An RNA-seq based comparative approach reveals the transcriptome-wide interplay between 3'-to-5' exoRNases and RNase Y

Abstract

RNA degradation is an essential process that allows bacteria to control gene expression and adapt to various environmental conditions. It is usually initiated by endoribonucleases (endoRNases), which produce intermediate fragments that are subsequently degraded by exoribonucleases (exoRNases). However, global studies of the coordinated action of these enzymes are lacking. Here, we compare the targetome of endoRNase Y with the targetomes of 3'-to-5' exoRNases from Streptococcus pyogenes, namely, PNPase, YhaM, and RNase R. We observe that RNase Y preferentially cleaves after guanosine, generating substrate RNAs for the 3'-to-5' exoRNases. We demonstrate that RNase Y processing is followed by trimming of the newly generated 3' ends by PNPase and YhaM. Conversely, the RNA 5' ends produced by RNase Y are rarely further trimmed. Our strategy enables the identification of processing events that are otherwise undetectable. Importantly, this approach allows investigation of the intricate interplay between endo- and exoRNases on a genome-wide scale.

Fig. 1. RNase Y processes RNAs after a guanosine.

a Representation of RNA end (5′ or 3′) profiling obtained by RNA sequencing (performed in biological triplicates). The RNA ends that were more abundant in the wild type (WT) and complemented rny deletion strain (∆rny::rny) than in the RNase Y deletion strain (∆rny) are annotated as rny_end. “NNNNNNN” represents a sequence processed by RNase Y. b The bar plot shows the number of 5′ or 3′ ends that were more abundant in the WT than in the ∆rny strain (see Methods). c RNase Y cleavage (scissors) generates two processing products. We never retrieved both the RNA fragments upstream and downstream of the cleavage site for the same processing event. d Schematic drawing of total and end (5′ or 3′) coverages from RNA sequencing, illustrating RNA 5′ “unique” (U) and 3′ “stepped” (S) end positions. e Proportion of RNA 5′ and 3′ ends classified as U and S. f Sequence and structure conservation of the identified 5′ and 3′ rny_ends. The logo was created from the alignment of all sequences 10 nt on each side of the identified ends. Error bars are automatically calculated by the WebLogo library and correspond to an approximate Bayesian 95% confidence interval. The minimum free energy (ΔG) was calculated at each nucleotide position using a sliding window of 50 nt over the entire genome. The average ΔG (kcal mol⁻¹) calculated for a window of 100 nt centred on the identified ends is depicted.

Fig. 2. 3′-to-5′ exoRNases trim RNAs generated by RNase Y processing.

Left bar plot: portion of RNA 3′ ends (3′ rny_ends) corresponding to 3′-to-5′ exoRNase start positions (bottom portion), 3′-to-5′ exoRNase stop positions (middle portion) and not associated with 3′-to-5′ exoRNase (top portion). Right bar plots: number of trimming starts (top) and stops (bottom) that correspond to 3′ rny_ends, which were uniquely produced by PNPase and YhaM or produced by two different 3′-to-5′ exoRNases.

Fig. 3. RNase Y-generated RNAs are degraded by PNPase until secondary structures are encountered.

a Upper panel: example of 3′ end coverage profiling from RNA sequencing. Middle panel: the RNA ends that were more abundant in the WT than in the ∆rny strain are indicated below the coverage and depicted with purple arrowheads (3′ rny_ends). The RNA ends corresponding to the trimming start and stop positions of exoRNases are depicted with green and red arrowheads, respectively (see Supplementary Fig. 1). The 3′ rny_ends were paired to 3′-to-5′ exoRNase stop positions (Supplementary Data 2 and Fig. 2, bottom). Bottom panel: the 3′-to-5′ exoRNases (‘pacman’ symbols) started trimming upon RNase Y (scissors) processing and stopped before the RNA termini. The 3′ rny_ends corresponding to the 3′-to-5′ exoRNase stop positions were compared with the exoRNase trimming start positions located downstream. The 3′-to-5′ exoRNase start position corresponds to the initial RNase Y processing position (Supplementary Data 3). b The logo, displaying the information (bits), was created from the alignment of all sequences surrounding the 19 identified PNPase trimming start positions. c Total coverage of SPy_0316 (encoding a putative transcriptional regulator) in WT obtained by RNA sequencing, and schematic representation of the locus. The grey rectangle indicates the region where the processing sites of RNase Y, PNPase and YhaM were identified. d 3′ end coverage of a portion of SPy_0316 in the WT, ∆rny, YhaM deletion mutant (∆yhaM) and PNPase deletion mutant (∆pnpA) strains. The coverage scales are indicated between brackets. RNase Y processed the RNA after a G, corresponding to the detected PNPase trimming start position. PNPase trimmed 34 nt of the SPy_0316 RNA 3′ end. This new RNA 3′ end was subsequently nibbled by YhaM. e RNA folding of the region 100 nt upstream of the 3′ rny_ends corresponding to YhaM trimming stop positions. YhaM started trimming after PNPase stopped, at the base of the stem loop structure, and consequently removed 2 nt from the RNA 3′ end. f Structure conservation at the 183 PNPase stop positions previously identified. The decrease in the minimum free energy (∆G, kcal mol⁻¹) is indicative of RNA structures likely preventing PNPase from degrading the RNA.

