Breast cancer is the leading cause of cancer-related deaths among women worldwide, and it exhibits diverse causes and clinical features (Harvey et al. 1999; Torre et al. 2016). Therefore, accurate classification of breast cancer subtypes is essential for determining appropriate treatment strategies. Subtypes of breast cancer can be classified based on the expression of estrogen receptor (ER) in tumor tissues. Approximately 70% of all breast cancers are estrogen receptor α (ERα)-positive, indicating potential responsiveness to endocrine therapies (Harvey et al. 1999). Targeting estrogen receptors has become the most effective approach for treating ERα-positive breast cancer patients, as endogenous estrogens play a crucial role in ERα-positive breast cancer cells. Selective estrogen receptor modulators (SERMs) such as tamoxifen and raloxifene competitively inhibit ER in breast cancer cells, exhibiting either agonistic or antagonistic behavior depending on the tissue types (Jaiyesimi et al. 1995). However, prolonged exposure to tamoxifen leads to chemoresistance, resulting in increased cell proliferation, cell migration, and an epithelial-to-mesenchymal transition (EMT) phenotype (Clarke et al. 1993; Kim et al. 2009; Bui et al. 2017). EMT is a critical step driving the early phase of cancer metastasis, characterized by the loss of epithelial cell-cell adherence junctions, reorganization of the actin cytoskeleton, downregulation of cell adhesion molecules such as E-cadherin, and upregulation of mesenchymal markers including N-cadherin, Snail, and vimentin (Christiansen and Rajasekaran 2006).

The status of ERα expression serves as a valuable clinical biomarker for the diagnosis and prognosis of breast cancer (Harvey et al. 1999). The typical ERα is a 66 kDa ligand-dependent transcription factor primarily localized in the nucleus. Wang et al (2005) identified a novel isoform of ERα with a molecular weight of 36 kDa, named ERα36. ERα36 differs from the full-length ERα66 as it lacks both of the two transcriptional activation domains (AF-1 and AF-2), while retaining the DNA-binding domain and partial dimerization and ligand-binding domains (Wang et al. 2005). ERα36 possesses an additional unique 27-amino-acid sequence that replaces the last 138 amino acids at the C-terminus (Lee et al. 2008). Several studies have reported that ERα36 is highly expressed in ERα66-negative breast cancer cells such as MDA-MB-231, MDA-MB-436, and SKBR3, while being minimally detected in ERα66-positive breast cancer cells including MCF-7, H3396, and T47D (Wang et al. 2006; Shi et al. 2009; Kang et al. 2011; Zhang et al. 2012; Li et al. 2013). Moreover, increased ERα36 levels have been associated with the development of acquired tamoxifen resistance (Shi et al. 2009; Zhang et al. 2012). Our recent study demonstrated that overexpression of ERα36 is linked to enhanced cell migration in MCF-7 cells through Hippo-dependent activation of yes-associated protein 1 (YAP1) (Park et al. 2022). However, one remaining question is how hormonally responsive breast cancers progress towards more aggressive and hormonally independent phenotypes. Previous studies have observed distinct morphological features and increased invasiveness in tamoxifen-resistant MCF-7 (TAMR-MCF-7) cells compared to their parental MCF-7 cells. Additionally, several studies, including our own (Li et al. 2013; Park et al. 2022), have reported the involvement of ERα36 in the progression of epithelial-mesenchymal transition (EMT) in TAMR-MCF-7 cells.

Long non-coding RNAs (lncRNAs) are a class of non-protein-coding RNA transcripts longer than 200 nucleotides (Kapranov et al. 2007). They function as activators, decoys, guides, or scaffolds for interacting proteins, DNA, and RNA (Wang and Chang 2011). Extensive research has demonstrated the critical role of lncRNAs in regulating various biological processes, including stem cell maintenance, proliferation, apoptosis, invasion, and metastasis of cancer cells (Gupta et al. 2010; Loewer et al. 2010; Venkatraman et al. 2013). Among numerous lncRNAs associated with cancer, homeobox transcript antisense intergenic RNA (HOTAIR) is one of the most commonly overexpressed lncRNAs in breast cancer, with a length of 2.2 kb (Rinn et al. 2007). Previous studies have shown that HOTAIR interacts with the Polycomb Repressive Complex 2 (PRC2), leading to changes in the gene expression profile associated with cancer metastasis (Gupta et al. 2010; Yoon et al. 2013). Aberrant upregulation of HOTAIR has been observed in tamoxifen-resistant breast cancer, and it directly interacts with the estrogen receptor protein, enhancing its transcriptional activity (Xue et al. 2016). In this study, we unveil the critical role of HOTAIR expression in ERα36-mediated EMT and metastasis of tamoxifen-resistant breast cancer.

