Induction of FGF receptor 2-IIIb expression and response to its ligands in epithelial ovarian cancer

Abstract

Epithelial ovarian cancers (EOCs) arise in the Ovarian Surface Epithelium (OSE). This tissue is a simple, poorly committed mesothelium which exhibits characteristics of epithelial and mesenchymal cells when grown in culture. In contrast, EOCs frequently exhibit properties of complex epithelial tissues of the female reproductive tract, such as oviductal, endometrial and cervical epithelia, and show induction of expression of epithelial markers such as E-cadherin. Fibroblast Growth Factor Receptor 2 isoform IIIb (FGF receptor 2-IIIb) is a spliced variant of FGF receptor 2 that binds the ligands FGF-1 and FGF-7 with high affinity, and is expressed exclusively by epithelial cells. We have studied the expression of FGF receptor 2-IIIb and its ligands in primary cultures of normal human OSE, EOC cell lines and snap frozen tissue from EOCs. Expression of FGF receptor 2-IIIb mRNA is undetectable in normal OSE, but is expressed in 16/20 (80%) of EOCs. FGFs 1 and 7 mRNAs are expressed in normal OSE, whilst only 4/20 (20%) and 12/20 (60%) of EOCs demonstrated expression for these ligands respectively. However, FGF-7 protein was detected in 70% (mean level = 0.7 ng/ml) of ascitic fluids obtained from patients with EOC. FGFs 1 and 7 stimulate DNA synthesis in EOC cell lines that express FGF receptor 2-IIIb. Moreover, DNA synthesis in these cell lines can be partially blocked by blocking antisera to FGFs 1 and 7. It is suggested that induction of expression of FGF receptor 2-IIIb may play a role in the development of EOCs by rendering the OSE susceptible to paracrine and/or autocrine stimulation by its requisite FGF ligands.

Introduction

Ovarian cancer is the leading cause of death from gynaecological malignancy. Despite its high incidence, the onset of the disease is usually asymptomatic, with most patients presenting with vague abdominal discomfort (Hernandez and Rosenheim, 1989). The majority of women diagnosed with this cancer exhibit an advanced, widely disseminated malignancy with a poor survival rate. Thus in most cases, the initial diagnosis is established at an advanced stage, when current therapy can only benefit a fraction of those affected (Westermann et al., 1997).

Over 90% of ovarian cancers are derived from the ovarian surface epithelium (OSE) (Scully, 1977), which is a modified form of the peritoneal mesothelium. This tissue is relatively uncommitted in comparison with other epithelial tissues. For example, it does not express markers of complex epithelial differentiation such as E-cadherin, and when explanted into culture it rapidly exhibits mesenchymal characteristics such as expression of vimentin and type III collagen, and invasion into matrigel (Kruk et al., 1994). Although some EOCs are thought to arise de novo in the surface epithelium, many are thought to derive from inclusion cysts formed by OSE that has become entrapped in the ovarian stroma as a result of ovulation (Scully, 1992, 1995). These inclusion cysts are frequently atypical with some displaying metaplasic alterations (Resta et al., 1993). The OSE secretes and responds to a variety of growth factors and cytokines, suggesting an integral role for these factors in maintaining normal ovarian physiology (Berchuck et al., 1993).

In contrast to many other solid tumours, epithelial ovarian carcinomas tend to become more highly differentiated than their tissue of origin, histologically resembling complex epithelia of the Mullerian duct such as epithelia of the fallopian tube, endometrium and cervix (Young et al., 1989). This increased commitment to a complex epithelial phenotype is reflected in molecular terms by induction of E-cadherin (Inoue et al., 1992; Veatch et al., 1994; Maines-Bandiera and Auersperg, 1997; Sundfeldt et al., 1997; Davies et al., 1998) and CA125 (Klug et al., 1984; Bast et al., 1995) expression in the majority of EOCs. Moreover, primary cultures of OSE from women with family histories of EOC also show phenotypes indicative of an increased commitment to an epithelial phenotype, including retention of cytokeratin expression, lower expression of type III collagen, lack of secretion of extracellular matrix and induction of CA125 expression (Dyck et al., 1996). Collectively, these results suggest that an increased commitment to an epithelial phenotype may be an early and functionally significant event in the neoplastic transformation of OSE cells.

The Fibroblast Growth Factors (FGFs) are a family of structurally related polypeptides, composed of at least 23 members, commonly identified as FGF-1 to 23 (Yamashita et al., 2000). They share between 35 and 50% amino acid sequence homology and act as mediators of a diverse range of developmental and physiological processes, both in vitro and in vivo. FGFs display mitogenic, chemotactic and angiogenic activity, inducing proliferation and differentiation in a wide range of tissues of epithelial, mesothelial, mesodermal and neuroectodermal origin (Burgess and Maciag, 1989; Basilico and Moscatelli, 1992; Friesel and Maciag, 1995).

