in vivo estrogenic potencies of 17β-estradiol, estrone, 17α-ethynylestradiol and nonylphenol

Aquatic Toxicology 66 (2004) 183–195

Comparative study on the in vitro/in vivo estrogenic potencies of 17␤-estradiol, estrone, 17␣-ethynylestradiol and nonylphenol K. Van den Belt, P. Berckmans, C. Vangenechten, R. Verheyen, H. Witters∗ VITO-Flemish Institute for Technological Research, Expertisecenter Environmental Toxicology, Boeretang 200 B-2400 MOL, Belgium Received 24 April 2002; received in revised form 4 September 2003; accepted 25 September 2003

Abstract The estrogenic activity of compounds was evaluated in a comparative approach both with in vitro and in vivo assays. By comparing simultaneously obtained experimental data, we evaluated the differences in response sensitivity (by EC10) and concentration–response relationships (including EC50) in order to get an idea about the predictive value of in vitro assays for in vivo estrogenic potencies or effects in fish. Two human estrogen receptor-based assays, the MVLN-assay (transformed MCF-7 human breast cancer cell line) and the yeast estrogen screen (YES-screen) were used for the in vitro evaluation of the estrogenic potencies. An in vivo model with the female zebrafish (Danio rerio) with plasma vitellogenin (VTG) as a biomarker for exposure and the ovarian somatic index (OSI) as an effect endpoint was used for the in vivo work. Compounds tested were 17␤-estradiol (E2), estrone (E1), 17␣-ethynylestradiol (EE2) and the alkylphenolic compound nonylphenol (NP). All compounds were found to be estrogenic in both in vitro assays and were able to induce VTG and to reduce the ovarian somatic index in female zebrafish. The MVLN-assay appeared up to 15 times more sensitive than the YES-screen. Concentration–response relationships, determined by EC10 and EC50 (concentration of test compound causing 10% or 50% effect compared to control) for VTG and OSI were of the same order of magnitude, indicating that VTG induction as an exposure biomarker can be predictive for effects on ovaries in females. We further demonstrated that for E1 and NP, the in vitro observed estrogenic potencies, based on EC50 values, were of the same order of magnitude as the in vivo estrogenic potencies. For EE2, a difference between in vitro and in vivo relative estrogenic potency was observed, being about 25 times more potent in vivo than could be expected based on the in vitro results. These experimental results showed the suitability of in vitro assays for screening purposes with qualitative assessment of estrogenicity, but they meanwhile point to the need of in vivo tests for an accurate hazard assessment for wildlife. © 2003 Elsevier B.V. All rights reserved. Keywords: 17␤-Estradiol; Estrone; 17␣-Ethynylestradiol; Fish; Cellular assays

1. Introduction The awareness that many compounds released into the environment have endocrine disrupting potencies, ∗ Corresponding author. Tel.: +32-14-335213; fax: +32-14-582657. E-mail address: [email protected] (H. Witters).

has grown over the past decade. The growing number of studies suggesting adverse effects of endocrine disruptors on reproduction and development in many wildlife species of fish (Munkittrick et al., 1991; Lyle et al., 1997), reptiles (Guillette et al., 1994; Sheehan et al., 1999), birds (Fry, 1995), mammals (Brouwer et al., 1989), has urged the development of screening and testing strategies for detecting adverse effects

0166-445X/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.aquatox.2003.09.004

184

K. Van den Belt et al. / Aquatic Toxicology 66 (2004) 183–195

on endocrine disruption in wildlife (EMWAT, 1997; Ankley et al., 1998; EDSTAC, 2000). A lot of research efforts have been put into the development of rapid in vitro assays to detect the estrogenic activity of chemicals. Examples of in vitro assays which have been successfully used to detect estrogenicity are the E-screen (Soto et al., 1995), the yeast estrogen screen (YES-screen) (Routledge and Sumpter, 1996), the MVLN-assay (Pons et al., 1990) and the ER-CALUX (Legler et al., 1999). These in vitro assays are rapid and cost effective and offer a suitable tool for the screening of numerous candidate pseudo-estrogens (Klotz et al., 1996; Zacharewski, 1997). In addition, they are more often used to assess the estrogenic potency of complex environmental matrices such as effluents of waste water treatment plants, surface waters or industrial effluents (Desbrow et al., 1998; Khim et al., 2001; Witters et al., 2001). The relevance of these assays for the evaluation of biological effects in intact organisms through xeno-estrogenic exposure might be limited since important processes such as uptake, bioaccumulation and metabolic activation or degradation of these compounds are not taken into account (Zacharewski, 1997; EMWAT, 1997; Tyler et al., 1998). Next to the in vitro work, a number of in vivo assays have been developed and applied for the assessment of endocrine disrupting activity of pure compounds or environmental samples. Within the group of in vivo test systems one might distinguish in vivo exposure and in vivo effect markers. The most frequently used in vivo exposure marker with regard to the aquatic environment, is the vitellogenin (VTG)-assay. VTG is the female yolk precursor protein which is synthesized in the liver and translocated to the ovaries (Bun and Idler, 1983). It is normally absent or present at very low levels in male fish but due to the constitutive presence of the gene, VTG can be induced upon estrogenic exposure (Sumpter and Jobling, 1995; Folmar et al., 1996; Jobling et al., 1996; Tyler et al., 1996). It was demonstrated that VTG can be a good biomarker for xeno-estrogenic exposure. However, the relationship between changes of VTG levels in both males or females and potential adverse impact on gonads, reproductive success or development of progeny is still uncertain. In vivo markers for effects on gonadal structure, reproductive function and long term reproductive success are most relevant and

