Nucleocytoplasmic Proteomic Analysis Uncovers eRF1 and Nonsense-Mediated Decay as Modifiers of ALS/FTD C9orf72 Toxicity

Abstract

The most common genetic cause of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) is a hexanucleotide repeat expansion in C9orf72 (C9-HRE). While RNA and dipeptide repeats produced by C9-HRE disrupt nucleocytoplasmic transport, the proteins that become redistributed remain unknown. Here, we utilized subcellular fractionation coupled with tandem mass spectrometry and identified 126 proteins, enriched for protein translation and RNA metabolism pathways, which collectively drive a shift toward a more cytosolic proteome in C9-HRE cells. Among these was eRF1, which regulates translation termination and nonsense-mediated decay (NMD). eRF1 accumulates within elaborate nuclear envelope invaginations in patient induced pluripotent stem cell (iPSC) neurons and postmortem tissue and mediates a protective shift from protein translation to NMD-dependent mRNA degradation. Overexpression of eRF1 and the NMD driver UPF1 ameliorate C9-HRE toxicity in vivo. Our findings provide a resource for proteome-wide nucleocytoplasmic alterations across neurodegeneration-associated repeat expansion mutations and highlight eRF1 and NMD as therapeutic targets in C9orf72-associated ALS and/or FTD.

Keywords: C9orf72; ETF1/eRF1; UPF1; amyotrophic lateral sclerosis; frontotemporal dementia; motor neurons; neurodegeneration; nonsense-mediated decay; nuclear invaginations; nucleocytoplasmic proteome.

Figure 1.. The C9-HRE Leads to Proteome-Wide N/C Redistribution.

(A) Experimental schematic used to identify the effects of the C9-HRE on the subcellular proteome. (B) WB for TAFII p130 and VINCULIN in the cytosolic (C) and nuclear (N) fractions of GFP-, C9×8- and C9×58-expressing cells. (C) Dot plots representing the cytosolic ratio obtained from LC-MS/MS analysis in GFP-expressing cells of bona fide cytosolic (black) and nuclear (gray) markers (n=3 independent biological replicates). (D) Bar plots representing the average number of redundant proteins identified in cells expressing GFP, C9×8 and C9×58 by LC-MS/MS. ANOVA; ns=not significant. (E) Violin plots showing the proteome-wide cytosolic ratio of the ~10,000 proteins identified in cells expressing GFP (0.528±0.003; mean ± SEM), C9×8 (0.548±0.003), and C9×58 (0.575±0.003). Kruskal-Wallis test. (F) Left: Representative confocal images of cells expressing N/C shuttling tdTomato reporter and GFP, C9×8-GFP or C9×58-GFP. Dashed lines mark the nuclear boundary. Right-top: N/C shuttling tdTomato reporter and workflow schematics. Right-bottom: Dot plot showing the fold change in the cytosolic/nuclear (C/N) ratio of the tdTomato signal in C9×58-expressing cells compared to controls. Individual cells are displayed as dots. Kruskal-Wallis test. (G) Violin plots showing the proteome-wide cytosolic ratio of the ~8,000 proteins identified in cells expressing GFP (0.58±0.004; mean ± SEM), ATX1–32Q (0.57±0.004), ATX1–84Q (0.61±0.004), HTT-25Q (0.60±0.004) and HTT-72Q (0.65±0.004). Kruskal-Wallis test. (H) Venn diagram showing the level of overlap of redistributed proteins in cells expressing C9×58, ATX1–84Q and HTT-72Q. In graphs, bars and horizontal lines represent mean ± SEM, and dotted lines mark the GFP-transfected cells mean value. Scale bars: 15 μm.

Figure 2.. The C9-HRE Causes a Redistribution of Proteins Involved in RNA Metabolism and Protein Translation.

