Integrative translational discovery of vascular risk factors in aging and dementia

Summary Alzheimer's disease (AD) is the most common cause of dementia and is characterized by accumulation of senile plaques and neurofibrillary tangles in the brain. Neuronal cell death and reactive gliosis are also components of neuropathology in AD patients resulting in significantly altered proportions of cell types in brain tissue of ADs when compared with unaffected controls. Studies investigating the brain transcriptome to identify genes and pathways associated with disease or intermediate phenotypes are abundant. However, heterogeneity with respect to the proportions of cell types that contribute to the bulk tissue transcriptome make it challenging to differentiate expression changes that are a consequence of neuropathology vs those that may be driving disease pathogenesis. This has important implications for the interpretation of studies, identification of therapeutic targets and understanding of the underlying pathways and cell types involved. Our parent and revision grants (RF1 AG051504, RF1 AG051504-01S2) aim to close this knowledge gap by generating transcriptome measures from sorted populations of cells isolated from brain tissue. The goal of this work is to identify cell-type marker genes for application of analytic algorithms to deconvolute cell type proportion from the bulk transcriptome. Single-cell type transcriptome data will also be used to investigate the role of sex and APOE genotype on genes and expression networks at the single-cell-type level. While sorted cell populations are more homogeneous than bulk brain tissue, there likely remains some heterogeneity due to cell type sub- populations, rare cell types, or specificity of targeted cell surface markers, and combinations thereof. The degree of this heterogeneity and the effect on the expression profile of sorted cell populations is not well understood. Furthermore the influence of selected reference datasets, cell-type marker genes and deconvolution algorithms on the accuracy of annotation and deconvolution for specific bulk tissue datasets are likewise not established. This administrative supplement proposal aims to address these knowledge gaps through use of 10X genomics Single Cell technology to characterize the heterogeneity of sorted populations of brain cells, and generate a benchmarked dataset against which cell type marker genes, reference datasets and deconvolution algorithms can be tested. The overarching goal of this proposal is to enhance and refine the utility of data from sorted populations of cells as a reference for use in annotation and deconvolution of bulk tissue gene expression datasets, and improve the accuracy of bulk tissue deconvolution for brain tissue. This work has implications for AD research and the broader neuroscience field.