Glucagon and cAMP Inhibit Cholesterol 7�-hydroxylase (CYP7A1) Gene Expression in HumanHepatocytes: Discordant Regulation of Bile Acid

Synthesis and GluconeogenesisKwang-Hoon Song and John Y. L. Chiang

The gene encoding cholesterol 7�-hydroxylase (CYP7A1) is tightly regulated to control bileacid synthesis and maintain lipid homeostasis. Recent studies in mice suggest that bile acidsynthesis is regulated by the fasted-to-fed cycle, and fasting induces CYP7A1 gene expressionin parallel to the induction of peroxisome proliferators-activated receptor � co-activator 1�(PGC-1�) and phosphoenolpyruvate carboxykinase (PEPCK). How glucagon regulatesCYP7A1 gene expression in the human liver is not clear. Here we show that glucagon andcyclic adenosine monophosphate (cAMP) strongly repressed CYP7A1 mRNA expression inhuman primary hepatocytes. Reporter assays confirmed that cAMP and protein kinase A(PKA) inhibited human CYP7A1 gene transcription, in contrast to their stimulation of thePEPCK gene. Mutagenesis analysis identified a PKA-responsive region located within thepreviously identified HNF4� binding site in the human CYP7A1 promoter. Glucagon andcAMP increased HNF4� phosphorylation and reduced the amount of HNF4� present inCYP7A1 chromatin. Our findings suggest that glucagon inhibited CYP7A1 gene expressionvia PKA phosphorylation of HNF4�, which lost its ability to bind the CYP7A1 gene andresulted in inhibition of human CYP7A1 gene transcription. In conclusion, this studyunveils a species difference in nutrient regulation of the human and mouse CYP7A1 gene andsuggests a discordant regulation of bile acid synthesis and gluconeogenesis by glucagon inhuman livers during fasting. (HEPATOLOGY 2006;43:117-125.)

The conversion of cholesterol to bile acids in theliver is a major pathway for removing excess cho-lesterol from the body and plays an important

role in maintaining cholesterol homeostasis.1,2 Imbalanceof cholesterol metabolism causes diseases such as athero-

sclerosis and gallstone disease. CYP7A1 is a liver-specificenzyme that catalyzes the first and rate-limiting step in thebile acid biosynthetic pathway.1 Expression of CYP7A1 isregulated mainly at the gene transcriptional level by manyfactors, including bile acids, diets, nutrients, cytokines,and hormones.3

Several nuclear receptors, including liver orphan recep-tor, �-fetoprotein transcription factor (also known asmouse liver–related homolog), hepatocyte nuclear factor4� (HNF4�), and pregnane X receptor, bind to theCYP7A1 gene and play important roles in regulation ofCYP7A1 gene transcription.1,4 HNF4� is the most abun-dant orphan nuclear receptor expressed in the liver thatbinds to a direct repeat with one base spacing (DR1) inthe bile acid response element II (BARE-II) (�144/�126) and stimulates CYP7A1 gene transcription.5 Mu-tation of this HNF4� binding site markedly reducedCYP7A1 promoter activity, and liver-specific conditionaldisruption of the HNF4� gene in mice drastically re-pressed CYP7A1 gene expression, indicating that HNF4�is crucial for CYP7A1 gene transcription and regula-

Abbreviations: HNF4�, hepatocyte nuclear factor 4�; CYP7A1, cholesterol 7�-hydroxylase; Luc, luciferase; BARE II, bile acid response element II; FXR, farnesoidX receptor; SHP, small heterodimer partner; PGC-1�, peroxisome proliferators-activated receptor � co-activator 1�; cAMP, cyclic adenosine monophosphate; PKA,protein kinase A; PEPCK, phosphoenolpyruvate carboxykinase; 8-Br-cAMP,8-Bromo-cyclic AMP; PCR, polymerase chain reaction; ChIP, chromatin immuno-precipitation.

From the Department of Biochemistry and Molecular Pathology, NortheasternOhio Universities College of Medicine, Rootstown, OH.

Received July 18, 2005; accepted August 29, 2005.Supported by NIH grants DK58379 and DK44442.Address reprint requests to: John Y. L. Chiang, Ph.D., Department of Biochem-

istry and Molecular Pathology, Northeastern Ohio Univ. College of Medicine, 4209State Route 44, P.O. Box 95, Rootstown, OH 44272. E-mail:jchiang@neoucom.edu; fax: 330-325-5915.

Copyright © 2005 by the American Association for the Study of Liver Diseases.Published online in Wiley InterScience (www.interscience.wiley.com).DOI 10.1002/hep.20919Potential conflict of interest: Nothing to report.

117

tion.5,6 Bile acid receptor farnesoid X receptor (FXR) in-hibits CYP7A1 gene transcription by inducing a negativenuclear receptor small heterodimer partner (SHP), whichthen inhibits �-fetoprotein transcription factor andHNF4� transactivation of the CYP7A1 gene.1 FXR reg-ulates a variety of genes involved in bile acid, lipoprotein,glucose, and triglyceride metabolisms.7

Although many studies about physiological responsesof CYP7A1 to cholesterol, bile acids, and hormones areconducted in animal models such as mouse and rat, verylittle is known regarding human CYP7A1 gene regula-tion. Moreover, recent studies reported that the CYP7A1gene is differentially regulated in different species. Nota-bly, a high-cholesterol diet induces CYP7A1 gene expres-sion in mice and rats with little change in plasmacholesterol levels, but cholesterol has no significant effector even suppresses CYP7A1 expression in rabbit, hamster,African green monkeys, and transgenic mice expressinghuman CYP7A1.8-11 The stimulation of CYP7A1 geneexpression by cholesterol is accomplished via liver orphanreceptor �, which binds to the CYP7A1 gene promoter inmice and rats, but not in humans.12,13

