CiRS-7 Enhances the Liquid-liquid Phase Separation of miRISC and Promotes DNA Damage Repair

ABSTRACT

Noncoding RNAs have been found to play important roles in DNA damage repair, whereas the participation of circRNA remains undisclosed. Here, we characterized ciRS-7, a circRNA containing over 70 putative miR-7-binding sites, as an enhancer of miRISC condensation and DNA repair. Both in vivo and in vitro experiments confirmed the condensation of TNRC6B and AGO2, two core protein components of human miRISC. Moreover, overexpressing ciRS-7 largely increased the condensate number of TNRC6B and AGO2 in cells, while silencing ciRS-7 reduced it. Additionally, miR-7 overexpression also promoted miRISC condensation. Consistent with the previous report that AGO2 participated in RAD51-mediated DNA damage repair, the overexpression of ciRS-7 significantly promoted irradiation-induced DNA damage repair by enhancing RAD51 recruitment. Our results uncover a new role of circRNA in liquid-liquid phase separation and provide new insight into the regulatory mechanism of ciRS-7 on miRISC function and DNA repair.

KEYWORDS:

• CircRNA
• ciRS-7
• condensation
• DNA repair
• liquid-liquid phase separation
• miRISC

Introduction

The microRNA (miRNA)-mediated gene silencing is an important post-transcriptional regulation that is integral to diverse physiological or pathological processes, including inflammation, differentiation, stress response, and the progression of cancer [1–3]. Generally, mature miRNAs are recruited to the Argonaute protein 2 (AGO2), forming the core of miRNA-induced silencing complex (miRISC), and bound to the partially complementary sequences in the 3’ untranslated regions (UTRs) of target mRNA, thereby inducing translation inhibition, deadenylation, decapping and decay of mRNAs [4–6]. The miRISC contains two core protein components, AGO2 and Argonaute bound GW182, termed trinucleotide repeat-containing protein (TNRC6) in mammalian [6–8]. Recently, liquid-liquid phase separation (LLPS) has been emerging as a key mechanism underlying the formation of membrane-less organelles and compartmentalization of biochemical reactions [9,10]. Sheu-Gruttadauria and MacRae et al. demonstrated that AGO2 possessed a PIWI domain composed of three tryptophan-binding regions making diverse interactions with the N-terminal, unstructured glycine/tryptophan (GW)-rich Argonaute binding domain (ABD) domain of TNRC6B [11]. Such multivalent interactions between AGO2 and TNRC6B drive miRISC undergo LLPS to form highly concentrated condensates, which further accelerate the miRISC-mediated mRNA deadenylation [12]. Previous studies have reported that AGO2 promotes homologous recombination (HR)-mediated DNA repair [13,14], whereas whether LLPS is required for its function remains unknown.

Circular RNAs (circRNAs) are single-stranded, covalently closed RNA molecules without a 5’ terminal cap structure and 3’ terminal polyadenylated tail, and therefore resistant to the deadenylation and decapping by miRISC [15,16]. CircRNAs could act as miRNA sponges that sequester miRNA from their target mRNA and prevent mRNA from miRISC-mediated suppression [17]. Specifically, ciRS-7 (also known as CDR1as), a highly expressed circRNA in human cells, has been identified as a sponge for miR-7 for it contains more than 70 selectively conserved miRNA binding sites [18,19]. Thomas et al. demonstrated that miR-7-binding sites within ciRS-7 facilitated the formation of ciRS-7-miR-7-AGO2 ternary complex [20]. However, whether ciRS-7-mediated multivalent interaction between miRISCs influenced the LLPS potential of miRISC was unknown. Here, we reported that the multiple binding between miR-7-containing miRISC and ciRS-7 increased the multivalent interaction among miRISCs, thereby enhancing the miRISC condensation, which further promoted AGO2-mediated DNA repair.

Materials and methods

Cell lines and transfections

HEK293T and HeLa cell lines were purchased from American Type Culture Collection (ATCC). All cell lines were mycoplasma-free and were authenticated using STR profiling by the provider ATCC. HEK293T and HeLa cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM, Gibco, ThermoFisher Scientific, Waltham, Massachusetts, USA) supplemented with 2 mM L-glutamine, 100 units/mL penicillin-streptomycin (15140122, Hyclone, South Logan, UT, USA) and 10% fetal bovine serum (FBS, Gibco), and cultured at 37°C with 5% CO₂. For routine cell culture passaging, trypsin-EDTA (25300062, Gibco) was used to detach cells from the cell culture flask or plates.

