A Characteristic Back Support Structure in the Bisphenol A-Binding Pocket in the Human Nuclear Receptor ERRγ

The endocrine disruptor bisphenol A (BPA) affects various genes and hormones even at merely physiological levels. We recently demonstrated that BPA binds strongly to human nuclear receptor estrogen-related receptor (ERR) γ and that the phenol-A group of BPA is in a receptacle pocket with essential amino acid residues to provide structural support at the backside. This led BPA to bind to ERRγ in an induced-fit-type binding mode, for example, with a rotated motion of Val313 to support the Tyr326-binding site. A similar binding mechanism appears to occur at the binding site of the BPA phenol-B ring. X-ray crystal analysis of the ERRγ-ligand-binding domain/BPA complex suggested that the ERRγ receptor residues Leu342, Leu345, Asn346, and Ile349 function as intrinsic binding sites of the BPA phenol-B, whereas Leu265, Leu268, Ile310, Val313, Leu324, Tyr330, Lys430, Ala431, and His434 work as structural elements to assist these binding sites. In the present study, by evaluating the mutant receptors replaced by a series of amino acids, we demonstrated that a finely assembled structural network indeed exists around the two adjacent Leu342-Asn346 and Leu345-Ile349 ridges on the same α-helix 7 (H7), constructing a part of the binding pocket structure with back support residues for the BPA phenol-B ring. The results reveal that the double-layer binding sites, namely, the ordinary ligand binding sites and their back support residues, substantiate the strong binding of BPA to ERRγ. When ERRγ-Asn346 was replaced by the corresponding Gly and Tyr in ERRα and ERRβ, respectively, the binding affinity of BPA and even 4-hydroxytamxifen (4-OHT) is much reduced. Asn346 was found to be one of the residues that make ERRγ to be exclusive to BPA.


Introduction
Bisphenol A (BPA) binds strongly to estrogen-related receptor c (ERRc), one of 48 human nuclear receptors [1,2]. Although ERRc, as well as its subtypes ERRa and ERRb, is an orphan receptor whose physiological ligand is unknown even today [3], its transcriptional activity seems crucial, especially during differential gene expression and development [4,5]. Also, it should be noted that one of the major physiological roles of the ERRs appears to control cellular energy metabolism [6].
The facts that ERRc is rich in both the fetal brain and the placenta and that ERRc is a probable candidate for involvement in prostatic growth and development seem to have important inferences for newborns [4,7,8]. Thus, the risk of exposure to BPA and its endocrine-disrupting activities has been a cause of concern, especially for fetuses, infants, and children, who are particularly vulnerable to the adverse effects of chemicals [9,10]. So-called lowdose effects of BPA are increasingly being identified for many organ tissues and systems in mice and rats in vivo [11][12][13][14][15][16], and such effects would be magnified in infants and children. Research into the effects of binding between BPA and ERRc has inevitably led to a hypothesis that the low dose effects of BPA are mediated through ERRc and its specific target gene(s) [2,17,18].
ERRc is constitutively active, exhibiting considerably high basal activity due to an activation conformation of the ligand-binding domain (LBD) with no ligand [1,[19][20][21]. Binding of BPA to ERRc has almost no effect on the original active conformation [22], and thus almost all of the basal activity is preserved [1,2]. ERRc is a BPA-specific receptor, but the ERRc-BPA complex behaves in the same manner as solo ERRc unaccompanied with BPA. This makes the hypothesis mentioned above very difficult to test. Although BPA-binding has no detectable influence on the ERR receptor conformation, our recent studies have explored, for instance, the fact that the isopropyl-methyl group of Val313 rotates approximately 120u to sustain the tight interaction in the BPA/ERRc-LBD complex and also to avoid a collision with the benzene A-ring of BPA [23]. Val313 is not a direct binding site of BPA, but a residue which assists or structurally supports the direct binding sites of BPA [24]. These results required us to inspect in detail all the BPA-binding sites of ERRc.
On the central sp 3 tetrahedral carbon atom (C*) of BPA, HO-C 6 H 4 -C*(CH 3 ) 2 -C 6 H 4 -OH, there are two methyl groups and two phenol groups, A and B, respectively (Fig. 1). From the X-ray crystal structure of the BPA/ERRc-LBD complex, a set of amino acid residues have been suggested to be in close proximity to the phenol-A group, and their roles as binding sites have been demonstrated by site-directed mutagenesis experiments [24]. Those include Glu275 (H3) and Arg316 (H5) for the phenolhydroxyl group and Leu268, Leu271, Leu309, and Tyr326 for the phenol-A benzene ring, in addition to Ile279, Ile310, and Val313 for structural support of these residues.
When the amino acid residues of the ligand-binding pocket (LBP) of ERRc were compared with those of ERRa and ERRb, it was found that Asn346 is replaced by Gly and Tyr in ERRa and ERRb, respectively. These replacements are the complete discrepancy seen in LBPs among all the ERR family (ERRa, b, and c). The substitution of Asn to Gly in ERRa would expand the cavity of LBP and possibly allow much more flexible adaptation of ligands, but would perhaps be unfavourable for a tight binding of the BPA phenol-B group. On the other hand, the AsnRTyr substitution in ERRb apparently makes the cavity much narrower, which could prevent the binding of BPA to ERRb.
In the present study, in order to better understand the characteristics of the binding of the BPA phenol-B ring to ERRc, we carried out a series of site-directed point mutagenesis modifications. In our previous study, we demonstrated that there is a Leu268 in close proximity to the benzene B-ring of BPA, and this Leu268 is positioned like a clamp or double-hook to tightly connect the phenol-A and -B rings of BPA. In the present study, we first carefully checked the X-ray crystal structure of the ERRc-LBD/BPA complex, including this Leu268, and especially with respect to the regions surrounding the BPA phenol-B ring. It was suggested that the Leu342, Leu345, Asn346, and Ile349 of ERRc function as essential binding sites of the phenol-B group of BPA. It should be noted that direct binding site receptor residues are present on the same a-helix 7 (H7) of ERRc-LBD (Fig. 1), forming the distinct adjacent ridges of Leu 342 -Asn 346 and Leu 345 -Ile 349 . On the other hand, it was suggested that Leu265, Leu268, Ile310, Val313, Leu324, Tyr330, Lys430, Ala431 and His434 work as structural elements to assist these binding sites. In order to prove this suggestion, we evaluated a series of site-directed mutant receptors, and we carried out receptor-binding assays for expressed ERRc-LBD mutant receptor proteins by using tritium-labelled BPA. In addition, the transcription activity of fulllength mutant receptors was evaluated by a reporter gene assay using the HeLa cell line. The results indicated that Leu342, Leu345, Asn346, and Ile349 indeed play essential roles in capturing a single BPA molecule, and that the supportive roles of other surrounding residues are essential for this BPA binding. We here describe in detail the structure-activity studies on these amino acid residues that form imperative BPA-binding sites in ERRc.
A series of ERRc mutants were prepared according to the manufacturer's instructions by using PfuTurbo DNA Polymerase (Stratagene, La Jolla, CA, USA) with pGEX-ERRc-LBD or pcDNA3.1-ERRc-Full as a template and a set of overlapping sense and antisense primer pairs. The mutations were introduced by PCR mutagenesis in a two-step reaction essentially as reported previously [24,25]. Each mutant LBD or full-length ERRc was amplified and cloned into the expression vector pGEX-6p-1 or pcDNA3.1(+) at the EcoRI and XhoI sites. The accuracy of all PCR product sequences was confirmed by using a CEQ TM 8800 Genetic Analysis System (Beckman Coulter, Fullerton, CA, USA).

