Regulation of 3-Hydroxy-3-methylglutaryl Coenzyme A Reductase Promoter by Nuclear Receptors Liver Receptor Homologue-1 and Small Heterodimer Partner

Cholesterol homeostasis in mammals involves pathways for biosynthesis, cellular uptake, and hepatic conversion to bile acids. Key genes for all three pathways are regulated by negative feedback control. Uptake and biosynthesis are directly regulated by cholesterol through its inhibition of the proteolytic activation of the sterol regulatory element binding proteins. The conversion of cholesterol into bile acids in the liver is regulated through the bile acid-dependent induction of the negatively acting small heterodimer partner nuclear receptor. In this report, we have shown that the small heterodimer partner also directly regulates cholesterol biosynthesis through inhibition of 3-hydroxy-3-methylglutaryl coenzyme A reductase but has no effect on low density lipoprotein receptor expression. This has significant metabolic significance, as it provides both a mechanism to independently regulate cholesterol synthesis from uptake (an essential regulatory feature known to occur in vivo) and a pathway for direct regulation of cholesterol biosynthesis by bile acids. This latter feature ensures that the early phase of bile acid synthesis (pre-cholesterol) is in metabolic communication with the later stages of the pathway to properly regulate whole pathway flux. This highlights an important regulatory feature that is shared with other key branched, multienzyme pathways, such as glycolysis, where pathway outflow through pyruvate kinase is regulated by the concentration of a key early intermediate, fructose 1,6-bisphosphate.

In the mammalian liver, cholesterol serves as a precursor in the synthesis of bile acids and as metabolite flow increases through the cholesterol pathway, bile acid production is increased. Bile acids act as feedback regulators of their biosynthesis by inhibiting the nuclear receptordependent activation of key bile acid biosynthetic target genes (1).
The nuclear receptor FXR 3 binds bile acids and induces the expres-sion of genes involved in bile acid export and the gene encoding another nuclear receptor, SHP (2,3). SHP lacks a DNA binding domain but interacts with the carboxyl-terminal activation domain of other DNAbound nuclear receptors and inhibits their activity. The best documented target of SHP repression is the LRH-1/FTF nuclear receptor, which binds DNA as a monomer and activates expression of the bile acid biosynthetic genes CYP7A1 (2,3) and CYP8B1 (4). Thus, as SHP levels rise in response to FXR-dependent activation, bile acid production is repressed through the negative effect of SHP. Cholesterol is an essential component of mammalian membranes, and its production is tightly controlled through the negative effects of cholesterol directly on the endoplasmic reticulum membrane proteins SREBPs and HMG-CoA reductase (5). As cholesterol accumulates, SREBP trafficking to and proteolytic activation in the Golgi are repressed, and the proteolytic release and subsequent degradation of HMG-CoA reductase from the endoplasmic reticular membrane is enhanced. The net effect is a decrease in both enzyme levels and metabolite flow through the pathway. This regulatory process is common to all mammalian cells.
However, bile acid production from cholesterol is unique to the mammalian liver, and it has been known for decades that bile acid feeding results in a similar down-regulation of cholesterol production in this organ (6). Traditionally, this has been attributed to the indirect effect that bile acids have through inhibiting CYP7A1, which would result in an increase in cholesterol followed by the predictable decline in mature SREBPs and HMG-CoA reductase activity.
Because the pathway leading to cholesterol in the liver corresponds to the "early" steps of bile acid synthesis, it would be more efficacious if a mechanism for the direct regulation of the early steps of the pathway directly through bile acids had evolved. In the current report, we present evidence supporting a direct mechanism for regulating cholesterol production by bile acids through the LRH-1/FTF and SHP nuclear receptors. We show that LRH-1/FTF activates and SHP represses HMG-CoA reductase transcription specifically, with no effect on LDL receptor expression. Overall, these studies define both a mechanism to independently regulate cholesterol synthesis from uptake, a key regulatory feature that has been documented in prior whole animal studies by Dietschy and co-workers (7). Additionally, these results also reveal a pathway for direct regulation of an early step in cholesterol biosynthesis by bile acids. This latter feature ensures that the early phase of bile acid synthesis (pre-cholesterol) is in metabolic communication with the later stages of the pathway to properly regulate whole pathway flux.

