Articles in PresS. Am J Physiol Cell Physiol (January 4, 2006). doi:10.1152/ajpcell.00478.2005
Molecular pathways regulating epidermal growth factor-induced epitheliomesenchymal transition in human ovarian surface epithelium Nuzhat Ahmed1,2, Sarah Maines-Bandiera3, Michael A Quinn1, Waldemar G. Unger3, Shoukat Dedhar4 and Nelly Auersperg3 1
Gynaecological Cancer Research Centre, Royal Women’s Hospital and Department of
Obstetrics and Gynaecology, University of Melbourne, Australia. 2Translational Proteomics, Baker Heart Research Institute, Australia. 3Department of Obstetrics and Gynaecology, University of British Columbia, Vancouver, Canada. 4BC Cancer Agency and The Prostate Centre at Vancouver Hospital and Department of Biochemistry and Molecular Biology, University of British Columbia, Vancouver, Canada.
Running title: Epithelio-mesenchymal transition in ovarian surface epithelial cells
Key words: Ovarian surface epithelium, EGF, migration, invasion, epithelio-mesenchymal transition Corresponding author: Dr Nuzhat Ahmed Gynaecological Cancer Research Centre Royal Women’s Hospital 132 Grattan Street, Carlton, Victoria 3053. Telephone: 61 3 9344 2616; Fax: 61 3 9344 2619 Email:
[email protected]
1 Copyright © 2006 by the American Physiological Society.
Abstract The ovarian surface epithelium is the pelvic mesothelium overlaying the ovaries. It is the precursor of the common epithelial ovarian carcinomas. In contrast to carcinomas in other physiological sites, where normal tissues of origin are stably epithelial and their neoplastic derivatives undergo epithelio-mesenchymal transition (EMT), normal ovarian epithelial cells (OSE) undergo EMT in response to environmental influences while epithelial ovarian cancer cells maintain differentiated epithelial phenotype. The present study examined the molecular mechanisms and possible physiological basis for the propensity of OSE to undergo EMT. At ovulation, ovarian surface epithelial layer is disrupted and the displaced fragments trapped in the ovarian cortex or the ruptured follicles are exposed to EMT-inducing factors derived from follicular fluid, platelets, stromal and luteal cells. We hypothesized that EMT may be a homeostatic mechanism that permits displaced OSE to assume a stromal phenotype within the ovarian cortex. Previously we have reported that epidermal growth factor (EGF) in conjunction with hydrocortisone (HC) is the EMT-inducing factor of OSE as shown by changes to a fibroblast-like morphology and growth pattern, increased collagen type III deposition and reduced keratin expression. We now report that EGF also increased cell motility, enhanced activities of secreted pro-matrix metalloproteinase 2 and 9, and enhanced expression and activation of extracellular signal regulated kinase (Erk) and integrin-linked kinase (ILK). Increased ILK expression correlated with the activation of protein kinase B (PKB)/Akt, phosphorylation of glycogen synthase kinase 3β (GSK-3β) and increased expression of cyclin E and cdk2 kinase. EGF withdrawal resulted in a more epithelial morphology and reversal of the EGF-induced activation of signaling pathways and pro-MMP activity. In contrast, treatment of EGF-treated cells with specific inhibitors of PI3 kinase, Mek or ILK inhibited the inhibitor-specific pathways. The inhibitors caused suppression of EGF-induced migration and
2
pro-MMP2/9 activities, but no change in the EGF-induced mesenchymal morphology. ILKsiRNA inhibited Akt phosphorylation and reduced pro-MMP2/9 activities, but had no effect on Erk activation or cell morphology. These results indicate that the EGF-induced morphological and functional changes in OSE are controlled by distinct signaling mechanisms working in cohort. EMT of OSE that is displaced by ovulation likely permits their survival and integration within the stroma under an altered, fibroblast-like identity. Failure of such mechanism(s) may lead to the formation of epithelium-derived inclusion cysts, which are known to be the preferential sites of malignant transformation.
3
Introduction Ovarian surface epithelium is a single flat-to-cuboidal layer of cells supported on the ovarian surface by a basement membrane and tunica albuginea. The cells are held together laterally by desmosomes and tight junctions (31).
Surface epithelial cells are continuous with the
mesothelium of the ovarian ligament and peritoneum. This single layer of epithelial cells contribute to the ovulatory process by lysis and reconstruction of the ovarian cortex, and is of major importance in gynecologic pathology since it is thought to be the source of 90% of ovarian neoplasms, the epithelial ovarian carcinomas (5, 28, 30). Ovarian surface epitheliumderived epithelial inclusion cysts are the preferred sites for the origin of epithelial ovarian cancer. These observations emphasize the need to determine the mechanisms regulating the postovulatory fate of ovarian surface epithelium, to better understand the initiation of its neoplastic transformation as well as the biology of ovulation.