ERRc-LBD protein expression
Wild-type and mutant ERRc-LBD proteins were expressed as glutathione S-transferase (GST)-fused preparations in E. coli BL21 as described previously [1,24,25]. The receptor protein was purified first by affinity chromatography using a column (106100 mm) of Glutathione-Sepharose 4B (GE Healthcare). After loading the protein solution, the column was incubated for 1 h at 4uC, and then washed three times with PBS containing 0.5% (v/v) Triton X-100 and once with sonication buffer (50 mM Tris-HCl (pH 8.0), 50 mM NaCl, 1 mM EDTA, and 1 mM dithiothreitol). The fusion protein was eluted with 50 mM Tris-HCl (pH 8.0) containing 20 mM glutathione (reduced form), which was removed by gel filtration on a column of Sephadex G-10 (156100 mm; GE Healthcare) equilibrated with 50 mM Tris-HCl (pH 8.0). The purity was confirmed by SDS-PAGE using 12.5% polyacrylamide gel and stained by Coomassie brilliant blue. The protein concentrations were determined by the Bradford method [26].

Circular dichroism (CD) spectra measurements
The GST-free ERRc-LBD proteins were prepared from their GST-fused proteins for the purpose of CD spectra measurement only. GST was removed by using a specific enzyme PreScission Protease (GE Healthcare) on an affinity column of Glutathione-Sepharose 4B. After loading the solution of GST-ERRc-LBD protein on the column, the resin was left to allow for incubation at 4uC for 4 h, and then eluted with a cleavage buffer of 50 mM Tris-HCl (pH 7.0) containing 150 mM NaCl, 1 mM EDTA, and 1 mM dithiothreitol. The eluate was diluted with the same buffer to obtain the solution of approximately 0.2 mM ERRc-LBD protein for CD measurements. The concentration was determined by the Bradford protein method [26]. The Tris-HCl buffer used was judged relevant for CD measurements of proteins.
CD spectra were recorded at 5-95uC with 10uC intervals on a JASCO J-725 spectropolarimeter (JASCO Co., Tokyo, Japan) in a cell of 1 mm path length. Spectra were acquired over the 195-300 nm range at a scan rate of 5 nm/min. Four scans were accumulated to obtain a mean spectrum. The buffer solution itself depicted a rather intense absorption below 200 nm, and thus CD spectra of GST-free ERRc-LBD proteins were normalized by subtracting the buffer scan recorded under the same conditions. The results were eventually analysed by the standard analysis software (JACSO) and expressed as the mean molar ellipticity [h].

