Characterization of a Novel Small Molecule Subtype Specific Estrogen-Related Receptor α Antagonist in MCF-7 Breast Cancer Cells

Background The orphan nuclear receptor estrogen-related receptor α (ERRα) is a member of the nuclear receptor superfamily. It was identified through a search for genes encoding proteins related to estrogen receptor α (ERα). An endogenous ligand has not been found. Novel ERRα antagonists that are highly specific for binding to the ligand binding domain (LBD) of ERRα have been recently reported. Research suggests that ERRα may be a novel drug target to treat breast cancer and/or metabolic disorders and this has led to an effort to characterize the mechanisms of action of N-[(2Z)-3-(4,5-dihydro-1,3-thiazol-2-yl)-1,3-thiazolidin-2-yl idene]-5H dibenzo[a,d][7]annulen-5-amine, a novel ERRα specific antagonist. Methodology/Principal Findings We demonstrate this ERRα ligand inhibits ERRα transcriptional activity in MCF-7 cells by luciferase assay but does not affect mRNA levels measured by real-time RT-PCR. Also, ERα (ESR1) mRNA levels were not affected upon treatment with the ERRα antagonist, but other ERRα (ESRRA) target genes such as pS2 (TFF1), osteopontin (SPP1), and aromatase (CYP19A1) mRNA levels decreased. In vitro, the ERRα antagonist prevents the constitutive interaction between ERRα and nuclear receptor coactivators. Furthermore, we use Western blots to demonstrate ERRα protein degradation via the ubiquitin proteasome pathway is increased by the ERRα-subtype specific antagonist. We demonstrate by chromatin immunoprecipitation (ChIP) that the interaction between ACADM, ESRRA, and TFF1 endogenous gene promoters and ERRα protein is decreased when cells are treated with the ligand. Knocking-down ERRα (shRNA) led to similar genomic effects seen when MCF-7 cells were treated with our ERRα antagonist. Conclusions/Significance We report the mechanism of action of a novel ERRα specific antagonist that inhibits transcriptional activity of ERRα, disrupts the constitutive interaction between ERRα and nuclear coactivators, and induces proteasome-dependent ERRα protein degradation. Additionally, we confirmed that knocking-down ERRα lead to similar genomic effects demonstrated in vitro when treated with the ERRα specific antagonist.


Introduction
ERRa is an orphan member of the superfamily of hormone nuclear receptors. The ERR subfamily consists of three members, ERRa, ERRb, and ERRc. ERRa was one of the first orphan receptors identified. It was found by using the DNA-binding domain (DBD) of Estrogen Receptor a (ERa) as a hybridization probe to screen recombinant DNA libraries [1]. Amino acid sequence comparison shows that apart from ERRb and ERRc, ERRa is more closely related to ERa and ERb than any other member of the superfamily of nuclear hormone receptors. ERRa and both ERa and ERb DNA Binding Domains share 70% amino acid identity. ERRa and ERa Ligand Binding Domains (LBD) share 36% amino acid identity; while ERRa and ERb LBD's share 37% amino acid identity [2,3]. In addition, although ERs and ERRs share a number of similar biochemical properties, ERRs do not bind 17b-estradiol (E2).
ERRa is known to bind to DNA as either a monomer or a dimer. ERRa can bind to estrogen-response elements (ERE) containing the recognition motif AGGTCAnnnTGACCT; ERRa also recognizes the single consensus half-site sequence TNAAGGTCA, referred to as an ERR-response element (ERRE) [4]. ERRa can bind the inverted repeat ERE as a dimer [5]. The binding of ERRa to an ERE or ERRE can lead to either a stimulatory or repressive event depending on the cell type, response element, context within a specific promoter, phosphorylation state of the receptor, potential ligands present, genomic context of ERRa (either competing or cooperating with ERa for binding), other receptors and coregulators present, and additional transcription factors involved [2]. Consequently, ERRs and ERs share common target genes (such as pS2, lactoferrin, and osteopontin) and exhibit cross-talk [6,7,8,9].
Whereas many other members of the steroid receptor superfamily are activated by ligand (including ERs), ERRs are constitutively active without the addition of a specific ligand. ERRa and ERRb have been shown to be constitutive activators of the classic ERE [10]. The authors also demonstrate that the p160 cofactors AIB1 (also known as SRC-3, NCoA3, ACTR, RAC3), GRIP1 (also known as SRC-2, NCoA2, TIF2) and SRC-1 (also known as NCoA1) potentiate the transcriptional activity by ERRa. It has been reported [9,10] using glutathione S-transferase (GST) pull down assays that ACTR (AIB1), SRC-1, and GRIP1 interact with the AF-2 domain of the LBD of ERRa without the addition of exogenous ligand. Moreover, fluorescence resonance energy transfer (FRET) assay has been used to demonstrate that SRC-1 and SRC-2 (GRIP-1) interact with all three ERRs without the addition of exogenous ligand. While ligands are not required for activation of ERR activity, there are known ligands which can modulate ERRs. Diethylstilbestrol (DES) antagonizes all three ERR isoforms whereas 4-hydroxytamoxifen (4-OHT) is an isoform specific inhibitor of ERRb and ERRc [11,12,13].
In addition to the p160 family of nuclear receptor coactivators that modulate ERR activity, another class of coactivators has also been reported. This class is made up of Proliferator-activated Receptor c Coactivator-1 a (PGC-1a) [14,15,16,17] and Proliferator-activated Receptor c Coactivator-1 b (PGC-1b) [18]. PGC-1a and PGC-1b are important regulators of genes that control many key aspects of metabolism including glucose uptake, gluconeogenesis, mitochondrial biogenesis, adipocyte cell fate specification, and adaptive thermogenesis [19]. PGC-1a interacts with ERRa and potentiates its transcriptional activity [14,15,16,17]. In a direct comparison of the binding affinities of SRC-1 and PGC-1a to bind ERRa, it has been shown that ERRa binds PGC-1a with 140-fold increased affinity in comparison to SRC-1 [20].
Our studies demonstrate that Compound A antagonizes ERRa transcriptional activity but shows little affect on ERRa mRNA levels. ERa mRNA and protein levels were not affected upon treatment with the ERRa antagonist, but other ERRa target genes such as pS2, osteopontin, and aromatase mRNA levels decreased upon treatment with the ERRa-subtype specific ligand. In addition, this ERRa tri-cyclic ligand antagonizes the constitutive interaction between ERRa and nuclear coactivators. We provide evidence that ERRa protein degradation is induced by the ERRa-subtype specific antagonist and this degradation is mediated though the ubiquitin 26S proteasome pathway. We report that the interactions between ERRa protein and the endogenous ERRa responsive gene promoters (ESRRA, ACADM and TFF1) are decreased by treatment with Compound A. Lastly, knocking-down ERRa by shRNA led to similar genomic effects seen when MCF-7 cells were treated with our ERRa antagonist.

