Estrogen Receptors α and β
Prevalence of Estrogen Receptor β mRNA in Human Vascular Smooth Muscle and Transcriptional Effects
Background—Estrogens have vascular effects through the activation of estrogen receptors (ERs). In addition to ERα, the first ER to be cloned, a second subtype called ERβ has recently been discovered.
Methods and Results—Using a reverse-transcriptase polymerase chain reaction assay that employs the same primer pair to simultaneously amplify ERα and ERβ transcripts, we found that ERβ is the ER form that is predominantly expressed in human vascular smooth muscle, particularly in women. The transcriptional effects of the 2 ERs in transfected HeLa cells differed. In response to 17β-estradiol, ERα is a stronger transactivator than ERβ at low receptor concentrations. However, at higher receptor concentrations, ERα activity self-squelches, and ERβ is a stronger transactivator. Tamoxifen has partial agonist effects with ERα but not with ERβ.
Conclusions—The protective effects of estrogens in the cardiovascular system of women may be due to the genomic effects of ERβ in vascular tissue.
Premenopausal women with normal estrogen levels rarely manifest coronary disease. In addition, the administration of exogenous estrogens to healthy postmenopausal women markedly reduces the incidence of coronary events.1 However, the mechanisms underlying this benefit are unknown. The modest cholesterol-lowering effect of estrogens does not explain their substantial protective effects.2 An alternative hypothesis is that estrogens protect against coronary heart disease through genomic mechanisms in the vascular bed. Estrogens inhibit the growth of vascular smooth muscle (VSM), which characterizes obstructive atherosclerotic lesions.3 4 Functional estrogen receptors (ERs) are present in VSM.5
The structure of ERα was elucidated in 1987.6 A second ER cDNA, designated ERβ, was cloned in l996.7 8 The 2 ERs have a similar affinity for 17β-estradiol, and they bind to the same estrogen response elements (EREs).9 However, ERβ differs from ERα in 2 important functional domains. First, the N-terminus containing the AF-1 (activation function-1) activation domain of ERβ has only 30% homology with that of ERα.7 8 This domain is critical to the partial agonist properties of some antiestrogens, and it may be involved in tissue specificity.10 Second, the hormone-binding domains (HBD) of ERα and ERβ have only 53% homology.7 8
Why estrogens enhance the proliferation of breast or uterine cells but inhibit the proliferation of VSM cells is an enigma. One possible explanation for this heterogeneity of estrogen action is that the vascular bed expresses structurally and functionally different ERs than the breast or uterus. ERα is the predominant receptor in the breast and uterus.11 However, because estrogens inhibit VSM proliferation in response to vascular injury in knockout mice that lack ERα,12 ERα is not essential to estrogen action in the vascular wall. Cynomolgus monkeys express both ERα and ERβ mRNA in coronary artery and cultured aortic smooth muscle cells.13 No information about the distribution of wild-type ERs in human VSM exists.
This study had 2 goals: to describe the relative expression of ERα and ERβ transcripts in VSM and to elucidate whether differences exist in the transcriptional activation and function of ERα and ERβ.
RNA Isolation and RT-PCR
VSM tissue from the human coronary artery, iliac artery, aorta, or saphenous vein was cultured for 72 hours and then frozen at −80°C, as described previously.14 Total RNA was isolated with TriReagent (Sigma).
RNA (0.5 to 1.0 μg) was heated to 70°C for 5 minutes, and 125 U of MuLV RT was added. Using the random hexamer oligo primer supplied in a GeneAmp RNA PCR kit (Perkin Elmer), an RT reaction was performed at 42°C for 60 minutes. After dilution of the RT cDNA product, PCR was performed with the following thermal cycles: 1 cycle at 94°C for 1 minute, followed by 35 cycles at 94°C for 15 s, 55°C for 25 s, and 72°C for 30 s, and 1 final extension cycle at 72°C for 7 minutes. Products were resolved on an 8% polyacrylamide gel in 1× Tris-borate/EDTA buffer (0.09 mol/L Tris-borate and 0.002 mol/L EDTA, pH 8.0).
