Monoclonal antibodies raised against B 16 melanoma cells in syngeneic mice were functionally screened for their ability to inhibit cell adhesion in tissue culture. Three of these antibodies (16/43, 16/77, 16/82), when preinjected into C57BL/6 mice, markedly reduced the number of experimental lung metastases produced by B 16 cells, possibly by interference with their adhesion to the lung endothelia. We now report that these monoclonal antibodies block in vitro attachment of the majority of human melanoma cell lines tested and also of carcinoma, neuroblastoma, and glioblastoma cells from both mice and humans but untransformed cell lines such as 3T3 mouse or MRC-5 human fibroblasts are not affected. The antibodies also react with mouse teratocarcinoma stem cells (F9, PCC4) but not with differentiated teratocarcinoma lines (PYS-2, 944). Furthermore, the antiadhesion activity of the antibodies could be quantitatively absorbed by intact human and mouse tumor cells but not by untransformed cells, suggesting that the corresponding antigens may represent tumor-associated cell surface components. Correspondingly, the antigens were found on simian virus 40-transformed 3T3 mouse fibroblasts and are expressed in a temperature-sensitive fashion in chicken fibroblasts transformed with a temperature-sensitive Rous sarcoma virus. On "immunoblots" of NaDodSO4-containing gels the three selected antibodies (16/43, 16/82, 19/1) were absorbed by antigens with molecular weights of 40,000 and 50,000.Monoclonal antibodies are highly specific reagents; in cancer research, they may eventually develop into powerful tools for diagnosis and therapy. However, few cases have been reported so far in which monoclonal antibodies were used with the goal of suppressing human tumors (1-4); most studies have dealt with human tumor cells in nude mice or with animal tumors (5-11). It is clear from these studies that the success of using monoclonal antibodies for cancer therapy ultimately depends on two factors: (i) whether the antibodies are specific for the tumor cells (i.e., are not absorbed by normal tissue) and (ii) whether the antibodies will also functionally interfere with tumor development (e.g., an antibody selected simply for binding to tumor cells might not be functionally active).We have recently developed procedures to find antibodies that fulfill these conditions. First, monoclonal antibodies were produced against tumor cells by immunization of syngeneic animals; this procedure enriches for antibodies directed against the "mutant" characteristics of the tumor cell-i.e., against tumor-associated antigens. Second, these monoclonal antibodies were screened in functional assays in vitro; this procedure increases the chance of finding antibodies that interfere with tumor development in vivo. Thus (human tongue carcinoma cells), TR138 (human larynx carcinoma cells), and MCF7 (human breast carcinoma cells) were obtained from E. B. Lane (London) (15). Chicken embryo fibroblasts (CEFs) transformed with Rous sarcoma virus (RSV) (SR-1) and ...