Truncated Estrogen Receptor α 46-kDa Isoform in Human Endothelial Cells
Relationship to Acute Activation of Nitric Oxide Synthase
Background— Estrogen acutely activates endothelial nitric oxide synthase (eNOS). However, the identity of the receptors involved in this rapid response remains unclear.
Methods and Results— We detected an estrogen receptor α (ERα) transcript in human endothelial cells that encodes a truncated 46-kDa ERα (Δ1a-hERα-46). A corresponding 46-kDa ERα protein was identified in endothelial cell lysates. Transfection of cDNAs encoding the full-length ERα (ERα-66) and Δ1a-hERα-46 resulted in appropriately sized recombinant proteins identified by anti-ERα antibodies. Confocal microscopy revealed that a proportion of both ERα-66 and hERα-46 was localized outside the nucleus and mediated specific cell-surface binding of estrogen as assessed by FITC-conjugated, BSA-estrogen binding studies. Both ERα isoforms colocalized with eNOS and mediated acute activation of eNOS in response to estrogen stimulation. However, estrogen-stimulated transcriptional activation mediated by Δ1a-hERα-46 was much less than with ERα-66. Furthermore, Δ1a-hERα-46 inhibited classical hERα-66–mediated transcriptional activation in a dominant-negative fashion.
Conclusions— These findings suggest that expression of an alternatively spliced, truncated ERα isoform in human endothelial cells confers a unique ability to mediate acute but not transcriptional responses to estrogen.
Received June 21, 2002; revision received August 28, 2002; accepted September 13, 2002.
The classical actions of estrogen are mediated via interaction with nuclear steroid receptors leading to transcriptional regulation of various estrogen-responsive genes. However, recent evidence points to alternative pathways that mediate acute estrogen responses via cell-surface estrogen receptors (ERs).1–4 An important example is the rapid activation of endothelial nitric oxide synthase (eNOS)1,5 via a pathway involving mitogen-activated protein kinase and PI-3 kinase.5,6 However, the exact identity of the receptors involved in these nonclassical estrogen responses remains unclear.
Some evidence suggests that a subpopulation of ERα is localized within caveolae and can mediate acute eNOS activation.1 However, acute responses to estrogen in cell lines not expressing the classical ERα (hERα-66)7 and maintained vascular responses to estrogen in ERα,8 ERβ,9 and double ER knockout mice10 have led investigators to consider the possibility of other cell-surface ERs. Alternative ERα proteins encoded by mRNA splice variants are potential candidates for mediation of rapid estrogen responses, because there is no evidence for genes encoding ER homologues other than ERβ.
Human ERα mRNA isoforms are generated by splicing of 5 alternative upstream exons (B through F) to a common acceptor site 70 nucleotides (at +16311,12) upstream of the translation start site in exon 1. This alternative splicing produces identical-length proteins but raises the potential for differential, and hence tissue-specific, transcriptional regulation. In addition, an exon-1 truncated transcript (designated Δ1a-hERα-46) has recently been demonstrated in a breast carcinoma cell line, resulting from splicing of the 5′UTR variant exon E or exon F directly to exon 2.13 In contrast to its full-length counterpart (hERα-66), recombinant Δ1a-hERα-46 has a predicted molecular weight of 46 kDa and lacks a transactivation domain, AF-1. This truncated ER acts as a competitive inhibitor of hERα-66–mediated transactivation in some cell contexts.13 Furthermore, an ERα-like protein of uncertain molecular identity, but similar size (45 kDa), was recently implicated in mediating acute estrogen responses in the EA.926 human endothelial cell line.7 We and others14 hypothesized that this protein could be translated from the Δ1a-hERα-46 mRNA,13 suggesting a possible role for alternatively spliced ERα transcripts in mediating acute estrogen responses.
Accordingly, we sought to characterize the roles of the putative Δ1a-hERα-46 and classical 66-kDa ERα isoforms in mediating acute estrogen responses in human endothelial cells. In particular, we aimed to investigate the relative expression of these transcripts, their cellular localization, and their role in the acute activation of eNOS by estrogen in human endothelial cells.
