Finding. Cicardi, Trotti, Tigano and colleagues put the ALS allele TDP-43 A315T into the endogenous locus of human cortical neurons and hung a Dendra2 tag on it. The cells lose neurites, fire wrong, and respire worse. They pile autophagy-dependent cytoplasmic aggregates. Nuclear splicing of the usual cryptic-exon targets still works until someone forces the protein out of the nucleus. Stress granules are already high and go higher with oxidants. The RNA chaperone Clip34 cuts aggregation and granule membership.
Why mitochondria are in the first paragraph
The respiration hit arrives while nuclear function is still standing. That is a gain-of-function timeline, not a TDP-43-left-the-nucleus-then-everything-died timeline. ALS mitochondrial papers that assume cryptic-exon loss as the start will mis-order this model.
Score 78. Human allele, respiration, a clean nuclear-versus-cytoplasmic split, a tool compound. Confidence is medium: cortical not motor, no named respiratory complex.
What to do with it If you time ALS mitochondria against TDP-43, start with this gain-of-function window. Test whether Clip34 returns oxygen consumption. Do not call it a cryptic-exon-first model.