Retinoids and their derivatives can suppress the development of cancer in animals and in humans. We and others have shown that retinoid X receptor (RXR)-selective retinoids or "rexinoids" suppress the development of breast cancer in several animal models with minimal toxicity. LGD1069 (Bexarotene) is a potent RXR-selective retinoid with reduced toxicity compared with naturally occurring retinoids. In this study, we investigated the expression of LGD1069-modulated biomarkers. We previously did cDNA array analysis of LGD1069-treated breast cells using Affymetrix microarrays. These studies identified many LGD1069-regulated genes, one of which was cyclooxygenase-2 (COX-2). Because COX-2 inhibitors have been shown to prevent cancer in other model systems, we investigated whether LGD1069 inhibits the expression of COX-2 in mammary tissue and in normal human mammary epithelial cells (HMEC). In mouse mammary tumor virus-erbB2 mice treated with LGD1069, there was a marked decrease of COX-2 expression in both normal and malignant mammary tissues. The effect of LGD1069 on COX-2 expression was also investigated in normal human breast cells. COX-2 expression was markedly reduced by treatment with LGD1069 at the RNA and protein level in normal HMECs; LGD1069 suppressed COX-2 promoter activity. We also showed that LGD1069 inhibited activator protein (AP-1)-dependent transcription in these breast cells, and that suppression of COX-2 expression was due to sequestration of CBP/p300. These results from in vivo and in vitro studies suggest that LGD1069, an RXR-selective retinoid, inhibits COX-2 expression by suppression of COX-2 transcription in part through transrepression of the AP-1 transcription factor. Thus, RXR-selective retinoids that inhibit AP-1 activity and suppress COX-2 expression may be particularly promising drugs for breast cancer prevention. Furthermore, such RXR-selective retinoids may be most useful in combination with antiestrogens for more effective prevention of breast cancer in women at high risk of this disease.