Effect-based tools for monitoring estrogenic mixtures: Evaluation of five in vitro bioassays

In vitro estrogen receptor transactivation assays (ERTAs) are increasingly used to measure the overall estrogenic activity of environmental water samples, which may serve as an indicator of exposure of fish or other aquatic organisms to (xeno)estrogens. Another potential area of application of ERTAs is to assist the monitoring of the potent steroids 17β-estradiol (E2) and 17α-ethinylestradiol (EE2) under the Water Framework Directive (WFD) watch-list mechanism. Chemical analysis of E2 and EE2 is currently hampered by limits of quantification being mostly above the proposed annual average Environmental Quality Standards (AA-EQS) of 0.4 and 0.035 ng/L, respectively. Sensitive ERTAs could circumvent current detection challenges by measuring total estrogenic activity expressed as E2-equivalent (EEQ) concentrations. However, the use of different ERTAs results in different EEQ concentrations for the same sample. Reasons for these differences are known, but it remains unclear how to use and interpret bioassay results in a harmonised way. The aim of this study was to compare the intra- and inter-day variability of EEQ measurements using five different ERTAs (YES, ERα-CALUX, MELN, T47D-KBluc and GeneBLAzer-ERα) with regard to their applicability as effect-based tools in environmental monitoring. Environmentally relevant artificial mixtures of (xeno)estrogens were prepared to represent samples with higher (i.e. multiple times the AA-EQS for E2) or lower pollution levels (i.e. around the AA-EQS for E2). Mixtures were tested either directly or following solid phase extraction (SPE). The SPE step was included, as environmental samples typically require enrichment before analysis. Samples were analysed repeatedly to test intra-day and inter-day variability. Estrogenicity was quantified using the 10% effect level (PC10) of the positive control (E2) and expressed as EEQ concentrations. The average coefficient of variation (CV) of EEQ concentrations for the five ERTAs and all samples was 32%. CV was lower for intra-day experiments (30%) compared to inter-day experiments (37%). Sample extraction using SPE did not lead to additional variability; the intra-day CV for SPE extracted samples was 28%. Of the five ERTAs, ERα-CALUX had the best precision and repeatability (overall CV of 13%).