You are here:
A COMPUTATIONAL LIBRARY OF THE BIOMOLECULAR TARGETS FOR TOXICITY: RECEPTORS IN THE ENDOCRINE SYSTEM
Citation:
Rabinowitz, J. R., S. B. Little, AND H. Fan. A COMPUTATIONAL LIBRARY OF THE BIOMOLECULAR TARGETS FOR TOXICITY: RECEPTORS IN THE ENDOCRINE SYSTEM. Presented at EPA Science Forum, Washington, DC, June 1-3, 2004.
Description:
A Computational Library of the Biomolecular Targets for Toxicity: Receptors in the Endocrine System
Authors: James R. Rabinowitz and Stephen B. Little, MTB/ECD/NHEERL/ORD, and Huajun Fan, Curriculum in Toxicology, University of North Carolina
Structure activity models for toxicity are usually developed from bioassay data for a series of (similar) chemicals. These models are usually based on the structure, properties and reactivities of the chemicals that have been tested and untested chemicals are evaluated based on these properties. Information about the biological system that is being assayed may influence the selection of the parameters in the model but is not specifically included in the model. Advances in a number of different fields have made it possible to include the biological system being assayed directly in the model. First, there has been an increase in the molecular level data about the mechanisms of toxicity and in many cases the biomolecular targets for toxicity and the molecules responsible for key transformation steps have been identified. Second, the methods for determining the three-dimensional structure of macromolecules have been improved and to some extent become routine. As a result, the structures of many biomolecular targets of toxicity have been determined. Third, a great deal of molecular modeling software has become available for characterizing the interaction between biopolymers and small molecules. Much of this software was developed for the design of new pharmaceuticals. Fourth, highly parallel computers have become available and in many ways these problems are naturally parallel. In order to apply these advances to an Agency problem, the capacity of molecules to bind to the estrogen and androgen receptor are being modeled. Crystal structures for these receptors with ligand bound are in the literature. The ligand is removed by computational means to create a target. The capacity of an untested chemical to bind to the receptor may be estimated from its best fit to the target. There are a number of crystal structures of the estrogen receptor with different ligand and cofactor in the literature. Studies comparing the results for different estrogen receptor targets show the importance of including receptor flexibility in the model. There is only a single structure available for the androgen receptor. In this case flexibility must be included directly, a much more computer intensive modeling effort. All these results will be shown. Ultimately, a library of biomolecular targets for toxicity may be developed.