Differential effects of sterol regulatory binding proteins 1 and 2 on sterol 12α-hydroxylase: SREBP-2 suppresses the sterol 12α-hydroxylase promoter

The most important pathway for the catabolism and excretion of cholesterol in mammals is the formation of bile acids. Improper regulation of this pathway has implications for atherosclerosis, cholesterol gallstone formation, and some lipid storage diseases. Sterol 12α-hydroxylase (12α-hydroxylase) is required for cholic acid biosynthesis. The α1-fetoprotein transcription factor FTF is crucial for the expression and the bile acid-mediated down-regulation of 12α-hydroxylase. Cholesterol, on the other hand, down-regulates expression of the 12α-hydroxylase gene. In this study, we show that the two sterol regulatory binding proteins (SREBPs) have opposite effects on the 12α-hydroxylase promoter. SREBP-1 activated the 12α-hydroxylase promoter, as it does with many other cholesterol-regulated genes. In contrast, SREBP-2 suppressed 12α-hydroxylase promoter activity. SREBP-1 mediates the cholesterol-down-regulation of 12α-hydroxylase promoter by binding to two inverted sterol regulatory elements found ∼300 nucleotides from the transcriptional initiation site. SREBP-2 mediated suppression of 12α-hydroxylase without binding to its promoter. Data are presented suggesting that SREBP-2 suppresses the 12α-hydroxylase promoter by interacting with FTF. This is the first report of a promoter responding oppositely to two members of the SREBP family of transcription factors. These studies provide a novel function and mode of action of a SREBP protein.

Cholesterol conversion to bile acids occurs via the "classic" (neutral) or the "alternative" (acidic) bile acid biosynthesis pathways (4). Cholic acid and chenodeoxycholic acid are the end products of these pathways and the major primary bile acids found in most vertebrates. Cholic acid is hydroxylated at position 12␣, whereas chenodeoxycholic acid is not. Sterol 12␣hydroxylase/CYP8B1 (12␣-hydroxylase) 1 is the specific enzyme for cholic acid synthesis and determines the ratio of cholic acid to chenodeoxycholic acid and thus the hydrophobicity of the circulating pool. This ratio of cholic to chenodeoxycholic acid has been postulated to play a role in cholesterol gallstone formation (5). Suppression of 12␣-hydroxylase by specific inhibitors has been suggested as a possible therapeutic strategy for dissolution of cholesterol gallstone (5). The alteration of cholic to chenodeoxycholic acid ratio affects biliary cholesterol and phospholipid secretion, thus altering intestinal cholesterol absorption and receptor-mediated lipoprotein uptake by the hepatocyte (6).
Bile acids negatively regulate the transcription of the cholesterol 7␣-hydroxylase (7␣-hydroxylase) gene that controls output from the classic pathway. Recent studies have delineated many of the factors involved in this regulation. Liver receptor homolog-1, also known as cytochrome P450 7␣Ϫhydroxylase promoter-binding factor, NR5A2 (7), or ␣ 1 -fetoprotein transcription factor (FTF) (8), has been proposed to be required for the transcription of the 7␣-hydroxylase gene (9,10). Bile acids activate the transcription of the gene encoding the small heterodimer partner 1 (SHP) via binding to the hormone receptor farnesoid X receptor. In turn, SHP has been proposed to dimerize with FTF and diminishes its activity on the 7␣-hydroxylase promoter by mechanisms that are not well understood (11,12).
Expression of the 12␣-hydroxylase gene is down-regulated by bile acids (13)(14)(15). We have recently shown that FTF is required for 12␣-hydroxylase promoter activity (16,17). FTF binds to two sites within the 12␣-hydroxylase promoter, and both sites are required for both promoter activity and bile acid-mediated regulation (16). We have also shown that SHP displaces FTF binding to its sites within the 12␣-hydroxylase promoter (17). Interestingly, cholesterol regulates these two key enzymes of the bile acid biosynthetic pathways in different directions. Specifically, 7␣-hydroxylase is activated upon cholesterol feeding (18), whereas 12␣-hydroxylase is repressed (15).
