Estrogen-Related Receptor A1 Transcriptional Activities Are Regulated in Part via the ErbB2/HER2 Signaling Pathway

We previously showed that (a) estrogen-related receptor A1 (ERRA1) down-modulates estrogen receptor (ER)–stimulated transcription in low ErbB2–expressing MCF-7 mammary carcinoma cells, and (b) ERRA and ErbB2 mRNA levels positively correlate in clinical breast tumors. We show here that ERRA1 represses ERA-mediated activation in MCF-7 cells because it failed to recruit the coactivator glucocorticoid receptor interacting protein 1 (GRIP1) when bound to an estrogen response element. In contrast, ERRA1 activated estrogen response element– and ERR response element–mediated transcription in ERA-positive, high ErbB2–expressing BT-474 mammary carcinoma cells, activation that was enhanced by overexpression of GRIP1. Likewise, regulation of the endogenous genes pS2, progesterone receptor, and ErbB2 by ERRA1 reflected the cell type–specific differences observed with our reporter plasmids. Importantly, overexpression of activated ErbB2 in MCF-7 cells led to transcriptional activation, rather than repression, by ERRA1. Two-dimensional PAGE of radiophosphate-labeled ERRA1 indicated that it was hyperphosphorylated in BT-474 relative to MCF-7 cells; incubation of these cells with anti-ErbB2 antibody led to reduction in the extent of ERRA1 phosphorylation. Additionally, mitogen-activated protein kinases (MAPK) and Akts, components of the ErbB2 pathway, phosphorylated ERRA1 in vitro . ERRA1-activated transcription in BT-474 cells was inhibited by disruption of ErbB2/epidermal growth factor receptor signaling with trastuzumab or gefitinib or inactivation of downstream components of this signaling, MAPK kinase/MAPK, and phosphatidylinositol-3-OH kinase/ Akt, with U0126 or LY294002, respectively. Thus, ERRA1 activities are regulated, in part, via ErbB2 signaling, with ERRA1 likely positively feedback-regulating ErbB2 expression. Taken together, we conclude that ERRA1 phosphorylation status shows potential as a biomarker of clinical course and antihormonaland ErbB2-based treatment options, with ERRA1 serving as a novel target for drug development. (Mol Cancer Res 2007;5(1):71–85) Introduction The steroid nuclear receptor estrogen receptor a (ERa), officially termed NR3A1 (1), is pivotally involved in the etiology of breast cancer. ERa mediates the effects of estrogens on transcription and is expressed at high levels in approximately three fourths of human breast tumors. It thereby serves as a critical biomarker of clinical course and target for therapy (reviewed in ref. 2). The orphan nuclear receptors estrogenrelated receptor a (ERRa; NR3B1), ERRh (NR3B2), and ERRg (NR3B3; ref. 1) exhibit a high degree of sequence similarity with ERa (reviewed in ref. 3). They do not bind naturally occurring estrogens, but share other biochemical activities with ERs, including binding to estrogen response elements (ERE; refs. 4-8). ERRs also bind to extended nuclear receptor half-site sequences resembling 5¶-TNAAGGTCA-3¶, referred to as ERR response elements (ERRE; refs. 5-7, 9, 10). ERRa mRNA levels are similar or greater than ERa mRNA levels in approximately one fourth of unselected human breast cancers, with the highest levels occurring in tumors lacking functional ERa (11). Additionally, ERRa mRNA levels correlate in breast cancers with those of ErbB2 (also called HER2/neu; ref. 11), a marker of tumor aggressiveness. Suzuki et al. (12) reported that f55% of human breast cancers are positive for ERRa by immunohistochemistry, with ERRapositive status being associated with greatly increased risk of recurrence and adverse clinical outcome. Thus, ERRa shows potential as a prognosticator and target for breast cancer therapy, with ERRa possibly playing an important role by substituting for ERa activities, especially in ErbB2-positive, ERa-negative, and tamoxifen-resistant tumors. Whereas ERa usually regulates gene expression in a ligandinducible manner, ERRa1, the 423-amino-acid major isoform encoded by the ESRRA gene (National Center for Received 7/26/06; revised 10/23/06; accepted 10/31/06. Grant support: USPHS grants P30-CA14520 (University of Wisconsin Comprehensive Cancer Center), P01-CA22443 (J.E. Mertz), T32-CA09135 (M.L. Farrell), T32-CA09681 (E.A. Ariazi), and CA89018 (Northwestern University Specialized Programs of Research Excellence in Breast Cancer; V.C. Jordan); U.S. Army Medical Research and Materiel Command grants DAMD1799-1-9452 (E.A. Ariazi), DAMD17-03-1-0347 (J.E. Mertz), W81XWH-05-10243 (J.E. Mertz), W81XWH-06-1-0500 (J.E. Mertz); and an Eli Lilly Fellowship (R.H. Lurie Comprehensive Cancer Center of Northwestern University). 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. Note: Current address for E.A. Ariazi: Fox Chase Cancer Center, Philadelphia, Pennsylvania. Dedicated in memory of Professor Jack Gorski. Requests for reprints: Janet E. Mertz, McArdle Laboratory for Cancer Research, University of Wisconsin-Madison, 1400 University Avenue, Madison, WI 53706. Phone: 608-262-2383; Fax: 608-262-2824. E-mail: [email protected] Copyright D 2007 American Association for Cancer Research. doi:10.1158/1541-7786.MCR-06-0227 Mol Cancer Res 2007;5(1). January 2007 71 on June 19, 2017. © 2007 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from Biotechnology Information accession NP_004442.3; refs. 4, 5), can constitutively activate transcription in the absence of ligand. ERRa1 interacts with peroxisome proliferator-activated receptor coactivator-1a (13, 14) and the p160 family of coactivators, including glucocorticoid receptor interacting protein 1 (GRIP1/SRC2; ref. 15), via a carboxyl-terminal coactivator-binding inverted LxLxxL motif (16). Bulky amino acid side chains almost completely fill the ERRa1 putative ligand-binding pocket (14, 17), with residue Phe recapitulating interactions analogous to ones provided by ligands, thereby promoting binding of coactivators (17). ERRa1 has been shown to bind the promoter of the estrogen-inducible pS2 gene (also called TFF1; ref. 18) and to modulate transcription of the estrogen precursor metabolizing genes aromatase (CYP19; ref. 19) and DHEA sulfotransferase (SULT2A1 ; ref. 20). It also modulates expression of the estrogen-responsive genes lactoferrin (4), osteopontin (21), and even ERRa itself (also called ESRRA ; refs. 22, 23). The effect of ERRa1 binding to a transcriptional response element can be either negative or positive depending on the specific cell type (8) and promoter context (24). For example, ERRa1 downmodulates E2-induced transcription in ERa-positive human mammary carcinoma MCF-7 cells by an active mechanism (8), yet activates gene transcription in ERa-negative human mammary carcinoma SK-BR-3 cells (19) and a variety of other cell lines, including human cervical carcinoma HeLa cells (8), human endometrial RL95-2 cells (4), human embryonic kidney 293 cells (21), and rat ROS 17/2.8 osteosarcoma cells (21). The factors determining whether ERRa1 activates or downmodulates transcription have yet to be fully identified. Epidermal growth factor receptor (EGFR) and ErbB2, members of the ErbB family of transmembrane receptor tyrosine kinases, signal in part through the MAPK and phosphatidylinositol-3OH kinase (PI3K)/Akt signaling pathways (25). Stimulation of these pathways can lead to activation of unliganded ERa (26), with overexpression of EGFR and ErbB2 implicated