CRISPR-Cas9 Screens Identify the RNA Helicase DDX3X as a Repressor of C9ORF72 (GGGGCC)n Repeat-Associated Non-AUG Translation

Abstract

Hexanucleotide GGGGCC repeat expansion in C9ORF72 is the most prevalent genetic cause of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). One pathogenic mechanism is the aberrant accumulation of dipeptide repeat (DPR) proteins produced by the unconventional translation of expanded RNA repeats. Here, we performed genome-wide CRISPR-Cas9 screens for modifiers of DPR protein production in human cells. We found that DDX3X, an RNA helicase, suppresses the repeat-associated non-AUG translation of GGGGCC repeats. DDX3X directly binds to (GGGGCC)n RNAs but not antisense (CCCCGG)n RNAs. Its helicase activity is essential for the translation repression. Reduction of DDX3X increases DPR levels in C9ORF72-ALS/FTD patient cells and enhances (GGGGCC)n-mediated toxicity in Drosophila. Elevating DDX3X expression is sufficient to decrease DPR levels, rescue nucleocytoplasmic transport abnormalities, and improve survival of patient iPSC-differentiated neurons. This work identifies genetic modifiers of DPR protein production and provides potential therapeutic targets for C9ORF72-ALS/FTD.

Keywords: ALS; CRISPR-Cas9 screen; DDX3X; FTD; RAN translation; RNA; helicase; neurodegeneration; repeat expansion.

Figure 1:. Genome-wide CRISPR-Cas9 knockout screens in human cells identify modifiers of DPR protein production from C9ORF72 GGGGCC repeats.

(A) The diagram of the C9R-EGFP reporter construct. (B) Flow cytometry of a single cell clone expressing C9R-EGFP in the GA frame. (C) Flow cytometry of a single cell clone expressing C9R-EGFP for RAN translation and RFP670 for AUG-translation. (D) Flow chart of the CRISPR-Cas9 screening procedure. RPE-1 reporter cells expressing Cas9 were infected with the lentiviral sgRNA KO library. The total infected cells before FACS, and the top GFP-high and GFP-low cell populations collected by FACS, were subjected to deep sequencing and statistical analysis for sgRNA distribution. (E) The volcano plot visualizes gene knockout enrichment in cells with different DPR expression levels. Red: genes conferring up-regulation of GA-GFP levels when knocked out (10% FDR). Blue: genes conferring down-regulation of GA-GFP levels when knocked out (10% FDR). (F) Functional enrichment of candidate modifiers identified from the genome-wide knockout screens. (G) Validation of screen hits in the independent dual-luciferase reporter cells by siRNA knockdown. HeLa Flp-In monocistronic reporter cells were induced to express translation reporters by doxycycline after two days of siRNA transfection, and luciferase activities were measured after another 24 hours. The Nanoluc luciferase represents RAN translation of C9ORF72 GGGGCC repeats (C9R-NLuc), and the Firefly Luciferase is produced by canonical translation (AUG-FLuc). The relative expression was compared to non-targeting siRNA control. Data are mean ± s.e.m. from three biological replicates. See also Figure S1, Table S1 and S2.

Figure 2:. NXF1-NXT1 RNA export pathway modulates DPR protein production.

(A) Schematic of the NXF1-NXT1 RNA export pathway. The components identified from the screens were colored with red, the gradient of which is proportional to the enrichment significance (casTLE score). (B) HeLa Flp-In monocistronic reporter cells were induced to express translation reporters by doxycycline after two days of siRNA transfection, and luciferase activities were measured after another 24 hours. NLuc signals were normalized to FLuc in each sample and the relative expression was compared to non-targeting siRNA control. *P<0.05, **P<0.01, ***P<0.001, two-tailed t-test. Data are mean ± s.e.m. from three biological replicates. See also Figure S1.

Figure 3:. DDX3X represses RAN translation of C9ORF72 GGGGCC repeats.

