Hydroxylated Metabolites of the Polybrominated Diphenyl Ether Mixture DE-71 Are Weak Estrogen Receptor-α Ligands

Background Polybrominated diphenyl ethers (PBDEs) are widely found in the environment and are suspected endocrine disruptors. We previously identified six hydroxylated metabolites of PBDE (OH-PBDEs) in treated mice. Objective We tested the hypothesis that OH-PBDEs would interact with and alter activity of estrogen receptor-α (ER-α). Methods We tested estrogenicity using two assays: 3H-estradiol (3H-E2) displacement from recombinant ER-α and induction of reporter gene (ERE-luciferase) in cultured cells. We incubated the PBDE mixture DE-71 with rat liver microsomes and tested the resultant metabolite mixture for estrogenic activity. We also determined relative estrogenic potential of individual hydroxylated PBDE congeners. Results Reporter gene activity was increased by DE-71 that had been subjected to microsomal metabolism. DE-71 did not displace E2 from ER-α, but all six of the OH-PBDE metabolites did. para-Hydroxylated metabolites displayed a 10- to 30-fold higher affinity for ER-α compared with ortho-hydroxylated PBDEs, and one produced a maximal effect 30% higher than that produced by E2. Coadministration of E2 and DE-71, or certain of its metabolites, yielded reporter activity greater than either chemical alone. Two ortho-OH-PBDEs were antiestrogenic in the reporter assay. Conclusions The observations—that the DE-71 mixture did not displace 3H-E2 from ER-α while the hydroxylated metabolites did—suggest that the weak estrogenic effects of DE-71 are due to metabolic activation of individual congeners. However, the behavior of DE-71 and its metabolites, when co-administered with E2, suggest a secondary, undetermined mechanism from classical ER-α activation.


Research
DE-71 is a commercial mixture of mostly tetra-and penta-brominated diphenyl ethers (PBDEs), which has been used extensively as a flame retardant (Agency for Toxic Substances and Disease Registry 2004). DE-71 and other similar commercial mixtures (known collectively as pentaBDE) were used almost exclusively as flame retardants in flexible polyurethane foam, a major component of bed mattresses and upholstered products. Production of DE-71 ceased in 2004(U.S. Environmental Protection Agency 2006, the same year that use of pentaBDE and other PBDE commercial mixtures was banned by the European Union (2003).
PBDEs are very stable compounds, and they are not chemically bonded to the material they are intended to protect from burning. As a result, they are widely found in environmental media (Hites 2004;Law et al. 2006) and can be found in human blood and milk (Furst 2006;Gomara et al. 2007;Lind et al. 2003;Main et al. 2007;Mazdai et al. 2003;Schecter et al. 2003). Some of the PBDE congeners most commonly found in human samples are BDE-47, BDE-99, and BDE-153 (Hites 2004;Gomara et al. 2007;Main et al. 2007;Mazdai et al. 2003). Two of these same congeners, the tetrabrominated BDE-47 and the pentabrominated BDE-99, are the main components of DE-71 (36% and 44% by weight, respectively); the hexabrominated BDE-153 is a minor component of DE-71 (4% by weight) (Qiu et al. 2007).
The prevalence of PBDEs in human tissue is of concern because these compounds are known to alter behavior, thyroid-hormone signaling, and sexual development in animals. Eriksson and colleagues found permanent aberrations in spontaneous behavior in rodents after developmental exposure to BDE-47 (Eriksson et al. 2001), BDE-99 (Eriksson et al. 2006;Viberg et al. 2004), or BDE-153 (Viberg et al. 2003). Serum thyroxine (T 4 ) was significantly decreased in several different experimental models and by different pentaBDE mixtures: in rats exposed to DE-71 either prenatally or postnatally (Ellis-Hutchings et al. 2006;Stoker et al. 2004;Zhou et al. 2002); in American kestrels (Falco sparverius) exposed in ovo to a mixture of BDE-47, BDE-99, BDE-100 and BDE-153 (Fernie et al. 2005); and in adult female rats exposed to the commercial pentaBDE mixture Bromkal 70-5 DE (Darnerud et al. 2007). The effect of PBDEs on T 4 levels may require metabolic activation because hydroxylated PBDEs, but not the nonhydroxylated congeners, are able to bind human transthyretin in vitro (Meerts et al. 2000). Other effects of pentaBDE mixtures or their congeners in experimental animals suggest estrogenic or antiandrogenic activity. In rats, developmental exposure to BDE-99 affected the regulation of estrogen target genes (Ceccatelli et al. 2006), impaired spermatogenesis (Kuriyama et al. 2005), and decreased circulating sex steroids and reduced anogenital distance in males (Lilienthal et al. 2006). Male rats exposed to DE-71 on postnatal days 23-53 had reduced seminal vesicle and ventral prostate weights and delayed puberty (Stoker et al. 2004).
