Defining an evolutionarily conserved role of GW182 in circular RNA degradation

Dear Editor, Circular RNAs (circRNAs) are covalently closed RNA molecules derived from thousands of protein-coding genes via “backsplicing”. In many cases, the “backsplicing” step can be trigged by the flanking complementary intronic repeat elements that efficiently bring the intervening splicing sites into close proximity. Most circRNAs are cytoplasmic, and nuclear export of mature circRNAs is regulated in a length-dependent manner. While the functions of circRNAs remain largely unknown, recent reports have revealed that some circRNAs can control gene expression by affecting transcription, acting as splicing regulators and mircoRNA sponges. It is also becoming evident that circRNAs are associated with several diseases such as cancer and brain disorders. Due to the lack of a defined 5′ or 3′ end, circRNAs are naturally more stable than their parental linear RNAs as they are not targets for the exosome or exonuclease. This was exemplified by the circRNAs derived from Drosophila dati and laccase2 gene or our previously described expression plasmids (Fig. 1a; Supplementary Fig. S1). Nevertheless, how circRNAs are degraded or what factors contribute to a surveillance pathway is largely unclear. To reveal the factors which are required for degradation of circRNAs, we employed a focused RNAi screening in Drosophila DL1 or S2 cells targeting 31 genes with known roles in RNA metabolism, and then assessed the levels of steady-state circRNAs by qRT-PCR (Fig. 1b; Supplementary Figs. 2–3. In line with previous study, knockdown of RNA decay factors Pop2 (also known as CAF1), Not1, and DCP2 led to significant accumulation of Vha68-1 mRNA (Supplementary Fig. S2b), but had no effect on the levels of steady-state circRNAs (Fig. 1b). We instead found that depletion of GW182 resulted in accumulation of both steady-state circdati and circlaccase2 transcripts (Fig. 1b). GW182 is a key component of P-body and RNAi machine as it facilitates the assembly of P-body and acts as a molecular scaffold bringing together RNA-induced silencing complexes and various mRNA decay enzymes. However, depletion of other P-body components or RNAi machine factors did not have effect on steady-state circRNA levels, indicating that P-body or RNAi machine does not affect circdati or circlaccase2 degradation (Fig. 1b). To examine whether GW182 exerts a general or limited role in controlling circRNA levels, we tested the levels of 12 additional circRNAs that were of varying length and exon counts in GW182-depleted DL1 cells (Fig. 1c; Supplementary Table S1). The levels of most steady-state circRNAs were significantly increased upon GW182 depletion. Importantly, the role of GW182 appears to be robust in affecting circRNA stability because (i) the levels of most nascent circRNAs were not affected upon GW182 depletion, suggesting that circRNA biogenesis is largely unaffected by GW182 (Fig. 1d; Supplementary Fig. S4a, b), (ii) circRNA accumulation was also observed in GW182depleted Drosophila S2 cells genome widely (Fig. 1e–g; Supplementary Figs. S4–S6), (iii) the enriched circRNAs were also verified using a secondary GW182 dsRNA (Supplementary Fig. S4c, d), (iv) overexpression of GW182 decreased the levels of steady-state circRNAs (Supplementary Fig. S4e), (v) depletion of GW182 had no effect on nuclear circRNA levels, while cytoplasmic circRNAs accumulated (Fig. 1h; Supplementary Fig. S7), and (vi) GW182 depletion had little effect on degradation of circRNAs’ parental mRNAs (Fig. 1f, g; Supplementary Figs. S4d, S6c–e). Taken together, these data suggested


