Molecular basis for the constitutive activity of estrogen-related receptor alpha-1.

Some orphan nuclear receptors, including estrogen-related receptor alpha-1 (ERRalpha-1), can activate gene transcription in a constitutive manner. Little is known about the molecular basis of the constitutive activity of these receptors. Our results from site-directed mutagenesis experiments have revealed that Phe-329 (analogous to Ala-350 in estrogen receptor alpha (ERalpha)) is responsible for the constitutive activity of ERRalpha-1. The ERRalpha-1 mutant F329A lost the transactivation activity and acted as a dominant negative mutant. The mammalian cell transfection experiments revealed that the ERRalpha-1 mutant F329A, like wild-type ERalpha, recognized toxaphene (an organochlorine pesticide) as an agonist. This compound was previously shown to be an antagonist of wild-type ERRalpha-1. On the other hand, like wild-type ERRalpha-1, the ERalpha mutant A350F was found to be constitutively active (as demonstrated by mammalian cell transfection and yeast two-hybrid assays). These results indicate that Phe-329 in ERRalpha-1 and Ala-350 in ERalpha play important roles in both ligand binding and transactivation function.

Based on ligand binding properties, nuclear receptors can be classified into three types. The first type of nuclear receptor, including estrogen receptor (ER) 1 and androgen receptor, is activated by specific ligands. The second type of receptor, similar to steroid and xenobiotic receptor, has a wide selectivity for ligands. The third type of receptor, like ERR␣-1, is transcriptionally active in the absence of exogenous ligand. The third type of receptor has similar structural domain arrangements to the first and the second types of receptors. Little is known as to why receptors such as ERR␣-1 are constitutively active. Our site-directed mutagenesis study has provided a molecular basis for the constitutive activity of ERR␣-1.
The cDNA for ERR␣-1 was first isolated by screening cDNA libraries using probes corresponding to the DNA-binding domain of the human ER␣ (1). Sequence alignment of ERR␣-1, ER␣, and ER␤ reveals a strong similarity. In the ligand binding region, the amino acid sequence of ERR␣-1 shows 36% identity when compared with ER␣ and 34% identity when compared with ER␤. However, ERR␣-1 does not bind to any of the major classes of steroids, including estrogens and androgens (1).
Vanacker et al. (2) suggested that ligands for ERR␣-1 might be present in fetal calf serum, based on results that showed a lack of the ERR␣-1 transactivation activity in ROS 17/2.8 cells under stripped serum conditions. Despite efforts from several laboratories, the physiological ligand for ERR␣-1 has not yet been identified. Experimentally, ERR␣-1 is transcriptionally active in the absence of exogenous hormone (see Fig. 2), and therefore, this receptor is generally considered to be a constitutive transcriptional activator protein. Utilizing yeast-based assays and mammalian transient transfection assays, we have recently found that two organochlorine pesticides, toxaphene and chlordane, can act as antagonists of ERR␣-1 that suppress the constitutive activity of ERR␣-1 (3). These findings are similar to those for the interaction of androstane metabolites with nuclear receptor CAR-␤ (4). In contrast to the antagonistic action on ERR␣-1, these pesticides have been reported to be weak agonists of ER␣ (5).

EXPERIMENTAL PROCEDURES
Materials-DNA sequencing kits were from United States Biochemical (Cleveland, Ohio). T4 kinase, T4 DNA ligase, and restriction endonucleases were purchased from New England Biolabs (Beverly, MA) and Roche Molecular Biochemicals. AmpliTaq polymerase was obtained from PerkinElmer Life Sciences. [ 14 C]Chloramphenicol ([D-threo-(dichloroacetyl-1-14 C)]) chloramphenicol (specific radioactivity, 55 mCi/ mmol) was from Amersham Pharmacia Biotech. The chloramphenicol acetyl transferase (CAT) expression vector, pUMSVOCAT, was a gift from Dr. K. Kurachi at the University of Michigan, Ann Arbor, Michigan. Oligonucleotide primers were synthesized in the DNA/RNA chemistry laboratory at the City of Hope. SK-BR-3 cells (ATCC (Manassas, VA)), derived from a human breast adenocarcinoma, were maintained in McCoy's 5A medium containing 10% fetal calf serum and glutamine. Estradiol and tamoxifen were purchased from Sigma. Toxaphene was kindly provided by Dr. Michael D. Shelby at the National Institute of Environmental Health Sciences. Dimethyl sulfoxide was from Mallinckrodt Chemical Works. The pSG5-GRIP1 was kindly provided by Dr. Michael R. Stallcup (University of Southern California, Los Angeles, CA). The ERE-TK-CAT plasmid was kindly provided by Dr. Ming-Jer Tsai (Baylor College of Medicine, Houston, TX).
