Hematopoietically Expressed Homeobox Is a Target of Farnesoid X Receptor in Chenodeoxycholic Acid–Induced Liver Hypertrophy

Xiangbin Xing,1,2* Elke Burgermeister,1* Fabian Geisler,1 Henrik Einwachter,¨ 1 Lian Fan,1,2 Michaela Hiber,1 Sandra Rauser,3 Axel Walch,3 Christoph Rocken,¨ 4 Martin Ebeling,5 Matthew B. Wright,5 Roland M. Schmid,1 and Matthias P.A. Ebert1

Farnesoid X receptor (FXR/Fxr) is a bile acid–regulated nuclear receptor that promotes hepatic bile acid metabolism, detoxification, and liver regeneration. However, the adap- tive 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 hypertro- phy. 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 enhancer element of intron-1 in the human and mouse HEX/Hex gene. A combi- nation of chromatin immunoprecipitation, electromobility shift assay, and transcrip- tional 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;49:979-988.)

iver enlargement (hepatomegaly) is an adaptive re- response. The induction of cytochrome p450 (CYP/Cyp) sponse to prevent toxicity to hepatocytes by en- enzymes and membrane transporters, as well as the Ldobiotics1 or xenobiotics.2 Chronic exposure to growth, differentiation, and cell volume of hepatocytes is endogenous bile acids, as in patients with and rodent regulated by members of the nuclear receptor superfamily models of cholestatic liver diseases,3,4 also activates this 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 Abbreviations: CDCA, chenodeoxycholic acid; CYP7A1/Cyp7a1, cholesterol- is a key player in the control of bile acid de novo synthesis, 7alpha-hydroxylase; FXR/Fxr, farnesoid X receptor; HEX/Hex, hematopoietically expressed homeobox; IR-1, inverted repeat-1; SHP/Shp, small heterodimer partner. excretion, and enterohepatic reabsorption. In rodents, a From the 1Department of Medicine II, Klinikum rechts der Isar, Technical high-fat diet, bile acids, and Fxr were linked to liver in- University Munich, Germany; 2 Department of Gastroenterology, The First Affili- jury, inflammation, and fibrosis.6,7 In contrast, studies in 3 ated Hospital of Sun Yat-sen University, Guangzhou, PR China; Institute of Fxr-deficient mice suggest a beneficial role of Fxr in pre- Pathology, Helmholtz Forschungszentrum Mu¨nchen, Neuherberg and 4 Institute of 8 Pathology, Charite, Berlin, Germany; 5 F. Hoffmann-La Roche, Basel, Switzer- vention of cholestasis, in liver regeneration upon partial land. hepatectomy,9 in antimicrobial defense within the gastro- Received May 14, 2008; accepted October 20, 2008. intestinal tract,10 and against tumor formation in the liv- *These authors contributed equally to this study. 11 Address reprints requests to: Prof. Matthias P.A. Ebert, M.D., Department of Med- er. Thus, FXR/Fxr is expected to be a protective and icine II, Klinikum rechts der Isar, Technical University of Mu¨nchen, Ismaningerstraße trophic factor beyond its classical metabolic functions. ϩ 22, D-81675 Mu¨nchen, Germany. E-mail: [email protected]; fax: 49-89- However, the target that underlie these functions 4140-2259. Potential conflict of interest: Nothing to report. are still under investigation. We show here that Hex (he- Copyright © 2009 by the American Association for the Study of Liver Diseases. matopoietically expressed homeobox), a key transcription Published online in Wiley InterScience (www.interscience.wiley.com). factor in vertebrate liver development,12 is a novel Fxr- DOI 10.1002/hep.22712 Additional Supporting Information may be found in the online version of this regulated gene that is induced during the liver response to article. CDCA in C57BL/6N mice.

979 980 XING ET AL. HEPATOLOGY, March 2009

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

Element Sequence

M00964 RRGGTYA N TRNM (PXR, CAR, LXR, FXR) M00767 GGGTBA A TRACCY (FXR inverted repeat-1) SHP Human GAGTTA A TGACCT (FXR inverted repeat-1) HEX Human CTGTATGAACGGAAAGGGTCA G GCTCTTT-CACTGCACAAGCCTGTTGAA Hex Dog CTGTATGAACGGAAAGGGTCA G GCTCTTT-CACTGCACAAGCTTGTTGAA Hex Cow CTGTATGAACGGAAAGGGTCA G GCTCTTT-CACTGCACAAGCTTGTTGAA Hex Mouse CTGTATGAACGGAAAGGGTCA G GCTCCTT-CACTGCACAAGCTTGTTGAA Hex Rat CTGTATGAACGGAAAGGGTCA G GCTCCTT-CACTGCACAAGCTTGTTGAA Hex opossum CTGTATGAACGGAAAGGGTCA G GCTCTTTTCGCTGCACAAACTTGTTGAA

