Hematopoietically expressed homeobox is a target gene of farnesoid X receptor in chenodeoxycholic acid–induced liver hypertrophy


  • Potential conflict of interest: Nothing to report.


Farnesoid X receptor (FXR/Fxr) is a bile acid–regulated nuclear receptor that promotes hepatic bile acid metabolism, detoxification, and liver regeneration. However, the adaptive pathways under conditions of bile acid stress are not fully elucidated. We found that wild-type but not Fxr knockout mice on diets enriched with chenodeoxycholic acid (CDCA) increase their liver/body weight ratios by 50% due to hepatocellular hypertrophy. Microarray analysis identified Hex (Hematopoietically expressed homeobox), a central transcription factor in vertebrate embryogenesis and liver development, as a novel CDCA- and Fxr-regulated gene. HEX/Hex was also regulated by FXR/Fxr and CDCA in primary mouse hepatocytes and human HepG2 cells. Comparative genomic analysis identified a conserved inverted repeat-1–like DNA sequence within a 300 base pair enhancer element of intron-1 in the human and mouse HEX/Hex gene. A combination of chromatin immunoprecipitation, electromobility shift assay, and transcriptional reporter assays demonstrated that FXR/Fxr binds to this element and mediates HEX/Hex transcriptional activation. Conclusion: HEX/Hex is a novel bile acid–induced FXR/Fxr target gene during adaptation of hepatocytes to chronic bile acid exposure. (HEPATOLOGY 2009.)

Liver enlargement (hepatomegaly) is an adaptive response to prevent toxicity to hepatocytes by endobiotics1 or xenobiotics.2 Chronic exposure to endogenous bile acids, as in patients with and rodent models of cholestatic liver diseases,3, 4 also activates this response. The induction of cytochrome p450 (CYP/Cyp) enzymes and membrane transporters, as well as the growth, differentiation, and cell volume of hepatocytes is regulated by members of the nuclear receptor superfamily such as farnesoid X receptor (FXR/Fxr).1, 3 FXR/Fxr binds DNA as a heterodimer with retinoid X receptor and is activated by chenodeoxycholic acid (CDCA).5 FXR/Fxr is a key player in the control of bile acid de novo synthesis, excretion, and enterohepatic reabsorption. In rodents, a high-fat diet, bile acids, and Fxr were linked to liver injury, inflammation, and fibrosis.6, 7 In contrast, studies in Fxr-deficient mice suggest a beneficial role of Fxr in prevention of cholestasis,8 in liver regeneration upon partial hepatectomy,9 in antimicrobial defense within the gastrointestinal tract,10 and against tumor formation in the liver.11 Thus, FXR/Fxr is expected to be a protective and trophic factor beyond its classical metabolic functions. However, the target genes that underlie these functions are still under investigation. We show here that Hex (hematopoietically expressed homeobox), a key transcription factor in vertebrate liver development,12 is a novel Fxr-regulated gene that is induced during the liver response to CDCA in C57BL/6N mice.


CDCA, chenodeoxycholic acid; CYP7A1/Cyp7a1, cholesterol-7alpha-hydroxylase; FXR/Fxr, farnesoid X receptor; HEX/Hex, hematopoietically expressed homeobox; IR-1, inverted repeat-1; SHP/Shp, small heterodimer partner.

Materials and Methods


Female wild-type (C57BL/6N; Charles River, Wilmington, MA) and Fxr knockout (Fxr-KO) (B6; 129XFVB-Nr1h4tm1Gonz/J; mixed C57BL/6N background; Jackson Laboratory, Bar Harbor, ME) mice (4 weeks, 14-19 g, n = 5 per group) were fed a chow diet (Altromin, Lage, Germany) with or without 1% (wt/wt) CDCA (Chemos GmbH, Regenstauf, Germany). Animal studies were conducted in agreement with ethical guidelines of the Technical University of Munich and approved by the government of Bavaria, Munich, Germany.


Chemicals were from Merck (Darmstadt, Germany) or Sigma (Taufkirchen, Germany). Antibodies were FXR/Fxr (sc-13063; Santa Cruz Biotechnology, Santa Cruz, CA); FXR/Fxr (A9033A, R&D Systems, Wiesbaden-Nordenstadt, Germany); HEX/Hex (H-4913; Sigma), lamin A/C (sc-20681; Santa Cruz Biotechnology), β-actin (AC-74; Sigma), acetyl-histone H3 (06-599; Upstate, Millipore GmbH, Schwalbach, Germany), cyclin D1 (SP4), Ki-67 (SP6) (both from DCS GmbH, Hamburg, Germany), and bromodeoxyuridine (BrdU; Serotec, Raleigh, NC).

