Br. J. Cancer (1981) 43, 1 SOMATIC-CELL HYBRIDS PRODUCING ANTIBODIES AGAINST CEA

Monoclonal antibodies to carcinoembryonic antigen (CEA) promise improved specificity for the measurement of this widely expressed human cancer antigen. A mouse monoclonal antibody binds weakly to CEA in perchloric acid extracts of tumour but strongly to CEA similarly isolated from serum, and its spectrum of cancer detection differs from conventional antisera. METHODS in somatic cell hybridization, introduced by Kohler & Milstein (1975), have made it possible to produce mono-clonal antibodies in tissue culture. Whilst the method has proved entirely successful for the production of antibodies against particulate antigens such as those on the cell surface (Trucco et al., 1978) and viruses (Koprowski et al., 1977), it has only recently been successfully applied to soluble tumour markers such as CEA (Accolla et al., 1979), hCG (Stahli et al., 1980) and alpha-foetoprotein (Tsung et al., 1980). Monoclonal antibodies against CEA are of special interest, since they promise to overcome particular difficulties associated with the antigenic heterogeneity of this glycoprotein (Rogers, 1976). Conventional antisera raised to purified CEA, even after removal of antibodies to known cross-reactive antigens such as NCA, contain a mixed population of antibodies (Keep & Rogers, 1979; Rogers & Keep, 1980). Different batches of antisera consequently vary in binding affinity for different forms of CEA or different antigens in the preparation , and this may impair or alter clinical specificity of the CEA assay. Mono-clonal anti-CEA on the other hand should be able to discriminate between small anti-genic differences expressed by the CEA complex, and provide tools to characterize CEA immunologically, to optimize and stabilize the specificity of the CEA assay and to test the concept that some antigenic 1 determinants on CEA or in CEA preparations may be of more clinical relevance as tumour markers than others (Rogers, 1976). Here we describe the first mono-clonal antibodies with a high specificity for circulating CEA. In this study BALB/c mice were immunized with 40 ,ug of highly purified CEA (R42), followed by a similar boost 5 weeks later (R42 was isolated by perchloric acid extraction of a pool of 6 metastatic colonic tumours as described previously Rogers et al., 1976). After removal of red cells by lysis, 108 spleen cells from these mice were fused with 107 mouse myeloma cells (P3-NSL/lAg 4-1, Flow Laboratories) in polyethylene glycol according to the method of Kohler & Milstein (1975) and distributed into 180 microtitre wells. The next day the medium was replaced with medium …


METHODS in somatic cell hybridization, introduced by Kohler & Milstein
, have made it possible to produce monoclonal antibodies in tissue culture. Whilst the method has proved entirely successful for the production of antibodies against particulate antigens such as those on the cell surface (Trucco et al., 1978) and viruses (Koprowski et al., 1977), it has only recently been successfully applied to soluble tumour markers such as CEA (Accolla et al., 1979), hCG (Stahli et al., 1980) and alpha-foetoprotein (Tsung et al., 1980).
Monoclonal antibodies against CEA are of special interest, since they promise to overcome particular difficulties associated with the antigenic heterogeneity of this glycoprotein (Rogers, 1976). Conventional antisera raised to purified CEA, even after removal of antibodies to known crossreactive antigens such as NCA, contain a mixed population of antibodies (Keep & Rogers, 1979;Rogers & Keep, 1980). Different batches of antisera consequently vary in binding affinity for different forms of CEA or different antigens in the preparation, and this may impair or alter clinical specificity of the CEA assay. Monoclonal anti-CEA on the other hand should be able to discriminate between small antigenic differences expressed by the CEA complex, and provide tools to characterize CEA immunologically, to optimize and stabilize the specificity of the CEA assay and to test the concept that some antigenic 1 determinants on CEA or in CEA preparations may be of more clinical relevance as tumour markers than others (Rogers, 1976). Here we describe the first monoclonal antibodies with a high specificity for circulating CEA.