Fig. 4. Initial RNase Y cleavage position validated by PNPase trimming start positions.

a Upper panel: example of 3′ end coverage profiling from RNA sequencing. Middle panel: the RNA ends that were more abundant in the WT than in the ∆rny strain are indicated below the coverage and depicted with purple arrowheads (3′ rny_ends). The RNA ends corresponding to the trimming start positions of exoRNases are depicted with green arrowheads (see also Supplementary Fig. 1). Bottom panel: The identified 3′ rny_ends were paired to 3′-to-5′ exoRNase start positions (Supplementary Data 4 and Fig. 2, top). The 3′-to-5′ exoRNase targeted the RNA fragments generated by RNase Y processing, but the exoRNase start position was nonetheless detectable in WT, indicating that a portion of the RNAs was not degraded. The 3′-to-5′ exoRNase start position corresponds to the initial RNase Y processing position (Supplementary Data 4). b Examples of RNAs identified by matching the 3′ rny_ends (purple arrowheads) with PNPase trimming start positions (green arrowheads) (see also Supplementary Fig. 3). For each RNA, the RNA 3′ end profile obtained from RNA sequencing in the WT, ∆rny and ∆pnpA strains is shown and the scales are indicated between brackets. The RNA 3′ ends generated by RNase Y (blue scissors) and eventually targeted by PNPase (‘pacman’ symbol) corresponded to decay intermediates. For rofA mRNA, RNase Y was also responsible for the generation of the 5′ end of the decay intermediate. For Spy_sRNA482963, another endoRNase (grey scissors) produced the 5′ end of the decay intermediate, which was previously identified as an RNA 5′ end that was more abundant in the ∆pnpA strain than in the WT strain.

Fig. 5. PNPase completely degrades the RNase Y-generated RNAs located upstream of the processing site.

a Example of 3′ and 5′ end coverage profiling by RNA sequencing. The RNA ends that were more abundant in the WT strain than in the ∆rny strain (purple arrowhead) and the RNA ends corresponding to trimming start positions of exoRNases (green arrowhead) are indicated below the coverage. The 5′ rny_ends were paired to 3′-to-5′ exoRNase start positions that were located at least 10 nt upstream (Supplementary Data 5). Subsequently to RNase Y activity, the 3′-to-5′ exoRNases completely degraded the RNA fragments upstream of the RNase Y processing. b Portion of RNA 5′ ends (5′ rny_ends) paired with 3′-to-5′ exoRNase start positions. All the trimming starts associated with 5′ rny_ends were PNPase trimming start positions.

Fig. 6. Characterization of RNase Y-generated RNA fragments in the WT strain.

a Representation of total, 5′ and 3′ end coverage profiles in the WT and ∆rny strains obtained by RNA sequencing and corresponding to RNA fragments produced by RNase Y. The 5′ and 3′ rny_ends are indicated with purple arrowheads. b The positions of the 5′ and 3′ rny_ends were compared by setting minimum and maximum distances of 40 and 1000 nt, respectively, between the ends. Each dot represents paired 5′ and 3′ rny_ends (Supplementary Data 6). c Northern blot analyses of RNA fragments in the open reading frames (ORFs) of murC, SPy_1551 and SPy_0316, generated by RNase Y and detectable only in the WT strain. The full blots, the loading controls and the RNA sequencing profile for each fragment are shown in Supplementary Fig. 5c–e, and the source data are provided as a Source Data file. Shown are the results of one representative northern blot analysis (n = 3). d Schematic representation of the generation of the short RNA fragments. RNase Y (blue scissors) is responsible for the production of both 5′ and 3′ fragment ends. The intermediate RNA fragment 3′ ends are, in 60% of the cases (Supplementary Data 6), subsequently trimmed by PNPase and/or YhaM (‘pacman’ symbol) from the start position (green arrowhead) until the stop position (red arrowhead).

