CDE-1 suppresses the production of risiRNA by coupling polyuridylation and degradation of rRNA

Background Modification of RNAs, particularly at the terminals, is critical for various essential cell processes; for example, uridylation is implicated in tumorigenesis, proliferation, stem cell maintenance, and immune defense against viruses and retrotransposons. Ribosomal RNAs can be regulated by antisense ribosomal siRNAs (risiRNAs), which downregulate pre-rRNAs through the nuclear RNAi pathway in Caenorhabditis elegans. However, the biogenesis and regulation of risiRNAs remain obscure. Previously, we showed that 26S rRNAs are uridylated at the 3′-ends by an unknown terminal polyuridylation polymerase before the rRNAs are degraded by a 3′ to 5′ exoribonuclease SUSI-1(ceDIS3L2). Results Here, we found that CDE-1, one of the three C.elegans polyuridylation polymerases (PUPs), is specifically involved in suppressing risiRNA production. CDE-1 localizes to perinuclear granules in the germline and uridylates Argonaute-associated 22G-RNAs, 26S, and 5.8S rRNAs at the 3′-ends. Immunoprecipitation followed by mass spectrometry (IP-MS) revealed that CDE-1 interacts with SUSI-1(ceDIS3L2). Consistent with these results, both CDE-1 and SUSI-1(ceDIS3L2) are required for the inheritance of RNAi. Conclusions This work identified a rRNA surveillance machinery of rRNAs that couples terminal polyuridylation and degradation.

In addition to guiding erroneous rRNA degradation, antisense ribosomal siRNAs (risiRNAs) silence pre-rRNAs through the nuclear RNAi pathway to suppress the accumulation of erroneous rRNAs in C. elegans [20,[24][25][26]. Erroneous rRNAs are usually oligouridylated at the 3′-ends and then degraded by the exoribonuclease SUSI-1(ceDis3L2). However, it is unclear which terminal uridyltransferase performs the untemplated addition of the 3′-end uracil. Identifying which PUP is involved in the 3′-uridylation of erroneous rRNAs and how it is involved will further our understanding of the quality control mechanism of cellular nucleic acids.
Here, we found that CDE-1 is specifically involved in suppressing risiRNA production. CDE-1 localizes to perinuclear granules in the germline and uridylates both Argonaute-associated 22G-RNAs, 26S, and 5.8S rRNAs at the 3′-ends. Interestingly, we found that CDE-1 interacts with SUSI-1(ceDIS3L2). Both CDE-1 and SUSI-1(ceDIS3L2) are required for the inheritance of RNAi. Therefore, our findings are consistent with the model that CDE-1 suppresses generation of risiRNAs by uridylating rRNA and recruiting SUSI-1(ceDIS3L2) to the rRNA.

Depletion of CDE-1 promotes risiRNA production
There are three RNA terminal uridylyltransferase genes, cde-1, pup-2, and pup-3, which are involved in RNA 3′-end uridylation in C. elegans. We previously showed that risiRNA was enriched in WAGO-4bound siRNAs in cde-1 mutants [8]. To further study the specificity and function of cde-1 in risiRNA production, we used the GFP::NRDE-3 transgene as a reporter. NRDE-3 is an Argonaute protein that transports siRNAs from the cytoplasm to the nucleus [27]. NRDE-3 localizes to the nucleus when it binds to siRNAs, but it accumulates in the cytoplasm when not bound to siRNA ligands. Disruption of the generation of endogenous siRNAs, for example, loss of function of eri-1 mutant results in relocalization of NRDE-3 from the nucleus to the cytoplasm. We crossed eri-1(mg366);GFP::NRDE-3 onto the pup mutant lines and found that the depletion of cde-1, but not pup-2 or pup-3, was able to redistribute NRDE-3 from the cytoplasm to the nucleus (Fig. 1a). We generated a single copy transgene CDE-1::mCherry by Mos1-mediated single-copy insertion (MosSCI) technology. This transgene was able to rescue the cde-1(tm936) defects and redistribute NRDE-3 from the nucleus to the cytoplasm (Additional file 1: Fig. S1 A). To exclude the possibility that PUP-2 and PUP-3 act redundantly to suppress siRNA generation, we generated pup-2;pup-3 double mutants. In the double mutants, NRDE-3 still accumulated in the cytoplasm (Additional file 1: Fig. S1 B). These data suggest that NRDE-3 was bound to newly generated siRNAs in cde-1 mutants.
To test whether the NRDE-3-bound siRNAs in cde-1 mutants contain risiRNA sequences, we used a risiRNA sensor expressing a his-72p::gfp::his-72 reporter fused to the 26S rRNA sequence (Fig. 1b). The sensor was expressed in wild-type N2 and eri-1(mg366) animals but silenced in cde-1(tm936) mutants. Furthermore, we quantified the amount of risiRNA by quantitative realtime PCR analysis and found that risiRNAs were increased in cde-1 mutants, but not in pup-2 or pup-3 mutants (Fig. 1c).
Base on the above results, we conclude that cde-1 likely acts as a suppressor of siRNA (susi) gene and suppresses the generation of risiRNAs.

