Epithelial Mesenchymal Transition and Cancer ... - Ingenta Connect

268

Current Cancer Drug Targets, 2010, 10, 268-278

Epithelial Mesenchymal Transition and Cancer Stem Cell-Like Phenotypes Facilitate Chemoresistance in Recurrent Ovarian Cancer N. Ahmed*,1,2,3,4, K. Abubaker1,3, J. Findlay1,2,4 and M. Quinn1,2 1

Women’s Cancer Research Centre, Royal Women’s Hospital, Melbourne, Australia; 2Department of Obstetrics and Gynaecology and 3Department of Surgery, University of Melbourne, Australia and 4Prince Henry’s Institute of Medical Research, Melbourne, Australia Abstract: Overcoming intrinsic and acquired chemoresistance is the major challenge in treating ovarian cancer patients. Initially nearly 75% of ovarian cancer patients respond favourably to chemotherapy, but subsequently the majority gain acquired resistance resulting in recurrence, cancer dissemination and death. This review summarizes recent advances in our understanding of the cellular origin and the molecular mechanisms defining the basis of cancer initiation and malignant transformation with respect to epithelial-mesenchymal transition (EMT) of ovarian cancer cells. We discuss the critical role of EMT frequently encountered in different phases of ovarian cancer progression and its involvement in regulating cancer growth, survival, migration, invasion and drug resistance. Using a model ovarian cancer cell line we highlight the relationship between EMT and the ‘migrating cancer stem (MCS) cell-like phenotype’ in response to drug treatment, and relate how these processes can impact on chemoresistance and ultimately recurrence. We propose the molecular targeting of distinct ‘EMT transformed cancer stem-like cells’ and suggest ways that may improve the efficacy of current chemotherapeutic regimens much needed for the management of this disease.

Keywords: Ovarian carcinoma, stem cell markers, EMT, migration, metastasis, differentiation, chemoresistance, recurrence. stem-like cells with the view to better management of this disease.

INTRODUCTION Epithelial ovarian cancer (EOC) is the fourth major cause of cancer mortality in women and constitutes 90% of all ovarian malignancies. Worldwide nearly 200,000 new cases are diagnosed and approximately 115,000 deaths occur annually from this devastating disease [1]. Due to lack of specific signs/symptoms and inadequate awareness about the disease on the part of general practitioners as well as the wider community, the disease is not diagnosed until advancedstages in most patients contributing to low overall cure rates. The five year survival for stage 1 patients is 90%, but it reduces to 30% in patients with advanced disease [2]. While surgery is the primary component of initial therapy in advanced-stage patients, almost all patients are treated with chemotherapy to eradicate the residual microscopic and macroscopic peritoneal metastases. Current treatment with paclitaxel and platinum based combination therapy results in a complete remission in 75% patients. Unfortunately, this remission is short lived lasting only for 16-21 months with subsequent recurrence and death as a consequence of metastatic spread [1]. In this review we discuss what is known about the initiation of ovarian cancer and the role of epithelial-mesenchymal transition (EMT) in facilitating the progression of this disease. We also shed light on the role of chemotherapy-induced EMT as one of the potential processes that may contribute to the development of chemoresistance and recurrence. In this context, we discuss the relationship between EMT and the acquisition of a ‘cancer stem celllike phenotype’ in response to chemotherapy. Emphasis has also been made on the development of new therapeutic approaches and how these would impact on EMT and cancer *Address correspondence to this author at the Women’s Cancer Research Centre, Royal Women’s Hospital, 20 Flemington Road, Parkville, Vic 3052, Australia; Tel: 61 3 8345 3734; E-mail: [email protected] 1568-0096/10 $55.00+.00