Presently, at least four FGF receptors (FGF receptor 1–4) have been discovered and together they regulate developmental processes in a number of tissues (Peters et al., 1992; Orr-Urtreger et al., 1993; Mason, 1994). The FGF receptors are tyrosine kinases which possess three extracellular immunoglobulin-like domains, a transmembrane region and a cytoplasmic split tyrosine kinase domain which is activated upon FGF binding. These receptors undergo alternative splicing generating a striking number of isoforms with ligand specificity being determined by deletions within the C-terminal sequences in the third immunoglobulin-like loop (Givol and Yayon, 1992). The existence of multiple ligands and receptors allows interaction between a single FGF receptor and a variety of growth factors. In the ovary, little is known about the role of the FGFs in maintenance of the normal OSE and the progression to carcinogenesis. Studies have indicated that ovarian tumours are able to produce FGF-2 (bFGF), a growth factor which can bind to the majority of FGF receptors, some of which are expressed in ovarian cancer (Di Blasio et al., 1993; Crickard et al., 1994).

Spliced variants of the FGF receptor-2 gene may be classified by their ability to bind the ligand FGF-7. FGF receptor 2-IIIb, for example, has the ability to bind both FGF-7 and FGF-1 with a high affinity due to its possession of the IIIb exon, whereas FGF receptor 2-IIIc binds FGF-1 and FGF-2 with a high affinity and expresses the IIIc exon (Miki et al., 1992). FGF-7 differs from other characterized FGF-related molecules, which are active on a broad range of cells types, in that its mitogenic activity is restricted to epithelial cells (Finch et al., 1989; Rubin et al., 1989). This is because FGF-7 can only act through the FGF receptor 2-IIIb isoform, which is expressed exclusively by epithelial cells. Given that FGF receptor 2-IIIb, like E-cadherin, is a marker of epithelial cells, we hypothesized that expression of FGF receptor 2-IIIb may be induced in EOCs. Thus, we have examined the expression of FGF receptor 2-IIIb and the expression and effects of its ligands in primary cultures of normal OSE, EOC cell lines and frozen tissue obtained from patients undergoing surgery for EOC. We now report that FGF receptor 2-IIIb expression is induced in the majority of EOCs and that FGF-7 is usually present in ascitic fluid. Moreover, we demonstrate that EOC cell lines expressing FGF receptor 2-IIIb can respond to FGFs 1 and 7.

Results

Identification of primary cultures

Primary cultures were successfully isolated from nine ovaries and their purity determined by staining with cytokeratins 8, 18 and 19 (Figure 1). Cultures consisting of >98% OSE cells were used for analysis.

Figure 1
figure1

Morphology and cytokeratin expression of normal OSE. (a) Photomicrograph of normal OSE grown in monolayer culture demonstrating typical cobblestone appearance of epithelial phenotype. (b) Photomicrograph of sub-confluent OSE cells at high power after methanol fixation, stained with cytokeratin-18 anti-serum to confirm epithelial phenotype of the cultures

Expression of FGF receptor 2-IIIb

Following cDNA synthesis and 40 cycles of PCR amplification, expression of a cDNA product corresponding to the mRNA encoding FGF receptor 2-IIIb was undetectable in an SV40-immortalized cell line derived from normal OSE and nine primary cultures of OSE (Figure 2a and Table 1). In contrast, FGF receptor 2-IIIb mRNA was detected in 16/20 (80%) of the snap frozen tumour samples (Figure 2a and Table 2) and 3/3 (100%) ovarian carcinoma cell lines (Figure 3a). A pair of GAPDH oligonucleotides was used as constitutive primers to ensure that GAPDH cDNA could be amplified from the cDNA samples (Figures 2c and 3d). The identity of the PCR products was confirmed by two methods. Firstly, selected bands were extracted from the gel and directly sequenced. Secondly, the remaining samples were Southern blotted and hybridized with 32P-labelled gene-specific internal oligonucleotides. The FGF receptor 2-IIIb specific oligonucleotide did not hybridize to the normal OSE primary cultures or cell lines, whereas 16/20 of the snap frozen ovarian tumour samples showed clear hybridization, in agreement with the ethidium bromide-stained gels (Figure 2b).

Figure 2
figure2

Expression of FGF receptor 2-IIIb in normal OSE and ovarian carcinomas. Total RNA was reverse transcribed and the cDNA was amplified by PCR using FGF receptor 2-IIIb-specific primers and stained with ethidium bromide. DNA was then blotted onto a nylon filter and hybridized with a 32P-labelled FGF receptor 2-IIIb-specific oligonucleotide. (a) Ethidium bromide stained agarose gel of products from normal OSE (lanes 1–4 and 15–18) and ovarian carcinomas (lanes 5–14 and 19–28) with FGF receptor 2-IIIb specific primers. The product of lane 14 was directly sequenced to confirm its identity. (b) Southern blot hybridization of lanes 1–13 and 15–28 of panel (a) with 32P-labelled FGF receptor 2-IIIb-specific oligonucleotide. (c) Ethidium bromide stained gel of the same samples amplified with GAPDH-specific primers

Table 1 Expression of FGF receptor 2-IIIb, FGF-1 and FGF-7 in normal OSE
Table 2 Expression of FGF receptor 2-IIIb, FGF-1 and FGF-7 in ovarian cancers
Figure 3
figure3

