Synergism of closely adjacent estrogen-responsive elements increases their regulatory potential

Recent gene transfer experiments have shown that an estrogen-responsive DNA element (ERE) GGTCANNNTGACC mediates the estrogen inducibility of the Xenopus laevis vitellogenin A1 and A2 genes as well as the chicken vitellogenin II gene. We report here on experiments that explain the estrogen regulation of the Xenopus vitellogenin B1 and B2 genes. In these genes, two ERE homologues, which have only low, if any, regulatory capacity on their own, act synergistically to achieve high estrogen inducibility. Furthermore we show that synergism of EREs is most efficient, when the two elements are closely adjacent and that it is lost when the synergistic elements are separated by 125 basepairs. In-vitro estrogen receptor binding experiments indicate that co-operative binding of estrogen receptors to closely adjacent EREs is not essential for synergism of ERE homologues that have no intrinsic regulatory capacity. Functional synergism of EREs is observed in the human estrogen-responsive MCF-7 cell line as well as in mouse fibroblasts (Ltk-) cotransfected with estrogen receptor expression vectors. Even expression of a truncated receptor protein lacking 178 amino acid residues of the amino-terminal end allows synergism, suggesting that the amino-terminal end preceding the DNA-binding domain of the estrogen receptor is not required.     

The uteroglobin promoter contains a noncanonical estrogen responsive element

Uteroglobin is expressed in various tissues of the rabbit under complex hormonal control. In the endometrium the uteroglobin gene is transcribed only in epithelial cells after administration of ovarian hormones. In this paper we demonstrate that within the promoter region of the rabbit uteroglobin gene, there is a functional estrogen-responsive element (ERE) located between -265 and -252. Hybrid constructions containing sequences of the uteroglobin promoter up to -299, linked to the chloramphenicol acetyltransferase gene of E. coli respond to estrogens in gene transfer experiments, whereas a deletion that removes half of the ERE does not. A synthetic oligonucleotide corresponding to the putative ERE is able to confer estrogen inducibility to an otherwise unresponsive promoter. Binding experiments with purified estrogen receptor from calf uterus reveal a DNase-I footprint over the ERE. Within this protected region six guanine residues that have been shown to be contacted by the receptor in other EREs are protected against methylation by dimethylsulfate in the presence of the estrogen receptor. We compare this ERE with the vitellogenin A2 ERE from Xenopus and find that the relative affinity of the uteroglobin ERE is slightly lower than that of the vitellogenin ERE. Thus, this uteroglobin ERE could be involved in physiological regulation of uteroglobin expression in the genital tract.     

A far upstream estrogen response element of the ovalbumin gene contains several half-palindromic 5'-TGACC-3' motifs acting synergistically.

We have identified an estrogen-responsive enhancer element (DH3 ERE) in the estrogen-induced DNAase I-hypersensitive region III of the chicken ovalbumin gene, which is located approximately 3.3 kb upstream from the mRNA start site and does not contain palindromic ERE. Four TGACC half-palindromic motifs, separated from each other by more than 100 bp, are responsible for conferring estrogen inducibility either to the proximal ovalbumin gene promoter or to heterologous promoters. Thus, widely spaced half-palindromic ERE motifs can act synergistically. Each half-palindromic motif was shown to bind the estrogen receptor (ER) with a low efficiency in vitro. However, two widely spaced half-palindromic motifs bound the ER cooperatively, much more efficiently than expected from binding to isolated half-ERE motifs. The ovalbumin promoter half-palindromic ERE motif located close to the TATA box was required for the activity of the distal DH3 ERE, but could be replaced by the binding sites of other transactivators.     

Introduction of estrogen-responsiveness into mammalian cell lines.

