Retinoic acid inhibits proliferation and steroid receptor gene expression in human breast cancer cell lines. Retinoic acid receptors (RAR)alpha, -beta, and -gamma are expressed in these cells and the expression of RAR alpha is significantly greater in estrogen receptor (ER)-positive cells. This study was undertaken to determine whether the same relationship between RAR alpha and ER gene expression was present in human breast cancers and to explore the possibility that the higher level of RAR alpha in ER-positive cells was due to estrogen regulation of RAR alpha gene expression. RAR alpha and ER mRNA expression were determined by Northern blot analysis in 116 primary breast tumors; 94 (81%) tumors were ER-positive and of these 87 (93%) were also RAR alpha-positive. The coexpression of ER and RAR alpha was statistically significant (P = 0.0052 by chi 2 contingency analysis). There was also a positive correlation (by linear regression analysis) between the levels of expression of ER and RAR alpha mRNA (r2 = 0.251, P = 0.0001), which confirmed the relationship previously documented in breast cancer cell lines and suggested that RAR alpha expression may be modulated in breast cancer in vivo by estrogens acting via the ER. The ability of estradiol to regulate RAR alpha gene expression was examined in vitro using T-47D cells which had been rendered sensitive to estrogen by repeated passage in steroid-depleted medium. Estradiol increased RAR alpha gene expression, but not that of RAR beta or RAR gamma, in a concentration-dependent manner, with the effect being maximal at 10(-10) M and less marked at higher concentrations. The effect was rapid, being detectable 1 h after and maximal 6 h after treatment with 10(-10) M estradiol. Co-treatment of cells with estradiol and antiestrogens (tamoxifen or ICI 164384, 4 x 10(-7) M for 6 h) inhibited the estradiol induction of RAR alpha gene expression, demonstrating that the effect was ER mediated. The estradiol sensitivity of the effect was underscored by the demonstration that addition of untreated serum to cells growing under steroid-depleted conditions was sufficient to induce maximal RAR alpha gene expression. This effect was totally abolished by addition of ICI 164384. In summary, the demonstration that estradiol increased RAR alpha mRNA levels in breast cancer cells supports the hypothesis that the correlation between RAR alpha and ER gene expression in breast tumors and breast cancer cell lines is due to estradiol augmentation of RAR alpha gene expression.