Cell lines EA.926, hMEC, COS-1, and MCF-7 were propagated according to standard protocols. Human umbilical vein endothelial cells (HUVECs) were isolated from single donors and used at passages 2 through 4. For estrogen starvation, cells were incubated in medium without phenol red or serum.
Reverse Transcriptase–Polymerase Chain Reaction
RNA was obtained from human endothelial-derived cell lines and HUVECs with the use of Trizol reagent (Life Technologies). MCF-7 cells were used as a positive control. Reverse transcriptase (RT) was carried out by random priming with 1 μg of DNAase-treated RNA (Promega). Polymerase chain reaction (PCR) was performed with the use of forward primers based in 5′ exons, as follows: 1A-F: 5′-GGAGCTGGCGGGGGGCGTTG-3′; 1BF: 5′- CGCGTTTAT-TTTAAGCCCAG- 3′; 1CF: 5′- CGGCCCTTGACTTCTACAAG-3′; 1DF: 5′-CTTCTTCACCTGAGAGAGCC-3′; 1EF: 5′ CAG-AGAAATAATCGCAGAGC-3′; 1FF: 5′-CCAAAACTGA-AAATGCAGGC-3′. The reverse primer, located in exon 2, as follows, was common to all reactions: 2R: 5′-CCTTGCAG-CCCTCACAGGAC-3′.
The Δ1a-hERα-46 pSG5 expression plasmid was generously provided by the Gannon laboratory (European Molecular Biology Laboratory, Heidelberg, Germany).13 hERα-66 pSG5 expression plasmid was created by subcloning the 5′ coding sequence of ERα (exons 1 through 4) from pEGFP-C1 hER (a gift from the Mancini laboratory, Baylor College of Medicine, Houston, Tex15) into pSG5-ER46 with the use of EcoRI and the HindIII site in exon 4.
The fusion constructs Δ1a-hERα-46-GFP and ERα-66-GFP were created by cloning from pSG5 into pEGFPN1 (Clontech Laboratories, Inc), with PCR used to generate an in-frame GFP fusion cDNA. The luciferase reporter gene (ERE)2-tk-Luc was a gift from the Scanlan laboratory (University of California, San Francisco).16 Plasmids were transfected into cells with Fugene 6 (Roche).
Immunoblotting and Fluorescent Confocal Microscopy of ERα-Related Proteins
Immunoblotting and immunofluorescence studies of ERα-related protein were performed with monoclonal antibodies to ERα C-terminal (F-10, Santa Cruz Biotechnology). eNOS immunostaining (Transduction laboratories) was performed on COS cells cotransfected with an eNOS expression plasmid and GFP-tagged ERα isoforms to examine the spatial relationships between eNOS and ERα. Bound primary antibodies were visualized with TRITC-conjugated goat anti-mouse secondary antibodies (Sigma). Cells were imaged with a Biorad MRC 1024 confocal laser-scanning microscope.
Determination of E2-Surface Binding With E2-Conjugated BSA-FITC
Cells were transiently transfected with Δ1a-hERα-46, hERα-66, or β-gal control plasmid. After 48 hours, cells were fixed and incubated in E2-conjugated BSA-FITC (Sigma) equating to ≈30 nmol/L of 17β-estradiol. Unconjugated BSA-FITC was used to examine for nonspecific binding. Cells were mounted in Vectorshield (Vector Laboratories) containing DAPI and TO-PRO-3 and visualized by confocal microscopy.
Determination of Estrogen-Induced eNOS Activation
To visualize NO production, we loaded cells with diacetylated DAF-2 (DAF-2DA: Calbiochem). This membrane-permeable dye is hydrolyzed intracellularly by cytosolic esterases releasing DAF-2, which is converted in the presence of NO into a fluorescent product, DAF-2 triazole.17 Forty-eight hours after cotransfection with Δ1a-hERα-46; ERα-66 or control; and eNOS or control plasmids, COS cells were incubated in 1 μmol/L DAF-2DA solution for 30 minutes at 37°C in the dark and then washed. Cells were stimulated for 15 minutes in 17β-estradiol (30 nmol/L) in the presence or absence of L-NMMA and then washed and fixed before examination under the confocal microscope.
