Human Estrogen Receptor Forms Multiple Protein-DNA Complexes*

A baculovirus expression system was used to over-produce the human estrogen receptor in insect cells. The estrogen receptor made in this system is full- length, binds estrogen specifically, and is recognized by a monoclonal antibody to the human estrogen recep- tor. The recombinant estrogen receptor binds the estrogen response element (ERE) in both the absence and presence of estrogen if the binding is carried out in the absence of Me. In the presence of Mg’+, the estrogen receptor binds the ERE in a hormone-dependent fash- ion. This effect is more pronounced at higher temperatures.

A baculovirus expression system was used to overproduce the human estrogen receptor in insect cells. The estrogen receptor made in this system is fulllength, binds estrogen specifically, and is recognized by a monoclonal antibody to the human estrogen receptor. The recombinant estrogen receptor binds the estrogen response element (ERE) in both the absence and presence of estrogen if the binding is carried out in the absence of Me.
In the presence of Mg'+, the estrogen receptor binds the ERE in a hormone-dependent fashion. This effect is more pronounced at higher temperatures.
Tamoxifen, a nonsteroidal anti-estrogen, is able to stimulate ERE binding to the same extent and under the same conditions as estradiol. Estradiol stimulates formation of an estrogen receptor-ERE complex with an increased mobility in native gels as compared with the complex formed without hormone or with tamoxifen.
These results demonstrate that specific DNA binding of the estrogen receptor is not absolutely dependent on the presence of hormone and that estradiol but not tamoxifen is able to induce a change in the estrogen receptor. This differential effect of estradiol and tamoxifen may be important in understanding the role of the receptor to activate target genes differentially.
Estrogen plays an important role in the growth and differentiation of a number of normal tissues in mammals as well as a pivotal part in the regulation of reproduction. In addition, the activity of the estrogen receptor significantly influences the behavior and treatment of greater than a third of human breast cancers (1). The human estrogen receptor is a member of a superfamily of nuclear receptors for small hydrophobic ligands including the steroid hormones, thyroid hormone, and retinoic acid (2). As a class, these receptors are transcription factors whose activity is regulated allosterically by hormone binding. A dissection of the molecular mechanism by which estrogen activates its receptor is central to our understanding of the complex biology governed by estrogen. The molecular cloning of the estrogen receptor has allowed significant advances in our understanding of the mechanism of its action (3-5). Utilizing transient co-transfection assays of the estrogen receptor and reporter plasmids, several functional domains of the estrogen receptor have been defined (6-8). These include a DNA-binding domain containing two zinc fingers, a hormone-binding domain (7), a hormonally inducible transactivation domain (9), and a constitutive transactivation domain (10). This domain model of the estrogen receptor and the other members of the steroid receptor family is borne out by studies in which functional chimeric receptors have been constructed which display the steroid-binding specificity of one receptor linked to the DNA-binding specificity of a second (11). Additionally, various domains of the estrogen receptor are also functional when linked to unrelated transcription factors (12) or even to a protein such as myc whose mechanism of action is unknown (13). In addition, the human estrogen receptor has been shown to be functional when expressed in yeast (14). The mechanism by which the estrogen receptor activates transcription is unknown. An important step in the activation is the recognition by the estrogen receptor of a specific DNA element termed the estrogen response element (ERE).' This element has been identified in the promoters of a number of estrogen-responsive genes. The ERE from the vitellogenin promoter consists of a 13-base pair palindromic sequence and represents a consensus binding site (15-M). An analysis of the specific estrogen receptor-ERE complexes will allow a greater understanding of the requirements for early steps in the regulation of gene expression by the estrogen receptor.
Tamoxifen, a triphenylethylene anti-estrogen, is widely used in the treatment of breast cancer, yet the mechanism by which tamoxifen antagonizes the action of estradiol is unknown. Tamoxifen competes with estradiol for binding of the estrogen receptor (19). In addition, a derivative of tamoxifen, 4-hydroxytamoxifen, has been shown to stimulate dimerization and sequence-specific DNA binding of the estrogen receptor (20). Thus, tamoxifen differs from estradiol in activity at a step distal to the induction of estrogen receptor-ERE complexes. An understanding of this difference will provide significant insight into the functioning of the estrogen receptor.