(A) Volcano plot representing the magnitude of change of the cytosolic ratio (x-axis) in C9×58-expressing cells compared to both controls (GFP- and C9×8-expressing cells). Y-axis represents the significance of the change (-log10 of the p-value). Proteins that exhibited a significant change were labeled in dark gray. (B) Left: pie chart displaying the percentage of redundant proteins identified by LC-MS/MS that demonstrated statistically significant subcellular redistribution in C9×58-expressing cells compared to controls. Right: Schematic showing the percentage of redistributed proteins that become more nuclear (red) or more cytosolic (green) in C9×58-expressing cells. (C) Top left: graphical representation of size-dependent N/C transport of proteins by passive diffusion or active transport. Top right: histogram showing the size distribution of proteins significantly redistributed in C9×58-expressing cells (black lined bars) compared to the global HEK-293 detected proteome (gray filled bars). Bottom: Pie charts exhibiting the percentage of proteins out of the global HEK-293 cell proteome detected (left) or out of the group of proteins redistributed in C9×58-expressing cells (right), that likely cross the nuclear envelope by passive (gray) or active (blue) transport. (D) Gene ontology (GO) analysis assessed for significant enrichment in biological processes of the 126 unique proteins redistributed in C9×58-expressing cells. Gray bars represent the number of proteins enriched for each pathway relative to the total number of redistributed proteins. P values and fold enrichment values for each cellular process are displayed on the right. (E) Graphical representation of the most significant (p<0.05) GO terms associated with redistributed proteins in C9×58-expressing cells. Proteins with higher (red) or lower (green) N/C ratios are depicted with their corresponding GO term and subcellular localization.

Figure 3.. Representative Redistributed Proteins are Genetic Modifiers of C9-HRE Toxicity in Drosophila Disease Models.

(A) Graph displaying the significantly redistributed proteins, as well as the direction and magnitude of the change observed in C9×58-expressing cells compared to controls. The y-axis represents the change in cytosolic ratio and the x-axis represents the change in nuclear ratio of proteins identified. The upper left gray quadrant includes proteins with a higher N/C ratio, while the lower right quadrant proteins with a lower N/C ratio. Target proteins selected for genetic interaction studies in Drosophila are highlighted in green (suppressors), red (enhancers) or orange (no effect) fonts. (B) Schematic illustrating the genetic interaction approach in Drosophila models of C9-HRE toxicity. Flies expressing RNAi for selected proteins were crossed with flies expressing 3, 30 or 36 G₄C₂ repeats, and the resultant level of toxicity was assessed by the level of suppression (green eye) or enhancement (red eye) of eye degeneration. (C) Bar plots displaying the level of eye degeneration in flies carrying C9×30 repeats with RNAi for ETF1, PRMT1, ENY2, TNPO3, CCT8, or SRSF1. Enhancers and suppressors of toxicity are represented by red and green bars, respectively. The basal level of eye degeneration in C9×30 flies is shown in yellow. Mann-Whitney U test. (D) Representative images of fly eyes from wild-type (GMR-GAL4 x EGFP) or C9-HRE (C9×30 or x36) mutant flies, with endogenous levels (EGFP), knockdown (ETF1-RNAi) or overexpression (ETF1-OE) of eRF1. Arrow heads indicate the presence of necrotic patches. (E) Bar plots displaying the level of eye degeneration in flies carrying C9-HRE (C9×36 repeats) with endogenous levels, knockdown (ETF1-RNAi) or overexpression (ETF1-OE) of eRF1. The basal level of eye degeneration in C9×36 flies is shown in orange; suppression and enhancement of C9-HRE toxicity is shown in green and red, respectively. Mann-Whitney U test. (F) Dot blot of poly-GR in control (GMR-GAL4), C9×36, and C9×36 flies with decreased (ETF1 RNAi) or increased (ETF1-OE) eRF1. Tubulin was used as a loading control. In graphs, bars represent mean ± SEM and the dotted lines mark the mean level of eye degeneration in C9-HRE flies.

Figure 4.. eRF1 is Redistributed in C9-ALS iPSC-Derived MNs.

