Oestrogen receptor-α (ERα) is a ligand-dependent transcription factor that primarily mediates oestrogen (E2)-dependent gene transcription required for mammary gland development. Coregulators critically regulate ERα transcription functions by directly interacting with it. In the present study, we report that ELF3, an epithelial-specific ETS transcription factor, acts as a transcriptional repressor of ERα. Co-immunoprecipitation (Co-IP) analysis demonstrated that ELF3 strongly binds to ERα in the absence of E2, but ELF3 dissociation occurs upon E2 treatment in a dose- and time-dependent manner suggesting that E2 negatively influences such interaction. Domain mapping studies further revealed that the ETS (E-twenty six) domain of ELF3 interacts with the DNA binding domain of ERα. Accordingly, ELF3 inhibited ERα’s DNA binding activity by preventing receptor dimerization, partly explaining the mechanism by which ELF3 represses ERα transcriptional activity. Ectopic expression of ELF3 decreases ERα transcriptional activity as demonstrated by oestrogen response elements (ERE)-luciferase reporter assay or by endogenous ERα target genes. Conversely ELF3 knockdown increases ERα transcriptional activity. Consistent with these results, ELF3 ectopic expression decreases E2-dependent MCF7 cell proliferation whereas ELF3 knockdown increases it. We also found that E2 induces ELF3 expression in MCF7 cells suggesting a negative feedback regulation of ERα signalling in breast cancer cells. A small peptide sequence of ELF3 derived through functional interaction between ERα and ELF3 could inhibit DNA binding activity of ERα and breast cancer cell growth. These findings demonstrate that ELF3 is a novel transcriptional repressor of ERα in breast cancer cells. Peptide interaction studies further represent a novel therapeutic option in breast cancer therapy.
The oestrogen receptor alpha (ERα) belongs to a large family of nuclear hormone receptors that mediate oestrogen (17β-oestradiol)-induced mammary epithelial cell proliferation and ductal formation required for mammary gland development [1–4]. Deregulated expression of ERα is therefore implicated in the development of ER-positive breast cancer [5,6]. Approximately 70% of breast tumours are ER-positive and respond well to anti-oestrogen therapy, whereas ER-negative breast tumours are poorly differentiated and display a worse prognosis [7–9]. Although selective ER modulators (SERMs) and anti-oestrogens are effective against ER-positive breast tumours, in many cases metastatic breast tumours eventually become resistant to this treatment .
ERα, a ligand-inducible transcription factor, contains three discrete domains namely, a core DNA binding domain (DBD), a hinge region and two activation function (AF1 and AF2) domains. The DBD is involved in recognizing oestrogen response elements (ERE) on ER target genes . The hinge region is a flexible coil that connects DBD and AF2 domain (also known as ligand binding domain, LBD), and is also shown to be involved in ERα dimerization [11,12]. Activation function 1 (AF-1) acts as hormone-independent, whereas activation function 2 (AF2) binds to oestrogen and mediates oestrogen-dependent actions . Upon ligand binding, ERα dimerizes and translocates into nucleus where it binds to EREs on target genes, and elicits transcriptional response [1,13]. However, ERα requires coregulators for its optimal activity [14,15]. Coregulators modulate oestrogen receptor transcriptional activity by modulating the chromatin structure [16–19]. Additional mechanisms also exist for regulating ER transcriptional activity. For instance, transcription factors like activator protein 1 (AP1), Sp1, nuclear factor-κB (NF-κB) and E2F1 can influence ERα-dependent gene transcription through their physical interactions with the receptor . We identified that ELF3, also a transcription factor, is an ERα interacting protein. However, the mechanism by which ELF3 influences ERα action remains largely unknown.
Epithelial-specific ETS transcription factor-1 (ELF3, also known as ESE-1) belongs to a family of ETS (E-twenty six) transcription factors which play a crucial role in various physiological processes . ELF3 is expressed in organs containing secretary epithelial cells including mammary gland [22,23]. It regulates transcription of a variety of genes that are involved in cellular transformation and inflammation [24–27]. Previously it was reported that p21 activated kinase (PAK1) phosphorylates ELF3 at Ser207 and regulates its stability and therefore, its cellular transformation activity . In support of this, deregulated activity and/or expression of ELF3 in human cancers have been reported [29–32]. In contrast to these reports, ELF3 is also shown to suppress tumour growth .
In this report, we show that ELF3 acts as a transcriptional repressor of ERα in breast cancer cells. ELF3 inhibits ERα transactivation by blocking ERα dimerization, thereby suppressing breast cancer cell proliferation. A small peptide sequence of ELF3 derived through functional interaction between ERα and ELF3 is sufficient to inhibit DNA binding activity of ERα and breast cancer cell growth. We further show that ELF3 is a transcriptional target of ERα in breast cancer cells. Results of the present study indicate ELF3 is a transcriptional repressor of ERα in breast cancer cells.