A recent study reported that bile acids impair HNF4�recruitment of peroxisome proliferators-activated recep-tor � co-activator 1� (PGC-1�) to both CYP7A1 andPEPCK genes and inhibits their gene transcription14 andsuggests that bile acid synthesis and gluconeogenesis arecoordinately regulated by bile acid feedback via a mecha-nism linked to the fasted-to-fed cycle.14 A similar studyshows that CYP7A1 mRNA expression increases in par-allel with PGC-1� in fasted mice.15 These results are indirect contradiction to the earlier reports that CYP7A1expression is repressed in piglets16 and rats17-19 duringfasting. It appears that nutritional status that alters hor-monal levels such as insulin and glucagon may regulateCYP7A1 expression. In the current study, we demon-strated that glucagon and cyclic adenosine monophos-phate (cAMP) repressed human CYP7A1 gene expressionvia protein kinase A (PKA) phosphorylation of HNF4�in primary human hepatocytes and HepG2 cells, the bestavailable models for studying human gene regulation.This result supports a species difference in regulation ofhuman CYP7A1 gene transcription by hormones and un-derscores the importance of studying human CYP7A1gene regulation.

Materials and Methods

Cell Culture. Primary human hepatocytes were iso-lated from human donors (HH1183, 54-year-old male;HH1189, 18-year-old male; HH1196, 56-year-old fe-male; HH1201, 69-year-old male; HH1205, 45-year-old

male; HH1209, 50-year-old female; HH1210, 35-year-old female; HH1211, 79-year-old male; HH1215, 36-year-old male; HH1220, 40-year-old female) and wereobtained from the Liver Tissue Procurement and Distri-bution System of National Institutes of Health (S. Strom,University of Pittsburgh, Pittsburgh, PA). Cells weremaintained in Hank’s modified medium modified Wil-liams E medium (Cambrex Bioscience, Inc., Walkersville,MD) supplemented with 0.1% gentamicin sulfate andamphotericin-B without insulin or dexamethasone. TheHepG2 and HEK293 cells were obtained from the Amer-ican Type Culture Collection (Manassas, VA). The cellswere maintained in a 1:1 mixture of Dulbecco’s modifiedEagle’s medium and F-12 (50:50; Life Technologies, St.Paul, MN) supplemented with 100 U/mL penicillinG/streptomycin sulfate (Celox, Hopkins, MN) and 10%heat-inactivated fetal bovine serum (Irvine Scientific,Santa Ana, CA).

Plasmids. The mammalian expression plasmids forHNF4�, wild-type PKA catalytic subunit, mutant PKA thatlacks the kinase activity, and pHNF4�-tk-Luc reporter thatcontains 4 copies of the HNF4� binding site in mouse tran-sthyretin gene fused upstream of the thymidine kinase (tk)minimal promoter and luciferase gene were as previouslydescribed.4,20,21 A series of human CYP7A1 promoter lucif-erase reporters were previously constructed.22 A humanphosphoenolpyruvate carboxykinase (PEPCK) promoter lu-ciferase reporter construct containing 2 kb upstream se-quence (pPEPCK�2,000/�73) was provided by RichardHanson (Case Western Reserve University, Cleveland,OH).23

RNA Isolation and Real-Time Quantitative Poly-merase Chain Reaction. Primary human hepatocyteswere treated with 100 nmol/L glucagon or 1 mmol/L8-Bromo-cyclic AMP (8-Br-cAMP) for a period from 1 to24 hours or as indicated. Total RNA was isolated usingTri-reagent (Sigma, St. Louis, MO) according to themanufacturer’s instructions. Reverse-transcription reac-tions were performed using RETROscript kit (Ambion,Austin, TX) following the manufacturer’s instructions.Approximately 2 �g total RNA from each sample wasreverse-transcribed, and aliquots of the cDNA were sub-jected to real-time quantitative polymerase chain reaction(PCR) with a Taqman Universal PCR Master Mix(Roche, Branchburg, NJ) following the manufacturer’sinstructions to detect CYP7A1, PEPCK, PGC-1�, andHNF4� mRNAs. Amplification of ubiquitin C was usedin the same reactions as an internal reference gene. Taq-man probes for real-time quantitative PCR were orderedfrom Assay on Demand (Applied Biosystems, Forest City,CA). All PCR reactions were done in triplicate. PCR am-plification was performed as follows: 50°C for 2 minutes,

118 SONG AND CHIANG HEPATOLOGY, January 2006

95°C for 10 minutes, 95°C for 15 seconds, and 40 cycles(1 minute each) at 60°C using an ABI PRISM 7500 se-quence detector (Applied Biosystems). Amplification datawere analyzed using the Sequence Detector version 1.7software (Applied Biosystems). Relative mRNA expres-sion levels were calculated using the ���Ct method rec-ommended by Applied Biosystems (User Bulletin no. 2,1997). Statistical analysis of real-time PCR results weredone using mean normalized cycle threshold (�Ct) valuesand the pooled standard deviation of the mean �Ct,which were analyzed by one-way ANOVA followed byTukey’s Honestly Significant post hoc test. A P value of lessthan .05 was considered as a statistically significant differ-ence. Experiments were repeated at least 3 times withdifferent cell preparations.

