Computational models of phytoestrogens, mycoestrogens, and diethylstilbestrol analogues

Phytoestrogens are plant-derived compounds that are known to induce estrogen activity and are important components of various foods including soy products, broccoli, and fruits and vegetables. Mixtures of phytochemicals are also marketed as dietary supplements to relieve postmenopausal symptoms. This study focuses on phytoestrogens, mycoestrogens, and diethylstilbestrol (DES) analogues that have been known to activate the alpha isoform of the human estrogen receptor (ER). We believe a structure-activity pattern exists within the chemical structure of the phytoestrogen and DES compounds that correlates to estrogen Receptor binding activity. Holographic Quantitative Structure-Activity Relationship (HQSAR) models with two-dimensional (2d) descriptors can describe this pattern, as well as three-dimensional (3d) models such as Comparative Molecular Field Analysis (CoMFA) and Comparative Molecular Similarity Indices Analysis (CoMSIA) which were produced using the Quantitative Structure-Activity Relationship (QSAR) method. Within this structure-activity relationship, we believe that these compounds bind to the estrogen receptor. The HQSAR model that was developed defines two-dimensional fragments (4-7 atom fragments) responsible for ER activity (Q2=0.758, R2=0.915). The 3d structures were used to develop the CoMFA and CoMSIA QSAR models that defined distinct structure features responsible for ER binding activity (Steric, Electrostatic, Hydrophobic). All relevant isomers and enantiomers were modeled to include the stereochemical nature of particular phytoestrogens in these 3D-QSAR models. The optimal CoMFA model displayed a predictive Q2 of 0.792 (R2=0.983) while the optimal CoMSIA model produced in a predictive Q2 of 0.831 (R2=0.933). Although the phytoestrogens have a similar backbone and rather rigid structure, the mycoestrogens have very little in common with the other classes of compounds and are more flexible than both phytoestrogens and diethylstilbestrol derivatives, however they persist to have an effect on the estrogen receptor. Therefore, we investigated how the conformation, flexibility, and general structure of the mycotoxins affect the binding that occurs between the mycotoxins, phytoestrogens, DES derivatives and the ER, and how this correlates to Estrogen Receptor binding activity. Isomers and enantiomers, Phytoestrogens, DES derivatives, and mycoestrogens were generated using the computer program Stereoplex, while the 3d structures were generated using the program Concord. Conformations of the structures were produced using Confort, and the removal of potentially identical compounds was performed by using a root means squared deviation of 0.75A as the minimum threshold of diversity for identifying duplicate compounds for the mycoestrogens. Comparison of the compounds to available crystal structures of the same Mycoestrogen from the Cambridge Structural Database was done to ensure the correct structures were produce, and the compounds where then aligned using FlexS with DES and 17beta-Estradiol (E2) as templates. The compounds were docked to the alpha isoform of the Estrogen Receptor from the 1ERE crystal structure using Surflex-Dock, and scored using CScore. Finally, docking and scoring results from the mycoestrogens suggested that the functional groups found on both phytoestrogens and diethylstilbestrol derivatives had an effect on the compounds activity when bound to the ER. Therefore, additional docking studies were performed to determine what effect they would have on the activity. The docking profiles show that very subtle differences exist within the compounds that allows them to mimic and bind to the estrogen receptor to allow activation of the receptor, and their alignment suggest features that are consistent to allow for mimicking 17beta-Estradiol. We conclude that the conformation searching, flexibility analysis, and the docking profiles focuses on features of the compounds that allow for activation of the estrogen receptor. This study has produced the most comprehensive QSAR models of ER binding activity for myco, phyto and stilbene estrogens to date. These three models have considerable potential to predict the ER binding activity of myco, phyto and stilbene compounds found through database screening or the analytical separation and identification of plant and fungal extracts.