Estrogen Receptor (cid:1) Regulates Expression of the Orphan Receptor Small Heterodimer Partner* □ S

Hormonal status can influence diverse metabolic pathways. Small heterodimer partner (SHP) is an orphan nuclear receptor that can modulate the activity of several transcription factors. Estrogens are here shown to directly induce expression of the SHP in the mouse and rat liver and in human HepG2 cells. SHP is rapidly induced within 2 h following treatment of mice with ethynylestradiol (EE) or the estrogen receptor (cid:1) (ER (cid:1) )-selective compound propyl pyrazole triol (PPT). SHP induction by these estrogens is completely absent in ER (cid:1) KO mice. Mutation of the human SHP promoter defined HNF-3, HNF-4, GATA, and AP-1 sites as important for basal activity, whereas EE induction required two distinct elements located between (cid:2) 309 and (cid:2) 267. One of these elements contains an estrogen response element half-site that bound purified ER (cid:1) , and ER (cid:1) with a mutated DNA binding domain was unable to stimulate SHP promoter activity. This ER (cid:1) binding site overlaps the known farnesoid X receptor (FXR) binding site in the SHP promoter, and the combination of EE plus FXR agonists did not produce an additive induction of SHP expression in mice. Surprisingly, induction of SHP by EE did not inhibit expression of the known SHP target genes cholesterol 7 (cid:1) -hydroxylase (CYP7A1) or sterol 12 (cid:1) -hydroxylase (CYP8B1). However, the direct regulation of SHP expression may

Estrogens exert biological effects in numerous organs throughout the body. The role of estrogens in reproductive biology, the prevention of postmenopausal hot flushes, and the prevention of postmenopausal osteoporosis are well established (1). Estrogens reduce low density lipoprotein (LDL) 1 cholesterol levels and elevate HDL cholesterol levels (2)(3)(4), although these beneficial lipid changes may not translate into favorable clinical results (5). Estrogens may also inhibit the development of colon cancer (5), inhibit the development of Alzheimer's disease (6), and inhibit development of cataracts (7,8). The multitude of estrogen responses matches the widespread distribution of estrogen receptors (ERs) throughout numerous organs, with ER␣ expression the highest in uterus, pituitary, kidney, and adrenal gland and ER␤ expression highest in ovary, uterus, bladder, and lung (9).
Although many of these actions of estrogens are due to the classic signaling pathway in which an ER binds to an estrogen response element in the promoter of a gene (10), it is now clear that many actions of estrogens are mediated by interaction of ER with other signaling pathways. For example, estrogens inhibit chronic and acute liver inflammation in the mouse by a mechanism that does not require ER activation of gene expression (11,12). This in vivo activity of ER correlates with the in vitro ability of ER to inhibit NFB signaling (13,14), likely through a coactivator competition mechanism (15). Similarly, ER can regulate gene expression via interaction with AP-1 response elements (16).
An additional mechanism by which ER could influence diverse signaling pathways is by altering expression levels of other transcription factors. For example, estrogen treatment induces expression of STAT5A in both the kidney (17) and liver (11). Intriguingly, recent microarray studies have demonstrated that chronic treatment with estrogens induces expression of the orphan short heterodimer partner (SHP) receptor in the mouse liver (11). SHP is known to be able to repress the activity of many nuclear hormone receptors in vitro, including ER␣ and ER␤ (18,19), estrogen receptor-related receptor ␥ (ERR␥) (20), androgen receptor (21), arylhydrocarbon receptor nuclear translocator (22), hepatocyte nuclear factor 4 (23,24), constitutive androstane receptor (25), retinoic acid receptor ␤ (25), retinoid X receptor (24), liver X receptor ␣ (26), and liver receptor homologous protein-1 (LRH-1) (27)(28)(29). Conversely, SHP stimulates the activity of nuclear factor-B (30), peroxisome proliferator-activated receptor ␣ (31), and peroxisome proliferator-activated receptor ␥ (32). Whether all these interactions occur in vivo is less clear, with only the activity of SHP in regulating bile acid synthetic gene expression having been characterized (33,34). Nevertheless, regulation of SHP expression has the potential to influence a wide array of physiological processes.
Estrogen regulation of SHP expression might thus contribute to many facets of estrogen biology. However, previous studies have not differentiated between a direct induction of SHP expression by estrogens and an indirect induction of SHP as a secondary consequence of prolonged treatment with estrogens. To discern between these two possibilities, both in vivo kinetic studies and in vitro transfection studies were used here to * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. characterize estrogen regulation of SHP expression. The results indicate that ER␣ directly stimulates expression of the SHP gene. inositol-1-phosphate synthase (IPS), intestinal trefoil factor (ITF), tissue factor (TF), and CYP3A11 in each individual animal were quantified by real-time PCR. All expression values were normalized for expression of GAPDH and are presented as the mean Ϯ S.E., with six animals per group. The mean expression of each gene in animals receiving a vehicle treatment and assayed at time 0 is defined as 1.0, with all other expression values reported relative to this value. C, ovariectomized mice were administered a single dose of 10 g/kg EE or 30 mg/kg GW4064 in methylcellulose Tween vehicle. At 1, 2, 4, 8, or 24 h following treatment, livers were removed and mRNA levels for SHP in each individual animal were quantified by real-time PCR. The mean expression of each gene in animals receiving a vehicle treatment and assayed at time 0 is defined as 1.0, with all other expression values reported relative to this value.

