After a variety of studies showed that certain chemicalsmight disrupt the sex hormonal systems of wildlife andhumans, the Organization for Economic Cooperationand Development (OECD), European countries, the United States and Japan initiated efforts to develop and validate assays and screen chemicals for their potential todisrupt the endocrine system of human and wildlife (CEC, 2004; EPA, 1998; Gelbke et al., 2004). Under theOECD activity, ‘‘Conceptual Framework for the Testingand Assessment of Endocrine Disrupting Chemicals’’ wasdeveloped. This framework has five components or levelseach corresponding to a different level of biological complexity (OECD, 2002). The in vitro assays such as receptorbinding assay are placed in level 2 in this framework toprovide mechanistic information and serve for screeningpurposes. As a tool to detect hormonal effects mediated thorough specific hormone receptors, especially for estrogen receptors (ERs), the application of the ER bindingassay has been investigated as a quick and cost-effective method.  Furthermore,  the  rodent uterotrophic assay, which has been placed in level 3 to provide data about single endocrine mechanism and effects, has also been recognized as an in vivo assay to detect estrogen receptormediated effects (Kanno et al., 2001; Owens and Koeter, 2003). 

Since the uterotrophic assay and ER binding assay bothdetect ER-mediated activities, a clear relationship betweenthe results from both two assays is expected. However, there are no studies that can demonstrate a clear relationship between two assays on a variety of chemical structures, although studies comparing at limited number ofchemical structures has been available (Takemura et al.,2005; Hong et al., 2005; Legler et al., 2002; Yamasakiet al., 2004). Since the receptor binding assay is consideredas one of the screening assays to determine if a chemicalhas potential to trigger receptor-mediated endocrine disruption such as ER and androgen receptor (AR). Theresults of receptor binding studies need to be related to biological relevance. The purpose of this study is to clarify therelationships between the results obtained in the in vitro ER binding assay and in vivo uterotrophic assay, and to determine the meaningful cutoff value from the ER bindingassay. For this purpose, we compared the results of receptor binding assay using human ERα(hERα) with immaturerat uterotrophic assay for 65 selected chemicals.

The chemical names, CAS numbers and sources of 65 test chemicals subjected in this study are listed in Table 1. Parabenes, benzophenones, biphenyls, diphenylmethanes, diphenylethylenes, phthalates, phenols and other chemicalstructure classes were included. All test chemicals used inthis study had more than 95% purity. Tris(hydroxy-methyl)aminomethane (Tris), phenylmethylsulfonyl fluo-ride (PMSF) and 17β-Estradiol were purchased from Sigma-Aldrich. Dimethylsulfoxide(DMSO), leupeptinhemisulfate monohydrate, sodium metavanadate, (NaVO3)dithiothreitol (DTT), glycerol and bovine serum albumin(BSA) were obtained from Wako Pure Chemical Industries,Ltd. and disodium salt dihydrate, ethylenediaminetetraace-tic acid (EDTA) and ethyleneglycol-bis(β-aminoethyl)-N,N,N',N'-tetraacetic acid (EGTA) were from Dojindo Laboratories Inc.

The receptor binding assay using recombinant humanestrogen receptor α(hERα) was conducted by the methodpreviously reported(Nakai et al., 1999; Yamasaki et al.,2004). Briefly, a recombinant hERα ligand binding domain(hERα-LBD) fused with glutathione-S-transferase (GST) was expressed in E. coli and purified using affinity chroma-tography technique. After the addition of sample solution(10μL) of varied concentration (1·10^-11 to 1·10^-4M as final concentrations) and [2,4,6,7,16,17-³H] 17β-estradiol([³H]-E2, 10μL; final conc. 0.5 nM, 155 Ci/mmol, Amar-sham Biosciences Corp.) in Tris-HCl (pH 7.4, 50μL) con-taining 1 mM EDTA, 1 mM EGTA, 1 mM NaVO₃,1mMDTT, 10% glycerol, 10 mg/mL BSA, 0.5 mM PMSF, and 0.2 mM leupeptin, a solution of recombinant hERα-LBD(30μL; final conc. 0.2 nM) was gently mixed. This mixture solution was incubated for 1 h at 25 C. Free radiolabelled ligands were removed by incubation with 0.4% dextran-coated charcoal (Sigma) (100μL) for 10 min at 4 C followed by filtration. The radioactivity of residual radiolabelled ligands bound to receptors in filtrate were measured by liquid scintillation counting. The assay was repeated more than three times for each test chemical.

The chemicals listed in Table 1 were tested in the immature rat uterotrophic assay according to the OECD drafttest guideline as previously reported (Yamasaki et al.,2004) in compliance with good laboratory practice (GLP).

Crj:CD (SD) IGS female rat pups (10-day old) purchased from Charles River Japan Inc. (Shiga, Japan) were weaned with their dams and individually housed until 19-days old. These immature female rats were weighted and weight-ranked to assign to each of the treated and control groups (6 rats/group). Three doses were used for each testchemical and the highest dose was set at the maximum tol-erance dose based on the results of dose-range finding test. The limit dose was set at 1000 mg/kg/day. Each group of six immature, 20-days old female rats received subcutaneous injections of a test chemical into their back for three consecutive days (4 mL/kg/day) for evaluation of estro-genic activity. The vehicle control group treated with oliveoil (s.c.) and the positive control group for estrogenic activity (s.c. 0.6μg/kg/day of 17α-ethynylestradiol (EE, Sigma)) were concurrently run for each test chemical. To evaluate anti-estrogenic activity in another set of treatment groups,EE (0.6μg/kg/day) was co-administered with the testchemical. The positive control group for anti-estrogenicactivity received subcutaneous co-administration of 1 mg/kg/day of tamoxifen (TAM, Sigma) and 0.6μg/kg/day of EE. The animals were sacrificed by bleeding from the abdominal vein under deep ether anesthesia 24 h after the final administration, and body weight and uterine weight of each animal were recorded.