PBDEs are suspected to behave as estrogens because of the similarity of their chemical structures and properties to other xenoestrogens, mainly the polychlorinated biphenyls (PCBs) (Hooper and McDonald 2000;Meerts et al. 2001;Pijnenburg et al. 1995). We have shown that DE-71 has weak estrogenic activity in vivo and in vitro (Mercado-Feliciano and Bigsby 2008). Because hydroxylated metabolites of a structurally similar class of halogenated aromatic pollutants, the PCBs, exert estrogenic effects (Carpenter 2006;Vakharia and Gierthy 2000), it may be reasonable to expect that hydroxylated forms of PBDEs would also be estrogenic. Others have shown that some PBDE congeners and certain synthetically hydroxylated congeners could exert estrogenic effects in cultured cells (Hamers et al. 2006;Meerts et al. 2001). In a recent in vivo study, BDE-47 had uterotrophic effects in immature rats (Dang et al. 2007), suggesting in vivo activation of this otherwise nonestrogenic PBDE (Meerts et al. 2001).
Estrogen response element-luciferase (EREluc) assays. We used two ERE-luciferase reporter gene systems: one transiently transfected and the other an established stably transfected cell line. For the transient system, ER-negative MDA-MB-231 breast cancer cells were plated in BM. Two days later cells were transfected using Tfx-20 (Promega, Madison, WI) with expression vectors for ER-α (HEG0; from P. Chambon, Université Louis Pasteur, Illkrich, France), the estrogen-responsive firefly luciferase reporter construct ERE2-pS2-luc (Long et al. 2001), and the control Renilla luciferase reporter construct pRL-TK (Promega). Cells were treated with test chemicals 1 hr after transfection, and assayed for luciferase activity after 18 hr. Results are expressed as the ratio of firefly luciferase to Renilla luciferase. For the stable reporter system, ER-positive BG1Luc4E2 ovarian cancer cells (Rogers and Denison 2000) were incubated in BG1-BM for 5 days before treatment; cells were then assayed for luciferase activity 18 hr after addition of the test compound.
In vitro generation of microsomal metabolites. We incubated DE-71 or E 2 with liver microsomes to obtain microsomal metabolites, following a procedure adapted from Bulger et al. (1978). Glucose-6-phosphate, glucose-6phosphate dehydrogenase (G6PD) and β-nicotinamide adenine dinucleotide phosphate (oxidized form, NADP+) were purchased from Sigma-Aldrich. DE-71 (1 mM) or E 2 (1 µM) were incubated for 24 hr with female rat liver microsomes (BD Biosciences Gentest, Woburn, MA) in a buffer that included an NADPH-generating system (50 mm Tris buffer, pH 7.5, 5 mM MgCl 2 , 12 mM glucose-6-phosphate, 0.4 mM NADP + , 2 units G6PD) in loose-capped tubes at 37°C with shaking. The incubation mixture was then centrifuged at 105,000 relative centrifugal force (RCF) at 4°C for 1 hr to remove microsomes. The hydroxylated organic fraction was extracted from the supernatant by solidstate extraction with ethanol elution using Sep-Pak Plus C18 cartridges (Waters Corp.; Milford, MA), then evaporated to dryness in vacuo and reconstituted in a volume of DMSO that would yield 10 mM PBDE or 10 µM estradiol, assuming 100% recovery. This extraction procedure was adapted from Yoshihara et al. (2004).