Drosophila cell culture, RNAi, and transfections
Drosophila DL1 and S2 cells were cultured at 25°C with Schneider's Drosophila medium (Sigma, S9895) plus 10% fetal bovine serum (HyClone, SH30910.03) and 1% (v/v) penicillinstreptomycin (Thermo Fisher Scientific, 15140122). dsRNAs from the DRSC (Drosophila RNAi Screening Center) were generated by in vitro transcription (MEGAscript kit, Thermo Fisher Scientific, AM1334) of PCR templates containing the T7 promoter sequence on both ends (Supplementary information, Table S3). To measure RNA half-lives, DL1 cells were treated with actinomycin D (1 μg/mL; Sigma, A4262) for the indicated amounts of time. For RNAi experiments in DL1 cells, 0.5 million of cells in 12-well dishes were bathed with 2 μg of dsRNA for 3 days.
For RNAi experiments in S2 cells, 1.5 million of cells in 12-well dishes were bathed with 8 μg of dsRNA for 3 days. RNA was isolated using Trizol (Thermo Fisher Scientific, 15596018) according to the manufacturer's instructions.
To generate plasmids for transfection into Drosophila DL1 and S2 cells, the indicated sequences were inserted into a pMK33/pMtHy-based vector between the metallothionein promoter (pMT) and the SV40 polyadenylation signal as described in Supplementary Plasmid Information.
For circRNA expression plasmids transfection, 2 million of Drosophila DL1 cells in 6-well plate were transfected with 2 μg plasmid using Effectene transfection reagent (QIAGEN, 301425). A final concentration of 500 μM copper sulfate (MACKLIN, C805782) was added for 3 h to induce expression followed by washing the cells twice with media containing 500 mM bathocuproine disulphonate (BCS; Sigma, B1125-500MG). For GW182 expression plasmids transfection, 0.5 million of Drosophila S2 cells in 6-well plate were transfected with 1 μg plasmid using Effectene transfection reagent (QIAGEN, 301425). A final concentration of 500 μM copper sulfate was added for 48 h to induce expression.

Nuclear and cytoplasmic fractionation
Cellular fractionation was performed as previously described 2 . In brief, Drosophila S2 cells were washed twice with PBS and resuspended with slow pipetting in 1 mL lysis buffer B (10 mM Tris-HCl pH 8, 140 mM NaCl, 1.5 mM MgCl2, 0.5% IGEPAL CA-630, 1 mM dithiothreitol, and 80 U/mL RNase inhibitor (Thermo Fisher Scientific, 10777019)). Nuclei were collected by centrifugation at 1,000g for 3 min at 4°C and the supernatant was saved as the cytoplasmic fraction.
Nuclei were resuspended in 1 mL lysis buffer B and 100 μL of detergent (3.3% (w/v) sodium deoxycholate, 6.6% (v/v) Tween 40) was added. Samples were slowly vortexed for 10 sec and incubated on ice for 5 min. Nuclei were then collected by centrifugation at 1,000g for 3 min and washed with 1 mL of lysis Buffer B. The final pellet (nuclear fraction) was resuspended in 1 mL Trizol. Efficient fractionation of nuclear and cytoplasmic RNAs was verified by testing levels of U6 snRNA and 18S rRNA.

RNA-seq library preparation and data analysis
Drosophila S2 cells were treated with either β-gal dsRNA or GW182 dsRNA for 3 days and then total RNA was extracted by using Trizol (Thermo Fisher Scientific, 15596018) according to the manufacturer's instructions. RNA-seq libraries were prepared by using TruSeq Stranded Total RNA Library Prep Gold kit (Illumina 20020598) following the manufacturer's instructions and were subsequently sequenced for 150 nt from both ends on Illumina HiSeq 2500 platform.
Sequencing reads were filtered by using fastp 11 to remove low quality bases and adaptor sequences from both ends of reads. The remaining reads were mapped to the Drosophila genome (BDGP6.22, downloaded from Ensembl) using TopHat2 (version 2.0.14) with default parameters 12 . For each library, gene expression levels were estimated by using Cufflinks version 2.2.1 13 based on Ensembl annotation (BDGP6.22, version 96). In order to compare the gene expression levels between β-gal dsRNA and GW182 dsRNA samples, the FPKM (Fragments Per Kilobase per Million mapped reads) of each gene was converted to TPM (Transcripts Per Million mapped reads) 14 . As most circRNAs were derived from protein coding genes, only protein coding genes were retained for further analysis.
To identify circRNAs with CIRI2 (version 2.0.6) 15 , RNA-seq reads were first mapped to Drosophila genome (BDGP6.22, downloaded from Ensembl) with bwa 16 and then the alignments were inputted to CIRI2 with default parameters. At least two junction reads in one of the libraries was required for a circRNA to be retained for further analysis. Expression levels of circRNAs were estimated by TPM (Transcripts mapped to back-splicing junctions Per Million uniquely mapped fragments). High confidence circRNAs were determined as those with TPM  0.1 in at least two samples. Only circRNAs derived from exon regions of individual protein coding gene according to Ensemble annotation (BDGP6.22, version 96) were retained for further analysis.