Estrogen-Dependent Transactivation in the Presence of Δ1a-hERα-46
The (ERE)2-tk-luc reporter gene construct was used to assess the effect of Δ1a-hERα-46 expression on the transcriptional activity of hERα-66 in the human endothelial cell–derived hMEC and EA.926 cell lines and in COS cells. Cells were grown in 24-well plates and transfected at 24 hours with a total of 1 μg of DNA per well (0.25 μg of (ERE)2-tk-luc; 0.25 μg of Δ1a-hERα-46 expression vector or control expression vector; 0.25 μg of hERα-66 expression vector or control expression vector; 0.25 μg of internal control β-galactosidase expression vector). Cells were stimulated in 30 nmol/L 17β-estradiol for 24 hours and then assayed for luciferase and β-galactosidase activity. Reporter gene activity was normalized for transfection efficiency according to the activity of the cotransfected reference control.
Δ1a-hERα-46 Is Expressed in Human Endothelial Cells
We first used RT-PCR to investigate the expression of ERα transcripts in human endothelial cells. Both truncated (Δ1a-hERα-46) and full-length ERα transcripts were present in human endothelial cells, including the EA.926 cell line, the hMEC microvascular endothelial cell line, and primary HUVEC cultures (Figure 1). In these cells, hERα transcripts were derived exclusively from exons 1E and 1F. The smaller RT-PCR product observed in human endothelial cells was confirmed by sequencing to result from alternative mRNA splicing from 5′ untranslated exons 1E or 1F directly to exon 2 (Figure 1) in a manner previously described for the truncated Δ1a-hERα-46 transcript in MCF-7 cells.13
Next, we sought to investigate whether appropriately sized ERα proteins corresponding to these hERα transcripts were present in endothelial cells. Western blots with an ERα C-terminal antibody revealed only a 46-kDa ERα protein in EA.926 cells in contrast to both 46- and 66-kDa ERα proteins in MCF-7 cells (Figure 2). This 46-kDa protein appeared of identical molecular weight to recombinant Δ1a-hERα-46 expressed in COS cells after transient transfection of the respective cDNA (Figure 2). Although some hERα-66 expression was detected by RT-PCR, repeated Western blots did not detect any ERα protein corresponding to the 66-kDa full-length isoform, consistent with observations by the Bender laboratory.7
Subcellular Localization of hERα46- and ERα66-GFP Fusion Proteins
We used ER-GFP constructs to investigate the subcellular localization of the 2 ERα isoforms in transfected COS cells. The GFP was tagged to the C terminal, ensuring a similar position relative to both ERα isoforms. Furthermore, the functional ability of the ERα66-GFP fusion protein to mediate estrogen-dependent transactivation of a target sequence was confirmed before additional investigation. Most hERα46- and hERα-66-GFP fluorescence was observed within the nucleus. However, significant proportions of both recombinant hERα-46 and hERα-66 were observed outside the nucleus and in association with the plasma membrane (Figure 3). Similar findings were observed in EA.926 cell and hMECs (data not shown). Nonpermeabilized, transfected cells immunostained for ERα with C terminal antibody suggested cell-surface localization of a proportion of the receptor, as well as C terminal integrity of the ERα component of the fusion proteins. Nontransfected COS cells or cells transfected with GFP-control vector were not recognized by the ERα antibody.
