Activation of Transcription by Estrogen Receptor α and β Is Cell Type- and Promoter-dependent*

Tamoxifen acts as a strong estrogen antagonist in human breast but as an estrogen agonist in the uterus. The action of tamoxifen is mediated through estrogen receptors (ERα and ERβ), which bind to a variety of responsive elements, to activate transcription. To examine the role of these varied elements in the response to antiestrogens, we studied the activation of a panel of differing promoters, by these compounds, in human breast, bone, and endometrial derived cell lines. No agonistic activity was observed in breast cells, whereas all antiestrogens, particularly tamoxifen, exhibited agonistic effects in uterine cell lines. All antiestrogens studied were agonistic in co-transfections of a collagenase reporter gene and ERβ, but tamoxifen alone was agonistic with ERα in (uterine) HEC-1-A cells. The ERα mediated, agonism of tamoxifen was not observed in primary cultures of human uterine stromal cells, whereas the ERβ-mediated agonism of all selective estrogen receptor modulators was present. This suggests that the two receptors operate by distinct pathways and that the response of cells to antiestrogens is dependent on the ER subtypes expressed.

Selective estrogen receptor modulators (SERMs), 1 such as tamoxifen, act through interaction with the estrogen receptor (ER) (1). Estrogens and SERMs bind to the ligand binding domain of the ER, allowing dimerization and binding to a palindromic estrogen response element (ERE) upstream of estrogen-sensitive genes. The bound dimer then acts to transactivate transcription (2). SERMs act by competing with estrogen for ER binding. However, estrogenic activity is inhibited in some tissues but unaffected in others (3)(4)(5). The cloning of a second estrogen receptor (6) (ER␤) raised the possibility of two different ER homodimers together with a heterodimer of the two ERs (7,8). The cloning of an N-terminal extended (9) and ligand binding domain insertion splice variants (10,11) of the human ER␤ suggests a wide range of possible homo-and heterodimers, each of which may possess differing ligand sensitivities. Many promoters have been identified that are estrogensensitive but lack an ERE. These include 1) the activation of genes possessing AP-1 elements, including collagenase (12)(13)(14), 2) the activation of expression of genes possessing Sp1 binding sites, which are often associated ERE half-sites (15)(16)(17)(18)(19)(20), 3) the activation of the TGF␤3 gene through a novel sequence termed the raloxifene response element (21,22), and more recently 4) the activation of genes possessing the antioxidant response element (23,24). The estrogen receptor has also been found to down-regulate the interleukin-6 gene by preventing the binding of NF-B (25)(26)(27).
The conformation of the ER allosterically varies depending on the DNA sequence bound (28) and thus will differ between response elements. Conformational changes may well affect both the ligand binding domain and its interaction with other proteins such as coactivators and corepressors (29 -32). Therefore, alternative estrogen signaling pathways allow for a broad range of activities to be produced by the same compound acting through differing response elements. However, the vast majority of work on the action of SERMs has concentrated on their action through EREs and has not considered the possible roles of these alternative estrogen responsive elements.
Tamoxifen is widely used as an adjuvant therapy in the treatment of women with breast cancer. In the breast, tamoxifen acts as an estrogen antagonist, reducing or preventing the proliferation of tumor cells (33). In contrast, in the uterus, this compound acts as an estrogen agonist, resulting in cell proliferation and in the long term, a 2-5-fold increase in endometrial tumors (34,35). In order to gain further insight into the effects of SERMs on the activation of ER␣ or ER␤ in cells derived from breast or endometrium, a panel of reporter constructs was assembled in which transcription was driven by a range of differing promoters and response elements. These constructs were then studied, in transient transfection assays, in human breast and uterine cell lines. The activities of SERMs were found to vary dramatically both between cell type and promoter construct studied. These results suggest that the effects of SERMs, in vivo, cannot be predicted by their actions on simple elements, such as the ERE, in isolation.
