Lc

Z

*C

BpH(zOwpmadbppafC�IpeGioCkHpt

Htpciceaarhpt

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

GASTROENTEROLOGY 2009;136:2316–2324

iver Zonation Occurs Through a �-Catenin–Dependent,-Myc–Independent Mechanism

OÉ D. BURKE,* KAREN R. REED,‡ TOBY J. PHESSE,‡ OWEN J. SANSOM,§ ALAN R. CLARKE,‡ and DAVID TOSH*

Centre for Regenerative Medicine, Department of Biology & Biochemistry, University of Bath, Claverton Down, Bath, UK; ‡School of Biosciences, University of

ardiff, UK; §The Beatson Institute, Garscube Estate, Glasgow, UK

nrelrmppfICs

cs7pr

dutntpSpaticwtf

bDgtnp

ACKGROUND AND AIMS: The Wnt pathway hasreviously been shown to play a role in hepatic zonation.erein, we have explored the role of 3 key components

Apc, �-catenin, and c-Myc) of the Wnt pathway in theonation of ammonia metabolizing enzymes. METH-DS: Conditional deletion of Apc, �-catenin, and c-Mycas induced in the livers of mice and the expression oferiportal and perivenous hepatocyte markers was deter-ined by polymerase chain reaction, Western blotting,

nd immunohistochemical techniques. RESULTS: Un-er normal circumstances, the urea cycle enzyme car-amoylphosphate synthetase I (CPS I) is present in theeriportal, intermediate, and the first few layers of theerivenous zone. In contrast, glutamine synthetase (GS)—nd nuclear �-catenin—is expressed in a complementaryashion in the last 1–2 cell layers of the perivenous zone.onditional loss of Apc resulted in the expression of nuclear-catenin and GS in most hepatocytes irrespective of zone.

nduction of GS in hepatocytes outside the normalerivenous zone was accompanied by a reduction in thexpression of CPS I. Deletion of �-catenin induces a loss ofS and a complementary increase in expression of CPS I

rrespective of whether Apc is present. Remarkably, deletionf c-Myc did not perturb the pattern of zonation. CON-LUSIONS: It has been shown that the Wnt pathway isey to imposing the pattern of zonation within the liver.erein we have addressed the relevance of 3 major Wnt

athway components and show critically that the zona-ion is c-Myc independent but �-catenin dependent.

epatocytes in the adult mammalian liver expressstructural and functional heterogeneity relative to

heir location in the liver acinus.1 Hepatocytes in theeriportal (afferent) zone have a higher capacity for glu-oneogenesis and fatty acid metabolism than hepatocytesn the perivenous (efferent) zone, which have a higherapacity for detoxification.2 Various mechanisms mayxplain the induction of acinar zonation. For intermedi-ry metabolism, the concentration gradients of oxygennd hormones, as well as nerve innervation, may beesponsible for induction and maintenance of hepatocyteeterogeneity.3 Although these gradients of function areronounced, perhaps the most striking example of zona-

ion is observed in the enzymes and pathways of ammo-

ia detoxification. Two systems are responsible for theemoval of ammonia in the liver—the urea cycle and thenzyme glutamine synthetase (GS). In adult mammalianiver, carbamoylphosphate synthetase I (CPS I) is theate-limiting enzyme in the urea cycle and is compart-

entalized in the mitochondria of hepatocytes. CPS Irotein is expressed in hepatocytes running from theeriportal zone toward the perivenous zone, but is absentrom the terminal 1–2 layers of perivenous hepatocytes.n contrast, GS is found in a complementary pattern toPS I and is only localized in the last 1–2 cell layers

urrounding the terminal hepatic venule (central vein).4

The Wnt genes encode a large family of secreted gly-oproteins, homologous to wingless in Drosophila, thatignal at the cell surface via �2 receptors: Frizzled, a-pass transmembrane domain-containing serpentinerotein and the low-density lipoprotein-related proteineceptor (LRP).

Transduction of the Wnt signal is mediated through 3istinct pathways: the canonical Wnt pathway dependentpon activation of �-catenin, the Wnt/Ca2 pathway, andhe Wnt/PCP (planar cell polarity) pathway.5 In the ca-onical pathway, the Wnt–Frizzled–LRP complex leads tohe recruitment and phosphorylation of the cytoplasmicrotein Disheveled, a key transducer of the Wnt signal.ubsequently, GSK3beta and Axin, 2 proteins that formart of a complex with the tumor suppressor proteindenomatous polyposis coli (Apc) that normally directshe phosphorylation and ubiquitination of �-catenin, arenhibited and degraded. In the absence of the Apc proteinomplex, �-catenin is free to translocate to the nucleus,here it can mediate transcription of Wnt target genes

hrough interaction with the transcription factors T-cellactor and lymphocyte enhancer factor. Signaling

Abbreviations used in this paper: Apc, adenomatous polyposis coli;NF, �-napthoflavone; CPS I, carbamoylphosphate synthetase I; Dkk1,ickkopf-1; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GS,lutamine synthetase; Ihh, Indian hedgehog; LRP, low-density lipopro-ein-related protein receptor; PEPCK, phosphoenolpyruvate carboxyki-ase; PFK, 6-phosphofructo-2-kinase; RT-PCR, reverse transcriptaseolymerase chain reaction; Rrm2, ribonucleotide reductase.

