Tamoxifen reverses epithelial–mesenchymal transition by demethylating miR-200c in triple-negative breast cancer cells

Human breast cancer cell lines MCF-7 (TCHu74) and MDA-MB-231 (TCHu227) were obtained from the Cell Bank of the Chinese Academy of Sciences (China). MCF-7/ADR cells derived from MCF-7 and cultured with 1 μg/mL adriamycin for at least 1 year and possessed adriamycin-resistance [13]. Human breast cancer cell lines MCF-7 and MCF-7/ADR cells were cultured in RPMI 1640 medium (GIBCO, Grand Island, NY), MDA-MB-231 cells were cultured in Leibovitz’s L-15 medium (GIBCO, Grand Island, NY, USA). Both RPMI 1640 and L15 medium were supplemented with 10% fetal calf serum (Gibco by Life Technologies, Cergy Pontoise, France), and penicillin (100 U/mL) and streptomycin (100 μg/mL). For MCF-7 cells, the medium additionally contained human-recombinant insulin (10 μg/mL) to maintain the endocrine dependency. For MCF-7/ADR cells, the medium additionally contained 1 μg/mL adriamycin to maintain the drug resistance property. The cells were cultured at 37 °C in a humidified atmosphere under 5% CO₂. The cells were subcultured every 2–4 days and harvested in the logarithmic phase of growth.

Tamoxifen (4-hydroxytamoxifen, 4-OHT/TAM) and 5-aza (5-aza-2′-deoxycytidine, 5-aza-dC) were purchased from Sigma-Aldrich (St Louis, Missouri, USA). Indicated cells were treated with 5 μmol/L TAM for 48 h. E-cadherin, vimentin, ER-α, PR, and Her-2 antibodies were purchased from Cell Signaling Technology (Beverly, MA, USA). Actin, p-gp, DNMT1 and DNMT3a antibodies were purchased from Santa Cruz Biotechnology (USA).

The effects of different agents on cell proliferation were measured using the MTT assay [14]. Briefly, indicated cells were seeded at 3 × 10⁴ per well in 96-well plates in quadruplicate and incubated overnight. Different concentrations of the test agents were then added and incubated for 5 days. Thereafter, 25 μL of MTT solution (5 mg/mL) was added to each well and the cells were incubated for another 4 h at 37 °C. After the incubation, the supernatants were removed carefully, and 200 μL of DMSO was added to each well. The cells were then lightly shaked for 10 min. Absorbance was measured at 570 nm in a Microplate Reader (Bio-Rad, CA, USA). Analysis of the obtained results was done using GraphPad Prism 5 computer program to evaluate cell proliferation rate and cytostatic rate. Untreated cells were used as controls.

Cells were washed in serum-free medium twice. The chemoinvasion assay was conducted using the 24-well chemotaxis chambers with 8 μm pores (Corning, NY, USA) according to the manufacturer’s instruction. Briefly, indicated cells were pretreated with or without TAM (5 μmol/L) for 48 h. Then, 2 × 10⁴cells were resuspended in fresh serum-free media and seeded into the upper chamber of a 24-well plate, while the lower chamber contained fresh culture media with 10% FBS as a chemo-attractant. The cells were allowed to invade for 24 h at 37 °C and the chambers were then washed with PBS. The cells on lower surface of the chamber were stained with 0.1% Giemsa stain solution for 2 h and counted in four different random fields at ×10 magnifications under electron microscope. Each experiment was performed at least three times.

Cells were washed twice with ice-cold PBS and solubilized in 1% Triton lysis buffer (50 mmol/L Tris-HCl, pH 7.4, 10 mmol/L EDTA, 100 mmol/L NaF, 150 mmol/L NaCl, 1% Triton X-100, 1 mmol/L PMSF 1 mmol/L Na₃VO₄, and 2 μg/mL aprotinin) on ice, then sonication and incubation at 4 °C for 30 min, followed by centrifugation at 12,000 g at 4 °C for 20 min. Then proteins were quantified according to BCA (Beyotime, China) method. Total proteins were subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and electronically transferred to nitrocellulose membranes (Immobilon-P, Millipore, Bredford, MA, USA). After blocking with 5% skim milk in TBST (10 mmol/L Tris, pH 7.4, 150 mmol/L NaCl and 0.1% Tween-20) for 1 h, the bands were incubated in the indicated primary antibodies at 4 °C overnight, followed by secondary antibodies incubated for 30 min at room temperature. After washing with TBST, the proteins were detected using an enhanced chemiluminescence reagent (SuperSignal Western Pico Chemiluminescent Substrate, Pierce, USA) and visualized with an ECL detection system (DNR Bio-Imaging Systems, Jerusalem, Israel) [15].

