ in Salmonella. b FadZ was enriched by Hfq-coIP. Genome browser screenshot showing the genomic locations of FadZ fragments under ESP condition. c The sequence and predicted secondary structure of FadZ. d Alignment of FadZ genomic sequences from representative enterobacterial genera. Conserved nucleotides were marked in red. The processed 5’ end and Rho-independent terminator were indicated. e Expression of FadZ in Salmonella wild-type and FadZ-deletion strains. Total RNA was isolated at indicated time points and analyzed on northern blot. 5S rRNA was probed as a loading control. f Northern blot analysis of FadZ expression in a strain carrying a temperature-sensitive RNase E allele (rne-TS) and the wild-type allele (rne-WT). Bacteria were grown in LB-broth to OD 1.0 at 28 °C and then shifted to 28 °C or 44 °C for 30 min. Total RNA was extracted and separated on a 6% PAA gel. g Predicted binding sites between FadZ and OMP genes using RNAhybrid. Conserved nucleotides among OMP genes were marked in red. Amino acids corresponding to the binding sites were shown below. The mutated nucleotides were indicated. h Regulation of abundant major porins by FadZ. Salmonella strains containing control vector pJV300 or plasmids constitutively expressing the indicated RNA fragments were grown overnight in LB. Total proteins were analyzed by SDS-PAGE, and total RNAs were analyzed by northern blotting. i Relative fluorescence for translational OMP-sfGFP reporters. A ΔfadZ strain carrying either the control plasmid pXG-1 or pXG10-sfGFP with an in-frame fusion to the indicated genes, together with pJV300 or pPL-FadZ as indicated. j Regulation of ompD::GFP by FadZ. k Regulation of ompC::sfGFP by FadZ. Graph bars (i–k) represent mean relative fluorescence. Error bars indicate standard deviations from n = 3 biological replicates. Source data are provided as a Source Data file.

Fig. 6. FadZ is induced by fatty acid metabolism as part of feed-forward regulatory loop.

a Expression of FadZ in Salmonella during growth in M9CA minimal medium supplemented with indicated carbon sources (final concentration of 0.1%). Glu, D-glucose; Oct, octanoic acid (C8); Ole, oleic acid (C18:1). b Regulation of major porins by FadZ in the presence of fatty acids. Salmonella strains containing control vector pJV300 or pZE12-FadZ were grown in M9CA minimal medium supplemented with 0.1% oleic acid. Total proteins were analyzed by SDS-PAGE. The OmpD bands are indicated by asterisks. c Expression of FadZ in wild-type and Δcrp during oleic acid shock at indicated time points. d, e β-galactosidase activities of a chromosomally encoded lacZ-transcriptional fusions to indicated promoters. Cells were grown in M9CA medium containing 0.1% glucose to an OD₆₀₀ of 1.0, and then resuspended in M9CA medium containing 0.1% oleic acid for 1 h before assay. Error bars indicate standard deviations (n = 3). Graph bars (d, e) represent average activities. Error bars indicate standard deviations from n = 3 biological replicates. f FadZ-mediated type I incoherent feed-forward loop (I1-FFL). TF, transcription factor upstream. Arrow refers to activation and bar refers to repression. Growth curve of Salmonella strains in M9CA medium containing 0.1% glucose (g) or 0.1% oleic acid (h). i Model showing that the 3’UTR-derived FadZ sRNA regulates a central regulatory hub OmpD. Salmonella major porin OmpD is a key mRNA regulatory hub regulated by twelve different sRNAs, which are under different transcriptional control. Dashed ovals indicate three sRNAs that potentially interact with ompD but showing no regulation. FadZ is the cognate sRNA in the fatty acid metabolism pathway. Extracellular long-/medium-chain fatty acids are transported into bacteria by a specialized porin FadL, and activate the expression of the fadBAZ mRNA via two master transcriptional regulators, CRP and FadR. The 3’UTR of mRNA is cleaved by RNase E to produce the FadZ sRNA. With the help of Hfq, FadZ basepairs to the porin mRNAs to repress their expression and trafficking into the Salmonella envelope. Data points (g, h) represent average OD₆₀₀ value. Error bars indicate standard deviations form n = 3 biological replicates. Source data are provided as a Source Data file.