Transfections of plasmids were conducted with Lipofectamine 2000 (11668019, Invitrogen, ThermoFisher Scientific) or FuGENE HD Transfection Reagent (E2311, Promega, Madison, Wisconsin, USA), while transfections of siRNA or miRNA mimics were conducted with Lipofectamine RNAiMAX (56532, Invitrogen) according to manufacturer’s protocol. The sequences of siRNAs used in this study are given in Table S1.

Construction of plasmids

The pmyc-EGFP-TNRC6B was a gift from Kumiko Ui-Tei (Addgene plasmid #42022; http://n2t.net/addgene:42022; RRID: Addgene_42022) [21], and EGFP-hAGO2 was a gift from Phil Sharp (Addgene plasmid #21981; http://n2t.net/addgene:21981; RRID: Addgene_21981) [22]. For recombinant protein purification, human cDNA of AGO2 (amino acids 1–859) was cloned from EGFP-hAGO2 plasmid and inserted into pQB3 vector (with His tag); human cDNA of TNRC6B-ABD domain (amino acids 437–1056) was cloned from pmyc-EGFP-TNRC6B plasmid and inserted in-frame before a C-terminal mEGFP into pGEX-6p-2 vector (with a GST tag).

The pcDNA3.0-ciRS-7, pcDNA3.0-ciRS-7-ir, and pcDNA3.0-ciRS-7-fs plasmids were constructed via the methods proposed by Thomas B. Hensen et al [18]. Briefly, the ciRS-7 exon was cloned by PCR using ciRS-7-fs F/R primers and inserted into pcDNA3.0 to generate pcDNA3.0-ciRS-7-fs. The ciRS-7 exon along with 1kb upstream and 200bp downstream to the nonlinear splice sites, was amplified using ciRS-7 forward/reverse (F/R) primers and inserted into pcDNA3.0 to generate pcDNA3.0-ciRS-7-ir. Then, an ~ 800bp DNA stretch of the upstream flanking sequence was amplified using ciRS-7-ir F/R primers and reversely inserted into the 3’ downstream of ciRS-7-ir in pcDNA3.0-ciRS-7-ir plasmid, thus generating the pcDNA3.0-ciRS-7 with ciRS-7 exons flanked by inverted repeats. The sequences of primers are given in Table S1.

Real-time quantitative PCR (qRT-PCR)

The total RNA of cells was extracted using TRIzol^TM reagent (15596–026, Invitrogen) according to the manufacturer’s instructions. Two micrograms of total RNA were used to synthesize cDNA using ReverTra Ace qPCR RT Master Mix with gDNA Remover (FSQ-301, TOYOBO). All the qRT-PCR were performed on a LightCycler480 System (Roche, Basel, Switzerland) using SYBR® Green Realtime PCR Master Mix (QPK-201, TOYOBO). Between duplicate wells, cycle threshold (Ct) values differing by less than 0.5 were used for further analysis. ΔCt was calculated by Ct of the target gene minus Ct of internal control, and the 2^−ΔCt value was used as the relative expression level. GAPDH was used as an internal control. The sequences of primers are given in Table S1.

Construction of HeLa cell line stably expressing ciRS-7

The pcDNA3.0-ciRS-7 plasmids were transfected into the HeLa cells with Lipofectamine 2000 (Invitrogen). Forty-eight hours after transfection, cells were selected with 2 μg/ml puromycin for 1 week, detached, and seeded into 96-well plates at 1 cell per well. Two weeks later, cell clones were amplified, and the overexpression of ciRS-7 was verified using qRT-PCR. GAPDH was used as an internal control. The sequences of primers are given in Table S1.

Live cell imaging and fluorescence recovery after photobleaching (FRAP)

HEK293T cells were seeded in a 35 mm glass-bottom dish twelve hours before transfection with the pmyc-EGFP-TNRC6B or EGFP-hAGO2 plasmid. Twenty-four hours after transfection, cells were subjected to live-cell imaging using a Zeiss LSM880 confocal microscope equipped with an incubation chamber providing a humidified atmosphere at 37°C with 5% CO₂.

For FRAP, a region of interest (ROI) for bleaching was labeled with a circle. Bleaching was performed at 100% power of 488 nm laser (scan speed, 0.67 μs; iterations, 1000) under 64× objective. Non-bleached control region was labeled at the same time. Images were captured every 1 s and the fluorescence intensities of ROI were collected. The dynamics of fluorescence intensity were calculated using ZEN3.1 (Blue edition, 3.1) relative to that of the pre-bleach time point.

