Publication of the International Union Against Cancer Publication de I’Union Internationale Contre le Cancer
Znt. J. Cancer: 50,373-381 (1992) 0 1992 Wiley-Liss, Inc.
ISOLATION AND CHARACTERIZATION OF A MONOCLONAL ANTIBODY, K1, REACTIVE WITH OVARIAN CANCERS AND NORMAL MESOTHELIUM Kai CHANG,Ira PASTANand Mark C. WILLINGHAM Laboratory of Molecular Biology, Division of Cancer Biology, Diagnosis and Centers, NCI, NIH, Bethesda, MD 20892, USA. We have isolated a new monoclonal antibody (MAb), K I ,that reacts with an epitope on the surface of human ovarian carcinoma cells. This antibody was generated by immunization of mice with periodate-treated human ovarian carcinoma (OVCAR-3) cells. These mice had been previously made tolerant with normal human kidney membranes. Spleen lymphocytes from these mice were selected prior t o fusion using a panning purification method on living OVCAR-3 cells. Initial screening of surface-reactive clones was performed in a single day using immunofluorescence on living OVCAR-3 cells, and secondary screening was performed using immunoperoxidase histochemistry on cryostat sections of normal human tissues and human tumors. The K I clone was subcloned and identified as an IgM isotype, but was subsequently isotype-switched to Is,,using a panning selection method. When evaluated by immunohistochemistry, the antigen reactive with K I was found in many ovarian non-mucinous tumors, as well as in squamous tumors of the esophagus, and cervical cancer. The only normal adult human tissues showing uniform reactivity with K I were the mesothelia of the peritoneal, pleural and pericardial cavities. There was also limited reactivity with epithelia of the trachea, tonsil and Fallopiantube. A similar tissue reactivity for K I was found in tissues from cynomolgus monkeys. K I reacted with many of the same tissues and tumors as the previously identified antibody OC 125, but several lines of evidence indicate that K I reacts with a different epitope and probably a different molecule, when compared t o OC 125. This evidence included assays employing immunofluorescence competition, doublelabel immunofluorescence, and solid-phase and live-cellradioimmunoassays. Since our data indicate that the antigen reactive with the K I antibody is a new molecular species, we have named the antigen CAKI. Unlike the shed antigen CAl25, CAKl was only cell-associated and was not found in the supernatant of cultured OVCAR-3 cells or in the blood of ovarian cancer patients. The K I antibody may be useful as a targeting agent for therapy and in the diagnosis of ovarian carcinoma, as well as some other human cancers.
Current therapy for most human cancers using surgery, radiation or chemotherapy is often unsuccessful unless the tumor is detected at an extremely early stage. Unfortunately, many human tumors are not detected until they have developed distant metastases. For some tumors, such as leukemias or ovarian cancer, their location and characteristics make metastasis a very early event. Therefore, much current interest has centered on methods of earlier diagnosis and on therapeutic approaches that can selectively attack small numbers of target tumor cells in metastatic sites. Immuno-targeting of therapeutic agents has a theoretically attractive potential for finding and destroying metastatic foci (Vitetta and Uhr, 1985; Pastan et al., 1986). Immunotoxins, immuno-directed radiotherapy or selective activation of immune mechanisms require the mediation of a highly selective targeting agent (Hellstrom and Hellstrom, 1985; Reisfeld and Cheresh, 1985). With the advent of MAb techniques (Kohler and Milstein, 1975), it has been possible to identify antibodies that react with the surface of tumor cells. In spite of the isolation of many such antibodies, very few have been found that are truly tumor-specific. OC12.5 (Bast et al., 1981) is an antibody that reacts with a high-molecular-weight glycoprotein complex antigen, CA125, that is shed from the surface of ovarian carcinomas. This antigen can be detected in the blood of ovarian carcinoma patients and provides a very useful diagnostic screening assay
(Bast et al., 1983). Such an antigen is less useful as a therapeutic target because of the large amount of circulating antigen in the blood-stream. We have attempted to isolate antibodies reactive with the surface of living, human ovarian carcinoma cells using a variety of methods for immunization and screening. We now describe a recently isolated antibody, K1, which reacts with many ovarian cancers and a limited number of normal tissues, and merits further evaluation as a therapeutic agent for ovarian cancers. We have named the molecule reactive with this antibody CAKl. MATERIAL AND METHODS
Cell culture Human ovarian tumor cell lines OVCAR-2, -3, -4, and -5, A-1847 and A-2780 were a gift from Dr. T. Hamilton (Fox Chase Cancer Center). The cell lines A431, AGS (CRL 1739), HTB20 (BT-474), HTB3O (SK-BR-3), HTB33 (ME-180), HTB77 (SKOV3), HTB103 (Kato 111), HeLa S3, HeLa, HT-29, KB, MDA-MB-468, MCF-7 and DU145 were obtained from the ATCC (Rockville, MD). OVCAR-3 cells and other cell lines were grown in RPMI 1640 medium supplemented with L-glutamine, penicillin, streptomycin and 5-10% fetal bovine serum (FBS) (GIBCO, Grand Island, NY).’ Tolerizationand immunization of mice Six-week-old female BALB/c mice were injected intraperitoneally (i.p.) with crude human kidney cell membranes ( 5 mg protein in 1ml). Ten minutes later, they were injected i.p. with 1.5 mg cyclophosphamide (Sigma, St Louis, MO) in 0.75 ml PBS. The same amount of cyclophosphamide was injected again at 24 hr and 48 hr. This entire procedure was repeated 2 more times at intervals of 2 weeks (Matthew and Sandrock, 1987). Three weeks after the final tolerization, the mice were immunized with OVCAR-3 cells. These cells were grown in 150-cmZflasks. The cells were next incubated in situ with 10 mM sodium periodate in 100 mM sodium acetate buffer (pH 4.5) at 23°C for 15 min, then with 50 p,g/ml goat anti-mouse IgG (H L) (affinity-purified) (Jackson ImmunoResearch, West Grove, PA) in 50 mM sodium borate, 100 mM sodium chloride (pH 8.5) at 37°C for 60 min. This procedure was employed to couple this anti-mouse IgG antibody to the cell surface in the hope of targeting these cells to antibody-bearing lymphocytes in sites with large numbers of immunocompetent cells. The cells were next treated with 0.5 mg/ml sodium borohydride in PBS for 3 min and scraped off the plastic, and then the suspension of cells in PBS (2.5 x lo6cells/0.5 ml) was injected i.p. into each tolerized mouse. These immunizations were repeated 3, 4 and 8 weeks after the first immunization. The final immunization (at 8 weeks) was carried out 4 days prior to the fusion procedure. The spleens of 2 immunized mice were removed and a single cell suspension was prepared for a panning selection procedure.
