Bile acids regulate gluconeogenic gene expression via small heterodimer partner-mediated repression of hepatocyte nuclear factor 4 and Foxo1.

Bile acid homeostasis is tightly controlled by the feedback mechanism in which an atypical orphan nuclear receptor (NR) small heterodimer partner (SHP) inactivates several NRs such as liver receptor homologue-1 and hepatocyte nuclear factor 4. Although NRs have been implicated in the transcriptional regulation of gluconeogenic genes, the effect of bile acids on gluconeogenic gene expression remained unknown. Here, we report that bile acids inhibit the expression of gluconeogenic genes, including glucose-6-phosphatase (G6Pase), phosphoenolpyruvate carboxykinase, and fructose 1,6-bis phosphatase in an SHP-dependent fashion. Cholic acid diet decreased the mRNA levels of these gluconeogenic enzymes, whereas those of SHP were increased. Reporter assays demonstrated that the promoter activity of phosphoenolpyruvate carboxykinase and fructose 1,6-bis phosphatase via hepatocyte nuclear factor 4, or that of G6Pase via the forkhead transcription factor Foxo1, was down-regulated by treatment with chenodeoxicholic acid and with transfected SHP. Remarkably, Foxo1 interacted with SHP in vivo and in vitro, which led to the repression of Foxo1-mediated G6Pase transcription by competition with a coactivator cAMP response element-binding protein-binding protein. These findings reveal a novel mechanism by which bile acids regulate gluconeogenic gene expression via an SHP-dependent regulatory pathway.

Bile acid metabolism plays a critical role for the elimination of excess cholesterol in the liver. This pathway is achieved via transcriptional regulation of cholesterol 7␣-hydroxylase (CYP7A1), the rate-limiting enzyme of bile acid biosynthesis (10,11). Upon bile acid stimulation, a member of nuclear receptors (NRs), farnesoid X receptor (FXR), potently activates transcription of the small heterodimer partner, SHP, which is an atypical orphan NR lacking a DNA-binding domain. Elevated SHP heterodimerizes and inactivates the liver receptor homologue-1 that is a transactivator of the CYP7A1 gene, eventually inhibiting bile acid production by repressing CYP7A1 transcription (12,13). SHP also antagonizes the functions of several NRs, including the glucocorticoid receptor (GR) (14) and HNF-4 (15). Given the important roles for GR and HNF-4 in controlling gluconeogenesis, these studies suggested that bile acid metabolism might link to glucose homeostasis.
In this study, we show that bile acids efficiently suppress the mRNA levels and promoter activities of G6Pase, PEPCK, and FBP1 in mouse liver and HepG2 cells, respectively. Remarkably, overexpression of SHP represses the gluconeogenic gene promoters by disrupting the association of CBP-Foxo1 or CBP-HNF-4. Our findings provide evidence that bile acids repress gluconeogenic gene transcription via the effect of the negative nuclear receptor SHP on HNF-4 or Foxo1.

EXPERIMENTAL PROCEDURES
Animal Studies-Eight-week-old male C57BL/6J mice (CLEA) were maintained on a fixed artificial light cycle (12 h of light and 12 of dark) on a standard diet or a diet supplemented with 1% cholic acid for 7 days.
Both groups were allowed free access to water. Mice were sacrificed 12 h after fasting, and the livers were removed for Northern blot analysis.
Northern Blot Analysis-Total RNA was isolated from mouse livers using ISOGEN (Nippon Gene), and the aliquots (10 g) were pooled from eight mice per diet. After electrophoresis, RNA was blotted onto GeneScreen Plus (PerkinElmer Life Sciences) and subsequently hybridized with specific 32 P-labeled cDNA probes. Relative expression levels were analyzed by utilizing a Typhoon 8600 (Molecular Dynamics) and standardized against ␤-actin controls. Mouse G6Pase (14 -682 bp, a kind gift from Dr. Atsushi Miyajima), mouse PEPCK (1119 -1869 bp), mouse FBP1 (221-1219 bp), and human SHP (1-774 bp) were used as hybridization probes. Cell Culture, Transfections, and Reporter Gene Assays-HepG2 and HEK293T cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, and transfections were performed by FuGENE6 reagents (Roche Applied Science). pRSV or pSV40-␤-galactosidase plasmid was included in each transfection experiment to control for the efficiency of transfection. To ensure equal DNA amounts, empty plasmids were added in each transfection. After transfection, cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum for 24 -48 h. If necessary, 5 h after transfection, medium was replaced with serum-free Dulbecco's modified Eagle's medium in the presence or absence of chenodeoxycholic acid (CDCA; Sigma) at the indicated concentrations, and cells were further incubated for 24 h. Luciferase activity was measured with Wallac (PerkinElmer Life Sciences) and normalized for ␤-galactosidase activity in the same sample.