Mutant Preparation-The mutants were generated by using a polymerase chain reaction (PCR)-based mutagenesis method described by Nelson and Long (6). Briefly, the PCR mutagenesis for the ERR␣1 mutant F329A used four primers: mutant Primer A for hERR␣1/F329A, TGTGACCTCGCTGACCGAGAG (a forward primer with the mutated bases underlined); Primer B, GGGGTACTAGTAACCCGGGATCCTCA-GTCCATCATGGCCTCGAG (a reverse hybrid primer containing a BamHI restriction site, a 3Ј-terminal 24-nucleotide sequence complementary to the ERR␣-1 cDNA in the reverse strand that encodes for the carboxyl terminus, and a 5Ј-terminal 20-nucleotide unique sequence); Primer C, GACGGATCCGAATTCATGTCCAGCCAGGTGGT-GGGCATTGAG (a forward primer containing an EcoRI restriction site and the cDNA sequence containing the first ATG site at the 5Ј-end); and Primer D, GGGGTACTAGTAACCCGGGA (a primer with a sequence that is identical to the 5Ј-terminal 20-nucleotide sequence of Primer B). PCR was carried out with a DNA Thermal Cycler 480 (PerkinElmer Life Sciences). For step 1, the reaction mixture (in 100 l) contained 2.5 * This work was supported by Grants ES08258 and CA44735 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. units of Taq polymerase, 1 pmol of pSG5-hERR␣-1 as template, and 100 pmol each of Primers A and B. The PCR cycles were 2 min at 95°C to denature the template DNA, 2 min at 45°C to allow the primers to anneal, and 2 min at 72°C for DNA extension and cycled 30 times. The PCR product was resolved over 1% agarose gel and extracted by using the Qiagen Gel Extraction Kit (Qiagen Inc., Chatsworth, CA). The Step 2 reaction contained about 0.6 pmol of the Step 1 product and 1 pmol of pSG5-hERR␣-1 and was run as a single cycle of 5 min at 95°C, 2 min at 45°C, and 10 min at 72°C. Primers C and D were then added (100 pmol of each), and 30 additional PCR cycles were completed. The final PCR products were extracted and digested with EcoRI and BamHI. The resulting full-length mutant cDNA for hERR␣1/F329A was ligated into the pSG5 expression vector. The sequence of the pSG5-F329A construct was confirmed by sequencing.
The ER␣ mutant A350F was also prepared by the PCR-based mutagenesis approach. Plasmid pIC-ERF (purchased from ATCC), which contains the complete coding sequence of human ER␣, was used as template. The sequences of Primers A(ER), B(ER), C(ER), and D(ER) to generate the full-length human ER␣ cDNA encoding hER␣/A350F are: 5Ј-CTGACCAACCTGTTCGACAGGGAG-3Ј, 5Ј-GGGGTACTAGTAAC-CCGGGCGGATCCTCAGACTGTGGCAGGGAAACCCTC-3, 5Ј-GCGG-AATTCGCCGCCGCCATGACCATGACCCTCCACACCAAAGC-3, and 5Ј-GGGGTACTAGTAACCCGGGC-3Ј, respectively. The PCR-amplified A350F-containing ER␣ cDNA fragment was digested with EcoRI and BamHI and subcloned into pSG5 vector through EcoRI and BamHI sites. The expression plasmid, pSG5-hER␣, for the expression of wildtype human ER␣, was generated as follows. The full-length coding region of human ER␣ was generated by PCR using Primers C(ER) and B(ER) (the same as those used in ER/A350F preparation) on template DNA, pIC-ERF. The PCR product was digested with EcoRI and BamHI and subcloned into pSG5 vector through EcoRI and BamHI sites.
