Retinoic acid (RA) plays complex roles in the development of breast cancer. While RA has been generally
perceived as preventing tumor formation, it has also been suggested to act in some models as a promoter of tumor progression. Little is known however about how RA is synthesized in the normal mammary gland and in tumors. We have shown that RA is made in mammary cells by a specific enzyme, RALDH3, which has a heterogeneous pattern of expression in breast tumors. Tumor types that are sensitive to the anti-proliferative effects of RA contain as a rule low levels of this protein, while tumors that express it to high levels are usually insensitive and may even use RA synthesis to their benefit. We have generated mouse models in which this enzyme is deleted and propose to use them to investigate its role in the normal mammary gland and in tumors that are promoted by RA.