Screening of the Antagonistic Activity of Potential Bisphenol a Alternatives Towards the Androgen Receptor Using Machine Learning and Molecular Dynamics Simulation

Zeguo Yang,Ling Wang,Yishuang Duan,Wenjuan Zhang,Shaochen Pang,Yunzhen Sun,Yong Liang,Huiming Cao
DOI: https://doi.org/10.2139/ssrn.4165250
2022-01-01
SSRN Electronic Journal
Abstract:Bisphenol A (BPA) can disturb the androgen receptor (AR)–mediated signaling pathway, which results in endocrine-disrupting effects and reproductive toxicity. However, the influence of the molecular structures of BPA alternatives on the antagonistic activity of the compounds towards AR remains unknown. Therefore, we developed a computational workflow combining machine learning (ML) and accelerated molecular dynamics simulation to predict the antagonistic activity of BPA alternatives towards AR and elucidate the underlying molecular mechanisms. The predictions indicate that more than 46% of the BPA alternatives may present the antagonistic activity of AR. Moreover, conformational flexibility analysis and binding free energy calculations suggest that BPA and bisphenol S (BPS) can bind to the binding pocket of the endogenous ligands of AR rather than the surface binding sites of AR. Compared to BPS, BPA induced the larger conformation changes of key helix H12 from an agonistic conformation to an antagonistic conformation. In summary, the integrated computational models developed in this study provide a tool for evaluating the antagonistic activity of BPA alternatives towards AR and the potential environmental and health risks of the compounds.
What problem does this paper attempt to address?