Estradiol Decreases IGF-1 and IGF-1 Receptor Expression in Rat Aortic Smooth Muscle Cells MECHANISMS FOR ITS ATHEROPROTECTIVE EFFECTS*

Insulin-like growth factor (IGF-1) is a potent mitogen for vascular smooth muscle cells. Both IGF-1 and its receptor have been shown to be highly expressed in atherosclerotic lesions. Here we investigated whether part of the vasculoprotective properties of E 2 may be mediated by its negative regulation of the IGF-1 system. HeLa cells, which do not contain endogenous estrogen receptors (ER), were transiently transfected with IGF-1R promoter constructs with or without a plasmid encoding human ER a or ER b and treated with 100 n M 17 b -estradiol (E 2 ) for 24 h. E 2 treatment decreased basal luciferase activity by 51%, and this effect was dependent on co-expression of ER a , whereas no repression was observed with ER b . A mutation within the DNA binding domain of the ER a abolished the repressor function of the ER receptor. Similarly, E 2 decreased IGF-1R tran- scription by 21% in rat aortic smooth muscle cells (RASMC), which express endogenous ER. This effect was specific for E 2 , because it was inhibited by an an- tiestrogen and because progesterone did not have any effect on IGF-1R expression in HeLa or RASMC transfected with progesterone receptor. Accordingly, E 2 de- creased IGF-1R and IGF-1 mRNA in RASMC by 47% and 33%. Western blot analysis and radioligand binding studies showed that E 2 also Blot Lysates to SDS-PAGE on 7.5% gels, and separated proteins transferred to polyvinylidene difluoride membranes. Blots blocked 5% incubated anti-IGF-1R b and secondary peroxidase-conjugated anti-rabbit an- tibody. Immunopositive bands were visualized by enhanced chemiluminescence. IGF-1 Radioimmunoassay— Specific IGF-1 immunoreactivity of cell-conditioned medium was determined as cell was dialyzed, and chromatographed using Bio-Gel P-30 polyacrylamide columns IGF-1 fractions were assayed using a poly-clonal anti-IGF-1 rabbit antiserum (kindly Standard curves were generated using human recombinant IGF-1. Thymidine Incorporation— RASMC were plated in 24-well plates in DMEM without Phenol Red alone or containing DCT FBS. After 48 h cells were treated with or without E 2 , for 24 h in complete medium. the las t 2 of theincubation three times min two times and lysed were measured by liquid scintillation spectrophotometry. All experi- ments were performed in quadruplicates.

Insulin-like growth factor (IGF-1) is a potent mitogen for vascular smooth muscle cells. Both IGF-1 and its receptor have been shown to be highly expressed in atherosclerotic lesions. Here we investigated whether part of the vasculoprotective properties of E 2 may be mediated by its negative regulation of the IGF-1 system. HeLa cells, which do not contain endogenous estrogen receptors (ER), were transiently transfected with IGF-1R promoter constructs with or without a plasmid encoding human ER␣ or ER␤ and treated with 100 nM 17␤-estradiol (E 2 ) for 24 h. E 2 treatment decreased basal luciferase activity by 51%, and this effect was dependent on co-expression of ER␣, whereas no repression was observed with ER␤. A mutation within the DNA binding domain of the ER␣ abolished the repressor function of the ER receptor. Similarly, E 2 decreased IGF-1R transcription by 21% in rat aortic smooth muscle cells (RASMC), which express endogenous ER. This effect was specific for E 2 , because it was inhibited by an antiestrogen and because progesterone did not have any effect on IGF-1R expression in HeLa or RASMC transfected with progesterone receptor. Accordingly, E 2 decreased IGF-1R and IGF-1 mRNA in RASMC by 47% and 33%. Western blot analysis and radioligand binding studies showed that E 2 also dose-dependently decreased IGF-1R protein expression in RASMC by 40% and 30%, respectively, and that IGF-1 protein was reduced by 43%. Repression of IGF-1R promoter activity by a combination of ER␣ and E 2 did not appear to be mediated via direct binding of ER to the IGF-1R promoter but rather by inhibition of SP1 binding to the IGF-1R promoter. Thus, E 2 down-regulates IGF-1R and IGF-1 expression in vascular smooth muscle cells. This may have important implications for the understanding of the beneficial effects of estrogen in the cardiovascular system.