Recombinant ER-α binding assay. Vehicle or test chemicals were incubated with 1 nM tritiated E 2 ( 3 H-E 2 ; Amersham Biosciences, Buckinghamshire, UK) and 0.6 nM recombinant ER-α (Panvera/Invitrogen, Madison, WI) in TE buffer (10 mM Tris, 1 mM EDTA, pH 7.5) at 4°C overnight. Hydroxylapatite (60% in TE buffer) was added, mixed well, and incubated for 15 min at room temperature. The resulting slurry was washed three times by centrifugation at 3,000 RCF at 4°C with TE buffer changed. Bound ligand was extracted by incubation of the slurry with absolute ethanol at 30°C for 10 min. Tritium ( 3 H) decay (counts per minute) was measured by liquid scintillation in a Beckman LS 5000 TD counter (Beckman-Coulter Inc., Fullerton, CA).
Statistics. All statistics were performed using GraphPad Prism, version 3.0a, for Macintosh (GraphPad Software, San Diego CA). For each analysis, we determined whether groups had unequal variances by Bartlett's test. Group averages with equal variances were compared to each other either by  Incubations with an incomplete NADPH generating system (lacking NADP+) were run in parallel and served as the negative control. The incubation products were tested in ERE-luciferase assays. One representative assay is shown; results are presented as mean ± SD (n = 4).
The results are representative of three similar assays. **p < 0.01, and # p < 0.001 compared with vehicle control. ## p < 0.001 compared with the same treatment without the complete NADPH generating system. one-way analysis of variance (ANOVA) with Tukey post-test or by unpaired t-test as appropriate. Group averages with unequal variances were compared to each other by t-test with Welch's correction. Groups treated with DE-71 or E 2 alone were compared with vehicle controls, and groups cotreated with DE-71 and E 2 were compared with controls treated with E 2 alone. All values are expressed as mean ± SD or SE as indicated. We considered groups statistically different if p < 0.05 by ANOVA with Tukey posttest or t-test (two-tailed). Dose-response studies were also subjected to regression analysis using a sigmoidal curve fitting model: where EC 50 is the median effective concentration. The modeled curve is shown in figures if R 2 > 0.8.

In vitro microsomal metabolism increases estrogenic character of DE-71.
To determine whether microsomal metabolism could increase estrogenic activity, we incubated DE-71 with female rat microsomes, with or without a complete NADPH generating system. The incubation product was then tested in ERE-luciferase assays, using either the transient transfection system with MDA-MB-231 cells or the stably transfected ER-α-positive BG1Luc4E2 cells. At 10 -5 M, DE-71 induced little ERE-luciferase activity when it was incubated in the buffer lacking an NADPH generating capacity (without NADP+; negative control), but the product of the complete incubation buffer produced a 4-fold increase in luciferase (Figure 1). Microsomal incubation decreased the activity of E 2 by 78%, an expected result because hydroxylation decreases the activity of E 2 (Martucci and Fishman 1993). Furthermore, the level of activity of the extract from the E 2 incubated without NADP+ indicates that recovery of compound by the extraction methods used and in three separate microsomal incubation experiments was 66-98%.

Phenolic metabolites found in blood of mice after treatment with DE-71.
In a previous study (Qiu et al. 2007), we examined blood serum from BALB/c mice that had been treated with DE-71 for 34 days and quantified identifiable phenolic metabolites that included two brominated phenols and six hydroxylated brominated diphenyl ethers (Table 1). These findings raise the possibility that the estrogenic effects seen earlier in mice and in culture (Mercado-Feliciano and Bigsby 2008) were due to the metabolites and not necessarily the original DE-71 congeners.
DE-71 and its phenolic metabolites activate the ERE. E 2 induces ERE-luciferase at concentrations above 10 -13 M in the stably transfected BG1Luc3E2 reporter cell culture system (Figure 2), with an EC 50 in the picomolar range (Table 2). Neat DE-71 and the metabolites found in mice were tested using BG1Luc3E2 cells to determine if they were able to activate ERE-mediated gene transcription (Figure 2), and their potencies and effectiveness were compared to that of E 2 ( Table 2). The potencies of each test compound were estimated by determining the compound's own EC 50 from its own maximal effect, and also by calculating its relative estrogenic potency determined from the concentration required to produce an effect equivalent to E 2 's EC 50 .