+
Panning selectionprocedure The suspension of spleen cells from 2 mice was incubated with living OVCAR-3 cells as follows: 100-cm diameter culture Received: July 4,1991 and in revised form September 19,1991.
374
CHANG ET AL.
dishes coated with Cell-Tak (Collaborative Research, Bedford, MA) were seeded with a monolayer of OVCAR-3 cclls the day before the procedure. The suspension of splenocytes (3 x lo7cells in 3 ml PBS with 5% FBS) was then placed on to this monolayer of OVCAR-3 cells and incubated for 60 rnin at 4°C. The unattached cells were removed with a pipette with gentle swirling of the dish, followed by addition of 5 ml PBS-FBS for 30 rnin at 4°C. These unattached cells were then discarded. The attached splenocytes were gently removed by repeated washing with a pipette, collecting each harvest of cells in a conical tube. The harvested splenocytes were fused with mouse myeloma P3X63-AG8.653 cells at a ratio of 5 to 1 using 50% PEG 4000 (GIBCO, Grand Island, NY) and plated in the central 60 wells of 96-well microtiter culture plates under HAT selection.
with 20% FBS in RPMI 1640 medium and again subcloned. The supernatants of these subclones were screened by immunofluorescence on living OVCAR-3 cells using rhodaminelabelled goat anti-mouse IgG (Fc-specific) (Jackson ImmunoResearch). IgG-secreting clones were selected and maintained in RPMI 1640 medium with 10% FBS. These clones were used to prepare ascites fluid in nude mice. K1 antibody was purified from either culture supernatant or ascites fluid using FPLCProtein A affinity chromatography. Using isotype-specific antimouse antibodies and immunofluorescence on living OVCAR-3 cells, K1 has been characterized as an IgG,, isotype antibody.
Other monoclonal antibodies OC125 and '251-labelledOC125 were either purchased from Centocor (Malvern, PA) and Amersham (Arlington Heights, IL), respectively, or were generous gifts from Drs. K. Lau and R.C. Knapp (Brigham and Women's Hospital), or Dr. R. Bast (Duke University Medical Center) (Bast et al., 1981, 1983). HB21, a mouse MAb reactive with the human transferrin receptor (Haynes et al., 19Sl), was prepared from hybridoma cells obtained from ATCC. 56, an IgG, h4Ab reactive with a rat-specific epitope (Hasamura et al., 1986), and used as a negative control, was a generous gift of Dr. S.-Y. Cheng (NCI).
Hybridoma culture and screening The somatic cell hybrids were grown in complete RPMI medium with 20% FBS with the addition of 20% conditioned medium from the mouse cell line J774 a.1 (ATCC). The initial clones were allowed to grow for 2 weeks prior to screening of the supernatants. Screening was performed as previously described (Willingham and Pastan, 1990) using a ScreenFast &amber (GIBCOIBRL, Gaithersburg, MD) with immunoflu- Immunofluorescence examination of living cells orescence microscopy on living OVCAR-3 cells. Then, 100 p1 Cells grown in 35-mm diameter culture dishes were incusupernatants were transferred using a 6-channel pipettor bated with antibodies at 4°C in BSA-PBS. Primary mouse (Flow, McLean, VA) and liquids were removed from the wells antibodies were used at 10 pg/ml in BSA-PBS for 30-60 rnin at using a 6-channel suction manifold (IntiCorp, San Rafael, 4°C and, after washing in PBS, were incubated with affinityCA). OVCAR-3 cells were plated on Cell-Tak in 150-mm purified goat anti-mouse IgG (H + L) conjugated to rhodamine culture dishes the day before screening. Each of these dishes (25 yg/ml) in BSA-PBS at 4°C for another 30 min. After were then assembled into a ScreenFast chamber at 4°C for 60 washing, the cells were fixed in formaldehyde and mounted in min and the supernatants were added to each well. Then the situ under coverslips in buffered glycerol. wells were washed, the chamber was disassembled and the cells in the open dishes were incubated with affinity-purified Immunoperoxidase histochemistry goat anti-mouse IgG (H + L) conjugated to rhodamine (25 Cryostat sections of normal human and monkey tissues and pg/ml in BSA-PBS) which was added for another 60 min. human tumors were prepared and processed for peroxidase After washing in PBS, the cells were fixed in 3.7% formalde- immunohistochemistry as described (Willingham, 1990). Briefly, hyde in PBS and covered with glycerol. The cells were then the cryostat sections of fresh-frozen tissue were thaw-mounted viewed with high-resolution epifluorescence optics on a Zeiss and air-dried onto Histostik (Accurate Chemical, Westbury, upright microscope using glycerol immersion. Of the original NY)-coated coverslips (22 x 22 mm), lyophilized for 4 hr and 480 growing clones in this fusion, 46 were selected for fixed in 100% acetone for 10 min at room temperature. All secondary screening based on the presence of bright surface subsequent antibody incubations were carried out in a diluent fluorescence on living OVCAR-3 cells. containing 0.1% saponin, 4 mg/ml normal goat globulin and Secondary screening was performed as described (Willing- PBS (sap-NGG-PBS). Primary antibodies (10 pg/ml) were ham, 1990) using immunoperoxidase histochemistry on cry- incubated with sections at room temperature for 60 min, then ostat sections of a composite tissue block containing normal washed and incubated with peroxidase-conjugated affinityhuman liver, kidney and colon, as well as tumors of the ovary (3 purified goat anti-mouse IgG (H L) (minimal