FLAG-tagged LRH-1/FTF and AF2 Domain Deletion (⌬AF-2)-PCR
oligos containing EcoRI and XbaI restriction enzyme sites were used to amplify human FTF (1-500 amino acids) from pCI FTF (a gift from Dr. Gregorio Gil, Virginia Medical College). A PCR oligo including base pairs to amino acid 466 of FTF along with the wild-type amino-terminal oligo were used to construct the AF-2 domain mutant. Digested PCR product was cloned into EcoRI-and XbaI-digested 2xFLAG pCDNA3.1 vector described previously (8). All sequences were confirmed by DNA sequencing. The construction of the FLAG-tagged SREBPs and documentation that the predicted fusion proteins are efficiently expressed in the transfected cells was also reported previously (8). CMV-SHP was from Dr. David Mangelsdorf (University of Texas Southwestern Medical Center) and CMV-HNF-4 was from Dr. Frances Sladek (University of California Riverside).
FTF DNA Binding Domain Mutant-The Stratagene QuikChange site-directed mutagenesis kit was used to introduce two point mutations, substituting an alanine for cysteine 1 in the "P-box" (9) of the DNA binding domain of FTF using the 2xFLAG FTF construct as template and oligos containing the relevant base pair change. The incorporation of the point mutation was confirmed by DNA sequencing. For glutathione S-transferase-FTF, PCR oligos containing EcoRI sites were used to amplify human FTF from pCI FTF, and the digested product was cloned into EcoRI-digested pGEX-2T. HMG-CoA reductase and LDL receptor promoter luciferase constructs have been described previously (10,11). The CYP8B1 promoter luciferase reporter construct (4) was a gift from Dr. Gregorio Gil (Virginia Medical College).
Mouse Studies and RNA Analyses-Wild-type and SHPϪ/Ϫ mice were housed, fed, and used in experiments as previously described (12), except as indicated in the figure legends. RNase protection assay and Northern blotting were performed with the indicated probes as previously described (12).
Immunoblotting Analysis of Protein Expression-These experiments were performed essentially as described previously (13). 293T cells were plated in normal medium (Dulbecco's modified Eagle's medium plus 10% (v/v) fetal bovine serum plus penicillin/streptomycin and glutamine) on day 0 at 450,000 cells/60-mm dish. On day 1, the cells were transfected with the appropriate plasmid construct along with a constant amount of CMV-␤-galactosidase expression plasmid. The cells were washed twice with 1ϫ phosphate-buffered saline on day 2 and refed with normal medium. On day 4, the cells were harvested into nuclear and cytoplasmic extract as described previously (13). Equal amounts of total protein normalized for transfection efficiency using co-transfected CMV-␤-galactosidase expression were analyzed by an SDS-polyacrylamide gel and transferred to nitrocellulose. Expression of specific proteins was detected by an antibody to the indicated epitope tag present.
Promoter Activation Studies-293T cells were plated on day 0 in normal medium at 350,000 cells/well of a 6-well plate. On day 1, the cells were transfected with luciferase reporter and protein expression plasmids by calcium phosphate co-precipitation. A CMV-␤-galactosidase expression construct was included in every transfection as a normalization control. 5 h post-transfection, the cells were washed twice with 1ϫ phosphate-buffered saline and new medium added. For transfections using exogenously expressed SREBPs, the cells were refed with normal medium as described above. For transfections using endogenously expressed SREBPs, cells were refed with either induced medium (defined serum-free medium from Invitrogen), or suppressed medium (induced medium containing 12 g/ml cholesterol and 1 g/ml 25-hydroxycholesterol) to suppress SREBP activity. 12 h after refeeding, the cells were harvested using cell lysis buffer (13), and cell extracts were used to measure activity for luciferase and ␤-galactosidase.
Recombinant FTF Protein Purification-Escherichia coli cells expressing glutathione S-transferase-FTF (4) were grown at 37°C to an OD of 0.6 and then induced with 1 mM isopropyl 1-thio-␤-D-galactopyranoside at 37°C for 3 h. The cells were harvested by sonication in NETN buffer (100 mM NaCl, 1 mM EDTA, 20 mM Tris, pH 8.0, 0.5% Nonidet P-40), and the soluble lysate was fractionated over a glutathione-agarose column and eluted with 10 mM glutathione. FTF protein fractions were identified by SDS-PAGE and Coomassie Blue staining, pooled, and dialyzed against 20 mM Tris, pH 8.0, 0.2 mM EDTA, and 50 mM KCl.