MATERIALS AND METHODS
Animal Studies-Ovariectomized 129 strain mice (20 g) or Sprague-Dawley rats (200 g) were purchased from Taconic Farms (Lexington, KY). ER␣KO CH knockout (obtained from Dr. Ken Korach (35)), ER␤ knockout (developed at Wyeth (36)), and ER␣ CH ER␤ knockout mice (developed at Wyeth (36)), all on a 129 background, were bred and ovariectomized at Wyeth. All animals were fed a casein-based diet (#8117, Test Diet, Richmond, IN) unless otherwise specified. All compounds tested were synthesized at Wyeth. Compounds given as daily subcutaneous injections were administered in 0.1 ml of 1:1 dimethyl sulfoxide/phosphate-buffered saline or 0.1 ml of 90% corn oil/10% ethanol. Compounds given orally by gavage were administered in 0.1 ml of 0.5% methylcellulose, 2% Tween 80. Compounds given orally in feed were added to the casein diet ground to a powder by mixing. For 5-day treatment studies, cholic acid (CA, Sigma, St. Louis, MO) was added to the ground casein diet. For 5-week cholic acid feeding studies, mice were fed a commercial high fat diet containing 0.5% CA (#21539, Purina, Richmond, IN). Animals were euthanized 2 h after the final treatment, generally 4 h after commencement of the light cycle. For kinetic studies, animals were treated at ϳ7:30 a.m. and euthanized at various times throughout the day. For cycling studies, CD-1 strain young adult female mice (Charles River, Wilmington, MA) were assessed daily by vaginal smear for two complete cycles before use. All animals were euthanized at midday between 11:00 a.m. and 2:00 p.m.

FIG. 2. ER␣ is required for induction of SHP.
A, ovariectomized wild type mice (black bars) or ovariectomized ER␣KO CH mice (gray bars) were administered a single subcutaneous injection of corn oil/ethanol vehicle (veh), 10 g/kg EE (EE), 10 g/kg EE plus 10 mg/kg ICI182780 (EE ϩ ICI), or 5 mg/kg PPT (PPT). Four hours later the livers were removed and mRNA levels for SHP, STAT5A, PgD syn, IPS, ITF, and CYP3A11 were quantified by real-time PCR. All expression values were normalized for expression of GAPDH and are presented as the mean Ϯ S.E. *, p Ͻ 0.01 for comparison to vehicle-treated wild type animals. B, ovariectomized wild type, ER␣KO CH , ER␤KO, or ER␣ CH ER␤KO mice were treated by daily subcutaneous injection of vehicle (veh), 10 g/kg E2 (E2), or 10 g/kg E2 plus 5 mg/kg ICI182780 (E2 ϩ ICI) for 6 weeks. Livers were removed 2 h following the last treatment, and expression of SHP was quantified by real-time PCR with GAPDH expression used for normalization. The data are presented as the mean Ϯ S.E., with the mean expression value in wild type animals treated with vehicle defined as 1.0. *, p Ͻ 0.01 for comparison to vehicle treated wild type animals.
during the third cycle, with livers removed for RNA analysis. All treatments were in accord with accepted standards of care as specified by the Wyeth animal care committee.
RNA Analysis-RNA was prepared from frozen organs or from cultured cells by using TRIzol reagent (Invitrogen, Carlsbad, CA) and was treated with DNase to remove contaminating genomic DNA. Following repurification by RNeasy spin columns (Qiagen, Valencia, CA), 100 ng of total RNA was assayed for gene expression by real time reverse transcription-PCR as previously described (12) using the primers and 6-carboxyfluorescein-labeled probes listed in Table I. GAPDH expression was monitored using standard rodent and human assay kits (#4308313 and #402869, respectively, Applied Biosystems, Foster City, CA). Statistical significance was determined by analysis of variance on log-transformed data using Huber weighting.
Cell Culture-HepG2 cells (American Type Culture Collection, Rockville, MD) or Hep89 cells (37) were maintained in high glucose Dulbecco's modified Eagle's medium (Invitrogen, Rockville, MD) supplemented with 100 units/ml penicillin, 100 g/ml streptomycin, 1ϫ glutamax, 1 mM sodium pyruvate, 0.1 mM nonessential amino acids, and 10% fetal bovine serum. For cycloheximide studies, Hep89 cells plated in deficient growth media (maintenance media without phenol red and supplemented with 10% charcoal-stripped fetal bovine serum) were grown to 80% confluence. The cells were treated overnight with deficient growth medium containing 1 nM ICI182780 to minimize residual estrogenic activity in the medium. The following day, the medium was supplemented with Me 2 SO vehicle or 10 g/ml cycloheximide (Sigma). After 30-min incubation at 37°C, the medium was further supplemented with Me 2 SO vehicle or 30 nM 17␤-estradiol (E2). After incubation for a final 4 h at 37°C, the cells were harvested for RNA preparation using TRIzol reagent.
