Sheep stromal-epithelial cell interactions and ovarian tumor progression

Abstract

Previous studies suggest that underlying ovarian stromal cues may regulate the ovarian surface epithelium. However, little is known about the interaction between ovarian stromal cells (OSC) and ovarian surface epithelial cells (OSE) under normal physiologic and pathologic conditions, largely because of the lack of a suitable model. In the current study, the OSC obtained from a sheep were immortalized with SV-40 T/t antigen (designated IOSC) and telomerase reverse transcriptase (designated IOSCH), followed by transfection with the oncogenic allele of the human H-Ras oncogene (designated IOSChR). IOSC cells transfected with H-Ras before immortalization with telomerase were designated IOSCRH. These sheep OSCs were used in both in vitro and in vivo model systems to evaluate mechanisms by which OSCs influence ovarian tumor progression. Normal sheep OSCs were found to inhibit the growth of SKOV3 and OVCAR3 human ovarian cancer cells, but not normal sheep OSE and human OSE cells (hOSE137 cells). In contrast, IOSChR and IOSCRH cells stimulated the growth of normal sheep and human OSE cells, as well as cancer cells. These findings were confirmed by in vivo studies. Our data provide compelling support for the importance of stromal-epithelial cell interactions during tumor progression, and show for the first time that immortalized and transformed OSCs promote growth of ovarian epithelial tumors. © 2007 Wiley-Liss, Inc.

Epithelial ovarian cancer (OC) is the leading cause of death due to gynecologic cancers in women in the United States.1 Approximately 22,420 new cases and 15,280 deaths will occur in 2007, and the overall 5-year survival rate is 45%. If diagnosed while still localized, the 5-year survival rate increases to 94%.2 Unfortunately, many ovarian tumors remain asymptomatic until an advanced stage, and only 19% of OCs are diagnosed while still confined to the ovary.1

OC can arise from any cell type of the ovary, including germ cells, granulosa or stromal cells; however, ∼90% of all OCs are epithelial and derived from a single cell layer of relatively pluripotent cells covering the surface of the ovary.3 Although the tumors derive from epithelial cells, their stromal cell component actively interacts with the tumor epithelium to support tumor progression.4 Stromal cell-epithelial cell interactions are important for tumor development, growth, angiogenesis and metastasis of many cancers including OC.5, 6 Unfortunately, few in vivo models for studying the role of ovarian stroma in the development of OC have been established. Although models that employ carcinogens have been used to induce ovarian tumorigenesis, animals rarely develop ovarian tumors spontaneously.7 Transgenic mouse models expressing promoters that drive expression of oncogenes in the ovarian surface epithelium have been developed recently to model OC.8

Transfection of oncogenic viruses that cause human cancers into normal cells produces cell lines capable of inducing cancer in animal models. These oncogenes include the large T antigen of simian virus 40, the adenovirus E1A and E1B proteins and the E6 and E7 proteins of certain human papillomaviruses.9 All of these viral proteins can bind to and inactivate the pRB and p53 tumor suppressor proteins.10, 11, 12 The coexpression of SV40 large T antigen and an oncogenic allele of the H-Ras gene in rodent fibroblasts results in transformation of the fibroblasts to cells capable of forming tumors. Conversely, similar attempts to transform normal human cells do not generally result in a transformed, tumorigenic cell phenotype13 The gene encoding the catalytic subunit of human telomerase (hTERT)14 combined with the SV40 early region9 yielded immortalized human fibroblasts, kidney epithelial cells, and human mammary epithelial cells.15 Subsequently, these immortalized cells acquired a tumorigenic phenotype upon introduction of the H-Ras oncogene.16

Multiple reports have shown that stromal cells are capable of regulating epithelial cell growth and tumor formation in tissues such as uterus, breast and prostate.17, 18, 19 Similarly, the intraovarian environment may regulate the biology of ovarian surface epithelia (OSE); however, little is currently known regarding the role of stroma on OSE cell growth.20 Primary ovarian stromal cells (OSCs) have been shown to increase OSE cell proliferation and growth response to estrogens, and additional functions of OSCs in OSE tumorigenesis have been suggested.21 Normal OSCs were shown to influence tumor growth and progression in nude mice. Other studies, however, suggest that normal ovarian stroma may promote the growth of normal OSE, but not transformed OC cells.22 OSCs produce a mixture of factors that can both promote and inhibit cell growth. Furthermore, normal ovarian and ovarian tumor cells respond differently to OSCs. The specific mechanisms by which OSCs influence ovarian tumor progression remain to be elucidated, but once discovered, they will provide insight into ovarian tumor biology.