To simultaneously amplify ERα and ERβ in the same PCR reaction, oligonucleotide primers were designed for PCR amplification of specific DNA fragments contained in both ERα and ERβ (Figure 1⇓). The forward primer sequence, ERF2, was AAGAGCTGCCAGGCCTGCCG, and the reverse primer sequence, ERR1, was GCCCAG-CTGATCATGTGAACCA. There was one mismatch in ERR1 for both ERβ (AT mismatch) and ERα (AG mismatch). The primer ERR1 had similar stability on both templates. The primer pair ERF2 and ERR1 generated a 382-bp fragment for ERα and a 346-bp fragment for ERβ. These primers were tested on ERα and ERβ cDNA plasmid clones to ensure that they generated the specific products targeted, with equivalent efficiency under the same amplification conditions.
Primers specific for ERα or ERβ only, but capable of distinguishing the wild-type from the variant forms of each, were also designed. The ERα-specific primers, a forward primer ERAF1 (GTCTCCGAGCCCGCTGATGCTACTGCAC) and a reverse primer ERAR1 (CGGATGCCCCTCCACGGCTAGTGG), generated a 1372-bp fragment. The ERβ-specific primers, forward primer ERBF1 (CGTGATGGAGGACTTGCACCCGCGAAGCAC) and reverse primer ERBR1 (TCCCTGGTGTGAAGCAAGTATCG-CTAGAACA), generated a 1212-bp fragment.
Wild-type human ERα expression vector was provided by Dr Pierre Chambon (Strasbourg, France). The human ERβ Bluescript KS construct was provided by Drs Eva Enmark and Jan-Ake Gustafsson (Stockholm, Sweden). The 1460-bp ERβ insert was recloned into the pSG5 expression vector (Stratagene) to generate pSG5-hER. The reporter plasmids were VIT-tk-CAT15 and ERE2-TATAtk-CAT.16
Transfection and Reporter Assays
Transfection experiments used 10 ng of human ERα expression vector or ERβ plasmid, 2 μg of ERE2-TATAtk-CAT or VIT-tk-CAT reporter, 3 μg of the β-galactosidase expression plasmid pCH-110 (Amersham Pharmacia Biotechnology) to correct for transfection efficiency, and 15 μg of Bluescribe carrier plasmid, for a total of 20 μg of DNA/plate.16 Cells were treated for 24 hours with 10 nmol/L 17β-estradiol and/or 100 nmol/L tamoxifen or ICI 182,780 (ICI Pharmaceuticals). Cell lysates were analyzed for CAT activity, as previously described,16 with chloramphenicol acetylation measured by thin-layer chromatography and quantified by phosphorimaging (Molecular Dynamics).
Statistical analysis was performed with SAS software (SAS Institute). The 2-tailed significance level was 0.05. Independent groups were compared using the unpaired t test, and dose response was assessed by linear regression.
Distribution of ERα and ERβ Transcripts in Human VSM
To investigate the relative levels of ERα and ERβ expression in human VSM, we designed primers that coamplified different transcript fragment lengths for each subtype in the same polymerase chain reaction (PCR). The combined ERα/β primer set (ERF2/ERR1; see Methods for a complete description) amplifies both a region of the cDNA encoding the DNA binding domain (DBD) and the hinge region of the 2 receptors that has a gap in ERβ when compared with the ERα sequence (Figure 1A⇑). A 382-bp product was generated from the cDNA encoding ERα, and a 346-bp product from the cDNA encoding ERβ. Equal efficiency of the PCR amplification reactions was confirmed using control plasmid DNA containing the recombinant cDNA of ERα or ERβ in molar ratios of 1:3 (Figure 1B⇑, lane 1), 1:1 (lane 2), and 1:2 (lane 3). Therefore, it was possible to simultaneously measure the relative reverse-transcriptase (RT) PCR product amount for each ER subtype, despite the exponential nature of PCR and the minimal variations in template sequences and reaction efficiencies.
ERα and ERβ RT-PCR products (determined using the combined ERα/β primer sets) in VSM from 12 patients are shown in Figure 2A⇓. Although both ERα and ERβ transcripts were present in all samples, some subsets expressed considerably more ERβ than ERα. The values of ERβ, expressed as a percentage of total ER, in samples of VSM from 8 male and 12 female subjects, are shown in the Table⇓. ERβ was more prevalent in the VSM from female subjects (67±12%) than in those from male subjects (51±15%; P=0.02). ERβ did not correlate with age in either sex.
