Widespread dysregulation of mRNA splicing implicates RNA processing in the development and progression of Huntington’s disease
Description
In Huntington’s disease (HD), a CAG repeat expansion mutation in the HTT gene drives a gain-of-function toxicity that disrupts mRNA processing. Although widespread dysregulation of gene splicing has been shown in human HD post-mortem brain tissue, post-mortem analyses are likely confounded by cell type composition changes in late stage HD, limiting the ability to identify dysregulation related to early pathogenesis. To study alternative splicing changes in early HD, we performed RNAsequencing analyses in an established isogenic HD neuronal cell model. We report cell type-associated and CAG length-dependent splicing changes, and find an enrichment of RNA processing genes coupled with neuronal function-related genes showing mutant HTT associated splicing. Comparisons with post-mortem data also identified splicing events associated with early pathogenesis that persist to later stages of disease, particularly in the striatum. Here, our results highlight splicing dysregulation in RNA processing genes in HD, leading to potential disrupted neuronal function and neuropathology.
Files
Steps to reproduce
Differentially splicing analysis was performed using the LeafCutter package. In brief, a unified splice site junction database was first generated by summarising sample junction read counts using bedtools. Splice junctions were then annotated using the Ensembl hg38 genome assembly and gtf as reference. Only splice junctions that have a splice site base sequence that are not repeat-masked and were identified in at least 3 samples were retained for downstream analysis. Splice junction clustering and read counting were performed using the `leafcutter_cluster_regtools.py` script with the `-C` option to include constitutively spliced junctions. Differential splice junction usage was then analysed using the `leafcutter_ds.R` script with option `--min_coverage=10` and the sample replicate number included as a confounding factor to account for batch effects. All pairs of biological groups-of-interest (cell line: NPC vs hESC; CAG mutant hESC: 45Q vs control, 81Q vs control; CAG mutant NPC: 45Q vs control, 81Q vs control; and CAG mutant Neuron: 45Q vs control, 81Q vs control) were tested independently. The NPC vs hESC comparison was performed by grouping all CAG length genotypes in each cell line.