Int. J. Cancer: 52,76-84 (1992) 0 1992 Wiley-Liss, Inc.

'

Publication of the InternationalUnion Against Cancer Publication de I'Union lnternationale Contre le Cancer

MORPHOLOGICAL, IMMUNOHISTOCHEMICAL AND BIOCHEMICAL CHARACTERIZATION OF 6 NEWLY ESTABLISHED HUMAN OVARIAN CARCINOMA CELL LINES V. MOBUS',5, C.D. GERHARZ?, U. PRESS',R. MOLL^, T. BECK',W. MELLIN4, K. POLLOW',P.G. KNAPSTEIN' and R. KREIENBERG' Departments of 'Obstetricsand Gynecology, 2Pathologyand 'Experimental Endocrinology, Universityof Mainz, 6500 Mainz; and 4Departmentof Pathology, Universityof Miinster, 4400 Miinster, Germany. Six permanent human tumor cell lines (OV-MZ- I to 6) were establishedfrom 6 patients with serous adenocarcinomasof the ovary. These cell lines were derived from both solid tumors and ascites, from pre-treated and untreated patients, and are available over a range of in vitro passage numbers. The tumor cells grow as monolayers and develop foci of "piled-up" cells in confluent cultures. Flow cytophotometry showed that all the lines exhibited DNA hyperdiploidywith DNA tetraploidy in one cell line and DNA aneuploidy in the other cell lines. The mean population doublingtime ranged from 24 to 52 hr. Transmission electron microscopy demonstrated that the tumor cells of all cell lines exhibited features of epithelial differentiation such as desmosomes and intracellular gland-like lumina. Immunocytochemical analysis showed that the co-expressionof cytokeratins and vimentin, which is a feature of ovarian serous cystadenocarcinomas in situ, was fully preserved in the majority of cell lines. The main cytokeratin polypeptides expressed were numbers 7,8, 17, I 8 and 19. The tumor-associated antigen CA- 125, but not CEA, was shed in the culture supernatant. This was in accordancewith FACScan analysis of the cell lines and the level of CA-I25 and CEA in the patients' serum. The estrogen and progesterone receptors were negative both in the cell lines and in the original tumors. These new ovarian carcinoma cell lines will be valuable models for further investigations into a variety of biological properties.

o 1992 Wiley-Liss,Inc. Ovarian cancer is difficult to detect at an early stage. Despite radical surgery and the clinical introduction of the potent anti-neoplastic agent cisplatin one decade ago, and nowadays its derivatives, the prognosis still remains poor. Cancer of the ovary remains the fourth leading cause of cancer deaths among women in most Western countries. One of the major clinical problems is the rapid emergence of drug resistance and associated cross-resistance. There is increasing interest, however, in the use of hormones, due to reports of variable success with hormonal therapy (Schwartz et al., 1982; Nash et al., 1989). The establishment of relevant experimental systems such as permanent cell cultures is therefore needed to gain more insight into these problems. In view of the relative difficulty of obtaining sterile tissue samples and transferring them successfully into culture, only a limited number of permanent human ovarian cancer cell lines have been described (Buick et al., 1985; Hill et al., 1987; Horowitz et al., 1985; Wolf et al., 1987; Van Niekerk et af., 1988; Crickard et al., 1989; Hills et al., 1989). Since cell lines may lose many of the tumor-specific properties of the original tumor during prolonged in vitro culture, we tried to establish new lines of our own. MATERIAL AND METHODS

Patient data OV-MZ-1. The cell line OV-MZ-1 was derived from ascitic fluid obtained from a 44-year-old patient 15 months after initial surgery with supra-cervical hysterectomy, bilateral salpingo-oophorectomy and partial omentectomy because of ovarian cancer FIGO stage 111. After 10 courses of chemother-

apy with CAP the patient was clinically in complete remission. A second-look laparatomy was performed with total omentectomy, douglasectomy and resection of the cervix. Histologic examination showed invasive tumoral residues and the patient received 6 more cycles of CP. Chemotherapy was stopped because of tumor progression. At this time, the cell line was established. Despite irradiation followed by second-line chemotherapy, the patient died 4 months later. OV-MZ-2. In September 1983, a 55-year-old patient had a laparatomy with hysterectomy, bilateral salpingo-oophorectomy and omentectomy because of ovarian cancer FIGO stage IV. She subsequently received 3 courses of cyclophosphamide. Further tumor progression was indicated by liver metastasis, and chemotherapy was changed to 11 courses of CAP, followed by 6 courses of CP. At this time the patient showed progression again, with extensive production of ascites, from which the line was established in January 1986. The patient died in April 1986. OV-MZ-3. In July 1984, the patient had a laparatomy with bilateral salpingo-oophorectomy and partial omentectomy. The disease stage was FIGO 111. Hysterectomy had been performed some years earlier. Post-operatively, the patient received 8 courses of chemotherapy with CAP. When the patient was clinically in complete remission, a second-look laparatomy with omentectomy and douglasectomy was performed in May 1985. Histological examination showed invasive residues of a papillary cystadenocarcinoma. Although chemotherapy (CP) was continued, the patient showed progression. The cell line was established from ascites in February 1986. The patient died in July 1986. OV-MZ-4. This cell line was established in October 1986 from the malignant ascites of a 64-year-old woman with ovarian cancer FIGO stage IV (liver metastasis). Some days after the ascites punction, a laparatomy with partial omentectomy and bilateral salpingo-oophorectomy was performed. Despite one course of chemotherapy with CAP, the tumor showed rapid progression and the patient died in December 1986. OV-MZ-5. A 57-year-old patient underwent exploratory laparatomy in August 1985, with no possibility of tumor debulking. At this time the cell line was established from ascites. The patient showed extended disease F I G 0 stage IV 'To whom correspondence and reprint requests should he addressed, at Department of Obstetrics and Gynecology, University of Mainz, Langenheckstr. 1,6500 Mainz. Germany. Abbreviations: CEA, carcinoembryonic antigen; CP, cisplatini cyclophosphamide; CAP, cisplatinianthracyclineicyclophosphamide; CK, cytokeratin; GFP, glial filament protein: IF, intermediate filament: MAb, monoclonal antibody; SN, culture supernatant; DI, DNA index; SDS-PAGE, sodium dodecylsulfate polyacrylamide gel electrophoresis. Received: November 19,1991 and in revised form March 25.1992.