Expression and purification of TNRC6B ABD-mEGFP fusion protein

GST-TNRC6B-ABD-mEGFP fusion protein was expressed in the Escherichia coli strain BL21 (DE3). Cultures were grown at 37°C, 220 rpm until the OD₆₀₀ reached 0.6–0.7 and protein production was induced with 0.5 mM isopropyl beta‐d‐thiogalactopyranoside (IPTG, 367-93-1, Sangon Biotech, Shanghai, China), followed by growth at 16°C, 200 rpm for 16–20 h. Cells were collected by centrifugation, resuspended in GST-lysis buffer (20 mM Tris/HEPES, pH 7.5, 150 mM NaCl, 10% glycerol, 1 mM dithiothreitol (DTT, 3483-12-3, Sangon Biotech) supplemented with 1 mM phenylmethanesulfonyl fluoride (PMSF), and underwent sonication lysis. The lysate was clarified by high-speed centrifugation (15000×g, 15 min, twice), and the supernatant was incubated with GST-tagged purification resin (P2253, Beyotime, Shanghai, China) at 4°C for 2 h in a gently rotate speed. The resin was washed three times with GST-lysis buffer. The GST-tagged protein was eluted by GST-lysis buffer supplemented with 25 mM glutathione (GSH, 70-18-8, Sangon Biotech) and then incubated with PreScission Protease (P2303, Beyotime) supplemented with 1 mM DTT at 4°C overnight to remove the GST tag. The excised protein was then applied to a Superdex200 10/300 gel filtration column (28990944, Cytiva, Marlborough, MA) using AKTA system (Cytiva), and the fractions containing ABD-mEGFP were pooled and concentrated in high salt buffer (20 mM HEPES, pH 7.5, 1 M NaCl, 2% glycerol, 1 mM DTT). Final proteins were flash-frozen in liquid N₂ and stored at −80°C. All purification steps were performed on ice or at 4°C.

Expression and purification of his-AGO2 recombinant protein

His-AGO2 protein was expressed in Spodoptera frugiperda (Sf9) insect cells using a baculovirus expression system. Sf9 cells were grown in Grace’s insect medium (Gibco) supplemented with 1% FBS to a density of 2 × 10⁶ cells per ml at 27°C in a shake flask, and then co-transfected pQB3-AGO2 plasmid and qBac-III (qBac Bacmid) with FuGENE HD Transfection Reagent (Promega) to generate recombinant baculoviruses which were collected at 4–5 days. The baculoviruses supernatants were amplified for multiple rounds with high titer and then used to infect new Sf9 cells at specific MOI for another 4–5 days. Cells were harvested by centrifugation and resuspended in a nondenaturing lysis buffer for ultrasonic lysis. The lysate was centrifugated, and the supernatant was incubated with Ni-NTA resin (P2229S, Beyotime) at 4°C overnight in a gently rotate speed. After several additional washes, the protein was eluted in elution buffer, and the eluted protein was further purified using a Superdex200 10/300 column (Cytiva). Purified His-AGO2 was concentrated in high salt buffer (20 mM HEPES, pH 7.5, 1 M NaCl, 2% glycerol, 1 mM DTT). Final proteins were flash frozen in liquid N₂ and stored at −80°C. All purification steps were performed on ice or at 4°C.

In vitro droplet assay

The purified recombinant TNRC6B-ABD-mEGFP protein was diluted to the indicated concentration in buffers containing 20 mM Tris-HCl (pH 7.4) and 150 mM NaCl with or without 5% polyethylene glycol (PEG8000) (89510, Sigma-Aldrich, St. Louis, Missouri, USA) in a volume of 20 μL. The purified AGO2 protein was added into the reaction volume at the presence of 5% PEG8000. The reaction mixtures were incubated at room temperature (RT) for 10 min and pipetted on glass slides for 2 min to make droplets settle down. Images were captured within 5 min using a Zeiss LSM880 confocal microscope with a 64× oil objective and further processed by ZEN software (Blue edition, 3.1).

