Tumor elimination by clustered microRNAs miR-306 and miR-79 via noncanonical activation of JNK signaling

JNK signaling plays a critical role in both tumor promotion and tumor suppression. Here, we identified clustered microRNAs (miRNAs) miR-306 and miR-79 as novel tumor-suppressor miRNAs that specifically eliminate JNK-activated tumors in Drosophila. While showing only a slight effect on normal tissue growth, miR-306 and miR-79 strongly suppressed growth of multiple tumor models, including malignant tumors caused by Ras activation and cell polarity defects. Mechanistically, these miRNAs commonly target the mRNA of an E3 ubiquitin ligase ring finger protein 146 (RNF146). We found that RNF146 promotes degradation of tankyrase (Tnks), an ADP-ribose polymerase that promotes JNK activation in a noncanonical manner. Thus, downregulation of RNF146 by miR-306 and miR-79 leads to hyper-enhancement of JNK activation. Our data show that, while JNK activity is essential for tumor growth, elevation of miR-306 or miR-79 overactivate JNK signaling to the lethal level via noncanonical JNK pathway and thus eliminate tumors, providing a new miRNA-based strategy against cancer.


Introduction
Cancer progression is driven by oncogenic alterations of intracellular signaling that lead to promotion of cell proliferation and suppression of cell death (Croce, 2008). The c-Jun N-terminal kinase (JNK) pathway is an evolutionarily conserved mitogen-activated protein (MAP) kinase cascade that regulates both cell proliferation and cell death in normal development and cancer (Bode and Dong, 2007;Eferl and Wagner, 2003). Indeed, JNK signaling can act as both tumor promoter and tumor suppressor depending on the cellular contexts (Bode and Dong, 2007;Bubici and Papa, 2014;Karin and Gallagher, 2005). Crucially, JNK signaling is often activated in various types of cancers (Bubici and Papa, 2014;Wu et al., 2019). Thus, accumulating evidence suggests that JNK signaling can be a critical therapeutic target for cancer. For instance, converting JNK's role from pro-tumor to antitumor within tumor tissue could be an ideal anticancer strategy.
Drosophila provides a superb model for studying the genetic pathway of cellular signaling and has made great contributions to understand the basic principle of tumor growth and progression (Enomoto et al., 2018;Tipping and Perrimon, 2014). The best-studied model of Drosophila malignant tumor is generated by clones of cells overexpressing oncogenic Ras (Ras V12 ) with simultaneous mutations in apicobasal polarity genes such as lethal giant larvae (lgl), scribble (scrib), or discs large (dlg) in the imaginal epithelium (Brumby and Richardson, 2003;Pagliarini and Xu, 2003). These tumors activate JNK signaling and blocking JNK within the clones strongly suppresses their tumor growth Uhlirova and Bohmann, 2006), indicating that JNK acts as a pro-tumor signaling in these malignant tumors. Conversely, clones of cells overexpressing the oncogene Src in the imaginal discs activate JNK signaling and blocking JNK in these clones results in an enhanced overgrowth (Enomoto and Igaki, 2013), indicating that JNK negatively regulates Src-induced tumor growth. Similarly, although clones of cells mutant for scrib or dlg in the imaginal discs are eliminated by apoptosis when surrounded by wild-type cells, blocking JNK in these clones suppresses elimination and causes tumorous overgrowth (Brumby and Richardson, 2003;Igaki et al., 2009), indicating that JNK acts as antitumor signaling in these mutant clones. Thus, JNK also acts as both pro-and antitumor signaling depending on the cellular contexts in Drosophila imaginal epithelium.
miRNAs are a group of small noncoding RNAs that suppress target gene expression by mRNA degradation or translational repression and have been proposed to be potent targets for cancer therapy. Indeed, several cancer-targeted miRNA drugs have entered clinical trials in recent years. For instance, MRX34, a miRNA mimic drug developed from the tumor-suppressor miR-34a, is the first miRNA-based anticancer drug that has entered phase I clinical trials for patients with advanced solid tumors (Beg et al., 2017;Hong et al., 2020). In addition, MesomiR-1, a miR-16 mimic miRNA that targets EGFR, has entered phase I trial for the treatment of thoracic cancers (Reid et al., 2013;van Zandwijk et al., 2017). Such miRNA-mediated anticancer strategy can be studied using the Drosophila tumor models. Indeed, in Drosophila, the conserved miRNA let-7 targets a transcription factor chinmo and thus suppresses tumor growth caused by polyhomeotic mutations (Jiang et al., 2018). In addition, miR-8 acts as a tumor suppressor against Notch-induced Drosophila tumors by directly inhibiting the Notch ligand Serrate (Vallejo et al., 2011). However, apart from these miRNAs that suppress growth of specific types of tumors, it is unclear whether there exist miRNAs that generally suppress tumor growth caused by different genetic alterations.
Here, using Drosophila tumor models and subsequent genetic analyses, we identified several tumor-suppressor miRNAs. Among these, miR-306 and miR-79, two clustered miRNAs located on the miR-9c/306/79/9b cluster, significantly suppressed growth of multiple types of JNK-activated tumors while showing only a slight effect on normal tissue growth. Mechanistically, miR-306 and miR-79 directly target RNF146, an E3 ubiquitin ligase that causes degradation of a JNK-promoting ADPribose polymerase Tnks, thereby overamplifying JNK signaling in tumors to the lethal levels via noncanonical JNK activation. Our findings provide a novel miRNA-based strategy that generally suppress growth of JNK-activating tumors.

