BT

DLA

SMS

*PCC

BccwsamlptcaMaathiawarhtdrt

Aps

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

GASTROENTEROLOGY 2008;135:2084–2095

ASIC—LIVER, PANCREAS, AND BILIARYRACT

evelopment of Nonalcoholic Steatohepatitis in Insulin-Resistantiver-Specific S503A Carcinoembryonic Antigen-Related Celldhesion Molecule 1 Mutant Mice

ANG JUN LEE,* GARRETT HEINRICH,* LARISA FEDOROVA,‡ QUSAI Y. AL–SHARE,* KELLY J. LEDFORD,*ATS A. FERNSTROM,* MARCIA F. MCINERNEY,§ SANDRA K. ERICKSON,� CARA GATTO–WEIS,¶ andONIA M. NAJJAR*

The Center for Diabetes and Endocrine Research and the Department of Physiology and Pharmacology, ‡Department of Medicine, and the ¶Department ofathology at the College of Medicine at the University of Toledo, Health Science Campus, Toledo, Ohio; §Department of Medicinal and Biological Chemistry at the

�

ollege of Pharmacy at the University of Toledo, Main Campus, Toledo, Ohio; and the Department of Medicine, University of California, and Veterans Affairs Medicalenter, San Francisco, California

tcWldai

fle“ptao

viisgP

ttgeNNPpR1w

See editorial on page 1860.

ackground & Aims: Liver-specific inactivation ofarcinoembryonic antigen–related cell adhesion mole-ule 1 causes hyperinsulinemia and insulin resistance,hich result from impaired insulin clearance, in liver-

pecific S503A carcinoembryonic antigen–related celldhesion molecule 1 mutant mice (L-SACC1). Theseice also develop steatosis. Because hepatic fat accumu-

ation precedes hepatitis, lipid peroxidation, and apo-tosis in the pathogenesis of nonalcoholic steatohepa-

itis (NASH), we investigated whether a high-fat diet, byausing inflammation, is sufficient to induce hepatitisnd other features of NASH in L-SACC1 mice.ethods: L-SACC1 and wild-type mice were placed onhigh-fat diet for 3 months, then several biochemical

nd histologic analyses were performed to investigatehe NASH phenotype. Results: A high-fat diet causedepatic macrosteatosis and hepatitis, characterized by

ncreased hepatic tumor necrosis factor � levels andctivation of the NF-�B pathway in L-SACC1 but not inild-type mice. The high-fat diet also induced necrosisnd apoptosis in the livers of the L-SACC1 mice. Insulinesistance in L-SACC1 fed a high-fat diet increased theepatic procollagen protein level, suggesting a role in

he development of fibrosis. Conclusions: A high-fatiet induces key features of human NASH in insulin-esistant L-SACC1 mice, validating this model as a toolo study the molecular mechanisms of NASH.

bout one third of adults in the United States arediagnosed with fatty liver disease, with 20%–30%

redicted to develop fibrosing steatohepatitis and 10%

howing the full spectrum of nonalcoholic steatohepati-

is (NASH). Incidence of the disease is expected to in-rease in parallel to the increased prevalence of obesity.1

ith NASH progressing to cirrhosis and/or hepatocellu-ar carcinoma and causing end-stage liver disease,2 theisease is projected to become the leading liver diseasend cause of liver transplantation as a result of cirrhosisn Western countries.

NASH is characterized by hepatic macrosteatosis, in-ammation, and fibrosis. Its pathogenesis is not fullylucidated, but the most prevalent mechanism is thetwo-hit” hypothesis.3 According to this hypothesis, he-atic steatosis initially develops (first hit) and predisposeso lipid peroxidation and inflammation, leading to hep-titis, apoptosis, fibrosis, and, ultimately, cirrhosis (sec-nd hit).

Activation of hepatic peroxisome proliferator-acti-ated receptor � (PPAR�)-dependent mechanisms dur-ng fasting increases transcription of enzymes involvedn fatty acid mitochondrial transport and �-oxidation,uch as carnitine palmitoyl transferase 1, to supportluconeogenesis. Some of these are co-regulated byPAR� co-activator 1� (PGC-1�),4 which is involved

Abbreviations used in this paper: CEACAM1, carcinoembryonic an-igen–related cell adhesion molecule 1; G-6-Pase, glucose-6-phospha-ase; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GSH,lutathione; HF, high-fat diet; L-SACC1, liver-specific S503A carcino-mbryonic antigen–related cell adhesion molecule 1 mutant mouse;ASH, nonalcoholic steatohepatitis; NF-�B, nuclear factor � B; NPC-1,iemann Pick type C1; PDK-4, pyruvate dehydrogenate kinase;EPCK, phosphoenolpyruvate carboxykinase; PPAR �,�, peroxisomeroliferator-activated receptor �,�; PGC-1�, PPAR� co-activator 1�;D, regular diet; SREBP1c, sterol regulatory element-binding proteinc; TNF�, tumor necrosis factor-�; UCP2, uncoupled protein 2; WT,ild-type mouse.

© 2008 by the AGA Institute0016-5085/08/$34.00

doi:10.1053/j.gastro.2008.08.007

mrc(sdptsgeIihpj

sbaNaodmcroqmltmlwplmNtttft

tmrsohhdrtp

ttdkhmN

aatwstR

wbsRaSWe((ptHatcrc

HgaNtK

bSmt

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

December 2008 INSULIN RESISTANCE IN NASH PATHOGENESIS 2085

ainly in promoting mitochondrial biogenesis andegulation of genes in the oxidative phosphorylationhain, such as the mitochondrial uncoupled protein-2UCP2), which reduces adenosine triphosphate synthe-is when activated by superoxides and the lipid peroxi-ation end products.5 Under conditions of obesity androlonged high-fat intake, excessive fatty acid oxida-ion and lipid �-peroxidation promote oxidativetress.6 Together with reduction of the mitochondriallutathione (GSH) defense system against the cytotoxicffect of tumor necrosis factor � (TNF�), this activates�B kinase (IKK)�-dependent nuclear factor � B (NF-�B)nflammatory pathways and causes insulin resistance,7

epatitis,8 and mitochondrial dysfunction. It also predis-oses to cell death and hepatocyte susceptibility to in-

ury, and progressive liver diseases such as NASH.9

Although NASH may develop in association with in-ulin resistance,10,11 the molecular relationship has noteen well delineated,12 in part owing to the lack of annimal model that replicates the human state adequately.o animal model has developed NASH spontaneously,

nd few may develop some of the clinical manifestationsf the disease.12,13 The methionine-choline– deficientiet induces fibrosing steatohepatitis. However, hu-an beings with NASH do not show methionine or

holine deficiency and this diet does not cause insulinesistance. The relevance of the leptin-deficient Ob/Obbese mouse in NASH pathogenesis also has beenuestionable because altered leptin signaling can itselfodulate inflammatory response, fibrosis, and hepatic

ipid metabolism.14 Insight provided by the phospha-ase and tensin homolog on chromosome 10 mutant

ouse also is limited because it is insulin sensitive andean, and it develops massive steatosis by comparisonith human NASH.15 The transgenic mouse with adi-ose tissue–specific expression of nuclear sterol regu-

atory element-binding protein 1c (SREBP-1c) displaysarked steatosis with a liver histology similar toASH.16 Because this mouse shows inherited lipodys-

rophy with hypoleptinemia and severe insulin resis-ance, it does not fully replicate the clinical manifes-ation of NASH. Thus, these experimental modelsailed to address adequately the role of insulin resis-ance in NASH pathogenesis.