Radio-ligand receptor-binding assays
Saturation binding. A saturation binding assay was conducted [27] using [ 3 H]BPA. The reaction mixture was incubated at 4uC for 2 h with the receptor proteins-GST-fused wild-type ERRc-LBD or its mutants-in 100 ml of binding buffer [10 mM HEPES (pH 7.5), 50 mM NaCl, 2 mM MgCl 2 , 1 mM EDTA, 2 mM CHAPS, and 2 mg/ml c-globulin]. The assay was performed with or without the addition of unlabelled BPA (final concentration of 1.0610 -5 M) to quantify the specific and nonspecific binding. After incubation with 100 ml of 1% dextran-coated charcoal (DCC) (Sigma-Aldrich) [28] in PBS (pH 7.4) for 10 min at 4uC, the DCC-absorbed free radio-ligand was removed by the direct vacuum filtration method using a 96well filtration plate (MultiScreen HTS HV, 0.45 mm pore size; Millipore, Billerica, MA, USA) for the bound/free separation [27]. Radioactivity was determined on a liquid scintillation counter (LS6500; Beckman Coulter, Fullerton, CA, USA).
The data on the specific binding of [ 3 H]BPA were first assessed by means of Scatchard plot analysis [29]. Then, these data were applied to a one-site binding hyperbola nonlinear regression analysis by the software package Prism (GraphPad Software Inc., La Jolla, CA, USA) to measure changes in the receptor density B max and equilibrium dissociation constant K d . The saturation binding assay was performed at least three times.
Competitive binding. BPA was dissolved in a binding buffer containing 0.1% dimethyl sulfoxide (DMSO). Competitive binding assays were performed in the presence of GST-fused wild-type ERRc-LBD or its mutants at the most appropriate concentration of each receptor. The reaction mixtures were incubated with [ 3 H]BPA (final concentration 5 nM) at 4uC for 2 h, and the bound/free separation was carried out by the DCC method as described above. Radioactivity was determined on a liquid scintillation counter TopCount NXT (PerkinElmer Life Sciences Japan Co., Ltd., Tokyo, Japan). To estimate the binding affinity, the IC 50 values (the concentrations for the half-maximal inhibition) were calculated from the dose-response curves evaluated by the nonlinear analysis program ALLFIT [30]. Each assay was performed at least three times.

Cell culture and transient transfection assays
HeLa cells were maintained in Eagle's Modified Eagle Medium (EMEM) (Nissui, Tokyo, Japan) in the presence of 10% (v/v) fetal bovine serum at 37uC under 5% CO 2 . HeLa cells were first seeded for 24 h at 5610 5 cells/dish (6 cm in diameter), and then transfected with luciferase reporter plasmid pGL3/36ERRE (3 mg), the expression plasmid of pcDNA3.1-ERRc-Full (wild-type or mutant) (2 mg), and 10 ng/dish of pSEAP plasmid as an internal control in the medium (5 ml in total) with Plus reagent (10 ml; Invitrogen) and Lipofectamine LTX (15 ml; Invitrogen) according to the manufacturer's protocol. Approximately 24 h after this transfection, cells were harvested and plated onto 96-well plates at a concentration of 5610 4 cells per well.
Luciferase activity was measured after 24 h at 37uC under 5% CO 2 by using a Luciferase Assay System (Promega, Madison, WI, USA) according to the manufacturer's instructions. Light emission was measured on a Wallac 1420 ARVOsx microplate reader (PerkinElmer). SEAP activity was assayed by using the Great EscAPe TM SEAP assay reagent (Clontech) according to the Fluorescent SEAP assay protocol [31,32]. Secreted embryonic alkaline phosphatase (SEAP) is a reporter widely used to study promoter activity or gene expression. Cells treated with 1% BSA/ PBS were used as a vehicle control. Values were computed as the fold inductions after normalization to SEAP activities. Each assay was performed in triplicate at least three times.

Statistical analysis
Data are presented as the mean 6 SD for indicated number of separate experiments. Statistical difference was determined by two-sided Student's t test. P-values less than 0.001 were considered significant.

Computational structural analysis of ERRc/bisphenol A complex
All molecular modeling studies were carried out using the molecular modeling software Discovery Studio 2.5 (Accelrys; San Diego, CA, USA). The protein structure of the ERRc-LBD/BPA complex was downloaded from the RCSB Protein Data Bank (PDB) (http://www.pdb.org/pdb/home/home.do) (PDB accession code: 2E2R) and hydrogen atoms were added correctly. ERRc-LBD/BPA complex was visualized and analyzed on the software. To identify amino acid residues constructed the LBP of ERRc-LBD, the residues were carefully checked for their spatial interrelationships with BPA molecule, especially noting the interatomic distance and angle.

