The authors have declared that no competing interests exist.
Conceived and designed the experiments: BS PS. Performed the experiments: BS. Analyzed the data: BS PS. Contributed reagents/materials/analysis tools: FK BNB PS. Wrote the paper: BS PS.
Consumers of fruits and vegetables are frequently exposed to small amounts of hormonally active pesticides, some of them sharing a common mode of action such as the activation of the human estrogen receptor α (hERα) or β (hERβ). Therefore, it is of particular importance to evaluate risks emanating from chemical mixtures, in which the individual pesticides are present at human-relevant concentrations, below their corresponding maximum residue levels. Binary and ternary iso-effective mixtures of estrogenic pesticides at effect concentrations eliciting a 1 or 10% effect in the presence or absence of 17β-estradiol were tested experimentally at the hERα in the yeast-based estrogen screen (YES) assay as well as in the human U2-OS cell-based ERα chemical-activated luciferase gene expression (ERα CALUX) assay and at the hERβ in the ERβ CALUX assay. The outcome was then compared to predictions calculated by means of concentration addition. In most cases, additive effects were observed with the tested combinations in all three test systems, an observation that supports the need to expand the risk assessment of pesticides and consider cumulative risk assessment. An additional testing of mixture effects at the hERβ showed that most test substances being active at the hERα could also elicit additive effects at the hERβ, but the hERβ was less sensitive. In conclusion, effects of the same ligands at the hERα and the hERβ could influence the estrogenic outcome under physiological conditions.
Many substances used as crop protection products possess hormonal activity, which may influence human health by imitating or disrupting endogenous hormones [
We investigated the effects of single pesticides (pirimicarb, propamocarb, fenhexamid, fludioxonil, chlorpyrifos, fenarimol, 2,4’-DDT and 4,4’-DDT) as well as selected binary and ternary mixtures of them at low effect concentrations in a β-galactosidase reporter gene assay, the broadly used Yeast-based Estrogen Screen (YES) assay, as well as in the human U2-OS cell-based ERα chemical-activated luciferase gene expression (ERα CALUX) assay. Full concentration-response curves were evaluated for the mathematical modeling, but the assessment of additivity was restricted to low effect concentrations (EC01 and EC10) in the range of human-relevant concentrations. Furthermore, the substances were screened in combination with a saturating concentration of 17β-estradiol (E2) to test for an E2 potentiating or an anti-estrogenic activity in the YES assay, and the anti-estrogenic substances were also tested for anti-estrogenic activity in the ERα CALUX assay.
Most studies have analyzed the mixture effects of pesticides at the hERα, while only a few reports have dealt with the effects of individual pesticides on the human estrogen receptor β (hERβ) and to our knowledge no study has investigated pesticide mixture effects at the hERβ. While the hERα frequently occurs in tissues related to reproductive activity (uterus, mammary gland), the hERβ is more widely distributed, and the ligand binding domains of the two isoforms slightly differ (59% homology) (reviewed by Gustafsson [
The well-known concept of concentration addition (CA), based on the work of Loewe and Muischnek [
The aim of this study was to evaluate the suitability of the established YES assay and the more recent ERα CALUX assay to identify estrogenic or anti-estrogenic effects at the hERα and to investigate mixture effects of estrogenic pesticides at low concentrations via CA. Additionally, the effects of the pesticides and mixtures at the hERβ, mostly acting as a counterpart of the hERα
17ß-estradiol (E2; CAS# 521-18-6; ≥ 98% purity), 17α-methyltestosterone (CAS# 58-18-4; 99.5% purity), 2,4’-dichlorodiphenyltrichloroethane (2,4’-DDT CAS# 789-02-6; 99.5% purity), 4,4’- dichlorodiphenyltrichloroethane (4,4’-DDT CAS# 50-29-3; 99.8% purity), 4-hydroxytamoxifen (4-HT; CAS# 68047-06-3; ≥ 98% purity), chlorpyrifos (CAS# 2921-88-2; 99,7% purity), corticosterone (CAS# 50-22-6; ≥ 98.5% purity), fenarimol (CAS# 60168-88-9; 99.9% purity), fenhexamid (CAS# 126833-17-8; 99,7% purity), fludioxonil (CAS# 131341-86-1; 99.9% purity), ICI 182,780 (ICI; CAS# 129453-61-8; > 98% purity), pirimicarb (CAS# 23103-98-2; 98.5% purity), propamocarb (CAS# 24579-73-5; 99.3% purity), resveratrol (CAS# 501-36-0, ≥ 99% purity), and tamoxifen (CAS# 10540-29-1; ≥99% purity) were purchased from Sigma Aldrich (Schnelldorf, Germany). By dissolving the chemicals in dimethyl sulfoxide (DMSO; CAS# 67-68-5; ≥ 99.5% purity; Carl Roth, Karlsruhe, Germany), dilution series were stored at -20°C in glass vials freeze-thawed for each experiment or batched in polypropylene vials and only thawed once.