Plasmids-A 1.4-kb region of the human SHP promoter spanning Ϫ1383 to ϩ19 was obtained through PCR amplification using 5Ј-TGAGAAAGATGCCCGGGACAGA-3Ј forward and 5Ј-GGGCTCGAGC-CCCTGGTTGGCTGGTGCTCA-3Ј reverse oligonucleotides with the human BAC clone RP11-285H13 (#MB11201, Invitrogen) as a template. This fragment was then digested with NheI and XhoI and cloned into the luciferase reporter gene pGL3-Basic (Promega, Madison, WI). Each of the human SHP promoter mutations used in this study were created by PCR amplifications using the Ϫ1383 SHP promoter as template followed by digestion with appropriate restriction enzymes and ligation into pGL3-Basic.
Transfections-HepG2 cells were seeded into 24-well plates (Falcon, Franklin Lakes, NJ) at a density of 150,000 cells/well in deficient growth media. After 24 h, the cells were transfected with 250 ng of SHP reporter construct, 30 ng of ER expression plasmid, and 100 ng of M19 ␤-galactosidase control expression plasmid using LipofectAMINE 2000 (Invitrogen) following the manufacturer's protocol. Six hours later, the cells were treated with Me 2 SO vehicle or 10 nM E2. Cell lysates in 1ϫ lysis buffer (Promega) were prepared 16 h later and assayed for luciferase and ␤-galactosidase activity using luciferase (Promega) and ␤-galactosidase (Tropix, Bedford, MD) assay systems. The normalized luciferase expression level of the pGL3-Basic plasmid lacking any promoter was defined as 1.0. All values reported are the mean of multiple independent assays.
Electrophoretic Mobility Shift Assays-Protein-DNA complexes were analyzed by incubation of baculovirus-expressed human ER␣ (Panvera, Madison, WI) with 32 P-labeled DNA probe containing the SHP Ϫ309/ Ϫ267 region followed by electrophoresis in low ionic strength polyacrylamide gels as described previously (37).

RESULTS
Recently we have demonstrated that treatment of ovariectomized mice with estrogens for 5 days induces expression of SHP in the liver (11). This induction by chronic estrogen treat- FIG. 3. ER␣ regulates SHP expression in the rat. Ovariectomized Sprague-Dawley rats were treated by daily subcutaneous injection of vehicle, 10 g/kg E2, or 5 mg/kg PPT for 6 weeks. Livers were removed 2 h following the last treatment and expression of SHP was quantified by real-time PCR with GAPDH expression used for normalization. The data are presented as the mean Ϯ S.E., with the mean expression value in wild type animals treated with vehicle defined as 1.0. *, p Ͻ 0.01 for comparison to vehicle-treated wild type animals.

FIG. 4. EE and PPT dose-response curves for induction of gene expression in the mouse liver.
A, ovariectomized mice were administered a single subcutaneous injection of increasing amounts of EE (0.3-100 g/kg, black circles) or PPT (0.1-10 mg/kg, open circles). Four hours following treatment, livers were removed and mRNA levels for SHP, STAT5A, PgD syn, IPS, and CYP3A11 in each individual animal were quantified by real-time PCR. All expression values were normalized for expression of GAPDH, and are presented as the mean Ϯ S.E., with six animals per group. The mean expression of each gene in animals receiving vehicle treatment is defined as 1.0, with all other expression values reported relative to this level. B, SHP mRNA levels were determined in CD-1 mice at proestrus (gray bar) or at estrus (black bar) by real-time PCR with GAPDH expression used for normalization. The data are presented as the mean Ϯ S.E., with the mean expression value in mice at estrus defined as 1.0. *, p Ͻ 0.01 for comparison to estrus mice. ment could be mediated either through a direct induction of SHP expression by ER, or through a secondary response subsequent to some other action of estrogen treatment. To delineate between these possibilities, expression of SHP and several other estrogen-regulated genes was determined at various times following treatment of ovariectomized mice with ethynylestradiol (EE) or the ER␣-selective agonist 4-propyl-1,3,5tris(4-hydroxyphenyl)pyrazole (PPT). Of the known estrogenresponsive genes in the liver, SHP induction was the most rapid, with peak increases in SHP mRNA levels occurring 2 h after estrogen treatment (Fig. 1A). The STAT5A gene also responded rapidly, with a peak induction at 4 h. Three other known estrogen-responsive genes, prostaglandin D synthase (PgD syn), inositol-1-phosphate synthase (IPS), and intestinal trefoil factor (ITF) all showed significantly slower induction kinetics than either SHP or STAT5A. The induction kinetics of SHP and STAT5A in the liver was equivalent to those seen for several genes, including STAT5A and tissue factor in the kidney (Fig. 1B), suggesting these genes are directly responsive to circulating estrogen levels. Finally, identical rapid kinetics for SHP induction occurred following the oral administration of either EE or GW4064 (Fig. 1C), an FXR agonist that directly activates the SHP promoter. These results together suggest that induction of SHP expression is a primary effect of treatment with estrogens.
