Monoclonal antibodies against tissue-nonspecific alkaline phosphatase. Report of the ISOBM TD9 workshop.

Nineteen monoclonal antibodies (MAbs) against tissue-nonspecific (liver/bone/kidney) alkaline phosphatase (TNALP) were investigated in the ISOBM TD-9 Workshop. These MAbs were generated with antigens obtained from human bone tissue (n = 9), human osteosarcoma cell lines (SaOS-2 and TPX; n = 7) and human liver tissue (n = 3). The evaluation included the following antigen forms: (a) commercially available preparations of human bone ALP (BALP) and liver ALP (LALP); (b) human BALP isoforms, B/I, B1 and B2; and (c) soluble secreted epitope-tagged recombinant human TNALP (setTNALP) expressed in COS-1, osteosarcoma (SaOS-2) and hepatoma (Huh2) cell lines. In addition, 16 TNALP mutant cDNAs corresponding to a wide spectrum of reported hypophosphatasia mutations were used in an attempt to map specific immunoreactive epitopes on the surface of the TNALP molecule. The TD-9 MAbs were evaluated by immunoradiometric (IRMA) assays, cross-inhibition and different enzyme immunoassay designs. No indications of explicit tissue discriminatory immunoreactivities of the investigated MAbs against TNALP were found. However, certain IRMA combinations of MAbs increased the specificity of BALP measurements. All MAbs bound to the three BALP isoforms B/I, B1 and B2, but none of the investigated MAbs were specific for any of the isoforms. Significant differences were, however, found in immunoreactivity between these isoforms, with cross-reactivities ranging from 21 to 109% between the two major BALP isoforms B1 and B2. Desialylation with neuraminidase significantly increased the MAb affinity for the BALP isoforms B/I, B1 and B2, and also decreased the observed differences in cross-reactivity between these isoforms. We suggest, therefore, that the MAb affinity is dependent on the amount/number of terminal sialic acid residues located at the five putative N-glycosylation sites. Based on the overall results, we present a putative three-dimensional model of the TNALP molecule with positioning of the four major antigenic domains (designated A-D) of the investigated MAbs. The TNALP molecule is depicted as a homodimer, hence most, but not necessarily all, epitopes are displayed twice. The antigenic domains were positioned with the following assumptions: domain A was positioned close to the active site since most of these MAbs interfered with the catalytic activity. Interestingly, both MAbs included in the commercial BALP kits were grouped with domain A. Moreover, 4 of the 5 putative N-glycosylation sites (with terminal sialic acid residues) are located within, or with close proximity to, domain A. Domain B was localized at the top flexible loop (crown domain) of the TNALP molecule. Domain C was clearly defined by the IRMA assay combinations and by site-directed mutants of TNALP to be close to residue E281, which is located near the fourth metal binding site, likely to be occupied by a calcium ion. Domain D was positioned close to residues A115, A162 and E174, but this domain was also close to the GPI anchor site. In conclusion, none of the 19 investigated TD-9 MAbs were entirely specific for BALP or LALP, thus indicating that all MAbs bind mainly to epitopes on the common protein core of BALP and LALP and/or common glycosylated epitopes. However, some MAbs (either single or in combination with other MAbs) work sufficiently well to measure BALP when the assayed samples do not contain elevated levels of LALP.     

Conserved epitopes in human and mouse tissue-nonspecific alkaline phosphatase. Second report of the ISOBM TD-9 workshop.

A panel of 19 monoclonal antibodies (MAbs) against human tissue-nonspecific (liver/bone/kidney) alkaline phosphatase (TNAP) was obtained through the ISOBM TD-9 workshop. In the present study, the reactivity of these MAbs has been characterized against mouse TNAP. A mouse embryonic stem cell line, frozen sections of long bones, alkaline phosphatase extracted from mouse bone, and serum were used as the source of TNAP for individual assays. Each MAb was tested for immunoreactivity to mouse TNAP by Western blot analysis, immunohistochemistry and enzyme immunoassay. Antibodies 314 and 315 reacted strongly with mouse TNAP in Western blots, while all other antibodies were negative. By immunohistochemistry, antibodies 314, 315 and 333 produced strong positive staining using frozen sections, while antibody 334 was moderately positive. Enzyme immunoassays indicated that MAb 333 was also able to bind to serum TNAP. These antibodies represent very useful reagents to study the pathophysiological expression of TNAP in mouse tissues and in mouse serum.     

