Abstract / Summary
Advances in transcriptomics have transformed our understanding of amyotrophic lateral sclerosis (ALS), a progressive neurodegenerative disease, revealing disrupted gene expression profiles and highlighting the multi-system biology of ALS. Despite major advances, transcriptomic studies have only begun to capture the complexity and the molecular hierarchy of transcriptomic alterations in ALS. To resolve and characterize the transcriptome in ALS, we performed a comprehensive reanalysis of bulk RNA sequencing from the New York Genome Center ALS Consortium cohort across five post-mortem tissues including motor and frontal cortex, cervical and lumbar spinal cord, and cerebellum. By deploying dual analytical pipelines - one reference-based to model canonical events and one de novo to detect transcript structural novelties - we disentangled the quantitative and qualitative architectures of ALS. Our reference-based analysis revealed that the ALS transcriptome is defined primarily by splicing failure rather than changes in gene expression. Aberrant splicing events, particularly intron retention, outnumbered differentially expressed genes by an order of magnitude. This widespread loss of fidelity disproportionately affected RNA-binding proteins (RBPs), suggesting a collapse in their autoregulatory feedback loops. Deconvolution of these signals identified distinct cellular vulnerabilities: transcriptional disruptions were enriched in glial cells in sporadic cases but in neuronal cells in C9ORF72-positive cases. Furthermore, we observed sex-specific dysregulation, with male individuals exhibiting greater disruption in guanosine triphosphatase signaling and ciliary organization pathways. In parallel, our de novo analysis uncovered a significant burden of ALS-specific RNA fusions that were absent in controls. Whole-genome sequencing of the same individuals, together with a larger reference population, confirmed that disease-specific fusions do not arise from genomic structural variants, indicating a transcriptional rather than genomic origin. PacBio long-read Iso-Seq in an independent post-mortem cerebellar cohort confirmed multiple ALS-specific fusions identified by short-read sequencing. Similarly, a junction-specific BaseScope assay detected one of the fusion transcripts, FAM69A-EVI5, in human ALS cerebellum, providing orthogonal evidence for the endogenous presence of an ALS-associated fusion transcript. ALS-specific fusion junctions also displayed a dramatic depletion of RBP motifs relative to canonical splice junctions, and the presence of these sparse disease-specific fusions was strongly correlated with severe splicing outliers in genes governing guanosine triphosphatase activity. Altogether, our results delineated a transcriptome characterized by aberrant splicing with tissue- and sex-specific changes and identified structural-variant-independent RNA fusions as an additional feature of RNA-processing dysregulation in ALS. This integrated view provides a mechanistic scaffold for the development of splicing-centered and RNA-structural therapeutic strategies for ALS.