CHARACTERIZATION OF HUMAN OVARIAN CARCINOMA CELL LINES

with liver metastasis. She subsequently received 10 courses of chemotherapy with the C P regimen. After partial remission, the disease again progressed, and the patient died in January 1987. OV-MZ-6. A 70-year-old patient had advanced cancer ( F I G 0 stage IV), with malignant effusion in the pleural cavity. Exploratory laparatomy was performed in March 1987 with no possibility of tumor debulking. The cell line was established from malignant ascites drained during this laparatomy. The patient received no further therapy, and died 6 weeks later.

Establishment of the cell lines In most cases we received the tumor from our own histology department: 43 tissue specimens were obtained and 6 lines successfully established, an overall success rate of 14%. After removal of fat and necrotic parts, tumor specimens were minced into pieces of approximately 1 mm3. We saw no advantage in exposing the tumor to enzymatic digestion. The cells in the supernatant were aspirated with a Pasteur pipette and the remaining tumor fragments minced again. The cells were subsequently Centrifuged 3 times at 1000 g for 10 min. Effusion cells were harvested by centrifugation at lOOOg for 10 min, twice re-suspended and centrifuged again. In one patient there was considerable contamination of the effusion with red blood cells, which were removed by Ficoll centrifugation. The remaining tumor cells were then transferred to T-30 flasks (NUNC, Roskilde, Denmark). Culture flasks were initially treated to selective trypsinization (trypsin 0.05% w/v, EDTA 0.02% w/v), to avoid fibroblast overgrowth. In addition, we attempted to remove the fibroblasts mechanically or to transfer the tumor cells selectively. Routine assays for Mycoplasma, fungi and bacterial contamination were negative. Culture conditions The cell lines were cultured on plastic at 37°C in a 5% COz and 95% air atmosphere. In the early phase of cultivation, the tumor cells were grown in CMRL-medium (GIBCO, Karlsruhe, Germany). When the tumor cells could be serially passaged, cells were grown in Dulbecco’s modified Eagle’s medium (DMEM, GIBCO). Both media were supplemented with 10% FCS (GIBCO), penicillin (100 U/ml), streptomycin (100 kg/ml), 1% (v/v) non-essential amino acids and sodium pyruvate and L-glutamine (2 mM). All cell lines were frozen in liquid nitrogen at early passages and at intervals of 5 passages during increasing passage numbers. Growth parameters The rate of cellular proliferation was measured in the cultures in logarithmic growth phase. Tumor cells ( 5 x los) were seeded in 25-cm2 culture flasks and re-fed on day 3. From day 3 to day 9, total cell numbers were determined in triplicates with the Neubauer hemocytometer. Preparation of cytosols; steroid receptor assay Cell line. One week before harvest, media were changed and DMEM supplemented with 5% charcoal-treated FCS was added. The cells were then suspended in cold assay buffer (10 mmolil phosphate buffer containing 10 mmol/l monothioglycerol, 1.5 mmolil EDTA, 3 mmol/l sodium azide and 10% glycerol, p H 7.5) and homogenized by ultrasonic disintegration of the cell membranes under permanent cooling. The homogenates were centrifuged at 105,OOOgfor 30 min at 4°C. The clear supernatant was used as cytosol. Original tissue specimen. Tissues were minced with a scalpel and homogenized in the above-mentioned buffer under permanent cooling using an Ultra Turrax homogenizer. Centrifugation was performed as described above.

77

Estrogen and progesterone receptor analysis of cytosols was performed with a dual label assay using 160 a-12sI-iodo-3,17estradiol (‘?“-E2) and 3H-R5020 as ligands, according to the method described by Grill et al. (1984). Tracer solution (100 k1) containing simultaneously lZsI-E2( 5 concentrations, range 0.25, 0.5, 1.0, 2.0 and 4.0 nmol/l) and 3H-R5020 (5 concentrations, range 0.5, 1.0,2.0,4.0 and 8.0 nmol/l) were incubated in duplicates overnight at 4°C with 100 kl of cytosol. Non-specific binding was assayed by the addition of 200 molar excesses of diethylstilbestrol and R5020 in a separate set of tubes. Bound and free steroids were separated by dextran-coated charcoal. Counting of radioactivity was performed using a dual label dpm program. The protein concentrations were determined according to Bradford (1976) using protein standard I (BioRad, CA). The binding data were calculated according to the method of Scatchard (1949) and were expressed as fmol receptor/mg of cytosol protein. Breast cancer cell line ZR-75-1 was used as a positive control. Flow cytophotometric DNA measurement Exponentially growing tumor cells were harvested and fixed in methanol. After centrifugation at 800 rpm for 5 min, cells were treated with 0.5% pepsin for 15 min, washed with sodium chloride solution and stained with DAPI (1 pg/ml in RPMI 1640); 2 x lo4cells were analyzed using a flow cytophotometer (FACSanalyzer, Becton-Dickinson, Heidelberg, Germany). The mean coefficient of variation (CV) was 4.1% in the tumor samples. Human peripheral-blood lymphocytes were used as a calibration standard for D N A diploidy. The ploidy level was expressed as the DNA index, the DNA index of diploid lymphocytes being 1.O. Cell-cycle analysis was performed as described by Dean and Jett (1974).