Synthetic estrogens have been reported to act as human pregnane X receptor agonists (38). However, neither EE nor PPT induced expression of CYP3A11, a sensitive marker for pregnane X receptor activation (39), at any time point (Fig. 1A).
To verify that SHP induction was mediated through ER␣, ovariectomized wild type mice or ER␣KO CH mice were treated with EE, EE plus the antagonist ICI182780, or PPT (Fig. 2). Induction of SHP expression by EE was inhibited by ICI182780, and neither EE nor PPT treatment induced SHP expression in the ER␣KO CH mouse. Similar results were obtained for the STAT5A, PgD syn, and IPS genes. However, the ITF gene remained responsive to EE and PPT induction in the ER␣KO CH mouse. This induction of ITF in the ER␣KO CH mouse could be blocked by ICI182780 (not shown) and is likely mediated by the aberrantly spliced ER␣ transcripts that are found in the ERaKO CH mouse, as has been described for genes in several other tissues (40,41). Chronic treatment of mice with 17␤-estradiol (E2) induced SHP expression in WT and ER␤KO mice but not in either ER␣KO CH or ER␣ER␤KO mice (Fig. 2B). Together, the induction of SHP expression by the ER␣-selective agonist PPT and the lack of induction of SHP by estrogens in the ER␣KO CH mouse suggest that ER␣ is responsible for the direct induction of SHP expression. Finally, the ability of ER␣ to induce expression of SHP was not limited to the mouse, because both EE and PPT induced expression of SHP in the rat (Fig. 3).
In these studies, the magnitude of gene induction by PPT was consistently greater than the magnitude of induction by EE. To determine whether this represented a fundamental difference between PPT, which activates only ER␣, and EE, which activates both ER␣ and ER␤, ovariectomized mice were treated with increasing amounts of either EE or PPT. Both EE and PPT produced very similar dose response curves for induction of SHP (Fig. 4A). The previous differences in the magnitude of SHP induction by PPT as compared with EE reflect simply the dose of compound administered. Approximately 10fold higher concentrations of PPT than of EE are required to produce the same degree of activation of a synthetic reporter construct in vitro (42), and the dose-response curves of EE and PPT induction of genes such as STAT5A, PgD syn, and IPS show the expected 10-fold shift. In contrast, the ED 50 of PPT for SHP induction was ϳ100-fold higher than the ED 50 of EE for SHP induction. This may suggest that the mechanism of ER␣ induction of SHP expression is different from the mechanism of ER␣ induction of these other genes.
The high sensitivity of SHP to induction by EE suggested that SHP expression could be regulated by natural fluctuations in estrogen levels. In the mouse, plasma E2 levels are typically about 3-fold higher at proestrus than at estrus (43,44). In concordance with these levels, mice at proestrus had 4-fold higher levels of liver SHP mRNA than did mice at estrus (Fig.  4B). Thus SHP expression is not only regulated by exogenous administration of estrogens but is also regulated by natural fluctuations of estrogen levels.
To determine whether E2 regulation of SHP gene expression also occurs in human hepatocytes, HepG2 cells, which have lost the ability to express ER␣, or Hep89, a derivative of HepG2 cells engineered to constitutively express ER␣ (37), were treated with E2. No regulation of SHP mRNA levels occurred in the HepG2 cells (Fig. 5A). SHP mRNA was constitutively elevated in the Hep89 cells relative to the HepG2 cells. Treatment of Hep89 cells with E2 stimulated SHP expression, whereas treatment with ICI182780 reduced SHP expression to the level found in HepG2 cells and also blocked E2 stimulation of SHP expression. To verify that the stimulation seen in Hep89 cells was due to ER␣, HepG2 cells were transiently cotransfected with a plasmid containing 1383 bp of the human SHP promoter (45) driving expression of a luciferase reporter gene along with either an empty expression vector or an ER␣ expression vector. Treatment with E2 or PPT induced SHP promoter activity only in the cells cotransfected with ER␣ (Fig. 5B).
Although the rapid induction of SHP by ER␣ in vivo suggested a direct activation of the SHP promoter, this was directly confirmed by monitoring the ability of E2 to induce SHP expression in cells treated with cycloheximide. Cycloheximide treatment of Hep89 cells itself increased SHP expression levels, as has been seen in HepG2 cells for at least one other estrogen-regulated gene, the proteinase inhibitor 9 gene (46). In the presence of cycloheximide, E2 treatment still resulted in a further augmentation of SHP expression (Fig. 5C). As a control for inhibition of protein synthesis, treatment of Hep89 cells with cycloheximide induced expression of ICAM-1, a known effect of cycloheximide inhibition of translation of the short-lived IB protein (47). Finally, ER␣ induction of SHP promoter activity required an intact DNA binding domain. Cotransfection of an expression plasmid encoding an ER␣ protein with the DNA binding specificity altered to that of the glucocorticoid receptor failed to activate the SHP promoter following E2 treatment (Fig. 5D), although this chimeric receptor did activate a glucocorticoid response element reporter plasmid following E2 treatment (data not shown). These results suggest a direct activation of the SHP promoter by ER␣, potentially through a direct binding mechanism.
