Associating cognitive reserve and brain‐wide gene expression regarding Alzheimer's disease

Background Cognitive reserve (CR) is susceptibility to brain damages, which impacts the development of Alzheimer’s disease (AD). Whereas CR is associated with certain subnetworks of brain connectivity, little research has been done on identifying genes that contribute to these subnetworks. In this study, we linked these subnetworks to regional variation in brain‐wide gene expression. We utilized Allen Human Brain Atlas (AHBA) in that it comprises gene expression data spanning the whole brain. Method We extracted a binary subnetwork regarding CR from AD patients. Then we drew out gene expression patterns in the subnetwork’s regions of interest (ROIs) and constructed a gene network through Weighted Gene Co‐expression Network Analysis (WGCNA). To research the distinctive trait of this sub‐region gene network (SGN), we also constructed a whole‐brain gene network (WGN). We detected modules (clusters of highly interconnected genes) of each network and identified their hub genes (genes with high intra‐modular connectivity). We sought non‐preserved, unique modules in SGN compared to WGN using the median rank statistic. Finally, the biological process ontology of the hub genes was investigated by functional enrichment analysis. Result We constructed gene networks based on gene expression data of 10544 genes. WGN consisted of 90 nodes of whole‐brain regions and SGN consisted of 49 nodes of the subnetwork ROIs. While 18 modules were detected in SGN, modules yellow, magenta, red and black were identified as unique modules. Module magenta included novel co‐expressed gene clusters, and the other modules combined genes that were classified as different modules in WGN. Apoptotic process, including apoptosis‐regulation and neuron death, was enriched distinctively in the 4 non‐preserved modules. Neuronal death due to apoptosis is a common pathological feature in the brains of AD patients. Conclusion We identified brain‐wide gene co‐expression modules that have unique specific connectivity patterns related to the CR subnetwork. This non‐preservation highlights the relevance of CR and gene expression patterns. Moreover, the modules were enriched for AD risk genes, further demonstrating the functional context of CR in AD. This study will provide the groundwork for novel insights and extensive research to associate brain‐wide gene expression and connectome‐wide phenotype regarding AD.