in the failure of antiestrogen therapy in both model systems (27-29) and clinical breast cancers (30-32). Thus, by analogy with ERa, we hypothesized that signaling via ErbB2 leads to phosphorylation of ERRa1, thereby modulating its activities. Findings in support of this hypothesis include the following: (a) ERRa1 can exist as a phosphoprotein (9); (b) human breast tumors that express high lev els of ErbB2 mRNA also frequently express high levels of ERRa mRNA (11); and (c) SK-BR-3 cells, in which ERRa1 functions as a constitutive activator (19), contain 2 orders of magnitude more ErbB2 mRNA than MCF-7 cells (33) in which it functions as a down-modulator of transcription (8). To test the validity of this hypothesis, we examined the effects of ErbB2 signaling on the transcriptional activity and phosphorylated state of ERRa1 in MCF-7 versus BT-474 cells, another mammary carcinoma cell line with very high ErbB2 levels (33). We found that overexpression of ERRa1 led to increased accumulation of pS2, progesterone receptor (PgR), and ErbB2 mRNAs in BT-474 cells and decreased accumulation of pS2 and ErbB2 mRNAs in MCF-7 cells. ERRa1 was hyperphosphorylated in BT-474 cells compared with MCF-7 cells and could be phosphorylated by MAPKs and Akts in vitro . Strikingly, ERRa1 transcriptional activity was stimulated by overexpression of activated ErbB2 in MCF-7 cells and inhibited in BT-474 cells treated with (a) the ErbB2 inhibitor trastuzumab (Herceptin; ref. 34), (b) the EGFR tyrosine kinase inhibitor gefitinib (Iressa, ZD1839; ref. 35), (c) the MAPK kinase (MEK) inhibitor U0126 (36), or (d) the PI3K inhibitor LY294002 (37). Thus, we conclude that ErbB2 signaling can modulate ERRa1 activities. Results Effects of E2 on Expression of ERRa in MCF-7 and BT-474 Cells ERRa1 can either repress or activate ERE-regulated expression (8), a target sequence to which it binds exclusively as a homodimer (38). Given our finding that ERRa mRNA levels positively correlate with those of ErbB2 and ErbB3 (11), we speculated that posttranslational modifications mediated by the ErbB2-directed pathway might contribute to regulation of ERRa1 activities. We studied here two ERa-positive breast cancer cell lines, MCF-7 and BT474, known to express low and high levels of ErbB2, respectively (39), to examine the effects of ErbB2 on ERRa1 activities. First, we measured endogenous ERRa1 and ERa protein levels and the effects of estrogen on these levels in these two cell lines. Lysates were prepared from MCF-7 and BT-474 cells cultured in estrogen-free medium and treated for 24 h with 100 pmol/L 17h-estradiol (E2), its vehicle ethanol as a control, 1 nmol/L E2, or 1 nmol/L E2 plus 1 Amol/L of the complete antiestrogen fulvestrant. The proteins in the lysates were separated by SDS-PAGE, blotted to a filter, and probed with antibodies specific for ERa, ERRa, and h-actin as an internal control (Fig. 1). As expected, ERa levels were similar in the two ERa-positive cell lines (Fig. 1, lane 1 versus lane 5), down-regulated following treatment with E2 (Fig. 1, lanes 2 and 3 versus lane 1 and lanes 6 and 7 versus lane 5), and further down-regulated in the presence of fulvestrant (Fig. 1, lane 4 versus lane 3 and lane 8 versus lane 7), a drug known to promote proteasome-mediated degradation of ERa (40). On the FIGURE 1. Immunoblot analysis showing endogenous expression of ERa and ERRa1 in MCF-7 and BT474 cells. Cells were cultured for 24 h in estrogen-free medium supplemented with ethanol, 100 pmol/L E2, 1 nmol/L E2, or 1 nmol/L E2 plus 1 Amol/L fulvestrant. Endogenous ERa, ERRa1, and h-actin were detected using primary antibodies specific for these proteins followed by IR fluorescent dye–conjugated secondary antibodies. ERa and ERRa1 protein levels normalized to h-actin are shown as units relative to the ERa level in the control-treated MCF-7 cells (lane 1). Ariazi et al. Mol Cancer Res 2007;5(1). January 2007 72 on June 19, 2017. © 2007 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from other hand, ERRa1 levels were not significantly affected by the presence of either of these ligands of ERa, except at the higher concentration of E2 in BT-474 cells (Fig. 1). Furthermore, the ERa/ERRa protein ratios were not significantly different between the MCF-7 and BT-474 cells under either the estrogen-free (3.3 versus 4.0, respectively) or 100 pmol/L E2 (1.5 versus 1.3, respectively) growth conditions. Thus, we can assume in the experiments presented below that differential effects on transcription were due to changes in the activities of ERRa1, not in its levels within the cells. ERRa1 Represses ERE-Regulated Transcription in MCF-7 Cells but Activates Transcription in BT-474 Cells We initially studied ERRa1-regulated transcription using minimal synthetic reporter genes containing (a) a palindromic ERE or an ERRE, (b) a TATA box, and (c) an initiator element rather than complex natural promoters. We did so to ensure observed effects were not due to indirect influences of other factors binding to other regulatory elements such as AP1or Sp1-binding sites. Cells were cotransfected with an ERE(5 )regulated or ERRE(5 )-regulated dual-luciferase reporter gene set, with the Renilla luciferase plasmid serving as an internal control for the firefly luciferase plasmid (Fig. 2A). Concurrently, cells were cotransfected in parallel with a TATAregulated dual-luciferase reporter set (Fig. 2A), which served as an external control for experimental conditions, the physiologic state of the cells, and non–specific effects on the basal transcriptional machinery. The effect and specificity of ERRa1 was evaluated by cotransfecting the cells with a plasmid expressing wild-type ERRa1, mutant ERRa1L413A/L418A, a variant defective in the carboxyl-terminal inverted LxLxxL motif that serves as a coactivator docking site (8), or their parental empty vector. The cells were also cotransfected with a plasmid that expressed GRIP1, a member of the p160 family of coactivators, or its empty parental plasmid. Afterward, the cells were cultured for 40 h in estrogen-free medium supplemented with the indicated compounds, harvested, and assayed for firefly and Renilla luciferase activity. As expected, treatment of the ERa-positive MCF-7 cells with 100 pmol/L E2 induced ERE-regulated transcription f19-fold (Fig. 2B, lane 7 versus lane 1). Introduction of exogenous wildtype ERRa1 or mutant ERRa1L413A/L418A led to a 68% and 71% reduction, respectively, in E2-stimulated transcription in these cells (Fig. 2B, lanes 9 and 11 versus lane 7). The finding that addition of ERRa1L413A/L418A did not lead to complete loss of the E2-stimulated activity indicates that some of the promoter sites probably remained occupied by ERa in these high ERa– expressing cells. Overexpression of the coactivator GRIP1 enhanced E2-stimulated transcription an additional 1.8-fold (Fig. 2B, lane 8 versus lane 7), most likely by stimulating the activity of ERa bound to the EREs. This finding indicates that GRIP1 was limiting in these cells. Nevertheless, overexpression of GRIP1 failed to overcome the down-modulation by ERRa1 of the ERa-stimulated transcription (Fig. 2B, lane 10 versus lane 8). Confirming and extending prior findings (8), we conclude that ERRa1 acted as a repressor in MCF-7 cells, downmodulating ERa-stimulated transcription. Importantly, whereas overexpression of the coactivator GRIP1 enhanced ERa activity, it failed to