Construction of Yeast Expression Plasmids-The yeast expression plasmids pGBT9-hER␣/HBD WT and pGBT9-hER␣/HBD A350F for the expression of Gal4DBD/hER␣HBD WT and Gal4DBD/hER␣HBD A350F fusion proteins were constructed as follows. The coding regions for wildtype and A350F mutant containing human ER␣ hormone-binding domains (amino acids 279 -595) were amplified by PCR using pSG5-hER␣ and pSG5-A350F as template DNA, respectively. The sequence of the forward primer is 5Ј-GCGGAATTCATGGAAGTGGGGTCTGCTGG-AGAC-3Ј, and the reverse primer is Primer B(ER), the same as the one In addition to pums-64/ϩ129CAT (2 g) and pSG5-GRIP1 expression plasmid (1 g), SK-BR-3 cells were transfected with pSG5-hERR␣-1 (1 g) or pSG5-F329A (1 g) as indicated. The CAT activity was measured after a 48-h transfection and plotted as the percentage of CAT activity from two independent experiments with each concentration in triplicate. As controls, the reporter activity in cells was transfected with the empty vector.
used for A350F preparation. The PCR products were digested with EcoRI and BamHI, and the digested DNA fragments were subcloned in frame into EcoRI-and BamHI-digested pGBT9 vector. The luciferase reporter plasmid that contains three copies of the ERE sequence, pGL3(ERE) 3 -luciferase, was constructed as follows. Two complimentary oligonucleotides for three tandem copies of ERE consensus sequences, 5Ј-AATTCCAGGTCAGAGTGACCTGAGCTAAAATACCAGGTCAGAG-TGACCTGAGCTAAAATACCAGGTCAGAGTGACCTGAGCTAAAAT-AC-3Ј and 5Ј-TCGAGTATTTTAGCTCAGGTCACTCTGACCTGGTATT-TTAGCTCAGGTCACTCTGACCTGGTATTTTAGCTCAGGTCACTCT-GACCTGG-3Ј, were synthesized, phosphorylated, annealed, filled in, and subcloned into pGL3 promoter vector through SmaI site. The sequences and the orientations of the inserts in all the constructs were confirmed by restriction digestion and DNA sequencing. The plasmid pGAD-GRIP1 322-1121 was the kind gift of Dr. Michael R. Stallcup (University of Southern California, Los Angeles).
Cell Transfection and CAT Assay-SK-BR-3 cells were transfected with the CAT plasmids by the use of Lipofectin (Life Technologies, Inc.) according to the manufacturer's instructions. The co-transfection exper-iments were performed 20 -24 h after seeding ϳ4 ϫ 10 5 cells per 60-mm tissue culture dish using 10 g of the test plasmid and 3 g of the plasmid pSV-␤-Gal, which was used to normalize the transfection efficiency. The overall amount of total DNA in all transfections was the same by including appropriate amounts of the empty vector, pSG5, in addition to specific amounts of the test plasmids indicated in each experiment. After overnight incubation, medium containing Lipofectin and DNA was removed, and the cells were cultured in the regular growth medium. After a 24-h incubation, the cells were harvested from the plates by scraping, pelleted by centrifugation, resuspended in 0.25 M Tris-HCl, pH 8.0, and disrupted by freeze-thawing three times. Aliquots of the lysate were used for an assay of ␤-galactosidase activity (7). The CAT activity in the cell extracts containing an equal amount of ␤-galactosidase activity from each sample was determined by the liquid scintillation counting method (8). Briefly, the appropriate amount of cell extracts was incubated in a reaction containing 14 C-labeled chloramphenicol and n-butyryl coenzyme A. The reaction products were extracted with a small volume of xylene. The xylene phase was mixed with scintillant and counted in a scintillation counter. The CAT activity was expressed as relative activity compared with that of the pUMSVO-CAT construct (1.0) and shown as the mean Ϯ S.E. of three independent transient transfection experiments performed for each construct.
