TDP-43 pathology is linked to motor neuron loss but is independent of stress granules in vivo.

Nuclear depletion and cytoplasmic aggregation of TDP-43 are pathological hallmarks of several neurodegenerative diseases, including amyotrophic lateral sclerosis (ALS), frontotemporal dementia (FTD), Alzheimer's disease, and the recently defined limbic-predominant age-related TDP-43 encephalopathy (LATE). Chronic activation of the integrated stress response (ISR) and persistence of stress granules, phase-separated assemblies proposed to function as a protective mechanism, have been hypothesized to initiate the formation of pathological TDP-43 inclusions observed in post-mortem neurons. However, recent clinical trials targeting the ISR and stress granule dissolution failed to demonstrate clinical benefit despite robust target engagement, calling this model into question. Here, we employ a recurrent hyperthermia paradigm to directly examine the relationship between stress granules and TDP-43 pathology in vivo. We find that RNA-binding proteins classically associated with stress granules persist as phase-separated cytoplasmic structures in spinal motor neurons of both non-transgenic and mutant TDP-43 mice. Importantly, these structures are reversible and spatially distinct from TDP-43 puncta. Moreover, in a mutant TDP-43 mouse model with an impaired acute stress granule response, stress exposure induces TDP-43 nuclear export and cytoplasmic accumulation. Recurrent stress in these mice leads to a selective loss of spinal α-motor neurons. Together, our findings demonstrate that TDP-43 nuclear clearance and cytoplasmic demixing occur independently of stress granules in vivo, challenging prevailing models of TDP-43 pathogenesis and highlighting important implications for therapeutic strategies targeting the ISR.