C9ORF72 GGGGCC repeat-associated non-AUG translation is upregulated by stress through eIF2α phosphorylation

Abstract

Hexanucleotide repeat expansion in C9ORF72 is the most frequent cause of both amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). Here we demonstrate that the repeat-associated non-AUG (RAN) translation of (GGGGCC) _n -containing RNAs into poly-dipeptides can initiate in vivo without a 5'-cap. The primary RNA substrate for RAN translation of C9ORF72 sense repeats is shown to be the spliced first intron, following its excision from the initial pre-mRNA and transport to the cytoplasm. Cap-independent RAN translation is shown to be upregulated by various stress stimuli through phosphorylation of the α subunit of eukaryotic initiation factor-2 (eIF2α), the core event of an integrated stress response (ISR). Compounds inhibiting phospho-eIF2α-signaling pathways are shown to suppress RAN translation. Since the poly-dipeptides can themselves induce stress, these findings support a feedforward loop with initial repeat-mediated toxicity enhancing RAN translation and subsequent production of additional poly-dipeptides through ISR, thereby promoting progressive disease.

Fig. 1

RAN translation of C9ORF72 hexanucleotide repeats can initiate with and without 5′-cap. a Schematic of monicistronic dual-luciferase reporters for RAN translation and canonical AUG translation. b HeLa Flp-In cells were induced to express translation reporters by doxycycline for 24 h. Relative RAN translation products from Frame-GA and Frame-GP were compared to no-repeat control. NLuc signals were normalized to FLuc in each sample. c Immunoprecipitation using MYC antibody from cells expressing Neg-NLuc or C9R-NLuc, followed by immunoblotting with GA or GP antibody. d Schematic of bicistronic reporters for cap-independent RAN translation and AUG translation. e Relative RAN translation products from Frame-GA and Frame-GP were compared to no-repeat control. NLuc signals were normalized to FLuc in each sample. f Immunoprecipitation using MYC antibody from bicistronic reporter cells followed by immunoblotting with GA or GP antibody. g Schematic timeline of siRNA transfection and induction of luciferase reporters in HeLa Flp-In cells. h Reporter cells were transfected with non-targeting siRNA or siRNA against cap-binding protein eIF4E. Immunoblotting of eIF4E showed the knockdown efficiency. GAPDH was blotted as internal control. i Expression of AUG-FLuc translation (left) and C9R-NLuc RAN translation (right) reporters in presence of eIF4E siRNA compared to non-targeting siRNA control. *P < 0.05, **P < 0.005, ***P < 10^–6, two-tailed t-test. j Reporter cells were treated with mTOR inhibitor Torin 1 for 24 h. Immunoblotting of phospho-4E-BP1 using antibodies recognizing different phosphorylation sites. β-actin was blotted as internal control. k Expression changes of AUG-FLuc and C9R-NLuc in bicistronic reporter cells under mTOR pathway inhibition by Torin 1 treatment. **P < 0.005, ***P < 0.0005, two-tailed t-test. l The relative translation levels of C9R-NLuc (Frame-GA and Frame-GP) with and without a functional 5′-cap. HeLa cells were transfected with in vitro transcribed monocistronic C9R-NLuc and Neg-NLuc RNA with either 5′-m⁷G cap or ApppG cap analog. The NLuc luciferase was normalized to RNA level in each condition, and each sample was compared with 5′-m⁷G capped C9R-NLuc in frame-GA (set as 100). Data are mean ± SEM from three biological replicates

Fig. 2

Cap-independent RAN translation is upregulated upon stress. a Schematic timeline of stress stimuli and induction of luciferase reporters in HeLa Flp-In cells. b Expression of bicistronic luciferase reporters was induced by doxycycline for 6 h in HeLa Flip-In cells. At the same time, cells were treated with arsenite or MG132 stimuli. Relative expression of AUG-FLuc and C9R-NLuc reporters was compared with no stress control. The luciferase signals were normalized to the total protein amount. Data are mean ± SEM from three biological replicates. ***P < 0.0005, two-tailed t-test. c Stress granules were detected with G3BP immunofluorescence in reporter cells treated with arsenite and MG132. d Immunoblotting using antibody recognizing phospho-eIF2α at the Ser51 site showed the upregulation under stress. GAPDH and total eIF2α were blotted as internal control

Fig. 3

Stress-induced RAN translation elevation is mediated through eIF2α phosphorylation. a Diagram of the integrated stress response (ISR) and the actions of small-molecular inhibitors of ISR. b Schematic timeline of stress stimuli and inhibitor treatments, and induction of luciferase reporters. c G3BP immunofluorescence of reporter cells treated with arsenite and MG132, without or with pretreatment of ISRIB and PERKi. Stress granules were detected with G3BP antibody. d Quantification of stress granule numbers per cell in c. More than 100 cells were counted in each condition. ***P < 0.0005, two-tailed t-test. e Fold change of relative RAN translation vs. AUG translation in bicistronic reporters under arsenite (left) and MG132 (right) stimuli, without or with pretreatment of ISRIB or PERKi compounds targeting the eIF2α pathway. NLuc signals were normalized to FLuc. **P < 0.005, ***P < 0.0005, two-tailed t-test. f Schematic timeline of cDNA transfection and inhibitor treatment (as in h), and induction of luciferase reporters. g Relative expression of C9R-NLuc vs. AUG-FLuc reporters upon transfection of eIF2α wild-type or S51D mutant were compared with negative control. NLuc signals were normalized to FLuc in each sample. *P < 0.05, **P < 0.005, two-tailed t-test. h Fold change of bicistronic RAN translation reporter expression upon eIF2α-S51D transfection, without or with ISRIB treatment. *P < 0.05, **P < 0.005, two-tailed t-test. Data are mean ± SEM from three biological replicates

