Colorectal carcinoma antigens detected by hybridoma antibodies

Somatic Cell Genetics, Vol. 5, No. 6, 1979, pp. 957-972

Colorectal Carcinoma Antigens Detected by Hybridoma Antibodies Hilary Koprowski, Zenon Steplewski, Kenneth Mitchell, Meenhard Herlyn, Dorothee Herlyn, and Peter Fuhrer The Wistar Institute of Anatomy and Biology, 36 Street at Spruce, Philadelphia, Pennsylvania 19104 Received 17 July 1979

One of us (Hilary Koprowski) had the privilegeand pleasure of collaborating with Boris Ephrussi during the latter's stay in the United States. In collaboration with Vittorio Defendi, we studied the expressionof polyoma-inducedantigen in the second somatic cell hybrid ever made (1) and of tumorigenicityin other hybrids between polyoma-transformedand normal mouse cells (2). It has been both astonishing and amazing to watch the progress made during the subsequent two decades in the field of somatic cell hybridization, and we deeply regret that Boris Ephrussi, who contributed so much to this progress, can no longer share our enthusiasm.

Ahstract--Hybridoma cells which secrete colorectal carcinoma-specific antibodies have been produced and used to study the antigenic structure of these tumor cells. Nineteen antibodies have been studied in detail, and 15 of these are colorectal carcinoma specific. Only two antibodies reactive with carcinoembryonic antigen (CEA) have been discovered and five other antibodies that react with distinct epitopes on the cell surface have been defined. Several antigens with distinct molecular characteristics have been shown to exist by use of hybridoma antibodies. Six hybridoma antibodies have been shown to mediate antibody-dependent cell-mediated cytotoxicity (ADCC).

INTRODUCTION W e have recently shown ( 3 - 5 ) that monoclonal antibodies produced by hybridomas derived from the fusion of mouse m y e l o m a cells with splenocytes of i m m u n i z e d mice display a specificity for surface antigen(s) of h u m a n m e l a n o m a and colorectal carcinoma. Precipitation of the t u m o r cell components by melanoma-specific h y b r i d o m a antibodies led to the observation that 957 0098-0366/79/1100-o9575o3.oo/0 9 1979 Plenum PublishingCorporation

958

Koprowski et al.

two antigenic moieties characterized by different molecular weights are detectable on human melanomas (Fuhrer et al., submitted for publication). Previously, we succeeded in establishing two hybridoma clones with specificity for human colorectal carcinoma (5). In this report we describe a large number of recently established hybridomas which secrete antibodies with specificity for the surface antigens of colorectal carcinoma cells. Thus, it is now possible to study a large panel of hybridomas which secrete homogeneous antibodies each of which reacts with antigens present on the cell surface of a panel of colorectal carcinomas. MATERIALS AND METHODS Colorectal Carcinoma Cell Lines. Eight colorectal carcinoma cell lines were obtained for these studies from A. Leibovitz of the Scott and White Clinic, Temple, Texas. These cell lines were established from freshly isolated, histologically confirmed surgical specimens and were classified into groups according to morphology and karyotype and were shown to synthesize carcinoembryonic antigen (CEA) (6). Other Human Cell Lines. Carcinoma cell lines derived from tumors other than those of colon or rectal origin were also obtained from the Scott and White Clinic. These included lines of breast (SW1403) and lung (SW1271) carcinomas. Human lung carcinoma cell lines A-549 and MBA9812 and normal human lung fibroblast lines HS0853 and HS0618 were obtained from the Cell Culture Laboratory, Naval Biosciences Laboratory, Oakland, California. Normal human fibroblast cell lines Flow 7000 and MRC-5 were obtained from Flow Laboratories, Rockville, Maryland. Sarcoma, astrocytoma, and lymphoid cell lines previously described (3, 5) were used with some melanoma lines established at The Wistar Institute (WM series) and with others obtained through the courtesy of Dr. Lloyd Old, Sloan Kettering Cancer Center, New York, New York (SK series). Production of Hybridomas. BALB/c mice were immunized for a secondary response as described (3). Three days prior to their sacrifice, mice were given an intravenous injection of 1 x 1 0 6 immunizing cells (SW1116); after sacrifice, a spleen cell suspension was prepared in the usual manner (7). Immune splenocytes were fused with the variant 653 of the myeloma cell line P3 x 63 Ag8 (established by John F. Kearney et al., J. Immunol., in press) as previously described with only one variation. Prior to fusion, cells were washed in Ca 2+- and Mg2+-free medium, and the fusion was performed in the absence of Ca and Mg ions. Hybrids were selected in a medium containing hypoxanthine/aminopterin/thymidine (HAT) after fused cells were seeded in wells of tissue culture plates (Linbro FB-16-24-TC). Approximately 20 days after fusion, single colonies were picked from each well and processed as

