Integrative network-based construction of ecologically relevant computational adverse outcome pathways for organic mercury-induced toxicity

Adverse outcome pathways (AOPs) describe mechanisms of toxicity by connecting molecular events with outcomes at higher levels of biological organization. Computational AOPs (cAOPs), constructed using existing toxicological data, can accelerate the early stages of AOP development, which is often a time-consuming and resource-intensive process. In this study, an integrative network-based framework was developed to construct cAOPs, with a particular focus on elucidating the toxicity of organic mercury in fish. First, 124 organic mercury compounds, associated fish toxicity endpoints, and proteins were curated from the Comparative Toxicogenomics Database and ECOTOX database, followed by molecular docking to identify novel interactions with 16 zebrafish protein orthologs. Subsequently, toxicity endpoints and identified molecular interactions were standardized and harmonized using established ontologies and AOP-Wiki. These data were integrated with event relationship information from AOP-Wiki and published literature to construct an organic mercury-associated toxicity network, comprising 197 nodes and 243 edges, which was subsequently filtered using AOP definition-based criteria to identify biologically relevant pathways. Further, these pathways were ranked based on their novelty with respect to existing AOPs in AOP-Wiki, resulting in the proposal of four novel cAOPs describing glutathione peroxidase binding or altered metallothionein levels leading to neurological manifestations or dysbiosis in fish. Overall, this study presents an integrative network-based framework for constructing cAOPs applicable to diverse contaminants and species, supporting new approach methodologies for toxicological risk assessment.