CDE-1 uridylates risiRNA
We first compared the small RNA expression profiles between wild-type and cde-1 mutant animals. Small RNAs were isolated from young adult animals. Although the depletion of cde-1 did not noticeably change the expression profile of different small RNA categories, risiR-NAs were enriched 4.7 fold in cde-1 mutant animals vs wild-type animals (Fig. 2a). We then immunoprecipitated GFP::NRDE-3 and deep sequenced the associated siRNAs. NRDE-3-bound risiRNAs were enriched 17 fold in cde-1 mutant animals, compared to what was observed in control animals (Fig. 2b).
Small RNAs associate with NRDE-3 and guide NRDE-3 to the target nuclear nucleic acids. In the presence of risiRNA, NRDE-3 accumulated in the nucleoli of cde-1 mutants (Fig. 2d). FIB-1 in C. elegans is encoded by an ortholog of the genes encoding human fibrillarin and Saccharomyces cerevisiae Nop1p [28,29]. FIB-1 localizes to the nucleolus in embryos. This result suggested that NRDE-3 translocated to the nucleoli and risiRNAs may thereby silence rRNAs in the nucleoli due to double depletions of cde-1 and eri-1.
Therefore, we conclude that CDE-1 and SUSI-1(ceDIS3L2) likely function as a protein complex to suppress risiRNAs production.

CDE-1 is involved in uridylation of rRNA
Previously, we showed that SUSI-1(ceDIS3L2) degrades oligouridylated rRNAs and suppresses the production of risiRNA [20]. To examine whether CDE-1 uridylates rRNAs, a 3′ tail-seq method was used to detect the oligouridylation of rRNA at the 3′-tail in the indicated genotypes (Fig. 4a). Total RNA was isolated from L3staged control and indicated animals, and then libraries were prepared by PCR amplification with a corresponding rRNA primer and a primer targeting the linker, which were followed by high-throughput sequencing (Fig. 4a).
The 3′-end of 26S rRNA was extensively modified by all four nucleotides compared to the annotated rRNA sequence [20]. We tried to detect the 3′-ends of 26S and 5.8S rRNA. We found that both 26S rRNA (Fig. 4b) and 5.8S rRNA (Fig. 4c) are well processed at the 3′-tails as the 3′-end nucleic acid of most RNAs is the same. the top ten putative interacting proteins identified by CDE-1 immunoprecipitation followed by mass spectrometry. WD score: A scoring metrix for identifying high-confidence candidate interacting proteins (HCIPs) http://besra. hms.harvard.edu/ipmsmsdbs/cgi-bin/tutorial.cgi. c The protein-protein interaction of CDE-1 and SUSI-1(ceDIS3L2) was assayed by coimmunoprecipitation followed by western blotting with the indicated antibodies. d SUSI-1(ceDIS3L2) accumulated in the cytoplasm. Images of CDE-1 and SUSI-1 expression in the germline cells at young adult stage Although we did not detect a dramatic change in the nontemplated addition of a single nucleotide, we observed a modest depletion of oligouridylation at the 3′tail of 26S rRNA, comparing cde-1(tm936) to wild-type N2 animals at 20°C. The depletion of oligouridylation of 26 rRNA can be rescued by a introduction of the CDE-1::GFP transgene (Fig. 4d). Previously, we have noticed that risiRNA was enriched in cold-shocked worms [20]. Here, we discovered a significant increase of oligouridylation at the 3′-tail of 26S rRNA, comparing wild-type N2 animals at 20°C to cold-shocked N2 worms at 4°C for 12 h. Meanwhile, cde-1 mutation blocked the increase triggered by the cold shock treatment (Fig. 4d). We also noticed a significant increase of oligouridylation at the 3′-tail of 26S rRNA, comparing susi-1(ust1) to wild-type N2 animals at 20°C (Fig. 4d). Furthermore, to uncover how CDE-1 and SUSI-1(ceDIS3L2) affect oligouridylation at the 3′-end of rRNA simultaneously, we knocked down susi-1 in cde-1(tm936) mutants, as these two genes are too close on the genome to be genetically manipulated. We detected a significant decrease of susi-1 mRNA level (Additional file 4: Fig. S4). However, there is no change of the oligouridylation at the 3′-tail of 26S rRNA, comparing to the untreated cde-1(tm936) worms. c, e Tail-seq data of 5.8S rRNAs from indicated animals at L3 stage. The numbers of reads with untemplated oligouridylation at the 3′-ends are indicated. Data are presented as the mean ± SD (n = 3, biological replicates). ns, not significant. *P < 0.05, **P < 0.01 (two-tailed Student's t test) As for 5.8S rRNA, we detected a dramatic change in the nontemplated addition of a single U nucleotide, which was accompanied by a significant increase of oligouridylation at the 3′-tail, comparing wild-type N2 animals at 20°C to cold-shocked N2 worms at 4°C. In the same way, depletion of cde-1 blocked the increase of mono-or oligouridylation triggered by cold shock. Similarly, the change tendency of uridylation at the 3′-tail of 5.8S rRNA in the indicated genotype or under cold shock treatment is consistent with that of 26S rRNA (Fig. 4e).
Taken together, these data suggest that cold shock leads to uridylation at the 3′-tail of rRNA, and CDE-1 is involved in uridylating 26S and 5.8S rRNAs.