Western blotting and Coomassie brilliant blue staining

After transfected with siRNA or plasmids for 48 h, HEK293T cells were lysed in RIPA buffer (25 mM Tris-HCl, pH 7.4, 150 mM NaCl; 1% NP-40; 0.5% Na-deoxycholate) supplemented with protease inhibitor cocktail (200-664-3, Roche, Basel, Switzerland). Proteins were separated by 8–12% SDS-PAGE, blotted onto PVDF membrane and incubated with primary antibodies (rabbit anti-AGO2, ab186733, Abcam, Cambridge, MA, USA, diluted in 1:2000; mouse anti-GAPDH 60,004–1-Ig, Proteintech, Rosemont IL, USA, diluted in 1:5000) overnight for 4°C, followed by incubation with secondary antibody (goat anti-rabbit IgG (H+L) HRP, GAR0072; goat anti-mouse IgG (H+L) HRP, GAM0072, MultiSciences, Hangzhou, China, both diluted in 1:5000) for 1 h at RT.

For Coomassie brilliant blue staining, approximately 200 μg of purified protein samples were diluted in 1×loading buffer (E151, Genstar, Beijing, China) and boiled in 100°C for 10 min. Proteins were separated by 8–12% SDS-PAGE, stained by Coomassie blue G250 (S19061, Yuanye Bio-Technology, Shanghai, China) for 1 h in RT at a low horizontal rolling speed, and finally washed for several times till the background of gel turned transparent.

Immunofluorescence (IF)

Cells were plated on coverslips in 24-well plate and transfected with siRNA (the sequences of siRNAs used in this study are given in Table S1) or plasmids for 48 h or irradiated (X-ray, 3 Gy) for the indicated time. Then, cells were fixed with 4% paraformaldehyde (DF0135, Leagene, Beijing, China) for 15 min at RT, washed three times with 1× PBS, and were blocked in blocking buffer (5% goat serum, 0.3% Triton X-100 in 1× PBS) for one hour at RT, followed by incubation of primary antibodies diluted in blocking buffer for 2 h at RT or 4°C overnight. After three times washes in 1× PBS, the samples were treated with secondary antibodies tagged with Alexa Fluor 555 (4413S, Cell Signaling Technology, CST, Danvers, MA, USA, diluted in 1:500) for one hour at RT in the dark. Cells were washed three times in 1× PBS and then stained with DAPI (D9542, Sigma-Aldrich, diluted in 2 μg/ml in 1× PBS) for 5 min. Finally, the glass slides were mounted in ProLong^TM Diamond Antifade Mountant (P36965, Invitrogen) before being visualized by LSM880 Zeiss confocal microscope. The primary antibodies, rabbit anti-AGO2 (ab186733, Abcam, diluted in 1:200), rabbit anti-γ-H2A.X (9718S, CST, diluted in 1:500), and rabbit anti-RAD51 (ET1705–96, HuaBio, Hangzhou, China, diluted in 1:100), were used in this study.

Quantification and statistical analysis

All foci or puncta numbers were counted manually. The data were represented as the means ± SD from independent experiments performed in triplicate. The differences between two groups were analyzed using an unpaired t test and one-way analysis of variance was used when more than two groups were compared. A P value of less than 0.05 was considered statistically significant. All statistical tests were two-sided and were performed using GraphPad Prism 8 (GraphPad Software Inc., San Diego, CA).

Results

miRISC proteins undergo LLPS in vivo and in vitro

We firstly confirmed the LLPS potential of TNRC6B and AGO2, the two core proteins of miRISC. Proteins with intrinsically disordered regions (IDRs) are more inclined to undergo LLPS. The PONDR (http://www.pondr.com/) analysis indicated that TNRC6B had a long IDR with a high average prediction score as 0.83 (Figure 1a, left panel), while AGO2 showed highly ordered with a low prediction score as 0.34 (Figure 1a, right panel). Ectopically expressed EGFP-TNRC6B and EGFP-AGO2 protein formed puncta in HEK293T cells (Figure 1b,c). EGFP-AGO2 protein formed big condensates but its number was much less than that of EGFP-TNRC6B (Figure 1c; Fig. S1A-C). The FRAP assay was performed to investigate the dynamic exchange between liquid-like condensate and its surroundings. As shown in Figure 1d, the fluorescence of EGFP-TNRC6B recovered shortly after photobleaching while EGFP-AGO2 puncta showed a very slight recovery, indicating that TNRC6B underwent LLPS in cells.