convert ERRa1 to an activator. To examine ERRa1 activity via EREs in the absence of ERa-stimulated transcription, the cotransfected cells were cultured in estrogen-free medium supplemented with (a) only the drug vehicle, ethanol (Fig. 2B, lanes 1-6), or (b) the complete antiestrogen fulvestrant along with 100 pmol/L E2 (Fig. 2B, lanes 13-18). Under either of these conditions, overexpression of wild-type ERRa1 led to a barely significant increase in transcription (Fig. 2B, lanes 3, 4, 15, 16 versus lanes 1, 2, 13, 14 , respectively). Thus, ERRa1 exhibited a very low level of activator activity in MCF-7 cells when ERa is absent. We likewise examined the ability of ERRa1 to modulate ERE-regulated transcription in BT-474 cells (Fig. 2D). In the absence of ERa-stimulated transcription, overexpression of wild-type ERRa1 led to an f5-fold increase in ERE-regulated transcription (Fig. 2D, lane 3 versus lane 1 and lane 15 versus lane 13). Overexpression of GRIP1 led to an additional 2-fold enhancement in ERRa1-stimulated transcription (Fig. 2D, lane 4 versus lane 3 and lane 16 versus lane 15) as well as a 2-fold enhancement in ERa-stimulated transcription (Fig. 2D, lane 8 versus lane 7). Thus, GRIP1 was limiting in the BT-474 cells. The ERRa1L413A/L418A mutant variant failed to enhance expression (Fig. 2D, lanes 5 , 6, 17 , and 18 versus lanes 1, 2, 13 , and 14 , respectively). Overexpression of wild-type ERRa1 did not significantly alter the level of ERE-regulated transcriptional activity when the BT-474 cells were incubated in the presence of 100 pmol/L E2 (Fig. 2D, lanes 9 and 10 versus lanes 7 and 8). This finding was likely due to ERRa1 simply substituting for ERa as another activator of transcription when it displaced ERa for binding the EREs. By contrast, the ERRa1L413A/L418A mutant led instead to a 50% to 60% reduction in E2-stimulated transcription (Fig. 2D, lanes 11 and 12 versus lanes 7 and 8). This reduction in expression was similar to the one observed in E2-stimulated ERE-regulated transcription by wild-type ERRa1 in MCF-7 cells. Hence, a mutant defective in docking coactivators mimicked in BT-474 cells the repressor activity of wild-type ERRa1 seen in MCF-7 cells. Thus, in contrast to the results observed in MCF-7 cells, ERRa1 activated ERE-regulated transcription in BT-474 cells, likely doing so in part via GRIP1 interaction with its carboxylterminal coactivator binding motif. ERRa1 also regulates transcription via ERREs, including the sequence 5¶-TCAAGGTCA-3¶. In sharp contrast to the effects observed on ERE-regulated expression, ERRE-regulated expression in MCF-7 cells was only slightly affected by either incubation with E2 (Fig. 2C, lanes 7 and 8 versus lanes 1 and 2) or overexpression of wild-type ERRa1 (Fig. 2C). Hence, neither ERa nor ERRa1 significantly affect ERRE-regulated transcription in MCF-7 cells. ERRE-regulated expression was also unaffected by incubation with 100 pmol/L E2 in BT-474 cells (Fig. 2E, lanes 7 and 8 versus lanes 1 and 2). However, independent of E2, ERRa1 activated ERRE-regulated transcription f2.7to 3-fold and 5.5to 6-fold in the absence and presence of GRIP1, respectively (Fig. 2E, lanes 3 and 4 versus lanes 1 and 2; lanes 9 and 10 versus lanes 7 and 8; lanes 15 and 16 versus lanes 13 and 14). Again, the ERRa1L413A/L418A mutant failed to induce transcription (Fig. 2E, lanes 5 and 6 versus lanes 1 and 2; lanes 11 and 12 versus lanes 7 and 8; lanes 17 and 18 ErbB2 Signaling Regulates ERRa1 Activity Mol Cancer Res 2007;5(1). January 2007 73 on June 19, 2017. © 2007 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from FIGURE 2. Differential transcriptional activity of ERRa1 in low ErbB2–expressing MCF-7 cells versus high ErbB2–expressing BT-474 cells. A. Reporter gene sets used in this study. B to E. MCF-7 and BT-474 cells were cotransfected with the ERE(5 ) or ERRE(5 )-regulated dual luciferase reporter gene sets along with ERRa1, ERRa1L413A/L418A, GRIP1, or empty parental plasmids as indicated. As an external normalization control, cells were also cotransfected in parallel for each condition indicated with the TATA-regulated dual-luciferase reporter set in place of the ERE(5 ) or ERRE(5 ) reporter sets. Cells were incubated for 40 h in estrogen-free medium supplemented with ethanol (EtOH ), 100 pmol/L E2, or 100 pmol/L E2 plus 100 nmol/L fulvestrant as indicated. Columns, mean of samples processed in triplicate; bars, SE. Data are presented relative to the luciferase activity present in the cells assayed in lane 1 of each figure. Hatched columns, cells cotransfected with the GRIP1 expression plasmid; solid columns, cells cotransfected with its empty parental plasmid. Ariazi et al. Mol Cancer Res 2007;5(1). January 2007 74 on June 19, 2017. © 2007 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from versus lanes 13 and 14), indicating dependence of ERRa1 activation via ERREs on coactivator docking. Thus, ERRa1 activated both EREand ERRE-regulated transcription in BT-474 cells. Vanacker et al. (7) previously reported that ERa can efficiently bind to and activate transcription via an ERRE. We observed here that E2 very weakly induced ERRE-regulated transcription. To address this discrepancy, we examined the responsiveness to E2 of the EREand ERRE-regulated reporter gene sets studied above in parallel with a previously described ERRa1-responsive reporter, pSFRE 3-Luc, which contains three copies of the same core extended ERE half-site driving an SV40 minimal early promoter (Fig. 2A; ref. 7). As expected, the ERE(5 )-regulated reporter efficiently responded to E2 in a concentration-dependent manner, with E2 maximally inducing ERE-regulated activity 14-fold in MCF-7 (Fig. 3A) and 12-fold in BT-474 cells (Fig. 3B). However, supraphysiologic concentrations of E2 up to 1 Amol/L maximally induced the ERRE (5 )-regulated reporter set by 2.1-fold and the SFRE 3regulated reporter set by 1.5-fold in MCF-7 cells (Fig. 3A). Similarly, the maximum level of ERRE(5 )and SFRE 3regulated reporter activity was 1.6-fold in BT-474 cells (Fig. 3B). Hence, ERREs exhibited only minimal responses to E2-stimulated ERa, effects that could have been indirect. To understand the reason for the lack of responsiveness of the ERRE/SFRE–regulated promoters to E2, we also examined whether ERa could bind to this ERRE sequence. Competition electrophoretic mobility shift assays were done using a radiolabeled, double-stranded ERE-containing oligodeoxynucleotide as probe; whole-cell extracts obtained from COS cells containing overexpressed ERa or ERRa1 as protein source; and unlabeled, double-stranded oligodeoxynucleotide containing an ERE, mutant ERE, or ERRE sequence as competitor (Fig. 3C). As expected, ERa efficiently bound the ERE (Fig. 3C, lanes 5-7 versus lane 4), but not the mutant ERE (Fig. 3C, lanes 8-10). Contrary to a prior report (7), ERa also failed to significantly bind the ERRE (Fig. 3C, lanes 11-13 versus lane 4). On the other hand, ERRa1 efficiently bound both the ERE (Fig. 3C, lanes 16-18 , versus lane 15) and the ERRE (Fig. 3C, lanes 22-24), but not the mutant ERE (Fig. 3C, lanes 19-21). Thus, we conclude that ERa does not significantly bind to the extended half-site ERRE consensus sequence 5¶-TCAAGGTCA-3¶, whereas ERRa1 can efficiently bind both the consensus sequence and at least some EREs. Differential Regulation of Endogenous Cellular pS2, PgR, and ErbB2 Genes by ERRa1 in MCF-7 versus BT-474 