Genome-wide association studies (GWAS) have identified more than 75 genetic risk loci for Alzheimer's Disease (AD), however for a substantial portion of these loci the genetic variants or genes directly involved in AD risk remain to be found. A GWAS locus defined by the index SNP rs1476679 in ZCWPW1 is one of the largest AD loci as the association signal spans 56 potential risk genes. The three most compelling candidate genes in this locus are ZCWPW1, PILRA and PILRB, based on genetic, transcriptomic, and proteomic evidence. We performed amplicon-based target enrichment and next-generation sequencing of the exons, exon-intron boundaries, and UTRs of ZCWPW1, PILRA and PILRB on an Illumina MiSeq platform in 1048 Flanders-Belgian late-onset AD patients and 1037 matched healthy controls. Along with the single-marker association testing, the combined effect of Sanger-validated rare variants was evaluated in SKAT-O. No common variants (n = 40) were associated with AD. We identified 20 validated deleterious rare variants (MAF < 1%, CADD score ≥ 20), 14 of which in ZCWPW1. This included 4 predicted loss-of-function (LoF) mutations that were exclusively found in patients (P = 0.011). Haplotype sharing analysis revealed distant common ancestors for two LoF mutations. Single-molecule long-read Nanopore sequencing analysis unveiled that all LoF mutations are phased with the risk haplotype in the locus. Our results support the recent report for the role of ultra-rare LoF ZCWPW1 variants in AD and suggest a potential risk mechanism for AD through ZCWPW1 haploinsufficiency.

No abstract