should be identified (Arcand-Hoy and Benson, 1998). These studies however have a high cost and are time consuming. Therefore, hazard identification strategies for the determination of potential endocrine disrupting effects of new and existing compounds do recommend the use of both rapid in vitro and in vivo assays in the screening- prioritization step (EMWAT, 1997; EDSTAC, 2000). In vivo screening is considered essential in the screening-prioritization stage as such a system integrates metabolism and all potential modes of action of a compound. These tests should include endpoints such as sex steroid levels, VTG induction and gonado-somatical index (GSI) and they should have a duration of at least 7–21 dayss (EMWAT, 1997). In vitro screening assays and structure activity relationships are considered useful at this screening level of assessment, especially if a particular mode of action is suspected. They however can not stand on their own and in their present stage of development, they are not sufficient to exclude concern for endocrine modulating effects (EMWAT, 1997). With regard to the current hazard identification strategies, this study intended to contribute to comparative evaluation and selection of relevant in vitro and in vivo tests for a few known xeno-estrogenic compounds. The question is raised whether in vitro tests, for reasons of cost and ethical matters, could be an alternative to in vivo tests to screen for estrogenic activity and priority ranking of numerous compounds. For the in vitro work, the YES-screen (Routledge and Sumpter, 1996) and the MVLN-assay (Pons et al., 1990) were used. An in vivo model with the zebrafish (Danio rerio), with plasma VTG concentration as exposure marker and ovarian somatic index (OSI) as effect marker (Van den Belt et al., 2001, 2002), was selected to evaluate the in vivo potency of the selected compounds. The sensitivity and response spectrum of the different assays for the selected compounds was investigated, being used as a measure to evaluate the predictive value of in vitro observed estrogenic potencies for in vivo effects by comparison of concentration–response curves. Both EC10 (sensitivity) and EC50 (potency) values (concentration of test compound causing 10% or 50% effect compared to control) were calculated as parameters of the response curve. Experiments were simultaneously performed in the same lab, using similar stock solutions, cell culture systems and stock of

K. Van den Belt et al. / Aquatic Toxicology 66 (2004) 183–195

organisms to allow accurate comparison and exclude variability which is not related to the test system itself.

2. Materials and methods 2.1. Test compounds Stock solutions of estradiol (1,3,5[10]-estratriene3,17␤-diol; minimum 98%; Sigma), estrone (1,3,5[10]estratriene-3-ol-17-one; minimum 99%; Sigma), ethynylestradiol (17␣-ethynyl-1, 3, 5[10]-estratriene3, 17␤-diol; minimum 98%; Sigma) and nonylphenol (NP) 4-nonylphenol; 99% GC, mixture of isomers; Acros) were prepared. The compounds were dissolved in methanol (pro analysis; Merck) and further diluted in methanol or dimethyl sulfoxide for application in the test systems. For each of the compounds, the same stock solutions were used in both the in vitro and in vivo experiments. 2.2. In vitro assays 2.2.1. MVLN-assay The MVLN-cell line was developed at INSERM (France) and was provided by Prof. Dr. Pons. MVLN-cells are human breast cancer cells (MCF-7 cells) which were stably transfected with a pVit-tk-Luc plasmide (Pons et al., 1990). The transfected cells contain an human estrogen response element (ERE) which is coupled to a luciferase reporter gene. Binding of compounds to the estrogen receptor leads to gene transcription of the luciferase reporter gene and the production of the luciferase enzyme. After addition of the Luciferase Assay Reagents, the light production is measured with a luminometer as a measure for the receptor binding of the tested compound. Cells were cultured in medium containing 46.5% D-MEM (Gibco), 46.5% Nutrient Mixture Ham’s F-12 (Gibco), 5% fetal calf serum (FCS, HyClone), 1% penicilline/streptomycine (Gibco) and 1% non-essential amino acids (Gibco). They were incubated at 37 ◦ C, 5% CO2 and 95% relative humidity. About 24 h before addition of the test compounds, the cells were seeded in a 48-well plate (220 000 cells per well) using OPTIMEM (Gibco) with 5% charcoal stripped FCS. For each dilution of the selected

185

compounds, four replicate wells were made. The final concentrations in the wells and the applied dilution series were: 3.2 × 10−13 to 1 × 10−9 M with a dilution factor of 5 for both E2 and EE2; 1.6 × 10−12 to 5 × 10−9 M with a dilution factor of 5 for E1; 1.25 × 10−7 to 4 × 10−6 M with a dilution factor of 2 for NP. On each plate, both a control and solvent control (0.1% methanol), four replicates each, were added. The plates were incubated for 19–20 h. After incubation, the cells were treated with Reporter Lysis Buffer, the plate was shaken for 25 min and afterwards frozen at −80 ◦ C. To analyze, the cells were thawed and the well contents were transferred to a conical 96-well plate for centrifugation. For the luciferase assay, 20 ␮l of the centrifugated cell lysates were transferred to a black flat-bottom 96-well plate and 100 ␮l of the Luciferase Assay Reagent was added. The luciferase activity was measured using a luminometer (Luminoskan, Labsystems). To assess toxicity of the compounds, the total protein content in 10 ␮l of the centrifugated cell lysates was measured using the Bradford Assay and treated cells were compared with control and solvent control conditions. All test compounds were tested in three independent assays. 2.2.2. YES-screen The YES-screen consists of an estrogen-inducible expression system in yeast cells (Saccharomyces cerevisiae). It was developed in the Genetics Department at Glaxo (UK) and was kindly provided by Prof. Dr. Sumpter. The DNA sequence of the human estrogen receptor is stably integrated into the yeast genome, which also contains expression plasmids carrying estrogen-responsive sequences controlling expression of the reporter gene lac-Z. Upon binding an active ligand, the estrogen-occupied receptor modulates gene transcription, and the reporter gene lac-Z is expressed producing the enzyme ␤-galactosidase, which is secreted into the medium. It then metabolizes the chromogenic substrate, chlorophenol red-␤-dgalactopyranoside (CPRG), from normally yellow into a red product, which can be measured by absorbance at 540 nm. The assay was performed according to the procedure described by Routledge and Sumpter (1996). Serial dilutions of the selected compounds were added in 96-well optically flat-bottom microtiter plates with six replicate wells per concentration. The

