Estrogenicity of styrene oligomers and assessment of estrogen receptor binding assays.

Polystyrene is frequently used in resins, and the styrene dimers and trimers eluted from polystyrene have been reported to have estrogenic activity (1). We have performed a number of in vitro and in vivo tests [i.e., estrogen receptor (ER) and androgen receptor binding assays, thyroid hormone receptor binding assays, human breast cancer cell line MCF-7 proliferation assays (E-SCREEN), uterotrophic assays in immature and ovariectomized rats, Hershberger assays, and prolactin release assays and steroidogenesis] and found no effects of styrene dimers or trimers on sex hormones in any of these assays (2–7). These results are supported by Fail et al. (8), who reported that mixtures of styrene oligomers did not show any estrogenic activity in the immature rat uterotrophic assay and the reporter gene assay. In addition, the Japan Environment Agency referred to their studies (9) and removed the styrene dimers and trimers from their list of endocrine disruptors (9). However, Ohyama et al. (10) reported that high concentrations of certain styrene dimers and trimers showed estrogenic effects in an ER binding assay and in the E-SCREEN assay. Recently, several assay systems have been used to assess endocrine-disrupting effects, but a few of these assay systems can cause false-positive reactions when test compounds are at high concentrations (11). To assess the accuracy of the ER binding assay system and the results of Ohyama et al. (10), and to ascertain the safety of styrene dimers and trimers, we used a solubility test and three ER binding assays (12) (Table 1). The ER binding assay, which detects the direct reactivity of ligand to a receptor, is the most standardized and simple test system for the detection of specific mechanisms of estrogenic activity. Using the radiosotope method (Method RI) as described previously (13,14), we observed that styrene dimers and trimers did not show statistically significant inhibitory action against the binding of [3H]-17β-estradiol (E2) to ER. We used Method A to detect the binding affinities of test samples to human ERα (hERα). Using a fluorescence polarization Screen-for-Competitor Kit ERα (Takara, Kyoto, Japan) as described by Bolger et al. (15), we measured the difference of polarization between fluorescence-labeled E2 (ES1) bound to ER and ES1 only. Styrene dimers and trimers did not show statistically significant inhibitory action against the binding of ESI to ER in this assay. We also used Method B, the method used by Ohyama et al. (10), to detect the binding affinities of test samples to the human recombinant ERα coated on the microplate by competition with fluorescence-labeled E2; this was performed using the Estrogen Receptor (α) Competitor Screening Kit (Wako PC, Osaka, Japan). Styrene dimers and trimers showed weak inhibitory effect on the binding of fluorescein E2 to hERα at 5 μmol/L, and their binding abilities were < 30% in this assay. To evaluate the ER binding assays themselves, we included vitamin D3, naphthalene, 5α-dihydrotestosterone, and testosterone in each of the three ER binding assays; none of these compounds bound ER in vitro (13,16,17). A cross-reaction between estrogen and androgens cannot occur in vivo unless the androgens are metabolized. In Method RI and Method A, these nonestrogenic compounds did not show any ability to bind to the ER. However, in Method B, these compounds showed binding affinity for the recombinant hERα coated on the microplate at such high concentrations that they did not dissolve, although the binding affinity of E2 was similar in each assay. These results suggest that Method B tends to detect false-positive effects and that it is less accurate at high concentrations because of a decline of specificity to estrogen at high concentrations at which compounds do not dissolve. The manufacturer’s instructions for the Estrogen Receptor (α) Competitor Screening Kit used for Method B say to “make sure there is no precipitation in the solution.” Styrene dimers and trimers are so hydrophobic that their solubility is very low in the buffer solutions used in each assay. On the basis of these results, styrene dimers and trimers have no affinity for ER in Methods