Cells Does ERRa1 also differentially regulate expression of endogenous cellular genes in a cell type–dependent manner? To begin to answer this question, we examined the promoter regions of cellular genes implicated in breast cancer for potential ERREs by searching a eukaryotic promoter database (41) and Genbank. The binding affinities of ERRa1 for these putative sites relative to a consensus ERRE were determined by semiquantitative competition EMSAs done with the consensus ERRE-containing double-stranded oligonucleotide serving as the radiolabeled probe DNA. ERRa1 bound these sites with a variety of affinities (Table 1). Interestingly, ERRa1 bound the PgR site 1 with a higher relative binding affinity (RBA, 1.85) than the reference ERRE although they contain the same ERRE extended half-site sequence. Thus, the precise context of an ERRE can modulate ERRa1 binding affinity for it. ERRa1 also bound quite well to the ErbB2 (RBA, 1.08), PgR site 2 (RBA, 0.90), and pS2 site 2 (RBA, 0.50) sequences. To examine the effects of ERRa1 on expression of the endogenous cellular genes pS2, PgR , and ErbB2 in MCF-7 and BT-474 cells, cells were cotransfected in parallel with plasmids encoding enhanced green fluorescent protein (EGFP) and either wild-type ERRa1 or its empty parental vector. After incubation for 48 h in estrogen-free medium to avoid effects due to ERa, EGFP-positive cells were isolated by fluorescence-activated cell sorting. RNA was purified from these EGFP-positive cells and assayed by quantitative real-time PCR for amounts of pS2, PgR, ErbB2, and ERRa mRNA relative to cellular 18S rRNA as an internal control (Fig. 4). Consistent with this protocol having worked successfully, ERRa mRNA levels were found to be 50to 67-fold higher in the cells (isolated by fluorescence-activated cell sorting) transfected with the ERRa1 expression plasmid compared with the ones transfected with the empty parental plasmid (data not shown). Whereas overexpression of ERRa1 in MCF-7 cells led to a modest decrease or no change in expression of these three genes (Fig. 4A-C, lane 2 versus lane 1), it led to a 2to 11-fold increase in their expression in BT-474 cells (Fig. 4A-C, lane 4 versus lane 3). Furthermore, with the basal level of ErbB2 mRNA already 5-fold higher in BT-474 cells than in MCF-7 cells (Fig. 4C, lane 3 versus lane 1), the resulting differential expression of ErbB2 increased to a highly significant 30-fold when ERRa1 was overexpressed (Fig. 4C, lane 4 versus lane 2). Thus, we conclude that ERRa1 differentially regulates endogenous target genes as well as synthetic reporter ones in a cell type–dependent manner. Extent of Phosphorylation of ERRa1 Correlates with its Ability to Activate Transcription To begin to determine the mechanism(s) of ERRa1 cell type–dependent activity, we examined ERRa1 phosphorylation status. MCF-7 and BT-474 cells were incubated with [P]Pi; ERRa1 was immunoprecipitated from protein extracts prepared from these cells; and its phosphorylated isoforms were resolved by two-dimensional PAGE. Although BT-474 cells were found to contain predominantly one highly phosphorylated isoform of ERRa1, MCF-7 cells contained several differentially phosphorylated isoforms of ERRa1 (Fig. 5A). Given the P label was roughly equally distributed among three isoforms of ERRa1 in the MCF-7 cells, the percentage of ERRa1 by moles in the most highly phosphorylated isoform in these cells was at most 15%. Thus, BT-474 cells contained a much larger percentage of their ERRa1 in a highly phosphorylated isoform than did the MCF-7 cells. The monoclonal antibody (mAb) 4D5 is the murine precursor of the humanized antibody trastuzumab. They share the same epitope-reacting regions, disrupting the ErbB2 3 http://www.epd.isb-sib.ch/. 4 http://www.ncbi.nlm.nih.gov/. ErbB2 Signaling Regulates ERRa1 Activity Mol Cancer Res 2007;5(1). January 2007 75 on June 19, 2017. © 2007 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from signaling pathway without affecting the overall amount of ERRa1 per cell (Fig. 7). Incubation of BT-474 cells with antibody 4D5 led to a significant reduction in the extent of ERRa1 phosphorylation, with the appearance of several phosphorylated isoforms of ERRa1 in a pattern somewhat similar to the one observed with the MCF-7 cells (Fig. 5A). The lower amount of phospo-labeled ERRa1 observed in the 4D5treated cells was due to this treatment being inhibitory to cell growth (data not shown). Thus, we confirmed prior reports that ERRa1 is a phosphoprotein (9, 38). We also conclude that ERRa1 was phosphorylated in vivo at several sites, with the extent of phosphorylation being cell type dependent and reduced by disruption of the ErbB2 signaling pathway. Importantly, the extent of phosphorylation correlated with the transcriptional activity of ERRa1: The partially phosphorylated isoforms present in MCF-7 cells likely functioned as repressors, whereas the highly phosphorylated isoform(s) present in BT474 cells probably functioned as activators. FIGURE 3. ERa activity on a palindromic ERE sequence compared with an extended half-site ERRE sequence. MCF-7 (A) and BT-474 (B) cells were cotransfected as described in Fig. 2 with the indicated dual-luciferase reporter gene sets and cultured in estrogen-free medium supplemented with ethanol. Concentrations of E2 range from 1 pmol/L to 1 Amol/L, or 100 pmol/L E2 plus 100 nmol/L fulvestrant as indicated. Cells were harvested 40 h later and assayed for luciferase activity, with normalization to both the internal and external reporter genes. Points, means of samples processed in triplicate; bars, SE. C. EMSAs showing ERRa1, but not ERa, binds the ERRE as well as the ERE with high affinity. Competition EMSAs were done with whole-cell extracts of COS cells transfected with plasmids expressing ERa or ERRa1 serving as protein source, a radiolabeled double-stranded oligonucleotide containing an ERE (5¶-taagcttAGGTCAcagTGACCTaagctta-3¶) serving as probe, and unlabeled double-stranded oligonucleotides corresponding to the sequences indicated below the gel serving as competitors. Capitalized letters within boxes, ERE and ERRE extended half-site sequences. White letters on black, mutations. Ariazi et al. Mol Cancer Res 2007;5(1). January 2007 76 on June 19, 2017. © 2007 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from Activated MAPKs and Akts Phosphorylate ERRa1 In vitro We next tested whether ERRa1 can serve in vitro as a substrate of MAPKs and Akts, downstream kinases in the ErbB2 signaling pathway. We incubated equal amounts of Escherichia coli –produced, carboxyl-terminal 6 His-tagged ERRa1 (ERRa1-His) with activated MAPK1, MAPK2, Akt1, or Akt2 in the presence of [g-P]ATP and resolved the resulting phosphorylated products by 4% to 12% gradient SDSPAGE. Myelin basic protein was included in each reaction as an internal control. Each of these four kinases phosphorylated ERRa1 in vitro (Fig. 5B). Interestingly, differences were observed in the mobilities of the phosphorylated proteins, consistent with multiple sites on ERRa1 being phosphorylated by the MAPKs (Fig. 5B, lanes 1 and 2) and fewer sites being phosphorylated by the Akts (Fig. 5B, lanes 3 and 4). To begin to localize sites of phosphorylation by 1-423 MAPK2, full-length glutathione S-transferase (GST)-ERRa11-423 and truncated variants of it were synthesized in and purified from E. coli . Equimolar amounts of each protein