Formalin-embedded breast cancer samples were obtained from the Department of Surgical Pathology of Tangshan People's Hospital. Immunohistochemical analysis was performed on paraffin-embedded sections using the EnVision kit (Dako, Glostrup, Denmark) following the manufacturer's protocols. Sections were boiled in retrieval solutions to expose the antigens. Polyclonal anti-BMP-6 (ab134723) and control primary antibodies were applied to the sections at a dilution of 1:50. The section-affixed slides were counterstained with hematoxylin, dehydrated and mounted.

Human breast cancer cell line MDA-MB-231 was maintained in Dulbecco's modified Eagle's medium (DMEM)-high glucose medium (Gibco) supplemented with 10% fetal bovine serum (FBS) (HyClone), penicillin and streptomycin. MDA-MB-231 cells were plated at a density of 2×10⁴ cells/well into 24-well plates for use in the luciferase assays, and at a density of 8×10⁴ into 6-well plates for western blotting and quantitative CHIP assays.

The constitutive vectors, pcDNA6B-BMP-6 (17), hMMP-1-WT and hMMP-1-mAP1 (10) were previously described. The full-length MMP-1 CDS was PCR amplified using the following primers: forward, 5′-AGCAAGCTTAGATGCACAGCTTTCCTCCAC-3′ and reverse, 5′-AGTCTCGAGTCAATTTTTCCTGCAGTTGAA-3′. The HindIII/XhoI site-tagged potential target sequence was then cloned into the HindIII/XhoI site of the pcDNA-6B vector.

The MDA-MB-231 cells were plated into 6-well plates and transfected with pcDNA6B-BMP-6 or pcDNA6B using Lipofectamine 2000 (Invitrogen). Transfected cells were resuspended in 10 ml of DMEM containing 10% FBS and seeded in a 10-cm-dish. Blaticidin-resistant clones were isolated over a period of 3–4 weeks. Overexpression of BMP-6 was confirmed by quantitative RT-PCR and western blotting.

The siRNA target sequences of human BMP-6 and MMP-1 were, 5′-GCGACACCACAAAGAGTTCTT-3′ and 5′-GAGTACAACTTACATCG TG-3′ as previously reported (9,18). Sense and antisense oligonucleotides with the internal loop were synthesized (Takara). These were annealed and ligated into the BamHI and HindIII sites of pSilencer 4.1-CMV neo (Ambion) to construct the BMP-6- and MMP-1-specific siRNA expression plasmids, according to the manufacturer's instructions. pSilencer 4.1-CMV neo expressing scrambled siRNAs (Ambion) were used as controls.

231-control and 231-BMP-6 cells were seeded into a 6-well plate. For the scratch-wound healing assay, linear scratches were made in cell sheets with a pipette tip. After 24 and 48 h of growth, images were captured of the cells using a microscope.

Transwell 24-well chambers with 8-mm pore size (Costar) were used as recommended by the manufacturer. The membranes were precoated with collagen matrix (Sigma), which was reconstituted by adding 0.5 ml serum-free medium to the well for 2 h. To assess the invasive ability of the cells, 2.5×10⁴ cells in 0.5 ml medium containing 1% FBS were placed into the upper compartment of the wells and 0.75 ml of medium containing 10% FBS was placed in the lower compartment. The Transwell chambers were incubated for 16 h at 37°C in a 5% CO₂ incubator. Cell penetration through the membrane was detected by staining the cells on the porous membrane with 0.25% crystal violet. To quantify the data, we washed the chamber twice with phosphate-buffered saline (PBS), and then used 33% acetic acid to wash off the excess crystal violet. Crystal violet remaining on the membranes was measured on a spectrophotometer at A570.

Total RNA was extracted from MDA-MB-231 cells that were treated with BMP-6 protein using the TRIzol reagent (Invitrogen). 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 MMP-1 (10), MMP-2 (10), MMP-9 (10), MMP-14 (19), TIMP-1 (20) and TIMP-2 (20) were amplified by quantitative PCR using the primers as reported. GAPDH was used as an internal control. Verification of the expression levels of genes was performed by quantitative RT-PCR using EvaGreen (Botium).

