Alzheimer’s disease (AD) is a progressively debilitating neurodegenerative disease that exacts a heavy toll on our social and health care systems. AD is presumably caused by pathogenic variants in a large number of genes working in concert with environmental factors and personal behaviors. However, despite years of worldwide efforts, only a few dozen AD-associated genes have been identified to date. Employing the gene-to-gene damaging variant rate (g2gDVR) analysis of published datasets containing 181,388 subjects we discovered 2114 potentially AD-associated genes, of which 711 were linked to the higher AD prevalence among African-Americans. The DVRs of these genes in AD population (55,460 subjects found in NIAGADS) were at least two-fold higher than those in the general population (125,748 subjects in gnomAD) with false discovery rate less than 1x10-12. The advantage of the g2gDVR method for studies of disease is that it focuses on variants expected to impact gene function, in contrast to methods based on SNP analysis, such as GWAS, that are gene-agnostic. In combination with transcriptome profiling of 21 brain regions from 2728 human AD and control samples (collected from various GEO datasets) we identified three of the 711 genes (GNB5, SNAP91 and AKAP11) that were candidate causative genes based on their reduced expression in both hippocampus and entorhinal cortex regions of AD patient brains. In the context of the APP/PSEN1 mouse model of AD, we further confirmed that Gnb5 heterozygosity exacerbates the formation of both amyloid plaques and neurofibrillary tangles, the hallmarks of AD brain neuropathology. These results suggested that the g2gDVR analysis method is highly effective in identification of AD-associated genes and is expected to be broadly applicable to other polygenic diseases.
The Intramural Research Program of the National Institute of Diabetes and Digestive and Kidney Diseases (ZIA DK043304-24) supported this research.