Candidate epitopes for measurement of hCG and related molecules: the second ISOBM TD-7 workshop

Participants of the Second International Workshop (WS) on human chorionic gonadotropin (hCG) of the International Society of Oncology and Biomarkers Tissue Differentiation 7 (ISOBM TD-7) have characterized in detail a panel of 69 antibodies (Abs) directed against hCG and hCG-related variants that were submitted by eight companies and research groups. Specificities of the Abs were determined using the First WHO International Reference Reagents for six hCG variants, i.e., hCG, hCGn, hCGβ, hCGβn, hCGβcf, and hCGα, which are calibrated in SI units, and hLH. Molecular epitope localizations were assigned to the ISOBM-mAbs by comparing ISOBM-Ab specificity, sandwich compatibility, and mutual inhibition profiles, to those of 17 reference monoclonal (m)Abs of known molecular epitope specificities. It appeared that 48 Abs recognized hCGβ-, 8 hCGα-, and 13 αβ-heterodimer-specific epitopes. Twenty-seven mAbs were of pan hCG specificity, two thereof with no (<0.1 %; epitope β₁), 12 with low (<1.0 %; epitopes β_2/4), and 13 with high (>>1 %; epitopes β_3/5) hLH cross-reactivity. The majority of hCGβ epitopes recognized were located in two major antigenic domains, one on the peptide chain of the tips of β-sheet loops 1 and 3 (epitopes β_2–6; 27 mAbs) and the second around the cystine knot (e.g., epitopes β₁, β₇, and β₁₀; 9 mAbs). Four mAbs recognized epitopes on hCGβcf-only (e.g., epitopes β₁₁ and β₁₃) and six mAbs epitopes on the remote hCGβ-carboxyl-terminal peptide (epitopes β₈ and β₉ corresponding to amino acids 135–144 and 111–116, respectively). For routine diagnostic measurements, methods are used that either detect hCG-only, hCGβ-only, or hCG together with hCGβ or hCG together with hCGβ and hCGβcf. Sandwich assays that measure hCG plus hCGβ and eventually hCGβcf should recognize the protein backbone of the analytes preferably on an equimolar basis, should not cross-react with hLH and not be susceptible to blunting of signal by nonmeasured variants like hCGβcf. Such assays can be constructed using pairs of mAbs directed against the cystine knot-associated epitope β₁ (Asp10, Asp60, and Gln89) in combination with epitopes β₂ or β₄ located at the top of β-sheet loops 1?+?3 of hCGβ involving aa hCGβ20-25?+?68-77. In summary, the results of the First and Second ISOBM TD-7 WSs on hCG provide the basis for harmonization of specificities and epitopes of mAbs to be used in multifunctional and selective diagnostic hCG methods for different clinical purposes.     

Serologic mapping and biochemical characterization of the carcinoembryonic antigen epitopes using fourteen distinct monoclonal antibodies

A series of 14 monoclonal anti-carcinoembryonic antigen (CEA, Mr 180,000) antibodies (MAbs) that show a strong degree of selective reactivity for human colon carcinomas versus normal adult tissues were used to construct a serological map of the CEA molecule. The MAbs were generated using extracts of colon carcinomas as immunogen and are thus given a COL designation. None of the 14 COL-MAbs tested were reactive with purified non-specific cross-reacting antigen (NCA, Mr 55,000) from normal lung, although some showed reactivity to human granulocytes. All the COL-MAbs tested were reactive with normal fecal antigen-2 (NFA-2, Mr 170,000); however, many of the COL-MAbs demonstrated a higher affinity constant to CEA than to NFA-2. Cross-competition radioimmunoassays classified the 14 COL-MAbs into 5 groups. The chemical nature of the COL-binding domains was tested using chemically or enzymatically treated CEA; all reacted with periodate-treated CEA and deglycosylated CEA, indicating that the COL-reactive epitopes appear to be of a proteinaceous nature. Heat treatment, reduction, alkylation, pepsin digestion or pronase treatment of CEA, however, gave differential results with respect to COL binding. Antibody titration experiments were carried out to define differential reactivities to colorectal carcinoma versus NCA-containing granulocyte extracts; these results were compared with results obtained using several anti-CEA MAbs that have been used in clinical trials. Granulocyte binding and biochemical studies showed that the COL MAbs may distinguish as many as 7 to 10 CEA epitopes.     

Immunohistochemical profiles of 30 monoclonal antibodies against cytokeratins 8, 18 and 19. Second report of the TD5 workshop.

