Expression quantitative trait loci (eQTL) analysis in a diverse Alzheimer disease cohort reveals ancestry-specific regulatory architectures.

Abstract

Generalization of Alzheimer disease (AD) genetic findings to ancestrally admixed populations, such as African Americans (AA) and Caribbean Hispanics from Puerto Rico (PR), is challenging as they are underrepresented in most studies. There is evidence of differing underlying genetic architecture of AD in these groups, but identification of associated DNA variants is just a first step in understanding underlying biology. We evaluated the gene regulatory architecture by expression quantitative trait loci (eQTL) analysis within and across ancestries in a diverse cohort. We performed RNAseq from peripheral blood and genotyping with the Illumina Global Screening Array in 537 individuals: 241 non-Hispanic Whites (NHW) (121 AD, 120 cognitively intact controls), 232 AA (115 AD, 117 controls) and 64 PR (34 AD, 30 controls). All individuals had phenotypic adjudication for consensus AD status and were over the age of 65 at last examination and blood draw. We ran eQTL analysis overall and within each ancestry accounting for age and sex. Results were filtered for variant-gene pairs with p-values ≤ 1x10-5. We identified 1846 eQTLs at 737 genes, 2383 eQTLs at 791 genes, and 979 eQTLs in 459 genes in AA, NHW and PR, respectively. 603 eQTLs (∼30%) were significant across all ancestries with the same target gene and direction of effect. Interestingly, >70% of the PR eQTLs were also significant in either or both of the AA and NHW. Among eQTLs in linkage disequilibrium (r2>=0.5) with a significantly associated marker from AD GWAS studies in NHW (Kunkle at el 2019) or AA (Kunkle et all 2020) were two specific to AA for HLA-DQB1 and C1QTNF4 and three to NHW for C2, CNN2, and PILRB. Analysis of case and control eQTL effects and the role of local ancestry in other admixed populations, including those with Amerindian ancestry, is ongoing. AA, NHW, and PR individuals have overlapping regulatory architecture that likely reflects the admixture in AA (20% European, 80% African) and PR (∼60% European, 20% African, and 20% Amerindian). These results underscore the importance of continuing to include diverse populations in AD genomics and functional studies to identify unique and shared architecture of AD.