bioRxiv2018
While Alzheimers disease (AD) is the most prevalent cause of dementia, complex combinations of the underlying pathologies have led to evolved concepts in clinical and neuropathological criteria in the past decade. Pathological AD can be decomposed into subsets of individuals with significantly different antemortem cognitive decline rates. Using transcriptome as a proxy for functional state, we preselected 414 expression profiles of clinically and neuropathologically confirmed AD subjects and age matched non-demented controls sampled from a large community based neuropathological study. By combining brain tissue specific protein interactome with gene network, we identify functionally distinct composite clusters of genes which reveal extensive changes in expression levels in AD. The average global expression for clusters corresponding to synaptic transmission, metabolism, cell cycle, survival and immune response were downregulated while the upregulated cluster had a large set of uncharacterized pathways and processes that may constitute an AD specific phenotypic signature. We identified four master regulators across all clusters of differentially expressed genes by enrichment analysis including TGIF1 and EGR3. These transcription factors have previously not been associated with AD and were validated in brain tissue samples from an independent AD cohort. We identify TGIF1, a transcriptional repressor as being neuroprotective in AD by activating co-repressors regulating genes critical for DNA repair, maintaining homeostasis and arresting cell cycle. In addition, we show that loss of EGR3 regulation, mediates synaptic deficits by targeting the synaptic vesicle cycle. Collectively, our results highlight the utility of integrating protein interactions with gene perturbations to generate a comprehensive framework for characterizing the alterations in molecular network as applied to AD.