Human breast cancer cell lines MCF7 and ZR75 were obtained from National Center for Cell Science, and maintained in RPMI 1640 medium with 10% FBS, 2 mM L-glutamine, 10 U/ml penicillin and 10 μg/ml streptomycin. HEK293T cells were obtained from National Center for Cell Science, and maintained in DMEM with 10% FBS, 2 mM L-glutamine, 10 U/ml penicillin and 10 μg/ml streptomycin. When necessary, cells were grown in IMEM supplemented with 2% dextran-charcoal stripped (DCC) serum for 48 h, and then cells were treated with 10 nM of E2 for desired time points. Stable expression or knockdown of ELF3 in MCF7 cells was carried out as described previously .
pcDNA-ELF3 and pcDNA-ERα, which express T7-ELF3 and T7-ERα respectively, were made as previously described [28,33]. ELF3shRNA (four different clones) and control shRNA plasmids were purchased from OriGene. The following ELF3 siRNAs were used in rescue experiments: 5′GCUACCUGUGUAUUGAAAUdTdT3′, 5′GGGCCUUUGGAUCGAAUAUdTdT3′. Various domains of ELF3 and ERα were PCR amplified using appropriate primers (Supplementary Table S1) and subcloned into pGEX4T1 vector. Similarly, green fluorescent protein (GFP)-ELF3 and ELF3ΔETS were generated and ligated to pEGFP-C1 and pcDNA3.1 vectors, respectively. HA-ERα was subcloned into pCMV-HA vector (Clontech). MCF7 genomic DNA was used as a template to amplify 3 kb promoter region of the human ELF3 gene by PCR amplification and then ligated to pGL3 basic vector (Promega) and sequence verified.
Cell proliferation and cytotoxicity assay
Cell proliferation was determined by MTT assay as described previously . To determine the cytotoxic effect of ELF3-derived peptides (wERIPE and mERIPE), cells were seeded in triplicate at a starting density of 3000 cells per well in a 96-well plate. The cytotoxic effect of tamoxifen on MCF7-vec or MCF7-T7-ELF3 was determined by MTT assay. After treating the cells with control, wild type or mutant peptides at various concentrations for 48 h, MTT assay was performed , and then IC50 values were calculated using sigma plot software (version 11).
Flow cytometry (FACS) analysis
Approximately 2 × 105 cells were treated with control, wild type or mutant peptides (2 mg/ml) for 18 h and fixed using ice cold 70% ethanol. Cells were then washed with PBS and treated with DNAse-free RNAse A (1 mg/ml) and propidium iodide (50 μg/ml) for 30 min. The stained cells were then analysed on a FACS Aria (BD Biosciences).
Quantitative real-time reverse transcription PCR analysis
RNA isolation, cDNA preparation and quantitative real-time reverse transcription PCR (qRT-PCR) were performed as described previously . Appropriate primers (Supplementary Table S1) were used for pS2, Cathepsin D, cMyc, CycD1 and GAPDH genes. Data are presented as fold change relative to the control sample.
GST pull-down assay
Using TNT kit (Promega), ERα and ELF3 were in vitro translated and GST pull-down assay was performed as described previously .
Western blotting and co-immunoprecipitation assays
Western blot and co-immunoprecipitation (Co-IP) were performed as described elsewhere . Blots were probed with the following protein specific antibodies: anti-ERα (Abcam), anti-ELF3 (Epitomics or R&D Systems), anti-β-actin (Sigma), anti-T7 (Novagen) and anti-GFP (Invitrogen). Approximately 1 mg of MCF7 cell lysate was immunoprecipitated with either control IgG or ELF3 antibody overnight at 4°C and immune complexes were separated on 8% SDS-PAGE and, proteins were then detected Western blotting (Bio-Rad Laboratories).
Colocalization studies were performed as described previously . Cells grown on cover slips were fixed with 4% paraformaldehyde and permeabilized by pre-chilled acetone and methanol (1:3). Following incubation with primary antibody at 4°C overnight and secondary antibodies conjugated with Alexa 546 (red) or Alexa 488 (green) dye (Molecular Probes) at room temperature for 1 h, cover slips were mounted on glass slides using DAPI (4′,6-diamidino-2-phenylindole). Fluorescent images were captured by Leica confocal microscope (Germany). For cellular uptake of biotin-labelled ELF3 peptides (wild type and mutant) by MCF7 cells, streptavidin-conjugated with Alexa 546 (green) was incubated and the images were captured by confocal microscope.