SUSI-1(ceDIS3L2) is required for the inheritance of RNAi
It was previously shown that CDE-1 is required for the inheritance of RNAi by uridylating WAGO-4-associated siRNAs [8,31]. Since CDE-1 interacts with SUSI-1(ceDIS3L2), we then asked whether susi-1 was also required for the inheritance of RNAi. We used a germline-expressed mex-5p::GFP::H2B (abbreviated as GFP::H2B) transgene as a reporter, which can inherit RNAi-induced gene silencing for multiple generations. Both hrde-1 and cde-1 were not required for exogenous gfp dsRNA to silence the GFP::H2B transgene in the parental generation, but they were essential for silencing in the F1 to F3 generations (Fig. 5). Similarly, susi-1(ceDis3L2) was not required for exogenous gfp dsRNA to silence the GFP::H2B transgene in the P0 generation but was necessary for silencing in F1 to F3 progeny. We conclude that susi-1(ceDis3L2) is required for the inheritance of RNAi.

Discussion
Misprocessed rRNAs are usually detected and degraded by surveillance machinery during ribosome biogenesis [13,14,23,32]. Previously, our lab identified a class of risiRNAs that downregulate pre-rRNAs through the nuclear RNAi pathway to suppress the accumulation of erroneous rRNAs. We identified a number of broadly conserved genes that are involved in rRNA processing and maturation. The depletion of these genes leads to an Fig. 5 Both CDE-1 and SUSI-1(ceDIS3L2) are required for the inheritance of RNAi. a mex-5p::GFP::H2B transgenic animals were exposed to bacteria expressing gfp dsRNA. F1 embryos were isolated and grown on control bacteria in the absence of further gfp dsRNA treatment. GFP expression in the indicated animals was imaged in the germline and oocytes. b The percentage of P0 to F3 animals expressing GFP was counted increase in risiRNAs. Thereafter, these genes are named suppressor of siRNA (susi) [20,24]. Among them, SUSI-1(ceDIS3L2) plays a vital role in the 3′-5′ degradation of oligouridylated rRNA fragments [20]. In this work, we further found that CDE-1 uridylates the 3′-end of 26S and 5.8S rRNAs and may recruit SUSI-1(ceDIS3L2) through protein-protein interactions. Therefore, we conclude that cde-1 is a new susi gene and suppresses the generation of risiRNAs.
Uridylation of the 3′-end of RNAs plays important functions in determining the fate of RNA [33,34]. For example, uridylation is an intrinsic step in the maturation of noncoding RNAs, including the U6 spliceosomal RNA or mitochondrial guide RNAs in trypanosomes [35]. Uridylation can also switch specific miRNA precursors from a degradative to a processing mode. This switch depends on the number of uracils added and is regulated by the cellular context [36,37]. However, the typical consequence of uridylation is accelerating the RNA degradation [21,38]. In this work, we showed that CDE-1 can uridylate 26S and 5.8S rRNAs and recruit the 3′-5′ exoribonuclease SUSI-1(ceDIS3L2), which may further promote the degradation of oligouridylated rRNAs. In the absence of either CDE-1 or SUSI-1(ceDIS3L2), erroneous rRNAs will accumulate in cells, which thereafter recruit the RNA-dependent RNA polymerases, including RRF-1 and RRF-2, to initiate risiRNA production (Fig. 6). risiRNAs then bind to both nuclear and cytoplasmic Argonaute proteins and silence rRNAs through both nuclear and cytoplasmic RNAi machinery. Therefore, risiRNA and the RNAi machinery, together with exoribonucleases, act to avoid the accumulation of potentially harmful or unnecessary erroneous rRNA transcripts [14,15,17,18,32,39,40]. 3′-end modifications play important roles in regulating the stability of siRNAs via distinct mechanisms as well. For example, methylation of the 3′-end inhibits uridylation and correlates with increased steady state levels of small RNAs [41,42]. In contrary, 3′ terminal uridylation may promote the degradation of siRNA [9,43]. CDE-1 uridylates endogenous siRNAs and modulates their binding affinity to CSR-1 and WAGO-4 [8,9]. PUP-2 has been reported to target let-7 miRNA [11,37]. Although PUP-3 has been validated as uridyl transferase, its targets are still unclear. Here, we found that CDE-1, but not PUP-2 or PUP-3, are engaged in suppressing risiRNA production. How these PUPs recognize their specific targets is still an enigma.