Antibodies and reagents

Anti-E-cadherin and anti-N-cadherin antibodies were supplied from BD Transduction (San Jose, CA). Anti-Snail antibody was purchased from Abcam (Cambridge, UK). Anti-ERα66, anti-Vimentin, anti-PCNA antibodies were obtained from Santa Cruz Biotechnology (Dallas, TX). Anti-ERα36 antibody (CY1109) was supplied from Cell Applications (San Diego, CA). Horseradish peroxidase (HRP)–linked anti-rabbit, or anti-mouse IgG were purchased from Cell Signaling Technology (Danvers, MA). Lincode control siRNA and lincode siRNA of human HOTAIR were obtained from Dhamarcon (Thermo Scientific, Waltham, MA). β-actin antibody and other reagents, including hematoxylin and eosin (H&E), were obtained from Sigma. Other reagents were obtained from Sigma (St. Louis, MO). pCMV-ER36 plasmid was kindly donated from Dr. ZhaoYi (Charlie) Wang (Creighton University Medical School, Omaha, USA).

Cell culture and establishment of tamoxifen-resistant MCF-7 cells

MCF-7 cells were cultured at 37℃ in 5% CO₂/95% air in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS), 100 units/mL penicillin, and 100 μg/mL streptomycin. TAMR-MCF-7 cells were established using the methodology reported previously (Kim et al. 2009) and the cells were cultured in phenol red-free DMEM containing 10% charcoal-stripped, steroid-depleted FBS (C/D FBS) (Hyclone, Logan, UT) and 3 μM 4-hydroxytamoxifen.

Generation of ERα66 overexpressing stable cell line

To generate the stable cell lines with ERα66 overexpression, amphotrophic retroviral supernatants were produced by transfection of the MSCV-GFP or MSCV-GFP-ER66 retroviral vectors into Phoenix packaging cells. Viral supernatants were harvested 72 h post-transfection, and viral stock was passed through a 0.45 μm filter. For each infection, 3 × 10⁴ cells of TAMR-MCF-7 were plated in six-well culture plates and incubated with viral stocks in a final volume of 2 mL (1 mL viral supernatant and 1 mL complete media). To enhance the efficiency of infection, 3 μg/mL Polybrene was added to the culture media. Remove the virus containing medium every 12 hours and repeat the infection as described above for 20 times.

Generation of ERα36 overexpressing stable cell line

MCF-7 cells were plated at a density of 1 × 10⁴ cells/24-well plate overnight, then transfected with an HA-tagged ERα36 expression vector driven by the CMV promoter or an empty expression vector using Lipofectamin 2000 (Invitrogen, Carlsbad, CA). 48 h after transfection, the transfected cells were replated and selected with G418 (800 μg/mL)-containing medium for 2 weeks. The medium was changed every three days until colonies appeared. Surviving single colonies were then picked and amplified, named as MCF7-ER36, and used for the experiments. The empty expression vector-transfected MCF-7 cells (MCF7-pcDNA3.1) were used as control.

siRNA knockdown assay

Lincode control siRNA and lincode siRNA of human HOTAIR were obtained from Dhamarcon and used to knockdown long non-coding RNA HOTAIR expression in target cells. Cells were grown in six-well dish, then transfected with lincode siRNA using Lipofectamine 2000 (Thermo Scientific, Waltham, MA) for 72 h.

Immunoblot analysis

Cells were washed with cold phosphate-buffered saline (PBS) and lysed in lysis buffer (150 mM Tris-Cl [pH.7.6], 10% NP-40, protease inhibitors and phosphatase inhibitors) and centrifuged at 16,000 g, 4℃. The fractionated proteins were separated in sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) and transferred to nitrocellulose membrane (Kim et al. 2023). The proteins were immunoblotted with specific primary and corresponding peroxidase-conjugated secondary antibodies.