In addition to the bile acid biosynthetic pathways, two other pathways are crucial in maintaining cholesterol homeostasis in mammals (i.e. the cholesterol biosynthetic pathway and the low density lipoprotein (LDL) receptor pathway). In the last several years, a family of transcriptional factors known as sterol regulatory element-binding proteins (SREBPs) have been shown to play crucial roles in controlling not only these two pathways but also adipogenesis, fatty acid biosynthesis, and insulin action (for reviews, see Refs. 19 and 20). SREBPs are transcriptional factors that are activated by proteolytic release from membranes. This occurs in sterol-depleted cells by two sequential proteolytic cleavages (21). Three different SREBPs have been described. SREBP-1a and -1c, are produced from a single gene through the use of alternate promoters that produce transcripts with different first exons. The third isoform, SREBP-2, is produced from a separate exon (19). The SREBPs are threedomain proteins of ϳ1150 amino acids. The NH 2 -terminal domain of ϳ480 amino acids and the COOH-terminal domain of ϳ590 amino acids project into the cytosol. The third domain of ϳ80 amino acids anchors them to the endoplasmic reticulum membrane. The NH 2 -terminal domains of SREBPs are transcriptional factors of the basic loop-helix-leucine zipper (bHLH-Zip) family (22). The NH 2 -terminal domains of all three SREBPs also contain a bHLH-Zip motif that mediates dimerization, nuclear entry, and DNA binding.
An important unresolved question is the specificity of these three SREBPs for regulating different genes. When SREBPs were first isolated, there was no evidence to suggest a functional difference between the three. Overexpression of the mature SREBPs (amino acids 1-480) in tissue culture cells also suggested that there was no functional difference, because all three SREBPs were able to activate genes that contained sterol regulatory elements (SREs) in their promoters. Later it was discovered that at lower levels of expression, approximately equal to physiological range, different SREBPs activate the same family of genes, but they do so in different proportions. In general, SREBP-2 appears to regulate genes involved in the cholesterol biosynthetic pathway, while SREBP-1a has a relatively greater effect on genes involved in fatty acid biosynthesis (23).
The addition of excess cholesterol has been shown to inhibit the processing of membrane-bound precursor forms of both SREBP-1 and -2 in experiments performed in both animals (24) and culture cells (25). However, when hamsters were fed a diet supplemented with a cholesterol lowering diet, both expression and proteolytic activation of SREBP-2 were increased (26). In contrast, expression of SREBP-1 was not altered, nor was processing of membrane-bound SREBP-1 accelerated in these animals. In fact, SREBP-1 processing was inhibited (26). Other experiments showed that SREBP-1c seems to be specifically involved in fat cell differentiation and lipid accumulation (27), and it also mediates insulin action in the liver (28). To date, there is only one study reporting that SREBPs mediate inhibition of gene transcription. In this case, microsomal triglyceride transfer protein gene transcription is suppressed by both SREBPs (29).
In the present study, we have shown that SREBP-1 and -2 have opposite effects on the 12␣-hydroxylase promoter. SREBP-1 activates the 12␣-hydroxylase promoter, as expected for a cholesterol-suppressed gene (15). Most significant, SREBP-2 suppresses the 12␣-hydroxylase promoter, and it does so independent of two SRE sites that we have located and characterized within the 12␣-hydroxylase promoter. This SREBP-2-mediated suppression of 12␣-hydroxylase promoter activity is also observed by manipulating the expression and processing of the endogenous SREBPs with an HMG-CoA re-ductase inhibitor. Additionally, we show evidence to support that SREBP-2 physically interacts with FTF and, as a result, suppresses the 12␣-hydroxylase promoter by inhibition of FTF activity, which is essential for 12␣-hydroxylase promoter activity. These observations suggest novel functions for SREBPs.

