Spontaneous malignant transformation of ovarian surface epithelial cells correlates with EMT alteration and stemness acquisition

In order to complete the physiological function of postovulatory repair during repeated ovulation, the ovarian surface epithelium (OSE) not only has to undergo epithelial-mesenchymal transition (EMT), but also possess the properties of somatic stem-like cells. However, there is no evidence to indicate that both EMT alteration and stemness acquisition are linked to epithelial ovarian carcinomas.Methods In this study, we established a cell model of spontaneous oncogenic transformation of mouse OSE (MOSE). The cell proliferation was assessed using clonogenic survival and soft agar. The self-renewal of cancer stem-like cells (CSCs) was determined by spheroid culture. CD44 + /CD117 + cells were analyzed using flow cytometer. The PCR array was used to determine the EMT-related mRNA level. Expression of pan-keratin, vimentin, E-cadherin, Snail1 and Slug were detected using western-blotting and immunofluorescences, respectively. The tumorigenesis were monitored by limiting dilution assay in vitro and in vivo .Results

These results indicated that both EMT alteration and stemness acquisition were closely correlated with the spontaneous malignant progression of MOSE cells. Our findings provide new insights into the future to combat epithelial ovarian carcinomas.

Background
Ovarian cancer is a deadly gynecologic malignancy with a five-year survival rate of less than 30% [1]. The poor prognosis is largely due to difficulties in its early diagnosis and the high rate of relapse as a result of chemoresistance. Epithelial ovarian carcinomas (EOC) account for 90% of all ovarian malignancies [2] and comprise multiple subtypes [3].
The early events of EOC development are poorly understood and the exact initiating cell population remains to be defined.
Ovarian surface epithelium (OSE), a single layer of squamous or cuboidal cells, plays physiological roles in both follicular rupture and subsequent ovarian remodeling by altering the motility and proliferation response necessary for extracellular matrix remodeling. OSE can undergo proliferative repair, and its normal regulation may contribute to the pathogenesis of EOC [4]. In addition, OSE has been postulated as the source of EOC base on pathological observations and experimental approaches [5][6][7][8].
The capacity of OSE that undergoes epithelial-mesenchymal transition (EMT) in response to postovulatory stimuli has been proposed to confer advantage to the postovulatory repair of the epithelia cells [9]. Hence, EMT is a part of normal OSE physiology and failure in this process may be the reasons of ovarian cancer initiation and metastasis.
Additionally, the cyclic pattern of repeated disruption and repair with complex remodeling associated with ovulation leads one to intuit the existence of a population of somatic stem cells that would be responsible for these processes. OSE with some stem-cell properties have been identified previously based on their slow proliferation [10]. In addition, OSE at the junction area contains cancer-prone stem cell niche [11]. Moreover, accumulating evidence supports that EMT and cancer stem cells (CSCs) play critical roles in the development of chemoresistance, tumor relapse and metastasis in ovarian cancer patients [12][13][14].
In order to complete the physiological function of postovulatory repair during repeated ovulation, OSE not only has to undergo EMT, but also possess the properties of somatic The isolation and culture of human and mouse ovarian surface epithelium have been described [24,27,28]. Therefore, taking advantage of the spontaneous transforming of OSE cells, we established a step-wise neoplastic transformation model of these cells from a premalignant phenotype to the malignant one. We also demonstrated that the alteration of EMT and acquisition of stem-like properties were associated with the neoplastic process of mouse OSE cells.

Cell isolation and culture
Mouse ovarian surface epithelial (MOSE) cells were isolated as described by Roby and Paul The total RNA was isolated from 10 6 cells using TRIZOL (Life Technologies). cDNA was synthesized from 500 ng of the total RNA using the RT 2 first strand fit (SABiosciences, Frederick, MD). After all the control tests, the samples were analyzed using the mouse EMT RT 2 profiler PCR array (PAMM-090Z, SABiosciences, Frederick, MD) that profiles the expression of 84 key genes. Altogether 84 different genes were simultaneously amplified in each sample. Five house-keeping genes (B2M, HPRT1, RPL13A, GAPDH, and ACTB), genomic DNA contamination control, reverse transcription control and positive PCR control were included in each PCR array. Briefly, the reaction mix was prepared from 2× SABiosciences RT 2 qPCR master mix and 102 ml of sample cDNA. Ten ml of this mixture was added into each well of the PCR array. PCR arrays were performed in 384-well plates on a 7500 real time PCR system (Applied Biosystem, Foster City, CA).

Spheroid culture
Sphere formation assay was performed using serum-free DMEM/F12 Medium, including 20 ng/mL mEGF, 20 ng/mL basic fibroblast growth factor, 100 U/mL penicillin and 100 µg/mL streptomycin, 2 mg/ml insulin, 4 mg/mL heparin sodium, and 6 mg/mL glucose. Late MOSE cells were plated at a density of 1000 cells in 35 mm dish and cultured for 10 days. Round cell clusters larger than 100 mm were judged as spheres. Data was presented as sphere formation efficiency (%) = (number of spherical formation/plated cells × 100%). Data points in figures represent three independent experiments.

