The RNA exosome complex degrades expanded hexanucleotide repeat RNA in C9orf72 FTLD/ALS.

Nucleotide repeat expansions in the C9orf72 gene cause frontotemporal lobar degeneration (FTLD) and amyotrophic lateral sclerosis (ALS). Transcribed repeat RNA accumulates within RNA foci and is also translated into toxic dipeptide repeat proteins (DPR). The mechanism of repeat RNA accumulation, however, remains unclear. The RNA exosome complex is a multimeric ribonuclease involved in degradation of defective RNA. Here, we uncover the RNA exosome as a major degradation complex for pathogenic C9orf72-derived repeat RNA. Knockdown of EXOSC10, the catalytic subunit of the complex, enhanced repeat RNA and DPR protein expression levels. RNA degradation assays confirmed that EXOSC10 can degrade both sense and antisense repeats. Furthermore, EXOSC10 reduction increased RNA foci and repeat transcripts in patient-derived cells. Cells expressing toxic poly-GR or poly-PR proteins accumulate a subset of small nucleolar RNA precursors, which are physiological substrates of EXOSC10, as well as excessive repeat RNA, indicating that arginine-rich DPR proteins impair the intrinsic activity of EXOSC10. Collectively, arginine-rich DPR-mediated impairment of EXOSC10 and the RNA exosome complex compromises repeat RNA metabolism and may thus exacerbate C9orf72-FTLD/ALS pathologies in a vicious cycle. ©2020 The Authors. Published under the terms of the CC BY NC ND 4.0 license.

Introduction

Frontotemporal lobar degeneration (FTLD) and amyotrophic lateral sclerosis (ALS) are currently incurable neurodegenerative disorders with distinct but overlapping phenotypes. The hexanucleotide repeat expansion in C9orf72 gene is the most common known genetic cause of FTLD/ALS (DeJesus‐Hernandez et al, 2011; Renton et al, 2011; Gijselinck et al, 2012). The hexanucleotide DNA repeat is bidirectionally transcribed into GGGGCC (G4C2) or CCCCGG (C4G2) repeat RNAs. The repeat RNA accumulates and forms RNA foci that sequestrate selective repeat RNA binding protein (RBP)s and thereby physiological function of the RBPs might be impaired. The G4C2 and C4G2 RNA repeats are even translated into five distinct dipeptide repeat protein (DPR) in the absence of AUG initiation codon through repeat‐associated non‐AUG dependent (RAN) translation (Ash et al, 2013; Gendron et al, 2013; Mori et al, 2013a,c; Zu et al, 2013). DPR toxicity has been extensively studied and validated in multiple disease models (May et al, 2014; Mizielinska et al, 2014; Zhang et al, 2016, 2018b). C9orf72 protein coded in the exons of C9orf72 has been functionally linked to vesicular trafficking and autophagy (Farg et al, 2014; Sellier et al, 2016; Webster et al, 2016; Yang et al, 2016; Aoki et al, 2017). C9orf72 haploinsufficiency exacerbated repeat‐dependent toxicity suggesting loss of C9orf72 function may also contribute to neurodegeneration in C9orf72 FTLD/ALS (Shi et al, 2018; Boivin et al, 2020; Zhu et al, 2020). Since the expansion hinder efficient transcription, mature C9orf72 mRNA transcript levels are consistently decreased in C9orf72 cases (DeJesus‐Hernandez et al, 2011; van Blitterswijk et al, 2015). In clear contrast, the RNA expression levels of the repeat flanking region are much increased in C9orf72 patients compared to that of the controls (Mori et al, 2013c). These contradicting findings imply that the repeat RNA accumulates despite inefficient transcription. As an explanation for this discrepancy, we hypothesized the C9orf72 repeat RNA might resist to efficient RNA degradation in C9orf72 FTLD/ALS.

The multiprotein complex RNA exosome (Fig EV1A) is involved in RNA quality control through degradation of defective RNAs. The RNA exosome can equip two different active centers for RNA degradation at the top [EXOSC10 (Rrp6 in yeast)] and bottom [DIS3 (Rrp44 in yeast) or DIS3L] of the complex (Kilchert et al, 2016). Combinations of catalytic components of the RNA exosome complex are different depending on its subcellular localization (Tomecki et al, 2010). EXOSC10 and DIS3 contain nuclear localization signals (Graham et al, 2009) and physiologically distribute in the nucleus. Within the nucleus, EXOSC10, but not DIS3, is strongly enriched in the nucleoli (Tomecki et al, 2010). Moreover, a DIS3 homologue DIS3L is largely restricted to the cytoplasm possibly due to its lack of nuclear localization signal (Tomecki et al, 2010).

Figure EV1

Knockdown of EXOSC10 with a second siRNA increases poly‐GA expression

Data information: All graphs are shown as mean ± SEM. ***P < 0.001; two‐tailed unpaired t‐test (C). A dot in graph represents a data point from single well of cell culture plate. Ct: non‐targeting siRNA. EXO10#7: another siRNA targeting EXOSC10.

A schematic diagram of the human RNA exosome complex which is composed of 11 subunits. Two catalytic subunits are shown in light and dark orange at the top and bottom of the complex. Non‐catalytic barrel‐like core region contains central tunnel where substrate RNA postulated to go through.

Knockdown of EXOSC10 using a second siRNA (EXO10#7) significantly increased generation of poly‐GA in HeLa cells expressing (G4C2) 80 repeats in a similar way as the first siRNA against EXOSC10 (EXO10#5). 3 independent experiments. Each experiment performed in duplicates.

In situ hybridization targeting sense RNA foci on control non‐repeat expressing cells (lacking specific FISH signal (red) when compared with Fig 1E). Scale bar = 10 μm.

Figure 1

Knockdown of EXOSC10 increases repeat RNA and poly‐GA expression

Data information: All graphs except (G) are shown as mean ± SEM. (G) is shown in box‐and‐whisker plot (the central band is the median the ends of the box represent the 1st and 3rd quartile, the whiskers extend from the ends of the box to the outermost data point that falls within “3rd quartile + 1.5 × interquartile range” to “1st quartile − 1.5 × interquartile range”). *P < 0.05, ***P < 0.001; ANOVA with Dunnett's post‐test vs. Ct (=control) (C) or two‐tailed unpaired t‐test (D, F, G). A dot in graph represents a data point from single well of cell culture plate (C, D, F). Ct: non‐targeting siRNA. EXO10#5: an siRNA targeting EXOSC10.Source data are available online for this figure.

Schematic representation of a plasmid expressing (G4C2) 80 repeats together with 5′ flanking of the C9orf72 repeat (113 bp) and artificially introduced 3′ 3xTAG.

siRNA‐mediated knockdown of EXOSC10 (EXO10), but not of DIS3 and DIS3L, significantly increased generation of poly‐GA in HeLa cells expressing (G4C2) 80 repeats. 3 independent experiments. Each experiment performed in duplicates. “ctrl” indicates a control vector which lacks the G4C2 repeat but still contains the 5′ flanking region and 3xTAG.

Increased G4C2 repeat RNA expression upon EXOSC10 knockdown on RT–qPCR. 4 independent experiments. Each experiment performed in duplicates.

Knockdown of EXOSC10 increased RNA foci in HeLa cells expressing (G4C2) 80 repeats visualized by in situ hybridization targeting sense RNA foci. In total, 25 (Ct) and 24 (EXO10#5) images were analyzed in 3 independent experiments. Scale bar = 10 μm.

Knockdown of EXOSC10 increased RNA foci intensity in HeLa cells expressing (G4C2) 80 repeats. Each dot represents average cellular RNA foci intensity (Arbitrary unit) of RNA foci‐positive cells in each randomly taken image. In total, 25 (Ct) and 24 (EXO10#5) images were analyzed in 3 independent experiments.