qRT-PCR
Complementary DNA for qRT-PCR was synthesized using PrimeScript RT Master Mix (Takara, RR036A) according to the manufacturer's instructions. qPCR was then performed in triplicate using FastSYBR Mixture (CWBIO, CW0955M). All qPCR primer sequences are provided in Supplementary Table S4 and S5.

Metabolic labeling of nascent RNA with 4sU and nascent RNA purification
Metabolic labeling of nascent RNA with 4sU and nascent RNA purification was performed as described previously with modifications 2, 17, 18 . Drosophila DL1 cells, treated with the indicated dsRNA for 3 days, were incubated with 250 μM 4sU (Sigma, T4509) for 5 min to label newly transcribed RNA. 20 μg 4sU labeled RNA was incubated in 500 µL biotinylation buffer (10 mM Tris pH 7.4, 1 mM EDTA) with 10 μg/mL MTSEA biotin-XX (Biotium, 90066; dissolved in dimethylformamide) at room temperature for 1.5 h. To verify that there was not significant variation across the biotinylation reactions, 2 ng of synthetic RNA was included in each reaction (Supplementary information, Table S5). Unbound MTSEA biotin was removed by equal volume Chloroform:Isoamyl alcohol (Sigma, C0549-1PT) extraction, and RNA was precipitated at 12,000g for 30 min at 4°C with 1:10 volume of 5 M NaCl and an equal volume of isopropanol.
The RNA pellet was washed with 80% ethanol and resuspended in 100 μL DEPC treated water.

Northern Blotting
Northern Blots using DIG Northern Starter Kit (Roche, 12039672910) were performed as previously described 19,20 . Digoxin-labeled RNA probe was prepared with the corresponding PCR products as templates for T7 transcription according to the manufacturer's instructions. Blots were viewed with a Bio-Rad ChemiDoc Imaging System. HIPK3 RNA probe sequence:

Statistical analyses
Statistical significance for comparisons of means was assessed by Student's t-test for both qRT-PCR and the percentage of circRNA junction reads in total uniquely mapped fragments. Statistical details and error bars are defined in each figure legend: P < 0.01 ( * * ) and P < 0.05 ( * ). Mann-Whitney U test was used to compare the statistical difference of gene expression fold change and the ratio of circRNA junction reads in total splicing reads.

Data availability
The RNA-seq data generated by this study have been deposited in the NCBI GEO database under the accession number GSE132100.         Supplementary Figure S6 Rep:               (a) Domain structure of Drosophila GW182 and its human homolog (TNRC6C).
Detailed sequence alignment is provided in Supplementary Protein Information. (b, c) Human HeLa cells were treated with a control siRNA or specific siRNAs to knock down TNRC6A, TNRC6B or TNRC6C for 2 days. qRT-PCR was then performed to assess the knock down efficiency (b) and measure levels of endogenous MALAT1, Hsp70, URH49 RNAs (c) or parental mRNAs of circRNAs (related to Figure 1j) (d).