Cell Lines and Reagents
The MCF-7 cell line (obtained from ATCC, Manassas, VA), and MCF-7/shRNA ERRa RNAi cell lines were maintained in EMEM with Earle's BSS and 2 mM L-glutamine that was modified to contain 1.0 mM sodium pyruvate, 0.1 mM nonessential amino acids, and 1.5 g/L sodium bicarbonate. It was also supplemented with 10% fetal bovine serum, 10 ug/ml bovine insulin, 100 units/ml of penicillin/streptomycin, at 37uC and 5% CO 2 . All cell lines were determined to be free of mycoplasma. Experiments that required maintenance of cells in ''stripped media,'' were washed with phosphate buffered saline, and then the media was changed to EMEM without phenol red that contained 10% charcoal dextran -treated FBS (CD-FBS).

Transient Transfection and Luciferase Assay
MCF-7 cells were maintained in phenol free EMEM/CD-FBS media for 4 days prior to performing transfections. 25,000 cells per well were plated into 96-well plates. 0.125 ug pGL2 Luc (Promega), or 0.125 ug p3xERE-TK-LUC [pGL2 plasmid (Promega, Madison, WI) containing 3 tandem repeats of the estrogen response element (ERE) sequence 59 -TTTGAT-CAGGTCACTGTGACCTCTAGAGT-39, placed upstream of a minimal herpes simplex thymidine kinase (TK) promoter directing the expression of the luciferase coding sequence (a generous gift from Tina Chang, Merck Research Laboratories, Rahway, NJ) and 0.0125 ug of phRL-TK renilla plasmid (Promega) were co-transfected in triplicate wells along with (when indicated) either 0.0625 ug pcDNA 3.1 (Invitrogen, Carlsbad, CA) or 0.0625 ug pcDNA3.1 hERRa (described below). Plasmids were diluted in OptiMEM (Invitrogen), the transfection reagent Lipofectamine LTX (Invitrogen) was added to the DNA solution and incubated for 25 minutes at room temperature. Next, the DNA-Lipofectamine complexes were added to the cells and incubated overnight. Subsequently, the cells (in triplicate) were treated with vehicle (DMSO), 100 pM 17-beta-estradiol (Sigma), 5 uM Compound A (Merck & Co, West Point, PA), 100 pM 17beta-estradiol/5 uM Compound A, or 1 uM ICI-182,780 (faslodex) (Tocris, Ellisville, MO) for 48 hours. Cells were then harvested and cell lysates were assayed for luciferase activity (renilla normalized) as per the manufacturer's directions by utilizing the Dual-Luciferase Reporter Assay System (Promega) and the Wallac Victor plate reader (Perkin Elmer, Wellesley, MA).

