Abstract
Loss of nuclear TDP-43 is a hallmark of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). Although TDP-43 is known to regulate RNA processing, including repression of cryptic exons, we currently lack a systems-level understanding of the consequences of TDP-43 loss. To address this, we generated multiomic datasets, including RNA-seq and proteomics, from human iPSC-derived neurons depleted of TDP-43. We found that differentially spliced genes, many expressing cryptic exons, had the greatest protein reductions. Surprisingly, nearly half of differentially expressed proteins were neither mis-spliced, nor differentially expressed genes; most of these also had no reported mis-splicing in seven additional post-mortem and iPSC-derived neuron datasets. Integrative network analysis identified a high-confidence disease-specific subnetwork of over 700 interacting proteins, enriched for mRNA processing, synaptic function, and autophagy. Comparison with post-mortem ALS and FTD samples revealed convergent protein and pathway disruptions. We experimentally validated network-predicted effects of cryptic splicing in ATG4B, STMN2, and DAPK1. Our analyses reveal new TDP-43-dependent molecular cascades and nominate central genes as potential ALS/FTD therapeutic targets.