First published online on April 27, 2005
Endocrine Reviews, doi:10.1210/er.2004-0027
Endocrine Reviews 26 (3): 465-478
Copyright © 2005 by The Endocrine Society
Reflections on the Discovery and Significance of Estrogen Receptor ß
Konrad F. Koehler,
Luisa A. Helguero,
Lars-Arne Haldosén,
Margaret Warner and
Jan-Åke Gustafsson
KaroBio AB (K.F.K.), Novum, SE-141 57 Huddinge, Sweden; and Department of BioSciences and Medical Nutrition (L.A.H., L.-A.H., M.W., J.-Å.G.), Karolinska Institutet, Novum, SE-141 57 Huddinge, Sweden
Correspondence: Address all correspondence and requests for reprints to: Jan-Åke Gustafsson, Department of BioSciences and Medical Nutrition, Karolinska Institutet, Novum, SE-141 57 Huddinge, Sweden. E-mail: jan-ake.gustafsson{at}mednut.ki.se
 |
Abstract
|
|---|
We have known for many years that estrogen is more than the female hormone. It is essential in the male gonads, and in both sexes, estrogen has functions in the skeleton and central nervous system, on behavior, and in the cardiovascular and immune systems. An important aspect of the discovery of estrogen receptor (ER) ß is that the diverse functions of estrogen can now be divided into those mediated by ER
and those mediated by ERß. Pharmacological exploitation of this division of the labors of estrogen is facilitated by the ligand-binding specificity and selective tissue distribution of the two ERs. Because the ligand binding domains of ER and ERß are significantly different from each other, selective ligands can be (and have been) developed to target the estrogenic pathway that is malfunctioning, without interfering with the other estrogen-regulated pathways. Because of the absence of ERß from the adult pituitary and endometrium, ERß agonists can be used to target ERß with no risk of adverse effects from chemical castration and uterine cancer. Some of the diseases in which there is hope that ERß agonists will be of benefit are prostate cancer, autoimmune diseases, colon cancer, malignancies of the immune system, and neurodegeneration.
- I. What Is an Estrogen?
- II. Ligand-Dependent Activation of ER
- III. Subtype Selective Ligands
- IV. Estrogen and the Female
- V. ERß in the Uterus
- VI. ERß in the Breast
- VII. The Prostate
- VIII. ERß in the Central Nervous System
- IX. ERß in Development of the Brain
- X. ERß and the Cardiovascular System
- XI. ERß and the Immune System
- XII. Molecular Basis of Agonism and Antagonism
- A. Direct antagonism
- B. Indirect antagonism
- C. Active/passive antagonism
- XIII. Selective Receptor Modulation
- XIV. Flexibility of the ER Binding Cavity Predicts the Development of Multiple Pharmaceuticals Targeting ER
- XV. Concluding Thoughts
|
I. What is an Estrogen?
|
|---|
BEFORE 1995, ESTROGENIC was synonymous with uterotropic, the ability to stimulate proliferation and induce expression of the progesterone receptor in the uterus. When estrogen receptor (ER) is inactivated (in ER –/– mice) (1), the uterus shows very little response to estrogen (2), and when there is no 17ß-estradiol (E2) (as in aromatase-deficient mice) (3), the uterus does not grow. In contrast, when ERß is inactivated, the uterus, if anything, is larger than it is in normal mice, and the response to E2 is stronger (4). To many, the lack of effect of ERß on E2-induced proliferation was proof that ERß is a nonfunctional receptor. We know that ER is the dominant receptor in the adult uterus, and this is why loss of ERß does not affect the response of the uterus to E2. ERß is expressed at high levels in other estrogen-target tissues such as the prostate (5), salivary glands (6), testis (7), ovary (8), vascular endothelium (9) and smooth muscle (10), certain neurons in the central (11, 12) and peripheral (13) nervous systems, and the immune system (14). In these tissues, which depend on E2 for maintenance of structure and/or function, E2 signals are mediated by ERß. In cell lines (15), E2 in the presence of ER elicits proliferation, but in the presence of ERß, it inhibits proliferation—one and the same hormone, two opposite effects.
This dual action of estrogen is most clearly demonstrated in HC11 cells, immortalized cells obtained from lactating mammary glands. These cells express both ER and ERß but do not proliferate in response to E2 (16). When they are treated with a selective ER agonist, these cells proliferate, and when they are treated with a selective ERß agonist, their growth is inhibited. E2 activates both ER and ERß, and the net result is no effect on growth (17). HC11 cells may be mimicking the situation in the lactating mammary gland in vivo, where high levels of both ER and ERß are coexpressed in almost every epithelial cell (18), but the gland is completely nonresponsive to the proliferative effects of E2 (18, 19).
