Estrogen Induces Estrogen-related Receptor α Gene Expression and Chromatin Structural Changes in Estrogen Receptor (ER)-positive and ER-negative Breast Cancer Cells*

Estrogen-related receptor α (ERRα), a member of the nuclear receptor superfamily, is closely related to the estrogen receptors (ERα and ERβ). The ERRα gene is estrogen-responsive in several mouse tissues and cell lines, and a multiple hormone-response element (MHRE) in the promoter is an important regulatory region for estrogen-induced ERRα gene expression. ERRα was recently shown to be a negative prognostic factor for breast cancer survival, with its expression being highest in cancer cells lacking functional ERα. The contribution of ERRα in breast cancer progression remains unknown but may have important clinical implications. In this study, we investigated ERRα gene expression and chromatin structural changes under the influence of 17β-estradiol in both ER-positive MCF-7 and ER-negative SKBR3 breast cancer cells. We mapped the nucleosome positions of the ERRα promoter around the MHRE region and found that the MHRE resides within a single nucleosome. Local chromatin structure of the MHRE exhibited increased restriction enzyme hypersensitivity and enhanced histone H3 and H4 acetylation upon estrogen treatment. Interestingly, estrogen-induced chromatin structural changes could be repressed by estrogen antagonist ICI 182 780 in MCF-7 cells yet were enhanced in SKBR3 cells. We demonstrated, using chromatin immunoprecipitation assays, that 17β-estradiol induces ERRα gene expression in MCF-7 cells through active recruitment of co-activators and release of co-repressors when ERRα and AP1 bind and ERα is tethered to the MHRE. We also found that this estrogen effect requires the MAPK signaling pathway in both cell lines.

Estradiol (E 2 ) 2 is required for the development and function of multiple tissue types and physiological systems in mammals, and it has been implicated in a range of pathological conditions, including the initiation and progression of breast cancer (Refs. 1-3 and references therein). The two genetically distinct nuclear estrogen receptors, ER-␣ and -␤, are thought to mediate most biological effects of E 2 (4). Upon binding of a single ligand molecule, the ERs homodimerize and interact directly with estrogen-response elements (ERE) in the regulatory sequences of estrogen target genes (5). In the absence of an ERE, the ERs may still engage via tethering to transcription factors that are bound to a target gene promoter (6 -8). ER binding ultimately induces gene expression by modifying the local chromatin structure and facilitating the assembly of a transcription complex in either case (9 -11). E 2 also elicits various physiological responses that cannot be totally explained by classic ERmediated actions in the nucleus (see review in 12 and references therein). For example, E 2 reportedly elicits rapid effects via crosstalk with intracellular signal transduction and growth factor signaling pathways (13,14). GPR30, a recently described G-proteincoupled receptor, is reported to mediate ER-independent estrogen signaling under certain conditions (15)(16)(17). Therefore, E 2 clearly acts through a multitude of complex signaling pathways that may or may not involve the nuclear ER forms.
In contrast to the ERs, the estrogen-related receptors (ERR) -␣, -␤, and -␥ all are orphan nuclear receptors with a high degree of sequence identity to the ERs but do not bind estrogens or any other known natural ligands (18,19). ERR␣ is ubiquitous among tissues of the developing embryo and adult (20 -22) and regulates an assortment of genes and physiological processes (23)(24)(25)(26). Like the ERs, ERR␣ binds to EREs but is capable of also binding an estrogen-related response element (ERRE) unique to the ERRs (27)(28)(29) in monomer or dimer form (30). Once bound to DNA, ERR␣ recruits co-regulators similar to those recruited by the ERs and thereby influences gene expression, indicating a considerable degree of functional conservation between the ERs and ERRs (31)(32)(33)(34)(35). In contrast, however, the ERRs are not ligand-dependent but are constitutively active (36 -38) and are able to modulate many ER-regulated physiological pathways and ER target genes (29, 35, 39 -41).
ER␣ is reportedly present in 75% of clinical breast tumor samples and correlates to a more favorable response to endo-crine therapy compared with ER␣-negative cancers (42). Interestingly, recent clinical studies implicate ERR␣ as having a potential role in breast cancer progression (43,44) based largely on its greater prevalence in ER␣-negative tumors and correlation with ErbB2, a known marker of aggressive tumors (45). Hence, positive ERR␣ expression is increasingly associated with an adverse clinical outcome and decreased chance of survival in breast cancer patients (46,47). ERR␣ may promote the growth of breast cancers by increasing local estrogen synthesis as it is reported to stimulate the expression of CYP19A1 (P450 aromatase) (48,49) and SULT2A1 (50), enzymes that contribute to estrogen biosynthesis. Furthermore, we have previously demonstrated that the ERR␣ gene, ESRRA, is an ER-dependent estrogen target in certain estrogen-responsive tissues and cell lines (21,51). At this time, however, the contribution of ERR␣ to hormone resistance and tumor progression in breast cancers remains unknown but may have vital clinical implications.
