Insulin Regulation of Cholesterol 7α-Hydroxylase Expression in Human Hepatocytes

Bile acid synthesis and pool size increases in diabetes, whereas insulin inhibits bile acid synthesis. The objective of this study is to elucidate the mechanism of insulin regulation of cholesterol 7α-hydroxylase gene expression in human hepatocytes. Real-time PCR assays showed that physiological concentrations of insulin rapidly stimulated cholesterol 7α-hydroxylase (CYP7A1) mRNA expression in primary human hepatocytes but inhibited CYP7A1 expression after extended treatment. The insulin-regulated forkhead box O1 (FoxO1) and steroid regulatory element-binding protein-1c (SREBP-1c) strongly inhibited hepatocyte nuclear factor 4α and peroxisome proliferator-activated receptor γ coactivator-1α trans-activation of the CYP7A1 gene. FoxO1 binds to an insulin response element in the rat CYP7A1 promoter, which is not present in the human CYP7A1 gene. Insulin rapidly phosphorylates and inactivates FoxO1, whereas insulin induces nuclear SREBP-1c expression in human primary hepatocytes. Chromatin immunoprecipitation assay shows that insulin reduced FoxO1 and peroxisome proliferators-activated receptor γ-coactivator-1α but increased SREBP-1c recruitment to CYP7A1 chromatin. We conclude that insulin has dual effects on human CYP7A1 gene transcription; physiological concentrations of insulin rapidly inhibit FoxO1 activity leading to stimulation of the human CYP7A1 gene, whereas prolonged insulin treatment induces SREBP-1c, which inhibits human CYP7A1 gene transcription. Insulin may play a major role in the regulation of bile acid synthesis and dyslipidemia in diabetes.

The liver plays a central role in lipid metabolism and maintaining whole body lipid homeostasis, which is dysregulated in metabolic syndrome (syndrome X), obesity, and diabetes (1). Bile acid synthesis in the liver is the predominant pathway for cholesterol catabolism and is regulated by cholesterol 7␣-hy-droxylase (CYP7A1) 2 (2). Bile acids are physiological agents that facilitate biliary cholesterol excretion, intestinal absorption of nutrients, and disposal of toxic metabolites. Bile acids are also signaling molecules that activate bile acid receptors to regulate bile acid synthesis and glucose metabolism (2). In vivo studies show that bile acid pool and excretion increase in diabetic human patients (3,4) and in experimental diabetic animals (5,6). Insulin treatment restores bile acid pool and synthesis to the normal levels. It has been reported that physiological concentrations of insulin (1.4 -14 nM) inhibit bile acid synthesis by down-regulation of CYP7A1 (7)(8)(9), sterol 12␣-hydroxylase (CYP8B1) (10), and sterol 27-hydroxylase (CYP27A1) gene transcription (9). We have previously reported that insulin plays a dominant role in the inhibition of CYP7A1 gene transcription (7,8). How insulin regulates human CYP7A1 and what factors mediate the insulin effects remain unknown.