Luciferase reporter gene assay
Cells were transfected with either pELF3-promoter-Luc or pERE-Luc and pRenilla-Luc plasmids (internal control) using Lipofectamine 2000 reagent. Twenty-four hour post transfection, cells were treated with or without E2 (10 nM) for 18 h, lysed in lysis buffer, and the luciferase activity was determined using dual luciferase reporter assay kit (Promega).
Cells were resuspened in five volumes of buffer A (10 mM HEPES, pH 7.9, 1.5 mM MgCl2, 10 mM KCl, 0.5 mM DTT, 1× protease inhibitor cocktail), for nuclei isolation. Nuclei were then resuspended in two packed nuclear volumes of buffer B (20 mM HEPES, pH 7.9, 25% glycerol, 0.42 M NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 1× protease inhibitor cocktail). ERE duplex oligos were end-labelled with [32P]ATP using polynucleotide kinase (PNK) enzyme. Nuclear extracts were incubated with radiolabelled DNA probes. For super shift assays, we used 1 μg of ERa antibody (Abcam). Samples were run on 4% polyacrylamide gels and electrophoresed at 150 V in 0.5× TBE buffer for 5 h at room temperature. Gels were dried and exposed to phosphoimager cassette for at least 16 h and then it was scanned by Typhon scanner.
ChIP assay was performed as described previously . Protein–DNA complexes were immunoprecipitated with either anti-ERα or IgG antibodies followed by protein A/G beads. After thorough washing, decross-linking, immunoprecipitated DNA was purified by phenol/chloroform (1:1) mixture, and then subjected to real-time qPCR analysis for desired genes using appropriate primers (Supplementary Table S1).
Peptide preparation and binding studies
Biotin-labelled ELF3 peptides were customarily synthesized by Peptide 2.0 and used for liposomal encapsulation as described previously . This process was repeated in a separate sample without peptides and was considered as control liposome. In vitro translated [35S]methionine-labelled full-length ELF3 was incubated with equal aliquots of GST or GST-ERα-C along with either wild type or mutant peptides (10 nM to 1 μM). The reaction mixture was diluted to 0.5 ml with GST binding buffer (25 mM Tris/HCl, pH 8.0, 50 mM NaCl, 10% glycerol, 0.1% Nonidet P-40). GST pull-down assays were performed as described previously .
Circular dichroism (CD) spectra analysis
Far UV CD spectra of the wERIPE and mERIPE peptides dissolved in methanol were recorded from 250 to 200 nm using a JASCO spectropolarimeter using 0.1-cm quartz cuvette and the spectral data were analysed.
Soft agar colony formation assay
Effect of ELF3-derived peptides on anchorage-independent growth capacity of MCF7 cells was determined by soft agar colony formation assay as described previously . MCF7 cells (1×104) transfected with control, mERIPE or wERIPE (1 μg/ml) were added to culture plates (35 mm) containing agar-medium mixture. Plates were incubated at 37°C for 3 weeks and colonies were manually counted under light microscope (Olympus).
In silico analysis of protein domain interactions
A de novo approach to modelling from a small number of amino acids was used for modelling of ELF3 peptides [36,37]. To find out the amino acid residues involved in the binding between DBD of ERα with ETS domain of ELF3, we developed 3D structures of wERIPE (wild type ELF3 peptide) and mERIPE (mutant ELF3 peptide), and performed the protein–protein docking in HEX (http://hexserver.loria.fr/). Oncomine, a publicly available cancer data base (Compendia Biosciences), was used for gene expression analysis. To analyse ELF3 expression in ER-negative compared with ER-positive breast tumours, two breast cancer datasets were exported from Oncomine [38,39]. Box plot was used to show log2-median centred ratio of ELF3 expression in ER-negative compared with ER-positive breast tumours.
All the experiments were performed at least 2–3 times for reproducibility. One-way ANNOVA tests were used to perform pair wise comparisons. All tests were performed using Prism software (Graph pad 5.0). P value <0.05 is considered as significant.
Identification of ELF3 as an ERα-interacting protein
Screening of a human mammary cDNA library using ERα as bait by a yeast two-hybrid approach as described previously  revealed that ELF3 was one of the several ERα-interacting proteins. The specificity of the interaction between ERα and ELF3 was confirmed by one-to-one two-hybrid interaction assay (Figure 1A). In vitro assays by GST pull-down experiments further demonstrated the interaction of GST-ELF3, but not GST, with [35S]methionine-ERα (Figure 1B). To ascertain the ELF3–ERα interaction in a more physiological context, MCF7 cells were treated with various concentrations of E2 and the cell lysates were analysed by Co-IP assay. Endogenous ERα readily co-immunoprecipitated with ELF3 in the absence of E2, but E2 treatment reduced the interaction between ERα and ELF3 in a dose-dependent manner (Figures 1C and 1D). Next, the dynamic interaction between ELF3 and ERα in MCF7 cells upon E2 treatment was analysed by Co-IP assay. As shown in Figures 1(E) and 1(F), ERα displayed a strong interaction with ELF3 in the absence of E2. However, ELF3–ERα interaction is gradually decreased over time (from 30 min to 9 h) upon E2 treatment, but partially restored after 18 h. Together the compelling evidence suggests the existence of a dynamic molecular interaction between ELF3 and ERα, which is negatively influenced by E2, in MCF7 cells.