Several studies, but not all, have suggested that estrogens are cardioprotective in postmenopausal women (1,2). The mechanisms of estradiol-induced reduction in the risk of coro-nary artery disease remain unclear. Although these atheroprotective effects of estrogen were principally attributed to the hormone's effects on serum lipid concentrations (3)(4)(5), recent findings suggest that the majority of the vasculoprotective effects of estrogen are due to direct effects on the vasculature (6). Direct effects of estrogens have been demonstrated in vitro and in vivo both in animal and human models. These include effects on gene expression (7,8), ion channel function (9,10), response to vasoactive substances (11)(12)(13)(14), as well as vascular smooth muscle cell proliferation and migration (14,15).
The possible involvement of insulin-like growth factor-1 (IGF-1) 1 and IGF-1 receptor (IGF-1R) in cardiovascular pathology has recently raised interest. In vitro data have shown that IGF-1 is a potent vascular smooth muscle cell (VSMC) mitogen (16,17), and several reports have documented that VSMCs express IGF-1 and its receptor (18 -20). We and others have shown that several growth factors up-regulate IGF-1R on VSMC and this ability of growth factors to increase the number of IGF-1R is likely critical for their mitogenic effects (17,(21)(22)(23)(24). Furthermore, regulation of IGF-1 binding proteins (IG-FBPs) by growth factors may be physiologically important (25).
Steroid receptors, including the estrogen receptors (ER) ␣ and ␤, mediate the specific response of cells to their respective ligands by virtue of their ability to bind cis-acting regulatory sequences termed steroid response elements (for review see Ref. 26). Although much is known about mechanisms of gene activation by ER, less information exists about repression of gene expression by ER. Although activation of genes by estrogens is typically mediated by binding of the activated receptor to the respective response element(s) present upstream of or within target genes, negative regulation by these hormones cannot always be explained by receptor-DNA interaction (27,28). To our knowledge, the inhibition of IL-6 in HeLa cells, lipoprotein lipase in 3T3-L1 cells, tumor necrosis factor ␣ in U937 cells, IGF-1 gene expression in primary rat osteoblasts, and the mannose-6 IGF-II receptor gene in breast cancer cells by estrogens are the only documented examples of repression by estrogens (29 -33). It was therefore of interest to us to explore the effects of estradiol on IGF-1R and IGF-1 expression in vascular cells such as RASMC and to determine the molecular mechanisms of ER-mediated action on IGF-1R gene expression.
We show that E 2 dose-dependently decreases IGF-1R and IGF-1 expression and that the antiproliferative activity of E 2 involves a down-regulation of IGF-1R and IGF-1. However, we found no direct binding of ER to sequences in the IGF-1R promoter that were sufficient to confer repression by ER in functional experiments. Nevertheless, results obtained from bandshift experiments indicated that there was an interaction between SP1 and ER, because ER decreased SP1 binding to the IGF-1R promoter. These data indicate that ER can modulate transcription from promoters that lack classical estrogen response elements (ERE) and have important implications for understanding cardiovascular effects of estrogens.
Cell Culture-RASMC (kindly provided by Dr. K. Griendling, Emory University, Atlanta, GA) were grown in DMEM supplemented with 10% heat-inactivated calf serum, 2 mM glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin and incubated at 37°C in a humidified 5% CO 2 atmosphere. HeLa cells were cultured under similar conditions with 5% fetal bovine serum. The ER-expressing, human breast tumor cells, MCF-7 (34), were maintained in DMEM and 10% fetal bovine serum. Prior to experiments, cell media were changed to DMEM without Phenol Red, containing dextran-coated, charcoal-treated, heat-inactivated (DCT) fetal bovine serum.
