Clin. exp. Immunol. (1991) 83, 92-95
ADONIS
000991049100017K
14C1, an antigen associated with human ovarian cancer, defined using a human IgG monoclonal antibody G. GALLAGHER, F. AL-AZZAWI*, L. P. WALSH & G. WILSON Immunology Research Group, University of Strathclyde, Glasgow, Scotland, and * Department of Obstetrics and Gynaecology, University of Cambridge, The Rosie Maternity Hospital, Cambridge, England
(Acceptedfor publication 10 August 1990)
SUMMARY We have selected a human EBV-transformed cell line from the involved lymph nodes of an ovarian cancer patient which secretes an IgG K antibody, able to recognize an antigen present on the surface of ovarian cancer cells. The antigen, termed 'I 4C I,' has previously been shown by immunohistological techniques to be present on the surface of the malignant cells within tumour specimens. Western blotting analysis has shown that the majority of primary ovarian cancer specimens and three continuous cell lines derived therefrom express 14C 1; other tissue types were negative. Preliminary biochemical characterization has been carried out, which shows that the 14C 1 antigen has a molecular weight range of 25-32 kD and an isoelectric point from pI 6 3 to 6-8. We believe that the 14CI antigen is immunologically relevant to ovarian cancer patients and may therefore represent a novel target for both active and passive immunotherapy. Keywords ovarian
cancer
14C1 human monoclonal antibodies immunotherapy
INTRODUCTION Epithelial cancer of the ovary is responsible for about 5000 deaths per year in the United Kingdom. While there are no generally available antigens or antibodies to aid the diagnosis or management of this tumour type, in recent years several mouse monoclonal antibodies have been described and developed as potential diagnostic agents and/or therapeutic tools in the monitoring and treatment of ovarian cancer (Bast et al., 1984; Umbach, 1984). One in particular, CA-125, is used extensively in the routine post-surgical monitoring of patients with ovarian cancer (Sekine et al., 1985) and another, HMFG-2, has been used for the treatment and imaging of i.p. residual disease following surgery (Ward et al., 1988). In addition, several other antibodies have been described which usually (Sakakibara et al., 1988), but not always (Mattes et al., 1987) define high molecular weight mucinous or carbohydrate antigens. Antibodies raised in mice are undoubtedly of great clinical potential, but they describe moieties that are antigenic in mice and so may not reflect the antigenicity of a particular tumour type in the human host. Working from the premise that antibody bound to tumour in situ will be specifically directed against that tumour, the production of human anti-cancer
antibodies from Epstein-Barr virus (EBV) transformed human lymphocytes has been investigated. A small number of such antibodies has been reported, particularly for breast and colonic tumours and melanoma (Haspel et al., 1985; Kan-Mitchell et al., 1986; Kjeldsen et al., 1988), with only one such report for ovarian cancer (Al-Azzawi, Govan & Stimson, 1987), which demonstrated the binding of the human antibody to the surface of cells in sections of primary malignant tissue. Here we describe the use of this antibody to define an antigen, which we term 14CI. This antigen is present on > 70% of ovarian epithelial tumours, but absent from a range of other tumour and normal material. We consider this antigen to be particularly important because it has been defined by a human antibody derived from the B cells in an involved lymph node, strongly suggesting that this antigen was the object of an antitumour immunological response at the time of resection. Such an antigen may represent a potent target for both active and passive immunotherapy in the post-surgical treatment of ovarian cancer.
MATERIALS AND METHODS Preparation of the anti-14CJ secreting cell line The cell line secreting anti-14CI antibody was prepared and selected as previously described (Al-Azzawi et al., 1987). Briefly, lymph nodes were obtained from a patient undergoing cytoreductive surgery for ovarian epithelial cancer at Stobhill General Hospital, Glasgow. Following dissociation of the nodes
Correspondence: G. Gallagher, University of Glasgow Department of Surgery, Queen Elizabeth Building, Glasgow Royal Infirmary, Glasgow G31 2ER, UK. F.A.A. present address: Department of Obstetrics and Gynaecology, Leicester Royal Infirmary, Leicester, UK.
92
14CJ and ovarian cancer in vitro, T cells were removed by two rounds of rosetting with AET-treated sheep erythrocytes and the resulting cells were exposed to live EBV (as the supernatant of the marmoset cell line, B95/8) for 2 h at 37°C. Following thorough washing, the cells were seeded at a density of 5 x 105/ml (200 ,l cultures) in 96well, flat-bottomed plates (Costar, Cambridge, MA). Wells were selected for the presence of antibody able to bind to the human ovarian cancer cell line OwMm I (a generous gift from Professor C. N. Hudson, St Bartholomew's Hospital, London, and made available to us by Professor W. H. Stimson, IRG, University of Strathclyde). Stable secretors were selected by multiple, repeated selection on this cell line. Subsequent analysis has shown that the 14CI antigen is expressed in cell lines derived in our laboratory from primary ovarian tumour tissue (below). The anti-14CI cell line secretes an IgGI with a single light chain (kappa), which we take as strong evidence of monoclonality (Gallagher et al., 1987). The line was maintained in Ham's F1O culture medium, containing 10% (v/v) fetal calf serum and 2 mm glutamine (complete medium; all media components were obtained from Biological Industries). For Western blotting analysis, supernatant from this line was diluted to 400 ng/ml IgG (as determined by ELISA) and used directly. Control experiments carried out with irrelevant human IgGIK (from a patient with myeloma) failed to detect the 14C1 antigen under identical analytical conditions.