EXPERIMENTAL PROCEDURES
Materials-Reagents used in DNA cloning and sequencing were from New England Biolabs, Roche Molecular Biochemicals, U.S. Biochemical Corp., or Invitrogen. Common laboratory chemicals were from Fisher, Sigma, or Bio-Rad. The luciferase promoterless vector, pGL3-Basic, was purchased from Promega. Oligonucleotides were prepared in the Medical College of Virginia DNA Synthesis Facility by the phosphoramidite method on an automated DNA synthesizer. pGL3/HMG-CoA Syn was made from pSyn CAT-1 (30), and it contains 368 nucleotides of the hamster HMG-CoA synthase promoter. pCS-A10 and pCS-2 contains the human SREBP-1a cDNA (amino acids 1-490) or the human SREBP-2 cDNA (amino acids 1-480), respectively, in the expression vector pCMV, and were a generous gift from Dr. Osborne (University of California, Irvine). pCDM8/mSHP, an expression plasmid that contains the full-length mouse SHP, was a gift from Dr. Moore (Baylor College of Medicine, Houston). pGEX4T3/hFTF was prepared by inserting the coding region for amino acids 1-501 of hFTF into pGEX4T3. Frame was checked by sequence analysis in all cases. pGL3/R12␣-865 and deletion mutant pGL3/R12␣-289 were prepared as described elsewhere (16).
General Methods-Standard recombinant DNA procedures were carried out essentially as described (16).
Transient Transfection and Luciferase Assays-HepG2 and CV-1 and McA-RH7777 cells were obtained from the American Type Culture Collection. 17-mm plates were used for transfections. HepG2 and McA-RH7777 cells were transfected with calcium phosphate using 2.0 g of total DNA, and CV-1 cells were transfected with Lipofectin (Invitrogen) using 750 ng of total DNA. HepG2 was transfected with 500 ng of the corresponding test plasmid, 5 ng of pCMV-Gal, and the indicated expression plasmid. After transfection, cells were incubated in fetal bovine serum-containing medium to suppress the processing of endogenous SREBPs. McA-RH7777 cells were transfected with 100 ng of test plasmid, and after transfection cells were incubated in lipoproteindeficient serum-containing medium in the presence or absence of 50 M mevinolin, supplemented with 50 M sodium mevalonate to provide the cells with the required nonsterol mevalonate-derived products. CV-1 cells were transfected with 25 ng of the corresponding test plasmid, 20 ng of pCMV-Gal, and the indicated amounts of pCS-A10 or pCS-2. After 16 h, the DNA was removed, and fresh fetal calf serum-containing medium was added. Cells were harvested 48 h later, and luciferase and ␤-galactosidase assays were performed with a kit from Tropix (Bedford, MA) according to the manufacturer's protocol. Average values are for the number of experiments indicated.
Expression and Preparation of GST Fusion Proteins-pGEX4T3/ hFTF contains the entire FTF-coding region placed in frame with the glutathione S-transferase (GST) coding region. Both GST alone and the GST-FTF fusion protein were expressed in E. coli BL21DE3 and induced with 2 mM isopropyl-1-thio-␤-D-galactopyranoside for 12 h at 37°C. Cells were harvested and lysed by sonication. Proteins were purified on glutathione-agarose and dialyzed against binding buffer (25 mM Hepes, pH 7.6, 100 mM NaCl, 20% glycerol, and protease inhibitors) and stored at Ϫ70°C until used.
Protein-Protein Interaction Assay-[ 35 S]Methionine-labeled target proteins were produced by expressing in vitro constructs containing either the entire SHP coding region or nucleotides encoding amino acids 1-480 from the human SREBP-2 using the TnT T7-coupled rabbit reticulocyte lysate system according to the manufacturer's protocol (Promega). E. coli expressed GST or GST-FTF and the 35 S-labeled proteins were incubated in binding buffer. The removal of nonspecifically bound proteins was performed by standard methods (32). Specifically bound proteins were eluted with 15 mM glutathione and analyzed by SDS-PAGE followed by PhosphorImager (Molecular Dynamics, Inc., Sunnyvale, CA) analysis.

