Integrative genomics and single-cell CRISPRi screening dissect Alzheimer GWAS non-coding variants regulating TSPAN14

Genome-wide association studies (GWASs) have uncovered many associations for human complex diseases, but functional dissection of the discovered loci has lagged behind. We present a variant-to-gene (V2G) mapping effort for Alzheimer disease (AD) leveraging the most recent AD GWAS meta-analyses. In this study, we integrated ten brain-relevant genomics datasets-including promoter Capture C, ATAC-seq, and RNA-seq from microglia, neurons, and astrocytes-to fine-map AD GWAS variants and identify effector genes. We then performed a single-cell CRISPRi Perturb-seq screen targeting 74 candidate regulatory regions in the human microglial cell line HMC3. Our V2G mapping effort identified 93 candidate causal variants and 94 effector genes (72 coding) for 35 AD loci. Our CRISPRi screen across ∼97,000 cells validated 21 variant-gene pairs. We showed that an intronic region at the TSPAN14 locus containing rs7080009, rs1870138, and rs1870137 is a microglial-specific enhancer activated by the AD-risk haplotype. CRISPR-mediated deletion of this region reduced TSPAN14 expression, disrupted cell-adhesion pathways, and lowered secretion of pro-inflammatory cytokines interleukin 6 (IL-6) and IL-8. Our study provides a systematic framework for mapping GWAS signals to effector genes in a cell-type-specific manner and identifies robust leads for in-depth functional investigations.