93
Table 1. Physical characteristics of the 14C1 antigen on human ovarian cancer cell lines OwMml, Ovan-4 and Ovan-5 Molecular weight Isoelectric point
OwMml Ovan-4
Ovan-5
(reduced) (kD)
(non-reduced)
32 32 26-32
ND 6 3-6-8 63-6-8
ND not determined.
washing in PBS containing 0.5% (v/v)/Tween-20 (PBS-T; Sigma), alkaline-phosphatase-conjugated, goat anti-human IgG (gamma-chain specific; Sigma) was diluted 1/500 in PBS-T and added for 1 h at room temperature, after which time the blot was again extensively washed. The washed blots were equilibrated in substrate buffer (100 mm Tris, 25 mm diethanolamine, 100 mM NaCl, 2 mM MgCl2; pH 9 55) and the bands were visualized by exposure to a substrate solution comprising 0 33 mg/ml nitrobluetetrazolium in substrate buffer, to each ml of which were added 6-7 p1 of 2 mg/ml phenazine methosulphate in water and 3-4 p1 of 40 mg/ml 5-bromo-4-chloro-3-indolyl phosphate (p-toluidine salt) in dimethylformamide, according to the method of Ey & Ashman (1976).
Preparation of tumour samples for electrophoretic analysis After histological confirmation of malignancy, a sample of nonnecrotic tissue was chopped finely and pressed through a nylon mesh to release tumour cells. These cells were lysed in a lysis buffer, comprising 0 5% (w/v) Nonidet P-40 and 0-1 mm PMSF in phosphate-buffered saline, pH 7-4 (PBS; all reagents from Sigma), at 4°C for 30 min. Cell membranes were prepared by centrifugation of the lysis mixture (400g, 10 min) and subjecting the supernatant to further centrifugation (1 5000 g, 45 min, 4°C). The supernatants from this procedure were designated 'membranes' and adjusted to 2 mg/ml protein (Bradford's method) before use in Western blotting experiments. Cell line membranes were prepared similarly. Serum and ascitic fluid samples were also adjusted to 2 mg/ml protein prior to electrophoresis.
Preliminary characterization of the 14CJ antigen Membrane preparations from three 14C 1-positive cell lines (OwMml, Ovan-4 and Ovan-5; unpublished data) were analysed by Western blotting, following SDS-PAGE or IEF separations. The results are summarized in Table 1. The 14CI antigen appeared as a double band at 32 kD in the Ovan-4 and OwMm 1 samples, and as a less-well-defined single band at 25-32 kD on Ovan-5. Both Ovan-4 and Ovan-5 gave a double band at pI 6-3-6-8 following IEF (OwMml not tested).
Western blotting Membrane and other preparations were subjected to SDSPAGE under reducing conditions, using an LKB Midget gel apparatus and 10% resolving gels of 0-75 mm thickness. Membrane samples were diluted to 0-5 mg/ml protein in sample buffer (62-5 mm Tris, 2% (w/v) SDS, 10% (v/v) glycerol, 4% (v/v) 2-mercaptoethanol, and 0 1 mg/ml bromophenol blue) and boiled for 5 min before loading 5-10 pg protein to each track. Alternatively, membrane proteins were separated by isoelectric focusing (IEF), using pre-cast, 1 0-mm thick gels (Pharmacia) in which a pH gradient of 3-5-9 5 was established. Following separation, proteins weere electrophoretically transferred to nitrocellulose membrane ('Hybond-C', Amersham International) using a BioRad Transblot apparatus, under 'wet' conditions in a transfer buffer comprising 20 mm Tris, 192 mM glycine, 1% (w/v) SDS and 25% (v/v) methanol. Transfer was conducted for at least 2 h at 100 V. Blots were 'blocked' by incubating in complete medium for 1 h at room temperature, then stained with 400 ng/ml anti- 14C1 antibody in complete medium for 90 min. Following thorough
Expression of the 14CJ antigen on primary human tissue We next examined the presence of the 14C1 antigen on a range of malignant and normal tissue, with particular reference to ovarian epithelial malignancies. The results are shown in Table 2. A clear, positive correlation between the presence of the 1 4C I antigen and diagnosis ofovarian cancer is shown, particularly in the cases of mucinous, serous and clear cell carcinomas; however, only one out of four specimens of poorly differentiated adenocarcinoma of the ovary was positive. One specimen of histologically benign mucinous cyst was also positive. The analysis of membranes from other tumour types and normal tissue showed a uniform lack of 14CI expression, with the exception of one specimen (of four) of colonic carcinoma, which was positive. Of particular interest in the negative group were the placentae, the benign ovarian serous and papilomatous cysts and the normal ovarian tissue. Sera from ovarian cancer patients, including those whose tumour tissue was positive, uniformly failed to contain the 14C I antigen (27 samples), and of four (malignant) ascitic fluids tested, only one contained the antigen.
RESULTS
G. Gallagher et al.
94
I
The 14C1 antigen presents as a 25-kD in primary tumour material Figure I shows a typical Western blot analysis of primary ovarian cancer material (mucinous cystadenocarcinoma), the antigen is detected as a single band (track c), at or around 25 kD; as previously stated, this band is not observed if irrelevant (myeloma-derived) IgGlK is used in place of the anti-14CI cell line for blotting. Note that another band is detected in all samples at 48 kD, which we have proved to be due to the detection of IgG heavy chain in the sample directly by the conjugate (however, the conjugate does not detect either human light chain).