Results and Discussion
Essential amino acid residues as structural elements specific for BPA-binding In order to successfully analyse the structure-activity relationship in proteins by replacing a series of amino acid residues with other amino acids, it is necessary to guarantee the purity of the protein expressed. In the present study, site-directed mutations were introduced into a series of amino acid residues on a-helix No. 7 (H7) of ERRc. These included Leu342, Leu345, Asn346, and Ile349. Their back support residues such as Leu265, Leu324, Tyr330, Lys430, and His434 were also mutated. The mutations were carried out by the PCR mutagenesis method [24,25]. For receptor-binding assays, mutant receptors were expressed as GSTfused ligand-binding domain (LBD) proteins in E. coli and purified by affinity-chromatography followed by gel-filtration. The purity of GST-fused ERRc-LBD was examined by SDS-PAGE (12.5% polyacrylamide gel), and all of the GST-ERRc-LBD mutants including the wild-type were judged to be sufficiently pure for use in the receptor-binding assays (Fig. S1). The purity of GST-free ERRc-LBD was also guaranteed by SDS-PAGE (Fig. S1).
Another structural inspection that can be used to guarantee the quality of expressed proteins is analysis of the conformational uniformity and stability of ERRc-LBD mutant receptors. This is particularly important for adequate evaluation and discussion of the structure-activity relationships between the wild-type and the mutant receptor proteins. In this study, the conformational quality of the expressed proteins was verified by the measurement of CD spectra. Because of the considerably large Cotton effects of GST, a protein with a mixed structure of a-helices and b-strands (Protein Data Bank accession code 2GSR) [33], we prepared GST-free ERRc-LBD proteins for the purpose of CD spectra measurement only, as reported in a previous study [24]. Measuring the temperature-dependent CD spectra (5uC-95uC at 10uC intervals), it was demonstrated that all the proteins that were mutated at the LBP did indeed hold a properly folded conformation without any denaturation/misfolding at the temperature region 5uC-35uC (Fig. S2), which is the standard condition for temperature in the ordinary regular receptor binding assays. The thermal unfolding curves showed the approximately 55uC of the structural melting point (Tm) for all of GST-free ERRc-LBD proteins (Fig. S2).
ERRc position 346 requires Asn, but not Gly and Tyr, for specific BPA-binding ERRa possesses Gly at position 402, which corresponds to Asn346 in ERRc, and thus we replaced ERRc-Asn346 by Gly. Although the resulting Asn346Gly-ERRc mutant receptor exhibited sufficient specific binding (ca. 70% of the total binding) for [ 3 H]BPA (Fig. S3), the equilibrium dissociation constant K d value of [ 3 H]BPA was 52.8 nM, approximately 11% of the binding affinity for the wild-type ERRc (Table 1). This clearly indicates that the AsnRGly substitution is greatly disadvantageous for the binding of BPA by ERRc, probably due to the lack of hydrogen bonding between Asn346 and the phenol hydroxyl group of BPA. The results strongly suggest that it would be difficult to induce BPA to bind to ERRa.
For the mutant receptors exhibiting a sufficient level of specific binding, the one-site binding hyperbola nonlinear regression analysis of saturation binding curves estimated the receptor density (B max : Table 1) for [ 3 H]BPA (Fig. S3), in addition to K d ( Table 1). In the homologous competitive binding assay using tritium-labeled [ 3 H]BPA and non-labeled BPA for the Asn346Gly-ERRc mutant receptor, almost the same result as described above for [ 3 H]BPA was obtained for non-labeled BPA (Fig. 2C). BPA was clearly weaker in this mutant receptor as compared with the wild-type ERRc, the IC 50 value being 42.4 nM [ca. 12% of the binding affinity (4.96 nM) for the wild-type ERRc] ( Table 2).
ERRb has Tyr at position 321 corresponding to Asn346 in ERRc. When this Tyr was placed at position 346 of ERRc, the resulting Asn346Tyr-ERRc exhibited a sufficient specific binding (ca. 60% of the total binding) for [ 3 H]BPA (Fig. S3). The K d value of [ 3 H]BPA was 31.5 nM, approximately 18% of the binding affinity for the wild-type ERRc (Table 1). In the competitive binding assay for this ERRc mutant receptor using [ 3 H]BPA as a tracer, non-labeled BPA exhibited an approximately 9-fold reduction in binding affinity ( Table 2). All these results indicate that BPA binds strongly to ERRc, but not to Asn346Tyr-ERRc, indicating that the AsnRTyr substitution at position 346 is disadvantageous for the acceptance of BPA. The structural conversion of the carboxyl amide CONH 2 (Asn) into the phenol C 6 H 4 -OH group (Tyr) is clearly disadvantageous for the binding of BPA.

Constructional importance of Tyr326 as a BPA-binding site
The para-hydroxyl group of Tyr at position 326 is in close proximity to the BPA phenol-B benzene ring. In the BPA/ERRc-LBD binding complex analysed by X-ray crystallography, the distance between the para-hydroxyl-oxygen atom and benzenecarbon atoms was estimated to be 3.1-4.9 Å in the complex structure [22]. This spacing appeared to allow the intermolecular interaction to a sufficient degree. It should be noted, however, that the Tyr326RPhe mutant receptor lacking the para-hydroxyl group of Tyr-b-phenol was found to be as active as the wild-type [24], indicating that there is no interaction between Tyr326 and the BPA phenol-B benzene ring. This result was somehow surprising, since the two are in extremely close proximity.
The fact that the Phe326 mutant receptor is as active as the wild type with Tyr326 implies definite nonalignment or independence of the BPA benzene B-ring from the Tyr326-para-hydroxyl group. BPA and Tyr326 do not undergo an OH/p interaction, which is one of the strongest molecular interactions. Because the bond angle of C2O2H in Tyr-b-phenol is approximately 105u, the Tyr-b-phenol-para-hydroxyl hydrogen atom must be distant from the BPA benzene B-ring. This causes the Tyr326-b-phenol to act like a wall blocking the pocket from accepting the BPA phenol Bgroup. This is in stark contrast to the fact that the C2-C3 edge of the Tyr326-b-phenol ring is in close proximity (3.75 Å ) to the benzene A-ring of BPA to undergo an edge-to-face p-p interaction, namely, the so-called CH/p interaction [24].