We determined whether ERα-GFP fusion proteins localized in the same way as ERα itself. The distribution of fluorescence was similar to that observed in nontransfected human endothelial cell line EA.926 when stained with primary anti-ERα antibody (data not shown), arguing against the observed distribution being a consequence of transfection or overexpression. The complete nuclear localization of hERα-66 after prolonged exposure to physiological levels of estrogen additionally suggested that the observed extranuclear localization of the ERα isoforms was not a result of the GFP disrupting nuclear localization signals (data not shown).
hERα-46 Associates With eNOS in Human Endothelial Cells
To investigate the potential role of hERα-46 in eNOS activation, we next investigated the association between hERα-46 and eNOS proteins with the use of fluorescent confocal microscopy. In cells incubated without estrogen, a proportion of both hERα-46 and hERα-66 was observed to colocalize with eNOS (Figure 4). Colocalization with eNOS remained after 10 minutes of stimulation with 30 nmol/L 17β-estradiol, but was lost after 24 hours of stimulation because of redistribution of ERα to the nucleus (data not shown). These results demonstrate that hERα-46 associates with eNOS in human endothelial cells in a similar manner to hERα-66.
Cells Expressing ERα-46 and ERα-66 Bind Estrogen at the Cell Surface
Because we observed recombinant Δ1a-hERα-46 at the cell membrane, we sought to investigate the ability of this protein to function as an estrogen receptor at this site. The cell-impermeable BSA-FITC tagged estradiol (E2coBSA-FITC: Sigma) was used to investigate estrogen binding at the surface of nonpermeabilized cells transfected with Δ1a-hERα-46, hERα-66, or control (β-gal) expression vectors. When transfected with the β-gal control vector, nonpermeabilized cells did not bind the cell-impermeable E2coBSA-FITC compound. However, cells transfected with Δ1a-hERα-46 or hERα-66 expression vectors bound E2coBSA-FITC at the cell surface at concentrations equivalent to 30 nmol/L 17β-estradiol (Figure 5). Preincubation with unlabeled estradiol (3 μmol/L) abolished E2coBSA-binding, and the FITC-BSA control compound did not bind, showing ligand specificity. These results suggest that hERα-46 functions as a specific estrogen-binding protein at the cell surface.
ERα-46 Mediates Estrogen-Induced eNOS Activation
Having observed estrogen binding by hERα-46 at the cell membrane and colocalization with eNOS, we next investigated its role in eNOS activation. Estrogen-induced eNOS activation was assessed in COS cells loaded with the NO-sensitive dye DAF-2DA. Confocal microscopy revealed that expression of Δ1a-hERα-46 or hERα-66 and eNOS resulted in estrogen-induced NO production not observed in the presence of either the ERα isoforms or eNOS alone (Figure 6). NO production was antagonized by the eNOS inhibitor L-NMMA. The NO-sensitive DAF-fluorescence was observed in a punctate pattern both in association with the membrane and in the cytoplasm. These data indicate that estrogen-induced NO production in endothelial cells is mediated by functional associations between hERα-46 or hERα-66 and eNOS.
ERα-46 and ERα-66 Mediate Differential Transcriptional Responses to Estrogen in Human Endothelial Cells
Because both hERα-46 and hERα-66 mediate acute eNOS activation, we next compared their role in mediating classical, transcriptional responses after longer estrogen administration. Prolonged estradiol treatment at physiological levels (30 nmol/L for 24 hours) resulted in loss of membrane-associated GFP fluorescence and loss of GFP-eNOS colocalization (data not shown). In transient transfection experiments with the (ERE)2-tk-luc reporter gene construct, hERα-66 mediated a marked transcriptional response to estrogen in hMECs, EA.926, and COS cell lines (Figure 7). In contrast, Δ1a-hERα-46 mediated only a fraction of this reporter gene expression in these cell lines (P<0.001; n=3). Furthermore, coexpression of Δ1a-hERα-46 and hERα-66 reduced estrogen-induced transactivation compared with that mediated by hERα-66 alone (P<0.01; n=3). These observations demonstrate that hERα-46 is a weak transcriptional activator compared with hERα-66 and acts as a competitive inhibitor of hERα-66–mediated transcriptional responses in these cells.
This study demonstrates expression of an exon 1–truncated hERα isoform, hERα-46, in human endothelial cells. This truncated ERα localizes to the cell membrane, specifically binds E2, and mediates acute estrogen-induced eNOS activation. Because hERα-46 mediates minimal estrogen-induced transcriptional activation and acts to competitively inhibit transcriptional activation by hERα-66, these findings provide new evidence for differential receptor-mediated mechanisms underlying acute and transcriptional estrogen signaling in human endothelial cells.