Colorectal Carcinoma Antigens

959

described previously (5). Two of the hybridoma antibodies included in these studies (1116-56-2 and 1083-17-1A) have already been described (5). Determination of Immunoglobulin Isotype. Isotype determinations were made by Ouchterlony immunodiffusion. Diffusion was carried out in 1% agarose in 0.02 M~Tris HC1 buffer, pH 8.0, and 0.1 M NaCI. Monospecific antisera were purchased (Bionetics, Bethesda, Maryland) and tested prior to use, to ensure specificity, with a panel of myeloma proteins. At the limits of sensitivity, the antisera were found to have the claimed specificity. Additive and Direct Binding Radioimmunoassays. All assays were performed by a previously described procedure (3). Target cells (5 x 105) were incubated with the hybridoma antibody for 1 h at room temperature. The cells were washed three times and then incubated with a second antibody. After 1 h the cells were again washed three times and then incubated with radiolabeled (t25I) rabbit antimouse F(ab')2. After three final washes, the amount of radioactivity present in the pellet was assessed by counting in a Packard gamma spectrometer. When direct radioimmunoassays were performed, the second incubation with hybridoma antibody was omitted. Antibody-Dependent Cell-Mediated Cytotoxicity (ADCC) Test. Spleen cells of normal (unimmunized) 6- to 12-week-old CBA mice were used as effector cells. For the preparation of single cell suspensions, spleens were pressed through a wire 60-mesh sieve into RPMI-1640 medium containing 5% fetal calf serum (FCS) and buffered with HEPES, pH 7.2, at a concentration of 5 raM. This medium was used throughout the test. Target cells were labeled with [51Cr]NazCrO4 (New England Nuclear, Boston, Massachusetts) for 15-18 h with 50 ~Ci 51Cr/106 cells and were added to wells of U-bottomed tissue culture microtiter plates at a concentration of 2 x 104 cells/50 ul medium. They were then covered with 50 #1 of supernatant from hybridoma cell cultures. Supernatant from P3 x 63 Ag8 (P3) cells was used as a control. Target cells were incubated with antibody for 2 h at 4~ washed once, and exposed to a suspension of effector cells at a ratio of 100 effector cells to one target cell. After further incubation of the plates for 6 h at 37~ in a humidified atmosphere of 5% CO2 and 95% air, half the supernatant was removed for determination of the released radioactivity. Maximal SlCr release was determined for each well by overnight incubation of the mixture with ZAP-Isotonin II (Coulter Diagnostics, Inc., Hialeah, Florida), final concentration 3.75%. In order to determine the level of spontaneous 51Cr release by the target cells, some plates were incubated with antibody and target cells in the absence of lymphocytes. Percentage of cytotoxicity (for both test and control samples) was calculated with the following formula: % cytotoxicity =

test or control release - spontaneous release maximal release - spontaneous release

x 100

900

Koprowski et al.

