Prostaglandin F2 alpha (PGF2 alpha) secretion is lowest at midcycle and highest on day 15 at luteolysis in the cycling guinea pig uterus and is inversely related to serum progesterone levels. An increase in 17-beta estradiol (E2) occurs only towards the end of the cycle. To investigate the effect of steroids on the control of uterine PGF2 alpha metabolism at the level of gene expression we established a primary cell culture model of day 15 cycling guinea pig endometrial cells. We cloned guinea pig cDNAs for cyclooxygenase 2 (COX-2), 15-hydroxyprostaglandin dehydrogenase (PGDH) that converts PGF2 alpha to biologically inactive 13,14-dihydro-15-keto PGF2 alpha (PGFM) and a fragment of cyclooxygenase-1 (COX-1). They were found to bear 87% and 90% homology at the amino acid level to their human counterparts for COX-2 and PGDH, respectively, retaining all functional sites. Purified epithelial and stromal cell subcultures were primed with medium containing either E2 or medroxyprogesterone acetate (MPA) for 24 h. They were then treated for a further 4 or 24 h either withdrawing the steroid, maintaining the priming steroid, or supplementing with both steroids, before harvesting conditioned media and RNA. Epithelial cells secreted 30-fold more PGF2 alpha compared with stromal cells (e.g. 7.8 +/- 0.7 vs. 0.26 +/- 0.09 pg/ng DNA.24 h), and PGF2 alpha secretion levels were approximately 15-fold higher than those of PGFM (e.g. 7.8 +/- 0.7 vs. 0.45 +/- 0.16 pg/ng DNA.24 h, for epithelial cells). COX-1 transcripts were low and unaffected by treatment in both cell types. COX-2 transcripts were more abundant in epithelial than stromal cells. Steroid-modulated, COX-2 dependent changes in PGF2 alpha secretion were observed. The addition of MPA to E2 primed cells caused a decrease in PGF2 alpha secretion and COX-2 messenger RNA levels after 4 h. Conversely, the addition of E2 to MPA primed epithelial cells led to an increase in PGF2 alpha secretion and COX-2 messenger RNA levels after 4 and 24 h. The withdrawal of E2 caused a fall in PGF2 alpha secretion and COX-2 transcripts after 24 h. In contrast, PGDH transcripts were more abundant in stromal than epithelial cells and were up-regulated by the addition of MPA to E2 primed cells. These in vitro observations are in keeping with the secretory profile seen in vivo in the cycling guinea pig uterus suggesting that 1) the fall of E2 and the coinciding rise in progesterone seen in the early cycle lead to a reduction in PGF2 alpha levels; and 2) the rise of E2 in the late cycle on a progesterone primed uterus is the stimulus for an increase in uterine PGF2 alpha production. Our findings suggest a differential role for uterine stroma and epithelium in vivo whereby the former acts to remove (via PGDH), and the latter to produce (via COX-2) biologically active prostaglandin.