In eukaryotes, genomic DNA is associated with histone proteins to form nucleosomes, the basic unit of chromatin (52). The structure of the nucleosome, and chromatin as a whole, is inherently prohibitive to gene expression because it precludes the access of transcription factors, such as the ERs and ERRs, to specific regulatory regions (53)(54)(55). Because ERR␣ is a constitutively active protein, transcriptional regulation of the ERR␣ gene may be a primary mechanism for modulating its activity. Hence, the chromatin structure of the ERR␣ gene warrants investigation. In this study, we used the breast cancer cell lines to characterize the nucleosomal arrangement in an area of the ERR␣ gene that is known to harbor an MHRE (51). Our results indicate that E 2 modulates the local chromatin architecture and histone acetylation levels around the MHRE region in both ERpositive (ER ϩ ) and ER-negative (ER Ϫ ) breast cancer cells. Further probing of mechanistic insights has shown that in ER ϩ cells ERR␣ gene expression requires ERR␣ and AP1 binding, along with ER␣ being tethered to the MHRE or a nearby AP1 element, favors recruitment of co-activators and release of corepressors, and ultimately leads to increased ERR␣ gene expression. Although the mechanisms by which E 2 modulates ERR␣ expression in ER Ϫ cells is not yet clear, our studies show that the nucleosome organization of the ERR␣ gene and E 2 -induced chromatin modification is similar to that in the ER ϩ cells. Furthermore, we have shown that the MAPK signaling pathway is required for estrogen action in both cell lines by using a MAPK inhibitor.
Cells Culture and Nuclei Preparation-MCF-7 cells were cultured in modified Eagle's medium. SKBR3 cells were cultured in Dulbecco's modified Eagle's medium. In both cases, the media were supplemented with 10% fetal bovine serum, and the cells were maintained in a humidified incubator at 37°C with 5% CO 2 . At least 2 days prior to all hormonal treatments, the cells were switched to phenol red-free modified Eagle's medium containing 5% charcoal-stripped fetal bovine serum. Treatments consisted of vehicle (ethanol), 10 nM E 2 , 1 M ICI, or 5 M PD98059 or combinations thereof for multiple time points as indicated by the individual experiment. Nuclei were prepared as described previously (58). Briefly, the cells were rinsed and detached from the dish using a rubber policeman in the presence of cold phosphate-buffered saline and pelleted by centrifugation at 750 ϫ g at 4°C for 5 min. The cell pellets were resuspended in homogenization buffer containing 0.1% Nonidet P-40 and incubated on ice for 2 min before being lysed by 5 strokes of pestle A in a Dounce homogenizer. Nuclei were sedimented through a 10% sucrose pad by centrifugation at 1,400 ϫ g for 20 min, washed with washing buffer to remove traces of Nonidet P-40, and collected by centrifugation at 750 ϫ g for 5 min. The nuclei preparations were then kept on ice until use.
Micrococcal Nuclease Analysis-Isolated nuclei were suspended in 200 l of washing buffer supplemented with 1 mM CaCl 2 and then digested with 0 -50 units of MNase for 5 min at 30°C. The reactions were stopped by the addition of 40 1 of 100 mM EDTA, 10 mM EGTA (pH 8.0). Genomic DNA was then purified, digested with the appropriate restriction enzyme (as indicated), and analyzed by Southern blot or reiterative PCR. The first control DNA (C1) was prepared by digestion of purified genomic DNA with 1 unit of MNase/ml for 5 min at 30°C. The second control DNA (C2) was prepared by double digestion of purified genomic DNA with MNase/ml and the appropriate restriction enzyme.
Southern Blotting and Reiterative PCR-For Southern blot analysis, 20 g of MNase-digested DNA per sample was separated on a 1.5% agarose gel, transferred to Hybond Nϩ membrane (GE Healthcare), and probed with 32 P-radiolabeled DNA fragments. Probe A was radiolabeled by PCR ( Fig. 1), and probe B was radiolabeled by random priming (Ready-to-Go beads; GE Healthcare) of the PstI/SphI genomic fragment ( Fig. 1) (51,56). Hybridization was carried out overnight at 65°C, and the blots washed and then exposed to x-ray film at Ϫ70°C.
Reiterative PCR was carried out as follows: 25 g of purified DNA digested with MNase and StuI was analyzed by linear PCR with a 32 P-labeled single strand primer (Table 1) corresponding to the Ϫ521/Ϫ510 region of the ERR␣ promoter (51, 56). The PCR-amplified products and sequencing control (50 ng of ERR␣ plasmid) were separated on 8% polyacrylamide denaturing gel and analyzed using a PhosphorImager (GE Healthcare). Band sizes were measured by using FragmeNT Analysis software (GE Healthcare).
Nuclear Protein Extraction, Cell Lysate Preparation, and Western Blot Analysis-MCF-7 and SKBR3 cells were treated as described above. Nuclear protein extracts were prepared by using the BD TM TransFactor extraction kit according to the supplier's instruction (Clontech) and used for ER␣ and ER␤ detection. Whole cell lysates were prepared for ERR␣ detection. Protein concentrations were determined (Bradford protein assay, Bio-Rad) and were separated by SDS-10% polyacrylamide gel and 4% stacking gel with Bio-Rad Miniprotein apparatus. After electrophoresis, the proteins were blotted onto polyvinylidene difluoride membrane (Invitrogen) and probed with antibodies specific to ER␣ and ERR␣.