In the first report of the TD5 workshop (TD5-1), the epitope specificities of 30 different monoclonal antibodies against cytokeratins 8, 18 and 19 were determined. This second report presents the immunohistochemical profiles of these antibodies using human appendix and normal skin for evaluation. Each antibody was tested by one or two different laboratories recruited from the Dutch Working Group on Immunohistochemistry and Cytochemistry. Eight different laboratories participated. The histological specimens were pretreated by the participants in three different ways for immunohistochemistry: microwave antigen retrieval in citrate buffer, enzymatic digestion to restore epitope exposure, no specific treatment (untreated paraffin-embedded samples), and tested blindly without knowledge of cytokeratin or epitope specificity of the antibodies at three different concentrations of 50, 10 and 1 microg/ml. Most of the tested antibodies (29/30) were useful in at least one pretreatment method, with microwave antigen retrieval being the most sensitive approach. For some antibodies, very high backgrounds were observed. Furthermore, it can be concluded that 11 MAbs performed well using all three staining protocols, including untreated paraffin-embedded sections. Interestingly, all the antibodies with documented selected specificity towards cytokeratin 8 (i.e. 178, 191, 199, 202 and 206) are reactive with an immunodominant region corresponding to amino acids 340-365 on cytokeratin 8, which evidently is well-suited as target for immunohistochemical interactions. Similarly, three antibodies with the same capacity to react with untreated samples had specificity against cytokeratin 19 (i.e. 179, 197 and 204) in the corresponding region in this filament, i.e. amino acids 311-335, or the KS 19.1 epitope. None of the six antibodies against the other major cytokeratin 19 epitope (BM 19.21) were found useful for immunohistochemistry on untreated samples. The overall conclusions from the present investigation are that all cytokeratin-8-specific antibodies with defined epitope specificities were very useful. Only one of the major two epitopes on cytokeratin 19 seems to be available for efficient immunohistochemistry. Cytokeratin 18 exposes some epitopes outside the immunodominant region reactive with the antibodies 190, 203 and 205 which can be used for untreated samples. The implications of these findings are of significance both for diagnostic histopathology and for the biology of tumor marker epitope expression in tissues.     

Adhesion to carcinoembryonic antigen by human colorectal carcinoma cells involves at least two epitopes

Carcinoembryonic antigen (CEA) may be involved in both cell-cell and cell-substrate adhesion. Our purpose was to determine whether epitopes involved in the homophilic binding of human colorectal carcinoma cells to CEA participated in adhesion to basement membrane proteins. Three human colorectal adenocarcinoma cell lines and one CHO cell line transfected with CEA cDNA were tested in a solid-phase adhesion assay. The 2 CEA-expressing carcinoma cell lines (KM-12c and CCL 188) and the transfectant, but not the parental CHO line, bound to CEA. The CEA-non-producing carcinoma line (Clone A) did not bind to CEA. All colorectal carcinoma cell lines, the transfectant and the parental CHO line bound to laminin, while the colorectal carcinoma lines bound to type-IV collagen. MAbs to epitopes on CEA that cross-react with non-specific cross-reacting antigen (NCA) inhibited adhesion of CEA-expressing cells to CEA. MAbs to non-cross-reactive epitopes of CEA did not block adhesion to CEA. When the inhibitory anti-CEA antibodies were compared in a competitive radioimmunoassay, 2 distinct epitopes were identified. Epitope I is in the N-terminal domain and defined by MAbs MN3, T84.1 and C110, whereas epitope II is located in the repeating loop domains and is recognized by antibodies MN15, PR3B10 and NPI. None of the antibodies to epitope I or II blocked adhesion by KM-12c or CCL 188 cells to laminin or type-IV collagen. Thus, at least 2 different regions on CEA participate in adhesion to CEA but not to collagen or laminin by CEA-expressing human colorectal carcinoma cells.     

Epitopes of carcinoembryonic antigen defined by monoclonal antibodies prepared from mice immunized with purified carcinoembryonic antigen or HCT-8R cells

A library of 18 monoclonal antibodies (MAbs) reactive with purified carcinoembryonic antigen (CEA) has been prepared. The specificity of these MAbs was tested and they have been separated into nine subgroups, each recognizing a different region of the CEA molecule. Seven MAbs from four of the groups also react with the nonspecific cross-reacting antigen. Some of the MAbs are directed against conformational determinants: three of the MAb groups bind poorly to sodium dodecyl sulfate-treated CEA, while five of the groups are not reactive with reduced and alkylated CEA. Three of the groups react with purified CEA but not with the cell surface CEA of HCT-8R cells, while the other groups react with both forms. The MAbs were tested for binding to fragments of CEA obtained by chemical cleavage and the groups of MAbs were found to react with different subsets of such fragments.     

