Abstract
Abstract
TDP-43 nuclear depletion and cytoplasmic aggregation are hallmarks of frontotemporal lobar degeneration (FTLD) and motor neuron disease, yet the events preceding depletion remain poorly characterized. PML nuclear bodies and the nucleolus are phase-separated protein quality control (PQC) compartments; SUMO1 is integral to PML nuclear body assembly and accumulates in the nucleolus under proteostatic stress. How they remodel in relation to TDP-43 pathobiology has not been examined at single-cell resolution in human tissue. We analyzed 18,742 anterior cingulate cortex neurons across 29 subjects (10 sporadic FTLD-TDP, 10 C9orf72-associated FTLD-TDP, 9 controls) by multichannel confocal immunofluorescence, scoring each neuron for five nuclear body phenomena and TDP-43 nuclear depletion. Three phenomena were elevated in nuclear TDP-43-preserved patient neurons relative to controls and returned to approximately control levels in depleted neurons, consistent with a protective role (control → preserved → depleted): SUMO1 nucleolar granules (3.2% → 13.9% → 5.0%), SUMO1 nucleolar puncta (0.2% → 11.0% → 3.1%), and PML nucleolar contacts (2.5% → 6.4% → 1.7%). In contrast, PML nuclear membrane contacts (0.1% → 2.6% → 19.9%) and macro-PML bodies (0.0% → 0.9% → 15.6%) increased progressively, and both were associated with depletion (odds ratio 9.2, 95% CI 6.1–14.0 and 20.2, 95% CI 13.9–29.4), consistently across sporadic and C9orf72-associated disease. Within nuclear TDP-43-preserved neurons, protection-associated phenomena rarely co-occurred with PML nuclear membrane contacts (odds ratio 0.39–0.54, all p < 0.05) but co-occurred with each other (odds ratio 3.3–4.0), defining two segregated remodeling programs; depletion fell from 15.3% in neurons bearing no protection-associated phenomenon to 2.3% in those bearing all three (Kendall's τ = −0.135). These findings identify opposing nuclear body remodeling programs in FTLD-TDP, expand the spectrum of pre-depletion changes, implicate phase-separated nuclear PQC compartments as determinants of neuronal fate, and raise the possibility that SUMO1 nucleolar accumulation is a general marker of neuronal stress.