Background
Epithelial ovarian cancer (EOC) is the second most common gynecological cancer and accounts for nearly half of all deaths associated with gynecological pelvic malignancies. Largely asymptomatic, over 70% of patients diagnosed with ovarian cancer at an advanced stage of the disease. Early detection is rare and screening programs in the general population have been unsuccessful. Recent studies have analyzed gene expression patterns to identify the molecular events involved in the development of cancer and to uncover diagnostic and prognostic markers. This approach, applied to ovarian cancer [
2‐
10], has resulted in the identification of several hundred genes differentially expressed between NOSE (normal ovarian surface epithelia) and EOC [
11]. In a previous study from our group [
1] several candidate genes that discriminate NOSE from EOC cells were identified and validated by real time RT-PCR. The differential expression of one of these candidates, bone morphogenic protein-2 (BMP-2), was further validated by immunohistochemistry (IHC) of patient tissue samples [
1].
The biological role of BMP-2 in ovarian cancer has not been elucidated. BMPs are members of the TGF-β superfamily, which play an important role in embryonic development events, such as gastrulation, neurogenesis, hematopoiesis and apoptosis [
12,
13]. Recent studies have suggested that some BMPs are implicated in cancer development [
14] as shown in breast and prostate cancer (reviewed in [
15,
16]). The effects of BMP-2 on cancer cells are controversial and are perhaps dependent on the tissue and environment where they are expressed [
17]. For example, BMP-2 has been shown to stimulate the growth of pancreatic carcinoma cells and prostate cancer cells in absence of androgen [
18,
19]. On the other hand, BMP-2 clearly inhibits the growth of tumor cells of many origins including cancers arising from thyroid, androgen-dependent prostate in presence of androgen, myeloma, gastric and pancreatic cells [
14,
18‐
22]. In cancer cells, BMP-2 was found to suppress apoptosis induced by TNFα or by serum deprivation [
23‐
25]. In ovarian cancer, overexpression of BMP-2, BMP-4 and BMP-7 mRNAs have been reported as dysregulated by microarray analyses [
1,
7,
8]. A recent study has demonstrated the involvement of BMP-4 in the epithelial mesenchymal transition in human ovarian cancer cells [
26]. Since BMP-2, along with family members BMP-4 and BMP-7, share the same receptors they may have similar effects. However, the binding affinity of BMPs on these receptors and subsequent receptor oligomerization are different which may lead to different downstream signaling and biological effects in response to BMPs [
15,
27].
BMP-2 acts via two types of serine/threonine receptors [
27]. Type I receptors are BMPR1a/Alk3 and BMPR1b/Alk6 and type II receptors are BMPR2 and ActRIIA. Type I receptors are phosphorylated by type II receptors after oligomerization occurs. Of the two signaling pathways for BMP, the Smad-dependent pathway appears to be the most important. Smad 1/5/8 are mediators of BMPRIa and BMPRIb whereas Smad6 and Smad7 are the inhibitory Smads of this pathway [
28] Phosphorylated Smad 1/5/8 forms a complex with Smad4 and translocate in the nucleus (review [
15]). The Smad-independent pathway activates TAK1, which can lead to MAPK activation as well as Akt and NF-kappaB activation [
29,
30]. The most characterized target genes of the BMP-2 signaling are
Id1 and
Smad6 that encode products promoting the growth regulation of BMPs. The signaling pathway induced by BMP-2 can be modulated by numerous antagonist proteins, such as Noggin, Cerbarus and Gremlin. These antagonists are secreted in the extracellular matrix. Previous results using Noggin [
26] and Chordin [
31] support the potential therapeutic role of these antagonists in ovarian cancer progression through the inhibition of BMP signaling. It has also been reported that
Gremlin gene expression is lower in ovarian cancer specimens compared to normal ovarian culture [
28].
In the present study, we focused on the role of BMP-2 in ovarian cancer. First, we examined the biological role of BMP-2 on three novel ovarian cancer cell lines (TOV-2223, TOV-1946, TOV-112D). These lines were selected since they do not express detectable levels of BMP-2, consequently, their sensitivity and response to recombinant BMP-2 protein was examined. The ability of BMP-2 to induce signaling pathways and expression of target genes was investigated. Functional assays were also performed to determine the in vitro behavior of these cell lines in response to BMP-2 treatment. Finally the association between BMP-2 and ovarian cancer patient survival was examined using ovarian cancer tissue array analysis.