In this study BALB/c mice were immunized with 40 ,ug of highly purified CEA (R42), followed by a similar boost 5 weeks later (R42 was isolated by perchloric acid extraction of a pool of 6 metastatic colonic tumours as described previously Rogers et al., 1976). After removal of red cells by lysis, 108 spleen cells from these mice were fused with 107 mouse myeloma cells (P3-NSL/lAg 4-1, Flow Laboratories) in polyethylene glycol according to the method of Kohler & Milstein (1975) and distributed into 180 microtitre wells. The next day the medium was replaced with medium containing hypoxanthine, aminopterin and thymidine, in which only hybrid cells remained viable. Supernatants from wells containing actively growing hybrid cells were then tested for anti-CEA activity by means of binding to 1251-labelled CEA using a double-antibody radioimmunoassay method. Antibody producing cells were cloned by dilution or single-cell transfer and viable cells propagated. 107 anti-CEA producing cells were injected i.p. into BALB/c mice to obtain ascites fluid MA-1 and serum MS-1.
Binding experiments with MA-1 and MS-1 were carried out using radio-labelled 1251-CEA (R42). This was prepared by labelling 20 ,ug of the CEA protein with 5 mCi of IMS 30, according to a modified method of Greenwood et al. (1963). The iodinated CEA was purified on Sephadex G-200 and the fractions with maximum binding to our conventional anti-CEA serum selected. The labelled CEA was used at a dilution of 1:200, equivalent to 200,000 ct/min. Binding studies with MA-1 and MS-1 showed that at a dilution of 1:1000 they were able to bind 110% and 170% of 0 3 ng of 1251-CEA (50 pi of diluted label) using an optimum dilution of 1:4 for the precipitating antibody (goat-anti-mouse). Maximum binding at a dilution of 1:100 was 2700, in contrast with the much greater binding (700 %) achieved at this dilution with both conventional anti-CEA (PKIG) and the monoclonal sera described by Accolla et al. (1979). The percentage binding of MA-1 at 1:1000 to 1251-CEA was found to be constant (I1I%) and independent of the concentration of label (Table I). This indicates that the MA-1 antibodies bind to a small sub-population of CEA molecules present in the label, or alternatively that the binding affinity was so low that as more label was added more of it bound to the antibody. These interpretations are consistent with other studies on the competitive binding of purified unlabelled CEA. In these experiments increasing amounts of standard CEA (R43, prepared in a similar manner to CEA-R42 but from a different pool of metastatic colonic tumours) were incubated with a mixture of monoclonal serum Comparative inhibition of 12511 CEA-MS-1 by CEA purified from perchloric acld extracts of 3 different pools (each comprising specimens from 6 patients) of liver metastases from colonic tumour (A, B, C) from a saline extract of a liver metastases from a single patient with a rectal tumour (D) and from a perchloric acid extract of a pool of 11 sera from patients with cancer and a raised CEA value (E). The CEA concentrations were measured in ng/ml by our conventional CEA assay. Curve C shows the inhibition in the range 360-10,000 ng/ ml of CEA (R43) which has been used as standard for the assays developed with monoclonal antibodies MS-1 and MA-1. For conx-enience an arbitrary unit has been adopted whereby 100 u/ml of MS-1 binding CEA produce 500% "displacement" of bound label and are equix alent to 8600 ng/ml of conventional CEA.
at 1:1000 dilution, the radiolabelled CEA and a commercial rabbit anti-mouse precipitating antibody (Dako Z109 at 1:40 dilution). It was demonstrated that a concentration of 8600 ng/ml of CEA was required to produce 5000 "inhibition" of bound label (Figure). The concentrations of CEA in ng/ml for these studies were measured by our routine CEA assay, using a conventional absorbed anti-CEA serum PK1G.
In comparison, a similar inhibition in our routine assay is produced by only 40 ng of CEA. Further studies have shown that 3 additional preparations of purified tumour CEA are comparatively poor inhibitors of the monoclonal MS--1 25J CEA system, requiring concentrations of CEA in excess of 8000 ng/ml to produce 5000 displacement of bound label (Figure). Because of the poor inhibition by CEA preparations an arbitrary has been adopted for use in radio assay with MS-I and MA-1. Thus MS-1 binding CEA are defined amount producing 50% displac bound label.