Scanning and transmission electron microscopy For scanning electron microscopy, the tumor cells were seeded on glass coverslips, fixed in situ by exposure to 2.5% phosphate-buffered glutaraldehyde solution (pH 7.2) and post-fixed in 2% osmium tetroxide solution. After dehydration in an ascending acetone series, the tumor cell monolayer was dried by the critical-point method and sputtered with gold. Electron photomicrographs were taken with a PSEM 501 Philips scanning electron microscope. For transmission electron microscopy, tumor tissue and tumor cells were fixed by exposure to 2.570 sodium-cacodylatebuffered glutaraldehyde solution (0.1 mol, pH 7.4) and postfixed in 1% sodium-cacodylate-buffered osmium tetroxide solution (0.1 mol, p H 7.4) prior to Epon embedding. These sections were contrasted with uranyl acetate and lead citrate. Electron photomicrographs were taken with an E M 410 Philips transmission electron microscope. Analysis of intermediate filament ( I F ) proteins by immunocytochemistry and gel electrophoresis Immunocytochemistry. Cells were grown on microscope slides to sub-confluent density, rinsed with PBS and fixed in methanol ( 5 min, -20°C) and subsequently briefly in acetone (-20°C). After air drying, the slides were stored at -20°C. The immunocytochemical methods used were those of indirect immunofluorescence microscopy and the indirect immunoperoxidase method, using standard techniques (Moll et al., 1988). The following MAbs against I F proteins were applied: (1) MAb Ks 18.174 specific for cytokeratin (CK) 18 (commercially available through Progen, Heidelberg, Germany); (2) MAb Ks 19.2.105 against CK 19 (Franke and Moll, 1987; Progen); (3) MAb CK 7 against CK 7 (Boehringer, Mannheim, Germany); (4) MAb AE 14 against CK 5 (Lynch et al., 1986; kindly provided by Dr. T.T. Sun, New York); ( 5 ) MAb E 3 against CK 17 (Troyanovsky et al., 1989; kindly provided by Dr. S.M. Troyanovsky, Moscow, Russia); (6) MAb 6B10 specific for CK

78

MOBUS ET AL.

4 (obtained from Euro-Diagnostics, Apeldoorn, The Netherlands); (7) MAb IT-Ks 20.2 against CK 20 (Moll et al., 1990; Progen). Since this MAb yielded negative results with all cell lines included in this study, it served also as a negative control. Cell lines derived from colorectal adenocarcinomas were positive with this antibody (results not shown); (8) MAb VIM-9 against vimentin (Pitz et al., 1987; Viramed, Martinsried, Germany); (9) h4Ab G-A-S against glial fibrillary (acidic) protein (GFP; Debus et al., 1983; Boehringer). Gel electrophoresis. Cytoskeletal residues obtained by subsequent extractions with a high-salt detergent buffer and a low-salt buffer (Achtstaetter et al., 1986) were analyzed by 2-dimensional gel electrophoresis using non-equilibrium p H gradient (NEPHG) electrophoresis in the first dimension and SDS-PAGE in the second dimension (Achtstaetter et al., 1986). Gels were stained using either Coomassie brilliant blue or a highly sensitive silver staining method (Achtstaetter et al., 1986).

Heterotransplantation in nude mice Tumorigenicity for all cell lines was tested by heterotransplantation in 6- to 8-week-old nu/nu mice (NMRI). The mice (n = 3 per tumor cell line) were obtained from the Versuchstieranstalt Hannover, Germany. Tumor cells were injected S.C. into both flanks at an inoculum size of 1 x lo7tumor cells each. Histological sections of the tumors were stained with hematoxylin and eosin. Determination of CA-125 and C E A In the supernatant. The secretion of CA-125 and CEA in the supernatant was determined in proliferating and resting cells. The supernatant of exponentially growing cell cultures was collected on days 3, 5 , 7 and 9 and the cells were counted. Resting cells were defined as cell cultures which had reached confluence for at least one week. After a cell count, the medium of resting tumor cells was totally renewed and the supernatant collected after 3 days. The supernatant was centrifuged at 1OOOg for 10 min and stored at -20°C until tested. The CA-125 level in the supernatant was determined by a solid-phase RIA (CIS, Dreieich, Germany), the CEA level by a solid-phase EIA (Hoffmann-LaRoche, Basel, Switzerland). By FACScan analysis. Murine monoclonal antibodies directed against CA-125 and CEA were purchased from CIS and Dianova-Immunotech (Marseille, France) respectively. Normal mouse IgGl was purchased from Becton Dickinson, FITC-conjugated goat anti-mouse IgG was obtained from Coulter (Krefeld, Germany). Quantitation of the tumorassociated surface antigens was performed on a FACScan cytofluorimeter. The cells were gently harvested by exposure to EDTA (0.05%) and filtered through a 30-km nylon mesh. Cells (1 x lo6) suspended in PBS were exposed to the specific or control antibody for 30 min on ice. The cells were then washed twice and re-suspended in PBS containing an appropriate FITC-conjugated second MAb. After another 30 min incubation on ice, cells were washed twice in PBS and analyzed at 488 nm. RESULTS

Figure 1 depicts the morphological aspects of the original ovarian tumors, which encompass a spectrum ranging from well differentiated serous cyst-adenocarcinomas to poorly differentiated adenocarcinomas. Table I summarizes some important characteristics of the cell lines. The lines have been in culture from 4 to 6 years, the passage number varying from 50 to 189. Three cell lines were established from pre-treated patients, the other 3 lines from untreated patients. There are 3 fast-proliferating cell lines (OV-MZ-2, -4 and -6) and 3 slowly proliferating cell lines (OV-MZ-1, -3 and -5).