GST Pull-down Assays-GST and GST fusion proteins were expressed in Escherichia coli BL21 (Promega) and purified using glutathione-Sepharose beads according to the manufacturer's instructions (Amersham Pharmacia Biotech). GST pull-down assays were performed by incubating cell extracts from HEK293T cells transfected with FLAG-Foxo1 or HA-SHP with equal amounts of various GST-Foxo1 or GST-SHP proteins immobilized on glutathione-Sepharose beads in the binding buffer (20 mM HEPES (pH 7.9), 150 mM KCl, 0.1% Nonidet P-40, 1 mM dithiothreitol, and protease inhibitors) for 4 h at 4°C. After washing the beads, the proteins were resolved by SDS-PAGE and analyzed by Western blot analysis.

RESULTS
Bile Acids Repress Gluconeogenic Gene Promoter Activities-To evaluate the effects of bile acids on hepatic gluconeogenic gene expression in vivo, C57BL/6J mice were treated for 7 days with a diet containing cholic acid, which is known to be a natural ligand of the nuclear receptor FXR (23). After overnight fasting, total RNA was isolated from the liver and subjected to Northern blot analysis. Consistent with previous re- ports (12,13), the treatment of cholic acid in mice induced SHP gene expression but not ␤-actin (Fig. 1A). On the other hand, the cholic acid diet substantially reduced the mRNA levels of gluconeogenic genes, including G6Pase, PEPCK, and FBP1 (Fig. 1A).
To elucidate the mechanism by which cholic acid down-regulates gluconeogenic gene expression, we performed reporter assays using human G6Pase-, PEPCK-, and FBP1-promoterluciferase constructs. HepG2 cells were transfected with each reporter construct and then treated with increasing concentrations of chenodeoxycholic acid (CDCA), which is a primary bile acid and induces SHP via the FXR-dependent pathway (12,13,24,25). As expected from the results in mice (Fig. 1A), CDCA led to a dose-dependent decrease in the three reporter gene activities (Fig. 1B, left), whereas no difference was observed in SV40 promoter activity (Fig. 1B, right). These results indicate that bile acids repress gluconeogenic gene expression in the transcriptional level.

Bile Acid-responsible Elements Are Identified in PEPCK and
G6Pase Promoters-Next, we sought to address the bile acidresponsible element(s) in the PEPCK and G6Pase promoters. The effect of CDCA on the 5Ј-deletions of the human PEPCK and G6Pase promoters was assayed in HepG2 cells (Fig. 2, A and B, left panels). As shown in Fig. 2A, right, CDCA reduced the activity of the PEPCK promoter-containing sequence Ϫ600 to ϩ7 (PEPCK/Ϫ600-luc). Although the promoter activity of 5Ј-deletion to Ϫ480 (PEPCK/Ϫ480-luc) was still repressed by CDCA treatment, this repressive effect was significantly diminished in the PEPCK/Ϫ335-luc construct, suggesting that the sequence Ϫ480 to Ϫ335 in the PEPCK promoter appears to contain the putative element, which is negatively regulated by CDCA. Similarly, CDCA repressed the G6Pase/Ϫ827-luc, G6Pase/Ϫ499-luc, and G6Pase/Ϫ237-luc constructs, but not the G6Pase/Ϫ154-luc construct (Fig. 2B, right), suggesting that the sequence Ϫ237 to Ϫ154 includes the bile acid response element in the G6Pase promoter. CDCA Represses Gluconeogenic Gene Promoter Activities via HNF-4 or Foxo1-Interestingly, the bile acid response elements identified in the PEPCK and G6Pase promoters overlap with the associate factor 1 (AF1) region (Ϫ441 to Ϫ410, PEPCK) and the insulin response sequence (IRS) (Ϫ189 to Ϫ157, G6Pase), respectively (Fig. 3A) (7,26,27). It has been shown that the AF1 region contains direct repeat sequences that correspond to a binding site for nuclear receptor HNF-4 (28). On the contrary, the IRS consensus has been identified as the binding sites for the forkhead transcription factor, Foxo1 (29,30). Because both of the transcription factors are involved in maintenance of gluconeogenic gene expression (1, 5), we hypothesized that bile acids may repress gluconeogenic gene promoter activities through HNF-4-or Foxo1-mediated transcriptions.
