Prediction and assessment of xenoestrogens mixture effects using the in vitro ERα-CALUX assay.

Abstract

Introduction: Many natural or synthetic compounds used in foods, dietary supplements, and food contact materials (FCMs) are suspected endocrine disruptors (EDs). Currently, scientific evidence to predict the impacts on biological systems of ED mixtures is lacking. In this study, three classes of substances were considered: i) phytoestrogens, ii) plant protection products (PPP) and iii) substances related to FCMs. Fourteen compounds were selected based on their potential endocrine activity and their presence in food and FCMs. Methods: These compounds were evaluated using an in vitro gene expression assay, the ERα-CALUX, to characterize their responses on the estrogen receptor alpha. Cells were exposed to fixed ratio mixtures and non-equipotent mixtures of full and partial agonists. The concentration-response curves measured for the three classes of compounds were characterized by variable geometric parameters in terms of maximum response (efficacy), sensitivity (slope) and potency (median effective concentration EC50). To account for these variations, a generic response addition (GRA) model was derived from mass action kinetics. Results: Although GRA does not allow us to clearly separate the concentration addition (CA) and independent action (IA) models, it was possible to determine in a statistically robust way whether the combined action of the chemicals in the mixture acted by interaction (synergy and antagonism) or by additive behavior. This distinction is crucial for assessing the risks associated with exposure to xenoestrogens. A benchmark dose approach was used to compare the response of phytoestrogen blends in the presence and absence of the hormone estradiol (E2). At the same time, 12 mixtures of 2-5 constituents including phytoestrogens, phthalates and PPPs in proportions close to those found in food products were tested. In 95% of cases, the response pattern observed showed a joint and independent effect of the chemicals on ER. Discussion: Overall, these results validate a risk assessment approach based on an additive effects model modulated by intrinsic toxicity factors. Here, the CA and IA approaches cannot be distinguished solely based on the shape of the concentration response curves. However, the optimized GRA model is more robust than CA when the efficacy, potency, and sensitivity of individual chemical agonists show large variations.