Abnormal RNA stability in amyotrophic lateral sclerosis

Abstract

Amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) share key features, including accumulation of the RNA-binding protein TDP-43. TDP-43 regulates RNA homeostasis, but it remains unclear whether RNA stability is affected in these disorders. We use Bru-seq and BruChase-seq to assess genome-wide RNA stability in ALS patient-derived cells, demonstrating profound destabilization of ribosomal and mitochondrial transcripts. This pattern is recapitulated by TDP-43 overexpression, suggesting a primary role for TDP-43 in RNA destabilization, and in postmortem samples from ALS and FTD patients. Proteomics and functional studies illustrate corresponding reductions in mitochondrial components and compensatory increases in protein synthesis. Collectively, these observations suggest that TDP-43 deposition leads to targeted RNA instability in ALS and FTD, and may ultimately cause cell death by disrupting energy production and protein synthesis pathways.

Fig. 1

RNA destabilization in ALS fibroblasts. a Schematic of Bru-seq (top) and BruChase-seq (bottom). BrU Bromouridine, Anti-BrU antibodies that recognize BrU. Example traces of RNA transcripts destabilized (b) or stabilized (c) in C9ALS fibroblasts. Blocks and lines denote gene and transcript structure, respectively. + strand genes are in green, – strand genes are in red. The traces represent RNA abundance at 0.5 h (top) and at 6 h (bottom) following Bru labeling. RPKM reads/kilobase of transcript/million mapped reads. Representative examples of RNA transcripts destabilized (d) or stabilized (e) in sALS fibroblasts. Scatter plot (f) and gene ontology (g) for 333 transcripts showing a change in stability ≥1.5-fold in C9ALS fibroblasts, in comparison to control (Cntl) cells. Stability is calculated as a ratio of transcript abundance at 6 h vs. 0.5 h. Scatter plot (h) and gene ontology (i) for 324 transcripts showing a change in stability ≥1.5-fold in sALS fibroblasts, compared to control cells. FDR false discovery rate. Cell lines used for these experiments are listed in Supplementary Table 1

Fig. 2

Abnormal RNA synthesis in ALS fibroblasts. Examples of RNA transcripts showing reduced (a) and increased (b) synthesis in C9ALS fibroblasts. The traces represent RNA abundance 0.5 h following Bru labeling. Representative schematics of RNA transcripts exhibiting reduced (c) and increased (d) synthesis in sALS fibroblasts. e Volcano plot for 65 transcripts showing significant changes in synthesis in C9ALS fibroblasts, in comparison to control (Cntl) cells. f Gene ontology was performed for transcripts displaying reduced synthesis in C9ALS fibroblasts. FDR false discovery rate. g Volcano plot for 28 transcripts exhibiting significant changes in synthesis in sALS fibroblasts, compared to control cells. Dotted lines in e and g depict an adjusted p value of 1 and fold change (FC) of 1.5. Cell lines used for these experiments are listed in Supplementary Table 1

Fig. 3

Conserved patterns of RNA destabilization in C9ALS iPSCs. Scatter plot (a), gene ontology (b), and STRING analysis (c) for the 956 transcripts showing changes in stability ≥1.5-fold in C9ALS iPSCs, in comparison to control (Cntl) cells. Higher resolution views for boxed areas in c are shown in d and e. Scatter plot (f) and gene ontology (g) for 865 transcripts with altered stability ≥1.5-fold in sALS iPSCs, in comparison to control cells. h Gene ontology analysis of RNA transcripts showing ≥1.5-fold reduction in C9ALS iPSCs vs. controls, acquired from the NeuroLINCS database. FDR false discovery rate. Cell lines used for these experiments, performed in duplicate, are listed in Supplementary Table 1. i−k Quantitative (q)RT-PCR in patient postmortem tissue, showing reduced abundance of RNAs related to mitochondrial oxidative phosphorylation (COX6B, COX6C-X1, COX5B, NDUFA1, NDUFA13) and the ribosome (RPL28, RPL38, RPS18) in C9ALS spinal cord (i), C9FTD frontal cortex (j), and sALS spinal cord (k). Graphs in i−k depict mean ± standard error. *p < 0.05, #p < 0.01, two-way ANOVA with Sidak’s multiple comparison test. All data in i−k were assembled from ≥3 biological replicates. Supplementary Table 3 lists the number of samples in each condition and clinical characteristics for each patient

Fig. 4

TDP-43 overexpression recapitulates RNA instability in control iPSCs. a Exogenous overexpression of EGFP and TDP-43 fused to EGFP (TDP43-EGFP) in control (Cntl) iPSCs. Scale bar, 100 µm. b Scatter plot for 1330 transcripts showing altered stability ≥1.5-fold in TDP43-EGFP overexpressing iPSCs, in comparison to EGFP-expressing cells, performed in triplicate. Gene ontology (c) and STRING analysis (d, excluding ubiquitin) for transcripts demonstrating altered stability in TDP43-EGFP overexpressing iPSCs. e−g Higher resolution views corresponding to boxed areas in d. h 180 transcripts were commonly destabilized in C9ALS and TDP43-EGFP-expressing cells. Among these transcripts, the ribosome and oxidative phosphorylation pathways were highly enriched by gene ontology. FDR false discovery rate

