Abstract
The genetics of late-onset Alzheimer's disease (AD) is complex due to the heterogeneous nature of the disorder. APOE*4 is the strongest genetic risk factor for AD. Genome-wide association studies have identified more than 30 additional loci, each having relatively small effect size. Known AD loci explain only about 30% of the genetic variance, and thus much of the genetic variance remains unexplained. To identify some of the missing heritability of AD, we analyzed whole-exome sequencing (WES) data focusing on non-APOE*4 carriers from two WES datasets: 720 cases and controls from the University of Pittsburgh and 7,252 cases and controls from the Alzheimer's Disease Sequencing Project. Following separate WES analyses in each dataset, we performed meta-analysis for overlapping markers present in both datasets. Among the four variants reaching the exome-wide significance threshold, three were from known AD loci: APOE/rs7412 (odds ratio (OR) = 0.40; p = 5.46E-24), TOMM40/rs157581 (OR = 1.49; p = 4.04E-07), and TREM2/rs75932628 (OR = 4.00; p = 1.15E-07). The fourth significant variant, rs199533, was from a novel locus on chromosome 17 in the NSF gene (OR = 0.78; p = 2.88E-07). NSF was also significant in the gene-based analysis (p = 1.20E-05). In the GTEx data, NSF/rs199533 is a cis-eQTL for multiple genes in the brain and blood, including NSF that is highly expressed across all brain tissues, including regions that typically show amyloid-β accumulation. Further characterization of genes that are affected by NSF/rs199533 may help to shed light on the roles of these genes in AD etiology.