Fresh breast cancer tissues of invasive ductal carcinoma (stage II) were obtained from the Tissue Banking Facility Jointly Supported by TMUCIH (Tianjin Medical University Cancer Institute and Hospital) & NFCR (National Foundation for Cancer Research). The pathological stage and nodal status were obtained from the primary pathology reports. The patients (8 ER positive and 8 ER negative) had a mean age of 52.8 ± 12.1 years and were recruited in the same department. This study was approved by the institutional ethics committee.

MCF-7 and MDA-MB-231 cells were maintained in DMEM-high glucose medium (GIBCO BRL, Grand Island, NY, USA) supplemented with 10% FBS (Hyclone, Logan, Utah, USA), penicillin, and streptomycin according to the recommendations of the American Type Culture Collection (ATCC). MDA-MB-231 cells were plated at a density of 2 × 10⁴ cells/well in 24-well plates and 8 × 10⁴ in 6-well plates for use in quantitative RT-PCR and luciferase assays, respectively. The cells were cultured in the presence or absence of 200 ng/ml rhBMP-6 (R&D Systems, Minneapolis, MN, USA) in DMEM supplemented with 5% FBS.

cDNA fragment encoding the full-length δEF1 sequence was prepared by PCR using the forward primer, 5'-CGCGGATCCAGGATCATGGCGGATGGC-3', and reverse primer, 5'-GAACCTCGAGCGAGCTTCATTTGTCTTCTCTTC-3'. The PCR products were digested with BamHI/XhoI, and cloned into pcDNA6B (Life Technologies, Grand Island, NY, USA). The E-cadherin promoter sequence (-308/+21) was obtained by PCR from human blood genomic DNA and cloned into the pGL4.10 vector (Promega, Madison, WI, USA) as described by Comijn et al. 17. Mutagenesis of the δEF1 sites in the human E-cadherin promoter was performed using the QuickChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) with the forward primer E2 box 1: 5'-gctgtggccggCAG

A

Total RNA was extracted from MCF-7 and MDA-MB-231 cells treated with or without 200 ng/ml rhBMP-6 for various times or from 16 breast cancer specimens using the TRIzol Reagent (Life Technologies, Grand Island, NY, USA). Total RNA (0.5 μg) from each sample was used for first strand cDNA synthesis (M-MLV Reverse Transcriptase, Promega, Madison, WI, USA). Specific products of human δEF1, human E-cadherin, and human BMP-6 were amplified by quantitative PCR using the following primers: δEF1, 5'-GGCCCCAGGTGTAAGCGC-3' (forward), and 5'-CAGGCCCCAGGATTTCTTG C-3' (reverse); E-cadherin, 5'-TGCTGCAGGTCTCCTCTTGG-3' (forward), and 5'-AGT CCCAGGCGTAGACCAAG-3' (reverse); BMP-6, 5'-CAACAGAGTCGTAATCA-3' (forward), and 5'-TTAGTGGCATCCACAAGCTCT-3' (reverse). GAPDH was used as an internal control. Verification of the expression levels of genes was performed by quantitative RT-PCR using EvaGreen (Botium, Hayward, CA, USA). The expression level was expressed as the threshold cycle (C_T) values of the target and reference gene-GAPDH, which is constitutively expressed and not regulated by treatment with rhBMP-6. Comparison and calculation of C_T values was used to determine the relative mRNA expression expressed as the fold change of target genes relative to the reference gene.

The following antibodies (Abs) were used: mouse monoclonal Ab against c-Myc (sc-40, Santa Cruz Biotechnology, CA, USA); goat polyclonal Ab against the N-terminal epitope of δEF1 (ZEB-E20, Santa Cruz Biotechnology, CA, USA); a mouse monoclonal Ab against E-cadherin (610181, BD Transduction Laboratories, KY, USA); and a mouse monoclonal Ab against FLAG-M2 (F-3165, Sigma, MO, USA).