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

Materials and Methods lands). The 930-bp human HEX proximal promoter (AL590080 [version .25] 24.700 to 25.630 bp) was Animals. Female wild-type (C57BL/6N; Charles cloned into pGL3-basic-luciferase (Promega GmbH, River, Wilmington, MA) and Fxr knockout (Fxr-KO) Mannheim, Germany). The 1428-bp full-length human (B6; 129XFVB-Nr1h4tm1Gonz/J; mixed C57BL/6N back- FXR cDNA (476 amino acids [aa], alpha 1 splice variant ground; Jackson Laboratory, Bar Harbor, ME) mice (4 containing the four-aa insertion MYTG, NM_005123)16,17 weeks, 14-19 g, n ϭ 5 per group) were fed a chow diet was inserted into pTarget (Promega). Transient transfection (Altromin, Lage, Germany) with or without 1% (wt/wt) and luciferase assays were performed as described.14 The CDCA (Chemos GmbH, Regenstauf, Germany). Animal FXR-RE electrophoretic mobility shift assay (EMSA) oligo- studies were conducted in agreement with ethical guide- nucleotide (Supporting Table 2) was generated based on the lines of the Technical University of Munich and approved consensus IR-1 DNA sequence AGG TCA t TGA CCT.18 by the government of Bavaria, Munich, Germany. Immunohistochemistry. Immunohistochemistry (IHC) Reagents. Chemicals were from Merck (Darmstadt, and hematoxylin&eosin staining was performed as de- Germany) or Sigma (Taufkirchen, Germany). Antibodies scribed.19 were FXR/Fxr (sc-13063; Santa Cruz Biotechnology, Reverse Transcription PCR and Quantitative PCR. Santa Cruz, CA); FXR/Fxr (A9033A, R&D Systems, Polymerase chain reactions (PCR) were performed as de- Wiesbaden-Nordenstadt, Germany); HEX/Hex (H-4913; scribed.19 Sigma), lamin A/C (sc-20681; Santa Cruz Biotechnology), DNA Microarray. Total RNA (1 ␮g) from wild-type ␤-actin (AC-74; Sigma), acetyl-histone H3 (06-599; Up- mice on control or CDCA diet (7 days) was subjected to state, Millipore GmbH, Schwalbach, Germany), cyclin D1 One-Cycle cRNA labeling (Affymetrix, Wycombe, UK) (SP4), Ki-67 (SP6) (both from DCS GmbH, Hamburg, and hybridized to a Mouse Genome 430A 2.0 Array (Af- Germany), and bromodeoxyuridine (BrdU; Serotec, Ra- fymetrix). GO-groups were identified by Gene Set En- leigh, NC). richment Analysis (GSEA) (http://www.broad.mit.edu/ Cell Culture. Human embryonic kidney HEK293 gsea) (Supporting Table 3). and hepatoma HepG2 cells (both from the American Comparative Genomics. Homologous genomic loci Type Culture Collection, Manassas, VA) and human (http://genome.ucsc.edu) were identified with BLAST Huh7 cells (JCRB Cell Bank, Japan) were maintained as (Basic Local Alignment Search Tool) and aligned using recommended by the suppliers. Primary mouse hepato- MLAGAN. Binding sites were predicted using matrices cytes were isolated as described.13 from TransFac 8.4 and processed following Rahmann et Oligonucleotides and Plasmids. Reporter plasmids al.20 (Table 1). were based on pTK-luciferase as described.14 A consensus Chromatin Immunoprecipitation. Chromatin Im- IR-1 FXR-responsive element (FXR-RE) GGG ACA t munoprecipitation (ChIP; Upstate, Millipore GmbH) TGA TCC15 from the human bile salt export pump was performed as described.14 (BSEP) Ϫ63/Ϫ50 base pair (bp) promoter sequence Electrophoretic Mobility Shift Assay. Cell extrac- (NT_005403) and a 300-bp fragment of the human tion and western blotting were performed as described.14 HEX intron 1 (AL590080 [version .25] 26.389 to 26.688 Digoxigenin Gel Shift kit (Roche Diagnostics GmbH, bp) (Supporting Table 1) were cloned into pTK-lucif- Mannheim, Germany) and LightShift௡ Chemilumines- erase. Deletion of the IR-1 GGG TCA g GCT CTT in cent EMSA kit (Pierce Biotechnology, Rockford, IL) were HEX intron1–pTK-luciferase was performed by Quick- used as recommended by the manufacturers. change Mutagenesis (Stratagene, Amsterdam, Nether- Statistics. Results are means Ϯ standard error (SE) HEPATOLOGY, Vol. 49, No. 3, 2009 XING ET AL. 981 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).

