Introduction

Top Summary Introduction Procedures Results Discussion References

Nuclear steroid/thyroid hormone receptors and their cognate hormonal ligands constitute a group of key mediators in the endocrine signaling pathways, playing an important role in the control of differentiation, growth, and metabolic homeostasis (Evans, 1988; Gronemeyer and Laudet, 1995). The receptors exercise control of these events by their function as hormone-activated transcription factors and modulators of gene expression in target cells (Beato, 1989; Gronemeyer, 1992). Classic signals to which this family of receptors respond are the steroid hormones, the thyroid hormones, and vitamins A and D, each interacting with a specific receptor of the family (Green and Chambon, 1988).

All receptors of the nuclear receptor superfamily have a similar architecture (Tsai and O'Malley, 1994). The amino-terminal A/B domain is involved in transactivation of gene expression. The C-domain contains a two-zinc finger structure, which plays an important role in receptor specific DNA-binding and receptor dimerization. The carboxyl-terminal ligand-binding domain (or E/F domain) is crucial for binding of receptor specific ligands, nuclear translocation, receptor dimerization, and modulation of target gene expression in association with corepressors and coactivators (Horwitz et al., 1996; Shibata et al., 1997).

Members of the nuclear steroid/thyroid hormone receptor family have been associated with various disorders and disease conditions such as cancer, osteoporosis, cardiovascular disease, inflammation, and metabolic disorders. A number of natural and synthetic hormonal drugs that modulate the function and activity of nuclear receptors are used for the treatment of major clinical indications, such as thiazolidinediones for treatment of type II diabetes (Lehman et al., 1995; Berger et al., 1996), glucocorticoids as anti-inflammatory drugs, estrogen antagonists for treatment of breast cancer, and androgen antagonists in prostate cancer therapy (Haynes et al., 1990).

A significant deal of attention has recently been focused on women's health issues and hormone replacement therapy. At the onset of perimenopause/menopause, women may have symptoms like hot flushes and urogenital tract complications (Lichtman, 1996). Furthermore, women may experience serious health risks such as the development of osteoporosis or cardiovascular disease during their postmenopausal life (Lichtman, 1996). The symptoms and health risks affecting many menopausal/postmenopausal women have been attributed to the loss of de novo production of the natural female sex hormone E2 (i.e., to an E2-deficient state).

To alleviate or prevent symptoms of the nature described above and the development of serious health risks, women are given estrogen replacement therapy in combination with a gestagen (Lichtman, 1996). However, current hormone replacement therapy is associated with certain serious concerns like fear for increased risk of breast or uterine cancer, indicating the need for development of safer therapy.

The effects of estrogen were long believed to be mediated by the ER cloned >10 years ago (Green et al., 1986; Greene et al., 1986). However, recently, a second ER, ER, was cloned from rat (Kuiper et al., 1996), mouse (Tremblay et al., 1997), and human (Mosselman et al., 1996; Enmark et al., 1997). The presence of two ERs, the "old" ER (renamed ER) and ER, reveals that the mechanism of biological action of estrogen and related synthetic drugs is more complex than previously thought. It also provides a unique opportunity for the development of improved modulators of estrogen action and the identification of new targets for estrogens.

In the current study, we compare the ligand binding and transcriptional responses of the human ER and the human ER, respectively, to various synthetic estrogen agonists and antagonists. The results of this study indicated that there are similarities between hER and hER in their responses to ligands but also that there are receptor-selective differences, which are of possible importance for design and synthesis of receptor-selective ligands for the development of drugs for hormone replacement therapy with improved in vivo efficacy and side effect profile.