Several components of the RNA exosome complex are already linked to neurodegenerative phenotypes. Homozygous or compound heterozygous genetic mutations in either EXOSC3, EXOSC8, or EXOSC9 has been linked to autosomal recessive developmental pontocerebellar hypoplasia, type1b (OMIM #614678) (Wan et al, 2012), type1c (OMIM #616081) (Boczonadi et al, 2014), or type1d (OMIM #618065) (Burns et al, 2018) presenting diffuse muscle weakness and progressive microcephaly. These evidences suggest proper RNA processing through the RNA exosome complex is indispensable for the development and survival of neurons. By using cellular models of C9orf72 FTLD/ALS, here we reveal that the RNA exosome complex, especially its catalytic component EXOSC10, is involved in degradation of the C9orf72 repeat RNA. This is further confirmed in RNA degradation assay and in patient‐derived cells. Expanded repeat RNA and/or DPRs have been reported to induce nucleocytoplasmic transport impairment and nucleolar stress (Freibaum et al, 2015; Zhang et al, 2015, 2018a; Mizielinska et al, 2017). We find frequent mislocalization of EXOSC10 in cells expressing arginine‐rich DPR in RAN translation‐dependent manner. Moreover, co‐immunoprecipitation assay reveals arginine‐rich DPR binds with EXOSC10. With these DPR, functional impairment of EXOSC10 is evident from the accumulation of its endogenous substrate, 3′extended small nucleolar RNA (snoRNA) precursors (Allmang et al, 1999a; Davidson et al, 2019). Strikingly, cells expressing arginine‐rich DPR show increased accumulation of co‐expressed repeat RNA. These results suggest arginine‐rich DPR causes functional impairment of the RNA exosome complex which leads to accumulation of repeat RNA and thus further deposition of RNA foci and DPR.

Results

The RNA exosome component EXOSC10 regulates C9orf72 G4C2 repeat RNA metabolism in a cellular model of C9orf72 repeat expansion

To examine potential contribution of the RNA exosome complex (Fig EV1A) on C9orf72 repeat RNA degradation, we applied previously developed repeat‐transfected cellular model that express both RNA foci and DPR from (G4C2)80 repeats (Mori et al, 2013c, 2016; Fig 1A). The reporter (G4C2)80 repeat construct has endogenous 113 bp 5′UTR region of the G4C2 repeat. This 113 bp 5′UTR region does not contain any ATG start codon but contain multiple stop codons in all three possible reading frames. Therefore, the translation of DPR only driven in a RAN translation‐dependent manner (Fig 1A). Knockdown of EXOSC10 but not of DIS3 and DIS3L significantly augmented generation of DPR (poly‐glycine‐alanine (GA)) in HeLa cells expressing (G4C2)80 repeats under the EF1 promotor (Figs 1B and C, and EV1B and C). RT–qPCR analysis confirmed a corresponding increase of G4C2 repeat RNA upon EXOSC10 knockdown (Fig 1D). Consistent with RT–qPCR analysis, EXOSC10 knockdown visualized more G4C2 repeat RNA foci (Figs 1E and F, and EV1D). Cellular RNA foci intensity was also increased upon EXOSC10 knockdown (Fig 1G). Conversely, ectopic expressions of increasing doses of siRNA‐resistant EXOSC10 rescued poly‐GA generation in a dose‐dependent manner upon knockdown of endogenous EXOSC10 (Fig 2A and B). Complementary expression of EXOSC10 also rescued repeat RNA expression (Fig 2C). Moreover, simple overexpression of EXOSC10 also suppressed repeat RNA expression in the absence of prior EXOSC10 knockdown (Fig 2D). Together, these results suggest that EXOSC10 is a significant determinant for repeat RNA metabolism and thus for DPR expression in our cellular model of C9orf72 repeat expansion.

Figure 2

Suppression of repeat RNA and poly‐GA by EXOSC10 expression

Data information: All graphs are shown as mean ± SEM. **P < 0.01, ***P < 0.001; ANOVA with Dunnett's post‐test vs. Ct (B) or with Turkey‐Kramer's post‐test (C), two‐tailed unpaired t‐test (D). A dot in graph represents a data point from single well of cell culture plate. Ct: non‐targeting siRNA. EXO10#5: an siRNA targeting EXOSC10.Source data are available online for this figure.

Dose‐dependent suppression of poly‐GA by exogenous EXOSC10 upon EXOSC10 knockdown. 3 independent experiments. Each experiment performed in duplicates.

Suppression of G4C2 repeat RNA by exogenous EXOSC10 on EXOSC10 knockdown. 5 independent experiments. Each experiment performed in duplicates. Concentration of transfected EXOSC10 plasmid was 0.2 ng/μl in 1 ml medium.

Overexpression of EXOSC10 decreases G4C2 repeat RNA. 3 independent experiments.

EXOSC10 degrades C9orf72 hexanucleotide repeat RNA in in vivo and in vitro RNA degradation assays

To confirm EXOSC10 indeed affect repeat RNA stability in cells, we performed in vivo RNA stability assay by treating repeat‐transfected HeLa cells with actinomycin D (a classic transcription inhibitor) to inhibit de novo transcription of repeat RNA. As expected, cells with EXOSC10 knockdown showed more stable expression of the G4C2 repeat RNA during experimental time course when compared to cells with non‐targeting siRNA knockdown (Fig 3A). This result suggests that EXOSC10 indeed regulates G4C2 repeat RNA metabolism in cells. Moreover, a similar effect was observed when 30 repeats of antisense CCCCGG repeat (C4G2)30 was used instead of (G4C2)80 (Fig 3B). This suggests EXOSC10 is also involved in the metabolism of antisense C4G2 repeat RNA in cells.

Figure 3

EXOSC10 degrades hexanucleotide repeat RNA in vivo and in vitro

Data information: Graph (A, B) are shown as mean ± SEM. Graph (D, F) are shown as mean ± 95% CI. # P = 0.1144, *P < 0.05, **P < 0.01, ***P < 0.001; two‐tailed paired t‐test at each time point (A, B), ANOVA with Dunnett's post‐test vs. hEXO10(−) at each time point (D) or two‐tailed unpaired t‐test at each time point (F). A dot in graph (A, B) represents a data point from single well of cell culture plate. A dot in graph (D, F) represents a data obtained from an aliquot of a reaction mixture sequentially obtained at each time point (0, 2.5, 5, 15, 30 min).

Under transcription inhibition with actinomycin D, cells with EXOSC10 knockdown (EXO10 KD) showed more stable expression of the repeat RNA compared to cells with control knockdown with non‐targeting siRNA (Ct KD). 7 and 6 independent experiments were performed.

Degradation of FAM‐labeled synthetic (G4C2)8 repeat RNA with recombinant human EXOSC10 (hEXO10) (Upper left). 3′ blocked FAM‐labeled G4C2 repeat RNA resisting 3′ exoribonuclease activity is not degraded by recombinant EXOSC10 (Lower left). Lack of recombinant EXOSC10 (Upper right) or Mg2+ ion which is essential co‐factor of EXOSC10 (Lower right) abolished the RNA degradation activity. 3 or 4 independent experiments. Quantification of the top bands is shown in (D).

Degradation of antisense C4G2 repeat RNA substrate with recombinant human EXOSC10 (Left). Without recombinant EXOSC10, no substrate degradation was observed (Right). 5 or 6 independent experiments. Quantification is shown in (F).