Electrophoretic Mobility Shift Assay-Single-stranded DNA oligos containing the potential FTF site at Ϫ300 were annealed for 1 h at 65°C and then end-labeled with 32 P for 1 h at 37°C. Purified FTF protein (25 ng) was incubated with 0.2 pmol of labeled HMG-CoA reductase probe on ice for 20 min and loaded onto a 5% acrylamide gel and run in 1ϫ Tris borate-EDTA for 1-2 h. The gel was fixed for half an hour in 10% (v/v) acetic acid and 10% (v/v) methanol, dried, and then exposed to film. For competition experiments, purified FTF protein (25 ng) was incubated on ice for 15 min with cold probe containing the indicated DNA sequence in 50-or 200-fold molar excess of 32 P-labeled HMG-CoA reductase probe. Labeled HMG-CoA reductase probe (0.2 pmol) was then added to the reactions, followed by a further incubation on ice for an additional 20 min. The reactions were loaded onto gels as described above.
Chromatin Immunoprecipitation Analysis-Chromatin immunoprecipitation was performed essentially as previously described (14) with the following minor modifications. 293T cells were transfected with an expression plasmid for Myc-tagged LRH-1 by Lipofectamine (Invitrogen). 5 h post-transfection, the cells were refed with defined serum-free medium as described above (minus sterols) to induce SREBP expression for 24 h. Formaldehyde cross-linking (1% (v/v)) was done for 9 min. After processing, the sonicated chromatin was obtained as described previously (14), and samples were preincubated with protein A-agarose beads and purified mouse IgG (50 g) for 1 h at 4°C on a rotator. Nonspecifically bound material was removed by pelleting the agarose beads, and supernatant fractions were incubated overnight at 4°C with 50 g of anti-LRH-1 antibody (Santa Cruz Biotechnology catalog number SC-25389x) for the ϩAb sample or 50 g of purified mouse IgG for the ϪAb sample, followed by incubation with blocked protein A beads for 2 h at 4°C on a rotator. After washing and reversing the cross-linking (14), the samples were analyzed by PCR. For the PCRs, 5 l of DNA from the LRH-1 precipitation was used, and PCR oligonucleotides that amplify a 250-bp fragment from the human HMG-CoA reductase or a 120-bp fragment from the human LDL receptor promoter were used, respectively. Amplification reactions were performed in triplicate at 30 cycles and monitored for amplification to ensure that the signals were in the linear range of the PCR. To analyze specific immunoprecipitation of Myc-tagged LRH-1, an immunoblot using an anti-Myc antibody (Santa Cruz Biotechnology catalog number SC-40) was performed on the material recovered after each immunoprecipitation.

RESULTS
Previous studies showed that bile acid-dependent inhibition of CYP7A1 is compromised in SHPϪ/Ϫ mice (12,15,16). Additionally, one of these studies also provided evidence that bile acid-dependent regulation of cholesterol metabolic genes, such as HMG-CoA reductase and the LDL receptor, might also be altered in SHPϪ/Ϫ mice (12). To more directly evaluate the effect of SHP on HMG-CoA reductase and LDL receptor gene expression, we compared their mRNA levels in wildtype and SHPϪ/Ϫ mice fed diets with and without cholic acid (CA) supplementation. The results in Fig. 1A demonstrate that treatment of wild-type mice with CA reduced the expression of mRNAs for both HMG-CoA reductase and the LDL receptor. Interestingly, in animals fed a normal chow diet, there was an increase in HMG-CoA reductase mRNA in SHPϪ/Ϫ compared with wild-type mice, and the suppression by CA feeding was blunted in the SHPϪ/Ϫ animals. In contrast, expression and CA suppression of LDL receptor mRNA were indistinguishable in wild-type and SHPϪ/Ϫ mice. These data suggest that SHP specifically inhibits expression of HMG-CoA reductase but not the LDL receptor.