Insulin is known to induce more than 100 genes but inhibit only a few hepatic genes involved in glucose metabolism (11). The consensus sequence of negative insulin response element (IRE), T(G/A)TTT(T/G)(G/T), has been identified in promoters of phosphoenolpyruvate carboxykinase (PEPCK), glucose-6-phosphatase, insulin-like growth factor binding protein-1, tyrosine aminotransferase, and apolipoprotein CIII genes (12). These IREs are known to bind hepatocyte nuclear factor 3␣ (HNF3␣ or FoxA1), but HNF3␣ does not mediate insulin action (12). Recently forkhead box O1 (FoxO1), a mammalian homolog of Caenorhabditis elegans DAF-16 has been identified as an insulin-regulated transcription factor that plays a critical role in mediating insulin inhibition of the PEPCK, glucose-6-phosphatase, insulin-like growth factor binding protein-1, and apolipoprotein CIII genes (13)(14)(15)(16)(17). Akt/protein kinase B, a downstream serine kinase of the insulin-signaling pathway, phosphorylates FoxO1 (18). The phosphorylated FoxO1 is excluded from the nucleus and degraded by the ubiquitin proteasome pathway, which results in inhibition of FoxO1 target genes (19,20). FoxO1 exerts a positive effect by binding to its target gene promoters (15). FoxO1 also regulates genes as a * This study was supported by NIDDK, National Institutes of Health Grants DK 58379 and DK44442 (to J. Y. L. C.). 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  DNA binding-independent co-activator or co-repressor of many transcription factors (21). Dysregulation of FoxO1 expression and function leads to hyperglycemia and hypertriglyceridemia in the mouse models of type I and type II diabetes (17). Insulin stimulates the activity of sterol regulatory elementbinding proteins (SREBPs), which are key regulators of lipid homeostasis and insulin actions (22). The predominant SREBP isoform, SREBP-1c in liver and adipose tissues, preferentially induces genes in fatty acid and triglyceride synthesis (23). A SREBP precursor (125 kDa) forms a complex with SREBP cleavage-activating protein in the endoplasmic reticulum. Insulin inducing gene-1 and -2a (Insig) anchor the SREBP⅐SREBP cleavage-activating protein complex to the endoplasmic reticulum membrane (24). When cellular sterol levels are low, Insigs release the SREBP cleavage-activating protein (SCAP)⅐SREBP complex and allow SCAP to escort SREBPs to the Golgi apparatus. Two sterol-regulated proteases act sequentially to release the N-terminal basic-helix-loop-helix leucine zipper domain. The mature SREBP (68 kDa) enters the nucleus, binds to the sterol regulatory element (SRE), and induces target gene transcription. Insulin stimulates liver orphan receptor ␣ (LXR␣) which induces SREBP-1c gene expression (25,26). Insulin also promotes SREBP-1c cleavage and increases insulin-stimulated lipogenesis by inhibition of Insig-2a expression in livers (11,24). Studies have shown that insulin-dependent proteolytic cleavage is crucial and sufficient for SREBP-1c activation (27).
In this study we investigated the insulin regulation of CYP7A1 gene transcription in human hepatocytes. Results indicate that short-term treatment of physiological concentrations of insulin rapidly induces, whereas prolonged treatment of insulin represses human CYP7A1 gene expression. This dual effect of insulin action may be mediated by two insulin sensitive transcription factors, FoxO1 and SREBP-1c. This study provides new insights into how bile acid synthesis is regulated by insulin under normal physiological conditions and how dysregulation of bile acid homeostasis may contribute to glucose and lipid abnormalities in diabetes.
Transient Transfection Assay-HepG2 cells were grown to ϳ80% confluence in 24-well tissue culture plates. Luciferase reporters and expression plasmids were transfected using Lipofectamine 2000 reagent (Invitrogen) following the manufacturer's instructions. The pcDNA3 empty vector was added to normalize the amounts of DNA transfected in each assay. Luciferase and ␤-galactosidase activities were assayed and expressed in relative luciferase units as described previously (28).
RNA Isolation and Quantitative Real-time PCR-Primary human hepatocytes were cultured in serum-free and insulin-free media for 24 h and treated with insulin (Sigma) as indicated. RNA isolation, reverse transcription reactions, and real-time PCR were performed as described previously (29). All TaqMan Probes, CYP7A1 (HS00167982), FoxO1 (HS00231106), SREBP-1 (HS00231674), HNF4␣ (HS00230853), and UBC (HS00824723), were ordered from Gene Expression Assays (Applied Biosystems Inc., Foster City, CA). PEPCK probe was custom made from Custom Gene Expression Assays (Applied Biosystems, Inc.). Ubiquitin C was used as an internal control for all PCR amplification reactions. Relative mRNA expression was quantified using the comparative Ct (⌬Ct) method and expressed as 2 Ϫ⌬⌬Ct . Each assay was done in triplicate and expressed as the mean Ϯ S.D.