crosssamples), stomach (1 sample), and colon (3 samples). Six reactivity to human serum proteins) (Jackson ImmunoReclones, including K1, were selected for their reactivity with search) (25 pg/ml 5% human AB serum). The sections were tumors but not with important normal tissues. They were then developed using diaminobenzidine-H,O, and counteranalyzed further using peroxidase histochemistry on many stained with Gill's hematoxylin prior to dehydration and more samples of human tumors and normal tissues (including mounting in Permount. cerebellum, cerebral cortex, lung, heart, bone marrow and small bowel). The K1 clone was selected for study and Antibody competition and double-labelling immunofluorescence Both K1 and OC125 antibodies were directly conjugated subcloned twice using limiting dilution cloning. with fluorescein isothiocyanate or tetramethylrhodamine isothiocyanate (Research Organics, Cleveland, OH). For fluoIsotyping and isotype switching The K1 clone originally selected secreted an IgM antibody. rescence competition experiments, OVCAR-3 cells plated on This was detected in the screening assay using anti-mouse IgG 35-mm dishes were cooled to 4°C in BSA-PBS (2 mg/ml BSA) because of cross-reactions with the light chains. This clone was and pre-incubated with 25-50 pg/ml unlabelled competitor converted to an IgG-secreting form using a panning-isotype antibody for 60 rnin (in BSA-PBS), then incubated with 10 switching method to be described in detail elsewhere. Briefly, pg/ml labelled antibody and 25-50 pg/ml unlabelled competibacteriological Petri dishes 100 mm in diameter were coated tor antibody together for another 60 min. After washing, the with 25-50 pgiml affinity-purified rabbit anti-mouse IgG, cells were fixed in formaldehyde. For double-labelling experiments, OVCAR-3 cells were first Fc-specific (Jackson ImmunoResearch) in 50 mM Tris, (pH 9.5). K1 hybridoma cells ( lo7 cells/plate) in 3 ml5% FBS-PBS incubated with rhodamine-labelled K1 antibody for 30 min (10 were incubated undisturbed in this dish at 4°C for 30 min, then pg/ml), then with FITC-labelled OC125 antibody (10 pg/ml) redistributed by tilting and swirling the dish. After another for 30 min. In other experiments, the order was reversed, and 30-60 rnin at 4"C, the suspended cells were removed by in some experiments the 2 antibodies were added together. All pipetting and discarded, and the attached cells were incubated of these double-label methods gave identical results.
+
+
375
OVARIAN CANCER MONOCLONAL ANTIBODY
Live-cell ELISA assay An ELISA assay was designed to determine whether K1 showed additive binding with other antibodies. Cells were incubated with various amounts of antibodies together, and the total binding of all mouse antibodies was detected using peroxidase-labelled goat anti-mouse IgG (25 pg/ml) (Jackson ImmunoResearch). The peroxidase activity in each well in these assays was detected using 100 p1 of 1 mg/ml 2,2'-azinobis(3-ethylbenzthiazoline-6-sulfonicacid) (Boehringer-Mannheim, Indianapolis, IN) in citrate buffer (pH 4.4) with 0.01% H,O,. The reaction was stopped by adding 100 pl 10% SDS, and the color reaction was quantitated using a microplate reader (Molecular Devices, Menlo Park, CA).
RESULTS
Generation of monoclonal antibody K l BALB/c mice were tolerized with a normal human kidney membrane preparation by a cyclophosphamide procedure (Matthew and Sandrock, 1987) designed to minimize the production of antibodies to antigens present on normal human cells. They were then immunized over many weeks with living OVCAR-3 cells. These cells had previously been treated with periodate and conjugated to an anti-mouse IgG antibody; this conjugation was performed to help target these human cells to areas rich in mouse-IgG-covered lymphocytes, areas that would be enriched with cells of the immune system. The spleen cells removed from these mice were then subjected to a panning selection (Wysocki and Sato, 1978) to enrich for splenocytes with surface-bound antibody that reacted with the surface of living OVCAR-3 cells. The selected splenocytes Solid-phase radioimmunoassays and competitive live-cell RIA Serum-free OVCAR-3 cell culture supernatant concentrate represented 1040% of the initial spleen cells, and they were was prepared by using PBS to wash 80% confluent OVCAR-3 then fused with mouse myeloma cells and grown under HAT cultures free of serum-containing medium, then incubating selection. Supernatants from the growing clones of hybridoma cells these cells in a serum-free medium (ABC medium, PAN DATA, Gainesville, FL) for 2 days at 37°C. The supernatant were screened in a single day for production of mouse medium was collected and spun at 10,000 g for 60 min to antibodies reactive with the surface of living OVCAR-3 cells eliminate cell debris, then concentrated by ultrafiltration using using immunofluorescence and ScreenFast chambers (Willinga Diaflo PM30 membrane (Amicon, Danvezs, MA). The ham and Pastan, 1990). Of the original 480 growing clones, 46, which were shown by immunofluorescence to have uniform resultant supernatant was concentrated 50- to 100-fold. surface localization, were selected for the secondary screening Serum-free OVCAR-3 culture supernatant concentrate, step using peroxidase immunohistochemistry. Of these 46, 6 CA125 standard antigen (Centocor, Amersham), and sera were