A C. elegans model of C9orf72-associated ALS/FTD uncovers a conserved role for eIF2D in RAN translation

A hexanucleotide repeat expansion GGGGCC in the noncoding region of C9orf72 is the most common cause of inherited amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). Potentially toxic dipeptide repeats (DPRs) are synthesized from the GGGGCC sequence via repeat associated non-AUG (RAN) translation. We developed C. elegans models that express, either ubiquitously or exclusively in neurons, a transgene with 75 GGGGCC repeats flanked by intronic C9orf72 sequence. The worms generate DPRs (poly-glycine-alanine [poly-GA], poly-glycine-proline [poly-GP]) and display neurodegeneration, locomotor and lifespan defects. Mutation of a non-canonical translation-initiating codon (CUG) upstream of the repeats blocked poly-GA production and ameliorated disease, suggesting poly-GA is pathogenic. Importantly, eukaryotic translation initiation factor 2D (eif-2D/eIF2D) was necessary for RAN translation. Genetic removal of eif-2D increased lifespan in both C. elegans models. In vitro findings in human cells demonstrated a conserved role for eif-2D/eIF2D in RAN translation that could be exploited for ALS and FTD therapy.

on eif-2D (gk904876) mutants and eif-2D (gk904876); C9 ubi animals. This analysis revealed that prominent locomotor defects (Fig. 5). In summary, this behavioral analysis complements our 2 6 6 molecular analysis (poly-GA detection) and survival assays (Fig. 4), providing compelling 2 6 7 evidence that loss of eif-2D ameliorates the pathogenic phenotype associated with RAN 2 6 8 translation of C9orf72 G4C2 nucleotide repeats. We next sought to determine whether the function of eif-2D/eIF2D in RAN translation and DPR 2 7 2 production is conserved from worms to human cells. To this end, we transfected HEK293 cells 2 7 3 with a bicistronic construct, in which the firefly luciferase (fLuc) gene is located in the first 2 7 4 cistron followed by a second cistron containing the 75 G4C2 repeats along with intronic 2 7 5 sequences that normally flank the repeats in the human C9orf72 locus (Fig. 6A). To monitor occur when the G4C2 repeats are located in the second cistron of a bicistronic construct 20, 37 .

8 0
Each bicistronic construct was co-transfected with plasmids carrying either control short hairpin 2 8 1 RNA (shRNA) or shRNA against eIF2D. First, we assessed the eIF2D protein levels via Western 2 8 2 blotting and observed a close to 50% reduction in HEK293 cells that received the eIF2D shRNA 2 8 3 compared to cells transfected with control shRNA (Fig. 6B-C). Next, we performed Western 2 8 4 blotting and luciferase assays and found that, similar to our C. elegans model, poly-GA and poly- increased sensitivity of the luciferase assay enabled us to detect poly-GR in these cells, albeit at 2 8 7 lower levels compared to poly-GA (Fig. 6E). Importantly, knock-down of eIF2D by shRNA 2 8 8 dramatically reduced (by at least 75%) the expression of all three DPRs (poly-GA, poly-GP, 2 8 9 poly-GR), suggesting eIF2D is required for RAN translation of GA, GP, and GR (Fig. 6B, D-E).