I
DISCUSSION
Table 2. Expression of 14CI on primary humaIn tumours (Western blotting)
Tumour type Serous carcinoma (ovary) Mucinous carcinoma (ovary) Clear cell carcinoma (ovary) Poorly differentiated adenocarcinoma (ovary) Malignant brenner (ovary)
Benign serous cyst (ovary) Benign papilomatous cyst (ovary) Benign mucinous cyst (ovary) Squamous cell carcinoma (cervix) Squamous cell carcinoma (vulva) Endometrial adenocarcinoma Colon carcinoma Stomach and oesophagus
Positive samples
Number tested
7 6 3 1
9 6 3 4
I
2 1 2 1
4
4
Placenta (term) Vagina (pre-pubescent) Vagina (post-menopausal) Ovary (pre-pubescent) Ovary (active) Ovary (post-menopausal)
2 2 I I
3
Serum (ovarian cancer: benign and malignant) Ascitic fluid (ovarian cancer: malignant)
27 1
4
kD
80 -; .:.,.:
48
26b. ...
..
'.
.
a b c Fig. 1. The 14C I antigen is present as a single band ofF25 kD on primary ovarian cancer tissue. Membrane preparations were clerived from tissue samples, and subjected to Western blotting as de: scribed. Lanes: a, vagina; b, endometrial adenocarcinoma; c, mucinoi us cystadenocarcinoma of the ovary.
This work is based on the premise that an enlarged lymph node represents an immunologically stimulated centre and therefore will contain activated, antigen-specific B cells. EBV transformation of such B cells from cancer patients should allow the examination of tumour immunogenicity, as perceived by the human host (Cote et al., 1986). We have used EBV transformation to prepare a human IgG-secreting cell line from the involved lymph nodes of a patient with ovarian epithelial cancer (Al-Azzawi et al., 1987) and demonstrated that it secretes an antibody able to recognize an antigen on this tumour with a high degree of selectivity. We have termed this antigen 14C1. The 14CI antigen has a molecular weight of 25-32 kD on primary tissue and established ovarian cancer cell lines and an isoelectric point of pI 6 3-6-8. Following SDS-PAGE, the 14C I antigen appears as a single band. The low molecular weight of the 14CI antigen sets it apart from ovarian-cancer-related antigens described previously, with the possible exception of that defined by the antibody MH99 (Mattes et al., 1987), which is a glycoprotein epithelial differentiation antigen, with subunits at 38 and 29 kD. This is within the scope of experimental variation between laboratories; the pI of the MH99 antigen was not reported. The expression of 14C1 seemed to be highly restricted to ovarian epithelial tumours in the limited series described here. Of 24 total ovarian tumours, 16 were positive (670%). Among the eight negative ovarian specimens, two were of the Brenner type (possible transitional epithelial cell origin) and three were of the poorly differentiated adenocarcinoma variety. These data suggest a clearly restricted distribution of the antigen to particular types of ovarian malignancy. Of the other tumour types tested, comprising 13 specimens taken from the colon, stomach, oesophagus, cervix, vulva and endometrium, only one (colonic) was positive (80%). None of the normal tissue specimens was positive, including term placentae, normal ovarian tissue and two out of three benign ovarian cysts. In addition, cell membranes from a range of human tumour cell lines, including Calu-6 (lung), HT-29 and Colo-320 (colorectal), T-24 (bladder), U937 (myeloid) and a human thymic epithelial line of our own derivation, also failed to express 14C1 (data not shown). The presence of the 1 4C I antigen on the specimen of benign mucinous cyst suggests one of two possible scenarios. Firstly, that the patient is borderline malignant and the presence of the 14C1 antigen has a diagnostic significance. Only time will tell, but present indications are that there are no signs of recurrence. Secondly, that the 14C1 antigen is present in mucinous cystic tissue and continues to be expressed if this tissue becomes
14CJ and ovarian cancer malignant. The expression of the tissue on malignant but not benign serous cystic tissue may be due to de novo expression or malignancy-related revelation of the antigen. This question, and that of change in expression on malignant over benign mucinous cysts will be answered when a cDNA probe for 14CI is prepared. While it is disappointing that the antigen is not truly 'tumour specific,' it is probably not realistic to expect that anything other than a virally induced tumour will express such a marker. The 14CI antigen is therefore 'tumour associated' and as such probably represents a cell-cycle protein whose overexpression is in some way associated with the development of the malignant state. In several cases, murine antibodies such as HMFG2 have shown good clinical potential in ovarian cancer, particularly for the treatment of recurrent ascites rather than solid metastatic nodules (Ward & Wallace, 1987). However, murine antibodies are known to solicit an 'anti-mouse' response from the human immune system (Pimm et al., 1985; Reynolds et al., 1986) and this could seriously diminish their clinical usefulness. Although such materials can be 'humanized,' the most satisfactory procedure would be to use a wholly human antibody. Such antibodies would offer the benefit of safe repeated administration, provided one of the appropriate specificity could be found. Much work has gone into the search for a human anti-tumour antibody, with some success (e.g. Kan-Mitchell et al., 1986) and some IgM materials have been considered for clinical use in a range of tumour types (Sikora et al., 1985) but only one is currently available for ovarian cancer (Al-Azzawi et al., 1987). The clinical potential of the anti- 14CI cell line as an imaging and/or targetting agent is strongly promoted by its restricted binding to serous and mucinous cystadenocarcinomas, with a less-well-defined affinity for clear-cell carcinoma (one specimen tested) and a lower percentage binding to poorly differentiated adenocarcinoma (one out of four). Our previous (Al-Azzawi et al., 1987) and more recent (Burch et al., 1990) histological studies clearly demonstrate that the antigen is present on the surface of malignant cells and that it can discriminate between tumour cells and surrounding stromal tissue; the present study uses immunochemical methods to demonstrate the presence of the antigen in membrane preparations. Both these observations enhance the potential use of the antibody as a therapeutic agent. Apart from the direct clinical relevance, it is perhaps more interesting to consider that the benefit of a human antibody may lie in the fact that it defines an antigen of immunological relevance to the patient, where an existing immune response has been demonstrated. 14C1 is such an antigen and may represent an important target for active immunotherapies, as part of the post-surgical management of ovarian cancer.
