A tumor-associated fibronectin isoform generated by alternative splicing of messenger RNA precursors.

Fibronectin (FN) represents the mixture of a number of structurally different molecules (isoforms) whose make-up varies depending on the FN sources. FN from cultured transformed human cells has a very different isoform composition with respect to its normal counterpart. In fact, SV-40-transformed WI-38VAI3 human fibroblasts produce high levels of a FN isoform (B-FN) which is very poorly expressed in their normal, WI-38, counterpart. We have recently demonstrated that the B-FN isoform derives from a differential splicing pattern of the FN primary transcript which leads, in transformed cells, to a high level expression of the exon ED-B (Zardi, L., B. Carnemolla, A. Siri, T. E. Petersen, G. Paolella, G. Sebastio, and F. E. Baralle. 1987. EMBO (Eur. Mol. Biol. Organ.) J. 6:2337-2342). Here we report on the production and characterization of a monoclonal antibody (BC-1) which recognizes an epitope within the protein sequence coded for by the ED-B exon. This monoclonal antibody makes it possible to carry out immunohistochemical analysis of the distribution of the ED-B-containing FN isoform (B-FN) in human tissues. The results show that while in normal, adult, human tissues total FN has a widespread distribution, the B-FN isoform is restricted only to synovial cells, to some vessels and areas of the interstitium of the ovary, and to the myometrium. On the contrary, the B-FN isoform has a much greater expression in fetal and tumor tissues. These results demonstrate that, in vivo, different FN isoforms have a differential distribution and indicate that the B-FN isoform may play a role in ontogenesis and oncogenetic processes.

The nature of the immunoreactive lipotropins in human plasma and tissue extracts.

This study was designed to establish definitively the nature of immunoreactive lipotropin (IR-LPH) in human plasma and tissue extracts. Using gel filtration, gel filtration under denaturing conditions, cationic exchange chromatography, immunoprecipitation, and radioimmunoassay, we have studied normal and tumorous human pituitaries, ectopic ACTH- and LPH-secreting tumors, plasma from normal subjects before and after dexamethasone administration, and plasma from patients with primary adrenal insufficiency and pituitary and nonpituitary ACTH- and LPH-secreting tumors. Except in the plasma and tumors of occasional patients with ectopic ACTH syndrome, the smallest IR-LPH appears to be lambda-lipotropin (lambdaLPH), which is often the predominant and occasionally the only IR-LPH present. The other major peptide appears to be betaLPH, a 91-amino acid molecule that contains lambdaLPH as its 1-58 sequence. Larger immunoreactive materials were observed in some specimens, but the "big" LPH in one plasma was shown to be lambdaLPH bound to IgG.The weak melanocyte-stimulating activity of LPH suggests that ACTH may be the principal pigmentary hormone in man. The fact that lambdaLPH, rather than betaLPH, is the predominant form in plasma suggests that the enkephalin-endorphin opiate peptides, which are contained in the "missing" 59-91 sequence from the betaLPH precursor of lambdaLPH, may be secreted in parallel with ACTH under both physiological and pathological conditions in man.

Correction of modal DNA values obtained by microspectrophotometry and tests for their shifts.

The technical problems in the application of statistical analyses of the nuclear DNA contents measured by plug or two-wavelength Feulgen microspectrophotometry were discussed. The skewness of the nuclear DNA content unimodally distributed in a cluster could be corrected by logarithmic transformation. A mean value of the logarithmically transformed DNA contents was a more logical characterization of cancer than a graphically obtained mode of nontransformed values. Equations were derived to estimate the number of cells in the G1- and S-phases of mitotic cycles contained within the cluster and to estimate the correction factor of the mean of transformed values of cells in the G1-phase. Thus statistical tests for a shift in the nuclear DNA content of the cells in question in the G1-phase from that of control cells were made possible.

Analysis of human cancer DNA for DNA sequences of human adenovirus type 4.

We investigated whether human adenovirus type 4 (Ad4) might cause cancer in humans. Cell culture and animal model studies indicated that Ad4-induced human tumors should contain Ad4 DNA sequences. Thus Ad4 DNA was labeled in vitro (sp act, approximately 10(7) 3H counts/min/microgram) and hybridized in liquid phase with DNA extracted from human tumors. No viral sequences were detected in 31 squamous cell carcinomas of the lung, 5 adenocarcinomas of the lung, 4 oat cell carcinomas of the lung, 4 carcinomas of the stomach, 10 carcinomas of the colon, 3 carcinomas of the kidney, 3 carcinomas of the breast, 2 carcinomas of the ovary, and 6 Hodgkin's lymphomas. Reconstruction experiments with added Ad4 DNA indicated that the probe detected about 1 copy of 5% of the Ad4 genome per tumor cell. Therefore, these data were strong evidence (but did not prove) that none of these particular tumors were induced by Ad4.