Results

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 Ͻ 0.05) in wild-type but not in Fxr-KO mice (Fig. 1A). CDCA-treated wild-type mice livers exhibited increased hepatocyte cell size, hepatocel- lular 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 200ϫ magnifi- cation), an indication of enlarged cell volume, was de- creased by approximately 20% (*P Ͻ 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 en- largement by CDCA. Immunohistochemistry revealed no significant differences for the proliferation markers Ki-67 (Fig. 1D), cyclin D1 (Fig. 1E), and for BrdU in- corporation (Fig. 1F). Quantitative PCR (qPCR) con- firmed lack of Fxr mRNA in Fxr-KO mice (Fig. 1G). The Fxr-target gene Cyp7a1 was significantly repressed in Fig. 1. Fxr mediates CDCA-induced liver hypertrophy in C57BL/6N CDCA-fed wild-type mice whereas derepressed in mice. (A) Liver/body weight ratio Ϯ SE after 8 weeks of chow enriched Fxr-KO mice (Fig. 1H). Shp remained unchanged in with 1% CDCA in wild-type (WT) and Fxr-KO mice (n ϭ 5 per group). CDCA-fed wild-type mice, a phenomenon which may be *P Ͻ 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 caused by secondary compensation mechanisms during 1% CDCA diet (with increased cell volume and anisokaryosis); Lower left long-term bile acid exposure (Fig. 1I). The qPCR for panel is Fxr-KO on control diet (with accumulation of lipid droplets); proliferative (c-myc, c-fos, cyclin D1) and inflammatory Lower right panel is Fxr-KO on 1% CDCA diet. (C) Cell numbers per ␣ field Ϯ SE at 200ϫ magnification (five fields per mouse) were calcu- (tumor necrosis factor , interleukin-6, intercellular cell lated as hypertrophy score from all four groups of mice (n ϭ 5 per adhesion marker-1) markers again revealed no significant group). *P Ͻ 0.05 versus chow-fed WT control. (D-F) Immunohistochem- changes (Supporting Fig. 1A). These data indicate that istry of liver sections against proliferation markers (D) Ki-67, (E) cyclin D1, and (F) BrdU. Positive nuclei are shown as percent Ϯ SE (n ϭ 5 long-term oral administration of CDCA to C57BL/6N mice per group). (G-I) qPCR detection of Fxr-target genes. Results from mice results in nonpathological liver hypertrophy that re- (G) exon 11 (deleted in KO) of Fxr, (H) Cyp7a1, and (I) Shp are quires Fxr. normalized against S12 and are shown as fold increase Ϯ SE (n ϭ 5 Hex Is a Candidate Fxr Target Gene and Up- mice per group) compared to WT mice on control diet. *P Ͻ 0.05 versus chow-fed WT control. Regulated During Liver Hypertrophy. To identify genes up-regulated during the early phase of CDCA-in- duced liver hypertrophy, we hybridized microarrays to they are from a mixed genetic background21 and were not RNA from livers of wild-type mice fed for 7 days with responsive to CDCA. Despite the shorter duration of ex- CDCA or control diet (Supporting Fig. 2). We excluded posure, the wild-type mice showed an approximately the Fxr-KO mice from this direct comparison because 17% increase (*P Ͻ 0.05) of liver/body weight ratio com- 982 XING ET AL. HEPATOLOGY, March 2009

tion. Indeed, our later studies (see below) indicate that Fxr also occupies the Hex 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 trans- porter Bsep, Ntcp, and Ostalpha (Supporting Fig. 1B) were not significantly regulated by CDCA; how- ever, 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 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 hepato- cytes that were stimulated ex vivo with CDCA (100 ␮M) or the synthetic agonist GW4064 (1 ␮M) (Fig. 3C,D) confirmed the up-regulation of Hex (two-fold to three-

Fig. 2. Identification of Hex as a CDCA-regulated Fxr-dependent target gene. (A) Liver/body weight ratio Ϯ SE after 7 days of chow diet enriched with 1% CDCA in WT and Fxr-KO mice (n ϭ 5 per group). *P Ͻ 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 Ϯ SE (n ϭ 5 mice per group) compared to WT mice on control diet. *P Ͻ 0.05 versus WT on control diet. pared to mice fed control diets (Fig. 2A) and also exhib- ited hepatocellular hypertrophy, albeit less pronounced than after the 8-week CDCA diet (data not shown). Path- way 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, sig- nal transduction, and developmental pathways (Support- ing Table 3). We thus focused on four genes (Hex, Tef, Rev-erb-beta, and Igfbp1) that are known as key regula- tors therein. Hex (4.5 Ϯ 1.9-fold in wild-type [WT] CDCA versus WT-control), Tef, and Rev-erb-beta ex- pression 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 indepen- dently of Fxr. Hex messenger RNA (mRNA) remained Ϯ Fig. 3. Fxr-dependent up-regulation of Hex in mouse livers and pri- increased (2.2-fold 0.3-fold in WT-CDCA versus mary mouse hepatocytes. (A,B) Western blotting of Hex protein in whole WT-control, data not shown) after 8 weeks, suggesting tissue lysates of mice livers (from three different animals per group) from that Hex induction is an early as well as a continuing event (A) 7 days and (B) 8 weeks of CDCA diet. (C,D) qPCR analysis reveals during the hepatic response. The elevated basal Hex ex- kinetics of Hex and Shp mRNA induction in primary hepatocytes of WT and KO mice by (C) CDCA (100 ␮M) and (D) GW4064 (1 ␮M). Values pression levels in Fxr-KO mice on control diet (Fig. 2B) are normalized to S12 and presented as fold Ϯ SE increase of mRNA suggest that unliganded Fxr may repress Hex transcrip- compared to vehicle-treated cells. *P Ͻ 0.05 versus vehicle. HEPATOLOGY, Vol. 49, No. 3, 2009 XING ET AL. 983

