You are here:
CHARACTERISTICS OF HIGH-QUALITY DATA FOR QSARS: AN ER CASE STUDY
Citation:
Tapper, M A., J. S. Denny, R C. Kolanczyk, B R. Sheedy, AND P. K. Schmieder. CHARACTERISTICS OF HIGH-QUALITY DATA FOR QSARS: AN ER CASE STUDY. Presented at Third Indo-US Workshop on Mathematical Chemistry, University of Minnesota-Duluth, Duluth, MN, August 2-7, 2003.
Description:
The USEPA is mandated by Congress to screen industrial chemicals and pesticides for potential endocrine activity. To evaluate this potential in fish, the affinity of chemicals for the rainbow trout estreogen receptor (rtER) is assessed. A subset of chemicals are also tested for their ability to induce ER-dependent protein production. Together this information provides biologically-relevant data for QSAR development, which is subsequently used to prioritize chemicals, among the thousands in commerce, to undergo testing as environmental estrogens.
A scientifically-defensible QSAR is one based upon high-quality empirical data that is interpretable in the context of the risk being predicted, with established criteria for data evaluation and acceptance. Therefore, criteria are being established for acceptance and interpretation of rtER binding data. Many chemicals bind rtER in a manner consistent with competitive displacement of the endogenous hormone 17B-estradiol from the receptor and with high relative binding affinity (RBA), while many more chemicals clearly show no affinity for the receptor. Additionally, many industrial chemicals have relatively weak, but measurable affinity for the receptor (0.0001 to 0.1% that of estradiol), but with the competitive nature of the binding suspect. The biological relevance of such low affinity binding is also unclear, but potentially significant in light of ubiquitous environmental distribution of some chemicals. We have developed a dual approach to determining binding character and biological relevance of low affinity xenobiotics. First, criteria for data acceptance and interpretation are established based on inspection of finding curves, including completeness, shape, and slope. Those chemicals producing incomplete curves, exhibiting less than mximum displacement of estradiol, or having apparent non-competitive binding interactions undergo further evaluation. Second, recognizing that binding information alone is of limited value additional tests are done to determine the relevance of suspect or extremely low rtER binding affinity in a more complex in vitro systemwhich includes ER-regulated protein production. In the male rainbow trout liver slice assay, tissue is exposed to varying concentrations of chemical and production of the vitellogenin mRNA from p-nonylphenol with an RBA of only 0.04%. Additional low affinity chemicals are under investigation. Throughout both tissue slice and binding assays target chemical concentrations are also measured to assure chemical identity and bioavailibity. The combined approach provides a high-quality, and toxicologically interpretable dataset for QSAR prediction of estrogenicity.