ERK/MAPK regulates ERRγ expression, transcriptional activity and receptor-mediated tamoxifen resistance in ER+ breast cancer

Abstract

Selective estrogen receptor modulators such as tamoxifen (TAM) significantly improve breast cancer-specific survival for women with estrogen receptor-positive (ER+) disease. However, resistance to TAM remains a major clinical problem. The resistant phenotype is usually not driven by loss or mutation of the estrogen receptor; instead, changes in multiple proliferative and/or survival pathways over-ride the inhibitory effects of TAM. Estrogen-related receptor γ (ERRγ) is an orphan member of the nuclear receptor superfamily that promotes TAM resistance in ER+ breast cancer cells. This study sought to clarify the mechanism(s) by which this orphan nuclear receptor is regulated, and hence affects TAM resistance. mRNA and protein expression/phosphorylation were monitored by RT-PCR and western blotting, respectively. Site-directed mutagenesis was used to disrupt consensus extracellular signal-regulated kinase (ERK) target sites. Cell proliferation and cell-cycle progression were measured by flow cytometric methods. ERRγ transcriptional activity was assessed by dual-luciferase promoter–reporter assays. We show that ERRγ protein levels are affected by the activation state of ERK/mitogen-activated protein kinase, and mutation of consensus ERK target sites impairs ERRγ-driven transcriptional activity and TAM resistance. These findings shed new light on the functional significance of ERRγ in ER+ breast cancer, and are the first to demonstrate a role for kinase regulation of this orphan nuclear receptor.

Abbreviations

EGF

epidermal growth factor

ER

estrogen receptor

ER+

estrogen receptor α-positive

ERE

estrogen receptor element

ERRγ

estrogen-related receptor γ

ERRE

estrogen-related receptor element

HA

hemagglutinin

MAPK

mitogen-activated protein kinase

Rb

retinoblastoma tumor suppressor

TAM

tamoxifen

Introduction

Worldwide, breast cancer is the most common cancer in women, with an estimated 1.38 million new cases diagnosed per year [1], and approximately 70% of breast cancers are estrogen receptor α-positive (ER+). ER+ breast cancer may be successfully treated using selective estrogen receptor modulators such as tamoxifen (TAM) [2], and the estrogen receptor (ER) is one of only two robust, reproducible biomarkers that are routinely used to make breast cancer treatment decisions in the clinic [3]. However, the development of TAM resistance is a pervasive problem that affects almost half of all women with ER+ breast cancer who are treated with TAM [4-6]. Typically, it is not loss or mutation of ER that causes resistance, but changes in proliferative and/or survival pathways in an ER+ breast tumor cell that over-ride the inhibitory effects of TAM. These frequently include alterations in receptor tyrosine kinases, cell-cycle regulatory proteins, and mediators of apoptosis.

In contrast to hormone-regulated nuclear receptors such as ER, 25 members of the nuclear receptor superfamily lack an identified ligand and are thus designated orphan nuclear receptors [7]. Orphan nuclear receptors display constitutive transcriptional activity, and have been implicated in numerous developmental and disease processes, including breast cancer [8]. A trio of estrogen-related receptors (ERRα, β and γ) are well-established transcriptional regulators of mitochondrial biogenesis and function, including fatty acid oxidation, oxidative phosphorylation and the tricarboxylic acid cycle [9, 10], in organs and tissues with high energy requirements, such as the heart and liver. Multiple studies have now shown that the ERRs alter metabolism and oncogene expression in breast and other cancer cells a way that promotes growth and proliferation [11, 12]. In non-transformed mammary epithelial cells, up-regulation of endogenous ERRγ after detachment from the extracellular matrix contributes to metabolic reprogramming, and, ultimately, the development of resistance to anoikis [13].

As their name implies, ERRs have broad structural similarity to classical ERs, but, being orphan nuclear receptors, have no (known) endogenous ligand and do not bind estrogen. ERRγ (official gene symbol ESRRG, NR3B3) is preferentially expressed in ER+ breast cancer [14]. Expression of endogenous ERRγ is up-regulated during acquisition of TAM resistance by ER+ invasive lobular breast cancer cells, and exogenous expression of ERRγ in this breast cancer type is sufficient to induce TAM resistance [15]. The level of ERRγ mRNA is also significantly increased in pre-treatment tumor samples from women with ER+ breast cancer whose cancer ultimately recurred following TAM treatment [8]. More recently, nuclear expression of ERRγ protein has been shown to correlate with lymph node-positive status in a small cohort of breast cancer patients [16], and gene-level amplification of ERRγ is significantly increased in lymph node metastases versus the primary breast tumor [17].

The goal of the current study is to better understand how ERRγ expression and activity are regulated, and how this regulation contributes to the TAM-resistant phenotype in ER+ breast cancer. We show here that (a) modulation of ERK activity directly affects ERRγ protein levels, (b) serines 57, 81 and 219 are required for ERK-mediated enhancement of ERRγ protein levels, and (c) mutation of these sites abrogates receptor-mediated TAM resistance and reduces transcriptional activity.