ACKNOWLEDGMENTS We would like to thank Ms J. Mitchell and Ms S. Webb for excellent research assistance. The work was supported by the Cancer Research Campaign (G.G., L.P.W.), the University of Strathclyde Research and Development Fund (G.W.) and the East Anglia Regional Health Authority (F.A.-A.) The anti-14Cl antibody described in this report is the property of the University of Strathclyde.
REFERENCES AL-AZZAWI, F., GOVAN, A.D. & STIMSON, W.H. (1987) Human antibodies to ovarian cancer antigens secreted by lymphoblastoid cell lines. J. Clin. Lab. Immunol. 22, 71-75.
95
BAST, R., KLUG, T., SCHAETZEL, E., LAVIN, P., NILOFF, J., GREBER, T., ZURAWSKI, V. & KNAPP, R. (1984) Monitoring human ovarian carcinoma with a combination of CA- 125, CA 19-9 and carcinoembryonic antigen. Am. J. Obstet. Gynaecol. 149, 553. BURCH, D., AL-AZZAWI, F., WALSH, L.P., WILSON, G. & GALLAGHER, G. (1990) Human antibodies and host recognition of ovarian carcinoma antigens. In Advances in the Application of Monoclonal Antibodies to Clinical Oncology (ed. by A. A. Epenitos) Chapman & Hall, (In press).
COTE, R.J., MORRISSEY, D.M., HOUGHTON, A.N., THOMSON, T.M., DALY, M.E., OETTGEN, H.F. & OLD, L.J. (1986) Specificity analysis of human monoclonal antibodies reactive with cell surface and intracellular antigens. Proc. natl Acad. Sci. USA, 83, 2959. EY, P.L. & ASHMAN, L.K. (1986) The use of alkaline phosphatase conjugated anti-immunoglobulin with immunoblots for determining the specificity of monoclonal antibodies to protein mixtures. Methods Enzymol. 121, 497. GALLAGHER, G., CHRISTIE, J.F., STIMSON, W.H., Guy, K. & DEWAR, A.E. (1987) T-24 B-cell differentiation factor induces immunoglobulin secretion in human B-cells without prior cell replication. Immunology, 60, 523. HASPEL, M.V., MCCABE, R.P., POMATO, N., JANESCH, N.J., KNOWLTON, J.V., PETERS, L.C., HOOVER, H.C. JR & HANNA, M.G. JR (1985) Generation of tumour-cell reactive human monoclonal antibodies using peripheral blood lymphocytes from actively immunised colorectal carcinoma patients. Cancer Res. 45, 395 1. KAN-MITCHELL, J., IMAM, A., KEMP, R.A., TAYLOR, C.R. & MITCHELL, M.S. (1986) Human monoclonal antibodies directed against melanoma associated tumour antigens. Cancer Res. 46, 2490. KJELDSEN, T.B., RASMUSSEN, B.B., ROSE, C. & ZEUTHEN, J. (1988) Human-human hybridomas and human monoclonal antibodies obtained by fusion of lymph node lymphocytes from breast cancer patients. Cancer Res. 48, 3208. MATTES, M.J., LOOK, K., FURUKAWA, K., PIERCE, V.K., OLD, L.J., LEWIS, J.L. JR & LLOYD, K.O. (1987) Mouse monoclonal antibodies to human epithelial differentiation antigens expressed on the surface of ovarian carcinoma ascites cells. Cancer Res. 47, 6741. PIMM, H.V., PERKINS, A.C., ARMITAGE, N.C. & BALDWIN, R.W. (1985) The characteristics of blood-borne radiolabels and the effect of antimouse IgG antibodies on localisation of radiolabelled monoclonal antibody in cancer patients. J. nucl. Med. 26, 101 1. REYNOLDS, J.C., CARRASQUILLO, J.A., KENAN, A.M., LORA, M.E., SUGARBAKER, P., ABRAMS, P., FOON, K., MULSHINE, J.L., ROTH, J., COLCHER, D., SCHLOM, J. & LARSON, S.M. (1986) Human anti-murine antibodies following immunoscintigraphy or therapy with radiolabelled monoclonal antibodies. J. nucl. Med. 27, 1022. SAKAKIBARA, K., UEDA, R., OHTA, M., NAKASHIMA, N., TOMODA, Y. & TAKAHASHI, T. (1988) Three novel mouse monoclonal antibodies, OM-A, OM-B and OM-C, reactive with mucinous type ovarian tumours. Cancer Res. 48, 4639. SEKINE, H., HAYES, D.F., OHNO, T., KEEFE, K.A., SCHAETZEL, E., BAST, R.C., KNAPP, R.C. & KUFE, D.W. (1 985) Circulating DF3 and CA 125 antigen levels in serum from patients with epithelial ovarian carcinoma. J. clin. Oncol. 3, 1355. SIKORA, C., ALDERSON, T., NETHERSELL, A. & SMEDLEY, H. (1985) Tumour localisation by human monoclonal antibodies. Oncol. Tumour Pharmacother. 2, 77. UMBACH, G. (1984) Review of tumour markers for ovarian cancer. Med. Hypotheses, 13, 329. WARD, B.G. & WALLACE, K. (1987) Localisation of the monoclonal antibody HMFG2 after intraperitoneal and intravenous injection into nude mice bearing subcutaneous and intraperitoneal human ovarian cancer xenografts. Cancer Res. 47, 4714. WARD, B.G., MATHER, S., SHEPHERD, J., CROWTHER, M., HAWKINS, L., BRITON, K. & SLEVIN, M.L. (1988) The treatment of intraperitoneal malignant disease with monoclonal antibody guided I-131 radiotherapy. Br. J. Cancer, 58, 658.
