Coordinated Control of Cholesterol Catabolism to Bile Acids and of Gluconeogenesis via a Novel Mechanism of Transcription Regulation Linked to the Fasted-to-fed Cycle*

Bile acid metabolism plays an essential role in cholesterol homeostasis and is critical for the initiation of atherosclerotic disease. However, despite the recent advances, the molecular mechanisms whereby bile acids regulate gene transcription and cholesterol homeostasis in mammals still need further investigations. Here, we show that bile acids suppress transcription of the gene (CYP7A1) encoding cholesterol 7α-hydroxylase, the rate-limiting enzyme in bile acid biosynthesis, also through an unusual mechanism not involving the bile acid nuclear receptor, farnesoid X receptor. By performing cell-based reporter assays, protein/protein interaction, and chromatin immunoprecipitation assays, we demonstrate that bile acids impair the recruitment of peroxisome proliferator-activated receptor-γ coactivator-1α and cAMP response element-binding protein-binding protein by hepatocyte nuclear factor-4α, a master regulator of CYP7A1. We also show for the first time that bile acids inhibit transcription of the gene (PEPCK) encoding phosphoenolpyruvate carboxykinase, the rate-limiting enzyme in gluconeogenesis, through the same farnesoid X receptor-independent mechanism. Chromatin immunoprecipitation assay revealed that bile acid-induced dissociation of coactivators from hepatocyte nuclear factor-4α decreased the recruitment of RNA polymerase II to the core promoter and downstream in the 3′-untranslated regions of these two genes, reflecting the reduction of gene transcription. Finally, we found that Cyp7a1 expression was stimulated in fasted mice in parallel to Pepck, whereas the same genes were repressed by bile acids. Collectively, these results reveal a novel regulatory mechanism that controls gene transcription in response to extracellular stimuli and argue that the transcription regulation by bile acids of genes central to cholesterol and glucose metabolism should be viewed dynamically in the context of the fasted-to-fed cycle.

A fundamental problem in biology is understanding the mechanisms by which extracellular stimuli transduce their signal to the nucleus and affect gene transcription. Genes are expressed in temporal and spatial pattern during development and can be regulated by hormonal and dietary stimuli or by pathogenic factors, thus contributing to disease initiation and progression. The appreciation of such mechanisms can help to unravel disease pathogenesis or even disclose targets for novel therapeutic interventions. A typical example is the mechanism whereby bile acids regulate their own synthesis by acting on transcription of the gene (CYP7A1) encoding cholesterol 7␣hydroxylase, the rate-limiting enzyme of the so-called "classic" pathway (1,2). CYP7A1 has been implicated in genetic susceptibility to atherosclerosis (3,4) because it controls the main route whereby cholesterol is removed from the body in mammals (2,5). The most relevant type of regulation of CYP7A1 is the feedback that bile acids exert on the expression of this gene (2). The complete understanding of such mechanisms represents an opportunity to describe unusual regulatory circuits controlling gene transcription and is of great value for potential biomedical applications in the design of new generations of drugs affecting cholesterol metabolism. Stroup et al. (6) first identified a bile acid-responsive element (BARE) 1 located at nucleotides Ϫ149 to Ϫ128 of CYP7A1, which contains hormone response element-like sequences recognized by several members of the nuclear receptor superfamily, such as hepatocyte nuclear factor-4␣ (HNF-4␣; NR2A1), chicken ovalbumin upstream promoter-transcription factor II (NR2F2), and ␣-fetoprotein transcription factor (FTF), also known as the CYP7A1 promoter-binding factor or liver receptor homolog-1 (NR5A2) (7)(8)(9)(10). The recent discovery of the bile acid receptor, farnesoid X receptor (FXR/BAR; NR1H4) (8,(11)(12)(13)(14), allowed an elegant model to be developed to explain the mechanisms of feedback regulation of CYP7A1 by bile acids (9,10,15). However, recent data from independent laboratories have shown that FXR, the small heterodimer partner (SHP; NR0B2), and FTF are not the only nuclear receptors that are responsible. In this regard, for instance, we have shown that bile acids inhibit the transcription of CYP7A1 by repressing the transactivation potential of HNF-4␣, a strong activator of CYP7A1, via a mitogen-activated protein (MAPK)-dependent signaling cascade (16). Furthermore, by using Shp Ϫ/Ϫ mouse models, Wang et al. (17) and Kerr et al. (18) recently demonstrated that bile acids can repress CYP7A1 transcription in an SHP-independent fashion, thus corroborating the observations made in in vitro models. These results raise the question of the molecular events underlying the additional regulatory mechanisms of CYP7A1 transcription by bile acids through the nuclear receptor HNF-4␣. A possible scenario could lie in the bile acid-induced dissociation of the HNF-4␣⅐coactivator complex. Nuclear receptors and transcription factors recruit coactivators to assemble the preinitiation complex efficiently on the core promoter of specific genes (19). Among the candidate coactivators possibly affected by bile acid signaling, we considered peroxisome proliferator-activated receptor-␥ coactivator-1␣ (PGC-1␣) and cAMP response elementbinding protein-binding protein (CBP). PGC-1␣ and CBP interact tightly with and coactivate HNF-4␣. The association of PGC-1␣ with HNF-4␣ is also a key event in the regulation of the gene encoding the gluconeogenic enzyme phosphoenolpyruvate carboxykinase (PEPCK) in response to the glucagon/cAMP cascade (20,21). Because HNF-4␣ is a key sensor of bile acid concentration in liver cells (16), we tested the hypothesis that bile acids could affect the association of HNF-4␣ with these transcription coactivators.