Quantitative polymerase chain reaction (qPCR)

Total RNA was extracted using Trizol reagent and treated with RNase-free DNase (Invitrogen). RNA-seq was performed by Macrogen, Inc. (Seoul, South Korea). For qPCR, cDNA was synthesized by reverse transcriptase kit (iNtRON, Seoul, South Korea) and qPCR was performed using specific primers as described below. The SYBR Green qPCR amplification was conducted with MiniOpticon real time PCR detection system (Bio-Rad laboratories Inc., Hercules, CA). Relative levels were calculated using the comparative CT method. Data were normalized to 18S.

Primers for real time PCR:

HOTAIR-F: GGTAGAAAAAGCAACCACGAAGC

HOTAIR-R: ACATAAACCTCTGTCTGTGAGTGCC

18S-F: AGGATCCATTGGAGGGCAAGT

18S-R: TCCAACTACGAGCTTTTTAACTGCA

Transwell migration assay

An in vitro cell migration assay was performed using a 24-well trans-well polystyrene membrane with 8 μm size pores (3422; Corning, Cambridge, MA) as described by previous report (Bui et al. 2017). Briefly, cells with serum-free media were seeded in the upper chamber of the trans-well plate and the lower chamber was filled with 600 μL serum-containing culture media. For gene silencing, cells were transiently transfected with Lincode control siRNA or Lincode HOTAIR siRNA for 48 hours before. Migrated cells to the lower filter side were analyzed.

Kaplan-Meier analysis for online survival calculation

Kaplan–Meier analysis of breast cancer (kmplot.com/analysis/) was performed as described previously (Gyorffy et al. 2012).

Immunocytochemistry

Cells were seeded in 10% complete media on coverslips for 24 h. Coverslips were fixed in 4% paraformaldehyde for 10 minutes at room temperature and then washed twice with 1 × PBS. Following washes, samples were permeabilized for 10 minutes in 0.1% Triton X-100, then washed with PBS. Samples were blocked with 10% horse serum (Invitrogen) for 1 hour at room temperature. Slides were incubated overnight at 4℃ with primary antibody. Samples were washed with PBS and incubated with secondary antibody for 1 hour at room temperature. For F-actin staining to visualize the cytoskeleton, samples were incubated with Alexa Fluor 546 phalloidin (Invitrogen) for 30 min. Slides were washed as above and stained with DAPI for 10 min at room temperature. Coverslips were mounted stored until imaging in the dark at 4℃. Images were acquired under iRiS™ Digital Cell Imaging System (Logos Biosystems, Annandale, VA).

RNA Binding Protein Immunoprecipitation (RIP) assay

RNA immunoprecipitation assay was performed by using the Magna RIP™ RNA-Binding Protein Immunoprecipitation Kit (Millipore, MA, USA) following the manufacturer’s protocol. Breast cancer cells were scraped off, then lysed in complete RIP lysis buffer, after which 100 μL of whole cell extract was incubated with RIP buffer containing magnetic beads conjugated with ERα36 antibody (CHI scientific, Maynard, MA), ERα66 antibody (Santa Cruz Biotechnology) and negative control Normal Rabbit IgG (Millipore). Samples were incubated with Proteinase K with shaking to digest the protein and then immunoprecipitated RNA was isolated. Purified RNA was then subjected to qPCR analysis to demonstrate the presence of the binding targets using respective primers.

Reporter gene assay

Cells were plated in phenol red-free DMEM containing 10% C/D FBS overnight, then transiently transfected with 1 μg of ERE-Luc reporter plasmid and 100 ng of phRL-SV40 plasmid (hRenilla luciferase expression for normalization) (Promega, Madison, WI) using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). After 6 hours, the medium was replaced with phenol red-free medium supplemented with 1% C/D FBS, and 10 nM 17β-estradiol and then incubated for 18 hours. A dual-luciferase reporter assay system (Promega, Madison, WI) was used to determine the promoter activity. Both firefly and Renilla luciferase activities in the cell lysates were measured using a luminometer (Tristar LB 941, Berthold Tech., Bad Wildbad, Germany). The relative luciferase activity was calculated by normalizing the promoter-driven firefly luciferase activity to that of hRenilla luciferase.

Intrasplenic injection model of liver metastases

Intrasplenic injection model of liver metastases were established as the methodology reported previously (Soares et al. 2014). Briefly, 5-week-old BALB/c athymic nude mice (Raon Bio Inc.) were inoculated with 1 × 10⁶ cells into spleen. After 5 weeks, the mice were sacrificed and the liver samples were analyzed by hematoxylin & eosin (H&E) staining method. All animal studies were performed according to the regulation and an approval of Seoul National University Institutional Animal Care and Use Committee (Approve # SNU-170618-1-2).