ADONIS
000991049100017K
14C1, an antigen associated with human ovarian cancer, defined using a human IgG monoclonal antibody G. GALLAGHER, F. AL-AZZAWI*, L. P. WALSH & G. WILSON Immunology Research Group, University of Strathclyde, Glasgow, Scotland, and * Department of Obstetrics and Gynaecology, University of Cambridge, The Rosie Maternity Hospital, Cambridge, England
(Acceptedfor publication 10 August 1990)
SUMMARY We have selected a human EBV-transformed cell line from the involved lymph nodes of an ovarian cancer patient which secretes an IgG K antibody, able to recognize an antigen present on the surface of ovarian cancer cells. The antigen, termed 'I 4C I,' has previously been shown by immunohistological techniques to be present on the surface of the malignant cells within tumour specimens. Western blotting analysis has shown that the majority of primary ovarian cancer specimens and three continuous cell lines derived therefrom express 14C 1; other tissue types were negative. Preliminary biochemical characterization has been carried out, which shows that the 14C 1 antigen has a molecular weight range of 25-32 kD and an isoelectric point from pI 6 3 to 6-8. We believe that the 14CI antigen is immunologically relevant to ovarian cancer patients and may therefore represent a novel target for both active and passive immunotherapy. Keywords ovarian
cancer
14C1 human monoclonal antibodies immunotherapy
INTRODUCTION Epithelial cancer of the ovary is responsible for about 5000 deaths per year in the United Kingdom. While there are no generally available antigens or antibodies to aid the diagnosis or management of this tumour type, in recent years several mouse monoclonal antibodies have been described and developed as potential diagnostic agents and/or therapeutic tools in the monitoring and treatment of ovarian cancer (Bast et al., 1984; Umbach, 1984). One in particular, CA-125, is used extensively in the routine post-surgical monitoring of patients with ovarian cancer (Sekine et al., 1985) and another, HMFG-2, has been used for the treatment and imaging of i.p. residual disease following surgery (Ward et al., 1988). In addition, several other antibodies have been described which usually (Sakakibara et al., 1988), but not always (Mattes et al., 1987) define high molecular weight mucinous or carbohydrate antigens. Antibodies raised in mice are undoubtedly of great clinical potential, but they describe moieties that are antigenic in mice and so may not reflect the antigenicity of a particular tumour type in the human host. Working from the premise that antibody bound to tumour in situ will be specifically directed against that tumour, the production of human anti-cancer
antibodies from Epstein-Barr virus (EBV) transformed human lymphocytes has been investigated. A small number of such antibodies has been reported, particularly for breast and colonic tumours and melanoma (Haspel et al., 1985; Kan-Mitchell et al., 1986; Kjeldsen et al., 1988), with only one such report for ovarian cancer (Al-Azzawi, Govan & Stimson, 1987), which demonstrated the binding of the human antibody to the surface of cells in sections of primary malignant tissue. Here we describe the use of this antibody to define an antigen, which we term 14CI. This antigen is present on > 70% of ovarian epithelial tumours, but absent from a range of other tumour and normal material. We consider this antigen to be particularly important because it has been defined by a human antibody derived from the B cells in an involved lymph node, strongly suggesting that this antigen was the object of an antitumour immunological response at the time of resection. Such an antigen may represent a potent target for both active and passive immunotherapy in the post-surgical treatment of ovarian cancer.
MATERIALS AND METHODS Preparation of the anti-14CJ secreting cell line The cell line secreting anti-14CI antibody was prepared and selected as previously described (Al-Azzawi et al., 1987). Briefly, lymph nodes were obtained from a patient undergoing cytoreductive surgery for ovarian epithelial cancer at Stobhill General Hospital, Glasgow. Following dissociation of the nodes
Correspondence: G. Gallagher, University of Glasgow Department of Surgery, Queen Elizabeth Building, Glasgow Royal Infirmary, Glasgow G31 2ER, UK. F.A.A. present address: Department of Obstetrics and Gynaecology, Leicester Royal Infirmary, Leicester, UK.
92
14CJ and ovarian cancer in vitro, T cells were removed by two rounds of rosetting with AET-treated sheep erythrocytes and the resulting cells were exposed to live EBV (as the supernatant of the marmoset cell line, B95/8) for 2 h at 37°C. Following thorough washing, the cells were seeded at a density of 5 x 105/ml (200 ,l cultures) in 96well, flat-bottomed plates (Costar, Cambridge, MA). Wells were selected for the presence of antibody able to bind to the human ovarian cancer cell line OwMm I (a generous gift from Professor C. N. Hudson, St Bartholomew's Hospital, London, and made available to us by Professor W. H. Stimson, IRG, University of Strathclyde). Stable secretors were selected by multiple, repeated selection on this cell line. Subsequent analysis has shown that the 14CI antigen is expressed in cell lines derived in our laboratory from primary ovarian tumour tissue (below). The anti-14CI cell line secretes an IgGI with a single light chain (kappa), which we take as strong evidence of monoclonality (Gallagher et al., 1987). The line was maintained in Ham's F1O culture medium, containing 10% (v/v) fetal calf serum and 2 mm glutamine (complete medium; all media components were obtained from Biological Industries). For Western blotting analysis, supernatant from this line was diluted to 400 ng/ml IgG (as determined by ELISA) and used directly. Control experiments carried out with irrelevant human IgGIK (from a patient with myeloma) failed to detect the 14C1 antigen under identical analytical conditions.