Luciferase Reporter Assay. For luciferase reporter as-say, HepG2 cells were plated in 24-well plates 24 hoursbefore transfection with reporter or expression plasmidsusing LipofectAMINE 2000 reagent (Life Technologies,Inc., Gaithersburg, MD) according to the manufacturer’sinstructions. Total DNA used in each transfection wasadjusted by adding the appropriate amount of pcDNA3vector. When studying the effect of 8-Br-cAMP, cellswere serum-starved for 24 hours before treatment. Lucif-erase activities are expressed as relative luciferase unit/�-galactosidase activity as previously described.4

In Vivo Phosphorylayion of HNF4�. Human pri-mary hepatocytes were treated with PKA inducers (100nmol/L glucagon or 1 mmol/L 8-Br-cAMP) for 6 hours.Cells were washed with phosphate-buffered saline andlysed in ice-cold modified radioimmunoprecipitationbuffer (50 mmol/L Tris-HCl, pH 7.4, 1% Nonidet P-40,0.25% Na�-deoxycholate, 150 mmol/L NaCl, 1 mmol/Lphenylmethylsulfonyl fluoride, 1 mmol/L EDTA, 1mmol/L sodium orthovanadate (Na3VO4), 1 mmol/LNaF), and Complete Protease Inhibitor cocktail (Sigma).Cells debris were centrifuged, and the supernatant dilutedwith 3 volumes of radioimmunoprecipitation bufferwithout NaCl was pre-cleared with whole rabbit serumadsorbed on Protein A-Sepharose beads (Amersham Bio-sciences, Piscataway, NJ) and subsequently subjected toimmunoprecipitation with 10 �g anti-HNF4� or non-immune serum overnight at 4°C. The beads were washedseveral times with radioimmunoprecipitation buffer con-taining 150 mmol/L NaCl, and the immuno-complexeswere subjected to electrophoresis on a 10% sodium dode-cyl sulfate polyacrylamide gel and then transferred to anitrocellulose membrane (Amersham Biosciences). En-hanced chemiluminescence Western blotting (AmershamBiosciences) was performed according to the manufactur-er’s instructions. Phosphor-HNF4� and HNF4� pro-teins were detected by incubation of blots with an anti-

phosphorprotein antibody (1:2,000 dilution; Zymed,South San Francisco, CA) and anti-HNF4� antibody (1:2,000 dilution; Santa Cruz Biotechnology, Santa Cruz,CA), respectively.

Chromatin Immunoprecipitation Assay. Chromatinimmunoprecipitation (ChIP) assays were performed using aChIP Assay kit (Upstate Cell Signaling Solutions, LakePlacid, NY) according to the manufacturer’s instructions.Human primary hepatocytes were treated with 100 nmol/Lglucagon or 1 mmol/L 8-Br-cAMP for various times, andchromatin was cross-linked in 1% formaldehyde and soni-cated as previously reported.4 Cell lysate solution (5%) inChIP dilution buffer was kept aside as “input.” Ten micro-grams HNF4� antibody (Santa Cruz Biotechnology) wasadded to precipitate DNA–protein complexes and non-im-mune IgG was used as a control. A 391-bp DNA fragment(�432 to �41) containing the BARE-I and BARE-II of theCYP7A1 promoter was PCR amplified for 30 cycles using 5�L DNA as template and analyzed on a 1.5% agarose gel.PCR primers for amplifying were as follows: 5�-ATCAC-CGTCTCTCTGGCAAAGCAC-3�; reverse primer: 5�-CCATTAACTTGAGCTTGGTTGACAAAG-3�.

Results

Glucagon and cAMP Repressed CYP7A1 mRNAExpression in Human Primary Hepatocytes. To un-derstand the nutrient regulation of human CYP7A1 geneexpression, we first investigated the effect of glucagon onCYP7A1 gene expression in primary cultures of humanhepatocytes. Treatment of human primary hepatocyteswith glucagon (100 nmol/L) caused a time-dependentdecrease in CYP7A1 mRNA levels to 10% of the controlin 24 hours (Fig. 1A). As expected, glucagon rapidly in-creased PEPCK mRNA levels by 10- to 20-fold, but didnot significantly change HNF4� mRNA levels. Interest-ingly, glucagon induced PGC-1� mRNA levels by 10-fold in 3 hours and 3-fold in 24 hours.

That glucagon induces cAMP, which acts as the secondmessenger that activates protein kinase A (PKA), whichphosphorylates the downstream targets of glucagon, hasbeen established. We hypothesized that the cAMP path-way might play a role in the glucagon-mediated repressionof CYP7A1. Therefore, human primary hepatocytes weretreated with 1 mmol/L 8-Br-cAMP, a stable analog ofcAMP, and analyzed for CYP7A1 mRNA levels. Asshown in Fig. 1B, CYP7A1 mRNA levels were stronglysuppressed by 8-Br-cAMP, similar to glucagon treatment.As a positive control experiment, 8-Br-cAMP inductedPEPCK and PGC-1� mRNA levels by approximately 15-to 20-fold. This compound had no significant effect onHNF4� mRNA expression.

HEPATOLOGY, Vol. 43, No. 1, 2006 SONG AND CHIANG 119

Because time-course study showed that glucagon and8-Br-cAMP strongly repressed CYP7A1 but inducedPEPCK and PGC-1� mRNA levels after 6 hours’ treat-ment, we analyzed the dose response of glucagon and8-Br-cAMP at this point of treatment. Treatment of hu-man primary hepatocytes with glucagon (1-100 nmol/L)or 8-Br-cAMP (0.01-1 mmol/L) showed a dose-depen-dent suppression of CYP7A1 mRNA levels; in contrast,these 2 compounds dose-dependently induced PEPCKand PGC-1� mRNA levels. HNF4� mRNA levels werenot affected by these 2 compounds (Fig. 1C-D).