ELF3 interacts with ERα both in vitro and in vivo
Mapping of ERα and ELF3 interacting domains
To map the domains in ERα that participate in the interaction with ELF3, in vitro GST pull-down assays were performed using GST fusions of various ELF3 domains namely Pointed (1–128 aa), trans-activation domain (TAD) (128–159 aa), serine aspartic acid-rich domain (SAR) (189–229 aa), AT hook domain (238–259 aa) and ETS domain (274–354 aa), and in vitro translated [35S]methionine-ERα. As shown in Figure 2(A), ERα bound to full-length ELF3 (1–371) and the ETS domain (274–354 aa). Conversely, in vitro translated [35S]methionine-ELF3 interacted strongly with C domain (176–251 aa, DNA binding domain) and weakly with D domain (251–305 aa, hinge region) of ERα (Figure 2B).
Mapping of protein domains involved in the interaction between ERα and ELF3
ELF3 acts as a transcriptional repressor of ERα
After establishing the interaction between ELF3 and ERα, we next examined the effect of ELF3 on the transcriptional activity of ERα. MCF7 cells were cotransfected with pERE-luc (an artificial oestrogen-responsive element-containing reporter) and either vector control or pcDNA-ELF3 plasmids. Ectopic expression of T7-ELF3 decreased E2-mediated transactivation function of ERα in a dose-dependent manner (Figure 3A). A similar repressive effect of ELF3 on ERα transactivation was also observed in ZR-75 cells (Figure 3B). To further substantiate these findings, ERE-Luc assay was also employed in HeLa cells, an ER-negative cell line, with or without ERα transfection. HeLa cells, in the absence of ectopic ERα expression, displayed below basal levels of ERE-luc activity and so no effect of ELF3 on luciferase activity was observed, whereas ectopic expression of ELF3 significantly decreased ERα transactivation function in ERα-transfected cells implying the direct repressive effect of ELF3 on ERα activity (Figure 3C). Next we verified whether ETS domain was responsible for ELF3’s repressive activity on ERα-dependent transcription. In concordance with the interaction data, ETS deletion mutant, i.e., ELF3ΔETS, which although predominantly localizes into nucleus (Supplementary Figure S1), significantly lost its ability to repress E2-dependent ERα transactivation (Figure 3D). To further investigate the role of endogenous ELF3 in ERα transactivation, ELF3 was knocked down (siRNA designed against 3′UTR of ELF3) in MCF7 cells and the ERE-luc activity was analysed. ELF3 knockdown indeed significantly increased the transactivation capacity of ERα (Figure 3E). However, ectopic expression of ELF3 siRNA-resistant T7-ELF3 in ELF3-knocked down cells partially restored the repressive effect of ELF3 (Figure 3E). The effect of ELF3 on the expression of endogenous ERα target genes in MCF7 cells upon treatment with E2 was assessed by real-time quantitative PCR to further corroborate the results of the luciferase assay. ELF3 ectopic expression indeed decreased the expression of both pS2 and Cathepsin D genes in MCF7 cells (Figure 3F). Similar results were also obtained for c-MYC and Cyclin D1 (CCND1), other ERα target genes (Supplementary Figure S2). Together these results indicate that ELF3 represses ERα transcriptional activity in breast cancer cells.
ELF3 represses ERα transcription function in breast cancer cells
ELF3 inhibits ERα recruitment to its target gene promoter by blocking its dimerization
As our domain mapping studies revealed the involvement of ETS domain in interaction with DNA binding domain (C domain) and hinge region (D domain) of ERα, we first tested whether ELF3 inhibits ERα binding to oligos containing ERE elements by gel shift (EMSA) assay. E2–ERα complex from control nuclear lysate (vec-treated) readily bound to ERE containing oligos (Figure 4A, lanes 2–4). However, supplementing wt-ELF3 (lanes 5–7), significantly decreased ERα binding to ERE oligos in a dose-dependent manner. We also examined the effect of ELF3ΔETS on ERα binding to ERE-oligos by gel shift (EMSA) assay. Wt-T7-ELF3 (lane 8), but not mutant T7-ELF3ΔETS (lane 9), decreased ERα binding to ERE oligos (Figure 4B). Furthermore, we observed a super shift with vector- as well as T7-ELF3ΔETS-transfected MCF7 nuclear lysates incubated with ERα antibody (lanes 11 and 13) but not with wt-T7-ELF3 lysate (lane 12). Next, promoter occupancy of ERα upon ELF3 knockdown (Supplementary Figure S3) in response to E2 treatment of the ERE elements within the endogenous pS2 and Cathepsin D gene promoters was analysed by ChIP assay. ELF3 knockdown significantly increased ERα recruitment to pS2 and Cathepsin D gene promoters as compared with control shRNA cells (Figure 4C).