were incubated with activated p42 MAPK and [gP]ATP and resolved by 12% SDS-PAGE (Fig. 5C). Phosphorylated heatand acid-stable protein regulated by insulin (PHAS-I) and GST-h-globin1-123 FIGURE 4. Effects of ERRa1 overexpression in MCF-7 (black columns) and BT-474 cells (hatched columns ) on expression of the endogenous cellular genes encoding pS2 (A), PgR (B), and ErbB2 (C) mRNA. Cells were cotransfected with pEGFP and the ERRa1 expression plasmid or its empty parent plasmid, pcDNA3.1 and incubated under estrogen-free conditions. Twenty-four hours later, EGFP-positive cells were isolated by fluorescence-activated cell sorting. The pS2, PgR, and ErbB2 RNAs in cells were analyzed by quantitative real-time PCR, with normalization to the 18S rRNA present in the same RNA samples. Data are shown relative to empty parental plasmid transfected MCF-7 cells (lane 1). Columns, mean of samples processed in quadruplicate; bars, SE. Table 1. ERRA1 Relative Binding of Affinities (RBAs) for Sequences in Promoters of Human Genes Implicated in Breast Cancer Gene Location* Oligonucleotide Sequence RBA PgR site 1 3,294 tcctaaggactgTCAAGGTCAtcaaatacaagg 1.85 ErbB2 3,441 aaaggaactttcCCAAGGTCAcagagctgagctc 1.08 Reference ERRE NA agcagtggcgatttgTCAAGGTCAcacagt 1.00 PgR site 2 5,166 tccttgctaaacCCAAGGTCAtaaatcttttctc 0.90 Erb 559 ggtgctcccactTAGAGGTCAcgcgcggcgtcg 0.56 pS2 site 2 407 tcccttccccctGCAAGGTCAcggtggccaccc 0.50 Cathepsin D 3,635 tggcatattgggTGAAGGTCAagggagtggcttc 0.49 IGF1R +272 gctccggctcgcTGAAGGTCAcagccgaggcgac 0.37 Human MDM2 +575 gggagttcagggTAAAGGTCAcggggccggggc 0.35 Prolactin 1,347 caaatttgaaacTAAAGGTCAcaggctgcttta 0.32 IGF2 site 2 6,479 ctgtcggcaggaACAAGGTCAccccttggcgttc 0.23 elk1 2,185 ctcccatctcacTTAAGGTCAaagccagggtcc 0.21 BRCA1 293 gtaattgctgtaCGAAGGTCAgaatcgctacctc 0.19 aromatase 99 cctgagactctaCCAAGGTCAgaaatgctgcaa 0.18 PgR site 3 5,912 aaaattgttttgTCTAGGTCAtttgcattttcac 0.14 EGF 396 caaataatgggcTGAAGGTGAactatctttact 0.14 pS2 site 1 266 gtaggacctggaTTAAGGTCAggttggaggaga 0.11 ERa 865 atgtttggtatgAAAAGGTCAcattttatattc 0.10 Abbreviation: NA, not applicable. *Location of the ERRE sequence relative to the gene transcription start site. cSequence found in reverse orientation in the natural promoter. ErbB2 Signaling Regulates ERRa1 Activity Mol Cancer Res 2007;5(1). January 2007 77 on June 19, 2017. © 2007 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from were assayed in parallel as positive and negative controls, respectively. As expected, activated MAPK efficiently phosphorylated PHAS-1, but not GST-h-globin1-123 (Fig. 5C, lane 1 versus lane 5 , respectively). MAPK phosphorylated each of the GSTERRa1 fusion proteins, with significantly more label incorporated into GST-ERRa11-423 than into GST-ERRa11-376 and GST-ERRa11-173 (Fig. 5C, lane 4 versus lanes 3 and 2). Thus, ERRa1 can serve as a substrate of MAPK1/2 and Akt1/2, with multiple phosphorylation sites likely present within the protein, including at least one within the carboxyl-terminal domain. Overexpression of ErbB2 in MCF-7 Cells Converts ERRa1 to an Activator Given that ERRa1 transcriptional activity correlated with the cell ErbB2 status (Figs. 2 and 4), we desired to test more directly whether altering cellular ErbB2 signaling would affect ERRa1 transcriptional activity. One approach we used was to cotransfect MCF-7 cells in parallel with (a) the ERE (5 )regulated and TATA-regulated dual luciferase reporter sets; (b) the expression plasmid encoding wild-type ERRa1; and (c) pErbB2Act, an expression plasmid encoding an activated (oncogenic) form of rat ErbB2 (rat neu; ref. 42) or its empty vector as a control. The cells were cultured in estrogen-free medium in the absence or presence of 1 Amol/L fulvestrant to prevent complications from ERa. As observed above (Fig. 2B), overexpression of ERRa1 alone led to minimal activation of ERE-regulated transcription (Fig. 6A, lane 2 versus lane 1 and lane 6 versus lane 5). Addition of ErbB2Act without exogenous ERRa1 stimulated ERE-regulated transcription f2-fold under both estrogen-free conditions and in the presence of fulvestrant (Fig. 6A, lane 3 versus lane 1 and lane 7 versus lane 5 , respectively). Hence, this activation by ErbB2Act was probably mediated via endogenous ERRa1, not ERa. Strikingly, overexpression of ERRa1 together with ErbB2Act stimulated ERE-regulated transcription almost 4-fold (Fig. 6A, lane 4 versus lane 1 and lane 8 versus lane 5). This effect of ErbB2Act occurred without a change in the level of ERRa1 (Fig. 6B, lanes 2, 4, 6 , and 8). Thus, we conclude that ERRa1 transcriptional activity can be altered by ErbB2 signaling. FIGURE 5. Phosphorylation of ERRa1 in situ and in vitro . A. Autoradiograms of two-dimensional gels showing the in situ phosphorylated states of ERRa1 in MCF-7 cells, BT-474 cells, and BT-474 cells incubated with a murine mAb to ErbB2 (HER2). MCF-7 and BT-474 cells were transfected in parallel with the wild-type ERRa1 expression plasmid. Twenty-seven hours later, the cells were metabolically labeled by incubation for 4 h in phosphate-free medium supplemented with [gP]ATP. Afterward, whole-cell extracts were prepared. ERRa1 was immunoprecipitated with an anti-GST-hERRa1 polyclonal antiserum and resolved by two-dimensional PAGE. The mAb 4D5– treated cells incorporated less P because they were growth inhibited; a second, longer exposure of this same gel is shown directly below the original one. B. In vitro phosphorylation of ERRa1 using activated MAPKs and Akts. Equal amounts of 6 His-tagged ERRa1 were incubated in parallel with activated MAPK1, MAPK2, Akt1, and Akt2 along with [gP]ATP. The products were resolved by 4% to 12% gradient SDS-PAGE and visualized by autoradiography. Myelin basic protein (MBP ) was included in the reactions as an internal positive control. C. Localization of sites of phosphorylation of ERRa1 in vitro by activated MAPK2. Equimolar amounts of the indicated GST-ERRa1 fusion proteins were incubated with activated MAPK2 and [gP]ATP. PHAS-I and GST-h-globin (molecular weight 41 kDa) were incubated likewise in parallel as positive and negative controls, respectively. The products were resolved by 12% SDS-PAGE and visualized with a PhosphorImager. Ariazi et al. Mol Cancer Res 2007;5(1). January 2007 78 on June 19, 2017. © 2007 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from Inhibition of ErbB2 Signaling Abrogates Transcriptional Activation by ERRa1 We next investigated whether blocking specific components within the ErbB2 signaling pathway led to inhibition of transcriptional activation by ERRa1. BT-474 cells were cotransfected with the reporter gene sets and expression plasmids as described in Fig. 2D (lanes 1-4). They were subsequently incubated for 40 h in estrogen-free medium supplemented with 20 Ag/mL nonspecific murine IgG as a control, the humanized anti-ErbB2 mAb trastuzumab at 20 Ag/ mL, or the small-molecule EGFR inhibitor gefitinib at 1 Amol/L (Fig. 7A). Gefitinib blocks transphosphorylation of ErbB2 by EGFR, thereby indirectly inhibiting ErbB2 (35). As expected, overexpression of ERRa1 in the IgG-treated cells led to a 4to 5-fold activation of ERE(5 )-regulated transcription (Fig. 7A, lanes 3 and 4 versus lanes 1 and 2). Incubation with trastuzumab led to an f85% reduction in ERE-regulated transcription by ERRa1 (Fig. 7A, lanes 7 and 8 versus lanes 3 and 4) to a level even below that observed in the presence of only endogenous ERRa1 