RI and A. Nevertheless, styrene dimers and trimers exhibited some affinity for the recombinant hERα in the Method B study, similar to that described by Ohyama et al. (10), but at high concentrations such that the compounds were not completely dissolved. This result is not because of the difference of sensitivity between rat ER and human ER, as shown in Method A with the use of hERα, but is caused by a decrease in specificity to estrogen because of the precipitation of test compounds. Ohyama et al. (10) reported that high concentrations of styrene dimers and trimers showed proliferative activity in the E-SCREEN assay. Cell proliferation can be induced by other growth factors, although proliferation of MCF-7 cell is basically E2 dependent (18–20), and the response to E2 in MCF-7 cells varies because of the various mutation of ER (21). Therefore, a false-positive response might only be shown in tests using proliferation as a target. The luciferase reporter gene assay, which indicates direct gene expression reactivity through the receptor, has been considered to be a more suitable assay for evaluating estrogenicity at the cellular level because of specificity to E2 response (22,23). Styrene dimers and trimers did not show any estrogenic effect in the E-SCREEN assay and the reporter gene assay in our previous study (6). In addition,

Polystyrene is frequently used in resins, and the styrene dimers and trimers eluted from polystyrene have been reported to have estrogenic activity (1). We have performed a number of in vitro and in vivo tests [i.e., estrogen receptor (ER) and androgen receptor binding assays, thyroid hormone receptor binding assays, human breast cancer cell line MCF-7 proliferation assays (E-SCREEN), uterotrophic assays in immature and ovariectomized rats, Hershberger assays, and prolactin release assays and steroidogenesis] and found no effects of styrene dimers or trimers on sex hormones in any of these assays (2-7). These results are supported by Fail et al. (8), who reported that mixtures of styrene oligomers did not show any estrogenic activity in the immature rat uterotrophic assay and the reporter gene assay. In addition, the Japan Environment Agency referred to their studies (9) and removed the styrene dimers and trimers from their list of endocrine disruptors (9). However, Ohyama et al. (10) reported that high concentrations of certain styrene dimers and trimers showed estrogenic effects in an ER binding assay and in the E-SCREEN assay. Recently, several assay systems have been used to assess endocrine-disrupting effects, but a few of these assay systems can cause false-positive reactions when test compounds are at high concentrations (11).
To assess the accuracy of the ER binding assay system and the results of Ohyama et al. (10), and to ascertain the safety of styrene dimers and trimers, we used a solubility test and three ER binding assays (12) ( Table 1). The ER binding assay, which detects the direct reactivity of ligand to a receptor, is the most standardized and simple test system for the detection of specific mechanisms of estrogenic activity.
Using the radiosotope method (Method RI) as described previously (13,14), we observed that styrene dimers and trimers did not show statistically significant inhibitory action against the binding of [ 3 H]-17β-estradiol (E 2 ) to ER.
We used Method A to detect the binding affinities of test samples to human ERα (hERα). Using a fluorescence polarization Screen-for-Competitor Kit ERα (Takara, Kyoto, Japan) as described by Bolger et al.
(15), we measured the difference of polarization between fluorescence-labeled E 2 (ES1) bound to ER and ES1 only. Styrene dimers and trimers did not show statistically significant inhibitory action against the binding of ESI to ER in this assay.
We also used Method B, the method used by Ohyama et al. (10), to detect the binding affinities of test samples to the human recombinant ERα coated on the microplate by competition with fluorescence-labeled E 2 ; this was performed using the Estrogen Receptor (α) Competitor Screening Kit (Wako PC, Osaka, Japan). Styrene dimers and trimers showed weak inhibitory effect on the binding of fluorescein E 2 to hERα at 5 µmol/L, and their binding abilities were < 30% in this assay.