The bisphenol A-binding site constructed by the Leu-Asn and Leu-Ile ridges
The binding site of the BPA phenol-B group in ERRc is made up of Leu342, Leu345, Asn346, and Ile349, being at a-helix H7 of ERRc-LBD (Fig. 1). Leu342 and Asn346 are on the same (i, i+4) ridge and the other Leu345 and Ile349 are on the adjacent (i+3, i+ 7) ridge in H7. Another important site is Tyr326 on the b-strand 1, and all these five receptor residues constrict a binding pocket for the BPA phenol-B group. In order to evaluate the contribution of H7 amino acid residues to the binding of BPA, Leu342, Leu345, Asn346, and Ile349 were mutated to Ala and some other structurally related amino acids.
Leu342. Leu342 was first replaced with Ala. The Leu342Ala-ERRc mutant receptor exhibited no specific binding for [ 3 H]BPA (Fig. S3) Fig. 2A) (Table 2). These results suggest that the Cb-branched side chains are not suitable to accept the BPA molecule at position 342 of ERRc.
Leu345. In our previous study, we demonstrated that, when BPA binds to ERRc Leu345 makes a back-and-forth rotation of its side chain isopropyl group to adopt a BPA molecule [23]. A strong hydrophobic interaction between the phenol-B benzene ring of BPA and the isopropyl group of Leu345 was eventually demonstrated. The Leu345RAla substitution, namely, the elimination of the isopropyl group from the Leu side chain bmethylene, resulted in a drop (approximately 3.5-fold) in the receptor affinity of [ 3 H]BPA (Table 1). When Leu345 was replaced with Val, the resulting mutant receptor exhibited a sharp drop in its affinity for [ 3 H]BPA; i.e., an almost 10-fold decrease in the affinity for [ 3 H]BPA in the saturation binding assay (Table 1) and an approximately 30-fold decrease in the affinity for [ 3 H]BPA in the [ 3 H]BPA/BPA homologous competitive receptor binding assay (Fig. 2B) (Table 2).
Although the isobutyl group of the Leu side chain could potentially make a back-and-forth rotation, the isopropyl group of the Val side chain cannot make such a rotation. Since Val's isopropyl side chain can rotate freely around the Ca-Cb axis, it would stay at the most stable location. It is highly likely that this location resulted in a sterical hindrance just due to its unfavourable rotatory x angle around the Ca-Cb axis, causing a drastic drop in the receptor affinity of BPA. A similar phenomenon was observed for the Leu345Ile mutant receptor, which showed an approximately 30-fold decrease in the binding of BPA in the homologous competitive binding receptor assay (Fig. 2B) (Table 2). Ile's side chain sec-butyl group also cannot rotate in a back-and-forth manner. All these results indicate that Leu345 is one of the intrinsic determining factors of BPA's highly specific binding ability for ERRc.
Asn346. The results of the mutation of Asn346 to Gly and Tyr are described above. In addition, in order to evaluate the structural importance of this Asn346 residue, it was also mutated to Ala. The resulting Asn346Ala-ERRc mutant receptor exhibited a fairly high specific binding (ca. 90% of the total binding) for [ 3 H]BPA (Fig. S3). However, this well-folded Asn346Ala-ERRc-LBD protein exhibited a distinctly reduced (2-4 times) binding affinity of BPA (Tables 1 and 2) (Fig. 2C), implying that Asn346's side chain carboxamide methyl group is engaged in the hydrogen bond with the hydroxyl group in the BPA's phenol-B. Gln is a homolog of Asn, and the difference in their side chains is one methylene chain in length; i.e., -CH 2 CONH 2 for Asn and -CH 2 CH 2 CONH 2 for Gln. Since the Asn346Gln-ERRc mutant receptor exhibited a 3-10 fold reduction in the affinity for BPA (Tables 1 and 2), this Asn346«BPA hydrogen bond was judged to be highly specific and effective.
Ile349. When Ile349 was replaced with Ala, the resulting Ile349Ala-ERRc exhibited a 3-4 fold reduction in affinity for [ 3 H]BPA and BPA (Tables 1 and 2). This indicates the importance of the sec-butyl side chain for Ile to bind a BPA molecule. When Ile349 was replaced with Val and Leu, the activity decrements were rather small: 8-65% for [ 3 H]BPA (Table 1) and 110-135% for BPA (Fig. 2D) (Table 2). However, the importance of Ile349 is beyond doubt, since each of the Ile349Ala-ERRc, Ile349Val-ERRc, and Ile349Leu-ERRc mutant receptors showed clearly lower B max values (Table 1). Collectively, the amino acid residues of Leu342, Leu345, Asn346, and Ile349, which are present at the a-helix H7 of ERRc, were judged to be authentic binding sites of BPA. Although their interactions with BPA molecules vary considerably, their significance is absolute and fundamental, as evidenced by the sitedirected mutagenesis experiments.
Amino acid residues that support the binding sites of BPA's phenol-B In our previous study, several amino acid residues were found to assist or structurally support the binding site amino acid residues, the side-chains of which participate in formation of a receptacle pocket for the BPA's phenol-A group [24]. Those included Ile279, Ile310, and Val313. In the present work, we also attempted to explore the surroundings of the amino acid residues of Leu342, Leu345, Asn346, and Ile349, which have been identified as the binding sites of BPA's phenol-B group. Ala431, which is located within 4.5 Å of the BPA's sp 3 methyl group, was found to be in close proximity to Leu345 (Fig. 3A), and was very likely to provide structural support for the binding site.
In the present study, Val313 was again identified to be important as a structurally supporting residue. When the region surrounding Ile349 was checked carefully, Val313 was found to make an important contribution to the structurally supportive environment (Fig. 3B). In fact, the Ala-substitution of Val313 resulted in a considerable reduction (approximately 5-fold) in the binding affinities of BPA (Table 3). It should be noted that, when the structure of ERRc-LBD with BPA was compared with the structure of ERRc-LBD without BPA, the isopropyl-methyl group of Val313 was rotated about 120u to prevent the collision with the BPA's phenol-A benzene ring [24]. Val313 was thus shown to be one of the key residues for construction of the BPA-binding site in ERRc-LBD, serving to hold the BPA-binding site amino acid residues in the proper positions. It is of note that there is no direct contact between Val313 and BPA.
Ile310 was found to also be one of the amino acids in the region surrounding Ile349, whereas Leu268 was one of amino acids in the region surrounding Leu342. As reported previously [24], Leu268 is a phenol-B binding site of BPA and is positioned like a clamp or double-hook to tightly connect both of BPA's phenol-benzene rings. Leu268 was found to be in close proximity to Leu342 (Fig. 3C). In this study, we further found that Leu265 and Tyr330 were back support residues of Leu268 (Fig. 3D). As a result, Leu268 serves both as a direct BPA-binding site and a back support residue, and Leu265 and Tyr330 serves only as a back support residue. Other newly speculated possible back support residues include Leu324 for Asp346 and Ile349, His434 for Leu342 and Leu345, and Lys430 for Leu345 (Fig. 3). Thus, Table 2. Receptor binding potency of BPA, 4-a-cumylphenol, and 4-OHT in the competitive binding assay using [ 3 H]BPA for human nuclear receptor ERRc and its mutants with site-directed mutagenesis in the BPA binding site amino acid residues.
Amino acid residues of ERRc receptors 1) Receptor binding potency IC 50 (nM)  Leu265, Leu324, Tyr330, Lys430, and His434 were mutated to Ala in order to determine their structural importance in the construction of BPA-binding pockets.
Leu265. Leu265 is present in the H3 a-helix, and we speculated that this Leu265 was one of the back support residues of Leu268 (Fig. 3C). As seen for the Leu342Ala-ERRc mutant receptor, Leu265Ala-ERRc exhibited no specific binding for [ 3 H]BPA (Table 1), indicating that Leu265 is critical for the binding to BPA. We could not perform both the saturation and competitive binding assays for [ 3 H]BPA. Since there is no direct contact between Leu265 and BPA, this Leu265 residue is judged to be a genuine back support residue of Leu268.
Leu324. Leu324 is on the S1 b-strand together with Tyr326. The S1 and S2 b-strands form a pleated sheet, which corresponds to a bottom plate of the ERRc ligand-binding pocket. Leu324 was speculated to be a back support residue of both Asn346 and Ile349 (Fig. 3E), and Leu324Ala-ERRc exhibited no specific binding for [ 3 H]BPA (Table 1). These results clearly imply that Leu324 is crucial for ERRc to bind BPA, particularly as an underlying support of Asn346 and Ile349.
Tyr330. Tyr330 is on the S2 b-strand, which forms a pleated sheet together with the S1 b-strand. The resulting pleated b-sheet, a bottom plate of the ERRc-LBP, has four amino acids (Leu324, Tyr326, Tyr330, and Met332) which stand close together inside the pocket (Fig. 3F). In addition to Leu265, we speculated that Tyr330 was a possible back support residue of Leu268 (Fig. 3C). When [ 3 H]BPA was examined in a saturation binding assay with the Tyr330Ala-ERRc mutant receptor, [ 3 H]BPA showed a more than 50-fold reduction in K d value (Table 1). In the homologous competitive binding assay of [ 3 H]BPA vs. BPA, BPA showed a more than 60-fold reduction in IC 50 value (Table 3). All these data indicate that Tyr330 is crucial as an underlying support of Leu268.
The importance of the Tyr-phenol hydroxyl group was proved by examination of the Tyr330Phe-ERRc mutant receptor. In the saturation binding assay, [ 3 H]BPA showed an approximately 2fold reduction in K d value (Table 1), and in the homologous competitive binding assay of [ 3 H]BPA/BPA, BPA showed an approximately 3-fold reduction in IC 50 value (Table 3). These results imply that the para-hydroxyl group of Tyr330 is an important structural element for the construction of ERRc-LBP on the bottom pleated b-sheet.
Lys430. Lys430 is present in the H11 a-helix, and we speculated that this Lys430 was a back support residue of Leu345 (Fig. 3G). When the Lys430Ala-ERRc mutant receptor was examined, [ 3 H]BPA showed only a 40% reduction in K d value in the saturation binding assay (Table 1), and a 70% reduction in IC 50 value in the homologous competitive binding assay of [ 3 H]BPA/BPA (Table 3). These results suggest that Lys430's back supporting ability is rather weak as compared to other residues. However, as shown in Table 1, the significance of Lys430 was evidenced by the much lower B max value of Lys430Ala-ERRc. These findings suggested that Lys430 is important for the construction of a sound and complete ERRc-LBP.
When 4-hydroxytamoxifen (4-OHT) was tested in a heterogeneous binding assay using [ 3 H]BPA for Lys430Ala-ERRc, it showed a rather stronger receptor binding potency (IC 50 = 4.10 nM) as compared with the potency (5.67 nM) for the wild-type ERRc (Table 3). This reinforced (about 40%) binding potency appeared to be due to the structural liberation of Leu345 from Lys430. Ala at position 430 might have been a better binding site for 4-OHT.
His434. His434 is present also in the H11 a-helix, and we speculated that this His434 was a back support residue of Leu342 and Leu345 (Fig. 3H). It was first replaced with Ala, and the resulting His434Ala-ERRc mutant receptor exhibited no specific binding for [ 3 H]BPA (Table 1). These results clearly imply that His434 is crucial for the binding of BPA by ERRc, particularly as a primary support of Leu342 and Leu345. When His was replaced Table 3. Receptor binding potency of BPA, 4-a-cumylphenol, and 4-OHT in the competitive binding assay using [ 3 H]BPA for human nuclear receptor ERRc and its mutants with site-directed mutagenesis in the back support residues of BPA binding sites.
Amino acid residues of ERRg receptors 1) Receptor binding potency IC 50 (nM) 1) Specifically mutated residues are designated in italics. 2) Because there was "no specific binding" in the saturation binding assay, the competitive binding assay could not be carried out.
3) The results of ERRc mutant receptors of Leu268, Ile310, and Val313 for BPA, except for those for 4-OHT and 4-a-cumylphenol, were re-collected from the previous report [24] for providing easiness in understanding. doi:10.1371/journal.pone.0101252.t003 with another aromatic amino acid, Phe, the resulting His434Phe-ERRc mutant receptor exhibited considerably reduced activities of BPA. When His434Phe-ERRc was examined, [ 3 H]BPA showed a 100% reduced K d value in the saturation binding assay (Table 1), and a 140% reduced IC 50 value in the homologous competitive binding assay of [ 3 H]BPA/BPA (Table 3). These results indicate that His434 makes an indispensable contribution to the back support, and thus Phe cannot replace it. The significance of His434 was further shown by the much lower B max value of His434Phe-ERRc (Table 1). When 4-a-cumylphenol was tested in the heterogeneous binding assay using [ 3 H]BPA for His434Phe-ERRc, this compound exhibited an almost unchanged high receptor binding potency (IC 50 = 5.12 nM), just like the potency (5.72 nM) for the wild-type ERRc (Table 3). This appeared to be due to the structural role of His434 as a primary support of Leu342 and Leu345, and perhaps Phe placed at position 434 provided better back support for these Leu residues, allowing them to accept the phenyl-B benzene ring of 4-a-cumylphenol. Interestingly, the opposite results were obtained for 4-OHT. 4-OHT exhibited a more than 40-fold reduction in receptor-binding potency (IC 50 = 234 nM) for His434Phe-ERRc as compared with that (5.67 nM) for the wildtype ERRc (Table 3). It is clear that Phe434 is quite unfavourable for the binding to 4-OHT, but not for the binding to 4-acumylphenol.