in the absence of the drug (Fig. 7A, lanes 1 and 2). This large reduction was probably due to treatment with trastuzumab altering as well the transcriptional activity of the endogenous ERRa1 (Fig. 7A, lanes 5 and 6 versus lanes 1 and 2). Incubation with gefitinib led to an even greater f90% reduction in ERE-regulated transcription by ERRa1 (Fig. 7A, lanes 11 and 12 versus lanes 3 and 4). Overexpression of GRIP1 largely failed to reverse the effect of the drug treatments (Fig. 7A, even-numbered lanes). Immunoblot analysis with an ERRa1-specific antiserum showed that incubation with the drugs had not affected accumulation of ERRa1 in the cells (Fig. 7C, lanes 1-3). Immunoblot analysis with antisera specific to the phosphorylated versus unphosphorylated forms of MAPK and Akt confirmed that these drug treatments had, indeed, inhibited activation of the MEK/MAPK and PI3K/Akt signaling pathways in these cells (Fig. 7D, lanes 1-3). Thus, we conclude that disruption of the ErbB2 signaling pathway with either trastuzumab or gefitinib prevented ERRa1 from functioning as an activator of ERE-regulated transcription in BT-474 cells, likely doing so in part by inhibiting addition of specific posttranslational phosphorylations of ERRa1 necessary for it to exist in its activator form. Moreover, blocking the ErbB2-directed cascade of signaling events rendered ERRa1 unresponsive to GRIP1-mediated coactivation. To test whether inhibition of ErbB2 signaling modulates ERRa1 activity by affecting the activities of downstream components in this pathway, we likewise examined the effects on ERRa1 activity of incubation of BT-474 cells with U0126 and LY294002, direct inhibitors of MEK and PI3K, respectively (see Fig. 8). In the cells treated with only DMSO, the solvent for these drugs, overexpression of ERRa1 led, as expected, to an f5-fold activation of ERE-regulated transcription (Fig. 7B, lane 3 versus lane 1). Incubation with 20 Amol/L U0126 led to an f50% reduction in ERRa1-induced transcription regardless of whether GRIP1 was also overexpressed (Fig. 7B, lanes 7 and 8 versus lanes 3 and 4). The effect of incubation with 20 Amol/L LY294002 was even greater, inhibiting ERRa1-mediated activation of ERE-regulated transcription by 75% to 85% (Fig. 7B, lanes 11 and 12 versus lanes 3 and 4). Again, immunoblot analysis showed that the drug-treated cells still accumulated ERRa1 (Fig. 7B, lanes 4 and 5), with U0126 having led to inhibition of MAPK phosphorylation without affecting Akt status (Fig. 7D, lane 4) and LY294002 having led to inhibition of Akt phosphorylation without affecting MAPK status (Fig. 7D, lane 5). Therefore, the MEK/MAPK and PI3K/Akt signaling pathways contribute to the ability of ERRa1 to activate ERE-regulated transcription in BT-474 cells. Discussion We showed here that ERRa1 down-modulated E2-induced ERE-regulated transcription in low ErbB2–expressing MCF-7 cells, doing so even when the coactivator GRIP1 was overexpressed (Fig. 2B). This inability of GRIP1 to overcome repression by ERRa1 was not due to lack of functionality because GRIP1 efficiently enhanced ERE-regulated expression mediated by an amino-terminal deleted variant of ERRa1 in these cells. Thus, the failure of wild-type ERRa1 to respond to GRIP1 must lie with its intrinsic properties in this cell line. On the other hand, wild-type ERRa1 functioned instead as a FIGURE 6. Overexpression of ErbB2Act (activated rat neu oncogene) leads to activation of ERRa1 in MCF-7 cells. A. MCF-7 cells were cotransfected with the ERE(5 )-regulated dual-luciferase sets described in Fig. 2, plasmids expressing ERRa1 or their empty expression plasmid, and ErbB2Act or its empty plasmid as indicated. Cells were harvested 48 h later. Columns, mean of samples processed in triplicate relative to the level present in the cells in lane 1; bars, SE. Solid columns, cells incubated in charcoal-stripped serum (CS-FBS ); cross-hatched columns, cells incubated in charcoal-stripped serum supplemented with E 6 mol/L fulvestrant (FUL ). B. Immunoblot analysis of ERRa1 present in MCF-7 cells treated as in A. The membrane was probed with antibodies specific to ERRa1 and h-actin followed by horseradish peroxidase–conjugated secondary antibodies and visualized by enhanced chemiluminescence and autoradiography. 5 E.H. Vu et al., submitted for publication. ErbB2 Signaling Regulates ERRa1 Activity Mol Cancer Res 2007;5(1). January 2007 79 on June 19, 2017. © 2007 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from ligand-independent activator of ERE-regulated transcription in BT-474 cells, activity that was further stimulated by overexpression of GRIP1 (Fig. 2D). Thus, ERRa1 transcriptional activity and ability to recruit GRIP1 is cell type dependent. ERRa1 also bound to (Fig. 3C, lanes 14-24) and activated transcription via ERREs (Fig. 2E) in BT-474 cells. Putative ERREs were identified in the promoter regions of multiple cellular genes implicated in breast cancer (Table 1), three of which, pS2, PgR , and ErbB2 , were shown here to be upregulated in response to ERRa1 in BT-474 cells, but not in MCF-7 cells (Fig. 4). This is the first report showing that ERRa1 affects ErbB2 expression, with ErbB2 mRNA levels activated almost 30-fold above the level observed inMCF-7 cell. Given that ErbB2 is an indicator of aggressive tumor growth, its regulation by ERRa1 provides another potential link for ERRa1 playing a role in the development of some breast cancers. We also showed here that ERRa1 can exist in multiple phosphorylated isoforms in vitro (Fig. 5B and C) and in situ (Fig. 5A). The extent of this phosphorylation was significantly greater, on average, in BT-474 cells than in MCF-7 cells and reduced by treatment of BT-474 cells with the murine version of trastuzumab (Fig. 5A). Furthermore, ERRa1 could serve directly as a substrate of activated MAPKs (Fig. 5B, lanes 1 and 2 , and C) and Akts (Fig. 5B, lanes 3 and 4) in vitro . Importantly, ERRa1 was shown to be a target of ErbB2 signaling: (a) Overexpression of an activated ErbB2 oncogene converted ERRa1 from a repressor to an activator of ERE-regulated transcription in MCF-7 cells (Fig. 6); and (b) disruption of ErbB2 signaling in BT-474 cells with trastuzumab, gefitinib, U0126, or LY294002 converted ERRa1 from an activator to a repressor of ERE-regulated transcription (Fig. 7). Therefore, we conclude that ERRa1 transcriptional activity is regulated, in part, by the ErbB2 signaling pathway affecting the precise state of phosphorylation of ERRa1 (Fig. 8). Cross-talk between ERa and ERRa1 ERa and ERRa1 can both bind EREs, competing for binding to them (refs. 6-8; Fig. 3C). Vanacker et al. (7) reported FIGURE 7. Effects of modulation of the ErbB2 signaling pathway on ERRa1-mediated activation of transcription. A and B. BT-474 cells were cotransfected as described in Fig. 2 with the ERE(5 )-regulated dual-luciferase reporter set along with the indicated expression plasmids and, likewise, in parallel with the TATA-regulated reporter set. They were incubated for 40 h in estrogen-free medium supplemented with (A) 20 Ag/mL nonspecific mouse IgG, 20 Ag/mL trastuzumab, or 1 Amol/L gefitinib as indicated; or (B) DMSO as the drug vehicle control, 20 Amol/L U0126, or 20 Amol/L LY294002 as indicated. Columns, mean of samples processed in triplicate relative to the levels present in the cells in lanes 1; bars, SE. Hatched columns, cells cotransfected with the