Electrophoretic Mobility Shift Assay (EMSA)-FoxO1 and FoxA1 were synthesized in vitro using the transcription/translation (TNT) system programmed with the FoxO1 and FoxA1 expression plasmids according to the manufacturer's instruction (Promega). [␣-32 P]dCTP (3000 Ci/mol) was obtained from PerkinElmer Life Sciences. Synthetic oligonucleotides of complimentary strands were labeled with [␣-32 P]dCTP and incubated with in vitro translated proteins (5 l) as described previously (28). Gels were dried and autoradiographed using a PhosphorImager 445Si (GE Healthcare). Sequences of the probes (tagged with GATC at the 5Ј end for labeling) used in the experiments are listed in the supplemental data. Oligonucleotides were synthesized by MWG Biotech (High Point, NC).
Site-directed Mutagenesis-A PCR-based QuikChange sitedirected mutagenesis kit (Stratagene, La Jolla, CA) was used for mutation of the reporter constructs; mutant rIRE2 sequence was introduced into rat p-344/luc plasmid, and the HNF4␣ binding site mutation was introduced into ph-1887/Luc plasmid as described previously (28).
Isolation of Cell Nuclei-Approximately 10 7 cells were washed 3 times with cold 1ϫ phosphate-buffered saline. Cells were collected by centrifugation at 4°C and re-suspended in 1 ml of buffer A (10 mM Tris-Cl, pH 7.5, 10 mM NaCl, 3 mM MgCl 2 ) containing protease inhibitors (Sigma) and incubated on ice for 10 min. 100 l of 10% Nonidet P-40 was then added, and cells were incubated for another 10 min on ice. Cells were lysed by passing through a 22-gauge needle 20 times. Cell lysates were centrifuged at 3000 rpm at 4°C for 10 min to precipitate the nuclei. Nuclei were washed 3 times in ice-cold buffer A and lysed in SDS lysis buffer for immunoblotting analysis.

Effect of Insulin on CYP7A1 and Transcription Factor mRNA Expression in Human Primary
Hepatocytes-We first studied the effect of insulin on CYP7A1 mRNA expression in human primary hepatocytes using quantitative real-time PCR. Fig. 1A shows that increasing doses of insulin (0.1 nM to 100 nM) rapidly induced CYP7A1 mRNA levels in 2 h, with a maximum induction of ϳ22-fold at the physiological concentration of 10 nM in this donor liver. Higher concentrations of insulin had less stimulatory effect on CYP7A1 mRNA expression. In contrast, insulin treatment for 6 h inhibited CYP7A1 mRNA by ϳ40% at 100 nM (Fig. 1B). As a positive control, insulin strongly reduced PEPCK mRNA expression in primary hepatocytes at both 2 and 6 h (Fig. 1, A and B). To further study the insulin effect on CYP7A1 mRNA expression, we treated primary hepatocytes with insulin (10 nM) for a period of time from 2 to 24 h. Quantitative real-time PCR analysis of primary hepatocytes from five donors showed that insulin induced CYP7A1 mRNA expression levels at 2 h by an average of ϳ9-fold and inhibited CYP7A1 mRNA at 6 and 24 h by ϳ36 and ϳ39%, respectively (supplemental Table 1). As a positive control, insulin strongly inhibited PEPCK mRNA in a time-dependent manner.
We also measured the mRNA expression levels of HNF4␣, the key regulator of human CYP7A1 gene transcription, and SREBP-1c and FoxO1, two transcription factors that are known to mediate the effects of insulin (Table 1, supplemental data). Insulin significantly induced HNF4␣ mRNA levels at 2 h by ϳ30% but inhibited it at 6 and 24 h. On the other hand, insulin induced SREBP-1c by about 2-fold, consistent with a previous report (30). Insulin had no effect on FoxO1 mRNA expression, consistent with previous studies that insulin regulates FoxO1 mainly at post-translational levels (15,18,19).
