Avoiding False Positives and Optimizing Identification of True Negatives in Estrogen Receptor Binding and Agonist/Antagonist Assays.

Abstract

The potential for chemicals to affect endocrine signaling is commonly evaluated via in vitro receptor binding and gene activation, but these assays, especially antagonism assays, have potential artifacts that must be addressed for accurate interpretation. Results are presented from screening 94 chemicals from 54 chemical groups for estrogen receptor (ER) activation in a competitive rainbow trout ER (rtER) binding assay and a trout liver slice vitellogenin mRNA expression assay. Results from true competitive agonists and antagonists, and inactive chemicals with little or no indication of ER binding or gene activation were easily interpreted. However, results for numerous industrial chemicals were more challenging to interpret, including chemicals with: (1) apparent competitive binding curves but no gene activation, (2) apparent binding and gene inhibition with evidence of either cytotoxicity or changes in assay media pH, (3) apparent binding but non-competitive gene inhibition of unknown cause, or (4) no rtER binding and gene inhibition not due to competitive ER interaction but due to toxicity, pH change, or some unknown cause. The use of endpoints such as toxicity, pH, precipitate formation, and determination of inhibitor dissociation constants (Ki) for interpreting the results of antagonism and binding assays for diverse chemicals is presented. Of the 94 chemicals tested for antagonism only two, tamoxifen and ICI-182780, were found to be true competitive antagonists. This report highlights the use of two different concentrations of estradiol tested in combination with graded concentrations of test chemical to provide the confirmatory evidence to distinguish true competitive antagonism from apparent antagonism.