186

K. Van den Belt et al. / Aquatic Toxicology 66 (2004) 183–195

final concentrations in the wells and serial dilutions were: 1 × 10−8 to 6 × 10−13 M for E2; 2.5 × 10−7 to 3.2 × 10−12 M for E1 and 5 × 10−8 to 6 × 10−13 for EE2, each with a dilution factor of 5. For NP, a dilution series with factor 2 was used and the final concentration in the wells ranged from 8 × 10−6 to 6.25 × 10−3 M. Each plate contained one row (n = 6 wells) of blanks with assay medium only and one row with solvent (0.5% methanol for E2, E1, EE2; 0.5% DMSO for NP) added to the medium. After 3 dayss of incubation at 32 ◦ C, the plates were read with a microplate reader (Multiskan Ascent, Labsystems) at 540 nm in comparison with measurements at 620 nm for turbidity (yeast growth). Changes of turbidity in treated cells compared to control cells gave an indication of toxicity. The assays were repeated three times for each of the selected compounds. 2.3. In vivo studies Adult zebrafish (age approximately 6–8 months) were purchased from a commercial dealer (Huybrechts NV Belgium who imported the fish from commercial breeders in USA or Singapore). Fish were observed for clinical health for at least 2 weeks before the experiments were initiated. The animals were kept under semi-static conditions in tap water, which was aerated to reduce the chlorine level (pH 8.0–8.5; hardness 140–145 mg/l CaCO3 ). The fish remained at a mean density of 1 g fish/l in 50 l glass aquaria, provided with an Eheim biofilter (Belcopet, Brugge, Belgium) at a temperature of 25–28 ◦ C and a 12 h dark/12 h light regime. The fish were fed a combination of dry flake food (Tetra brand) and live brine shrimp nauplia, Artemia salina or waterfleas, Daphnia magna on a daily basis. Groups of seven females and three males were exposed for a period of 3 weeks in 8 l full glass aquaria using aerated tap water as dilution medium. The three males were added to the all female experimental groups in order to avoid effects of a 3 weeks sex separation period on the ovarian somatical index. Based on results from previous dose-range finding studies with the selected test compounds and their reported in vitro estrogenic potencies, the following concentrations were used: 0–0.028–0.057–0.11–0.46–0.91–1.83 nM E2; 0–0.11–0.46–0.91–1.83–3.66 nM E1; 0–0.0071– 0.014–0.028–0.057–0.11 nM EE2 or 0–0.14–0.57–

1.13–2.27 ␮M NP. The methanol concentration in the aquaria was 10 ␮l/l and possible effects of the solvent on OSI or VTG were evaluated in a separate solvent control group. All exposures were performed in duplex (n = 2 experiments). Five dayss before the end of the exposure, the males were removed from the aquaria in order to avoid spawning of the females (for standardization purposes). After the 3 weeks exposure, all female fish were killed with 500 mg/l MS222 (3-aminobenzoicacidethylester, Sigma), weighed and from each individual about 5–15 ␮l of blood was collected from the caudal vein. Blood was collected in freshly Li-heparinised (300 U/ml) hematocrit micro capillary tubes, rinsed with a 1 mM solution of the protease inhibitor phenylmethylsulfonyl fluoride (Sigma), and centrifuged at 11 000 rpm for 10 min. After centrifugation, the plasma was retained, diluted 1/100 in phosphate buffered saline (PBS, Gibco) and kept at −20 ◦ C until it was analyzed for VTG. After the collection of blood, fish were dissected and the gonads were weighed by use of a precision balance (Perkin-Elmer, Autobalance AD-2, 0.001 mg). The gonado-somatic indices were calculated as follows: GSI = (weight of the gonads in mg)/(weight of the total body in mg) × 100. 2.4. Total protein analyses The total protein concentrations in diluted plasma samples (zebrafish assay) and centrifugated cell lysates (MVLN-assay) were determined using the Bradford Assay (Bio-Rad). This assay is based on the observation that the absorbance maximum for an acidic solution of Coomassie Brillant Blue G-250 shifts from 465 to 595 nm when binding to protein occurs. The measurements were performed in 96-well plates, using 10 ␮l of the samples (1/100 diluted plasma or centrifugated cell lysates). 2.5. VTG analyses A combination of gel electrophoresis and densitometry was used to measure VTG in the collected plasma samples. A detailed description of this methodology with performance characteristics is given in one of our previous publications (Van den Belt et al., 2003). Electrophoresis was carried out on 4–20% Tris–glycine gels (NOVEX) in a minisystem

K. Van den Belt et al. / Aquatic Toxicology 66 (2004) 183–195

(Xcell, NOVEX). Plasma samples were diluted in sample buffer (Tris–glycine, NOVEX), heated for 5 min at 100 ◦ C and gels were loaded with 5 ␮g protein per lane per sample compared to MW standards (NOVEX). Gels were run under non-reducing conditions for 80 min at 125 V. Next, the gels were stained with 0.1% Coomassie Blue for 30 min and destained in tap water for about 2 dayss. A Western blot using a rabbit anti-carp-VTG polyclonal antibody (Brunel University, UK) that cross reacts with zebrafish VTG (Tyler et al., 1996) allowed to identify two large protein bands with a molecular weight of respectively 193 and 134 kDa. These bands were recognized as VTG in the plasma protein profile of fish enriched with VTG due to intra-peritoneal injection with E2 (10 mg E2/kg per week during 3 weeks) while absent in sham-injected fish. To quantify VTG in the experimental samples, gels were analyzed using the Kodak Image Analyses version 3 software package. The gels were scanned with a CCD camera (SONY), stored under UVP Image store 500 and the sum intensity of the bands identified as VTG (193–134 kDa) was determined. Using a standard curve with BSA (0.078–2.5 ␮g BSA per lane), the VTG content of the selected bands was quantified and the percent of VTG compared to the total plasma protein content was calculated. 2.6. Chemical analyses The fish were exposed to the test compounds for 3 weeks under semi-static conditions with a daily renewal of 80% of the test volume in order to keep the actual concentration as close as possible to the nominal ones. In order to evaluate the loss of the compound over a 24 h period, the actual concentration of each compound was monitored in three exposure conditions with high, medium and low levels: 1.84–0.46–0.057 nM for E2; 3.66–0.92–0.11 nM for E1; 0.11–0.029–0.014 nM for EE2 and 2.27–0.57– 0.14 ␮M for NP. During the 3 weeks exposure period, once a week, samples were taken directly after a renewal and again before the next renewal. Subsequent dilutions of the stock solutions of the different compounds used for the spiking of the aquaria were also analyzed. E2, EE2 and E1 were analyzed using liquid chromatography and mass spectrometry (LCMS). The LC (Hewlett-Packard GmbH) consisted