9 0
In contrast, the protein levels of fLuc, which is expressed by the first cistron via canonical AUG- summary, these results suggest that eIF2D is necessary for DPR production (poly-GA, poly-GP, DPRs from G4C2 repeat-containing RNA. The discovery of RAN translation of C9orf72 nucleotide repeats combined with the reported 3 0 9 toxicity of DPRs in model systems have focused the ALS/FTD field on this unconventional form 3 1 0 of translation, as well as on DPRs as potential therapeutic targets. To study the molecular 3 1 1 mechanisms underlying RAN translation, we initially developed a C. elegans model that carries 3 1 2 a transgene encoding 75 copies of the G4C2 repeat under the control of a ubiquitous promoter. to DPR toxicity because mutation of a non-canonical CUG initiation codon decreased poly-GA 3 1 6 production and ameliorated locomotor and lifespan defects in C9 ubi animals without affecting 3 1 7 C9orf72 mRNA levels. Moreover, a second C. elegans model (referred to as C9 neuro ) that 3 1 8 generates poly-GA exclusively in neurons displayed a similar phenotype. Importantly, we found 3 1 9 that knock-out of the eukaryotic translation initiation factor eif-2D/eIF2D suppressed poly-GA 3 2 0 expression and ameliorated lifespan defects in both C9 ubi and C9 neuro models. Lastly, our in 3 2 1 vitro findings in human cells combined with our C. elegans data argue for a conserved role of  The C9 ubi and C9 neuro worm models offer insights into DPR toxicity 3 2 6 An unresolved issue in the field of ALS and FTD research is the relative contribution of each of 3 2 7 the five DPRs (poly-GA, poly-GP, poly-GR, poly-PR, poly-PA) to disease pathogenesis. In this 3 2 8 paper, two lines of evidence indicate that poly-GA, the most readily visible DPR in the brain and disease phenotypes in C. elegans: (1) poly-GA aggregates and appears to be the most abundantly 3 3 1 expressed DPR in our worm models, and (2) mutation of the non-canonical CUG initiation 3 3 2 codon decreased poly-GA production and ameliorated locomotor and lifespan defects in C9 ubi 3 3 3 and C9 neuro animals. Importantly, these in vivo results significantly extend previous reports 3 3 4 demonstrating that this same CUG is required in vitro for poly-GA translation 20,29,30,31 . In is that poly-GA causes cellular toxicity through proteosomal impairment and protein 3 3 9 sequestration 5 . A recent publication reports that in an in vitro system poly-GA spreads cell-to- associated ALS/FTD 43 . However, a zebrafish model that produces poly-GA did not exhibit motor 3 4 5 axon toxicity 44 , presumably due to low poly-GA expression.

4 6
Although our C9 ubi worm model produces a small amount of poly-GP, this DPR is 3 4 7 thought to be relatively nontoxic 5 . In contrast, the toxicity of arginine-containing DPRs, constructs driving expression of poly-GR or poly-PR caused toxicity in muscles and neurons 46 . Furthermore, initiation of poly-GR translation has been reported to occur at the same upstream 3 5 1 CUG codon used for poly-GA production, but with subsequent frameshifting into the GR reading 3 5 2 frame 30 . However, two recent papers showed that translation initiation of poly-GR did not occur 3 5 3 In addition to eIF2D, RPS25 is the only other factor known to be required for the 3 9 9 expression of these three DPRs, albeit to variable degrees, i.e., RPS25 knock-down in vitro 4 0 0 reduced poly-GA by 90%, poly-GP by 50%, and poly-GR by 30% 13 . Collectively, these and 4 0 1 other studies 16 suggest that a number of different factors play critical roles in RAN translation.

0 2
Although the present study uncovered a requirement for eif-2D/eIF2D in RAN translation,  Two previous studies reported DPR-associated toxicity in C. elegans by using either a codon-4 1 2 optimized construct that carries no G4C2 repeats, or a construct that carries 66 G4C2 repeats 22, 46 .

1 3
The latter study strongly indicated that repeat associated non-AUG (RAN) translation occurs in 4 1 4 the worm. However, the non-canonical (non-AUG) initiating codon and translation initiation 4 1 5 factor remained elusive. By generating C. elegans animals that carry 75 G4C2 repeats flanked by with previous reports in yeast and flies 16,22 , strongly suggest that the molecular machinery 4 2 0 required for RAN translation is evolutionary conserved, offering an opportunity to use simple 4 2 1 model systems for the discovery of RAN regulators.

2 2
In summary, we leveraged the specific strengths of the C. elegans model and 4 2 3 discovered that the non-canonical translation initiation factor eif-2D/eIF2D is required for RAN influences RAN translation is thus an important future endeavor.   The template for the generation of the plasmids used in the C9 ubi , Δ C9 ubi, and UAG ubi animals 4 6 6 have been described in an in vitro study 20 . In C9 ubi animals, the plasmid has 75 G4C2 repeats in the reading frame of poly-GA, which is followed by the unc54 3' UTR. In Δ C9 ubi animals, the 4 7 0 construct is identical but lacks the G4C2 repeats. In UAG ubi animals, the construct is identical 4 7 1 but the non-canonical translation initiating codon CUG is mutated to UAG. In C9 ubi , Δ C9 ubi , 4 7 2 UAG ubi animals, the following promoter sequence of the snb-1 gene was used: The unc54 3' UTR was cloned into the XhoI-SacI site of the plasmids using the primer 4 9 2 sets shown in Supplementary Table 1. Fig. 1A shows the constructs with the snb-1 promoter and 4 9 3 the upstream CUG, which is the translation initiation codon for poly-GA (see Results).