To determine if exposure to estrogens could alter the ERα/ERβ ratio, samples of VSM from 8 male and 12 female patients were treated for 72 hours with 17β-estradiol or vehicle. No consistent differences existed in the subtype ratios between estrogen-treated sets and controls. Studies of samples from 3 patients found no differences in ER ratios between tissue immediately frozen when obtained and tissue incubated for 72 hours in medium (data not shown).
The combined primer set quantifies relative amounts of each subtype, but it cannot discriminate between wild-type, full-length ERα or ERβ and HBD deletion variants of these subtypes. ERα variants are common in VSM.14 Therefore, we also used primers designed to amplify longer lengths of either ERα or ERβ specifically (Figure 1A⇑). These included ERAF1/ERAR1 for ERα and ERBF1/ERBR1 for ERβ, which included the N-terminus, DBD, hinge region, and HBD (Figure 2B⇑). The specific primers were tested using plasmid DNA containing ERα or ERβ to confirm that the PCR generated full-length products (data not shown). The major transcript encoding ERβ of the sample shown in Figure 2B⇑ is full-length (open arrow), although a small band representing a lower molecular weight variant (solid arrow) can be discerned. However, the ERα transcripts are present predominantly as variant forms (solid arrows): there is little full-length, wild-type ERα present (open arrow). Similar results were obtained in the other 19 samples of VSM (data not shown). Therefore, the results with the combined ERα/ERβ primer set underestimate the ratio of wild-type ERβ to wild-type ERα transcripts.
Transcription by ERα and ERβ: Promoter and Ligand Specificity
To examine differences in promoter specificity between ERα and ERβ, ER-negative HeLa cells were cotransfected with either the ERE2-TATAtk-CAT (Figure 3A⇓) or vitellogenin (VIT-tk-CAT) (Figure 3B⇓) promoter/reporter constructs, as well as increasing concentrations (1 to 100 ng) of ERα or ERβ expression vectors. The cells were treated with vehicle or 10 nmol/L 17β-estradiol. The chloramphenicol transferase (CAT) transcription driven by each promoter was measured. Substantial estradiol-induced transcription occurred in cells with low levels of ERα; however, as the receptor concentration increased, transcription decreased due to a phenomenon called “self-squelching.” The mechanism underlying self-squelching is unknown, but a similar effect of the progesterone receptor A-isoform requires an intact HBD.17 Unlike ERα, ERβ had no transcriptional activity with ERE2-TATAtk-CAT (Figure 3A⇓). With VIT-tk-CAT, estradiol induced similar transcriptional activity in ERβ and ERα at low concentrations (1 and 5 ng cDNA; Figure 3B⇓). However, at higher receptor concentrations, because ERβ did not self-squelch, it was a more potent transactivator than ERα. The ERE2-TATAtk-CAT promoter construct contains 2 consecutive, synthetic EREs upstream of a minimal thymidine kinase promoter linked to the CAT gene.16 The VIT-tk-CAT promoter contains 2 repeats of the ERE found in the vitellogenin A1 gene promoter.15 Differences in the EREs and flanking sequences probably account for the differences in promoter recognition by the 2 ERs. Specifically, the VIT-tk-CAT promoter has several potential AP-1 sites that could modulate transcriptional activation by the 2 ERs.18 Such differences in specific gene transcription between the 2 receptors may underlie the variation in responses to 17β-estradiol observed in different tissues, including VSM.
To further analyze possible differences in the transcriptional activities of ERα and ERβ, we studied the differential effects of antiestrogens. Some antiestrogens possess partial agonist activity (ie, tamoxifen), and others are pure antagonists (ie, ICI 182,780). HeLa cells were cotransfected with ERα or ERβ and with the reporters ERE2-TATAtk-CAT (Figure 4A⇓) or VIT-tk-CAT (Figure 4B⇓). The transfected cells were treated with 17β-estradiol, tamoxifen, or ICI 182,780, either alone or in combination. Compared with 17β-estradiol, tamoxifen is a partial agonist on the ERE2-TATAtk-CAT reporter when bound to ERα; under these conditions, tamoxifen did not suppress the agonist effects of 17β-estradiol (Figure 4A⇓). ICI 182,780 had no agonist effect on this promoter and strongly suppressed the transcription induced by 17β-estradiol. ERβ did not induce the transcription of ERE2-TATAtk-CAT with any ligand (Figure 4A⇓).