Figure 1. miRISC proteins underwent LLPS in vivo. a. Bioinformatics analysis of intrinsically disordered regions (IDR) of TNRC6B (left penal) and AGO2 (right penal) using the online tool, PONDR (www.pondr.com). The purple line represented the predicted score and regions above 0.5 represented the IDR. The ABD domain (residues, 437aa-1056aa) of TNRC6B possessed a high average score. b. Exogenous EGFP-TNRC6B showed numerous liquid-like puncta in vivo. HEK293T cells were transfected with plasmids for 24 h and then fixed and stained with DAPI for immunofluorescence before observation using confocal microscopy. c. Exogenous EGFP-AGO2 showed several puncta in the cytoplasm of HEK293T cells. d. FRAP assays of EGFP-TNRC6B and EGFP-AGO2 in HEK293T cells. The bleached punctum was labeled by a white circle. Four puncta from different replications were analyzed (n = 4). The dynamics fluorescence intensities were recorded and plotted in the curve. Data were presented as mean ± SD.

To further verify the LLPS properties of TNRC6B in vitro, we purified the N-terminal ABD (437–1056 aa) of TNRC6B (Figure 2a; Fig. S2), which was highly disordered (Figure 1a) and responsible for the interaction with the tryptophan-binding pockets of AGO2 [11]. Consistent with the results in cells, purified recombinant ABD-mEGFP protein formed spherical droplets in the presence of PEG8000, a molecular crowding agent (Figure 2b). The FRAP results showed that approximately 60% of ABD-mEGFP droplets were mobile and dynamically exchanged with the surrounding (Figure 2c). Moreover, the addition of purified His-AGO2 protein significantly enhanced the phase separation of ABD-mEGFP, presenting as a phase transition from liquid-like into solid-like property (Figure 2d-e). Together, these results indicated that the highly disordered ABD domain contributed to the LLPS of TNRC6B, while AGO2 enhanced the solid-like property of condensates.

Figure 2. miRISC proteins undergo LLPS in vitro. a. Coomassie brilliant blue staining of purified TNRC6B-ABD-mEGFP recombinant protein. b. Droplets formation assay of purified TNRC6B-ABD-mEGFP protein under distinct protein concentration (2 μM-20 μM) in buffers (20 mM Tris-HCl, pH 7.4, 150 mM NaCl) with or without 5% PEG8000. c. FRAP of TNRC6B-ABD-mEGFP droplets in vitro. The bleached punctum was labeled with a red circle, and the unbleached region was labeled with a yellow circle. Three droplets from different replications were analyzed (n = 3). The dynamics fluorescence intensities were recorded and plotted in the curve. Data were presented as mean ± SD. d. Coomassie brilliant blue staining of purified His-AGO2 protein. e. Condensation of TNRC6B-ABD-mEGFP protein incubated with BSA or His-AGO2 protein.

ciRS-7 promotes miRISC condensation

ciRS-7 is the first functionally characterized circRNA containing more than 70 conserved miRNA target sites, which may enhance the multivalent interactions between ciRS-7 and miRISC. To investigate the effects of ciRS-7 on LLPS of miRISC, we constructed the pcDNA3.0-ciRS-7 plasmid to overexpress ciRS-7, and two variant plasmids lacking inverted flanking sequence, pcDNA3.0-ciRS-7-fs, and pcDNA3.0-ciRS-7-ir, as negative controls [18] (Figure 3a-b). Interestingly, the overexpression of ciRS-7 had a slight effect on EGFP-TNRC6B and AGO2 expression, while significantly increasing the number of EGFP-TNRC6B droplets (Figure 3d, upper panel) and AGO2 foci (Figure 3d, lower panel). Moreover, although the silencing of ciRS-7 (Figure 3e) had no impact on the expression of AGO2 protein (Figure 3f), it reduced both EGFP-TNRC6B droplets (Figure 3g, upper panel) and endogenous AGO2 foci (Figure 3g, lower panel). These results indicate that ciRS-7 promotes the LLPS of miRISC.