chosen for their reactivity with human tumors but lack of from 16 ovarian cancer patients were assayed for specific reactivity with normal liver, kidney and colon. One of these 6 binding with K1 and OC125 antibodies using a sandwich clones, K1 was saved and further characterized using immunosolid-phase radioimmunoassay. One hundred microliters, con- histochemistry on many normal tissues and human tumors. taining 0.5 pg K1, OC125, or a control mouse antibody, were The initial K1 clone was of the IgM isotype, but this was added to each well of a 96-well plate (12-strip styrene RIA switched to an IgG, isotype using a panning selection method plate) (GIBCO) and allowed to dry overnight at room temper- and subsequently recloned 4 times. ature. The wells were quenched to minimize non-specific adsorption of proteins by the addition of 250 pI of 4% BSA in Immunojluorescence reactivity of K l with cell lines PBS for 1 hr at 37°C. After washing, the wells were incubated Figure 1 shows the homogeneously strong surface binding of with 100 p1 of CA125 standard antigen from 7.5 U/ml to 500 K1 on cultured cells using immunofluorescence. K1 reacted U/ml, undiluted patient serum samples or triplicate dilutions strongly with OVCAR-3, KB, A-1847 and AGS cells, but not of OVCAR-3 supernatant for 1 hr at 37°C. The unbound with SK-BR-3 cells. OC125 also reacted strongly with antigens were removed by washing with 1% BSA-PBS 5 times OVCAR-3 cells, but did not react with A-1847 or AGS cells; it and the wells were then incubated with 100 pI of 1251-labelled did react with SK-BR-3 cells. Table I shows the reactivities of OC125 (6000 cpm/100 pl) (Centocor) or 100 pl of 1251-labelled K1 with many different cultured human carcinoma cell lines in K1 (10,000 cpm/100 pl) (Lmfstrand, Gaithersburg, MD) for 2 this assay, and compares these reactivities with those of the to 12 hr at 23°C. The wells were then washed extensively with previously isolated antibody OC125, reactive with the CA125 1% BSA-PBS. The individual wells were separated from the antigen (Bast et al., 1981; Davis et al., 1986). K1 reacts strongly strip, and radioactivity was counted in a gamma counter with many cell lines derived from ovarian cancers, including OVCAR-2, 3, 4 and 5 , and A-1847 and HTB77 (SKOV3). It (Beckman Gamma 5500B, Irvine, CA). also reacts with HeLa and 2 HeLa variants (HeLa S3 and KB Live-cell RIA was carried out using lZSI-IabelledK1 or cells), originally derived from a cervical carcinoma. While OC125 (70-100,000 cpm/100 pl/well). OVCAR-3 cells plated OC125 reacts with some of these ovarian-tumor cell lines, it in Cell-Tak-coated wells of a 96-well microtiter plate were does not react with A-1847 and reacts poorly with KB cells. incubated with 5% BSA/RPMI 1640 at 37°C for 1 hr, then Neither antibody reacts with A-2780, another ovarianincubated with unlabelled K1 or OC125 antibody as a compet- carcinoma-derived cell line. In addition, K1 reacts well with itor in triplicates from 0.003 ng/ml to 50 pg/ml before and the cell lines AGS and HTB103 (KatoIII) originally derived during the incubation with iodinated antibody (either K1 or from gastric carcinomas, whereas OC125 fails to react with OCl25) (2 hr, 23°C). these lines. K1 fails to react with a number of other cell lines, such as HTB30 (SK-BR-3) which shows a strong reaction with OC125. These data indicate that K1 reacts well with nonTrypsin, proteinase K, neuraminidase and periodate treatments mucinous ovarian-carcinoma-derived cell lines, and with lines Live OVCAR-3 cells were incubated with enzymes for 2 hr derived from cervical and gastric tumors. These data also at 37°C as follows: with neuraminidase (type X, Sigma) at 0.2 illustrate many differences in reactivity between K1 and U/ml in 50% PBS with 50% 0.1 mM acetate buffer (pH 5.0); OC125, indicating that they react with different epitopes. with 2% porcine pancreatic trypsin (type IX, Sigma); or with 0.2 mg/ml proteinase K, in 0.1 M Tris-HCI, 50 mM CaCl,,.pH Immunoperoxidase histochemistry in normal tissues 7.8. Protease-treated cells were reattached to poly-L-lysineK1 and OC125 were also examined for reactivity with treated dishes prior to immunofluorescence examination with normal human and monkey tissues using peroxidase immunoK1 or OC125 antibodies. Cells were treated with periodate histochemistry in cryostat sections. Figure 2 shows examples of using 10 mM NaIO, in 100 mM acetate buffer, pH 4.5 for 15,30 the localization of K1 and OC125 in normal human mesothelia. Figure 3 shows the localization of K1 and OC125 in other and 60 min at room temperature.
376
CHANG ET AL. TABLE I - IMMUNOFLUORESCENCE LOCALIZATION OF K1 AND OC125 ON HUMAN CULTURED CELL LINES
Cell line
OVCAR-2 (ovarian ca.) OVCAR-3 (ovarian ca.) OVCAR-4 (ovarian ca.) OVCAR-5 (ovarian ca.) A-1847 (ovarian ca.) HTB77(SKOV3) (ovarian ca.) A-2780 (ovarian ca.) HTB33 (ME-180) (cervical ca.) HeLa (cervical ca.) HeLa S3 (cervical ca.) KB (cervical ca.) AGS (CRL 1739) (gastric ca.) ' HTB103 (Kato 111) (gastric'ca.) ' A431 (epidermoid ca.) HT-29 (colon ca.) HTB30 (SK-BR-3) (breast ca.) HTB20 (BT-474) (breast ca.) MDA-MB-468 (breast ca.) MCF-7 breast ca.) DU145 [prostate ca.)
++ ++ het ++ het +
++
+ + het (50% -) +++ het ++++ het
++S
+- + het (30%) ++ het (10% ++) -
++ ++ +++ ++ ++
-
++ ++
- ( < 5 % ++) -
-
+++ (
Znt. J. Cancer: 50,373-381 (1992) 0 1992 Wiley-Liss, Inc.