L-SACC1 mice with liver-specific overexpression ofhe dominant-negative S503A phosphorylation-defective

utant of CEACAM1 (the CarcinoEmbryonic Antigen-elated Cell Adhesion Molecule 1) develop hepatic steato-is with increased hepatic triglyceride output and visceralbesity,17 resulting from impaired insulin clearance andyperinsulinemia. This shows that CEACAM1 promotesepatic insulin clearance in a phosphorylation-depen-ent manner. CEACAM1 also acts as a co-inhibitoryeceptor after T-cell activation to inhibit inflamma-ion.18 Because this function depends on the phos-

horylation of immunoreceptor tyrosine-based inhibi- e

ion motifs within the cytoplasmic domain in additiono SHP1,19 we investigated whether L-SACC1 miceevelop a NASH-like phenotype under conditionsnown to trigger inflammation, such as prolongedigh-fat (HF) intake.20 We herein report that L-SACC1ice show key features of obesity-related humanASH when fed HF for 3 months.

Materials and MethodsAnimal MaintenanceAnimals were kept in a 12-hour dark/light cycle

nd fed standard chow ad libitum. All procedures werepproved by the Institutional Animal Care and Utiliza-ion Committee. Six-month-old genetically matchedild-type (WT) and L-SACC1 female mice were fed a

tandard chow (regular diet [RD]) or an HF diet (45%otal calories from fat) for 3 months (Cat #D12451;esearch Diets Inc, New Brunswick, NJ).

Metabolic PhenotypeAfter an overnight fast, mice were anesthetized

ith sodium pentobarbital at 11:00 am. Whole venouslood was drawn from the retro-orbital sinuses to mea-ure fasting glucose levels using a glucometer (Accu-chek;oche Applied Science, Indianapolis, IN), serum insulinnd leptin levels by radioimmunoassays (Linco Research,t Charles, MO), serum free fatty acids (NEFA C kit;ako, Osaka, Japan), triglyceride levels (Infinity Triglyc-

rides; Sigma, St Louis, MO) and cholesterol levelsPointe Scientific, Canton, MI), and free cholesterolWako). Serum alanine aminotransferase (ALT) and as-artate aminotransferase (AST) levels were measured perhe manufacturer’s instructions (Biotron Diagnostic,emet, CA). Visceral adipose tissue was excised, weighed,

nd visceral adiposity was expressed as a percentage ofotal body weight. Hepatic total cholesterol and freeholesterol were measured using the Infinity cholesteroleagent (Thermo Electron, Waltham, MA) and a freeholesterol reagent (Wako), respectively.21

Liver HistologyHistologic examination was established using

&E of formalin-fixed, paraffin-embedded liver. The de-ree of steatosis and lobular inflammation was graded on0 –3 scale, with 3 being the highest, according to theASH scoring system, which recently was proposed by

he National Institute for Diabetes and Digestive andidney Diseases–NASH Clinical Research Network.22

Liver StainingDeparaffinized and rehydrated slides were incu-

ated in a 0.1% solution of Sirius Red (Direct Red 80;igma) in saturated picric acid in the dark for 1 hour andounted with resin. For the terminal deoxynucleotidyl

ransferase–mediated deoxyuridine triphosphate nick-

nd labeling assay, sections were stained with ApopTag

PtstLpmCespLG

BBam3f

GeytrA

wif8tmmt

AsCaBzIC(p(TtdMc

T

AABVRFSSSSHSHHHI

a

b

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

2086 LEE ET AL GASTROENTEROLOGY Vol. 135, No. 6

lus Peroxidase Apoptosis Detection Kit (Chemicon In-ernational, Temecula, CA) per the manufacturer’s in-tructions. For immunohistochemical analysis, liver sec-ions were blocked in 1.5% horse serum (Vectoraboratories, Burlingame, CA) for 30 minutes androbed with fluorescein isothiocyanate– conjugated anti-ouse antibodies to F4/80 (1:20; eBioscience, San Diego,A), and CD3 or CD4 (1:100; BD Pharmingen, San Di-

go, CA) for 1 hour in the dark. Nuclei were counter-tained with propidium iodide. Eight to 10 random fieldser section were counted. Images were captured using aeica TCS SP5 broadband confocal microscope (Wetzlar,ermany).

GSH AssayReduced glutathione was measured using the

ioxytech GSH-400 kit (OXISResearch, Portland, OR).23

riefly, liver tissue was homogenized in 5% metaphosphoriccid before adding a chromogenic reagent (4-chloro-1-ethyl-7-trifluromethyl-quinolinium methylsulfate) and

0% NaOH, followed by incubation at room temperatureor 10 minutes in the dark and measurement at A400 nm.

Nitrite AssaySerum nitrite levels were measured using the

riess reagent system (Promega, Madison, WI). Briefly,qual amounts of sulfanilamide solution and N-1-napth-lethylenediamine dihydrochloride solution were addedo equal volumes of serum, and allowed to incubate atoom temperature for 10 minutes before measurement at540 nm.

able 1. Phenotype Characterization of Age- and Sex-Matche

LT level, �mol/L 42ST level, �mol/L 14ody weight, g 20isceral fat, % bwt 0.1andom blood glucose level, mg/dL 67asting blood glucose level, mg/dL 58erum leptin level, mg/dL 11erum insulin, pmol/L 73erum free fatty acid level, mEq/L 0.3erum triglyceride level, mg/dL 40epatic triglyceride, mg/g protein 23erum cholesterol level, ng/dL 61epatic total cholesterol, �g/mg protein 15epatic free cholesterol, �g/mg protein 11epatic cholesterol esters, �g/mg protein 3.5

n vivo cholesterol synthesis, �mol 3H20 incorporation/h/gWhole body 0.8Liver 4.7Heart 0.1Small intestine 1.7

P � .05 vs WT-RD. Values are expressed as mean � SEM.