To examine more direct effect of EXOSC10 on G4C2 and C4G2 repeat RNA degradation, in vitro RNA degradation assay was performed. N‐terminally GST‐tagged recombinant human EXOSC10 was mixed with synthetic 5′FAM‐labeled (G4C2)8 repeat RNA or (C4G2)8 repeat RNA. Since EXOSC10 is a 3′exoribonuclease, 3′ ‐labeled (G4C2)8 repeat RNA is also tested as negative control (blocked substrate). The mixtures containing one of each substrate were then incubated in the presence of recombinant RNase inhibitor (to prevent unintentional RNA degradation through potentially contaminating RNase A, B, and C) and analyzed with TBE‐Urea gel electrophoresis (Fig 3C–F). Recombinant EXOSC10 shows time‐dependent degradation of synthetic G4C2 repeat RNA, while 3′ ‐labeled blocked G4C2 repeat RNA is not degraded at all. Lack of recombinant EXOSC10 or Mg2+ ion (essential co‐factor for EXOSC10 activity) abolished the RNA degradation activity, further validating EXOSC10‐mediated G4C2 repeat degrading activity (Fig 3C and D). Moreover, recombinant EXOSC10 also shows time‐dependent degradation of antisense C4G2 repeat RNA (Fig 3E and F). These in vivo and in vitro results provide further evidence that EXOSC10 contributes to the degradation of C9orf72 sense and antisense repeat RNA.

EXOSC10 reduction increases the RNA foci and repeat transcripts in fibroblasts from C9orf72 repeat expansion carriers

Repeat RNA expressed from plasmid‐based reporter contains artificial sequence from linker region, TAG region, and plasmid backbone. Next, we asked whether EXOSC10 degrades the endogenous sense and antisense repeat RNA in human primary cells. To do so, EXOSC10 knockdown was performed on fibroblasts from three C9orf72 patients (Mori et al, 2016; Zhou et al, 2017). The demographics of these patients were previously described (Mori et al, 2016; Zhou et al, 2017) and summarized in Appendix Fig S1A. The sense and antisense RNA foci could be visualized through in situ hybridization in C9orf72 fibroblasts (Figs 4A and EV2A) as previously described (Donnelly et al, 2013; Mori et al, 2016). Western blot analysis confirmed efficient suppression of EXOSC10 upon siRNA‐mediated knockdown (Fig EV2B and C). Upon EXOSC10 knockdown, %cells with G4C2 repeat RNA foci and number of G4C2 repeat RNA foci per cell increased (Fig 4B and C, Appendix Fig S1B). Similarly, %cells with antisense C4G2 repeat RNA foci and the number of C4G2 repeat RNA foci per cell increased with EXOSC10 knockdown (Fig 4D and E, Appendix Fig S1B). Consistent with in situ hybridization analysis, strand‐specific RT–qPCR analysis targeting 3′ flanking region of endogenous G4C2 repeat (=5′ flanking region of C4G2 repeat) revealed increased endogenous sense and antisense repeat RNA expression upon EXOSC10 knockdown (Fig 4F and G). These results suggest that EXOSC10 contributes the metabolism of disease‐associated sense and antisense repeat RNA with endogenous sequence contexts in C9orf72 patient‐derived cells.

Figure 4

Knockdown of EXOSC10 increases RNA foci and repeat RNA transcripts in cells derived from C9orf72 mutation carriers

Data information: Graph (B–E) are shown as mean ± 95% CI. Graph (F, G) are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001; two‐tailed unpaired t‐test (B–G). Single dot in graph (B–E) represents an average value from 10 different cells from a coverslip. A dot in graph (F, G) represents a data point from single well of cell culture plate. Ct: non‐targeting siRNA. EXO10#5: an siRNA targeting EXOSC10. Additional information is summarized in Appendix Fig S1.Source data are available online for this figure.

In situ hybridization. G4C2 RNA foci were detected in a C9 patient‐derived fibroblast but not in a control‐derived fibroblast with non‐targeting siRNA treatment (left panels, control/si Ct; middle panels, C9/si Ct). Knockdown of EXSOC10 increased RNA foci (right panels, C9/siEXOSC10#5). Nuclei were stained with DAPI (blue). White arrows indicate nuclear RNA foci. The green arrows show cytoplasmic RNA foci. Scale bar = 10 μm.

Quantification of %cell with G4C2 RNA foci. 3 independent knockdown experiments were performed. 10 DAPI‐positive cells per experiment were randomly selected, and z‐stack images were obtained. In total, z‐stack images from 90 cells per group (control siRNA or EXOSC10 siRNA) were counted for RNA foci.

Quantification of the number of G4C2 RNA foci/cell. 3 independent knockdown experiments were performed. 10 DAPI‐positive cells per experiment were randomly selected, and z‐stack images were obtained. In total, z‐stack images from 90 cells per group (control siRNA or EXOSC10 siRNA) were counted for RNA foci.

Quantification of %cell with C4G2 RNA foci. 3 independent knockdown experiments were performed. 10 DAPI‐positive cells per experiment were randomly selected, and z‐stack images were obtained. In total, z‐stack images from 90 cells per group (control siRNA or EXOSC10 siRNA) were counted for RNA foci.

Quantification of the number of C4G2 RNA foci/cell. 3 independent knockdown experiments were performed. 10 DAPI‐positive cells per experiment were randomly selected, and z‐stack images were obtained. In total, z‐stack images from 90 cells per group (control siRNA or EXOSC10 siRNA) were counted for RNA foci.

Increased endogenous repeat RNA expression upon EXOSC10 knockdown in strand‐specific RT–qPCR. Ten independent knockdown experiments (Case 1 N = 4, Case 2 N = 4, Case 3 N = 2) were performed.

Figure EV2

Efficient knockdown of EXOSC10 using siRNA in fibroblasts with C9orf72 repeat expansion

Data information: Graph is shown as mean. A dot in graph represents a data point from single well of cell culture plate. Ct: non‐targeting siRNA. siEXO10#5: siRNA targeting EXOSC10.

In situ hybridization. C4G2 RNA foci were detected in a patient‐derived fibroblast but not in a control‐derived fibroblast with non‐targeting siRNA treatment (left panels, control/si Ct. middle panels, C9/si Ct). Knockdown of EXOSOC10 increased RNA foci (right panels, C9/siEXOSC10#5). Nuclei were stained with DAPI (blue). White arrows indicate nuclear RNA foci. The green arrows show cytoplasmic RNA foci. Scale bar = 10 μm.

Efficient knockdown of EXOSC10 using siRNA in fibroblasts from 3 C9orf72 repeat carriers. β‐actin is shown as a loading control. 2 independent experiments.

DIS3 compensates G4C2 repeat RNA degradation in the absence of EXOSC10

EXOSC10 plays two major roles in RNA degradation. One is direct substrate degradation via its intrinsic exoribonuclease activity through the DEDDy and HRDC domains (Januszyk et al, 2011). The other is efficient guidance of substrate RNA into the central tunnel of barrel‐like structure which is formed by catalytically inert nine core subunits called Exo1‐9 and thus facilitate DIS3‐dependent substrate RNA degradation (Fig EV1A; Kilchert et al, 2016; Wasmuth & Lima, 2017).

Once passed through the barrel‐like structure, RNA substrate is expected to be degraded by DIS3 (Bonneau et al, 2009; Malet et al, 2010; Drazkowska et al, 2013; Makino et al, 2013; Wasmuth et al, 2014). While DIS3 knockdown alone did not show clear effect on poly‐GA expression (Fig 1B and C), we wondered DIS3 activity might be hidden behind the excessive activity of EXOSC10. To test the notion, we performed additive knockdown of DIS3 on top of EXOSC10 knockdown. As expected, double knockdown of EXOSC10 and DIS3 resulted in a significant increase in poly‐GA expression compared to single EXOSC10 knockdown (Fig 5A and B). This was further confirmed at RNA expression level (Fig 5C). Furthermore, overexpression of DIS3 suppressed EXOSC10 knockdown‐induced G4C2 repeat RNA accumulation (Fig 5D). These results suggest that DIS3 can compensate G4C2 repeat RNA degradation in the absence of EXOSC10, while EXOSC10 plays primary role in the RNA exosome complex dependent C9orf72 repeat RNA degradation. To further characterize EXOSC10's mechanism of action, we generated catalytically dead version of EXOSC10 (containing D313N + E315Q + D371N triple mutations on its catalytic center) (Januszyk et al, 2011). Although 3′exoribonuclease activity is lost with these mutations, catalytically dead EXOSC10 still be able to promote DIS3‐mediated RNA degradation in yeast model possibly through facilitating substrate access of the exosome complex (Wasmuth & Lima, 2012).