Methods
Cell culture and reagents
The TOV-2223, TOV-1946 and TOV-112D cell lines, developed from long term passages of serous ovarian cancer samples as described previously [
32,
33], were grown at 37°C in 5% CO
2 and in OSE consisting of 50:50 medium 199:105 supplemented with 5% fetal bovine serum (FBS) and 2 μg/ml Gentamicin. All reagents used for cell culture media were purchased from Wisent (Qc, Canada). Human recombinant BMP-2 (355-BM-010/CF) and mouse Noggin (#1967-NG-025/CF) were supplied by R&D system (Mineapolis, MN, USA). TNF-α was obtained from Roche Applied Science (Indianapolis, IN). BMP-2-pCMV6-XL4 was purchased from Origene (Rockville, MD) and cloned into pcDNA3.1 (Invitrogen Life Technologies, Carlsbad, CA) as a
NotI fragment. The pcDNA3.1-BMP-2 gene was sequenced to confirm the correct insertion of BMP-2 cDNA in the pcDNA3.1 vector.
Primary cultures, tumor samples and patient characteristics
Tumor samples were collected from surgeries performed at the Centre hospitalier de l'Université de Montréal (CHUM). An independent pathologist assigned histopathology and tumor grade according to International Federation of Gynecology and Obstetrics (FIGO) criteria. A gynecologic oncologist reviewed tumor stage and residual disease. Normal tissues were obtained from tumor-free participants that have undergone oophorectomy. Primary cell cultures from normal ovarian surface epithelia (NOSE) and EOC samples were established as described [
34,
35]. Cells in primary culture were maintained in OSE media supplemented with 10% (v/v) fetal bovine serum (FBS), 2.5 ug/mL amphotericin B and 50 μg/mL gentamicin [
34]. The tumor samples used for the tissue array studies are presented in Table
1. Tissue selection criteria for this study was based on all histopathologies from chemotherapy-naïve patients having provided informed consent with all samples having been collected between 1993–2003. Clinical data were extracted from the Système d'Archivage des Données en Oncologie (SARDO) that includes entries on tumor grade and stage, treatment and clinical outcomes such as the progression-free interval as defined by RECIST criteria and survival. No correlation between age of embedded paraffin tissues and antibody staining intensity on the tissue array was identified.
Table 1
Composition of the ovarian cancer tissue array
Serous | 19 | 3 | 17 | 6 | 4 | 7 | 2 |
Endometrioid | 25 | 18 | 6 | 9 | 8 | 4 | 1 |
Clear cell | 15 | 10 | 5 | 2 | 2 | 8 | 3 |
Mucinous | 25 | 20 | 2 | 1 | 0 | 2 | 19 |
Mixed cells | 5 | 2 | 3 | 0 | 0 | 4 | 1 |
ELISA
Culture supernatants from confluent cellular monolayers were centrifuged at 3000 rpm for 10 min and frozen at -80C until further use. All ascites fluids were re-centrifuged for 10 min at 8000 rpm before performing ELISAs. After centrifugation, samples were tested by ELISA for secreted mature BMP-2 (item DBP200, R&D System) concentration according to the manufacturer's instructions. The limit of detection for BMP-2 was 30 pg/ml.