Fig. 5

Reduced abundance of mitochondrial proteins in C9ALS iPSCs. a Of the oxidative phosphorylation proteins detected by MS (n = 61), 52% were significantly reduced in C9ALS iPSCs, and 16% were similarly reduced in sALS iPSCs compared to controls (Cntl). b Fold change in each of 61 oxidative phosphorylation proteins. c Cumulative change in the abundance of oxidative phosphorylation proteins in ALS iPSCs. d 45% and 3% of the 95 cytoplasmic ribosomal proteins detected by MS in ALS iPSCs were significantly increased in C9ALS and sALS iPSCs, respectively, compared to controls. e Fold change for the 95 cytoplasmic ribosomal proteins in ALS iPSCs. f Cumulative change in the abundance of cytoplasmic ribosomal proteins in ALS iPSCs. g 72 separate mitochondrial ribosomal proteins were detected by MS in ALS iPSCs; 69% and 3% were reduced in abundance in C9ALS and sALS iPSCs, respectively. h Fold change for the 72 mitochondrial ribosomal proteins in ALS iPSCs. i Cumulative change in the abundance of mitochondrial ribosomal proteins in ALS iPSCs. FDR false discovery rate. All experiments performed in triplicate, with two lines /condition. Plots in c, f, i show mean ± standard deviation. *p < 0.0001; #p < 0.05, one-way ANOVA with Benjamini−Hochberg correction for multiple observations

Fig. 6

Unbiased proteomics confirms mitochondrial protein deficit in C9ALS iPSCs. Gene ontology (a) and STRING analysis (b) for the 806 proteins reduced ≥10% in C9ALS iPSCs, demonstrating enrichment for components of oxidative phosphorylation, mitochondrial ribosome, and RNA degradation pathways. Higher magnification views of boxed clusters in b are shown in c−e. Gene ontology (f) and STRING analysis (g) for the 961 proteins increased ≥10% in C9ALS iPSCs, showing enrichment for components of the proteasome, amino acid biosynthesis, and RNA transport pathways. Higher magnification views of boxed clusters in g are shown in h−j. FDR false discovery rate. All experiments performed in triplicate with two lines/condition

Fig. 7

Correlating RNA stability and protein abundance in ALS iPSCs. Linear regression of RNA stability (as measured by the RNA stability index for each transcript, or the abundance at 6 h/0.5 h) and protein abundance determined by MS. A significant (p < 0.0001) association was detected between RNA stability and protein abundance in control, C9ALS and sALS iPSCs for all transcripts (top row, a−c) and for those involved in oxidative phosphorylation (Ox/phos, bottom row, g−i). However, no such relationship was identified for ribosomal protein-encoding transcripts and their corresponding proteins (middle row, d−f). Representative scatter plots are shown from one line each of control, C9ALS and sALS iPSCs; identical results were obtained upon examination of other lines. p value determined by extra sum-of-squares F test

Fig. 8

Mitochondria morphology and protein synthesis in ALS patient-derived cells. a Mitochondrial morphology in control (Cntl) and C9ALS fibroblasts expressing mito-GFP. BF brightfield. Scale bars = 20 µm. Mitochondrial form factor (a measure of mitochondrial complexity, b) and aspect ratio (an estimate of circularity, c), or both (d) in fibroblasts expressing mito-GFP. Morphological analysis of mitochondrial form factor (e), aspect ratio (f), or both (g) in iPSCs stained with the mitochondrial dye TMRE. *p < 0.05, **p < 0.001 by one-way ANOVA with Tukey’s test. n > 20 cells/group. Results in b−d were combined from four lines each of control and C9orf72 fibroblasts, while e, f were combined from two lines each of control and C9orf72 iPSCs, performed in duplicate. Plots in b, c, e, f show median (horizontal line), interquartile range (box) and maximum/minimum (vertical lines). Graphs in d and g show mean ± standard error. h Bioenergetics analyses demonstrated greater reductions in oxygen consumption rate (OCR) upon addition of 900 nM FCCP, a decoupling agent, to C9ALS and sALS fibroblasts in comparison to controls. n = 8 (Control), 8 (sALS), and 7 (C9ALS) lines/group, as described in Supplementary Table 1. Plot in h shows mean ± 95% confidence interval. i iPSCs from controls and patients carrying C9orf72 mutations displayed elevated protein synthesis by SUnSET. *p < 0.01 by two-sided Kolmogorov−Smirnov test. Inset shows a scatter plot of normalized anti-puromycin counts (mean ± standard error) from control and C9orf72 mutant iPSCs. *p = 0.0129, unpaired t test. n = 2 lines each of control and C9ALS iPSCs, in three separate replicates. j Cumulative distribution function for fractional recovery of mCherry fluorescence in control and C9orf72 iPSCs at 3.5 h. *p < 0.01, two-sided Kolmogorov−Smirnov test. Inset illustrates a scatter plot of fractional recovery (mean ± standard error) in control and C9orf72 mutant iPSCs. *p < 0.01, unpaired t test. n = 2 lines each of control and C9ALS iPSCs, combined from three replicates