The standard test for an estrogen, i.e., stimulation of growth of the uterus, is, of course, still a good test for an ER agonist, but there is no single corresponding good test for an ERß agonist. In fact, there may not be such a thing as a single good ERß agonist. What is emerging is an array of ERß-selective agonists, each influencing a specific profile of genes. Although we know what is a consensus estrogen response element (ERE), most estrogen-responsive genes do not contain perfect consensus sequences, and the transcriptional activity of ER or ERß on various EREs is influenced by the chemical structure of the estrogenic ligand. Hall and Korach (20) have evaluated the activities of ER and ERß on four different EREs (vitellogenin A2, human pS2, lactoferrin, and complement 3) in the presence of E2, phytoestrogens, and xenoestrogens. In terms of transactivation by ER and ERß, the vitellogenin and lactoferrin promoters were not discriminatory. The pS2 and C3 EREs were most responsive to ERß but very weakly to ER. In addition, the transcriptional activity of either receptor on any promoter varied with the ligand.
Another factor influencing selectivity of ER ligands is that the influence of ERs on transcription is not confined to EREs. ERs modulate transcription at activator protein-1 (21) and specificity protein 1 (22) sites and interact with the nuclear factor-
B pathway (23). The action of the two receptors at these sites can be opposite to each other, but this depends on cellular context, and it is not possible to predict how ER or ERß will influence transcription at these sites. The notion that ERß is a weaker trans-activator than ER is based on data obtained with consensus EREs and has to be qualified to include the fact that the relative activity of these two receptors as trans-activators varies with the promoter. It should, therefore, be possible in the future to develop selective ER and ERß ligands based on actions on target tissues and even target genes.
|
II. Ligand-Dependent Activation of ER
|
|---|
ER and ERß belong to the nuclear hormone receptor family, many of whose members are ligand-activated transcription factors that regulate gene expression in a cell- and promoter-specific manner. These receptors contain an N-terminal DNA binding domain and a C-terminal ligand binding domain (LBD) (reviewed in Ref. 24). Binding to its cognate ligand causes a change in receptor conformation that results in dimerization and binding to specific promoter sequences of DNA called hormone response elements (HREs). The activated receptor/DNA complex then recruits other cofactors from the nucleus, which results in transcription of DNA downstream from the HRE into mRNA and eventually protein, which causes a change in cell function.
There are two activation domains in ER (Fig. 1
), an N-terminal ligand-independent activation function (AF-1) and a C-terminal ligand-dependent activation function (AF-2) (25). These act synergistically to recruit various coactivator proteins to the DNA/ER complex. Depending on cellular and promoter context, coactivator binding exclusively to AF-1 leads to either partial activation or no activation of ER (26). Full activation generally requires that coactivators bind simultaneously to both AF-1 and AF-2 (27). In contrast to ER, ERß appears to have diminished AF-1 activity while, at the same time, it possesses a fully functional AF-2 (28).
View larger version (24K):
[in this window]
[in a new window]
|
FIG. 1. Structural domains of the human ER and ERß. The 1D amino acid sequence is represented by colored bars (N terminus to the left, and C terminus to the right) with starting and ending amino acid residue numbers of each of the domains A-F denoted above the bars. Regions of the receptors responsible for specific functions (transcriptional activation, nuclear localization, dimerization, DNA-binding, coactivator and corepressor binding) are indicated by lines below the bars. NDT, N-terminal domain; DBD, DNA binding domain. [Adapted from C. M. Klinge: Steroids 65:227–251, 2000 (117 ). Copyright 2000, with permission from Elsevier.]
|
|
Coactivator proteins contain one or more nuclear receptor interaction domains with the canonical LXXLL motif, which directly interacts with AF-2. Numerous structures have been obtained of the LBDs of various nuclear receptors, including ER (29, 30, 31, 32), complexed with both agonist and coactivator peptides. In these crystal structures, the LXXLL motif is found to adopt an -helical conformation and binds to a cleft on the surface of the LBD formed by helices 3, 5, and 12 (H3, H5, H12) (Fig. 2A). Agonists bind to an internal cavity of the receptor that stabilizes the overall conformation of the LBD and, in particular, promote a conformation of H12 that favors coactivator recruitment.
View larger version (38K):
[in this window]
[in a new window]
|
FIG. 2. Depiction of experimental crystallographic structures. A, Human ER LBD complexed with the agonist diethylstilbestrol (top, PDB accession code 3ERD) (29 ). B, Human ER LBD complexed with the mixed agonist/antagonist 4-hydroxytamoxifen (middle, 3ERT) (29 ). C, Human PPAR complexed with the antagonist GW-6471 (bottom, 1KKQ) (106 ). The protein is depicted in green, except for helix-12, which is colored cyan; ß-sheets, yellow; the GR interacting protein (GRIP) coactivator peptide, red (top); and the SMRT corepressor peptide, magenta (bottom). Ligand atoms are white (carbon), red (oxygen), blue (nitrogen, not visible), and green (fluorine, not visible). These figures were created using the PyMol molecular graphics system [W. L. DeLano. The PyMOL Molecular Graphics System (2002), DeLano Scientific, San Carlos, CA. http://www.pymol.org].
|
|
|
III. Subtype Selective Ligands
|
|---|
Because ER and ERß have markedly different tissue distributions, one way to achieve tissue selectivity is through subtype selective ligands that preferentially bind to one or the other receptor. Although the two receptor isoforms share only 56% identity in the LBD, the residues that line the binding cavity are much more highly conserved, and only two amino acid differences are found (Fig. 3). The first difference is the amino acid that resides below the D-ring pocket of the binding cavity where ER contains Met-421, while the corresponding residue in ERß is Ile-373. A second amino acid difference is found above the D-ring pocket where ER contains Leu-384, while the corresponding residue in ERß is Met-336. These are fairly conservative amino acid substitutions given that side chains of these residues occupy approximately the same volume and possess approximately the same hydrophobicity. Nevertheless, these amino acid residue differences impart subtle differences between the two receptor binding cavities that can be exploited in the development of subtype selective ligands (33). Furthermore, amino acid differences between the two isoforms further removed from the binding cavity may also contribute to subtype selectivity.