Expression of FGF receptor 2-IIIb, FGF-1 and FGF-7 in ovarian carcinoma cell lines. Total RNA was reverse transcribed and the cDNA was amplified by PCR using gene-specific primers and stained with ethidium bromide. (a) Ethidium bromide stained agarose gel of the ovarian carcinoma cell lines MDAH-2774, SK-OV-3 and 41 M (lanes 1–3 respectively) with FGF receptor 2-IIIb-specific primers. Water and genomic DNA were included as controls (lanes 4 and 5 respectively). (b) Ethidium bromide stained agarose gel of the ovarian carcinoma cell lines MDAH-2774, SK-OV-3 and 41 M (lanes 1–3 respectively) with FGF-7-specific primers. Water and genomic DNA were included as controls (lanes 4 and 5 respectively). (c) Ethidium bromide stained agarose gel of the ovarian carcinoma cell lines MDAH-2774, SK-OV-3 and 41 M (lanes 1–3 respectively) with FGF-1-specific primers. Water and genomic DNA were included as controls (lanes 4 and 5 respectively). (d) Ethidium bromide stained agarose gel of the ovarian carcinoma cell lines MDAH-2774, SK-OV-3 and 41 M (lanes 1–3 respectively) with GAPDH-specific primers. Water and genomic DNA were included as controls (lanes 4 and 5 respectively)

Expression of FGF receptor 2-IIIb in the ovarian cancer cell lines, SK-OV-3, 41/M and MDAH-2774, was confirmed by Northern blotting analysis (Figure 4).

Figure 4
figure4

Northern blotting analysis of FGF receptor 2-IIIb expression in ovarian carcinoma cell lines. Poly (A)-RNA was blotted onto a nylon filter and hybridized with a 32P-labelled FGF receptor 2-IIIb-specific or GAPDH-specific cDNA transcript. (a) Northern blot hybridization of the ovarian carcinoma cell lines SK-OV-3, 41/M and MDAH-2774 (lanes 1–3 respectively) with an FGF receptor 2-IIIb specific probe. (b) Northern blot of the same samples with GAPDH-specific probe

Expression of ligands for FGF receptor 2-IIIb

Following 40 cycles of amplification by PCR, a cDNA product corresponding to the mRNA encoding FGF-7 was detected in 1/3 of the ovarian carcinoma cell lines (Figure 3b), 9/9 primary cultures of OSE (Figure 5a and Table 1), an SV40-immortalized normal OSE cell line and 12/20 (60%) of the snap frozen malignant tumour samples snap frozen tumour tissue (Figure 5a and Table 2). FGF-1 mRNA was detected in all of the cell lines, both normal and tumorigenic (Figures 3c, 5c and Table 1). However, only 4/20 (20%) of the malignant samples demonstrated expression of FGF-1 (Figure 5c and Table 2). Again, oligonucleotides specific for GAPDH amplification were used as constitutive primers (Figures 3d and 5e). To confirm the identity of these ligand cDNAs, the selected PCR products were extracted from the agarose gel and directly sequenced. The remaining PCR products were Southern blotted and hybridized to 32P-labelled gene-specific internal oligonucleotides. Hybridization was detected in all of the normal OSE samples for both FGF-7 and FGF-1 (Figure 5b, d). In addition, 12/20 of the malignant tissue samples demonstrated hybridisation to FGF-7, whilst only 4/20 of the snap frozen EOC samples demonstrated hybridization to the FGF-1 oligonucleotide, in agreement with the ethidium bromide stained gels (Figure 5b, d).

Figure 5
figure5

Expression of FGF-1 and FGF-7 in normal OSE and ovarian carcinomas. Total RNA was reverse transcribed and the cDNA was amplified by PCR using FGF-1 or FGF-7 gene-specific primers and stained with ethidium bromide. DNA was then blotted onto a nylon filter and hybridized with a 32P-labelled FGF-1 or FGF-7-specific oligonucleotide. (a) Ethidium bromide stained agarose gel of normal OSE (lanes 1–4; 15–18) and ovarian carcinomas (lanes 5–14; 19–28) with FGF-7 specific primers. The product of lane 14 was directly sequenced to confirm its identity. (b) Southern blot hybridization of lanes 1–13 and 15–28 of panel (a) with 32P-labelled FGF-7-specific oligonucleotide. (c) Ethidium bromide stained agarose gel of normal OSE (lanes 1–4; 15–18) and ovarian carcinomas (lanes 5–14; 19–28) with FGF-1 specific primers. The product of lane 14 was directly sequenced to confirm its identity. (d) Southern blot hybridization of lanes 1–13 and 15–28 of panel (c) with 32P-labelled FGF-1-specific oligonucleotide. (e) Ethidium bromide stained gel of the same samples with GAPDH-specific primers

Detection of FGF-7 in ascitic fluid

A reproducible ELISA method was employed to determine the concentration of FGF-7 in ascitic fluid. A linear relationship was found between increasing concentration of recombinant FGF-7 and OD at 450 nm (Figure 6). FGF-7 protein was detectable in 38/54 (70%) of ascitic fluids by this ELISA. The mean level of FGF-7 detected was 0.7 ng/ml, and 9/54 (17%) of the samples contained levels of FGF-7 greater than 2.0 ng/ml. The concentrations of FGF-7 in the ascitic fluids of patients from whom snap frozen tumour tissue was obtained is presented in Table 2; where ascitic fluid was present and available, FGF-7 protein was detectable in 6/9 (67%) of the samples. FGF receptor 2-IIIb mRNA was also detected in the tumour tissue in 4/6 of the patients, including two patients where the concentration of FGF-7 in the ascitic fluid exceeded 2 ng/ml.