Identification of miR-306 and miR-79 as novel tumor-suppressor miRNAs
To identify novel antitumor miRNAs in Drosophila, we focused on 37 miRNA clusters or miRNAs that are highly expressed in Drosophila eye-antennal discs (Chung et al., 2008). Using the Flippase (FLP)-Flp recognition target (FRT)-mediated genetic mosaic technique, each miRNA was overexpressed in clones of cells expressing Ras V12 with simultaneous mutations in the apicobasal polarity gene dlg (Ras V12 /dlg -/-) in the eye-antennal discs, the best-studied malignant tumor model in Drosophila (Pagliarini and Xu, 2003; Figure 1C; compare to Figure 1). We found that overexpression of miR-7, miR-79, miR-252, miR-276a, miR-276b, miR-282, miR-306, miR-310, miR-317, miR-981, miR-988, or the miR-9c/306/79/9b cluster in Ras V12 /dlg -/clones dramatically suppressed tumor growth ( Figure 1C  Eye-antennal disc bearing GFP-labeled clones of indicated genotypes (A and B, 5 days after egg laying, C-G, 7 days after egg laying). (H) Quantification of clone size (% of total clone area per disc area in eye-antennal disc) in (A-G). Error bars, SD; ****p<0.0001 by one-way ANOVA multiple-comparison test. (I) Pupation rate of flies with indicated genotypes. Data from three independent experiment, n > 30 for each group in one experiment; error bars, SD. (J) Eclosion rate of flies with indicated genotypes. Data from three independent experiment, n > 30 for each group in one experiment; error bars, SD. (K, L) Adult eye phenotype of flies with indicated genotypes. (M-P) Eye-antennal disc bearing GFP-labeled clones of indicated genotypes (5 days after egg laying). (Q) Quantification of clone size (% of total clone area per disc area in eye-antennal disc) of (M-P). Error bars, SD; **p<0.01, ****,p<0.0001 by one-way ANOVA multiple-comparison test.
The online version of this article includes the following source data and figure supplement(s) for figure 1: Source data 1. Quantitative data for Figure 1.
These data indicate that miR-306 and miR-79 are tumor-suppressor miRNAs that only mildly suppress normal tissue growth but specifically block tumor growth in Drosophila imaginal epithelium.