To confirm the specificity of the PKA signaling path-way in mediating glucagon inhibition of CYP7A1 geneexpression, several specific inhibitors of signaling path-ways were used to treat human primary hepatocytes 1hour before addition of glucagon (100 nmol/L). Real-time PCR analysis indicated that pretreatment of a PKAinhibitor, H-89 (10 �mol/L) prevented glucagon inhibi-tion of CYP7A1 mRNA expression (Fig. 1E). However,there was no significant effect of Ro 31-8220 (an inhibitorof protein kinase C), PD98059 (an inhibitor of mitogen-activated protein kinase, MAPK), SB-203580 (an inhib-itor of stress-activated protein kinase), and SP600125 (aninhibitor of c-jun N-terminal kinase) on CYP7A1 mRNAexpression. These results indicated that the PKA pathwayactivated by glucagon specifically inhibited CYP7A1mRNA expression in human primary hepatocytes.

PKA Repressed Human CYP7A1 Promoter Activ-ity. To further study the effect of the PKA pathway inrepressing CYP7A1 gene transcription, transient transfec-tion assay of human CYP7A1-luciferase reporter(CYP7A1-Luc) activity was performed in HepG2 cells.Treatment of 8-Br-cAMP (1 mmol/L) repressed the

4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™Fig. 1. Expression of CYP7A1, PEPCK, PGC-1�, and HNF4� mRNA in

human primary hepatocytes treated with glucagon (A,C) and 8-Br-cAMP(B,D). (A) Human primary hepatocytes (HH1196, HH1201, HH1205,HH1209) were treated with glucagon (100 nmol/L) for the times indi-cated. (B) Human primary hepatocytes (HH1183, HH1189, HH1196)were treated with 8-Br-cAMP (1 mmol/L) for the times indicated. (C)Human primary hepatocytes (HH1196, HH1201, HH1205, HH1209)were treated with glucagon in the concentrations indicated for 6 hours,and (D) human primary hepatocytes (HH1183, HH1189, HH1196) weretreated with 8-Br-cAMP at the concentrations indicated for 6 hours. (E)Human primary hepatocytes (HH1209, HH1210, HH1211) were treatedwith glucagon inhibitors H-89 (10 �mol/L), Ro 31-8220 (1 �mol/L),PD98059 (20 �mol/L), SB-203580 (25 �mol/L), or SP600125 (25�mol/L) 1 hour before adding glucagon (100 nmol/L). Glucagon (100nmol/L) or 8-Br-cAMP (1 mmol/L) alone was treated as a control. TotalRNA was isolated for real-time quantitative polymerase chain reactionanalysis of CYP7A1, PEPCK, PGC-1�, and HNF4� mRNA levels. Numbersabove the bars indicate the levels of mRNA relative to controls. Datarepresent the mean � SD of at least 3 individual experiments. CYP7A1,cholesterol 7�-hydroxylase; PEPCK, phosphoenolpyruvate carboxykinase;PGC-1�, peroxisome proliferators-activated receptor � co-activator 1�;HNF4�, hepatocyte nuclear factor 4�, 8-Br-cAMP, 8-Bromo-cyclic AMP.

120 SONG AND CHIANG HEPATOLOGY, January 2006

CYP7A1 reporter activity by approximately 60%, butstimulated a human PEPCK reporter (PEPCK-Luc) ac-tivity by approximately 10-fold (Fig. 2A-B). This PEPCKreporter has all important response elements (i.e., glu-cocorticoid receptor, CCAAT Response Element Bindingprotein, HNF4, and forkhead transcription factor Foxo1)that are known to regulate PEPCK gene transcription. Wenext co-transfected an expression plasmid for PKA cata-lytic subunits to test the role of PKA in mediating gluca-gon and cAMP effect on human CYP7A1 reporteractivity. As shown in Fig. 2C, addition of increasingamounts of wild-type PKA expression plasmids markedlyinhibited CYP7A1 reporter activity, whereas an expres-sion plasmid for mutant PKA20 did not affect the reporteractivity. As a positive control, wild-type PKA stimulatedPEPCK-Luc activity, whereas the mutant PKA had noeffect (Fig. 2D). These data demonstrated that activationof PKA repressed CYP7A1 but stimulated PEPCK genetranscription, consistent with the effect of cAMP onCYP7A1 and PEPCK mRNA expression (Fig. 1).

PKA Repression of Human CYP7A1 Is LiverSpecific. CYP7A1 is only expressed in the hepatocytes.To test whether glucagon/cAMP/PKA-mediated repres-sion of CYP7A1 is liver specific, we studied the effect ofPKA on CYP7A-Luc reporter activity in the liver-derivedHepG2 cells and kidney-derived HEK293 cells. In HepG2cells, co-transfection of the wild-type PKA expression plas-mid strongly repressed CYP7A1-Luc reporter activity, butthe mutant PKA that lacks the kinase activity did not haveany effect on CYP7A1 reporter activity (Fig. 3A). InHEK293 cells, the basal reporter activity is much lower thanin HepG2 cells. Co-transfection of either a wild-type or mu-tant PKA expression plasmid did not significantly modulateCYP7A1 reporter activity (Fig. 3B). In contrast, PEPCK re-porter activity was strongly stimulated by wild-type PKA inboth cell types, and mutant PKA had no effect (Fig. 3C-D).These data indicated that PKA-mediated repression ofCYP7A1 expression was liver specific, whereas PEPCK,which is expressed in both liver and kidney, was regulated byPKA in both liver and kidney cells.

