Acetaldehyde Inhibits PPARγ via H2O2-Mediated c-Abl Activation in Human Hepatic Stellate Cells

AH

EVA*IIO

BtfitmgvicAwcawaTvaTwascocfPmssbg

Fattemoiai

GASTROENTEROLOGY 2006;131:1235–1252

cetaldehyde Inhibits PPAR� via H2O2-Mediated c-Abl Activation in Humanepatic Stellate Cells

LISABETTA CENI,*,‡ DAVID W. CRABB,§ MARCO FOSCHI,� TOMMASO MELLO,¶ MIRKO TAROCCHI,*ALENTINO PATUSSI,*,# LUCA MORALDI,** RENATO MORETTI,** STEFANO MILANI,*,‡ CALOGERO SURRENTI,*,‡,# andNDREA GALLI*,‡

Gastroenterology Unit, Department of Clinical Pathophysiology, ‡Center of Excellence for Research, Transfer and Higher Education, DENOthe, and �Department ofnternal Medicine, University of Florence, Florence, Italy; §Departments of Medicine and Biochemistry and Molecular Biology, Indiana University School of Medicine,ndianapolis, Indiana; ¶FiorGen Foundation, Florence, Italy; and #Centro Alcologico Regionale and **First Unit of General Surgery and Transplantation, Azienda

uc(iamHflph

tHpovtttaitcCc

adaaf

mpvmcvrcl

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

ackground & Aims: Accumulating evidence indicateshat acetaldehyde (AcCHO) is one of the main mediators ofbrogenesis in alcoholic liver disease. AcCHO stimulates syn-hesis of fibrillar collagens in hepatic stellate cells, but the

olecular events directly involved in the activation of collagenenes are debatable. Methods: Peroxisome proliferator-acti-ated receptor � (PPAR�) is a nuclear receptor that is expressedn stellate cells, and its activation by specific ligands inhibitsollagen synthesis. In this study, we evaluated the effects ofcCHO on PPAR� transcriptional activity and its correlationith the AcCHO-induced collagen synthesis in hepatic stellate

ells. Results: AcCHO treatment inhibited ligand-dependentnd -independent PPAR� transcriptional activity, and this effectas correlated with an increased phosphorylation of a mitogen-ctivated protein kinase site at serine 84 of the human PPAR�.ransfection of the PPAR�Ser84Ala mutant completely pre-ented the effect of AcCHO on PPAR� activity and in parallelbrogated the induction of collagen gene expression by AcCHO.he effect of AcCHO on PPAR� activity and phosphorylationas blocked by extracellular signal–regulated kinase (ERK) 1/2nd protein kinase C (PKC)� inhibitors as well as by catalase,uggesting that hydrogen peroxide is involved in the molecularascade responsible for PPAR� phosphorylation via activationf the PKC�/ERK pathway. Furthermore, inhibition of c-Ablompletely abrogated the effect of AcCHO on either PPAR�unction or collagen synthesis; in addition, expression of thePAR�Ser84Ala mutant prevented the profibrogenic signalsediated by c-Abl activation. Conclusions: Our results

howed that the induction of collagen expression by AcCHO intellate cells is dependent on PPAR� phosphorylation inducedy a hydrogen peroxide–mediated activation of the profibro-enic c-Abl signaling pathway.

ibrosis is a key histologic feature that characterizes theprogression of alcohol-induced hepatic injury in chronic

lcohol abusers.1 The mechanisms whereby ethanol administra-ion induces liver damage and excess collagen deposition inhese tissues remain incompletely understood. Accumulatingvidence indicates that acetaldehyde (AcCHO), the major activeetabolite of ethanol oxidation, is one of the main mediators

f fibrogenesis in alcoholic liver disease.2 In fact, AcCHO-mod-fied epitopes correlate with progression of liver fibrosis both inlcoholic patients and in alcohol-fed animals.3,4 Furthermore,

late synthesis of fibrillar-forming collagens and structural gly-oproteins of extracellular matrix in hepatic stellate cellsHSCs).5 These cells are mesenchymal pericytes with character-stic intracytoplasmatic lipid droplets rich in retinyl esters andre currently considered the primary source of extracellularatrix components in the liver.6 When cultured on plastic,SCs undergo a spontaneous transformation from the resting

at-storing phenotype into highly proliferative myofibroblast-ike cells, thereby mimicking the process of activation thatrevails in vivo after chronic injuries, including long-term alco-ol consumption.7

The precise molecular events directly involved in the activa-ion of collagen genes by AcCHO are quite complex. In humanSCs, AcCHO induces the transcription of the �1(I) and �2(I)rocollagen genes by a mechanism that requires the activationf a protein kinase C (PKC)-dependent pathway, which is in-olved in a rapid increase of the steady state levels of AP-1ranscription factors.8,9 In turn, AP-1 activation was postulatedo be involved in the AcCHO-induced expression of the basicranscription element binding protein, which is able to trans-ctivate the rat �1(I) collagen promoter.10 Although AcCHO canncrease DNA binding of NF-1 and C/EBP� transcription fac-ors to their specific cis-acting regulatory sites located in mouseollagen promoters,11,12 the intracellular signals regulating Ac-HO-induced collagen gene expression in humans remain un-

ertain.The peroxisome proliferator-activated receptor � (PPAR�) is

member of the nuclear receptor superfamily of ligand-depen-ent transcription factors that is predominantly expressed indipose tissue, where it has been shown to have a key role indipogenesis and in regulation of insulin resistance.13 PPAR�orms a heterodimer with retinoid X receptor and alters tran-

Abbreviations used in this paper: AcCHO, acetaldehyde; DMSO, di-ethyl sulfoxide; ERK, extracellular signal–regulated kinase; HSC, he-

atic stellate cell; JNK, c-Jun-N-terminal kinase; MAP, mitogen-acti-ated protein; MEK, mitogen-activated protein kinase kinase; MOI,ultiplicity of infection; PEG-CAT, polyethylene glycol–conjugated

atalase; PKC, protein kinase C; PPAR, peroxisome proliferator-acti-ated receptor; PPRE, peroxisome proliferator response element; RGZ,osiglitazone; siRNA, small interfering RNA; SDS-PAGE, sodium dode-yl sulfate/polyacrylamide gel electrophoresis; SFIF, serum-free/insu-in-free medium; TGF, transforming growth factor; wt, wild-type.© 2006 by the American Gastroenterological Association (AGA) Institute

0016-5085/06/$32.00

spoBlttkg

tabmczHc

hhwbstdcpop

aNLBP(Ahdwa(abPpamMoS(

h

tcuopUcIfsiotcebgmttFwpsipti�P1Messt(mldhpbc

tfpasgPppegg

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

1236 CENI ET AL GASTROENTEROLOGY Vol. 131, No. 4

cription of numerous target genes after binding to specificeroxisome proliferator response elements (PPRE), consistingf a hexameric direct repeat separated by a single nucleotide.14

esides the role of PPAR� in adipocyte differentiation and inipid and glucose metabolism, additional biological implica-ions of this receptor have been recently identified.15 PPAR�ranscriptional regulation is involved in cytokine and chemo-ine expression, control of cell growth, motility, and pro-rammed death.16,17

PPAR� is expressed in human HSCs, and its expression andranscriptional activity are reduced during cell activation in vivond in vitro.18,19 In activated stellate cells, activation of PPAR�y specific ligands inhibits growth factor–induced proliferation,igration, and cytokine production.19,20 Furthermore, we re-

ently showed that PPAR� activators such as antidiabetic thia-olidinediones inhibited collagen gene expression in culturedSCs and prevented collagen deposition in different toxic and

holestatic animal models of liver fibrosis.21

In consideration of the recent observation that AcCHO in-ibits the transcriptional activity of PPARs in cultured ratepatocytes,22 the present study was designed to determinehether a derangement of PPAR� transcriptional activity mighte involved in ethanol-induced collagen synthesis. The resultshow that in human HSCs, AcCHO inhibits PPAR� transcrip-ional activity at the posttranslational level by an H2O2-depen-ent phosphorylation of the receptor and that activation of-Abl is involved in the signaling cascade responsible for PPAR�hosphorylation. Thus, these data suggest that the impairmentf PPAR� function may be a critical event in the fibrogenicrocess induced by long-term alcohol consumption.

Materials and MethodsMaterialsCulture media, trypsin, all restriction endonucleases,

nd DNA-modifying enzymes were from GIBCO (Grand Island,Y). Stractan, Nycodenz, and protein A Sepharose were fromife Technologies (Milan, Italy). Platelet-derived growth factorB, TPA, UO186, PD98059, Gö6976, rottlerin, myristylatedKC�, calphostin C, �V1-2, and �V1-1 were from Calbiochem

La Jolla, CA). Nitrocellulose membranes (Hybond) were frommersham (Milan, Italy). Pronase was from Boehringer Mann-eim (Monza, Italy). Antibodies against phosphorylated andiphosphorylated extracellular signal–regulated kinase (ERK1/2)ere from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodiesgainst PKC� and anti-Tyr(P) were from Upstate BiotechnologyUpstate Biotechnology, Inc, Charlottesville, VA). Antibodiesgainst �2-microglobulin were from Abcam (Abcam Inc, Cam-ridge, MA). Imatinib mesylate was from Novartis (Novartisharmaceuticals, East Hanover, NJ). Rosiglitazone (RGZ) andioglitazone were from GlaxoSmithKline (Welwyn, England)nd Takeda Chemicals (Tokyo, Japan), respectively. Radioactiveaterial was purchased from New England Nuclear (Boston,A). Matrigel was from Becton Dickinson (Bedford, MA). All

ther reagents were from Sigma Chemical Co (Milan, Italy).prague–Dawley rats were purchased from Charles RiverComo, Italy).

Isolation and Culture of HSCsHSCs were obtained from wedge sections of normal

uman liver unsuitable for transplantation, after approval by t

he local ethics committee. After combined 0.5% pronase/0.05%ollagenase tissue digestion, human HSCs were isolated byltracentrifugation over 4 gradients (1.111/1.080/1.058/1.053)f Stractan and characterized as previously described.23 HSCurity (as estimated by the autofluorescence of the cells byV-excited fluorescence microscopy and by the ability to ex-

lude propidium iodide) was �95%. Cells were cultured inscove’s modified Dulbecco’s medium supplemented with 20%etal bovine serum, 2 mmol/L glutamine, 0.1 mmol/L nones-ential amino acids, 1 mmol/L sodium pyruvate, 0.6 U/mLnsulin, and 1% antibiotic-antifungal solution. Primary culturesf HSCs were allowed to grow to confluence, subcultured byrypsinization (0.025% trypsin/0.5 mmol/L EDTA), and thenultured in the same medium as previously described. Somexperiments were performed in quiescent cells cultured on aasement membrane–like substrate (Matrigel) using the “thinel” method (50 �L/cm2 of growth surface) according to theanufacturer’s instructions. To assess the effect of different

reatments, confluent cells were made quiescent by placinghem in serum-free/insulin-free medium (SFIF) for 24 hours.or ethanol and AcCHO incubation, airtight culture dishesere used to prevent evaporative loss. Some experiments wereerformed in the presence of kinase inhibitors. These sub-tances have been previously tested in HSCs, and their specific-ty at the concentrations used in the present study has beenreviously shown as follows: PD98059 (mitogen-activated pro-ein kinase kinase [MEK] inhibitor), 30 �mol/L; UO126 (MEKnhibitor), 50 �mol/L; calphostin (pan-PKC inhibitor), 1mol/L; Gö6976 (PKC� inhibitor), 100 nmol/L; myristylatedKC�, 40 nmol/L; rottlerin (PKC� inhibitor), 2 �mol/L; �V1-1,�mol/L; �V1-2, 1 �mol/L; and imatinib mesylate, 1 �mol/L.ost substances were dissolved in dimethyl sulfoxide (DMSO)

xcept for myristylated PKC�, �V1-1, �V1-2, and imatinib me-ylate, which were dissolved in water. Preliminary experimentshowed that cellular toxicity of the different treatments used inhe present study can be disregarded for the following reasons:1) lack of trypan blue staining of the cells during the experi-

ental period within the concentration used and (2) lack ofactate dehydrogenase leakage from cells into the culture me-ium using the cytotoxicity detection kit (Boehringer Mann-eim, Mannheim, Germany). In this study, experiments wereerformed on activated �-smooth muscle actin—positive cellsetween the first and third serial passages or in quiescent cellsultured on Matrigel, using lines obtained from 12 patients.