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 con- served mechanism between human and mouse. The HEX Gene Contains a Functional Highly Con- served FXR Binding Site in Intron 1. We then ana- lyzed the human HEX gene for potential FXR binding sites. The proximal promoter region is GC-rich and con- tains multiple binding elements for SP1, hepatocyte nu- clear factor 3␤, SMADs and other transcription factors.12,24 The intronic regions harbor conserved bind- ing sites for hematopoietic MYB/ETS/GATA factors (in- tron 1)25,26 and enhancer elements for embryonic liver development (intron 3).27 Using comparative bioinfor- matics, we identified a hexameric IR-1–like DNA ele- ment (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 nucle- otide 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 ␮M CDCA and sub- jected to ChIP with a polyclonal acetyl-histone H3 anti- serum followed by genomic qPCR. Primers (Supporting Table 1) were designed to interrogate the distal (Ϫ1000/ Ϫ500) and proximal (Ϫ500/ϩ1) promoter regions, the transcriptional start site (ϩ1/exon1) and the two con- Fig. 4. HEX is also expressed and induced by CDCA in HepG2 human served regions in intron 1 and intron 3 (Fig. 5A). For hepatoma cells. (A) Western blotting of FXR and ␤-actin in whole-cell control, we interrogated the IR-1 (GAGTTA A lysates of HepG2 and Huh7 cells. (B) RT-PCR for FXR, HEX, and ␤-actin 28 in HepG2 and Huh7 cells. (C,D) qPCR. Dose response of HEX mRNA TGACCT) in the human SHP1. Specific amplification induction in (C) HepG2 cells but not in (D) Huh7 cells after 8 hours of products were visualized (Fig. 5B, left panel) together CDCA. Values are normalized to S12 and presented as -fold Ϯ S.E. (n ϭ with quantification of normalized cycle threshold values 3 independent experiments) increase of HEX mRNA versus vehicle- (Fig. 5B, right panel). Compared to vehicle-treated cells treated cells. **P Ͻ 0.01 versus vehicle. (lane 1), CDCA (lane 2) led to the acetylation of the chromatin along several stretches of the HEX locus in- fold) and Shp (four-fold to eight-fold) in WT but not cluding at the IR-1 site in intron 1. No pull-down of Fxr-KO mice. We also detected FXR protein (Fig. 4A) genomic DNA was observed with uncoupled agarose and mRNA (Fig. 4B) in human hepatoma HepG2 cells beads (lane 3) or control immunoglobulin G (lane 4). but not in Huh7 cells, whereas HEX mRNA (Fig. 4B) was ChIP with a polyclonal antiserum to FXR (Fig. 5C) de- expressed in both cell lines. This finding contrasts to re- tected increased binding of FXR to the SHP IR-1 element ports detecting FXR in Huh7 cells22,23 but may be due to as well as to the HEX intron 1 and 3 regions in presence of different sources or cultivation conditions. We exploited CDCA (lane 2) compared to vehicle controls (lane 1). No the lack of FXR in Huh7 cells as a human FXR null substantial FXR binding was detected for the proximal hepatocyte system. Indeed, whereas qPCR revealed that promoter regions. ChIP was then repeated in lysates from HEX, SHP (Fig. 4C), and BSEP (not shown) were up- primary mouse hepatocytes treated for 24 hours with 100 regulated by CDCA in HepG2 cells and CYP7A1 expres- ␮M CDCA. Fxr binding to the well-conserved intronic sion was repressed (not shown), neither of these genes region 1 was detected in hepatocytes from WT animals, 984 XING ET AL. HEPATOLOGY, March 2009

Fig. 5. The genomic HEX/Hex locus is targeted by CDCA and FXR/Fxr in HepG2 cells and mouse hepa- tocytes. (A) Primer sites for ChIP in the human/mouse genomic HEX/Hex locus. (B-D) ChIP. HepG2 cells (B, representative experiment; C, n ϭ 3 independent pas- sages) and primary hepatocytes from wild-type and Fxr-KO mice (n ϭ 3 per genotype) (D) were treated for 24 hours with vehicle (lane 1) or 100 ␮M CDCA (lane 2). Immunoprecipitation was performed with acetyl- H3-histone, rabbit polyclonal FXR/fxr antiserum, con- trol immunoglobulin G, or no antibody (empty bead control). Left panels: Genomic qPCR reactions (end- point: 40 cycles) visualized by gel electrophoresis. Right panels: CT-values from qPCR of immunoprecipi- tated DNA were normalized to the CT-values of input DNA and are expressed as fold Ϯ SE increase of pull-down by CDCA compared to (B) empty bead or (C,D) vehicle controls. *P Ͻ 0.05 versus vehicle. whereas no increase in binding was evident in Fxr-KO man HEX Intron 1 (HEX-In1) (Supporting Table 2). animals (n ϭ 3 per group) (Fig. 5D). These data indicate Increased protein binding to both FXR-RE and HEX-In1 that CDCA increases the accessibility of the chromatin was visible in nuclear extracts from FXR-transfected (lane promoting binding of FXR/Fxr to intronic regions of the 2) compared to untransfected (lane 1) cells (Fig. 6B). The HEX/Hex gene. complex (lane 2) was competed by an excess of unlabeled FXR Binds Directly to the IR-1–Like Motif in the oligonucleotides (lane 3), rabbit polyclonal antiserum HEX Intron 1. To explore the direct interaction of FXR (lane 4) or monoclonal antibody (right panel: lane 5) with the HEX IR1-like element, we performed EMSAs. against FXR. We then examined the effect of CDCA on HEK293 cells, which are devoid of endogenous FXR,29,30 FXR binding affinity to HEX-In1 (Fig. 6C). A constitu- were transfected with either WT-FXR expression plasmid tive, though weak, binding of a protein complex was evi- or empty vector (EV). Ectopic FXR was exclusively local- dent in untransfected, vehicle-treated (lane 1) or CDCA- ized to the nucleus (Fig. 6A). EMSAs were calibrated stimulated (lane 2) cells. Binding intensity was increased using a bona fide IR-1 consensus (FXR-RE) oligonucleo- in FXR-transfected cells (lane 3) and further increased by tide18 and compared to an oligonucleotide corresponding CDCA-stimulation of ectopic FXR protein (lane 4). Ad- to the central region of the IR-1–like element in the hu- dition of unlabeled HEX-In1 (lanes 5,6) competed the HEPATOLOGY, Vol. 49, No. 3, 2009 XING ET AL. 985