	Experimental Procedures

Top Summary Introduction Procedures Results Discussion References

Materials. E2, 17-Estradiol, tamoxifen, 4-OH-tamoxifen, genistein, bisphenol A, and DES were from Sigma-Aldrich Sweden AB. Raloxifene and ICI 164,384 were synthesized according to published procedures (Jones et al., 1984; Bowler et al., 1989). [³H]E2 was purchased from New England Nuclear Research Products (Boston, MA). MEM culture media, FCS, L-glutamine, OptiMEM, lipofectamin, G418, and gentamicin were purchased from Gibco Life Technologies (Stockholm, Sweden). Hygromycin B was from Calbiochem (LabKemi AB, Sweden). Phenol red-free Coon's/F12 medium was from SVA (Uppsala, Sweden). The SRC 3000 serum substitute was purchased from Tissue Culture Services (Botolph Claydon, England). Rabbit reticulocyte lysate was from Promega (Scandinavian Diagnostic Services, Falkenberg, Sweden). The chemiluminescence substrate CSPD was purchased from Tropix (Boston, MA).

Ligand competition binding assay. Full-length recombinant hER and hER were produced at high levels using the baculovirus expression system (Luckow and Summers, 1989). Receptor protein was prepared from the nuclear fraction as described previously (Barkhem et al., 1991) and extracted with buffer containing 17 mM K₂HPO₄/3 mM KH₂PO₄, pH 7.9, 400 mM KCl, 1 mM MgCl₂, 0.5 mM EDTA, 6 mM monothioglycerol, and 8.7% glycerol. The concentration of extracted hER was 400 pmol/ml nuclear extract and 800 pmol/ml extract for hER (determined by specific binding of [³H]E2).

Vector constructs. The cDNAs encoding the full-length human ER (Greene et al., 1986) and the human ER (Enmark et al., 1997) were cloned into the BamHI and XbaI sites in the mammalian expression vector pMT-hGH (Friedman et al., 1989; Alksnis et al., 1991) after excision of human growth hormone coding sequences. The human ER cDNA encodes the wild-type 66-kDa receptor protein, and the human ER cDNA used in all experiments in this study encodes the 485-amino acid residues long form with a molecular mass of ~55 kDa.

DNA transfections. All transient or stable transfections were done by using the OptiMEM/lipofectamin procedure according to the supplier's recommendations (Gibco Life Technologies).

Generation of stable ER and ER reporter cell lines. The 293 cells (CRL-1573; American Type Culture Collection, Rockville, MD) routinely cultured in MEM supplemented with 10% FCS and 2 mM L-glutamine were first transfected with 2.5 µg of the pERE2-ALP reporter vector and 0.2 µg of the drug resistance vector pSV2-Neo (Southern and Berg, 1982) using the lipofectamin procedure according to the supplier's recommendations. A G418-resistant clone mix (293/ERE2-ALP) was isolated and used in a second round stable transfection with 0.5 µg of pMT-hER and pMT-hER, respectively, together with 0.1 µg of the drug resistance vector pKSV-Hyg (Gritz and Davies, 1983). Individual hygromycin B-resistant clones were isolated. One stable clone each of 293/hER and 293/hER reporter cells was chosen for further study in response to various estrogen agonists and antagonists.

Assay procedure for hormonal effects on 293/hER and 293/hER reporter cells. Approximately 25 × 10³ cells per well were seeded onto 96-well culture plates in 100 µl of Coon's/F12 supplemented with 10% FCS (stripped twice using dextran-coated charcoal) and 2 mM L-glutamine. At 24 hr later, conditioned medium was replaced with 100 µl Coon's/F12 supplemented with 5% serum substitute, 2 mM L-glutamine, gentamicin (50 µg/ml), and hormonal substances as indicated in the figure legends. In all experiments, cells have been exposed to hormones for 72 hr before harvest and analysis for effect on reporter gene expression. Triplicate determinations of reporter protein levels in the conditioned media for each concentration of compound have been performed in all experiments.