View larger version (93K):
[in this window]
[in a new window]
|
FIG. 3. Modeled complexes of (A) ER selective 3, 17-dihydroxy-19-nor-17--pregna-1,3,5(10 )-triene-21–16-lactone (16-lactone estradiol) (122 ) (top) and (B) ERß-selective 8ß-vinyl estradiol (122 ) ligands (bottom) bound to ER and ERß. These models are adopted from Hillisch et al. (39 ) and are based on the ER/estradiol crystallographic structure (PDB accession code 1ERE) (113 ). The structure of the estradiol ligand in the crystallographic complex was mutated to either estradiol lactone or 8-vinyl estradiol using the build option of Maestro version 7.0 (Schrödinger, LLC, New York, NY). Hydrogen atoms were then added to the complex in Maestro (hydrogen atom treatment: All-atom with No-Lp), and the resulting models were energy minimized using the MMFF force field of MacroModel version 9.0 (Schrödinger) holding position of protein nonhydrogen atoms fixed while optimizing the positions of all ligand and hydrogen atoms. The protein is represented by green tubes as are critical conserved residues (Glu-353, Arg-394, and His-524), which make hydrogen bonding interactions (dashed yellow lines) with the ligand. Residues that differ between ER (Leu-384 and Met-421) and ERß (Met-336 and Ile-373) are represented by magenta and orange tubes, respectively. There is a repulsive steric interaction between the lactone and Ile-373 in ERß, which explains the ER selectivity of the 16-lactone estradiol ligand. Conversely, there is a repulsive steric interaction between the vinyl group and Leu-384 in ER, which accounts for the ERß selectivity of the 8ß-vinyl estradiol ligand.
|
|
Shortly after the discovery of ERß (34), it was found that certain estrogenic compounds display modest subtype binding and/or efficacy selectivity (35, 36). For example, the phytoestrogen genistein is about 30-fold ERß-selective (36, 37). In view of the relatively conservative amino acid differences in the binding cavity and the plasticity of both LBDs, it has been difficult to rationalize the binding selectivity of these modestly selective ligands. The crystallographic structure of ERß complexed with genistein (38) revealed no obvious reason why this ligand is ß-selective. In the ERß-genistein complex, the polar 5-hydroxy-4-oxo functionality of the bound ligand is adjacent to Met-336, which is slightly more polar than the hydrophobic Leu-384 in ER. In addition, the longer Met-336 in ERß is closer to the chromenone ring of genistein compared with the branched Leu-384 in ER, resulting in better packing in the ERß-genistein complex and hence higher affinity.
Schering AG (Berlin, Germany) described two isoform-selective ligands, which are modified estradiol derivatives (39). The first is an ER-selective ligand, estradiol 16-lactone (Fig. 3A
). The bulky lactone ring of the bound ligand is in close proximity to the Met-421 in ER and Ile-373 in ERß. The more flexible Met-421 residue in ER can apparently adopt a conformation that accommodates the lactone moiety of the ligand, whereas in ERß, the less flexible Ile-373 cannot avoid a significant steric clash with the lactone. Hence, this ligand displays significant binding selectivity for ER. In contrast, 8ß-vinyl estradiol (Fig. 3B) shows substantial binding selectivity for ERß. In the ER ligand-binding cavity, there is a steric clash between the vinyl group of the ligand and Leu-384. In the ERß binding pocket, the vinyl group encounters the more flexible Met-336, and this accounts for the ß-selectivity of this ligand. A number of additional subtype-selective ligands have been described. These include: PPT, which has been reported to be 410-fold selective in binding to ER over ERß (40) and which behaves as an agonist through ER and as an antagonist through ERß; oxathiin-6-ol (41), an ER-selective SERM (SERAM); the ERß-selective agonists DPN (42); Japan Tobacco indole (43); benzoxazole from AstraZeneca (44); indenoquinoline from Akzo Nobel (45); cyclopentachromene from Lilly (46); cyclopentaindene from Merck (47); and WAY-358 and ERB-041 from Wyeth (48, 49) (Fig. 4).
A site-directed mutagenesis experiment demonstrated that a single mutation in ER, changing amino acid 421 from Leu to Met, substantially increased the affinity of the mutant receptor for DPN, and conversely, a mutation in ERß converting Met-336 to Leu decreased the affinity for DPN (50). Therefore, most of the selectivity of DPN can be attributed to the Leu-421 (ER)/Met-336 (ERß) amino acid difference, which is located in the ß (top) face of the binding cavity. Residue differences at the beginning of H3 also appeared to have a secondary contribution to selectivity. Furthermore, the lowest energy docking of the S enantiomer of DPN to ERß places the nitrile group of the ligand in close proximity to the Met-336 residue (50). Hence, a more favorable interaction between the nitrile group of DPN and Met-336 in ERß accounts for the ß-selectivity of this ligand.