In what follows, we show for the first time that bile acids suppress the transcription of CYP7A1 by blocking the association of HNF-4␣ with the coactivators PGC-1␣ and CBP. Furthermore, we found that PEPCK, a target of the HNF-4␣⅐PGC-1␣ complex that mediates the response to starvation (21), is subject to the same regulation by bile acids in vitro and in vivo. We also provide strong evidence that Cyp7a1 is stimulated in parallel with Pepck in fasted mice. This work reveals bile acids as key regulators of glucose and lipid metabolism.
Cell Cultures and Transient Transfection Assays-HepG2 cells were maintained in Dulbecco's modified Eagle's medium/F-12 nutrient mixture (1:1) supplemented with 10% heat-inactivated dextran/charcoalstripped fetal calf serum. Cells were transfected by the calcium phosphate coprecipitation technique as described previously (16). Typically, coprecipitates contained 100 ng of reporter plasmid, 50 ng of receptor vector, 250 ng of coactivator expression vector or an equivalent amount of empty carrier vector (pcDNA3), and 300 ng of pCMV␤ (Clontech, Palo Alto, CA) in 24-well plates. Transfected cells were incubated for 16 h in serum-free medium in the presence of 25 M chenodeoxycholic acid (CDCA) or 10 M FXR ligand GW4064 (kindly provided by Dr. Krister Bamberg) or an equivalent amount of vehicle (0.1% ethanol or Me 2 SO, respectively). In some experiments, transfected cells were also treated with 1 mM 8-bromo-cAMP for 8 h. As a negative control for the specificity of CDCA effects on gene transcription, we also performed experiments in which transfected cells were treated with 25 M ursodeoxycholic acid (UDCA). Luciferase and ␤-galactosidase assays were performed as described previously (16) and are expressed as means Ϯ S.D. of triplicate samples. Each experiment was performed at least three times.
Real-time Quantitative Reverse Transcription-PCR-HepG2 cultures were treated with the indicated combinations of 25 M CDCA and 1 mM 8-bromo-cAMP for 8 h in serum-free medium. Triplicate samples were pooled, and total RNA was extracted with the Absolutely RNA reverse transcription-PCR miniprep kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions. About 1 g of total RNA from each sample was reverse-transcribed with Superscript II (Invitrogen s.r.l., Milan, Italy) following the manufacturer's instructions, and aliquots of the cDNAs (corresponding to ϳ10 ng of the original RNA) were subjected to real-time quantitative PCR with a SYBR Green kit (QIAGEN S.p.A., Milan) following the manufacturer's instructions to detect CYP7A1, PEPCK, HNF-4␣, SHP, and PGC-1␣ mRNAs. 18 S rRNA was used as the housekeeping gene for sample normalization and was amplified in separate wells within the same plate. Primers for real-time PCRs were designed with Primer Express software (Applied Biosystems, Monza-Milan, Italy) and optimized to work in a two-step protocol (40 cycles of amplifications each consisting of a denaturation step at 95°C for 15 s and an annealing/extension step at 60°C for 60 s). The oligonucleotides used for real-time PCR were synthesized by Eurogentec (Seraing, Belgium). 2 The specificity of the amplified products was monitored by performing melting curves at the end of each amplification. All of the amplicons generated a single peak, thus reflecting the specificity of the primers. Experiments were repeated at least twice with different cell preparations.