RESULTS
We recently localized and characterized two overlapping FTF sites within the 12␣-hydroxylase promoter (16). Further examination of the 12␣-hydroxylase promoter sequence revealed two potential SRE elements in a reversed orientation. Fig. 1 shows an abbreviated 12␣-hydroxylase promoter sequence highlighting these elements. Fig. 2 shows an alignment of these two SREs with other well characterized SREs found in the human LDL receptor (33) and the hamster HMG-CoA synthase (30). We characterized the potential function of the two 12␣-hydroxylase SREs by performing electrophoresis mobility shift assays with a DNA probe containing both sites using human recombinant SREBP-1a and SREBP-2 as protein source. Fig. 3A shows that both SREBPs bind to the 12␣hydroxylase probe (lanes 1 and 8). Specificity of this binding was shown by competition with a probe containing the LDL receptor promoter SRE (lanes 2, 3, 9, and 10), the same probe (lanes 4, 5, 11, and 12), and the same 12␣-hydroxylase probe with the two SRE mutated (lanes 6, 7, 13, and 14). The wild type 12␣-hydroxylase probe competed better than the LDL probe, suggesting that the 12␣-hydroxylase promoter has higher affinity for SREBPs than the LDL receptor promoter, probably due to the existence of two SREs in the 12␣-hydroxylase promoter. To demonstrate that both 12␣-hydroxylase SREs bind SREBPs, each site was mutated independently and used in electrophoresis mobility shift assay experiments. Fig.  3B shows that individual mutation of each SRE diminished but did not abolish binding. Both sites needed to be mutated to completely abolish binding to either SREBP-1a or -2 (lanes 5 and 10) in agreement with the direct binding experiments (Fig.  3A). These two SREs here localized are the only binding sites for SREBPs that we could detect. We could not find binding of SREBPs to the FTF sites within the 12␣-hydroxylase sites (Ϫ63 to Ϫ48) (data not shown).
To study the in vivo functionality of the 12␣-hydroxylase SRE sites, we cotransfected HepG2 cells with a 12␣-hydroxylase promoter/luciferase reporter gene and an expression plas-mid containing a cDNA encoding the mature form of either SREBP-1a or SREBP-2. Fig. 4A shows that SREBP-1a activates the 12␣-hydroxylase promoter in a concentration-dependent manner, as expected for a SRE containing promoter. Sur-   2-7 and 9 -14) and compared with binding in the absence of competitor (lanes 1 and 8). 12␣M, the SRE-1&2 double mutant referred to as 12␣M-1&2 under "Experimental Procedures." The arrow points to the retarded band. The gel corresponding to SREBP-2 was exposed for double the length of time as SREBP-1. B, five different probes were used, the SRE site from the LDL receptor as positive control (lanes 1 and 6), the wild type 12␣hydroxylase sequence from nucleotides Ϫ338 to Ϫ303 (lanes 2 and 7), and the three indicated mutants (lanes 3-5 and 8 -10).
prisingly, however, SREBP-2 had an opposite effect and suppressed the 12␣-hydroxylase promoter up to 5-fold. To ensure the functionally of the SREBP expression plasmids, we performed the same experiment but using the HMG-CoA synthase promoter (30). This promoter is known to be activated by both SREBP-1 (34) and SREBP-2 (35). Fig. 4B shows that the mature form of both SREBPs activated the HMG-CoA synthase promoter in the same system where we observed the opposite effect of SREBP-1 and -2 on the 12␣-hydroxylase. Background values (pGL3) were not affected by either SREBP.
To study whether the effect of SREBPs could be observed in a nonliver cell line, we performed similar overexpression experiments in CV-1 cells. As we reported earlier (16), the 12␣hydroxylase promoter is inactive in CV-1 cells when no transcription factor is cotransfected (Fig. 5A). Overexpression of SREBP-1a activated the 12␣-hydroxylase promoter in a concentration-dependent manner up to 40-fold (Fig. 5A). However, overexpression of SREBP-2 showed very slight, if any, activation. In control experiments, both SREBPs activated the HMG-CoA synthase promoter up to 25-fold (Fig. 5B), as expected.
