17-Epiestriol, an Estrogen Metabolite, Is More Potent Than Estradiol in Inhibiting Vascular Cell Adhesion Molecule 1 (VCAM-1) mRNA Expression*

17-β estradiol (17-β E2) attenuates the expression of vascular cell adhesion molecule 1 (VCAM-1)in vivo at physiological levels (pg/ml), whereas supraphysiological concentrations of 17-β E2 (ng/ml) are required in vitro. We assessed whether a metabolite of estrogen, which could only be generated in vivo, might be a more potent inhibitor of VCAM-1 expression and thereby explain this discrepancy. We report here that 17-epiestriol, an estrogen metabolite and a selective estrogen receptor (ER) β agonist, is ∼400× more potent than 17-β E2 in suppressing tumor necrosis factor (TNF) α-induced VCAM-1 mRNA as well as protein expression in human umbilical vein endothelial cells. Genistein, an ERβ agonist, at low concentrations (1 and 10 nm) also suppressed TNFα-induced VCAM-1 mRNA expression. These actions of 17-epiestriol and genistein were significantly attenuated in the presence of the estrogen receptor antagonist ICI-182780. Other estrogenic compounds such as ethinyl estradiol and estrone did not have any effect on TNFα-induced VCAM-1 expression at the concentrations tested. We further show that, 1) 17-epiestriol induces the expression of endothelial nitric-oxide synthase mRNA and protein, 2) 17-epiestriol prevents TNFα-induced migration of NFκB into the nucleus, 3) NG-nitro-l-arginine methyl ester, an inhibitor of NO synthesis, abolishes 17-epiestriol-mediated inhibition of TNFα-induced VCAM-1 expression and migration of NFκB from the cytoplasm to the nucleus. Our results indicate that 17-epiestriol is more potent than 17-β E2 in suppressing TNFα-induced VCAM-1 expression and that this action is modulated at least in part through NO.

The mechanism by which estrogens attenuate the development of atherosclerosis is not known, although various actions of estrogens have been suggested to mediate this effect (1). We have previously demonstrated that 17-␤ estradiol (17-␤ E 2 ) in vivo inhibits the adhesion of monocytes to endothelial cells of the rabbit aorta (2). We also demonstrated that following a cholesterol-enriched diet to ovariectomized rabbits, expression of VCAM-1 1 protein was induced in the aorta, and this was attenuated by administration of 17-␤ E 2 (2). Treatment of cultured rabbit aortic endothelial cells with 17-␤ E 2 also attenuated the lysophosphatidylcholine-induced expression of VCAM-1 protein. However, the concentrations of 17-␤ E 2 required to suppress VCAM-1 expression in vitro were in the supraphysiological range when compared with the physiological levels required for in vivo studies. We therefore postulated that one mechanism by which only relatively low concentrations of 17-␤ E 2 were required for in vivo studies when compared with in vitro studies was that 17-␤ E 2 might be converted to a more potent metabolite in vivo (2).
Some effects of estrogens are mediated through non-genomic mechanisms, whereas others require transcriptional activation of genes (3). These latter actions usually require that estrogens combine with specific receptors. Two types of estrogen receptors (ER) have been described. They are the "classical" ER, ER␣ and the "novel" more recently described ER␤ (4). Both ER␣ and ER␤ have been detected in endothelial cells and in vascular smooth muscle cells (5,6). The affinity of various estrogenic compounds for the two ER subtypes markedly differ. 17-␤ E 2 binds to ER␣ and ER␤ with similar affinity. 17-␣ ethinyl estradiol (EE), a synthetic estrogen widely used in oral contraceptive formulations, has an ER␣-selective agonist potency, whereas 17-epiestriol, an estrogen metabolite, and genistein, a phytoestrogen, at nM concentrations have ER␤-selective agonist potency (7).
The present work was therefore undertaken to test whether certain estrogenic compounds, including an estrogen metabolite, with different agonistic potencies for the two types of estrogen receptors were more potent than 17-␤ E 2 in suppressing VCAM-1 mRNA expression and the potential mechanism(s) involved. Information obtained from these studies could lead to the development of compounds that attenuate atherogenesis with fewer side effects than those observed with 17-␤ E 2 .
DNA, Denhardt's solution, and SDS were obtained from Sigma; the steroids 17-␤ E 2 , EE, and 17-epiestriol were also obtained from Sigma with ϳ99% purity as assessed by thin layer chromatography. Fetal bovine serum (FBS) charcoal/dextran-treated was obtained from Hy-Clone Laboratories. FBS was obtained from Atlas, Fort Collins, CO, and ICI-182780 was obtained from Tocris Cookson Ltd., Ballwin, MO. The human NFB p65 Nushift kit (2006760) was obtained from Geneka Biotechnology (Montreal, Canada).