Chromatin Immunoprecipitations-HepG2 cells transfected with the pcDNA3-HA-PGC-1␣ plasmid were treated with 25 M CDCA or vehicle (three plates/group) for 12 h in serum-free medium, and chromatin was cross-linked with 1% formaldehyde for 10 min at room temperature. Cross-linking was stopped with 125 mM glycine, and cells were washed with ice-cold phosphate-buffered saline, scraped, and swollen on ice for 10 min in phosphate-buffered saline containing protease inhibitors (Roche Diagnostics S.p.A.). Cell extracts were prepared in SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl (pH 8.1), and protease inhibitors). Extracts were sonicated with a microtip on ice to obtain DNA fragments ranging from 200 to 1000 bp (three pulses for 30 s with the power set to 5 in a Heat Systems sonicator). Soluble chromatin fragments were centrifuged, diluted 10 times with chromatin immunoprecipitation dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl (pH 8.1), and 167 mM NaCl), precleared with protein G-Sepharose containing 200 g/ml sonicated salmon sperm DNA, and immunoprecipitated with the indicated antibodies for 14 h at 4°C. Immunocomplexes were captured on protein G-Sepharose; washed sequentially with low salt wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8.1), and 150 mM NaCl), high salt wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8.1), and 500 mM NaCl), LiCl wash buffer (0.25 M LiCl, 1% Nonidet P-40, 1% deoxycholate, 1 mM EDTA, and 10 mM Tris-HCl (pH 8.1)); and washed twice with 10 mM Tris-HCl and 1 mM EDTA (pH 8). Samples were eluted with freshly prepared 50 mM Tris-HCl, 1 mM EDTA, and 1% SDS (pH 8.1) at 65°C for 10 min, and after adjusting the concentration of NaCl to 200 mM, cross-linking was reverted at 65°C for 5 h. Following treatment with 10 g/ml RNase A and 20 g/ml proteinase K for 1 h at 45°C, genomic DNA fragments were extracted with phenol/chloroform and precipitated with ethanol. Specific genomic DNA fragments from immunoprecipitated samples and inputs were quantitated by real-time PCR with SYBR Green as indicated above. As a negative control for the specificity of CDCA effects on recruitment of coactivators to the CYP7A1 and PEPCK genes, we also performed experiments treating HepG2 cells with 25 M UDCA.
The antibodies used were anti-CBP C terminus (Upstate Biotechnology, Inc., Lake Placid, NY), anti-HA, and anti-RNA polymerase II (Santa Cruz Biotechnology). Primer sets were designed to amplify the following human genomic DNA regions: CYP7A1 BARE (from nucleotides Ϫ252 to Ϫ113 relative to the cap site), CYP7A1 upstream region (from nucleotides Ϫ1935 to Ϫ1833 relative to the cap site), CYP7A1 TATA box (from nucleotides Ϫ90 to ϩ26 relative to the cap site), CYP7A1 3Ј-untranslated region (from nucleotides 2009 to 2141 of the mRNA), PEPCK hormone response unit (HRU) (from nucleotides Ϫ518 to Ϫ263 relative to the cap site), PEPCK upstream region (from nucleotides Ϫ1318 to Ϫ1236 relative to the cap site), PEPCK TATA box (from nucleotides Ϫ77 to ϩ72 relative to the cap site), and PEPCK 3Ј-untranslated region (from nucleotides 2298 to 2408 of the mRNA). 2 Animal Studies-Male mice (C57BL/6J, 8 weeks old, five to nine animals/group) were maintained on a normal light cycle (7:00 a.m. to 7:00 p.m.) on a standard chow diet or fed a diet supplemented with 1% cholic acid for 8 days. Both groups had free access to food and water. To test the effect of fasting on gene expression, mice fed the standard chow diet were starved for 15 h and compared with mice fed the same diet. Both groups had free access to water. Livers were collected between 9:00 and 10:00 a.m., and total RNA was extracted with an RNeasy midi-kit (QIAGEN S.p.A.) and treated with RNase-free DNase to remove possible contaminating genomic DNA following the manufacturer's instructions. Specific mRNA quantitation was performed by realtime quantitative PCR as described above.
Statistical Analyses-Statistical analyses were performed with Student's t test, setting the significance at p Ͻ 0.05.

Bile Acids Repress HNF-4␣ Transcription Activity and CYP7A1 Expression via an FXR-independent Pathway-We
have previously shown that bile acids repress the transcription of CYP7A1 by affecting the transactivation potential of the liver-enriched orphan nuclear receptor HNF-4␣ via an MAPKdependent pathway (16). We wanted to verify whether the inhibitory effect of bile acids on HNF-4␣ activity is secondary to the initiation of the FXR/SHP cascade. To this end, we used a Gal4-based assay in HepG2 cells with the DNA-binding domain of the yeast transcription factor Gal4 fused with HNF-4␣ and observed that CDCA repressed the activity of this nuclear receptor (Fig. 1A, left panel). Surprisingly, the FXR agonist GW4064 (23) did not affect the transactivation potential of HNF-4␣ (Fig. 1A, left panel). The specificity of this effect was assessed in a similar assay with the Gal4-VP16 construct, which was not affected by CDCA or GW4064 (Fig. 1A, right  panel). Control experiments showed that CDCA and GW4064 did activate FXR-dependent transcription (Fig. 1B) and repressed the transcription of CYP7A1 (Fig. 1C). In parallel transfection assays, the more hydrophilic bile acid UDCA did not affect the transcription of CYP7A1 or the activity of Gal4-HNF-4␣ (data not shown). These results establish that bile acids down-regulate CYP7A1 transcription via both FXR-dependent and -independent pathways, the latter typically involving the nuclear receptor HNF-4␣.