The resulting data from the ER binding assay were analyzed using GraphPad Prism®, Version 4 (GraphPad Soft-ware, Inc.), and the IC₅₀ value for each test chemical wasobtained by logistic equation. The relative binding affinity (RBA) of each test chemical was calculated using the following equation:

The Dunnet test was used to analyze the data from theuterotrophic assay. When the significant increase of uterine weights in agonism assay or significant decrease in antago-nism assay were observed, the test chemical was evaluatedas estrogenic or anti-estrogenic, respectively.

The lowest effective dose (LED,µmol/kg/day), the low-est dose showing a statistically significant effect in thisassay, was employed as a quantitative parameter in thiscomparison study with logRBAs from the in vitro ER binding assay, and the correlation coefficients (r²) and its P values were calculated by GraphPad Prism® version 4 (GraphPad Software, Inc.). Also, contingency analyseswere performed to calculate concordance (the rate agree-ment of the results among assays), false-negative (the rateof negatives in the ER binding assay identified as positivein the uterotrophic assay) and false-negative rates (the rateof positives in the ER binding assay identified as negativesin the uterotrophic assay).

The results of the in vitro binding assay to hERα for 65 chemicals are shown in Table 1.

RBA values were obtained for 47 of 65 chemicals. The highest and lowest logRBAs were 1.57 of clomiphene citrate and 3.49 of diisononyl phthalate, respectively.

The remaining 18 chemicals were regarded as negatives (non-binders) in the concentration tested. Among them, although 4 chemicals, i.e., ethyl p-hydroxybenzoate, di(n-propyl)phthalate, 17α-methyltestosterone and 4,4’-dim-ethoxytriphenylmethane showed 20–50% displacementand they were regarded as negatives in this study.

Sixty-five chemicals were tested by immature rat utero-trophic assay in both estrogenic and anti-estrogenic assaysystems. The test chemical was evaluated as estrogenic or anti-estrogenic if the uterine weights were significantlyincreased in the estrogenic assay or decreased in the anti-estrogenic assay. In such cases, the log lowest effectivedoses (logLED,µmol/kg/day) were shown in Table 1.

Based on this evaluation, 31 and 25 chemicals were identified as estrogenic and anti-estrogenic in immature ratuterotrophic assay, respectively. Twenty-one, including alldiphenylmethanes tested, exhibited both estrogenic andanti-estrogenic responses. On the other hand, none of thephthalates tested in this study have either estrogenic noranti-estrogenic.

The logRBAs obtained from the in vitro ER bindingassay were compared with logLEDs from the uterotrophicassay. As shown in Fig. 1, the logRBAs were well correlated with both logLEDs in estrogenic and anti-estrogenicassay systems r²= 0.67 (n= 28, P< 0.0001) and 0.79 (n= 23, P< 0.0001), respectively. The lowest logRBA that can detect estrogenic or anti-estrogenic response in the uterotrophic assay was 2.63 (RBA = 0.00233%) of p-tert-butylphenol, respectively.

The results of ER binding and uterotrophic assays basedon the evaluation of the ER related response as detectable (positive) or not (negative) are compared in Table 2-1. Therates of concordance, false negative and false positive for all chemicals tested was 66%, 14% and 57%, respectively.

Fig. 1. Relationships between logLED and logRBA values. a, Relation-ship between estrogenic logLEDs and logRBAs b, Relationship between anti-estrogenic logLEDs and logRBAs

Fig. 2. Changes of indexes by contingency analysis depending on cutoff of RBA. The rates of concordance, false negative and false positive areshown as open circle with solid line, closed triangle with solid line andopen triangle with dash line, respectively.

Seventeen chemicals found to bind to the ER were neitherestrogenic nor anti-estrogenic in the uterotrophic assay.Among these chemicals, 3 benzothiazoles and 6 phthalateswere included, and the logRBAs of the remaining 8 chemicals were relatively weak ranging from –3.36 to –1.34. Fivechemicals were classified as non-binders that showed estro-genic or anti-estrogenic responses in the uterotrophic assay. All 3 non-ER binder chemicals that showed estrogenicity inthe uterotrophic assay were androgens (testosterone enan-thate and 17α-methyltestosterone) and p-diethylamino-benzaldehyde. The other 2 non-ER binders that showedanti-estrogenic effects were atrazine and benzanthrone.

Based on the comparison above, the ER binding assayseemed to have higher sensitivity than immature rat utero-trophic assay. In order to determine the lowest biologicallyeffective binding potency in the ER binding assay, the relationship between RBA cutoff value and the rates of concordance, false negative and false positive from contingencyanalysis were investigated (Fig. 2). As mentioned above,the lowest RBA that showed estrogenic or anti-estrogenicresponses in the uterotrophic assay was 2.63 (RBA = 0.00233) of p-tert-butylphenol in this study. Whenthis value is used as a cutoff value, the rates of concordanceand false positive rate ratios were refined at 82% and 23%,respectively without increasing the false-negative rate (Table 2-2). This cutoff achieved the best concordance andlowest false-negative ratios as shown in Fig. 2.