Pathology of Ovarian Cancer The initiating events in EOC are poorly understood. EOC is thought to arise directly from the surface epithelium of the ovary and its associated inclusion cysts or indirectly from benign ovarian lesions derived from the inclusion cysts [3]. The ovarian surface is composed of peritoneal mesothelial cells and is separated from the underlying stroma by a basement membrane. The cells are fragile and easily detached simply by handling of the ovary and are commonly lost from human surgical specimens. This underlines our limited knowledge about the pathophysiology of these cells and their contribution to the development of EOC. In females of all ages, particularly in postmenopausal women, surface epithelial inclusion cysts are commonly observed within the ovarian parenchyma [4]. The mechanism of their formation is still not known and there are several theories that support the formation of these cysts. The ‘incessant ovulation theory’ of ovarian cancer development suggests that these cysts are formed when the released ovarian surface epithelial cells resulting from the follicular rupture during the ovulation get trapped into the ovarian parenchyma due to repair defects on the ovarian surface [5]. As the postovulatory ovarian surface epithelial cells have the ability to transform to fibroblast-like migratory cells (for reasons still not known) [3], the trapping of released cells may potentially result due to a deficit in a migratory EMT-like process resulting from defective wound repair [3, 6]. These findings suggest that EMT is a homeostatic mechanism to maintain intact ovarian surface epithelium, the lack of which may result in the formation of epithelial inclusion cysts [3]. The evidence that the inclusion cysts are the originators of ovarian carcinomas may be supported by immunohistochemical © 2010 Bentham Science Publishers Ltd.

EMT and Stem Cell-Like Phenotypes Facilitate Chemoresistance in Ovarian Cancer

identification of several ovarian carcinoma markers (CA125, CA19-9, E-cadherin, placental-like alkaline phosphatase) which are observed up to 10 times more frequently in inclusion cysts than on the surface epithelium [7, 8]. Moreover, most early carcinomas of the ovary appear to be confined within the organ without the involvement of the surface epithelium, suggesting that the inclusion cysts are more prone to the development of cancer than the surface epithelium on its own [4]. Hence, inclusion cysts are the preferred sites for dysplastic changes that lead to ovarian cancer. Consistent with that, ovarian surface epithelial cells from women with a family history of ovarian carcinoma have been shown to have reduced capacity to undergo EMT and contain more inclusion cysts than women with no family history of a carcinoma [3]. This suggests a link between the lack of induction of EMT in the postovulatory ovarian surface epithelial cells and the initiation of ovarian cancer. Aberrant differentiation is a unique aspect of ovarian cancer biology as tumors acquire the complex differentiation pattern of fallopian tube (serous carcinoma), endometrium (endometroid carcinoma), endocervix (mucinous carcinoma) and vagina (clear cell carcinoma) [4]. Each subtype has identifiable precursor lesions and multiple early genetic alterations [9]. Hence, mutations in the early progenitor cells in the inclusion cysts may be the contributing factors in the initiation of ovarian carcinoma. Two groups of ovarian tumors are distinguished based on clinical, pathological and genetic studies [9, 10]. Type I tumors include low-grade micropapillary serous carcinomas, mucinous, endometrioid as well as clear cell carcinomas demonstrating genetic instability and mutations in K-ras, BRAF, -catenin, PI3-kinase pathway and/or PTEN [10]. As these genetic alterations occur early in the transformation process, the contribution of these genes defines the tumorigenic process as well as histological differentiation. In contrast, type II carcinomas consist mostly of high-grade serous carcinomas, carcinosarcomas and undifferentiated carcinomas. They display high levels of genetic instability and are characterized by mutation of p53 [10]. In addition, substantial numbers of type II carcinomas also develop outside the ovary (peritoneum, fallopian tube). Hence, one can propose that the morphological diversity of ovarian cancers can arise from either ‘ovarian cancer stem cells’ or ‘early progenitor cells’ undergoing transforming mutations resulting in ‘multi-potent cancer stem cells’ responsible for the origin of various subtypes of EOC. Understanding ovarian cancer stem cells will provide unifying insights into the progression of this histologically diverse disease. TRANSITION FROM EPITHELIAL TO MESENCHYMAL PHENOTYPE AND THE PROGRESSION OF CANCER EMT is defined by a complex molecular and cellular programme by which epithelial cells lose their differentiated characteristics such as cell-cell adhesion, apical-basal polarity and lack of cell motility and gain mesenchymal features including motility, invasiveness and increased resistance to apoptosis [11-13]. Several observations in human tumors and experimental animal models have provided convincing evidence for the physiological role of EMT in tumor progres-