Real-time RT-PCR
MCF-7 cells were maintained in phenol free EMEM/CD-FBS media for 4 days prior to drug treatments (in triplicate) and MCF-7/shRNA ERRa RNAi cells were maintained in normal media containing whole serum (described above). Total RNA was extracted from MCF-7 cells treated with either DMSO (Control) or 5 uM Compound A for 24 or 48 hours. The RNA samples were DNase I (Ambion Inc., Austin, TX) treated and cDNA was synthesized (High-Capacity cDNA Archive Kit, Applied Biosystems). Real-time RT-PCR was performed with an ABI 7900 HT sequence detection system (Applied Biosystems, Foster City, CA). Primer/probe sets for target genes: human ERRa (ESRRA) (Hs00607062_gH), human ERa (ESR1) (Hs00174860_m1), human PGC-1a (PPARGC1A) (Hs00173304_m1), human PDK4 (PDK4) (Hs00176875_m1), human osteopontin (SPP1) (Hs00959010_m1), human pS2 (TFF1) (Hs00170216_m1), human ACADM (ACADM) (Hs00163494_m1) and 18S rRNA endogenous control (4308329) were purchased from Applied Biosystems. The housekeeping gene18S rRNA was used as the internal quantitative control for normalization. Relative gene expression was calculated with the DDCt method as outlined in the Applied Biosystems User Guide. In brief, the threshold cycle (C T ) values for the target gene and reference (18S) were determined by ABI PRISM Sequence Detection System software. Mean C T values and standard deviations were calculated in Microsoft Excel. DC T was calculated by DC T = C T target2C T reference. After the mean and standard deviation of the DC T 's value were determined, DDC T = DC T test sample2DC T calibrator sample. Next, the standard deviations of the DDC T values were calculated and finally, the fold-differences were determined by the DDC T , expressed as 2 2DDCT .

ERRa Antibody
The ERRa specific peptide sequence: AGPLAVAGGPRK-TAAPVN, was synthesized by EvoQuest Custom Antibody Services (Invitrogen). The peptide was then coupled to a hapten carrier (keyhole limpet hemocyanin) for immunization. Polyclonal antibodies were generated in New Zealand white rabbits and the polyclonal ERRa specific antibody (pAb ERRa) was purified by affinity chromatography.

In Vitro Expression of ERa and ERRa
Full-length human ERa (pcDNA3.1 hERa) and full-length human ERRa (pcDNA3.1 hERRa) proteins were expressed by a TnT coupled reticulocyte lysate system as per the manufacturer's recommended conditions (Promega Corporation) for use as positive controls with Western Blotting. In addition, [ 35 S]methionine was added to the transcription/translation reaction for radiolabeled hERRa that was used in the biotinylated pull-down assays.

Biotinylated Pull-down Assay
ProFound Pull-Down Biotinylated Protein:Protein Interaction Kit (Pierce, Rockford, IL) was utilized and manufacture's instructions were followed. Briefly, streptavidin beaded agarose (washed with 16 TBS) was incubated with biotinylated protein in 16 TBS at 4uC in provided spin column for 1 hour on a rocking platform and then columns were centrifuged. Biotin blocking solution was added, samples were incubated and centrifuged. Biotin blocking step was repeated once followed by washes with 16 TBS. In vitro translated [ 35 S]-labeled protein in 16 TBS was added along with DMSO (control), 10 mM Compound A, or 10 mM DES to biotinylated protein bound to streptavidin beaded agarose. The samples were then incubated for 4 hours at 4uC on a rocking platform. The bound protein was washed with 16 TBS and the beads were collected by centrifugation. The bound protein was eluted in SDS sample buffer, loaded into a 10% NuPage Bis-Tris Gel (Invitrogen) and analyzed by phosphorimaging (Typhoon 9400, ImageQuant TL software, GE Healthcare

Chromatin Immunoprecipitation (ChIP) Assay
After 4 days of growing the MCF-7 cells in EMEM without phenol red that contained 10% CD-FBS, the cells were treated with DMSO or 5 uM Compound A in duplicate for 24 and 48 hours. Subsequently, the cells were fixed according to Genpathway, Inc. cell fixation protocol which can be found at www.genpathway.com and the chromatin immunoprecipitation were carried as described (43)   was utilized to obtain whole cell extractions by following the manufacturer's protocol. Protein concentration was determined with the DC Protein Assay (Bio-Rad, Hercules, CA) and 20 ug of nuclear protein extracts or 40 ug of whole cell protein extracts were loaded into a 10% NuPage Bis-Tris Gel (Invitrogen). After electrophoresis, the proteins were transferred to nitrocellulose membrane (Invitrogen). Western blotting was carried out by utilizing pAb ERRa (described above) or ERa (G-20, Santa Cruz Biotechnology, Santa Cruz, CA), ECL rabbit IgG, HRP-linked whole antibody (from donkey) (GE Healthcare, Piscataway, NJ), and Western Lightning Chemiluminescence Reagent Plus (Perkin Elmer, Boston, MA). Equal loading of nuclear protein extracts per lane was assessed by Coomassie Brilliant Blue staining of a gel run in parallel while equal loading of whole cell protein extracts per lane was assessed by stripping the nitrocellulose membrane and reprobing with b-actin monoclonal antibody (Sigma). Densitometric quantification of protein levels from three independent experiments were performed by using the AlphaEase FC Imaging Software program (Alpha Innotech, San Leandro, CA). MCF-7/ shRNA ERRa RNAi cell line nuclear protein extraction, protein concentration determination, electrophoresis, protein transfer, and Western blotting are described above.