Plasmids and Transfection-The full-length promoter of the IGF-1R p(Ϫ2350/ϩ640-Luc) and the shorter construct p(Ϫ476/ϩ640-Luc) were a generous gift from Dr. H. Werner (National Institutes of Health, Bethesda, MD). Deletion fragments were made from the full-length promoter construct and subcloned upstream of the firefly luciferase cDNA (35). The plasmids encoding human ER␣ (HEG0), HE82, human ER␤ (pCMV-hER␤), and pCMV-SP1 were kind gifts from Dr. P. Chambon (Strasbourg, France), Dr. S. Mader (Montreal, Canada), Dr. J.-Å. Gustafsson (Huddinge, Sweden), and Dr. R. Tjian (University of California at Berkeley, CA), respectively. The following constructs have previously been described: the empty vector pSG5 (36), and XETL (37). In brief, HEG0 carries human wild-type ER␣ cDNA, HE82 contains human ER␣ cDNA with a mutation in the DNA binding domain resulting in the recognition of a glucocorticoid response element instead of an ERE (38), and the reporter plasmid XETL expresses firefly luciferase under control of one vitellogenin A2 ERE upstream of the herpes simplex virus thymidine kinase promoter. The coding sequence of the human progesterone receptor B (PRB) was subcloned into the unique EcoRI site of pSG5 resulting in pSG5/hPR. HeLa cells were plated in 24-well and RASMC in 12-well plates and transfected with 1 g of reporter plasmid and 5 ng of pRL-TK per well with or without 200 ng of HEG0, pSG5/hPR, HE82, pCMV-hER␤, or pCMV-SP1 with LipofectAMINE reagent. 20 h after transfection, the DNA-containing medium was changed and the cells were treated with or without E 2 (100 nM) or progesterone (100 nM) for 12-24 h. In some experiments transfected cells were incubated with OHT (1 M) or ICI (0.1 M) for 1 h prior to the addition of E 2 . Luciferase activity was measured with the Dual-Luciferase kit according to the manufacturer's recommendations. Firefly luciferase activity was normalized to the internal control Renilla luciferase (Luc/Ren).
RNase Protection Assays-RNase protection assays were performed as described previously (18). In brief, 20 g of total RNA was hybridized with [ 32 P]UTP-labeled antisense IGF-1R and IGF-1 riboprobe and cohybridized with an 18 S probe (Ambion, Austin, TX). After overnight hybridization at 42°C and RNase digestion, samples were proteinase K-treated, phenol-extracted, and analyzed by 6% polyacrylamide/8 M urea denaturing gel electrophoresis. Densitometric analyses were per-formed using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
Radioligand Binding Studies-Radioligand binding assays were performed as described previously (39). Briefly, RASMC cultured under DCT serum conditions treated with or without E 2 in 24-well plates were incubated with 0.1 nM 125 I-IGF-1 and 0 -0.1 M unlabeled IGF-1 for 90 min at room temperature. Cells were washed in ice-cold binding buffer and solubilized in 0.2 N NaOH before counting. All assays were performed in duplicate for each experimental point. Data were analyzed using the LIGAND program.
Western Blot Analysis-Prior to the experiments, cultured RASMC were switched to Phenol Red-free DMEM containing DCT FBS for 48 h before adding E 2 at various concentrations for 24 h. Cells were washed in ice-cold phosphate-buffered saline and lysed as previously published (35). Lysates were subjected to SDS-PAGE on 7.5% gels, and separated proteins were transferred to polyvinylidene difluoride membranes. Blots were blocked with 5% dry milk and incubated with anti-IGF-1R␤ antibody and secondary peroxidase-conjugated donkey anti-rabbit antibody. Immunopositive bands were visualized by enhanced chemiluminescence.
Thymidine Incorporation-RASMC were plated in 24-well plates in DMEM without Phenol Red alone or containing DCT FBS. After 48 h cells were treated with or without E 2 , for 24 h in complete medium. 1 Ci/ml of [ 3 H]thymidine was added during the last 2 h of the incubation period. Cells were washed three times with ice-cold phosphate-buffered saline, incubated for 30 min in 10% trichloroacetic acid on ice, washed two times in ice-cold 95% ethanol, and lysed in 0.2 N NaOH. Samples were measured by liquid scintillation spectrophotometry. All experiments were performed in quadruplicates.
Electrophoretic Mobility Shift Assay (EMSA)-The human ER, PRB, and HE82 proteins were synthesized in vitro in TNT-T7-coupled rabbit reticulocyte lysates. Nuclear extracts from untransfected MCF-7 and RASMC, or HeLa cells transfected with HEG0, were incubated with recombinant human SP1, NFB, or ER␣ proteins in binding buffer containing 10 mM HEPES, pH 7.9, 10% glycerol, 100 mM KCl, 2 mM dithiothreitol, 0.1 mM EDTA, 5 mM MgCl 2 , 2 g of poly(dI-dC), 0.3 g/l bovine serum albumin, and 32 P-labeled DNA in a final volume of 20 l at room temperature. Preincubations containing ligand, antibody, and/or cold competitor (200-fold excess) as indicated were performed at room temperature for 15 min. After the incubation step the probe was added and binding conducted for additional 20 min. Reaction mixtures were loaded onto a 6% PAGE gel in 0.5 ϫ Tris borate-EDTA (TBE). The following oligonucleotide and its complement were used as labeled probes and cold/unlabeled competitors: ERE, 5Ј-GATCTCTTTGAT-CAGGTCACTGTGACCTGACTTTG-3Ј. The probe for the IGF-1R promoter extended from nucleotides Ϫ476/ϩ21.