187

of a HP 1100 series quaternary pump, a vacuum degasser and a thermo stated column compartment. Using a Gilson sampler changer (model 222), 50 ml sample fractions were extracted with an OSP-2 on a C18 cartridge (LiChrospher 60 RP-18, 10 ␮m). The eluted samples (70/30 methanol/water) were divided on an analytical column (Alltima, C18, 5 ␮m) and analyzed by MS (Micromass Limited), type quatro II with an electro spray interface. The method linearity was evaluated by putting 50 ml water samples on the analytical column through on-line solid-phase extraction. The method was linear up to 300, 500 and 1000 ng/l for respectively EE2, E2 and E1. The LOD (3× blanc) was 1 ng/l for E1, 3 ng/l for EE2 and 7 ng/l for E2. Concentrations of E2, E1 or EE2 in solvent control samples were all below the limit of detection. NP was analyzed using fluid chromatography with UV-detection. The fluid chromatograph consisted of a L6200A pump, a L5025 thermo stated column, an AS-4000 auto sampler and a L4500 diodearray detector (Merck Hitachi). NP was separated on a reversed phase column (Alltima C18, 5 ␮m) and quantified at 275 nm. For samples with high NP content (100–2000 mg/l), 10 ␮l samples were injected on the analytical column. The method was linear up to 2500 mg/l. For samples with low NP content (10– 1000 ␮g/l), 25 ml sample fractions were extracted with an OSP-2 on a C18 cartridge (LiChrospher 60 RP-18, 10 ␮m). The total extracts were transferred to the analytical column. The method was linear up to 1200 ␮g/l. The LOD (3× blanc) was 13 ␮g/l. 2.7. Statistics Two way ANOVA was used to detect significant differences between and within the replicate experiments. Data were analyzed using Statistica version 5.1 (Statsoft Inc., USA). Correlation analyses between OSI and VTG data were performed with the Excel software package (Microsoft, USA). To evaluate the response sensitivity and estrogenic potency, respectively the 10% (EC10) and 50% (EC50) effect concentrations were calculated for the different endpoints using non linear regression with a sigmoid curve fitting (Graphpad Software Inc., version 2.01, 1994–1995). For the calculation of the EC values for OSI, the OSI data were transformed to percent reduction compared to the average of the control fish. All

188

K. Van den Belt et al. / Aquatic Toxicology 66 (2004) 183–195

calculations of EC10 and EC50 were performed for each of the replicate experiments with in vitro assays (n = 3) or replicate in vivo experiments (n = 2).

3. Results 3.1. Exposure conditions The actual concentrations measured in the different serial dilutions of stock solutions, used for the spiking of the aquaria, were 61–83% for E2, 91–100% for E1, 86–100% for EE2 and 82–91% for NP of the nominal concentrations. In the water samples collected directly after a renewal, the actual concentrations at the time of analyses were only 23–60%, 30–68%, 61–100% and 20–62% of the nominal concentrations for respectively E2, E1, EE2 and NP. Actual concentrations found in samples collected 24 h after a renewal decreased to

13–45%, 13–31%, 25–73% and 16–37% of the desired nominal concentrations for respectively E2, E1, EE2 and NP. In general, the actual concentrations of EE2 corresponded better with the desired nominal concentrations compared to this for E2, E1 and NP. Since actual concentrations measured in the stock solutions of all test compounds were high (>60–100% of nominal concentrations), these differences confirm the environmental persistence of EE2 (Ternes et al., 1999a,b). The reduction in actual concentration after 24 h could be explained by uptake by the fish together with degradation and adsorption to aquaria and debris. 3.2. Assay results The results of both the YES-screen and the MVLN-assay for the selected compounds are presented in Fig. 1. In both in vitro assays, no indications of toxicity of the selected compounds were found.

Fig. 1. Results of the YES (a) and the MVLN (b) in vitro screenings assay for estrogenic activity for E2, E1, EE2 and NP expressed as percent of maximum response, based on the re-calculated testsignal for E1, EE2 and NP as a function of positive control E2 (average ±S.E., n = 3 replicate experiments).

K. Van den Belt et al. / Aquatic Toxicology 66 (2004) 183–195

189

Table 1 In vitro assay results In vitro

E2

E1

EE2

NP

YES EC10 (nM) EC50 (nM)

0.039 ± 0.007 0.153 ± 0.032

0.062 ± 0.057 0.390 ± 0.057

0.036 ± 0.007 0.174 ± 0.040

403 ± 22 1276 ± 58

MVLN EC10 (nM) EC50 (nM)

0.003 ± 0.0002 0.015 ± 0.002

0.010 ± 0.002 0.078 ± 0.014

0.002 ± 0.0006 0.010 ± 0.004

128 ± 12 463 ± 14

The EC10 and EC50 (average ± S.E., n = 3 replicate experiments), calculated based on the in vitro assay results, are given for the different compounds tested.

The results show that all compounds have the ability to bind to the estrogenic receptor and therefore can be considered estrogenic. However, from the position of the different curves it is clear that there is a difference in response for gene transcription after ER-receptor binding. For each of the replicate assays (n = 3), EC10 and EC50 values were calculated and average values with standard error for the different compounds tested are given in Table 1. Based on the calculated EC10 and EC50 values, the MVLN-assay showed higher sensitivity and potency for the four compounds tested. The in vivo concentration–response studies with female zebrafish were performed in duplex and since two-way ANOVA demonstrated no significant difference between the two experiments for the different compounds tested, the data from the separate experiments were pooled for presentation in Fig. 2. The EC10 and EC50 values given in Table 2 are averages with standard errors of the values calculated for the individual replicate experiments (n = 2).

It is demonstrated that an exposure to the natural female steroids E2 and E1, the synthetic estrogen EE2 or the alkylphenol NP, resulted in a significant concentration-dependent reduction in OSI compared to control females after a 3 weeks exposure period. The average OSI observed for the solvent control females was not significantly different from the different average values for control females indicating that the observed reduction of OSI in treatment groups can not be caused by using 0.1% methanol as solvent. The concentration at which a significant reduction in OSI was observed (open bars) was different for the selected compounds. For the natural steroids these concentrations were 0.46 and 3.66 nM for respectively E2 and E1. For EE2, a concentration as low as 0.029 nM was sufficient to significantly reduce the OSI. In contrast, for NP a significant OSI reduction was observed at an exposure concentration of 2270 nM. In addition to the in vivo effect marker OSI, the effects on the in vivo exposure marker for estrogenic exposure, VTG, are also presented in Fig. 2.