Bile acid feeding induces SHP through the bile acid-activated nuclear receptor FXR, but bile acids also have pleiotropic effects. Therefore, to more directly evaluate SHP and FXR in the regulation of HMG-CoA reductase, we analyzed the effects of a synthetic FXR agonist on HMG-CoA reductase and LDL receptor expression in wild-type and SHPϪ/Ϫ mice (Fig. 1B). In wild-type animals, the synthetic FXR agonist GW4064 decreased HMG-CoA reductase expression, but the effect was lost in the SHPϪ/Ϫ animals (Fig. 1B, compare lanes 1, 2 and 5, 6 with lanes 3, 4 and 7, 8). In contrast, the FXR agonist had no effect on the expression of LDL receptor mRNA in either wild-type or SHPϪ/Ϫ animals. As an additional control, similar administration of a synthetic agonist for retinoid X receptor LG00268 had no effect on mRNA levels for either HMG-CoA reductase or the LDL receptor (lanes 9 -12).
These results suggest that SHP specifically inhibits HMG-CoA reductase expression. For known SHP target genes, such as CYP7A1 and CYP8B1, inhibition occurs by interfering with the activation by the nuclear receptor LRH-1/FTF. To determine whether a similar mechanism was functioning for HMG-CoA reductase, we first evaluated whether LRH-1/FTF could bind to the endogenous HMG-CoA reductase promoter in cellular chromatin using a chromatin immunoprecipitation assay. An LRH-1 expression vector was transfected into 293 cells, and formaldehyde cross-linked chromatin was treated with control IgG or with an antibody to LRH-1. The LRH-1 antibody did precipitate the LRH-1 protein specifically (immunoblot in Fig. 2), and the DNA associated with the immunoprecipitation pellets was analyzed by PCR for the presence of the promoters for either HMG-CoA reductase or the LDL receptor as a control. The PCR results demonstrated that LRH-1 protein bound specifically to the endogenous HMG-CoA reductase promoter, and it was not associated with the LDL receptor promoter chromatin (Fig. 2).
In transient transfection assays, the activation of HMG-CoA reductase by SREBPs was always significantly lower compared with other target genes analyzed in parallel, suggesting there was some additional Human embryonic kidney 293T cells were transfected with an expression vector for LRH-1/FTF and processed for chromatin immunoprecipitation analyses as described under "Materials and Methods." A, binding to HMG-CoA reductase promoter. Oligonucleotides for the human HMG-CoA reductase (HMGR) promoter were used in PCR analyses. The input titration represents serial 3-fold dilutions of the input DNA performed in duplicate. After the immunoprecipitation with a control IgG or LRH-1 (LRH IP) antibody, the recovered material (5 l) was analyzed by PCR and resolved by neutral polyacrylamide gel electrophoresis and stained with ethidium bromide. The size for the HMG-CoA reductase promoter PCR product is 256 bp. Pictures of the gels are shown at the top, sample intensities (S.I.) were measured using quantity one software (Bio-Rad), and duplicate values were averaged and plotted on the graph below the gel pictures. Error bars are included to indicate the range for the two duplicate values. B, same as described for A, except that oligonucleotides designed to amplify the human LDL receptor (LDLR) promoter were used. Shown are input (in) or samples after the immunoprecipitation protocol were also analyzed for protein recovery using the LRH or control IgG. A picture of the developed immunoblot is also shown. 5% of the input was used for the analysis, and 1% of the immunoprecipitation samples was used. Wild-type (ϩ/ϩ) and SHP null mice (Ϫ/Ϫ), 5 animals/group, were fed a control diet (Con) or a diet supplemented with 0.5% cholic acid (CA) for 12 weeks. Total liver RNA was pooled from all mice in each group, and 20 g were used in a standard RNase protection assay as described previously (12). Red, HMG-CoA reductase. LDLR, low density lipoprotein receptor. The graphs show the normalized RNA levels relative to the actin signal, and the wild-type control-fed group was set at 1.0. Wt, wild-type; KO, knock-out. B, HMG-CoA reductase regulation by the FXR ligand is impaired in SHPϪ/Ϫ mice. Shown is a Northern blot. Two wild-type (ϩ/ϩ) or two SHP (Ϫ/Ϫ) mice were fed either a control (CON) diet or a diet supplemented with the FXR or retinoid X receptor (RXR) agonist by oral gavage for 1 day (12). RNA from each animal (20 g) was resolved in separate lanes and probed with 32 P-labeled HMG-CoA reductase (Red) or LDLR cDNA probes. The band representing HMG-CoA reductase RNA levels shows a slight migration anomaly across the gel, likely because of a small current imbalance as different samples migrated through the gel.