93
Table 1. Physical characteristics of the 14C1 antigen on human ovarian cancer cell lines OwMml, Ovan-4 and Ovan-5 Molecular weight Isoelectric point
OwMml Ovan-4
Ovan-5
(reduced) (kD)
(non-reduced)
32 32 26-32
ND 6 3-6-8 63-6-8
ND not determined.
washing in PBS containing 0.5% (v/v)/Tween-20 (PBS-T; Sigma), alkaline-phosphatase-conjugated, goat anti-human IgG (gamma-chain specific; Sigma) was diluted 1/500 in PBS-T and added for 1 h at room temperature, after which time the blot was again extensively washed. The washed blots were equilibrated in substrate buffer (100 mm Tris, 25 mm diethanolamine, 100 mM NaCl, 2 mM MgCl2; pH 9 55) and the bands were visualized by exposure to a substrate solution comprising 0 33 mg/ml nitrobluetetrazolium in substrate buffer, to each ml of which were added 6-7 p1 of 2 mg/ml phenazine methosulphate in water and 3-4 p1 of 40 mg/ml 5-bromo-4-chloro-3-indolyl phosphate (p-toluidine salt) in dimethylformamide, according to the method of Ey & Ashman (1976).
Preparation of tumour samples for electrophoretic analysis After histological confirmation of malignancy, a sample of nonnecrotic tissue was chopped finely and pressed through a nylon mesh to release tumour cells. These cells were lysed in a lysis buffer, comprising 0 5% (w/v) Nonidet P-40 and 0-1 mm PMSF in phosphate-buffered saline, pH 7-4 (PBS; all reagents from Sigma), at 4°C for 30 min. Cell membranes were prepared by centrifugation of the lysis mixture (400g, 10 min) and subjecting the supernatant to further centrifugation (1 5000 g, 45 min, 4°C). The supernatants from this procedure were designated 'membranes' and adjusted to 2 mg/ml protein (Bradford's method) before use in Western blotting experiments. Cell line membranes were prepared similarly. Serum and ascitic fluid samples were also adjusted to 2 mg/ml protein prior to electrophoresis.
Preliminary characterization of the 14CJ antigen Membrane preparations from three 14C 1-positive cell lines (OwMml, Ovan-4 and Ovan-5; unpublished data) were analysed by Western blotting, following SDS-PAGE or IEF separations. The results are summarized in Table 1. The 14CI antigen appeared as a double band at 32 kD in the Ovan-4 and OwMm 1 samples, and as a less-well-defined single band at 25-32 kD on Ovan-5. Both Ovan-4 and Ovan-5 gave a double band at pI 6-3-6-8 following IEF (OwMml not tested).
Western blotting Membrane and other preparations were subjected to SDSPAGE under reducing conditions, using an LKB Midget gel apparatus and 10% resolving gels of 0-75 mm thickness. Membrane samples were diluted to 0-5 mg/ml protein in sample buffer (62-5 mm Tris, 2% (w/v) SDS, 10% (v/v) glycerol, 4% (v/v) 2-mercaptoethanol, and 0 1 mg/ml bromophenol blue) and boiled for 5 min before loading 5-10 pg protein to each track. Alternatively, membrane proteins were separated by isoelectric focusing (IEF), using pre-cast, 1 0-mm thick gels (Pharmacia) in which a pH gradient of 3-5-9 5 was established. Following separation, proteins weere electrophoretically transferred to nitrocellulose membrane ('Hybond-C', Amersham International) using a BioRad Transblot apparatus, under 'wet' conditions in a transfer buffer comprising 20 mm Tris, 192 mM glycine, 1% (w/v) SDS and 25% (v/v) methanol. Transfer was conducted for at least 2 h at 100 V. Blots were 'blocked' by incubating in complete medium for 1 h at room temperature, then stained with 400 ng/ml anti- 14C1 antibody in complete medium for 90 min. Following thorough
Expression of the 14CJ antigen on primary human tissue We next examined the presence of the 14C1 antigen on a range of malignant and normal tissue, with particular reference to ovarian epithelial malignancies. The results are shown in Table 2. A clear, positive correlation between the presence of the 1 4C I antigen and diagnosis ofovarian cancer is shown, particularly in the cases of mucinous, serous and clear cell carcinomas; however, only one out of four specimens of poorly differentiated adenocarcinoma of the ovary was positive. One specimen of histologically benign mucinous cyst was also positive. The analysis of membranes from other tumour types and normal tissue showed a uniform lack of 14CI expression, with the exception of one specimen (of four) of colonic carcinoma, which was positive. Of particular interest in the negative group were the placentae, the benign ovarian serous and papilomatous cysts and the normal ovarian tissue. Sera from ovarian cancer patients, including those whose tumour tissue was positive, uniformly failed to contain the 14C I antigen (27 samples), and of four (malignant) ascitic fluids tested, only one contained the antigen.
RESULTS
G. Gallagher et al.
94
I
The 14C1 antigen presents as a 25-kD in primary tumour material Figure I shows a typical Western blot analysis of primary ovarian cancer material (mucinous cystadenocarcinoma), the antigen is detected as a single band (track c), at or around 25 kD; as previously stated, this band is not observed if irrelevant (myeloma-derived) IgGlK is used in place of the anti-14CI cell line for blotting. Note that another band is detected in all samples at 48 kD, which we have proved to be due to the detection of IgG heavy chain in the sample directly by the conjugate (however, the conjugate does not detect either human light chain).
