Single-cell eQTL mapping reveals convergent glial–neuronal risk architecture in Parkinson’s disease

Synucleinopathies affect ∼15 million people and are classically divided into neuronal (Parkinson’s disease (PD), dementia with Lewy bodies) and glial (multiple system atrophy) disorders. Here we challenge this dichotomy. We functionally fine-map 90 PD GWAS signals across nine cell types in cortex and substantia nigra using disease-context, population-scale single-nucleus eQTL meta-analysis (N = 1,197), bulk brain eQTL analysis (N = 1,182), and Mendelian randomization. A stringent causal framework integrates single-nucleus allelic imbalance (snASE) with orthogonal validation. We identify 125 functional risk genes for 50 loci—nearly doubling supported genes—and assign genes and cell types to over half of GWAS signals. Unexpectedly, 51% of risk genes are regulated in glia, particularly oligodendrocytes and their precursors. Across cell types, risk converges on a shared glial–neuronal vesiculopathy network. These findings uncover a convergent glial-neuronal risk architecture and establish a single-cell atlas for context-aware gene discovery and precision therapeutics for PD.