Analysis of the SHP promoter region up to Ϫ1383 failed to identify any good matches to a consensus estrogen receptor response element (ERE), although several ERE half-sites were identified. Analysis of promoter elements required for basal and ER␣-induced SHP expression in hepatocytes was performed by cotransfection of a series of SHP promoter mutations (delineated in Fig. 6) into HepG2 cells (Fig. 7 and Supplemental Table SII). Four major elements (HNF-4, HNF-3, GATA, and AP-1 as shown in Fig. 6) driving basal expression were identified. Thus, removal of an HNF-4 consensus element at Ϫ550 by a 5Ј deletion (/Ϫ560 compared with Ϫ529, Fig. 7A), by an internal deletion (del Ϫ580/Ϫ529, Fig. 7C) or by destruction of the HNF-4 site with an SpeI restriction site (LS-554/Ϫ548, Fig. 7D) all reduced basal activity by about 2-fold. Second, removal of an HNF-3 consensus element at Ϫ465 by a 5Ј deletion (/Ϫ481 compared with /Ϫ429, Fig. 7A), by a 3Ј deletion (del Ϫ462/Ϫ126 compared with del Ϫ468/Ϫ126, Fig 7B), by an Introduction of an SpeI site between the HNF-3 and GATA elements (LS Ϫ461/Ϫ455) did not alter basal promoter activity. Fourth, removal of an AP-1 site at Ϫ265 by a 5Ј deletion (/Ϫ279 compared with /Ϫ229, Fig. 7A), by a 3Ј deletion (Ϫ260/Ϫ126 compared with Ϫ267/Ϫ126, Fig. 7B), by an internal deletion (del Ϫ267/Ϫ179, Fig. 7C), or by destruction of the AP-1 site with an SpeI site (LS Ϫ266/Ϫ261, Fig. 7D) all lowered basal promoter activity by 2-to 4-fold. Thus the major elements regulating basal SHP expression in HepG2 cells appear to be HNF-3, GATA, AP-1, and HNF-4 sites, in approximately that order of importance.
None of these elements was required for E2 induction of the SHP promoter. Thus promoter constructs in which the HNF-4 site (LS Ϫ554/Ϫ549), the HNF-3 site (LS Ϫ467/Ϫ462), the GATA site (LS Ϫ455/Ϫ450), or the AP-1 site (LS Ϫ266/Ϫ261) were disrupted all were still induced by E2 (Fig. 8C). Removal of either the HNF-3 or GATA site actually increased induction by E2, potentially because of an increased contribution of the E2-responsive element to the remaining SHP promoter activity in the absence of these constitutive elements. Analysis of the series of 5Ј deletions suggested that the region between Ϫ316 and Ϫ229 was involved in E2 induction (Fig. 8A). Similarly, the 3Ј series of deletions suggested that the E2 response element was located between Ϫ307 and Ϫ273 (Fig. 8B). However, in both series of mutations, the loss of E2 induction occurred over a series of progressive deletions, and a complete sudden loss of E2 induction did not occur. To further characterize this region in detail, a series of clustered point mutations covering the SHP promoter between Ϫ309 and Ϫ261 were assayed for E2 induction (Fig. 8C). No single mutation could be identified that completely eliminated E2 induction. However, a set of mutations at the 5Ј end of this region (LS Ϫ309/Ϫ304, LS Ϫ307/ Ϫ302E, and LS Ϫ303/Ϫ298) and another set of mutations at the 3Ј end of this region (LS Ϫ285/Ϫ280, LS Ϫ278/Ϫ273, and LS Ϫ272/Ϫ267) partially reduced E2 inductions. When a mutation from the 5Ј set was combined with a mutation from the 3Ј set (LS Ϫ307/Ϫ302E and Ϫ278/Ϫ273), E2 induction of the SHP promoter was completely abolished. A combination of two mutations that individually did not alter E2 induction (LS Ϫ297/Ϫ292 and Ϫ266/Ϫ261E) also did not reduce E2 induction. Thus E2 induction of the SHP promoter relied upon two distinct but closely linked sites.
The previous finding that an intact DNA binding domain was required for ER␣ activation of the SHP promoter suggested that ER␣ might bind to one of these two regions. Purified human ER␣ bound to an oligonucleotide containing the Ϫ309 to Ϫ267 region of the SHP promoter (Fig. 9A). This binding could be competed by an authentic ERE, but not by a mutated ERE, and inclusion of an antibody directed against ER␣ produced a supershifted complex. Binding analysis of the series of mutations tested in HepG2 cells revealed that the LS Ϫ285/Ϫ280 mutation alone was able to disrupt binding of ER␣ to this fragment of the SHP promoter (Fig. 9B). This mutation destroys an ERE half-site (GGTCA, Fig. 6) present within this region. ER␣ direct recognition of this ERE half-site likely contributes to E2 induction of the SHP promoter, although other interactions must also contribute to this induction based upon the transfection results.