X-ray crystallographic analysis of the ERB-041 ligand reveals an alternative mechanism for achieving ß-selectivity (48). In the crystallographic structures of ERB-041 complexed with ERß, the fused heterocyclic ring occupies the C, D-ring pocket of the receptor, and the 7-vinyl group is adjacent to the Ile-373 residue. The authors attribute the selectivity of ERB-041 to two factors: 1) better packing of the Met-336 residue in ERß vs. Leu-384 in ER against the benzoxazole ring of the ligand; 2) increased steric repulsion between the 7-vinyl group of the ligand and Met-421 in ER relative to Ile-373 in ERß.
|
IV. Estrogen and the Female
|
|---|
It is frequently claimed that nature does not care about women after the reproductive years, hence the rapid aging effects on many organ systems in women after the menopause. The case can be made that the opposite is true. Exposure to the potent proliferative effects of estradiol (via ER) is necessary for reproduction but is associated with health risks and is limited to a part of a woman’s life. Other actions of estrogen are mediated by ERß, and there are numerous estrogens in plants that activate this receptor even when the ovary ceases to produce E2. As the Western diet becomes more and more refined, our intake of phytoestrogens is reduced. Diets rich in unrefined grains, fruits, and vegetables contain phytoestrogens of diverse chemical structures with varying degrees of selectivity for ER and ERß, and there is much evidence that such diets are associated with lower risks of cancer, inflammatory diseases, diabetes, and cardiovascular diseases (51, 52). It is probably best that, to obtain the full benefits of phytoestrogens, we should not focus on a single source of phytoestrogens (e.g., soy beans) or on the purified phytoestrogens themselves. Perhaps we should learn a lesson from ancient Chinese who understood the importance of combinations of multiple plant components in their medicines. Because we have accepted that vitamins A, B, C, D, and E must come from our diet, we should be able to accept the idea that phytoestrogens are vitamins as well.
|
V. ERß in the Uterus
|
|---|
The uterus and pituitary gland are special in that ERß is expressed during development and ER is expressed when the tissue matures (53, 54). The selective expression of ER in the pituitary (55) has enormous consequences for the use of estrogens, particularly in males. E2 via ER in the pituitary turns off gonadotropin secretion and silences the gonads, resulting in chemical castration. In females, this is not a problem because the exogenous E2 simply replaces the hormone secreted by the ovary. In males, it is a significant problem because testosterone synthesis is inhibited. When suitable ERß-selective agonists are developed, they could be useful in treatment of prostate cancer, cardiovascular, or central nervous system diseases, without the risk of affecting the pituitary.
In the immature uterus, ER and ERß are expressed at comparable levels in the epithelium and stroma; E2 treatment decreases ERß in the stroma and increases ER in both compartments (4). In the uterus of immature ERß –/– mice, progesterone receptor and the proliferation marker, Ki-67, are expressed at higher levels than in immature wild-type mice, and the ERß –/– uterus exhibits exaggerated responsiveness to E2 (4). It appears that ERß keeps the uterus quiescent before the ovary begins to secrete E2. The question then is how does ERß keep the uterus silent in the absence of E2? We suggested that the ligand for ERß is 5-androstane-3ß, 17ß-diol (3ßAdiol), which is known to be secreted by the prepubertal ovary (56). In the mature uterus, ERß plays a role in cervical ripening, which is essential for parturition (57), and in decidualization, which is essential for implantation of the fetus (58).
In terms of diseases of the urogenital tract, ERß may be involved in some common and debilitating diseases, e.g., endometriosis, urinary incontinence, and uterine prolapse. Endometriosis is inappropriate, E2-dependent growth of uterine tissue in the abdominal cavity. It is very painful and is common in young women during the reproductive years (59). Studies show that the ratio of ER to ERß mRNA is increased in endometriotic tissue (60). Because of its antiproliferative effects in the uterus, ERß may have a role to play in controlling growth of endometriotic lesions. Uterine prolapse and urinary incontinence are not uncommon in elderly women who are not replaced with E2. Although the reasons for this syndrome are not completely understood, there are abnormalities of connective tissue structure and/or repair in response to stress and loss of elastic recoil in the cardinal ligaments (61). Ligaments of prolapsed uteri are characterized by a higher expression of collagen III and tenascin and lower quantities of elastin (62) and ERß (63). The decrease in the ratio of collagen I/(III+V) seen in postmenopausal women who are not on HRT is thought to reduce the tensile strength of collagen and increase susceptibility to anterior vaginal wall prolapse. The role of E2 in uterine prolapse/urinary incontinence was made clear during phase 3 trials with the selective ER modulator, levormeloxifene, for treatment of osteoporosis. Levormeloxifene binds to both ER and ERß and has been shown to decrease ERß mRNA in some brain areas (64). In a clinical trial on the use of levormeloxifene for the treatment of osteoporosis, uterovaginal prolapse occurred in 7% and urinary incontinence in 17% of the women in the levormeloxifene groups, and the trial was stopped (65). Interestingly, the uterus is not the only organ where ERß plays a role in elastic recoil; one characteristic of ERß –/– mice is loss of elastic recoil in the lung (66). It is possible that ERß is involved in elastin and collagen homeostasis in the body either through transcriptional effects on the elastin and collagen genes or through regulation of some of the proteases that are involved in the degradation of these proteins (67).