In this study, primary OSE and OSC cell lines derived from a biopsy of a normal sheep ovary were immortalized with SV40 large T antigen and hTERT, and subsequently transformed with human H-Ras oncoprotein. The normal, immortalized and transformed OSE or OSC cell lines were used to investigate ovarian stromal-epithelial cell interactions and ovarian tumor progression in vitro and in vivo.

Material and methods

Sheep ovarian cell isolation, culture and characterization

Laparoscopy was used to monitor the process of isolating OSE and OSC cells from a sheep. OSE cells from sheep were obtained by scraping the surface epithelium with a biopsy scraper; OSCs were obtained by cutting a piece of the same ovary. The cells were cultured in NOE with 5% FBS under a 5% CO₂-95% air atmosphere at 37°C. Individual clones were isolated from cell cultures for further purification of a homogeneous cell population. These cells were subjected to immunohistochemistry staining for cytokeratin and vimentin to verify epithelial or stromal origin. SKOV3 and OVCAR3 human OC cells were cultured in the media described previously.23 Human normal OSE cells (hOSE 137 and T29) have been described previously.24, 25 Because these normal surface epithelial cells cannot survive for greater than several passages in the absence of immortalization, normal human OSE cells were immortalized with telomerase (T29 cells) or were subjected to p53 knockdown by RNA interference (hOSE137), in order to maintain the cells in culture.

Generation of cell lines and retrovirus infection

Sheep ovarian surface stromal cells were infected first with retrovirus containing SV40 large T antigen cDNA to make SV40 immortalized stromal cells (IOSC). Then IOSC was infected sequentially, first with retrovirus containing a full-length hTERT cDNA (to generate the IOSCH cell lines) and next with pBabe-puro-HRAS^V12 (to create the ISOChR cell lines). IOSC cells were transfected H-Ras before immortalization with telomerase to create IOSCRH cells. Briefly, amphotropic retroviral packaging Phoenix cells were transfected with pBabe-zeocin-SV40 T/t, pBabe-hygro-hTERT and pBabe-puro-HRAS^V12 by calcium phosphate precipitation, then the supernatants were used to infect the OSCs. Infected cells were selected in zeocin (500 μg/ml for 10 days), hygromycin (100 μg/ml for 10 days) or puromycin (0.5 μg/ml for 10 days). The detailed protocol for retrovirus production has been described.26 The antibody used to detect HRAS was rabbit polyclonal C20 (SC-520) (Santa Cruz Biotechnology), 1:500 dilution. The antibody used to detect SV40 large T antigen was from BD Bioscience, 1:200 dilution. Western blot was carried out as previously described.

Telomerase activity assay

Telomerase activity was measured with the TeloTAGGG Telomerase PCR ELISA kit (Roche Applied Science), using the protocol described by the manufacturer. Briefly, stromal cells were lysed in lysis buffer for 30 min on ice. Then lysates were centrifuged and supernatants were collected. Cell lysates were heated at 85°C for 10 min to destroy telomerase activity. Then 3 μl of heat inactivated or nonheated cell lysates were added into reaction mixture. Reactions were performed using the following conditions: primer elongation at 25°C for 30 min and inactivation of telomerase activity at 94°C for 5 min; followed by PCR amplification: denaturation at 94°C for 30 sec, annealing at 50°C for 30 sec, polymerization at 72°C for 30 sec, 30 cycles and a final extension at 72°C for 10 min. An aliquot of the PCR product was denatured and hybridized to a digoxigenin-(DIG)-labeled, telomeric repeat-specific detection probe. The resulting product was immobilized via the biotin labeled primer to a streptavidin-coated microplate. The immobilized PCR product was then detected with an antibody against DIG (anti-DIG-POD) that is conjugated to peroxidase. Finally, the probe was visualized by peroxidase metabolism of tetramethylbenzidine to form a colored reaction product.

Soft agar assay

Approximately 1.5 × 10⁴ cells were suspended in 1 ml of ovarian stromal cell culture medium with 0.35% agarose (Life Technologies, Rockville, MD), and the suspension was placed on top of 2 ml of solidified 0.7% agarose in culture medium. Triplicate cultures for each cell type were maintained for 14 days at 37°C in an atmosphere of 5% CO₂ and 95% air, and fresh medium was added after 1 week. Colonies >50 μm in diameter were counted after 2 weeks. These experiments were repeated twice.