To test this possibility, we generated the two types of mutated reporters. In the human PEPCK promoter, the direct repeat motif in the AF1 region was converted to the GAL4 recognition site (upstream activation sequence, UAS) (Fig. 3A,  left). Alternatively, the consensus sequences of IRSs were mutated in the human G6Pase promoter (Fig. 3B, left). After confirming that gAF1-luc largely diminishes transactivation by HNF-4 and the mtIRS-luc construct completely abolishes the response to Foxo1 (data not shown), we investigated whether AF1 in the PEPCK or IRS in the G6Pase promoters coincides with bile acid response elements. Compared with the PEPCK and G6Pase wild-type reporters, CDCA-dependent repression was substantially abrogated in the gAF1 and mtIRS constructs (Fig. 3, A and B, right panels). These findings clearly indicated that HNF-4 and Foxo1 are target transcription factors for CDCA-dependent repression in gluconeogenic gene expression.
To further clarify the repressive mechanism of CDCA, HepG2 cells were cotransfected with PEPCK-or FBP1-promoter-luciferase constructs together with HNF-4 expression plasmid following incubation with CDCA. In the absence of CDCA, HNF-4 stimulated the promoter activities of PEPCK and FBP1 5-7-fold, whereas CDCA drastically inhibited HNF-4-mediated transactivation of the promoters (Fig. 3C). Likewise, in the G6Pase promoter, CDCA efficiently inhibited Foxo1-mediated transcription in a dose-dependent manner (Fig. 3D). Taken together, these results suggest that CDCA down-regulates the transactivation function of HNF-4 and Foxo1 in gluconeogenic gene expressions.
SHP Represses Gluconeogenic Gene Promoter Activities via HNF-4 or Foxo1-Bile acids have been shown to induce SHP protein via the FXR-dependent pathway (12, 13, 24, 25). SHP dimerizes with a subset of nuclear receptors and represses their transactivation functions (31). To assess whether CDCA represses gluconeogenic gene promoter activities through FXR/ SHP-dependent pathway, a dominant-negative FXR mutant (W469A), which contains ligand-and DNA-binding domains but lacks transactivation function (21,22), was cotransfected with G6Pase-, PEPCK-, and FBP1-promoter-luc constructs. As shown in Fig. 4A, FXR (W469A) canceled repression by CDCA in these promoter activities, suggesting that CDCA repressed gluconeogenic gene transcription via FXR-dependent pathway. Moreover, we confirmed that the mRNA levels of SHP are increased by CDCA in HepG2 cells (32). These findings prompted us to examine the effects of SHP on the gluconeogenic gene promoters. To this end, increasing amounts of SHP expression plasmids were cotransfected into HepG2 cells together with the G6Pase-, PEPCK-, and FBP1-promoter reporter constructs. As shown in Fig. 4B, SHP significantly repressed the gluconeogenic gene promoters in a dose-dependent manner (left panel), whereas SHP had no effect on SV40 promoter activity (right panel). Several studies have reported that bile acids repress human CYP8B1 and CYP27A1 transcription by attenuating the transactivation activity of HNF-4 through the interaction with SHP (33,34). In addition, our recent study demonstrated that bile acids repress HNF-4-mediated human angiotensinogen gene expression in an SHP-dependent manner (32). We therefore explored whether SHP could antagonize HNF-4-mediated gluconeogenic gene transcription. HNF-4 and increasing amounts of SHP expression plasmids were cotransfected into HepG2 cells together with the PEPCK-and FBP1-promoter-luc plasmids. As shown in Fig. 4C, SHP completely inhibited HNF-4-induced activation of the PEPCK and FBP1 promoters. Moreover, we examined the effect of SHP on the Foxo1-mediated G6Pase transcription. As expected, SHP potently repressed Foxo1-stimulated G6Pase promoter activity in a dose-dependent manner (Fig. 4D). These results suggest that bile acids down-regulate the promoter activity of PEPCK and FBP1 via HNF4 or that of G6Pase via Foxo1 in an SHP-dependent fashion.
