Cutaneous melanoma is often resistant to chemo- and radiotherapy. This resistance has recently been demonstrated to be due, at least in part, to high activating transcription factor 2 (ATF-2) activity in these tumors. In concordance with these reports, we found that B16 mouse melanoma cells had higher levels of ATF-2 than immortalized, but non-malignant mouse melanocytes. In addition, the melanoma cells had a much higher amount of phosphorylated (active) ATF-2 than the immortalized melanocytes. In the course of determining how retinoic acid (RA) stimulates activating protein-1 (AP-1) activity in B16 melanoma, we discovered that this retinoid decreased the phosphorylation of ATF-2. It appears that this effect is mediated through p38 MAPK, because RA decreased p38 phosphorylation, and a selective inhibitor of p38 MAPK (SB203580) also inhibited the phosphorylation of ATF-2. Since ATF-2 activity appears to be involved in resistance of melanoma to chemotherapy, we tested the hypothesis that treatment of the melanoma cells with RA would sensitize them to the growth-inhibitory effect of taxol. We found that pretreatment of B16 cells with RA decreased the IC50 from 50 nM to 1 nM taxol. On the basis of these findings and our previous work on AP-1, we propose a model in which treatment of B16 cells with RA decreases the phosphorylation of ATF-2, which results in less dimer formation with Jun. The "freed-up" Jun can then form a heterodimer with Fos, resulting in the increased AP-1 activity observed in RA-treated B16 cells. Shifting the balance from predominantly ATF-2:Jun dimers to a higher amount of Jun:Fos dimers could lead a change in target gene expression that reduces resistance to chemotherapeutic drugs and contributes to the pathway by which RA arrests proliferation and induces differentiation.