© 2009 by the AGA Institute0016-5085/09/$36.00

doi:10.1053/j.gastro.2009.02.063

tl

oitpsdwomstmatgstAececletts

itptAplce�opprticuretdmtc

sstliMar

HwgAbtaip

pSuf

e4Cccsows0a(tuppSif(rLC

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

June 2009 �-CATENIN REGULATES HEPATIC ZONATION 2317

hrough this pathway is very much dependent upon theevel of �-catenin within the cell.6

Wnt signaling pathways play a central role in embry-nic development, maintenance of adult stem cells (eg,

ntestinal stem cells), and the progression of humanumors.7,8 In the last few years, several reports have im-licated Wnt signaling in hepatogenesis. Ober et al9 de-cribed the requirement in zebrafish for mesodermallyerived Wnt signals, acting through the canonical path-ay, for specification of the hepatic endoderm. Becausef the early embryonic lethality of �-catenin mutantice, a similar role for Wnts in mammalian hepatic

pecification has not yet been demonstrated. However,he temporal regulation of �-catenin in the developing

ouse liver is implicated in biliary cell differentiationnd in promoting liver growth and hepatocyte matura-ion.10,11 A role for �-catenin in regulating hepatocyterowth in adult liver is supported by the observation thatome human liver tumors exhibit redistribution and ac-ivation of �-catenin owing to mutations in Axin1, Axin2,pc, and �-catenin.12–14 Perhaps more fascinating, how-ver, is the demonstration that liver tumors in micearrying activating mutations in either �-catenin or Ha-rasxhibit altered zonal gene expression.15 That is, tumorsarrying �-catenin mutations express GS throughout theesion, whereas Ha-ras tumors are negative for GS butxpress E-cadherin (a periportal marker).15 It is observa-ions such as this, from liver tumors, that first suggestedhat �-catenin may be required for mediating the expres-ion of zone-specific genes.

Apc is involved in the degradation of �-catenin andnactivation results in the accumulation of �-catenin,hereby mimicking the activation of the Wnt signalingathway. Inactivation of Apc in the liver alters the zona-ion of ammonia-metabolizing enzymes suggesting thatpc may be involved in maintaining differences betweeneriportal and perivenous hepatocytes.16 Conditional de-

etion of Apc caused a switch in the phenotype of theells: Periportal hepatocytes lost expression of urea cyclenzymes and gained the perivenous phenotype. Blocking-catenin signaling by ectopic adenoviral overexpressionf the Wnt antagonist Dickkopf-1 (Dkk1) induced aeriportal-like phenotype in perivenous cells.16 These ex-eriments established a key role for the Wnt pathway inegulating hepatic zonation. To address the relevance ofhe key components of the Wnt pathway, we have exam-ned the roles of both �-catenin and c-Myc in zonation byonditionally deleting both genes in vivo. The strategy wesed relies on mice in which the promoter element of theat cytochrome CYP1A1 gene is used to regulate Crexpression. The CYP1A1 element is normally transcrip-ionally silent, but can be up-regulated on exposure toefined lipophilic xenobiotics, which bind to a cytoplas-ic aryl hydrocarbon receptor, allowing it to translocate

o the nucleus, where it complexes with other factors to

reate a high-affinity DNA-binding protein.17 This tran- b

criptional complex binds to specific DNA recognitionequences present in the CYP1A1 promoter to initiateranscription. Cre-mediated recombination is obtained iniver and small intestine.18 We determined the effects ofndividual conditional deletion of Apc, �-catenin, and c-

yc on the normal zonal expression of CPS I and GS. Welso extend these observations to mice with combinato-ial deletions of either Apc and �-catenin or Apc and c-Myc.

Materials and MethodsMouse LinesAll experiments were performed under the UK

ome Office guidelines. Outbred male mice from 6 to 12eeks of age were segregated for the C57BJ and S129enomes. The loxP flanked alleles used were as follows:pc,18 �-catenin,19 c-Myc,20 and AhCre.19 To induce recom-ination, mice were given 3 daily intraperitoneal injec-ions of �-napthoflavone (bNF) 80 mg/kg body weightnd livers were harvested either 4 or 21 days after the firstnjection. Mice were humanely killed by a Schedule 1rocedure and the livers removed for processing.

Tissue Fixation, Embedding, and ProcessingDissected liver tissue was fixed for 1 hour in 4%

araformaldehyde or overnight in 10% formalin (Sigma,t Louis, Mo) at 4°C depending on the antibody to besed (see below). After fixation, tissue samples were trans-

erred to 70% ethanol and embedded in paraffin wax.

Immunohistochemical Analysis of AdultMouse LiverSamples were dehydrated through an increasing

thanol series, embedded in paraffin wax and sectioned at�m. Immunoperoxidase staining for CPS I, GS, and

YP2E1 (4% PFA) LECT2 and �-catenin (formalin) wasarried out as follows. Sections were dewaxed in Histo-lear for 7 minutes. This step was then repeated. Theections were then rehydrated through a decreasing seriesf ethanol (100%, 100%, 95%, 90%, 70%, 50%). Sectionsere rinsed in dH2O, washed twice in phosphate-buffered

aline (PBS) for 5 minutes each and permeabilized with.5% triton X 100 for 30 minutes. High temperaturentigen retrieval was performed in citrate or EDTACYP2E1 only) buffer for 30 minutes at 90°C and sec-ions allowed to cool to room temperature for 30 min-tes. After two 5-minute washes in PBS, endogenouseroxidases were quenched with the DAKO Envisioneroxidase block (Glostrup, Denmark) for 10 minutes.ections were washed in PBS and blocked for 30 minutes

n 2% Roche (Mannheim, Germany) blocking buffer be-ore addition of the following antibodies: anti-mouse GS1:400; Transduction Laboratories, San Diego, Calif), anti-abbit CPS (1:1,000; a kind gift of Wouter Lamers),ECT2 (1:500; a kind gift from Satoshi Yamagoe), andYP2E1 (1:500; a kind gift of Magnus Ingelman-Sund-

erg) in blocking buffer overnight at 4°C. Immunostain-

ictmEaeathsfc

(eUsCfap3euTgv(TTGgGA

CGRnA

ldaCsmlmcdbtce

dm�mc(ca1t

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

2318 BURKE ET AL GASTROENTEROLOGY Vol. 136, No. 7

ng for �-catenin (1:50; Transduction Laboratories) wasarried out as previously described.21 Excess primary an-ibody was removed by washing 3 times in PBS for 10

inutes each. Sections were incubated with the DAKOnvision peroxidase-labeled anti-mouse or rabbit second-ry antibody polymer for 30 minutes. The DAB substrat-– chromogen mixture was added to the sections andllowed to develop for 10 minutes. The reaction waserminated in dH2O and the sections counterstained withematoxylin where appropriate. Specimens were ob-erved using a Leica DMRB microscope. Image collectionrom the Leica was made with a Spot camera and imagesollated into figures in Photoshop.