Localisation of monoclonal antibodies reacting with different epitopes on carcinoembryonic antigen (CEA)--implications for targeted therapy.

Antibody targeting has potential for selective delivery of cancer therapy. However, there is a wide variation in the degree of antibody localisation in individual patients with colorectal adenocarcinoma. Colorectal adenocarcinomas are composed of glandular structures separated from fibrovascular stroma by a basal lamina which may represent a significant barrier to extravasated antibody. Basement membrane-associated CEA epitopes may be more accessible to antibodies than those which are cytoplasmic or lumenal. We have investigated by immunohistochemistry and in vivo localisation, the extent to which distribution of antigen epitopes influences targeting. Two monoclonal antibodies (A5B7 and EA77) recognising non-overlapping CEA epitopes were reacted immunohistochemically with samples of 39 tumours. Intensity and site of reaction were assessed for basement membrane, cytoplasmic or lumenal surface association. 125I-labelled antibodies were injected into nude mice bearing LS174T tumour. Per cent injected activity per gram was measured in tumour and normal tissues, 24, 72 and 168 h later. Tissues reacted immunohistochemically for CEA were autoradiographed to assess the relationship of injected antibody to target antigen. Immunohistochemistry showed that A5B7 antibody favours basement membrane aspects of malignant glands; in contrast, EA77 concentrated generally on lumenal surfaces. In vivo localisation showed that per cent inj.act g-1 in tumour for A5B7 reached 36.5% at 24 h. EA77 localised to a lesser extent (9.1% at 24 h), despite a longer circulatory half-life. Autoradiography combined with immunohistochemistry showed A5B7 reacting with antigen close to vasculature after 24 h, slowly penetrating deeper parts of the tumour by 72 h. In contrast, EA77 was confined mainly to fibrovascular stroma, showing little labelling of antigen-positive tumour cells. Localisation differences between A5B7 and EA77 may partly be due to accessibility of epitopes on tumour cells.     

Conformational epitopes specific to carcinoembryonic antigen defined by monoclonal antibodies raised against colon cancer xenografts

Four monoclonal antibodies (MoAbs) reactive with carcinoembryonic antigen (CEA) were obtained by hybridizing mouse myeloma cells (P3-X63-Ag8-U1) with spleen cells from nude mice (BALB/c, nu/nu) that had rejected transplanted human colonic adenocarcinomas Co-3 and Co-4 following intraperitoneal injection of spleen cells from immunocompetent mice (BALB/c). By solid-phase RIA with purified CEA and its related antigens, NCC-CO-413 (IgG2a, kappa) was shown to react with NCA and BGP-I as well as with CEA, whereas the reactivities of three other MoAbs, NCC-CO-308 (IgG1, kappa), -432 (IgG1 lambda), and -411 (IgG1, kappa) were limited to CEA. Immunohistochemical reactivities of these MoAbs to colonic carcinomas, granulocytes, and liver bile canaliculi on acetone-fixed paraffin-embedded sections ("AMeX" sections) confirmed the specificities of these MoAbs shown by the solid-phase RIA. By competition solid-phase RIA, the epitopes recognized by NCC-CO-308 and -432 were shown to be shared or located close to each other, whereas the other MoAbs were shown to recognize different epitopes. Thus, two epitopes specific to CEA and one shared by NCA and BGP-I as well as CEA were identified. Furthermore, reactivities of MoAbs with the two CEA-specific epitopes were easily abolished by heat denaturation or reduction of CEA, as revealed by solid-phase RIA and SDS-PAGE-immunoblotting, indicating that these two CEA-specific epitopes are based on the conformational structure of the CEA molecule.     

Antigenic sites in carcinoembryonic antigen

The epitope reactivities of 52 well-characterized monoclonal antibodies (Mabs) against carcinoembryonic antigen from 11 different research groups were studied using competitive solid-phase immunoassays. About 60% of all possible combinations of Mabs as inhibitors and as the primary binding antibody were investigated. The inhibition data were analyzed by a specially developed computer program "EPITOPES" which measures concordance and discordance in inhibition patterns between Mabs. The analysis showed that 43 of the 52 Mabs (83%) could be classified into one of five essentially noninteracting epitope groups (GOLD 1-5) containing between four and 15 Mabs each. The epitopes recognized by the Mabs belonging to groups 1 to 5 were peptide in nature. With one or two possible exceptions non-classifiable Mabs were either directed against carbohydrate epitopes (4 Mabs) or were inactive in the tests used. Within epitope groups GOLD 1, 4, and 5 two partially overlapping subgroups were distinguished. Mabs with a high degree of carcinoembryonic antigen specificity generally belonged to epitope groups GOLD 1 and 3.     