DE-71 was able to significantly induce ERE-luciferase at test concentrations ≥ 5 × 10 -5 M, reaching the same effectiveness as E 2 at 10 -4 M ( Figure 2B). The calculated EC 50 for DE-71 was 3.7 × 10 -5 M. Because the maximal effect of DE-71 was very close to the maximal effect of E 2 , its EC 50 (3.7 × 10 -5 M) and its estrogen equivalency potency (EEP; DE-71 metabolites are weak ER-α ligands Environmental Health Perspectives • VOLUME 116 | NUMBER 10 | October 2008  3.9 × 10 -5 M) were similar. One metabolite, 4´-OH-BDE-17 was clearly more potent than DE-71, with an EC 50 in the micromolar range (Table 2). 4´-OH-BDE-17 had a relative estrogenic potency approximately 10-fold that of DE-71, and it was more effective than DE-71 or E 2 , reaching an estimated maximal effect 30% higher than E 2 ( Figure 2C; Table 2). 4´-OH-BDE-49 had an EC 50 similar to that of DE-71 but a much lower efficacy; its maximal effect did not even reach the EC 50 for E 2 ( Figure 2E; Table 2). Another para-hydroxylated metabolite, 4-HO-BDE-42, appears to have been more potent than DE-71, but because of limited availability of this compound, the analysis was not carried out with a sufficient span of concentrations to allow an accurate estimate of the EC 50 ( Figure 2D).

DE-71 phenolic metabolites modify ERE activation by E 2 .
Because DE-71 and several of its hydroxylated metabolites were able to either activate and/or displace 3 H-E 2 from ER-α, We cotreated BG1Luc3E2 cells with 10 -11 M E 2 and one of the chemicals of interest to determine if the PBDEs were able to modify E 2induced ERE-luciferase activity. The two bromophenol metabolites found in mice, 2,4-DBP and 2,4,5-TBP, were not tested because they had no significant effect in either the ERE-induction or the 3 H-E 2 displacement assays. DE-71 induced ERE-luciferase beyond the maximal effect of E 2 alone ( Figure 4B). The same was true for 4´-OH-BDE-17 and 4´-OH-BDE-49 ( Figure 4C,E). Another PBDE tested, 4-OH-BDE-42, appeared to induce EREluciferase above the E 2 maximum, but the effect was not statistically significant ( Figure 4D). The only meta-OH-PBDE tested, 3-OH-BDE-47, was not estrogenic by itself but was able to potentiate the ERE-luciferase induction of E 2 (Figures 2G and 4G). The two orthohydroxylated BDE metabolites exhibited a biphasic dose-response curve in the E 2 cotreatment assay. At high concentrations, 2´-OH-BDE-28 and 6-OH-BDE-47 were able to antagonize the effect of E 2 . 6-OH-BDE-47 was the more potent antagonist, showing an effect at 5 × 10 -6 M, whereas antagonism by 2´-OH-BDE-28 was observed only at 5 × 10 -5 M ( Figure 4F,H). Both of these metabolites also seem to potentiate the effect of 10 -11 M E 2 at lower concentrations in a manner similar to other PBDEs tested, although this effect was statistically significant only for 2´-OH-BDE-28. However, because only DE-71 and 2´-OH-BDE-28 were tested at a concentration of ≥ 5 × 10 -5 M, it is possible that other hydroxylated PBDEs have the same biphasic behavior in the ERE-luciferase assay. The protein content of the culture wells at the time of harvest (not shown) indicated that there was no toxicity produced by 6-OH-BDE-47 or 2´-OH-BDE-28 at these concentrations (5-50 µM) when tested alone or in combination with E 2 . Interestingly, both OH-BDEs found to be estrogen antagonists were able to displace 3 H-E 2 from ER-α ( Figure 3F,H), but when administered alone they did not induce significant ERE-luciferase ( Figure 2F,H).