miR-306 and miR-79 suppress tumor growth by promoting cell death
We next investigated the mechanism by which miR-306 and miR-79 suppress tumor growth. Immunostaining of Ras V12 /dlg -/or Ras V12 /lgl -/tumors with anti-cleaved DCP-1 antibody revealed that expression of the miR-9c/306/79/9b cluster, miR-306, or miR-79 in tumor clones significantly increased the number of dying cells ( Figure 2F and Figure 2-figure supplement 1E). In addition, blocking cell death in tumor clones by overexpressing the caspase inhibitor baculovirus p35 canceled the tumor-suppressive activity of miR-306 or miR-79, while p35 overexpression alone did not affect growth of normal tissues or Ras V12 /dlg -/tumors ( Figure 2G-N, quantified in Figure 2O). These data indicate that the miR-9c/306/79/9b cluster, miR-306, or miR-79 suppresses tumor growth by inducing cell death. Importantly, overexpression of these miRNAs alone did not cause cell death in normal tissue ( Figure 2P-S, quantified in Figure 2T), suggesting that miR-306 or miR-79 cooperates with a putative tumor-specific signaling activated in Ras V12 /dlg -/or Ras V12 /lgl -/tumors to induce synthetic lethality.

miR-306 and miR-79 suppress tumor growth by enhancing JNK signaling
We thus examined whether Ras activation or cell polarity defect cooperates with miR-306 or miR-79 to induce cell death. Overexpression of the miR-9c/306/79/9b cluster, miR-306, or miR-79 in Ras V12expresing clones did not affect their growth (Figure 3-figure supplement 1A-D, quantified in Figure 3-figure supplement 1E), indicating that Ras signaling does not cooperate with these miRNAs. Notably, overexpression of these miRNAs in dlg -/clones significantly reduced their clone size ( Figure 3A-E, quantified in Figure 3F). In addition, blocking cell death by overexpression of p35 canceled the ability of these miRNAs to reduce dlg -/clone size ( Figure 3G-O, quantified in Figure 3P), suggesting that these miRNAs block dlg -/clone growth by promoting cell death. These data show that miR-306 or miR-79 cooperates with loss of cell polarity to induce synthetic lethality.
We then sought to identify the polarity defect-induced intracellular signaling that cooperates with miR-306 or miR-79 to induce cell death. It has been shown that clones of cells mutant for cell polarity genes such as dlg activate JNK signaling via the Drosophila tumor necrosis factor (TNF) Eiger (Brumby and Richardson, 2003;Igaki et al., 2009). We found that overexpression of miR-306 or miR-79 alone moderately activated JNK signaling in the eye-antennal discs, as visualized by anti-p-JNK antibody      It has been shown that the severity of the reduced-eye phenotype depends on the levels of JNK activation and subsequent cell death (Igaki et al., 2002;Igaki et al., 2006;Igaki et al., 2009), suggesting that miR-306 and miR-79 enhance Eiger-mediated activation of JNK signaling. Indeed, blocking JNK signaling by overexpression of a dominant-negative form of Drosophila JNK Basket (Bsk DN ) canceled the tumor-suppressive activity of miR-306 or miR-79 against Ras V12 /dlg -/or Ras V12 /lgl -/tumors ( Figure 3R-W, quantified in Figure  miR-306 and miR-79 enhance JNK signaling stimulated by different upstream signaling We next examined whether miR-306 or miR-79 suppresses growth of other types of tumors with elevated JNK signaling via an Eiger-independent mechanism. Overexpression of an activated form of the Drosophila PDGF/VEGF receptor homolog (PVR act ) results in JNK activation and tumor formation in the wing disc (Wang et al., 2016a) and eye-antennal disc ( Figure 4B, compare to Figure 4A, quantified in Figure 4F). This tumor growth was significantly suppressed by overexpression of the miR-9c/306/79/9b cluster, miR-306, or miR-79 ( Figure 4B-E, quantified in Figure 4F). In addition, the size of clones overexpressing the oncogene Src64B in the eye-antennal disc ( Figure 4H, compare to Figure 4G, quantified in Figure 4F), which activate JNK signaling (Enomoto and Igaki, 2013), was significantly reduced when the miR-9c/306/79/9b cluster, miR-306, or miR-79 was coexpressed ( Figure 4H-K, quantified in Figure 4L). Moreover, nonautonomous overgrowth of surrounding wildtype tissue by Src64B-overexpressing clones (Enomoto and Igaki, 2013) was significantly suppressed by coexpression of these miRNAs ( Figure 4M-P, quantified in Figure 4Q). Furthermore, the size of clones mutant for an RNA helicase Hel25E or an adaptor protein Mahj, both of which are eliminated by JNK-dependent cell competition when surrounded by wild-type cells Tamori et al., 2010), was significantly reduced when these miRNAs were coexpressed ( . We further examined whether expression of these miRNAs enhances normally occurring JNK activity during development. The pnr-GAL4 driver strain specifically expresses GAL4 in the wing discs in a broad domain corresponding to the central presumptive notum during metamorphosis (Ishimaru et al., 2004;Zeitlinger and Bohmann, 1999). Knocking down Hep, the Drosophila JNK kinase, using the pnr-GAL4 driver generates a split-thorax phenotype caused by reduced JNK signaling (Ishimaru with anti-cleaved Dcp-1 antibody (P-S and P'-S', 5 days after egg laying). (T) Quantification of dying cells in GFP-positive clone area in (P-S). Error bars, SD; n.s., p>0.05 (not significant) by one-way ANOVA multiple-comparison test.
The online version of this article includes the following source data and figure supplement(s) for figure 2: Source data 1. Quantitative data for Figure 2.
Source data 2. Genotypes for Figure 2 and     , 2004). On the contrary, ectopic expression of Hep or Eiger using pnr-GAL4 generates a smallscutellum phenotype caused by elevated JNK signaling (Ma et al., 2013;Xue et al., 2007). Similarly to Hep or Eiger, ectopic expression of miR-306 or miR-79 using pnr-GAL4 resulted in a small-scutellum phenotype ( These data suggest that miR-306 and miR-79 broadly enhance JNK signaling activity stimulated by different upstream signaling.