Histology

The liver was harvested immediately after sacrifice and fixed in 10% formalin. Fixed samples were embedded in paraffin, and sections of 3 μm were stained with hematoxylin and eosin (H&E) for microscopic analysis. For the histological examination, all of the metastatic lesions were observed by HistoQuant (3D HISTECH, Budapest, Hungary).

Data analysis

Data are presented as mean ± S.D. or S.E. Student’s t-test was used to analyze differences between experimental groups. Values of p < 0.05 or p < 0.01 or p < 0.005 were considered statistically significant.

Association of higher ERα66 gene expression with improved survival in tamoxifen-treated breast cancer patients

One potential mechanism contributing to the development of antiestrogen resistance is the downregulation of ERα66 (Melchor et al. 1990; Zhao et al. 2011; Li et al. 2013). To investigate the association between ERα66 gene expression and tamoxifen response, we analyzed a large cohort of ERα66-positive breast cancer patients from publicly available data (Gyorffy et al. 2012). Kaplan-Meier analysis demonstrated a statistically significant correlation between high ERα66 expression and improved relapse-free survival (RFS, p = 0.04) and distant metastasis-free survival (DMFS, p = 0.04) specifically in response to tamoxifen treatment (Fig. 1A, upper). Several studies have reported high expression of ERα36 in tamoxifen-resistant breast cancer (Shi et al. 2009; Li et al. 2013). Although not statistically significant, Kaplan-Meier curve analysis indicated a tendency for high ERα36 expression to be associated with poorer relapse-free survival (RFS, p = 0.28) and distant metastasis-free survival (DMFS, p = 0.12) in response to tamoxifen treatment (Fig. 1A, lower).

Figure 1. Expression of ERα66 and ERα36 in MCF-7 and TAMR-MCF-7 cells. (A) Kaplan-Meier analyses for relapse-free survival (RFS) of the cohort of patients with ERα66-positivity (upper, left), receiving tamoxifen treatment only without chemotherapy. Affymetrix ID for ERα66 used was 215552_s_at. The cut-off value used in analysis was 316 and the expression range of the probe was 4–2,922. Patient number for low ERα66 (black) and high ERα66 (red) is presented under the following months. Kaplan-Meier analyses for distant metastasis-free survival (DMFS) of the cohort of patients with ERα66-positivity (upper, right), receiving tamoxifen treatment only without chemotherapy. Affymetrix ID for ERα66 used was 211234_x_at. The cut-off value used in analysis was 161 and the expression range of the probe was 7–3,924. Patient number for low ERα66 (black) and high ERα66 (red) is presented under the following months. Kaplan-Meier analyses for RFS of the cohort of patients with ERα36-positivity (lower, left), receiving tamoxifen treatment only without chemotherapy. Affymetrix ID for ERα36 used was 205767_at. The cut-off value used in analysis was 27 and the expression range of the probe was 1–8,932. Patient number for low ERα36 (black) and high ERα36 (red) is presented under the following months. Kaplan-Meier analyses for DMFS of the cohort of patients with ERα36-positivity (lower, right), receiving tamoxifen treatment only without chemotherapy. Affymetrix ID for ERα36 used was 205767_at. The cut-off value used in analysis was 26 and the expression range of the probe was 1–8,932. Patient number for low ERα36 (black) and high ERα36 (red) is presented under the following months. (B) Expression of ERα66 and ERα36 in tamoxifen-resistant (TAMR-MCF-7) cells and parental MCF-7 cells. Total cell lysates, cytosolic fraction and nuclear fraction of MCF-7 and TAMR-MCF-7 cells were subjected to western blot analyses. Actin or proliferating cell nuclear antigen (PCNA) were used as loading controls for cytoplasmic and nuclear fractions, respectively. (C) Immunocytochemistry of ERα66 and ERα36. MCF-7 and TAMR-MCF-7 cells were stained with ERα66 (upper, green) or ERα36 (lower, red). The results are shown as representative image.

Subsequently, we examined the expression and localization of ERα66 and ERα36 in parental MCF-7 and TAMR-MCF-7 cells. Consistent with previous reports (Li et al. 2013), we observed a significant downregulation of ERα66 and upregulation of ERα36 in TAMR-MCF-7 cells compared to parental MCF-7 cells (Fig. 1B). Immunoblot analysis of cytosolic and nuclear fractions revealed that ERα66 was distributed in both the cytosol and nucleus, with higher levels detected in the nucleus of MCF-7 cells (Fig. 1B). In contrast, ERα36 was predominantly present in the cytosolic fraction, with minimal amounts in the nucleus (Fig. 1B).