Cell Culture-For isolation of human umbilical vein endothelial cells (HUVEC), umbilical cords from female fetus were selected as we observed increased expression of ER␣ and ER␤ mRNA when compared with those obtained from male fetus. HUVEC were isolated from freshly collected umbilical cords (female fetuses) as previously described (8) and cultured on 0.1% gelatin-coated 75-mm flasks in M199 medium supplemented with 20% FBS, 5 g/ml endothelial-derived growth factor, 100 g/ml heparin, 100 units of penicillin G sodium, 100 g/ml streptomycin sulfate, 0.25 g/ml amphotericin-B and 10 mM HEPES buffer, pH Ϸ 7.5. Prior to the experiments, cells were shifted to phenol red-free M199 medium supplemented with 2% charcoal/dextran-treated FBS and the antibiotics mentioned above. Only second or third passage cells were utilized in all the current studies.
Reverse Transcription-PCR for ER␣ and ER␤ mRNA-RNA extraction and purification were done using RNeasy mini kit as described by the manufacturer's protocol (Qiagen, Chatsworth, CA). Total RNA (3 g per sample) was subjected to reverse transcription reaction with 50 units of Moloney murine leukemia virus reverse transcriptase at 42°C for 25 min as previously described (9). The resulting cDNA samples were PCR-amplified using Gene Amp RNA PCR kit (PerkinElmer Life Sciences) according to the manufacturer's protocol. Oligonucleotide primers were designed for simultaneous PCR amplification (9) of specific DNA fragments contained in both ER␣ and ER␤ using the following set of primers: forward primer sequence, 5Ј-AAG AGC TGC CAG GCC TGC CG-3Ј; reverse primer sequence, 5Ј-GCC CAG CTG ATC ATG TGA ACC A-3Ј.
Northern Blot Analysis for VCAM-1 and Endothelial Nitric-oxide Synthase (eNOS) mRNA-10 g of total RNA was loaded on 1% agarose-formaldehyde gel, electrophoresed, and transferred to Hybond membrane (Amersham Biosciences) overnight by capillary action. The RNA was UV-cross-linked using GS Gene Linker (BioRad). VCAM-1 (Research Genetics) and eNOS (a gift from Dr. Thomas Michel, Brigham and Women's Hospital, Boston, MA) cDNAs were labeled with [␣-32 P]dCTP (ICN Biochemicals) using the random priming method. Membranes were prehybridized at 42°C overnight followed by hybridization with respective labeled probes for another 24 h at 42°C. The membranes were washed twice at room temperature with 2ϫ sodium chloride and sodium citrate (SSC) buffer and 0.5% SDS followed by washing at 65°C for 30 min twice with 0.5ϫ SSC and 0.1% SDS, and respective bands were quantitated using a Phosphoimager (Molecular Dynamics). The VCAM-1 band was normalized with the GAPDH band as an internal control.
Data Analysis-Values are expressed as mean Ϯ S.D. obtained from three separate experiments in each group. Differences between groups were assessed by one-way analysis of variance and Newman-Keuls multiple comparison test where appropriate. p values Ͻ 0.05 were considered as significant.

Effect of TNF␣ on VCAM-1 mRNA Expression and the Mod-
ulating Role of ER Agonists-Second and third passage HUVEC, which were utilized for all our studies, expressed both ER␣ and ER␤ mRNA (data not shown). No basal VCAM-1 mRNA expression was detected in the unstimulated cells or those treated with the different estrogenic compounds alone. The structures of the different estrogens studied are shown in Fig. 1. Preliminary experiments indicated that HUVEC when exposed to 10 ng/ml of TNF␣ for 4 h led to maximal VCAM-1 mRNA expression. This concentration and time period were utilized for all the experiments.

17-Epiestriol and VCAM-1 Expression
was gradually lost, and the VCAM-1 expression started returning to baseline values. However, the magnitude of VCAM-1 expression at 3 nM of 17-epiestriol was still significantly less than that observed with TNF␣ alone. Estriol at similar concentrations to that of 17-epiestriol had no effect (data not shown). Genistein (1 and 10 nM) attenuated VCAM-1 mRNA expression (Fig. 5). The calculated IC 50 (mean Ϯ S.D.) for genistein was 7.16 Ϯ 3 nM. The effects of all these estrogen receptor agonists were significantly attenuated in the presence of the ER antagonist ICI-182780 (1 M), and the results for 17-epiestriol are shown in Fig. 6.

