Novel Alzheimer risk genes determine the microglia response to amyloid-β but not to TAU pathology

Background: Thousands of SNPs associated with risk of Alzheimer9s disease (AD) in genome-wide association studies (GWAS) do not reach genome-wide significance. When combined, they contribute however to a highly predictive polygenic risk score. The relevance of these subthreshold risk genes to disease, and how their combined predictive power translates into functionally relevant disease pathways, is unknown. We investigate here at the genome-wide level and in an unbiased way to what extent AD risk genes show altered gene expression in the context of increasing amyloid-β or Tau pathology in mouse models of AD. Methods: We used an existing GWAS data set to generate lists of candidate AD genes at different levels of significance. We performed transcriptomic analysis on wild-type and transgenic APP/PS1 (APPtg) and Thy-TAU22 (TAUtg) mouse models at early and late stage of disease. We used unbiased weighted gene co-expression network analysis (WGCNA) to identify clusters of co-regulated genes responsive to amyloid-β or TAU pathology. Gene set enrichment was used to identify clusters that were enriched for AD risk genes. Findings: Consistent and significant enrichment of AD risk genes was found in only one out of 63 co-expression modules. This module is highly responsive to amyloid-β but not to TAU pathology. We identify in this module 18 AD risk genes (p-value=6.5e-11) including 11 new ones, GPC2, TREML2, SYK, GRN, SLC2A5, SAMSN1, PYDC1, HEXB, RRBP1, LYN and BLNK. All are expressed in microglia, have a binding site for the transcription factor SPI1 (PU.1), and become significantly upregulated when exposed to amyloid-β. A subset regulates FC-gamma receptor mediated phagocytosis. Interpretation: Genetic risk of AD is functionally translated into a microglia pathway responsive to amyloid-β pathology. This insight integrates aspects of the amyloid hypothesis with genetic risk associated to sporadic AD.