Assay for human placental ALP. The level of ALP expressed from the ERE2-ALP reporter vector in the stably transformed 293/hER and 293/hER reporter cells was determined by a chemiluminescent assay in which a 10-µl aliquot of heat-treated (at 65° for 40 min) conditioned cell culture medium was mixed with 200 µl of assay buffer (10 mM diethanolamine, pH 10.0; 1 mM MgCl₂, and 0.5 mM disodium 3-[4-methoxyspiro(1,2-dioxetane-3,2'-(5'-chloro)tricyclo[3.3.1.1.^3,7]decan)-4-yl]phenyl phosphate) (Alksnis et al., 1991; Nilsson et al., 1993) in white microtiter plates (Dynatech Laboratories In Vitro AB, Stockholm, Sweden) and incubated at 37° for 20 min before being transferred to a microplate format luminometer (Luminoskan; Labsystem, Helsinki, Finland). The setting of the Luminoskan was integral measurement with 1-sec reading of each well. The ALP activity is expressed in LU, which is directly proportional to the level of ALP expressed from the cells.

Calculation of degree of agonism in percent of E2. The degree of agonism in percent of E2 for each concentration of ligand was calculated as {[response of ligand (LU)  background (LU)]: [maximum response to E2 (LU)  background (LU)]} × 100.

	Results

Top Summary Introduction Procedures Results Discussion References

Stable subtype-specific ER reporter cell lines. The human 293 kidney epithelial cell line was chosen as the host cell line for the generation of stable reporter cells for screening of compounds that act via the human ERs  (hER) and  (hER) because as 293 cells did not respond to E2 in transient transfections with an ERE-driven reporter vector (Nilsson S, unpublished observations). The generation of the ER and ER reporter cell lines, respectively, was done in two steps: first, the establishment of a G418-resistant reporter vector clone mix (293/ERE2-ALP), and, second, a second transfection with the expression vectors encoding the human ER and human ER proteins, respectively, and the drug-resistance vector pKSV-Hyg. Of several hygromycin B-resistant cell clones, one clone of each receptor subtype, 293/hER and 293/hER, showing the best signal-to-noise ratio and sensitivity to increasing concentration of E2, was chosen for further studies. Both the 293/hER cells and 293/hER cells showed a dose-dependent transactivation of ALP reporter gene expression by E2, whereas the reporter clone mix 293/ERE2-ALP did not show any response to E2, as expected for a cell that does not express any endogenous or exogenous ER (Fig. 1). The level of receptor protein, hER and hER, respectively, expressed by the stable cells is <4000 receptors/cell as determined by ligand equilibrium binding assay (data not shown).

View larger version (24K): [in this window] [in a new window]   Fig. 1.   Dose-response to E2 in the 293/hER, 293/hER, and 293/ERE2-ALP reporter vector clone mix. The different genetically engineered clones and clone mix were exposed to the indicated concentrations of E2 for 72 hr before assay for the amount of ALP reporter protein expressed as described in the text (5 × 1012 M indicates no added E2; solvent only). Values are the mean from triplicate determinations for each concentration of E2, with error bars (standard deviation) indicated for each value. For some concentrations, error bars are not visible because they were smaller than the symbol.

View larger version (28K): [in this window] [in a new window]   Fig. 2.   Response in the 293/hER and 293/hER reporter cell lines to increasing concentrations of different commercially available estrogens. The two stable reporter cell lines, 293/hER and 293/hER were treated with (A) E2, (B) DES, (C) bisphenol A, (D) 17-ethynyl,17-estradiol, (E) 16,17-Epiestriol, and (F) genistein at the concentrations indicated. The level of ALP reporter protein expressed was assayed 72 hr after the addition of the ligands. The response is displayed as percent of the maximum response to E2 in 293/hER and 293/hER. The response values for each concentration of ligand are the mean of triplicate determinations with the mean ± standard deviation for each value indicated. For some concentrations, error bars are not visible because they were smaller than the symbol.