Figure 5

DIS3 compensates repeat RNA degradation in the absence of EXOSC10

Data information: All graphs are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001; ANOVA with Turkey–Kramer's post‐test (B, C, D, F). A dot in graph represents a data point from single well of cell culture plate. Ct: non‐targeting siRNA. EXO10#5: an siRNA targeting EXOSC10. DIS3#11: an siRNA targeting DIS3. “(−)” indicates mock transfection of backbone vector.Source data are available online for this figure.

Double knockdown of EXOSC10 and DIS3 significantly increased poly‐GA and repeat RNA expression compared with single knockdown of EXOSC10 in cells expressing (G4C2) 80 repeats. 3 independent experiments. Each experiment performed in duplicates.

Overexpression of DIS3 compensated EXOSC10 knockdown‐induced repeat RNA accumulation. 3 independent experiments. Each experiment performed in duplicates.

Catalytically dead mutant EXOSC10 rescued EXOSC10 knockdown‐induced GA increase. 3 independent experiments. Each experiment performed in duplicates.

We then compared rescue activity of wild‐type and catalytically dead EXOSC10 on (G4C2)80 repeats expressing cells under EXOSC10 knockdown. In this condition, we found the catalytically dead EXOSC10 still be able to rescue the EXOSC10 knockdown‐induced GA increase to a similar degree with wild‐type EXOSC10 (Fig 5E and F). This result provides further support for dual actions of EXOSC10 on repeat RNA degradation, i.e., RNase activity of its own (directly shown through in vitro RNA degradation assay) and promotion of substrate access through the RNA exosome complex which facilitates DIS3‐mediated RNA degradation (Wasmuth & Lima, 2012).

Arginine‐rich DPR expressed from hexanucleotide repeat expansion contributes dysregulated extranucleolar distribution of EXOSC10

The five DPR species have very distinct biophysical properties. Poly‐GA DPR is the most abundant, highly hydrophobic and aggregation‐prone DPR (Mackenzie et al, 2013; Mori et al, 2013c). Arginine (R)‐rich DPRs namely poly‐glycine‐arginine (GR) and poly‐proline‐arginine (PR) accumulate within the RNA granules and nucleoli in cellular models (Kwon et al, 2014; Wen et al, 2014; Lee et al, 2016). Since the expanded G4C2 repeat RNA and DPRs are reported to induce nucleocytoplasmic transport defect and nucleolar stress (Freibaum et al, 2015; Zhang et al, 2015, 2018a; Mizielinska et al, 2017), we next asked whether expression of the repeat RNA and DPR affects intracellular localization of EXOSC10. Cells expressing (G4C2)80 under strong CMV promoter (Fig 6A) accumulate either poly‐GA, poly‐GP, or poly‐GR or any combinations of those with poly‐GA most abundant (Mori et al, 2016). Not all cells with transfected repeat express all the three DPR (Mori et al, 2016). In this condition, expression levels of endogenous EXOSC10 were not altered (Fig EV3A–D). To analyze intracellular distribution of EXOSC10, HA‐tagged DPR and endogenous EXOSC10 were double‐immunostained. In the analysis, we monitored one of the three DPR species potentially expressed from the RAN translation reporter. This approach could enrich an effect of particular DPR on intracellular distribution of EXOSC10. Interestingly, while EXOSC10 physiologically accumulates in nucleolus (Tomecki et al, 2010), cells lacking nucleolar condensation of EXOSC10 were significantly increased among HA‐tagged poly‐GR but not HA‐tagged poly‐GA or HA‐tagged poly‐GP expressing cells (Fig 6C, D, E, G and H). To test the effect of another arginine‐rich DPR from antisense repeat strand, poly‐proline‐arginine (PR), we overexpressed 30 repeats of antisense C4G2 ((C4G2)30) (Fig 6B) in cells (Fig 6F). Again, endogenous expression levels of EXSOC10 were not changed (Fig EV3E and F). Like poly‐GR expressing cells, HA‐tagged poly‐PR expressing cells showed diffuse nuclear misdistribution of EXOSC10 (Fig 6F and I). Taken together, these results suggest arginine‐rich DPRs (poly‐GR and poly‐PR) expressed from G4C2 and C4C2 repeat contribute to the dysregulated distribution of EXOSC10.

Figure 6

Arginine‐rich DPR modulates intracellular distribution of endogenous EXOSC10

Data information: All graphs are shown as mean ± SEM. **P < 0.01, ***P < 0.001; ANOVA with Dunnett's post‐test vs. mock (H) or two‐tailed unpaired t‐test (I). A dot in graph represents a data from single well of cell culture plate.Source data are available online for this figure.

HeLa cells were transfected with one of the G4C2 repeat plasmids (A), antisense C4G2 repeat plasmid (B), or control mock plasmid. “mock” indicates a control vector which lacks the repeat but still contains the 5′ flanking region and 3xTAG.

HA‐tagged poly‐GA, poly‐GP, poly‐GR or poly‐PR (red), and EXOSC10 (green) were immunostained with anti‐HA and anti‐EXOSC10 antibodies. Nuclei were stained with DAPI (blue). Scale bar = 10 μm.

EXOSC10 (green) was immunostained with anti‐EXOSC10 antibody in cells transfected with control mock plasmid. Nuclei were stained with DAPI (blue). Scale bar = 10 μm.

Quantification of the frequency of cells with mislocalized EXOSC10. Only DPR‐positive cells were quantified (except mock‐transfected cells expressing no DPR. In this case, all cells in the picture were quantified). 3 independent experiments. (Total cell counts: (H) mock 300 cells, GA 112 cells, GP 69 cells, GR 53 cells, (I) mock 445 cells, PR 159 cells).

Figure EV3

Expression of sense and antisense repeats did not change endogenous EXOSC10 expression

Data information: All graphs are shown as mean ± SEM. ANOVA with Dunnett's post‐test vs. ctrl (B, D) or two‐tailed unpaired t‐test (F). A dot in graph represents a data point from single well of cell culture plate.

80 repeats of (G4C2) expression did not affect expression level of EXOSC10 in HeLa cells (A, B) or SH‐SY5Y cells (C, D). Four independent experiments in HeLa and 3 independent experiments in SH‐SY5Y cells. Each experiment performed in duplicates. N/T indicates no transfection control. Ctrl indicates mock transfection with corresponding backbone vector.

30 repeats of (C4G2) expression did not affect expression level of EXOSC10 in SH‐SY5Y cells. Four independent experiments. N/T indicates no transfection control. Ctrl indicates mock transfection with corresponding backbone vector.

