A keratin epitope that is exposed in a subpopulation of preneoplastic and neoplastic mouse mammary epithelial cells but not in normal cells.

Three monoclonal anti-keratin antibodies, AE1, AE3, and AE4, were used to compare the expression of keratins in normal, preneoplastic, and malignant mouse mammary epithelial cells growing in primary culture. In indirect immunofluorescence, AE1 did not stain normal cells but did stain a minority of preneoplastic and carcinoma cells. AE3 reacted with a subpopulation of epithelial cells in both the normal and abnormal cultures, except for certain cultures from one type of tumor wherein all of the epithelial cells were reactive. AE4 decorated an elaborate keratin filament network in all cultured mammary epithelial cells, regardless of neoplastic state. In double-label immunofluorescence, a guinea pig anti-keratin antiserum, which reacts preferentially with myoepithelial cells, exhibited coincident staining with AE1 in the tumor cultures and AE3 in the normal and most tumor cultures, indicating that the cells recognized by the antibodies in these populations were myoepithelial. Immunoblot experiments with cytoskeletal polypeptides extracted from the normal and tumor cells demonstrated that the set of keratins recognized by each monoclonal antibody was essentially the same in all of the cells except for a Mr 40,000 component that was present in normal cells but either absent or diminished in the cancer cells. Thus, while normal cells had Mr 40,000 and 50,000 keratins recognized by AE1, the epitope detected by this antibody was apparently concealed or "masked" in situ. AE3 reacted in immunoblots with a major keratin group (Mr 54,000-55,000) and a minor keratin (Mr 57,000), while AE4 reacted only with the Mr 54,000-55,000 keratin species. Because immunofluorescence with AE4 showed that the Mr 54,000-55,000 keratin group was present in all mammary epithelial cells, the AE3-reactive epitope must be masked in the majority of normal and tumor cells. The data therefore showed that epitopes on three major keratins, the Mr 40,000, 50,000, and 54,000-55,000 group, were "masked" in normal cells, whereas in tumor cells "masking" involved primarily the Mr 54,000-55,000 keratin. Attempts to "unmask" the epitopes recognized by AE1 in normal cells or to increase the number of cells reactive with AE3 in the normal and tumor cultures failed. Thus, certain cultured preneoplastic and neoplastic mammary cells with a myoepithelial phenotype have an altered organization of keratins that is manifested by a keratin antigenic determinant which is visible by immunocytochemistry in the abnormal cells but not in normal mouse mammary cells. This is the first demonstration that the immunoreactivity of keratins can be modified during neoplastic progression of epithelial cells.