Cell Culture.

Human embryonic kidney HEK293 and hepatoma HepG2 cells (both from the American Type Culture Collection, Manassas, VA) and human Huh7 cells (JCRB Cell Bank, Japan) were maintained as recommended by the suppliers. Primary mouse hepatocytes were isolated as described.13

Oligonucleotides and Plasmids.

Reporter plasmids were based on pTK-luciferase as described.14 A consensus IR-1 FXR-responsive element (FXR-RE) GGG ACA t TGA TCC15 from the human bile salt export pump (BSEP) −63/−50 base pair (bp) promoter sequence (NT_005403) and a 300-bp fragment of the human HEX intron 1 (AL590080 [version .25] 26.389 to 26.688 bp) (Supporting Table 1) were cloned into pTK-luciferase. Deletion of the IR-1 GGG TCA g GCT CTT in HEX intron1–pTK-luciferase was performed by Quickchange Mutagenesis (Stratagene, Amsterdam, Netherlands). The 930-bp human HEX proximal promoter (AL590080 [version .25] 24.700 to 25.630 bp) was cloned into pGL3-basic-luciferase (Promega GmbH, Mannheim, Germany). The 1428-bp full-length human FXR cDNA (476 amino acids [aa], alpha 1 splice variant containing the four-aa insertion MYTG, NM_005123)16, 17 was inserted into pTarget (Promega). Transient transfection and luciferase assays were performed as described.14 The FXR-RE electrophoretic mobility shift assay (EMSA) oligonucleotide (Supporting Table 2) was generated based on the consensus IR-1 DNA sequence AGG TCA t TGA CCT.18


Immunohistochemistry (IHC) and hematoxylin&eosin staining was performed as described.19

Reverse Transcription PCR and Quantitative PCR.

Polymerase chain reactions (PCR) were performed as described.19

DNA Microarray.

Total RNA (1 μg) from wild-type mice on control or CDCA diet (7 days) was subjected to One-Cycle cRNA labeling (Affymetrix, Wycombe, UK) and hybridized to a Mouse Genome 430A 2.0 Array (Affymetrix). GO-groups were identified by Gene Set Enrichment Analysis (GSEA) (http://www.broad.mit.edu/gsea) (Supporting Table 3).

Comparative Genomics.

Homologous genomic loci (http://genome.ucsc.edu) were identified with BLAST (Basic Local Alignment Search Tool) and aligned using MLAGAN. Binding sites were predicted using matrices from TransFac 8.4 and processed following Rahmann et al.20 (Table 1).

Table 1. Sequence Alignment of Predicted* FXR/Fxr-Binding Elements in Intron 1 of Genomic HEX/Hex Loci
  • *

    Based on MLAGAN/TransFac 8.4 matrices (M00964 and M00767).

M00767GGGTBAATRACCY (FXR inverted repeat-1)
SHP HumanGAGTTAATGACCT (FXR inverted repeat-1)

Chromatin Immunoprecipitation.

Chromatin Immunoprecipitation (ChIP; Upstate, Millipore GmbH) was performed as described.14

Electrophoretic Mobility Shift Assay.

Cell extraction and western blotting were performed as described.14 Digoxigenin Gel Shift kit (Roche Diagnostics GmbH, Mannheim, Germany) and LightShift® Chemiluminescent EMSA kit (Pierce Biotechnology, Rockford, IL) were used as recommended by the manufacturers.


Results are means ± standard error (SE) from five individual animals per group or at least three independent cell experiments. P values were calculated using one-way analysis of variance. Data were analyzed by SPSS 13.0 and Graphpad Prism (Version 4.0).


Fxr Is Required for Induction of Liver Hypertrophy by CDCA.