Current Cancer Drug Targets, 2010, Vol. 10, No. 3

269

sion [13]. The EMT programme in malignant cells is triggered by various signals, some of which are released in the form of cytokines and growth factors by the cells constituting the stroma of the tumors [14]. The transforming growth factor (TGF) family of cytokines are the best characterized EMT inducers in cancer pathogenesis [15]. TGF also influences the activities of other growth factors such as platelet derived growth factor (PDGF) and fibroblast growth factor (FGF) [16-18] some of which can act independently or in concert to trigger the EMT programme [19]. Moreover, hypoxia can induce EMT in tumors through the up regulation of hypoxia-inducible factor-1 (HIF) and repression of E-cadherin expression by enhancing the transcription of Snail1 and Twist1, activation of the Notch and NFK pathways and induction of DNA hypomethylation [20]. Several recent studies have recognized the role of non-coding RNAs in activating EMT programme in cancers [21]. In this context, the role of miR-200 and miR-205 families has been well characterized in regulating EMT [22]. Evidence of EMT in Ovarian Cancer Clinical manifestation of advanced-stage EOC includes the ovary and omentum with diffuse intraperitoneal metastases and malignant ascites [23, 24]. The early steps of EOC progression involve disruption of the ovarian tumor capsule and shedding of malignant cells from the primary tumors into the peritoneum where they survive as single cells, multicellular aggregates or spheroids free floating in ascites. Under this scenario, localized proteolytic degradation of mesothelial extracellular matrix facilitates the migration of floating cells allowing them to anchor as secondary lesions on pelvic organs and at a later stage, metastasize to distant organs [25]. Most of the women with advanced-stage ovarian cancer have localized peritoneal spread with no clinically apparent hematogenous metastases, indicating that the spread of ovarian cancer follows a unique pattern distinct from other cancers [26]. Hence, processes such as cell-cell, cell-ECM, localized intraperitoneal migration and invasion of the peritoneal lining by floating single cells or spheroids play a dominant role in cancer progression. Unlike most other carcinomas that dedifferentiate during neoplastic progression, most ovarian carcinomas persist as differentiated epithelial tumors. It has been suggested that primary EOC may undergo an EMT-like process during localized invasion in the peritoneum and retain mesenchymal features in advanced tumors [24]. This mainly occurs as a result of fewer intercellular junctions resulting from reduced cell-cell adhesion due to loss of E-cadherin function or expression and gain in N-cadherin expression [23]. Loss of Ecadherin expression and/or function with the acquisition of mesenchymal phenotype has been clinically associated with poor prognosis in many tumors and down regulation of Ecadherin has been shown to result in a mesenchymal phenotype [27]. Even though, E- and N-cadherin expression are central to EMT, in recent years it has become evident that ‘incomplete or partial EMT’ can occur in tumors whereby cancer cells retain epithelial features and cell-cell contacts while still able to migrate and invade like in a classical EMT process [12]. This intermediate phenotype is observed in ovarian surface epithelium which expresses epithelial markers including keratin, desmosomes, tight junctions, laminin,

270 Current Cancer Drug Targets, 2010, Vol. 10, No. 3

Ahmed et al.

collagen type IV and apical microvilli, as well as mesenchymal markers such as vimentin and collagen types I and III [3]. Ovarian surface epithelium expresses low levels of Ecadherin, mostly in the inclusion cysts while the epithelial integrity is maintained primarily by N-cadherin, consistent with the epithelial-mesenchymal phenotype of the tissue [3]. Although enhanced E-cadherin expression has been demonstrated in well-differentiated tumors and in the tumor cells of the ascites, its expression is most commonly lost in metastasis and an increased trend of retention of N-cadherin expression in metastatic lesions has been reported [24]. Even though an inverse relationship between E-cadherin expression and invasiveness of tumor cells has been described, this is not always a universal feature in all carcinomas, especially in ovarian tumor ascites spheroids [24, 25, 28]. E-cadherin expressing ovarian carcinoma spheroids has been shown to adhere to and invade the surrounding mesothelium [24, 29, 30]. Spheroids undergo reduced proliferation and limited drug penetration resulting in decreased susceptibility to chemotherapy [31] and hence, mimic traits of stem-like cells. Over all, collective data from recent literature suggests that the EMT-like process plays a critical role in the progression of the disease. CANCER STEM PROGRESSION

CELLS

(CSCs)