ELF3 inhibits ERα DNA binding capacity and recruitment to target gene promoters by affecting its dimerization
ERα is a ligand-dependent transcription factor. Upon ligand binding it undergoes conformational dependent dimerization to be recruited and to associate with target gene chromatin [40,41]. To test whether the inhibitory activity of ELF3 on ERα recruitment to target gene chromatin was due to its ability to affect ERα dimerization, we cotransfected pcDNA-ERα (T7-ERα), pCMV-ERα (HA-ERα) and either GFP or GFP-ELF3 into HEK293T cells, which express neither ERα nor ELF3 endogenously, and Co-IP assay was performed after treating the cells with E2 (10 nM) for 1 h. HA-ERα readily co-immunoprecipitated with T7-ERα in GFP-transfected cells upon E2 treatment (lane 4) but not in GFP-ELF3-transfected cells (lane 6) (Figure 3C). Thus, ELF3-mediated repression of ERα transcriptional activity could be attributed to its ability to inhibit ERα dimerization in vivo (Figure 4D).
ELF3 inhibits oestrogen-dependent breast cancer cell proliferation
To further investigate the functional relationship between ERα and ELF3 with respect to breast cancer cell growth, cell proliferation assay was performed under conditions of ELF3 ectopic expression as well as knockdown in MCF7 cells. Preliminary cell cycle analysis using flow cytometry (FACS) indicated that ELF3 suppresses cell proliferation (reduced S phase cells; vec compared with ELF3: 20% compared with 16%) (Supplementary Figure S4). After confirming the ectopic expression of either T7-ELF3 or T7-ELF3ΔETS (Figure 5A, inset) in MCF7 cells, cells were treated with either ethanol or E2 (10 nM) for 1–5 days and cell proliferation was determined. ELF3 ectopic expression indeed decreased cell proliferation compared with either control vec-transfected or T7-ELF3ΔETS-transfected cells. Conversely, ELF3 knockdown increased E2-dependent cell proliferation in MCF7 cells (Figure 5B). Because ELF3 displayed inhibitory activity towards ERα functionality, we tested whether it confers tamoxifen (an ERα antagonist) resistance in MCF7 cells. Over two-fold (∼2.4) increased IC50 value in ELF3-transfected cells as compared with control cells indicates ELF3 over expression confers tamoxifen resistance in MCF7 cells (Figure 5C). Together, these results suggest that ELF3 suppresses ERα-mediated cell proliferation and confers tamoxifen resistance in MCF7 cells.
ELF3 suppresses ERα-mediated breast cancer cell proliferation
ELF3 peptides inhibit DNA binding activity of ERα
Having established the functional interaction between ELF3 and ERα, we investigated mapping the minimal region of ELF3 that participates in interaction with DBD of ERα. A highly conserved region between 331 and 347aa (aa, amino acid) of ELF3 is involved in interaction with DBD of ERα (Figure 6A). Protein secondary structure analysis predicted a helical structure for this 17aa peptide named ‘wERIPE’ (wild type ERα-interacting protein of ELF3). CD analysis further confirms predominantly a helical structure for wERIPE (Supplementary Figure S5). Molecular docking studies further indicated that Arg334 and Gly347 (corresponds to Arg4 and Gly17 of wERIPE) of ETS domain of ELF3 were involved in hydrogen bonding with Asn217 and Tyr219 (corresponds to Asn11 and Tyr13 of DBD domain of ERα (Supplementary Table S2); and Gln226 was located in ‘D box’ region (corresponds to Gln48 of DBD domain of ERα) in the DBD of ERα respectively (Figure 6B, top panel). Residues present in ‘D box’ have been shown to be involved in receptor dimerization [42,43]. Replacement of the amino acids, Tyr7, Ile11, Leu12, Ile14 and Val15 with Pro, Ala, Pro, Arg and Ala respectively, in this 17 aa peptide of ETS domain (now referred to as mERIPE, mutant ERα-interacting protein of ELF3), resulted in significant loss of peptide helical structure (Supplementary Figure S5) and also its interaction with critical residues of D box domain in the DBD of ERα (Figure 6B, lower panel) (Supplementary Figure S6). Because ELF3 is interacting with ERα through the ETS domain, we assumed that 17aa peptide could compete with ELF3 to bind to ERα. GST pull down assay was performed to test this hypothesis, by incubating [35S]ELF3, GST-ERα-C and increasing concentrations of either wt- or mt-peptides. As shown in Figure 6(C), wERIPE could inhibit the interaction between ELF3 and ERα effectively at 1 μM concentration, whereas the mERIPE failed to do so. Next we examined the effects of wERIPE and mERIPE on the DNA binding activity of ERα by EMSA. wERIPE indeed displayed more inhibitory activity towards ERα’s DNA binding than mERIPE (Figure 6D). Together, these results indicate that wERIPE could inhibit the DNA binding activity of ERα and potentially suppress ERα functions.
