An eight-compound mixture but not corresponding concentrations of individual chemicals induces triglyceride accumulation in human liver cells.

In real life, organisms are exposed to complex mixtures of chemicals at low concentration levels, whereas research on toxicological effects is mostly focused on single compounds at comparably high doses. Mixture effects deviating from the assumption of additivity, especially synergistic effects, are of concern. In an adverse outcome pathway (AOP)-guided manner, we analyzed the accumulation of triglycerides in human HepaRG liver cells by a mixture of eight steatotic chemicals (amiodarone, benzoic acid, cyproconazole, flusilazole, imazalil, prochloraz, propiconazole and tebuconazole), each present below its individual effect concentration at 1-3 µM. Pronounced and significantly enhanced triglyceride accumulation was observed with the mixture, and similar effects were seen at the level of pregnane-X-receptor activation, a molecular initiating event leading to hepatic steatosis. Transcript pattern analysis indicated subtle pro-steatotic changes at low compound concentrations, which did not exert measurable effects on cellular triglycerides. Mathematical modeling of mixture effects indicated potentially more than additive behavior using a model for compounds with similar modes of action. The present data underline the usefulness of AOP-guided in vitro testing for the identification of mixture effects and highlight the need for further research on chemical mixtures and harmonization of data interpretation of mixture effects.

More than additive effects on liver triglyceride accumulation by combinations of steatotic and non-steatotic pesticides in HepaRG cells.

The liver is constantly exposed to mixtures of hepatotoxic compounds, such as food contaminants and pesticides. Dose addition is regularly assumed for mixtures in risk assessment, which however might not be sufficiently protective in case of synergistic effects. Especially the prediction of combination effects of substances which do not share a common adverse outcome (AO) might be problematic. In this study, the focus was on the endpoint liver triglyceride accumulation in vitro, an indicator of hepatic fatty acid changes. The hepatotoxic compounds difenoconazole, propiconazole and tebuconazole were chosen which cause hepatic fatty acid changes in vivo, whereas fludioxonil was chosen as a hepatotoxic substance not causing fatty acid changes. Triglyceride accumulation was analyzed for combinations of steatotic and non-steatotic pesticides in human HepaRG hepatocarcinoma cells. Investigations revealed a potentiation of triglyceride accumulation by mixtures of the steatotic compounds with the non-steatotic fludioxonil, as compared to the single compounds. Mathematical modeling of combination effects indicated more than additive effects for the tested combinations if the method by Chou was applied, and a decrease in EC50 values of the steatotic compounds when applied in mixtures. Use of an adverse outcome pathway (AOP)-driven testing strategy for liver steatosis showed interactions of the test compounds with the nuclear receptors AHR, CAR and PXR, as well as a downregulation of ACOX2. An ACOX2-dependent mechanism underlying the observed mixture effect could not be verified using a siRNA approach. By contrast, a toxicokinetic interaction was identified including an inhibition of the metabolic enzyme CYP3A4 by fludioxonil and a decreased metabolic conversion of the CYP3A4 substrate difenoconazole when used in mixture experiments. In conclusion, an interaction by a steatotic and a non-steatotic compound at the toxicokinetic level on the endpoint triglyceride accumulation in vitro was described.

Combinations of LXR and RXR agonists induce triglyceride accumulation in human HepaRG cells in a synergistic manner.

Activation of nuclear receptors (NR), for example the retinoid-X-receptors (RXR) or the liver-X-receptors (LXR), plays a crucial role as the molecular initiating event in the adverse outcome pathway for liver steatosis. The downstream biological consequences of NR interactions are still not fully understood, especially with multi-receptor-activating compounds and their mixtures. While the default assumption for mixture risk assessment is dose addition, the potential of combinations of synthetic RXR agonists to exert synergistic effects has been shown in the context of NR activation studies. The fact that RXR and LXR are heterodimerization partners raises the question whether combinations of LXR and RXR agonists may cause synergistic effects. Compounds with defined properties were chosen to examine their interactions regarding the activation of RXR and LXR, as well as the steatosis-related key events target gene activation and triglyceride accumulation, using the human HepaRG liver cell model. Synergistic effects were determined for cellular triglyceride accumulation, especially at high compound concentrations, as evaluated using five different mathematical models. Altered LXRα activation in the presence of RXR agonists was observed, and synergistic effects on LXR target genes were identified as a presumably underlying mechanism of the observed synergistic effect. These findings challenge the general validity of dose addition as the default assumption for mixture effects, and point toward the need for a mode of action-based risk assessment for chemical mixtures.

Mixture effects of chemicals: The difficulty to choose appropriate mathematical models for appropriate conclusions.

Many different approaches have been proposed to evaluate and predict mixture effects. From a regulatory perspective, several guidance documents have been recently published and provide a strategy for mixture risk assessment based on valuable frameworks to investigate potential synergistic effects. However, some methodological aspects, e.g. for considering mathematical models, are not sufficiently defined. Therefore, the aim of this study was to examine the usefulness of five main mathematical models for mixture effect interpretation: theoretical additivity (TA), concentration addition (CA), independent action (IA), Chou-Talalay (CT), and a benchmark dose approach (BMD) were tested using a fictional data set depicting scenarios of additivity, synergism and antagonism. The synergism and antagonism scenarios were split in x-axis and y-axis synergism/antagonism, meaning a shift of the curve on x-axis or y-axis. The BMD approach was the only model which showed a perfect correspondence for dose addition. Regarding synergism and antagonism, all approaches correspond well for the x-axis synergism and antagonism with only few exceptions. In contrast, some limitations were observed in the particular scenarios of y-axis synergism and antagonism. Therefore our results show that each model has advantages and disadvantages, and that therefore no single model appears the best one for all kinds of application. We would recommend instead the parallel use of different models to increase confidence in the result of mixture effect evaluation.