P � .05 HF vs RD. Values are expressed as mean � SEM.

Thiobarbituric Acid Reactive SubstancesAssayLipid peroxidation was measured as described13

ith modifications. Briefly, liver tissue was homogenizedn 1.15% KCl before centrifugation at 10,000 rpm at 4°Cor 10 minutes. The supernatant (100 �L) was added to.1% sodium dodecyl sulfate, 20% acetic acid, and 0.8%hiobarbituric acid, followed by heating at 95°C for 60

inutes. n-butanol-pyridine (15:1) was added to the coolix before centrifugation at 4000 � g for 10 minutes and

he upper layer was measured at A532 nm.

Western AnalysisTotal tissue lysates or 10 �g serum lysates (for

poB) were analyzed by 4%–12% gradient sodium dodecylulfate–polyacrylamide gel electrophoresis (Invitrogen,arlsbad, CA) before Western analysis with polyclonalntibodies against PPAR�, PGC-1�, UCP2 (Santa Cruziotechnology, Santa Cruz, CA), cytochrome p450 en-yme (Chemicon International), ApoB48/100 (Chemiconnternational), fatty acid synthase,24 Niemann Pick type1 (NPC-1) (Abcam, Cambridge, MA), procollagen

SouthernBiotech, Birmingham, AL), and p65 NF-�Bhosphoserine (Ser 536), p65 NF-kB, and protein kinase C

PKC�) phosphothreonine-Thr 410/403 (Cell Signalingechnology, Danvers, MA), in addition to monoclonal an-

ibodies against PPAR�, PKC�, glyceraldehyde-3-phosphateehydrogenase (GAPDH) (Santa Cruz), and actin (Sigma).embranes were incubated with horseradish-peroxidase–

onjugated anti–immunoglobulin G antibody and proteins

and L-SACC1 Mice Fed an RD or HF Diet for 3 Months

WT SACC1

HF RD HF

1.01 41.0 � 1.15 54.2 � 4.88a 50.3 � 4.02a

9.68 160. � 6.27 185. � 11.2a 170. � 7.31a

0.66 28.3 � 3.10b 27.1 � 1.76a 36.4 � 1.71a,b

0.08 5.50 � 2.26b 4.96 � 1.33a 12.4 � 2.85a,b

6.74 115. � 9.40b 108. � 5.78a 130. � 7.50a,b

13.1 58.0 � 4.57 45.7 � 4.40 87.0 � 18.2a,b

1.08 33.5 � 11.8b 65.6 � 22.2a 98.0 � 33.7a

2.55 153. � 22.6b 153. � 23.4a 188. � 28.2a

0.01 0.52 � 0.07b 0.68 � 0.08a 0.74 � 0.08a

4.68 52.3 � 12.4 63.4 � 3.27a 27.8 � 3.49a,b

2.81 36.5 � 3.07b 34.2 � 3.41a 49.3 � 3.58a,b

1.18 72.0 � 5.36 62.7 � 3.04 76.1 � 9.751.21 15.7 � 0.62 13.2 � 0.51 15.8 � 1.270.93 10.7 � 0.50 6.83 � 0.65a 8.55 � 0.73a

0.55 4.97 � 0.62 6.36 � 0.46a 7.28 � 1.00a

0.10 0.69 � 0.110.72 9.99 � 1.28a

0.10 0.06 � 0.050.16 2.07 � 0.09

d WT

RD

.4 �0. �.5 �8 �.5 �.3 �.6 �.0 �8 �.2 �.6 �.2 �.4 �.9 �7 �

9 �2 �0 �0 �

wPa

Tblnc(am

gbt(I

GTrGaC

cPn

SsPc

FheLRaaPoGlwSrlbiMfm�

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

December 2008 INSULIN RESISTANCE IN NASH PATHOGENESIS 2087

ere detected by enhanced chemiluminescence (Amershamharmacia Biotech, Buckinghamshire, United Kingdom)nd quantified by densitometry.

Northern AnalysisLiver messenger RNA (mRNA) was extracted with

rizol (Invitrogen) and the MicroPoly(A) Pure kit (Am-ion, Austin, TX) before probing with Random Primed–

abeled complementary DNAs (cDNAs) (Roche) for car-itine palmitoyl transferase 1, phosphoenolpyruvatearboxykinase (PEPCK), pyruvate dehydrogenate kinasePDK-4), glucose-6-phosphatase (G-6-Pase), SREBP1c,nd TNF� before reprobing with GAPDH cDNA to nor-alize for the amount loaded.

Real-Time Polymerase Chain ReactionRNA was extracted using RNeasy Mini Kits (Qia-

en, Valencia, CA) before TNF� cDNA was synthesizedy M-MLV reverse transcriptase, and its level was quan-itated by Absolute SYBR Green real-time PCR mastermixThermo Fisher Scientific, Waltham, MA) in a Bio-Rad-cycler (Hercules, CA). TNF� level was normalized with

igure 1. Northern analysis ofepatic enzymes. Liver wasxtracted from fed WT and-SACC1 mice, which had beenD-fed or HF-fed for 3 months,nd mRNA was purified and an-lyzed sequentially for (A) Cpt-1,dk4, and Gapdh levels. (B) An-ther aliquot was analyzed for-6-Pase and Gapdh mRNA

evels. (C and D) Other aliquotsere analyzed for Pepck andrebp1c, respectively, before

eprobing with Gapdh mRNAevels. Bands were quantitatedy densitometry and presented

n the graph as arbitrary units.ore than 5 samples were used

rom each feeding group ofice. Values are mean � SE. *P

.05 HF vs RD.

APDH using the following primers: TNF� forward:TCTGTCTACTGAACTTCGGGGTGATCGGTCC, TNF�

everse: GTATGAGATAGCAAATCGGCTGACGGTGT-GG, GAPDH forward: CCAGGTTGTCTCCTGCGACT,

nd GAPDH reverse: ATACCAGGAAATGAGCTTGA-AAAGT.

Statistical AnalysisData were analyzed with Statview software (Aba-

us, Berkeley, CA) using one-factor analysis of variance. Avalue of less than .05 was considered statistically sig-

ificant.

ResultsHF Diet Exacerbates the MetabolicDerangement in L-SACC1 MiceAs previously reported,17,25 9-month-old female L-

ACC1 mice develop higher body weight and visceral obe-ity by comparison with their WT counterparts (Table 1).rolonged HF intake for 3 months increases visceral fat and,onsequently, serum leptin levels, in both groups of mice,

wii

l

1acap

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

2088 LEE ET AL GASTROENTEROLOGY Vol. 135, No. 6

ith a stronger effect on WT mice (Table 1). Moreover, HFncreases serum insulin and free fatty acid levels by approx-mately 2-fold in WT, but not L-SACC1, mice (Table 1).

