RESPONSE OF NORMAL AND ONCOGENE-TRANSFORMED HUMAN MAMMARY EPITHELIAL-CELLS TO TRANSFORMING GROWTH-FACTOR-BETA-1 (TGF-BETA-1) - LACK OF GROWTH-INHIBITORY EFFECT ON CELLS EXPRESSING THE SIMIAN-VIRUS 40 LARGE-T ANTIGEN

The relationship between the expression of selected oncogenes having d ifferent modes of action and the loss of the capacity to respond in vi tro to transforming growth factor-beta 1 (TGF-beta 1) was analyzed in human mammary epithelial cells (MEC). Primary MEC cultures from health y donors and the spontaneously immortalized MCF-10A cell line were use d as normal controls. Various assays (employing both complete and chem ically defined media) were used: short-term DNA synthesis, long-term c ell proliferation under anchorage-dependent and -independent condition s, expression of surface-differentiation molecules. Whereas primary ME C and the MCF-10A cell line were fully responsive to the growth-inhibi tory activity of TGF-beta 1 under different test conditions, MEC trans formed by c-Ha-ras, c-erbB2, int-2 or SV40-large-T antigen were not in hibited by TGF-beta 1 in a short-term DNA-synthesis assay. However, in anchorage-dependent conditions TGF-beta 1 inhibited the proliferation of all lines investigated, with the exception of SV40-T-antigen-trans formed MEC. The colony-formation assay in soft agar revealed that all lines, but not those expressing the int-2 or the SV40-T-antigen genes, were inhibited by TGF-beta 1. Neutralizing antibody to TGF-beta had n o significant effects on oncogene-transformed lines, suggesting that t he endogenous production of an active form of this growth factor is no t a major determinant in MEC transformation by the oncogenes investiga ted. The only observed effect of TGF-beta 1 on selected surface-differ entiation molecules was that normal MEC produced increased levels of t he human milk fat globule antigen-1. Thus it appears that the response of MEC to TGF-beta 1 is consistently attenuated by the insertion of a variety of oncogenes and that it is abolished only by the expression of the SV40-large-T antigen. Whereas no single in vitro assay was capa ble of accurately reflecting the actual responsiveness of different li nes, the growth-curve assay in anchorage-dependent conditions was the best single predictive test. (C) 1994 Wiley-Liss, Inc.