Growth assays with stromal conditioned medium

Stromal conditioned medium was collected as previously described. Briefly, cell lines IOSC, IOSCH and IOSCRH were plated in T75 flasks. When the cells reached confluence, they were washed with serum-free medium and incubated in 10 ml of serum-free medium for 24 hr at 37°C. Medium was collected and centrifuged, and cell debris was discarded. The supernatant was filtered through a 0.2 μM membrane filter and stored at −80°C until use. Cell growth was analyzed by quantifying [³H] thymidine incorporation into newly synthesized DNA or by methylthiazole tetrazolium (MTT) assay for cell viability using Promega's CellTiter 96 AQueous One Solution Cell Proliferation Assay, according to the manufacturer's directions.

For the [³H] thymidine incorporation assays, OC or normal epithelial cells were plated at 2 × 10⁴/ml in 0.1 ml DMEM medium containing 0.1% calf serum in 96-well plates. After 24 hr, cells were treated for 48 hr with no growth factor (control), or 40 ng/ml epidermal growth factor (EGF), or 100% stromal conditioned medium (SCM) obtained from various OSC cell lines. This treatment duration corresponds to the approximate length of a cell cycle. After treatment, 0.1 ml of DMEM containing 0.5 μCi [³H] thymidine was added to each well, and the cells were incubated for 4 hr at 37°C. Plates were washed once with ice-cold PBS, fixed with 100% methanol for 10 min on ice and washed once with 1 ml of distilled water. One milliliter of ice-cold 5% trichloroacetic acid was added, and the plates were placed on ice for 10 min and washed once with distilled water. This treatment was then repeated. Following solubilization with 0.3 M NaOH at 37°C for 30 min, the cell lysates (400 μl in NaOH solution) were added to scintillation solution and counted in a β-scintillation counter.22 The amount of [³H] thymidine incorporated into cells in each well was calculated. Quadruplicates were tested for each sample.

Nude mouse injections

Female four- to six-weeks-old athymic nude mice (^nu/nu) on a Balb/c background were obtained from NCI, and used within 2 weeks of arrival. The care and manipulation of mice were in accordance with the approved guidelines of University of Texas M. D. Anderson Cancer Center for the ethical treatment of animals. Mice were given injections of SKOV3 cells alone or in combination with ovarian stromal cell lines (OSC, IOSCHR). Cells were injected subcutaneously on one side of the back of the mouse. Experiments were designed so that cancer cells (i.e. SKOV3, 5 × 10⁶) alone or the same number of cancer cells plus ovarian stromal cell combinations were injected together (n = 5 mice per group). To prepare cells for injection, cultures were trypsinized for 5 min at 37°C. After gentle removal using a glass pipette, cells were centrifuged, resuspended in PBS and counted on a hemocytometer. Cell suspensions were diluted in Hank's balanced salt solution and placed into a syringe for injection. All injected mice contained an observable mass at the site of injection that steadily decreased in size until ∼Day 7 post-injection. After Day 7, an observable tumor mass developed and grew in size. Mice were monitored daily and tumor area was determined every other day for a total of 40 days. Tumor area was determined by multiplying the length of the tumor by its width.

Histopathology and immunohistochemistry

Formalin-fixed, paraffin-embedded tumor tissues were cut into 4 μm-thick sections, which were attached to slides, deparaffinized, rehydrated and stained with hematoxylin and eosin (H&E) for histopathological analysis. Mitotic rates were determined by a pathologist, who counted 5 fields in different areas throughout the 40× high power field. The mitotic rate was determined by the average of the counts from the 5 high power fields.

For pan-cytokeratin and vimentin immunostaining of paraffin-embedded sections, slides were treated with xylene 2 × 10 min and 100% ethanol 2 × 2 min. Endogenous peroxidase activity was blocked by incubating for 10 min in 3% H₂O₂. After rinsing 2–3 times in water, slides were placed in a plastic coplin jar filled with Retrievagen A solution (BD Bioscience) and heated in water bath at 89°C for 10 min. Then slides were slowly cooled down to room temperature for 20 min. Slides were washed with 2–3 changes of water and 10% goat normal serum was added to block nonspecific biding at 37°C for 30 min. Slides were incubated with mouse antipan-cytokeratin IgG (Sigma, 1:600 dilution) or mouse antivimentin IgM antibody (Sigma, 1:200 dilution) at 37°C for 1 hr. After washing the slides with PBS 3 times, a 3,3′-diaminobenzidine kit was used to detect secondary antibody binding. Control sections were incubated in the absence of primary antibody. Results were considered valid only if these controls were negative.