Consistent with insulin resistance,17,25 random glucoseevel is higher in RD-fed L-SACC1 than WT mice (Table

). An increase in hepatic Pepck and G-6-Pase mRNA levelst fasting,17,25 but not postprandial state, mice fed ahow-diet (RD) (Figure 1) suggests that L-SACC1 micere geared toward gluconeogenesis, and hence are predis-osed to develop fasting hyperglycemia in response to

Figure 2. Hepatic lipid metabo-lism. (A) Liver lysates from morethan 3 mice from each of the 4feeding groups (RD- and HF-WT, and RD- and HF–L-SACC1)were analyzed by immunoblot-ting (IB) with �-fatty acid syn-thase (FAS), �-PPAR�, and �–NPC-1 antibodies followed byreprobing (reIB) with �-actin or�-GAPDH to normalize for theamount of protein loaded. (B) Atotal of 10 �g serum proteinsfrom more than 5 mice per feed-ing group were analyzed byWestern blotting with �-ApoBantibody to detect ApoB100(�500 kilodaltons) and ApoB48(�200 kilodaltons), which werequantitated by densitometry andpresented in the bar graph as ar-bitrary units. Values are mean �SE. *P � .05 HF vs RD. †P � .05vs WT-RD. (C) Liver sectionsfrom more than 5 mice of each ofWT-RD (block 1), WT-HF (block2), L-SACC1–RD (block 3), andL-SACC1–HF (block 4) were an-alyzed by H&E staining for lipid

accumulation.

timpGai(

ti(h

FhspbcmmtaS.fgbN(A(t(oqpu

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

December 2008 INSULIN RESISTANCE IN NASH PATHOGENESIS 2089

he HF diet. In fact, HF causes an approximately 2-foldncrease (Table 1) in the fasting glucose level of L-SACC1

ice together with a significant increase in hepatic post-randial mRNA levels of gluconeogenic enzymes: Pepck,-6-Pase, and Pdk-4 (Figure 1). In contrast, HF does notlter the fasting blood glucose level in WT mice. Instead,t causes a more marked increase in fed glucose level�2-fold) than in L-SACC1 mice (Table 1).

igure 3. Characterization ofepatic inflammation. (A) Liverections from more than 5 miceer feeding group were analyzedy H&E staining for inflammatoryell infiltration. (B) Hepatic Tnf�RNA levels from more than 3ice per feeding group were de-

ermined by Northern analysis,s in Figure 1. Values are mean �E. *P � .05 HF vs RD. †P �

05 vs WT-RD. (C) Liver lysatesrom more than 3 mice per feedingroup were analyzed by Westernlotting (IB) with �-phospho–F-�B (upper gel) and reprobing

reIB) with �–NF-�B (lower gel).ctivation of protein kinase C

PKC)� was analyzed similarly forhe amount of phosphorylatedupper gel) relative to protein levelf PKC� (lower gel). Bands wereuantitated by densitometry andresented in the graph as arbitrary

nits.

HF Diet Alters Lipid Metabolism inL-SACC1 Mice

As previously reported,17 hepatic triglyceride con-ent is higher in L-SACC1 than in WT mice (Table 1), butn response to HF it is increased to the same extent�1.8-fold) in both mouse groups (Table 1). Becauseyperinsulinemia increases the transcription of lipogenic

eciStea(nLegH(2lnh(2ncwoi

shTerlHsNL

eii0imHRh

c(iaisR

thWQti(P0

rpltHeppW

Ftaas

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

2090 LEE ET AL GASTROENTEROLOGY Vol. 135, No. 6

nzymes,26 this could result, at least in part, from in-reased de novo lipogenesis, as suggested by the approx-mately 2- to 3-fold increase in hepatic mRNA levels ofrebp-1c (Figure 1C) and the protein level of one of itsargets, fatty acid synthase (Figure 2A). Basal proteinxpression of PPAR�, a key regulator of lipogenesis indipocytes, is higher in L-SACC1 relative to WT livers�3-fold) (Figure 2A; RD-fed L-SACC1 vs WT) and doesot increase further by HF (Figure 2A; HF-fed vs RD-fed-SACC1 mice). The increase in hepatic de novo lipogen-sis in L-SACC1 mice correlates with higher serum tri-lyceride levels when mice are fed RD (Table 1), but notF, which causes a marked decrease in serum triglyceride

Table 1) and ApoB100/ApoB48 protein levels (FigureB). The latter suggests reduced hepatic output of trig-

ycerides by HF in L-SACC1 mice. In WT mice, HF doesot increase serum triglyceride levels despite increasingepatic production, as indicated by increased Srebp-1c

Figure 1), fatty acid synthase, and PPAR� levels (FigureA). With serum ApoB100/Apo48 protein content beingormal (Figure 2B), a reduction in serum triglycerideould result from substrate redistribution from liver tohite adipose tissue, as suggested by increased visceralbesity and increased release of free fatty acids and leptin

n HF-fed WT mice (Table 1).Although total body and intestinal in vivo cholesterol

ynthesis are unchanged, hepatic cholesterol synthesis isigher (�2-fold) in chow-fed L-SACC1 mice (Table 1).his leads to an increase in hepatic content of cholesterolsters with a reciprocal decrease in free cholesterol levels,esulting in a net balance of total cholesterol levels iniver and serum from L-SACC1 mice (Table 1). Although

F does not change hepatic cholesterol levels, it reducesignificantly (�50%) the hepatic protein content ofPC-1, a late endosomal cholesterol traffic protein, in-SACC1, but not WT, mice (Figure 2A).H&E staining of liver sections reveals that HF differ-

ntially modifies the histologic manifestation of steatosisn L-SACC1 and WT mice (Figure 2C). HF markedlyncreases hepatic steatosis in L-SACC1 mice (score, 1.33 �.28 vs 0.30 � 0.12 in RD-fed mice; P � .05), and changes

ts histology from microsteatosis (Figure 2C, block 3) toacrosteatosis (Figure 2C, block 4). In WT mice, however,F increases steatosis (score, 0.58 � 0.30 vs 0.00 � 0.00 inD-fed mice; P � .05), while preserving its microvesicularistology (Figure 2C, block 2).