To confirm the subcellular localization of ERα66 and ERα36, immunocytochemistry was performed on both cell lines. Consistent with the immunoblotting results, ERα66 staining was predominantly detected in the nuclei of MCF-7 cells but was undetectable in TAMR-MCF-7 cells (Fig. 1C, upper). Conversely, ERα36 staining showed stronger intensity in the cytoplasm of TAMR-MCF-7 cells (Fig. 1C, lower).

EMT phenotypes in ERα36 overexpressing MCF-7 cells

To further investigate the role of ERα36, we established a stable MCF-7 cell line that constitutively expressed ERα36 (MCF7-ER36). Immunoblot analysis confirmed significantly increased levels of ERα36 and a sharp decrease in ERα66 expression in MCF7-ER36 cells compared to control MCF7-pcDNA3.1 cells (Fig. 2A). Phase contrast microscopy images revealed that control MCF-7 and MCF7-pcDNA3.1 cells exhibited a typical epithelial cell morphology, while TAMR-MCF-7 and MCF7-ER36 cells displayed morphological changes characterized by a spread and spindly form resembling mesenchymal cells (Fig. 2B).

Figure 2. EMT features in ERα36-overexpressing MCF-7 cells. (A) Expression of ERα66 and ERα36 in ERα36-overexpressing (MCF7-ER36) cells and control MCF7-pcDNA3.1 cells. Total cell lysates, cytosolic fraction and nuclear fraction of MCF-7 and TAMR-MCF-7 cells were subjected to western blot analyses. Actin or proliferating cell nuclear antigen (PCNA) were used as loading controls for cytoplasmic and nuclear fractions, respectively. (B) Representative figure of morphological characteristics of MCF-7 cells, TAMR-MCF-7 cells, MCF7-pcDNA3.1 cells and MCF7-ER36 cells in vitro culturing. (C) Immunofluorescence staining of cell–cell junction protein E-cadherin and the actin cytoskeleton. MCF-7 cells, TAMR-MCF-7 cells, MCF7-pcDNA3.1 cells and MCF7-ER36 cells were stained with E-cadherin (green), Phalloidin (red) as well as DAPI (blue) and pictures were taken at ×40 magnification. (D) Immunoblot analyses of EMT markers in parental MCF-7 cells, TAMR-MCF-7 cells, MCF7-pcDNA3.1 cells and MCF7-ER36 cells. Actin was used as a loading control.

These cell types were subsequently stained with rhodamine-phalloidin and E-cadherin. Phalloidin staining revealed a reorganization of actin filaments, which are essential for the regulation of cell migration, in MCF7-ER36 and TAMR-MCF-7 cells (Fig. 2C). Furthermore, E-cadherin, an epithelial junction marker, was predominantly localized at cell-cell contacts in MCF-7 and MCF7-pcDNA3.1 cells, while the intensity of E-cadherin staining was reduced in both MCF7-ER36 and TAMR-MCF-7 cells (Fig. 2C), indicating the presence of EMT phenotypes in ERα36 overexpressing cell types. Western blot analyses confirmed a significant upregulation of typical EMT markers such as N-cadherin, Snail, and Vimentin, accompanied by a downregulation of E-cadherin in both MCF7-ER36 and TAMR-MCF-7 cells (Fig. 2D). These findings suggest that the expression of ERα36 is involved in the progression of EMT in ER-positive breast cancer cells.