Figure 3. ciRS-7 promoted miRISC condensation. a. The diagram of ciRS-7-fs, ciRS-7-ir, and ciRS-7 expression vectors, referred to the paper of Thomas b. Hensen et al [18]. The ciRS-7-fs consisted of the ciRS-7 exon without flanking sequences and splice sites; ciRS-7-ir contained the 1 kb upstream and 200 bp downstream endogenous flanking genomic sequences; and ciRS-7 has an inverted upstream sequence in the 3’ terminal, as illustrated by the directional bars. CMV, cytomegalovirus promoter; pA, polyadenylation signal; SA, splice acceptor; SD, splice donor. b. Quantitative analysis of ciRS-7 expression in HEK293T cells transfected with pcDNA3.0-ciRS-7-ir, pcDNA3.0-ciRS-7-fs, pcDNA3.0-ciRS-7, and empty vector (EV). For (b), data from 2 replications were presented; Data were presented as mean ± SD. c. Effects on AGO2 and TNRC6B protein level in HEK293T cells with transient overexpression ciRS-7. d Effects of ciRS-7 overexpression on exogenous EGFP-TNRC6B puncta (upper penal) and endogenous AGO2 foci (lower penal) in HEK293T cells. Quantitative counts of EGFP-TNRC6B puncta and AGO2 foci per cell were presented. For EGFP-TNRC6B puncta, n = 18 (EV), n = 21 (ciRS-7-fs), n = 19 (ciRS-7-ir), n = 18 (ciRS-7); for AGO2 foci, n = 39 (EV), n = 139 (ciRS-7-fs), n = 80 (ciRS-7-ir), n = 132 (ciRS-7). e. Quantitative analysis of ciRS-7 expression in HEK293T cells transfected with siRNAs for ciRS-7. For (E), data from 4 replications were presented; Data were presented as mean ± SD. f. Effects on AGO2 protein level in HEK293T cells with suppression of ciRS-7. g. Effects of ciRS-7 silencing on EGFP-TNRC6B puncta (upper penal) and endogenous AGO2 foci (lower penal) in HEK293T cells. Quantitative counts of EGFP-TNRC6B puncta and AGO2 foci per cell were presented. For EGFP-TNRC6B puncta, n = 20 (siNC), n = 20 (siciRS-7-1), n = 20 (siciRS-7-2); for AGO2 foci, n = 172 (siNC), n = 136 (siciRS-7-1), n = 62 (siciRS-7-2). Data were presented as mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

miR-7 enhances the phase separation of miRISC

ciRS-7 was characterized as a miRNA sponge that was bound by a large number of miRNAs, especially miR-7. Next, we further elucidated whether miR-7 mediated the multivalent interactions between ciRS-7 and miRISC, and thereby evaluated its condensation. Although it had no influence on AGO2 protein level (Figure 4a-b), transfection of miR-7 significantly enhanced the puncta formations of EGFP-TNRC6B (Figure 4c) and AGO2 (Figure 4d) in HEK293T cells. Consistently, inhibiting miR-7 with its antisense oligonucleotides moderately inhibited the number of AGO2 foci but had no influence on EGFP-TNRC6B droplets (Figure 4e-f), while downregulation of miR-7 by silencing DICER1 (Fig. S3A-C) moderately reduced the number of EGFP-TNRC6B droplets and AGO2 foci (Figure 4g-h), suggesting that other miRNAs apart from miR-7 was involved in the regulation of miRISC LLPS. These results indicate that miR-7 enhances the LLPS of miRISC.

Figure 4. miR-7 accelerated phase separation of miRISC. a. Verification of miR-7 mimics or anti-miR-7 by qRT-PCR. Data were presented as mean ± SD. b. Effects on AGO2 protein level in HEK293T cells transfected with miR-7 mimics or miR-7 antisense oligonucleotide. c-d. Effects of miR-7 overexpression on exogenous EGFP-TNRC6B puncta (c) and endogenous AGO2 foci (d) in HEK293T cells. Quantitative counts of EGFP-TNRC6B puncta (n = 20 for siNC; n = 20 for miR-7) or AGO2 foci (n = 37 for siNC; n = 47 for miR-7) per cell were presented. Data were presented as mean ± SD. e-f. Effects of anti-miR-7 on exogenous EGFP-TNRC6B puncta (e) and endogenous AGO2 foci (f) in HEK293T cells. Quantitative counts of EGFP-TNRC6B puncta (n = 38 for anti-NC; n = 34 for anti-miR-7) or AGO2 foci (n = 115 for anti-NC; n = 91 for anti-miR-7) per cell were presented. Data were presented as mean ± SD. g-h. Effects of inhibiting total miRNAs generation by siDICER1 on exogenous EGFP-TNRC6B puncta (g) and endogenous AGO2 foci (h) in HEK293T cells. Quantitative counts of EGFP-TNRC6B puncta (n = 28 for siNC; n = 30 for siDICER1-2) or AGO2 foci (n = 87 for siNC; n = 83 for siDICER1-2) per cell were presented. Data were presented as mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