Despite the importance of estrogen-mediated effects in the vasculature, the identity of estrogen receptors that mediate acute effects of estrogen in endothelial cells has remained unclear. The classical ERα-66 copurifies with and activates eNOS in isolated caveolae from ovine pulmonary arterial endothelial cells.1 However, our findings now suggest an important role for a novel splice variant of ERα as an additional mediator of acute estrogen responses in human endothelial cells. Immunoidentification confirmed the predominance of the 46-kDa protein in EA.926 cells, the identical molecular weight of the recombinant Δ1a-hERα-46. Primary HUVECs also express a 46-kDa ERα variant.7
Use of hERα-46 and hERα-66 GFP fusion proteins enabled us to determine the localization of these hERα isoforms in living cells in relation to eNOS. The peripheral and membrane-associated localization of hERα-46 and the partial colocalization with eNOS demonstrated by the hERα46-GFP fusion protein was indistinguishable from that of hERα-66, previously shown by ultracentrifugation and immunofluorescence studies to localize to plasma membrane caveolae.1
We used a non–GFP tagged Δ1a-hERα-46 expression vector to examine for the role of this truncated ERα isoform in acute and transcriptional responses to estrogen. Nontransfected COS cells posses no ERα immunoreactivity and no transcriptional or acute response to estrogen. However, COS cells expressing Δ1a-hERα-46 bound estrogen at the cell surface and demonstrated acute eNOS activation in response to estrogen in a similar manner to those expressing ERα-66. Furthermore, these findings clarify previous observations in EA.926 cells, in which only a 45-kDa ERα-like protein was detected, despite intact acute responses to estrogen.7 Our demonstration of Δ1a-hERα-46 expression in several other human endothelial cell types now suggests that the exon 1 truncated ERα isoform plays important roles in endothelial estrogen signaling.
Specific binding of BSA-estradiol conjugates to the surface of cells expressing ERα-66 or Δ1a-hERα-46 is additional evidence supporting the ability of both isoforms to function as estrogen receptors in the mediation of acute, surface-related effects. A concern with the use of BSA-estradiol compounds is the possibility of incomplete conjugation that allows binding of free estradiol.18 However, this is only a limitation in studies addressing the acute physiological effects of BSA-estradiol administration. In our binding studies, any unconjugated estrogen would decrease the FITC signal seen at the cell surface by direct competition, as shown by addition of excess unconjugated estradiol. We used multiple additional controls, including BSA-FITC, together suggesting that membrane binding in cells expressing ERα isoforms is estrogen-specific.
The expression of hERα in endothelial cells has important implications for acute versus transcriptional effects of estrogen signaling. Because hERα-46 is only a weak mediator of transcriptional responses to estrogen and acts to inhibit the strong transcriptional activation mediated by hERα66, the predominant expression of hERα46 in endothelial cells would shift the balance of estrogen signaling toward acute rather than transcriptional responses. Differential expression of alternatively spliced ERα variants may therefore provide a novel mechanism for modulation of estrogen responses in the vascular endothelium.
Our identification of ERα46 as an important mediator of acute estrogen effects in human endothelial cells highlights the importance of recent findings in ERα knockout mouse models. Alternative ERα transcripts in the mouse, corresponding to human Δ1a-hERα-46, may explain residual estrogen responsiveness in ERα knockout mice (αERKO), in which only the first coding exon was targeted for disruption.19 In these mice, a short ER transcript is expressed that encodes an N-terminal truncated, mutant ER that binds estrogen but possesses significantly reduced estrogen-dependent transcriptional activity compared with that of the wild-type ER,20 resulting in the same functional domain changes as our findings with ERα-46 in human endothelial cells. Indeed, the vascular responses to estrogen that are preserved in these αERKO mice8 are abolished in a new knockout mouse with total deletion of ERα gene.21–23 Specifically, expression of alternatively spliced transcripts lacking the AF-1 domain in αERKO mice seems to mediate estrogen-induced endothelial NO production that is absent in the complete knockout.24 These insights from mouse models suggest that mouse homologs of Δ1a-hERα-46 are sufficient to mediate the acute endothelial effects of estrogen and in principle support our observations of the functional effects of hERα-46 in human endothelial cells, although the 46-kDa truncated ERα protein was not readily detected in mouse aortic lysates.24 Additional studies are required to investigate possible roles for truncated ERs in mediating acute and transcriptional effects of estrogen in human pathophysiology.