Vector Constructs and Transfection ofCultured Human HSCsCells were transfected by calcium phosphate precipita-

ion18 at a density of 5 � 105 cells/60-mm dish with theollowing vectors: 2.5 �g of PPRE (ARE-7)3-tk-luciferase re-orter plasmid (containing 3 copies of the PPRE from thedipocyte lipid-binding protein [aP2] gene ligated to a herpesimplex thymidine kinase promoter upstream of a luciferaseene),24 5 �g of human wild-type (wt) or mutated (Ser84Ala)PAR�1 expression plasmid, and 2.5 �g of 3.5-kilobase humanro-�2(I) collagen promoter/CAT construct (pMS3.5/CAT).25

SV40-Luc control vector or pSV2CAT (vectors containing SV40arly promoter and enhancer sequences that drive the luciferaseene or the chimeric chloramphenicol acetyl transferase [CAT]ene, respectively) (5 �g) were used as internal controls for

2emQJnCSiCbTwwohbmmaT3mpb

(KltHfuaihaa1twwpbgTppopfcprthh(

kTasiwptm4ppabTB

miMta

dt(NsiEifom(apt

rmbtfwcmtugtA

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

October 2006 EFFECTS OF ACETALDEHYDE ON PPAR� ACTIVITY 1237

0%) was evaluated in preliminary experiments using anti-lucif-rase antibodies (Promega, San Luis, CA). Site-directedutagenesis of pcmx-PPAR�1 was conducted using theuikChange Site– directed mutagenesis kit (Stratagene Inc, La

olla, CA) following the manufacturer’s instructions. The oligo-ucleotide used in mutagenesis was CAAAGTGGAGCCTG-AGCTCCACCTTATTATTCTGAGAAGACTC, and it changeder84 to Ala. The total amount of DNA transfected was normal-

zed with a carrier DNA (pcDNA3.1; Invitrogen Corp, Carlsbad,A). Four hours later, the cells were exposed to phosphate-uffered saline (PBS) containing 15% glycerol for 3 minutes.he cells were rinsed twice with PBS, and fresh SFIF mediumas added. Twelve hours after transfection, cells were treatedith either ethanol or AcCHO with or without PPAR� ligandsr their vehicle (DMSO). Twenty-four hours later, the cells werearvested, washed twice with PBS, and lysed in 150 �L of auffer containing 25 mmol/L Tris, pH 7.8, 2 mmol/L EDTA, 20mol/L dithiothreitol, 10% glycerol, and 1% Triton X-100. Fiftyicroliters of cell extract was incubated with the luciferase

ssay reagent based on the original protocol of de Wet et al.26

he number of relative light units was determined with a-second delay and a 30-second incubation. CAT activity waseasured as described previously.27 The conversion of chloram-

henicol to its acetylated products was quantified on an Ambiseta scanner (Ambis System, San Diego, CA).

Immunoprecipitation and Western BlottingCells were scraped and homogenized in ice-cold buffer

pH 7.4) consisting of 50 mmol/L Tris, pH 7.4, 150 mmol/LCl, 1% Triton X-100, 1 mmol/L EDTA, 5 mmol/L N-ethylma-

eimide, and 0.2 mmol/L phenylmethylsulfonyl fluoride. Pro-ein content was determined according to the Lowry method.28

omogenates were divided into aliquots and stored at �80°Cor later use. Cell extracts (50 �g/lane) were boiled for 3 min-tes in Laemmli sample buffer, separated by gel electrophoresis,nd electroblotted onto nitrocellulose filters. Nonspecific bind-ng sites were blocked by incubating nitrocellulose sheets for 1our in PBS containing 5% low-fat dry milk. For ERK1/2ctivation assays, nitrocellulose sheets were then incubated withnti-tyrosine phosphorylated ERK1/2 (p42MAPK/p44MAPK)(1:0,000) antibody. For PKC� activation assay, cell lysate, ob-ained as described previously, was incubated for 45 minutesith 20 �L of protein A Sepharose beads incubated beforehandith 1 �g anti-PKC� for 4 hours at 4°C. The immunocom-lexes were then solubilized in Laemmli depolymerizationuffer and subjected to sodium dodecyl sulfate/polyacrylamideel electrophoresis (SDS-PAGE) using 9% polyacrylamide gel.yrosine-phosphorylated PKC� was detected using anti-phos-hotyrosine at a dilution of 1:1000. For immunodetection ofrotein, the membrane was probed with anti-PKC� at a dilutionf 1:500. For PPAR� protein expression, nuclear extracts (40 �grotein) were prepared as described in the following text andractionated in 12% SDS-PAGE. Proteins were detected by in-ubating with anti-PPAR� antibodies (1:100 dilution) (rabbitolyclonal H-100; Santa Cruz Biotechnology) overnight atoom temperature, followed by the appropriate secondary an-ibody conjugated with horseradish peroxidase (1:1,000) for 2ours at room temperature. Blots were developed with an en-anced chemiluminescence detection system (ECL plus) kit

Pharmacia Bioscience, Arlington Heights, IL). s

c-Abl Kinase AssayCells were lysed for 30 minutes at 4°C in 750 �L of

inase buffer (50 mmol/L Tris [pH 7.4], 150 mmol/L NaCl, 1%riton X-100, 0.1% SDS, 1% sodium deoxycholate, 0.1 TIU/mLprotinin, 50 �g/mL phenylmethylsulfonyl fluoride, 1 mmol/Lodium vanadate, and 1 mg/mL leupeptin). Extracts were clar-fied, and equivalent proteins were incubated overnight at 4°Cith anti-Abl (K12; Santa Cruz Biotechnology). Immune com-lexes were collected with protein A Sepharose and washedwice in kinase lysis buffer (25 mmol/L Tris [pH 7.4], 10

mol/L MgCl2, 1 mmol/L dithiothreitol) before incubation in0 �L kinase buffer containing 5 �mol/L adenosine triphos-hate, 2 �g GST-Crk, and 0.5 �Ci/rxn [�32P]adenosine triphos-hate. The kinase reaction was allowed to proceed for 5 minutest 37°C, stopped with 40 �L (�2 concentration) Laemmliuffer, and visualized by autoradiography after SDS-PAGE.otal c-Abl proteins were detected using an antibody from BDiosciences-PharMingen (no. 554148).

Isolation of Nuclear Protein Extract andDNA Binding AssayNuclear proteins were isolated from HSCs based on a

icropreparation method.29 The nuclear extract was suspendedn 20 mmol/L HEPES (pH 7.9), 420 mmol/L NaCl, 1.5 mmol/L

gCl2, 0.2 mmol/L EDTA, 25% glycerol, 0.5 mmol/L dithio-hreitol, and 0.2 mmol/L phenylmethylsulfonyl fluoride, andliquots were frozen in liquid nitrogen and stored at �70°C.

Electrophoretic mobility shift assays were performed by ra-iolabeling double-stranded oligonucleotides corresponding tohe PPRE ARE-7 of the mouse adipocyte lipid-binding proteinaP2) (5=-TGCACATTTCACCCAGAGAGAAGGGATTGA-3=).30

uclear extracts (20 �g) were incubated with 1–2 �g of non-pecific competitor DNA (poly [dIC]) in binding buffer contain-ng 10 mmol/L HEPES, pH 7.9, 60 mmol/L KCl, 1 mmol/LDTA, and 7% (vol/vol) glycerol on ice for 15 minutes. Where

ndicated, specific competitor oligonucleotides were added be-ore the addition of labeled probe and incubated for 15 minutesn ice. For supershift assays, antibodies were added and theixture was incubated an additional 1–2 hours. Labeled probe

20,000 cpm) was added last and the reaction incubated andditional 15 minutes on ice. Reaction mixtures were electro-horesed on a nondenaturing 4% acrylamide gel and subjectedo autoradiography.

RNA Extraction and Northern BlotHybridizationTotal RNA was extracted by guanidinium-phenol-chlo-

oform methods of Chomczynski and Sacchi31 with minorodifications.23 Cellular messenger RNA levels were measured

y Northern blot hybridization. Ten micrograms of denaturedotal RNA per sample was electrophoresed in a 1% agarose/3%ormaldehyde gel and transferred to a Nylon membrane. Filtersere prehybridized and then hybridized overnight at 42°C with

omplementary DNA probe of rat �1(I) procollagen.31 Comple-entary DNA probe was labeled with [32P]deoxycytidine 5=-

riphosphate to a specific activity of 2–5 � 108 cpm/�g DNA bysing a random primer kit (Amersham, Little Chalfont, En-land). Equal loading of the samples was checked by reprobinghe blots with a probe encoding the ribosomal protein 36B4.fter hybridization, filters were washed 4 times in 2� standard

aline citrate/0.1% SDS at 65°C and then exposed to Kodak

Xb

UM2cCGmm[mafistralT

asatRm

Ss3fgbimmiiSP

raAiwSpm

b(iftSdwa3fdiguvcuctp

cRtvsTcaTTGapfiapfitcrmamDpct(vf(ia

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

1238 CENI ET AL GASTROENTEROLOGY Vol. 131, No. 4

AR-5 film (Kodak, Milan, Italy) at �70°C. The hybridizationands were quantitated by scanning laser densitometry.

Reverse-Transcription Polymerase ChainReactionOne microgram of RNA was reversed transcribed with 2

of Moloney murine leukemia virus reverse transcriptaseMLV (Life Technologies Inc) at 42°C for 60 minutes in a

0-�L mixture in the presence of random hexamers. The nu-leotide bases used were sense primer 5=-TCTGGCCCAC-AACTTTGGG-3= and antisense primer 5=-CTTCACAAGCAT-AACTCCA-3= (base pairs 1275–1294) for human PPAR�. Twoicroliters of reverse-transcribed mixture was subjected to poly-erase chain reaction in a 20-�L mixture (10 mmol/L Tris-HCl

pH 8.3], 50 mmol/L KCl, 2.0 mmol/L MgCl2, 0.01% gelatin, 20mol/L deoxynucleotide triphosphate, 0.5 U of Taq polymer-

se [Life Technologies Inc], and 0.25 pmol of primers). Thirty-ve cycles of reaction at 94°C for 50 seconds, 60°C for 45econds, and 72°C for 90 seconds were performed by DNAhermal cycler (Perkin-Elmer Cetus, Norwalk, CT). Efficiency ofeverse transcription was controlled in each sample by polymer-se chain reaction amplification of human �2-microglobu-in (sense 5=-GCAAAAGATGAGTATGCCTG-3=, antisense 5=-TCACTCAATCCAAATGCGG-3=).

Collagen AssayProcollagen type I was determined in culture media by

n enzyme-linked immunoassay method as previously de-cribed.23 An anti-monkey procollagen I N-terminal peptidentibody that cross-reacted with human protein was used forhe assay. Standards and samples were assayed in triplicate.esults were expressed as micrograms of procollagen type I pericrograms of cellular DNA.

In Vivo Labeling and Analysis of32P-Labeled PPAR�Transfected cells were serum starved for 24 hours in

FIF, pretreated with phosphate-free medium for 1 hour, andubsequently incubated in 0.8 mCi of [32P]orthophosphate at7°C for 3 hours. Cells were incubated with kinase inhibitorsor 30 minutes, followed by addition of either AcCHO orrowth factors. AcCHO stimulation proceeded for 15 minutesefore removal of the media and cell lysis. Cells were harvested

n radioimmune precipitation lysis buffer (10% glycerol, 137mol/L NaCl, 1% Nonidet, 0.5% deoxycholate, 0.1% SDS, 20mol/L Tris, pH 8.0, 2 mmol/L EDTA, complete protease

nhibitors, and 20 mmol/L NaVO4). Whole cell extracts weremmunoprecipitated with anti-PPAR� antibody and protein Aepharose for 16 hours at 4°C and resolved in 10% SDS-PAGE.hosphoproteins were visualized by autoradiography.

Adenoviral TransductionRecombinant adenoviral vectors were constructed from

eplication-deficient adenovirus type 5 with deletions in the E1nd E3 genes as described previously32 and obtained using thedEasy Adenoviral Vector System (Stratagene Inc) The catalyt-

cally inactive mutant of MEK is characterized by a substitutionith Glu and Asp, respectively, of the regulatory residueser218 and Ser222 and by deletion of predicted �-helix encom-assing residues 32–51.32 AdPKC�K376R is characterized by a

y converting the invariant lysine in the catalytic domainamino acid 376) to an arginine.33 The double proline mutationn the regulatory SH2-CD linker (P242E/P249E, c-AblKA) con-erring constitutive activity to c-Abl and the kinase dead coun-erpart (K290M, c-AblKD) were obtained with the QuikChangeite-Directed Mutagenesis Kit (Stratagene Inc) as previouslyescribed by Barilà and Superti-Furga.34 Recombinant virusesere propagated in 50 T175 flasks of HEK 293 cells infected atmultiplicity of infection (MOI) of 5. Cells were recovered

6 – 48 hours after infection and viruses released by 5 cycles ofreeze thawing. All viral preparations were purified by CsClensity gradient centrifugation, dialyzed, and stored at �70°C

n 10 mmol/L Tris-HCl, pH 7.4, 1 mmol/L MgCl2, and 10%lycerol. Titers of viral stocks were determined by plaque assaysing HEK 293 cells. Stellate cells were transduced with adeno-iral vectors at an MOI of 60 in 1 mL of Iscove’s mediumontaining 2% fetal bovine serum. Transduction efficiency eval-ated by �-galactosidase staining was up to 90%. After 1 hour,ulture medium was added to the plate and cells were main-ained for 12 hours before transient transfection by calciumhosphate precipitation as described previously.