Cell Culture and Transfections-Human breast-derived, MCF-7 (ATCC) cells were maintained in Dulbecco's modified Eagle's medium/ F-12 (1:1) (Life Technologies, Inc.) supplemented with Glutamax I (Sigma) and 10% fetal calf serum (Sigma). Human osteoblast-derived, MG-63 (ATCC) cells were maintained in Eagle's modified minimal essential medium (Sigma) supplemented as above. Human uterus-de-rived, HEC-1-A (ATCC) cells were maintained on McCoy's 5A medium (Sigma) supplemented as detailed above. All cell lines were established to be mycoplasma-free both before and after the study was completed. Primary cell cultures were prepared as described previously (41) and maintained in Dulbecco's modified Eagle's medium/F-12 (Life Technologies) supplemented with Glutamax I (Sigma), 10% fetal calf serum (Sigma), 1ϫ antibiotic/antifungal solution (Sigma), and 7.5% sodium bicarbonate. All media were Phenol Red-free, and all serum estrogen was stripped using dextran-coated charcoal powder (Sigma). All transfections were performed using Fugene 6 (Roche Molecular Biochemicals) at a ratio of 1:1.5, DNA:Fugene. Cells were dosed 4 h after transfection and harvested 24 h later. Cells were lysed in reporter lysis buffer (Promega); ␤-galactosidase activity was determined using a ␤-galactosidase assay kit (Promega), and luciferase activity was determined by a luciferase kit (Promega).
Protein Extraction and Western Blotting-Total cell proteins were extracted as described previously (42). Protein concentrations were determined using a protein determination kit (Sigma). Proteins were separated on SDS-polyacrylamide gels using a Mini Protean II (Bio-Rad) gel kit according to the manufacturer's instructions. Proteins were transferred to Hybond ECL membrane (Amersham Pharmacia Biotech) using a Mini-Trans-Blot Electrophoretic Transfer Cell (Bio-Rad). Blots were blocked with 10% defatted milk protein overnight at 4°C. Blots were then washed with TBS-T20 (TBS plus 0.1% Tween 20) three times for 15 min and two times for 5 min. Blots were probed with primary antibody (ER␣, 1:1000 (Novocastra Laboratories NCL-ER-CF11, antihuman, mouse monoclonal to full-length human ER␣); ER␤, 1:5000 (Upstate Biotechnologies, Inc., 06-629 anti-rat, rabbit polyclonal to amino acids 54 -71 of rat ER␤)) in TBS-T20 containing 1% defatted milk protein for 1 h. Filters were washed as before and then incubated with secondary antibody (ER␣ anti-mouse (Sigma), 1:40,000; ER␤ anti-rabbit (Amersham Pharmacia Biotech), 1:40,000) in TBS-T20 for 30 min. Filters were rinsed and washed three times in TBS-20 and then probed with streptavidin 3°antibody (1:2000) in TBS for 20 min before rinsing and washing three times for 5 min in TBS-T20. Detection was then performed by enhanced chemiluminescence (Amersham Pharmacia Biotech).

SERMs Do Not Act as Agonists, in the Context of a Wide
Range of Promoters, in the Breast-derived, MCF-7 Cell Line-Studies on the effects of SERMs have, generally, concentrated on signaling through the consensus ERE, and the effect of these compounds on alternative estrogen signaling pathways has yet to be fully assessed. Stimulation of a panel of six diverse estrogen-responsive reporter constructs (as described under "Materials and Methods") by 17␤-estradiol, 4-hydroxytamoxifen, raloxifene, and faslodex (ICI 182, 780) has been studied. Initially, to validate the transfection system, the activation of the ERE reporter (pERE-TK-Luc) was investigated over a wide range of doses of the above compounds in MCF-7 cells. This cell line has high endogenous levels of both ER subtypes, so these were not co-transfected in this study. As expected, 17␤-estra-diol was found to be strongly agonistic over a wide range of doses. A strong antagonistic dose response was also observed with all three antiestrogens examined (Fig. 1).