The cells were seeded in Lab-Tek chamber slides (Nunc S/A, Polylabo, Strasbourg, France). The cells were treated with or without TAM (5 μmol/L) for 72 h and fixed with 4% paraformaldehyde for 15 min, permeabilized with 0.2% Triton X-100 for 5 min, blocked with 5% bovine serum albumin (BSA) in 1× PBS for 1 h at room temperature and then incubated with E-cadherin and vimentin antibodies for 1 h. Then Alexa Fluor 546-conjugated goat anti-rabbit IgG or Alexa Fluor 488-conjugated goat anti-rabbit IgG (Molecular Probes) were added in blocking solution for 1 h at room temperature in the dark. 4′6-diamidino-2- phenylindole was used to stain nuclei for 5 min. After mounted with the Slow Fade Antifade Kit (Molecular Probes, Eugene, OR, USA), the cells were visualized by fluorescence microscopy (BX61, Olympus, Japan) [16].

The cells were cultured and harvested at the indicated times. Total RNA was extracted from cells using the RNeasy mini kit (Qiagen, Carlsbad, CA, USA). For miRNAs, The One Step PrimeScript^® miRNA cDNA Synthesis Kit (Takara, Japan) was used for RNA reverse transcription. Relative expression of miRNAs was calculated via the comparative cycle threshold (Ct) method, and the expression of small nuclear RNA U6 was used as reference. The sequence-specific forward primers for mature miR-200c was: 5′-ACACTCCAGCTGGGTAATACTGCCGGGTAA-3′ and for U6 internal control was forward (5′-GCTTCGGCAGCACATATACTAAAAT-3′) and reverse (5′-CGCTTCACGAATTTGCGTGTCAT-3′), respectively. The Uni-miR qPCR Primer was included in the kit. SYBR^® Premix Ex Taq™ II (Perfect Real Time) (Takara, Japan) was used for monitoring the amount of miRNA. The PCR conditions were 30s at 95 °C, followed by 45 cycles at 95 °C for 5 s and 58 °C for 25 s. The relative amount of the target RNA was calculating by 2^-ΔΔCt method. The detailed method was described in our previous studies [16].

MCF-7 and MCF-7/ADR cells were cultured without insulin or adriamycin for 3 days. The expression levels of miRNAs were quantified using GeneChip miRNA Array (Affymetrix, Santa Clara, CA, USA) according to the manufacturer’s instructions by Gene Tech Biotechnology Company (Shanghai, China) [16]. In brief, total RNA (1 μg) was extracted with miRNeasy Mini Kit (Qiagen, Germany) and labeled with a FlashTag Biotin RNA Labeling kit (Genisphere, Hatfield, PA, USA). Then the labeled RNA was injected onto the microarrays and incubated at 48 °C for 16 h. After washing and staining, the signals were obtained using a GeneChip Scanner 3000 7G (Affymetrix, Santa Clara, CA). Data was normalized using the RMA algorithm. The PCA and unsupervised clustering of microarray analysis were shown in Additional files 1 and 2. QC report and raw data of microarray analysis were shown in Additional files 3 and 4. The microarray result was uploaded to GEO database. Data from this microarray is available at GSE96821.

The si-DNMT1 and si-DNMT3a and corresponding negative control were designed and synthesized by RiboBio (Guangzhou, China) and stored at −80 °C before use. Cells were transiently transfected with si-DNMT1 and si-DNMT3a and corresponding negative control using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer’s protocol.

5-week-old female Balb/c nude mice were from SLAC Laboratory Animal Co., Ltd. (Shanghai, China). Mice were randomly allocated to two groups (n = 6 per group). 1 × 10⁶ MDA-MB-231 cells were injected into the tail vein of the control group. 1 × 10⁶ MDA-MB-231 T cells (MDA-MB-231 cells pretreated with 5 μmol/l TAM for 5 d) were injected into the tail vein of the treatment group and given with TAM orally every other day for 2 months. Mice were killed by cervical dislocation according to the protocol filed with the Guidance of Institutional Animal Care and Use Committee of China Medical University, and lung tissue was taken out for HE staining to ensure metastasis focals. Experimental research on mice complied with the Guidance of Institutional Animal Care and Use Committee of China Medical University, and had been approved by the ethics committee of China Medical University.