Direct Bile Acid Regulation of HMG-CoA Reductase Promoter
protein required that was missing (17). Based on the results presented above, we reasoned that this missing protein might be LRH-1/FTF. To test this idea, we performed a transient transfection assay in 293 cells using a culture protocol that activates the processing of endogenous SREBPs (14). Here, companion dishes of transfected cells are cultured with medium containing or lacking regulatory sterols, and endogenous SREBPs are cleaved from their membrane location and accumulate in the nucleus in the sterol-depleted samples (18,19).
Consistent with our earlier studies, the HMG-CoA reductase promoter was activated ϳ2-fold by this sterol depletion protocol (Fig. 3 compare lane 5 with 6). When an expression plasmid for LRH-1/FTF was co-transfected under sterol-depleted conditions, there was a significant increase in expression of the HMG-CoA reductase promoter (Fig.  3, lane 7), and this stimulation was specifically inhibited when an SHP expression plasmid was added on top of the LRH-1/FTF construct (lanes 8 -9). However, transfection of the SHP expression plasmid alone had no effect on the modest activation by endogenous SREBPs (Fig. 3, compare lane 6 with 10 and 11). This result suggests that SHP does not inhibit SREBP-mediated activation but only affects the promoter stimulation mediated by LRH-1/FTF.
For controls, we also analyzed the LDL receptor and CYP8B1 promoters (Fig. 3, lanes 1-4 and 12-17). Similar to our previous studies (17), the sterol depletion protocol resulted in a higher degree of activation of the LDL receptor promoter, and consistent with the studies in the SHPϪ/Ϫ mice, there was no effect of LRH-1/FTF or SHP on LDL receptor promoter activity. However, as a positive control, LRH-1/FTF addition stimulated the CYP8B1 promoter, and this was inhibited by the addition of SHP.
In the above experiments using sterol depletion, all three SREBPs were released from the membrane and accumulated in the nucleus; therefore, it was unclear whether LRH-1/FTF functions to enhance the activation of all three SREBPs or whether there is a preference for one of the three SREBP isoforms. To address this, we performed transient promoter activation experiments in cells cultured in the presence of exogenous sterols to suppress the activation of endogenous SREBPs. Addi-tionally, we transfected expression plasmids for each of the mature SREBP isoforms, alone or together with the LRH-1/FTF expression construct (Fig. 4A). Transfection of either SREBP-1a or -2 expression vectors resulted in an ϳ2-fold activation of the HMG-CoA reductase promoter luciferase reporter, and in each case, the addition of the LRH-1/ FTF construct further stimulated promoter activity significantly. The addition of the SREBP-1c expression plasmid alone had no effect on the   1 g) for the individual SREBP isoforms (BP-2, BP-1a, BP-1c) alone or in combination with expression vectors for LRH-1/FTF or SHP as indicated. Luciferase expression was normalized to an internal control ␤-galactosidase expression plasmid, and fold activation was calculated as in Fig. 3. B, 293T cells were transfected as in Fig.  3 and in A, except an expression construct for HNF-4 was included in the place of LRH-1/FTF (0.1 g). 293T cells were transfected with a luciferase reporter construct for HMG-CoA reductase and expression constructs for SREBP-2 and LRH-1/FTF or ⌬AF-2, as indicated, along with the internal control CMV-␤galactosidase. Fold activation was calculated as described in the other figure legends. An immunoblot for wild-type and ⌬AF-2 expression using equal amounts of total cell protein (25 g) and an antibody to the FLAG epitope is shown in the inset. B, the wild-type (wt) and DNA binding domain mutant (DBDm) were individually transfected into 293T cells with luciferase reporter plasmids for HMG-CoA reductase (RED) or CYP8B1 and cultured in the presence (ϩ) or absence (Ϫ) of regulatory sterols to activate the endogenous SREBPs. Fold regulation was calculated as described in the legend to Fig. 3. Equal amounts of whole cell protein (25 g) were analyzed for protein expression from the DNA binding domain mutant construct and wt LRH-1/FTF using an immunoblotting procedure with the FLAG epitope antibody and is shown at the top of the figure. N corresponds to a sample analyzed from mock-transfected cells.
promoter by itself, consistent with previous reports where SREBP-1c is a weak activator compared with SREBP-1a or -2 (8). However, the addition of the LRH-1/FTF expression construct resulted in a 3-fold stimulation. This was consistently above the small stimulation that resulted by the addition of the LRH-1/FTF expression plasmid alone (Fig. 4A,  compare lanes 8 and 9 with 11).