Identification of the IREs in the CYP7A1 Gene-FoxO1 is known to bind to the consensus IRE, T(G/A)TTT(T/G)(G/T), in the promoters of glucose-6-phosphatase, PEPCK, insulinlike growth factor binding protein-1, and tyrosine aminotransferase and mediates the inhibitory effect of insulin on these  A glucose-6-phosphatase IRE probe was used as a positive control. Unprogrammed TNT lysates (T3 and T7) were used as negative controls for nonspecific bindings. Sequences of IRE probes used are described in the supplemental data. C, EMSA of in vitro synthesized FoxO1 and FoxA1 with 32 P-labeled rat CYP7A1 IRE2 and mutant IRE2 probes. Competition assays were done with a 100-fold excess of unlabeled probes. Mutations in rIRE2 probes is shown. G6Pase, glucose 6-phosphatase. D, effects of FoxO1 on rat CYP7A1 reporter activities. WT, wild type rat CYP7A1 reporter (p-344/luc). RLU, relative light units. Mutant, rat CYP7A1 reporter (p-344/luc) containing single nucleotide mutation in IRE2 sequence. Reporter (0.2 g) and the indicated expression plasmids (0.1 g) were transfected into HepG2 cells. Transient transfection and luciferase assays were performed as described "Experimental Procedures." A mutation in rat IRE2 is shown. Statistical analysis was performed by Student's t test; *, significant, p Ͻ 0.05. genes (31). Analysis of the nucleotide sequences in the CYP7A1 promoter identified four putative IREs in the rat gene and two putative IREs in the human gene ( Fig. 2A). Only the IRE located in the bile acid response element-I (Ϫ81 to Ϫ75) is conserved in rat and human CYP7A1 genes. We then performed an EMSA to test if FoxO1 bound to these IREs in the CYP7A1 gene. Fig. 2B shows that FoxO1 bound strongly to the rat IRE2 (rIRE2) but did not bind to rIRE1, rIRE3, rIRE4, human (h) IRE1 and hIRE2 despite the similarity in nucleotide sequences ( Fig. 2A). FoxO1 binding to rIRE2 was competed out by a large excess of unlabeled rIRE2 probe and a glucose-6-phosphatase probe that contained three FoxO1 binding sites (Fig. 2C) (16). This rIRE2 also bound FoxA1 (HNF3␣) as expected. A single mutation of G to T in the rIRE2 abolished Foxo1 binding (Fig. 2C). To test the effect of FoxO1 on rat CYP7A1 gene transcription, reporter assays were performed in HepG2 cells with a rat CYP7A1 promoter/luciferase construct. Co-transfection of a FoxO1 expression plasmid stimulated rat CYP7A1 reporter activity by ϳ4-fold (Fig. 2D, left panel). When a single mutation that abolished FoxO1 binding was introduced into the rat CYP7A1 reporter, the basal reporter activity was reduced by more than 80%. Co-transfection of FoxO1 failed to stimulate the mutant reporter activity (Fig. 2D, left panel). This mutant reporter did respond to HNF4␣ and LXR␣ because it retained the HNF4␣ and LXR␣ binding sites as in the wild type plasmid (Fig. 2D,  right panel).
FoxO1 Inhibited Human CYP7A1 Reporter Activity by Blocking HNF4␣ Trans-activation Activity-The effect of FoxO1 on human CYP7A1 reporter activity was studied using transfection assays. In contrast to the stimulatory effect on the rat CYP7A1 reporter, co-transfection of FoxO1 inhibited human CYP7A1 reporter activities of ph-1887/Luc, ph-371/Luc, and ph-150/Luc by ϳ70% (Fig. 3A). Further deletion of the sequence downstream of Ϫ150 abolished the inhibitory effect of FoxO1. Thus, the inhibitory effect of FoxO1 required the region from Ϫ150 to Ϫ135, which was previously mapped as the bile acid response element-II that contained an HNF4␣ binding site (32). It has been reported that FoxO1 interacts with the DNA binding domain of HNF4␣ and inhibits HNF4␣ transactivation activity and its DNA binding (21). To test if the HNF4␣ binding site was involved in FoxO1 regulation of the human CYP7A1 gene, the HNF4␣ binding site in the ph-1887/ Luc reporter was mutated. Fig. 3B shows that the HNF4␣ binding site mutant reporter (mHNF4␣-ph-1887/luc) had a much lower basal activity, and FoxO1 did not inhibit the mutant reporter activity. These results suggest that FoxO1 might inhibit human CYP7A1 gene by inhibiting HNF4␣ trans-activation activity. A mammalian one-hybrid assay was used to study the effect of FoxO1 on HNF4␣ trans-activation of a Gal4 (5ϫUAS)/TK/Luc reporter. Fig. 3C shows that Gal4-HNF4␣ activation of the GAL4 reporter activity was drastically increased by PGC-1␣, and FoxO1 dose-dependently inhibited reporter activity (Fig. 3C). These results suggest that in addition to interfering with HNF4␣ binding to DNA, FoxO1 might also block HNF4␣ and PGC-1␣ interaction. The effect of FoxO1 on HNF4␣ and PGC-1␣ co-activation of human CYP7A1 reporter activity (Fig. 3D) was also studied. HNF4␣ or PGC-1␣ stimulated human CYP7A1 reporter activity by less than 100%, likely the result of abundant HNF4␣ expression in HepG2 cells. FoxO1 strongly inhibited the human CYP7A1 reporter activity stimulated by HNF4␣ and/or PGC-1␣. Taken together, these results suggest that FoxO1 could act as a DNA binding independent co-repressor of HNF4␣ and interfere with HNF4␣ and PGC-1␣ co-activation of CYP7A1 gene transcription.