Table 2 In vivo assay results In vivo

E2

E1

EE2

NP

VTG EC10 (nM) EC50 (nM)

0.303 ± 0.101 0.642 ± 0.048

0.514 ± 0.191 0.753 ± 0.115

0.011 ± 0.003 0.021 ± 0.0007

699 ± 80 1121 ± 115

OSI EC10 (nM) EC50 (Nm)

0.208 ± 0.073 0.880 ± 0.227

0.721 ± 0.605 1.720 ± 0.660

0.023 ± 0.0007 0.027 ± 0.0006

816 ± 479 1337 ± 403

−0.61 (n = 91; P < 0.001)

−0.40 (n = 74; P < 0.001)

−0.52 (n = 61; P < 0.001)

−0.63 (n = 53; P < 0.001)

OSI − VTG Correlation, r

The EC10 and EC50 (average ± S.E., n = 2 replicate experiments) calculated based on the in vivo assay results, and the correlation coefficient, r between OSI and VTG are given for the different compounds tested.

190

K. Van den Belt et al. / Aquatic Toxicology 66 (2004) 183–195

Fig. 2. Results of the in vivo exposure of adult female zebrafish. Effects of a 3 weeks exposure to E2, E1, EE2 or NP on OSI and plasma VTG levels are presented, pooled data from two experiments (average ± S.E., n = 14 fish) (open bars and symbols indicate significant differences compared to control values: P < 0.05 with one-way ANOVA).

K. Van den Belt et al. / Aquatic Toxicology 66 (2004) 183–195

191

Table 3 Response sensitivities EC10 ratio

E2

E1

EE2

NP

YES/MVLN OSI/VTG OSI/YES OSI/MVLN VTG/YES VTG/MVLN

13.2 [8.1–20.6] 0.7 [0.3–1.4] 5.4 [2.5–10.0] 70.9 [38.3–108.5] 7.8 [3.8–14.3] 103.5 [58.7–155.8]

5.0 [1.3–10.7] 1.4 [0.2–4.1] 11.7 [1.3–61.7] 58.8 [7.3–152.4] 8.4 [3.5–32.9] 42.0 [20.2–81.3]

14.5 [6.4–47.7] 2.1 [1.6–3.0] 0.60 [0.50–1.1] 9.3 [6.5–26.7] 0.30 [0.19–0.61] 4.4 [2.4–14.9]

3.1 1.2 2.0 6.4 1.7 5.4

[2.5–4.0] [0.43–2.1] [0.79–3.4] [2.2–12.1] [1.4–2.2] [4.1–7.3]

Response sensitivities of the used assays for the different compounds tested, expressed as ratio’s of average EC10 values (n = 2 or 3). Ranges are calculated as follows: (minimum EC10 YES/maximum EC10 MVLN–maximum EC10 YES/minimum EC10 MVLN), etc. of the n = 2 or 3 individual EC10 values.

A significant concentration-dependent increase in the plasma VTG levels in female zebrafish was observed upon exposure to all of the selected compounds. Correlation analyses demonstrated that for all compounds tested, the observed induction of VTG was significantly correlated with the observed reduction in OSI (see Table 2). As for the in vitro assays, EC10 and EC50 values were calculated for the two separate in vivo experiments and averages with standard errors are given in Table 2. The YES/MVLN and OSI/VTG EC10 ratios were calculated to evaluate which of the respectively in vitro and in vivo assays were the most sensitive for the different compounds tested (Table 3). The other EC10 ratios given in Table 3 were calculated to compare the sensitivity of the selected in vivo assays with the sensitivity of the in vitro assays used. The estrogenic potency relative to E2 of the different compounds tested, is given in Table 4. This relative estrogenic potency is expressed as the EC50-E2/EC50-compound X ratio for the different compounds as observed in the different assays. The ratios in Tables 3 and 4 were calculated based on the average of the different EC values calculated for the individual assays. To allow an evaluation

of the variability of these ratios, minimum–maximum ranges were calculated.

4. Discussion The objective of this comparative study was to evaluate and compare the sensitivity and estrogenic potency measured by both in vitro and in vivo assays for a number of selected compounds with known estrogenic activity. All of the tested compounds showed a clear estrogenic activity in both selected in vitro assays (Fig. 1). The MVLN-assay appears to be more sensitive than the YES-screen for all test compounds, based on EC10 values. Depending on the compound tested, the MVLN-assay is about 3–15 times more sensitive than the YES-screen (Table 3). Factors such as cell wall permeability, differences in receptor levels and protein proteolysis mechanisms, non-receptor cell-specific factors, metabolic capabilities, multidrug resistance, efflux pumps and endogenous yeast binding proteins have been suggested as possible explanations for differences between yeast based assays and mammalian assays (Zacharewski, 1997). Although all

Table 4 Relative estrogenic potencies EC50 ratio

E2

E1

YES MVLN OSI VTG

1 1 1 1

0.4 0.2 0.5 0.8

[0.21–0.68] [0.12–0.30] [0.28–1.0] [0.69–1.1]

EE2

NP

0.9 [0.44–2.1] 1.6 [0.71–4.6] 32.6 [23.8–41.8] 30.6 [26.4–33.7]

0.0001 [0.00008–0.0002] 0.00003 [0.000002–0.00004] 0.0007 [0.0004–0.0012] 0.0006 [0.0005–0.0007]

Relative estrogenic potencies compared to E2 expressed as ratio’s (EC50 E2/EC50 compound) for average (n = 2 or 3) EC50 values. Ranges are calculated as follows: (minimum EC50 E2/maximum EC50 compound X–maximum EC50 E2/minimum EC50 compound X) of the n = 2 or 3 individual EC50 values.