In vitro and in silico analysis of interaction between ERα and ELF3-derived peptides (ERIPEs)
ERα-interacting peptides of ELF3 affect breast cancer cell growth
As in vitro studies indicated the inhibitory effect of ERα-interacting peptides of ELF3 (ERIPEs) towards ERα DNA binding activity, we next explored the effect of ELF3 peptides on breast cancer cell growth. We examined the cellular uptake of wERIPE and mERIPE (biotin-labelled) in MCF7 cells by confocal microscopy. As shown in Figure 7(A), both wERIPE and mERIPE were localized to both cytoplasm and nuclear compartments. The cytotoxic effect of these peptides on ER-positive (MCF7 and ZR-75) and ER-negative (MDA-MB231) breast cancer cells was also analysed. wERIPE showed ∼5-fold more cytotoxicity than mERIPE (IC50: 0.15 μg/ml, wERIPE compared with 0.75 μg/ml, mERIPE) in MCF7 cells (Table 1). Similar observations were made in ZR-75-1 cells. However, no cytotoxic effects of either wERIPE or mERIPE in MDA-MB231 cells were observed (Table 1). We next treated the MCF7 cells with either wERIPE or mERIPE for 24 h, and analysed the cell cycle stages by FACS assay to verify whether the cytotoxic effect of ELF3-derived peptides was due to arrest of cells at sub-G1 stage (apoptotic). As shown in Figure 7(B), wERIPE-treated cells displayed more apoptosis (∼40%, sub-G1 cells) than in either control (5%) or mERIPE-treated cells (15%). Furthermore, wERIPE-treated MCF7 cells showed increased PARP proteolysis (Figure 7C, lane 3) compared with either vehicle or mERIPE (lanes 1–2) treated cells (Figure 7C). In light of these findings, we analysed the effect of ELF3-derived peptides on anchorage-independent growth ability of MCF7 cells. wERIPE significantly reduced the colony forming ability of MCF7 cells than control or mERIPE (Figure 7D). Together these results indicate that wERIPE could inhibit breast cancer cell growth partly by inducing cellular apoptosis.
ERα interacting peptides of ELF3 (ERIPEs) affect breast cancer cell growth
|S. No.||Peptide||Cell type||IC50 IJQ/ml|
|S. No.||Peptide||Cell type||IC50 IJQ/ml|
ELF3 is an oestrogen-inducible gene
Because ELF3 displayed repressive activity towards ERα-dependent transcription and functions, we reasoned that increasing the expression of endogenous ELF3 may provide an option to treat ER-positive breast cancers. In view of this, we analysed human ELF3 promoter. Surprisingly, we found eight ½ ERE sites spanning 3 kb promoter region (202007562–202011137 bp) of ELF3 gene which is located on chromosome 1q32.1 (Supplementary Figures S7 and S8). Gene expression Omnibus (GSO) data sets (GDS1326; GDS4061) indicated the induction of ELF3 in response to E2 (Supplementary Figure S9). We cloned the 5′ UTR of 3 kb in length upstream of the transcriptional start site of the ELF3 gene into a pGL3-Luciferase reporter vector and a luciferase assay was performed in MCF7 cells to confirm whether ELF3 is an E2 responsive gene. As shown in Figure 8(A), ELF3 promoter-luciferase activity was significantly increased upon E2 treatment as compared with the untreated cells. Further a significant increase in ELF3 transcript and protein levels upon E2 treatment in MCF7 cells was observed (Figures 8B and 8C). To further confirm the direct involvement of ERα in ELF3 transcription, MCF7 cells were treated with E2 in presence or absence of ICI-182,780 (an ERα antagonist) and ELF3 transcripts levels were analysed by RT-PCR. As shown in Figure 8(D), ICI treatment decreased the ELF3 levels in E2 treated cells. ERα occupancy at ½ ERE sites on ELF3 promoter in response to E2 treatment was further analysed by ChIP assay using ERα antibody. ERα readily recruited to all ½ ERE sites on ELF3 gene promoter, albeit with different affinities (Figures 8E and 8F). Thus, these results indicate that ELF3 is an E2 response gene in breast cancer cells.