Tamoxifen is also an agonist in the presence of ERα on the VIT-tk-CAT reporter (Figure 4B⇑). Again, ICI 182,780 is a pure antagonist. When bound to ERβ with VIT-tk-CAT, tamoxifen lacked partial agonist activity and, unlike ICI 182,780, it did not suppress the agonist effects of 17β-estradiol.
L7/SPA is a novel transcriptional coactivator that enhances the partial agonist effect of tamoxifen but has no effect on transactivation by estrogens or pure antagonists.19 To determine whether any ERβ-mediated latent partial agonist activity by tamoxifen could be exposed, this antiestrogen was tested on both ER subtypes in the presence of an L7/SPA expression vector using the VIT-tk-CAT reporter (Figure 5⇓). Coexpression of L7/SPA increased the partial agonist effects of tamoxifen on ERα to levels that exceeded those obtained with 17β-estradiol. However, even in the presence of L7/SPA, no agonist effects of tamoxifen were detectable in cells transfected with ERβ constructs. Thus, a fundamental difference exists in the 2 ERs with respect to the actions of antiestrogens.
Effect of ERβ on Transcriptional Effects of ERα
Because ERα and ERβ are coexpressed in VSM, it is important to determine whether the presence of one ER subtype can modify the transcriptional effects of the other. We examined the effect of ERβ on ERα-controlled transcription of the VIT-tk-CAT reporter in cells treated with tamoxifen (Figure 6⇓). A 5-fold molar excess of ERβ substantially reduced the agonist effect of tamoxifen on ERα (P=0.005 for linear dose response and P=0.02 for dose of 0 versus 2.5). Therefore, at least one transcriptional effect of ERα can be inhibited by the presence of ERβ.
Considerable clinical information has supported the notion that estrogens protect against atherosclerosis, particularly coronary artery disease. Atherosclerotic-mediated events are rare in premenopausal women, and exogenous estrogen reduces their incidence in healthy postmenopausal women.1 However, in the Heart and Estrogen-progestin Replacement Study (HERS),20 a randomized, placebo-controlled protocol in postmenopausal women with preexisting coronary artery disease, treatment with 0.625 mg of conjugated equine estrogens with 2.5 mg of medroxyprogesterone acetate failed to reduce coronary events and increased thromboembolic events. Although it is possible that the large body of previous evidence showing that estrogens protect the vascular bed is incorrect, other explanations for the findings in the HERS study deserve consideration. (1) In light of the increase in thromboembolic events in the HERS study, an increased tendency to thrombosis due to the hormone therapy may have offset the beneficial effects in the vascular wall. (2) Evidence from animal studies indicates that concomitant administration of the progestin medroxyprogesterone acetate may have adverse vascular effects that oppose the protective effects of estrogens.21 22
Although the relationship between estrogens and vascular disease is poorly understood, evidence exists that estrogens have genomic effects on the vascular wall, which is probably mediated through ERs. These effects include the inhibition of VSM growth and migration in vitro and the inhibition of arterial intimal hyperplasia in vivo.3 4 Without dietary manipulation or vascular trauma, ovarian ablation in sheep induces aortic intimal hyperplasia, which can be prevented by the administration of an estrogen.23
A seeming paradox of estrogenic effects is the observation of growth inhibition in the vascular bed but growth stimulation in other target organs, such as the breast and uterus. Our demonstration that the newly discovered ERβ subtype is the most prevalent receptor mRNA in human VSM offers a possible explanation for the important differences in the effects of estrogens on the vascular bed compared with other organs. We offer evidence that ERβ differs in its transcriptional activation and effects from ERα, which is the most prevalent receptor in the breast and uterus.11 We propose that the physiological effects of estrogens or antiestrogens in a tissue depend to a considerable extent on the type of ER expressed in that tissue.