We introduced estrogen responsiveness as a new characteristic into rat hepatoma, mouse Ltk- and human HeLatk-cells by transfecting the human estrogen receptor (ER) cDNA. To measure the estrogen response we used Xenopus vitellogenin gene A2 constructs linked to the bacterial CAT gene. Transient cotransfections of the ER cDNA and the vitellogenin gene-CAT constructs containing the estrogen responsive element (ERE) lead to a hormone dependent induction of CAT activity whereas cotransfected vitellogenin gene constructs lacking the ERE are not inducible. Stable transfections of ER cDNA into Ltk- cells give rise to cell clones that are estrogen responsive as shown by transfection of various vitellogenin gene-CAT constructs. These results prove that the transfected ER is biologically active and is sufficient to make a cell estrogen responsive.     

Functional interaction of hybrid response elements with wild-type and mutant steroid hormone receptors.

Steroid hormone receptors can be divided into two subfamilies according to the structure of their DNA binding domains and the nucleotide sequences which they recognize. The glucocorticoid receptor and the progesterone receptor (PR) recognize an imperfect palindrome (glucocorticoid responsive element/progesterone responsive element [GRE/PRE]) with the conserved half-sequence TGTYCY, whereas the estrogen receptor (ER) recognizes a palindrome (estrogen responsive element) with the half-sequence TGACC. A series of symmetric and asymmetric variants of these hormone responsive elements (HREs) have been tested for receptor binding and for the ability to mediate induction in vivo. High-resolution analysis demonstrates that the overall number and distribution of contacts with the N-7 position of guanines and with the phosphate backbone of various HREs are quite similar for PR and ER. However, PR and glucocorticoid receptor, but not ER, are able to contact the 5'-methyl group of thymines found in position 3 of HREs, as shown by potassium permanganate interference. The ER mutant HE84, which contains a single amino acid exchange, Glu-203 to Gly, in the knuckle of ER, creates a promiscuous ER that is able to bind to GRE/PREs by contacting this thymine. Elements with the sequence GGTCAcagTGTYCT that represent hybrids between an estrogen response element and a GRE/PRE respond to estrogens, glucocorticoids, and progestins in vivo and bind all three wild-type receptors in vitro. These hybrid HREs could serve to confer promiscuous gene regulation.     

Binding of type II nuclear receptors and estrogen receptor to full and half-site estrogen response elements in vitro.

The mechanism by which retinoids, thyroid hormone (T3) and estrogens modulate the growth of breast cancer cells is unclear. Since nuclear type II nuclear receptors, including retinoic acid receptor (RAR), retinoid X receptor (RXR) and thyroid hormone receptor (TR), bind direct repeats (DR) of the estrogen response elements (ERE) half-site (5'-AGGTCA-3'), we examined the ability of estrogen receptor (ER) versus type II nuclear receptors, i.e. RARalpha, beta and gamma, RXRbeta, TRalpha and TRbeta, to bind various EREs in vitro . ER bound a consensus ERE, containing a perfectly palindromic 17 bp inverted repeat (IR), as a homodimer. In contrast, ER did not bind to a single ERE half-site. Likewise, ER did not bind two tandem (38 bp apart) half-sites, but low ER binding was detected to three tandem copies of the same half-site. RARalpha,beta or gamma bound both ERE and half-site constructs as a homodimer. RXRbeta did not bind full or half-site EREs, nor did RXRbeta enhance RARalpha binding to a full ERE. However, RARalpha and RXRbeta bound a half-site ERE cooperatively forming a dimeric complex. The RARalpha-RXRbeta heterodimer bound the Xenopus vitellogenin B1 estrogen responsive unit, with two non-consensus EREs, with higher affinity than one or two copies of the full or half-site ERE. Both TRalpha and TRbeta bound the full and the half-site ERE as monomers and homodimers and cooperatively as heterodimers with RXRbeta. We suggest that the cellular concentrations of nuclear receptors and their ligands, and the nature of the ERE or half-site sequence and those of its flanking sequences determine the occupation of EREs in estrogen-regulated genes in vivo .     

The human estrogen receptor can regulate exogenous but not endogenous vitellogenin gene promoters in a Xenopus cell line.