ELF3 is an E2 inducible gene in breast cancer cells
ELF3 expression correlates with ER status in breast cancers
E2-dependent expression of ELF3 in breast cancer cells predicted a potential positive correlation between ELF3 and ERα expression in breast cancer cell lines or subclasses of breast tumours where this pathway played a significant role. We first analysed the expression of ELF3 in ER-positive and ER-negative breast cancer cell lines to test this hypothesis. Western blot analysis indicated the co-expression of ELF3 and ERα in ER-positive breast cancer cells (MCF7 and ZR-75) but not in ER-negative breast cancer cells (MDA-MB231 and MBA-MD435) (Figure 9A). Next we focused on microarray datasets that classified tumour samples based on ER subtypes. Two datasets obtained from the Oncomine data base showed moderately high expression of ELF3 in ER-positive breast cancer subtype over ER-negative, indicating that ELF3 expression in response to E2-ERα observed is functionally relevant to ER-positive breast cancers (Figure 9B). These results together suggest that ELF3 is an E2 response gene. ELF3 thus expressed feeds back to suppress ERα functions in breast cancer cells (Figure 9C).
ELF3 expression correlates with ER status in breast cancers
Because the majority of breast cancers are ERα-positive, and existing therapies to treat breast cancer are not effective, understanding the molecular mechanisms that regulate ERα functions in breast cancer cells is of continued interest. Furthermore, as coregulators are proven to be critical factors in regulating ERα functions both in normal development as well as in cellular transformation, considerable amount of interest led to identifying several ERα coregulators . With this background, we identified ELF3 as an ERα interacting protein that represses ERα transactivation functions in breast cancer cells. Mechanistic studies revealed that ELF3 controls ERα transactivation functions by blocking ERα dimerization and thereby, its DNA binding activity towards ERα target genes. Another important finding from the present study through molecular docking studies is the identification of a 17 amino acid synthetic peptide derived from ELF3, which we named as ERIPE, that could inhibit ERα DNA binding activity and ER-positive breast cancer cell growth.
The role of ETS family factors in mammary gland development and breast cancer has been well documented . For instance, ETS-1, which has been shown to promote breast tumour metastasis, is an ERα coactivator in breast cancer cells . ERG, another ETS factor, is reported to inhibit ERα transcriptional activity implying an inherent association of ETS activity with ERα . In the present study, we determined the functional interaction between epithelial-specific ETS factor, ELF3, and ERα in breast cancer cells. As the functionality of ERα is dependent on its ability to bind and transcriptionally up-regulate the genes that are involved in cell proliferation to cause cellular transformation [1,4], we report that ELF3 inhibits breast cancer cell proliferation by repressing ERα transcriptional activity. We found a strong interaction between ERα and ELF3 in MCF7 cells. But oestrogen treatment alleviated such interaction implying E2 induced conformational change in the receptor might result in ELF3 dissociation from ERα. The other possibility could be the recruitment of coactivators to the receptor upon treatment with E2. Furthermore, decreased interaction of ELF3 with ERα upon E2 treatment is attributed to the reduced repressive activity of ELF3 towards ERα transcriptional functions in response to E2 (Figures 3–5). Remarkable in this respect is the involvement of ETS domain in interaction with DNA binding domain of ERα implying that ELF3-mediated transcriptional repression at ERα target genes occurs through inhibition of ERα DNA binding activity. This is not the case for ERG inhibiting ERα functions, where ERG does not physically interact with ERα . Rather, N-terminal and C-terminal transactivation domains of ERG are involved in repressive functions . This argues that the transcriptional repressive activity displayed by ELF3 on ERα is different from ERG and offers a repertoire of inhibitory mechanisms to control ERα activity in breast cancer cells.