Fig. 2. Effects of 8-Br-cAMP and PKA on human CYP7A1-Luc and PEPCK-Luc reporter activity. HepG2 cells were transfected with CYP7A1-luciferasereporter (CYP7A1-Luc) (A) and PEPCK-luciferase reporter (PEPCK-Luc) (B). After transfection, cells were serum-starved for 24 hours in the presenceor absence of 8-Br-cAMP (1 mmol/L). HepG2 cells were transfected with CYP7A1-Luc (C) and PEPCK-Luc (D) with increasing amounts of wild-typeor mutant PKA expression plasmids (50, 100, and 200 ng). Luciferase activity was normalized to �-galactosidase activity. All experiments were donein duplicates, and data represent the mean � SD of 3 individual experiments. 8-Br-cAMP, 8-Bromo-cyclic AMP; PKA, protein kinase A; CYP7A1,cholesterol 7�-hydroxylase; PEPCK, phosphoenolpyruvate carboxykinase.

Fig. 3. Cell-specific regulation of CYP7A1 promoter reporter activity by PKA. HepG2 (A,C) and HEK 293 (B,D) cells were transfected withCYP7A1-Luc (A-B) or PEPCK-Luc (C-D) with increasing amounts of wild type (WT) or mutant (MT) PKA expression plasmids (100 and 200 ng).Luciferase activity was normalized to �-galactosidase activity. All experiments were done in duplicates, and data represent the mean � SD of 3individual experiments. CYP7A1, cholesterol 7�-hydroxylase; PKA, protein kinase A; PEPCK, phosphoenolpyruvate carboxykinase.

HEPATOLOGY, Vol. 43, No. 1, 2006 SONG AND CHIANG 121

Identification of a Region Conferring cAMP Inhi-bition of the Human CYP7A1 Gene. We next at-tempted to identify potential sequences conferringglucagon/cAMP inhibition of the CYP7A1 gene. A seriesof 5�-deletion constructs of human CYP7A1-Luc reporterwere used in transfection assay. Deletion of humanCYP7A1 promoter sequence from �1178 to �150 didnot affect 8-Br-cAMP inhibition of reporter activity (Fig.4A). However, further deletion to �135 abolished therepressive effect of 8-Br-cAMP, suggesting that the regionbetween �150 and �135 conferred the negative effect ofcAMP. Interestingly, this region was previously identifiedas a BARE-II, which contains a DR-1 sequence for bind-ing of HNF4�.24

To further confirm that this DR-1 site mediates thecAMP response, mutations were introduced into theDR-1 sequence of the ph-298CYP7A1-Luc reporter(WT-298, Fig. 4B). This mutant reporter (MT-298) didnot respond to cAMP. These results suggested that gluca-gon/cAMP-activated pathway repressed CYP7A1 genetranscription through the HNF4� binding site.

PKA Represses HNF4� Transcriptional Activity.Because HNF4� is involved in mediating PKA suppres-sion and PKA-dependent phosphorylation of HNF4� isknown to reduce its DNA-binding and transactivationactivity,25 we studied the effect of PKA on HNF4� tran-scriptional activity of a heterologous luciferase reportercontaining 4 copies of a HNF4� binding sequence (frommouse transthyretin promoter) fused to a minimal TKpromoter (pHNF4-tk-Luc). As shown in Fig. 5A, co-transfection of HNF4� drastically stimulated this heter-ologous reporter activity. Addition of increasing amountsof the wild-type PKA plasmid drastically reduced reporteractivity. In contrast, a mutant PKA, which has no kinase

activity, did not have any effect on the reporter activity.Co-transfection of HNF4� stimulated human CYP7A1-Luc reporter activity by 2-fold. Wild-type but not mutantPKA strongly inhibited the reporter activity stimulated byHNF4�. Interestingly, increasing the amount of wildtype PKA repressed HNF4�-stimulated CYP7A1 pro-moter reporter activity below the basal activity, suggestingthat PKA abolished the activity stimulated by the endog-enous HNF4� present in HepG2 cells (Fig. 5B). Thesedata suggested that PKA might phosphorylate HNF4�DNA binding domain or activation function domains(AF1 and AF2), and resulted in inhibition of HNF4�transactivation of the CYP7A1 gene.

Glucagon and cAMP Increase Phosphorylayionof HNF4�. To determine whether glucagon and 8-Br-cAMP affect phosphorylation of HNF4�, primary hu-man hepatocytes were treated with glucagon or 8-Br-

Fig. 4. Identification of the sequence responsive to PKA in the human CYP7A1 gene. (A) A series of deletion constructs of human CYP7A1-luciferasereporters were transfected into HepG2 cells and treated with 8-Br-cAMP (1 mmol/L) or a corresponding amount of vehicle under the serum-starvedcondition. Basal reporter activities of each reporter constructs are 4.2 � 106, 5.0 � 106, 3.8 � 106, 1.2 � 106, 8.2 � 105, 2.1 � 105, 1.6 �105, and 1.0 � 105 relative luciferase unit/�-gal, for �1187, �371, �298, �185, �150, �135, �80, and pGL3, respectively. (B) Mutationalanalysis of human CYP7A1 promoter. Wild-type (WT) and mutant (MT) DR-1 sequences of human CYP7A1 are indicated. Wild-type and mutantreporters were transfected into HepG2 cells and treated with 8-Br-cAMP (1 mmol/L) or a vehicle in the serum-starved condition. Luciferase activitywas normalized to �-galactosidase activity. Basal reporter activities of each construct are 3.8 � 106, 1.0 � 104, 2.1 � 105, and 1.0 � 105 relativeluciferase unit/�-gal, for WT-298, MT-298, WT-135, and pGL3 basic, respectively. All experiments were done in duplicate, and data represent themean � SD of 3 individual experiments. PKA, protein kinase A; CYPK7A1, CYP7A1, cholesterol 7�-hydroxylase; 8-Br-cAMP, 8-Bromo-cyclic AMP.