ciRS-7 promotes DNA damage repair

Given that AGO2 had been found to facilitate homologous recombination-mediated DNA repair by promoting RAD51 recruitment [13], we wondered if ciRS-7 played roles in DNA damage repair. We constructed the HeLa cells with stable overexpression of ciRS-7 (Sup Figure S4A) and exposed the cells to X-ray to induce DNA double-strand breaks (DSBs). Both the γ-H2A.X (pS139-H2A.X) and RAD51 were detected using IF. Upon irradiation, γ-H2A.X rapidly accumulated in the nucleus at 0.5 hour and was continuously cleared at 6 and 20 h. In agreement with our hypothesis, 20 h after irradiation, cells overexpressing ciRS-7 contained fewer foci of γ-H2A.X (Figure 5a), while ciRS-7-silencing cells contained more γ-H2A.X foci than control cells (Figure 5c), indicating that ciRS-7 promoted DNA repair. Consistently, the overexpression of ciRS-7 also enhanced the recruitment of RAD51 to DNA damage site (Figure 5b) while silencing siRS-7 reduced RAD51 foci (Figure S4B). Meanwhile, inhibiting miR-7 by its antisense oligonucleotide suppressed the clearance of radiation-induced γ-H2A.X (Figure 5d; Figure S4C). The results suggested that the ciRS-7 might promote RAD51 accumulation to enhance radiation-induced DNA damage repair.

Figure 5. ciRS-7 promoted AGO2-mediated DNA damage repair. a. Detection of γ-H2A.X by immunofluorescence assay in HeLa-ciRS-7 or HeLa-EV cells at different time points (0.5 h, 6 h, 20 h) after irradiation. Quantitative counts of γ-H2A.X foci per cell or cells with more than 5 γ-H2A.X foci were presented. For γ-H2A.X foci number, EV: n = 97 (-), 107 (0.5 h), 84 (6 h), 93 (20 h); ciRS-7: n = 110 (-), 102 (0.5 h), 99 (6 h), 84 (20 h). Data were presented as mean ± SD from 3 replications. b. Detection of RAD51 by immunofluorescence assay in HeLa-ciRS-7 or HeLa-EV cells at different time points (0.5 h, 6 h, 20 h) after irradiation. Quantitative counts of RAD51 foci per cell or cells with more than 5 RAD51 foci were presented (n = 3). For RAD51 foci number, EV: n = 101 (-), 116 (0.5 h), 117 (6 h), 88 (20 h); ciRS-7: n = 117 (-), 150 (0.5 h), 135 (6 h), 112 (20 h). Data were presented as mean ± SD from 3 replications. c. Detection of γ-H2A.X by immunofluorescence assay in HeLa cells transfected with NC or siciRS-7 at 20 h after irradiation. Quantitative counts of γ-H2A.X foci per cell were presented (n = 50 for NC; n = 50 for siciRS-7). Data were presented as mean ± SD. d. Detection of γ-H2A.X by immunofluorescence assay in HeLa cells treated with anti-NC or anti-miR-7 at 20 h after irradiation. Quantitative counts of γ-H2A.X foci per cell were presented (n = 47 for anti-NC; n = 47 for anti-miR-7). Data were presented as mean ± SD. HeLa cells were exposed to 3 Gy X-ray. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, no significance.

Discussion

In this study, we characterized the circRNA ciRS-7 and miR-7 promoted the LLPS of miRISC. Meanwhile, we found that ciRS-7 promoted radiation-induced DNA repair.

Multivalent interactions among macromolecules are the driving force of LLPS [23,24]. LLPS has been found to be regulated by factors that change multivalent interaction, including post-translational modification, mutation, and binding partners [25–27]. Specifically, the binding of DNAs or RNAs is commonly the driving force of condensation of DNA/RNA-binding proteins [28,29]. For example, the exogenous DNA binding to cGAS protein robustly induces the formation of liquid-like cGAS condensates for which cGAS is activated [30]; lncRNA NEAT1 binds to NONO and SFPQ proteins to induce the formation of liquid-like membraneless paraspeckle [31]; lncRNA SNHG9 and its associated phosphatidic acids interact with the C-terminal domain of LATS1, promoting LATS1 phase separation [32]. Although many lncRNAs have been reported as a regulator of protein LLPS, the participation of circRNA in LLPS is largely unknown. In this study, we uncover that ciRS-7, a circRNA containing over 70 putative miRISC binding sites, enhances miRISC phase separation, which provides new insight into the role of circRNA in LLPS formation.