SREBP-1c Inhibited Human CYP7A1 Gene Transcription by Blocking HNF4␣ Recruitment of PGC-1␣-To test if SREBP-1c was involved in mediating insulin effect on human CYP7A1, the effect of SREBP-1c on human CYP7A1 reporters was studied by transfection assays in HepG2 cells. Co-transfection of the nuclear form of SREBP-1c strongly inhibited the human CYP7A1 reporter (ph-1887/Luc) by more than 80% (Fig. 4A). Deletion analyses of the human CYP7A1 promoter localized a region from Ϫ150 to Ϫ135 that mediated the inhibitory effect of SREBP-1c (Fig. 4A). We then tested the effect of SREBP-1c on the HNF4␣ binding site mutant reporter (mHNF4␣-ph-1887/luc). As shown in Fig. 4B, mutation of the HNF4␣ binding site markedly reduced the basal reporter activity, and SREBP-1c failed to inhibit the mutant reporter activity. These results suggest that SREBP-1c might inhibit human CYP7A1 by inhibiting HNF4␣ trans-activation of CYP7A1. SREBP-1c has been shown to interact with HNF4␣-LBD and block its recruitment of PGC-1␣ to the PEPCK gene (33). Thus, a mammalian onehybrid assay was used to study the effect of SREBP-1c on PGC-1␣ and Gal4-HNF4␣ co-activation of a GAL4 5ϫUAS/ TK/Luc reporter. Fig. 4C shows that SREBP-1c inhibited Gal4-HNF4␣-LBD and PGC-1␣ co-activation of reporter activity in a dose-dependent manner. Furthermore, SREBP-1c strongly inhibited the basal activity of a human CYP7A1 reporter and the activity stimulated by HNF4␣ and PGC-1␣ (Fig. 4D). Thus, SREBP-1c could act as a DNA binding independent co-repressor of HNF4␣ and interfere with HNF4␣ and PGC-1␣ co-activation of CYP7A1 gene transcription.

Western Blot Analysis of Effects of Insulin on FoxO1, SREBP-1c, and HNF4␣ Protein Expressions in Primary Human
Hepatocytes-To study effect of insulin on phosphorylation of FoxO1 and expression of SREBP-1c protein in primary human hepatocytes, immunoblot analysis of FoxO1 and SREBP protein expression was conducted. Fig. 5A shows that without insulin treatment, FoxO1 was not phosphorylated and was mainly located in the nucleus. Insulin treatment for 30 min led to phosphorylation of FoxO1 and its rapid nuclear exclusion and degradation. Immunoblot analysis did not detect SREBP-1c precursor (125 kDa) expression in hepatocytes (data not shown). Insulin treatment increased the amount of mature SREBP-1c (68 kDa) in cell lysates in a time-dependent manner (Fig. 5B). Insulin treatment did not significantly affect HNF4␣ protein levels in hepatocytes (Fig. 5C). The time courses of insulin inactivation of FoxO1 and induction of SREBP-1c in hepatocytes correlate well with quantitative real-time PCR results on CYP7A1 expression and support our hypothesis that insulin inactivates FoxO1 to rapidly induce CYP7A1 gene expression and that prolonged insulin treatment induces SREBP-1c to inhibit CYP7A1 gene expression.