(A) After two days of siRNA transfection, HeLa Flp-In dual-luciferase reporter cells were induced to express translation reporters by doxycycline and luciferase activities were measured after another 24 hours. Top: NLuc signals were normalized to total protein in each sample and the relative expression was compared to non-targeting siRNA control. ****P<0.0001, two-tailed t-test. Data are mean ± s.e.m. from three biological replicates. Bottom: Immunoblotting of DDX3X showed the knockdown efficiency. β-actin was blotted as internal control. (B) FLAG-DDX3X and GFP constructs were co-transfected into the GP reporter cells. After 24 h induction, the 20% highest GFP-expressing cells, 20% medium GFP-expressing cells, and GFP-negative cells were collected by FACS for subsequent assays. Top: NLuc signals were normalized to total protein in each sample and the relative expression was compared to negative control. *P<0.05, **P<0.01, two-tailed t-test. Data are mean ± s.e.m. from three biological replicates. Bottom: Immunoblotting of DDX3X showed the overexpression level. * non-specific product. β-actin was blotted as internal control. (C) C9R-NLuc and FLuc total RNAs were measured by qRT-PCR in control or DDX3X siRNA transfected reporter cells. *P<0.05, two-tailed t-test. Error bars represent s.e.m. in three biological replicates. (D) Cells were fractionated to separate nucleus and cytoplasm. The levels of nuclear marker GAPDH pre-mRNA and cytosolic marker mitochondria RNA MT-RNR1 (Right), as well as C9R-NLuc and FLuc RNA (Left), were measured by qRT-PCR and normalized to GAPDH mRNA in each fraction. The ratio of cytosolic/nuclear RNA showed the sub-cellular distribution of each RNA. Error bars represent s.e.m. in three biological replicates. (E) In vitro transcribed C9R-Nluc RNA (frame-GP) with either 5’-m⁷G cap or ApppG cap analogue was transfected into either control or DDX3X knockdown HeLa cells. Top: The NLuc luciferase was normalized to RNA level in each condition. **P<0.01, two-tailed t-test. Data are mean ± s.e.m. from three biological replicates. Bottom: Immunoblotting of DDX3X showed the knockdown efficiency. β-actin was blotted as internal control. (F) Reduction of DDX3X enhances the polysome association of C9R-NLuc RNA. Left: UV absorbance (254 nm) profile of sucrose gradient fractionations of cytoplasmic extracts from HeLa RAN translation reporter cells. The distribution of DDX3X across the ribosome profile was shown by immunoblotting. Middle and right panels: Quantification of C9R-NLuc RNA (middle) and AUG-FLuc RNA (right) distribution in polysome gradients. Each RNA was normalized to the spike-in CLuc internal control, and the relative levels in each fraction were calculated as percentage of the total levels from all the fractions. Data are mean ± s.e.m. from three biological replicates. (G) siRNA and plasmid DNA encoding FLAG-tagged wild type or mutant DDX3X were co-transfected into HeLa Flp-In GP reporter cells. Immunoblotting of DDX3X showed the knockdown efficiency and exogenous expression levels. NLuc signals were normalized to total protein in each sample and the relative expression was compared to non-targeting siRNA control. Data are mean ± s.e.m. from three biological replicates. (H) Reporter cells were treated with increasing dosage of DDX3X ATPase inhibitor RK-33 for 24 h. Immunoblotting showed the total DDX3X levels did not change significantly. NLuc signals were normalized to total protein in each sample and the relative expression was compared to DMSO treatment control. Data are mean ± s.e.m. from three biological replicates. See also Figure S2, S3 and S4.

Figure 4:. DDX3X directly binds GGGGCC repeat RNA.

(A) Affinity pulldown of biotin-labeled repeat RNAs and their interacting proteins from HeLa cell extract, followed by immunoblotting with DDX3X antibody. (B) MBP and MBP-tagged DDX3X wild type or G302V mutant were expressed and purified from bacteria, separated on SDS-PAGE and stained with Coomassie blue. (C) Time-dependent RNA-stimulated ATPase activity of 1 pmol DDX3X WT or G302 mutant in the presence of 2 pmol RNA and 0.1 mmol ATP was measured by quantification of phosphate release. Data are mean ± s.e.m. from three biological replicates. (D-G) EMSA of in vitro transcribed radiolabeled RNA repeats incubated with increasing doses of purified DDX3X proteins. (D) EMSA for MBP-DDX3X WT with (GGGGCC)₄₀ RNA. (E) EMSA for MBP with (GGGGCC)₄₀. (F) EMSA for MBP-DDX3X G302V with (GGGGCC)₄₀. (G) EMSA for MBP-DDX3X WT with antisense repeat (CCCCGG)₄₀ RNA. See also Figure S4.