Arginine‐rich DPR interacts with EXOSC10 and inhibits G4C2 repeat RNA metabolism in a cellular model of C9orf72 FTLD/ALS

Because multiple components of the RNA exosome complex are among the extended list of poly‐GR or poly‐PR interactome (Lopez‐Gonzalez et al, 2016; Boeynaems et al, 2017), we next asked whether endogenous EXOSC10 binds to arginine‐rich DPRs. First, nuclear and cytoplasmic fractions of cells expressing GFP only or GFP fused with codon optimized versions of DPR [each of poly‐GA89, GR89, PR89 (Fig 7A)] were obtained. Cells with these constructs express GFP‐fused DPR from conventional translation through Kozak sequence and ATG‐initiation codon. Optimal nuclear/cytoplasmic separation was validated with Western blot using antibodies against HDAC1 (nuclear marker) and GAPDH (cytoplasmic marker) (Fig 7B). In our system, GFP‐poly‐PR89 selectively accumulates within nuclear fraction, while GFP, GFP‐poly‐GA89 and poly‐GR89 are distributed throughout nucleus and cytoplasm (Fig 7C and E). The nuclear or cytoplasmic fractions were served for immunoprecipitation using GFP‐trap beads. Western blot analysis revealed that in cytoplasmic fraction where GFP‐poly‐PR89 is absent, EXOSC10 accumulated only in GFP‐poly‐GR fraction (Fig 7C and D). In nuclear fraction, EXOSC10 are selectively detected in GFP‐poly‐GR89 and GFP‐poly‐PR89 precipitants but not in control GFP only or GFP‐poly‐GA89 precipitants (Fig 7E and F). Expression levels of EXOSC10 were not altered by the presence of GFP‐poly‐GR89, GFP‐poly‐PR89, and GFP‐poly‐GA89 (Fig EV4A and B). These results suggest preferential binding of EXOSC10 with arginine‐rich DPRs, and such interaction may contribute functional impairment of the RNA exosome complex.

Figure 7

Arginine‐rich DPRs have preferential interaction with endogenous EXOSC10

Data information: All graphs are shown as mean ± SEM. *P < 0.05, ***P < 0.001; ANOVA with Dunnett's post‐test vs. GFP (D, F). A dot in graph represents a data point from single well of cell culture plate.Source data are available online for this figure.

Schematic diagrams of codon optimized (lacking extensive G4C2 repeat) plasmids expressing N‐terminally EGFP fused 89 repeats of either poly‐GR, poly‐PR, or poly‐GA under the control of CMV promoter.

Immunoblot validation of successful fractionation using antibodies against HDAC1 (nuclear marker) and GAPDH (cytoplasmic marker).

Co‐immunoprecipitation of GFP‐DPR interacting proteins in cytoplasmic fraction using GFP‐trap beads. Immunoprecipitants were served for immunoblot with anti‐EXOSC10 antibody. The same membrane was reprobed with anti‐GFP antibody.

Quantification of the relative intensities of “EXOSC10 in co‐IP fraction”/“EXOSC10 in input”. 3 independent experiments.

Co‐immunoprecipitation of GFP‐DPR interacting proteins in nuclear fraction using GFP‐trap beads followed by immunoblot using anti‐EXOSC10 antibody. The same membrane was reprobed with anti‐GFP antibody.

Quantification of the relative intensities of “EXOSC10 in co‐IP fraction”/“EXOSC10 in input”. 3 independent experiments.

Figure EV4

Western blot and filter trap analysis for EXOSC10 and GFP‐DPR expressions

GFP‐DPR expression did not affect endogenous expression of EXOSC10 in HeLa cells. EXOSC10 expression was analyzed by Western blotting. GFP and GFP‐DPRs were analyzed by filter trap. Identical samples were loaded on (A) and (B).

Cells co‐transfected EF1‐based (G4C2)80 repeat together with either GFP, GFP‐GR177, or GFP‐PR166 at 1:2 ratio were analyzed by filter trap and Western blotting. Identical samples were loaded on (C) and (D–G). With this EF1‐based (G4C2)80 repeat construct, RAN translation‐dependent poly‐GP or poly‐GR expressions were below the detection limit as previously described.

To examine whether arginine‐rich DPRs indeed alter intracellular RNA metabolism through EXOSC10, we analyzed endogenous substrates of EXOSC10: short 3′extended snoRA48 and snoRA68 precursors (Davidson et al, 2019). To do so, qPCR primers targeting 3′extended region of snoRNAs were designed. First, we confirmed that EXOSC10 knockdown increased the expression of endogenous 3′extended snoRA48 and snoRA68 precursors (Fig 8B and D). This verified these snoRNA precursors are indeed substrates of EXOSC10. Importantly, GFP‐poly‐GR177 or GFP‐poly‐PR166 expression augmented expressions of these snoRNA precursors (Fig 8A, C and E). These results suggest arginine‐rich DPR causes functional impairment of EXOSC10. To further validate our model that GR/PR inhibits EXOSC10‐dependent repeat RNA degradation, we co‐expressed EF1 promoter‐based (G4C2) 80 repeats and each of GFP‐fused arginine‐rich DPRs (Figs 1A and 8A, and EV4C–G). Strikingly, the expression of GFP‐poly‐GR177 or GFP‐poly‐PR166 induced augmented accumulation of G4C2 repeat RNA (Fig 8F). Together, these results indicate that arginine‐rich DPR impedes C9orf72 repeat RNA metabolism through the inhibition of EXOSC10/the RNA exosome complex.

Figure 8

Arginine‐rich DPR causes functional impairment of EXOSC10/the RNA exosome complex

Data information: All graphs are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001; two‐tailed unpaired t‐test (B, D) or ANOVA with Dunnett's post‐test vs. GFP (C, E, F). A dot in graph represents a data point from single well of cell culture plate.

Schematic diagrams of codon optimized (lacking extensive G4C2 repeat) plasmids expressing N‐terminally EGFP fused 177 repeats of poly‐GR or 166 repeats of poly‐PR under the control of CMV promoter.

Increased expression of 3′extended snoRA48 precursor upon EXOSC10 knockdown (B) or GFP‐poly-GR177/GFP-poly‐PR166 expression (C) on RT–qPCR. 6 independent experiments. Each experiment performed in duplicates.

Increased expression of 3′extended snoRA68 precursor upon EXOSC10 knockdown (D) or GFP‐poly-GR177/GFP‐poly-PR166 expressions (E) on RT–qPCR. 6 independent experiments. Each experiment performed in duplicates.

RT–qPCR analysis showed increased cellular G4C2 repeat RNA expression from co‐transfected EF1 (G4C2)80 repeat construct (Fig 1A) in the presence of GFP, GFP-poly‐GR177, or GFP‐poly-PR166. 3 independent experiments. Each experiment performed in duplicates.

Discussion

Here, we have shown the RNA exosome complex confers degradation of expanded G4C2 repeat RNA in C9orf72 FTLD/ALS. The repeat RNA could be toxic by itself through sequestration of specific RNA binding proteins (Donnelly et al, 2013; Lee et al, 2013; Mori et al, 2013b; Sareen et al, 2013; Xu et al, 2013; Cooper‐Knock et al, 2014). Especially, association between antisense RNA foci and TDP‐43 pathology has been reported (Cooper‐Knock et al, 2015; Aladesuyi Arogundade et al, 2019). At the same time, the repeat RNA is a source of highly toxic DPRs (Mori et al, 2013c; May et al, 2014; Mizielinska et al, 2014; Zhang et al, 2016, 2018b). Therefore, reduction of the repeat RNA could potentially be therapeutic option in C9orf72 FTLD/ALS. In our condition, EXOSC10 is most relevant to the repeat RNA metabolism among the three catalytic subunits of the RNA exosome complex [i.e., EXOSC10, DIS3, and DIS3L (Allmang et al, 1999b; Mitchell et al, 1997; Tomecki et al, 2010)]. Reduced EXOSC10 increases the levels of the repeat RNA which leads to elevated DPR production and accumulation. Not surprisingly, these effects were canceled by exogenous expression of EXOSC10. In vivo RNA stability assay using transcription inhibitor actinomycin D confirmed EXOSC10 knockdown compromised both sense and antisense repeat RNA degradation. Moreover, recombinant EXOSC10 efficiently degrades synthetic sense and antisense repeat RNA. These results strongly support EXOSC10‐mediated degradation of C9orf72 repeat RNA. Crucially, in fibroblast derived from patients with C9orf72 repeat expansion, we confirmed an accumulation of both sense and antisense RNA foci and repeat RNA transcripts upon EXOSC10 knockdown. This implicates in vivo relevance of EXOSC10 in C9orf72 FTLD/ALS context. In vivo and in vitro RNA degradation assays differ in the enzymatic composition. In in vivo assay, EXOSC10 is expected to work as a component of the RNA exosome complex, while in vitro assay we used recombinant EXOSC10 alone as a catalytic machinery. The antisense repeat RNA may be slightly more resistant to EXOSC10‐mediated degradation in vivo (Figs 3A and B, and 4F and G) and in vitro (Fig 3C–F). Reported structural difference between sense and antisense repeat RNA might underlie the observed difference (Fratta et al, 2012; Reddy et al, 2013; Haeusler et al, 2014; Dodd et al, 2016).