Reverse Transcriptase Polymerase ChainReactionReverse transcriptase polymerase chain reaction

RT-PCR) was performed as described.22 Total RNA wasxtracted from livers using the TRI reagent (Sigma, Poole,K) and first strand complementary DNA was synthe-

ized using the Omniscript RT kit (Qiagen, Valencia,alif). The PCR reactions were carried out under the

ollowing conditions: 94°C for 2 minutes, denaturationt 92°C for 30 seconds, annealing at 56°C (except forhosphoenolpyruvate carboxykinase [PEPCK], 58°C) for0 seconds, and extension at 72°C for 45 seconds. Prim-rs were obtained from Invitrogen and primer sequencessed as follows: APC (forward, 5= TTGATGGAATGTGCTT-GGAA3=; reverse, 5= CACAAGGCTTCCTGGTCTTT3=), Ar-inase 1 (forward, 5= GGTTCTGGGAGGCCTATCTT 3=; re-erse, 5= TTATGGTTACCCTCCCGTTG 3=), �-cateninforward, 5= GCAACCCTGAGGAAGAAGAT 3=; reverse 5=TAGCTCCTTCCTGATGGAG 3=), CPS I (forward, 5=GAGTGGGTCTGCCATGAAC 3=; reverse, 5= TGGACATT-AATGGCCCAGA 3=), glyceraldehyde-3-phosphate dehydro-

enase (GAPDH; forward, 5= TTGTC AGCAATGCAT CCT-CACCA 3=; reverse, 5= GTCTCCTG TGACTTCAAC-

GCAAC 3=), GS (forward, 5= TTTATCTTG- o

ATCGGGTGTG3=; reverse, 5= TTGATGTTGGAG-TTTCGTG 3=), GLT1, glutaminase 2, PEPCK,NAse4, and RHBG (as previously published16). Theumber of cycles was 20 for GAPDH, 30 for �-catenin,PC, and PEPCK and 25 for all other primers.

ResultsAhcre Mice and RecombinationTo investigate the role of �-catenin signaling in

iver zonation, we took advantage of a system for intro-ucing specific gene mutations into the epithelia of thedult murine liver by the transcriptional regulation ofre recombinase. In a transgenic line (Ahcre), cre expres-

ion is inducible from a cytochrome P450 promoter ele-ent that is transcriptionally up-regulated in response to

ipophilic xenobiotics such as bNF. We initially deter-ined the efficiency of Cre-mediated recombination, Ah-

re animals were crossed to the R26R reporter strain andouble-heterozygote animals received bNF (80 mg/kgody weight). �-Galactosidase staining was used to assesshe extent of recombination in whole-mount livers. Re-ombination at a lacZ reporter locus showed extensivexpression of �-galactosidase in the liver (Figure 1A).

Hepatic Deletion of Apc and �-CateninThe Ahcre model provides a simple route for intro-

ucing specific gene mutations into the hepatocytes of theouse liver. To confirm the published roles for Apc and

-catenin in the maintenance of hepatic zonation of am-onia-metabolizing enzymes, Ahcre� mice were inter-

rossed with mice carrying loxP flanked Apc allele:Apc580SApcfl/fl) or loxP flanked �-catenin allele: CatnbtmKem (�-ateninfl/fl).18,19 Appropriate adult mice were identifiednd injected intraperitoneally with 3 doses of bNF over a-day period. The livers were removed after 4 days andhe tissue processed for changes in expression of markers

Figure 1. (A) X-gal–stainedliver from Ahcre mouse. (B–F)Hematoxylin and eosin stainingof wild-type (B), Apc homozy-gous (C), �-catenin homozy-gous (D), Apc/�-catenin doublehomozygous (E), and c-Myc ho-mozygous (F) liver sections.

f zonation using RNA and protein analysis. Histologic

asuamT�tdi

bAa3rr

otAitpccn�ttAiv

Fca�A(fApRePbpG�am

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

June 2009 �-CATENIN REGULATES HEPATIC ZONATION 2319

nalysis on hematoxylin and eosin–stained liver sectionshowed no discernable differences in morphology (Fig-re 1B–E). The Ahcre�Apcfl/fl mice rapidly deterioratedfter recombination possibly owing to hyperammone-ia and hyperglutaminemia as previously reported.16

he Ahcre�c-mycfl/fl, Ahcre� �-Cateninfl/fl and Ahcre�Apcfl/fl

-cateninfl/fl mice do not suffer any ill health after induc-ion with bNF, whereas the Ahcre�Apcfl/flc-mycfl/fl miceeveloped signs of ill health identical to those observed

n Ahcre�Apcfl/fl mice.23

To confirm loss of Apc, we examined Apc expressiony PCR within the liver of different recombinants:hcre�Apc�/�, Ahcre�Apcfl/�, and Ahcre�Apcfl/fl (Figure 2A)nd Ahcre�Apc�/�, �-Cateninfl/�, and �-Cateninfl/fl (FigureA) after bNF treatment. Using primers flanking theecombination site within Apc, where the forward and

igure 2. Complementaryhanges in the zonation of GSnd CPS 1 in the liver after-naphthoflavone induction ofhCre� Apcfl/fl mice for 4 days.