Sequential interaction of cellular signaling pathways is operative during ovarian follicular development, ovulation and the post-ovulatory lutenization period (34).
During the pre-
ovulatory phase preferential growth of the ovulatory follicle brings it into close apposition with the ovarian surface epithelium (31). Ovarian surface epithelium produces lysosomal enzymes and urokinase as well as tissue type plasminogen activator just prior to ovulation (18) and mice deficient in the plasminogen activator gene functions have been shown to have a lower ovulatory efficiency (20).
Urokinase plasminogen activator (uPA) stimulates the release of
tumor necrosis factor α from thecal endothelium that progressively induces MMP expression and other inflammatory response (29).
Hence, collagenolysis precedes ovulation and is
accompanied by the exfoliation and displacement of a discrete region of ovarian surface epithelium close to the dome of the ruptured ovulatory follicle. In the post-ovulatory phase,
4
some of the exfoliated, displaced OSE undergo apoptosis (27), while others undergo repair and mend the surface injury (26). In addition, OSE might undergo EMT in response to ovulationinduced inflammatory mediators and migrate from the ovarian surface into stroma, or get trapped in the ruptured follicle where they are exposed to EGF and other EMT-inducing factors derived from the follicular fluid, blood, platelets and luteal cells. Hence, the process of repair and wound healing of scarred ovarian surface epithelium and the reshaping of displaced OSE to a mesenchymal migratory form occurs consecutively as part of the post-ovulatory ovarian remodeling process. While EMT presumably permits OSE to become incorporated into the ovarian stroma, epithelial inclusion cysts are derived from ovarian surface epithelium that does not undergo EMT within the stromal environment (4). Histopathological examination suggests that early malignant changes frequently occur in ovarian surface epithelium-lined crypts and epithelial inclusion cysts rather than on the ovarian surface (5), suggesting a potential link between the process of ovulation and the initiation of ovarian carcinoma.
Furthermore, the
risk of developing epithelial ovarian cancer increases with the frequency of ovulation (4). It is reasonable to assume, therefore, that EMT of OSE might reduce the chance of ovarian surface epithelium-derived malignant transformation. This phenomenon must be distinguished from EMT of advanced epithelial ovarian carcinomas, where EMT likely enhances invasiveness in a manner similar to other types of cancer.
Studies of the early events of ovarian carcinogenesis have been hampered by the minute amount of recoverable OSE from the human ovary and also by their limited life span in culture conditions where their epithelial phenotype is maintained. EGF has been described as a potent mitogen for human OSE (39). Human OSE has a significant expression of EGF receptor
5
(EGFR) in vivo and in culture (6). At ovulation, OSE are exposed to EGF released from platelets within blood clots and subsequently produced by stromal and luteal cells (23). In addition, transforming growth factor α (TGFα) binds to and activates EGFR in ovarian stroma and thecal cells (16). HC by itself is not mitogenic for OSE cells but it enhances the effect of EGF (39).
EMT, characterized by the dissociation of epithelial cells from epithelial sheets to migratory fibroblast-like cells is an important event in embryonic morphogenesis, gynaecological physiology and malignant transformation (7, 9, 44). However, EMT by cancer cells does not follow an orderly program and is different from physiological and developmental EMT (46). EMT induced by growth factors and cytokines requires reprogramming of epithelial cells in order to reshape for locomotion and invasion (24). In addition to increased motility the process can also induce proteolytic digestion of basement membranes upon which epithelia resides (24). Local expression of growth factors such as TGF-β, EGF, IGF, FGF-2 can initiate and facilitate this process by binding receptors with ligand-inducible intrinsic kinase activity (17). TGF-β induced response involves an increase in EGF receptors on the epithelial cell surface, and EGF or other EGFR ligands can assist in completing the response by initiating a diverse range of signaling pathways, the major ones being the Ras-Erk, Rac-Jnk-p38MAPK, PLCγ-1, PI3 kinase and downstream Akt pathways (12, 13, 48).
Depending on the cell type this
activation may result in a number of biological responses including mitogenesis, motility, protein secretion and differentiation.
6
To understand the physiology of OSE in the post-ovulatory microenvironment it is important to elucidate the response of these cells to stimuli present in that environment.