For studying the interaction of ERR␣-1 with ligands, SK-BR-3 cells were initially cultured in growth medium containing charcoal/dextrantreated serum for 1 week. Compounds were added 24 h after cDNA transfection and incubated for another 24 h. Cells were harvested after FIG. 5. A, transactivation analysis of ER␣ mutant A350F. HeLa cells were transfected with a luciferase reporter plasmid ((ERE)3SV40_Luc, 0.5 g) and pSG5-A350F (0.1 g). The transfected cells were incubated with Me 2 SO (DMSO) or 100 nM E2 in Me 2 SO for 48 h. After cells were washed twice with 1 ϫ phosphate-buffered saline, the luciferase activity was measured. For experiments that involved GRIP1, the cells were also transfected with pSG5-GRIP1 (1 g). B demonstrates that wildtype ER␣ interacts with GRIP1 only in the presence of E2. The cells were transfected with (ERE)3SV40_Luc (0.5 g), pSG5-ER␣ (0.1 g), and when indicated, pSG5-GRIP1 (1 g). After incubation with Me 2 SO or 100 nM E2 for 48 h, the reporter activities were measured.

FIG. 6.
A, transactivation analysis of ER␣ mutant A350F in SK-BR-3 cells. B demonstrates that wild-type ER␣ interacts with GRIP1 only in the presence of E2. The experimental conditions are identical to those described in Fig. 5. DMSO, Me 2 SO. two washes with 1 ϫ phosphate-buffered saline, and the CAT activity was measured.
ER Functional Assay-Transactivation analysis of wild-type ER␣ and its mutant A350F was performed using HeLa cells and SK-BR-3 cells as the host cells. The cells were transfected with a luciferase reporter plasmid pGL3(ERE) 3 -luciferase (0.25 g) alone or along with 10 ng of pSG5-hER␣ or pSG5-A350F using Lipofectin in Opti-MEM medium. Five h after exposure to Lipofectin/DNA, the Lipofectin/DNAcontaining medium was removed, and the cells were cultured in 5% charcoal/dextran-treated fetal bovine serum and 100 nM E2-containing medium. Twenty-four h after transfection, the cells were harvested, lysed, and subjected to protein assay. An aliquot of cell lysates containing the same amount of protein from all the samples was used for luciferase assay according to the manufacturer's instructions (Promega). The luciferase activities were expressed as the mean and standard deviation of three independent experiments.
Yeast Two-hybrid Assay-The interaction between the wild-type ER␣ hormone-binding domain or its mutant A350F with GRIP1 was demonstrated by the yeast two-hybrid assays. The yeast strain Y187 was co-transformed with pGBT9-hER␣/HBD WT or pGBT9-hER␣/ HBD A350F and pGAD-GRIP1 322-1121 . The reporter ␤-galactosidase activities in transformants were measured following the published procedure (7).
Computer Modeling-Homology models (9) were built using software marketed by Molecular Simulations Inc. (San Diego, CA). The agonistbound form of ERR␣-1 was modeled using the DES-bound form of the estrogen receptor as a template (Protein Data Bank 3ERD, Ref. 10). The antagonist form was modeled using the tamoxifen-bound form (3ERT).

RESULTS AND DISCUSSION
We reviewed the amino acid sequence alignment between ERR␣-1 and ER␣ (1) and compared the corresponding residues in ERR␣-1 with those in the ER␣ ligand-binding site that are in direct contact with E2, as shown by Tanenbaum et al. (11) (Fig.  1). We have found that 9 out of the 19 residues are identical. Seven residues are conservative changes. One critical difference is between Phe-329 in ERR␣-1 and Ala-350 in ER␣ (Fig.  1). It was thought that this difference might be the cause of the differences in ligand binding properties between the two receptors. The phenyl group of Phe-329 in ERR␣-1 could provide steric hindrance to prevent the binding of E2 to this receptor. To test our hypothesis, we prepared ERR␣-1 mutants in which Phe-329 was replaced by the residues in ER␣, i.e. F329A as well as the ER␣ mutant A350F.
Studies from our laboratory have revealed that ERR␣-1 upregulates aromatase expression in human breast tissue by binding to a regulatory element, S1, that is situated near promoters I.3 and II of the human aromatase gene (12). Aromatase is an enzyme that converts androgens to estrogens. The DNA mobility shift assay revealed that the levels of the protein-S1 complexes for the mutant F329A and wild-type ERR␣-1 were very similar (results not shown), supporting our prediction that Phe-329 is situated in the ligand-binding domain, not in the DNA-binding domain.