Additionally, regardless of which SREBP was analyzed, the addition of the SHP expression construct inhibited only the LRH-1/FTF-mediated effect, because the magnitude of activity after repression by SHP was equal to that stimulation by each SREBP alone. These results indicate that LRH-1/FTF can function with all three SREBPs to activate the HMG-CoA reductase promoter and that SHP inhibition only affects the LRH-1/FTF stimulatory effect.
Because SHP is known to inhibit activation by other nuclear receptors, such as HNF-4 (20), and because overexpression of transcription factors in transient assays may exaggerate normal physiological effects, as a control, we analyzed whether HNF-4 could activate the HMG-CoA reductase promoter along with SREBPs (Fig. 4B). Transfection of an HNF-4 expression construct in place of LRH-1/FTF had no effect on SREBP-dependent activation of the HMG-CoA reductase promoter, whereas it efficiently activated the HNF-4 target gene CYP8B1 (Fig. 4B).
LRH-1/FTF, similar to other nuclear receptors, requires its carboxylterminal AF-2 activation domain and a zinc finger DNA binding motif to activate target genes. To determine whether these critical functions are required for activation of HMG-CoA reductase, we deleted the AF-2 domain or introduced a point mutation at a critical cysteine residue of the DNA binding domain P-box (9) to alanine to inhibit DNA binding. Despite the fact that both of these mutant proteins were expressed efficiently in the transfected cells, neither one was able to activate the HMG-CoA reductase promoter like the wild-type protein (Fig. 5). Thus, both of the crucial nuclear receptor functional domains are required and suggest that LRH-1/FTF likely binds directly to the HMG-CoA reductase promoter.
In scanning the DNA sequence of the HMG-CoA reductase promoter used in these studies, we noted two putative recognition sites that are conserved between the hamster, mouse, and human promoters (Fig.  6A). To test whether these sites are important for LRH-1/FTF-mediated activation, we deleted them from the luciferase reporter construct. The activation studies shown in Fig. 6B show that deletion of the two sites resulted in a severe blunting of LRH-1/FTF-mediated activation, but these truncations had little effect on overall promoter activity or on stimulation by SREBPs alone (Fig. 6B) (21). Next, the DNA site at Ϫ300 was tested for DNA binding directly using recombinant LRH-1/FTF protein in an electrophoretic mobility shift assay (Fig. 6C). Recombinant LRH-1/FTF bound to this HMG-CoA reductase promoter site specifically, and a mutation that changed the sequence away from the predicted consensus recognition site failed to compete efficiently for binding. Additionally, oligonucleotides containing the LRH-1/FTF site from the CYP8B1 promoter or the cold wild-type HMG-CoA reductase DNA oligos used for the electrophoretic mobility shift assay competed efficiently for binding. Thus, DNA binding directly to the HMG-CoA reductase promoter is required for the LRH-1/FTF stimulatory effect.

DISCUSSION
In our previous studies, we noted that the magnitude of stimulation by sterol depletion or the addition of exogenous SREBP expression constructs was very modest for the HMG-CoA reductase promoter compared with the activation achieved with other SREBP target genes ana- showing the positions of the two putative LRH-1 sites relative to already characterized sites for SREBP, nuclear factor-Y (NF-Y), and cAMP-response-element binding protein (CREB) (10). At the bottom is a lineup of the genomic DNA from the Ϫ300 region of the hamster, human and mouse promoters showing the conservation of this putative LRH-1 site. The human site was also shown to bind directly to the LRH-1/FTF protein (S. Datta and T. F. Osborne, unpublished data). B, the wild-type and two deletion reporter constructs for the HMG-CoA reductase promoter (see A, B, and C in panel A) were analyzed for activation by SREBP-2 and LRH-1/FTF as indicated and as described in the previous figure legends. C, purified recombinant FTF (25 ng) was used in an electrophoretic mobility shift assay with 32 P-labeled probe containing the putative LRH-1/FTF response element from Ϫ300 in the HMG-CoA reductase promoter, and where indicated (lanes 3-8), a molar excess (50 or 200ϫ) of unlabeled, competitor (Comp.) DNAs were included in the binding reaction as described under "Materials and Methods." wt, wild-type hamster HMG-CoA reductase LRH-1/FTF site; mt, HMG-CoA reductase mutant competitor with a single base mutation that alters the putative LRH-1/FTF site response element; 8B, oligonucleotides containing a known LRH-1/FTF site promoter, the mouse CYP8B1 promoter.