We have used recombinant human estrogen receptor made in a baculovirus expression system to study the role of estrogen in the stimulation of sequence-specific DNA binding in uitro.
A model is proposed in which the estrogen receptor-ERE complex can exist in at least three distinct states and in which these are responsible for the differences in biological activity of estrogen, estrogen withdrawal, and tamoxifen. at 20 "C for 20 min. The protein-DNA complexes were separated on 4% (30:1, acrylamide to bisacrylamide) polyacrylamide gels using Tris/glycine buffer, pH 8.3 (25). The gels were dried and subjected to autoradiography.

Expression of the Human Estrogen Receptor in Insect
Cells-A baculovirus expression system (21) was used to overexpress the human estrogen receptor (ER). The plasmid HE0 (7) was used as the source of the human ER cDNA. It has been shown recently that this plasmid contains a point mutation in the hormone-binding domain of the ER (Gly*" for VaYoO) which decreases its affinity for estradiol (26). The coding region for the ER was cloned into the transfer vector pVL941 under the control of the polyhedron promoter for expression in insect cells (21). To obtain the recombinant viral stock, the resulting plasmid pVLER was co-transfected with wild-type baculovirus AcNPV DNA into Sf9 cells. The ER recombinant virus (AcNPV-ER) was identified using plaque hybridization, and the positive plaques were purified through three further rounds of plaque purification. The approximate expression level of the human ER in this system was determined by an SDS-polyacrylamide gel analysis of total protein extracts made from infected cells (Fig. 1). A prominent band at M, 67,000 was present in extracts from the ER recombinant virus-infected cells (lane 3) but was not evident in extracts from the mock infected or wild-type virusinfected cells (lanes 1 and 2, respectively). We estimate that the ER represents 3% of the total protein in infected cells. In order to confirm that this protein being overexpressed in the system was in fact the human ER, Western blotting was performed using the D75 monoclonal antibody made against human ER (Fig. 2). A prominent band of 67 kDa was observed in the ER recombinant extract (lane 4) whereas no crossreacting material was seen in either mock infected or wildtype-infected cell extracts (lanes 2 and 3, respectively). Several species that migrated as a series of bands slower than that of the full-length ER reacted with the D75 antibody (lane 4). These may represent post-translational modifications of the ER protein.
Extracts from Sf9 cells infected with the ER recombinant or control virus were assayed for estrogen binding using the dextran-coated charcoal method (19). There was no specific estrogen binding in either mock infected or control virusinfected cell extracts. of a specific estrogen response element. Extracts of the ER recombinant virus-infected cells generated a specific DNAprotein complex (Fig. 3,  Formation of the estrogen receptor-ERE complex using the "'P-labeled ERE probe either in the absence ((-) lanes 1 and 3) of estradiol or the presence ((+) lanes 2 and 4) of 0.1 /JM estradiol (E2) assayed by gel mobility.
Binding was carried out at 20 "C for 20 min followed by either 15 min of additional incubation at 20 "C in the absence of Me (lanes 1 and 2) or 15 min of additional incubation at 37 "C in the presence of 5 mM Mg".
of the ERE, EREmtl; an unrelated sequence from the promoter of the Xenopus estrogen receptor, XERES"; and a sequence from the human ER promoter, ERCOUP," which contains a sequence homologous to the COUP sequence found in the ovalbumin promoter (27). These sequences contain a perfect half-site of the palindromic ERE. The specific complex formed between the ER and the '"P-labeled ERE probe (Fig.  3, lane 1) was effectively competed by a 20-fold molar excess of unlabeled ERE (Fig. 3, lane 2). This complex was not significantly competed by the other oligonucleotides tested (Fig. 3, lanes 3-5). These results indicate that the human ER overexpressed in this system recognizes the ERE specifically. It is important to note that a fragment containing an ERE half-site, ERCOUP, did not bind the ER and must have an affinity at least 20-fold lower than the palindromic ERE. The initial experiments testing the specific binding of the ER were carried out at 20 "C in the absence of Mg"+. A specific ER.ERE complex was observed in both the presence and absence of estradiol (Fig. 4, lanes 1 and 2). In addition, there was no significant difference in the total amount of specific complex in the presence or absence of estradiol over a wide range of DNA and protein concentrations (data not shown). The most visible effect of addition of estradiol in these experiments was a small but reproducible increase in the mobility of the complex (Fig. 4, lanes 1 and 2). This increase in mobility was observed over the entire range of probe and protein concentrations tested. The increase in mobility was not due to the chemical presence of the steroid, as the nonsteroidal estrogen diethylstilbestrol had the same effect (data not shown).