The CpG island was predicted by the MethPrimer software. MethPrimer accepts a DNA sequence as input, performs a digital bisulfite conversion of the input sequence, and then picks primers on the converted sequence. Results of primer selection are delivered through a Web browser in text and graphic views (http://​www.​urogene.​org/​cgi-bin/​methprimer/​methprimer.​cgi) [17]. The miR-200c promoter region used an island size of 3000 nucleotides, a GC percentage of at least 50% and an observation/expectation CpG ratio of more than 0.6.

All the presented data were expressed as the mean ± SD and the representative results were confirmed in at least three independent experiments. Statistical comparisons were calculated by Student’s t-test. P < 0.05 was considered statistically significant. IC₅₀ values were calculated by nonlinear regression analysis using GraphPad Prism 5 software.

Usually, TNBC cells are divided into two types: cells that were initially TNBC and those that transformed from HR⁺ breast cancer cells [18, 19]. Here we used both kinds of TNBC cells to fully illustrate the character of TNBC [20, 21], with HR⁺ breast cancer cell MCF-7 taken as control. To confirm the character of breast cancer cells that we use, the Her2, PR, ER-α expression were determined in three different breast cancer cell lines. MCF-7 cells are ER⁺/PR⁺/Her2⁻ with epithelial morphology (Fig. 1a, b); MCF-7/ADR and MDA-MB-231 cells are TNBC (ER⁻/PR⁻/Her2⁻) with mesenchymal morphology (Fig. 1b). We also examined the EMT markers, E-cadherin and vimentin (Fig. 1c). We then examined the migration ability of the three breast cancer cell lines. The mesenchymal TNBC cells were more likely to metastasize than the epithelial TNBC cells and non-TNBC cells, (P < 0.05; Fig. 1d), which was facilitated by the EMT process.

MCF-7, MCF-7/ADR and MDA-MB-231 breast cancer cell lines were cultured overnight and treated with TAM (0 [control], 1, or 5 μmol/L) for 48 h. Dramatic morphological changes were observed in mesenchymal TNBC cells MCF-7/ADR and MDA-MB-231 at 5 μmol/L (Fig. 2a), which transformed from characteristic spindle-shaped mesenchymal cells to epithelial cells with tight junctions. However, this morphological change was less obvious in cells treated with 1 μmol/L TAM. The MCF-7 cells retained their epithelial character after treated with TAM (Additional file 5). Therefore, we selected 5 μmol/L TAM as our treatment concentration (Fig. 2a).

These results were also confirmed by detection of EMT protein markers and immunofluorescence tests. Protein samples from MDA-MB-231 and MCF-7/ADR cells treated with TAM showed increased E-cadherin and decreased vimentin compared to control group (Fig. 2b, c). TAM also dramatically decreased the cell migration ability to 59.4 ± 3.8% (MDA-MB-231) and 43.2 ± 5.7% (MCF-7/ADR cells) compared to that of the control group (Fig. 2d).

We also used a murine model to test the effect of TAM on migration ability of mesenchymal TNBC cells. MDA-MB-231 cells were pretreated with TAM for 5 days, and then injected into tail vein of mice, which also received TAM by gavage every other day. The mice were sacrificed at 8 weeks, and their lungs were removed, sectioned into pathological samples, and stained with hematoxylin-eosin (HE) staining to visualize metastases (Fig. 2e). The rate of migration to lungs of treated mice was 66.7%, compared with 100% of the control group (Fig. 2e). These data thus imply that TAM decreased migration ability by reversing EMT process of TNBC cells.

Given that EMT affects chemosensitivity of breast cancer [17], we hypothesized that reversing EMT in mesenchymal TNBC cells might increase their sensitivity to chemotherapeutic agents. We evaluated the sensitivity of commonly used drugs, epirubicin and 5-Fluorouracil (5-Fu), in breast cancer cells by MTT assay, and found that after 5 days’ pre-treatment with TAM, sensitivity to epirubicin increased significantly in MDA-MB-231 cells (Fig. 3a), with the IC₅₀ of 12.05 μmol/l, whereas the IC₅₀ of the control group could not be reached due to chemoresistance. Similarly, sensitivity to 5-Fu increased in MDA-MB-231 cells (IC₅₀: 220.4 μmol/l; Fig. 3b). These results thus indicate that TAM increased the sensitivity of mesenchymal TNBC cells.