Additional questions remain as to how and why erroneous rRNAs could be recognized by CDE-1. Our Fig. 6 A working model of risiRNA biogenesis in C. elegans. The erroneous cellular rRNAs are scrutinized and suppressed through a number of mechanisms. Erroneous rRNAs are uridylated by CDE-1 and degraded by exoribonucleases such as SUSI-1(ceDis3L2). The disruption of CDE-1 or SUSI-1(ceDis3L2) results in the accumulation of erroneous rRNAs that thereafter recruit RdRPs to synthesize risiRNA. A risiRNA-mediated feedback loop silences rRNA expression through RNAi machinery and compensates for the disruption of the degradation of erroneous rRNA transcripts previous work found that either the modification errors or processing errors of rRNAs trigger the generation of risiRNAs [24]. How these different kinds of errors are sensed and scrutinized is still unknown. Deciphering the intricate interaction network of CDE-1 or other TUTases is key to fully understanding the effect of RNA uridylation. In addition, CDE-1 was previously reported required for RNAi inheritance [8,31]. The underlying mechanism remains unclear. Our previous data suggested that CDE-1 may mediate RNAi inheritance by regulating the binding of 22GRNAs to WAGO-4 [8].
Here, we found that CDE-1 interacted with SUSI-1(ceDIS3L2), another protein required for the inheritance of RNAi. Further elucidating the function of SUSI-1(ceDIS3L2) and CDE-1 will shed light on the mechanism of RNAi inheritance.

Conclusions
RNA uridylation is a potent and widespread posttranscriptional regulator of gene expression. The untemplated terminal polyuridylation has a decisive impact on RNA's fate. In this article, we further decipher the molecular mechanism of risiRNA generation in C. elegans. We provide genetic, cell biological, and biochemical evidence documenting the function of CDE-1, which uridylates rRNAs at 3′-ends. By using IP-MS technology, we found that CDE-1 interacts with SUSI-1(ceDIS3L2), a 3′-5′ exoribonuclease that we previously found to suppress risiRNA production. CDE-1 also uridylates Argonaute-associated risiRNAs. Interestingly, both CDE-1 and SUSI-1(ceDIS3L2) promote RNAi inheritance. In addition, this study sheds new light on a complicated surveillance network by combining uridylation, degradation, and RNAi-mediated gene silencing to maintain rRNA homeostasis.

Strains
Bistol strain N2 was used as the standard wild-type strain. All strains were grown at 20°C unless otherwise specified. The strains used in this study were listed in Additional file 5: Table S1.

Quantification of the subcellular location of NRDE-3
The subcellular localization of NRDE-3 was quantified as described previously [20]. Images were collected on a Leica DM4B microscope.

Quantitative RT-PCR
RNA was isolated from the indicated animals and subjected to DNase I digestion (Thermo Fisher). cDNA was generated from the isolated RNA using a GoScript™ Reverse Transcription System (Promega) according to the vendor's protocol. qPCR was performed using a MyIQ2 real-time PCR system (Bio-Rad) with an AceQ SYBR Green Master mix (Vazyme). The primers used in RT-qPCR were listed in Additional file 6: Table S2. eft-3 mRNA was used as an internal control for sample normalization. Data analysis was performed using a comparative threshold cycle (ΔΔCT) approach.