MDA-MB-231 cells were solubilized in lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1% Triton ×-100, 0.5% sodium deoxycholate) supplemented with aprotinin (10 μg/ml), leupeptin (10 μg/ml), and PMSF (1 mM). The lysates were cleared by centrifugation and the protein concentrations were determined using the BCA Protein Assay Kit (Pierce, Rockford, IL, USA). Proteins were separated by SDS-PAGE and transferred onto nitrocellulose membranes (Amersham Biosciences, Piscataway, NJ, USA). Western blotting was performed using standard techniques and immunoreactive bands were detected by chemiluminescence (ECL, Amersham Biosciences, Piscataway, NJ, USA).

MDA-MB-231 cells were co-transfected with wild-type or mutant human E-cadherin promoter constructs and different amounts of δEF1 expression plasmids (1.5, 3.0, and 6.0 μg/well) in 6-well plates using Lipofectamine 2000 (Invitrogen, Austin, TX, USA). Cells were treated with rhBMP-6 (200 ng/ml) for 24 h after transfection. Lysates were prepared 24 h after treatment. The luciferase activity was then measured using the Dual-Luciferase Reporter Assay System (Promega, Madison, WI, USA) according to the manufacturer's instructions. Luciferase activity was normalized using the Renilla luciferase activity.

The target sequence of the siRNA is 5'-TGATCAGCCTCAATCTGCA-3' for human δEF1 as previously reported 36. The sense and antisense oligonucleotides with the internal loop were synthesized (TaKaRa, Shiga, Japan). These were annealed and ligated into the BamHI and HindIII sites of pSilencer 4.1-CMVneo (Ambion, Carlsbad, CA, USA) to construct the δEF1-specific siRNA expression plasmid according to the manufacturer's instructions. pSilencer 4.1-CMVneo expressing a scrambled siRNA (Ambion, Carlsbad, CA, USA) was used as a control. Transient transfection with siRNAs was performed with Lipofectamine 2000 (Invitrogen, Austin, TX, USA) in MDA-MB-231 cells. G418-resistent clones were isolated over a period of 3–4 weeks. Down-regulation of δEF1 was confirmed by western immunoblot analysis.

MDA-MB-231 cells were grown for 24 h up to 80% confluence in the presence or absence of 200 ng/ml rhBMP-6. Cells were cross-linked with 1% formaldehyde and processed using the Chromatin Immunoprecipitation (ChIP) Assay Kit (Updates, Lake Placid, NY, USA). The following antibodies (10 μg) were used: anti-ZEB, a polyclonal antibody raised against N-terminal epitopes of δEF1, or unrelated anti-FLAG control antibody (F-3165, Sigma, MO, USA) with 100 μg of chromatin per CHIP. Purified immunoprecipitated DNA was used for quantitative PCR reactions. The primers used for this analysis were as follows: 5'-AGGCTAGAGGGTCACCGCGTC-3' (forward), and 5'-GCTTTGCAGTTCCGACGCCAC-3' (reverse). Copy numbers for the DNA fragments (-175 to +21) of E-cadherin promoter in each anti-ZEB sample with or without BMP-6 induction were determined and compared to copy numbers of the DNA fragment without IP (input DNA). Anti-FLAG antibody was used as a control for IP reactions. The percentage of the input was then calculated. The final value was the percentage input obtained with specific antibody minus the percentage input obtained with anti-FLAG control antibody. The dissociation curve was determined for each quantitative PCR to ensure that a single band was produced. Each data point represents three independent samples.

MCF-7 and MDA-MB-231 breast cancer cells feature different capacities for invasiveness and metastasis in vivo and, as such, have been intensively used as tumor models during past few years 44. Our preliminary studies indicated that expressions of BMP-6 and δEF1 were inversely related in MCF-7 and MDA-MB-231 cells. MDA-MB-231 cells, which express high level of δEF1 mRNA, exhibited very low level of BMP-6 (Figure 1). In contrast, MCF-7 cells had markedly reduced δEF1 mRNA levels (Figure 1). Moreover, δEF1 has recently been identified as a direct transcriptional repressor of E-cadherin in these cancer cells 36. We therefore performed RT-PCR to examine the expression of E-cadherin in MCF-7 and MDA-MB-231 cells. Consistent with the previously reported results 1336, E-cadherin mRNA expression was high in MCF-7 cells, but undetectable in MDA-MB-231 cells (Figure 1).