Binding of 4-a-cumylphenol and 4-hydroxytamoxifen
In this study we demonstrated that one of the two phenolhydroxyl groups of BPA makes a hydrogen bond with Asn346. The other phenol-hydroxyl group simultaneously forms hydrogen bonds with Glu275 and Arg316 [24,25]. 4-a-Cumylphenol, HO-C 6 H 4 -C(CH 3 ) 2 -C 6 H 5 , lacks one of the two phenol-hydroxyl groups of BPA, and it was found to bind to ERRc as potently as BPA [2,23]. This immediately raises the question of which hydrogen bond holds 4-a-cumylphenol in the ligand-binding pocket of ERRc. The X-ray crystal structure of the complex of 4a-cumylphenol/ERRc-LBD revealed that the phenol-hydroxyl group of 4-a-cumylphenol makes strong hydrogen bonds with Glu275/Arg316 of ERRc, whereas the phenyl group undergoes a hydrophobic interaction with the Leu345-isopropyl group [23]. The fact that 4-a-cumylphenol is as potent as BPA was thus substantiated by the similarity of their binding modes in the ERRc-binding pocket. However, one important structural issue arose; that is, 4-a-cumylphenol did not require the back-and-forth rotation of the Leu345-isobutyl group. In other words, BPA requires the back-and-forth rotation of the Leu345-isobutyl group merely because of the existence of the phenol-B hydroxyl group.
Given these facts, we evaluated the binding activities of 4-acumylphenol for a series of ERRc mutant receptors. It is clear that all the mutant receptors showed a reduced ability to bind 4-acumylphenol just as observed for BPA (Tables 2 and 3). The Asn346 mutant receptors were the only exception, and their activity decrements observed for 4-a-cumylphenol were much smaller (1.7-3.4 times) than those for BPA (4.0-11.3 times) ( Table 2). Since 4-a-cumylphenol lacks the BPA's phenol-B hydroxyl group, which is engaged in a hydrogen bond with Asn346, the smaller affinity drops must have been due to the lack of this hydrogen bond (Fig. 4) (Table 2). It should be noted that Leu345 is the most important among the residues for BPA binding, because the drop in affinity was the largest for the Leu345 mutant receptors (Table 2). This implies that Leu345 is responsible for the recognition followed by the strong affinity binding of BPA's phenol-B group. 4-Hydroxytamoxifen (4-OHT) works as an inverse agonist, which suppresses the constitutive high activity of ERRc [1,34,35]. Thus, 4-OHT is a different type of ligand and its binding site appears to be shared only partly with BPA. The molecular volume of 4-OHT is 319.9 Å 3 , while that of BPA is 187.6 Å 3 , indicating that the molecular size of 4-OHT is approximately 1.7-fold larger than that of BPA. When 4-OHT was tested in the competitive binding assays for a series of ERRc mutant receptors with [ 3 H]BPA as a tracer, essentially comparable activity profiles were observed as for BPA (Tables 2 and 3). The results suggested that 4-OHT and BPA share the binding site such as Leu342, Leu345, Asn346, and Ile349 residues on the H7 a-helix.