|
VI. ERß in the Breast
|
|---|
For the past 30 yr or so, ER has been indispensable in classifying the hormone sensitivity of breast cancers and in determining whether women will respond to the ER antagonist, tamoxifen (68). Tamoxifen is useful only in breast cancers that express ER. When ERß was discovered in 1995, there were doubts as to whether it could have a role in breast cancer because responsiveness to tamoxifen was so clearly dependent on the presence of ER. ER is essential for ductal growth, and in ER –/– mice there is very little growth of mammary ducts (1). However, despite its clear effects on ductal proliferation, ER is never colocalized in cells with proliferation markers (18, 19). This led to the dogma that E2-stimulated ductal proliferation was indirectly mediated by growth factor secretion from the stroma (69). The stroma of the adult human and rodent mammary glands expresses ERß but not ER (18, 19, 70), so if proliferation is mediated by estrogen-stimulated release of stromal growth factors, these would be dependent on ERß signaling. This is clearly not the case because in ERß –/– mice, there is normal ductal branching in the mammary glands (71). To solve this riddle, we reexamined the proposition that stromal growth factors are responsible for E2-stimulated ductal growth. We found that ER in the epithelial cells is the receptor that receives the proliferation signal from E2, but very early in G1, ER is lost from the nucleus (18). Thus, ER is never coexpressed with Ki-67, cyclin A, or proliferating cell nuclear antigen. If cellular DNA is labeled with bromodeoxyuridine, ER is not detectable during DNA synthesis but is reexpressed in bromodeoxyuridine-labeled daughter cells (18).
This still leaves us with the question of what is the role of ERß in the mammary gland. We know now that ERß is the more abundant ER in the normal breast and examination of the ductal epithelium of ERß –/– mice suggests that it is a prodifferentiative factor (71). Some, but not all, proliferating cells express ERß, indicating that ERß is not essential for proliferation. In vitro experiments with MCF7 cells suggest that ERß is antiproliferative (15). ER-positive MCF7 breast cancer cells respond to E2 with increased proliferation. When ERß is introduced into these cells, E2-induced proliferation is inhibited (15). Ductal cells in the mammary gland appear to be one example of cells where ER and ERß oppose each other on proliferation and the proliferative response to E2 is determined by the ratio of ER/ ERß. The functions of ERß in the breast are probably related to its antiproliferative as well as its prodifferentiative functions. Studies on the ERß –/– mouse mammary gland (71) showed that ERß regulates levels of several proteins characteristic of differentiated cells. These include the adhesion molecules, E-cadherin and integrin 2; the gap junction protein, connexin 32; and the tight junction protein, occludin. The presence of ERß in a breast cancer may be an indication of a more benign cancer, because degree of differentiation is the decisive factor in the aggressiveness and invasiveness of cancers.
Although there are several published studies on the expression of ER and ERß in breast cancer, the sample size in these studies has been small, and it will be some time before clinically relevant conclusions can be drawn. Very early after the discovery of ERß, in collaboration with Charles Coombes at Cancer Research-UK Laboratories Imperial College (London, UK), we measured ER and ERß in 100 frozen breast cancer samples and compared these profiles with those of normal breasts and fibrocystic breasts. Sucrose density gradient centrifugation with E2 as ligand was used to assess binding to ER and ERß and Western blotting to confirm the presence of the proteins. In 1998, this study was of no interest to the major cancer journals, so some of the results were published in a review in Endocrine Related Cancers in 1999 (72). What we found in the study was that ERß is dominant in normal and fibrocystic breasts and in medullary cancer. Ductal cancer grade 1 was characterized by a loss of ERß and a high expression of ER, grade 3 was characterized by a loss of both receptors, whereas grade 2 had some cancers expressing one or the other receptor and some expressing both receptors. Our data are compatible with the clinical observations that grade 1 cancers are responsive to tamoxifen and grade 3 cancers, i.e. those with no ER, are not responsive. The absence of ER in medullary cancers explains why these cancers do not respond to tamoxifen, but the presence of ERß offers the promise that ERß-selective ligands might be of some use in treatment of these cancers. If ERß is prodifferentiative and is lost from breast cancer, one novel strategy in breast cancer treatment could be the use of inducers of ERß as well as ERß-selective agonists.
|
VII. The Prostate
|
|---|
People frequently ask why we were looking for an ER in the prostate. The truth is that because of the confusing endocrinology of the prostate, George Kuiper, then a postdoctoral fellow, was looking for a second androgen receptor. Some of us were interested in discovering any nuclear receptor that would help us to understand the biology of 3ßAdiol, a dihydrotestosterone (DHT) metabolite whose biology we had been studying for several years (73, 74, 75). It turns out that ERß, the androgen receptor, 3ßAdiol, and 3ßAdiol hydroxylase are all part of a novel pathway, important for regulation of prostate growth and differentiation (75). Study of this pathway led us to a very heretic conclusion, i.e., that the major biological function of DHT is not that it is a better androgen than is testosterone but that it is the precursor of 3ßAdiol, the second estrogen in the body. The synthesis of this estrogen from DHT in the prostate provides the endogenous ligand for ERß and permits ERß to exert its antiproliferative effects on the prostate.