MMP-1 protein expression levels were measured with an MMP-1 ELISA kit (cat. #QIA55; Calbiochem) according to the manufacturer's instructions. To prepare samples for ELISA, cells were grown to 80% confluency and transfected with MMP-1-pcDNA6B, si-MMP-1 or treated with 2 µM MMP-1 inhibitor (cat. # 444250; EMD Chemicals, Gibbstown, NJ, USA) for 24 h; supernatants from the 48-h incubation were collected and concentrated 10-fold, using Amicon Ultra-4 centrifugal filters (Merck Millipore, Billerica, MA, USA). MMP-1 amounts were calculated as ng/ml protein.

Preparation of total cell extracts and western blotting with the appropriate antibodies was performed as previously described (13). The following antibodies were used: rabbit polyclonal Ab against MMP-1 (sc-30069; Santa Cruz), mouse monoclonal Ab against actin (A-4700; Sigma), rabbit polyclonal Ab against c-Fos (sc-52), and rabbit polyclonal Ab against c-Jun (sc-1694) (both from Santa Cruz).

MDA-MB-231 cells were co-transfected with wild-type or mutant human MMP-1 promoter constructs into 24-well plates using Lipofectamine 2000. Cells were treated with TPA (100 ng/ml), curcumin (20 µM) or BMP-6 (100 ng/µl) for 6 h after transfection. Lysates were prepared and the luciferase activity was then measured using the Dual-Luciferase Reporter Assay system (Promega) according to the manufacturer's instructions. Luciferase activity was normalized using the Renilla luciferase activity.

ChIP assays were performed using reagents commercially obtained from Upstate Biotechnology (Lake Placid, NY, USA), essentially according to the manufacturer's instructions. The antibodies and primers used in these experiments were previously reported (10).

To better understand the expression of BMP-6 in breast cancer, we collected 36 cases of advanced breast tumor specimens with lymph node metastasis, and 32 cases of paired or unpaired normal tissues. We detected the BMP-6 expression level in breast cancer and normal breast tissues by immunohistochemical staining. The expression of BMP-6 was represented by the numbers or percentages of negative cases. The results showed that the rate of BMP-6-negative expression was 30.56% in breast cancer, but was 9.58% in the normal tissues (Table I and Fig. 1). These results from clinical samples suggest that the absence of BMP-6 expression in a certain percentage of the breast cancer tissues may be closely related with the metastasis of certain breast cancer cases.

To confirm the potential role of BMP-6 in cell migration and invasion in vivo, MDA-MB-231 cells were stably trans-fected with the BMP-6 expression plasmid (231-BMP-6) or the empty vector (231-control). The overexpression of BMP-6 was confirmed by quantitative RT-PCR and western blotting (Fig. 2A). The morphology of 231-control and 231-BMP-6 breast cancer cells was detected under a light microscope. As shown in Fig. 2B, the 231-BMP-6 cells were arranged more closely than the 231-control cells. The results of the wound healing test showed that the motile ability of the 231-BMP cells was weaker than that of the 231-control cells (Fig. 2C). Furthermore, we examined the effect of BMP-6 on the ability of breast cancer cells to migrate through an artificial basal membrane using the Boyden chamber assay. The results of cell migration and invasion assays indicated that MDA-MB-231 cells overexpressing BMP-6 presented a distinct decrease in the number of migrating and invading cells in comparison to the empty-vector control (Fig. 2D).

To understand the molecular changes, we performed quantitative RT-PCR to examine the effect of BMP-6 on MMPs and TIMPs in the MDA-MB-231 cells. MMP-1 has been described as a mediator that controls the ability of breast cancer growth, invasion and distant metastasis (9,10). In the present study, MDA-MB-231 cells were treated with different doses (50, 100 and 200 ng/ml) of BMP-6 protein. Total RNAs were collected at 24 h. As shown in Fig. 3A, BMP-6 down-regulated MMP-1 expression in a dose-dependent manner. Then, MDA-MB-231 cells were treated with BMP-6 (100 ng/ml). Total RNAs were collected at 0, 12, 24 and 48 h. In the MDA-MB-231 cells treated with BMP-6 (100 ng/ml) for 12 h, an up to 40% decrease in the expression of MMP-1 mRNA was noted when compared to the basal level (Fig. 3B). The inhibition was further decreased to ~60% between 24 and 48 h (Fig. 3B). The expression of MMP-1 was significantly lower in the 231-BMP-6 cells compared with the level in the control group at both the mRNA and protein levels (Figs. 3C, 5A and B). Additionally, we examined BMP-6-regulated expression levels of another three MMP members, MMP-2, MMP-9 and MMP-14 in the MDA-MB-231 cells. However, as shown in Fig. 3D–F, BMP-6 slightly downregulated MMP-2, MMP-9 and MMP-14 expression at the mRNA level. The MMP system is counterbalanced by a group of four tissue inhibitors of metalloproteinases (TIMPs) that have varying specificities for individual MMPs (21). Thus, we further assessed the potential effect of BMP-6 on TIMP-1, TIMP-2, TIMP-3 and TIMP-4. The results of quantitative RT-PCR showed that BMP-6 only slightly upregulated the mRNA levels of TIMP-1 and TIMP-2 (Fig. 3G and H). BMP-6 resulted in an up to 3-fold increase in the expression of TIMP-1 mRNA at 48 h, compared to the basal level (Fig. 3G). TIMP-3 and TIMP-4 mRNA levels were not significantly affected (data not shown). The above observations suggest there may be a dual effect of BMP-6 in promoting MDA-MB-231 cell migration, through concurrently downregulating MMPs and upregulating TIMP-1 and TIMP-2. However, inhibition of MMP-1, which was specifically and significantly mediated by BMP-6, contributed to induce the metastasis of breast cancer.