ChIP Assay of HNF4␣ Binding and Recruitment of SREBP-1c, FoxO1, and PGC-1␣ to the Human CYP7A1 Chromatin-To further confirm the roles of FoxO1 and SREBP-1c in mediating insulin regulation of human CYP7A1 gene transcription, ChIP assays were performed using primary human hepatocytes treated with 10 nM insulin for a period of time from 1 to 24 h. An antibody against FoxO1, SREBP-1c, HNF4␣, or PGC-1␣ was used to immunoprecipitate chromatin complexes for ChIP assays using primers designed to amplify a 391-bp fragment of human CYP7A1 promoter region that included an HNF4␣ binding site. In the absence of insulin, ChIP assays detected FoxO1 and HNF4␣ in the CYP7A1 chromatin (Fig. 6A). Because FoxO1 does not bind to the human CYP7A1 promoter, FIGURE 3. Effect of FoxO1 on human CYP7A1/Luc reporter activity. A, CYP7A1 reporter constructs (0.2 g) were co-transfected with expression plasmids (0.1 g) into HepG2 cells, and luciferase activity (relative light units (RLU)) was analyzed 40 h after transfection. RLU, relative light units. B, HNF4␣ site mutant reporter (mHNF4-ph1887-Luc) was transfected in HepG2 cells. pcDNA3 empty vector or FoxO1 expression plasmid was co-transfected as indicated. C, mammalian one-hybrid assay. The Gal4/luciferase reporter 5ϫUAS/TK/Luc (0.2 g) was co-transfected with Gal4 empty vector (Ϫ) (0.1 g), Gal4-HNF4␣ (0.1 g), PGC-1␣ (0.1 g), and increasing amounts of FoxO1 expression plasmid (0.05-0.2 g) as indicated. D, a human CYP7A1 reporter, ph-371-Luc (0.2 g), was co-transfected with FoxO1, HNF4␣, and/or PGC-1␣ (0.1 g) as indicated into HepG2 cells. Each assay was performed in triplicate, and statistical analysis was performed using Student's t test; *, significant, p Ͻ 0.05. it must have been recruited to the CYP7A1 chromatin by HNF4␣. Insulin treatment rapidly abolished FoxO1 association with the CYP7A1 chromatin, consistent with FoxO1 inactivation and nuclear exclusion by insulin. In contrast, in the absence of insulin SREBP-1c was not recruited to the CYP7A1 chromatin. Insulin treatment time-dependently increased SREBP-1c but decreased PGC-1␣ recruitment to CYP7A1 chromatin. These results are consistent with the induction of nuclear SREBP-1c by insulin in hepatocytes (Fig. 5B). Interestingly, insulin treatment for 1 h increased HNF4␣ and PGC-1␣ binding to CYP7A1 chromatin. Because FoxO1 is known to interact with HNF4␣ and inhibit its DNA binding activity (21), decreased FoxO1 protein in the nucleus allows HNF4␣ to bind to CYP7A1 chromatin. Taken together, our ChIP assay results suggest that insulin rapidly decreases FoxO1 in the nucleus, which leads to increased HNF4␣ and PGC-1␣ binding and trans- . Effect of SREBP-1c on human CYP7A1 reporter activity. A, human CYP7A1 reporter constructs (0.2 g) were co-transfected with pcDNA3 empty vector or SREBP-1c expression plasmids (0.1 g) into HepG2 cells, and luciferase activities were analyzed 40 h after transfection. RLU, relative light units. B, HNF4␣ site mutant reporter (mHNF4-ph1887/Luc) (0.2 g) was co-transfected with 0.1 g of pcDNA3 or SREBP-1c expression plasmid into HepG2 cells. C, mammalian one-hybrid assay. The Gal4/luciferase reporter, 5ϫUAS/TK/Luc (0.2 g) was co-transfected with Gal4 empty vector (Ϫ) (0.1 g), Gal4-HNF4␣ (0.1 g), PGC-1␣ (0.1 g), and increasing amounts of SREBP-1c expression plasmid (0.05-0.2 g) as indicated. D, human CYP7A1 reporter construct, ph-371-Luc (0.2 g), was co-transfected with 0.1 g each of HNF4␣ and PGC-1␣ and/or SREBP-1c expression plasmid as indicated. Each assay was performed in triplicate, and statistical analysis was performed using Student's t test; *, significant, p Ͻ 0.05. activation of the CYP7A1 gene. Insulin induces SREBP-1c, which is recruited to CYP7A1 chromatin to block HNF4␣ recruitment of PGC-1␣, and results in inhibiting CYP7A1 gene transcription.