miR-306 and miR-79 enhance JNK signaling activity by targeting RNF146
We next sought to identify the mechanism by which miR-306 and miR-79 enhance JNK signaling by searching for the target gene(s) of these miRNAs. The clustered miRNAs often target overlapping sets of genes and thus co-regulate various biological processes (Kim et al., 2009;Wang et al., 2016b;Yuan et al., 2009). Given that miR-306 and miR-79 are located on the same miRNA cluster, we searched for the common targets of these miRNAs using the online software TargeyScanFly 7.2 (http://www. targetscan.org/fly_72/) and found 11 mRNAs that were predicted to be targets of both miR-306 and miR-79 ( Figure 5A). We then examined whether knocking down of each one of these candidate genes could activate JNK signaling in Drosophila wing discs, where a clear JNK activation was observed when miR-306 or miR-79 was overexpressed ( Figure 5-figure supplement 1). As a result, we found that knocking down of RNF146, but not any other available RNAis for the candidate genes, resulted in JNK activation ( Figure 5B and C, Figure 5-figure supplement 2). The RNF146 mRNA had putative target sites of miR-306 and miR-79 in its 3′UTR region ( Figure 5D). To confirm that RNF146 mRNA is a direct target of miR-306 and miR-79, we performed a dual-luciferase reporter assay in Drosophila S2 cells using wild-type RNF146 3′UTR (RNF146 WT) or mutant RNF146 3′UTR bearing mutations at the putative binding site of miR-306 (RNF146 m1) or miR-79 (RNF146 m2) ( Figure 5D). We found that miR-306 and miR-79 reduced wild-type RNF146 3′UTR expression but did not affect respective mutant RNF146 3′UTR ( Figure 5E), indicating that miR-306 and miR-79 directly target RNF146 3′UTR ( Figure 5D). We also confirmed that overexpression of miR-306 or miR-79 reduced the endogenous levels of RNF146 protein ( Figure 5F, quantified in Figure 5G) and that suppression of miR-306 and miR-79 functions by using miRNA sponges increased the endogenous levels of RNF146 protein in the adult eyes ( Figure 5-figure supplement 3A, quantified in Figure 5-figure supplement 3B).
The online version of this article includes the following source data and figure supplement(s) for figure 3: Source data 1. Quantitative data for Figure 3.