ISOLATION AND CHARACTERIZATION OF A MONOCLONAL ANTIBODY, K1, REACTIVE WITH OVARIAN CANCERS AND NORMAL MESOTHELIUM Kai CHANG,Ira PASTANand Mark C. WILLINGHAM Laboratory of Molecular Biology, Division of Cancer Biology, Diagnosis and Centers, NCI, NIH, Bethesda, MD 20892, USA. We have isolated a new monoclonal antibody (MAb), K I ,that reacts with an epitope on the surface of human ovarian carcinoma cells. This antibody was generated by immunization of mice with periodate-treated human ovarian carcinoma (OVCAR-3) cells. These mice had been previously made tolerant with normal human kidney membranes. Spleen lymphocytes from these mice were selected prior t o fusion using a panning purification method on living OVCAR-3 cells. Initial screening of surface-reactive clones was performed in a single day using immunofluorescence on living OVCAR-3 cells, and secondary screening was performed using immunoperoxidase histochemistry on cryostat sections of normal human tissues and human tumors. The K I clone was subcloned and identified as an IgM isotype, but was subsequently isotype-switched to Is,,using a panning selection method. When evaluated by immunohistochemistry, the antigen reactive with K I was found in many ovarian non-mucinous tumors, as well as in squamous tumors of the esophagus, and cervical cancer. The only normal adult human tissues showing uniform reactivity with K I were the mesothelia of the peritoneal, pleural and pericardial cavities. There was also limited reactivity with epithelia of the trachea, tonsil and Fallopiantube. A similar tissue reactivity for K I was found in tissues from cynomolgus monkeys. K I reacted with many of the same tissues and tumors as the previously identified antibody OC 125, but several lines of evidence indicate that K I reacts with a different epitope and probably a different molecule, when compared t o OC 125. This evidence included assays employing immunofluorescence competition, doublelabel immunofluorescence, and solid-phase and live-cellradioimmunoassays. Since our data indicate that the antigen reactive with the K I antibody is a new molecular species, we have named the antigen CAKI. Unlike the shed antigen CAl25, CAKl was only cell-associated and was not found in the supernatant of cultured OVCAR-3 cells or in the blood of ovarian cancer patients. The K I antibody may be useful as a targeting agent for therapy and in the diagnosis of ovarian carcinoma, as well as some other human cancers.
Current therapy for most human cancers using surgery, radiation or chemotherapy is often unsuccessful unless the tumor is detected at an extremely early stage. Unfortunately, many human tumors are not detected until they have developed distant metastases. For some tumors, such as leukemias or ovarian cancer, their location and characteristics make metastasis a very early event. Therefore, much current interest has centered on methods of earlier diagnosis and on therapeutic approaches that can selectively attack small numbers of target tumor cells in metastatic sites. Immuno-targeting of therapeutic agents has a theoretically attractive potential for finding and destroying metastatic foci (Vitetta and Uhr, 1985; Pastan et al., 1986). Immunotoxins, immuno-directed radiotherapy or selective activation of immune mechanisms require the mediation of a highly selective targeting agent (Hellstrom and Hellstrom, 1985; Reisfeld and Cheresh, 1985). With the advent of MAb techniques (Kohler and Milstein, 1975), it has been possible to identify antibodies that react with the surface of tumor cells. In spite of the isolation of many such antibodies, very few have been found that are truly tumor-specific. OC12.5 (Bast et al., 1981) is an antibody that reacts with a high-molecular-weight glycoprotein complex antigen, CA125, that is shed from the surface of ovarian carcinomas. This antigen can be detected in the blood of ovarian carcinoma patients and provides a very useful diagnostic screening assay
(Bast et al., 1983). Such an antigen is less useful as a therapeutic target because of the large amount of circulating antigen in the blood-stream. We have attempted to isolate antibodies reactive with the surface of living, human ovarian carcinoma cells using a variety of methods for immunization and screening. We now describe a recently isolated antibody, K1, which reacts with many ovarian cancers and a limited number of normal tissues, and merits further evaluation as a therapeutic agent for ovarian cancers. We have named the molecule reactive with this antibody CAKl. MATERIAL AND METHODS
Cell culture Human ovarian tumor cell lines OVCAR-2, -3, -4, and -5, A-1847 and A-2780 were a gift from Dr. T. Hamilton (Fox Chase Cancer Center). The cell lines A431, AGS (CRL 1739), HTB20 (BT-474), HTB3O (SK-BR-3), HTB33 (ME-180), HTB77 (SKOV3), HTB103 (Kato 111), HeLa S3, HeLa, HT-29, KB, MDA-MB-468, MCF-7 and DU145 were obtained from the ATCC (Rockville, MD). OVCAR-3 cells and other cell lines were grown in RPMI 1640 medium supplemented with L-glutamine, penicillin, streptomycin and 5-10% fetal bovine serum (FBS) (GIBCO, Grand Island, NY).’ Tolerizationand immunization of mice Six-week-old female BALB/c mice were injected intraperitoneally (i.p.) with crude human kidney cell membranes ( 5 mg protein in 1ml). Ten minutes later, they were injected i.p. with 1.5 mg cyclophosphamide (Sigma, St Louis, MO) in 0.75 ml PBS. The same amount of cyclophosphamide was injected again at 24 hr and 48 hr. This entire procedure was repeated 2 more times at intervals of 2 weeks (Matthew and Sandrock, 1987). Three weeks after the final tolerization, the mice were immunized with OVCAR-3 cells. These cells were grown in 150-cmZflasks. The cells were next incubated in situ with 10 mM sodium periodate in 100 mM sodium acetate buffer (pH 4.5) at 23°C for 15 min, then with 50 p,g/ml goat anti-mouse IgG (H L) (affinity-purified) (Jackson ImmunoResearch, West Grove, PA) in 50 mM sodium borate, 100 mM sodium chloride (pH 8.5) at 37°C for 60 min. This procedure was employed to couple this anti-mouse IgG antibody to the cell surface in the hope of targeting these cells to antibody-bearing lymphocytes in sites with large numbers of immunocompetent cells. The cells were next treated with 0.5 mg/ml sodium borohydride in PBS for 3 min and scraped off the plastic, and then the suspension of cells in PBS (2.5 x lo6cells/0.5 ml) was injected i.p. into each tolerized mouse. These immunizations were repeated 3, 4 and 8 weeks after the first immunization. The final immunization (at 8 weeks) was carried out 4 days prior to the fusion procedure. The spleens of 2 immunized mice were removed and a single cell suspension was prepared for a panning selection procedure.