Effects of Ala-substitutions on the constitutive biological activity
ERRc is a constitutively active self-activating nuclear receptor, and thus ERRc exhibits almost the full activity without ligand [1,22,36]. For a series of Ala-substituted ERRc mutant receptors, the effects of mutations on such constitutive activity were evaluated by means of the luciferase reporter gene assay, and the results are shown in Fig. 5. When the mutant receptors Leu342Ala-ERRc, Leu345Ala-ERRc, Asn346Ala-ERRc, and Ile349Ala-ERRc were assayed, Asn346Ala-ERRc was immediately found to retain approximately 75% of the constitutive activity. In contrast, the other three mutant receptors showed clearly reduced activity: to 30-35% of the activity shown by the wild-type ERRc (Fig. 5A). These results suggested that a series of amino acids, the pair of Leu342 and Asn346 plus the pair of Leu345 and Ile349, both of which are present in H7, was structurally crucial for the binding of BPA, but the effect of this series on the constitutive activity of ERRc was rather moderate.
With respect to the constitutive activity, ERRc must retain its activation conformation, in which a-helix 12 sits on the ligandbinding pocket at an appropriate position. BPA can stay inside the LBP without changing this activation conformation. Most of the amino acid residues inside ERRc-LBP are simultaneously the binding site of BPA. In addition to Leu342, Leu345, Asn346, and Ile349, the residues Leu268, Leu271, Glu275, Leu309, Arg316, and Tyr326 are also involved. Their Ala-substituted ERRc mutant receptors were tested previously and the effects of mutations on the constitutive activity were also moderate. Occasionally, however, the Ala-substitutions of back support amino acids were found to cause profound damage to the constitutive activity. For instance, Ile310Ala-ERRc and Val313Ala-ERRc were almost completely inactive [24]. This result was also reproduced in the present study as shown in Fig. 5B. Since the conformational damage caused by the Ile310RAla and Val313RAla substitutions was found to be irreversible even with a high concentration of BPA, these amino acid residues are crucial for building the activation conformation of ERRc.
As to other back support amino acids, the Ala-substituted mutant receptors were also tested. The effects of these mutations on the constitutive activity were also moderate, as shown in Fig. 5B. Leu265Ala-ERRc and Tyr330Ala-ERRc exhibited approximately 45% and 30% of the constitutive activity of the wild-type ERRc, respectively, while both Leu324Ala-ERRc and His434Ala-ERRc exhibited approximately 70% of the wild-type level of constitutive activity. On the other hand, Lys430Ala-ERRc showed almost the full constitutive activity of the wild-type ERRc. All these results clearly indicate that the residual importance of the constitutive activity varies considerably, probably reflecting the difference in structural significance in building of the activation conformation.