To address the biological significance of the ERR␣-1 mutation, we co-transfected the SK-BR-3 human breast cancer cells with the expression plasmid for ERR␣-1 (or its mutant) along with a CAT reporter plasmid containing the aromatase genomic fragment having promoter I.3 and the ERR␣-1 binding element S1 (i.e. pums-64/ϩ129CAT, as described in Ref. 12). Although wild-type ERR␣-1 enhanced promoter I.3-driven CAT activity as reported previously (12), the mutant F329A reduced the basic promoter activity (Fig. 2). Furthermore, F329A suppressed the wild-type ERR␣-1 activity when they were coexpressed. These results indicate that F329A, acting as a dominant negative mutant, inhibited the activity of wild-type ERR␣-1. As reported previously (3), the nuclear coactivator GRIP-1 augments the enhancer activity of ERR␣-1 without the need of exogenous ligand. However, the F329A activity was not affected by the co-expression of GRIP-1 (Fig. 3). The latter result supports our prediction that Phe-329 is situated in the ligand-binding site. It is known that coactivators such as GRIP-1 interact with nuclear receptors at the ligand-binding domain when the receptor is in an agonist binding conformation. It was not expected that toxaphene, an antagonist for ERR␣-1 (3), could act as an agonist for F329A (Fig. 4). Toxaphene increased the F329A activity in a dose-dependent manner, whereas the same compound decreased the wild-type activity. Therefore, the ERR␣-1 mutant F329A becomes ER␣-like in that it recognizes toxaphene as a weak agonist. However, GRIP-1 was found not to enhance F329A activity in the presence of toxaphene, indicating that GRIP-1 could not recognize toxaphene-bound F329A (results not shown).
To better understand the importance of Phe-329 in ERR␣-1, we made and characterized the ER␣ mutant A350F. Our mammalian cell transfection experiments (in both HeLa and SK-BR-3 cells) have revealed that the ER␣ mutant A350F is transcriptionally active without the addition of E2 (Figs. 5 and 6).

FIG. 7. Demonstration of the interaction between the ER␣ mutant A350F and GRIP1 by yeast two-hybrid assays. A shows that
Gal4DB-A350F is transcriptionally active in yeast strain Y187 in the absence of Gal4 activation domain (AD) constructs. E2 was not required for the transactivation activity. The activity increased significantly when yeast was co-transformed with Gal4 AD-GRIP1 and Gal4DB-A350F. B shows that the reporter ␤-galactosidase activity can only be measured when yeast was co-transformed with Gal4DB-ER␣ (wildtype) and Gal4 AD-GRIP1 and when E2 (100 nM) was added. DMSO, Me 2 SO.
In fact, E2 was not able to further increase the transactivation activity, indicating that E2 cannot bind to this ER␣ mutant. Our experiments have also found that A350F can recruit GRIP-1 in the absence of ligands (Figs. 5 and 6). These results indicate that the ER␣ mutant A350F becomes ERR␣-1-like in that it is transcriptionally active in the absence of exogenous hormone.
Our yeast two-hybrid assay has found that the ER␣ mutant A350F, but not wild-type ER␣, has the ability to transactivate a heterologous promoter such as GAL4 promoter. It was found that a chimeric receptor consisting of the DNA-binding domain of GAL4 fused to the ligand-binding domain of ER␣ with the A350F mutation was transcriptionally active on a GAL4 reporter (Fig. 7). Ala-350 in ER␣ is within an autonomous activation domain AF-2␣ (amino acids 282-351) (14).