In order to determine the conditions for hormone-dependent binding of the ER, the effects of incubation temperature and ionic environment were investigated. When complexes preformed by incubation at 20 "C and in the absence of Mg2 were incubated further in the presence of 5 mM Mg2+ and at 37 "C, a significant estradiol requirement for binding was observed (Fig. 4, lanes 3 and 4). In the absence of hormone, the ER-ERE complex was not stable under these conditions. In the presence of estradiol, these changes in incubation conditions did not significantly alter the total amount of complex formed or its increased mobility in native gels (Fig.  4, lanes 2 and 4).
The effects of temperature and ionic environment on the hormonal dependence of the binding of the ER to the ERE were next assayed independently. In the absence of Mg2+, DNA binding was observed in both the presence and absence of estradiol at all of the temperatures tested (Fig. 5). At 4 "C (lanes 1 and 2) and 20 "C (lanes 3 and 4) the level of binding was almost equivalent with or without estradiol, but at 30 "C (lanes 5 and 6) and 37 "C (lanes 7 and 8) a significant dependence of binding upon estradiol was observed. The previously noted shift in the mobility of the complex upon addition of estradiol was reproduced at all temperatures tested. The hormonal dependence of binding was tested when reactions were incubated at different ionic strengths (Fig. 6). At 20 "C, an increased Mg2+ concentration restored the estradiol dependence of ER binding. This effect was observed at 5 mM Mg2 and was more significant at 10 mM Mg2' (Fig. 6, lanes 7-10). In general, high KC1 concentrations decreased the level of binding in both the presence and absence of estradiol although a differential effect was observed at 400 mM KC1 (Fig. 6, lanes  15 and 16).
Full hormone dependence for binding of the ER required incubation at both elevated temperature and Mg2' concentrations. The effects of temperature and Mg2+ concentration on the hormonal dependence of ER binding were apparently additive. The effects of increasing the temperature from 20 to 30 "C and the Mg2+ concentration from 0 to 5 mM summed to Formation of the estrogen receptor-ERE complex using the "P-labeled ERE probe either in the absence ((-) lanes I, 3,5, and 7) of estradiol or the presence ((+) lanes 2,4, 6, and 8) of 0.1 pM estradiol (E2) assayed by gel mobility.
Binding was carried out in the absence of Mg*+ for 20 min at 4 "C (lanes I and 2), at 20 "C (lanes 3 and 4), at 30 "C (lanes 5 and 6), or at 37 "C (lanes 7 and 8). account for the total binding at the higher temperature and ionic conditions. Tamorifen Induces a Novel ER -ERE Complex-The ability of tamoxifen, a nonsteroidal anti-estrogen, to stimulate the binding of ER to the ERE was also analyzed using the gel mobility assay (Fig. 7). Significantly, tamoxifen was able to stimulate ERE binding to the same extent as estradiol even in the presence of MgZ+ and at elevated temperature (Fig. 7B,  lanes 2-5). However, tamoxifen did not induce the increase in the mobility of the ER. ERE complex which was observed with estradiol (Fig. 7, A and B, lanes 2-5). When increasing concentrations of tamoxifen were added to a constant satu-Multiple Estrogen Receptor-DNA Complexes rating amount of estradiol in the binding reaction, the mobility of the complex shifted to the slower moving form (Fig. 7,   A and B, lanes 6-8). The requirement for a loo-fold higher concentration of tamoxifen than estradiol is due to its approximately loo-fold lower affinity for the ER (19). Thus, tamoxifen induces a specific ER. ERE complex with a mobility in native gels which is similar to that formed in the absence of hormone at low temperature and ionic strengths. It differs from this complex in that it is stable at high temperatures and ionic strengths. This novel complex with properties of both the complex formed in the absence of hormone and the estradiol-induced complex may contribute to the weak estrogenie properties of tamoxifen reported in some systems (28). DISCUSSION Unraveling of the complex biology governed by estrogen requires an understanding at a molecular level of the mechanism by which estrogen regulates the ER as a transcription factor. The receptor must be modified by the binding of the steroid, and it is important to determine its various states.