Comparison of procoagulant activities in extracts of normal and malignant human tissue.

Cancer procoagulant A (CPA) was originally described in extracts of tumor tissue, but whether this represented a quantitative and/or a qualitative difference from procoagulant activity in normal tissue extracts was not clear. Procoagulant activity was quantitated in extracts of 12 matched normal and malignant human tissue samples from the large intestine, breast, lung, and kidney. The specific activity of procoagulants in the tumor extracts was not greater than that in the extracts of normal tissue. Two enzymatic characteristics of CPA that distinguish it from tissue thromboplastin are its inhibition by diisopropylfluorophosphate (DFP) and its lack of dependence on factor VII. These specific tests were used to evaluate qualitative differences between procoagulants from normal and malignant intestinal tissues. In the paired normal and malignant tissue extracts, all tumor samples were inhibited by DFP and were active in factor VII-depleted bovine plasma (F7D-BP). In contrast, the extracts of normal tissue were insensitive to DFP and, except for one extract, were inactive in F7D-BP. Four of 9 other tumor extracts (44%) were positive for both of these tests for CPA, whereas the other 5 extracts were positive for only one of the two tests. The results suggest that extracts of normal and malignant tissues contained similar levels of procoagulant. However, malignant tissue contained a procoagulant enzymatically different from normal tissue thromboplastin. Furthermore, most of the malignant tissue extracts seemed to contain little or no thromboplastin.

Demonstration of an altered S-adenosylmethionine synthetase in human malignant tumors xenografted into athymic nude mice.

S-Adenosylmethionine synthetase isoenzymes (EC 2.5.1.6) were studied in human malignant tumors xenografted into athymic nude mice and were studied in normal human and rat tissues. The tumors included 7 melanomas; 4 colon carcinomas; 3 each of mammary, cervical, and ovary carcinomas; 2 each of lung carcinomas and sarcomas; and 1 each of lymphoma and stomach and nasopharyngeal carcinomas. The presence of an altered intermediate Michaelis constant (Km) isoenzyme, previously shown to be a unique characteristic of rat neoplastic tissues, in these tumors was invariably detectable, though the presence of the low-Km isoenzyme characteristic of the enzyme of normal tissues was also evident in many of these tumors. Only tumors demonstrating high rates of growth were devoid of the low-Km isoenzyme. The low-Km isoenzyme was the only enzyme detectable in many normal human and rat tissues. Thus this isoenzyme was probably the common enzyme of all tissues. Some additional organ-specific isoenzymes were found, e.g., in the liver and lactating mammary gland. The low-Km isoenzyme in malignant tumors was altered to assume a considerably higher Km value, and this alteration appeared to be a common aberration of malignant tumors.

Opioid receptors and endogenous opioids in diverse human and animal cancers.

Receptor binding studies demonstrated specific high-affinity, saturable binding of a number of opioid ligands to a wide variety of neural and nonneural human and animal tumors. Radioimmunoassays revealed the presence of beta-endorphin and methionine-enkephalin in these tumors. Both methionine- and leucine-enkephalin were detected in tumor tissue by immunocytochemistry, with immunoreactivity related to the cortical cytoplasm of tumor cells, but not to cell nuclei. Endogenous opioids and receptors were found in benign and malignant tumors representative of ectodermal, mesodermal, and endodermal origin. Receptors and endogenous opioid peptides were present in tumors from many different species, including those transplanted into nude mice. These results suggest that opioid receptors and endogenous opioids are fundamental features of human and animal cancers.

RNA:DNA ratios in human tumors.

A study has been made of the quantitative relation existing between the RNA:DNA ratios in benign and in malignant human tumors. It has been demonstrated that a significant difference (Student's t-test) existed between malignant and benign human tumors within the different groups studied: a) malignant and benign breast tumors (t=2.22, P less than 0.05); b) Hodgkin's disease in spleen and its controls (t=2.39, P less than 0.02); and c) malignant and benign thyroid tumors (t=4.23, P less than 0.01). The diagnostic value of this finding is discussed.