Ligands with receptor selective potency or efficacy. Among a number of commercially available ligands with known estrogen activity, 17-ethynyl-17-E2 (Fig. 2D) displayed the largest hER selectivity with a 35-fold (range, 35-60-fold) higher potency in transactivating ALP gene expression in 293/hER cells compared with 293/hER cells. 16,17-Epiestriol, on the other hand, displayed the largest hER selectivity with a 7-fold lower EC₅₀ value for hER than for hER (Fig. 2E). The phytoestrogen genistein showed receptor selective efficacy, being more efficacious than E2 in the 293/hER cells (range, 107-130% of E2) but demonstrating only partial agonism via hER in the 293/hER cells (Fig. 2F). However, genistein was slightly more potent via hER compared with hER, which is in agreement with its higher relative binding affinity to ER than to ER (Kuiper et al., 1997) (in a regular ligand binding assay, the IC₅₀ value for genistein in competition with [³H] E2 for binding to hER and hER is 30 nM and 1 µM, respectively).

Effect of partial estrogens/antiestrogens. The potency and agonist/antagonist activity of tamoxifen, 4-OH-tamoxifen, ICI 164,384, and raloxifene were tested in the 293/hER and 293/hER cells, respectively (Fig. 3, A-D). All four ligands showed a partial agonist activity in the 293/hER cells (Fig. 3, A and C) but only estrogen antagonism in the 293/hER cells (Fig. 3B). In the 293/hER cells, these ligands behaved as inverse agonists, suppressing the background reporter gene expression in a dose-dependent fashion. The relative agonism in percent of E2 in the 293/hER was in the range of 9-13% for tamoxifen, 4-OH-tamoxifen, and raloxifene, respectively, and ~3% for ICI 164,384 (Table 1).

View larger version (30K): [in this window] [in a new window]   Fig. 3.   Agonist/antagonist response to tamoxifen, 4-OH-tamoxifen, raloxifene, and ICI 164,384 in the presence or absence of 0.5 nM E2 in 293/hER and 293/hER. The effect on ALP gene expression to increasing concentrations of tamoxifen, 4-OH-tamoxifen, raloxifene, and ICI 164,384, respectively, in the absence (, , , ) or presence (, , , ) of 0.5 nM E2 in (A) 293/hER cells and (B) 293/hER cells. C, Agonist function of tamoxifen, 4-OH-tamoxifen, raloxifene, and ICI 164,384 is a magnification of their agonism displayed in A. D, Antagonist potency of raloxifene in the 293/hER and 293/hER reporter cell lines is shown for comparison. The response values are the mean values from triplicate samples for each concentration of ligand. For some concentrations, error bars are not visible because they were smaller than the symbol.

View this table: [in this window] [in a new window]   TABLE 1 Character of tamoxifen, 4-OH-tamoxifen, ICI 164,384, and raloxifene as estrogen agonists, antagonists, or both in the 293/hER and 293/hER reporter cells, respectively

View larger version (41K): [in this window] [in a new window]   Fig. 4.   Effect of increasing concentrations of raloxifene and ICI 164,384 on the agonist potency of E2 in the 293/hER and 293/hER reporter cell lines. The cells were exposed for 72 hr with no or 1, 5, 50, or 500 nM raloxifene or ICI 164,384 and increasing concentrations of E2 (1012 to 105 M; 1013 stands for no E2 added, only solvent or solvent and the indicated concentrations of raloxifene or ICI 164,384). Effect on the EC50 value for E2 in the presence of increasing concentrations of raloxifene in (A) 293/hER cells and (B) 293/hER cells and in the presence of increasing concentrations of ICI 164,384 in (C) 293/hER cells and (D) 293/hER cells. The values represent the mean of triplicate samples for each concentration of ligand with the mean ± standard deviation for each value indicated. For some concentrations, error bars are not visible because they were smaller than the symbol.

View larger version (21K): [in this window] [in a new window]   Fig. 5.   Effect of raloxifene and ICI 164,384 on the affinity of E2 for hER and hER. The apparent equilibrium binding constant (Kdapp) of [3H]E2 for hER and hER in the presence of low, fixed concentrations of raloxifene and ICI 164,384 was determined as described in the text. Binding parameters for [3H]E2 (including the estimated error in the determination) were obtained from nonlinear curve fitting of binding data to the Hill equation using Kaleidagraph 3.08 (Synergy Software, Reading, PA). Representative data from one of three independent experiments are shown.