AND

TUMOR

Genetic and epigenetic alterations in adult tissue-specific stem cells have been suggested to lead to malignant transformation into cancer progenitor cells also designated as cancer initiating or CSCs [32]. This cancer progenitor model of tumor progression is supported by the fact that the small number of progenitor cells isolated from patient’s malignant specimens can give rise to the bulk of the tumor containing cancer cells at a more advanced stage of differentiation [33]. Some recent reports have also demonstrated a very small sub-population of cancer progenitor cells isolated from wellestablished cancer cell lines maintained under undifferentiated or poorly differentiated conditions during long term cultures [34]. In such a model both primary tumors and metastatic lesions should display a similar genetic profile as both populations are derived from the same lineage of cancer stem cells. The histological heterogeneity of ovarian cancer can be explained by the existence of multipotent cancer stem cells arising from somatic ovarian stem cells. According to that model, ovarian somatic stem cells would be expected to divide asymmetrically yielding two daughter cells one destined towards differentiation while the other undergoing selfrenewal. Repeated asymmetric self-renewal may lead to accrual of mutations over time which ultimately may lead to the development of histological subtypes of progenitor cells leading to sub-type specific malignant progression. Among the methods most frequently used to identify cancer stem cells are the fluorescence-activated cell sorting (FACS) using specific antibodies directed against one or several stem cell-like surface markers [35-37] and the Hoechst dye efflux sorting of the side population of cancer stem-like cells (SP) [38, 39]. Numerous putative markers are currently under investigation to enrich the stem cell population. Among the most common cell surface markers used for the isolation and characterization of cancer stem cells are

CD44+CD24- in breast cancers [40], CD44+21+ integrinCD133+ in prostate [41], CD133+ in glioblastomas [42], colon [43] and prostate [44] cancers, 21+ integrin in prostate cancer [45], EpCAM+ in hepatocellular carcinomas [46] and many others. Besides these, several nuclear stem cell markers such as Bmi-1, Nanog, Oct-4 [43] and the expression of multi-drug-resistant proteins and related members of the ATP-binding cassette (ABC) transporter family, such as ABCG2 which facilitate the efflux of chemotherapy drugs [47] are used to characterize the isolated cancer stem cell populations. To date few studies that have revealed the presence of CSCs in primary ovarian and ascites tumors cells have focused on the identification and characterization of a stem cell population with a phenotype of CD117+/CD44 + [48], CD44+/MyD188+ [49] or anoikis resistant tumor spheres [50] or on the identification of a side population cells [51]. CD133+ cells were more frequently observed in primary ovarian tumors than in their metastatic lesions or in normal ovaries [52], and the stem cell factor (c-KIT, CD117) is over expressed in chemoresistant ovarian carcinomas [53] and a higher percentage of CD44+ cells were observed in metastatic tumors [49]. Therefore, one can postulate that the isolation of CD44+/CD117+/CD133+ cells from ovarian tumors should result in the identification of enriched ovarian CSCs. These studies would provide an initial platform to understand the much needed biology of CSCs of the ovary. However, a future aim will be to use novel stem cell markers so that clinically relevant targets can be identified. The role of CD44 has been well characterized in several tumors [54]. CD44+/high CD24-/low fractions of breast cancer cells have significantly greater tumorigenic potential in vivo than CD24+/highCD44-/low cell fractions [40]. The use of CD24 and CD44 cell surface markers for the cancer stem-like cell identification has recently been assessed by serial analysis of gene expression (SAGE) from normal and neoplastic human breast tissues [55]. The expression profiles of stem-like cells from both normal and neoplastic breasts were very similar and both tissue types expressed several stem cell markers. In spite of the similarities, significant differences were found between normal and breast cancer stem cells. Unlike normal stem cells that represented a very small population of the tissue mass, cancer stem cells constituted a significant portion of the tumor mass. In breast cancer, CD44+/CD24- phenotype constituted 12-60% of the whole tumor mass, whereas in colon, CD133+ phenotype stem cells ranged between 3.8-24.6% of the total tumor cells [56]. A recent mathematical calculation determining the probability of tumor growth after treatment demonstrates that cancer stem cells may represent the majority of tumor cells in advancedstage tumors [57]. Hence, cancer stem cells generating new tumors may not be as rare as previously thought and enriching for a relatively common cell phenotype [such as CD44, epithelial cell adhesion molecule (EpCAM), 21 integrin, etc] with a function different than in the normal cells could be important. Sphere Formation and the Cancer Stem Cell Phenotype of the Ovary In vitro enrichment and propagation of cancer stem cells is achieved by growing cells in an unattached condition in the form of ‘spheres’. In brain tumors, CD133+ cells can