DISCUSSION
In the current study we investigated the insulin regulation of human CYP7A1 gene transcription in human hepatocytes. Quantitative real-time PCR analyses have revealed that physiological concentrations of insulin strongly and rapidly induce CYP7A1 mRNA expression. Interestingly, prolonged insulin treatment results in inhibition of human CYP7A1 mRNA expression. Results indicate that the dual effect of insulin action on human CYP7A1 is likely mediated through two insulin-regulated transcription factors, FoxO1 and SREBP-1c. FoxO1 binds to the IRE in the rat CYP7A1 gene promoter and stimulates rat CYP7A1 reporter activity. However, this IRE is not conserved in the human CYP7A1 promoter. Our results sug- . Chromatin immunoprecipitation assays of the effect of insulin on FoxO1, SREBP-1c, and HNF4␣ recruitment to CYP7A1 chromatin. A, human primary hepatocytes (HH1308) were treated with 10 nM insulin for the time indicated. Cells were harvested for ChIP assay as described "Experimental Procedures." An antibody against HNF4␣, SREBP-1, or FoxO1 was used to immunoprecipitate chromatin for PCR amplification and analysis. 10% of cell lysate was used as input. Non-immune IgG was used as negative controls. B, ChIP assays of the effect of insulin and T0901317 on HNF4␣ recruitment of PGC-1␣ to CYP7A1 chromatin. HepG2 cells in 100 mm dishes were transfected with HA-tagged PGC-1␣ (10 g). Cells were then treated with insulin (100 nM) or T0901317 (1 M) for 24 h. Chromatin was precipitated (IP) with an antibody against HA tag or HNF4␣ for ChIP assays. 10% of cell lysate was used as input. Non-immune IgG was used as negative controls. C, ChIP assays of HepG2 cells overexpressing SREBP-1c or FoxO1. HepG2 cells in a 100-mm dishes were transfected with 10 g of HA-tagged PGC-1␣, pcDNA3 empty vector, and FoxO1, or SREBP-1c expression plasmid was overexpressed as indicated. Cells were harvested for ChIP assay as described under "Experimental Procedures." An antibody against HA tag or HNF4␣ was used to immunoprecipitate chromatin for PCR amplification and analysis. 10% of cell lysate was used as input. Non-immune IgG was used as a negative control.
gest that FoxO1 functions as a co-repressor that inhibits the human CYP7A1 gene by interfering with the HNF4␣ and PGC-1␣ co-activation of the gene. The ChIP assay reveals that FoxO1 is present in CYP7A1 chromatin without insulin treatment, and FoxO1 may interact with HNF4␣ to block its recruitment of PGC-1␣ to CYP7A1 chromatin (Fig. 6A). Upon insulin treatment, FoxO1 is released, and PGC-1␣ is recruited by HNF4␣ to transiently stimulate CYP7A1 mRNA expression in human hepatocytes. Prolonged insulin treatment induces the mature form of SREBP-1c, which is recruited to CYP7A1 chromatin to inhibit HNF4␣ and PGC-1␣ trans-activation of the human CYP7A1 gene. This study suggests that CYP7A1 is a target gene of both FoxO1 and SREBP-1c, which play critical roles in regulating bile acid synthesis in response to insulin levels in hepatocytes. Bile acids are known to activate a nuclear receptor farnesoid X receptor, which induces a negative nuclear receptor small heterodimer partner to inhibit nuclear receptor-regulated gene transcription (34). Several recent reports have shown that bile acids play critical roles in the regulation of glucose and lipid metabolism (35)(36)(37). During the postprandial state, serum insulin levels increase transiently to stimulate glucose metabolism. According to our study, insulin may rapidly induce CYP7A1 and bile acid synthesis to facilitate nutrient absorption in the intestine during the postprandial state. Bile acids inhibit the PEPCK gene and prevent hyperglycemia (38). During the post-absorptive state and starvation, glucagon levels increase to inhibit CYP7A1 and bile acid synthesis while stimulating gluconeogenesis to prevent hypoglycemia (39).