RNA preparation and Quantitative PCR
Total RNA from cell lines was prepared using the RNeasy kit from Qiagen (Qiagen Inc., ON, Canada). The cDNA synthesis was done according to the protocol of the SuperScript™ First-Strand Synthesis System for real time PCR (Invitrogen Life Technologies, Carlsbad, CA) with a starting amount of 2 μg RNA and reverse transcription performed with random hexamers. The PCR reaction was performed with a Rotor-gene 3000 Real-Time Centrifugal DNA Amplification System (Corbett tumor tissues Research, NSW, Australia). The Quantitect™ SYBR Green PCR (Qiagen) reaction mixture was used according to the manufacturer's instructions. Serial dilutions were performed to generate a standard curve for each gene tested in order to define the efficiency of the real time PCR reaction and a melt curve was done to confirm the specificity of the reaction. Based on the strong stability of ERK1 gene expression in ovarian cancer tissue, it was chosen as an internal control [
1]. All experiments, including positive and negative controls, were performed in triplicate. The PCR primers targeted exonic sequences that were interrupted by at least one intron. The amplicons were sequenced to verify their specificity for the targeted genes. Primers were: Id1 fw 5'-cggaatctgagggagaacaag, rev 5'-ctgagaagcaccaaacgtga; Smad6 fw 5'-gagctgagccgagagaaaga, rev 5'-agatgcacttggagcgagtt: Snail fw 5'-gagtggttcttctgcgctac, rev 5'-cagagtcccagatgagcatt; Wnt5a fw 5'-gcgcgaagacaggcatcaaag, rev 3'-ggcgttcaccacccctgctg; Erk1 fw 5'-gcgctggctcacccctacct, rev 5'-gccccagggtgcagagatgtc, BMPR1a fw 5'-cttattcagctgcctgtggt, rev 5'-attcttccacgatccctcct; BMPR1b fw 5'-tacaagcctgccataagtgaagaagc, rev 5'-tcatcgtgaaacaatatccgtctg and BMPR2: fw 5'-gctaaaatttggcagcaagc, rev 5'-cttgggccctatgtgtcact.
Western blot analysis
Cells were lysed with cold lysis buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM DTT/1 mM NaF/0.5% NP-40/0.5 mM PMSF/0.2 mM sodium orthovanadate/2 μg/ml of aprotinin, leupeptin and pepstatin), and the lysate boiled in loading buffer, separated by SDS-PAGE, and transferred onto a nitrocellulose membrane. Membranes were saturated with 5% (w/v) milk/PBS/0.1% Tween 20. Immunodetection was done as described in the ECL kit protocol (Amersham Pharmacia): i.e. incubated 2 h at room temperature with specific antibody, washed with PBS and incubated for another 30 min at room temperature with peroxidase-conjugated antibodies (Santa-Cruz Biotechnology Inc.). Western-blot analysis was performed with Erk1 (Santa Cruz Biotechnology Inc, CA), Smad1/5/8 (A-14 Santa Cruz Biotechnology Inc.), phospho-Erks (Cell Signaling, Beverly, MA, USA), phospho-Smad1/5/8 (Cell Signaling), p65 (Santa Cruz Biotechnology Inc), Akt (Santa-Cruz Biotechnology Inc.), phospho-Akt and phospho-S-536-p65 (Cell Signaling) and beta-Actin (AbCam, MA, USA) antibodies. All experiments were performed in triplicate with the TOV-2223 cell line and at least twice for the TOV-1946 and TOV-112D cell lines.
Cytoplasmic and nuclear extracts
To prepare cellular extracts, 5 × 106 cells were washed twice in cold PBS buffer and resuspended in lysis buffer containing 10 mM Tris pH 7.9/10 mM NaCl/5 mM MgCl2/10 mM sodium orthovanadate/0.5 mM PMSF/10 μg/ml of the protease inhibitors (PMSF, pepstatin, leupeptin and aprotinin). After swelling the cells for 30 min on ice, 0.1% Nonidet P-40 and 10% glycerol (v/v) were added and the lysates centrifuged for 1 min at 4°C and 5,000 rpm. Supernatants consisting of cytoplasmic extracts were carefully decanted for cytoplasmic extracts. Nuclei pellets were resuspended for 1 h in 40 μl of lysis buffer containing 10 mM Tris pH 7.9/400 mM NaCl/0.1 EDTA/0.5 mM DTT/5% glycerol/0.5 mM PMSF/10 μg/ml protease inhibitors. Particulate matter was eliminated by centrifugation for 10 min at 13,000 × g at 4°C. Protein concentrations were determined using the Bradford method.
Transfection and luciferase reporter assay
Cells were plated in 96-well plates and at 70–80% confluence (approximately 5 × 10
4 cells), they were co-transfected with 0.2 μg of DNA and 200 ng of a constitutively active Renilla luciferase (pCMV-RL) (Promega, WI, USA) by the lipofectamine method (Invitrogen Life Technology). After 6 h, cells were washed in fresh medium and incubated overnight. Cells were stimulated for 16 h with BMP-2 or TNFα and were then assayed for luciferase activity using the dual luciferase reporter assay system (Promega). The 3enh-κb-CONA-luc carries a firefly luciferase gene under the control of a trimeric repeat of the κB consensus [
36].