Fig. 4

TDP-43 prion-like domain promotes RAN translation through phospho-eIF2α. a GFP fluorescence and G3BP immunofluorescence of HeLa Flp-In cells transfected with GFP control, TDP43-WT, or TDP43-F4, without or with the treatment of ISRIB or PERKi. b Quantification of stress granule numbers per GFP-positive cell in a. More than 100 cells were counted in each condition. ***P < 0.0001, two-tailed t-test. c Immunoblotting using antibody recognizing phospho-eIF2α showed its upregulation with TDP43-F4 expression, compared to GFP and TDP43-WT controls. GAPDH and total eIF2α were blotted as internal control. d After 1 day of transfection cells were induced to express translation reporters by doxycycline, and luciferase activities were measured after another 24 h. NLuc signals were normalized to FLuc in each sample and the relative expression was compared to GFP transfection control. *P < 0.05, two-tailed t-test. e Cells were treated with ISRIB or PERKi after transfection with TDP43-F4. Immunoblotting with antibodies recognizing phospho-eIF2α, eIF2α, phospho-PERK, and GAPDH. f Fold change of bicistronic RAN translation reporter expression upon TDP43-F4 expression, without or with the treatment of ISRIB or PERKi. *P < 0.05, **P < 0.005, two-tailed t-test. Data are mean ± SEM from three biological replicates

Fig. 5

RAN translation of GGGGCC repeats from C9ORF72 spliced intron. a Schematic of bicistronic splicing dual-luciferase reporters with C9ORF72 exons and intron. C9R-NLuc reporter is located in intron 1 and AUG-FLuc reporter is in exon 2 fused with the original C9ORF72 start codon. The positions of primers amplifying different RNA products were labeled. b Relative RAN translation products from Frame-GA and Frame-GP were compared to no-repeat control, after 24 h induction in HeLa Flp-In cells. NLuc signals were normalized to FLuc in each sample. c Immunoprecipitation using MYC antibody from cells expressing Neg-NLuc, C9R-NLuc of frame-GA, or GP in bicistronic splicing reporters, followed by immunoblotting with GA or GP antibody. d The relative translation level of C9R-NLuc in monocistronic, bicistronic, and bicistronic splicing reporters. The NLuc luciferase level in each cell line was normalized to the NLuc RNA level, which was quantified by qRT-PCR. e After 24 h induction of reporter genes, cells were fractionated to separate nucleus and cytoplasm. The levels of reporter spliced mRNA and unspliced pre-mRNA, Nluc (amplifying excised intron and unspliced pre-mRNA), GAPDH pre-mRNA, and the mitochondria RNA MTRNR1 were measured by qRT-PCR and normalized to GAPDH mRNA in each fraction. The ratio of cytosol/nuclear RNA showed the subcellular distribution of each RNA. The nuclear marker GAPDH pre-mRNA is depleted in cytosol fraction (left) and the cytosol marker MTRNR1 is highly enriched in cytosol fraction (middle). NLuc RNA shows more cytosolic distribution than pre-mRNAs (right). f The diagram of experiment design and TRAP methodology to isolate ribosome-associated RNAs. g Total, cytosol, and ribosome-associated RNAs were isolated for qRT-PCR. NLuc and reporter pre-mRNA were measured and normalized to GAPDH mRNA in each fraction. The ratios of ribosome-associated/cytosol RNA (left) and ribosome-associated/total RNA (right) show that the Nluc intron RNA is associated with ribosomes, but the pre-mRNA is not. Data are mean ± SEM from three biological replicates

Fig. 6

Translation of intronic repeats does not require 5′-cap and can be upregulated by stress. a HeLa reporter cells were transfected with non-targeting siRNA or siRNA against cap-binding protein eIF4E. Immunoblotting of eIF4E showed the knockdown efficiency. β-actin was blotted as internal control. b Expression of AUG-FLuc translation (left) and C9R-NLuc RAN translation (right) reporters in presence of non-targeting siRNA or siRNA against eIF4E. **P < 0.005, ***P < 0.0005, two-tailed t-test. c Bicistronic splicing reporter cells were treated with mTOR inhibitor Torin 1 for 24 h. Immunoblotting of phospho-4E-BP1 using antibodies recognizing different phosphorylation sites. β-actin was blotted as internal control. d Expression of AUG-FLuc and C9R-NLuc in bicistronic splicing reporter cells with the treatment of mTOR inhibitor Torin 1. **P < 0.005, ***P < 0.0005, two-tailed t-test. e Fold change of relative RAN translation vs. AUG translation of the bicistronic splicing reporters under arsenite (left) and MG132 (right) stimuli, without or with pretreatment of ISRIB or PERKi. NLuc signals were normalized to FLuc in each sample. *P < 0.05, **P < 0.005, ***P < 0.0001, two-tailed t-test. f Fold change of RAN translation upon eIF2α-S51D expression, without or with the treatment of ISRIB. *P < 0.05, **P < 0.005, two-tailed t-test. g Diagram for a feedforward loop with C9ORF72 (GGGGCC)_n RAN translation being enhanced by initial repeat-mediated toxicity through eIF2α phosphorylation pathway. Data are mean ± SEM from three biological replicates