Identification of functional non-coding variants affecting Alzheimer’s disease risk by Massively Parallel Reporter Assay

The majority of genetic risk variants for late-onset Alzheimer’s disease (LOAD) reside within non-coding genomic regions, suggesting they exert pathogenic effects by disrupting transcriptional regulatory programs. To systematically identify functional variants, we performed an unbiased, high-throughput Massively Parallel Reporter Assay (MPRA) in the THP-1 human monocytic cell line, screening 2,231 SNPs across 24 LOAD GWAS loci. We identified 62 variants exhibiting significant allele-specific transcriptional regulatory output, including rs636317 in the MS4A locus. The risk allele of rs636317 disrupts a CTCF binding motif, altering fine-scale chromatin looping. To validate these findings, we employed CRISPR/Cas9 to generate an allele-specific deletion of the rs636317 T allele in H9 human embryonic stem cells. Monocytes differentiated from these edited cells displayed increased expression of MS4A4E , MS4A6A , and TREM2 , along with highly elevated levels of soluble TREM2 (sTREM2). These results provide a mechanistic link between a non-coding genetic variant, transcriptional regulation of the MS4A family, and TREM2 -mediated immune responses in Alzheimer’s disease pathogenesis.