I
DISCUSSION
Table 2. Expression of 14CI on primary humaIn tumours (Western blotting)
Tumour type Serous carcinoma (ovary) Mucinous carcinoma (ovary) Clear cell carcinoma (ovary) Poorly differentiated adenocarcinoma (ovary) Malignant brenner (ovary)
Benign serous cyst (ovary) Benign papilomatous cyst (ovary) Benign mucinous cyst (ovary) Squamous cell carcinoma (cervix) Squamous cell carcinoma (vulva) Endometrial adenocarcinoma Colon carcinoma Stomach and oesophagus
Positive samples
Number tested
7 6 3 1
9 6 3 4
I
2 1 2 1
4
4
Placenta (term) Vagina (pre-pubescent) Vagina (post-menopausal) Ovary (pre-pubescent) Ovary (active) Ovary (post-menopausal)
2 2 I I
3
Serum (ovarian cancer: benign and malignant) Ascitic fluid (ovarian cancer: malignant)
27 1
4
kD
80 -; .:.,.:
48
26b. ...
..
'.
.
a b c Fig. 1. The 14C I antigen is present as a single band ofF25 kD on primary ovarian cancer tissue. Membrane preparations were clerived from tissue samples, and subjected to Western blotting as de: scribed. Lanes: a, vagina; b, endometrial adenocarcinoma; c, mucinoi us cystadenocarcinoma of the ovary.
This work is based on the premise that an enlarged lymph node represents an immunologically stimulated centre and therefore will contain activated, antigen-specific B cells. EBV transformation of such B cells from cancer patients should allow the examination of tumour immunogenicity, as perceived by the human host (Cote et al., 1986). We have used EBV transformation to prepare a human IgG-secreting cell line from the involved lymph nodes of a patient with ovarian epithelial cancer (Al-Azzawi et al., 1987) and demonstrated that it secretes an antibody able to recognize an antigen on this tumour with a high degree of selectivity. We have termed this antigen 14C1. The 14CI antigen has a molecular weight of 25-32 kD on primary tissue and established ovarian cancer cell lines and an isoelectric point of pI 6 3-6-8. Following SDS-PAGE, the 14C I antigen appears as a single band. The low molecular weight of the 14CI antigen sets it apart from ovarian-cancer-related antigens described previously, with the possible exception of that defined by the antibody MH99 (Mattes et al., 1987), which is a glycoprotein epithelial differentiation antigen, with subunits at 38 and 29 kD. This is within the scope of experimental variation between laboratories; the pI of the MH99 antigen was not reported. The expression of 14C1 seemed to be highly restricted to ovarian epithelial tumours in the limited series described here. Of 24 total ovarian tumours, 16 were positive (670%). Among the eight negative ovarian specimens, two were of the Brenner type (possible transitional epithelial cell origin) and three were of the poorly differentiated adenocarcinoma variety. These data suggest a clearly restricted distribution of the antigen to particular types of ovarian malignancy. Of the other tumour types tested, comprising 13 specimens taken from the colon, stomach, oesophagus, cervix, vulva and endometrium, only one (colonic) was positive (80%). None of the normal tissue specimens was positive, including term placentae, normal ovarian tissue and two out of three benign ovarian cysts. In addition, cell membranes from a range of human tumour cell lines, including Calu-6 (lung), HT-29 and Colo-320 (colorectal), T-24 (bladder), U937 (myeloid) and a human thymic epithelial line of our own derivation, also failed to express 14C1 (data not shown). The presence of the 1 4C I antigen on the specimen of benign mucinous cyst suggests one of two possible scenarios. Firstly, that the patient is borderline malignant and the presence of the 14C1 antigen has a diagnostic significance. Only time will tell, but present indications are that there are no signs of recurrence. Secondly, that the 14C1 antigen is present in mucinous cystic tissue and continues to be expressed if this tissue becomes
14CJ and ovarian cancer malignant. The expression of the tissue on malignant but not benign serous cystic tissue may be due to de novo expression or malignancy-related revelation of the antigen. This question, and that of change in expression on malignant over benign mucinous cysts will be answered when a cDNA probe for 14CI is prepared. While it is disappointing that the antigen is not truly 'tumour specific,' it is probably not realistic to expect that anything other than a virally induced tumour will express such a marker. The 14CI antigen is therefore 'tumour associated' and as such probably represents a cell-cycle protein whose overexpression is in some way associated with the development of the malignant state. In several cases, murine antibodies such as HMFG2 have shown good clinical potential in ovarian cancer, particularly for the treatment of recurrent ascites rather than solid metastatic nodules (Ward & Wallace, 1987). However, murine antibodies are known to solicit an 'anti-mouse' response from the human immune system (Pimm et al., 1985; Reynolds et al., 1986) and this could seriously diminish their clinical usefulness. Although such materials can be 'humanized,' the most satisfactory procedure would be to use a wholly human antibody. Such antibodies would offer the benefit of safe repeated administration, provided one of the appropriate specificity could be found. Much work has gone into the search for a human anti-tumour antibody, with some success (e.g. Kan-Mitchell et al., 1986) and some IgM materials have been considered for clinical use in a range of tumour types (Sikora et al., 1985) but only one is currently available for ovarian cancer (Al-Azzawi et al., 1987). The clinical potential of the anti- 14CI cell line as an imaging and/or targetting agent is strongly promoted by its restricted binding to serous and mucinous cystadenocarcinomas, with a less-well-defined affinity for clear-cell carcinoma (one specimen tested) and a lower percentage binding to poorly differentiated adenocarcinoma (one out of four). Our previous (Al-Azzawi et al., 1987) and more recent (Burch et al., 1990) histological studies clearly demonstrate that the antigen is present on the surface of malignant cells and that it can discriminate between tumour cells and surrounding stromal tissue; the present study uses immunochemical methods to demonstrate the presence of the antigen in membrane preparations. Both these observations enhance the potential use of the antibody as a therapeutic agent. Apart from the direct clinical relevance, it is perhaps more interesting to consider that the benefit of a human antibody may lie in the fact that it defines an antigen of immunological relevance to the patient, where an existing immune response has been demonstrated. 14C1 is such an antigen and may represent an important target for active immunotherapies, as part of the post-surgical management of ovarian cancer.