Limiting dilution assay
Limiting dilution assay was performed to measure the number of cells required to generate at least 1 tumor sphere/well as previously described [29]. Briefly, serial two fold dilutions of M-I (from 0 to 5000 cells) and M-L cells (from 0 to 500 cells) were sorted into ultra-low 96-well plates with 6 wells per dilution, respectively. Cultures were fed 50 ml of CSCs medium every 2 days until day 10. Fraction of wells without spheres (y-axis) was plotted against the number of cells plated per well (x-axis). Regression lines were plotted and x-intercept values calculated, which represent the number of cells required to form at least 1 tumor sphere in every well.
Secondary antibodies were purchased from Amersham Biosciences (Piscataway, NJ) and the immune complexes were detected using an enhanced chemiluminescence (Millipore, Boston, MA) method according to the manufacturer's instructions.

Immunofluorescences
Cells were plated into micro-slides, and then fixed in 4% paraformaldehyde. Cells were permeabilized with 0.5% Triton followed by blocking with 5% BSA for at least 30 minutes at room temperature (RT). Incubations with primary antibodies were carried out for overnight at 4℃. Appropriate secondary antibodies conjugated to either AlexaFluor 488, AlexaFluor 543, or AlexaFluor 635 were carried out for 30 min at RT. Nuclear was counterstained with 4',6-diamidino-2-phenylindole (DAPI). Image was observed using SP5 confocal laser scan microscope (Leica, Wetxlar, Germany).

Flow cytometry assay
For stem cell marker analysis, 10 6 cells were resuspended in PBS, incubated with fluorescence-conjugated monoclonal antibodies against CD44-FITC and CD117-PE (BD Pharmingen) at 4℃ for 10 min, and washed by PBS for 3 times. The samples were analyzed by FC500 flow cytometer (Beckman Coulter, Fullerton, CA).

Histology and immunohistochemistry
Resected ovaries and tumors were collected from mice, fixed in 4% paraformaldehyde.
Routine paraffin embedding of tissues and hematoxylin and eosin staining of tissue sections were performed by histopathological laboratories of Soochow University (Suzhou, China). For immunohistochemistry, tissues were cut into 5 µm sections. Slides were deparaffinized and rehydrated in an ethanol series. The sections were treated with 3% H 2 O 2 /MET for 20 min and boiled with citrate buffer (0.01 M, pH 6.00), then incubated overnight with indicated primary antibodies at 4℃. After that, followed with PBS washed for 3 times, the slides were then incubated with secondary antibodies with HRP. The slides were incubated with DAB solution, and then counterstained with hematoxylin according to standard protocols.

Statistical analysis
Data were presented as mean ± standard deviation (SD), and were analyzed with one way analysis of variance (ANOVA) and Student's t-test. P <0.05 were considered statistically significant.

It is well known that primary epithelial cells stop replication after ~10 passages in vitro.
Owing to the specific physiological role of ovarian remodeling during ovulation, it is possible that OSE continues to proliferate when subcultured in vitro. We wondered whether establishing a cell model of spontaneous epithelial ovarian cancer represented distinct transitional states of neoplastic progression. We first isolated primary MOSE cells To further elucidate these transitional stages, we determined the capacity of anchorageindependent growth, an in vitro hallmark of neoplastic transformation of cells. Both early (<20 passages) and intermediate (21-80 passages) MOSE cells were unable to form colonies in soft agar. Notably, late-passage (>81) MOSE cells were capable of forming >30 mm colonies but at diminished efficiency compared with human ovarian cancer SKOV-3 cells ( Fig. 2A). Compared with MOSE cells at earlier stages, late-passage MOSE cells exhibited dramatically increased plating efficiency and growth rate, another proliferative parameter that is often associated with neoplastic change (Fig. 2B and C). The average chromosomal number observed in early passages MOSE cells was 49, and 36% of cells at this stage were hyperdiploid. But as the cells progressed, there was a trend toward increased numeric abnormalities. In the late-passage cells, the average chromosome number was 64, with 43% were near-tetraploid and hypertetraploid (Fig. 2D). Therefore, based on morphological change, chromosomal number and proliferating ability, we defined three sequential stages of transformed MOSE cells as M-E (≤20 passages), M-I (21-80 passages) and M-L (≥81 passages) cells, respectively.