Statistics-All experiments were performed at least three times. Statistical significance was measured by Student's t test. A value of p Ͻ 0.05 was considered statistically significant.

Effect of E 2 on IGF-1R Promoter Activity in HeLa or Rat
Aortic Smooth Muscle Cells-To measure the effect of E 2 on IGF-1R gene expression, HeLa cells, which do not contain endogenous ER, were transiently transfected with the fulllength IGF-1R promoter reporter construct with or without cotransfecting HEG0, a plasmid encoding human ER␣. In HeLa cells estradiol treatment (100 nM for 24 h) decreased basal luciferase activity to 49 Ϯ 5% (Fig. 1A). This effect was ER-dependent, because E 2 did not reduce basal IGF-1R expression in HeLa cells when human ER was not cotransfected. Similarly, E 2 decreased by 21% IGF-1R transcription in RASMC-expressing endogenous ER (p ϭ 0.005) (Fig. 1B). Using specific primers for rat ER␣ and rat ER␤, we found both transcripts in RASMC (data not shown) as has been previously published by others (40 -43). The fact that E 2 stimulated transactivation of a minimal ERE promoter reporter construct in RASMC without transfecting HEG0 (data not shown) supports the notion that the endogenous ERs were functional as has been previously shown by others (41).
Accordingly, these effects of E 2 appeared to be specific, because progesterone did not have any effect on IGF-1R expression in HeLa, or RASMC transfected with progesterone receptor (data not shown). In addition, the E 2 antagonist ICI 164,384 reversed the reduction in IGF-1R promoter activity induced by E 2 , demonstrating a specific ER-mediated effect, whereas the partial antagonist OHT acted in a synergistic way by further decreasing luciferase activity (Fig. 1C).
Interestingly, the repression of IGF-1R promoter activity by E 2 was abrogated when smaller IGF-1R promoter deletion mutants were used. Indeed, the reduction in luciferase activity was maintained with the construct p(Ϫ476/ϩ21) and p(Ϫ416/ ϩ21), however, the reduction disappeared with p(Ϫ330/ϩ21), suggesting that the E 2 -responsive region was located 5Ј of base pair Ϫ330 in the IGF-1R promoter (data not shown). Importantly, HE82, an ER mutant carrying a mutation within the DNA binding domain and thus recognizing a glucocorticoid response element instead of an ERE, was unable to repress expression from the IGF-1R promoter, suggesting that an ER with an intact DNA binding domain is required (Fig. 1D).
Since the identification of a second ER subtype, termed the ER␤ (44,45), much research has been focused on the potentially distinct role of ER␣ and ER␤ in vasculoprotection. It was therefore of interest to determine whether the reduction in IGF-1R transcription by the combination of E 2 and ER␣ was subtype-specific or whether it could also be observed using ER␤. Most interestingly, E 2 did not repress IGF-1R transcription when HeLa cells were transfected with human ER␤, which suggests a ER␣ subtype-specific effect (Fig. 1D). E 2 Decreases IGF-1 and IGF-1R mRNA Levels-To confirm the results obtained in transfection studies, endogenous IGF-1 and IGF-1R mRNA levels were measured by RNase protection assay in RASMC treated with or without E 2 . In agreement with the transfection studies, E 2 significantly and dose-dependently reduced basal levels of IGF-1 and IGF-1R by 47% and 33%, respectively ( Fig. 2A), whereas OHT had similar effects as E 2 (Fig. 2B). Similarly to the transcriptional assays, the antiestrogen ICI reversed the decreasing effect of E 2 on IGF-1 and IGF-1R (Fig. 2B) (Fig. 3A). Similarly, E 2 (100 nM) reduced basal IGF-1 binding sites by approximately 30% as measured by radioligand binding studies, further confirming the results seen in Western blots (per- centage change in IGF-1R number: control ϭ 100 Ϯ 0% and E 2 ϭ 71 Ϯ 7%, respectively, n ϭ 4). In addition, IGF-1 protein levels in RASMC were also significantly reduced by E 2 (43% reduction with 100 nM E 2 ) as measured by RIA of cell-conditioned medium (Fig. 3B).
