In silico toxicity evaluation of dioxins using structure–activity relationship (SAR) and two-dimensional quantitative structure–activity relationship (2D-QSAR)

• Published in: Archives of toxicology Vol. 93; no. 11; pp. 3207 - 3218
• Main Authors: Yang, Hong, Du, Zhe, Lv, Wen-Juan, Zhang, Xiao-Yun, Zhai, Hong-Lin
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.11.2019; Springer Nature B.V
• Subjects: Algorithms; Aromatic compounds; Artificial neural networks; Binding; Biomedical and Life Sciences; Biomedicine; Dioxins; Environmental behavior; Environmental Health; Human behavior; Mathematical models; Molecular Toxicology; Neural networks; Occupational Medicine/Industrial Medicine; Organic compounds; Persistent organic pollutants; Pharmacology/Toxicology; Pollutants; Polychlorinated dibenzodioxins; Polychlorinated dibenzofurans; Predictions; Risk taking; Robustness (mathematics); Structure-activity relationships; Toxicity; Two dimensional models; Dioxins; SAR; Heuristics; POPs; 2D-QSAR; pEC; pEC50
• ISSN: 0340-5761; 1432-0738; 1432-0738
• DOI: 10.1007/s00204-019-02580-w

Prediction of pEC 50 values of dioxins binding with the aryl hydrocarbon receptor (AhR) is of great significance for exploring how dioxins induce toxicity in human body and evaluating their environmental behaviors and risks. To reveal the factors that influence the toxicity of dioxins, provide more accurate mathematical models for predicting the pEC 50 values of dioxins, and supplement the toxicity database of persistent organic pollutants, qualitative structure–activity relationship (SAR) and two-dimensional quantitative structure–activity relationship (2D-QSAR) were used in this study. The research objects in this study were 60 organic compounds with pEC 50 values and 162 compounds without pEC 50 values, which included polychlorinated dibenzofurans (PCDFs), polychlorinated dibenzo-p-dioxins (PCDDs), and polybrominated dibenzo-p-dioxins (PBDDs). The qualitative structure–activity relationship (SAR) was performed first and concluded that halogen substitutions at any of the 2, 3, 7, and 8 sites increased the pEC 50 value of the compound. Moreover, two-dimensional quantitative structure–activity relationship (2D-QSAR) models were established by employing multiple linear regression (MLR) method and artificial neural network (ANN) algorithm to investigate the factors affecting the pEC 50 values of dioxins molecules. MLR was used to establish the well-understood linear model and ANN was used to establish a more accurate non-linear model. Both models have good fitting, robustness, and predictive ability. Importantly, the ability of dioxins binding to AhR is mainly determined by molecular descriptors including E1m, SM09_AEA (dm), RDF065u, F05 [Cl–Cl], and Neoplastic-80. In addition, the pEC 50 values of the 162 dioxins without toxicity data were predicted by MLR and ANN models, respectively.