ACKNOWLEDGMENTS We would like to thank Ms J. Mitchell and Ms S. Webb for excellent research assistance. The work was supported by the Cancer Research Campaign (G.G., L.P.W.), the University of Strathclyde Research and Development Fund (G.W.) and the East Anglia Regional Health Authority (F.A.-A.) The anti-14Cl antibody described in this report is the property of the University of Strathclyde.
REFERENCES AL-AZZAWI, F., GOVAN, A.D. & STIMSON, W.H. (1987) Human antibodies to ovarian cancer antigens secreted by lymphoblastoid cell lines. J. Clin. Lab. Immunol. 22, 71-75.
95
BAST, R., KLUG, T., SCHAETZEL, E., LAVIN, P., NILOFF, J., GREBER, T., ZURAWSKI, V. & KNAPP, R. (1984) Monitoring human ovarian carcinoma with a combination of CA- 125, CA 19-9 and carcinoembryonic antigen. Am. J. Obstet. Gynaecol. 149, 553. BURCH, D., AL-AZZAWI, F., WALSH, L.P., WILSON, G. & GALLAGHER, G. (1990) Human antibodies and host recognition of ovarian carcinoma antigens. In Advances in the Application of Monoclonal Antibodies to Clinical Oncology (ed. by A. A. Epenitos) Chapman & Hall, (In press).
COTE, R.J., MORRISSEY, D.M., HOUGHTON, A.N., THOMSON, T.M., DALY, M.E., OETTGEN, H.F. & OLD, L.J. (1986) Specificity analysis of human monoclonal antibodies reactive with cell surface and intracellular antigens. Proc. natl Acad. Sci. USA, 83, 2959. EY, P.L. & ASHMAN, L.K. (1986) The use of alkaline phosphatase conjugated anti-immunoglobulin with immunoblots for determining the specificity of monoclonal antibodies to protein mixtures. Methods Enzymol. 121, 497. GALLAGHER, G., CHRISTIE, J.F., STIMSON, W.H., Guy, K. & DEWAR, A.E. (1987) T-24 B-cell differentiation factor induces immunoglobulin secretion in human B-cells without prior cell replication. Immunology, 60, 523. HASPEL, M.V., MCCABE, R.P., POMATO, N., JANESCH, N.J., KNOWLTON, J.V., PETERS, L.C., HOOVER, H.C. JR & HANNA, M.G. JR (1985) Generation of tumour-cell reactive human monoclonal antibodies using peripheral blood lymphocytes from actively immunised colorectal carcinoma patients. Cancer Res. 45, 395 1. KAN-MITCHELL, J., IMAM, A., KEMP, R.A., TAYLOR, C.R. & MITCHELL, M.S. (1986) Human monoclonal antibodies directed against melanoma associated tumour antigens. Cancer Res. 46, 2490. KJELDSEN, T.B., RASMUSSEN, B.B., ROSE, C. & ZEUTHEN, J. (1988) Human-human hybridomas and human monoclonal antibodies obtained by fusion of lymph node lymphocytes from breast cancer patients. Cancer Res. 48, 3208. MATTES, M.J., LOOK, K., FURUKAWA, K., PIERCE, V.K., OLD, L.J., LEWIS, J.L. JR & LLOYD, K.O. (1987) Mouse monoclonal antibodies to human epithelial differentiation antigens expressed on the surface of ovarian carcinoma ascites cells. Cancer Res. 47, 6741. PIMM, H.V., PERKINS, A.C., ARMITAGE, N.C. & BALDWIN, R.W. (1985) The characteristics of blood-borne radiolabels and the effect of antimouse IgG antibodies on localisation of radiolabelled monoclonal antibody in cancer patients. J. nucl. Med. 26, 101 1. REYNOLDS, J.C., CARRASQUILLO, J.A., KENAN, A.M., LORA, M.E., SUGARBAKER, P., ABRAMS, P., FOON, K., MULSHINE, J.L., ROTH, J., COLCHER, D., SCHLOM, J. & LARSON, S.M. (1986) Human anti-murine antibodies following immunoscintigraphy or therapy with radiolabelled monoclonal antibodies. J. nucl. Med. 27, 1022. SAKAKIBARA, K., UEDA, R., OHTA, M., NAKASHIMA, N., TOMODA, Y. & TAKAHASHI, T. (1988) Three novel mouse monoclonal antibodies, OM-A, OM-B and OM-C, reactive with mucinous type ovarian tumours. Cancer Res. 48, 4639. SEKINE, H., HAYES, D.F., OHNO, T., KEEFE, K.A., SCHAETZEL, E., BAST, R.C., KNAPP, R.C. & KUFE, D.W. (1 985) Circulating DF3 and CA 125 antigen levels in serum from patients with epithelial ovarian carcinoma. J. clin. Oncol. 3, 1355. SIKORA, C., ALDERSON, T., NETHERSELL, A. & SMEDLEY, H. (1985) Tumour localisation by human monoclonal antibodies. Oncol. Tumour Pharmacother. 2, 77. UMBACH, G. (1984) Review of tumour markers for ovarian cancer. Med. Hypotheses, 13, 329. WARD, B.G. & WALLACE, K. (1987) Localisation of the monoclonal antibody HMFG2 after intraperitoneal and intravenous injection into nude mice bearing subcutaneous and intraperitoneal human ovarian cancer xenografts. Cancer Res. 47, 4714. WARD, B.G., MATHER, S., SHEPHERD, J., CROWTHER, M., HAWKINS, L., BRITON, K. & SLEVIN, M.L. (1988) The treatment of intraperitoneal malignant disease with monoclonal antibody guided I-131 radiotherapy. Br. J. Cancer, 58, 658.