192

K. Van den Belt et al. / Aquatic Toxicology 66 (2004) 183–195

compounds are estrogenic in both assays, they do show a difference in relative estrogenic potency, expressed as a shift to the higher or lower concentration range of the concentration–response curve (Fig. 1). Values obtained for relative estrogenic potencies (relative to EC50 of E2; Table 4) in the YES-screen, demonstrate that EE2 appears to be more or less equipotent to E2. E1 and NP are respectively about 2.5 and 10 000 times less potent than E2. For the MVLN-assay, analogue relative estrogenic potencies were found with EE2 being about 1.6 times more potent, E1 about five times less potent and NP about 33 000 times less potent than E2 (Table 4). The observed relative estrogenic potencies are in agreement with literature data. Coldham et al. (1997) reported a relative estrogenic potency compared to E2 (1) of 0.88 and 0.09 for respectively EE2 and E1 evaluated with a recombinant yeast cell bioassay. Reported relative potencies (E2 = 1) for NP in recombinant yeast cell assays were 0.0022 (Coldham et al., 1997) and 0.007 (Routledge and Sumpter, 1996). Legler et al. (2002) reported relative estrogenic potencies of 1.2, 0.2 and 0.000015 in the ER-CALUX for respectively EE2, E1 and NP. The latter data are highly comparable with those obtained with the MVLN-assay in this study. The current in vivo exposures demonstrated that all compounds tested, including the xeno-estrogen NP, are estrogenic in female zebrafish since a highly significant induction of the estrogen-specific biomarker VTG was found (Fig. 2). From both the position of the VTG induction curves in Fig. 2 and the EC50 ratios given in Table 4, it is clear that EE2 is a very potent inducer of VTG with a relative potency of about 30 times higher than E2. E2 and E1 were found almost equipotent for induction of VTG while the relative potency for NP was much lower (Table 4). The high potency of EE2 for VTG induction has also been reported in adult male rainbow trout (Oncorhynchus mykiss) by Jobling et al. (1996). An exposure to 0.0067 nM (2 ng/l) EE2 for a period of 3 weeks was sufficient to elevate the VTG levels in males a million-fold (Jobling et al., 1996). The results of a 3 weeks comparative response study with male fathead minnow (Panter et al., 1998) indicated that although E1 (0.118 nM or 31.8 ng/l) induced VTG at a lower concentration, E2 (0.367 nM or 100 ng/l) was more potent than E1 at higher concentrations, causing greater VTG induction and testicular inhibition. In a 14 days

exposure study with female juvenile rainbow trout, an EC50 for VTG induction of 0.055 nM (15 ng/l) was found for E2 (Thorpe et al., 2000), which is about 10 times lower than the EC50 for E2 (0.642 nM) observed in our study with zebrafish. The ability of NP to induce VTG is well documented in a number of fish species. Jobling et al. (1996) reported for NP a LOEC of 92 nM (20.3 ␮g/l) for induction of VTG in male rainbow trout which is about 7.5 times lower than the EC10 we observed in female zebrafish. In female juvenile rainbow trout an EC50 of 73 nM (16 ␮g/l) NP for VTG was found in a 14 days exposure study (Thorpe et al., 2000) which is about 15 times lower than the EC50 calculated for female zebrafish in the current study. It should be further elucidated to which extent differences in species sensitivity and sex specific differences can explain the differences for VTG induction between our observations on female zebrafish and other fish species mentioned in the literature. In addition to VTG, the effects of the selected compounds on OSI, a commonly used endpoint in reproduction studies with fish (Arcand-Hoy and Benson, 1998) were evaluated. The OSI values in control fish are comparable to those reported in zebrafish by others (Tyler et al., 1996; Örn et al., 1998). Significant reduction of OSI was observed after the exposure to any of the selected compounds, including the non-steroidal compound NP (Fig. 2). The relative potencies expressed as EC50 ratios given in Table 4 indicate that E1 is two times less potent than E2 for causing OSI reduction while EE2 is about 30 times more potent than E2. Opposite to this, NP was found to be about 1400 times less potent than E2 for causing similar effects on OSI. The ovaries of females with severely reduced OSI (<5%) were characterized macroscopically by a reduction in size and an absence of maturing oocytes, which is likely to impair the future reproduction success of these females. The observed effects on OSI and ovarian macroscopic structure are in correspondence with previous studies with EE2 (Van den Belt et al., 2001, 2002). Miles-Richardson et al. (1999) too, showed that the ovaries of females exposed to 10 nM E2 for 2 weeks were characterized by a predominance of pre-vitellogenic oocytes and atretic follicles. Comparison of the sensitivity of both in vivo endpoints, based on the EC10 ratio (Table 3) indicates that both endpoints show approximately the same sensitivity. Investigations that focus on a

K. Van den Belt et al. / Aquatic Toxicology 66 (2004) 183–195

relationship between biomarker changes (e.g. VTG) and reproductive effects (e.g. OSI) could provide useful information to model the impact of xeno-estrogens for risk assessment purposes (Arcand-Hoy and Benson, 1998; Kramer et al., 1998). The observed decrease in OSI following exposure to the different compounds was significantly (P < 0.001) correlated to the increase in VTG levels. This is a confirmation of similar concentration–response curves and demonstrates that besides being a marker for in vivo estrogenic exposure, VTG can potentially be used as a predictive marker for effects on ovarian structure. Results reported by Kramer et al. (1998) and our previous studies (Van den Belt et al., 2001, 2002) support these findings providing evidence for disruption of ovarian function. In fathead minnows exposed to E2, Kramer et al. (1998) reported a significant correlation between VTG induction in females and reduced egg production. A significant relationship between reduced OSI and reduced spawning capacity in zebrafish was observed in a 3 weeks exposure study with EE2 (0, 5, 10 and 25 ng/l) (Van den Belt et al., 2001). The value of in vitro screening assays as a predictive screening tool for in vivo estrogenic activity and effects has been shown. Indeed they detected and confirmed the estrogenicity of the compounds as observed through studies with fish and did not show false negative results. However, our results demonstrated that when risk assessment is requested, in vivo testing is essential given the observed quantitative differences between in vitro and in vivo testing, which are likely the consequence of intrinsic characteristics of the test. Besides indicating that a compound can bind to the E2 receptor, one of the important features of in vitro screening assays is the evaluation of the estrogenic potency relative to E2. The relative estrogenic potencies presented in Table 4 show that for EE2 the in vitro and in vivo relative estrogenic potencies differ with about a factor 30. This difference is less pronounced for NP while for E1, there is a fair correspondence between the in vitro and in vivo relative estrogenic potencies. The difference between the in vitro and in vivo relative estrogenic potencies for E1 are of the same order of magnitude as the differences between the minimum and maximum ratios for the different assays. Legler et al. (2002) reported analogue findings with EE2 being 100 times more potent than E2 in a transgenic zebrafish model while only 1.2 times more potent in