Migration and metastasis by ERα36 overexpression in MCF-7 cells

We previously demonstrated that TAMR-MCF-7 and ERα36 overexpressing cells exhibit typical EMT phenotypes and a high capacity for migration (Kim et al. 2009; Bui et al. 2017; Park et al. 2022). In this study, we confirmed that TAMR-MCF-7 and MCF7-ER36 cells possessed significantly higher in vitro migratory ability compared to parental MCF-7 or MCF7-pcDNA3.1 cells, as determined by transwell migration assays (Fig. 3A). Furthermore, we established an intrasplenic injection model of liver metastasis to examine whether ERα36-overexpressing MCF-7 cells induce liver metastasis in vivo. Five-week-old BALB/c athymic nude mice were divided into four groups: mice bearing control MCF-7 cells (group 1), mice bearing TAMR-MCF-7 cells (group 2), mice bearing MCF7-pcDNA3.1 cells (group 3), and mice bearing MCF7-ER36 cells (group 4) (Fig. 3B). There was no significant difference in body weight among these four groups. Neither mice implanted with MCF-7 nor MCF7-pcDNA3.1 cells exhibited tumor formation in the spleens or liver metastases (Fig. 3B). However, mice implanted with either TAMR-MCF-7 or MCF7-ER36 cells developed aggressive tumor formation in the spleen and showed an increased susceptibility to macroscopic metastases on the liver surface (Fig. 3B). To assess the metastatic tumor burden, liver sections were stained with H&E. As shown in Fig. 3C, liver tissues in group 2 and group 4 exhibited a severe micrometastatic hepatic tumor burden, whereas none were observed in group 1 and group 3 (Fig. 3C). Four out of five (80%) mice bearing TAMR-MCF-7 cells and three out of five (60%) mice bearing MCF7-ER36 cells developed metastases to the liver (Fig. 3D). These results suggest that EMT-derived metastases in tamoxifen-resistant breast cancer may result from ERα36 overexpression.

Figure 3. Enhanced cell migration and metastasis by ERα36 overexpression in MCF-7 cells. (A) Transwell migration assays demonstrating the increase on migratory ability of tamoxifen-resistant and ERα36-overexpressing MCF-7 cells compared to control cells. Representative microscopy (×20) images of MCF-7 cells, TAMR-MCF-7 cells, MCF7-pcDNA3.1 cells and MCF7-ER36 cells (left). The average number of migrated cells per field among different experimental groups (right). Data represent the mean ± SD of three replicates (^##p < 0.01, significant difference versus parental MCF-7; **p < 0.01, significant difference versus control MCF7-pcDNA3.1). (B) Intrasplenic injection model of liver metastases was performed (upper). Representative images of spleens, livers and macroscopic metastases identified on the surface of livers (lower). (C) Representative images of H&E-stained liver sections from each group. Darker purple region showing metastasized tumor. (D) Incidence of liver metastasis. 80% (4/5) and 60% (3/5) of mice-bearing TAMR-MCF-7 and MCF7-ER36 cells with developed liver metastases, respectively. None of mice-bearing MCF-7 or MCF7-pcDNA3.1 resulted in tumor formation as well as metastasis.

Involvement of the long non-coding RNA HOTAIR in EMT

The long non-coding RNA (lncRNA) HOTAIR has been identified as a potent predictor of metastasis and drug resistance and is associated with EMT (Tang et al. 2013; Wu et al. 2014). Previous studies have demonstrated that HOTAIR is involved in conferring tamoxifen resistance to MCF-7 cells (Rinn et al. 2007; Yoon et al. 2013). Furthermore, the HOTAIR promoter contains multiple estrogen response elements (EREs) for binding of ER (Bhan and Mandal 2015; Wu et al. 2015).

To measure the fold changes in HOTAIR expression, we performed qPCR analysis. As expected, HOTAIR was highly upregulated in both TAMR-MCF-7 and ERα36-overexpressing MCF-7 cells (Fig. 4A). To further validate the interaction between HOTAIR and ERα, we performed RIP assays on extracts from four different breast cancer cell lines using antibodies against ERα36 or ERα66. RNA levels in the immunoprecipitates were determined by qPCR. Consistent with a previous report (Yoon et al. 2013), we found that ERα66 antibody-driven immunoprecipitates contained significant amounts of HOTAIR, particularly in ERα66-positive breast cancer MCF-7 and MCF7-pcDNA3.1 cells. Similarly, HOTAIR was preferentially enriched by ERα36-immunoprecipitates in ERα66-negative breast cancer TAMR-MCF-7 and MCF7-ER36 cells (Fig. 4B). Additionally, upregulation of ERα66 or downregulation of ERα36 in TAMR-MCF-7 cells resulted in a reduction of HOTAIR levels (Fig. 4C). These results suggest that HOTAIR expression is directly mediated by both ERα36 and ERα66 isoforms.