ERβ is the Prevalent Wild-Type ER mRNA in Human VSM
Because reliable monoclonal antibodies for human ERβ are not commercially available, we used RT-PCR to quantify transcript expression for both ERα and ERβ in human VSM. Our method allows the cDNA encoding each subtype to be amplified in the same reaction tube. Because the amplified products differ in size, they can be distinguished from one another on electrophoretic gels, and the relative percentage of each ER subtype in a sample can be quantified. Because this method cannot distinguish between wild-type and exon-deletion variants of ERα and ERβ, we also performed separate PCR reactions using primers that recognized longer transcripts specific for either ERα or ERβ.
We demonstrated that although both ERβ and ERα mRNA were present in all 20 subjects we studied, the more prevalent wild-type receptor mRNA was ERβ. In women, ERβ was present in higher quantities in most samples, and in men, ERβ and ERα were present in approximately equal quantities. Qualitative analysis of the mRNAs with specific primers demonstrated that ERβ was encoded primarily by full-length transcripts, but ERα transcripts included substantial amounts of putative exon-deletion variants.
The importance of variant ER messages due to exon deletions is unclear. In cells transfected with variants of ERα, stimulation with estrogens could result in anomalous transcriptional activities that dominantly inhibit or enhance the effects of wild-type ERα.14 24 25 There has been no confirmation that the ERα variant transcripts are translated into proteins in vivo. Thus, ERα exon-deletion variants may have no physiological effects. If so, the wild-type receptors would determine the estrogenic effects in VSM, and ERβ would be the major receptor subtype mediating transcription.
Potential Influence of Differences in Structure of ERα and ERβ on Function
Human ERβ differs significantly from ERα at the N-terminus (including the AF-1 region), the hinge region, and the HBD (including differences in the AF-2 region and F domain at the far C-terminus; Figure 1⇑). Although the functional domains of ERβ have not yet been delineated, these structural differences probably account for differences in transcriptional activities. The AF-1 region of ERα is responsible for the partial agonist effects of tamoxifen,10 26 and we found that tamoxifen lacks partial agonist activity on ERβ. Structural differences between the 2 subtypes in the AF-2 region and the F domain may also contribute to the differing effects of antiestrogens on the 2 receptors. In crystallographic analyses, the F domain generates an α-helix (helix 12), whose appropriate folding over the surface of the ligand-activated HBD is critical in imparting agonist versus antagonist information to the transcriptional machinery.27 28 The amino acid composition of the ERβ F domain differs considerably from that of ERα.7 8
Our data also allowed us to deduce important functional information about the hinge region between the DBD and HBD. We recently isolated a transcriptional coactivator (termed L7/SPA) that binds to the hinge region of ERα and enhances the agonist activity of antiestrogens, such as tamoxifen.19 Thus, the hinge region may be an important site for protein-protein interactions. Because tamoxifen-occupied ERβ is not influenced by L7/SPA (Figure 5⇑), the ERβ hinge region functions differently from that of ERα, which has only 19% homology with ERβ.
Possible Functional Significance of the Prevalence of ERβ in VSM
Although low levels of wild-type ERα induce strong estrogenic responses, self-squelching limits transcriptional activity at higher receptor levels. Because ERβ does not self-squelch, at high concentrations, ERβ is a more potent transactivator than ERα. Cotransfection of ERβ did not suppress estradiol-induced, ERα-dependent transcription (data not shown).
However, tamoxifen bound to ERβ suppressed the agonist effects of tamoxifen-occupied ERα. Complicating any analysis is the observation that the inhibitory potency of antagonists and their agonist effects are dependent on promoter context and cell-type specificity.29 We found that with ERα, tamoxifen had partial agonist effects on the ERE2-TATAtk-CAT promoter but not on the VIT-tk-CAT promoter. Tamoxifen-occupied ERβ lacks agonist effects on both promoters. Nevertheless, tamoxifen-occupied ERβ can have agonist effects on some promoters containing AP-1 sites.18
Potential Clinical Importance of ERβ Activation in VSM
Inhibition by estrogens of VSM growth in vitro can be prevented by the antiestrogen ICI 182,780 and by actinomycin D, an inhibitor of transcription.4 This observation supports the hypothesis that the effects of estrogens in VSM are primarily mediated through classic ER-dependent transcriptional mechanisms. Interestingly, in ERα-deficient mice (ERKO), estrogens inhibit VSM proliferation in response to vascular injury.12 Because ERβ is expressed in the vasculature of these mice, this receptor subtype seems to be sufficient for mediating the antiproliferative effects of estrogens.