193

the ER-CALUX. The higher in vivo estrogenic potency of EE2 compared to E2, can be explained by the higher binding affinity of EE2 for the E2 receptor as observed in the channel catfish (Nimrod and Benson, 1997). Furthermore, the presence of the 17␣-ethynyl group makes EE2 less susceptible for metabolisation than endogenous steroids and it is subject to intensive enter hepatic recycling (Guengerich, 1990). The results from this comparative study showed that in vitro screenings assays for estrogenic activity can qualitatively assess in vivo estrogenic activity of compounds for fish. However, the predictive value, based on EC10 and EC50 values of in vitro assays might be quantitatively less accurate for compounds where metabolic properties play a crucial role during in vivo exposures. The limitation of in vitro assays is also that they do not account for differences in uptake of the compound, bioaccumulation or interactions involving the induction of binding proteins such as sex hormone binding globulins that may modulate the uptake and metabolisation of sex steroids or steroid mimics (Zacharewski, 1997). It is evident that this comparative approach should be performed with an extended list of more xeno-estrogenic and non-estrogenic compounds in order to gather data, to make correlations and eventually develop prediction models. Until now, it remains a prerequisite that in vivo studies are performed for hazard and risk assessment of potential estrogen disrupting compounds.

Acknowledgements The author received a PhD fellowship at Vito, Mol in Belgium. We like to thank G. Van Ermen and A. Borburgh for scientific advice and R. Mannaerts for the chemical analyses, performed in the Vito laboratories for analytical measurements. We are also grateful to G. Geukens and B. Roosen for their technical support. The authors finally thank Dr. Tyler for the supply of the carp-VTG antibody.

References Ankley, G., Mihaich, E., Stahl, R., Tillitt, D., Colborn, T., MacMaster, S., Miller, R., Bantle, J., Campbell, P., Denslow, N., Dickerson, R., Folmar, L., Fry, M., Giesy, J., Gray, E.L.,

194

K. Van den Belt et al. / Aquatic Toxicology 66 (2004) 183–195

Guiney, P., Hutchinson, T., Kennedy, S., Kramer, V., LeBlanc, G., Mayes, M., Nimrod, A., Patino, R., Petersen, R., Purdy, R., Ringer, R., Thoams, P., Touart, L., Van Der Kraak, G., Zacharewski, T., 1998. Overview of a workshop on screening methods for detecting potential (anti) estrogenic/androgenic chemicals in wildlife. Environ. Toxicol. Chem. 17, 68–87. Arcand-Hoy, L.L., Benson, W.H., 1998. Fish reproduction, an ecologically relevant indicator of endocrine disruption. Environ. Toxicol. Chem. 17, 49–57. Brouwer, A., Reijnders, P.J.H., Koeman, J.H., 1989. Polychlorinated biphenyl (PCB)-contaminated fish induces vitamin A and thyroid deficiency in the common seal (Phoca vitulina). Aquat. Toxicol. 15, 99–106. Bun Ng, T., Idler, D.R., 1983. Yolk formation and differentiation in teleost fishes. In: Hoar, W.S., Randall, D.J. (Eds.), Fish Physiology, vol. IXA. Academic Press, USA, pp. 373–397. Coldham, N.G., Mehul, D., Sivapathasundaram, S., McDonnell, D.P., Conor, C., Sauer, M.J., 1997. Evaluation of a recombinant yeast cell estrogen screening assay. Environ. Health Perspect. 105, 734–742. Desbrow, C., Routledge, E.J., Brighty, G.C., Sumpter, J.P., Waldock, M., 1998. Identification of estrogenic chemicals in STW effluent. 1. Chemical fractionation and in vitro biological screening. Environ. Sci. Technol. 32, 1549–1558. EDSTAC, 2000. Report to Congress. Endocrine Disrupting Screening Program-Environmental Protection Agency. EMWAT, 1997. In: Tattersfield, L., Matthiesssen, P., Campbell, P., Grandy, N., Länge, R. (Eds)., SETAC-Europe, OECD, EC Expert Workshop on Endocrine Modulators and Wildlife: Assessment and Testing, Veldhoven, The Netherlands, 10–13 April 1997. SETAC-Europe, Brussels, Belgium, p. 126. Folmar, L.C., Denslow, N.D., Rao, V., Chow, M., Crain, A., Enblom, J., Marcino, J., Guillette Jr. L.J., 1996. Vitellogenin inductions and reduced serum testosteron concentrations in feral male carp (Cyprinus carpio) captured near a major metropolitan sewage treatment plant, Environ. Health Perspect. 104, 1096–1101. Fry, D.M., 1995. Reproductive effects in birds exposed to pesticides and industrial chemicals. Environ. Health Perspect. 103 (Suppl. 7), 165–171. Guengerich, F.P., 1990. Metabolism of 17␣-ethynylestradiol in humans. Life Sci. 47, 1981–1988. Guillette, L.J., Gross, T.S., Mason, G.R., Matter, J.M., Percival, H.F., Wood-ward, A.R., 1994. Developmental abnormalities of the gonads and abnormal sex-hormone concentrations in juvenile alligators from contaminated and control lakes in Florida. Environ. Health Perspect. 102, 680–688. Jobling, S., Sheanan, D., Osborne, J.A., Matthiesen, P., Sumpter, J.P., 1996. Inhibition of testicular growth in rainbow trout (Oncorhynchus mykiss) exposed to alkylphenolic chemicals. Environ. Toxicol. Chem. 15, 194–202. Khim, J.S., Lee, k.T., Villeneuve, D.L., Kannan, K., Giesy, J.P., Koh, C.H., 2001. In vitro bioassay determination of dioxinlike and estrogenic activity in sediment and water from Ulsan bay and its vicinity, Korea. Arch. Environ. Contam. Toxicol. 40, 151–160.