RNA Isolation, Reverse Transcription, and Real Time RT-PCR-Total RNA was extracted using the RNeasy mini kit according to the supplier's protocol (Qiagen). For RT-PCR analyses, cDNA was synthesized as described previously (59). Real time RT-PCR was performed for ERR␣ and ␤ 2 -microglobulin gene with Applied Biosystems 7900HT real time PCR System and Taqman Universal PCR Master Mix (Applied Biosystems). Average cycle threshold (C t ) values for ERR␣ were calculated and normalized to C t values for ␤ 2 -microglobulin. Power SYBR Green PCR Master Mix (Applied Biosystems) was used for pS2, c-Myc, and ␤ 2 -actin genes. Average C t values for pS2 or c-Myc were calculated and normalized to C t values for ␤ 2 -actin. All the primer and probe sequences are presented in Table 1.
Restriction Enzyme Hypersensitivity Analysis-Cell nuclei were prepared after experimental treatment and digested in vivo with MslI, ApoI, or ApaI at 30°C for 15 min as described previously (58). The genomic DNA was then purified and digested in vitro with StuI or SphI. The digested DNA fragments were then amplified by reiterative primer extension as described above. The PCR-extended products were purified by phenol/chloroform extraction, ethanol precipitation, and analyzed on 8% denaturing polyacrylamide gels. The bands densities were measured by Kodak 1D program, and the percentage of cleavage was calculated as in vivo band density relative to the total band density.
In Vitro Protein Translation, Nuclear Protein Preparation, and Electrophoresis Mobility Shift Assay (EMSA)-ER␣ and ER␤ were in vitro transcribed and translated using the TNT Coupled Reticulocyte Lysate Systems according to the supplier's instructions (Promega, Madison, WI). Nuclear proteins of the MCF-7 cells were prepared with the TransFactor extraction kit (Clontech). The MHRE fragment was cut from the AAB-CAT reporter (51) by NheI and XhoI digestion, gel-purified, and used as the probe in EMSA. The double-stranded MHRE oligos were labeled with [ 32 P]dGTP by fill-in with Klenow large fragment of DNA polymerase I. The antibodies for supershift experiments for ER␣ were TE111.5D11 from Neo-Marker (Fremont, CA), ER␤, clone 9.88, IgM, from Sigma, and human lactoferrin antibody prepared in our laboratory (60). The EMSA procedure has been described previously (21,29).
Chromatin Immunoprecipitation (ChIP) Assay-Native ChIP was used to evaluate histone modifications in cells following treatments (61). Briefly, isolated nuclei were digested with MNase until the chromatin became mono-to trinucleosome in size. ChIP analyses were carried out by using the ChIP assay kit (Upstate Biotechnologies) with minor modifications. Native chromatin was diluted with ChIP dilution buffer and precleared with 80 l of a salmon sperm DNA-protein A-agarose for 30 min with agitation at 4°C. Immunoprecipitation was performed overnight (10 -14 h) at 4°C with specific antibodies or with rabbit IgG as a control. After immunoprecipitation, 60 l of salmon sperm DNA-protein A-agarose were added for 1 more h at 4°C to capture the immune protein-DNA complexes. The immunoprecipitants were sequentially washed according to instructions, and the immune complexes were eluted with 1% SDS in 0.1 M NaHCO 3 , and the DNA was purified by phenol/ ethanol extraction and precipitation. To study the transcription factor and co-regulator occupancy, standard ChIP was performed in which the chromatin was first cross-linked with 1% formaldehyde at 37°C for 20 min. The crude cell lysates were sonicated to shear the chromatin to 400 -1200-bp size as verified by the agarose gel electrophoresis, and the ChIP assays were performed. After elution, the chromatin was reverse cross-linked with 0.2 M NaCl at 65°C for 6 h and then the DNA was purified. In Re-ChIP experiments, complexes were eluted by incubation with 10 mM dithiothreitol at 37°C for 30 min,

For ChIP assay
Forward 5Ј-GTC AGT GCA GGA CAG CCC GCG TGA G-3Ј Reverse 5Ј-GAT AGG GCC CGG ACG GAG AAA GC-3Ј diluted 1:50ϫ in ChIP dilution buffer, followed by re-immunoprecipitation with the second antibodies (62) or rabbit IgG as a control. DNA product was detected by PCR with specific forward and reverse primer sequences (

RESULTS
The ERR␣ Gene MHRE Is Assembled Over a Single Nucleosome-We demonstrated previously that estrogen stimulation of the ERR␣ expression is mediated through the MHRE located between bp Ϫ709 to Ϫ649 of the ERR␣ promoter (51,63). To study the influence of estrogen on the chromatin dynamics in this portion of the ERR␣ promoter, we had to first determine the nucleosome positions around the MHRE region. Two probes (A and B) were selected to map nucleosome positions between bp Ϫ1205 and Ϫ41 relative to the transcription start site (Fig.  1A). To map nucleosome positions from the 5Ј end of the 1-kb target region, nuclei from MCF-7 cells were digested with increasing concentrations of MNase; the DNA was purified, cut with StuI, blotted, and hybridized with probe A (Fig. 1B). A similar process was used to map the nucleosome position from the 3Ј end but instead cutting with SphI and hybridizing the resulting blot with probe B (Fig. 1C). Two control DNA samples were included, one with MNase digestion only (C1) and the other digested again with the restriction enzyme (C2).