To study whether the two SREs located between Ϫ331 and Ϫ308 in the 12␣-hydroxylase promoter are involved in the SREBPs effects, we studied the effect of SREBP overexpression on a deletion promoter construct at Ϫ289. This shorter promoter was not activated by SREBP-1a, but SREBP-2 had the same suppression on the deleted construct and the wild type (Fig. 6), which indicates that this SREBP-2-mediated suppres-sion of 12␣-hydroxylase promoter does not require binding to the SREs and that the DNA element(s) mediating that suppression is within Ϫ289 nt of the transcriptional initiation site. We then mutated the two potential SREs shown in Fig. 1. Data in Fig. 7 show that both SREs mediate the activation of 12␣- hydroxylase by SREBP-1a. In addition, either of those two sites mediates the activation, since both SREs need to be mutated to eliminate the SREBP-1a activation. We performed similar experiments with SREBP-1c and found a similar effect as 1a, except that the activation through the two SREs was weaker (data not shown).
We attempted to localize the DNA element mediating the SREBP-2 suppression of 12␣-hydroxylase promoter activity by using the same deletion constructs and block mutation used to localize the FTF element (16). We co-transfected each of those mutants into HepG2 cells together with pCS2. All 12␣-hydroxylase promoter mutants were equally suppressed upon overexpression of the SREBP-2 nuclear form except mutants that had the FTF site mutated, which were inactive, and consequently they could not be suppressed (data not shown). These results suggested that SREBP-2 suppresses 12␣-hydroxylase promoter activity through the FTF element, but no SREBP-2 binding could be detected (data not shown), suggesting that SREBP-2 could cause suppression by a direct interaction with the FTF protein.
We have shown that the 12␣-hydroxylase promoter can be activated in nonliver cells, such as CV-1, by overexpressing FTF (16). To test the hypothesize that SREBP-2 suppresses the 12␣-hydroxylase by interacting with FTF, we activated the 12␣-hydroxylase promoter in CV-1 cells by cotransfecting it with FTF. Then we included increasing amounts of either SREBP-1a or -2 expression plasmids in the transfection. Fig.  8A shows that expression of FTF activated the 12␣-hydroxylase promoter ϳ12-fold. Most important, expression of SREBP-2 suppressed the FTF-mediated activation of the 12␣-hydroxylase promoter up to 5-fold. SREBP-1a overexpression did not suppress the FTF-activated 12␣-hydroxylase promoter but actually had a synergistic effect (Fig. 8B) as expected, since SREBP-1a by itself activates the 12␣-hydroxylase promoter (Fig. 5A). As a further control, we observed no effect of SREBP-2 on the SREBP-1a-activated 12␣-hydroxylase promoter (Fig. 8C). The SREBP-2-mediated suppression of 12␣hydroxylase promoter was overcome by transfecting higher levels of FTF (Fig. 9).
Recently, it has been shown that cultured rat hepatoma cells (McA-RH7777) mimic the expression pattern of rat liver by expressing SREBP-1c as opposed to SREBP-1a (36). The high level of SRBP-1c expression is abolished when cholesterol synthesis is blocked by an HMG-CoA reductase inhibitor. This is due to the requirement of LXR and its ligands for the expression of SREBP-1c (37,38). Because of the low intracellular cholesterol levels when cells are incubated in the presence of an HMG-CoA reductase inhibitor, processing of SREBP-2 is enhanced (36), thus creating a situation in terms of mature SREBPs similar to that when mature SREBP-2 is overexpressed (i.e. high levels of SREBP-2 and virtually nonexistent mature SREBP-1) (36). When we incubated McA-RH7777 cells in the presence of 50 M mevinolin, 12␣-hydroxylase promoter activity was decreased by approximately 2-fold (Fig. 10). As a control, we showed that under the same conditions HMG-CoA FIG. 8. SREBP-2, but not SREBP-1, suppresses FTF-dependent, but not SREBP-1-dependent, activation of the 12␣-hydroxylase promoter. A, CV-1 cells were cotransfected with the 12␣-hydroxylase wild-type promoter and the indicated amounts of pCMV-FTF and pCS-2 (mature SREBP-2). B, CV-1 cells were cotransfected with the 12␣-hydroxylase wild-type promoter and the indicated amounts of pCMV-FTF and/or pCS-A10 (mature SREBP-1a) and/or pCS-2 (mature SREBP-2). Relative transcription was determined by normalizing luciferase activity to ␤-galactosidase activity. The data were normalized to the activity of the 12␣-hydroxylase promoter in the absence of any expression plasmid. Values represent the averages of three experiments Ϯ S.D. synthase promoter was activated more than 2-fold, since it is activated by SREBP-2 (35).