Adenovirus-Delivered Small Interfering RNAWe adopted the AdEasy-1 system (Stratagene Inc) to

onstruct the AdSiRNA targeting c-Abl in human HSCs. TheNA polymerase III H1-RNA gene promoter was cloned into

he promoter-less shuttle vector pShuttle to obtain the newector designed pShuttle-H1, which can drive the expression ofmall interfering RNA (siRNA) in recombinant adenovirus.35

he 64-nucleotide oligonucleotides encoding human cAbl-spe-ific siRNA were designed as previously described by Ohba etl36 (5=-GATCTCCCGCAACTACATCACGCCAGTttcaagagaAC-GGCGTGATGTA GTTGCTTTTTGGAAA-3= and 5=-AGCTT-TCCAAAAAGCAACTACATCACGCCAGT tctcttgaaACTGGC-TGATGTAGTTGCGGGA-3=). These oligonucleotides were

nnealed and subcloned to the BglII and HindIII sites ofShuttle-H1 to get pShuttle-H1-cAbl. Subcloning was con-rmed by EcoRI digestion, and the inserted sequences werenalyzed by dideoxy sequencing. A recombinant adenovirus wasroduced by a double recombination event between cotrans-ormed adenoviral backbone plasmid pAdEasy-1 and a linear-zed pShuttle-H1-cAbl following the manufacturer’s instruc-ions. Briefly, pShuttle-H1-cAbl was linearized with PmeI andotransformed with pAdEasy-1 into BJ5183 cells by electropo-ation. Positive clones were selected and confirmed by DNA

iniprep and PacI digestion. Plasmids from correct clones weremplified by transforming into DH5� cells followed by DNAaxiprep (Qiagen, Hilden, Germany). The resulting adenoviralNA (AdH1-cAbl or AdH1-empty) was linearized with PacI andurified by ethanol precipitation. A total of 1.5 � 106 packagingells (HEK 293) were plated in a 25-cm2 flask the day beforeransfection and transfected by 24 �g of Dosper liposomeRoche, Basel, Switzerland) with 6 �g of PacI linearized adeno-iral DNA. The next day, the medium containing the trans-ection mix was replaced with 6 mL of growth mediumDulbecco’s modified Eagle medium). Transfected cells werencubated for an additional 7 days, and virus was harvestednd amplified according to the manufacturer’s instructions

stTdoebdiSe

httda1teltuacmnsvr5

cN�

7atnl2(paacg(

PubaerHdtatmaHe

FHccieabc[ptwaestd

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

October 2006 EFFECTS OF ACETALDEHYDE ON PPAR� ACTIVITY 1239

Measurement of Hydrogen Peroxide ReleaseIntracellular production of H2O2 was assayed as de-

cribed previously.37 Briefly, cells were starved for 48 hours andhen incubated for the indicated times with 85 �mol/L AcCHO.hree minutes before the end of the incubation time, 2=,7=-ichlorofluorescein diacetate was added to a final concentrationf 5 �mol/L. This compound was converted by intracellularsterases to 2=,7=-dichlorofluorescein, which was then oxidizedy H2O2 to give a highly fluorescent compound. Cells were thenetached from the substrate by trypsinization and analyzed

mmediately by flow cytometry by using a BD Biosciences FAC-can flow cytometer equipped with an argon laser lamp (FL-1;mission, 488 nm; band pass filter, 530 nm).

Animal ExperimentsMale Sprague–Dawley rats weighing 180 –200 g were

oused in plastic cages with a wire-mesh floor providing isola-ion from a hygienic bed and were exposed to a 12-hour con-rolled light cycle. They were fed with Good Laboratory Practiceiet (Nossan, Milano, Italy). Experiments were performed inccordance with the institutional ethical guidelines. A total of20 rats, divided into 12 groups, were killed to isolate HSCs athe indicated time points after an intraperitoneal injection ofither 2 mL/kg of CCl4 or its vehicle (olive oil). Imatinib mesy-ate (50 mg · kg�1 · day�1) or its vehicle (water) was adminis-ered by gavage beginning 3 days before CCl4 injection and dailyntil the rats were killed. Rat HSCs were isolated from controlnd treated animals by in situ perfusion with pronase andollagenase according to the method of Knook et al38 withinor modifications as described previously. Activities of ala-

ine aminotransferase and aspartate aminotransferase were as-ayed in venous blood samples obtained from the inferior cavaein 24 hours after CCl4 treatment and determined by theoutine method used in the hospital laboratory (Technicon RA00; Miles, Inc, Tarrytown, NY).

Statistical AnalysisResults are expressed as mean � SD. Group means were

ompared by analysis of variance, followed by the Student–ewman–Keuls test if the former was significant. A P value of.05 was considered statistically significant.

ResultsEthanol Metabolism–Derived AcCHO InhibitsPPAR� Transcriptional Activity in HSCsThe activity of a PPAR�-responsive reporter gene (ARE-

3-tk-luciferase) was used as an index of PPAR� function inctivated HSCs. As previously described,18 activated HSCs con-ain low amounts of transcriptionally active PPAR�, and bothatural and synthetic ligands such as 15d-PGJ2 and thiazo-

idinediones (RGZ and pioglitazone) induced reporter activity- to 3-fold over controls treated with vehicle alone (DMSO)Figure 1A). Cotransfection with human PPAR� expressionlasmid increased reporter activity more than 2-fold in HSCs,nd a further induction was obtained when PPAR� ligands weredded to PPAR�-transfected cells. Ethanol at the physiologi-ally relevant concentration of 20 mmol/L inhibited both li-and-independent and -stimulated PPAR� activity in HSCs

The dose dependency of ethanol-induced inhibition ofPAR� transcriptional activity was further characterized (Fig-re 1B). Ethanol showed a clear dose-dependent inhibition ofasal and RGZ-stimulated PPAR� activity, both in the presencend absence of transfected PPAR�. Ethanol treatment did notxert any effect in control experiments using a tk-luciferaseeporter (not shown). In consideration of the fact that humanSCs have the capacity to metabolize ethanol via the alcoholehydrogenase/aldehyde dehydrogenase enzymatic pathway,39

he alcohol dehydrogenase inhibitor 4-methylpyrazole and theldehyde dehydrogenase inhibitor cyanamide were used to de-ermine if the effect of ethanol on PPAR� was dependent on its

etabolism. Neither compound affected ARE-73-tk-luciferasectivity in activated HSCs in control experiments (not shown).owever, 4-methylpyrazole completely prevented the effect of

thanol on PPAR� function, while cyanamide augmented the

igure 1. Effect of ethanol on PPAR� transcriptional activity inSCs. Serum-starved human HSCs were cotransfected using the cal-ium phosphate procedure with the reporter ARE-73-tk-luc and with thehloramphenicol acetyl transferase expression vectors (pSV2-CAT) as

nternal controls for transfection efficiency. pcmx-PPAR� or pcmxmpty vector were also cotransfected where indicated. The totalmount of DNA transfected in each set of cells was maintained constanty using a carrier DNA (pcDNA3.1). Twelve hours after transfection,ells were treated with (A) PPAR� ligands (RGZ, 20 �mol/L; pioglitazonePGZ], 20 �mol/L; 15d-PGJ2, 5 �mol/L) or its vehicle (DMSO) in theresence or absence of ethanol (20 mmol/L) or (B) increasing concen-rations of ethanol (5–50 mmol/L). After 24 hours of incubation, cellsere harvested for luciferase and CAT assay as described in Materialsnd Methods. Data are expressed as mean � SD from 5 experiments,ach of which was performed in duplicate. *P � .05 or higher degree ofignificance versus control. Open circles, P � .05 versus pcmx-PPAR�ransfected cells with no ligands; closed circles, P � .02 or higheregree of significance versus no ethanol treatment conditions.

nhibitory effect (Figure 2A). This suggests that AcCHO gener-

aa

PtaslbTo

nsAHtDRtWAelswcoRps

mytTttmAtmwwept

gcpmtrmCcp

i

d(aiS(tp

Ftf7(Pwpch0hMostee

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

1240 CENI ET AL GASTROENTEROLOGY Vol. 131, No. 4

ted from ethanol was responsible for the inhibition of PPAR�ctivity.

AcCHO Inhibits PPAR� Function via aPhosphorylation-Dependent MechanismThe role of AcCHO as mediator of the ethanol-induced

PAR� inhibition was confirmed by the effect of low concen-rations of exogenous AcCHO (50 –150 �mol/L), which wereble to inhibit ligand-dependent and -independent PPAR� tran-criptional activity (Figure 2B). There was no evidence of cellu-ar toxicity at these doses of AcCHO as documented by trypanlue exclusion and lactate dehydrogenase leakage (not shown).o understand how AcCHO impaired PPAR� function, it was

igure 2. Effect of ethanol and its metabolism on PPAR� transcrip-ional activity in HSCs. Serum-starved human HSCs were cotrans-ected using the calcium phosphate procedure with the reporter ARE-3-tk-luc, the chloramphenicol acetyl transferase expression vector

pSV2-CAT) as internal control for transfection efficiency, and pcmx-PAR� or pcmx empty vector. Twelve hours after transfection, cellsere treated with (A) RGZ (20 mmol/L) or its vehicle (DMSO) in theresence or absence of ethanol (20 mmol/L) or (B) an increased con-entration of AcCHO (50–150 �mol/L). Inhibitors of alcohol and alde-yde dehydrogenase (4-methylpyrazole and cyanamide) were used at.1 mmol/L where indicated. After 24 hours of incubation, cells werearvested for luciferase and CAT assay as described in Materials andethods. Data are expressed as mean � SD from 5 experiments, eachf which was performed in duplicate. *P � .05 or higher degree ofignificance versus (A) no ethanol or (B) no AcCHO treatment condi-ions. Open circles, P � .03 or higher degree of significance versusthanol-treated cells with no inhibitors; closed circles, P � .001 versusthanol-treated cells plus cyanamide.

f interest to study the effect of AcCHO on the ability of t

uclear proteins extracted from HSCs to bind a PPAR� re-ponse element (ARE-7) in electrophoretic mobility shift assays.s shown in Figure 3A, nuclear extracts from control-activatedSCs contained proteins that retarded the ARE-7 oligonucleo-

ide (Figure 3A, lane 1). Treatment with RGZ enhanced PPAR�NA binding in parallel with the increase of PPAR� messengerNA and protein expression as documented by a semiquanti-

ative reverse-transcription polymerase chain reaction andestern blotting (Figure 3A, lane 2). As expected, the PPAR�/

RE-7 complex was strongly induced when nuclear proteinsxtracted from PPAR�-transfected HSCs were used (Figure 3A,ane 4). The specificity of this band was confirmed by itsignificant reduction when the reaction mixture was incubatedith antibodies against PPAR� (Figure 3A, lanes 3 and 5) or by

ompetition with 200 molar excess of an unlabeled oligonucle-tide (not shown). Treatment with AcCHO did not affect eitherGZ- or pioglitazone-dependent up-regulation of PPAR� ex-ression and DNA binding, either in the presence or the ab-ence of transfected PPAR� (Figure 3B and C).

PPAR� activity is regulated at the posttranslational level byitogen-activated protein (MAP) kinase– dependent phosphor-

lation of serine 84 (equivalent to serine at 82 amino acids ofhe mouse PPAR�1), which reduces its transcriptional activity.40

hus, we tested the effect of AcCHO on PPAR� phosphoryla-ion by in vivo labeling with [32P]orthophosphate in PPAR�-ransfected HSCs. Cell lysates were prepared after AcCHO treat-

ent and immunoprecipitated with PPAR�-specific antibodies.s shown in Figure 3D, PPAR� was weakly phosphorylated in

he absence of AcCHO in activated HSCs. However, after 15inutes of treatment with AcCHO, PPAR� phosphorylationas markedly increased. Similarly, receptor phosphorylationas also induced by platelet-derived growth factor and phorbol

ster (TPA). Time course analysis indicated that PPAR� phos-horylation occurred 5 minutes after AcCHO treatment andhen returned to control levels after 6 hours (Figure 3E).

In consideration that AcCHO induces collagen synthesis andene expression in fully activated HSCs5,9,12 but not in quies-ent cells,41 we tested the effect of AcCHO on PPAR� phos-horylation in freshly isolated HSCs cultured on a basementembrane–like substratum (Matrigel). In this culture condi-

ion, HSCs maintain a quiescent phenotype characterized byounded shape, high density of vitamin A– containing droplets,

inimal proliferation, and no synthesis of collagen type I.42

onfirming previous data,43,44 AcCHO was not able to induceollagen synthesis in quiescent HSCs and did not modify thehosphorylation status of the receptor (not shown).

These data indicate that AcCHO might impair PPAR� activ-ty at the posttranslational level in activated HSCs.

PPAR� Phosphorylation Is Involved inAcCHO-Induced Collagen SynthesisTo verify that AcCHO inhibition of PPAR� activity was

ependent on phosphorylation of Ser84, the Ser84 ¡ Ala mutantpcmx-PPAR(Ser84Ala)) was created by site-directed mutagenesisnd was cotransfected with the PPAR�-responsive reportern activated HSCs. Similar to the wild-type construct, theer84 ¡ Ala mutant induced reporter activity more than 4-foldFigure 4A). AcCHO repressed PPAR� activity in wild-typeransfected HSCs, but the activity induced by transfection withcmx-PPAR�(Ser84Ala) was resistant to AcCHO-mediated inhibi-

rtppwoepW

aoePrCcPcv

cPicTivm�cfS(pP

k

Fst(megOrcPactWw

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

October 2006 EFFECTS OF ACETALDEHYDE ON PPAR� ACTIVITY 1241

To confirm that Ser84 was the phosphorylated residue inesponse to AcCHO treatment, the Ser84 ¡ Ala mutant wasransfected into HSCs and in vivo labeling was performed in theresence or absence of 85 �mol/L AcCHO treatment. As ex-ected, the phosphorylation of the wild-type PPAR� (PPAR�(wt))as enhanced by AcCHO treatment whereas phosphorylationf the mutant was unaffected (Figure 4B, upper panel). In allxperiments, similar amounts of both mutant and wild-typeroteins were expressed in the transfected HSCs, as shown byestern blot analysis (Figure 4B, lower panel).Because PPAR� was shown to modulate collagen synthesis

nd gene expression,19 –21,45 the effect of increased cellular levelf mutant and wild-type PPAR� on AcCHO-induced collagenxpression was tested (Figure 5). To obtain a high level ofPAR� receptor expression, HSCs were transduced by adenovi-us carrying PPAR� constructs. As previously described,5 Ac-HO increased procollagen type I production about 3-fold over

ontrol in human HSCs (Figure 5A). Infection with pAd-PAR�(wt) reduced AcCHO-induced procollagen about 30%ompared with control cells infected with p-Ad-�gal. Con-

igure 3. Effect of AcCHO on PPRE binding and PPAR� phosphorerum-starved human HSCs transfected using the calcium phosphateransfection, cells were treated with RGZ (20 �mol/L) or pioglitazone (P85 �mol/L). After 24 hours of incubation, nuclear proteins were extrac

obility shift assays were performed with a probe consisting of a doulement ARE-7. Nuclear extracts were incubated with 20,000 cpm of thel. Where indicated, antibodies against PPAR� (Ab) were used to confine microgram of total RNA extracted from transfected HSCs was reve

eaction using specific primers for PPAR� and �2-microglobulin (�2) ashain reaction products were electrophoresed on ethidium bromide–coPAR� and �2-microglobulin protein levels in the nuclear extract. (D andnd then incubated in phosphate-free media with 0.8 mCi of [32P]orthophoncentrations of AcCHO (85–150 �mol/L), platelet-derived growth facime points. Lysates were immunoprecipitated with PPAR� antibodies

estern blots with PPAR� antibodies showing PPAR� levels in transfhich was performed in duplicate, are shown.