Foxo1 Interacts with SHP in Vivo and in Vitro-To further confirm whether Foxo1 is a regulatory target whereby bile acids inhibit the G6Pase promoter, additional reporter assays were conducted using a 3ϫIRS model promoter construct that contains the cognate Foxo1 binding elements. Similar to that observed on the G6Pase promoter (Fig. 4, C and D), the treatment of CDCA and the cotransfection of SHP attenuated the activity of the 3ϫIRS model promoter (Fig. 5, A and B). These data strongly indicated that Foxo1 is a direct target of SHP. To elucidate the mechanism underlying the SHP-dependent inhibition of Foxo1 activity, we used coimmunoprecipitation assays to test whether Foxo1 interacts with SHP. HEK293T cells were cotransfected with FLAG-Foxo1 with or without HA-SHP expression plasmids. Immunoprecipitations were performed with anti-FLAG antibody, and the precipitates were analyzed by Western blot with anti-FLAG and anti-HA antibodies. As shown in Fig. 5C, the Foxo1-SHP complexes were readily detected when both FLAG-Foxo1 and HA-SHP were cotransfected, indicating that Foxo1 can associate with SHP in vivo. To further determine which portions of Foxo1 and SHP are essen- tial for their interactions, GST pull-down assays were conducted using HA-SHP transfected into HEK293T cells and several deletion mutants of GST-Foxo1 protein expressed in bacteria (Fig. 5D, upper panel). As illustrated in Fig. 5D, lower panel, SHP tightly bound the C terminus of Foxo1 (410 -652 amino acids). Alternatively, Foxo1 exclusively bound the N terminus of SHP (1-91 amino acids) (Fig. 5E). These data demonstrate that Foxo1 is a target of SHP in vivo and in vitro. SHP Displaces CBP by Competing for Interaction with Foxo1-To elucidate the mechanism underlying the repression of Foxo1-mediated transcription by SHP, we employed mammalian one-hybrid analysis using the GAL4 DNA-binding domain fused to the C terminus of Foxo1 (GAL4-Foxo1-CT). This region contains the minimum transactivation domain of Foxo1 (35) and the binding region of p300 (36). Furthermore, we confirmed that it corresponds to the binding region of CBP 2 and SHP (Fig. 5E). Thus, these observations raised the possibility that SHP may compete with the interaction of Foxo1 and CBP. HepG2 cells were cotransfected with a GAL4 reporter gene containing the five GAL4 recognition sites upstream of the adenovirus major late promoter (5ϫUAS-MLP-luc) and expression plasmids for either the GAL4 DNA-binding domain (GAL4-DBD) as a negative control or GAL4-Foxo1-CT. Significantly, CDCA and SHP inhibited transactivation of GAL4-Foxo1-CT (Fig. 6, A and B), suggesting that SHP interfered with Foxo1 transactivation, probably by dissociating CBP from Foxo1.
To verify this possibility, we investigated whether SHP prevents physical interaction between Foxo1 and CBP. HEK293T cells were cotransfected with FLAG-Foxo1 together with increasing amounts of SHP expression plasmid, and the cell lysates were immunoprecipitated with anti-FLAG antibody (Fig. 6C). As shown in Fig. 6C, SHP interfered with the binding of Foxo1 to endogenous CBP in vivo in a dose-dependent manner. These results suggested that disruption of the Foxo1-CBP complex by SHP led to attenuating Foxo1-mediated transactivation.