To evaluate the ER binding assays themselves, we included vitamin D 3 , naphthalene, 5α-dihydrotestosterone, and testosterone in each of the three ER binding assays; none of these compounds bound ER in vitro (13,16,17). A cross-reaction between estrogen and androgens cannot occur in vivo unless the androgens are metabolized. In Method RI and Method A, these nonestrogenic compounds did not show any ability to bind to the ER. However, in Method B, these compounds showed binding affinity for the recombinant hERα coated on the microplate at such high concentrations that they did not dissolve, although the binding affinity of E 2 was similar in each assay. These results suggest that Method B tends to detect false-positive effects and that it is less accurate at high concentrations because of a decline of specificity to estrogen at high concentrations at which compounds do not dissolve. The manufacturer's instructions for the Estrogen Receptor (α) Competitor Screening Kit used for Method B say to "make sure there is no precipitation in the solution." Styrene dimers and trimers are so hydrophobic that their solubility is very low in the buffer solutions used in each assay. On the basis of these results, styrene dimers and trimers have no affinity for ER in Methods RI and A. Nevertheless, styrene dimers and trimers exhibited some affinity for the recombinant hERα in the Method B study, similar to that described by Ohyama et al. (10), but at high concentrations such that the compounds were not completely dissolved. This result is not because of the difference of sensitivity between rat ER and human ER, as shown in Method A with the use of hERα, but is caused by a decrease in specificity to estrogen because of the precipitation of test compounds. Ohyama et al. (10) reported that high concentrations of styrene dimers and trimers showed proliferative activity in the E-SCREEN assay. Cell proliferation can be induced by other growth factors, although proliferation of MCF-7 cell is basically E 2 dependent (18-20), and the response to E 2 in MCF-7 cells varies because of the various mutation of ER (21). Therefore, a false-positive response might only be shown in tests using proliferation as a target. The luciferase reporter gene assay, which indicates direct gene expression reactivity through the receptor, has been considered to be a more suitable assay for evaluating estrogenicity at the cellular level because of specificity to E 2 response (22,23). Styrene dimers and trimers did not show any estrogenic effect in the E-SCREEN assay and the reporter gene assay in our previous study (6). In addition, at high concentrations at which test compounds were precipitated, cells indicated an abnormal response in the luciferase activity of control plasmids and in morphology (data not shown). To construct a stable assay system, we used HeLa cells transfected with an hERα expression plasmid derived from normal human liver ERα. In this assay system, styrene dimers and trimers did not show any increase in E 2 -dependent luciferase transcription activity. These results agreed with the result of the ER binding assay. We presume that styrene dimers and trimers had no binding affinity to ER and they did not affect E 2 -dependent transcription. As a result, in our comparison of three ER binding assays using estrogenic and nonestrogenic compounds, it appeared that Method RI and Method A were useful for evaluating binding affinity for the ER, but Method B, similar to the method of Ohyama et al. (10), tended to indicate false-positives in high concentrations in which test chemicals were insoluble; this reduced the specificity of ER to E 2 . Based on our present results and previous reports (2-7), we found no endocrine-disrupting activities in styrene dimers and trimers eluted from polystyrene-containing instant noodle containers.

Estrogenicity of Styrene Oligomers: Response to Ohno et al.
The main point of the letter by Ohno et al.
is that styrene oligomers have no estrogenic activity, that our statement about "some styrene oligomers having binding affinity for hERα" was inaccurate, and that the MCF-7 cell proliferation assay is useless in detecting estrogenicity.
It seems that Ohno et al. have misunderstood our article. We are confident that the results of the MCF-7 cell proliferative assay and the binding assay of styrene oligomers to hERα in our paper are accurate.
If our binding assay indicated false positives in the range of concentrations in which test chemicals were insoluble, the inhibition by 1e,3e,5a-triphenylcyclohexane (ST-6) and 1e,3e,5e-triphenylcyclohexane (ST-7) would also increase, but no binding activity was observed for ST-6 and ST-7 at any concentration tested. This method (Ohno et al.'s Method B) showed an Environmental Health Perspectives • VOLUME 110 | NUMBER 7 | July 2002

A 385
Correspondence increase in the inhibition of binding by some soluble styrene oligomers but no effect by the same chemicals at insoluble concentrations. Ohno et al.'s Method B showed no ED 30 values of the styrene oligomers at > 10 µmol/L. Therefore, Ohno et al.'s Method B also indicated no effect by the styrene oligomers at insoluble concentrations except testosterone, 5αdihydrotestosterone, vitamin D 3 , and naphthalene. It seems that testosterone, 5αdihydrotestosterone, vitamin D 3 , and naphthalene used by Ohno et al. had special characteristics for the competitive binding assay kit (Wako, Osaka, Japan).
The MCF-7 cell proliferation assay is a recognized method for estrogenic screening. Ohno et al. overemphasize other growth factors. All of the styrene oligomers we tested did not have proliferative activity (1). ST-6 and ST-7 had no proliferative activity at all, but the proliferative potency of ST-3 and ST-4 was comparable with that of bisphenol A. Moreover we confirmed that OH-tamoxifen, an antagonist, inhibited cell proliferation by ST-1, ST-3, ST-4, ST-5, SD-3, and SD-4 (2). Recently, we reported that ST-1 and ST-4 were estrogenic in the reporter gene assay using MVLN cells established by stable transfection with the luciferase gene (3). Moreover we found that some other styrene oligomers were also estrogenic in this reporter gene assay (2).
We are confident that our paper (1) does not include any inaccurate results.
EHP regrets the errors.