Bile Acids Repress CYP7A1 Transcription by Targeting the HNF-4␣⅐PGC-1␣ Complex-To assess how bile acids affect the transcription machinery of CYP7A1, we studied the role of transcription coactivators. We hypothesized that bile acids affect the transcription machinery of CYP7A1 by impairing the formation of the HNF-4␣⅐coactivator complex. As a candidate coactivator involved in this process, we considered PGC-1␣, a stress-inducible coactivator strongly interacting with HNF-4␣ that is implicated in the transcription regulation of genes playing a key role in liver, skeletal muscle, and adipose tissue metabolism (21,24,25). We asked whether bile acids affect the recruitment of PGC-1␣ by HNF-4␣ to the CYP7A1 promoter when they reach a threshold concentration in hepatocytes. To test this hypothesis, we studied protein/protein interactions by co-immunoprecipitation experiments. HepG2 cells were transfected with an expression vector for HA-tagged PGC-1␣ and treated with CDCA or vehicle; whole cell extracts were immunoprecipitated with anti-HA monoclonal antibody. Captured complexes were analyzed by SDS-PAGE and Western blotting with an antibody against HNF-4␣ to evaluate the effect of CDCA on the association between PGC-1␣ and HNF-4␣. CDCA strongly hindered the physical association of HNF-4␣ with PGC-1␣ ( Fig. 2A, upper panel). Control experiments showed that CDCA did not change the protein levels of HA-PGC-1␣ and HNF-4␣ ( Fig. 2A, middle and lower panels). As expected from previous data (21), when immunoprecipitation was performed with the N-terminal portion of HNF-4␣, no interaction was detected (Fig. 2B), whereas PGC-1␣ interacted with the Cterminal part of HNF-4␣, and CDCA prevented this interaction. To further assess whether bile acids exert their action at the level of HNF-4␣ or other cofactors, we modified the coimmunoprecipitation experiment by adding in vitro synthe- sized HNF-4␣ to HepG2 cell extracts. In this case, we observed no change in the HNF-4␣/PGC-1␣ interaction with lysates from HepG2 cells treated with CDCA (Fig. 2C), suggesting that bile acids may elicit a structural or post-translational modification of the nuclear receptor that alters its affinity for PGC-1␣.
We next tested in transient transfection assays the functional relevance of the inhibition of the HNF-4␣/PGC-1␣ interaction by bile acids. HepG2 cells were transfected with the human CYP7A1 promoter-luciferase chimeric gene (22) and then challenged with CDCA, 8-bromo-cAMP (a stable analog of cAMP), or a combination of the two stimuli. CDCA strongly repressed the activity of the human CYP7A1 promoter (Fig.  2D). In contrast, cAMP strongly stimulated the transcription of luciferase driven by the CYP7A1 promoter (Fig. 2D). However, when transfected HepG2 cells were incubated in the presence of both stimuli, CDCA failed to repress the promoter activity of the CYP7A1 construct (Fig. 2D). The SV40-driven luciferase gene did not respond to any of the tested stimuli. These observations suggest that by inducing the expression of PGC-1␣ (see To confirm this hypothesis, we measured the transcription output of the CYP7A1 promoter-luciferase gene cotransfected with the expression vector for PGC-1␣ in the presence or absence of bile acids. As expected, CDCA robustly depressed the CYP7A1 promoter-driven transcription of the luciferase reporter gene (Fig. 2E). However, the ectopic expression of PGC-1␣ prevented the effect of CDCA on CYP7A1 transcription (Fig. 2E) and stimulated the basal promoter activity of CYP7A1, consistent with the previous experiment showing stimulation of the CYP7A1 promoter in the presence of cAMP. To assess whether PGC-1␣ mediates these effects directly on HNF-4␣, we performed Gal4-based assays, cotransfecting the PGC-1␣ expression vector. CDCA repressed the transcription activity of the Gal4-HNF-4␣-(1-455) protein (Fig. 2F, upper left  panel). The overexpression of PGC-1␣ boosted Gal4-HNF-4␣- dependent transcription and, most importantly, completely abolished the effect of bile acids (Fig. 2F, upper left panel). However, PGC-1␣ failed to stimulate the basal transcription activity of the C-terminal deletion mutant Gal4-HNF-4␣-(1-249), which lacks the interface for the interaction with coactivators recruited by the ligand-binding domain (LBD) (21), and did not prevent the repression by CDCA (Fig. 2F, upper right  panel). Conversely, CDCA inhibited the activity of the N-terminal deletion mutant of HNF-4␣, Gal4-HNF-4␣-(130 -455), whereas the overexpression of PGC-1␣ prevented the effect of the bile acid on the transcription activity of the receptor and stimulated the basal activity of the truncated receptor (Fig. 2F,  lower left panel). This result is consistent with the ability of the LBD of the receptor to interact with PGC-1␣ (21). Taken together, these experiments indicate that bile acids repress the transcription of CYP7A1 by impairing the protein/protein interaction of PGC-1␣ with the LBD of HNF-4␣, thus preventing the assembly of factors that promote the transcription initiation of CYP7A1. These data also suggest that two regions of HNF-4␣ could bestow the responsiveness of the receptor to bile acids, the C-terminal domain being PGC-1␣-dependent and the N-terminal domain being PGC-1␣-independent.