RNF146 promotes Tnks degradation
We next investigated the mechanism by which downregulation of RNF146 by miR-306 or miR-79 enhances JNK signaling activity. It has been shown in Drosophila embryos, larvae, wing discs, and adult eyes that loss of RNF146 upregulates the protein levels of Tnks (Gultekin and Steller, 2019;Wang et al., 2019), a poly-ADP-ribose polymerase that directly mediates poly-ADP ribosylation of JNK, which triggers K63-linked poly-ubiquitination of JNK and thereby promotes JNK-dependent apoptosis in Drosophila (Feng et al., 2018;Li et al., 2018). In addition, loss of RNF146 was shown to enhance rough-eye phenotype caused by Tnks overexpression (Gultekin and Steller, 2019). These observations raise the possibility that downregulation of RNF146 by miR-306 or miR-79 enhances JNK signaling via upregulation of Tnks. Indeed, as reported previously (Feng et al., 2018;Li et al., 2018), Western blot analysis revealed that overexpression of Tnks induces phosphorylation of JNK (JNK activation) in S2 cells ( Figure 6A, lane 2 vs. lane 1, quantified in Figure 6C). Notably, coexpression of RNF146 significantly downregulated Tnks protein level and suppressed Tnks-induced JNK phosphorylation ( Figure 6A, lane 3 vs. lane 2, quantified in Figure 6B and C). Moreover, knocking down of RNF146 or overexpression of miR-306 or miR-79 significantly upregulated Tnks protein level and promoted JNK phosphorylation ( Figure 6D, quantified in Figure 6E and F, Figure 6-figure supplement 1A, quantified in Figure 6-figure supplement 1B and C). These data support the notion significant), **p<0.01 by two-tailed Student's t-test. (F) Lysates of adult heads of indicated genotypes were subjected to Western blots using indicated antibodies. (G) Quantification of relative levels of RNF146 protein in (F) from three independent experiments. Error bars, SD; *p<0.05, **p<0.01 by one-way ANOVA multiple-comparison test. (H-J) Eye-antennal disc bearing GFP-labeled clones of indicated genotypes (7 days after egg laying). (K) Quantification of clone size (% of total clone area per disc area in eye-antennal disc) of (H-J). Error bars, SD; ****p<0.0001 by two-tailed Student's t-test. (L-M) Eye-antennal disc bearing GFP-labeled clones of indicated genotypes (7 days after egg laying). (N) Quantification of clone size (% of total clone area per disc area in eye-antennal disc) of (L-M). Error bars, SD; n.s., p>0.05 (not significant) by two-tailed Student's t-test. (O-R) Eye-antennal disc bearing GFP-labeled clones of indicated genotypes (7 days after egg laying). (S) Quantification of clone size (% of total clone area per disc area in eyeantennal disc) of (O-R). Error bars, SD; *p<0.05, **p<0.01 by one-way ANOVA multiple-comparison test.
The online version of this article includes the following source data and figure supplement(s) for figure 5: Source data 1. Quantitative data or raw data for Figure 5.
Source data 2. Genotypes for Figure 5 and Figure 5-figure supplements 1-6.           Figure 6K). Due to the fact that overexpression of Tnks alone resulted in larger clone size than Ras V12 /dlg -/-+Tnks clone ( Figure 6H and J, quantified in Figure 6K), our data support the notion that Tnks suppresses growth of Ras V12 /dlg -/tumors by cooperating with JNK signaling. Consistent with the data shown above, overexpression of Tnks rescued the lethality of flies bearing Ras V12 /dlg -/tumors in the eye-antennal discs ( Figure 6L and  M).
Finally, we sought to clarify the mechanism by which downregulation of RNF146 upregulates Tnks. A pervious study has shown that Tnks protein levels were significantly higher in Rnf146 mutant background than in wild-type (Gultekin and Steller, 2019). However, this upregulation of Tnks can be caused by either elevated Tnks protein synthesis or reduced Tnks protein degradation. We thus examined the possibility that RNF146 promotes degradation of Tnks. Blocking new protein synthesis in S2 cells by the protein synthesis inhibitor cycloheximide (CHX) resulted in a time-dependent depletion of Tnks protein with a half-life of less than 3 hr ( Figure 6N, quantified in Figure 6P). This depletion of Tnks was significantly retarded when RNF146 was knocked down ( Figure 6O, quantified in Figure 6P). These data indicate that endogenous RNF146 promotes degradation of Tnks protein. Taken together, our data show that miR-306 or miR-79 directly targets RNF146, thereby leading to elevation of Tnks protein that induces noncanonical activation of JNK signaling ( Figure 6Q).