The estrogen- and progesterone-receptor content of the cell lines was always negative, corresponding to the receptor status of the original tumor. As a positive control, we determined the receptor content of the breast cancer cell line ZR-75-1. We found an estrogen-receptor concentration of 12 fmol/mg, that of progesterone receptor being 832 fmol/mg. In the nude mice, 4 out of 6 cell lines were tumorigenic. These features have been shown to be reproducible for low and high passage numbers.

Flow cytometric D N A analysis Flow cytophotometry showed that all the cell lines exhibited DNA hyperdiploidy, with D N A tetraploidy in the cell line OV-MZ-1 and D N A aneuploidy in the other cell lines. The DNA index ranged between 1.4 and 4.4. Only the cell lines OV-MZ-1 and -6 were evaluable for S-phase analysis and showed an S-phase fraction of 35% vs. 13%. The remaining 4 cell lines showed multiple DNA stem lines, which impeded further analysis of the cell-cycle distribution. Morphological characteristics of the cell lines All cell lines proved to be strictly anchorage-dependent, growing as an adherent cell layer and forming multilayered sheets, clusters or papillary projections in confluent cultures. Electron microscopy scanning of the tumor cells of OV-MZ-1 to -4 showed that the cell shape was polygonal (Fig. 2, a d ) and formed loosely juxtaposed cell colonies (OV-MZ-1 and -2) or epithelial-like cell complexes (OV-MZ-3 and -4). The tumor cells of MZ-OV-S and -6 exhibited a more fibroblast-like appearance, with spindle-shaped tumor cells (Fig. 2, e, f). Microvillus-like cytoplasmic protrusions could be seen in all cell lines, and the cell surface of OV-MZ-2 in particular was studded with cytoplasmatic blobs and finger-shaped microspikes (Fig. 26). Transmission electron microscopy showed only minor differences between the different cell lines. The tumor cells were often separated by slit-like intercellular spaces bridged by cytoplasmic interdigitations and typical desmosomes. Occasionally the tumor cells exhibited intracellular lumina, indicating their origin from gland-forming adenocarcinomas. The cytoplasm of the tumor cells was rich in polyribosomes, mitochondria and intermediate filaments, whereas profiles of rough endoplasmatic reticulum were seldom seen. Monoparticulate glycogen deposits were often observed, and some tumor cells exhibited intracellular desmosomes as well as fat droplets. The nuclei were moderately irregular in shape, predominantly round to oval, showing finely scattered heterochromatin and multiple nucleoli. Patterns of IFproteins Results concerning expression of IF proteins, including the individual CK polypeptides, are summarized in Table 11. Immunocytochemical staining generally produced a distinctly fibrillar cytoplasmic pattern. Immunocytochemical and gel electrophoretic results were in good agreement. Immunocytochemically, C K 18 was detected uniformly in essentially all the cells of all the cell lines studied. This CK was therefore consistently present in gel electrophoresis, together with its partner CK 8. CK 19 was detected in most of the cells of all cell lines (Fig. 3). Quantitative differences were observed for CK 7, although this CK could be detected in most cell lines. The expression of CK 17 was distinctly heterogeneous both among the cell lines and among different cells of individual cell lines, with grossly varying proportions of immunostained cells. No other CK polypeptide could be detected by gel electrophoresis

CHARACTERIZATION OF HUMAN OVARIAN CARCINOMA CELL LINES

79

FIGURE1 - Morphological aspects of the original tumors by light microscopy: well-differentiated serous cystadenocarcinomas giving rise to OV-MZ-2 (O), OV-MZ-3 (c) and OV-MZ5 ( e ) ;moderately differentiated serous cystadenocarcinomas giving rise to OV-MZ-1 ( u ) and OV-MZ-4 ( d ) ;poorly differentiated serous cystadenocarcinoma giving rise to OV-MZ-6 (f). Scale bar, 100 p n .

(Fig. 3). Using immunocytochemistry, however, only a few cells expressing CK 4 and, even more rarely, CK 5 were occasionally found, while the reactions for CK 20 were always completely negative (not shown).

Interestingly, most of the cell lines studied expressed the mesenchymal I F protein vimentin in addition to CK polypeptides (Fig. 3a). Results from immunocytochemistry showed that the majority of cells were usually vimentin-positive, but in

80

MOBUS E T AL. TABLE 1 -CLINICAL AND BIOLOGICAL CHARACTERISTICS OF THE CELL LINES AND THE ORIGINAL TUMOR Cell line

Initiation of culture

Source

OV-MZ-1

1985

ascites

OV-MZ-2

1986

ascites

OV-MZ-3 OV-MZ-4 OV-MZS OV-MZ-6

1986 1986 1985 1987

solid ascites ascites ascites

Pre-treatment

Survival of patients in months

Median doubling time (hrs)

23

38

10 courses cp CAP 3 courses of cyclophosphamide 11 courses CAP 6 courses CP 8 courses CAP no no no

Hormone-receptor content of original tumor cell line

n.d.