The trans-activation of the promoter panel was then investigated in the MCF-7 cells. The ERE construct (pERE-TK-Luc) was strongly activated, over control levels, by estradiol, while tamoxifen, raloxifene, and faslodex all exhibited strong antagonistic activity (Fig. 2a). Faslodex was a particularly strong antagonist, reducing activity to less than 1% of the control. In transfections with the complement C3 reporter (pC3-Pst-T1-Luc), estradiol again acted as an agonist, and both raloxifene and faslodex acted as strong antagonists. However, tamoxifen did not produce a response different to that of the controltreated cells (Fig. 2b). Previous studies found that in HepG2 cells, transfected with this construct, tamoxifen produced an agonistic response (36). Using the collagenase promoter (pCol73-Luc), strong estradiol agonism was observed, while all three SERMs showed no significantly different response to the control treatment (Fig. 2c). A similar response was also observed using the TGF␣ promoter construct (pXP1-TGF1.1e-Luc) (Fig. 2d). These results show that, while no agonistic activity of antiestrogens was observed in any promoter context, the degree of antagonism, particularly with respect to tamoxifen, varied considerably. No significant activation of the raloxifene response element (p␤3-499-Luc) or adrenomedullin (p-LCF-1543-Luc) reporters (data not shown) was observed in this cell line with any treatment. Activation of these constructs was found in all other cell lines studied; therefore, it was assumed that tissue-specific properties of the MCF-7 cell line prevented expression, rather than any inherent fault with the reporter constructs.
Effects of SERMs on the Activity of the Promoter Panel in the Osteoblast-derived Cell Line MG-63 Co-transfected with either ER␣ or ER␤-The osteoblast cell line MG-63 has been suggested as a model cell line for bone remodeling and is thought to express negligible levels of ERs. However, in Western blots of these cells, both ER␣ and ER␤ expression could be detected (Fig. 3). Since we wished to compare the activity of the ER␣ with that of ER␤, the presence of these endogenous receptors was a potential problem. To test the ability of the endogenous ERs, expressed in this cell line, to activate the reporter genes, reporter constructs were transfected into this cell line without receptor co-transfection. No estradiol-inducible transcriptional activity could be observed in these experiments (data not shown), and it is therefore assumed that in subsequent experiments trans-activation is a result solely of the co-transfected ER.
In the MG-63 cells, all six of the reporter constructs were responsive to estradiol or antiestrogens in co-transfections with at least one of the two ERs (Fig. 4). In general, estradiol was found to be an agonist for both ER␣ and ER␤, although the degree of agonism did vary significantly. However, estradiol failed to enhance ER␤ activity in the context of the adrenomedullin promoter (p-LCF-1543-Luc) and was a very weak agonist of ER␣ activity in the context of the ERE promoter (pERE-TK-Luc). The later result suggests that even on a very simple response element such as an ERE the two ER subtypes may have distinct activities.
Both raloxifene and faslodex were either found to be antagonists or had no effect relative to control on five of the six promoters studied. In contrast, in transfections of the collagenase promoter (pCol73-Luc) (Fig. 4c) both compounds were ineffectual with ER␣ but strong agonists with ER␤. The effects of tamoxifen, while mostly antagonistic, were less predictable, since significant agonistic activity was observed in several promoter contexts. In transfections with the TGF␣ reporter construct (pXP1-TGF1.1e-Luc) slight tamoxifen agonism was only observed with ER␣ (Fig. 4d), but with the TGF␤3 promoter (p␤3-499-Luc) strong agonism was observed with both receptors (Fig. 4e). This suggests that tamoxifen agonism may be both promoter-and ER subtype-specific.
When the adrenomedullin reporter (p-LCF-1543-Luc) was co-transfected with ER␣, strong estradiol agonism was observed (Fig. 4f). Both raloxifene and faslodex displayed antagonistic activities, while tamoxifen had no significant activity. Interestingly, in co-transfections with ER␤ there was no sig-nificant activity, relative to control, in any treatment group. These results suggest that the agonistic activity of ER ligands is receptor subtype-specific.