A) RT-PCR analysis of cDNArom wild-type and AhCre�

pc�/fl and Apcfl/fl mice theerivenous markers GS, Glt1,NAse4 and the periportal mark-rs CPS, Arg1, Glutaminase 2,EPCK, and GAPDH. (B) Westernlotting analysis for total �-cateninrotein, GS, CPS, Arg1, andAPDH. (C) Immunostaining for-catenin, GS, CPS, CYP2E1,nd Lect2 on liver sections fromice with the indicated genotype.

everse primers sit on exons 13 and 15, respectively, we �

bserved a band of �362-bp corresponding to the wild-ype allele and 1 of �167 bp representing the recombinedpc580S allele lacking exon 14 (Figure 2A). There was no

ncrease in �-catenin RNA in Ahcre�Apcfl/fl livers; however,he level of �-catenin was moderately elevated at therotein level (Figure 2B). To determine whether the in-reased levels of total �-catenin protein reflected an in-rease in unphosphorylated (activated/nuclear) �-cate-in, we performed immunohistochemistry. Normally,-catenin is strongly localized to the membrane of hepa-

ocytes; however, weak nuclear staining can be seen in theerminal perivenous hepatocytes (Figure 2C). In thehcre�Apcfl/fl mice, we observed strong nuclear �-catenin

n cells of all 3 zones, thus confirming that Apc inacti-ation induces the accumulation of unphosphorylated

-catenin and translocation to the nucleus (Figure 2C).

acno

1Rmobo

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

2320 BURKE ET AL GASTROENTEROLOGY Vol. 136, No. 7

Zonation of Carbamoylphosphate SynthetaseI and Glutamine SynthetaseAccompanying the increase in nuclear �-catenin is

n up-regulation of genes found in perivenous hepato-ytes (GS, glutamate transporter [Glt1], and the ammo-ia transporter [RhBG], RNAse4) and a down-regulation

f periportal genes (CPS I, the urea cycle enzyme arginase A

, Glutaminase 2, and PEPCK; Figure 2A). QuantitativeT-PCR analysis confirmed these observations (Supple-entary Figure 1). We also determined the protein levels

f GS, CPS I, and arginase 1 (Figure 2B) by Westernlotting and densitometry and these reflected the changesbserved at the mRNA level (Supplementary Figure 2).

Figure 3. Complementarychanges in the zonation of GSand CPS I in the liver after�-naphthoflavone induction ofAhCre� Apcfl/fl �-cateninfl/fl and�-cateninfl/fl mice for 21 days. (A)RT-PCR analysis. (B) Westernblotting. (C) Immunostaining forperivenous and periportal mark-ers as for Figure 2.

lthough the PCR and Western blotting data provide an

istptatCa1CppaIeCrCcbotGep

ll(atCmauCtshGahicpbahrt

ewcgmpsdzCndAopfl

wa1

lluemol3scMe(aob

mtc

muathP

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

June 2009 �-CATENIN REGULATES HEPATIC ZONATION 2321

ndication of the changes in the relative levels of expres-ion, they do not tell us whether there are complemen-ary changes in the zonation of CPS I and GS ineriportal and perivenous hepatocytes. To investigatehis possibility, we examined the expression of periportalnd perivenous cell markers in liver sections of wild-ype and mutant mice that were injected with bNF.PS I and GS displayed a normal zonation in wild-typend Ahcre�Apcfl/� animals, namely that GS is located in–2 cell layers surrounding the central vein and thatPS I is located in a complementary fashion in peri-ortal, intermediate, and the first few layers of theerivenous zone. However, in Ahcre�Apcfl/fl mice, CPS Ind GS expression are profoundly altered (Figure 2C).n Ahcre�Apcfl/fl mice, the zonation of GS is dramaticallyxtended into the hepatocytes of the periportal zone andPS I is reduced in a complementary fashion. A similar

esult is obtained when other perivenous markers,YP2E1 and Lect2, were examined, suggesting that the

hanges in gene expression are not restricted to just GS,ut extend to other markers. Although Lect 2 is a markerf zonation, we have shown that it is not important forhe maintenance of zonation because normal zonation ofS is retained in Lect2-null mice (data not shown). Het-

rozygous Ahcre�Apcfll� mice showed an indistinguishablehenotype to that of control animals.

Deletion of �-catenin Induces a Loss ofPerivenous Gene Expression

In Ahcre� �-cateninfl/fl mice, we saw a reducedevel of �-catenin at the RNA and protein levels andack of nuclear �-catenin in perivenous hepatocytesFigure 3) and expression of perivenous markers isbsent or substantially reduced. We performed quan-itative RT-PCR analysis for GS, Glt1, RHBG, RNase 4,PS1, and arginase 1. GS, Glt1, RHBG, and RNase 4RNA levels were down-regulated, whereas CPS 1 and

rginase genes were up-regulated (Supplementary Fig-re 1). At the protein level, GS was down-regulated andPS1 and arginase 1 were up-regulated (Supplemen-

ary Figure 2). This suggests that �-catenin normallyuppresses the periportal phenotype in perivenousepatocytes as well as maintaining the expression ofS and other perivenous markers. To demonstrate an

bsolute requirement for �-catenin in the regulation ofepatic zonation, we examined the phenotypic changes

n livers from different genetic backgrounds includingombined deletion of Apc and �-catenin and the ap-ropriate controls (Figure 3). In mice with deletion ofoth genes, there is a loss of the perivenous phenotypend expression of the periportal phenotype in terminalepatocytes confirming that �-catenin is the criticalegulator of perivenous gene expression (Supplemen-

ary Figures 1 and 2). c

c-Myc Deletion Does Not Alter the Zonationof CPS I and GS in the LiverGiven that we have recently shown that Wnt activity is

ntirely dependent on functional c-Myc in the small intestine,e next tested the c-Myc dependency of zonation.24 We deleted

-Myc on both an Apc heterozygous and an Apc-deficient back-round. Recombination in these mice was induced by treat-ent with bNF and the livers were removed after 21 days for

rocessing. Histologic analysis on hematoxylin and eosin–tained liver sections after deletion of c-Myc alone showed noiscernable differences in morphology (Figure 1F). To analyzeonation, we stained liver sections for �-catenin, GS, CPS I,yp2E1, and Lect2 (Figure 4). After deletion of c-Myc alone, theormal zonation persisted, showing zonation to be indepen-ent of c-Myc. Furthermore, analysis of mice mutant for bothpc and c-Myc showed that the loss of zonation after deletionf Apc occurred independently of c-Myc. The expression oferivenous genes GS, Glt1, RHBG, and RNAse4 in Ahcre�Apcfl/

c-mycfl/fl mice was up-regulated; CPS 1 and arginase 1 expressionere both down-regulated. Protein levels of GS, CPS, andrginase 1 also reflect similar changes (Supplementary Figuresand 2).