Discussion
PBDEs are suspected to behave as estrogens because of the similarity of their chemical structure to other xenobiotics, mainly the PCBs (Crews et al. 1995;Ulbrich and Stahlmann 2004;Winneke et al. 2002). Furthermore, hydroxylated metabolites of PCBs have been shown to exert estrogenic effects (Blair et al. 2000;Kuiper et al. 1997). Therefore, it may be reasonable to expect that hydroxylated forms of PBDEs are also estrogenic. Our previous findings indicate that the PBDE mixture DE-71 is estrogenic in vitro and in vivo, although much less potent than E 2 (Mercado-Feliciano and Bigsby 2008). Here, we show estrogenic and antiestrogenic effects of the phenolic metabolites of DE-71 by an interaction with ER-α. Meerts et al. (2001) tested 17 PBDE congeners for estrogenic activity in an EREluciferase assay (ER-CALUX; Legler et al. 1999). Two of the congeners in DE-71, BDE-28 and BDE-100, were mildly estrogenic; BDE-100 was the most potent of the PBDEs although not the most effective. Using the same ER-CALUX bioassay, Hamers et al. (2006) showed weak estrogenic activity for the DE-71 congeners BDE-28, BDE-47, and BDE-100, but not for the pentaBDE mixture Bromkal 70-5DE. Results from Meerts et al. (2001) and Hamers et al. (2006) agree in the relative potency of these chemicals, and both groups agree that the EC 50 for BDE-100 is in the micromolar range. In the present study we showed that DE-71 increased expression of ERE-luciferase reporter gene in BG1Luc3E2 cells, with potency similar to that of BDE-28 and BDE-100 in the ER-CALUX assay (Hamers et al. 2006;Meerts et al. 2001). The difference between our ERE-luciferase results and those of Hamers et al. (2006) can be accounted for by the fact that we tested higher concentrations. It may also be that the BG-1 cells we used and the T47D cells of the ER-CALUX assay differ in their ability to metabolically activate the various PBDE congeners.
In the present study, we observed metabolic activation of DE-71 to an estrogenic product in vitro, and previous experiments demonstrated that  congeners are hydroxylated in the mouse (Qui et al. 2007), a chemical modification that could increase their estrogenic activity. In the in vitro experiments, DE-71 was preincubated with rat liver microsomes, imitating the classical experiments by which Kupfer and Bulger demonstrated the metabolic activation of the DE-71 metabolites are weak ER-α ligands Environmental Health Perspectives • VOLUME 116 | NUMBER 10 | October 2008 Figure 4. ERE-luciferase induction in BG1Luc4E2 cells after coteatment with E 2 . See "Material and Methods" for details. Squares indicate data for cotreatment with 10 -11 M E 2 and the specified chemical; data are normalized to the maximal effect of E 2 (set at 1.0). Each curve is the mean of 3-5 independent dose-response studies, except for 2-OH-BDE28, 6-OH-BDE47, and 4-OH-BDE49 for which only 2 assays were performed. Error bars indicate SE. Modeled data for E 2 -only dose-response (blue dashed line) are shown in all charts for comparison. *p > 0.5; **p > 0.01; and # p > 0.001 compared with the maximal effect of E 2 alone. proestrogen methoxychlor (Bulger et al. 1978;Kupfer and Bulger 1979). Preincubation of DE-71 with microsomes under enzyme-activating conditions increased its estrogenic activity. Mammalian liver microsomes are rich in cytochrome P450 (CYP450), a group of isoenzymes responsible for metabolism of many endogenous and exogenous chemicals including estrogens (reviewed by Bigsby et al. 2005). The biological activities of environmental chemicals have been found to be either increased or decreased by specific CYP450s (Goldstein and Faletto 1993), and some of these chemicals are known to be converted into estrogens by CYP450 metabolism (Bulger et al. 1978;Kohno et al. 2005;Kupfer and Bulger 1979;Morohoshi et al. 2005). Based on findings by Qui et al. (2007) and others (Malmberg et al. 2005;Marsh et al. 2006), the DE-71 congener BDE-47 seems to be the source of activated OH-PBDEs in laboratory rodents; BDE-47 itself has been found to have little estrogenic activity (Hamers et al. 2006;Meerts et al. 2001). Others have also found OH-PBDEs in wild marine animals that could be BDE-47 metabolites (Kelly 2006;Verreault et al. 2005;Verreault et al. 2007). However, the source of OH-BDEs in the marine environment can be both natural and anthropogenic, because some marine organisms produce natural brominated compounds (Vetter 2006). We observed both estrogenic and antiestrogenic effects in our study. We found that metabolites of DE-71 hydroxylated at the ortho position could act as antiestrogens. Although cotreatment with DE-71 produced a larger effect than the maximal response to E 2 in the present study using a reporter gene assay, in another bioassay, based on MCF-7 cell proliferation, we observed both an agonist effect when DE-71 was administered alone and an antiestrogenic effect when coadministered with E 2 (Mercado-Feliciano and Bigsby 2008). Because both MCF-7 cells and ovarian cancer cells express CYP450 enzymes (Deloia et al. 2008;Leung et al. 2007;Peters et al. 2004), it may be that the effects in cell proliferation and gene expression were due to metabolites generated in culture. The difference in responses between cell proliferation and reporter gene assays could result from generation of different metabolites in each assay due to differences in treatment duration (10 days vs. 18 hr, respectively) and/or the predominant CYP450 isoenzyme activities in each kind of tissue (mammary vs. ovarian cancers). In addition, the estrogenic activity seen by other investigators for individual congeners in reporter gene bioassays (Hamers et al, 2006;Meerts et al. 2001) could be due to metabolic activation of those compounds. Thus, it is likely that the observed biological activity of DE-71 resulted from the sum of estrogenic and antiestrogenic activities of metabolites produced from individual BDE congeners.