Antibody to epiglycanin and radioimmunoassay to detect epiglycanin-related glycoproteins in body fluids of cancer patients.

By means of a radioimmunoassay, which utilized [125I]-epiglycanin and anti-epiglycanin antiserum induced in rabbits by injections of viable TA3-Ha ascites cells with Freund's complete adjuvant, picogram quantities of epiglycanin could be detected. Anti-epiglycanin antiserum was similarly produced in allogeneic mice. Unlabeled epiglycanin lost the capacity to compete with [125I]epiglycanin in the radioimmunoassay as a result of periodate oxidation or incubation with endo-alpha-N-acetyl-D-galactosaminidase (Diplococcus pneumoniae), an enzyme found to cleave only the disaccharide beta-D-galactopyranosyl-(1----3)-2-acetamido-2-deoxy-D-galactose chain from serine or threonine residues in epiglycanin. Glycosylhydrolases known to cleave alpha-D-mannose, beta-D-galactose (1,4-linked), beta-N-acetyl-D-glucosamine, and alpha-N-acetyl-D-galactosamine did not reduce the activity of epiglycanin. Neuraminidase enhanced the activity twofold to fivefold. The finding that little or no activity was demonstrated by the disaccharide, the reduced disaccharide, or other glycoproteins containing the same disaccharide chain suggested that the antigenic determinant probably involved the disaccharide and a unique amino acid sequence at the site of its attachment. By means of the radioimmunoassay epiglycanin cross-reactive antigens were detected in the peritoneal or pleural fluid and in the sera of patients with metastatic cancer. Lower concentrations of epiglycanin-like antigen(s) were found in the peritoneal fluid of patients with hepatitis or liver cirrhosis but not in normal serum.

Discriminant analysis of trace element distribution in normal and malignant human tissues.

Discriminant analysis of 16 trace element levels measured by ultramicro energy dispersive X-ray fluorescence in malignant and histologically normal human breast, colon, and lung tissues is shown to be a potentially valuable methodology for making malignant-normal and tissue-type classifications. Linear composites of trace elements producing optimal malignant-normal discriminations are found to differ with respect to the number and identity of elements included in the composite for breast, colon, and lung tissues. Nine-, 10-, and 11-element discriminant functions produced overall classification accuracies of 98% for breast, 100% for colon, and 100% for lung tissues, respectively. Elements found to be most important in distinguishing between malignant and normal tissues are Ca, Rb, and Zn in breast, Ca, Zn, and Fe in colon, and Fe, Mn, and Cu in lung samples. Three-group discriminations between breast, colon, and lung tissues were 85% accurate using trace element levels in paired malignant-normal tissues and 91% accurate using trace element levels in tumor tissues only.

Copper- and zinc-containing superoxide dismutase and manganese-containing superoxide dismutase in human tissues and human malignant tumors.

Superoxide dismutases might conceivably protect against both ionizing radiation and free radical-producing antibiotic antitumor drugs. Copper- and zinc-containing superoxide dismutase (CuZn superoxide dismutase) and manganese-containing superoxide dismutase (Mn superoxide dismutase) were specifically assayed in human malignant tumors and for comparison in human tissues. The tumors possessed less CuZn superoxide dismutase than did the more metabolically active tissues, but there was a large overlap between the tissue and the tumor levels. Mn superoxide dismutase was found in all tumors, and the ratio between the activities of CuZn superoxide dismutase and Mn superoxide dismutase was not different from that of the normal tissues. Human tumors are thus different from tumors from other species which have been reported to be deficient or very low in Mn superoxide dismutase. There was no obvious relation between sensitivity to ionizing radiation and content of the enzymes among the tumors and the tissues, nor did tumor types known to be responsive to radical-producing drugs possess less CuZn superoxide dismutase or Mn superoxide dismutase than other tumors.

Specific keratins as molecular markers for neoplasms with a stratified epithelial origin.

The expression of specific keratin polypeptides in human neoplasms was investigated by the immunoblot technique using monoclonal anti-keratin antibodies. Mr 50,000 and 58,000 keratins, recognized by AE1 and AE3 antibodies, respectively, were detected only in carcinomas of stratified epithelial origin, but not in carcinomas derived from simple epithelia. No keratin was detected in nonepithelial tumors including melanoma, lymphoma, neurofibroma, and sarcoma. The results suggest that the Mr 50,000 and 58,000 keratins provide useful molecular markers for identifying neoplasms of stratified squamous epithelial origin.