Targets of �-CateninWe determined the dependency of �-catenin in the

iver by quantitative RT-PCR, of 3 genes selected from pub-ished microarray analysis.21,25 We chose the catalytic sub-nit of ribonucleotide reductase (Rrm2), an isozyme of thenzyme 6-phosphofructo-2-kinase (Pkfk�3) and the signalingolecule Indian hedgehog (Ihh). We found that the expression

f all 3 genes was significantly increased (�2-fold) in Apcfl/fl

ivers compared with wild-type livers (Supplementary Figure). In contrast Rrm2, Pkfk�3 and Ihh mRNA levels wereignificantly down-regulated in �-Cateninfl/fl animals, indi-ating that these genes are regulated by �-catenin. In c-

ycfl/fl livers, Ihh levels were not significantly altered; how-ver, Rrm2 and Pkfk�3 levels were significantly reducedSupplementary Figure 3). We therefore conclude that Ihh isdirect target of �-catenin and expression is not dependentn c-Myc, whereas Rrm2 and Pkfk�3 expression is regulatedy �-catenin through a c-Myc–dependent mechanism.

DiscussionTo elucidate the role of Wnt/�-catenin pathway in

aintenance of the normal liver, we investigated the dele-ion of the signaling intermediates: Apc, �-catenin, and-Myc alone and in combination.

Until recently, the molecular mechanisms responsible foraintenance of metabolic zonation in the liver were poorly

nderstood. At least some of the gradients of function, suchs those of carbohydrate and xenobiotic metabolism, arehought to be regulated by the gradients of oxygen andormones that run across the liver lobule.2 Evidence fromerret et al has implicated the Wnt/�-catenin system in the

ontrol of hepatic zonation of ammonia metabolizing en-

zAetacihtCtAMo

WaGtcrpopspsv

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

2322 BURKE ET AL GASTROENTEROLOGY Vol. 136, No. 7

ymes.16 The authors showed using conditional deletion ofpc that the zonation can be altered so that the perivenousxpression of GS is extended to the more proximal peripor-al hepatocytes. This change in zonation is accompanied by

complementary reduction in the expression of the ureaycle enzymes. Injection of a viral expression vector encod-ng the Wnt antagonist, Dkk1, also reduced GS and en-anced CPS I expression.16 We wished to determine in detailhe role of Apc, �-catenin, and c-Myc in liver zonation. There/loxP system was used with the Cre transgene driven by

he inducible Cyp1A1 promoter. In adult mice we deletedpc, �-catenin, and the downstream target of �-catenin, c-yc. With disruption of the Apc gene, nuclear �-catenin was

bserved in the majority of cells, indicating activation of m

nt signaling in liver. Gene expression analysis revealed anlteration in a number of genes, including an increase inS, the glutamate transporter Glt1, and the ammonia

ransporter RhBG,26 markers typical of perivenous hepato-ytes. GS and Glt 1 (along with OAT) have previously beeneported as �-catenin target genes27; hence, changes in ex-ression after deletion of Apc are consistent with previousbservations. For the first time, we also examined the peri-ortal genes and found a complementary loss of genes,uggesting a switch in phenotype from periportal toerivenous-like hepatocytes. These data are consistent withtudies of altered GS expression in mice carrying an acti-ated �-catenin transgene,27 �-catenin knockout mice, and

Figure 4. Disruption of zonationin Apc/c-Myc double homozy-gous liver occurs independent ofc-Myc. Comparison of AhCre�

Apcfl/� c-Myc�/�, Apcfl/� c-Mycfl/fl,and Apcfl/fl c-Mycfl/fl liver by immu-nostaining after �-naphthoflavoneinduction for 21 days.

ice ectopically overexpressing Dkk1.16

tdcetlcnttit

opasmmiaohpzrtuei�dte�Itca

WrtusddMfz

ppsh

Wocmccsi

dAfiis

emCtvdMeErcrocrdd

aG1

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

June 2009 �-CATENIN REGULATES HEPATIC ZONATION 2323

Critically, we have now extended these studies to examinehe expression of periportal and perivenous markers aftereletion of either �-catenin or c-Myc. Deletion of �-cateninauses a complete loss of GS and extension of the CPS Ixpression pattern to the hepatocytes surrounding the cen-ral vein. Previously, �-catenin has been conditionally de-eted in adult mouse liver with a loss of GS and some of theytochrome P450 enzymes.28 However, CPS I expression wasot examined. The results of the present study demonstratehat �-catenin normally suppresses the periportal genes inhe terminal perivenous cells and support the idea thatnhibition of Wnt signaling is required for maintenance ofhe periportal phenotype.16

We determined the expression of 3 genes that are targetsf �-catenin: Rrm2, Pfkf�3, and Ihh. Ribonucleoside-diphos-hate reductase is composed of 2 subunits termed RRM1nd RRM2. RRM2 can function as an inhibitor of Wntignaling by suppressing the activity of �-catenin,29 so Rrm2

ay form part of a regulatory loop to regulate �-catenin–ediated transcription. 6-Phosphofructo-2-kinase (PFK-2)

s encoded by 4 genes designated Pfkf�1– 4 and is a potentctivator of the glycolytic pathway.30 An increasing gradientf glycolytic activity runs from periportal to perivenousepatocytes.31 Because Pfkf�3 is regulated by �-catenin, it isossible that Pfkf�3 may control glycolysis in the perivenousone. The Hedgehog pathway is essential for maintainingesident hepatic progenitors in mouse liver, plays a role inhe ductular response to primary biliary cirrhosis, and isp-regulated in human hepatocellular carcinoma.32–34 Lev-ls of all 3 genes (Rrm2, Pfkf�3, and Ihh) were up-regulatedn AhCre�Apcfl/fl livers and down-regulated in AhCre�

-Cateninfl/fl livers. We propose that Rrm2 and Pfkf�3 areirect targets of c-Myc in agreement with previous observa-ions.35,36 Therefore, down-regulation of Rrm2 and Pfkf�3xpression observed in c-Mycfl/fl liver may be mediated by-catenin via a c-Myc–dependent mechanism. In contrast,

hh is up-regulated by �-catenin and is likely to be a directarget because the expression was not significantly altered in-Mycfl/fl liver. However, it is not known how the Hedgehognd Wnt pathways interact within the liver lobule.

c-Myc has been proposed as a critical mediator of thent pathway and indeed deficiency of c-Myc completely

epresses the phenotype of Apc deficiency in the intes-ine.24 Remarkably, the expression of CPS I and GS werenaffected by the loss of c-Myc in either a wild-typeetting or after loss of Apc. Our results therefore begin toeconvolute the Wnt-mediated control of zonationownstream of �-catenin, such that we demonstrate c-yc– dependent transcriptome changes to be irrelevant

or establishing zonation of ammonia-metabolising en-ymes in the liver.