Effect of 17-Epiestriol on TNF␣-induced VCAM-1 Protein
Expression-We next examined the effect of 17-epiestriol on TNF␣-induced VCAM-1 protein expression. HUVEC were preincubated with different concentrations of 17-epiestriol for 48 h. Six hours after the addition of TNF␣, the cells were fixed and assayed for cell-bound VCAM-1 protein by ELISA as described previously (11). A significant reduction in VCAM-1 protein was observed at a 300-pM concentration of 17-epiestriol (Fig. 7). Consistent with the observation related to VCAM-1 mRNA, TNF␣-induced VCAM-1 protein expression also showed a biphasic response to 17-epiestriol.
Intermediate Role of Nitric Oxide in the Action of 17-Epiestriol-To determine whether NO played an intermediate role in attenuating TNF␣-induced VCAM-1 expression, we first analyzed the effect of 17-epiestriol on eNOS expression. Treatment of HUVEC with 17-epiestriol showed a biphasic response on eNOS protein expression. The maximal increase (68 Ϯ 8%) was observed at 300 pM for both eNOS protein (Fig. 8) and mRNA expression (35%, data not shown). We next assessed the effect of three different concentrations of 17-epiestriol on TNF␣-induced VCAM-1 mRNA expression in the absence and presence of L-NAME (3 ϫ 10 Ϫ4 M), an inhibitor of NO synthesis. This concentration of L-NAME has been previously demonstrated by us to significantly attenuate basal-and agonist-stimulated NO release by endothelial cells (12). L-NAME either alone or along with TNF␣ did not affect VCAM-1 expression. 17-epiestriol (100 and 300 pM and 1 nM) attenuated VCAM-1 mRNA expression, and this effect was not observed in the presence of an inhibitor of NO synthesis (Fig. 9).
Effect of 17-Epiestriol on TNF␣-induced NFB Activation-To determine whether 17-epiestriol regulated TNF␣-induced VCAM-1 expression by inhibiting NFB migration to the nucleus, we performed gel-shift assays using oligonucleotides corresponding to the tandem-B sites on the VCAM-1 promoter. We also assessed whether NO played an intermediate role in this action by 17-epiestriol. 17-epiestriol at concentrations of 100 pM (data not shown) and 300 pM (Fig. 10, lane 5) significantly decreased the intensity of the shifted band produced by nuclear extracts obtained from HUVEC treated with TNF␣. However, 10 nM of 17-epiestriol was less effective in decreasing the intensity of the shifted band when compared with 300 pM (Fig. 10, lane 6). Specificity of the complexes was determined by competition with an excess of unlabelled oligonucleotide and with a labeled mutant oligonucleotide when no band was observed (Fig. 10, lane 4). The specificity was further confirmed by supershift analysis (Fig. 10, lane 3) with an affinity-purified polyclonal antiserum to p65. Nuclear translocation of NFB following treatment with 17-epiestriol in the presence of L-NAME was similar to that observed in the absence of 17-epiestriol and in the presence of L-NAME alone (Fig. 10, lanes 7 and 8). This suggests that the inhibitory effect of 17-epiestriol on the translocation of NFB to the nucleus following stimulation of HUVEC with TNF␣ was mainly dependent on NO synthesis. DISCUSSION The primary objective of this study was to assess whether some selected estrogenic compounds, including a phytoestrogen, a synthetic estrogen, and an estrogen metabolite with different agonistic potencies for ER␣ and ER␤ were more potent than 17-␤ E 2 in attenuating TNF␣-induced VCAM-1 expression of HUVEC in vitro. 17-␤ E 2 is the major circulating unconjugated estrogen in premenopausal women. It binds to human ER␣ and ER␤ with approximately similar affinity (7) and was therefore used as the prototype with which the comparisons to the other estrogens were made. EE, a synthetic estrogen, was selected since it is widely used as the estrogenic component in oral contraceptives and has ER␣selective agonistic potency (7). 17␣ epiestriol was selected because it is an estrogen metabolite (13), which has an ER␤selective agonist potency (7). Estriol was selected as its concentration markedly increases during pregnancy and is structurally very similar to 17␣ epiestriol except that the 17-OH group in estriol is in the 17-␤ position, whereas that of 17␣ epiestriol is in the ␣ position (Fig. 1). This difference makes it a very weak agonist for both ER␣ and ER␤ (7). The isoflavone phytoestrogen genistein binds to ER␤ almost as well as 17-␤ E 2 , whereas its affinity to ER␣ is considerably lower than that of 17-␤ E 2 (14).