Although less prominent, DIS3 is also significantly involved in the degradation of repeat RNA. An additive knockdown of DIS3 on EXOSC10 knockdown boosted repeat RNA and DPR expression. Moreover, DIS3 overexpression compensated EXOSC10 knockdown‐induced repeat RNA accumulation. These results argue that the RNA exosome complex as a whole works as a G4C2 and C4G2 repeat RNA degradation machinery. EXOSC10 has the two proposed functions on RNA degradation. One is RNase activity of its own, and another is promotion of substrate loading into tunnel of the barrel‐like core of the RNA exosome complex. Mechanistically, we found even the catalytically dead version of EXOSC10 can rescue EXOSC10 knockdown‐induced GA increase, possibly through facilitating RNA degradation by DIS3. This suggests EXOSC10's promotion of substrate loading plays significant role in repeat RNA degradation. However, EXOSC10's own 3′exoribonuclease catalytic activity should not be ignored. One reason is because single knockdown of DIS3 did not affect GA expression (Fig 1C). Since DIS3 knockdown is expected to cancel the consequence of EXOSC10's substrate loading effect, catalytic activity of EXOSC10 seems to have ability to compensate it. Another argument is that we did confirm repeat RNA degradation with purified EXOSC10 via its 3′exoribonuclease activity in vitro. These findings suggest both EXOSC10's inherent RNase activity and substrate loading activity works in corporative manner in the context of repeat RNA degradation.

We found intranuclear localization of EXOSC10 is misregulated in cells expressing arginine‐rich DPRs in a RAN translation‐dependent manner. Because cells with the RAN translation reporter construct express repeat RNA potentially together with single to multiple DPR species in a various combination, we could not exclude the possibility that observed mislocalization in arginine‐rich DPR expressing cells is due to compound effects of arginine‐rich DPR and co‐expressed repeat RNA/the other DPRs. Still, frequency of cells with EXOSC10 mislocalization was significantly increased in cells selected with arginine‐rich DPR positivity compared with that of control mock‐transfected cells, but not in cells selected with poly‐GA or GP positivity (Fig 6). This finding implicates the potential importance of arginine‐rich DPR in EXOSC10 mislocalization. Since C9orf72 repeat RNA and DPRs are known to induce nucleocytoplasmic transport defect and nucleolar stress, mislocalization of EXOSC10 could be the downstream effect of these cellular dysregulation. Mislocalization of EXOSC10 may cause its functional impairment; however, at the moment we have no direct evidence on this point. In accord with recent reports of DPR interactome (Lopez‐Gonzalez et al, 2016; Boeynaems et al, 2017), we confirmed that arginine‐rich DPRs have preferential interaction with endogenous EXOSC10. The EXOSC10‐DPR interactions may cause functional impairment of the RNA exosome complex. Indeed, 3′extended snoRA48 and snoRA68 precursors, previously reported as endogenous substrates of EXOSC10, accumulated in poly‐GR and poly‐PR expressing cells. Moreover, the expression of G4C2 repeat RNA is significantly increased when co‐expressed with poly‐GR or poly‐PR. Thus, it is tempting to speculate that DPR‐mediated functional impairment of the RNA exosome complex leads to impaired nuclear RNA quality surveillance and thus accumulation of unusual repeat RNA (Fig 9). Interestingly, a recent report revealed repetitive element transcripts are elevated in the brain of C9orf72 ALS/FTD patients (Prudencio et al, 2017). Since the repeat RNA transcripts escaped from nuclear RNA quality surveillance can accumulate as RNA foci or even exported to cytoplasm where it would be translated into DPR, DPR‐mediated functional impairment of the RNA exosome complex may lead to inefficient metabolism of the repeat RNA and further accumulation of DPR. Consequently, this pathological downward spiral may exacerbate neurodegeneration in C9orf72 FTLD/ALS.

Figure 9

Conceptual summary

Impairment of EXOSC10 with arginine‐rich DPR compromises repeat RNA degradation through the RNA exosome complex. This accelerates repeat RNA accumulation (as RNA foci) and RAN translation‐dependent DPR production. The produced DPR in turn compromises EXOSC10/the RNA exosome complex and thus accelerates the pathological process of C9orf72 FTLD/ALS.

Materials and Methods

Cell culture

HeLa cells were cultured in DMEM containing 10% FCS and penicillin/streptomycin (P/S). SH‐SY‐5Y cells were cultured in DMEM/F‐12 containing 10% FCS and P/S.

Patient‐derived fibroblasts

We used fibroblast cell lines from three different C9orf72 ALS cases. These cells were originally obtained based on written informed consent in accordance with the Helsinki convention as previously described (Mori et al, 2016; Zhou et al, 2017). Cells were grown in DMEM supplemented with 20% FCS and P/S. The presence of the expanded C9orf72 repeat was confirmed by repeat prime PCR (Akimoto et al, 2014).

Plasmids

The EF1 promoter‐driven and CMV promoter‐driven (G4C2) 80 repeats vectors are described previously (Mori et al, 2013c, 2016). CMV promoter‐driven (C4G2) 30 repeats vector is described previously (Nihei et al, 2020). EXOSC10 cording sequence was subcloned from pT7‐V5‐SBP‐C1‐HsRRP6 (Addgene plasmid # 64916, a gift from Prof. Elisa Izaurralde) into HindIII/BamHI site of the pcDNA3.1 hygro (+) vector (Invitrogen). To generate a siRNA‐resistant EXOSC10, several silent mutations in siRNA target sequence (J‐010904‐05 Human EXOSC10 ACGAAAAGCUCUUGAAUUG) were introduced by site‐directed mutagenesis inverse PCR using primers CTGAACTGCCAGGAATTTGCAGTTGACTTG and CAGTTTTTCATTGAGTTCCACGAGTTCATC. To generate a catalytically dead version of EXOSC10 (triple point mutations; D313N + D371N + E315Q), we performed sequential site‐directed mutagenesis. Point mutation was introduced using the following primers: TGCCAGGAATTTGCAGTTAACTTGGAGCACCACTCT and AGAGTGGTGCTCCAAGTTAACTGCAAATTCCTGGCA for D313N, GGTCTTTCATGGTGCTGATTCAAACATAGAATGGCTACAG and CTGTAGCCATTCTATGTTTGAATCAGCACCATGAAAGACC for D371N, GAATTTGCAGTTAACTTGCAGCACCACTCTTACAGGAGC and GCTCCTGTAAGAGTGGTGCTGCAAGTTAACTGCAAATTC for E315Q. Human DIS3 was cloned into pcDNA3.1hygro(+) vector (Invitrogen) at BamHI/NotI sites. Cording sequence was obtained through nested PCR using cDNA from HeLa cells and following primer sets. (1st PCR: CTCCGGGGTTAGGCGTATTC and CCCTGAAGTTGCTGTCCTGT, 2nd PCR: CGCGGATCCGCCACCATGCTCAAGTCCAAGACGTTC and CGCGCGGCCGCCTATTTTCCAAGCTTCATCTTC). Sequence was verified with Sanger sequence.

pEGFP‐GR89, PR89, GA89, GR177, and PR166 vectors were constructed as previously described (Bennion Callister et al, 2016) with slight modifications (Nihei et al, 2020). DNA oligo coding 12 repeats of each DPR, but avoiding GGGGCC or CCCCGG repeat sequence, were designed using alternative codons. Restriction sites for XhoI‐FokI and BbsI‐XbaI were placed at the beginning and end of the oligo DNA. The codon optimized 11 repeat cording DPR sequences were isolated through digestion with FokI and BbsI and then ligated into BbsI‐linearized parental 12 repeat DPR cording vector to double the repeat length. This process was repeated until reached 177 repeat lengths were achieved (12 + 11 = 23 repeats, 23 + 22 = 45 repeats, 45 + 44 = 89 repeats, 89 + 88 = 177 repeats). We failed to obtain PR177 due to instability of repetitive sequence during cloning. DNA sequence was verified by Sanger sequence.