The antiproliferative effects of ERß were evident to us when we examined ERß –/– mice. In the ventral prostates of these mice, there were foci of epithelial hyperplasia (5, 76). We found that the proliferation in the epithelium was higher and the apoptosis index was lower in prostates of mice lacking ERß (76). Furthermore, with markers of basal, intermediate and fully differentiated cells we concluded that, in the absence of ERß, the epithelium does not fully differentiate and a large fraction of the epithelial cells retain the capacity to proliferate. These findings led us to make two further predictions: 1) ERß ligands will be useful in regulating prostate growth and will be useful in treatment or prevention of prostate cancer; and 2) use of drugs to block 5-reductase (finasteride) is not a wise strategy because, in preventing formation of DHT, we also prevent formation of 3ßAdiol and thus remove the endocrine pathway that limits prostate growth and promotes differentiation. Both predictions turned out to be correct. In 2004, Eli Lilly and Co. (Indianapolis, IN) announced that the use of ERß-selective ligands reduces normal prostate growth and inhibits tumor formation when prostate cancer cells are injected sc into nude mice (77). In 2003, the results of the Prostate Cancer Prevention Trial were published (78). In this study, more than 18,000 healthy volunteers were randomized into two arms, finasteride 5 mg daily and placebo, respectively. As expected, the incidence of prostate cancer was lower in the finasteride arm but, unexpectedly, the incidence of Gleason scores 7–10 was higher. This increase in more aggressive, less differentiated cancers has caused a lot of concern in the medical community (79) and prompted us to suggest (80) that in order for patients to experience the benefits of loss of DHT without the unwanted effects of the loss of 3ßAdiol, they should be given ERß agonists, such as soy phytoestrogens, along with 5-reductase inhibitors.
|
VIII. ERß in the Central Nervous System
|
|---|
In addition to its influences on development, plasticity, and survival of neurons, estrogen has effects on several neurotransmitter systems in the brain. It is an important regulator of serotonergic, dopaminergic, and cholinergic neurons. ER is the predominant receptor subtype in the basal forebrain cholinergic neurons of the adult rat brain (81) where it is thought to enhance cognitive functions by modulating the production of acetylcholine. In primates, ER is present in neurons of the basal forebrain but is not expressed in cholinergic neurons (82). The issue of whether estrogen replacement therapy decreases the likelihood of developing Alzheimer’s disease remains unresolved, with some epidemiological studies suggesting beneficial effects of estrogen replacement and others showing no effect (83, 84, 85). Although the possible mechanisms of estrogen’s protective effect are still under investigation, estrogen is probably involved in prevention or slowing of neurodegenerative processes and not in reversing neurodegeneration. This means that the time at which estrogen replacement is initiated is likely to be an important factor in determination of how efficacious estrogen will be in protecting the brains of postmenopausal women.
In the anterior dorsal raphe nucleus, estradiol increases the amount of the serotonin receptor mRNA and the serotonin transporter mRNA. ERß but not ER is expressed in serotonergic neurons in the dorsal raphe nucleus of the mouse (86). This is the system that is targeted by selective serotonin reuptake inhibitors (SSRI) in the treatment of depression. There are still some questions about species differences in ERß expression in the dorsal raphe nucleus. The rat appears to be different from the mouse in that there is no ERß in the rat dorsal raphe (87), but in primates, ERß is expressed in serotonergic neurons (88). In cynomolgus monkeys, phytoestrogens from soy, which are relatively selective for ERß, improve mood and enhance serotonergic transmission in the dorsal raphe (89). It is therefore possible that in the future, novel antidepressant drugs will be designed on the basis of their ERß agonistic activity in the dorsal raphe nucleus.
The dopaminergic system is also estrogen responsive, and estrogen can prevent or modulate insults to dopaminergic neurons. Within the nigrostriatal dopaminergic system, estrogen modulates tyrosine hydroxylase activity, dopamine metabolism, and dopamine receptors (90, 91, 92, 93). Parkinson’s disease is more prevalent in males, and long-term estrogen therapy enhances dopaminergic responsiveness in postmenopausal women. Our own studies have shown that in ERß –/– mice, neuronal survival throughout life is compromised so that by 2 yr of age there is a remarkable amount of neurodegeneration, particularly in the substantia nigra (94, 95). Thus, estrogens may influence the course of Parkinson’s disease, and E2 deficiency after the menopause may explain why there is a high incidence of late-onset neuropsychiatric disorders in females.