The YES assay developed by Routledge and Sumpter [
A cytotoxic effect was defined as a decrease in optical density of the cell suspension at 690 nm by more than 30% when compared to the solvent control [
Substances that could not be fully solubilized in the YES assay were tested in a cell-free control plate, and the turbidity, as a benchmark for insolubility, was measured at a wavelength of 690 nm. This was necessary to distinguish between a real growth induction of the yeast cells by the tested substance and an artifact resulting from the insolubility of the compound at higher concentrations, since in both cases an increase of turbidity in the cell-containing assay would be observed. It should be taken into account that an increased turbidity could also mask cytotoxicity.
The ERα and ERβ CALUX assay were performed to evaluate the effect of the test substances on the hERα and hERβ as described by van der Burg et al. [
In the case of anti-estrogenicity assessment in the ERα CALUX assay, test substances that showed an anti-estrogenic activity in the YES assay, fenhexamid and fludioxonil, were tested in combination with 3 pM E2 (~ EC50). One plate per experiment included a full dose-response curve of the competitive hERα antagonist tamoxifen (3 nM to 10 μM), 1 nM 4-HT as additional anti-estrogenic control and 10 μM resveratrol as negative control, all also combined with 3 pM E2.
The raw optical density values obtained at 690 nm in the YES assay were subtracted as background from the values obtained at 540 nm and thereafter the corresponding solvent control values were also subtracted. All data points were normalized to the effect of 1 nM E2 (~ EC100), and mean values were generated in the YES assay. The data from the ERα and ERβ CALUX assays were evaluated in a similar manner: Solvent controls were subtracted from the raw luminescence values and all mean values were normalized to 0.1 nM E2 (~ EC100) in the ERα CALUX assay and 30 nM (~ EC100) in the ERβ CALUX assay, with one exception: When fludioxonil and fenhexamid were tested for anti-estrogenic effects in the ERα CALUX assay, the values were normalized to 3 pM E2.