Our results indicate that 17-␤ E 2 did not affect VCAM-1 mRNA expression at lower concentrations (100 and 300 pM), whereas at higher concentrations (1-300 nM) it decreased VCAM-1 mRNA in a concentration-dependent manner (Figs. 2  and 3). This action of 17-␤ E 2 in vitro has been studied by other investigators; however, the results are controversial. In studies using HUVEC, 17-␤ E 2 suppressed the induction of VCAM-1 mRNA expression induced by either lipopolysaccharide (2,15) or by interleukin -1␤ (8, 16). In contrast, Cid et al. (18) observed that 17-␤ E 2 did not affect basal VCAM-1 expression but caused a 30-50% increase in the presence of TNF␣ and had no effect when HUVEC were stimulated with interleukin-1. It has also been reported that 17-␤ E 2 had no effect on the TNF␣-induced expression of VCAM-1 in cultured endothelial cells (19). These contrasting observations may be due to differences in the extent to which the endothelial cells used in the different studies expressed the estrogen receptors, differences in the concentration of 17-␤ E 2 utilized, or to differences in the agents used to induce VCAM-1 expression.
In this regard, in both the present study and in those previously reported (8,15,16), the concentrations of 17-␤ E 2 required for suppression of VCAM-1 expression in vitro were in the supraphysiological range (1 nM-10 M). The physiological concentration of 17-␤ E 2 is in the pM range, whereas that of 17-epiestriol has not been reported in humans.
In contrast, EE, which is a more potent and selective agonist for ER␣ when compared with 17-␤ E 2 , did not attenuate TNF␣induced VCAM-1 mRNA expression at any of the concentrations studied (Figs. 2 and 3). One potential explanation for the absence of any effect of EE may be that ER␤ mediates the estrogen-induced decrease in VCAM-1 mRNA expression, whereas ER␣ has a negligible role. This may be the case as 17-epiestriol, which has an ER␤-selective agonistic potency, decreased VCAM-1 mRNA expression and the IC 50 was ϳ400ϫ lower than that observed with 17-␤ E 2. At present, we cannot explain why higher concentrations of 17-epiestriol were gradually less effective. A similar biphasic effect of 17-epiestriol on VCAM-1 protein expression (Fig. 7) and a shift of NFB to the nucleus (Fig. 10) was also observed. Estriol, which is a very weak agonist for both ER␣ and ER␤ (7), did not affect VCAM-1 mRNA expression. This indicates that the alignment of the 17-OH group in the ␣ position is an important requirement for 17-epiestriol for its selective ER␤ agonistic activity. The ability of the various estrogens to attenuate VCAM-1 mRNA expression was not observed in the presence of the anti-estrogen 1C1-182780. This confirmed that the action of the estrogens were receptor-mediated. Further and more specific assessment of the precise ER type involved in mediating the VCAM-1 mRNA expression by estrogens will have to await the discovery of specific ER␣ and ER␤ receptor antagonists. Genistein at nanomolar concentrations has binding affinity to ER␤, which is similar to that of 17-␤ E 2, whereas its affinity to ER␣ is considerably less (14). Our observation that genistein at nM concentrations (1 and 10 nM) decreased VCAM-1 mRNA expression supports the concept that ER␤ is most likely involved in the attenuation of VCAM-1 expression in endothelial cells and that ER␣ probably has very little role in this phenomenon. At these concentrations, genistein has very little tyrosine kinase inhibitory activity (14) and, hence, the effects are likely due to its action on the ER␤.
No estrogen-responsive element in the promoter region of the VCAM-1 gene has yet been identified (20), and other indirect mechanisms need to be considered. The VCAM-1 promoter contains consensus binding sites for the nuclear transcription factor NFB, the GATA family of transcription factors, and an AP-1 site (15). NFB is a proinflammatory transcription factor up-regulating several genes involved in endothelial activation (21). It has recently been demonstrated that a very high concentration of 17-␤ E 2 (10 M) suppressed VCAM-1 expression indirectly by inhibiting the nuclear translocation and DNA binding of NFB, and this was also associated with a reduction of AP-1 and GATA transcription factors binding to the VCAM-1 promoter (15). These investigators also demonstrated that in human endothelial cells, NFB inhibition by E 2 was associated with decreased IB-␣ degradation.