Fig. 5. Effect of PKA on HNF4� transcriptional activity. HepG2 cellswere transfected with indicated pHNF4-tk-Luc (A) or hCYP7A1-Luc (B)and HNF4� expression plasmid (200 ng) with increasing amounts ofwild-type (WT) or mutant (MT) PKA expression plasmids (50, 100, and200 ng). Luciferase activity was normalized to �-galactosidase activity.All experiments were done in duplicates, and data represent the mean �SD of 3 individual experiments. PKA, protein kinase A; HNF4�, hepato-cyte nuclear factor 4�.

122 SONG AND CHIANG HEPATOLOGY, January 2006

cAMP for 6 hours, and cellular extracts wereimmunoprecipitated with an anti-HNF4� antibody.HNF4� and phosphorylated-HNF4� in immunopre-cipitants were analyzed by Western blot with an anti-HNF4� antibody and anti-phosphorprotein antibody,respectively. As shown in Fig. 6, treatment of primaryhepatocytes with glucagon or 8-Br-cAMP resulted in amarked increase in the amount of phosphorylatedHNF4� in the immunoprecipitants detected by an anti-phosphor antibody, but did not affect the amount ofHNF4� in the precipitants. This result suggests that glu-cagon and 8-Br-cAMP strongly stimulated HNF4� phos-phorylation, but did not affect the expression of HNF4�protein in primary human hepatocytes.

Glucagon and cAMP Dissociate HNF4� From Hu-man CYP7A1 Chromatin. We then studied the effect ofglucagon and 8-Br-cAMP on HNF4� binding toCYP7A1 chromatin using primary human hepatocytesfor ChIP assay. As shown in Fig. 7A, glucagon treatmentrapidly decreased the amounts of CYP7A1 chromatinprecipitated by an antibody against HNF4� in a time-dependent manner (Fig. 7A). No signal was detectedwhen non-immune immunoglobulin G was used for im-munoprecipitation. Moreover, 8-Br-cAMP also rapidlydecreased the amount of human CYP7A1 chromatin pre-cipitated by an anti-HNF4� antibody (Fig. 7B). Takentogether, these results indicated that glucagon and cAMPdecreased HNF4� binding to human CYP7A1 chroma-tin in human primary hepatocytes.

DiscussionIn the current study, we demonstrated that glucagon

and cAMP repressed CYP7A1 gene expression levels in atime- and dose-dependent manner in human primaryhepatocytes, and the PKA signaling pathway mediatedthe inhibitory effect of glucagon and cAMP. We provided

the strong evidences that the glucagon/cAMP/PKA sig-naling increased the phosphorylation state of HNF4�and reduced its DNA binding and transactivation of theCYP7A1 gene.

PKA has been reported to phosphorylate the A-box ofthe HNF4� DNA-binding domain and decrease itsDNA-binding activity.25 AMP kinase phosphorylatedHNF4� on serine 304 and reduced HNF4� transactiva-tion activity by inhibiting its homodimerization and sta-bility.26 Phosphorylation of tyrosine residues on HNF4�affected its nuclear localization and transactivation.27

This study further supports our hypothesis that HNF4�plays a critical role in regulating CYP7A1 gene transcrip-tion and that reducing HNF4� binding and trans-acti-vating activity by physiological regulators such as bileacids, cytokines, and glucagon results in inhibition of theCYP7A1 gene.28 Diverse intracellular signaling pathwayshave been shown to mediate bile acid feedback inhibitionof the CYP7A1 gene. Bile acids are able to activate theprotein kinase C and cytokine signaling pathways, whichactivate the MAPK/c-jun N-terminal kinase pathway andinhibit the CYP7A1 gene.7 This study adds the glucagon/cAMP/PKA signaling pathway to a network of signalingpathways that inhibits the CYP7A1 gene. These signalingmechanisms are independent of the FXR/SHP pathwayin bile acid feedback regulation of bile acid synthesis.7

These signaling pathways converge to regulate HNF4�,the most important regulator of the CYP7A1 gene.

Our results that glucagon and cAMP inhibit humanCYP7A1 gene expression are consistent with the earlier

Fig. 7. Glucagon and 8-Br-cAMP treatments decrease in vivo bindingof HNF4� to the CYP7A1 chromatin. Human primary hepatocytes(H1211, HH1215, HH1220) were treated with glucagon (100 nmol/L)(A) or 8-Br-cAMP (1 mmol/L) (B) for the times indicated and subjectedto formaldehyde cross-linking. Chromatin fragments were prepared bysonication and immunoprecipitated with anti-HNF4� antibody, and pro-moter sequence containing HNF4� binding DR-1 element was analyzedby PCR using primer sets specific for the CYP7A1 promoter. Cell lysatesolution (5%) in ChIP dilution buffer was kept aside as “Input.” Datarepresent 1 of 3 separate experiments. 8-Br-cAMP, 8-Bromo-cyclic AMP;HNF4�, hepatocyte nuclear factor 4�; PCR, polymerase chain reaction.