Ligand binding triggers conformational change in the ERα to promote its dimerization [40,48]. Dimerization is an important step for ERα binding to DNA and so its transcriptional activity [49,50]. Therefore, blocking of ERα dimer formation has grave implications for its transcriptional activity and breast cancer cell growth. Both LBD as well as hinge region is shown to participate in ERα dimerization [42,43]. Proteins like Calmodulin bind at hinge region and enhance ERα dimerization [11,12]. In the present study, employing HEK293 cells, which express neither ERα nor ELF3, we demonstrate that ectopically expressed GFP-ELF3 inhibits interaction of T7-ERα with HA-ERα i.e., ERα dimerization. Furthermore, residues present in ‘D box’ (D for dimerization), which is located in DNA binding domain, have been shown to be involved in receptor dimerization [42,43]. Our molecular docking studies suggest that Arg334 and Gly347 of ETS domain of ELF3 are involved in hydrogen bonding with Asn217, Tyr219 and Gln226 of DBD of ERα. Of these residues, Gln226 was located in ‘D box’ of DBD domain of ERα. It implies that ELF3 may inhibit ERα transcription functions by blocking its dimerization through its direct interaction with D box residues. As our in vitro studies show weak interaction of ELF3 with hinge region of ERα, the possibility of hinge region-mediated ERα dimerization aided by Calmodulin protein cannot be ruled out. Our observation in the present study sugggests that inhibition of DNA binding capacity of ERα on to ERE elements by ELF3 involves direct interaction of ETS domain of ELF3 with DBD of ERα that impedes receptor dimerization, DNA binding and subsequently its transcription activity (Figure 10).
Model illustrating the repressive activity of ELF3 on ERα functionality
Approximately 70% of the breast cancers are ER-positive [7–9]. Therefore, SERMs have been one of the important therapeutic options for treating oestrogen receptor-dependent breast cancers . One such drug is tamoxifen, which inhibits ERα functions by blocking ligand binding and suppresses ER-dependent transcription and, therefore it is being used widely to treat breast cancers. However, longer treatments with tamoxifen resulted in the development of drug resistance and posed increasing risk of endometrial cancers [10,52]. Given these clinical-associated problems, developing alternative approaches to treating breast cancer has become a challenging task. Exploring the domain mapping studies, we identified a 17 aa region of ELF3 called ERIPE that could inhibit ERα DNA binding activity. Further, liposome encapsulated ERIPE is being transported into breast cancer cells tested and exhibited the cytotoxic effect in MCF7 cells, an ER-positive breast cancer cells, but not in MDA-MB231 cells, an ER-negative cell line. Hence ERIPE can be developed into a drug to treat ER-positive cancers. However, the efficacy of this peptide in vivo needs further investigation.
A plethora of oestrogen target genes have been identified and characterized [13,53]. Many of these gene products are directly or indirectly associated with mammary gland development and their expression has correlated with breast cancer. In this report, we identified ELF3 as an oestrogen target gene in breast cancer cells. The finding that ELF3 is an ER-responsive gene suggests the existence of a negative regulatory loop between the two proteins in which E2-dependent up-regulation of ELF3 feeds back to inhibit ERα function and down-regulate the oestrogenic response in breast cancer cells (Figure 9C).
In conclusion, we identified that ELF3 opposes ERα functions via a novel mechanism in which ELF3 impedes receptor dimerization and thereby its transcriptional activity. A peptide derived from ELF3, ERIPE, displayed antagonistic effects on ERα functions and breast cancer cell growth. Therefore, this peptide has enormous therapeutic potential for use in breast cancer therapy.
Vijaya Narasihma Reddy Gajulapalli and Bramanandam Manavathi designed experiments, data analysed and wrote the manuscript. Vijaya Narasihma Reddy Gajulapalli, Venkata Subramanyam Kumar Samanthapudi and Saratchandra Singh Khumukcham performed experiments. Madhusudana Pulaganti, Vijaya Lakhsmi Malisetty, Lalitha Guruprasad and Suresh Kumar Chitta analysed protein–peptide docking and bioinformatic.
We acknowledge DST-PURSE, UOH-DBT-CREBB, UGC-UPE2 and DST-FIST for providing the research facilities at University of Hyderabad. Authors acknowledge Dr Prakash Prabhu, Department of Biotechnology, University of Hyderabad, for his help in CD analysis. Authors acknowledge the help of Dr Suresh Yenugu, Department of Animal Biology, and University of Hyderabad, for editing of the manuscript.
This work was supported by the Department of Biotechnology, India [grant numbers BT/PR11114/BRB/10/635/2008, BT/PR8764/MED/97/104/2013 and BT/MED/30/SP11273/2015]; the Innovative Young Biotechnologist Award, India [grant BT/01/IYBA/2009]; the Council for Scientific and Industrial Research, India [grant number 37(1359)/09/EMR-II]; and the Department of Science and Technology, Ministry of Science and Technology, Govt. of India [grant number SB/S0/BB/013/2013] to B.M.
activation function 1
DNA binding domain
epithelial-specific ETS transcription factor-1/E74 like ETS transcription factor-3
oestrogen receptor alpha
oestrogen response element
ERα-interacting peptide of ELF3
green fluorescent protein
ligand binding domain
quantitative real-time reverse transcription PCR
selective oestrogen receptor modulator