Immunohistochemistry for Ki67 (clone MIB-1, DAKO M7240, dilution 1:75) was carried out with antigen retrieval by pressure cooking in 0.01 M citrate buffer at pH 6.0, and detection was via DAKOCytomation enVision/HRP kit K4004. Only distinct nuclear staining of invasive carcinoma cells was used for scoring via the light microscope.12

Fluorescence in situ hybridization

Tumors that formed in nude mice were excised and placed in formalin. Fixed tumors were embedded in paraffin and cut into 7 and 4 μM sections. Tumor sections of 4 μM were stained with eosin and hemotoxylin as previously described. Tumor sections of 7 μM were assayed for the presence of mouse, sheep and human cells using fluorescence in situ hybridization (FISH) with species-specific genomic probes. Paraffin-embedded sections fixed on slides were treated by the following protocol: xylene 2 × 10 min; 100% ethanol 2 × 10 min, then air dry; 0.1 M NaSCN at 80°C for 1 min; H₂O 2 × 5 min; 1 mg/ml RNAase at 37°C for 20 min; H₂O 2 × 5 min; 200 μl of 20 mg/ml proteinase K in 20 ml PBS (plus 100 μl 10% SDS) at 37°C for 30 min; H₂O 2 × 5 min; apply 10 μl of 1 μM 4,6-diamino-2-phenylindole (DAPI) in distilled water to counterstain and check for adequate digestion of tissue sections. Slides were then rinsed in H₂O 2 × 5 min, dehydrated in a 70, 85 and 100% ethanol series and air dried. Probes used for FISH were mouse genomic DNA extracted from mouse tail, human genomic DNA purchased from Promega and sheep genomic DNA extracted from sheep blood. Each DNA sample was labeled by nick translation as previously described (Current Protocols in Human Genetics, Unit 4.5, 2005) with Alexa 594-5-dUTP (Molecular Probes). Probe sizes ranged between 300 and 2,000 bp when electrophoresed on a nondenaturing 1% agarose gel. Each labeled probe (10 ng/ml in 20 μl of hybridization buffer; final concentration of 50% formamide, 10% dextran sulfate and 2×SSC [sodium chloride-sodium citrate buffer] at pH 7) was denatured at 72°C for 10 min. The slides were also denatured in 70% formamide in 2×SSC at 72°C for 5min. Then denatured probes in hybridization buffer were applied to the denatured slides, covered with 22 mm² cover slips and incubated at 37°C overnight. Following hybridization, the slides were soaked in 2×SSC for 5 min to facilitate removal of the cover slips. Slides were washed 3 times in washing buffer (50% formamide, 2×SSC, pH 7) at 45°C, then 1 time in 2×SSC at 45°C, once in 0.1 M phosphate buffer with 0.05% NP-40 at pH 8 and a final rinse in distilled water before air drying. All washes were 10 min each. The tissue sections were counterstained with 1 μM DAPI in Mounting Medium (VECTASHIELD® HardFSet™ Mounting Medium; Vector Laboratories) for preserving fluorescence, which hardens after cover-slipping to facilitate handling and storage. The slides were then stored at −20°C until they were used. Slides were read on an Olympus IX51 fluorescence microscope.

Statistics

Data are expressed as mean ± SE. For significance analyses, data were analyzed by analysis of variance (ANOVA) or t-test, and p < 0.05 was considered significant.

Results

SV40 and H-Ras protein expression and telomerase activity in OSCs

SV40 large T antigen expression (97 kDa; Fig. 1a) and H-Ras oncoprotein expression (21 kDa; Fig. 1b) were observed in immortalized, H-Ras-transformed IOSCH (designated IOSCRH) cells, but not in the normal parent OSC cells. Detectable telomerase activity is defined as the threshold at which the absorbance of target cells compared to heat-inactivated controls is higher than 0.2 A₄₅₀ nm – A₆₉₀ nm units by ELISA assay using a TeloTAGGG Telomerase kit. Telomerase activity was higher in IOSCRH cells (light bars) than in parental stromal cells (Fig. 1c). There was no difference in telomerase activity between IOSCRH (IOSC cells first transfected with H-Ras, then with telomerase) and IOSChR (IOSC cells first immortalized with telomerase, and then transfected with H-Ras). Heat inactivated extracts from each group were used as negative controls (dark bars). After OSCs were immortalized and transfected, cell viability was determined by MTT assay. Enhanced cell proliferation was observed in IOSC cells (Fig. 1d). There was no significant difference among OSC and IOSC, OSC and IOSCh or OSC and IOSCR (paired t-test). This figure also shows that normal sheep OSC proliferation decreased between 4 and 7 days of culture, whereas IOSC, IOSCh and IOSCR cells continued to proliferate during this time period.