Fig. 7. RNase Y produces decay intermediates degraded by PNPase.

a Schematic representation of decay intermediates generated by RNase Y (blue scissors) and rapidly degraded by PNPase (‘pacman’ symbol). RNase Y produced the decay intermediate 5′ end, which was identified as an RNA 5′ end that was more abundant in the ∆pnpA strain than in the WT strain. The decay intermediate 3′ end, corresponding to the PNPase trimming start position, was probably produced by an unidentified endoRNase (grey scissors). b Sequence conservation around the 5′ ends (5′ pnpA_ends) and 3′ ends (PNPase starts) of the decay intermediate from the 185 decay fragments previously identified, present in the ∆pnpA strain and not in the WT strain. c Northern blot analyses of decay intermediates (in pyrH and SPy_2197) in the WT, ∆rny, ∆pnpA and ∆pnpA∆rny strains. The RNase Y-generated fragments degraded by PNPase (purple arrows) are indicated. Shown are the results of one representative northern blot analysis (n = 3). The full blots, loading controls, RNA sequencing-based RNA end profile with the detected 5′ and/or 3′ rny_ends, and trimming start positions are shown in Supplementary Fig. 6 and the source data are provided as a Source Data file.

Fig. 8. RNase Y initiates the exoRNase-mediated degradation of putative regulatory 5′ UTRs.

a Left, models of decay of putative regulatory elements. RNase Y (blue scissors) processes regulatory 5′ UTR elements, producing decay intermediates that are subsequently degraded by PNPase (‘pacman’ symbol) (see also Supplementary Fig. 7). b Northern blot analyses of putative T-boxes (serS and thrS) in the WT, ∆rny, ∆pnpA and ∆pnpA∆rny strains. The full-length (FL) and RNase Y-generated decay intermediates (purple arrowheads) are indicated. Shown are the results of one representative northern blot analysis (n = 3). The full blots, the loading controls, the RNA end profile with the detected 5′ and/or 3′ rny_ends obtained by RNA sequencing, and the trimming start and stop positions are shown in Supplementary Fig. 7 and the source data are provided as a Source Data file.

Fig. 9. The concerted action of RNase Y and PNPase is responsible for the differential RNA stability of the rsmC-cdd-bmpA operon.

a Schematic representation of the rsmC-cdd-bmpA operon; the location of the promoter, terminator and probes used in the northern blot analyses and predicted RNA sizes are shown. b 5′ and 3′ end RNA sequencing coverages in the WT and ∆rny strains (for the 5′ end) and in the WT and ∆pnpA strains (for the 3′ end) of a region comprising portions of the cdd and bmpA ORFs and the intergenic region between the two genes. The coverage scale is indicated between brackets. The 5′ rny_end and the PNPase trimming start position identified in the cdd-bmpA intergenic region are depicted with purple and green arrowheads, respectively. c RNA 5′ end in the cdd-bmpA intergenic region in the WT and ∆rny strains, generated by RNase Y (scissors) and identified by primer extension analysis. The primer used is depicted with an arrow and binds upstream of the RNase Y processing and PNPase start positions (‘pacman’ symbol). The size of the expected cDNA product is indicated. Shown are the results of one representative primer extension experiment (n = 3). d, e The stability of rsmC-bmpA_cdd, cdd and bmpA RNAs was determined by northern blot analyses up to 45 or 8 min after the addition of rifampicin in the WT and ∆rny strains or in the WT and ∆pnpA strains. Shown are the results of one representative northern blot analysis (n = 3). The 16 rRNA was used as a loading control. Source data are provided as a Source Data file.

Fig. 10. Fate of the RNAs cleaved by RNase Y in S. pyogenes.

RNase Y processing occurs preferentially after a G. The two processing products generated from the same molecule were never detected together. Subsequently to RNase Y activity, three different events are conceivable. a Both generated RNA fragments are degraded by exoRNases and/or endoRNases; hence, this activity would be undetectable in our experimental setting. b The RNA fragments upstream of the RNase Y processing position were detected (i.e., 3′ ends), but not the downstream products (i.e., 5′ ends). Among the 130 RNAs detected, we demonstrated that 52% of the RNA products were trimmed by 3′-to-5′ exoRNases (mainly PNPase and/or YhaM). The remaining 48% were targeted either by several 3′-to-5′ exoRNases or by unidentified RNases. c The RNAs downstream of the RNase Y processing position were detected (i.e., 5′ ends), but not the upstream products (i.e., 3′ ends). Since 87.4% of the detected 5′ ends were mapped after G, we deduced that these RNAs were not further trimmed. The remaining 12.6% of the detected 5′ ends were not located after a G; therefore, we hypothesized that these ends were likely targeted by the 5′-to-3′ exoRNase J1.