HF Diet Induces Hepatitis in L-SACC1 MiceH&E staining of liver sections reveals that HF

auses scattered inflammatory infiltration in L-SACC1Figure 3A, block 4 vs 3; score, 1.25 � 0.28 vs 0.40 � 0.19n RD-fed mice; P � .05), with altered hepatocellularrchitecture. In contrast, WT mice show few, if any,nflammatory islands (Figure 3A, block 2 vs 1), with noignificant score difference (0.42 � 0.15 vs 0.50 � 0.00 in

D-fed mice; P � .05), and maintain normal cell archi- fi

ecture in response to HF diet. Consistently, HF increasesepatic Tnf� mRNA levels (�2-fold) in L-SACC1, but notT, mice as assessed by Northern analysis (Figure 3B).uantitative reverse-transcription polymerase chain reac-

ion reveals that HF diet also increases Tnf� mRNA levelsn white adipose tissue derived from L-SACC1 mice8.86% � 1.42% vs 3.83% � 0.99% Gapdh in RD-fed mice;

� .023), but not WT mice (1.54% � 0.03% vs 1.99% �.05% Gapdh in RD-fed mice; P � .05).

Consistent with a role for TNF� in activating IKK-�, aedox-sensitive kinase that up-regulates proinflammatoryathways,27 HF markedly activates NF-�B in L-SACC1

iver, as indicated by an approximately 4-fold increase inhe phosphorylation of NF-�B (Figure 3C). In contrast,F does not activate NF-�B in WT mice to the same

xtent (Figure 3C). In support of PKC� playing an im-ortant role in NF-�B activation,28 HF increases phos-horylation (activation) of PKC� in L-SACC1, but notT, liver (Figure 3C).

igure 4. Immunohistochemical analysis of inflammation. Liver sec-ions were analyzed with (A) fluorescein isothiocyanate–conjugatednti-F4/80 antibody and (B) fluorescein isothiocyanate–conjugatednti-CD3 and anti-CD4 antibodies. Experiments were repeated on 3ections per mouse. Counts are mean positively stained cells per visible

eld � SE. *P � .05 HF vs RD.

mvLaLtiaCe

PWwfpvs1f

ai(i5b(tlwsbsc5mpdLdm

Fttw(�a(iolpaBtg(rVH

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

December 2008 INSULIN RESISTANCE IN NASH PATHOGENESIS 2091

In further support of increased release of TNF� fromacrophages, immunostaining analysis with F4/80 re-

eals a larger macrophage pool in the liver of HF-fed-SACC1 mice (Figure 4A, block 4 vs 1–3, green stain). Inddition, HF induces CD3�/CD4� T-cell population in-SACC1, but not WT, mice as assessed by immunohis-ochemical analysis (Figure 4B). Taken together, the datandicate that HF diet induces hepatitis in L-SACC1 mice,nd that this may implicate an increase in TNF� andD4� T-cell population, both of which are critical play-

rs in NASH pathogenesis.

HF Diet Causes Oxidative Changes inL-SACC1 MiceWestern analysis reveals that hepatic PPAR�,

GC-1 1�, and UCP2 levels are higher in L-SACC1 thanT mice by approximately 2-fold (Figure 5A). Consistentith increased activation of PPAR� by HF,29 prolonged

at feeding induces PPAR� hepatic protein levels by ap-roximately 2-fold in both mouse groups (Figure 5A, HFs RD lanes), and induces Cpt1 mRNA levels with a moretatistically significant effect in L-SACC1 mice (FigureA). This suggests increased fatty acid oxidation in HF-

igure 5. Nitroso-redox evalua-ion. (A) Liver lysates from morehan 3 mice per feeding groupere analyzed by immunoblotting

IB) sequentially with �-PPAR�,-uncoupled protein-2 (UCP2),nd �-PPAR� co-activator 1�

PGC1�) antibodies before reprob-ng (reIB) with �-actin (upper groupf gels). Another aliquot was ana-

yzed similarly with �-cytochrome450 enzyme (CYP2E1) antibodynd �-actin (lower group of gels).ands were quantitated by densi-

ometry and presented in the barraph as arbitrary units. (B) Liver

TBARS)/MDA and GSH, and se-um was assayed for nitrite levels.alues are mean � SE. *P � .05F vs RD. †P � .05 vs WT-RD.

ed L-SACC1 mice. i

The protein level of hepatic cytochrome p450 enzyme,member of the microsomal cytochrome p450 that is

nvolved in the metabolism of long-chain fatty acidslipooxygenation) and microsomal lipid �-peroxidation,s comparable in RD-fed L-SACC1 and WT mice (FigureA). HF increases cytochrome p450 enzyme protein levely approximately 3-fold in L-SACC1, but not WT, miceFigure 5A). Moreover, lipid peroxidation, as assessed byhiobarbituric acid reactive substance concentration iniver lysates, is highly increased in HF-fed as comparedith RD-fed L-SACC1 and WT mice (Figure 5B). This

uggests initiation of oxidative changes in L-SACC1 livery HF diet.30 In support of this hypothesis, the level oferum nitrite, a nitric oxide oxidation product, is de-reased in L-SACC1, as compared with WT mice (FigureB), suggesting lower nitric oxide bioavailability in theseice, as in other models of obesity.31 Consistent with the

ositive effect of TNF� on nitrite production,32 HF in-uces a more marked increase in serum nitrite level in-SACC1 than WT mice (Figure 5B, HF vs RD) andecreases hepatic GSH levels in L-SACC1, but not WT,ice (Figure 5B). Thus, HF promotes a nitroso-redox

mbalance in L-SACC1 mice.

(imtHdth6voa

b

LoicW

odmsvhAWio

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

2092 LEE ET AL GASTROENTEROLOGY Vol. 135, No. 6

HF Diet Induces Necrosis and Apoptosis inL-SACC1 MiceWestern analysis of liver lysates indicates that poly-

adenosine diphosphate-ribose) polymerase is cleaved intots smaller 85-kilodalton subunit only in HF-fed L-SACC1

ice, but not in the other groups (Figure 6A). This suggestshat early apoptosis develops in L-SACC1 mice only afterF feeding.33 This observation is supported by terminaleoxynucleotidyl transferase–mediated deoxyuridineriphosphate nick-end labeling staining, which detectsistologic changes only in HF-fed L-SACC1 mice (FigureB, block 4 [arrowheads] vs blocks 1–3). H&E staining re-eals that livers from HF-fed L-SACC1 mice, but notthers, also show necrosis (Figure 6B, block 5 [arrowheads]nd Figure 3, block 4).

Development of Early Fibrosis inL-SACC1 MiceLiver lysates from L-SACC1 mice show increased

asal procollagen protein content (Figure 7A, RD-fed o

-SACC1 vs WT), with a NASH-like chicken-wire patternf the Sirius red stain (Figure 7B, block 3). In addition to

nducing procollagen content (Figure 7A), HF increaseshicken-wire fibrogenic changes in L-SACC1, but not

T, mice (Figure 7B, block 4 vs 3).