The identified ER␣ binding site comprises one arm of the inverted repeat IR-1 FXR binding site within the SHP promoter, suggesting that simultaneous occupancy of this site by both ER␣ and FXR should not occur. To determine whether ER␣ and FXR agonists could produce an additive induction of SHP expression, ovariectomized mice were fed a diet containing an ER agonist (EE), an FXR agonist (GW4064), or both together. Both EE and GW4064 individually stimulated expression of SHP (Fig. 10). However, the addition of EE to GW4064 treatment did not produce any further stimulation of SHP expression. EE induction of STAT5A, PgD syn, IPS, and ITF was not altered by treatment with GW4064. To verify that these results were not unique to GW4064, ovariectomized mice were fed either a control diet or a diet containing cholic acid (to activate FXR) along with subcutaneous EE treatment. EE activated SHP expression in the mice consuming the control diet but not in mice consuming the cholic acid-containing diet (Fig.  11A). The same result was obtained if the study period was extended to 5 weeks to ensure equilibrium had been obtained (Fig. 11A). The inability of ER␣ to activate SHP expression in animals in which FXR was activated is consistent with the overlapping ER␣ and FXR response elements identified by the transfection analysis and binding studies.
One of the known in vivo effects of SHP induction is to repress expression of the bile acid synthetic pathway genes CYP7A1 and CYP8B1. However, treatment with EE for either 5 days or 5 weeks induced expression of SHP, but no corresponding reduction in CYP7A1 or CYP8B1 mRNA levels occurred (Fig. 11A). Similar results were obtained using E2 (data not shown). Induction of SHP mRNA levels by either 5-day or 5-week treatment with CA induced SHP slightly more than did the estrogen treatment, yet CA treatment completely abolished CYP7A1 and CYP8B1 expression. To determine whether the degree of SHP induction produced by EE was sufficient for repression of bile acid synthetic genes, ovariectomized mice were fed diets containing increasing amounts of CA (Fig. 11B). Addition of 0.1% CA was sufficient to repress expression of CYP7A1 and CYP8B1 by 80 -90%, with this amount of CA inducing SHP expression to a lesser extent than did EE treat-FIG. 8. Elements mediating ER␣ induction of SHP promoter activity. HepG2 cells were cotransfected with 250 ng of SHP luciferase reporter plasmid containing various regions of human SHP promoter sequence, 30 ng of pcDNA3 ER␣ expression plasmid, and 100 ng of M19 ␤-galactosidase control plasmid. After treatment with either vehicle or 10 nM E2 for 16 h, cell extracts were prepared and assayed for luciferase and ␤-galactosidase activities. All luciferase values were normalized for ␤-galactosidase expression. The -fold regulation was calculated as the ratio of normalized luciferase activity in E2-treated cells to the normalized luciferase activity in vehicle-treated cells. The -fold regulation for the control plasmid pGL3-Basic was defined as 1.0 for each independent study. All plasmids were assayed in at least three independent transfections. A, results from a series of 5Ј progressive truncations of the SHP promoter. B, results from a series of 3Ј progressive truncations of the upstream SHP promoter joined to the proximal SHP promoter at position Ϫ126. C, results from a series of clustered point mutations produced by introduction of either SpeI or EcoRI (denoted by E) restriction sites into the SHP promoter.
FIG. 9. Binding of ER␣ to the SHP promoter. A, a 32 P-labeled DNA probe containing the Ϫ309 to Ϫ267 region of the SHP promoter was incubated with 0.33 g of recombinant human ER␣ (lanes 2-5). Competition experiments were performed by inclusion of a 100-fold excess of wild type (lane 3) or mutated (lane 4) vitellogenin ERE oligonucleotides in the incubation. Supershift analysis was performed by the addition of an affinity-purified monoclonal antibody against human ER␣ in the incubation (lane 5). Protein-DNA complexes were resolved by electrophoresis in low ionic strength polyacrylamide gels. B, 32 Plabeled DNA probes containing the Ϫ309 to Ϫ267 region of the SHP promoter (WT) or this region with the indicated clustered point mutations were incubated with 0.33 g of recombinant human ER␣ and subjected to gel electrophoresis. ment. Thus, although the magnitude of SHP induction by EE would appear to be sufficient to produce a significant reduction in CYP7A1 and CYP8B1 expression, no such reduction occurred.

DISCUSSION
Previously we have demonstrated that chronic dosing of mice with 17␤-estradiol increased expression of SHP (11). Here, we demonstrate that this is due to a direct induction of SHP promoter activity mediated by ER␣. In cultured HepG2-expressing ER␣, E2 induced expression of SHP in the presence of cycloheximide, and cotransfection of an ER␣ expression plasmid conferred estrogen responsiveness upon a reporter plasmid driven by the human SHP promoter. Estrogen regulation of the SHP promoter was localized to between Ϫ309 and Ϫ267 and was distinct from four major sites (HNF-4, HNF-3, GATA, and AP-1) controlling basal SHP promoter activity in HepG2 cells. This region contained two elements important for maximal induction by estrogens, with one element located between Ϫ309 andϪ298 and a second element located between Ϫ285 and Ϫ267. Both elements are highly conserved among the human, mouse, and rat promoters (Fig. 6B). The second element contained an ERE half-site that bound to purified ER␣ and contributed to the induction of the SHP promoter by EE treatment of HepG2 cells. The related nuclear hormone transcription factor ERR␥ can also bind to ERE half sites (48) and activates the SHP promoter (20). However, ERR␥ activation does not utilize this ERE half-site but, rather, utilizes an SF1 site located at Ϫ80 in the SHP promoter (49).