ersely, the effect of AcCHO on procollagen type I synthesis was b

ompletely abrogated in cells infected with the mutantPAR�(Ser84Ala). Northern blot analysis confirmed the stronger

nhibitory effect of the Ser84 ¡ Ala mutant on AcCHO-inducedollagen gene expression than wild-type receptor (Figure 5B).o further link PPAR� phosphorylation to collagen synthesis

nduced by AcCHO, HSCs were transfected with a CAT reporterector driven by a 3.5-kilobase human pro-�2(1) collagen pro-oter (pMS-3.5/CAT). In transduced HSCs, the addition of 85mol/L AcCHO increased the promoter activity 3.5-fold over

ontrol. This activity was reduced about 38% in cells cotrans-ected with pAd-PPAR�(wt), whereas cotransfection with theer84 ¡ Ala mutant abrogated the effect of AcCHO completelyFigure 5C). Taken together, these results suggest that therofibrogenic effect of AcCHO was prevented by blockingPAR� phosphorylation.

PKC�/ERK Signaling Pathway Is Involved inPPAR� Phosphorylation by AcCHOSerine at residue 84 of the human PPAR� is a MAP

inase consensus recognition site that becomes phosphorylated

n. (A–C) PPRE binding activity was evaluated in nuclear extracts fromcedure with pcmx-PPAR� or pcmx empty vector. Twelve hours after0 �mol/L) or its vehicle (DMSO) in the presence or absence of AcCHOs described in Materials and Methods. (Upper panels) Electrophoretictrand oligonucleotide, which contains a copy of the PPAR� responsebe and then analyzed by electrophoresis through a 4% polyacrylamidee presence of PPAR� in the PPRE binding complexes. (Middle panels)anscribed using random hexamers and amplified by polymerase chainribed in Materials and Methods. The reverse-transcription polymeraseing agarose gel. Lower panels represent Western blot analysis showingSCs were transfected with pcmx-PPAR�, serum starved for 24 hours,ate for 3 hours. Subsequently, the cells were treated with (D) increasing(10 ng/mL), or TPA (100 ng/mL) for 15 minutes or (E) for the indicated

ubjected to SDS-PAGE and autoradiography. Lower panels representcells. Representative data from 3 independent experiments, each of

ylatiopro

GZ) (2ted able-se prorm thrse trdescntain

E) Hosph

tor BBand sected

y ERK1/246 and c-Jun-N-terminal kinase (JNK).47 To verify the

stsatp6a

enod

FaHcishwfisf.tPrwwstWt3a

FcccPmp�ooAHRMbdss�pT�lit

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

1242 CENI ET AL GASTROENTEROLOGY Vol. 131, No. 4

ignaling pathways involved in AcCHO inhibition of PPAR�ranscriptional activity, we used specific kinase inhibitors. Ashown in Figure 6A, 2 different MEK1/2 inhibitors, PD98059nd UO126, completely abrogated the inhibition of PPAR�ranscriptional activity mediated by AcCHO (Figure 6A) andrevented receptor phosphorylation induced by AcCHO (FigureB). On the contrary, inhibition of JNK by SP600125 did notffect AcCHO regulation of PPAR� function (not shown).

Investigations over the past decade have shown that manyffects of ethanol on stellate cell function are closely intercon-ected with PKC-dependent pathways. Confirming previousbservations in rats,48 we found that human HSCs express 4

igure 4. Ser84 ¡ Ala PPAR� mutant is resistant to AcCHO-medi-ted transcriptional repression and phosphorylation. (A) Serum-starvedSCs were cotransfected with the reporter ARE-73-tk-luc and with thehloramphenicol acetyl transferase expression vector (pSV2-CAT) as

nternal control for transfection efficiency. The human PPAR� expres-ion vector (pcmx-PPAR�(wt)) or the Ser84 ¡ Ala mutant (pcmx-PPAR�(Ser84Ala)) were also cotransfected where indicated. Control cellsere cotransfected with pcmx empty vector. Twelve hours after trans-

ection, cells were treated with AcCHO (85 �mol/L). After 24 hours ofncubation, cells were harvested for luciferase and CAT assay as de-cribed in Materials and Methods. Data are expressed as mean � SDrom 5 experiments, each of which was performed in duplicate. *P �002 or higher degree of significance versus control. (B) Cells wereransfected with the wild-type PPAR� expression vector (pcmx-PAR�(wt)) or with the Ser84 ¡ Ala mutant (pcmx-hPPAR�(Ser84Ala)), se-

um starved for 24 hours, and then incubated in phosphate-free mediaith 0.8 mCi of [32P]orthophosphate for 3 hours. Subsequently, the cellsere treated with or without AcCHO (85 �mol/L) for 15 minutes. Ly-ates were immunoprecipitated with PPAR� antibodies and subjectedo SDS-PAGE and autoradiography. The lower panel represents a

estern blot using PPAR� antibodies to demonstrate levels of wild-ype and mutant PPAR� in transfected HSCs. Representative data fromindependent experiments, each of which was performed in duplicate,re shown.

ifferent PKC isoforms: PKC�, PKC�, PKC�, and PKC� (not i

igure 5. PPAR� phosphorylation is involved in AcCHO-inducedollagen synthesis. (A) Human HSCs were transduced with adenoviralonstructs at an MOI of 60 for 12 hours. After 24 hours in SFIF medium,ells were treated with 85 �mol/L AcCHO for an additional 24 hours.rocollagen type I was determined in culture by an enzyme-linked im-unoassay method as described in Materials and Methods. Data, ex-ressed as micrograms per microgram of cellular DNA, are mean valuesSD of 5 experiments performed in triplicate. *P � .02 or higher degree

f significance versus no AcCHO treatment. °P � .05 or higher degreef significance versus pAd-�gal–infected cells treated with AcCHO. (B)fter 6 hours of treatment with AcCHO, total RNA was extracted fromSCs transduced with adenoviral constructs. Ten micrograms of totalNA were used for Northern blot analysis as described in Materials andethods. The size of messenger RNA transcripts (expressed in kilo-ases) is indicated on the right. Representative data from 3 indepen-ent experiments, each of which was performed in duplicate, arehown. (C) After 12 hours of transduction with the adenoviral con-tructs, HSCs were cotransfected with the 3.5-kilobase human pro-2(I) collagen promoter/CAT construct (pMS3.5/CAT) and with theSV40-Luciferase vector as internal control for transfection efficiency.wenty-four hours after transfection, cells were treated with AcCHO (85mol/L) for an additional 24 hours and then harvested for CAT and

uciferase assay. The data are mean � SD for 4 experiments performedn triplicate. P � .05 or higher degree of significance versus no AcCHOreatment conditions. Open circles, P � .05 or higher degree of signif-

stHush(PPiiPtadaMeuc

rHt

EvieuppuhLt3(wup

aa

Fu(aaoeH[wP ent e

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

October 2006 EFFECTS OF ACETALDEHYDE ON PPAR� ACTIVITY 1243

hown). To examine whether AcCHO affects PPAR� transcrip-ional activity via a PKC-dependent pathway, we pretreatedSCs with pan- or isoform-specific PKC inhibitors for 30 min-tes before exposing the cells to 85 �mol/L AcCHO. Of allubstances tested, the pan-inhibitor calphostin, the PKC� in-ibitor rottlerin, and the PKC� translocation inhibitor peptide

�V1-1) completely prevented the inhibitory effect of AcCHO onPAR� activity (Figure 6C) and blocked AcCHO-inducedPAR� phosphorylation (Figure 6D), whereas the PKC� inhib-

tor Gö6976, myristylated PKC�, and PKC� translocation inhib-tor peptide (�V1-2) had no effect on AcCHO regulation ofPAR� activity and phosphorylation. To further substantiatehe role of PKC� and ERK1/2 in AcCHO inhibition, we evalu-ted PPAR� transcriptional activity in cultured HSCs trans-uced with wild-type and dominant negative MEK and PKC�denoviral constructs. As expected, dominant negative pAd-EKK97M and pAd-PKC�K376R completely prevented the AcCHO

ffect on reporter activity, whereas a negligible effect was doc-mented in HSCs when they were transduced with wild-typeonstructs (Figure 7A).

It has been shown that PKC and ERK1/2 play a key role inegulating AcCHO-induced collagen synthesis in activatedSCs.49 Western blot analysis and immunoprecipitation were

igure 6. ERK1/2 and PKC� are involved in AcCHO-mediated effesing the calcium phosphate procedure with the reporter ARE-73-t

pSV2-CAT) as internal control for transfection efficiency. pcmx-PPAR� ofter transfection, cells were treated with RGZ (20 �mol/L) or its vehicle (nd MEK inhibitors was evaluated by preincubating cells with each inhif incubation, cells were harvested for luciferase and CAT assay as desxperiments, each of which was performed in duplicate. *P � .01 or higuman HSCs were transfected with pcmx-PPAR�, serum starved fo

32P]orthophosphate for 3 hours. Subsequently, the cells were treatedith PPAR� antibodies and subjected to SDS-PAGE and autoradiograpPAR� levels in transfected cells. Representative data from 3 independ

hus performed to evaluate the effect of AcCHO on PKC� and t

RK1/2 activation in these cellular culture systems. PKC� acti-ation was evaluated by monitoring tyrosine phosphorylationn response to AcCHO in anti-PKC� immunoprecipitates. Theffects of AcCHO on ERK1/2 phosphorylation were determinedsing antibodies that specifically recognize the active tyrosine-hosphorylated forms of ERK1 and ERK2. PKC� was stronglyhosphorylated 5 minutes after AcCHO treatment (Figure 7B,pper panel). This state of activation progressively declined by 3ours and returned to control levels after 6 hours of treatment.ikewise, AcCHO induced an increase of ERK1/2 phosphoryla-ion after 5 minutes, with a peak at 10 minutes that declined at0 minutes and then returned to control levels after 60 minutesFigure 7C, upper panel). Equal loading of protein in each laneas confirmed by probing membranes with anti-ERK1/2 (Fig-re 7C, lower panels) and anti-PKC� antibodies (Figure 7B, loweranels).

Hydrogen Peroxide Activation of c-Abl IsRequired for AcCHO Repression of PPAR�Activity via the PKC�/ERK PathwaysRecent evidence has indicated that AcCHO induces the

ctivation of type I collagen genes by an H2O2-dependent mech-nism.50 Thus, we assessed whether the inhibition of PPAR�

PPAR� function. (A and C) Serum-starved HSCs were cotransfectedand with the chloramphenicol acetyl transferase expression vectorx empty vector was also cotransfected where indicated. Twelve hours) in the presence or absence of AcCHO (85 �mol/L). The effect of PKCr vehicle for 30 minutes before the addition of AcCHO. After 24 hours

d in Materials and Methods. Data are expressed as mean � SD from 5egree of significance versus no ethanol treatment conditions. (B and D)hours, and then incubated in phosphate-free media with 0.8 mCi ofcCHO (85 �mol/L) for 15 minutes. Lysates were immunoprecipitatedwer panels represent Western blotting with PPAR� antibodies showingxperiments, each of which was performed in duplicate, are shown.

ct onk-lucr pcm

DMSObitor ocribeher dr 24with Ahy. Lo

ranscriptional activity by AcCHO was mediated by H2O2 pro-

dpba

tpitP

hiitret

y

Fpf7vPTooeAhS.ms0wnaad

FAAdwCpcphippspMpi(PtEtew

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

1244 CENI ET AL GASTROENTEROLOGY Vol. 131, No. 4

uction in HSCs. As shown in Figure 8, the H2O2 scavengerolyethylene glycol– conjugated catalase (PEG-CAT) preventedoth AcCHO-induced inhibition of PPAR� activity (Figure 8A)

igure 7. Phosphorylation of PKC� and ERK1/2 in response tocCHO and effect of PKC� and MEK dominant negative constructs oncCHO-mediated inhibition of PPAR� activity. (A) HSCs were trans-uced with adenoviral constructs at an MOI of 60 for 12 hours and thenere cotransfected with the reporter ARE-73-tk-luc and with the pSV2-AT as internal control and with pcmx-PPAR� expression vector orcmx empty vector as indicated. Twenty-four hours after transfection,ells were treated with RGZ (20 �mol/L) or its vehicle (DMSO) in theresence or absence of AcCHO (85 �mol/L). After an additional 24ours, cells were harvested for luciferase and CAT assay as described

n Materials and Methods. The data are mean � SD for 4 experimentserformed in triplicate. *P � .005 or higher degree of significance versusAd-�gal–infected cells with no AcCHO treatment. (B and C) Serum-tarved HSCs were treated with AcCHO (85 �mol/L) for the indicatederiod of time. Cell lysates were obtained as described in Materials andethods. (B) For PKC� phosphorylation assay, proteins were immuno-recipitated with an anti-PKC� antibody. The immunoprecipitates were

mmunoblotted against an anti-phosphotyrosine antibody (Anti P-Tyr).C) For ERK1/2 activation assay, proteins were separated by SDS-AGE, transferred to nitrocellulose, and then incubated with antibody

hat recognizes the active tyrosine-phosphorylated forms of ERK1 andRK2. Equal loading of protein in each lane was confirmed by probing

he membrane with anti-PKC� or anti-ERK1/2 antibodies (lower pan-ls). Representative data from 3 independent experiments, each ofhich was performed in duplicate, are shown.

nd receptor phosphorylation (Figure 8B). The role of H2O2 in d

he impairment of PPAR� function was further confirmed byarallel experiments in which the addition of exogenous H2O2

nduced PPAR� phosphorylation and inhibited its transcrip-ional activity. Both of these effects were blocked by antioxidantEG-CAT.