Figure 6
figure6

Detection of recombinant FGF-7 by ELISA. Typical standard curve used in the detection and quantification of recombinant FGF-7 (rhFGF-7). A linear relationship between concentration of rhFGF-7 and OD at 450 nm was obtained over the range 0–2 ng/ml

Co-expression of E-cadherin protein and FGF receptor 2-IIIb mRNA in ovarian carcinomas

Although the number of tumours was small (20 cases), an attempt was made to determine whether induction of FGF receptor 2-IIIb expression is related to expression of the epithelial marker E-cadherin. Hence, we studied the expression of E-cadherin by immunocytochemistry. Sixteen tumours scored >11 for E-cadherin expression, whilst four tumours demonstrated a score of between 1–11 (Table 2). There appeared to be a correlation between E-cadherin immunostaining and FGF receptor 2-IIIb mRNA expression and, where tumours were scored as >11 or <11 for E-cadherin expression, this relationship achieved statistical significance (P = 0.012; Fisher Exact Test) (Table 3).

Table 3 Relationship between expression of FGF receptor 2-IIIb and E-cadherin in ovarian carcinomas

Ovarian cancer cell lines expressing FGF receptor 2-IIIb respond to FGFs 1 and 7

Primary cultures of normal OSE did not respond to FGFs 1 or 7 over a range of 100 pg/ml to 10 ng/ml; whereas addition of 10% fetal calf serum (FCS) induced a highly significant (3–5-fold) induction of DNA synthesis (Figure 7a). In contrast, the 41 M, MDAH-2774 and SK-OV-3 ovarian carcinoma cell lines responded to FGFs 1 and 7 over a similar range of concentrations; the most significant induction of DNA synthesis occurred at 10 ng/ml (2–3-fold induction) (Figure 7b). Administration of higher concentrations of these ligands did not produce a greater stimulation of DNA synthesis. The response of MDAH-2774 cells to FGF-1 at 10 ng/ml was particularly striking with the induction of DNA synthesis at these concentrations of ligand exceeding that of 10% FCS.

Figure 7
figure7

Response of primary cultures of normal OSE and ovarian carcinoma cell lines to FGF-1 and FGF-7. (a) Primary cultures of ovarian surface epithelium were incubated in serum-free medium (control) and stimulated with FGFs 1 or 7 at a range of concentrations. (b) Ovarian carcinoma cell lines were incubated in serum free medium (control) and stimulated with either FGF-1 or FGF-7 at a range of concentrations. Positive controls were stimulated with 10% (v/v) FCS. The incorporation of [3H]-thymidine was determined 24 h later by scintillation counting. *Significant induction of DNA synthesis (P<0.001, students t-test)

To determine whether induction of DNA synthesis in ovarian carcinoma cell lines occurs, at least in part, by an autocrine mechanism, the effects on DNA synthesis of blocking antisera to these ligands was investigated in the 41 M cell line. DNA synthesis was inhibited by 60% in the presence of 15 μg/ml of FGF-7 blocking antiserum and 75% in the presence of 100 μg/ml of FGF-1 blocking antiserum (Figure 8b), when the cells were assayed in serum free medium. DNA synthesis in primary cultures of normal OSE was not affected by addition of these antibodies (Figure 8a). The blocking antisera did not produce a cytotoxic effect sufficient to cause loss of cell viability; trypan blue exclusion was maintained when cells were exposed to the concentrations of blocking antisera used in the assay. Moreover, following the removal of blocking antisera the cells plated normally and continued to proliferate.

Figure 8
figure8

Effects of a neutralizing antibody to FGF-1 and FGF-7 on proliferation of normal OSE and an ovarian carcinoma cell line. (a) Primary cultures of ovarian surface epithelium were incubated in serum-free medium (control) and treated with neutralizing antibody to either FGF-1 (100 μg/ml), FGF-7 (10 μg/ml or 15 μg/ml respectively) or 10% FCS. (b) The ovarian carcinoma cell line, 41 M, was incubated in serum-free medium (control) and treated with neutralizing antibody to FGF-1, FGF-7 or 10% (v/v) FCS. The incorporation of [3H]-thymidine was determined 24 h later by scintillation counting. *Significant induction of DNA synthesis (P<0.001, students t-test); **Significant reduction of DNA synthesis (P<0.001, students t-test)

Discussion

In this report, we have demonstrated that expression of the mRNA for FGF receptor 2-IIIb is induced in the majority of epithelial ovarian cancers. We have also demonstrated that FGF receptor 2-IIIb is translated into a functional receptor protein in three EOC cell lines; FGF-7 signalling occurs exclusively through the FGF receptor 2-IIIb isoform and this ligand can significantly induce DNA synthesis in EOC cell lines that express FGF receptor 2-IIIb. FGF-1 can also signal through FGF receptor 2-IIIb and this ligand also stimulates proliferation of these cell lines. However, FGF-1 is more promiscuous and it is possible that this ligand also signals through other FGF receptor isoforms in ovarian cancers. Little is known about the expression of other FGF receptor isoforms in ovarian cancer. However, FGF-2 has been shown to induce proliferation of several ovarian cancer cell lines and this ligand can signal through the majority of FGF receptor isoforms, which also bind FGF-1. Therefore, it can be presumed that at least some ovarian cancers express one or more of these FGF receptor isoforms. However, in this study we were unable to stimulate three primary cultures of normal human OSE and a human OSE cell line with FGF-1. Therefore, at least in normal OSE cells, these receptor isoforms are either not expressed, not functional, or do not mediate DNA synthesis.