Two questions arise in relation to Wnt regulation oferiportal and perivenous phenotype: (1) Which Wnts areroduced in the liver? and (2) What is the source of the Wntignal responsible for activating �-catenin in the perivenous

epatocytes? Monga et al have shown that a number of

nts were present in the adult mouse liver.37 Wnts mayriginate either from hepatocytes or from the endothelialells of the central veins. Two lines of evidence suggest thisay be the case. First, Gebhardt et al have shown that

o-culture of the epithelial cell line RL-ET-14 induces GS inultured GS-negative periportal hepatocytes.38 The seconduggests that direct interaction with the central veins maynduce GS position-specific expression.39

In the present study, we have directly determined theependence of the adult liver on �-catenin by using thehcre model. There is an absolute requirement for �-catenin

or the maintenance of GS zonation. In the future, it will bemportant to distinguish between targets of �-catenin andnduction of a different gene program, and to determine theource of Wnts involved in metabolic zonation.40

The Wnt/�-catenin system has been shown to play anssential role in the maintenance of the stem cell compart-ent and in cell proliferation in a number of organs.41

onditional deletion of Apc in the intestine prevents migra-ion and differentiation of intestinal epithelial cells.23 Pre-ious studies have also suggested roles for �-catenin in liverevelopment and cancer (reviewed in Thompson andonga42). For example, Monga et al10 identified temporal

xpression of �-catenin in embryonic mouse liver between10 and 14 and down-regulation of �-catenin causes aeduction in proliferation and cell survival. Herein, we haveonfirmed and extended observations that identify a criticalole for Wnt pathway components in normal liver physiol-gy. In particular, we demonstrated that �-catenin is theritical mediator of zonation in the liver. Given the dualoles of the Wnt pathway in normal and diseased liver, theeconvolution of this pathway has clear ramifications forissecting the mechanisms underlying both processes.

Supplementary Data

Note: To access the supplementary materialccompanying this article, visit the online version ofastroenterology at www.gastrojournal.org, and at doi:0.1053/j.gastro.2009.02.063.

References

1. Gebhardt R. Metabolic zonation of the liver: regulation and implica-tions for liver function. Pharmacol Ther 1992;53:275–354.

2. Tosh D, Albert KGMM, Agius L. Glucagon regulation of gluconeogen-esis and ketogenesis in periportal and perivenous rat hepatocytes.Biochem J 1988;256:197–204.

3. Jungermann K, Keitzmann T. Oxygen: modulator of metabolic zona-tion and disease of the liver. Hepatol 2000;31:255–260.

4. Gebhardt R, Lindros K, Lamers WH, et al. Hepatocellular heteroge-neity in ammonia metabolism: demonstration of limited colocaliza-tion of carbamoylphosphate synthetase and glutamine synthetase.Eur J Cell Biol 1991;56:464–467.

5. Kikuchi A, Kishida S, Yamamoto H. Regulation of Wnt signaling byprotein–protein interaction and post-translational modifications. ExpMol Med 2006;38:1–10.

6. Miller JR, Hocking AM, Brown JD, et al. Mechanism and function ofsignal transduction by the Wnt/beta-catenin and Wnt/Ca2� path-

ways. Oncogene 1999;18:7860–7872.

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

4

4

4

R

RUeCU

A

fPSW

C

F

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

2324 BURKE ET AL GASTROENTEROLOGY Vol. 136, No. 7

7. Cadigan K, Nusse R. Wnt signaling: a common theme in animaldevelopment. Genes Dev 1997;11:3286–3305.

8. Pinto D, Clevers H. Wnt, stem cells and cancer in the intestine. BiolCell 2005;97:185–196.

9. Ober EA, Verkade H, Field HA, et al. Mesodermal Wnt 2b signalingpositively regulates liver specification. Nature 2006;442:688–691.

0. Monga SP, Monga HK, Tan X, et al. Beta-catenin antisense studiesin embryonic liver cultures: role in proliferation, apoptosis, andlineage specification. Gastroenterology 2003;124:202–216.

1. Hussain SZ, Sneddon T, Tan X, et al. Wnt impacts growth anddifferentiation in ex vivo liver development. Exp Cell Res 2004;292:157–169.

2. Taniguchi K, Roberts LR, Aderca IN, et al. Mutational spectrum ofbeta-catenin, AXIN1 and Axin2 in hepatocellular carcinomas andhepatoblastomas. Oncogene 2002;21:4863–4871.

3. Nhieu JT, Renard CA, Wei Y, et al. Nuclear accumulation of mutatedbeta-catenin in hepatocellular carcinoma is associated with in-creased cell proliferation. Am J Pathol 2000;155:703–710.

4. Colnot S, Decaens T, Niwa-Kawakita M, et al. Liver-targeted disrup-tion of Apc in mice activates beta-catenin signaling and leads tohepatocellular carcinomas. Proc Natl Acad Sci U S A 2004;101:17216–17221.

5. Hailfinger S, Jaworski M, Braeuning A, et al. Zonal gene expressionin murine liver: lessons from tumors. Hepatology 2006;43:407–414.

6. Benhamouche S, Decaens T, Godard C, et al. Apc tumor suppres-sor gene is the “zonation-keeper” of mouse liver. Dev Cell 2006;10:759–770.

