Computational Molecular Modeling Studies of the Interactions of Estrogens with Their Receptors and Intracellular Binding Protein PDIp

The endogenous estrogens are vitally important female sex hormones with diverse biological functions. Disruptions of their actions contribute to the pathogenesis of a number of disease states in humans, such as endocrine malfunction, infertility, and development of cancers. Therefore, it is important to be able to predict whether a given chemical may have the potential to alter estrogen's actions through binding to the estrogen receptors (ERs) or other estrogen-binding proteins in the body. As opposed to some of the widely-used methods such as biochemical assays and crystallography, computational molecular modeling methods have the potential advantages of low cost, high speed, and high throughput. My dissertation research sought to explore the potential usefulness of computational molecular modeling tools in studying the interactions of various estrogen derivatives (e.g., endogenous estrogen metabolites, non-aromatic steroids, and synthetic antiestrogens) with human ERalpha and ERbeta as well as with a recently-identified intracellular estrogen-binding protein.;As described in SECTION I of this dissertation, I estabolished and validated a molecular docking approach for studying the binding interactions of estrogens with human ERalpha and ERbeta, and then I used this method to determine the binding characteristics of 27 structurally-similar estrogen derivatives with the ligand binding domains (LBDs) of human ERs. I found that while the binding modes of these estrogen derivatives are very similar to that of 17beta-estradiol (E2), there are distinct subtle differences. In the case of A-ring E2 derivatives, for instance, the small differences in the length of the hydrogen bonds formed between the phenolic 3-hydroxyl group of the estrogens and the ERs were found to be a major determinant of their overall binding affinity. This study reveals some of the structural features of the binding interactions of steroidal ligands with human ERs.;All previously characterized endogenous estrogens are steroids with their A-ring being an aromatic ring. Based on the computational analysis of the binding characteristics of various aromatic estrogens for human ERs as described above, I tested an intriguing hypothesis that some of the non-aromatic androgen metabolites or precursors may also be able to bind to human ERs. This work is described in SECTION II of my dissertation. With the aid of the computational molecular modeling tools together with a number of in vitro ER binding assays, I identified, for the first time, several non-aromatic metabolites or synthetic precursors of endogenous androgens that can bind to human ERalpha and ERbeta with physiologically-relevant binding affinity, and these non-aromatic steroids can also activate the ERs to elicit estrogenic responses in human cancer cell lines. These results lead to the suggestion that some of the endogenously-formed androgen precursors or metabolites may serve as male-specific ER modulators.;It is known that compounds such as ICI-172,780, which possesses a long linear side chain attached to the C-7alpha position of E2, could serve as effective ER pure antagonists. The studies described in SECTION III of my dissertation sought to test the hypothesis that estrogen analogs with a shorter but bulky side chain may also be able to function as effective ER antagonists, and theoretically, these compounds may be structurally more stable. Our laboratory designed and synthesized nine of these novel estrogen analogs. Four of them were identified as having a strong antiestrogenic activity in ER-positive human breast cancer cells. Computational molecular modeling studies found that these compounds could tightly bind the ERalpha LBD similar to ICI-182,780, which helps to explain the mechanism of their antiestrogenic actions.;Recently, our laboratory identified, for the first time, the pancreas-specific protein disulfide isomerase (PDIp) as a novel intracellular E2-binding protein with the binding site located in its b-b' domain. As described in SECTION IV of my dissertation, I used computational molecular modeling methods to determine the detailed E2-binding site structures of the human PDIp protein. Molecular docking analysis predicted the binding site in the hydrophobic pocket between the b and b' domains. The hydrogen bond formed between the 3-hydroxyl group of E2 and PDIp-His278 is indispensable for the binding interaction. Selective point mutations of relevant amino acid residues and selective modifications of the ligand structures both confirmed this predicted binding mode. Altogether, these results precisely define, for the time, the E2-binding site structure of human PDIp.;As a whole, the results of my dissertation projects offer important insights into the three-dimensional structural characteristics of the binding interactions of various estrogen analogs with the human ERalpha, ERbeta, and PDIp. These studies provide a platform for the future development of an automated docking-based computational approach that can screen numerous environmental compounds for their potential ability to bind human ERs as well as other estrogen binding proteins in the body.