Analysis of ERa and ERRa Protein Levels
Stable Transfection with shRNA ERRa and shRNA (-) Plasmids 0.5 ug of four unique SureSilencing shRNA plasmids (Super-Array Bioscience Corporation, Frederick, MD) specific for human ERRa and a negative control were transfected separately into 40,000 MCF-7 cells per well (24-well plate). shRNA (-) and shRNA ERRa plasmids were under the control of the U1 promoter and also contain GFP. Plasmids were diluted in OptiMEM (Invitrogen), the transfection reagent Lipofectamine LTX (Invitrogen) was added to the DNA solution and incubated for 25 minutes at room temperature. Next, the DNA-Lipofectamine complexes were added to the cells, incubated for 24 hours, and then cells were replenished with fresh media. Subsequently, transfected cells were expanded.

Statistics
Error bars represent standard error of the mean (SEM) between replicates of a given experiment. Comparisons between two groups were made by analysis of variance (ANOVA) followed by a student t-test at 0.05 significance level with P values indicated.

Compound A Inhibits Constitutive Transcriptional Activity of ERRa
ERRa binding to ERE's and subsequent constitutive transactivation has been shown [10,13,26]. In addition, Kraus et al [26] has reported that ERRa can modulate estrogen responsiveness and effectively compete with ERa for the binding to EREs. To test whether our novel ERRa ligand antagonizes the constitutive transcriptional activity of ERRa, we cotransfected MCF-7 breast cancer cells with the reporter plasmid p3xERE-TK-Luc or the ERE-negative control plasmid (pGL2 Luc) together with the control parental vector pcDNA3.1 and phRL-TK renilla plasmid for normalization. The co-transfected cells were then treated with vehicle (DMSO), 100 pM E2, 5 uM Compound A, 100 pM E2/ 5 uM Compound A, or 1 uM ICI -182,780. After 48 hours cells were harvested and cell lysates were assayed for luciferase activity. Cells transfected with the p3xERE-TK-Luc and treated with DMSO demonstrated constitutive (basal level) activity with a 73fold increase in transcriptional activity above cells transfected with pGL2 Luc and treated with DMSO (Fig. 1A). MCF-7 cells transfected with the p3xERE-TK-Luc and treated with E2 had a 6.3-fold increase in transcriptional activity verses the basal level. Cells treated with the ERRa antagonist, Compound A are transcriptionally repressed, 0.73-fold (or 27% decrease) below basal level (P = 0.032) (Fig. 1A inset).
To study the effect of the ERRa-subtype specific antagonist on estrogen dependent transcriptional activity, cells were treated with E2 in combination with the ERRa antagonist (Fig. 1A). When cells were treated with estrogen plus Compound A there was a reduction in transcriptional activation. Additionally, the estrogen receptor selective antagonist, ICI -182,780, greatly reduces any transactivation suggesting that ER is needed for estrogen stimulated expression to occur at the ERE (Fig. 1A). Thus, taken together, the functionality of the interconnections/cross-talk of ERRa and ERa at an ERE is illustrated.
Would having more ERRa present lead to greater antagonism by Compound A? To answer this question, a similar cotransfection experiment described above was performed and the ERRa expression plasmid (pcDNA3.1 hERRa) was co-transfected (instead of control parental vector pcDNA3.1) with either p3xERE-TK-Luc or pGL2 Luc. Over-expressing ERRa in MCF-7 cells leads to decreased down modulation within all treatment groups relative to luciferase activity (compare Fig. 1A to 1B). Cells transfected with the p3xERE-TK-Luc+pcDNA 3.1 hERRa demonstrated constitutive (basal level) activity by conferring an 114-fold increase in transcriptional activity above cells transfected with pGL2 Luc and treated with vehicle (Fig. 1B). MCF-7 cells transfected with the p3xERE-TK-Luc+pcDNA 3.1 hERRa and treated with 100 pM E2 conferred a 2.6-fold increase in transcriptional activity verses the basal level. Moreover, while treating cells with 5 uM Compound A lead to a 27% decrease below the basal constitutive activity in the original experiment (Fig. 1A); over expressing ERRa leads to a 45% decrease in transactivation upon treatment with 5 uM Compound A (P = 0.001) (Fig. 1B inset). Similarly, 5 uM Compound A plus 100 pM E2 also led to a 33% decrease. Therefore, while transfecting ERRa into MCF-7 cells leads to an overall decreased modulation of estrogen responsiveness (also previously reported [26]), a larger window of antagonism was also demonstrated with more ERRa present. Over expressing ERRa in MCF-7 cells nearly abolishes transactivation upon treatment with 1 uM ICI 182,780 (Fig. 1B).