Results

The level of ERRγ mRNA (ESRRG) is increased in pre-treatment tumor samples from women with ER+ breast cancer who relapse within 5 years of TAM treatment [8, 18]. Using the KM Plotter tool [19] to test whether there is an association between ERRγ and other clinical parameters in additional patient populations with longer follow-up time, we found that high expression of ESRRG (upper versus lower tertile) is significantly associated with worse overall survival in ER+ breast cancer patients who received TAM as their only endocrine therapy (Fig. 1A, hazard ratio 2.44, log-rank P = 0.035). MCF7/RR cells are a TAM-resistant variant of MCF7 cells [20] that depend on heightened signal transduction through networks regulated by NFκB [21] and glucose-regulated protein 78 [22] for maintenance of the resistance phenotype. Quantitative RT-PCR analysis showed that expression of ERRγ (Fig. 1B) is increased in resistant MCF7/RR cells versus sensitive parental MCF7 cells. However, MCF7 cells have a mean cycle threshold (C_T) > 35, indicative of very low expression outside the optimal range of TaqMan gene expression assays; the mean C_T for MCF7/RR cells is 33. We subsequently performed non-quantitative RT-PCR for ESRRG in independent samples of MCF7 and MCF7/RR cells alongside a human ERRγ ORF cDNA clone (Fig. 1C). Although ESRRG mRNA is detectable in both cell lines, the signal intensity observed in approximately 400 ng cDNA is 40–50% less than that obtained from 800 pg of plasmid. Western blotting showed that MCF7 and MCF7/RR cells have undetectable ERRγ protein in 67 μg of whole cell lysate, while 25 ng of purified ERRγ protein is observed (Fig. 1D). These data show that MCF7 and MCF7/RR cells express very low levels of receptor mRNA, and that endogenous ERRγ protein is not readily detected in these cells by the available commercial antibodies.

Figure 1

ERRγ expression in ER+ breast tumors and breast cancer cells. (A) Expression of ESRRG in ER+, TAM-treated breast tumors is associated with worse overall survival. HR = hazard ratio, calculated using the KM Plotter tool [19]. (B) Relative expression of ESRRG normalized to that of RPLP0 in MCF7 and MCF7/RR cells by quantitative RT-PCR. The mean cycle threshold (C_T) value for parental (MCF7) cells is indicated. Values are means ± SD for n = 3 replicates from a representative assay performed independently twice. Asterisks indicate a statistically significant difference compared with MCF7 cells (P ≤ 0.01, Student's t test). (C) Expression of ESRRG and RPLP0 in MCF7 and MCF7/RR cells by non-quantitative RT-PCR. The upper and lower arrows indicate ESRRG and RPLP0 amplicons, respectively. ‘Plasmid’ indicates the ERRγ ORF cDNA clone. (D) Expression of ERRγ protein in MCF7 and MCF7/RR cells by western blot analysis. The asterisk indicates a non-specific band detected by the ERRγ antibody. ‘Purified protein’ indicates human ERRγ transcript variant 2. β-actin represents a loading control.

We therefore adapted an exogenous expression model comprising MCF7 cells transiently transfected with hemagglutinin (HA)-tagged ERRγ (HA-ERRγ) [15, 23] to determine the mechanism(s) by which this orphan nuclear receptor, when expressed, may modulate the TAM-resistant phenotype. Post-translational modifications such as phosphorylation play essential roles in the regulation of many proteins, including nuclear receptors. At least eight phosphorylation sites have been shown to regulate expression or activity of classical (ligand-regulated) ER [24], and a number of these have clinical significance in women with breast cancer who are treated with TAM [4, 25]. In the absence of identified ligand(s), the activity of orphan receptors is thought to be particularly sensitive to regulation by phosphorylation [26-30]. ERK hyperactivation has been associated with TAM resistance in vivo and in vitro [31, 32], and inhibition of its upstream regulator MEK improves the anti-tumor activity of the steroidal anti-estrogen drug fulvestrant in ER-positive ovarian cancer [33]. Therefore, we tested whether the activity of ERK or the two other major members of this kinase family (Jun N-terminal kinase, JNK, and p38) directly affect exogenous ERRγ expression in MCF7 cells (Fig. 2A, left). The minimal consensus sequence required for phosphorylation of a substrate by any member of the mitogen-activated protein kinase (MAPK) family is the dipeptide motif S/T-P [34], and ERRγ contains four serines but no threonines that meet these criteria: amino acids 45, 57, 81 and 219. Pharmacological inhibition of phosphorylated ERK (pERK) by U0126 strongly reduces exogenous HA-ERRγ levels, but inhibitors of p38 (SB203580) or JNK (SP600125) do not. Furthermore, co-transfection with a mutant, constitutively active, form of MEK (MEKDD [35]) increases pERK expression and enhances HA-ERRγ levels (Fig. 2B), as does co-transfection with wild-type ERK2 (Fig. 2C). Stimulating MCF7 cells with epidermal growth factor (EGF) also increases pERK expression and enhances exogenous HA-ERRγ levels, and these effects are blocked by co-treatment with U0126 (Fig. 2D). Finally, pharmacological inhibition of pERK by U0126 inhibits exogenous HA-ERRγ expression in a second ER+ breast cancer cell line, SUM44 (Fig. 2E). These data strongly suggest that ERRγ is positively regulated by ERK.