Having found that BMP-6 inhibits MMP-1 expression in MDA-MB-231 cells, we next assessed whether BMP-6 is a true suppressor of MMP-1 transcription using reporter gene assays. As shown in Fig. 4B, a reporter gene assay showed that BMP-6 significantly inhibited human MMP-1 promoter activity of the wild-type −159/+19 reporter gene. The inhibition was >50% following BMP-6 treatment, compared to that of the control. However, knockdown of BMP-6 using RNA interference resulted in increased luciferase activity of the human MMP-1 promoter, compared to that of the si-control (Fig. 4B). In order to clarify the regulatory mechanism by which BMP-6 affects MMP-1 expression, we analyzed the MMP-1 promoter using the online tools TRANSFAC (http://www.generegulation.com) and thTESS (http://www.cbil.upenn.edu/cgi-bin/tess/tess). Based on our online analysis, we found that there was an AP-1 binding site (CATGAGTCAG) at the position −70/−60 of the MMP-1 promoter (Fig. 4A). Our previous results indicated that BMP-6 inhibited the expression of microRNA-21 via reduction in the expression of δEF1 and c-Fos/c-Jun (22). In the present study, luciferase assay results indicated that TPA (activator of the AP-1 signaling) upregulated the promoter activity of MMP-1, whereas curcumin (inhibitor of AP-1 signaling) repressed its activity in the MDA-MB-231 cells (Fig. 4C), suggesting a potential effect of the AP-1 pathway to mediate MMP-1 transcription. Thus, the AP-1 element on the human MMP-1 promoter was mutated to generate the MMP-1-mAP-1 construct (10). The results of the luciferase assay demonstrated that the mutation of the AP-1 element on human MMP-1 promoter totally depleted its response to TPA, curcumin and BMP-6 (Fig. 4D). We investigated whether BMP-6 affects c-Fos or c-Jun expression at the protein level. Western blotting experiments demonstrated that both c-Fos and c-Jun expression was obviously inhibited in the 231-BMP-6 cells (Fig. 4E). Importantly, the ChIP assays indicated that c-Fos/c-Jun was able to bind to the MMP-1 promoter during basal conditions in an AP-1 site-dependent manner (Fig. 4F). The result of the q-CHIP assay showed that the binding of c-Fos/c-Jun to the MMP-1 promoter was decreased by BMP-6 induction (Fig. 4G).

Having clarified how BMP-6 inhibits MMP-1 expression, we investigated whether MMP-1 influences the biological effect of BMP-6 in breast cancer cells. MMP-1 has been reported to induce invasion in MDA-MB-231 cells (9). We investigated whether MMP-1 could rescue BMP-6-inhibited breast invasion in vitro. 231-control and 231-BMP-6 cells were transfected with the MMP-1 expression plasmid or an siRNA expression plasmid targeting MMP-1 or treated with an MMP-1 inhibitor. Expression of MMP-1 was confirmed by ELISA and western blotting (Fig. 5A and B). Boyden chamber assay showed that MMP-1 induced MDA-MB-231 cell invasion in vitro. In contrast, MMP-1 knockdown by RNA interference or the MMP-1 inhibitor exhibited an opposite effect (Fig. 5C and D). Moreover, MMP-1 decreased BMP-6-inhibited invasion of MDA-MB-231 cells (Fig. 5C and D).