In conclusion, this study identifies a novel, exon 1–spliced ERα in human endothelial cells. The exon 1 truncated ERα is similar to ERα-66 in its subcellular localization and ability to mediate estrogen binding at the surface of the membrane in a ligand-specific fashion, leading to estrogen-induced eNOS activation. However, its inability to mediate transcriptional responses in endothelial cells suggests a unique role of this truncated ER in mediating differential acute versus transcriptional responses to estrogen in the vascular endothelium.
This work was supported by grants from the British Heart Foundation and from the Wellcome Trust. Dr Figtree was supported by the Rhodes Trust and by Trinity College, University of Oxford.
Stefano GB, Prevot V, Beauvillain JC, et al. Cell-surface estrogen receptors mediate calcium-dependent nitric oxide release in human endothelia. Circulation. 2000; 101: 1594–1597.
Haynes MP, Sinha D, Russell KS, et al. Membrane estrogen receptor engagement activates endothelial nitric oxide synthase via the PI3-kinase-Akt pathway in human endothelial cells. Circ Res. 2000; 87: 677–682.
Russell KS, Haynes MP, Sinha D, et al. Human vascular endothelial cells contain membrane binding sites for estradiol, which mediate rapid intracellular signaling. Proc Natl Acad Sci U S A. 2000; 97: 5930–5935.
Karas RH, Hodgin JB, Kwoun M, et al. Estrogen inhibits the vascular injury response in estrogen receptor beta-deficient female mice. Proc Natl Acad Sci U S A. 1999; 96: 15133–15136.
Karas RH, Schulten H, Pare G, et al. Effects of estrogen on the vascular injury response in estrogen receptor α, β (double) knockout mice. Circ Res. 2001; 89: 534–539.
Flouriot G, Brand H, Denger S, et al. Identification of a new isoform of the human estrogen receptor-α (hER-α) that is encoded by distinct transcripts and that is able to repress hER-α activation function 1. EMBO J. 2000; 19: 4688–4700.
Mendelsohn ME. Nongenomic, ER-mediated activation of endothelial nitric oxide synthase: how does it work? What does it mean? Circ Res. 2000; 87: 956–960.
Paech K, Webb P, Kuiper GG, et al. Differential ligand activation of estrogen receptors ERα and ERβ at AP1 sites. Science. 1997; 277: 1508–1510.
Lubahn DB, Moyer JS, Golding TS, et al. Alteration of reproductive function but not prenatal sexual development after insertional disruption of the mouse estrogen receptor gene. Proc Natl Acad Sci U S A. 1993; 90: 11162–11166.
Dupont S, Krust A, Gansmuller A, et al. Effect of single and compound knockouts of estrogen receptors α (ERα) and β (ERβ) on mouse reproductive phenotypes. Development. 2000; 127: 4277–4291.
Pare G, Krust A, Karas RH, et al. Estrogen receptor-α mediates the protective effects of estrogen against vascular injury. Circ Res. 2002; 90: 1087–1092.
Brouchet L, Krust A, Dupont S, et al. Estradiol accelerates reendothelialization in mouse carotid artery through estrogen receptor-α but not estrogen receptor-β. Circulation. 2001; 103: 423–428.
Pendaries C, Darblade B, Rochaix P, et al. The AF-1 activation-function of ERα may be dispensable to mediate the effect of estradiol on endothelial NO production in mice. Proc Natl Acad Sci U S A. 2002; 99: 2205–2210.