The IIIb isoform of FGF receptor 2 has been reported to be restricted to epithelial cells (Finch et al., 1989; Rubin et al., 1989). Our findings that FGF receptor 2-IIIb is not expressed by normal OSE, but is induced in ovarian cancer gives further support to the concept that the OSE becomes more firmly committed to a complex epithelial phenotype during the process of malignant transformation. The opposite is true of most epithelial tissues, where loss of epithelial differentiation and histotypic organisation are among the earliest histologically identifiable changes in carinogenesis. The paradoxical differentiation of ovarian carcinomas may be a consequence of the pluripotential nature of the OSE which expresses markers of mesenchymal and simple epithelial differentiation in culture and retains the competence of its embryological precursor, the coelomic epithelium, to differentiate to complex Mullerian duct epithelial cell types. It has been reported previously that other markers of epithelial differentiation are induced in ovarian cancer development, including the CA125 (Klug et al., 1984; Bast et al., 1995) and E-cadherin (Sundfeldt et al., 1997) proteins. E-cadherin has been proposed to act as a ‘master gene’ that activates and regulates the expression of other genes that are required for epithelial differentiation. In bladder cancer, expression of E-cadherin and FGF receptor 2-IIIb are closely related (Diez de Medina et al., 1999). Although the numbers were small (n = 20), our study suggests that E-cadherin and expression of FGF receptor 2-IIIb mRNA are also related in ovarian cancer; the EOCs expressing the highest levels of E-cadherin protein expressed FGF receptor 2-IIIb mRNA. Transfection of E-cadherin was sufficient to induce mesenchymal-to-epithelial transition (Auersperg et al., 1999) and the tumorigenic phenotype in SV40-immortalized human OSE cells (Ong et al., 2000). It is tempting to speculate that induction of the tumorigenic phenotype may be mediated, at least in part, by signalling through aberrantly expressed epithelial-specific growth factor receptors such as FGF receptor 2-IIIb. This may occur by an autocrine mechanism, as normal OSE and many ovarian cancers express the mRNAs for FGFs 1 and 7.

Metaplastic changes to more epithelial phenotypes and an accompanying induction of E-cadherin expression is frequently seen in surface invaginations, inclusion cysts and areas of dysplastic epithelium, suggesting that these events may be among the earliest events in malignant transformation of the ovarian epithelium. If induction of FGF receptor 2-IIIb also occurs at such an early stage, it may result in proliferation through a paracrine or autocrine mechanism. FGF-7 has been shown to be an important paracrine factor in normal ovarian physiology; FGF-7 produced by thecal cells can induce proliferation of granulosa cells and therefore plays a role in follicular expansion (Parrott et al., 1994). Ovarian epithelial cells in inclusion cysts may be in close proximity to these cells, and may also be susceptible to paracrine stimulation by FGF-7. It is not known as yet whether FGF-1 acts as a paracrine growth factor in the ovary.

In this study, we have shown that normal OSE expresses the mRNA for FGFs 1 and 7, but expression of these ligands is lost in a proportion of EOCs. This may also be related to the epithelial differentiationof OSE in ovarian cancer, because these ligands, especially FGF-7, are more commonly expressed by mesenchymal cells (Finch et al., 1989; Rubin et al., 1989). However, at least 50% of the ovarian cancers in our study retained either FGF-1 and/or FGF-7 mRNA expression, indicating the potential of these ligands to act as autocrine growth factors in EOCs. Indeed, we have demonstrated that FGFs 1 and 7 can act as autocrine factors in at least one EOC cell line by partially inhibiting its proliferation with blocking antibodies to FGFs 1 and 7. Furthermore, we have detected abundant quantities of FGF-7 in the majority of ascitic fluids of patients suffering from ovarian cancer. The identification of FGF-7 as a growth factor in ascitic fluid may have important implications for the treatment of patients with ovarian cancer since EOCs normally metastasise by detachment from the ovary into the peritoneal cavity, with subsequent attachment to the peritoneal mesothelium and invasion of abdominal organs. Ascitic fluid bathes EOC cells in the peritoneal cavity, and inhibiting EOC proliferation by administration of FGF receptor 2-IIIb antagonists may potentially limit the spread of the disease and subsequent morbidity and mortality. Another member of the FGF family, FGF-10, has recently been cloned, has a similar sequence to FGF-7 and also functions through FGF receptor 2-IIIb (Lu et al., 1999). However, it is not known whether this ligand is expressed by EOCs or is present in ascitic fluid. In addition to their roles as mitogens, FGFs have also been implicated in other processes associated with neoplasia including motility, invasion and angiogenesis. Ovarian cancer cells must acquire all of these phenotypes in order to form life-threatening secondary tumours. Signalling by FGFs through aberrantly expressed FGF receptor 2-IIIb may, at least in part, modulate these processes. We are currently investigating the effects of FGFs 1 and 7 on these processes in ovarian cancer cell lines in vitro and in xenograft models in vivo.

In summary, we have shown that FGF receptor 2-IIIb is frequently expressed in epithelial ovarian cancer, whereas the ovarian surface epithelium does not express this receptor isoform. The induction of expression of this epithelial-specific receptor isoform may be related to the increased commitment of the OSE to an epithelial phenotype when it undergoes malignant transformation. Epithelial ovarian cancer cell lines that express FGF receptor 2-IIIb mRNA can respond to two of its ligands, FGFs 1 and 7, by an autocrine and/or paracrine mechanisms. Inhibiting this growth signalling pathway may be a potential therapeutic option in patients with epithelial ovarian cancer.