Figure 4. Binding of ERα with long non-coding RNA HOTAIR and its roles in EMT in breast cancer cells. (A) Relative expression of lncRNA HOTAIR in TAMR-MCF-7 vs. MCF-7 cells, MCF7-ER36 vs. MCF7-pcDNA3.1 cells as analyzed by qRT-PCR. Data represent mean ± SD of 5 seperate samples (***p < 0.005, control = 1). (B) ERα protein binding with lncRNA HOTAIR. Cells were subjected to RIP assay using an anti-ER antibody or IgG control. IP-enriched RNA was then analyzed by qPCR. Left panel, RIP with anti-ERα66 antibody, anti-ERα36 antibody, preimmune IgG or 10% input from MCF-7 and TAMR-MCF-7 cell extracts. Right panel, RIP with anti-ERα66 antibody, anti-ERα36 antibody, preimmune IgG or 10% input from MCF7-pcDNA3.1 and MCF7-ER36 cell extracts. Data represent mean ± SD of 3 separate samples (***p < 0.005 significant as compared to 10% Input of control cells; ^#p < 0.05, ^###p < 0.005 significant as compared to control cell lysates incubated with anti-ERα36 antibody; ^$p < 0.05, ^$$$p < 0.005 significant as compared to control cell lysates incubated with anti-ERα66 antibody). (C) Expression fold changes of lncRNA HOTAIR following knockdown of ERα36 (left) or overexpression of ERα66 (right) in tamoxifen-resistant MCF-7 cells. Data represent mean ± SD of 5 separate samples (*p < 0.05 significant as compared to TAMR-mock cells; ***p < 0.001 significant as compared to control TAMR-GFP cells). (D) Western blot analysis of ERα36 and ERα66 protein levels following treatment of cells with lincode control siRNA (siControl) and lincode siRNA of human HOTAIR (siHOTAIR). GAPDH was used as control. (E) Immunoblot analyses of EMT markers in TAMR-MCF-7 cells and MCF7-ER36 cells transfected with siControl and siHOTAIR. GAPDH was used as control. (F) Transwell migration assays demonstrating the reduction on migratory ability in TAMR-MCF-7 and MCF7-ER36 cells transfected with siHOTAIR compared to cells transfected with siControl. Representative microscopy (×20) images of these cells (left). The average migrated-cell number per field in different experimental groups (right). Data represent the mean ± SD of three replicates (***p < 0.005, significant difference versus cells transfected with siControl).

To gain further insight into which forms of ERα are controlled by HOTAIR, we introduced HOTAIR siRNA to silence HOTAIR expression. We observed that HOTAIR knockdown downregulated the ERα36 protein level, whereas ERα66 expression did not show any significant difference (Fig. 4D). Next, to investigate whether downregulation of HOTAIR affects the EMT phenotype, we assessed several EMT markers. Immunoblot results revealed the upregulation of the epithelial molecule E-cadherin and the downregulation of the mesenchymal marker Vimentin following silencing of the lncRNA HOTAIR (Fig. 4E). Knockdown of HOTAIR lncRNA also significantly suppressed the migratory capability of TAMR-MCF-7 and MCF7-ER36 cells (Fig. 4F). These data suggest that HOTAIR participates in the maintenance of the mesenchymal phenotype in breast cancer cells.

It has been reported that critical milestones in the phenotypic modification of ERα-positive breast tumors consist of the loss of hormone-dependency and increased metastatic potential (Harvey et al. 1999; Kim et al. 2009). This study demonstrated the upregulation of ERα36 and loss of expression of ERα66 in tamoxifen-resistant MCF-7 cells. Furthermore, ERα36 was predominantly distributed in the cytoplasm of TAMR-MCF-7 cells, while ERα66 was mainly expressed in the nucleus of MCF-7 cells. These results are consistent with a previous report showing that ERα36-expressing HEK-293 cells exhibited 50% of ERα36 fractionates with the plasma membrane, 40% with the cytosol, and 10% with nuclei (Wang et al. 2006), and suggest a possible regulatory interaction between the two types of ERα. Several studies revealed that ectopic expression of ERα36 reduced mRNA transcripts of ERα66 (Lin et al. 2010; Li et al. 2013). Because ERα36 lacking the transcription activator domain is mainly localized in intracellular membrane with the interaction with chaperone gp96 (Hou et al. 2015), ERα36 can stimulate diverse signaling pathways including epidermal growth factor receptor, mitogen-activated protein kinase, and phosphatidylinositol 3-kinase/Akt (Wang et al. 2018). Hence, it is possible that ERα36-mediated non-genomic pathway activation may affect the expression of ERα36 protein.

