Differential expression of estrogen receptor-β (ERβ) in human pituitary tumors:: Functional interactions with ERα and a tumor-specific splice variant

The mitogenic and regulatory effects of estrogen (E-2) in adenohypophysial cells are known to be mediated through the nuclear estrogen receptor (ER alpha). Expression of ER alpha and several of its messenger ribonucleic acid (RNA) alternate splice variants has been shown to be restricted to prolactinomas and gonadotroph tumors. However, little is known about gene expression patterns of the novel nuclear hormone receptor ER beta in the neoplastic pituitary. ER beta has high homology to ER alpha in the DNA- and ligand-binding domains, but encodes a distinct transcriptional activating function-1 (AF-1) domain. Using RT-PCR analysis of total RNA from 38 human pituitary adenomas, we found that ER beta messenger RNA was coexpressed with ER alpha and its splice variants in 60% of prolactinomas, 100% of mixed GH/PRL tumors, and 29% of gonadotroph tumors. ER beta gene expression was not limited to ER alpha-positive tumor subtypes, however, and was also found in 100% of null cell tumors, 80% of somatotroph tumors, and 60% of corticotroph tumors. Because ER beta is coexpressed with ER alpha and its splice variants in prolactinomas and gonadotroph tumors, we functionally characterized the potential interactions between ER beta and ER alpha. We also examined the potential cooperative effects on ER beta-mediated gene expression of a tumor-specific truncated Delta 5ER alpha splice variant that has been shown to be coexpressed in the majority of ER alpha-positive tumors. This exon 5 splice variant encodes the AF-1 domain as well as regions critical for DNA binding and nuclear localization, but lacks the Ligand-binding and AF-2 domains. Mammalian expression vectors encoding ER alpha, Delta 5ER alpha, and/or ER beta complementary DNAs were transiently transfected along with an E-2 response element promoter-luciferase (ERELuc) reporter into human ER alpha/ER beta-negative osteosarcoma U2-OS cells. ER beta was less potent than ER alpha in activating E-2-stimulated ERELuc activity (4- vs. 14-fold relative to basal control levels). However, when Delta 5ER alpha was coexpressed with ER beta or ER alpha, E-2-stimulated ERELuc activity was markedly increased to 8- and 57-fold, respectively, relative to basal control levels when each full-length isoform was expressed alone. Finally, coexpression of ERP with ER alpha did not significantly alter the E-2-stimulated ERELuc activity induced by ER alpha alone. Cotreatment with tamoxifen markedly inhibited all E-2-stimulated ERELuc responses to baseline levels. Together, these data suggest that ER beta has a minor role in mediating E-2 responses in ER alpha-positive tumors, but may be the main mediator of E-2-stimulated gene expression when expressed alone in somatotroph, corticotroph, and null cell tumors. This low, but significant, level of ER beta trans-activation potential may he enhanced by coexpression of Delta 5ER alpha in neoplastic pituitary. Therefore, E-2-mediated gene expression in normal and neoplastic pituitary appears to be highly dependent on the expression of ER alpha and ER beta isoforms, which have varying transcriptional activities.