|
IX. ERß in Development of the Brain
|
|---|
In ERß –/– mouse fetuses, newborn neurons generated in the ventricular zone cannot find their way to their correct position in the layers of the cortex, and many die on the way (95). This leads to a neuronal deficit in layers three and four in the somatosensory cortex in the brains of adult mice. The mechanisms behind this defect in neuronal migration have been intensively investigated. We found by microarray analysis that several genes involved in neuronal migration were poorly expressed in the fetal ERß –/– mouse brain (Table 1). This list of genes does not tell us whether we were simply confirming that there was a migration deficit or whether we are looking at directly ERß-regulated genes. Failure of neurons to migrate does not necessarily mean that the ERß in neurons is responsible for neuronal migration. There could be indirect effects, such as defects in radial glial cells or in secretion of glucocorticoids from the fetal adrenal cortex where ERß is expressed. Before it can be determined whether neuronal ERß might be involved in the embryonic brain, ERß-expressing cells in the developing brain will have to be identified.
|
X. ERß and the Cardiovascular System
|
|---|
E2 is implicated in many vascular disorders. The risk of cardiovascular disease increases after the decline in E2 at menopause, and E2 replacement relieves perimenopausal hot flushes. E2 may have direct effects on endothelial and smooth muscle cells in the blood vessels and may have indirect effects through its actions in the lung and nervous system. One very controversial issue has been the question of whether or not there are ERs in the heart muscle. Neonatal cardiomyocytes in culture express both ER and ERß in their nuclei, but these receptors are not detectable until E2 is added to the culture medium (96). Some studies have reported that ERß cannot be detected in the adult heart (97), whereas in others, ERß is abundant in the nucleus (98, 99, 100) and in one study ERß was exclusively mitochondrial (101). In our own studies, we have been unable to detect either ER or ERß in cardiac muscle of mice with RT-PCR, Western blotting, or immunohistochemistry (102). Nor can we detect any ER in fetal mouse hearts. ERß is very abundant in the alveoli of the lung, and ER is abundant in the large bronchi. Both receptors are detected in the lungs in heart-lung preparations where the heart muscle is completely devoid of signals. Unless the ERs in the heart are in some way modified so as to be unrecognizable by the antibodies used, we have to conclude that these receptors are not expressed in the heart muscle.
|
XI. ERß and the Immune System
|
|---|
In the immune system, the division of labor between ER and ERß is very clear. ERß functions early in the bone marrow to regulate proliferation of the progenitor cells and loss of ERß in ERß –/– mice leads to myeloproliferative disease (14). ER has its important functions in the spleen and thymus, and loss of ER leads to autoimmune disease with attendant glomerulonephritis (103). The dual nature of estrogen in the immune system has been known for years. During the reproductive years, women suffer from systemic lupus erythematosus, and after menopause they suffer from inflammatory diseases such as arthritis. It is interesting that loss of E2 in aromatase knockout mice leads to Sjögren’s syndrome, an autoimmune disease affecting exocrine glands (104). It is very likely that selective ER modulators will be used to treat autoimmune diseases, rheumatoid arthritis, systemic lupus erythematosus, and certain malignancies of the immune system. Indeed, Wyeth-Ayerst has already developed ERß agonists showing promising effects in animal models of rheumatoid arthritis and inflammatory bowel disease (105).
|
XII. Molecular Basis of Agonism and Antagonism (Fig. 5)
|
|---|
A. Direct antagonism
Antagonists and agonists bind to the same internal cavity of the ER LBD, but antagonists sterically prevent H12 from adopting the agonist conformation. Instead, antagonists induce a positioning of H12 over the site where coactivators would normally contact the LBD, and thus coactivator recruitment to AF-2 is blocked (Fig. 2B). Shiau et al. (30) have used the term "active" to describe antagonists that directly displace H12; however, we prefer the term "direct" to avoid confusion with "active/passive" terminology that has been used to describe the behavior of steroidal antagonists of the glucocorticoid and progesterone receptors (see Section XII.C) (Table 2).
View larger version (22K):
[in this window]
[in a new window]
|
FIG. 5. Chemical structure of agonists and antagonists. The side chains of direct antagonists that displace helix 12 are highlighted by ovals.
|
|
No crystallographic structures of ER complexed with a corepressor peptide are yet available, but a crystal structure of peroxisome proliferator-activated receptor (PPAR) 
complexed with the antagonist GW-6471 and the corepressor peptide silencing mediator of retinoid and thyroid receptors (SMRT) is available (106), and it is probably a good approximation of the ER complex because the amino acid residues involved in corepressor recruitment are highly conserved across the members of the nuclear receptor family (106). In the PPAR complex (Fig. 2C), the corepressor occupies the same cleft as do coactivators, but H12 forms less extensive contacts with the remainder of the LBD, and this means a less stable cofactor/receptor complex (106). However, there are more extensive hydrophobic interactions and additional hydrogen bonds between the LBD and corepressor (106), and these additional stabilizing interactions are sufficient to overcome the destabilization of H12. Conversely, in the agonist conformation of the LBD, the coactivator LXXLL motif is perfectly accommodated in the coactivator binding cleft.