Fig. 6. The IR-1–like motif in the human HEX Intron 1 binds FXR and promotes CDCA-mediated transactivation of a heter- ologous 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 ϭ IR-1 consensus; HEX-In1 ϭ In- tron 1 of the HEX gene) and unlabeled competitors (HEX-In1-WT ϭ wild type; -MUT ϭ mutant; FXR-RE ϭ IR-1 consen- sus), (ϩa) rabbit polyclonal or (ϩb) 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 tran- siently cotransfected with pTK-luciferase re- porter plasmids and FXR expression plasmid (FXR-WT) or empty vector (EV), re- spectively, and were stimulated for 24 hours with CDCA (0, 30, and 100 ␮M). (H) HepG2 cells transfected as in (G). Lucif- erase values are normalized to protein con- tent and indicated as fold increase Ϯ standard error (n ϭ 3 independent experi- ments) compared to vehicle-treated con- trols. *P Ͻ 0.05 versus vehicle. lower band of the doublet, which was therefore termed to HEX-In1 was detectable, which was not significantly the specific one. This was further proven using a mutant increased by ligand treatment (lanes 1,2) (Fig. 6F), but HEX-In1 oligonucleotide (Fig. 6D) as competitor. The which was specific as evident by a concentration-depen- lower band of the doublet was competed by the wild-type dent reduction of signal by unlabeled HEX-In1 oligonu- (lane 5,6) but not by the mutant (lane 7,8) HEX-In1 cleotide (lanes 3-8) or FXR antibody (not shown). These compared to FXR-transfected CDCA-stimulated cells data show that FXR binds to the IR-1–like element in the without competitor (lane 3,4) (Fig. 6D). The unlabeled HEX Intron 1. consensus FXR-RE also acted as an efficient competitor CDCA Increases FXR-Mediated Transactivation of (Fig. 6E), indicating that the protein complex that binds a Heterologous Promoter Through HEX Intron 1. To to HEX Intron-1 contains FXR. In HepG2 cells express- determine FXR-mediated transactivation on the HEX In- ing endogenous FXR, constitutive binding of a protein(s) tron-1, we constructed a series of reporter constructs. The 986 XING ET AL. HEPATOLOGY, March 2009