	Discussion

Top Summary Introduction Procedures Results Discussion References

In this report, we have shown that the two ER subtypes,  and , respond in a similar fashion to some ligands but that there also are receptor-specific responses. The overall homology between hER and hER is 47% with only ~18% homology in their amino-terminal domains and 59% homology in their ligand binding domains (Enmark et al., 1997). The highest homology between hER and hER is in the DNA binding domain with identical amino acid sequence in the P-box, the primary determinant for sequence specific recognition of a hormone response element (Umesono and Evans, 1989).

The ligand binding cavity of hER is delimited by amino acid residues interspersed between amino acid residue 342 and amino acid residue 547 (Brzozowski et al., 1997). The homology between hER and hER in this region of their ligand binding domains is very high, especially with reference to amino acid residues indicated in the hER structure to make direct contacts with the ligand or lining the cavity (Brzozowski et al., 1997). It therefore is not surprising that E2 and other ligands bind to hER and hER with approximately similar affinity. However, ligands with higher affinity for hER than for hER, and vice versa, have been described (Kuiper et al., 1997). One such example is genistein, which has been shown to have a significantly higher affinity for rat and human ER than for hER (~30-fold) (Kuiper et al., 1997) (in a regular ligand binding assay, the IC₅₀ value for genistein in competition with [³H] E2 for binding to hER and hER is 30 nM and 1 µM, respectively). However, the degree of receptor-selective difference in affinity was not mirrored by the same degree of receptor-selective potency in the cell-based gene transcription assay. The potency of genistein in the 293/hER reporter cells was only 4-5-fold higher than that in the 293/hER cells, and in contrast to expectations, genistein was only a partial agonist via the  receptor but (at least) a full agonist when the transcriptional response was mediated by ER. As reported by others, the gene modulatory effect of a receptor after binding of a ligand depends on the conformational change of the receptor induced by the ligand and the subsequent events, including receptor dimerization, receptor/DNA interaction, formation of a preinitiation complex, and recruitment of and interaction with coactivators and other transcription factors (Beekman et al., 1993; McDonnell et al., 1995; Katzenellenbogen et al., 1996; Shibata et al., 1997). Based on our current knowledge of what is required for a ligand to be recognized by the cellular transcription machinery as a full agonist, it is obvious that genistein most likely does not induce an optimal agonist conformation of the ER protein, thus explaining its partial agonist activity and possibly also its relatively smaller receptor-selective gene modulatory potency compared with its receptor-selective binding affinity.

As expected, the well known partial agonists/antagonists tamoxifen, 4-OH-tamoxifen, and raloxifene displayed a low but significant estrogenic activity in the 293/hER reporter cells (Fig. 3, A and C). Even the pure antagonist ICI 164,384 induced a low degree of transcriptional activity when bound to hER (Fig. 3C). However, none of these ligands displayed any degree of estrogen agonism in the 293/hER reporter cells; only antagonism was displayed (Fig. 3B). Most likely, all four ligands inactivated the ligand-dependent AF2 function of both hER and hER, and therefore the most likely explanation for the observed differences in transcriptional response to these ligands is to be found in the amino-termini of hER and hER, respectively. Recently, it was described that the partial agonism of tamoxifen is mediated by a slightly different part of the ER AF1 region than required for E2 (McInerney and Katzenellenbogen, 1996). A plausible explanation for the absence of any reporter gene transactivation activity by hER in the presence of tamoxifen, 4-OH-tamoxifen, and raloxifene is that hER lacks this particular function of hER AF1.