Nuclei digested in vivo produced a ladder when stained with ethidium bromide, each band indicating the presence of a nucleosome (data not shown). As expected,  3-7). Probe B was labeled by random priming of the PstI/SphI fragment and used in Southern blotting and hybridization. The controls and marker are the same as described in B. D, fine mapping of the MHRE nucleosome position by reiterative primer extension as described under "Experimental Procedures." Nuclei (lanes 7-9) were digested with increasing amounts of MNase (0 -30 units/ml). The genomic DNA (lane 6) was digested with 5 units/ml MNase and serves as a control (con). DNA was purified from MNase-digested nuclei, digested again with StuI, and analyzed by reiterative primer extension. Plasmid h55 containing the ERR␣ gene promoter was sequenced at the MHRE region and produced the sequencing tracks as a reference (lanes 1-4). Line graph was created by using Kodak1D software on the band intensity of lane 9.
Southern blots with probe A or probe B detected at least three nucleosome boundaries with an overlapping region around bp Ϫ480 to Ϫ778 (Fig. 1, B and C), indicating that the MHRE region of the ERR␣ promoter is organized into regular nucleosome arrays. The ERR␣ promoter produced comparable results using the same mapping technique in SKBR3, HK2, and HepG2 cell lines (data not shown).
MNase digestion provides only an estimation of nucleosome positions. Therefore, high resolution PCR-based mapping was carried out to better demarcate the MHRE nucleosome position (Fig. 1D). The ERR␣ gene promoter was sequenced to pro-vide a landmark (Fig. 1D, lanes  1-4). Reiterative PCR products identified two clusters of MNase sensitivity (Fig. 1D, lanes 6 -9, arrows) centering on bp Ϫ628 and Ϫ774 (147 bp), suggesting the expected nucleosomal size of ϳ146 bp. These two sites correspond to the nucleosome containing the MHRE and were consistent with the sites identified by Southern blot (Fig. 1, B and C). Based on probe locations and fragment sizes, we concluded that the ERR␣ MHRE is organized into a single nucleosome.
Estrogen Induces ERR␣ Gene Expression in ER-positive MCF-7 and ER-negative SKBR3 Cells-Estrogen stimulation of the ERR␣ in estrogen-responsive tissues and cell lines has been shown previously (21,51). To determine its sensitivity to estrogen stimulation in an ER-negative environment, we compared ERR␣ expression in MCF-7 cells, a well characterized ER-positive breast cancer cell line, and SKBR3 cells, an ER-negative breast cancer cell line reported to possess Gprotein-dependent estrogen signaling (64). The ER expression profile in each cell line was verified by RT-PCR ( Fig. 2A, upper panels) and Western blotting ( Fig. 2A, lower panels). Two well documented estrogen target genes, PS2 and c-MYC, were used as positive controls for estrogen treatments. Each was induced by 10 nM E 2 in MCF-7 cells and exhibited a temporal pattern of induction consistent with published reports. Likewise, the pS2 and c-Myc genes were unresponsive to E 2 in the ER-negative SKBR3 cells (Fig. 2B).
A similar treatment regimen (10 nM E 2 over 24 h) indicated that ERR␣ is induced in a biphasic fashion in MCF-7 cells (Fig. 2C, upper panel). Two distinct peaks in ERR␣ expression, each Ͼ2-fold relative to untreated cells, were observed at 0.5 and 4 h after E 2 treatment, and each was followed by a precipitous return to the basal level.
Interestingly, E 2 induction of ERR␣ in SKBR3 cells exhibited a strikingly different pattern, with a single peak of 2.6-fold at 8 h post-E 2 treatment followed by a precipitous decline thereafter (Fig. 2C, upper panel). Western blotting analyses confirmed that ERR␣ levels were increased in both MCF-7 and SKBR3 cells after estrogen stimulation (Fig. 2C, lower  panel). A dose-response curve indicated that maximum induction of ERR␣ expression in SKBR3 cells occurred at 10 nM E 2 (Fig. 2D), similar to that described previously in MCF-7 cells (51). E 2 Induces MHRE Nucleosome Modification in Both MCF-7 and SKBR3 Cells-We hypothesize that E 2 induction of ERR␣ involves chromatin remodeling around the MHRE, making it more accessible to transcription factors. Such remodeling of the region can be revealed via increased sensitivity to restriction endonucleases digestion, such as those for MslI and BstI which are within the MHRE, and ApaI which is located close to the 3Ј-nucleosome boundary (Fig. 3A). MCF-7 cells were treated with E 2 for 15-120 min, and hypersensitivity at the MslI sites was examined using the oligo D primer (Fig. 3A). In untreated cells, oligo D primer produced a single DNA fragment extending to the StuI site at bp Ϫ1215, indicating that both MslI sites in the MHRE were not accessible. Within 15 min after E 2 exposure, however, both MslI sites exhibited increased sensitivity, as indicated by the production of two small PCR fragments by oligo D (MslI at Ϫ666 and MslI at Ϫ689) (Fig. 3B). As expected, the ApoI site located outside the nucleosome was not affected by E 2 treatment (Fig. 3C). The percent cleavage at each of the MslI sites was comparable after E 2 treatment and exhibited no change from 15 to 120 min of E 2 exposure (Fig. 3, B and C).