Compound A Suppresses Expression of ERRa Target Genes
Since Compound A was shown to inhibit the constitutive transcriptional activity of ERRa (Fig. 1), we next wanted to examine the effects of Compound A on target gene expression at the mRNA level. Quantitative real-time RT-PCR was used to measure ERRa (ESRRA) target gene mRNA levels in MCF-7 cell treated with 5 uM Compound A for either 24 or 48 hours. No reduction in steady-state ERRa or ERa (ESR1) mRNA levels was measured after 24 and 48 hours of treatment with the ligand (Fig. 2, S1). On the contrary, other known ERRa target genes including medium-chain acyl coenzyme (ACADM) [4], aromatase (CYP19A1) [27], pyruvate dehydrogenase kinase 4 (PDK4) [28], osteopontin (SPP1) [7], and pS2 (TFF1) [6] were all down modulated upon treatment with the compound for 48 hours (Fig. 2, S1). Additionally, peroxisome proliferator-activated receptor coactivator-1a (PGC-1a) (PPARGC1A), although not considered an ERRa target gene, is known to bind and interact with ERRa [14,29], was robustly down modulated upon treatment with the ERRa antagonist in comparison to cells treated with vehicle alone (Fig. 2, S1).

Compound A Decreased Constitutive Interactions of ERRa with Nuclear Coactivators AIB1, GRIP-1 and PGC-1a
Our group has previously reported an IC 50 of 170 nM for Compound A in an ERRa LBD/PGC-1a coactivator homogenous time-resolved fluorescence interaction assay [21]. To further study this inhibitory effect we performed biotinylated pull-down assays with AIB1, GRIP-1, and PGC-1a nuclear coactivators to look at the effects of Compound A on ERRa/nuclear coactivator constitutive interaction. The receptor interaction domains (RID) of AIB1 (aa 557-773), GRIP-1 (aa 565-798), and PGC-1a (aa 1-338) were expressed, biotinylated, and purified. Full-length human ERRa protein was expressed and 35 S[methionine] labeled. After incubating ERRa, nuclear coactivator, and 10 uM ligand, a standard streptavidin bead/biotinylated pull-down assay was carried out (see Materials and Methods). The constitutive interaction of ERRa with nuclear coactivators AIB1, GRIP-1, or PGC-1a was unaffected by the presence of DMSO, but was considerably reduced upon treatment with Compound A or the known ERRa antagonist DES [11,13] (Fig. 3). Upon treatment with Compound A, AIB1 showed a 35% reduction in comparison to the vehicle (DMSO) treated sample (Fig. 3A). Similar results were also seen with nuclear coactivators GRIP-1 (Fig. 3A) or PGC-1a with either Compound A or DES (Fig. 3B). The nuclear coactivator PGC-1a exhibited the greatest release with an 81% reduction (Fig. 3B). The reduction of nuclear coactivator levels upon treatment with the ERRa antagonist demonstrates disruption of the constitutive interaction between ERRa and these coactivators.
To further support the finding that Compound A disrupts the constitutive interaction between ERRa and nuclear coactivators, we also performed ChIP assays after MCF-7 cells were treated with vehicle (DMSO) or 5 uM Compound A for 24 and 48 hours. The cells were fixed, chromatin was immunoprecipitated with anti-GRIP-1 antibody, and quantitative real-time PCR (Q-PCR) was performed with primers targeting well-studied estrogenrelated receptor response elements (ERREs) in ERRa (ESRRA) [30,31], ACADM (ACADM) [4,32], and pS2 (TFF1) promoters [6,33]. At 24 hours, treatment with Compound A significantly decreased association of ERRa/GRIP-1 (P,0.001) (Fig. 3C top) to a region in ESSRA (ERRa gene), while at 48 hours decreased association events of ERRa/GRIP-1 to genomic regions flanking ESSRA, ACADM, and TFF1 (DMSO vs. Cmpd A, P,0.001) (Fig. 3C bottom) were detected. Luc ERE (empty vector control) or p3xERE-TK-LUC and phRL-TK renilla plasmid were cotransfected into MCF-7 cells. After indicated cell treatment, cells were harvested and cell lysates were assayed for luciferase activity (renilla normalized). Cells transfected with the p3xERE-TK-Luc showed constitutive activity compared to the empty vector pGL2 Luc treated with vehicle. MCF-7 cells transfected with the p3xERE-TK-Luc and treated with E2 conferred a 6.3-fold increase in transcriptional activity while cells treated with Compound A are significantly repressed, 0.73-fold (or 27% decrease) below basal level (*, P = 0.032) (Fig. 1A inset) or when treated with Compound A in combination with E2, Compound A still represses the transcriptional effect conferred by ERRa. ICI -182,780 also greatly reduces any transactivation. (B) The same experiment (described above) was performed along with either pcDNA 3.1 (empty vector control) or pcDNA3.1 hERRa as indicated. Over-expressing ERRa in MCF-7 cells increases down modulation (relative luciferase activity) within all treatment groups. Cells transfected with the p3xERE-TK-Luc+pcDNA 3.1 hERRa conferred an 114-fold increase in transcriptional activity above cells transfected with pGL2 Luc and treated with vehicle. Cells transfected with the p3xERE-TK-Luc+pcDNA 3.1 hERRa and treated with E2 exhibited a 2.6-fold increase in transcriptional activity in comparison to pGL2 Luc+pcDNA3.1 hERRa. Over expressing ERRa leads to a 45% decrease in transactivation upon treatment with Compound A (P = 0.001) (Fig. 1B inset). Similarly, Cmpd A+E2 led to a 33% decrease and over expressing ERRa in MCF-7 cells still led to nearly abolishing transactivation upon treatment with ICI 182,780. Results are expressed as the normalized luciferase activity (mean6SEM) of three independent experiments performed in triplicate. Differences in luciferase activity between vehicle (DMSO) and Cmpd A were measured by ANOVA followed by a student t-test with a 0.05 significance level. *, P = 0.032 and **, P = 0.001. doi:10.1371/journal.pone.0005624.g001