(A) Representative WB bands for eRF1 in nuclear and cytosolic fractions of MN cultures derived from a control and a C9-ALS iPSC line. Bottom: Heat map showing the average eRF1 protein level normalized to subcellular fraction-loading controls (VINCULIN and LMNA/C, for cytosolic and nuclear fractions, respectively) in MNs derived from 3 healthy control and 3 C9-ALS patient iPSC lines (n=5–7 independent differentiations). (B) Left: Representative confocal images of healthy control and C9-ALS patient MNs immunolabeled for eRF1 (red), MAP2 (green) and DNA (blue). Dashed lines mark the nuclear boundary. Right: 3-D reconstruction of control (top) and C9-ALS patient (bottom) MN nuclei shown on the left. (C) Dot plot displaying the fold change in the N/C ratio of eRF1 signal quantified in MNs derived from 3 controls and 3 C9-ALS patient iPSC lines. Individual cells are represented as empty-filled circles, bars represent mean ± SEM, and the dotted line marks the mean N/C ratio in control samples. Mann-Whitney U test. (D) Representative FISH+ICC confocal image of a C9-ALS patient MN nucleus immunolabeled for G₄C₂ RNA foci (green), eRF1 (red) and DNA (blue). Dashed lines mark the inset region magnified in the right bottom merged image. (E) Pie charts displaying the percentage of nuclear eRF1 signal that colocalizes with C9-HRE RNA foci (left) and the percentage of C9-HRE RNA foci that colocalize with eRF1 signal (right). (F) Representative structured-illumination microscopy (SIM) image of a C9-ALS patient MN nucleus immunolabeled for eRF1 (red) and LMNB1 (green). (G) Representative serial confocal images of a C9-ALS patient MN nucleus immunolabeled for eRF1 (red) and LMNB1 (green). Arrow heads indicate ERF1+ puncta localized within LMNB1+ nuclear envelope invaginations. (H) Pie chart displaying the percentage of nuclear eRF1 signal that colocalizes with LMNB1+ nuclear envelope invaginations in C9-ALS patient MNs. (I) Regular (top) or expansion (bottom) confocal microscopy images of human iPSC-derived MN nuclei immunolabeled for LMNB1 (green) and DNA (blue). (J) Representative expansion microscopy images of healthy control (top) and C9-ALS patient (bottom) MN nuclei immunolabeled for eRF1 (red), LMNB1 (green), and DNA (blue). (K) Top: Representative expansion microscopy image of a C9-ALS patient MN nucleus immunolabeled for puromycin incorporation (red), LMNB1 (green), and DNA (blue). Bottom: Representative expansion microscopy image of C9-ALS patient MN nucleus immunolabeled for KDEL (red), LMNB1 (green), and DNA (blue). Scale bars: 25 (I), 10 (B left, K), 5 (B right, D, F, G, J), 1 (D inset) μm. N: Nucleus; C: Cytosol.

Figure 5.. eRF1 is Redistributed in Postmortem C9-ALS Tissue.

(A) Representative IHC confocal images of layer V neurons immunolabeled for eRF1 (green), MAP2 (red) and DNA (blue) in motor cortex tissue from non-neurological controls and C9-ALS patients. Dashed lines mark the nuclear boundary. (B) Dot plot displaying the fold change in the N/C ratio of eRF1 signal observed in cortical neurons of 3 non-neurological age-matched controls and 3 C9-ALS patients. Mann-Whitney U test. (C) Representative IHC confocal images of layer V neurons immunolabeled for LMNB1(red), MAP2 (green), and DNA (blue) in motor cortex tissue from non-neurological controls and C9-ALS patients. (D) Dot plot displaying the fold change in the nuclear area occupied by LMNB1+ invaginations in cortical neurons of 3 non-neurological age-matched controls and 3 C9-ALS patients. Mann-Whitney U test. In graphs, bars represent mean ± SEM and the dotted lines mark the mean in control samples. Scale bars: 10 μm.

Figure 6.. Protein Translation and mRNA Degradation Through NMD in C9-HRE Expressing Cells.