Figure 2

Effect of MEK and ERK on ERRγ protein levels. (A) Inhibition of ERK, but not p38 or JNK, reduces exogenous ERRγ expression. MCF7 cells were transiently transfected with the pSG5 empty vector or HA-ERRγ, then treated with DMSO, 5 μm U0126 (MEK inhibitor), 25 μm SB203580 (p38 inhibitor) or 10 μm SP600125 (JNK inhibitor) for 24 h prior to lysis and western blot analysis. The left panels show HA-ERRγ levels, phosphorylated ERK (pERK) and total ERK from a representative experiment repeated at least twice. The right panels show total and phosphorylated p38 and JNK (p-p38 and pJNK, respectively) from the same experiment. (B) Constitutively active mutant MEK enhances ERRγ protein levels. MCF7 cells were transiently co-transfected with HA-ERRγ and either the MEKDD construct or additional pSG5 empty vector. The asterisk indicates the transfected MEKDD construct. (C) Exogenous wild-type ERK2 enhances ERRγ protein levels. MCF7 cells were transiently co-transfected with HA-ERRγ and either MEKDD, wild-type HA-tagged ERK2, or additional pSG5 empty vector. The asterisk indicates the transfected MEKDD construct. The arrow and the ^ symbol indicate transfected HA-ERRγ and HA-ERK2, respectively. (D) EGF-mediated enhancement of ERRγ protein levels is reversed by concomitant ERK inhibition. MCF7 cells were transiently transfected with HA-ERRγ, then cultured under low-serum conditions for 20 h before treatment with DMSO, 25 ng·mL⁻¹ EGF or 25 ng·mL⁻¹ EGF plus 5 μm U0126 for 2 h. (E) Inhibition of ERK reduces exogenous ERRγ expression in a second ER+ breast cancer cell line. SUM44 cells were transiently transfected with the pSG5 empty vector or HA-ERRγ, then treated with DMSO or 5 μm U0126 for 22 h. β-actin represents a loading control.

The putative ERK phosphorylation sites in ERRγ are either located in the N-terminal activation function 1 region of the protein (amino acids 45, 57 and 81), or in the hinge region downstream of the DNA binding domain (amino acid 219). Tremblay et al. [36] have shown that ERRγ and ERRα are regulated via a phosphorylation-dependent SUMOylation motif. Phosphorylation at ERRγ S45 directs SUMOylation at K40, leading to repression of ERRγ transcriptional activity, and, when this serine is mutated to alanine (S45A), ERRγ expression and transcriptional activity are enhanced. Therefore, we generated two variants of ERRγ by site-directed mutagenesis: S45A (part of the phosphorylation-dependent SUMOylation motif) or S57,81,219A (unknown function). In contrast to wild-type and S45A ERRγ, levels of the S57,81,219A variant are decreased by 70% compared to wild-type ERRγ (Fig. 3A). To determine whether these three serine residues are required for the MEK/ERK-mediated increase in ERRγ levels, wild-type or S57,81,219A ERRγ was co-transfected with MEKDD (Fig. 3B). Consistent with data presented in Fig. 2B, activated MEK increases wild-type ERRγ by approximately threefold. However, MEKDD is unable to enhance levels of the triple serine mutant. Similarly, treatment with U0126 reduces wild-type HA-ERRγ levels by 70% (consistent with Fig. 2A), but has no further effect on S57,81,219A ERRγ (Fig. 3C). Serines 57, 81 and 219 therefore appear to be required for regulation of ERRγ protein levels by ERK, and their mutation to alanine reduces basal receptor expression.

Figure 3

Contribution of serines 57, 81 and 219 to ERRγ protein levels. (A) Concomitant serine to alanine mutations at residues 57, 81 and 219 reduce basal HA-ERRγ levels. (B) Inclusion of MEKDD fails to increase the protein levels of S57,81,219A HA-ERRγ. (C) Erk inhibition does not reduce S57,81,219A HA-ERRγ protein levels. MCF7 cells were transiently transfected and treated with 5 μm U0126 or DMSO for 24 h as indicated prior to lysis and western blot analysis. β-actin represents a loading control. The densitometric values for the ratio of HA to β-actin were normalized to the level of wild-type receptor in the absence of treatment (1.0). Data are from representative experiments that were performed independently at least three times.