A monkey antigen crossreacting with carcinoembryonic antigen, CEA.

Normal monkey tissues were found to contain an antigen which crossreacts immunologically with the carcinoembryonic antigen (CEA) of the human digestive tract. The monkey antigen reacted with complete or partial identity to the normal crossreacting antigen (NCA) in humans when tested in immunodiffusion against anti-CEA or anti-NCA. Extracts of monkey tissues inhibited in radioimmunoassays measuring human NCA. It is possible that monkey foetuses and colonic tumours contain CEA.     

Differential expression of carcinoembryonic antigen in early gastric adenocarcinomas versus benign gastric lesions defined by monoclonal antibodies reactive with restricted antigen epitopes

Murine monoclonal antibodies (MAbs) reactive with distinct epitopes on carcinoembryonic antigen (CEA) have been analyzed systematically by radioimmunoassays, Western blotting, and immunohistochemical assays to define CEA expression in adenocarcinomas, benign lesions, and normal tissues of the stomach. Each of four COL-MAbs (COL-1, COL-4, COL-6, and COL-12) reacted preferentially with cell extracts of adenocarcinomas versus those of normal mucosae in solid-phase radioimmunoassays. Using Western blotting analyses MAbs COL-1, COL-4, COL-6, and COL-12 detected only the Mr 180,000 molecule characteristic of CEA in adenocarcinoma of the stomach; no reactivity was observed in an extract of normal gastric mucosa. Antibody competition radioimmunoassays were then carried out to define relations among COL-MAbs using 125I-radiolabeled MAbs, and nonradiolabeled MAbs as competitors. A spectrum of formalin-fixed, paraffin-embedded normal, benign, and malignant tissue sections of the stomach were examined for immunoreactivities with COL-MAbs using immunohistochemical assays to define whether the COL-MAbs were able to detect CEA expression in early foci of gastric carcinomas. All of the COL-MAbs generally demonstrated selective reactivities to adenocarcinomas (n = 40) versus benign lesions (n = 15) and normal mucosae (n = 6) of the stomach. From 72 to 100% of adenocarcinomas at early stage (n = 18) were reactive with the COL-MAbs, suggesting that these MAbs might serve as immunohistochemical diagnostic tools to detect early foci of gastric carcinoma. The data reported here indicate that the COL-MAbs can potentially be utilized as radioimmunological and immunohistochemical adjuncts to differentiate early adenocarcinomas from normal mucosae or benign lesions of the stomach on the basis of differential CEA expression.     

Enhancement of carcinoembryonic antigen expression by interferon.

Human interferon decreased DNA but not RNA synthesis in a human colon carcinoma cell line, WiDr; in addition, there was a two- to three-fold increase in the expression of a tumor-associated antigen, carcinoembryonic antigen. In contrast, interferon had no effect on a normal human diploid cell line, WI-38. Thus, in addition to its anti-cellular effect against tumor cells, interferon can also modulate tumor antigenicity.     

Immunological heterogeneity of carcinoembryonic antigen: purification from meconium of an antigen related to carcinoembryonic antigen

Two antigens cross-reactive with carcinoembryonic antigen (CEA) and distinct from the nonspecific cross-reacting antigen were identified in meconium by double immunodiffusion with a conventional goat anti-CEA antiserum. These two antigens together competitively inhibited cross-reacting antibodies against them in CEA radioimmunoassay and contributed to the measurement of meconium CEA levels which averaged 6 times higher than that determined with anti-CEA specific antibody. A purification method for one of these antigens, tentatively designated meconium antigen, is described and uses a combination of ethanol fractionation, ion-exchange and molecular sieve chromatography, and adsorption to an immunoadsorbent containing a cross-reactive murine monoclonal antibody to CEA. Preliminary characterization of the purified meconium antigen showed it to be a glycoprotein, migrating as an alpha-globulin and having a molecular size similar to that of CEA (Mr 185,000 versus 200,000). Antigenically, it lacks at least one determinant present on CEA and differs further from CEA by being weakly reactive with concanavalin A and resistant to proteolytic digestion with Pronase E. Although these properties of meconium antigen suggest that it may be nonspecific cross-reacting antigen 2, additional chemical and antigenic studies are required to establish its relationship to CEA and other CEA-related antigens in meconium.     