To determine the effect of TAM on miRNAs that mediate regulation of EMT in TNBC cells, we compared miRNA microarray analysis of TNBC cells (MCF-7/ADR) with non-TNBC cells (MCF-7). After normalization, we found significantly different miRNA expression between the two cell lines, using Volcano Plot filtering of 400 miRNAs (Fig. 4a). Consistent with a recent report [22], miR-200c expression, which is commonly acknowledged to affect EMT, showed dramatic difference between the two cell lines. Relative fold change was 128.6-fold in MCF-7 cells in the expression of miR-200c compared with MCF-7/ADR cells (Fig. 4b). To verify the result of microarray analysis, we performed quantitative real-time PCR for the two cell lines. Consistent with the microarray analysis result, the relative fold change was 34.508-fold in MCF-7 cells in the expression of miR-200c compared with MCF-7/ADR cells (Fig. 4c). However, after treatment with 5 μmol/L TAM for 48 h, miR-200c expression restored in the mesenchymal MCF-7ADR cells; this treatment did not affect miR-200c expression in non-TNBC cells MCF-7 (Fig. 4d). These results suggested that TAM could induce increased miR-200c in mesenchymal TNBC cells.

We hypothesized that hypermethylation of the miR-200c promoter region might have caused low miR-200c expression in mesenchymal TNBC cells. This mechanism of miRNA down-regulation is closely related to DNA hypermethylation in breast cancer [23]. As miRNA expression is affected by methylation of CpG islands in their promoter regions, we used an online CpG island predictor to confirm the hypothesis [24]. Sequence analysis of miR-200c promoters revealed the presence of a highly-expressed CpG island upstream of the miR-200c cluster (blue, Fig. 5a). This prediction indicated that miR-200c expression might be affected by demethylation of its promoters. To evaluate the demethylation effect on miR-200c expression, we examined miR-200c expression after treatment with 5-azacytidine (5-AZA), a specific inhibitor of DNA methylation, for 72 h. Quantitative real-time PCR revealed that the relative fold change of miR-200c was 2.23 ± 0.67 -fold after treatment with 5-AZA for 3 days, compared with negative control, using MDA-MB-231 cells (P < 0.05; Fig. 5d). Then EMT phenotypic change was detected by immunofluorescence. Expression of E-cadherin increased and vimentin decreased in MDA-MB-231 cells (Fig. 5b) accompanied with decreased migration ability. Transwell assay also showed that after pretreatment of 5-AZA, cell migration ability decreased to 39 ± 3.57% compared to that of control, in MDA-MB-231 cells. (Figure 5c). These results suggested that 5-AZA also reversed EMT via increased miR-200c expression, and decreased migration ability in TNBC cells.

Our data demonstrated that TAM could change the morphology of mesenchymal TNBC cells, and reverse the EMT process by regulating the expression of miR-200c. A previous study showed that miRNA expression was tightly controlled by DNA methylation [25]. Therefore, we use western blot method to determine the expression of DNA-methyltransferase (DNMT, an essential enzyme in the process of DNA methylation), to evaluate DNA methylation levels. Methylation status differs among breast cancer cell lines; mesenchymal TNBC cells showed higher methylation status than MCF-7 cells (Fig. 6a). We discovered that the DNMT expression in mesenchymal TNBC cells decreased after 48 h’ treatment with TAM (Fig. 6b). Western blot revealed significant decreased DNMT1 and DNMT3a in MDA-MB-231 and MCF-7/ADR cells. To confirm that miR-200c is regulated by DNMTs, we used siRNAs to double-knock DNMT1 and DNMT3a. Compared with scrambled siRNA control cells, knockdown of DNMT1 and DNMT3a significantly increased the miR-200c expression (Fig. 6c), accompanying with decreased vimentin and increased E-cadherin in MDA-MB-231 and MCF-7/ADR cells (Fig. 6d). Together, these data suggested that TAM increased miR-200c by partially inhibition of DNMT1 and DNMT3a to reverse EMT in mesenchymal TNBC cells.