Effects of SERMs on the Activity of the Promoter Panel in the Endometrial Derived Cell Line HEC-1-A Co-transfected with ER␣ and/or ER␤-There have been reports that ER␣ has been detected in HEC-1-A cells, together with ERE binding activity in cell extracts and estrogen-stimulated growth (43). Western blots found both ER␣ and ER␤ to be present, although at levels much below those seen either in MCF-7 cells (data not shown) or in the uterus (Fig. 3). Despite these low levels of endogenous receptors, no transactivation was detected with any of the six reporter constructs when transfected without additional estrogen receptors (data not shown). The activity of the reporter genes in subsequent co-transfections is therefore taken to be due to the transfected receptor alone rather than to endogenous receptors.
Uterine cells have been reported to express low levels of the ER␤ mRNA in addition to high levels of ER␣ (44), but the relative levels of the two proteins remain unknown. This raises the possibility of cells, in this tissue, containing active ER␣/␤ heterodimers. Therefore, to investigate the roles of the two receptors, HEC-1-A cells were co-transfected with ER␣, ER␤, or an equimolar ratio of the two. In co-transfections of the ERE promoter construct and ER␣, strong estradiol agonism was observed (Fig. 5a). Neither tamoxifen nor raloxifene was significantly different from control, while faslodex was an antagonist. In co-transfections with ER␤, although estradiol remained an agonist, both tamoxifen and raloxifene also became agonistic, while faslodex had no significant effect. This suggests that as with the MG-63 cells (Fig. 4a), even on a simple promoter, such as the pERE-TK-Luc, the agonism of antiestrogens differs markedly between the two ER subtypes. When the two receptors were co-transfected, in a 1:1 ratio, the pattern of agonism observed closely resembled that of the ER␤ transfection rather than that of ER␣.
In the HEC-1-A (uterine) cell line, co-transfections of the complement C3 reporter (pC3-Pst-T1-Luc) with ER␣ resulted in strong estradiol agonism. Both raloxifene and faslodex produced a strongly antagonistic effect, while tamoxifen displayed no significant activity (Fig. 5b). These results closely resemble those seen in the MG-63 cells. Co-transfections with ER␤ gave very weak expression from this construct. Estradiol agonism was observed, however, and both tamoxifen and raloxifene were found to be strongly antagonistic. Faslodex produced no significant effect relative to control. When the two receptors were coexpressed, the pattern of agonism observed using this reporter construct closely resembled that seen with ER␣ rather than that seen with ER␤; however, the antagonism of faslodex is not as strong as observed with ER␣ alone.
After co-transfection with ER␣ and the collagenase reporter construct (pCol73-Luc), estradiol produced a weak agonistic response (Fig. 5c), and tamoxifen gave a stronger agonistic response, while the response to raloxifene or faslodex was not significantly different from control. In co-transfections with ER␤ and this reporter some weak estradiol agonism was observed, but all three antiestrogens displayed significantly greater agonism than estradiol. When the two receptors were coexpressed, a pattern of agonism was observed closely resembling that seen with ER␤ alone. The strong increase in the agonism of both raloxifene and faslodex in cells expressing ER␤ raises the possibility of a population of cells responsive to these compounds in the uterus.
Transfections using the TGF␣ reporter construct (pXP1-TGF1.1e-Luc) produced results very similar to those observed in the bone-derived MG-63 cell line. In ER␣ co-transfections, estradiol produced a strong agonistic activity, and tamoxifen was weakly agonistic, while both raloxifene and faslodex exhibited an antagonistic response (Fig. 5d). In ER␤ co-transfections, estradiol was again agonistic, while all other treatments produced no significant effect. Co-transfection of the two receptors produced a pattern of activity resembling that of ER␣; however, the estradiol agonism was weaker than that seen with ER␣ alone.