Next, we analyzed the antagonistic character of raloxifene and ICI 164,384 in the 293/hER and 293/hER reporter cells, respectively, and compared this with their characteristics as modulators of the K_dapp of [³H]E2 for hER and hER, respectively. Both ligands behaved as competitive antagonists (Fig. 4) and competitive receptor binders to E2 for both hER and hER (Fig. 5). However, already low concentrations of raloxifene caused a dramatic shift in the EC₅₀ value for E2 in the 293/hER cells (Fig. 4A) compared with the raloxifene effect in the 293/hER cells (Fig. 4B) and the ICI 164,384 effect in the 293/hER cells (Fig. 4C). The effect of raloxifene on the potency of E2 in the 293/hER cells was confirmed by the increased K_dapp of E2 for hER, especially at concentrations of raloxifene above 0.2 nM (Fig. 5). Similar concentrations of raloxifene had only moderate effect on the K_dapp of E2 in binding to hER. Although ICI 164,384 affected the K_dapp of E2 for hER more than for hER (Fig. 5), that effect was not reflected in the cells in which ICI 164,384 caused a ~4 order of magnitude shift in the EC₅₀ value for E2 in both 293/hER and 293/hER cells (Fig. 4, C and D). However, ICI 164,384 was a more potent antagonist than raloxifene in the 293/hER cells (Table 1), shifting the EC₅₀ value of E2 ~4 orders of magnitude at its highest concentration compared with raloxifene, which shifted the EC₅₀ value of E2 ~3 orders of magnitude (Fig. 4, B and D). Thus, although raloxifene showed both an hER-selective affinity (Fig. 5) and potency as antagonist (Figs. 3D and Fig. 4, A and B), ICI 163,384 showed hER-selective affinity (Fig. 5) but no receptor-selective antagonist potency (Table 1 and Fig. 4, C and D). The explanations for the hER-selective affinity and antagonist potency of raloxifene and the hER-selective affinity of ICI 164,384 are difficult but may be examined when the three-dimensional structure of hER has been determined. That ICI 164,384 showed a similar antagonist potency for both hER and hER (Table 1 and Fig. 4, C and D) despite its hER-selective affinity (Fig. 5) may be explained by the assumption that ICI 164,384 interferes with the homodimerization function of both hER (Fawell et al., 1990) and hER with equal potency.

Although some of the differences between hER and hER in response to partial estrogens could be explained by the lack in hER of the part of AF1 that is responsible for mediating the partial agonism of, for example, tamoxifen when bound to hER, it cannot explain all the differences between hER and hER in their transcriptional responses to various agonists and antagonists described in this report. The affinity of a ligand for a particular receptor and the conformational change induced by a ligand after its binding to the receptor are two important parameters that determine the transcriptional potency and agonist/antagonist character of a ligand. Both affinity and ligand-induced conformational change of the receptor depend, in part, on the three-dimensional structure of the ligand and its hydrophobic/hydrophilic character and, in part, on the volume and shape of the ligand binding cavity and the type of amino acid residues lining the cavity. As mentioned, the differences between hER and hER in the part of LBD that makes up the ligand binding cavity of hER are small, explaining why many ligands have a similar affinity and biological character when bound to hER or hER. Nevertheless, there are a number of ligands (Kuiper et al., 1997) (Nilsson S, unpublished observations) displaying receptor-selective affinity or biological character (e.g., genistein) that binds to hER with a ~30-fold higher affinity than to hER. However, the most striking receptor-selective difference with genistein perhaps is its partial agonism mediated by hER but slight superagonist character (range, 107-130% of E2) when bound to hER. This receptor-selective character of genistein and, in particular, its - versus -selective difference in efficacy are difficult to explain by just comparing the primary sequences of the LBDs of hER and hER, respectively. A possible explanation is that the ligand binding cavity of hER is more different from the cavity of hER than can be expected based on comparisons between primary amino acid sequences. Another explanation might be that hER has somewhat different requirements for coactivators than hER, a difference that may only become apparent with particular ligands like genistein.

In conclusion, there are good reasons to believe that it will be possible to develop selective hER and hER ligands. Such ligands may form the basis of novel pharmaceutical agents, providing significant advantages over currently available drugs for hormone replacement therapy in women.