Experimental Validation of Bioinformatically Prioritized Therapeutic Hypotheses and Targets for Alzheimer’s Disease

Abstract

AbstractBackgroundThe complex etiology of Alzheimer’s disease (AD) challenges therapeutic development efforts by obscuring which altered biological processes are optimal candidates for intervention. We have created a workflow to prioritize molecular targets based on signatures of AD risk from patient data and assemble candidate targets into functional groupings from which we can articulate sets of hypotheses. Prioritized data‐derived hypotheses to date include mitochondrial dysfunction and chronic neuroinflammation. We present an approach to rapidly assess whether the hypothesized driver genes affect the predicted biological outcomes in the appropriate cellular contexts.MethodImmortalized mouse BV2 cells were used as a microglial surrogate. To test the candidate hypotheses in an AD‐relevant context, we examined both wild‐type BV2 cells and cells where Psen2 expression has been stably knocked down. To perturb the nominated target genes, each cell line was transfected with one of a panel of 29 siRNAs targeting the genes or a control siRNA prior to phenotypic assessment. We tested mitochondrial function using both MitoTracker TMRM staining and the AlamarBlue assay. For immune function, we tested phagocytic activity using pHrodo Green Zymosan Bioparticles and the transcriptional output from NF‐kappaB using a luminescent reporter. Finally we measured proteomic differences between the siRNA treated cells and control cells to assess the impact of each perturbation on the expression of protein sets targeted by the hypotheses.ResultOf the targets screened, 28% are found to exhibit mitochondrial phenotypes in both assays and 38% are found to exhibit immune phenotypes in both assays. Hits from the immune assays are generally more specific to the immune hypothesis targets (70%) whereas the hits from the mitochondrial assays are more evenly divided between both hypothesis areas (50%). Examining the proteomic results highlights the pathways that are altered by each perturbation and we align these with patient data to identify perturbations that affect the targeted pathways.ConclusionWe have established a platform to rapidly assess biologically‐informed and interpretable hypotheses and targets derived from large‐scale data resources for disease relevance. The targets validated by these experiments will be prioritized for further resource development within the TREAT‐AD consortium.