Estrogen-induced nuclease hypersensitivity was also detected at the ApaI site when using oligo A in reiterative PCR. The ApaI site lies 3Ј boundary of the nucleosome containing the MHRE (Fig. 3A). Interestingly, this E 2 -induced hypersensitivity could be blunted by the pure anti-estrogen ICI (Fig. 3D). These results demonstrate that estrogeninduced chromatin modifications within the ERR␣ promoter in MCF-7 cells is mediated by ER␣.
Because ERR␣ induction by E 2 in ER-negative SKBR3 cells exhibited a markedly delayed peak relative to MCF-7, our evaluation of structural changes in the chromatin in these cells was equally delayed until 2-6 h after E 2 exposure. Nonetheless, E 2 -induced hypersensitivity around the MslI site was obvious within 2 h after E 2 exposure (Fig. 4A), whereas the ApoI site outside the MHRE nucleosome showed no change (Fig. 4B). Surprisingly, E 2 -induced hypersensitivity around the ApaI site was not blocked by ICI as was observed in the MCF-7 cells but enhanced (Fig. 4C). The ICI-induced increase of hypersensitivity at the ApaI site in SKBR3 cells was consistent with the increased expression of ERR␣ gene after ICI or ICI and E 2 treatment (Fig. 4D). E 2 Induces Histone Acetylation and Co-regulator Exchange in the MHRE Nucleosome-Acetylation of lysine residues in histone tails reduces their affinity for DNA, relaxing the arrangement of chromatin structure and thereby facilitating access of transcription factors and RNA polymerase II (53,55). We therefore chose to evaluate the acetylation of histones H3 and H4 of the MHRE nucleosome using native ChIP assay at various time points after E 2 exposure (Fig. 5A). In MCF-7 cells, histone H3 acetylation occurred within 5 min of E 2 exposure and Cells were either treated with vehicle or 10 nM E 2 for 15-120 min. Isolated nuclei were digested with MslI (in vivo), and the purified DNA was digested with StuI (in vitro) again. The DNA fragments were analyzed by reiterative primer extension with oligo D. C, isolated nuclei from E 2 -treated cells were digested with ApoI (in vivo) and purified DNA was digested with StuI (in vitro). The DNA fragments were analyzed by reiterative primer extension with oligo D. D, ICI blocks E 2 -induced restriction enzyme hypersensitivity. Cells were either treated with vehicle, 10 nM E 2 , or 10 nM E 2 ϩ 1 M ICI for 2 h. Nuclei were digested with ApaI, and purified DNA was digested with SphI again. The DNA fragments were amplified by reiterative primer extension with oligo A. Percent cleavage ϭ in vivo band intensity/(total band densities) from three experiments are presented as means Ϯ S.E. * indicates p Ͻ 0.05 for vehicle versus E 2 -treated cells; con, control.
increased dramatically to a peak within 15 min, followed by a gradual decline to basal levels within 1 h (Fig. 5B). E 2 -induced acetylation of histone H4 followed a similar trend but was less robust (Fig. 5B). Loss of histone acetylation but not total histone at the MHRE was verified using an antibody against the C terminus of the histone H3 (Fig. 5C). These data demonstrate that rapid but transient acetylation of histone H3 and H4 of the MHRE nucleosome precedes active transcription in MCF-7 cells.
We examined whether events occur in the ER Ϫ cells after E 2 treatment consistent with the delayed expression observed earlier in SKBR3 cells (Fig. 5D). Although the ER-positive MCF-7 cells exhibited maximum histone acetylation within 15 min of E 2 exposure (Fig. 5B), histone acetylation at MHRE vicinity in SKBR3 cells was on by 15 min but persisted up to 120 min (Fig.  5D). We also found that RNA polymerase II (pol II) and phosphorylated pol II (p-pol II) were recruited to the MHRE nucleosome in E 2 -treated SKBR3 cells in a temporal pattern consistent with the 120-min peak of histone acetylation (Fig. 5E) which was subsequently followed by an increase in ERR␣ expression.