The experiments shown above suggest that SREBP-2 suppresses 12␣-hydroxylase promoter through FTF, a required transcription factor for 12␣-hydroxylase expression. To demonstrate that SREBP-2 and FTF physically interact, we performed GST pull-down assays between 35 S-labeled SREBP-2 and a GST-FTF fusion protein. Fig. 11 shows that SREBP-2 and FTF specifically interact (lanes 5 and 6). This interaction is not as strong as the interaction between FTF and SHP (11,12) that we used as a positive control. DISCUSSION The data presented in this study provide evidence for an opposite effect of SREBP-1 and -2 on the 12␣-hydroxylase promoter. SREBPs are transcriptional activators that have been revealed to play crucial roles in controlling not only the choles-terol biosynthetic and uptake pathways but also adipogenesis, fatty acid synthesis, and insulin action (for reviews, see Refs. 19 and 20). In all cases studied to date, overexpression of the mature SREBPs activate transcription of all genes regulated by SREBPs except for the microsomal triglyceride transfer protein gene, where all SREBPs act as negative regulators of its transcription (29). The 12␣-hydroxylase promoter, reporter here, is the first one that has been shown to be regulated in the opposite manner by different members of the SREBP family.
The existence of two inverted SREs in the rat 12␣-hydroxylase promoter (Figs. 1 and 2) suggested that these sites could be involved in the cholesterol-mediated suppression of the rat 12␣-hydroxylase gene expression (15), since they are implicated in the regulation by cholesterol of many genes involved in its metabolism and uptake. SREs act as binding sites for the proteolytically activated factors, SREBPs (19). In all genes studied to date, the mature forms of all three members of the SREBP family activate SRE-containing genes in tissue culture cells, with the exception of the microsomal triglyceride transfer protein gene. However, in vivo SREBP-2 favors the cholesterol biosynthetic pathway, while SREBP-1a has a relatively greater effect on fatty acid biosynthesis, and SREBP-1c is less active (23). Indeed, both SREBP-1a and -2 bind to both SREs found on the 12␣-hydroxylase promoter with about equal intensity (Fig.  3), and both 12␣-hydroxylase SREs are bound by SREBPs with higher affinity than by the LDL receptor SRE (Fig. 3). In line with these binding data and previous studies on other promoters, overexpression of the mature SREBP-1a activated the 12␣hydroxylase promoter in HepG2 cells, although not as strongly as the HMG-CoA synthase promoter (Fig. 4B), and this activation was eliminated when both 12␣-hydroxylase SREs were deleted or mutated (Figs. 6 and 7). Surprisingly, SREBP-2 suppressed 12␣-hydroxylase promoter activity when its mature form was overexpressed in HepG2 cells (Fig. 4A).
A similar suppression of 12␣-hydroxylase promoter activity was observed by manipulating the levels of endogenous SREBPs. In liver cells, SREBP-1c is the predominant transcript made from the SREBP-1 gene, and it requires LXR and its oxysterol ligands to be transcribed (37,38). McA-RH7777 hepatoma cells mimic this pattern of SREBP-1 expression (36). When these cells are incubated in the presence of compactin, an HMG-CoA reductase inhibitor, mature SREBP is essentially absent, whereas mature SREBP-2 increases (36) due to the low intracellular cholesterol levels, creating, in effect, a similar SREBP expression pattern as when the mature SREBP-2 was transfected. Under those conditions, we also observed a decrease in 12␣-hydroxylase promoter activity (Fig. 10), in contrast with an expected activation of the HMG-CoA synthase promoter.