To stimulate chronic bile acid exposure in mouse liver, C57BL/6N wild-type and Fxr-KO mice were fed a chow diet enriched with 1% (wt/wt) CDCA for 8 weeks. CDCA increased the liver/body weight ratio by approximately 50% (*P < 0.05) in wild-type but not in Fxr-KO mice (Fig. 1A). CDCA-treated wild-type mice livers exhibited increased hepatocyte cell size, hepatocellular anisokaryosis, double-layered liver cell plates, and “tile/plaster” formation indicative of hypertrophy (Fig. 1B). Neither infiltration of inflammatory cells nor any other damage-related signs of bile acid–induced toxicity such as fibrosis or steatosis were observed. The absolute number of hepatocytes per visual field (at 200× magnification), an indication of enlarged cell volume, was decreased by approximately 20% (*P < 0.05) in livers from CDCA-fed wild-type mice compared to littermates on control diet (Fig. 1C). The livers of Fxr-KO mice, while exhibiting steatosis (lipid droplets, enlarged nuclei) which naturally develops in this strain,21 did not show liver enlargement by CDCA. Immunohistochemistry revealed no significant differences for the proliferation markers Ki-67 (Fig. 1D), cyclin D1 (Fig. 1E), and for BrdU incorporation (Fig. 1F). Quantitative PCR (qPCR) confirmed lack of Fxr mRNA in Fxr-KO mice (Fig. 1G). The Fxr-target gene Cyp7a1 was significantly repressed in CDCA-fed wild-type mice whereas derepressed in Fxr-KO mice (Fig. 1H). Shp remained unchanged in CDCA-fed wild-type mice, a phenomenon which may be caused by secondary compensation mechanisms during long-term bile acid exposure (Fig. 1I). The qPCR for proliferative (c-myc, c-fos, cyclin D1) and inflammatory (tumor necrosis factorα, interleukin-6, intercellular cell adhesion marker-1) markers again revealed no significant changes (Supporting Fig. 1A). These data indicate that long-term oral administration of CDCA to C57BL/6N mice results in nonpathological liver hypertrophy that requires Fxr.

Figure 1.

Fxr mediates CDCA-induced liver hypertrophy in C57BL/6N mice. (A) Liver/body weight ratio ± SE after 8 weeks of chow enriched with 1% CDCA in wild-type (WT) and Fxr-KO mice (n = 5 per group). *P < 0.05 versus chow-fed WT control. (B) Hematoxylin & eosin staining of liver. Upper left panel is WT on control diet; Upper right panel is WT on 1% CDCA diet (with increased cell volume and anisokaryosis); Lower left panel is Fxr-KO on control diet (with accumulation of lipid droplets); Lower right panel is Fxr-KO on 1% CDCA diet. (C) Cell numbers per field ± SE at 200× magnification (five fields per mouse) were calculated as hypertrophy score from all four groups of mice (n = 5 per group). *P < 0.05 versus chow-fed WT control. (D-F) Immunohistochemistry of liver sections against proliferation markers (D) Ki-67, (E) cyclin D1, and (F) BrdU. Positive nuclei are shown as percent ± SE (n = 5 mice per group). (G-I) qPCR detection of Fxr-target genes. Results from (G) exon 11 (deleted in KO) of Fxr, (H) Cyp7a1, and (I) Shp are normalized against S12 and are shown as fold increase ± SE (n = 5 mice per group) compared to WT mice on control diet. *P < 0.05 versus chow-fed WT control.

Hex Is a Candidate Fxr Target Gene and Up-Regulated During Liver Hypertrophy.