DiscussionThe mechanisms underlying NASH and the role

f insulin resistance in its pathogenesis are poorly un-erstood, in part because of the lack of an experimentalodel replicating the full spectrum of the disease.12 We

how that L-SACC1 mice, which show insulin resistance,isceral obesity, and hepatic steatosis,17 also develop earlyepatic fibrosis together with a mild increase in serumLT and AST levels (Table 1), and hepatic TNF� levels.hen fed HF, they develop other key features of NASH,

ncluding macrosteatosis followed by hepatitis, lipid per-xidation, oxidative stress, and hepatocellular injury.

As in human NASH,11 HF induces hepatic fatty acid

Figure 6. Analysis of cell in-jury. (A) Liver lysates from morethan 5 mice were analyzed byimmunoblotting (IB) with �-poly-(adenosine diphosphate-ribose)polymerase (PARP) before rep-robing with �-tubulin. (B) Liversections from all 4 feedinggroups were analyzed by termi-nal deoxynucleotidyl trans-ferase–mediated deoxyuridinetriphosphate nick-end labeling(blocks 1–4) and H&E staining(block 5). Injured cells are em-phasized by arrowheads (blocks4 and 5).

xidation and gluconeogenesis in L-SACC1 mice and

coosa

scspacip

aamctioapi

et

alImiL

apglapacicmTaw

aCttfiptpc

hiwimrwsraN

aftldesstl

Fwfm

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT

December 2008 INSULIN RESISTANCE IN NASH PATHOGENESIS 2093

auses fasting hyperglycemia. The increase in fatty acidxidation fails to decrease hepatic triglyceride content,wing to increased triglyceride production and decreasedecretion, as indicated by an increase of Srebp-1c mRNAnd a reduction of ApoB100 levels,34 respectively.

CEACAM1 mutation elevates hepatic cholesterol synthe-is and esterification in the endoplasmic reticulum, whichould be facilitated by an increased supply of de novoynthesized fatty acids in the absence of CEACAM1.24 It isossible that the elevation in esterification occurs to preventn increase in free cholesterol level and protect against itsytotoxic effect.35,36 This assigns a role to CEACAM1 inntegrating cholesterol metabolism and fatty acid syntheticathways.

It is unlikely that an increase in cholesterol ester playsrole in the progression to steatohepatitis because this

ppears to require accumulation of free cholesterol initochondria.37 We did not measure mitochondrial free

holesterol content, but reduced NPC-1 levels may par-ition it from cytosolic lipid droplets to mitochondria38

n HF-fed L-SACC1 mice. As suggested by null mutationf NPC1,37 this could deplete mitochondrial GSH storesnd increase sensitivity to the cytotoxic effect of theroinflammatory cytokine, TNF�, the level of which is

ncreased in these mice.An increase in hepatic Tnf� mRNA could result from

levation in the pool of resident macrophages in response

igure 7. Analysis of fibrosis. (A) Liver lysates from more than 3 miceere analyzed by immunoblotting (IB) with �-procollagen antibody be-

ore reprobing (reIB) with �-tubulin. (B) Liver sections from more than 5ice of each feeding group were analyzed by Sirius red staining.

o increased free fatty acid supply39 and excessive lipid m

ccumulation in the hepatocyte, which could induce aocal inflammatory response in HF-fed L-SACC1 mice.40

t is possible that the intrahepatic inflammatory milieu isodulated by the release of adipokines, as suggested by

ncreased Tnf� mRNA in visceral white adipose tissue of-SACC1, but not WT, mice.In agreement with experimental animal models13,41

nd human beings42,43 with metabolic derangement, lipideroxidation also is increased in HF-fed L-SACC1. To-ether with a more marked increase in serum nitriteevels, this suggests that HF causes nitroso-redox imbal-nce and oxidative stress, which may pave the way foreroxidative events associated with necrotic damage44

nd apoptosis45 in L-SACC1 mice. We propose that HFauses these pathologies in L-SACC1, not WT, mice bynducing hepatic TNF�46 and increasing sensitivity to itsytotoxic effect through alterations of the GSH-baseditochondrial defense system. Consistent with a role forNF�-dependent activation of IKK-� in oxidative stressnd inflammation of NASH,39 HF activates NF-�B path-ays to a larger extent in L-SACC1 than WT mice.Steatohepatitis involves a Th1 cytokine response, char-

cterized by increased cytokine release from intrahepaticD4� T cells.20,47,48 Although the activation state of

hese cells was not examined, we observed an increase inheir population in L-SACC1 mice, as has been reportedor Ob/Ob mice.40 Because CEACAM1 has been implicatedn the anti-inflammatory response to T-cell activation,18 it isossible that inactivating CEACAM1 in hepatocytes limitshe CEACAM1-dependent inhibitory responses in T lym-hocytes and leads to a robust inflammatory response toytokines in L-SACC1 mice.

Chow-fed L-SACC1 mice show insulin resistance andepatic steatosis. Based on the normal physiology of

nsulin action, one would predict that insulin resistanceould not be associated with increased hepatic triglycer-

de content. This prediction is borne out in the LIRKOouse.49 However, unlike LIRKO, L-SACC1 mice have

educed, but not absent, insulin-receptor signaling,hich may lead to the peculiar admixture of insulin

ensitivity (in the context of increased lipogenesis) andesistance (in terms of increased fatty acid oxidation andltered glucose homeostasis) that is characteristic ofASH.In support of the hypothesis that insulin resistance is

n independent predictor for fibrosis in NASH,50 chow-ed L-SACC1 mice develop NASH-like fibrogenic changeshat can be attributed to the increase in both TNF� andeptin.46 In fact, increased leptin levels do not play airect role in the pathogenesis of NASH,51 but can exac-rbate the effects of TNF�.52 In addition, L-SACC1 micehow basal mild increases in AST and ALT levels. Con-istent with observations in a rabbit model of steatohepa-itis,13 HF intake does not further increase AST and ALTevels. With the majority of NASH patients showing nor-

al AST levels,53 this suggests that altered AST levels in

LlwopcNact

dihcoirim

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

BA

SIC–LIV

ER,

PA

NCREA

S,A

ND

BILIA

RY

TRA

CT

2094 LEE ET AL GASTROENTEROLOGY Vol. 135, No. 6

-SACC1 mice may closely parallel the mildly increasedevels in basal TNF�, leptin, and fibrogenic changes,hich in turn are likely to be associated with visceralbesity, insulin resistance, and hepatic steatosis. Weropose that these cellular, metabolic, and fibrogenichanges precondition the mice to develop the overtASH phenotype in parallel to increased inflammatory

ctivation and oxidative stress when cued by an in-rease in the cytotoxic effects of TNF� in response tohe HF diet.