Antagonists of estrogen action such as tamoxifen offer the ability to probe the activity of the ER under alternate conditions and may offer clues to the mechanism of ER action. We have produced the human ER in a baculovirus expression system and have studied the conditions influencing the formation of the ER. ERE complex.
At least three distinct ER.ERE complexes exist. In the absence of hormone, a complex forms between the ER and the ERE if Mg*+ is omitted from the binding reaction. Either raising the temperature of the binding reaction or addition of Mg2+ suppresses the formation of this ER.ERE complex. Binding of the ER requires dimerization, and the physical interactions being affected by these changes could either be responsible for dimerization or DNA binding. The formation of this ER.ERE complex implies that there may be some level of ERE occupancy even in the absence of estrogen in uiuo. This may function in some genes to control the basal level of transcription and may allow these genes to be transcriptionally poised for the addition of an estrogenic signal. Recent work with the thyroid hormone receptor has shown that it too can bind its response element in the absence of hormone (29,30). In view of this, it is noteworthy that in certain cell types, ER-dependent transcriptional activation is seen in the absence of exogenously added estrogen (26). Thus, the ER.ERE complex may form in uivo in the absence of estrogen and provide an additional level of gene regulation.
The ER.ERE complex formed in the presence of estradiol can be distinguished from the no hormone complex on the basis of two criteria. These are a differential sensitivity of the complexes to Mg2+ and an increased mobility of the estradiolinduced complex in native gels. The complex formed in the presence of estradiol was shown to be more stable than that formed without hormone. Only the former complex was stable at 37 "C and in the presence of 5 mM Mg". It should be noted that under these latter conditions, binding of ER to the ERE is estrogen dependent. The basis for this increased stability is unclear, but it may involve stabilization of ER dimers by hormone under these conditions. Alternately, the effect may be on the stability of the ER monomer. The increased mobility of the estradiol-induced complex is independent of its stability, as tamoxifen is able to induce a complex that is stable to Mg*+ but shows a mobility that is similar to the no hormone complex. Thus, this change in mobility signals an estradiolinduced change in the ER and may reflect the activation of the hormone-inducible transactivation domain (9).
The ER. ERE complex formed in the presence of tamoxifen is distinct from either the no hormone complex or the estradiol complex. The tamoxifen complex displayed the same stability to temperature and Mg2+ as did the estradiol complex. Interestingly, the mobility of the tamoxifen complex was more similar to the no hormone complex. This would imply that the effect of tamoxifen differs from the effect of estradiol in a way distinct from the induction of specific DNA binding by the ER. This difference is reflected by the increased mobility of the estradiol-induced complex in native gels and may signal differential activation of the hormone inducible transactivation domain (9). For example, the steroid may induce a conformation that strongly activates transcription whereas tamoxifen may stabilize binding but not activate transcription in the same fashion and may even inhibit transcription. Thus, tamoxifen could exhibit its anti-estrogenic effects on promoters that require the estrogen-inducible change in the transactivation domain for activity. In these cases, tamoxifen would be a competitive inhibitor of estrogen action or may bind and suppress transcription initiation. In cases in which both estrogen-dependent DNA binding and induction of the hormone-responsive transactivation domain are essential, tamoxifen would be expected to be a transdominant suppressor. Under other conditions and in certain cell types, tamoxifen is known to have estrogenic activity (28). This may be due to the ability of tamoxifen to induce stable ERE binding under conditions that do not favor estrogen-independent binding of the ER. In these cases a second non-estrogen-dependent transactivation domain contained in the amino-terminal portion of the ER may be utilized (10). Thus, the apparent paradox of tamoxifen having both antagonistic and agonistic properties would be explained on the basis of its ability to stimulate the formation of a complex that has properties of both the no hormone and estradiol-induced complexes.