Fig. 6. Glucagon and 8-Br-cAMP treatments increase phosphorylationof HNF4� in human primary hepatocytes. Human primary hepatocytes(H1211, HH1215) were treated with glucagon (100 nmol/L) or 8-Br-cAMP (1 mmol/L), and cellular extracts were immunoprecipitated (IP)with rabbit anti-HNF4� antibody. Western blot analysis was performedusing anti-phosphorprotein antibody. Western blot analysis for HNF4�using goat anti-HNF4� antibody was performed to confirm the equalamounts of HNF4� precipitant. 8-Br-cAMP, 8-Bromo-cyclic AMP; HNF4�,hepatocyte nuclear factor 4�.

HEPATOLOGY, Vol. 43, No. 1, 2006 SONG AND CHIANG 123

reports that glucagon or cAMP decreased CYP7A1 ex-pression in rat primary hepatocytes29 and fasting repressesCYP7A1 expression in rats and piglets.16-19 Reduction ofbile acid synthesis by fasting could result in increasinghepatic cholesterol levels, which is consistent with thefasting-induced serum total cholesterol levels in healthynon-obese humans.30 However, these results are in con-tradiction to two recent reports that fasting and cAMPinduce both CYP7A1 and PEPCK expression14,15 and thehypothesis that bile acid synthesis and gluconeogenesisare coordinately regulated by bile acids in response to thefasting-to-fed cycle.14 These two studies were conductedin mice, and the species differences between mice andhumans in regulation of CYP7A1 and bile acid synthesisare well documented.7 A most recent study reported thatCDCA and an FXR agonist GW4064 induce PEPCKmRNA expression in mouse livers, rat liver cell lines, andrat and human primary hepatocytes.31 This is also in com-plete opposite to bile acid inhibition of PEPCK reportedby De Fabiani et al.14 The previous results of glucagon/cAMP effect on CYP7A1 mRNA expression in rat pri-mary hepatocytes are consistent with this study in humanprimary hepatocytes, suggesting that primary hepatocytesare suitable for studying CYP7A1 regulation.

According to this study, we propose that glucagon re-leased from the pancreas during fasting may inhibitCYP7A1 gene transcription and reduce bile acid synthesisand bile acid pool size in human liver. This should resultin an inactivation of the FXR/SHP pathway during fast-ing. Glucagon markedly induces PGC-1�, which is a po-tent co-activator of CCAAT response element bindingprotein and glucocorticoid receptor that strongly upregu-late PEPCK gene transcription during fasting to maintainpostprandial glucose levels and prevent hypoglycemia.32

Because bile acid synthesis requires adenosine triphos-phate and bile acids are not needed during fasting, there isno physiological benefit for upregulating bile acid synthe-sis during fasting as suggested by De Fabiani et al.14 Onfeeding, the glucagon level is reduced and CYP7A1 ex-pression is de-repressed to produce bile acids for fat ab-sorption in the intestine. Bile acids are quantitatively re-absorbed in the ileum and transported to the liver toactivate the FXR/SHP pathway and other signaling path-ways that inhibit bile acid synthesis. The FXR/SHP path-way also may inhibit PEPCK and prevent hyperglycemiaand diabetes during the postprandial period. The entero-hepatic recirculation of bile acids is an important physio-logical process that tightly controls the bile acid pool sizeto protect hepatocytes from bile acid toxicity during thefasting and fed states. Similarly, the fasting and feedingcycle regulates glucose metabolism and homeostasis. Byvirtue of their physiological roles in nutrient absorption

and distribution, bile acids may play a role in controlglucose and lipid metabolism to maintain metabolic ho-meostasis.

In summary, our current study showed a mechanism oftranscriptional regulation of bile acid synthesis by gluca-gon in humans. Inhibition of CYP7A1 gene expressionand bile acid synthesis during fasting by glucagon/cAMP/PKA signaling may ensure the maximal activation of glu-coneogenesis to maintain glucose homeostasis and energymetabolism. This study establishes an important role forglucagon regulation of bile acid synthesis.

References1. Chiang JY. Bile acid regulation of gene expression: roles of nuclear hor-

mone receptors. Endocr Rev 2002;23:443-463.2. Russell DW, Setchell KD. Bile acid biosynthesis. Biochemistry 1992;31:

4737-4749.3. Chiang JY. Regulation of bile acid synthesis. Front Biosci 1998;3:d176-

d193.4. Li T, Chiang JY. Mechanism of rifampicin and pregnane X receptor inhi-

bition of human cholesterol 7 �-hydroxylase gene transcription. Am JPhysiol Gastrointest Liver Physiol 2005;288:G74-G84.

5. Stroup D, Chiang JY. HNF4 and COUP-TFII interact to modulate tran-scription of the cholesterol 7�-hydroxylase gene (CYP7A1). J Lipid Res2000;41:1-11.

6. Hayhurst GP, Lee YH, Lambert G, Ward JM, Gonzalez FJ. Hepatocytenuclear factor 4alpha (nuclear receptor 2A1) is essential for maintenance ofhepatic gene expression and lipid homeostasis. Mol Cell Biol 2001;21:1393-1403.

7. Chiang JY. Regulation of bile acid synthesis: pathways, nuclear receptors,and mechanisms. J Hepatol 2004;40:539-551.

8. Xu G, Salen G, Shefer S, Tint GS, Nguyen LB, Parker TT, et al. Regula-tion of classic and alternative bile acid synthesis in hypercholesterolemicrabbits: effects of cholesterol feeding and bile acid depletion. J Lipid Res1998;39:1608-1615.

9. Horton JD, Cuthbert JA, Spady DK. Regulation of hepatic 7�-hydroxy-lase expression and response to dietary cholesterol in the rat and hamster.J Biol Chem 1995;270:5381-5387.

10. Rudel L, Deckelman C, Wilson M, Scobey M, Anderson R. Dietary cho-lesterol and downregulation of cholesterol 7�-hydroxylase and cholesterolabsorption in African green monkeys. J Clin Invest 1994;93:2463-2472.