B. Indirect antagonism
There is another class of nuclear receptor antagonists that have molecular volumes roughly comparable to agonists and lack a side chain that sterically displaces H12. Examples include the AR antagonist flutamide, the ERß antagonist tetrahydrochrysene (30), and the mineralocorticoid receptor antagonist progesterone (107). Because H12 is not directly displaced by this class of ligands, we refer to this type of antagonism as "indirect." H12 of the LBD appears to be in equilibrium between two conformations: the first favors coactivator recruitment (Fig. 2A), and the second blocks coactivator binding (Fig. 2B). Agonists shift the conformational equilibrium strongly in the former direction, whereas indirect antagonists apparently do not stabilize this conformation. For those ER ligands that lack side chains and thus do not displace H12, the key determinant of whether they function as agonists or indirect antagonists is their positioning with respect to His-524 (30). His-524 is also involved in stabilizing H12 in the agonist conformation through its hydrophobic contacts with Met-528, which stabilize Val-533 and Val-534 on the loop connecting H11 and H12.
The conformational equilibrium position for H12 differs between ER and ERß. In ERß, the antagonist conformation is favored more than it is in ER. Hence, the ligand tetrahydrocrysene functions as an antagonist when bound to ERß and an agonist when bound to ER (30). The agonist orientation of H12 in ERß is less stable than it is in ER. In ER, the phenolic hydroxyl group of Tyr-537 at the beginning of H12 (hydrogen bond acceptor), forms a hydrogen bond with Asn-348. In ERß, Tyr-537 is replaced by a lysine that is unable to function as hydrogen bond acceptor at physiological pH. This additional hydrogen bonding interaction, present in the agonist conformation of ER but absent in ERß, stabilizes the agonist conformation of ER relative to ERß.
C. Active/passive antagonism
Two types of nuclear receptor antagonists have been described by Wagner et al. for the glucocorticoid (GR) and progesterone receptors based on a two-step "active inhibition" model (108). The first step involves competitive inhibition of agonist binding (which we refer to as "passive" antagonism). A second step involves binding of the receptor/antagonist complex to HREs to produce a transcriptionally inactive complex that is able to block the binding of activated receptors to DNA ("active" antagonism). Depending on the conformation of the receptor induced by the bound antagonist, the antagonist may function solely as a passive antagonist or, in addition, as an active antagonist. For example, the ligand RU-486 behaves as an active antagonist of GR, whereas close analogs of RU-486 such as RTI 3021-012 and 3021-022 behave primarily as passive antagonists. The active antagonist RU-486 induces a unique conformation in the GR, which causes nuclear translocation, binding to HREs, and recruitment of nuclear receptor corepressors SMRT and NCoR. The passive antagonists RTI 3021-012 and 3021-022, in contrast, induce alternative GR conformations with impaired nuclear translocation, DNA binding, and ability to bind corepressors. Under certain cellular and promoter contexts, passive antagonists may display partial agonist properties. In addition, because active antagonists block the binding of activated receptors to DNA, active antagonists generally possess higher antagonist potency (i.e., biochemical efficiency) than passive antagonists (109).
At least in some cellular and promoter contexts, the mixed agonist/antagonist tamoxifen and raloxifene appear to function as active antagonists (110). In addition, TAS-108 has been reported to promote a conformation of ER capable of binding to DNA and corepressors (111) and therefore by definition is an active antagonist. In contrast, the "pure" antagonist ICI-182,780-ER complex, does not bind DNA nor does it recruit corepressor proteins. Insight into the lack of corepressor binding to the ER-ICI-182,780 complex was gained by examination of the crystallographic structure (112). ICI-182,780 is a derivative of E2 that binds in a flipped orientation relative to normal E2 binding mode (113). The 7 substituent then essentially occupies the location normally occupied by the 11ß position of E2 in its typical binding mode and hence is positioned to displace H12. In addition to displacing H12, the end of the long hydrophobic side chain occupies the coactivator/corepressor binding cleft (Fig. 2, B and C, respectively), hence the binding of both coactivator and corepressors is blocked.
|
XIII. Selective Receptor Modulation
|
|---|
Depending on the cellular and promoter context, certain ER ligands can function either as agonists or antagonists. For example, the high affinity ER ligand, tamoxifen, functions as an antagonist in breast and agonist in the uterus (111). In contrast, raloxifene functions as an antagonist in both tissues. One mechanism contributing to the tissue-selective effects of tamoxifen is the ratio of coactivators to corepressors expressed in the tissue. In particular, steroid receptor coactivator-1 is not coexpressed with ER in the mammary epithelium, but the two proteins are coexpressed in the endometrium (111), and the levels of expression of steroid receptor coactivator-1 are higher in the endometrium than in the breast. Increased expression of coactivators shifts the conformational equilibrium of the ER/tamoxifen complex in the agonist direction. In contrast to tamoxifen, raloxifene has a longer side chain that appears to more strongly maintain the conformational equilibrium in the antagonist direction, regardless of the coactivator expression levels.
Binding of different agonists may cause small shifts in the position of H12 and thereby modulate the affinity of coactivators for nuclear receptors. Thus, ligands span a spectrum ranging from partial to full to super agonism. Furthermore, different coactivators may induce slightly different conformations in the coactivator binding cleft, and hence different agonists may induce small changes in the conformation of the LBD that would lead to preferential association of certain coactivators (114, 115, 116). Because coactivator populations may differ between cell types, this provides an additional mechanism for selective receptor modulation.
|
XIV. Flexibility of the ER Binding Cavity Predicts the Development of Multiple Pharmaceuticals Targeting ER
|
|---|
Top
Abstract
I. What is an...