GRIP1 expression plasmid; solid columns, cells cotransfected with its empty parental plasmid. C. Immunoblot analysis of overexpressed ERRa1 in BT-474 cells treated as in A and B. The membranes were probed with antibodies that react specifically with ERRa1 and h-actin as a control. Reactive proteins were visualized by enhanced chemiluminescence and autoradiography. Lanes 1 to 3, samples from A; lanes 4 and 5, samples from B done on a different day. D. Immunoblot analysis of the phosphorylation status of Akt and MAPK in BT-474 cells treated as in A and B. The membranes were probed with antibodies that react specifically with phosphorylated Akt (p-Akt ), total Akt, phosphorylated p42/44 MAPK (p-MAPK ), and total p42/44 MAPK. Ariazi et al. Mol Cancer Res 2007;5(1). January 2007 80 on June 19, 2017. © 2007 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from that ERa can also bind to an ERRE, activating transcription 6to 10-fold through this sequence in an E2-stimulated manner. In contrast, we observed only minimal (i.e., 1.5to 2-fold) activation of two different ERRE/SFRE–regulated reporters in MCF-7 and BT-474 cells following addition of E2, conditions that led to 12to 14-fold activation of our minimal EREregulated reporter (Fig. 3A and B). This finding was expected because ERa was unable to significantly bind a consensus ERRE (Fig. 3C, lanes 1-13). Differences between our experiments and the previous report include the following: (a) use of an ERa-negative rat osteosarcoma cell line transfected with an ERa expression plasmid for the transcription assays instead of breast carcinoma cell lines that endogenously express high levels of ERa; and (b) use of ERa-programmed reticulocyte lysates for the protein source for the EMSAs instead of wholecell lysates of COS cells transfected with an ERa expression plasmid. Thus, the ERa protein levels in our reporter gene assays and EMSAs were probably significantly lower than they were in the previously reported experiments. We conclude that ERa probably does not significantly interact with an ERRE when present at physiologic concentrations; however, it remains possible that ERa exhibits a low affinity for some ERREs when its concentration is nonphysiologically high. Because ERRa1 exhibits a strong preference over ERa for binding to ERREs, there probably exists specific ERRa1regulated genes that could serve as biomarkers of ERRa1 activities. Coregulators of ERRa1 GRIP1 has been shown to recognize the nuclear receptor box within the COOH terminus of ERRa1, enhancing transcriptional activity (16). Why, then, did GRIP1 fail to significantly enhance ERRa1 transcriptional activity in MCF-7 cells (Fig. 2B and C)? Barry et al. (10) have reported that the exact sequence of the nine-nucleotide extended half-site sequence and the state of phosphorylation of ERRa1 (38) affect whether ERRa1 preferentially binds to an ERRE as a monomer or homodimer; ERRa1 acts as a repressor when bound as a monomer because it cannot recruit coactivators such as PGC-1a. However, the transcriptional activity of ERRa1 and its ability to recruit the coactivator GRIP1 to an ERE was shown here to be dependent on cell type (Fig. 2B versus D) and ErbB2 status (Fig. 6). Since ERRa1 binds to an ERE only as a homodimer (Fig. 3C; ref. 38), an alternative mechanism(s) must also exist by which ERRa1 transcriptional activity can be regulated. Based on the data presented here, we hypothesize that the recruitment of coactivators such as GRIP1 is determined, at least in part, by the phosphorylation status of specific amino acid residues within ERRa1 (Fig. 8). In addition to ERRa1, GRIP1 itself is also a phosphoprotein target of EGFR signaling via MAPK whose phosphorylation is required for full activity (43). Thus, the effects of activated ErbB2 and the drug inhibitors of EGFR/ErbB2 signaling on transcription (Figs. 6 and 7) could have been due to changes in the phosphorylation status of GRIP1 and ERRa1. PGC-1a can also function as a strong coactivator of ERRa1 (13). However, it is not present in the mammary cell lines studied here (data not shown). SRC-1 and SRC-3/AIB1 have also been shown to stimulate ERRa1 activity in transient transfection assays in some mammalian cell lines, albeit only modestly (15, 16). Thus, GRIP1 is likely the major, physiologically relevant coactivator of ERRa1 in mammary cells. Hence, we hypothesize that ERRa1/GRIP1 complexes likely substitute for ERa/AIB1 complexes as the major activators driving ERE-regulated expression in some breast cancers, especially ERa-negative ones. In these cases, drugs that specifically disrupt these complexes may serve as a novel therapy. Ligand-Independent Regulation of ERRa1 Activities via ErbB2 Signaling Based on the results presented here, we hypothesize the following model for regulation of ERRa1 activities (Fig. 8). In cells expressing ErbB2 at low levels (e.g., MCF-7), ERRa1 exists, on average, in a minimally phosphorylated state in which it binds EREs as a homodimer, yet fails to respond to GRIP1dependent coactivation. Thus, it inhibits transcription. In cells expressing ErbB2 at high levels (e.g., BT-474), ErbB2, as either a homodimer or heterodimer with other ErbB family members, signals additional or alternative phosphorylations of ERRa1, at least in part, through MEK/MAPK and PI3K/Akt signaling pathways. This highly phosphorylated form of ERRa1 binds to both EREs and ERREs as a homodimer, activating transcription via interactions with cellular coactivators such as GRIP1. Thus, changes in specific sites of phosphorylation of ERRa1 induced via the ErbB2 signaling pathway convert ERRa1 between repressor and activator of transcription. The precise mechanism(s) that regulates the interaction between GRIP1 and ERRa1 is still unclear. Barry et al. (38) have proposed that ERRa1 can switch between monomer and homodimer, with only the homodimer form binding coactivators. Another possibility is that the repressor domain(s) of ERRa1 interacts with cellular corepressors, blocking binding of coactivators. Castet et al. (44) recently reported that the corepressor RIP140 inhibits ERRa-mediated trans-activation of ERE-dependent expression. Other data consistent with a corepressor(s) regulating the activity of ERRa1 include (a) identification of a repressor domain within the NH2-terminal region of ERRa1 (45) and (b) overexpression of ERRa1E97G,A98S,A101V, FIGURE 8. Model for modulation of transcription by ERRa1 via the ErbB2 signaling pathway. Only a few of the numerous players in the ErbB2 signaling pathways are indicated, along with the steps in these pathways blocked by the drugs used in this study. See text for details. ErbB2 Signaling Regulates ERRa1 Activity Mol Cancer Res 2007;5(1). January 2007 81 on June 19, 2017. © 2007 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from a variant of ERRa1 that fails to bind DNA due to mutations in its DNA-binding domain P-box, derepressing ERE-regulated transcription in MCF-7 cells, presumably by sequestering a corepressor (8). A third possibility is that changes in phosphorylation lead to changes in other posttranslational modifications in ERRa1, thereby affecting the coregulators with which it interacts. Consistent with this latter hypothesis is the recent finding that ERRa1 is also sumoylated; mutation of one of these sites of sumoylation also affects ERRa1 transcriptional activity. These three mechanisms are not mutually