+
Panning selectionprocedure The suspension of spleen cells from 2 mice was incubated with living OVCAR-3 cells as follows: 100-cm diameter culture Received: July 4,1991 and in revised form September 19,1991.
374
CHANG ET AL.
dishes coated with Cell-Tak (Collaborative Research, Bedford, MA) were seeded with a monolayer of OVCAR-3 cclls the day before the procedure. The suspension of splenocytes (3 x lo7cells in 3 ml PBS with 5% FBS) was then placed on to this monolayer of OVCAR-3 cells and incubated for 60 rnin at 4°C. The unattached cells were removed with a pipette with gentle swirling of the dish, followed by addition of 5 ml PBS-FBS for 30 rnin at 4°C. These unattached cells were then discarded. The attached splenocytes were gently removed by repeated washing with a pipette, collecting each harvest of cells in a conical tube. The harvested splenocytes were fused with mouse myeloma P3X63-AG8.653 cells at a ratio of 5 to 1 using 50% PEG 4000 (GIBCO, Grand Island, NY) and plated in the central 60 wells of 96-well microtiter culture plates under HAT selection.
with 20% FBS in RPMI 1640 medium and again subcloned. The supernatants of these subclones were screened by immunofluorescence on living OVCAR-3 cells using rhodaminelabelled goat anti-mouse IgG (Fc-specific) (Jackson ImmunoResearch). IgG-secreting clones were selected and maintained in RPMI 1640 medium with 10% FBS. These clones were used to prepare ascites fluid in nude mice. K1 antibody was purified from either culture supernatant or ascites fluid using FPLCProtein A affinity chromatography. Using isotype-specific antimouse antibodies and immunofluorescence on living OVCAR-3 cells, K1 has been characterized as an IgG,, isotype antibody.
Other monoclonal antibodies OC125 and '251-labelledOC125 were either purchased from Centocor (Malvern, PA) and Amersham (Arlington Heights, IL), respectively, or were generous gifts from Drs. K. Lau and R.C. Knapp (Brigham and Women's Hospital), or Dr. R. Bast (Duke University Medical Center) (Bast et al., 1981, 1983). HB21, a mouse MAb reactive with the human transferrin receptor (Haynes et al., 19Sl), was prepared from hybridoma cells obtained from ATCC. 56, an IgG, h4Ab reactive with a rat-specific epitope (Hasamura et al., 1986), and used as a negative control, was a generous gift of Dr. S.-Y. Cheng (NCI).
Hybridoma culture and screening The somatic cell hybrids were grown in complete RPMI medium with 20% FBS with the addition of 20% conditioned medium from the mouse cell line J774 a.1 (ATCC). The initial clones were allowed to grow for 2 weeks prior to screening of the supernatants. Screening was performed as previously described (Willingham and Pastan, 1990) using a ScreenFast &amber (GIBCOIBRL, Gaithersburg, MD) with immunoflu- Immunofluorescence examination of living cells orescence microscopy on living OVCAR-3 cells. Then, 100 p1 Cells grown in 35-mm diameter culture dishes were incusupernatants were transferred using a 6-channel pipettor bated with antibodies at 4°C in BSA-PBS. Primary mouse (Flow, McLean, VA) and liquids were removed from the wells antibodies were used at 10 pg/ml in BSA-PBS for 30-60 rnin at using a 6-channel suction manifold (IntiCorp, San Rafael, 4°C and, after washing in PBS, were incubated with affinityCA). OVCAR-3 cells were plated on Cell-Tak in 150-mm purified goat anti-mouse IgG (H + L) conjugated to rhodamine culture dishes the day before screening. Each of these dishes (25 yg/ml) in BSA-PBS at 4°C for another 30 min. After were then assembled into a ScreenFast chamber at 4°C for 60 washing, the cells were fixed in formaldehyde and mounted in min and the supernatants were added to each well. Then the situ under coverslips in buffered glycerol. wells were washed, the chamber was disassembled and the cells in the open dishes were incubated with affinity-purified Immunoperoxidase histochemistry goat anti-mouse IgG (H + L) conjugated to rhodamine (25 Cryostat sections of normal human and monkey tissues and pg/ml in BSA-PBS) which was added for another 60 min. human tumors were prepared and processed for peroxidase After washing in PBS, the cells were fixed in 3.7% formalde- immunohistochemistry as described (Willingham, 1990). Briefly, hyde in PBS and covered with glycerol. The cells were then the cryostat sections of fresh-frozen tissue were thaw-mounted viewed with high-resolution epifluorescence optics on a Zeiss and air-dried onto Histostik (Accurate Chemical, Westbury, upright microscope using glycerol immersion. Of the original NY)-coated coverslips (22 x 22 mm), lyophilized for 4 hr and 480 growing clones in this fusion, 46 were selected for fixed in 100% acetone for 10 min at room temperature. All secondary screening based on the presence of bright surface subsequent antibody incubations were carried out in a diluent fluorescence on living OVCAR-3 cells. containing 0.1% saponin, 4 mg/ml normal goat globulin and Secondary screening was performed as described (Willing- PBS (sap-NGG-PBS). Primary antibodies (10 pg/ml) were ham, 1990) using immunoperoxidase histochemistry on cry- incubated with sections at room temperature for 60 min, then ostat sections of a composite tissue block containing normal washed and incubated with peroxidase-conjugated affinityhuman liver, kidney and colon, as well as tumors of the ovary (3 purified goat anti-mouse