To identify genes up-regulated during the early phase of CDCA-induced liver hypertrophy, we hybridized microarrays to RNA from livers of wild-type mice fed for 7 days with CDCA or control diet (Supporting Fig. 2). We excluded the Fxr-KO mice from this direct comparison because they are from a mixed genetic background21 and were not responsive to CDCA. Despite the shorter duration of exposure, the wild-type mice showed an approximately 17% increase (*P < 0.05) of liver/body weight ratio compared to mice fed control diets (Fig. 2A) and also exhibited hepatocellular hypertrophy, albeit less pronounced than after the 8-week CDCA diet (data not shown). Pathway analysis of the microarray data (Supporting Fig. 2) revealed no changes in gene clusters related to ribosome/protein synthesis, peroxisome proliferation, cell cycle, or apoptosis rates, all of which are potential mechanisms leading to increased liver weight (Supporting Fig. 3). The majority of gene changes mapped to GO-groups defining metabolic (adaptive detoxification), transcriptional, signal transduction, and developmental pathways (Supporting Table 3). We thus focused on four genes (Hex, Tef, Rev-erb-beta, and Igfbp1) that are known as key regulators therein. Hex (4.5 ± 1.9-fold in wild-type [WT] CDCA versus WT-control), Tef, and Rev-erb-beta expression was up-regulated by CDCA (Fig. 2B,C,D), whereas Igfbp1 was reduced (not shown). Hex induction by CDCA was lost in Fxr-KO mice while the increase in Tef and Rev-erb-beta expression remained (Fig. 2B,C,D), suggesting that their response to CDCA occurs independently of Fxr. Hex messenger RNA (mRNA) remained increased (2.2-fold ± 0.3-fold in WT-CDCA versus WT-control, data not shown) after 8 weeks, suggesting that Hex induction is an early as well as a continuing event during the hepatic response. The elevated basal Hex expression levels in Fxr-KO mice on control diet (Fig. 2B) suggest that unliganded Fxr may repress Hex transcription. Indeed, our later studies (see below) indicate that Fxr also occupies the Hex locus in the absence of ligand. As before, cytochrome p450 7a1 (Cyp7a1) was repressed by CDCA in WT mice and was derepressed in Fxr-KO mice (Fig. 2E). Shp (Fig. 2F) and the hepatic bile acid transporter proteins Bsep, Ntcp, and Ostalpha (Supporting Fig. 1B) were not significantly regulated by CDCA; however, their expression was higher in wild-type than in Fxr-KO mice, suggesting Fxr plays a role in regulation of their basal expression. On the protein level, Hex was also strongly increased by CDCA in WT but not in Fxr-KO mice both after 7 days (Fig. 3A) and 8 weeks (Fig. 3B). Analysis by qPCR on total RNA from primary hepatocytes that were stimulated ex vivo with CDCA (100 μM) or the synthetic agonist GW4064 (1 μM) (Fig. 3C,D) confirmed the up-regulation of Hex (two-fold to three-fold) and Shp (four-fold to eight-fold) in WT but not Fxr-KO mice. We also detected FXR protein (Fig. 4A) and mRNA (Fig. 4B) in human hepatoma HepG2 cells but not in Huh7 cells, whereas HEX mRNA (Fig. 4B) was expressed in both cell lines. This finding contrasts to reports detecting FXR in Huh7 cells22, 23 but may be due to different sources or cultivation conditions. We exploited the lack of FXR in Huh7 cells as a human FXR null hepatocyte system. Indeed, whereas qPCR revealed that HEX, SHP (Fig. 4C), and BSEP (not shown) were up-regulated by CDCA in HepG2 cells and CYP7A1 expression was repressed (not shown), neither of these genes were altered in the FXR-negative Huh7 cells (Fig. 4D). These data suggested that the CDCA-mediated FXR/Fxr-dependent regulation of the HEX/Hex gene shares a conserved mechanism between human and mouse.

Figure 2.

Identification of Hex as a CDCA-regulated Fxr-dependent target gene. (A) Liver/body weight ratio ± SE after 7 days of chow diet enriched with 1% CDCA in WT and Fxr-KO mice (n = 5 per group). *P < 0.05 versus chow-fed WT control. (B-F) qPCR verification of candidate and classical Fxr-target genes in mouse livers from the DNA array chip. Results are normalized against S12 and shown as fold increase ± SE (n = 5 mice per group) compared to WT mice on control diet. *P < 0.05 versus WT on control diet.

Figure 3.

Fxr-dependent up-regulation of Hex in mouse livers and primary mouse hepatocytes. (A,B) Western blotting of Hex protein in whole tissue lysates of mice livers (from three different animals per group) from (A) 7 days and (B) 8 weeks of CDCA diet. (C,D) qPCR analysis reveals kinetics of Hex and Shp mRNA induction in primary hepatocytes of WT and KO mice by (C) CDCA (100 μM) and (D) GW4064 (1 μM). Values are normalized to S12 and presented as fold ± SE increase of mRNA compared to vehicle-treated cells. *P < 0.05 versus vehicle.

Figure 4.

HEX is also expressed and induced by CDCA in HepG2 human hepatoma cells. (A) Western blotting of FXR and β-actin in whole-cell lysates of HepG2 and Huh7 cells. (B) RT-PCR for FXR, HEX, and β-actin in HepG2 and Huh7 cells. (C,D) qPCR. Dose response of HEX mRNA induction in (C) HepG2 cells but not in (D) Huh7 cells after 8 hours of CDCA. Values are normalized to S12 and presented as -fold ± S.E. (n = 3 independent experiments) increase of HEX mRNA versus vehicle-treated cells. **P < 0.01 versus vehicle.

