Mutations in presenilin-1 (PSEN1) and presenilin-2 (PSEN2) cause the majority of familial Alzheimer's disease (FAD) cases, and their discovery entrenched the amyloid cascade hypothesis as the field's dominant framework for over thirty years. The prevailing interpretation holds that FAD mutations impair gamma-secretase proteolytic activity, shifting cleavage of amyloid precursor protein toward longer, aggregation-prone Aβ42 fragments, thereby initiating neurodegeneration. This misinterpretation has guided decades of therapeutic development, produced dozens of failed clinical trials, and cost an estimated 42. 5 billion without achieving disease modification (Cummings et al. , 2022). We argue here that FAD mutations cause neurodegeneration not by disrupting gamma-secretase proteolytic activity, but by disrupting two parallel classes of evolutionarily ancient, gamma-secretase-independent non-proteolytic function: a lysosomal-autophagic arm governing lysosomal acidification, autophagic flux, and protein aggregate clearance; and an ER-mitochondrial calcium arm governing calcium homeostasis between the ER and mitochondria. These two tracks operate through distinct primary mechanisms and are supported by distinct lines of evidence, but converge on the same downstream outcome: proteostatic failure; a calcium-dependent secondary pathway additionally links the two tracks at the level of autophagic flux, with nuclear calcium deficiency reducing Sestrin2 expression and constitutively locking mTORC1 to the lysosomal membrane, silencing CLEAR network-driven autophagy in human FAD neurons and patient fibroblasts independently of the direct V0a1 lysosomal acidification mechanism. Two converging lines of evidence are used to support this hypothesis. First, in Dictyostelium discoideum, a social amoeba separated from humans by approximately one billion years of evolution, presenilin orthologs are required for lysosomal acidification, autophagic flux, and protein aggregate clearance in an organism without Aβ, APP, and vertebrate nervous system; crucially, catalytically inactive presenilin, incapable of any proteolytic cleavage, fully rescues every one of these phenotypes, demonstrating that the conserved functions are structurally independent of proteolysis (Sharma et al. , 2019). Second, FAD-equivalent mutations in C. elegans that preserve full gamma-secretase/Notch signaling capacity nevertheless cause mitochondrial calcium overload and neuronal dysfunction, genetically separating the two functions and demonstrating that non-proteolytic PSEN disruption causes neuronal pathology in the complete absence of Aβ (Ashkavand et al. , 2025). Independently consistent with this hypothesis, rare coding variants in PSEN1, including known pathogenic FAD mutations, are enriched in Parkinson's disease patients at an odds ratio of 54. 2 (95% CI 18. 8-156. 1), constraining any account of PSEN pathogenicity to mechanisms that operate beyond amyloid-specific disease. Together, these lines of evidence demonstrate that PSEN dysfunction causes neurodegeneration through mechanisms that cannot be reduced to Aβ42/Aβ40 ratio shifts. This reframing has direct therapeutic consequences. The correct targets are not Aβ secretases but rather the ancient PSEN-dependent machinery: lysosomal acidification (v-ATPase), ER-mitochondria calcium homeostasis, and autophagic flux. We outline the mechanistic evidence, explain why gamma-secretase-centric therapy was mechanistically wrong, and propose a research agenda focused on proteostasis restoration as the path toward disease modification in FAD.
Arthur Stewart (Thu,) studied this question.