DHX8 regulates degradation of RNA by RNautophagy

Abstract

RNautophagy is an intracellular degradation pathway in which RNA is directly taken up by lysosomes. The cytoplasmic regions of the lysosomal membrane proteins, LAMP2C and SIDT2, can interact with consecutive guanine sequences in RNA, mediating the uptake of RNA during RNautophagy. RNautophagy has also been implicated in the clearance of expanded CAG-repeat mRNA and RNA foci associated with polyQ disease. However, the mechanisms of RNA uptake during RNautophagy remain unclear. Here, we screened for proteins that bind consecutive guanine sequences and identified RNA helicase DHX8 as a binding partner. DHX8 interacts with SIDT2 and is partially localized to the cytoplasmic side of the lysosomal membrane. We found that DHX8 regulates intracellular RNA degradation via SIDT2-dependent RNautophagy but not via macroautophagy. RNA binding, but not ATPase activity, of DHX8 is likely to be important for regulating RNA degradation. DHX8 also contributes to the clearance of pathogenic CAG repeat mRNA and RNA foci, and the levels of both soluble protein and insoluble high-molecular-weight aggregates of expanded polyQ tracts. Our findings provide insights into the mechanisms underlying the regulation of intracellular RNA degradation, autophagic pathways, and possibly the pathogenesis of repeat RNA-related disorders.

Graphical Abstract

Figure 1.

Screening of RNA helicases that are involved in RNA degradation via lysosome. (A) Oligonucleotide DNA sequences used for pull-down assays. (BandC) Interaction of the endogenous proteins with poly-dA (15-mer) or poly-dG (15-mer) in mouse brain lysates. Pull-down assays were performed using oligo DNA (1 nmol). Interacting SIDT2 was detected by immunoblotting (B). Silver stain analysis of the interacting proteins (C). (D) Interaction proteins were identified using LC/MS-MS analysis. Scatterplot of the proteins for which there was a significant difference in binding to poly-dG and poly-dA in both of two independent experiments. Data represent the log2 of the abundance ratio of proteins interacting with poly-dG to that with poly-dA. RNA helicases which were reported to be present on lysosomes or not were indicated red or purple, respectively. (E) Schematic of the experimental method for measuring intracellular degradation of endogenous RNA in MEFs transfected with siRNAs. Cells were labeled with [³H] uridine, grown in medium with unlabeled uridine, and chased for 24 h. Radioactivity was expressed as a percentage of degraded RNA. (F) RNA degradation in MEFs transfected with indicated siRNAs. Data are mean ± SEM (n = 4). (G–I) Levels of DHX8 and EIF4A1 proteins in MEFs transfected with indicated siRNA were analyzed using immunoblotting. Data are mean ± SEM (n = 4). (J–M) RNA degradation in MEFs transfected with indicated siRNAs and treated with or without CQ (50 μM). Data are mean ± SEM. Numbers within the bars indicate n. P-values are from Dunnett’s multiple comparisons test (F, H, and I) or Tukey’s multiple comparisons test (J–M); CQ, chloroquine.

Figure 2.

Pathway involved in DHX8-mediated RNA degradation is SIDT2-dependent pathway, not macroautophagy. (A–H) RNA degradation in Atg5 KO (A, B, E, and F) or Sidt2 KO (C, D, G, and H) MEFs transfected with indicated siRNAs and treated with or without CQ (50 μM). Data are mean ± SEM. Numbers within the bars indicate n. P-values are from Tukey’s multiple comparisons test (A–H).

Figure 3.

DHX8 mediates RNautophagy and cooperates with SIDT2 independent of enzymatic activity of DHX8. (A–C) Schematic of in vitro RNA uptake assay using isolated lysosomes (A). MEFs were transfected with control or Dhx8 siRNA. After 72 h, lysosomes were isolated by density gradient ultracentrifugation. Lysosomes were incubated with 5 μg total RNA (purified from mouse brains) for 3 min at 37°C in the presence of ATP, and then lysosomes were precipitated by centrifugation, and RNA levels in the solution outside the lysosomes were measured (B). RNA uptake levels were calculated by subtracting the RNA levels in the solution outside the lysosomes from the RNA input levels (C). Data are mean ± SEM (n = 3). (D) Schematic of the experimental method for measuring intracellular degradation of endogenous RNA in N2a cells overexpressing SIDT2 and/or DHX8. (E) Overexpression of SIDT2 and/or DHX8 (WT or R620A) in N2a cells was confirmed using immunoblotting. (F and G) RNA degradation in N2a cells overexpressing indicated proteins. Data are mean ± SEM. Numbers within the bars indicate n. (Hand I) N2a cells were overexpressed with DHX8–HB or DHX8–R620A–HB and were UV crosslinked. The DHX8 and covalently crosslinked DHX8–RNA complex were separated by SDS–PAGE and detected using immunoblotting (H). The DHX8–RNA complex levels were normalized by the DHX8–HB levels (I). Data are mean ± SEM (n = 3). P-values are from unpaired t test (C and I) or Tukey’s multiple comparisons test (F and G).

Figure 4.