The production of hydrogen peroxide by AcCHO in cultureduman HSCs is shown in Figure 9. We observed a rapid increase

n intracellular oxidant production starting at 5 minutes, peak-ng between 10 and 20 minutes, and slowly decreasing thereaf-er. Such an increase in intracellular oxidants, revealed by theedox-sensitive fluorescent dye 2=,7=-dichlorofluorescein diac-tate, is most likely due to H2O2, to which this probe is selec-ively sensitive.51

It has been shown that H2O2 may induce tyrosine phosphor-lation of PKC� by activation of c-Abl.52,53 We then tested the

igure 8. H2O2 is the mediator of AcCHO effect on PPAR� phos-horylation and activity. (A) Serum-starved human HSCs were cotrans-

ected using the calcium phosphate procedure with the reporter ARE-3-tk-luc and with the chloramphenicol acetyl transferase expressionector (pSV2-CAT) as internal control for transfection efficiency. pcmx-PAR� or pcmx empty vector was also cotransfected where indicated.welve hours after transfection, cells were treated with RGZ (20 �mol/L)r its vehicle (DMSO) in the presence or absence of AcCHO (85 �mol/L)r H2O2 (25 �mol/L). The effect of the H2O2 scavenger, PEG-CAT, wasvaluated by preincubating cells with this enzyme for 30 minutes beforecCHO or H2O2 treatments. After 24 hours of incubation, cells werearvested for luciferase and CAT assay. Data are expressed as mean �D from 5 experiments, each of which was performed in duplicate. *P �

01 or higher degree of significance versus no AcCHO or H2O2 treat-ent. (B) Human HSCs were transfected with pcmx-PPAR�, serum

tarved for 24 hours, and then incubated in phosphate-free media with.8 mCi of [32P]orthophosphate for 3 hours. Subsequently, the cellsere treated with AcCHO or H2O2 for 15 minutes. Lysates were immu-oprecipitated with PPAR� antibodies and subjected to SDS-PAGEnd autoradiography. Lower panels represent Western blot with PPAR�ntibodies showing PPAR� levels in transfected cells. Representativeata from 3 independent experiments, each of which was performed in

iapiceAdnrmt

maom1g1E(EPi

oargaipP(ptg

PunuAeSdvaop

Aa

Ftua(ohvtcdcswpSmwwhR

FSiwMc

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

October 2006 EFFECTS OF ACETALDEHYDE ON PPAR� ACTIVITY 1245

nvolvement of c-Abl in AcCHO inhibition of PPAR� activitynd phosphorylation. The c-Abl inhibitor, imatinib mesylate,revented inhibition of PPAR� activity and phosphorylation

nduced by AcCHO (Figure 10A and B). To confirm the role of-Abl in the pathway controlling PPAR� function, we alsovaluated the effect of AcCHO on the activity of the luciferaseRE-7 reporter in HSCs transduced with wild-type or kinase-ead c-Abl. As shown in Figure 10C, expression of c-AblKD, butot of c-Ablwt, prevented AcCHO inhibition of PPAR�-inducedeporter activity. In agreement with these results, AcCHO treat-

ent was not able to induce PPAR� phosphorylation in c-AblKD

ransfected HSCs (Figure 10D).The antagonizing effects of c-Abl inhibition on AcCHO-

ediated PPAR� phosphorylation led us to explore the kinasectivity of c-Abl following AcCHO treatment in HSCs. Additionf AcCHO resulted in enhanced c-Abl kinase activity within 10inutes, which returned to basal levels after 1 hour (Figure

1A). Pretreatment with imatinib mesylate and PEG-CAT abro-ated c-Abl activity induced by either AcCHO or H2O2 (Figure1B and C). In parallel, both imatinib and PEG-CAT inhibitedRK activation as well as tyrosine phosphorylation of PKC�

Figure 12A and B). Furthermore, rottlerin treatment inhibitedRK activation by AcCHO (Figure 12A, lane 7), confirming thatKC� and ERK are in the linear pathway responding to AcCHO

n HSCs.In view of these results, we evaluated the effect of inhibition

f c-Abl/PKC� signaling pathways on collagen expression inctivated HSCs. As shown in Figure 13, both imatinib andottlerin, as well as PEG-CAT, inhibited AcCHO-induced colla-en type I synthesis and gene expression (Figure 13A and B). Inddition, as previously shown,49 pretreatment with the MEKnhibitor PD98059 also inhibited AcCHO-induced collagen ex-ression and ERK activation but had no effect on c-Abl andKC� activation (not shown). Furthermore, kinase-dead c-Abl

pAd-c-AblKD) expression in adenoviral transduced HSCs com-letely abrogated the fibrogenic effect of AcCHO, whereas wild-ype c-Abl (pAd-c-Ablwt) did not modify the induction of colla-

igure 9. AcCHO induces generation of H2O2 in cultured HSCs.erum-starved HSCs were treated with 85 �mol/L of AcCHO for the

ndicated time points. Hydrogen peroxide production was evaluatedith 2=,7=-dichlorofluorescein diacetate as described in Materials andethods. The data are mean � SD for 4 experiments performed in tripli-

ate. *P � .01 or higher degree of significance versus control (time 0).

en synthesis by AcCHO (Figure 13C and D). w

In addition, to support the involvement of c-Abl kinase inPAR� phosphorylation and collagen synthesis by AcCHO, wesed an adenovirus-delivered siRNA to silence c-Abl. The ade-ovirus packaging of hairpin siRNA targeted against c-Abl wassed to infect cultured HSCs. As shown in Figure 14A, indH1-cAbl but non-AdH1-empty transduced HSCs, c-Abl wasfficiently silenced both at 24 and 48 hours postinfection.ilencing of c-Abl had no effect on ligand-dependent and -in-ependent PPAR� transcriptional activity but completely pre-ented the effect of AcCHO on PPAR� transcriptional activitynd phosphorylation (Figure 14B and C). In parallel, depletionf c-Abl abrogated the profibrogenic effect of AcCHO both atrotein and messenger RNA levels (Figure 14D and E).

Collectively, these results indicate that c-Abl activation bycCHO is required to induce collagen gene expression viactivation of the PKC�/ERK pathway.

igure 10. c-Abl activation is involved in AcCHO-mediated inhibi-ion of PPAR� function. (A) Serum-starved HSCs were cotransfectedsing the calcium phosphate procedure with the reporter ARE-73-tk-lucnd with the chloramphenicol acetyl transferase expression vectorpSV2-CAT) as internal control for transfection efficiency. pcmx-PPAR�r pcmx empty vector was also cotransfected as indicated. Twelveours after transfection, cells were treated with RGZ (20 �mol/L) orehicle in the presence or absence of AcCHO (85 �mol/L). The effect ofhe c-Abl inhibitor, imatinib mesylate, was evaluated by preincubatingells for 30 minutes before addition of AcCHO. (C) Cells were trans-uced with adenoviral constructs at an MOI of 60 for 12 hours and thenotransfected as described previously. *P � .03 or higher degree ofignificance versus no AcCHO treatment. (B) HSCs were transfectedith pcmx-PPAR�, serum starved for 24 hours, and then incubated inhosphate-free media with 0.8 mCi of [32P]orthophosphate for 3 hours.ubsequently, the cells were treated with AcCHO (85 �mol/L) for 15inutes. The effect of imatinib was evaluated by preincubating cellsith the drug for 30 minutes before the addition of AcCHO. (D) Cellsere transduced with adenoviral constructs at an MOI of 60 for 24ours, and then in vivo labeling was performed as described previously.epresentative data from 3 independent experiments, each of which

FHH(Meds

F�Papibw

FgAitaMD*mgMp3aMtlivAC

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

1246 CENI ET AL GASTROENTEROLOGY Vol. 131, No. 4

igure 11. Both AcCHO and H2O2 induced c-Abl activation inSCs. Serum-starved HSCs were treated with AcCHO (85 �mol/L) or2O2 (25 �mol/L) for 15 minutes (B, C) or for the indicated period of time

A). c-Abl kinase activity was determined as described in Materials andethods (upper panels). Total c-Abl protein was determined by West-

rn blot analysis (lower panels). Representative data from 3 indepen-ent experiments, each of which was performed in duplicate, are

cpendent experiments performed in duplicate are shown.

igure 12. c-Abl activation by AcCHO is necessary to activate the PKC�/ERK1/2 pathway. Serum-starved HSCs were treated with AcCHO (85mol/L) for 15 minutes. Cell lysates were obtained as described in Materials and Methods. The effect of kinase inhibitors and the antioxidantEG-CAT were evaluated by preincubating cells with each inhibitor or vehicle for 30 minutes before the addition of AcCHO. (A) For ERK1/2 activationssay, proteins were separated by SDS-PAGE, transferred to nitrocellulose, and then incubated with antibody that recognizes the active tyrosine-hosphorylated forms of ERK1 and ERK2. (B) For PKC� phosphorylation assay, proteins were immunoprecipitated with an anti-PKC� antibody. The

mmunoprecipitates were immunoblotted against an anti-phosphotyrosine antibody (Anti P-Tyr). Equal loading of protein in each lane was confirmedy probing the membrane with anti-PKC�, anti-ERK1/2 antibodies (lower panels). Representative data from 3 independent experiments, each of

igure 13. c-Abl activation is involved in AcCHO-induced collagenene expression. (A) Serum-starved HSCs were treated with 85 �mol/LcCHO for 24 hours, and the effect of kinase inhibitors and the antiox-

dant PEG-CAT was evaluated by preincubating cells 30 minutes beforehe addition of AcCHO. Procollagen type I was determined in culture byn enzyme-linked immunoassay method as described in Materials andethods. Data, expressed as micrograms per microgram of cellularNA, are mean values � SD of 5 experiments performed in triplicate.

P � .01 versus AcCHO-treated control cells. (B) After 6 hours of treat-ent with AcCHO, total RNA was extracted from HSCs. Ten micro-rams of total RNA was used for Northern blot analysis as described inaterials and Methods. The size of messenger RNA transcripts (ex-ressed in kilobases) is indicated at the right. Representative data fromindependent experiments, each of which was performed in duplicate,re shown. (C) Cells were transduced with adenoviral constructs at anOI of 60 for 12 hours. After 24 hours in SFIF medium, cells were

reated with 85 �mol/L AcCHO for 24 hours and subsequently procol-agen type I was determined in culture media by an enzyme-linkedmmunoassay as described in Materials and Methods. Data are meanalues � SD of 5 experiments performed in triplicate. *P � .02 versuscCHO-treated control cells. (D) After 6 hours of treatment with Ac-HO, total RNA was extracted from HSCs infected with adenoviralonstructs for Northern blot analysis. Representative data from 3 inde-

srd

skSwlPAt

FgMFabEbEipsWcwmfmcp

FPwoTTAspc((l.Vs0wiPeA8wdeAAvi

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

October 2006 EFFECTS OF ACETALDEHYDE ON PPAR� ACTIVITY 1247

PPAR� Phosphorylation Is Involved in thec-Abl Profibrogenic Pathway in HSCsTo confirm that PPAR� is a downstream target of c-Abl

ignaling, we examined whether the PPAR� phosphorylation-esistant mutant interfered with collagen gene expression in-

igure 14. Silencing of c-Abl abrogates the effect of AcCHO onPAR� function and collagen synthesis. Human HSCs were incubatedith recombinant virus (AdH1-c-Abl or AdH1-empty vector) at an MOIf 60 in 1 mL of Iscove’s medium containing 2% fetal bovine serum. (A)he silencing of c-Abl was monitored by Western blot analysis. (B)welve hours after infection, cells were transfected with the reporterRE-73-tk-luc and with the chloramphenicol acetyl transferase expres-ion vector (pSV2-CAT) as internal control for transfection efficiency.cmx-PPAR� or pcmx empty vector was also cotransfected as indi-ated. Twelve hours after transfection, cells were treated with RGZ20 �mol/L) or its vehicle (DMSO) in the presence or absence of AcCHO85 �mol/L). After 24 hours of incubation, cells were harvested foruciferase and CAT assay as described in Materials and Methods. *P �05 or higher degree of significance versus no AcCHO treatment. (C)irally transduced HSCs were transfected with pcmx-PPAR�, serumtarved for 24 hours, and then incubated in phosphate-free media with.8 mCi of [32P]orthophosphate for 3 hours. Subsequently, the cellsere treated with AcCHO (85 �mol/L) for 15 minutes. Lysates were

mmunoprecipitated with PPAR� antibodies and subjected to SDS-AGE and autoradiography. Representative data from 3 independentxperiments, each of which was performed in duplicate, are shown. (D)fter 24 hours in SFIF medium, viral transduced HSCs were treated with5 �mol/L AcCHO for 24 hours and subsequently procollagen type Ias determined in culture media by an enzyme-linked immunoassay asescribed in Materials and Methods. Data are mean values � SD of 5xperiments performed in triplicate. *P � .01 versus AcCHO-treateddH1-empty transduced HSCs. (E) After 6 hours of treatment withcCHO, total RNA was extracted from HSCs transduced with adeno-iral constructs for Northern blot analysis. Representative data from 3ndependent experiments performed in duplicate are shown.