In type II diabetes, FoxO1 nuclear exclusion is impaired due to insulin resistance (17). Hyperinsulinemia may inhibit CYP7A1 and bile acid synthesis by inducing SREBP-1c, which stimulates lipogenesis and contributes to hypertriglyceridemia. Inhibition of bile acid synthesis stimulates gluconeogenesis and lipogenesis. In type I diabetes, lack of insulin activates FoxO1, which stimulates gluconeogenesis, but inhibits bile acid synthesis and also causes hyperglycemia and dyslipidemia. All these metabolic changes are caused by insulin and may contribute to the development of diabetic dyslipidemia and cardiovascular diseases.
It should be noted that we have demonstrated distinct species differences in FoxO1 and SREBP-1c regulation of rat and human CYP7A1 gene transcription. FoxO1 induces rat CYP7A1 by binding to an IRE (TGTTTAC) in the rat gene promoter, but FoxO1 does not bind and inhibit the human CYP7A1 gene. Analysis of homologous CYP7A1 genes reveals that this rat IRE sequence is also conserved in mouse and hamster genes. In the mouse CYP7A1 gene this IRE is located in the same region and is likely a FoxO1 binding site. A recent report shows a 2-fold higher expression of CYP7A1 mRNA in FoxO1 transgenic mice, suggesting induction of mouse CYP7A1 by FoxO1 (40). In the hamster CYP7A1 gene this IRE is located upstream of a putative IRE that binds HNF3␣ and HNF4␣ (41). SREBP-1c also differentially regulates the rat and human CYP7A1 genes. SREBP-1c binds to the E-box (CANNTG) and SRE motifs and activates gene transcription (42). However, SREBP-1c also can function as a DNA binding-independent repressor by interfering with HNF4␣ interaction with PGC-1␣ and inhibiting PEPCK gene transcription (33). This study shows that the same mechanism also inhibits the human CYP7A1 gene. In contrast, we found that SREBP-1c strongly stimulated rat CYP7A1 reporter activity (data not shown). Analysis of the rat CYP7A1 gene promoter sequence reveals three E-boxes and several GC-rich sequences located downstream of Ϫ344 of the rat CYP7A1 promoter. These E-box motifs are not conserved in the human CYP7A1 gene. Further study is needed to identify the SREBP-1c binding site in the rat CYP7A1 gene and the mechanism by which FoxO1 and SREBP-1c regulates the rat CYP7A1 gene. Previously we have reported that in contrast to the induction of CYP7A1 mRNA expression in mouse liver, LXR␣ does not stimulate human CYP7A1 gene transcription because the human CYP7A1 gene lacks a LXR␣ response element (43). Species differences in regulation of the CYP7A1 gene and lipid metabolism underscore the importance of studying human CYP7A1 gene regulation in human hepatocyte models.
In summary, we studied the mechanisms of insulin regulation of CYP7A1 gene transcription mediated by FoxO1 and SREBP-1c. Our results suggest that insulin has a dual effect on human CYP7A1 gene transcription; that is, a rapid induction at physiological concentrations to control glucose and lipid metabolism, and later, inhibition to maintain glucose and lipid homeostasis. Understanding the complex mechanism of insulin regulation of bile acid synthesis and lipid homeostasis will lead to developing novel drug therapies for treating diabetic dyslipidemia.