Compounds may act as endocrine disruptors through a number of mechanisms, including indirectly by altering metabolism of endogenous hormones. Hamers et al. (2008) found that BDE-47 and several of the OH-PBDE metabolites that we found to have either estrogenic or antiestrogenic activity in culture also inhibit estrogen sulfotransferases (E2SULT) in vitro. Such an effect would translate into an increased activity of administered E 2 and could explain the ability of 3-OH-BDE-47 to increase ERE-luciferase expression above the level induced by E 2 alone when BG-1Luc4E2 cells are cotreated with both chemicals, even though 3-OH-BDE-47 does not induce ERE-luciferase by itself. However, the E 2 dose used in cotreatment with 3-OH-BDE-47 was already high enough to reach the maximal effect in this system; therefore, it is unlikely that making more E 2 available by inhibiting E2SULT would increase ERE-luciferase signaling. Furthermore, it is not known whether the BG-1Luc4E2 cells express E2SULT, and we observed no significant potentiation of the estrogenic effect by 4-OH-BDE-42, another metabolite that behaves as a potent inhibitor of E2SULT in vitro. A weaker E2SULT inhibitor, 6-OH-BDE-47, actually produced an antiestrogenic effect in our system. Thus, it is unlikely that altered E2SULT activity explains the additional estrogenic effect produced by 3-OH-BDE-47.
The levels of DE-71 needed to have estrogenic effects in our studies (micromolar range) are much higher than the highest concentrations found so far in human blood serum (0.1-5 nM; Mazdai et al. 2003). However, there is little current information on the levels of OH-PBDEs in human serum or the role that human enzymes (especially CYP450) may play in the formation of DE-71 metabolites. Rodent tissues do not have the same CYP450 activities as human tissues (Bogaards et al. 2000); therefore, the metabolites we found in mouse serum (Qui et al. 2007) or those found by Marsh et al. (2006) in rat feces may not be representative of metabolites formed in humans. Several of the PBDE metabolites we found in mice (4´-OH-BDE-17, 6-OH-BDE-47, 3-OH-BDE-47, and 4-OH-BDE-42) have been found in human serum samples from children working at a municipal waste disposal site (Athanasiadou et al. 2008) at much lower concentrations (< 0.1 nM) than are required to cause estrogenic effects in culture (≥ 1 µM) or associated with slight estrogenic effects in mice (> 40 nM; Table 1) following approximately 1 month of treatment (Mercado-Feliciano and Bigsby 2008). It is reasonable to expect that PBDE exposure for municipal waste workers is due in part to contact with DE-71 and other pentaBDE mixtures in discarded consumer products. However, Athanasioadou et al. (2008) found an additional PBDE metabolite in humans, 4-OH-BDE-90, that we did not observe in mice, and our own unpublished studies indicate that 5-OH-BDE-47 and 5´-OH-BDE-99 are major metabolites in human blood (Qui et al., in press); the estrogenic activity of these compounds has not been tested.
In summary, the observations that the DE-71 mixture does not displace 3 H-E 2 from ER-α-while the hydroxylated metabolites do-suggest that the weak estrogenic effects of DE-71 are due to metabolic activation of individual congeners. However, the behavior of DE-71 and some of its metabolites when coadministered with E 2 suggest a secondary, undetermined mechanism of action different from classical ER-α activation.