Normalized concentration-response data from the YES, ERα and ERβ CALUX assays were analyzed by nonlinear regression after the x-axis with the test substance concentrations was log-transformed. The best-fit approach established by Scholze et al. [
Fixed mixtures ratios were calculated to be iso-effective [
YES EC01 | ERα CALUX EC01 | ERβ CALUX EC01 | YES EC10 | ERα CALUX EC10 | ERβ CALUX EC10 | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
substance | M | [CI] | M | [CI] | M | [CI] | M | [CI] | M | [CI] | M | [CI] |
propamocarb | - | - | 2.09E-07 | [1.66E-07-3.72E-07] | 1.15E-05 | [3.55E-06-2.69E-05] | - | - | 7.24E-07 | [6.31E-07-9.77E-07] | 2.63E-05 | [1.41E-05-2.95E-05] |
chlorpyrifos | 7.94E-08 | [3.09E-08-4.57E-06] | 6.17E-07 | [2.14E-07-1.07E-06] | - | - | - | - | 3.47E-06 | [2.63E-06-4.27E-06] | - | - |
fenarimol | 1.17E-05 | [3.89E-06-2.95E-05] | 7.94E-07 | [6.61E-07-1.15E-06] | 3.63E-06 | [1.35E-06-7.94E-06] | 2.19E-05 | [1.17E-05-3.72E-05] | 2.82E-06 | [2.63E-06-3.31E-06] | 1.35E-05 | [9.12E-06-2.57E-05] |
fludioxonil | 8.32E-07 | [4.42E-08-9.22E-07] | 3.63E-07 | [3.16E-07-5.50E-07] | 8.91E-07 | [1.00E-07-1.23E-06] | 2.63E-05 | [6.92E-06-3.80E-05] | 1.26E-06 | [1.12E-06-1.48E-06] | 2.09E-06 | [5.89E-07-2.40E-06] |
fenhexamid | 2.19E-07 | [1.51E-08-6.92E-07] | 6.31E-07 | [2.95E-07-8.71E-07] | 1.17E-06 | [5.89E-07-2.88E-06] | 2.14E-05 | [2.04E-06-2.63E-05] | 2.63E-06 | [1.82E-06-3.02E-06] | 5.01E-06 | [3.98E-06-5.62E-06] |
4,4‘-DDT | 3.47E-07 | [1.51E-07-6.76E-07] | 4.57E-07 | [3.63E-07-5.13E-07] | 1.02E-06 | [5.50E-07-1.82E-06] | 8.32E-06 | [6.61E-06-1.12E-05] | 1.07E-06 | [9.55E-07-1.15E-06] | - | - |
2,4‘-DDT | 3.31E-08 | [1.82E-08-6.76E-08] | 4.07E-08 | [2.45E-08-6.46E-08] | 5.50E-08 | [8.32E-09-1.86E-07] | 4.68E-07 | [3.55E-07-6.92E-07] | 1.02E-07 | [7.41E-08-1.45E-07] | 6.03E-07 | [4.27E-07-1.23E-06] |
Comparison of the EC01 and EC10 values, the effect concentration needed to elicit a 1 or 10% effect of 1 nM E2 with the approximate 95% confidence interval [CI] of the single substances in the corresponding test systems. Pirimicarb was excluded, since it did not show an effect in any test system.
The hERα was trans-activated concentration-dependently by E2 in the YES assay (
Regression models with the indicated EC50 concentrations and the 95% confidence bands for E2 in the (A) YES, (B) ERα CALUX and (C) ERβ CALUX assays.
All tested pesticides showed estrogenic activity with full concentration-response curves in the YES and ERα CALUX assays with the exception of pirimicarb, which was negative in both assays, and propamocarb, which was only active in the ERα CALUX assay (
Regression models of the test substances in the (A) YES, (B) ERα CALUX and (C) ERβ CALUX assays. Substances not eliciting an effect are not shown.
Regression models of the test substances in the (A) YES (with 1 nM E2) and the (B) ERα CALUX assay (with 3 pM E2). Substances not eliciting an effect are not shown.
Fenhexamid and fludioxonil showed concentration-dependent anti-estrogenic effects when combined with 1 nM E2 in the YES assay (
In the ERβ CALUX propamocarb, fenarimol, fludioxonil, fenhexamid, 4,4’-DDT and 2,4’-DDT showed a dose-dependent estrogenic activity at the hERβ (
The concentrations of the test substances to be tested were limited depending on their solubility. A decreased solubility led to an increase of turbidity/optical density measured at 690 nm in the yeast-based assay or was determined by microscopic examination in the CALUX assays. Concentrations leading to insolubility were excluded in the CALUX assays, but not in the YES assay. In the CALUX assays it was possible to generate full concentration-response curves at lower concentrations. It should be pointed out that fenarimol and chlorpyrifos only showed reporter gene activity in the YES assay when applied in concentrations leading to insolubility; nevertheless, these were tested in the YES assay (100 μM fenarimol, 100–1000 μM chlorpyrifos, 100–500 μM fludioxonil, 100 μM fenhexamid, 50–500 μM 2,4’- and 4,4’-DDT). Cytotoxicity, measured by performing the LDH leakage assay, was not observed in the case of the U2-OS cells (data not shown). A decrease of turbidity in the YES assay was assumed to be equivalent to an inhibition of yeast cell growth. Fenhexamid decreased turbidity in the YES assay by more than 30% in comparison to the solvent control at concentrations ≥ 500 μM and the concentrations were therefore excluded (data not shown).