Figure 1. Determination of the expression of SV40 T/t antigen, the expression of H-Ras oncoprotein and activity of hTERT in IOSCRH cells. (a) Expression of SV40 large T antigen protein (97 kDa) was detected in ISOCRH, but not in their parent OSC. (b) Expression of H-Ras oncoprotein (21 kDa) was detected in ISOCRH cells. No expression was detected in their parent OSC. (c) Telomerase activity was detected in IOSCRH but not in OSC cells. *, p < 0.05 compared to OSC. Heat inactivated extracts from each group were used as negative controls (dark bars). (d) An MTT assay was used to quantitate cell viability in various OSC cell lines (OSC, IOSC, IOSCh and IOSCR) on Day 0, Day 1, Day 4 and Day 7. The plate was read at OD₄₉₂ and the mean and standard deviation were calculated. There was no significant difference in MTT activity among the groups. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Generation of transformed stromal cells in vitro and induction of tumorigenesis in vivo

A soft agar colony formation assay was performed to identify anchorage-independent growth of IOSCRH and their parent cells. Colonies greater than 50 μm in diameter were counted after 2weeks. Approximately 1 × 10⁴ IOSCRH cells produced 65 ± 8 colonies while their normal parent cells failed to produce colonies. Nude mice inoculated with IOSCRH cells alone showed no tumor formation (data not shown), indicating that the IOSCRH cells were transformed but not tumorigenic.

Effect of stromal conditioned media on the growth of normal and neoplastic OSE cells in vitro

The effects of SCM collected from normal OSC or IOSCRH cells on the growth of sheep OSE or human OC cells were evaluated by quantifying [³H] thymidine incorporation into newly synthesized DNA as described previously.22 EGF was used as a positive control to stimulate all cell types. The results were performed in triplicate wells for each experimental condition, and the experiment was repeated at least 3 times. Conditioned media obtained from normal OSCs inhibited OC cell growth (SKOV3; Fig. 2a or OVCAR3; Fig. 2b), but not normal human OSE cells (hOSE137; Fig. 2c) or sheep OSE cells (Fig. 2d). In contrast, conditioned media obtained from transformed stromal cells (IOSCRH) stimulated normal human (Fig. 2c) or sheep OSE (Fig. 2d) cell growth. The effects of SCM collected from various OSC cells (OSC, IOSC, IOSCh and IOSChR) on the growth of human normal ovarian cells or cancer cells are summarized in Figure 2e. OSC media only inhibited SKOV3 cell proliferation (p = 0.021; t-test). Media obtained from various immortalized or/and transformed OSC cells stimulated both cancer and normal cell proliferation. MTT cell viability assays were used to verify the results of these thymidine incorporation assays. Cell proliferation was significantly decreased in both cancer cell lines, SKOV3 and OVCAR3 (Figs. 3a and 3b), but not in the normal human OSE cell lines hOSE 137 and T29 (Figs. 3c and 3d) using conditioned media obtained from normal sheep OSCs. In agreement with the tritiated thymidine assays, cell proliferation was significantly increased in the p53-null normal ovarian cell line hOSE137 using conditioned media obtained from IOSCRH cells (Fig. 3c).

Figure 2. The effect of stromal conditioned medium collected from OSC, IOSCRH and IOSChR cells on the proliferation of SKOV3 cells (a), OVCAR3 cells (b), hOSE137 (human normal ovarian epithelial cells; c) and sOSE (sheep normal ovarian epithelial cells; d). Conditioned media obtained from normal OSCs inhibited SKOV3 [(a) OSC; **, p = 0.003] and OVCAR3 [(b) OSC; **, p = 0.0021] human OC cell proliferation, compared to cell culture media with no conditioned medium added. There was no inhibition by OSC conditioned medium of proliferation of hOSE137 normal human ovarian epithelial cells or sheep OSE cells (c, d). Conditioned media obtained from transformed ISOCRH stimulated normal human hOSE137 [(c) **, p = 0.0074] and sheep OSE cells [(d) ***, p < 0.001]. Epidermal growth factor (EGF) was used as a positive control. OSC conditioned media significantly decreased the proliferation of SKOV3 cells (p = 0.021; t-test), but not hOSE137 cells (e). IOSC conditioned media significantly stimulated the proliferation of SKOV3 cells (p = 0.004) and hOSE137 cells (p = 0.040). IOSCh conditioned media stimulated the proliferation of hOSE137 cells, but not SKOV3 cells. IOSCRH conditioned media significantly decreased the proliferation of both cancer (p = 0.011) and normal cells (p = 0.007).

Figure 3. The effects of stromal conditioned media on the growth of normal and neoplastic OSEs in vitro were further confirmed by an MTT cell viability assay. Conditioned media obtained from normal sheep OSCs decreased the viability of SKOV3 cells [(a) OSC; **, p = 0.0063] and OVCAR3 cells [(b) OSC; ***, p = 0.0007] compared to cell culture medium without added conditioned medium. There was no significant inhibition by normal sheep OSCs of viability of the normal human OSE cells lines hOSE137 (c) and T29 (d). Conditioned media obtained from transformed ISOCRH increased the viability of normal human hOSE137 cells [(c) ***, p < 0.0001].