Compound A Induces ERRa Protein Degradation
It has been well established that different ER ligands have different effects on ERa protein stability and degradation. For example, at 48 hours 4-hydroxytamoxifen (4-OHT) or idoxifene increases ERa protein levels, estradiol (E2) or ICI -182,780 induces protein degradation, while others like raloxifene display little effect [34,35]. Also, given that ERa in some contexts is most likely needed for ERRa activation to occur (Fig. 1), we examined the expression status of ERRa after MCF-7 cells were treated with vehicle (DMSO) or the selective estrogen receptor modulators (SERMs) 10 nM E2, 10 nM 4-OHT, and 10 nM ICI-182,780 for 24 and 48 hours. Vehicle (DMSO) does not effect protein stability, while as previously reported [35] E2 or ICI-182,780 leads to degradation while 4-OHT leads to an increase in ERa (Fig. 4A  top). Interestingly, 4-OHT and ICI-182,780 treated MCF-7 cells do not alter ERRa levels, while there is a 30% increase in ERRa after 48 hour treatment with E2 (Fig. 4A bottom). To study the effects of ERRa and ERa stability after treatment with the ERRa ligand Compound A, MCF-7 cells were treated with vehicle (DMSO) or 5 uM Compound A for 12, 24, and 48 hours. Nuclear extracts were isolated and ERRa and ERa protein levels were analyzed by Western blot. After a 12 hour incubation with Compound A, a 20% reduction of ERRa protein was seen (compare lanes 3 and 4, Fig. 4B); while after 24 hours a 27% reduction (compare lanes 5 and 6) and after 48 hours a 74% reduction (compare lanes 7 and 8) of ERRa was detected. Treating cells with the ERRa antagonist for 12, 24, or 48 hours yielded negligible ERa protein level changes (Fig. 4B).
To determine whether down regulation of the ERRa protein is mediated by the ubiquitin proteasome pathway, we treated MCF-7 cells with vehicle (DMSO), 5 uM Compound A, the proteasome inhibitor MG132 (1 uM), or 5 uM Compound A plus 1 uM MG132 for 36 hours. Whole cell extracts were isolated and ERRa protein levels were analyzed by Western blot. Cells treated with the ERRa antagonist (lane 3, Fig. 4C) exhibited a 51% reduction in comparison to vehicle (lane 2). Addition of MG132 slightly reduced ERRa (lane 4 compared to lane 2) while addition of MG132 blocks ERRa degradation caused by Compound A (lane 5 compared to lanes 2, 3, and 4) (Fig. 4C). Thus, our results suggest that Compound A down-regulation of ERRa involves ubiquitinmediated proteolysis.

Treatment with the ERRa Antagonist Decreased Association between ERRa and ERRa Targeted Promoters
In order to investigate the effect of Compound A on ERRa binding at the promoter region of ERRa target genes (ESRRA, ACADM, and TFF1), chromatin Immunoprecipitation (ChIP) assays were performed after MCF-7 cells were treated with DMSO, 3 pM 17b-estradiol (E2), or 5 uM Compound A for 24 and 48 hours. The cells were fixed, chromatin was immunoprecipitated with anti-ERRa antibody, and quantitative real-time PCR (Q-PCR) was performed with primers targeting well characterized estrogen-related receptor response elements (ERREs) in ERRa (ESRRA) [30,31], ACADM [4,32], and pS2 (TFF1) promoters [6,33]. At 24 hours, treatment with Compound A had little or no effect on ERRa association with these target genes (Fig. 5A), while at 48 hours decreased association of ESRRA, ACADM, or TFF1 (P,0.001) (Fig. 5B) was demonstrated, Figure 3. Constitutive interaction of ERRa and nuclear coactivators is reduced upon treatment with ERRa antagonist. Biotinylated pull-down assays with nuclear coactivators AIB1, GRIP-1 (Fig. 3A), and PGC-1a (Fig. 3B). Lane 1 molecular weight ladder. Lane 2,   (Fig. 4A top). Neither 4-OHT or ICI-182,780 treated cells altered ERRa levels at 24 or 48 hours, while there is a 30% increase in ERRa after 48 hour treatment with E2 (Fig. 4A bottom). (B) MCF-7 cells were treated with vehicle (DMSO) or Compound A for 12, 24, and 48 hours. ERRa and ERa protein levels (nuclear extracts) were analyzed by Western blot. After a 12 hour incubation with Compound A, a 20% reduction of ERRa protein was seen; while after suggesting a decreased association between ERRa and ERRa targeted promoters.

