Work

Poison exon splicing in the human brain provides a new paradigm for understanding and targeting neurological disorders

Year: 2026

Type: article

Abstract: Poison exons are highly conserved exons whose inclusion creates premature termination codons and triggers nonsense-mediated decay (NMD) of the mature transcript. Despite their important role in post-transcriptional regulation, poison exons remain poorly annotated due to the lack of systematic transcriptome-wide approaches. A comparative analysis of 954 eukaryotic transcriptomes reveals that Homo Sapiens exhibits the highest enrichment of NMD-targeted isoforms. By systematically predicting and annotating splicing events that generate premature termination codons, we identify 9,814 poison exons in the human genome. Using RNA-seq datasets from GTEx and BrainSpan, we analyze poison exon splicing across tissues and developmental stages and identify 83 poison exons uniquely found in the human brain, 973 with brain-specific differential splicing, and 1,284 differentially spliced during brain development. By integrating ClinVar annotations with SpliceAI predictions, we identify 1,625 pathogenic variants associated with neurological disorders that are predicted to affect the splicing of 743 poison exons, and we functionally validate the impact of a subset of them using CRISPR prime-editing in human cells. Functional characterization of the intronic NDUFAF6 c.420 + 784C > T variant associated with Leigh syndrome demonstrates increased poison exon inclusion and a significant reduction in protein levels in iPSC-derived neurons, with consequent impaired mitochondrial membrane potential compared to controls. Together, these findings provide functional evidence that the c.420 + 784C > T variant promotes aberrant poison exon inclusion, leading to reduced NDUFAF6 protein expression and mitochondrial dysfunction. Our findings highlight poison exons as pivotal regulators of gene expression in the human brain and support their therapeutic targeting in neurological diseases.

Source: Genome biology

DOI: 10.1186/s13059-026-04292-8