If, as we propose, the predominant ER in a tissue determines the local physiological effects of estrogens, the clinical implications of the high prevalence of ERβ in the vascular bed are considerable. The differential effects of ERβ activation in the vascular bed offer an attractive hypothesis to explain why estrogens are inhibitory in this tissue but stimulatory in others.
This work was supported by grants HL55291, HL57144, CA26860, and DK58238 from the National Institutes of Health. We thank Dr A.D. Robertson for performing the statistical analysis.
- Received August 30, 1999.
- Revision received November 4, 1999.
- Accepted November 15, 1999.
- Copyright © 2000 by American Heart Association
Chen S, Li H, Durand J, Oparil S, Chen Y-F. Estrogen reduces myointimal proliferation after balloon injury of rat carotid artery. Circulation. 1996;93:577–84.
Horwitz KB, Horwitz LD. Canine vascular tissues are targets for androgens, estrogens, progestins, and glucocorticoids. J Clin Invest. 1982;69:750–758.
Kuiper GGJM, Enmark E, Pelto-Huikko M, Nilsson S, Gustafsson JA. Cloning of a novel estrogen receptor expressed in rat prostate and ovary. Proc Nat Acad Sci U S A. 1996;93:5925–5930.
Pace P, Taylor J, Suntharalingam S, Coombes RC, Ali S. Human estrogen receptor beta binds DNA in a manner similar to and dimerizes with estrogen receptor alpha. J Biol Chem. 1997;272:25832–25838.
McInerney EM, Katzenellenbogen BS. Different regions in activation function-1 of the human estrogen receptor required for antiestrogen- and estradiol-dependent transcription activation. J Biol Chem. 1996;27:24172–24178.
Hodges YK, Richer JK, Horwitz KB, Horwitz LD. Variant estrogen and progesterone messages in human vascular smooth muscle. Circulation. 1999;99:2688–2693.
Hovland AR, Powell RL, Takimoto GS, Tung L, Horwitz KB. An N-terminal inhibitory function, IF, suppresses transcription by the A-isoform but not the B-isoform of human progesterone receptors. J Biol Chem. 1998;273:5455–5460.
Paech K, Webb P, Kuiper GGJM, Nilsson S, Gustafsson J-A, Kushner PJ, Scanlan T. Differential ligand activation of estrogen receptors ERα and ERβ at AP1 sites. Science. 1997;277:1508–1510.
Hulley S, Grady D, Bush T, Furberg C, Herrington D, Riggs B, Vittinghoff E. Randomized trial of estrogen plus progestin for secondary prevention of coronary artery disease in postmenopausal women: heart and estrogen/progestin replacement study (HERS) research group. JAMA. 1998;280:605–613.
Selzman CH, Gaynor JS, Turner AS, Johnson SM, Horwitz LD, Whitehill TA, Harken AH. Ovarian ablation alone promotes aortic intimal hyperplasia and accumulation of fibroblast growth factor. Circulation. 1998;98:2049–2054.
Berry M, Metzger D, Chambon P. Role of the two activating domains of the oestrogen receptor in the cell-type and promoter-context dependent agonistic activity of the anti-oestrogen 4-hydroxytamoxifen. EMBO J. 1990;9:92811–92818.
Nichols M, Rientjes JM, Stewart AF. Different positioning of the ligand-binding domain helix 12 and the F domain of the estrogen receptor accounts for functional differences between agonists and antagonists. EMBO J. 1998;17:765–773.
Watanabe T, Inoue S, Ogawa S, Ishii Y, Hiroi H, Ikeda K, Orimo A, Muramatsu M. Agonistic effect of tamoxifen is dependent on cell type, ERE-promoter context, and estrogen receptor subtype: functional difference between estrogen receptors alpha and beta. Biochem Biophys Res Commun. 1997;236:140–145.