300 bp fragment of the human HEX Intron-1 sequence clinical use,3 we performed our study with CDCA to was cloned into the pTK-luciferase vector and transiently stimulate a systemic Fxr response. cotransfected with either FXR-WT or empty vector (EV) In this setting, we identified HEX/Hex as a novel into HEK293 cells (Fig. 6G). Luciferase activity was de- CDCA-regulated FXR/Fxr target gene in both mouse termined upon a 24-hour CDCA treatment (30 and 100 hepatocytes and human hepatoma HepG2 cells, indica- ␮M). CDCA activated HEX Intron-1-pTK-driven re- tive of a conserved mechanisms in these species. We show porter gene expression in FXR-WT-transfected (two-fold that CDCA promotes chromatin acetylation and FXR/ to six-fold versus vehicle, *P Ͻ 0.05) but not in EV- Fxr binding to an IR-1–like element embedded in a transfected HEK293 cells. As positive control, luciferase highly conserved enhancer region in the gene’s first in- activity in cells transfected with FXR-RE-pTK-luciferase tron. Interestingly, this 300 bp region also contains sev- was also dose-dependently induced by CDCA (5-fold to eral myb and GATA binding sites directly downstream to Ͻ 12-fold versus vehicle, *P 0.05) in an FXR-dependent the IR-1–like element that act as active enhancers in cells manner. No significant activity was observed using the of hematopoietic but not hepatic origin.25,26 A truncated enhancer-less pTK-luciferase (Fig. 6G) or a pGL3-lucif- HEX In-1 oligonucleotide lacking the first myb site di- erase reporter plasmid harboring the human HEX proxi- rectly adjacent to the IR-l (Supporting Table 2) allowed mal promoter (not shown). Deletion of both IR-1 half- FXR binding as for the full-length HEX In-1 (not sites by mutagenesis of the HEX Intron-1-pTK-luciferase shown), suggesting that this myb site is not essential. It is ⌬ plasmid ( HEX-In1) abrogated the response to CDCA intriguing to propose that the IR-1–like element appears in FXR-transfected cells (Fig. 6G). These findings corrob- functionally independent of the first myb site25,26 and orated that the IR-1 element is responsible for the may thus serve as an organ-specific switch in regulation of CDCA-mediated and FXR-mediated induction of HEX cells of hematopoietic versus hepatic origin in response to Intron-1–driven reporter gene expression. In HepG2 developmental or metabolic signals. cells (Fig. 6H) or primary mouse hepatocytes (data not Hex is critical for the differentiation of the columnar shown), the CDCA inducibility of the reporter constructs hepatic endoderm to the multilayered liver bud during was less pronounced than in HEK293 cells, a phenome- vertebrate development.12,41 CDCA overloading of the non that is consistent with the EMSA results and may be liver may thus reactivate this early morphogenic process explained by the different molecular context of cofactors and enable expansion of adult hepatocytes. This hypoth- and transcriptional coregulators in the cell types used. esis is supported by Hex regulation in the adult under Discussion stress conditions, such as upon partial hepatectomy in rats42 or viral infection of rhesus monkeys.43 In our mi- In this study, we identified the homeobox factor HEX/ croarray, a cluster of sonic hedgehog (Shh) target genes44 Hex as a novel bile acid–regulated FXR/Fxr-target gene. is down-regulated by CDCA as exemplified by Tsc22 C57BL/6N mice upon short-term and long-term feeding domain family 3 (glucocorticoid-induced leucine zipper) of a chow diet enriched with 1% (wt/wt) CDCA devel- (Supporting Fig. 2, Supporting Fig. 4A, Supporting Ta- oped nonpathological Fxr-dependent liver enlargement ble 3). Shh promotes proliferation, fibrosis, and carcino- (hepatomegaly) characterized by hepatocellular hypertro- genesis upon hepatic injury45,46 and acts antagonistic to phy. This finding contrasts with other studies using diets Hex in liver embryogenesis.41 Shh activates cyclin D1, enriched with 0.1% to 1% (wt/wt) cholic acid (CA) whereas Hex inhibits nuclear export and translation of which evoked liver weight loss and bile acid–related hep- cyclin D1 mRNA.47 CDCA, through up-regulation of atotoxicity.31-33 However, others reported either minor or Hex and down-regulation of Shh signaling, may shift the no toxicity of orally applied CA8,11,21,34-36 or CDCA.37-40 response in favor of hepatocyte hypertrophy (Supporting This divergence may be caused by variations in duration Fig. 4B). Hex may also be more directly linked to Fxr in of treatment doses, and pharmacological properties (e.g., bile acid homeostasis by the convergence of Hex48 and hydrophobicity) of the bile acid types which differ in their Fxr/Shp49,50 in regulation of Ntcp. Future experiments metabolic conversion rates to secondary bile acids in vivo5 and in their binding affinities and cofactor recruitment to using conditional knockout or transgenic mice treated FXR/Fxr.5 Similar to our results, a previous study9 with FXR/Fxr ligands have to further investigate the func- showed that a diet enriched with 0.2% (wt/wt) CA in- tional link between Hex, Shh, and liver hypertrophy. creased liver size in C57BL/6 wild-type mice in as short as Acknowledgment: We are grateful to Minhu Chen 3 to 5 days and that this required Fxr. This finding already for collaborations. We thank Eric Niesor and Martin proposed specific mechanisms leading to this trophic re- Benson for discussion and advice, Jo¨rg Mages for microar- sponse. Because CA is only a weak Fxr agonist5 and not in ray studies, and Hans-Peter Ma¨rki for supply of ligands. HEPATOLOGY, Vol. 49, No. 3, 2009 XING ET AL. 987