The insulin-resistant L-SACC1 mice are predisposed toeveloping NASH in response to HF. This could be ow-

ng to altered lipid metabolism, which leads to reducedepatic GSH-based defense against TNF�, and to in-reased inflammatory response to TNF� in the absencef the anti-inflammatory effect of CEACAM1. By show-

ng a dual role for CEACAM1 in inhibiting inflammatoryesponse and reducing steatosis in liver, our observationsndicate that a reduction in CEACAM1 may serve as a

olecular link between insulin resistance and NASH.

References

1. Farrell GC, Larter CZ. Nonalcoholic fatty liver disease: from ste-atosis to cirrhosis. Hepatology 2006;43:S99–S112.

2. Hui JM, Kench JG, Chitturi S, et al. Long-term outcomes ofcirrhosis in nonalcoholic steatohepatitis compared with hepatitisC. Hepatology 2003;38:420–427.

3. Day CP, James OF. Steatohepatitis: a tale of two “hits”? Gastro-enterology 1998;114:842–845.

4. Yoon JC, Puigserver P, Chen G, et al. Control of hepatic glucone-ogenesis through the transcriptional coactivator PGC-1. Nature2001;413:131–138.

5. Echtay KS, Murphy MP, Smith RA, et al. Superoxide activatesmitochondrial uncoupling protein 2 from the matrix side. Studiesusing targeted antioxidants. J Biol Chem 2002;277:47129–47135.

6. Chavin KD, Yang S, Lin HZ, et al. Obesity induces expression ofuncoupling protein-2 in hepatocytes and promotes liver ATP de-pletion. J Biol Chem 1999;274:5692–5700.

7. Arkan MC, Hevener AL, Greten FR, et al. IKK-beta links inflamma-tion to obesity-induced insulin resistance. Nat Med 2005;11:191–198.

8. Dela Pena A, Leclercq I, Field J, et al. NF-kappaB activation, ratherthan TNF, mediates hepatic inflammation in a murine dietarymodel of steatohepatitis. Gastroenterology 2005;129:1663–1674.

9. McClain CJ, Mokshagundam SP, Barve SS, et al. Mechanisms ofnon-alcoholic steatohepatitis. Alcohol 2004;34:67–79.

0. Marchesini G, Marzocchi R. Metabolic syndrome and NASH. ClinLiver Dis 2007;11:105–117.

1. Abdelmalek MF, Diehl AM. Nonalcoholic fatty liver disease as acomplication of insulin resistance. Med Clin North Am 2007;91:1125–1149, ix.

2. Green RM. NASH—hepatic metabolism and not simply the met-abolic syndrome. Hepatology 2003;38:14–17.

3. Otogawa K, Kinoshita K, Fujii H, et al. Erythrophagocytosis by livermacrophages (Kupffer cells) promotes oxidative stress, inflamma-tion, and fibrosis in a rabbit model of steatohepatitis: implicationsfor the pathogenesis of human nonalcoholic steatohepatitis. Am JPathol 2007;170:967–980.

4. Leclercq IA, Farrell GC, Schriemer R, et al. Leptin is essential forthe hepatic fibrogenic response to chronic liver injury. J Hepatol

2002;37:206–213.

5. Kurlawalla-Martinez C, Stiles B, Wang Y, et al. Insulin hypersen-sitivity and resistance to streptozotocin-induced diabetes in micelacking PTEN in adipose tissue. Mol Cell Biol 2005;25:2498–2510.

6. Nakayama H, Otabe S, Ueno T, et al. Transgenic mice expressingnuclear sterol regulatory element-binding protein 1c in adiposetissue exhibit liver histology similar to nonalcoholic steatohepa-titis. Metabolism 2007;56:470–475.

7. Poy MN, Yang Y, Rezaei K, et al. CEACAM1 regulates insulinclearance in liver. Nat Genet 2002;30:270–276.

8. Gray-Owen SD, Blumberg RS. CEACAM1: contact-dependent con-trol of immunity. Nat Rev Immunol 2006;6:433–446.

9. Nagaishi T, Pao L, Lin SH, et al. SHP1 phosphatase-dependent Tcell inhibition by CEACAM1 adhesion molecule isoforms. Immu-nity 2006;25:769–781.

0. Li Z, Soloski MJ, Diehl AM. Dietary factors alter hepatic innateimmune system in mice with nonalcoholic fatty liver disease.Hepatology 2005;42:880–885.

1. Erickson SK, Lear SR, Dean S, et al. Hypercholesterolemia andchanges in lipid and bile acid metabolism in male and femalecyp7A1-deficient mice. J Lipid Res 2003;44:1001–1009.

2. Bondini S, Kleiner DE, Goodman ZD, et al. Pathologic assess-ment of non-alcoholic fatty liver disease. Clin Liver Dis 2007;11:17–23.

3. Kellogg AP, Wiggin TD, Larkin DD, et al. Protective effects ofcyclooxygenase-2 gene inactivation against peripheral nerve dys-function and intraepidermal nerve fiber loss in experimental dia-betes. Diabetes 2007;56:2997–3005.

4. Najjar SM, Yang Y, Fernstrom MA, et al. Insulin acutely decreaseshepatic fatty acid synthase activity. Cell Metab 2005;2:43–53.

5. Dai T, Abou-Rjaily GA, Al-Share’ QY, et al. Interaction betweenaltered insulin and lipid metabolism in CEACAM1-inactive trans-genic mice. J Biol Chem 2004;279:45155–45161.

6. Horton JD, Goldstein JL, Brown MS. SREBPs: activators of thecomplete program of cholesterol and fatty acid synthesis in theliver. J Clin Invest 2002;109:1125–1131.

7. Cai D, Yuan M, Frantz DF, et al. Local and systemic insulinresistance resulting from hepatic activation of IKK-beta and NF-kappaB. Nat Med 2005;11:183–190.

8. Manicassamy S, Gupta S, Huang Z, et al. Protein kinase C-theta-mediated signals enhance CD4�T cell survival by up-regulatingBcl-xL. J Immunol 2006;176:6709–6716.

9. Patsouris D, Reddy JK, Muller M, et al. Peroxisome proliferator-activated receptor alpha mediates the effects of high-fat diet onhepatic gene expression. Endocrinology 2006;147:1508–1516.