This study strongly suggests that binding of SREBP-2 to the DNA is not required to suppress the 12␣-hydroxylase promoter and that SREBP-2 mediates this suppression indirectly, upon interaction with FTF. This is based on the following observations. First, SREBP-2 suppressed the wild type 12␣-hydroxylase promoter as well as mutants that have been deleted or mutated of their SRE sites (Fig. 6), the only SREs that were located within the 12␣-hydroxylase promoter. Second, SREBP-2 suppressed 12␣-hydroxylase promoter mutants where the whole promoter was mutated in blocks (16) (data not shown). All of these mutants were suppressed by SREBP-2 except the ones that had the FTF sites mutated, which have no activity and consequently cannot be suppressed. Third, SREBP-2 suppresses the 12␣-hydroxylase promoter when its expression is solely dependent on FTF (Fig. 8) as shown with nonliver cells (16).
The mechanism of action suggested by the experiments shown in this study (i.e. SREBP-2 suppressing the FTF activation of the 12␣-hydroxylase by physically interaction between the two proteins) is reminiscent of how bile acids down-regulate expression of the 7␣-hydroxylase and 12␣-hydroxylase genes. In this case, the bile acid-activated farnesoid X receptor induces expression of the SHP gene, which in turn interacts with the FTF protein, suppressing its activity on the 7␣-hydroxylase promoter (11,39) as well as the 12␣-hydroxylase promoter (17). However, we have observed some differences for the mechanism of suppression of 12␣-hydroxylase promoter by SHP or SREBP-2. We have shown that SHP displaces FTF binding to its sites (17), whereas in experiments performed in parallel using either in vitro made or recombinant SREBP-2, FTF binding to its sites was not displaced (data not shown). Whether the same or different FTF domain is involved in the interaction with SHP and SREBP-2 remains to be studied.
Interestingly, the 7␣-hydroxylase promoter also requires FTF for activity (9, 10), but we have observed that SREBP-2 has no effect on the 7␣-hydroxylase promoter (data not shown). How then is specificity provided to the 12␣-hydroxylase promoter such that SREBP-2 suppresses the 12␣-hydroxylase promoter but not the 7␣-hydroxylase? This problem of specificity also exists for the SHP-mediated suppression of 7␣-hydroxylase transcription, since SHP is known to interact with and suppress many other nuclear receptors such as hepatocyte nuclear factor 4 (39), estrogen receptor (40), and retinoid X receptor (41). We speculate that the 12␣-hydroxylase and 7␣hydroxylase FTF sites, although homologous, are different enough so that FTF bound to the 12␣-hydroxylase promoter can interact with SREBP-2, whereas FTF bound to the 7␣hydroxylase promoter cannot. Indeed, the 12␣-hydroxylase and the 7␣-hydroxylase FTF sites are quite different in sequence and in function. There are two overlapping FTF sites in the 12␣-hydroxylase promoter (16) and only one in the 7␣-hydroxylase promoter (9,16). Both sites are required for 12␣-hydroxylase promoter activity, since mutation of either site alone renders an inactive 12␣-hydroxylase promoter (16). Furthermore, substituting the 12␣-hydroxylase FTF site with the 7␣hydroxylase FTF element also renders the 12␣-hydroxylase promoter inactive (data not shown).
The physiological role and significance of the SREBP-2 mediated suppression of 12␣-hydroxylase expression is unclear.
Under conditions of low liver intracellular cholesterol, such as hamsters or mice fed mevinolin (an HMG-CoA reductase inhibitor) and colestipol (a bile acid binding resin), the amount of nuclear SREBP-2 increased (24,26,42). Concomitantly, the nuclear SREBP-1 decreased to undetectable levels, due in part to the requirement by the SREBP-1c promoter for ligand (oxysterols)-occupied liver X receptors (36,43). If indeed this higher than normal SREBP-2/SREBP-1 ratio suppresses 12␣hydroxylase expression, as this study strongly suggests, the ratio of cholic acid to chenodeoxycholic acid would diminish, which would increase bile acid pool hydrophobicity. Hydrophobic bile acids are stronger suppressors of 7␣-hydroxylase transcription (44) due to the higher affinity of hydrophobic bile acids for farnesoid X receptor (45). This higher hydrophobic bile acid pool would result in lower 7␣-hydroxylase activity that would decrease cholesterol conversion to bile acids and save cholesterol for more crucial needs. Further characterization of this novel SREBP-2-mediated suppression of gene expression should provide answers to these questions and a further understanding of the biology of SREBPs.