Methylation analysis of the carbohydrate portion of carcinoembryonic antigen.

The carbohydrate structural units of carcinoembryonic antigen samples isolated from four different tumors were quantitated using gas chromatography-mass spectrometery after methylation and subsequent conversion to their alditol acetates. Different carcinoembryonic antigen preparations showed some quantitative but no qualitative differences in the structural units present. The results indicate that a large portion of the fucose residues in the glycoprotein were linked to N-acetylglucosamine and that most of the branching mannose residues were probably linked to three N-acetylglucosamine residues.     

Specific adhesion of carcinoembryonic antigen-bearing colorectal cancer cells to immobilized carcinoembryonic antigen.

Recent characterization of the genomic structure of carcinoembryonic antigen (CEA) is consistent with that of a cellular adhesion molecule. To examine this function in colorectal cancer, the adherence of cell lines to microtiter wells coated with CEA and well-described adhesive molecules was determined. The CEA-positive cell line LoVo and the CEA-devoid cell line H-Meso-1 did not differ in adherence to the extracellular matrix proteins laminin, collagen and fibronectin, whereas LoVo cells adhered to CEA (10 micrograms/well) in a specific manner (43% bound cells vs. 1.5% bound cells with BSA or alpha-acidglycoprotein controls, P less than 0.01) while H-MESO-1 showed no adhesion to CEA (less than 0.6% bound cells). This adhesion of LoVo cells to CEA was not affected by co-incubation of cells with EDTA, sodium azide, or at 23 degrees C. However, the CEA to CEA adhesive interaction was inhibited by a monoclonal antibody directed against an epitope in the N-terminal domain of the CEA molecule, and decreased by enzymatic removal of CEA from the LoVo cell membrane. The extent of adhesion to immobilized CEA by four CEA-producing cell lines (LoVo, HT29, LS174T and LS174-S), correlated with membrane CEA expression as determined by FACS analysis. The results of these experiments add support to the concept that CEA may function as a specific homotypic cellular adhesion molecule for colorectal cancer cells.     

Alpha-foetoprotein and carcinoembryonic antigen in germ cell neoplasms.

Serum alpha-foetoprotein (AFP) and serum carcinoembryonic antigen (CEA) levels were measured, serially whenever possible, in 70 patients attending the Institute of Radiotherapy, Rotterdam, on account of testicular (65) or ovarian (4) germ cell tumours or, in one case, an endodermal sinus (yolk sac) tumour in the mediastinum. In 15 patients the disease was active; in the others it was in remission. Patients with active disease had raised serum AFP levels which correlated well with disease activity; no patient without evidence of active disease had raised serum AFP levels. None of the patients with active disease was found to have raised serum CEA levels. There was no correlation between serum AFP and CEA levels in patients with germ cell neoplasms, but good correlation between serum AFP levels and disease activity. Serum CEA levels did not correlate with disease activity, and serial determinations would therefore not be useful in monitoring progress in this group of diseases.     

Immunohistochemical localization and molecular characteristics of three monoclonal antibody-defined epitopes detectable on carcinoembryonic antigen (CEA)

Three murine monoclonal antibodies (MAbs) reactive to different epitopes on CEA were selected according to their ability to bind to various human tissue sections. The most selective MAb, BW 431/31, bound to the majority of colon carcinomas but only faintly to normal colon mucosa. In addition to the tissues stained by MAb BW 431/31, MAb BW 250/183 reacted with granulocytes, colon mucosa and faintly with pancreatic ducts. The third MAb, BW 374/14, reacted with granulocytes, colon mucosa, strongly with pancreatic ducts and with alveolar and bronchial epithelium. The antigenic determinants recognized by the 3 MAbs in human tissue sections were resistant to formaldehyde fixation and paraffin embedding as well as to periodic acid oxidation and neuraminidase treatment. The last two treatments suggest that the epitopes are protein in nature. Using MAb affinity chromatography, 3 antigen preparations were isolated from a human colon carcinoma xenograft with an approximate molecular weight of 180 kd. These preparations were shown to bear the epitopes from each of the MAbs and from a polyclonal antiserum specific for purified CEA. Furthermore, the ability of MAb BW 431/31 to localize its antigenic determinant in vivo on a human colon carcinoma xenograft is evaluated and its possible application in the patient is suggested.