0. Robertson G, Leclercq I, Farrell GC. Nonalcoholic steatosisand steatohepatitis. II. Cytochrome P-450 enzymes and oxida-tive stress. Am J Physiol Gastrointest Liver Physiol 2001;281:G1135–G1139.

1. Saraiva RM, Minhas KM, Zheng M, et al. Reduced neuronal nitricoxide synthase expression contributes to cardiac oxidative stressand nitroso-redox imbalance in ob/ob mice. Nitric Oxide 2007;16:331–338.

2. Alloatti G, Penna C, De Martino A, et al. Role of nitric oxide andplatelet-activating factor in cardiac alterations induced by tumornecrosis factor-alpha in the guinea-pig papillary muscle. Cardio-vasc Res 1999;41:611–619.

3. Simbulan-Rosenthal CM, Rosenthal DS, Iyer S, et al. Transientpoly(ADP-ribosyl)ation of nuclear proteins and role of poly(ADP-ribose) polymerase in the early stages of apoptosis. J Biol Chem1998;273:13703–13712.

4. Charlton M, Sreekumar R, Rasmussen D, et al. Apolipoproteinsynthesis in nonalcoholic steatohepatitis. Hepatology 2002;35:898–904.

5. Tabas I. Consequences of cellular cholesterol accumulation: ba-sic concepts and physiological implications. J Clin Invest 2002;

110:905–911.

3

3

3

3

4

4

4

4

4

4

4

4

4

4

5

5

5

5

cAu

aDtmL

gDUS

December 2008 INSULIN RESISTANCE IN NASH PATHOGENESIS 2095

6. Soccio RE, Breslow JL. Intracellular cholesterol transport. Arte-rioscler Thromb Vasc Biol 2004;24:1150–1160.

7. Mari M, Caballero F, Colell A, et al. Mitochondrial free cholesterolloading sensitizes to TNF- and Fas-mediated steatohepatitis. CellMetab 2006;4:185–198.

8. Shen WJ, Patel S, Natu V, et al. Interaction of hormone-sensitive lipase with steroidogenic acute regulatory protein:facilitation of cholesterol transfer in adrenal. J Biol Chem2003;278:43870–43876.

9. Crespo J, Cayon A, Fernandez-Gil P, et al. Gene expression oftumor necrosis factor alpha and TNF-receptors, p55 and p75, innonalcoholic steatohepatitis patients. Hepatology 2001;34:1158–1163.

0. Bigorgne AE, Bouchet-Delbos L, Naveau S, et al. Obesity-inducedlymphocyte hyperresponsiveness to chemokines: a new mecha-nism of fatty liver inflammation in obese mice. Gastroenterology2008;134:1459–1469.

1. Enriquez A, Leclercq I, Farrell GC, et al. Altered expression ofhepatic CYP2E1 and CYP4A in obese, diabetic ob/ob mice, andfa/fa Zucker rats. Biochem Biophys Res Commun 1999;255:300–306.

2. Emery MG, Fisher JM, Chien JY, et al. CYP2E1 activity before andafter weight loss in morbidly obese subjects with nonalcoholicfatty liver disease. Hepatology 2003;38:428–435.

3. Lieber CS. CYP2E1: from ASH to NASH. Hepatol Res 2004;28:1–11.

4. Schulze-Osthoff K, Beyaert R, Vandevoorde V, et al. Depletionof the mitochondrial electron transport abrogates the cytotoxicand gene-inductive effects of TNF. EMBO J 1993;12:3095–3104.

5. Shimizu S, Eguchi Y, Kamiike W, et al. Involvement of ICE familyproteases in apoptosis induced by reoxygenation of hypoxichepatocytes. Am J Physiol 1996;271:G949–G958.

6. Carter-Kent C, Zein NN, Feldstein AE. Cytokines in the pathogen-esis of fatty liver and disease progression to steatohepatitis:implications for treatment. Am J Gastroenterol 2008;103:1036–1042.

7. Kremer M, Hines IN, Milton RJ, et al. Favored T helper 1

response in a mouse model of hepatosteatosis is associated V

with enhanced T cell-mediated hepatitis. Hepatology 2006;44:216–227.

8. Tiegs G. Cellular and cytokine-mediated mechanisms of inflam-mation and its modulation in immune-mediated liver injury.Z Gastroenterol 2007;45:63–70.

9. Biddinger SB, Hernandez-Ono A, Rask-Madsen C, et al. Hepaticinsulin resistance is sufficient to produce dyslipidemia and sus-ceptibility to atherosclerosis. Cell Metab 2008;7:125–134.

0. Angulo P, Keach JC, Batts KP, et al. Independent predictors ofliver fibrosis in patients with nonalcoholic steatohepatitis. Hepa-tology 1999;30:1356–1362.

1. Chalasani N, Crabb DW, Cummings OW, et al. Does leptin play arole in the pathogenesis of human nonalcoholic steatohepatitis?Am J Gastroenterol 2003;98:2771–2776.

2. Manco M, Marcellini M, Giannone G, et al. Correlation of serumTNF-alpha levels and histologic liver injury scores in pediatricnonalcoholic fatty liver disease. Am J Clin Pathol 2007;127:954–960.

3. Ong JP, Elariny H, Collantes R, et al. Predictors of nonalcoholicsteatohepatitis and advanced fibrosis in morbidly obese pa-tients. Obes Surg 2005;15:310–315.

Received May 29, 2008. Accepted August 14, 2008.Address reprint requests to: Sonia M. Najjar, PhD, College of Medi-

ine, University of Toledo, Health Science Campus, 3000 Arlingtonvenue, Mail Stop 1008, Toledo, Ohio 43614. e-mail: sonia.najjar@toledo.edu; fax: (419) 383-2871.The authors thank Dr Sandrine Pierre for advice on lipid peroxidation

ssay, Dr R. Mark Wooten for his guidance on inflammation analysis, andr Andrea Kalinoski for help with confocal microscopy. The authors also

hank Anthony DeAngelis, Jill M. Schroeder-Gloeckler, Sadeesh Ra-akrishnan, Jehnan Liu, and Jennifer Kalisz (Najjar laboratory), and Steve

ear (Erickson laboratory) for excellent technical assistance.S.J.L. and G.H. contributed equally to this work.The authors disclose the following: This work was supported by

rants from the National Institutes of Health (DK 54254 to S.M.N. andK072187 to S.K.E.), the American Diabetes Association (S.M.N.), thenited States Department of Agriculture (USDA 38903-02315 to.M.N. and M.F.M.), and by a Merit Award from the Department of

eteran’s Affairs (to S.K.E.).

BA

SIC–L

IVER

,PA

NCREA

S,A

ND

BIL

IARY

TRA

CT