Figure 5:. Partial loss of bel function enhanced expanded GGGGCC repeat toxicity in Drosophila.

(A) Representative images of external eye phenotypes of adult files. (G₄C₂)₅₈ expression driven by the GMR-Gal4 driver induces a rough eye phenotype. Partial loss of bel function either through the mutant allele (bel^EY08943) or RNAi knockdown (bel^JF02884 and bel^GL00205) enhances (G₄C₂)₅₈ toxicity. UAS-GFP was used as control for UAS-RNAi lines. Partial loss of bel by itself has no phenotype. The genotypes of all flies examined here are described in detail in the Methods section. (B) Quantification of the rough eye phenotypes of flies with various genotypes as presented in Panel (A). ****P < 0.0001, by chi-square analysis. The number of flies analyzed for each genotype ranges from 50 to 200. (C) Repeat RNA level was measured by qRT-PCR and is shown to be not changed by RNAi knockdown (bel^JF02884 and bel^GL00205). Values are mean ± s.e.m. from four independent experiments, analyzed by one-way ANOVA. (D) ELISA analysis of poly(GP) protein levels in the heads of 5-day-old flies. GP levels are increased by RNAi knockdown of bel expression. Values are mean ± s.e.m. of six independent experiments, analyzed by one-way ANOVA with Dunn’s post hoc test. See also Figure S5.

Figure 6:. DDX3X modulates DPR protein levels and DPR-mediated toxicity in patient cells.

(A) DDX3X in two C9ORF72-ALS patient lymphoblast cell lines was knocked down by two individual shRNAs. (Top) Poly-GP was measured by ELISA. Different shapes of data points represent different lymphoblast lines. ***P<0.001, two-tailed t-test. Data are mean ± s.e.m. from two lines with two biological replicates each. (Bottom) Immunoblotting of DDX3X showed the knockdown efficiency. β-actin was blotted as internal control. (B) C9ORF72-ALS iPSCs were transfected with either non-targeting control or DDX3X siRNA. (Top) Poly-GP was measured by ELISA. Different shapes of data points represent different iPSC lines. ****P<0.0001, two-tailed t-test. Data are mean ± s.e.m. from three C9ORF72-ALS cell lines and two with two biological replicates. (Bottom) Immunoblotting of DDX3X showed the knockdown efficiency. β-actin was blotted as internal control. (C) iPSNs were infected with lentivirus expressing either DDX3X or GFP as negative control. (Top) Poly-GP was measured by ELISA. Different shapes of data points represent different lines. *P<0.05, two-tailed t-test. Data are mean ± s.e.m. from three control and three C9ORF72-ALS cell lines and one with two biological replicates. (Bottom) qRT-PCR of DDX3X RNA showed the fold of overexpression compared to the endogenous level in each line. Data are mean ± s.e.m. from two technical replicates. (D) Immunofluorescence of exogenously expressed FLAG-tagged DDX3X, Ran-GFP reporter, and MAP2 as neuronal marker in iPSC-derived neurons. The scale bar represents 5 μm. (E) Quantification of neuronal C/N Ran ratios in (D) showed increased cytoplasmic Ran levels in C9ORF72-ALS iPSN compared to control, and overexpression of DDX3X decreased cytoplasmic Ran. Neurons expressing both Ran-GFP and FLAG-DDX3X were measured and compared with the negative control. Data points of different colors represent individual cells from different cell lines. 25-40 neurons were quantified at each condition. ****P<0.0001, one-way ANOVA with Dunn’s post hoc test. Data are mean ± s.d. from three C9ORF72-ALS iPSN lines and three CTRL iPSN lines. (F) Glutamate-induced excitotoxicity of C9ORF72 iPSNs was significantly higher than control iPSNs. Exogenous expression of DDX3X prevents glutamate-induced C9ORF72 iPSN cell death. Each data point represents one individual line. About 500 neurons were quantified at each condition. ***P<0.001, ****P<0.0001, two-way ANOVA with Tukey’s post hoc test. Data are mean ± s.d. from two CTRL iPSN lines and three C9ORF72-ALS iPSN lines. See also Figure S6.