Proteomic comparison of human neural cell-derived extracellular vesicles and parental cells from Alzheimer's disease and cognitively normal individuals.

Circulating brain-derived extracellular vesicles (BDEVs) have emerged as promising biomarkers for neurodegenerative diseases, including Alzheimer's disease (AD). However, it remains unclear to what extent extracellular vesicles (EVs) proteomes reflect the molecular states and disease-associated alterations of their parent brain cell types. Here, using a multi-line human induced pluripotent stem cell (hiPSC) platform derived from three AD and three cognitively normal (CN) donors, we generated neurons, astrocytes, microglia, and oligodendrocytes, and performed paired proteomic profiling of each cell type and its secreted EVs. We systematically compared protein profiles to evaluate cell-EV similarity, disease-associated features, and concordance with proteomic datasets from human AD brain tissue. Across all four lineages, EV proteomes showed extensive overlap with parent cells (>97% overlap; Jaccard index: 0.69-0.80) while also displaying lineage-specific functional biases. Under AD versus CN comparisons, EVs exhibited larger effect sizes and retained a higher number of differentially expressed proteins (DEPs) when applying the same fold-change criteria, yielding clearer AD-CN separation than their parent cells. Importantly, EV DEPs showed higher concordance with human AD brain proteomic signatures (EVs: 2,134 DEPs; cells: 816 DEPs). Finally, amyloid precursor protein (APP)-derived peptides, including amyloid-β (Aβ), were preferentially enriched in neuron- and oligodendrocyte-derived EVs, and AD EVs showed elevated Aβ42, p-Tau217 and p-Tau181 relative to CN EVs. Together, these data indicate that cell type-resolved EV proteomes largely recapitulate parent cell identity while sensitively capturing AD-relevant molecular alterations, supporting EV-based strategies for early diagnosis and monitoring of AD and potentially other neurodegenerative disorders.