exclusive. Other kinase/phosphatase signaling pathways probably also target ERRa1, leading to changes in ERRa1 activities via alterations in its specific sites of phosphorylation. For example, Barry and Giguère (38) recently reported that protein kinase Cy, an enzyme whose activity is stimulated by epidermal growth factor or phorbol 12-myristate 13-acetate, can phosphorylate ERRa1 within its DNA-binding domain, thereby enhancing both the binding of ERRa1 homodimers to ERREs and transcription. They further showed that stimulation of ERRa1 phosphorylation by incubation of their MCF-7 cells with phorbol 12-myristate 13-acetate can lead to an f2-fold activation of transcription of the pS2 gene via an ERRE present within its promoter. Likely, numerous cellular kinases and phosphatases activated through signaling pathways can affect specific sites of phosphorylation that exist within the A/B, C (38), and E/F (Fig. 5C) domains of ERRa1. Depending on which of these multiple specific sites becomes phosphorylated, ERRa1 functions as a repressor or activator to modulate expression of numerous EREand ERRE-regulated cellular genes. Role of ERRa in Breast Cancer The ErbB family of tyrosine kinase receptors signals diverse pathways that play roles in the development of aggressive breast cancers and their resistance to antihormonal therapy. Hence, factors whose activities are both estrogenindependent and sensitive to disruptors of ErbB2 signaling likely contribute to some tamoxifen-resistant and ER-negative breast cancers. ERRa1 meets the following criteria: (a) the activator form of ERRa1 can functionally substitute for ERa in ErbB2-overexpressing cells (Figs. 2D and 6), and (b) blockade of ErbB2 signaling or its downstream effectors, e.g., MEK/MAPK or PI3K/Akt, leads to conversion of ERRa1 from an activator to a repressor, eliminating the ability of ERRa1 to substitute for ERa (Fig. 7). Thus, ER-positive breast tumors expressing high levels of ErbB2 along with the activator form of ERRa will likely not respond well to hormonal-blockade therapies; rather, they may respond well instead to ErbB2-based therapies such as trastuzumab. Given that ERRa1 likely down-modulates the activity of ERa in some ERa-positive tumors while it functionally substitutes for ERa in other tumors leading to estrogen-independent activation of key genes involved in breast cancer, ERRa1 and its phosphorylation status should be evaluated as biomarkers of prognosis and determinants of specific therapeutic treatments. Moreover, ERRa may have use, in itself, as a target for a new class of drugs, possibly for use in combination with some current therapies. Materials and Methods Cell Lines MCF-7/WS8 mammary carcinoma cells were used in all studies in which MCF-7 cells are indicated; they were clonally derived from MCF-7 cells by selection for sensitivity to growth stimulation by E2 (46, 47). BT-474 cells were obtained from the American Type Culture Collection (Manassas, VA). Both cell lines were maintained in estrogenized medium (i.e., phenol redcontaining RPMI 1640, 10% whole fetal bovine serum, 6 ng/mL insulin, 2 mmol/L glutamine, 100 Amol/L nonessential amino acids, and 100 units of penicillin and streptomycin per milliliter). Two days before reseeding of cells for an experiment, the medium was changed to phenol redand estrogen-free medium containing charcoal-stripped fetal bovine serum (48). The monkey kidney COS-M6 cell line was cultured as previously described (8). Cells were maintained at 37jC in a humidified 5% CO2 atmosphere. Cellular Treatment Agents E2 (Sigma-Aldrich, St. Louis, MO) and the complete antiestrogen fulvestrant (ICI 182,780, Faslodex; a generous gift from AstraZeneca, Macclesfield, United Kingdom) were dissolved in ethanol. Control, nonspecific murine IgG (reagent grade, Sigma-Aldrich) was dissolved in PBS. Trastuzumab (Herceptin; purchased from the Lurie Cancer Center pharmacy) was dissolved in bacteriostatic water. A hybridoma cell line that secretes the mAb 4D5, a murine precursor of trastuzumab directed against the ectodomain of ErbB2 (HER2), was obtained from the American Type Culture Collection; IgG was purified from the ascites fluid of a mouse inoculated with this cell line. Gefitinib (Iressa, ZD1839; a generous gift from AstraZeneca) was initially dissolved in DMSO, followed by further dilution in ethanol. U0126 (Promega, Madison, WI) and LY294002 (Promega) were dissolved in DMSO. All test agents were added to the medium at a 1:1,000 (v/v) dilution. Plasmids Plasmid pcDNA3.1-hERRa1 (ERRa1), encoding the fulllength, 423-amino-acid major human isoform of ERRa, and plasmid pcDNA3.1-hERRa1L413A/L418A (ERRa1L413A/L418A), encoding a variant of ERRa1 defective in the carboxylterminal coactivator-binding LxLxxL motif, have been previously described (8). Plasmid pcDNA3.1-hERRa11-376 (ERRa11-376), generated by PCR-based subcloning, encodes a carboxyl-terminal truncated variant of ERRa1 lacking the coactivator-binding LxLxxL motif. Plasmid pcDNA3-GRIP1 encodes the coactivator GRIP1 (49). The replication-defective retrovirus pJRneu has been previously described (42); it encodes the activated form of the rat neu oncogene (ErbB2Act). Plasmid pEGFP encodes enhanced green fluorescent protein (TaKaRa; Clontech, Palo Alto, CA). Plasmids pTA-ffLuc and pTA-srLuc, containing TATA-box basal promoter firefly and Renilla luciferase reporter genes, respectively, were constructed by insertion via HindIII linkers of the nucleotides 31 to +31 region of the herpes simplex virus thymidine kinase promoter into pGL3-Basic and phRG-B (Promega), respectively. Plasmid pERE(5 )TA-ffLuc, containing five tandem copies of the consensus palindromic ERE, was Ariazi et al. Mol Cancer Res 2007;5(1). January 2007 82 on June 19, 2017. © 2007 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from constructed by insertion into the XhoI and BglII sites of pTA-ffLuc of the oligodeoxynucleotides 5¶-tcgagagAGGTCActgTGACCTctgagagAGGTCActgTGACCT ctctcagAGGTCActgTGACCTctgcgagAGGTCActgTGACCTctgcgagAGGTCActgTGACCTcta-3¶ and 5¶-gatctagAGGTCAcagTGACCTctcgcagAGGTCAcagTGACCT ctcgcagAGGTCAcagTGACCTctgagagAGGTCAcagTGACCTctctcagAGGTCAcagTGACCT ctc-3¶. Plasmid pERRE(5 )TAffLuc, containing five tandem copies of a consensus ERRE, was constructed by insertion into pTA-ffLuc of the oligodeoxynucleotides 5¶-tcgagtcaAGGTCAgagctcgtcaAGGTCActgcagctcaAGGTCAgctagcgtcaAGGTCAgcatgcgtcaAGGTCAa-3¶ and 5¶-gatctTGACCTtgacgcatgcTGACCTtgacgctagcTGACCTtgagctgcagTGACCTtgacgagctcTGACCTtgac-3¶. EREs and ERREs are underlined, with core ERE half-sites indicated in uppercase letters. The previously described plasmid pSFRE 3Luc (refs. 6, 50; generous gift of Jean-Marc Vanacker) contains three copies of a consensus ERRE/SFRE (5¶-TCAAGGTCA-3¶) upstream of the SV40 minimal early promoter regulating firefly luciferase expression. Plasmid phRL-TK contains the herpes simplex virus thymidine kinase promoter driving expression of a humanized Renilla luciferase (Promega). Transient Transfection Reporter Gene Assays Transient transfection assays were done using a dualluciferase system (Promega) in which pTA-srLuc served as an internal control (Fig. 2A). The basal TATA promoter–regulated set (Fig. 2A) was transfected in parallel with the EREor ERRE/SFRE–regulated sets as an external control for the physiologic state of the cells and effects of test agents on basal transcription activity. All data were normalized to both of