EGF was
previously shown to induce several components of EMT in OSE (36). In the present study, we therefore examined further phenotypic characteristics and the molecular pathways regulating EGF/HC-induced EMT in OSE. We report that EGF/HC-induced EMT encompasses increased motility and pro-MMP2/9 activity in OSE. Furthermore, we show that this EMT is dependent on the consecutive activation of Erk and ILK pathways, these pathways act in concert, and that the inhibition of either of these pathways has no effect on the complementary pathway. As a result there is an inhibition of EGF-induced functions without change in EGF-induced morphology. These results support the hypothesis that EMT of OSE that are trapped in the postovulatory follicle may be important for preventing the formation of epithelial inclusion cysts and thus provide protection from the initiation of neoplastic transformation.
7
Materials and Methods
Cells and Culture Methods: Appropriate ethical permits were obtained as required by the University of British Columbia, Canada. All women who participated in the study were premenopausal and between the ages of 29 to 49 years (mean age 36 years). No correlation with clinical parameters could be established as information about their hormonal status, stage in the menstrual cycle or use of oral contraceptives was not available. The cell culture method was used as described previously (36). Briefly, fragments of OSE were scraped from the ovarian surface during laparoscopic procedures for nonmalignant gynecologic conditions.
The
fragments were incubated in 1.0 ml of culture medium [199/MCDB105/10% fetal bovine serum (FBS)] in 35-mm culture dishes pre-coated with FBS at 370C in 5% CO2. After 2 h, 1.0 ml of more medium was added to the culture dishes. At confluence cells were sub-cultured with 0.06% trypsin and 0.01% ethylenediamenetetra-acetic acid. In certain cases OSE was grown in the presence of epidermal growth factor (EGF, 10 ng/ml) and hydrocortisone (HC, 1.0 µg/ml).
Preparation of cell lysates OSE growing in standard medium or treated with EGF/HC were lysed for 30 min on ice in lysis buffer (1% Nonidet P-40, 50 mM Hepes, pH 7.4, 150 mM NaCl, 2 mM EDTA, 2 mM phenylmethylsulfonyl fluoride, 1 mM orthovanadate, 1 mM NaF, 10 µg/ml aprotonin, 10 µg/ml leupeptin). Cell extracts were centrifuged at 10,000 g for 20 min with the resulting supernatant being the cell lysates used in assays. In some cases cells were treated with inhibitors PD 98059 (10 or 20 µM, Sigma, MO, USA) or LY 294002 (20 or 40 µM, Calbiochem, CA, USA) or KP 392 (50 or 100 µM, Kinetek, Vancouver, Canada) for 24 h before the preparation of cell lysates. The relative protein concentration of the cell lysates was
8
determined using a commercial protein assay kit with BSA standards according to the manufacturer’s instruction (BioRad Laboratories, USA).
Western blotting Cell lysates containing equal amounts of protein were electrophoresed on 10% SDS-PAGE gels under non-reducing conditions and transferred to immobilin-P membranes (Millipore, MA, USA). Antibodies used to probe Western blots were monoclonal anti-ILK (Becton and Dickinson, CA, USA), phospho-Erk, total Erk, phospho-Akt, total Akt, phospho-GSK3β, total GSK3 β (Cell Signalling, MA, USA), cyclin E (Upstate Biotechnology, NY, USA) and β-actin (Sigma, MO, USA). Bands were visualised by using peroxidase-labelled secondary antibody and an ECL detection system (Amersham, Buckinghamshire, UK) according to the manufacturer’s instructions.
Preparation of conditioned medium Cells were allowed to grow in 25-cm2 flasks in a standard medium or EGF/HC containing medium until they were 80-90% confluent. Serum-free conditioned medium was prepared as described previously (2). Protein estimation on the conditioned medium was performed using a commercial protein assay kit with BSA standards according to the manufacturer’s instruction (BioRad Laboratories, CA, USA).
Zymography Pro-MMP2 and pro-MMP9 activities in the conditioned medium of OSE were analysed using 10% SDS-gelatin (1mg/ml final concentration) zymography under non-reducing conditions as described previously (2). Gelatinolytic activity attributed by pro-MMP-2 and pro-MMP-9 was confirmed by activation with APMA (2 mM), a known activator of MMP or 1:10 phenanthroline (2mM) an inhibitor of MMP activation as described previously (2).
9
Wounding assay Four cases of low passage OSE were grown in standard medium with or without EGF/HC for 2 to 5 passages. Cells were grown to confluence, put on low serum (0.1%) for 24 h before wounding. Inhibitors were added 2 h before wounding and monolayers were wounded with the tip of a sterile 200 µl pipette. Dimethyl sulfoxide (DMSO) was used as a control vehicle. The wound was marked and measurements were taken using an ocular micrometer.
Ten
representative fields were marked and measured. Wounds were measured again after 6 h of wounding. Data were analyzed by one-way ANOVA (Analysis of Variance) followed by Bonferroni test. All data were considered significantly different from each other at p