Tumor,genesis in nude mice

Current passage

50

31 24 2 17 1

'0.negative ( < 10 fmolimg cytosol protein).-n.d., not done.

lines OV-MZ-1 and -5 only a few vimentin-expressing cells were detected. The antibody against G F P yielded negative results.

Heterotransplantation Four out of six cell lines (OV-MZ-1, -2, -4 and -6) produced slow-growing tumors after S.C. transplantation of 1 x lo7viable tumor cells in nude mice. The tumors of the cell lines OV-MZ-2 and -4 grew in all 3 animals inoculated, the cell lines OV-MZ-1 and -6 produced tumors in 2 / 3 animals. The tumors reached a diameter of 6 to 8 mm in 8 to 12 weeks and the latency period between transplantation and the first positive evidence of tumor growth was 2 to 3 weeks. OV-MZ-3 and -5 tumor cells did not give rise to transplant tumors within the observation period of 12 weeks. Histologically, the transplant tumors derived from OVMZ-1 and OV-MZ-2 were classified as undifferentiated carcinomas composed of solid masses of tumor cells. OV-MZ-4 tumor cells formed moderately differentiated tubulo-papillary adenocarcinomas, closely resembling the original tumor. OVMZ-6 produced poorly differentiated adenocarcinomas exhibiting irregular glandular spaces. Psammoma bodies which had been observed in some of the original tumors were never observed in transplant tumors. Determination of CA-125 and CEA Table I11 shows the expression of the tumor-associated antigen CA-125 in vivo and in vitro. In 3 patients, the CA-125 serum level was determined pre-operatively. One patient exhibited an elevated level with 326 Uiml, 2 patients with diagnostic second-look operations and clinically complete remission had normal CA-125 values. At the time of relapse, CA-125 was strongly elevated in all 3 patients. CA-125 secretion of proliferating cells in the supernatant varied from 4 to 95 U l m l i l x lo6 cells. For resting cells, the secretion varied between 0.5 to 1058 U / m l / l x lo6 cells. The level of CA-125 in culture medium alone was 3.5 U/ml. Two lines (OV-MZ-2 and -4) did not show any relevant secretion. These 2 lines were also negative with FACScan analysis, whereas for the CA-125-positive cell lines the percentage of positive cells varied from 32.6% to 77.7%. Interestingly, the original tumor of OV-MZ-2 yielded highly elevated levels of CA-125 in the patient's serum, whereas the cell line derived from this tumor failed to express CA-125. For the cell lines which were positive for the secretion of CA-125, we found a nearly linear correlation between cell number and CA-125 level in the supernatant. In contrast, the negative cell lines never showed any secretion at all despite increasing cell numbers. Figure 4a shows the combined histograms of the positive cell line OV-MZ-3, Figure 4b,c the histograms of the cell line OV-MZ-2, which is weakly positive

at low passage number and definitely negative at high passage number. Results of the CEA determination are not shown. CEA serum levels of all patients were always in the normal range, and thus CEA could not be detected in any cell line. DISCUSSION

Six human ovarian carcinoma cell lines have been newly established and investigated in terms of morphological and biological properties. The morphological and ultrastructural features both of the cell lines and of the corresponding transplant tumors in nude mice are consistent with their derivation from serous cystadenocarcinoma of the ovary. The ploidy status of ovarian carcinoma has been shown to be a very constant feature (Iversen and Skaarland, 1987) and an important factor in predicting prognosis. In contrast, S-phase fraction is often subject to considerable intra-tumoral variation and seems to have no relationship to prognosis (Brescia et al., 1990), with the exception of DNA-diploid tumors (Kallioniemi et al., 1988). The D N A indices of ovarian cancer cell lines have been reported to range from 1.1 to 2.9 (Giancotti et al., 1989; Hills et al., 1989). In our study, one cell line was tetraploid (OV-MZ-l), the other 5 lines being aneuploid. With a D N A index of 4.4, the ploidy level of the cell line OV-MZ-3 by far exceeded not only the ploidy level of all ovarian cancer cell lines described in the literature, but also the ploidy level of primary ovarian cancers (Brescia et al., 1990). Due to the presence of multiple D N A stem lines in most cell lines, cell-cycle analysis could be performed for only 2 cell lines, exhibiting a dominant S-phase (35%) for OV-MZ-1 and a lower S-phase (13%) for OVMZ-6. Scanning electron microscopy showed that some of the cell lines (OV-MZ-3 and -4) were composed of tightly juxtaposed epithelial-like cell clusters, whereas other cell lines (OV-MZ-5 and -6) exhibited a more fibroblast-like appearance. Ultrastructurally, the epithelial nature of the cell lines was confirmed by the presence of numerous desmosomes and occasional intracelM a r gland-like spaces. The inability to produce transplant tumors in nude mice, which became evident for the cell lines OV-MZ-3 and OV-MZ-5, does not argue against the carcinomatous derivation of these cell lines, since this phenomenon has been reported in other human ovarian tumor cell lines (Hill et al., 1987). Comparing the histomorphological features of the original tumors with those of the corresponding transplant tumors in nude mice, a loss of histomorphological differentiation became evident for the cell lines OV-MZ-1 and -2, which suggests a selection process during in vitro culture. The consistent expression of several CK polypeptides in all cell lines of this study provides further support for their

CHARACTERIZATION OF HUMAN OVARIAN CARCINOMA CELL LINES

81

RCURE 2 - Morphological aspects of human ovarian carcinoma cell lines by scanning electron microscopy: loosely juxtaposed polygonal cells in OV-MZ-I (a) and OV-MZ-2 (b):epithelial-like cell complexes in OV-MZ-3 (c) and OV-MZ-4 (d); fibroblast-like, spindle-shaped tumor cells in OV-MZ-5 (e) and OV-MZ-6 (f). Scale bar, 10 km.

epithelial nature and derivation. Moreover, the expression patterns of the individual IF proteins fit very nicely to their derivation from (non-mucinous) ovarian carcinomas. This becomes evident when the present results are compared with

the results of a recent analysis of IF protein expression of common epithelial tumors of the ovary (Moll et al., 1991). Both the in vivo tumors of that study and the cell lines presented here share the basic features of IF-protein expression, i.e., the

82

MOBUS ETAL.

