328
X. Yang et al.
four types of noncovalent interactions except for cation–π interactions were observed
between the aliphatic compounds and hTTR. The missing cation–π interaction can
explain why those compounds exhibit moderate hTTR binding potency.
To date, many target-based virtual screening strategies and methods have been
developed [78, 108–112]. In addition, some target-based virtual screening software
has also been developed, including VirtualToxLab™ [113]. We will not attempt here
to cover these advances in detail.
16.6 Conclusions and Future Directions
In this contribution, we have described how to reveal the underlying mechanism of
endocrine disruptor action by employing the molecular modeling methods. Various
aspects were discussed including (1) preparation of the 3D structure of biomacromolecules and EDC molecules, (2) obtaining EDC–biomacromolecule complexes,
and (3) probing the mechanism. A variety of examples were included in the presentation.
Some future directions are proposed as below:
Improving the modeling methods. Tremendous progress had been made for the
target-based methods. However, the efficiency and accuracy of the modeling methods
(e.g., the search algorithms and scoring functions) need to be improved in the future.
Nonreceptor-mediated toxicity pathways. It was clear that EDCs could interfere
with multiple steps of hormone regulation, including biosynthesis and metabolism,
plasma binding, receptor activation/inhibition. Thus far, most researches in EDCs
have focused on modeling the disruption effects of nuclear receptors. There is a need
to pursue more efforts in modeling the interaction between EDCs and nonreceptormediated target in the future.
Attention to species diversity. The structure and function of biomacromolecules in
the endocrine system were conserved across different species. However, the sequence
similarity showed considerable species variation, which may result in different binding affinity and interaction mechanisms of EDCs with target from different species.
Molecular modeling protocols or guidelines. Lots of software or tools for targetbased methods were available. In this situation, how to correctly use those tools and
obtain consistent, reproducible results became a priority problem. Chen conducted
a critical survey on whether we should trust the results of docking studies [68]. His
results indicated that questionable docking results could be observed, even in highprofile journals. How do we increase the accuracy and scientific rigorousness of the
modeling results? Issuing a general molecular modeling protocol or guideline may be
one solution. To date, several official guidelines for toxicant-based modeling methods
have been issued to guide the development and use of ligand-based predictive model.
For example, the OECD has published a guidance document on the validation of
(quantitative) structure–activity relationship ((Q)SAR) models [114] and guidance on
grouping of chemicals [115]. Meanwhile, the European Chemicals Agency (ECHA)
has issued Read-Across Assessment Framework (RAAF) [116]. However, no official
X. Yang et al.
four types of noncovalent interactions except for cation–π interactions were observed
between the aliphatic compounds and hTTR. The missing cation–π interaction can
explain why those compounds exhibit moderate hTTR binding potency.
To date, many target-based virtual screening strategies and methods have been
developed [78, 108–112]. In addition, some target-based virtual screening software
has also been developed, including VirtualToxLab™ [113]. We will not attempt here
to cover these advances in detail.
16.6 Conclusions and Future Directions
In this contribution, we have described how to reveal the underlying mechanism of
endocrine disruptor action by employing the molecular modeling methods. Various
aspects were discussed including (1) preparation of the 3D structure of biomacromolecules and EDC molecules, (2) obtaining EDC–biomacromolecule complexes,
and (3) probing the mechanism. A variety of examples were included in the presentation.
Some future directions are proposed as below:
Improving the modeling methods. Tremendous progress had been made for the
target-based methods. However, the efficiency and accuracy of the modeling methods
(e.g., the search algorithms and scoring functions) need to be improved in the future.
Nonreceptor-mediated toxicity pathways. It was clear that EDCs could interfere
with multiple steps of hormone regulation, including biosynthesis and metabolism,
plasma binding, receptor activation/inhibition. Thus far, most researches in EDCs
have focused on modeling the disruption effects of nuclear receptors. There is a need
to pursue more efforts in modeling the interaction between EDCs and nonreceptormediated target in the future.
Attention to species diversity. The structure and function of biomacromolecules in
the endocrine system were conserved across different species. However, the sequence
similarity showed considerable species variation, which may result in different binding affinity and interaction mechanisms of EDCs with target from different species.
Molecular modeling protocols or guidelines. Lots of software or tools for targetbased methods were available. In this situation, how to correctly use those tools and
obtain consistent, reproducible results became a priority problem. Chen conducted
a critical survey on whether we should trust the results of docking studies [68]. His
results indicated that questionable docking results could be observed, even in highprofile journals. How do we increase the accuracy and scientific rigorousness of the
modeling results? Issuing a general molecular modeling protocol or guideline may be
one solution. To date, several official guidelines for toxicant-based modeling methods
have been issued to guide the development and use of ligand-based predictive model.
For example, the OECD has published a guidance document on the validation of
(quantitative) structure–activity relationship ((Q)SAR) models [114] and guidance on
grouping of chemicals [115]. Meanwhile, the European Chemicals Agency (ECHA)
has issued Read-Across Assessment Framework (RAAF) [116]. However, no official