We have identified and characterized three human estrogen receptor (ER) mutants, which, at low concentrations, are capable of blocking the intracellular activity of wild type ER. The mutants, a truncated ER (ER1-530), a point mutant (L540Q), and a frameshift (S554fs), were generated by random chemical mutagenesis of the ER hormone binding domain and screened first for low activity in a yeast selection system. In transient co-transfection assays using ER-deficient Chinese hamster ovary cells, all three mutants exhibited less than 10% of the transcription activation activity of wild type ER, and when co-expressed with wild type ER, each of the mutants effectively suppressed the ability of wild type ER to activate transcription of an estrogen-regulated reporter plasmid. When equal amounts of plasmid encoding the ER mutants and wild type ER were used, S554fs, ER1-530, and L540Q suppressed the activity of wild type ER by 80, 55, and 75%, respectively. At a ratio of 1 part S554fs to 10 parts wild type ER, transcription was still inhibited by 40%. Western blot analysis showed that all three mutants were expressed at approximately the same level as wild type ER. Suppression of transcription was specific for ER, since the mutants did not inhibit progesterone receptor-mediated transcription. Not all mutations leading to inactive ER confer the dominant negative phenotype, as five ER mutants rendered transcriptionally inactive by point mutations between residues 516 and 524 of the ER hormone binding domain were poor inhibitors of wild type ER activity. Binding studies showed that the L540Q and S554fs dominant negative mutants bound 17 beta-estradiol with wild type affinity (Kd = 0.3-0.5 nM), whereas ER1-530 exhibited a 15-fold reduction in affinity for estradiol. The three dominant negative ERs showed significant ability to interact with the estrogen response element (ERE) in promoter interference assays, but ER1-530 and S554fs displayed little or no binding to the ERE in gel mobility shift assays where higher affinity for the DNA may be required for the receptor-ERE complex to remain associated during the electrophoresis. These data support the idea that, in all three mutants, it is loss of function of the COOH-terminal transactivation domain which leads to the dominant negative phenotype. S554fs, a powerful dominant negative mutant, is a good candidate for further studies aimed at suppressing the estrogen-dependent growth of human breast cancer cells.