In the context of the TGF␤3 reporter (p␤3-499-Luc) and ER␣, weak agonism was observed with both estradiol and raloxifene, while tamoxifen was strongly agonistic and faslodex gave no significant response (Fig. 5e). In cells co-transfected with ER␤, estradiol was a strong agonist and faslodex a weak agonist, while all other compounds produced no significant response. Interestingly, when the two receptors were coexpressed, a novel pattern of agonism emerged. Both estradiol and faslodex produced weak but significant agonism (p Ͻ 0.005), while both tamoxifen and raloxifene produced a strongly agonistic effect (p Ͻ 0.001).
Co-transfections using the adrenomedullin promoter (p-LCF-1543-Luc) failed to produce any tamoxifen-specific agonism. Irrespective of receptor context, estradiol produced strong agonism, while all three SERMs produced no significant response (Fig. 5f). Co-transfection of the receptors together ablated the estradiol response. The lack of tamoxifen-specific response is surprising, since this gene was originally detected as an up-regulated RNA in differential display studies of primary endometrial cell cultures treated with tamoxifen.
When the two estrogen receptor isoforms were co-transfected, the response to treatment in some cases resembled that seen with ER␣ alone (Fig. 5, a and c) and in some cases resembled that with ER␤ alone (Fig. 5, b and d), and in some cases a novel effect was produced (Fig. 5, e and f). This suggests that not only is the coexpression of ER␤ in this tissue capable of affecting the response of ER␣ but also that expression may give rise to a novel pattern of responses to SERMs. We attempted to delineate the level of ER␤ expression required to produce an agonistic response to raloxifene. The expression of ER␤ relative to that of ER␣ was titrated into a set of co-transfections with the collagenase reporter construct (pCol73-Luc) (Fig. 6). In this experiment, raloxifene agonism was found to remain weak up to an approximately equimolar ratio of the two receptors. Increasing ER␤ expression further produced a rapid increase in raloxifene agonism and a further increase in the agonism of tamoxifen. These results suggest that in a uterine cell, expressing both receptors, significant raloxifene agonism would only be observed if the predominantly expressed ER subtype were ER␤. Whether such cells exist in vivo is, at present, unknown.

The SERM Agonism of Promoters in Primary Cultured Uterine Cells Differs from That Observed in Immortalized Cells-
One of the most striking differences observed in the above transfections was the different agonistic and antagonistic properties of antiestrogens in transfections with the ERE and the collagenase reporter construct. To investigate these differences, the responses of these two promoters were studied in FIG. 3. Western blot of protein extracts from HEC-1-A. MG-63 and primary uterine stromal cells were probed with antibodies to ER␣ and ER␤. Total cellular protein from each cell line was extracted as described (42). Protein concentrations were determined using a protein determination kit (Sigma). Proteins were separated by SDS-polyacrylamide gel electrophoresis and transferred to Hybond ECL membrane by electroblotting. Blots were blocked with 10% defatted milk protein overnight. After washing, blots were probed with 1) ER␣ primary antibody followed by anti-mouse secondary or 2) ER␤ primary antibody followed by anti-rabbit secondary. Filters were washed and probed with streptavidin tertiary antibody, and detection was performed by enhanced chemiluminescence. Positive control is an extract of rat uterine total protein. The arrow indicates the 68-kDa marker; hER␣ is ϳ65 kDa, and hER␤ is ϳ59 kDa. primary uterine stromal cell cultures. After co-transfection of the ERE reporter together with ER␣ into these cells, strong estradiol agonism was observed (Fig. 7a). Tamoxifen had no effect, relative to control, while both raloxifene and faslodex were antagonistic. In co-transfections with ER␤, estradiol was again agonistic, while tamoxifen, raloxifene, and faslodex produced no significant effect. These results are very similar to the pattern of agonism observed in the HEC-1-A cells. However, when the collagenase reporter was co-transfected with ER␣, the strong tamoxifen agonism, observed in the HEC-1-A cells, was absent (Fig. 7b). Estradiol had a strongly agonistic effect, while both faslodex and raloxifene had no significant effect, relative to control. When the primary cells were transfected with ER␤, estradiol was found to have weakly agonistic effects, while tamoxifen, raloxifene, and faslodex were all agonistic. This result is very similar to the effects of these compounds in the HEC-1-A transfections. This suggests that the signaling via the ER␣ and the ER␤ may be mechanistically divisible, with an ER␤ pathway common to the HEC-1-A and primary cells, while the ER␣ pathway varies between the two cell types. DISCUSSION We have analyzed the responses of a panel of reporter constructs based on both classical EREs and "alternative estrogenresponsive elements" to treatment with estrogens and antiestrogens in a variety of cell lines. Striking differences in the response of this panel were found between the three cell lines studied. In the MCF-7 (breast-derived) cells, no detectable agonist activity of any antiestrogen was found. However, the degree of antagonism did vary between the promoters. In this cell line, no response of the TGF␤3 (p␤3-499-Luc) and adrenomedullin (p-LCF-1543-Luc) promoters, to any treatment, was detected. However, these two constructs contain no inserted strong promoter or enhancer regions and so may not be sensitive enough to respond to treatment in this cell type.