Quantitation and characterization of gamma-interferon receptors on human tumor cells.

The vast majority (71 of 77) of human tumor cells derived from various tissue origins were found to express specific membrane receptors for gamma-interferon (IFN-gamma). Six receptor-negative tumors were found among leukemic cells of lymphoid origin. Scatchard analysis with 125I-labeled human recombinant IFN-gamma revealed a similar binding affinity with a mean dissociation constant (Kd) of around 2 X 10(-11) M not only for various established cell lines, but also for leukemic and carcinoma cells derived from biopsy material. In contrast to similar KdS, large differences in the number of expressed IFN-gamma membrane receptors were found on distinct tumor cells of the same cell type ranging from a few hundred up to 2 X 10(4) for both carcinoma cells and leukemic cells. For comparison, the IFN-gamma receptor number on normal lymphocytes (mean, approximately 300/cell) and normal bone marrow cells (mean, approximately 1000/cell) was consistently found to be low. Cross-linking of membrane-bound 125I-IFN-gamma with disuccinimidyl suberate and subsequent sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis revealed, in both leukemia and carcinoma cells, three distinct complexes with molecular weights of approximately 70,000, 92,000, and 160,000, suggesting the existence of IFN-gamma receptor subunits. A dimeric structure of the functional IFN-gamma receptor with an estimated molecular weight of about 128,000 +/- 10,000 is proposed. Together with the Scatchard analysis, these data suggest the existence of a single class of high affinity IFN-gamma receptors in tumor cells of distinct tissue origin.

Location and distribution of difucoganglioside (VI3NeuAcV3III3Fuc2nLc6) in normal and tumor tissues defined by its monoclonal antibody FH6.

The distribution of a novel difucoganglioside (6B ganglioside, NeuAc alpha 2----3Gal beta 1----4[Fuc alpha 1----3]GlcNAc beta 1----3Gal beta 1----4[Fuc alpha 1----3]GlcNAc beta 1----3Gal beta 1----4Glc beta 1----1Cer) in various normal adult and fetal tissues, as well as in cancer tissues, has been studied by immunoperoxidase staining with a specific monoclonal antibody, FH6, directed to this antigen. A large variety of embryonic and fetal tissues (stomach, colon, small intestine, pancreas, esophagus, lung, and heart) showed a diffuse, weakly positive staining, particularly in the epithelial layer, up to the 70th to 80th day of gestation. However, no staining was observed in various normal adult tissues, including gastrointestinal and glandular epithelial tissues which were stained positively by antibody N-19-9 (directed to sialyl-Lea) or CSLEXI (directed to sialyl-Lex). FH6-positive loci were limited to the proximal convoluted tubuli in kidney and granulocytes. In contrast, 44 of 76 cases of cancer tissue tested, including gastric, colonic, lung, breast, and renal cancers, showed clearly positive staining. The intensity of staining in gastric and colonic cancer tissues by FH6 antibody was weaker and less frequent, although the incidence of positive staining for lung (50%) and breast cancer (86%) was significantly higher than that of the antigen stained by monoclonal antibody FH4 (Y. Fukushi, S. Hakomori, and T. Shepard, J. Exp. Med., 159: 506-520, 1984), which is directed to the asialo core of the FH6 antigen. The antigen levels in the serum of patients with various cancers, inflammatory diseases, and normal subjects were determined by radioimmunoassay. The antigen level was found to be significantly higher in the serum of some patients with cancer, particularly lung, liver, and pancreatic cancers, as compared with the serum levels in other types of cancer, noncancerous diseases, and normal subjects.

Isolation and characterization of proteoglycans from human nonepithelial tumors.