Bile Acids Impair the Recruitment of CBP by HNF-4␣-Because the N-terminal region of HNF-4␣ (amino acids 1-249) maintained the response to CDCA, we looked into the possibility that another coactivator interacting with this domain of the receptor may also be involved in this phenomenon. The coactivator CBP/p300 apparently interacts with both the N-and C-terminal regions of HNF-4␣ (26,27), and we therefore tested the effect of bile acids on the HNF-4␣⅐CBP complex in similar co-immunoprecipitation assays in HepG2 cells cotransfected with the HA-tagged CBP expression vector. CDCA strongly decreased the interaction of CBP with HNF-4␣ (Fig. 3A, upper  panel). Control Western blots showed that CDCA affected neither the levels of HA-CBP nor the expression of HNF-4␣ (Fig.  3A, middle and lower panels). We then analyzed the interaction of CBP with the N-and C-terminal domains of HNF-4␣ by immunoprecipitations performed with Gal4-HNF-4␣ fusion proteins and confirmed that they both interacted with CBP (Fig. 3B); most importantly, however, we found that CDCA prevented the interaction of CBP with both portions of the receptor (Fig. 3B). As with PGC-1␣, co-immunoprecipitation of in vitro synthesized HNF-4␣ with HA-CBP from control and CDCA-treated cells showed no change in the HNF-4␣⅐CBP complex (Fig. 3C).
Transcription analysis in HepG2 cells demonstrated that the overexpression of CBP prevented the feedback regulation of the CYP7A1 gene by CDCA, but, unexpectedly, did not affect the basal activity of the CYP7A1 promoter (Fig. 3D). On the other hand, the ectopic expression of CBP blocked the effect of CDCA on all of the Gal4-HNF-4␣ constructs (Fig. 3E) as distinct from PGC-1␣. CBP therefore also seems to be involved in the regulation of CYP7A1 transcription by bile acids. CBP stimulated the basal transcription activity of full-length HNF-4␣ and the C-terminal deletion mutant bearing amino acids 1-249 (Fig.  3E, upper panels), but not that of the N-terminal deletion mutant containing amino acids 130 -455 (Fig. 3E, lower  left panel).
Bile Acids and cAMP Regulate the mRNA Levels of CYP7A1-To confirm the physiological significance of our observations, we measured the mRNA levels of CYP7A1, SHP, PGC-1␣, and HNF-4␣ in HepG2 cells treated with CDCA, 8-Br-cAMP, or a combination of the two by real-time quantitative PCR. CDCA reduced the levels of CYP7A1 mRNA (Fig. 4A) and, at the same time, enhanced the expression of SHP (Fig. 4B) as previously reported (9, 10); CDCA did not decrease the levels of PGC-1␣ (Fig. 4C) and HNF-4␣ (Fig. 4D) mRNAs. Conversely, cAMP increased PGC-1␣ mRNA (Fig. 4C) due to the cAMP response element in the promoter of this coactivator (20). Consistent with the increase in PGC-1␣ expression, cAMP also enhanced the mRNA levels of CYP7A1, indicating that the cholesterol 7␣-hydroxylase gene is a target of the glucagon/ cAMP cascade in liver cells (Fig. 4A). Remarkably, when CDCA was added in combination with cAMP, it failed to repress the transcription of CYP7A1, confirming the importance of PGC-1␣ in the feedback regulation of CYP7A1 by bile acids (Fig. 4A). As expected, the FXR ligand GW4064 decreased the mRNA level of CYP7A1 and, at the same time, increased that of SHP, but did not reduce the mRNA levels of PGC-1␣ and HNF-4␣.