Cell proliferation
To determine the effect of BMP-2 on cell growth, 5 × 104 cells were plated into six well culture plates. After allowing the cells to adhere overnight they were treated with recombinant BMP-2 and/or Noggin. Two, four and six days later, cells were detached with trypsin and counted in the presence of 0.05% Trypan blue using a hemacytometer. Untreated cells were used as controls.
Migration assays
Cells were grown to confluence in 6 well culture plates. Using a pipet tip, a wound was produced in the monolayer at two different positions on the plate. The adherent monolayer was then washed two times in PBS to remove non-adherent cells and media/FBS was added with or without BMP-2. After 20 or 40 hrs the open wound surface area was quantified by digital images taken under phase contrast microscopy. All experiments were repeated at least twice.
Spheroids were formed using a modification of the hanging droplet method [
37]. Briefly, 4 × 10
3 cells were resuspended in 16 μl of OSE/FBS media supplemented with 50 ng/ml BMP-2 and placed on the cover of a 150 mm tissue culture plate. The cover was placed over a plate that contained 15 ml of OSE to prevent dehydration of the hanging droplets. Spheroid formation was monitored after four and ten days, and representative spheroids were photographed. Untreated cells were used as controls.
Tissue array and immunohistochemistry (IHC)
A tissue array containing 94 cores of ovarian epithelial tissues was built (Table
1, [
1]). A detailed protocol is described in Le Page et al, [
1]. Briefly, the tissue array was heated at 60°C for 30 min, de-paraffinized in toluene and rehydrated in a gradient of ethanol. Antigen retrieval was done in 90°C citrate buffer (0.01 M citric acid + 500 ul Tween-20/L adjusted to pH 6.0) (J.T. Baker Philipsburg, NJ) for 15 min. The tissue was blocked with a serum-free reagent (DakoCytomation Inc., Mississauga, ON) and incubated with BMP-2 antibodies (Santa-Cruz Biotechnology, CA, USA) overnight at 4°C in a humid chamber. Optimal antibody concentration was determined by serial dilutions. Endogenous peroxidase activity was quenched by treatment with 3% H
2O
2. The array was incubated with a secondary biotinylated antibody (DakoCytomation Inc.) followed by incubation with a streptavidin-peroxidase complex (DakoCytomation Inc.) for 10 min at room temperature. Reaction products were developed using diaminobenzidine containing 0.3% H
2O
2 as a substrate for peroxidase and nuclei were counterstained with diluted hematoxylin. Epithelial zones were scored according to the intensity of staining (value of 0 for absence, 1 for weak, 2 for moderate, 3 for high and 4 for very high intensity). Each array was independently analyzed in a blind study by two independent observers.
Statistical analysis
For survival and progression-free disease analyses, we used the Cox regression survival model with time dependent covariate and Kaplan-Meier curves coupled with the log rank test. Receiver operating characteristics (ROC) curves were generated for each marker to define a threshold of expression corresponding to the best sensitivity and specificity for patient survival. A threshold of BMP-2 intensity = 4 appeared optimal. For Cox regression analysis, the markers were treated as categorical variables based on the threshold of expression. All statistical analyses were performed using SPSS software, version 11.0 (SPSS Inc., Chicago, IL, USA).
Discussion
In this study we attempted to clarify the role of BMP-2 in ovarian cancer. An initial report highlighted the overexpression of BMP-2 in primary cultures of ovarian cancer cells and in the tissues of ovarian cancer patients [
1]. Using three different cell lines, we report different
in vitro and
in vivo effects of BMP-2 on epithelial ovarian cancer cells. The three cell lines selected for this study expressed receptors for BMP-2 and were responsive to BMP-2 stimulation as seen by the activation and phosphorylation of Smad1/5/8 transcription factors as well as the gene expression of
Id 1,
Snail and
Smad6. However, although the signaling pattern was similar in all cell lines, they did not show the same biological activities in the presence of BMP-2. Only the TOV-2223 cell line showed a reduce proliferation rate in the presence of BMP-2 and was not influenced when cultured in 3D spheroid conditions. In contrast, the motility of all cell lines was stimulated in presence of BMP-2. Further work needs to be done to define particular characteristic of each ovarian cancer cell line that determine response to BMP-2. These results suggest that the effects of BMP-2 on ovarian cancer cells may be complex and dependent on the particular cellular context. The heterogeneity in response to BMP-2 is unlikely related to the histopathological subtype since TOV-2223 and TOV-1946, which respond differently to the presence of BMP-2, are both derived from a serous subtype.