Klotz, D.M., Beckman, B.S., Hill, S.M., McLachlan, J.A., Walters, M.R., Arnold, S.F., 1996. Identification of environmental chemicals with estrogenic activity using a combination of in vitro assays. Environ. Health Perspect. 104 (Suppl. 10), 1084– 1089. Kramer, V.J., Miles-Richardson, S., Pierens, S.L., Giesy, J.P., 1998. Reproductive impairment and induction of alkalinelabile phosphate, a biomarker of estrogen exposure, in fathead minnows (Pimephales promelas) exposed to waterborne 17␤estradiol. Aquat. Toxicol. 40, 335–360. Legler, J., van den Brink, C.E., Brouwer, A., Murk, A.J., van der Saag, P.T., Vethaak, A.D., van der Burg, B., 1999. Development of a stably transfected estrogen receptor-mediated luciferase reporter gene assay in the human T47D breast cancer cell line. Toxicol. Sci. 48, 55–66. Legler, J., Zeinstra, L.M., Schuitemaker, F., Lanser, P.H., Bogerd, J., Brouwer, A., Vethaak, A.D., De Voogt, P., Murk, A.J., van der Burg, B., 2002. Comparison of in vivo and in vitro reporter gene assays for short-term screening of estrogenic activity. Environ. Sci. Technol. 36 (20), 4410–4415. Lyle, C.M., Frid, C.L.J., Gill, M.E., McCormick, D., 1997. Abnormalities in the reproductive health of flounder Platichthyus flesus exposed to effluent from a sewage treatment works. Mar. Pollut. Bull. 34, 34–41. Miles-Richardson, S.R., Kramer, V.J., Fitzgerald, S.D., Render, J.A., Yamini, B., Barbee, S.J., Giesy, J.P., 1999. Effects of waterborn exposure of 17-␤ estradiol on secondary sex characteristics and gonads of fathead minnows (Pimephales promelas). Aquat. Toxicol. 47, 129–145. Munkittrick, K.R., Portt, C.B., Van Der Kraak, G.J., Smith, I.R., Rokosh, D.A., 1991. Impact of bleached kraft mill effluent on population characteristics, liver MFO activity and serum steroid levels of a lake Superior white sucker (Catostomus commersoni) population. Can. J. Fish Aquat. Sci. 48, 1371– 1380. Nimrod, A.C., Benson, W.H., 1997. Xenobiotic interaction with and alteration of channel catfish estrogen receptor. Toxicol. Appl. Pharmacol. 147, 381–390. Örn, S., Andersson, P.L., Förlin, L., Tysklind, M., Norrgren, L., 1998. The impact on reproduction of an orally administered mixture of selected PCBs in zebrafish (Danio rerio). Arch. Environ. Contam. Toxicol. 35, 52–57. Panter, G.H., Thompson, R.S., Sumpter, J.P., 1998. Adverse reproductive effects in male fathead minnows (Pimephales promelas) exposed to environmentally relevant concentrations of the natural oestrogens, oestradiol and oestrone. Aquat. Toxicol. 42, 243–253. Pons, M., Gagne, D., Nicolas, J.C., Mehtali, R., 1990. A new cellular model of response to estrogens: a bioluminescent test to characterize (anti)estrogen molecules. Biotechnology 9, 450– 459. Routledge, E.J., Sumpter, J.P., 1996. Estrogenic activity of surfactants and some of their degradation products assessed using a recombinant yeast screen. Environ. Toxicol. Chem. 15, 241–248. Sheehan, D.M., Willingham, E., Gaylor, D., Bergeron, J.M., Crews, D., 1999. No threshold dose for estradiol-induced sex reversal

K. Van den Belt et al. / Aquatic Toxicology 66 (2004) 183–195 of turtle embryos: how little is too much? Environ. Health Perspect. 107, 155–159. Soto, A.M., Sonnenschein, C., Chung, K.L., Fernandez, M.F., Olea, N., Serrano, F.O., 1995. The E-screen assay as a tool to identify estrogens an update on estrogenic environmental pollutants. Environ. Health Perspect. 103 (Suppl. 7), 113– 122. Sumpter, J.P., Jobling, S., 1995. Vitellogenesis as a biomarker for estrogenic contamination of the aquatic environment. Environ. Health Perspect. 103 (Suppl. 7), 173–178. Ternes, T.A., Stumph, M., Mueller, J., Haberer, K., Wilken, R.D., Servos, M., 1999a. Behavior and occurrence of estrogens in municipal sewage treatment plants. I. Investigations in Germany, Canada and Brazil. Sci. Total Environ. 225, 81– 90. Ternes, T.A., Kreckel, P., Mueller, J., 1999b. Behavior and occurrence of estrogens in municipal sewage treatment plants. II. Aerobic batch experiments with activated sludge. Sci. Total Environ. 225, 91–99. Thorpe, K.L., Hutchinson, T.H., Hetheridge, M.J., Sumpter, J.P., Tyler, C.R., 2000. Development of an in vivo screening assay for estrogenic chemicals using juvenile rainbow trout (Oncorhynchus mykiss). Environ. Toxicol. Chem. 19, 2812– 2820.

195

Tyler, C.R., van der Eerden, B., Jobling, S., Panter, G., Sumpter, J.P., 1996. Measurement of vitellogenin, a biomarker for exposure to oestrogenic chemicals, in a wide variety of cyprinid fish. J. Comp. Physiol. B 166, 418–426. Tyler, C., Jobling, S., Sumpter, J.P., 1998. Endocrine disruption in wildlife: a critical review of the evidence. Crit. Rev. Toxicol. 28, 319–363. Van den Belt, K., Verheyen, R., Witters, H., 2001. Reproductive effects of ethynylestradiol and 4t-octylphenol on the zebrafish (Danio rerio). Arch. Environ. Contam. 41 (4), 458–467. Van den Belt, K., Wester, P., van der Van, L., Verheyen, R., Witters, H., 2002. Effects of ethynylestradiol on the reproductive physiology in zebrafish (Danio rerio): time dependency and reversibility. Environ. Toxicol. Chem. 21 (4), 767–775. Van den Belt, K., Verheyen, R., Witters, H., 2003. Comparison of vitellogenin responses in zebrafish and rainbow trout following exposure to environmental estrogens. Ecotoxicol. Environ. Safety 56 (2), 271–281. Witters, H., Vangenechten, C., Berckmans, P., 2001. Detection of estrogenic activity in Flemish surface waters using a in vitro recombinant assay with yeast cells. Wat. Sci. Technol. 43, 117–123. Zacharewski, T., 1997. In vitro bioassays for estrogenic substances. Environ. Sci. Technol. 31, 613–623.