Materials and methods

Primary culture of human OSE and cell lines

Primary cultures of human ovarian surface epithelium were established according to the method of Auersperg (Kruk et al., 1990). Briefly, ovaries were scraped using a cell scraper to release the ovarian epithelium which is only tenuously attached to the underlying stroma. The sheets of epithelium removed were transported in Hanks Buffered Saline Solution (HBSS) (Life Technologies, Paisley, UK) before centrifugation. They were then resuspended in medium comprising 40% (v/v) RPMI (Life Technologies) and 40% (v/v) Ham's F-12 (Life Technologies) with 10% (v/v) added FCS (Life Technologies). Once established the cultures were maintained in this medium in an atmosphere of 5% CO2, at 37°C. Primary cultures were established from women undergoing oophorectomy for benign disease (hysterectomy and oophorectomy for menorrhagia). All patients gave written consent and approval was granted from the local ethics committee. Characterization of the cultures was achieved by staining methanol-fixed preparations of the cultures with antibodies to cytokeratins 8 (clone LE41), 18 (clone LE61) and 19 (RCK108) (Dako, Ely, UK) (Figure 1). Using co-staining techniques with propidium iodide to stain all nuclei, an estimate of the epithelial content could be made. In all cases the cultures had a greater than 98% content of epithelial cells.

The IOSE-van cell line was obtained from Dr Nelly Auersperg, University of British Columbia, Vancouver, Canada and is derived from normal OSE immortalized with SV40. The ovarian carcinoma cell lines 41-M, SK-OV-3 and MDAH-2774 were originally isolated from ascitic fluid in patients with ovarian carcinoma and were obtained from the ATCC (American Type Culture Collection, Mancissa, USA). All cell lines were grown in 90% (v/v) Dulbecco's Modified Eagle's Medium (DMEM) (Life Technologies) supplemented with 10% (v/v) FCS and 1% (v/v) penicillin and streptomycin (Life Technologies).

Tissue specimens

Twenty tumour samples were obtained from twenty patients who were undergoing primary surgery for ovarian cancer at the Royal Victoria Infirmary, Newcastle, UK. Diagnosis of ovarian cancer was confirmed by histopathological examination. These samples were snap frozen in liquid nitrogen at time of surgery. Pathology details are listed in Table 2. The patients had an average age of 64.5 with range 39–82. Tumours were undifferentiated (1), poorly differentiated (1), papillary (2), mucinous (2), endometrioid (3), borderline (4) or serous (7) with stages I (6), III (10) or IV (4).

RT–PCR

Total RNA was extracted from both cell cultures and tissue samples using the RNeasy Mini Kit (Qiagen, Crawley, UK). cDNA was prepared from 1 μg of total RNA from each sample using Superscript II (Life Technologies). Oligonucleotide primers for amplifying FGF receptor 2-IIIb, FGF-1, FGF-7 and GAPDH were synthesised by VH Bio Ltd, Newcastle, UK (Table 4). The Polymerase Chain Reaction (PCR) was carried out in a 40 μl reaction mixture containing 2 μl cDNA, 200 μM dNTPs, 0.2 μl forward and reverse primers and 0.5 μl; AdvanTaq DNA Polymerase (Clontech, Basingstoke, UK) for 40 cycles of amplification in a thermal cycler (Perkin Elmer, Beaconsfield, UK) with 1 min denaturation, at 94°C, 1 min annealing at 54°C and 1 min chain extension at 72°C. Selected PCR products were directly sequenced to confirm their identity.

Table 4 Primers for RT–PCR and Southern blotting

Southern blotting

Specific internal oligonucleotide probes were designed to hybridize to the genes of interest (Table 4). Ethidium bromide stained PCR products were blotted onto Hybond N+ membrane (Amersham Pharmacia Biotech, Little Chalfont, UK) and probed with the relevant internal oligonucleotide. Following pre-hybridization the membranes were hybridized at 50°C for 2 h with the corresponding 32P-radiolabelled oligonucleotide probe in a buffer containing 5 × SSC (3 M sodium chloride, 0.3 M sodium citrate; pH 7.0) (Sigma, Dorset, UK), 5 × Denharts solution and 0.5% (w/v) SDS (sodium dodecyl sulphate) (Sigma) (Leung et al., 1997). The membranes were washed in successive washes of 2 × SSC/0.1% (w/v) SDS twice at 50°C for 15 min each and 1 × SSC 0.1% (w/v) SDS twice for 15 min. Hybridization products were visualized by autoradiography.

Northern blotting

Poly (A)-RNA was extracted from three ovarian carcinoma cell lines using an Oligotex direct mRNA Kit (Qiagen). Aliquots of RNA (10 μg) were fractionated on a 1.5% agarose -formaldehyde gel and transferred onto hybond N+ membrane. Membranes were pre-hybridized at 42°C for 2 h in a hybridization buffer containing 50% formamide, 2 × Denhardt's Solution, 5 × SSPE, 0.1% SDS and 100 μg/ml salmon sperm DNA. Subsequently, the pre-hybridized membranes were probed for 12 h at 42°C with a previously sequenced FGF receptor 2-IIIb PCR product labelled with [α-32P]dCTP (Amersham Pharmacia Biotech). Following hybridization, the membranes were washed in successive washes of 2 × SSC 0.1% (w/v) SDS twice for 10 min, 0.2 × SSC 0.1% (w/v) SDS once for 15 min and 0.1% SSC 0.1% (w/v) SDS twice for 15 min. All washes were carried out at 65°C. Hybridization products were visualized by autoradiography. The same filter was stripped and rehybridized with a previously sequenced GAPDH PCR probe.