grow successfully in an unattached neurosphere-like form while CD133- cells cannot [42]. Cells from glioblastomas cultured as neurospheres expressed numerous stem cell-like markers and were highly tumorigeneic and resistant to radiation [58]. Similar findings were obtained for breast cancer cells enriched in CD44+/highCD24-/low populations [40] and ovarian cancer cells enriched in CD44+ and CD117+ populations [48]. This method of enriching stem cell populations in a sphere culture raises several questions. Firstly, the formation of tumor spheres is directly dependent on autocrine regulation by tumor cells and is devoid of paracrine interactions with supporting cells (stromal, endothelial, infiltrating white blood cells, etc) and the extracellular matrix (ECM) which makes the ‘stem cell niche’. Hence, one can question if the sphere cultures without any interacting ECM or supporting cells can imitate the true phenotype of cancer stem cells. Even though the scenario of enriching the cancer stem cell population by sphere formation may not seem relevant to many solid tumors, it is applicable to ovarian cancer, where metastasis is significantly dependent on the survival of shed tumor cells which survive as cellular aggregates or spheroids with subsequent attachment to the peritoneal lining of the abdomen. Under these circumstances one can assume that the tumor spheroids imitating ‘ovcaspheres’ are in direct contact with the ascites microenvironment which can be described as a ‘cancer stem cell niche’ due to the availability of growth factors and circulating stromal and other supporting cells (such as endothelial cells, tumor associated mononuclear cells, etc). These ‘ovcaspheres’ are slow proliferating, resistant to anoikis and survival threatening insults - characteristics common in cancer stem-like cells. As the cellular processes enabled in the spheroids or ‘ovcaspheres’ are similar to cancer stem cells, one can question if the ascites environment is ideal as a ‘cancer stem cell niche’ facilitating the enrichment of ‘stemness’ genes in self-renewable chemoresistant population of tumor cells in recurrent patients. These observations may link cancer stem cells to recurrence and chemoresistance, a major hurdle in the management of advanced-stage ovarian cancer. Moreover, recent studies have demonstrated proteolytic processing of E-cadherin to soluble E-cadherin ectodomain (sE-cadherin) found in the ascites of ovarian cancer patients [59, 60]. In addition, the ‘sEcadherin-rich’ ovarian tumor microenvironment has been shown to facilitate cancer cell dissemination by inducing EMT-like processes [60]. These data suggest that the role of EMT in ovarian cancer dissemination is critical for primary tumor cells to disperse into the peritoneum for subsequent attachment to the secondary sites. This induction of EMT in the peritoneum of ovarian cancer patients can be correlated with the paradigm of migrating cancer stem (MCS)-cells hypothesized for colorectal cancer progression [61]. In the MCS-cell concept, a decisive step for tumor cells to migrate at the tumor-host interface triggers aberrant signals that lead to the induction of EMT. According to this concept most of the cancer stem-like cells are located at the invading edge of the tumors. In ovarian cancer metastasis one of those decisive steps may be the cleavage of E-cadherin that facilitates the induction of EMT resulting in the migration of tumor cells. Interestingly, both EMT and acquisition of ‘cancer stem-like phenotype’ are regulated by fundamental embryonic pathways both of which are aberrantly activated in cancer [62]. Hence, both EMT and MCS-cell concept are closely