Bile Acids Regulate the Rate-limiting Enzyme of Gluconeogenesis in the Liver-The demonstration that bile acids target the protein/protein interaction between HNF-4␣ and the coac- tivators PGC-1␣ and CBP/p300 prompted us to investigate whether this type of regulation may also apply to other hepatic genes. The liver gene PEPCK, which encodes the enzyme catalyzing the rate-limiting step of gluconeogenesis, is under the control of the HNF-4␣⅐PGC-1␣ complex, which is recruited to a composite HRU (20,21). We found that CDCA strongly reduced the mRNA level of PEPCK, whereas cAMP increased it (Fig.  5A). However, CDCA did not decrease the level of PEPCK mRNA when cells were simultaneously treated with cAMP (Fig. 5A). We confirmed these results by measuring the transcription output of the PEPCK promoter in the presence or absence of bile acids. CDCA strongly repressed the promoter activity of the PEPCK promoter-luciferase fusion gene (Fig.  5B); however, when PGC-1␣ was cotransfected, the basal promoter activity increased 3-fold, and CDCA no longer affected the PEPCK promoter (Fig. 5B). Similarly, the overexpression of CBP prevented the CDCA inhibition of PEPCK transcription (Fig. 5B), although CBP did not stimulate the basal promoter activity of PEPCK, as in the case of the CYP7A1 promoter (Fig.  3D). As expected, the FXR agonist GW4064 did not decrease the mRNA level (Fig. 5A) and transcription (Fig. 5C) of PEPCK, demonstrating that bile acids can affect gene transcription also through an FXR-independent signaling pathway. In parallel transfection assays, the more hydrophilic bile acid UDCA did not affect the transcription of PEPCK (data not shown).
To prove that HNF-4␣ is the target of the signaling cascade elicited by CDCA that leads to the repression of PEPCK transcription, we performed reporter gene assays with mutants of the PEPCK promoter in which the AF1 and AF3 sequences, the binding sites for HNF-4␣, were substituted by Gal4-binding sites (21). This substitution blunted the CDCA-mediated repression of transcription (mutants gAF1, gAF3, and gAF1/3) (Fig. 5D). However, the ectopic expression of Gal4-HNF-4␣, which can bind to the Gal4-binding sites introduced in the PEPCK promoter, restored the repression by bile acids (Fig.  5D). The mutant of the AF2 sequence, which brings about a negative response of PEPCK to insulin (28), retained the response to bile acids. It is noteworthy that the basal transcription activity of the gAF1 and gAF3 mutants was lower than that of the wild-type promoter and that the cotransfection of Gal4-HNF-4␣ strongly elevated the basal transcription of these constructs. Altogether, our results indicate for the first time that bile acids returning to the liver also regulate PEPCK and that, by analogy with the CYP7A1 promoter, the nuclear receptor HNF-4␣ and the coactivators PGC-1␣ and CBP play a crucial role in this type of regulation.
Bile Acids Decrease the Recruitment of PGC-1␣, CBP, and RNA Polymerase II to the CYP7A1 and PEPCK Promoters in the Context of Chromatin-To definitely prove that CDCA down-regulates the transcription of CYP7A1 and PEPCK by selectively impairing the recruitment of PGC-1␣ and CBP by HNF-4␣ binding to these two promoters in the native chroma-  's t test). D, mutation analysis of the HRU in the PEPCK promoter. The AF1, AF2, and AF3 sites in the PEPCK promoter were replaced by a Gal4-binding site and tested for response to CDCA in HepG2 cells. The specificity of the CDCA effect was assessed using the SV40 promoter-luciferase construct. *, p Ͻ 0.05; and #, p Ͻ 0.05 versus Gal4-cotransfected cells and control vehicle, respectively (Student's t test). wt, wild type; RLU, relative light units. tin environment, we carried out chromatin immunoprecipitation assays with extracts from HepG2 cells treated with CDCA or vehicle. Cell extracts were immunoprecipitated with antibodies selective for PGC-1␣ and CBP, and the amount of immunoprecipitated DNA fragments containing the BARE in CYP7A1 and the HRU in PEPCK was quantitated by real-time PCR. CDCA significantly decreased the amount of PGC-1␣ and CBP recruited to the BARE and HRU of CYP7A1 and PEPCK, respectively (Fig. 6, A and B, upper panels). The more hydrophilic bile acid UDCA did not affect the recruitment of PGC-1␣ and CBP to the CYP7A1 and PEPCK promoters (data not shown). No signal was detected with the upstream sequences of the two genes (data not shown), indicating that these antibodies immunoprecipitated selective regions of these two genes associated with PGC-1␣ or CBP. Moreover, CDCA decreased the amount of RNA polymerase II, PGC-1␣, and CBP recruited to the core promoters (Fig. 6, A and B, middle panels) as well as the amount of these proteins found downstream in the 3Јuntranslated regions (lower panels), reflecting the reduction of gene transcription. Therefore, these results demonstrate that bile acids act upon the transcription of these two genes also in the native context of chromatin by decreasing the amount of these coactivators and RNA polymerase II associated with specific regions of these genes.
