Using In Vitro and Machine Learning Approaches to Determine Species-Specific Dioxin-like Potency and Congener-Specific Relative Sensitivity among Birds for Brominated Dioxin Analogues

Abstract

There is a paucity of experimental data regarding dioxin-like toxicity of polybrominated dibenzo-p-dioxins/ dibenzofurans (PBDD/Fs) and non-ortho polybrominated biphenyls (PBBs). In this study, avian aryl hydrocarbon receptor 1 (AHR1)-luciferase reporter gene assays were used to determine their species-specific dioxin-like potencies (DLPs) and congenerspecific interspecies relative sensitivities in birds. The results suggested that DLPs of the brominated congeners for chicken-like (Ile324_Ser380) species did not always follow World Health Organization toxicity equivalency factors of their chlorinated analogues. For ring-necked pheasant-like (Ile324_Ala380) and Japanese quail-like (Val324_Ala380) species, the difference in DLP for several congeners was 1 or even 2 orders of magnitude. Moreover, molecular docking and molecular dynamics simulation were performed to explore the interactions between the brominated congeners and AHR1-ligand-binding domain (LBD). The molecular mechanics energy (E MM ) between each congener and each individual amino acid (AA) residue in AHR1−LBD was calculated. These E MM values could finely characterize the final conformation of species-specific AHR1−LBD for each brominated congener. Based on this, mechanism-driven generalized linear models were successfully built using machine learning algorithms and the spline approximation method, and these models could qualitatively predict the complex relationships between AHR1 conformations and DLPs or avian interspecies relative sensitivity to brominated dioxin-like compounds (DLCs). In addition, several AAs conserved among birds were found to potentially interact with species-specific AAs, thereby inducing species-specific interactions between AHR1 and brominated DLCs. The present study provides a novel strategy to facilitate the development of mechanism-driven computational prediction models for supporting safety assessment of DLCs, as well as a basis for the ecotoxicological risk assessment of brominated congeners in birds.

Wildlife ecological risk assessment in the 21st century: Promising technologies to assess toxicological effects

“…Another recent investigation exemplified why molecular docking studies require validation. Zhang et al (2021) performed enzyme induction assays (the selected indicator of species sensitivity) and molecular docking/molecular dynamics modeling to explore the effects of brominated analogs of chlorinated DLCs on species representing AhR‐based avian sensitivity groups. They found that the two lines of evidence did not correlate in a simple manner, and machine learning was needed to explore the relationships between molecular conformations and species sensitivities.…”

Optimizing the Quality of Machine Learning for Identifying the Share of Biogenic and Fossil Carbon in Solid Waste

“…Omitting the excessive redundant information using PCA reduced the subsequent computation time and fixed the dimensionality curse and data sparsity problems. 47 For example, the time required for RFR (87.50 s) is orders of magnitude greater than that of Scaler+PCA+RFR (0.92 s) (Figure 6C).…”