Common and rare variant analyses implicate JARID2 in cerebral tau deposition

Genetic research on Alzheimer's disease (AD) has primarily focused on amyloid-β (Aβ), with fewer studies exploring tau pathology. We performed common variant GWAS on tau-PET SUVRs from A4 (n = 311 preclinical AD) and ADNI (n = 375 across diagnostic groups) cohorts. We complemented this with locus-specific rare variant analyses in 1561 individuals across five cohorts. Genetic findings were evaluated using circulating plasma proteins from the UK Biobank Pharma Proteomics Project (n = 54,129). Polygenic risk scores (PRS) for tau and amyloid-SUVR were tested for association with AD. GWAS identified two loci: rs78636169 (P = 5.76 × 10^-10) in JARID2 and rs7292124 (P = 2.20 × 10^-8) near ISX. Rare-variant analysis in JARID2 revealed chr6:15257832:A:G (P = 7.08 × 10^-05) in harmonized analysis and chr6:15492808:C:T (P = 1.65 × 10^-09) in rare-variant meta-analysis. Pleiotropy analyses suggested limited overlap between tau- and amyloid-related genetic signals. Gene-based analysis highlighted JARID2, a component of the PRC2 multi-protein complex. Mendelian randomization analysis identified LRRFIP1, a protein that binds with PRC2, as potentially causally linked to tau pathology. Amyloid-PRS, but not tau-PRS, was associated with AD clinical status, with age-dependent effects in APOE-ε4 carriers. Leveraging both GWAS and a large rare-variant cohort, we identified JARID2 as a candidate gene associated with tau pathology and observed patterns consistent with partially distinct roles of Aβ and tau in AD progression.