ELISA for FGF-7

Expression of FGF-7 in ascitic fluid was studied using a sandwich ELISA (R&D Systems, Oxon, UK) using the method recommended by the manufacturer with minor modifications (Mehta et al., 2000). Briefly, 3 μg/ml of rabbit anti-mouse antibody (Dako), diluted in 0.05 M Tris.HCl pH 7.6, was coated onto a 96-well plate (Dynex, Virginia, USA) and incubated overnight. 0.9 μg/ml of mouse anti-FGF-7 antibody in phosphate buffered saline (PBS), pH 7.6, was added to each well and incubated for 30 min. Unbound antibody was removed by washing and blocked with 3% (w/v) bovine serum albumin (BSA) and 5% (w/v) sucrose in PBS. 150 ng/ml biotinylated goat anti-FGF-7 antibody (diluted in 0.05 M Tris.HCl, pH 7.6) was added to each well and incubated for 1 h. Strep ABC Complex/HRP was added as per the manufacturers instructions (Dako) for 30 min and the substrate, tetramethylbenzidine (TMB) with hydrogen peroxide, was added for 10 min. The reaction was stopped with 1 M sulphuric acid (Life Technologies) and the plate read at 450 nm. A calibration curve was constructed using varying concentrations of recombinant FGF-7 (Sigma) (Figure 6).

Expression of E-cadherin protein

Formalin-fixed, paraffin-embedded blocks of tumour samples from twenty patients undergoing surgery for advanced ovarian cancer were provided by Department of Pathology, Royal Victoria Infirmary, Newcastle, UK (Table 2). Sections were stained with an anti-human E-cadherin mouse monoclonal antibody (clone HECD-1) (R&D Systems) as previously described, with minor modifications (Davies et al., 1998). Briefly, sections were dewaxed in xylene and antigen retrieval was carried out in 1.2 mM Ethylenediamine Tetraacetic acid (EDTA) buffer, pH 8.0, for 1 min in a boiling pressure cooker. Following washing in PBS, sections were treated with 3% hydrogen peroxide in methanol solution to remove endogenous peroxidase activity. Blocking was carried out in normal rabbit serum for 10 min and 1 μg/ml E-cadherin antibody was added to the sections for 30 min at room temperature. 70 μg/ml of a rabbit anti-mouse biotin-tagged IgG secondary antibody (Dako) was applied for 30 min. Bound antibody was detected using the Vectastain ABC method as per the manufacturers instructions (Vectastain, Peterborough, UK). Diaminobenzidine tetrachloirde (Sigma) was the chromagen used to detect the bound antibody. Sections were viewed by microscopy and reviewed by two independent observers (IAS and BRD). A scoring system was applied and any differences of opinion were resolved by discussion.

DNA synthesis assays

Cells were plated out into the middle-60 wells of 96-well plates (Corning, High Wycombe, UK) at a density of 1 × 104 cells/well, and cultured in 90% (v/v) DMEM supplemented with penicillin, streptomycin and 10% (v/v) FCS for 3 h at 37°C. Cells were then washed twice with HBSS and subsequently cultured in serum-free DMEM containing 250 μg/ml BSA (Sigma). After 24 h, growth factors or 10% FCS were added to each well and cells cultured for a further 24 h at 37°C. 1 μCi/ml [3H]-thymidine (Amersham Pharmcia Biotech) and 1 μM of non-radiolabelled thymidine were then added to each well for a further 12 h at 37°C. Cells were washed with HBSS, detached by trypsinization and the DNA precipitated with 100 μl 10% (w/v) trichloroacetic acid (Sigma) at 4°C for 2 h. The amount of incorporated [3H]-thymidine was then measured by β-scintillation counting (Microbeta Plate, Wallac, Finland). In experiments where blocking antibodies were used, the blocking antibodies to FGFs 1 and 7 (R&D Systems) were diluted in serum-free DMEM containing 250 μg/ml BSA and the cells incubated for 24 h before adding [3H]-thymidine and assaying as described above.

References

  1. Auersperg N, Pan J, Grove BD, Peterson T, Fisher J, Maines-Bandiera S, Somasiri A, Roskelley CD . 1999 Proc. Natl. Acad. Sci. USA 96: 6249–6254

  2. Basilico C, Moscatelli D . 1992 Adv. Cancer Res. 59: 115–165

  3. Bast Jr RC, Boyer CM, Xu FJ, Wiener J, Dabel R, Woolas R, Jacobs I, Berchuck A . 1995 J. Cell. Biochem. Suppl. 23: 219–222