XhoI‐FokI‐GA12‐BbsI‐XbaI: 5′‐CTCGAGGGATGTTGAATTCTGGTGCTGGCGCGGGAGCAGGCGCTGGTGCTGGTGCAGGAGCGGGTGCGGGAGCTGGTGCCGGCGCAGGTGCTGTCTTCGGATCCTAGTCTAGA‐3′

XhoI‐FokI‐GR12‐BbsI‐XbaI: 5′‐CTCGAGGGATGTTGAATTCTGGTCGTGGACGTGGACGAGGTCGAGGTCGAGGTCGTGGACGTGGTCGAGGTCGAGGTCGTGGACGTGGTCGTGTCTTCGGATCCTAGTCTAGA‐3′

XhoI‐FokI‐PR12‐BbsI‐XbaI: 5′‐CTCGAGGGATGTTGAATTCTCCGCGACCTCGACCGCGGCCACGCCCACGCCCTCGGCCCAGACCACGTCCTAGGCCCAGACCCAGACCGCGAGTCTTCGGATCCTAGTCTAGA‐3′

siRNA‐mediated knockdown and plasmid transfection

The following siRNAs were obtained from Dharmacon: ON‐ TARGETplus Human non‐targeting siRNA D‐001810‐01, J‐010904‐05 Human EXOSC10 ACGAAAAGCUCUUGAAUUG, J‐010904‐07 Human EXOSC10 GAAGUGACAUGUACAUUCU, J‐015405‐11 Human Dis3 UGAUGAAGAUCGUGCGCGA, J‐015405‐12 Human Dis3 AGGUAGAGUUGUAGGAAUA, J‐015333‐05 Human Dis3L GAACAAGGGCCACCACUUA, J‐015333‐06 Human Dis3L UGACGGAGUUAUUUAUUCA. Five or ten picomoles of each siRNA was reverse‐transfected using RNAiMax (Thermo Fisher Scientific) and OPTI‐MEM when 24‐well plate format was used. After overnight incubation, media were exchanged and 1 ng/μl (500 μl when 24‐well plate format is used) of plasmids was transfected with lipofectamine LTX with plus reagent (Thermo Fisher Scientific) in OPTI‐MEM into HeLa cells unless otherwise stated. On SH‐SY5Y cells, lipofectamine 3000 with P3000 reagent (Thermo Fisher Scientific) in OPTI‐MEM was used alternatively. For rescue experiments, 1 ng/μl of repeat expressing plasmids and 0.006, 0.02, 0.06, and 0.2 ng/μl for dose–response experiment, and 0.2 ng/μl for single dose experiment of rescue constructs were co‐transfected. Media were exchanged after 4‐ to 6‐h incubation. Cells were harvested 2 days after plasmid transfection for Western blot unless otherwise indicated.

Antibodies

The following antibodies were used for Western blots (WB), immunofluorescence (IF), and Filter trap (FT): anti‐DYKDDDDK(FLAG) Tag (Cell Signaling #2368S) WB 1/1,000, anti‐myc clone 9E10 (Santa Cruz) WB 1/1,000, IF 1/100, anti‐HA Tag clone 3F10 (Roche) WB 1/1,000, IF 1/100, anti‐GFP(B‐2) (Santa Cruz) WB 1/500, FT 1/500, anti‐beta‐actin (Sigma‐Aldrich) WB 1/1,000, FT 1/1,000, anti‐EXOSC10 (ATLAS ANTIBODIES) WB 1/500, IF 1/2,000, anti‐DIS3 (Proteintech) WB1/2,000, anti‐DIS3L (Santa Cruz) WB 1/500, anti‐GAPDH (Proteintech) WB 1/20,000, anti‐HDAC1 (Proteintech) 1/1,000, anti‐PR (Proteintech) WB 1/500, FT 1/500, anti‐GR (Proteintech) WB 1/1,000, FT 1/1,000.

Reverse transcription and quantitative polymerase chain reaction (RT–qPCR)

HeLa cells expressing (G4C2)80 or (C4G2)30 repeats were harvested 1–2 days after plasmid transfection. Total RNA was prepared using the RNeasy and Qiashredder kit (Qiagen). RNA preparations were treated with Turbo DNA‐free kit (Thermo Fisher Scientific) to minimize residual DNA contamination. Two micrograms of RNA was used for reverse transcription with M‐MLV Reverse Transcriptase (Promega) using oligo‐(dT) 12–18 primer (Invitrogen). RT–qPCR was performed using the ViiA7 Real‐Time PCR System (Applied Biosystems) with TaqMan technology. Primers and probes were designed (IDT) for 3′ TAG region of repeat constructs (repeat TAG primer for sense repeat). Primer 1: TCT CAA ACT GGG ATG CGT AC, Primer 2: GTA GTC AAG CGT AGT CTG GG, Probe/56‐FAM/TG CAG ATA T/Zen/C CAG CAC AGT GGC G/3IABkFQ/. For antisense repeat primer and probe set (C4G2 antisense tag) were used. Primer 1: TCT CAA ACT GGG ATG CGT, Primer 2 GTC CTT GTA GTC AAG CGT AGT C, Probe/56‐FAM/TC GTA TGG G/Zen/T ACC CGG GTT TGC/3IABkFQ/. A primer/probe set for Human ACTB, Hs.PT.39a.22214847 (IDT) was used as endogenous control. Each sample was paired with no reverse transcription controls showing < 1/210 (ΔCT > 10) signal when compared to reverse transcribed samples, thus excluding contamination of plasmid DNA‐derived signal. Each biological sample (=a sample from a well of culture plate) was analyzed in duplicate or triplicate. Signals of repeat RNA‐derived cDNA were normalized to ACTB cDNA according to the ΔΔC T method. Alternatively, patient‐derived fibroblasts were harvested 3 days after siRNA‐mediated knockdown. For the strand‐specific quantification of endogenous repeat RNA in fibroblast, total RNA was collected as described above. 3′ proximity region of endogenous G4C2 repeat was targeted for RT–qPCR as previously described (Mori et al, 2013c). SuperScript IV Reverse Transcriptase (Thermo Fisher Scientific) and strand‐specific primers (either sense RT primer CAATTCCACCAGTCGCTAGA or antisense RT primer CTGCGGTTGCGGTGCCTGC) were used for strand‐specific reverse transcription. The following set was used for qPCR; Primer 1: AAG AGG CGC GGG TAG AA, Primer 2: CAG CTT CGG TCA GAG AAA TGA G, Probe: /56‐FAM/CT CTC CTC A/Zen/G AGC TCG ACG CAT TT/3IABkFQ/ (Mori et al, 2013c). Obtained signals were normalized according to the ΔCT method. PCR product was validated by Sanger sequencing.