We next compared S57,81,219A ERRγ to the wild-type receptor with regard to its ability to induce TAM resistance. We first used 5-bromo-2′-deoxyuridine (BrdU) incorporation analyzed by fluorescence activated cell sorting (FACS) to measure changes in DNA synthesis (S phase) following 4-hydroxytamoxifen (4HT) treatment in MCF7 cells transiently transfected with empty vector (control), wild-type or mutant ERRγ (Fig. 4A). As expected, 4HT reduces DNA synthesis by 50% in control (pSG5-transfected) cells. Wild-type ERRγ confers resistance to 4HT (P < 0.05), but S57,81,219A ERRγ does not. We then tested whether 4HT-mediated induction of the cyclin-dependent kinase inhibitors p21 and p27, markers of G₀/G₁ arrest that are essential for TAM-mediated growth inhibition [37, 38], is altered by exogenous ERRγ. Similar to its effect on the estrogen receptors [39], 4HT increases the expression of both wild-type and S57,81,219A ERRγ (Fig. 4B). However, the approximately 1.5- and 1.3-fold induction of p21 and p27, respectively, by 4HT in empty vector-transfected cells is reduced or blocked by exogenous expression of wild-type ERRγ but not mutant ERRγ. We also measured total and phosphorylated levels of the retinoblastoma tumor suppressor (Rb), a target of active cyclin D1/cyclin-dependent kinase complexes and another indicator of G₁ cell-cycle progression. Results obtained regarding the role of Rb in the TAM response and resistance are somewhat contradictory. Some studies report a reduction in pRb in responsive cells following TAM treatment, while others show that loss or down-regulation of total Rb is associated with TAM resistance in cell culture models, xenografts and premenopausal women with ER+ breast cancer [40, 41]. Under vehicle-treated conditions, we observed a strong induction of total and pRb by wild-type but not S57,81,219A ERRγ. When treated with 4HT, the ratio of pRb to total Rb in wild-type ERRγ-expressing cells is increased approximately twofold versus vehicle treatment, and this is due to a robust decrease in total Rb. In the presence of S57,81,219A ERRγ, pRb remains essentially constant, but total Rb is increased when these cells are treated with 4HT. Together, these data show that S57,81,219A ERRγ is impaired in its ability to promote TAM resistance, and suggest that this may be due (at least in part) to altered regulation of cell-cycle progression by mutant versus wild-type receptor.

Figure 4

Effect of the S57,81,219A mutation on the TAM response. (A) Inhibition of BrdU incorporation by 4HT is reversed by wild-type but not S57,81,219A HA-ERRγ. MCF7 cells were transiently transfected as shown, treated with ethanol (vehicle) or 1 μm 4HT for 48 h, then incubated with BrdU for an additional 18–20 h before fixation and staining for HA and BrdU. The dashed line indicates BrdU incorporation in vehicle-treated cells (set to 1.0). Values are means ± SD for n = 3 independent assays. The asterisk indicates a statistically significant difference (P < 0.05, post hoc Dunnett's test after one-way ANOVA). The difference between pSG5 and S57,81,219A mutant HA-ERRγ was not statistically significant (n.s.). For transfections with wild-type or S57,81,219A ERRγ, the data presented are for HA-positive FACS-sorted cells only. For transfections with the empty vector pSG5 control, the data presented are for all cells in the population. (B) The 4HT-mediated induction of cell-cycle inhibitors p21 and p27 is reversed by wild-type but not S57,81,219A HA-ERRγ, and the phosphorylation state of Rb is differentially affected by wild-type versus mutant receptor. MCF7 cells were transiently transfected as shown, then treated with ethanol or 2.5 μm 4HT for 21 h prior to lysis and western blot analysis. β-actin represents a loading control. The densitometric values for the ratio of the indicated proteins to β-actin in 4HT-treated conditions were normalized to the level of their expression in the absence of treatment (1.0) for each transfected construct; for pRb Ser780, the ratio of phosphorylated to total Rb signal (which was then normalized to β-actin) is shown. Data are from a representative experiment that was performed independently three times.

ERRγ directly regulates transcription by binding to estrogen response elements (EREs) or estrogen-related response elements (ERREs). Deblois et al. identified a hybrid ERRE/ERE as the major binding site for ERRα in breast cancer [42]. Because S57,81,219A ERRγ does not induce TAM resistance, we tested whether this mutant shows impaired promotion of transcriptional activity at all three response elements. In MCF7 cells, the activity of mutant S57,81,219A ERRγ versus wild-type ERRγ is significantly reduced by approximately 30% on the ERRE (Fig. 5A) and ERE (Fig. 5B). For the first time, we show that ERRγ also stimulates transcription from the hybrid ERRE/ERE (Fig. 5C). However, the activity of the S57,81,219A mutant ERRγ at this hybrid element is decreased versus the wild-type receptor by < 10%. In contrast, the S57,81,219A mutant ERRγ shows a 30–40% reduction in transcriptional activity at all three response elements in ER+ breast cancer cell line SUM44 (Fig. 5D–F). These data demonstrate that ERK-mediated stabilization of ERRγ positively regulates receptor transcriptional function, and suggest that this is most relevant to ERRE- and ERE-driven activity.