We also examined co-regulator occupancy at the MHRE nucleosome in MCF-7 cells using a standard ChIP assay, including a formaldehyde fixation step to ensure the detection of loosely bound proteins (Fig. 6A). Our results indicate that p300 is recruited to the MHRE nucleosome within 5 min of E 2 treatment, followed by CBP, the latter exhibiting two distinct periods of peak binding at 15 and 60 min. Interestingly, the co-repressor RIP140 was present on the MHRE nucleosome prior to E2 treatment but was quickly released within 15 min after E2 exposure. The recruitment of p-pol II to the MHRE was trailing behind CBP; it peaked at 30 and 120 min (Fig. 6B). These results are consistent with the current model of cyclical exchange between co-regulators at the chromatin level during estrogen regulation of target genes (62,65).

Estrogen Induces ERR␣ Expression in MCF-7 Cells via an ER␣-mediated Mechanism-The
ERs do not bind the ESRRA23-response element located within the MHRE region (33). We confirmed these findings by EMSA using in vitro transcribed/translated ER␣ and ER␤ (Fig. 7A, lanes 2 and 3). However, nuclear protein extract (NPE) from ER-positive MCF-7 cells clearly contains a component(s) capable of binding the MHRE (Fig. 7A, lanes 4 -9, arrows). When combined with ER␣ or ER␤, the NPE of MCF-7 cells continues to produce a complex with the MHRE; and furthermore, antibodies to ER␣ supershift this complex (Fig. 7A, compare lanes 7-9). Collectively, these data indicate the following: (a) MCF-7 nuclei contain protein(s) capable of binding the MHRE, and (b) this complex recruits ER␣ but not ER␤.
ChIP analyses were performed to further confirm the tethering of ER␣ to the ERR␣ MHRE in MCF-7 cells after E 2 treatment. Repeated experiments indicated that endogenous ER␣ is recruited to the MHRE within 15 min, peaks at 30 min, recedes to base line within 60 min but returns again to maximum levels at 120 min, suggestive of a cyclical binding (Fig. 7B).
ERR␣ and ERR␥ are known MHRE-binding proteins (63, 66) and therefore candidate components of the MHRE binding complex detected in MCF-7 NPEs. The ER␣ could potentially be tethered to the MHRE and co-occupy the region with DNAbound ERRs (29) or, alternatively, may be tethered to AP1 (67)(68)(69) or SP1 (7, 70) complexes bound to their consensus binding sites located upstream as well as within the MHRE nucleosome. To discern among these possibilities, we first used ChIP assays to demonstrate that ERR␣ and AP1 interact with their corresponding binding elements in the nucleosome (Fig. 7C). ERR␣

. Estrogen induces MHRE nucleosome modification in SKBR3 cells.
A, cells were treated with E 2 for 2, 4, or 6 h, and the same MslI restriction enzyme hypersensitive assays were performed as described in Fig. 3B. B, cells were treated with E 2 for 2 h, and the ApoI site was examined as described in Fig. 3C. C, ICI enhances E 2 -induced restriction enzyme hypersensitivity. Cells were treated with vehicle, 10 nM E 2 , 1 M ICI, or 10 nM E 2 and 1 M ICI for 8 h. The restriction enzyme ApaI hypersensitivity was analyzed as described in Fig. 3D. D, ICI induces ERR␣ gene expression. Top panel, Taqman real time PCR. Total RNAs were prepared from SKBR3 cells 8 h after treatment, and the ERR␣ mRNA was measured. Lower panel, Western blotting. Cells were treated with E 2 , ICI, or E 2 and ICI for 24 h, and the whole cell lysate was prepared. The experiments were repeated three times. Data are presented as means Ϯ S.E. ␤-Actin serves as the loading control. * indicates p Ͻ 0.05 for vehicle versus treated cells; con, control.
binding is largely constitutive but clearly increased by E 2 , peaking 45 min after treatment and then gradually decreasing to below base-line levels by 90 min. This suggests a dynamic relationship between ERR␣ and the MHRE in response to E 2 treatment. Interestingly, AP1 is not constitutively bound to the MHRE nucleosome but becomes so within 15 min after E 2 treatment, and cycles in 45-min intervals thereafter for at least 120 min. Furthermore, maximum AP1 recruitment to the MHRE precedes that of ER␣, the latter peaking 30 min after E 2 treatment. Additionally, we have shown by using ChIP and re-ChIP experiments that ER␣ is a co-occupant with both AP1 and ERR␣ on the MHRE nucleosome (Fig. 7, D and E), although its occupation with the latter appears more robust. These data confirm that the ERR␣ is an estrogen-responsive transcription factor in MCF-7 cells, as well as indicates that ER␣ may serve as a transducer for this estrogen action via co-occupancy with DNA-bound ERR␣ and AP1 proteins. E 2 -induced ERR␣ Expression Can Be Blocked by ERK Inhibitor-Estrogen triggers a variety of second messenger signaling events independent of the ERs in addition to the classical ER-mediated gene activation (12). Among the many potential pathways, we first tested an inhibitor of ERK pathway, PD98059, in ER-negative SKBR3 cells. The result showed that the E 2 -induced expression was completely blocked by the ERK inhibitor (Fig. 8A). Similar experiments were performed on the ER-positive MCF7 cells with a different treatment regime that matches the ERR␣ expression. Like with the SKBR3 cells, the ERK inhibitor also blocked the E 2 -induced expression (Fig. 8B). Because ER␣ is functionally regulated via phosphorylation by several protein kinases (71), it is expected that the suppression of the MAPK signaling pathway in ER-positive cells blunts ERmediated estrogen action. These results indicate that estrogen regulation of the ERR␣ gene expression in ER-positive and ERnegative breast cancer cells is measurable and depends on MAPK signaling pathways.