In Vivo Regulation of Cyp7a1 and Pepck by Bile Acids and in the Fasted State in Mice-To substantiate the in vivo relevance of the results obtained with cell cultures, we performed experiments with mice fed a 1% cholic acid diet and with mice undergoing an overnight fast. In the first experiment, the analysis of mouse Cyp7a1 and Pepck gene expression, performed by real-time quantitative PCR, demonstrated that bile acids decreased the mRNA levels of these two genes in vivo (Fig. 7A) and, at the same time, enhanced the expression of SHP (data not shown). In the second experiment, the mRNA levels of Cyp7a1 and Pepck in fasted mice were strongly increased (Fig.  7B) along with PGC-1␣ (data not shown), therefore confirming that these two genes are positively regulated in vivo by the glucagon/cAMP cascade triggered during the fasted state. These data provide strong in vivo evidence that bile acids negatively regulate the expression of key genes in cholesterol and glucose metabolism and establish for the first time that Cyp7a1 is stimulated in fasted mice along with Pepck. Taken together, these results suggest that the coordinated regulation of these two genes by bile acids may be linked to the fasted-tofed cycle. DISCUSSION In this study, we have provided multiple evidences for a novel molecular mechanism of transcription regulation of key genes in cholesterol catabolism and in gluconeogenesis (Fig. 8). According to this model, bile acids affect the recruitment of the coactivators PGC-1␣ and CBP by HNF-4␣ to the promoters of CYP7A1 and PEPCK.
Based on these observations, it is necessary to reconsider and broaden our view on the way bile acids affect gene transcription. Bile acids tightly control their own synthesis by multiple signaling cascades that converge on the promoter of the gene CYP7A1, which encodes the rate-limiting enzyme of the classic pathway. Such stringent control is necessary because an excessive output of bile acids may be detrimental to several tissues, so multiple biochemical mechanisms have evolved to fine-tune their synthesis. The FXR/SHP/FTF and HNF-4␣/PGC-1␣/CBP pathways may complement each other for this purpose. 3 The other remarkable finding in this study is that bile acids not only affect their own synthesis, but also control the transcription of genes involved in other metabolic pathways such as gluconeogenesis (Fig. 8). Considering that glucose output from the liver is often increased in type 2 diabetes (29), contributing to the exacerbation of this disease, our results on the inhibition of PEPCK transcription by bile acids provide an opportunity to exploit this signaling cascade as a potential target for novel antidiabetic agents.
The regulation of CYP7A1 and PEPCK by opposing stimuli such as bile acids and cAMP, the first mimicking postprandial conditions and the latter the fasted state, suggests an intriguing hypothesis according to which the fasted-to-fed cycle regulates apparently unrelated metabolic pathways in a coordinated fashion. In light of these observations, we propose that the regulation of gene transcription by bile acids should be viewed dynamically in the context of the fasted and fed states. After a prolonged fasting period, the transcription of CYP7A1 and PEPCK raises probably as a combination of the stimulation by the glucagon/cAMP cascade and the concomitant decrease in the concentration of bile acids returning to the liver. On one hand, this may help to prepare the gastrointestinal tract for the digestion and absorption of fats in a subsequent meal and, on the other, to increase gluconeogenesis and to buffer the falling plasma concentration of glucose during the fasted state. Conversely, in the fed state, as the concentration of bile acids fluxing through the enterohepatic circulation increases, the reduction of CYP7A1 and PEPCK transcription may be secondary to the drop in the glucagon level and to the direct inhibition elicited by bile acids, which are massively secreted into the duodenum and return to the liver at higher concentrations than during a prolonged fasting period. This is the first clear evidence demonstrating that CYP7A1 transcription is enhanced during the fasted state. Previous reports showed contradicting results that may be explained by the different experimental systems used in those studies (30,31). However, our in vitro and in vivo results strongly argue for this type of regulation. In this respect, PGC-1␣ is the key factor contributing to the stimulation of CYP7A1 and the other target gene, PEPCK, because it is strongly induced by the glucagon/ cAMP cascade (20).