Figure 5

Effect of the S57,81,219A mutation on ERRγ transcriptional activity. MCF7 and SUM44 cells were transiently co-transfected with pSG5 empty vector, wild-type HA-ERRγ or S57,81,219A HA-ERRγ plus ERRE-driven promoter–reporter luciferase constructs (A, D), ERE-driven promoter–reporter luciferase constructs (B, E) or ERRE/ERE-driven promoter–reporter luciferase constructs (C, F) and the Renilla internal control construct for 24 h (MCF7) or 48 h (SUM44) prior to lysis and luciferase assay. Values are means ± SD for the luciferase:Renilla ratio of n = 3 replicate wells from a representative assay performed three times independently. Asterisks indicate statistically significant differences (P ≤ 0.001, one-way ANOVA with post hoc Tukey's test).

Discussion

In this study, we show that ERRγ protein levels are enhanced or stabilized by active ERK, map this activity to three serine residues, and demonstrate that impairment of ERRγ phosphorylation at these sites reduces receptor-mediated TAM resistance and transcriptional activity in ER+ breast cancer cells. We propose that ERK-mediated phosphorylation of ERRγ is a key determinant of TAM resistance in ER+ breast cancer cells in which this receptor is expressed and drives the resistant phenotype.

To our knowledge, this is the first demonstration of direct, functional consequences of phospho-regulation of a member of the ERR family. Ariazi et al. showed that ERRα transcriptional activity in ER+ breast cancer cells was enhanced by human epidermal growth factor receptor 2 (HER2) endogenous amplification (BT474) or exogenous expression (MCF7), and that pharmacological inhibition of AKT or MAPK reduces this activity [26]. They also provided evidence, via in vitro kinase assays using GST-tagged ERRα constructs, that multiple receptor sites (particularly in the C-terminus) may be phosphorylated by AKT and MAPK. However, Chang et al. reported that, in SKBR3 cells (an HER2-amplified, ER-negative breast cancer cell line), expression of endogenous ERRα target genes is repressed by AKT inhibitors but not MAPK inhibitors through regulation of the co-activator peroxisome proliferator-activated receptor-gamma coactivator 1beta (PGC1β) [43]. Moreover, they found that mutation of the proposed phosphorylation sites in ERRα had no effect on promotion of transcriptional activity by the receptor, which is in direct contrast to our finding that mutation of three ERK consensus sites in ERRγ significantly impairs transcriptional activity and receptor-mediated TAM resistance. The possibility that ERRα and ERRγ, despite their high sequence similarity and overlapping target genes, have differential functions in breast cancer has recently gained considerable traction [11, 44], and is one that our future studies will address, particularly with respect to selection of ERE- and ERRE-containing endogenous target genes (see below).

We were surprised by the apparent specificity of ERK for positive regulation of ERRγ in ER+ breast cancer cells. All three members of the MAPK family (ERK, JNK, p38) phosphorylate the same S-P core motif, but our data show that only pharmacological inhibition of ERK reduces ERRγ protein levels. It should be noted that, under these experimental conditions, p38 and JNK are expressed but their activation (phosphorylation) is minimal (Fig. 2A, right). We therefore cannot rule out the possibility that, in other contexts, ERRγ may be regulated by these members of the MAPK family.

It is not yet clear how inhibition of ERK, or the S57,81,219A ERRγ mutation, ultimately lead to a decrease in receptor levels. A possible explanation is a change in proteasomal-mediated degradation of the receptor, i.e. phosphorylation of serines 57, 81 and/or 219 by ERK slows or prevents ubiquitination and degradation of ERRγ. Our data showing that a brief (2 h) stimulation with EGF is sufficient to enhance HA-ERRγ expression is consistent with a proteasomal degradation mechanism. Similar to what was observed here, MEK/ERK-mediated stabilization of the GLI Family Zinc Finger 2 (GLI2) oncoprotein results in reduced ubiquitination of GLI2 that requires intact glycogen synthase kinase 3 beta (GSK3β) phosphorylation sites [45]. Parkin is the only E3 ubiquitin ligase that has so far been shown to ubiquitinate ERRγ (and other members of the ERR family) [46], but knowledge of whether/how Parkin is affected by ERK signaling in breast cancer is limited. In neurons, Parkin and MAPKs act in opposition to regulate microtubule depolymerization [47], and Parkin has been reported to bind microtubules in several breast cancer cell lines and stabilize their interaction with paclitaxel, leading to enhanced sensitivity to this chemotherapeutic drug [48]. In MCF7 cells, exogenous Parkin expression also independently attenuates cell proliferation by causing G₁ arrest [49]. Future studies will determine whether ERK-dependent regulation of ERRγ requires the Parkin and ubiquitin/proteasome pathway.