DISCUSSION
There is mounting clinical evidence linking ERR␣ expression to tumor progression and less favorable prognoses in breast cancer patients (43,44,46,47). We have previously demonstrated that estrogens induce ERR␣ expression in MCF-7 cells, an ER␣-positive breast cancer cell line. We have been able to replicate these findings as well as demonstrate estrogen induction of ERR␣ expression in SKBR3 cells, an ER-negative breast cancer cell line. Furthermore, we have demonstrated that estrogen causes structural remodeling of the nucleosome containing Cells were treated with vehicle or 10 nM E 2 for various times. Acetylation status of histone H3 and H4 was determined by native ChIP assay. DNA purified from the immunoprecipitated chromatin was amplified by real time PCR. C, total histone H3 was determined by native ChIP assay. D, time-dependent histone acetylation in SKBR3 cells. Cells were treated with E 2 for 15 and 120 min. The status of histone H3 and H4 acetylation was determined by native ChIP assay. con, control. E, recruitment of RNA pol II and phosphorylated RNA polymerase II (p-pol II) to the MHRE nucleosome in SKBR3 cells. Cells were treated with 10 nM E 2 for various time points, and the standard ChIP assays were performed to detect the presence of pol II and p-pol II on the MHRE nucleosome. The PCR products of the ChIP assays from repeated experiments are presented as means Ϯ S.E.
an MHRE in the ERR␣ gene promoter in both cell lines, as evident by increased hypersensitivity to restriction enzymes and increased histone acetylation. We also demonstrated that estrogen induces ERR␣, AP1, and ER␣ occupancy at the MHRE nucleosome and recruits co-activator and RNA polymerase II to this region in MCF-7 cells, whereas in SKBR3 cells, we showed the recruitment of RNA polymerase II. Although ERR␣ induction requires the MAPK pathway in both cell types, the rapid response to estrogen in the ER␣-positive MCF-7 cells may contribute by the tethering of ER␣ to ERR␣ and AP-1, each bound to their respective response elements.
Chromatin structure is an inherent barrier that must be overcome to allow transcription and gene expression. Hormoneresponse elements in the promoters of steroid hormone target genes are often distinctively positioned in the nucleosome in such a way that they are not easily accessible by transcription factors and transcription initiation complex (72,73). In the presence of hormone, however, activated receptors are bound to the nucleosome, which in turn recruits co-regulators that act to further modify the core histone tails and produce chromatin structure that is more permissible. These events allow the recruitment of additional transcription factors and the generation of a transcription initiation complex.
The promoter of the ubiquitously expressed ERR␣ gene is no exception. It is organized such that a functional MHRE lies within a single nucleosome, which also harbors consensus AP1 (6) and SP1 (74,75) sites just downstream. This arrangement suggests the two regulatory elements may cooperate to bring about estrogen stimulation of the ERR␣ gene. Indeed, we have demonstrated that ERR␣ and AP1 are integrally involved in ER␣-mediated estrogen induction of the ERR␣ expression in MCF-7 cells. We believe that SP1 also participates in estrogenstimulated activity of the ERR␣ gene. By using EMSA, ChIP, and re-ChIP approaches, we have shown that ER␣ is present in the protein complex assembled on the MHRE nucleosome region of the ERR␣ gene in MCF-7 cells in response to estrogen. Recruitment of ER␣ to the MHRE nucleosome after E 2 treatment was cyclical but exhibited a temporal pattern such that peak binding occurred at 120 min, coinciding with maximum AP1 occupancy, which then led to increased acetylation of histones H3 and H4, recruitment of p300, and the release of the constitutively bound co-repressor RIP140. Therefore, ERR␣ binding to the MHRE in its own promoter is a decisive step toward inducing the overall dynamic exchanges necessary to allow for its expression in ER-positive breast cancer cells.
The organization of the ERR␣ MHRE nucleosome may largely contribute to estrogen responsiveness because it contains three distinct but potentially estrogen-influenced regulatory elements, an MHRE, AP1, and SP1. Modification of any one of these elements through ERR␣ or even alternative signaling pathways could initiate a chain reaction of chromatin modification and co-regulator recruitment, thus beginning activation of transcriptional activity. We have produced evidence that estrogen clearly induces a more "open" chromatin configuration around the MHRE, as illustrated by increased restriction enzyme hypersensitivity assay in both MCF-7 and SKBR3 cells. Interestingly, the two MslI sites within the MHRE appear less accessible than the ApaI site located just outside the MHRE but still within the nucleosome. This difference could be an artifact of differential activities between the two restriction enzymes, or FIGURE 6. Estrogen induces co-regulator dynamics and recruitment of RNA polymerase II in the MHRE nucleosome in MCF7 cells. Cells were treated with vehicle or 10 nM E 2 for 5-60 min, and the standard ChIP assay was performed. A, co-regulator dynamics. Specific antibodies to co-activators p300 and CBP and co-repressor RIP140 were used in the study. B, RNA polymerase II recruitment. Specific antibody to phosphorylated polymerase II was used. DNA was amplified by PCR, and the relative band intensity of repeated experiments was determined by using the Kodak 1D software. Data are presented as means Ϯ S.E. possibly because of ERR␣ masking the MslI sites when bound to the MHRE.