Similar effects with BMP-4, as observed here with BMP-2, have recently been reported in ovarian cancer cell lines and ovarian cancer primary cultures [
26,
39]. We observed that BMP-2 slightly reduced the proliferation of TOV-2223 cells but had no effect on TOV-112D and TOV-1946 cells suggesting that some cell lines are resistant to the anti-proliferative activity of BMP-2. In the same way, BMP-4 has also been reported to slightly reduce the proliferation of SKOV3 ovarian cancer cells, as well as some primary cultures of ovarian cancer cells while other ovarian primary cultures were not sensitive to this protein [
39]. The reason why some ovarian cancer cells are resistant to this anti-proliferative effect is unknown. We also observed similar increases in motility in cells treated with BMP-2 as reported by others with BMP-4 [
26]. Since BMP-2 and BMP-4 bind the same type I and type II BMP receptors, it is not surprising to notice similarities in their induction of signaling pathways. A strategy based on the single inhibition of either BMP-2 or BMP-4 may not be sufficient to reduce the tumorigenic effect driven by Smad1/5/8 signaling. In contrast, targeting several BMPs by the use of extracellular antagonists such as Chordin, Noggin or Gremlin may be more effective. Preliminary results shown here with Noggin and by others using Noggin [
26] and Chordin [
31] support the potential therapeutic role of these antagonists in ovarian cancer progression through the inhibition of BMP signaling. It will be of great interest to test Noggin, Chordin, Cerberus or Gremlin as
in vivo potential tumor suppressors in xenograph models.
We also observed that some malignant cells from ascites samples overexpressed BMP-2 compared to cells from solid tumor samples of the same patients. The motility of cancer cells is an important factor determining the metastatic spread of tumors. As ascites tumor cells are detached from the primary tumor site and may have acquired a metastatic potential, this observation suggests that BMP-2 may be associated or involved in the process of evading tumor cells from the primary site to the omentum. In line with this hypothesis, we observed that BMP-2 stimulates the
in vitro migration of ovarian cancer cell lines. In addition several reports have shown a role of BMP-2 in invasion of lung, prostate, breast cancer cells and BMP-4 in ovarian cancer [
21,
40,
41]. To confirm the role of BMP-2 in the metastatic process of ovarian cancer cells, additional
in vivo assays would be required. Metastasis is a major cause of cancer related mortality. The fact that patients with higher expression of BMP-2 in ovarian tissues have shorter survival supports a role for BMP-2 in the motility of ovarian cancer cells and aggressiveness of ovarian tumors. Further functional assays are required to determine the exact role of BMP-2 in these biological processes.
Acknowledgements
The authors are very grateful to the staff and patients at the Gynecologic Oncology Service at the Hôpital Notre-Dame for providing the samples. We thank Lise Portelance, Louise Champoux, Jean-Simon Diallo and Jason Madore for their assistance. MZ was supported by studentships from Canderel and Marc Bourgie funds of the Institut du cancer de Montréal, and Faculté des études supérieures de l'Université de Montréal.
This work was supported by a grant from the Canadian Institutes of Health Research (CIHR) to A.-M.M.-M., P.N.T. and D.M.P. Tumor banking was supported by the Banque de tissus et de données of the Réseau de recherche sur le cancer of the Fonds de la Recherche en Santé du Québec (FRSQ), affiliated with the Canadian Tumor Repository Network (CTRNet).
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
Conception, coordination and design of the study: CLP, PT and AMMM. Financial support to: PT, DMP and AMMM. Collection and analysis of clinical data: CLP, MdeL and DMP. Collection and analysis of molecular data: CLP, MP, MZ and LM. Collection and Assembly of data: CLP. Data analysis and interpretation: CLP, MdeL, MZ and LM. Manuscript writing: CLP and AMMM.