The ERE (pERE-TK-Luc) and the complement C3 reporter (pC3-Pst-T1-Luc) constructs are both regulated by EREs and yet were found to possess differing responses to SERM treatment (Figs. 4, a and b, and 5, a and b). In the MCF-7 cells, SERMs were less antagonistic in transfections with the complement C3 than the ERE construct, whereas in the HEC-1-A cells this situation was reversed. The pERE-TK-Luc construct contains a single consensus ERE linked to thymidine kinase promoter, while the complement C3 promoter contains at least two divergent ERE-like elements (36). It has been suggested that sequence variation within the ERE (45) or flanking sequence (46), bending of DNA (47), and ERE separation (48) may all influence ER-ERE interaction. Therefore, superficially similar response elements may have differing responses to SERMs, making the action of the compounds difficult to predict in vivo.
The greatest differences in responses to SERMs were observed in transfections containing the collagenase promoter (pCol73-Luc). In the MG-63 cells, co-transfections with ER␣ resulted in weak tamoxifen agonism, whereas in HEC-1-A cells tamoxifen was more agonistic than estradiol. In co-transfections with ER␤ in either cell line, all three antiestrogens were agonistic, while estrogen had little effect. The effects of SERMs in uterine cells agree with previous observations in Ishikawa cells (12,13). When the two receptor subtypes were coexpressed, all three antiestrogens were, again, agonistic. To investigate this interaction further, we studied the activation of the collagenase reporter by various concentrations of the two receptors, in HEC-1-A cells (Fig. 6). Our results suggest that ER␤ expression greater than that of ER␣ was required to elicit raloxifene and faslodex agonism. The agonism of tamoxifen in AP-1 containing promoter contexts has been proposed as a potential mechanism for this compound's activity in the uterus (12). This could potentially occur via ER␣-or ER␤-mediated pathways. However, ER␤ is expressed at low levels in the uterus (44), and neither raloxifene nor faslodex exhibit agonist activity, as observed in the ER␤ pathway in vitro. Therefore, if the agonism of tamoxifen, in vivo, is mediated through AP-1 activation, it is via the ER␣ rather than the ER␤ pathway.
To investigate the activation of collagenase reporter (pCol73-Luc) in a situation closer to that in vivo, we studied the activation of this construct in primary uterine stromal cells. Activity in co-transfections with ER␤ and the collagenase reporter was similar to that seen in the HEC-1-A cells. However, in ER␣ co-transfections, the tamoxifen agonism, observed in cell lines, was not found. If tamoxifen agonism via AP-1 interaction does occur through an ER␣-specific pathway, these results suggest that this response is more sensitive in cell lines than in primary cultured cells. Tamoxifen may therefore only act as an agonist on uterine cells that express a "tamoxifen-sensitive AP-1 phenotype." Co-transfections of the TGF␣ reporter (pXP1-TGF1.1e-Luc) produced results similar to those observed using the complement C3 construct. The expression of TGF␣ has been suggested to be regulated by divergent EREs (37) and several Sp1 binding sites (17,49). Recently, the RAR␣ promoter, also thought to be regulated by ERE-Sp1 interactions (50), was found to possess reverse pharmacology when co-transfected with ER␤, in HepG2 cells (51). We did not detect this effect with ER␤ and the TGF␣ reporter construct in the cell lines studied here. It therefore remains unclear whether the novel response reported with the RAR␣ promoter is unique to that gene or a more common property of ERE/Sp1-containing promoters.