Silencing of ERRa Decreases mRNA levels of ERRa Target Genes but not ERa
Does reduction of ERRa expression lead to similar effects seen by antagonizing/down-regulating ERRa with Compound A? Four different (1-4) plasmids (SuperArray Bioscience Corporation) expressing short hairpin RNAs (shRNA) specific for ERRa under the control of the U1 promoter and containing the GFP marker gene were transfected separately into MCF-7 cells. Additionally, an shRNA expressing a scrambled artificial sequence that does not match any human, mouse or rat gene was transfected and used as the negative control. MCF-7/shRNA ERRa3 cells underwent four rounds of fluorescent activated cell sorting (FACS) to enrich for GFP expressing cells (Fig. 6A). Similar results were seen with MCF-7/ shRNA ERRa2 (data not shown). ERRa mRNA expression was measured (Fig. 6B-top panel, S2) and as a higher number of GFP expressing cells were sorted and isolated (rounds I-IV), a decrease in ERRa mRNA levels were detected. While only 51% reduction of ERRa (versus the negative control) was measured with cells that underwent two rounds of FACS, after 4 rounds the enriched population exhibited a statistically significant (P,0.05) 79% knockdown. Similar results were seen with MCF-7/shRNA ERRa2 (data not shown). Along with ERRa, ACADM and PGC-1a mRNA expression levels were also determined and statistically significant (P,0.05) reduction of ACADM and PGC-1a was measured in both after 4 rounds of FACS (Fig. 6B, S2). ERRa protein expression was measured (Fig. 6C) and MCF-7/shRNA ERRa3 cells exhibited 69% less protein versus the negative control, while MCF-7/shRNA ERRa2 ERRa protein levels were reduced by 59%. Similarly to when MCF-7 cells were treated with Compound A (Fig. 2) knocking-down ERRa by shRNA led to significant decreases (P,0.05) in expression of ERRa target genes aromatase (CYP19A1), osteopontin (SPP1), and pS2 (TFF1) while ERa (ESR1) levels were not affected (Fig. 6D, S3).
First, Compound A inhibits the constitutive transcriptional activity of both endogenous and ectopically expressed ERRa (Fig. 1A). When ERRa was overexpressed in MCF-7 cells, a greater window of repression by Compound A was exhibited, and the overall estrogen responsiveness (measured by an ERE reporter construct) was down-modulated (Fig. 1B) -an interesting event first reported by Kraus and colleagues [26]. Our previous research reports that Compound A specifically binds ERRa [21]. Compound A did not exert a direct effect on ERa either by modulating mRNA expression or altering protein stability. Furthermore, while Compound A does not modulate ERRa mRNA expression, it induces proteasome-dependent ERRa protein degradation (Fig. 2 &  4C). Lanvin and colleagues recently reported the ERRa inverse agonist XCT790 does not act on ERa or ERRa mRNA level, nor does it modify ERa protein stability, but it also induces proteasome 24 hours a 27% reduction and after 48 hours a 74% reduction of ERRa was exhibited. Additionally, treating cells with ERRa antagonist for 12, 24, or 48 hours yielded negligible ERa protein level changes (Fig. 4B bottom). (C) MCF-7 cells were treated with vehicle (DMSO), Compound A, MG132, or Compound A plus MG132 for 36 hours. ERRa protein levels (whole cell extracts) were analyzed by Western blot. Cells treated with Compound A exhibited a 51% reduction in comparison to vehicle. Addition of MG132 blocks ERRa degradation caused by Compound A. Equal loading of nuclear protein extracts per lane was assessed by Coomassie blue staining of gels (Fig. 4A,B) while additionally; equal loading of whole cell protein extracts per lane was assessed by stripping the nitrocellulose membrane and re-probing with b-actin monoclonal antibody (Fig. 4C). Densitometric quantification of protein levels is described in Materials and Methods. All Western blots included human full-length ERa and/or ERRa in vitro translated proteins which were used as positive controls. Results shown are representative of three independent experiments. doi:10.1371/journal.pone.0005624.g004 dependent ERRa protein degradation [23]. Furthermore, when MCF-7 cells were treated with Compound A for 12 hours and 24 hours, modest 20% and 27% reduction in protein levels are seen (Fig. 4B). But at 48 hours, a robust 74% reduction was observed. When MCF-7 cells were treated with Compound A for 48 hours, followed by ChIP performed with anti-ERRa antibody there was a significant decreased association to the three promoters (ESRRA, ACADM, and TFF1) (Fig. 5B). Therefore, based on our ERRa ChIP (Fig. 5B) and our Western blot data (Fig. 4B, C), the decrease in association of ERRa at the promoter region of ERRa target genes, is most likely due to protein degradation of ERRa caused by Compound A.