11. Agellon LB, Drover VA, Cheema SK, Gbaguidi GF, Walsh A. Dietarycholesterol fails to stimulate the human cholesterol 7�-hydroxylase gene(CYP7A1) in transgenic mice. J Biol Chem 2002;277:20131-20134.

12. Chiang JY, Kimmel R, Stroup D. Regulation of cholesterol 7�-hydroxy-lase gene (CYP7A1) transcription by the liver orphan receptor (LXR�).Gene 2001;262:257-265.

13. Goodwin B, Watson MA, Kim H, Miao J, Kemper JK, Kliewer SA. Dif-ferential regulation of rat and human CYP7A1 by the nuclear oxysterolreceptor liver X receptor-alpha. Mol Endocrinol 2003;17:386-394.

14. De Fabiani E, Mitro N, Gilardi F, Caruso D, Galli G, Crestani M. Coor-dinated control of cholesterol catabolism to bile acids and of gluconeogen-esis via a novel mechanism of transcription regulation linked to the fasted-to-fed cycle. J Biol Chem 2003;278:39124-39132.

15. Shin DJ, Campos JA, Gil G, Osborne TF. PGC-1� activates CYP7A1 andbile acid biosynthesis. J Biol Chem 2003;278:50047-50052.

16. Lewis DS, Oren S, Wang X, Moyer ML, Beitz DC, Knight TJ, et al.Developmental changes in cholesterol 7�- and 27-hydroxylases in thepiglet. J Anim Sci 2000;78:943-951.

17. Bjorkhem I, Akerlund JE. Studies on the link between HMG-CoA reduc-tase and cholesterol 7�-hydroxylase in rat liver. J Lipid Res 1988;29:136-143.

124 SONG AND CHIANG HEPATOLOGY, January 2006

18. Noshiro M, Nishimoto M, Okuda K. Rat liver cholesterol 7�-hydroxylase:pretranslational regulation for circadian rhythm. J Biol Chem 1990;265:10036-10041.

19. Nguyen LB, Shefer S, Salen G, Ness G, Tanaka RD, Packin V, et al.Purification of cholesterol 7�-hydroxylase from human and rat liver andproduction of inhibiting polyclonal antibodies. J Biol Chem 1990;265:4541-4546.

20. Maurer RA. Both isoforms of the cAMP-dependent protein kinase cata-lytic subunit can activate transcription of the prolactin gene. J Biol Chem1989;264:6870-6873.

21. Park YY, Kim HJ, Kim JY, Kim MY, Song KH, Cheol Park K, et al.Differential role of the loop region between helices H6 and H7 within theorphan nuclear receptors small heterodimer partner and DAX-1. Mol En-docrinol 2004;18:1082-1095.

22. Wang DP, Stroup D, Marrapodi M, Crestani M, Galli G, Chiang JY.Transcriptional regulation of the human cholesterol 7�-hydroxylase gene(CYP7A) in HepG2 cells. J Lipid Res 1996;37:1831-1841.

23. Chakravarty K, Wu SY, Chiang CM, Samols D, Hanson RW. SREBP-1c andSp1 interact to regulate transcription of the gene for phosphoenolpyruvatecarboxykinase (GTP) in the liver. J Biol Chem 2004;279:15385-15395.

24. Crestani M, Sadeghpour A, Stroup D, Galli G, Chiang JY. Transcriptionalactivation of the cholesterol 7�-hydroxylase gene (CYP7A) by nuclearhormone receptors. J Lipid Res 1998;39:2192-2200.

25. Viollet B, Kahn A, Raymondjean M. Protein kinase A-dependent phos-phorylation modulates DNA-binding activity of hepatocyte nuclear factor4. Mol Cell Biol 1997;17:4208-4219.

26. Hong YH, Varanasi US, Yang W, Leff T. AMP-activated protein kinaseregulates HNF4� transcriptional activity by inhibiting dimer forma-tion and decreasing protein stability. J Biol Chem 2003;278:27495-27501.

27. Ktistaki E, Ktistakis NT, Papadogeorgaki E, Talianidis I. Recruitmentof hepatocyte nuclear factor 4 into specific intranuclear compartmentsdepends on tyrosine phosphorylation that affects its DNA-binding andtransactivation potential. Proc Natl Acad Sci U S A 1995;92:9876-9880.

28. Chiang JY. Bile acid regulation of hepatic physiology: III. Bile acids andnuclear receptors. Am J Physiol Gastrointest Liver Physiol 2003;284:G349-G356.

29. Hylemon PB, Gurley EC, Stravitz RT, Litz JS, Pandak WM, Chiang JY, etal. Hormonal regulation of cholesterol 7�-hydroxylase mRNA levels andtranscriptional activity in primary rat hepatocyte cultures. J Biol Chem1992;267:16866-16871.

30. Savendahl L, Underwood LE. Fasting increases serum total cholesterol,LDL cholesterol and apolipoprotein B in healthy, nonobese humans. JNutr 1999;129:2005-2008.

31. Stayrook KR, Bramlett KS, Savkur RS, Ficorilli J, Cook T, Christe ME, etal. Regulation of carbohydrate metabolism by the farnesoid X receptor.Endocrinology 2005;146:984-991.

32. Yoon JC, Puigserver P, Chen G, Donovan J, Wu Z, Rhee J, et al. Controlof hepatic gluconeogenesis through the transcriptional coactivator PGC-1.Nature 2001;413:131-138.

HEPATOLOGY, Vol. 43, No. 1, 2006 SONG AND CHIANG 125