X-ray structural analyses of the ER␣ ligand-binding domain (10,11,15) have revealed that helix 12 assumes different conformations in response to ligand binding. In the presence of E2 and other agonists, helix 12 binds to a hydrophobic groove, the floor of which consists of ligands and side chains contributed by helices 3, 5/6, and 11. Antagonists such as tamoxifen partially fill the helix 12 groove, thus preventing helix 12 from assuming a conformation that recruits coactivator proteins (10). Results obtained from the analysis of the two mutants described in this paper allow us to conclude that, although we cannot exclude the possibility of the existence of a physiological ligand for ERR␣-1, this orphan receptor is transcriptionally active in the absence of ligands. A molecular model of wild-type ERR␣-1 reveals that the side chain of Phe-329 might be able to mimic bound ligand because it partially fills the binding pocket. This allows helix 12 to assume an active conformation capable of recruiting coactivator protein GRIP-1. As a result, the receptor is constitutively active (always turned on). This hypothesis has been evaluated by introducing a phenylalanine residue at the analogous position in ER␣ (A350F). As shown in Fig. 8A, the phenyl side chain fills nearly half the estrogen binding cavity, thus completing the floor of the groove in which helix 12 normally binds when the agonist is present. As predicted, the A350F mutant is constitutively active.
Because the putative ligand cavity of wild-type ERR␣-1 appears to be filled with side chains (most notably Phe-329), we predict that toxaphene binds in the hydrophobic groove normally occupied by helix 12 when it assumes the active conformation (Fig. 8B). Toxaphene therefore acts as an antagonist of wild-type ERR␣-1, displacing helix 12 and preventing the recruitment of coactivator. However, when space is created in the utative ligand cavity via the mutation of Phe-329 to Ala, our model suggests that toxaphene can fill the cavity (Fig. 8C) and thus serve as an agonist, which it does when tested experimentally. However, the cavity is still not large enough to accommodate E2 or tamoxifen, which confirms our observation that these compounds are not ligands for wild-type ERR␣-1 or the F329A mutant (data not shown). In addition, helix 12 does not assume a correct conformation for recruiting coactivator protein GRIP-1.
For classical receptors, ligand binding is essential for the transcription activity. Our results provide a structural basis as to why some orphan receptors such as ERR␣-1 are transcriptionally active in the absence of exogenous hormone. In ERR␣-1, the phenyl side chain of Phe-329 is situated in the ligand binding pocket that places the AF-2 domain in an active conformation. In addition, site-directed mutagenesis in the pro- FIG. 8. Computer modeling analysis of ER␣ and ERR␣-1. A, ER␣ (Protein Data Bank 3ERD) with alanine 350 mutated to phenylalanine (purple). This phenyl side chain fills a portion of the estrogen binding cavity (dotted region) and completes the binding surface for helix 12 (red), allowing it to assume an active conformation. B, model of wild-type (WT) ERR␣-1 showing toxaphene (yellow) in the helix 12 groove (the region occupied by the dimethyl amino side chain of tamoxifen when bound to ER␣). When bound in this location, toxaphene displaces helix 12, which then occupies the groove reserved for coactivator binding. It therefore acts as an antagonist. C, model of ERR␣-1 mutant (F329A). The removal of the bulky phenyl side chain creates a ligand cavity large enough to accommodate toxaphene but not estrogen. Here, toxaphene acts as an agonist. posed ligand binding pocket of ERR␣-1 modifies the binding properties of toxaphene, strengthening the conclusion that this pesticide is a ligand of this receptor. Toxaphene is among the 12 persistent organic pollutants identified by the United Nations Environment Program as requiring urgent attention. Its antagonistic effect on ERR␣-1 should not be overlooked.
In addition to Phe-329 in ERR␣-1 and Ala-350 in ER␣, there are two other nonconserved differences between the ligandbinding site of ERR␣-1 and that of ER␣. They are Gly-403 in ERR␣-1, Phe-425 in ER␣, Val-493 in ERR␣-1, and Gly-521 in ER␣ (Fig. 1). The ERR␣-1 mutant G403F has been generated and was found to have similar properties to the wild-type ERR␣-1, indicating that this position is not as important as Phe-329. As expected, the function of the double mutant F329A/G403F is similar to F329A (results not shown). E2 was found not to be a ligand for the ERR␣-1 mutant G403F or F329A/G403F. The ERR␣-1 mutant V493G was also generated recently and found to be constitutively active. Neither E2 nor toxaphene acts as a ligand of this mutant. Although it is not unexpected that E2 is not a ligand of V493G, the results suggest that Val-493 is important for toxaphene binding. Addi-tional mutagenesis experiments are being carried out to further characterize the role of Val-493 in toxaphene binding.