uced by c-Abl. To address this question, collagen gene expres- c

ion was measured in activated HSCs cotransduced with theinase-active c-Abl (pAd-c-AblKA) and with the wild-type orer84 ¡ Ala mutated PPAR� viral constructs. Transductionith pAd-c-AblKA increased c-Abl activity and ERK phosphory-

ation in HSCs cotransduced with either pAd-PPAR�(wt) or pAd-PAR(Ser84Ala) (Figure 15A and B), whereas expression of pAd-c-blKA induced a significant PPAR� phosphorylation in HSCs

ransduced with pAd-PPAR�(wt) but had a negligible effect in

igure 15. PPAR� is a downstream target of c-Abl–mediated fibro-enic signaling. HSCs were transduced with adenoviral constructs at anOI of 60 for 12 hours as described in Materials and Methods. (A and

) For c-Abl activity, cells were immunoprecipitated with antibodiesgainst c-Abl and immune complexes were incubated with kinaseuffer containing the specific c-Abl substrate GST-Crk. (B and G)RK1/2 activation was detected in total protein extracts using an anti-ody against the active tyrosine-phosphorylated forms of ERK1 andRK2. (C and H) For detection of phosphorylated PPAR�, cells were

ncubated in phosphate-free media with 0.8 mCi of [32P]orthophos-hate. Lysates were immunoprecipitated with PPAR� antibodies andubjected to SDS-PAGE and autoradiography. Lower panels representestern blot with PPAR� antibodies showing PPAR� levels in infected

ells. (D and I) For Northern blot analysis of collagen type I, total RNAas extracted from infected HSCs after 24 hours of culture in SFIFedium. Representative data from 3 independent experiments per-

ormed in duplicate are shown. (E and J) Procollagen type I was deter-ined in culture media of adenoviral-infected HSCs after 24 hours of

ulture in SFIF medium. Data are mean values � SD of 5 experimentserformed in triplicate. *P � .01 versus pAd-�gal–infected cells. Open

Hitcemap(sv

oaasea(pHcaoup

rpfpauniddatn8

Hmp

tAipcmsmPnrPo

cm

htPDhfAacwrtgi

Fcr(oa3ymtpaaappf

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

1248 CENI ET AL GASTROENTEROLOGY Vol. 131, No. 4

SCs infected with pAd-PPAR(Ser84Ala) (Figure 15C). Accord-ngly, transduction with kinase-active c-Abl strongly inducedhe steady-state messenger RNA level and protein synthesis ofollagen type I in HSCs expressing wild-type PPAR�, whereas noffect was documented in cells expressing the Ser84 ¡ Alautant (Figure 15D and E). Furthermore, the effect of c-Abl

ctivation on PPAR� phosphorylation and collagen type I ex-ression was completely abrogated by rottlerin and PD98059

Figure 15F–J), suggesting that the increased collagen expres-ion related to PPAR� phosphorylation by c-Abl requires acti-ation of the PKC�/ERK pathway.

To show that PPAR� phosphorylation and c-Abl activationccur in vivo during liver injury, we isolated HSCs from ratsfter a single intraperitoneal injection of CCl4 in the presence orbsence of a concomitant oral administration of imatinib me-ylate. As previously shown,54 collagen �1(I) messenger RNAxpression, analyzed by Northern blot, increased between 48nd 72 hours after a single intraperitoneal injection of CCl4

Figure 16A). As shown in Figure 16C (upper panel), serinehosphorylation in PPAR� immunoprecipitates was absent inSCs isolated from control animals, whereas it was signifi-

antly increased at 48 and 72 hours after CCl4 treatment. Oraldministration of imatinib abrogated both the marked increasef collagen gene expression induced by treatment of CCl4 (Fig-re 16B, upper panel) and in parallel prevented PPAR� phos-horylation (Figure 16D, upper panel).

Because c-Abl becomes autophosphorylated on multiple ty-osine residues on activation,55 we tested c-Abl tyrosine phos-horylation in anti–c-Abl immunoprecipitates in HSCs isolatedrom CCl4-treated rats. We showed that c-Abl was stronglyhosphorylated after CCl4 injection (Figure 16C, upper panel)nd that imatinib treatment abrogated this effect (Figure 16F,pper panel). On the contrary, administration of imatinib didot prevent the increase of �–smooth muscle actin expression

n HSCs isolated from CCl4-treated animals (not shown) andid not protect from CCl4-induced hepatocellular damage, asocumented by the unmodified levels of transaminases (CCl4:lanine aminotransferase, 517.93 � 31.3 U/L; aspartate amino-ransferase, 957.71 � 51.7 U/L; CCl4 � imatinib: alanine ami-otransferase, 653.69 � 52.1 U/L; aspartate aminotransferase,31.74 � 59.5 U/L).

Collectively, these results suggest that c-Abl is activated inSCs during liver injury and that its inhibition by imatinibay be responsible, at least in part, for the reduction of PPAR�

hosphorylation and collagen gene expression.

DiscussionThe mechanisms whereby long-term alcohol consump-

ion results in hepatic fibrogenesis are not entirely understood.ccumulating evidence from in vivo and in vitro studies has

ndicated that PPAR� has a primary role in maintaining restinghenotype in HSCs, and its transcriptional activation by spe-ific ligands can inhibit collagen gene expression in activatedyofibroblast-like cells.18 –21 In the present study, we have ob-

erved that the profibrogenic stimulus induced by AcCHO, theain product of ethanol oxidation, involves the inhibition of

PAR� transcriptional activity in HSCs. The presence of etha-ol reduced the ability of PPAR�-specific agonists to activate aeporter construct in the presence or absence of transfectedPAR�. This effect was mediated by AcCHO, because inhibition

ompletely, while the aldehyde dehydrogenase inhibitor cyana-ide enhanced the effect of ethanol.A similar transcriptional repression was previously shown in

epatocytes where AcCHO impaired PPAR� transcriptional ac-ivity and DNA binding.22 On the contrary, dysfunction ofPAR� function in HSCs was not related to inhibition of itsNA binding, suggesting that ethanol metabolism impairedepatic PPAR function with different mechanisms and in iso-

orm- and cell-specific manners. In fact, we found that in HSCs,cCHO exerted its action to counteract PPAR� transcriptionalctivity at posttranslational levels by phosphorylation of theonsensus MAP kinase site at serine 84 of the human PPAR�1,hich is equivalent to position 112 of mouse PPAR�2. Previous

eports indicated that phosphorylation of this site reducedranscriptional activity of PPAR� and, in parallel, blocked li-and-dependent adipogenic activity in preadipocytes.47,56 Themplications of this posttranslational modification of PPAR� in

igure 16. Both PPAR� phosphorylation and c-Abl activation oc-ur in vivo during CCl4-induced liver injury. HSCs were obtained fromats at the indicated time points after an intraperitoneal injection of CCl42 mL/kg) in the presence or absence of concomitant oral administrationf imatinib mesylate (50 mg · kg�1 · day�1) as described in Materialsnd Methods. (A and B) For Northern blot analysis of collagen type I and6B4, 20 �g of total RNA was used. (C and D) For PPAR� phosphor-lation assay, proteins extracted from freshly isolated HSCs were im-unoprecipitated with anti-PPAR� antibodies. The immunoprecipi-

ates were immunoblotted with anti-phosphoserine (Anti P-Ser) (upperanels) or anti-PPAR� (lower panels) antibodies. (E and F) For c-Ablctivation assay, proteins were immunoprecipitated with an anti–c-Ablntibody and then the immunoprecipitates were immunoblotted withnti-phosphotyrosine (Anti P-Tyr) (upper panels) or anti–c-Abl (loweranels) antibodies. These results are representative of 3 different ex-eriments performed in triplicate in different cell preparations obtained

rom different animals.

hese cells are significant. Various growth factors and cytokines,

sotpfPt

ptioaspt

rcatc

owlidTtcgce

qptium

imvatmsflICwp

rawHJ

ta

ttpCPaystlip

avsTtPicdbscStH

tsAHatsCaPtuetr

oankcrgkhssf

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

October 2006 EFFECTS OF ACETALDEHYDE ON PPAR� ACTIVITY 1249

uch as platelet-derived growth factor, epidermal growth factor,r tumor necrosis factor �, could affect transcriptional activa-ion of numerous genes involved in lipid metabolism via PPAR�

hosphorylation, thus contributing to the antiadipogenic ef-ects of these agents.57,58 In addition, the genetic prevention ofPAR� phosphorylation preserves insulin sensitivity in the set-ing of diet-induced obesity in mice.59

In human-activated HSCs, inhibitors of MAP kinase signalathways prevented the ability of platelet-derived growth factoro down-modulate PPAR� transcriptional activity.18 Moreover,n rat HSCs, Sung et al have recently shown that the expressionf the Ser82 ¡ Ala PPAR� mutant abolished the inhibitoryction of tumor necrosis factor � on PPAR� transactivation,60

uggesting the importance of PPAR� phosphorylation in therofibrogenic and proinflammatory responses mediated byhese factors in HSCs.

Here we show that the Ser84 ¡ Ala PPAR� mutant wasesistant to AcCHO-mediated repression of PPAR� activity andompletely prevented the increase of collagen type I synthesisnd gene expression induced by AcCHO in HSCs, indicatinghat this mutant may function as a dominant repressor ofollagen type I transcription.

The antifibrotic effect of PPAR� activation has been previ-usly shown in human HSCs and other myofibroblast-like cells,here specific PPAR� ligands, such as antidiabetic thiazo-

idinediones, inhibit the increase of collagen gene expressionnduced by transforming growth factor (TGF)-�1 in a receptor-ependent manner.21,61,62 Interestingly, Han et al showed thatGF-�1, similar to other growth factors, induced phosphoryla-

ion of PPAR� via a MAP kinase signaling pathway and de-reased PPAR� transcriptional activity.63 These data, taken to-ether with our results, suggest that phosphorylation of PPAR�

ould represent a common pathway by which different factorsxert their profibrogenic activity.

Interestingly, AcCHO did not induce collagen synthesis inuiescent HSCs41,43,44 and was not able to modulate PPAR�

hosphorylation in these cells. The molecular events involved inhe unresponsiveness of quiescent HSCs to fibrogenic stimuli64

ncluding AcCHO remain speculative, but the inability to mod-late the posttranslational status of PPAR� might be one of theechanisms involved.Our results also establish a primary role for ERK1/2 kinase

n AcCHO-mediated regulation of PPAR� activity. In fact, treat-ent with MEK inhibitors as well as transduction with adeno-

irus carrying an MEK dominant negative construct completelybrogated both PPAR� phosphorylation and inhibition of itsranscriptional activity. The ERK1/2 pathways are central to

any signal transduction processes and constitute the majorystem through which growth factor receptors transduce proin-ammatory and profibrogenic signals to the nucleus in HSCs.65

n agreement with Svegliati-Baroni et al,49 we found that Ac-HO induced ERK1/2 activity in HSCs, and this was correlatedith the up-regulation of collagen gene expression, at least inart, via inactivation of PPAR� transcriptional activity.

Recently, it has been shown that AcCHO activates JNK1/2 inat HSCs.66 This member of stress-activated protein kinases isble to phosphorylate the MAP kinase site of PPAR�,46,60 and itas suggested to regulate the synthesis of collagen type I in ratSCs treated with AcCHO.10 However, our data show that

NK1/2 is unlikely to be involved in AcCHO-mediated inhibi- m

ion of PPAR� because the specific inhibitor, SP600125, did notffect AcCHO regulation of PPAR� function.

Certain insights have arisen from the finding that AcCHOreatment in HSCs is associated with activation of PKC8,9,67 andhat PKC is linked to ERK1/2 activation in response to somerofibrogenic stimuli, including those mediated by Ac-HO.49,68 –70 Here, we found that the selective inhibition ofKC� by rottlerin or by expression of PKC� dominant negativebrogated AcCHO inhibition of PPAR� activity and phosphor-lation and impaired ERK1/2 activation as well as collagenynthesis by AcCHO. Moreover, AcCHO was able to induceyrosine phosphorylation of PKC� with kinetics closely corre-ated to the time-dependent activation of ERK1/2. These datandicate that PKC� acts upstream of PPAR� phosphorylation byarticipating in phosphorylation of ERK1/2.

