An analysis of peripheral blood whole-genome transcriptome for identification of Alzheimer's disease-related genes

Abstract

Alzheimer’s disease (AD), the most common form of dementia, has pathophysiological characteristics, including amyloid plaques, cerebral amyloid angiopathy, and neurofibrillary tangles. Although several studies analyzing gene expression data have uncovered valuable patterns, most gene expression data were obtained from biopsy or autopsy-based samples, which are difficult to extrapolate to clinical settings. Blood gene expression profiling has been considered to be a useful surrogate for the brain transcriptomic signature. This dissertation discusses the blood AD-related genes and their association with the blood diabetes mellitus (DM)- and cardiovascular disease (CVD)-related genes. The first part of dissertation is the identification of the blood AD-related genes using statistical methods, protein-protein interaction (PPI), transcription factor (TF) databases, and the Convergent Functional Genomics (CFG) method. The blood AD-related genes using statistical methods and CFG generated the best performance for AD prediction, and were enriched with inflammation, mitochondria, Wnt signaling, and ubiquitin pathways. The second part is to identify shared blood transcriptomic signatures between AD and DM. Blood AD and DM gene expression datasets were combined and a co-expression network was used to construct modules consisting of genes with similar expression patterns. We selected one module related with both AD and DM status, where five genes were identified as dysregulated transcription factors using a gene regulatory network (GRN) based on Pearson’s correlation coefficient and PPI. The third part is to identify the shared disease-related signatures between AD and CVD. We have utilized statistical methods, PPI, TF, disease-gene relationship databases, and SNP that are related to disease status or RNA expression to identify several candidate diseaserelated gene (DRG) sets. We selected informative DRG sets that had a high accuracy for disease prediction. We then selected the conserved DRG sets by comparing them with cell type-specific DRGs, yielding the actual AD- and CVD-related genes, respectively. Using the GRN constructed, we identified two upstream genes. Collectively, the findings in the present dissertation suggested that the AD-related genes obtained from blood gene expression data are useful in predicting the AD classification, and may contribute to reveal common pathophysiology of AD with chronic diseases, such as DM and CVD