Effect of normal OSCs and transformed OSCs on epithelial tumor growth in nude mice

Female athymic nude mice (nu/nu; 6-week-old) were inoculated subcutaneously in the right flank with SKOV3 cells (5 × 10⁶) alone, in combination with normal OSC (5 × 10⁶ cells) or in combination with IOSChR (5 × 10⁶ cells). Each treatment group contained 5 mice that were observed daily for tumor growth and the volume of each tumor was calculated twice weekly for a total of 30–40 days. This experiment was repeated 3 times. A representative experiment is depicted in Figure 4a. There was a significant difference between the SKOV3 plus OSC group and the SKOV3 group (p = 0.059), as well as between the SKOV3 plus IOSCRH group and the SKOV3 group (p = 0.0001). Forty percent of the mice in the group receiving SKOV3 cells plus normal OSC failed to develop a tumor. Smaller tumors were obtained from SKOV3 OC cells coinjected with normal sheep OSC (29.94 ± 27.80 mm² at Day 36 post-injection) as compared to tumors cultivated from SKOV3 only (112.33 ± 69.11 mm²) or tumors cocultivated with IOSCRH (143.88 ± 47.75 mm²), suggesting that the normal stromal cells inhibited tumor growth. Conversely, enhanced tumor growth was observed when SKOV3 cells were coinjected with IOSCRH as compared to SKOV3 cells cultivated without OSC or with normal OSC.

Figure 4. The effects of stromal conditioned media on the growth of normal and neoplastic OSEs in vitro was further evaluated in vivo (a). Smaller tumor sizes were observed when SKOV3 OC cells were coinjected with normal sheep OSC (SKOV3 + OSC) as compared to tumors injected with SKOV3 cells only (SKOV3) or tumors coinjected with IOSCRH (SKOV3 + ISOCRH), suggesting that the normal stromal cells inhibited tumor growth. Conversely, enhanced tumor growth was observed when SKOV3 cells were coinjected with IOSCRH as compared to SKOV3 cells injected without OSC or with normal OSC. There were significant treatment (p < 0.0014), time (p < 0.001) and treatment × time (p < 0.0003) effects. When stained with H&E (b–d), tumors derived from SKOV3 + OSCs (1), SKOV3 cells (4) and SKOV3 + IOSCRH cells (7) displayed similar morphologic features. Greater than 75% of the cells showed intense, diffuse cytoplasmic staining for vimentin in tumors derived from SKOV3 + OSCs (2), SKOV3 cells (5) and SKOV3 + IOSCRH cells (8). Approximately, 20–50% of the tumor cells displayed intense, multifocal cytoplasmic staining for pancytokeratin in tumors derived from SKOV3 + OSCs (3), SKOV3 cells (6) and SKOV3 + IOSCRH cells (9).

Gross, histologic and immunohistochemical characterization of resultant epithelial tumors

Tumors obtained from the above 3 groups were subjected to routine histological examination and immunohistochemistry (Figs. 4b and 4c). Histologically, all tumors that developed across the 3 groups displayed similar morphologic features with various tumor sizes. Smaller tumor sizes or no tumor were observed in the mice inoculated with SKOV3 cells plus OSC cells (Fig. 4b; 1–3) compared with the mice inoculated with SKOV3 cells only (Fig. 4c; 4–6). The largest tumor sizes were observed in mice inoculated with SKOV3 cells plus IOSCRH cells (Fig. 4d; 7–9) compared with the mice inoculated with SKOV3 cells only (Fig. 4c; 4–6). They consisted of large packets of markedly anaplastic cells separated by prominent fibrovascular stroma. Each tumor displayed a mixture of dense, compact cells that lacked cytoplasmic clearing and larger cells with abundant clear cytoplasm that frequently appeared vacuolated. All tumors demonstrated a high mitotic rate (8–12/high power field, 40×), pronounced nuclear atypia and frequent macrokaryosis. Necrosis was present in all tumors to a varying degree.

The results of immunohistochemical staining were also similar among the 3 tumor groups. Intense, diffuse cytoplasmic staining for vimentin was present in greater than 75% of the tumor cells (Fig. 4d; 8) in mice inoculated with SKOV3 plus IOSCRH cells. Intense, multifocal cytoplasmic staining for pancytokeratin was observed in ∼20–50% of the tumor cells (Fig. 4d; 9).