Proteoglycans (PGs) and glycosaminoglycans (GAGs) were identified in myogenic and fibrogenic tumors. More PGs containing mainly chondroitin sulfate could be detected in malignant tumors (leiomyosarcomas) than in benign tumors (leiomyomas and fibromas). Two groups of PGs were detected in the malignant tumors by ion-exchange chromatography and gel chromatography. One group was a large molecule with chondroitin sulfate side chains, seemingly composed of two or more subpopulations that were eluted from Sepharose CL-4B with a kav of 0.45. After removal of GAG side chains from the PG by chondroitinase AC digestion, core molecules with molecular weights greater than 200,000 were obtained. Another PG detected was a fraction of small PG eluted from Sepharose CL-4B with a kav of 0.45. It consisted of a core molecule with a molecular weight approximately equal to 48,000 and GAG side chains containing chondroitin sulfate-dermatan sulfate hybrids. The mixed sequence of L-iduronic acid with D-glucuronic acid in the same GAG chain was demonstrated by the formation of a small proportion of tetrasaccharide after chondroitinase AC digestion. In the benign tumors, the large PG was found only in very small amounts, and PG detected was composed mainly of the small one eluted from Sepharose CL-4B with a kav of 0.45. Its core protein had a molecular weight of approximately equal to 46,000, which was similar to that of small PG obtained from leiomyosarcomas, but its GAG side chains were composed mainly of dermatan sulfate containing small amounts of glucuronic acid. The results suggest that the core molecules of small PGs from both benign and malignant tumors are the products of the same gene but that they are processed in a different manner to form proteoglycans with different types of GAG chains.

Detection of ganglioside GD2 in tumor tissues and sera of neuroblastoma patients.

A murine monoclonal antibody (monoclonal antibody 126) produced against cultured human neuroblastoma cells (LAN-1) was found to be specifically directed to a disialoganglioside (GD2) antigen preferentially expressed on both cell lines and tissues derived from melanoma and neuroblastoma. In enzyme-linked immunosorbent assays, monoclonal antibody 126 failed to react with leukemic and lymphoblastoid cells as well as with a variety of carcinoma and sarcoma cell lines. Immunohistological analysis by the immunoperoxidase technique revealed strong reactivity of monoclonal antibody 126 with frozen and formaldehyde-fixed neuroblastoma and melanoma tissues. Tissues from patients with glioma or with small cell cancer of the lung showed faint staining, whereas those from individuals with sarcoma, lymphoma, and a variety of other neoplasms proved to be negative. Sera of neuroblastoma patients showed significantly elevated GD2 levels compared to normal children (p less than 0.001) and children with other tumors (p less than 0.001) as determined by a quantitative competitive enzyme-linked immunosorbent assay. Furthermore, the GD2 serum level of one neuroblastoma patient, when followed serially, was found to correlate with progression of disease, suggesting the potential usefulness of this assay for the diagnosis and monitoring of neuroblastoma.

Inhibition of human DNA polymerase alpha by alpha 1-antichymotrypsin.

alpha 1-Antichymotrypsin (ACT), which is known as an efficient serum protease inhibitor and is detected in tumor cell nuclei, was found to inhibit the activity of DNA polymerase alpha purified from human stomach adenocarcinoma. The concentration of ACT required for 50% inhibition was 1.0 mg/ml and the manner of its inhibition showed the partially competitive relationship between ACT and DNA in the assay system. Furthermore the removal of ACT by anti-ACT antibody lost its antichymotryptic and anti-DNA polymerase activities in parallel. On the other hand, it did not inhibit the activity of human DNA polymerase beta. Other human serum proteins including serum albumin, alpha 1-acid glycoprotein, alpha 1-antitrypsin, and immunoglobulin G as well as other protease inhibitors such as leupeptin, pepstatin, phenylmethylsulfonyl fluoride, and chymostatin did not affect the activity of DNA polymerase alpha. Furthermore ACT heated at 60 degrees C did not inhibit DNA polymerase alpha, although it could still bind to DNA as well as native ACT. It was therefore concluded that the inhibitory action of ACT on DNA polymerase alpha was a direct phenomenon unrelated to its protease inhibitory or DNA binding activities.

Presence of fucosyl residues on the oligosaccharide antennae of membrane glycopeptides of human neuroblastoma cells.