7. Matsushita N, Sogawa K, Ema M, et al. A factor binding to thexenobiotic responsive responsive element (XRE) of P-450 1A1 geneconsists of at least 2 helix-loop-helix proteins, Ah receptor and Arnt.J Biol Chem 1993;268:21002–21006.

8. Shibata H, Toyama K, Shioya H, et al. Rapid colorectal adenomaformation initiated by conditional targeting of the Apc gene. Science1997;278:120–123

9. Ireland H, Kemp R, Houghton C, et al. Inducible cre-mediated con-trol of gene expression in the murine gastrointestinal tract: effect ofloss of beta-catenin. Gastroenterology 2004;126:1236–1246.

0. Muncan V, Sansom OJ, Tertoolen L, et al. Rapid loss of intestinalcrypts upon conditional deletion of the Wnt/Tcf-4 target gene c-Myc.Mol Cell Biol 2006;26:8418–8426.

1. Sansom OJ, Reed KR, Hayes AJ, et al. Loss of Apc in vivo immedi-ately perturbs Wnt signalling, differentiation and migration. Genesand Development 2004;18:1385–1390.

2. Burke ZD, Shen C-N, Ralphs KL, et al. Characterization of liverfunction in transdifferentiated hepatocytes. J Cell Physiol 2006;206:147–159.

3. Reed KR, Athineos D, Meniel VS, et al. B-catenin deficiency, but notMyc deletion, suppresses the immediate phenotypes of APC loss inthe liver. Proc Natl Acad Sci U S A 2008;105:18919–18923.

4. Sansom OJ, Meniel VS, Muncan V, et al. c-Myc deletion rescues Apcdeficiency in the small intestine. Nature 2007;446:676–679.

5. Tan X, Behari J, Cieply B, et al. Conditional deletion of beta-cateninreveals its role in liver growth and regeneration. Gastroenterology2006;131:1561–1572.

6. Weiner ID, Miller RT, Verlander JW. Localization of the ammoniumtransporters, Ph B glycoprotein and Ph C glycoprotein, in the mouseliver. Gastroenterology 2003;124:1432–1440.

7. Cadoret A, Ovejero C, Terris B, et al. New targets of beta-cateninsignalling in the liver are involved in the glutamine metabolism.Oncogene 2002;21:8293–8301.

8. Loeppen S, Koehle C, Buchmann A, et al. A beta-catenin-dependentpathway regulates expression of cytochrome P450 isoforms in

mouse liver tumors. Carcinogenesis 2005;26:239–248.

9. Tang LY, Deng N, Wang LS, et al. Quantitative phosphoproteomeprofiling of Wnt3a-mediated signaling network: indicating the involve-ment of ribonucleoside-diphosphate reductase M2 subunit phos-phorylation at residue serine 20 in canonical Wnt signal transduc-tion. Mol Cell Proteomics 2007;6:1952–1967.

0. Duran J, Navarro-Sabate A, Pujol A, et al. Overexpression of ubiqui-tous 6-phosphofructo-2-kinase in the liver of transgenic mice resultsin weight gain. Biochem Biophys Res Commun 2008;365:291–297.

1. Jungermann K, Katz N. Functional specialization of different hepa-tocyte populations. Physiol Rev 1989;69:708–764.

2. Sicklick JK, Li YX, Melhem A, et al. Hedgehog signaling maintainsresident hepatic progenitors throughout life. Am J Physiol Gastroin-test Liver Physiol 2006;290:G859–870.

3. Jung Y, McCall SJ, Li YX, et al. Bile ductules and stromal cellsexpress hedgehog ligands and/or hedgehog target genes in primarybiliary cirrhosis. Hepatology 2007;45:1091–1096.

4. Sicklick JK, Li YX, Jayaraman A, et al. Dysregulation of the Hedge-hog pathway in human hepatocarcinogenesis. Carcinogenesis2006;27:748–757.

5. Kuschak TI, Kuschak BC, Taylor CL, et al. c-Myc initiates illegitimatereplication of the ribonucleotide reductase R2 gene. Oncogene2002;21:909–920.

6. Osthus R, Shim H, Kim S, et al. Deregulation of glucose transporter1 and glycolytic gene expression by c-Myc. J Biol Chem 2000;275:21797–21800.

7. Zeng G, Awan F, Otruba W, et al. Wnt’er in the liver: expression ofWnt and frizzled genes in mouse. Hepatology 2007;45:195–204.

8. Schrode W, Mecke D, Gebhardt R. Induction of glutamine syn-thetase in periportal hepatocytes by co-cultivation with a liver epi-thelial cell line. Eur J Cell Biol 1990;53:35–41.

9. Kuo FC, Darnell JE Jr. Evidence that interaction of hepatocytes withthe collecting (hepatic) veins triggers position-specific transcriptionof the glutamine synthetase and ornithine aminotransferase genesin the mouse liver. Mol Cell Biol 1991;11:6050–6058.

0. Burke ZD, Tosh D. The Wnt/beta-catenin pathway: master regulatorof acinar zonation? Bioessays 2006;28:1072–1077.

1. Reya T, Clevers H. Wnt signalling in stem cells and cancer. Nature2005;434:843–850.

2. Thompson MD, Monga SP. WNT/beta-catenin signalling in liverhealth and disease. Hepatology 2007;45:1298–1305.

Received June 10, 2008; Accepted February 26, 2009.

eprint requestsAddress requests for reprints to: David Tosh, Centre for

egenerative Medicine, Department of Biology & Biochemistry,niversity of Bath, Claverton Down, Bath BA2 7AY, United Kingdom.-mail: d.tosh@bath.ac.uk; fax: (�44) 1225 386779; or to Alanlarke, School of Biosciences, University of Cardiff, Cardiff CFIO 3AX,nited Kingdom. e-mail: clarker@cf.ac.uk; fax: (44) 1225 386779.

cknowledgmentsZDB and KRR contributed equally to this work.The authors thank the Wellcome Trust and Cancer Research UK

or financial support. We also thank Professor Wouter Lamers,rofessor Satoshi Yamagoe, and Professor Magnus Ingelmannundberg for antibodies and Mark Bishop, Lucy Pietzka, Irynaithington, and Derek Scarborough for technical assistance.