271

related but have distinct cellular processes which may be crucial for the progression of ovarian cancer. ASSOCIATION OF EMT AND CSCS: A MERGER FOR POTENTIAL CHEMORESISTANCE IN OVARIAN CANCER Treatment for advanced EOC is difficult because of the inability to completely resect the diffuse tumor spread within the peritoneum and the eventual resistance of tumor cells to chemotherapy. Despite favourable initial responses, most of these tumors relapse and eventually become chemo resistant. Platinum-based drugs (cisplatin) in combination with paclitaxel are the first-line chemotherapeutic agents used for the treatment of EOC [1]. Cisplatin exerts its effect by forming intra and inter-strand platinum-DNA cross-links which if inadequately repaired kills cancer cells [63]. Paclitaxel on the other hand exerts its effect by stabilizing the microtubules and induction of cell-cycle arrest in G2-M phase and activation of apoptotic signals [64]. Several clinical trials have been undertaken in which cisplatin in combination with other chemotherapy agents have been tried for advancedstage ovarian cancer patients [2]. Unfortunately, these treatment protocols had little success with a progression-free survival interval of only 16-22 months in the majority of cases. In short, the present chemotherapy regime on the survival of ovarian cancer patients is disappointing and a better understanding of the mechanism by which chemoresistance occurs is necessary to improve treatment outcomes. Recent literature suggests molecular and phenotypic associations between chemoresistance and the acquisition of an EMT-like phenotype in cancer cells [65]. In recent reports, acquisition of an EMT phenotype has been demonstrated in gemcitabine resistant pancreatic cancer cells [66], oxaliplatin resistant colorectal cancer cells [67], paclitaxel resistant ovarian cancer cells [68] and tamoxifen-resistant breast cancer cells [69]. Consistent with these findings, residual breast cancers after conventional therapy have been shown to display a mesenchymal phenotype and tumor initiating features [70]. This is also consistent with the frequent expression of stem cell and EMT markers in circulating tumor cells of metastatic breast cancer patients [71]. These studies suggest that EMT and cancer stem-like cell phenotypes are closely related [72]. However, it is not clear whether the CSCs responsible for initiating tumors are the same as the ones that survive treatment and regrow as tumor after chemotherapy. Nonetheless, the concept that the induction of EMT can generate cancer stem-like cells is supported by recent studies in breast cancer. One nicely demonstrated the induction of expression of EMT and stem cell-like programmes by TGF treatment [73], and another showed an increase in the number of stem-like cells by the introduction of transcription factors such as Twist1 or Snail1, responsible for the repression of E-cadherin [27, 74]. Consistent with these observations, hypomethylation of genes specifying transcription factors that programme stem cell phenotypes leads to EMT in cancer cells [75]. In a most recent report, EMT inducing E-cadherin transcriptional repressors Snail and Slug have been shown to impose acquisition of the CSC-like phenotype and chemoresistance in ovarian cancer cells by overcoming p53-mediated apoptosis [76]. These findings suggest that the

272 Current Cancer Drug Targets, 2010, Vol. 10, No. 3

Ahmed et al.

Fig. (1). Morphological change in OVCA 433 cells in response to cisplatin: (A) parental OVCA 433 cells; (B) 5 days treatment with cisplatin (5μg/ml); (C) 1 month treatment with cisplatin (5μg/ml) (OVCA 433-LT); and (D) passage 6 in normal medium after 1 month treatment with cisplatin (OVCA 433-LT). Phase contrast images are representative of 5 independent experiments. Magnification-100 X; scale= 50μm.

initiation of the EMT programme in ovarian and other cancer cells may be critical for the acquisition of stem cell-like characteristics resulting in acquired chemo resistance. Cisplatin Induced EMT Generates Ovarian Cancer Stem-Like Cells: A Study on the OVCA 433 Cell Line as an Experimental Model The human ovarian OVCA 433 cell line, originally derived from the ascites of an advanced-stage ovarian cancer patient [77], was treated with cisplatin (5μg/ml) for 1 month. OVCA 433 cells were maintained in normal growth medium for several passages (OVCA 433-LT cells). Cisplatin treated cells displayed a mesenchymal morphology (Fig. 1B) within 48 hours which was maintained for several passages (~passage 16) even after the withdrawal of the drug (Fig. 1D). Almost all OVCA 433 cells surviving cisplatin treatment (5μg/ml) displayed elongated, irregular fibroblastoid morphology (Fig. 1 C) compared to rounded cobblestone epithelial appearance of parental OVCA 433 cells (Fig. 1A). After the removal of the drug and a lag period of 3 weeks in normal medium, the growth of mesenchymal cells (OVCA 433-LT) resumed, and within 3-4 passages, the proliferation rate was similar to that of parental untreated cells. Consistent with the morphological change, the epithelial marker Ecadherin was greatly reduced while the expression of mesenchymal markers such as N-cadherin and vimentin were greatly enhanced in OVCA 433-LT cells (Fig. 2). In addi-

tion, OVCA 433-LT cells demonstrated increased migration (wound healing assay) compared to control untreated cells (Fig. 3). Treatment with cisplatin also resulted in enhanced expression of CD44, 2 integrin subunit (no change in 1 integrin subunit), ABCG2 transporter and CD117 compared

Fig. (2). Expression of EMT markers in response to cisplatin: (A) Western blot analysis of EMT markers (E-cadherin, N-cadherin & vimentin) in the cell lysates of OVCA 433-LT (passage 6 in normal medium) compared to parental untreated cells was performed as described previously [81]. Protein loading was checked by stripping the membrane and staining with anti--actin. This is representative of three independent experiments.



273

Fig. (3). Migration of OVCA 433 cells in response to cisplatin: (A) The migratory response of parental untreated cells and OVCA 433-LT cells was determined by the wound healing assay as described previously [81]. Phase contrast images are representative of 3 independent experiments. *p