The reduction in S57,81,219A mutant ERRγ protein levels, and its failure to induce TAM resistance or promote cell-cycle progression in MCF7 cells, is not perfectly correlated with impaired transcriptional activity. S57,81,219A mutant ERRγ is significantly less active at ERRE and ERE sites. However, Fig. 5C shows that the activity of the S57,81,219A mutant at the hybrid ERRE/ERE element is surprisingly close to that of wild-type in MCF7 cells, but is reduced by 30% in SUM44 cells (Fig. 5F). Because these divergent results were obtained using identical, plasmid-borne heterologous promoter constructs (three tandem ERRE/ERE sequences functioning as enhancers of the SV40 core promoter) under similar experimental conditions, we hypothesize that this context-dependent difference in mutant ERRγ activity may be due to a difference in either the repertoire of co-regulatory proteins, or the expression of ERα, in MCF7 versus SUM44 cells. The latter possibility is interesting in light of what is known about the interplay between family members ERRα and ERα at these hybrid response elements. Using serial ChIP assays, Deblois et al. showed that, in MCF7 cells, ERRα and ERα cannot simultaneously occupy these hybrid sites, and reduction of ERα by siRNA enriched ERRα binding to these sequences in the promoter regions of FAM100A and ENO1 [42]. We previously reported that SUM44 cells show high basal expression of ERα [15], representing a threefold enrichment of mRNA and protein levels versus MCF7 cells (P < 0.001, data not shown). This may mean that, where competition with ERα is limited (i.e. in MCF7 cells), S57,81,219A mutant ERRγ is more readily recruited to ERRE/ERE sites. However, S57,81,219A mutant ERRγ is still unable to fully induce TAM resistance in MCF7 cells, and shows compromised activity at ERE inverted repeats and the ERRE half site in these cells. This implies that phosphorylated, wild-type ERRγ may preferentially activate ERE- and ERRE-regulated target genes to promote the TAM-resistant phenotype.

Experimental procedures

Cell lines, culture conditions and reagents

ER-positive, tamoxifen-responsive MCF7 cells were originally obtained from Marvin Rich (Karmanos Cancer Institute, Detroit, MI, USA). The ER-positive, tamoxifen-resistant variant of MCF7 (MCF7/RR cells) were a kind gift from W.B. Butler (Indiana University of Pennsylvania, Indiana, PA, USA) [20]. ER-positive, tamoxifen-responsive SUM44 cells have been described previously [15]. All cells tested negative for Mycoplasma spp. contamination, and were maintained in a humidified incubator with 95% air and 5% carbon dioxide. MCF7 and MCF7/RR cells were cultured in modified improved minimal essential medium (Life Technologies, Grand Island, NY, USA) with phenol red (10 mg·L⁻¹) supplemented with 5% fetal bovine serum. SUM44 cells were cultured in serum-free Ham's F12 medium containing 1.25 mg·L⁻¹ phenol red, with insulin, hydrocortisone and other supplements as described previously [15, 50].

4-hydroxytamoxifen (Sigma, St Louis, MO, USA) was dissolved in 200-proof ethanol, stored as a 10 mm stock at −20 °C, and used at the concentrations indicated. The MEK inhibitor U0126, the JNK inhibitor SP600125 and the p38 inhibitor SB203580 (Tocris Bioscience, Ellisville, MO, USA) were dissolved in dimethyl sulfoxide (DMSO), stored as 10 or 50 mm stocks at −20 °C, and used at the concentrations indicated. Poly-l-lysine was purchased from Sigma. Recombinant human EGF was purchased from PeproTech (Rocky Hill, NJ, USA), and used at the concentration indicated.

Expression constructs and reporter plasmids

An ORF cDNA clone for human ERRγ (AB020639.1) was purchased from GeneCopoeia (Rockville, MD, USA). Wild-type, HA-tagged murine ERRγ (pSG5-HA-ERR3, 100% protein sequence identity to human ERRγ transcript variant 1) has been described previously [15, 23]. The serine to alanine variants S45A and S57,81,219A were generated using a QuikChange Lightning site-directed mutagenesis kit (Stratagene, La Jolla, CA, USA), confirmed by automated DNA sequencing (GENEWIZ, South Plainfield, NJ, USA), and have been deposited at Addgene (Cambridge, MA, USA; plasmid numbers 37849 and 37850, respectively). Amino acid numbers correspond to transcript variant 1. Plasmids encoding constitutively active MEK (pBabe-puro-MEK-DD [51]) and wild-type HA-tagged ERK2 (pCDNA-HA-ERK2 WT [52]) were obtained from Addgene (plasmid numbers 15268 and 8974, respectively).