Although estrogen regulation of gene expression has been typically considered as being mediated by a canonical inverted repeat ERE, a growing number of genes are known to require additional elements for estrogen regulation. For example, ERE half-sites in combination with SP1 sites can confer estrogen regulation to multiple genes (50), whereas an AP-1 site is required for function of an ERE within the pS2 gene promoter (51). Neither of these mechanisms are likely for the SHP promoter, because the upstream region has poor homology to ei-ther SP1 or AP-1 binding sites. The one functional AP-1 site in the SHP promoter as identified here and by others (52) is clearly dispensable for estrogen regulation of SHP promoter activity. Interestingly, SREBP-1 is required for estrogen induction of LDL receptor expression (53), and our analysis of the human LDL receptor promoter suggests the presence of an ERE half-site ϳ15 nucleotides upstream from the SREBP-1 binding site (data not shown). The corresponding position relative to the ERE half-site is within the Ϫ309/Ϫ298 element, and this element has partial homology to the SREBP-1 consensus binding site (54). However, this element also contains a base change known to disrupt SREBP-1 binding in the LDL receptor promoter (55). Furthermore, cycloheximide treatment rapidly diminishes SREBP-1 protein levels (56), suggesting that cycloheximide treatment likely would have blocked E2 induction of the SHP promoter if SREBP-1 were required. Finally, many genes that utilize half-ERE sites to confer regulation have multiple functional half-sites, ranging from two active half-sites in the prothymosin ␣ promoter (57) to numerous active sites in the NHE-RF gene promoter (58). The Ϫ309/ Ϫ298 element contains a near match (AGGgCA) to an ERE half-site. Although this site did not demonstrate binding by ER␣ in gel shift assays, the ability of ER␣ to interact with this site may be greater in the context of the full promoter with associated transcription factors bound.
The direct activation of the SHP promoter by ER␣ in cultured cells was reflected by activation of the SHP promoter in vivo. Thus, either subcutaneous or oral administration of EE or the ER␣-specific agonist PPT rapidly induced SHP expression, with peak SHP expression occurring at 1-2 h following treatment. Interestingly, a set of transcription factors exhibited these rapid inductions. For example, STAT5A was also rapidly induced by estrogen treatment. These kinetics were similar to those seen for gene induction in the kidney (17) and for immediate response genes in the uterus (59). The induction of SHP was blocked by ICI182780 treatment and was completely absent in ER␣KO CH mice. Although the plasma concentrations of EE and PPT were not measured in these studies, the inductions of SHP and STAT5A in the liver as well as STAT5A and tissue factor in the kidney give a pattern very similar to the plasma concentration profile following administration of a single dose of EE to women, suggesting that expression levels of these genes all rapidly reflect circulating estrogen levels. Whereas all genes in the kidney were rapidly induced, in the liver several genes such as PgD syn, IPS, ITF, and others had much slower kinetics of induction. Because SHP is demonstrated here to be a direct target of ER␣, and the presence of a perfect ERE in the mouse STAT5A promoter (not shown) suggests it may also be a direct target of ER␣, it may be that estrogen induction of a small number of transcription regulators such as SHP and STAT5A mediate the subsequent slower induction of many additional genes.
The ER binding site in the SHP promoter utilizes one of the inverted repeats of the FXR binding site (29), suggesting that FXR and ER␣ should not be able to simultaneously occupy this site. In concordance with this, EE treatment failed to further induce expression of SHP in mice in which FXR was activated either by treatment with GW4064 or by consumption of a diet containing cholic acid. Interestingly, the human SHP promoter has also been shown to bind LXR at a DR-4 site, which also overlaps the FXR IR-1 element (60), in a manner analogous to the ER␣ binding site. Thus at least three nuclear hormone receptors, FXR, LXR, and ER␣, all bind to the same region of the human SHP promoter. In the mouse and rat, the LXR DR-4 binding site is disrupted by a single nucleotide change that disrupts LXR activation. Thus no activation of SHP expression occurs after treatment of rodent hepatocytes with an LXR agonist (60). Whether ER␣ interaction with the human SHP promoter influences LXR activation of SHP expression remains to be determined.