The HEX Gene Contains a Functional Highly Conserved FXR Binding Site in Intron 1.

We then analyzed the human HEX gene for potential FXR binding sites. The proximal promoter region is GC-rich and contains multiple binding elements for SP1, hepatocyte nuclear factor 3β, SMADs and other transcription factors.12, 24 The intronic regions harbor conserved binding sites for hematopoietic MYB/ETS/GATA factors (intron 1)25, 26 and enhancer elements for embryonic liver development (intron 3).27 Using comparative bioinformatics, we identified a hexameric IR-1–like DNA element (GGGTBA A TRACCY) within the reverse strand of intron 1 (GGGTCA G GCTCTT) (Accession number AL590080 version.25, Position 26.384 to 26.396) that clusters within the region containing the MYB/ETS/GATA-binding sites.25, 26 The best match to a consensus IR-1 was evident within the first half-site (Table 1). This candidate IR-1 FXR/Fxr element is 100% identical in human, dog, cow, and opossum and has only one nucleotide change in mouse and rat. To analyze FXR binding to these candidate regions, we performed ChIP on the human genomic HEX locus in HepG2 cells. HepG2 cells were treated for 24 hours with 100 μM CDCA and subjected to ChIP with a polyclonal acetyl-histone H3 antiserum followed by genomic qPCR. Primers (Supporting Table 1) were designed to interrogate the distal (−1000/−500) and proximal (−500/+1) promoter regions, the transcriptional start site (+1/exon1) and the two conserved regions in intron 1 and intron 3 (Fig. 5A). For control, we interrogated the IR-1 (GAGTTA A TGACCT) in the human SHP1.28 Specific amplification products were visualized (Fig. 5B, left panel) together with quantification of normalized cycle threshold values (Fig. 5B, right panel). Compared to vehicle-treated cells (lane 1), CDCA (lane 2) led to the acetylation of the chromatin along several stretches of the HEX locus including at the IR-1 site in intron 1. No pull-down of genomic DNA was observed with uncoupled agarose beads (lane 3) or control immunoglobulin G (lane 4). ChIP with a polyclonal antiserum to FXR (Fig. 5C) detected increased binding of FXR to the SHP IR-1 element as well as to the HEX intron 1 and 3 regions in presence of CDCA (lane 2) compared to vehicle controls (lane 1). No substantial FXR binding was detected for the proximal promoter regions. ChIP was then repeated in lysates from primary mouse hepatocytes treated for 24 hours with 100 μM CDCA. Fxr binding to the well-conserved intronic region 1 was detected in hepatocytes from WT animals, whereas no increase in binding was evident in Fxr-KO animals (n = 3 per group) (Fig. 5D). These data indicate that CDCA increases the accessibility of the chromatin promoting binding of FXR/Fxr to intronic regions of the HEX/Hex gene.

Figure 5.

The genomic HEX/Hex locus is targeted by CDCA and FXR/Fxr in HepG2 cells and mouse hepatocytes. (A) Primer sites for ChIP in the human/mouse genomic HEX/Hex locus. (B-D) ChIP. HepG2 cells (B, representative experiment; C, n = 3 independent passages) and primary hepatocytes from wild-type and Fxr-KO mice (n = 3 per genotype) (D) were treated for 24 hours with vehicle (lane 1) or 100 μM CDCA (lane 2). Immunoprecipitation was performed with acetyl-H3-histone, rabbit polyclonal FXR/fxr antiserum, control immunoglobulin G, or no antibody (empty bead control). Left panels: Genomic qPCR reactions (endpoint: 40 cycles) visualized by gel electrophoresis. Right panels: CT-values from qPCR of immunoprecipitated DNA were normalized to the CT-values of input DNA and are expressed as fold ± SE increase of pull-down by CDCA compared to (B) empty bead or (C,D) vehicle controls. *P < 0.05 versus vehicle.

FXR Binds Directly to the IR-1–Like Motif in the HEX Intron 1.