IgG (H L) (minimal crosssamples), stomach (1 sample), and colon (3 samples). Six reactivity to human serum proteins) (Jackson ImmunoReclones, including K1, were selected for their reactivity with search) (25 pg/ml 5% human AB serum). The sections were tumors but not with important normal tissues. They were then developed using diaminobenzidine-H,O, and counteranalyzed further using peroxidase histochemistry on many stained with Gill's hematoxylin prior to dehydration and more samples of human tumors and normal tissues (including mounting in Permount. cerebellum, cerebral cortex, lung, heart, bone marrow and small bowel). The K1 clone was selected for study and Antibody competition and double-labelling immunofluorescence Both K1 and OC125 antibodies were directly conjugated subcloned twice using limiting dilution cloning. with fluorescein isothiocyanate or tetramethylrhodamine isothiocyanate (Research Organics, Cleveland, OH). For fluoIsotyping and isotype switching The K1 clone originally selected secreted an IgM antibody. rescence competition experiments, OVCAR-3 cells plated on This was detected in the screening assay using anti-mouse IgG 35-mm dishes were cooled to 4°C in BSA-PBS (2 mg/ml BSA) because of cross-reactions with the light chains. This clone was and pre-incubated with 25-50 pg/ml unlabelled competitor converted to an IgG-secreting form using a panning-isotype antibody for 60 rnin (in BSA-PBS), then incubated with 10 switching method to be described in detail elsewhere. Briefly, pg/ml labelled antibody and 25-50 pg/ml unlabelled competibacteriological Petri dishes 100 mm in diameter were coated tor antibody together for another 60 min. After washing, the with 25-50 pgiml affinity-purified rabbit anti-mouse IgG, cells were fixed in formaldehyde. For double-labelling experiments, OVCAR-3 cells were first Fc-specific (Jackson ImmunoResearch) in 50 mM Tris, (pH 9.5). K1 hybridoma cells ( lo7 cells/plate) in 3 ml5% FBS-PBS incubated with rhodamine-labelled K1 antibody for 30 min (10 were incubated undisturbed in this dish at 4°C for 30 min, then pg/ml), then with FITC-labelled OC125 antibody (10 pg/ml) redistributed by tilting and swirling the dish. After another for 30 min. In other experiments, the order was reversed, and 30-60 rnin at 4"C, the suspended cells were removed by in some experiments the 2 antibodies were added together. All pipetting and discarded, and the attached cells were incubated of these double-label methods gave identical results.
+
+
375
OVARIAN CANCER MONOCLONAL ANTIBODY
Live-cell ELISA assay An ELISA assay was designed to determine whether K1 showed additive binding with other antibodies. Cells were incubated with various amounts of antibodies together, and the total binding of all mouse antibodies was detected using peroxidase-labelled goat anti-mouse IgG (25 pg/ml) (Jackson ImmunoResearch). The peroxidase activity in each well in these assays was detected using 100 p1 of 1 mg/ml 2,2'-azinobis(3-ethylbenzthiazoline-6-sulfonicacid) (Boehringer-Mannheim, Indianapolis, IN) in citrate buffer (pH 4.4) with 0.01% H,O,. The reaction was stopped by adding 100 pl 10% SDS, and the color reaction was quantitated using a microplate reader (Molecular Devices, Menlo Park, CA).
RESULTS
Generation of monoclonal antibody K l BALB/c mice were tolerized with a normal human kidney membrane preparation by a cyclophosphamide procedure (Matthew and Sandrock, 1987) designed to minimize the production of antibodies to antigens present on normal human cells. They were then immunized over many weeks with living OVCAR-3 cells. These cells had previously been treated with periodate and conjugated to an anti-mouse IgG antibody; this conjugation was performed to help target these human cells to areas rich in mouse-IgG-covered lymphocytes, areas that would be enriched with cells of the immune system. The spleen cells removed from these mice were then subjected to a panning selection (Wysocki and Sato, 1978) to enrich for splenocytes with surface-bound antibody that reacted with the surface of living OVCAR-3 cells. The selected splenocytes Solid-phase radioimmunoassays and competitive live-cell RIA Serum-free OVCAR-3 cell culture supernatant concentrate represented 1040% of the initial spleen cells, and they were was prepared by using PBS to wash 80% confluent OVCAR-3 then fused with mouse myeloma cells and grown under HAT cultures free of serum-containing medium, then incubating selection. Supernatants from the growing clones of hybridoma cells these cells in a serum-free medium (ABC medium, PAN DATA, Gainesville, FL) for 2 days at 37°C. The supernatant were screened in a single day for production of mouse medium was collected and spun at 10,000 g for 60 min to antibodies reactive with the surface of living OVCAR-3 cells eliminate cell debris, then concentrated by ultrafiltration using using immunofluorescence and ScreenFast chambers (Willinga Diaflo PM30 membrane (Amicon, Danvezs, MA). The ham and Pastan, 1990). Of the original 480 growing clones, 46, which were shown by immunofluorescence to have uniform resultant supernatant was concentrated 50- to 100-fold. surface localization, were selected for the secondary screening Serum-free OVCAR-3 culture supernatant concentrate, step using peroxidase immunohistochemistry. Of these 46, 6 CA125 standard antigen (Centocor, Amersham), and sera were chosen for their reactivity with human tumors but lack of from 16 ovarian cancer patients were assayed for specific reactivity with normal liver, kidney and colon. One of these 6 binding with K1 and OC125 antibodies using a sandwich clones, K1 was saved and further characterized using immunosolid-phase radioimmunoassay. One hundred microliters, con- histochemistry on many normal tissues and human tumors. taining 0.5 pg K1, OC125, or a control mouse antibody, were The initial K1 clone was of the IgM isotype, but this was added to each well of a 96-well plate (12-strip styrene RIA switched to an IgG, isotype using a panning selection method plate) (GIBCO) and allowed to dry overnight at room temper- and subsequently recloned 4 times. ature. The wells were quenched to minimize non-specific adsorption of proteins by the addition of 250 pI of 4% BSA in Immunojluorescence reactivity of K l with cell lines PBS for 1 hr at 37°C. After washing, the wells were incubated Figure 1 shows the homogeneously strong surface binding of with 100 p1 of CA125 standard antigen from 7.5 U/ml to 500 K1 on cultured cells using immunofluorescence. K1 reacted U/ml, undiluted patient serum samples or triplicate dilutions strongly with OVCAR-3, KB, A-1847 and AGS cells, but not of OVCAR-3 supernatant for 1 hr at 37°C. The unbound with SK-BR-3 cells. OC125 also reacted strongly with antigens were removed by washing with 1% BSA-PBS 5 times OVCAR-3 cells, but did not react with A-1847 or AGS cells; it and the wells were then incubated with 100 pI of 1251-labelled did react with SK-BR-3 cells. Table I shows the reactivities of OC125 (6000 cpm/100 pl) (Centocor) or 100 pl of 1251-labelled K1 with many different cultured human carcinoma cell lines in K1 (10,000 cpm/100 pl) (Lmfstrand, Gaithersburg, MD) for 2 this assay, and compares these reactivities with those of the to 12 hr at 23°C. The wells were then washed extensively with previously isolated antibody OC125, reactive with the CA125 1% BSA-PBS. The individual wells were separated from the antigen (Bast et al., 1981; Davis et al., 1986). K1 reacts strongly strip, and radioactivity was counted in a gamma counter with many cell lines derived from ovarian cancers, including OVCAR-2, 3, 4 and 5 , and A-1847 and HTB77 (SKOV3). It (Beckman Gamma 5500B, Irvine, CA). also reacts with HeLa and 2 HeLa variants (HeLa S3 and KB Live-cell RIA was carried out using lZSI-IabelledK1 or cells), originally derived from a cervical carcinoma. While OC125 (70-100,000 cpm/100 pl/well). OVCAR-3 cells plated OC125 reacts with some of these ovarian-tumor cell lines, it in Cell-Tak-coated wells of a 96-well microtiter plate were does not react with A-1847 and reacts poorly with KB cells. incubated with 5% BSA/RPMI 1640 at 37°C for 1 hr, then Neither antibody reacts with A-2780, another ovarianincubated with unlabelled K1 or OC125 antibody as a compet- carcinoma-derived cell line. In addition, K1 reacts well with itor in triplicates from 0.003 ng/ml to 50 pg/ml before and the cell lines AGS and HTB103 (KatoIII) originally derived during the incubation with iodinated antibody (either K1 or from gastric carcinomas, whereas OC125 fails to react with OCl25) (2 hr, 23°C). these lines. K1 fails to react with a number of other cell lines, such as HTB30 (SK-BR-3) which shows a strong reaction with OC125. These data indicate that K1 reacts well with nonTrypsin, proteinase K, neuraminidase and periodate treatments mucinous ovarian-carcinoma-derived cell lines, and with lines Live OVCAR-3 cells were incubated with enzymes for 2 hr derived from cervical and gastric tumors. These data also at 37°C as follows: with neuraminidase (type X, Sigma) at 0.2 illustrate many differences in reactivity between K1 and U/ml in 50% PBS with 50% 0.1 mM acetate buffer (pH 5.0); OC125, indicating that they react with different epitopes. with 2% porcine pancreatic trypsin (type IX, Sigma); or with 0.2 mg/ml proteinase K, in 0.1 M Tris-HCI, 50 mM CaCl,,.pH Immunoperoxidase histochemistry in normal tissues 7.8. Protease-treated cells were reattached to poly-L-lysineK1 and OC125 were also examined for reactivity with treated dishes prior to immunofluorescence examination with normal human and monkey tissues using peroxidase immunoK1 or OC125 antibodies. Cells were treated with periodate histochemistry in cryostat sections. Figure 2 shows examples of using 10 mM NaIO, in 100 mM acetate buffer, pH 4.5 for 15,30 the localization of K1 and OC125 in normal human mesothelia. Figure 3 shows the localization of K1 and OC125 in other and 60 min at room temperature.
376
CHANG ET AL. TABLE I - IMMUNOFLUORESCENCE LOCALIZATION OF K1 AND OC125 ON HUMAN CULTURED CELL LINES
Cell line
OVCAR-2 (ovarian ca.) OVCAR-3 (ovarian ca.) OVCAR-4 (ovarian ca.) OVCAR-5 (ovarian ca.) A-1847 (ovarian ca.) HTB77(SKOV3) (ovarian ca.) A-2780 (ovarian ca.) HTB33 (ME-180) (cervical ca.) HeLa (cervical ca.) HeLa S3 (cervical ca.) KB (cervical ca.) AGS (CRL 1739) (gastric ca.) ' HTB103 (Kato 111) (gastric'ca.) ' A431 (epidermoid ca.) HT-29 (colon ca.) HTB30 (SK-BR-3) (breast ca.) HTB20 (BT-474) (breast ca.) MDA-MB-468 (breast ca.) MCF-7 breast ca.) DU145 [prostate ca.)
++ ++ het ++ het +
++
+ + het (50% -) +++ het ++++ het
++S
+- + het (30%) ++ het (10% ++) -
++ ++ +++ ++ ++
-
++ ++
- ( < 5 % ++) -
-
+++ (