TABLE I1 - EXPRESSION OF IF PROTEINS I N OVARIAN CANCER CELL LINES' Cell line

Stratified-epithelial cytokeratins CK4 CK5 CK17

CK7

Simple-epithelial cytokeratins CK8* CK18 CK19

Glial

Vimentin

CK20

filament protein

OV-MZ-1 ++ ++ +++ +++ +++ (+) +++ +++ +++ - +++ OV-MZ-2 (+) OV-MZ-3 +++ +++ +++ ++ (+) +++ OV-MZ-4 ++ + +++ +++ + + 3 ++3 OV-MZ-5 ++ ++ +++ +++ +++ + + +++ +++ ++ - +++ + (+) + OV-MZ-6 'As determined by immunocytochemistry and confirmed by 2-dimensional gel electrophoresis.2Determined by gel electrophoresis ~nly.-~Microcolonypattern.--, negative; (+), < 5% of cells positive; +, 5 to 20% of cells positive; + +, 21 to 80% of cells positive, + + +, > 80% of cells positive. ovarian carcinomas (Moll et al., 1991). This is a particularly good example of the frequently noted conservative nature of the cell-type-specific expression (or non-expression) of CK polypeptides, not only during malignant transformation and tumor progression, but also during the establishment of cell lines (Moll eta/., 1982). The high level of vimentin expression in 4/6 cell lines is well in line with the fact that serous ovarian carcinomas already express vimentin in vivo, a feature characteristic of only a limited spectrum of carcinoma types (Moll et al., 1991). Besides the afore-mentioned common features of CK expression, it has been observed that, with certain CK polypeptides, quantitative as well as qualitative differences exist among individual ovarian carcinoma cell lines. This indicates that there are differences in the degree of differentiation, which may reflect either differences among the original tumors o r FIGURE 3 - Two-dimensional gel electrophoresis of cytoskeletal changes that have occurred during the establishment and proteins of OV-MZ-6 cells (a) and OV-MZ-1 cells (b), using non-equilibrium pH gradient (NEPHG) electrophoresis in the maintenance of the cell lines. Nevertheless, our results indifirst dimension and SDS-PAGE in the second dimension. Gels cate that no fundamental changes of CK and IF-protein were stained with Coomassie brilliant blue (a) or a silver staining expression patterns have occurred during the process of method (b). The cytokeratins present are designated by numbers long-term in vitro culture. according to the catalogue of human cytokeratins (Moll et d., CA-125 is elevated in the serum of approximately 80% of 1982); V, vimentin; A, endogenous actin; P, 3-phospho-glyceroki- patients with epithelial ovarian cancer (Soper et a/., 1990). If nase from yeast added as marker polypeptide. Note the similarity elevated at the time of diagnosis, CA-125 serum levels correof the CK polypeptide patterns, while biochemically detectable late with the course of disease in 80 to 95% of patients. CA-125 amounts of vimentin are present only in OV-MZ-6 cells. is consistently the most important tumor marker in ovarian cancer, though only a few studies have been performed on the expression of tumor markers in ovarian cell lines (Buick et al., TABLE -THE TUMOR-ASSOCIATED ANTIGEN CA- 125 (Uiml) IN PATIENTS' SERUM AND IN CELL CULTURE 1985; Ishiwata and Ishiwata, 1985; Hills et al., 1989) and they have reported that 60 to 90% of these cell lines were positive CA-125 in the supernatant of Serum value FACScan for CA-125. Our results show that 4 out of 6 lines were weakly Cell lines pre. prolifer;ting rest,ng c e l l s ~ positive or even highly positive for CA-125. In 3 cases, these results relapse cells (day 3) cells ldav 71 were in accordance with the patient's serum level. In contrast, the cell line OV-MZ-2 failed to express CA-125 in vitro, 52.6 786 21.4 76 OV-MZ-1 20' although the serum level at the time of relapse was high (2686 0.5 0 2686 6 OV-MZ-2 n.d. U/ml). Interestingly, we found a low percentage of CA-12575.9 1150 95 1058 OV-MZ-3 2 9 positive cells in the low but not in the high passage number. 0 n.d. OV-MZ-4 n.d. 4 3 15 49 32.6 n.d. OV-MZ-5 326 This observation suggests clonal selection of CA-125-negative 77.7 n.d. 4.5 25 OV-MZ-6 n.d. tumor cells in long-term in vitro culture conditions and ex'Results expressed in Ulmlll x lohcells.-2Serum value before cludes the alternative possibility that the lack of expression of CA-125 originates from non-antigen-expressing tumor cells. A diagnostic second-look operation.-n.d., not done. corresponding loss of CA-125 expression on prolonged culturing of an ovarian cancer line was also observed by Van Niekerk clear-cut predominance of simple-epithelial CKs (mainly CKs, etal. (1988). It is important to note that we observed the loss of 8, 18, 19) over stratified-epithelial CKs, the frequent expres- CA-125 expression only in 1 out of 5 cell lines. sion of CKs 7 and 17, and the common co-expression of The CA-125-positive cell lines showed a nearly linear vimentin. This indicates that important components of the correlation between cell number and CA-125-antigen sheddifferentiation program of the original carcinomas have been ding into the SN, the average production per cell remaining maintained during the establishment of continuous cell lines. stable during exponential cell growth. This observation conOf particular interest is the consistent absence in all cell firms that the determination of the CA-125 serum level is lines of CK 20, a newly described simple-epithelial CK polypep- highly suited for monitoring the course of the disease and that tide (Moll et al., 1990). I n vivo this CK has been found in elevated serum levels may directly reflect the tumor mass. mucinous but not in serous, endometrioid and anaplastic Comparison of cell lines OV-MZ-3 and -6, which from FACS-

