A tiered screening strategy based on extensive virtual fractionation and elucidation was developed to simplify identification of toxicants in complex environments. In tier1-virtual fractionation, multivariate analysis (MVA) was set up as an alternative of physical fractionation. In tier2-virtual structure elucidation, in-house quantitative structure–retention relationship (QSRR) models and toxicity simulation methods were developed to simplify nontarget identification. The efficiency of the tiered virtual strategy was tentatively verified by soil samples from a chemical park contaminated by antiandrogenic substances. Eight out of 18 sites were detected as antiandrogenic, while none of them exhibited androgenic agonist potencies. Sixty-seven peaks were selected for further identification by MVA, among which over 90% were verified in androgenic fractions in traditional effect-directed analysis (EDA). With 579 tentative structures generated by in silico fragmentation, 74% were elucidated by QSRR and 65% were elucidated by in silico toxicity prediction. All prior peaks were identified at different confidence levels with over 40% of the identified peaks above confidence level 2b, which has been increased over 40% with less than half of the time spent compared to traditional EDA. Such a combination of tiered virtual screening methods provides more efficient and rapid identifications of key toxicants at contaminated sites.

•MD-based separation was the most reliable among the three qualitative simulations.•The MD-based identification model could identify the activeness of 90.5% of PBDEs.•A new workflow including qualitative and quantitative simulations was generated.•The workflow could avoid “false positive” that exist in the QSAR prediction.Recently, great attention has been paid to the identification and prediction of the androgen disrupting potencies of polybrominated diphenyl ethers (PBDEs). However, few existing models can discriminate active and inactive compounds, which make the quantitative prediction process including the quantitative structure-activity relationship (QSAR) technique unreliable. In this study, different grouping methods were investigated and compared for qualitative identification, including molecular docking and molecular dynamics simulations (MD). The results showed that qualitative identification based on MD, which is lab-independent, accurate and closer to the real transcriptional activation process, could separate 90.5% of active and inactive chemicals and was preferred. The 3D-QSAR models built as the quantitative simulation method showed r2 and q2 values of 0.513 and 0.980, respectively. Together, a novel workflow combining qualitative identification and quantitative simulations was generated with processes including activeness discrimination and activity prediction. This workflow, for analyzing the antagonism of androgen receptor (AR) of PBDEs is not only allowing researchers to reduce their intense laboratory experiments but also assisting them in inspecting and adjusting their laboratory systems and results.Download high-res image (130KB)Download full-size image

•A database covering all existing crystal structures of 39 nuclear receptors was built.•An inverse docking method was developed to predict highly vulnerable NRs.•Some rarely reported targets (e.g. LRH-1) are suggested to be vulnerable NRs.Some pollutants can bind to nuclear receptors (NRs) and modulate their activities. Predicting interactions of NRs with chemicals is required by various jurisdictions because these molecular initiating events can result in adverse, apical outcomes, such as survival, growth or reproduction. The goal of this study was to develop a high-throughput, computational method to predict potential agonists of NRs, especially for contaminants in the environment or to which people or wildlife are expected to be exposed, including both persistent and pseudo-persistent chemicals. A 3D-structure database containing 39 human NRs was developed. The database was then combined with AutoDock Vina to develop a System for Predicting Potential Effective Nuclear Receptors (SPEN), based on inverse docking of chemicals. The SPEN was further validated and evaluated by experimental results for a subset of 10 chemicals. Finally, to assess the robustness of SPEN, its ability to predict potentials of 40 chemicals to bind to some of the most studied receptors was evaluated. SPEN is rapid, cost effective and powerful for predicting binding of chemicals to NRs. SPEN was determined to be useful for screening chemicals so that pollutants in the environment can be prioritized for regulators or when considering alternative compounds to replace known or suspected contaminants with poor environmental profiles.Download high-res image (98KB)Download full-size image

Some hydroxylated polybrominated diphenyl ethers (HO-PBDEs), that have been widely detected in the environment and tissues of humans and wildlife, bind to thyroid hormone (TH) receptor (TR) and can disrupt functioning of systems modulated by the TR. However, mechanisms of TH disrupting effects are still equivocal. Here, disruption of functions of TH modulated pathways by HO-PBDEs was evaluated by assays of competitive binding, coactivator recruitment, and proliferation of GH3 cells. In silico simulations considering effects of coregulators were carried out to investigate molecular mechanisms and to predict potencies for disrupting functions of the TH. Some HO-PBDEs were able to bind to TR with moderate affinities but were not agonists. In GH3 proliferation assays, 13 out of 16 HO-PBDEs were antagonists for the TH. In silico simulations of molecular dynamics revealed that coregulators were essential for identification of TH disruptors. Among HO-PBDEs, binding of passive antagonists induced repositioning of H12, blocking AF-2 (transactivation function 2) and preventing recruitment of the coactivator. Binding of active antagonists exposed the coregulator binding site, which tended to bind to the corepressor rather than the coactivator. By considering both passive and active antagonisms, anti-TH potencies of HO-PBDEs could be predicted from free energy of binding.

Hydroxylated and methoxylated polybrominated diphenyl ethers (HO-/MeO-PBDEs) have received increasing attention for their potential endocrine disrupting activities and widely environmental distribution. However, little information is available for the anti-androgenic activities, and the molecular mechanism of interactions with androgen receptor (AR) is not fully understood. In the present study, cell line assay and computational simulation were integrated to systematically explore the molecular mechanism of interactions between chemicals and AR. The metabolites with similar molecular structures exhibited different anti-androgenic activity while none of them showed androgenic activity. According to the multisystem molecular dynamics simulation, minute differences in the structure of ligands induced dramatic different conformational transition of AR-ligand binding domain (LBD). The Helix12 (H12) component of active ligands occupied AR-LBD could become stable, but this component continued to fluctuate in inactive ligands occupied AR-LBD. Settling time and reposition of H12 obtained in dynamics process are important factors governing anti-androgenic activities. The related settling times were characteristic of anti-androgenic potencies of the tested chemicals. Overall, in our study, the stable reposition of H12 is characterized as a computational mark for identifying AR antagonists from PBDE metabolites, or even other various environmental pollutants.

The increased incidences of disorders of male reproductive tract as well as testicular and prostate cancers have been attributed to androgenic pollutants in the environment. Drinking water is one pathway of exposure through which humans can be exposed. In this study, both potencies of androgen receptor (AR) agonists and antagonists were determined in organic extracts of raw source water as well as finished water from waterworks, tap water, boiled water, and poured boiled water in eastern China. Ten of 13 samples of source water exhibited detectable AR antagonistic potencies with AR antagonist equivalents (Ant-AR-EQs) ranging from <15.3 (detection limit) to 140 μg flutamide/L. However, no AR agonistic activity was detected in any source water. All finished water from waterworks, tap water, boiled water, and poured boiled water exhibited neither AR agonistic nor antagonistic activity. Although potential risks are posed by source water, water treatment processes effectively removed AR antagonists. Boiling and pouring of water further removed these pollutants. Phthalate esters (PAEs) including diisobutyl phthalate (DIBP) and dibutyl phthalate (DBP) were identified as major contributors to AR antagonistic potencies in source waters. Metabolites of PAEs exhibited no AR antagonistic activity and did not increase potencies of PAEs when they coexist.