The ERE-containing promoter reporter construct (3xERE-luciferase) has been described previously [15, 53]. To generate the ERRE-containing reporter construct (3xERRE-luciferase [54]) and the hybrid ERRE/ERE-responsive reporter construct (3xERRE/ERE-luciferase [42]), oligonucleotides were synthesized by Integrated DNA Technologies (Coralville, IA, USA), annealed, and cloned into KpnI/BglII-digested pGL3-Promoter vector(Promega, Madison, WI, USA) using standard techniques. The oligonucleotide sequences are 5′-CCGGACCTCAAGGTCACGTTCGGACCTCAAGGTCACGTTCGGACCTCAAGGTCAGGATCCA-3′ (ERRE forward), 5′-gatctGGATCCTGACCTTGAGGTCCGAACGTGACCTTGAGAACGTGACCTTGAGGTCCGggtac-3′ (ERRE reverse), 5′-CCGGACCTCAAGGTCACCTTGACCTCGTTCGGACCTCAAGGTCACCTTGACCTCGTTCGGACCTCAAGGTCACCTTGACCTGGATCCA-3′ (ERRE/ERE forward) and 5′-gatctGGATCCAGGTCAAGGTGACCTTGAGGTCCGAACGAGGTCAAGGTGACCTTGAGAACGAGGTCAAGGTGACCTTGAGGTCCGggtac-3′ (ERRE/ERE reverse). Bold text indicates consensus ERRE sequences, italic text indicates consensus ERE sequences, and small letters indicate KpnI and BglII sites. Correct annealing and insertion were confirmed by automated DNA sequencing (GENEWIZ), and the corresponding plasmids have been deposited at Addgene (plasmid numbers 37851 and 37852, respectively).

Clinical data

The KM Plotter tool (http://kmplot.com/analysis/) [19] was used to evaluate ERRγ mRNA expression in publicly available breast cancer gene expression data (Affymetrix (Santa Clara, CA, USA) ProbeID 207981_s_at) from 65 patients selected on the basis of the following parameters: overall survival, upper versus lower tertile of ESRRG expression, ER-positive tumors (including those for which ER+ status is extrapolated from gene expression data), TAM as the only form of endocrine therapy, and any chemotherapy.

RT-PCR

RNA was extracted from subconfluent monolayers of exponentially growing cultures using an RNeasy Mini kit (Qiagen, Valencia, CA, USA). One microgram of total RNA was DNase-treated and reverse-transcribed using SuperScript II and other reagents from Life Technologies. Quantitative RT-PCR was performed for individual cDNA samples (fivefold dilution) using TaqMan gene expression assays for ESRRG and ribosomal protein large P0 (RPLP0) as described previously [15]. Standard (non-quantitative) RT-PCR was performed on 400 ng cDNA or 800 pg of the human ERRγ ORF cDNA clone using primers designed to amplify ESRRG or RPLP0 and TaqSelect DNA polymerase from Lucigen (Middleton, WI, USA) under the following PCR conditions: 94 °C for 2 min, 35 cycles of 94 °C for 30 s, 54 °C for 30 s and 72 °C for 1 min 24 s, a final extension at 72 °C for 10 min, then held at 4 °C. The primers used were 5′-GGAGGTCGGCAGAAGTACAA-3′ (ESRRG forward), 5′-GCTTCGCCCATCCAATGATAAC-3′ (ESRRG reverse), 5′-ACCATTGAAATCCTGAGTGA-3′ (RPLP0 forward) and 5′-AATGCAGAGTTTCCTCTGTG-3′ (RPLP0 reverse). The lengths of the PCR products were 241 and 187 bp for ESRRG and RPLP0, respectively..

Transient transfection and immunoblotting

Cells were seeded on six-well, 12-well or 100 mm plastic tissue culture dishes 1 day prior to transfection with the indicated expression constructs using Lipofectamine 2000 or Lipofectamine LTX (Life Technologies) or JetPrime (VWR, Radnor, PA, USA) according to the manufacturer's instructions. For transfections using Lipofectamine 2000, wells were pre-coated with poly-l-lysine. Transfection complexes were removed (and, where indicated, 4HT or kinase inhibitors were added) at 4–6 h post-transfection. For the growth factor stimulation experiment, at 4–6 h post-transfection, the cells were rinsed briefly twice in sterile, pre-warmed NaCl/P_i and cultured under low-serum conditions (0.5% fetal bovine serum) overnight (approximately 20 h) before treatment with EGF in the presence or absence of U0126 for 2 h. For both transfected and non-transfected cells, wells and dishes were lysed in modified radioimmunoprecipitation assay buffer [55] supplemented with CompleteMini protease inhibitor and PhosSTOP phosphatase inhibitor tablets (Roche Applied Science, Penzburg, Germany). Polyacrylamide gel electrophoresis and protein transfer were performed as described previously [15, 55]. Nitrocellulose membranes blocked in either 5% non-fat dry milk or 7.5% BSA in Tris-buffered saline plus Tween for ≥ 1 h were incubated overnight at 4 °C with primary antibodies for phosphorylated Erk1/2 (1 : 1000), total Erk1/2 (1 : 1000), total MEK (1 : 1000), phosphorylated JNK (1 : 5000), total JNK (1 : 500), phosphorylated p38 (1 : 1000), total p38 (1 : 1000), phosphorylated Rb Ser780 (1 : 1000) and total Rb (1 : 1000) (all from Cell Signaling, Beverly, MA, USA), ERRγ (1 : 100, ab82319) (Abcam, Cambridge, MA, USA), p21 (1 : 300, sc-756) and p27 (1 : 500, sc-528) (Santa Cruz Biotechnology, Dallas, TX, USA) or the HA epitope tag (1 : 500, HA.11 clone 16B12) (Covance, Princeton, NJ, USA). For ERRγ detection, 25 ng of purified protein corresponding to human ERRγ transcript variant 2 (Origene, Rockville, MD, USA) was run alongside 67 μg of whole-cell lysates. As a loading control, all membranes were re-probed using β-actin primary antibody (1 : 5000–1 : 10 000; Sigma) for ≥ 1 h at room temperature [15]. Horseradish peroxidase-conjugated secondary antibodies (1 : 5000) were used, and enhanced chemiluminescent detection were performed as described previously [15].