On this basis we wanted to assess the possibility that estrogens may have increased the synthesis of NO, which in turn may have led to the suppression of DNA binding of transcription factors, thereby leading to inhibition of VCAM-1 expression. There is a lot of similarity between the actions of estrogens and NO on various transcription factors that modulate VCAM-1 expression. NO donors inhibited cytokine-induced expression of VCAM-1 and repressed VCAM-1 gene transcription in part by inhibiting nuclear binding protein NFB (22,23). NO also stabilized the NFB inhibitor, IB-␣ (24), and cellular treatment with NO donor compounds also inhibited AP-1 binding to DNA (25).
We previously demonstrated that estrogens can increase NO production (12), and subsequently others have demonstrated that it can occur through both genomic (26) and non-genomic (27,28) mechanisms. We therefore decided to assess whether 17-epiestriol at concentrations that attenuated VCAM-1 mRNA expression also increased eNOS protein and whether this was at least in part due to increased eNOS mRNA. Our results indicated that 17-epiestriol also had a biphasic effect on eNOS protein expression. 17-Epiestriol at a concentration that maximally suppressed TNF␣-induced VCAM-1 mRNA expression also increased eNOS protein (Fig. 8) and mRNA, suggesting that NO may play an intermediary role in this action of 17-epiestriol. This was confirmed by observing that the suppression of VCAM-1 mRNA expression by 17-epiestriol was not observed following inhibition of NO synthesis (Fig. 9). Our results also suggest that 17-epiestriol-induced NO production played an intermediate role in inhibition of the nuclear translocation and DNA binding of NFB in HUVEC as this inhibitory effect of 17-epiestriol was significantly attenuated in the presence of an inhibitor of NO synthase. Higher concentrations (1 and 10 nM) of 17-epiestriol attenuated the eNOS protein expression when compared with 300 pM, and this may explain the biphasic response of 17-epiestriol on VCAM-1 mRNA and protein expression. Our study did not address the precise molecular mechanism(s) by which NO production stimulated by 17-epiestriol inhibited the translocation of NFB to the nucleus. It is also possible that the differences in actions observed between estrogenic compounds with different affinities for ER␣ and ER␤ receptors could be mediated by interaction with the AP-1 site. The interactions of ER␣ and ER␤ with the AP-1 site could lead to signaling in opposite ways and thus may play opposite roles in gene regulation (29). It would also be interesting to assess whether the reduction of AP-1 and GATA binding to the VCAM-1 promoter by estrogens is also mediated by an increase in NO production, thereby indicating an important intermediary role of NO in the regulation of some transcription factors by estrogens.
In conclusion, our results indicate that estrogens most likely inhibited VCAM-1 mRNA and protein expression by an action mediated by ER␤ and not by ER␣ and that NO played an intermediate role in this process. Our results further suggest that 17-epiestriol, an estrogen metabolite, is more potent than 17-␤ E 2 . This may be one potential explanation as to why much lower concentrations of 17-␤ E 2 are required to suppress VCAM-1 expression in vivo (where various metabolites of estrogens are formed) when compared with those required in vitro.
These results may have important clinical implications. An important area of research with relation to estrogen replacement therapy is identifying the role of ER␣ and ER␤ in mediating the antiatherosclerotic action of estrogens. It has been demonstrated that 17-␤ E 2 can inhibit responses to vascular injury in an acute carotid artery endothelium-denuded model in ovariectomized ER␣ Ϫ/Ϫ (30) and ER␤ Ϫ/Ϫ (31) animals to the same extent as in wild type animals. On this basis it has been suggested that ER␣ and ER␤ can complement each other in vivo to mediate the vasoprotective action of estrogens following vascular injury. Recently, it has been demonstrated that genistein, which has selective binding affinity to ER␤, provided vasoprotective actions that were equally efficacious to that observed with 17-␤ E 2 in an acute carotid artery endotheliumdenuded rat model and yet were devoid of effects on the uterus, which mainly have ER␣ receptors (32). In the acute carotid artery endothelium-denuded model, in contrast to the atherosclerosis model, the endothelial cells are denuded, and it would therefore be difficult to assess the actions of estrogens exerted specifically via the endothelial cells. It would therefore be interesting to assess whether 17-epiestriol is as effective or more potent than 17-␤ E 2 in attenuating atherogenesis and yet is devoid of effects on the reproductive system when compared with 17-␤ E 2. Further studies are in progress to assess this in vivo in an animal model. Our results demonstrating the biphasic action of 17-epiestriol in modulating VCAM-1 gene and protein expression indicate that estrogens may have paradoxical effects in inflammatory processes. This may be one potential explanation for the controversy in relation to the effects of estrogen replacement therapy on cardiovascular morbidity (17).