Although estrogen induces measurable chromatin remodeling, independent of the ER status of the cell type, the mechanisms involved may differ. In MCF-7 cells, estrogen-induced ERR␣ expression occurred in two distinct waves. Our results indicate that the first wave of estrogen induction follows a current model of ER-mediated transcriptional activation where ER␣ indirectly interacts with a complex bound to the MHRE. Whether this or another mechanism is responsible for the second wave of induction is unclear. Estrogen induction of ERR␣ FIGURE 7. ER␣ tethered to MHRE nucleosome in MCF-7 cells. A, EMSA. The in vitro transcribed and translated ER␣ and ER␤ and nuclear protein extract (NPE) from MCF-7 cells were incubated with labeled MHRE oligos at room temperature individually or in combination. The protein complexes were detected by EMSA on a nondenaturing gel. Antibodies (Ab) used in supershift experiments were ER␣, ER␤ (commercial), and human lactoferrin (83). Retarded protein complexes are indicated by arrows, and the supershifted band is marked (SS). B, E 2 -induced ER␣ occupancy on the MHRE nucleosome. Cells were treated with 10 nM E 2 for 15-120 min. The presence of ER␣ on the chromatin was immunoprecipitated with the ER␣ antibody in ChIP assay. DNA products were determined by real time PCR, and a representative PCR gel picture is shown on top. C, E 2 induces AP1 and ERR␣ recruitment to the MHRE nucleosome. Loading of ER␣, AP1, and ERR␣ on the MHRE nucleosome were determined by standard ChIP assay. The PCR product was scanned and presented as the relative density from two experiments. The data are presented as means Ϯ S.E. D, co-occupancy of ER␣ with AP1 on the MHRE. Cells were treated with 10 nM E 2 for 30 min, and the presence of AP1 and ER␣ on the MHRE nucleosome was detected by standard ChIP assay (upper panel). The presence of ER␣ in the immunoprecipitated AP1 complexes is shown by re-ChIP assay (lower panel). E, co-occupancy of ER␣ with ERR␣ on the MHRE. Same experiment as in D, but an ERR␣ antibody was used in the first ChIP assay. in ER␣-negative SKBR3 cells was much delayed, as were the expected changes in chromatin structure, histone acetylation, and protein recruitment. Interestingly, ICI did not block E 2induced transcriptional activity in the ER-negative breast cancer cells but rather enhanced it (Fig. 4D). This result was supported by the restriction enzyme hypersensitivity assay of the MHRE nucleosome treated with both E 2 and ICI (Fig. 4C). These findings establish the estrogen responsiveness of ERR␣ gene in an ER-negative environment.
The G-protein-coupled receptor, GPR30, has been recently implicated in mediating E 2 actions in certain cell-specific contexts (76). For example, phytoestrogens such as genistein and quercetin stimulate the estrogen-responsive c-FOS gene in ERnegative SKBR3 cells via GPR30 (77). Furthermore, the selective estrogen receptor modulators tamoxifen and ICI are reported to function as agonists for GPR30 in SKBR3 cells (76). Although the potential role of GPR30 in the E 2 -induced ERR␣ expression in SKBR3 cells is not known, we certainly recognize it as a candidate. We have demonstrated that E 2 induction of ERR␣ relies on the MAPK signaling pathway as it is susceptible to a MAPK inhibitor. Stimulation of the MAPK pathway by estrogen can lead to the activation of ERR␣ via phosphorylation (22,78). Phosphorylated ERR␣ binds DNA and interacts with co-activators more efficiently. Therefore, it is possible that ERR␣ enhances its own expression through autoregulation (30,33).
We are only beginning to understand the clinical importance of ERR␣ expression in breast cancer. The association of ERR␣ expression in breast cancer with a poor clinical outcome suggests it plays a role in tumor progression and aggressiveness (43,44,46). Evidence of similar correlates that appear in cancers of the prostate (79), ovary (80), colon (81), and endometrium (82) continue to emerge. Because its function is generally constitutive, ERR␣ could potentially induce ER␣ target genes in hormonally responsive cancers in an estrogen-independent manner. In addition, ERR␣ and ER␣ may antagonize each other's function on certain promoters (40,57). The correlation between overexpression of ERR␣ and CYP19A1 (49), a downstream target of ERR␣, in breast cancer cells suggests that ERR␣ may contribute to increased local production of estrogens in mammary glands. Likewise, ERR␣ reportedly increases SULT2A1 expression and therefore may increase the pool of aromatase substrates (50). These findings suggest a potential feed forward loop in which local production of estrogen increases ERR␣ expression in aggressive breast tumors.