T his study is an attempt to determine whether estrogen could directly regulate human gonadotropin-releasing hormone (GnRH) gene expression. Human GnRH expression vectors were constructed by fusing various 5' flanking regions of the human GnRH gene upstream of the luciferase reporter gene (LUC) or the thymidine kinase promoter linked to the chloramphenicol acetyltransferase reporter gene (CAT). These constructs were transiently transfected into a human choriocarcinoma cell line (JEG-3) and LUC or CAT activity was measured after either no treatment or treatment with various concentrations of estradiol. A stimulatory estrogen response element (ERE) was localized to a 32-bp region between -547 and -516 bp. To determine whether estrogen receptor bound to this region of the gene, we performed DNase I footprinting using purified calf uterine estrogen receptor. DNase I footprinting demonstrates a strong footprint between -567 and -514 bp of the human GnRH gene. In addition, an avidin-biotin complex DNA-binding assay demonstrated that a biotinylated DNA fragment containing -541 to -517 bp of the human GnRH gene bound 35S-labeled estrogen receptor as well as a biotinylated DNA fragment containing the xenopus vitellogenin ERE. On the other hand, the negative control biotinylated DNA fragment derived from adenovirus 5 bound insignificant amounts of 35S-labeled estrogen receptor. Both the GnRH ERE and vitellogenin ERE bound 35S-labeled estrogen receptor with high affinity (approximately 1 nM). These data indicate that the human GnRH gene contains an ERE sufficient to mediate a stimulatory response to estrogen in heterologous cells. Based upon these data we hypothesize that the human GnRH gene might also be directly regulated by estrogen in the hypothalamus, and that this regulation may explain the GnRH hypersecretion observed at the time of the preovulatory luteinizing hormone (LH) surge.