Conclusion
Based on the results from the X-ray crystal structure analysis of the BPA/ERRc-LBD complex, we carried out an investigation to identify the binding site of the BPA phenol-B ring, by evaluating the receptor binding and biological activities of Ala-substituted mutant receptors. We confirmed that the amino acid residues Leu268, Leu342, Leu345, Asn346, and Ile349 form a receptacle ligand-binding pocket for the BPA phenol-B ring. We also confirmed that a number of amino acid residues, such as Leu265, Leu268, Ile310, Val313, Leu324, Tyr330, Lys430, Ala431 and His434, functioned as back support residues of these binding sites. The present results reveal that the strong binding of BPA to ERRc is supported by the double-layer binding sites.  . The results of luciferase-reporter gene assay to evaluate the biological activity of ERRc and its Ala-substituted mutant receptors. The percentage relative potencies of a series of mutant receptors were measured against the basal constitutive activity of the wild-type ERRc receptor (100%). (A) Binding site residues, and (B) Back support residues. An internal control that distinguishes the transcriptional level from variations in transfection efficiency was achieved by co-transfecting a second plasmid that constitutively expresses an activity that can be clearly differentiated from SEAP. The mark ''*'' shows that no basal constitutive activity was observed for Ile310Ala and Val313Ala. The assays were performed in triplicate at least three times (P,0.0001). doi:10.1371/journal.pone.0101252.g005

Supporting Information
GST-free ERRc-LBD proteins. (A) Wild-type ERRc, (B) Leu324Ala ERRc, and (C) Leu342Ala ERRc. CD spectra in the 200-260 nm UV region are shown by the mean molar ellipticity [h] (degrees?cm 2 /dmol). Thermal unfolding curves were depicted, by plotting the mean molar ellipticity [h] at 222 nm. Alasubstitutions of Leu324 (back support residue of Asn346 and Ile349) and Leu342 (direct binding site of BPA's phenol-B group) resulted in inactivity in binding BPA. (TIF) Figure S3 Receptor-binding assays of tritium-labeled bisphenol A (BPA) for the Leu342, Leu345, Asn346, and Ile349 mutant receptors of GST-fused ERRc-LBD. (A) Saturation binding assays with the curves of total binding (filled circle), non-specific binding (filled square), and specific binding (open circle). (B) Scatchard plot analyses showing a single binding mode with a binding affinity constant (K d ) and receptor density (B max ). No Scatchard plot analysis was carried out for the Leu342Ala-ERRc mutant receptor because of its lack of specific binding in the saturation-binding assay. All the saturation binding assays using [ 3 H]BPA were carried out at least three times and a representative result that afforded K d and B max values close to the means is shown for each mutant receptor. (TIF)