The percentage of ADCC was calculated according to the equation: % ADCC = % cytotoxicity of text sample - % cytotoxicity of control sample The Student's t test was used to calculate whether differences in cytotoxicity between the test and control samples were statistically significant. Radioiodination of Colorectal Carcinoma Cells. Cell surfaces were radiolabeled with iodine 125 by the lactoperoxidase method (8). Adherent monolayers of colorectal carcinoma cells were labeled in situ. Radiolabeled cells were washed three times with phosphate-buffered saline (PBS) and solubilized by incubation for 15 min at 4~ in 1 ml of NET buffer (0.15 M NaCI, 0.05 M Tris, 5 mM EDTA, 0.2 mM phenylmethylsulphonyl fluoride, pH 7) containing 0.5% NP-40 (Shell Chemical Co., London, England). Solubilized material was centrifuged at 100,000 g for 30 min and absorbed serially with two 0.5 ml volumes of packed Staphylococcus aureus Cowan I (SaCI) previously washed with NET buffer containing 0.05% NP-40 and 1 mg/ml bovine serum albumin (BSA) (Pentex, Miles Laboratories, Inc., Elkhart, Indiana). SaCI-cleared preparations were stored at -70~ Radiolabeled cell surface antigens were precipitated from NP-40 solubilized cells by hybridoma antibodies. Antigen precipitation was accomplished by mixing 100 #1 of SaCI-cleared, solubilized cell supernatant with 200 #1 of hybridoma antibody-containing culture supernatant. Following incubation for 30 min at 4~ hybridoma antibody reaction mixtures were added to 50-#1 volumes of packed SaCI to which 100/~g of a purified rabbit anti-mouse immunoglobulin (Ig) had previously been added. SaCI pellets were then suspended into reaction mixtures, incubated for 1 h at 4~ and centrifuged at 10,000 g for 5 min. The SaCI pellets were washed three times with 0.05% NP-40 NET, transferred to new tubes and eluted with 50 #1 of Laemmli reducing buffer (9). Biosynthetic Labeling of Colorectal Carcinoma Cells. Cells were grown for 48 h in methionine-free medium 199 (GIBCO, Grand Island, New York, New York) supplemented with 50 #Ci [75Se]selenomethionine (Amersham Searle, Arlington Heights, Illinois). Cell monolayers were then washed and solubilized as described for iodinated cells. Cell lysates were cleared and processed as described above. Radioautography. Radioautography was performed by exposure of RP x omat film for 2-6 days with a DuPont Cronex Lightning plus intensifying screen. Film was developed in a standard manner.

Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis (SDSPAGE). SDS-PAGE was performed by the method of Laemmli (9). Dissolved antigens were electrophoresed on 10% polyacrylamide SDS gets (acrylamide-bis ratio 38:1). Apparent molecular weights of precipitated

Colorectal Carcinoma Antigens

961

antigens were determined by comparison of the mobilities of the observed bands with the mobilities of a set of proteins whose molecular weights have been well characterized. Assessment of Binding to CEA. A CEA assay kit was purchased from H o f f m a n n - L a Roche, Nutley, New Jersey. Precipitation of the radiolabeled C E A was attempted with each hybridoma antibody and with the standard H o f f m a n n - L a Roche goat anti-CEA antiserum. Precipitation of hybridoma antibodies was achieved by the addition of rabbit anti-mouse Ig. Goat antibody was precipitated by H o f f m a n n - L a Roche zirconyl phosphate gel. Precipitated antigens were solubilized and fractionated by S D S - P A G E to confirm the molecular weights of the antigens precipitated. RESULTS Spleen cells from a mouse immunized with S W l 1 1 6 cells were fused with cells of myeloma cell line P3 • 63 Ag8 (653 variant). Of 76 hybridoma colonies isolated (Table 1), 50 were found to secrete antibody which bound to various kinds of human cells. Of these 50 "human-specific" antibodies, 19 showed a specificity for human tumor cells. The binding specificities of the antibodies secreted by the 19 hybridomas and by two hybridomas, 1116-56-2 and 1083-17-1A, produced previously (5) are shown in Table 2. It may be seen that eight hybridoma antibodies bound to cells of all colorectal carcinomas and that two of these also bound to melanoma cells and one (1116NS-20) bound to nine melanomas, one astrocytoma, and one breast carcinoma. Three other hybridomas secreted antibodies that reacted with cells of seven out of eight colorectal carcinomas, and one of these also reacted with cells of a melanoma cell line. The remaining ten hybridomas secreted antibodies that bound to cells of six, five, or four out of eight colorectal carcinomas; one of these also reacted with cells of one out of ten melanomas tested. None of the 21 antibodies bound to lung carcinoma cells,

Table 1. Specificitiesof hybridoma antibodies produced by fusion of cells from a single mouse

spleen with mouse myelomavariant P3 • 63 Ag8 (653)

Isotypeproduced IgM IgG1 IgG2a IgG2b IgG3 None detected TOTAL

Total colonies isolated 23 19 2 6 6 20 76

Colonies secreting antibody with specificities for colorectal carcinoma cells only

Coloniessecreting antibodieswith specificity for human tumor cells

6

2

9 0 0 0 0 15

2 0 0 0 0 4

962

Table 2.