The core of miRISC is comprised of the miRNA, AGO2, and TNRC6B, the latter containing a large proportion of repeated and unstructured sequences (Figure 1a). Indeed, we found that compared to AGO2, TNRC6B was more prone to undergo LLPS in vivo (Figure 1d). The purified ABD domain of TNRC6B also formed spherical droplets in vitro (Figure 2b). Meanwhile, we found that the introduction of AGO2 turned the liquid-like ABD droplets into partially gel-like condensates in vitro (Figure 2d). Consistent with our results, Sheu-Gruttadauria and MacRae et al. reported the formation of large clusters of AGO2-ABD [11]. Probably, the miRISC condensates are mainly driven by TNRC6B phase separation and further stabilized by interacting with the other core protein, like AGO2.

ciRS-7 interacts with miRISC in a miRNA-dependent manner. Theoretically, the abundant miRNA binding sites of ciRS-7 may serve as a scaffold that congregates miRISC, miR-7, and ciRS-7 into highly-clustered compartmentalization, promoting the condensates of miRISC. In agreement with our hypothesis, the overexpression of ciRS-7 increased the TNRC6B and AGO2 foci in cells, whereas silencing ciRS-7 reduced the condensates (Figure 3d-g). Interrupting total miRNAs generation by siDICER1 also impaired TNRC6B droplets (Figure 4g), suggesting that other miRNAs apart from miR-7 may be involved in miRISC phase separation regulation that needs further elucidation.

As a circRNA, ciRS-7 is well-known as a miRNA sponge, which sequesters miRNAs from target mRNAs to prevent miRNA-mediated gene silencing. Recently, more and more evidence showed that ciRS-7 functioned in a miRNA-independent manner. Eva Hernando et al. reported that ciRS-7 interacted with IGF2BP3 and sequestered it from target mRNAs, suppressing IGF2BP3-mediated Melanoma invasion and metastasis [33]. Additionally, recent studies have illustrated that ciRS-7 could sense DNA damage signals and preserve p53 function as a tumor suppressor in glioma, which exerted a new insight into the link between ciRS-7 and DNA damage response [34]. Here, we proposed that ciRS-7 may also function by enhancing the condensation of miRISC. Our previous study showed that the phase separation of DNA repair factors promoted radiation-induced DNA damage repair [35,36]. Considering that AGO2 facilitates homologous recombination-mediated DSB repair [13], it is reasonable to propose that ciRS-7-induced miRISC condensates might further promote the DSB repair (Figure 5a-b).

Meanwhile, our study has some limitations, and several points need to be explored in the future work. Firstly, due to technical problem, we didn’t examine the colocalization of AGO2, TNRC6B, miR-7 and ciRS-7 in the LLPS condensates. Secondly, an in vitro LLPS assay using purified miRISC complex and synthesized ciRS-7 is absent to support the conclusion that ciRS-7 enhances the LLPS of miRISC. Additionally, as TNRC6B was a large protein that was difficult to express and purify, we only used purified recombinant ABD domain instead of full-length TNRC6B in the in vitro assay.

Author contributions

X.B.W. and X.J.F. designed and supervised this study. Y.L.W., L.L.F., J.S., W.Y.C., S.Y.B., S.M.B., G.D.Z. and R.Z.W. performed the experiments. L.L.F., J.S., and Y.L.W. wrote the manuscript. J.Z., X.B.W. and X.J.F. revised the manuscript. All authors read and approved the final manuscript.

Research involving human participants and/or animals

This article does not contain any studies with human participants performed by any of the authors.

Acknowledgments

We appreciated the technical support and generous help from Dr. Hao Nan (Northwest A&F University) and Dr. Zhi-Heng Deng (Tsinghua University) on protein purification.

Disclosure statement

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19491034.2023.2293599.

Additional information

Funding

The work was supported by the National Key R&D Program of China [2022YFA1105300, 2022YFC2503700]; The National Science Fund for Distinguished Young Scholars [82225040]; the National Science Fund for Excellent Young Scholars [82122057]; the Natural Science Foundation of China [82103770, 81972750, 82373513, 82171163, 82302307]; Guangdong Natural Science Funds for Distinguished Young Scholars [2021B1515020022, 2023B1515020090]; Guangdong Science and Technology Project [2022A1515012363]; Fundamental Research Funds for the Central Universities [22qntd3602]; and the National Postdoctoral Program for Innovative Talents of China [BX20220359]; Beijing Bethune Charitable Foundation [flzh202102].