After establishing the relationship between BMP-6, E-cadherin, and δEF1 in cancer cell lines, we moved on to validate this finding in breast cancer tissue samples. We collected 16 tumor specimens (T1–T8 were ER positive and T9–16 were ER negative) and analyzed their relative mRNA levels using quantitative RT-PCR. As shown in Figure 2b, an inverse correlation between BMP-6/E-cadherin and δEF1 is clearly evident in ER^- cancer specimens (7 out of 8: T9, T10, T11, T13, T14, T15, T16). Only in T12 was the low δEF1 level not accompanied by an increased level of E-cadherin/BMP-6. In ER⁺ cancer specimens, a clear inverse correlation was evident in T1, T3, T7, and T8 (Figure 2a). However, in the other ER⁺ tumor samples such an inverse correlation was not as evident. In addition, our results revealed that expression of BMP-6 in ER⁺ breast tumor specimens was comparatively higher than in ER^- cases. However, for expression of δEF1 in ER⁺ and ER^- breast tumor specimens, this relationship was reversed; here ER⁺ tumors had lower δEF1 mRNA levels than ER^-. Together, these data suggest a strong inverse relationship between the expressions of BMP-6/E-cadherin and δEF1, which could contribute to the invasiveness and metastatic capacity of breast cancer cells.

To assess whether a loss of E-cadherin expression in MDA-MB-231 cells is related to the presence of δEF1 suppression as well as a deficiency of BMP-6, the cells were cultured in the presence or absence of rhBMP-6. Total RNA was extracted at 0, 3, 12, 24, 48, 72, and 96 h following BMP-6 treatment. Quantitative RT-PCR analysis demonstrated that BMP-6 treatment for 72 h resulted in an up to 60% decrease in the expression of δEF1 mRNA, compared to the basal level (Figure 3a). Meanwhile, BMP-6 treatment for 96 h increased the expression of E-cadherin more than 8-fold over the level at baseline (Figure 3b). In the sequence of time, the BMP-6-induced down-regulation of δEF1 transcription occurred at as early as 3 h; whereas up-regulation of E-cadherin was observed only after 12 h of BMP-6 induction (Figure 3a and 3b), suggesting that BMP-6-induced down-regulation of δEF1 may well be a pre-requisite to allow the expression of E-cadherin.

To verify these findings, western blot was performed to determine BMP-6-modulated expression of δEF1 and E-cadheirn at the protein level. This time, MDA-MB-231 cells were cultured in the presence or absence of 200 ng/ml rhBMP-6 and total protein lysates were collected at 24 and 48 h following treatment. Western blot analysis revealed that BMP-6 treatment for 48 h significantly reduced the level of δEF1 protein (Figure 3c) with the expression of E-cadherin being up-regulated concurrently (Figure 3d).

With the knowledge that BMP-6 represses δEF1 transcription and causes up-regulation of E-cadherin in MDA-MB-231 cells, we moved on to examine whether BMP-6 is a bona fide stimulator of E-cadherin expression in MDA-MB-231 cells using reporter gene assays. The study showed that BMP-6 significantly stimulated E-cadherin promoter activity of the wild-type -308/+21 reporter gene, compared to the empty-vector transfected control cells (Figure 4b, lane 1 verse 2). Furthermore, we examined whether δEF1 overexpression could repress the BMP-6-induced transcriptional activation of E-cadherin. MDA-MB-231 cells were co-transfected with E-cadherin promoter reporter and δEF1 expression plasmid in the presence of rhBMP-6. Overexpression of δEF1 was confirmed by western blot (Figure 4a). As anticipated, δEF1 expression significantly inhibited the promoter activity of E-cadherin being activated by rhBMP-6 (Figure 4b, lane 2 verse 4), confirming a role for δEF1 in suppressing transcriptional activation of E-cadherin by inducers such as BMP-6.

