Brain multiomic profiling identifies tau-related transcriptomic dysregulation in Alzheimer’s disease

Abstract

Abstract Identification of gene expression changes in postmortem brain tissue of Alzheimer’s disease donors compared to controls has implicated numerous biological pathways for Alzheimer’s disease pathophysiology. Nonetheless, there is still limited understanding of how gene expression dysregulation underpins specific proteinopathies core to Alzheimer’s disease. Here, we investigate brain transcriptomic changes in a well-characterized cohort of Alzheimer’s disease donors to identify genes and networks that associate with Alzheimer’s disease endophenotypes, including neuropathology measures (Braak stage, Thal phase and cerebral amyloid angiopathy score) and Alzheimer’s disease-related brain protein levels (Apolipoprotein E, Amyloid-β 40, Amyloid-β 42, tau and phospho-Tau). Bulk transcriptome measures were collected from the temporal cortex tissue of 477 Alzheimer’s disease donors. Following quality control, transcriptome-wide association studies were performed for each endophenotype. We used weighted gene co-expression network analysis to build co-expression networks and integrated transcriptome with epigenetic and genetic data from the same donors. We detected a total of 5740 Bonferroni-significant temporal cortex gene associations with Alzheimer’s disease endophenotypes, most of which were with brain tau levels. We discovered tau-associated co-expression modules enriched in known and novel Alzheimer’s disease pathways. We found that a beneficial (or neutral) brain biochemical state of higher total tau and lower phospho-Tau is associated with increased levels of synaptic, DNA damage/repair, nucleic acid metabolism and myelin processes. In contrast, in a detrimental state with lower total and higher phospho-Tau, there is upregulation of vascular and immune, and downregulation of mitochondrial and myelin pathways. There are brain gene expression perturbations that are associated with Alzheimer’s disease endophenotypes. While some of these associations are common across multiple endophenotypes, many are distinct for different Alzheimer’s disease-related proteins. Based on these findings, we propose a hypothetical model of dynamic brain gene expression changes that track with progressive Alzheimer’s disease proteostasis. These expression changes hold potential to serve as dynamic, precision biomarkers of brain Alzheimer’s disease progression. This study demonstrates the potential of integrative multi-omics and deep Alzheimer’s disease endophenotypes in well-characterized brain tissues to precisely uncover the complex biology of Alzheimer’s disease.