(A) Top: schematic of de novo protein translation assay performed in iPSC-derived MNs. Bottom: representative image of HPG fluorescent labeling of de novo protein synthesis in an ISL1–2+ MN. (B) Dot plot displaying the level of HPG incorporation in MNs derived from 3 control and 3 C9-ALS iPSC lines. Kruskal-Wallis test. (C) Dot plot displaying the level of HPG incorporation in MNs transfected with scrambled (siScr) or ETF1 siRNAs. MNs derived from 1 control and 2 C9-ALS iPSC lines. Kruskal-Wallis test; ns=not significant. (D) Representative confocal images of healthy control and C9-ALS patient MNs immunolabeled for total UPF1 (green), pUPF1 (red), and DNA (blue). (E) Dot plot displaying the fold change in pUPF1/UPF1 signal in control and C9-ALS MNs. (Mann-Whitney U test. (F) Pie charts displaying NMD activation levels defined by pUPF1/UPF1 quartiles (Q1–4) observed in control (top) and C9 (bottom) MNs. (G) Top: Representative WB for SMG1 in MN cultures derived from 3 control and 3 C9-ALS iPSC lines. GAPDH was used as a loading control. Bottom: bar plots showing the fold change in the SMG1/GAPDH ratio in control and C9-ALS MN cultures. T test. (H) Left: Representative confocal images of C9-ALS patient MNs transfected with scrambled or ETF1 siRNAs and immunolabeled for total UPF1 (green), pUPF1 (red), and DNA (blue). Right: dot plot displaying the fold change in pUPF1/UPF1 signal in C9-ALS MNs transfected with scrambled or ETF1 siRNAs. n=3 independent differentiations; Mann-Whitney U test. In graphs, empty-filled circles represent individual cells (B, C, E, H) or biological replicates (G). Bars represent the mean ± SEM and dotted lines mark the mean level in control samples (B, C, G, H) or Q1–4 quartile limits (E). Scale bars: 25 (D, H), 20 (A) μm.

Figure 7.. C9orf72 mRNA is Targeted for Degradation by NMD.

(A) Dot plot displaying the fold change in the N/C ratio of eRF1 signal quantified in C9-ALS MNs with (+) or without (−) C9-HRE RNA foci. n=2 independent differentiations; Mann-Whitney U test; p*<0.05; **<0.01; ***<0.001. (B) Representative FISH+ICC confocal images of C9-ALS patient MNs transfected with scrambled or ETF1 siRNA and immunolabeled for G₄C₂ RNA foci (green), MAP2 (red) and DNA (blue). Dashed lines mark the nuclear boundary and dotted lines mark the neuronal soma. (C) Bar plots showing the number of nuclear (bottom) and cytosolic (top) C9-HRE RNA foci per cell in MNs derived from 1 control and 3 C9-ALS iPSC lines, transfected with scrambled or ETF1 siRNAs. Wilcoxon signed rank. (D) Representative FISH+ICC confocal images of healthy control and C9-ALS patient MNs transfected with scrambled or ETF1 siRNAs, and immunolabeled for poly(A) RNA (green), MAP2 (red) and DNA (blue). Dashed lines mark the nuclear boundary and dotted lines mark the neuronal soma. (E) Dot plot showing the N/C ratio of poly(A) RNA in control and C9-ALS MNs transfected with scrambled or ETF1 siRNAs. n=2 independent differentiations; Kruskal-Wallis test, p=ns: not significant; **<0.01, ****<0.0001;. (F) Left graph: Bar plots showing fold change in the level of C9orf72-intron retention in MN cultures derived from 3 control and 3 C9-ALS iPSC lines as measured by semi-quantitative RT-PCR. Bar plots comparing intron retention in C9-ALS MNs transfected with scrambled (Scr) vs. ETF1 (middle graph) or vs. UPF1 (right graph) siRNAs. Mann-Whitney (left and right graphs), and t test (middle graph). (G) Representative images of fly eyes from wild-type (GMR-GAL4 x EGFP) or C9-HRE (C9×30 or x36) mutant flies, with endogenous levels (EGFP), knockdown (UPF1-RNAi) or overexpression (UPF1-OE) of UPF1. Arrow heads indicate the presence of necrotic patches. (H) Bar plots displaying the level of eye degeneration in flies carrying C9-HRE (C9×30 or x36 repeats) with endogenous levels, knockdown (UPF1-RNAi) or overexpression (UPF1-OE) of UPF1. The basal level of eye degeneration in C9×30 or x36 flies is shown in yellow and orange respectively; suppression and enhancement of C9-HRE toxicity is shown in green and red, respectively. Mann-Whitney U test. (I) Dot blot of poly-GR in C9×36, and C9×36 flies with decreased eRF1 (ETF1-RNAi) or UPF1 (UPF1-RNAi). Tubulin was used as a loading control. In graphs, empty-filled circles represent individual cells (A, C, E), biological replicates (F) or analyzed flies (H). Bars represent the mean ± SEM, and dotted lines mark the mean levels in control/reference samples. Scale bars: 10 μm.