Direct Bile Acid Regulation of HMG-CoA Reductase Promoter
lyzed in parallel (17). In contrast, HMG-CoA reductase gene expression was activated very robustly when SREBPs were overexpressed in mice (22). Although there might be additional differential post-initiation regulatory actions on the mRNAs that may partially account for these differences, the results are also consistent with a model where an additional protein was missing in our transient transfection assays for the HMG-CoA reductase promoter.
The current studies were initiated when we noted that regulation of HMG-CoA reductase was aberrant in SHPϪ/Ϫ mice. Additional studies presented here further support this idea and suggest that LRH-1/FTF is the missing protein. Additionally, the SHP effect exhibits specificity for HMG-CoA reductase, because neither the FXR agonist nor SHP itself had any effect on LDL receptor expression in any of the assays we utilized. Thus, the direct regulation through bile acids and SHP (Fig. 7, left) is specific to the cholesterol synthetic pathway. Our studies have focused on HMG-CoA reductase, because it is considered the classic rate-controlling enzyme of the pathway and because our earlier studies suggest there was a missing component in our transient expression assays. Whether additional enzymes of the pathway are similarly regulated remains to be determined.
When bile acid levels rise, CYP7A1 is inhibited and pathway flux is repressed. Without any alteration to earlier steps in the pathway, cholesterol levels would rise, which would inhibit SREBP maturation (Fig. 7, right). However, because the LRH-1/FTF activation of HMG-CoA reductase is inhibited by SHP, the studies presented here provide the first evidence that, in addition to the indirect effect of bile acids through cholesterol, they also have a direct inhibitory action on the expression of HMG-CoA reductase (Fig. 7, left). This indicates that the early and late sectors of the pathway are in metabolic communication with each other to more quickly adapt to changes in pathway influx and outflow. There is a similar regulatory mechanism in glycolysis, where the product of phosphofructokinase, fructose 1,6 bisphosphate, which measures early flux into the pathway, is a positive regulator of pyruvate kinase, which controls pathway outflow (23).
Spady et al. (7) report that endogenous cholesterol biosynthesis and cholesterol uptake through the LDL receptor pathway are independently regulated in the livers of mice. In these studies, the addition of the bile acid sequestrant cholestyramine to the diet significantly increased hepatic sterol synthetic rates, whereas LDL clearance rates were not altered relative to control fed mice. Cholestyramine reduces bile acid reabsorption and would effectively deplete the endogenous hepatic FXR agonist pool, which would be predicted to decrease SHP levels. In fact, we have documented that SHP expression is repressed by feeding mice a similar bile acid sequestrant. 4 In another study (24), Sheng et al. mentioned that feeding mice a bile acid sequestrant alone was ineffective at increasing nuclear levels of SREBPs in mice.
Taken together, these two studies indicate that the mechanism by which bile acid sequestrants increase sterol biosynthesis cannot solely be explained by an increase in nuclear SREBP levels. The current studies demonstrating that expression of HMG-CoA reductase is activated by LRH-1/FTF and repressed by SHP, without any change in LDL receptor expression, provides a reasonable molecular explanation that connects both of these important earlier studies together.
The LRH-1/FTF DNA sites are conserved in the human HMG-CoA reductase promoter. Therefore, our results also suggest that FXR agonists might be effective when combined with statins to treat hypercholesterolemia in humans. Statin therapy results in a compensatory upregulation of HMG-CoA reductase gene expression as the liver attempts to compensate for the decreased sterol production. The addition of an FXR ligand may work synergistically with statins to prevent this response through inhibiting activation by LRH-1/FTF.