Studies in other cell types such as dermal fibroblasts, mes-ngial cells, and cardiac myofibroblasts have shown PKC� in-olvement in TGF-� and endothelin 1–mediated collagen bio-ynthesis.71–73 Runyan et al have shown that the blockade ofGF-�1–induced PKC� activation inhibits Smad3 transcrip-

ional activity.72 This is a very intriguing finding, becausePAR� inhibits Smad transcriptional activity via a physical

nteraction with Smad3 in human smooth muscle cells.74 Re-ent results have shown that Smad transcriptional activity in-uced by TGF-�1 in dermal fibroblasts is significantly inhibitedy expression of Ser84 ¡ Ala PPAR� mutant (Ceni et al, manu-cript in preparation), suggesting that PPAR� phosphorylationould modulate profibrogenic signals via interaction with themad pathway. This possibility, although feasible, needs to beested and is currently being investigated in AcCHO-treatedSCs.Several lines of evidence support the hypothesis that oxida-

ive stress plays a key role in the development of hepatic fibro-is.75,76 Greenwel et al showed a direct connection betweencCHO-induced fibrogenesis and oxidative stress. In mouseSCs, AcCHO induced excess formation of H2O2, and catalase,

n H2O2-enzyme scavenger, prevented AcCHO-elicited collagenype I gene up-regulation.50 In agreement, in our experimentaletting, AcCHO induced a rapid production of H2O2 and PEG-AT prevented the inhibition of PPAR� transcriptional activitynd receptor phosphorylation induced by AcCHO. Moreover,EG-CAT completely inhibited both AcCHO-induced activa-ion of the PKC�/ERK pathway and AcCHO-mediated up-reg-lation of collagen gene expression. The treatment with exog-nous H2O2 induced PPAR� phosphorylation and inhibited itsranscriptional activity in human HSCs, thereby confirming theole of oxidative stress in the mechanism of PPAR� inactivation.

Previous reports have shown that PKC� is phosphorylatedn tyrosine and activated in cells treated with H2O2 by a mech-nism dependent on activation of c-Abl protein tyrosine ki-ase.52,53,77,78 c-Abl, which belongs to the nonreceptor tyrosineinase family, distinct from the Src family, is implicated inellular oxidative stress responses and is involved in cell-cycleegulation and integrin signaling.79 Recently, it has been sug-ested that c-Abl may have a potential role in regulating cyto-ine-mediated fibrosis. In both renal and lung fibroblasts, in-ibition of c-Abl by imatinib mesylate blocks the TGF-�1ignaling pathway and prevents bleomycin-induced lung fibro-is and renal fibrogenesis in obstructive nephropathy.80,81 Weound that AcCHO induced c-Abl activity via an H2O2-mediated

echanism and that inhibition of c-Abl, by imatinib mesylate

acpiatocc

bglt

gwAsmtpim

Pitmsitii

epviviabstTtfm

1

1

1

1

1

1

1

1

1

1

2

2

2

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

1250 CENI ET AL GASTROENTEROLOGY Vol. 131, No. 4

s well as by silencing of c-Abl gene or expression of kinase-dead-Abl, completely abrogated the AcCHO effect on PPAR� and inarallel prevented collagen type I up-regulation. Imatinib also

nhibited AcCHO-induced tyrosine phosphorylation of PKC�nd ERK1/2 activation, showing that in response to AcCHO,he PKC�/ERK1/2 pathway is activated primarily downstreamf c-Abl. The further observation that PEG-CAT prevented-Abl activation suggests that H2O2 is the main mediator of-Abl activation by AcCHO.

Consistent with the evidence that TGF-� modulates profi-rogenic signals by c-Abl activation80,81 and induces procolla-en type I expression by an H2O2-mediated mechanism,82 it isikely that multiple fibrogenic pathways function through ac-ivation of c-Abl.

The involvement of PPAR� phosphorylation in the profibro-enic signal mediated by c-Abl activation was shown in HSCshere coexpression with the phosphorylation-resistant Ser84 ¡

la PPAR� mutant completely repressed collagen type I synthe-is induced by expression of the kinase-active c-Abl. Further-

ore, the inhibition of the cAbl-induced collagen type I syn-hesis and gene expression by rottlerin as well as by PD98059rovides convincing proof that PKC� and ERK1/2 are located

n a linear downstream pathway in the profibrogenic signalsediated by c-Abl activation.Finally, we have shown that both c-Abl activation and

PAR� phosphorylation occur in rats during acute liver injurynduced by intraperitoneal injection of CCl4 and that bothhese events are abrogated in animals treated with imatinib

esylate in parallel to the prevention of collagen gene expres-ion. Although we cannot exclude that imatinib exerts its actionn vivo via inhibition of tyrosine kinases other than c-Abl,83

hese data suggest that activation of c-Abl and the cascadencluding PPAR� phosphorylation might be a pathway involvedn the profibrogenic response to different hepatic injury.

In conclusion, we have shown that the induction of collagenxpression by AcCHO in stellate cells is dependent, at least inart, on PPAR� phosphorylation and that H2O2-mediated acti-ation of the cAbl/PKC�/ERK1/2 signaling pathway is involvedn this fibrogenic process. Collectively, these observations pro-ide insight into the molecular mechanisms of the alcohol-nduced liver fibrosis and emphasize relevant pathophysiologicnd clinical implications. In recent years, activation of PPAR�y specific ligands such as thiazolidinedione molecules has beenuggested for treatment of chronic hepatic diseases, includinghose induced by long-term alcohol consumption.21,84 – 86

herefore, knowledge of the molecular mechanisms of PPAR�ranscriptional inactivation during chronic tissue injury mayacilitate the development of new rational strategies to design

ore effective antifibrotic therapies based on PPAR� agonists.

References

1. Olaso E, Friedman L. Molecular regulation of hepatic fibrogen-esis. J Hepatol 1998;29:836–847.

2. Lieber CS. Alcoholic liver disease: new insights in pathogenesislead to new treatments. J Hepatol 2000;32(suppl):113–128.

3. Holstage A, Bedossa P, Poynard T, Kollinger M, Chaput JC, Hou-glum K, Chojkier M. AcCHO-modified epitopes in liver biopsyspecimens of alcoholic patients: localization and association withprogression of liver fibrosis. Hepatology 1994;19:367–374.

4. Niemelä O, Parkkila S, Ylä-Herttuala S, Villanueva J, Ruebner B,

tion and fibrogenesis in micropig model of alcohol-induced liverdisease. Hepatology 1995;22:1208–1214.

5. Casini A, Cunningham M, Rojkind M, Lieber CS. Acetaldehydeincreases procollagen type I and fibronectin gene expression incultured rat fat-storing cells through a protein synthesis-depen-dent mechanism. Hepatology 1991;13:758–765.

6. Friedman SL. The cellular basis of hepatic fibrosis. N Engl J Med1993;328:1828–1835.

7. Eng FJ, Friedman SL. New insights into hepatic stellate cellsactivation: the simple becomes complex. Am J Physiol 2000;279:G7–G11.

8. Casini A, Galli G, Salzano R, Ceni E, Franceschelli F, Rotella CM,Surrenti C. Acetaldehyde induces c-fos and c-junproto oncogenesin fat-storing cell cultures through protein kinase C activation.Alcohol Alcohol 1994;29:303–304.

9. Anania FA, Womack L, Potter JJ, Mezey E. Acetaldehyde en-hances murine �2(I) collagen promoter activity by Ca2�-indepen-dent protein kinase C activation in cultured rat hepatic stellatecells. Alcohol Clin Exp Res 1999;23:279–284.

0. Chen A, Davis BH. The DNA binding protein BTEB mediatesacetaldehyde-induced, jun N-terminal kinase-dependent �I(I) col-lagen gene expression in rat hepatic stellate cells. Mol Cell Biol2000;20:2818–2826.

1. Anania F, Potter JJ, Rennie-Tankersley L, Mezey E. Effects ofacetaldehyde on nuclear protein binding to the nuclear factor Iconsensus sequence in the �2(I) collagen promoter. Hepatology1995;21:1640–1648.

2. Attard FA, Wang L, Potter JJ, Rennie-Tankersley L, Mezey E.CCAAT/enhancer binding protein � mediates the activation of themurine �1(I) collagen promoter by acetaldehyde. Arch BiochemBiophys 2000;378:57–64.

3. Kintscher U, Law RE. PPAR-mediated insulin sensitization: theimportance of fat versus muscle. Am J Physiol Endocrinol Metab2005;288:E287–E291.

4. Kliewer SA, Umesono K, Noona DJ, Heyman RA, Evans RM.Convergence of 9-cis retinoic acid and peroxisome proliferatorsignaling pathway through heterodimer formation of their recep-tors. Nature 1992;358:771–774.

5. Debril M-B, Renaud J-P. Fajas L, Auwerx J. The pleiotropic func-tions pf peroxisome proliferator-activated receptor . J Mol Med2001;79:30–47.

6. Kernsten S, Desvengre B, Wahli W. Roles of PPARs in health anddisease. Nature 2000;405:421–424.

7. Delerive P, Fruchart JC, Staels B. Peroxisome proliferator-acti-vated receptors in inflammation control. J Endocrinol 2001;169:453–459.

8. Galli A, Crabb D, Price D, Ceni E, Salzano R, Surrenti C, Casini A.Peroxisome proliferator-activated receptor transcriptional regu-lation is involved in platelet-derived growth factor-induced prolif-eration in human hepatic stellate cells. Hepatology 2000;31:101–108.

9. Marra F, Efsen E, Romanelli R, Caligiuri A, Pastacaldi S, BatignaniG, Bonacchi A, Caporale R, Laffi G, Pinzani M, Gentilini P. Ligandsof peroxisome proliferator-activated receptor modulate profibro-genic and proinflammatory actions in hepatic stellate cells. Gas-troenterology 2000;119:466–478.

0. Miyahara T, Schrum L, Rippe R, Xiong S, Yee HF, Motomura K,Nania FA, Willson TM, Tsukamoto H. Peroxisome proliferator-activated receptors and hepatic stellate cells activation. J BiolChem 2000;275:35715–35722.

1. Galli A, Crabb DW, Ceni E, Salzano R, Mello T, Svegliati-Baroni G,Ridolfi F, Trozzi L, Surrenti C, Casini A. Antidiabetic thiazo-lidinediones inhibit collagen synthesis and hepatic stellate cellactivation in vivo and in vitro. Gastroenterology 2002;122:1924–1940.

2. Galli A, Pinaire J, Fisher M, Dorris R, Crabb DW. The transcrip-

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

4

4

4

4

4

4

4

4

4

4

5

5

5

5

5

5

5

5

5

5

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

October 2006 EFFECTS OF ACETALDEHYDE ON PPAR� ACTIVITY 1251

vated receptor � is inhibited by ethanol metabolism. J Biol Chem2001;276:68–75.

3. Casini A, Pinzani M, Milani S, Grappone C, Galli G, Jezequel AM,Schuppan D, Rotella CM, Surrenti C. Regulation of extracellularmatrix synthesis by transforming growth factor �1 in humanfat-storing cells. Gastroenterology 1993;105:245–253.

4. Galli A, Ceni E, Crabb DW, Mello T, Salzano R, Grappone C, MilaniS, Surrenti E, Surrenti C, Casini A. Antidiabetic thiazolidinedionesinhibit invasiveness of pancreatic cancer cells via PPAR inde-pendent mechanisms. Gut 2004;53:1688–1697.

5. Inagaki Y, Truter S, Ramirez F. Transforming growth factor-�stimulates �2(I) collagen gene expression through a cis-actingelement that contains an Sp-1 binding site. J Biol Chem 1994;259:14828–14834.

6. de Wet JR, Wood KV, DeLuca M, Helinski DR, Subramani S. Fireflyluciferase gene: structure and expression in mammalian cells.Mol Cell Biol 1987;7:725–737.

7. Galli A, Stewart M, Dorris R, Crabb D. High level expression ofRXR� and the presence of endogenous ligands contribute toexpression of a peroxisomal proliferator receptor-responsivegene in hepatoma cells. Arch Biochem Biophys 1998;354:288–294.

8. Lowry OH, Rosebrough NJ, Farr AL, Randal RJ. Protein measure-ment with the folin phenol reagents. J Biol Chem 1951;195:265–275.

9. Andrews NC, Faller DV. A rapid micropreparation technique forextraction of DNA-binding proteins from limiting numbers of mam-malian cells. Nucleic Acids Res 1991;19:2499.

0. Tontonoz P, Hu E, Graves RA, Budavari AL, Spiegelman BM.mPPAR2: tissue-specific regulator of an adipocyte enhancer.Genes Dev 1994;8:1224–1234.

1. Chomczynski P, Sacchi E. Single-step method of RNA isolation byacid guanidinium thiocyanate-phenol-chloroform extraction. AnalBiochem 1987;162:156–159.

2. Foschi M, Chari S, Dunn MJ, Sorokin A. Biphasis activation ofp21 RAS by endothelin-1 sequentially activates the ERK cascadeand phosphatidylinositol 3-kinase. EMBO J 1997;16:6439–6451.

3. Li W, Yu JC, Shin DY, Pierce JH. Characterization of proteinkinase C- (PKC-) ATP binding mutant. J Biol Chem 1995;270:8311–8318.

4. Barilà D, Superti-Furga G. An intramolecular SH3-domain interac-tion regulates c-Abl activity. Nat Genet 1998;18:280–282.

5. Shen C, Buck AK, Liu X, Winkler M, Reske SN. Gene silencing byadenovirus-delivered siRNA. FEBS Lett 2003;539:111–114.

6. Ohba H, Zhelev Z, Bakalova R, Ewis A, Omori T, Ishikava M,Shinohara Y, Baba Y. Inhibition of bcr-abl and/or c-abl geneexpression by small interfering, double-stranded RNAs. Cancer2004;101:1390–1403.

7. Bae YS, Kang SW, Seo MS, Baines IC, Tekle E, Chock PB, RheeSG. Epidermal growth factor-induced generation of hydrogen per-oxide. Role in EGF receptor-mediated tyrosine phosphorylation.J Biol Chem 1997;272:217–221.

8. Knook DL, Sefflaar AM, DeLeeuw AM. Fat storing cells of the liver.Exp Cell Res 1982;139:468–471.

9. Casini A, Pellegrini G, Ceni E, Salzano R, Parola M, Robino G,Milani S, Dianzani MU, Surrenti C. Human hepatic stellate cellsexpress class I alcohol dehydrogenase and aldehyde dehydroge-nase but not cytochrome P4502E1. J Hepatol 1998;28:40–45.