Koprowski et ai.

Bindingof hybridoma-secretedantibodies to colorectalcarcinoma and other cell lines Carcinoma cell lines

Fibroblasts

SV40 IsoColoMela- Astro- Sar- transAntibody type CEAa rectal Lung Breast noma cytoma coma formed Fetal Adult llI6NS-3d 3,M Pos. 8/8 c 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 ll16NS-33a 3,M Neg. 8/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 1116NS-22 yl Neg. 8/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 1116-56-2b 3,1 Neg. 8/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 1083-17-1Ab 3,1,3,2a Neg. 8/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 ll16NS-33b 3,M Neg. 8/8 0/3 0/1 2/i0 0/2 0/2 0/1 0/2 0/2 1116NS-29 ~1 Neg. 8/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 1116NS-20 yl Pos. 8/8 0/3 1/1 9/10 1/2 0/2 0/1 0/2 0/2 llI6NS-38e ~1 Neg. 7/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 1116NS-43 ~1 Neg. 7/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 lll6NS-10 7M Neg. 7/8 0/3 0/1 1/10 0/2 0/2 0/1 0/2 0/2 1116NS-38c yM Neg. 6/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 Ill6NS-3a 3,1 Neg. 6/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 Il16NS-38a TM Neg. 6/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 lI16NS-38b 3,M Neg. 6/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 1116NS-36 F 3,1 Neg. 5/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 lll6NS-3b 71 Neg. 5/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 1116NS-19 ~1 Neg. 5/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 lI16NS-52a 3,1 Neg. 5/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 ll16NS-52b 3,1 Neg. 5/8 0/3 0/1 1/10 0/2 0/2 0/1 0/2 0/2 1116NS-15 3,M Neg. 4/8 0/3 0/1 0/10 0/2 0/2 0/1 0/2 0/2 aCEA = carcinoembryonicantigen; assayed with Hoffmann-LaRoche kit. bAntibodies secretedby hybridomasproducedby fusion with 3,1secretingP3 • 63Ag8 myeloma cells (5). CNumber of cell lines showingbinding/total number of cell lines tested. sarcoma cells or any fibroblast line tested. The rather remarkable specificity of the hybridoma antibodies observed in the direct binding test was further confirmed by the results of absorption studies with hybridoma clone 108317-1A. In addition to results already published (5), antibody produced by clone 1083-17-1A was extensively tested for absorption to a variety of noncolon tumor lines such as renal, lung, bladder, and larynx carcinomas; astrocytomas; melanomas; and other tumor cell lines as well as by normal and EB virus-transformed cells (Knut et al., to be published). None of these or any xenogeneic tissue decreased the reactivity of 1083-17-1A antibody for colorectal carcinomas. Reaction of Hybridoma Antibodies with CEA. Only two antibodies reacted with the iodinated CEA preparation (1116NS-3d and 1116NS-20) (Table 2). The molecular nature of the precipitated antigens differed in each case: l l l 6 N S - 3 d precipitated molecules with the characteristic molecular weight of intact CEA (180,000 daltons) (10); 1116NS-20 precipitated a complex group of antigens, including species with lower molecular weights in the 130,000- and 60,000-dalton region as well as 180,000-dalton molecules.

Colorectal Carcinoma Antigens

963

Isotypic Distribution of Monoclonal Antibodies. With the exception of 1083-17-1A, which reacted with human cells and secreted IgG2a, no IgG2a, IgG2b, or IgG3 showed specificity for human tumor cells. As shown in Tables 1 and 2, eight of 23 IgM producing hybridomas were specific for human tumor cells and six of these were specific for colorectal carcinoma. In the case of IgG1 producing hybridomas, 12 of 20 were specific for human tumor cells and 10 of these were specific for colorectal carcinomas. Hybridoma clone 1083-17-1A was obtained by the fusion of mouse splenocytes with the original P3 • 63 Ag8 myeloma cells; the IgG1 which this hybridoma secreted is therefore of mouse myeloma origin. Clone 1083-17-1A remains the only anticolorectal carcinoma antibody of the IgG2a sublcass which shows a specificity for antigenic determinants expressed by cells of colorectal carcinoma (see above). Classification of Colorectal Carcinomas by Reactivity of their Cells with a Panel of Hybridoma Antibodies. As shown in Table 3, antigenic determinants present on the cell surface of three out of eight colorectal carcinomas were recognized by all hybridoma antibodies; these tumors were