Iso-effective combinations of different estrogenic pesticides resulted in a concentration-dependent increase of reporter gene activity in the YES assay (
Concentration-response function | EC01 | EC10 | |||||||
---|---|---|---|---|---|---|---|---|---|
mixture | RM | predicted M | observed M [CI] | predicted M | observed M [CI] | ||||
Weibull | 23.39 | 4.46 | 0 | 0.81 | 4.90E-07 | 5.89E-07 [5.01E-07-7.41E-07] | 1.86E-06 | 2.00E-06[1.86E-06-2.19E-06] | |
Weibull | 23.18 | 4.41 | 0 | 0.84 | 5.01E-07 | 5.50E-07 [4.90E-07-7.76E-07] | 1.95E-06 | 1.86E-06 [1.86E-06-2.24E-06] | |
Weibull | 20.96 | 3.99 | 0 | 0.68 | 5.37E-07 | 4.90E-07 [4.07E-07-6.31E-07] | 2.29E-06 | 1.95E-06 [1.82E-06-2.14E-06] | |
Weibull | 27.65 | 5.30 | 0 | 0.43 | 5.50E-07 | 1.20E-06 [8.51E-07-1.86E-06] | 2.45E-06 | 3.39E-06 [3.09E-06-4.17E-06] | |
Weibull | 24.47 | 4.66 | 0 | 0.76 | 3.98E-07 | 6.76E-07 [5.25E-07-7.76E-07] | 1.45E-06 | 2.14E-06 [1.95E-06-2.29E-06] | |
Weibull | 26.43 | 5.09 | 0 | 0.76 | 4.07E-07 | 9.12E-07 [7.94E-07-1.10E-06] | 1.51E-06 | 2.63E-06 [2.51E-06-2.82E-06] |
Concentration-response function | EC01 | EC10 | ||||||||
---|---|---|---|---|---|---|---|---|---|---|
mixture | RM | predicted M | observed M [CI] | predicted M | observed M [CI] | |||||
logit | 18.07 | 3.42 | - | 0 | 0.15 | 2.69E-07 | 8.71E-07 [1.12E-08-7.24E-06] | 2.69E-05 | 8.32E-06 [2.63E-06-7.08E-05] | |
probit | 11.05 | 2.14 | - | 0 | 0.15 | 3.89E-07 | 1.35E-06 [2.57E-08-5.89E-06] | 3.89E-05 | 1.07E-05 [3.80E-06-1.10E-04] | |
logit | 16.80 | 3.19 | - | 0 | 0.15 | 5.62E-07 | 7.94E-07 [2.63E-09-5.75E-06] | 1.10E-04 | 8.91E-06 [3.98E-06-3.72E-05] |
Concentration-response function | EC01 | EC10 | |||||||
---|---|---|---|---|---|---|---|---|---|
mixture | RM | Predicted M | Observed M [CI] | Predicted M | Observed M [CI] | ||||
logit | 19.33 | 3.80 | 0 | 0.59 | 1.02E-06 | 6.92E-07 [5.37E-07-1.74E-06] | 3.09E-06 | 3.09E-06 [2.75E-06-5.62E-06] | |
logit | 16.88 | 3.36 | 0 | 0.58 | 1.07E-06 | 5.89E-07 [3.31E-07-1.62E-06] | 3.55E-06 | 3.09E-06 [2.45E-06-5.62E-06] | |
Weibull | 26.92 | 5.72 | 0 | 0.19 | 4.47E-06 | 6.03E-06 [3.80E-06-8.13E-06] | 1.23E-05 | 1.74E-05 [1.17E-05-2.29E-05] | |
Weibull | 79.24 | 16.21 | 0 | 0.22 | 3.72E-06 | 8.32E-06 [4.37E-06-9.12E-06] | 1.12E-05 | 1.20E-05 [1.15E-05-1.91E-05] |
In the present study mostly additive effects of pesticide mixtures at low effect concentrations in the YES, the ERα CALUX as well as the ERβ CALUX assays were observed. The predicted EC01/10 values not lying in the range of the 95% CIs of the experiments in the ERα CALUX (