To further address the specificity of δEF1 in regulation of E-cadherin promoter, we mutated two δEF1-binding sites identified in the human E-cadherin promoter (-308/+21), individually or in combination (Figure 5). The mutant E-cadherin promoter constructs were then co-transfected with δEF1 expression plasmid in the presence of rhBMP-6. A measurement of luciferase activity indicated that δEF1 expression inhibited BMP-6-induced reporter activity of the E-cadherin promoter harboring mutations in either E-box 1 or E-box 3 (Figure 4b, lane 6 verse 8, lane 10 verse 12), an effect similar to that of δEF1 on the wild-type E-cadherin promoter. However, mutations within both E-box elements abolished the repressive effect of δEF1 on the BMP-6-induced reporter gene activity (Figure 4b, lane 13 verse 15, lane 14 verse 16). These data suggested that ectopic expression of δEF1 can attenuate BMP-6-induced transactivation of the E-cadherin promoter through either the E-box 1 or E-box 3 binding sites in MDA-MB-231 cells. Mutation of both E-boxes is required to efficiently interfere with the repressive effect of δEF1 on BMP-6-regulated E-cadherin expression. Moreover, the double mutation of E-boxes not only abolishes the repressive effect of δEF1, it also blunts the induction effect by BMP-6 on E-cadherin promoter. Collectively, we can conclude that in MDA-MB-231 cells, the BMP-6 medicated activation of E-cadherin gene is closely associated with the ability of E-cadherin promoter to bind to its putative suppressors (such as δEF1) through its E-box 1 and E-box 3.

Naturally we were interested in investigating whether knockdown of δEF1 using RNA interference (RNAi) would result in an activation of the promoter. To do so, a small interfering RNA (siRNA) targeting δEF1 or a scrambled control siRNA were stably transfected into MDA-MB-231 cells, followed by treatments with 200 ng/ml rhBMP-6. The knockdown of δEF1 expression was confirmed by western blot (Figure 6a), accompanied by a significant increase in the promoter activity of E-cadherin with or without BMP-6 treatment, compared to the cells transfected with the control siRNA (Figure 6b). The data verified that depletion of δEF1 in MDA-MB-231 cells is sufficient to allow a native expression of the silenced E-cadherin gene. Furthermore, a knockdown of δEF1 abolished the E-cadherin promoter stimulation by rhBMP-6 (Figure 6b), suggesting that the stimulatory effect of BMP-6 on E-cadherin transcription occurs indirectly. Down-regulation of δEF1 is a pre-requisite for BMP-6 to induce the expression of E-cadherin. While the δEF1 suppression of the BMP-6-stimulated E-cadherin promoter activity was dose-dependent, double mutation of E-boxes causes a total abolishment of the δEF1 effect, given that the molecule was unable to exhibit a repression at any dose when both E-boxes were mutated (Figure 7).

As previously demonstrated, we have shown that BMP-6 can down-regulate the repressor, δEF1, to effectively stimulate the E-cadherin promoter. Considering that δEF1 is directly associated with the E-cadherin proximal promoter in vivo, acting as a direct transcription repressor 36, we wished to study this physical interaction in response to BMP-6 by quantitative CHIP assay. Total protein lysates from MDA-MB-231 cells treated with or without 200 ng/ml rhBMP-6 were further incubated with a δEF1 antibody directed against the δEF1 amino-terminus in order to precipitate endogenous δEF1. As shown in Figure 8a, δEF1 specifically bound to the -175/+21 region of the E-cadherin promoter, and the binding was significantly reduced by BMP-6 treatment (Figure 8b). This data supports the notion that BMP-6 stimulation of E-cadherin promoter takes place through both a reduction of δEF1 expression and a physical removal of δEF1 from binding to its cognate DNA elements.