FACS analysis of bromodeoxyuridine (BrdU) incorporation

MCF7 cells were seeded in poly-l-lysine-coated six-well plastic tissue culture plates at a density of 2.5 × 10⁵ cells per well, respectively, 1 day prior to transfection with 4 μg HA-ERR3, the S57,81,219A variant or empty vector (pSG5) using Lipofectamine 2000. Four to 6 h post-transfection, transfection complexes were removed and cells were treated with 1 μm 4HT or ethanol vehicle. Forty-eight hours later, BrdU was added to a final concentration of 10 μm for an additional 18–20 h. Cells were fixed and stained using an allophycocyanin BrdU flow kit with 7-amino actinomycin D (BD Pharmingen, San Jose, CA, USA) according to the manufacturer's instructions with one modification: during incubation with the allophycocyanin-conjugated anti-BrdU antibody, cells were co-stained with Alexa Fluor 488-conjugated anti-HA antibody (Covance) at dilutions of 1 : 50–1 : 100. Fluorescence-activated cell sorting (FACS) was performed on a FACSAria instrument (BD Pharmingen). For wild-type- and mutant-transfected cells, data are presented for only HA-positive (i.e. Alexa Fluor 488-stained) cells; for empty vector-transfected cells, data for all sorted cells are presented.

Promoter–reporter luciferase assays

MCF7 and SUM44 cells were seeded in poly-l-lysine-coated 24- and 12-well plastic tissue culture plates at 7.5 × 10⁴ and 2.0 × 10⁵ cells per well, respectively. The following day, cells were co-transfected with 500 or 1000 ng HA-ERR3, the S57,81,219A variant or empty vector (pSG5), plus 290 or 580 ng of the 3xERE-, 3xERRE- or 3xERRE/ERE-luciferase constructs, and 10 or 20 ng of pRL-SV40-Renilla, respectively. Transfection complexes were removed, and the medium was replaced 4–6 h post-transfection. At 24 h (MCF7 cells) and 48 h (SUM44 cells) post-transfection, cells were lysed and analyzed for dual-luciferase activity as described previously [15].

Image analysis and statistics

imagej (http://rsbweb.nih.gov/ij/) was used to perform densitometry. All statistical analyses were performed using graphpad prism 5.0c for Mac (GraphPad Software Inc., La Jolla, CA, USA), with the exception of the hazard ratio and log-rank P value in Fig. 1A, which were generated by the KM Plotter tool. All values are means ± SD, and statistical significance is defined as P ≤ 0.05. The results of quantitative RT-PCR, BrdU incorporation and promoter–reporter luciferase assays were analyzed by Student's t test or one-way analysis of variance (ANOVA) with post hoc Tukey's or Dunnett's multiple comparison tests.

Acknowledgements

These studies were supported by an American Cancer Society Young Investigator Award (IRG-97-152-16), a Department of Defense Breast Cancer Research Program Concept Award (BC051851), and a Career Catalyst Research Grant from Susan G. Komen for the Cure (KG090187) to R.B.R., as well as by start-up funds from a Lombardi Comprehensive Cancer Center (LCCC) Support grants (P30-CA-51008; Principal Investigator Louis M. Weiner), U54-CA-149147 (Principal Investigator Robert Clarke) and HHSN2612200800001E (co-Program Directors Robert Clarke and Subha Madhavan). M.M.H. was supported by a Lombardi Comprehensive Cancer Center Tumor Biology Training grant (T32-CA-009686; PI Anna T. Riegel) and a Post Baccalaureate Training in Breast Cancer Health Disparities Research grant (PBTDR12228366; PI Lucile L. Adams-Campbell). Technical services were provided by LCCC the Flow Cytometry, Genomics & Epigenomics and Tissue Culture Shared Resources, which are also supported by P30-CA-51008. The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute, the National Institutes of Health, the American Cancer Society, the Department of Defense, or Susan G. Komen for the Cure. We would like to thank Stephen Byers, Robert Clarke, Katherine Cook-Pantoja, Karen Creswell, Tushar Deb, Hayriye Verda Erkizan, Mary Beth Martin, Ayesha N. Shajahan-Haq and Geeta Upadhyay (Departments of Biochemistry and Molecular & Cellular Biology, Oncology, and Pathology, Georgetown University) for sharing reagents, helpful discussions and intellectual insights, and/or critical reading of the manuscript.

Author contribution

MMH and RBR designed and performed experiments, analyzed the data, and wrote the paper. HT and CCS performed experiments.