0. Camp HS, Tarufi SR. Regulation of peroxisome proliferator-acti-vated receptor by mitogen-activated protein kinase. J Biol Chem1997;272:10811–10816.

1. Friedman SL. Acetaldehyde and alcoholic fibrogenesis: fuel to thefire, but not the spark. Hepatology 1990;12:609–612.

2. Friedman SL, Roll FJ, Boyles J, Arenson DM, Bissel MD. Mainte-

cytes by basement membrane matrix. J Biol Chem1989;264:10756–10762.

3. Maher JJ, Zia S, Tzagarakis C. Acetaldehyde-induced stimulationof collagen synthesis and gene expression is dependent oncondition of cell culture: studies with rat lypocytes and fibro-blasts. Alcohol Clin Exp Res 1994;18:403–409.

4. Sato T, Gordon RG, Murayama N, Kato R, Tyson CA. Influence ofacetaldehyde and ethanol on rat hepatic lipocyte characteristicsin primary culture. Cell Biol Int 1995;19:879–881.

5. Yavron S, Chen Li, Xiong S, Wang J, Rippe RA, Tsukamoto H.Peroxisome proliferators activated receptor suppresses proxi-mal �1(I) collagen promoter via inhibition of p300-facilitated NF-Ibinding to DNA in hepatic stellate cells. J Biol Chem 2005;280:40650–40659.

6. Camp HS, Rarufi SR, Leff T. c-jun N-terminal kinase phosphory-lates peroxisome proliferator activated receptor 1 and nega-tively regulates its transcriptional activity. Endocrinology 1999;140:392–397.

7. Hu E, Kin JB, Sarraf P, Spiegelman BM. Inhibition of adipogenesisthrough MAP kinase-mediated phosphorylation of PPAR. Sci-ence 1996;274:2100–2103.

8. Britton RS, Bacon BR. Intracellular signalling pathways in stellatecell activation. Alcohol Clin Exp Res 1999;23:922–925.

9. Svegliati-Baroni G, Ridolfi F, Di Sario A, Saccomanno S, Bendia E,Benedetti A, Greenwel P. Intracellular signaling pathways in ac-etaldehyde-induced collagen and fibronectin gene expression inhuman hepatic stellate cells. Hepatology 2001;33:1130–1140.

0. Greenwel P, Dominguez-Rosales JA, Mavi G, Rivas-Estilla AM,Rojkind M. Hydrogen peroxide: a link between acetaldehyde-elicited �1(I) collagen gene up-regulation and oxidative stress inmouse hepatic stellate cells. Hepatology 2000;31:109–116.

1. Chiarugi P, Pani G, Giannoni E, Taddei L, Calovitti R, Raugei G,Symons M, Borrello S, Galeotti T, Ramponi G. Reactive oxygenspecies as essential mediators of cell adhesion: the oxidativeinhibition of a FAK tyrosine phosphatase is required for celladhesion. J Cell Biol 2003;161:933–944.

2. Su X, Wu F, Datta R, Kharbanda S, Kufe D. Interaction betweenprotein kinase C ad c-Abl tyrosine kinase in the cellular re-sponse to oxidative stress. J Biol Chem 2000;275:7470–7473.

3. Kumar S, Bharti A, Mishra NC, Raina D, Kharbanda S, Saxena S,Kufe D. Targeting of c-Abl tyrosine kinase to mitochondria in thenecrotic cell death response to oxidative stress. J Biol Chem2001;276:17281–17285.

4. Ikejima K, Honda H, Yoshikawa M, Hirose M, Kitamura T, Takei Y,Sato N. Leptin augments inflammatory and profibrogenic re-sponses in the murine liver induced by hepatotoxic chemicals.Hepatology 2001;34:288–297.

5. Tanis KQ, Veach D, Duewel HS, Bornmann WG, Koleske AJ. Twodistinct phosphorylation have additive effects on Abl family ki-nase activation. Mol Cell Biol 2003;23:3884–3896.

6. Adams M, Reginato MJ, Shao D, Lazar MA, Chatterjee VK. Tran-scriptional activation of peroxisome proliferator-activated recep-tor is inhibited by phosphorylation at the consensus mitogen-activated protein kinase site. J Biol Chem 1997;272:5128–5132.

7. Ron D, Braiser AR, McGehee RE, Habener JF. Tumor necrosisfactor-induced reversal of adipocytic phenotype of 3T3-L1 cells ispreceded by a loss of nuclear CCAT/enhancer binding protein(C/EBP). J Clin Invest 1992;89:223–233.

8. Sorrero G, Mills D. Physiological role of epidermal growth factoron adipose tissue development in vivo. Proc Natl Acad Sci U S A1991;88:3912–3916.

9. Rangwala SM, Rhodes B, Shapiro JS, Rich AS, Kim JK, ShulmanGI, Kaestner KH, Lazar MA. Genetic modulation of PPAR phos-phorylation regulates insulin sensitivity. Dev Cell 2003;5:657–

6

6

6

6

6

6

6

6

6

6

7

7

7

7

7

7

7

7

7

7

8

8

8

8

8

8

8

tF@

eR

pFafa

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

1252 CENI ET AL GASTROENTEROLOGY Vol. 131, No. 4

0. Sung CK, She H, Xiong S, Tsukamoto H. Tumor necrosis factor-�inhibits peroxisome proliferator-activated receptor at a post-translational level in hepatic stellate cells. Am J Physiol Gas-trointest Liver Physiol 2004;286:G722–G729.

1. Burgess HA, Daugherty LE, Thatcher TH, Lakatos HF, Ray DM,Redonnet M, Phipps RP, Sime PJ. PPAR agonists inhibit TGF-�induced pulmonary myofibroblast differentiation and collagen pro-duction: implication for therapy of lung fibrosis. Am J Physiol LungCell Mol Physiol 2005;288:L1146–1153.

2. Zafiriou S, Stanners SR, Saad S, Polhill TS, Poronnik P, PollockCA. Pioglitazone inhibits cell growth and reduced matrix produc-tion in human kidney fibroblast. J Am Soc Nephrol 2005;16:638–645.

3. Han J, Hajjiar DP, Tauras JM, Feng J, Gotto AM, Nicholson AC.Transforming growth factor-�1 (TGF-�1) and TGF-�2 decreaseexpression of CD36, the type B scavenger receptor, throughmitogen-activated protein kinase phosphorylation of peroxisomeproliferator-activated receptor-. J Biol Chem 1999;2:1241–1246.

4. Hellerbrand C, Jobin C, Licato LL, Sarbor RB, Brenner DA. Cyto-kines induce NF-�B in activated but not in quiescent rat hepaticstellate cells. Am J Physiol 1998;275:G269–G278.

5. Pinzani M, Marra F. Cytokine receptors and signalling in hepaticstellate cells. Semin Liver Dis 2001;21:397–416.

6. Anania FA, Womack L, Jiang M, Sexena NK. Aldehydes potentiate�2(I) collagen gene activity by JNK in hepatic stellate cells. FreeRadic Biol Med 2001;30:846–857.

7. Khabanda KK, Shubert KA, Wyatt TA, Sorrell MF, Tuma DJ. Effectof malondialdehyde-acetaldehyde protein adducts on the proteinkinase C secretion of urokinase-type plasminogen activator inhepatic stellate cells. Biochem Pharmacol 2002;63:553–556.

8. Tourkina E, Gooz P, Pannu J, Bonner M, Scholz D, Hacker S,Silver RM, Trojanowska M, Hoffman S. Opposing effects of pro-tein kinase � and protein kinase � on collagen expression byhuman lung fibroblasts are mediated via MEK/ERK and caveolinsignalling. J Biol Chem 2005;280:13879–13887.

9. Ueda Y, Hirai S, Osada S, Suzuki A, Mizuno K, Phno S. Proteinkinase C activates MEK-ERK pathway in a manner independent ofRas and dependent of Raf. J Biol Chem 1996, 271:23512–23519.

0. Takahashi M, Matsui A, Inao M, Mochida S, Fujiwara K. ERK/MAP-dependent PI3K/Akt phosphorylation though VEGFR-1 afterVEGF stimulation in activated hepatic stellate cells. Hepatol Res2003;26:232–236.

1. Jimenez SA, Gaidarova S, Saitta B, Sandorfi N, Herrich DJ, Rosen-bloom JC, Kucich U, Abrams WR, Rosenbloom J. Role of proteinkinase C-delta in the regulation of collagen gene expression inscleroderma fibroblasts. J Clin Invest 2001;108:1395–1403.

2. Runyan CE, Schnaper HW, Poncelet AC. Smad3 and PKC medi-ate TGF-beta1-induced collagen I expression in human mesangialcells. Am J Physiol Renal Physiol 2003;285:F413–F422.

3. Chintalgattu V, Katwa LC. Role of protein kinase C in endothelin-induced type I collagen expression in cardiac myofibroblast iso-lated from the site of myocardial infarction. J Pharmacol Exp Ther2004;311:691–699.

4. Fu M, Zhang J, Zhu X, Myles DE, Wilson TM, Liu X, Chen YE.Peroxisome proliferator-activated receptor gamma inhibits trans-forming growth factor beta-induced connective tissue growth fac-tor expression in human aortic smooth muscle cells by interfering

5. Parola M, Robino G. Oxidative stress-related molecules and liverfibrosis. J Hepatol 2001;35:297–306.

6. Gomez JA, Molero X, Vaquero E, Alonso A, Salas A, MalageladaJR. Vitamin E attenuates biochemical and morphological featuresassociated with development of chronic pancreatitis. Am J Gas-trointest Liver Physiol 2004;287:G162–G169.

7. Sun X, Majumder P, Shioya H, Wu F, Kumar S, Weichselbaum R,Kharbanda S, Kufe D. Activation of the cytoplasmic c-Abl tyrosinekinase by reactive oxygen species. J Biol Chem 2000;275:17237–17240.

8. Li B, Wang X, Rasheed N, Hu Y, Boast S, Ishii T, Nakayama K,Nakayama KI, Goff SP. Distinct roles of c-Abl and Atm in oxidativestress response are mediated by protein kinase C. Gene Dev2004;18:1824–1837.

9. Hantschel O, Superti-Furga G. Regulation of the c-Abl and BCR-Abltyrosine kinases. Nat Rev Mol Cell Biol 2004;5:33–44.

0. Daniels CE, Wikes MC, Edens M, Kotton TJ, Murphy SJ, LimperAH, Leof EB. Imatinib mesylate inhibits the profibrogenic activityof TGF-� and prevents bleomycin-mediated lung fibrosis. J ClinInvest 2004;114:1308–1316.

1. Wang S, Wilkes MC, Leof EB, Hrschberg R. Imatinib mesylateblocks a non-Smad TGF-� pathway and reduces renal fibrogen-esis in vivo. FASEB J 2005;19:1–11.

2. Garcia-Trevijano ER, Iraburu MJ, Fontana L, Dominguez-RosalesJA, Auster A, Covarrubias-Pinedo A, Rojking M. Transforminggrowth factor �1 induces the expression of �1(I) procollagenmRNA by a hydrogen peroxide-C/EBP�-dependent mechanism inrat hepatic stellate cells. Hepatology 1999;29:960–970.

3. Aono Y, Nishioka Y, Inayama M, Ugai M, Kishi J, Uehara H, IzumiK, Sone S. Imatinib as a novel antifibrotic agent in bleomycin-induced pulmonary fibrosis in mice. Am J Respir Crit Care Med2005;171:1279–1285.

4. Neuschwander-Tetri BA, Brunt EM, Wehmeir KR, Oliver D, BaconBR. Improved non-alcoholic steatohepatitis after 48 weeks oftreatment with the PPAR ligand rosiglitazone. Hepatology 2003;38:1008–1017.

5. Sanyal AJ, Mofrad PS, Contos MJ, Sargeant C, Luketic VA, Ster-ling RK, Stravitz RT, Shiffman ML, Clore J, Mills AS. A pilot studyof vitamin E versus vitamin E and pioglitazone for the treatmentof non-alcoholic steatohepatitis. Clin Gastroenterol Hepatol2004;2:1107–1115.

6. Tomita K, Azuma T, Kitamura N, Nishida J, Tamiya G, Oka A,Inokuchi S, Nishimura T, Suematsu M, Ishii H. Pioglitazone pre-vents alcohol-induced fatty liver in rats though up-regulation ofc-Met. Gastroenterology 2004;126:873–885.

Received July 5, 2005. Accepted July 5, 2006.Address requests for reprints to: Andrea Galli, MD, PhD, Gastroen-

erology Unit, Department of Clinical Pathophysiology, University oflorence, Viale Morgani 85, 50134 Firenze, Italy. e-mail: a.gallidfc.unifi.it; fax: (39) 055-4271297.Supported by grants from Ministero dell’ Istruzione, dell’ Università,della Ricerca (MUIR), Regione Toscana (TRESOR project), Cassa diisparmio di Firenze (CRF), and FiorGen Foundation.The authors thank Dr J. Mizukami for human PPAR�1 expression

lasmid, Dr D. Barilàfor c-Abl constructs, W. Li for PKC� constructs, Dr. Ramirez for pMS-3.5/CAT, Dr D. Schuppan for anticollagen antibody,nd Prof G. Gabbiani for anti—�–smooth muscle actin antibody; Pro-essors A. Casini, W. Bosron, and M. Serio for many helpful commentsnd suggestions; and R. Salzano, R.A. Ross, and D. Price for technical