Effect of Estradiol on Serum-induced DNA Synthesis-IGF-1 is a potent mitogen, and a functional IGF-1R is required for the mitogenic effects of various growth factors (22,24). To determine whether the reduced levels of IGF-1 and IGF-1R expression induced by E 2 could explain the reduced DNA synthesis observed after E 2 treatment (14, 46), we measured [ 3 H]thymidine incorporation in confluent RASMC. As shown in Table I, E 2 dose-dependently reduced DNA synthesis under serum conditions by approximately 50%. Exogenous addition of IGF-1 (50 -100 ng/ml), however, was not able to reverse the E 2 -induced decrease in thymidine incorporation (data not shown).
ER␣ Does Not Bind Directly to the IGF-1R Promoter-The IGF-1R promoter contains multiple SP1 sites in both the 5Јflanking and 5Ј-untranslated regions (47). Consensus EREs consist of an inverted repeat of the palindrome GGTCA separated by a 3-base pair spacer (27,48,49). However, no evident ERE is present within the IGF-1R promoter. In preliminary experiments, we tested our EMSA conditions by determining classical ER binding to its consensus response element and concurrent supershift with anti-ER antibodies (Fig. 4A). To investigate the binding of ER to the IGF-1R promoter sequences, the ER-regulated promoter construct p(Ϫ476/ϩ21) was used. Nuclear extracts from MCF-7 cells formed retarded bands with the 32 P-labeled IGF-1R probe, which could be supershifted by anti-ER␣ antibody (data not shown). However, neither in vitro synthesized or purified ER bound the IGF-1R promoter probe, whereas SP1 and NFB proteins formed a retarded band complex (data not shown). Interestingly, the intensity of the two main SP1-dependent bands was significantly reduced by co-incubation with in vitro synthesized ER or purified ER (Fig. 4B), whereas both ER preparations had no effect on the NFB-dependent band (data not shown). Although ER protein diminished the SP1/probe band, in vitro translated PRB or HE82, human ER␣ carrying a mutation in the DNA binding domain, had no effect, indicating not only an ERspecific effect but also an effect dependent on a conserved ER DNA binding domain (Fig. 4B).  Regulation of IGF-1 and IGF-1R by Estradiol IGF-1R transcription by E 2 may be due to ER-SP1 proteinprotein interaction. Indeed, transient overexpression of SP1 blunted the ER/E 2 -induced decrease in IGF-1R transcription using HeLa cells (Fig. 5).

DISCUSSION
Premenopausal women have less coronary artery disease than do men. However, the incidence of the disease rises markedly after menopause, and hormone replacement therapy may reduce the risks to premenopausal levels (55)(56)(57)(58). Until recently, the atheroprotective effects of estrogen were attributed principally to the hormone's effects on serum lipid concentrations. However, it is now evident that E 2 has other vasoprotective effects, such as increasing vasodilation via stimulation of nitric oxide synthase (59), decreasing plasma concentrations of renin and ACE (60), and decreasing vascular expression of the Ang II AT 1 receptor gene (61). These vascular effects of estradiol are likely to play an important role in the atheroprotective effects of the hormone. In addition our studies demonstrate that 17␤-estradiol modulates IGF-1 and IGF-1R mRNA and protein levels in RASMC. Although previous reports have shown that estrogen may alter expression of IGFBPs and IGF-1R mRNA and IGF-1 binding sites in human breast cancer cells (33,62,63), we provide the first evidence that estradiol inhibits IGF-1 and IGF-1R expression in vascular smooth muscle cells. Together with the inhibiting effect of estrogen on vascular smooth muscle cell proliferation, these findings suggest a possible mechanism for the observation that estradiol has antiatherogenic properties in vivo and in vitro.