Fucosyl residues linked alpha 1 leads to 3 or 4 to N-acetylglucosamine were found in large amounts on glycopeptides from the membranes of human tumor cells of neurectodermal origin but not on membrane glycopeptides from human fibroblasts. The fucosyl residues were detected by release of radioactive fucose from the glycopeptides with an almond alpha-L-fucosidase specific for fucosyl alpha 1 leads to 3(4)-N-acetylglucosamine. In other studies, the linkage was shown to be alpha 1 leads to 3 by nuclear magnetic resonance analysis (U. V. Santer, M. C. Glick, H. van Halbeek, and J. F. G. Vliegenthart. Carbohydr. Res., 118: in press, 1983). Glycopeptides containing these fucosyl residues from four human neuroblastoma cell lines were defined by binding to immobilized lectins. In addition, the glycopeptides from one human neuroblastoma cell line, CHP-134, were further characterized by enzyme degradation and columns calibrated for size and charge. The antennary position of fucosyl alpha 1 leads to 3-N-acetylglucosamine on the glycopeptides was demonstrated by the use of exoglycosidases and endoglycosidase D, since complete degradation to yield fucosyl-N-acetylglucosaminylasparagine was obtained only after treatment with almond alpha-L-fucosidase prior to the sequential degradation. Fucosyl alpha 1 leads to 3-N-acetylglucosamine was present on most size and charge classes of membrane glycopeptides and therefore was not limited to a few glycoproteins. Since the almond alpha-L-fucosidase cleaves fucosyl residues from glycoproteins, the physiological effects of the increased specific fucosylation on human tumors of neurectodermal origin can be examined.

Relationship of monoclonal antibody binding to estrogen and progesterone receptor content in breast cancer.

Three monoclonal antibodies--H59, H71, and H72--which react with human breast cancers have been developed using the estrogen-dependent human breast cancer cell line, ZR-75-1, as the immunogen. H59 bound only to estrogen receptor-positive, estrogen-regulated breast cancer cells in culture, whereas H71 and H72 bound breast cancer cells irrespective of the estrogen receptor content. All three antibodies have minimal cross-reactivity with non-breast tissue culture cell lines. The three antigens appear to be glycoproteins located on the cell surface. H59 and H72 antigens bound preferentially to the apical surface of duct cells and may be secreted; H71 antigen demonstrated no evidence of an apical orientation or secretion. The binding of the antibodies to fixed cryosections from 152 breast cancer and 111 benign breast disease specimens has been evaluated using a radioimmunoassay. Eighty-five % of breast cancer and almost 100% of benign disease specimens were bound by at least one antibody. H59 bound 39%, H71 bound 51%, and H72 bound 65% of cancer specimens. Estrogen receptor and progesterone receptor analyses were obtained on 141 specimens. H59 bound almost exclusively to tumor specimens which contained estrogen and/or progesterone receptor, but not to all receptor-positive tumors. Therefore, the H59 antigen appeared to be present on a subset of estrogen receptor-positive tumors. Considering that it bound only to estrogen-regulated cells in culture, the antigen may be estrogen regulated, and its presence may predict a response to hormone therapy. H71 and H72 recognized cell surface differentiation antigens but bound tumor specimens regardless of the receptor content. These antibodies may be useful as independent variables for predicting response to therapy and prognosis of patients with breast cancer.

Mucin-like carcinoma-associated antigen defined by three monoclonal antibodies against different epitopes.

Three monoclonal antibodies [MAb], b-8, b-12, and b-15, have previously been shown to react with mammary carcinomas and with a restricted set of cells in normal human tissues [C. Stähli et al., Experientia (Basel), 41: 1377-1381, 1985; H. R. Zenklusen et al., Virchows Arch. Abt. A Pathol. Anat., 413: 3-10, 1988]. They are shown here to recognize the same high molecular weight acid soluble glycoprotein antigen. Lectin binding, biolabeling, and deglycosylation experiments demonstrate that it contains O-linked carbohydrate side chains with sialic acid and hexoses including fucose, galactose, and/or galactosamine but little if any mannose. These properties, typical of mucin-like glycoproteins, agree with the antigen expression on mucin-secreting epithelial surfaces (H. R. Zenklusen et al., Virchows Arch. Abt. A Pathol. Anat., 413:3-10, 1988). The antigen is thus named mucin-like carcinoma-associated antigen (MCA). The three MAb are shown to bind to three different epitopes on MCA. Two of these epitopes (MCA-b-8 and MCA-b-15) are O-linked carbohydrates, and one (MCA-b-15) contains sialic acid. The epitope MCA-b-12 is of peptide nature. Of various two-site sandwich enzyme immunoassays composed of different combinations of the three MAb, the one with MAb b-12 in both positions is selected for a serum assay. Analyses of tumor patients' sera demonstrate that this MCA enzyme immunoassay can be of use as a tumor marker assay for mammary carcinomas. The parameter MCA enzyme immunoassay is shown to differ from other parameters described in the literature.