Brood size
Synchronized L3 worms were individually placed onto fresh NGM plates, and the progeny numbers were scored.

Deep sequencing of small RNAs and bioinformatic analysis
Total RNAs and the Argonaute-associated RNAs were isolated from the indicated animals and subjected to small RNA deep sequencing using an Illumina platform (Novogene Bioinformatics Technology Co., Ltd.), as previously described [20].
For Argonaute-associated RNAs, synchronized worms were sonicated in sonication buffer (20 mM Tris-HCl, pH 7.5, 200 mM NaCl, 2.5 mM MgCl 2 , and 0.5% NP40). The eluates were incubated with TRIzol reagent and then precipitated with isopropanol. The precipitated RNA was treated with FastAP Thermosensitive Alkaline Phosphatase (Thermo Scientific), re-extracted with TRIzol, and treated with T4 Polynucleotide Kinase (T4 PNK, Thermo Scientific) in the presence of 1 mM ATP before library construction.
The Illumina-generated raw reads were first filtered to remove adaptors, low-quality tags, and contaminants to obtain clean reads by Novogene. Clean reads ranging from 18 to 30 nt were mapped to the transcriptome assembly WS243 using Bowtie2 with default parameters. The number of reads targeting each transcript were counted by custom Perl scripts. The number of total reads mapped to the genome minus the number of total reads corresponded to sense rRNA transcripts (5S, 5.8S, 18S, and 26S), which was used as the normalization number, to exclude the possible degradation fragments of sense rRNAs.

Proteomic analysis
Proteomic analysis was conducted as previously described [45]. Briefly, mixed-stage transgenic worms expressing CDE-1::GFP were resuspended in equal volumes of 2× lysis buffer (50 mM Tris-HCl pH 8.0, 300 mM NaCl, 10% glycerol, 1% Triton X-100, Roche®cOmplete EDTA-free Protease Inhibitor Cocktail, 10 mM NaF, and 2 mM Na 3 VO 4 ), and lysed in a FastPrep-24 5G homogenizer. The lysate supernatant was incubated with anti-GFP antibody, which was linked to beads, for 1 h at 4°C. The beads were then washed three times with cold lysis buffer. The GFP immunoprecipitates were eluted with chilled elution buffer (100 mM glycine-HCl, pH 2.5). Approximately 1/8 of the eluates were subjected to western blotting analysis. The rest of the eluates were precipitated with TCA or cold acetone and dissolved in 100 mM Tris (pH 8.5), with 8 M urea. The proteins were reduced with TCEP, alkylated with IAA, and finally digested with trypsin at 37°C overnight. LC-MS/MS analysis of the resulting peptides and MS data processing approaches were conducted as previously described [46]. A WD scoring matrix was used to identify high-confidence candidate interacting proteins.
The RNAs were reverse transcribed with the following primer: Universal 3′ linker RT: 5′-GCTGGAATTCGCGGTTAAATCCTTGGTGCC CGAGTGT-3′. The cDNAs were PCR amplified (16 cycles) with the primers 26S rRNA-F: 5′-CAGATCAC TCTGGTTCAATGTC-3′ and universal 3′ linker RT primer, or with the primers 5.8S rRNA-F: 5′-GGTT GCATCGAGTATCGATGAA-3′ and universal 3′ linker RT primer, gel purified and then deep sequenced using an Illumina platform, according to the manufacturer′s instructions, by Novogene (Beijing, China). The number of reads with distinct 3′-end modifications was counted by custom Perl scripts. Three biological replicates were conducted in each genotype or treatment.

RNAi inheritance assay
Synchronized L1 animals of the indicated genotypes were exposed to bacteria expressing gfp dsRNA. F1 embryos were collected by hypochlorite/NaOH treatment and grown on OP50 bacteria. The GFP expression levels in both the parental generation (P0) and the progeny (F1-F3) were visualized and scored. Images were collected with a Leica DM4B microscope system.

Statistics
Bar graphs with error bars represent the mean and SD. All of the experiments were conducted with independent C. elegans animals for the indicated N replicates. Statistical analysis was performed with two-tailed Student's t tests or unpaired Wilcoxon tests. The threshold for Student's t test P values was set to 0.05 or 0.01.