onflicts of interestThe authors disclose no conflicts.

unding

Supported by the Wellcome Trust and Cancer Research UK.

cRI(AGGGGaucog(GAAwi

S4

b[cadp

mTmtogcT0i�CT3worsEa

pmccS4

p�Adl(p

June 2009 �-CATENIN REGULATES HEPATIC ZONATION 2324.e1

Supplementary Materials and Methods

Quantitative Real-Time ReverseTranscriptase Polymerase Chain Reaction

Quantitative real-time reverse transcriptase polymerasehain reaction (qRT-PCR) was carried out using theoche Light Cycler and SYBR premix Ex taq (Takara Bio

nc). Primer sequences used were as follows: PFKF�3forward: 5= GAAGCCATGGACGACTTCAT3=; reverse: 5=ACAGCCTGGTGACAGATGA 3=), RRM2 (forward: 5=ATAGAGCAGGAGTTCCTCA 3=; reverse: 5= AACAG-ACTGAGGTAGCTAC 3=), and IHH (forward: 5= CTT-CCTACAAGCAGTTCAG 3=; reverse: 5= TGAGTCTC-ATGACCTGGAA 3=); all other primers are as listed

bove. The relative amounts of mRNA were calculatedsing the �� CT method. The CT value of a gene is theycle number during the PCR at which a certain thresh-ld T of amplification is reached. The CT values wereenerated by the Roche Lightcycler and the mean valuesC� T) calculated for target genes and the reference geneapdh. The fold change for each target gene inhcre�Apcfl/fl, Ahcre��-Cateninfl/fl, Ahcre�Apcfl/fl�-Cateninfl/fl,hcre�c-Mycfl/fl, and Ahcre�Apcfl/flc-Mycfl/fl liver comparedith wild-type controls was calculated using the follow-

ng equations:

�CT � C�T�target� � C�T�reference�, and

�� CT � �CT�control� � �CT�experimental�.

tatistical analysis was performed using Graph Pad Prismsoftware.

Western BlottingProtein extracts were prepared from liver samples

y rapid freeze–thaw in lysis buffer (0.5 mol/l Tris-HClpH.6.8] containing a 1:100 dilution of protease inhibitorocktail [Sigma, St Louis, Mo]). Samples were centrifugedt 14,000 rpm for 10 minutes at 4°C to remove cellularebris and quantified using the Bio-Rad (Hercules, Calif)

rotein assay reagent. Protein extracts were denatured by P

ixing with a 2 � SDS-PAGE loading buffer (0.125 mol/lris-HCl [pH 6.8], 4% [w/v] SDS, 20% [v/v] glycerol, 0.2ol/l DTT, 0.02% [w/v] bromophenol blue) and heated

o 100°C for 5 minutes. We separated 10 �g of proteinn a 5% or 10% Criterion precast Tris-HCl polyacrylamideel (Bio-Rad) and transferred on to a BioTraceNT nitro-ellulose membrane (Pall Corporation, Pensacola, Fla).he membrane was blocked using 5% (w/v) Marvel in.1% (v/v) PBS-Tween and subsequently probed. Antibod-

es were obtained and diluted as follows: anti-mouse-catenin (1:5000), anti-mouse GS (1:15000), anti-rabbitPS (1:15000), anti-mouse Arginase 1 (1:2,000; BDransduction Laboratories). Anti-mouse glyceraldehyde--phosphate dehydrogenase (GAPDH; 1:2,000; Ambion)as used as a loading control. Secondary antibodies werebtained and diluted as follows: peroxidase labeled anti-abbit or anti-mouse both at 1:2000 (Amersham Bio-ciences, Bucks, UK). The signal was detected with theCL TM Western blotting analysis system (Amersham)nd developed on Hyperfilm (Amersham).

DensitometeryDensitometric analysis of western blotting was

erformed using the Scion Image software. Band densityeasurements were corrected against a GAPDH-loading

ontrol for each sample and the fold change calculated inomparison with levels observed in wild-type animals.tatistical analysis was performed using Graph Pad Prismsoftware.

Selection of �-Catenin Targets FromMicroarray AnalysisWe identified potential �-catenin targets from

reviously published microarray data using the criteria of2-fold up-regulation after the conditional deletion ofPC48 and �2-fold down-regulation after the conditionaleletion of �-catenin.49 Three genes that are linked to

iver development, Wnt signaling and/or liver cancerRrm2, pkfkb3, and Ihh), were selected from our list ofotential �-catenin targets for further analysis by qRT-

CR.

SaA

Sai

2324.e2 BURKE ET AL GASTROENTEROLOGY Vol. 136, No. 7

upplementary Figure 1. Fold change in periportal and perivenous gene expression. Bar charts show the fold change in expression of CPS1,rginase1, GS, Glt1, RHBG, RNAse4, and �-catenin in wild-type AhCre�Apcfl/fl, AhCre� �-cateninfl/fl, AhCre�c-Mycfl/fl, AhCre�Apcfl/flc-Myc fl/fl, and

hCre� Apcfl/fl �-Cateninfl/fl animals. Statistically significant changes and P values are indicated (*).

upplementary Figure 2. Fold change in periportal and perivenous protein levels. Bar charts show the fold change in the levels of CPS1,rginase1, GS, and �-catenin in wild-type AhCre� Apcfl/fl, AhCre� �-cateninfl/fl, AhCre�c-Mycfl/fl, AhCre�Apcfl/flc-Mycfl/fl, and AhCre� Apcfl/fl �-caten-

nfl/fl animals. Statistically significant changes and P values are indicated (*).

SpPa

June 2009 �-CATENIN REGULATES HEPATIC ZONATION 2324.e3

upplementary Figure 3. Fold change in �-catenin target gene ex-ression. Bar charts show the fold change in expression of Rrm2,fkf�3, and Ihh in wild-type AhCre� Apcfl/fl, and AhCre� �-cateninfl/fl

nimals. Statistically significant changes and P values are indicated (*).