Ribosome profiling reveals novel regulation of C9ORF72 GGGGCC repeat-containing RNA translation

Abstract

GGGGCC (G₄C₂) repeat expansion in the first intron of C9ORF72 causes amyotrophic lateral sclerosis and frontotemporal dementia. Repeat-containing RNA is translated into dipeptide repeat (DPR) proteins, some of which are neurotoxic. Using dynamic ribosome profiling, we identified three translation initiation sites in the intron upstream of (G₄C₂) repeats; these sites are detected irrespective of the presence or absence of the repeats. During translocation, ribosomes appear to be stalled on the repeats. An AUG in the preceding C9ORF72 exon initiates a uORF that inhibits downstream translation. Polysome isolation indicates that unspliced (G₄C₂) repeat-containing RNA is a substrate for DPR protein synthesis. (G₄C₂) repeat-containing RNA translation is 5' cap-independent but inhibited by the initiation factor DAP5, suggesting an interplay with uORF function. These results define novel translational mechanisms of expanded (G₄C₂) repeat-containing RNA in disease.

Keywords: ALS; C9ORF72; DAP5; frontotemporal dementia; ribosome profiling; ribosome stalling; uORF translation.

FIGURE 1.

Ribosome profiling of C9ORF72. (A) iPSCs derived from three patients with expanded G₄C₂ repeats (C9) and three controls (cntrl) were treated with cycloheximide (CHX) or homoharringtonine (HHT) and then processed for ribosome profiling of the C9ORF72 RNA. Total read counts are provided in Supplemental Table 1. (B) An area of intron 1 upstream of G₄C₂ repeats was expanded to reveal two clusters of ribosome footprints. Also shown is an expanded area of intron 1 from a C9 patient following lactimidomycin D (LTM) treatment, which has an effect similar to HHT and ribosome profiling. (C) iPSC-derived neurons from a C9ORF72 patient and a healthy control were treated with HHT followed by ribosome profiling. The relevant area of intron 1 is shown. (D) A reporter construct containing a portion of C9ORF72 exon 1, C9ORF72 intron 1, G₄C₂₍₇₀₎, nano luciferase, a 3′ portion of C9ORF72 intron 1, the 5′ end of C9ORF72 exon 2, and firefly luciferase was expressed in HEK293T cells for ribosome profiling. (E) HEK cells were transfected with the plasmid noted above, or an identical one in which G₄C₂₍₇₀₎ was removed, followed by HHT treatment and ribosome profiling. Three ribosome footprints were detected on RNA derived from both plasmids, and are designated 1 (at 222), 2 (at 269), and 3 (at 281). The dip in the signal on the G₄C₂₍₇₀₎ RNA is due to a point mutation of an A for a G after 18 repeats in the reporter, which does not affect the reading frame. (F) Mapping of start sites with codon resolution shows that footprint 1 contains a putative CUG start site and is in-frame with poly(GP). Footprint 2 contains a putative CUG start site and is in-frame with poly(GA); however, a stop codon is immediately adjacent to this CUG codon. Footprint 3 contains a putative CUG start site in-frame with poly(GA).

FIGURE 2.

Functional analysis of translation start sites based on ribosome profiling. (A) The CUG in ribosome footprint 3 was mutated to CCG and the plasmid was transfected into HEK cells for 24 h. Nano and firefly luciferase activities and nano and firefly RNA levels (by RT-qPCR) in the poly(GA) and poly(GP) frames were measured and their relative translational activities (activity to RNA) were plotted. (B) The CUGs in the three ribosome footprints were mutated to CCG and nano and firefly activities and their RNAs in the poly(GA) and poly(GP) frames were measured and plotted as above. (C) The AUG at the end of exon 1 was mutated to UAA and nano and firefly luciferase activities in the poly(GA) and poly(GP) frames were measured and plotted as above. (*) P < 0.05; (**) P < 0.01, (***) P < 0.001, (****) P < 0.0001 (Student's t-test). (D) uORF formation and alignment of amino acid sequences of the uORF in human, orangutan, and macaque.

FIGURE 3.

Substrate for C9ORF72 intron translation. (A) Lysates from HeLa Flp-In cells stably expressing the reporter plasmid were ultracentrifuged through 0.5 M and 1 M sucrose to pellet polysomes. Parallel experiments were performed when lysates were treated with EDTA, which dissociated polysomes. (B) Spliced RNA was measured from exon 1 to exon 2 and was far more abundant in our input compared to unspliced RNA (pair 3), relative to hygromycin (hyg) RNA expressed from the same plasmid. (C) RNA from the pellets as well as input was subjected to RT-qPCR with the primers in intron 1 and exon 1 (pair 1), within intron 1 (pair 2), and intron 1 and exon 2 (pair 3). (D) The ratios of the RT-qPCR products using primer pair 1 (exon 1 and intron 1) to primer pair 2 (both in intron 1) in the polysome pellet and total input RNA is shown. (E) Lysates from iPSCs C9#1 (also referred to as C9 26#6) were ultracentrifuged through 0.5 M and 1 M sucrose to pellet polysomes. RNA from the pellets was subjected to RT-qPCR with the primers in intron 1 and exon 1 (pair 1), within intron 1 (pair 2), and intron 1 and exon 2 (pair 3). Parallel experiments were performed when lysates were treated with EDTA, which dissociated polysomes. (F) The ratios of the RT-qPCR products using primer pair 1 (exon 1 and intron 1) to primer pair 2 (both in intron 1) in the polysome pellet and total input RNA are shown. All experiments were performed in biological triplicate. (*) P < 0.01, (***) P < 0.001 (Student t-test).

FIGURE 4.

Mechanism of C9ORF72 intron translation. (A) A hairpin (−50 kcal/mol) or an identical size control sequence CAA(21) that forms no hairpin was inserted two bases downstream from the cap in the reporter construct. The plasmids were transfected into HEK cells followed by determination of nano and firefly luciferase activities and RNA levels. (B) The constructs shown in Figure 1D were transfected into HEK cells followed by HHT treatment as well as 4EGI, a small molecule inhibitor of the eIF4E–eIF4G interaction. 4EGI was applied to cells at 100 µM for 3 h. Ribosome profiling was then performed as in Figure 1E. (C) A HeLa Flp-In cell line expressing the construct depicted in Figure 1D was transduced with lentiviruses expressing a nonspecific sequence, negative siRNA (nsi) or shRNAs for DAP5, eIF3d, or eIF4E. Nano and firefly luciferase activities and RNAs were determined. These proteins as well as vinculin were then western blotted and quantified. (D) C9ORF72 and GAPDH RNA immunoprecipitation was performed with DAP5 antibody or nonspecific IgG on HeLa Flp-In cells. RNA values were normalized to DAP5, which was set at 1. All experiments were performed in biological triplicate. (*) P < 0.01, (***) P < 0.001 (one-way ANOVA). (E) A model for C9ORF72 GGGGCC repeat-containing RNA translation. Ribosome profiling shows initiating ribosomes on 3 CUG codons upstream in the intron of G₄C₂(n). Two of the CUG codons are in-frame with poly(GA) and one in-frame with poly(GP). A uORF initiating at an AUG codon at the end of exon 1 is a negative regulator of translation. Recruitment or positioning of ribosomes on the uORF AUG may be facilitated by the eIF4G-like protein DAP5. Ribosome stalling may occur on the G₄C₂(n) hairpins formed by the repeat expansion. Created with BioRender.com.

Sandra Almeida

Emily E. Stackpole

Heleen M. van ‘t Spijker