DHX8 interacts with SIDT2 and partly localizes to the cytoplasmic side of the lysosomal membrane. (Aand B) Cellular localization of endogenous DHX8 and LAMP1–EGFP in MEFs (A). Boxed regions are enlarged (A, lower panels); scale bar: 10 μm. Line scans show profiles of fluorescence intensity against line distance (B). (C and D) Cellular localization of endogenous DHX8 and SIDT2–EGFP in MEFs (C). Boxed regions are enlarged (A, C, and E, lower panels); scale bar: 10 μm. Line scans (D). (E–G) Cellular localization of endogenous DHX8 and SIDT2–EGFP in N2a cells (E). Boxed regions are enlarged (E, lower panels); scale bar: 10 μm. Line scans (F and G). (H) N2a cells overexpressing SIDT2 and DHX8–HB or untagged DHX8 were subjected to streptavidin (SA) pull-down assays. Interacting proteins were detected using immunoblotting. (I) Schematic representations of the full-length (WT) and truncated DHX8 constructs are shown with their respective domain compositions: NT (N-terminal fragment containing RS and S1 domains), CT (C-terminal fragment containing RecA1, RecA2, WH, Ratchet, and OB domains), RecA (fragment containing only RecA1 and RecA2 domains), and WRO (fragment containing WH, Ratchet, and OB domains). RS, arginine/serine-rich domain; S1, S1 domain; RecA1, RecA-like domain 1; RecA2, RecA-like domain 2; WH, winged helix domain; Ratchet, ratchet domain; OB, oligonucleotide/oligosaccharide-binding fold. Numbers indicate amino acid positions. (Jand K) N2a cells overexpressing SIDT2–FLAG and indicated DHX8–HB variants were subjected to streptavidin (SA) pull-down assays. Interacting proteins were detected using immunoblotting. (L–  N) Simultaneous interaction of the overexpressed SIDT2–FLAG and DHX8 with oligonucleotides in N2a lysates. Pull-down assays were performed using 1 nmol of poly-dG (15-mer). Interacting proteins were detected by immunoblotting. Data are mean ± SEM (n = 3). P-values are from unpaired t test (M and N).

Figure 5.

DHX8 mediates degradation of CAG repeat RNA. (A) Schematic representation of the constructs used in the experiment: HTTex1 with CAG-22 repeats (WT) tagged with EGFP, HTTex1 with CAG-145 repeats (mutant associated with Huntington’s disease) tagged with EGFP. (B) Experimental scheme for measuring mRNA levels and mRNA degradation of HTTex1 constructs using Tet-Off system. MEFs were seeded at –96 h time point, followed by plasmid transfection at the –72 h time point and siRNA transfection at the –48 h time point. mRNA levels of HTTex1 constructs were assessed at 0 h time point using qPCR. For mRNA degradation measurements, transcription of the constructs was inhibited by adding Dox at 0 h, mRNA levels of HTTex1 constructs were assessed at 0 and 6 h, and RNA degradation (%) was calculated as described in “Materials and methods” section. (C) MEFs were transfected with HTTex1 constructs, and then Dhx8 was knocked down using siRNA. The mRNA levels of HTTex1–CAG-22–EGFP or HTTex1–CAG-145–EGFP were measured. (Dand E) WT or Sidt2 KO MEFs were transfected with expression vectors for HTTex1–145–EGFP, and then Dhx8 was knocked down using siRNA. The mRNA levels (D) and mRNA degradation (E) of HTTex1–145–EGFP were measured. Data are mean ± SEM. Numbers in parentheses indicate n. P-values are from Tukey’s multiple comparisons test.

Figure 6.

DHX8 and SIDT2 cooperate in the clearance of cytoplasmic RNA foci of HTTex1–eCAGr. (A) Schematic representation of the experimental design to visualize mRNA localization of CAG-47 repeat (eCAGr) construct. The construct contains an eCAGr region followed by 12 × MS2 hairpins. The YFP-dMCP binds to the MS2 hairpins, allowing visualization of mRNA, and therefore localization and formation of RNA foci can be assessed. (B) Cellular localization of eCAGr mRNA, endogenous DHX8, and LAMP1–BFP in MEFs treated with or without Bafilomycin A1 (20 nM). Arrowheads indicate eCAGr RNA foci. Boxed regions are enlarged (right panels). Representative confocal images are shown; scale bar: 10 μm. (C and D) Imaging of eCAGr mRNA in WT or Sidt2 KO MEFs transfected with control or Dhx8 siRNAs (C). Lysosomes were stained with LysoTracker Red. Arrowheads indicate eCAGr RNA foci (C). Boxed regions are enlarged (C, lower panels). Representative confocal images are shown; scale bar: 10 μm. The number of cytoplasmic foci per cell was counted and shown as truncated violin plots with median and quartiles (D). P-values are from Tukey’s multiple comparisons test (D).

Figure 7.

DHX8 knockdown increases both soluble and insoluble expanded polyQ proteins. (A–E) MEFs overexpressing HTTex1–Q22–EGFP or HTTex1–Q145–EGFP were transfected with control or Dhx8 siRNAs. The cells were harvested and fractionated into 1% Triton X-100-soluble (A–C) and -insoluble (D and E) fractions, and the samples were analyzed using immunoblotting. Relative levels of soluble HTTex1-Q22 (B), 145-EGFP (C), and insoluble HTTex1-145-EGFP high molecular weight aggregates (E) were quantified. Data are mean ± SEM (n = 3). (Fand G) MEFs (F) and N2a cells (G), both stably expressing ATXN3-Q79 were transfected with control or Dhx8 siRNAs and incubated for 72 h. ATXN3-eCAGr mRNA levels were then measured. Data are mean ± SEM (n = 3). (H–K) MEFs and N2a cells stably expressing ATXN3-Q79 were transfected with control or Dhx8 siRNA and incubated for 72 h. The cells were harvested and fractionated into 1% Triton X-100-soluble (H and I) and -insoluble (J and K) fractions, and the samples were analyzed using immunoblotting. Relative levels of soluble ATXN3–Q79 (I) and insoluble ATXN3–Q79 (K) were quantified. Data are mean ± SEM (n = 3). P-values are from unpaired t test.