AOP-Based Analysis of Curated Transcriptomic Data Reveals a Context-Dependent Core Mechanism of PFAS-Induced Liver Steatosis
Nicoletta D’Alessandro, Luca Mannino, Giusy del Giudice, Laura Ylä-Outinen, Noora Perho, Laura A. Saarimäki, Emanuele Di Lieto, Zeyad Al-Abdulraheem, Marcella Torres Maia, Simo Inkala, Lorenzo Campini, Lena Möbus, Jack Morikka, Antonio Federico, Angela Serra, Dario GrecoAbstract
Toxicology faces the need to shift from generalized hazard evaluation toward precision approaches that account for the impact of the exposed biological systems. This need is particularly evident for per- and polyfluoroalkyl substances (PFAS), a highly persistent and diverse chemical class whose multiorgan apical toxicities are well documented, yet whose mechanistic understanding remains fragmented. To address this gap, a comprehensive toxicogenomic collection covering multiple PFAS and biological systems is curated, harmonized, and standardized. Through systematic integration of these data with the Adverse Outcome Pathway framework, biological context-aware key event networks that capture the progression from molecular initiating events to apical outcomes are reconstructed. Additionally, new transcriptomic profiles from macrophages exposed to seven PFAS are generated to address the critical but under-investigated immune-related context. Analysis reveals that PFAS toxicity arises from shared early molecular perturbations that diverge across biological systems to produce organ-specific outcomes where immune-related processes consistently emerge as central contributors across multiple contexts. In the liver emerges a conserved mechanistic core underlying PFAS-induced steatosis activated through system-specific pathways shaped by PFAS physicochemical properties and biological context. Overall, this work provides a framework for advancing precision toxicology, enabling rapid, mechanistically grounded, and context-aware PFAS hazard characterization, and supporting the prioritization of uncharacterized PFAS based on shared and context-dependent mechanistic patterns.