Transfection of a human estrogen receptor cDNA expression vector (HEO) into cultured Xenopus kidney cells confers estrogen responsiveness to the recipient cells as demonstrated by the hormone dependent expression of co-transfected Xenopus vitellogenin-CAT chimeric genes. The estrogen stimulation of these vit-CAT genes is dependent upon the presence of the vitellogenin estrogen responsive element (ERE) in their 5' flanking region. Thus, functional human estrogen receptor (hER) can be synthesized in heterologous lower vertebrate cells and can act as a trans-acting regulatory factor that is necessary, together with estradiol, for the induction of the vit-CAT constructs in these cells. In addition, vitellogenin minigenes co-transfected with the HEO expression vector also respond to hormonal stimulation. Their induction is not higher than that of the vit-CAT chimeric genes. It suggests that in the Xenopus kidney cell line B 3.2, the structural parts of the vitellogenin minigenes do not play a role in the induction process. Furthermore, no stabilizing effect of estrogen on vitellogenin mRNA is observed in these cells. In contrast to the transfected genes, the endogenous chromosomal vitellogenin genes remain silent, demonstrating that in spite of the presence of the hER and the hormone, the conditions necessary for their activation are not fulfilled.     

Chromatin studies reveal that an ERE is located far upstream of a vitellogenin gene and that a distal tissue-specific hypersensitive site is conserved for two coordinately regulated vitellogenin genes.

Estrogen induces the expression of three vitellogenin genes in chicken hepatocytes. To survey the vitellogenin III (VTGIII) gene region for possible distal regulatory sequences, we identified tissue-specific hypersensitive (HS) sites within a 45 kb chromatin region spanning this gene. Five constitutive HS sites were found to mark the VTGIII gene region in hormone-naive hepatocytes. Strikingly, the constitutive HS site located 5.5 kb upstream of the VTGIII gene and a previously identified HS site located within the coordinately regulated VTGII gene mapped to nearly identical copies of a 72 bp sequence. Moreover, it would appear that there has been evolutionary pressure to retain specifically this 72 bp of VTGII-like sequence near the VTGIII gene subsequent to the VTGIII and VTGII genes becoming unlinked approximately 16 Myr ago. Two additional sets of HS sites were induced in the VTGIII gene region in response to estrogen. One set mapped immediately upstream of the gene in the vicinity of what we show to be a functional estrogen response element (ERE). The other induced HS site mapped 7.5 kb upstream of the gene. This far-upstream region was sequenced and was found to contain two imperfect ERE consensus sequences spaced 88 bp apart. In transient expression assays neither of these individual imperfect ERE sequences was functional, but a fragment spanning both sequences behaved as a strong ERE. In contrast to this synergism between imperfect ERE sequences, the presence of an NF-1 binding site 23 bp away from the more distal imperfect ERE sequence was not sufficient to render the latter a functional ERE in our assays.     

Two functional estrogen response elements are located upstream of the major chicken vitellogenin gene.

We used a transient-expression assay to identify two estrogen response elements (EREs) associated with the major chicken vitellogenin gene (VTGII). Each element was characterized by its ability to confer estrogen responsiveness when cloned in either orientation next to a chimeric reporter gene consisting of the herpes simplex virus thymidine kinase promoter and the chloramphenicol acetyl transferase-coding region. Deletion analyses indicated that sequences necessary for the distal ERE resided within the region from -626 to -613 (nucleotide positions relative to the VTGII start site) whereas those necessary for the proximal ERE were within the region from -358 to -335. These distal and proximal elements contain, respectively, a perfect copy and an imperfect copy of the 13-base-pair sequence that is an essential feature of the EREs associated with two frog vitellogenin genes. These chicken VTGII EREs mapped near regions that were restructured at the chromatin level when the endogenous VTGII gene was expressed in the liver in response to estradiol. These data suggest a model for the tissue-specific expression of this estrogen-responsive gene.