Besides PGC-1␣, CBP is also involved, although the two coactivators play different roles in the feedback regulation of CYP7A1 and PEPCK by bile acids. PGC-1␣ interacts exclusively with the C-terminal domain of HNF-4␣ and is displaced upon exposure of hepatocytes to bile acids. Conversely, CBP interacts with both the N-and C-terminal regions of the receptor. The involvement of PGC-1␣ and CBP in this type of regulation appears to be specific because the overexpression of other coactivators such as steroid receptor coactivator-1 and transcription intermediary factor-2 did not overcome the effect of bile acids on the HNF-4␣-dependent transcription of these genes (data not shown). At present, we still cannot explain how bile acids can induce the dissociation of these coactivators from HNF-4␣. However, the observation that in vitro synthesized HNF-4␣ does not dissociate from coactivators present in extracts from cells challenged with bile acids argues for possible post-translational or structural modifications of the receptor induced by bile acids. In this regard, it should be mentioned that the interaction of CBP with the N-terminal domain of HNF-4␣ determines the acetylation of lysine residues within the nuclear localization sequence, which is crucial for nuclear localization, DNA binding, and interaction with CBP itself (32). It is possible that bile acids may prevent the acetylation of these lysine residues and consequently decrease the activity of the receptor as well as its affinity for CBP. It will be interesting to verify whether similar or other post-translational modifications can affect the recruitment of coactivators in the C-terminal domain of HNF-4␣ and whether bile acids can somehow cause these modifications.
It was somewhat surprising to find that the overexpression of CBP could not enhance the transcription of these two genes, as opposed to PGC-1␣. This may reflect the requirement of PGC-1␣ docking to the LBD of HNF-4␣ as a prerequisite for the proper interaction of CBP with and coactivation of HNF-4␣ through the LBD (33). The endogenous levels of PGC-1␣ in HepG2 cells may therefore be insufficient to cooperate with CBP in the coactivation of HNF-4␣ via the LBD. On the other hand, the interaction of CBP with the N-terminal fragment of HNF-4␣ does not require the presence of PGC-1␣ (27); thus, it could be activated by CBP in the Gal4-based assay. However, in the context of the PEPCK and CYP7A1 promoters, the interaction of CBP with the N-terminal domain of HNF-4␣ does not seem to be sufficient in itself to enhance the expression of these two genes, but requires the presence of PGC-1␣.
The data presented here open the question as to which of the two bile acid-elicited signaling pathways is more relevant for the negative regulation of CYP7A1 and PEPCK gene transcription. The following considerations suggest that the HNF-4␣ pathway may be more critical for these two genes. First, bile acids can still decrease the mRNA levels of Cyp7a1 in Shp Ϫ/Ϫ 3 While this manuscript was under revision, Holt et al. (38) reported a novel mechanism of transcription regulation by bile acids mediated by FXR and fibroblast growth factor-19, a secreted growth factor that signals through the fibroblast growth factor receptor-4 cell-surface receptor tyrosine kinase and in turn strongly suppresses the expression of CYP7A1 in hepatocytes through a c-Jun N-terminal kinase-dependent pathway.  II). B, bile acid-induced repression of CYP7A1 and PEPCK transcription is secondary to the dissociation of the HNF-4␣⅐coactivator complex that causes the faulty recruitment of general transcription factors and activation of RNA polymerase II. According to this model, the effect of bile acids does not depend on the activation of CYP7A1 transcription by oxysterols via liver X receptor (9) or PEPCK transcription by glucocorticoid hormones (34). mice. Second, PEPCK does not contain liver receptor homolog-1/FTF-binding sites, and its inhibition by bile acids relies solely on HNF-4␣ binding to the AF1 and AF3 sequences in the promoter. Thus, the PEPCK promoter may be considered as a natural "variant" of the CYP7A1 BARE that responds to bile acids exclusively via HNF-4␣ (Fig. 8). This would also explain why the selective FXR ligand GW4064 did not repress the transcription of PEPCK because FTF, the target of SHP, does not interact with this promoter. This result is in conflict with a recent report showing that PEPCK is a novel target for SHP inhibition (34). However, it should be recalled that the negative effect of SHP was observed in cells stimulated by glucocorticoid hormones, whereas here we studied the effect of bile acids and the selective FXR agonist on PEPCK transcription under basal conditions. Third, because inhibition via FXR/SHP/FTF requires de novo synthesis of SHP through ligand-bound FXR, it is likely that the inhibition mediated by HNF-4␣ precedes others in time because it is modulated by rapid events involving an MAPK pathway (16). This view is supported by the indication that HNF-4␣ seems to undergo regulation of its own activity via a yet unidentified post-translational modification as an essential prerequisite for the bile acid-induced dissociation of the HNF-4␣⅐coactivator complex. Several groups have already shown that post-translational modifications can affect the activity of HNF-4␣ in different ways (32,(35)(36)(37). The FXR/SHP pathway may, however, have a role in reinforcing and supporting the repression already established through HNF-4␣.
In conclusion, we have described a novel mechanism of transcription regulation affecting key genes in bile acid synthesis and gluconeogenesis that explains why Shp Ϫ/Ϫ mice are still responsive to bile acids. The discovery that bile acids also repress the transcription of PEPCK, the rate-limiting enzyme of gluconeogenesis, will stimulate new studies on the coordinated control of different metabolic pathways. This study discloses new target mechanisms for the design of novel treatments of metabolic diseases such as hyperlipidemia and diabetes.