Discussion
In this study, we have identified the clustered miRNAs miR-306 and miR-79 as novel antitumor miRNAs that selectively eliminate JNK-activated tumors from Drosophila imaginal epithelia. Mechanistically, miR-306 and miR-79 directly target RNF146, an E3 ligase that promotes degradation of a poly-ADPribose polymerase Tnks, thereby leading to upregulation of Tnks and thus promoting JNK activation ( Figure 6K). Importantly, this noncanonical mode of JNK activation has only a weak effect on normal tissue growth but it strongly blocks tumor growth by overactivating JNK signaling when tumors already possess elevated JNK signaling via the canonical JNK pathway ( Figure 6K). Given that tumors or premalignant mutant cells often activate canonical JNK signaling, miR-306 and miR-79 can be novel ideal targets of cancer therapy.
Our study identified several putative co-target genes of miR-306 and miR-79 ( Figure 5A). Interestingly, some of these genes (Atf3, chinmo, and chn) have been reported to be involved in tumor growth in Drosophila. Atf3 encodes an AP-1 transcription factor that was shown to be a polarity-loss responsive gene acting downstream of the membrane-associated Scrib polarity complex (Donohoe expressing indicated protein and dsRNA targeting indicated gene. (E, F) Quantification of relative Tnks-myc levels (E) and p-JNK (F) levels in (D) from three independent experiments. Error bars, SD; *p<0.05, ***p<0.001 by one-way ANOVA multiple-comparison test. (G-J) Eye-antennal disc bearing GFP-labeled clones of indicated genotypes (G, H, 5 days after egg laying, I, J, 7 days after egg laying). (K) Quantification of clone size (% of total clone area per disc area in eye-antennal disc) of (G-J). Error bars, SD; ****p<0.0001 by one-way ANOVA multiple-comparison test. Source data 1. Quantitative data or raw data for Figure 6 (part 1).
Source data 2. Quantitative data or raw data for Figure 6 (part 2).
Source data 3. Genotypes for Figure 6.    , 2018). Knockdown of Atf3 suppresses growth and invasion of Ras V12 /scrib -/tumors in eyeantennal discs (Atkins et al., 2016). Chinmo is a BTB-zinc finger oncogene that is upregulated by JNK signaling in tumors (Doggett et al., 2015). Although loss of chinmo does not significantly suppress tumor growth, overexpression of chinmo with Ras V12 or an activated Notch is sufficient to promote tumor growth in eye-antennal discs (Doggett et al., 2015). Chn encodes a zinc finger transcription factor that cooperates with scrib -/to promote tumor growth (Turkel et al., 2013). Although we found that knockdown of these genes did not activate JNK signaling, it is possible that these putative target genes also contribute to the miR-306/miR-79-induced tumor suppression.
Intriguingly, it has been reported that miR-79 is downregulated in Ras V12 /lgl-RNAi tumors in Drosophila wing discs (Shu et al., 2017). Given that miR-306 is located in the same miRNA cluster with miR-79, it is highly possible that miR-306 is also downregulated in tumors. This suggests that tumors have the mechanism that downregulates antitumor miRNAs for their survival and growth. Future studies on the mechanism of how tumors regulate these miRNAs would provide new understanding of tumor biology.
Our study uncovered the miR-306/79-RNF146-Tnks axis as noncanonical JNK enhancer that selectively eliminates JNK-activated tumors in Drosophila. Considering that miR-9, the mammalian homolog of miR-79, is predicted to target mammalian RNF146 ( Figure 5-figure supplement 6) and that JNK signaling is highly conserved throughout evolution, it opens up the possibility of developing a new miRNA-based strategy against cancer.