To explore the direct interaction of FXR with the HEX IR1-like element, we performed EMSAs. HEK293 cells, which are devoid of endogenous FXR,29, 30 were transfected with either WT-FXR expression plasmid or empty vector (EV). Ectopic FXR was exclusively localized to the nucleus (Fig. 6A). EMSAs were calibrated using a bona fide IR-1 consensus (FXR-RE) oligonucleotide18 and compared to an oligonucleotide corresponding to the central region of the IR-1–like element in the human HEX Intron 1 (HEX-In1) (Supporting Table 2). Increased protein binding to both FXR-RE and HEX-In1 was visible in nuclear extracts from FXR-transfected (lane 2) compared to untransfected (lane 1) cells (Fig. 6B). The complex (lane 2) was competed by an excess of unlabeled oligonucleotides (lane 3), rabbit polyclonal antiserum (lane 4) or monoclonal antibody (right panel: lane 5) against FXR. We then examined the effect of CDCA on FXR binding affinity to HEX-In1 (Fig. 6C). A constitutive, though weak, binding of a protein complex was evident in untransfected, vehicle-treated (lane 1) or CDCA-stimulated (lane 2) cells. Binding intensity was increased in FXR-transfected cells (lane 3) and further increased by CDCA-stimulation of ectopic FXR protein (lane 4). Addition of unlabeled HEX-In1 (lanes 5,6) competed the lower band of the doublet, which was therefore termed the specific one. This was further proven using a mutant HEX-In1 oligonucleotide (Fig. 6D) as competitor. The lower band of the doublet was competed by the wild-type (lane 5,6) but not by the mutant (lane 7,8) HEX-In1 compared to FXR-transfected CDCA-stimulated cells without competitor (lane 3,4) (Fig. 6D). The unlabeled consensus FXR-RE also acted as an efficient competitor (Fig. 6E), indicating that the protein complex that binds to HEX Intron-1 contains FXR. In HepG2 cells expressing endogenous FXR, constitutive binding of a protein(s) to HEX-In1 was detectable, which was not significantly increased by ligand treatment (lanes 1,2) (Fig. 6F), but which was specific as evident by a concentration-dependent reduction of signal by unlabeled HEX-In1 oligonucleotide (lanes 3-8) or FXR antibody (not shown). These data show that FXR binds to the IR-1–like element in the HEX Intron 1.

Figure 6.

The IR-1–like motif in the human HEX Intron 1 binds FXR and promotes CDCA-mediated transactivation of a heterologous promoter. (A) Fractionation and western blotting of untransfected (HEK) and FXR-transfected (HEK-FXR) HEK293 cells; (B-F) EMSAs with digoxigenin-labeled (DIG) or biotin-labeled (BIO) oligonucleotides (FXR-RE = IR-1 consensus; HEX-In1 = Intron 1 of the HEX gene) and unlabeled competitors (HEX-In1-WT = wild type; -MUT = mutant; FXR-RE = IR-1 consensus), (+a) rabbit polyclonal or (+b) mouse monoclonal FXR-antibodies directed against the DBD of FXR in nuclear extracts from (B-E) transfected HEK293 and (F) HepG2 cells; (G) HEK293 cells were transiently cotransfected with pTK-luciferase reporter plasmids and FXR expression plasmid (FXR-WT) or empty vector (EV), respectively, and were stimulated for 24 hours with CDCA (0, 30, and 100 μM). (H) HepG2 cells transfected as in (G). Luciferase values are normalized to protein content and indicated as fold increase ± standard error (n = 3 independent experiments) compared to vehicle-treated controls. *P < 0.05 versus vehicle.

CDCA Increases FXR-Mediated Transactivation of a Heterologous Promoter Through HEX Intron 1.

To determine FXR-mediated transactivation on the HEX Intron-1, we constructed a series of reporter constructs. The 300 bp fragment of the human HEX Intron-1 sequence was cloned into the pTK-luciferase vector and transiently cotransfected with either FXR-WT or empty vector (EV) into HEK293 cells (Fig. 6G). Luciferase activity was determined upon a 24-hour CDCA treatment (30 and 100 μM). CDCA activated HEX Intron-1-pTK-driven reporter gene expression in FXR-WT-transfected (two-fold to six-fold versus vehicle, *P < 0.05) but not in EV-transfected HEK293 cells. As positive control, luciferase activity in cells transfected with FXR-RE-pTK-luciferase was also dose-dependently induced by CDCA (5-fold to 12-fold versus vehicle, *P < 0.05) in an FXR-dependent manner. No significant activity was observed using the enhancer-less pTK-luciferase (Fig. 6G) or a pGL3-luciferase reporter plasmid harboring the human HEX proximal promoter (not shown). Deletion of both IR-1 half-sites by mutagenesis of the HEX Intron-1-pTK-luciferase plasmid (ΔHEX-In1) abrogated the response to CDCA in FXR-transfected cells (Fig. 6G). These findings corroborated that the IR-1 element is responsible for the CDCA-mediated and FXR-mediated induction of HEX Intron-1–driven reporter gene expression. In HepG2 cells (Fig. 6H) or primary mouse hepatocytes (data not shown), the CDCA inducibility of the reporter constructs was less pronounced than in HEK293 cells, a phenomenon that is consistent with the EMSA results and may be explained by the different molecular context of cofactors and transcriptional coregulators in the cell types used.