Tumorigenicity of MOSE cells in vivo
To assess oncogenic transformation of MOSE cells in vivo, we subcutaneously injected them into athymic nude mice. As expected, M-L cells formed tumors in three of three animals within four weeks post-injection. In contrast, M-E cells did not form any tumor by 12 weeks post-injection (Fig. 3A). Although M-I cells formed smaller nodules with average volume of 17 mm 3 , HE staining showed only infiltrated lymphocytes ( Fig. 3B and C).
However, M-L cells developed relative fast growing tumors with an average volume of 1475.86 mm 2 , and showed malignant cells with a large nuclear to cytoplasmic ratio and neovascularization (Fig. 3). These tumor nodules showed high level of both Pan-keratin and Vimentin, indicating that they were similar in phenotype to that of epithelial ovarian cancer [9, 19, 30] (Fig. 3D). These results showed that MOSE cells, over time,  Mmp3, Col3a1, and Esr1. These genes were strongly reduced in malignant progression of MOSE cells (Fig. 4A, 4B and 4C). Furthermore, the mRNA levels of EMT-induced transcription factors (Snail, Slug, Tcf4 and Tgfb2) were significantly up-regulated, and those of epithelial markers (Krt19 and Krt7) were down-regulated in M-L cells (Fig. 4A).
Indeed, the results of immunofluorescence studies also showed an increased expression of Snail and Slug in M-L cells, compared with M-E cells, as demonstrated by western-blotting analysis (Fig. 4D). Interestingly, level of E-cadherin and b-catenin in M-I cells increased dramatically (Fig. 4E). Collectively, as illustrated by the mesenchymal morphology of M-L cells shown in Figure 1A, these results indicated that the neoplastic transformation of MOSE was accompanied by continuous EMT-inducing signals including notable increase of Snail and Slug.

Stem-cell-like properties were correlated with neoplastic transformation of MOSE cells
There is recent evidence that OSE at the junction area contains a cancer-prone stem cell niche [10, 11, 31]. We wondered whether the neoplastic transformation of normal surface epithelial cells correlates with acquisition of stemness. So, the stem cell-like properties of MOSE cells were monitored during their long-term culture in vitro. The percentages of CD44 + and CD117 + , markers for ovarian cancer stem cells, were found to be up-regulated post neoplastic transformation of MOSE cells (Fig. 5A). Consistently, mRNA of stem-related genes, such as Nanog and Sox2, gradually increased and KLF4 decreased during malignant progression of MOSE cells (Fig. 5B). Interestingly, the M-L cells displayed obviously selfrenewal properties, which formed larger and more spheres from a single cell suspension under serum-free medium. The sphere forming efficiency (SFE) was markedly higher in M-L cells, compared with that in M-E and M-I cells (Fig. 5C). These findings suggested that stem-like cells in ovarian surface epithelium might be involved in the progression of

spontaneous neoplastic transformation of MOSE cells in vitro.
A single tumor sphere resulting from the proliferation of a single cancer stem cell coupled with limiting dilution analysis allows for the determination of the minimal frequency of repopulating tumor sphere cells within the cell population. Figure 5D  M-E cells represent the earliest transitional state of our model, characterized by typical cobblestone-epithelial appearance, a slow growth rate, and inability to anchorage-independent growth. They favor strong cell-substratum as well as cell-cell contact.
Interestingly, in this stage, the stem and/or progenitor cell markers, such as CD44 and CD117 were often detected, even though in low proportion. The existence of somatic stem cells is likely to reside in distinct tissue compartments of the In accordance with other reports [19,24], we also found that MOSE cells exhibited both mesenchymal and epithelial markers continuously. This mixed phenotype enables the OSE cells to respond rapidly to a variety of environmental, hormonal, and stress factors, but it is also thought to contribute to the onset of neoplastic transformation by rendering the cells more susceptible to transition from an epithelial to a mesenchymal-like phenotype.
M-L cells represent the most aggressive transitional stage and display features similar to tumor cells, such as increased proliferation, and above 50% triploid and tetraploid cells. There are several characteristics of EOC which indicate that it may be a stem cell-driven disease. Such as, EOC can generate differentiated subtypes that recapitulate the histology of other normal gynecologic tissues and the high rate of chemoresistance and recurrence after successful initial treatment. In clinical and experimental studies, the high level of stem-cell markers, such as Nanog, ALDH-1, CD44, CD117, CD133 and Sox-2, were found to be associated with poor outcomes of EOC patients [30,[37][38][39][40]. CSCs contribute to the aggressive behavior of EOC [38,41]. Our findings also suggest that acquisition of stemness is associated with malignant spontaneous transformation of OSE cells in long-term culture.
The authors declare that they have no competing interests.

Funding
This study was supported by the National Natural Scientific Funding of China (grant No. 81872622, 81673151, U1832140, and 81372979), and BL is the Principle Investigator. The funding source had no role in the study design, data collection and analysis, the decision to publish, or the preparation of this manuscript.     Limiting dilution assay showed that M-L cells possessed higher capability of