83

CHARACTERIZATION OF HUMAN OVARIAN CARCINOMA CELL LINES

(IOUpassage)

OV-Hi!-2

OV-HZ-3

OV-HZ-2 (high passage)

380

FL I

FL 1

Control Cft- 125

~

_-_I._._._.__.

Control CA-125

FL I

Events: 10000 Events: lo0013

FIGURE 4 - Combined histograms of the CA-125-positive cell line OV-MZ-3 (passage number 65) and the CA-125-negative cell line OV-MZ-2 at low ( 5 ) and high (178) passage number.

can analysis were both highly positive, demonstrated substantial differences in their secretion behavior. There was pronounced antigen shedding of the cell line OV-MZ-3, whereas the cell line OV-MZ-6 showed almost no shedding. Thus, a negative CA-125 serum level in patients with a large tumor burden is explainable either by a CA-125-negative tumor or by lack of secretion in CA-125-positive tumors. CEA shows elevated serum levels in only 30 to 40% of ovarian cancer, especially in poorly or mucinous differentiated carcinomas and advanced disease. Hills et al. (1989) and Horowitz et al. (1985) reported CEA-positive cell lines derived from ovarian serous cystadenocarcinoma. In contrast, all our cell lines were negative for CEA in accordance with the patients’ negative serum level. Furthermore, all 6 lines were negative for the expression of estrogen and progesterone receptors. As far as could be determined, this was in accordance with the original tumor. There have, however, been previous reports on the presence of steroid hormone receptors in ovarian cancer cell lines (Wolf et al., 1987: Crickard et al., 1989) as well as reports of responses to estrogen and antiestrogen treatment in cell lines (Nash et af., 1989).

Despite a loss of differentiation in 2 xenotransplanted tumors, the preservation of IF patterns and the expression of tumor-associated CA-125 indicate that our cell lines retained distinctive features of the original ovarian tumors. These facts contradict the assumption that permanent tumor cell lines will inevitably de-differentiate during long-term culture. The lines therefore will be highly suitable in screening assays of novel platinum-containing chemotherapeutic agents or new cytostatic metallocene complexes such as titanocene dichloride or vanadocene dichloride. Furthermore, the role of membrane glycoproteins and the cellular level of glutathione can be studied in sensitive and in drug-resistant cell lines. Four of the six lines have been xenografted in the nude mouse, thus providing the possibility of comparative in vitro and in vivo experiments. ACKNOWLEDGEMENTS We thank Ms. A. Knorr and Ms. D. Thomas for excellent technical assistance and Dr. G. Hoffmann for his cooperation in providing biopsy material. We are grateful to Ms. B. Kemmeter for careful typing of the manuscript.

REFERENCES

ACHTSTAE~TER, T., HATZFELD, M., QUINLAN, R.A., PARMELEE, D. and FRANKE, W.W., Separation of cytokeratin polypeptides by gel electrophoresis and chromatographic techniques and their identification by immunoblotting.Meth. EnzymoL, 134,355-371 (1986). BRADFORD, M.M., A rapid and sensitive method for the quantification of microgram quantities of protein utilizing the principle of proteindye hinding.Anal. Bioclietn., 72,248-254 (1976). BRESCIA.R.J.. BARAKAT,R.A., BELLER,U., FREDERICKSON, G., SUHRLAND. M.J.. DUBIN, N. and DEMOPOULOS, R., The prognostic significance of nuclear DNA content in malignant epithelial tumors of the ovary. Cancer, 65,141-147 (1990). BUICK. R.N., POLLAND, R. and TRENT, J.M., Comparative properties of five h u m a n ovarian adenocarcinoma cell lines. Cancer Rex, 45,

specific for glial fibrillary acidic (GFA) protein and for each of the neurofilament triplet polypeptides. Diflerenriation, 25, 193-203 (1983). FRANKE, W.W. and MOLL,R., Cytoskeletal components of lymphoid organs. Differentiation, 36, 145-163 (1987). GIANCOTI, F.R., DORSETT, B.H., WEAVER, S.C.. BHARATHUR, R., IOACHIM, H.L. and BARBER,H.R.K., Description of an endometrioid ovarian cancer cell line. Gynecol. Oncol., 35,330-337 (1989). GRILL, H.-J., MANZ.B., BELOVSKY, 0. and POLLOW, K., Criteria for the establishment of a double-labelingassay for simultaneousdetermination of estrogen and progesterone receptors. Oncobgy, 41, 25-32 (1984).

