This study utilizes the mammalian two-hybrid system to examine the role of ligand in the dimerization of human progesterone receptor (hPR). The GAL4 DNA-binding domain and the herpes simplex virus VP16 transactivation domain were fused to the amino terminus of full-length hPR (both the A and B isoforms) to produce chimeric proteins. PR dimerization was detected by the ability of cotransfected GAL4/PR and VP16/PR chimeras in COS cells to induce expression of a reporter gene under the control of GAL4-binding sites (pG5CAT). Hormone agonist-dependent interactions were observed between the two like isoforms of PR (A-A and B-B) and between PR-A and PR-B (A-B), indicating that hormone can stimulate the formation of the three possible dimeric forms of PR within cells. In contrast, neither type I (ZK98299) nor type II (RU486, ZK112993) progestin antagonists stimulated interaction between these same hybrid PR proteins. However, activation of the VP16/PR chimera by antagonists on a progesterone response element-controlled reporter gene (DHRE-E1b-CAT) was only a fraction (4-13%) of that stimulated by agonist R5020. One possibility for the failure to detect an induction in the two-hybrid assay is antagonist-induced repression of the activity of the VP16/PR fusion protein rather than a failure of antagonists to stimulate interaction between the hybrid proteins. To test this idea, an UP-1 carboxyl-terminal truncation mutant of PR was used to construct the two-hybrid proteins. PR-UP-1 selectively binds antagonists, but not agonists, and is fully activated in response to antagonists. Both types of progestin antagonists stimulated interactions between GAL4/PR(UP-1) and VP16/PR(UP-1) hybrid proteins, indicating that antagonists are capable of stimulating PR dimerization in cells and do not function by disrupting or preventing dimerization. To determine whether PR bound to an antagonist can dimerize in whole cells with PR bound to agonist, GAL4/PR(UP-1) was paired in the two-hybrid assay with a VP16/PR fusion protein harboring a point mutation in PR at amino acid 722 (Gly-Cys) that specifically binds progestin agonist but not antagonist. Neither R5020 nor RU486 alone stimulated interaction between these ligand-specific PR hybrid proteins. However, strong interaction was detected by addition of both agonist and antagonists, indicating the formation of mixed ligand heterodimers and that both PR partners require ligand for dimerization to occur. Based on electrophoretic gel mobility shift assays (EMSAs), these heterodimers appear to have substantially reduced DNA binding activity. Progestin antagonists inhibit agonist activation of PR at concentrations that are too low to be accounted for by a simple competition mechanism for binding to PR. We propose that antiprogestin inactivation of PR in trans by heterodimerization contributes to the biological potency of these compounds.