RT–qPCR for 3′ext‐snoRA48, 68

HeLa cells were harvested 2 days after siRNA‐mediated knockdown or 1 days after plasmid transfection (GFP, GFP‐GR177, GFP‐PR166). Total RNA was extracted and analyzed as described above except random hexamer (Invitrogen) was used for reverse transcription. The following set was designed and used for 3′ext‐snoRA48, Primer 1: CAA AGC AAC TCC TTG TTC ATC C, Primer 2: GGC TAC AAT ACC ACC TCT TTA AAG, Probe/56‐FAM/TG GTT CAT G/Zen/C CTT GGA CAC ATA GGT/3IABkFQ/; and for 3′ext‐snoRA68, Primer 1: GCA AAC AGC AAG CGG ATC TT, Primer 2 CCC TGC AGT CAC TGG CC, Probe/56‐FAM/CC CTC AAA G/Zen/T GAA TTT GGA GGT TCC ACA/3IABkFQ/. PCR product was validated by Sanger sequencing.

In vivo RNA stability assay using Actinomycin D

HeLa cells were cultured on 12‐well culture plates and reverse‐transfected with indicated siRNAs (non‐targeting control or targeting EXOSC10) (day 0). The next day cells were transfected with repeat plasmid for 6 h (day 1). On day 2, 10 μg/ml actinomycin D pretreatment was started at time point −1 h to abolish transcription. In the presence of actinomycin D, cells were then sequentially collected at indicated time points (0, 0.5, 1, 2, and 3 h) and frozen immediately. Since actinomycin D treatment may drastically affect total amount/composition of cellular RNA, equal amount of in vitro transcribed EGFP RNA using RiboMAX Large Scale RNA Production System‐T7 (Promega) was added each well just prior to RNA purification as a normalization control. Total RNA was extracted and analyzed as described in RT–qPCR section. The following set was used for qPCR targeting EGFP, Primer 1: GCA CAA GCT GGA GTA CAA CTA, Primer 2: TGT TGT GGC GGA TCT TGA A, Probe: /56‐FAM/AG CAG AAG A/Zen/A CGG CAT CAA GGT GA/3IABkFQ/ qPCR signals of repeat RNA‐derived cDNA were normalized for time point 0 h, and normalized expressions of repeat RNA/EGFP are shown according to the ΔΔC T method.

In situ hybridization

In situ hybridization was performed as previously described (Mori et al, 2016) with slight modifications. 4% paraformaldehyde‐fixed and 0.2% Triton X‐100 perforated cells on glass coverslips were rinsed 2× SSC and then incubated in prehybridization solution (40% formamide (Life Technologies, 15515‐026), 2× SSC, 2.5% BSA) at 57°C for 30 min. Cells were then incubated with hybridization solution (40% formamide, 2× SSC, 0.8 mg/ml tRNA (Roche), 0.8 mg/ml single strand salmon sperm DNA (Sigma‐Aldrich, D7656), 0.16% BSA, 8% Dextran sulfate (Sigma‐Aldrich), 1.6 mM Ribonucleoside‐vanadyl complex (New England Biolabs, S1402S), 5 mM EDTA, 100 ng (for HeLa), or 10 ng (for Fibroblasts)/ml 5′ Cy3‐labeled (CCCCGG) × 4 probe or 5′ Cy3‐labeled (GGGGCC) × 4 probe (IDT; DeJesus‐Hernandez et al, 2011) at 57°C (for sense foci) or at 37°C (for antisense foci) overnight. The following day, cells were sequentially washed in 40% formamide, 0.5× SSC for three times 10 min each at 57°C and then with 0.5× SSC three times 10 min each at room temperature. After a brief rinse with PBS, cells were washed 0.05% Tween20 in 1× PBS for three times 5 min each. Nuclei were counterstained with 0.1 μg/ml of DAPI for 20 min and then washed with 0.05% Tween20 in 1× PBS for three times 5 min each. Glass coverslips were mounted using Prolong Diamond antifade (Life Technologies) and analyzed with Leica DMI400B confocal microscopy with LAS AF software (Leica). For quantification of RNA foci on fibroblast, cells were analyzed with LSM710 (Zeiss) microscope with a 63×/1.20 W Korr M27 oil immersion objective. Z‐Stacked images of 1 μm thickness were sequentially taken to cover the whole nucleus at 4.0 times digital zoom. Number of RNA foci within DAPI‐positive area was manually counted at 3D presentation at Maximum projection mode using Zen2011 software (Zeiss).

%cells with RNA foci in HeLa cell were manually counted from RNA foci signals and DAPI signals of single plane images. Quantification of RNA foci intensity in HeLa cells was performed using ImageJ software. Binary images from DAPI counterstaining were used for the nuclear regions. RNA foci intensity of a cell was defined as [“average Cy3 signal in DAPI‐positive area of RNA foci‐positive cell” subtracted with “average Cy3 signal in DAPI‐positive area of RNA foci‐negative cell”] times “DAPI‐positive area of a RNA foci‐positive cell”. RNA foci intensities were quantified from three independent experiments.

co‐IP experiment

HeLa cells expressing GFP or GFP‐poly‐DPR (GA89, GR89, PR89) were harvested 1 days after plasmid transfection. Cells were lysed and separated into nuclear and cytoplasmic fractions using NE‐PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific).

These fractions were incubated with GFP‐TRAP A beads (Chromo Tek) overnight at 4°C and subsequently precipitated by centrifugation at 2,500 g for 2 min at 4°C. Beads were washed in ice‐cold dilution buffer (10 mM Tris–HCl pH 7.5, 150 mM NaCl, 0.5 mM EDTA) three times (2,500 g for 2 min at 4°C) and resuspended in 2× Laemmli buffer. The protein samples were heated at 100°C for 10 min and served for SDS–PAGE.

RNA degradation assay using recombinant EXOSC10

RNA degradation assay was performed as previously described (Januszyk et al, 2011) with slight modifications. 5′ 6‐FAM‐labeled RNA oligonucleotides were synthesized and HPLC‐purified (FASMAC). 48‐nt (G4C2)8, 3′‐blocked 48‐nt (G4C2)8, and (C4G2)8 oligonucleotides have the following sequences: FAM‐GGGGCCGGGGCCGGGGCCGGGGCCGGGGCCGGGGCCGGGGCCGGGGCC, FAM‐GGGGCCGGGGCCGGGGCCGGGGCCGGGGCCGGGGCCGGGGCCGGGGCC‐ , FAM‐CCCCGGCCCCGGCCCCGGCCCCGGCCCCGGCCCCGGCCCCGGCCCCGG. Exoribonuclease activities were assessed in a 100 μl reaction mixture of 10 mM Tris–HCl (pH 8.0), 10 mM DTT, 5 mM MgCl2, 1 U/ml RNase inhibitor (NEB), 3 nM RNA, and 10 nM GST‐fused full‐length human EXOSC10 protein (Abcam) at 37°C. Time‐course assays were conducted by removing 10 μl aliquots at each time point (0, 2.5, 5, 15, 30 min). Reactions were quenched by the addition of 10 μl of loading buffer (95% formamide, 20 mM EDTA, 1% 10× loading buffer (TaKaRa)) and heated at 80°C for 30 s. Samples (20 μl each in volume) were loaded onto a 15% polyacrylamide–7 M urea gel for electrophoresis. Gels were imaged using a ChemiDoc Touch MP (Bio‐Rad).

Statistics

Statistical analysis was performed using JMP Pro 13.2.0 software.

Author contributions

KM conceived the project, supervised the study, performed experiments, and wrote manuscript with input from YK. YK and TY performed most experiments with the help of SG. YK analyzed the data, performed statistical analysis, and prepared all figures. KM and MI coordinated the study. All authors read and approved the final manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Appendix

Click here for additional data file.

Expanded View Figures PDF

Click here for additional data file.

Review Process File

Click here for additional data file.

Source Data for Figure 1

Click here for additional data file.

Source Data for Figure 2

Click here for additional data file.

Source Data for Figure 4

Click here for additional data file.

Source Data for Figure 5

Click here for additional data file.

Source Data for Figure 6

Click here for additional data file.

Source Data for Figure 7

Click here for additional data file.