D., HARKING, L.K., HILL,B.T.. WHELAN,R.D.H.. GIBBY. E.M., SHEER, 3668-3676 (1985). SHELLARD, S.A. and RUPNIAK, H.T., Establishment and characterizaCRICKARD, K., NIEDBALA.M.J., CRICKARD, U..YOONESS, M., SAND- tion of three new human ovarian carcinoma cell lines and initial evaluation of their potential in experimental chemotherapy studies. BERG, A.A., OKUYAMA, K., BERNACKI, R.J. and SATCHIDANAND, S.K., Characterization of human ovarian and endometrial carcinoma cell Int. J. Cancer, 39,219-225 (1987). lines established on extracellular matrix. Gynecol. Oncol., 32, 163-173 HILLS,C.A., KELLAND, L.R., ABEL.G.. SIRACKY, J., WILSON, A.P. and ( 1989). HARRAP, K.R., Biological properties of ten human ovarian carcinoma DEAN,P.N. and JET, J.H., Mathematical analysis of DNA distribu- cell lines: calibration in vifro against four platinum complexes. Brit. J. tions derived from flow microfluorometry. J. Cell BioL, 60, 523-527 Cancer, 59,527-534 (1989). (1 974). HOROWITZ, A.T., TREVES,A.J., VOSS, R., OKON,E., FUKS,Z., DEBLIS, E., WEBER,K. and OSBORN, M., Monoclonal antibodies DAVIDSON, L. and BIRAN,S., A new human ovarian carcinoma cell

84

MOBUSET AL.

line: establishment and analysis of tumor-associated markers. Oncol42,332-337 (1985). ISHIWATA, J. and ISHIWATA, C . ,CA-125 levels in the conditioned media of the gynecological tumor cell lines. Acra ohst. gyriaec. Jpn., 37, 147-148 (1985). IVERSEN. 0.-E. and SKAARLAND, E., Ploidy assessment of benign and malignant ovarian tumors by flow cytometry-a clinicopathologic study. Cancer, 60,82-87 (1987). KALLIONIEMI, O.P., PUNNONEN R., MATTILA, J., LEHTINEN, M. and KOIVULA, T., Prognostic significance of DNA index, multiploidy and S-phase fraction in ovarian cancer. Cancer, 61,334-339 (1988). LYNCH, M.H., O’GLIIN. W.M., HARDY,C., MAK,L. and SUN,T.-T., Acidic and basic hair/nail (“hard”) keratins: their co-localization in upper cortical and cuticle cells of human hair follicle and their relationship to “soft” keratins.J. CellBiol., 103,2593-2606 (1986). MOLL,R., ACHTSTAETTER, T., BECHT.E., BALCAROVA-STANDER, J., ITTENSOHN, M. and FRANE, W.W., Cytokeratins in normal and malignant transitional epithelium: maintenance of expression of urothelial differentiation features in transitional cell carcinomas and bladder carcinoma cell lines.Am. J. Pathol., 132, 123-144 (1988). MOLL,R., FRANKE, W.W., SCHILLER, D.L., GEIGER, B. and KREPLER, R., The catalog of human cytokeratins: patterns of expression of specific cytokeratins in normal epithelia. tumors and cultured cells. Cell, 31, 11-24 (1982). MOLL, P., PITZ, S., LEVY, R., WEIREL,W., FRANKE.W.W. and CZERNOBILSKY, B., Complexity of expression of intermediate filament proteins, including glial filament protein, in endometrial and ovarian adenocarcinomas. Hum. Pathol., 22,989-1001 (1991). MOLL, R., ScHiLLER, D.L. and FRANKE.W.W., Identification of protein I1 of the intestinal cytoskeleton as a novel type-I cytokeratin

OD,

with unusual properties and expression patterns. J. Cell Biol., 111, 567-580 (1990). NASH,J.D., OZOLS, R.F., SMYTH, J.F. and HAMILTON. T.C.. Estrogen and anti-estrogen effects on the growth of human epithelial ovarian cancer in vitro. Ohstet. Gynecol., 73, 1009-1016 (1989). NIEKERK, C.C. VAN, POELS,L.G., JAP, P.H.K., SMEETS,D.F.C.M., THOMAS, C.M.G., RAMAEKERS, F.C.S. and VOOIJS, G.P., Characterization of a human ovarian carcinoma cell line, OTN 14, derived from a mucinous cystoadenocarcinoma. Int. J. Cancer, 42, 104-111 (1988). PITZ, S., MOLL.R., STORKEL,S. and THOENES, W., Expression of intermediate filament proteins in sub-types of renal-cell carcinomas and in renal oncocytomas. Distinction of two classes of renal-cell tumors. Lab. Invest.. 56,642-653 (1987). SCATCHARD, D.G., The attraction of proteins for small molecules and ions.Ann. N . Y A c a d . Sci.. 51, 660-672 (1949). SCHWARTZ, P.E., GATING. G., MACLUSKY, N., NAFTOLIN, F. and EISENFELD, A,, Tamoxifen therapy for advanced ovarian cancer. Ohstet. Gynecol., 59,583-588 (1982). SOPER,J.T., HUNTER, V.J.. DALY,L.. TANNER, M., CREASMAN, W.T. and BAST,R.C., Pre-operative serum tumor-associated antigen levels in women with pelvic masses. Obstet. Gynecol., 75,249-254 (1990). TROYANOVSKY, S.M., GUELSTEIN,V.I., TCHIPYSHEVA, T.A., KRUTOVSKIKH, V.A. and BANNIKOV, G.A.. Patterns of expression of keratin 17 in human epithelia: dependency on cell position. J. Cell Sci., 93,419-426 (1989). WOLF, C.W., HAYWARD, J.P., LAWRIE, S.S.. BUCKTON, K., MCINTYRE, M.A., ADAMS,D.J., LEWIS,A.D., SCOTT,A.R.R. and SMYTH,J.F., Cellular heterogeneity and drug resistance in two ovarian andenocarcinoma cell lines derived from a single patient. Inf.J. Cancer, 39,695-702 (1987).