Materials and methods
Fly stocks threshold of the fluorescence. Total clone area/disc area (%) in the eye-antennal disc was calculated using ImageJ and Prism 8 (GraphPad).
Plasmid and in vitro transcription of dsRNA pAc5.1/V5-His vector (Thermo Fisher Scientific, Cat #V411020) was used to construct plasmids for expressing proteins or miRNAs in Drosophila S2 cells. The RNF146 or Tnks ORF was amplified from fly cDNAs via PCR. The RNF146 ORF was cloned into the EcoR І-Xho І site of the pAc5.1/V5-His vector. The Tnks ORF carrying a myc tag at its 5′-end was cloned into the Kpn І-Xho І site of the pAc5.1/ V5-His vector. Extended region of miR-306 (-184 to +136) or miR-79 (-124 to +131) was amplified from fly cDNAs via PCR and cloned into the Kpn І-EcoR І site of the pAc5.1/V5-His vector.

Cell culture and transfection
Drosophila S2-ATCC cells (RRID:CVCL_Z232) was obtained from American Type Culture Collection (ATCC). Its identity was confirmed by visual inspection of the cell morphology and its growth kinetics in Schneider's Drosophila medium (Thermo Fisher Scientific, Cat #21720024)/10% fetal bovine serum (FBS) and penicillin/streptomycin. A mycoplasma test is usually not done for S2 cells.
For transfection assay, S2 cells were plated in 100 mm plates or six-well plates and grown overnight to reach 70% confluence. After that, DNA plasmids or dsRNAs were transfected into the cells using FuGene HD transfection reagent (Promega, Cat #PRE2311) according to the manufacturer's protocol. The protein synthesis inhibitor CHX (Santa Cruz Biotechnology, Cat #SC-3508) was used at 50 μg/ml.

Dual-luciferase reporter assay
The psiCHECK-2 vector (Promega, Cat #C8021) was used to construct plasmids for dual-luciferase reporter assay. RNF146 3′UTR or its mutant was cloned into the Xho І-Not I site of the psiCHECK-2 vector. Renilla luciferase activity and firefly luciferase activity were measured using GloMax-Multi Jr Single-Tube Multimode Reader (Promega) according to the manufacturer's protocol.

Statistical analysis
When comparing two groups, statistical significance was tested using a Student's t-test. When comparing multiple groups, statistical significance was tested using a one-way ANOVA multiple-comparison test.

Additional files
Supplementary files • Transparent reporting form

Data availability
All relevant data are within the paper and its Supporting Information files. All the numerical data that are represented as a graph in a figure are provided in the Source Data file.