There was a high level of concordance between the ERα CALUX and the YES assay regarding the identification of estrogenic effects of single substances. With the exception of pirimicarb and propamocarb, all tested pesticides trans-activated the hERα in both test systems. Pirimicarb was negative in both bioassays. There were certain differences in the identification of estrogenic and anti-estrogenic activity in the selected reporter gene assays: Propamocarb was identified as an estrogenic compound with an effect at the hERα in the ERα CALUX assay but not in the YES assay. Both substances were applied alone in the YES and in the ERα CALUX assay and in combination with a saturating concentration of 1 nM E2 in the YES assay. As known from the scientific literature, pirimicarb [
We detected an anti-estrogenic activity of fenhexamid and fludioxonil when applied in combination with E2 in the YES assay (
Supramaximal estrogenic effects of substances eliciting higher putative effects than E2 itself were observed (
To our knowledge, this is the first study to test mixtures of pesticides for additive effects not only at the hERα but also at the hERβ level. Most pesticides being active at the hERα were also active at the hERβ, and additive effects were observed when applied as mixtures. Although the pesticide mixtures were less efficient at the hERβ, one should analyze their effects at both receptor subtypes, since after an exposure one would expect that the compounds interact with both receptors in humans, the receptor subtype ratio influencing the estrogenic outcome. Additive effects of the tested pesticides at the hERα were mostly observed in the YES assay as well as in the ERα CALUX assay, an observation that supports the assumption of additivity of pesticides sharing a mode of action. The used
(TIFF)
(TIFF)
(TIFF)
Regression models with 95% confidence bands; dashed end of the regression model line stands for concentrations at which the turbidity of the yeast suspension was reduced; S4A–S4F Fig show experiments in the YES assay with (A) 1 mM chlorpyrifos applied together with 1 nM E2 and increasing concentrations of 4-hydroxytamoxifen; (B) 1 mM chlorpyrifos applied together with 1 nM E2 and increasing concentrations of ICI 184,780; (C) 100 μM fenarimol applied together with 1 nM E2 and increasing concentrations of 4-hydroxytamoxifen; (D) 100 μM fenarimol applied together with 1 nM E2 and increasing concentrations of ICI 184,780; (E) 100 μM fenarimol applied together with increasing concentrations of 4-hydroxytamoxifen; (F) 100 μM fenarimol applied together with increasing concentrations of ICI 184,780; (G) 60 μM fenhexamid applied together with increasing concentrations of tamoxifen were tested in the ERα CALUX assay.
(TIFF)
for an iso-effective binary mixture of fludioxonil and fenhexamid in the ERα CALUX assay, based on their individual EC10 values.
(PDF)
(PDF)
Iso-effective mixtures based on EC01/EC10 or EC101/EC110 values of the single compounds.
(PDF)
(PDF)
(PDF)
(PDF)
(PDF)
(PDF)
(PDF)
(PDF)
(PDF)
This study was financially supported by the Federal Institute for Risk Assessment in Berlin, Germany (grant FK.3 1329–484 6433898).