SHP has been suggested to regulate the activity of numerous transcription factors in transfection assays. However, in animals the only clearly defined regulatory effect for SHP is in the bile acid feedback pathway in which FXR induction of SHP inhibits expression of CYP7A1 and CYP8B1 (28,29). This has been most clearly demonstrated in the SHP KO mouse, in which expression of CYP7A1 is constitutively elevated (33,34). SHP repression of CYP7A1 expression is mediated by inhibition of ␣ 1 -fetoprotein transcription factor (LRH-1) activation of the SHP promoter (28,29), whereas SHP repression of CYP8B1 expression appears to differ between species, with SHP inhibition of HNF-4␣ activity responsible for repression of human CYP8B1 promoter activity (23) and SHP inhibition of ␣ 1 -fetoprotein transcription factor responsible for repression of rat CYP8B1 promoter activity (61). Surprisingly, although the FIG. 11. EE induction of SHP does not repress CYP7A1 or CYP8B1 expression. A, ovariectomized mice were fed either a casein-based diet (Ϫ) for 5 days, a casein-based diet supplemented with 0.3% cholic acid (CA, ϩ) for 5 days, a control diet (Ϫ) for 5 weeks, or a high fat diet containing 0.5% CA (ϩ) for 5 weeks. All animals received daily subcutaneous injections of vehicle (Ϫ) or 10 g/kg EE (ϩ). After 5 days of treatment, livers were removed 1 h after the last subcutaneous treatment. After 5 weeks of treatment, livers were removed 24 h after the last subcutaneous treatment. Levels of SHP, CYP7A1, and CYP8B1 mRNA were quantified by real-time PCR with normalization for GAPDH expression. Values shown are the mean Ϯ S.E. determined from six animals per group, with gene expression in animals consuming the control diet and treated with vehicle for either 5 days or 5 weeks defined as 1.0. *, p Ͻ 0.01 for comparison to control diet vehicle animals. B, ovariectomized mice were fed a casein diet supplemented with various concentrations of CA. After 5 days of treatment, liver mRNA levels for SHP, CYP7A1, and CYP8B1 were determined by real-time PCR with normalization for GAPDH expression. Values shown are the mean Ϯ S.E. of determinations, with 12-18 animals total per group. The expression level of each gene in the absence of CA supplementation was defined as 1.0. *, p Ͻ 0.01 for regulation by CA. magnitude of induction of SHP by EE was similar to that produced by 0.1% cholic acid (which resulted in 90% repression of expression of CYP7A1 and CYP8B1), EE treatment did not produce the expected repression of either CYP7A1 or CYP8B1 expression. Similar results were found for E2 treatment (data not shown), and there was no correlation between SHP expression and CYP7A1 and CYP8B1 repression in cycling animals (data not shown). Some FXR ligands such as GW4064 both induce SHP expression and repress CYP7A1 and CYP8B1 expression (29,33), whereas others such as guggelsterone induce SHP expression but fail to repress either CYP7A1 or CYP8B1 expression (62), as seen here for EE. It is not yet clear whether a similar pattern will emerge for ER␣, with some ER␣ ligands both inducing SHP and repressing CYP7A1 and CYP8B1.
The basis for the lack of repression of CYP7A1 and CYP8B1 repression despite the induction of SHP in EE-treated mice remains to be determined. In SHP KO mice, cholic acid still represses CYP7A1 expression via alternative mechanisms (33,34). SHP induction by itself might not be sufficient to mediate repression of CYP7A1 and CYP8B1 without these additional signaling pathways. FXR activation could provide additional signals such as production of a SHP ligand with the ability to confer SHP repression upon CYP7A1 and CYP8B1 promoters. A second possibility is that ER sends a stimulatory signal to the CYP7A1 and CYP8B1 promoters to countervail the SHP repression signal. However, ER␣ does not stimulate activity of either the human CYP7A1 or CYP8B1 promoters in cotransfection assays, and EE treatment of mice did not increase mRNA levels for either LRH-1 or HNF-4␣ (data not shown). A third possible explanation relies on the ability of liganded ER␣ to bind to SHP, which then represses ER␣ activity (18,19,63). If the affinity of SHP for liganded ER␣ were significantly greater than for either LRH-1 or HNF-4␣, then the SHP produced by ER␣ activation might be sequestered by the active ER␣ and thereby be unable to mediate repression of LRH-1 or HNF-4␣ activity. In this model, it would be expected that EE treatment could reverse bile acid-mediated repression of CYP7A1 and CYP8B1 expression, but only a very small effect in this direction was seen in the studies performed here. Whatever the basis, it appears that in vivo mechanisms have been provided to prevent cross-talk between SHP utilization in bile acid signaling pathways and in estrogen signaling pathways.
SHP is expressed in many tissues of the body aside from liver (25,45), suggesting that regulation of bile acid synthesis is unlikely to be its sole role. Estrogen treatment can regulate SHP expression in organs other than the liver and coexpression of ER␤ in HepG2 cells can induce SHP promoter activity (data not shown), suggesting that all in vivo estrogenic regulations of SHP need not necessarily be mediated through ER␣. The conservation of estrogen regulation between rodents and human suggests estrogen regulation of SHP in these organs may play a role in the biology of estrogen. For example, postmenopausal women have increased amounts of white adipose tissue, and hormone replacement therapy decreases the level of white adipose tissue (64 -66). Similarly, white adipose tissue is increased by ovariectomy (67) or in ER␣KO mice (68). Inactivating mutations in the SHP gene have been associated with mild obesity in the Japanese population (69). Given the direct regulation of SHP by ER␣ demonstrated here, these findings suggest the hypothesis that regulation of SHP levels are an important part of the mechanism by which estrogens regulate adipose tissue levels. Further analysis of the role of estrogen induction of SHP in specific organs other than the liver will be necessary to delineate the contributions of SHP to estrogen physiology.