Preliminary radioimmunoassai ments developed with MS-1 have marked loss of binding activity oi a saline extract of liver meta, rectal tumour at 85°C for 30 mi II), demonstrating that the deti with which MS-1 binds is present heat-labile and heat-stable n Further studies with extracts tumours are required to see whe is general, though similar findi] recently been obtained on a v CEA extracts, using conventional (Keep & Rogers, 1979). Like con anti-CEA, MS-1 is also unable t guish different chemical forms separated by concanavalin A chromatography. In these pr experiments, liver metastases of tumour was extracted (a) with acid, (b) with saline and (c) with s subsequently heat-treated at 85' min. Each extract was applied to of con A-Sepharose and eluted v ing CEA sodium acetate containing IM NaCl (Fracat-treated tion 1), 01M sodium borate in 0dIM ic rectal phosphate buffer pH 6-0 (Fraction 2A); fractionthe remaining fractions 2B, 3 and 4 were affinity eluted respectively with 2%, 10% and 20% methyl glucoside in the acetate Heat-buffer as described elsewhere (Keep et al., treated 1978). Each fraction was then assayed in a extract double-antibody assay developed with 47 6 MS-1. The presence of MS-1 binding CEA 205 was demonstrated in all fractions ( Table   1o04 II) suggesting that the MS-1 binding 16 0 determinant is not affected by hetero-990 geneity of the carbohydrate residues in 55r0 CEA, and is probably situated in the internal part of the CEA glycoprotein The CEA (Rogers, 1976). Studies with several perbtained by chloric acid extracts of tumour CEA have body assay IS-1.
shown that the Con A-binding profile of MS-i-binding CEA is similar to that standard obtained with conventional CEA where unitage very little non-binding CEA is found (Keep immunoet al., 1978). However, in contrast to the 100 u of results with perchloric acid extracts, a as that significant fraction of MS-i-binding CEA, ement of isolated by saline extraction of tumour at neutral pH, was not bound by Con Ay experi-Sepharose (Table II). This difference has shown a been seen previously with foetal colon n heating CEA, but not with tumour CEA, using stases of conventional antisera (Keep et al., 1978) in (Table and further suggests that MS-1 may be ,erminant specific for a unique sub-population of t on both CEA molecules, possibly of foetal type. nolecules. Nine normal-tissue extracts tested, inof other cluding a lung, a liver, a spleen and 6 -ther this colon extracts all from separate indings have viduals, contain very little MS-i-binding -ariety of antigen with values ranging from 1 8-4-9 1 antisera u/g of tissue. It is significant that extracts ventional of normal spleen and lung, which contain to distina high content of the normal crossof CEA reacting antigen NCA, do not inhibit the affinity MS-1-1251-CEA binding any more strongly eliminary than the normal colon extracts, showing ' a rectal that the specificity of MS-1 is probably )erchloric not directed to NCA. ,aline and In contrast to CEA extracted from°C for 30 tumour tissue, CEA isolated by perchloric a column acid extraction of a pool of II sera from vith 01m patients with advanced colonic cancer "displaced" the MS-i-bound label easily (Figure). In this case the serum CEA, measured as 1000 ng/ml on our routine assay, produced "displacement" in the monoclonal system in excess of 80%. This suggests that serum CEA either has a larger sub-population of MS-1-binding CEA or has a higher overall binding affinity for MS-I than is found in CEA extracted from tumours. These differences could result from chemical modifications of CEA by extraction from the serum of cancer patients or by the predominance in the serum of CEA which has been degraded or altered in the liver. The latter is more likely, since double-antibody radioimmunoassays developed with monoclonal sera have confirmed the high specificity of both MS-1 and MA-1 for a CEA component in untreated sera from patients with various cancers. Thus 30/102 (29%) of samples from patients with colon cancer, 16/42 (38%) of samples from patients with rectal cancer, 16/31 (51%) of samples from patients with gastric cancer and 12/34 (35%) of patients with prostatic cancer produced raised values (> 15 u/ml). Of 144 samples from normal subjects, only one was raised. Compared to CEA assays with conventional antisera, the "monoclonal" assays were much less sensitive to CEA extracted from tumour tissue.
The results described here demonstrate that monoclonal antibodies bind to an antigen present in highly purified CEA from various tumours and with our radiolabelled CEA as used in our routine assay. They are also consistent with the concept that different immunological forms of CEA exist which can be distinguished by appropriate antisera. However, in view of the immunological heterogeneity of CEA and the polyspecific nature of many conventional anti-CEA sera (Rogers & Keep, 1980) we cannot exclude the possibility that these monoclonal sera are recognizing a unique antigen not related to CEA, but present in purified CEA preparations. If this is the case the antigen is clearly related to malignancy since it is prevalent in the serum of patients with cancer but not detected in significant amounts in sera from normal individuals and extracts of normal lung, spleen, liver and colon. The study suggests that the deployment of different monoclonal sera raised against purified CEA may help to establish whether some antigenic determinants in these preparations are of particular clinical interest. Support of the Medical Research Council is gratefully acknowledged.