Because IGF-1 is an important migration factor and mitogen for smooth muscle cells (64 -66) and because IGF-1 expression is increased after balloon injury (67) and in atherosclerotic lesions (68), it was of interest to us to study the effects of estradiol on IGF-1 and IGF-1R expression in vascular smooth muscle cells. Initially, we focused on transient transfection assays using IGF-1R promoter constructs (35). HeLa cells were initially chosen because of their lack of ER. Thus, these cells provided a useful model in which to study ER-independent and ER-dependent effects on the IGF-1R promoter. Our results show that E 2 diminished IGF-1R transcription via an ER-dependent pathway. Similar results were also found in COS-1 cells (data not shown). Not only was ER required for the repressor effect of E 2 on IGF-1R transcriptional activity, but the effect was also specific for E 2 , because the antiestrogen ICI 164,384 completely abrogated the repression and progesterone had no effect, in the presence of PRB, on IGF-1R transcription. Further evidence that a functional ER was necessary for the  4 and 5), or control IgG (lane 6). B, recombinant ER␣, in vitro translated HEG0, pSG5/hPR, or HE82 were allowed to interact with a 0.497-kilobase DNA fragment from the IGF-1R promoter in the presence or absence of SP1 protein and analyzed by electrophoresis on a 6% polyacrylamide gel. transrepression of E 2 on the IGF-1R promoter, was provided by studies using the mutant ER construct HE82. This mutant consist of a wild-type ER in which three amino acids in the first zinc finger have been replaced by the equivalent amino acids of the first zinc finger of the glucocorticoid receptor changing the DNA binding specificity of HE82 to that of the glucocorticoid receptor (38).
In agreement with the above described findings, E 2 reduced IGF-1R transcription in RASMC, which do express endogenous ER. Also E 2 dose-dependently decreased IGF-1R mRNA and protein when compared with control. We have previously shown that small changes in the number of IGF-1R have major effects on cell growth (69). Thus, this reduction in IGF-1R may be physiologically relevant and may explain the decrease in DNA synthesis observed.
Little information is available regarding the relationship between E 2 and IGF-1 in vascular smooth muscle cells. Most studies have focused on the reproductive organs and other estrogen-sensitive cells. In these cells E 2 is often related to an increase in IGF-1 and IGF-1R signaling by sensitizing cells to the mitogenic effects of IGF-1 as shown in breast cancer cells (62) and in vivo in the uterus (70,71). However, in RASMC we observed a reduction in IGF-1 mRNA and protein expression induced by E 2 . This is in agreement with reports where oral E 2 replacement therapy in postmenopausal women induced a marked decrease in serum IGF-1 levels (72,73). The observed decrease in IGF-1 mRNA is also in agreement with the study of McCarthy et al. (32), using primary rat osteoblasts, however, it is in contrast to the reports from Ernst and Rodan (74). Interestingly, OHT further potentiated the depressor effect of E 2 on IGF-1 and IGF-1R mRNA in RASMC and in transient transfection studies using HeLa cells, whereas ICI blunted the E 2 response. The finding that OHT similarly to E 2 decreased IGF-1 and IGF-1R mRNA expression is in good agreement with reports that OHT can act as a partial agonist of ER, depending on cell context (75). However, E 2 clearly down-regulated IGF-1 and IGF-1R protein expression, consistent with its effect on IGF-1 and IGF-1R mRNA levels. Because IGF-1 is a potent mitogen, the decrease in IGF-1 and IGF-1R could at least partially explain the inhibitory effect of E 2 on DNA synthesis. Thus, although exogenous IGF-1 failed to reverse the inhibitory effect of E 2 on DNA synthesis, this is quite possibly due to the persistent reduction in IGF-1R. Indeed, we have previously shown that IGF-1R density is a critical determinant of vascular smooth muscle cell growth responses (39).
Bioactivity of IGF-1 is modulated by several high affinity binding proteins (IGFBPs) present in the vasculature (76,77). These IGFBPs control the distribution of IGF-1 between extracellular and cellular compartments and can also alter IGF-1 bioactivity by modulating its interaction with its receptor (78). The major two IGFBPs found to be secreted by RASMC were IGFBP-2 and IGFBP-4 (25). However, E 2 had no effect on IGFBP secretion when compared with control, suggesting that the E 2 inhibitory response in DNA synthesis is rather related to its depressor effect on IGF-1 and IGF-1R expression than on the expression of inhibitory binding proteins (data not shown).