Table 3. Classification of colorectal carcinoma cell lines Binding in RIA" of hybridorna supernatant to cells of colorectal carcinoma Antibody 1116-56-2 c 1083-17-lA c lll6NS-3d

1116NS-33a 1116NS-22 ll16NS-33b 1116NS-29 1116NS-20 ll16NS-38e 1116NS-43 lll6NS-10 1116NS-38c lll6NS-3a ll16NS-38a ll16NS-38b 1116NS-36 lll6NS-3b 1116NS-19 1116NS-52a ll16NS-52b 1116NS-15

Tumor group

S W l l I 6 b SW1222 SW1463 1174 3357 2987 1529 4478 2622 5247 4588 1704 1974 3621 2t93 5585 2986 3128 3428 1052 6121 5449 5219 966 I

1571 4275 5297 2274 6819 3590 8302 4099 2603 1194 3014 3916 2333 3580 3652 788 1616 3763 6404 5212 1270 I

1336 4585 3387 2176 5128 2221 2542 3071 2154 1447 2258 2283 4426 2258 2621 1886 1110 2664 3205 3318 885 I

SW403

SW948

SW1083

SW837

SW1345

2186 5226 873 650 1809 956 5253 2029 2025 3267 2486 2499 2222 2378 2657 975 0 3686 3214 3618 0 ]I

1556 3331 763 2933 1158 1044 4553 3620 1996 806 1827 1994 1986 1591 2124 0 0 2238 3473 3336 0 II

2173 4684 1844 500 2839 1308 5048 3750 640 1331 620 1004 0 1202 1219 820 570 0 0 0 0 III

500 3555 1212 II12 2137 1495 2672 2206 0 1417 1606 0 784 0 0 0 702 0 0 0 0 iII

1447 5121 2407 1607 4068 2160 3758 4137 1391 0 0 0 0 0 0 0 0 0 0 0 732 III

aRIA - radioimmunoassay; values >_ 500 cpm only. bSW1116 cell line was used for immunization of mice. CAntibodies secreted by hybridomas produced with P3 • 63Ag8 myeloma cell line (5) and cells from animals immunized with SW1116 (1116-56-2) or SW1083 (1083-17-1A).

Low b Low Low High High High Intermediate

1116NS-3d 1116-56-2 lll6NS-3a 1116NS-19 lI16NS-52a 1083-17-1A 1116NS-33b

-

lll6NS-3d

+ -

+c

1116-56-2

+ +

-

+

lll6NS-3a

+ +

+ + +

1116NS-19

3,1 "y1 "~'1 ~,l,,y2 a 3;1 3'I

16.3" 15.1 18.2 22.1 18.0 16.8

SW1116

13.3 12.7 10.5 13.9 9.3 N.T)

SW1222 10.6 6.7 7.7 22.1 13.1 7.3

SW948

Colorectal carcinoma

a% 5tCr release c a l c u l a t e d as described in text. bN.T.-not tested.

II16NS-3a 1116NS-19 1l16NS-52a 1083-17-1A 1116NS-29 1l16NS-52b

Antibody

Isotype 9.2 0 0 7.1 0 0

SW837 0 0 0 0 0 0

Lung carcinoma SW1271

0 0 0 0 0 0

0 0 0 0 0 0

WM9

Melanoma SW691

A D C C a g a i n s t cells of:

0 0 0 0 0 0

Astrocytoma SW1088

+

+ + + + +

0 0 0 0 0 0

+ + -

1116NS-33b

0 0 0 0 0 0

Flow 7000

Fibroblasts W138

1083-17-1A

T a b l e 5. Biological a c t i v i t y of a n t i c o l o r e c t a l c a r c i n o m a a n t i b o d i e s as m e a s u r e d b y A D C C

+ +

+ + + +

llI6NS-52a

~ by RIA. bLow-avidity antibodies bound at a level of