Identification of murine, ovine and human derived cells via fluorescence in situ hybridization

Tumors obtained from nude mice may be comprised of cells from 3 different species: human OC cells (SKOV3), sheep OSCs (OSC or IOSCRH) and murine stromal cells recruited by human OC cells. FISH was used to confirm the source of tumor constituent cells, using species-specific chromosomal probes to distinguish human, sheep and murine cells (Figs. 5a–5c). After each probe hybridized to normal human, sheep and mouse ovarian tissues, the specificity and lack of cross-species hybridization was confirmed (data not shown). FISH analysis was employed to evaluate tumor biopsies obtained at Day 7 (data not shown) or Day 40 post-injection (Figs. 5a–5c). A DAPI DNA stain identified the nuclei of viable cells. Increased fluorescence intensity of sheep probes was observed in mice inoculated with SKOV3 plus IOSCRH cells (Fig. 5a; 2), in comparison to inoculation with SKOV3 plus OSC (Fig. 5a; 1); SKOV3 alone showed no fluorescence with a sheep probe (data not shown). Decreased numbers of murine stromal cells could be detected in mice inoculated with SKOV3 plus IOSCRH cells (Fig. 5b; 4) than in mice inoculated with SKOV3 cells alone (Fig. 5b; 3) using a mouse-specific fluorescently-labeled probe. SKOV3 tumors had the ability to recruit mouse stromal cells to support their growth (Fig. 5b; 3–4) and this recruitment was decreased in SKOV3 cells coinnoculated with IOSCRH cells. Increased intratumoral stroma cells with a lower density of SKOV3 cells were detected in mice innoculated with SKOV3 plus IOSCRH cells (Fig. 5c; 6), compared to innoculation with SKOV3 cells alone (Fig. 5c; 5), using human-specific fluorescently-labeled probes. Furthermore, the relationship of Ki67 immunohistochemical detection with clinicopathologic parameters was evaluated in the 3 groups. Ki67 is a nuclear protein that is tightly linked to the cell cycle. Increased Ki67 expression was observed in mice injected with SKOV3 plus IOSCRH cells, compared with mice inoculated with SKOV3 cells only (Fig. 5d) or SKOV3 plus OSC cells.

Figure 5. Using a species-specific FISH assay, tumors derived from SKOV3 + OSC and SKOV3 + IOSCRH were stained with species-specific probes including sheep fluorescent DNA probes (a; 1, 2), mouse fluorescent DNA probes (b; 3, 4) and human fluorescent DNA probes (c; 5, 6). Increased fluorescence intensity of sheep probes was observed in mice innoculated with SKOV3 plus IOSCRH cells (a; 2), in comparison to inoculation with SKOV3 plus OSC (a; 1). Decreased numbers of murine stromal cells could be detected in mice innoculated with SKOV3 plus IOSCRH cells (b; 4) than in mice innoculated with SKOV3 cells alone (b; 3). Increased intratumoral stroma cells with a lower density of SKOV3 cells were detected in mice innoculated with SKOV3 plus IOSCRH cells (c; 6), compared to inoculation with SKOV3 cells alone (c; 5). Ki67 antigen expression was evaluated to determine the Ki67 proliferation index (d). There was increased Ki67 antigen expression in SKOV3 coinjected with IOSCRH cells (SKOV3 + IOSCRH), in comparison to SKOV3 coinjected with OSC cells (SKOV3) (p = 0.0345).

Discussion

In this study, we report the establishment of immortalized and transformed sheep epithelial and stromal cell lines containing SV40 large T antigen, telomerase and the H-Ras oncogene. Cell immortalization and transformation is a key step for tumorigenesis,14 and several human cell culture models of tumorigenesis, including OC,26 have been established. Although these models do not reproduce all aspects of human OC, our model did bring about a transformed phenotype of sheep ovarian cells. For example, although SV40 large T antigen is unlikely to be involved in the genesis of either human or sheep OC, its introduction into normal sheep OSCs did nonetheless increase cell viability as measured by the MTT assay (Fig. 1d). This result is similar to results observed in an earlier study using human OSE cells transfected with SV40 large T antigen.27 In this study, SV40 large T antigen transfection resulted in higher proliferation rates and saturation density of the OSE.

Sheep ovarian cells have several advantages over other models of OC. Normal sheep OSC and OSE can grow for longer periods than human primary ovarian cells in culture without immortalized or transfection, which affords greater opportunities to follow stromal-epithelial cell interactions that are important in ovarian tumor progression. Human primary cells are generally difficult to culture for extended periods without immortalization or transfection. Another advantage of our sheep ovarian cell model is that all reagents and antibodies used for human ovarian cell studies can be applied to studies using sheep cells. These cells can later be introduced into sheep, which are similar in size to humans and can therefore be studied with human instrumentation that is currently being used for whole-body imaging and surgery. Finally, the species-specific fluorescent probes we developed can be used to investigate additional aspects of tumor-stromal interactions in this ovarian cell model in the future.

