Abstract
Late-onset Alzheimer's disease (AD) arises in part from a complex genetic architecture dominated by common, low-penetrance variants, many of which are enriched in glial cells and remain mechanistically unresolved. Unlike the rare coding mutations that contribute to early-onset AD, these common variants often lie in noncoding regions, complicating efforts to link genetic risk to cellular function. Emerging evidence suggests that many glial-enriched risk genes contribute to disease by disrupting communication between glia and neurons. Such interactions are essential for preserving synaptic health and modulating immune responses to pathology. Understanding how polygenic variation perturbs these pathways requires integrative strategies that combine large-scale postmortem brain datasets with experimentally tractable human cellular models. In this review, we highlight recent progress in decoding the cellular impact of AD risk variants through the lens of glial-neuronal communication. We first illustrate how human brain studies have mapped cell-type-specific gene expression and intercellular networks associated with genetic risk. We then discuss how human stem cell-derived co-culture and 3D models are being used to test these hypotheses in controlled experimental systems. As a case study, we focus on CLU (Clusterin), a well-replicated risk locus that modulates glial inflammation, lipid exchange, and neuronal vulnerability. Together, these studies build a scalable, human-centric framework for linking genotype to function and point toward new opportunities for therapeutic discovery rooted in intercellular biology.