References 21. Sinal CJ, Tohkin M, Miyata M, Ward JM, Lambert G, Gonzalez FJ. Targeted disruption of the nuclear receptor FXR/BAR impairs bile acid 1. Zollner G, Marschall HU, Wagner M, Trauner M. Role of nuclear recep- and lipid homeostasis. Cell 2000;102:731-744. tors in the adaptive response to bile acids and cholestasis: pathogenetic and 22. Landrier JF, Eloranta JJ, Vavricka SR, Kullak-Ublick GA. The nuclear therapeutic considerations. Mol Pharm 2006;3:231-251. receptor for bile acids, FXR, transactivates human organic solute transport- 2. Dai X, Souza AT, Dai H, Lewis DL, Lee CK, Spencer AG, et al. PPARal- er-alpha and -beta genes. Am J Physiol Gastrointest Liver Physiol 2006; pha siRNA-treated expression profiles uncover the causal sufficiency net- 290:G476-G485. work for compound-induced liver hypertrophy. PLoS Comput Biol 2007; 23. Anisfeld AM, Kast-Woelbern HR, Meyer ME, Jones SA, Zhang Y, Wil- 3:e30. liams KJ, et al. Syndecan-1 expression is regulated in an isoform-specific 3. Chen J, Raymond K. Nuclear receptors, bile-acid detoxification, and cho- manner by the farnesoid-X receptor. J Biol Chem 2003;278:20420-20428. lestasis. Lancet 2006;367:454-456. 24. Denson LA, McClure MH, Bogue CW, Karpen SJ, Jacobs HC. 4. Liu Y, Binz J, Numerick MJ, Dennis S, Luo G, Desai B, et al. Hepatopro- HNF3beta and GATA-4 transactivate the liver-enriched homeobox gene, tection by the farnesoid X receptor agonist GW4064 in rat models of intra- Hex. Gene 2000;246:311-320. and extrahepatic cholestasis. J Clin Invest 2003;112:1678-1687. 25. Donaldson IJ, Chapman M, Kinston S, Landry JR, Knezevic K, Piltz S, 5. Pellicciari R, Costantino G, Fiorucci S. Farnesoid X receptor: from struc- et al. Genome-wide identification of cis-regulatory sequences control- ture to potential clinical applications. J Med Chem 2005;48:5383-5403. ling blood and endothelial development. Hum Mol Genet 2005;14: 6. Qin P, Borges-Marcucci LA, Evans MJ, Harnish DC. Bile acid signaling 595-601. through FXR induces intracellular adhesion molecule-1 expression in 26. Sato A, Keng VW, Yamamoto T, Kasamatsu S, Ban T, Tanaka H, et al. mouse liver and human hepatocytes. Am J Physiol Gastrointest Liver Identification and characterization of the hematopoietic cell-specific en- Physiol 2005;289:G267-G273. 7. Vergnes L, Phan J, Strauss M, Tafuri S, Reue K. Cholesterol and cholate hancer-like element of the mouse hex gene. J Biochem (Tokyo) 2004;135: components of an atherogenic diet induce distinct stages of hepatic inflam- 259-268. matory gene expression. J Biol Chem 2003;278:42774-42784. 27. Rodriguez TA, Casey ES, Harland RM, Smith JC, Beddington RS. Dis- 8. Guo GL, Lambert G, Negishi M, Ward JM, Brewer HB Jr, Kliewer SA, et tinct enhancer elements control Hex expression during gastrulation and al. Complementary roles of farnesoid X receptor, pregnane X receptor, and early organogenesis. Dev Biol 2001;234:304-316. constitutive androstane receptor in protection against bile acid toxicity. 28. Goodwin B, Jones SA, Price RR, Watson M, McKee D, Moore L, et al. A J Biol Chem 2003;278:45062-45071. regulatory cascade of the nuclear receptors FXR, SHP-1, and LRH-1 re- 9. Huang W, Ma K, Zhang J, Qatanani M, Cuvillier J, Liu J, et al. Nuclear presses bile acid biosynthesis. Mol Cell 2000;6:517-526. receptor-dependent bile acid signaling is required for normal liver regen- 29. Wang S, Lai K, Moy FJ, Bhat A, Hartman HB, Evans MJ. The nuclear eration. Science 2006;312:233-236. hormone receptor farnesoid X receptor (FXR) is activated by androsterone. 10. Inagaki T, Moschetta A, Lee YK, Peng L, Zhao G, Downes M, et al. Endocrinology 2006;147:4025-4033. Regulation of antibacterial defense in the small intestine by the nuclear bile 30. Nakahara M, Furuya N, Takagaki K, Sugaya T, Hirota K, Fukamizu A, et acid receptor. Proc Natl Acad SciUSA2006;103:3920-3925. al. Ileal bile acid-binding protein, functionally associated with the farne- 11. Yang F, Huang X, Yi T, Yen Y, Moore DD, Huang W. Spontaneous soid X receptor or the ileal bile acid transporter, regulates bile acid activity development of liver tumors in the absence of the bile acid receptor farne- in the small intestine. J Biol Chem 2005;280:42283-42289. soid X receptor. Cancer Res 2007;67:863-867. 31. Teng S, Piquette-Miller M. Hepatoprotective role of PXR activation and 12. Soufi A, Jayaraman PS. PRH/Hex: an oligomeric transcription factor and MRP3 in cholic acid-induced cholestasis. Br J Pharmaol 2007;151:367- multifunctional regulator of cell fate. Biochem J 2008;412:399-413. 376. 13. Geisler F, Algul H, Paxian S, Schmid RM. Genetic inactivation of RelA/ 32. Wang R, Lam P, Liu L, Forrest D, Yousef IM, Mignault D, et al. Severe p65 sensitizes adult mouse hepatocytes to TNF-induced apoptosis in vivo cholestasis induced by cholic acid feeding in knockout mice of sister of and in vitro. Gastroenterology 2007;132:2489-2503. P-glycoprotein. HEPATOLOGY 2003;38:1489-1499. 14. Burgermeister E, Schnoebelen A, Flament A, Benz J, Stihle M, Gsell B,et 33. Wang L, Han Y, Kim CS, Lee YK, Moore DD. Resistance of SHP-null al. A novel partial agonist of peroxisome proliferator-activated receptor- mice to bile acid-induced liver damage. J Biol Chem 2003;278:44475- gamma (PPARgamma) recruits PPARgamma-coactivator-1alpha, pre- 44481. vents triglyceride accumulation, and potentiates insulin signaling in vitro. 34. Shih DM, Kast-Woelbern HR, Wong J, Xia YR, Edwards PA, Lusis AJ. A Mol Endocrinol 2006;20:809-830. role for FXR and human FGF-19 in the repression of paraoxonase-1 gene 15. Ananthanarayanan M, Balasubramanian N, Makishima M, Mangelsdorf expression by bile acids. J Lipid Res 2006;47:384-392. DJ, Suchy FJ. Human bile salt export pump promoter is transactivated by 35. Miyata M, Tozawa A, Otsuka H, Nakamura T, Nagata K, Gonzalez FJ, et the farnesoid X receptor/bile acid receptor. J Biol Chem 2001;276:28857- al. Role of farnesoid X receptor in the enhancement of canalicular bile acid 28865. output and excretion of unconjugated bile acids: a mechanism for protec- 16. Huber RM, Murphy K, Miao B, Link JR, Cunningham MR, Rupar MJ, et tion against cholic acid-induced liver toxicity. J Pharmacol Exp Ther 2005; al. Generation of multiple farnesoid-X-receptor isoforms through the use 312:759-766. of alternative promoters. Gene 2002;290:35-43. 36. Yamagata K, Daitoku H, Shimamoto Y, Matsuzaka H, Hirota K, Ishida J, 17. Zhang Y, Kast-Woelbern HR, Edwards PA. Natural structural variants of the nuclear receptor farnesoid X receptor affect transcriptional activation. et al. Bile acids regulate gluconeogenic gene expression via small het- J Biol Chem 2003;278:104-110. erodimer partner-mediated repression of hepatocyte nuclear factor 4 and 18. Laffitte BA, Kast HR, Nguyen CM, Zavacki AM, Moore DD, Edwards Foxo1. J Biol Chem 2004;279:23158-23165. PA. Identification of the DNA binding specificity and potential target 37. Miyata M, Matsuda Y, Tsuchiya H, Kitada H, Akase T, Shimada M, et al. genes for the farnesoid X-activated receptor. J Biol Chem 2000;275: Chenodeoxycholic acid-mediated activation of the farnesoid X receptor 10638-10647. negatively regulates hydroxysteroid sulfotransferase. Drug Metab Pharma- 19. Ebert MP, Fei G, Kahmann S, Mu¨ller O, Yu J, Sung JJ, et al. Increased cokinet 2006;21:315-323. beta-catenin mRNA levels and mutational alterations of the APC and 38. Matsukuma KE, Bennett MK, Huang J, Wang L, Gil G, Osborne TF. beta-catenin gene are present in intestinal-type gastric cancer. Carcinogen- Coordinated control of bile acids and lipogenesis through FXR-dependent esis 2002;23:87-91. regulation of fatty acid synthase. J Lipid Res 2006;47:2754-2761. 20. Rahmann S, Muller T, Vingron M. On the power of profiles for transcrip- 39. Cheng X, Buckley D, Klaassen CD. Regulation of hepatic bile acid trans- tion factor binding site detection. Stat Appl Genet Mol Biol 2003;2:Arti- porters Ntcp and Bsep expression. Biochem Pharmacol 2007;74:1665- cle7. 1676. 988 XING ET AL. HEPATOLOGY, March 2009