SHP Displaces CBP by Competing for Interaction with HNF-4 -Finally, we investigated how SHP represses HNF-4-mediated promoter activities. Because SHP competes with the association between HNF-4 and a coactivator, SRC-3 (15), we performed mammalian one-hybrid analysis using the GAL4 DNA-binding domain fused to HNF-4 (GAL4-HNF-4). HepG2 cells were cotransfected with a 5ϫUAS-MLP-luc and expression plasmids for either the GAL4-DBD or GAL4-HNF-4. Significantly, SHP inhibited GAL4-HNF-4-mediated promoter activity (Fig. 7A), suggesting that SHP interfered with HNF-4 transactivation, probably by dissociating coactivators. Because it was reported that HNF-4 interacts with CBP (20, 37), we tested whether SHP could compete with the interaction of HNF-4 with CBP. As shown in Fig. 7A, CBP coactivated HNF-4-dependent reporter activities. Interestingly, we found that SHP interferes with CBP-mediated HNF-4 transactivation (Fig. 7A). To further investigate the mechanism by which SHP suppresses HNF-4-mediated transactivation, we investigated whether SHP prevents the physical interaction of HNF-4 with CBP. To this end, HEK293T cells were cotransfected with FLAG-HNF-4 together with empty vector or SHP expression plasmid and the cell lysates were immunoprecipitated with anti-FLAG antibody (Fig. 7B). As shown in Fig. 7B, SHP interfered with the binding of HNF-4 to endogenous CBP in vivo. These findings suggest that SHP inhibits HNF-4-mediated transactivation of PEPCK and FBP1 promoters by disrupting the HNF-4-CBP complex. DISCUSSION In the present study, we have made four major findings. First, bile acids down-regulate G6Pase, PEPCK, and gene expression in mice (Fig. 1). Second, CDCA and SHP repress HNF-4-or Foxo1-stimulated promoter activities of the gluconeogenic genes (Figs. 2-4). Third, SHP antagonizes the transactivation function of Foxo1 through direct interaction (Fig. 5). Finally, the mechanism underlying SHP-dependent inactivation of Foxo1 and HNF-4 is due to the dissociation of CBP from them ( Fig. 6 and Fig. 7, A and B). Based on these results and previous reports in which bile acids bind and activate the FXR/RXR heterodimer that leads to the induction of SHP (12,13), we summarize a model for bile acid-mediated down-regulation of the gluconeogenic genes via induced SHP (Fig. 7C). According to this model, the increased levels of SHP displace CBP by competing for interaction with Foxo1 on the G6Pase promoter and HNF-4 on the PEPCK and FBP1 promoters.
More recently, De Fabiani et al. (38) have shown that bile acids repress CYP7A1 and PEPCK gene expression. They, however, argued that the suppressive effect was accomplished by an SHP-independent dissociation of the HNF-4-coactivator complex. Compared with their study, we showed that bile acids repressed the expression of the three gluconeogenic genes in an SHP-dependent manner. Furthermore, SHP directly targeted and inhibited transactivating abilities of Foxo1, which is a key regulator of G6Pase gene transcription (3). Combined with their findings and ours, it seems likely that there are at least two distinct pathways in bile acid-dependent suppression of gluconeogenic gene expression through SHP-dependent and -independent mechanisms.
Downstream of the insulin-signaling pathway, the Foxo subfamily members are phosphorylated by protein kinase B, which leads to their exclusion from the nucleus (39,40). In addition to phosphorylation, we previously reported that Foxo factors are finely regulated by multiple modifications, namely acetylation (19) and ubiquitination (41). These modifications cause transcriptional repression (19) or proteasomal degradation in Foxo proteins (41,42). Alternatively, a recent study (43) has demonstrated that bile acids regulate C/EBP␤, CREB, and c-Jun function by their phosphorylation through extracellular signalregulated kinase and c-Jun NH 2 -terminal kinase pathways. In view of these findings, it will be of interest to verify the additional possibility that bile acids can stimulate the phosphorylation, acetylation, or ubiquitination of Foxo1 via a signal-dependent cascade.
Excessive hepatic glucose production by gluconeogenesis is a major contributor of both fasting hyperglycemia and exaggerated postprandial hyperglycemia. In vivo genetic analyses have shown that Foxo1 regulates the expression of G6Pase and PEPCK genes in response to insulin (1). In addition, PGC-1 activates an entire program of key gluconeogenic enzymes and is controlled by Foxo1 in an insulin-dependent fashion (44). Hence, the suppression of gluconeogenesis remains a very attractive pharmaceutical target in diabetes (45). Thus, our finding that bile acids inhibit expression of gluconeogenic genes for PEPCK, FBP1, and G6Pase through the interaction of SHP with HNF-4 or Foxo1 may provide a step toward the discovery of an antidiabetic drug to control hepatic gluconeogenesis.