  4. Berchuck A, Kohler MF, Boente MP, Rodriguez GC, Whitaker RS, Bast Jr RC . 1993 Cancer 71: 545–551

  5. Burgess WH, Maciag T . 1989 Annu. Rev. Biochem. 58: 575–606

  6. Crickard K, Gross JL, Crickard U, Yoonessi M, Lele S, Herblin WF, Eidsvoog K . 1994 Gynecol. Oncol. 55: 277–284

  7. Davies BR, Worsley SD, Ponder BA . 1998 Histopathology 32: 69–80

  8. De Medina SG-D, Popov Z, Chopin DK, Southgate J, Tucker GC, Delouvee A, Thiery JP, Radvanyi F . 1999 Oncogene 18: 5722–5726

  9. Di Blasio AM, Cremonesi L, Vigano P, Ferrari M, Gospodarowicz D, Vignali M, Jaffe RB . 1993 Am. J. Obstet. Gynecol. 169: 1517–1523

  10. Dyck HG, Hamilton TC, Godwin AK, Lynch HT, Maines-Bandiera S, Auersperg N . 1996 Int. J. Cancer 69: 429–436

  11. Finch PW, Rubin JS, Miki T, Ron D, Aaronson SA . 1989 Science 245: 752–755

  12. Friesel RE, Maciag T . 1995 FASEB J. 9: 919–925

  13. Givol D, Yayon A . 1992 FASEB J. 6: 3362–3369

  14. Hernandez E, Rosenheim NB . 1989 Manual of Gynecologic Oncology Churchill Livingstone: London pp 57–74

  15. Inoue M, Ogawa H, Miyata M, Shiozaki H, Tanizawa O . 1992 Am. J. Clin. Pathol. 98: 76–80

  16. Klug TL, Bast Jr RC, Niloff JM, Knapp RC, Zurawski Jr VR . 1984 Cancer Res. 44: 1048–1053

  17. Kruk PA, Maines-Bandiera SL, Auersperg N . 1990 Lab. Invest. 63: 132–136

  18. Kruk PA, Uitto VJ, Firth JD, Dedhar S, Auersperg N . 1994 Exp. Cell. Res. 215: 97–108

  19. Leung HY, Mehta P, Gray LB, Collins AT, Robson CN, Neal DE . 1997 Oncogene. 15: 1115–1120

  20. Lu W, Luo Y, Kan M, McKeehan WL . 1999 J. Biol. Chem. 274: 12827–12834

  21. Maines-Bandiera SL, Auersperg N . 1997 Int. J. Gynecol. Pathol. 16: 250–255

  22. Mason IJ . 1994 Cell 78: 547–552

  23. Mehta PB, Robson CN, Neal DE, Leung HY . 2000 J. Urol. 164: 2151–2155

  24. Miki T, Bottaro DP, Fleming TP, Smith CL, Burgess WH, Chan AM, Aaronson SA . 1992 Proc. Natl. Acad. Sci. USA 89: 246–250

  25. Ong A, Maines-Bandiera SL, Roskelley CD, Auersperg N . 2000 Int. J. Cancer 85: 430–437

  26. Orr-Urtreger A, Bedford MT, Burakova T, Arman E, Zimmer Y, Yayon A, Givol D, Lonai P . 1993 Dev. Biol. 158: 475–486

  27. Parrott JA, Vigne JL, Chu BZ, Skinner MK . 1994 Endocrinology 135: 569–575

  28. Peters KG, Werner S, Chen G, Williams LT . 1992 Development 114: 233–243

  29. Resta L, Russo S, Colucci GA, Prat J . 1993 Obstet. Gynecol. 82: 181–186

  30. Rubin JS, Osada H, Finch PW, Taylor WG, Rudikoff S, Aaronson SA . 1989 Proc. Natl. Acad. Sci. USA 86: 802–806

  31. Scully RE . 1977 Am. J. Pathol. 87: 686–720

  32. Scully RE . 1992 Ovarian Cancer 2: Biology, diagnosis, management Chapman and Hall: London pp 199–205

  33. Scully RE . 1995 J. Cell. Biochem. Suppl. 23: 208–218

  34. Sundfeldt K, Piontkewitz Y, Ivarsson K, Nilsson O, Hellberg P, Brannstrom M, Janson PO, Enerback S, Hedin L . 1997 Int. J. Cancer 74: 275–280

  35. Veatch AL, Carson LF, Ramakrishnan S . 1994 Int. J. Cancer 58: 393–399

  36. Westermann AM, Beijnen JH, Moolenaar WH, Rodenhuis S . 1997 Cancer Treat. Rev. 23: 113–131

  37. Yamashita T, Yoshioka M, Itoh N . 2000 Biochem. Biophys. Res. Commun. 277: 494–498

  38. Young RH, Clement PB, Scully RE . 1989 Diagnostic and Surgical Pathology Sternberg, SS. (ed.) Raven Press: New York pp 1655–1734

Download references

Acknowledgements

We would like to thank Dr Anne Wilson (Oncology Research Laboratory, Derby, UK) for providing some of the ovarian cancer ascitic fluid and Mrs Barbara Innes for technical assistance with the E-cadherin immunocytochemistry. This work was supported by a William Ross PhD studentship from the Cancer Research Campaign (CRC).

Author information

Correspondence to Barry R Davies.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Steele, I., Edmondson, R., Bulmer, J. et al. Induction of FGF receptor 2-IIIb expression and response to its ligands in epithelial ovarian cancer. Oncogene 20, 5878–5887 (2001) doi:10.1038/sj.onc.1204755

Download citation

Keywords

  • FGF receptor 2
  • ovarian cancer
  • ovarian surface epithelium

Further reading