The adrenomedullin gene was detected as a tamoxifen-induced mRNA species in uterine cells (39). Subsequent analysis of the expression of this gene found it to be a novel, endothelial, growth factor expressed in both uterine endometrial and stromal cells. However, a construct based on this promoter (p-LCF-1543-Luc) (38), although responsive to estradiol, had no response to tamoxifen in any cell type studied, including primary uterine stromal cells (data not shown). Unlike the previous study, we co-transfected ERs into the cells; therefore, it is possible that the tamoxifen agonism is reliant on a particular pattern of ER expression. Interestingly, despite activation in FIG. 7. Transfection of the reporter constructs pERE-TK-Luc or pcol73-Luc in a primary uterine stromal cell line. Monolayer cultures of primary uterine stromal cells, in six-well plates, were co-transfected with either 1) pERE-TK-Luc (0.25 g) or 2) pCol73-Luc (0.25 g) together with the internal control plasmid pCMV␤ (0.25 g) and either an expression plasmid for hER␣ or hER␤ (0.25 g) using the transfection reagent Fugene 6. Dosing and analysis were as described in the legend to Fig. 4. Results represent the mean Ϯ S.E. response to estradiol in experiments with either ER singly when the two receptors were co-transfected, no activation was found with any compound studied. This further suggests that the regulation of the expression of this gene is highly dependent on ER isoform expression. If this protein is a tamoxifeninduced growth factor in vivo, then it represents a further possible mechanism of tamoxifen agonism and warrants further study.
In transfections of the TGF␤3 (p␤3-499-Luc) construct, we were unable to detect any raloxifene agonist activity with either estrogen receptor subtype, in bone-derived MG-63 cells (Fig. 4e). This construct (52) has previously been found to be strongly induced by raloxifene, but not estradiol, in MG-63 cells when coexpressed with ER␣ (21). The reason for this difference is unclear. In HEC-1-A cells, strong raloxifene agonism was found only after coexpression of both ER subtypes. This suggests that activity from this promoter is strongly dependent on the ER subtype expressed. The mechanism by which the ER interacts with this element is still unknown, but direct binding to the DNA seems unlikely (22). Further investigation is required to understand the ER's interaction with this promoter and its role in the activity of raloxifene.
To study activity of the promoter panel in cells expressing both ER subtypes, both receptors were co-transfected, in equimolar ratios, into the HEC-1-A cell line. We found that some promoter constructs display a pattern of responses similar to that of ER␣ alone (i.e. the complement C3 promoter), some display a pattern similar to that of ER␤ alone (i.e. ERE and collagenase promoters), and some display novel patterns (i.e. raloxifene response element and adrenomedullin promoters). This suggests that while some promoters bind one homodimer preferentially, heterodimer binding may give rise to novel responses to antiestrogens. Therefore, cells expressing certain combinations of receptors may be particularly vulnerable to the agonistic effect of SERMs.
Our results show that antiestrogens, and in particular tamoxifen, are more agonistic in uterine than breast cells. However, the strong tamoxifen agonism, observed with some reporter constructs in uterine cell lines, is not observed in primary cultures of uterine stroma. Co-transfections of the two ER subtypes suggest that signaling occurs via a distinct pathway for each receptor and that the binding specificity of the two homodimers is promoter-specific. Further studies are required to investigate the mechanistic difference in signaling by ER␣ and ER␤ (particularly in the context of AP-1 activation) and the observed difference in these pathways between cell lines and primary cultures.