Our results provide evidence of the role of transformed stromal cells in stimulating epithelial cell growth. However, unlike observations derived from a bovine model of tumorigenesis in which normal stromal cells were not detected 1 week post-injection,22 we did detect the presence of normal sheep stromal cells at 7 days and 30 days after coinjection with SKOV3 cells. Normal OSC cells from sheep are capable of inhibiting SKOV3 tumor growth; whereas, transformed stromal cells stimulate normal epithelial cell growth in vitro and epithelial ovarian tumor cell growth in vivo. Sheep stromal cells transfected with SV40 large T antigen, telomerase and H-Ras became immortalized and transformed, as indicated by their ability to grow in culture to high passage numbers in our laboratory, and by colony formation in a soft agar assay. However, the 2 separate clones obtained from IOSChR or IOSCRH cells showed significantly lower colony formation ability than SKOV3 OC cells, and no tumors formed in mice injected with IOSChR or IOSCRH cells alone (4 individual experiments have been done in our laboratory with a total of 40 nude mice). Another finding in our study was that IOSCRH conditioned media significantly stimulated hOSE137 cells, which contain a silenced p53 gene (Figs. 2c and 3c), as well as hOSE151 cells (data not shown), but not T29 (Fig. 3d) and T72 normal ovarian cells immortalized with telomerase (data not shown). These findings indicate that oncogenic H-Ras signaling is coupled to tumor suppressor pathways, whose activation can trigger a senescence-like permanent growth arrest (cellular senescence) in normal cells. Understanding the mechanisms by which oncogenic Ras signaling induces senescence may help to develop novel therapeutic interventions.

Interestingly, greater expression of the Ki-67 marker was detected in tumors from mice injected with IOSCRH plus SKOV3 cells than in mice injected with SKOV3 cells alone (Fig. 5d) or SKOV3 plus OSC cells, which indicated that transformed OSC cells promote the proliferative capacity of tumors. In contrast to the situation with stromal cells, sheep epithelial (OSE) cells transfected with SV40 large T antigen, telomerase and H-Ras not only became immortalized and transformed, but also were tumorigenic in the same nude mouse model used with transformed stromal cells. Mice injected with IOSChR cells (10 mice per experiment, repeated twice) developed ascites (5–8 ml per mouse) and multiple tumors on Day 35, especially when cells were injected intraperitoneally rather than subcutaneously. Collectively these results suggest that 2 independent clones of transfected OSCs underwent transformation and promoted tumor proliferation, but were not tumorigenic in nude mice in the absence of epithelial tumor cells. Similar transfections carried out in ovarian epithelial cells, on the other hand, produced tumorigenic ovarian cells.

To date there is little evidence that the stromal cell component of OCs is transformed. However, there is strong evidence that stromal cells in the tumor bed of epithelial cancers are activated.28 A recent study of normal ovarian stromal fibroblasts found that secretion of the RAS-induced chemokine growth-regulated oncogene 1 (Gro-1) by preneoplastic OSE cells induced senescence in normal fibroblasts, and that senescent fibroblasts induced by Gro-1 can promote tumor growth.29Gro-1 links RAS signaling to the senescence of stromal fibroblasts and ovarian tumorigenesis. These studies indicate that the presence of the epithelial tumor influences the biology of the stromal cell.

Models for human cancer are valuable to study molecular pathways of neoplastic transformation and tumorigenesis in vivo, and they serve as tools for selected preclinical trials. OC is not genetically and phenotypically uniform, and therefore one model system will never be enough to recapitulate various forms of the disease. It is unrealistic to expect that one model can mimic all aspects of human OCs. Here we provide a model other than mice that may be useful for OC studies. Additional genes in future studies can also be introduced into this animal model to evaluate their effect on ovarian metabolism or OC.

In conclusion, our studies have established that normal stromal cells stimulate normal ovarian epithelial cell growth, but inhibit ovarian tumor development. Only low levels of OSC are detected in tumor masses. In contrast, transformed stromal cells support both normal and ovarian tumor epithelial cell growth, and lead to increased levels of stroma in tumor masses. Our data support the concept that stromal cells are required for tumor progression, and the model we established here can be used to gain an understanding of the role of cellular and extracellular components that surround and interact with OC cells. Our system can also be used to explain how stromal-epithelial interactions control and promote tumor growth. The availability of normal or transformed sheep stromal or epithelial cell lines will provide tools to advance invivo studies of the association between stroma and epithelia in the ovary and to study tumorigenesis in OC.