Deconstructing and modeling the single cell architecture of the Alzheimer brain

Project Summary/Abstract A diagnosis of Alzheimer's disease dementia (AD) represents the confluence of multiple different molecular processes affecting different cell types in different regions of the aging brain to cause neuronal loss, diminished communication among brain regions, and cognitive decline that ultimately leads to a syndromic diagnosis of AD. In our earlier work, we generated RNA sequence data from the dorsolateral prefrontal cortex (DLPFC), a hub of cognitive and mood circuits, in two prospective cohorts of aging, the Religious Order Study and Memory and Aging project designed to be analyzed jointly (ROSMAP). In our funded AMP-AD project, we explore the relation of these data to AD and cognitive decline, using a molecular network modeling approach that identified sets of critical co-expressed genes that we validated to be involved in AD in independent samples. Further, we used these directional network maps to identify novel regulator genes for AD that we confirmed at the protein level in the target brain tissue and in human iPSC model systems. However, these results explain only a small fraction of the disease biology. In this proposal, we address this limitation in two important ways: (1) we generate single nucleus data to uncover the detailed neuronal and non-neuronal cell population architecture of the aging brain and (2) we bring in additional pathologic and new imaging outcome measures that capture the effect of neurovascular disease. Getting a transcriptome-wide profile on each of the 2000 nuclei profiled in each of 500 brains, we propose to generate an unparalleled dataset of over 1 million individual transcriptomes from the human brain, which will offer a qualitatively different, more refined exploration of the human cortex in AD. Coupled with neuroimaging outcome measures, this dataset will create an ideal platform to integrate the AMP-AD and M2OVE-AD projects since these 500 subjects have all relevant outcome measures for both sets of studies. Finally, we extend the single nucleus profiling to two key validation systems: iPSC-derived neurons, astrocytes, and microglia as well as the new mouse models created by the MODEL-AD project. These reference data create an integrated framework of high-resolution transcriptomic data that will facilitate the selection of an optimal model system for validation and drug discovery to follow up results in AMP-AD and M2OVE-AD.