40. Wang DQ, Tazuma S, Cohen DE, Carey MC. Feeding natural hydro- 46. Omenetti A, Popov Y, Jung Y, Choi SS, Witek RP, Yang L, et al. The philic bile acids inhibits intestinal cholesterol absorption: studies in the hedgehog pathway regulates remodeling responses to biliary obstruction in gallstone-susceptible mouse. Am J Physiol Gastrointest Liver Physiol rats. Gut 2008;57:1275-1282. 2003;285:G494-G502. 47. Topisirovic I, Culjkovic B, Cohen N, Perez JM, Skrabanek L, Borden KL. 41. Bort R, Signore M, Tremblay K, Martinez Barbera JP, Zaret KS. Hex ho- The proline-rich homeodomain protein, PRH, is a tissue-specific inhibitor meobox gene controls the transition of the endoderm to a pseudostratified, cell of eIF4E-dependent cyclin D1 mRNA transport and growth. EMBO J emergent epithelium for liver bud development. Dev Biol 2006;290:44-56. 2003;22:689-703. 42. Tanaka T, Inazu T, Yamada K, Myint Z, Keng VW, Inoue Y, et al. cDNA 48. Denson LA, Karpen SJ, Bogue CW, Jacobs HC. Divergent ho- cloning and expression of rat homeobox gene, Hex, and functional char- meobox gene hex regulates promoter of the Na(ϩ)-dependent bile acid acterization of the protein. Biochem J 1999;339(Pt 1):111-117. cotransporter. Am J Physiol Gastrointest Liver Physiol 2000;279: 43. Djavani M, Topisirovic I, Zapata JC, Sadowska M, Yang Y, Rodas J, et al. G347-G355. The proline-rich homeodomain (PRH/HEX) protein is down-regulated in 49. Denson LA, Sturm E, Echevarria W, Zimmerman TL, Makishima M, liver during infection with lymphocytic choriomeningitis virus. J Virol Mangelsdorf DJ, et al. The orphan nuclear receptor, shp, mediates bile 2005;79:2461-2473. acid-induced inhibition of the rat bile acid transporter, ntcp. Gastroenter- 44. Ingram WJ, Wicking CA, Grimmond SM, Forrest AR, Wainwright BJ. ology 2001;121:140-147. Novel genes regulated by Sonic Hedgehog in pluripotent mesenchymal 50. Eloranta JJ, Jung D, Kullak-Ublick GA. The human Naϩ-taurocholate cells. Oncogene 2002;21:8196-8205. cotransporting polypeptide gene is activated by glucocorticoid receptor 45. Huang S, He J, Zhang X, Bian Y, Yang L, Xie G, et al. Activation of the and peroxisome proliferator-activated receptor-gamma coactivator-1al- hedgehog pathway in human hepatocellular carcinomas. Carcinogenesis pha, and suppressed by bile acids via a small heterodimer partner-depen- 2006;27:1334-1340. dent mechanism. Mol Endocrinol 2006;20:65-79.