ER-mediated transactivation is a complex process regulated by ligand-dependent or ligand-independent mechanisms (reviewed in Ref. 79) and by interactions with coactivators and/or co-repressors, by binding directly to various DNA elements or by indirectly enhancing DNA binding via protein-protein interaction (80). In our studies, ER-dependent repression of the IGF-1R promoter, unlike transactivation of ERE-containing reporter and EMSA probe by the same combination, did not appear to be mediated via high affinity binding of the ER to the IGF-1R promoter probe, because no direct binding of purified ER alone to the IGF-1R promoter was observed. Instead, our results suggested that ER inhibited SP1 binding to the IGF-1R promoter. This would suggest that protein-protein interactions between ER and SP are responsible for the inhibitory action of ER. Indeed, the IGF-1R promoter contains putative consensus sequences for SP1 (81) but also other regulatory elements like Egr-1 (82), AP-2 (83), platelet-derived growth factor-responsive element (84). This is somewhat in contrast to the previous finding of Porter et al. (51), who described an enhancement of SP1 binding to SP1 consensus sequence by co-incubating with ER protein. However, this could likely be promoter-and cellspecific. To confirm the inhibitory effect of ER on SP1 binding to IGF-1R promoter, we transiently overexpressed SP1 and showed that this blocked the ability of E 2 to decrease IGF-1R transcription. In addition, an intact DNA binding domain of ER was required for the effects of SP1 binding to the IGF-1R promoter, even if direct binding of ER to the same promoter was not detected. The 3-amino acid mutation within the DNA binding domain of HE82 may thus be sufficient to prevent interaction with SP1 bound to the IGF-1R promoter.
Interactions between ligand-activated steroid hormone receptors and specific genomic sequences are well described and are believed to be the dominant mechanism whereby this class of hormones exerts its molecular effects (27). ERs preferentially bind to a 5-bp palindromic DNA sequence, the ERE. This element can be sufficient to enhance transcription, as can "imperfect" or half-EREs, which can be located a great distance from the transcription start site (85). Although traditionally thought to be enhancing, there are accumulating reports of ER negatively affecting transcription (29,86,87). As does ours, these reports speculate about ERE interactions. Ray et al. (86), studied the negative effect of E 2 on IL-1-mediated IL-6 gene activation. Although inhibition involved the ER, high affinity binding of ER to the IL-6 promoter could not be demonstrated nor did recombinant ER bind to the promoter fragment in gel mobility shift assays. Likewise, there was no binding observed between ER and the promoter 1 of the rat IGF-1 gene (32). This could be explained by an interaction between ligand-activated ER and other trans-activating factors, as has been described for the transcription factor AP1 (88 -90), SP1 (50 -52, 91), and NFkB (86,92,93). It appears that in the absence of obvious and typical EREs in the promoters of negatively regulated target genes, the ER may function as repressor by antagonizing the activity of positively regulating transcription factors without direct ER-DNA contacts (29,86).
Recently, a second ER, ER␤, has been discovered (44,45). Although ER␣ and ER␤ share a high degree of identity in their ligand binding and DNA binding domain, and although both have similar affinities for E 2 and recognize the same consensus ERE, they do respond differentially to partial agonist antiestrogens in transactivation assays (44,45,94,95). In addition, ER␤ has biological roles that are distinct from those of ER␣ as evidenced with the different phenotypes of the ␣ERKO and ␤ERKO mice (96,97). Expression of ER␣ has been reported to increase in rabbit cardiac allografts (98), whereas ER␤ is upregulated, but not ER␣, in rat endothelial cells after carotid artery injury (99). Our results showing that only ER␣ was able to repress IGF-1R transcription in HeLa cells, are among the first to show a differential effect of E 2 via ER␣ or ER␤. Indeed, Paech et al. (89), have shown a differential ligand activation of ER␣ and ER␤ at AP1 sites. However, at this point we cannot rule out that ER␣/ER␤ heterodimers are involved in the repression of IGF-1 or IGF-1R in RASMC, because ER␤ transcripts have been found in these cells (41,42,44).
In summary, our results demonstrate that E 2 inhibits IGF-1R and IGF-1 mRNA and protein expression in vascular smooth muscle cells. This decrease may explain the inhibitory effect of E 2 on DNA synthesis and its antiproliferative effects on vascular cells and may, thus, offer one mechanism by which estradiol retards atherosclerosis in premenopausal women.