EMT is indicated as a phenotypic conversion linked with invasion and metastasis (Christiansen and Rajasekaran 2006). A hallmark of EMT is losing expression of E-cadherin, a key cell-cell coherence molecule, which is recognized as a caretaker of the epithelial phenotype (Kalluri and Weinberg 2009). Accumulating evidence indicates the association of ERα status with the EMT phenotype in tamoxifen-resistant breast cancer cells (Shi et al. 2009; Zhao et al. 2011; Li et al. 2013; Zhang and Wang 2013; Wu et al. 2014). Loss of ERα66 in ERα36-overexpressing breast cancer cells results in EMT progression characterized by remarkable changes in the expression profile of EMT markers and the reorganization of F-actin (Wu et al. 2014). Consistent with these findings, our study demonstrated the downregulated expression of ERα66, subsequently markedly induced cellular phenotypic changes accompanied by the downregulation of E-cadherin and elevation of several mesenchymal markers in both tamoxifen-resistant MCF-7 and ERα36-overexpressing MCF-7 cells. Consequently, these cells also possessed higher metastatic potential compared to their parental cell types in vitro and in vivo. ERα66 was reported to suppress the expression of the nuclear transcription factor Snail, a negative transcription factor of E-cadherin gene expression (Fearon 2003), which implies that ERα66 may functionally play a crucial role in maintaining epithelial features of breast cancer cells. Hence, it is possible that the functions of ERα36 on EMT-derived metastases are dependent on the defect in ERα66 transactivation.

Increasing evidence points to the long non-coding RNA (lncRNA) HOTAIR as a key regulatory factor in the molecular mechanisms underlying the development and progression of cancer (Woo and Kingston 2007), as well as cancer invasiveness and metastasis (Loewer et al. 2010; Liu et al. 2013). Previously, HOTAIR was shown to be directly regulated by ERα, and its upregulation promotes ligand-independent ER activation, along with conferring tamoxifen resistance (Yoon et al. 2013). The HOTAIR promoter contains a variety of transcription factor binding sites, including ERE, Sp1 binding site, hypoxia response element, and AP-1 binding site, among others, indicating the complicated regulation mechanism of HOTAIR (Bhan et al. 2014). Our study was in accordance with these findings, showing a significant elevation of HOTAIR levels in tamoxifen-resistant MCF-7 cells. We also observed a significant increase in HOTAIR expression in ERα36-overexpressing MCF-7 cells, which have low endogenous levels of ERα66 and high levels of ERα36. Considering our data showing that ERα36 knockdown or ERα66 overexpression reduces HOTAIR expression in TAMR-MCF-7 cells, ERα36 would be a more potent activator of HOTAIR gene expression compared to ERα66. It has been reported that G protein-coupled estrogen receptor 1 (GPER), stimulated by 4-hydroxytamoxifen, results in the upregulation of HOTAIR via the suppression of miR148a (Tao et al. 2015). Moreover, GPER physically interacting with ERα36 acts as a coregulator in nuclear factor-κB-mediated gene transcription (Pelekanou et al. 2016). Hence, ERα36-mediated regulation of HOTAIR expression could be under the control of the GPER pathway.

Our study further demonstrated that both ERα isoforms, ERα66 and ERα36, can interact with HOTAIR, and that HOTAIR selectively controls the expression of ERα36. HOTAIR directly interacts with chromatin-modifying proteins, such as the enzymatic subunit of polycomb repressive complex 2 (PRC2) and lysine-specific demethylase 1A (LSD1), and recruits them to the target gene loci to suppress their transcription via H3K27 trimethylation (PRC2 activity) and H3K4 demethylation (LSD1 activity) (Kaneko et al. 2010; Tsai et al. 2010). Previous studies have shown that HOTAIR negatively regulates E-cadherin gene expression by interacting with PRC2 (Venkatraman et al. 2013; Bhan et al. 2014). Consistent with those findings, we demonstrated that the suppression of HOTAIR in TAMR-MCF-7 cells and ERα36-overexpressing MCF-7 cells led to the reversal of EMT and inhibition of their migratory capability. In summary, our work highlights the reciprocal roles of ERα36 and HOTAIR in the EMT-derived metastasis of tamoxifen-resistant breast cancer.

The authors declare that they have no conflict of interest.

This research was funded by the National Research Foundation of Korea (NRF) grant (NRF-2021R1A4A1021787).