It has been previously demonstrated that the ERR antagonist DES interferes with the constitutive interaction between ERRc and the nuclear coactivator GRIP-1 [13]. Therefore, to investigate whether or not Compound A could antagonize the constitutive interaction between ERRa and nuclear coactivators we performed biotinylated pull-down assays with AIB1, GRIP-1, and PGC-1a nuclear coactivators. Our data suggest that Compound A promotes nuclear coactivator release between the ERRa and AIB1, GRIP-1, or PGC-1a (Fig. 3A,B).
To extend the finding that Compound A is acting like an antagonist and disrupts the constitutive interaction between ERRa and nuclear coactivators, ChIP assays were carried out after MCF-7 cells were treated with vehicle or Compound A. At 24 hours, a significant decrease in the interaction between GRIP-1/and ERRa (DMSO vs. Compound A, P,0.001) (Fig. 3C top) was exhibited. In comparison, when MCF-7 cells were treated with the ERRa antagonist for 24 hours, only a 27% reduction in ERRa protein levels were seen (Fig. 4B). Thus the ability of Compound A to directly interfere with cofactor association is demonstrated by ChIP. At 48 hours, decreased binding of ERRa (ESRRA), ACADM, and pS2 (TFF1)/GRIP-1 (DMSO vs. Cmpd A, P,0.001) (Fig. 3C bottom) was demonstrated and at 48 hours, a robust 74% reduction is reported (Fig. 4B). Therefore, at 48 hours, the significant (DMSO vs. Compound A, P,0.001) decrease in association between GRIP-1 and ERRa target genes is most likely due to degradation of ERRa protein.
In order to confirm that antagonizing/down-modulating of ERRa by Compound A is specifically acting through ERRa, we used shRNA plasmids to knock-down endogenous ERRa. After four rounds of FACS, excluding ERa (ESR1), all other genes were significantly reduced (P,0.05) in MCF-7/shRNA ERRa3 cells (Fig. 6B), corroborating the effects seen when MCF-7 cells were treated with the ERRa-subtype specific ligand, Compound A, for 24 hours (Fig. 2). Furthermore, treatment of MCF-7 breast cancer cells with Compound A, leads to inhibition of growth; similarly, MCF-7/shRNA ERRa3 cell growth is also slowed when compared to the control MCF-7/shRNA (-) cell line [39].
It has been well demonstrated that targeting ERa with such drugs as tamoxifen and ICI-182,780 (faslodex) has lead to successful therapy [40,41]. Therefore, future studies should include treating ERa (+) cells with combinations of tamoxifen, faslodex or an aromatase inhibitor with Compound A. Additionally, it would be of great interest to treat both ERa (-) and tamoxifen resistant cells with Compound A. In the future it will be interesting to study the effects of knocking-down or antagonizing both ERa and ERRa in breast cancer cells. Intriguingly, Lanvin and coauthors recently reported that the XCT790, an ERRa selective inverse agonist, plus the pure anti-estrogen ICI-182,780 potentiates the ICI-182,780 induced ERa degradation inferring XCT790 may enhance the efficacy of ICI-182,780 in breast cancer treatment. Additionally, XCT790+ICI-182,780 dramatically enhanced ERRa degradation verses ERRa degradation induced by just XCT790 [23].
Based on supporting data in the literature an ERRa specific antagonist shows exciting potential as a novel therapy to treat breast cancer. ERRa is expressed in numerous human breast cancer cell lines, breast tumors, and in breast adipose tissue [6,37,38]. ERRa expression in human breast carcinomas is significantly associated with an increased risk of disease recurrence or poor clinical outcome [38]. It has been reported that ERRa expression is associated with an adverse, aggressive tumor phenotype correlating with ERBB2 (HER2, NEU) overexpression [37]. Additionally, the ERRa ligand DES slows breast cancer cell growth at high concentrations (in vitro) [6] and in the past has been used to treat breast cancer in clinical settings [42]. ERRa antagonists may also be used for ER-negative cancers. BT-20/ shRNA ERRa knock-down cell lines were also established and sorted by FACS. Interestingly, after the second round of GFP enrichment, BT-20 (ER-negative) [43] cells carrying shRNA ERRa plasmid, stopped growing (data not shown).