Biological Domain Network Analysis Identifies Potential Triadic Interaction between Mitochondria, Nuclear Epigenetics, and Synaptic Viability

Abstract

AbstractBackgroundThe AD endophenotypes initially described in the Common Alzheimer’s Disease Research Ontology (CADRO) were augmented and transformed into computationally implementable definitions for 19 discrete Biological Domains. These AD biological domains were employed in the development of molecular interaction networks depicting the signaling and protein‐protein interaction as a central bioinformatics resource for the TREAT‐AD consortium.Method19 AD biological domains were identified by iterative literature culling and model testing based on gene set enrichment analysis and gene ontology term identification. All 19 biological domains are defined by an exhaustive set of GO terms, allowing scientists everywhere to use the same criteria to define their large datasets in an unbiased manner. Network models were constructed from the leading‐edge genes from all AD‐risk enriched GO terms within each biological domain. The network models incorporate data from over 1700 transcriptomic brain studies and 1100 proteomic brains studies examining shifts in gene expression. These are employed as a reference tool for the formulation of specific hypotheses.ResultThe analysis of mitochondrial gene expression changes at both the transcriptomic and proteomic level identify deficits in both mitochondrial energy production and metabolism along with a decrease in acetyl‐CoA production, a necessary element for many reactions including histone acetylation. This suggests that mitochondrial metabolic tone may modify epigenetic regulation through lysine acetyltransferase (KATs) independently from ATP production. Examination of decrements in KAT activity from other model systems suggest a particularly robust impact on synaptic gene expression, implying a potential interaction between mitochondrial metabolism and synaptic gene expression independent of energy production. Intriguingly, some elements of a protein module identified by TREAT‐AD large‐scale deep proteomic analysis found to be most tightly associated with AD pathology and cognitive decline, are tightly associated with this model.ConclusionThe biological domains were constructed as a computational resource to facilitate hypothesis generation within the TREAT‐AD consortium, and here are implemented to highlight one emerging interaction between domains that may imply new mechanisms of disease regulation coupling between mitochondria and synaptic viability through shifts in epigenetic regulation.