In this study, we identified the homeobox factor HEX/Hex as a novel bile acid–regulated FXR/Fxr-target gene. C57BL/6N mice upon short-term and long-term feeding of a chow diet enriched with 1% (wt/wt) CDCA developed nonpathological Fxr-dependent liver enlargement (hepatomegaly) characterized by hepatocellular hypertrophy. This finding contrasts with other studies using diets enriched with 0.1% to 1% (wt/wt) cholic acid (CA) which evoked liver weight loss and bile acid–related hepatotoxicity.31-33 However, others reported either minor or no toxicity of orally applied CA8, 11, 21, 34-36 or CDCA.37-40 This divergence may be caused by variations in duration of treatment doses, and pharmacological properties (e.g., hydrophobicity) of the bile acid types which differ in their metabolic conversion rates to secondary bile acids in vivo5 and in their binding affinities and cofactor recruitment to FXR/Fxr.5 Similar to our results, a previous study9 showed that a diet enriched with 0.2% (wt/wt) CA increased liver size in C57BL/6 wild-type mice in as short as 3 to 5 days and that this required Fxr. This finding already proposed specific mechanisms leading to this trophic response. Because CA is only a weak Fxr agonist5 and not in clinical use,3 we performed our study with CDCA to stimulate a systemic Fxr response.

In this setting, we identified HEX/Hex as a novel CDCA-regulated FXR/Fxr target gene in both mouse hepatocytes and human hepatoma HepG2 cells, indicative of a conserved mechanisms in these species. We show that CDCA promotes chromatin acetylation and FXR/Fxr binding to an IR-1–like element embedded in a highly conserved enhancer region in the gene's first intron. Interestingly, this 300 bp region also contains several myb and GATA binding sites directly downstream to the IR-1–like element that act as active enhancers in cells of hematopoietic but not hepatic origin.25, 26 A truncated HEX In-1 oligonucleotide lacking the first myb site directly adjacent to the IR-l (Supporting Table 2) allowed FXR binding as for the full-length HEX In-1 (not shown), suggesting that this myb site is not essential. It is intriguing to propose that the IR-1–like element appears functionally independent of the first myb site25, 26 and may thus serve as an organ-specific switch in regulation of cells of hematopoietic versus hepatic origin in response to developmental or metabolic signals.

Hex is critical for the differentiation of the columnar hepatic endoderm to the multilayered liver bud during vertebrate development.12, 41 CDCA overloading of the liver may thus reactivate this early morphogenic process and enable expansion of adult hepatocytes. This hypothesis is supported by Hex regulation in the adult under stress conditions, such as upon partial hepatectomy in rats42 or viral infection of rhesus monkeys.43 In our microarray, a cluster of sonic hedgehog (Shh) target genes44 is down-regulated by CDCA as exemplified by Tsc22 domain family 3 (glucocorticoid-induced leucine zipper) (Supporting Fig. 2, Supporting Fig. 4A, Supporting Table 3). Shh promotes proliferation, fibrosis, and carcinogenesis upon hepatic injury45, 46 and acts antagonistic to Hex in liver embryogenesis.41 Shh activates cyclin D1, whereas Hex inhibits nuclear export and translation of cyclin D1 mRNA.47 CDCA, through up-regulation of Hex and down-regulation of Shh signaling, may shift the response in favor of hepatocyte hypertrophy (Supporting Fig. 4B). Hex may also be more directly linked to Fxr in bile acid homeostasis by the convergence of Hex48 and Fxr/Shp49, 50 in regulation of Ntcp. Future experiments using conditional knockout or transgenic mice treated with FXR/Fxr ligands have to further investigate the functional link between Hex, Shh, and liver hypertrophy.


We are grateful to Minhu Chen for collaborations. We thank Eric Niesor and Martin Benson for discussion and advice, Jörg Mages for microarray studies, and Hans-Peter Märki for supply of ligands.